Advances in Cancer Research Volume 42

ADVANCESINCANCER RESEARCH VOLUME 42

Contributors to This Volume Annerose Anders

Barbara B. Knowles

Fritz Anders

Karl Lennert

Angelika Barnekow

Paul A. Marks

Robert D. Cardiff

David Y. Mason

Peter C. Doherty

Richard A. Rifkind

Alfred C. Feller

Manfred Schartl

Richard K. Gershon

Michael Sheffery

Douglas R. Green

Harald Stein

Peter J. Wettstein

ADVANCES IN CANCERRESEARCH Edited by

GEORGE KLEIN Department of Tumor Biology Karolinska lnstitutet Stockholm. Sweden

SIDNEY WEINHOUSE Fels Research Institute Temple University Medical School Philadelphia, Pennsylvania

Volume 42- 7 984

ACADEMIC PRESS, INC. (Harcourt Brace Jovanovich, Publishers) Orlando San Diego New York London Toronto Montreal Sydney Tokyo

COPYRIGHT @ 1984, BY ACADEMIC PRESS,INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN ANY F O R M OR BY ANY MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR A N Y INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER.

ACADEMIC PRESS, INC.

Orlando, Florida 3288 7

United Kingdom Edition publislted by ACADEMIC PRESS, INC. ( L O N D O N ) LTD. 24/28 Oval Road, London N W l 7DX

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 52 - 1 3 3 6 0

ISBN 0-12-006642-4

PRINTED IN THE UNITED STATES OF AMERICA 84 85 86 87

9 8 7 6 5 4 3 2 1

CONTENTS

CONTHIHUTOHS TO VOI.UME42 . . . . . . . . . . . . . . . . . . .

ix

Immunological Surveillance of Tumors in the Context of Major Histocompatibility Complex Restriction of T Cell Function PETEII

c. D O I I E H N .

BAHBAHA

B . KNOWLES. A N D

PETEH

J . WETTSTFXN

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Imtnunological Surveillance: Historical Aspects and Early Speculations Concerning Transplantation Antigens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I11. Criticisms of Iintnunological Surveillance 1970 . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . Evidence That T Cell Surveillance Is Important i n Some Tumor Systems . . . . . V . MHC-Restricted T Cell Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............................ VI . The Molecular Nature of the MHC ransplantation Antigens . . . . . . . VII . T Lymphocyte Effectors and Tumor VIII . Biological Models for Recognition of Minimal Changes on Cell Surface: T Cell Responses to SV40 TSTA and Single Minor H Antigens . . . . . . . . . . . . . . IX . The Expression of MHC Antigens on Tumor Cells . . . . . . . . . . . . . . . . . . . . . . . . ..... X . MHC Phenotype and Susceptiliility to Cancer . . . . . . . . . . . XI . General Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............................ XI1. Summary . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2 5 8 12 16 21 32 36 50 51 53 54

lmmunohistological Analysis of Human Lymphoma: Correlation of Histological and Immunological Categories HAHALD STEIN. K A R L. LENNEHT. ALFREII

c. FELLEH.A N D

DAVIII Y . MASl)N

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1 . Distinction of Malignant Lympliotna from Other Neoplasms . . . . . . . . . . . . . . . . 111. Division of Malignant Lymphoma into Hodgkin’s Lymphoma. Non-Hodgkin’s Lymphoma and True Histiocytic Sarcoma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . B Cell Lymphomas . . . . . . . . . . . . ... ... V . T Cell Lymphomas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . .

.

67 69 71 73 117 140 142

vi

CONTENTS

Induced Differentiation of Murine Erythroleukemia Cells: Cellular and Molecular Mechanisms RICIIARD

A. RIFKIND,M I C I I A E L SEIEFFERY, A N D

PAUL

A. MARKS

I. Introduction . . . ...................................... 11. Terminal Cell Division . . . . . . . . . . . . . . . . . . . . . . . . . Induced Differentiation . . . . .

149 151

. . . . . . . . . . . . . 155

. . . . . . . . . . . . . . .. . . . . . . . . . . .. References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

160 162 163

Protoneoplasia: The Molecular Biology of Murine Mammary Hyperplasia ROBERTD. CARDIFF Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mouse Mammary Tumor System . . . . . , . , . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Development of Transplantable Hyperplastic Outgrowth Lines Characterization of Hyperplastic Outgrowth Lines. . . . . . . . . . . . . . . . . . . . . . . . . Restriction Endonuclease Mapping. . ............................... The Molecular Biology of Mouse Mammary Neoplasia . . . . . . . . . . . . . . . . . . . . . Origin and Evolution of Mouse Mammary Tumors . . . . . . . . . . . . . . . . . . . . . . . . The HAN Is Protoneoplastic . . . . . . . . . . . . . . . IX. The Role of MuMTV in Mous X. Summary. . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I. 11. 111. IV. V. VI. VII. VIII.

167 168 170 171 172 178 181 185 186 187 188

Xiphophorus as an in Wvo Model for Studies on Normal and Defective Control of Oncogenes FRITZANDERS,MANFRED SCNARTL, ANCELIKA BARNEKOW, AND ANNEROSE ANDERS I. Introduction and Historical Background.. . . . 11. Ubiquity of Oncogenes in Purebred Animals

191 194 211 230 IV. Oncogenes and Regulatory Genes.. . . . . . . . . . . V. Theoretical Considerations on a General Concept of Neoplasia.. . . . . . . . . . . . . 26 1 VI. Conclusions. . . . . . . . . . . . . . . . . . . . 268 268 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Contrasuppression: The Second Law of Thymodynamics, Revisited DOUGLAS R. GREENA N D RICHARD K. GERSHON I. Introduction.. . . . . . . . . . . .. . . . .. . . . .. . .. .. . . . . . . . . . . . . . . . . . . . . . . . .. . . .

11. Defining Cell Circuits.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

278 280

vii

CONTENTS

111. Defining Contrasuppression ..... IV. Defining a Specific Cont Phenotype to Its Cellula V. Functional Activity of the Contrasuppressor Circuit . . . . . . . . . . . . . . . . , , . , , . , VI. Conditions That Influence the Generation and/or Activation of Contrasuppressor Cells . VII. Immutiological Consequ ssor Circuit. . , VIII. Contrasuppression and Tumor Immuri ............. IX. Human Examples of Contrasuppressioi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X. Contrasuppression in the Future. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I

.

.

.

.

.

.

.

.

.

.

.

.

.

..........

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

I . . . . . . . . . . . . . . . . . . . . . .

282 282 285 286 30 1 312 318 320 322

Addendum 1: Heterogeneity of Contrasuppressor T Cell Function and Addendum 2: Relation of the Allogeneic Effect to Contrasuppression . . . . . . . .

INDEX. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CONTENTS OF PHEVIOUS VOLUMES . . . . . . . . . . . . . . . . . . . . . . . . . , . . , . . , , , . . , . . ,

326 33 1

337 343

This Page Intentionally Left Blank

CONTRIBUTORS TO VOLUME 42 Numbers in parentheses indicate the pages on which the authors’ contributions begin

( l g l ) , Genetisches Institut, Justus-Liebig-Universitiit Giessen, 0-6300 Giessen, Federal Republic of Germany FRITZ ANDERS (191), Genetisches lnstitut, Justus-Liebig-Universitat Giessen, 0-6300 Giessen, Federal Republic of Germany ANGELIKABAHNEKOW (191), Institut f u r Virologie, Justus-Liebig-Universitiit Giessen, 0-6300 Giessen, Federal Republic of Germany ROBERT D . CARDIFF (167), Department of Pathology, University of Calqornia School of Medicine, Davis, Calqornia 95616 PETER c. DOHERrY (I), Department of Experimental Pathology, The John Curtin School of Medical Research, Canberra ACT 2601, Australia ALFREDC. FELLEH (67), lnstitute of Pathology, Christian Albrecht University, 2300 Kiel, Federal Republic of Germany RICHARD K. GERSHON (277), Department of Pathology and the Howard Hughes Medical Institute for Cellular Immunology, Yale University School of Medicine, New Haven, Connecticut 06510 R. GREEN (277), Department of Pathology, Yale University School DOLJCLAS of Medicine, New Haven, Connecticut 06510 BARBARA B. KNOWLES (l),The Wistar Institute, Philadelphia, Pennsylvania ANNEROSE A N D E R s

19104

LENNERT (67), Institute of Pathology, Christian Albrecht University, 2300 Kiel, Federal Republic of Germany PAULA. M A R K S (149), DeWitt Wallace Research Laboratory and the SloanKettering Dicision, Graduate School of Medical Sciences, Meimrial Sloan-Kettering Cancer Center, New York, New York 10021 DAVID Y. MASON (67), Department of Haematology, John Radclqfe Hospital, Oxford OX3 9DU, England RICHARDA. RIFKIND(149), DeWitt Wallace Research Laboratory and the Sloan-Kettering Division, Graduate School of Medical Sciences, Memorial Sloan-Kettering Cancer Center, New York, New York 10021 MANFRED SCHARTL(191), Genetisches Institut, Justus-Liebig-Universitat Giessen, 0-6300 Giessen, Federal Republic of Germany MICHAELSHEFFERY(149), DeWitt Wallace Research Laboratory and the Sloan-Kettering Division, Graduate School of Medical Sciences, Memorial Sloan-Kettering Cancer Center, New York, New York 10021 HARALDSTEIN (67), Institute of Pathology, Christian Albrecht University, 2300 Kiel, Federal Republic of Germuny PETER J. WETSTEIN(l),The Wistar Institute, Philadelphia, Pennsylvania 19104 KAHL

ix

This Page Intentionally Left Blank

IMMUNOLOGICAL SURVEILLANCE OF TUMORS IN THE CONTEXT OF MAJOR HISTOCOMPATIBILITY COMPLEX RESTRICTION OF T CELL FUNCTION Peter C. Doherty, Barbara B. Knowles, and Peter J. Wettstein Department of Experimental Pathology, The John Curtin School of Medical Research. Canberra, Australia, and The Wistar Institute, Philadelphia, Pennsylvania

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Immunological Surveillance: Historical Aspects and Early Speculations Concerning Transplantation Antigens. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Immunological Surveillance 1970. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Early Ideas about the Role of the Major Histocompatibility Complex and Immunological Surveillance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Alloreactivity and Immunological Surveillance 1970. . . 111. Criticisms of Immunological Surveillance 1970. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Nude Mice and Natural Surveillance. . . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Immunopotentiation and T Cell Subsets. . . . . . . . . . . IV. Evidence That T Cell Surveillance Is Important in Some Tumor Systems.. . . . . . A. Viruses, Tumors, and Host B. Immunosuppression and Tr V. MHC-Restricted T Cell Recogn .............................. A. T Cell Specificity for M H C B. The T Cell Repertoire and Zr G e n e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . The Molecular Nature of the M H C . . . . . . . . . . . A. Class I Genes: Recognition and Polymorphism . . . . . . . . . . . . . . . . . . . . . . . . . . B. Class 11 Glycoproteins: Posttranslational Associations and Diversity VII. T Lymphocyte Effectors and Tumor-Specific Transplantation Antigens . . . . . . . . . A. T Cell Specificity for Viruses.. , . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. What Are Tumor-Specific Transplantation Antigens? . . . . . . . . . . . . . . . . . . . . . VIII. Biological Models for Recognition of Minimal Changes on Cell Surface: T Cell Responses to SV40 TSTA and Single Minor H Antigens IX. The Expression of MHC Antigens on Tumor Cells . . . . . . . . . . . . . . . . . . . . . . . . . . A. Class I MHC Antigens and Tumorigenicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Mutations in Class I MHC Genes and T Cell Surveillance . . . . . . . . . . . . . . . . C. “Alien” Class 1 M H C Antigens on Tumor Cells . . . . . . . . . . . . . . . . . . . . . . . . . D. Cross-Reactions at the T Cell Level between TSTA and Class I M H C Antigens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Expression of Class I1 MHC Antigens on Tumor Cells . . . . . . . . . . . . . . . . . . . X. MHC Phenotype and Susceptibility to Cancer. . . . . . . . . . . . . . . . XI. General Concepts , . . , , . . , . . , , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... ................................. XII. Summary References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2 3 3 4 5 6 7 8 8 10 12 12 13 16 17 19 21 22 27 32 36 36 42 44 45 47 50 51 53 54

I. Introduction

Analyzing the role of major histocompatibility complex (MHC) glycoproteins in controlling the recognition of cell membrane components by thy1 ADVANCES IN CANCER RESEARCH, VOL. 42

Copyright Q 1984 by Academic Press, Inc. All rights of reproduction in any form reserved. ISBN 0-12-006642-4

2

PETER C. DOHERTY ET AL.

mus-derived lymphocytes (T cells) has, over the past decade, been a major obsession for many cellular immunologists. Considerable progress has been made, though, because of the difficulties of studying what are essentially surface interactions between functionally different cell populations, some important ideas and experiments are in dispute. Even so, a general consensus exists for many key points and there are a number of thoughtful, extensive reviews dealing with the basic biology of MHC restriction (e.g., Shearer and Schmitt-Verhulst, 1977; Miller, 1979; Wagner et al., 1980; Sprent et al., 1980; Snell, 1981; Klein and Nagy, 1983). Yet another article of this type would be of relatively little value, and very boring both to write and to read. We thus take as a starting point our own earlier reviews of the topic (Zinkernagel and Doherty, 1979; Doherty, 1980a,b, 1983; Doherty and Bennink, 1980a), and do not attempt to include all relevant references. The intention of the present account is thus to concentrate on the role of the MHC restriction phenomenon in immunological surveillance, particularly T cell surveillance, of tumors. No attempt is made to deal in detail with the whole problem of surveillance against cancer. The emphasis is on T cell recognition, antigenicity, the nature of MHC antigen expression on tumor cells, and the implications of such findings for the operation (or subversion) of immunological surveillance. Trying' to review the tumor literature in this way has proved an intriguing, though sometimes perplexing, exercise: it would be difficult to overemphasize the complexity of tumor systems! II. Immunological Surveillance: Historical Aspects and Early Speculations Concerning Transplantation Antigens

There is a need to put the useful facets of the immunological surveillance concept in a contemporary way so that, in the minds of tumor biologists, the baby is not thrown out with the somewhat muddy bath water. According to Humphrey (1981) the baby may be 75 years old and the father is Paul Ehrlich. The immunological surveillance hypothesis was proposed before the discovery of MHC restriction. Immunological surveillance, more particularly T cell surveillance, needs to be reargued in the context of the necessity of thymus-derived lymphocytes to recognize antigens that might be unique to the tumor in association with MHC glycoproteins. It now seems very likely that the major function of the cytotoxic T lymphocyte (CTL) set is the monitoring of cell-surface structural integrity throughout the body and the elimination of cells expressing evidence of abnormal phenotype (Doherty and Zinkernagel, 1975; Doherty et a l . , 1976; Shearer and Schmitt-Verhulst, 1977; Zinkernagel and Doherty, 1979; Mills and North, 1983). The most appropriate term in the English language to describe

T CELL S U R V E I L L A N C E A N D M H C

3

this role is “immunological surveillance” (Zinkernagel and Doherty, 1974b), or T cell surveillance. However, the use of “immunological surveillance” is, in the minds of most workers in the cancer field, preempted by the powerful statement of Burnet (1970) who tied the phylogeny and ontogeny of T lymphocyte function almost exclusively to the need to destroy malignant cells. This must now be regarded as historical (Mitchison and Kinlen, 1980).There has been a total (and continuing) revolution in our understanding of the role and specificity of T cells since 1970, when Burnet’s book was published. The term immunological surveillance needs, therefore, to be restated in a contemporary context if we are to retain it as a useful generalization. Some aspects of Burnet’s original formulation must be deemphasized, while others were of undoubted predictive value.

A. IMMUNOLOGICAL SURVEILLANCE 1970 Burnet (1970) did not intend that the immunological surveillance concept should assume the status of a rigid hypothesis, or dogma. However, his arguments were compelling and there was a ready audience for any optimistic statement concerning the possibility of iininunological intervention in cancer. We should allow Burnet to speak for himself. The immunological surveillance concept is, in essence, “that a major function of the immunological mechanism in inarninals is to recognize and eliminate foreign patterns arising in the body by somatic mutation or some equivalent process. Froin the point of view of survival, this is important primarily as a means by which the appearance of malignancy may be effectively cut short. . . . The thymus dependent system of irnmunocytes will be almost solely responsible for surveillance, antibody and antibody-producing cells having an almost negligible role.” The arguments thus serves to focus our attention onto the need to eliminate abnormal cells, and states that this is a function of T cells. We should have little to quarrel with in this aspect of Burnet’s formulation. The emphasis on malignancy, which was also stressed by Thomas (1959), may need to be moderated. However, the concept of T cell surveillance against tumor cells should not be discarded in the process.

B. EARLYIDEASABOUT

THE ROLEOF T H E MAJOR HISTOCOMPATIBILITY COMPLEX A N D IMMUNOLOGICAL SURVEILLANCE

The interpretation of the nature of cell-surface surveillance was revolutionized by the discovery and analysis of (MHC) restriction of virus and hapten-specific cytotoxic T cell function in late 1973 (Section V). This may ultimately be shown to have vindicated the earlier proposal of Lawrence

4

PETER C. DOHERTY ET

AL.

(reviewed by Lawrence, 1974) that many T cells may be specific for “self + x. Lawrence suggested that “self’ might be the so-called transplantation antigens. However, Lawrence’s ideas did not gain general acceptance at the time (as early as 1959): Burnet (1970)did not refer to him and those of us who came later to the field were completely unaware of his perceptive speculations. This is at least in part due to the fact that many of the studies with transfer factor, the nature of which Lawrence was trying to explain, were equivocal and there was no good experimental system available for analyzing his model. Even so, more recent work with both MHC restriction and T cell factors may support the case that Lawrence was on the right track, bnth technically and intellectually. The concept that there was some relationship between viruses and the MHC had also been explored conceptually, in a somewhat different way, by Snell (1968) who sought to explain the extraordinary polymorphism of the MHC genes by postulating that H-2 antigens served as receptors for potentially lethal viruses. Total elimination of a species during the course of a raging pandemic would be avoided if the MHC glycoproteins of some individuals did not bind the virus in question. More recent findings have removed most of the need to make this argument. Even so, if it turns out that self + x is true (Zinkernagel and Doherty, 1974b), hell’s idea may be applicable to whether or not a particular virus can generate an appropriately immunogenic interaction with MHC products on the surface of the stimulator/target cell (Doherty and Zinkernagel, 1975). However, it is very obvious that these speculations of both Lawrence (1974) and Snell(l968) had no influence on either Burnet’s formulation of the immunological surveillance concept, or on the field of transplantation in general. The first is evident in Burnet’s writing, the second in the fact that no major research group turned its attention to the subject of infectious viruses and the transplantation antigens. ”

C. ALLOREACTIVITYAND IMMUNOLOGICAL SURVEILLANCE 1970 In 1970, the most impressive known functions of the T cell were concerned with graft rejection and alloreactivity. People thinking about T cell recognition (Amos et d., 1971; Bodmer, 1972; Burnet, 1973) were immensely impressed with both the potency of the alloreactive response, and with the extremely high frequency of effector T cells which were apparently specific for foreign transplantation antigens (Simonsen, 1967; Wilson et d., 1968). Why should this be so? What possible raison d’etre could there be for the evolution of a mechanism for rejecting cells from other individuals of the same, or a different, mammalian species? The available evidence thus suggested that T lymphocytes were focused

T CELL SURVEILLANCE AND MHC

5

onto the surface of other cells by the presence of structures seen as non-self, the so-called transplantation antigens. In an evolutionary context, this might be thought to reflect a need for the development of a mechanism for avoiding mutual parasitism by members of the same species. Rapid homograft rejection is found in all vertebrate species above the amphibia (Du Pasquier and Miggiani, 1973), and incompatibility associated with cell killing is seen for organisms as primitive as the corals and sponges (Hildeman et a l . , 198Oa,b). It is still not inconceivable that these interactions between primitive precursors of the immunocytes and cells bearing early analogs of the transplantation antigens may, as animals became more complex, have driven the evolution of the thymus and the T cell system. However evolutionary pressure may not have been principally exerted by the need to avoid mutual parasitism and to control somatic mutation: infectious disease has probably played a major role (Zinkernagel and Doherty, 1979). Burnet (1970) reviewed the then available information and proposed that the contemporary equivalent of the mutual rejection process observed in more primitive life forms (Hildeman et al., 1980a,b) served two main functions. The first was to prevent the transmission of malignant cells between individuals of the same species who happened to make close, physical contact (Burnet, 1973). The second was to eliminate cells that expressed evidence of somatic mutation, which might be associated with lack of biological control. It is easy, with hindsight, to criticize these proposals. Why should spontaneously arising tumors of nonviral origin not be transmitted horizontally within inbred mouse populations? Could we assume that mutations which resulted in escape from normal growth control would, of necessity, modify the histocompatibility antigens? However, as late as 1971 (Amos et al., 1971), despite immense experimental effort and the existence of journals dedicated to the subject, alternative speculations on the biological role of the strong transplantation antigens were so little grounded in experimental evidence as to have no impact on the field in general. Burnet’s formulation of immunological surveillance has, however, generated continuing debate and enquiry.

Ill. Criticisms of Immunological Surveillance 1970

The evidence against the immunological surveillance concept (Burnet,

1970)was clearly summarized in a provocative editorial by Moller and MO1ler (1976), which introduced a volume of Transplantation Reviews (Vol. 28) concerned with the topic. It may be useful to examine some of these criticisms in the light of our current understanding of host response.

6

PETER C. UOHERTY ET A L .

A. NUDE MICE A N D NATURAL SURVEILLANCE

Much has been made of the observation that congenitally athymic (nu/nu) mice are less likely to develop spontaneous tumors than are their phenotypically normal (nu/ +) littermates (Rygaard and Povlsen, 1976). However, not all long-term studies of nu/nu mouse populations have given this result (Outzen et al., 1975; Prehn, 1976a,b; Stutman, 1981; Parker et al., 1982). Also, there is no debate that the nu/nu mouse is much more susceptible to virus-induced (polyoma) oncogenic process (Allison et al., 1974; Allison, 1980)and that elimination of the polyoma-transformed cell may be due to T cell surveillance function. It is now amply documented that the nu/nu mice have much more potent natural killer (NK) cell populations than are found in their n u / + littermates (Herberman and Holden, 1978; Herberman et al., 1979). It also seems reasonable to assume that the NK cell may perform a generalized surveillance function, which serves to eliminate spontaneously emerging tumors at the stage when they are still very small (Kiessling and Wigzell, 1979; Roder and Haliotis, 1980): NK cells are present in the circulation and are capable of mediating effector function following a short period of stimulation. The T cell, on the other hand, requires specific stimulation and clonal expansion, a process which does not lead to the generation of cytotoxic effectors for 5 or 6 days. Perhaps this function of NK cells should not, at the present stage of understanding, be referred to as immunological surveillance. Until there is clear evidence of an adaptive NI( response, with clonal expansion, memory, and specificity that is more than continental (if not global) the NK cell may not legitimately be regarded as an immunocyte. It is more appropriate to speak of NK surveillance, “natural surveillance,” or “natural resistance” (Stutman, 1981). However, the lines between these two systems should not be drawn too rigidly. Cloned CTL may be induced to show the specificity patterns characteristic of NK cells following incubation in the presence of excessive quantities of lymphokines (Brooks, 1983). This may reflect the expression of a second NK-type receptor on the T cells (Binz et al., 1983). Might such events occur in the physiological microenvironment of the tumor? The essential point concerning the nutnu mouse is that it cannot be regarded as an animal model for simple lack of T cells. A powerful, if often inadequate compensatory mechanism exists in the NK cells and nonspecifically activated macrophages (Nelson et al., 1981; Herberman and Ortaldo, 1981). Also, growing nu/nu spleen cells in the presence of interleukin 2 (11-2) may result in the emergence of functional T cells (Hunig and Bevan, 1980b; Stotter et al., 1980).

T CELL S U R V E I L L A N C E A N D M H C

7

The one fact that is undisputed is that the nu/nu mouse is much more susceptible to many infectious processes, which points to the major role of T cell surveillance mechanisms in acute encounters with viruses, bacteria, and larger parasites. The defect in host responses is clearly exemplified by the fact that many human tumors (Fogh et al., 1982) and somatic cell hybrids made with human tumors grow unchecked in nulnu mice, while these are readily rejected by their nu/+ littermates (Koprowski et al., 1978). Even so, the fact that the nulnu is not obviously more likely to develop spontaneous tumors points to the basic importance of natural surveillance mechanisms (Stu tman , 1981). B. IMMUNOPOTENTIATION A N D T CELLSUBSETS We were also, in 1970, unaware of the fact that there are functionally different sets of T cells, broadly classified as cytotoxic (CTL), helpers (Th), and suppressors (Ts). Effector function for the CTL component is mediated by direct contact between T cell and target and results in lysis of the target. However helper T cells may not interact with the tumor cell as such (unless it expresses class I1 MHC antigens, see Section IX,E), but with tumorspecific components that have been processed by an appropriate stimulator macrophage or dendritic cell. The helper then elaborates growth factors which, if secreted in an actual tumor, might (perhaps) serve to promote the growth of that tumor. This would not, for instance, be a problem for the case of normal somatic cells infected with a lytic virus. It is thus not too surprising, even without invoking the complexities of T cell circuitry (Cantor et al., 1978; Gershon et al., 1981), that there are reports in the literature (Prehn 1976b, 1983)that improving the availability of T cell precursors results in enhanced tumor growth. Further analysis of this question may be made by assessing the consequences of transferring primed, Lyt 1 - 2 + (CTL, Ts) and Lyt 1+ 2 - (Th) to T cell-deficient animals bearing tumor foci. In addition, the Ts subset may be eliminated by pretreating the donors with a small dose of cyclophosphamide (30 mg/kg, Gershon, 1975). Also, the availability of the new monoclonal anti-Lyt reagents has considerably facilitated such studies. Experiments with the influenza virus model have shown that the Lyt 1 + 2 - subset promotes a delayedtype hypersensitivity response which is of no obvious benefit to the animal, while the Lyt 1 - 2 + cells are involved in eliminating virus-infected cells and in terminating the disease process (Ada et a l . , 1981). The Lyt 1 - 2 + terminology used here refers to experiments done using complement-mediated lysis to remove one or another cell population: such lymphocytes can be shown to express small amounts of Lyt 1 when examined on the fluorescence-activated cell sorter (Bluestone and Hodes, 1983).

8

PETER C. DOHERTY ET AL.

Combining the Lyt 1 - 2 + and Lyt 1 + 2 - populations may tend to improve the capacity of a mouse to destroy, for instance, influenza virusinfected alveolar macrophages and lung epithelium. Both T cells and activated macrophages, which may differentiate from blood-bourne monocytes recruited (at least in part) by factors secreted by the Lyt 1 2- population, are necessary to eliminate foci of virus infection in tissues. However, in the case of a tumor, it is not beyond the bounds of possibility that these two sets of T cells might work against each other, with the CTL removing transformed cells and the Th subset elaborating growth factors. The net consequence could be to potentiate the immunoselection of somatic mutants that are no longer seen by the CTL (Section IX). Any further approach to the problem of immunopotentiation needs to be done using functionally defined T cell subsets (Mills and North, 1983) or cloned populations of lymphocytes. +

IV. Evidence That T Cell Surveillance Is Important in Some Tumor Systems

Before discussing in detail the interactions between MHC antigens, T cells, and tumors, it seems appropriate to first summarize evidence supporting the idea that immunological surveillance does operate, at least sometimes, to limit the emergence of malignancy. The obvious defects in immunological surveillance have received considerable attention (Stutman, 1975; Moller and Moller, 1976; Prehn, 1976a,b),and there is no point in attempting to review this information again here. However, it will be argued later that many “holes” in the T cell surveillance network can be readily explained in the context of current knowledge of the limitations of MHC-restricted T cell recognition (Sections, V, B, IX).

A. VIRUSES,TUMORS,A N D HOSTRESPONSE It is obvious that T cell surveillance mechanisms may be very effective at preventing the emergence (or mediating the rejection) of virus-induced tumors in mice, rats, and rabbits (reviewed by Levy and Leclerc, 1977; Zinkernagel and Doherty, 1979; Kreider and Bartlett, 1981). A more detailed account of instances where suppression (or modification) of an individual MHC gene may allow the growth of a particular virus-induced tumor is given in Sections IX,A and B. The Epstein-Barr virus (EBV) causes infectious mononucleosis (IM) of man, and is also implicated as the triggering agent in at least two human tumors: Burkitt’s lymphoma and nasopharyngeal carcinoma (reviewed by Nilsson and Klein, 1982; Ada, 1982). The latter disease is, at least in the Singapore Chinese, slightly more prevalent in individuals of the HLA-B4,

T CELL SURVEILLANCE A N D M H C

9

A-27 phenotype. The virus is involved in the induction of continuous proliferation of human B cells which, during IM, appear in relatively small numbers in the peripheral circulation (Klein, 1979). Most of the lymphocytes circulating in IM are T cells, and EBV-immune T cells are known to interact with EBV-infected B cells in a class I MHC-restricted fashion (Rickinson et d . , 1980; Moss et a l . , 1980). The general situation for IM, and for children who are too young to develop clinical mononucleosis, is probably that the immune T cells eliminate EBV-infected B cells before they reach a fully transformed state (Thorley-Lawson et al., 1977; Moss et a l . , 1977). The induction of lyniphomagenesis seems to depend on a multistep process involving chromosomal translocations in the region of C-myc and the activation of the transforming gene Blym-1 (Diamond et a l . , 1983). The EBV model thus seems to be a clear instance of the effective operation of T cell surveillance mechanisms. In addition, those B cells that do become transformed may then be held in check by EBV-immune T cells (Moss et al., 1977; Thorley-Lawson, 1980). For instance, newly diagnosed patients with nasopharyngeal carcinoma show a significant impairment of EBV-specificT cell-mediated immunity when compared with long-term survivors and EBV-immune controls who do not have the disease (Moss et d . , 1983). Much more severe disease occurs in individuals with genetically determined immunodeficiency (Purtillo, 1981), or, as may be the case with Burkitt’s lymphoma, those suffering chronic B cell proliferation as a result of constant exposure to malaria (Klein, 1979). Another example of high tumor incidence associated with infectious processes is Kaposi’s sarcoma in homosexual men expressing the HLA DR 5 phenotype (Friedman-Kien et al., 1982). The acquired immunodeficiency syndrome (AIDS) recognized in such individuals is associated with a relative drop in numbers of T cells of the helper-inducer (OKT4+) subset, with concurrent infection by a number of agents including cytomegalovirus and Pneumocystis spp. (Gottlieb et a l . , 1981; Friedman-Kien et al., 1982; Fauci, 1982). In addition, at least 25-40% of people with AIDS (cf. 1% in the population at large) have evidence of present or past exposure to the human T cell leukemia virus (HTLV) which infects OKT4+ lymphocytes (Gallo and Reitz, 1982; Essex et al., 1983a; Gelman et d.,1983; Gallo et d . , 1983; Barr6-Sinousi et al., 1983). The question is whether these infections are cause or effect in AIDS. Other possibilities are that the disease is induced in passive homosexuals by repeated intravascular exposure to allogeneic lymphocytes, human sperm, or seminal fluids (Shearer, 1983; Navarro and Hagstrom, 1982), or that taking amyl nitrite has an adjuvant effect (Durack, 1981). However, the fact that AIDS is also recognized in nonhomosexual drug addicts and in hemophiliacs receiving clotting factors indicates that a

10

PETER C . DOHERTY ET AL.

blood-borne infectious agent must be involved (Marx, 1983a). Hemophiliacs also have a very high incidence of antibodies to HTLV (Essex et al., 198313). One possibility is that the development of Kapsoi’s sarcoma is influenced by the host response to OKT4+ lymphocytes that are persistently infected with HTLV (Barrk-Sinousi et al., 1983). For instance, Levy and Ziegler (1983) have suggested that induction of the tumor results from continued secretion of angiogenesis-generating factors by cells attempting to compensate for the immune disorder. If this is true, AIDS may reflect iinmunostirnulation rather than a failure of basic T cell surveillance mechanisms. A similar argument may be made for the etiology of Burkitt’s lymphoma. Alternatively, Kaposi’s sarcoma may reflect iinmunosuppression as a result of removal of potential responder cells of the OKT4+ phenotype by HTLVspecific CTL (Mitsuya et d., 1983). Uncontrolled host response to cells persistently infected with viruses may also constitute what seems to be oncogenic process. Rouse and colleagues (1973) suggested many years ago that the T cell proliferation and invasion that characterises Marek’s disease in the chicken reflects the CTL response to tissues infected with the virus. The fact that there is a very good vaccine that prevents infection with the Marek’s disease herpesvirus (Biggs, 1975) may give some hope for breaking the cycle in such infections/proliferative processes.

B. IMMUNOSUPPRESSION A N D TRANSPLANTATION There is now ample evidence that the incidence of some tumors is dramatically increased in people who have received organ transplants with accompanying immunosuppressive therapy (reviewed by Kinlen et al., 1979; Mitchison and Kinlen, 1980; Penn, 1981; Shiel, 1982). However the occurrence of many solid tumors is not greatly modified by this process, and Hodgkins’ disease is relatively rare in transplant patients (2%, cf. 34% of lymphomas in other individuals; Penn, 1981). This is of interest, as Hodgkins’ disease is one of the few tumors for which there is a clear association between susceptibility and HLA type (Section X). On the other hand, the frequency of non-Hodgkins lymphoma in transplant patients is 45-100 times higher than that seen in the population at large. At least some of these cases are associated with EBV (Purtillo, 1981). The most coininon neoplasm found in transplant patients in areas where there is high exposure to ultraviolet (UV) light is squamous cell carcinoma (Shiel, 1982). In a series of 290 Caucasian patients receiving cadaveric renal transplants in the subtropical city of Brishane 28 developed cancer, with the incidence of skin cancer in this group being 93% (Hardie et al., 1980). It thus

T C E L L SURVEILLANCE A N D M H C

11

seems likely that some skin cancers are normally kept in check by T cell surveillance mechanisms. This idea is supported by a series of elegant experiments in the mouse model, showing that UV-induced skin tumors are highly antigenic when transferred into normal, syngeneic recipients (reviewed by Kripke, 1981). Under these conditions, strong CTL responses are seen which distinguish between individual UV-induced tumors. However, in the primary tumorbearing host, this effector T cell response is apparently modulated by antigen-specific suppressor T cells. The truly surprising feature of this suppression is that UV-irradiated animals are unresponsive to UV-induced (though not to methylcholanthrene-induced) tumors of different histological types and (from the aspect of CTL response) distinct antigenicity. The establishment of this “tolerant” state may depend on preferential expansion of T suppressors in the absence of appropriate presentation of tumor components by antigen-processing Langerhans cells in skin, which are damaged by UV irradiation (Streilein and Bergstresser, 1980; Elmers et al., 1983). The development of malignancy at a higher incidence in transplant patients is apparently not due solely to the immunosuppressive drugs used to facilitate acceptance of the graft. A follow-up of 3823 renal transplant patients that were treated with azathioprine, cyclophosphamide, and chlorambucil showed a 60 times increase in non-Hodgkins lymphoma and an excess (23X normal) of squamous cell carcinoma (Kinlen et d., 1979). The comparable finding for 1349 people given immunosuppressive drugs alone was a 12 times increase in non-Hodgkins’ lymphoma and a 6 times increase in squamous cell carcinoma. Observations of this type have led to the concept that continued antigenic stimulation resulting from the presence of the graft contributes to the onset of malignancy. A similar case has been made for EBV, malaria, and Burkitts’ lymphoma, and for HTLV and Kaposi’s sarcoma (Section IV,A). The other feature of non-Hodgkins’ lymphoma in transplant patients is the high incidence of tumor localization to the central nervous system (Penn, 1981). This could reflect that the immunosuppressive drugs used damage the blood-brain barrier, in addition to possible subversion of T cell surveillance mechanisms which might be expected to limit metastasis. Conclusions: The analysis of both human and experimental tumors for which there is known, or strongly suspected, viral etiology indicates that T cell surveillance mechanisms are centrally involved in limiting the emergence of such tumors. Additional evidence for the efficacy of T cell monitoring against some tumors is the high incidence of non-Hodgkins’ lymphoma and squamous cell carcinoma in immunosuppressed transplant patients. In both situations, continued antigenic stimulation by the graft or by virally modified cells may be an important contributing factor potentiating the

12

PETER C . DOHERTY ET AL.

induction of carcinogenesis. Perhaps it will eventually be shown that all tumors that are subject to T cell surveillance involve viruses. However, there is currently no evidence that this is the case for squamous cell carcinoma in man, and the antigenic variability of UV-induced tumors in the mouse (Kripke, 1981) would not seem to support such a “pan-virus” hypothesis.

V. MHC-Restricted T Cell Recognition

The basic fact of MHC-restricted T cell recognition is that T lymphocytes are constrained to interact with cell surface, rather than with free antigen, by the need to see one or another MHC glycoprotein. The helper-inducer (Th) and delayed-type hypersensitivity (DTH) T cells are targeted onto the Class I1 MHC antigens, while cytotoxic effectors interact with the Class I glycoproteins. The exception to the rule may be the suppressors (Ts), at least some categories of which can bind to free antigen. A number of exhaustive reviews of MHC-restriction are already available (Section I). This account concentrates on aspects that are relevant to the present discussion.

A. T CELL

SPECIFICITY FOR

MHC

DETERMINANTS

The nature of MHC-restricted T cell recognition is most readily summarized for the CTL, as the interaction involves (at least at the effector stage) only two cells-the lymphocyte and the target. For simplicity of discussion it is convenient to designate different MHC glycoproteins as A, B, and C, and foreign non-MHC antigens as x, y, z. Other self-determinants are referred to as S,, S,. The phenomenon was discovered, and the term MHC-restriction coined, when it was found that virus and hapten-specific CTL that were sensitized in the context of A + x were not lytic for targets expressing B + x, A y, or A + S, (Doherty and Zinkernagel, 1975; Zinkernagel and Doherty, 1974a; Shearer, 1974). This led to the concept that the lymphocytes (in the virus model) were recognizing “altered self,” which might be thought of as “changes in the H-2 antigens produced by the process of virus synthesis or as some complex of viral and H-2 antigen” (Zinkernagel and Doherty, 1974b). The former possibility was made much less likely b y the discovery that CTL responsive to minor H determinants, particularly the male HY antigen, were also MHC restricted (Bevan, 1975; Gordon et al., 1975). Thus the generation of A + x as an immunogenic entity does not depend irrevocably on either an infectious process or on direct derivitization of the MHC glycoprotein with hapten (Forinan et al., 1977). However, the possibility that noncovalent

+

13

T CELL S URVE IL L ANCE A N D M H C

interactions between A and x on the cell surface lead to allosteric changes in A that are recognized by the CTL is still not excluded (Section VI1,B). The central fact that either A, or some variant of A, must be recognized by the T cell specific for A + x has now been rigorously established. The nature of x is much less clearly understood, and will be discussed later (Section VII). Recent molecular studies (Section VI) using gene transfection protocols have shown definitively that MHC-restricted T cells are interacting with the N and C1 domains of the MHC glycoprotein (Ozato et al., 1983; Reiss et al., 1983). Experiments with T cell clones (Sherman, 1980, 1982; Hurwitz et al., 1983a,b) and the use of monoclonal antibodies to block T cell function indicate that different determinants on N and C I may be recognized by T cells specific for A + x or A + y (Fischer-Lindahl and Lemke, 1979; Blanden et al., 1979; McMichael et al., 1980). Much of this blocking with monoclonal antibodies is probably steric, and does not reflect that the Ig molecules are binding to the exact site seen by the T cell (Allouche et al., 1982). Though the effector CTL are all targeted onto Class I MHC determinants, the exclusive specificity of individual T cells for a particular Class I MHC glycoprotein has been somewhat overstated. The reason for this is that the T cell response is operationally determined by the stimulating antigen(s), A + x. This means that a population of T cells is selected, with the various clonal elements presumably having differing affinities for A x. The overall impression is of a T cell response that is very precisely directed at A + x. However when cloned T cell lines are derived from the population it is found that, though many are apparently only lytic for cells expressing A x, a few will show a degree of cross-reactivity with B x (Cerottini, 1980). Also, a cloned line has been generated in an (A x B)F, situation which will recognize cells presenting either A + x or B y (Hunig and Bevan, 1982). In addition, there are numerous instances where T cells selected to interact with A x will also lyse targets expressing an unrelated alloantigen, C (Section IX,D). The overall impression that emerges is that the major constraint governing T cell effector function is the requirement for the lymphocyte to have sufficient affinity for the MHC glycoprotein on the target cell. Any T cell developing in an A individual must obviously be tolerant of A. Stimulation to effector function will presumably only occur when the cell surface is modified by expression of x, and the T cell has specificity for A + x. However, in the final analysis, this may only serve to focus the T cell onto A. In the case of an alloantigen B, which is not encountered during T cell ontogeny, recognition of the MHC glycoprotein alone is presumably sufficient.

+

+

+

+

+

B. THE T CELLREPERTOIREA N D Zr GENES Analysis of the T cell repertoire has been hampered by our lack of understanding of the nature of the T cell receptor. This situation may be largely

14

PETER C;. DOHERTY ET A L .

remedied by the time that the present article is published. Earlier ideas that the binding site on the T cell receptor is encoded by immunoglobulin V,, genes (Janeway et al., 1976; Binz and Wigzell, 1977; Eichmann, 1978) do not seem to have stood the test of time (Kurosawa et al., 1981; Jensenius and Williams, 1982; Marrack and Kappler, 1982; Keinp et al., 1983; Kraig et al., 1983). It was always difficult to understand how, given the fact of MHCrestricted T cell recognition, identical genes could be encoding both Ig V,, regions and T cell receptor binding sites involved in antibody and CMI responses to the same antigen (Doherty et nl., 1977a). At a minimum, the T cell would need to be using a different part of the Ig repertoire (Doherty and Bennink, 1979). Current studies with T cell clones and hybridomas (Haskins et al., 1983; Reinherz et al., 1983)indicate that the receptor inay consist of two associated chains, each of MW 40,000-50,000, in both mouse and man. These chains are linked b y intermolecular disulfide bonds (Samelson et al., 1983). This disulfide-linked heterodimer apparently constitutes the clonotypic, MHCrestricted T cell receptor. However, it seems that, at least in the human system, this is further cornplexed with the inonomorphic T3 chain (MW 20,000-25,000) which inay recognize a constant region of the MHC glycoprotein. Sepharose-bound monoclonal antibodies to either the clonotypic heterodimer or to the conserved T3 can activate T cell proliferation (Meuer et al., 1983). The genes coding for one chain of the T cell receptor have now been cloned and sequenced (Marx, 198313;Yanagi etul., 1984; Hedrick et al., 1984). Evidence has been found of considerable homology with Ig light chain, though the two may have diverged before speciation to mouse and man. There are also similarities between Ig and MHC genes (Section VI). Perhaps genes encoding all these molecules with some recognition function constitute a supergene family descended from a common ancestral pool (Williams, 1984). Application of this molecular technology might be expected to result in a rapid resolution of the problem of the T cell receptor and the T cell repertoire. However, at this stage, we are still in the situation that any discussion of T cell repertoire is a simplistic analysis cast in the sense of interactions between the lymphocyte and the target, or stimulator. The constant molecular entity that is definitively identified as a participant is the MHC gene product (A, B, or C) on the surface of the target (Section VI). We also know that the target must be modified by expression of a neoantigen (x, y, and z), and that other self components (Sl, S,) are immunogenic when used to sensitize MHC-identical but minor H-antigen-different siblings or congenic mouse strains (Section VIII). Furthermore, any discussion of affinity can only refer to this cell-cell interaction, not to the characteristics of molecular binding events (Owen et al., 1982a).

T CELL SUHVEILLAN(:E A N D M H C

15

The physiological development of the T cell repertoire in an (A x B)F, individual is constrained by the need to generate a capacity for immune responsiveness directed at self MHC components A or B presented in the context of neoantigens x or y. However, at the same ti.ne it is essential that effector T cells that can focus onto A + S,, or B + S, do not emerge as this would lead to autoimmunity. The potential spectrum of T cell responsiveness is thus influenced b y the need to ensure self tolerance to A + S,, A S,, B S,, B + S,. The consequence is that, if A S, is cross-reactive with B x, there will Ile a specific “hole” in the T cell repertoire for B + x. This was first discussed by Langman (1978) and Schwartz (1978) and is considered by Klein and Nagy (1983)to be the main mechanism governing failure of T cell responsiveness, or immune response (Zr) gene effects mapping to the MHC (Benacerrafand McDevitt, 1973). There is some debate as to whether these “tolerance-induction” events occur principally in bone marrow, in thymus, or in both sites (Besedovsky et al., 1979; PhillipsQuagliata, 1980; Doherty and Bennink, 198011). Considerable disagreement also exists as to whether or not the extent ofthe T cell repertoire is expanded to operate principally in the context of A following events occurring in, for instance, the thymus ofan [(A x B)F, + A] radiation chimera (Zinkernagel, 1978; Bevan and Fink, 1978). This idea has been discussed at length (reviewed by Zinkernagel and Doherty, 1979; Howard, 1980; Doherty and Bennink, 1980b; Doherty et al., 198lb) and has its genesis in an early suggestion of Jerne’s (1971) that the T cell repertoire in an A animal is derived principally via mutational events occurring as a result of the developing thymocytes recognizing A in thymus. Klein and Nagy (1983) refer to the concept as the “individualization hypothesis” and, after considering the available evidence in great detail, are dubious about the usefulness of the idea. However, the jury is still out on the individualization hypothesis. The resolution will probably not come from further cellular immunology experiments, but from the analysis of genes coding for the T cell receptors by the molecular biologists (Marx, 198311). If the individualization hypothesis is not true the repertoire may appear to be germline, being essentially similar in A and B mice which are congenic for other non-MHC genes (Marrack and Kappler, 1982). Otherwise, there should be dramatic differences between T cells developing in, for instance, [(A x BF,) + A] and [(A x BF,) + B] radiation chimeras. This could obviously lead to considerable variation in the spectrum of immune responsiveness. Whether or not a CMI response can develop thus depends on there being a specific receptor configuration on the T cell which interacts with an approx on the target. Variations in the magnitude of priate organization of A response associated with a particular neoantigen (x) presented in individuals with different M H C types (A and B) are referred to as “Zr gene” effects (Benacerraf and McDevitt, 1972). The “Zr genes” were, for many years,

+

+

+

+

+

16

PETER C. DOHERTY ET AL.

solely the province of those working with Class I1 MHC-restricted responses. As a consequence, the unfortunate description “ I region” was adopted for the loci encoding the Class I1 genes in the mouse. The latecomers to the field, investigating the Class I MHC-restricted CTL, distanced themselves from this terminology. The reason was that the then accepted paradigm proposed that the Zr genes encoded all, or part of, the T cell receptor(s) (Benacerraf and Katz, 1975). However, as a result of the somewhat different insights (reviewed by Matzinger, 1981; Klein and Nagy, 1983; Robertson, 1983) that could be generated with the CTL system (Zinkernagel and Doherty, 1974b; Doherty and Zinkernagel, 1975) the “Zr gene-lymphocyte receptor” idea was soon abandoned (Benacerraf, 1978) by most, though not all, workers in the field. As a result, it then seemed appropriate to extend the use of “lr gene” to also cover CTL responses (Zinkernagel et al., 1978; Doherty et al., 1978). The essential point is that both the Class I and Class I1 MHC genes are Zr genes. If the MHC glycoproteins have any receptor-acceptor role, this probably operates at the level of a physiological interaction between A and x on the plasma membrane of the target cell. The proof that the Zr gene product is indeed the MHC glycoprotein on the target/stimulator cell has now been provided abundantly from both genetic mapping and gene transfection experiments (Jones et al., 1978; Section VI). However, until we have a better understanding of how the molecular entities A and x are organized with respect to one another on cell surface (Section VII), it is apparent that any “Zr gene” effect can be defined only in the context of the available T cell receptor specificities. Again, the analysis of the germline T cell repertoire, the role of clonal deletion or suppression to ensure self-tolerance, and the selection associated with physiological differentiation processes occurring in the pre-, intra-, and postthymic environments is central to our understanding of T cell recognition and responsiveness. At this stage the most rigorously defined information about the nature of Zr gene effects is provided by the current molecular analysis of the MHC genes and glycoproteins. VI. The Molecular Nature of the MHC

The application of contemporary techniques in protein chemistry and molecular biology over the past few years has resulted in tremendous advances in our understanding of the MHC (reviewed by Coligan et al., 1981; Nathenson et al., 1981; Hood et al., 1982; Winoto et al., 1983; Steinmetz and Hood, 1983). The field is moving very rapidly: the present account will obviously be somewhat dated by the time that this review is published.

T CELL SURVEILLANCE A N D M H C

17

A. CLASSI GENES:RECOGNITION A N D POLYMORPHISM The Class I MHC glycoproteins are two-chain structures (Coligan et a l . , 1981) consisting of a more variable transmembrane polypeptide of M W 45,000 encoded within the MHC (mouse chromosome 17) and an attached, relatively conserved P,-microglobulin subunit (MW 11,500) which is encoded on mouse chromosome 2. The heavy chain comprises three external hydrophilic domains (N, C1, and C2) each of about 90 amino acid residues, a hydrophobic transmembrane segment (approximately 40 residues), and a short cytoplasmic region (about 30 residues). There are two carbohydrate prosthetic groups linked to asparagine residues on N and C1, while the P2microglobulin is nonglycosylated. The recognition event involving cytotoxic T cells seems to be concerned with the N and C1 domains, which are most remote from the cell plasma membrane, while monoclonal antibodies may bind to N, C l , and C2 (Ozato et a l . , 1983; Reiss et al., 1983). It is now clear from DNA sequence studies that there are many more Class I genes than would have been expected from knowledge that the transplantation antigens (H-2K, D, and L) can act as restricting elements for cytotoxic T cells (Steinmetz et d., 1982; Hood et al., 1982; Winoto et al., 1983). The MHC encompasses about 2 cM of DNA, which may include from 2000 to 4000 kilobases. There are currently 36 known Class I MHC genes in the mouse that can be divided by restriction map analysis into 13 gene clusters. However, only 5 of these 36 genes map to the H - 2 region (about 500 kb of DNA between H-2K and H-2D,L). The other 31 map to the TLa complex, which is also part of the MHC (about 2000 kb of DNA to the right of H-2D, L) but is thought not to be associated with self-MHC-restricted T cell recognition. One cluster of 7 genes has been localized to the Q a , 2,3 region (Steinmetz et al., 1982) and there is evidence that Qa determinants may serve as targets for alloreactive T cells (Forman and Flaherty, 1978). The other Class I genes in the TLa region have no known function, though it is possible that at least some of them may code for differentiation antigens that are expressed transiently at stages throughout ontogeny. It may be of interest to probe tumor lines for the expression of such genes. One of the major questions about the MHC has concerned the extreme polymorphism of the Class I, as distinct from the Class II, genes encoding the strong transplantation antigens (Amos et al., 1971; Doherty and Zinkernagel, 1975). Analysis of the 36 Class I genes with cDNA probes specific for the 5’ and 3’ ends reveals that the exon encoding the third external domain (C2) is much more conserved than those encoding the first and second (N and C1) external domains (Steinmetz et al., 1982). This fits well with knowledge of MHC-restricted CTL specificity that has been gleaned from experiments with H-2 mutant mice (Section IX,B). The central role of the external

18

PETEH C. DOHERTY ET AL.

segments of the Class I glycoprotein in CTL recognition has now been shown conclusively with recombinant genes constructed from restriction endonuclease fragments of cloned H-2D“ and H-2Ld genes, so as to exchange the exons coding for N and C1. This “exon-shuffling” protocol has been used to demonstrate that T cell recognition of allogeneic, and virus-infected syngeneic (to H-2d) mouse L cells that have been transfected with the recombinant H-2“ genes is concerned with N and C1, not C2 (Ozato et al., 1983; Reiss et al., 1983). It now seems likely that the extreme polymorphism of the Class I MHC genes (more than 100 alleles at H-2K and H-2D) is largely generated from rearrangements of existing gene sequences within the MHC (Steinmetz et al., 1982; Pease et al., 1983). Analysis of the H-2K””1 mutant series has provided much of the evidence for the proposition that polymorphism results mainly from block transfer of sequences from other Class I MHC genes (Pease et al., 1983; Schulze et al., 1983). Point mutation seems an unlikely explanation, as the complex substitutions found in the H-2Kb?I1mutants require sequential, clustered nucleotide base changes. In addition, “new” DNA sequences found in the mutants may also be identified in genes coding for other Class I glycoproteins (e.g., H-2L“) (Evans et al., 1982). The mechanisms underlying M HC polymorphism is currently thought to reflect copy transfer analogous to gene conversion in yeast (Lopez de Castro et al., 1982; Schulze et al., 1983). This involves a nonreciprocal recombination event, by which a particular segment of one gene becomes incorporated into the corresponding portion of another related, but nonidentical gene, (Baltimore, 1981). Both genes retain their integrity and physical location, but a nonreciprocal alteration occurs in the structure of one partner. The existence of 36 different Class I MHC genes would seem to allow ample scope for such interactions to occur. Selective pressures to do with the need for H-2KD-restricted CTL recognition (Doherty and Zinkernagel, 1975; Doherty, 1980a,b) may be the reason why such extensive variation in Class I MHC glycoproteins is normally only recognized for the relatively small H-2 complex, and not for the 31 genes associated with the TLa complex. The fact that the spectrum of self seen by self-monitoring T cells may be redefined for each individual by the range of H-2KD molecules encountered during ontogeny (Zinkernagel, 1978; Bevan and Fink, 1978; Doherty and Bennink, 1980b) means that a mouse expressing any variant H-2KD glycoprotein that can potentially be recognized by self-monitoring T cells is likely to survive and reproduce. However, if the Class I genes mapping to the TLa complex encode differentiation antigens involved (for instance) in organogenesis, it is likely that these genes would need to be highly conserved (Klein, 1975). The tumor cell lines expressing the so-called “alien” MHC glycoproteins (Section IX,C)

T CELL S U R V E I L L A N C E A N D M H C

19

might well be reexamined from the aspect that such gene conversion, possibly also involving sequences in the TLu complex, inay have occurred. In fact, a major function of the Class I genes which do not map to H-2K,D,L may be to provide genetic material to ensure a high level of polymorphism at these loci. Conclusions: There are 36 known Class I genes in the MHC, 5 of which map to the H - 2 region and may be involved in T cell function. The extreme polymorphism associated with the external N and C1 domains of Class I MHC glycoproteins mapping to H-2K, D, I, (or HLA-A,B) may be explained by mechanisms involving block transfer of sequences (gene conversion) between Class I genes already present in the genome. The finding that extensive polymorphism is limited to the relatively sinall number of genetic loci mapping to the H-2, as distinct from the TLa complex, inay reflect selective pressures operating via M HC-restricted T cell function.

B. CLASSI1 GLYCOPROTEINS: POSTTRANSLATIONAL ASSOCIATIONS AND

DIVEHSITY

The Class I1 MHC glycoproteins serve as focusing sites for T cells of the helper-inducer subset, and have long been known to be involved in determining levels of immune response (reviewed by McDevitt, 1981; Hedrick et al., 1982; Matis et al., 1982; Klein and Nagy, 1983). Those of mouse (reviewed by McNicholas et al., 1982a,b; Mathis et al., 1983) and man (reviewed by Shackelford et al., 1982; Larharniner et al., 1982; Lee et al., 1982) are structurally comparable, and are considered together for the purpose of this discussion. The HLA-DR antigen coinprises two, noncovalently associated glycoprotein subunits. Unlike the Class I MHC glycoproteins, both are integral membrane proteins that have their COOH-terminal regions on the cytoplasmic aspect of the cell membrane. The heavy (MW 33,000-44,000) and light (MW 27,000-30,000) chains consist of two external domains each of about 90 or more residues (a1and a2, p l and p2), a transmembrane region of about 20 to 30 residues and a cytoplasmic region of 10 to 15 residues. The heavy chain is largely invariant, while the light chain seems to carry the major polymorphic determinants. Detailed analyses correlating structure and function have been made for the mouse Ia glycoproteins, found mainly on B lymphocytes and macrophages. In mice homozygous for, for instance, the H-2k haplotype, two distinct Ia glycoproteins are found (Mathis et al., 1983) each consisting of an OL (MW 34,000) and a p (MW 29,000) subunit. The A complex (A,:AB) maps to the I-A region of the MHC, while the B complex (E,:EB) is encoded at both I-E (E,) and I-A (Ep). A variety of mouse strains (of the H-2”, H-2>,

20

PETER C. DOHERTY ET AL.

H-2q and H-2q haplotypes) fail to express the E complex: this can be correlated with diminished immune responsiveness to some antigens. The reasons for absence of the E complex include a deletion of the 5‘ end of E , (b and s), aberrant E, mRNA (Q,and a defect in E , RNA processing or stability (Mathis et al., 1983). It seems that the E , gene, which exhibits much less sequence diversity than A,, is of limited selective advantage, while defective A, genes are not found and A, may thus be essential for survival. A major mechanism for generating Ia polymorphism is that posttranlational chain reassociations can occur in mice heterozygous at the H - 2 complex, between A,:A, and E,:E,. This allows the possible expression of as many as 8 different Ia complexes, a situation that cannot occur with the Class I glycoproteins because one chain (&,-microglobulin) is essentially invariant. Such F, Ia antigens might be thought to provide a mechanistic basis for heterozygote advantage. However, difficulties arise for this argument when use is made of parental strains that do not express the E , gene (McNicholas et al., 1982a; Matis et a l . , 1982). Also, even in situations where both E , genes are fully functional, the level of expression of any complex in the F, may be only half that found in either parent. Even so, combining a and p chains from mice which are both low responders to a particular antigen (e.g., pigeon cytochrome c) may result in the formation of a new a$ complex which is associated with high responsiveness. Perhaps the responder T cells are recognizing “junction zones” between the two chains, or novel allosteric changes induced in one chain as a result of binding to the other chain. Another point that is of considerable interest in the evolutionary context has emerged from recent molecular studies (Kaufman and Strominger, 1982; Korman et al., 1982; McNicholas et al., 1982b; Larhammer et al., 1982). It seems that there is significant sequence homology between the a2 domains of E , and DR,, &-microglobulin, and regions of the immunoglobulin molecules where tertiary folds occur. These similarities are obvious for the relatively invariant portion of the Class I1 molecule that is closest to the cell membrane, while there is no homology in the more polymorphic p chain Nterminal region which bears the carbohydrate. The overall interpretation is, however, that such homologies indicate a common phylogenetic origin for Class I and Class I1 MHC glycoproteins and the immunoglobulins. Conclusions: The H-2 Ia antigens of the mouse are complexes (A,:Ap, E,:E,) of two trans-membrane glycoproteins encoded at Z-A or I-E. Novel complexes can be formed in heterozygotes due to the association of a and p chains from each H-2 haplotype. This allows 8 different Ia configurations in the F,, though mice of some H-2 types do not phenotypically express the E , gene and the number of possible Ia specificities is thus decreased. The formation of such complexes in the heterozygote may allow the emergence of high responsiveness to a particular antigen that is identified with low respon-

T C E L L SURVEILLANCE A N D M H C

21

siveness in either parent. The HLA-DR glycoproteins of man seem comparable in every way to H-2 Ia, though there is accumulating evidence that they may show a much higher level of polymorphism. VII. T Lymphocyte Effectors and Tumor-Specific Transplantation Antigens

Analysis of the inflammatory infiltrate in experimentally induced tumors indicates that there is selective invasion by CTL that are principally reactive to the tumor in question (Gillespie et al., 1978; Chapdelaine et al., 1979; Brunner et al., 1981; Ting and Yang, 1982). For instance, in the MSVMoLV model tumor-specific CTL precursors are present at a frequency of % Lyt 2 + T cells in the tumor mass comparable with 1/42 for the same set of lymphocytes in peripheral blood (Brunner et al., 1981). The existence of tumor-specific CTL in human cancer may also be inferred from studies with EBV (Section IV,A) and from the observation that precursors in peripheral blood can be stimulated to mediate specific lysis of autologous tumor cell lines or recently excised tumors (Vose and Bonnard, 1982; Vanky and Klein, 1982; Mukherji and MacAlister, 1983). There is thus no obvious reason to doubt the reality of tumor-specific CTL responses in at least some categories of cancer, though the generation of appropriate effector populations may be modulated by the involvement of suppressor T cells (Berendt and North, 1980; Greene, 1980; Frost et al., 1982; Ting and Zhang, 1983). Understanding these interactions may constitute the main hope for the useful application of the cellular immunology approach for the control of malignancy. Experiments with both virally and chemically induced tumors indicate that the majority of such tumor-specific CTL populations are MHC restricted (Levy and Leclerc, 1977; Kaneko et al., 1978; Greenberg et al., 1981; Poupon et d.,1981; Green et al., 1980, 1982; Lannin et al., 1982; Colombo et al., 1983; Ahrlund-Richter et al., 1983; Korngold and Doherty, 1984, and Sections IV and IX). There may be exceptions to this rule (Stutman and Chen, 1978; Giorgi et al., 1982) though, even if evidence of MHC restriction (Lee,a requirement for a particular allelic Class I MHC glycoprotein) cannot be found this does not exclude the possibility that a more constant region of the MHC molecule is being recognized (Plata, 1982). Also, the most thoroughly studied exception to the MHC restriction rule concerns Ig+ plasmacytoma cells (Burton et al., 1977). Perhaps, despite other evidence to the contrary (Kaneko et al., 1978), the Ig molecules alone can serve as the elements that focus the effectors onto cell surfaces (Snodgrass et al., 1981; Giorgi et al., 1982). The assertion that most, if not all, tumor-specific CTL must operate via interaction with self MHC glycoprotein still seems a reasonable one.

22

PETEH C . DOHERTY ET AI..

Any discussion of T cell specificity for tumors must also include consideration of the helper-inducer (Th) lymphocytes. Generation of a potent Class I MHC-restricted tumor-specific CTL response is probably dependent on the concurrent development of Class I1 MHC-restricted Th populations (Fujiwara et ul., 1980; Gomard et a l . , 1981; Yu and Bernstein, 1982; Mills and North, 1983). The possibility thus exists that T cell surveillance mechanisms will operate effectively only if the neoantigen(s) characteristic of the tumor is appropriately presented in the context of both Class I and Class I1 MHC glycoproteins (Sections V and VI). The latter will, unless the tumor is H-21a or HLA-DR positive (Section IX, E), probably also require the reprocessing of tumor-specific components by antigen-processing macrophages or dendritic cells (Biasi et a l . , 1983). A considerable amount of information is available concerning the specificity of both CTL and Th populations for MHC glycoproteins (Sections V, VI, IX). What is known of the antigenic entities that we describe as non-self, or tumor-specific?

A. T CELLSPECIFICITY FOR VIRUSES Some insight into the complexity of the problem of understanding the nature of T cell specificity for neoantigens presented on cell surface may be gained by considering the examples provided by conventional infectious viruses, such as the influenza A viruses (Doherty et a l . , 19778; Zinkernagel and Rosenthal, 1981; Greenspan et a l . , 1983). The influenza A viruses have been of general interest as they provide a naturally occurring biological system for reassortment of various antigenic components (Webster et a l . , 1982). However, the capacity to use genetic engineering technology to express isolated viral genes in cells (Benjamin, 1983) is now making this approach historical. Even so, it is worth considering the topic briefly as a cautionary tale for those attempting to analyze T cell specificity for tumorassociated surface antigens.

1 . The Complexity of a Well-Churacterised System: Influenza A Viruses and T Cell Recognition The influenza A viruses are enveloped viruses with single-stranded RNA of nonmessage polarity (reviewed in Kilbourne, 1975; Palese, 1977; Webster et al., 1982). The genome consists of 8 segments which may segregate independently when two different influenza A viruses are grown in the same cell. This provides a simple method of generating “recombinants,” which reassort the various virus proteins. The surface of the virus particle (or virion) presents two major glycoproteins, the hemagglutinin (HA) and neuraminidase (N), which appear as spikes on electron microscopy. The HA molecules are organized as trimers, while the N are tetramers. Protection afforded by

T C E L L SURVEILLANCE A N D M H C

23

serum antibody is principally directed at the HA, though the N may also be involved. Differences between influenza A virus subtypes are reflected in lack of serological cross-reactivity between the HA and N, there being a complete absence of reciprocal neutralization between viruses designated HAlNl and HA3N2. However, the internal (in the virion) matrix (M) and ribonucleoprotein (RNP) antigens of HAlNl and HA3N2 viruses are very similar, though they are different from those found in the influenza B virus. Effector CTL generated in response to infection with influenza A viruses are functionally MHC restricted and are not lytic for H-2 compatible targets infected with influenza B viruses (Effros et al., 1977; Doherty et al., 1977a; Zweerink et al., 1977; McMichael and Askonas, 1978). However, though a relatively small proportion of the CTL clones that are generated following infection of mice appear to be HA specific (Effros et al., 1977; Braciale et al., 1981a), the great majority (about 90%) do not distinguish between targets infected with HAlNl and HA3N2 viruses (Askonas et al., 1982; Owen et al., 1982). This is surprising, as the HA and N molecules are by far the predominant viral proteins presented on the surface of the infected cell. Is it possible that many of these cross-reactive CTL are recognizing a more conserved region of the HA (Koszinowski et al., 1980) which is serologically silent? However, CTL generated following infection with influenza A and influenza B viruses show reciprocal specificity, yet different influenza A HA molecules may have no greater sequence homology with each other than with influenza B HA glycoproteins (Krystal et al., 1982). Another possibility is that the cross-reactive, influenza A-specific CTL may be recognizing internal viral components expressed on the plasma membrane of infected target cells. Analysis with monoclonal antibodies indicates that significant amounts of the influenza virus RNP antigen are present on cell surface (Virelizier et al., 1977; Yewdell et al., 1981). Similarly, even smaller quantities of M protein may be expressed (Hackett et al., 1980; Yewdell et al., 1981). However, any conclusion that these relatively conserved M and RNP antigens are indeed recognized by the CTL would be premature. Monoclonal antibodies directed at these proteins have not yet been shown to block T cell-mediated lysis (J. Yewdell and J. Bennink, personal communication). The use of recombinant viruses to analyze recognition patterns for the relatively rare CTL clones that seem to be HA-subtype specific has also provided surprises. The experiments of Townsend and Skehel (1982) indicate that the genes responsible for the determinants recognized by such lymphocytes segregate independently from those coding for both the HA and N glycoproteins. Another “HA-specific” CTL clone has been shown to recognize an antigenic entity that is in some way influenced by genes coding for both the RNP and a viral polymerase (Bennink et al., 1982). Again, it has

24

m r m c. DOHEHTY ET AL.

not been possible to block influenza-specific CTL clones with monoclonal antibodies directed at the HA glycoprotein, though some inhibition has been observed in experiments with bulk T cell populations (Effros et al., 1979). The situation for HA-specific Th clones is somewhat different. Two separate studies have shown that at least a proportion of such T cells are recognizing relatively conserved (encompassing the HA1 but not the HA3 glycoproteins) regions of the HA molecule, which may tend to be serologically inert (Lamb et al., 1982; Hackett et al., 1983). Perhaps this divergence between the CTL and Th populations reflects that stimulation of the Th clones requires that the antigen be processed and presented on the surface of an appropriate macrophage or dendritic cell. On the other hand, the antigenic entity that is seen by the CTL may need to be inserted “of itself,” in this case as a result of an infectious process, in the plasma membrane of the target. Conclusions: The predominant antigenic entities expressed on the surface ofcells infected with influenza A viruses are the viral HA and N glycoproteins. However, conclusive evidence that either of these inolecules is recognized by most (or any) influenza-immune CTL has not been easy to generate, though it is apparent that some Th clones are reactive to HA determinants. In addition, analysis with recombinant viruses indicates that the antigen(s) of interest to the few CTL clones which appear to be HA-specific may not be coded for solely by the genes which specify the viral HA or N. Perhaps this reflects the evolutionary need to focus CTL populations onto cell surface by ensuring both that their receptors are not saturated with free virus, and that they will not be blocked by serum antibodies that bind to viral determinants expressed on the plasma membrane of the target. The lesson is that demonstration of a novel glycoprotein on the surface of a tumor cell with (for instance) monoclonal antibody offers no guarantee that the molecule in question will be recognized by surveillance T cells. Also, simple correlation of T cell specificity with expression of a particular non-self glycoprotein may be misleading.

2 . Tumor Viruses Most of the studies concerned with assessing the contribution of various viral components to the recognition of tumor cells by CTL have utilized retrovirus models (Reviewed by Levy and Leclerc, 1977; Zinkernagel and Doherty, 1979). Attempts at blocking CTL function with serum antibodies directed against viral gp45, pr60, p30, p15, p12, and p10 have been unsuccessful (Gomard et al., 1978). Some inhibition of cytotoxicity was observed with one antiserum specific for gp70, but several other antibodies to this glycoprotein have been without effect (Gomard et al., 1978; D. Zarling et al., 1978; Green et al., 1980). However, monoclonal antibodies to H-2Db will inhibit a range of MSV-MoLV-specific CTL clones (Weiss et al., 1981). Using Friend virus-transformed cell lines Collins et al. (1980)found that

T C E L L SURVEILLANCE A N D M H C

25

the magnitude of CTL-mediated lysis was directly proportional to the amount of viral gp70 (coded for by the enu gene) found on cell surface. Furthermore, target cells transfected with the enu gene have now been shown to be recognized by at least a proportion of Moloney-MuLV-specific CTL (Flyer et al., 1983), though it is possible that other viral proteins may also be recognized by MuLV-specific T cells. Other studies with the Gross virus model correlated CTL specificity with expression of the Gross associated cell surface antigen (GCSA), encoded by the gag gene of Gross-MuLV (Green, 1980). Bulk CTL populations were found to be highly cross-reactive for targets transformed by viruses of the Friend-Moloney-Rauscher (FMK) groups, but are much less likely to recognize tumors expressing GCSA (Plata and Lilly, 1979). Analysis of these specificity patterns using CTL clones showed that more than 50% were lytic only for the immunizing tumor, whether FMR- or Gross-induced, while less than 10% were cross-reactive (Plata, 1982). Monoclonal antibodies to the P30 antigen encoded by the gug gene of Gross-MuLV have now been shown to block cytolysis by at least some clones of virus-specific CTL (Plata et al., 1983). The specificity patterns for CTL in the retrovirus models thus tend to correlate with a requirement for the presence of identifiable molecular entities. Patterns of CTL specificity (Campbell et al., 1983) also correlate well with known cross-reactivities for the large T antigen expressed on the surface of papovavirus-transformed cells (Soule et al., 1982). Priming with the large T antigen alone is capable of inducing tumor-specific immunity against SV40transformed tumor cells in mice (Chang et al., 1979). Effector CTL clones generated in the SV40 model may, or may not, mediate lysis of targets that are transformed with the human BK virus and thus bear a cross-reactive T antigen (Campbell et al., 1983). In addition, an SV40-specific CTL clone was blocked by a monoclonal antibody to the T antigen (Pan and Knowles, 1983). Such CTL recognize targets transfected with isolated segments of the genes encoding the T antigen, with individual T cell clones being specific for different regions of the viral glycoprotein (Gooding and O’Connell, 1983). It thus seems apparent that the large T antigen is involved in the generation of the SV40-TSTA. Further discussion of the MHC restriction patterns governing recognition and tuinorigenicity in the SV40 model is presented elsewhere (Sections VIII and IX). Considerable emphasis has been placed in the retrovirus models on determining whether or not the virus, or a viral component, is physically associated with some Class I MHC glycoprotein (reviewed by Zinkernagel and Doherty, 1979; Giorgi et al., 1982). This was stimulated by the “altered self” hypothesis (Zinkernagel and Doherty, 19741-3)and by early experiments of Schrader et ul. (1975) which showed that Rauscher virus gp70 and H2 antigens would cocap on the surface of EL4 tumor cells. These cocapping studies

26

PETER C . DOHERTY ET A L .

were confirmed by D. Zarling et al. (1978) but not by Goinard et al. (1978) using the MSV model. Also, Fox and Weissman (1979) failed to demonstrate coprecipitation of viral gp70 and H-2 molecules, while Honeycutt and Gooding (1980) found the opposite result when low concentrations of detergent were used for membrane solubilization. Experiments using rat cell lines transformed with adenovirus 2 have produced strong evidence of an association between the adenovirus early protein (E19) and Class I MHC glycoprotein (Kvist et d., 1978; Kampe et d., 1983). An antiserum against the Class I antigen heavy chain coprecipitates E19. Another interesting aspect is that the total amount of p,-microglobulin is relatively reduced in these cell lines, perhaps reflecting that less pzmicroglobulin is bound by the Class I glycoprotein heavy chain as a result of conformational changes induced by the association with E l 9 (Kampe et al., 1983). An intriguing series of observations was made by Bubbers et al. (1978) who showed that Friend virus particles isolated from serum had selectively incorporated the MHC glycoproteins (H-2Kk and H-2Db) that were associated with CTL responsiveness (Blank and Lilly, 1977). Successful demonstration of the presence of these H-2 molecules required that the virus first be dissociated. Earlier studies of Hecht and Summers (1976) had also found that H-2Kk is present in budding vesicular stomatitis virus (VSV) and Hale (1980) demonstrated coprecipitation of H-2k and the VSV G protein. However, H-2k is associated with minimal CTL responsiveness in mice infected with VSV (Zinkernagel and Rosenthal, 1981). The significance of findings that mature virions, or particular viral components, are physically bound to one or another MHC glycoprotein is thus not always clear. Cocapping of viral and H-2 components (Senik and Neauporte-Sautes, 1979) could reflect that the molecules which seem to be associated are in some way attached to cominon elements in the actin-myosin skeleton of the cell (Bourguignon et aZ., 1978), but this does not explain the coprecipitation experiments. Another approach that has been taken to analyze host-response in the retrovirus models is immunization with subcellular fractions of the tumor (Klein et ul., 1983). However, these experiments have not, to date, been principally oriented toward understanding the nature of the specific molecular entities involved in T cell recognition, though they do indicate that at least part of the antigenic structure that is recognized is of viral origin. Reconstituted membranes from the MSV-MoLV-transformed MBL-2 cell line have also been used to stimulate secondary CTL in uitro in the presence of added 11-2 (Duprez et al., 1983). The immunogenicity of these preparations is enhanced by adding insoluble cellular matrix proteins, which contribute to the formation of vesicles that are thought to improve the characteristics of antigen presentation. This makes the point that the organization

T CELL S U R V E I L L A N C E A N D M H C

27

of molecules in plasma membrane is important in T cell recognition, though it adds little in this regard to earlier observations that such vesicles need to incorporate both Class I MHC and neoantigenic determinants in order to optimally stimulate CTL (Finberg et al., 197811; Hale et al., 1980). Conclusions: Correlations can be found between a requirement for expression of a particular retrovirus or papovavirus gene product and the specificity of recognition by cloned CTL. It seems apparent that the antigenic entity seen by SV40-immune CTL involves the large T antigen, and at least some retrovirus specific CTL are reactive to cell surface changes induced by expression of the env gene product gp70. There is also a body of evidence that retrovirus and Class I MHC glycoproteins may associate on the cell surface, but this is strongly disputed by some workers. Other experiments have demonstrated a close association between the early adenovirus protein E l 9 and rat Class I MHC glycoproteins.

B. WHATARE TUMOR-SPECIFIC TRANSPLANTATION ANTIGENS? We must face the situation that, with the possible exception of some plasmacytomas (Giorgi et a l . , 1982), recognition of tumors by CTL probably requires that the determinants unique to the tumor are presented in the context of self Class I MHC glycoproteins. The simplest case that can be envisaged is that the MHC glycoprotein itself is in some way modified as a consequence of the oncogenic change. The possibility that the MHC genes are altered by mutation, by gene conversion, or are not expressed following transformation (Section VI) is discussed in detail elsewhere (Section IX). It is not currently known whether such “mutant” MHC molecules will always be recognized as foreign by surveillance T cells. Obviously, if the latter were the case, the cells bearing them would tend to be eliminated and the tumor cells would not escape from immunological surveillance. The “mutants” that have been studied in depth to date were all selected on the basis of graft rejection in v i m , which is obviously a measure of T cell recognition and thus skews the analysis toward molecules expressing changes that are immunogenic. Experimentally, self MHC glycoproteins can be made antigenic by direct derivitization with haptens (Forman et al., 1977). The altered MHC molecule is then readily perceived as non-self by the CTL (Shearer and SchmittVerhulst, 1977; Sherman et al., 1979). Despite considerable effort, no success was achieved in attempts at demonstrating that these MHC-restricted, hapten-specific CTL clones have any measurable affinity for the hapten alone (von Boehmer and Haas, 1981). The T cells are apparently specific for the haptenated-self molecule. Thus, though such CTL can discriminate between MHC-identical targets modified with TNP and DNP (Forman, 1977),

28

PETER C . DOHERTY ET AL.

use of a “simple” entity such as a hapten does not allow us to make any precise conclusions about the nature of T cell specificity. A contrary conclusion might be drawn from studies with purified DTH T cells (Moorhead, 1981) and cloned Th lines (Lamb et al., 1983) which show that incubation with hapten, or with synthetic peptides of the influenza virus HA, is suppressive. Perhaps the antigen in question is binding directly into the T cell receptor, and an inhibitory “signal” results when there is no concurrent recognition of M HC glycoprotein on an appropriate dendritic cell (Van Voorhis et al., 1983) capable of delivering a stimulatory signal. However, the other possibility is that the antigen first interacts with MHC molecules on the surface of the T cells themselves, which then present a complex A x that is recognized by the clonotypic receptors on other cells. The antigen-presenting lymphocytes might be thought of as being comparable to the “veto” cells postulated by Miller (1980), which are considered to promote self-tolerance to the antigens that they bear by delivering an “offsignal” to the T cells that recognize them. Evidence in support of this idea is provided by the finding that incubation with antiserum to Ia prevents the tolerizating effect of the synthetic viral HA (Lamb et al., 1983; M. Feldman, personal communication), indicating that lymphocytes in the cloned T cell population are presenting antigen to each other in an MHC-restricted way. Also, suppression of CTL-priming in (A x B)F, mice that is specific for A + x, but not for B + x, results from the injection of T cells bearing A + x (Fink et al., 1983). If x was being recognized as an independent entity, lymphocytes reactive to B + x would also be affected. One possibility that has yet to be adequately investigated is that the insertion of foreign molecules into cell membrane, or the nonconvalent interaction between such a molecule and an MHC glycoprotein in the same membrane, will modify the MHC glycoprotein so that it is perceived as nonself by CTL mediating surveillance (Doherty and Zinkernagel, 1975; Cohen and Eisen, 1977). Such interactions could be of low affinity, perhaps reflecting recurrent association and dissociation, and might involve only a limited number of the MHC molecules at any one time. Satisfactory demonstration of allosteric change in, perhaps, only a minority population of a particular molecular species on cell surface obviously poses considerable problems. A different approach to the analysis of events occurring on the plasma membrane is to measure the extent of fluorescence resonance energy transfer between cell-surface molecules. Using this method Damjanovich et aZ. (1983) found evidence of energy transfer between Class I MHC glycoproteins and concanavalin A binding sites, indicating that these two entities are in close proximity. However such interactions are not particularly stable and the molecules do not cocap. In addition, analysis of the rotational properties of MHC-antigen-antibody complexes indicates that the MHC glycopro-

+

29

T CELL SURVEILLANCE AND M H C

teins are not clustered. The picture that emerges is that the MHC glycoproteins are essentially monomeric, with little lateral mobility in the plasma membrane. The latter may reflect interactions with underlying skeletal elements. It seems logical that tight, irreversible interactions between MHC molecules and other entities expressed on cell surface should not be the rule. We know that A + S, is operationally synonymous with A + x when perceived by T cells from an H-2 congenic (A) but lion H-2 different mouse. If A S,, A + S, etc. should constitute stable complexes the net result could be saturation of the available MHC glycoprotein on cell surface, which might in turn lead to failure to recognize neoantigen (e.g., virus) in an MHC-restricted way. However it is also known that, though the available MHC molecules may be derivatized with TNP, virus-immune CTL can still recognize such target cells (Biddison et al., 1977). Perhaps the observation that high levels of CTL responsiveness may commonly be associated with the particular Class I MHC glycoproteins that are turned-over and shed most rapidly (Emerson et a l . , 1980) has something to do with the saturation kinetics of the interactions between A and x, y, S, etc. Another way out of this dilemma is suggested by Kampe et al. (1983), who propose that complexes of E l 9 adenovirus protein and rat Class I MHC glycoprotein are taken into the cell in coated vesicles, with the E l 9 being degraded in lysosomes and the Class I antigen being returned to the cell surface. Evidence for the idea that the MHC glycoproteins are altered was not found when target cells modified in various ways were assayed using bulk populations of alloreactive CTL (Zinkernagel et a l . , 1977). However, the question needs to be examined again in the light of newer information concerning the possible cross-reactive recognition of self + x and alloantigen by CTL clones (summarized in Section IX,D). An obvious experiment is to test the susceptibility of the same target cell modified with different viruses and haptens to lysis by CTL clones generated following priming with alloantigen (Glasebrook et a l . , 1981). The implication of this idea is, of course, that the only entity that the CTL perceives is a change in the configuration of the MHC glycoproteins. However, even if such allosteric changes can lead to T cell recognition, this may not be the sole mechanism by which an MHC glycoprotein might seem to be altered. There are also indications that MHC and tumor-associated molecules may form rather stable complexes on cell membrane. Experimental findings which support this idea for the virus models are summarized in Section VI1,A above. Other evidence that such associations may occur is also available for mouse lymphomas (Fujimoto et a l . , 1973; Callahan et al., 1979). The question is then if the CTL is recognizing a direct change in the MHC glycoprotein as a result of these protein-protein interactions [must the tu-

+

30

PETER C . DOHERTY ET AL.

mor-specific component be a protein, even glycolipids may be involved (Lipinski et al., 1982)?].Alternatively, the T cell receptor may be specific for a ‘junction-zone” between the two molecules. Yet another possibility is that the recognition unit is comprised of two chains, one of which recognizes the MHC component while the other interacts with the tumor-associated antigen. Evidence is accumulating that the T cell receptor is indeed a two-chain structure (Section V,B) and that the genes coding for these two chains do not segregate independently following the fusion of T cell clones specific for A x and B y: hybrids with specificity for A y do not emerge (Kappler et al., 1981). The other possibility that has received considerable attention in the past is that the T cell expresses two separate receptors, R, and R,, which recognize A and x independently in the plasma membrane of the target. This model does not explain why a hybrid between separate clones recognizing A x and B y does not bind to cells expressing only A + y (Kappler et al., 1981). Also, it is difficult (Doherty et al., 1977b) to understand the nature of the interaction between the T lymphocyte and target if the binding events R, A and R, - X are quite independent. Presumably the plasma membrane of the lymphocyte expresses multiple copies of both R, and R,, and there would similarly be numerous A and x determinants on the surface of the target. The avidity of the interactions R, - A and R, - x would have to be of approximately equal strength, otherwise the lymphocyte would bind quite effectively to cells expressing either A alone or x alone: this is not the case (Zinkernagel and Doherty, 1975). Why should m binding events involving R, - A plus n binding events involving R, - x be more avid (at the level of cell-cell interaction) than (m n) R, - A or (m n) R, - x? A way out of this difficulty is to argue that one receptor (R,) is cryptic and is expressed only on the cell surface following the interaction R, - x. However, Rock and Benacerraf (1983) have now described an MHC-restricted T cell hybridoma specific for A x which has measurable affinity for A. Also, any two receptor model has difficulty in explaining how CTL clones reactive to A + x can also recognize alloantigen B (Section IX,D). The concept that B is recognized via R, in a single receptor mode does not fit with other observations that one T cell clone interacts with both A + x and B y, but not with A or B alone (Hunig and Bevan, 1982), or that A T cells can be sensitized with B x after removal of precursors that recognize B (Doherty and Bennink, 1979; Klein and Nagy, 1983). Current thinking is thus much oriented toward the idea that A and x are associated on the target cell, with the complex (or modification of A) being recognized by a single T cell receptor. If this is true, the major constraint governing the immunogenicity (for CTL) of any tumor-associated cell surface antigen is whether or not it can form an

+

+

+

+

+

+

+

+

+

+

T CELL SURVEILLANCE A N D M H C

31

appropriate interaction with at least one of the Class I MHC glycoproteins present on that cell (Doherty and Zinkernagel, 1975). In this case, “appropriate” means that A + x can only be perceived as foreign if the T cell repertoire incorporates R, + (discussed in Section V). The immunogenic region of the configuration A + x would then, in the present context, be the tumor-specific transplantation antigen. It may also be the case that R A t x expressed on a T cell lymphoma can constitute a tumor-specific antigen (Allison et al., 1982). General support for the R A f x model for T cell recognition is available from experiments with clones and hybridomas of the Th subset. The fact that this class of lymphocytes recognizes antigenic entities that have first been processed by appropriate stimulator cells has allowed investigators working with these systems to use a range of well-defined proteins, including insulins, lysozymes, myoglobins, and cytochromes (Barcinski and Rosenthal, 1977; Corradin and Chiller, 1979; Maizels et al., 1980; Berzofsky et al., 1982; Hedrick et al., 1982; Abromson-Leeman and Cantor, 1983). The overall conclusion is that the neoantigen and Class I1 MHC glycoproteins are not recognized independently, but must be associated at least during the course of the interaction between the T cell and the antigen-presenting stimulator. Furthermore, these experiments have added the important conclusion that Th cells are discriminating one, or several, amino acid substitutions located at a particular conformational site on the foreign protein. Are these changes expressed in the region of the molecule that is in closest proximity to the determinants recognized by the T cell on the Class I1 MHC glycoproteins? Alternatively, are different allosteric changes induced in the MHC glycoprotein when particular sites in the associated neoantigen are changed? Conclusions: The speculative conclusion that can be drawn from a variety of studies of T cell specificity is that the CTL are recognizing Class I MHC glycoproteins that are in some way associated with neoantigen, in this case tumor-related surface antigen. The possibility is discussed that “tumor-specific transplantation antigen” is essentially an operational definition reflecting the association on the surface of the tumor cell between the MHC glycoprotein (A) and the tumor molecule (x) on the one hand, and the T cell receptor repertoire for A x on the other. The nature of A + x is presumably dictated as much by the characteristics of the particular MHC glycoprotein A as by the structure of the neoantigen x. This interaction may, in the context of the available T cell repertoire, constitute the basis of immune response gene effects. Immunogenicity may thus be associated with an identifiable molecule of tumor origin, but the precise nature of the TSTA recognized by the CTL (or Th) clones may, perhaps, reflect the interaction with MHC glycoproteins.

