Advances in
CANCER RESEARCH Volume 76
This Page Intentionally Left Blank
Advances in
CANCER RESEARCH Volume 76
E...
58 downloads
1169 Views
15MB 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
Advances in
CANCER RESEARCH Volume 76
This Page Intentionally Left Blank
Advances in
CANCER RESEARCH Volume 76
Edited by
George F. Vande Woude ABL-Basic Research Program National Cancer Institute Frederick Cancer Research and Development Center Frederick, Maryland
George Klein Microbiology and Tumor Biology Center Karolinska lnstitutet Stockholm, Sweden
ACADEMIC PRESS San Diego London Boston New York Sydney Tokyo Toronto
This book is printed on acid-free paper. @
Copyright 0 1999 by ACADEMIC PRESS All Rights Reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system. without permission in writing from the Publisher. The appearance of the code at the bottom of the first page of a chapter in this book indicates the Publisher’s consent that copies of the chapter may be made for personal or internal use of specific clienls. This consent is given on the condition, however. that the copier pay the stated per copy fee through the Copyright Clearance Center. Inc. (222 Rosewood Drive. Danvers. Massachusetts 01923). for copying beyond that permitted by Sections 107 or 108 of the U S . Copyright Law. This consent does not extend to other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works, or for resale. Copy fees for pre-1999 chapters are as shown on the title pages. If no fee code appears on the title page. the copy fee is the same as !or current chapters. 0065-230X/99 530.00
Academic Press a division of Hurcourt Brnce & Company
525 B Street. Suite 1900, San Diego. California 92101-4495, USA http:l/www.apnet.com
Academic Press
24-28 Oval Road, London NW 1 7DX. U K http:l/wu w.hbuk co uklapl international Standard Book Number: 0- 12-006676-9
PRIhTED LN THE UNITED STATES OF AMERICA 9 9 W 0 1 0 2 0 3 0 4 E B 9 8 7 6 5 4 3 2
1
Contents
Contributors to Volume 76 ix
Fibronectin and Its Integrin Receptors in Cancer Erkki Ruoslahti I. Introduction
1
11. Reduced Adhesiveness Is Needed for Detachment and Migration
3
111. Anchorage Dependence and Anoikis 5 IV. Cell Migration and Invasion 10 References 16
Myb and Oncogenesis Brigitte Canter a n d Joseph S. Lipsick I. Introduction 21 11. The M y b Genes 22 111. Structural and Functional Features of the Myb Proteins IV. Regulation of v-Myb and c-Myb 41 V. Transcriptional Regulation by v-Myb and c-Myb 46 VI. The Myb-Chromatin Connection 50 References 52
36
c-Src, Receptor Tyrosine Kinases, and Human Cancer Jacqueline S. Biscardi, David A. Tice, a n d Sarah 1. Parsons I. Introduction
61
11. Receptor Tyrosine Kinases and Human Cancers
63 111. c-Src and c-Src Family Members in Human Cancers 78 IV. Mechanisms of c-Src Action 89 V. Potential Therapeutic Applications of c-Src/HERl Interactions References 103
102
V
vi
Contents
Epidemiology of Kaposi’s Sarcoma-Associated Herpesvirushluman Herpesvlrus 8 Thomas F. Schulz I. 11. 111. IV. V.
Introduction 121 KSHV Phylogeny and Molecular Epidemiology 122 Geographic Distribution 124 KSHV Prevalence in Risk Groups for HIV-1 Transmission Transmission of KSHV 141 VI. Association of KSHV with Disease 145 VII. Conclusion 153 References 154
137
Consensus on Synergism between Cigarette Smoke and Other Environmental Carcinogens in the Causation of Lung Cancer Amold E. Reif a n d Timothy Heeren 1. Introduction 161 11. Testing the Significance of a Finding of Synergism 165 III. Carcinogenic Synergism and Public Health 172 IV. Previous Findings on Synergism Involving Cigarette Smoke 176 V. Multistep Carcinogenesis 177 VI. Varying the Time Frame of Data Collection 180 VII. Conclusion 182 References 182
Carcinogenesis and Natural Selection: A New Perspective to the Genetics and Epigenetics of Colorectal Cancer JadeBreivik a n d Gustav Gaudemack I. Introduction 187 11. Evolution and Cancer 188 111. The Microsatellite Instability Pathway 192 IV. The Chromosomal Instability Pathway 196 V. MIN versus CIN 199 VI. DNA Methylation and the Epigenetics of Cancer 200 W. Location-Related Carcinogenic Environments 206 VIII. Conclusion and Perspectives 208 References 209
Antitumor lmmunity at Work in a Melanoma Patient Pierre G . Coulie, Hideyuki Ikeda, Jean-Francois Baurain, a n d Rita Chiari I. htroduction 214
Contents 11. Melanoma Patient LB33 and Melanoma Cell Lines 216 111. Autologous CTLs against MEL.A Cells 218 IV.Identificationof Antigens Recognized by CTLs on MEL.A Cells 223 V. The MEL.B Cells 227 VI. A New Class of Antitumor CTL 232 VII. Conclusions 238 References 239
Index 243
This Page Intentionally Left Blank
Contributors
Numbers in parentheses indicate the pages on which the authors’ contributions begin.
Jean-Franqois Baurain, Catholic University of Louvain, Cellular Genetics Unit, B-1200 Brussels, Belgium (213) Jacqueline S. Biscardi, Department of Microbiology and Cancer Center, University of Virginia Health Sciences Center, Charlottesville, Virginia 22908 (61) Jarle Breivik, Section for Immunotherapy, The Norwegian Radium Hospital, N-0310 Oslo, Norway (187) Rita Chiari, Catholic University of Louvain, Cellular Genetics Unit, B-1200 Brussels, Belgium (213) Pierre G. Coulie, Catholic University of Louvain, Cellular Genetics Unit, B1200 Brussels, Belgium (213) Brigitte Ganter, Department of Pathology, Stanford University School of Medicine, Stanford, California 94305 (21) Gustav Gaudernack, Section for Immunotherapy, The Norwegian Radium Hospital, N-0310 Oslo, Norway (187) Timothy Heeren, Department of Epidemiology and Biostatistics, Boston University School of Public Health, Boston, Massachusetts 021 18 (161) Hideyuki Ikeda, Division of Cellular Signaling, Institute for Advanced Medical Research, Keio University School of Medicine, Shinjuku-ku, 160 Tokyo, Japan (213) Joseph S . Lipsick, Department of Pathology, Stanford University School of Medicine, Stanford, California 94305 (21) Sarah J. Parsons, Department of Microbiology and Cancer Center, University of Virginia Health Sciences Center, Charlottesville, Virginia 22908 (61) Arnold E. Reif, Mallory Institute of Pathology, Boston University School of Medicine, Boston, Massachusetts 02118 (161) Erkki Ruoslahti, Cancer Research Center, The Burnham Institute, La Jolla, California 92037 (1)
X
Contributors
Thomas F. Schulz, Molecular Virology Group, Department of Medical Microbiology,The University of Liverpool, Liverpool L69 3GA, United Kingdom (121) David A. Tice, Department of Microbiology and Cancer Center, University of Virginia Health Sciences Center, Charlottesville, Virginia 22908 (61)
Fibronectin and Its Integrin Receptors in Cancer Erkkl Ruoslahtl Cancer Research Center The Burnham Institute LA Jolla, California 92037
I. Introduction 11. Reduced Adhesiveness Is Needed for Detachment and Migration 111. Anchorage Dependence and Anoikis IV. Cell Migration and Invasion A. Tumor Cells in Circulation-Possible Antimetastatic Effect of Fibronectin in Plasma B. Site-SpecificMetastasis-Vascular Specificities References
The adhesive extracellular matrix protein fibronectin and its integrin receptors play important roles at several stages of tumor development. Tumor cells are generally less adhesive than normal cells and deposit less extracellular matrix. The loosened matrix adhesion that results may contribute to the ability of tumor cells to leave their original position in the tissue. Normal cells, when detached, stop growing and undergo anoikis (apoptosiscaused by loss of adhesion). Integrin-activated pathways mediated by focal adhesion kinase (FAK) and the adapter protein Shc seem to be particularly important in anchorage dependence; many oncoproteins are capable of shunting these pathways. Malignant cells circumvent anchorage dependence with the help of oncoproteins. Once invading tumor cells have gained access to the circulation, adhesion to the endothelia and other tissue components facilitates the establishment of tumor colonies at distant sites. Specific tissue affinitiesmay underlie the tendency of some tumors to metastasize preferentially to certain tissues. Interfering with tumor cell attachment with integrin-binding peptides has been shown to he an effective antimetastatic strategy in animal experiments. Tumor angiogenesisis yet another aspect of malignancy wherein extracellular matrices and integrins are important. Angiogenic endothelial cells in tumor vessels depend on the av family of integrins for survival. Inhibiting angiogenesis with compounds that block the activity of av integrins, and targeting drugs into tumors through these integrins, show promise as new anticancer strategies. 0 1999 Academic Press.
I. INTRODUCTION Cancer is a disease of tissue architecture. In forming tissues and organs during development, cells specialize and migrate to their appropriate places Advances in CANCER RWURCH
0065-230W99 $30.00
Copyright B 1999 by Academic Press. All rights of reproduction in any form reserved.
2
Erkki Ruoslahti
in an orderly way. After this process is completed, the body plan is strictly maintained, except in cancer. Cancerous cells acquire the ability to breach tissue barriers, trespassing into adjacent tissues and distant sites in the body. This process, metastasis, is the most devastating aspect of cancer. Metastatic cancer has usually reached so many places that cure by surgery becomes impossible. For that reason, invasion into normal tissue and metastasis are the hallmarks of malignancy. A benign tumor that is not removed can get very large; the cells that make up such a tumor obviously overproliferate, but unlike malignant cancer cells, they do not invade or metastasize. Adhesive interactions are thought to play a major role in the construction of the body plan during development. These interactions comprise an area code system that guides cells to their appropriate locations in the body and anchors them there. Adhesion is also important in the maintenance of the body plan. Thus, it is not surprising that cell adhesion molecules-integrins and their extracellular matrix ligands in particular-are important in cancer. Integrins are the main class of cell adhesion receptors for extracellular matrices and can also serve as cell-cell adhesion receptors (Hynes, 1992; Springer, 1994; Ruoslahti, 1996). Integrins are membrane proteins that consist of an 01 and a p subunit, each with a molecular mass in the 100- to 200kDa range. Both subunits span the cell membrane, with most of the polypeptide outside the cell. The cytoplasmic domains of the integrin subunits, with one exception, are short, 30- to 50-amino acid peptides. There are 15 known 01 subunits and 8 p subunits, which combine into some 25 different integrins. The P l , p2, and a v subunits can pair with a particularly large number of possible companion subunits, each defining its subfamily among integrins. The extracellular ligand-binding specificity of an integrin is generated jointly by the a and p subunits of the integrin. Integrins display specificity on several levels. First, they are expressed in a cell-type- and stage-specific manner. Thus, one group of integrins is associated with migration and proliferation in various types of cells. These “emergency integrins” include a 5 p 1 , avp3, and avp6 (Sheppard, 1996). These integrins may be particularly important in cancer. Many other integrins are selectively expressed in a certain cell type or a few cell types. Examples of cell-type-specific integrins include aIIbp3 in platelets and a 6 p 4 in epithelial cells. Another level of integrin specificity is manifested in their ligand binding. Many of the integrins bind the RGD cell attachment sequence, but they recognize that sequence differentially in the context of various extracellular matrix proteins, such that some bind primarily to fibronectin and others to vitronectin (Ruoslahti, 1996). At yet another level of specificity, individual integrins mediate distinct signals into the interior of the cell (Schwartz et al., 1995; Clark and Brugge, 1995; Juliano and Haskill, 1993). Integrins have been shown to be signaling molecules capable of generating both common signals and signals that are specific for individual integrins. It
Fibronectin and Its Integrin Receptors
3
has also been found that information can flow in the reverse direction through integrins; their ligand-binding ability is regulated from inside the cell (Chan and Hemler, 1993; Zhang et al., 1996; Kolanus et al., 1996; Hughes et al., 1997). In this article, I review the current understanding of the role of integrins and their ligands in the development and dissemination of cancer. Also discussed are a number of emerging integrin-based cancer treatments.
11. REDUCED ADHESIVENESS IS NEEDED
FOR DETACHMENT AND MIGRATION To be able to emigrate to another tissue, cancer cells have to detach from their original location, invade a blood or lymphatic vessel, travel in the circulation to a distant site, and establish a new cellular colony (Fig. 1).At every one of these steps, they must escape a number of controls that keep normal cells in place. The first step in cancer invasion is likely to be loosening of the adhesive restraint between cells. Both cell-extracellular matrix adhesion and cell-cell adhesion contribute to keeping cells in place. Indeed, adhesion molecules that mediate these interactions are frequently missing or compromised in cancer cells. The first adhesion anomaly discovered was the loss of fibronectin matrix in malignantly transformed cells in the early 1970s (Ruoslahti, 1988). Fibronectin is the prototype cell attachment protein found in the matrix surrounding normal cells. The fibronectin matrix mediates cell adhesion and anchorage through a number of fibronectin-binding integrins, including a5pl. The fibronectin matrix appears to be an important constraint on cells, because its restoration by various means suppresses cell migration and tumorigenicity. Fibronectin does not form matrix spontaneously-the cell uses its a5p1 integrin to capture secreted fibronectin and convert it into the fibrils that are then deposited into the matrix. Forced expression of the a5pl integrin in tumor cells reduces their motility and tumorigenicity (Giancotti and Ruoslahti, 1990). Conversely, a decrease in a5pl expression increases the tumorigenicity of CHO cells (Schreiner et al., 1991). Some of the a5pl integrin effect may be related to a signaling role of the integrin (Varner et al., 1995; Varner and Cheresh, 1996). However, the increase in fibronectin matrix assembly that accompanies an increase in a5pl expression or activity (Giancotti and Ruoslahti, 1990; Wu et al., 1996) appears to be responsible for most of the effect. Thus, expression of fibronectin from a transfected cDNA can convert tumorigenic cells into nontumorigenic ones (Wu et al., 1996), and the
4
Erkki Ruoslahti
Fig. 1 Critical stages in tumor metastasis.
fibrillar form of fibronectin, superfibronectin, has a tumor-suppressive effect in vitro and in vivo. I discuss superfibronectin later on in this review. Malignant transformation by oncogenes generally reduces cellular adhesiveness, and this may be one critical element in tumorigenesis. At least two oncogenes, Src and Ras, can impair integrin activity (Akamatsu et al., 1996; Hughes et al., 1997). The Src oncogene can directly phosphorylate the p l integrin subunit, causing loss of p l integrin ligand-binding affinity (Akamatsu et a/., 1996). Src also phosphorylates a number of signaling molecules associated with integrins; the effect appears to be to shunt the integrin sig-
Fibronectin and Its lntegrin Receptors
5
naling pathways that control anchorage dependence (below). The influence of Ras on integrins was discovered in experiments designed to reveal cDNAs capable of reducing integrin affinity for their ligands (Hughes et af., 1997). Importantly, by impairing the a5pl integrin, both Src and Ras reduce fibronectin matrix deposition. In addition, oncogenic Src reduces fibronectin synthesis by the cells transformed with it (Adams et al., 1977). All these changes contribute to reduced adhesiveness and the increased propensity to migrate characteristic of transformed cells. Among the cell-cell adhesion molecules, E-cadherin behaves as a tumor suppressor. It is often lost on malignant transformation, either because of disabling mutations in its gene or in genes for the cytoplasmic proteins, catenins, that activate E-cadherin (Birchmeier, 1995). Alternatively, the expression of the E-cadherin gene can become down-regulated. Moreover, by manipulating this molecule in cultured cancer cells, one can change the cells’ ability to invade tissues and form tumors. Blocking the function of E-cadherin can turn a cultured lineage of cells from noninvasive to invasive. Conversely, restoring E-cadherin to cancer cells that lack it can negate their ability to form tumors when they are injected into mice. The a5pl integrin and N-cadherin coordinately regulate cell proliferation and migration (Huttenlocher et al., 1998). Moreover, integrin-linked kinase (ILK), a kinase that binds to the cytoplasmic domains of integrin p subunits and regulates integrin activity, also influences the activity of the LEF-l/TNF transcription factor, which is associated with p-catenin (Novak et al., 1998). Thus, both cell-matrix and cell-cell adhesion molecules individually and concordantly develop abnormalities that permit cancer cells to leave their original tissue site.
111. ANCHORAGE DEPENDENCE AND ANOIKIS The loss of tissue attachment discussed above creates a dilemma for a cancer cell about to invade; cells need attachment to be able to grow and even to survive. This leads to one of the most fundamental requirements in cancer progression, that cancer cells must become independent of anchorage. Fibroblasts dissociated from their extracellular matrix become arrested in their growth (Folkman, 1978). This phenomenon has been recognized for many years and is known as anchorage dependence on growth. The arrested growth of detached fibroblasts is reversible; on reattaching they regain their ability to proliferate. More recent is the realization that anchorage dependence is an integrin-mediated event; the cell attachment must be through integrins-other cell surface molecules will not suffice (Meredith et al., 1993; Frisch and Francis, 1994; Re et al., 1994; Boudreau et al., 1995). Moreover,
6
Erkki Ruoslahti
endothelial and epithelial cells, when displaced from the extracellular matrix, undergo apoptosis. The likely biological purpose of this response, termed anoikis (Frisch and Francis, 1994), is that it would cause detached cells to die before they could reattach at new locations and disturb normal tissue architecture. Evidence now indicates that to satisfy the anchorage dependence requirement it is not enough for cells to attach to a matrix through any integrin, they have to attach through the correct integrin. The a5pl integrin may be particularly important in providing proper attachment. Thus, experiments show that cells engineered to express the a5Pl integrin as their fibronectin receptor survive for an extended period when cultured on fibronectin in the absence of serum, whereas cells with an alternative fibronectin receptor, olvpl, undergo apoptosis under the same conditions (Zhang et aL, 1995; O’Brien et al., 1996). No other integrin among several tested was able to substitute for d p l , even when the test cells were cultured on matrix proteins that serve as ligands for those integrins (Zhang et al., 1995; Z. Zhang and E. Ruoslahti, unpublished). The reason for the survival of the a5plexpressing cells appears to be that they express increased amounts of the antiapoptotic protein Bcl-2 (Zhang et al., 1995). Other integrins may be needed under other conditions. Thus, angiogenic endothelial cells depend on the avP3 integrin and epithelial cells on laminin-binding integrins in vivo (Brooks, 1995).The antiapoptotic effect of av@3in endothelial cells may be routed through NF-KB(Scatena et al., 1998). Figure 2 depicts some of these pathways. The integrin-selective survival signals are likely to provide a further safeguard for the maintenance of the integrity of normal tissues, because cells of one type would not be able to survive a change of location by attaching to a new place with a different matrix. All these results show that a molecular explanation for anchorage dependence is beginning to take shape, although much still remains to be learned. Neoplastic cells develop changes that enable them to circumvent the integrin signaling requirement and become independent of their anchorage. It is important to note that I am using here the terms anchorage dependence and anchorage independence as manifested in suspension culture. The classical definition of these terms derives from behavior of cells in semisolid media. Although that method of determining anchorage dependence correlates with in vivo tumorigenicity (Kahn and Shin, 1979), there are exceptions. For example, transforming growth factor @ (TGFP)can support the growth of anchorage-dependent, nontumorigenic cells in semisolid media, apparently because TGFP increases the expression of fibronectin and the integrins that bind to fibronectin (MassaguP, 1987).The ability of cells to survive and grow in suspension without any support is less likely to be altered by TGFP. Cancer cells get around the requirement of integrin-mediated attachment by shunting the integrin signaling pathways that control anchorage depen-
7
Fibronectin and Its Integrin Receptors
f
Ras
9
n
I
f
-\
- extracellular xp
matrix
Fig. 2 Integrin pathways preventing anoikis and some of their oncoprotein substitutes. See text for references.
dence. They can also protect themselves against anoikis by up-regulating an antiapoptotic protein, such as Bcl-2 (Zhang et al., 1995). Oncogenes often have as one of their activities the ability to confer a cell with anchorage independence. In some cases, this may be the only activity of an oncogene. One can distinguish between an oncogenic activity that affects both growth and anchorage from an effect that is directed only at anchorage by determining whether the transformed cells require serum for growth in culture. When the oncoprotein activity affects only anchorage dependence, the cells are able to survive and grow in suspension, but they require serum (or growth factors). The anchorage-related signalling pathways overtaken by oncoproteins originate at subcellular structures known as focal adhesions. Focal adhesions are specialized cell-substrate contacts, in which integrin clusters link actin filaments to the substrate or to extracellular matrix. As a result of the integrin and cytoskeletal clustering, various signaling molecules become concentrated in focal adhesions at the cytoplasmic surface of the cell membrane. Thus, focal adhesions appear to contain the highest concentration of proteins phosphorylated at tyrosine residues, a hallmark of signaling molecules (Burridge and Chrzanowska-Wodnicka, 1996). The high reactant concentration in focal adhesions and the proximities of multiple components necessary to complete signal transduction through a pathway facilitate efficient signaling. One important proximity is that of the cell membrane, because many signaling molecules require membrane association to be active
8
Erkki Ruoslahti
(e.g., Ras) or have their enzymatic substrates in the membrane leg., phosphoinositol-3' (PI-3') kinase]. The formation of focal adhesion on integrin-mediated cell attachment results in the activation of a number of protein tyrosine kinases. These kinases include focal adhesion kinase (FAK),a cytoplasmic protein tyrosine kinase that binds directly to integrin cytoplasmic domains and plays a central role in integrin signaling (Parsons, 1996). Activated FAK connects to the Ras/ Faf/MAP kinase mitogenic pathway through the adapter protein Grb-2 (Schlaepferet al., 1994) and FAK activation also leads to an activation of PI3' kinase (Chen and Guan, 1994). The activation of FAK on the binding of integrins to their extracellular matrix ligands and the ensuing cell spreading may be the critical requirement that makes a normal cell anchorage dependent; without FAK activation, the cell will not grow and may undergo apoptosis. The evidence that suggests this role for FAK is that FAK confers anchorage independence to epithelial [Madin-Darby canine kidney (MDCK)]cells without affecting their other growth properties (Frisch et al., 1996), and that reducing FAK expression with antisense cDNA can induce anoikis in previously anchorage-independent celis (Hungerford et al., 1996). However, there is some controversy in the literature regarding the importance of FAK in anoikis. An alternative pathway involving the adapter protein Shc has been proposed as the conduit of this effect (Wary et al., 1996). Interestingly, this pathway is activated only by certain integrins, not all of them. FAK and Shc, and as yet unidentified pathways, all of which converge on Grb2, cooperate in mediating the activation of the extracellular signal-regulated kinase/mitogen-activated protein (ERWMAP) kinase pathway through Ras and Raf-1 (Schlaepfer et al., 1994). The activation of the phosphatidyl inositol-3' kinase (PI3-K) seems to be more important in anoikis than the MAP kinase pathway (Marte et al., 1997; Frisch and Ruoslahti, 1997), but it may be that this pathway may also respond to cell attachment through FAK and other signaling molecules. FAK is associated with the oncoproteins Src, p13OCUs,a recently characterized docking protein, and certain cytoskeleton-associated proteins such as paxillin and cortactin, all of which become activated by phoshorylation on cell attachment and spreading (Parsons, 1996; Polte and Hanks, 1997).One important activity of various oncoproteins appears to be that they cause downstream activation of the integrin-FAK pathway so that integrin-mediated adhesion is no longer needed for cell survival and growth. This makes the cells anchorage independent. Some tumor cells express elevated levels of FAK (Brunton et al., 1997), and this may be at least partly responsible for loss of anchorage dependence. The oncogenic variants of Src cause increased phosphorylation of several focal adhesion components, including FAK and pl3OCas (Kanner et aL, 1990). The phosphorylation of these proteins is independent of cell adhesion in Src-transformed cells (Schlaepfer and Hunter,
Fibronectin and Its Integrin Receptors
9
1996; Vuori et al., 1996), and, therefore, presumably generates a “false” anchorage signal. The oncogenic adapter protein v-Crk may function similarly, because it causes increased phosphorylation of p13OcuSand changes the subcellular localization of p13OcUs (Matsuda et al., 1990; Nievers et al., 1997). Ilk is a recently discovered serinehhreonine kinase that may play a role in integrin regulation of malignancy (Hannigan et al., 1996). Ilk binds to the cytoplasmic domain of the PI integrin subunit and the binding suppresses kinase activity of Ilk. When Ilk is overexpressed and active, it promotes malignant transformation and anchorage independence. The ability to circumvent the anchorage requirement may be as important an activity of oncoproteins as is their ability to stimulate cell growth. The FAK-Ras-MAPK and Shc-Ras-MAPK pathways are examples of how oncogene activation can substitute for cell attachment. How the signals that couple loss of cell attachment to inhibited growth and apoptosis are carried into the nucleus is not entirely clear. It may be simply the lack of the positive signals discussed above, but there is also evidence that an “empty” (not extracellular ligand-bound) integrin can generate a negative signal (Varner et al., 1995). In their natural environment tissue cells are in contact with extracellular matrix, rather than a cell culture substrate. Malignantly transformed cells generally elaborate less extracellular matrix around themselves than do normal cells, but they may need some matrix to be able to grow. It seems that a dense matrix and a total loss of matrix will both prevent a cell from growing and migrating, and that there is an optimal amount of matrix for these functions. Thus, forced expression of the a5Pl integrin causes abundant fibronectin matrix production and suppresses tumorigenicity (Giancotti and Ruoslahti, 1990),whereas preventing cultured tumor cells from making any fibronectin matrix inhibits their growth (Saulineret al., 1996) and migration (Bourdoulous et al., 1998). The link between the matrix and the regulation of growth and survival is likely to be the actin cytoskeleton. Actin stress fiber organization in spread cells depends on fibronectin matrix (Bourdoulous et al., 1998). On the other hand, a close correlation exists between actin stress fiber formation and fibronectin matrix assembly. Thus, disrupting the cytoskeleton with cytochalasin D inhibits fibronectin matrix assembly, whereas lysophosphatidic acid (LPA), which induces stress fiber formation, enhances it (Zhang et al., 1994). LPA probably acts through the small GTPase Rho, because inhibiting Rho with C3 transferase inhibits the LPA-induced matrix assembly (Zhong et al., 1998). Rho is a member of the Ras superfamily of GTPases that controls the cytoskeletal rearrangements accompanying cell spreading. Integrins provide the connection that makes the extracellular matrix and the cytoskeleton interdependent. Interaction of integrins with the actin cy-
10
Erkki Ruoslahti
toskeleton i s also essential for matrix assembly (Hynes, 1992) and truncation of the PI cytoplasmic domain blocks fibronectin matrix assembly by preventing the interaction of the ligand-occupied integrin with the cytoskeleton (Wu e t a / . , 1996). One indication of the importance of the extracellular matrix and cytoskeleton in cell growth, migration, and malignancy is the evidence that farnesyl transferase inhibitors may suppress malignancy through Rho, as well as Kas (Lebowitz et a/., 1997). Like Ras, Rho requires farnesylation to be active, and it has now been found that Rho may be an important target for the drugs that were designed as inhibitors of Ras farnesylation. In the nucleus, the growth inhibitory signals from the loss of cell attachment appear to be mediated by changes in nuclear cyclidcdk complexes. The two main changes are reduced expression of cyclin D at the protein level, and a loss of cyclin E/Cdk2 activity that is secondary to an up-regulation of the p21Cip and p27Kip inhibitors of cyclidcdk kinases (Assoian and Marcantonio, 1996; Fang et al., 1996; Shulze et al., 1996; Zhu et al., 1996).Removal of extracellular matrix has similar effects (Bourdoulous et al., 1998). Importantly, the cyclin E/Cdk2 complex stays active in anchorage-independent cells that are not attached. One mechanism whereby this is accomplished by a malignant cell is increased expression of cyclin E, which neutralizes the inhibitors (Fang et al., 1996), but other mechanisms, such as displacement of the Cdk inhibitor from the nucleus into the cytoplasm (Orend et al., 1998), are likely to play a role as well. The end result is that malignant cells can proliferate regardless of attachment to a substrate. In contrast, removal of all fibronectin matrix, despite the similarity of some of its effects to those from loss of substrate attachment, does not seem to be as easily circumvented by tumor cells as is anchorage dependence (Sauliner et al., 1996; Bourdoulous et al., 1998). Perhaps, the matrix also controls pathways other than the induction of cyclin inhibitors. This dual role of cell attachment and matrix formation may explain why fibronectin and its receptors do act as classical tumor suppressors (Taverna et al., 1998); downregulation of their functions enhances the malignant behavior of cells, but their complete absence is a disadvantage to a tumor cell.
IV. CELL MIGRATION A N D INVASION Anchorage dependence is only one of the constraints that a cancer cell must overcome to move away from its original site. Epithelial cells are separated from the underlying tissue by a basement membrane, a thin layer of specialized extracellular matrix. Basement membranes form a barrier that most types of normal cells cannot breach; the main exception is leukocytes,
Fibronectin and Its Integrin Receptors
11
whose ability to penetrate basement membranes and other tissue barriers makes it possible for them to reach sites of inflammation. Tumor cells are also adept at penetrating basement membranes and invading tissues. Moreover, highly metastatic cells generally invade more effectively than do those from nonmetastatic tumors in assays that model the invasive process in vitro, and their invasive properties can be suppressed with integrin-blocking peptides (Gehlsen et al., 1988). Apparently, integrin-mediated cell adhesion provides the traction for tumor cell migration (Albelda et al., 1990).Various proteases that tumor cells make or activate are important for tissue penetration by tumor cells (Stetler-Stevensonet al., 1993; Zou et al., 1994). Integrins control the expression of some of these proteases (Werb et al., 1989; Seftor et al., 1992).They also regulate the cell surface localization of another protease important in tumor invasion, urokinase, by directing its receptor to focal adhesions (Seftor et al., 1992; Pollanen et al., 1987). Proteases on cancer cell surfaces may have roles other than helping the cell take down matrix barriers; some cell surface proteases process growth factors (Blobel, 1997). Another activity that may be important in invasion relates to deposition of fibrin around tumor nodules (Brown et al., 1988).Fibrin, together with the fibronectin that binds to it, forms a provisional wound matrix (Gailit and Clark, 1994).Like wound matrix, the new matrix around the tumor nodules may provide a favorable substrate for cells to migrate into, facilitating invasion. The d p l , avf33, and other a v integrins are among the integrins that play a prominent role in cell migration associated with tissue remodeling and cancer. Thus, for example, skin epithelial cells express the a v integrin subunit at the leading edge of the epidermis migrating into a wound. The a v is paired with the f35 and f36subunits in these cells (Sheppard, 1996; Zambruno et al., 1995; Gailit et al., 1994). The asp1 integrin is also expressed in migrating, but not resting, keratinocytes (Zambruno et al., 1995). Melanomas appear to depend on the avp3 integrin. This integrin has been shown to be important for the migration, survival, and tumorigenicity of melanoma cells (Montgomery et al., 1994). These melanoma cells lose their ability to form tumors if they are selected not to express the avp3 integrin, and they regain that ability when avf33 expression is restored by cDNA transfection. These changes are thought to depend on a role of avp3 in maintaining the survival of the melanoma cells; they die by apoptosis if their avp3 integrin is not engaged in substrate attachment. That the a v p 3 integrin is important for progression of melanomas in vivo is further suggested by a correlation that has been found between avp3 expression and invasiveness of the tumors; those melanomas that have progressed to the vertical growth mode (penetrating deeper into the dermis) express avp3 (Albelda et al., 1990; Natali et al., 1997). Although integrins mediate cell migration, their role in tumor invasion is
12
Erkki Ruoslahti
unlikely to depend solely on this activity. Rather, their signaling functions are probably important. In this regard, the avp3 and a5pl integrins have special properties. As discussed above, cell attachment through these integrins activates the intracellular signaling molecule Shc (Wary et af., 1996). The connection to Shc is shared by the collagen-binding integrin u l p l . Shc activates the RadMAP kinase growth-regulatory pathway, and the ability of the a5Pl and a v integrins to activate that pathway through Shc may explain why these integrins are associated with tissue regeneration and tumor invasion. Despite their similarities, the avla5 class of integrins also possesses individually distinct signaling pathways. Thus, avp3 is associated with certain growth factor receptors and can enhance their function, whereas asp1 can control apoptosis through the antiapoptosis protein Bcl-2. avp3 is associated in cells with the intracellular pathways that mediate platelet-derived growth factor (PDGF),insulin, and insulin-like growth factor signaling pathways (Bartfeld etaf., 1993; Schneller etaf., 1997). Highly activated forms of the insulin receptor and PDGFP receptor coimmunoprecipitate with the avp3 integrin, and the mitogenic and chemotactic activities of these growth factors are enhanced in cells bound to a subratrate through this integrin (Vuori and Ruoslahti, 1994; Schneller et al., 1997; Jones et al., 1996). Thus, the avp3-growth factor receptor association may play a role in promoting tumor cell survival, growth, and invasion. As discussed below, the uvp3 integrin is also important in tumor angiogenesis. Having acquired invasive properties and having left its original position, a tumor cell (or a group of tumor cells) is poised to gain access to the circulation. This process involves penetrating the endothelial basement membrane that surrounds small blood vessels, and then the endothelial cell layer. The penetration of blood vessels is likely to be aided by poor integrity of small blood vessels near the advancing tumor cells: tumors cause angiogenesis, and the newly formed vessels have discontinuities (Folkman, 1995).