+

32

PETER C. DOHERTY ET AL.

VIII. Biological Models for Recognition of Minimal Changes on Cell Surface: T Cell Responses to SV40 TSTA and Single Minor H Antigens

Mouse non-H-2 histocompatibility (H) antigens comprise a complex system of cell surface alloantigens encoded by polymorphic, autosomal, and sex-linked genes. These molecules have been detected primarily by graft rejection as indicated by the designation “histocompatibility,” and the analysis has been facilitated by the selection of congenic resistant strains defining single non-H-2H loci (reviewed in Snell and Stimpfling 1966; Bailey, 1975). The rationale for considering the non-H-2H systems as models for TSTA is obviously that single, non-H-2 antigenic differences might reasonably be thought to be recognized via comparable immunological pathways. In fact, it is not impossible that some of the minor H antigens may be products of endogenous retroviruses. The murine non-H-2H antigen system is encoded by greater than 40 autosomal and sex-linked polymorphic genes (Bailey, 1975; Graf€and Bailey, 1973). However the various cell-surface molecules encoded by these genes have not, with the possible exception of H-Y, been identified serologically or biochemically. This minimum estimate of H gene number is derived from the identification of single non-H-2H loci through the selection of congenic resistant strains which differ from background strains by limited numbers of H genes, in the optimal case a single H gene. Non-H-2H antigens so defined have been detected in uiuo primarily by tissue transplantation techniques including skin grafting and tumor transplantation (Counce et aZ., 1956; Graff and Bailey, 1973). Although a limited number of reports have suggested the production of non-H-2H antigenspecific antibodies upon repeated challenges (Zink and Heyner, 1977; Long et aZ., 1981), the bulk of experimental evidence suggests that the immune response to non-H-2H alloantigens is limited to T cells. A similar situation applies for the immune response to SV40 TSTA (Tevethia et al., 1974). More recently, in uitro assays have been developed for the detection of both nonH-2H antigens and SV40 TSTA. In addition the non-H-2H antigens were one of the systems used early on to demonstrate H-2 restriction of cytolysis: Bevan (1975, 1976) observed that responder spleen cells from mice that were primed with non-H-2 incompatible spleen cells and boosted in primary mixed lymphocyte culture (MLC) generated effector T cells specific for multiple non-H-2H antigens when tested in the cell-mediated lympholysis (CML) assay. The cytotoxic T cells generated were restricted in effector function by the H-2 genotypes of the responder, MLC stimulator, and CML target populations. Perhaps of more importance for the analysis of single non-H-2H antigens, Simpson and co-workers (Gordon et aZ., 1975; Simpson

T C E L L SURVEILLANCE A N D M H C

33

and Gordon, 1977) observed the generation of H-2-restricted cytotoxic effectors specific for the male-specific antigen H-Y. More recently it has been shown that cytotoxic effectors can be generated which are specific for H-3, H-4 (Wettstein and Frelinger, 1980), and H-7 alloantigens (Wettstein and Frelinger, 1977). An additional facet of the T cell response to non-H-2H antigens has been elucidated with the observations that H-2-restricted T cells from primed inice will proliferate in primary and secondary MLC when presented with H-7 (Wettstein and Frelinger, 1977), H-4 (Wettstein, 1982), H-3, and H-Y antigens (Wettstein, 1981). A dominant characteristic of the T cell response to non-H-2H antigens is the preferential presentation of single non-H-2H antigens by H-BK/D molecules. It was initially observed that the H-Y antigen is presented preferentially to H-2”restricted T cells via H-2D” rather than H-2K” (Gordon et al., 1975, Simpson and Gordon, 1977). Subsequently, it was found that the H-7.1 alloantigen is also presented preferentially through D” (Wettstein and Frelinger, 1977) whereas the H-4.2 alloantigen is presented through H-2K” (Wettstein and Frelinger, 1980). Further, the H-37.3 antigen is presented preferentially in association with H-2D“ (P. J. Wettstein, unpublished observations). These observations thus indicate that non-H-2H antigens are presented to cytotoxic T cells through a single H-2KID molecule shared by the high responder (and most effective) antigen-presenting cells. As a rule, within a given H - 2 haplotype only one H-2KID molecule presents a single non-H-2H antigen. This is not always the case for the response of SV40 TSTA which is recognized in association with both KID molecules in H-2”, H-2’, and H-2f mice, but in association with only the K or D molecule in H-2“, H-2k, and H-2q (Pfizenmaier et al., 1980b). Another exception may be the H-3.1 antigen which is presented effectively by both H-2Kb and H-2DL)(Wettstein and Frelinger, 1980). However, it is possible that the congenic strain combination employed for detecting the H-3.1 alloantigen actually defines two nonH-2H antigens, one of which is restricted by H-2K” and the second by H-2D1’ (Roopenian and Click, 1980). The major point to be made regarding preferential presentation is that changing the H - 2 haplotype clearly alters the spectrum of non-H-2H antigens which are represented in such a way as to induce a response. Taken in the context of the concept of immunological surveillance, mutational changes in major histocompatibility complex (MHC)-linked genes important in regulating the immune response may alter the spectrum of TSTAs which can be recognized and allow a particular species to remain abreast of changes in the tumor antigens (Section VI1,B). The appropriate system for such analysis is the panel of H - 2 K / D mutants whose altered gene products differ from wild-type molecules by single or double amino acid interchanges.

34

PETER C. DOHEHTY ET A L .

Such an analysis has been performed using the Khfn series responses to SV40 TSTA (Pan et al., 1982). Of the nine Kb mutant strains tested, only the Kbllll mutant did not produce K-end-restricted SV40 TSTA specific CTL upon immunization, and SV40 transformed Kb71L1cells were not lysed by K6 SV40 TSTA effectors. Although this result can he interpreted to show that the bml mutation is at a critical site which no longer a1)ows association of H-2K and SV40 TSTA, the Kb7111gene product does not share a single fully expressed determinant with Kb (Melief et al., 1980) and the response to other antigens in association with K1”IL1(see below) are anomalous. Of more critical interest, is the finding that CTLs from SV40 TSTA immunized Kbna mutant mice, containing a mutation in the N domain of the H-2 molecule, recognize SV40 TSTA in association with the KhnLHgene product, but not the wild-type K” product. SV40 TSTA specific wild-type CTLs, on the other hand, do recognize, to some degree, the KbnLRSV40 TSTA positive target cells. The H-2KbId-restricted, SV40-specific CTL may reflect a narrower receptor repertoire than the K6 CTL, or the one-way cross reactive lysis may reflect an immunodominant response to SV40 TSTA in association with the unaltered Db gene product, since both K” and D6 are SV40 TSTA associative alleles. A strong SV40 TSTA response to SV40 TSTA in association with one allele can dramatically alter that in association with another (Pan and Knowles, 1983): it is tempting to speculate that the alteration in the Kb71L8molecule renders the D6-associated response immunodominant. The H-2K” mutants were also tested for their ability to present the H-4.2 and H-3.1 alloantigens to H-BK”-restricted cytotoxic T cells generated in primary MLC from primed spleen cell populations (Wettstein, 1982). It was found that the K” mutants differ greatly in their capacity to present the H-4.2 alloantigen to @‘--restricted cytotoxic effector T cells. Targets expressing Kb7r11,KBnd, K61f18,KbnL1I are not capable of presenting H-4.2, while those of the Kb7,*, Kbnl*, Kb7IL9,and K b n L 1 O phenotype effectively presented H-4.2. These results generally concur with those obtained in similar CML testing of Kb-restricted, H-3.1-specific cytotoxic effectors. All targets expressing the mutant alleles which presented H-4.2 also presented H-3.1. However, unlike the response of H-4.2, targets expressing H-2K1>7pswere capable of presenting H-3.1 to H-2K6 restricted T cells, albeit to a relatively lower level than other mutant H-2Kb alleles which presented H-3.1. It thus seems that mutations occurring at the H-2K locus can, in certain instances, alter the iinmunological functions associated with the respective wild-type molecules. The differential presentation of the H-4.2 and H-3.1 alloantigens by H-2Kb”” indicates that these changes in immunological function are, to some extent, antigen-specific. Further, the ability of a target expressing a mutant H-2K6 allele to present the H-4.2 and H-3.1 alloantigens correlates with the position of the respective mutant-associated amino acid interchange

T CELL S U R V E I L L A N C E A N D M H C

35

in the K6 molecule (Nairn et ul., 1980). That is, the amino acid interchanges in the N and C1 domains occurring in the linear sequence which is generated by the disulfide bond in the C1 domain have the most deleterious efiects on the presentation of the H-4.2 and H-3.1 alloantigens to wild-typerestricted effector T cells. Two alternative explanations may be advanced for the differential control of presentation of non-H-2H antigens to cytotoxic effectors by mutant H-2K” molecules. First, mutants that do not present H antigens to wild-type-restricted effectors express mutant molecules which have been altered such that they are no longer recognized as “self’ by wild-type H-2K1’-restricted effectors. Second, the mutant H-2K” alleles expressed in targets that do not present H antigens to wild-type-restricted targets have been changed by virtue of their respective mutations from a high responder “K”” allele to a low responder allele. This alternative draws its support from the Class I1 MHC l r gene systems in which low or nonresponder l r gene products do not present the respective antigens to I region-matched responder T cells (Schwartz et ul., 1978). In order to distinguish between these two alternatives, mutant K6 mice were tested for their ability to generate H-4.2-specific cytotoxic effectors when presented with H-4.2 in the context of “self’ mutant K” molecules. Responder and stimulator combinations were devised to exclude H antigen barriers which were not H-4.2 (bm‘x’ x BIO.A): F, mice were immunized with B10-H-2~~H-4” spleen cells and boosted in MLC with (bm‘x’ x B10H-2”H-4”)F1stimulators. BIO.A (KkDd)mice do not respond to H-4.2 nor do BlO-H-2”H-4” stimulators present H-4.2 in ljitro (Wettstein and Frelinger, 1977). The 4.2-specific effectors generated were tested for cytolytic potential in CML assays (Wettstein and Melvold, 1983). It seems that responders expressing KIJ1114, Kb71J, Kbln6, and Kl)llli, generated H-4.2-specific effectors with efficiency similar to responders expressing wild-type Kb, while responders expressing K b n l l , KbnL3, and KbtrLfldid not generate H-4.2-specific effectors. The K b l l l l l mice showed intermediate levels of cytotoxic effector function. Also all H-4.2-specific effectors were H-2K” restricted with no evidence that, in the absence of a response mapping to H-2Kb a subpopulation of Db-restricted effectors was expanded. Duplicate experiments were performed with primed responder spleen cells donated by recipients of 3-5 sets of H-4.2-incompatible skin grafts, a regimen which is more effective for priming than a single injection of H-4.2-incompatible spleen cells. As in the previously described experiments, responders expressing the K b l l d , K1”lS, Kb11L6, and K[j711yalleles generated K”-restricted, H-4.2-specific, cytotoxic effectors, while those expressing the K1>lldand K b l f l l oalleles did not generate H-4.2 specific effectors, and Kblla and K L l r l l l responders generated intermediate levels of effectors.

36

PETER C. DOHERTY ET A L .

In all experiments H-4.2-specific effectors generated by Kbvl* responders cross-reacted with BIO targets which do not express H-4.2. Subsequent CML analyses indicated that the observed cross-reaction was specific for wild-type H-2Kb alone (P. J. Wettstein, unpublished data). These observations are of interest as they offer a comprehensive analysis of a panel of H - 2 K mutants from the aspect of capacity to regulate the immune response to a single foreign antigen. Most importantly, the ability of a mutant H-2Kb molecule to present H-4.2 to wild-type restricted effectors correlated in, magnitude with the ability of responders expressing these mutant molecules to respond to H-4.2. Further, these findings indicate that the amino acid sequence in the N and C1 domains generated by the disulfide bond in the C1 domain is important in regulating the immune response to the H-4.2 alloantigen. Conclusions: Analysis of the CTL response to SV40 TSTA and to single, minor H antigens provides clear evidence that relatively small alterations in the amino acid sequence of an MHC glycoprotein are associated with dramatic changes in the spectrum of antigen presentation and immune recognition. It is generally, though not invariably, found that only one of the Class I restricting elements (H-SK,D,L) is associated with T cell responsiveness to a single minor H antigen. This is found less commonly for SV40 TSTA. Whether or not a relatively small change in cell-surface phenotype will result in recognition by CTL cannot, at this stage, be predicted. The fact that many of the minor H antigens are not recognized in association with particular MHC glycoproteins indicates that a significant proportion of novel structures on the cell surface may be immunologically silent, depending on the MHC phenotype of the individual concerned. The implications of this for tumor surveillance are obvious. It should also be realized that most “mutations” in Class I MHC genes are difficult to detect by serological analysis of the relevant molecule, and could thus escape detection on tumor cells. IX. The Expression of MHC Antigens on Tumor Cells

The current situation is thus that any TSTA is defined operationally by a set of responder T cell clones (Sections VII,B, VIII). The molecular nature of the entity unique to the particular tumor that is recognized by these T cells is not clearly understood. However, we do know that the T cell must interact with one or another MHC glycoprotein on the target. The following account explores links between quantitative and qualitative aspects of MHC antigen expression, T cell surveillance, and tumor survival.

A. CLASSI MHC ANTIGENSA N D TUMORIGENICITY The embryonal carcinoma cells (EC) derived from murine teratocarcinomas do not express MHC antigens (Artzt and Jacob, 1976) and are

T C E L L SURVEILLANCE A N D M H C

37

tumorigenic even in allogeneic hosts. Such cells are not, when modified or infected, recognized by either hapten-specific or virus-immune CTL (Forman and Vitetta, 1975; Zinkernagel and Oldstone, 1977; Doherty et al., 1977b). Embryonal carcinoma cells differentiate into cells that are H2K/D positive. When F9 EC cells, that contain the C6 plasmid (SV40, herpes simplex thymidine kinase, pBR322), but do not express the SV40 early gene products, are induced to differentiate, both allogeneic and SV40-TSTA specific lymphocytes lyse the differentiating derivatives in a temporally synchronous fashion that correlates with serologic detection of H - 2 K D gene products and SV40 T-antigen (Knowles et al., 1980). In this case, both the TSTA and the H-BK/D gene products become simultaneously expressed on differentiation, so that tumor control cannot be correlated with expression of either gene product separately. Correlation between cytotoxic T cell recognition and tumor development has been extensively studied for several transforming viruses. SV40 can transform mouse cells in uitro which may then form tumors in immunodeficient, but not syngeneic immunocompetent mice. Among the six haplotypes examined (b, d, f,K , 4, and s) a strong SV40 TSTA CTL response is found in association with K"D" and K k , a moderate response in association with D", and no response in association with K", D k , and K q (Pfizenmaier et al., 1980b). Following prolonged passages in uiuo or in uitro, some SV40 transformed cell lines become tumorigenic in immunocompetent mice. Gooding (1982) derived a tumorigenic C3H SV40 transformed cell line by in uiuo passage in irradiated mice and transfer to syngeneic mice. Analysis of the Class I MHC gene products on the tumorigenic cell line revealed normal expression of Dk,but no expression of K k . Since SV40 is only recognized in association with K k in the C3H mouse, a correlation between lack of expression of the restriction element for the SV40 TSTA-specific CTLs and tumorigenicity seems to exist. All of the other SV40-transformed cell lines that are tuinorigenic in immunocompetent animals are of the H-2" haplotype; The H - 2 D d gene product is expressed in those cell lines that have been examined (Pan and Knowles, unpublished). A further exploration of the ability of mice of the H - 2 k and H2" haplotype to control tumor growth has been attempted in the H-2 congenic strains B10. BR and BlO. D2. Injection of SV40 into these mice results in the appearance of tumors in those of the H-2", but not the H 2 k haplotype (Abramczuk et al., 1984). The CTL response to SV40 TSTA in H-2D" mice is extremely weak (Knowles et al., 1979) or undetectable (Gooding, 1979). It thus seems that tumorigenicity of SV40-transformed cell lines in H-2" mice appears to result in the escape of fully transformed cells from inefficient immune surveillance. A major point of interest is that, in both H - 2 k and H-2" mice, the SV40 TSTA is recognized only in association with either the K ( K k ) or D(Dd)restriction element. Escape from efficient immune surveillance requires mutation of only one re-

38

PETER C . DOHERTY ET A L .

striction element ( K k ) or inefficient recognition in association with only one

KD gene product. No H2b SV4O-transformed cells have been reported to be tumorigenic in syngeneic immunocompetent mice: H2“ mice recognize SV40 TSTA in association with both the K” and Db molecules. A further demonstration of the efficacy of T cell-mediated immunological surveillance is provided by the work of Meruelo (Meruelo et al., 1978; Meruelo, 1979, 1980)with the radiation leukemia virus (RadLV). Resistance to RadLV-induced oncogenic process is controlled via genes in the H-2D region of the mouse MHC. Almost immediately following intrathymic injection of RadLV the expression of H-2D-coded glycoproteins is increased on the surface of thymocytes from resistant (H-2Dd) but not from susceptible (H-2Ds) mice. Resistance apparently reflects that the elevation in the levels of H-2Dd is accompanied by a strong CTL response, which operates to eliminate the transformed cells. The converse situation is found for mice expressing the susceptible (H-2Ds) phenotype, where H-2D-encoded glycoproteins can no longer be detected on the surface of the RadLV-transformed cell populations by the time that overt leukemia develops. These H-2Dnegative tumor cells seem not to be recognized by RadLV-immune CTL, and are poorly immunogeneic (in syngeneic systems) when compared with H-2d-positive cultures derived from mice of the resistant (H-2d) phenotype. Evidence for a correlation between quantitative differences in CTL effector function and variations in the levels of H-2Dd glycoprotein expression is also available from a coinparative analysis of different leukemic cell lines induced by the Gross murine leukemia virus (Plata et d.,1981). In addition, susceptibility to the Friend murine leukemia virus (MuLV) has long been known to be in some way controlled by MHC genes (Lilly, 1968)which map to the H-2D region (Chesebro et al., 1974). As with the other systems, this has been correlated with CTL responsiveness (Chesebro and Wehrly, 1976; Blank and Lilly, 1977). Thus, with both the RadLV and MuLV models, there seems to be a direct relationship between expression of appropriate MHC genes, CTL responsiveness, and resistance to oncogenic process. Differential expression of H-2b glycoproteins was studied in three sublines of the B16 melanoma that had been cultured in vitro. One line was able to kill allogeneic hosts, was not lysed by anti-H-2“ cytotoxic effectors, and did not express detectable amounts of H-2K”. Similar correlations were made for another line, which became more malignant on continued in vitro passages. The conclusion reached was that metastatic potential was inversely correlated with the level of H-2 antigen expression (Nanni et aZ., 1983). Experiments where tumor cells are exposed to the selective influence of an allogeneic, or semiallogeneic, host environment also support the above concept. Methylcholanthrene (MCA)-induced sarcomas of (H-2KkDd x H-2KsDs)F, origin were selected by passage in parental-strain mice (Klein

T C E L L SURVEILLANCE A N D M H C

39

and Klein, 1958). This resulted in the emergence of tumor populations which no longer expressed the H - 2 haplotype allogeneic to the selecting environment. A similar analysis, with comparable results, was also done with MoLV-induced lymphomas (Bjaring and Klein, 1968). Later experiments with the MCA-induced series (Ahrlund-Richter et al., 1982) gave evidence for MHC-restricted rejection of these tumors, when lymphocytes and tumor cells were injected together into sublethally (400 rad) irradiated recipient mice. In another study, an MCA-induced sarcoma of the (H-2KkDd x H-2KdD“)F, phenotype was selected by serial, intraperitoneal transplantation of ascites populations in the H-2KdDd parent (Kerbel et al., 1978). A highly metastatic subline was isolated from these mice and found to be negative for the H-2Kk alloantigen. The change appeared to be both stable and irreversible, though the investigators could not be absolutely certain that the metastatic “variant” had not been newly induced (perhaps by a virus) in the selecting DBA/2 (H-2KdDc1)host environment. There is a need to check such cell lines, using contemporary molecular probes to see whether the “missing” MHC genes are still present and down-regulated. In somewhat the converse of the above experiment Ostrand-Rosenberg et al. (1983) found that 402Ax teratocarcinoma cells, which are normally MHC negative, are induced to express surface H-2“ glycoproteins when grown in an allogeneic host. This change, which is not observed for teratocarcinoma cells passaged in syngeneic mice, leads to rejection of the tumor in the MHC-incompatible situation. Also, the expression of MHC antigens is in some way dependent on the presence of alloreactive T cell populations which operate, perhaps, by secreting y interferon (Morris et al., 1982; Wallach et al., 1982). From the foregoing, it would appear that there is a clear-cut, inverse relationship between the expression of MHC glycoproteins and tumorigenicity. However, this is not always the case. Segal and colleagues (De Batselier et al., 1980; Katzav et al., 1983a,b) have studied a sarcoma (T10) induced with MCA in an (H-2KbDb X KkDk)F, mouse. This tumor grows locally (LT10) and generates spontaneous lung metastasis (M-T10). All T10 lines are negative for the expression of both H-2Kk and H-2K“: are these two alleles associated with M HC-restricted CTL responsiveness? This would certainly fit with the evidence that we have discussed above. However, the surprising finding is that L-T10 expresses H-2Db but not H-2Dk, while M-T10 expresses both H-2Db (though at a lower level than L-T10) and H-2Dk. In every case, the emergence of a metastatic variant of T10 is accompanied by the expression of H-2Dk. One interpretation of these findings is to argue (for the T10 line) that neither H-2Dk nor H-2D“ is capable of presenting a putative TSTA to MHCrestricted CTL. Perhaps the H-2Dk glycoprotein (or gene products encoded

40

PETER C . DOHERTY ET A L

at the H-2D- end) is involved in the induction of suppression. The MTlO line is, in addition, so refractory to T cell-mediated attack in the in oiuo situation that it is not eliminated by H-2K1)Db mice, even though it expresses the H-2Dk alloantigen and can be lysed in vitro by H-2Dk-specific CTL generated in a mixed lymphocyte culture. Similar findings were recorded for the H-2K“D” 3LL Lewis lung carcinoma (Isakov et al., 1983). Metastatic capacity for this tumor, which arose spontaneously in C57BL mice, is correlated with loss of H-2Kb and increased levels of H-2Dt’. An intriguing property of this tumor is that it will grow in both H-2-compatible and H-2-different recipients, but will produce spontaneous metastasis only in syngeneic animals. Alloreactive CTL specific for H-2D” may thus be able to prevent metastasis, without being capable of limiting the growth of the solid tumor. Earlier experiments of Haywood and McKhann (1971) also found a positive correlation between tumorigenicity and levels of MHC expression for MCA-induced tumors. The more immunogenic, and less rapidly metastatic, tumor lines expressed lower levels of surface H-2 glycoproteins. However none of the tumors was completely negative for any of the H-2 antigens that were assayed. Also, the method of immunization may have influenced the outcome of the study. Mice were injected intramuscularly in the leg with viable tumor cells, and the leg bearing the tumor was then amputated 7 days later. The level of immunity detected would thus depend on the numbers of potential effector T cells which had not localized to the tumor mass at the time of amputation: if T cells had entered the tumors expressing large amounts of H-2 antigen more rapidly, the investigators may have achieved a state of clonal deletion of effectors by surgical means. In addition, the capacity of the tumors to metastasize was measured in mice that has first seen irradiated with 350 rads. Though the tumor lines expressing lower levels of H-2 antigens migrated more slowly to lung, all of those tested eventually metastasized and killed the recipient mice. The possibility that we should be considering the total amount of the MHC antigen present in the cell, rather than just the quantity that can be detected on the cell surface under in uitro culture conditions, is raised by the experiments of Dennis et ul. (1981).Working again with MCA-induced tumors they found that a highly metastatic line expressed just as much surface H-2K and H-2D glycoproteins as a minimally tumorigenic variant, but was considerably less effective in both stimulating a TNP-specific CTL response and in acting as a target for TNP-immune effectors. However, much higher levels of MHC antigen were found in isolated endoplasmic reticulum and plasma membrane fractions from the nonmetastatic variant. Perhaps these internal MHC components may be rapidly induced to appear

T C E L L SURVEILLANCE A N D M H C

41

on the cell surface under conditions of immune confrontation, exposure to interferon, and so forth. Hecent spectacular findings of Schrier et al. (1983) and Bernards et (11. (1983) have greatly strengthened the case for the idea that tumor clones escape froin CTL surveillance by monitoring the expression of Class I MHC glycoproteins. Hat cells transformed by the highly oncogenic adenovirus 12 are very tumorigenic in syngerieic adult rats and show a complete absence of the rat Class I MHC glycoprotein RT1.A. Lack of expression is not due to failure to synthesize p,-microglobulin, but reflects a mechanism regulating the mRNA necessary for production of the HT1 molecule. The susceptibility of these tumor cells to CTL-mediated lysis is, as a consequence, greatly reduced. Other tumor lines transformed with adenovirus 5 , which do not modulate the expression of HTl.A, are equally oncogenic in nu/nu mice but are readily eliminated by iininunocoinpetent rats. These studies thus provide a clear correlation between tumor growth loss of Class I MHC glycoprotein expression and defective T cell surveillance function. A similar decrease in H-2 mHNA has been found for murine tumors induced by Moloney MuLV which have lost some, or all, Class I MHC glycoproteins. The level of mRNA was reduced 30-fold in such a tumor which was phenotypically H-2 negative (Baldacci et a l . , 1983). Loss of H-2 expression thus seeins to result from changes in the transcription of H - 2 genes, which presumably reflects the operation of some kind of regulatory process. Conclusions: There are a number of instances where survival of a subset of tuinor cells depends on modulation of the expression of a particular Class I MHC gene that is associated with CTL responsiveness. Presumably this reflects selection in uiuo by exposure to effector lymphocytes. However the correlation between loss of MHC glycoprotein expression and tumor survival is riot invariant, arid situations are described where acquisition of a more metastatic phenotype is associated with enhanced cell-surface expression of a Class I MHC glycoprotein though this also involves loss of another Class I MHC molecule. The questions that seeiii to be appropriate when making such analyses are whether (1)the particular MHC antigen that is apparently missing, or present in larger amounts than normal, is associated with CTL responsiveness, and (2) the levels of MHC glycoprotein expressed on cell surface under in uitro culture conditions reflect the total amount of MHC antigen present in the cell, which might be induced under conditions of immunological confrontation? In summary, the level of expression of a particular Class I MHC glycoprotein may be assigned only functional significance in the light of associated patterns of MHC-restricted T cell responsiveness. Many of the experiments

42

PETER C . DOHEHTY ET AL.

which correlate levels of MHC antigen expression, tumorigenicity, and metastatic potential can be accommodated within this conceptual framework. However, one other factor that should be considered is that the H-2 molecules may also be involved in promoting intercellular adhesiveness, at least for fibroblasts (Bartlett and Edidin, 1978; Zeleny et al., 1978).

B. MUTATIONSI N CLASSI MHC GENES A N D T CELLSUHVEILLANCE It is possible that tumors could escape from immunological surveillance by mutating Class I MHC genes. The selective pressure would, of course, be provided by the self-monitoring T cells that are specific for the putative TSTA presented in association with the wild-type Class I glycoprotein. There is ample documentation that quite small changes in Class I MHC antigens (Nairn et al., 1980; Ewenstein et al., 1980; Yamaga et al., 1983) dramatically modify the spectrum of MHC-restricted T cell recognition (Zinkernagel, 1976; Blanden et al., 1976; McKenzie et al., 1977; Zinkernagel and Klein, 1977; Forman and Klein, 1977; Doherty et al., 1981a; Wettstein, 1982; Pan et al., 1982; Wettstein and Melvold, 1983). Many such mutants cannot be differentiated from the wild-type when tested using polyclonal antisera (Klein, 1978), and could thus easily be missed on tumor cells. The presence of such mutant Class I MHC molecules in tumor populations has been clearly shown by in uitro selection experiments using monoclonal antibodies (Holtkainp et al., 1979; Hajan, 1980). The LDHB cell line is a spontaneous lymphoma of (C3H x DBA/S)F, origin: cells expressing variants of the H-2Kk molecule which will no longer bind a particular monoclonal antibody (but still react strongly with the appropriate alloantiserum) are found in cultures of this tumor at a frequency of between 1OWS and 10W6 (Holtkamp et al., 1981). The variant studied in most detail by Holtkainp et d.(1981) did not serve as an MHC-restriction target for TNP-specific CTL sensitized in the wild-type situation, though it could be recognized by alloreactive T cells. Similarly, Potter et al. (1983) examined variants that had been iminunoselected for changes at H-2Dc1(Hajan, 1980). One of these was no longer lysed by an anti-H-2Dd cell line, while another CTL clone specific for fluorescinated H-2Dd did not react with two of the mutants that had been derivatized with the hapten. There was again a divergence in recognition by self-monitoring and alloreactive T cells. Also, the mutant molecules were shown to retain serological specificity patterns characteristic of H-2Dc1when tested with other monoclonal antibodies, though the T cell recognition spectrum had changed. A somewhat different example of change in the nature of a Class I MHC

-r CELL

SURVEILLANCE AN D M H C

43

glycoprotein comes from experiments with the LT-85 alveologenic adenocarcinoma (Martin et al., 1977). This tumor was transplacentally induced with N-ethyl nitrosourea in C3HF6/HeN (C3HF) mice which express a “mutant” H-2K(H-2Kkv1)glycoprotein differing from the wild-type H-2Kk found in the C3H/HeN (C3H) strain (Martin et d., 1978). The LT-85 tumor grows better in the C3H than i n the C3HF, from which it is derived. Analysis of tryptic peptides indicates that the H-2Kk molecule present on the tumor is complementary in some regions to H-2Kk, and in others to H-2Kkv1(Callahan et al., 1983). Furthermore, 40% of tumors so induced in C3HF mice seem to be immunologically identical to LT-85 (Martin et al., 1977). The suggestion is that the atypical H-2K molecule expressed on LT-85 may arise by recombination, probably involving gene conversion, rather than by point mutation (Evans et al., 1982; Callahan et al., 1983). central question that needs to be answered if we are to seriously consider the possibility that the escape of tumor cells from immunological surveillance reflects the mutation of Class I genes to give an MHC restriction phenotype that is not recognized by self-monitoring T cells. We need to know whether such an event can occur without the mutant MHC glycoprotein becoming essentially allogeneic. The Class I mutants, such as the H-2KbTf1series, that have been generated in uiuo were all selected on the basis of reciprocal skin graft rejection (Bailey and Kohn, 1965; Egorov, 1967; Melvold and Kohn, 1976). The rapidity of this rejection correlates well with the magnitude of the alloreactive CTL response generated by cross-stimulation between mutant and mutant, mutant and wild-type (Klein, 1978; Melief et al., 1980; Sherman, 1982). In turn, both the graft rejection and alloreactive CTL profiles show a broadly inverse relationship to the degree of crossreactivity found for self MHC-restricted CTL (Melief et al., 1980; Hunvitz et al., 1983a,b). Obviously, if loss of MHC restriction phenotype is characteristically accompanied by the generation of an essentially alloreactive response to the mutant glycoprotein, cells expressing such an MHC antigen would not be positively selected so as to be present at high frequency in the in uiuo situation. This may be a somewhat simplistic interpretation if, as claimed by some, tumor cells can indeed express alien histocompatibility antigens (Section IX,C). Conclusions: It is theoretically possible that selection for cells expressing mutant Class I MHC glycoproteins could enable a tumor to escape from immunological surveillance. Such changes would probably be missed in an analysis made with heterogeneous antisera, though they might well be detected by using an appropriate panel of monoclonal antibodies or DNA probes. However, while results from in uitro systems may make this type of analysis seem worthwhile, we need to know whether tumor clones expressing such changes can actually exist at high frequency in uiuo: would they be

44

PETER C. DOHERTY ET AL.

perceived as allogeneic and consequently be eliminated by the tumor-bearing host? C. “ALIEN” CLASSI MHC ANTIGENS O N TUMOR CELLS There is a considerable literature supporting the viewpoint that genetically inappropriate, or alien, histocompatibility antigens may be expressed on the surface of some tumor cells (reviewed by Festenstein and Schmidt, 1981). The evidence has been derived from several different laboratories, and the topic has been the subject of two international meetings (Bortin and Truitt 1980, 1981). The presence of an “alien” MHC antigen in a tumor cell population is generally demonstrated by showing that antibody to a particular H-2 “private” specificity will bind to, and perhaps precipitate (Kubota and Manson, 1983), a glycoprotein which would norinally be associated with a completely different MHC haplotype. In addition, some experiments indicate that these “alien” MHC antigens may also be recognized as restricting elements by, for instance, virus immune T cells (Schirrmacher et al., 1980). The conceptual basis for the “alien” MHC idea has not rested on the “altered self’ (or self + x) hypothesis (Bach and Bortin, 1981), but on the speculation of Bodmer (1972, 1973) that the apparent polymorphism of the Class I M HC glycoproteins reflects the operation of controlling regulator genes rather than true allelism. The idea was that the genes coding for, for instance, the extensive range of H-2K glycoproteins might all be aligned along the chromosome, with a particular gene being switched on by a gerinline regulator. The presence of an “alien” MHC glycoprotein on a tumor cell would thus result from disruption of normal regulation associated with the transformed state, and read-out of DNA that was already present (but not expressed) in cells with a more differentiated phenotype. There are precedents for this “control of gene expression” model from other biological systems and, even for the purpose of this hypothesis, Bodmer (1973) did not exclude the possibility that some Class I genes are indeed allelic. Unfortunately, the “control of gene expression” idea has not been supported by the recent DNA sequence studies (Steinmetz et al., 1982). No evidence has been found for the presence of “alien” Class I MHC genes in the MHC haplotypes that have been sequenced to date. However, some of the information supporting the operation of copy mechanisms (analogous to gene conversion) to give rise to the H-2Khfr1mutants (Pease et al., 1983, Section VI) might provide a means whereby different MHC glycoproteins could emerge in tumor cells. Even so, it should be bourne in mind that the most distinct of the H-2Kh mutants (H-2K”””) is not readily distinguished serologically from the wild-type H-2Kb, while the “alien” MHC antigens are all detected with antisera.

T CELL SURVEILLANCE A N D M H C

45

The other problem, that is well recognized by workers in the area (Bach and Bortin, 1981), is that some of the tumor lines are not what they are thought to be. Evidence for subpopulations of cells that express independent enzyme isotypes identical to those associated with mice that normally express the “alien” MHC glycoproteins would seem to support the idea that contamination has occurred (Schirrniacher et al., 1981). However, at least some of the findings cannot be readily discounted in this way (Festenstein and Schmidt, 1981; Kubota and Manson, 1983). The onus is on those working in the area to promote the necessary molecular studies, using tumor clones that are thought to express “alien” MHC glycoproteins but still show enzyme isotypes characteristic of the host. The DNA probes and techniques for such analyses now seem to be available, and there is little point in continuing extensive biological experimentation with these systems until this is resolved. Another possibility that needs to be considered is that some of the socalled “alien” MHC glycoproteins reflect posttranslational changes induced, for instance, by infection with vaccinia virus (Garrido et al., 1977). Evidence is available that at least a proportion of cell-surface MHC glycoproteins may associate with viral components (including vaccinia virus) at the plasma membrane (see Section VII). Such interactions could potentially modify the serological characteristics of a particular MHC antigen. For example, binding one monoclonal antibody to a viral glycoprotein may alter the binding profile for a second, or a third monoclonal antibody (Lubeck and Gerhard, 1982; Clegg et al., 1983). This might reflect that these protein-protein interactions have led to allosteric changes which modify serological specificity. Lewis and Bishop (1983)have suggested that the capacity of an antiserum to H-2D“ to precipitate H-2Dk from the H-2KkDkK36.16 tumor is due to an interaction between endogenous viral antigens and the MHC glycoprotein. The possibility should not be discounted that some of the descriptions of “alien” M HC glycoproteins are actually evidence for the existence of “altered self.” Conclusions: The presence of “alien” MHC glycoproteins, reflecting the expression of a completely different Class I allele on tumor cells, is controversial. The application of currently available molecular biology techniques should resolve the problem. The possibility that posttranslational changes lead to the development of “altered self’ configurations should be considered.

D. CROSS-REACTIONS AT T H E T CELLLEVELBETWEEN TSTA A N D CLASSI MHC ANTICENS Much of the interest in the “alien” MHC antigen work (see above) has derived from observations that priming lymphocyte populations with pooled

46

PETER C. DOHERTY ET AL.

allogeneic cells from a variety of MHC-different individuals leads to the development of cytotoxic effectors which can recognize, and eliminate, tumor cells that are syngeneic to the host (reviewed by Bach et al., 1980; Parmiani et al., 1982). In the first of these experiments J. Zarling et al. (1978) sensitized lymphocytes in vitro with pooled normal cells from 20 unrelated individuals: when this protocol was used with cases of leukemic reticuloendotheliosis the cytotoxic lymphocytes that were generated were lytic for the patients own peripheral leukemic cells, but not for normal T cells from the same person. The alloantigen-activated T cells had obviously detected a determinant, or determinants, unique to the tumor population. A similar approach has been used to stimulate effectors that are lytic for cells from solid tumors derived from both man and mouse though, in the latter case, NK cells were also shown to be operating (Paciucci et al., 1980; Strausser et al., 1981). In the human situation, strong priming of tumor-reactive T cells may be seen with pooled, allogeneic stimulators, but little evidence of effector function is observed when lymphocytes from a single M HC-different individual are used for sensitization (Bach et al., 1980). A more definitive analysis may be made using genetically defined mouse strains. Parmiani et al. (1982) found that immunity to a Moloney-virus induced BALB/c lymphoma YC9 (H-2d) could be stimulated by priming with cells from C3HF (H-2k), C3H.SW (H-2“)), and B10-background lines, but not with the non-H2 background congeneic BALB. K (H-2k) or BALB. B (H-2“). The allogeneic sensitization effect might thus be thought to require differences in both MHC and nonMHC genes, coding for minor histocompatibility antigens. However, the determinants that are unique to YC8 are also recognized by T cells that are stimulated with MHC-compatible B10-D2 and DBAI2 lymphocytes (Parmiani et d.,1982; Sensi et d.,1983). Thus, from these experiments, the more important difference seems to rest in the non-M HC background, though effector function may be seen following stimulation with lymphocytes expressing both minor and major incompatibilities. These results from Parmianis’ laboratory (Parmiani et al., 1982; Sensi et al., 1983) fit very well with newer knowledge about the nature of T cell specificity (see Sections V and VII). Any specific immune response reflects the combined operation of many different clones of T cells and B cells. “Specificity” is thus a population phenomenon. Just as with monoclonal antibodies, cloned T cell lines do not interact solely with the “antigen” against which they are sensitized but with a particular limited epitope on that molecule. As a result, many T cell clones have been found to show quite unexpected cross-reactivity patterns. A number that were selected for reactivity to self + x (where x may be a minor H antigen, a hapten, or a virus) are also lytic for one or another targets expressing a particular alloantigen in the

T CELL SURVEILLANCE A N D M H C

47

absence of x (von Boehmer et al., 1979; Weiss et al., 1980; Braciale et ul., 1981b). The converse is also true, with alloreactive CTL clones showing specificity for self x (Schawaller et al., 1980). The debate continues as to whether the alloreactive T cell repertoire is totally inclusive in that specific for self + x, or whether there are indeed alloreactive T cells that are not capable of being involved in self-monitoring functions. Ample evidence is also available from studies with bulk T cell populations that the two repertoires (if such there are) overlap (Bevan, 1977; Lemonnier et a l . , 1977; Finberg et al., 1978a,b; Pfizenmaier et al., 1980b). The overall spectrum of self + x seen by alloreactive T cells may be shown (using chimeras) to obey the “rules” proposed for thymic restriction of T cell specificity: alloreactive T cell populations from [(A x B)F, + A] chimeras show greater cross-reactivity for A + x than for B x, and the converse is also true (Hunig and Bevan, 1980a). In addition, the patterns of cross-reactivity may be extended to different self-restricted T cells with at least one (A x B)F, T cell clone being capable of interacting with both A + x and B + y, where x and y are different non-MHC antigens (Hunig and Bevan, 1982). Taken together, the above experiments which show extensive, unpredicted cross-reactivities for both cloned T cell lines and bulk T cell populations, provided a reasonable explanation for the finding that priming with pooled, allogeneic cells may promote tumor immunity. The observation that such pools are much more effective stimulators than are cells from a single MHC-different individual presumably means that emergence of the relevant effector(s) depends on chance cross-reactivity. Such results cannot, in isolation, be considered to justify the viewpoint that “alien” MHC antigens are expressed on many tumor cells. Conclusions: The observation that priming with pooled allogeneic cells may stimulate the generation of effector T cells capable of recognizing tumors expressing self x may be explained in the context of current knowledge of T cell specificity and the “altered self’ hypothesis. These findings do not necessarily provide any support for the idea that tumor cells express germ-line-coded “alien” MHC antigens.

+

+

+

E. EXPRESSION OF CLASSI1 MHC ANTIGENSO N TUMOR CELLS The Class I MHC glycoproteins are expressed, though at varying concentrations, in most tissues throughout the body (Klein, 1975). However, the Class I1 MHC antigens (H-2Ia, HLA-DR) are normally only found on a much more limited range of cell types. These include subsets of T and B lymphocytes, monocyte/macrophages, dendritic cells, and Langerhans cells, in fact the various cellular components involved in the stimulation and regulation of immune responses. Other cells, such as endothelium, may be

48

PETER C. DOHERTY ET AL.

induced to express Ia determinants when exposed to y interferon (Pober et a l . , 1983b). Furthermore, such Ia+ endothelial cells are recognized by immune T cells (Pober et al., 1983a). We have interpreted this differential expression of Ia glycoproteins as reflecting a functional economy in the immune system (Doherty and Zinkernagel, 1975; Doherty and Bennink, 1980a). Recognition of Ia + x on an antigen presenting cell by a Th cell will result in the secretion of 11-2, which promotes the clonal expansion of antigen-specific T and B lymphocytes. The successful operation of this cell circuitry obviously requires that the various constituents of a particular immune response be recruited so as to be in close proximity in an appropriate anatomical niche, such as lymph node or spleen, where factors operating at short range are present at high concentration. This focusing presumably reflects, at least in part, the distribution of antigen presenting cells that express the Class I1 M HC glycoproteins recognized by the Th subset. The situation in pathological states may be somewhat different and the topic is, at this stage, one that has received relatively little attention. We know, for instance, that massive antibody production can continue for considerable periods in virus-infected brain tissue that has been invaded by a variety of inflammatory cell types (reviewed by Doherty, 1982). Such responses may well involve the establishment of Th, stimulator, B cell circuits in the site of pathology. Perhaps expression of la antigens on endothelium is important for T cell recruitment (Pober et d . , 1983a,b). Similar events may also be presumed to be possible in tumors that are infiltrated with lymphocytes and macrophages. However, some classes of tumors add another order of complexity in that the transformed cells themselves express Class I1 MHC glycoproteins. The spectrum, and possible significance, of this phenomenon merits consideration. In general, Class I1 MHC glycoproteins tend to be found on tumor cells derived from lineages that would be expected to present such determinants. Tumor lines of B cell and monocyte origins often express Class I1 antigens encoded, in the mouse, in both I-A and IE/C (McKean et al., 1981; Lanier and Warner, 1981). In addition, a percentage of murine leukemias may be Ia+ without concurrent surface expression of either Ig or Thy 1.2 (Chesebro et al., 1976). Similarly, many human leukemias express HLA-/DR antigens, often in the absence of Ig markers (reviewed by Winchester et al., 1977; Greaves and Janossy, 1978). The amount of cell-surface Ia may also vary through the cell cycle, with evidence for differential expression of glycoproteins encoded in I-A and IE/C (Lanier and Worner, 1981). However, perhaps the most significant finding concerning Ia+ B cell and monocyte tumors is that a proportion of such lines can be shown to function as antigenpresenting cells for Ia-restricted T cell responses (McKean et al., 1981;

T C E L L SURVEILLANCE A N D M H C

49

Walker et d . ,1982; Kappler el a / ., 1982). Not all Ia+ B cell lines can act in this way (Walker et al., 1982), but the fact remains that many of these tumors can both bind and stimulate 11-2 secreting T cells (Kappler et a l . , 1982). This cell-surface Ia inay serve to focus effectors (DTI-I subset?) involved in elimination of leukemia cells (Forni et al., 1976). Extensive serological evidence is also available that a consideral)le spectrum of primary, solid tumors of man express enhanced levels of suiface Ialike (HLA-DH) molecules: these include melanoma, rectum adenocarcinoma, differentiated liver adenocarcinoma, meningioma, ganglioneuroblastoma, and glioblastoma (Natali et al., 1980a,b). Similar findings have been recorded for continuous cell lines from melanoma (Winchester et al., 1978; Pollack et al., 1980, 1981), hepatoma, bladder carcinoma (Pollack et al., 1981),and medullary (but not ductal) breast carcinoma (Natali et a l . , 1983). Careful analysis of the DK antigens expressed on melanoma cells, coinpared with those present on donor B lymphocytes froin the individuals concerned, indicates that the DR specificities found are genetically appropriate and do not in any sense represent “alien” MHC determinants (Pollack et al., 1980, 1981; Wilson et al., 1981). What, if anything, is the biological significance of Class I1 MHC antigen expression on tumor cells? Priming inbred guinea pigs with a syngeneic Ia lekueinia results in subsequent rejection of both Ia+ and Ia- sublines. However, when an Ia-negative variant is used for sensitization, no protection is observed on challenge (Forni et al., 1976). The presence of the Ia glycoprotein thus seems to make the tumor immunogenic. Perhaps this explains why the I-ILA-DH+ medullary breast carcinoma has a more favorable prognosis than the HLA-DH- ductal carcinoma (Natali et al., 1983). On the other hand, inalignant melanomas express HLA-DR determinants while benign nevi do not (Natali et al., 19811)).The HLA-DR+ melanoma cells are poor stimulators of allogeneic T cell responses (Pollack et nl., 1980; Pollack, 1981) though this could reflect that they are not able to deliver a suitable signal (11-1) to potential responder T cells. Conclusions: Many tumors express Class I1 MHC glycoproteins. Some are derived from lineages which are commonly I a + , while others (such as melanoma) presumably originate froin precusors that are HLA-DR negative. Some tumors of B cell origin can stimulate Ia-restricted syngeneic T cell responses, while this has not been shown for the melanomas. No general picture emerges of the possiible consequences of the interactions between such tumors and T cells restricted by Class I1 MHC determinants. Speculations concerning rejection on the one hand, and growth promotion on the other are obvious (Section 111, B). Any analysis in experimental systems should include concurrent studies of the presence or absence of a Class I MHC restricted CTL response. The possibility remains that most instances +

50

PETER C . IIOHERTY ET AL.

of Ia and HLA-DR expression on tumor cells simply reflect subtle genetic changes underlying the uncoupling of proliferation and differentiation (Greaves et al., 1981), and are of no more direct biological significance. X. MHC Phenotype and Susceptibility to Cancer

The case that has been developed throughout this review is that effector T lymphocytes capable of recognizing tumor cells are reactive to A + x, where A is an MHC glycoprotein and x is some component of tumor origin (Sections V and VII). Correlations of susceptibility to a particular forin of cancer and MHC phenotype might thus be thought to emerge in situations where x is constant. For instance, in some virus-induced leukemias in inice a clear relationship is found between oncogenic process and H-2 type (Lilly and Pincus, 1973). In this case, x is presumably of viral origin and might thus be identical for all tumors caused by that virus (Sections IV,A and VII,A). Susceptibility to lung tumors in mice, which are induced transplacentally by injecting the mother with N-ethyl-N-nitrosourea, is also associated with MHC type (Oomen et al., 1983). However, the incidence of most human cancers is not obviously a function of the MHC phenotype (Stastny et al., 1983; Svejgaard et d . ,1983). There is quite a strong correlation between HLA-A1 and incidence of Hodgkins’ disease (Hors and Dausset, 1983). Hodgkins’ disease may not be a good candidate for operation of T cell surveillance mechanisms, as immunosuppressed transplant patients do not show an increased incidence of this tumor (Section IV,B). Kaposi’s sarcoma emerges with high frequency in HLADR5 homosexuals suffering from acquired immunodeficiency, which may be a consequence of infection with human T cell leukemia virus (Section IV,A). Also, susceptibility to some types of inelanoma has been correlated with HLA type (Pellegris et al., 1982). The lack of widespread evidence that HLA type is an important risk factor in most human cancers could thus reflect that the tumor-specific component (x) which may, or may not, contribute to an iininunogenic configuration of A + x is not constant. This s e e m to be the case for UV-induced skin cancers or mice, which difier considerably between H-2 identical, congenic mice (Kripke, 1981). The other problem is that, in a (A x B)F, individual, A + x may cross-react with B + S,, where S, is a non-MHC self-component encountered during T cell development. This will result in there being a “hole” in the T cell repertoire. People of identical MHC types will differ greatly in the spectrum of non-MHC genes that they express. The capacity to mount a tumor-specific response might thus also be expected to vary independently of MHC phenotype. The topic of T cell specificity is discussed in greater detail in Sections V and VII. +

51

T CELL SURVEIL,LANCE A N D M H C

Most instances of a good correlation between MHC phenotype and human disease susceptibility patterns have emerged for conditions where there is a strong possibility that an element of autoimmunity may be involved (Stastny et al., 1983; Svejgaard et al., 1983). Such diseases often involve long-term infectious processes, or there may be other indications that autoimmunity is triggered by an infection (Geczy et d.,1980). Perhaps this reflects that an immunogenic entity (x) associated with the pathogen is cross-reactive (A x) with B S,, where S, is a self component that is normally cryptic in an (A x B)F, individual. Conclusions: There is little evidence that the incidence of most forms of human cancer is related to M HC phenotype. This may reflect both the nature of TSTA, and of the T cell response. However, failure to find correlations of this type is not surprising, and in no way diminishes the central role of MHC-restricted T cell function in immunological surveillance.

+

+

XI. General Concepts

The central fact governing surveillance of cell surface by T lymphocytes is that both stirnulation and effector function are constrained by the need to interact with M H C glycoproteins (A, B, or C). In the self situation (as distinct from alloreactivity) cell-mediated immunity develops only if the target cell is modified by the expression of some neoantigen (x or y) which x can be recognized by a T lymphocyte with sufficient affinity for A (Zinkernagel and Doherty, 19741); Doherty and Zinkernagel, 1975).Generation of a strong CMI response inay also require that neoantigens be seen in association with both Class I and Class I1 M HC glycoproteins, to stimulate cytotoxic effector and helper-inducer T cells, respectively. Such mechanisms operate well when there is a gross antigenic challenge as occurs, for instance, in virus infections. So many novel determinants are introduced that it is unlikely that there would be a total absence of potential responder T cells. The cell-mediated immune system may, in the phylogenetic sense, have been selected principally to deal with infectious processes. However the situation with tumors, other than those which are virus induced, may be that apparently novel structures on cell surface have been expressed elsewhere in the body during ontogeny. Also, even if a tumor cell does present a neoantigen, it is by no means certain that a suitably immunogenic configuration of A + x will result. There are several reasons why such a response might fail to emerge in an individual of the (A x B)F, M H C phenotype. One is that x inay not represent a totally new molecule, perhaps being an epitope encoiintered at some earlier stage of development by the T cell precursor. A further coilsideration is that A + x is very similar to B S,, where S, is a non-MHC self component present throughout ontogeny. A

+

+

52

PETEH C. DOHEHTY ET AL,

+

state of tolerance will exist for B S,, and thus for A + x, though A + x may well be recognised in an (A x C)F, individual. Yet another possibility is that A and x cannot generate a suitably antigenic confo;mation under any circumstances. Demonstration of a new antigenic entity by, for instance, monoclonal antibody on the surface of a tumor cell provides no guarantee that the inolecule in question will be seen as foreign by effector T cells. It is thus salutary to consider the fact that the analysis of T cell responses to single, minor H antigens provides numerous instances of nonresponsiveness to particular neoantigens in the context of one or another MHC glycoprotein. This may be a major reason for maintaining a high level of MHC gene polymorphism: a normal person will have at least four, functional Class I MHC glycoproteins encoded by the two alleles mapping to each of HLA-A and HLA-B. Perhaps, though the phylogenetic development of CMI was driven by the need of complex organisms to overcome the gross insult of infectious disease, the maintenance of a high level of MHC gene polymorphism reflects that the system is also fine-tuned in an attempt to deal with the type of minimal antigenic change that might be associated with many tumors. The argument may be made that cancer is mostly a disease of the aged and that any such mechanism would exert only minimal selective pressure. However, the MHC-T cell interaction is probably at least as phylogenetically primitive as the amphibia and the divergence between reproductive age and total life span may be much less for most other primates and mammals than it is for man. Even so, it must be remembered that there is a penalty inherent in expressing too many different Class I MHC genes in any one individual: the greater the range, the more the likelihood that A x will cross-react with B S, with consequent development of nonresponsiveness. Therefore any argument concerning polymorphism must ultimately be formulated at the population level. Even if A x is perceived as non-self by effector T cells, there are still ways in which the tumor can escape from this form of immunological surveillance. One mechanism is iminunoselection of cells which modulate x (Uyttenhove et a l . , 1983): this is ditricult to analyze on other than an operational level as we are often somewhat ignorant of the molecular nature of x. The other possibility, which can be assessed more readily, is that clones of tumor cells will emerge which no longer express the MHC glycoprotein associated with the T cell response in question. Instances of this occurring have been clearly documented in experimental systems, and it may be of analytical value to continue to look at this question with human tumors. An apparently contradictory situation exists in one experimental model where enhanced expression of a particular MHC antigen is associated with greater malignancy. However, these tumor cells also lose another MHC glycoprotein. The other possibility worth considering in this regard is that “muta-

+

+

+

T C E L L SUHVEILLANCE A N D M M C

53

tional” change in the MHC genes (A’) which may be serologically silent, can alter the MHC antigen so that Al + x is no longer immunogenic. Again the T cell response itself will provide the selective pressure for such events to emerge. Molecular biology now offers the tools for such analyses, and the new information on gene conversion provides a mechanism. It is thus easy to understand why T cell surveillance against tumors may, at best, be an imperfect defense mechanism. The possibility that specific T cell suppression may be involved (Russell et ul., 1983) is outside the scope of this review, though it is dealt with at length by Naor (1979), Gerniain and Benacerraf (1980), and Kripke (1981). Even if new molecules are expressed on the surface of the tumor cell these may not be recognized by T lymphocytes. Alternatively, tuinor clones which modulate the expression of an antigenic entity may be immunoselected, perhaps by cytotoxic and helperinducer T cells working in opposition. Even so, T cell surveillance does seem to operate in man to contain, at a minimum, virus-induced tumors and some skin cancers. The limitations of the T cell surveillance system must be recognized, but there is good cause for considering parameters bearing on MHC-restricted T cell recognition when attempting to analyze the host response to any tumor. XII. Summary

The immunological surveillance hypothesis was formulated prior to the realization of the fact that an individual’s effector T cells generally only see neoantigen if it is appropriately presented in the context of self MHC glycoproteins. The biological consequence of this mechanism is that T lymphocytes are focused onto modified cell-surface rather than onto free antigen. The discovery of MHC-restricted T cell recognition, and the realization that T cell-mediated immunity is of prime importance in promoting recovery from infectious processes, has thus changed the whole emphasis of the surveillance arugment. Though the immunological surveillance hypothesis generated considerable discussion and many good experiments, there is no point in continuing the debate in the intellectural context that seemed reasonable in 1970. It is now much more sensible to think of “natural surveillance” and “T cell surveillance,” without excluding the probability that these two systems have elements in common. We can now see that T cell surveillance probably operates well in some situations, but is quite ineffective in many others. Part of the reason for this may be that the host response selects tumor clones that are modified so as to be no longer recognized by cytotoxic T cells. The possibility that this reflects changes in MHC phenotype has been investigated, and found to be the case, for some experimental tumors. In this regard, it is worth remembering that many “mutations” in

54

PETER C . DOHERTY ET AL.

MHC genes that completely change the spectrum of T cell recognition are serologically silent. The availability of molecular probes for investigating the status of MHC genes in tumor cells, together with the capacity to develop cloned T cell lines, monoclonal antibodies to putative tumor antigens, and cell lines transfected with genes coding for these molecules, indicates how T cell surveillance may profitably be explored further in both experimental and human situations.

ACKNOWLEDGMENTS We thank Drs. G . L. Ada, M. Banyard, K. Blank, E. Herber Katz, and I. Ramsliaw for advice and criticism, and the US National Institutes of‘ Health for siipport (AI15412).

REFERENCES Abramczuk, J., Pan, S . , and Knowles, B. B. (1984).J. Virol., 49, 540-548. Ahrornson-Leenian, S . R., and Cantor, H. (1983).J. E x p . Med. 158, 428-431. Ada, G. L. (1982).In “Host Factors in Human Carcinogenesis” (H. Bartsch, B. Armstrong, and W. Davis, eds.), pp. 223-239. International Agency for Cancer Research, Lyon. Ada, G. L., Leung, K. N., and Ertl, H. (1981). Zinrnunol. Rev. 58, 5-24. Ahrlund-Richter, L., Klein, E . , and Merino, F. (1982). Zrnniunogenetics 15, 53-62. Ahrlund-Richter, L . , Merino, F., and Klein, E . (1983). Imnunogenetics 18, 159-163. Allison, A. C. (1980). In “Viral Oncology” ( G . Klein, ed.), pp. 481-487. Raven, New York. Allison, A. C., Monga, G. N., and Hammond, V. (1974). Noture (London)252, 746-747. Allison, J. P., McIntyre, B. W., and Bloch, D. (1982).J. Zmnmnol. 129, 2293-2300. Allouche, M . , Bennink, J. R., Mckearn, T. J.. and Doherty, P. C. (1982). Cell. Zmrnunol. 68, 193-198. Amos, D. B., Bodmer, W. F., Ceppellini, R., Condliffe, P. G . , Dausset, J., Fahey, J. L., Goodman, H. C . , Klein, G., Klein, J., Lilly, F., Mann, 11. L., McDevitt, H., Nathenson, S . , Palm, J., Reisfeld, R. A., Rogentine, N. G . , Sanderson, A. R., Shreffler, D. C . , Simonsen, M., and Van Rood, J. J. (1971). Fed. Proc., Fed. Am. Soc. E x p . B i d . 31, 10871104. Artzt, K., and Jacob, F. (1976). Transplantation 17, 632-634. Askonas, B. A , , and Webster, R. G. (1979). Eur. J. Zmmunol. 10, 151-156. Askonas, B. A,, Mullbacher, A , , and Ashman, R. B. (1982). Zinrnunology 45, 79-84. Bach, F. H . , and Bortin, M. M. (1981). Transplant. Proc. 13, 1975-1978. Bach, F. H . , Pacincci, P. A., MacPhail, S., Sondel, P. M., Alter, B. J . , and Zarling, J. M. (1980). Transplant. Proc. 12, 2-7. Bailey, D. W. (1975). Zrnrnunogenetics 2, 249-256. Bailey, 13. W . , and Kohn, H. I. (1965). Genet. Res. 6, 330-340. Baldacci, P., Pozo, F., Gisselbrecht, S . , and Kourilsky, P., (1983).J. Exp. Med. 158, 12941306. Baltimore, D. (1981). Cell 24, 592-594. Barcinski, M. A., and Rosenthal, A. (1977).J. Exp. Med. 145, 726-742. Barr6-Sinoussi. F., Chermann, J . C . , Rey, F., Nugeyre, M. T., Chamaret, S., Gruest, J., Dauguet, C . , Axler-Bin, C., Vezinet-Brun, S . , Rouzioux, C., Rozenbaum, W., and Montagnier, L. (1983). Science 220, 868-871. Bartlett, P. F., and Edidin, M. (1978).J. Cell R i d . 77, 377-388.

T C E L L SURVEILLANCE A N D M H C

55

Benacerraf, B. (1978). J. Znitnutiol. 120, 1809-1812. Benacerraf, B.. and Katz. 1). H. (1975). Ado. Cancer Res. 21, 121-173. Benacerraf, B., and McDevitt, H. 0. (1973). Science 175, 27.3-279. Benjamin, T. L. (1983). RBA Rec;. Cancer, in press. Bennink, J. R., Yewdell, J. W., and Gerhard, W. (1982). Nature (London)296, 75-76. Berendt, M. J., and North, R. J. (1980).J. E x p . Med. 151, 69-80. Bernards, R., Schrier, P. I., Houweling, A . , Bos, J. L., van der Eb, A. J . , Zijlstra, M.. and Melief, C. J. M . (1983). Nature (London) 305, 776-779. Berzofsky, J. A., Buchenmyer, 6. K., and Hicks, G. (1982).J. ltnmunol. 128, 737-741. Besedovsky, H. O., Ilel Rey, A., and Sorkin, E. (1979). J . Ex),. Med. 150, 1351-1358. Bevan, M. J. (1975). Nature (London)256, 419-421. Bevan, M . J. (1976). J . E x p . Med. 143, 1283-1288. Bevan, M . J. (1977). Proc. Natl. Acad. Sci. U . S . A . 74, 2094-2098. Bevan, M. J., and Fink, P. (1978). l n i t n i t n d . Reo. 42, 4-19. Biasi, G., Saggin, L.. Dazzi, F., Ile Rossi, A., and Collavo, I). (1983).J , Itnnzunol. 130, 14471450. Biddison, W. 11. E., Snodgrass, H. R., Bennink, H . , EfTros, R. B., and Doherty, P. C. (1977). J . E x p . Med. 146, 617-622. Biggs, P. M . (1975). Rr. J . Cancer S t q q d . ZZ 31, 152-15. Binz, H., and Wigzell, H. (1977). Contemp. Top. Initnunobiol. 7, 113-177. Binz, H., Fenner, M., Frei, D., and Wigzell, H. (1983).J. E x p . Med. 157, 1252-1260. Bjaring, B., and Klein, G. (1968). J. Natl. Cancer lnst. 41, 1411-1429. Blanden, R. V., Dunlop, M . B. C., Iloherty, P. C., Kohn, H. I., and McKenzie, I. F. C. (1976). Ztnnutnogenetics 3, 541-548. Blanden, R. V., Mullbacher, A , . and Ashman, H. B. (1979).J , Exp. Med. 150, 166-173. Blank, K., and Lilly, F. (1977). Nature (London) 269, 808-809. Bluestone, J. A , , and Hodes, R . J. (1983). Inzmuno/. Today 4, 256-259. Bodnier, W. F. (1972). Nature (London) 237, 139-145. Bodmer, W. F. (1973). Transplant. Proc. 5, 1471-1475. Boehmer, H., von, and Haas, W. (1981). Zttmiunol. Reo. 54, 27-56. Boehiner, H., von, Hengartner, H., Nabholz, M., Lernhardt, W., Schrier, M. H., and Haas, W. (1979). Eur. J . ltntnunol. 9, 5925-597. Bortin, M. M., and Truitt, R. L., eds. (1980). Transplant. Proc. 12, 1-218. Bortin, M. M., and Truitt, R. L., eds. (1981). Transplant. Proc. 13, 1751-1978. Bourguignon, L. Y. W., Hyman, R., Trowbridge, I., and Singer, S. J. (1978). Proc. Nod. Acad. Sci. U . S . A . 75, 2406-2410. Braciale. T. J., Andrew, M. E., and Braciale, V. L. (1981a). J. E x p . Med. 153, 910-923. Braciale, T. J., Andrew, M. E . , and Braciale, V. L. (198111).J. Exp. Med. 153, 1371-1376. Brooks, C. G. (1983). Nature (London) 305, 155-158. Brunner, K. T., Macllonald, H. R., and Cerottini, J.X. (1981). J. E x p . Med. 154, 362-373. Bulhers, J . E., Chen, S., and Lilly, F. (1978). J. E x p . Med. 147, 340-351. Burakoff, S. J , , Germain, R. N., Dorf, M. E., and Benacerraf, B. (1975). Proc. Natl. Acad. Sci. U . S . A . 73, 625-629. Burnet, F. M. (1970). “Immunological Surveillance.” Pergamon, Oxford. Burnet, F. M. (1973). Nature (London) 245, 359-361. Burton, R. C., Chism, S. E., and Warner, N . L. (1977).J . Imnunol. 188, 971-980. Callahan, G . N . , Allison, J. P., Pellegrino, M. A,, and Reisfeld, R. A. (1979).J . Ztntnunol. 122, 70-74. Callahan, G. N., Pardi, I)., Giedlin, M. A,, Allison, J. P., Morizet, D. M . , and Martin, J. W. (1983). J . Zmmunol. 130, 471-474.

56

PETER C . DOHERTY ET AL.

Campbell, A. E., Foley, F. L., and Tevethia, S . (1983).J. Immune/. 130, 490-492. Cantor, H., McVdy-Boudreau, L., Hugenberger, J., Naidorf, K., Shen, F. W., and Gershon, R. K. (1978).J. Exp. Med. 147, 1116-1125. Cerottini, J.-C. (1980). Prog. Inmuno/. 4, 622-631. Chang, C., Martin, R . G., Livingston, J. M . , Lubrosky, S . W., Hu C.-P., and Mora, P. T. (1979). J . Virol. 29, 69-75. Chapdelaine, J . M., Plata, F., and Lilly, F. (1979).J. E x p . Med. 149, 1531-1536. Chesebro, B., and Wehrly, K. (1976).J. E x p . Med. 143, 85-99. Chesebro, B., Wehrly, K . , and Stimpfling, J. (1974). J . E x p . Med. 140, 1457-1467. Chesebro, B., Wehrly, K., Chesebro, K . , and Dortis, J. (1976).J . Itnmunol. 117, 1267-1274. Clegg, J . C., Chanas, A. C., and Could, E . A. (1983).J. Cen. Virol. 64, 1121-1126. Cohen, R. K., and Eisen, H. N . (1977). Cell Zmmunol. 32, 1-9. Coligan, J. E., Kindt, T. J., Uehara, H . , Martinko, J., and Nathanson, S. 6 . (1981). Nature (London) 291, 35-39. Collins, J. K., Brett, W. J , , and Chesebro, B. (1980). J. Zmmuno/. 125, 1318-1324. Coloinbo, M. P.,Rodolfo, M., atid Parmiani, G . (1983).J. Zmmunogenet. 10, 337-340. Corradin, G . , and Chiller, J. M. (1979).J. E x p . Med. 149, 436-447. Counce, S . , Smith, P., Barth, R., and Snell, G . D. (1956). Ann. Surg. 144, 198-204. Damjanovich, S . , Tron, L., Szollosi, J . , Zidovetzki, R . , Vaz, W. L. C., Regateiro, F., ArndtJovin, 1). J., and Jwin, T. M . (1983). Proc. Natl. Acad. Sci. U . S . A . 80, 5985-5989. De Batselier, P., Katzav, S., Gorelik, E., Feldman, F., and Segal, S . (1980). Nature (London) 288, 179-181. Dennis, J. W., Donaghue, T. P., Carlow, D. A,, and Kerbel, R. S . (1981). Cancer Res. 41, 4010-4019. Diamond, A , , Cooper, 6 . M., Ritz, J., and Lance, M. (1983). Nature (London)305, 112-116. Doherty, P. C. (1980a). Prog. Zmrnunol. 4, 563-576. Doherty, P. C. (1980b). I n “Strategies of Immune Regulations” (A. Cunningham and E. Sercarz, eds.), pp. 103-120. Academic Press, New York. Doherty, P. C. (1982). Proc. Znt. Neuroimtnunol. Con$, Ist, in press. Doherty, P. C. (1983). I n “Genetics of the Immune Response” (E. and G . Moller, eds.), pp. 91-109. Plenum, New York. Doherty, P. C., and Bennink, J. R. (1979). J. Exp. Med. 149, 150-157. Doherty, P. C., and Bennink, J. R. (198Oa). Fed. Proc., Fed. Am. Soc. E x p . Biol. 40, 218-221. Doherty, P. C., and Bennink, J. R. (1980b). Scand. J. Zmmunol. 12, 271-280. Doherty, P. C., and Zinkernagel, R. M. (1975). Lancet 1, 1406-1409. Iloherty, P. C., Blanden, R. V., and Zinkernagel, R. M. (1976). Transplant Reo. 29, 89-124. Doherty, P. C., Effros, R. B., and Bennink, J. (1977a). Proc. Natl. Acad. Sci. U . S . A . 74, 12091213. Doherty, P. C., Solter, D., and Knowles, B. B. (1977b). Nature (London) 266, 361-362. Doherty, P. C., Effros, R. B., Bennink, J. R . , and Gerhard, W. (1977~).Perspect. Virol. 10, 73-88. Doherty, P. C., Biddison, W. E., Bennink, J. R., and Knowles, B. B. (1978).J. Exp. Med. 148, 534-543. Doherty, P. C., Bennink, J. R., and Wettstein, P. J. (1981a).J . Zmmunol. 126, 131-133. Doherty, P. C., Korngold, R . , Schwartz, D. H., and Bennink, J. R. (1981b). Zmmunol. Reo. 58, 37-72. Du Pasquier, L., and Miggiano, V. C. (1973). Transplant. Proc. 5, 1457-1461. Duprez, V., Mescher, M . F., and Burakoff, S . J. (1983).J. Imrnunol. 130, 493-495. Durack, D. T. (1981). New Engl. J. Med. 305, 1465-1467. Effros, R. B . , Doherty, P. C., Gerhard, W., and Bennink, J. (1977).J. Exp. Med. 145,557-568.

T C E L L SURVEILLANCE A N D M H C

57

Effros, R. B., Frankel, M. E., Gerhard, W., and Iloherty, P. C. (1979).J. Inmruno/. 123, 13431356. Egorov, I. K. (1967). Genetika 9, 136-144. Eichmann, K. (1978). Ado. Imtnunol. 26, 195-254. Elmers, C. A., Bergstresser, P. R., Tigelaar, R. E., Wood, P. J., and Streilein, J . W. (1983).1. E x p . Med. 158, 781-794. Emerson, S. G., Murphy, D. B., and Cone, R. E. (1980).J. Exp. Med. 152, 783-795. Essex, M . , McLane, M. F., Lee, T. H., Falk, L., Howe, C. W. S., Mullins, J. I., Calwadilla, C., and Francis, D. P. (1983a). Science 220, 862-864. Essex, M., McLane, M. F., Lee, T. H . , Tachibana, N . , Mullins, J . I., Kreiss, J . , Kasper, C. K., Poon, M.-C., Landay, A,, Stein, S. F., Francis, 11. P., Cabradilla, C., Lawrence, 1). N . , and Evatt, B. L. (19831)).Science 221, 1061-1OCi4. Evans, G. A,, Margulies, D. H., Camerini-Otero, R. D., Ozato, K.. and Seidman, J . (1982). Proc. Natl. Acad. Sci. U . S . A . 79, 1994-1998. Ewenstein, B. M . , Uehara, H., Nisizawa, T., Melvold, K. M., Kohn, H. I., and Nathenson, S. 6. (1980). Initnunogenetics 11, 383-395. Fauci, A. S. (1982). Ann. Intern. Med. 96, 777-779. Festenstein, H . , and Schmidt, W. (1981). Itninuno/. Reo. 60, 85-127. Finberg, R., Burakoff, S., Cantor, H., and Benacerraf, B. (1978a).Proc. Natl. Acad. Sci. U . S . A . 75, 5145-5149. Finberg, R., Mescher, M . , and Burakoff, S. J. (1978b). J. E x p . Med. 148, 1620-1627. Fink, P. J., Weissman, 1. L., and Bevan, M. J. (1983).J. E x p . Med. 157, 141-154. Fischer-Lindahl. K., and Lemke, H. (1979). Eur. J. Inittrunol. 9, 526-536. Flyer, D. C., Burakoff, S. J . , and Faller, D. V. (1983). Nature (London)305, 815-818. Fogh, J . , Tiso, J . , Orfeo, T., Fogh, J. M . , llaniels, W. P., and Sharkey, F. E. (1982). In "Proceedings of the Third International Workshop of Nude Mice" (N. D. Reed, ed.), pp. 447-456. Fischer, Berlin. Forman, J. (1977). J. Exp. Med. 146, 600-605. Forman, J . , and Flaherty, L. (1978). Zrntriuriogenetics 6, 227-233. Forman, J . , and KIein, J. (1977). Ittit~iuno~enetics 4, 183-193. Forman, J., and Vitetta, E. (1975). Proc. Not/. Acad. Sci. U . S . A . 72, 3661-3665. Forman, J . , Vitetta, E. S., and Hart, I). A. (1977). J. Zmniuno/. 118, 803-808. Forni, G . , Shevach, E. M . , and Green, I. (1976).J. E x p . Med. 143, 1067-1081. Fox. R. I., and Weissman, I. L. (1979). J. Itnttiuno/. 122, 1697-1704. Friedman-Kien, A . E., Lauhenstein, L. J . , Rulkstein, P., Buimovici-Klein, E., Marmor, M . , Stahl, R., Spigland, I., Kim, K. S . , and Zolla-l'azner, S. (1982). Ann. Intern. Met/. 96, 693-700. Frost, P.,Prete, P., and Kerbel, R. (1982). I t i t . J. Cuncer 30, 211-217. Fuijimoto, S . , Chen, C., Sabbadini, E . , and Sehon, A. (1973).J. Initnuno/. i l l , 1093-1100. Fujiwara, H . , Hamaoka, T., Shearer, 6. M . , Yainanrato, T., and Terry, W. I>. (1980). J. Itntnunol. 124, 863-869. Gallo, R. C., and Reitz. M. S. (1982).J. Nut/. Cancer Inst. 69, 1209-1214. Gallo, R. C., Sarin, P. S., Gelmann, E. P . , Robert-Guroff, M . , Richardson, E . , Kalyanaraman, V. S., Mann, 11.. Sidhu, 6. I)., Stahl, R. E., Zolla-Pazner, S., Leibowitch, J., and Popovic, M. (1983). Science 220, 865-867. Garrido, F., Schmidt, W., and Festenstein, H. (1977). J. Ini~mrnogenet.4, 115-125. Geczy, A. F., Alexander, K., Bashir, H. N . , and Edmonds, J. P. (1980). J. E x p . Med. 152. (SUPPI.),331-340. Gelmann, E. P., Popovic, M . , Blayney, D., Masur, H., Sidhu, G., Stahl, R. E . , and Gallo, R. C. (1983). Science 220, 862-865.

58

PETER C. DOHERTY ET AL.

Germain, R . N., and Benacerraf, B. (1980). Springer Sent. Immunopathol. 3 , 93-127. Gershon, R. K. (1975). Contemp. Top. Zmmunohiol. 3, 1-40. Gerslion, R. K . , Eardley, D. D., Durum, S., Green, D. R . , Shen, F.-W., Yamanchi, K., Cantor, H., and Murphy, D. B. (1981).J. E x p . Med. 153, 1533-1546. Gillespie, G. Y., Hanson, C. B., Haskins, R . G . , and Russell, S. W. (1978).J . Immunol. 119, 564-570. Giorgi, J. V., Burton, R. C., Scott, I)., and Warner, N. L. (1982). Int. J. Cancer 29, 119126. Glasebrook, A. L., Sarmiento, M., Loken, M. R . , Dialynas, D. P., Quintans, J . , Eisenberg, L., Lutz, C. T., Wilde, D., and Fitch, F. W. (1981). Immunol. Reo. 54, 225-266. Gomard, E . , Levy, J. P., Plata, F., Henin, Y., Ihprez, V., Bismuth, A., and Reme, T. (1978). Eur. J. Zmniunol. 8, 228-236. Gomard, E., Wybier Fracpi, J . , Siminler, M. C., McKenzie, I. F. C . , and Levy, J. P. (1981).J. Zmmunol. 127, 2291-2295. Gooding, L. R . (1979).J. Immunol. 118, 920-927. Gooding, L. R. (1982).J. Inimtrnol. 129, 1306-1312. Gooding, L. R . , and O’Connell, K. A. (1983).J. Iniinunol. 131, 2580-2586. Gordon, R. D., Simpson, E . , and Samelson, L. E. (1975).J . Exp. Med. 142, 1108-1120. Gottlieh, M . S.,Schroff, R . , Schanker, H. M . , Weissnian, J . D., Fan, P. T., Wolf, R. A , , and Saxon, A. (1981). New Engl. J. Med. 305, 1425-1431. Graff, R. J., and Bailey, 1). W. (1973). Trunsplant. Reo. 15, 26-49. Greaves, M. F., and Janossy, G. (1978). Biochini. Biophys. Acta 516, 193-230. Greaves, M. F., Delia, D., Robinson, J . , Sutherland, R . , and Newman, R. (1981). Blood Cells 7, 257-280. Green, W. R . (1980).J. Intmunol. 125, 2584-2590. Green, W. R . , Nowinski, R . C., and Henney, C. S. (1980).J. Zmmunol. 125, 647-655. Greenberg, P. D., Chewer, M. A . , and Fefer. A. (1981).J. Zmnwnol. 126, 2100-2103. Greeiie, M. I . (1980). Contemp. Top. Zmmunohiol. 11, 81-116. Greene, M . I., Perry, L. L., Kinney-Thornas, E., and Benjamin, T. L. (1982).J. Immunol. 128, 732-736. Greenspan, N. S., Schwartz, 1). H., and Ilolierty, P. C. (1983). In “Advances in Host Defence Mechanisms” (A. S. Fauci, ed.), in press. Hackett, C. J., Askonas, B. A., Webster, R. G . , and Van Wykke, K. (1980).J. Exp. Merl. 151, 1014-1025. Hackett, C. J . , Dietzschold, B., Gerhard, W., Christ, B., Knorr, R . , Gillessen, D., and Melchers, F. (1983).J. E x p . Med. 158, 294-302. Hale, A. H. (1980). Iminrrnogenetics 10, 469-479. Hale, A. H., Ruebush, M. J . , and Harris, D. T. (1980).J , Zmmunol. 125, 428-430. Hampar, B. (1981). Ado. Cancer Res. 35, 27-47. Hardie, 1. R . , Strong, R. W., Hartley, L. J . C., Woodruff, P. W. H., and Clunie, 6 . J . (1980). Surgery 87, 177-183. Haskins, K., Kuho, R . , White, J . , Pigeon, M., Kappler, J . , and Marrack, P. (1983).1. Exp. Med. 157, 1149-1169. Haywood, G . R., and McKhann, C. F. (1971).J. Exp. Med. 133, 1171-1187. Heclit, T. T., and Summers, I). F. (1976).J. Virol. 19, 833-845. Hedrick, S. M . , Matis, L. A . , Hecht, T. T . , Satnelson, L. E., Longo, D.L., Heber-Katz, E . , and Schwartz, R. H. (1982). Cell 30, 141-152. Hedrick, S. M., Nielsen, E. A., Kavaler, J., Colien, D. I., and Davis, M. M. (1984). Nature (London) 308, 153-158. Herberman, R . B., and Holden, H. T. (1978). Ado. Cancer Res. 27, 305-377. Herherman, R . B., and Ortaldo, J. R. (1981). Science 214, 24-30.

T C E L L SURVEILLANCE A N D M H C

59

Herberinan. R. B., Djeu, J. Y., Kay, H. I)., Ortaldo, J. R., Riccardi, C., Bonnard, G. I)., Holden, H. T.. Fagnani, R., Santoni, A . , and Puccetti, P. (1979). Imniunol. Reo. 44, 43-70. Hildeniann, W. H., Bigger, C. H., Johnston, I. S . , and Jokiel, P. L. (1980a). Transplantation 30, 362-367. Hildeinann, W. H . , Jokiel, P. L., Bigger, C. H . , and Johnston, I. S. (198011).Transplantation 30, 297-301. Hood, L., Steiniiietz, M . , and Goodenow, R. (1982). Cell 28, 685-687. Holtkamp, B., Fischer-Lindahl. K . , Segall, M . , and Rajewsky, K. (1979). Zmmunogenetics 9, 405-421. Holtkamp, B., Cramer, M . , Lemke, H., and Riijewsky, K. (1981). Nature (London)289, 66-68. Honeycutt, P. J . , atid Gooding, L. R. (1980). Eur. J . Initnunol. 10, 363-370. Hors, J., and Dausset, J. (1983). Initnuno/. Rea. 70, 166-192. Howard, J. C. (1980). Nature (London) 286, 15-16. Humphrey, J. H. (1981). Zmtnunol. Toduy 2, v. Hunig, T. R., and Bevan, M . J. (1980a). J . E x p . Med. 151, 1288-1298. Hunig, T., and Bevan, M. J. (1980b).J . Exp. Med. 152, 688-702. Hunig, T. R., and Bevan, M . J. (1982).J. Esp. Med. 155, 111-125. Hurwitz, J. L., Pan, S., Wettstein, P. J., and Doherty, P. C. (1983a). Zinmrrnogenetics 17, 7987. Hurwitz, J. L., Wettstein, P. J., and Doherty, P. C. (19831)).J. Itnmunol. 131, 471-474. Isakov, N . , Katzav, S., Feldman, M . , and Segal, S. (1983).J. Natl. Cancer Inst. 71, 139-145. Janeway, C. A , , Jr.. Wigzell, H., and Binz, H. (1976). Scond. J. Imtnunol. 5, 993-1001. Jensenius, J. C., and Williams, A. F. (1982). Nature (London) 300, 583-588. Jerne, N. K. (1971). Eur. J . Itnmunol. 1, 1-9. Jones, P. P., Murphy, D. B., and McDevitt, H. P. (1978). J. Exp. Med. 148, 925-939. KSmpe, O. , Larhanimer, D., Wiman, K., Schenning, L., Claesson, K., PaPbo, S., HyldigNielsen, J. J., Rask, L., and Peterson, P. A. (1983). In “Genesis of the Immune Response” (E. Moller and G. Mdler, eds.), pp. 61-79, Plenum, New York. Kaneko, Y., Natsuume-Sakai, S.,and Migita, S. (1978).1. Itnmunol. 121, 427-437. Kappler, J. W., Skidmore, B., White, J . , and Marrach, P. (1981).J. Exp. Med. 153, 1198-1214. Kappler, J., White, J., Degniann, D., Mustain, E., and Marrack, P. (1982). Proc. Natl. Acad. Sci. U . S . A . 79, 3604-3607. Katzav, S., I>e Batselier, P., Tartakovsky, B., Feldman, M., and Segal, S. (1983a). J . Nutl. Cancer Inst. 71, 317-324. Katzav, S., Tartakovsky, B., I)e Batselier, P., Isakov, N . , Feldman, M., and Segal, S. (19831)). Transplant. Proc. 15, 162-170. Kanfiiian, J. F., and Strominger, J. L. (1982). Nature (London)297, 694-697. Kemp. I). J . , Adams. J. M., Mottram, P. L., Thomas, W. R., Walker, I. D., and Miller, J. F. A. P. (1983). J . Exp. Med. 156, 1848-1853. Kerbel, R. S., Twiddy, R. F., and Robertson, 1). M. (1978). Znt. J. Cancer 22, 583-594. Kiessling, R., and Wigzell, H. (1979). Ado. Cancer Res. 24, 165-208. Kilbourne, E. D., ed. (1975). “The Influenza Viruses and Influenza,” pp. 1-14. Academic Press, New York. Kinlen, L. J., Sheil, A. G . , Peto, J., and Dell, R. (1979). Brit. Med. J. ii, 1461-1466. Klein, B. Y., Devens. B., Deutsch, O., Ahituv, A . . Frenkel, S., Kobrin, B. J . , and Naor, D. (1983). Tratqdant. Proc. 13, 790-797. Klein, G. (1979). Proc. Natl. Acad. Sci. U . S . A . 76, 2442-2446. Klein, G . , and Klein, E. (1958). J. Cell. Corny). Physiol. 52, (Suppl. l), 125-167. Klein, J. (1975). “Biology of the Mouse Histocompatibility-2 Complex.” Springer-Verlag, Berlin and New York.