Tumor Cells in Circulation-Possible Effect of Fibronectin in Plasma
A.
Antimetastatic
Cancer usually metastasizes through the blood circulation or the lymphatics. Experimental metastasis studies suggest that very few cells survive the stay in the circulation. When cultured tumor cells are injected intravenously into mice, only a small minority of them successfully establish a colony in tissues, usually in the lungs, of the animal (Stetler-Stevensonet al., 1993). In one study, patients with intraperitoneal tumors were provided relief from the accumulation of ascites by providing a shunt that allowed the ascitic fluid to drain back into the circulation. Along with the ascites, vast
Fibronectin and Its Integrin Receptors
13
numbers of tumor cells were introduced into the circulation in this procedure. Surprisingly, those patients who survived for extended periods of time did not have extensive new metastases, as one might have expected (Tarin et al., 1984).Thus, it appears that the blood is an unfavorable environment for tumor cells. The vulnerability of malignant cells in the circulation is likely to have many underlying reasons, most of them unknown, but some of them are adhesion related. There may be natural defense mechanisms to eliminate tissue cells, such as tumor cells, that have entered the circulation. Plasma fibronectin may be one of the factors facilitating the elimination of circulating tumor cells. Saba and Cho (1977) injected a purified plasma fraction rich in fibronectin together with tumor cells and found less hematogenous metastasis in the fibronectintreated mice. Superfibronectin (sFN), a recently discovered polymeric form of fibronectin (Morla et al., 1994), inhibits cell migration in cell culture and has striking antimetastatic activities in vivo (Pasqualini et al., 1996). Treatment of tumor cells with sFN in vitro renders the cells nontumorigenic on subsequent injection into mice. More importantly, systemic treatment of mice with sFN strongly inhibits spontaneous metastasis from subcutaneously implanted tumors. A wide variety of tumor types respond to sFN in this manner, including human breast, colon, and ovarian carcinomas, melanoma, and 0steosarcoma implanted into nude mice. The mode of action of sFN requires further study, but when used to treat cells in suspension, it makes them incapable of adhering to any extracellular matrix substrate (Pasqualini et al., 1996). It may be that sFN coats tumor cells that are in transit in vivo, preventing them from attaching to the sites where metastases would otherwise form, and that the tumor cells incapacitated in this manner would then be susceptible to elimination by natural defense mechanisms. One of these mechanisms may be uptake of the sFN-coated tumor cells by the reticuloendothelial system, as has been postulated for fibronectin (Saba and Cho, 1977). sFN has been shown to gain access to the blood after an intrapecitoneal injection (Pasqualini et al., 1996). The fact that sFN does not affect the size of the primary tumor, only suppresses metastasis, also supports the assumption that it may be exerting its effects on circulating tumor cells. Peptides containing the RCD cell attachment sequence also have antimetastatic effects. These peptides mimic the ligands of a number of integrins, and they prevent the binding of integrins to their natural ligands (Ruoslahti, 1996). When RGD peptides are coinjected with tumor cells, fewer colonies will appear in the lungs than if the peptide is omitted, or if an inactive control peptide is injected (Humphries et al., 1986; Hardan et al., 1993; Pasqualini and Ruoslahti, 1996). One reason for this activity of the RGD peptides may be that they may prevent circulating tumor cells from binding
14
Erkki Ruoslahti
platelets (Nierodzik et al., 1992).Platelets can provide growth factors for the tumor cells and may heip them become lodged in small blood vessels, a location favorable for extravasation of the tumor cells (Stetler-Stevenson et al., 1993). However, the RGD peptide effect on experimental metastasis is also seen in mice whose platelets have been depleted (Humphries et al., 1988), indicating that other RGD-dependent adhesion or signaling events are important as well. As discussed later, binding of tumor cells to the endotheliurn of blood vessels is a likely target of the peptide inhibition. Blocking integrin activity may also reduce malignancy through signaling events (Weaver et al., 1997). Current techniques make it possible to detect and quantitate tumor cells in the circulation of patients. Polymerase chain reaction (PCR) detection of mRNAs for epithelial proteins that are not present in normal circulating cells has been particularly useful in this regard (Gross et al., 1995; Jaakkola et al., 199.5; Hoon et al., 1995).As techniques evolve, it should be possible to learn more about the factors that influence the spread of cancers through the bloodstream.
B. Site-Specific Metastasis-Vascular Specificities Various metastatic cancers spread preferentially to certain tissues. Circulatory patterns explain much of this selectivity, because circulating tumor cells usually get trapped in the first vascular bed they encounter, the lungs or the liver. Accordingly, the lungs are the most common site of metastasis for some tumors, and the liver for others. The simplest mechanism for circulating tumor cells to attach to the endothelium is to become physically trapped in small blood vessels, the diameter of which is smaller than that of the tumor cell or a cluster of tumor cells. Tumor cells can produce factors that make platelets aggregate around them. The platelet connection may be the reason why antiplatelet drugs have anticancer effects in some experimental systems (Hardingham et al., 1993). Physical trapping of cancer cells in the blood vessels a t the site of metastasis does not explain all site-specific metastasis; some types of cancer show a striking preference for organs other than those that receive the venous blood. For example, prostate cancer metastasized almost exclusively into the bones. More than 100 years ago, Paget formulated an explanation for the site-specific metastasis of cancers. He postulated that the metastasis preferences would be influenced both by the frequency at which tumor cells are delivered to a given site (“seed”) and by the suitability of the tissue environment for the tumor (“soil”).Specific affinity between the adhesion molecules on tumor cells and those on the endothelium of blood vessels in the preferred tissues would cause more seeding of tumor cells in the preferred tissues than
Fibronectin and Its lntegrin Receptors
15
would be the case otherwise. Given the ability of adhesion receptors to elicit cellular signals, cell adhesion could be contributing to a favorable “soil” as well. It has been shown that the metastatic spread of tumor cells can be directed to a predetermined site by an adhesion molecule. Using transgenic mice expressing the leukocyte adhesion molecule E-selectin in their vasculature has shown an altered pattern of metastatic distribution of cells that express the selectin ligand Le” (Biancone et al., 1996).The selectin system is one that leukocytes use to find inflammatory sites, at which they exit the circulation to enter tissues. Not surprisingly, lymphomas show preferential tissue homing that depends on adhesion molecules that are expressed specifically in certain lymphoid organs and that mediate leukocyte homing into these tissues (Weissman, 1994).Among the fibronectin receptors, the a4pl integrin has been shown to redirect metastasis into the bone marrow. Intravenously injected Chinese hamster ovary cells formed colonies in the lungs of nude mice, whereas forced expression of a4pl in these same cells caused them to metastasize to bones in addition to the lungs (Matsuura et al., 1996).Antibody inhibition experiments showed that the ligand for a4pl in the bone marrow is VCAM-1, rather than fibronectin. Other studies have shown that tumor cells bind preferentially to endothelial cells from their preferred tissue site of metastasis (Auerbach et al., 1991),suggesting that endothelial specificities are likely to direct natural tumor homing as well. A peptidase and an ion channellike receptor have been identified as tumor-homing receptors in the lungs (Johnson et al., 1993;Elble et al., 1997),but little else is known about such receptors. We have developed a new method for studying specific features of the vasculature in different tissues. The method is in vivo screening of peptide libraries for peptides that direct the phage to home to a selected tissue. We looked for and found phage that bound preferentially to blood vessels in mouse brain and kidney (Pasqualini and Ruoslahti, 1996).Additionally, we have identified specific homing peptides for a number of other organs and tissues (Rajotte et al., 1998).The extensive diversity of endothelia revealed by these studies has led us to postulate that most, perhaps all, tissues display individual markers in their vasculature. The nature of these vascular “addresses” is obviously of great interest, and studies are underway to use the homing peptides to isolate the relevant target molecules. Endothelial cell vessels that are undergoing angiogenesis differ from endothelial cells in resting blood vessels in that they express a number of proteins that are not expressed at detectable levels in established blood vessels (Brooks, 1994;Martiny-Baron and Marmy, 1995).We have also used the phage technique to find peptides capable of homing into tumor blood vessels and have identified several peptide motifs that selectively direct phage into tumors (Arap et al., 1998;W. Arap, R. Pasqualini, and E. Ruoslahti, un-
16
Erkki Ruoslahti
published results). One such peptide binds selectively to the QV integrins avp3 and avp5 (Koivunen et al., 1995; Pasqualini et al., 1997; Arap et al., 1998). The avp3 integrin (and avp5) is a known marker of angiogenic vessels (Varner and Cheresh, 1996; Brooks, 1994). Originally discovered as an RGD-directed receptor for vitronectin (Pytela et al., 1985), avP3 is now known to bind to a number of extracellular matrix proteins including, under some circumstances, fibronectin (Ruoslahti, 1996). The avP3 integrin is not only a marker of angiogenic vessels, it is functionally important in the angiogenic process; growing endothelial cells require avp3 for survival. Antibodies and soluble peptides capable of inhibiting the binding of avp3 to its extracellular matrix ligands cause endothelial cell apoptosis and destruction of the neovasculature (Brooks, 1995; Chen et al., 1997). We have used peptides that bind to a v integrins to show that phages carrying these peptides home into tumors in a highly selective manner (Arap et al., 1998). More recently, tumor imaging was accomplished by directing liposomes to tumor blood vessels with the help of anti-avp3 (Sipkins et al., 1998).These results indicate that the QV integrins are present on the luminal surface of the tumor vessels and that they, therefore, can be used to target drugs, cells, liposomes, and other therapeutic devices into tumors. In summary, studies on integrin-mediated cell attachment to extracellular matrices have greatly increased our understanding of central cell biological phenomena, such as anchorage dependence, and have generated a number of possible approaches to new therapies of cancer.
ACKNOWLEDGMENTS The author’s work is supported by the following grants from the National Institutes of Health, Department of Health and Human Services: CA28896, CA62042, CA67224, and Cancer Center Support Grant CA30199.
REFERENCES Adams, S., Sobel, M. E., Howard, B. H., Olden, K., Yamada, K. M., de Crombrugghe, B., and Pastan, I. (1977). Proc. Natl. Acud. Scr. U.S.A. 74,3399-3403. Akamatsu, K., Ichihara-Tanaka, K., Ozono, K., Kamiike, W.,Matsuda, H., and Sekiguchi, K. (1996). Cancer Res. 56,4541-4546. Albelda, S. M., Mette, S. A., Elder, D. E., Stewart, R., Damjanovich, L., Herlyn, M., and Buck, C. A. (1990). Cancer Res. 50,6757-6764. Arap, W., Pasqualini, R., and Ruoslahti, E. (1998). Science 279,377-380. Assoian, R. K., and Marcantonio, E. E. (1996).J. Clin. Invest. 98,2436-2439.
Fibronectin and Its lntegrin Receptors
17
Auerbach, R., Lu, W. C., Pardon, E., Gumkowski, F., Kaminska, G., and Kaminski, M. (1991). Cancer Res. 51,394-399. Bartfeld, N. S., Pasquale, E. B., Geltosky, J. E., and Languino, L. R. (1993).J.Biol. Cbem. 268, 17270-17276. Biancone, L., Araki, M., Araki, K., Vassalli, P., and Stamenkovic, I. (1996).J. Exp. Med. 183, 58 1-587. Birchmeier, W. (1995). Biofisays 17,97-99. Blobel, C. P. (1997). Cell 90,589-592. Boudreau, N., Sympson, C. J., Werb, Z., and Bissell, M. J. (1995).Science 267, 891-893. Bourdoulous, S., Orend, G., MacKenna, D. A., Pasqualini, R., and Ruoslahti, E. (1998).J.Cell Biol. 143,267-276. Brooks, P. C. (1994).Cell 79, 1157-1164. Brooks, P. C. (1995).]. Clin. Invest. 96, 1815-1822. Brown, L. F., Asch, B., Harvey, V. S., Buchinski, B., and Dvorak, H. F. (1988).Cancer Res. 48, 1920-1925. Brunton, V. G., Ozanne, B. W., Paraskeva, C., and Frame, M. C. (1997). Oncogene 14, 283-293. Burridge, K., and Chrzanowska-Wodnicka,M. (1996).Annu. Rev. Cell Dev. Biol. 12,463-519. Chan, B. M., and Hemler, M. E. (1993).J. Cell Biol. 120,537-543. Chen, C. S., Mrksich, M., Huang, S., Whitesides, G. M., and Igber; D. E. (1997).Science 276, 1425-1428. Chen, H.-C., and Guan, J.-L. (1994). Proc. Natl. Acad. Sci. U.S.A. 91,10148-10152. Clark, E. A., and Brugge, J. S. (1995).Science 268,233-239. Elble, R. C., Widom, J., Gruber, A. D., Abdel-Ghany, M., Levine, R., Goodwin, A., Cheng, H. C., and Pauli, B. U. (1997).J. Biol. Cbem. 272,27853-27861. Fang, E, Orend, G., Watanabe, N., Hunter; T., and Ruoslahti, E. (1996).Science 271,499-502. Folkman, J. (1978).Nature (London) 273,345-349. Folkman, J. (1995).Nature Med. 1,27-31. Frisch, S . M., and Francis, H. (1994).J. Cell Biol. 124, 619-626. Frisch, S. M., and Ruoslahti, E. (1997). Curr. Opin. Cell Biol. 9, 701-706. Frisch, S. M., Vuori, K., Ruoslahti, E., and Chan-Hui, P. Y. (1996).J. Cell Biol. 124, 619626. Gailit, J., and Clark, R. A. F. (1994). Curr. Opin. Cell Biol. 6,717-725. Gailit, J., Welch, M. P., and Clark, R. A. (1994).J. Invest. Dermatol. 103,221-227. Gehlsen, K. R., Argraves, W. S., Pierschbacher, M. D., and Ruoslahti, E. (1988).J. Cell Biol. 106,925-930. Giancotti, F. G., and Ruoslahti, E. (1990). Cell 60, 849-859. Gross, H.-J., Verwer, B., Houck, D., Hoffman, R. A., and Recktenwald, D. (1995).Proc. Natl. Acad. Sci. U.S.A. 92,537-541. Hannigan, G. E., Leung-Hagesteijn, C., Fitz-Gibbon, L., Coppolino, M. G., Radeva, G., Filmus, J., Bell, J. C., and Dedhar, S. (1996).Nature (London) 379,91-96. Hardan, I., Weiss, L., Hershkoviz, R., Greenspoon, N., Alon, R., Cahalon, L., Reich, S., Slavin, S., and Lider, 0. (1993).Int. J. Cancer 55,1023-1028. Hardingham, J. E., Kotasek, D., Farmer, B., Butler, R. N., Mi, J.-X., Sage, R. E., and Dobrovic, A. (1993).Cancer Res. 53,3455-3458. Hoon, D. S., Wang, Y., Conrad, A. J., Schmid, P., Garison, D., Kuo, C., Foshag, L. J., Nizze, A. J., and Morton, D. L. (1995).J.Clin. Oncol. 13,2109-2116. Hughes, P. E., Renshaw, M. W., Pfaff, M., Forsyth, J., Keivens, V. M., Schwartz, M. A., and Ginsberg, M. H. (1997). Cell 88,521-530. Humphries, M. J., Olden, K., and Yamada, K. M. (1986).Science 233,467-470. Humphries, M. J., Yamada, K. Y., and Olden, K. (1988).J. Clin. Invest. 81,782-790.
18
Erkki Ruoslahti
Hungerford, J. E., Compton, AM.T., Matter, M. L., Hoffstrom, B. G., and Otey, C. A. (1996). J . Cell Biol. 135, 1383-1390. Huttenlocher, A., Lakonishok, M., Kinder, iM.,Wu, S., Truong, T., Knudsen, K. A., and Horwitz, A. F. (1998).J. Cell Biol. 141, 515-526. Hynes, R. 0. (1992). Cell 69, 11-25. Jaakkola, S., Vornanen, T., Leinonen, J., Rannikko, S., and Stenman, U.-H. (1995).Clin. Chem. 41,182-186. Johnson, R. C., Zhu. D., Augustin-Voss, H. G., and Pauli, B. U. (1993).I. Cell Biol. 121, 1423-1432. Jones, J. I., Prevette, T., Gockerman, A., and Clemmons, D. R. (1996). Proc. Natl. Acad. Sci. U.S.A. 93,2482-2487. Juliano, R. L., and Haskill, 5 . (1993).]. Cell Biol. 120, 5777-585. Kahn, P., and Shin, S. I. (1979).1. Cell Biol. 82, 1-16. Kanner, S. EL, Reynolds, A. B., Vines, R. R., and Parsons, J. T. (1990). Proc. Nutl. Acud. Sci. U.S.A. 87, 3328-3332. Koivunen, E., Wang, B., and Ruoslahti, E. (1995).Bmtechnology 13, 265-270. Kolanus, W., Nagel, W., Schiller, B., Zeitlmann, L., Godar, S., Stockinger, H., and Seed, B. ( 1 996). Cell 86,233-247. Lebowitz, P. F., Sakamuro, D., and Prendergast, G. C. (1997). Cancer Res. 57, 708-713. Marte, B. M., Rodriguez-Vicuna, P., Wennstrom, S., Warne, P. H., and Downward, J. (1997). Cum Biol. 7, 63-70. Martiny-Bacon, G., and Marme, D. (1995). Curr. Opirr. BiotechnoL 6,675-680. Massag&, J. (1987).Cell 49, 437-438. Matsuda, M., Mayer, B. J., Fukui, Y., and Hanafusa, H. (1990).Science 248, 1537-1539. Matsuura, N., Puzon-McLaughlin, W., hie, A,, Morikawa, Y., Kakudo, K., and Takada, Y. (1996).Am. J. Puthol. 148, 55-61. Meredith, J. E., Fazeli, B., and Schwartz, M. A. (1993).Mol. Biol. Cell 4, 953-961. Montgomery, A. hi. P., Reisfeld, R. .4.,and Cheresh,D. A. (1994).Proc. Nutl. Acud. Sci. U.S.A. 91,8856-8860. Morla, A., Zhang, Z., and Ruoslahti, E. (1994).Nature (London) 367, 193-198. Natali, P. G., Hamby, C. V., Felding-Habermann, B., Liang, B., Nicotra, M. R., Di Filippo, F., Giannarelli, D., Temponi, M., and Ferrone, S. (1997). Cancer Res. 57, 1554-1560. Nierodzik, M. I.. R., Kajumo, F., and Karpatkin, S. (1992). Cancer Res. 52, 3267-3272. Nievers, M. G., Birge, R. B., Greulich, H., Verkleij, A. J., Hanafusa, H., and van Bergen en Henegouwen, 1’. M. P. (1997).1. Cell Sci. 110,389-399. Novak, A., Hsu, S.-C., Leung-Hagestein, C., Radeva, G., Papkoff, J., Montesano, R., Roskelley, C., Grosschedl, R., and Dedhar, S. (1998).Proc. Nutl. Acud. Sci. U.S.A. 95,4374-4379. O’Brien, V.. Frisch, S. M., and Juliano, R. L. (1996).Exp. Cell Res. 224,208-213. Orend, G., Hunter, T., and Ruoslahti, E. (1998). Oncogene 16, 2575-2583. Parsons, J. T. (1996). Cum Opin. Cell Biol. 8, 146-152. Pasqualini, R., and Ruoslahti, E. (1996). Nuture (London) 380, 364-366. Pasqualini, R., Bourdoulous, S., Koivunen, E., Woods, Jr., V. I., and Ruoslahti, E. (1996).Nuture Med. 2, 793-799. Pasqualini, R., Koivunen, E., and Ruoslahti, E. (1997).Nuture Biotechnol. 15,542-546. PBllanen, J., Saksela, O., Salonen, E.-M., Andreasen, P., Nielsen, L., Dan@,K., and Vaheri, A. (1987).I . Cell Biol. 104, 1085-1096. Poltc, T. R., and Hanks, S. K. (1997). 1. Biol. Chem. 272, 5501-5509. Pyetela, R., Pierschbacher, M. D., and Ruoslahti, E. (1985). Cell 40, 191-198. Kajotte, D., Arap, W., Hagedorn, IM., Koivunen, E., Pasqualini, R., and Ruoslahti, E. (1998). J . Clm. Invest. (in press).
Fibronectin and Its Integrin Receptors
19
Re, F., Zanetti, A., Sironi, M., Polentarutti, N., Lanfrancone, L., Dejana, E., and Colotta, F. (1994).J. Cell Biol. 127, 537-546. Ruoslahti, E. (1988).Annu. Rev. Biochem. 57, 375-413. Ruoslahti, E. (1996).Annu. Rev. Cell Dev. Biol. 12, 697-715. Saba, T. M., and Cho, E. (1977).J. Reticuloendothel. SOL. 22, 583-596. Sauliner, R., Bhardwaj, B., Klassen, J., Leopold, D., Rahimi, N., Tremblay, E., Mosher, D., and Elliott, B. (1996).Exp. Cell Res. 222, 360-369. Scatena, M., Almeida, M., Chaisson, M. L., Fausto, N., Nicosia, R. E, and Giachelli, C. M. (1998).J. Cell Biol. 141, 1083-1093. Schlaepfer, D. D., and Hunter, T. (1996).Mol. Cell. Biol. 16,5623-5633. Schlaepfer, D. D., Hanks, S. K., Hunter, T., and Van der Geer, P. (1994).Nature (London) 372, 786-79 1. Schneller, M., Vuori, K., and Ruoslahti, E. (1997).EMBO J. 16, 5600-5607. Schreiner, C., Fisher, M., Hussein, S., and Juliano, R. L. (1991). Cancer Res. 51, 1738-1740. Schwartz, M. A., Schaller, M. D., Ginsberg, M. H. (1995). Annu. Rev. Cell Dev. Biol. 11, 549-99. Seftor, R. E. B., Seftor, E. A., Gehlsen, K. R., Stetler-Stevenson,W. G., Brown, P. D., Ruoslahti, E., and Hendrix, M. J. C. (1992).Proc. Natl. Acud. Sci. U.S.A. 89,1557-1561. Sheppard, D. (1996).BioEssays 18,655-660. Shulze, A,, Zerfass-Thome, K., Berges, J., Middendorp, S., Jansen-Diirr P., and Henglein, B. (1996).Mol. Cell. Biol. 16,4632-4638. Sipkins, D. A., Cheresh, D. A., Kazemi, M. R., Nevin, L. M., Bednarski, M. D., and Li, K. C. P. (1998).Nature Med. 4,623-626. Springer, T. A. (1994). Cell 76,301-314. Stetler-Stevenson,W. G., Aznavoorian, S., and Liotta, L. A. (1993). Annu. Rev. Cell Biol. 9, 541-573. Tarin, D., Price,J. E., Kettlewell, M. G., Souter, R. G., Vass, A. C., and Crossley, B. (1984).Cuncer Res. 48,3584-3592. Taverna, D., Ullman-Culleri, M., Rayburn, H., Bronson, R. T., and Hynes, R. 0. (1998).Cuncer Res. 58, 848-853. Varner, J. A., and Cheresh, D. A. (1996).Curr. Opin. Cell Biol. 8, 724-730. Varner, J. A., Emerson, D. A., and Juliano, R. L. (1995).Mol. Biol. Cell 6, 725-740. Vuori, K., and Ruoslahti, E. (1994).Science 266,576-1578. Vuori, K., Hirai, H., Aizawa, S., and Ruoslahti, E. (1996).Mol. Cell. Biol. 16,2606-2613. Wary, K., Mainiero, F., Isakoff, S., Marcantonio, E., and Giancotti, F. (1996). Cell 87, 733743. Weaver, V. M., Petersen, 0. W., Wang, F., Larabell, C. A., Briand, P., Darnsky, C., and Bissell, M. J. (1997).J. Cell Biol. 137,231-245. Weissman, I. L. (1994). Cell 67,207-216. Werb, Z.,Tremble, P. M., Behrendtsen, O., Crowley, E., and Damsky, C . H. (1989).J.Cell Biol. 109,877-889. Wu, C., Hughes, P. E., Ginsberg, M. H., and McDonald, J. A. (1996).Cell Adhesion Commun. 4,149-158. Zambruno, G., Marchisio, P. C., Marconi, A., Vaschieri, C., Melchiori, A., Giannetti, A., and DeLuca, M. (1995).J. Cell Biol. 129, 853-865. Zhang, Q., Checovich, W. J., Peters, D. M., Albrecht, R. M., and Mosher, D. F. (1994).J. Cell Biol. 127, 1447-1459. Zhang, Z., Vuori, K., Reed, J. C., and Ruoslahti, E. (1995).Proc. Nutl. Acud. Sci. U.S.A. 92, 6161-6165. Zhang, Z . , Vuori, K., Wang, H.-G., Reed, J. C., and Ruoslahti, E. (1996).Cell 85, 61-69.
20
Erkki Ruoslahti
Zhong, C., Chrzanowska-Wodnicka, M., Brown, J., Shaub, A., Belkin, A. M., and Burridge, K. (1998).1. Cell Biol. 141, 539-551. Zhu, X., Ohtsubo, M., Bohmer, R. M., Roberts, J. M., and Assoian, R. K. (1996).J. Cell Biol. 133,391403. Zou, Z., Anisowicz, A., Hendrix, M. J., Thor, A., Neveu, M., Sheng, S., Rafidi, K., Seftor, E., and Sager, R. (1994). Science 263,526-529.
Myb and Oncogenesis Brigitte Canter and loseph S. Upsick Department of Pathology Stanford University School o f Medicine Stanford, California 94305
I. Introduction 11. The Myb Genes A. Discovery of v-Myb and c-Myb B. Oncogenic Activation of c-Myb C. Expression and Function of c-Myb during Development and Differentiation
D. c-My6 in Cell Growth and Cell Death E. Regulation of c-Myb Expression F. Myb-Related Genes 111. Structural and Functional Features of the Myb Proteins A. The Myb Repeat B. The Myb DNA-Binding Domain C. Transcriptional Activation Domain and Heptad Leucine Repeat D. Regulation by the Carboxyl Terminus IV. Regulation of v-Myb and c-Myb A. Sites of Phosphorylation B. Interactions with Other Proteins V. Transcriptional Regulation by v-Myb and c-Myb A. Genes Activated by v-Myb and c-Myb B. Transcriptional Repression by Myb Proteins VI. The Myb-Chromatin Connection References
I. INTRODUCTION This review will focus on v-My6 and its normal cellular counterpart c-My6 with an emphasis on the biological and biochemical functions of their protein products. Both the v-Myb and c-Myb proteins are nuclear, bind directly to DNA, have short half-lives, and can regulate gene expression. As various eukaryotic genome projects proceed, new members of the My6 gene family are discovered at an increasingly rapid rate, particularly in plants. In addition, genetic and biochemical analyses of differentiation, transcriptional regulation, chromosome function, and the cell cycle have identified additional My6-related genes. However, outside of the signature 50-amino acid Myb repeats that constitute its DNA-binding domain, the c-Myb protein has Advances in CANCER RESEARCH 0065-23OW99$30.00
Copyright 8 1999 by Academic Press. All rights of reproduction in any form reserved.
22
Brigitte Canter and Joseph S. Lipsick
significant homology only with two other proteins of vertebrates, A-Myb and B-Myb. A single closely related protein is present in the sea urchin and in the fruit fly. In this review we draw upon our current knowledge of Mybrelated genes and proteins insofar as they inform us about v-Myb and c-Myb. Other reviews about v-Myb and c-Myb are also available (Graf, 1992; Introna et ai., 1994; Ness, 1996; Shen-Ong, 1990; Thompson and Ramsay, 1995; Wolff, 1996). More information about the evolution of the Myb repeats and gene family can be found elsewhere (Lipsick, 1996), as can more detailed reviews of the vertebrate B-Myb gene (Saville and Watson, 1998) and the Myb-related genes of plants (Martin and Paz-Ares, 1997).
11. THE My6
GENES
A. Discovery of v-My6 and c-My6 The first member of the Myb gene family to be identified was the v-Myb oncogene of the avian myeloblastosis virus (AMV), which causes rapidly fatal monoblastic leukemia in chickens (Baluda and Reddy, 1994). This acutely transforming retrovirus is unusual in that it causes only leukemias in vivo and transforms only cells of the monocyte/macrophage lineage in culture. Indeed, the first evidence for oncogene specificity came from the observation that AMV could transform macrophages but not fibroblasts in culture, whereas Rous sarcoma virus did the converse, although both viruses could replicate in both cell types (Baluda, 1963; Durban and Boettiger, 1981). As is the case for other acutely transforming retroviruses, AMV arose via two recombinations between a replication-competent retrovirus and sequences of cellular origin (Klempnauer et al., 1982; Rushlow et al., 1982a). In particular, the enu gene of the myeloblastosis-associated virus type 1 (MAV-1) has been replaced by the v-Myb oncogene in AMV (Perbal et al., 1985). The 5' transduction event appears likely to have occurred at the DNA level because AMV has retained a portion of a c-Myb intron. The remainder of vMyb is an intronless cDNA copy of seven internal exons of c-Myb, the last of which is incomplete. As a result of this transduction process, v-Myb is expressed as a spliced subgenomic viral mRNA. Conditional alleles have been used to demonstrate that v-Myb is required for both the initiation and the maintenance of the transformed phenotype (Engelke et al., 1997; Moscovici and Moscovici, 1983). The first six amino acids of AMV v-Myb are encoded by the gag sequences upstream of the viral splice donor site. The last eleven amino acids of v-Myb are encoded by the 3' end of the env gene. However, neither the gag- nor the env-encoded amino acids are required for oncogenic transformation (Tbanez
Myb and Oncogenesis
23
and Lipsick, 1988; Lipsick and Ibanez, 1987). The MAVIAMV viruses have an unusual U3 region in their long terminal repeats (LTRs) and the MAV virus has a propensity to cause nephroblastomas and osteopetrosis in chickens, unlike other Rous-related helper viruses that generally cause longlatency lymphomas (Rushlow et al., 1982b). However, these unusual LTR sequences are not absolutely required for transformation by v-Myb in cell culture, even though there appears to be strong selection for them when AMV is passaged in chickens (Engelke and Lipsick, 1994). The 48-kDa v-Myb protein is truncated relative to the normal 75-kDa cMyb protein at both its amino and carboxyl termini (Figs. 1 and 2 ) (Gerondakis and Bishop, 1986; Rosson and Reddy, 1986). In addition, v-Myb contains 10 amino acid substitutions relative to the homologous region of c-Myb (a previously described eleventh substitution was not identified in c-Myb cDNA sequences). A second acutely transforming avian leukemia virus, E26, causes a rapidly fatal erythroblastosis in vivo and transforms multipotent hematopoietic progenitors in culture (Graf et al., 1992; Moscovici et al., 1983).E26 encodes a tripartite 135-kDa Gag-Myb-Ets fusion protein (Leprince et al., 1983; Nunn et al., 1983). c-Myb and the c-Ets protooncogene are present on different chicken chromosomes and it remains unclear whether E26 arose by sequential viral transduction of two different cellular genes or by transduction of a preexisting chromosomal translocation (Symonds et al., 1986). The Myb segment encoded by the E26 virus has even larger truncations than v-Myb of AMV and also contains one amino acid substitution that is different than any of those present in v-Myb of AMV (Fig. 2) (Nunn et al., 1984). The shorter Myb fragment present in E26 is only weakly transforming in culture without Ets and can be complemented by a variety of other oncogenes, including tyrosine kinases that result in autocrine growth factor production (Metz et al., 1991). In contrast, the Myb and Ets open reading frames cannot complement one another well in trans and there is a strong selection for the production of Myb-Ets fusion proteins in vivo from viruses that encode these proteins separately (Metz and Graf, 1991).
B. Oncogenic Activation of c-My6 Truncation of the c-Myb protein is required for the efficient transformation of myelomonocytic cells in culture (Gonda et al., 1989; Grasser et al., 1991). In general, amino-terminal truncations are more effective than carboxy-terminal truncations in this regard (Dini and Lipsick, 1993). Interestingly, although the amino acid substitutions in v-Myb are not required for oncogenic transformation, they strongly influence the phenotype of the transformed cells (Dini et al., 1995; Introna et al., 1990; Stober-Grasser and Lipsick, 1988). Cells transformed by AMV resemble monoblasts that are
-regulation
-
acidic
n
C-Myb
DNA binding
activation
Myb repeats I R l I R Z I R 3 1
I I
n
1
I
acidic
n I
llUl
I
negative regulation
~
n H .I
1 01
4
exon9A L
L
n n R 1
~ 1
I
m
u
)
1
E
m
D
A-Myb B-Myb
Urch Myb Dros Myb
Fig. 1 Topography of Myb proteins. The chicken c-Myb, A-Myb, and B-Myb proteins, the sea urchin Myb protein, and the Drosophilu Myb protein were aligned using the MACAW program (Schuler et ul., 1991). Exon YA was not included in the c-Myb sequence. Boxes indicate regions of statistically significant similarity. Shading indicates the degree of similarity within each box based on mean scores. v-Myb indicates the portions of c-Myb that are present in the oncoprotein encoded by AMV. R1, R2, and R3 indicate the three Myb repeats. HLR indicates a heptad leucine repeat in c-Myb that has also been referred to as the “leucine zipper.” See the text for a more detailed discussion.
Myb and Oncogenesis
25
committed to differentiate into macrophages. Indeed, either phorbol esters or liganded retinoic acid receptor-a can promote the differentiation of AMVtransformed monoblasts despite the continued presence of v-Myb (Pessano et al., 1979; Smarda et al., 1995; Symonds et al., 1984). Amino acid substitutions within both the DNA-binding domain and the transcriptional activation domain of AMV v-Myb are required for the monoblast phenotype (Dini et al., 1995; Introna et al., 1990). Cells transformed by a variant of vMyb that lacks all of these substitutions have a myelomonoblastic phenotype and bear cell surface markers of both the granulocytic and monocyte/macrophage lineage. In addition, the reversion of single amino acid substitutions within AMV v-Myb can result in the transformation of cells with a promyelocytic phenotype when grown in the presence of chicken myeloid growth factor (cMGF), a chicken cytokine related to mammalian granulocyte colony stimulating factor (G-CSF)(Introna et d., 1990). It has been shown that constitutive expression of full-length c-Myb can also cause transformation of bipotential myelomonocytic cells in culture (Ferrao et al., 1995; Fu and Lipsick, 1997). However, the cells grow more slowly and differentiate more frequently than do cells transformed by truncated c-Myb proteins. On the other hand, protein truncation is not required for the transformation of chick neural retinal cells by c-Myb (Garrido et al., 1992). Although endogenous c-Myb is not expressed in hematopoietic cells transformed by v-Myb or truncated forms of c-Myb, additional experiments have shown that the presence of c-Myb is compatible with transformation by v-Myb and that v-Myb does not repress the endogenous c-Myb gene (Lipsick, 1987; Smarda and Lipsick, 1994). These results suggest that v-Myb transforms hematopoietic cells that have turned off c-Myb expression as part of their normal differentiation process, but that the absence of c-Myb expression is not required for transformation. The c-Myb protooncogene has been further implicated in cancer as a result of retroviral insertional mutagenesis in two well-characterized experimental systems. In the presence of an inflammatory response, the replicationcompetent, oncogene-deficient Moloney murine leukemia virus (MuLV) causes myelomonocytic tumors rather than the more typical thymic lymphomas (Wolff et al., 1988).In virtually all of these myelomonocytic tumors (formerly known as plasmacytoid lymphosarcomas), there is an insertion of the retrovirus within the endogenous c-My6 locus (Shen-Ong et al., 1984, 1986). Most of these insertions result in amino-terminal or, less frequently, in carboxy-terminal truncations of the c-Myb protein that are very similar to those found in the AMV and E26 oncoproteins. However, some tumors contain insertions that are predicted to result in the deletion of only 38 residues from the carboxyl terminus of c-Myb (Fig. 2) (Nazarov and Wolff, 1995). In a second experimental system, the injection of a replication-competent,
26
Brigitte Canter and Joseph S. Lipsick
oncogene-deficient avian leukosis virus into day 1 2 chicken embryos causes nonbursal B cell lymphomas with a remarkably short latency (Kanter et al., 1988; Pizer and Humphries, 1989). In these tumors there is invariably a retroviral insertion in the endogenous c-My6 locus that results in a much smaller amino-terminal truncation than that found in AMV, E26, or the MuLV-induced tumors (Fig. 2). Interestingly, when a recombinant retrovirus expressing a cDNA with this 20-residue amino-terminal truncation was injected into chickens, sarcomas and carcinomas as well as lymphomas were induced (Jiang et al., 1997).Detailed analyses of tumor progression strongly suggest that the insertional activation of c-Myb alone is not sufficient for tumor formation in either the chicken or mouse model system (Belli et al., 1995; Pizer et al., 1992). Other experiments have also implicated c-My6 in oncogenesis. An avian retrovirus containing a carboxy-terminal truncation of the c-Myb protein was reported to cause muscular fibrosarcomas in chickens (Press et al., 1994).The resulting tumors were monoclonal and of rather long latency, unlike most tumors caused by acutely transforming retroviruses. Retroviral insertion into c-My6 was also found in a Marek’s disease chicken T lymphoma cell line (Le Rouzic and Perbal, 1996). This is of historical interest because
Fig. 2 Alignment of Myb protein sequences. Myb protein sequences were aligned using the CIIJSTALW implementation of the Parsimony After Progressive Alignment (PAPA) method (Feng and Doolittle, 1990; Thompson et al., 1994),with some additional adjustments based on blocks identified with the MACAW local alignment program (Schuler et ul., 1991). Shading was then performed using the BOXSHADE program. Black shading indicates identical residues at a given position; gray shading indicates similarity. Hu, Human; Ch, chicken; Xe, Xenopus; urchin, sea urchin; Dros, Drosophilu. Exon 9A-encoded sequences were included for human and chicken c-Myb, but not Xenopus c-Myb. The XX near the carboxyl terminus of Xenopus B-Myb indicates an additional 62 residues that were unaligned and were not included in the numbering scheme in order to conserve space. CKII, Casein kinase I1 phosphorylation site; FL, amino-terminal truncation of the murine FL variant of c-Myb; ALV, amino-terminal truncation caused by avian leukosis virus insertion in B cell lymphomas; R1, R2, and R3 indicate the Myb repeats; AMV, amino- and carboxy-terminal truncations of v-Myb of AMY E26, amino- and carboxy-terminal truncations of v-Myb of E26 leukemia virus; cys, an essential cysteine required for transcriptional activation and oncogenic transformation by AMV v-Myb that has been proposed to be subject to redox regulation; GSKIII, a peptide phosphorylated in v-Myb in vivo and by glycogen synthase kinase Ill in vitro; shaded hexagons, two heptad hydrophobic repeats, the darker of which has been referred to as the “leucine zipper”; exon 9A, residues encoded by an alternatively spliced exon in c-Myb; 1120 and 1151, carboxy-terminal truncations in mutants of AMV v-Myb; asterisks, highly conserved putative phosphorylation sites for proline-directed kinases; PS and BN, deletions in c-Myb that together activate transcription and oncogenic transformation; EVES, sequence that mediates interaction with the DNA-binding domain in yeast two-hybrid assays and that also contains a serine that is phosphorylated in c-Myb in vivo and by MAPK in vitro; F-MuLV, carboxy-terminal truncations caused by Friend murine leukemia virus in some myelomonocytic tumors; Dm myb’, position of a glycine that is mutated to a serine in a temperature-sensitive mutant of Drosophila Myb.
27
Myb and Oncogenesis
..
CKlI Hu-c-Myb Ch-c-Myb Xe-0-Myb Ru-A-Myb Ch-A-Myb Xe-A-Myb Eu-B-Myb Ch-B-Myb Xe-B-Myb Urchin-Myb Droa-Myb
Eu-c-Myb Ch-c-Myb Xe-c-Myb Bu-A-Myb Ch-A-Myb Xe-A-Myb Hu-B-Yyb Ch-B-Myb Xe-B-Myb Urchin-Myb D ro a -Myb
Eu-c-Myb Ch-c- Myb Xe-c-Myb Hu-A-Myb Ch-A-Myb Xe-A-Myb Ru-B-Myb Ch-B-Myb Xo-B-Myb Urchin-Myb Dr o s-M yb
Hu-c-Myb Ch-c-Myb Xe-c-Myb Hu-A-Myb Ch-A-Myb Xe-A-Myb Hu-B-Myb Ch-B-Myh Xe-B-Myb Urchin-Myb Dros-Myb
Hu-c-Myb Ch-c-Myb Xe-c-Myb Hu-A-Myb Ch-A-Myb Xe-A-Myb Hu-B Myb C h-B- Myb Xa-B My b Urchin-Myb Dr o s-Myb
-
-
196 196 193 191 191 190 187 187 187 194 240
Fig. 2
FL ALV V
V
acidic
Brigitte Canter and JosephS. Lipsick
28
acidic
Siu-c-Myb Ch-c-Myb Xe-c-Myb Bu-A- Myb Ch-A-Myb Xe-A-Myb Bu-B-Myb Ch-E-Myb Xe -B- My b Urchin-Myb D r 0 B -M yb
Ku-c-Uyb Ch-c-Myb Xe-c-Myb Eu-A-Myb Ch A-Mvb XeIA-M%b Xu-B-Myb Ch-B-Myb Xa-B-Myb Urchin-Myb Droa-Myb
287 251 248 240 251 297 "leucine zipper" '2I' w-)
Bu-c-Myb Ch-c-Myb Xe-c-Myb Eu-A-Myb Ch-A-Myb Xe-A-Myb Bu-B-Uyb Ch-B-Myb Xe-B-Myb Urchin-Myb Drc8-Myb
374 375 362 340 345 338 292 284 273 282 329
Ku-c-Myb Ch-c-Myb Xe-c-Myb Bu-A-Mpb Ch-A-Myb Xe-A-Myb Bu-B-Myb Ch-B-Myb Xe-8-Myb Urchin-Myb Drc8-Myb
428 429 386 393 398 390 345 338 316 330 364
Ku-c-Myb Ch-c-Myb Xe-c-Myb HU-A-Myb Ch-A-Myb Xe-A-Myb EU-E-Myb Ch-B-Uyb Xa-E-Myb Urchin-Myb DroB-Myb
474 415
m'
w-)
+...................... "..........._.I ".I..
#
1151 V?..""
.......".............
.....................................................................
exon 9A..... .....-... ............... ............................................................. --..............-.............................. "
"
"
GR-ALQrQ-----QR1GNgTKPAGSPSPRVNK~-GT-AVQLQ----- EGGAS~LCRPPGLPISNLSKT~&--
I
~-LMRIQ-----ENLGAMECQFNVSLV Wt-LMRIQ-----ENrR1UrCQINVSVn
~-WI--------E-BISPDCALNSCLV ~QLQASEQQQVLPPRQPS~LVPSVTZYR ~~~-QT--------PSKPTPSLPNVTBYR TB-MV------TDKPQ&SN--VTBIR
~%-~T--------EMNTKQs----IDIR
-----________---------------
* *
*
... ....................".".......-............................ I
386
441
446 429 404 385 362 373 374
Fig. 2 (continued)
.*
SSLD-PPKV-L~PARES---SPP~-SPKS-LSASQGS----
29
Myb and Oncogenesis
* Hu-c-Myb Ch-c-Myb Xe-c-Myb Hu-A-Myb Ch-A-Myb Xa A-Myb Hu-H-Myb Ch-B-Myb Xe-8-Myb Urchan-Myb Dros Mvb
524 524 387 492 497 474 157
Hu-c-Myb Ch-c-Myb Xr-c-Myb Hu-A-Myb Ch-A-Myb Xe-A-Myb Hu-B-Myb Ch-B-Myb Xa-B-Myb Urchin-Myb Dros-Myb
583 583
iiu-c-Myb Ch-c-Myb Xe-c-Myb Hu-A- M y b Ch-A-Myb Xe-A-Myb Eu-B-Myb Ch-B-Myb Xa-8-Myb Urchin-Myb Dros-Myb
639 639 503 604 609 585 563 546 522 537 501
Bu-c-Myb C h-c -Myb Xa-c-Myb Hu-A-Myb Ch-A-Myb Xe-A-Myb Hu-B -Myb C h-B- M y b Xe-B-Myb Urchin-Myb Dros-Myb
718 718 583 705
440
416 433 394
447 551
556 532 514 197 473 483 443
*
710 680 652 638 632 642 596
Fig. 2 (continued)
*
AMV rPS
*
30
Brigitte Ganter and JosephS. Lipsick
AMV arose in a Marek’s disease virus-infected chicken (Baluda and Reddy, 1994). Expression of a v-My6 transgene under control of the T cell-specific CD2 promoter caused late-onset T cell lymphomas in laboratory mice (Badiani et al., 1996). These experiments together imply that additional genetic alterations are required for oncogenic transformation by activated forms of c-My6. The human c-My6 gene is located at 6q24 (Harper et al., 1983). Thus far, no consistent chromosomal translocations involving c-My6 have been identified in human leukemias or lymphomas. However, the level of c-My6 expression is quite high in most leukemias and lymphomas with an immature phenotype (Westin et al., 1982). In addition, the levels of c-My6 expression have been correlated with prognosis in human breast and colon carcinomas (Greco et al., 1994; Guerin et al., 1990; Torelli et al., 1987). Interestingly, the c-My6 gene is amplified in a number of human colon carcinoma, pancreatic carcinoma, but not leukemia cell lines (Alitalo et al., 1984; Henderson and Wolman, 1988; Wallrapp et al., 1997). In particular, it was recently found that the c-My6 gene is amplified in 10% of primary pancreatic carcinomas and that high levels of c-My6 expression are also observed in the majority of all pancreatic carcinomas (Wallrapp et al., 1997). It has also been suggested that mutations within the first intron of c-My6 may correlate with increased expression in human colon cancers by altering gene expression (Thompson et al., 1997).
C. Expression and Function of c-MyG during Development and Differentiation The normal c-My6 protooncogene is expressed at high levels in immature hematopoietic cells of all lineages, and its expression decreases as the cells differentiate (Chen, 1980; Duprey and Boettiger, 1985; Gonda and Metcalf, 1984; Kirsch et al., 1986; Ramsay et al., 1986; Westin et al., 1982). Constitutive expression of c-My6 can block the differentiation of various established erythroid and myeloid cell lines, suggesting that c-Myb may control hematopoietic differentiation (Clarke et al., 1988; Selvakumaran et al., 1992; Smarda and Lipsick, 1994; Todokoro et al., 1988). Consistent with these observations, laboratory mice with a homozygous mutation in c-My6 die in utero due to a failure of fetal liver hematopoiesis (Mucenski et al., 1991). In contrast, the earlier yolk sac hematopoiesis appears to be normal in these mice, as does the development of megakaryocytes within the fetal liver. Additional studies have suggested that c-Myb function is required for T lymphocyte development. In particular, a transgene that produces a fusion of the c-Myb DNA-binding domain and the Drosophila Engrailed repressor
Myb and Oncogenesis
31
domain can prevent normal thymocyte maturation (Badiani et al., 1994). Further studies have shown that this transgene causes apoptosis (Taylor et al., 1996). The expression of c-My6 has also been detected in immature epithelial cells in a variety of tissues, including the colon, respiratory tract, skin, and retina (Queva et al., 1992; Sitzmann et al., 1995).The role of c-Myb in the normal development of these tissues remains unclear at present. Transgenic mice bearing a c-Myb gene driven by the ubiquitously expressed p-actin promoter displayed no thymic abnormalities, but did develop degenerative abnormalities in skeletal and cardiac muscles (Furuta et al., 1993). In Xenopus, a c-Myb homolog is expressed throughout development and continues in adult tissues, with the highest levels in the intestine, heart, liver, lung, and ovary (Amaravadi and King, 1994).
D. e M y 6 in Cell Growth and Cell Death In addition to its role in regulating differentiation, c-Myb has also been implicated in control of the cell cycle. When resting lymphocytes are stimulated to divide, c-Myb mRNA and protein expression begin in the late GI phase of the cell cycle and continue into the S phase (Lipsick and Boyle, 1987; Torelli et al., 1985). Interestingly, cell lines that represent B lymphocytes at different stages of maturation appear to regulate c-My6 expression differently with respect to the cell cycle (Catron et al., 1992). In addition, the constitutive expression of c-Myb can rescue v-Myb-transformed cells from phorbol ester-induced differentiation and cell cycle arrest in both GI and G, (Smarda and Lipsick, 1994). Surprisingly, the DNA-binding domain of cMyb alone is sufficient for such rescue (Engelke et al., 1995a). Interestingly, experiments with a temperature-sensitive mutant of the E26 virus suggest that cells must traverse S phase for the Gag-Myb-Ets fusion protein to alter cell morphology after a shift to the permissive temperature (Beug et al., 1987). It has also been reported that c-Myb can permit fibroblasts to progress through the cell cycle in the absence of exogenous IGF-1, apparently by inducing endogenous IGF-1 (Reiss et al., 1991; Travali et al., 1991). Although c-Myb expression has been reported in normal fibroblasts, others have failed to reproduce this result (Catron et al., 1992; Thompson et al., 1986). Experiments with antisense oligonucleotides have suggested that c-Myb expression is required for the progression of hematopoietic cells into the S phase of the cell cycle (Gewirtz et al., 1989). However, other investigators have raised serious doubts about the specificity of this technique (Burgess et al., 1995). c-Myb has also been proposed to prevent apoptosis by directly regulating
32
Brigitte Canter and loseph S. Lipsick
the expression of the Bcl-2 gene (Frampton et al., 1996; Salomoni et al., 1997; Taylor et al., 1996). However, other investigators have found that Bcl2 expression persists after functional inactivation of a hormone-inducible form of the c-Myb protein (Hogg et al., 1997). Still other investigators have suggested that c-My6 promotes rather than inhibits apoptosis in neuronal cells (Estus et al., 1994).It therefore remains unclear whether a primary role of c-My6 is to prevent or accelerate apoptosis, or rather whether c-Myb regulates apoptosis indirectly simply by maintaining cells in the cycle in a fashion similar to the action of many extracellular growth factors.
E. Regulation of c-My6 Expression The regulation of c-My6 expression has remained rather enigmatic (Boise et al., 1992). In addition to regulation at the level of transcriptional initiation, evidence has been presented for regulation at the level of transcriptional elongation within the first intron, as had previously been proposed for the cMyc gene (Bender et al., 1987; Watson, 1988). However, in the case of cMyc the conclusions reached by analyzing nuclear run-on experiments have generally not been supported by more detailed analyses of mRNAs in whole cells (Krumm et al., 1992; Strobl and Eick, 1992). Rather than being regulated by attenuation at some distance from the promoter as initially thought, c-Myc now appears to be regulated by the release of RNA polymerase engaged near the start site in a fashion reminiscent of a proposed mechanism of action of the lac repressor of Escherichia coli (Lee and Goldfarb, 1991). Similar experiments have not yet been reported for c-Myb. The c-My6 promoter has no classical TATA motif and transcription appears to initiate at multiple sites within the promoter (Bender and Kuehl, 1986; Dvorak et al., 1989; Watson et al., 1987).An additional promoter has also been identified within the first intron of c-Myb (Jacobs et al., 1994).Although numerous proteins have been reported to bind to the c-My6 promoter and first intron, it remains unclear which, if any, of these proteins and binding sites is relevant in vivo (Calabretta and Nicolaides, 1992; McCann etal., 1995; Phan et al., 1996; Reddy and Reddy, 1989; Sullivan et al., 1997; Toth et al., 1995). For example, the c-Myb protein has been reported to either repress or activate its own promoter in transient assays (Guerra et al., 1995; Nicolaides et al., 1991). In this regard, an 8-kb fragment of chicken genomic DNA extending upstream from the first intron was unable to recapitulate the regulation of the endogenous c-Myb gene in a number of different cell lines, suggesting the presence of more widely dispersed regulatory elements (Y. Vaishnav and J. s. Lipsick, unpublished). The advent of homologous recombination in murine embryonic stem cells and various “knock-in” strategies should eventually help to answer these questions. A
Myb and Oncogenesis
33
region of particular interest is a 124-nucleotide motif within the c-My6 promoter that is 92% identical in chicken and mouse (Urbanek et al., 1988). Also, a consensus E2F-binding site is present within the c-My6 promoter and, by analogy with B-My6, may be required for cell cycle regulation (DeGregori et al., 1995; Mudryj et al., 1990).
F. My6-Related Genes In mammals, birds, and amphibians, two My6-related genes in addition to c-My6 (A-My6 and B-My6) have been identified (Nomura et al., 1988). The c-My6 and A-My6 genes appear to have arisen from a recent gene duplication event following an earlier duplication of a B-My6-like gene that is more closely related to the sole My6-related genes of sea urchin and Drosophilu (Fig. 3). Like c-My6, the expression of A-My6 appears to be rather tissue specific. The A-My6 gene is expressed in mammals at high levels in the developing central nervous system, in germinal center B lymphocytes, in mammary gland ductal epithelium, and in testis (Golay et al., 1998; Mettus et al., 1994; Trauth et al., 1994). A-Myb has also been reported to cooperate with c-Myc in driving smooth muscle cells into S phase (Marhamati et al., 1997). In Xenopus, A-My6 expression is high in mitotic spermatogonial cells, but ceases on meiosis (Sleeman, 1993). Laboratory mice with a homozygous mutation in the A-My6 gene are viable but display a failure of spermatogenesis and of mammary gland development in response to pregnancy (Toscani et al., 1997). The role of A-My6 in malignancies has not been extensively investigated, but mice that widely expressed A-My6 transgenes developed follicular hyperplasia of the spleen and lymph nodes due to a proliferation of B cells with a germinal center phenotype (DeRocco et al., 1997). The B-My6 gene is expressed throughout mouse development, unlike either A-My6 or c-My6 (Sitzmann et al., 1996).The expression of B-My6 appears to correlate with cell division during embryogenesis. When quiescent cultured fibroblasts enter the cell cycle, B-My6 mRNA is induced in late G, and early S phase (Lam et ul., 1992). The B-My6 promoter contains a binding site for the E2F transcription factor that negatively regulates gene expression during Go and early G, (Lam and Watson, 1993; Zwicker et al., 1996). Either p107 or p130 but not Rb is required for this negative regulation (Hurford et al., 1997). Other genes that are similarly regulated include cdc2, cyclin A, tbymidylate synthetase, ribonucleotide reductase, and E2 F1 . Interestingly,constitutive expression of B-My6 has been reported to bypass p53-induced G, arrest, even though p21 induction occurs (Lin et al., 1994). These results suggest that the regulation of B-My6 expression may be a critical checkpoint in G,. In addition, the peak expression of B-Myb in S phase
Brigitte Canter and loseph S. Lipsick
34
A-Myb
Fig. 3 Phyiogenetic tree of Myb protein sequences. Myb protein sequences were aligned using CLUSTALW and a boot-strapped tree was generated (Thompson et al., 1994). Hum, Human; Chi, chicken; Xen, Xenopus; urch, sea urchin; Dros, Drosophila; Dicty, Dictyostelium; Asper, Aspergillus nidulans; Scerv, Saccharomyces cerevisiae.
and its short half-life raise the possibility that the essential function of B-Myb may be transcriptional repression or even nontranscriptional. Experiments with antisense nucleic acids are consistent with the idea that B-Myb expression is essential for cell cycle progression (Arsura et ul., 1992; Sala and Calabretta, 1992). However, these studies are subject to the same caveats noted above for similar studies of c-Myb. No mutations of the B-Myb gene have yet been reported in vertebrates. There is a single Myb-related gene in Drosophilu that is more closely related to B-Myb than to A- or c-Myb (Figs. 2 and 3). In particular, the Drosophilu Myb protein lacks the central acidic region and the heptad leucine repeatIFAETL regions. Drosophilu Myb expression is seen throughout embryonic development and generally correlates with cell division, similar to B-Myb expression in vertebrates (Katzen et ul., 1985). However, Drosophilu Myb does not appear to be expressed in the larval tissues that undergo endoreduplication in which repeated S phases without M phases are
Myb and Oncogenesis
35
used to create polyploid nuclei. Temperature-sensitivemutants of Drosophila My6 result in lethality at several different points during development (Katzen and Bishop, 1996). Drosophila My6 may be required in the GJM transition because cell cycles that fail during late wing development at the nonpermissive temperature can be restored by ectopic expression of cdc2 or string (cdc25) (Katzen et ul., 1998). There appears to be only one sea urchin My6-related gene that is closely related to B-My6 (Figs. 2 and 3). It was discovered as a transcriptional repressor of a specialized actin gene during development (Coffman et al., 1997). Two My6-related genes have appeared in the Caenorhabditis elegans genome project database thus far. One, contained in cosmid D1081, is a homolog of Schizosaccharomycespombe cdc5, a gene that is required in the G, phase of the cell cycle (Ohi et al., 1994, 1998). Cdc5 homologs are also present in vertebrates, Drosophila, and green plants (Bernstein and Coughlin, 1997; Hirayama and Shinozaki, 1996; Ohi et al., 1998). The other My6related gene in C. eleguns, contained in cosmid F32H2, appears to be more closely related to c-Myb than to Cdc5. However, its DNA-binding domain is more evolutionarily distant from c-Myb than that of a Myb identified in the cellular slime mold Dictyostelium discoideum (Stober-Grasser et al., 1992; S. McCann and J. S. Lipsick, unpublished). These results suggest that another gene more closely related to c-My6 is likely to be present in the nematode C. elegans. Our analysis of the sequences and functional domains of the Myb proteins of animals suggests a model for Myb evolution. A single Myb gene most similar to Drosophila My6, sea urchin My6, and vertebrate B-My6 underwent a gene duplication event during the genesis of vertebrates. One copy of this gene was selected for retention of its function and is the modern-day B-Myb that is required in all cell types. The second copy of this gene drifted and acquired a central transcriptional activation domain that was then selected for a more specialized function in specific tissue(s). This second gene then underwent another duplication, giving rise to modern-day AMy6 and c-Myb. These two genes were then selected for a similar function in different specialized tissue(s) and have therefore both retained their central transcriptional activation domains. Such a model in which a gene has undergone two rounds of duplication and divergence during vertebrate evolution is consistent with observations in other regulatory gene families (Sidow, 1996).It will therefore be of interest to determine the repertoire of My6 genes in the “intermediate” species that have not yet been examined, including tunicates, Amphioxus, jawless fish, shark, and bony fish. Preliminary data suggest that tunicates have a single My6 gene whereas bony fish have three My6-related genes (E. Chen, S . McCann, and J. S. Lipsick, unpublished).
36
Brigitte Canter and loseph S. Lipsick
Ill. STRUCTURAL AND FUNCTIONAL FEATURES OF THE MYB PROTEINS The c-Myb protein is present in the cell nucleus and contains three main functional domains-an amino-terminal DNA-binding domain, a central transcriptional activation domain, and a carboxy-terminal regulatory domain (Fig. 1)(Sakura et al., 1989).
A. The
Myb Repeat
The family of Myb proteins is defined by the presence of a highly conserved Myb domain (repeat) of approximately 50 amino acids (Klempnauer and Sippel, 1987). Each repeat contains three conserved tryptophans, spaced 18 or 19 amino acids apart, that form a hydrophobic core and that play a critical role in sequence-specific DNA binding (Anton and Frampton, 1988; Kanei-Ishii et al., 1990; Saikumar et af., 1990).Each repeat folds into three well-defined helices that form a helix-turn-helix-related structural motif as was shown by nuclear magnetic resonance (NMR)spectroscopy for single cMyb repeats and for the c-Myb repeats R2R3 complexed with DNA (Frarnpton et af., 1991; Gabrielsen et a!., 1991; Ogata et al., 1992, 1994-1996). However, the b-Myb R2R3 are somewhat different (McIntosh et al., 1998).
B. The M y b DNA-Binding Domain The c-Myb DNA-binding domain (DBD) consists of three of tandem imperfect Myb repeats as described above ( R l , R2, and R3) (Gonda et al., 1985; Klempnauer and Sippel, 1987).The vertebrate A-Myb and B-Myb, the Drosophila Myb, and the Dictyostelium Myb proteins have conserved all three of these Myb repeats (Lipsick, 1996). During the genesis of v-Myb from c-Myb, deletion of part of the first repeat ( R l )has occurred (Gerondakis and Bishop, 1986; Rosson and Reddy, 1986) (Fig. 2). Such deletions correlate well with efficient oncogenic transformation of myelomonocytic cells (Dini and Lipsick, 1993). However, despite the lack of R1, v-Myb localizes to the cell nucleus and binds to DNA in a fashion similar to that of c-Myb. The minimal region of c-Myb that is both necessary and sufficient for sequencespecific DNA binding has been narrowed down to repeats R2 and R3 (Gabrielsen et af., 1991; Garcia et af., 1991; Howe et al., 1990; Oehler et al., 1990). Unlike many other sequence-specific DNA-binding proteins, the Myb proteins bind to DNA as monomers. Although repeat R1 was not shown to be involved in sequence-specificDNA recognition, it has been shown to in-
Myb and Oncogenesis
37
crease the affinity for DNA and the stability of the Myb-DNA complex (Dini and Lipsick, 1993; Ebneth etal., 1994b; Tanikawa etal., 1993).DNA binding has also been suggested to be regulated by the redox state of a highly conserved Cys at position 130 within repeat R2 (Guehmann et al., 1992; Myrset et al., 1993).In other experiments substitution of the homologous Cys by a Ser in v-Myb did not diminish DNA binding in vitro, but did abolish oncogenic transformation and reduce transcriptional activation in animal cells and in yeast (Chen and Lipsick, 1993; Grasser et al., 1992). Selection of a pool of chicken genomic DNA fragments with bacterially produced v-Myb led to the initial identification of the recognition sequence PyAAC(T/G)G(Myb recognition element, or MRE) (Biedenkappet al., 1988). Gratifyingly,three such sites were found 5’ of the transcriptional start site within the promoter of the E26-inducible mim-1 gene (Ness et al., 1989). Similar consensus sequenceswere later found by random oligonucleotideselection with either v-Myb or c-Myb proteins (Howe and Watson, 1991; Weston, 1992). Structural data imply that R3 and R2 cooperatively recognize the AAC core and the last G residue, respectively (Ogata et al., 1994). The third helix of each repeat is the recognition helix that makes specific DNA contacts within the major groove, although the precise mode of recognition by repeats R2 and R3 differs. The linker between R2 and R3 appears to be critical for DNA binding as well (Hegvold and Gabrielsen, 1996). Both v-Myb and c-Myb show greatly reduced affinity for DNA when the recognition sequence is methylated, as often occurs in inactive promoter regions (Klempnauer, 1993). Circular permutation assays have suggested that the Myb proteins bend DNA on binding (Saikumar et al., 1994). However, it remains controversial whether this technique is a valid measure of DNA bending (Hagerman, 1996; Sitlani and Crothers, 1998). A more extensive analysis of the MRE has revealed that the downstream flanking sequence is important for binding as well as for trans activation by the Myb proteins (B. Ganter, S . Chao, and J. S. Lipsick, unpublished). The v-Myb protein, but not the c-Myb protein, requires a stretch of six Ts (motif #2) downstream of the PyAACT/GG site (motif #1) for efficient in vitro binding to the strong mim-1 A site. The presence of repeat R1 within the cMyb DBD allows c-Myb to bind efficiently to the MRE without a downstream motif #2. This observation may explain why c-Myb can bind to a bigger pool of target sites compared to v-Myb. Although the T-stretch improves binding on either strand, it is required on a specific strand for transcriptional activation by both v-Myb and c-Myb. These results suggest that the local promoter structure is very important in this process. A stretch of adjacent T-A base pairs can adopt a Z-DNA-like confrontation that can result in a local bend at the junction of B- to Z-DNA (Wu and Crothers, 1984). Therefore, we suggest that the Myb proteins may require a specific local DNA conformation to induce transcriptional activation. Amino terminal to the three Myb repeats are 30 residues that include a ca-
38
Brigitte Canter and loseph S . Lipsick
sein kinase I1 (CKII) phosphorylation site and a run of acidic amino acids (10 of 15 are Glu or Asp) (Fig. 2). The presence of acidic residues but not the CKII sites is conserved among all known animal Myb proteins. Truncation of the first 30 residues of c-Myb removes both the CKII site and the acidic residues and results in increased DNA binding and transcriptional activation, but not in transformation of myelomonocytic cells in culture (Dini and Lipsick, 1993). However, retroviral insertional mutagenesis in nonbursal B cell lymphomas in chickens removes the first 20 residues of c-Myb, and a retrovirus producing this protein is oncogenic in animals (Jiang et al., 1997).Together these results suggest that the amino-terminal acidic residues and R1 regulate both DNA binding and cell proliferation, but that individual mutations may be cell-type specific in their transforming abilities. In this regard, many published experiments with variants of murine c-Myb called FL lack the first 1 7 residues of c-Myb, raising questions about whether this protein and various mutations derived from it are influenced by this potentially activating N-terminal mutation (Gonda et af., 1989). One of two published chicken c-My6 cDNA clones predicts an additional 60-codon open reading frame upstream of the initiation codon found in all other c-My6 clones from different species (Rosson and Reddy, 1986). Some investigators have reported that this additional open reading frame is specific for the thymus and results from intermolecular recombination with an unlinked gene that encodes an RNA splicing factor (Vellard et al., 1992). However, most other investigators have failed to detect either the hybrid mRNA or the predicted thymus-specific protein product. Therefore, an alternative hypothesis that must be considered is that this upstream open reading frame resulted from the ligation of two different cDNA fragments during the generation of a particular cDNA clone.