60

PETER C . DOHEHTY ET A L .

Klein, J. (1978). Adu. Imrnunol. 26, 55-146. Klein, J., and Nagy, Z. A. (1983). Ado. Cancer Res. 37, 233-313. Knowles, B. B., Koncar, M., Pfizenmaier, K., Solter, D., Aden, D., and Trinchieri, G. (1979). J . Immunol. 122, 1798-1806. Knowles, B. B., Pan, S . , Solter, D., Linnenbach, A,, Croce, C., and Heubner, K. (1980). Nature (London) 288, 615-618. Kobayashi, H . , Takeichi, N., and Kuzumaki, N. (1978). “Xenogenization of Lymphocytes, Erythroblasts and Tumor Cells” (Vol. 10). Hokkaido University Medical Library Series. Koprowski, H . , Steplewski, Z., Herlyn, D., and Herlyn, M. (1978). Proc. Natl. Acad. Sci. U . S . A . 75, 3405-3409. Kornian, A. J.. Auffray, C . , Schanibock, A., and Strominger, J. L. (1982).Proc. Natl. Acad. Sci. U . S . A . 79, 6013-6107. Koriigold, R., and Doherty, P. C. (1984). Scand. J. Immunol. 19, 175-180. Koszinowski, U . A., Allen. H., Gething, M. J., Waterfield, M. D., and Klenk, H. D. (1980).J. E x p . M u d . 151, 945-958. Kraig, E., Kronenberg, M., Kapp, J. A , , Pierce, C. W., Abruzzini, A. F., Sorenson, C. M., Samelson, L. S . , Schwartz, R. H., and Hood, L. E. (1983).J. Exp. Med. 158, 192-209. Kreider, J. W., and Bartlett, 6. L. (1981). Ado. Cancer Res. 35, 81-110. Kripke, M . L. (1981). Adu. Cancer Res. 34, 69-106. Krystal, M., Elliot, R. M . , Benz, E. W., J r . , Young, J. F., and Palese, P. (1982). Proc. Natl. Acad. Sci. U . S . A . 79, 4800-4804. Kulmta, K., and Manson, L. A. (1983). Int. J . Cancer 31, 655-660. Knrosawa, Y., voii Boehmer, H . , Haas, W., Sakono, H . , Trauneker, A., and Tonegawa, S. (1981). Nature (London) 290, 565-570. Kvist, S., Ostberg, L., P e r s o n , H., Philipson, L., and Peterson, P. A. (1978).Proc. Natl. Acad. Sci. U . S . A . 75, 5674-5678. Lamb, J. R., Eckels, D. D . , Lake, P., Woody, J. N . , and Green, N. (1982). Nature (London) 300, 66-68. Lamb, J. R., Skidmore, B. T., Gree, N . , Chiller, J. M., and Feldman, M. (1983).J. Exp. Med. 157, 1434-1447. Langman, R. E. (1978). Reo. Physiol. Biochetn. Pharnuzcol. 81, 1-35. Lanier, L. L., and Warner, N. L. (1981).J. Itntnunol. 126, 626-631. Lannin, D. R., Yu, S . , and McKhann, C. F. (1982).J. Immunol. 128, 263-268. Larhammer, D., Gustafsson, K., Claesson, L., Bill, P., Wiman, K., Scheming, L., Sundelin, J., Widmark, E., Peterson, P. A., and Rask, L. (1982). Cell 30, 153-161. Lawrence, H. S. (1974). Haroey [Act. 68, 239-350. Lee, J. S . , Trousdale, J., and Bodmer, W. F. (1982).Proc. Natl. Acad. Sci. U . S . A . 79,545-549. Lemonnier, F . , Burakoff, S . , Germain, R. N., and Benacerraf, B. (1977).Proc. Natl. Acad. Sci. U.S.A. 74, 1229-1233. Levy, J . P., and Leclerc, J. C. (1977). Ado. Cancer Res. 24, 1-66. Levy, J. A,, and Ziegler, J. L. (1983). Lancet ii, 78-81. Lewis, C. M., and Bishop, C. E. (1983). Zmniunology 50, 433-440. Lilly, F. (1968).J. E x p . Med. 127, 465-473. Lilly, F., and Pincus, T. (1973). Ado. Cancer Res. 17, 231-277. Lipinski, M., Nudelman, E . D . , Wiels, J.. and Parsons, M. (1982).J. Immunol. 129, 23012304. Long, P. M., Lafiise, W. P . , and David, C. S. (1981).J. Itnmunol. 127, 825-828. Lopez d e Castro, J . A., Strominger, J. L., Strong, 11. M., and Orr, H. T. (1982). Proc. Natl. Acad. Sci. U . S . A . 79. 3813-3817.

T C E L L SURVEILLANCE AND M H C

61

Lubeck, M., and Gerhard, W. (1982). Virology 118, 1-7. McDevitt, H . 0. (1981).J. Immunogenet. 8, 287-295. McDevitt, H . O., and Benacerraf, B. (1969). Ado. Zmmunol. 11, 31-74. McKean, D. J., Infante, A. J.. Nilson, A., Kimnto. M., Fathman, C. G . , Walker, E . , and Warner, N. L. (1981).J. E x p . Med. 154, 1419-1431. McKenzie, I. F. C . , Pang, T., and Blanden, R. V. (1977). Immunol. Reo. 35, 185-230. McMichael, A. J., and Askonas, B. A. (1978). Eur. J. Inimunol. 8, 705-711. McMichael, A. J., Parham, P., Brodsky, F. M., and Pilch, J. R . (1980). J. E x p . Med. 152 (Suppl.), 195-203. McNicholas, J., Steinmetz, M., Hunkapiller, T., Jones, P., aiid Hood, I,. (1982a). Science 218, 1229- 1232. McNicholas, J. M., Murphy, D. B., Matis, L. A , , Schwartz, R. H . , Lerner, E. A , , Janeway, C. A , , Jr., and Jones, P. P. (198211).J. E x p . Med. 155, 490-507. Maizels, R. M., Clarke, J. A., Harvey, M. A., and Sercarz, E. (1980). Eur. J. Znimunol. 10, 509-5 15. Marrack, P., and Kappler, J. (1982).J. E x p . Med. 157, 404-418. Martin, W. J., Gipson, T. G., and Rice, J. M. (1977). Nature (1,ondon) 265, 738-739. Martin, W. J., Gipson, T. G . , Conliffe, M. A , , Cotton, W. G . , Dove, L. F., and Rice, M. (1978).J. Immunogenet. 5, 255-260. Marx, J . L. (1983a). Science 220, 806-809. Marx, J. L. (1983b). Science 221, 1278-1279. Mathis, 1). J., Benoist, C . , Williams, V. E., 11, Kanter, M., and Mcllevitt, H. 0. (1983). Proc. Natl. Acad. Sci. U . S . A . 80, 273-277. Matis, L. A , , Jones, P. P., Murphy, D. B., Hedrick, S. M., Lerner, E. A , , Janeway, C. A,, Jr., McNicholas, J. M., and Schwartz, R. H. (1982).J. Exp. Med. 155, 508-523. Matzinger, P. (1981). Nature (London) 292, 497-501. Melief, C. J. M., D e W a d , L. P., Van Iler Meulen, M. Y., Melvold, R. W., and Kohn, H. I. (1980).J. E x p . Med. 151, 993-1013. Melvold, R. W., and Kohn, H. I. (1976). Immunogenetics 3, 185-191. Meruelo, D. (1979).J. Exp. Med. 149, 898-909. Meruelo, D. (1980).J. Immunogenet. 7, 81-90. Meruelo, D., Nimlestein, S., Lieberman, M., and McDevitt, H. 0. (1978).J. E x p . Med. 147, 470-487. Meuer, S. C . , Cooper, D . A,, Hodgdon, J. C., Hussey, R. E . , Fitzgerald, K. A,, Schlossman, S. F . , and Reinherz, E. L. (1983). Science 222, 1239-1242. Miller, J. F. A. P. (1979). Ado. Cancer Res. 29, 1-44. Miller, R. G . (1980). Nature (London) 287, 544-546. Mills, C. D . , and North, R. J . (1983).J. E x p . Med. 157, 1448-1460. Misko, I. S., Moss, D. J., and Pope, J. H. (1980). Proc. Natl. Acad. Sci. U . S . A . 77,4247-4250. Mitchison, N. A,, and Kinlen, L. A. (1980). Prog: Immunol. 4, 641-650. Mitsuya, H . , Matis, L. A,, Megson, M., Bunn, P. A , , Murray, C . , Mann, D. L., Gallo, R. C . , and Broder, S. (1983).J. Exp. Med. 158, 994-999. Moller, G., and Mdler, E. (1976). Transplant. Reo. 28, 3-16. Moorhead, J. W. (1981).J . E x p . Med. 154, 1811-1826. Morris, A. G., Lin. Y. L., aiid Askonas, B. A. (1982). Nature (London)295, 150-152. Moss, 1). J . , Scott, W., and Pope, J. H. (1977). Nature (London) 268, 735-736. Moss, D. J.. Chan, S. H., Burrows, S . R., Chew, T. S., Kane, R. A., Staples, J. A., and Kunaratnam, N. (1983). Znt. J. Cancer 32, 301-305. Mukherji, B., and MacAlister, T. J. (1983).J . E x p . Med. 158, 240-245. Nairn, R . , Yamaga, K., and Nathenson, S. (1980). Annu. Reo. Genet. 14, 241-277.

62

PETER

<:.

DOHEHTY ET A L .

Nanni, P., Colombo, M. P., De Giovanni, C., Lollini, P.-L., Nicoleti, G . , Parmiani, G., and Prodi, G. (1983).J. Ztnmunogenet. 10, 361-370. Naor, D. (1979). Ado. Cancer Res. 29, 45-125. Natali, P. G . , Cordiali-Fei, P., Di Filippo, F., Quaranta, V., Pellegrino, M. A., and Ferrone, S. (1981a). Clin. Immunol. Zrnmunopathol. 19, 250-259. Natali, P. G . , De Martino, C., Quaranta, V., Pellegrino, M. A., and Ferrone, S. (1981b). Zmrnunogenetics 12, 409-413. Natali, P. G . , Giacomini, P., Bigotti, A., Imai, K., Nicotra, M. R . , Ng, A. K., and Ferrone, S . (1983). Cancer Res. 43, 660-668. Nathenson, S. G., Uehara, H., and Ewenstein, B. M. (1981).Annu. Reo. Biochem. 50, 10251052. Navarro, C., and Hagstrom, J. N. C. (1982). New Engl. J. Med. 306, 933. Nelson, D. S., Nelson, M., Farram, E., and Inoue, Y. (1981).Aust. J. E x p . Biol. Med. Sci. 59, 229-262. Ng, A. K., Pellegrino, M. A., Imai, K., and Ferrone, S. (1981). J. Zmmunol. 127, 443-447. Nilsson, K., and Klein, G . (1982). Ado. Cancer Res. 37, 319-380. Oonien, C. J. M., Ilemant, P., Hart, A. A. M . , and Emmelet, P. (1983).Znt. J. Cancer 31,447454. Ostrand-Rosenherg, S . , Cohn, A. L., and Sandoz, J. W. (1983).J. Zmmunol. 130, 2969-2973. Outzen, H. C., Caster, R. P., Eaton, G . J., and Prehn, H. T. (1975).J. Reticuloendothel. Soc. 17, 1-9. Owen, J. A., Allouche, M., and Doherty, P. C. (1982a). Cell Imrnunol. 67, 49-59. Owen, J., Greenspan, N., and Doherty, P. C. (198211).I n “Transplantation Antigens” (P. Parham and J . Strominger, eds.), Chapman Hall, London, Recep. Rec. 14, 85-117. Ozato, K., Evans, G. A., Shykind, B., Margulies, D. H., and Seidman, J. (1983). Proc. Natl. Acad. Sci. U .S.A. 80, 2040-2043. Paciucci, P., Macphail, S., Zarling, J. M., and Bach, F. H. (1980).J. Zmmunol. 124, 370-375. Palese, P. (1977). Cell 10, 1-10, Pan, S. H., and Knowles, B. B. (1983). Virology 125, 1-7. Pan, S. H., Wettstein, P. J . , and Knowles, B. B. (1982).J. Imrnunol. 128, 243-246. Parker, J. W., Joyce, J., and Pattengale, P. (1982). In “Proceedings of the Third International Workshop on Nude Mice” (N. D. Heed, ed.), pp. 347-357. Fischer, Berlin. Parmiani, G., Sensi, M. L., Carbone, G . , Colombo, M., Peirotti, M . , Ballinari, D., Hilgers, J . , and Hilkens, J. (1982). Znt. J. Cancer 29, 323-332. Pease, L. R . , Schulze, D. H., PfalTenbach, G . M., and Nathenson, S. G . (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 242-246. PeHegris, G . , Illeni, M. T., Rovini, M. T., Vaglini, M., Cascinelli, N., and Ghidoni, A. (1982). Znt. J. Cancer 29, 621-623. Penn, I. (1981). Transplant. Proc. 12, 736-738. Pfizenmaier, K., Jung, H., Kurrle, R . , Hollinghoff, M., and Wagner, H. (1980a). Irnmunogenetics 10, 395-404. Pfizenmaier, K., Pan, S. H., and Knowles, B. B. (1980b).J. Zrnrnunol. 124, 1888-1891. Phillips-Quagliata, J. M. (1980). Cell Immunol. 49, 120-134. Plata, F. (1982).J. Exp. Med. 155, 1050-1062. Plata, F . , and Lilly, F. (1979).J. E x p . Med. 150, 1174-1186. Plata, F., Tilkin, A.-F., Levy, J.-P., and Lilly, F. (1981).J. E x p . Med. 154, 1795-1810. Plata, F., Kalil, J., Zilber, M.-T., Fellows, M . , and Levy, D. (1983).J. Zmmunol. 131, 25512556. Pober, J. S . , Collins, T., Gimbrone, M. A., Jr., Cotram, R. S., Gitlin, J. D., Fiers, W.,

T C E L L SURVEILLANCE A N D M H C

63

Calyberger, C.. Krensky, A. M., Burakoff, S. J., and Reiss, C. S. (1983a).Nature ([Andon) 305, 726-729. Pober, J. S., Gimbrone, M. A . , Jr., Cotran, P. S., Reiss, C. S., Burakoff, S. J., Fiers, W., and Auk, K. A. (l983b).J. Exp. Med. 157, 1339-1353. Pollack, M . S. (1981). Transplant. Proc. 13, 1947-1950. Pollack, M. S., Livingston, P. O., Fogh, J., Carey, T. E., Oettgen, H. F., and Dupont, B. (1980). Tissue Antigens 15, 249-256. Pollack, M. S., Heagney, S. D., Livingston, P. O., and Fogh, J. (1981).J. Natl. Cancer Inst. 66, 1003-1012. Potter, T. A., Palladino, M. A., Wilson, D. B., and Rajan, T. V. (1983).J. Exp. Med. 158, 10611076. Poupon, M. F., Payelle, B., and Lespinats, G. (1981).J. Zrnmunol. 126, 2342-2346. Prehn, R. T. (1976a).Ado. Cancer Res. 23, 203-238. Prehn, R. T. (1976b). Transplant. Reo. 28, 34-42. Prehn, R. T. (1983). Znt. J. Zmrnunophanmcol. 5, 255-257. Purtillo, D. T. (1981). Ado. Cancer Res. 34, 279-312. Rajan, T. V. (1980). Zmmunogenetics 10, 423-431. Reinherz, E. L., Meuer, S. C., Fitzgerald, K. A , , Hussey, R. E., Hodgdon, J . C . , Acuto, O., and Schlossman, S. F. (1983). Proc. Natl. Acad. Sci. U . S . A . 80, 4104-4108. Reiss, C. S., Evans, G . A., Margulies, D. H., Seidnian, J. G., and Burakoff, S. J . (1983). Proc. Natl. Acad. Sci. U . S . A . 80, 2709-2712. Rickinson, A. B., Wallace, L. E . , and Epstein, M. A. (1980). Nature (London) 283, 865-867. Robertson, M. (1983). New Sci. 99, 198-200. Rock, K. L., and Benacerraf, B. (1983).J. Exp. Med. 157, 359-364. Roder, J. C., and Haliotis, T. (1980). Zrnrnunul. Today 1, 96-100. Roopenian, D. C., and Click, R. E. (1980). Zmrnunogenetics 10, 333-341. Rouse, B. T., Wells, R. J. H., and Warner, N. L. (1973).J . Zrnrnunol. 110, 534-539. Russell, P. S . , Chase, C. M., and Burton, R. C. (1983).J. Z.tnmunol. 130, 951-957. Rygaard, J., and Povlsen, C. D. (1976). Transplant. Reo. 28, 43-61. Samelson, L. E., Gerniain, R. N . , and Schwartz, R. H. (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 6972-6976. Schawaller, R., Rollinghoff, M., and Wagner, H. (1980). Scand. J. Zrnmunul. 11, 449-453. Schirrmacher, V., Garrido, F., Hiibsch, D., Garcia-Olivares, E., and Koszinowski, U. (1980). Transplant. Proc. 12, 32-37. Schirrniacher, V., Robinson, P., Altevogt, P., Radhruch, A , , Garrido, F., Koszinowski, U., Krog, H.-U., and Lundin, L. (1981). Transplant. Proc. 13, 1819-1823. Schmidt, W., Atfield, G., and Festenstein, H. (1979). Zmrnunogenetics 8, 311-321. Schrader, J. W., Cunningham, B. A., and Edelman, G. M. (1975). Proc. Natl. Acad. Sci. U . S . A . 72, 5066-5070. Schrier, P. I . , Bernards, R., Vaessen, R. T. M. J., Houweling, A., and van der Eb, A. J. (1983). Nature (London) 305, 771-775. Schulze, D. H., Pease, L. R., Geier, S. S., Reyes, A. A., Sarmiento, L. A., Wallace, R. B., and Nathenson, S. G. (1983). Pruc. Natl. Acad. Sci. U . S . A . 80, 2007-2011. Schwartz, R. H. (1978). Scand. J . Zmmunol. 7, 3-10. Schwartz, R. H., Yano, A , , and Paul, W. E. (1978). Zmmunol. Reo. 40, 153. Senik, A,, and Neauporte-Sautes (1979).J. Zmmunol. 122, 1461-1467. Sensi, M. L., Parenza, M., and Parmiani, G. (1983).J. Natl. Cancer Inst. 70, 291-297. Shackelford, D. A,, Kaufman, J. F., Korman, A. J.. and Strominger, J. L. (1982). Imrnunol. Reo. 66, 138-187. Shearer, G . M. (1974). Eur. J . Zrnrnunol. 4, 527-533.

64

PETER

<:.

DOHERTY ET A L .

Shearer, G . M. (1983). lmmunol. Today 4, 181-185. Shearer, G. M., and Schmitt-Verhulst, A. (1977). Ado. Immunol. 25, 55-91. Sherman, L. A. (1980).J. Exp. Med. 151, 1386-1397. Sherman, L. A. (1982). J . E x p . Med. 155, 380-389. Sherman, L. A,, Mescher, M. F . , and Burahoff, S. J. (1979).J . Zmmunol. 123, 501-502. Shiel, A. G . R. (1982). I n “Tissue Transplantation” (P. J. Morris, ed.), Vol. 3, pp. 242-252. Churchill, Edinburgh. Simonsen, M. (1967). Cold Spring Harbor Symp. Quant. B i d . 32, 519-522. Simpson, E., and Gordon, R. D. (1977). Immunol. Rev. 35, 59-75. Snell, G. D. (1968). Folio Biol. (Praque) 14, 355-358. Snell, G. D. (1981). “Mamma~ianGenetics and Cancer.” The Jackson Laboratory Fiftieth Anniversary Symposium, pp. 241-272, Liss, New York. Snell, G . D., and Stimpfling, J. H. (1966). In “Biology of the Laboratory Mouse” (E. L. Green, ed.), pp. 457-491. McGraw-Hill, New York. Snodgrass, H. R . , Bosma, M. J., and Wilson, D. B. (1981).J. Exp. Med. 154, 491-500. Soule, H. R., Lanford, R . E., and Butel, J. S. (1982). Int. J. Cancer 29, 337-344. Sprent, J . , Korngold, R . , and Molnar-Kimber, K. (1980). Springer Semin. Immunopathol. 3, 213-245. Stastny, P., Ball, E. J., Dry, P. J., and Nunez, 6. (1983). Zminunol. Rev. 70, 113-153. Steinmetz, M., and Hood, L. (1983). Science 222, 727-733. Steinmetz, M., Winoto, A., Minard, K., and Hood, L. (1982). Cell 28, 489-498. Stotter, H., Rude, E., and Wagner, H. (1980). Eur. J. lmmunol. 10, 719-722. Strausser, J. L., Mazumder, A., Grimm, E. A., Lotze, M. T., and Rosenberg, S. A. (1981).J. Zminunol. 127, 266-271. Streilein, J. W., and Bergstresser, P. R. (1980). Transplantation 30, 319-323. Stutman, 0.(1975). Ado. Cancer Res. 32, 261-422. Stutman, 0. (1981). Nandb. Cancer Zminunol. 7, 1-25. Stutman, O. , and Shen, F. W. (1978). Nature (London)276, 181-182. Svejgaard, A., Platz, P., and Ryder, L. P. (1983). Immunol. Reo. 70, 193-218. Tevethia, S. S., Blasek, J.. Wanek, J., and Goldstein, A. (1974).J . lmniunol. 113, 1417-1421. Thomas, L. (1959). In “Cellular and Humoral Aspects of the Hypersensitive States” (H. S. Lawrence, ed.), pp. 529-533, Harper (Hoeber), New York. Thorley-Lawson, D. A. (1980). J. lmmunol. 124, 745-751. Thorley-Lawson, D. A,, Chess, L., and Strominger, J. L. (1977).J. Exp. Med. 146, 495-508. Ting, C.-C., and Yang, S. S. (1982). lnt. J. Cancer 30, 625-632. Ting, C.-C., and Zhang, S.-R.(1983). Znt. J. Cancer 32, 385-391. Townsend, A. R . M . , and Skehel, J. J. (1982). Nature (London) 300, 655-657. Uyttenhove, C., Maryanski, J., and Boon, T. (1983).J . E x p . Med. 157, 1040-1052. Vanky, K., and Klein, E. (1982). Int. J. Cancer 29, 547-553. Van Voorhis, W. C., Valinsky, J . , Hoffman, E., Luban, J., Hair, L. S., and Steinman, R. M. (1983). I . E x p . Med. 158, 174-191. Virelizier, J. L., Allison, A. C . , Oxford, J . S., and Schild, G . C. (1977). Nature (London) 226, 52-53. Vose, B. M., and Bonnard, G . I). (1982). lnt. J. Cancer 29, 33-39. Wagner, H., Pfizenmaier, P., and Rollinghoff, M. (1980). Adv. Cancer Res. 31, 77-124. Walker, E., Warner, N. L., Chestnut, R., Kappler, J., and Marrack, P. (1982). J . Zmmunol. 126, 2164-2169. Wallach, D., Fellous, M., and Revel, M. (1982). Nature (London) 299, 833-836. Webster, R . G . , Laver, W. G . , Air, G . M., and Schild, G . C . (1982). Nature (London) 296, 115-121.

T C E L L SURVEILLANCE A N D M H C

65

Weiss, A., Brunner, K. T., MacDonald, H. R., and Cerottini, J.-C. (1980)./. Exp. Med. 152, 1210-1225. Weiss. A., Maclhnald, H. R., Cerottini, J.-C., and Brunner, K. T. (1981).J. Ztnmunol. 126, 482-485. Wettstein, P. J. (1981). Zmn~unogenetics14, 241-251. Wettstein, P. J. (1982). J. Imtnunol. 128, 2629-2633. Wettstein, P. J., and Frelinger, J. A. (1977). J . E r p . Med. 146, 1356-1366. Wettstein, P. J., and Frelinger, J. A. (1980). Zrninunogenetics 10, 211-225. Wettstein, P. J., and Melvold, R. W. (1983). Initnunogenetics 17, 109-123. Williams, A. F. (1984). Nature (London) 308, 108-109. Wilson, B. S., Indiveri, F., Molinaro, G. A., Quaranta, V., and Ferrone, S. (1980). Transplant. Proc. 12, 125-129. Wilson, D. B., Blyth, J. L., and Nowell, P. C. (1968).J. Exp. Med. 128, 1157-1181. Winchester, R . J., Ross, G. D., Jarowski, C. I., Wang, C. Y., Halper, J . , and Broxmeyer, H. E. (1977). Proc. Natl. Acad. Sci. U . S . A . 74, 4012-4016. Winchester, R. J., Wang, C. Y . , Gibofsky, A., Kunkel, H. G., Lloyd, K. O., and Old, L. J. (1978). Proc. Natl. Acad. Sci. U . S . A . 75, 6235-6239. Winoto, A., Steinmetz, M., and Hood, L. (1983). Proc. Natl. Acad. Sci. U . S . A . 80, 3425-3429. Yamaga, K. M., PfatTenbach, G. M., Pease, L. R., McGovern, D., Nisizawa, T., Melvold, R. W., Kohn, H. I., and Nathenson, S. G. (1983). Zrnrnunogenetics 17, 31-41. Yanagi, Y., Yosikakai, Y., Legget, K . , Clark, S. P., Aleksander, I., and Mak, T. W. (1984). Nature (London) 308, 145-149. Yewdell, J. W., Frank, E., and Gerhard, W. (1981). J. Ztnniunol. 126, 1814-1519. Yu, A. S., and Bernstein, I. D. (1982). J. Zmsnunol. 129, 856-860. Zarling, D. A , , Keshet, I., Watson, A , , and Bach, F. H. (1978).Scand. J. Imtnunol. 8, 497-508. Zarling, J. M., hobins, H. I., Raich, P. C., Bach, F. H., and Bach, M. L. (1978). Nature (London) 274, 269-271. Zeleny, V., Matousek, V., and Lengerova, A. (1978). 5, 41-47. Zink, G. L., and Heyner, S. (1977). Itnnzunogenetics 4, 257-266. Zinkernagel, R. M. (1976). J. E x p . Med. 143, 437-443. Zinkernagel, R. M. (1978). Znimunol. Rea. 42, 202-270. Zinkernagel, R. M., and Iloherty, P. C. (1974a). Nature (London) 248, 701-702. Zinkernagel, R. M . , and Doherty, P. C. (1974b). Nature (London) 251, 547-548. Zinkernagel, R. M., and Doherty, P. C. (1975).J. E x p . Med. 141, 1427-1436. Zinkernagel, R. M . , and Doherty, P. C. (1979). Arlu. Zmnzunol. 27, 51-177. Zinkernagel, R. M., and Klein, J. (1977). Zrntnutiogenetics 4, 581-590. Zinkernagel, R. M., and Oldstone, M. B. A. (1977). Proc. Natl. Acad. Sci. U . S . A . 73, 36663670. Zinkernagel, R. M., and Rosenthal, I<. L. (1981). Ztnniunol. Rea. 58, 131-155. Zinkernagel, R. M., Adler, B., and Althage, A. (1977). Ztntnunogenetics 5, 367-378. Zinkernagel, R. M., Althage, A , , Cooper, S., Kreeby, G., Klein,.'l A., Sefton, B., Flaherty, L., Stimpfling, J.. Shremer, D., and Klein, J. (1978). J. E x p . Med. 148, 592-606. Zur Hausen, H. (1980). Adu. Cancer Res. 33, 77-107. Zweerink, H. J . , Askonas, B. A , , Millican, D., Courtneidge, S. A,, and Skehel, J. J . (1977). Nature (London)267, 354-356.

This Page Intentionally Left Blank

IMMUNOHISTOLOGICALANALYSIS OF HUMAN LYMPHOMA: CORRELATION OF HISTOLOGICAL AND IMMUNOLOGICAL CATEGORIES Harald Stein,’ Karl Lennert,’ Alfred C. Feller,* and David Y. Masont *Institute of Pathology, Christian Albrecht University, Kiel. Federal Republic of Germany, and tDepartment of Haematology. John Radcliffe Hospital, Oxford. England

I. Introduction ............ .... 67 11. Distinction o eoplasrns . . . . . . . . . . . . . . . . . . 69 111. Division of Malignant Lymphoma into Hodgkin’s Lymphoma, Non-Hodgkin’s 71 Lymphoma, and True Histiocytic Sarcoma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 IV. B Cell Lymphomas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 A. Classification by Iinr f the Morphological T B. Classification by Mo with Antigen Profile ................... 80 V. T Cell Lymphomas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 A. Correlation of Clinically and Morphologically Defined T Cell Lymphomas with Antigenic Profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 B. Prethymic and Thymic (TdT-Positive) Lymphoblastic LymphomaILeukemia . 120 C. Peripheral T Cell LyrnphomadLeukemias 127 VI. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 142 References

I. Introduction

For many years human lymphomas have been the subject of much dispute among histopathologists. This debate has sprung from the great variation in morphological appearance And clinical behavior of these neoplasms, and also from the difficulty in distinguishing between neoplastic and reactive lymphoid disorders. Morphological studies gave rise to a number of different classification systems, culminating in 1966 in Rappaport’s scheme. However, it became evident by this time that further progress in the classification of lymphoma was impossible using purely morphological criteria. One means by which this impasse could be resolved was by application of the knowledge which was emerging from immunological research laboratories concerning the existence of two lymphocyte populations (T and B cells), and the ability of lymphoid cells to react to a variety of stimuli (e.g., antigens, mitogens) by transforming into large blast cells capable of division. It appeared possible that different types of malignant lymphomas might represent neoplasms arising from one or the other of these two cell lineages at different stages of transformation. In an early attempt to relate morphological lymphoma entities to immu67 ADVANCES IN CANCER RESEARCH. VOL. 12

Copyright D 1984 I’y Arddroiic Prera. Inc. All rights nf reproduction m any fnrm reserved ISBN 0 - 1 2 - W 2 - 4

68

HARALD STEIN ET AL.

nological data the immunoglobulin (Ig) content of tissue homogenates from malignant lymphoma biopsies was investigated. These results (Stein et al., 1972, 1973a, 1974a) indicated that tumors previously termed “reticulum cell sarcomas” (and therefore deemed to arise from the reticuloendothelial system) were capable of Ig production and could hence be classified as Bimmunoblastic lymphoma (Stein et n l . , 1974a,b). This technique also allowed the entity of lymphoplasmacytic/cytoid lymphoma (LP immunocytoma) (rich in Ig) to be more clearly recognized and to be distinguished from other types of lymphomas, notably chronic lymphocytic leukemia (in which Ig is present at only low levels) (Stein et al., 1973a, 1974a, 1975). On the basis of these findings the Kiel lymphoma classification scheme was drafted (Lennert et al., 1975; Gerard-Marchant et al., 1974). The publication of this scheme coincided with the appearance of the Lukes and Collins independently produced classification system (Lukes and Collins, 1974, 1975), which was similar in many respects to the Kiel scheme (Lennert et d . ,1983). The Lukes and Collins scheme was based on a comparison of the nuclear profiles of normal and neoplastic lymphoid cells, as seen by the camera lucida technique. Concurrently with the development of techniques for analyzing Ig in tissue homogenates, studies of B and T lymphocyte surface markers began to be applied to the characterization of malignant lymphoid neoplasms. These investigations were restricted initially to those lymphoproliferative disorders in which there is leukemic spread (e.g., CLL, Sezary syndrome, etc.) since this facilitated the studies of neoplastic cells in suspension (Preud’ Homme and Seligmann, 1972; Aisenberg and Bloch, 1972). Later, however, suspensions prepared from solid lymphoid neoplasms began to be investigated. In retrospect it can be seen that many of‘ these cell suspension studies were of limited value. The analysis of isolated lymphoma cells cannot provide any information on the topographical relationship between different cellular elements, e.g., it is impossible to distinguish between cells which have come from areas heavily infiltrated with lymphoma and those extracted from normal residual lymphoid areas. Furthermore the preparation of cell suspensions introduces the risk that cells will be lost and that certain cell types will be selectively extracted, e . g . , reactive nonneoplastic cell populations may be spuriously overrepresented in the final cell suspension. Thus the development of techniques which enable neoplastic cells to be characterized in situ (i.e., in tissue sections) rather than in cell suspension was a major advance. The earliest studies in this field were reported by Shevach et al. (1973), who applied rosetting techniques to frozen sections. However the fine specificity (in terms of the ability to visualize labeling of individual cells) of this method and also the number of different surface markers which could be detected were very limited. In 1974 Taylor and Burns reported that cytoplasmic Ig could be revealed

H U M A N LYMPHOMA

69

by immunoperoxidase staining of paraffin sections and much interest was generated in the possibility of analyzing neoplastic lymphoid cells by this method. However, a major limitation of this approach was that only a minority of lymphomas contain intracytoplasmic immunoglobulin in substantial quantities (i.e., those which show differentiation toward the terminal plasma cell stage). Smaller quantities of intracytoplasmic immunoglobulin may be found in lymphomas derived from earlier B cell maturation stages (e.g., germinal center cell tumors), but the quantities present are frequently too low to be detected in formalin-fixed paraffin-embedded material. Furthermore early studies overlooked the frequency with which exogenous serum immunoglobulin diffuses into cells during tissue processing, giving rise to spuriously polyclonal labeling reactions (Mason et al., 1980). The introduction in 1976-1978 of methods for staining surface immunoglobulin on neoplastic cells in cryostat sections hence represented a major technical advance (Hoffmann-Fezer et al., 1976; Warnke and Levy, 1978). In this type of preparation membrane immunoglobulin and other surface antigens are well preserved, and it thus becomes possible to demonstrate the monoclonal nature of many lymphoid neoplasms (Hoffmann-Fezer et al., 1978, 1979, 1981; Warnke and Levy, 1978). The technical aspects of this procedure were optimized by Stein et al. (1980, 1982b), who achieved iminunoperoxidase labeling substantially stronger than had been reported previously. This opened the possibility of analyzing with a high degree of sensitivity the expression of membrane antigens on neoplastic cells and also in the same preparation to visualize cell morphology and topographical cellular relationships. By a fortuitous coincidence this period also saw the increasingly wide availability of monoclonal antibodies directed against human white cell differentiation antigens. These reagents greatly extended the repertoire of surface markers which could be detected on human lymphoid tumors and immunostaining of these antigens by the recently developed immunoperoxidase labeling methods represented a powerfd new analytical approach. These methods have now been used in the authors’ laboratories for the past 3 years to analyze the full spectrum of human lymphoproliferative disorders. In the present article the results of these ongoing studies are reviewed. It should be emphasized that this review is based primarily on information from the authors’ laboratory. References are given to a number of studies pelformed in other laboratories which are directly relevant to our own results. However, we have not attempted a full review of the current literature in this field and readers may wish to refer to publications from other centers. II. Distinction of Malignant Lymphoma from Other Neoplasms

It is usually an easy matter for histopathologists to distinguish malignant lymphoma on morphological grounds from neoplasms of nonlymphoid ori-

70

HARALD STEIN ET AL.

TABLE I IMMUNOHISTOLOCICAL DISTINCTION OF LYMPHOMAS FROM CARCINOMAS WITH MONOCLONAL ANTIBODIES ~~

Antigens"

Lymphoma

Carcinoma

Leukocyte common antigen Cytokeratins Milk fat globule Carcinoembryonic antigen

+b

--/(+) -

+/-

HLA-DRC

+/-

-I(+)

-

+I+/-

4 For details of the monoclonal antibodies detecting these individual antigens see Table 111. b +, All cases positive; +I-, majority of cases positive; +I( -), occasional cases negative; -I(+), occasional cases positive; -, all cases negative. c The results indicate that HLA-DR is not a reliable marker for the distinction between malignant lymphoma and carcinoma, since this antigen is occasionally expressed on carcinomas, while being absent from approximately 10%of B cell lymphomas of large cell type and nearly all T cell lymphomas.

gin. However, a variety of nonlymphoid malignancies may, on occasion, show histological similarity to malignant lymphoma, especially of large cell type. In particular difficulties may arise over the distinction between malignant lymphoma and anaplastic carcinoma. Recently these diagnostic problems have been greatly simplified by the use of antibodies which can distinguish between white cells and epithelial cells. Many of these antibodies are monoclonal reagents specific for antigens which could not previously be detected with polyclonal antisera. Of particular value have been antibodies against leukocyte common antigen (present on most lymphomas, absent from carcinomas) and against epithelial intermediate filaments (present in carcinomas, but absent from lymphomas). Details of these and other markers are given in Table I. One obstacle to the widespread use of these cell lineage-specific antibodies for routine tissue diagnosis lies in the fact that the antigens they detect are destroyed by formalin fixation and histological processing procedures. However, three monoclonal antibodies which react with the antigen common to all white cells (leukocyte common antigen), with milk fat globule antigen and the carcinoembryonic antigen (CEA) not only in frozen sections but also in routinely processed paraffin-embedded sections have recently been reported. Using these antibodies Gatter et al. (1984) recently reported on 37 neoplasms which had been referred for immunohistological typing because of difficulty in deciding on morphological grounds whether the tumors were primitive carcinomas or malignant lymphomas. In all but

H U M A N LYMPHOMA

71

one case the diagnosis could be established on the basis of the staining reactions of the two antibodies. Twenty-nine cases expressed leukocyte common antigen and were negative for milk fat globule antigen and CEA, indicating that they were of white cell origin and hence probably lymphomas. The seven remaining cases showed the opposite marker profile identifying them as being of probable epithelial origin, i.e., carcinomas. These results suggest that the majority of neoplasms of questionable origin seen in routine histological practice may be correctly classified by immunohistological staining, even when only routinely processed paraffin-embedded tissue is available. They also illustrate that the majority of these cases are of lymphoid origin, a finding which may reflect the relative unfamiliarity of the routine pathologist with the full range of morphological appearances of lymphomas. Ill. Division of Malignant Lymphoma into Hodgkin’s Lymphoma, Non-Hodgkin’s Lymphoma, and True Histiocytic Sarcoma

Malignant lymphomas (ML) show a wide range of morphological features which prompted many pathologists in the past to classify them on the basis of these appearances. Although much misinterpretation occurred, it was generally agreed to divide ML into three main categories: lymphomas derived from lymphoid cells (NHL), neoplasms derived from macrophages/histiocytes, and Hodgkin’s disease. Within the last 13 years a number of cell lineage markers were found which have made it possible to investigate the validity of these basic morphology groupings. Table I1 shows the results of our analysis of more than 250 malignant lymphomas. This investigation confirmed that the tumor cell involved in Hodgkin’s lymphoma is different from those involved in nonHodgkin’s lymphoma and true histiocytic sarcoma, not only in morphology but also in antigenic profile. Furthermore, these studies provide evidence that a tumor cell of a single basic type proliferates in all histological types of Hodgkin’s lymphomas (Stein et ul., 1982b). In contrast at least two different cell populations are involved in non-Hodgkin’s lymphomas, namely B and T cells. Until 1970 it was widely believed (e.g., Lennert, 1964; Rappaport, 1966) that a significant number of non-Hodgkin’s lymphomas, especially those of large cell type, were histiocytic in origin (“reticulosarcoma/histiocytic lymphoma”). However, in the early 1970s the measurement of Ig extracted from lymphoma tissue samples revealed that the majority (60-70%) of those large cell lymphomas (“reticulum cell sarcoma”) contained greatly increased amounts of IgM suggesting that they were of lymphoid origin (Stein et al., 1972, 1974c; Stein, 1978). This finding was confirmed subsequently with cell suspension and immunohistological studies using an extensive panel of B, T, and macrophage-associated markers (Habeshaw et al., 1979a; Warnke et al.,

TABLE I1 ANTIGENIC REACTIVITYOF NEOPLASTIC CELLSIN THE MAJOR CATEGORIES OF MALIGNANT LYMPHOMAS

Lymphoma categories

B cell associated antigen To15

Non-Hodgkin's lymphoma B-tyPe T-tyPe Hodgkin's disease Large cell tumors (True) histiocytic sarcomas

143 0

5= 0 0

" UCHTl

SIg

T cell associated antigens"

134 0 0 0 0

0 70 15 4e 0

HLA-DR

141 10 36

9 2

SternbergReed cell associated antigen Ki-1

Macrophage associated antigensb

2c

0

Od

0 0 0 2

40

9

0

Lysozyme

al-Antitrypsin

Number of cases

0 0 0

0 0 13

152

0 2

9 2

70 40

9f 2

(detects T3 antigen) and/or OKT11. Antimonocyte 1, antimonocyte 2, OKM1, and S-HCM. c These two SIg+ and Ki-1 cases may represent the neoplastic equivalent of Epstein-Barr virus-positive lymphoblastoid cell lines, which are also positive for SIg and Ki-1. Some cases in another series of T cell lymphomas showed variable numbers of Ki-1-positive cells. Those cases were of the pleomorphic subtype (Stein et al., 1984). e The Ki-l+ large cell lymphomas expressing T cell antigens are probably not plain T cell lymphomas but rather neoplasms related to Hodgkin's disease since the Hodgkin's and Stemberg-Reed cells in 15 of the 40 Hodgkin's disease cases also showed coexpression of Ki-1, T cell antigens, and HLA-DR. f Ki-1-positive, al-antitrypsin-negative neoplasms were not included; the al-antitrypsin-negative cases (213 of all Ki-1-positive cases) showed the same surface marker heterogeneity as the ul-antitrypsin-positive cases. +

H U M A N LYMPHOMA

73

1980; Stein et al., 1980, 1981b). In the series of malignant lymphomas presented here only 2 out of 250 cases of malignant lymphoma proved to be macrophage related on the basis of marker studies. This low frequency of histiocytic sarcoma is in line with the data published by Lukes et al. (1978) but is in sharp contrast to the findings presented by Van der Valk et al. (1982a) and other groups. The latter authors reported a histiocytic origin for approximately 10- 15% of all non-Hodgkin’s lymphoma. Isaacson and Wright (1978) and lsaacson et al. (1983) have also reported a high frequency of histiocyte-derived lymphomas; however, this finding was based on a study of gastrointestinal lymphomas, particularly those arising in celiac disease patients. Both groups used lysozyme and a,-antitrypsin as macrophage markers, but stressed that lysozyme was positive in only a minority of cases whereas a,-antitrypsin was detectable in most cases. To clarify these contradictory results, we have investigated all our cases for the presence of a,-antitrypsin. This constituent was found to be absent from all cases showing clear-cut B and T cell phenotype but to be present in two lysozyme-positive tumors. G. Pallesen (personal communication, June 1978) obtained very similar results on immunostaining more than 100 cases of non-Hodgkin’s lymphomas for the presence of a,-antitrypsin. Furthermore, Hodgkin and Sternberg-Reed cells in approximately one-third of all cases of Hodgkin’s disease (Isaacson, 1979; Stein et al., 1982b), and also the neoplastic cells in nine unclassifiable lysozyme-negative large cell tumors (lacking a clear-cut B or T cell phenotype) were positive for a,-antitrypsin. The majority of these latter cases resembled malignant histiocytosis in morphology while a smaller proportion showed some morphological similarity to immunoblastic lymphoma. All of these lysozyme-negative a,-antitrypsinpositive large cell tumors expressed the Sternberg-Reed cell-associated antigen Ki-1. Our conclusion is that the neoplastic cells in these tumors are closely related to the tumor cells of Hodgkin’s disease and thus may belong to the group of Hodgkin’s lymphoma rather than to the true histiocytic sarcoma or non-Hodgkin’s lymphoma group. A report of further details of the antigen profile, the morphology, and clinical behavior of these tumors has been submitted. The remainder of this review will be restricted to cases of lymphomas which are derived from either B or T cells. IV. B Cell Lymphomas

In attempting to correlate immunological typing with conventional morphological classification two approaches may be adopted: one may categorize the tumors on immunological grounds into homogeneous groups showing identical phenotype and then examine the morphological features of neoplasms in each of these groups, or one may undertake the reverse operation,

TABLE 111 ANTIBODIES USED IN MORPHOLOGICALCLASSIFICATION OF B CELLLYMPHOMAS

Antibody

Specificity

Molecular weight

PD7l26 2Bll F10, F10-8a-4 TU35

Leukocyte common antigen Leukocyte common antigen Leukocyte common antigen HLA-DR

200,000 200,000 200,000 28,000/34,000

To15 F8-10-13

All B cells All B cells. subset of T cells

140,000 215.000

2 Anti-IgM

IgM

Anti-IgD Anti-IgG Anti-IgA Anti+ Anti-A VIL-A1 TUl

IgD

C3RTo5 Leu-1

TU33 Tll/Lyt 3

IgA Kappa Lambda Common ALL antigen (CALLA) Follicular mantle lymphocytes, subset of follicular dendritic cells C3b receptor All T cells, B-CLL, centrocytic lymphoma, follicular mantle lymphocytes weakly Vast majority of T cells Sheep erythrocyte receptor

100,000 45.000

Reference Wamke et al. (1984) Warnke et al. (1984) Dalchau et al. (1980) Ziegler et al. (1982) Stein et al. (1982a) Dalchau and Fabre (1981) Bethesda Research Laboratory (BRL) BRL BRL BRL BRL BRL Knapp et al. (1982) Ziegler et al. (1981)

205,000 65-69,000

Gerdes et al. (1982) Engleman and Levy (1980) (B-D)

130,000 55,000

Ziegler et al. (1984) Verbi et al. (1982) Kamoun et al. (1981)

Equivalent or identical antibodies DAKO-LC L243 (BD=), DAKOHLA-DR DAKO-pan-B

DAKO-IgM DAKO-IgD

DAKO-lambda

J5 (Ritz et al., 1981b) DAKO-C3b receptor DAKO-T1, OKTl

Leu-5 (BD)

UCHTl

All T cells

19,000

TUl4

All immature and majority of mature T cells

40,000

Beverley and Callard (1981) Ziegler et al. (1984)

TU33

Mature T cells Inducer/helper T cells, Langerhans cells. macrophages Suppressorlcytotoxic T cells

110,000 62,000

Ziegler et al. (1984) Evans et al. (1981)

Leu-3a OKT8 Leu-7 T-ALL2 NA1/34 4 VI

Antimonocyte 1

Antimonocyte 2 OKMl S-HCL.3 R4l2.3 Antilysozyme Anti-a l-antitrypsin Ki-1

Natural killer cells Interdigitating reticulum cells, cortical thymocytes Interdigitating reticulum cells, cortical thymocytes, Langerhans cells Monocytes, macrophages, small subset of lymphoid cells, renal podocytes and tubule cells, etc. Monocytes, macrophages Granulocytes, monocytes, macrophages Macrophages, hairy cell leukemia cells, granulocytes (weakly) Follicular dendritic reticulum cells. splenic marginal zone cells Lysozyme al-Antitrypsin Sternberg-Reed cells, “Ki-1” cells in normal lymphoid tissue, lymphoblastoid cell lines hut not Burkitt lymphoma lines

BD. Becton Dickinson.

33,000

Reinherz et d. (1980)

OKT3

WT1 (Tax et al. (1981); 3A1 (Haynes et a l . , 1979); DAKO-T2 OKT12 OKT4 Leu-2a (BD), DAKOT8

Abo and Balch (1981) BRL 49,000

McMichael et al. (1979)

OKT6

BRL

90,000/150,000

BRL Breard et ~ l (1980) .

90,000/150,000

Schwarting et al. (1984) Naiem et al. (1983)

120,000/130,000

DAKOPATTS DAKOPAITS Schwab et al. (1983)

DAKO-DRCI

76

HARALD STEIN ET AL.

i. e., to classify morphologically and then to examine the immunological features. In the following, B cell lymphomas are considered from these two points of view. The antibodies on which the immunophenotype analysis was based are listed in Table 111.

A. CLASSIFICATION BY IMMUNOPHENOTYPE In order to group non-Hodgkin’s lymphoma of B type showing identical immunological phenotype together we selected the presence of the B cell antigen To15, the presence and class of surface Ig, the expression of common ALL antigen (CALLA), T U l , Leu-1, and S-HCL3 antigens and the presence of dendritic reticulum cells (DRC). This approach which yielded seven major categories and the results are shown in Table IV. Among 180 non-Hodgkin’s lymphomas of B type, 115 clearly fitted into one of these seven immunological types. The immunological categories were then correlated with morphology (last column of Table IV). It is evident that two immunological types (Types I1 and 111) were consistently associated with a specific morphological category of lymphoma (i.e., hairy cell leukemia; follicular centroblasticcentrocytic lymphoma). Type V (Burkitt-like immunophenotype) and Type VI cases (pre-B cell phenotype) were all linked morphologically with lymphoblastic lymphoma. However, Type V contained nine cases of Burkitt’s lymphoma, one case with “lymphoblastic lymphoma B-non-Burkitt” morphology and one case with “lymphoblastic lymphoma unclassified.” In Type VI the cases were of non-Burkitt types (both “unclassified” type and “B-nonBurkitt type”). There was also a good (although less than absolute) correlation between Types I and IV and their morphological features. Type I covered all B chronic lymphocytic leukemia cases but was not specific for this disease entity since it also contained 3 of 11 lymphoplasmacytic/cytoid immunocytomas. Nearly all cases of centrocytic lymphoma and one case of immunoblastic lymphoma were found in Type IV. The last type (VII) included approximately half of the large cell lymphomas, two centrocytic lymphomas, and four LP lymphomas, indicating that the Type VII marker profile was most characteristic of large cell lymphomas. Among the total of 180 B,cell lymphomas investigated 34 could not be fitted into one or other of the seven major immunohistological categories. These 34 cases showed a wide variety of different antigenic profiles and hence did not cluster into obvious groups. However, it is important to note that immunohistological typing identified, with two exceptions, categories of lymphoma which corresponded to groups already distinguished in the Kiel classification. This substantiates the view that the lymphoma types described in the Kiel classification represent true biological entities rather than arbitrary subjective groupings.

TABLE IV IMMUNOLOGICAL CATEGORIZATION OF B CELL NON-HODGKIN'S LYMPHOMA (RELATIONSHIPTO HISTOLOGICAL CLASSIFICATION)"~b Immunological type

B cell antigen To15

IgM

I I1

+d

V

+ + + +

VI

+

VII

+I-

111

IV

IgD

IgG

CALLA

TUl

Leu-1

S-HCL3

-

Presence of DRCC

Histological type

~

~~

+

+

-

+

+

-

+I4-1-

+I-

i/-I+

-

-

i

-/+

+/-

+/-

-

-

i

-

+

-

i+ -

-

-

-

i i

-

-

i

i/-

+

-/+

-

-

i/-/i -I+

-

-

-

-

None or few None Spherical meshwork Diffuse or nodular meshwork None or few

Some cases show meshwork pattern

16/18 B-CLL; 3/11 LPL 12/12 HCL 22/22 follicular CB-CC 16/18 CC; 1/18 IB 919 LB (Bu) 115 LB(BnonBu) 1/10 LB(U) 3/10 LB (U) 415 LB (B non-Bu) 1/10 LB (U) 9/23 CB; 11/18 IB; 2/18 CC; 4/11 LPL;

a B-CLL, Chronic lymphocytic leukemia; LPL, lymphoplasmacyticlcytoid lymphoma; HCL, hairy cell leukemia; follicular CB-CC, follicular centroblastic-centroytic lymphoma; CC, centrocytic leukemia; IB, immunoblastic lymphoma; LB(Bu), lymphoblastic lymphoma, Burkitt type;

LB(BnonBu), lymphoblastic lymphoma, B non-Burkitt-type; LB(U), lymphoblastic lymphoma, unclassified. b Immunological data were based on an analysis of 180 cases. The majority of cases were also classified according to the Kiel scheme (see last column) although cases in which the morphological category was not clear cut have been excluded. DRC, Dendritic reticulum cells. d +, All cases positive; +/-, majority of cases positive; -/+, majority of cases negative; all cases negative.

-.

78

HARALD STEIN ET AL.

TABLE V. SUMMARY OF IMMUNOCWOCHEMICAL LABELING REAC-

hntigenslantibcdies Leukocyte common antigen HLA-DR B cell antigen (To15) B cell-associated antigen (F8-11-13)C SIgMd SIgD SIgC

SIY S-kappa S-lambda Kappa and lambda kappa:lambda ratio SIg negative cytoplasmic Igd CALLA TO1 C3b receptor Leu-l/TI TO33 DRC antigen (R4/23) TIULyt 3 UCHTl/OKT3 TU14/T2 Leu 3a/T4 T8 Leu 7 T-ALL2/NA1/34 Antimonocyte I Antimonocyte 2 OKMl S-HCU Associated cells Meshwork of follicular dendritic cells T zones Interdigitating cells Fnllicular growth pattern

B CLL (n=l8)

Prolymphocytic leukemiab (n=5)

18 18 18 18

17 16 0

0 10

7 0 1.4:1

0 0

0 18

6

5 5

na 5 3 1 0 0

4 0 04 0 na 0 0

0 2e o/ 10 0 0 0 0 0 0 0 0 0

na 4 2 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0

in 9

Hairy cell leukemia* (n=12)

Lyniphoplasmacytic cytoid lymphonia (n=ll)

Centroblasticcentrocytic lymphoma (n=22)

12

11

22

12 12 12

11 11

22

11

22

n

10

5

7

2 1

3 0

5 7 0 0.7:l 0 n.a.

7 3

10 7 11 1 11 8 3 1.3 1 0 n.a. 22

2 0 0 0

1

2:I 0 I1 0

7 4

22

16

22 0 0

0 0/7 12

6 3 0 0 0 0 0 0 0 0 0 0 0 0/3

0 0 0/10 0 0 0 0 0 0 0 0 0

0

0

0

22

0 0 0

0 0 0

0 0 0

22

0 0 0 0 0 0 0 0 0 0

8

22 ~~

The majority of the entries refer to the labeling reactions of neoplastic cells within the biopsies. The reactions of associated cells are indicated in the lower portion of the table. b Four cases of prolymphocytic leukemia and six of the cases of hairy cell leukemia were labeled as cell smears rather than in cryostat sections. Antibody F8-11-13 detects a variant of the leukocyte common antigen which is preferentially (but not exclusively) expressed on B cells (Dalchau and Fabre, 1981). d SIgM, SIgD, etc. indicate that the staining was performed on cryostat sections, and that a

79

H U M A N LYMPHOMA

TIONS OF

NON-HODGKIN’S LYMPIIOVA OF B CELLTYPE(152 CASES)“

Centrocytic lymphoma (n= 18)

Centroblastic lymphoma (n=23)

Immunohlastic lymphonla B type (n=IR)

Lyniphoblastic lymphoma, Burkitt type (n=9)

Lymphoblastic lymphoma, unclassified

18

23

18

18

19

I8

21 20

16 16

18 16 15 1 0

12

5 12 0

9

041 0 na 0 0 11 17 13 0 0 015 0110 0 0 0 0 0 0 0 0

4 4

I 6 0 1.5:l I

Multiple myeloma

(n=4)

Lymphohlastic lymphoma, B-nonBurkitt type (a=7)

9

4

I

0

9 9

4 4

14

9

2

I I 5

0 0 0

10 5 3

8

I

3 0 0

1 1 1 0

0 0 4 0

4

0 0 0 0

I

4

3 0

I 0

1 0

1 0 4:1 5d

I I 5 I 14.1 4 n.d. 4

1.3.1

1:l 3 n.d. 4 1

Oil0 0 0 0 0 0 0 0 0

3 1 0 0 0 015 0 0 0 1 0 0 0 0 0

0 n.d 9 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

16

R

0

2

2

5 0 5

0 0 0

0 0 0

0 2

n.a. 9

5 2 I 0 0 0 015

1

4 n.d. 4 0 0 2 0

(n=5)

5

0 0

1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0

0

0

0 0 0

0 0 0

0 0 0

I 0 0 1 0 0 0 0 0 0 0 0 0 0 oi 1

0 0 0 0

0 0 0 0 0 0

therefore the majority of positive reactions represent surface Ig. However intracytoplasmic Ig would also be visualized in those cases of NHL in which the neoplastic cells contain intracellular Ig (lymphoplasmacytic/cytoidlymphoma and multiple myeloma). “Cytoplasmic Ig” refers to the reactions of paraffin-embedded tissue sections, in which membrane Ig is not detected. e In these two cases it was impossible to be certain that this antigen was present on the tumor cells rather than on the numerous infiltrating T cells.

80

HARALD STEIN ET AL.

B. CLASSIFICATION BY MORPHOLOGY A N D CORHELATION OF THE MORPHOLOGICALTYPESWITH ANTIGENPROFILE The immunological features of cases in the different histological types of the Kiel classification were reviewed. The results (Table V) indicate that all B cell lymphomas of low grade malignancy consistently express HLA-DR, the B cell-associated antigens detected by antibodies To15 and F8-11-13 and surface immunoglobulin; whereas among high grade lymphomas of centroblastic and immunoblastic type, HLA-DR, B cell antigens, or surface immunoglobulins are missing in a number of cases. In the following sections the main characteristics of each histological lymphoma category will be discussed separately. 1 . Chronic Lyinphocytic Leukemia of B Cell Type (B-CLL) B-CLL in the Kiel classification is roughly equivalent to “Group A malignant lymphoma, small lymphocytic” of the Working Formulation (The NonHodgkin’s Lymphoma Pathologic Classification Project, 1982). However, there is a difference in that Group A also includes cases showing plasmacytoid differentiation; such cases would be categorized in the Kiel classification as lymphoplasmacytic/cytoid lymphoma. B-CLL is usually characterized by a slowly progressive lymphocytosis and infiltration of the bone marrow and moderate lymph node enlargement; splenomegaly usually follows after some months or years. Only in a minority of cases does splenic enlargement occur without lymph node enlargement (Dighiero et al., 1979). B-CLL in the Kiel classification appears generally homogeneous in terms of morphology, as illustrated in Fig. l a and b. However, there may be variations in the presence and size of pseudofollicles (see Fig. la). These structures consist of medium sized and large cells (prolymphocytes and paraimmunoblasts) and their presence or absence forms the basis of the following subclassification of B-CLL (Lennert and Mohri, 1978): (1)diffuse type with only occasional paraimmunoblasts and prolymphocytes; (2) pseudofollicular type with clearly recognizable pseudofollicles containing paraimmunoblasts

FIG. 1. Chronic lymphocytic leukemia of B cell type (B-CLL). (a) Typical histological appearance. Note the pseudofollicular pattern (not seen in T-CLL). Giemsa, X30. (b) Higher magnification showing that the pseudofollicles are made up of large lymphoid cells. The cells surrounding the pseudofollicles resemble follicular mantle lymphocytes in morphology. Giemsa, X200. (c) Frozen section immunostained for kappa chains. Most of the cells are positive. X 100. (d) Frozen section immunostained with a monoclonal antibody R4/23 which recognizes follicular dendritic reticulum cells (FDRC). A small focus of a meshwork of solitary, residual FDRC is seen in the upper half of the figure. All other cells are unstained. X50.

H U M A N LYMPHOMA

81

82

HARALD STEIN ET AL.

TABLE VI THEANTIGENIC PROFILE OF B-CLL A N D SMALL LYMPHOCVTIC LYMPHOMA I N PREVIOUS STUDIES REPORTED

Reference

Number of cases

Aisenberg et al. (1983) Gajl-Peczalska et al. (1982) Habeshaw et al. (1983) Janossy et al. (1983)

7 21 11 60

0

b

IgM

I@

IgG

gp67*

2 n.i.b 1

n.i. n.i.

6 16

60

n.i.

1 n.i. 1 n.i.

7 60

Detected by monoclonal antibodies Leu 1, OKT1, T1, T101, or RFA-1. n.i., no information.

and prolymphocytes; and (3) a tumor-forming type with extensive areas of prolymphocytes. All but two cases in our present series of 18 cases of B-CLL belonged to the pseudofollicular subtype, the remaining two cases being of diffuse type. The immunohistological marker profile (Table V) proved to be highly constant, characterized by the expression of HLA-DR, B cell antigens To15, F8-11-13 antigen, T U l , and Leu-1, frequent expression of surface IgM (Fig. Ic) and surface IgD, and the absence of easily detectable cytoplasmic immunoglobulins, CALLA or antigen S-HCL3. A comparison of our data with that reported by other groups is given in Table VI. The results of Janossy et al. (1983)are in agreement in that all cases of B-CLL expressed the T cell-associated antigen (gp65-69) recognized by antibodies T1, anti-Leu-1, T102, and RFA-1. In other series this antigen was found less constantly [i.e., in only 29 of the 39 cases reported by Aisenberg et al. (1983); Habeshaw et al. (1983); Gajl-Peczalska et al. (1982); see Table VI]. Our data also differ in other aspects from the results reported by Aisenberg et al. (1983)and by Habeshaw et al. (1983); in both these studies only one or two 1g.M-positivecases are reported (Table VI) whereas all but one of our cases were clearly positive for this surface marker [as were all of the 60 cases reported by Janossy et al. (1983)l. These discrepancies may reflect differences in sensitivity of the immunocytochemical detection systems used in individual laboratories. The series of 21 small lymphocytic lymphoma cases investigated immunohistologically by Gajl-Peczalska et al. (1982) deserves some special mention since they were classified histologically according to the Working Formulation (as Type A). All the cases were surface Ig positive as were all our 18 cases of B-CLL; in contrast to our series, however, only 16 of the 21 cases expressed the gp65-69 T cell-associated antigen. This discrepancy probably reflects the inclusion by Gajl-Peczalska et al. (1982) of lymphomas showing

H U M A N LYMPHOMA

83

plasmacytoid differentiation. These neoplasms are separately categorized (as lymphoplasmacytic/cytoid lymphoma) in the Kiel classification, and, as noted below, often lack gp65-69 (Leu-1). Caligaris-Cappio et al. (1984) recently presented data suggesting that a small subgroup of B-CLL patients have an unusually good prognosis due to the very slowly progressive course of the disease. This subgroup was characterized by the simultaneous expression of RFA-1 and RFA-4 antigen. In our B-CLL series there were two membrane markers which showed an inconstant expression, namely C3b receptors (detected by monoclonal antibody C3RTo5) and TU33. There was no positive or negative correlation between the expression of the two markers and it remains an open question whether they identify significant subtypes of B-CLL. It may be added that in seven of the nine TU33-positive B-CLL cases the TU33 expression was very weak; this is of relevance in relation to centrocytic lymphoma (see below). The pseudofollicles present in 16 of our 18 B-CLL cases deserve comment. These structures have been interpreted in the past as areas of proliferation and/or differentiation. The former assumption could not be consistently confirmed using a monoclonal antibody (Ki-67) which is specific for a proliferation-associated nuclear antigen [since there was variation from case to case (manuscript in preparation)]. However, the use of other antibodies provided evidence for the view that pseudofollicles represent areas of local differentiation, i.e., TU22, a monoclonal antibody specific for HLA-D (or HLA-DC), produced much stronger staining of pseudofollicle cells than of the surrounding small lymphocytes, while the staining for IgD showed (in at least a proportion of cases) the opposite pattern. Using a sensitive threestage immunoperoxidase technique it was possible to obtain a weak cytoplasmic labeling in paraffin-embedded sections for IgM and J chain of pseudofollicular cells but not of small lymphocytes. In uitro phorbol ester stimulation of CLL cells is reported to decrease SIg, increase CIg, and increase secretory mRNA synthesis (Totterman et al., 1981). These findings argue against pseudofollicle cells being at an earlier stage of maturation than the surrounding small lymphocytes and suggest instead that they represent cells which were partially differentiated toward antibody-secreting cells, i.e., immunoblasts. However, full elucidation of the nature of pseudofollicles must await the results of labeling with additional monoclonal antibodies. In this context monoclonal antibody FMC7 would be of interest; it has been reported that this antibody reacts only with prolymphocytic leukemia cells and with B-CLL cells in “prolymphocytoid” transformation (Catovsky et al., 1981; Catovsky, 1982). It would therefore be of interest to investigate whether this antibody reacts selectively with pseudofollicles in B-CLL. The Normal Counterpart of B-CLL Cells. One of the major aims underlying immunological phenotyping of B-CLL cells is to relate these cells to a normal B cell differentiation stage or subpopulation. Comparison of the

84

HARALD STEIN ET AL.

TABLE VII COMPARISON OF T I ~ EANTIGENIC PROFILEO F B-CLL A N D FOLLICULAR MANTLE LYMPHOCYTES

SIgM SIgI) Cytoplasmic Ig CALLA TU 1 Leu-1 a 3'

c

B-CLL

FML

+a

++b

-

-

+ +

(+P

+

++

+

Weak to moderate staining. Strong staining. Very weak but distinct staining.

morphological and immunological phenotype of B-CLL cells with that of normal B cells (Table VII) shows that B-CLL cells most closely resemble follicular mantle lymphocytes, a subpopulation of recirculating B cells (Howard et al., 1972). Both B-CLL cells and follicular mantle lymphocytes express surface IgM, surface IgD and TUl, and lack easily detectable amounts of cytoplasmic immunoglobulin and CALLA. However, this similarity does not hold true for other antigens. Leu-1 is found to be present on B-CLL cells and absent from follicular mantle lymphocytes; and TU33 is present on the neoplastic cells in a proportion of B-CLL cases but is consistently absent from follicular mantle lymphocytes. In contrast C3b receptors are always present on follicular mantle lymphocytes, but are expressed in only 30% of B-CLL samples. Recently Caligaris-Cappio et al. (1982) described a small population of B cells in normal tonsils and lymph nodes whose phenotype was similar to that of typical B-CLLs [low density of surface Ig, gp65-69 positive (i.e., Leu-1), mouse erythrocyte receptor positive]. These authors argued that this small B cell population is the normal counterpart of B-CLL cell. However, recently, by maximizing the sensitivity of immunoenzyinatic staining, we have been able to demonstrate gp65-69 in small amounts on normal follicular mantle lymphocytes (Fig. 2). Hence the apparent differences between B-CLL cells and follicular mantle lymphocytes may be quantitative rather than qualitative.

2 . Prolymphocytic Leukemia of B Cell Type (B-PLL) In 1974 Galton et al. described prolymphocytic leukemia, a lymphoproliferative disorder which may be distinguished from typical B-CLL on both clinical and morphological grounds.

H U M A N LYMPHOMA

85

FIG.2. Frozen section of tonsil imniunostained with the monoclonal antibody anti-Leu-1, using the sensitive inultilayer alkaline phosphatase-antialkaline phosphatase (APAAP) method (Cordell et d., 1984).The T cells both within and outside the B cell follicle are strongly positive. Note that in addition to the strongly stained T cells, the cells ofthe follicular mantle (FM) show a weak but distinct staining. In contrast the B cells of the germinal center (GC), the border of which is indicated by a dashed line, are negative. The staining of follicular mantle lymphocytes by anti-Leu-1 (and other antibodies against the gp65 69,000 antigen) was not seen in the past because of the lower sensitivity of the detection systems used.

In contrast to typical B-CLL, splenic enlargement is frequently seen at presentation, while lymph node enlargement is either absent or minimal, except as a terminal development. The cytological features of prolymphocytic leukemia are shown in Fig. 3. The great majority of prolymphocytic leukemias are B cell-derived. However, these cases differ in membrane phenotype from B-CLL’since FMC7 and surface Ig is expressed at high density, while mouse red blood cell

86

HARALD STEIN ET AL.

Fig. 3. Blood smear from a case of prolymphocytic leukemia of B cell type. Note the prominent centrally located nucleoli and the relatively abundant cytoplasm of the leukemic cells. May-Griinwald-Giemsa. x800.

receptors are only weakly expressed or absent (Catovsky et al., 1981; Catovsky, 1982). All of our five cases of prolymphocytic leukemia also differed from B-CLL in the absence of TUl, an antigen consistently expressed on typical B-CLL cases (Table V). Future studies of a larger number of cases are required to show whether TUl is constantly absent in B prolymphocytic leukemia and thus a valuable reagent for distinguishing B prolymphocytic leukemia from B-CLL. Gobbi et al. (1983) reported that the gp65-69 antigen detected by Leu-1, T1, etc. is absent from prolymphocytic leukemia. However, two of our five cases expressed this antigen. Recently it was shown that treatment of B-CLL cells with phorbol esters converts the B-CLL immunophenotype (negative for FMC7 and RFA4, and weakly positive for SIg) into a B-PLL-like phenotype (positive for FMC7 and RFA4, and strongly positive for SIg) (Caligaris-Cappio et al., 1984). This suggests that B-CLL and B-PLL are closely related leukemias in which the proliferating cells differ by only one or a few differentiation steps. This finding also makes the occasional transformation of a B-CLL into B-PLL understandable. 3. Hairy Cell Leukemia Hairy cell leukemia is a distinctive disease entity which may be distinguished from other malignant lymphomas by the following characteristic

H U M A N LYMPHOMA

87

features: a chronic clinical course, marked splenomegaly, pancytopenia, and an infiltrate of morphologically distinctive hairy cells in peripheral blood, bone marrow, and spleen. The morphological features of hairy cells are shown in Fig. 4a. Although hairy cell leukemia can often be diagnosed without difficulty on the basis of clinical data and cytology, difficulties arise in some cases, e.g., when the neoplastic cells lack the typical surface projections, or are present in the peripheral blood in low numbers. It is usually even more difficult to identify hairy cells on morphological criteria alone in tissue sections. For these reasons, and since the cell origin of hairy cell leukemia has yet to be identified, many research workers have sought additional diagnostic criteria in hairy cell leukemia. Tartrate-resistant acid phosphatase (Yam et al., 1971) is generally held to be the most valuable of these features. However, this test is not entirely specific, since acid phosphatase in hairy cells is not always tartrate resistant and there are a variety of lymphomas in which tartrate-resistant acid phosphatase can occur, e.g., chronic T cell leukemia (Catovsky and Costello, 1979), prolymphocytic leukemia (Catovsky et al., 1974a; Pallesen et al., 1979; Stein et al., 1981b), S6zary’s syndrome, etc. For a long time it was not clear whether the neoplastic cells in hairy cell leukemia were related to the monocyte/macrophage or the B cell lineage. Although it is now proven that hairy cell leukemia cells are capable of synthesizing immunoglobulins (e.g., Cohen et al., 1979; Rieber et al., 1979; Golde et al., 1977) and show Ig gene rearrangement (Korsmeyer et aZ., 1983) and are thus B cell derived, the normal counterpart of this highly characteristic cell has yet to be identified among the different B cell subpopulations. This situation has prompted several groups to produce monoclonal antibodies against hairy cell leukemia cells. Information is available from two groups (Posnett et al., 1982; Schwarting et al., 1984) on the reactivity of such monoclonal antibodies. None of these antibodies (HC1, HC2; S-HCL1 and S-HCL3)was specific for hairy cells. Nevertheless, these reagents may be of diagnostic value in differentiating hairy cell leukemia from B and T cell lymphomas (see also Table V). Antibody S-HCL3 is particularly valuable in this context (although it also reacts with macrophages) since it gives strong staining of all neoplastic cells in hairy cell leukemia (12 cases tested; Fig. 4b and c) and shows no labeling of neoplastic cells of non-Hodgkin lymphomas of other types (88 cases tested). In keeping with the proven B cell nature of hairy cell leukemia all our cases of hairy cell leukemia have shown strong staining with the B cell-specific antibody To15. Thus, although hairy cell leukemia cannot at present be positively identified by a single monoclonal antibody, it may be diagnosed by the use of two antibodies in combination, i.e., the B cell-specific antibody To15 (or other B cell-specific antibodies such as S-HCLl) and S-HCL3 (Schwarting et al.,

88

HARALD STEIN ET AL.

FIG. 4. Hairy cell leukemia. (a) Typical cytological appearance of the neoplastic cells in a blood smear. Note the abundant pale cytoplasm with a poorly defined margin. A normal lymphocyte (double arrow) and a normoblast (single arrow) are seen on the left. Blood smear stained with May-Grunwald-Giemsa. X 750. (b) Blood smear immunostained with the monoclonal antibody S-HCL3 using the APAAP method. The hairy cells are strongly positive whereas a normal lymphocyte (arrowed) is negative. X750. (c) Frozen section of a liver infiltrated by hairy cell leukemia cells stained with the monoclonal antibody S-HCL3. The leukemic infiltrates in the sinusoids and portal areas are strongly positive and contrast with the unlabeled hepatocytes. An even stronger staining of the leukemic cells was seen with the pan-B cell antibody To15 (not shown). x 160. (d) Frozen section of the same liver as shown in c immunostained with the monoclonal anti-T cell antibody UCHT1. Only scattered T cells are positive: the leukemic cells are unstained. X260.

H U M A N LYMPHOMA

89

1984; Falini et al., 1984). Hairy cell leukemia is the only type of lymphoid neoplasm which we have found to react with both To15 and S-HCL3 simultaneously. Besides these positive immunological features, hairy cell leukemia differs from other non-Hodgkin lymphomas in that it consistently lacks C3b receptors (recognized by TO^), TUl, Leu-1, and TU33 (Stein et al., 1981a; this study, and Falini et al., 1984). The available data thus indicate that hairy cell leukemia possesses a distinctive antigenic profile. Since the antigens recognized by To15 and S-HCL3 are expressed on all the neoplastic cells in any individual case, the diagnostic value of these antibodies for identifying hairy cell leukemia cells is clearly greater than that of tartrate-resistant acid phosphatase. The fact that antibody S-HCL3 also reacts strongly with macrophages again raises the question as to the relationship of hairy cells to cells of the mononuclear phagocyte system. However, a careful evaluation of cytocentrifuged peripheral blood mononuclear cells revealed that antibody S-HCL3 reacts not only with monocytes but also with a small fraction of lymphoid cells. Whether this S-HCLS-positive lymphoid cell population represents the normal equivalent of hairy cell leukemia cells is an interesting possibility which requires further study.

4 . Lympho plamcyticlcytoid Lymphoma (LP lmmunoc ytoma) LP immunocytoma resembles B-CLL in its slowly progressive clinical course, but differs in that it has a much more variable clinical and morphological picture (for clinical details see Heinz et al., 1979). The significant morphological features are shown in Fig. 5a. Our previous investigations have shown that in contrast to both B-CLL (a proliferation of nonsecretory B cells of probable follicular mantle lymphocyte origin) and plasmacytoma (a proliferation of secretory B cells), LP immunocytoma is characterized by a mixed proliferation of both nonsecretory and secretory B cells (Stein et al., 1973b, 1974a, 1975; Lennert et al., 1975; Stein, 1976). The secretory B cells (plasmacytoid cells) can be identified by their content of easily detectable cytoplasmic Ig in formalin-fixed and paraffin-embedded sections. However, the diagnostic importance of this feature is not widely appreciated, e.g., in the Working Formulation LP immunocytoma is not listed as a separate lymphoma type, but instead is included under categories “A, malignant lymphoma, small lymphocytic” and “F, malignant lymphoma, diffuse, mixed, small and large.” One reason why pathologists are reluctant to accept LP immunocytoma as a separate lymphoma type may lie in the fact that it is very difficult on purely morphological grounds to distinguish it from B-CLL. It was therefore of interest to investi-

FIG.5 . Lymphoplasmacytic lymphoma. (a) Typical histological appearance. Note the mixture of lymphocytes and plasmacytic cells (arrowed). Gieinsa. x 750. (I,) Frozen section immunostained for kappa light chains. The lymphoid cells show a ringlike staining whereas the plasmacytic cells, which are preferentially localized in the vicinity of trabeculae, show a strong labeling of the whole cytoplasm. x 100. (c) Frozen section from the same case as shown in b iminunostained for lambda light chains. Only a few scattered cells are positive. x 100.

H U M A N LYMPHOMA

91

gate whether LP immunocytoina exhibits a distinctive antigenic profile detectable with monoclonal antibodies. Table V shows that all the cases of LP immunocytoma studied contained cells with easily detectable cytoplasmic Ig (Fig. 5b and c). This cytoplasmic Ig was restricted to one light chain type in all cases (including the one that showed a weak positive staining for kappa and lambda chains in frozen sections). The evaluation of the other antigenic markers clearly reveals that among the 11 cases there were only 2 with a phenotype identical to B-CLL (surface IgM positive, surface IgD positive, T U l positive, Leu-1 positive). These two cases may represent atypical variants of B-CLL with the capacity for limited differentiation toward plasma cells. Whether or not this interpretation is correct, it is important to note that all other cases of LP immunocytoma differed clearly from B-CLL in lacking Leu-1 and TU33. Four cases of LP immunocytoma also lacked T U l , providing a further indication of its difference from B-CLL. If LP immunocytoma were simply a variant of B-CLL, namely a form of BCLL without a differentiation arrest, one would expect the nonsecretory Blymphocyte population to be reactive with Leu-1 and T U l (like B-CLL cells) and only those cells showing evidence of maturation toward plasma cells to be negative. However, no such heterogeneity of labeling was seen in 9 cases of LP immunocytoma studied, suggesting that the majority of cases of LP iininunocytoma as defined in the Kiel classification are derived from a clearly distinct B cell population. It was also suggested that those LP immunocytoinas that contain centroblast-like and centrocyte-like cells may be derived from germinal center cells (Stein, 1976, 1978; Wright and Isaacson, 1983). However, the present study does not confirm this assumption since all cases of LP immunocytoma, including those with centroblast-like and centrocyte-like cells, proved, in contrast to follicular CB-CC and reactive germinal center cells, to be CALLA negative. A potential cell of origin in LP immunocytoina would be the B cell population found in the pulp, especially that of medullary cords of lymph nodes since this region is a very active site in plasma cell generation. In this context it may be of interest that comparison of the antigenic profile of the Leu-1-negative LP immunocytoma cases with that of other lymphoma types shows the closest resemblance to that of large cell lymphomas, and in particular iminunoblastic lymphomas. Taken together, our data indicate that it is justified to separate at least Leu-1-negative cases of LP immunocytoma from other categories of lymphoma.

5 . Centroblastic-Centrocytic Lymphoinu (CB-CC) Centroblastic-centrocytic lymphomas imitate the reactive germinal center in their follicular growth pattern (Fig. 6a) and cellular composition (i.e., the

92

HAHALD STEIN ET AL.

FIG. 6. Follicular centroblastic-centrocytic lymphoma. (a) Characteristic histological appearance at low power magnification. Giemsa. x36. (b) Histological appearance at higher magnification. The tumor is composed of centroblasts (double arrowed) and centrocytes. Ciemsa. XA50.

presence of both centroblasts and centrocytes; Fig. 6b). A small number of cases show a diffuse growth pattern, but such cases will not be considered here since they have yet to be investigated inimunohistologically. In the Working Formulation CB-CC appears under four different lymphoma types. The historical reason for this is that Rappaport (1966) did not recognize follicular lymphomas as a separate lymphoma entity but rather as a growth pattern which could be encountered in all types of lymphoma. He also did not feel that there was a close relationship between neoplastic nodular proliferation and normal lymphoid follicles. In consequence his classification stated that each of the four cytological types of lymphoma (small lymphocytic, poorly differentiated, mixed lymphocytic-histiocytic, and histiocytic) could occur in a nodular form. This division has been adopted in principle in the Working Formulation but the cytological types have been renamed according to the Lukes terminology. The first evidence suggesting that there was a link between follicular lymphoma and normal germinal centers was provided by Lennert and col-

HUMAN LYMPHOMA

93

leagues. One of us (Lennert, 1961, 1964) defined cytologically two major types of lymphoid cells (now called centroblasts and centrocytes) in the normal germinal center and found that these two cell types are present in the neoplastic follicles of all follicular lymphomas. In consequence the term “centroblastic-centrocytic” lymphoma was coined. In addition, follicular dendritic reticulum cells, a constituent of all normal germinal centers, could be identified by electron microscopy in neoplastic follicles (Lennert and Niedorf, 1969). Immunological studies performed on cell suspensions prepared from lymphoma biopsies and on frozen sections using the EAC adherence assay provided further evidence that follicular lymphomas were derived from germinal center cells. These data are not discussed here since they have been reviewed on numerous occasions elsewhere (e.g., Mann et al., 1979; Stein, 1978; Stein et al., 1981b). However, recent immunohistological studies (Stein et al., 1981a, 1982a) have provided new and more direct evidence that follicular CB-CC lymphoma represents the neoplastic equivalent of follicular hyperplasia. These investigations have been more informative than previous cell suspension studies since they have enabled the topographical relationship of cells to be studied in situ (i.e., to distinguish clearly constituents of the neoplastic follicules from intervening normal tissue) and have also made it possible to study the follicular dendritic reticulum cell population, a cell type which does not readily enter suspension. The results of our immunohistological studies of 22 cases of follicular CBCC presented in Table V showed that this tumor type consists of follicularlike structures filled with neoplastic B cells (Fig. 7), as demonstrated by staining for the B cell antigen To15 and by the restriction of surface Ig to one light chain type (with three exceptions discussed below). The neoplastic B cells present in the follicles differ in their antigenic profile from the cells of all other lymphoma types (see Table V). They constantly expressed CALLA (at low density) and C3b receptors (in varying density and on a varying percentage of the cells) and lacked both Leu-1 (with one exception) and TU33. The Leu-l-positive case was morphologically a borderline case between follicular CB-CC and centroblastic lymphoma. The expression of CALLA was usually sharply limited to the cells of the neoplastic follicules (Fig. 8a). The latter finding has also been reported by other groups (Ritz et al., 1981; Habeshaw et al., 1983). This finding was poorly understood until we found that normal germinal center cells also express CALLA with the same density and distribution (Fig. 8b) (Stein et al., 1982a). In the majority of our cases there was a more or less pronounced lymphocyte mantle zone (corona) around the neoplastic follicules, and the B cells present in these zones were either neoplastic (as evidenced by the restriction of the light chains to the same type as found in the neoplastic

94

HARALD STEIN ET AL.

FIG.7. Frozen section of a follicular centroblastic-centrocytic lymphoma immunostained for IgM. The tumor cells in the neoplastic follicles and some cells in between are stained. X36.

~

follicles), reactive (mosaic-like expression of kappa and lambda chains), or showed a mixed pattern with a preponderance of cells expressing the same light chain type as the neoplastic cells of the follicle. In line with previous studies (Warnke and Levy, 1978) the neoplastic B cells expressed surface Ig. Follicular CB-CC proved to be the only B cell lymphoma in which the surface Ig was as frequently IgG as IgM. For reasons that are not entirely clear the demonstration of gamma chains on the neoplastic follicular cells was often only convincingly possible using monoclonal anti-IgG antibodies, and a number of cases found in a previous study (Stein et al., 1981a) (using polyclonal antibodies) to be IgG negative turned out to be positive when studied with monoclonal anti-IgG antibodies (Stein et al., 1982a, and this study). IgD was present on the neoplastic cells in 10 of our 22 cases. The expression of IgD on neoplastic germinal center cells is in contrast to reactive germinal center cells in normal lymphoid tissue which are constantly surface IgD negative. However, we have observed three cases of lymphadenitis with very prominent follicular hyperplasia, in which the majority of the reactive germinal center cells expressed variable amounts of surface IgD. In most surface IgD-positive biopsies, the distribution pattern of surface IgD was different from that of surface IgM and the corresponding light

H U M A N LYMPHOMA

95

FIG. 8. Immunostaining for common ALL antigen (CALLA). (a) Follicular centroblasticcentrocytic lymphoma showing labeling of the neoplastic follicles. X50. (b) A reactive B cell follicle. Staining is confined to the germinal center. In contrast to the neoplastic follicles seen in (a), there is some diffuse intracellular staining in the reactive germinal center. This is due to a staining of follicular dendritic reticulum cells (FDRC) for CALLA as revealed by imrnunostaining of isolated FDRC. The FDRC present in follicles of centroblastic-centrocytic lymphoma appear to be unreactive for CALLA. Therefore staining for CALLA in centroblastic-centrocytic lymphoma produces a sharp ring-like labeling of the neoplastic follicle cells (as seen in a). F M , Follicular mantle. X 76.