C. Transcriptional Activation Domain and Heptad Leucine Repeat v-Myb and c-Myb can activate transcription of model reporter genes that contain multiple Myb-binding sites upstream of a minimal promoter (Ibanez and Lipsick, 1990; Klempnauer et al., 1989; Nishina et af., 1989; Weston and Bishop, 1989). This transcriptional activation can be detected both in animal cells and in budding yeast (Chen and Lipsick, 1993; Punyammalee et al., 1991; Seneca et af., 1993). The transcriptional activation (TA) domain of the c-Myb protein has been mapped near the center of the protein. The limits of the TA domain are not well-defined and appear to depend on the cell line and reporter construct used (Chen et al., 1995; Ibanez and Lipsick, 1990; Kalkbrenner et al., 1990; Sakura et af., 1989). Fusion proteins that contain a heterologous DNA-binding domain and various fragments of vMyb were used to identify a small central acidic domain as the region re-
Myb and Oncogenesis
39
sponsible for transcriptional activation (Weston and Bishop, 1989). However, other studies have shown that this small central domain is not sufficient for transcriptional activation in the context of the native v-Myb protein (Ibanez and Lipsick, 1990). Rather, v-Myb contains several redundant regions that in various combinations are sufficient for transcriptional activation (Chen et al., 1995; Fu and Lipsick, 1996). Furthermore, none of the acidic residues within this central domain are necessary for transcriptional activation either in animal cells or in budding yeast. However, all of these activation regions together are required for oncogenic transformation by vMyb. The central acidic domain is highly conserved among the c-Myb and A-Myb proteins, but to a lesser degree if at all in the B-Myb, urchin Myb, or Drosophila Myb proteins (Figs. 1and 2). This observation is consistent with the failure to demonstrate transcriptional activation by the latter three proteins except under special conditions or in particular cell lines (Foos et al., 1992; Hou et al., 1997; Lane et al., 1997; Mizuguchi et al., 1990; Tashiro et al., 1995; Watson et al., 1993; Ziebold et al., 1997; (J. Manak and J. S. Lipsick, unpublished). In this regard, studies in yeast have shown that -1% of random bacterial open reading frames score positively as eukaryotic transcriptional activation domains, even though that is not their real function (Ma and Ptashne, 1987).The identification of an endogenous gene that is directly activated by B-Myb or Drosophila Myb therefore remains an important unanswered question. Carboxy-terminal to the central acidic domain of c-Myb is a region retained in AMV v-Myb that contains a heptad leucine repeat (HLR) that has been referred to as the “leucine zipper” (Kanei-Ishii et al., 1992). This term was originally coined to describe the dimerization domains of B-ZIP transcription factors such as C/EBP, GCN4, Fos, and Jun, which adopt a specific interdigitating, parallel coiled-coil structure (Landschulz et al., 1988; O’Shea et al., 1989). However, no similar structure has yet been demonstrated for c-Myb (Ebneth et al., 1994a). This heptad leucine repeat of cMyb has been proposed to function as a negative regulator because substitution of specific leucines with proline residues activates the protein, possibly by inhibiting dimerization (Nomura et al., 1993). In contrast, mutational analyses of v-Myb have shown that this region is essential for both transcriptional activation and oncogenic transformation by v-Myb (mutant 1120 versus 1151in Fig. 2), although the leucine residues are not required for these functions (Fu and Lipsick, 1996; Ibanez and Lipsick, 1988,1990). Although substitution of these leucines with alanine did not abolish transformation in culture, leukemogenicity in chickens was abolished, presumably due to temperature sensitivity of the mutant proteins (Bartunek et al., 1997).These results are consistent with the observation that v-Myb of E26, which lacks this region, transforms only weakly in the absence of fusion with the Ets protein (Metz and Graf, 1991).Interestingly, the leucine, isoleucine, and methionine residues of this leucine zipper are not all conserved in Xenopus c-Myb or the
40
Brigitte Canter and JosephS.Lipsick
closely related A-Myb proteins of mammals, birds, and amphibians (Fig. 2). However, other amino acids in this region that are required for transformation and transcriptional activation by v-Myb are highly conserved among the c-Myb, A-Myb, and to a lesser degree the B-Myb proteins (EFAETLQLID). Of additional interest, the alternatively spliced exon 9A of c-Myb inserts approximately 120 amino acids (E9A) just carboxy terminal to these conserved sequences prior to the final leucine of the zipper (Rosson et al., 1987; Shen-Ong, 1989). This larger protein, including E9A-encoded residues, represents 20% or less of the total c-Myb protein in cell types examined thus far. In contrast, sequence motifs encoded by this alternatively spliced exon of c-Myb are present in the major forms of A-Myb, B-Myb, urchin Myb, and Drosophila Myb proteins. B-My6 has been reported to display similar alternative splicing of this exon, albeit to a far lesser degree than c-Myb (Kamano et al., 1995). Exon 9A-encoded residues are not present in either the AMV or E26 v-Myb proteins. However, other studies have shown that the presence of exon 9A is compatible with oncogenic transformation (Woo et al., 1998). Furthermore, exon 9A appears to increase the transcriptional activation by c-Myb proteins that retain either their normal amino or carboxyl termini.
D. Regulation by the Carboxyl Terminus The carboxyl terminus of c-Myb that is deleted in v-Myb has been highly conserved during evolution and appears to function as a regulator of the remainder of the protein (Figs. 1 and 2). Truncation of this region occurs in the v-Myb oncoproteins of AMV and E26, and also occurs as a result of retroviral insertional mutagenesis in some murine myeloid leukemias. A recent publication suggests that carboxy-terminal truncation may activate cMyb by increasing protein stability (Bies and Wolff, 1997). However, a variety of other studies have not revealed significant steady-state differences in the abundance of strongly and weakly transforming mutants of c-Myb (Dini and Lipsick, 1993; Grasser et al., 1991; Hu et al., 1991). Many experiments have suggested that the carboxyl terminus of c-Myb is a negative regulator (NR) of transcriptional activation. First, truncation of this domain results in increased transcriptional activation of model reporter genes bearing Myb-binding sites (Hu et al., 1991; Sakura et al., 1989). Second, replacement of the c-Myb DNA-binding domain with that of the yeast GAL4 protein results in a complete lack of transcriptional activation (Dubendorff et al., 1992; Kalkbrenner et al., 1990). However, a carboxyterminal truncation similar to that of v-Myb strongly activates such GAL4-Myb fusion proteins. The deletion of two nonadjacent regions within the carboxyl terminus (PS and BN in Fig. 2) was required for the activation of GAL4-Myb fusion proteins in lieu of a complete truncation (Duben-
Myb and Oncogenesis
41
dorff et al., 1992). The same double deletion can also significantly activate the oncogenic potential of c-Myb (D. M. Wang and J. S. Lipsick, unpublished). Interestingly, the most conserved region of the carboxyl terminus lies between these two negative regulatory regions, suggesting that it has a different function. However, it is also possible that this region is important in negative regulation but that its structure is perturbed by these adjacent deletions. An analysis of LexA-Myb fusion proteins further demonstrated that the carboxyl terminus of c-Myb inhibited transcriptional activation but neither nuclear transport nor DNA binding. In addition, it was shown that the carboxyl terminus could specificallyinhibit the transcriptional activation domain of c-Myb or v-Myb in trans. These results led to the hypothesis that the c-Myb protein is regulated by intramolecular interactions. Additional experiments have suggested that the carboxyl terminus can also inhibit DNA binding by c-Myb (Ramsay et al., 1991). This was proposed to occur by homodimerization via the heptad leucine repeat (Nomura et al., 1993). The same group of investigators have mapped the inhibition of DNA binding to two nonoverlapping regions of the carboxyl terminus that flank but do not contain the heptad leucine repeat (Tanaka et al., 1997). On the other hand, other investigators have failed to observe inhibition of DNA binding by the carboxyl terminus (Krieg et al., 1995). Further experiments have suggested that there may be a direct regulation of the Myb DNA-binding domain by the carboxyl terminus. In particular, the amino and carboxyl termini of c-Myb were shown to score positively for protein-protein interaction in yeast two-hybrid and phage display assays (Dash et al., 1996; Kiewitz and Wolfes, 1997). Surprisingly, assays of transcriptional activation by various Myb proteins in budding yeast have demonstrated that the carboxyl terminus of c-Myb increases rather than inhibits transcriptional activation in this system (Chen and Lipsick, 1993; Seneca et al., 1993). These latter experiments have therefore led to a model in which negative regulation by the carboxyl terminus of c-Myb requires additional specific animal cell protein(s) not present in yeast. In support of such a model, other intestigators have reported that the carboxyl terminus of c-Myb can increase transcriptional activation in trans in some assays, presumably by titration of a limiting negative regulator (Vorbrueggenet al., 1994).
IV. REGULATION OF V-Myb AND c+Myb A. Sites of Phosphorylation The regulation of transcription factors by phosphorylation has been described for many different proteins (Karin, 1994). Both v- and c-Myb are phosphorylated at multiple sites in vivo, and at least some of these sites seem
42
Brigitte Ganter and Joseph S . Lipsick
\
I*
Fig. 4 Phosphorylation sites in c-Myb and v-Myb. The locations of potential phosphorylation sites in c-hilyb and v-Myb are indicated by asterisks. DBD, DNA-binding domain; R1, R2, and R3, Myb repeats; TA, transcriptional activation domain; NR, negative regulatory domain; HLR, heptad leiicine repeat; CK 11, casein kinase 11; PKA, protein kinase A; GSK 111, glycogen synthase kinase Ill; MAPK, mitogen-activated protein kinase. The dashed line indicates that a similar serine is present in v-Myb, but its phosphorylation by PKA has not yet been examined.
to be of functional importance (Fig. 4). At the N terminus of c-Myb, two ser-
ines (Ser-11 and Ser-12) have been mapped as in vivo phosphorylation sites and these two sites can be modified in vitro by casein kinase I1 (CKII) (Luscher eta/., 1990). Similar sequences are also found in Drosophila Myb and in vertebrate A-Myb, but not in B-Myb or urchin Myb (Fig. 2). Phosphorylation of Ser-1 1 and -12 was initially reported to reduce the DNA-binding activity of c-Myb, and substitution of these two serines by alanines resulted in a decreased cooperativity with NF-M (Luscher et al., 1990; Oelgeschlager et al., 1995). On the other hand, experiments with a shorter form of c-Myb that lacks the entire carboxyl terminus imply just the opposite, namely, that phosphorylation of Ser-11 and -12 by CKII increases DNA-binding activity (Ramsay et al., 1995).The same workers reported that in addition to the Nterminal CKII sites, Ser-116 within the second repeat of c-Myb can be phosphorylated by cyclic AMP-dependent protein kinase A (PKA) in vitro. Again it was suggested that phosphorylation of this site positively affects DNA binding by the c-Myb proteins. In general, it seems that CKII (Ser-11 and Ser12, absent in v-Myb) and PKA (Ser-116) potentially regulate c-Myb function through the N-terminal domain. However, no regulation of this phosphorylation has yet been reported in intact cells. Furthermore, mutation of the CKII sites does not cause oncogenic transformation in culture or in animals (Dini and Lipsick, 1993; Jiang et al., 1997). In addition to the amino-terminal phosphorylation sites (CKII and PKA sites), a number of other phosphorylation sites are present in c-Myb and in
Myb and Oncogenesis
43
v-Myb. Eight potential sites of phosphorylation by MAP kinases (proline-directed protein kinases) conserved between the avian, murine, and human Myb proteins are clustered in or near the carboxy-terminal negative regulatory domain of c-Myb (Aziz et al., 1993). Seven of these conserved sites are deleted in both the E26 and AMV viral oncoproteins. The p42""Pk kinase can phosphorylate avian c-Myb but not v-Myb in vitro on Ser-533, as analyzed by two-dimensional tryptic phosphopeptide mapping (Aziz et al., 1993).The same site is also phosphorylated in vivo and mutation of this serine causes increased transcriptional activation by c-Myb on some promoters, but not others (Aziz et al., 1995; Miglarese et al., 1996; Vorbrueggen et al., 1996). Therefore, it was suggested that modulation of c-Myb function might occur by phosphorylation of specific sites within the negative regulatory domain by MAP kinases. Because MAP kinases are localized both to the nucleus and to the cytoplasm, one might envision that MAP kinases regulate either the nuclear localization of c-Myb, alter the affinity of c-Myb for chromatin, or modulate the function of c-Myb by influencing inter- and intramolecular interactions of the carboxy-terminal negative regulatory domain. In vivo, other kinases, such as p44""pk and which show overlapping substrate specificity with p42""pk kinase (Hall and Vulliet, 1991),might also be involved in the phosphorylation and therefore regulation of c-Myb function. Consistent with this hypothesis, a mitosis-specific phosphorylation of c-Myb has also been reported (Luscher and Eisenman, 1992). For B-Myb the story is somewhat different. It was found that B-Myb is specifically phosphorylated during S phase (Robinson et al., 1996). Furthermore, the presence of ectopically expressed cyclin A stimulates transcriptional activation by B-Myb as well as its ability to promote the entry of cells into the S phase of the cycle (Ansieau et al., 1997; Lane et al., 1997; Sala et al., 1997; Ziebold et al., 1997). A similar increased transcriptional activation by A-Myb but not c-Myb in the presence of cyclin A has also been reported (Ziebold and Klempnauer, 1997).However, others have reported that exogenous cyclin A can stimulate both v-Myb and c-Myb transcriptional activation in a nonspecific fashion (Ganter et al., 1998). The mapping of cyclin A/cdk2 phosphorylation sites within these Myb proteins and an analysis of relevant mutations will therefore be of great interest. A cluster of serine and threonine phosphorylation sites was identified in AMV v-Myb (amino acids 267 to 303) by two-dimensional tryptic peptide mapping and antipeptide antibodies (Bading et al., 1989; Boyle et al., 1991). These sites can be phosphorylated efficiently in vitro by glycogen synthase kinase I11 (GSK-111)(Woodgett, 1991).However, although mutation of these phosphorylation sites greatly reduced the isoelectric heterogeneity of v-Myb, no alternations were observed in transcriptional activation or oncogenic transformation (Fu and Lipsick, 1996). The role of the homologous sites in c-Myb remains to be determined.
44
Brigitte Ganter and Joseph S. Lipsick
B. Interactions with Other Proteins A number of putative Myb-binding proteins (Fig. 5) have been identified. In this regard, several lines of evidence had suggested that specific proteinprotein interactions are important for the function of v-Myb and c-Myb. First, a series of deletions and linker insertions within v-Myb identified mutants that entered the nucleus, bound to DNA, but nevertheless failed to activate transcription or cause oncogenic transformation (Ibanez et a/., 1988; Ibanez and Lipsick, 1988, 1990; Lane et al., 1990). Second, mutations in both the DNA-binding and transcriptional activation domains of v-Myb are required to confer a monoblastic phenotype on transformed cells (Dini et a1.,1995; Introna et al., 1990). Three of the four amino acid substitutions within the v-Myb DNA-binding domain are predicted to lie on the surface of the protein that faces away from the DNA (Ogata et al., 1994). Third, although subsets of various transcriptional activation domains of v-Myb are sufficient for activating the expression of model reporter genes, the presence of all of these activation domains is required for oncogenic transformation (Chen et al., 1995; Fu and Lipsick, 1996). These results suggest that the regulation of multiple cellular target genes by distinct protein-protein interactions is likely to be required for transformation. Fourth, the observation that the carboxyl terminus of c-Myb functions as a negative regulator in animal cells but not in yeast suggests that specific animal cell proteins are required for this regulation (Chen and Lipsick, 1993; Seneca et al., 1993). Several proteins have been reported to interact directly with c-Myb. First, cyclin D was shown to inhibit transcriptional activation by v-Myb and oncogenic variants of c-Myb with similar amino-terminal truncations, but not by c-Myb (Ganter et a/., 1998). Truncation of the first Myb repeat ( R l ) was required for full inhibition by cyclin D. Surprisingly, this inhibition was CDK independent, a theme echoed by studies of a Myb-related cell cycle regulator DMP-1 that is also inhibited by cyclin D in a CDK-independent fashion (Hirai and Sherr, 1996; Inoue and Sherr, 1998). A second protein, Cyp40, was shown to inhibit DNA binding in vitro by c-Myb but not v-Myb (Leverson and Ness, 1998). This protein is a cyclophilin that contains a prolyl isomerase domain and a series of TPR repeats. Interestingly, the mutations present in the v-Myb DNA-binding domain specifically prevented inhibition of DNA binding by Cyp40. Because cyclosporin A inhibits the prolyl isomerase activity of Cyp40 and its ability to block DNA-binding by the c-Myb DNA-binding domain, it will be of interest to determine the effects of cyclosporin and other prolyl isomerase inhibitors on transcriptional activation by v-Myb and c-Myb. The central transcriptional activation domains of c-Myb and A-Myb have been reported to interact directly with the closely related CBP and p300 coactivators (Dai et al., 1996; Facchinetti et al., 1997; Kiewitz and Wolfes, 1997;
45
Myb and Oncogenesis
Chicken c-Myb
Domain of protein-protein interaction .Interaction with CBP -Interactionwith c y c h D
I
- Interaction with pl00
I
-Interaction with HF3 -Interaction with BS69
-
192
38
-Interaction with plM)
-Regionsof intramolecular interaction
317 I 342
192
38
192
-
192
426
-
-
561
375 I 405
513
563
Fig. 5 Sites of protein-protein interaction in c-Myb. Domains of c-Myb that have been reported to interact with other cellular proteins are indicated by black bars. The numbers indicate residues in chicken c-Myb without exon 9A.
Oelgeschlager et al., 1996).These proteins were initially discovered as PKAdependent coactivators of the CREB transcription factor and as cellular proteins that coprecipitated with the adenovirus E l A transforming protein (Goldman et al., 1997). Over the past few years a wide variety of sequencespecific transcription factors have been reported to bind directly to CBP and p300, including the retinoic acid receptor, the glucocorticoid receptor, Jun, Fos, STAT proteins, p53, Rb, and c-Myb. Interestingly, both CBP and p300 were reported to act as histone acetylases (Bannister and Kouzarides, 1996; Ogryzko et al., 1996). Various experiments have led to a model in which CBP/p300 are limiting coactivators in the cell and competition for them by different classes of activators (e.g., glucocorticoid receptor and Fos/Jun) results in inhibitory cross-talk between different signaling pathways (Kamei et al., 1996). Such a model is supported by the observation that CBP demonstrates haplo-insufficiency in humans-one mutant copy of the gene results in the multiple developmental defects that constitute the Rubinstein-Taybe syndrome (Petrij et al., 1995). In the case of c-Myb, it remains unknown whether CBP and p300 binding alone are necessary or sufficient for transcriptional activation or oncogenic transformation. In addition, it will be interesting to determine whether the previously reported inhibition of v-Myb transformation by liganded retinoic acid receptor is the result of such competition for CBP and/or p300 (Smarda et al., 1995). Several different proteins have been reported to bind directly to or mimic the carboxyl terminus of c-Myb. First, a protein that interacts with the heptad
46
Brigitte Canter and Ioseph S. Lipsick
leucine repeat of c-Myb was identified by pull-down assays and the relevant gene was later cloned (Favier and Gonda, 1994; Tavner et al., 1998). This p 160 protein appears to be present predominantly in the nucleolus and has some sequence similarity to known nucleolar proteins. However, Myb proteins appear to be specifically excluded from the nucleolus (Boyle et al., 1984; Klempnauer et al., 1984). In some hematopoietic cells, p160 is processed to a shorter cytoplasmic p67 form. In transient transfections, cDNAs encoding p67 but not p l 6 0 inhibited transcriptional activation by c-Myb. Second, a plOO protein that contains an EVES motif similar to that in the carboxyl terminus of c-Myb has also been reported to bind to the c-Myb DNA-binding domain and inhibit transcriptional activation by c-Myb (Dash et al., 1996).The EVES motif is a site for phosphorylation of c-Myb by MAP kinase, raising the possibility of a reversible and regulatable binding of plOO to c-Myb. The p100 protein was originally discovered in a yeast two-hybrid screen with the Epstein-Barr virus EBNAZ protein (Tong et al., 1995). In contrast to the results with c-Myb, p l 0 0 was reported to increase transcriptional activation by EBNA2. The EVES motif is not highly conserved among other Myb proteins, including A-Myb, B-Myb, and Drosophila Myb, implying that it may be a specific regulator for c-Myb alone. Third, the BS69 protein was recently identified in a yeast two-hybrid screen using the carboxyl terminus of c-Myb as bait (N. Collins and J. S. Lipsick, unpublished). This protein was initially identified as an E l A-binding protein than can inhibit transcriptional activation by E1A (Hateboer et al., 1995). We have found that BS69 can also inhibit transcriptional activation by fulllength c-Myb, but not by carboxy-terminal truncations of c-Myb. The vertebrate BS69 protein is present within the nucleus and contains several motifs common to other transcriptional regulators, including a PHD finger, a BKOMO-like domain, and a MYND domain that is also found in MTG8, Nervy, and DEAF-1. Furthermore, BS69 has been highly conserved during evolution, suggesting that perhaps it may regulate a variety of other Myb proteins via their highly conserved carboxyl termini, including A-Myb, BMyb, and Drosophila Myb (J. Manak and J. S. Lipsick, unpublished). The functional importance of all these putative Myb-binding proteins remains to be tested by appropriate genetic analyses.
V. TRANSCRIPTIONAL REGULATION BY V-Myb AND C-Myb A consensus DNA-binding sequence for the v-Myb protein was identified by analysis of a pool of random genomic DNA fragments that were selected with bacterially produced Ah4V v-Myb protein (Biedenkapp et al., 1988). Subse-
Myb and Oncogenesis
47
quently several laboratories showed that concatemers of this consensus sequence PyAAC(T/G)Gcan confer v-Myb- and c-Myb-dependentinducibility to a variety of test promoters (Ibanez and Lipsick, 1990; Klempnauer et al., 1989; Nishina et al., 1989; Weston and Bishop, 1989). In addition, similar binding sites were found within the promoter of mim-I, a cellular gene that is directly regulated by the E26 and c-Myb proteins (Ness et al., 1989). However, v-Myb and c-Myb can also activate promoters that do not contain Myb-binding sites, presumably by more indirect mechanisms (Engelke et al., 1995b; Ibanez and Lipsick, 1990; Kanei-Ishii et al., 1994,1997; Klempnauer et al., 1989).
A. Genes Activated by vcMyb and e M y b Oncogenic transformation of myelomonocytic cells by v-Myb and truncated forms of c-Myb initially appeared to correlate well with their ability to activate transcription of model reporter genes (Hu et al., 1991; Lane et al., 1990), suggesting that cell transformation depends on the activation of crucial target genes. However, the v-Myb protein encoded by the naturally occurring isolate of AMV is an extremely weak transcriptional activator and stronger activation does not correlate well with transformation (Engelke et al., 1995b). On the contrary, mutants of AMV v-Myb have been identified that activate transcription better than the wild-type protein but fail to transform myelomonocytic cells oncogenically in culture (Chen et al., 1995). Transcriptional activation by mutants of c-Myb also does not strictly correlate with transformation (Dini and Lipsick, 1993).Furthermore, although the acidic region of the HSV VP16 protein can substitute for the acidic region of v-Myb in oncogenic transformation, a simple v-Myb-VP16 fusion protein that strongly activates transcription does not itself cause oncogenic transformation (Engelke et al., 1995b; Frampton et al., 1993). These results suggest a model in which v-Myb has multiple transcriptional regulatory domains that are required for the regulation of specific “target” genes, all of which are required for oncogenic transformation (Chen et al., 1995). Because the Myb proteins bind to a rather small recognition site (AACNG), such potential recognition sites’can be found in almost any random piece of DNA of 1 kb or longer. For example, the commonly used E. coli plasmid pUC contains several Myb-binding sites. Therefore, the presence of such sites alone within a promoter does not necessarily imply that the gene in question is really regulated by Myb even if transient transfection studies seem to support such a conclusion. Nevertheless, many publications have identified genes thought to be regulated by v-Myb and c-Myb solely on the basis of DNA-binding and transient transfection assays. A more complete listing and discussion of genes proposed to be regulated by Myb proteins is provided in a review by Ness (1996).We will limit our discussion here to those genes for
48
Brigitte Canter and loseph S. Lipsick
which genetic evidence has been provided for direct regulation by Myb proteins. The first Myb-regulated gene to be identified was mim-1, a promyelocytespecific gene that is inducibly regulated by a temperature-sensitive mutant of the E26 Gag-Myb-Ets protein (Ness et al., 1989). This gene can also be induced by c-Myb but not by v-Myb of AMV, and it is not expressed in AMV- , transformed monoblasts, excepting for the AMV-transformed BM2 cell line (Dini et al., 1995; Queva et al., 1992). The promoter of the mim-1 gene contains three closely spaced Myb-binding sites and is strongly activated by vMyb and c-Myb in transient transfection assays. Only the strongest of these three Myb-binding sites is required for this activation in transient assays. Although c-Myb is expressed in many different types of hematopoietic and lymphoid cells, mim-1 gene expression is only detectable in granulocytic cells that constitute a small subset of those cell types that express c-My6 (Queva et al., 1992). Furthermore, some cells that express mim-1 in the developing embryo do not express c-My6. These observations were partially reconciled when it was shown that the myeloid-specific transcription factor NF-M, a B-ZIP protein that is the homolog of the mammalian C/EBP-P or NF-IL-6 protein (Sterneck et al., 1992), is required for the induction of mim-1 expression by Myb proteins (Ness et al., 1993). Remarkably, the introduction of c-Myb and NF-M into nonhematopoietic cells is sufficient to induce expression of the endogenous mim-1 gene (Burk et al., 1993; Ness et al., 1993). It has been reported that the Myb and C/EBP proteins interact directly via their DNA-binding domains even in the absence of DNA (Mink et al., 1996). Interestingly, proteins of the Myb and C/EBP families have also been reported to synergize in activating the promoters of other myeloid-specific genes (Burk et al., 1997). Another approach for identifying Myb-regulated genes has been to analyze a v-Myc-transformed chicken macrophage cell line that contains a hybrid E26/AMV v-Myb protein fused to the hormone-binding domain of the human estrogen receptor (Burke and Klempnauer, 1991). The molecular cloning of cDNAs that are differentially expressed in the presence or absence of estrogen has led to the identification of three additional Myb-regulated genes-chicken lysozyme, adenosine receptor 2B, and tom-1 (Burk et al., 1997; Worpenberg, et al., 1997). The latter gene is coregulated by Myb and C/EBP, as described above for mim-1. A similar approach has utilized a murine myeloid cell line transformed by a carboxy-terminal truncation of c-Myb fused to the hormone-binding domain of the estrogen receptor (Hogg et al., 1997). In these experiments the c-kit growth factor receptor gene was found to be a direct target of regulation by Myb. This is consistent with the presence of functional Myb-binding sites within the c-kit promoter region (Yamamoto et al., 1993). In contrast, the expression of c-Myc and Cdc2, which have also been proposed to be di-
Myb and Oncogenesis
49
rectly regulated by c-Myb, was not altered by withdrawal of estrogen. Furthermore, Bcl2, an antiapoptosis gene also proposed to be directly regulated by c-Myb, was still expressed after withdrawal of estrogen and apoptosis did not occur. Myelomonocytic cells transformed by AMV v-Myb do not require exogenous chicken myeloid growth factor, whereas those transformed by v-Myb proteins lacking the amino acid substitutions in either the DNA-binding or transcriptional activation domains are growth factor dependent (Dini et al., 1995).Indeed, transformation by AMV v-Myb appears to drive an autocrine loop that results in the production of cMGF by the transformed cells themselves. These results suggested that AMV v-Myb might either directly or indirectly regulate the promoter of the cMGF gene. Recent studies have shown that the cMGF promoter is directly regulated by the homeobox-containing GBX2 protein, and that the G B X 2 gene is regulated by AMV v-Myb (Kowenz-Leutzet al., 1997). G B X 2 was directly regulated by an AMV-E26 Myb-estrogen receptor fusion protein that contains only a single amino acid substitution present in the DNA-binding domain of v-Myb, which by itself does not confer the full AMV v-Myb phenotype. In contrast, G B X 2 was not induced by c-Myb in the absence of additional signal transduction. The authors concluded that mutations in the DNA-binding domain of AMV v-Myb render it independent of signaling events that are normally required for cMyb to activate GBX2. The discovery of this Myb/GBX2/growth factor axis is the first example of gene regulation by v-Myb that clearly contributes to oncogenic transformation.
B. Transcriptional Repression by Myb Proteins The failure of transcriptional activation by v-Myb to correlate well with oncogenic transformation raises the possibility that repression as well as activation may be important in this process. An analysis of the effects of Dtype cyclins on transcriptional activation by v-Myb favors the importance of repression in oncogenic transformation. In particular, cyclins D1 and D2 specifically inhibit transcription when activated through the v-Myb DNAbinding domain, but not the c-Myb DNA-binding domain (Ganter et al., 1998). The D-type cyclins belong to a subfamily of cyclins, proteins that are thought to govern transitions through distinct phases of the cell cycle by regulating the activity of cyclin-dependent kinases (CDKs) (Sherr, 1993). The D-type cyclins are strongly implicated in controlling progression through the G,/G, phase of the cell cycle. However, the inhibition of v-Myb by D-type cyclins appears to be independent of CDKs. This was shown by utilizing a dominant negative CDK4 mutant and a cyclin D1 mutant that cannot bind to its cyclin kinase partners (Ganter et al., 1998).
50
Brigitte Ganter and Joseph S. Lipsick
The biological relevance of these observations was supported by experiments showing that when a v-Myb-transformed monoblast cell line is induced to differentiate with phorbol esters, cyclin D levels go down and transcriptional activation by v-Myb goes up. These results suggested that repression rather than activation of some genes may correlate with v-Myb transformation. Consistent with this finding, a recent analysis of the Mybbinding sites within the N-Ras promoter revealed that both c-Myb and v-Myb can function as repressors by significantly reducing promoter activity in transient assays (Ganter and Lipsick, 1997). Furthermore, c-Myb has been shown to repress the c-ErbB2 promoter by direct competition with the TATA-binding protein in similar assays (Mizuguchi et al., 1995). The regulation of Myb-related proteins by D-type cyclins may be a more general phenomenon, because the distantly related Myb protein, DMPl, has recently been isolated in a yeast two-hybrid screen using cyclin D2 as the bait (Hirai and Sherr, 1996). Interestingly, cyclin D inhibits DNA binding by DMP1, also in a CDK-independent fashion (Inoue and Sherr, 1998). Furthermore, it has recently been shown that the estrogen receptor could be activated by cyclin D in a CDK- and ligand-independent fashion, arguing that cyclin D may have a wider role in transcriptional regulation than previously thought (Zwijsen et al., 1997).
VI. THE Myb-CHROMATIN CONNECTION Several additional transcriptional regulators have been identified that contain more distantly related Myb repeats (Table I). These Myb-repeat-containing proteins can be grouped into different families, including the telobox family (Bilaud et al., 1996), the SANT-domain family of transcriptional regulators (Aasland et al., 1996),and the transcription terminator family (Reeder and Lang, 1997). The telobox proteins include the recently identified human and mouse telomere-binding proteins TRFl and TRF2 (Bilaud et af., 1997; Broccoli et al., 1997; Chong et af., 1995), TBFl (Bilaud et al., 1996; Brigati et af., 1993), and the S. pombe Tazl protein (Cooper et al., 1997). Amino acid sequence comparison revealed that all of these proteins contain one or two distantly related Myb repeats. In addition, the three-dimensional X-ray structure of the major telomere-binding protein in budding yeast, RAPlp, unexpectedly revealed a Myb-related DNA-binding domain as a protein fold for telomeric DNA recognition (Konig et al., 1996; Konig and Rhodes, 1997). The SANT-domain-containing proteins include the SWISNF component SWI3 (Peterson and Herskowitz, 1992; Wang et al., 1996; Yoshinaga et al., 1992),the ISWI protein that participates in nucleosome remodelling and that also has significant similarity to SWI2 (Tsukiyama et al.,
Table I Myb Domain-Containing Proteins Involved in Transcription and Chromatin Remodeling" Protein family/name
Organism
Myb family A-Myb Mammals, birds, amphibians B-Myb Mammals, birds, amphibians c-Myb Mammals, birds, amphibians D-Myb Drosophila V-Myb AMV, E26 virus Telobox proteins Taz 1 Schizosaccharomyces pombe TBFl Saccharomyces cerevisiae TRFl Human, mouse TRF2 Human, mouse SANT-domain-containing proteins ADA2 Human S. cerevisiae B"
Saccharomyces cerevisiae
N-COR
Mouse
RSC-8 Saccharomyces Drosophila, human, S. cerevisiae I-SWI proteins SWI3 proteins Human, S. cerevisiae Transcription termination signals Rebl Saccharomyces cerevisiae, S. pombe TTFl Human, mouse Other proteins RAP1 SNAP190
Saccharomyces cerevisiae Human
Protein complex
Function
DNA binding
Transcriptional regulator, cell cycle control? Transcriptional regulator, cell cycle control? Transcriptional regulator, cell cycle control? Transcriptional regulator, cell cycle control? Leukemogenesis
Yes Yes Yes Yes Yes
Regulation of telomere length ? Regulation of telomere length Prevention of telomere fusion
Yes Yes Yes Yes
Transcriptional adapter, binds to histone acetyltransferase (GCNS) Subunit of TFIIIB
No No
Transcriptional corepressor
No
Chromatin remodeling Chromatin remodeling Chromatin remodeling
No No No
RNA pol I termination complex RNA pol I termination comp1ex
RNA pol I terminator
Yes
RNA pol I terminator
Yes
Telomere complex SNAPc complex
Regulation of telomere length, silencing snRNA transcriptional initiation by RNA pol I1 and I11
Yes Yes
Telomere protein Telomere protein Telomere protein Telomere protein
complex complex complex complex
ADA and SAGA complex RNA pol 111 initiation complex N-CoR/mSin3/mRPD3 corepressor complex RSC complex NURF complex SWI-SNF complex
"For a review of Myb-related proteins in plants, see Martin and Paz-Ares (1997).
52
Brigitte Canter and loseph S . Lipsick
1995), the ADA component ADA2 that complexes with the GCN5 histone acetylase (Berger et al., 1992; Candau et al., 1996), the corepressor N-CoR (Horlein et al., 1995), and the B subunit of TFIIIB (Kassavetis et al., 1995), a component of the RNA polymerase I11 initiation complex. The transcription terminator signal proteins with Myb-related sequences include the yeast Rebl protein and the mouse mTTF1 (Evers etal., 1995;Ju etal., 1990; Lang and Reeder, 1993). Interestingly, although all of these proteins are either involved in transcriptional regulation or chromatin remodeling, some of them have no intrinsic DNA-binding activity. The conservation of the Myb domain during the evolution of different transcriptional regulators/chromatin effectors that do not bind to DNA suggests another function for the Myb domain such as in specific protein-protein interactions that alter DNA accessibility in chromatin. We believe that carefully designed experiments will reveal whether the Myb domain has such a general function in chromatin remodeling, telomere binding, and transcription. Such non-DNA-binding functions of the Myb domain are likely to be important in understanding oncogenic transformation by v-Myb and c-Myb, as well. We therefore look forward to learning what other secrets the Myb proteins still harbor.
ACKNOWLEDGMENTS We thank the members of our laboratory, past and present, and our colleagues in the world of My6 research for their scientific efforts and their helpful discussions. Research in our own laboratory was supported by the National Cancer Institute of the United States Public Health Service. BG was supported in part by the Swiss National Science Foundation.
REFERENCES Aasland, R., Stewart, A. F., and Gibson, T. (1996). Trends Biochem. Sci. 21, 87-88. Alitalo, K., Winqvist, R., I h , C. C., de la Chapelle, A., Schwab, M., and Bishop, J. M. (1984). Proc. Natl. Acad. Sci. USA 81,4534-4538. Amaravadi, L., and King, M. W. (1994). Oncogene 9,971-974. Ansieau, S., Kowenz-Leutz, E., Dechend, R., and Leua, A. (1997).]. Mol. Med. 75,815-819. Anton, I. A., and Frampton, J. (1988). Nature (London) 336, 719. Arsura, M., Introna, M. Passerini, F., Mantovani, A., and Golay, J. (1992). Blood 79,2708-2716. Aziz, N., Wu, J., Dubendorff, J. W., Lipsick, J. S., Sturgill, T. W., and Bender, T. P. (1993). Oncogene 8,2259-2265. Aziz, N., Miglarese, M. R., Hendrickson, R. C., Shabanowitz, J., Sturgill, T. W., Hunt, D. F., and Bender, T. P. (1995).Proc. Natl. Acad. Sci. U.S.A. 92,6429-6433. Badiani, P., Corbella, P., Kioussis, D., Marvel, J., and Weston, K. (1994). Genes Dev. 8, 770-782.
Myb and Oncogenesis
53
Badiani, P. A., Kioussis, D., Swirsky, D. M., Lampert, I. A., and Weston, K. (1996). Oncogene 13,2205-2212. Bading, H., Beutler, C., and Moelling, K. (1989).Oncogene 4, 33-38. Baluda, M. A. (1963). CSH Symp. Quant. Biol. 27,415-425. Baluda, M. A., and Reddy, E. P. (1994). Oncogene 9,2761-2774. Bannister, A. J., and Kouzarides, T. (1996).Nature (London) 384, 641-643. Bartunek, P., Karafiat, V., Dvorakova, M., Zahorova, V., Mandikova, S., Zenke, M., and Dvorak, M. (1997).Oncogene 15,2939-2949. Belli, B., Wolff, L., Nazarov, V., and Fan, H. (1995).J. Virol. 69,5138-5141. Bender, T. P., and Kuehl, W. M. (1986).Proc. Natl. Acud. Sci. U.S.A. 83, 3204-3208. Bender, T. P., Thompson, C. B., and Kuehl, W. M. (1987).Science 237, 1473-1476. Berger, S. L., Pina, B., Silverman, N., Marcus, G. A., Agapite, J., Regier, J. L., Triezenberg, S. J., and Guarente, L. (1992).Cell 70,251-265. Bernstein, H. S., and Coughlin, S. R. (1997).J. Biol. Cbem. 272, 5833-5837. Beug, H., Blundell, P. A., and Graf, T. (1987). Genes Dev. 1,277-286. Biedenkapp, H., Borgmeyer, U., Sippel, A. E., and Klempnauer, K. H. (1988).Nature (London) 335,835-837. Bies, J., and Wolff, L. (1997). Oncogene 14,203-212. Bilaud, T., Koering, C. E., Binet-Brasselet, E., Ancelin, K., Pollice, A., Gasser, S. M., and Gilson, E. (1996).Nucleic Acids Res. 24, 1294-1303. Bilaud, T., Brun, C., Ancelin, K., Koering, C. E., Laroche, T., and Gilson, E. (1997). Nature Genet. 17,236-239. Boise, L. H., Gorse, K. M., and Westin, E. H. (1992). Oncogene 7,1817-1825. Boyle, W. J., Lampert, M. A., Lipsick, J. S., and Baluda, M. A. (1984).Proc. Natl. Acad. Sci. U.S.A. 81,4265-4269. Boyle, W. J., van der Geer, P., and Hunter, T. (1991).Methods Enzymol. 201, 110-149. Brigati, C., Kurtz, S., Balderes, D., Vidali, G., and Shore, D. (1993). Mol. Cell. Biol. 13, 1306-1314. Broccoli, D., Smogorzewska,A., Chong, L., and de Lange, T. (1997).Nature Genet. 17,231-235. Burgess, T. L., Fisher, E. F., Ross, S. L., Bready, J. V., Qian, Y. X., Bayewitch, L. A., Cohen, A. M., Herrera, C. J., Hu, S. S., Kramer, T. B., et al. (1995).Proc. Natl. Acad. Sci. U.S.A. 92, 4051-4055. Burk, O., and Klempnauer, K. H. (1991).EMBO J. 10,3713-3719. Burk, O., Mink, S., Ringwald, M., and Klempnauer, K. H. (1993).EMBOJ. 12,2027-2038. Burk, O., Worpenberg, S., Haenig, B., and Klempnauer, K. H. (1997). EMBO J. 16, 13711380. Calabretta, B., and Nicolaides, N. C. (1992). Crit. Rev. Eukaryot. Gene Expr. 2,225-235. Candau, R., Moore, P. A., Wang, L., Barlev, N., Ying, C. Y., Rosen, C. A., and Berger, S. L. (1996).Mol. Cell. Biol. 16,593-602. Catron, K. M., Purkerson, J. M., Isakson, P. C., and Bender, T. P. (1992).J. Immunol. 148, 934-942. Chen, J. H. (1980).]. Virol. 36,162-170. Chen, R. H., and Lipsick, J. S. (1993).Mol. Cell. Biol. 13,4423-4431. Chen, R. H., Fields, S., and Lipsick, J. S. (1995). Oncogene 11, 1771-1779. Chong, L., van Steensel, B., Broccoli, D., Erdjument-Bromage, H., Hanish, J., Tempst, P., and de Lange, T. (1995).Science 270, 1663-1667. Clarke, M. F., Kukowska-Latallo, J. F., Westin, E., Smith, M., and Prochownik, E. V. (1988). Mol. Cell. Biol. 8, 884-892. Coffman, J. A., Kirchhamer, C. V., Harrington, M. G., and Davidson, E. H. (1997).Development 124,4717-4727. Cooper, J. P., Nimmo, E. R., Allshire, R. C., and Cech, T. R. (1997). Nature (London) 385, 744-747.
54
Brigitte G a n t e r a n d Joseph S. Lipsick
Dai. P., Akimaru, H., Tanaka, Y., Hou, D. X., Yasukawa, T., Kanei-Ishii, C., Takahashi, T., and Ishii, S. (1996). Genes Dew. 10,528-540. Dash, A. B., Orrico, F. C., and Ness, S. A. (1996). Genes Dev. 10, 1858-1869. DeGregori, J., Kowalik, T., and Nevins, J. R. (1995). Mol. Cell. Biol. 15,4215-4224. DeRocco, S . E., Iozzo, R., Ma, X. P., Schwarting, R., Peterson, D., and Calabretta, B. (1997). Proc. Natl. Acad. Sci. U.S.A. 94,3240-3244. Dini, P. W., and Lipsick, J. S. (1993). Mol. Cell. Biol. 13, 7334-7348. Dim, P. W., Eltman, J. T., and Lipsick, J. S. (1995).]. Virol. 69, 2515-2524. Dubendorff, J. W., Whittaker, L. J., Eltman, J. T., and Lipsick, J. S. (1992). Genes Dev. 6, 2524-2535. Duprey, S. P., and Boettiger, D. (1985). Proc. Natl. Acad. Sci. U.S.A. 82,6937-6941. Durban, E. M., and Boettiger, D. (1981). Proc. Natl. Acad. Sci. U.S.A. 78,3600-3604. Dvorak, M., Urbanek, P., Bartunek, P., Paces, V., Vlach, J., Pecenka, V., Arnold, L., Travnicek, M., and Riman, J. (1989). Nucleic Acids Res. 17, 5651-5664. Ebneth, A,, Adermann, K., and Wolfes, H. (1994a). FEBS Lett. 337,265-268. Ebneth, A., Schweers, O., Thole, H., Fagin, U., Urbanke, C., Maass, G., and Wolfes, H. (1994b). Biochemtstry 33, 14586- 14593. Engelke, U., and Lipsick, J. S. (1994).1. ViroI. 68, 2752-2755. Engelke, U., Smarda, J., and Lipsick, J. S. (1995a). Oncogene 11, 735-741. Engelke, U., Whittaker, L., and Lipsick, J. S. (1995b). Virology 208, 467-477. Engelke, U., Wang, D. M., and Lipsick, J. S. (1997).J. Virol. 71,3760-3766. Estus, S., Zaks, W. J.. Freeman, R. S., Gruda, M., Bravo, R., and Johnson, Jr., E. M. (1994).]. Cell Biol. 127, 1717-1727. Evers, R., Srnid, A , , Rudloff, U., Lottspeich, F., and Grummt, 1. (1995). EMBO J. 14, 1248-1256. Facchinetti, V., Loffarelli, L., Schreek, S., Oelgeschlager, M., Luscher, B., Introna, M., and Golay, J. (1997). Biochern.]. 324, 729-736. Favier, D., and Gonda, T. J. (1994). Oncogene 9,305-311. Feng, D.-F., and Doolittle, R. F. (1990). Methods Enzymol. 183, 375-387. Ferrao, P., Macmillan, E. M., Ashman, L. K., and Gonda, T. J. (1995). Oncogene 11, 1631-1638. Foos, G., Grimm, S., and Klempnauer, K. H. (1992).EMBO 1. 11,4619-4629. Frampton, J., Gibson, T. J., Ness, S. A., Doderlein, G . , and Graf, T. (1991). Protein Eng. 4, 891-901. Frampton, J., Kouzarides, T., Doderlein, G., Graf, T., and Weston, K. (1993). EMBO]. 12, 1333-1341. Frampton, J., Ramqvist, T., and Graf, T. (1996). Genes Dev. 10, 2720-2731. Fu, S. L., and Lipsick, J. S. (1996).1. Virol. 70, 5600-5610. Fu. S. L., and Lipsick, J. S. (1997). Cell Growth Differ. 8, 35-45. Furuta, Y., Aizawa, S., Suda, Y., Ikawa, Y., Nakasgoshi, H., Nishina, Y., and Ishii, S. (1993). Transgenic Res. 2, 199-207. Gabrielsen, 0. S., Sentenac, A., and Frornageot, P. (1991).Science 253, 1140-1143. Ganter, B., and Lipsick, J. S. (1997). Oncogene 15, 193-202. Ganter, B., Fu, S., and Lipsick, J. S. (1998).EMBO]. 17, 255-268. Garcia, A,, LaMontagne, K., Reavis, D., Stober-Crasser, U., and Lipsick, J. S. (1991). Oncogene 6,265-273. Garrido, C., Grasser, F., Lipsick, J. S., Stehelin, D., and Saule, S. (1992). I. Virol. 66, 6773 -6776. Gerondakis, S., and Bishop, J. M.(1986). Mol. Cell. Biol. 6, 3677-3684. Gewirtz, A. M., Anfossi, G . , Venturelli, D., Valpreda, S., Sims, R., and Calabretta, B. (1989). Science 245, 180-183. Golay, J., Broccoli, V., Lamorte, G . , Bifulco, C., Parravicini, C., Pizzey, A., Thomas, N. S . , Delia, D.. Ferrauti, l?, Vitolo, D., and Introna, M. (1998).]. lrnrnunol. 160, 2786-2793.
Myb and Oncogenesis
55
Goldman, P. S., Tran, V. K., and Goodman, R. H. (1997). Recent Frog. Horn. Res. 52, 103-120. Gonda, T. J., and Metcalf, D. (1984).Nature (London)310,249-251. Gonda, T. J., Gough, N. M., Dunn, A. R., and de Blaquiere,J. (1985).E M B O J . 4,2003-2008. Gonda, T. J., Buckmaster, C., and Ramsay, R. G. (1989).E M B O J. 8, 1777-1783. Graf, T. (1992). Curr. Opin. Genet. Dev. 2,249-255. Graf, T., McNagny, K., Brady, G., and Frampton, J. (1992). Cell 70,201-213. Grasser, F. A., Graf, T., and Lipsick, J. S. (1991).Mol. Cell. Biol. 11,3987-3996. Grasser, F. A., LaMontagne, K., Whittaker, L., Stohr, S., and Lipsick, J. S. (1992).Oncogene 7, 1005-1009. Greco, C., Gandolfo, G. M., Mattei, F., Gradilone, A., Alvino, S., Pastore, L. I., Casale, V., Casole, P., Grassi, A., Cianciulli, A. M., et ul., (1994).Anticancer Res. 14, 1433-1440. Guehmann, S., Vorbrueggen, G., Kalkbrenner, F., and Moelling, K. (1992).Nucleic Acids Res. 20,2279-2286. Guerin, M., Sheng, Z. M., Andrieu, N., and Riou, G. (1990).Oncogene 5,131-135. Guerra, J., Withers, D. A., and Boxer, L. M. (1995).Blood 86,1873-1880. Hagerman, P. J. (1996).Proc. Natl. Acad. Sci. U.S.A. 93,9993-9996. Hall, F. L., and Vulliet, P. R. (1991).Curr. Opin. Cell. Biol. 3, 176-184. Harper, M. E., Franchini, G., Love, J., Simon, M. I., Gallo, R. C., and Wong-Staal, F. (1983). Nature (London)304, 169-171. Hateboer, G., Gennissen, A., Ramos, Y. F., Kerkhoven, R. M., Sonntag-Buck, V., Stunnenberg, H. G., and Bernards, R. (1995).E M B O J . 14,3159-3169. Hegvold, A. B., and Gabrielsen, 0. S. (1996).Nucleic Acids Res. 24,3990-3995. Henderson, A. S., and Wolman, S. R. (1988). Cytogenet. Cell Genet. 47,95-97. Hirai, H., and Sherr, C. J., (1996).Mol. Cell. Biol. 16, 6457-6467. Hirayama, T., and Shinozaki, K. (1996).Proc. Natl. Acad. Sci. U.S.A. 93, 13371-13376. Hogg, A., Schirm, S., Nakagoshi, H., Bartley, P., Ishii, S., Bishop, J. M., and Gonda, T. J. (1997). Oncogene 15,2885-2898. Horlein, A. J., Naar, A. M., Heinzel, T., Torchia, J., Gloss, B., Kurokawa, R., Ryan, A., Kamei, Y., Soderstrom, M., Glass, C. K., and Rosenfeld, M. G. (1995). Nature (London) 377, 397-404. Hou, D. X., Akimaru, H., and Ishii, S. (1997).FEBS Lett. 413, 60-64. Howe, K. M., and Watson, R. J. (1991).Nucleic Acids Res. 19, 3913-3919. Howe, K. M., Reakes, C. F., and Watson, R. J. (1990).E M B O J . 9,161-169. Hu, Y. L., Ramsay, R. G., Kanei-Ishii, C., Ishii, S., and Gonda, T. J. (1991). Oncogene 6, 1549-1553. Hurford, Jr., R. K., Cobrinik, D., Lee, M. H., and Dyson, N. (1997). Genes Dev. 11, 1447-1463. Ibanez, C. E., and Lipsick, J. S. (1988).]. Virol. 62,1981-1988. Ibanez, C. E., and Lipsick, J. S. (1990).Mol. Cell. Biol. 10,2285-2293. Ibanez, C. E., Garcia, A., Stober-Grasser,U., and Lipsick, J. S. (1988).J.Virol. 62,4398-4402. Inoue, K., and Sherr, C. J. (1998).Mol. Cell. Biol. 18, 1590-1600. Introna, M., Golay, J., Frampton, J., Nakano, T., Ness, S. A., and Graf, T. (1990). Cell 63, 1289-1297. Introna, M., Luchetti, M., Castellano, M., Arsura, M., and Golay, J. (1994). Semin. Cancer Biol. 5,113-124. Jacobs, S. M., Gorse, K. M., and Westin, E. H. (1994).Oncogene 9,227-235. Jiang, W., Kanter, M. R., Dunkel, I., Ramsay, R. G., Beemon, K. L., and Hayward, W. S. (1997). J. Virol. 71,6526-6533. Ju, Q. D., Morrow, B. E., and Warner, J. R. (1990).Mol. Cell. Biol. 10, 5226-5234. Kalkbrenner, F., Guehmann, S., and Moelling, K., (1990).Oncogene 5,657-661. Kamano, H., Burk, B., Noben-Trauth, K., and Klempnauer, K. H. (1995). Oncogene 11, 2575-2582.
56
Brigitte Ganter and loseph S. Lipsick
Kamei, Y., Xu, I.., Heinzel, T., Torchia, J., Kurokawa, R., Gloss, B., Lin, S. C., Heyman, R. A., Rose, D. W., Glass, C. K., and Rosenfeld, M. G. (1996). Cell 85, 403-414. Kanei-Ishii, C., Sarai, A,, Sawazaki, T., Nakagoshi, H., He, D. N., Ogata, K., Nishirnura, Y., and Ishii, S. (1990f.J.Biol. Cbem. 265, 19990-19995. Kanei-Ishii, C., MacMillan, E. M.,Nomura, T., Sarai, A., Ramsay, R. G., Aimoto, S., Ishii, S., and Gonda, T. J. (1992).Proc. Nut/. Acud. Sci. U.S.A. 89,3088-3092. Kanei-Ishii, C., Yasukawa, T., Morimoto, R. I., and Ishii, S. (1994). 1. Biol. Cbem. 269, 15768- 1577.5. Kanei-Ishii, C., Tanikawa, J., Nakai, A,, Morimoto, R. I., and Ishii, S. (1997). Science 277, 246 -24 8. Kanter, ,M. R., Smith, R. E., and Hayward, W. S. (1988).J.Virol. 62, 1423-1432. Karin, h4. (1994).Cum Opin. CellBiol. 6, 415-424. Kassavetis, 6 . A., Nguyen, S. T., Kobayashi, R., Kurnar, A., Geiduschek, E. P., and Pisano, M. (1995).Proc. Natl. Acad. Sci. U.S.A. 92, 9786-9790. Karzen, A. L., and Bishop, J. M. (1996). Proc. Nutl. Acud. Sci. U.S.A. 93, 13955-13960. Katzen, A. L.. Kornberg, T. B., and Bishop, J. M. (1985). Cell 41,449-456. Katzen, A. L., Jackson, J., Harmon, B. I’., Fung, S. M., Ramsay, G., and Bishop, J. M. (1998). Genes Dev. 12,831-843. Kiewitz, A., and Wolfes, H. (1997).FEBS Lett. 415,258-262. Kirsch, I. R., Berrness, V., Silver, J., and Hollis, G. F. (1986).J. Cell Biochem. 32, 11-21. Klempnauer, K. H . (1993). Oncogene 8, 111-115. Klempnauer, K. H., and Sippel, A. E. (1987). EMBO 1.6,2719-2725. Klempnauer, K. H., Gonda, T. J., and Bishop, J. M. (1982). Cell 31,453-463. Klempnauer, K. H., Symonds, G., Evan, G. I., and Bishop, J. M. (1984). Cell 37,537-547. Klempnauer, K. H., Arnold, H., and Biedenkapp, H. (1989). Genes Dev. 3, 1582-1589. KSnig, P., and Rhodes, D. (1997).Trends Biochem. Sci. 22,43-47. Konig, P., Giraldo, R., Chapman, L., and Rhodes, D. (1996). Cell 85, 125-136. Kowenz-Leutz, E., Herr, P., Niss, K., and Leutz, A. (1997). Cell91, 185-195. Krieg, J., Oelgeschlager, M., Janknecht, R., and Luscher, B. (1995). Oncogene 10, 22212228. Krumm, A., Media, T., Brunvand, M., and Groudine, M. (1992). Genes Dev. 6, 2201-2213. Lam, E. W., and Watson, R. J. (1993).EMBO]. 12,2705-2713. Lam, E. W., Robinson, C., and Watson, R. J. (1992). Oncogene 7, 1885-1890. Landschulz, W. H., Johnson, P. F., and McKnight, S. L. (1988).Science 240,1759-1764. Lane, S., Farlie, P., and Watson, R. (1997). Oncogene 14,2445-2453. Lane, T., Ibanez, C., Garcia, A,, Graf, T., and Lipsick, J. (1990).Mol. Cell. Biol. 10,2591-2598. Lang, W. H., and Reeder, R. H. (1993).Mol. Cell. Biol. 13,649-658. Lee, J., and Goldfarb, A. (1991).Cell 66, 793-798. Leprince, D., Gegonne, A,, Coll, J., de Taisne, C., Schneeberger, A., Lagrou, C., and Stehelin, D. (1983).Nature (London) 306,395-397. Le Rouzic, E., and Perbal, B. (1996).J. Virol. 70, 7414-7423. Leverson, J. D., and Ness, 5. A. (1998).Mol. Cell 1,203-211. Lin, D., Fiscella, ‘M., O’Connor, P. M., Jackman, J., Chen, M., Luo, L. L., Sala, A., Travali, S., Appella, E., and Mercer, W. E. (1994). Proc. Nutl. A d . Sci. U.S.A. 91, 10079-10083. Lipsick, J. 5 . (1987).1. Virol. 61, 3284-3287. Lipsick, J. S. (1996). Oncogene, 13, 223-235. Lipsick, J. S., and Boyle, W. J, (1987).Mol. Cell. Biol. 7, 3358-3360. Lipsick, J. S., and Ibanez, C. E. (1987).J. Virol. 61,933-936. Luscher, B., and Eisenrnan, R. N. (1992).1. Cell Bid. 118,775-784. Luscher, B., Chrisrenson, E., Litchfield, D. W., Krehs, E. G., and Eisenrnan, R. N. (1990).Nuture (London) 344,517-522. Ma, J.. and Ptashne, M. (1987).Cell51, 113-119.
Myb and Oncogenesis
57
Marhamati, D. J., Bellas, R. E., Arsura, M., Kypreos, K. E., and Sonenshein, G. E., (1997).Mol. Cell. Biol. 17,2448-2457. Martin, C., and Paz-Ares, J. (1997).Trends Genet. 13,67-73. McCann, S., Sullivan,J., Guerra, J., Arcinas, M., and Boxer,, L. M. (1995).J. Biol. Chem. 270, 23785-23789. McIntosh, P., Frenkiel, T., Wollborn, U., McCormick, J., Klempnauer, K., Feeney, J., and Carr, M. (1998).Biochemistry 27,9619-9626. Menus, R. V., Litvin, J., Wali, A., Toscani, A., Latham, K., Hatton, K., and Reddy, E. P. (1994). Oncogene 9,3077-3086. Metz, T., and Graf, T. (1991).Cell 66,95-105. Metz, T., Graf, T., and Leutz, A. (1991).E M B O J. 10,837-844. Miglarese, M. R., Richardson, A. F., Aziz, N., and Bender, T. P. (1996). J. Biol. Cbem. 271, 22697-22705. Mink, S., Kerber, U., and Klempnauer, K. H. (1996).Mol. Cell. B i d . 16, 1316-1325. Mizuguchi, G., Nakagoshi, H., Nagase, T., Nornura, N., Date, T., Ueno, Y., and Ishii, S. (1990). J. Biol. Cbem. 265,9280-9284. Mizuguchi, G., Kanei-Ishii, C., Takahashi, T., Yasukawa, T., Nagase, T., Horikoshi, M., Yamamoto, T., and Ishii, S. (1995).J. Btol. Chem. 270,9384-9389. Moscovici, M. G., and Moscovici, C. (1983).Proc. Natl. Acad. Sci. U.S.A. 80, 1421-1425. Moscovici, M. G., Jurdic, P., Samarut, J., Gazzolo, L., Mura, C. V., and Moscovici, C. (1983). Virology 129, 65-78. Mucenski, M. L., McLain, K., Kier, A. B., Swerdlow, S. H., Schreiner, C. M., Miller, T. A., Pietryga, D. W., Scott, W. J., Jc, and Potter, S. S. (1991).Cell 65,677-689. Mudryj, M., Hiebert, S. W., and Nevins, J. R. (1990).E M B O J. 9,2179-2184. Myrset, A. H., Bostad, A., Jamin, N., Lirsac, P. N., Toma, F., and Gabrielsen, 0. S. (1993). E M B O J. 12,4625-4633. Nazarov, V., and Wolff, L. (1995).J. Virol. 69,3885-3888. Ness, S. A. (1996).Biochim. Biophys. Acta 1288, F123-F139. Ness, S. A., Marknell, A., and Graf, T. (1989).Cell 59, 1115-1125. Ness, S. A., Kowenz-Leutz, E., Casini, T., Graf, T., and Leutz, A. (1993). Genes Dev. 7, 749 -759. Nicolaides, N. C., Gualdi, R., Casadevall, C., Manzella, L., and Calabretta, B. (1991). Mol. Cell. Biol. 11, 6166-6176. Nishina, Y., Nakagoshi, H., Imamoto, F., Gonda, T. J., and Ishii, S. (1989).Nucleic Acids Res. 17,107-117. Nomura, N., Takahashi, M., Matsui, M., Ishii, S., Date, T., Sasamoto, S., and Ishizaki, R. (1988).Nucleic Acids Res. 16, 11075-11089. Nomura, T., Sakai, N., Sarai, A., Sudo, T., Kanei-Ishii, C., Ramsay, R. G., Favier, D., Gonda, T. J., and Ishii, S. (1993).]. Biof. Chem. 268,21914-21923. Nunn, M. F., Seeburg, P. H., Moscovici, C., and Duesberg, P. H. (1983). Nature (London) 306, 391-395. Nunn, M., Weiher, H., Bullock, P., and Duesberg, P. (1984). Virology 139,330-339. Oehler, T., Arnold, H., Biedenkapp, H., and Klempnauer, K. H. (1990).Nucleic Acids Res. 18, 1703-1710. Oelgeschlager, M., Krieg, J., Luscher-Firzlaff,J. M., and Luscher, B. (1995).Mol. Cell. B i d . 15, 5966-5974. Oelgeschlager,M., Janknecht, R., Krieg, J., Schreek, S., and Luscher, B. (1996).E M B O J. 15, 2771-2780. Ogata, K., Hojo, H., Aimoto, S., Nakai, T., Nakamura, H., Sarai, A., Ishii, S., and Nishimura, Y. (1992).Proc. Natl. Acad. Sci. U.S.A. 89,6428-6432. Ogata, K., Morikawa, S., Nakamura, H., Sekikawa, A., Inoue, T., Kanai, H., Sarai, A., Ishii, S., and Nishimura, Y. (1994). Cell 79, 639-648.