96

HAKALD STEIN ET AL.

chain. The surface IgD staining tended to be weak on the tumor cells at the central area of the follicles and stronger on the tumor cells at the rim of the follicles or in the follicular mantle-like zones. The immunohistological demonstration of normal “bystander cells” in follicular CB-CC underlines the very striking similarity between the architecture found in follicular CB-CC and follicular hyperplasia. In 22 cases of follicular CB-CC a sharply defined meshwork of FDRCs was demonstrated within the neoplastic follicles by the anti-FDRC antibody R4/23 (Fig. 9a and b) or anti-C3b receptor antibody To5, closely resembling that present in reactive germinal centers (Fig. 9c and d). This pattern of cellular arrangement is thus pathognomic for follicular CB-CC. In nearly all cases the staining of the FDRC processes for C3b receptor (antibody To5) was considerably stronger than that of the neoplastic B cells. This finding makes it necessary to reinterpret the results of complement receptor analysis by the EAC-adherence test, a technique in which the binding of complement-coated erythrocytes (EAC) to frozen sections is evaluated. In this test EAC bind predominantly to neoplastic follicles in follicular lymphomas and to germinal centers in follicular hyperplasia (Shevach et al., 1973; Stein et al., 1978). This was interpreted as indicating that neoplastic cells in follicular lymphomas and normal germinal center cells express C3 receptors. However, the immunohistological findings described above (i.e., labeling with monoclonal anti-C3b receptor) suggest that EAC adherence preferentially detects C 3 receptors present at high density on follicular dendritic reticulum cells. In order to confirm this assumption experimentally we processed tonsil cell suspensions under conditions that do not allow follicular dendritic reticulum cells to be recovered. More than 50% of the lymphoid cells were B cells capable of rosetting with EAC in suspension. A cell pellet was prepared and snap frozen, and cryostat sections prepared. No binding of EAC to these sections was observed, under conditions which allow strong labeling of germinal centers in normal tonsil sections. It was hence concluded that EAC binding to germinal centers does indeed represent C3b receptor expression on FDRC rather than on lymphoid cells. A second feature constantly associated with follicular CB-CC is the presence around the neoplastic follicles of a more or less well-developed T zone. The number of T cells present in these areas varies from cases to case. Blood vessels and sometimes interdigitating reticulum cells (identifiable with the monoclonal antibody T-ALL2 and NA1/34, Stein et al., 1981a, 1982c), as well as varying numbers of B cells, are also present in the T zone. The distribution of T cells of helper and suppressor phenotypes was very similar to that found in follicular hyperplasia, with the exception that suppressor T cells were present more frequently within neoplastic follicles than within normal follicles, and that the ratio of T4 to T8 cells (helpersuppressor)

FIG.9. Immunostaining for follicular dendritic reticulum cells (FDRC) with monoclonal antibody R4/23. (a) Follicular centroblastic-centrocytic lymphoma. In this case the monoclonal antibody R4/23 reveals tightly packed FDRC within the neoplastic follicles. x36. (b) Another case of centroblastic-centrocytic lymphoma. Here the antibody R4/23 demonstrates a less dense meshwork of FDRC in the neoplastic follicles. x56. (c) Normal tonsil at low magnification. The monoclonal antibody R4/23 reveals a dense meshwork of FDRC within the B cell follicles. X36. (d) Normal tonsil at a higher magnification. The dense meshwork of FDRC demonstrated by the monoclonal antibody R4/23 is clearly seen in the germinal center and in the follicular mantle. x 100.

98

HARALD STEIN ET AL.

showed a shift toward an excess of T8 cells. Dvoretsky et al. (1982) reported that the percentage of T cells in neoplastic follicles (20%)was virtually identical to that seen in nonneoplastic follicles of stimulated lymph nodes or tonsils but that, in contrast to nonneoplastic follicles, 67% of these Leu-l-positive T cells lacked the T4 antigen. However, no other T cell reactive monoclonal antibodies were used in this study and double staining for Leu-1 and T cell antigens was not performed, so that the significance of this finding remains to be clarified. The immunohistological similarities between neoplastic and reactive follicles are not confined to B cells, FDRC, and T cells. It is also striking that the distribution pattern of Leu-7-positive cells (putative natural killer cells), and of macrophages and reticulin-associated antigens in neoplastic follicles mimics that of reactive germinal centers. It should be added that in 3 of our 22 cases of follicular CB-CC the individual B cells of the neoplastic follicles seem to stain for both kappa and lambda chains. This phenomenon did not appear to be due to a technical artifact (since it was repeatable on numerous occasions) and the possibility that these cases represented follicular hyperplasia misdiagnosed as follicular CB/CC lymphoma was also ruled out. The same finding has also been observed by other investigators (e.g., Habeshaw et al., 197913, and personal communication). We have at present no explanation for this unexpected finding. Taken together, our immunohistological data clearly indicate that follicular lymphoma of centroblastic-centrocytic type, as defined in the Kiel classification, is a separate biological entity.

6 . Centrocytic Lymphoma Centrocytic lymphoma is defined as a lymphoma consisting solely of germinal center cells of centrocytic type (see for morphological details Fig. 10a). In classifications other than the Kiel scheme this type of lymphoma is not treated as a separate entity. In the Lukes-Collins classification (Lukes and Collins, 1974, 1975), for example, it has been included with follicular center cell lymphoma, small and large cleaved, which is largely equivalent to CBCC in the Kiel classification. This alternative view of the classification of CC reflects the fact that the neoplastic cells in CC show a morphological resemblance to those seen in CB-CC. Indeed there is a further similarity between the two entities, i.e., the presence of a meshwork of FDRC [detectable with antibodies R4/23 and anti-C3b receptor antibody (Stein et al., 1981a, 1982a)l in 16 of our 18 centrocytic lymphoma cases (Table V and Fig. l l a and b). However, on more detailed immunohistological analysis, it becomes apparent that there are many differences between centrocytic lymphoma and follicular CB-CC.

Fic;. 10. Centrocytic lymphoma. (a) Typical histological appearance. Note the irregular shape of the nuclei of the neoplastic cells. Gieinsa. x800. (b) Frozen section immunostained for IgM. Nearly all cells are positive. The labeling reaction for lambda light chains produced a similar staining whereas the reaction for kappa light chains was negative on the tumor cells. x 100. (c) Frozen section of the same case as shown in 11 iminnnostained with the monoclonal antibody OKT11. The tumor cells are negative. Among the tumor cells are scattered strongly stained T cells. x50.

100

HARALD STEIN ET AL.

FIG. 11. Iinmunostaining for folliciilar dendritic reticulum cells (FDRC) in two cases of centrocytic lymphoma using inonocloiial antibody R4/23. Case 1 (a) shows a diffuse meshwork of FDRC whereas in case 2 (I)) the FDRC meshwork is nodular. Note that the FDRC meshwork in the latter case is not as well demarcated as in follicular centroblastic-centrocytic lymphoma. (a) x20. (b) x36.

In contrast to follicular CB-CC, the heavy chain class expressed in centrocytic lymphoma was always IgM, the only exception being a case which expressed IgD together with IgG. Interestingly, this case was the only one without any light chain expression. In five of the IgD-positive cases IgD expression was very weak. The ratio of kappa-positive to lambda-positive cases (5:12) differed from the pattern seen in other B cell lymphomas (in which kappa-positive cases predominate) except prolymphocytic leukemia. A similar finding has been reported by Habeshaw et al. (1983). The light chain staining of nearly all cells in centrocytic lymphoma was restricted to one light chain indicating that centrocytic lymphoma does not usually contain significant numbers of nonmalignant cells (as are frequently found in follicular CB-CC). One of the most important differences from follicular CB-CC concerns the antigens CALLA, Leu-1, T U l , and TU33. Leu-1 was present in 16 of 18 cases of centrocytic lymphoma, whereas it was detectable in only 1 of the 22 cases of follicular CB-CC. As mentioned above the single Leu-l-positive follicular CB-CC case was morphologically on the borderline with cen-

HUMAN LYMPHOMA

101

troblastic lymphoma. TU33 was strongly expressed on the tumor cells of 13 of 18 cases of centrocytic lymphoma and was missing in all cases of follicular CB-CC. T U l was detectable in none of the centrocytic lymphoma cases, but was present in 16 of the 22 cases of follicular CB-CC. Furthermore, all centrocytic lymphoma cases with typical morphology (16 of 18) lacked follicular mantle-like zones and T zones and contained only a relatively small number of T cells among the neoplastic cells (Fig. 1Oc). Two cases were not quite typical for centrocytic lymphomas in morphology since they exhibited a partially follicular growth pattern and contained a few larger cells. These two cases were thus morphologically borderline cases between centrocytic lymphoma and follicular CB-CC; however, the antigen profile of these two cases corresponded fully with that of typical centrocytic lymphomas. These two cases prompted us to compare the relationship between the distribution of FDRC and the growth pattern in our cases of centrocytic lymphoma. As expected it was noted that the more nodular the FDRC meshwork pattern the more nodular the morphological growth pattern. In this context it is of interest that the two cases of centrocytic lymphoma that contained only a few FDRC (in the form of small foci) were negative for Leu-1 and TU33. However, the morphology of these two cases was typical of centrocytic lymphoma, showing that the antigen profile (TUl negative, Leu-1 positive) and the presence of a FDRC meshwork is not absolutely correlated with centrocytic-type morphology. Whether these two Leu-lnegative (and TUSSnegative) and FDRC meshwork-negative cases represent a different lymphoma type or only a variant of typical centrocytic lymphoma remains to be shown. The frequency of Leu-1-negative lymphomas of centrocyte-type morphology may vary from country to country. Recently Pallesen (personal communication) analyzed six lymphomas with typical centrocytic lymphoma morphology and found four of them to be negative for Leu-1 and TU33. A comparison of our data with that of other authors is difficult if not impossible, because of the use of a different classification scheme (GajlPeczalska et al., 1982), the use of cell suspensions rather than tissue sections, or the use of immunoperoxidase detection systems which are of relatively low sensitivity. In conclusion our immunological data support the view that centrocytic lymphoma is a separate entity, clearly distinguishable in most cases from other lymphoma categories. Cellular Origin of Centrocytic Lymphoma. The normal cellular counterpart of the neoplastic cell is not as evident in centrocytic lymphoma as it is in the case of follicular CB-CC (see above). Arguments in favor of a derivation of centrocytic lymphoma cells from germinal center cells are (1) the mor-

102

HARALD STEIN ET AL.

TABLE VIII EVIDENCE O F T w o TYPES OF CENTROCYTES Centrocyte Type 1 Occurring as a major population in normal germinal centers Present in follicular CB-CC Present in centrocytic lymphoma B cell antigen To15 SIgM or SIgC Heavy chain class restricted to SIgM CALLA TU 1 C3b receptor (C3RTo5) Leu-1 TU33

+ + -

+ + + +/+

Centrocyte Type I1

+ + -

+ +/-

+

+/-

phological similarity between the two cell types, and (2) the presence of the meshwork of FDRC among the centrocytic lymphoma cells. The discrepancies lie in the findings that (1)centrocytic lymphoma cells constantly express Leu-1 and in a majority of cases TU33 (antigens which in reactive tissue are exhibited by T cells but by few if any centrocytic cells); (2) centrocytic lymphoma cells are devoid of CALLA (present on normal centroblasts and centrocytes); and finally (3) centrocytic lymphoma cells frequently express delta chains (which are absent from reactive centrocytes). Centrocytic lympxoma cells could be obviously derived from follicular mantle zone cells but the differences in morphology and the constant absence of T U l from centrocytic lymphoma cells argues against such a relationship (the T U l antigen being constantly expressed on follicular mantle lymphocytes). These differences in phenotype between centrocytic lymphoma cells and normal B cells may be explained either by changes in antigen expression associated with malignant transformation ‘or by derivation from an as yet unidentified B cell subset. We favor the latter alternative (Stein et al., 1982a), and suggest, for reasons set out in Table VIII, that there are two types of centrocytes, and that centrocytic lymphoma derives from “Type 11” centrocytes.

7 . Centroblastic Lymphoma and Immunoblastic Lymphoma Centroblastic lymphoma and immunoblastic lymphoma will be discussed together. The “classical” morphological features of these two disorders, as defined in selected cases, are clearcut and easily distinguishable (see Figs. 12 and 13). However, in individual cases it is often difficult or impossible to distinguish between them on histological grounds. The reason for this dificulty lies in the fact that a significant number of large cell lymphomas are

H U M A N LYMPHOMA

103

F I G 12. Typical histological appearance of centrohlastic lymphoma. The tumor cells usually contain two or three nucleoli which tend to be located at or close to the nuclear membrane. Cytoplasm is sparse and basophilic. Gienisa. x800. FIG. 13. Typical histological appearance of imtnunoblastic lympliotna. The tumor cells usually contain only one nucleolus which tends to be found at the center of the nucleus. The cytoplasm is mediiini broad to abundant and basophilic. Giemsa. X800.

104

HARALD STEIN E T AL.

composed of a mixed population of cells some of which resemble centroblasts, and others immunoblasts. These “mixed cases” are included in the Kiel scheme (Lennert and Mohri, 1978) under centroblastic lymphoma and designated “polymorphic centroblastic lymphoma. ” This distinguishes them from “monomorphic centroblastic lymphoma” in which there is a predominance of cells with a centroblast-like morphology. Monomorphic centroblastic lymphoma seems to be roughly equivalent to “group G, malignant lymphoma, diffuse, large cell, noncleaved” of the Working Formulation; however, it is evident that centroblastic lymphoma of the Kiel classification and “malignant lymphoma, diffuse, large cell, noncleaved cell” of the Working Formulation are not completely congruent since centroblastic lymphoma comprises cases with diffise and follicular growth patterns whereas “malignant lymphoma, diffuse, large cell, noncleaved cell” does not. Furthermore, we would stress that immunohistological phenotyping shows that some of the “malignant lymphoma, diffise, large cell, noncleaved cell” turn out to be of T cell type. This may partly be a reflection of the widespread preference for hematoxylin and eosin staining: Giemsa staining reveals finer morphological features and makes it possible to evaluate the structure not only of nuclei but also of the cytoplasm. Immunoblastic lymphoma corresponds to “group H, malignant lymphoma, large cell, immunoblastic, including plasmacytoid” in the Working Formulation. The difficulty in morphologically differentiating centroblastic lymphoma from immunoblastic lymphoma has caused a great deal of confusion, leading a number of lymphoma experts to suggest that the two categories should be included in a single group of “large cell lymphomas.” However, there is evidence from follow-up studies that centroblastic lymphoma and immunoblastic lymphoma differ in survival times and response to treatment (Meusers et al., 1979). The studies of Strauchen et al. (1978) can also be taken as evidence in this direction, although a different classification scheme was used. Given this disagreement concerning the relationship between the two types of lymphoma, it is obviously pertinent to investigate whether immunohistological phenotyping with a large panel of monoclonal antibodies can enable centroblastic lymphoma to be clearly separated from immunoblastic lymphoma. The results of our approach are given in Table V. We included in our series only those cases of centroblastic lymphoma which were monomorphic or showed a minimal degree of cellular polymorphism. This immunohistological study only revealed minor differences between centroblastic lymphoma and immunoblastic lymphoma. Centroblastic lymphoma was more often surface Ig negative and more frequently expressed

H U M A N LYMPHOMA

105

CALLA and T U l . Centroblastic lymphoma was associated in some cases with a more or less well-developed FDRC meshwork, whereas this was not found in any of the immunoblastic lymphoma cases. The presence of the FDRC meshwork in centroblastic lymphoma was associated in six cases with a follicular or vaguely follicular architecture. However, apart from these minor (and possibly fortuitous) differences, the antigenic profiles of centroblastic lymphoma and immunoblastic lymphoma had much in common. In both types there were cases which lacked HLA-DR, B cell antigens, or surface immunoglobulin. In no cases were B cell antigens and surface Ig simultaneously lacking on the tumor cells. In each morphological lymphoma type under discussion there was one case that lacked both B cell antigens, i.e., To15 and F8-11-13; these two cases also lacked HLA-DR but expressed surface immunoglobulin (the centroblastic lymphoma case IgM lambda and the immunoblastic case IgA kappa). The case-to-case variation in phenotype among both centroblastic and immunoblastic lymphomas makes it difficult to interpret the data obtained. One possibility is that centroblastic lymphoma and immunoblastic lymphoma are not fundamently distinct entities and that differences in morphology and antigenic profile reflect only the type of intraclonal diversity which occurs in all types of tumors (e.g., squamous cell carcinoma of the skin, chronic myeloid leukemia, etc.). The alternative view is that the minor differences seen are indeed a reflection of true differences in cellular origin but that currently available monoclonal antibodies are not adequate to clearly establish this distinction. It may be added that centroblastic lymphoma and immunoblastic lymphoma vary greatly in the number of infiltrating host cells (e.g., macrophages and T cells). In some cases the admixture of macrophages can be so high that there is a risk of misinterpreting them as being neoplastic. This may partly account for the frequency with which histiocytic tumors appeared in lymphoma classifications in the past. Warnke et al. (1980)investigated a series of “large cell lymphomas of nonT type” and identified two phenotypically different tumor types: (1) HLADR positive and surface immunoglobulin positive and (2) HLA-DR positive and surface Ig negative (and T cell antigen negative). They interpreted the second group as probably being of B cell nature. They stressed the importance of distinguishing these two groups since the surface Ig-negative group had a better prognosis and responded better to therapy than did the other group. In this context it may be important to note that eight lymphoma cases with an immunoblastic lymphoma-like morphology which we studied did not fit into either the B immunoblastic lymphoma or T immunoblastic (see below) lymphoma group. These eight cases expressed HLA-DR and lacked surface Ig and T cell antigens and thus resemble the cases reported by

106

HARALD STEIN ET AL.

Warnke et al. (1980). All eight cases proved to be positive for the SternbergReed cell-associated antigen Ki-1, suggesting that they are more closely related to Hodgkin’s disease than to B or T cell neoplasms. More details of these lymphomas will be published.

8 . Multilobated Lymphomas of B Cell Type, a Variant of Centroblastic Lymphoma? Our series of non-Hodgkin lymphomas contained two cases characterized by the presence of many cells with multilobated nuclei very similar to those described by Pinkus et al. (1979) and Weinberg and Pinkus (1981) as a variant of T cell lymphoma. Immunohistological phenotyping revealed that both our cases were of B cell nature (B cell antigen To15 positive and surface Ig positive), like the case reported by Pileri et al. (1982), indicating that multilobated nuclei are not restricted to lymphomas of T cell derivation. The morphological and antigenic phenotype of these two cases was most similar to that of normal and neoplastic centroblasts, leading to the view that these cases are probably a multilobated variant of centroblastic lymphoma.

9. B Cell Lymphomas with the Morphology of True Histiocytic Sarcoma-a New Lymphoma Entity? Among the 200 lymphoid neoplasms studied there were three that we and other histopathologists classified on morphological grounds as typical cases of true histiocytic sarcoma. The histological features of these cases are shown in Fig. 14a-c. To our surprise, the immunohistological studies revealed a clearcut B cell antigenic profile (see Table IX). All three cases were positive for the B cell antigen To15 and two for the B cell antigen detected by antibody F8-11-13. The tumor cells showed a strong staining for one light chain type in two cases and a very weak staining in one case. In Case 1 the heavy chain staining was not interpretable because of high level of background staining, but in the other two cases a proportion of the tumor cells were clearly positive (although only weakly) for one heavy chain class. All the macrophage-associated markers studied were absent except OKM 1 (which represents the C3bi receptor) in Case 2. However, it has already been shown that O K M l can be expressed by some tumors of undoubted B cell nature, e.g., hairy cell leukemia (Jansen et al., 1981). Since all three cases showed a very similar morphological and immunological phenotype (except T U l and OKMl expression in Case 2) the question arises as to whether the three cases represent a new, as yet unidentified, type of B cell lymphoma. The existence of these three unusual cases also emphasizes that B cell lymphomas encompass a very broad morphological

FIG. 14. Three large cell lymphomas with the morphological appearance usually considered typical of true histiocytic sarcomas. In fact each of these tumors showed a B cell phenotype on iminunohistological analysis (see Table IX). All three cases have in common highly polymorphic nuclei and abundant pale-staining cytoplasm. (a) H and E-stained paraffin section. x 800. (b) H and E-stained plastic section. x 800. (c) Giemsa-stained paraffin section. X800.

108

HARALD STEIN ET AL.

TABLE IX ANTIGENPROFILE O F TIIREE CASESMORPHOLOGICALLY DIAGNOSED AS TRUEHISTIOCVTIC SARCOMA Case 1

K

A

+

CALLA TU 1 C3bR (C3RTo5) Leu-1 TlULyt 3 UCHTl OKMl Antiinonocyte 1 Antimonocyte 2 Lysozyme aL-Antitrypsin

-

(1

b

n.d.

Case 2

Case 3

n.d.

-

n.i., not interpretable. ii.d., not done.

spectrum, and include morphological features that are generally thought to be specific for macrophages. In this context our immunohistological findings in seven cases of lymph node tumors, collected by Van der Valk and colleagues, may be of interest. Five of the seven neoplasms were categorized as malignant histiocytosis but only one reacted with antibodies against macrophage-associated markers, the other four cases expressing the B cell antigen To15. The tumor cells of the remaining two cases were thought to be derived from follicular dendritic reticulum cells (Van der Valk et aZ., 1982b), but also proved to be To15 positive and to lack macrophage reactive markers. They were also nonreactive with the FDRC specific antibody (R4/23). This finding excludes a close relationship of the tumor cells of these two cases to FDRC and proves their B cell nature.

10. Lyniphoblastic Lymphomas (LB) Lymphoblastic lymphoma are morphologically defined as a group of lymphoid neoplasms that have in common the proliferation of predominantly

H U M A N LYMPHOMA

109

FIG.15. Lyinphoblastic lymphoma of Burkitt-type. (a) Histological appearance at low power. Note the numerous “starry sky” macrophages scattered among the tumor cells. Giemsa. ~ 7 5 . (b) Histological appearance at high power. Three macrophages, containing nuclear debris, are seen. Giemsa. X800. (c) Frozen section immunostained with a pan-B cell antibody To15. The tumor cells (lower two-thirds of the picture) are positive as well as the B cells in the uninvolved adjacent lymphoid tissue (upper one-third). The starry sky macrophages are unstained. ~ 7 5(d) . Sequential frozen section immunostained with the anti-C3b receptor antibody C3RTo5. The starry sky macrophages among the tumor cells and the B cells in the uninvolved adjacent lymphoid tissue (upper fifth of the picture) are strongly labeled while the tumor cells are unstained. x 100.

110

HARALD STEIN ET AL.

medium sized “blast cells” with scanty, more or less basophilic cytoplasm (see Figs. 15, 16, and 18).All but two of the lymphomas in our series which were classified morphologically as lymphoblastic lymphomas expressed the antigen profile of B cells of an early differentiation stage (for details see below). This profile is clearly different from that of the B cell neoplasms discussed above, all of which derive from B cells at later differentiation stages. This finding justifies the inclusion of lymphoblastic lymphomas in a single group, separated from other types of non-Hodgkin lymphoma. It should be noted, however, that lymphoblastic lymphoma is neither morphologically nor immunologically homogeneous. The first subgroup of lymphoblastic lymphoma to be identified (on the basis of its morphological and epidemiological features) was Burkitt’s lymphoma (Fig. 15b). A second subtype was separated on the basis of the focal expression of acid phosphatase reactivity (Leder, 1965; Catovsky et al., 1974b; Lennert et al., 1975; Stein et al., 1976) and later on the presence of convoluted nuclei in at least some of the tumor cells (Lukes and Collins, 1974; Barcos and Lukes, 1975). There have subsequently been attempts to classify the remainder of the lymphoblastic lymphomas and to relate them to a cell of origin. In Kiel two categories, “lymphoblastic lymphoma, B non-Burkitt” and “lymphoblastic lymphoma, unclassified” were distinguished. “Lymphoblastic lymphoma, B non-Burkitt” shares with the lymphoblastic lymphoma of Burkitt type strong cytoplasmic basophilia and a cohesive growth pattern (Fig. 16a) but lacks other characteristics associated with Burkitt’s lymphoma (e.g., the presence of “starry sky” macrophages). “Lymphoblastic lymphoma, unclassified” (Fig. 16b) comprises those lymphoblastic lymphomas which cannot be related on the basis of enzyme cytochemistry and/or morphology to the B cell or T cell lineages. a. Correlation of Morphologically Defined Subtypes (LB Burkitt type, LB-B non-Burkitt Type, LB Unclass$ed Type) with Antigen Profile. To check the validity of morphological/enzyme cytochemical subtyping of lymphoblastic lymphoma, we correlated these morphological subtypes with their antigenic profiles and vice versa. The results are given in Tables X and

XI. The immunological phenotyping of nine cases of “lymphoblastic lymphoma, Burkitt type” confirmed that this type consistently shows a highly constant immunophenotype characterized by the expression of B cell antigen To15 (Fig. 15c), surface IgM, CALLA, and the absence of T U l , C3b receptors (Fig. 15d), Leu-1, and TU33. Although there were only seven cases available it is of interest that none of the cases of “lymphoblastic lymphoma, B non-Burkitt” was of T cell origin. In one case the phenotype was consistent with the immunophenotype of lymphoblastic lymphoma of Burkitt type (“Burkitt-like”) (see below); the

H U M A N LYMPHOMA

111

FIG.16. Typical histological appearance of lyinphoblastic lymphoma of B non-Burkitt type (a) and of unclassified type (I)). Giemsa. ~ 8 0 0 .

TABLE X IMMUNOLOGICAL TYPINGOF LYMPHOBLASTIC LYMPHOMA" Morphological type

Number of cases

HLA-DR TU35

Pan B To15

SIg

CALLA VIL-A1

T assoc.Ag

TU14

Leu-1

Immunophenotype ~

Burkitt type B-non-Burkitt type

+" + +

9-9

7-Ek +

+

Unclassified type

10-i

Convolutedlacid phosphatase

20

-

20

+ ;-) -1

+

+

+ + + + +

+

+

-

+ +

+ + +

-

-

-

-

-

-

+/-

+ +

-

-

-

-

-

-

-I+

-

-

-I+

+ +

~~

Burkitt-like Burkitt-like Pre-B cell Unclassified (2 more mature B cell)

+I(-)

Burkitt-like Pre-B cell Prethymic or thymic

+4-)

Prethymic or thymic

a +, All cases positive; +/(-), occasional negative cases; +I-, majority of cases positive; -I+, majority of cases negative; -I(+), occasional positive cases; -, all cases negative.

TABLE XI CORRELATION OF THE IMMUNOPHENOTYPES OF LYMPHOBLASTIC LYMPHOMAS WITH THEIRMORPHOLOGICALSUBTYPES (KIEL CLASSIFICATION)

HLA-DR

Pan B To15

SIg

CALLA VIL-A1

T cell Ag TU 14

Leu-1

Burkitt-like

+a

+

+

+

-

-

Pre B cell

+

+

-

--/(+)

-

-

Immunophenotype c

I

W

Prethymic and thymic

Number of cases

Burkitt B non-Burkitt Unclassified

’< -I+

+

Morphological type of lymphoblastic lymphoma

+I(-)26 <2:

B non-Burkitt Unclassified Unclassified Convoluted/acid phosphatase

a +, All cases positive; +I(-), occasional negative cases; -/+, majority of cases negative; -/(+), occasional positive cases; -, all cases negative.

114

HARALD STEIN ET AL.

remaining four cases were of pre-B cell type (defined as B cell antigen To15 positive, surface Ig negative, CALLA positive). Originally two additional cases had been classified morphologically in the “lymphoblastic lymphoma, B non-Burkitt” group; however, their antigenic phenotype did not fit into the Burkitt-like type or pre-B cell type because of the absence of CALLA and presence of Leu-1 in one case. Reevaluation of the morphology of these cases with a knowledge of their antigenic phenotype led to the conclusion that both cases are probably related to germinal center cell lymphoma of centroblas tic type. The immunological typing of 10 cases of “lymphoblastic lymphoma, unclassified type” revealed that more than half of the cases expressed a T cell phenotype, three a pre-B cell phenotype, and one a Burkitt-like phenotype. The 20 cases of lymphoblastic lymphoma of T type diagnosed on the basis of the presence of cells with convoluted nuclei and of pronounced focal phosphatase acitvity (lymphoblastic lymphoma of convoluted/acid phosphatase type) all presented an antigen profile consistent with T cell origin. Taking the data of Tables X and XI together we can conclude the following.

1. Using the Pan B cell antibody To15 and T cell antibody TUl4 all cases of lyinphoblastic lymphoma/leukemia could be identified as B or T cell derived. This means that the former classification of lymphoblastic lymphoma/leukemia into three types, namely non-B non-T (CALLA type), B cell type, and T cell type is no longer valid. All cases of non-B non-T could be classified as either B cell or T cell related. 2. The immunophenotype “Burkitt-like” is associated, in more than 80% of the cases, with Burkitt type morphology, but may also rarely be seen in cases of lymphoblastic lymphoma, B non-Burkitt,” and “lymphoblastic lymphoma, unclassified” types (see Table XI). 3. The immunologic pre-B cell phenotype is associated with the inorphology of either “lymphoblastic lymphoma, B non-Burkitt” type, or “lymphoblastic lymphoma, unclassified,” type, but not with “lymphoblastic lymphoma convolutedlacid phosphatase” type. 4. On purely morphological grounds it is not possible to distinguish between pre-B cell and Burkitt-like B immunophenotype except that the latter is usually associated with Burkitt type morphology. 5 . “Convoluted nuclei” and “pronounced focal acid phosphatase reactivity” are reliable criteria for identifying T cell lyinphoblastic lymphomas if used in conjunction. However, it should be emphasized that their assessment requires some experience and that they allowed only two-thirds of the T cell lymphoblastic lymphomas to be separated from lyinphoblastic lymphoma of pre-B cell or Burkitt-like immunophenotype. From these findings it appears that subtyping of lymphoblastic lymphoma

HUMAN LYMPHOMA

115

(with the exception of Burkitt’s lymphoma) should be based on antigenic profile and enzyme cytochemical criteria rather than on morphology alone. b. Lymphoblastic Lymphoma of Burkitt Type (LB Burkitt Type). Lymphoblastic lymphoma, Burkitt type includes both endemic (true) Burkitt’s lymphomas and lyinphoblastic lymphomas with an identical morphology occurring sporadically in nonendeinic areas including the United States and Europe. The morphological features are illustrated in Fig. 15a and b. Cell surface studies performed on cell suspensions have already revealed that the immunological properties of Burkitt type lymphoblastic lymphoma are highly homogeneous (e.g., Stein et a l . , 1981b) and this finding is confirmed by labeling with monoclonal antibodies. The antigen profile (see Table V) is characterized by a constant expression of HLA-DK, B cell antigen To15, F8-11-13 antigen, surface immunoglobulin, strong expression of CALLA, and the constant lack of all other markers. The constant expression of CALLA (Ritz et al., 1981) and the absence of Leu-1 is in agreement with other studies (Gajl-Peczalska et a l . , 1982). In the past there has been no consensus as to whether Burkitt type lymphoma cells can express surface IgD in addition to surface IgM: our data confirmed that in a minority of cases surface IgD is indeed expressed. Cellular origin. Although the morphological and immunological phenotype of LB Burkitt type is well defined, and although it is certain that this tumor is derived froin B cells, it is not yet clear to which normal B cell differentiation stage the LB Burkitt type cell is related. The observed tendency of LB Burkitt type cells to home to gerininal centers suggests that the tumor cells are derived from small noricleaved germinal center cells (Mann et al., 1976). However, Kerndrup and Pallesen (1981) reported that in LB Burkitt type an infiltration of the germinal centers is preceded by infiltration of the mantle zones. Furthermore, the expression of large amounts of CALLA suggests that the tumor cells may be derived from B cells at a relatively early differentiation stage. Because of the lack of markers specific for precursor cells of germinal center and mantle zone cells, hypotheses concerning the cell of origin of Burkitt cells cannot at present be proven or disproven. c. Lymphoblastic Lymphomas with a “Burkitt-like” Immunophenotype. The “Burkitt-like” immunophenotype is HLA-DR positive, surface Ig positive, CALLA positive, Leu-1 negative, T U l negative, and TU33 negative (the last two not shown in the tables). As mentioned above and shown in Table XI all but 2 of the 11 cases with this iminunophenotype had the morphological features of lyinphoblastic lymphoma of Burkitt type. These two cases were indistiiiguishable in morphology from lymphoblastic lymphomas showing a pre-B cell phenotype. This raises the question of how such cases will behave clinically, i.e., in keeping with their phenotype as exam-

116

HAKALD STEIN ET AL.

FK:. 17. Lymphoblastic lymphoma of unclassified type. (a) Frozen section immunostained with monoclonal anti-HLA-DR antibody TU35. The tumor cells are strongly positive as are the interdigitating reticulum cells in the uninvolved area (upper right corner); the T cells in the uninvolved area are unstained. X75. (b) Adjacent frozen section immunostained for CALLA with monoclonal antibody VIL-A1. The tumor cells are strongly labeled whereas all cells in the uninvolved area (right upper corner) are negative. x 150.

ples of Burkitt lymphoma, or according to their morphological features as common ALL. This question cannot at present be answered but it is under investigation. d . Lymphoblastic Lymphoinus with a “Pre-B Cell” Immunophenotype. Seven of our cases of lymphoblastic lyinphoina were HLA-DR positive (Fig. 17a), surface Ig negative, CALLA positive (Fig. 17b), and T cell antigen negative. This pattern corresponds to the “null” or “common” ALL phenotype as described by Greaves et al. (1975, 1981) and many others. Vogler et al. (1978), Greaves et al. (1979), Brouet et al. (1979), and others reported that approximately 20-40% of the ALL of this phenotype contain small amounts of cytoplasmic mu chains, unaccompanied by light chains, and are thus related to the earliest recognized stage in the B cell lineage (the pre-B cell). In frozen and routinely processed paraEin-embedded sections it is not possible to detect small amounts of cytoplasmic mu chains and we were therefore not able to check whether our five cases expressed the classical phenotype of pre-B cells. However, the fact that the cells of all five lympho-

H U M A N LYMPHOMA

117

blastic lymphoma cases reacted strongly with the B cell-specific antibody To15 indicates their close relationship to the B cell lineage and thus relates them to a pre-B cell (HLA-DR positive, surface Ig negative, cytoplasmic mu chain positive, CALLA positive, B cell antigen positive) or a very early pre-B cell (HLA-DR positive, surface Ig negative, cytoplasmic mu chain negative, CALLA positive, and B cell antigen positive) differentiation stage. Studies performed on a larger number of blood and bone marrow samples (unpublished results) have also revealed that the great majority of ALL cases exhibiting “null” or “common” ALL phenotype express To15, confirming that most, if not all, cases of lymphoblastic lymphoma/leukemia which lack T cell markers are derived from B cell precursors. Recent studies performed with other B cell reactive monoclonal antibodies (Nadler et al., 1981; Greaves et al., 1981) as well as studies on rearrangement of C mu region genes (Korsmeyer et al., 1981) support this.

V. T Cell Lymphomas

The study of lymphoma cell surface markers has allowed T cell-derived tumors to be objectively distinguished from B cell neoplasms and has enabled immunophenotype to be correlated with morphology. It was found that T cell neoplasms often exhibit very irregular nuclear configurations, variously described as “cerebriform,” “convoluted,” and “multilobated,” or ,, as resembling “maple leaves, walnut seeds,” “jelly fish,” and “embryos.” However, until now the existence of distinct clinico-pathological subtypes of T cell lymphoma, based on these peculiar nuclear configurations, has not been clearly established. It is obviously of interest to relate the morphology of T cell lymphomas to the functional potential of the neoplastic cells. Such investigations were initially restricted to the few research laboratories capable of performing functional assays. More recently the generation of a wide range of anti-T cell antibodies has facilitated these studies. A further advance in the characterization of T cell lymphomas has come from the demonstration that T cell lymphomas are endemically distributed in Southwestern Japan, suggesting the possibility of transmission by an infectious agent (Takatsuki et al., 1977; Uchiyama et al., 1977; the T- and B-cell Malignancy Study Group, 1981). Shortly afterward circulating antibodies to a retrovirus (designated ATLV) (Hinuma et al., 1981, 1982) and then the retrovirus itself was discovered in all the endemic T cell lymphoma cases studied (Miyoshi et al., 1981) and the same retrovirus (designated HTLV) was independently isolated in the United States from a patient with cutaneous T cell lymphoma (Poiesz et al., 1980, 1981). Since that time many attempts have been made to correlate ‘I

LABELING REACTIONS

OF

TABLE XI1 NEOPLASTIC CELLS A N D ASSOCIATED CELLS I N T CELL LYMPHOMAS/LEUKEMIAS

-

Pleomorphic T cell lymphoma

c

m

Nonendemic cases T-ALLI T-LBO Cells Neoplastic TU14 (3A1,WTl) UCHTllT3 TlI/Lyt 3 Leu-11T1 TU33 NA1134lT6 T4ILeu-3a T8 Leu-7 TdT

(4b

26/26 23/26 16/26 22/26 12/26 15/26 11/26 10126 0112 515

Large

T-IB

L. of plasmacytoid cells

(4

(b)

(b)

(4

(4

(4

(b)

(4

(4

(4

(4

(4

212 212 212 212 212 012 212 012

0115 15/15 15/15

515 13/13 17/17

317

0120 13/15 30131 27/31

015

0113 8/13 5/13

018 718 018 117

417 717 717 517 617 017 717 017

417 717 717 517 617 017 617 117

214 414 414 414 314 014 214 014

315

0115 0115 15/15

0111 10111 10111 10111 10111 0111c 11111 0111 018

0115

0117

T-CLL

Cutaneous Tcell L MF T-PLL S B Z ~ I Y S.

617 717

ATLL

Small Medium

30131 0131

515

515 015 015

415 515 515 015

215 115

011 011 111 011 111 011 011

CALLA (VIL-A1) HLA-DR (TU35) B cell antigen To15 SIg Associated cells FDRC (R4123 positive) IRC (NA1134 positive)

-

14/16 2/26 0126 0126

012 012 012 012

0115

0120

212

0120

212

015

017 117 017 017

017 217 017 017

014 114 014 014

115 115 015 015

111 111 01 1 011

-

-

317

416

214

315

011

-

-

417

317

214

114

o/ 1

0113

018 018 018 018

2/11 2l1ld 0111 0111

4/31 0111 0131

-e

-

-

0111

-

-

-

11/11

0113

0 T-ALL, Acute lymphoblastic leukemia of T type; T-LB, lymphoblastic lymphomas of T-type; T-CLL, chronic lymphocytic leukemia of T type; TPLL, prolymphocytic leukemia of T type; M F , mycosis fungoides; ATLL, adult T cell lymphomalleukemia; T-IB, imniunoblastic lymphonia of T type; FDRC, follicular dendritic reticuluni cells; IRC, interdigitating reticulum cells. b (a) own studies; (b) studies by Catovsky et ~ l (1983); . (c) studies by Takatsuki et d. (1982) and Yamada (1983). c Skin biopsies were investigated in all 11 cases, and many of the NA1/34-positive dendritic cells were hence probablv Langerhans cells and not interdigitating reticulum cells. d Large numbers of Langerhans cells were present in these cases and the possibility could not be excluded that HLA-DR was present on these cells rather than on the tumour cells. No data available.

120

HARALD STEIN ET AL.

antigenic profile in T cell lymphomas not only with morphology but also with the presence of ATLVIHTLV.

A. CORRELATION OF CLINICALLY A N D MORPHOLOGICALLY DEFINED T CELLLYMPHOMAS WITH ANTIGENICPROFILE Table XI1 shows the results of an attempt to correlate morphological features with antigenic profile in T cell lymphomas. The results indicate that among the many surface markers investigated there is none that is restricted to a single morphological lymphoma category. Furthermore, it is evident from Table XI1 that the immunological phenotypes of most T cell lymphomas show considerable variability and overlap. However, there is one additional marker, although not T cell specific, which correlates well with morphology: viz. the presence of deoxynucleotidyl transferase (TdT), a feature selectively associated with lymphoblastic morphology. Its occurrence is restricted in normal cells of the T lineage to cortical thymocytes and in neoplasms to thymic-derived and prethymic “lymphoblasts” (reviewed by Bollum, 1979; Janossy et al., 1980; Brouet and Seligmann, 1981; Habeshaw et al., 1979a). In consequence, there is general agreement that T cell lymphomas can be divided into two main groups:

1. TdT-positive cases, representing thymic and prethymic T cell lymphoblastic lymphomaIleukemia, i. e., a proliferation of “medium sized blasts (lymphoblasts),” the phenotype of which resembles that of thymocytes and their precursors. Such neoplasms have an aggressive clinical course and affect mainly children and young adults. 2. TdT-negative peripheral, or postthymic T cell lymphomas/leukemia, i.e., proliferations of morphologically more mature T cells. These conditions are seen mainly in adults. B. PRETHYMIC A N D THYMIC (TdT-POSITIVE) LYMPHOBLASTIC LYMPHOMA~LEUKEMIAS In earlier studies thymic neoplasms were diagnosed on the basis of binding of sheep erythrocytes (“E rosetting”), the demonstration of localized phosphatase reactivity, rosetting with C3 coated erythrocytes, reactivity with polyclonal anti-T cell antisera, and the detection of increased amounts of TdT in conjunction with lymphoblastic morphology, convoluted nuclei (Fig. 18a), and the presence of a mediastinal mass (reviewed by Catovsky et al., 1982). These studies showed that lymphoblastic leukemia of T type (TALL) varies considerably in marker profile from case to case and does not differ clearly from T cell lymphoblastic lymphoma of T type (T-LB).This is in

FIG. 18. Thymic lymphoblastic lymphoma. (a) Typical histological appearance. Some cells showing nuclear convolution are indicated by arrows. Giemsa. X800. (b)Frozen section immunostained with monoclonal anti-T cell antibody TUl4. All tumor cells are strongly labeled, but a residual B cell follicle is negative. x 150. (c) Adjacent frozen section immunostained with the monoclonal anti-HLA-DR antibody TU35.The cells of the residual B cell follicle are positive whereas the tumor cells are negative. x 150.

122

HARALD STEIN ET AL.

TABLE XI11 GREAVES' S<:IIEME"OF AHHANCINCT-LYMPIIOBl.ASTIC LYMPIIOMAS IN RELATION TO NORMAL DIFFERENTIATION

I (n=4) T8 T4 T11 T6 (NAll34) T1 (Leu-1) T3 Clinical diagnosis ALL LB

I1 (n=11)

+

111 (n=35) +/+/-

+

+/+Wb

+W

+W

IV (n=t3)

+ + + + +W

+/-

4c

10 1

31 4

1 7

Greaves et al. (1981). 6 w, c

Weak.

Number of cases.

keeping with the view that T-ALL is a leukemic manifestation of T-LB. The only difference between the leukemic and nonleukemic cases appeared to be that T-LB is more often associated with an anterior mediastinal mass than is T-ALL (Greaves et al., 1981; Stein et al., 1981~). The heterogeneity of the marker profile in T-ALL and T-LB has prompted attempts to relate the marker profile to stages of thymic differentiation. Using C3 receptors and sheep E rosettes, three stages of thymic differentiation have been proposed (Stein and Muller-Hermelink, 1977): an early or prethymic stage (sheep E receptor negative, C3 receptor positive), a prothymocyte stage (sheep E receptor positive, C3 receptor positive), and a mature stage (sheep E receptor positive, C3 receptor negative). The early thymocyte and the prothymocyte types of disease appeared to be particularly rich in focally distributed acid phosphatase. In T-LB the predominant phenotype was that of prothymocytes [sheep E receptor positive, C3 receptor positive (Stein et al., 1981c)l. More recently the availability of an extensive panel of T cell reactive monoclonal antibodies has provided the opportunity for a more detailed analysis of T-ALL/T-LB in relation to normal T cell differentiation. These studies (Reinherz et al., 1980; Bernard et al., 1981; Greaves et al., 1981) have indicated an even greater heterogeneity of T-ALL/T-LB than was evident in earlier studies. The categories of T-ALLIT-LB, representing different differentiation stages, proposed by Greaves et al. (1981) are listed in Table XIII. These authors defined four major subsets of T cells, two of which represented prethymic differentiation stages and two intrathymic stages.

123

H U M A N LYMPHOMA

TU35 (anti-HLA-DR) TU33 T8 T4 T11 CALLA NA1/34 Leu-1 UCHTl TUl4 Clinical diagnosis ALL

LB ALLILB M ediastinal inass

" Number

+

+ +/-

+ +I+/-

+

+

1 1' 1 0 1

1

+ + 1 1 0 0

+I+I-

+ +

+ + +

014 214

1 0

214 014

1 2

f

+/+/-

+/-

+

+

+/-

+/+/-

+

+ + + 1 1 0 1

+ + +

11s 315 015 5/51

114 314 014 212

of cases.

The results of our own studies are presented in Tables XI1 and XIV. A major finding was the detection of an antibody (TU14) that reacts with all cases of T-ALLIT-LB (see also Fig. 18b), but is unreactive with all B cell neoplasms and with a proportion of mature T cell lymphomas. The reactivity of cells or normal thyinus with TUl4 is shown in Fig. 19. Similar findings had been obtained with the monoclonal antibody WT1 (Tax et ul., 1981) and 3A1 (Catovsky et a l . , 1983), both of which proliably identify the same 40,000 molecular weight protein as TUl4. These fiiidiiigs are important because there is 110 other reagent which shows this type of selective reactivity with early T cells. However, it has to be stressed that TU14 cannot be used to distinguish prethymic and thymic neoplasms from those derived from later (postthymic or peripheral) T cell lyinphoinas since it reacts with approximately 70-80% of the latter type of neoplasm (Table XII). The i~ii~nunocytoche~nical reactions were used as the basis of an attempt to categorize our cases of T-ALL/LB into groups of different differentiation stages. As can be seen from Table XIV, this led to a slightly different and extended scheme compared to that proposed by Greaves et al. (1981) (Table XIII); this was due to the use of antibodies TU14, TU33, and UCHTl in place ofT3; UCHTl and T3 are believed to detect the same 19,000 MW antigen, but UCHTl produces much stronger staining. The TU33 antigen

124

HAKALD STEIN ET AL.

Fic:. 19. Frozen section of a norinal thyinus gland froin a 3-year-old child inimunostained with the monoclonal anti-T cell antibody TU14. All cortical thymocytes are stained, some inore strongly than others. 11) the mrclulla ii minor popiilatioii of 1al)eletl thymocytes is seen. X250.

appears later than other T cell-associated antigens during intrathymic differentiation (see Fig. 22). Eight major subsets of prethymic and thymic T cells were identified. Category VII (the common cortical thymic phenotype) is characterized by an antigen profile identical to that of nearly all normal cortical thymocytes. Categories I-VI are defined by increasing completeness of this composite antigen profile and probably represent earlier (and possibly prethymic) phenotypes. This appears to be true especially for the three T-ALL cases which expressed TU14 alone and may therefore be derived from early/prethymic T cells in the bone marrow. Surprisingly, two of these cases were HLA-DR

H U M A N LYMPHOMA

125

FIG.20. Normal thymus from a 3-year-old child. (a) Frozen section immunostained with the monoclonal anti-HLA-DR antibody TU35. This antibody reveals a dense dendritic meshwork of nonlyinphoid cells in the cortex (the medulla is not shown in this picture). The lymphoid cells among the meshwork are unstained. However, in the spherical area at the center of the field the dendritic meshwork is missing and the lymphoid cells show ring-like labeling. x50. (b) Touch imprint of thymus tissue immunostained for HLA-DR with monoclonal antibody TU35. The cortical thyinocytes (their origin being confirmed by immunostaining adjacent touch imprints with monoclonal antibody NA1/34), macrophages (single arrow), and thymic nurse cells (double arrows) are positive. x350. (c) Adjacent touch imprint immunostained with the monoclonal antiepithelial antibody TU25. Thymic nurse cells are positive, but the cortical thymocytes are unstained. ~ 3 5 0 .

positive. HLA-DR-positive T-ALL has not been described previously, but normal lymphoid stem cells are believed to be HLA-DR positive, and this may explain the appearance of this antigen on these two cases. However, we have shown that weak HLA-DR expression on a varying percentage of normal cortical thymocytes (Fig. 20a and b) can be demonstrated by staining frozen sections and touch imprints of thymus (instead of suspended cells as used in other studies). This finding emphasizes that HLA-DR expression may potentially occur at various stages of T cell maturation. Greaves et al. (1981)interpreted in their study all cases expressing either

126

HARALD STEIN ET AL.

T8 or T4 as being more immature than those with both T8 and T4 on their surface membranes (corresponding to our Category VII). However, the demonstration of the late cortical thymocyte marker TU33 on some of these cases suggests that a proportion of T8-positivelT4-negative or T8-negative/T4-positive cases represents a differentiation stage later and not earlier than that of common cortical thymocytes. On the whole our immunophenotypical studies confirm the work of others (Greaves et al., 1981; Catovsky et al., 1983) showing that T-ALL and T-LB are overlapping diseases, but that the two diseases tend to show arrest at a different maturation stage, i.e., most T-ALL have a prethymic or immature thymic phenotype whereas T-LB tends to be associated with a mature cortical thymic phenotype. These differences in membrane phenotype may be important from a clinical standpoint since it is reported that T lymphoblastic lymphomas with NA1/34-positive (T6-positive) and/or T4-positive, T8positive cells do not manifest bone marrow infiltration early in the disease (Bernard et al., 1981). The assessment of T cell lymphomas for the expression of CALLA revealed-in agreement with other studies (e.g., Greaves et al., 1981)-that this antigen was largely restricted to T-ALL/T-LB, with only four exceptions in our series of 65 cases of peripheral (postthymic) T cell neoplasms (see Table XI1 and below). It has previously been suggested (Thiel et al., 1980, 1981) that lymphoma or leukemic cells which simultaneously express T antigens and CALLA represent a hybrid form of neoplasm, lying between common ALL (C-ALL) and T-ALL (so-called C/T-form of ALL). The results in our series of cases are hence surprising, in that CALLA expression was not associated with an early differentiation stage but rather with a more mature thymic phenotype (Table XIV). This suggests that pre-T cells are, in contrast to pre-B cells, CALLA negative and that T cells acquire CALLA during intrathymic differentiation. By immunostaining normal fetal and postnatal thymus glands we were able to demonstrate that CALLA-positive cells do indeed exist in the thymic cortex (Fig. 21b). The number of CALLA-positive cells increased with age of gestation. However, their number is surprisingly low in view of the relatively high percentage of CALLA-positive cases of T-ALL/T-LB. Taken together, these studies of the phenotype of T-ALL/T-LB reveal a greater variety of antigen expression than is seen in normal thymus glands (see Fig. 22). Hence, while normal thymocytes may be subdivided into only three differentiation stages (large thymic blasts, cortical thymocytes, and medullary thymocytes) (e.g., Janossy and Prentice, 1982), as many as eight different categories of T-ALL/T-LB can be distinguished. It is possible that this greater complexity of neoplastic T cells is an illusion, reflecting neoplasia- and proliferation-associated changes in surface markers. However,

H U M A N LYMPHOMA

127

FIG.21. Normal thymus gland from a 3-year-old child (a) Frozen section immunostained with monoclonal antibody NA1/34. The cortical thymocytes show a strong staining. In the medulla only 10-258 of the lymphoid cells are labeled. X 100. (b) Frozen section iminunostained for CALLA with monoclonal antibody VIL-A1. Only a small percentage of thymocytes, principally localized in the cortex (C), are labeled. M , Medulla. X 100.

er, the authors prefer the explanation that the categories revealed by immunocytochemical analysis do indeed represent different maturation stages and that the study of T cell neoplasms provides a valuable means of identifying small subsets of early thymocytes or prethymocytes which are too few in number to be detected in the normal thymus.

C. PERIPHERAL T CELLL Y M P H O M A ~ L E U K E M I A S

As mentioned above, no monoclonal antibodies have been described which react selectively with any of the morphologically defined subtypes of peripheral T cell lymphoma/leukemia (e.g., S6zary syndrome, T-CLL). Hence, the classification of T cell lymphomas and leukemias used in this review is based on multiple criteria, i. e., immunophenotype, morphology, clinical behavior, and evidence of viral etiology (e.g., the demonstration of antiviral antibodies, or the presence of viral antigens and/or DNA sequences). On this basis, peripheral T cell neoplasms may be separated into the following variants: lymphoblastic lymphoma, chronic lymphocytic leukemia, prolymphocytic leukemia, cutaneous lymphoma, T-zone lymphoma, pleomorphic lymphoma with endemic and nonendemic subtypes, immu-

128

HARALD STEIN ET AL.

@D

TUlL? H L A - D R ?

,one narrow 3rethymo cyte

large thymic blast

10.5 - 5%)

small and medium sized cortical t hymocyt e (60-80%)

T U 33 medullary thymocyte ( 15 - 20%)

TUlL 7

inducer helper

Leu - 1 UCHTl TUlL +/-

111 Leu - 1 suppressor1 UCHT 1 cytotoxtc TUlL

peripheral T cell

stages are defined on the basis of reactivity with monoclonal antibodies. Modified from Janossy and Prentice (1982).

noblastic lymphoma, and lymphoma of plasmacytoid T cells (see table). These different categories are described in detail below.

1. T Lymphoblastic LymphomalLeukemia (T-LBIT-ALL) Several studies have reported that cases of T-LBIT-ALL with the phenotype of peripheral T cells (TdT negative, dot like acid esterase positive) are

H U M A N LYMPHOMA

129

occasionally encountered (Stein et al., 1981~).However, their rarity is indicated by the fact that no such cases were found among the 58 cases of T-LB and T-ALL reported by Greaves et nl. (1981) or our own series of 26 cases.

2 . Chronic Lymphocytic Leukemia of T Type (T-CLL) The term “T-CLL” has been widely used by many authors to encompass a variety of T cell-derived leukemic neoplasms including T-PLL (Brouet et al., 1975; Toben and Smith, 1977; Boumsell et al., 1981), T cell lymphomas with blood involvement (Reinherz et al., 1979), and T cell leukemia/lymphoma of Japanese and blacks, designated adult T cell lymphoma/leukemia (ATLL) (Toben and Smith, 1977; Lennert et al., 1982). In accordance with Catovsky et al. (1982), we think that the term T-CLL should be restricted to those leukemias in which the neoplastic cells cytologically resemble normal small T lymphocytes, i.e., cases showing marked nuclear irregularity or pronounced cellular pleomorphism have to be analyzed further to determine whether they belong in the other categories of T cell peripheral neoplasms detailed below. Two types of T-CLL may be distinguished; the neoplastic cells in the first type are positive for T4 antigen (i.e., are of helper type), contain acid esterase (Stein et al., 1981c) and dipeptidylaminopeptidase IV (Feller and Panvaresch, 1981), and possess small irregular nuclear protrusions. The latter feature led to the name “knobby type” (Levine, 1981).The two T-CLL cases in our series of T cell lymphomas/leukemias were of this type, whereas all 15 cases of T-CLL reported by Catovsky et al. (1983) were of the second type (Table XII). In the second type nuclei are round and regular, and azurophilic granules are seen in the cytoplasm (Fig. 23a). However, in some cases nuclei can be fairly irregular in shape and/or azurophilic granules may be missing (Feller, 1983). Acid esterase (Stein et al., 1981c) and dipeptidylaminopeptidase IV (Feller and Panvaresch, 1981) are absent. Catovsky et al. (1983) have reported that the phenotype is that of a T suppressor cell. However, in functional assays the neoplastic cells from most cases do not suppress B cell differentiation in uitro (Hoffman et al., 1982; Thien et al., 1982) but function instead in most instances as killer cells [antibody-dependent cellular cytotoxicity (ADCC)] (Humke et al., 1982). This is in keeping with the fact that hypogammaglobulinemia is not a common clinical feature in these patients (Catovsky et al., 1983). A few cases of T-CLL have been found to express OKMl (which represents the C3bi receptor), with or without T8 antigen. The cells of these latter cases functioned in uitro as natural killer cells as well as being capable of ADCC (Rumke et al., 1982; Schlimok et al., 1982). These differences probably reflect the different functional capacities of specific T cell subsets. However, it is not clear how they relate to the

130

HARALD STEIN ET AL.

FIG. 23. Blood smears from two types of leukemic T cell disorder. (a) Typical cytological appearance of T%positive, T4-negative T cell leukemia (cytotoxic/suppressor phenotype). Note the abundant pale cytoplasm and the presence of azurophilic granules (see arrowed cell). MayGriinwald-Giemsa. x 800. (b) Cytological appearance of circulating SBzary cells. They are characterized by a cerebriform nucleus. May-Griinwald-Giemsa. x800.

hematological manifestations often seen in such cases of T-CLL, e.g., neutropenia and/or red cell hyperplasia. Many of these cases of T4-negative, T8-positive T-CLL have only a modest lymphocytosis at presentation, and may be clinically stable for many years. Hence the truly malignant nature of this disorder has sometimes been questioned. However, the observation that spontaneous regressions never occur and that in some patients a clear progression to a more aggressive neoplasm takes place (Catovsky et al., 1982; Hoffman et al., 1982) suggests that this disorder represents a true neoplastic proliferation. This view is further substantiated by the finding that the surface marker profile of the cells in this disorder differs from that of normal T suppressor cells, i.e., the T cell-associated 40,000 molecular weight protein recognized by antibodies 3A1 (and probably also by TU14 and WT1) is absent from the neoplastic cells, while being present on 85% of normal peripheral T cells including all

H U M A N LYMPHOMA

131

T8-positive cells (Haynes et d . , 1981; Catovsky et al., 1983). This suggests that T8-positive T-CLL derives from a T cell subset which is present in normal individuals in only very small numbers. However, this observation requires confirmation in a larger number of cases. 3. Prolymphocytic Leukemia of T Type (T-PLL) This condition is characterized by splenomegaly, lymphadenopathy, and skin lesions (Catovsky et al., 1982). According to Catovsky et al. (1983), it is the only peripheral T cell leukemia that is consistently positive for the T cellassociated 40,000 molecular weight protein recognized by monoclonal antibodies 3A1 (and probably also by TU14 and WTl), a marker commonly found on T-ALL cells and on normal thymic blast cells (see above). Further features include the coexpression in some cases of T4 and T8 (as occurs in cortical thymocytes) despite the “postthymic” phenotype of the neoplastic cells (T6 negative, TdT negative), and the lack of helper function in uitro despite a T4-positive phenotype (Rumke et al., 1982).

4 . Cutaneous T Cell Lymphoma [Sbzary Syndrome (SS) and Mycosis Fungoides ( M F ) ] SS and M F are discussed together since both disorders are characterized b y the proliferation of lymphoid cells that exhibit epidermotropisrn and have irregular deeply convoluted (cerebriform) nuclei. In SS these cerebriform cells are referred to as Skzary cells (Fig. 23b) and in M F as Lutzner cells. In M F larger cells may also occur; they are called mycosis cells. In both conditions neoplastic cells infiltrate the epidermis either singly or in small groups, forming Pautrier’s microabscesses (Fig. 24a). These lymphoid cell infiltrates are responsible for generalized exfoliative erythroderma in SS and for scaly skin eruptions in MF, often progressing to plaques of tumor cell infiltration. The main difference between SS and M F is that SS is associated with a leukemic blood picture, and many authors consider that SS may be a leukemic variant of M F . Immunologically SS and M F have been described as peripheral T cell lymphomas with the surface antigen profile (Kung et al., 1981; Boumsell et al., 1981) and in uitro function o f T helper cells (Broder et al., 1976; Berger et al., 1979). However, rare exceptions have been reported in which the cerebriform lymphoid cells showed a T suppressor immunophenotype and/or acted in uitro as T suppressors (Hopper and Haren, 1980; Haylies et al., 1982). In our phenotypic analysis (Table XII) the circulating Skzary cells showed the classical marker profile of helper T cells referred to above, i.e., reacting with antibodies T3, anti-Leu-1, T11, T4 but giving no reactions with T8. One case of SS was remarkable since it did not express the sheep erythrocyte receptor (i,e., was T11 negative). This case was also unreactive with both T4

FIG. 24. Mycosis fungoides. (a) Histological appearance in tlie skin. Neoplastic cells with cerebriform nuclei infiltrate tlie epidermis, tending to cluster and form Pautrier’s microabscesses. Giemsa. X800. (I)) Frozen section immunostained with the monoclonal anti-T cell antibody UCHT1. The majority of the cells that infiltrate the dermis and epidermis are labeled. X500. (c) Frozen section immunostained with the monoclonal antibody NA1/34, which react with cortical tliymocytes, Langerhans cells, and interdigitating reticulum cells. The staining reveals large numbers of NAU34-positive Langerhans cells and/or interdigitating reticulum cells in the epidermis (arrowed) and in the tumor infiltrates in the dermis. X56.

HUMAN LYMPHOMA

133

and T8 but expressed Leu-7, an antigen which is reported to be present on natural killer (NK) cells (Abo and Balch, 1981). Beyond that we noticed two cases in which only about 50-60% of S6zary cells expressed the sheep erythrocyte receptor. The phenotype of M F is very similar to that of SS. All but one case stained positively with the pan-T cell reagents anti-Leu-1 and UCHTl/T3 (Fig. 24b). The one case negative for Leu-1 and UCHTl/T3 was histologically of tumor forming type, and was atypical in that the typical feature of epidermotropism was absent and a high number of mitotic figures were seen. In our series all cases showed the phenotype of T-helper/inducer cells. We found two cases expressing the common ALL antigen. One case was histologically of tumor-forming type, and the great majority of neoplastic cells were CALLA positive. The other case was classical M F with marked epidermotropism in which about 40-50% of lymphoid cells showed CALLA positivity. Another very common feature of M F is the presence of large numbers of NA1/34- (or OKT6-) positive and HLA-DR-positive Langerhans and/or interdigitating reticulum cells in the epidermis as well as in the tumor infiltrates in the corium (Fig. 24c). The presence of HLA-DR on these cells may make it difficult in individual cases to decide whether the tumor cells themselves express this antigen. Though these cells are a common feature in this disease, their occurrence in subepidermal layers of the skin is not of great diagnostic value because these cells can also be plentiful in the corium in biopsies of chronic dermatitis. Finally it is noteworthy that all our cases of MF showed a negative reaction for the T cell associated 40,000 molecular weight protein recognized by TUl4, while this antigen was expressed on the cells of three out of seven cases of SS. This finding agrees with others pointing to a difference between the neoplastic cells involved in SS and MF.

5. T Zone Lymphoma We originally chose the name “T zone lymphoma” for this group of tumors because our first cases showed selective involvement of T cell areas of lymphoid tissue, associated with a striking preservation (or even hyperplasia) of B cell follicles (Fig. 25) (Lennert et al., 1975; Lennert and Mohri, 1978). Furthermore, the characteristic components of the T zone, i.e., interdigitating reticulum cells and epithelioid venules, were present between the neoplastic cells. However, as more cases accumulated, it became evident that good preservation of B cell follicles was not a constant feature and could not be regarded as typical of this lymphoma category. In addition, the number of interdigitating reticulum cells and epithelioid venules varies greatly from

FIG.25. T zone lymphoma. (a) Typical histological appearance at low magnification. Although the neoplastic cells are confined to the pale interfollicular areas (T zones) the distribution of the residual normal follicles (F)may simulate a follicular lymphoma. Giemsa. X 14. (b) Histological appearance of the neoplastic interfollicular area (T zone) at a higher magnification. Polymorphic small and medium-sized lymphoid cells predominate. Giemsa. X800. (c) Frozen section immunostained with the monoclonal anti-T cell antibody OKT11. The neoplastic cells surrounding the lymphoid follicle (right upper corner) are strongly stained. x 150.(d) Adjacent frozen section immunostained with monoclonal anti-lgD. The cells of the lymphoid follicle are positive whereas the surrounding neoplastic T cells are unstained. x 150.

H U M A N LY M P H O M A

135

case to case, and these elements may also be prominent in T cell lymphomas of other types. Hence, we felt that this T cell lymphoma category is probably best defined on the basis of the cytology of the tumor cells, i.e., nuclear irregularity and variable cell size, rather than on the distribution patterns of other, nonneoplastic elements (Lennert et d., 1982). In consequence we included T zone lymphoma in the category of pleomorphic T cell lymphoma (see below for further details). However, at a recent conference on Japanese and European T cell lymphomas held in Kiel in September 1983 Japanese lymphoma experts reported that in Japan typical cases of T zone lymphoma are ATLV negative in contrast to the pleomorphic T cell lymphomas without lymphoid follicles. Although this observation requires confirmation by further studies, it suggests that pleomorphic T cell lymphomas containing lymphoid follicles and significant numbers of epithelioid venules (i.e., T zone lymphomas) are distinct from, rather than identical to, pleomorphic T cell lymphomas which lack lymphoid follicles and show a low density of epithelioid venules. Furthermore, T zone lymphomas can contain clear cells while such cells have not been observed in (ATLV-positive) pleomorphic T cell lymphoma without lymphoid follicles. 6. Pleoinorphic T Cell L y m p h o m This group of T cell lymphomas is characterized by a proliferation of T cells of varying size with very irregularly shaped nuclei. Pleomorphic T cell lymphoma cannot be easily classified within the Kappaport and International Working Formulation schemes. This lymphoma type is classified as diffuse mixed, diffuse poorly differentiated, and diffuse large lymphoid, which cut across traditional Rappaport categories and also across categories in the Working Formulation. On the basis of virological studies at least two types or groups of pleomorphic T cell lymphomas can be distinguished: endemic, ATLV-positive cases and nonendemic, ATLV-negative cases. a. Endevnic, ATLV-Positive Pleomorphic T Cell Lymphoma [Adult T Cell LymphomalLeukemia (ATLL)]. This type of pleomorphic T cell lymphoma is usually called adult T cell lymphoma/leukemia (ATLL) and this is used in the discussion below. ATLL is a rapidly progressive disease characterized by lymphadenopathy, frequent hypercalcemia, and skin involvement. In 7 0 4 0 % of cases, peripheral blood contains numerous neoplastic cells which have characteristic pleomorphic nuclei (Hanaoka et al., 1979; Kikuchi et al., 1979). This marked nuclear irregularity is also seen in lymph node sections which show diffuse replacement by the neoplastic process. Histologically, ATLL has been subclassified into three groups, i. e., pleomorphic, medium-sized, and large cell type (Kikuchi et al., 1982). The nuclear polymorphism is most pronounced

136

HARALD STEIN ET AL.

in the medium-sized category (Hanaoka et al., 1979; Hanaoka, 1982; Kikuchi et al., 1979, 1982). Epidemiological and virological studies provide evidence that ATLL is a distinct disease entity of viral etiology. The places of birth of patients with ATLL in Southwestern Japan show a striking clustering (Takatsuki et d., 1977; Uchiyama et al., 1977). All cases of ATLL have been associated with the presence within the neoplastic cells of a human retrovirus, designated adult T cell leukemia virus (ATLV), and antibodies against the major structural core protein (p24) of ATLV are constantly present in the serum of these patients (Hinuma et al., 1981; Hinuma, 1982; Miyoshi et al., 1981). The same virus had been isolated previously in Gallo’s laboratory (Poiesz et al., 1980, 1981) from both fresh cells and T lymphoblastoid cell lines from two black patients suffering from a disease identical to ATLL, (initially categorized as a “cutaneous T cell lymphoma”). More recently, ATLL has been detected in West Indian and Caribbean patients (Catovsky et al., 1982), and the virus also appears to be endemic in this area. As Table XI1 shows, the immunophenotype is identical in both Japanese and Caribbean cases: T3 positive, T11 positive, T4 positive, T8 negative. Although all immunological studies agree that ATLL cells exhibit a T helper cell phenotype, most functional studies have shown that they suppress rather than induce B cell differentiation in uitro (Yamada, 1983; Hattori et al., 1981). This finding was initially difficult to understand; however, recent studies have shown that T4 cells contain two subsets: one helps B cells to differentiate while the other is normally inert but acquires potent suppressor activity following pokeweed mitogen stimulation (Thomas et al., 1981, 1982). Hence, ATLL cells probably derive from this second T4-positive subset, in contrast to classical Sezary and mycosis fungoides which arise from the first T4-positive subset. All five cases of ATLL investigated by Catovsky et al. (1983) were negative for the T cell-associated 40,000 molecular weight protein detected with the antibody 3A1. Recently we were able to immunostain a larger series of Japanese cases of ATLL. All cases proved to be negative for the T cellassociated 40,000 molecular weight protein (detected by TUl4); the absence of this antigen may therefore enable ATLV-positive T cell lymphomas to be distinguished from the majority of the nonendemic pleomorphic T cell lymphomas (see below). b. Nonenakmic, ATLV-Negative Pleomorphic T Cell Lymphoma. This category has received different names in the literature: immunoblastic lymphoma of T cell type (Lukes and Collins, 1974), malignant lymphoma of peripheral T cell type (Waldron et al., 1977), T zone lymphoma (Lennert et al., 1975; Lennert and Mohri, 1978; Helbron et al., 1979), mixed

H U M A N LYMPHOMA

137

FIG.26. Schematic diagram of the different cellular variants of pleomorphic T-cell lymphoma (small, medium-sized. and large). For comparison, two norinal B lymphocytes, a plasma cell, and two mast cells are shown at the left and two interdigitating reticulum cells (IRC) at the right. From Lennert et u2. (1982).

“blastic”/cytic T cell lymphoma (Stein et al., 1981c), and adult T cell lymphoma/leukemia (Suchi et al., 1979). The relationship between nonendemic pleomorphic T cell lymphoma and T zone lymphoma is discussed above. Three cellular variants can be distinguished: small, medium-sized, and large. The cytological characteristics of these variants are illustrated in Figs. 26 and 27a. Table XI1 shows that all nonendemic cases of pleomorphic T cell lymphoma were positive for UCHTl and T11 (Fig. 27b), whereas the T cell markers Leu-1, Lyt 1, TU33, and TU14 were less constantly expressed. Fifteen out of 18 cases were T4 positive and T8 negative and one case was T8 positive and T4 negative. The T8-positive case was found among the medium-sized variants and did not show any morphological differences from the other medium-sized cases in histological sections. However, in imprints prepared from the biopsy of this case the tumor cells exhibited a pale cytoplasm containing azurophilic granules like the T8-positive T-CLL cases. There were no specific differences in the antigenic profile between the three variants, except a tendency for the large cell variant to show a more incomplete T cell phenotype than the smaller and medium-sized variants (i.e., to lack the “full house” of T cell markers). Immunohistological staining with anti-FDRC antibody produced an unexpected finding. In more than one-third of the cases neoplastic areas con-

FIG. 27. Pleomorphic T cell lymphoma, medium-sized. (a) Giemsa-stained plastic embedded section. Note the deeply cleft nuclei with sinuous outline and particularly the concave aspect of their margins. Some nuclei resemble jellyfish (arrowed). X800. (11) Frozen section immunostained with a monoclonal anti-T cell antibody TU33. The majority of cells are positive. X 100. (c) Frozen section immunostained with the monoclonal antibody R4/23 reactive with follicdar dendritic reticulum cells (FDRC). Patchy FDRC meshworks are demonstrated. X75. (d) Frozen section immunostained with the monoclonal antibody NA1/34, revealiug negative tumor cells interspersed with some intensely stained interdigitating reticulum cells. x 75.

H U M A N LYMPHOMA

139

tained an ill-defined meshwork of FDRC, without associated B cells (Fig. 27c). These FDHC meshworks were not visible in conventionally stained sections. The presence of dendritic reticulum cells in pleomorphic T cell lymphoma cannot be used as a criterion for distinguishing it from other lymphoma types, since similar patchy dendritic reticulum cell meshworks are seen in other types of lymphoma, e.g., T-CLL, mycosis fungoides, T immunoblastic lymphoma, and in angioiminunoblastic lymphadenopathy (roughly equivalent to lymphogranulomatosis X). In half of the cases the pleomorphic T cell lymphomas contained significant numbers of NA1/34-positive interdigitating reticulum cells (Fig. 27d). Our nonendeinic cases of pleoinorphic T cell lymphoma (except the T8positive and T4-negative case) resemble ATLL both morphologically and immunologically. Furthermore, it is known that both types of T cell neoplasm show a poor response to therapy and a short survival. This raises the question of whether nonendemic pleoinorphic T cell lymphomas and ATLL derive from the same subset of T cells (although differing in etiology). An analogy might be drawn with the two types of B-lymphoblastic lymphoma (i.e., endemic EBV-positive African cases and nonendemic EBV-negative European and American cases). All cases of pleomorphic T cell lymphomas investigated to date in Kiel have proved negative for serum antibodies to HTLV. However, the exact relationship of pleomorphic T cell lymphoma to ATLL remains to be clarified by further immunological and functional investigations. The immunological data presented in Table XI1 show one difference: all cases of ATLL investigated by Catovsky et al. (1983) and recently in our own laboratory (to be published) were negative for the T cellassociated 40,000 molecular weight protein, whereas in nonendemic pleomorphic T cell lymphoma this antigen was found in more than 50% of the cases. In a recent paper (Lennert et a l . , 1982) we had included lympho-epithelioid lymphoma (Lennert’s lymphoma) in the pleoinorphic T cell lymphoma category. It appears to be quite clear, however, that other lesions (e.g., LP immunocytoma) may show a similar morphologic picture. Hence the T cell nature of each case diagnosed as lympho-epithelioid lymphoma must be proven.

7. T Iintnunoblastic Lyinphoina T cell lymphomas of this type are characterized by a proliferation of large basophilic cells with round or oval nuclei. Nucleoli are usually smaller than those seen in B iminunoblastic lymphoma and often multiple. The cytoplasm is moderately to intensely basophilic. Only 5 of our 71 cases of T cell neoplasms were of this type. The results of our immunohistological investiga-

140

HARALD STEIN ET AL.

tions (Table XII) showed that three cases were of T helper and one of T suppressor phenotype, and one lacked both of these markers. It is noteworthy that three of the five T immunoblastic lymphomas contained ill-defined meshworks of dendritic reticulum cells, whereas interdigitating reticulum cells were detectable in significant numbers in only one case.

8. T Cell Lymphoma of Clear Cell Type It is not yet clear whether T cell lymphomas composed predominantly of clear cells represent a separate entity or rather variants of other lymphoma types. Our own experience is limited since only 2 of our 71 cases were of this type. These two cases showed a helper T cell phenotype and revealed a patchy meshwork of dendritic reticulum cells in the absence of B cell follicles.

9. Lymphoma of Plusmucytoid T Cells This is a recently described type of T cell lymphoma derived from plasmacytoid T cells. The two reported cases (Miiller-Hermelink et al., 1983) were associated with myeloid leukemia. Plasmacytoid T cells occur in clusters within T zones in 10% of lymph nodes which show reactive lymphoid hyperplasia (Vollenweider and Lennert, 1983). These T cells are characterized by a moderate amount of rough endoplasmic reticulum (ergastoplasm) and a distinctive immunophenotype (see Table XII). The moderate amount of ergastoplasm gives this special T cell subset its plasmacytoid appearance, which is easily recognizable following Giemsa staining (Fig. 28; although not in hematoxylin and eosin stained sections). These cells presumably possess secretory functions, but their product has yet to be identified. VI. Conclusion

The analysis of human tissue lymphoma biopsies by immunohistological techniques provides the opportunity of investigating whether previously established histological categories can be correlated with different patterns of antigen expression. The results reported above are gratifying, in that many previously recognized lymphoma categories do indeed appear to represent immunologically distinct entities. In particular the concept, central to the Kiel classification, that many non-Hodgkin’s lymphomas, both diffuse and follicular, are derived from germinal center cells, is confirmed in this study. Not only are the surface antigens expressed on the neoplastic cells in these lymphomas similar to those seen on normal germinal center cells, but there is also a meshwork of FDRC present in most cases, providing additional evidence for their germinal center origin. However, it should be noted that immunological phenotyping does reveal a few differences in sur-

HUMAN LYMPHOMA

141

FIG. 28. Typical histological appearance of a lymphoma of plasmacytoid T cells. Note the eccentric position of the nuclei within the slightly basophilic cytoplasm of the tumor cells. Giemsa-stained plastic ernbedded section. x 800.

face antigen expression from that seen on normal germinal center cells and it remains to be seen whether these discrepancies reflect alterations in phenotype associated with neoplastic transformation or the derivation of these tumors from subtypes of germinal center cells. Immunohistological analysis has also shed new light on the diffuse lymphomas. However, numerous questions in this field remain to be answered. A larger series of lymphoblastic lymphoma cases requires to be studied. No clear-cut distinction can be drawn at present between the two major large cell categories found in the Kiel classification (centroblastic and immunoblastic). T cell lymphomas have also proved, as a result of immunological staining, to be more heterogeneous and complex than had previously been suspected. Finally the nature of lymphomas expressing Ki-1 antigen, and their relationship to Hodgkin’s disease, requires to be elucidated. The results reported above thus indicate that immunohistological studies, while providing much new insight into human lymphoma, also raise many new and fascinating questions. The next few years are likely to see a rapid expansion of our knowledge in this field and the resolution of many questions which currently perplex lymphoma pathologists concerning the origin of neoplastic lymphoid cells.

142

HARALD STEIN ET AL.

ACKNOWLEDGMENTS The work described in this article was supported by grants from the Deutsche Forschungsgemeinschaft SFB 111, project CLI, and the Leukaemia Research Fund, U.K. This review was compiled during the tenure by Prof. Stein of a Leukaemia Research Fund Senior Research Fellowship at the Nufield Department of Pathology, University of Oxford, John Radcliffe Hospital, Oxford.

REFERENCES Abo, T., and Balch, C. M. (1981).J. Zmmunol. 127, 1024. Aisenberg, A. C., and Blocb, K. J. (1972). N. Engl. J. Med. 287, 272. Aisenberg, A. C., Wilkes, B. M . , and Harris, N. L. (1983). Blood 61, 469. Barcos, M. P., and Lukes, R. J. (1975). In “Conflicts in Childhood Cancer. An Evaluation of Current Management” (L. F. Sinks and J. 0.Godden, eds.), pp. 147-178. Liss, New York. Berger, C. L., Warburton, D., Raefat, J., et al. (1979). Blood 53, 642. Bernard, A., Boumsell, L., Reinherz, E. G., et al. (1981). Blood 57, 1105. Beverley, P. C. L., and Callard, R. E. (1981). Eur. J. Zmmunol. 11, 329. Bollum, F. J. (1979). Blood 54, 1203. Boumsell, L., Bernard, A., Reinherz, E. L., Nadler, L. M., Ritz, J., Coppin, H., Richard, Y., Dubertret, L., Valensi, F., Degos, L., Lemerle, J.. Flandrin, G., Dausset, J . , and Schlossman, S. F. (1981). Blood 57, 526. Breard, J. M., Reinherz, E. L., Kung, P. C., Goldstein, G., and Schlossmann, S. F. (1980).J. Immunol. 124, 1943. Broder, S . , Edelson, R. L., Lutzner, M. A., et 01. (1976).J . Clin. Znoest. 58, 1297. Brouet, J. C., and Seligmann, M. (1981). Pathol. Res. Pract. 171, 262. Brouet, J. C., Flandrin, G., Sasprotes, M., et ~ l (1975). . Lancet ii, 890. Brouet, J. C., Preud’Homme, J.-L., Penit, G . , et al. (1979). Blood 54, 269. Caligaris-Cappio, F., Gobbi, M., Bofill, M., and Janossy, G. (1982).J. Exp. Med. 155, 623. Caligaris-Cappio, F., Gobbi, M., Bergui, L., Campana, D., Laura, F., Fierro, M. T., and Foa, R. (1984). Br. J. Haematol. 56. Catovsky, D. (1982). Nouo. Reo. Fr. Hemutol. 24, 343. Catovsky, D . , and Costello, G. (1979). Basic Appl. Histochem. 23, 255. Catovsky, D., Millain, E., Okos, A,, and Galton, D. A. G. (1974a). Lancet ii, 751. Catovsky, D., Pettit, J. E . , Galetto, J., et d. (1974b). Br. J. Haemutol. 26, 29. Catovsky, D., Cherchi, M., Brooks, D., Bradley, J., and Zola, H. (1981). Blood 58, 406. Catovsky, D., Linch, D. C., and Beverley, P. C. L. (1982). Clin. Haemutol. 11, 661. Catovsky, D., San Miguel, J. F., Soler, J., Matutes, E., Melo, J. V., Bourikas, G . , and Haynes, B. F. (1983).J. Exp. Clin. Cancer Res. 2, 229. Cohen, H. J.. George, E. R., and Kremer, W. B. (1979). Blood 53, 764. Cordell, J. L., Falini, B., Erber, B. N., Ghosb, A. K., Abdulaziz, Z., Macdonald, S., Pulford, K. A. F., Stein, H.,and Mason, D. Y. (1984).J. Histochem. Cytochem. 32, 219. Dalchau, R., and Fabre, J. W. (1981).J. E x p . Med. 153, 753. Dalchau, R . , Kirkley, J., and Fabre, J. W. (1980). Eur. J. Zmmunol. 10, 737. Dighiero, G., Charron, D., Debre, P., et al. (1979). Br. J. Haemutol. 41, 169. Dvoretsky, P., Wood, G. S., Levy, R., and Warnke, R. A. (1982). Hum. Pathol. 13, 618. Engleman, E. G., and Levy, R. (1980). Clin. Res. 28, 502A. Evans, R. L., Wall, D. W., Platsoucas, C. D., Siegal, F. P., Fikrig, S. M., Testa, C. M., and Good, R. A. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 544. Falini, B., Schwarting, R . , Erber, W., Posnett, D. N., Martelli, M. F., Grignani, F., Zuccaccia, M., Gatter, K. C., Stein, H., and Mason, D. Y. (1984). Submitted.

H U M A N LYMPHOMA

143

Feller, A. C. (1983). 2nd Workshop European Lymphoma Study Group, Southampton. Feller, A. C . , and Parwaresch, M. R. (1981). J. Cancer Res. Clin. Oncol. 101, 59. Gajl-Peczalska, K. J., Bloomfield, G. D., Frizzera, G., Kersey, J. H., and Le Bien, T. W. (1982). In “B and T Cell Tumors” (E. S. Vitetta, ed.), pp. 63-67. Academic Press, New York. Galton, D. A. G., Goldmann, J. M., Wiltshaw, E., Catovsky, D., Henry, K., and Goldenberg, G. 1. (1974). Br. J. Haematol. 27, 7. Gatter, K . C . , Alcock, C . , Heryet, A., Pulford, K. A , , Taylor-Papadimitriou, J . , Stein, H., and Mason, 11. Y. (1984). Am. J. Clin. Pathol., in press. Gerard-Marchant, R., Hamlin, I., Lennert, K., Rilke, F., Stansfeld, A. G . , and van Unnik, J. A. M. (1974). h n c e t II, 406. Gerdes, J., Naieni, M . , Mason, D. Y . , and Stein, H. (1982). Immunology 45, 645. Gobbi, M., Caligaris-Cappio, F., and Janossy, 6. (1983). Br. J. Haetnntol. 54, 393. Golde, 11. W., Stevens, R. H . , Quan, S. G . , et al. (1977). Br. J. Ifaematol. 35, 359. Greaves, M. F., Brown, G., Rapson, N . T., and Lister, R. A. (1975). Clin. Zmmtrnol. Zinmunopathol. 4 , 67. Greaves, M. F., Verbi, W., Vogler, B. L., et al. (1979). Leuk. Res. 3, 353. Greaves, M. F., Rao, J., Hariri, G . , Verbi, W., Catovsky, D., Kung, P., and Goldstein, G. (1981). Leuk. Res. 5, 281. Habeshaw, J. A,, Catley, P. F., Stansfeld, A. G . , et al. (1979,). Br. J. Cancer 39, 566. Habeshaw, J . A , , Catley, P. F., Stansfeld, A. G., and Brearley, R . L. (19791)).Br. J . Cancer40, Habeshaw, J. A., Bailey, D., Stansfeld, A. G., and Greaves, M. F. (1983). Br. J. Cancer 47, 327. Hanaoka, M. (1982). GANN Monogr. Cancer Res. 28, 1. Hanaoka, M.,Sasaki, M.. Matsumoto, H., Tankawa, M., Yarnabe, H., Tomimoto, K., Tasaka, C., Fujiwara, H., Uchiyama, T.. and Takatsuki, K . (1979). Acta Pathol. J p n . 29, 723. Hattori, T., Uchiyatna, T., Toibana, T., Takatsuki, K., and Uchino, T. (1981). Blood 58, 645. Haynes, B. F., Eisenbarth, G., and Fauci, A. (1979). Proc. Natl. Acad. Sci. U . S . A . 76, 5829. Haynes, B. F., Metzgar. R . S., Minna, J. D., and Bunn, P. A. (1981). N . Engl. J. Med. 304, 1319. Haynes, B. F., Hensley, L. L., and Jegasothy, B. V. (1982). Blood 60, 463. Heinz, R., Stacher, A., Theml, H., Pralle, H.. Bremer, K., Brunswicker, F., Burkert, M., Common, H., Fiille, H. H., Griineisen, A , , Hermann, F., Leopold, H., Lifters, R . , Nowicki, L., Niirnberger, R., Rengs-Hausen, H., Ruhl, U., Schoengen, A , , Schmidt, M., Wirthmuller, R . , and Schwarze, E.-W. (Kiel Lymphoma Study Group). (1979). Meet. Znt. Soc. Haematol. Eur. Afr. Dio., 5th. Zfamhurg Aug. 26-31. Helbron, D., Brittinger, G., and Lennert, K. (1979). Blut 39, 117. Hinuma, Y. (1982). GANN Monogr. Cancer Res. 28, 221. Hinuma, Y., Nagata, K., Hanaoka, M., Nakai, M., Matsumoto, T., Kinoshita, K.-I., Shirakawa, S., and Miyoshi, I. (1981). Proc. Natl. Acad. Sci. U . S . A . 78, 6476. Hinuma, Y . , Komoda, H., Chosa, T., Kondo, T . , Kobakura, M., Takenaka, T., Kikuchi, M., Ichimaru, M., Yonoki, K., Sato, I., Matsuo, R., and Hanaoka, M. (1982). Znt. J . Cancer 29, 631.

Hoffmann-Fezer, G., Rodt, H., Eulitz, M., and Thierfelder, S. (1976). J. Zmmunol. Methods 13, 261. Hoffinann-Fezer, G., Pielsticker, K., Rodt, H., and Thierfelder, S. (1978). Verh. Dtsch, Ges. Pathol. 52, 371. Hoffmann-Fezer, C., Thierfelder, S., Pielsticker, K., and Rodt, H. (1979). Leuk. Res. 3, 297. Hoffmann-Fezer, G., Loehrs, U., Rodt, H. V., and Thierfelder, S. (1981). Cell Tissue Res. 216, 361.

144

HARALD STEIN ET AL.

Hoffman, F. M., Smith, D., and Hocking, A. (1982). Clin. E x p . Immunol. 49, 401. Hopper, J. E., and Haren, J. M. (1980). Clin. Zmmunol. Irnmunopathol. 17, 43. Howard, J. C., Hunt. S. V., and Gowans, J. L. (1972).J. E x p . Med. 135, 200. Isaacson, P. (1979).J. Clin. Pathol. 32, 802. Isaacson, P., and Wright, D. H. (1978). Hum. Pathol. 9, 661. Isaacson, P., Wright, D. H., and Jones, D. B. (1983). Cancer 51, 80. Janossy, G., and Prentice, H. G. (1982). In “Clinics in Haematology” (W. Wolvey, ed.), pp. 631-660. Saunders, Philadelphia, Pennsylvania. Janossy, G., Bollum, F. J., Bradstock, K. F., and Ashley, J . (1980). Blood 56, 430. Jansen, J., Schuit, H. R. E., Meijer, C. J. L. M., Le Bien, T. W., Hijmans, W., and Kersey, J. H. (1981). In “Leukemia Markers” (W. Knapp, ed.), pp. 179-192. Academic Press, New York. Kamoun, M., Martin, P. J., Hansen, J. A,, Brown, M. A., Siadak, A. W., and Nowinski, R. C. (1981).J. E x p . Med. 153, 207. Kerndrup, G., and Pallesen, G. (1981). Scand. J. Haematol. 27, 99. Kikuchi, M., Mitsui, T., Matsui, N., Sato, E., Tokunaga, M., Hasui, K., Ichimaru, M., Kinoshita, K . , and Kamihira, S. (1979).Jpn. J. Clin. Oncol. 9 (Suppl.), 407. Kikuchi, M., Mitsui, T., Eimoto, T., Toyooka, R., and Nishiuchi, M. (1982). GANN Monogr. Cancer Res. 28, 37. Knapp, W., Majdic, O., Bettelheim, P., and Liszka, K. (1982). Leuk. Res. 6, 137. Korsmeyer, S. J., Hieter, P. A., Ravetch, J. V., et al. (1981). In “Leukemia Markers” (W. Knapp, ed.), pp. 85-97. Academic Press, New York. Korsmeyer, S. J., Greene, W. C., Cossman, J., Hsu, S. M., Jensen, J. P., Neckers, L. M., Marshall, S. L., Bkhashi, A., Depper, J. M., Leonard, W. J., J a e , E. S., and Waldmann, T. A. (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 4522. Kung, P. C., Berger, C. L., Goldstein, G., et al. (1981). Blood 57, 261. Leder, L.-D. (1965). Klin. Wochenschr. 43, 795. Lennert, K. (1961).In “Handbuch der speziellen pathologischen Anatomie und Histologie” (0. Lubarsch, F. Henke, R. Roessle, and E. Uehlinger, eds.), Band I, Dritter Teil “Lymphknoten. Diagnostik in Schnitt und Ausstrich: Bandteil A: Cytologie und Lymphadenitis.” Springer-Verlag, Berlin and New York. Lennert, K. (1964).In “Pathologie der Halslymphknoten,” pp. 83-96. Springer-Verlag, Berlin and New York. Lennert, K., and Mohri, N. (1978).In “Handbuch der Speziellen Pathologischen Anatomie und Histologie” (E. Uehlinger, ed.), Band I, Dritter Teil “Malignant Lymphomas other than Hodgkin’s Disease” (K. Lennert et al.), pp. 111-469. Springer-Verlag, Berlin and New York. Lennert, K., and Niedorf, H. R. (1969). Virchows Arch. Abt. B 4 , 148. Lennert, K., Mohri, N . , Stein, H., and Kaiserling, E. (1975). Br. J. Haematol. 31 (Suppl.), 193. Lennert, K., Stein, H., Feller, A. C., and Gerdes, J. (1982). In “B and T Cell Tumors” (E. S. Vitetta, ed.), pp. 9-28. Academic Press, New York. Lennert, K., Collins, R. D., and Lukes, R. J. (1983). Histopathology 7, 549. Levine, A. (1981). Proc. Int. Lymph. Conf., 2nd, Athens. Lukes, R. J., and Collins, R. D. (1974). Recent Results in Cancer Res. 46, 18. Lukes, R. J., and Collins, R. D. (1975). Br. J. Cancer 31 (Suppl. 11), 1. Lukes, R. J., Taylor, C. R., Parker, J. W., Lincoln, T. L., Pattengale, P. K., andTindle, B. H. (1978). Am. J. Pathol. 90, 461. McMichael, A. J . , Pilch, J. R., Galfre, G., et al. (1979). Eur. J. Imniunol. 9, 205. Mann, R. B., JaKe, E. S., Brylan, R. C., Nanba, K., Frank, M. M., Ziegler, J. L., and Berard, C. W. (1976). N . Engl. J . Med. 295, 685.

H U M A N LYMPHOMA

145

Mann, R. B . , Jaffe, E. S., and Berard, C. W. (1979). Am. J. Pathol. 94, 103. Mason, D. Y., Bell, J. I., Christensson, B., and Biberfeld, P. (1980). Zmmunology 40, 235. Meusers, P., Bartels, H., Brittinger, G . , Common, H . , Diihmke, E., Fiille, H. H., Gunzer, U., Heinz, R., Kiinig, E., Musshoff, K. A., Pralle, H., Schmalhorst, U., Theml, H., Kriiger, G. R. F., and Lennert, K. (1979). N . Engl. J. Med. 301, 384. Miyoshi, I., Kubohishi, I., Yoshimoto, S., Akagi, T., Ohtsuki, Y., Shiraishi, Y., Nagata, K., and Hinuma, Y. (1981). Nature (London) 294, 770. Muller-Hermelink, H. K., Stein, H., Steinmann, G., and Lennert, K. (1983). Am. J . Surg. Pathol., 7, 849. Nadler, L. N . , Ritz, J., Reinherz, E. L., and Schlossmann, S. F. (1981). In “Leukemia Markers” (W. Knapp, ed.), pp. 3-17. Academic Press, New York. Naiem, M . , Gerdes, J . , Adulaziz, Z., Stein, H., and Mason, D. Y. (1983).J. Clin. Pathol. 36, 167. Pallesen, G . , Madsen, M . , and Pedersen, B. B. (1979). Scand. 1. Haenmtol. 22, 407. Pileri, S., Brandi, G., Rivano, M . T., Govoni, E., and Martinelli, G . (1982). Tumori 68, 543. Pinkus, 6. S., Said, J. W., and Hargreaves, H. (1979). Am. J. Cfin. Pathol. 72, 540. Poiesz, B. J., Ruscetti, F. W., Gazdar, A. F . , Bunn, P. A , , Hinna, J. D., and Gallo, R. C. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 7415. Poiesz, B. J., Ruscetti, F. W., Reitz, M . S., Kalyanaraman, V. S., and Gallo, R. C. (1981). Nature (London) 294, 268. Posnett, D. N., Chiorazzi, N., and Kunkel, H. 6. (1982).J. Clin.Inoest. 70, 254. Preud’Homtne, J. L., and Seligmann, M . (1972). Blood 40, 777. Rappaport, H. (1966). “Atlas of Tumor Pathology,” Sect. 3, Fasc. 8. U.S.Armed Forces Institute of Pathology, Washington, D.C. Reinherz, E. L., Nadler, L. M., Rosenthal, D. S . , et al. (1979). Blood 53, 1066. Reinherz, E. L., Kung, P. C., Goldstein, G., Levey, R. H., and Schlossman, S. F. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 1588. Rieber, E. P., Hadam, M. R., Linke, R. P., et al. (1979). Br. J. Haematol. 42, 175. Ritz, J . , Lee, M. N . , Bhan, A. K., Notis-McConarty, J.. Pesando, J. M., and Schlossman, S. F. (1981a). Blood 58, 648. Ritz, J . , Pesando, J. M., Sallan, S. E., Clavel, L. A., Notes-McConarty, J . , Rosenthal, P., and Schlossman, S. F. (1981b). Nature (London) 283, 583. Rumke, H. C., Miedema, F., ten Berge, I. J. M., et al. (1982). J. Itnmunol. 129, 419. Schlimok, K. G., Thiel, E., Rieber, E. P., et al. (1982). Blood 59, 1157. Schwab, U., Stein, H., Gerdes, J , , Lemke, H . , Kirchner, H . , Schaadt, M., and Diehl, V. (1983). Nature (London)299, 65. Schwarting, R., Stein, H., and Wang. (1984). Blood, in press. Shevach, E. M., Jaffe, E. S., and Green, J. (1973). Transplant. Reo. 16, 3. Stein, H. (1976). Zmtnun. Znfekt. 4, 52 and 95. Stein, H. (1978). In “Handbuch der Speziellen Pathologischen Anatomie und Histologie” (E. Uehlinger, ed.), Band I, Dritter Teil “Malignant Lymphomas other than Hodgkin’s Disease” (K. Lennert, et al.), pp. 529-656. Springer-Verlag, Berlin and New York. Stein, H., and Muller-Hermelink, H. K. (1977). Br. J. Haematol. 36, 225. Stein, H . , Lennert, K., and Parwaresch, M. R. (1972). Lancet 11, 855. Stein, H., Kaiserling, E., and Lennert, K. (1973a). In “Leukamien und maligne Lymphome” (A. Stacher, ed.), pp. 195-201. Urban & Schwarzenberg, Munich. Stein, H., Kaiserling, E., Lennert, K., and Panvaresch, M. R. (197313). Klin. Wochenschr. 51, 389. Stein, H., Kaiserling, E., and Lennert, K. (1974a). Znt. Congr. Znt. Acad. Pathol., loth, Hamburg Ahstr. No. 122, p. 58.

146

HARALD STEIN ET AL.

Stein, H., Kaiserling, E . , and Lennert, K. (1974b). In “Modern Trends in Human Leukemia” (R. Neith, R. C. Gallo, S. Spiegelmann, and J. R. F. Stohlman, eds.), Vol. 14, pp. 97-106. Lehmanns, Munich. Stein, H., Kaiserling, E., and Lennert, K. (1974~).Virchows Arch. Abt. A 364, 51. Stein, H., Kaiserling, E., and Lennert, K. (1975). Meet. Eur. Afr. Dio. Int. Soc. Haematol., 3rd. London Abstr. No. 19, p. 5. Stein, H., Petersen, N., Gaedicke, G., Lennert, K., and Landbeck, G. (1976). Int. 1. Cancer 17, 292. Stein, H., Siemssen, U., and Lennert, K. (1978). B r . 1. Cancer 37, 520. Stein, H., Bonk, A,, Tolksdorf, G., Lennert, K., Rodt, H., and Gerdes, J. (1980).]. Histochern. Cytochem. 28, 746. Stein, H., Mason, D. Y., Gerdes, J., Ziegler, A., Naiem, M., Wernet, P., and Lennert, K. (1981a).In “Leukemia Markers” (W. Knapp, ed.), pp. 99-108. Academic Press, New York. Stein, H., Staudinger, M., Tolksdorf, G., and Lennert, K. (1981b).]. Cancer Res. Clin. Oncol. 101, 29. Stein, H., Tolksdorf, G., and Lennert, K. (1981~).Pathol. Res. Pract. 171, 197. Stein, H., Gerdes, J., and Mason, D. Y. (1982a). Clin. Haeratol. 11, 531. Stein, H., Gerdes, J., Schwab, U., Lemke, H . , Mason, D. Y . , Ziegler, A,, Schienle, W., and Diehl, V. (1982b). Int. 1. Cancer 30, 445. Stein, H., Lennert, K., Mason, D. Y., Gerdes, J., Ziegler, A., Naiem, M., and Wernet, P. (1982~).In “Advances in Comparative Leukemia Research 1981” (1). S. Yohn and J. R. Blakeslee, eds.), pp. 479-486. Elsevier, Amsterdam. Stein, H., Mason, D. Y., Gerdes, J., O’Connor, N., Wainscoat, J., Pallesen, G., Gatter, K., Falini, B., Delsol, G . , Lemke, H., and Lennert, K. (1984). Submitted. Strauchen, J. A,, Young, R. C., De Vita, V. T., Jr., Anderson, T., Fantone, J. C., and Berard, C. W. (1978). N . Engl. 1. Med. 299, 1382. Suchi, T., Tajima, K., Nanba, K., Wakasa, H., Mikata, A,, Kikuchi, M., Mori, S., Watanabe, S . , Mohri, N., Shamoto, M., Harigaya, K., Itagaki, T., Matsuda, M., Kirino, Y., Takagi, K., and Fukunaga, S. (1979). Acta Pathol. Jpn. 29, 755. Takatsuki, K., Uchiyama, T., Sagawa, K., and Yodoi, J. (1977). In “Topics in Hematology” (S. Seno, F. Takaku, and S. Irino, eds.), pp. 73-77. Excerpta Medica, Amsterdam. Takatsuki, K., Uchiyama, T., Ueshima, Y., Hattori, T., Toibana, T., Tsudo, M., Wano, Y., and Yodoi, J. (1982). GANN 28, 13. Tax, W. J. M., Willems, H. W., Kibbelaar, M. D. A., et al. (1981). Prot. Biol. Fluids 29, 701. Taylor, C. R., and Burns, J. (1974).1. Clin. Pathol. 27, 14. The Non-Hodgkin’s Lymphoma Pathologic Classification Project (1982). Cancer 49, 2112. The T- and B- Cell Malignancy Study Group (1981).]pn. 1. Clin. Oncol. 11, 15. Thiel, E. , Rodt, H., Huhn, D., Netzel, B., Grosse-Wilde, H ., Ganeshaguru, K., and Thierfelder, S. (1980). Blood 56, 759. Thiel, E. , Rodt, H., Huhn, D., Stiinkel, K., Gutensohn, W., and Thierfelder, S. (1981). In “Leukemia Markers” (W. Knapp. ed.), pp. 471-474. Academic Press, New York. Thien, S. L., Catovsky, D . , Oscier, D., et al. (1982). Clin. E x p . Imniunol. 47, 670. Thomas, Y., Rogozinski, L., Irigoyen, 0. H., et al. (1981).1. Exp. Med. 154, 459. Thomas, Y., Rogozinski, L., Irigoyen, 0. H . , et al. (1982). 1. Itnntunol. 128, 1386. Toben, H. R., and Smith, R. G. (1977). Clin. Exp. Immunol. 27, 292. Totterman, T. H., Nilson, K., Glaesson, L., Simonsson, B., and Aman, P. (1981). Hum. Lymphocyte Dqfer. 1, 3. Uchiyama, T., Yodoi, J., Sagawa, K., Takatsuki, K., and Huchino, H. (1977). Blood 50, 481. Van der Valk, P., Meijer, C. J. L. M., Willemse, R., van Oosterom, A. T., Spaander, P. J., and te Velde, J. (1982a). Thesis, University of Leiden.

H U M A N LYMPHOMA

147

Van der Valk, P., Ruiter, I). J . , den Ottolander, 6.J., te Velde, J., Spaatider, 1’. J., and Meijer, C. J. L. M. (198211). Histopathology 6, 269. Verhi, W., Greaves. M . F . , Schneider, C . , Koubek, K., Janossy. G . , Stein, H . , Kung, P., and Goldstein. G. (1982). Eur. J. Imnunol. 12, 81. Vogler, L. B., Christ, W. M . , Bockmann, 1). E . , et a / . (1978). N. Engl. J. Med. 298; 872. Vollenweider, R . , and Lennert, K . (1983). Virchotcs Arch. Cell Pathol. 44, 1. Waldron, A., Leech, H., Glick, D., et nl. (1977). Cancer 40, 1604. Warlike, R . , and Levy, R. (1978). N. Etigl. J. M e d . 298, 481. Warnke, R., Miller, R., Grogan, T., Pederson, M . , Dilley, J., and Levy, H. (1980). N . Engl. J. M e d . 303, 293. Warlike, R. A,, Gatter, K . C . , Falini, B . , Hildreth, P.,Woolston, R.-E., Pulford, K . , Cordell, J. L., Cohen, B . , De Wolf-Peeters, C., and Mason, D. Y. (1983). N. Engl. J. Med. 309, 1275. Weinberg, I). S., and Pinkus, G. S. (1981). Am. J. Clin. Pathol. 76, 190. Wright, P. G., and Isaacson, P. 6. (1983). In “Biopsy Pathology of the Lymphoreticular System” (L. S. Gottlieb, A. M . Neville, and F. Walker, eds.), pp. 268-284. Chapman & Hall, London. Yam, L. T., Li, C. Y., and Lam, K . W. (1971). N . Engl. J. Med. 284, 357. Yamada, Y. (1983). Blood 61, 192. Ziegler, A , , Stein, H., Muller, C . , and Wernet, P. (1981).I n “Leukemia Markers” (W. Knapp, ed.), pp. 113-116. Academic Press, London. Ziegler, A , , Uchaiiska-Ziegler, B . , Zerithen, J . , and Weriiet, P. (1982). Sumztic Cell Genet. 8, 775. Ziegler, A,, ef 01. (1984). I n preparation.

This Page Intentionally Left Blank

INDUCED DIFFERENTIATION OF MURINE ERYTHROLEUKEMIA CELLS: CELLULAR AND MOLECULAR MECHANISMS Richard A. Rifkind, Michael Sheffery, and Paul A. Marks DeWitt Wallace Research Laboratory and the Sloan-Kettering Division, Graduate School of Medical Sciences, Memorial Sloan-Kettering Cancer Center. New York. New York

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Terminal Cell Division . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Characteristics of Inducer-Mediated Differentiation . . . . . . . . . . . . . . . . . . . . . . B. Commitment to Terminal Cell Division C. Commitment to Terminal Cell Division D. Commitment Involves a Change for Which Cells Retain a “Memory” . . . . . . E. Protein p53 and the Onset of Terminal Cell Division. . . . . . . . . . . . . . . . . . . . . 111. Gene Expression, DNA St during Induced Differentiation A. Murine Globin Gene 1 B. Transcriptional Regulation of Globin Genes during Development and Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Posttranscriptional Regulation of Globin Gene Expression . . . . . . . . . . . . . . . .

149 151 151

IV. Differentiation and the Cell Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

160

‘on . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..........................

162 162 163

References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

153 154

156 157

I. Introduction Erythropoiesis constitutes one of the best characterized developmental lineages in higher organisms; the progressive developmental stages have been identified and aspects of the control mechanisms regulating development, including the effects of hemopoietic growth factors, are increasingly well defined (Marks and Rifkind, 1978). Terminal cell differentiation in this lineage has been described in terms of morphological and biochemical changes, as well as events at the molecular level, including the generation of differentiation-specific mRNAs (globin mRNAs) and other differentiationspecific products (Marks and Rifkind, 1978; Terada et al., 1972; Ramirez et al., 1975; Sassa, 1980). Molecular probes for the definition of specific gene expression in this lineage are readily available. The murine erythroleukemia cell (MELC), a virus-transformed precursor approximating, in terms of its developmental potential, the colony forming 149 ADVANCES IN CAh’CER RESEARCH. VOL 42

Copyright Q 1984 by Academic Press. Inc All right\ of rr(lroduction in any form reserved ISRN n-18-o(IFc14z-4

150

R I C H A R D A. R I F K I N D ET AL.

unit for erythropoiesis (CFUe) stage of the erythroid lineage, provides yet additional advantages for the study of developmentally significant regulatory mechanisms during erythroid cell differentiation (Marks and Rifkind, 1978). These transformed cells, blocked in their differentiation at a particular moment in the developmental history of the erythroid cell lineage, can be cloned and retain their normal developmental potentiality. This provides a unique opportunity for defining the cellular and molecular phenotype of a cell arrested at a particular developmental stage in its lineage. Indeed, as will be developed in this article, a number of features of chromatin and DNA structure, in the globin gene domains, have been identified in MELC which appear to characterize this developmental stage of the erythroid cell lineage. MELC can be induced, by exposure to any of a variety of chemical agents, to initiate their developmental program in a fashion which, by all tests so far applied, appears similar to the normal developmental history of nontransformed erythroid precursor cells. These cells provide an opportunity to examine, at a cellular and molecular level, events which accompany, and which may regulate, the initiation of terminal cell differentiation, including the expression of differentiation-specific genes and the initiation of terminal cell divisions. Among these, as will be developed below, is a complex set of changes of chromatin structure in the globin gene domains, which appear to be significant in the induced expression of the genes during differentiation. Additional advantages of the MELC system include the availability of different inducing agents, capable of initiating different patterns of expression of the characteristics of induced differentiation (Marks and Rifkind, 1978; Tanaka et al., 1975; Reuben et al., 1976; Nude1 et al., 1977a,b), and the availability of' variant cell lines selected for their resistance to various inducing agents and which are providing powerful tools for exploring relationships among the molecular changes which accompany induced differentiation (Ohta et al., 1976). Many aspects of the biological events which occur during induced MELC differentiation have been explored in a number of laboratories. These include, for example, factors determining morphogenetic changes (Volloch and Housman, 1982), changes in cell membrane properties, including ion fluxes (Mager and Bernstein, 1978a,b; Smith et al., 1982), cyclic nucleotide metabolism (Gazitt et al., 1978b), iron transport and heme synthesis (Sassa, 1980), and others. In this article we shall concentrate on three features of induced MELC differentiation dealing with the regulation of expression of developmentally specific genes and the relationships between differentiation and the cell cycle.

1. Studies concerning the mechanisms regulating the initiation of termi-

M U R I N E ERYTHROLEUKEMIA CELLS

151

nal cell division (commitment), which is a characteristic feature of the development of both normal and transformed erythroid cells. 2. The nature of mechanisms regulating the expression of differentiationspecific genes during induced differentiation; these studies, in large part, address the control of globin gene expression at the transcriptional and posttranscriptional levels, both of which may be modulated during induced differentiation. 3. Studies on the relationships between the cell division cycle, DNA synthesis, globin gene replication, and induced cell differentiation.

II. Terminal Cell Division The nature of the factors regulating cell growth in eukaryotic cells remains one of the central unresolved issues in biology today. One requirement for such a study is a homogeneous cell population which can be readily manipulated with respect to the transition from unlimited cell growth to the terminal cell divisions characteristic of many differentiating cell lineages. MELC meet this criterion and provide one attractive model for study of this problem.

A. CHARACTERISTICS OF INDUCER-MEDIATED DIFFERENTIATION MELC can be maintained in suspension culture for an essentially unlimited number of passages under appropriate conditions (Marks and Rifkind, 1978) and can be induced by a variety of agents (Table I), including TABLE 1 INDUCERS OF MELC E R Y T I ~ R ODIFFERENTIATION^ ID Polar compounds Fatty acids DNA intercalators Modified bases Phosphodiesterase inhibitors Ion-flux agents Physical agents Posttranscription-acting agent

Dirnethyl sulfoxide (MeZSO) Hexamethylene bisacetamide (HM BA) Butyric acid Actinornycin Azacytidine Methylisoxanthine Ouabain UV, X ray Hemin

a This is an abbreviated tabulation of agents which have been demonstrated to induce MELC to express characteristics of erythroid differentiation. A more complete list has been presented elsewhere (Marks and Rifkind, 1978; see also Mager and Bernstein, 1978b; Creusot et al., 1982; Ross and Sautner, 1976; Reuben et al., 1980).

152

RICHARD A. RIFKIND ET AL.

Me,SO, HMBA, and butyric acid, to express characteristics of erythroid cell differentiation including the onset of terminal cell division, as well as the expression of differentiation-specific genes.

B. COMMITMENT TO TERMINAL CELLDIVISION One requirement for studying regulation of the onset of terminal cell division is an assay, at the single cell level, for the loss of capacity for cell proliferation. Such an assay has been developed for MELC (Gusella et al., 1976; Fibach et al., 1977). MELC are exposed to inducer in suspension culture, aliquots are removed after appropriate intervals and grown as a single cell suspension in semisolid medium in the absence of inducer. The growth and developmental potential of each cell is scored by examining the progeny colonies after several days. Large colonies which do not stain for hemoglobin (benzidine-negative colonies) are the progeny of uncommitted MELC. Small colonies (less than 32 cells) of benzidine-positive cells are the progeny of cells which had become committed to their program of terminal cell division (and their differentiation as erythroid cells). A small proportion of cells give rise to “mixed” colonies consisting of both benzidine-reactive and benzidine-negative cells. Analogous patterns of commitment have been observed in studies of other cell lines capable of induced terminal cell differentiation, such as the rat myoblast (Nadal-Ginard, 1978), human keratinocyte (Rheinwald, 1979), and mouse lymphocyte (Milner, 1977). Commitment, defined as the ability to express this transition from a stage of unlimited proliferation to the stage of terminal division, is dependent on the nature, as well as on the duration of exposure to and concentration of the inducer (Reuben et al., 1980). Commitment can be detected as early as 12 hr in culture with certain inducers; almost 100% of the population can be committed to terminal cell division by 48 hr in culture with HMBA, probably the most effective inducer identified (Reuben et al., 1980).

C. COMMITMENT TO TERMINAL CELL DIVISION INVOLVES A MULTISTEPPROCESS Several lines of evidence suggest that the commitment of MELC is a multistep process. Inducer-mediated commitment to terminal cell division exhibits different kinetics with different inducers (Reuben et al., 1980; Nude1 et al., 1977b; and Marks et al., 1979). Variants of MELC have been developed which show altered patterns of response to inducers with respect to terminal cell division, including variants which fail to exhibit inducermediated commitment although they express other characteristics of ery-

M U R I N E ERYTHROLEUKEMIA CELLS

153

throid cell differentiation (Harrison, 1977; Marks et al., 1983; Eisen et al., 1978; Pragnell et al., 1980). Studies with a number of inhibitors of induced differentiation provide additional evidence. Levenson et al. (1979), examining the response of differentiating MELC to the nucleotide analog, cordycepin, which inhibits accumulation of polyadenylated mRNA, found evidence that in Me,SOmediated commitment there is a step which is sensitive to inhibition by this agent. The rate limiting step for commitment appears, however, to be insensitive to inhibition by cordycepin but sensitive to inhibition by cycloheximide, an inhibitor of protein synthesis (Levenson and Housman, 1979). Based upon work from our laboratory, analyzing patterns of induction in normal and variant MELC cell lines, it has been suggested that there is a precommitment (“initiation”) phase during which certain inducer-mediated metabolic changes occur, including, perhaps, alterations in membrane permeability, cell volume, and CAMP concentration (Marks and Rifkind, 1978; Eisen et al., 1978; Gazitt et al., 1978a,b; Mager and Bernstein, 1978a,b; Smith et al., 1982). This is followed by a period during which changes occur which appear to involve the accumulation of a factor or factors which may be responsible for the commitment to terminal cell division and, perhaps, for the expression of genes characteristic of terminal differentiation (Chen et al., 1982). There then follow those changes which characterize expression of the terminal erythroid cell phenotype (Marks and Rifkind, 1978; Harrison, 1977; Marks et al., 1982), including accumulation of newly synthesized globin mRNA (Ross et al., 1972), the sequential induction of heme-synthesizing enzymes (Sassa, 1980), the progressive loss of the capacity for cell division (Gusella et al., 1976; Fibach et al., 1977), and other changes (Boyer et al., 1972; Ostertag et al., 1972; Eisen et al., 1977). A CHANCE FOR WHICHCELLS D. COMMITMENT INVOLVES RETAIN A “MEMORY”

Tumor promoters, such as 12-0-tetradecanoylphorbol-13-acetate(TPA), are potent inhibitors of inducer-mediated differentiation of MELC (for review see Marks et al., 1982) as well as other cell lineages (Diamond et al., 1977; Ishii et al., 1978). When MELC are exposed to both HMBA and TPA, the phorbol ester suppresses the onset of terminal cell division as well as the accumulation of newly synthesized a-and (3-globin mRNA and hemoglobin (Marks et al., 1982). If MELC are transferred from medium containing both inducer (HMBA) and inhibitor (TPA) into medium without either agent, such cells retain, for a period of time, a “memory” of their prior exposure to the inducer (Fibach et al., 1979). Although the nature of this memory, at a molecular level, remains to be defined, the memory for prior exposure to

154

RICHARD A. R I F K I N D ET AL.

inducer can persist for more than one cell cycle, and may reflect the accumulation of that factor (or factors) noted above. The phenomenon of memory for HMBA-mediated commitment has also been demonstrated using another potent inhibitor of induced differentiation, the glucocorticoid, dexamethasone (Santoro et al., 1978; Tsiftsoglou et al., 1979; Osborne et a l . , 1982). Dexamethasone suppresses expression of inducer-mediated MELC terminal cell division, as well as inhibiting the accumulation of cytoplasmic a- and P-globin mRNAs, globins, and hemoglobins (Chen et al., 1982; Lo et a l . , 1978; Scher et al., 1978; Mierendorfand Mueller, 1981). We have demonstrated that dexamethasone inhibits commitment to terminal cell division at a stage in the multistep process of inducer-mediated MELC commitment which is not rate limiting for this process (Chen et a l . , 1982). It has been shown, as well, that dexamethasone inhibits the accumulation of nuclear globin mRNA sequences, suggesting that the steroid acts to block the inducer-mediated increase in globin gene expression (Marks et al., 1982). However, Shaul et al. (1981)have suggested that TPA inhibits the transport of globin mRNA to the cytoplasm during culture of MELC with Me,SO, implicating a step involved in the transport or stability of mRNA. It remains to be determined whether inhibition of inducer-mediated MELC commitment by TPA and dexamethasone reflects an action of these agents at a transcriptional or a posttranscriptional level and how the inhibitors’ effects on globin gene expression are related to their effects in inhibiting commitment to terminal cell division.

E. PROTEINp53 A N D

THE

ONSETOF TERMINAL CELLDIVISION

Little is known of the molecular or biochemical mechanisms which are specifically implicated in the signal for the actual onset of termination of cell division, as noted above. HMBA-mediated MELC commitment to terminal cell division is coordinated with the expression of other markers of erythroid differentiation, among which we have focused our attention on the globin genes. Inducers which initiate commitment to terminal cell division also initiate a complex series of changes in the structure of globin gene chromatin and increase the rate of transcription at the al-and Pmaj-globin gene loci (Sheffery et al., 1982, 1983a; Profous-Juchelka et al., 1983). However, the accumulation of globin mRNA is not itself sufficient to initiate commitment to terminal cell division. Hemin, for example, increases the rate of a-and Pglobin mRNA and hemoglobin accumulation (Ross and Sautner, 1976), but does not induce commitment to terminal cell division (Gusella et al., 1980; Profous-Juchelka et a l . , 1983). On the other hand, imidazole, an inhibitor of heme accumulation, blocks Me2SO-induced accumulation of hemoglobin,

M U R I N E ERYTHROLEUKEMIA C E L L S

155

but does not block commitment of terminal cell division (Gusella et al., 1982). The first manifestation of termination of cell division in MELC appears to be a transient prolongation of the G, phase of the cell cycle which can be detected in the first cell cycle which follows one complete S phase in inducer (Terada et al., 1977). Eventually, after 4-5 cell divisions, induced MELC are permanently arrested in G,. Evidence has accumulated suggesting that synthesis of a labile protein in early G , may control the entry of cells into S phase (Rossow et al., 1979). A nuclear protein (p53), initially recognized by its elevated levels in transformed cells (DeLeo et al., 1979)and by its capacity to bind to SV40 T antigen (Lane and Crawford, 1979; Linzer and Levine, 1979), has been identified as possibly important in determining normal cell cycle progression from G, to S (Milner and Milner, 1981; Mercer et al., 1982). It has also been suggested that p53 protein levels may be regulated in relation to a cell’s status in the differentiation of a cell lineage (Oren et al., 1982; Chandrasekarau et al., 1982). We have recently shown, by microscopic immunofluorescence, flow microfluorimetry, and immunoprecipitation after [35S]methionine incorporation, that during induced differentiation of MELC there is a decrease in p53 synthesis and in the cell content of this protein (Shen et al., 1983). Although the role of p53 protein in MELC or other cells has not been established with certainty, the principal relationship established by these studies is between down-regulation of p53 synthesis and the loss of cellular proliferative capacity. Hemin, which can induce globin gene expression, but does not induce commitment to terminal cell division, does not induce a fall in p53 synthesis or content, while a commitment-resistant cell variant (Rl), which can express certain characteristics of the MELC developmental program (Marks et al., 1983), also fails to decrease its p53 content in response to inducer (Shen et al., 1983). Taken together, these observations suggest that down-regulation of p53 protein is a part of the coordinate program of events which occur during induced MELC differentiation, and may prove to be significant to the onset of termial cell division. Ill. Gene Expression, DNA Structure; and Chromatin Configuration during Induced Differentiation

The murine erythroleukemia cell is a model cell system well suited to the study of mechanisms which control the coordinated expression of sets of unlinked differentiation-specific genes during development. The globin genes provide a paradigm for such a study; induction results in a 10- to 20fold increase in the rate of accumulation of globin mRNAs, regulated to a large degree by increased transcription (Hofer et al., 1982), although post-

156

RICHARD A. R I F K I N D ET

AL.

transcriptional levels of control have also been implicated (Profous-Juchelka et al., 1983). Two types of hemoglobin, Hblnaj (containing a-and P1naj-globinpolypeptides) and Hblnirl (a-and P’”’”-polypeptides) can be found in MELC. In uninduced MELC there is a low but detectable level of Hbmill,reflecting a low level of expression of the a-and plllill-globingenes (Nudel et al., 1977b). Polar inducers, such as M e 2 S 0 and HMBA, initiate accumulation of more HblnaJ than HbInin; the fatty acid inducers (such as butyric acid) induce roughly equal amounts of both proteins, while hemin induces, predominately, accumulation of Hblllill(Nudel et d.,1977b; Curtis et d.,1980).

A. MURINEGLOBINGENEDOMAINS Detailed physical maps of the murine a-and P-globin gene domains have been extensively characterized (Konkel et al., 1978, 1979; Nishioka and Leder, 1979). A 60 kb region of DNA containing the p-globin genes and a 40 kb region containing a-globin genes have been cloned into h-bacteriophage vectors (Leder et al., 1980). Within the 60 kb p domain, located on chromosome 7, seven loci have been identified which correspond to p-globin genes or pseudogenes. Identified loci include (in the 5’ to 3’ order of transcriptional polarity) the pln“J-and P1nill-globingenes. The 40 kb region contains (in the 5’ to 3’ direction) a-embryonic globin genes and the al-and a2-adult globin genes. These are not linked to the (3 domain, and are found on chromosome 11 (Leder et al., 1981). The complete nucleotide sequence of the al-globin gene has been determined (Nishioka and Leder, 1979). Of the several globin “pseudogenes” (which do not appear to be expressed), the best characterized are the a3-and a,-globin genes. These genes are not physically linked to the embryonic, a 1 and a2 genes, and have been located on chromosomes 15 and 17, respectively (Leder et al., 1981). B. TRANSCRIPTIONAL REGULATIONOF GLOBIN GENESD U RI N G DEVELOPMENT A N D DIFFERENTIATION Much evidence suggests that globin gene expression during development and differentiation is regulated at the level of gene transcription (Groudine et al., 1981; Groudine and Weintraub, 1981; Weintraub et al., 1981; Landes and Martinson, 1982; Landes et al., 1982; Villeponteau et al., 1982). Transcriptional regulation of globin gene expression in the mouse has been best characterized in MELC. Using the nuclear RNA chain elongation assay, Hofer et u1. (1982)have demonstrated a 10- to 20-fold increase in plnaj-globin gene transcription during induced differentiation. Transcription initiates

M U R I N E ERYTHHOLEUKEMIA CELLS

157

predominately near the cap site, occurs predominately off the coding strand, and extends approximately 1.5 kb beyond the poly(A) addition site before terminating. Likewise transcription of the al-globin gene increases some 10to 20-fold during induction (Sheffery et al., 1984), initiating at or near the cap site and terminating, apparently, in a region 50 to 250 bp 3' to the putative polyadenylation site. An increase i n a,-globin gene transcription can be detected within 2 to 3 cell cycles (about 36 hr) following exposure to the inducer, HMBA; the rate of transcription continues to increase for at least 48 hr. C. POSTTRANSCRIPTIONAL REGULATION OF GLOBIN G E N EEXPRESSION Whereas HMBA, Me,SO, and other agents induce the program of MELC differentiation in a coordinate fashion, including globin gene expression and coininitment to terminal cell division, it has been demonstrated (see above) that exposure to hemin leads to accumulation of globin mRNA (and hemoglobins) but not to loss of cellular proliferative capacity. It has recently been shown that the increase in globin mRNA content detected in hemin-treated MELC is not accompanied by an increase in globin gene transcription, as measured by the technique of nuclear RNA chain elongation (ProfousJuchelka et ul., 1983). Hemin appears to act by a posttranscriptional effect upon a low but constitutive level of globin gene transcription in uninduced MELC. This effect of heinin may be mediated by changes in the processing, transport, or stability of such transcripts. The significance of this heminmediated, posttranscriptional control mechanism has yet to be determined. It should be noted, however, that induction by HMBA or other agents is characterized by an early and brisk activation of the heme synthetic enzyme system and the accumulation of cellular heme (Ebert and Ikawa, 1974; Sassa, 1976; Fibach et al., 1979).The rate of globin inRNA accumulation in induced MELC may, then, be the product of both transcriptional and posttranscriptional mechanisms of control.

D. THE ROLE o~ CHROMATIN STRUCTURE I N GLOHIN G E N EREGULATION It is likely that changes in chromatin structure, mediated in part at least by chroinosoinal proteins, play an important role in regulating the transcription of specific genes (Elgin, 1981; Weisbrod, 1982). A number of biochemical features distinguish active chromosomal regions from bulk chromatin at both the DNA and protein level. Several of these have been examined in detail with respect to the expression of globin genes during induced MELC differ-

158

RICHARD A . RIFKIND ET AL.

entiation. These include the pattern of DNA methylation, the sensitivity of chromatin to digestion by DNase I, and the appearance of sites or regions that are hypersensitive to digestion by a number of endonucleases including DNase I and S1 nuclease. Certain patterns of chromatin and DNA structure, found in uninduced MELC, do not change upon induction of terminal cell differentiation; these include the pattern of DNA methylation about both the a1and PlllaJ genes and the overall sensitivity of chromatin containing these genes, to DNase I digestion (Sheffery et al., 1982). In addition, a number of DNase I hypersensitive regions can be detected in uninduced MELC, in the region of both the Plnaj and a1 genes (Sheffery et al., 1982, 1983). These include a site located within the second intron (IVS-2) of the plnaj gene, and sites 5’ to the a,-globin gene cap site. Collectively these molecular features appear to be characteristic of a transformed erythroid cell precursor arrested in its development at approximately the CFUe stage of erythroid cell differentiation. This molecular phenotype is stably propagated in the continuously replicating, uninduced MELC; and from a functional point of view appears to characterize an erythroid precursor with at best a low, constitutive level of globin gene transcription. Each of these features of DNA and chromatin structure has been examined in detail during induced MELC differentiation.

I. DNA Methylation The pattern of DNA methylation has been cited as a heritable molecular characteristic which, at least in many instances, distinguishes expressed and unexpressed genes. We have examined the pattern of DNA methylation in the region of the plnaj- and a,-globin genes during HMBA-mediated differentiation (Sheffery et al., 1982). Cytosine methylation in the nucleotide sequence, CCGG, was assayed by the use of the methyl-sensitive isoschizomer-pair of restriction enzymes, MspI and HpaII, and other restriction enzymes. There are relatively few potentially methylated sites in the MELC globin gene domains which can be assayed by these restriction enzymes. Of the sites assayed near the plnaj-globin gene, one site is fully methylated, one partially methylated, and one is unmethylated in uninduced MELC. Most cites, but not all, assayed near the a-globin genes are unmethylated in uninduced cells. No detectable change in the pattern of methylation around either gene was observed during HMBA-mediated differentiation. Globin genes, both a1and pmaJ,in nonerythroid mouse tissues display distinctly more methylation than they do in MELC (S. Einheber and M. Sheffery, unpublished observations). It would appear likely that, within the limits of resolution of this assay, the pattern of globin gene methylation in the erythroid cell lineage is established and stably propagated at a developmental stage in erythropoiesis prior to the stage represented by the MELC.

MURINE ERYTHROLEUKEMIA CELLS

159

2. DNase 1 Sensitivity Studies of the accessibility to DNase I of MELC chromatin, in the globin gene domains, has suggested that the chromatin of uninduced cells is already in a configuration compatible with transcriptional activity (Miller et al., 1978). It has been demonstrated (Sheffery et al., 1982)that the pmaJ-and alglobin gene-associated chromatin regions are distinctly more sensitive to digestion by DNase I than is the Iga (immunoglobulin) gene (which is not expressed by cells in the erythroid lineage), in both uninduced and induced MELC. As in the case of the methylation pattern, it appears that the globin gene domains have been selectively modified during the developmental history of the erythroid lineage, establishing the globin-related chromatin in a potentially active configuration. These changes, as in the case of the methylation pattern, are stably propagated in uninduced MELC.

3. Nuclease Hypersensitivity Sites We have obtained evidence for alterations in chromatin structure which are specifically associated with inducer-mediated activation of globin gene transcription (Sheffery et al., 1982, 1983). During HMBA-induced differentiation, sites displaying a 6- to 10-fold increase in DNase I sensitivity appear in chromatin regions near the 5‘ end of the a 1and pmajgenes. The DNase I hypersensitive site which appears near the pmaJ-globin gene maps to an approximately 200 base pair region in the 5’ flanking region of that gene. That hypersensitivity site which becomes detected near the al-globin gene likewise is located in the region 5’ to the a1cap site. The changes in chromatin structure which are revealed by nuclease probes during induced differentiation are more complex, however, than simply the reconfiguration of sites at the 5’ end of the globin genes. As noted already, it has been demonstrated that there is a DNase I hypersensitive site, located within the second IVS of the pmaJ gene, which can be detected in uninduced MELC. The nuclease hypersensitivity at this site disappears during HMBA-mediated differentiation, replaced by the new hypersensitivity site which lies 5’ to the pmaJcap site. Although this complex change in chromatin configuration normally takes place in a coordinate fashion, just prior to the initiation of globin gene transcription, it has been shown (Sheffery et a l . , 1983) that in an inducer-resistant variant these two changes in chromatin configuration can be dissociated. In the R 1 variant of MELC, which is resistant to the inducing effect of HMBA on globin gene expression and commitment (Marks et al., 1983), HMBA mediates the disappearance of that hypersensitivity site located in IVS-8 but fails to generate the new site, located 5’ to the PmaJcap site, and fails to initiate transcription at the pmaJ gene.

160

RICHARD A. RIFKIND ET A L .

At the a, domain there is also a complex pattern of chromatin reconfiguration that occurs during induced differentiation (Sheffery et al., 1984). The uninduced MELC, overlapping DNase I and S1 nuclease-sensitive sites are detected 5' of the a-globin gene cap site. During induction the nuclease sensitivity of these sites increases and new, nonoverlapping DNase I and S1 sites develop, one approximately 300 base pairs 5' of the a,-globin cap site, and the other mapping to a region virtually coincident with the cap site itself. None of these changes in nuclease sensitivity occurs in the HMBAresistant MELC variant (Rl), suggesting that there may be significant differences between the chromatin-associated events that take place in the aand (3-globin gene domains during induced differentiation. IV. Differentiation and the Cell Cycle

Inducer-related events associated with the cell cycle, with DNA synthesis, and, in particular, with replication of the globin genes appear to be important in the induction of MELC differentiation and accelerated transcription of the globin genes (Gambari et al., 1979; Brown and Schildkraut, 1979; Epner et al., 1981; Levy et al., 1975). Studies in this laboratory have shown that inducing agents must be present during at least one cell cycle to initiate differentiation (Levy et al., 1975)and that late GI-early S may be the period of particular significance for subsequent globin gene expression (Gambari et al., 1978, 1979). Since DNA replication has been implicated in the transition from an inactive to an active chromosome structure in several systems (Groudine and Weintraub, 1981; Holtzer et al., 1972; Dienstman and Holtzer, 1975), it has been speculated that the changes in globin gene chromatin configuration which are associated with enhanced globin gene expression in MELC may be introduced during replication of the genes, and that the presence of the inducer at a restricted time in the cell division cycle may be critical to the reconfiguration of these gene domains. In this context, it has been found that genes whose expression is stimulated during MELC differentiation (the a- and (3-globin genes) are replicated during early S (Epner et al., 1981; Furst et al., 1981). Each of these aspects of induced differentiation is discussed below.

A. THE CELLCYCLEA N D INDUCED DIFFERENTIATION Evidence suggesting a relationship between the cell cycle (possibly DNA synthesis), and the onset of erythroid cell differentiation, particularly the onset of globin gene expression, in normal and transformed erythroid cells, has come from a number of studies. It was demonstrated a number of years ago that one effect of erythropoietin on normal erythropoietin-sensitive tar-

M U H I N E EHYTHROLEUKEMIA CELLS

161

get cells (probably the CFUe) is an acceleration of DNA synthesis which precedes the accumulation of globin messenger RNA (Djaldetti et a l . , 1972; Rainirez et al., 1975). Studies using inhibitors of DNA synthesis have added to this evidence implicating DNA synthesis and the cell cycle in the initiation of globin gene expression during normal erythropoiesis (Marks et al., 1977). Additional evidence has come from studies on MELC differentiation. For example, when cultures are initiated at high cell densities, the time required for cell doubling is prolonged and the extent of differentiation is decreased; under these conditions, the higher the rate of cell proliferation, the greater the proportion of cells induced to differentiate (Rifiind and Marks, 1982). More direct evidence for a role of the cell cycle in induced MELC differentiation comes from studies with synchronized cell populations, employing several independent methods for the induction of synchronized cell growth. Using 2 mM thymidine to inhibit progression of the cells past the G,/S interface, Levy et al. (1975) demonstrated that induction requires exposure to inducing agents as the cells move through at least one cell cycle. This conclusion has been reached, as well, using synchronization by nutrient deprivation (McClintock and Papaconstantinou, 1974) and by the use of a temperature-sensitive cell cycle mutant (Harrison, 1977). Geller et al. (1978) showed that commitment to initiate terminal cell division and other features of the differentiated phenotype is accomplished most quickly by cultures of MELC synchronized in the G, or G, phase of the cell cycle, suggesting that early S phase is important for induction. More recently, using thymidine, thymidine plus hydroxyurea, or centrifugal elutriation for the synchronization of MELC in G, or at the G,/S interface, it was demonstrated that exposure to inducer during one cell cycle resulted in the initiation of accumulation of globin messenger RNA during the subsequent cell cycle, beginning during G,. The critical time in the cell cycle for subsequent initiation of globin messenger RNA accumulation is G, or early S (Gambari et al., 1978, 1979). Finally, using the phorbol ester, an inhibitor of induced differentiation, it was demonstrated (Gambari et al., 1980) that the TPAsensitive period for inhibition of globin synthesis is confined to that cell cycle in the presence of inducer which precedes the onset of globin mRNA accumulation. Evidence suggesting that it is the onset of globin gene transcription, and not merely the accumulation of globin messenger RNA, that is sensitive to a cell cycle-related effect of inducers has been obtained from studies examining the kinetics of globin mRNA accumulation during induction with various inducing agents. When M ELC are induced to differentiate by polar-planar compounds such as HMBA and Me,SO, the first accumulation of globin mRNA is detected approximately 12 hr after the initiation of culture with

162

RICHARD A. RIFKIND ET AL.

inducer (the cell division cycle is approximately 12 hr). On the other hand, when MELC are induced to accumulate globin mRNA by exposure to hemin, mRNA accumulation can be detected in less than 6 hr and under certain circumstances in as little as 2 hr, a time too short for a significant cell cycle requirement (Nude1 et al., 1977b). As already described (see above) the initiation of globin mRNA accumulation by hemin is accomplished by a posttranscriptional effect upon a constitutive low rate of globin gene transcription and is accompanied by no detectable increase in transcription at the globin gene loci. Taken together these observations suggest that the cell cycle dependence of inducer effect is related to the transcription-modulating effects of these agents; an agent which modifies globin gene expression through a posttranscriptional mechanism, such as hemin, functions in a manner which is, apparently, independent of the cell division cycle.

B. GLOBINGENEREPLICATION Taken together, studies using both inducers of differentiation and inhibitors of differentiation have suggested that there is a requirement for the cell cycle for induced differentiation and the critical period may lie in late G, or early S phase. Based upon speculation that reconfiguration of globin geneassociated chromatin during replication in the presence of chemical inducers of differentiation may be critical for induced gene transcription, studies were undertaken to define the time in the cell cycle during which both a-and pglobin genes undergo DNA replication (Epner et al., 1981). For these studies, centrifugal elutriation (based upon cell volume) was employed for cell synchronization in order to avoid artifacts introduced by the use of inhibitors of DNA synthesis. Newly replicated DNA sequences were prepared from synchronized cells cultured for short periods with 5-bromodeoxyuridine, and the BUdR-containing, newly replicated DNA isolated by CsCl gradient centrifugation and analyzed by hybridization with cloned probes for the aand p-globin gene sequences. Both a- and p-globin gene sequences, in MELC, are replicated early in S phase, whereas ribosomal RNA gene sequences are replicated throughout the cell cycle. These observations are consistent with the speculation that gene replication may be an important event, implicated in the reconfiguration of globin genes required for the activation of transcription. V. Summary

Study of inducer-mediated differentiation of murine erythroleukemia cells provides insights into the cellular and molecular mechanisms implicated in cell differentiation. The loss of proliferative capacity is revealed to be a

M U R I N E ERYTHROLEUKEMIA CELLS

163

complex multistep process during which the cells progress through a series of stages, including a precoinmitment “initiation” stage, a stage suggestive of the accumulation of commitment-related factors, and, finally, a stage of expression of the characteristics of the differentiated state. Cell cycle arrest in G, phase of the cell cycle may, in part at least, be related to downregulation of protein p53 synthesis. Expression of induced differentiation is accompanied by an acceleration of transcription at the globin loci, and possibly by posttranscriptional modulation of globin mRNA accumulation, as well. Cells at the stage of erythroid cell development represented by the transformed, differentiation-arrested MELC, have acquired a unique DNA structure and chromatin configuration around the globin genes which distinguish them from other, nonerythroid cells; additional complex changes in chromatin configuration accompany, and probably precede, inducer-mediated acceleration of globin gene transcription during terminal differentiation. Passage through G, and early S phase of the cell cycle, in the presence of inducer, is critical for subsequent globin gene expression and may be important in establishing the chromatin reconfiguration required for gene expression.

ACKNOWLEDGMENTS These studies were supported, in part, by grants from the National Cancer Institute (PO1 CA-31768 and CA-08748) and the Bristol-Myers Cancer Research Program.

REFERENCES Boyer, S. H., Wuu, K. D., Noyes, A. N . , Young, R., Scher, W., Friend, C., Preisler, H., and Bank, A. (1972). Blood 40, 823-835. Brown, E., and Schildkraut, C. (1979).J . Cell. Physiol. 99, 261-278. Chandrasekarau, K., Mora, P. T., Nagarajan, L., and Anderson, W. B. (1982).J . Cell. Physiol. 113, 134-140. Chen, Z. X.,Banks, 1.. Rifkind, R. A,, and Marks, P. A. (1982). Proc. Natl. Acad. Sci. U . S . A . 79, 471-475. Creusot, F., Acs, G . , and Christman, J. K. (1982).J . Biol. C h e m 257, 2041-2048. Curtis, P., Finnigan, A. C., and Rovera, 6. (1980).J . B i d . Chem. 255, 8971-8974. DeLeo, A. B., Jay, G., Apella, E., IhBois, G . C., Law, L. W., and Old, L. J . (1979). Proc. Natl. Acad. Sci. U . S . A . 76, 2420-2424. Diamond, L., O’Brien, T. G., and Rovera, G. (1977). Nature (London)269, 247. Dienstman, S. R., and Holtzer, H. (1975). In “Cell Cycle and Cell Differentiation” 0. Reinert and H. Holtzer, eds.), pp. 1-25. Springer-Verlag. Berlin and New York. Djaldetti, M., Priesler, H., Marks, P. A , , and Rifkind, R. A. (1972).J . B i d . Chem. 247, 731735. Ebert, P. S., and Ikawa, Y. (1974). Proc. SOC. E x p . B i d . Med. 146, 601-604. Eisen, H., Nasi, S., Georgopoulos, C. P., Arndt-Jovin, D., and Ostertag, W. (1977). Cell 10, 689-695.

164

RICHARD A. RIFKIND ET AL.

Eisen, H., Keppel-Bellivet, F., Georgopoulos, C. P . , Sassa, S., Granick, J . , Pragnell, I., and Ostertag, W. (1978). Cold Spring Harbor ConJ Cell Pro16 5, 277-294. Elgin, S. C. R. (1981). Cell 27, 413-415. Epner, E., Rifkind, R. A , , and Marks, P. A. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 30583062. Fibach, E., Reuben, R . C., Rifkind, R. A , , and Marks, 1’. A. (1977).Cancer Res. 37,440-444. Fibach, E., Yamasaki, H., Weinstein, I. B., Marks, P. A., and Rifkind, R . A. (1978).Cancer Res. 38, 3685-3688. Fibach, E., Gambari, R . , Shaw, P. A , , Maniatis, G . , Reuben, R . C., Sassa, S.,Rifind, R. A , , and Marks, P. A. (1979).Proc. Natl. Acad. Sci. U . S . A . 76, 1906-1910. Furst, A , , Brown, E. H . , Braunstein, J. D., and Schildkraut, C. L. (1981).Proc. Natl. Acad. Sci. U.S.A. 78, 1023-1027. Gambari, R . , Terada, M . , Bank, A , , Rifkind, R. A., and Marks, P. A. (1978).Proc. Natl. Acad. Sci. U.S.A. 75, 3801-3804. Gambari, R., Marks, P. A . , and Rifkind, R. A. (1979).Proc. Natl. Acad. Sci. U.S.A. 76, 45114515. Gambari, R., Fibach, E., Rifkind, R. A., and Marks, P. A. (1980). Biochem. Biophys. Res. Conimun. 94, 867-874. Gazitt, Y., Deitch, A. D., Marks, P. A , , and Rifkind, R . A. (197th). E r p . Cell Res. 117, 413420. Gazitt, Y., Reuben, R. C . , Deitch, A. D., Marks, P. A., and Rifkind, R. A. (197%). Cancer Res. 38, 3779-3783. Geller, R., Levenson, R., and Housrnan, I). (1978).J. Cell. Physiol. 95, 213-222. Groudine, M . , and Weintraub, H. (1981).Cell 24, 393-401. Groudine, M., Peretz, M . , and Weintraub, H. (1981). Mol. Cell. Biol. 1, 281-288. Gusella, J., Geller, R., Clarke, B., Weeks, V., and Housman, D. (1976). Cell 9, 221-229. Gusella, J. F., Weil, S. C., Tsiftsoglou, A. S . , Volloch, V., Neurnann, J . R., Keys, C., and Housman, D. E. (1980). Blood 56, 481-487. Gusella, J. F., Tsiftsoglou, A. S., Volloch, V., Weil, S. C., Neumann, J.. and Housman, D. E. (1982).J. Cell. Physiol. 113, 179-185. Harrison, P. R. (1977). Znt. Reo. Biochem Biochem. Cell O f f e r . 15, 227-268. Hofer, E., Hofer-Warbinek. R., and Darnell, J. E., Jr. (1982).Cell 29, 887-893. Holtzer, H., Sanger, J . W., Ishikawa, H., and Strahs, K. (1972). Cold Spring Harbor Symp. Quant. B i d . 37, 549-566. Ishii, D., Fibach, E., Yamasaki, H., and Weinstein, I. B. (1978). Science 200, 556. Konkel, 1). A., Tilghman, S. M . , and Leder, P. (1978).Cell 15, 1125-1132. Konkel, D. A., Maizel, J. V., and Leder, P. (1979). Cell 18, 865-873. Landes, 6. M . , and Martinson, H. G . (1982).J. Biol. Chem. 257, 11002-11007. Landes, G . M . , Villeponteau, B., Prihyl, T. M., and Martinson, H. G . (1982).J. Biol. Chem. 257, 11008-11014. Lane, D. P., and Crawford, L. V. (1979). Nature (London) 278, 261-263. Leder, P . , Hansen, J., Konkel, D., Leder, A , , Nishioka, Y., andTalkington, C. (1980). Science 209, 1336-1342. Leder, A., Swan, D., Ruddle, F., D’Eustachio, P., and Leder, P. (1981).Nature (London)293, 196-200. Levenson, R . , and Housman, D. (1979).J. Cell Biol. 82, 715-725. Levenson, R . , Kerner, J., and Housman, D. (1979). Cell 18, 1073-1078. Levy, J., Terada, M., Rifkind, R. A., and Marks, P. A. (1975).Proc. Natl. Acad. Sci. U.S.A. 72, 28-32. Linzer, D. I., and Levine, A. J. (1979).Cell 17, 43-52.

MURINE ERYTHROLEUKEMIA CELLS

165

Lo, S. C., Aft, R., Ross, J., and Mueller, G. C. (1978). Cell 15, 447. McClintock, R. P., and Papaconstantinou, J. (1974). Proc. Natl. Acad. Sci. U . S . A . 71, 45514555. Mager, D., and Bernstein, A. J. (1978a).J. Cell. Physiol. 94, 275-285. Mager, D., and Bernstein, A. J. (197%). Supramol. Struct. 8, 431-438. Marks, P. A , , and Rifkind, R. A. (1978). Annu. Reo. Biochern. 47, 419-448. Marks, P. A,, Rifkind, R. A., Bank, A,, Terada, M., Maniatis, G., Reuben, R. C., and Fibach, E. (1977). In “Growth Kinetics and Biochemical Regulation of Normal and Malignant Cells” (B. Drewinko and R. M. Humphrey, eds.), pp. 329-346. Williams & Wilkins, Baltimore, Maryland. Marks, P. A , , Rifkind, R. A,, Bank, A , , Terada, M . , Gambari, R., Fibach, E., Maniatis, G., and Reuben, R. C. (1979). In “Cellular and Molecular Regulation of Hemoglobin Switching” (G. Stamatoyannopoulos and A. W. Neinhuis, eds.), pp. 437-455. Grime & Stratton, New York. Marks, P. A , , Rifkind, R. A., Garnbari, R., Epner, E., Chen, Z. X., and Banks, J . (1982). Curr. Top. Cell. Regul. 21, 198-201. Marks, P. A , , Chen, Z. X.. Banks, J., and Rifkind, R. A. (1983). Proc. Natl. Acad. Sci. U . S . A . 80, 2281-2284. Mercer, W. E., Nelson, D., DeLeo, A. B., Old, L. J . , and Baserga, R. (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 6309-6312. Mierendorf, R. C., and Mueller, G. C. (1981).J. Biol. Chern. 256, 6736-6741. Miller, D. M., Turner, P., Nienhuis, A. W., Axelrod, D. E., and Gopulakrishnan, T. V. (1978). Cell 14, 511-524. Milner, J. (1977). Nature (London) 268, 441-442. Milner, J . , and Milner, S. (1981). Virology 112, 785-788. Nadal-Ginard, B. (1978). Cell 15, 855-864. Nishioka, Y., and Leder, P. (1979). Cell 18, 857-882. Nudel, U., Salmon, J. D., Terada, M., Bank, A,, Rifkind, R. A , , and Marks, P. A. (1977a). Proc. Natl. Acad. Sci. U.S.A. 74, 1100-1104. Nudel, U., Salmon, J., Fibach, E . , Terada, M., Rifkind, R. A., and Marks, P. A. (1977b). Cell 12, 463-469. Ohta, Y., Tanaka, M . , Terada, M . , Miller, 0. J., Bank, A., Marks, P. A., and Rifkind, R. A. (1976). Proc. Natl. Acad. Sci. U.S.A. 73, 1232-1236. Oren, M., Reich, N . C., and Levine, A. J. (1982). M o l . Cell. B i d . 2, 443-449. Osborne, H. B., Bakke, A. C., and Yr, J. (1982). Cancer Res. 42, 513-518. Ostertag, W., Melderis, H . , Steinheider, G., Kluge, N . , and Dube, S. (1972).Nature (London) New B i d . 239, 231-234. Pragnell, I. B., Arndt-Jovin, D. J . , Jovin, T. M . , Fogg, B., and Ostertag, W. (1980). Exp. Cell Res. 125, 459-470. Profous-Juchelka, H. R., Reuben, R. C., Marks, P. A , , and Rifkind, R. A. (1983). Mol. Cell. B i d . 3, 229-233. Ramirez, F., Gambino, R., Maniatis, G. M . , Rifkind, R. A , , Marks, P. A., and Bank, A. (1975). J . B i d . Cheni. 250, 6054-6085. Reuben, R. C., Wife, R. L., Breslow, R., Rifkind, R. A , , and Marks, P. A. (1976). Proc. Natl. Acad. Sci. U.S.A. 73, 862-866. Reuben, R. C., Rifkind, R. A,, and Marks, P. A. (1980). Biochim. Biophys. Acta 605, 325-346. Rheinwald, J. G. (1979). Int. Reo. Cytol. Suppl. 10, 25-33. Rifkind, R. A,, and Marks, P. A. (1982).I n “Differentiation In Vitro” (M. M. Yeoman and D. E. S. Truman, eds.), pp. 183-192. Cambridge Univ. Press, London and New York. Ross, J., and Sautner, D. (1976). Cell 8, 513-520.

166

RICHARD A. RIFKIND ET AL.

Ross, J.. Ikawa, Y., and Leder, P. (1972). Proc. Natl. Acad. Sci. U.S.A. 69, 3620-3623. Rossow, P. W., Riddle, V. G. H., and Pardee, A. B. (1979). Proc. Natl. Acad. Sci. U.S.A. 76, 4446-4450. Santoro, M. G., Benedetto, A., and J&e, B. B. (1978). Biochem. Biophys. Res. Commun. 85, 1510-1517. Sassa, S . (1976).J . E r p . Med. 143, 305-315. Sassa, S . (1980). In “In Vivo and In Vitro Erythropoiesis: The Friend System” (G. B. Rossi, ed.), pp. 219-228. Elsevier, Amsterdam. Scher, W., Tsuei, D., Sassa, S., Price, P., Gabelmar, N., and Friend, C. (1978). Proc. Natl. Acad. Sci. U.S.A. 75, 2851-3858. Shaul, Y., Ginzburg, I., and Aviv, H. (1981). Eur. J. Biochem. 114, 591-595. Sheffery, M., Rifkind, R. A,, and Marks, P. A. (1982).Proc. Natl. Acad. Sci. U.S.A. 79, 11801184. Sheffery, M., Rifkind, R. A., and Marks, P. A. (1983). Proc. Natl. Acad. Sci. U.S.A.80, 33493353. Sheffery, M . , Marks, P. A,, and Rifkind, R. A. (1984).J. Mol. Biol. 172, 417-436. Shen, D.-W., Real, F. X., DeLeo, A. B., Old, L. I., Marks, P. A , , and Rifkind, R. A. (1983). Proc. Natl. Acad. Sci. U.S.A. 80, 5919-5922. Smith, R. L., Macara, I. G., Levenson, R., Housman, D., and Contley, L. (1982). J . Biol. Chem. 257, 773-780. Tanaka, M., Levy, J., Terada, M., Breslow, R., Rifkind, R. A., and Marks, P. A. (1975). Proc. Natl. Acad. Sci. U.S.A. 72, 1003-1006. Terada, M., Cantor, L., Rifkind, R. A., Bank, A., and Marks, P. A. (1972). Proc. Natl. Acad. Sci. U.S.A. 69, 3575-3579. Terada, M., Fried, J.. Nudel, U., Rifkind, R. A., and Marks, P. A. (1977). Proc. Natl. Acad. Sci. U.S.A. 74, 248-252. Tsiftsoglou, A. S., Gusella, J. F., Volloch, V., and Housman, D. E. (1979). Cancer Res. 39, 3849-3855. Villeponteau, B., Landes, G. M., Pankvatz, M. J., and Martinson, H. G. (1982).J . Biol. Chem. 257, 11008-11014. Volloch, V., and Housman, D. (1982).J . Cell B i d . 93, 390-394. Weintraub, H., Larsen, A,, and Groudine, M. (1981). Cell 24, 333-344. Weisbrod, S . (1982). Nature (London) 297, 289-295.

PROTONEOPLASIA:THE MOLECULAR BIOLOGY OF MURINE MAMMARY HYPERPLASIA Robert D. Cardiff Department of Pathology, University of California School of Medicine, Davis, California

I . Introduction . . . . . . . . . . . . . 11. Mouse Mammary Tumor Sy 111. The Development of Transplantable Hyperplastic Outgrowth Lines. . . . . . . . . . . . IV. Characterization of Hyperplastic Outgrowth Lines V. Restriction Endonuclease Mapping. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. General Comments.. . . . . . B. MUMTV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. The Molecular Biology of Mouse Mammary Neoplasia . . . . . . . . . . . . . . . . . . . . . . . A. Tumors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Hyperplasias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Origin and Evolution of Mouse Mammary Tumors . . . . VIII. The HAN Is Protoneoplastic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IX. The Role of MuMTV in Mouse Mammary Tumorigenes X. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References ........................................

167 168 170 171 172 172 175 178 178 180 181 185 186 187 188

1. Introduction

The earliest stages of the neoplastic process are commonly regarded as incipient carcinomas, or carcinomas in situ, already committed to a malignant course. An alternate viewpoint is that the early stages consist of tissues which are more susceptible to carcinogenic stimuli but are not neoplastic nor necessarily obliged to progress to malignancy. These conflicting concepts are important because they influence scientists designing experiments and clinicians treating patients (Foulds, 1969). Fortunately, excellent animal models of neoplastic progression exist which form the conceptual basis for any such action (Foulds, 1958, 1969; Farber and Sporn, 1976). The most extensively used model of neoplastic progression is probably the mouse mammary tumor system because the mouse mammary gland contains a distinctive focal epithelial hyperplasia called the hyperplastic alveolar nodule (HAN) which is prone to develop into a mammary carcinoma (Nandi and McGrath, 1973; Cardiff et al., 1977; Medina, 1978). Since the HAN can be isolated and characterized independent of other mammary tissues, its role in mammary tumorigenesis has been thoroughly documented and it is considered the prototypic “preneoplastic” lesion (Cardiff et a l . , 1977). Recent 167 ADVANCES IN CANCER RESEARCH. VOL. 42

Copyright 0 1984 by Academic Press, Inc. All rights of reproduction in any form reserved. ISBN 0-12-006642-4

168

ROBERT D . CARDIFF

technical advances have permitted analysis of this intriguing lesion at a molecular level. The molecular studies have reinforced many of the earlier concepts based upon the more traditional experimental approaches and have provided new insight into mammary tumorigenesis. The purpose of this review is to analyze the recent molecular literature in the context of the biology of the mouse mammary tumor system. Evidence will be presented to support the following conclusions: (1) premalignant mammary hyperplasias are the result of critical alterations in cell DNA, (2) the alterations in DNA lead to a clonal cell proliferation, (3) tumors are direct subclonal descendents of hyperplastic cells which have undergone further alterations in DNA, and (4) the major role of the mouse mammary tumor virus (MuMTV) is in the formation of nodules, not in the formation of tumors. These conclusions indicate that mouse mammary hyperplasias represent the initial neoplastic event and are an essential component of the neoplastic progression. For these reasons, the HAN might be better considered as protoneoplastic, rather than “preneoplastic.”

II. Mouse Mammary Tumor System

Domesticated and feral mice spontaneously develop mammary tumors with an incidence which is dependent upon genetic background, hormonal mileau, infection with MuMTV, and a variety of other factors (Nandi and McGrath, 1973; Gardner and Rasheed, 1982). The tumors are the end result of a stepwise process associated with several types of mammary hyperplasias (Medina, 1978). DeOme (1967) proposed a two step model of this system. First, normal epithelial cells are transformed into nodule cells and emerge as hyperplastic alveolar nodules (HAN) in a process referred to as nodulogenesis (Fig. 1). MuMTV CARCINOGENS

NORMAL

-%

HAN

’

.TUMOR

FIG. 1. A modification of the standard schematic representation of the mouse mammary tumor model (Cardiff et al., 1977). The three types of mammary epithelium represented are normal, hyperplastic (HAN), and tumor. Step 1 is referred to as nodulogenesis and step 2 is referred to as tumorigenesis. The serial transplanted HAN, referred to as a hyperplastic outgrowth (HPO), is also represented in its position relative to the other tissues. MuMTV and other carcinogens were thought to act in both nodulogenesis and carcinogenesis.

PROTONEOPLASIA

169

Second, nodule cells are transformed and emerge as tumors in a process called tumorigenesis (Cardiff et al., 1977). The model has been adapted to human breast cancer (Cardiff et al., 1977) and expanded to accommodate Fould’s concept of regression and Medina’s ductal hyperplasias (Medina, 1982). The HANS are key elements in mammary tumorigenesis. They are multifocal hyperplasias of the lobuloalveolar epithelium which require increased levels of estrogen, corticosteroids, and prolactin for development but can be maintained on constitutive levels of these hormones (Bern and Nandi, 1961). In particular, they appear to be more sensitive to constitutive levels of prolactin since removal of prolactin stimulation results in the regression of most HANS (Welsch and Nagasawa, 1977). HANS are more frequently observed in mouse strains with a high mammary tumor incidence than in low incidence strains (Nandi and McGrath, 1973). When HANS and normal ducts are transplanted into gland-cleared mammary fat pads, only the outgrowths of HANS develop tumors (DeOme et al., 1959). This not only proves that HANS are precursors to tumors but also provides an operational definition of “preneoplasia,” i. e., higher risk of tumorigenesis after isolation, transplantation, and growth (Cardiff et al., 1977). This rigorous operational criterion of preneoplasia is not as well fulfilled by any other experimental model system. Nodule transformation results in immortality since HANS can be indefinately maintained as serial transplants (Daniel et al., 1968; Young et al., 1971). The outgrowths of hyperplastic tissue in gland-free fat pads are referred to as hyperplastic outgrowths, usually abbreviated either as HPO or HOG (Medina, 1973; Ashley et al., 1980b) (Fig. 1).The immortality of the hyperplastic mammary tissues distinguishes them biologically from transplants of normal mammary epithelium that die out after only 5-6 transplant generations (Daniel et al., 1968; Young et d., 1971). In spite of their immortalization, mammary hyperplasias have a limited growth potential in that they do not invade local tissues and do not grow outside of the fat pad. This distinguishes mammary hyperplasias from the mammary adenocarcinomas which invade local tissue, metastasize to the lung, and grow outside of the fat pad (Medina and Asch, 1980). Since HANS can be serially transplanted, HPO lines have been developed and used to characterize “preneoplastic” mammary tissue (Medina, 1973; Ashley et al., 1980b). Each HPO line has distinctive morphological, virological, and biological characteristics involving, most importantly, their tendency to progress to malignancy. Thus HPO lines have been developed with tumor incidence ranging from 3 to 100% indicating that each hyperplastic tissue has a unique tumorigenic potential (Medina, 1973; Ashley et al., 1980b). In spite of this difference and the large number of phenotypic char-

170

ROBERT D. CARDIFF

acters catalogued for HPO lines, no reliable indicator of tumor potential has been found (Medina, 1978; Medina and Asch, 1980). The inability to find a reliable prognostic determinant is frustrating because it would be a major asset in predicting the clinical significance of comparable lesions in humans (Wellings et al., 1975; Cardiff et al., 1977). Two other biological characteristics of mammary hyperplasias are important to consider: (1) their pluripotential nature and (2) their increased susceptibility to carcinogens. The pluripotential nature of the HAN population has been demonstrated by subdividing a single HAN and transplanting each subdivision into a separate fat pad. This results in morphologically and biologically diverse outgrowths (DeOme et al., 1961). For example, subdivision of one HPO line in the third transplant generation resulted in three distinct sublines (Ashley et al., 1980b). Further, selective transplantation of morphologically aberrant areas from a single HPO can result in aberrant sublines (Cardiff et al., 1983). These data support the pluripotential nature of mammary hyperplasias. The increased susceptibility of hyperplastic cells to carcinogens has been the subject of numerous studies (Medina, 1973, 1978). Exposure to exogenous hormones, chemical carcinogens, virus, or radiation increases the tumor incidence of most HPO lines and usually decreased the tumor latent period (Medina, 1973). Exposure of normal mammary cells to the same carcinogens frequently results in an increased tumor incidence but the latent period is relatively longer. For example, no tumors appeared in the mammary glands of BALB/c mice treated with hormones, virus, or chemical carcinogens although many tumors did develop in the HPO transplanted into the same mice (Ashley et al., 1982). Radiation exposure most dramatically illustrates the increased susceptibility to carcinogenesis (Medina, 1973). Irradiation of BALB/c or BALBIcfC3H HPOs resulted in a dose-dependent increase in tumor incidence (Faulkin et al., 1982). In contrast, irradiated normal BALB/c mammary ducts did not develop any tumors after 2 years. Ill. The Development of Transplantable Hyperplastic Outgrowth Lines

The transplantation technique used to develop HPO lines is important in determining the characteristics and the stability of the lines and must be considered in the interpretation of the results. Therefore, this technique will be discussed in some detail. The first step in transplantation is the preparation of the recipient fat pad. At 3 weeks of age, the mouse mammary tree has begun to grow out from the nipple but fills only 10-20% of the mammary fat pad. The developing mammary tree can be identified under a dissecting microscope and removed by

PROTONEOPLASIA

171

surgical excision or electrocautery, leaving the mammary fat pad devoid of mammary epithelium. This leads to a fat pad referred to as gland-free or gland-cleared. Other mammary tissue can then be transplanted into the gland-free fat pad without interference from the host mammary gland. In the next step, the operator examines the donor HPO tissue under a microscope and dissects out the “most typical” area of tissue for transplantation. The “typical area” is removed as a strip from one edge of the HPO. The strip of mammary tissue is then cut into 1-2 mm cubes and each cube transplanted into a recipient gland-free fat pad. The transplanted tissue fills the fat pad in 10-12 weeks and the procedure is repeated as required. As can be appreciated from this description, the selection and transplantation procedure is totally dependent upon the judgment and ability of the operator. In our laboratory, only two experienced operators are allowed to maintain the stock HPO lines. The experienced operator can establish a stable transplant line in 3-4 transplant generations. The selection system is heavily biased toward a preconceived uniformity. One can easily understand how successive transplantation of YICMMI of the total population could quickly dilute out and eliminate rare or atypical subpopulations, creating a stable, uniform population. However, variants in the population can also be identified and transplanted (Cardiff et a l . , 1983). IV. Characterization of Hyperplastic Outgrowth Lines

A number of outgrowth lines have been developed. Two series of HPO lines, the D series and the Z series, have been used most extensively in molecular studies and will be discussed below. The D series was developed by transplantation of hormone-induced HANS from BALB/c mice (Medina, 1973). The BALB/c mouse does not carry exogenous MuMTV and has a low tumor incidence and the D series HPOs do not express MuMTV antigens nor contain exogenous MuMTV (Ashley et al., 1980a). The most extensively studied line, D1, initially had a low tumor incidence and was used in the classical biological experiments which established the fundamental characteristics of mammary hyperplasia (Medina, 1973; Cardiff et a l . , 1977). Later passages of D1 were also free of exogenous virus but the tumor incidence increased from 5 to 80% and its susceptibility to carcinogenic stimuli was altered (Ashley et al., 1982). Other HPO lines from the D series have been characterized and extensively used to delineate the nature of preneoplastic breast tissue but will not be discussed here (Medina, 1978). The Z series of HPO lines was developed by transplantation of three BALB/cfC3H HANS into BALB/c gland-free fat pads (Ashley et d . ,1980b). The Z series HPOs carry and express exogenous MuMTV. One HPO line

172

ROBERT D. CARDIFF

(25) was divided in the third transplant generation into three distinctive lines ( Z ~ CZ5c1, , and Z5d). The 25 group is interesting because each member has a different tumor incidence despite their common origin. Z5c has the highest tumor incidence at 50%, Z5cl has 44% tumors, and Z5d has only 15% tumors. The two other Z series HPOs have similar growth rates but 24 has a tumor incidence of 90% while 23 has 5% tumors (Ashley et al., 1980b). Perhaps the most iinportant aspect of these transplantation systems is that the hyperplastic tissues can not only be experimentally isolated but can also be expanded in any given generation. This provides sufficient tissue mass for most biochemical and virological analyses. Thus, the mouse mammary hyperplasias are unique among experimental models of “preneoplasia” in that they provide the only opportunity to study such pure tissue populations that are precursors to malignancy.

V. Restriction Endonuclease Mapping

A. GENERAL COMMENTS Restriction endonuclease digestion of DNA in conjunction with the Southern transfer technique (Southern, 1975) and specific radiolabeled DNA probes has become a powerful analytical tool. Much of the data considered in subsequent sections of this article comes from Southern blot analysis. Although the technique is widely applied, the nuances of interpretation may not be readily apparent to scientists not routinely using the technique. It is appropriate, therefore, to discuss some ofthe more general aspects of Southern blot analysis. The D N A is first extracted in high-inolecular-weight form using standard techniques. The DNA is then digested with restriction endonucleases that cleave DNA at specific sequences 4 to 6 nucleotides in length (Roberts, 1976), creating a series of DNA “restriction” fragments whose size is dependent upon the distance between successive restriction sites (Fig. 2). The restriction fragments are then separated on the basis of their relative molecular weight by electrophoresis in agarose gels. After denaturing the DNA in situ, the fragments are transferred from the agarose gels to nitrocellulose filter paper by the ingeniously simple Southern blotting technique. The agarose gel is placed on top of adsorbent filter paper that has its ends dipped in a buffer solution. Nitrocellulose paper is placed on top of the gel and a stack of dry paper towels is put on top of the nitrocellulose. The buffer is drawn by capillary action through the agarose and nitrocellulose into the paper towels. This action transfers the singlestranded DNA fragments to the nitrocellulose where they bind.

173

PROTONEOPLASIA

5’

t

GAG

t

P

P

O

E

POL

t

ENV

3’

t

P

B

I

t

t

P

kP

P?

C

A

- m H - H w

II

m DNA‘~P

A

5’

t

I

B?

3’

t

t

B

B?

B

A

B

A

.1

C

-

B

m

B FIG. 2. The M u M T V provirus (boxed areas) is represented intergrated into mouse D N A (wavy lines) (I). The provirus has long terminal repeats (LTR) at the 5’ and 3’ ends of the proviral molecule. The proviral sequences have five specific PstI sites (P; panel B) and two BnmNI sites (B; panel A). When the respective enzymes are used the proviral D N A is cleaved into a series of D N A fragments whose length is dependent upon the distance between the cleavage sites (11). The D N A fragments, such as fragment C in A or fragments B, C, D, E in B, which contain only MuMTL’ proviral DNA, are called “internal fragments.” Fragments which contain both mouse and proviral D N A (fragments A and B in A and fragment B in B) are hostviral fragments. As can lie appreciated, the patterns of the autoradiograph is dependent upon the size of the fragments and the probe used. When a DNA“,’ is used all fragments are visualized. However, when a DNALTHis used only the LTRs are detected.

174

ROBERT D. CARDIFF

The nitrocellulose paper is baked dry at 80°C in vucuo which causes the DNA to bind very tightly and then the filter is hybridized with a radiolabeled DNA probe for a period of 2-3 days. The filter is washed, dried, and exposed to X-ray film. When the film is developed, the autoradiography will have a dark band wherever probe has bound to a homologous DNA sequence (Fig. 3). If multiple homologous sequences occur, multiple bands will appear. The autoradiographic bands are indicators of homologous restriction fragments and the overall pattern of the bands is referred to as the restriction pattern.

FIG.3. An autoradiograph of a Southern blot showing the distribution of MuMTV DNA in uninfected GH liver tissue and MuMTV infected GR tumors. The probe was composed of cloned DNAs representative of most of the viral genome. Note that the tumors have more MuMTV restriction fragments than the uninfected liver.

PHOTONEOPLASIA

175

The quality of the results is dependent not only on the extent of DNA homology but also upon the completeness of enzymatic digestion of the DNA, the conditions of electrophoresis, the efficiency of transfer, the purity and concentration of the probe, the conditions of hybridization, and the sensitivity of the autoradiography. The most important single factor is the probe. Early MuMTV probes were produced using RNA purified from virions. A calf thymus DNA primed reverse transcription reaction was used to obtain “representative” radiolabeled cDNA (DNAreP). While this reaction did produce cDNA representative of the entire genome (Ashley et al., 1980a), the genome was not equally represented. .In retrospect, the long terminal repeats (LTR) were underrepresented in most cDNA probes. Further, many “purified” viral RNAs were contaminated with a variety of host sequences. Repetitive host sequences and ribosomal RNAs have been particularly troublesome problems but rarely acknowledged in print. Most of the probe problems have been solved with the development and wide dissemination of molecularly cloned MuMTV DNAs which can be nick translated and used as probes. (Majors and Varmus, 1981). The cloned DNAs are highly specific, can be labeled to a high specific activity, and used at high concentrations. On the other hand, the use of recombinant DNA has introduced new problems. The most common problem has been contamination of buffers with plasmids. Almost every laboratory has discovered an unexpected new autoradiographic band in Southern blots only to find that the band comes from a contaminant. In some cases, such experiences have caused embarrassing retractions of published data. When a MuMTV probe is applied under stringent conditions to the nitrocellulose paper following Southern transfer, it will hybridize only to homologous sequences. As emphasized in Fig. 2, each restriction fragment thus identified may contain only MuMTV sequences or it may contain MuMTV and host flanking sequences. Interpretation of the pattern of restriction fragments requires knowledge of the MuMTV restriction map.