58
Brigitte Canter and JosephS. Lipsick
Ogaca, K., Morikawa, S., Nakamura, H., Hojo, H., Yoshimura, S . , Zhang, R., Aimoto, S., Arnetani, Y., Hiram, Z., Sarai, A., et al. (1995).Nature Struct. Biol. 2,309-320. Ogata, K., Kanei-Ishii. C., Sasaki, IM., Hatanaka, H., Nagadoi, A., Enari, M., Nakamura, H., Nishimura, Y., Ishii, S., and Sarai, A. (1996).Nature Strut. Bzol. 3, 178-187. Ogryzko, V. V., Schiltz, R. L., Russanova, V., Howard, B. H,, and Nakatani, Y. (1996).Cell 87, 953-959. Ohi, R., McCollum, D., Hirani, B., Den Haese, G. J., Zhang, X, Burke, J. D., Turner, K., and Gould, K. I.. (19941.E M B O J . 13,471-483. Ohi, R., Feokistova, A., McCann, S., Valentine, V., Look, A. T., Lipsick, J. S., and Gould, K. L., (1998).Mot. Cell. Biol. 18,4097-4108. O'Shea, E. K., Rutkowski, R., and Kim, P. S., (1989).Science 243,538-542. Perbal, B., Lipsick, J. S., Svoboda, J., Silva, R. F., and Baluda, M. A. (198q.J.Virol. 56,240-244. Pessano, S., Gazzolo, L., and Moscovici, C. (1979).Microbiologica, 2, 379-392. Peterson, C. L., and Herskowitz, I. (1992). Cell 68,573-583. Petrij, F., Giles, R. H., Dauwerse, H. G., Saris, J. J., Hennekarn, R. C., Masuno, M., Tommerup, N., van Ommen, G. J., Goodman, R. H., Peters, D. J., et a[., (1995).Nature (London) 376, 348-351. Phan, S. C., Feeley, B., Withers, D., and Boxer, L. M. (1996).Mol. Cell. Biol. 16,2387-2393. Pizer, E., and Hurnphries, E. H. (1989).J . Virol. 63, 1630-1640. Pizer, E. S., Baba, T. W., and Humphries, E. H. (1992).J. Virol. 66, 512-523. Press, R. D., Reddy, E. P., and Ewert, D. L., (1994).Mol. Cell. Biol. 14,2278-2290. Punyammalee, B., Crabeel, M., de Lannoy, C., Perbal, B., and Glansdorff, N. (199.1).Oncogene 6,11-19. Queva, C., Ness, S. A., Crasser, F. A., Graf, T., Vandenbunder, B., and Stehelin, D. (1992).Development 114,125-1 33. Ramsay, R. G., Ikeda, K., Rifkind, R. A., and Marks, P. A. (1986).Proc. Natl. Acad. Sci. U.S.A. 83,6849-6853. Ramsay, R. G., Ishii, S., and Gonda, T. J. (1991).Oncogene 6, 1875-1879. Ramsay, U.G., Morrice, N., Van Eeden, P., Kanagasundaram, V., Nomura, T., De Blaquiere, J., Ishii, S., and Wettenhall, R. (1995).Oncogene 11, 2113-2120. Reddy, C. D., and Reddy, E. P. (1989).Proc. Natl. Acad. Sri. U.S.A. 86,7326-7330. Reeder, R. H., and Lang, W. H. (1997). Trends Biochern. Sci. 22,473-477. Reiss, K., Ferber, A., Travali, S., Porcu, P., Phillips, P. D., and Baserga, R. (1991).Cancer Res. 51,5997-6000. Robinson, C., Light, Y., Groves, R., Mann, D., Marias, R., and Watson, R. (1996).Oncogene 12,1855-1864. Rosson, D., and Reddy, E. P. (1986).Nature (London) 319,604-606. Rosson, D., Dugan, D., and Reddy, E. P. (1987). Proc. Natl. Acad. Sci. U.S.A. 84, 31713175. Rushlow, K. E., Lautenberger, J. A., Papas, T. S . , Baluda, M. A., Perbal, B., Chirikjian, J. G., and Reddy, E. P. (1982a).Science 216, 1421-1423. Rushlow, K. E., Lautenberger, J. A., Reddy, E. P., Souza, L. M., Baluda, M. A., Chirikjian, J. G., and Papas, T. S. (1982b).J. Virol. 42, 840-846. Saikumar, P., Murali, R., and Reddy, E. P. (1990).Proc. Natl. Acad. Sci. U.S.A. 87,8452-8456. Saikumar, P., Gabriel, J. L., and Reddy, E. P. (1994).Oncogene 9, 1279-1287. Sakura, H., Kanei-Ishii, C., Nagase, T., Nakagoshi, H., Gonda, T. J., and Ishii, S. (1989).Proc. Natl. Acad. Sci. U.S.A. 86,5758-5762. Sala, A., and Calabretta, B. (1992).PTOC.Natl. Acad. Sci. U.S.A. 89, 10415-10419. Sala, A., Kundu, iM., Casella, I., Engelhard, A., Calabretta, B., Grasso, L., Paggi, M. G., Giordano A., Watson, R. J., Khalili, K., and Peschle, C. (1997).Proc. Natl. Acad. Sci. U.S.A. 94, 532-536.
Myb and Oncogenesis
59
Salomoni, P., Perrotti, D., Martinez, R., Franceschi, C., and Calabretta, B. (1997).Proc. Natl. Acad. Sci. U.S.A. 94, 3296-3301. Saville, M. K., and Watson, R. J. (1998).Adv. Cancer Res. 72, 109-140. Schuler, G. D., Altschul, S . F., and Lipman, D. J. (1991). Proteins Struct. Funct. Genet. 9, 180-190. Selvakumaran, M., Liebermann, D. A., and Hoffman-Liebermann, B. (1992).Mol. Cell. Biol. 12,2493-2500. Seneca, S., Punyammalee, B., Sureau, A., Perbal, B., Dvorak, M., and Crabeel, M. (1993).Oncogene 8,2335-2342. Shen-Ong, G. L. (1989). Curr. Top. Microbiol. Immunol. 149, 71-76. Shen-Ong, G. L. (1990).Biochim. Biophys. Actu 1032,39-52. Shen-Ong, G. L., Potter, M., Mushinski, J. F., Law, S., and Reddy, E. P. (1984).Science 226, 1077-1080. Shen-Ong, G. L., Morse, H. C. D., Potter, M., and Mushinski, J. F. (1986).Mol. Cell. Biol. 6, 380-392. Sherr, C. J. (1993). Cell 73, 1059-1065. Sidow, A. (1996).Curr. Opin. Genet. Dev. 6, 715-722. Sitlani, A., and Crothers, D. M. (1998).Proc. Nutl. Acud. Sci. U.S.A. 95, 1404-1409. Sitzmann, J., Noben-Trauth, K., and Klempnauer, K. H. (1995). Oncogene 11,2273-2279. Sitzmann, J., Noben-Trauth, K., Kamano, H., and Klempnauer, K. H. (1996). Oncogene 12, 188 9- 1894. Sleeman, J. P. (1993). Oncogene 8,1931-1941. Smarda, J., and Lipsick, J. S. (1994).Oncogene 9,237-245. Smarda, J., Sugarman, J., Glass, C., and Lipsick, J. (1995).Mol. Cell. Biol. 15,2474-2481. Sterneck, E., Muller, C., Katz, S., and Leutz, A. (1992).E M B O J . 11, 115-126. Stober-Gasser, U., and Lipsick, J. S. (1988).J. Virol. 62, 1093-1096. Stober-Grasser, U., Brydolf, B., Bin, X., Grasser, F., Firtel, R. A., and Lipsick, J. S. (1992).Oncogene 7,589-596. Strobl, L. J., and Eick, D. (1992).E M B O J . 11,3307-3314. Sullivan, J., Feeley, B., Guerra, J., and Boxer, L. M. (1997).J. Biol. Chem. 272, 1943-1949. Symonds, G., Klempnauer, K. H., Evan, G. I., and Bishop, J. M. (1984). Mol. Cell. Biol. 4, 2587-2593. Symonds, G., Quintrell, N., Stubblefield, E., and Bishop, J. M. (1986).I. Virol. 59, 172-175. Tanaka, Y., Nomura, T., and Ishii, S. (1997).FEBS Lett. 413, 162-168. Tanikawa, J., Yasukawa, T., Enari, M., Ogata, K., Nishimura, Y., Ishii, S., and Sarai, A. (1993). Proc. Nutl. Acad. Sci. U.S.A. 90,9320-9324. Tashiro, S., Takemoto, Y., Handa, H., and Ishii, S. (1995).Oncogene 10, 1699-1707. Tavner, F. J., Simpson, R., Tashiro, S., Favier, D., Jenkins, N. A., Gilbert, D. J., Copeland, N. G., Macmillan, E. M., Lutwyche, J., Keough, R. A., Ishii, S., and Gonda, T. J. (1998). Mol. Cell. Biol. 18, 989-1002. Taylor, D., Badiani, P., and Weston, K. (1996).Genes Dev. 10,2732-2744. Thompson, M. A., and Ramsay, R. G. (1995).BioEssays 17,341-350. Thompson, C. B., Challoner, P. B., Neiman, P. E., and Groudine, M. (1986).Nature (London), 319,374-380. Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994).Nucleic Acids Res. 22,4673-4680. Thompson, M. A., Flegg, R., Westin, E. H., and Ramsay, R. G. (1997). Oncogene 14, 17151723. Todokoro, K., Watson, R. J., Higo, H., Amanuma, H., Kuramochi, S., Yanagisawa, H., and Ikawa, Y. (1988).Proc. Natl. Acud. Sci. U.S.A. 85,8900-8904. Tong, X., Drapkin, R., Yalamanchili, R., Mosialos, G., and Kieff, E. (1995).Mol. Cell. Biol. 15,4735-4744.
60
Brigitte Ganter and joseph S. Lipsick
Torelli, G., Selleri, L., Donelli, A., Ferrari, S., Emilia, G., Venturelli, D., Moretti, L., and Torelli, U. (1985).Mol. Cell. Biol. 5,2874-2877. Torelli, G., \'enturelli, D., Colo, A., Zanni, C., Selleri, L., Moretti, L., Calabretta, B., and Torelli, U. (1987).Cancer Res. 47,5266-5269. Toscani, A., Mettus, R. V., Coupland, R., Simpkins, H., Litvin, J., Orth, J., Hatton, K. S., and Reddy, E. P. (1997).Nature (London) 386, 713-717. Toth, C. R., Hostutler, R. F., Baldwin, Jr., A. S., and Bender, T. P. (1995).J. Biol. Chem. 270, 7661-7671. Trauth, K., Mutschler, B., Jenkins, N. A., Gilbert, D. J., Copeland, N. G., and Klempnauer, K. H. (1994).EMBO]. 13,5994-6005. Travali, S., Reiss, K., Ferber, A., Petralia, S., Mercer, W. E., Calabretta, B., and Baserga, R. (1991).Mol. Cell. Biol. 11, 731-736. Tsukiyarna, T., Daniel, C., Tamkun, J., and Wu, C. (1995).Cell 83, 1021-1026. Urbanek, P., Dvorak, M., Bartunek, P., Pecenka, V., Paces, V., and Travnicek, M. (1988).Nucleic Acids Res. 16, 11521-11530. Vellard, M., Sureau, A., Soret, J., Martinerie, C., and Perbal, B. (1992).Proc. Natl. Acad. Sci. U.S.A. 89, 2511-2515. Vorbrueggen, G., Kalkbrenner, F., Guehrnann, S., and Moelling, K. (1994).Nucleic Acids Res. 22,2466-2475. Vorbrueggen, G., Lovric, J., and Moelling, K. (1996).Biol. Chem. 377, 721-730. Wallrapp, C . , Muller-Pillasch, F., Solinas-Toldo, S., Lichter, P., Friess, H., Buchler, M., Fink, T., Adler, G., and Gress, T. M. (1997).Cancer Res. 57, 313.5-3139. Wang, W., Xue, Y., Zhou, S., Kuo, A., Cairns, B. R., and Crabtree, G. R. (1996). Genes Dev. 10,2117-2130. Watson, R. J. (1988).Oncogene 2,267-272. Watson, R. J., Dyson, P. J., and McMahon, J. (1987).E M B O J . 6, 1643-1651. Watson, R. J., Robinson, C., and Lam, E. W. (1993).Ntrcleic Acids Res. 21,267-272. Westin, E. H., Gallo, R. C., Arya, S. K., Eva, A., Souza, L. M., Baluda, M. A,, Aaronson, S. A,, and Wong-Staal, F. (1982).Proc. Natl. Acad. Sci. U.S.A. 79,2194-2198. Weston, K. (1992).Nucleic Acids Res. 20, 3043-3049. Weston, K., and Bishop, J. M. (1989).Cell 58, 85-93. Wolff, L. (19961. Crit. Rev. Oncogen. 7,245-260. Wolff, L., Mushinski, J. E, Shen-Ong, G. I.., and Morse, H. C. D. (1988).J. Immunol. 141, 681-689. Woo, C. H., Sopchak, L., and Lipsick, J. S. (1998).]. Virol. 72,6813-6821. Woodgett, J. R. (1991).Trends Biochem. Sci. 16, 177-181. Worpenberg, S., Burk, O., and Klempnauer, K. H. (1997).Oncogene 15,213-221. Wu, H.-M., and Crothers, D. M. (1984). Nature (London) 308,509-513. Yamamoto, K., Tojo, X., Aoki, N., and Shibuya, M. (1993).Jpn.J.Cancer Res. 84,1136-1144. Yoshinaga, S . K., Peterson, C. I.., Herskowitz, I., and Yarnarnoto, K. R. (1992).Science 258, 1.598-1604. Ziebold, U., and Klempnauer, K. H. (1997).Oncogene 15,1011-1019. Ziebold, U., Bartsch, O., Marais, R., Ferrari, S., and Klempnauer, K. H. (1997).Curr. Biol. 7, 253-260. Zwicker, j., Liu, N., Engeland, K., Lucibello, F. C., and Muller, R. (1996). Science 271, 1595-1597. Zwijsen, R. M., Wientjens, E., Klompmaker, R., van der Sman, J., Bernards, R., and Michalides, R. J. (1997).Cell 88,405-415.
cmSrc, Receptor Tyrosine Kinases, and Human Cancer JacquelineS. BLcardi,*David A. Tice,* and Sarah J. Parsons? Department of Microbiology and Cancer Center University of Virginia Health Sciences Center Charlottesville, Virginia 22908
I. Introduction 11. Receptor Tyrosine Kinases and Human Cancers A. Hepatocyte Growth FactorlScatter Factor Receptor B. Colony-Stimulating Factor-1 Receptor C. Fibroblast Growth Factor Receptors D. Platelet-Derived Growth Factor Receptor E. Epidermal Growth Factor Receptor F. HER2lneu G. HER Family Members and Estrogen Receptor Interactions 111. c-Src and c-Src Family Members in Human Cancers A. c-Src Structure and Mechanisms of Regulation B. Evidence for the Involvement of c-Src in Human Cancers C. c-Src Family Members and Human Cancers D. Nonreceptor Tyrosine Kinases Related to c-Src Family Members and Human Cancers IV. Mechanisms of c-Src Action A. Evidence for Involvement of c-Src in Signaling through Receptor Tyrosine Kinases B. Targets of c-Src V. Potential Therapeutic Applications of c-SrclHER1 Interactions References
I. INTRODUCTION Since the discovery that tyrosine kinases are among the transforming proteins encoded by oncogenic animal retroviruses, it has been speculated that this family of enzymes may contribute to the development of human malignancies. However, evidence supporting that hypothesis has been slow to evolve, largely because early emphasis was placed on examining human tumors for genetic alterations in protooncogenes encoding these enzymes. Such alterations have proved rare or nonexistent. Instead, investigations have fo‘Equal contributions were made by these authors. tTo whom correspondence may be addressed. Advances in CANCER RESEARCH 0065-23OW99 $30.00
Copyright 0 1999 by Academic Press. All rights of reproduction in any form reserved.
62
lacqueline S. Biscardi et al.
cused on determining levels of expression and posttranslational mechanisms of regulation of these proteins, particularly as they relate to signaling pathways that modulate growth, adhesion, invasion, and motility. Two classes of tyrosine kinases have emerged as potentially important players in promoting the evolution of human tumors: receptor kinases (RTKs) and nonreceptor tyrosine kinases of the c-Src family. Elevated levels of both these classes of tyrosine kinases can be found in a large number of tumors in a strikingly similar pattern of aberrant cooverexpression, suggesting that the two families may cooperate with one another during oncogenesis. Indeed, in model tissue culture systems, overexpression of receptor alone can result in malignant transformation when a continuous source of ligand is provided. However, overexpression of c-Src alone is non- or weakly oncogenic. These results indicate that c-Src, if it plays a role in tumorigenesis, most likely mediates its effects through RTKs. Demonstrations that c-Src physically associates with a number of RTKs in a ligand-dependent fashion provided some of the first evidence for functional cooperativity between these families of proteins. Subsequent studies showed that in complex, the two kinases reciprocally affect one another’s behavior, such that c-Src can be regarded both as a regulator of RTKs and as a cotransducer of signals emanating from them. c-Src is capable of physically associating with the receptors for platelet-derived growth factor (PDGF), prolactin, epidermal growth factor (EGF),colony-stimulating factor-1 (CSF-l), fibroblast growth factor (FGF), and hepatocyte growth factor/scatter factor (HGF/SF), as well as with the HER2/neu and Sky tyrosine kinases (this review and Toshima et al., 1995; Berlanga et al., 1995), all of which are postulated to play a role in the genesis and/or progression of various human cancers. Although c-Src and its family members are also known to participate in signaling events elicited by heterotrimeric G protein-coupled receptors (Malarkey et al., 1995) and neuronal ion channels (Ely et al., 1994; Holmes et al., 1996; Yu et al., 1997; van Hoek et al., 1997), this review focuses on the interactions of c-Src and Src family members with RTKs because of the growing documentation of the interactions between these proteins in human malignancies. First, a summary is presented, naming the RTKs that are most frequently implicated etiologically in human cancers and that have been shown to interact with c-Src. This summary includes a short review of the physical characteristics of the receptors, their molecular mechanisms of signaling, and their putative roles in specific cancers. Second, evidence is discussed for the involvement of c-Src and Src family members in human tumor development, and third, a synopsis is outlined showing the molecular mechanisms by which c-Src and its family members have been found to interact with receptors and other targets. Finally, we will speculate on the prospects for developing novel therapies based on these interactions.
63
c-Src and Receptor Tyrosine Kinases
11. RECEPTOR TYROSINE
A N D HUMAN CANCERS
KINASES
Figure 1 depicts the structural features of several classes of RTKs that interact with c-Src. All consist of an extracellular ligand-binding domain that bears motifs characteristic of the type of receptor (e.g., repeated immunoglobulin-like motifs for the PDGF and FGF receptors or cysteine-rich motifs in the EGF family of receptors), a transmembrane segment, a tyrosine
Fig. 1 Structures of receptor tyrosine kinase families known to associate with c-Src. All receptors are transmembrane glycoproteins that function as receptors for polypeptide growth factors. Structurally, these molecules are composed of large, extracellular domains that exhibit characteristic ligand-binding motifs, as well as transmembrane, juxtamembrane, catalytic, and C-terminal domains. Ligand binding induces dimerization, enzymatic activation, and autophosphorylation on specific tyrosine residues in the C-terminal domains. These phosphorylated tyrosine residues serve as docking sites for signaling molecules that transmit biological signals from the extracellular milieu to the nucleus. In the platelet-derived growth factor (PDGF) and fibroblast growth factor (FGF) receptor families, the kinase domain is interrupted by an insert that contains additional docking sites. EGFR, Epidermal growth factor receptor; CSF-IR, colony-stimulating factor-1 receptor; HGF/SFR, hepatocyte growth factor/scatter factor receptor; HER, human epidermal growth factor receptor.
64
lacqueline S.Biscardi et al.
kinase catalytic domain, and a carboxy-terminal region that contains sites of autophosphorylation. Binding of ligand causes dimerization of the receptor, activation of tyrosine kinase activity, and (trans) autophosphorylation of specific C-terminal tyrosine (Tyr) residues (reviewed in Heldin, 1996; Weiss et al., 1997), which in turn serve as docking sites for a variety of signaling molecules that contain SH2 domains (Pawson and Schlessinger, 1993), including phospholipase Cy (PLCy), phosphatidylinositol-3 kinase (PI-3) kinase), GTPase-activating protein of Ras (RasGAP), phosphotyrosine phosphatases (PTPases), Janus kinasedsignal transducers and activators of Transcription (JAK/STATS), adapter proteins (including Shc, Grb, Nck), and members of the c-Src family of tyrosine kinases (reviewed in Erpel and Courtneidge, 1995; Heldin, 1996). Signals are subsequently transmitted to the nucleus via several pathways, including the JAK/STAT and the Grb2/SOS/Ras/Raf/MEK/MAP kinase cascades (reviewed in Bonfini et al., 1996; Denhardt, 1996). Members of the STAT and MAP kinase families translocate from the cytoplasm to the nucleus and induce changes in gene expression, which bring about a variety of functional outcomes, such as mitogenesis, morphogenesis, and motility. The contribution of c-Src to downstream signaling from these RTKs has been the subject of growing interest, with emphasis on how c-Src may contribute to transformation and maintenance of the cancerous phenotype that is dependent on and induced by the receptors. In this treatise, a total of five RTK families and their putative roles in development of malignancy will be considered. The first four, receptors for HGFISF, CSF-1, FGF, and PDGF, are implicated as etiological agents in a wide variety of human cancers, and their ability to influence processes such as cytoskeletal changes, cell motility, and angiogenesis are thought to contribute to the metastatic potential of tumors. The fifth group, members of the EGF receptor family (HERl-4), will be discussed in the context of breast cancer, along with the estrogen receptor. This steroid hormone receptor plays a pivotal role in the etiology of breast cancer and growing evidence indicates its ability to reciprocally interact with c-Src and members of the HER family of RTKs.
A. Hepatocyte Growth FactodScatter Factor Receptor The Met tyrosine kinase is the receptor for hepatocyte growth factor/scatter factor (Bottaro etal., 1991; Naldini et al., 1991). This receptor was first identified as the product of the human oncogene, tpr-met, which was isolated from a chemically treated human cell line by the NIH3T3 gene transfer method (Cooper et al., 1984; Park et al., 1987). The normal cellular receptor is composed of two subunits, a 145-kDa p chain, which spans the cell
c-Src and Receptor Tyrosine Kinases
65
membrane and possesses ligand-binding and tyrosine kinase activity, and a 50-kDa a chain, which resides extracellularly and is covalently bound to the p subunit through disulfide linkages (Gonzatti-Haces et al., 1988). Related family members include the Sea and Ron RTKs (Ronsin et al., 1993; Huff et al., 1993).Each member of the Met family possesses two tandemly arranged, degenerate YVH/NV motifs in the C-terminal tail of the receptor, which are capable of binding the SH2 domains of the signaling molecules PI-3 kinase, PTPase 2, PLCy, c-Src, and Grb2/Sos (Ponzetto et al., 1994). Mutations in these motifs (H1351N) result in increased transforming ability but decreased metastasis (Giordano etal., 1997), a phenomenon that is linked to the creation of an additional Grb2 binding site and hyperactivation of the Ras pathway. HGF/SF is produced by cells of mesodermal origin and acts on epithelial and endothelial cells, eliciting numerous biological responses, including cell motility, growth, morphogenesis, differentiation, and angiogenesis (Kan et al., 1991; Rubin et al., 1991; Halaban et al., 1992).Which response is elicited in part depends on the cell type, developmental stage, and tissue context (Weidner et al., 1993; Kanda et af., 1993; Zhu et al., 1994; Rosen and Goldberg, 1995; Grano et al., 1996). For example, Met signals through STAT3 to induce the formation of branched tubule structures in Madin-Darby canine kidney (MDCK) cells, a hallmark of angiogenesis (Boccaccio et ul., 1998). HGF binding to primary human osteoclasts and osteoblasts triggers receptor kinase activity and autophosphorylation in both cell types. However, in osteoclasts, HGF binding is accompanied by increased levels of intracellular calcium, activation of c-Src, changes in cell shape, stimulation of chemotaxis, and DNA replication, whereas osteoblasts respond simply by undergoing DNA synthesis (Grano et al., 1996). Furthermore, osteoclasts also express HGF, but osteoblasts do not. This finding suggests that an autocrine loop may be responsible for signaling in osteoclasts, whereas a paracrine mechanism is functional in osteoblasts. HGF/SF and the Met receptor have been implicated in several types of human cancer. Met is overexpressed in gastric, ileal, colorectal, and thyroid papillary carcinomas, as well as in osteogenic sarcoma (Di Renzo et al., 1991, 1992; Rosen et al., 1994; Grano et al., 1996). The level of Met expression, as measured by intensity of Met immunofluorescence, has also been shown to correlate with grade of malignancy in primary human brain tumors (Koochekpour et al., 1997). In the case of ovarian carcinoma, Met levels can be regulated by the cytokines interleukin la (IL-la),IL-6, and tumor necrosis factor a (TNFa),thereby providing a physiological mechanism by which overexpression of Met can be achieved (Moghul et al., 1994). Approximately 14% of patients with papillary renal carcinoma have germ-line alterations in the Met receptor (Schmidt et al., 1997). Receptors bearing these mutations have been shown in NIH3T3 cells to result in increased tyrosine kinase activity of the receptors and Met-mediated focus formation
66
lacqueline S . Biscardi et d.
and tumors in nude mice, thus providing direct evidence for the ability of mutationally altered Met to function as an oncogene (Jeffers et al., 1997). The ability of HGF/SF to “scatter” cells and to increase their motility is strongly suggestive of a role for this ligand in tumor cell invasion. Indeed, several lines of evidence link HGF/SF to stimulation of the urokinase plasminogen activator (UPA) system, a cascade of proteases thought to promote release, extravasation, and migration of tumor cells. That the UPA cascade is critical for cell migration is supported by the findings that UPA -/- mice are unable to recruit migrating cells in response to inflammation (Gyetko et al., 1996), and do not support the growth and metastasis of experimental melanomas (Min et al., 1996). Shapiro et al. (1996) also showed that blocking interaction of UPA with its receptor results in decreased angiogenesis and tumor spread. The link between UPA and HGF was made when Jeffers et al. ( 1996b) reported that stimulation of the urokinase proteolytic system occurred concomitantly with HGF/SF-induced invasion and metastasis of human tumor cells. Rosen and Goldberg (1995) also demonstrated that HGF/SF is capable of stimulating angiogenesis in a rat cornea neovascularization assay. Together, these studies provide compelling evidence that HGF/SF are capable of promoting tumor progression by enhancing invasion and angiogenesis. Further evidence for a role for HGF/S/Met receptor in tumor invasiveness and angiogenesis comes from the findings that high titers of HGF/SF in invasive breast cancers are factors for relapse and death (Yamashita et al., 19941, that HGF/SF treatment of glioma cell lines stimulates proliferation and invasion (Koochekpour et al., 1997), and that invasive bladder carcinomas possess higher HGFISF titers than d o noninvasive cancers (Joseph et al., 199s). In addition, the Met receptor is overexpressed in several types of tumor stroma, including bladder wall, vascular smooth muscle, and vascular endothelial cells (Rosen and Goldberg, 1995),suggesting a paracrine signaling mechanism between tumor cells and the underlying stroma. Thus, HGF/SF and Met interactions may promote metastasis by enhancing proliferation via autocrine or paracrine routes, stimulating the expression of plasminogen activators, and triggering angiogenesis. (Rong et al., 1992; Kanda et al., 1993; Bellusci et a/., 1994; Jeffers et al., 1996a,b).
B. Colony-Stimulating Factor- I Receptor c-Fms, the cellular homolog of the viral oncogene v-Fms (Sherr et al., 198S), is the receptor for colony-stimulating factor-1, which stimulates the proliferation and differentiation of macrophages, osteoclasts, and placental trophoblasts (Sherr, 1990; Roth and Stanley, 1992; Insogna et al., 1997). That CSF-1 is critical for the development of mononuclear phagocytes was
c-Src and Receptor Tyrosine Kinases
67
shown by studies in mi,ce that fail to express functional CSF-1: these mice exhibit an osteopetrotic phenotype and lack osteoclasts and macrophages (Wiktor-Jedrzejczaket al., 1990, 1991). The CSF-1 receptor is expressed in placenta (Pollard et al., 1987; Regensstreif and Rossant, 1989; Hume et al., 1997),osteolasts (Insogna etal., 1997), and cells of monocyte lineage (Woolford et al., 1985),whereas the ligand, CSF-1, is produced by fibroblasts, myoblasts, osteoblasts, bone marrow stromal cells, and endothelial cells (Sherr, 1990; Roth and Stanley, 1992). Such independent distribution of ligand and receptor underlies the importance of cell-cell interactions in regulating receptor function. c-Fms bears sequence and structural similarity to the steel receptor, c-Kit, and to the receptors for FGF, PDGF, and Flt3/FLK2 (Hanks etal., 1988; Rosnet and Birnbaum, 1993).The unique feature of this group is that each member possesses an “insert” region within its kinase domain. The downstream targets of c-Fms include PI-3 kinase, STAT 1, and PLCy, all of which bind to phosphorylated tyrosine residues within the kinase insert portion of the molecule (Varticovskiet al., 1989; Shurtleff et al., 1990; Reedijk et al., 1990; Novak et al., 1996; Bourette et al., 1997). Bourette et al. (1997)have shown that sequential activation of the PI-3-kinase-dependent and PLCy-dependent signaling pathways is required to initiate the differentiation process of myeloid cells. c-Src has also been shown to associate with c-Fms and to be activated on binding of CSF-1 to the receptor. Complex formation between c-Src and c-Fms is thought to occur via the SH2 domain of c-Src and a juxtamembrane phosphotyrosyl residue on the receptor (Courtneidge et al., 1993; Alonso et al., 1995). In osteoclasts, phosphorylation of c-Src in response to CSF-1 stimulation occurs concomitantly with rearrangements of the actin cytoskeleton and spreading of the cells, suggesting that c-Src may be involved in regulating these processes (Insogna et al., 1997). Overexpression of c-Fms in NIH3T3 or Rat2 fibroblasts or in various tumor cells results in transformation, growth in soft agar, and tumor formation in nude mice (Rettenmeier et al., 1987; Taylor et al., 1989; van der Geer and Hunter, 1989; Favot et al., 1995). These findings demonstrate the oncogenic potential of overexpressed c-Fms. As described above, c-Fms and CSF1 are normally not expressed in the same cell type. However, coexpression is seen in tumors of the pancreas, endometrium, stomach, lung, and breast, and in acute myeloid leukemia, hairy cell leukemia, and Hodgkin’s lymphoma (Rambaldi et al., 1988; Kacinski et al., 1990; Paietta et al., 1990; Baiocchi et al., 1991; Kauma et al., 1991; Bruckner et al., 1992; Filderman et al., 1992; Storga et al., 1992; Tang et al., 1992; Leiserowitz et al., 1993; Till et al., 1993; Burthem et al., 1994; Berchuck and Boyd, 1995). Coexpression correlates with poor patient prognosis, most likely due to the establishment of an autocrine loop (Kacinski et al., 1990; Tang et al., 1992). Evidence suggests that such an autocrine loop contributes not only to tumor
68
Jacqueline S. Biscardi et al.
cell proliferation but also to invasiveness (Bruckner et al., 1992; Filderman et al., 1992; Burthem et al., 1994). In this regard, coexpression of c-Fms and its ligand in endometrial cancers correlates with a more advanced stage and with increased myometrial invasion (Leiserowitz et al., 1993). Moreover, CSF-1 stimulation results in the expression of UPA in lung tumors, Lewis lung carcinoma cells, and in NIH3T3 cells transfected with c-Fms (Filderman et al., 1992; Favot et al., 1995; Stacey etal., 1995). Together, these findings suggest that, like HGF/SF/MetR, deregulation of c-FmsKSF-1 interactions has the potential of contributing to the metastatic process in a variety of human cancers.
C. Fibroblast Growth Factor Receptors The FGF receptors comprise a large family that is encoded by four separate genes, each oi which can be alternatively spliced. Each receptor is also capable of binding several different ligands, resulting in a complex array of possible receptor/ligand pairs (Johnson and Williams, 1993). All receptors for FGF possess extracellular ligand-binding domains, which contain immunoglobulin-like repeats, and bipartite, intracellular tyrosine kinase domains (Lappi, 1995).Which signaling molecules are recruited varies with cell type and receptor/ligand pair. For example, in NIH3T3 cells (Zhan et al., 1994) FGFR 1 and c-Src physically associate following ligand binding, and activation of the receptor triggers the c-Src-dependent phosphorylation of the actin-binding protein, cortactin. Because cortactin is localized to cortical actin, particularly at the leading edge of a migrating cell (Wu et al., 1991; Maa et al., 1992; Wu and Parsons, 1993), its phosphorylation is speculated to influence cell motility and invasiveness. In other studies, ligand stimulation of FGFR 1 and FGFR 3 on C6 rat myoblasts results in activation of the p21Ras and MAPK pathway (Klint et al., 1995; Kanai et al., 1997). In these same cells, activation of the FGFR 3 receptor alone causes an increase in phosphorylation of PLCy but a decrease in c-Src phosphorylation (Kanai et al., 1997). FGF receptors are ubiquitously expressed during embryogenesis, but their presence is restricted after birth (Wanaka et al., 1991; Peters et al., 1992, 1993; Pastone et al., 1993). As a family, FGFs have mitogenic, nonproliferative, and antiproliferative effects. Which response is elicited is determined by the ligand, the type of cell exposed to the ligand, and the particular isoform of the receptor expressed on that cell (Schweigerer et al., 1987; Sporn and Roberts, 1988). For example, FGF 2 promotes survival of cultured neurons (Walicke, 1988), whereas FGF 1 and FGF 2 stimulate growth of fibroblasts, oligodendrocytes, astrocytes, smooth muscle cells, endothelial
c-Src and Receptor Tyrosine Kinases
69
cells, and retinal epithelial cells (Burgess and Maciag, 1989). FGFs can also act as chemotactic factors for fibroblasts and glial cells (Senior et af., 1986). Basic FGF (bFGF, or FGF 2) induces neurite outgrowth in embryonic chick ciliary ganglion cells (Schubert et al., 1987) and can mediate cellular migration in experimental systems (Sato and Rifkin, 1988).Treatment of cultured vascular endothelial cells with FGF 2 induces the formation of blood capillary-like tubules, a finding that suggests FGFs may play a role in angiogenesis (Montesano et al., 1986; Slavin, 1995). In this regard, a large literature is beginning to accumulate in support of a role for FGFs in angiogenesis, because they have been demonstrated to stimulate endothelial cell division, migration, release of proteolytic enzymes, and capillary formation (Slavin, 1995). In addition to these functions in normal cells, FGFR family members are implicated in the progression of a variety of human cancers. FGFs are thought to act as autocrine growth factors for melanomas, gliomas, and meningiomas (Lappi, 1995), and their levels are elevated in many different tumor types (Nguyen et al., 1994).FGF receptors are also overexpressed in human tumors. For example, 10% of human breast tumors exhibit amplifications of chromosomal regions encoding FGF receptors (Adnane et af., 1991), and FGFR 4 mRNA levels are frequently elevated in breast cancer cells as compared to normal tissue (Lehtola et af., 1993; Ron et al., 1993; Penault-Llorca et al., 1995). Some evidence also suggests that differential expression of FGFR isoforms can influence the propensity of a cell to undergo malignant transformation. In normal fetal and mature brain, FGFR 1, which possesses three immunoglobulin-like extracellular repeats, is expressed. However, in astrocytic tumors, an increase in the expression of an FGFR with two immunoglobulin-like domains is observed. This form has increased affinity for acidic and basic FGF (Shing et al., 1993). Changes in FGFR expression also occur during the conversion of normal or hyperplastic prostatic epithelium to malignant tumor tissue, where the increased expression of an alternatively spliced form of FGFR 2, which has a higher affinity for bFGF, appears to create an autocrine stimulatory loop (Wang et af., 1995). FGFs, along with other factors, are often secreted by tumors, and their increased extracellular abundance is linked to enhanced invasiveness (Klagsbrun et af., 1976; Libermann et af., 1987; Wadzinski et af., 1987; Folkman et al., 1988). In breast tumor cells, FGFR 4 activation results in membrane ruffling, a morphological change that is associated with metastasis (Johnston et al., 1995). In in vitro invasion assays, FGF 2 induces the migration of bovine capillary endothelial cells through placental tissue in a dose-dependent manner (Mignatti et al., 1989), and bFGF stimulates production of metalloproteinases in human bladder cancer cell lines, an event associated with increased invasiveness of the cells (Miyake et al., 1997). Moreover,
70
JacquelineS . Biscardi et al.
bFGF-dependent, sustained activation of MAPK correlates with the scattering of neuroepithelioma cells (van Puijenbroek et al., 1997).Together, these studies suggest that FGFs and FGFRs play important roles in human cancer progression by promoting the metastatic process.
D. Platelet-Derived Growth Factor Receptor The PDGFR has two isoforms, a and p, which differ in their preferences for binding homo- or heterodimers of the A and B forms of the PDGF ligand (Yarden et al., 1986; Claesson-Welsh et al., 1989; Claesson-Welsh and Heldin, 1989; Heldin and Westermark, 1990; Ross et al., 1990; Matsui et al., 1993). Both receptor isoforms consist of an extracellular domain that contains immunoglobulin-like motifs, transmembrane and juxtamembrane regions, a catalytic domain with an insert, and a C-terminal tail (Heldin and Westermark, 1990; Ross et al., 1990). Signaling molecules, which include PI-3 kinase (Kazlauskas and Cooper, 1989; Auger et al., 1989; Coughlin et al., 1989), PLCy (Kumjian et al., 1989; Meisenhelder et al., 1989; Wahl et al., 1989; Morrison et al., 1990), RasGAP (Molloy et al., 1989; Kaplan et al., 1990; Kazlauskas et al., 1990), and the Src family members c-Src, Fyn, and c-Yes (Kypta et al., 1990) (Twamley et al., 1992), bind phosphorylated tyrosine residues in the C-terminal tail, the kinase insert, and the juxtamembrane region via their SH2 domains. Interestingly, the same downstream effectors in different cell types can elicit different cellular responses. For example, in human hepatoma cell lines, PLCy and PI-3 kinase can independently transmit mitogenic signals (Valius and Kazlauskas, 1993), whereas in C H O cells a precise balance exists between migration-promoting signaling via PLCy and PI-3 kinase and migration-inhibitory signaling via RasGAP (Kundra et al., 1994). Phosphorylation of Tyr-988 in the carboxy terminus of the 01 receptor is associated with induction of chemotaxis, whereas phosphorylation of Tyr-768 and Tyr-1018 negatively regulates this process (Yokote et al., 1996). These results suggest that the different phosphorylation sites serve as binding sites for unique signaling molecules that influence cellular behavior in different ways. This hypothesis is further supported by studies in smooth muscle cells showing that PDCF-induced activation of PLCy is associated with actin disassembly and chemotaxis, whereas an independent signaling pathway, probably involving small GTPases such as Rho, appears to mediate the proliferative effect of PDGF in this system (Bornfeldt et al., 1995). PDGF receptors and their ligands regulate a wide spectrum of normal cellular processes in cells of mesenchymal and endothelial origin. These processes include differentiation, proliferation, survival, and migration. For example, the receptor for PDGF 01 is necessary for the development of neur-
c-Src and Receptor Tyrosine Kinases
71
a1 crest cells (Soriano, 1997) and alveolar branching in the lung (Souza et al., 1995);the PDGF p receptor is required for proper development of the cardiovascular and renal systems (reviewed in Betsholtz, 1995). The PDGF p receptor is also found in mesenchymal tissue of the developing trachea and intestine and in the endothelium of blood vessels, where it is thought to play a role in regulating mesenchymal-epithelial interactions (Shinbrot et al., 1994).In addition, the PDGF 01 receptor is required for the maximal chemotactic effect of PDGF on lung fibroblasts (Osornio-Vargas et al., 1996). Numerous studies suggest that various PDGF and PDGFR isoforms are also involved in the genesis or maintenance of human cancers. The PDGFR is overexpressed in human pancreatic cancer (Ebert et al., 1995), primary and metastatic melanomas (Barnhill et al., 1996), and in mesothelioma cell lines (Versnel et al., 1994; Langerak et al., 1996).PDGFR expression is also seen in many neural crest-derived human tumors, including neuroblastoma and Ewing’s sarcoma (Matsui et al., 1993), in basal cell carcinoma (Ponten et al., 1994), and in tumors of the lung and pituitary (Leon et al., 1994; Vignaud et al., 1994). PDGF and its receptors are not normally expressed in epithelial cells, but their aberrant expression in tumors of this origin suggest that they could be involved in the oncogenic process. The situation is made more complex by the fact that some tumors express one or both forms of the ligand and no receptor(s) or vice versa, suggesting that both autocrine and paracrine signaling loops are involved in PDGF-mediated growth of tumors. For example, autocrine signaling loops have been shown to contribute to the growth of human esophageal carcinomas (Juang et al., 1996), mesotheliomas (Langerak et at., 1996), malignant melanomas (Barnhill et ul., 1996), gliomas, and glioblastomas (Potapova et al., 1996). However, results from Coltrera et al. (1995) show that PDGF may also function in a paracrine fashion in some human breast tumors. Their studies revealed that PDGF p is expressed in breast epithelium and tumor tissues, and the receptor is present in stromal fibroblasts. A similar situation appears to exist in ovarian cancer (Versnel et al., 1994), in lung tumors (Vignaud et al., 1994), and in basal cell carcinomas (Ponten et al., 1994). The ability of PDGF to induce chemotaxis may also play a role in tumor cell metastasis. For example, expression of the receptor for PDGF 01 in Lewis lung carcinoma cells increases their metastatic potential, whereas expression of the receptor truncated at the kinase domain reverses this effect (Fitzer-Attas et al., 1997). Potapova et al. (1996) demonstrated that in human glioblastoma cells, which express both the PDGF p receptor and its ligand, further expression of PDGF p results in tumor formation in nude mice and increased metastasis. These examples support the idea that in addition to mediating normal cell migration, aberrant expression or activation of PDGF receptors in tumor cells can contribute to their proliferative and invasive properties.
72
Jacqueline S. Biscardi et nl.
E. Epidermal Growth Factor R e c e p t o r The human epidermal growth factor receptor, HER1, belongs to a family of human RTKs that includes HER2/neu, HER3, and HER4 (Ullrich and Schlessinger, 1990). All members of this family are transmembrane tyrosine kinases that possess an extracellular domain with two cysteine-rich repeats, an intact catalytic domain, and a C-terminal tail that binds SH2-containing signaling effectors on activation of the receptor. Ligands for HERl include epidermal growth factor (EGF), transforming growth factor-a (TGF-a), betacellulin (Riese et al., 1996), and epiregulin (Komurasaki et al., 1997). No specific ligand for HER2 has yet been defined, but HER3 and HER4 can be activated by a family of alternatively spliced ligands, called heregulins (HRG) (reviewed in Hynes and Stern, 1994). Each member of the HER family is capable of heterodimerizing with other members of the family, thereby providing a means by which HER2, although it lacks a ligand, can signal. Such dimerization appears to occur in a hierarchical order, wherein the HER2/3 interaction is the most preferred and the HER1/4 interaction the least preferred (Pinkas-Kramarski et al., 1996). Studies in 32D hematopoetic cells, which do not express any HER family members, show that heterodimers have more potent mitogenic activity than do homodimers and that HER3 heterodimers are the most transforming. However, when HERl is present, signaling through this receptor dominates over other members of the family (Pinkas-Kramarski et al., 1996). The focus of our discussion will be on the roles of HERl and HER2 in human cancer, because large bodies of literature exist for each. HER3 and HER4 are more recent additions to the family, and characterization of them with respect to their possible involvement in human cancers is just beginning. However, it is important to note that overexpression of HER3 has been detected in some breast cancers (Lemoine et al., 1992) and in papillary thyroid carcinomas (Faksvag et al., 1996). Thus, the possibility exists that homo- or heterodimerization of HER3 or HER4 with HERl or HER2 mediates tumorigenic signaling in a manner similar to that of HERl and HER2. A major role for HERl is its involvement in normal human development. It affects many stages, from postfertilization to sexual maturation. For example, HERl and its ligand, TGF-a, control proliferation of blastocoel cells as well as embryoluterine signaling and implantation (Rappoll et al., 1988; Dardik and Schultz, 1991; Arnholdt et al., 1991; Zhang et al., 1992).HERl is also necessary for development of embryonic lung, skin, palate (Lee and Han, 1990), and hair follicles (Hansen et al., 1997). During puberty, HERl and the estrogen receptor together regulate the differentiation of normal breast epithelium and uterine and vaginal growth (Nelson et al., 1991; Ignar-Trowbridge et al., 1992). Loss of control of these interactions is thought to play a role in the genesis of human tumors, and the diversity of tissues
c-Src and Receptor Tyrosine Kinases
73
that are regulated developmentally by HERl is reflected in the spectrum of tissues and cell types in which HERl is thought to play an oncogenic role. That HERl can function as an oncoprotein was demonstrated by the ability of NIH3T3 cells, engineered to overexpress HERl and held in the continual presence of EGF, to become transformed and develop tumors in nude mice (Velu et al., 1987; DiFiore et al., 1987a; Di Marco et al., 1989). HERl effector substrates include the adaptor proteins Shc (Pelicciet al., 1992; RuffJamison et al., 1993)and Grb2, which feed into the well-defined Ras/MAPK signaling pathway (Li et al., 1993; Egan et al., 1993; Rozakis-Adcock et al., 1993), as well as PLCy (Rhee, 1991), c-Cbl (Levkowitz et al., 1996), eps 8, and eps 15 (Fazioli et al., 1992,1993aYb). Reports from several laboratories show that on activation, HERl physically associates with the c-Src nonreceptor tyrosine kinase in both normal fibroblasts and in a variety of tumor cell lines (Luttrell et al., 1994; Maa et al., 1995; Sat0 et al., 1995; Stover et al., 1995; Biscardi et al., 1998a). Complex formation with c-Src occurs concomitantly with enhanced phosphorylation of receptor substrates, suggesting that c-Src may act to increase the receptor’s tyrosine kinase activity, thus enhancing the potential for cellular transformation and tumorigenesis (Maa et al., 1995; Tice et al., 1998; Biscardi et al., 1998a,b). This hypothesis was tested directly using a panel of C3HlOT1/2 murine fibroblasts that were engineered to overexpress HERl and c-Src,-either alone or in combination. Cells overexpressing both HERl and c-Src were found to produce synergistically larger and more numerous tumors in nude mice and colonies in soft agar than those produced by cells overexpressing either HERl or c-Src alone (Maa et al., 1995). These findings represent the first causal evidence for cooperativity between c-Src and HERl in tumorigenesis. What is the evidence for involvement of HERl in the genesis of human tumors? Aberrant expression, overexpression, or truncation of HERl has been demonstrated to occur in a variety of human cancers, including benign skin hyperplasia, glioblastoma, and cancers of the breast, prostate, ovary, liver, bladder, esophagus, larynx, stomach, colon, and lung (Harris et al., 1992; Khazaie et al., 1993; Scambia et al., 1995). In patients with ovarian cancer, overexpression of HERl correlates with a decreased response to chemotherapy and decreased survival (Scambia et al., 1995; Fischer-Colbrie et al., 1997), suggesting that HERl plays a proactive role in ovarian tumor progression. HERl overexpression also appears to play a role in the etiology of glioblastomas. Forty percent of glioblastomas exhibit amplification of the HERl gene (Khazaie et al., 1993), but in these tumors, overexpression is not the only abnormality regarding HER1. An alternatively spliced form of the receptor, termed EGFRvIII, is also frequently observed (Libermann et al., 1985; Yamazaki et al., 1988; Tuzi et al., 1991; Chaffanet et al., 1992). This form of the receptor lacks nucleotides 275-1075, which encode a large
74
lacqueline S. Biscardi et al.
portion of the extracellular domain (Humphrey et al., 1990; Ekstrand et al., 1992; Wong et al., 1992, and displays constitutive activity, perhaps due to its inability to be controlled by ligand (Ekstrand et al., 1994). Existing evidence suggests that EGFRvIII signals differently than wild-type receptor, prefering the PI-3 kinase pathway (Moscatello et al., 1998) to the Ras/MAPK pathway (Montgomery et al., 1995; Moscatello et al., 1998).In addition to glial tumors, one study showed that EGFRvIII is present in 16% of non-small-cell lung carcinomas (Garcia et al., 1993) as well as in 86% of medulloblastomas, 78% of breast cancers, and 73% of the ovarian cancers examined (Moscatello et al., 1995). In contrast, EGFRvIII has not yet been detected in normal tissue, a finding that provides compelling evidence for an oncogenic role for this form of the receptor. A link between HERl and breast cancer has also emerged in recent years. Amplification or overexpression of the genes encoding one or more of the HER family members is estimated to occur in approximately 67% of human breast cancers (Harris et al., 1992), with overexpression of HERl detected in approximately 30% of patients (Battaglia et al., 1988; Delarue et al., 1988; Bolla et al., 1990; Koenders et al., 1991; Toi et al., 1991; Harris et al., 1992). Elevated levels of HERl are also associated with loss of estrogen-dependent growth (Klijn et al., 1993), suggesting a role for HERl in the later stages of tumor progression. In addition to transformation and proliferation, studies from several laboratories suggest that HERl also enhances the invasive potential of tumor cells. Overexpression of HERl has been shown to result in an increased ability of rat mammary carcinoma cells to migrate through matrigel (Lichtner et al., 1995; Kaufmann et al., 1996), and higher levels of HERl are found in tumor tissue at metastatic sites as compared to primary sites (Sainsbury et al., 1987; Toi et al., 1991). Both these findings are supportive of a role for HERl in metastasis.
E HER2lneu Like HERl, activated HER2 possesses an intracellular tyrosine kinase domain as well as C-terminal phosphotyrosines that are capable of binding downstream substrates, such as PLCy, PI-3 kinase, Grb7, p120 RasGAP, p190, RhoGAP, c-Src, Shc, PTPlD, PTPlB, eps-8, and Tob, an antiproliferative protein (Hynes and Stern, 1994; Matsuda et al., 1996; Liu and Chernoff, 1997).Because both HERl and HER2 appear to activate similar downstream signaling pathways in experimental cell systems, it is unclear how specificity of signaling is achieved. The most likely explanation is that activation of a particular signaling pathway is dependent on cell type and on the
c-Src and Receptor Tyrosine Kinases
75
subset of HER family members and effector molecules available at any given time. However, a few examples of specific substrates have been reported, such as the c-Cbl adaptor protein for HERl (Levkowitz et al., 1996) and paxillin and a protein of 23 kDa (p23) for HER2 (Romano et al., 1994). HER2 is expressed in all tissues except the hematopoietic system (De Potter et al., 1990; Press et al., 1990). Studies using mice that are deficient in HER2, HER4, or the HER3/4 ligand, HRG, demonstrate that signaling through HER2 heterodimers is necessary for proper cardiac and neural development (Meyer and Birchmeier, 1995; Gassmann et al., 1995; Lee et al., 1995). A great deal of evidence from both experimental systems and human patients also points to the involvement of HER2 in malignant transformation. In certain tumors, it has been found that HER2 can be overexpressed up to 100 fold, due to gene amplification (Hynes and Stern, 1994).This finding, coupled with the fact that overexpression of HER2 alone, without the addition of agonist for HER family members, can induce focus formation in cultured fibroblasts (Hudziak et al., 1987; DiFiore et al., 1987b) suggests that overexpression of HER2 is capable of inducing oncogenic activity in the human. In addition, overexpression of HER2/neu in PC-3 prostate cancer cells has been shown to result in an increased incidence of metastasis after orthotopic introduction (Zhau et al., 1996). Whereas amplification of the gene encoding HER2 is found in 10-30% of breast, ovarian, and gastric tumors (Hynes and Stern, 1994), tumors of the lung, mesenchyme, bladder, and esophagus contain high levels of HER2 protein but no gene amplification, suggesting that both transcriptional and posttranscriptional mechanisms are responsible for increased HER2 levels (Kraus et al., 1987; Hynes et al., 1989; King et al., 1989; Kameda et al., 1990). HER2 is apparently involved in the genesis of many types of human tumors, but its role has been most well-characterized in breast cancer. Increased levels of HER2 protein appear to correlate with poor patient prognosis (Slamon et al., 1987, 1989; Paik et al., 1990; Gusterson et al., 1992) and a loss of responsiveness to the,antiestrogen, tamoxifen (Nicholson et al., 1990; Wright et al., 1992; Klijn et al., 1993). In transgenic mouse models, HER2/neu was demonstrated to induce mammary tumors when expression was targeted to the mammary gland by the use of the murine mammary tumor virus promotor (Muthuswamy et al., 1994). These HER2/neu tumors contain increased levels of c-Src and c-Yes kinase activity as compared to normal, surrounding tissue (Muthuswamy et al., 1994; Muthuswamy and Muller, 1995). Furthermore, c-Src was found to coimmunoprecipitate with HER2/neu (Muthuswamy et al., 1994), suggesting that c-Src cooperates with HER2 as well as with HERl in regulating malignant progression. Because HER2/neu is most frequently localized to the primary tumor mass in the murine model and is found in earlier stage in situ carcinomas in humans
76
jacqueline S. Biscardi et al.
(van de Vijver et al., 1988; Paik et al., 1990; Lin et al., 1992; Barnes et al., 1992), it is speculated that this molecule is involved in earlier stages of breast cancer than is HER1.
G. HER Family Members a n d Estrogen Receptor Interactions Increasingly compelling data are accumulating that point to interactions among the estrogen receptor (ER), HER1, HERZ, and c-Src as being major factors in the development of human breast cancer. The ER is a steroid hormone receptor of 6 7 kDa that dimerizes and becomes activated as a transcription factor on binding of estrogen (Mangelsdorf et al., 1995). Functional domains of the ER include an amino-terminal A/B region, which is responsible for ligand-independent transcriptional activation; a central DNA-binding domain; and a carboxy-terminal E/F hormone-binding domain, which is responsible for estradiol-induced transcription (Tsai and O’Malley, 1994; Beato et al., 1995). In addition to the well-characterized ER 01 isoform, a p isoform, which has differing transcriptional properties and expression patterns, has been discovered (Paech et al., 1997). Loss of ER responsiveness in human breast tumors correlates with overexpression of HERl and with a poorer patient prognosis (Fitzpatrick et al., 1984; Sainsbury et al., 1985; Dauidson et al., 1987; Nicholson et al., 1988). The mechanism by which a breast tumor cell loses responsiveness to estrogen is unclear, but this event may be regulated in part by interactions with HER family members and/or c-Src. Cross-talk between growth factor receptor tyrosine kinases and the ER was first demonstrated by IgnarTrowbridge and co-workers (1992, 1993), who showed that treatment of cells with EGF activates the transcriptional activity of the ER and that this effect is dependent on the amino-terminal A/B domain of the ER. The ER also appears to have the ability to affect expression of the EGF receptor. In ER-positive breast cancer cells, estradiol treatment increases HERl mRNA levels (Yarden et al., 1996). This effect may be directly mediated by the ER, because the HERl promoter has sequences that share loose homology with the estrogen response element (ERE)and can bind human ER (Yarden et al., 1996). Conversely, HERl can affect ER expression. Overexpression of TGF-cx in the ER-positive ZR75-1 breast cancer cell line, along with prolonged treatment of these cells with antiestrogens, results in loss of the ER, whereas treatment of parental ZR75-1 cells with antiestrogens alone has little effect (Clarke et al., 1989; Agthoven et al., 1992). These results are interpreted to mean that continual and concomitant stimulation of HERl and ER can cause a reduction in ER expression. In this regard, it is through that expres-
c-Src and Receptor Tyrosine Kinases
77
sion of HERl and ER and mutually exclusive, because most breast tumors that overexpress HERl lack functional ER (Fitzpatrick et al., 1984; Sainsbury et al., 1985; Davidson et al., 1987; Nicholson et al., 1988). Because breast tumors that do not show this inverse expression tend to be HERl/ER positive rather than HERUER negative, it has been suggested that overexpression of HERl precedes loss of the ER (Koenders et al., 1991; Dittadi et al., 1993; Chrysogelos and Dickson, 1994). It is unclear how overexpression or activation of HERl leads to loss of ER expression. One possible mechanism may involve signaling to MAP kinase. HERl activation results in the phosphorylation of the ER on Ser-118, a phosphorylation that is required for hormone-independent transcriptional activity of the ER (Kato et al., 1995). Ser-118 is also thought to be a target for MAP kinase, because studies using dominant negative Ras and MEK demonstrated a loss of this phosphorylation concomitantly with a loss of EGF-dependent transcriptional activation (Bunone et al., 1996). Autocrine stimulatory loops involving TGF-a and HERl are known to exist in breast cancer, thus it is speculated that the continued stimulation of the ER via the HERl/MAP kinase pathway leads to its down-regulation and eventual loss. Estradiol is also known to induce phosphorylation of the ER (Auricchio et al., 1987).In addition, Arnold et al. (1995a)reported that the ER is basally phosphorylated on Y537 in vivo. The role of the Y537 phosphorylation is controversial. Early studies showed that tyrosine phosphorylation of the ER activates its hormone-binding activity (Migliaccio et al., 1989) and that phosphorylation of Y537 is required for binding of the ER to the ERE (Arnold and Notides, 1995; Arnold et al., 1995b). Further in vitro studies demonstrated that c-Src is able to phosphorylate Y537 and that this phosphorylation is necessary for homodimerization of the ER and for binding of estradiol (Arnold et al., 1995a, 1997).In agreement with these findings, Castoria et al. (1996) reported that a non-hormone-binding form of the ER found in mammary tumors can be converted to a hormone-binding form by irt vitro phosphorylation with a calcium/calmodulin-regulated kinase, which is thought to be a c-Src family member. However, additional studies in which Y537 was mutated to various amino acids suggest that phosphorylation of Y537 per se is unnecessary for estradiol-mediated activation of the ER but may be important in ligand-independent (i.e., growth factor-mediated) activation (Weis et al., 1996; Lazennec et al., 1997). Although c-Src is capable of phosphorylating the ER, the ER may also influence c-Src activity. Estradiol treatment has been shown to increase c-Src tyrosine phosphorylation and kinase activity in MCF7 breast cancer cells (Migliaccio et al., 1993, 1996) and to stimulate kinase activity of c-Src and its related family member, c-Yes, in colon carcinoma cells (Di Domenico et al., 1996). Ligand-independent down-regulation of the ER may also be mediated by
78
JacquelineS . Biscardi et al.
HER2 signaling pathways. Pietras et al. (1995) showed that overexpression of HER2 in MCF7 cells leads to estrogen-independent growth and ERE transcriptional activation. Furthermore, treatment of these cells with HRG, which stimulates HER2-dependent signaling via HER2/3 heterodimers, induces tyrosine phosphorylation and down-regulation of the ER. Other investigators have shown that HRG treatment inhibits the expression of ER in ER-positive breast cancer cells and can revert the estradiol-mediated decrease in HER2 expression (Grunt et al., 1995).Taken together, these results suggest that in experimental cell systems, HER2 and the ER are expressed in a mutually exclusive manner. However, in human breast tumors, the situation is less clear, with some reports indicating an inverse correlation between HER2 and ER expression (Adnane et al., 1989; Borg et al., 3 990) and others indicting no such correlation (Slamon et al., 1989; Bacus et al., 1996).
111. c4rc AND c-Src FAMILY MEMBERS IN HUMAN CANCERS
A. c-Src Structure and Mechanisms of Regulation c-Src is the cellular, nontransforming homolog of v-Src, the oncoprotein encoded by the chicken retrovirus, Rous sarcoma virus. c-Src is a 60-kDa tyrosine kinase that is composed of six domains: an N-terminal membraneassociation domain, a “Unique” domain, SH3 and SH2 domains, a catalytic domain, and a negative regulatory domain (Fig. 2, see color plate). Although c-Src is cytosolic, it localizes to intracellular membranes, including the plasma membrane and membranes of endosomes and secretory vesicles within the cytosol (Parsons and Creutz, 1986; Kaplan et al., 1992; Resh, 1994). It is tethered to these membranes by the combined action of an N-terminal, covalently linked myristate moiety, salt bridges between basic amino acids in the N terminus and phosphates of the lipid backbone, and noncovalent interactions with integral or associated membrane proteins (Resh, 1994).Membrane localization of c-Src is required for its ability to participate in growth factor receptor-mediated signaling in normal cells (Wilson et al., 1989). The function of the Unique domain is not well-defined. However, based on the fact that it exhibits the greatest sequence divergence among family members of all the domains (Brown and Cooper, 1996), it is speculated to specify protein-protein interactions that are unique to individual Src family members. The SH3 and SH2 domains mediate the binding of c-Src with other signaling proteins through proline-rich or phosphotyrosine-containing regions on target proteins, respectively (Pawson and Schlessinger, 1993). The major regulatory region of the enzyme is a short domain at the extreme C terminus of
Fig. 2 Structure of c-Src. c-Src is the prototype of a large family of cytoplasmic tyrosine kinases that associate with cellular membranes through lipid modifications at their N termini. As a linear molecule, the relationship between the various domains can be seen: an N-terminal membrane association domain that contains the site of myristylation, a Unique domain that exhibits the widest sequence divergence among family members of any of the domains, an Srchomology-3 (SH3) domain that binds poly(pro1ine) motifs on target molecules, an Src-homology-2 (SH2) domain that binds phosphotyrosine residues on target molecules, an SH2/kinase linker, the catalytic domain, and the negative regulatory domain that contains the predominant site of tyrosine phosphorylation on the inactive molecule (Y527 in chicken, Y531 in human). The three-dimensional orientation of the molecule, lacking the membrane-association and Unique domains, is depicted as a ribbon diagram. Reprinted with permission from Nature, Xu et al. (1997), and from Michael Eck (configured from the atomic coordinates provided on the Web). Copyright 1997 Macmillan Magazines Limited. The enzymatic activity of c-Src is regulated by the coordinated effects of target proteins binding to or covalent, posttranslational modifications of the SH3, SH2, and negative regulatory domains on the catalytic domain, as described in the text.
This Page Intentionally Left Blank
c-Src and Receptor Tyrosine Kinases
79
the molecule, which harbors a Tyr residue that becomes phosphorylated (Y530 in human c-Src; Y527 in chicken c-Src) by a C-terminal Src kinase, CSK (Okada et al., 1991). Phosphorylated Y527/530 (pY527) is capable of binding its own SH2 domain in a manner that inhibits kinase activity without physically blocking the catalytic site, as shown in Fig. 2 (Yamaguchi and Hendrickson, 1996; Sicheri et al., 1997; Xu et al., 1997). Binding of tyrosine-phosphorylated cellular proteins to the SH2 domain is thought to destabilize the intramolecular pY527/SH2 domain interaction and induce a conformational change that results in enzymatic activation. Structural studies have revealed that the SH2 and SH3 domains collaborate in their binding of respective protein partners, thereby cooperatively influencing the activity of the enzyme (Eck et al., 1994). Furthermore, crystallographic analysis has shown that sequences just N terminal to the catalytic domain (termed the SH2-kinase linker) comprises a loop structure that functions as a “pseudo” SH3 binding site (Yamaguchi and Hendrickson, 1996; Sicheri et al., 1997; Xu et al., 1997). Together, the intramolecular phosphotyrosine/SH2 and linkerlSH3 interactions direct a conformation that presses the linker against the backbone of the catalytic domain and renders the protein inactive. As with the SH2 domain, binding of signaling proteins to the SH3 domain is thought to release the constraints of the linker/SH3 interaction on the kinase domain, resulting in activation of catalytic activity. Mutation of Y527 to F or deletion of the C-terminal regulatory domain (as in v-Src) results in a constitutively active protein that phosphorylates target proteins in an unregulated fashion and induces cellular transformation and oncogenesis (Cartwright et al., 1987; Kmiecik and Shalloway, 1987; Piwnica-Worms et al., 1987; Reynolds et al., 1987). In normal cells, c-Src is nononcogenic or only weakly so, even when it is overexpressed (Shalloway et al., 1984; Luttrell et al., 1988). However, under certain conditions (growth factor stimulation or translocation; outlined below), the enzyme can become activated, either via dephosphorylation of pY527 or by binding of signaling proteins to the N-terminal half of the protein. Activation is most frequently a transient event, and c-Src, in contrast to v-Src, is thought to respond to negative control by rephosphorylation of Y527 or by the release of binding proteins and the resumption of intramolecular interactions. It has been the conjecture of many investigators that the transient nature of c-Src activation often prevents its detection. In fact, the possibility exists that little or no activation above basal levels is necessary for catalysis, if the substrate is properly positioned near the catalytic cleft. Thus, another “regulator” of c-Src activity may well be its intracellular localization and, at a finer level, its appropriate juxtaposition to substrate within a signaling complex. Identification of c-Src substrates and proteins that bind its SH2 and SH3 domains is now critical for further understanding of the role c-Src and its family members play in biological processes.