B. MuMTV Several restriction maps of MuMTV DNA have been published (Shank et al., 1978; Cohen et al., 1979b; Varmus et al., 1979; Cohen and Varmus, 1980; Fanning et al., 1980b; Puma et al., 1982). Our purposes are best served by concentrating on the differences between endogenous and exogenous MuMTVs and upon fragments which contain both host and viral sequences. For example, the enzyme BamHI cuts MuMTV-2 twice in the envelope region of the viral genome. The secondary restriction sites are somewhere in the host flanking DNA. Thus, BamHI digestion results in three restriction fragments for each provirus, an internal fragment, a 5’ and a 3’ fragment (Fig. 2). The size of the outer restriction fragment is entirely

176

ROBERT D . CARDIFF

dependent upon the distance from the MuMTV BamHI site to the next BamHI site in the host DNA. The host DNA contributes most to the size of the fragment. Thus, differences in the fragment length are directly attributable to the length of the host flanking sequences. Different mouse strains may have unique restriction patterns because they have a different complement of proviruses (Cohen and Varmus, 1979). Restriction fragments containing both MuMTV and host sequences will be referred to as MuMTV-host fragments (Fig. 2). Other enzymes will cut MuMTV DNA more times giving more “internal” fragments composed entirely of viral sequences (Cohen et al., 1979b; Shank et al., 1979; Fanning et al., 198Ob). PstI and BainHI have been particularly useful because they provide internal restriction fragments which are unique for the exogenous C3H viruses (Fig. 2). For example, PstI cleaves the exogenous MuMTV(C3H) provirus five times creating a unique 4.2 kb internal fragment (Cohen et al., 1979b) (Fig. 2). Since the endogenous proviruses of BALB/c and C3H mice do not contain a 4.2 kb restriction fragment, the presence of this PstI fragment indicates that the tissue is infected by the exogenous C3H virus (Cohen et al., 1979b). BamHI is also useful since it cuts MuMTV(C3H) DNA twice in the envelope region providing a unique 1.3 kb fragment which is also diagnostic of acquired proviral DNA in the BALB/c genome (Cardiff et al., 1981). The BamHI has the added advantage of giving sufficiently large 3’ and 5’ MuMTV fragments that the relative size of the host flanking sequences can also be determined (Cardiff et al., 1983) (Fig. 2). Several other exogenous MuMTVs have been characterized by Southern blot analysis (Fanning et al., 1980a; Puma et al., 1982; Etkind et al., 1982). For example, the exogenous MuMTV(RII1) has different PstI sites and can thus be distinguished from MuMTV(C3H) (Etkind et al., 1982). On the other hand, viruses which are carried as both genetic elements and infectious milk agents, such as GR-mtv-2 in GR mice or C3Hf virus in C3H mice, represent special problems. Their genomes cannot be completely mapped because the endogenous provirus obscures the map. Their restriction maps can be completed by infection of heterologous cells or by analysis of unintegrated full length double-stranded DNAs produced in vitro by reverse transcription of virion RNA (Puma et al., 1982). The detection of acquired (exogenous) MuMTV provirus DNA and the characterization of the host flanking DNA by Southern blots is facilitated using the cloned MuMTV LTR as a probe (DNALTR).The LTR is the long terminal repeat found at both ends of the proviral genome (Fig. 2) (Majors and Varmus, 1981; Donehower et aZ., 1983). Thus, cleavage of the proviral structural genes with most enzymes generates 3‘ and 5‘ fragments containing at least one LTR and the flanking host sequences (Fig. 2). Differences in

PHOTONEOPLASIA

177

lengths of the restriction fragment are largely dependent upon the lengths of the host flanking DNA. This makes the LTR an ideal probe for identifying MuMTV-host fragments and comparing relative length of host restriction fragments (Cardiff et al., 1983). Armed with these probes, the appropriate restriction enzymes and mouse DNA, there have been many investigations of MuMTV infected tissues (reviewed in Fanning and Cardiff, 1984). One of the basic tenets emerging from these studies has been that MuMTV-induced tumors have quasi-clonal or clonal dominant populations. This claim is based on Southern blot analysis and on a series of assumptions which are not always clearly stated or appreciated. The basis for these claims is outlined below. When MuMTV proviruses are in the same location in all cells, such as in the case with genetically transmitted or endogenous proviruses, the MuMTV restriction pattern of all tissues will be identical and the autoradiographic bands will be of equal density (Varmus et al., 1978). When a cell is infected by exogenous MuMTV, the acquired provirus integrates randomly or, at least, in many different positions in the mouse genome. Random integration results in many infected mammary cells, each containing one or more proviruses in a different location. Since the exogenous provirus occupies no single cellular integration site at a high frequency, the acquired MuMTV-host restriction fragments cannot be detected in the DNA from heterogeneous or polyclonal tissues such as the lactating mammary glands (Cohen et al., 1979b). In contrast to DNA blots from polyclonal MuMTV-infected lactating mammary gland, Southern blots of DNA from MuMTV-infected tumors do have extra MuMTV fragments (Cohen et al., 1979a) (Fig. 3). Since exogenous MuMTV integrates randomly, the most likely explanation for the additional MuMTV-host flanking fragments is that the tumor arose from just one cell or several cells (Varmus et al., 1978). That is, mouse mammary tumors are composed of clonal, quasi-clonal, or clonal dominant cell populations (Cohen et al., 1979a; Cohen and Varmus, 1980; Cardiff et aZ., 1981). Several problems arise in the interpretation of MuMTV restriction patterns in neoplasms. The criticism most frequently heard is that many newly acquired fragments are submolar in relation to the endogenous MuMTV fragments. However, the newly acquired fragments would not be expected to be equimolar since MuMTV generally integrates into only one of the two homologous chromosomes (Nusse and Varmus, 1982). The relative concentration is further diluted with the uninfected mesenchymal and inflammatory cells normally found in the tumor. Further, the exogenous MuMTV-host fragments in some tumors have different densities, implying subpopulations. The variable autoradiographic densities of acquired bands within the same tumor are difficult to explain, however, some “submolar” fragments have been

178

ROBERT D. CARDIFF

maintained without apparent change for many transplant generations (Cardiff et al., 1983). Perhaps some “submolar” bands may be due to deletions in the region of the provirus being probed (Nee1 et al., 1981). It cannot be determined on the basis of a single blot whether such variability indicates multiple clones or a single clone with multiple subclones. A second, but frequently encountered, problem is the theoretical “undetected” MuMTV restriction fragment. If a new restriction fragment is not observed, how can one be assured that it is not hidden or masked by comigration with another band? For example, the initial EcoRI digests of some GR tumor DNA failed to reveal additional proviruses (Fanning and Cardiff, 1984). Only after digestion of the same DNA with other enzymes did the newly acquired fragments appear (Fanning et al., 1980a). Clearly, the additional proviral sequences were hidden by the endogenous MuMTV DNA. Comigration of bands raises a second problem when trying to compare one tissue with another. Does simple comigration of bands from two tissues prove that they are the same fragments? The answer is clearly no, digestion with multiple enzymes is required to prove identity. In summary, although the Southern blot analysis has proven to be a very powerful tool in the study of mammary tumorigenesis, its interpretation requires careful consideration of the variables in the system.

VI. The Molecular Biology of Mouse Mammary Neoplasia

A. TUMORS Keeping in mind the technical caveats outlined above the molecular biology of the mouse mammary tumor system can now be considered. These studies are essentially confined to the analysis of MuMTV nucleic acids. The initial liquid phase nucleic acid hybridization analysis detected comparable MuMTV copy number in all inbred laboratory strains of mice (Varmus et al., 1972). With selective prehybridization, RNA probes could be developed which distinguished between exogenous and endogenous viruses (Drohan et al., 1977; Michalides et al., 1978). Very careful studies using liquid phase hybridization and hybridization kinetics established that MuMTV-induced tumors have more copies of MuMTV DNA than normal tissues (Morris et al., 1977). Using Southern blot analysis, the unique 4.2 kb PstI fragment of MuMTV(C3H) provirus was detected in the DNA of BALB/cfC3H lactating mammary glands but no new host-virus fragments were detectable (Cohen et al., 1979b). In contrast, BALBIcfC3H tumor DNAs contained more MuMTV restriction fragments than liver or lactating mammary gland DNA. Indeed,

PROTONEOPLASIA

179

the two basic tenets of the molecular biology of the mammary tumor system are based on these results: (1) MuMTV DNA is amplified in virus-induced tumors and (2) tumors are composed ofclonal dominant populations (Cohen et al., 197913). Increased numbers of newly acquired MuMTV restriction fragments have been found in all virus-induced tumors including BALB/cfC3H, C3H, GR, CSHf, RIII, and BALB/cNIV tumors (Puma et al., 1982; Fanning et al., 1980a; Etkind et al., 1982; Cohen and Varmus, 1980; Etkind and Sarkar, 1983; Altrock et al., 1982; Drohan et al., 1982; Groner and Hynes, 1980; Hynes et al., 1980; Macinnes et d., 1981; Morris et al., 1982). The GR tumors were interesting since the initial studies failed to detect new restriction fragments (Groner and Hynes, 1980). However, for the reasons previously indicated, even GR tumors were eventually found to have amplified MuMTV proviruses (Fanning et al., 1980a). Comparisons of many different tumors failed to reveal the same MuMTV-host or “tumor-specific” fragments. This was interpreted to mean that there were no preferred MuMTV integration sites or loci. However, the mere presence of clonal dominant populations implied that a selective type of host-virus interaction was required in the formation of tumors. Recently, two domains of mouse DNA, referred to as int-1 and int-2, have been identified in virus-induced tumors which are frequently occupied by exogenous MuMTV provirus DNA (Nusse and Varnus, 1982; Peters et al., 1983). The int-1 site was identified by cloning DNA from a C3H tumor which had only one additional provirus. The host sequences flanking the acquired provirus were then used to select the normal, uninterrupted region of DNA from a BALB/c library. Using probes from this region, 19 of 26 C3H mammary tumors examined showed MuMTV provirus integrated within 19 kb of the original site. The integration of MuMTV into this locus, detected by an altered restriction fragment length, is accompanied by int-1 mRNA expression in tumors but not normal mammary gland (Nusse and Varmus, 1982). The second site, int-2, was identified by DNA cloned from BALB/c DNA and was found to be occupied by newly integrated MuMTV DNA in 17 of 40 BR6 mouse tumors examined (Peters et al., 1983). These experiments establish that MuMTV integration into certain domains occurs repeatedly in association with murine neoplasia. The integration of MuMTV is not apparently limited to a single site but may occur in a number of different areas within the domain. This may account for the difficulty experienced previously in defining a “tumor-specific” host-virus restriction fragment. These data also imply that MuMTV acts in much the same manner as do some other slow-acting RNA tumor viruses, such as the avian leukosis virus (ALV), that is, by activating cellular oncogenes. ALV provirus integrates

180

ROBERT D. CARDIFF

close to the cellular myc gene (c-myc) resulting in the expression of c-myc mRNA (Hayward et al., 1981; Nee1 et al., 1981; Payne et al., 1982). Originally, all ALV proviruses appeared to integrate “upsteam” from c-myc leading to a downstream-promotion model. However, ALV has been found in other relationships to the c-myc gene so that a more general model might be insertion-activation or insertion-mutagenesis (Hayward et al., 1981; Varmus, 1982a). In mouse mammary tumorigenesis, MuMTV appears to insert close to a given host sequence and to be related to the expression of these host sequences. Obviously, the transcribed region is a potential mammary oncogene and, thus, of extreme interest. B. HYPERPLASIAS The premalignant mammary hyperplasias have not been as extensively characterized. However, enough reports have been published to evaluate the general direction of research in this critical area. The Z series of BALBIcfC3H HPO lines described above have been examined using Southern blot analysis (Cardiff et al., 1981). All five HPO lines contained exogenous MuMTV fragments demonstrating that additional MuMTV sequences and clonal dominant populations can occur in premalignant hyperplasias as well as malignant neoplasms. The three HPOs from HAN 25 all shared several MuMTV fragments but also had other unique fragments. The other two lines, 23 and 24, had their own unique MuMTV restriction patterns. The restriction fragment patterns were stable over 11 transplant generations. Acquired MuMTV restriction fragments have also been reported in GR plaques and HPOs (Fanning et al., 1982; Michalides et al., 1982a). In one study, the acquired fragments were difficult to identify in GR plaques implying that the plaques comprise a more heterogeneous population (Fanning et al., 1982). However, another study revealed abundant bands in GR hormone-dependent tumors (Michalides et al., 1982a). In a more detailed analysis, acquired MuMTV fragments were found in the DNA of an isolated BALB/cfC3H HAN demonstrating that MuMTV DNA amplification and clonal dominant cell populations were not simply an artifact of the transplantation system (Cardiff et al., 1983). Individual HANS transplanted into gland-cleared fat pads developed primary HPOs which had additional MuMTV sequences identified by LTR probes. Further, each primary HPO had its own restriction pattern. When a single HAN was subdivided and each piece was transplanted into a separate fat pad, the primary HPOs shared multiple restriction fragments when analyzed with any one of several restriction enzymes (Cardiff et al., 1983). However, each primary outgrowth also had different, or unique,

PROTONEOPLASIA

181

fragments. Further, in spite of shared restriction patterns with many enzymes, at least one enzyme was found whose restriction fragment pattern failed to show a common band in one set of related HPOs derived from a single HAN. This perplexing pattern suggested very closely related but not identical fragments exist in each subdivision of the HAN. If both hyperplasias and tumors are characterized by acquired MuMTV restriction fragments, what is the relationship between the two tissues? The first direct comparison involved the Southern blot analysis of GR hormoneindependent tumors which arose from hormone-dependent tumors (Michalides et al., 1982a). The hormone-independent tumors had additional MuMTV-host fragments together with the same restriction fragments as the dependent precursors. This pattern was interpreted as being consistent with genetic divergence from a common origin. Similar observations were made in the BALB/cfC3H system (Cardiff et al., 1983). A tumor from one of the HPOs can be compared with the same HPO froin which it arose. In 47 such comparisons, all tumors had the same major acquired MuMTV fragments as the parental HPO. However, all tumors had one or more additional fragments. Several tumor-HPO pairs had nearly identical restriction fragment patterns but the tumors had at least one minor additional band found using one enzyme which was apparently masked using several other enzymes. The tumor-specific restriction fragments were unique to each tumor, even in multiple tumors arising from the same transplant (Cardiff et al., 1983). These findings have far reaching implications concerning the origin and evolution of mammary neoplasia, the concept of preneoplasia, and the role of MuMTV in these processes. They can be best understood in the context of the biology of the mouse mammary tumor system as will be presented in the subsequent sections. VII. Origin and Evolution of Mouse Mammary Tumors

The recent rediscovery of tumor heterogeneity has focused attention on the therapeutic and conceptual challenge confronting the clinician and tumor biologist attempting to deal with the enormous diversity found within a single tumor (Owens et al., 1982). The mouse mammary tumor system has been one of the primary models of tumor heterogeneity (Heppner et al., 1981; Dexter et al., 1978). On the other hand, the mouse mammary tumor system also serves as an example of clonality (Cohen et al., 1979b; Varmus et al., 1979; Cardiff et al., 1983). These apparently conflicting viewpoints can be reconciled by distinguishing between the origin and the evolution of a neoplasm. In order to develop this theme, each model will be discussed. Early work on the HAN emphasized its biological diversity (DeOme et al.,

182

ROBERT D. CARDIFF

1961; Cardiff et al., 1977). As described in Section VI,B, transplantation of subdivisions of individual HANS or HPOs can result in outgrowths with different morphological and biological characteristics. Several cell lines and subclones with different phenotypes, karyotypes, biological behavior, responses to drugs, and genotypes have been developed from a single BALB/cfC3H tumor (Heppner et UI!. , 1978). Clearly, the original tumor cells had the capacity for great diversity. What is not clear is whether the diverse cells existed in the original tumor or evolved in subsequent passages. Cells from the same GR tumor had different restriction patterns (Morris et al., 1982). Each of several cell lines cloned from another single tumor had different growth and transplant potentials (Macinnes et al., 1981). Each clone had a unique MuMTV restriction pattern but careful scrutiny of the published Southern blots reveals at least one acquired restriction fragment was common to all clones. Southern blot analyses of a number of premalignant and malignant mammary tissues have also revealed new and unique MuMTV fragments in each tissue (Cardiff et al., 1981, 1983). These experiments emphasize the naturally occurring heterogeneity found in neoplasias. While the evidence that mouse mammary neoplasias are heterogeneous is substantial, extensive data suggest that tumors are homogeneous populations derived from selected subpopulations. The only point which is not entirely clear at this time is the exact size of the cell population which becomes neoplastic. Is it one cell, several cells, or a large number of cells? Let us examine the evidence for clonality. The HAN is, by definition, a focal lesion. Therefore, many MuMTVinfected cells in the mouse mammary gland do not participate in the formation of a HAN. In this sense, the HAN is a subset of the total population. When the HAN or its HPO is transplanted, the earliest detectable evidence of malignancy is a microfocal “tumorlet” (Ashley et al., 1980b). Further, small tumors can be readily observed grossly as focal lesions within outgrowths. Thus, the tumor arises out of a focal area of the HPO. Clearly many of the cells in the HPO do not participate in the formation of any given tumor. Tumors, in this sense, arise as a subset of the hyperplastic cell population. This principle could also be demonstrated by transplanting the virusnegative BALB/c HPO D1 into BALB/cfC3H mice that carry the infectious C3H MuMTV (Ashley et al., 1980a). Only a fraction of the D 1 HPO cells became productively infected so that each outgrowth was a mixture of infected and uninfected cells (Ashley et al., 1980a). As predicted from the studies using the sex-linked, polymorphic enzymes, polyclonal tumors would be expected to reflect the mixed cell composition of the precursor population. In contrast, monoclonal tumors should be composed of a single

PHOTONEOPLASIA

183

cell lineage (Nowell, 1976; Fialkow, 1977). Tumors arising from the D1 HPO transplanted into BALB/cfC3H were composed of either entirely uninfected or entirely infected cell populations. No tumors of obvious mixed cell origin were identified. The Southern blot analysis corroborated these data strongly supporting a clonal origin of mammary tumors. Another line of evidence is also based upon Southern blot analysis. The most striking feature of premalignant and malignant tissues is the appearance of acquired MuMTV restriction fragments. These host-virus fragments have been found in all MuMTV-induced neoplasms and indicate that the neoplasms are composed of homogeneous clonal dominant populations. The presence of MuMTV restriction fragments common to all subdivisions of the same tumor or the same HAN reinforces the concept of a common origin. However, because of the technical limitations of restriction analysis, outlined above, the restriction data fall short of proving a monoclonal origin of tumors or HANS. The experiments with the integration loci provide yet another line of evidence (Nusse and Varmus, 1982; Peters et al., 1983). The MuMTV integrates into only one chromosome so that each tumor has a normal germline int-1 or int-2 locus and a locus interrupted by proviral integration. In the published blots, many of the rearranged and unrearranged int-1 and int-2 fragments are nearly equamolar. This can happen only if the tumors are monoclonal proliferations. The other tumors with submolar rearranged bands might be explained by dilution from a large component of mesenchymal cells carrying two normal gerinline homologous chromosomes (Nusse and Varmus, 1982). The weight of the evidence strongly supports a clonal origin of mouse mammary neoplasms. The data supporting heterogeneous populations in neoplasms can be reconciled with the data supporting a clonal origin by recognizing that both lines of evidence are essentially correct and by distinguishing between the origin and the evolution of neoplasms. We offer the following model (Fig. 4). The initial event appears to take place in the normal mammary epithelium. As a result a critical change in the host DNA such as the integration of MuMTV into a critical genetic domain, perhaps int-1 or int-2, a cell gains a selective growth advantage. The progeny of that cell proliferate to emerge from the general population as an immortalized HAN or plaque. These hyperplasias are hormone dependent, prolactin dependent in the case of the HAN, and estrogen dependent in the case of the plaque. Even though the HAN is probably derived from a single cell, the HAN population is apparently capable of undergoing many further genetic changes which result in biologically heterogeneous populations. Of the many cells in the general hyperplastic population, only one or a few undergo secondary DNA changes which result in malignant transformation.

184

ROBERT D. CARDIFF

.++

FIG. 4. A schematic diagram of mouse mammary neoplasia which incorporates the recent molecular data. The relative status of each tissue is depicted in relation to neoplastic progression (abscissa) and growth (ordinate). Progression is divided into the number of theoretical genetic mutations (hits) required for nodulogenesis (1) and tumorigenesis (2). Arrows demonstrate possible divergent pathways.

These “tumor-transformed” cells have additional selective growth advantages and emerge as a malignant neoplasm, generally with additional MuMTV genotypic markers. The concept of a clonal origin and divergent evolution has been applied to other tumors (Nowell, 1976),but has never been as clearly demonstrated at the DNA level as in the mouse mammary tumor system. The presumptive monoclonal origin in the mammary tumor system suggests that nodule transformation is very infrequent. The problem with such rare events is that they may be very difficult to detect and correctly interpret. Although the phenomena may be detected when the neoplasm has achieved sufficient mass, it may be impossible to distinguish cause from effect. If tumors are also the result of subclonal expansion from the nodule population, the malignant transformation must be equally rare and subject to the same interpretative constraints. The divergent evolution of mammary tumors indicated by rapid changes in both phenotype and genotype demonstrates the great plasticity of mammary tumors. One must be impressed with the seemingly endless variations of the restriction patterns found in tumors and HANS. For example, the HPO lines continue producing tumors after 20 transplant generations and no two tumors from any given generation have the same genotype (Cardiff et al., 1983). Either the HPO populations are endlessly diverse at the outset or

PHOTON EOPLASIA

185

they keep producing new malignant subclones. Most likely they continually produce new transforinants, an important characteristic of preinalignant hyperplasias. The heterogeneity of tumors serves to warn the investigator of the hazards inherent in interpreting any set of empirical observations. Phenomenon found in tumor cells, no matter how different from normal cells, may not be evidence of a causative or initiating event. They may merely reflect the evolutionary divergence of tumors. The proper definition of origin entails the study of the initial events, unencumbered by a vast evolutionary divergence. The mouse inaininary tumor system has the advantage in that initial events can be studied in the WAN.

VIII. The HAN Is Protoneoplastic

The original two step model of maminary tumorigenesis regards the HAN as a “preneoplastic” tissue (DeOme, 1967). When the concept was developed, the mouse mammary tumor was simply considered a neoplasm and HANSwere logically considered precursor or preneoplastic states. The term “preneoplasia” had the added advantage of designating the relative position of such tissues in the pathway to malignancy. Further, the term “preneoplasia” was sufficiently neutral that it could be supported without extended debate. However, our knowledge of the molecular and cellular biology of the mouse mammary tumor system has now progressed to the point that a more accurate term is needed. The term preneoplasia has the disadvantage of implying that the tissues are antecedent to the first step in the neoplastic progression (Foulds, 1969). Even normal tissues, it could be argued, are in this sense preneoplastic. Several lines of evidence challenge these concepts of “preneoplasia. ” First, HAN cells are immortal and can be serially transplanted while normal mammary tissues cannot (Daniel et ul., 1968; Young et d., 1971). For example, the D1 HPO line has now been carried in transplantation for over 15 years (Medina and DeOme, 1968; Ashley et al., 1980a). Immortality is one of the major biological characteristics which distinguish HAN cells from normal. Second, the HAN cells are genetically altered as evidenced by the presence of acquired MuMTV restriction fragments in the outgrowth DNA (Cardiff et d . , 1981, 1983). The presence of acquired MuMTV provirus is a reproducible, acquired genetic change in these somatic cells and implies that the HANS are composed of clonal dominant populations. Clonal dominance, in turn, implies that these changes in the cell genotype are critical to

186

ROBERT 1). CARDIFF

the neoplastic process, perhaps even responsible for the immortalization of the tissue. Further, the Southern blot analysis strongly suggests that tumors are direct derivatives of the hyperplastic cells. This relationship makes the mouse mammary hyperplasias an integral and critical part of the neoplastic progression that ends in the malignant tumor. The HAN is clearly the first step. On the other hand, it is equally clear that not all cells in the hyperplastic populations are committed to malignant transformation. In fact, tumors arise from HAN as distinct clonal subpopulations. The vast majority of cells in the outgrowths do not progress to malignancy but remain in a latent or dormant state. Apparently, having attained the first step, the outgrowth cells are in a state of heightened susceptibility but do not progress until “hit” by a second event. With the newer evidence, the term “preneoplasia” becomes inadequate. In search for a more appropriate term, we suggest protoneoplasia be applied to the high risk mammary hyperplasias because this implies that they are the original, or first, step in the neoplastic progression and that the protoneoplastic mammary hyperplasia is an integral but nonobligate part of neoplastic progression. IX. The Role of MuMTV in Mouse Mammary Tumorigenesis

There is no question that MuMTV is associated with mouse mammary tumorigenesis; however, its precise role in tumorigenesis has been elusive. If a low tumor incident strain such as BALBIc is infected with MuMTV either by injection or foster nursing, tumors will rapidly develop (Nandi and McGrath, 1973; Altrock and Cardiff, 1979). The tumor incidence is proportional to the amount of virus given, the amount of virus expressed, and the number of cells infected (Cardiff and Young, 1980). Since MuMTV is expressed in normal mammary cells but not in all malignant cells and is associated with a prolonged tumor latency period, it is unlikely that MuMTV carries a oncogene analogous to those found in the rapidly transforming sarcoma viruses. Since mouse mammary tumors are associated with clonal dominant proliferations the genetic events leading to mouse mammary tumorigenesis must be rare. The finding of preferred integration regions makes the insertion-mutation model attractive (Varmus, 1982a,b). A major biological question becomes, at which step, nodulogenesis or tumorigenesis, does MuMTV exert its effect or does it work at both? Early experiments favored the concept that MuMTV acts in both steps of mammary tumorigenesis (Medina, 1973, 1982). The nodule-inducing virus

PROTONEO PLAS IA

187

(NIV) first described in C3Hf was subsequently transferred to BALB/c where it resulted in a high HAN incidence and a low incidence of mammary tumors (Young et al., 1984). The apparent disassociation of nodule induction from tumor induction led to the hypothesis that two types of MuMTV were required, NIV causing nodules and a second virus required for tumorigenesis (Medina, 1973). A second line of evidence favoring the involvement of MuMTV in tuinorigenesis was developed by transplanting the BALB/c HPO D1 and D2 into virus-infected BALB/cfC3H mice. This resulted in an increased tumor incidence in the HPO lines (Medina and DeOme, 1970); however, no virological studies were done. These and other experiments support the hypothesis that MuMTV influences both nodulogenesis and tumorigenesis. Evidence leading to a different interpretation has slowly accumulated. For example, when the D1 experiments were repeated, transplantation into BALB/cfC3H resulted in a decreased D1 tumor incidence and only 50% of the tumors were virus infected (Ashley et al., 1980a). These experiments suggested that the D l was no longer susceptible to the influence of MuMTV and challenged current thinking about the role of MuMTV in tumorigenesis. The observations that the DNAs of HPOs and tumors derived from HPOs have the same MuMTV restriction fragments suggest that MuMTV is involved in at least the initiation of nodules. Since some tumors have very few additional MuMTV DNA fragments, the major and most critical host-virus interactions probably occur at the level of the HPO. Since some MuMTVinduced tumors have only minor changes in MuMTV DNA as compared to the HPO DNA, MuMTV integration must have initiated the first step but may not be important in the second step (Cardiff et al., 1983) (Fig. 4). All this suggests that MuMTV is primarily involved in the induction of protoneoplastic HANS, a hypothesis with several important implications. First, the mouse mammary tumor virus should be renamed the “mouse mammary hyperplasia virus (MuMHV). Second, the virus, while not directly transforming, induces host genes which immortalize tbe cell in a manner similar to the myc gene (Land et al., 1983). Finally, genes independent of MuMHV are critical in the malignant transformation of the protoneoplastic mammary cell. ”

X. Summary

The mouse mammary tumor virus has provided a window into the inner workings of the mammary epithelial cell at the earliest stages of neoplasia. Techniques of molecular biology permitted us to look through that window revealing a new biology which deserves consideration as a model for mammary tumorigenesis in all species.

188

ROBERT D. C A R D I F F

According to this model the neoplastic process originates in a single mammary epithelial cell as a result of a critical genetic alteration, such as integration of MuMTV (MuMHV) into a key site in the mouse genome (Fig. 4). The genetic alteration immortalizes the cell and provides it with selective growth advantages which result in a clonal proliferation. This original proliferation emerges as the protoneoplastic mammary hyperplasia. The protoneoplastic cells have limited growth potential and are not obligated to undergo malignant transformation but they represent a genetically evolving population highly susceptible to full blown malignancy after exposure to carcinogens. Protoneoplastic cells which undergo further genetic alterations that provide additional selective growth advantages proliferate and emerge as malignant tumors. The genetic alterations are sometimes reflected by changes in viral DNA but this is not essential and most mouse mammary tumors probably do not occur as the result of new host-virus interactions. No doubt the current work on the mouse “int” loci will help define the genes responsible for the induction and maintenance of the protoneoplastic state. Since such host genes have proven so ubiquitous, one must also predict that analogous genes will be found in human mammary protoneoplasias. Detection of such sequences may help distinguish protoneoplastic processes from nonneoplastic, low risk hyperplasias in the human breast. Finally, the gene or genes involved in the more lethal malignant transformation await elucidation. Based on past and current progress one can be sure that the mouse mammary tumor system will help point the way.

ACKNOWLEDCMENTS This article is dedicated to Dr. K. B. DeOme, the man who started us thinking about the problem. I appreciate the many colleagues who shared their reprints and preprints with me. The editorial comments of many colleagues are greatly appreciated. Special thanks are due Dr. M. B. Gardner for his meticulous editing and useful discussion. M y thanks to Drs. H. C. Outzen, A. Freeman, J. Hilgers, F. Meyers, and B. Edwards for discussion and comments. I thank Ms. Faith Betti for her cheerful and competent preparation of the manuscript.

REFERENCES Altrock, B. W., and Cardiff, R. D. (1979).J . Natl. Cancer Inst. 63, 813-820. Altrock, B. W., Cardiff, R. D., Puma, J. P., and Lund, J. K. (1982).J . Natl. Cancer Inst. 68, 1032-1041. Ashley, R. L., Cardiff, R. D., and Fanning, T. G. (1980a).J . Natl. Cancer Inst. 65, 977-986. Ashley, R. L., Cardiff, R. D., Mitchell, D. J., Lund, J. K., and Faulkin, L. J. (19801,). Cancer Res. 40, 4232-4242. Ashley, R. L., Cardiff, R. D., Pratt, T. S . , and Faulkin, L. J . (1982).1. Natl. Cancer Inst. 69, 639-645.

PROTONEOPLASIA

189

Bern, H. A., and Nandi, S. (1961). Prog. E r p . Tumor Res. 2, 90-144. Cardiff, R. D., and Young, L. J . T. (1980). Cell Prolg 7, 1105-1114. Cardiff, R. D., Wellings, S. H., and Faulkin, L. J. (1977). Cancer 39, 2734-2746. Cardiff, H. D., Fanning, T. G., Morris, D. W., Ashley, R. L., and Faulkin, L. J. (1981). Cancer Res. 41, 3024-3029. Cardiff, R . D., Morris, D. W., and Young, L. J. T. (1983).J . Natl. Cancerlnst. 71, 1011-1019. Cohen, J. C. (1980). Cell 19, 653-662. Cohen, J. C., and Varmus, H. E. (1979). Nature (London)278, 418-423. Cohen, J. C., and Varmus, H. E. (1980).J. Virol. 35, 298-308. Cohen, J. C., Majors, J. E., and Varmus, H. E. (1979a).J. Virol. 32, 483-496. Cohen, J. C., Shank, P. R . , Morris, V. L., Cardiff, H. D., and Varrnus, H. E. (1979b).Cell 16, 333-345. Daniel, C. W., DeOme, K. B., Young, L. J . T., Blair, P. B., and Faulkin, L. J. (1968). Proc. Natl. Acad. Sci. U.S.A. 61, 53-60. DeOme, K. B. (1967). I n “Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability” 0. Neyman, ed.), pp. 649-655. Univ. of California Press, Berkeley, California. DeOnie, K. B., Faulkin, L. J., Bern, H. A., and Blair, P. B. (1959). Cancer Res. 19, 515-520. DeOme, K. B . , Blair, P. B., and Faulkin, L. J. (1961).Acta Union Int. Cancerum 17, 973-982. Dexter, D. L., Kowalski, H. M . , Glazer, B. A., Fligiel, Z., Vogel, R., and Heppner, G. H. (1978). Cancer Res. 38, 3748-3763. Donehower, L. A,, Fleurdelys, B., and Hager, 6. L. (1983).J. Virol. 45, 941-949. Drohan, W. N . , Kettman, R., Colcher, D., and Schloln, J. (1977).J. Virol. 21, 986-995. Drohan, W. N . , Cardiff, R. D., Lund, J. K., and Schlorn, J. (1980).Cancer Res. 40,2316-2322. Drohan, W. N . , Benade, L. E., Graham, D. E., and Smith, 6. H. (1982).J. Virol. 43, 941949. Etkind, P. R., and Sarkar, N . H. (1983).J. Virol. 45, 114-123. Etkind, P. R., Szabo, P., and Sarkar, N. H. (1982).J. Virol. 41, 855-867. Fanning, T. G., and Cardiff, R. D. (1984). Ado. Virol. Oncol. 4, 71-94. Fanning, T. G., Puma, J. P., and Cardiff, R. D. (1980a).J . Virol. 36, 109-114. Fanning, T. G., Puma, J. P., and Cardiff, R. I). (1980b). Nucleic Acids Res. 8, 5715-5723. Fanning, T. G., Vassos, A. B., and Cardiff, R. D. (1982).J. Virol. 41, 1007-1013. Farber, E. , and Sporn, M. D., eds. (1976). Cancer Res. 36, 2475-2706. Faulkin, L. J., Mitchell, D. J., Caridff, R. D., and Goldman, M. (1982). Conf. Radioprotect. Anticarcinogens, 1st p. 10 (Abstr.). Fialkow, P. J. (1977). Prog. Cancer Res. Ther. 3, 439-453. Foulds, L. (1958).J. Chronic Dis. 8, 2-37. Foulds, L. (1969). “Neoplastic Development.” Academic Press, New York. Gardner, M . B., and Rasheed, S. (1982). Znt. Reo. E x p . Pathol. 23, 209-267. Groner, B . , and Hynes, N . E. (1980).J . Virol. 33, 1013-1025. Hayward, W. S . , Neel, B. 6.. and Astrin, B. G. (1981). Nature (London)290, 475-480. Heppner, G. H . , Shapiro, W. R., and Rankin, J . K. (1981). Pediatr. Oncol. 1, 99-116. Hynes, N . E . , Groner, B., Diggleman, H., Van Nie, R., and Michalides, R. (1980). Cold Spring Harbor Syuap. Quant. B i d . 44, 1161-1168. Land, H., Parada, L. F . , and Weinberg, R. A. (1983). Nature (London)304, 596-602. Macinnes, J. I . , Chan, E. C. M. L., Percy, D. H., and Morris, V. L. (1981). Virology 113, 119129. Majors, J. E. , and Varmus, H. E. (1981). Nature (London)289, 253-259. Medina, D. (1973). Methods Cancer Res. 7, 3-53.

190

ROBERT D. CARDIFF

Medina, D. (1978). In “Breast Cancer” (W. L. McGuire, ed.), Vol. 11, pp. 47-102. Plenum, New York. Medina, D. (1982). In “The Mouse in Biomedical Research” (H. L. Foster, 1980; J. D. Small and J . G. Fox, eds.), Vol. 4, pp. 373-396. Academic Press, New York. Medina, D., and Asch, B. B. (1980). In “Cell Biology of Breast Cancer” (C. McGrath, M. J. Brennan, and M. A. Rich, eds.), pp. 363-371. Academic Press, New York. Medina, D., and DeOme, K. B. (1968).J. Natl. Cancer Inst. 40, 1303-1308. Medina, D., and DeOme, K. B. (1970).J . Natl. Cancer Inst. 45, 353-363. Michalides, R . , VanDeemter, L., Nusse, R., and Van Nie, R. (1978). Proc. Natl. Acad. Sci. U . S . A . 75, 2368-2378. Michalides, R., Wagenaar, E., and Sluyser, M. (1982a). Cancer Res. 42, 1154-1158. Michalides, R., Wagenaar, E., Hilkens, J., Hilgers, J., Groner, B., and Hynes, N. E. (1982b). J . Virol. 43, 819-829. Morris, V. L., Mederios, E., Ringold, G. M., Bishop, J. M., and Varmus, H. E. (1977).J . Mol. B i d . 114, 73-91. Morris, V. L., Gray, D. A , , Jones, R. F., Chan, E. M. L., and McGrath, C. M. (1982). Virology 118, 117-127. Nandi, S . , and McGrath, C. M. (1973). Ado. Cancer Res. 17, 353-414. Neel, B. G., Hayward, W. S . , Robinson, H. L., Fang, J., and Astrin, S. M. (1981). Cell 23, 323-334. Nowell, P. C. (1976). Science 194, 23-28. Nusse, R . , and Varmus, H. E. (1982). Cell 31, 99-109. Owens, A. H., Coffey, D. S., and Baylin, S . B., eds. (1982). “Tumor Cell Heterogeneity; Origin and Implications.” Academic Press, New York. Payne, G. S., Bishop, J. M., and Varmus, H. E. (1982). Nature (London) 295, 209-213. Peters, G., Brookes, S . , Smith, R., and Dickson C. (1983). Cell 33, 369-377. Puma, J. P., Fanning, T. G., Young, L. J. T., and Cardiff, R. D. (1982).J. Virol. 43, 168-165. Roberts, R. J. (1976). CRC Crit. Reu. Biochem. 4, 123-164. Shank, P. R . , Cohen, J. C., Varmus, H. E., Yamamoto, K. R., and Ringold, G. M. (1978).Proc. Natl. Acad. Sci. U.S.A. 75, 2112-2116. Southern, E. M. (1975).J . Mol. B i d . 38, 503-517. Varmus, H. E. (1982a). Cancer Suroeys 1, 309-319. Varmus, H. E. (1982,). Science 216, 812-820. Varmus, H . E., Bishop, J. M., Nowiinski, R. C., and Sarker, N. H. (1972). Nature (London) 238, 189-190. Varmus, H. E., Cohen, J. C., Shank, P. R., Ringold, G. M., Yamamoto, K. R., Cardiff, R. D., and Morris, V. L. (1978). ICN-UCLA Symp. Mol. Cell. Biol. 11, 161-179. Wellings, S. R . , Jensen, H. M., and Marcus, R. G. (1975).J . Natl. Cancer Inst. 55, 231-273. Welsch, C. W., and Nagasawa, H. (1977). Cancer Res. 37, 951-963. Young, L. J. T., Medina, D., DeOme, K. B., and Daniel, C. W. (1971).Exp. Gerontol. 6, 4956. Young, L. J. T., DeOme, K. B., Pitelka, D. R., Blair, P. B., and Cardiff, R. D. (1984). Cancer Res. (in press).

XIPHOPHORUS AC AN IN VlVO MODEL FOR STUDIES ON NORMAL AND DEFECTIVE CONTROL OF ONCOGENES’ Fritz Anders, Manfred Schartl, Angelika Barnekow, and Annerose Anders Genetisches lnstitut and lnstitut fur Virologie (FB Hurnanrnedizin). Justus-LiebigUniversitatGiessen, Giessen, Federal Republic of Germany

I. Introduction and Historical Background. . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . 191

11. Ubiquity of Oncogenes in Purebred Animals Derived ild Populations . . . A. Biology and Taxonomy of Xiphophorus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

B. Insusceptibility of Animals from Wild Populations to Development of Neoplasia . . . . . . . . , . . . . . . . . , . , . . , . . . . . . . . . . . . . . . , , , C. Occurrence of Neoplastically Transformed Cells in Animals from Wild Populations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........... D. Competence for Neoplastic Transformation . . . . . ogenes in An E. Cellular Homologs of R from Wild Populations. . ......... ................ ingdoni . . . . , , , . , . . . . . . . . . . . , . . . . . . . F. The c-src Oncogene in 111. Defective Control of Oncogenes in Hybrids . . . . . . . . . . , . . . . . . . . . . . . . . . . . . . . ,

194 194 197 199 204

C. Assignment of Neoplasia to Chromosomes . . . . . .

207 207 211 211 213 . . . . . . . . . . . . . . . 218

...........

. . . . . . . . . . . . . . . . . . . 220

E. Tissue Specificity or Ti 228 230 A. The Oncogene.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 . . . . . . . . . . . . . . 242 249 D. The Differentiation Gene .......... 25 1 26 1 26 1 . . . . . . . . . . . . . 26 1 262 . . . . . . . . . . . . . . . . 263 268 268 ........................................

I. Introduction and Historical Background

Oncogenes are genes that code for neoplastic transformation and possibly for the maintenance of the neoplastic state of a cell. Although they have been ‘Dedicated to Peter Karlson on the occasion of his 65th birthday. 191 ADVANCES IN CANCER RESEARCH, VOL. 42

Copyright D 1984 by Academic Press. Inc. All rights of reproduction in any form reserved. ISBN 0-12-M)6642-4

192

FRITZ ANDERS ET AL.

deduced from the inheritance of certain tumors many decades ago by geneticists (see Strong, 1958; Lynch, 1967; Bauer, 1968; Heston, 1974), their general significance for tumor development remained widely unnoticed until they became identified as constituents of the genomes of tumor viruses and cells in recent years by virologists (see Bishop, 1982a,b). The history of the discovery of the oncogenes considered in this article can be traced back to the year 1928, when Kurt Kosswig in Muenster, Georg Haeussler in Heidelberg, and Myron Gordon in New York found that certain hybrids between the ornamental platyfish (Xiphophorus muculatus) and the ornamental swordtail (Xiphophorus helleri) are capable of developing melanoma spontaneously. Subsequently these authors showed that the melanomas of the hybrids originated from black spots that were inherited from the platyfish following Mendelian laws. Both the spots and the melanomas were formally assigned to Mendelian genes which were designated as “color genes” (because of the black coloration of the spots and of the melanomas), or “macromelanophore genes” (because of the giant pigment cells found in the spots and in the melanomas). The enhancement of the spots to melanomas was initially believed to be the result of enhanced color gene expressivity in the pigment cells exerted by “intensifier genes” that were assumed to be contributed to the hybrid genome by the swordtail. Thus, positive control of the color genes was believed to be the major cause for melanoma formation in Xiphophorus (Kosswig, 1929, 1937; Gordon, 1931, 1958). Sporadically occurring hybridization-conditioned tumors involving tissues other than the pigment cell system, e.g., ocular tumors (Gordon, 1947b), thyroid tumors (Aronowitz et al., 1951; Berg et al., 1953), and kidney tumors (Baker et al., 1953), were difficult to interpret in terms of formal genetics. Our group started its research on cancer in the platyfish-swordtail system in 1957 when Gordon and Kosswig provided us with some of their fish stocks. During the following 10 years it became clear that development of melanoma and some other types of neoplasia in the fish is due to a certain gene that is endowed with the capacity to mediate neoplastic transformation (F. Anders, 1967). This gene is an oncogene by definition (A. Anders et al., 1973a,b; see Heston, 1974), and was designated as “tumor gene” (Tu) (F. Anders et al., 1974). In contrast to the initial concept of positive control of “color genes” mentioned above it was shown that Tu is normally under negative control by certain linked and nonlinked regulatory genes ( R genes). Following hybridization, chromosomes carrying nonlinked R genes may be replaced by chromosomes lacking them: the oncogene Tu, then may become derepressed to a certain degree and may mediate neoplasia. Thus, spontaneous tumor formation in platyfish-swordtail hybrids was finally recognized as a problem of negative regulation of an oncogene (F. Anders, 1967). Some positive control on tumor growth detected in the derepressed Tu

CONTROL OF ONCOGENES

193

system could be related to metabolites such as amino acids which may serve as nutrient factors in the transformed cells (F. Anders et al., 1962b, 1963, 1969; F. Anders and Klinke, 1965; M . Sieger et al., 1968; F. Sieger et al., 1969), and to hormonal influences (Siciliano and Perlmutter, 1972; A. Anders et al., 1973a; A. Schartl et al., 1982). During the second decade of our studies we found that Tu is not only responsible for the spontaneous development of neoplasms in hybrids but also for germ line mutation-conditioned tumors as well as for the large variety of neurogenic, epithelial, and mesenchymal neoplasms (Abdo, 1979) that can be triggered in somatic cells by mutagenic (A. Anders et al., 1973a; Schwab et al., 1978a,b, 1979; Schwab and A. Anders, 1981) and chromatindamaging agents (F. Anders et al., 1981a), and by tumor promoters (A. Schartl et al., 1982). While genetically conditioned and environmentally triggered neoplasia in Xiphophorus became rather well understood in terms of formal genetics, population genetics, cytogenetics, and developmental genetics (A. Anders and F. Anders, 1978; F. Anders, 1981; F. Anders et al., 1981b; Prescott and Flexer, 1982), its molecular basis remained extremely resistant to any elucidation. New directions for our research arose when we related the concept of the oncogene of endogenous viruses freshly developed by Bentvelzen (1972) to our earlier concept of the tumor gene in the fish genome (see the discussions in A. Anders et al., 1973a; Heston, 1974; Kollinger et al., 1979). The appropriate experiments were undertaken once tumor virologists realized that the oncogenes of certain tumor-mediating retroviruses such as the src oncogene from Rous sarcoma virus (RSV) of chicken are also present in the noninfected cells of the host organisms which are not taxonomically related to the host (see Bishop, 1982a,b). In Xiphophorus the cellular counterparts (c-onc genes) of 10 different avian and mammalian retroviral oncogenes (u-onc genes) were detected in a cooperative work of the laboratories of the Genetics Institute (Giessen) with the laboratories of H. Bauer (Giessen) and R. Gallo (Bethesda), by M . Schartl, A. P. Czernilofsky, and G. Franchini (see F. Anders, 1982; M . Schartl and Barnekow, 1982). One of these onc-genes, namely c-src (the cellular counterpart of the oncogene of Rous sarcoma virus, u-src), was studied in more detail and its expression could be correlated with that of the oncogene Tu (Barnekow ct al., 1982; M . Schartl ct al., 1982; Bauer et al., 1982; F. Anders, 1982, 1983; F. Anders et al., 1983). The present article aims to unify our populational, morphological, developmental, and cell biological findings obtained during research on the biology of the oncogene Tu of Xiphophorus. Furthermore, it aims to show that neoplasia of multicellular animals including humans results from elimination, deletion, impairment, or insufficicncy of regulatory genes that nor-

194

FRITZ ANDERS ET A L .

-RK)Soto la Marim - X x@wdurn Rio k n u c o - X

vanatus

0

Mexico city

x mOcUktUS

Rio Lancetila

\ FIG. 1 Mae of Mexico and adiacent parts of Central America showing- the distribution of these species of Xiphophorus which were mainly used in this study. For details see Kallman (1975).

mally control the oncogene, or from the introduction of uncontrolled accessory oncogenes into the genome. Many earlier and recent findings cited in this article were to date unpublished. II. Ubiquity of Oncogenes in Purebred Animals Derived from Wild Populations

We shall first report some basic facts on Xiphophorus (see Rosen and Bailey, 1963; Kallman, 1975) that are important for an understanding of the model character of our system for studies on neoplasia in general. A. BIOLOGYA N D TAXONOMY OF Xiphophorus

Xiphophorus is a viviparous genus of topminnows inhabiting ditches, rivers, lakes, brooks, ponds, and pools in the Atlantic coastal drainage systems from northern Mexico southeast to northern Honduras (Fig. 1). The peninsula of Yucatan is not inhabited by these animals indicating that the ancestors of the recent genus Xiphophoms were already present in the pliocenic Central America before Yucatan emerged from the sea. Xiphophoms reproduces in closed populations and has evolved into innu-

CONTROL OF ONCOGENES

/

195

gordoni couchianus xiphidium variatus evelynae milleri maculatus andersi nigrensis pygmaeus montezumae cortezi clemenciae alvarezi ~~PMH~~ helleri

'

signum

FIG. 2. Cladogerietic divergence of the presently described species of the genus Xiphophorus. From Radda (1980), modified.

merable, phenotypically distinguishable groups which are isolated geographically or ecologically (see Atz, 1962; Zander, 1967). Seventeen of these groups which differ clearly from each other by their phenotype have been classified as species (Rosen, 1979; Radda, 1980) (Fig. 2). They can be arranged in species groups, as follows: the maculatus species group (the platyfish), e.g., X . mucuZatus (Fig. 3A), X . uuriatus, and X . couchianus; the helleri species group (the swordtails), e.g., X . helleri (Fig. 3B), X . clemenciae, and X . signum; and the montezutnae species group (the montezuma swordtails), e.g., X . cortezi, X . montezumae, and X . pygmueus. Some of these species are endemic to only an extremely restricted area, occupying at best a few spring pools or lakes; others inhabit a single river system; and still others, such as X . inaculatics and X . helleri which are the main species of this research, can be found from the Rio Jainapa (Mexico, near Veracruz; northwest from Yucatan), eastward to the Belize River (British Honduras; southeast from Yucatan), which represents a distance of over 3500 km (Fig. 1). Some species, e.g., X . maculatus and X . helleri, are sympatric, for instance in the Belize River and in the Rio Jamapa; but interspecific hybrids have never been found in the natural habitat (see Kallman and Atz, 1967; Zander, 1967).

196

FRITZ ANDERS ET AL.

FIG. 3. Male (top) and female (bottom) of (A) Xiphophorus maculatus (from a population of the Rio Usumacinta) and (B) Xiphophorus helleri (from Rio Lancetilla) (up to about 8 cm in length).

CONTROL OF ONCOGENES

197

The animals of all taxonomic groups of Xiphophorus can be hybridized in the laboratory, and all hybrids are fertile. This, together with the findings that the degree of enzyme polymorphism is low (A. Scholl, 1973; A. Scholl and F. Anders, 1973a; E. Scholl, 1977; Morizot and Siciliano, 1982), that pairing of the chromosomes in the hybrids during meiosis is normal (Siegmund, 1982; Kollinger and Siegmund, 1981), and that genome organization shows conformity in all cases tested (Schwab, 1982b; Herbert, 1983), led us to the conclusion that the relationship between these taxonomic groups known as species in the literature, actually is at the level of geographical and ecological populations and races comparable to the populations and races of most of the wild animal species and the human species (F. Anders et al., 1981b). Even the large morphological differences between X . helleri and X . muculutus (compare Fig. 3A with 3B) that culminate in the development or lack, respectively, of a “sword” (an elongation of the lower part of the tail fin in the adult males) is of minor taxonomic value: following treatment of newborn X . muculatus with methyl-testosterone a sword develops indicating that even the sword-lacking fish carry the genetic information for the development of the sword like their sword-carrying taxonomic counterparts (Dzwillo, 1964; A. Schartl, 1981).

B. INSUSCEPTIBILITY OF ANIMALSFROM WILDPOPULATIONS TO DEVELOPMENT OF NEOPLASIA Tens of thousands of individuals from wild populations of the different taxonomic groups of Xiphophorus have been collected by several authors (Gordon, Kallman, Borowsky, Siciliano, Zander, and ourselves), but no tumors were detected. Furthermore, almost all species listed in Figs. 1 and 2 have been inbred or bred in closed stocks in our laboratories (e.g., X . helleri from Rio Lancetilla and X . maculatus from Hio Jamapa, collected by Gordon in 1939; inbred for about 90 and 130 generations, respectively), but no tumors developed. The pure-bred descendants of the wild populations also proved to be highly insensitive to carcinogens such as X rays (A. Anders et al., 1971, 1973a,b; Pursglove et al., 1971; Pursglove, 1972; Haas, 1981), Benzo(a)pyrene (Maas, 1967), N-niethyl-N-nitrosourea (MNU) (Schwab et al., 1978a,b; C. R. Schmidt, 1983), N-ethyl-N-nitrosourea (ENU), diethylnitrosamine (DENA), dimethyl sulfoxide (DMSO) (C. K. Schmidt, 1983; Herbert, 1983), and to tumor promoters such as 12-O-tetradecanoylphorbol-13-acetate (TPA) (Schwab, 1982a; C. R. Schmidt, 1983; Herbert, 1983), 17-methyltestosterone, and other steroids (A. Schartl, 1981; A. Schartl et al., 1982)as well as to potential tumor promoters such as saccharine, cyclamate and diazepam (C. R. Schmidt, 1983; Herbert, 1983). The first part of Table I

TABLE I NEOPLASIA I N Xiphophorus 1 YEAR AFTER TREATMENT WITH MNUa A N D X h y s b Number of survivors

X Rays

MNU Pure-bred X. maculatus (Rio Jamapa) X. oariutus (Rio Panuco) X. riphidiurn (Ro Soto la Marina) X. helleri (Rio h c e t i l l a ) X. cortezi (Rio Axtla)

410 -100 100 415 100

3405 -500

-

-100 -2000 -100

Number of neoplasms MNU

X Rays

0 0

0

0

F2-F,;

BCI-BC24

8258

3587

-13,500 10-3 M; four times for 1 hr in %week intervals. lo00 R; three times for 45 min in 6-week intervals.

0

-lo00

470

F1

0 0 0

0 0

-7200 Hybrids

0

18(4%) 826(10%)

0 163(5%) 1007 (7.5%)

CONTROL OF ONCOGENES

199

summarizes material, methods, and results of a recent broad-scale treatment experiment of this kind: about 7200 pure-bred individuals survived treatment with X rays and MNU, but none developed neoplasia (for germ line mutation-conditioned neoplasms that develop in the offspring of the treated wild animals see Section IV,A,2). C. OCCURRENCE OF NEOPLASTICALLY TRANSFORMED CELLSI N ANIMALS FROM WILD POPULATIONS While neoplasms have never been observed in wild Xiphophorus there are strong indications that neoplastic transformation of a certain number of cells is a common process in these animals (F. Anders et al., 1980). Up to the present, however, this process could only be determined in pigment cells of the skin which, because of their natural pigmentation, can easily be observed in vivo. Evidence for transformation of pigment cells in the nontumorous wild fish comes from studies on certain population-specific and species-specific patterns of black spots (Fig. 4). Gross inspection of the spots showed that they are composed of abnormal melanophores which differ from the regular melanophores of the skin by their enormous size and their heavy black pigmentation (Figs. 5 and 6). Gordon (1958) has considered them as an additional type of melanophore in the pigment cell system, namely the “macromelanophore,” which develops following a process of “macromelanophore differentiation.” Out of 9000 adults of X . muculatus collected by him, 1879 had these macromelanophore spots. Morphological, ultrastructural, biochemical, and developmental studies revealed similarities and differences between the regular and the abnormal melanophores which led to the disclosure of the true nature of the abnormal melanophores and of the spots. The regular rnelanophores of the skin (Fig. 5 ) can easily be recognized by their lobulated or dendritic shape and by their content of melanin (BeckerCarus, 1965; Lueken et al., 1973; E. R. Schmidt, 1978). They contain completely melanized melanosomes but lack other cytoplasmic structures such as the endoplasmatic reticulum and the Golgi complexes (Weissenfels et al., 1970; U. Vielkind, 1972). Tyrosinase activity has not been found in the melanophores. They do not proliferate. These features suggest that the melanophores have reached the final stage of pigment cell differentiation. After having reached a certain age, they are removed by macrophages and become replaced by younger melanophores (see Fig. 8). It is important to note that the melanophores are spaced singly and keep a certain distance between each other (Lueken and Kaeser, 1972; Lueken et al., 1973; E. R. Schmidt, 1978). One can assume, therefore, that the dis-

200

FHIlZ ANDEHS ET AL.

FIG.4. Patterns of spots in Xiphophorus consisting of terminally differentiated neoplastically transformed pigment cells. (A) “Spotted dorsal” of X . maculatus from Rio Jamapa;(B) “Spotted” pattern of the same population; (C) “Nigra” of X . maculatus from Belize River; (D) “Lineatus” and “Punctatus” patterns of X . uariatus from Rio Panuco; (E) “Flecked” of X . xiphidium from Rio Soto La Marina, (F) “ D a b b e d of X . helleri from Belize River. See Figs. 1 and 2.

CONTROL OF ONCOGENES

20 1

FIG. 4D-F.

tances between the particular melanophores are controlled by a mechanism like density-dependent regulation in the sense proposed by Holley (1975)for control of cell growth in cultures. The ahnomnul melanophores of the spots of the skin (Fig. 6A and B), also lobulated and heavily pigmented, can easily be recognized by their enor-

FIG.5. Regular melanophores in the skin. Note the equal distances between the cells.

FIG.6. The accumulations of neoplastically transformed melanophores (Tr melanophores) in the skin. (A) Two interlacing Tr melanophores in the early development of a spot; note the surrounding distance-regulated regular melanophores. (B) Spot, composed of many Tr melanophores.

CONTROL OF ONCOGENES

203

mous size as compared to that of the normal melanophores (Breider and Seeliger, 1938; F. Anders et al., 1980). They are multinucleated and highly endopolyploid, and can enlarge up to 0.5 mm in diameter. Like the regular melanophores and abnormal melanophores are not able to divide, show no tyrosinase activity (Schlage, 1978; U. Vielkind et al., 1977), contain only mature melanosomes, and lack the other cytoplasmatic structures (U. Vielkind, 1972, 1976) indicating that they have also reached their final stage of cell differentiation. After they have reached a certain age, they are also removed by macrophages and replaced by younger abnormal melanophores (Lueken et al., 1973; U. Vielkind, 1976; E. R. Schmidt, 1978; Diehl, 1982) (see Fig. 8, Tr cells). The radically different feature of the abnormal melanophores as compared to the regular melanophores is, however, that they are not subjected to distance-dependent regulation. Their lobules and dendrites interlace (Fig. 6A) and the cells grow onto each other, thus forming compact three-dimensional accumulations of some hundred cells each, which appear as the heavily pigmented black spots in the skin (Fig. 6B). As generally accepted by tumor cell biologists, normal cells after having reached a certain density become terminally differentiated and stop dividing, whereas tumor cells are not regulated by density dependency; they remain poorly differentiated and continue to proliferate (see Pierce and Wallace, 1971; Holley, 1975; Prescott, 1976; Prescott and Flexer, 1982). Dysfunction of density-dependent regulation is a fundamental process underlying the change of a cell from the normal to a neoplastic state. The abnormal pigment cells of Xiphophorus, however, although not regulated by density dependency or distance dependency, respectively, are completely differentiated and are incapable of dividing further. These facts suggest that the early pigment cells, after being neoplastically transformed, become restrained from proliferation by genes that constrain them to differentiate to the final stage of abnormal melanophores. Dysfunction of distance-dependent regulation of the abnormal pigment cells, therefore, appears reminiscent of an early transformation event that was posttransformationally “neutralized” by cell differentiation. We will show later that differentiation of the transformed cells is exerted by a certain “differentiation gene” (see Section IV, D). There are also indications for neoplastic transformation of pigment cells in nonspotted animals, for instance in X . inaculatus from Rio Usumacinto. Studies on late embryos and neonates of these animals have shown that during the time in which the immune system matures, single heavily pigmented cells that occur in the skin are rejected as pigmented aggregates of dead cell masses from the surface of the skin (Fig. 7) (M. Schartl, 1979). This process is interpreted at present as an autoimmune rejection of transformed pigment cells. It can be suppressed by the immune-suppressive steroid

204

FRITZ ANDERS ET AL.

FIG. 7. Rejected aggregates of dead pigment cell inasses (arrows) presumably originating from transformed cells. The fish is a neonate of X. rnuculatus from Rio Usumacinto. Adults are shown in Fig. 3A.

prednisolon. Treatment with testosterone (A. Schartl, 1981) or BUdR (HaasAndela, 1978) leads to an acceleration of the rejection of the melanin aggregates. D. COMPETENCE FOH NEOPLASTIC TRANSFORMATION The occurrence of neoplastically transformed inelanophores (Tr melanophores; all transformed cells are designated Tr cells in the following sections) in the fish raises the question of the stage of differentiation in which the pigment cells are capable of changing from the norinal state to the neoplastic state. This stage of cell differentiation is considered the competent stage for neoplastic transformation, To detect this stage we studied first the norinal differentiation of the pigment cells. According to Fitzpatrick and Lerner (1953) and Bagnara and Hadley (1973), we denote the very early coininon precursors of all types of pigment cells as chromntoblasts, the embryonic cells potentially capable of producing inelanine as mlnnoblasts, the mature but not yet terminally differentiated melanin-producing cells as melanocytes, and the terminally differentiated melanin-containing cells as inelanophores (for these latter cell-types see Section 11,C). The pigment cells of vertebrates originate froin neural crest cells (DuShane, 1938; Weston, 1970). I n Xiphophorus (see Fig. 8) (Humm and Young, 1956; F. Anders et nl., 1979a; Diehl, 1982) at the fourth day of embryonic life the neural crest cells start migrating. Those entering their

205

CONTROL OF ONCOGENES NORMAL MELANOPBORE

MALIGNANT AND BENIGN MELANOMA; SPOTS

PATTERN

Different proportions of the different stages of c e l l differentiatlon

Homeostasls between the dlfferent stages of cell differentlatlon

Melanoohores / Macrophages \Tr

I‘ Melanocytes

-, .z

+

CJ

fi Melanoblasts

Diff

z=

‘

Melanophores I \

Tr Melanocytes

I

c I - g e n e s -c J b

-

Al

1 7 system

FIG.8. Schematic presentation of the differentiation of normal and neoplastically transformed pigment cells. S, I, and A melanoblasts are stem, intermediate, and advanced melanoblasts, respectively. The T r cells represent the transformed cells. Only I melanoblasts are competent for neoplastic transformation. Tu,tumor gene (oncogene) (see Section IV,A); R M ~ ,regulatory , gene for control of Tu in the melanophore system (see Section IV,B); R N ~ RE^^, ~ , RM,,.~, regulatory genes controlling Tu in the nervous cell system, the epithelial tissues and the mesenchymal tissues, respectively; Rco, compartment-specific regulatory genes (see Section IV,C); g, “golden” gene that blocks pigment cell differentiation; Dqf, differentiation gene (see Section IV,D); I genes, intensifier genes, which support proliferation of poorly differentiated transformed pigment cells. Macrophages attack melanophores and Tr melanophores. From A. Anders and F. Anders (1978), modified.

final locations, including the corium of the skin and the extracutaneous connective tissues of the peritoneum, pericardium, dura mater, fascia abdominalis, chorioidea, and vascular structures (Peter, 1982, 1983), become determined to differentiate to pigment cells. These cells are considered to be chromtoblasts, which are the common precursors of all pigment cells (chromtophores) including pterinophores, purinophores, and melanophores. Those chromatoblasts committed to differentiate to melanophores give rise to stem cells for melanophore differentiation (S melanoblasts). These may reproduce throughout the life of the fish, but may also differenti-

206

FRITZ ANDERS ET AL.

ate to an intermediate stage of melanoblasts ( I melanoblasts). All cells that have reached this stage continue differentiation to the most advanced stage of melanoblasts (A melanoblasts). These cells can be distinguished from their precursors by their reaction to dopa. The first A melanoblasts occur in the skin of the 5-day-old embryos. They may differentiate within about 15 hr to melanocytes. The melanocytes differentiate to melanophores, the first of which occur in the 11-day-old embryo. In contrast to their precursors, they do not divide and are removed by macrophages after they have reached a certain age. A homeostasis exists between the different stages of melanophore differentiation, which is controlled by distance-dependent regulation (see Section I1,C). Evidence for the restriction of competence for neoplastic transformation to a certain stage of pigment cell differentiation comes from two lines of observations (F. Anders et a l . , 1972). (1)A certain mutant (golden gg) (A. Anders et al., 1973a), in which the melanophore differentiation is almost completely blocked at the stage of the S melanoblasts (see gin Fig. 8), fails to differentiate Tr cells, indicating that S melanoblasts as well as earlier stages (chromatoblasts, neural crest cells) are noncompetent for neoplastic transformation. (2) A melanoblasts, melanocytes, and melanophores, which can easily be recognized, have never been found to undergo neoplastic transformation. These cells appear to be too advanced in the normal process of differentiation, and therefore have lost the competence for transformation. The cells competent for neoplastic transformation, therefore, are the I melanoblasts. The I melanoblasts, after being transformed to TrI melanoblasts, differentiate to the easily recognizable, proliferating TrA melanoblasts. These Tr cells differentiate to the heavily pigmented, endopolyploid Tr melanocytes which proceed to the terminal stage of differentiation of transformed pigment cells, represented by the abnormal melanophores (see Section 11,C), i.e., the Tr melanophores. As will be shown later, spots and melanomas differ in the different proportions of Tr cells found in the different stages of pigment cell differentiation: Tr melanophores represent the predominant cells of the spots and the benign melanomas while TrA melanoblasts and Tr melanocytes represent the majority of cells of the malignant melanoma. We assume that competence to neoplastic transformation is, in general, restricted in all organisms to a certain stage of cell differentiation in any one of the organisms systems, and that Tr cells in any case may differentiate to a terminal stage, in which they, although still neoplastically transformed, are no longer deleterious to the organism; this assumption is supported by many observations from under systems (see Pierce and Wallace, 1971; Sachs, 1978, 1982).

CONTROL OF ONCOGENES

207

E. CELLULAR HOMOLOGS OF RETROVIRALONCOGENES I N ANIMALS FROM WILD POPULATIONS. Our interest in the cellular homologs of retroviral oncogenes in Xiphophorus came from several loosely connected lines of observations, and from the literature: (I) the ubiquity of transformed cells in all individuals of the fish (Section II,C), (2) the inheritance of genetic information for neoplastic transformation (Sections III,C,D), and (3) the identification of c-src, i.e., the cellular homolog of the Rous sarcoma virus transforming gene, in the genome of various vertebrates such as uninfected chicken (Stehelin et al., 1976), mouse, calf, humans, and salmons (Spector et al., 1978). These facts raised the question whether cellular oncogenes (c-onc genes), equivalent or even homologous to retroviral oncogenes (u-onc genes), are ubiquitously present in Xiphophorus, and whether they play any role in neoplasia. The cellular counterparts of all 10 u-onc genes tested by molecular hybridization have been detected in DNA from X . helleri and X . maculatus with u-onc probes (Fig. 9) carried out by M. Schartl and Czernilofsky (1982) and M. Schartl and Franchini (1982) (see F. Anders, 1982; M. Schartl and Barnekow, 1982). From one of these c-onc genes, namely the c-src, the RNA transcript was identified (M. Schartl and Franchini, 1984), and a phosphoprotein was detected that, as shown by its antigenic, biophysical, and enzymatic properties is closely related to the transforming protein of the RSV u-src oncogene (pp60"-src) (Fig. 10): it was immunoprecipitated with antiserum against p ~ 6 0 " - has ~~~ an, estimated molecular weight of 60,000, and is associated with a kinase activity which phosphorylates the tyrosine residue in the heavy chain of anti-pp6OsrC IgG. This phosphoprotein is, therefore, assumed to be the product of c-src of the fish, i. e., the pp60c-src (Barnekow et al., 1982). As measured by means of the pp60c-src kinase assay according to Collet and Erikson (1978) (see Collet et al., 1980) (Fig. ll), c-src is active in all individuals of all wild populations of the different taxonomic groups of Xiphophorus tested (Table 11). Brain tissue shows always a considerably higher kinase activity as compared to that of skin, liver, spleen, and testes. Muscle tissue showed no or a barely detectable kinase activity (Fig. 12). The fact that the oncogenes c-src is active in the nonneoplastic cells of the pure-bred animals indicates that neither the oncogene itself nor its product is necessarily correlated with tumor development (Barnekow et al., 1982; M. Schartl et al., 1982).

F. THE c-src ONCOGENE IN

THE

ANIMALKINGDOM

The identification of c-src in the genome of various vertebrates such as chicken (Stehelin et al., 1976), mouse, calf, humans, salmon (Spector et al.,

208

A

FRITZ ANDERS ET AL.

B c-onc

Protypic virus

c-erb

Avian erythroblastosis virus ( 0 )

c-src

Rous sarcoma virus (b)

c-myc

~ v i a nm y e l o c y t o m a t o s i s

c-abl

Abelson nurine leukemia virus

c-fes

Gardner-Arnstein

c-myb

Avian myeloblastosis virus

virus(^)

feline sarcoma virus

c-rasH Harvey murine sarcoma virus

c-ras'

Kirsten murine sarcoma virus

c-sis

Simian sarcoma virus

c-yes

Y73 sarcoma virus

FIG.9. Cellular honiologs of viral oncogenes detected in Xiphophorus. (A) Southern blots of Xiphophorus total genoinic DNA digested with EcoRI, hybridized to nick-translated viral onc probes (a, u-erb; b, u-src; c, u-myc). (B) c-onc genes so far found in Xiphophorus. Viral onc probes were gifts from R. C. Gallo and K. Toyoshima, in collaboration with G. Franchini.

1978), and four different species of Xiphophorus from eight different localities (Barnekow et al., 1982; M . Schartl et al., 1982) (Table 11) led us to a more systematic search for this oncogene in additional taxonomic groups of the animal kingdom, and in the plant kingdom. First, different fish genera of the family of Poeciliidae that are more or less taxonomically related to Xiphophorus were investigated. All fish tested showed a pp60c-src kinase activity indicating that c-src is present in their genome (Barnekow et al., 1982). Second, c-src has been detected partly by molecular hybridization, partly by serological identification of pp60c-src, and partly by the tyrosine kinase activity of pp60c-src in a large variety of other metazoa ranging from mammals, coelenterates (Fig. 13), down to the sponges (Schartl and Barnekow, 1982). Our results together with those of Spector et al. (1978)obtained with humans, calf, and salmon and those of Shilo and Weinberg (1981) obtained with Drosophila are listed in Table 111. All unicellular eukaryotes tested, including the protists Euglena, Cryp-

CONTHOL OF ONCOGENES

209

FIG. 10. Ilemonstration of "2P-labeled ppG0c-sVC in fish brain extracts. Left track: Imrnunoprecipitation of fish brain extracts, labeled before with ["2P]Pi in oitro, with antisera from RSV tumor-bearing rabbits (TBR serum). The M, 50,000 protein is presumably a degradation product of ppGOC-Sr". Hight track: Iininunoprecipitation of the same extracts with normal rabbit serum. Both G0K and 52K proteins could not be detected. From Barnekow et al. (1982).

tomonas, Chlorogoniuin, Paramecium, Tetrahymena as well as the colony forming Volvox, were negative for the tyrosine phosphorylating kinase activity. The same is true for Trichoplax adhaerens, which is regarded to represent an intermediate form between the protozoan and metazoan organization. Algae and higher plants also showed no kinase activity. The ubiquity of c-src in metazoa raises the idea that this homolog of the viral oncogene v-src might have still unknown basic functions closely related to the evolution of the inulticellular organization of the animals, and that neoplasia might be a character that is closely related to this evolution. As the sponges are known to have evolved in the proterozoicum, the origin of the csrc oncogene has to be estimated at over 1.5 x lo9 years ago. At present several groups are trying to determine whether a cellular oncogene, such as c-src, is capable of mediating neoplastic transformation like its viral counterpart (Oskarsson et al., 1980; Pulciani et al., 1980; Reddy et al., 1982; de Feo et al., 1981; Takeya and Hanafusa, 1982; de Klein et al., 1982; Tabin et al., 1982). If this should be proven we suggest that all individuals of all metazoa are endowed with the capacity to develop neoplasia. Support for this idea comes from the fact that neoplasia is distributedalthough sporadically-in all groups of multicelIular animals (Huxley, 1958;

210

FRITZ ANDERS ET AL.

P;[I

ATP

5 05 - PAGE

32P- I g G H E A V Y CHAIN ( 5 3 K ) . D E T E C T E D B Y AUTORADIOGRAPHY

FIG. 11. Assay for pp60c-sI~ckinase activity according to Collet and Erikson (1978). See Barnekow et al. (1982).

Dawe and Harshbarger, 1969, 1975; Krieg, 1973; Kraybill et al., 1977; Dawe et al., 1981; Kaiser, 1981). Ubiquity of oncogenes in the wild animals on the one hand and the infrequent occurrence of neoplasia in these animals on the other hand raises the question of the mechanisms that protect the majority of the individuals of all metazoans from the action of their own oncogenes (F. Anders, 1981; F. Anders et al., 1979a,b, 198lb). The Xiphophorus model provides an opportunity to contribute to the study of this problem. This TABLE I1 EXPI<ESSION OF pp60r-Src-ASSOC1ATEDKINASE IN BRAINEXTHACTS OF Xiphophorus OF DIFFERENT PHOVENANCti Species

X. X. X. X. X. X. X. X.

helleri helleri inaculatus maculatus maculatus corteai

variatus oariatus a

+ +, 100-200

Population

Kinase activitya

Belize River Rio Lancetilla Belize River Rio Jamapa Rio Usumacinta Rio Axtla Rio Coy Rio Panuco

+++ +++ +++ ++ ++ +++ ++ ++

cpm/mg protein; + + +, 200-400 cpm/rng protein.

CONTROL OF ONCOGENES

211

FIG. 12. Demonstration of pp60L'-Srcassociated protein kinase activity in extracts of different fish organs. 53K = 32P-labeled IgG heavy chain, detected by autoradiography (see Fig. 11). Autoradiograms show equal amounts of protein for each sample. Note that brain shows the highest kinase activity. See Fig. 28.

contribution comes from studies on defective control of the oncogenes which has been observed in certain hybrids (see Section 111). Ill. Defective Control of Oncogenes in Hybrids

In contrast to the purebred descendents of wild populations of Xiphophorus that are highly insusceptible to neoplasia (see Section II,B), certain interpopulational, interracial, and interspecific hybrids are susceptible.

A. SUSCEPTIBILITY OF HYBRIDS TO DEVELOPMENT OF NEOPLASIA Following treatment with carcinogens, depending on the wild populations used for hybridization, about 1 to 4% of the first hybrid generation (F,) develop neoplasia. Tumor incidence increases in the second hybrid generation (F,) and the first backcross generation (BC,) up to about 8% and remains stable in the succeeding generations, which in the case of the backcrosses have been tested up to BC,, (Table I, second part). As compiled from the results of several investigators in our laboratories (A. Anders et al., 1973a, 1983; Schwab et al., 1978a,b, 1979; Kollinger et al.,

212

FRITZ ANDERS ET AL.

FIG.13. Identification of phospho-amino acids in "1'-labeled IgG heavy chain of TBR serum precipitated sea anenione extracts and subsequent perforniance of the kinase assay. Note that the phosphorylation site is exclusively tyrosine. See legends Figs. 10-12, and Barnekow et al. (1982).

1979; Haas, 1981; C. R. Schmidt, 1983), 805 of 10,195 (8%) hybrids which survived treatment with M N U and X rays developed a large variety of different neoplasms (Table IV; Fig. 14). Most of the neoplasms were classified as neurogenic and mesenchymal, with melanoma, neuroblastoma, and fibrosarcoma being the predominating types (Schwab et a l . , 1978a,b; Schwab and A. Anders, 1981; Abdo, 1979). Epithelial neoplasms were less frequent but comprised those with the largest diversity. Some individuals developed several tumors of different types, for instance melanoma, neuroblastoma, and rhabdomyosarcoma. Almost all tissues including the respiratory, endocrine, exocrine, excretory, reproductive, gastrointestinal, re-

CONTROL OF ONCOGENES

213

TABLE 111

c-src Mammals Humans Calf Rat Mouse Birds Chicken Quail Frogs

Xenopus

Bony fish Poeciliidae

Xiphophorus Girardinus Poecilia Belonesox Heterandria Xenotuca Flat fish Sea robin Mackerel Roach Gudgeon Salmon Codfish Cichlid

IN

EUKAHYOTES Cartilaginous fish Shark Jawless fish Lamprey Acrania

Amphioxus Insects Cockroach

Mollucs Cuttle fish (?) Coelenterates Sea anemone Sponges Marine sponge Freshwater sponge

Drosophila

ticuloendothelial, hepatobiliary, skeletal, muscular, nervous cell, and pigment cell system developed tumors. The tumors show some fish-specific features, but their specific structure and growth are essentially identical to that of the corresponding tumors of other vertebrates (see Schlumberger and Luck6, 1948; Schlumberger, 1957; Scarpelli, 1969; Abdo, 1979) including humans (Sobel et al., 1975; Riehl et al., 1984). €3. HYBRIDIZATION AS A STEPTOWARD NEOPLASIA

The occurrence of individuals susceptible to neoplasia among hybrids brings about the question whether this phenomenon is unique for Xiphophorus or represents a more or less general phenomenon in the animal kingdom. A survey of the literature (F. Anders, 1968, 1981; F. Anders et al., 1979b, 198lb,c) and the experience of pathologists (see Weiss, 1972) illustrates the general rule that (1)the incidence of spontaneously developing neoplasms is low in pure-bred animals from wild populations, and neoplasia is difficult to induce in these same animals, while (2) the incidence of spontaneously developing neoplasms is high in animals of hybrid origin, and neoplasia is easily inducible in these animals (e.g., naturally occurring or experimentally produced interspecific and interpopulational hybrids, domestic, laboratory, and ornamental animals, pets) (see Stunzi, 1972; Hayes, 1978). Many examples of neoplasia in hybrids and improved breeds of hybrid origin in the animal kingdom have been cited in the literature: (1)insects: experimental hybrids and laboratory stocks, e.g., Drosophila (Gateff, 1978,

214

FRITZ ANDERS ET AL.

TABLE IV NEOPLASMI N Xiphophorus HYBRIDS(F2-F,; B C L - B C ~ ~1 )YEAR AFTER TREATMENT WITH MNU A N D x R A Y S a ~ b ~ C ~ d

Number of animals which developed neoplasmse Type of neoplasm

MNU

Neurogenic Melanoma (benign) Melanoma (malignant) Neuroblastoma Epithelial Squamous cell carinoma Epithelioma Carcinoma (low-differentiated) Carcinoma (high-differentiated) Adenocarcinoma (kidney) Adenocarcinoma (thyroid) Papilloma Hepatonia Acanthoma Mesenchymal Fibrosarcoma Rhabdom yosarcoma Lymphosarcoma Re ticulosarcoma Total

135 138 84

6 19 3 2 8 2 9 5 3 190 331 4

642

X Rays

Percentage incidence based on total number of survivors MNU

X Rays

491

7

2.12 2.09 1.27

2.6 0.95 0.2

0 6 4 5 2 3 0 1 0

78

0.09 0.28 0.05 0.03 0.12 0.03 0.14 0.07 0.04

0 0.17 0.11 0.14 0.05 0.08 0 0.03 0

12.36

2.87 0.5 0.01 0.06

0.17 0.05 0 0

:i }

' 0

I

0

163

805

Classification according to Mawdesley-Thomas (1975). Total number of survivors: 10,195 (MNU, 6608; X rays, 3587). MNU, 10W3M; four times for 1hr in 2-week intervals (Schwab et al., 1979).X Rays, lo00 R; three times for 45 min in 6-week intervals (Pursglove et al., 1971). Out of 10,195, 805 (7.9%)hybrids developed neoplasia. Several animals developed several different kinds of neoplasms. @

1982), Solenobia (Seiler et al., 1958); (2) fish: naturally occurring hybrids, e.g., Lake Ontario hybrid carp (Leatherland and Sonstegard, 1978), ornamental fish, e. g., red swordtail (Xiphophorus), guppy (Poecilia reticulata) (Sato et al., 1973), Cirardinus (A. Schartl et a l . , 1982), orange medaka (Oryzias)(Takayama and Ishikawa, 1977), goldfish (Ishikawa et al., 1978a), ornamental carp (Ishikawa and Takayama, 1977), domestic fish bred for economic reasons, e.g., domestic carp (Ishikawa et al., 1978b), domestic trout (Halver and Mitchel, 1967; Sinnhuber et al., 1977); (3) birds: ornamental

CONTROL OF ONCOGENES

215

hybrids, e.g., musk duck x mallard duck (Crew and Koller, 1936), peacock x guinea fowl (Poll, 1920), improved breeds of domestic chicken (Weiss, 1972);and (4) mammals: interspecific hybrid mice, e.g., Mus musculus x M . buctrianus (Little, 1947), laboratory mice strains, e.g., blue ribbon mice (Heston and Vlahakis, 1968), hybrids between laboratory rat strains, e.g., BALB/c x NZB (Warner et al., 1974), hybrids between laboratory rat strains, e.g., Sprague-Dawley x Long-Evans (Gross and Dreyfuss, 1979), domestic dogs, e.g., boxer dog (Stunzi, 1972; Weiss, 1972; Denlinger et al., 1978), cats. cattle, swine, e.g., Sinclair swine (Hook et al., 1979), horses, e.g., Lippizaner (Gebhard and Niebauer, 1979), etc. The phenomenon of introducing susceptibility to neoplasia by means of hybridization is not limited to the animal kingdom. Susceptibility to neoplasia has also been observed in a large variety of plant hybrids, especially in cultivated plants that are mainly bred by hybridization. Hybrids of cabbage, lilies, tobacco, tomatoes, calanchoe, thorn-apples, poplar, etc., are wellknown examples (see Ahuja, 1965; F. Anders, 1968; F. Anders et al., 1981b,c; Beiderbeck, 1977; Braun, 1978). While there are no data on the relation between hybridization and cancer in human beings comparable to those in animals and plants, it is interesting to speculate whether the many facts on the high tumor incidence in humans that do not agree with the concept of the primacy of environmental factors and life style in carcinogenesis (Burch, 1976; Higginson, 1969; Maugh, 1979; Oeser and Koeppe, 1979; Koeppe, 1980) may be explained by civilizationconditioned interpopulational and interracial hybridization in our preceding generations. Such speculations are probably of little value in the fight against cancer. They might, however, help to find those factors that make a particular individual susceptible or insusceptible to cancer, or sensitive or insensitive to carcinogens (Heston, 1974). Interpopulational and interracial human mating (we use the term population and race in the sense of “Mendelian population” and “Mendelian race” proposed by Dobzhansky, 1937) may have occurred at any time and place. Because of the high mobility of humans as compared to other mammals one should expect high values of genetic heterogeneity. Various estimates based on enzyme variation showed that heterogeneity in humans (Schull, 1979) is comparable to that of domestic mammals, such as cats (O’Brien, 1980), but is about 6 times higher than that of wild macaques, about 10 times higher than that observed in big wild mammals such as elk, moose, polar bear, black bear, and elephant seal, and about twice as high as that of most feral rodents studied so far (Fuerst et ul., 1977). On the basis of these data one could assume that the high tumor incidence in humans might also be related to hybridization like that in domestic animals. There are also some data on chromosomal heteromorphisms in human

FIG. 14. Xiphophorus ~ ~ c u ~ Q ~ u helleri s - X . hybrids exhibiting A: MNU-triggered fibrosarcoma; B: different MNU-triggered fibrosarcorna (a) and rhabdornyosarcorna (b);C: MNU-trig216

gered signet ring cell carcinoma; D: spontaneous amelanotic melanoma on an albino (note vascularization). From C. R . Schmidt (1983). 217

218

FRITZ ANDERS ET AL.

populations that might be useful for estimates of heterogeneity within and between different populations. According to such estimates it appears that, for instance, Japanese populations exhibit a low degree of Q- and C-band chromosome heteromorphisms, whereas Americans have a much higher degree of this heteromorphisms, with blacks showing even more variability than whites (Lubs et al., 1977; Yamada and Hasegawa, 1978). The same gradation is also reported for the incidence of neoplasia in Japanese and Americans (Higginson, 1969; see Maugh, 1979; Muir and Nectoux, 1982; Young and Pollack, 1982). One is tempted to propose that the chromosomal heteromorphism reflects differences in the degree of heterogeneity and therefore differences in tumor incidence between the Japanese and the white and black United States populations. We suggest that these differences in tumor incidence are due to different degrees of interpopulational and interracial matings in nations thereby affecting genetic heterogeneity as it does in Xiphophorus. On the other hand, the extremely low tumor incidence of active Mormons and Seventh-Day Adventists, as compared to total United States whites (see Cairns et al., 1980; Gardner, 1980)might be due to the biological homogeneity of their populations (which favors insusceptibility to cancer) rather than to environmental factors. The same could apply to the low tumor incidence in Japan as compared to that of the other industrial nations. We feel these differences in tumor incidence cannot be explained sufficiently by environmental carcinogenic influences, because the environment differs only to a low degree. They also cannot be explained sufficiently by racial differences: natural selection will not favor one race and discriminate against the other but it will work against susceptibility to cancer in all populations and all races (for details see F. Anders, 1981; F. Anders et al., 1979b, 1981b,c). Neoplasia in Xiphophorus may serve as a model system suitable for studying how much hybridization between members of differently evolved human populations might have influenced tumor incidence in our modern industrial nations.