80
jacqueline S.Biscardi et al.
The majority of the studies leading to the above model have been conducted in animal tissue culture systems and are just now being applied to the study of c-Src in human tumors. In the following section evidence is presented for the involvement of c-Src in the genesis of human tumors, with particular emphasis on its putative role in colon, breast, lung, and myeloid tumors.
B. Evidence for the Involvement of c-Src in Human Cancers Like the RTKs, many lines of evidence are suggestive of a role for c-Src in the genesis and progression of multiple types of human cancer. This evidence is both genetic and biochemical in nature and has been generated by studies of cultured tumor cell lines and surgically generated tumor tissue. Together these studies have implicated c-Src as an etiological agent for the development of neuroblastomas, myeloproliferative disorders (including myeloid leukemia), and carcinomas of the colon, breast, lung, esophagus, skin, parotid, cervix, and gastric tissues. Interestingly, although alterations of cSrc have been described at both the gene and protein levels in various cancer tissues, the changes are quite variable and include both increases and decreases in gene copy number and in protein levels and specific enzyme activities. Taken at face value, these findings suggest multiple ways in which c-Src can contribute to the oncogenic process, both as a dominantly acting oncogenic protein and as a negatively acting tumor suppressor. However, the multitude of changes could also reflect fortuitous alterations that do not contribute to the ultimate malignant phenotype. There may also be technical reasons for the variability in the findings, such as the different probes used for genetic analysis and the different antibodies and cell extraction conditions used for biochemical analysis. It is clear that further work needs to be done to clarify these issues and attempts made to minimize technical problems. Of particular importance to future studies will be the development and characterization of good animal and tissue culture models to test the hypotheses derived from analyses of human tumor tissues, whereby the contribution of individual genes or proteins can be evaluated for their oncogenic potential against a normal cell background rather than against a heterogeneous background of unknown numbers and types of genetic alterations that occur in every human tumor. 1 . GENETIC EVIDENCE
With the identification of the first protooncogenes came a plethora of studies examining the genomic content of multiple human tumors for deletions, amplifications, and diverse rearrangements in chromosomes containing pro-
c-Src and Receptor Tyrosine Kinases
81
tooncogenes. For the most part, these studies identified few if any gross changes in the c-Src gene, which maps to the q arm of chromosome 20. Furthermore, gene expression studies, employing a variety of techniques to measure steady-state levels and newly synthesized mRNA have also revealed few changes in c-Src-specific mRNA (Bishop, 1983; Slamon et al., 1984). These findings led many investigators to conclude that c-Src played a minor (if any) role in the genesis of human tumors. Not until researchers began examining protein levels and specific enzyme activities did evidence for the involvement of c-Src begin to emerge. However, there were a few exceptions to the general rule described above, and one in particular is noteworthy. Four groups have identified a deletion of 16-21 cM in the long arm of chromosome 20 [del(20q)] as a recurring, nonrandom abnormality in malignant myeloid disorders, including nonlymphocytic leukemia and polycythemia (Simpson, 1988; Roulston et al., 1993; Hollings, 1994; Asimakopolous et al., 1994). This deletion maps between 20q11.2 and 20q13.3, a region that encodes the c-Src protooncogene (Hollings, 1994). The notion that deletion of a chromosomal region is signatory for a tumor suppressor gene suggests that, if c-Src is a critical gene in this deletion, it behaves as a negative regulator of cell growth, not as a dominant oncogene, as is commonly believed. That c-Src may have some tumor suppressor-like characteristics in myeloid cells is supported by the finding of several groups (Barnekow and Gessler, 1986; Gee et al., 1986) that c-Src expression levels increase during myeloid differentiation. If c-Src plays a critical role in promoting differentiation and maintaining the postmitotic state, then loss of such an activity might permit cells to once again acquire proliferative activity-the hallmark of a tumor suppressor.
2. BIOCHEMICAL EVIDENCE By far the bulk of evidence supporting a role for c-Src in the development of human tumors comes from biochemical studies, wherein the levels of cSrc protein and tyrosine kinase activity have been examined in hundreds of human tumors and compared to normal tissue controls. As will be discussed in more detail below, in some tumor specimens, high enzymatic activity is accompanied by high protein level, yielding little or no change in specific activities, whereas in others, protein levels are only slightly or modestly elevated, and the specific activity of the enzyme is increased. In yet other examples, high protein levels are accompanied by low enzymatic activity. However, the overall conclusion is that in a very high percentage (>50% and approaching 100% in some studies) of human tumors of many different tissue types, c-Src activity is altered (usually elevated) and that this alteration occurs in early to middle stages of tumor progression and is maintained or increased throughout progression to metastasis.
82
JacquelineS.Biscardi et al.
These findings raise questions as to the mechanism of c-Src activation and the mechanisms by which protein levels are elevated (especially in light of the few instances of increases in c-Src-specific mRNA production). The consensus at the present time is that changes in c-Src specific activity in human tumors are due to posttranslational events and not to mutations of the gene. Using RNase protection and restriction fragment-length polymorphism assays to detect activating mutations of c-Src in a spectrum of human tumors, Wang et al. (1991)were unable to detect mutations at codons known to contribute to the oncogenicity of v-Src and c-Src (namely, codons 98, 381,444, and 530 in the human c-Src sequence). These findings led the investigators to conclude that mutational activation is not the mechanism of enhancement of c-Src-specific kinase activity. On the other hand, DeSeau et al. (1987) described differential activation of c-Src in normal colon cells versus colonic tumor cells depending on the conditions of extract preparation, i.e., whether the lysis buffers contained the proteins tyrosine phosphatase inhibitor, vanadate, and/or high concentrations of ionic and nonionic detergents. From these results, one could deduce that tyrosine phosphorylation of c-Src or other cellular proteins and proteidprotein interactions play a role in regulating not only c-Src activity but also its stability and abundance. Indeed, structural studies on the c-Src molecule described above would support this notion. However, so as not to think that the issue is resolved, studies by Watanabe et af. (1995) indicate that in 18 cancer cell lines, elevated activities of c-Src and c-Yes (a Src-related family member) are accompanied by correspondingly elevated levels of C-terminal Src kinase, the protein that phosphorylates Y530 in human c-Src and negatively regulates c-Src kinase activity. These findings suggest that CSK may not have an antioncogenic role to play in tumor progression or that dephosphorylation of Y530 is not required for activation of c-Src. Here the focus is on three different carcinomas-colon, breast, and lungfor which substantial amounts of data are accumulating to indicate a role for c-Src in their development. That these represent three of the four most common forms of cancer in adults (prostate cancer being the fourth) suggests that c-Src may be a more formidable player in tumorigenesis than had previously been appreciated.
3 . COLON CANCER Utilizing c-Src-specific antibodies and an immune complex-based tyrosine kinase assay, a number of investigators have reported that c-Src-specific tyrosine kinase activity (total activity relative to total c-Src protein in an immune complex) is elevated in colon cancer. In panels of colon cancers examined by Rosen et al. (1986), Bolen et al. (1987a,b), and Cartwright et al.
c-Src and Receptor Tyrosine Kinases
83
(1989),c-Src was found to exhibit elevated kinase activity, ranging from -2to 40-fold above that found in normal colon tissues or cultures of normal colon mucosal cells. In some cases this increase could be accounted for by increases in protein levels, but in other instances it could not, indicating an increase in specific kinase activity. These results suggest that either elevation in c-Src protein and/or activation of c-Src may contribute to the genesis of human colon tumors. Indeed, additional studies by Lundy et al. (1988) and Cartwright et al. (1990,1994) demonstrated increased kinase activity in premalignant epithelia of ulcerative colitis and in early-stage colonic polyps as compared to adjacent normal mucosa. In the latter study, activity was highest in malignant polyps and in >2-cm benign polyps that contained villous structure and severe dysplasia. Thus, c-Src activity is found to be elevated in early stages of colon cancer and this elevation is associated with those polyps that are at greatest risk for developing cancer. Talamonti et al. (1993) also demonstrated incremental increases in c-Src activity and protein level as the tumors progressed, with the greatest increases seen in metastatic lesions. Increases in specific kinase activity were also observed, with liver metastases exhibiting an average increase of 2.2-fold over normal mucosa, whereas extrahepatic metastases demonstrated an average 12.7-fold increase. These results support the idea that c-Src may play multiple roles in tumor progression. A number of studies have been done to determine if c-Src indeed plays a causal role in tumor development. Herbimycin A, an inhibitor of Src family kinases, was shown to inhibit the growth in monolayer of seven colon tumor cell lines as compared to one cell line from normal colonic mucosa, CCL239 (Garcia et al., 1991). In another study, blockage of the myristylation modification of Src family members in a panel of human colon adenocarcinoma tumor cell lines by N-fatty acyl glycinal compounds was shown to prevent localization of c-Src to the plasma membrane and to depress colony formation of these cell lines in soft agar and cell proliferation assays (Shoji et al., 1990). Tumor necrosis factor (TNF-a-mediated growth inhibition of human colorectal carcinoma cell lines was accompanied by a reduction in the activity of s-Src (Novotny-Smith and Gallick, 1992).And last, using an antisense expression vector specific for c-Src, Staley et al. (1997) demonstrated that expression of c-Src antisense in HT29 human colon adenocarcinoma cells resulted in slower proliferation and slower growing tumors in nude mice as compared to the parental control. Together, these studies are consistent with a causative role for c-Src in colon cancer progression. How could c-Src be functioning to promote progression of colonic tumors? Using an in vitro progression model based on the PC/AA premalignant colonic adenoma cell line, Brunton et al. (1997) demonstrated that in the conversion from adenoma to carcinoma, levels of both the EGF receptor and FAK protein increased, while the expression and activity of c-Src were
84
JacquelineS. Biscardi et al.
unaltered. However, EGF induced motility in the carcinoma cells, but not in the adenoma cells, and this increase was accompanied by an EGF-induced increase in c-Src kinase activity, relocalization of c-Src to the cell periphery, and phosphorylation of FAK. The authors interpret these findings to indicate that c-Src is not the driving force for tumor progression, but cooperates with other molecules (such as EGFR and FAK) in the process. Other investigators have observed that adhesion of HT29 human colon carcinoma cells to E-selectin results in a decrease in c-Src activity (Soltesz et al., 1997), suggesting that, on release from substratum restrictions, c-Src activity is restored or elevated. In a related study, Empereur et al. (1997)generated evidence for cooperativity between c-Src and HGF/SF in developing invasive properties of the PC/AA cell line. Specifically, introduction of activated c-Src or polyoma middle-T antigen (which requires c-Src for oncogenic activity) into the adenoma PC/AA cell line induced conversion of the adenoma to carcinoma, overexpression of the HGF receptor, and an invasive capacity in the presence of HGF. Thus, current evidence suggests that one mechanism by which c-Src promotes colonic tumor progression is by cooperating with components of the cell adhesion/motility machinery. Similar conclusions were reached by Mao et al. (1997), who demonstrated activation of c-Src in response to EGF or HGF treatment of human colon cancer cells with high metastatic potential. 4. BREAST CANCER As with colon cancer, a number of early investigations reported elevated c-Src activity in human breast cancers (Jacobs and Rubsamen, 1983; Rosen et al., 1986; Lehrer et al., 1989). In several reports the elevation in activity was not accompanied by elevated levels of c-Src protein, suggesting an activation of the protein. However, Koster et al. (1991), using a screening method based on in vitro synthesis of cDNA copied from total cellular RNA of tumor tissue, found that 25-30% of the analyzed tumors showed significant elevations in expression of several protooncogenes, including c-Src. Using immune complex kinase assays, immunoblotting, and immunohistochemical approaches, Verbeek et al. (1996) and Biscardi et al. (1998a) demonstrated that increases in c-Src kinase activity are almost invariably accompanied by increases in c-Src protein levels and little if any change in specific kinase activity. Interestingly, the immunohistochemical studies of Verbeek et al. (1996) showed that in malignant cells, the majority of c-Src appeared to be concentrated around the nucleus, whereas in normal cells, it is distributed more evenly in the cytoplasm. The discrepancies between the more recent data and the earlier data may reflect changes in the quality of the antibodies and the more quantitative analyses performed in the recent studies. In total, the current evidence indicates that few “activations” of c-
c-Src and Receptor Tyrosine Kinases
85
Src occur in breast tumor cells; rather, elevations in protein levels appear to be the major cause of the increases in c-Src kinase activity. In a recent study involving 72 breast cell lines and tumor biopsies, tyrosine kinase activity was found to be elevated in 100% of the samples, as compared to normal tissue controls, and c-Src tyrosine kinase accounted for 70% of the total cytosolic activity (Ottenhoff-Kalff et al., 1992). The same group performing that study had previously found that the level of cytosolic protein tyrosine kinase activity parallels the malignancy in breast tumors (Hennipman et al., 1989) and that the majority of this activity is precipitated by anti-c-Src antibodies. These results provide compelling correlative evidence that c-Src plays a key role in the development of breast cancer. In agreement with this conclusion, Lehrer et al. (1989) and Koster et al. (1991) also note that elevated c-Src kinase activity is most frequently found in tumors that are progesterone receptor negative. Because loss of progesterone receptor is a histochemical marker for later stage tumors, c-Src activity appears to increase as the tumor progresses in severity. To directly assess the effect of mammary gland-specific expression of c-Src, Webster et al. (1995) established transgenic mice that carried a constitutively activated form of c-Src under the transcriptional control of the murine mammary tumor virus long terminal repeat. Female transgenic mice exhibited a lactation defect and frequently developed mammary epithelial hyperplasias, which occasionally progressed to frank neoplasias. The authors interpret these results to mean that expression of activated c-Src in the mammary gland is not sufficient for induction of mammary tumors-that some other event must take place for frank neoplasias to occur. That c-Src can play more than a bystander role in tumor development, however, was demonstrated by the experiments of Guy et al., (1994),wherein mice transgenic for the polyoma virus middle-T antigen under the control of the murine mammary tumor virus long terminal repeat developed tumors when in a genetic background positive for c-Src, but not when in a background null for c-Src. Similar results were obtained by Amini et al. (1986a), who used c-Src antisense expression vectors to demonstrate that c-Src is required for transformation of rat FR3T3 cells by polyoma middle-T antigen in tissue culture. Together, these studies indicate that c-Src is necessary but not sufficient for tumor development in the mammary gland. 5. LUNG CANCER
Lung cancer is the leading cause of cancer death in the United States. Small cell lung cancer (SCLC) accounts for 20-25% of all bronchogenic carcinoma and is associated with the poorest 5-year survival of all histologic types. c-Src expression was found to be elevated in 60% of all lung cancers (Mazurenko et al., 1991b), when biopsy material of tumors, metastases, and
86
Jacqueline S . Biscardi et al.
“normal” surrounding tissues from patients with different histological types of stomach and lung cancer, melanoma, and other malignancies were analyzed by immunoblotting and immunohistochemistry. A breakdown of the lung histologic types exhibiting increased c-Src expression revealed that cSrc protein was elevated in SCLC and atypical carcinoid tumors, as well as in non-small-cell tumors, such as adenocarcinoma, bronchoalveolar, and squamous cell lung cancer (Mazurenko et al., 1991a). In these studies no analysis of c-Src kinase activity was reported. Somewhat contrasting results were reported by authors of a study in which 60 human cell lines used by the National Cancer Institute for the random screening of potential anticancer drugs were analyzed for c-Src kinase activity. In this study SCLC-derived cell lines had a low activity, whereas non-small-cell lung tumors exhibited activity that was greater than that observed in colon cancer cells, which are considered to have high c-Src activity (Budde et al., 1994). The findings from these studies are strongly supportive of other investigations, concluding that c-Src is frequently overexpressed in SCLC and other types of lung cancer (Cook et al., 1993).
6 . OTHER CANCERS Many other tumor types exhibit elevations in c-Src kinase activity or proteinlmRNA levels, including neuroblastomas (Bjelfman et al., 1990)and carcinomas of the esophagus (Jankowski et d.,1992; Kumble et d.,1997), gastric tract (Takekura et al., 1990), parotid gland (Bu et al., 1996), ovary (Budde et al., 1994), and skin (Kim et al., 1991). With regard to skin cancers, a study carried out in a mouse model of epidermal tumor promotion described activation of erbB2 and c-Src in phorbol ester-treated mouse skin as a possible mechanism by which phorboi esters promote skin tumors in mice. Activation of erbB2 and c-Src kinase is also observed in the epidermis of TGFa transgenic mice, where expression of human TGFa was targeted to basal keratinocytes (Xian et al., 1997). In cervical cancer, evidence is beginning to emerge for cis activation of cellular protooncogenes (including c-Src) by integration of human papillomavirus DNA into the genome of cervical epidermal cells (Durst et al., 1987). In tissue culture studies using primary hamster embryo cells, infection with other DNA tumor viruses, such as SV40, adenovirus, or bovine papillomavirus, also results in increases in the specific activity of c-Src (Amini et al., 198613).
C. c-Src Family Members and Human Cancers c-Src is the prototype for a family of nonreceptor protein tyrosine kinases, for which novel members are regularly being identified. Current members
c-Src and Receptor Tyrosine Kinases
87
include c-Src, Fyn, c-Yes, Lck, Hck, Lyn, c-Fgr, Blk, and Yrk (Brickell, 1992; Sudol et al., 1993; Brown and Cooper, 1996). All members have the same overall structure and minimally contain Unique, SH3, SH2, and kinase domains. The greatest sequence divergence occurs in the Unique domain, thus its name. Not all members are linked to lipids at the N terminus, nor are all negatively regulated by a C-terminal domain that includes the Tyr-530 homolog of human c-Src. c-Fgr, Lck, Hck, and Blk are expressed predominantly in cells of hematopoietic lineage, whereas c-Src, c-Yes, Fyn, Lyn, and Yrk are more ubiquitous. All members have been implicated in various signal transduction pathways, and with the mounting evidence for involvement of c-Src in the genesis of multiple human cancers, the question arises as to whether close relatives of c-Src may also be implicated in these diseases. If so, there are other questions that warrant investigation: Is more than one family member involved in the genesis of the same tumor type? Do c-Src family members fulfill overlapping or unique functions in promoting tumor formation and progression? Are there members of this family that are expressed exclusively in tumors as compared to normal tissue? A review of the literature reveals a paucity of information with regard to any of these issues. It is not clear whether this paucity reflects the unavailability of useful and appropriate reagents to investigate the questions, or whether studies have been conducted and few have uncovered evidence for c-Src family member involvement. Although the following description is not meant to be comprehensive, it does suggest that family members in addition to c-Src may be involved in the genesis of human and certain animal tumors. 1. GENETIC EVIDENCE
In humans, sequences related to the human c-Yes gene were found to be amplified in a single primary gastric cancer out of 22 cases that were examined (Seki et al., 1985). The sequences were amplified four- to fivefold, but normal stomach tissue adjacent to the tumor tissue in the same patient showed no amplification. In the dog, a protooncogene related to the human c-Yes gene was detected as restriction fragment-length polymorphisms (RFLPs) in a Southern blot analysis of genomic DNA from six canine primary mammary tumors in a screen employing seven protooncogene probes. These RFLPs were 0.1 to 1.0 kb shorter than the normal gene, suggesting the occurrence of chromosomal rearrangements and possible deregulation of gene expression, leading to tumorigenesis (Miyoshi et al., 1991). Melanoma formation in the fish Xiphophorus is a genetic model for the function of tyrosine kinases in tumor development. In malignant melanomas from these fish, elevated levels of c-Yes and Fyn activity have been detected as compared to normal tissue (Hannig et al., 1991). Fyn has also been found to coprecipitate with the Xiphophorus melanoma receptor kinase (Xmrk),the molecule
88
lacqueline S. Biscardi et al.
that is responsible for the formation of hereditary malignant melanoma in this lower vertebrate (Wellbrook et al., 1995). These results suggest that Xmrk may function at least in part through Fyn in melanoma formation.
2. BIOCHEMICAL EVIDENCE In studies similar to those conducted for c-Src, evidence for the involvement of other c-Src family members in the etiology of human cancers is emerging, but at a much slower pace than that for c-Src. Elevated c-Yes tyrosine kinase activity has been detected in premalignant lesions of the colon that are at greatest risk for developing cancer (Pena et al., 1995). In this study, the activity of c-Yes in such adenomas was 12- to 14-fold greater than activity in adjacent normal mucosa. Similar results were obtained when mRNA levels of nine protooncogenes in colonic tissue from patients with inflammatory bowel disease (IBD) were measured. The steady-state level of cYes-encoded mRNA was considerably higher in IBD patients resected for colon cancer than in patients resected for active chronic IBD or in controls (Alexander et al., 1996). These results suggest that expression of this gene may be a marker for development of colon cancer in IBD. Finally, in rodents, the action of the transforming proteins of mouse and hamster polyomaviruses (middle-T antigens) is mediated in part through c-Src family kinases, with preferential action of hamster T antigen for Fyn (Brizuela et al., 1995). c-Src family members have also been implicated in the genesis of diseases involving Epstein-Barr Virus (EBV), such as Burkitt’s lymphoma, Hodgkin’s disease, and nasopharyngeal cancer. All of these diseases involve abnormal proliferation of B cells. EBV encodes two transformation-associated proteins, LMPl and LMP2, that are integral membrane proteins. LMP2 mRNA is the only EBV-specific message detected in B lymphocytes from individuals harboring EBV latent infections. LMP2 protein also associates with c-Src family tyrosine kinases, LMPl, and other unidentified proteins, suggesting that the association of these two EBV-encoded membrane proteins could create a macromolecular complex mediating constitutive B lymphocyte activation through normal cell signal transduction pathways (Longnecker, 1994). In human malignant melanoma and other cancers, aberrant expression of basic fibroblast growth factor (bFGF) causes constitutive autocrine activation of its cognate receptor and autonomous growth of tumor cells in culture (see above). Expression of a dominant-negative mutant of the FGF receptor (lacking the kinase domain) was found to suppress tumor formation in nude mice and markedly reduce c-Src family kinase activity in melanoma cells (Yayon et af., 1997). Together these studies suggest that c-Src family kinases play an important role in maintenance and/or progression of malignant melanoma.
c-Src and Receptor Tyrosine Kinases
89
D. Nonreceptor Tyrosine Kinases Related to c-Src Family Members and Human Cancers Several new nonreceptor tyrosine kinases have been isolated from human breast cancer cells. A cDNA encoding a 54-kDa hhosphoprotein, called Rak, was cloned from human breast cancer cells (Cance et al., 1994). This protein shares 51% identify with c-Src and contains SH3,SH2,kinase, and negative regulatory domains. However, it has some properties that are distinct from c-Src, such as its predominant expression in epithelial cells, its lack of a myristylation site, and its almost exclusive localization to the nucleus. However, like c-Src, Rak is overexpressed in subsets of primary human epithelial tumors, suggesting that it may play a role in development of human cancer. Another protein, named Brk (breast tumor kinase), appears to be expressed exclusively in breast tumor tissue as opposed to normal mammary epithelium (Barker et al., 1997). Approximately two-thirds of breast tumors express appreciable levels, and 27% of these overexpress Brk 5- to 40-fold or more. When overexpressed in fibroblasts or mammary epithelial cells, Brk sensitizes cells to the action of EGF and also induces a partial transformed phenotype (Kamalati et af., 1996). These findings suggest that Brk is a functionally important factor in the evolution of breast cancer.
IV. MECHANISMS OF c-Src ACTION A. Evidence for Involvement of c-Src in Signaling
through Receptor Tyrosine Kinases The elevated levels of c-Src expression and/or activation in a wide spectrum of human tumors suggest that c-Src is contributing in some way to the neoplastic phenotype. That c-Src is overexpressed in many of the same tumors in which specific RTKs are also often overexpressed suggests that the two classes of tyrosine kinases may functionally interact to promote tumorigenesis. Many of the RTKs are oncogenic when overexpressed or inappropriately expressed, as described above. The question that follows is whether c-Src is required for the oncogenic capabilities of overexpressed RTKs, or whether c-Src enhances or contributes to RTK-mediated oncogenesis by any means. This latter question was in part addressed when it was shown that cooverexpression of c-Src with HER1 in a mouse fibroblast model resulted in synergistic increases in tumor volume, as compared to tumors developed by cells overexpressing only one of the pair of kinases (Maa et al., 1995). These results provided direct evidence for the enhancing effect of cSrc on receptor-transforming ability, and suggested that a similar synergism
90
jacqueline S. Biscardi et al.
may be occurring in human tumors that cooverexpress c-Src and the EGFR or other RTKs. Targets of c-Src action can be inferred from an analysis of its known intracellular substrates, which, besides the cell surface receptors, are almost exclusively proteins that regulate actin cytoskeleton dynamics. Thus, c-Src appears to have the capability of affecting both mitogenic growth pathways and morphogenic pathways that influence cell/matrix and cell/cell interactions, motility, invasiveness, and metastasis. Here the focus is on studies that are beginning to reveal the molecular interactions between c-Src and its substrates (as they relate to malignancy) and the effects phosphorylation by c-Src have on their functions. It is becoming clear that c-Src is an obligate partner in mediating mitogenic signaling of at least two RTKs, specifically the PDGF and EGF receptors, and that in the case of EGFR, c-Src mediates tumorigenic signaling as well. This new information, in turn, can be used to design novel diagnostics and therapeutics to interdict the symbiotic relationship between c-Src and the RTKs. A number of different growth factor receptors that have been shown to associate with or activate c-Src or Src family members were enumerated in Section I. These included receptors for PDGF, CSF-1, HGF/SF, and EGF, as well as HER2. With the exception of the PDGFR and EGFR, little is known about the role of c-Src in signaling through these receptors, other than the fact that c-Src either associates with the receptor or is activated following specific ligand stimulation. Therefore, we focus our discussion on c-Src interactions with the PDGF and EGF receptors. The data implicating c-Src in PDGF-dependent signaling will be briefly summarized, this being the subject of several other reviews. The bulk of our attention will then be focused on the mechanism of interaction between c-Src and EGFR family members. 1 . ROLE FOR
c-Src IN SIGNALING FROM THE PDGFR
The first evidence that c-Src participates in PDGFR signaling came from the work of Ralston and Bishop (1985), who first observed that c-Src becomes activated on PDGF stimulation. Kypta et al. (1990) later demonstrated that c-Fyn and c-Yes are also activated in a PDGF-dependent manner. Activation of c-Src was shown to be accompanied by a translocation of c-Src from the plasma membrane to the cytosol (Walker et al., 1993), a process that may be linked to internalization of the receptor. PDGF stimulation was also shown to stimulate transient association of Src family members with the PDGFR (Kypta et af., 1990). Association between Src family members and the receptor is believed to involve phosphotyrosine-SH2 interactions, because the SH2 domain of c-Fyn is required for binding to the receptor in vitro (Twamley et al., 1992) and mutation to phenylalanine of Y579 and Y581 in the juxtamembrane region of the receptor results in a decrease in both PDGF-induced c-Src activation and binding to the receptor in
c-Src and Receptor Tyrosine Kinases
91
vivo (Mori et al., 1993). These data and results from in vitro peptide binding studies (Alonso et al., 1995) suggest that Y579 and Y581 directly mediate binding of Src family members to the PDGFR. Interaction of c-Src with the PDGFR appears to have consequences for both c-Src and the PDGFR. Hansen et al. (1996) have shown that Y934 in the kinase domain of the PDGFR is phosphorylated by c-Src both in vitro and in vivo. Expression of a receptor harboring a phenylalanine substitution at residue 934 in intact cells results in a decreased mitogenic signal and an increase in chemotaxis and motility, along with enhanced PLCy tyrosine phosphorylation. These data suggest that phosphorylation of Y934 by c-Src positively regulates mitogenesis, while negatively regulating cell motility, possibly via a PLCy-mediated pathway. Activation of c-Src by PDGF is also accompanied by the appearance of novel phosphorylations on c-Src, including two serine phosphorylations, S12 and an unidentified S residue (Gould and Hunter, 1988)), and one tyrosine phosphorylation, Y138 (Broome and Hunter, 1997). Y138 is located in the SH3 domain of c-Src, and phosphorylation of this residue diminishes the ability of peptide ligands to bind the SH3 domain in vitro. Mutation of Y138 or Y133 to phenylalanine or complete deletion of the SH3 domain reduces the mitogenic effect of PDGF (Erpel et al., 1996; Broome and Hunter, 1996). The hypothesis that Src family members are required for PDGF-dependent signaling is supported by the inhibitory effects of kinase-inactive c-Src or an antibody specific for the C-terminal domain of Src family members on PDGF-induced BrdU incorporation into newly synthesized DNA (Twamley-Steinet al., 1993).
2. ROLE OF c-Src IN SIGNALING FROM EGFR In our laboratory, initial attempts to detect EGF-induced alterations in cSrc kinase activity or physical association between c-Src and the EGFR in a panel of nontransformed avian and rodent cell lines were negative, or yielded inconsistent results (Luttrell et al., 1988). Therefore, a direct test of the involvement of c-Src was undertaken, in which wild-type (wt) and mutational variants of c-Src were overexpressed in C3HlOT1/2 mouse fibroblasts, and the effect of overexpression of these variants on EGF-induced [3H]thymidine incorporation was examined. Overexpression of wt c-Src resulted in a two- to fivefold increase in [3H]thymidine incorporation above Neo-only controls (Luttrell et al., 1988), whereas overexpression of c-Src harboring inactivating mutations in the kinase, SH2, or rnyristylation domains resulted not only in a reduction in the enhanced effect of overexpressed wt c-Src but also in a dominant negative effect on endogenous, EGFinduced DNA synthesis (Wilson et al., 1989). These results indicated not only that c-Src is required for mitogenesis stimulated by EGF, but also that c-Src kinase activity, an intact SH2 domain, and membrane association are
92
lacqueline S.Biscardi et al.
necessary to fulfill the role of c-Src in the process. These findings were corroborated by studies in NIH3T3 cells, in which a decrease in EGF-induced BrdU incorporation was observed on microinjection of antibodies to c-Src family members or introduction of a kinase-inactive c-Src cDNA into cells (Roche et al., 1995). c-Src was also shown to affect EGF-induced tumorigenesis (Maa et al., 1995). In C3H10T1/2 cells, coexpression of c-Src and the HERl results in synergistic increases in proliferation, colony formation in soft agar, and tumorigenicity in nude mice, as compared to cells overexpressing c-Src or HERl alone. Furthermore, under conditions of receptor and c-Src overexpression, an EGF-inducible complex between the proteins can be detected. Enhanced tumor growth correlates with the ability of c-Src to associate stably with the receptor, the appearance of two novel tyrosine phosphorylation sites on the receptor, and enhanced phosphorylation of the receptor substrates, Shc and PLCy. These findings suggest that c-Src association with and phosphorylation of the receptor results in hyperactivation of the receptor and enhanced mitogenic signaling to downstream effectors. Subsequent investigations have revealed that the kinase activity of c-Src is required for the biological synergy between c-Src and overexpressed HERl (Tice et al., 1998). Kinase-defective c-Src, when expressed in a cell line overexpressing HER1, acts in a dominant negative fashion to inhibit EGF-dependent colony formation in soft agar and tumorigenicity in nude mice. The effects of both wt and kinase-defective c-Src are very striking, with the single wt c-Src or HERl overexpressors forming barely detectable tumors in nude mice (