C. ASSIGNMENTOF NEOPLASIATO CHROMOSOMES To study the crucial differences between the fish that are insusceptible to neoplasia and those that develop neoplasia following treatment with carcinogens, we attempted to assign the susceptibility to specific tumors to specific chromosomes. For this purpose 65 defined genotypes of X . m c ulatus, X . xiphidium, X . variatus, X . cortezi, X . helleri, and their hybrids were employed (Schwab, 1980; Schwab et al., 1978a,b, 1979; Schwab and Anders, 1981). These genotypes exhibit, or lack, specific color patterns (Gordon, 1947a; Atz, 1962; Kallman, 1968; Kallman and Atz, 1967; A. Anders et

CONTROL OF ONCOGENES

F.- hvbrrd

219

x b / / e r i- d

Backcross generation

FIG. 15. Crossing scheme showing the assignment of carcinogen-triggered neoplasms (melanomas, neuroblastomas, epitheliomas, carcinomas, fribromas, sarcomas) to backcross segregants exhibiting a reddish coloration together with pale black stripes. Both pattern of coloration and susceptibility to neoplasia (sensitivity to carcinogens) in the backcross hybrids depend on the same chromosome which, in this experiment, is an X chromosome inherited from Xiphophorus macultatus.

al., 1973a,b) or enzyme markers (A. Scholl, 1973; A. Scholl and Anders, 1973a,b; Ahuja et al., 1980; E. Scholl, 1979; Siciliano and Wright, 1976; Morizot and Siciliano, 1979, 1982)which are due to the expression of specific genes located on different chromosomes. We used mainly backcrosses, which were selectively bred for a specific phenotypic marker, and thereby for a specific chromosome. Such backcross generations (BC) segregate into 50% animals carrying the chromosome marker, and 50% lacking this chromosome. Depending on the wild populations used in the crossings, neoplasia could be assigned specifically to defined chromosomes. The crossing scheme of Fig. 15 shows an example: in the cross of the insusceptible X . nmculatus exhibiting a pale reddish coloration and some longitudinal dark stripes, with the also insusceptible X . helleri exhibiting a homogeneous gray coloration, almost exclusively the BC segregants exhibiting the reddish coloration and the stripes were sensitive to the carcinogens while the homogeneous gray segregants were, with few exceptions, insensitive. The susceptible fish de-

220

FRITZ A N D E R S ET AL.

veloped those kinds of neoplasms which are listed in Table IV; some of these hybrid individuals developed several tumors of different tissue types, for instance fibrosarcoma and rhabdomyosarcoma (Fig. 14B), or melanoma, neuroblastoma, and carcinoma. This experiment clearly shows that the loci coding for the pattern of coloration and for susceptibility to develop different neoplasms are linked to the same chromosome. This is not to say that all BC segregants exhibiting the reddish coloration and the stripes develop neoplasia; but the fish in which the neoplasms developed belong almost exclusively to that group of BC segregants that exhibits the reddish coloration and the stripes. In the present experiment both characters, susceptibility to neoplasia and the color pattern, belong to linkage group I (the sex chromosomes); they also contain loci coding for the enzymes adenosine deaminase, glucose-6-phosphate dehydrogenase, and 6-phosphogluconate dehydrogenase (Morizot et al., 1977). Analysis of sex determination of the fish used in this experiment revealed that the sex chromosome to which the different kinds of neoplasms could be assigned is the X chromosome of X . muculatus (A. Anders and F. Anders, 1963; Schwab, 1980). Other crossing protocols using different pure-bred strains of Xiphophorus have shown that the chromosome on which susceptibility to develop neoplasia depends may be a Y chromosome, or an autosome. Other examples, again, have shown that susceptibility to neurogenic and epithelial neoplasms depends on a certain chromosome (e.g., and X chromosome) while susceptibility to neoplasms of mesenchymal origin depends on another chromosome (e.g., a Y chromosome or an autosome) (Schwab et al., 1978b; Schwab and Anders, 1981).

D. ASSIGNMENTOF CANCER SUSCEPTIBILITY TO ONCOGENES A N D REGULATORY GENES The assignment of carcinogen-triggered neoplasia to specific chromosomes raised the question whether the neoplasms could be assigned to specific genes. The pursuit of this problem proved to be extremely difficult because the development of carcinogen-triggered neoplasia is a rather rare event (see Table I) and, although assignable to chromosomes, would require immense efforts for the production of sufficient experimental material for a finer genetic analysis. The rationale for the design of the experiments that raised the opportunity to assign the carcinogen-triggered neoplasms to genes is based on studies on the occurrence of hereditary neoplasms such as melanomas, pterinophoromas (pigment cell tumors that consist predominantly of drosopterine-containing cells; Henze et al., 1977; Hempeters et al., 1981), neuroblastomas, thyroid carcinomas, kidney adenocarcinomas, and reticulosarcomas, in certain hybrids.

CONTROL O F ONCOGENES

221

1. Oncogenes and Regulatory Genes in Animals Developing Neoplasia Spontaneously The individuals of laboratory hybrid populations of Xiphophorus (wild hybrid populations do not exist; Kalhnan and Atz, 1967; Zander, 1967; see Section II,A) that develop neoplasia spontaneously are extremely rare. They can, however, be propagated in large numbers at the will of the experimenter, because they and their descendants develop neoplasia exactly following Mendelian predictions (Kosswig, 1937; Gordon, 1947a, 1958; F. Anders, 1967; Kallman, 1975). Hundreds of thousands of individuals that develop neoplasia following Mendelian laws have been studied. To illustrate the assignment of cancer susceptibility to distinct genes we used melanoma because development of this type of tumor can easily be observed. Even a singly transformed pigment cell can be distinguished from a regular pigment cell by gross examination of the living animal (see Section II,C, D). Furthermore, morphological, ultrastructural, and histochemical markers specific to the successive stages of pigment cell differentiation facilitate the detection of the first events involved in neoplastic transformation (see Section I1,D). These facts, and the ease with which crossings in Xiphophorus can be accomplished, provided the background for the design of the two series of crossing experiments that elucidated the existence of oncogenes and regulatory genes by formal genetics (F. Anders, 1967) (Figs. 16 and 17). (1)The animals used for the initial crosses were females of a mutant stock of X . inaculatus (platyfish) derived from the wild population of Rio Jamapa (Fig. 16A), and males of X . helleri (swordtail) from Rio Lancetilla (Fig. 16B; see Section 11,A).The platyfish infrequently exhibit spots consisting of transformed pigment cells that are terminally differentiated (Tr melanophores) whereas the swordtails are never spotted (see Section 11,C). The development of the spots is restricted to the dorsal fin and to the skin of the posterior part of the body of the platyfish, and is X chromosome linked. Crosses of the platyfish (A) with the swordtail (B) result in F, hybrids (C) that develop uniformly in all individuals melanomas consisting mainly of Tr melanophores that are similar to those of the spots of the parental platyfish. These melanomas are benign. They occur only in those compartments of the body where the platyfish parent infrequently exhibits the spots. In older F, animals, the melanoma of the dorsal fin and that of the posterior part of the body combine to form a large superficially spreading benign melanoma. Backcrosses of the F, hybrids with the swordtails as the recurrent parent (D) result in offspring (BC,) exhibiting three types of segregants: 25% of the BC, (E) develop benign melanoma like that of the F,, 25% (F)develop malignant melanoma consisting mainly of incompletely differentiated Tr cells which invade other tissues (except for brain, gonads, intestine) and kill the fish,

FIG. 16. Crossing procedure for the production of melanoma developing hybrids of Xiphophorus. (A) X. maculatus from Rio Jamapa; some small spots in the skin of the dorsal fin and the side of the body are visible. Spots consist of terminally differentiated neoplastically transformed pigment cells. (B) X. helleri from Rio Lancetilla, always lacking spots. (C) F1 hybrid developed benign melanoma instead of spots (100%of the FJ. (D) X. helleri from B used in the backcross as the recurrent parent. (E) Backcross hybrid with developing benign melanoma (25% of the BC generation). (F) Backcross hybrid with developing malignant melanoma (25%of the BC generation. (G, H) Backcross hybrids that do not develop melanoma (50% of the BC generation). From F. Anders et al. (1984). modified.

CONTROL OF ONCOGENES

223

FIG.17. Crossing procedure for the suppression of melanoma in Xiphophorus. (A) Malignant melanoma bearing backcross hybrid according to Fig. 16F. (B) X. maculatus according to Fig. 16A as the recurrent parent. (C) Quasi-FLexhibiting benign melanoma. (D)X . maculatus as the recurrent parent. (E, F, G, H) Backcross hybrids (quasi-X. maculatus) exhibiting spots only. From F. Anders et al. (1984), modified.

whereas 50% (G and H) develop neither spots nor melanomas. Further backcrosses (not shown in Fig. 16) of the fish carrying benign melanoma, with the swordtail, result in a BC, that exhibits the same segregation pattern as the BC,. The same applies for further backcrosses of this kind. Backcrosses of the fish carrying the malignant melanoma with the swordtail show a different result: 50% of the BC segregants develop malignant melanoma, whereas the remaining 50% are melanoma free; benign melanomas do not occur. Whenever melanomas occur in these crossing experiments, they develop in both the compartment of the dorsal fin and the compartment of the posterior part of the body. (2)In contrast to the above results (Fig. 17A-H), backcrosses of malignant melanoma bearing hybrids (a) with platyfish (B) (like those of Fig. 16A, but lacking the spot factor by deletion of a Giemsa band; see Section IV,A,3) result in a quasi-F, that segregates in 50% animals displaying benign melanoma (C) and 50% exhibiting neither melanomas nor spots (not shown in the

224

FRITZ ANDERS ET AL.

Est-1

c-SfC

F, hybrid

X RDf’RPD‘RMe/’TU

- - - - - - -- DiffEst-1 - - -- - - -

-

Est-1-

C-SIT

--

X . helleri

a BCl

--

_--------

D/ff

3

I

i

-

““‘”’-.

.

‘Tu X

- - - - - -- ----------

c - -- - - - - - -

- - - - - - - - - I

--- ---- -DiffFst-7

D/ff

1291

8 :II:I--:I - - - - - - -- -

FIG. 18. Crossing scheme according to Fig. 16, which displays the genetic conditions for the “spontaneous” development of spots, benign melanoma, and malignant melanoma. Solid lines, chromosomes of X. maculatus; dashed lines, chromosomes of X. helleri; Tu, tumnor gene; R’M,!, impaired regulatory gene specific to the pigment cell system; R r g and R‘,f, impaired regulatory genes controling Tu in the compartments of the posterior part of the body (Pp) and the dorsal fin (DO; D v f , regulatory gene controlling differentiation of neoplastically transformed cells; Est-I, locus for esterase-1 ofX. maculutus. Note that Tu, R ’ M e , , R ’ p p , and R’Dfare linked to the X chromosome, whereas Dgf and Est-I are linked to an autosome. E s t , esterase profiles (polyacrylamide gels stained using a-naphthyl-acetate as substrate). Note Est-1 isozyme in the X. maculatus, the F I hybrid, and half of the backcross segregants, which include fish with benign melanoma and others without melanoma (see arrows pointing to the gels). Data from Ahuja et al. (1980).c-src (cellular counterpart of the viral src oncogene), pp60C-srckinase activity expressed as counts per minute/milligram protein. Note basic and excessive activity and correlation between c-src expression and Tu expression. Data from Barnekow et al. (1982); M. Schartl et al. (1982); see Section IV,A,5. Diff, activity of Dqfexpressed as the replacement of guanine in position 34 (first position of the anticodon) of the tRNAs for Asp, Asn, His, and Tyr by labeled guanine, pinol Gua incorporated/A260 tRNA. Note that high values indicate high contents of guanosine (G),whereas low values indicate high values of queuosine (Q) in position 34. Data from Dess (1983); Kersten et ul. (1983); see Section IV,D,4. Arrows between the backcross segregants, design of transplantation experiments described in Section IV,D,3.

scheme). Further backcrosses of the benign melanoma bearing quasi-F, hybrids with the platyfish (D) result in spotted and nonspotted fish that are similar to the purebred platyfish (E, F, G , H). Genetically they segregate into those that inherit the capability to develop melanoma after crossings with swordtails and those that do not inherit the capability to develop melanoma (not shown in the scheme). These two genotypes are hard to distinguish phenotypically. The results obtained in these two series of crossing experiments revealed several genetic components that are involved in melanoma formation, or

CONTROL OF ONCOGENES

225

protection from melanoma, respectively. They shall be out1:ned in the following, and discussed in more detail in Section IV. The platyfish genome (see Fig. 18) contains the genetic information for the development of spots which apparently is lacking in the swordtail genome. Since the spots consist of transformed pigment cells (Section II,C), this information is considered to be encoded by an oncogene, Tu. From about 70 structural changes involving crossovers, deletions, duplications, and translocations, we know that Tu is normally located at the end of the X chromosome of X . maculatzcs (A. Anders et al., 1973b; Forster and F. Anders, 1977; Ahuja et al., 1979; Chatterjee et al., 1981), and is apparently also responsible for the large variety of the carcinogen-triggered neurogenic, epithelial, and mesenchyinal neoplasms (A. Anders and F. Anders, 1978; Schwab, 1980; Schwab et al., 1978a,b, 1979) listed in Table IV (in other crossings the susceptibility to neoplasms of mesenchymal origin could be assigned to a separate chromosome; see Sections II1,C and E). The protection of the platyfish from the activity of its own Tu oncogene is apparently exerted by regulatory genes. Following hybridization with the swordtail, Tu becomes deregulated, indicating that the swordtail lacks not only the oncogene Tu but also the regulatory genes (F. Anders, 1967). Several types of regulatory genes can formally be deduced from the outcome of the crossing experiments. (1)The restriction of crossing-conditioned neoplasia to melanoma indicates the presence of a Tu-linked pigment cell-specific regulatory gene which, as known from mutagenesis studies (A. Anders et al., 1973a; A. Anders and F. Anders, 1978; Ahuja, 1979), is impaired by mutation. Out of the tissue-specific regulatory genes only the impaired pigment cell-specific regulatory gene for Tu, i.e., R f M e l is , shown in the scheme (Fig. 18). (2) The restriction of the crossing-conditioned melanomas to the posterior part of the body and to the dorsal fin indicates the presence of Tulinked regulatory genes that are specific to certain compartments of the body. In additional experiments corresponding to those shown in Figs. 1618, depending on the genotype (mutant) of the platyfish used for the initial crosses, melanoma develops, for instance, in the anterior part of the body, the anal fin, tail fin, the mouth, the eye, the peritoneum, the meninx etc. Thirteen compartments have been identified that correspond to different genes which in turn correspond to sites of the body where the melanomas occur (A. Anders and F. Anders, 1978). Of these compartment-specific regulatory genes which have been designated in total as R,, (Fig. 8) only the impaired dorsal fin-specific regulatory gene, i.e., R’Dp and the impaired posterior part-specific regulatory gene, i . e . , RIP,,, are shown in the scheme. (3)The clearcut 1:l segregation between the BC hybrids bearing malignant melanomas and those bearing benign melanomas indicates the existance of a prominent regulatory gene derived from X . mculatus that is nonlinked to the oncogene Tu. Since the benign melanomas consist, in contrast to the

226

FRITZ ANDERS ET AL.

malignant melanoma, predominantly of terminally differentiated transformed pigment cells (U. Vielkind, 1972, 1976; Diehl, 1982; F. Anders et al., 1980) this regulatory gene is considered to be involved in differentiation of these cells, and was, therefore, designated as D i f j The Dif-containing chromosome can easily be detected by an esterase marker, Est-1 which is closely linked to D$f (Siciliano and Wright, 1976; E. Scholl, 1977; Ahuja et al., 1980; Morizot and Siciliano, 1982). Additional genes of minor importance involved in melanoma formation have been identified but are not taken into consideration in this study. Following crossings and backcrossings according to Fig. 18, the chromosomes of the platyfish (continuous lines) are replaced by the homologous chromosomes of the swordtail (broken lines), resulting in the gradual disintegration of the regulatory gene system for Tu. Thus the Tu hybrids develop spontaneously benign melanoma if some regulatory genes such as Diff are still present in the system, and malignant melanoma if the regulatory genes are lacking. If Tu is lacking, no melanomas occur. In contrast, after backcrossings of the melanoma-bearing hybrids with the platyfish as the recurrent parent (see Fig. 17) the chromosomes carrying regulatory genes for Tu are reintroduced into the descendants. This results in a reconstruction of the original regulatory gene system that suppresses the activity of Tu.

2. Oncogenes and Regulatory Genes in Animals Requiring Carcinogenic Triggers for the Development of Melanoma To relate the genes that are responsible for the development of spontaneous melanoma to the genes that are involved in the development of the carcinogen-dependent neoplasms, we replaced the RtDfR f P pRtMel Tu chromosome by another one the crucial difference of which is that R,,, is nonmutated and active (RDfR f g sRMelTU) (Fig. 19). Because RMelis inherited along with Tu, melanoma does not develop spontaneously in the hybrids. Following treatment with carcinogens, those hybrids carrying the RDf R f g s RMelTu chromosome but lacking the nonlinked regulatory genes including Diff are highly sensitive to the carcinogens because development of melanoma requires only impairment or deletion of the crucial pigment cell-specific RMelgene in a melanophore precursor at the side of the body. Crosses between two BC hybrids of this genotype were the basis for the establishment of a strain homozygous for the RDf R f B sRMelTu chromosome (Fig. 20). Because of the fact that each of the two copies of the Tu oncogene is repressed by its own linked RMel which acts in cis position only (see Section IV, B) the incidence of animals developing melanoma following treatment with carcinogens potentially doubles. These animals are highly suitable as test animals for mutagenic carcinogens in the water (F. Anders et al., 1981d, 1983; A. Anders et aZ., 1983; Schmidt, 1983).

227

CONTROL OF ONCOGENES

Est-1

C-SK

c-SfC

Esf-l-

FIG. 19. Crossing scheme displaying the genetic conditions of susceptibility to carcinogendependent neoplasia. The highly susceptible genotype is extremely sensitive to the carcinogenic (mutagenic) inducer. See legend to Fig. 18 for explanation of abbreviations. R'B.r, impaired regulatory gene controlling Tu in the compartment of the entire side of the body (Bs).

H I G H L Y SENSITIVE TESTER S T R A I N

FOR M U T A G E N I C CARCINOGENS

FIG. 20. Breeding procedure of an established tester strain which is very highly sensitive to mutagenic carcinogens. Both X and Y chromosome contain Tu and RM,+ This strain was derived from the highly susceptible segregants shown in Fig. 19. See legends to Figs. 18 and 19 for explanation of abbreviations.

228

FRITZ A N D E R S ET A L .

MNU or X rays

4

MNU or X rays

4

F. hvbrid-t

X . hdleri - 8

x rays-( MNU or

Backcross generation Frc. 21. Schematic presentation of crossings and hxkcrossings between X. oariatus (Fig. 413) and X. helleri (Fig. 3B) with X. helleri as the recurrent parent that have shown spontaneous development of benign melanoma in 50% of the BC generation, and a high rate of carcinogentriggered malignant melanoma, iieuroblastoina and different kinds of benign and malignant epithelial tumors in the same segregants. Carcinogen-triggered neoplasms of inesenchyinal origin occur in both the melanomatous and the nonmelanomatons segregants indicating that they are genetically independent from neurogenic and epithelial neoplasms.

E. TISSUESPECIFICITYOR TISSUENONSPECIFICITY OF ONCOGENES A N D REGULATORY GENES The hybrid segregants that were highly susceptible for carcinogen-triggered melanoma (see Fig. 19) were also susceptible to many kinds of carcinogen-triggered neurogenic, epithelial, and mesenthymal neoplasm (see Fig. 15; Sections II1,A and D). Furthermore, many hybrid individuals treated with carcinogens, in addition to the melanoma, developed multiple neoplasms such as neuroblastoma, retinoblastoma, carcinoma, and sarcoma (Abdo, 1979; Kollinger, 1980; Schmid, 1983). Therefore, the development of the different carcinogen-triggered neoplasms apparently depends on the same Tu (Schwab et al., 1978a,b). Additional information about the assignment of the different kinds of neoplasms to the oncogene Tu came from crosses and backcrosses between X . variatus (Fig. 4D) and X . helleri (Fig. 3B) (Fig. 21) that correspond to those between X . maculatus and X . helleri (see Fig. 19). In this experiment 50% of

CONTROL OF ONCOGENES

229

FIG.22. Backcross hybrids hetween X . ~ a r i u t ~ and r s X . laelkri according to Fig. 21. (A) Fish exhibiting crossing-conditioned (germ line hereditary) superficial spread benign melanoma, and MNU-triggered invasive inalignant melanoma. (B) Fish exhibiting the crossing-conditioned benign melanoma only.

the animals of the BC generation develop superficial spreading benign melanoma but maintain their normal vitality and fertility; the melanoma is not iiivasive in these animals (Altmaier and F. Anders, 1968). The remaining 50% of the animals of the BC generation are melanoma free. After treatment with X rays and MNU the melanomatous segregants may develop foci of malignant melanoma on the skin and even on the superficial spreading benign melanoma (Fig. 22). Furthermore, these animals may develop different kinds of neurogenic (mostly neuroblastoma) and epithelial neoplasms. The crossing-conditioned benign melanoma as well as the carcinogen-triggered malignant melanoma, neurogenic, and epithelial neoplasms can be assigned to the same chromosome. Neoplasms of meseiichymal origin, however, develop in both the melanomatous and the nonmelanomatous segregants. Further crossing analysis of the potential of the fish to develop neoplasia following the carcinogenic trigger has shown that the mesenchymal tumors are mediated by oncogenes other than those that mediate the neurogenic (including melanoma) and epithelial neoplasms (Schwab et al., 1979; Schwab and A. Anders, 1981). Additional studies are required before we can decide whether (1) Tu is

230

FRITZ ANDERS ET AL.

tissue-nonspecific whereas tissue specificity of the tumor comes from Tulinked tissue-specific regulatory genes, or (2) whether there are different onc genes that are under the control of linked regulatory genes specific to the tissue-specific onc genes. It would be extremely important to know whether the 10 different onc genes (see Fig. 9) that have been identified in the genome of Xiphophorus by molecular hybridization are involved in neoplasia, especially in the problem of tissue-specificity or tissue nonspecificity of the genetically determined oncogene Tu (see Section IV,A,5). IV. Oncogenes and Regulatory Genes

Gordon (1947a, 1958) and Kosswig (1937) have shown that the gene system involved in melanoma formation in Xiphophorus is highly polygenic. Xray mutagenesis studies have confirmed these results, and revealed that the constituents of the polygenic system are distributed throughout all chromosomes (Pursglove et al., 1971; Pursglove, 1972; A. Anders et al., 1971; F. Anders et al., 1981a). However, among these genes there are only few prominent genes that predominatly determine whether melanoma develops or not (Section 111,D). These genes are Tu, the oncogene; RMel,the pigment cell-specific regulatory gene; R,, (e.g., RDp R,,), the series of compartment-specific regulatory genes (Df = dorsal fin; Pp = posterior part of the body); and Dijjj the differentiation gene. Studies on 48 structural changes of sex chromosomes of X . mculatus, X . uariatus, and X . xiphidium (19 deletions, 14 duplications, 4 translocations, and 11 X-Y crossovers) indicate that Tu, at least as far as the sex chromosomes are concerned, is located terminally, and that RMel and the R,, series are closely linked to Tu (Fig. 23) (A. Anders et al., 1973a,b; and unpublished data). Some of the major chromosome aberrations involving loss and translocation of both Tu and its linked regulatory genes controlling Tu in the melanophore system (the region including R,,, RMMel, and Tu), were, in addition to their genetic identification, also observed cytologically (Ahuja, 1979; Ahuja et al., 1979). Dijjf is not linked to Tu and is located together with the loci for esterases-1 and -4, and malatedehydrogenase-2 on an autosome corresponding to linkage group V according to Morizot and Siciliano (1983).

A. THE ONCOGENE

1. General Features of Tu

All deletions of Tu are nonlethal in both the heterozygous and the homozygous state, indicating that this sex chromosome-linked oncogene is not

23 1

CONTROL OF ONCOGENES

___-

RMes

sex

//

RNerv

Pter

RCo

Tu

RMe 2

I

I

1

+ V

10

FIG.23. Preliminary map of the sex chromosonies (X and Y) of the platyfish (X. mcuZatu.s, X. onriatus, and X . xiphidiuni; Section II,A) based on 48 structural changes (11 X-Y crossovers, 19 deletions, 14 duplications, 4 translocations) and Gienisa banding studies, in which oncogene Tu and its regulatory genes are involved. The brackets indicate the regions within which the structural changes occurred (7, 10, 48); one translocation separated Tu from all linked regulatory genes (Section IV,A,2, Fig. 24). sex, sex determining region; Rnf,,, REpi,RNErc,sets of regulatory genes controlling Tu in mesenchymal, epithelial tissues, and in the nervous cell system. Pter, pterinophore locus; RC:~,, region containing at least 13 compartment-specific regulatory genes such as R D f (dorsal fin-specific), R p p (posterior part-specific) etc. (Section IV,C); RMt,l. melanophore-specific regulatory gene. The unit including Rc;<,,R M e / ,and Tu corresponds to the specific “color genes” known as Sd (spotted dorsal), S p (spotted), etc. in the literature. The unit including Pter corresponds to the so-called “erythrophore genes” known as Dr (dorsal red), Ar (anal red), Or (orange), etc. in the literature (see Kalhnan, 1975).Additional regulatory genes for Tu are distributed throughout other chromosomes, e.g., Diff(nonlinked to Tu). From A. Anders and F. Anders (1978), modified.

essential for the fish. One could, for instance, assume that additional copies of Tu present in the autosomes may compensate for the loss of the sex chromosome-linked Tu locus according to a gene dosage compensation mechanism, which permits normal cellular functions to continue. Such a function might possibly be the normal outburst and control of cell reproduction in different tissues at different periods during different stages of embryonic and postembryonic development as well as in regeneration processes. Impairment or loss of control of this function might give rise to the initial phase of tumor formation. Nevertheless, the only function of Tu known so far is that it mediates neoplastic transformation. Up to the present we have not found any mutation of Tu itself although one may expect this to be possible based on the findings of the laboratories of Weinberg (Tabin et al., 1982) and Barbacid (Reddy et al., 1982) that mutation converts a silent onc gene to the transforming state. Tu of Xiphophorus displays its transforming activity as a normal nonmutated gene which, however, is deregulated following elimination, deletion, or impairment of regulatory genes (see Section IV,B,3). Tu might be related to virus particles found in the crossing-conditioned and in the MNU-triggered neuroblastomas and melanomas (Kollinger et al., 1979). Normally these particles are not present in the neoplasms but occur

232

FRITZ ANDEHS ET AL.

following the treatment of BUdR. Up to the present there is no indication that these particles may trigger neoplastic transformation or may stimulate tumor growth. In most neoplasms the virus particles resemble small virions of the DNA tumor virus SV40 or polyoma virus. In addition to these particles, B-type- and C-type-like particles were also found. One could speculate that Tu normally governs normal functions in the cell as mentioned above, but may also generate virus particles, if the cell is treated with an inducing chemical. According to Gillespie and Gallo (1975), who refer to the B-type and C-type particles, such a gene would be a “class I gene” that can create a “class I virus.” Normal tissues of nontumorous and tumorous fish as well as the tumors showed RNA-dependent DNA polymerase activity. This activity, however, could not be assigned to the supposed endogenous virus reverse transcriptase activity (Lueke, 1984; Lueke and F. Anders, 1983). Regardless of any future findings, that might relate Tu to an endogenous oncogenic virus, Tu acts, is regulated, and is inherited as a chromosomal gene of the natural gene pool of the Xiphophorine fish (Section 111,D). Tu was transferred by purified DNA (Haas-Andela, 1978; J. Vielkind et al., 1982): donor DNA of various fish strains carrying copies of deregulated Tu was injected into recipient embryos lacking these Tu copies. Up to 8% of the treated embryos and the resulting young fish developed colonies of neoplastically transformed cells. If embryos were treated with DNA from strains lacking these Tu copies, no such colonies were observed (see Section IV,B). In accordance with Breider (1939) we found that Tu may mediate melanoma formation in albinos (Fig. 14D). The developing amelanotic melanomas revealed defective premelanosomes and a lack of melanosomes and melanin (U. Vielkind, 1972, 1976; J. Vielkind et al., 1971). The occurrence of the albino melanomas indicates that Tu acts independently from melanin synthesis, and that the black pigmentation of the melanotic melanoma is an epiphenomenon of melanoma development. 2. The Completely Deregulated Oncogenic Effect of Tu in the Pigment Cell System

Information about the genuine oncogenic effect of Tu comes from a balanced laboratory stock carrying a lethal Tu translocation (Fig. 24). The Tu gene, minus all linked regulatory genes, became translocated from an X chromosome of X. muculatus to an autosome of X. helleri and, in its new position, is no longer controlled by its formerly linked regulatory genes that act in cis position only (A. Anders et al., 1973a,b). Nonlinked regulatory genes are not present in the system except for D@ The Tu-carrying progeny of the stocks segregate into 50% animals carrying Dqf which survive, whereas the corresponding 50% lacking Dqf are lethal. As a consequence of the

CONTROL OF ONCOGENES

A

_ _ _ _ - ------- - Tu

x

4,P t e r ,R C o , RMe I

233

FIG.24. Translocation of Tu from the X chromosome (X) o f X . inuculutus to an autosome (A) of X . helleri. Note separation of Tu from its linked regulatory genes (see Fig. 23). F, female determining region of the X chromosome. See legend to Fig. 23 for further abbreviations.

transmission of the unrestrained Tu through the germ line, the pigment cell precursors become transformed in the embryo as soon as they become competent for neoplastic transformation by cell differentiation (F. Anders et al., 1979a). During the first days of embryogenesis, differentiation of pigment cells still undergoes the normal course. After the embryo is 5 days old, some single cells become transformed, and at a later time about 10-20 dividing transformed melanoblasts appear in the peduncle of the tail fin. These differentiate within about 15 hr to transformed melanocytes (Fig. 25A), which represent the predominant cells of the growing melanoma. During the further development of the embryo, neoplastic transformation contineus in all areas where pigment cell precursors become competent (Fig. 25B) and the melanoma grows by both transformation and proliferation, thus developing into a “whole body melanoma” (Fig. 25C), which will kill the fish before or shortly after birth. The development of melanoma in the early embryo reflects the genuine oncogenic effect of the completely derepressed Tu on the pigment cell system. These observations suggest to us that Tu exerts important normal functions in cytodifferentiation and proliferation in the early embryo that are related to the neural crest where the pigment cell precursors originate. Moreover, we assume that in normal embryogenesis these functions become switched off or choked by the regulatory genes before the fifth day of embryonic life. If, however, the regulatory genes (i.e., the entire switch in the lethal Tu translocation) are lacking, Tu continues to exert its early embryo-specific functions which, as an extension of the cellular development in the early embryo, appears as transformation of the competent cells to the neoplastic state. The assumption of normal nononcogenic functions of Tu in early embryogenesis raises the question regarding the genes that might exert these functions in the animals lacking Tu, such as swordtails used in the crossing experiments (Section 111,D) or the deletion animals mentioned above. This

234

FRITZ ANDERS ET AL.

FIG.25. Completely deregulated oncogenic effect of the oncogene Tu in the melanophore system following Tu translocation shown in Fig. 24. (A) Ten-day-old embryo (3 mm long) exhibiting some Tr melanocytes at the peduncle of the tail fin. (B) Same fish, 5 days later (4 mm long). (C) Neonate of the same genotype (6 mm long).

question leads to the problem of indispensable and accessory copies of the oncogenes, and to the problem of oncogene dosage.

3. Indispensable and Accessory Copies of Tu In total, about 30 deletions of Tu copies have been observed in our Tucontaining stocks of maculatus, variatus, xiphidium, helleri, and

x.

x.

x.

x.

CONTROL OF ONCOGENES

235

X. montezume. All deletions, even the loss of a Tu-containing Giemsa band in the homozygous condition in the female, or in the heterozygous state in the male (observed in the X chromosome of X. maculutus) have no detectable effect on the viability of the fish (Ahuja, 1979; Ahuja et ul., 1979). This observation and the fact that wild populations may lack Tu in total led us to the conclusion that the Tu oncogene considered so far is accessory for the fish. This does not imply that the accessory Tu is lacking normal functioning. One could, for instance, assume that certain copies of Tu are present in the genome which are indispensable and may compensate the loss of the accessory Tu loci. On the other hand, one could also expect that indispensable copies of Tu might mediate neoplastic transformation after disturbance of their regulation. Support for the assumption of the existence of indispensable Tu copies comes from the following experiment. Platyfish carrying the X chromosomal deletion of the Giemsa band that includes the accessory Tu (Ahuja, 1979) were crossed with the swordtail according to the procedure outlined in Figs. 16 and 18. As was expected no X chromosomal inherited tumors developed spontaneously in the hybrid offspring, but after treatment with MNU in rare cases, the backcross hybrids developed different kinds of neoplasms (including melanoma) that could be assigned partly to the deleted X chromosome of X . maculatus (42/408) and partly to autosomes of unknown origin (20/470). Thus it appears that the platyfish, besides the easily detectable accessory Tu copies, also contains indispensable ones that require more intricate experiments for their detection (Schmidt, 1983). Accessory copies of the oncogenes may be present but are not essential. If they are present, special regulatory gene systems are required for their control. We introduced up to 10 known potentially tumorigenic copies of Tu (four copies in homozygous duplications in both X chromosomes of the female, six copies in three nonhomologous pairs of autosomes) together with their linked regulatory genes into a genome containing nonlinked regulatory genes, but no tumors developed spontaneously nor could any effect on viability be observed. We assume that the Tu copies present in a certain genome are not strongly limited in number if their control is maintained by regulatory genes.

4 . Oncogene Dosuge More information about control of the Tu oncogene comes from studies on oncogene dosage compensation and oncogene dosage effect. We shall refer first to dosage compensation: An X chromosome showing the Tu deletion according to Fig. 24 (XDe'; the chromosome originates from X. maculutus but occurred in an X. maculutus-X. helleri hybrid) was introduced by introgressive breeding into a stock of X. muculutus, the X chromosome of which contains the accessory Tu (XTu) which, because of the impairment of its

236

FRITZ ANDERS ET AL.

linked regulatory genes RMrl and R,,-, exhibits the "spotted dorsal" phenotype (see Fig. 18). Both X chromosomes are identical except for the Tu deletion. The autosomal regulatory system for Tu is complete. The resulting animals contain none, one, or two copies of the accessory Tu (Fig. 26A-C). The animals lacking the Tu oncogene (A) do not show any neoplastically transformed cell, whereas the littermates containing one (B) or two (C) Tu copies, due to the impairment of RMF,and ED,-, exhibit spats in the dorsal fin that are phenotypically identical. If the experiment was modified by using autosomes of X . helleri that lack the nonlinked regulatory genes for Tu as the genetic background (Fig. 27A-C), the animals lacking the Tu oncogene, as expected, do also not show any neoplastically transformed cell (A), whereas the littermates containing one Tu copy exhibit malignant melanoma (B) while those having inherited the double dosage of Tu develop extreme malignant melanoma (C). Oncogene dosage compensation (Fig. 26B and C ) and oncogene dosage efTect (Fig. 27B and C) have been observed in many experiments of this kind (F. Anders and Klinke, 1966). 5. Tu and c-src

After the cellular homologs of retroviral oncogenes, particularly c-src (Section II,E), and the Mendelian oncogene Tu (Section II1,D) were identified in all individuals of all groups of Xiphophorus tested, we combined the molecular studies on c-src with the genetic studies on Tu. Based on the assumption that the activity of the pp60r-"'" associated phosphokinase monitors the activity of the c-src oncogene, we carried out comparisons between kinase activity (Section II,E), i.e., c-src expression, and tumor development, i.e., Tu expression (Section 111). In somatic mutationconditioned neoplasms (melanomas and others) triggered by M N U and X rays (Section II1,A) we found elevated levels of kinase activity as compared to the corresponding nontumorous tissues (M. Schartl et al., 1983). The results are compatible with those obtained by other authors in other systems (see Jacobs and Kubsamen, 1983), but still left it unclear to determine whether the kinase activity in tumors is causally related to neoplasia or is only one of the numerous epiphenomena associated with neoplasia (see Section IV, D). More informative data come from studies on hybrids developing melanoma spontaneously according to Mendelian laws (Section IIl,D,1; Figs. 16 and 18). These hybrids show tissue-specificity in kinase activity like the nontuinorous fish (see Fig. 12). This activity is high in melanoma and in brain, and varies genotype specifically in both melanoma and brain in the same direction (Fig. 28) (Barnekow et al., 1982; M . Schartl et al., 1982). The genotype-specific similarity of kinase activity in brain and melanoma was not found in the animals developing the somatic mutation-conditioned neo-

CONTROL OF ONCOGENES

237

FIG.26. Gene dosage compensation in the “spotted dorsal” phenotype (Tu expression in the dorsal fin) in littermates that are genetically identical except for the dosage of the accessory Tu. Tu is controlled hy the complete notilinked regulatory gene system. (A) No T u , 110 spots. (B) One copy of T i t , spots in the dorsal fin. ( C )Two copies of Tu, spots do trot differ from those of B.

238

FRITZ ANUERS ET AL.

Eic. 27. Gene dosage effect of the same Tu as used in the experiments shows in Fig. 26 except that the nonlinked regulatory gene system is lacking. (A) No Tu, no neoplastically transformed cell. (B) One copy of Tu, malignant melanoma originates at the dorsal fin. (C) Two copies of Tu, extreme malignant melanoma develops.

CONTROL O F ONCOGENES

239

FIG. 28. Demonstration of pp60C-"'"-associated protein kinase activity in extracts of brain and melanoma. Track 1: brain from henign melanoma bearing BC hybrids. Track 2: brain from malignant melanoma bearing BC hybrids. Track 3: benign melanoma. Track 4: malignant melanoma. 53K indicates the heavy chain of immunoglobulin G of the pp60c-srr inimunocomplex. Note that kinase activity in both brain and melanoma changes in the same direction. BC hybrids according to Figs. 16, 18, and 36. See legends to Figs. 11 and 12 for kinase assay.

plasms indicating that there is no secondary interdependence between kinase activity in brain and melanoma ( M . Schartl et al., 1983). Hence, we could determine pp60c-srcassociated protein kinase activity mainly in brain extracts and relate the activity observed to the expression of Tu ascertained by the development of melanoma (M. Schartl et al., 1982). The possibility that the differences in kinase activity measured in the fish of different Tu genotypes are due to epiphenomena of the melanoma appears unlikely. Therefore, the results reflect the actual genetic activity of the c-src oncogene in the nontumorous brain tissue of the tumorous and nontumorous fish. To study possible relations between Tu-conditioned neoplasia and c-src

240

FRITZ ANDERS ET AL.

FIG.29. Correlation between gene dosage effect of Tu (specified as the phenotype of the tumor) and gene dosage effect of c-src (pp60C-srckinase activity expressed as counts per niinute/milligram protein) in littermates containing (A) no accessory Tu, (B) single dose of Tu,(C) double dose of Tu. The genetic background of the fish is identical; the pigment cell-specific R,,, which is linked to Tu is impaired by germ line mutation. From F. Anders et al. (1984). modified.

CONTROL OF ONCOGENES

Tn gene complex corre\pnnding to phentype"

striped" dab bed(, dabbed

24 1

Dosage of the accessory Tuc No Tu (cpm/mg)

One dosage (cdmd

1)oul)le dosage (cpmling)

90 (18p 170 (11) 200 (5)

2oo (4) 190 (10) 260 (3)

390 (8) 390 (51) 1240 (4) ~

Three to eight I)rains per nieasurement were used. Different gels each were measured. c One gel each was measured. The Tu copy of striped originates froin the X and Y chromosomes of X . maculatus from Rio Janiapa, Mexico; the linked Rnlp,and R B are ~ impaired resulting in the phenotype shown in Fig. 29B and C (this is a mutant of X . nmculatus shown in Fig. 19). The Tu copy of dabbed originates from an autosome of X. helleri froin Belize River, British Honduras; Tu control is partly impaired in the compartment of the side of the body. f Phenotype and genotype are similar to those of e , but X . helleri originates from Rio Lancetilla, Mexico. g Total number of hains is indicated in parentheses. a

expression we took advantage of the three genetic experiments outlined in Figs. 18, 19, and 29. In the experiment outlined in Fig. 18, the pure-bred X . maculatus carrying two repressed copies of the accessory Tu, as well as the pure-bred X . helleri and the BC hybrids lacking the accessory Tu, display the same activity of c-src kinase (about 300 cpm/mg protein). This activity we interpret to be the basic expression of c-src. In contrast, the melanoma-bearing hybrids which contain the derepressed T u show an increase of c-src activity, with the malignant melanoma-bearing BC hybrids displaying the highest activities (about 600 cpm/mg protein). In the experiment outlined in Fig. 19, all pure-bred and hybrid animals, irrespective of the dosage of the accessory Tu, but dependent on their nontumorous state maintained by several nonlinked regulatory genes or by a single linked regulatory gene (see the highly susceptible genotype), display a similar uniform base-level c-src activity. In littermates (Fig. 29 A-C), however, genetically identical except for the absence (A) or presence of one (B) or two (C) partially derepressed accessory Tu copies, p ~ 6 0 ' - displays "~~ a kinase activity that increases stepwise and in parallel with the dosage of Tu. This Tu gene dosage, in turn, determines whether the animals will develop no tumors or slow or rapidly growing

242

FRITZ ANDE HS ET A L .

tumors. Table V shows additional experiments of the same kind that yielded similar results. The main results of these experiments are that the nontumorous fish display a basic expression of c-src, which, in the tumorous fish, may increase stepwise under two conditions, i.e., one, the stepwise derepression ofan accessory Tu, and two, the stepwise introduction of additional copies of a derepressed accessory Tu. These findings suggest several possible interpretations of the relationship between Tu and c-src: (1) Tu might be independent from c-src, and the correspondence between both Tu and c-src is due to linkage relationships. (2) the c-src might represent a regulatory gene for Tu or vice versa. (3) Tu might be identical to c-src, and this oncogene can code for a large variety of neoplasms. (4) Tu might consist of different oncogenes responsible for different kinds of neoplasia, and c-src is one of these genes. At present, we cannot make a firm interpretation; additional data are required. Two points favor the idea that the accessory Tu oncogene is composed of several c-onc genes homologous to the retroviral wont genes. These are (1)the fact that all types of neoplasia including epithelial, neurogenic, and mesenchymal neoplasms could be assigned, for instance, to an accessory Tu located on a particular Giemsa band of an X chromosome, and (2)the assumption that each of the different c-onc genes identified by molecular hybridization codes for a different type of neoplasia. Those crossing experiments, in which the backcross hybrids segregated into one group of animals susceptible to epithelial and neurogenic neoplasms, and into the other group susceptible to mesenchymal tumors (see Section III,E) could be explained by the assumption that part of the tissue-specific onc genes that normally might compose Tu were translocated to a nonhomologues chromosome.

B. THE PIGMENT CELL-SPECIFIC REGULATORY GENE 1. Significance of Regulatory Genes in Neoplasia

Emphasis is being placed at present in cancer research on the molecular characterization, amplification, rearrangement, mutation, and overexpression of oncogenes, and on their normal and abnormal functions. We know from the Xiphophorus model, on the other hand, that the most important process involved in neoplasia in these animals is probably loss, deletion, impairment, or any other dysfunction of the regulatory gene system permitting the abnormal expression of the ubiquitously present oncogenes. Deregulation of an oncogene caused by spontaneous or environmentally induced molecular or structural changes in the regulatory gene system is the biological basis for the development of neoplasia in Xiphophorus. It is these regulatory genes and not the Tu oncogene itself that the carcinogens act on

CONTROL OF ONCOGENES

243

when they trigger neoplasia in Xiphophorus (A Anders et al., 1973a,b; A. Anders and F. Anders, 1978). Based on this insight our group has considered the possibility (Ahuja and F. Anders, 1976, 1977) that the specific chromosome abnormalities which have been found to be associated with chemically (Mitelman and Levan, 1972; Levan et al., 1974) and virally (Mitelman et al., 1972) induced neoplasms of mammals, and with certain neoplasms of humans (Rowley, 1973, 1974, 1977; Zech et al., 1976; Mark, 1977; Sandberg, 1980), might arise as a result of molecular change, chromosome translocation, or deletion of tissuespecific regulatory genes which control oncogene expression in those tissues. Thus, impairment or deletion of regulatory genes, on the one hand, presumably leads to neoplastic development, whereas loss or impairment of oncogenes, on the other hand, would lead to a decrease of the tumor-mediating potential in a specific tissue. Recent results on chromosomal location of oncogenes and on specific chromosome aberrations associated with certain tumors (Dalla-Favera et al., 1982a,b; de Klein et al., 1982; Nee1 et al., 1982; Rowley, 1982, 1983; Taub et al., 1982) may also be interpreted by the assumption that the development of many human neoplasms results from the release of oncogenes from their regulatory genes caused by chromosome aberration and impairments.

2 . General Features of Rndel The pigment cell-specific regulatory gene RMelwas first detected by Kallman (1970) when he observed an unequal crossover that separated a T u oncogene-containing chromosome segment from its linked RMe,. Mutagenesis studies (A. Anders et al., 1973a,b) confirmed that RMel is closely linked to Tu (Fig. 23). R M , [controls Tu in the cis position only, i.e., the R M p l gene controlling the closely linked Tu of a specific chromosome does not influence the Tu located on the homologous chromosome or any other chromosome. Based on these findings we developed an extensive breeding program to produce donor and recipient fish for gene transfer (transfection) experiments. These were expected to furnish new data on the functional interrelationship between RMel and Tu.

3. Cotran.$er of R&f,lel and Tu by DNA Injection Total genomic DNA extracted from nontumorous male gonads of laboratory stocks of X . iiwculatus carrying accessory copies of Tu and normal, impaired, or deleted RMel genes was injected into the neural crest region of early embryos (0.05 to 0.9 pg DNA per embryo) of X . helleri which apparently lack both the accessory Tu copies and the RM,,/genes (Fig. 30) (HaasAndela, 1978; J. Vielkind, 1979; J. Vielkind et a l . , 1982). In some of the experiments DNA from animals that contained six additional but rigidly

244

I;nrrz

ANDERS ET AL.

FIG. 30. Outline of the transfectioii experiments. See text. From Haas-Andela (1978) arid J. Vielkind et ( I / . (1982). modified.

repressed accessory Tzi copies was also used. The injected DNA maintains its high molecular weight for about 2 hr (Schwab, 1974; Schwab et a l . , 197611) and thereafter becomes degraded to pieces which are too sinall to contain genetic information (J. Vielkind, 1971; J. Vielkind et d., 1973a; Schwab et d . , 197611). Therefore, the timing of DNA injection with the appropriate developmental state of the embryo, i.e., the stage just prior to the migration of neural crest cells and their descendants to their final location in the skin and the extracutaneous tissues (Sections II,D and IV,C) is very important for the outcome of the experiments. A neural crest cell or its early descendant still close to the neural crest region may take up the donor DNA, and may divide and differentiate to I melanoblasts, the only pigment cell precursors that are competent to the transforining TZLactivity (Section 11,D; Fig. 8). These cells eventually may become transformed to TrI melanoblasts. Additional proliferation of the Tr cells may amplify the transfected Tu in the pigment cell population, and the expected result should become visible after differentiation of the still nonpigmented TrI melanoblasts to the heavily

CONTROL OF 0NCXK;E N ES

245

pigmented and large Tr nielanocytes and Tr inelanophores that form black cell colonies on the greenish-gray normal skin. Similar colonies consisting of transformed pigment cells were also expected to occur in extracutaneous tissues that normally are also populated by regular inelanophores (see Section IV,C). Most of the transforrnants (embryos and young fish) developed single Tr cells or small colonies ofTr cells (Fig. 31A) that were indistinguishable from those of the first stages of inherited inelanoina in embryos and young fish (Fig. 31C) (Haas-Andela, 1978; J. Vielkind et al., 1982), and from those developing from transplanted stein-inelanoblasts containing an active Tu (Fig. 31B) (M. Schartl, 1979). It is important to note that most of the Tr cell colonies observed had about the same size indicating that they had undergone the same number of cell divisions. Several transformants exhibited large amounts of inelanoina cells that grew along the meninx priinitiva of the spinal cord (Fig. 32) and killed the embryos. The transforming activity of the DNA of different donor breeds (Fig. 33AE) lacking the accessory Tu and RAI1,/(A), or exhibiting phenotypically different degrees of control of Tu by RAIPl(B-D) was compared. If a donor was used which lacks RA,<,,Tu (A), no recipient exhibited transformed pigment cells. If a donor strain was used that carries a Tu slightly derepressed in the pigment cell system by a “weak” mutatioii of its linked R M F Ii.e., , R’,+,pl Tu (B), then 0.4% of the recipients developed colonies of neoplastically transformed cells. If the DNA originated from a strain that carries a Tu derepressed to a greater degree, due to a “stronger” mutation of RMMe,, i.e., R M I , (C), / the incidence of recipients exhibiting transformed pigment cells increased to 2.6%. If, finally, a donor was used in which the Tu lacks the RM,,/(due to a chroniosoinal translocation) (D) this incidence increased to 6.3%. DNase degradation prior to injection eliminates the transforming activity of the DNA (E). DNA froin animals carrying additional but rigidly repressed accessory T u copies (not shown i n the figure) did not influence the incidence of transformants. These results obtained by the gene transfer experiments in Xiphophorus may be helpful in our efforts to analyze the processes leading to neoplasia. Besides the fact that the information for neoplastic transformation, presumably Tu itself, was transfered via total genoinic DNA injection, it is important to note that the transforming donor DNA did not originate froin tumor cells but froin the nonneoplastic testes indicating that ongocenes must not necessarily be changed or amplified in order to acquire the transforming potential. The inany oncogene transfection experiments accoinplished during the last years by several authors with other systems (Der et ul., 1982; Goldfarb et ul., 1982; Krontiris et al., 1981; Perucho et al., 1981; Reddy et al., 1882; Tabin et al., 1982) in which DNA extracts from tumors were used

FIG. 31. Colonies of neoplastically transformed pigment cells in embryos and fiy normally not capable of developing transformed cells (A and B). (A) Tr melanocytes induced by injection of Tu DNA into an early embryo. From Haas-Andela (1978). (B) Control experiment by M. Schartl(l979):Tr melanocytes differentiating from stem cells of melanoma transplanted into an embryo. (C) Unmanipulated control. Tr melanocytes normally differentiating in Tu-containing genotypes capable of developing transformed cells. Note that the Tr cells are all alike. For details see J. Vielkind et al. (1982). See also Fig. 38.

FIG.32. Twelve-day-old embryo (3 mm long) raised in oitro with pigment cells of the embryos populating the dura inater of the spinal cord. (A) Transformant of a transfection experiment outlined in Fig. 30. Note the neoplastically transformed cells. (B) Control with regular pigment cells. From Haas-Andela (1978).

248

FHITZ ANDEHS ET AL.

D O N O R S

NO, OF R E C I P I E N T S

Fic. 33. Transfection activity of donor DNA extracted from male gonads of fish differing in gene dosage of Tu,and in Tu control by the pigment cell-specific regulatory gene Rnfrl. See text. Data from Haas-Andela (1978) and J. Vielkind et a l . , (1982).

under the expectation that tumor DNA differs from DNA from normal tissues should be, in our opinion, reconsidered under the viewpoint of repression and derepression of oncogenes exerted by intact and defective regulatory genes. The main factors responsible for neoplasia are, in view of our results on Xiphophorus, not the onc genes, but their regulatory genes. This view is supported by the fact that the incidence of transformants, i. e., the incidence of the transformation events mediated by Tu, was independent from the number of Tu copies in the donor DNA (we tested DNA

CONTROL OF CjNCOGENES

249

containing up to eight copies), but was exclusively dependent on the degree of impairment of RA,<,,(see Fig. 33). In this view it appears also reasonable to assume that RMel,if present, is so closely linked to Tu that both Tu and RMPl were apparently always cotransferred (Vielkind et al., 1982). Although the donor DNA originated from fish exhibiting different degrees of Tu expression (Fig. 33B, C , and D), the transformed cells of the recipients looked all alike, and the Tr cell colonies all were about the same size. This indicates that neoplastic transformation of a certain cell is more an all-ornothing process than a multistep process that releases this cell and its descendants from negative control by regulatory genes. On the other hand, growth of the tumor is neither influenced by Tu nor by the intact or impaired Rbfd

C. COMPARTMENT-SPECIFIC RECULATOHY GENES Most of the neoplasms are preferentially located in certain compartments of the body of the fish. For instance, the different kinds of carcinomas, as well as sarcomas, and melanoinas originate preferentially in the area around the peduncle of the dorsal fin and tail fin; epitheliomas develop frequently in the region around the posterior part of the operculum (see Figs. 15-18, 21, 26, and 27). The coinpartmentation has been studied in more detail mainly by means of X-ray-induced germ line mutations (a total of 28) that affect one or several sites of the crossing-conditioned inelanomas (Fig. 34A-H). These melanomas develop, for instance, in the tail fin (A), the dorsal fin (B), in both tail fin and dorsal fin ( C ) ,in the anal fin (D), in the tail fin, dorsal fin, anal fin, mouth tip, and the posterior part of the side of the body (E; mutations of five compartment-specific regulatory genes are involved), in the anterior and posterior parts of the side of the body (F), in all compartments except for the mouth, belly, eye, dorsal fin, and tail fin (G), or even in all compartments of the body (H). The phenotypes of additional combinations of impaired Rc,, genes were depicted earlier (A. Anders and F. Anders, 1978). The compartment-specific distribution of these melanomas is inherited according to the segregation of the parental Tu-carrying chromosome, indigenes are linked to Tu, and structural changes cating that the respective Rc;<> of the chromosome have verified that this linkage is very close (see Fig. 23). At least 14 genes corresponding to 14 different compartments have been identified (Fig. 35). They represent regulatory genes that were designated R,, in total and RAIl(anterior part), R,, (posterior part), RDf(dorsal fin), RTf (tail fin), etc. in reference to each specific body compartment. Intact R,, genes repress Tu, and impaired R,, genes permit Tu activity. They act,

FIG.34. Examples of compartmentation depending on compartment-specificregulatory genes (Rco genes) which, in case of impairment, permit compartment-specific melanoma formation. See Fig. 35 and text.

CONTROL OF ONCOGENES

25 1

FIG.35. Compartments for the development of melanomas in the fish (see Fig. 34) that are due to compartment-specific regulatory genes ( R c o genes). Mo, mouth; Ey, eye; Bm, brain membrane; Ap, anterior part; Df, dorsal fin; Pt, peritoneum; Af, anal fin; Sc, spinal cord; Pp, posterior part; Cr, crescent region; Pd, peduncle of the tail fin; Cf, caudal fin stripe; Tf, tail fin. From A. Anders and F. Anders (1978).

however, in the cis position only, indicating that the compartment-specific regulation of Tu exerted by the R,, genes acts at the DNA level. In the active state the R,, genes appear to delay the differentiation of pigment cells in the stem cell stage (S melanoblasts; see R,, in Fig. 8; Diehl, 1982). Additional mechanisms that are not understood provide the fish with differentiating pigment cells that mostly escape neoplastic transformation, but very exceptionally may be transformed. If, however, one or several R,, genes are impaired by mutation, compartment-specific melanomas develop in the hybrids and are inherited according to Mendelian prediction. There are many observations cited in the literature indicating that human melanomas are also preferentially located in certain areas of the body (Olsen, 1966). Xiphophorus, therefore, may provide a model to study this phenomenon. D. THE DIFFERENTIATION GENE

1 . General Features As shown in Section III,D and Figs. 16 and 18 benignancy and malignancy in the hybridization-conditioned melanomas depend upon the presence or absence, respectively, of the chromosome carrying DijJ Linkage relationship studies using biochemical markers for this chromosome, i.e., the esterase Est-1 and the isozyme A of the glyceraldehyde-3-phosphatedehydrogenase, have confirmed that the DijjJcarrying chromosome, like the accessory Tu, is derived from the platyfish. Animals from those wild popula-

252

FRITZ ANDEHS ET AL.

FIG.36. Melanoma bearing backcross segregants according to the schematic drawings of Fig. 18. (A) Benign melanoma bearing fish containing one dose of the differentiation gene Dq$ (B) Malignant melanoma bearing littermates that lack the O f , gene.

tions that apparently lack the accessory Tu oncogene also apparently lack the Dgfgene (E. Scholl, 1977; Ahuja et al., 1980). D$fmay be present in double dosage, single dosage, or be lacking, and the Tu-containing animals carrying two, one, or no D$f can easily be distinguished by gross examination of their phenotype. If D$f is present in double dosage (D$f/D$j) the oncogene Tu phenotypically may express at the most some occasional spots which are considered to be extreme benign melanomas (Section 11,C); if present in single dosage (D$f/-) the Tu oncogene may express benign melanoma. If, however, D$f is lacking (-/-) the oncogene Tu expresses almost always malignant melanoma.

2. DgJDependent Characters To study the basic difference between the benign and the malignant state of the melanoma, thousands of melanoma bearing BC segregants have been produced according to the crossing scheme shown in Fig. 18, and all showed a clearcut 1:1 segregation into animals developing spontaneously benign or malignant melanomas (Fig. 36; Table VI). BC hybrids that require the carcinogenic trigger for the development of carcinoma (see Fig. 19) show also a clearcut D$f effect: The D$f-carrying segregants (identified by esterase 1) that are moderately susceptible to neoplasia develop benign melanoma, whereas the D$f-lacking segregants that are extremely susceptible develop malignant melanoma (Fig. 37).

253

CONTROL OF ONCOGENES

TABLE VI SIXHEGATION OF FISIICARHYING BENIGN A N U MALIGNANT MELANOMA IN BC GENERATIONS Melanoma Stock from which Tu and D f f originated Spotted Spotted dorsal Spotted dorsal mutant Spotted dorsal mutant Stripe sided mutant Lineatus mutnnt

SP

Sd Sd' SdW Sr'

Li'

BC generation

Benign Off/ -

Malignant

1-4 1-12 1-4 1-5 1 1-2

173 1.358 1.270 1.831 263 232

207 1.344 1.235 1.806 25 1 228

Total

5.118

5.071

-I-

The cytological, fine structural, biochemical, and biological data obtained by the comparison of benign (with DiLfj and malignant (without DiLff) melanoma are listed in Table VII. The data show that the cells of the benign melanoma are well differentiated whereas those of the malignant melanoma are poorly differentiated, and that differentiation of the Tr cells is controlled by DiLff This, together with the findings on pigment cell differentiation (Section 11,D; Fig. 8) led us to the following conclusions: (1) ifDiLffis lacking, the majority of the melanoma cells persist in the stage of the poorly differentiated, continuously dividing TrA melanoblasts and Tr melanocytes, and only a few cells differentiate to Tr melanophores which are incapable of dividing and, at a certain age, are removed by macrophages; (2) if, however, the D$f is present, the majority of the melanoma cells become terminally differentiated to Tr inelanophores, whereas only a few cells remain in the stage of TrA melanoblasts and Tr melanocytes. Terminal differentiation and removal of the melanoma cells are antagonistic to the permanent supply of melanoma cells from S melanoblasts by transformation, and the melanoma renders benign. This observation is in accordance with the outcome of experiments in which newborn BC segregants were treated with methyltestosterone which strongly promotes pigment cell differentiation from the precompetent to the competent state for neoplastic transformation (A. Schartl, 1981): both, DiLff-carrying and DiLff-lacking newborn segregants, developed malignant melanomas that killed them. A benign period of the melanoma in the DiLff-carrying fish was not observed. Consequently, the level of D$ff-controlled benignancy is not at the premalignant stage but at the malignant stage of the tumor. On the other hand, if one considers the stepwise disintegration of the regulatory gene system for Tu, which is re-

254

FRITZ ANDERS ET AL.

FIG. 37. Backcross hybrids according to Fig. 19 after MNIJ treatment. (A) Diff-carrying segregant (identified by esterase-I) developing benign melanoma. (B) Dfl-lacking segregant (identified by lack of esterase-1) developing malignant melanoma.

quired for neoplasia, and considers elimination, deletion, or impairment, respectively, of the Dijjf as the last step of disintegration, benignancy might appear as the premalignant state of a malignant melanoma. The existence of animals carrying malignant melanoma on the surface of a benign melanoma like that shown in Fig. 22 might support this concept; actually, however, such a malignant melanoma is an additional, independently derived malignant melanoma that initiates following a mutation of Dijjfin an early pigment cell precursor. The concept of the change of a tumor from the benign to the malignant state which is generally accepted in cancer research could not be verified in Xiphophorus, because such a change has never been observed in Xiphophorus, and is even unlikely to exist. Malignant melanoma (and the malignant state of other neoplasms) in Xiphophorus is malignant from the very beginning, even in its initial stage consisting of one or only several cells (Section 11,D).

255

CONTROL OF ONCOGENES

TABLE VII THE GENEDVj I N Xiphophorus Diff/ -

-1-

Benign melanoma Well differentiated Slow growing Noninvasive Nonlethal Difficult to transplant No vascularization Difficult to trigger by promoters

Malignant melanoma Poorly differentiated Fast growing Invasive Lethal Easily transplantable Vascularization Sensitive to promoters (testosterone, CAMP, corticotropin, BUdR, TPA, cyclamate, saccharin, phenobarbital, etc.) No regression

Regression following testosterone treatment, etc. Weak effect of external factors on growth rate

No effect of nutrient factors Many macrophages Tr melanophores are prevailing Endopolyploid and multinucleated Mature melanosomes Lack of ER and Golgi complexes Low enzyme activities

Drastic effect of external factors on growth rate (temperature, salinity, UV, etc.) Drastic effect of nutrient factors (amino acids) Few macrophages TrA melanoblasts and Tr melanocytes are prevailing Diploid and uninucleated

Less complex glycosphingolipids Low rate of thymidine incorporation

Immature melanosomes Well-developed ER and Golgi complexes High enzyme activities (tyrosinase, LDH B4, MDH etc.) More complex glycosphingolipids High rate of thymidine incorporation

Tumorous and nontumorous tissues Low c-src activity

Tumorous and nontumorous tissues High c-src activity

Reference

U. Vielkind (1976) A. Anders et al. (1973a.b)

M. Schartl et al. (1981); C. R. Schmidt (1983); Herbert (1983)

A. Schartl et al. (1982)

F. Anders et al. (1962b); F. Sieger et al. (1969)

M. Sieger et al. (1968); F. Anders et al. (1969); F. Sieger et al. (1969) Diehl (1982) U. Vielkind (1976)

Schwab et al. (1976a); E. Scholl (1977); U. Vielkind et 01. (1977) Felding-Habermann et al. (1983) M. Sieger et al. (1968); J. Vielkind et al. (1973b)

M . Schartl et al. (1982)

256

FRITZ ANDEHS ET AL.

TABLE VII (Continued) DiEl Dispersed chromatin High rate of DNA nuclease digestion Low pteridiiie contents First position of the anticodon of tRNAAspP, tRNAA", tRNATyr, tRNA1iiScontains predominantly Queuosine DVj product is diffusible

-1Condensed chromatin Low rate of DNA nuclease digestion High pteridine contents First position of the anticodon of tRNAAsp, tRNAAS",tRNATyr, tRNAILiScontains predominantly guanosine No product

Reference F. Ariders el a!. (1981a); Heil (1983) Heil (1983) Henze et al. (1977) Dess (1982); Kersten et a1 (1983)

M . Schartl (1979)

3. A Dijferentiation-Promoting Dijf-Dependent Product The most convincing data supporting Dijjdependent control of pigment cell differentiation come from transplantation experiments which were designed according to the arrows drawn in Fig. 18 between the BC hybrids ( M . Schartl, 1979). The crucial transplantation experiments, including the composition of secondary chimeras by fusion of parts of early embryos, have shown that pigment cell precursors present in the transplant taken from fish carrying Tu but lacking Dijf (material of still tumor-free early embryos of the malignant melanoma developing genotype) become transformed and remain incompletely differentiated Tr cells if transplanted into embryos lacking Tu and Difj the resulting animals develop malignant melanoma. If, however, the pigment cell precursors of the same genotype were transplanted into Tulacking embryos that contain the Dijf, the cells of the developing melanoma become terminally differentiated and regain their distance regulation (Fig. 38): these resulting animals develop extreme benign melanomas which regress and eventually may become removed by macrophages. Thus the effect of Dqf on the differentiation of the neoplastically transformed pigment cells can be traced to a diffusible substance. The nature of this substance is unknown at present.

4 . Modijied tRNAs in Dijf-Dependent Differentiation There is considerable evidence for the involvement of modified nucleosides containing tRNAs in the process of cell differentiation in eubacteria, slime molds, and normal neoplastic tissues of vertebrates (Nishimura and Kuchino, 1979; Kersten, 1982a, 1983; Katze et al., 1983; see Nass,

CONTROL OF ONCOGENES

257

FIG.38. Secondary chimera composed by transplantation of tissues containing precursor cells of malignant melanoma that originated from a young backcross hybrid containing Tu hut lacking Dqf(malignant melanoma developing BC segregant according to Fig. 18), to an embryo lacking Tu but containing Dqf (nontumorous BC segregant in Fig. 18 at bottom left). Note terminal differentiation and distance (density) regulation of the transformed cells of the transplant according to the Dqfgenotype of the host. From M . Schartl (1979).

1983). The modified nucleosides occur at well-defined positions in specific tRNAs (Fig. 39A) (see Kersten, 1982b; Spritzel and Gauss, 1982). Many studies were focused on a family of tRNAs including tRNAAsn,tRNAAsp, tRNAHis, and tRNATyr which may contain queuosine (Q) instead of guanosine (G)in the first position of the anticodon (position 34; see arrow in Fig. 39A). The Q nucleoside (7-{[(4,5-cis-dihydroxy-2-cyclopenten-l-yl)-amino]methyl}-7-deazaguanosine) is unique in that its purine skeleton is modified to a 7-deaza structure (Fig. 39B). Eubacteria synthesize the base queuine de nouo (see Nishimura, 1983) whereas vertebrates are supplied with queuine by nutrition or the intestinal flora (Reyniers et al., 1981), and queuine itself is inserted into the nucleotide chain of tRNA by exchange with guanine by tRNA-guanine transglycosylases. The more the differentiation progresses, the more G is replaced by Q in position 34 (see Nass, 1983). The method to estimate the G:Q ratio in a given population of the tRNA family consisted of following the replacement of guanine in position 34 by a labeled guanine exerted by a guanine transglycosylase (insertase) of E . coli (Okada and Nishimura, 1979; Dess, 1982, 1983; Kersten et al., 1983). The results obtained in Xiphophorus by measurement of [3H]guanine incorporation in the tRNAs for Asn, Asp, His, and Tyr, differing in the ratio of G:Q in position 34 are summarized in Fig. 40. The graphs show the kinetics of the exchange of G 34 of the tRNA family by [3H]Gua, which is the reaction used to evaluate the amount of (Q)-tRNA (Okada and Nishimura,

cm o k

ml I m 2A

a& m5C

A

-0

t6A mt6A i6A msi6A

I Gm

Cm rncrn5~ crncm5~

NH

B

HO

OH

CONTROL OF ONCOGENES

259

1979). The fish genotypes and phenotypes are identical to those shown in Fig. 18. For each measurement several hundred individuals were used. In accordance with the findings of many investigators working in other differentiation systems (see Nass, 1983) [3H]guanine incorporation is high if the tRNAs are prepared from malignant melanomas that consist predominantly of poorly differentiated cells. In contrast, the incorporation is low if the tRNAs are derived from benign melanomas that consist predominantly of well-differentiated cells. The tRNAs for Asn, Asp, His, and Tyr of the malignant melanomas, therefore, are G-rich, whereas those of the benign melanomas are Q-rich (Fig. 40B). To decide whether the distinct difference in G:Q ratios between benign and malignant melanoma is Diff dependent or represents an epiphenomenon of benignancy and malignancy, the skin of nontumorous littermates that segregate into animals carrying Dijf and lacking Dijf like the tumorous fish in a 1:1 ratio (see Fig. 18) was used for the measurement (Fig. 40C). The Dijflacking segregants (specified by the lack of Est-1) had always higher amounts of Q-lacking tRNA than the Dijfcarrying ones. The skin of corresponding segregants derived from crossings between stocks other than that used in Fig. 40 showed the same differences in [3H]guanine incorporation (Dess, 1982; Kersten, 1982a; F. Anders, 1982). The skin of the parent animals used for the initial crosses showed the same differences: X . helleri that lacks the Dijfhas a high [3H]Gua incorporation (i.e., is G-rich) whereas X. muculatus that contains the Dijfhas a lower ["]Gus incorporation (i.e., is Q-rich). Similar differences, although not as pronounced as in the skin and tumors, were also found in the nontumorous liver. From these results we conclude that the differences of G:Q ratios between benign and malignant melanoma are no epiphenomena of benignancy and malignancy, but are very closely related to the primary effect of Dijfthat in tumorous fish converts the malignant to the benign state. The differences in the functional properties of Q-containing and Q-lacking tRNAs requires further elucidation. The (Q)tRNAs are suggested to prefer condons NAU over NAC, whereas the Q-lacking tRNAs read NAC and NAU equally well (Nishimura, 1983). This can be an important mechanism in the regulation of translation. For eukaryotic tRNATyrit has been shown that the Q-lacking species reads a terminator codon, probably UAG (Bienz and Kubli, 1981). Therefore the Q-lacking and Q-containing tRNAs of vertebrates might select mRNAs for translation by a regulatory mechanism, a

FIG. 39. (A) General cloverleaf structure of tRNA and positions of modified nucleosides. Positions and abbreviations of modifications in Sprinzel and Gauss (1982). (B) Structure of queuosine. From Kersten et ~ l (1983). .

FIG.40. Incorporation of [SHIguanine in position 34 of tRNA for Asp, Asn, His, and Tyr of Xiphophorus catalyzed by tRNA-guanine-transglycosylase (insertase) of E. coli. The graphs show the kinetics of the exchange of 6 3 4 of tRNA by [3H]Gua, a reaction used to evaluate the amount of ( Q - ) tRNA (Okada and Nishimura, 1979). A.B.C, according to the fish shown below the curves. These fish correspond to those shown in Fig. 18. High incorporation of [3H]Gua in the Dqf-lacking animals corresponds to low content of Q, whereas low incorporation of [3H]Gua in D@-containing X. helleri; (0) X . mnculatus. (B) Melanoma of BC segregants: (0) animals corresponds to high content of Q . (A) Skin of pure-bred Xiphophorus: (0) malignant; (H)benign. (C) Skin of nonmelanomatous BC segregants: (A) lacking Di& containing Dijf (identified by esterase-1). Note that comparable D@-containing animals always have a lower G content and higher Q content than Dqfilacking ones. See Fig. 18: The pairs of Dijjfand Dilff- values (10 and 15; 15 and 20; 10 and 15) are based on these and additional data of this kind. Data from Kersten et al. (1983).

(A)

CONTROL OF ONCOGENES

261

process already demonstrated for bacteria and designated as termination transcription control (Yanofsky, 1976; for details see Kersten et al., 1983). V. Theoretical Considerations on a General Concept of Neoplasia

A number of theories have been put forth concerning the origin of neoplasia; some have suggested that tumors are caused by viruses (Huebner and Todaro, 1969), while others have proposed that altered control mechanisms of cell differentiation are the main cause (Markert, 1968; Pitot, 1968), often brought about by promoters (Cairns, 1982; Weinstein et al., 1982). Further, it has been suggested that somatic mutations induced by physical or chemical agents might have relevance to the origin of tumors (Boveri, 1929; Bauer, 1928), but there are also genetically conditioned tumors that arise apparently independent of any external influence (Knudson, 1973, 1982). Evaluating these theories, it appears that, regardless of what causes neoplasia, three basic events leading to neoplasia are always the same, namely (1) neoplastic transformation, followed by (2) restrained cell differentiation, and (3) continued cell proliferation. This implies that the principle underlying neoplasia must be also always the same (Holley, 1975; Ahuja and F. Anders, 1976). A. T H ECOMMON BASISOF NEOPLASIAI N METAZOA There is a considerable accnmulation of evidence that the basic prerequisite of tumor formation is a genetic factor which, under normally regulated conditions probably exerts essential but unknown normal functions in the multicellular organization of apparently all metazoa (Sections 11,F and 111,B); if deregulated, however, it appears as an oncogene or an accumulation of oncogenes, respectively (Section IV,A). Whether the oncogene is primarily a cellular gene or is contributed as a viral oncogene which is anyway derived from a cellular gene, is of minor importance in this context.

B. THE COMMON BASISOF NEOPLASIAOF DIFFERENT TISSUES While the development of the large variety of mesenchymal, epithelial, and neurogenic neoplasms observed in vertebrates is mostly considered as completely independent, the Xiphophorus model provides some new perspectives for a unified consideration of this diversity. The different kinds of neoplasia are possibly mediated by a common Tu oncogene, or set of oncogenes that compose Tu, which is probably present in all tissues (Fig. 8). Tissue specificity of the activity of Tu, however, comes apparently from

262

FRITZ A N D E R S ET AL.

tissue-specific sets of regulatory genes (RMes;REpi;RNero; RMel in Fig. 8) which, if impaired or deleted, permit further events leading to the neoplasms in the respective tissue(s). In addition, there also seems to exist a genetic mechanism that controls Tu independently from the respective tissue. This is derived from the observation that, following treatment with a carcinogen, fish like the BC hybrids shown in Figs. 15 and 21, may develop multiple tumors such as melanoma, neuroblastoma, rhabdomyosarcoma, and epithelioma. Our results, therefore, may unify the origin of the different kinds of neoplasia in Xiphophorus to a common principle. It would be worthwhile to examine whether the unity of neoplasia suggested in the Xiphophorus model may be also valid for neoplasia in other vertebrates including humans.

C. THE COMMON BASISOF TUMORETIOLOGY Since there are many factors that regulate the oncogene, many events are required for the disturbance of this regulation: the process leading to neoplasia is, necessarily, a multistep process. This is not to say that the final event, i.e., neoplastic transformation of a cell itself, is a multistep process. The Xiphophorus model has rather shown that neoplastic transformation of a particular cell is an all-or-nothing-process which represents the last step of a long chain of events that dismantle the regulatory gene system for Tu. This dismantlement may start with a series of hybridization, may accumulate through many generations, and, finally, may complete it in a somatic cell during the period from youth to senescence of a particular individual. The sequence of events leading to the last step, however, may be specific to the particular individual, and to the nature of the carcinogenic trigger, thus providing the basis for the singularity of a particular tumor. The laboratory fish stocks derived from wild populations require a large variety of events to dismantle the regulatory gene system for the Tu oncogene. They develop neoplasia only if the many steps leading to neoplasia are completed by a combination of different carcinogenic influences, such as (1) elimination of regulatory genes from the germ line by selective mating, (2) impairment or deletion of regulatory genes in the germ line by mutagens, (3) impairment or deletion of the last active regulatory gene in a somatic cell by mutagens, and (4)promotion of noncompetent cells to competence for neoplastic transformation. It is easy to see that the specific cancer etiology that finally comes to our notice after the animals are treated is determined by the next to last step completing the multistage process permitting that last step, i.e., neoplastic transformation that initiates tumor formation (A. Anders and F. Anders, 1978).

CONTROL O F ONCOGENES

263

D. A UNIFIEDVIEWO F TUMOR ETIOLOGY So far we have recognized six main types of tumor etiology all of which are based on deregulation of the Tu oncogene. These types are neoplasia conditioned by (I)carcinogen-dependent somatic mutation, (2)carcinogen-dependent germ line mutation, (3) crossing-conditioned gene elimination, (4) promoter-dependent cell differentiation, (5) carcinogen-dependent chromatin condensation, and (6) increase of ongocene dosage in somatic cells or germ line. This will be demonstrated in genotypes showing a high susceptibility to melanoma based on the presence of one copy of an accessory Tu that is repressed by only one regulatory gene (Fig. 41; for genotype see Fig. 19). 1 . Neoplasia Conditioned by Carcinogen-Dependent Somatic Mutation

If the only regulatory gene present in the system remains unchanged (Fig. 41, left), or becomes impaired or deleted in a postcompetent cell like an A melanoblast (Aa), no melanoma will develop. Melanoma will, however, develop if the mutation of the regulatory gene occurs in an I melanoblast (Ab). This cell is competent and becomes neoplastically transformed. Following the processes of cell division and cell differentiation, the TrI melanoblast gives rise to an easily detectable Tr cell clone. These Tr cells may continue to divide, but finally differentiate to Tr melanophores; these, after having reached a certain age, are removed by macrophages. Because less differentiated Tr cells are not present, the development of the melanoma stops. Eventually the melanoma regresses. As is demonstrated in the scheme, the origin of such a somatic mutation-conditioned melanoma is unicellular and its growth proceeds only by proliferation. Following complete removal, we observe no relapse. This was expected because the system lacks supply with cells competent for transformation. The mutation of the regulatory gene may also occur in an S melanoblast (Ac), or an earlier precompetent pigment cell. This cell remains nontransformed and may multiply over a long period (“latent period”) as a normal stem cell. Those descendants reaching the stage of competence by differentiation are transformed simultaneously. After some cell divisions, paralleled by cell differentiation, they become visible as a large cell clone consisting of hundreds or thousands of dividing TrA melanoblasts and Tr melanocytes, which give rise to the melanoma. Those melanoma cells, that complete differentiation to Tr melanophores, are attacked and removed by macrophages. Those S melanoblasts, however, which do not further differentiate may reproduce identically throughout the further life of the fish and may serve as a permanent source of I melanoblasts, which then becomes neoplastically transformed. As explained by means of the scheme, the origin of such a melanoma is multicellular, although it can be traced back to a single

FIG.41. Differentiation of normal and neoplastically transformed pigment cells in different etiological types of neoplasia. See text. From F. Anders et ~ l (1981b). .

CONTROL OF ONCOGENES

265

mutational event in a somatic cell. It grows by both permanent transformation and proliferation of the descendants of the mutated cell. After complete removal of the melanoma there may or may not be a relapse. We assume that relapse occurs if some mutated stern cells in the surroundings of the melanoma escape removal and differentiate to the competent I melanoblasts, which become neoplastically transformed. No relapse, however, has ever been observed after extirpation of entire fins or large parts of the skin that carry the localized clonal melanomas.

2. Neoplasia Conditioned h y Carcinogen-Dependent Germ Line Mutation The same genotype that develops the somatic mutation-conditioned melanomas may also produce germ line mutation-conditioned neoplasms (Fig. 41B) (A. Anders et n l . , 1981; Chatterjee et al., 1981). In this case melanoma develops following carcinogen-induced impairment or deletion of the only regulatory gene in a germ line cell, and is inherited according to Mendelian rules. As a consequence of the inheritance of the mutation through the germ line, the Tu oncogene becomes active in the developing progeny as soon as the melanophore precursor cells differentiate to the competent I melanoblasts, and melanoma develops “spontaneously” in embryos and young individuals. This type of neoplasm is of multicellular origin and grows by both transformation and proliferation. In the event that it is restricted to a certain compartment of the body (see Section IV,C) removal is possible. The melanoma shown, however, relapse, because stem cells still remain available for differentiation to the competent cells, which become transformed. Although the somatic mutation-conditioned and the germ line mutationconditioned melanomas are apparently of completely different etiology (they are hereditary or nonhereditary tumors, respectively) their cells must be genetically identical if one considers that they originate from the mutation of the same regulatory gene and the activity of the same Tu oncogene. Morphological, pathological, and cytological studies, however, unexpectedly revealed many differences between these two types of melanomas. For instance, germ line mutation-conditioned melanomas lack chromosome aberrations whereas such aberrations are typical for somatic inutation-conditioned melanoma (Chatterjee et al., 1981). These differences, we believe, represent epiphenomena of tumor development and may be useful for diagnosis and prognosis of tumors of unknown etiology in the fish.

3. Neophsia Conditioned by Crossings Although the etiology of this type of tumor (Fig. 41C) is quite different from that of the germ line mutation-conditioned melanoma, both these types of melanoma are, however, closely related (Fig. 41, compare B and C). The solitary regulatory gene present in the system, if nonlinked to Tu, can be

266

FRIm ANDERS ET AL.

eliminated by crossings with animals that do not carry this gene (Figs. 16 and 18).Neoplasms of this type are always passed on to the offspring if the crucial regulatory gene remains lacking, and are not passed on (and therefore appear nonhereditary) if chromosomes lacking the regulatory gene are resubstituted by those carrying them. Like the germ line mutation-conditioned melanomas, the crossing-conditioned ones occur as the cells become competent for neoplastic transformation. The resulting tumor is of multicellular origin and grows and invades the nontumorous tissues by both transformation and proliferation. Following complete removal, which is possible in the case of compartment-specific tumors, they show relapse because stem cells differentiate to the competent stage and become transformed. In our laboratory we harvest tissues of melanoma of this type several times during the life of the fish. Harvest of the melanoma results in a postponement of death caused by neoplasia.

4 . Neoplasia Conditioned by Promoter-Dependent Cell Dqferentiation Neoplasia of this type can be induced in fish of certain genotypes the Tu of which is pigment cell-specifically derepressed by impairment or deletion of the only regulatory gene, but cannot become active because melanophore differentiation is delayed in the stage of the noncompetent S melanoblasts (see g and R,, in Fig. 8, and D in Fig. 41) (A. Anders et aZ., 1983).Chemical agents and drugs, such as 27-methyltestosterone and other steroids (A Schartl, 1981; A. Schartl et al., 1982; M. Schartl et al., 1981), cyclic AMP, corticotropin, BUdR, 12-0-tetradecanoylphorbol-13-acetate(TPA), saccharine, cyclamate, diazepam, and others (Herbert, 1983; Schmidt, 1983), as well as general environmental changes, such as the decrease of the temperature and the increase of the salinity of the water in the tank, promote almost simultaneously the differentiation of large amounts of the noncompetent cells to competent onces, which subsequently become neoplastically transformed. X Rays and N-methyl-N-nitrosourea (MNU), which are powerful mutagenic carcinogens, may also trigger neoplasms of this type. It appears, however, that in this case neither of these agents acts as a mutagen, but as a differentiation-promoting agent, such as methyltestosterone, cyclic AMP, low temperature, etc., which certainly are not mutagens. The melanoma of this type of etiology may develop simultaneously within the whole skin and inside the body wherever the differentiation of the pigment cell precursors is promoted from the noncompetent to the competent stage. They are of multicellular origin and grow permanently by both transformation of normal cells and proliferation of Tr cells. They are nonhereditary. We use the fish developing neoplasia of this kind of etiology as tester fish for tumor promoters (A. Anders et al., 1983; Schmidt, 1983).

CONTROL OF ONCOGENES

267

5. Neoplasia Conditioned by Carcinogen-Dependent Chromatin Condensation

The principle underlying this etiological type of neoplasia was first recognized when pure-bred spotted X . maculatus from Rio Jamapa were treated as embryos with X rays. The developing animals exhibited an enhancement of spots to benign melanoma (Pursglove, 1972; Pursglove et al., 1971) and have inherited this change of phenotype for about 45 generations. Genetic analysis showed that all chromosomes were involved in the phenotypic change (A. Anders et al., 1971), and phenogenetic, cytological (Lueken and Knoll, 1968), electron microscopic, and biochemical studies revealed a correlation between the hereditary enhancement of oncogene expression and a hereditary change in interphase chromatin appearance from a dispersed to a condensed state (F. Anders et al., 1981a; Heil, 1983). Similar experiments were done by treatment of the hybrid fish carrying only one regulatory gene for its Tu oncogene with several carcinogens. Almost all treated fish developed hereditary melanomas indicating that the regulatory gene is affected rather by a general hereditary change of the chromatin than by mutation (Schmidt, 1983). Neoplasia based on a general change of the genome following treatment with X rays and chemicals has also apparently been observed in mice (Nomura, 1982). Additional studies are required for an understanding of this phenomenon. 6. Neoplasia Conditioned by Additional lntroduction of Uncontrolled Oncogenes Many genotypes have been produced from the Tu oncogenes which are derepressed either by mutation-conditioned impairment or hybridizationconditioned elimination of the regulatory gene, but these animals only developed some neoplastically transformed cells instead of malignant melanoma. Following inbreeding, the offspring carrying the double dosage of Tu (the fish homozygous for Tu) developed malignant melanoma (F. Anders and Klinke, 1966). The genetic situation is similar to that of the oncogene dosage effect shown in Figs. 27 and 29. It is easy to see that the etiology of hybridization-conditioned and that of inbreeding-conditioned neoplasms is closely related. The crucial event in the inbreeding-conditioned neoplasms is, however, the introduction of an additional oncogene into the system. Introduction of additional oncogenes that are derepressed is also the basis of experimental oncogene transfer by DNA injection of embryos that led to the induction of neoplastically transformed cells (Section IV,A,3). Additional types of tumor etiology are conceivable, for instance neoplasia

268

FRITZ ANDERS ET A L .

following the deficiency of repair of damage to DNA or following the deficiency of immune surveillance. These and other deficiencies, however, are involved in control of tumor growth rather than in the actual process of neoplastic transformation. VI. Conclusions

The Xiphophorus tumor system has provided the opportunity to reduce the enormous complexity of cancer etiology to a few biological elements basically involved in neoplasia. The development of a tumor requires an oncogene which, after impairment, deletion, or elimination of its regulatory genes is permitted to mediate neoplastic transformation. Emphasis is being placed today in cancer research on the actual oncogenes themselves, but, in our opinion, the most important genes involved in neoplasia are these regulatory genes. However, although detected by classical genetics in the Xiphophorus system, these genes are not at present open to a more finely detailed molecular biological analysis. Their actual mode of action is therefore still far from being understood. ACKNOWLEDGMENTS This work has been generously supported by the Deutsche Forschungsgemeinschaft since 1957. The authors also gratefully acknowledge the support given by the President of the JustusLiebig-Universitit Giessen, and by the Bundesminister fur Forschung und Technologie through many years. Thanks are due to K5te Klinke, Birgit Krauskopf, Kristine Kruger, and Helene Schifer-Pfeiffer for valuable help in this research. Thanks are also due to Scott Robertson, University of British Columbia, Vancouver, for critical reading of the manuscript.

REFE HEN c: E s Abdo, S. (1979). Thesis, University of Giessen. Ahuja, M. R . (1965). Q . Reu. Biol. 4, 329-340. Ahuja, M. R. (1979). In “Antiviral Mechanisms in the Control of Neoplasia” (P. Chandra, ed.), pp. 17-37. Plenum, New York. Ahuja, M. R., and Anders, F. (1976). Prog. E x p . Tumor Res. 20, 380-397. Ahuja, M. R . , and Anders, F. (1977). Recent Ado. Cancer Res. Cell Biol. Tumor Virol. 1, 103117. Ahuja, M. R., Lepper, K., and Anders, F. (1979). Experiinentia 35, 28-29. Ahuja, M . R . , Schwab, M., and Anders, F. (1980).I . Hered. 71, 403-407. Altmaier, A , , and Anders, F. (1968). Verh. Dtsch. Zool. Ces. Zool. Anz. 32 (Suppl.), 225-238. Anders, A,, and Anders, F. (1963). Z. Vererbrungslehre 94, 1-18. Anders, A,, and Anders, F. (1978). Biochim. Biophys. Acta 516, 61-95. Anders, A,, Anders, F., and Pursglove, D. L. (1971). Experientia 27, 931-932. Anders, A , , Anders, F., and Klinke, K. (1973a). In “Genetics and Mutagenesis of Fish” (J. H. Schroder, ed.), pp. 53-63. Springer-Verlag. Berlin and New York. Anders, A., Anders, F., and Klinke, K . (1973b). In “Genetics and Mutagenesis in Fish” (J. H. Schroder, ed.), pp. 33-52. Springer-Verlag, Berlin and New York.

CONTROL OF ONCOGENES

269

Anders, A,, Kollinger, G . , and Chatterjee, K. (1981). In “Pigment Cell 1981” (M. Seiji, ed.), pp. 501-506. Univ. of Tokyo Press. Anders, A., Schmidt, C.-R., Herbert, A,, and Anders, F. (1983). In “Prufung von Cheniikalien auf Kaneerogenitit, Mutagenitat und Teratogenitit,” pp. 253-268. Gesellschaft fur Strahlen und Umweltforschung, Munchen. Anders, F. (1967). Experientiu 23, 1-10, Anders, F. (1968). Zentrulbl. Vet. Med. R 15, 29-46. Anders, F. (1981). Klin. Wochenschr. 59, 943-956. Anders, F. (1982). In “Biochemistry of Differentiation and Morphogene. pp. 91-1 15. Springer-Verlag, Berlin and New York. Anders, F. (1983). I n “Modern Trends in Human Leukemia V” (R. Neth, R. C. Gallo, M . F. Greaves, M. A. S. Moore, and K . Winkler, eds.), pp. 186-206. Springer-Verlag. Berlin and New York. Anders, F., and Klinke, K. (1965). Z. Vererbungslehre 96, 49-65. Anders, F., and Klinke, K. (1966). Verli. Dtsch. Zoo/. Ges. Zool. Anz. 30 (Suppl.), 391-401. Anders, F., Anders, A., and Klinke, K. (1962a). Verh. Dtsch. 2001.Ges. 2001.Anz. 26 (Suppl.), 97-118. Anders, F., Vester, F., Klinke, K., and Schuhmaclier. H. (1962b). Biol. Zentrulbl. 81, 45-65. Anders, F., Drawert. F., Klinke, K . , and Reuther, K. H. (1963). Experientiu 19, 219-224. Anders, F., Sieger, M., and Klinke, K. (1969). Experientia 25, 871-814. Anders, F., Klinke, K., and Vielkind, U. (1972). B i d . Zeit 2, 35-45. Anders, F., Anders, A., and Vielkind, U. (1974). Int. Cancer Congr., I l t h , Florence 3, 305 (Abstr.). Anders, F., Diehl, H . , Schwali, M., and Anders, A. (1979,). “Pigment Cell” (S. N. Klaus, ed.), Vol. 4, pp. 142-149. Karger, Basel. Anders, F., Scholl, E., and Schartl, M. (197911).In “Neue Perspektiven in der Medizin” (H. Porcher and K. Theurer, eds.), pp. 38-100. Enke, Stuttgart. Anders, F., Diehl, H., and Scholl, E. (1980). Linnean Soc. Sy~np.Ser. 9, 211-224. Anders, F . , Chatterjee, K., Schwal), M . , Scholl, E., and Anders, A. (1981a). Am. Zool. 21, 535-548. Anders, F., Schartl, M . , and Scholl, E. (198lb). I n “Phyletic approaches to Cancer” (C. J. Ilawe, ed.), pp. 289-309. Japan Scientific Societies Press, Tokyo. Anders, F . , Scholl, E., and Schartl, M. (1981~).In “Pigment Cell 1981” (M. Seiji, ed.), pp. 491-499. Univ. of Tokyo Press. Anders, F., Schwab, M . , and Scholl, E. (1981d). In “Short-term Tests for Chemical Carcinogens” (H. F. Stich and R. H. C . San, eds.), pp. 399-407. Springer-Verlag, Berlin and New York. Anders, F., Schartl, M., and Barnekow, A. (1984).J . Nut/. Cancer Inst., in press. Aronowitz, O., Nigrelli, R. F., and Gordon, M. (1951). Zoologica 36, 239-242. Atz, J. W. (1962). Zoo/ogicu 47, 153-181. Bagnara, J. T., and Hadley, M. E. (1973). “Chro~natophoresand Color Change.” Prentice-Hall, New York. Gorhnian, R. F., Nigrelli, R. F., and Gordon, M. (1953). Genetics 38, Baker, K. F., Berg, 0.. 655. Barnekow, A , , Schartl, M . , Anders, F., and Bauer, H. (1982). Cancer Res. 42, 2429-2433. Bauer, H . , Barnekow, A , , Boschek, C. B., Friis, H. R., and Zieniiecki, A. (1982). I n “Biochemistry of Differentiation and Morphogenesis” (L. Jaenicke, ed.), pp. 74-90. SpringerVerlag, Berlin and New York. Bauer, K. H. (1928). “Mutationstheorie der Gescliwulstentstehitng.” Springer-Verlag, Berlin and New York.

270

FRITZ ANDERS ET AL.

Bauer, K. H. (1968). In “Aktuelle Probleme aus dem Gebiet der Cancerologie 11” (H. Lettrb and G. Wagner, eds.), pp. 63-84. Springer-Verlag, Berlin and New York. Becker-Carus, C. (1965). Z. Wiss. Zool. A 172, 37-103. Beiderbeck, R. (1977). “Pflanzentumoren.” Ulmer, Stuttgart. Bentvelzen, P. (1972).In “RNA Viruses and Host Genome in Oncogenesis” (P. Emmelot and P. Bentvelzen, eds.), pp. 309-337. North-Holland Publ., Amsterdam. Berg, O., Edgar, M., and Gordon, M. (1953). Cancer Res. 13, 1-9. Bienz, M., and Kubli, E. (1981). Nature (London)294, 188-190. Bishop, J. M. (1982a). Sci. Am. 246, 69-78. Bishop, J. M. (1982b). Ado. Cancer Res. 37, 1-32. Boveri, T. (1929). “The Origin of Malignant Tumors.” Williams & Wilkins, Baltimore, Maryland. Braun, A. C. (1978). Biochim. Biophys. Acta 516, 167-191. Breider, H. (1939). Z. Wiss. Zool. 152, 107-128. Breider, H., and Seeliger, R. (1938). 2. Wiss. 2001.A 151, 243-285. Burch, P. R. J. (1976). “The Biology of Cancer: A New Approach.” MTP, Lancaster. Cairns, J. (1982). In “Carcinogenesis and Biological Effects of Tumor Promoters” (E. Hecker, N. E. Fusenig, W. Kunz, F. Marks, and H. W. Thielmann, eds.), pp. 647-651. Raven, New York. Cairns, J., Lyon, J. L., and Skolnick, M. (1980). “Cancer Incidence in Defined Populations.” Cold Spring Harbor Press, Cold Spring Harbor, New York. Chatterjee, K., Kollinger, G., Schmidt, C. R . , Anders, A,, and Anders, F. (1981). Cancer Genet. Cytogenet. 3, 195-209. Collet, M . S . , and Erikson, R. L. (1978). Proc. Natl. Acad. Sci. U.S.A. 75, 2021-2024. Collet, M. S . , Purchio, A. F., and Erikson, R. L. (1980). Nature (London) 285, 167-169. Crew, F. A., and Koller, P. (1936). Proc. R . Soc. (Edinburgh) 56, 210. Dalla-Favera, R., Bergni, M., Erikson, J., Patterson, D., Gallo, R. C., and Croce, C. M. (1982a). Proc. Natl. Acad. Sci. U.S.A. 79, 7824. Dalla-Favera, R., Gallo, R. C., Giallongo, A , , and Croce, C. (1982b). Science 218, 686. Dawe, C. D., and Harshbarger, J. C. (1969). Natl. Cancer Znst. Monogr. 31. Dawe, C. I., and Harshbarger, J. C. (1975). In “The Pathology of Fishes” (W. E. Ribelin and G. Migaki,eds.), pp. 871-894. Univ. of Wisconsin Press, Madison, Wisconsin. Dawe, C. J., Harshbarger, J. C., Kondo, S., Sugimura, T., andTakayama, S. (1981). “Phyletic Approaches to Cancer.” Japan Scientific Societies Press, Tokyo. DeFeo, D., Gonda, M . A., Young, H. A., Chang, E. H., Lowt, D. R., Scolnick, E. M., and Ellis, R. W. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 3328-3332. de Klein, A., van Kessel, A. G., Grosveld, G., Bartram, C. R., Hagemeijer, A., Bootsma, D., Spurr, N. K., Heisterkamp, N., Groffen, J., and Stephenson, J. R. (1982).Nature (London) 300, 765-767. Denlinger, R. H., Koestner, A., and Swenber, J. A. (1978). Cancer Res. 38, 1711-1717. Der, C. J . , Krontiris, T. G., and Cooper, G. M. (1982). Proc. Natl. Acad. Sci. U . S . A . 79, pp. 3637-3640. Dess, G. (1982). Thesis, Univ. Giessen. Dess, G . (1983). Thesis, Univ. Erlangen. Diehl, H. (1982). Thesis, Univ. Giessen. Dobzhansky, T. (1937). “Genetics and the Origin of Species.” Columbia Univ. Press, New York. Dushane, G . P. (1938).I . Exp. Zool. 78, 485-503. Dzwillo, M. (1964). Mitt. Hamb. Zool. Mus. 62, 15-22.

CONTROL OF ONCOGENES

27 1

Felding-Habermann, B., Anders, A., and Wiegandt, H. (1983). Znt. Symp. Glycoconjugates, 7th, Lund (Abstr.). Fitzpatrick, T. B., and Lerner, A. B. (1953). Science 117, 640-642. Forster, W., and Anders, F. (1977). Zool. Anz. 198, 167-177. Fuerst, P. A , , Chakraborty, R., and Nei, M. (1977). Genetics 86, 455-483. Gardner, E. J. (1980). In “Prevention and Cure of Hereditary Cancers” 0. Cairns, J. L. Lyon, and M. Skolnick, eds.), pp. 365-377. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Gateff, E. (1978). Science 200, 1448-1459. Gateff, E. (1982). Ado. Cancer Res. 37, 33-74. Gebhard, K., and Niebauer, F. (1979). Eur. Workshop Melanin Pigment., 4th London (Abstr.). Gillespie, D., and Gallo, R. C. (1975). Science 188, 802-811. Goldfarb, M., Schimizu, K., Perucho, M., and Wigler, M. (1982). Nature (London)296,404409. Gordon, M. (1928). J. Hered. 19, 551-556. Gordon, M. (1931). Am. J. Cancer 15, 1495-1523. Gordon, M. (1947a). Ado. Genet. 1, 95-132. Gordon, M. (1947b). J . Natl. Cancer Znst. 7, 87-92. Gordon, M. (1958). Ann. N.Y. Acad. Sci. 71, 1213-1222. Gross, L.,and Dreyfuss, Y. (1979). Proc. Natl. Acad. Sci. U.S.A. 76, 5910-5913. Haas, J. (1981). Thesis, Univ. Giessen. Haas-Andela, H. (1978). Thesis, Univ. Giessen. Halver, J. E., and Mitchel, I. A. (1967). Res. Rep. U.S. Fish Wildlije Sero. 70, 1-99. Haussler, G. (1928). Klin. Wochenschr. 7, 1561-1562. Hayes, H. M. (1978). Eur. J. Cancer 14, 1299-1308. Heil, M. (1983). Thesis, Univ. Giessen. Henze, M., Rempeters, G., and Anders, F. (1977). Comp. Biochem. Physiol. 56B, 33-64. Herbert, A. (1983). Thesis, Univ. Giessen. Heston, W. E. (1974).J . Hered. 65, 262-272. Heston, W. E., and Vlahakis, G. (1968). J . Natl. Cancer Znst. 40. 1161-1166. Higginson, J. (1969). Proc. Can. Cancer Conf., 8th, pp. 40-75. Holley, R. W. (1975). Nature (London) 258, 487-490. Hook, R. R., Aultman, M. D., Adelstein, E. H., Oxenhandler, R. W., Milliken, L. E., and Middleton, C. C. (1979). Znt. J . Cancer 24, 668-672. Huebner, R. J., and Todaro, G. J. (1969). Proc. Natl. Acad. Sci. U.S.A. 79, 1087-1094. Humm, D.G., and Young, R. S. (1956). Zoologica 41, 1-10, Huxley, J. (1958). “Biological Aspects of Cancer.” Allen & Unwin, London. Ishikawa, T., and Takayama, S. (1977). Ann. N.Y. Acad. Sci. 298, 330-341. Ishikawa, T., Prince Masahito, Matsumoto, J., and Takayama, S. (1978a).J . Natl. Cancer Znst. 61, 1461-1470. Ishikawa, T., Prince Masahito, and Takayama, S. (1978b). Cancer Res. 38, 3954-3959. Jacobs, C., and Riibsamen, H. (1983). Cancer Res. 43, 1696-1702. Kaiser, H. E. (1981). “Neoplasms-Comparative Pathology of Growth in Animals, Plants and Man.” Williams & Wilkins, Baltimore, Maryland. Kallman, K. D. (1968). “Enjoy your Platys and Swordtail.” Pet Library, New York. Kallman, K. D. (1970). Genetics 64, 32. Kallman, K. D. (1975). Handb. Genet. 4, 81-132. Kallman, K. D., and Atz, J. W. (1967). Zoologica 51, 107-135. Katze, J. R., Beck, W. T., Cheng, C. S., and McCloskey, J. A. (1983). In “Recent Results in

272

FRITZ ANDERS ET AL.

Cancer Research: Modified Nucleosides and Cancer” (G. Nass, ed.), pp. 146-159. Springer-Verlag, Berlin and New York. Kersten, H. (1982a). I n “Biochemistry of Differentiation and Morphogenesis” (L. Jaenicke, ed.), pp. 116-119. Springer-Verlag, Berlin and New York. Kersten, H. (1982b). In “Biochemistry of S-adenosylmetliionine and Related Compounds” (E. Usdein, R. T. Borchardt, and R. C. Creveling, eds.), pp. 357-369. Macmillan, New York. Kersten, H. (1983).In “Recent Results in Cancer Research: Modified Nucleosides and Cancer” (G. Nass, ed.), pp. 255-263. Springer-Verlag, Berlin and New York. Kersten, H., Schachner, E., Dess, G., Anders, A , , Nishimura, S., and Shindo-Okada, N. (1983). In “Biochemical and Clinical Aspects of Pteridines” (Wachter, Curtius, and Pfleiderer, eds.), pp. 367-382. De Gruyter, Berlin. Knudson, A. G. (1973). Ado. Cancer Res. 17, 317-352. Knudson, A. G. (1982). Proc. Lnt. Cancer Congr., L3th, Seattle (Abstr.). Koeppe, P. (1980). Biol. Med. 9, 99-110. Kollinger, G. (1980). Thesis, Univ. Giessen. Kollinger, G., and Siegmund, E. (1981). Verh. Dtsch. Zool. Ges. 74, 206. Kollinger, G., Schwab, M., and Anders, F. (1979).J . Cancer Res. Clin. Oncol. 95, 239-246. Kosswig, C. (1928). Z. Indukt. Abstamm. Vererb. Lehre 47, 150-158. Kosswig, C. (1929). Der Ziichter 1, 152-157. Kosswig, C. (1937). B i d . Gen. 13, 276-293. Kraybill, H. F . , Dawe, C. J . , Harshbarger, J. C., and Tardiff, R. 6. (1977).Ann. N.Y. Acad. Sci. 298. Krieg, K. (1973). B i d . Zentralbl. 92, 617-623. Krontiris, T. G., and Cooper, G. M. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 1181-1184. Leatherland, J. F . , and Sonstegard, R. A. (1978). Cancer Res. 38, 3164-3173. Levan, G., Ahlstrom, U., and Mittelman, F. (1974). ffereditas 77, 263-280. Little, C. C. (1947). B i d . Reo. 22, 315. Lubs, H. A , , Kimberling, W. J., Hecht, F., Patil, S. R . , Brown, J., Gerald, P., and Summit, R. L. (1977). Nature (London) 268, 631-632. Lueke, W. (1984). Thesis, Univ. Giessen. Lueke, W., and Anders, F. (1983). Znt. Pigment Cell Conf., 12th 131, (Abstr.). Lueken, W., and Knoll, C. (1968). Experientia 24, 595-596. Lueken, W., and Kaeser, U. (1972). Experientia 28, 1340-1341. Lueken, W., Schmidt, E. R., and Lepper, K. (1973).In “Genetics and Mutagenesis in Fish” (J. H. Schroder, ed.), pp. 139-160. Springer-Verlag, Berlin and New York. Lynch, H. T. (1967). Recent Results Cancer Res. 12. Maas, G. (1967). Thesis, Univ. Saarbrucken. Mark, J. (1977). Ado. Cancer Res. 24, 165-222. Markert, C. L. (1968). Cancer Res. 28, 1908-1914. Maugh, T. H . (1979). Science 205, 1363-1366. Maugh, T. H. (1980).J. tiered. 71, 2-8. Mawdesley-Thomas, L. E. (1975). In “The Pathology of Fishes” (W. E. Ribelin and G. Migaki, eds.), pp. 805-870. Univ. of Wisconsin Press, Madison, Wisconsin. Mittelman, F., and Levan, G. (1972). Mereditas, 71, 325-334. Mittelman, F., Mark, J., Levan, G., and Levan, A. (1972). Science 176, 1340-1341. Morizot, D. C., and Siciliano, M. J. (1979). Genetics 93, 947-960. Morizot, D. C., and Sicliano, M. J. (1982). Genetics 102, 539-566. Morizot, D. C., and Siciliano, M. J. (1983).J . Natl. Cancer Inst. 71, 809-813. Morizot, D. C., Wright, D. A,, and Siciliano, M. J. (1977). Genetics 86, 645-656. Muir, C. S . , and Nectoux, J . (1982). In “Cancer Epidemiology and Prevention” (D. Schotten-

CONTROL OF ONCOGENES

273

feld and J. F. Fraumeni, eds.), pp. 119-137. Saunders Company, Philadelphia, Pennsylvania. Nass, 6. (1983). “Recent Results in Cancer Research. Modified Nucleosides and Cancer.” Springer-Verlag, Berlin and New York. Neel, G. B., Jhanwar, S. C., Chaganti, R. S. K., and Hayward, W. S . (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 7842-7846. Nishimura, S. (1983). Prog. Nucleic Acid Res. Mol. B i d . , in press. Nishimura, S., and Kuchino, Y. (1979). Gann 24, 245-262. Nomura, T. (1982). Nature (London) 296, 575-577. O’Brien, S. J. (1980). 1. Hered. 71, 2-8. Oeser, H., and Koeppe, P. (1979). “Krehs: Schicksal oder Verschulden?” Thieme, Stuttgart. Okada, N . , and Nishimura, S. (1979). 1. B i d . Chem. 254, 3062-3066. Olsen, 6. (1966). Actu Chir. Scand. 60, 149-162. Oskarsoii, M . , McClements, W. L., Blair, D. G., V. Maize], J., and Vande Woude, G. F. (1980). Science 207, 1222-1224. Perucho, M., Goldfarh, M., Shimixu, K., Lama, C., Fogh, J., and Wigler, M. (1981). Cell27, 467-476. Peter, R. U. (1982). Eur. Workshop Melanin Pigment., 4th, Edinburgh (Abstr.). Peter, R. U. (1983). Thesis, Univ. Giessen. Pierce, G. B., and Wallace, C. (1971). Cancer Res. 31, 127-134. Pitot, H. C. (1968). Cancer Res. 28, 1880-1887. Poll, H. (1920). Beitr. Pathol. Anat. 67, 40. Prescott, 11. M. (1976). Ado. Genet. 18, 99-177. Prescott, 11. M., and Flexer, A. S. (1982). “Cancer: The Misguided Cell.” Sinauer, Sunderland, Massachusetts. Pulciani, S., Santos, E . , Lanver, A. V., Long, L. K., Aronson, S. A,, and Barbacid, M. (1980). Nature (London)300, 539-541. Pursglove, D. L. (1972). Thesis, Univ. Giessen. Pursglove, 11. L., Anders, A , , Doll, G., and Anders, F. (1971). Experientia 27, 695-697. Radda, A. C. (1980). Aquearia 27, 39-44. Reddy, E. P., Reynolds, R. K., Santos, E., and Barbacid, M. (1982).Nature (London)300, 149152. Rempeters, G., Henze, M., and Anders, F. (1981). Comp. Biochern. Physiol. 69B, 9198. Reyniers, J. P., Pleasants, J. R., Wostman, B. S., Katze, J. R., and Farkas, J. R. (1981).1.Biol. Chem. 256, 11591-11594. Riehl, R., Schartl, M., and Kollinger, G. (1984).1.Cancer Res. Clin. Oncol. 107, 21-31. Roseii, D. E. (1979). Bull. Am. Mus. Nut!. Hist. 162, 267-376. Rosen, 1).E., and Bailey, R. M. (1963). Bull. Am. Mus. Natl. Hist. 126, 1-176. Rowley, J. D. (1973). Nature (London) 243, 290-293. Rowley, J. D. (1974).1.Natl. Cancer Inst. 52, 315-320. Rowley, J. D. (1977). In “Genetics of Human Cancer” 0. J. Mulvihill, R. W. Miller, and J. F. Fraumeni, eds.), pp. 125-136. Raven, New York. Rowley, J. D. (1982). Science 216, 749-751. Rowley, J. D. (1983). Nature (London)301, 290-291. Sachs, L. (1978). Nature (London)274, 535-539. Sachs, L. (1982). In “Biochemistry of Differentiation and Morphogenesis” (L. Jaenicke, ed.), pp. 3-15. Springer-Verlag, Berlin and New York. Sandberg, A. A. (1980). “The Chromosome in Human Cancer and Leukemia.” Elsevier, Amsterdam.

274

FRITZ ANDERS ET AL.

Sato, S., Matsushiam, T., Tanaka, N., Sugimura, T., and Takashima, F. (1973).J. Natl. Cancer Inst. 50, 767-776. Scarpelli, D. G. (1969). I n “Fish in Research’ (0.W. Neuhaus and J. E. Nalver, eds.), pp. 4548. Academic Press, New York. Schartl, A. (1981). Thesis, Univ. Giessen. Schartl, A,, Schartl, M., and Anders, F. (1982). In “Carcinogenesis and Biological Effects of Tumor Promoters” (E. Hecker, N. E. Fusenig, W. Kunz, F. Marks, and H. W. Thielmann, eds.), pp. 427-434. Raven, New York. Schartl, M. (1979). Thesis, Unio. Giessen. Schartl, M . , and Barnekow, A . (1982). Dqferentiation 23, 109-114. Schartl, M., and Franchini, G. (1982). Unpublished (see Anders et al., 1983). Schartl, M., and Franchini (1984). In preparation. Schartl, M., and Czernilofsky, A. P. (1982). Unpublished (see Anders, F. 1982a). Schartl, M., Schartl, A,, and Anders, F. (1981). In “Pigment Cell 1981” (M. Seiji, ed.), pp. 517-524. University of Tokyo Press. Schartl, M., Barnekow, A., Bauer, H., and Anders, F. (1982). Cancer Res. 42, 4222-4227. Schartl, M., Barnekow, A., Schmidt, C. R., Bauer, H., and Anders, A. (1983). In preparation. Schlage, W. (1978). Thesis, Univ. Giessen. Schlumberger, H. G. (1957). Cancer Res. 17, 823-832. Schlumberger, H. G., and Luck& B. (1948). Cancer Res. 8, 657-712. Schmidt, C. R . (1983). Thesis, Univ. Giessen. Schmidt, E. R. (1978). Roux Arch. Entwicklungsmech. 184, 115-134. Scholl, A. (1973). In “Genetics and Mutagenesis in Fish” (J. H. Schroder, ed.), pp. 277-299. Springer-Verlag, Berlin and New York. Scholl, A,, and Anders, F. (1973a). In “Genetics and Mutagenesis of Fish” (J. H. Schroder, ed.), pp. 301-313. Springer-Verlag, Berlin and New York. Scholl, A,, and Anders, F. (1973b). Arch. Genet. 46, 121-129. Scholl, E. (1977). Thesis, Univ. Giessen. Scholl, E. (1979). Mitt. Hamb. Zool. Mus. Inst. 76, 489-490. Schull, W. J. (1979).J. Toricol. Enoiron. Health 5, 17-25. Schwab, M. (1974). Thesis, Univ. Giessen. Schwab, M. (1980). Thesis for Habilitation, Univ. Giessen. Schwab, M. (1982a). In “Carcinogenesis and Biological Effects of Tumor Promoters” (E. Hecker, N . E. Fusenig, W. Kunz, F. Marks, and H. W. Thielmann, eds.), pp. 417-426. Schwab, M. (1982b). Mol. Gen. Genet. 188, 410-417. Schwab, M., and Anders, A. (1981). In “Neoplasms-Comparative Pathology of Growth in Animals, Plants, and Man” (H. E. Kaiser, ed.), pp. 451-459. Williams & Wilkins, Baltimore, Maryland. Schwab, M., Ahuja, M. R., and Anders, F. (1976a). Comp. Biochem. Physiol. 54B, 197-199. Schwab, M., Vielkind, J., and Anders, F. (1976b). Mol. Gen. Genet. 144, 151-158. Schwab, M., Haas, J., Abdo, S., Ahuja, M. R . , Kollinger, 6.. Anders, A,, and Anders, F. (1978a). Experientia 34, 780-782. Schwab, M., Abdo, S., Ahuja, M. R., Kollinger, G., Anders, A., and Anders, F. (1978b). Z. Krebsforsch. 91, 301-315. Schwab, M., Kollinger, G., Haas, J., Ahuja, M. R., Abdo, S., Anders, A., and Anders, F. (1979). Cancer Res. 39, 519-526. Seiler, J., Puchta, O., Brunold, E., and Rainer, M. (1958). Roux Arch. Entwicklungsmech. 150, 199-372. Shilo, B. Z., and Weinberg, R. A. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 6789-6792. Siciliano, M. J., and Perlmutter, A. (1972). J. Natl. Cancer Inst. 49, 415-421.

CONTROL OF ONCOGENES

275

Siciliano, M. J., and Wright, D. A. (1976). Prog. Exp. Tumor Res. 20, 398-411. Sieger, F., Sieger, M . , Priissing, R., and Anders, F. (1969). Experientia 25, 778-779. Sieger, M., Sieger, F., and Anders, F. (1968). Verh. Dtsch. Zool. Ges. Zool. Anz. 32(Suppl.), 2 10-225. Siegmund, E. (1982). Thesis, Univ. Giessen. Sinnhuber, R. O., Hendricks, J. D., Wales, J. H., and Putman, G. B. (1977).Ann. N.Y. Acad. Sci. 298, 389-408. Sobel, H. J. Marquet, E . , Kallman, K. D., and Corley, G. J. (1975). In “The Pathology of Fishes” (W. E. Ribelin and 6. Migaki, eds.), pp. 945-981. Univ. of Wisconsin Press, Madison, Wisconsin. Spector, D. H., Varmus, H. E., and Bishop, J. M. (1978). Proc. Natl. Acad. Sci. U.S.A. 75, 4102-4106. Sprinzl, M., and Gauss, D. H. (1982). Nucleic Acids Res. 10, rl-r55. Stehelein, D., Varmus, H. E., and Bishop, J. M. (1976). Nature (London) 260, 170-173. Strong, L. C. (1958). Ann. N.Y. Acad. Sci. 71, 807-1241. Stiinzi, H. (1972). Heintierpraxis 17, 66-70. Tabin, C. J . , Bradley, S. M., Bargmann, C. I., Weinberg, R. A., Papageorge, A. G., Scolnick, E. M., Dhar, R., L o w , D. R., and Chang, E. H. (1982).Nature (London)300, 143-149. Takayama, S., and Ishikawa, T. (1977). lARC Sci. Publ. 16, 271-286. Takeya, T., and Hanafiusa, H. (1982).J . Virol. 44, 12-18. Taub, R., Kirsch, I., Morton, C., Lenoir, G., Swan, D., Tronick, S . , Aaronson, S., and Leder, P. (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 7837-7841. Vielkind, J. (1971). Thesis, Univ. Giessen. Vielkind, J. (1979). Thesis for Habilitation, Univ. Giessen. Vielkind, J., Vielkind, U., and Anders, F. (1971). Cancer Res. 31, 868-875. Vielkind, J.. Vielkind, U., and Anders, F. (1973a).Z. Krebsforsch. 80, 103-112. Vielkind, J., Schwab, M., and Anders, F. (1973b).In “Genetics and Mutagenesis in Fish” 0. H. Schroder, ed.), pp. 123-137. Springer-Verlag. Berlin and New York. Vielkind, J . , Haas-Andela, H . , Vielkind, U . , and Anders, F. (1982).Mol. Gen. Genet. 185,379389. Vielkind, U. (1972). Thesis, Univ. Giessen. Vielkind, U. (1976).J . Exp. Zool. 196, 197-204. Vielkind, U., Schlage, W., and Anders, F. (1977). Z. Krebsforsch. 90, 285-299. Warner, N. L., Potter, M., and Metcalf, D. (1974). U/CC Tech. Rep. Ser. Genf. 13. Weinstein, I. B., Horowitz, A. D., Mufson, R. A , , Fisher, P. B., Ivanovic, V., and Greenbaum, E. (1982). In “Carcinogenesis and Biological Effects of Tumor Promoters” (E. Hecker, N . E. Fusenig, W. Kunz, F. Marks, and H. W. Thielmann, eds.), pp. 599-613. Raven, New York. Weiss, E. (1972). In “Allgemeine Pathologie” (A. Frei, ed.), pp. 295-347. Paray, Berlin. Weissenfels, N., Schifer, D., and Bretthauer, R. (1970). Virchows Arch. Abt. B 5, 144-158. Weston, J. A. (1970). Ado. Morphog. 8, 41-114. Yamada, K., and Hasegawa, T. (1978). Hum. Genet. 44, 89-98. Yanofsky, C. (1976). I n “Molecular Mechanisms in the Control of Gene Expression” (D. P. Nierlich, W. J . Rutter, and C. J. Fox, eds.), pp. 75-86. Academic Press, New York. Young, J . L., and Pollack, E. S. (1982). In “Cancer Epidemiology and Prevention” (D. Schottenfeld and J . F. Fraumeni, eds.), pp. 138-165. Saunders, Philadelphia, Pennsylvania. Zander, C. D. (1967). Mitt. Hamb. Zool. Mus. Inst. 64, 87-125. Zech, L., Haglund, V., Nilsson, K., and Klein, G. (1976). Int. J . Cancer 17, 47-56.

This Page Intentionally Left Blank

CONTRASUPPRESSION: THE SECOND LAW OF THYMODYNAMICS, REVISITED Douglas R. Green and Richard K. Gershon' Department of Pathology and the Howard Hughes Medical Institute for Cellular Immunology, Yale University School of Medicine, New Haven, Connecticut

I. Introduction ............. 11. Defining Cell ................................................... A. Tecliniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Cellular Make-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Defining Contrasuppression. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Defining a Specific Contrasuppressor Circuit and Assigning a Unique Phenotype to Its Cellular and Molecular Members.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , A. The Inducer Cell .... B. The Transducer Cell.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , , . , , , , C. The Effector Cell . . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . . . . . . , . . . . . I). Cellular Interactions in the Action of an Antigen-Specific Contrasuppressor Factor in Cell-Mediated Immunity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Functional Activity of the Contrasuppressor Circuif VI. Conditions That Influence the Generation and/or Activation of Contrasuppressor Cells. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Ontogeny . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. The Role of the Antigen Preset ............... C. Special Antigenic Determinants . . . . . . . . . . . . . . . . . D. Antigen Dose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Genetics. . . . . . ...... ells. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Crossreactivity of the Ly-2 Contrasuppressor Inducer Cell . . . . . . . . . . . . . . . . H. Adjuvants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Immunological Consequences of Activation of the C ............... A. Hyperimniunity . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Adoptive Transfer of Immune Response.. . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . C. Microenvironmental Immune Regulation D. Immune Response to Malaria.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Recovery from Trauma Associated Suppression . . . . . . . . . . . . . . . . . . . . . . . . . . F. Immune Regulation in Old A g e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Autoimmunity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII. Contrasuppression and Tumor Immunity A. Theoretical Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Contrasuppression i n Tumor Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Contrasuppression in Tumor Therapy. . . . . . D. Further Evidence for Contrasuppression in Tumor Immunity . . . . . . . . . . . . .

278 280 280 28 1 282 282 282 283 283 284 285 286 286 289 293 296 297 297 298 300 301 30 1 303 305 308 310 310 310 312 312 313 313 314

'Richard K. Gershon died on July 11, 1983. This article is affectionately dedicated to his memory.

277 ADVANCES IN CANCEH HESEAHCH, \'OL 42

Copyright B 1984 by Academic Press, Inc. All rights of rtJprndoction in any form reserved. ISBN 0-12-oofi(j42-4

278

D O U G L A S R. G R E E N A N D RICHARD K . GERSHON

E. Class I Antigens in Contrasuppression and Tumor Immunity . . . . . . . . . . . . . . 315 F. Contrasuppression in Enhancement of Lymphoid Tumor Development. . . . . 317 IX. Human Examples of Contrasuppression. ..................... X. Contrasuppression in the Fut References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Addendum 1: Heterogeneity of Contrasuppressor T Cell Function and Specificity . . . . . . . . . . . . . . . . . . . . . . . . . . . 326 Addendum 2: Relation of the Allogeneic Effect of Contrasuppression . . . . . . . . . . 331

To he a great scientist you have to be a mental athlete. You have to think until it hurts, and then you have to keep on thinking through the pain. Then maybe you’ll find some new answers. When you get up to bat swing for the fences. (R. K. Gershon)

I. Introduction2

Contemporary physics and most of modern science proceeds on the assumption that the universe can be described by a finite number of generalized statements from which specific systems can be deduced by logical (and mathematical) manipulation. In the eighteenth and nineteenth centuries the laws of universal gravitation and motion were taken as axioms of a mathematical system capable of predicting the behavior of the physical universe. Vitalists were quick to point out, however, that this rigid deterministic view must either deny the reality of biological novelty or else claim that it can be derived from mechanical laws. They proposed instead that the principles of life were not reducible to physical interactions. The Second law of Thermodynamics was taken as eivdence for the vitalist interpretation. While an increase in entropy should occur with time, evolution has led instead to the appearance of more ordered, better organized systems. The solution to this apparent paradox is that the Second Law refers to closed systems whereas the evolution of organic matter is an open system dependent upon external energy sources. Today we understand that while the processes of life are compatible with the laws of physics they cannot be derived from them. Boundary conditions which direct biological activity (such as the sequence of bases in a segment of DNA) harness the laws of inanimate nature but are themselves irreducible to those laws. The discipline of biology, therefore, is greatly concerned with the elucidation of these boundary conditions, these controlling principles of

life.

2The historical and philosophical views expressed in this introduction have been more elaborately expressed by Polyani (1968) and others (Mercer, 1981).

CONTRASUPPRESSION

279

In the study of the immune system the search for boundary conditions has produced an emphasis on the regulation of immune responses. An early attempt to express a principle of iminunoregulation was the “Second Law of Thymodynamics” (Gershon, 1974a) which proposed that for any positive (helper) effect there was a corresponding negative (suppressive) influence. Both positive and negative regulatory influences come under the domain of T cell function (hence “Thymodynamics”), the suppressor T cells opposing the helper T cells. More recently another aspect of this “law” has been described. That is, many suppressive influences are associated with corresponding antagonistic effects which can serve to balance suppression. These “contrasuppressive” influences are yet another T cell function. We can now construct a slightly different picture of immunoregulation which still contains the original notion of symmetrical influences. Two types of regulatory T cells act upon the central helper T cell [the “conductor of the immunological orchestra” (Green and Gershon, 1981)].One acts to down-regulate the helper cell’s activity while the other can block this down-regulation. The latter is referred to as a “contrasuppressor cell.” It is the opposing dynamic interaction between this cell (or group of cells) and the suppressor cell that forms one basis of immune homeostasis. One could propose an analogy in another biological system. Motoneurons in the ventral horn of the spinal column are regulated by a number of signals. An impulse set up in an input axon leading to an excitatory presynaptic terminal (PT) will elicit an evoked potential in the motoneuron. In our analogy the PT represents the helper T cell. Inhibitory interneurons terminating at the PT can depolarize the PT without effecting the motoneuron directly. The result is inhibition of the excitatory signal, analogous to suppression of the helper T cell. This negative activity is, in turn, antagonized by inhibitory connections to the inhibitory interneuron itself, a connection which can provide disinhibition of the inhibitory input from the interneuron to the motoneuron (Shepherd, 1979). In principle then, disinhibition and contrasuppression are analogous activities. The modes of action, however, are dfierent. Whereas disinhibition operates by down-regulation of an inhibitory activity, contrasuppression works to block the negative signal at its target (see Section V). Nevertheless, the regulatory symmetry found in both systems argues in favor of a unifying biological principle of regulation. We do not propose to define such a unifying principle (which would be a true biological boundary condition) but will instead set our sights at describing one small aspect: the role of contrasuppressor T cells in the regulation of the immune response. We begin our discussion by considering the analysis of immune regulatory elements in general.

280

DOUGLAS H . G R E E N A N D HICHARD K. GEHSHON

II. Defining Cell Circuits

At the present time there is a great deal of information on the nature of the cellular and molecular elements that compose suppressor cell circuits. However, before this wealth of information was accumulated over the last decade or so, we were naively unaware of the complexity of these suppressor cell circuits. At the present time we have achieved the state where we now are clearly and definitely aware of our ignorance. We are, however, as naive of how contrasuppressor cells a i d their molecular mediators work as we were in our studies of suppressor cells in the early 1970s, but we are at a great advantage in dissecting contrasuppressor cell circuits because of the advances made in the past decade in the technologies needed for this type of dissection.

A. TECHNIQUES The major technological breakthrough that allowed us to dissect and understand suppressor cell circuits steins from the pioneering work of Cantor and Boyse (1975). These workers showed that the genetic‘prograin that is expressed in fully differentiated cells combines information for function and cell surface phenotype. Thus when a cell’s “luxury gene”3 is activated, there is a concomitant change in glycoproteins or other molecules that are expressed on the cell’s surface. Often these cell surface markers have allelic 3111 rnulticellular organisms, cells ditferentiate to the point where their activity extends beyond making materials that are mecessary for the cells own survival. They also make materials that help other cells within the organism to survive. Those genes that code for products which a cell makes to help the survival of other cells that compose the multicellular organisms are called luxury genes. This “altruistic” activity of cells is not really confined to multicellular organisms. Some bacteria make toxins that help them colonize certain microenvironnients. The cells that made the toxins (which could be considered luxury gene products) die during their production of the toxin and thus sacrifice themselves for the greater good of their genetically identical brethren. A classical luxury gene product of multicekdar organisms would be hemoglobin. The best studied luxury gene product of the immune system is immunoglobin. The Cantor-Boyse notion would predict that when the precursor of an antibody forming cell is induced to activate its genetic program for making antibody a change in the cell surface phenotype would also occur, which could be used to separate the differentiated daughter cell from its precursor parental cell. This prediction of the Cantor-Boyse hypothesis has been verified (Moller, 1983). The demonstration by Cantor and Boyse that the “Ly” alloantigens marked cells with distinct functions within the T cell system (Cantor and Boyse, 1975; Huber et al., 1976) gave investigators a new potent tool for removing several levels of ignorance. Subsequent production of other allo-antisera, as well as monoclonal antibodies, has allowed a further dissection of the imniunoregulatory apparatus. W e expect that the use of these techniques in the near fiiture, in combination with modern cloning techniques a i d production of T cell hybrids, will allow the generation of considerably more information in the next decade.

CONTRASUPPRESSION

28 1

variants and, by using genetically appropriate mouse strains, antisera that recognize the allelic variants can be made and used to isolate cells that express unique cell surface combinations of these alloantigens. When this is done one finds that the isolated cells have a single, or several closely related, luxury gene function(s).

B. CELLIJLARMAKE-UP Relevant to the discussion at hand is the composition of cellular circuits that seem to be specifically involved in activating effector cells with defined functions. Regardless of the programmed function of any immunoregulatory circuit so far elucidated, the basic structure seems to be consistent. In each case there is an inducer population that produces a signal which is received by a “transducer cell”4 that then helps to activate the circuit’s effector cells. At least some of the cells in the inducer and transducer populations can be separated from other cells with inducer or transducer function by virtue of unique cell surface antigenic phenotypes. In the murine system it would appear that inducer cell activity is invested in the 30%of T cells that express the Ly-1 alloantigen and fail to express the Ly-2 alloantigen (Ly-1 cells). Cells with transducer function express both the Ly-1 and the Ly-2 alloantigens (Ly-1,2 cells). Some of them also express other differentiation antigens that indicate that at least some subsets of transducer cells have been programmed during their differentiation in terms of which “luxury” gene they will eventually be induced to express. However, most Ly-1,2 cells have not yet been shown to bear cell surface determinants that indicate whether their final luxury gene program has already been activated. Perhaps these cells may act as precursor cells for any of the more differentiated cells. Because they have not yet expressed their final gene program and have the potential to differentiate into cells with opposing functions depending on the nature of the inducing signals they receive, they have been referred to as “hermaphrocytes” (Gershon et al., 1976). Most, if not all immunoregulatory circuits described have been found to be composed of these cellular elements: an inducer, a transducer, and an effector cell. 4The term “transducer cell” was coined by C. A. Janeway, Jr. It is an excellent name because it implies that this cell or group of cells is involved in conveying the inducer cell’s message to an effector cell. Two mechanisms of transduction have been proposed. Transducer cells can receive signals from inducer cells and differentiate into effector cells (McDougal et al., 1980). In other cases, tranducer cells make products which amplify the direct communication between the inducer and the effector cells (Tada and Okumiira, 1979).

282

DOUGLAS R . GREEN A N D RICHARD K. GERSHON

III. Defining Contrasuppression

The idea of contrasuppresion is not new. In fact its existence was postulated shortly after the discovery that some T cells play an active role in suppressing immune responses (Gershon, 1974b). It was initially noted that the induction of “infectious immunological tolerance” (which in fact was simply the deliverance of a suppressor signal from one T cell to another) could be blocked if the target cells (the ones which were to become suppressed) were stimulated with a crossreacting antigen. Stimulation with the crossreacting antigen produced no measurable helper effect, but could completely abrogate the acceptance of the suppressor cell signal. Thus, cells responsible for the phenomenon of contrasuppression were assigned the following attributes: (1) they deliver no measurable helper effect in the absence of suppressor cell signals; and (2) the amount of “help” that they can give is only equal to that amount that the suppressor cells have removed. Thus, a contrasuppressor cell is in effect an antisuppressor cell, which, with the help of hindsight, should have been theoretically predicted to be needed as a homeostatic balance for suppression. It is likely that there is more than one immunoregulatory circuit which is involved in balancing the activity of suppressor cells. However, at this stage only one has been worked out in detail. In this system the contrasuppressor cells not only perform the two requisite functions listed above, but in addition have the ability to confer an increased resistance to suppression upon helper T cells (Green et aZ., 1981a; Green and Gershon, 1982). The duration of this resistance to suppression is not known.

IV. Defining a Specific Contrasuppressor Circuit and Assigning a Unique Phenotype to Its Cellular and Molecular Members

A. THE INDUCER CELL The nature of the initiator cell that interacts with the antigen presenting cell to start the contrasuppressor cell circuit going has not yet been defined. However, it is known that the initiation event leads to the activation of an IJ + , Ly-1-,2+ (Ly-2) cell (Gershon et d.,1981) that, together with an appropriate I-J+ transducer cell activates I-J +, Ly-l+ ,2- (Ly-1) effector cells (Gershon et al., 1981; Yamauchi et al., 1981). Calling this I-J+, Ly-2 cell an inducer cell violates the rules stated above, that induction or initiation of immune responses is confined to cells that express the Ly-l phenotype. Either this system is an exception to the rule or, more likely, there is a Ly-1 inducer or initiator cell that we have not yet uncovered and the I-

CONTRASUPPRESSION

J

283

, Ly-2 cell, which we are presently calling an inducer cell may be part of a larger circuit. It is important to note that we have only found this cell by examination of previously activated systems. Although we will continue to refer to this cell as an inducer cell (because it does express an inducing function), we will do so tentatively, because of the generalization about Ly-1 cells made above. What is known about the contrasuppressor inducer cell besides its cell surface pehnotype? (1)It makes an I-J+ product as well as an antigen binding product (Yamauchi et al., 1981). It is not known whether these two attributes (expression of I-J and antigen binding) are invested in a single molecule. It is likely, if one can generalize from what has been learned about other I-J+ regulatory products, that the two attributes are carried by two separate polypeptides (Yamauchi et al., 1982a; Flood et al., 1982). (2) The antigenic specificity of the antigen binding moiety can be shown, by absorption studies, to be significantly more crossreactive than are either suppressor inducer or suppressor effector molecules. This particular point is extremely important because (a) at least one cell in the circuit would have been predicted by the early studies (Gershon, 1974b) to be crossreactive; (b) the crossreactive nature of this cell can help explain some otherwise poorly understood immunologic phenomena (see Section V1,G); and (c) it gives the experimenter the ability to separate suppressor and contrasuppressor factors made by Ly-2 cells by absorbing the contrasuppressor material on crossreactive antigens (Yamauchi et al., 1981). +

B. THE TRANDUCER CELL The I-J+ contrasuppressor inducer factor made by the I-J+, Ly-2 cell has, up to now, only been shown to work when there is a functional I-J+, Ly-1,2 contrasuppressor transducer cell present (Yamauchi et al., 1981). It is not known whether there is a direct signal from the Ly-2 inducer cell to the Ly-1 effector cell. The I-J determinant the transducer cell bears is serologically related (using monoclonal antibodies) to the I-J determinants found on the Ly-2 inducer cell and its cell-free product. This I-J determinant is serologically distinguishable from the I-J determinants found on cells of the suppressor circuit (Yamauchi et al., 1982b).

C. THE EFFECTOR CELL The effector cell of the contrasuppressor circuit we have defined expresses the Ly-1+, 2 - , I-J phenotype (Green et al., 1981a). Its I-J marker reacts with the same monoclonal antibodies as do the I-J markers of the other cells and their products that compose the contrasuppressor circuit (Yamauchi et +

284

DOUGLAS R. G R E E N A N D RICHARD K . GERSHON

al., 1982b). This effector cell also makes an I-J cell-free product which can +

replace the cellular biologically activity (Green, 1981). It is not yet established whether this factor has an antigen binding moiety, but from other studies (see Section VI1,A) the cell that makes the effector product can have antigen specificity. This I-J , Ly-1 contrasuppressor effector cell and its product are unable, as far as we can tell, to deliver a helper signal to cells that are depleted of T cells or to immune responses being induced by Ly-1 cells in the absence of Ly-2 cells. However, when Ly-2 suppressor cells are added to the system, the contrasuppressor effector cell and molecule(s) can negate the added suppressor signal. When it does this the response does not exceed that which occurs in the absence of suppressor cells in the system. Thus, this effector cell qualifies as a contrasuppressor cell by the criteria listed above. Another characteristic which quite convincingly distinguishes contrasuppressor effector cells from helper cells is that the murine contrasuppressor cell binds to the Vicia uillosa lectin and can be eluted from it with specific sugars (Green et al., 1981a). Helper cells do not bind in any significant numbers to this lectin. This differential activity, i.e., binding to the V. uillosa lectin, may be a general characteristic of contrasuppressor systems of all mammals, including humans. If so, it could help lead to the isolation and definition of these important regulatory cells in human disease. +

D. CELLULAR INTERACTIONS I N THE ACTION O F A N ANTIGEN-SPECIFIC CONTRAS u PPRE SSOR FACTOR I N C E LL-ME DIATE D IMMUNITY Ptak et al. (1984a)have described a contrasuppressor factor which specifically binds antigen and functions to render immune T cells resistant to the suppressive effects of tolerized recipients or T suppressor factors (TsF). The factor has no effect on cells previously exposed to T suppressor factor. Like the TsF in this system, the T contrasuppressor factor reacts with antigenprimed, I-J+ , Ly-2 T cells (Ptak et al., 1984b). This Ly-2 T cell population must be primed, but need not be primed to the same antigen for which the factor is specific. An interaction of the contrasuppressor factor with its specific antigen (conjugated to naive spleen cells) induces nonspecifically primed 1-J+, Ly-2 T cells to produce a factor with contrasuppressive effects on immune Ly-1 cells used for adoptive transfer. The latter factor appears antigen nonspecific, although the possibility that antigen specificity may be derived from cells in the target population is an open possibility. Nevertheless, the cellular interactions for the function of the T contrasuppressor factor directly parallel interactions described for TsF in this system. Research is in progress to relate this circuit to the contrasuppressor circuit described above.

CONTHASUPPHESSION

285

V. Functional Activity of the Contrasuppressor Circuit Given its characteristic behavior there are two major sites at which the contrasuppressor effector cell could work. It could inhibit suppressor cells or it could make the helper cells resistant to the signals delivered by the suppressor cell. There is no evidence supporting the former mechanism [and it is, in fact, contra-indicated (Green, 198l)I. We have data,’however, substantiating the latter mechanism (Green and Gershon, 1982; Green, 1981, 1984). A double chambered, periscopic Marbrook vessel was constructed. Primed Ly-1 cells were added to chamber 1 (lower) and either nothing, Ly-2 suppressor cells, Ly-1 contrasuppressor cells, or a mixture of suppressor and contrasuppressor cells were added to the second (upper) chamber. Chambers 1 and 2 were separated by a membrane with a 0.1 Frn pore size. The apparatus was cultured in appropriate medium for 48 hr, after which time the Ly-1 cells in chamber 1 were retrieved and experiments were done to determine if any of the cells in chamber 2 had influenced their subsequent behavior. It was found that the inclusion of suppressor cells in chamber 2 decreased the helper activity of the Ly-1 cells in chamber 1. This reduction of helper activity was eliminated when the contrasuppressor cells were included with the suppressor cells in chamber 2. The inclusion of the contrasuppressor cells did something other than simply inhibit suppressor cells. This was determined by asking whether the helper cells recovered from the apparatus that had suppressor and contrasuppressor cells in chamber 2 expressed a differential sensitivity to subsequent suppressor signals than did the Ly-1 cells cultured in chamber 1 when either nothing, or contrasuppressor cells only, were present in chamber 2. In the latter two instances, the retrieved Ly-1 helper cells were as sensitive as freshly isolated Ly-1 helper cells (which had not been in the chamber at all) to suppressor signals. However, the Ly-1 cells from chamber 1 of the apparatus that had both suppressor and contrasuppressor cells in chamber 2 were resistant to suppressor signals given to them subsequently. This resistance was relative in that when we added more suppressor cells we achieved suppression. From these results one can draw several conclusions.

1. Contrasuppressor cells do nothing to the helper cells in the absence of suppression. This leads to conclusion 2. 2. The signal needed to cause the helper cells to develop resistance to suppression requires an interaction between suppressor and contrasuppressor cells. We think the likely explanation for this finding is that the activity of the contrasuppressor cell is not autonomous. Like other immunological effector cells, it needs an inducing signal to function and, in this

286

DOUGLAS R. GREEN A N D R I C H A R D K . GERSHON

case, the inducing signal came from the suppressor cell population. Alternatively, since it has been shown that some molecular complexes which effect immunoregulation are composed of molecules made by two different cells expressing different Ly phenotypes (Ptak et al., 1982) it is possible that the contrasuppression was effected by such a molecular complex. 3. Contrasuppressor cells do not act simply by shutting off the suppressor cells, since suppressor cell activity was required to increase the resistance of the helper cells to fresh suppressor cell signals. 4. The contrasuppressor effector cell has only a limited time frame in which it can work. Once the suppressor cell signal becomes manifest in its target cell, the contrasuppressor effector cell does not have the ability to reverse that signal. This was observed when contrasuppressor cells were added to helper cells which had already been suppressed. Under these conditions no relief of the suppressed state could be obtained (unpublished observations). In summary we can define that part of the contrasuppressor circuit so far uncovered as being composed of an I-J , Ly-2 inducer cell which makes an I-J antigen-binding “crossreactive” molecule(s), acting through an I-J transducer cell to produce an 1-J+, Ly-1 effector cell which also makes an IJ + effector molecule. All of the I-J+ cells and molecules within the circuit react with a monoclonal anti-I-J antibody that does not react with several other types of I-J+ cells (Yainauchi et al., 1982b). The ultimate activity of the effector cell in this circuit is to (1)negate the signals given by suppressor cells and, in fact, (2) turn them into signals which make the Ly-1 helper target cell resistant to further suppressor cell signals even when the contrasuppressor effector cell has been removed from the system. The duration of this period of suppressor cell resistance is, at present, unknown. +

+

+

VI. Conditions That Influence the Generation and/or Activation of Contrasuppressor Cells

A. ONTOGENY It has been known for many years that if one places adult inurine spleen cells in culture under conditions where they will thrive but adds no extra inducing signals, the cultured cells will promiscuously develop into potent suppressor cells (Janeway et al., 1975; Hodes and Hathcock, 1976). We noted that if neonatal cells (from mice 5-12 days of age) were used instead of adult cells, in the same type of cultures, promiscuous contrasuppression was generated (Green et a l . , 1981a). However, neonatal animals have been

CONTRASUPPRESSION

287

shown to possess a fully active suppressor cell circuit (Murgita et al., 1978; Mosier and Johnson, 1975; Ptak et al., 1979). Why don’t adult spleen cells generate contrasuppressor cells in culture to overcome promiscuous suppression? By mixing cells of neonates and adults prior to culture it was found that adult spleen contain T cells capable of regulating the generation of contrasuppressor cells. The time in ontogeny optimal for the development of contrasuppressor cells precedes the appearance of the cells which regulate the generation of contrasuppressive activity. We have recently found a monoclonal antibody which can identify the cells which down-regulate contrasuppressor cells (Green et al., 1983). Preliminary evidence suggests that these cells start to emerge from the thymus around day 8 postbirth reaching significant numbers after 14 days. Appropriate treatment of adult T cells with this monoclonal reagent, prior to putting the adult spleen cells into culture, results in the development of promiscuous contrasuppressor activity similar to that found when neonatal cells are put into culture. Thus, there appears to be a period in the development of the organism during which contrasuppressor activity may be the dominant immunoregulatory force. One would suspect that there is an earlier period, before contrasuppressor cells emerge, when suppression is the dominant regulatory activity. Unfortunately, since spleen cells from mice less than 5 days old do not thrive well in culture, this type of analysis (generation of promiscuous immunoregulatory activity) could not be used to determine whether contrasuppressor cells emerged from the thymus at a defined time after suppressor cells did. Ptak and his colleagues have put forth evidence that indicates that there may be a major emergence of contrasuppressor cells from the thymus on approximately day 2-3 after birth in the CBA mouse strain (ShowronCendrzak et al., 1983). While spleen cells taken from mice before day 3 after birth were very potent suppressors of a graft-versus-host (GVH) response, thymocytes from 2-day-old mice would interfere with suppression produced by the newborn spleen cells. Spleen cells from 3 day or older mice had no ability to suppress the GVH activity. However, if the mice were thymectomized before day 3, the splenic suppressor activity persisted for many weeks. These results indicate that contrasuppressor cells start to emerge in significant numbers from the thymus between days 2 and 3 in ontogeny. Ptak and his colleagues have been able to determine that the day 3 spleen cell population which has lost its suppressive activity could be made to act like the suppressive day 2 spleen cell population by treatment with an anti-1J reagent to remove contrasuppression. The notion that contrasuppressor cells appear early followed by the appearance of cells that regulate their activity is supported by the experimental findings of Smith and colleagues (1982, 1983a,b). They found that adult mice which were thymectomized 3 days after birth were susceptible to induction

288

DOUGLAS R. GREEN A N D RICHARD K . GERSHON

of autoimmune disease by administration of polyclonal B cell activators. If thymectorny was done before day 3 or not done at all, autoimmune disease did not develop. Thus, one could speculate that significant numbers of contrasuppressor cells had emerged from the thyinus by day 3 and that the cells which regulate the contrasuppressor cells had not yet emerged. This notion was tested by asking whether different levels of contrasuppressor activity could be found in the autoimmune mice as compared to the various controls. In most of the cases, cultured spleen cells from the experimentally induced autoimmune mice developed promiscuous contrasuppressor activity (like neonatal animals), while none of the control mice exhibited this effect. Thus the observations of Smith et al. fit with those of Ptak and colleagues. The available evidence indicates that three types of regulatory cells emerge at different times during ontogeny. Initially, the predominant cell is a suppressor cell, most likely one that has been described (Murgita et al., 1978) and closely resembles cells in the “feedback suppressor circuit. ” There is a subsequent wave of contrasuppressor cells, which emerge several days after birth. Their emergence is followed by a wave of cells which we shall refer to as “level 2” suppressors, which have a specialized ability to regulate contrasuppression (Green et al., 1983). This scenario, while somewhat hypothetical at this point, is attractive because it offers an explanation for the high incidence of transcient allergy in juveniles (Pearlman, 1973), as well as for their well recognized reduced morbidity to infectious diseases. Thus, our findings would predict that the immunoregulatory balance should be skewed toward contrasuppression before level 2 suppressors appear and at this time in development there should be a high level of immune resistance which carries with it, unfortunately, a high risk of developing allergies or perhaps even autoimmune phenomena. A possible selectional advantage would accrue if the contrasuppressor system could be activated by antigen-antibody complexes that inhibit helper cell activity. If so, then the infant could gain iminunological memory in its contrasuppressor cell pool while still protected from pathogens by maternal antibodies. The juvenile learns to respond better to pathogens in adult life if his immune system has learned of their existence while under the protection of maternal antibodies (Goidl and Siskind, 1974). This possible role of the contrasuppressor system is supported by some phenomenological observations. Antibody-antigen complexes in antibody excess completely abrogate the ability of the recipient mouse to make an antibody response. However, when that transiently unresponsive mouse is tested, after the passive antibody has all been eliminated, its immunologic memory indicates that it had indeed reacted to the antigen (Tite and Playfair, 1978). This observation leads us into a disucssion of the major variables that can determine whether contrasuppressor cells will be activated.

CONT RAS UP P RE S S ION

B. THE ROLE OF

THE

289

ANTICEN PRESENTING CELL

1 . I g - Antigen Presenting Cells There are several types of cells with a denclritic morphology which appear to be ideally suited for presenting antigen to the immune system. One of these types of cells is found in the skin and is referred to as a Langerhans cell. Another is a dendritic cell found in the spleen by Steininan and coworkers (Nussenzweig et a l . , 1981). Both of these cells express very high levels of I-A controlled material and, in addition, the splenic dendritic cell also expresses high levels of H-2D. Whether Langerhans cells also express high levels of Class 1 antigens is not presently known. Both these types of cells have the capacity of producing what could be called “suppressor cell resistant immunity” (Ptak et a l . , 1980; Britz et al., 1982). Splenic macrophages and peritoneal exudate cells, although also I-A+, do not have the same capacity. Mice that are innoculated intravenously with hapten-labeled peritoneal exudate cells or splenic macrophages develop a profound state of immunological unresponsiveness, as measured in contact sensitivity responses to the specific hapten. These cellular innocula have also been shown to be highly efficient at activating suppressor cells (Claman et a l . , 1980). Further, it can be shown that these cellular innocuola have the capacity to induce immunity, as measured by contact sensitivity, if the suppressor system of the recipient is temporarily inactivated by one of several mechanisms (Greene and Benacerraf, 1980). However, hapten-labeled Langerhans cells or dendritic cells do not require the inactivation of the suppressor circuit to be immunogenic. Their immunogenicity in mice that have intact suppressor systems is not due to a failure to activate suppressor cells. This can be shown by demonstrating that immunity will ensue even when a highly tolerogenic compound such as trinitrobenzene-sulfonic acid (TNBSA) is innoculated intravenously, and concomitantly, with TNP-labeled dendritic cells. It may be that these dendritic cells activate iininunoregulatory cells with contrasuppressive acitivity. This interpretation is supported by some other studies of Ptak et al. (1981) that showed that iv innoculation of the ordinarily tolerogenic hapten-labeled peritoneal exudate cells, along with a small inoculuin of promiscuous contrasuppressor cells (generated from neonatal mice as described in Section VI,A), not only failed to induce a state of immunological tolerance, but rather produced an immune state which was quantitatively similar to that found in optimally immunized animals. Thus, activation of cells with contrasuppressor activity can change a tolerogen into an immunogen. This is a very potent effect and argues for the benefit of having contrasuppressor cells activated in order to get good immune responses. These results can also explain an interesting phenomenon described by

290

DOUGLAS I<. CH EEN A N D RICHARD K . CJEHSHON

Macher and Chase (1969). They showed that if an area of the skin which had been painted with a sensitizing reagent was removed a short time after painting, tolerance to the antigen was generated. If a significantly longer period of time took place before the skin site was excised, immunization occurred. The sensitizing agent therefore had two effects: a tolerogenic one and an immunogenic one (like injecting mice with both TNP-labeled Langerhans cells and TNBSA). When the skin is painted with a sensitizing agent a significant amount of material probably binds to a variety of blood-bourne cells and molecules circulating to the spleen where it activates suppressor cells. However, Langerhans cells are also haptenated and if allowed to reach the draining nodes activate “suppressor resistant immunity.” In essence the Macher-Chase experiments were quite similar to the ones of Ptak et al. (1981) described above, with a simple refinement made by Ptak; instead of painting the skin he attached the hapten to the purified Langerhans cells. In another system Braley-Mullen (1980) showed that when antigen (SIII penumococcal polysaccharide) coupled spleen cells containing both splenic macrophages (tolerogenic signals) and dendritic cells (immunogenic signals) were injected intravenously suppressor cells were generated. Their activity could only be demonstrated, however, if an Ly-1 cell which “neutralized” suppression was removed. Similarly, treatment of recipients with a low dose of cyclophosphamide also served to reveal suppressor cells. The injection of antigen-coupled spleen cells into untreated animals induced the generation of Vicia villosa lectin adherent, I-J , Ly-1 T cells capable of blocking suppressor cell activity (Braley-Mullen, 1983). Such observations have important practical implications for vaccine research as well as for further understanding the induction of immunoregulatory subsets. These series of experiments support the idea that the development of immunity is a question of balance between suppression and contrasuppression with the helper cell as the central focus of the opposing regulatory forces. +

2 . The Role of lmmunoglobin during the Initial Antigen Presenting Event Since we have already built a hypothesis on the basis of a postulate that antigen-antibody complexes preferentially activate contrasuppressor cells, it behooves us to see if there is any more direct evidence than the phenomenology described in Section V1,A. Good evidence that immunoglobin is somehow involved in activating cells of the contrasuppressor circuit stems from studies done in conjunction with C. A. Janeway, Jr. (Janeway et al., 1981; and unpublished observations). Those studies have shown that mice, which have their immunoglobin system suppressed from birth by treatment with an anti-p chain antiserum, fail to develop promiscuous contrasuppressor cells at any of the time periods so far

CONTRASUPPRESSION

29 1

tested. Thus they are extraordinarily different from the spleen cells of normal mice in this regard. These findings are reminiscent of some experiments done in a collaboration between our laboratory and that of M . Cooper. It was found that a group of anti-p suppressed mice failed to develop any delayed type hypersensitivity response to SRBC independent of the dose of antigen that was used for immunization (Burrows et al., 1976). Control mice had a normal dose optima (see Section VI, D for further discussion). However, when both groups of mice were pretreated with low doses of cyclophosphamide they made equivalent delayed type hypersensitivity responses. These results suggested that anti-p suppressed B cell-deficient mice might have excessively high levels of suppressor cells, which was subsequently disproved by Janeway et al. (1981). Now, in conjunction with the latest findings, it seems likely that the reason the anti-p suppressed mice could not develop delayed type hypersensitivity in the absence of treatment with some agent that removes suppressor cells was because they had not developed adequate contrasuppressor cells. Assuming that anti-p chain serum had no anti-T cell activity in it (none could be shown) these findings strongly suggest an inducing role for B cells in contrasuppressor cell activation. L’age-Stehr and colleagues (1980) have shown that intravenous injection of syngeneic or even autochthonous purified B cell blasts into mice can lead to a transient GVH effect which is mediated by Ly-1 cells. Removal of immunoglobin from the surface of the B cell blasts prevents the induction of this transcient autoimmunity. The exact nature of the cell responsible for the induction of autoimmunity in these animals was not determined, but with all of the other types of evidence presently at hand, one might postulate that the B cell blasts were activating contrasuppressor cells. This interpretation would fit in well with the work of Smith et al. (1982, 1983a) discussed above (Section VI,A) that show that polyclonal B cell activators worked through the action of contrasuppressor cells in leading to autoimmunity.

3. The Role of Presentation of Antigen on Mouse Immunoglobulin As discussed, antigen-antibody complexes are capable of inducing both suppression (Rao et al., 1980) and immunologic memory (Tite and Playfair, 1978), perhaps related to activation of contrasuppression. Hamaoka et al. (1977) utilized an immunization protocol involving antigen conjugated to mouse immunoglobulin to induce immunity in the (apparent)absence of any suppressor cell activity. Animals so immunized became resistant to stimuli which would otherwise induce suppression, that is, they rejected and became immune to tumor cells coupled with the specific antigen (discussed in Section IX). Based on these observations, Ptak et al. (1983a,b) immunized

292

DOUGLAS R . GREEN A N D RICHARD K . GERSHON

mice with TNP-anti-TNP conjugate complexes. Ly-1 T cells from the spleens of these mice were found to secrete a potent TNP-specific contrasuppressor factor. The interaction of this factor with immune cells to interfere with suppressive influences is discussed above (Section IV, D). 4. Blocking of Gruft-versus-Host-lnduced Suppression by T Cells That

Have Been Exposed to Antigen-Antibody Complexes The observations made on immunization by antigen-antibody complexes are akin to another series of experiments involving innoculation with antiH-2 antibodies. Hurtenbach et al. (1981) have induced suppressor cells that can block the in vitro generation of allogeneic killer cells. They do this by subjecting F, animals to nonlethal graft-versus-host (GVH) reactions and find the suppressor cells in the spleens of the mice undergoing the GVH reaction. They have also found that if they inject antibody against H-2 determinants in the F, mouse they not only fail to find the GVH induced suppressor cells, but they find T cells that can transfer the resistance to GVH induced suppression to otherwise normal animals. Thus, the antigen-antibody complexes in the F, animals lead to the production of a T cell that inhibits or acts as a courterforce to GVH-induced suppression. The phenotype of the responsible cell is presently being tested to see if it bears the markers of cells in the contrasuppressor circuit. We have given emphasis to this experiment by Hurtenbach and colleagues because they have shown that antigen-antibody complexes activate a T cell that can actively block suppression. However, it has been noted many times before that antigen-antibody complexes, if of the correct isotype and given in the right molar ratios, are augmentative of immune responses (Heyman et al., 1982). There is no evidence in these other systems that the augmentative effect is mediated by suppressor inhibition. However, if the Hurtenbach finding is an example of a general phenomenon, we suspect that it is likely that contrasuppression may indeed play an important role in other types of antigen-antibody complex mediated immunoaugmentation. It is also well known that many antigen-antibody complexes also activate suppressor cells (Rao et al., 1980). As discussed above and below, there seems to be very few conditions that lead solely to the activation of contrasuppressor cells and not suppressor cells. The factors that activate one system tend to activate the other. Which effect will dominate seems to depend on factors such as antigen dose, the nature of the antibody in the antigen-antibody complexes, the site of inoculation of antigen, etc. One most important factor in the activation of contrasuppression is imprinting, i.e., having the antigen presented on specialized antigen presenting cells (as discussed in Section VI,B,l above).

CONTRASUPPRESSION

293

5 . The Potential Action of Contrusuppressor Cells in Either lnhibiting o r Breaking Allotype-Specific Suppression

F, mice which are neonatally exposed to anti-allotype antibodies become temporarily suppressed, in that the antibody they produce does not bear the allotype to which the anti-allotype serum was directed (Herzenberg et a l . , 1976). This suppressive effect of the anti-allotype serum usually wanes quite early. However, in SJL x BALB/c F, mice half of the offspring develop a state of chronic suppression such that the allotype inay never be expressed. This chronic suppression of allotype can be transferred to normal mice by suppressor T cells. Occasionally, one of the chronically suppressed inice can “escape” suppression. When such an animal begins to produce very small amounts of an antibody that expresses the previously suppressed allotypic marker (as little as l%),one finds that their T cells develop the ability to block the transfer of allotypic specific suppression (L. Herzenberg, personal communication). Thus, this is another instance where a cell with contrasuppressor-like activities can be shown to be activated in association with the recognition of markers on immunoglobulin molecules; small amounts of antibody bearing the suppressed allotype inay be responsible for the contrasuppressive effects. However, it is possible that the contrasuppressor cells are responsible for the “antibody leak. ” A simple experiment to decide between these alternatives is to inject the chronically allotype suppressed mice with small amounts of antibody expressing the suppressed allotype and see if it is the antibody that directly activates the contrasuppressor cell.

c. SPEC I A L ANTIGENIC: DETERMINANTS 1 . Hen E g g Lysozyme The notion that T cells with different functions see different epitopes on complex molecules (P-galactosidase) was first proposed by Sercarz and coworkers. Sercarz has also obtained evidence that helper and suppressor cells inay see different epitopes on hen egg lysozyme (HEL), an antigen under l r gene control (Sercarz et d., 1978). The reason “nonresponder” mice fail to make an antibody response to the intact molecule is due to activation of their suppressor cells by a sinall fragment (N-C) of the molecule. A larger fragment of the molecule, lacking three amino-terminal residues (AP-HEL) fails to induce these suppressor cells and “nonresponder” mice can become “responders” if immunized with this molecular fragment (Wicker et a l . , 1982). Thus, Sercarz et al. have a molecule with separate determinants, one of which induces help and another that induces suppression. In collaboration with Sercarz we proposed that genetically controlled unresponsiveness may

294

D O U G L A S R. G R E E N A N D R I C H A R D K . G E R S H O N

be due to the failure of nonresponder mice to make contrasuppressor cells to the “suppressor” determinant. W e hyperimmunized responder mice with the intact antigen or one of the two fragments described above. T cells from the intact HEL and the N-C hyperimmunized animals were capable of interferring with the suppression of H E L primed helper cells (assays were done using nonspecific suppressor cells and the double Marbrook intermediate culture described in Section V). However, T cells from mice immunized with the AP-HEL (this is the fragment that activates helper cells) had no ability to contrasuppress. These preliminary results performed by D. R. Green in the laboratory of Dr. E. Sercarz at UCLA suggest that the difference between responder and nonresponder mice has nothing to do with their ability to recognize “helper” determinants on the molecule. It is their differential sensitivity to the special epitope of the small N-C fragment. In nonresponder mice this determinant activates suppressor cells predominantly, while in responder mice it might activate contrasuppressor cells. Thus, there is evidence that suppressor and contrasuppressor cells are related to one another by features other than name. It appears that they both “see” the same specialized epitopes on molecules. Perhaps the way these epitopes are presented to the immune system determines whether suppression or contrasuppression will win out (see Section VI,B).

2 . GAT The conclusion given above might have been inferred from studies of nonresponder mice in the GAT (random copolymer of glutamic acid, alanine, and tyrosine) system where it was found that one simply had to inactivate suppressor cells to turn a nonresponder into a responder mouse (Kapp et al., 1975). This indicates that the helper system of the nonresponder mouse was perfectly normal, but that the force that opposes suppression was lacking. Interestingly, like the H E L system, the GAT molecule could be manipulated in a way that made it immunogenic in strains that were ordinarily nonresponders. This was accomplished by decreasing the number of tyrosine residues in the polymer (Schwartz et al., 1976). 3. Glycophorin The notion of epitopes with specificity for T cell subsets has been extended to studies using more conventional antigens like sheep red blood cells (SRBC). Fresno et al. (1982) have shown that a purified SRBC specific suppressor molecule recognizes only the glycophorin determinant on the SRBC, yet is capable of suppressing the response to the entire array of antigens presented on the SRBC. Iverson and colleagues (1983a) have isolated a potent suppressor factor that circulates in the serum of SRBC hyperimmune mice using an anti-idiotypic reagent. They have found that like the

CONTRASUPPRESSION

295

clonal factor of Fresno et al. the major portion of the circulating T suppressor material is glycophorin specific. Thus, there are two independent lines of evidence that suggest that glycophorin carries on it a suppressor cell specific (or preferential) epitope. In another series of experiments, Schiff (1980)found that the activation of suppressor cells by immunization with glycophorin is a dose-dependent phenomenon. If one immunizes cells in Mishell-Dutton cultures with a wide range of doses of glycophorin one can find a dose that activates cells which inhibit suppression but which do not deliver a conventional helper signal, and are likely to be contrasuppressor cells. These studies, taken together with those in the HEL system, indicate that suppressor and contrasuppressor cells may see the same, or closely related epitopes on complex antigens. Helper T cells see a different and perhaps more diverse array of epitopes.

4 . Rationale f o r Special Epitopes These findings raise a philosophical question. Why should suppressor and contrasuppressor see specialized epitopes on a given molecule? We offer as an explanation the notion that specialized self-markers have coevolved with the immune system and were selected because they help to prevent autoimmunity. This notion would predict that mutants having glycophorin-negative red blood cells should have a significant autoimmune hemolytic anemia in contrast to appropriate control populations. At the present time too few mutants have been identified to test this hypothesis. Why should a marker that serves as a “self’-recognition determinant in the autochthonous host also serve as a suppressor determinant on cells and molecule of a foreign host? We could postulate that determinants specific for suppressor cells are, like other self-marking molecules such as H-2, highly crossreactive. Therefore foreign variants of self-marking molecules (being crossreactive) would activate self-recognizing suppressor cells with a greater degree of intensity than would non-self-“crossreactive” determinants. We would also suggest that there may be some chemical relatedness of selfmarking epitopes found on different types of molecules and cells. Why should suppressor and contrasuppressor cells see the same epitope? This may have to do with parallel development of suppression and contrasuppression during evolution. That is, in the autochthonous host some molecules have been identified as self during ontogeny and are not reacted to. In foreign hosts the modifications of the molecule which make it foreign would also lead to an overreaction of the cells that respond to it. This view would predict that allogeneic effect factor might express significant amounts of contrasuppressive activity (see Addendum 2).

296

D O U G L A S R. GREEN A N D RICHARD K . GERSHON

D. ANTIGENDOSE It is clear in many systems that there is an optimal dose of antigen that produces a good immune response while doses of antigen on either side of the optima tend to be tolerogenic and/or lead to activation of suppressor cells. Several examples of this type of finding have already been discussed (Section VI,C). These findings could be interpreted to mean that contrasuppressor cells respond only in a narrow dose range (which may be optimal for getting the antigen onto the specialized antigen presenting cells discussed in Section VI,B), while suppressor and helper cells have no such stringent requirements. Evidence in support of this notion stems from studies that have used low doses of cyclophosphamide (Cy) to incapacitate suppressor cells before immunization with various doses of antigen. In such studies, Askenase and colleagues (1975) found that injecting mice intravenously with an optimal dose of SRBC led to the production of a strong DTH response in the immunized mice. Mice immunized with 1/10 or 10 times the optimal number of SRBC made rather poor responses. However, the Cy-pretreated mice made the same optimal or maximum response when immunized with all three doses of antigen. The question arose at that time as to why elimination of suppressor cells failed to result in an increased response at the optimal dose of antigen. The results imply that suppressor cells, or at least Cysensitive suppressor cells, were not activated at that antigen dose. It was hard to imagine a mechanism whereby suppressor cells would become activated by low doses of antigen, and then become inactivated by increasing the dose of antigen but become activated again by further increasing the dose of antigen. It now seems likely that the optimal antigen dose had activated contrasuppressor cells (see Section VI,B on DTH in B cell-deficient mice) so that physical removal of the suppressor cells had no effect because the host had, by virtue of activation of the contrasuppressor system, functionally eliminated them itself. Similar findings have been made with the glycophorin system described above (Section VI,C,3), that is, that suband supraoptiinal doses of glycophorin preferentially activate suppressor cells while at the optimal dose no suppression is seen. The question then arises: Is the contrasuppressor cell only activated in a narrow range of antigen dose for the reasons noted above, or is contrasuppression only efficient at overcoming the type of suppression one gets at low doses and less efficient at overcoming the suppression produced by high doses of antigen? This question brings 11s back to the notion of level 1 and level 2 suppressors as discussed (Sections VI,A and V1,F). Having defined level 2 suppressors with a monoclonal reagent (Green et al., 1983) (discussed in Section VI,A), we would favor the latter interpretation. This interpretation is also consistent with the ontogenic studies that indicate that “level 1”

suppressor cells appear first, contrasuppressor cells follow, and “level 2” suppressors follow later (Section V1,A). It may be that high doses of antigen are required to activate level 2 suppressors which are then capable of regu lat i 11g con t ras ti p p re s sion . There is another reason to favor this explanation which leads us into an area where further insights allout the role of contrasuppression in determining iininunc responsiveness can be gleaned.

E. GENETICS Tada and colleagues (Taniguchi et a l . , 1976) have reported that mice with B10 background genes will not respond to a particular suppressor factor. They have determined that the failure of the mice to respond is due to their lack of expression of receptors for the suppressor product. Yamauchi et al. (1981) have shown that mice with B10 background genes can be made to respond normally to a (probably different) suppressor factor if any of the components of the contrasuppressor circuit, discussed above (Section IV), are removed from the system. It is likely that the “Tada” suppressor molecule is one that overcomes contrasuppression under normal circumstances but that under the conditions of their studies the acceptor cells for that molecule were not present. The same type of conclusions can be drawn from studies on mice with A strain background genes. These inice share the same phenotypic defect as do mice with B10 background genes but the lesson in the A strains is that they fail to produce the “Tada” type of suppressor factor (Taniguchi et al., 1976). Using a standard immunizing regimen we have found that, like B10 mice, A strain mice can be made sensitive to suppression by removal of components of the contrasuppressor circuit (Green, 1984). Thus, in situations where the “Tada type” of suppressor factor or the cells which are receptive to it are limiting, contrasuppression is dominant. Suppression will only be seen when the contrasuppressor circuit is quenched.

F. “LEVEL2” SUPPRESSOK CELLS The results discussed above are analogous to the ones reported (Section V1,A) where it was shown that removal of “level 2” suppressors made the spleen cells of adults behave like neonatal cells in culture. The suppressor factor that Tada and colleagues were studying is significantly different from the ones we have been studying, in that ( I ) it is made by a cell with a different phenotype (I-J+ , Ly-2 cell), (2) it has an acceptor cell, a cell with a different phenotype (an Ly-1,2 cell), and (3) it shows different genetic restrictions. Analysis of the data froin Tada’s group might suggest that they are

298

D O U G L A S I<. GREEN A N D R I C H A R D K . G E R S H O N

dealing with a suppression amplijiyying system which could be akin to “level 2” suppressors, i.e., amplified suppressor cells. It is on these theoretical grounds that we postulate that the antigen dose effect described above is due to the activity of “level 2” suppressors at high doses of antigen and that these “level 2” suppressors can, under certain conditions, overcome the delivery of contrasuppressor signals. To be consistent with any notion of symmetry in biological systems we might postulate that there should exist such a thing as “level 2” contrasuppression. There is some very interesting preliminary evidence that there may indeed be two types of contrasuppressor cells; the crossreactive contrasuppressor cell that we have discussed and an antigen-specific one. These may turn out to be members of two different circuits rather than different cell types within a circuit (see Addendum 1).

G. CROSSREACTIVITY OF T H E Ly-2 CONTRASUPPRESSOR INDUCER CELL There is another theoretical point which derives from the observation that at least one cellular member of the contrasuppressor circuit(s) has a broader crossreactive specificity than do helpers and (at least) “level 1” suppressor cells. The notion is related to experimental observations made by a large number of workers, of which Weigle and colleagues (Weigle, 1967) have been in the forefront: the ability to break immunologic tolerance by immunization with a crossreactive antigen. To the best of our knowledge, this very important generalized finding has never been adequately explained. One hypothesis is that a second helper cell, capable of recognizing a determinant on the crossreactive antigen (not present on the tolerogen) stimulates B cells to make antibody to other determinants on either molecule. This explanation completely fails to explain how, once tolerance has been broken, the animal is fully capable of reacting to the original tolerogen in a highly immunogenic way (Weigle, 1967; Benjamin and Weigle, 1970). All other explanations offered depend upon a theoretical cell (either a T cell of a B cell) that makes a crossreactive product and fails to become tolerant and/or suppressed by determinants on the crossreactive antigen. There are major problems with this explanation. No reason has been put forward to explain why the crossreactive cell should have preferentially escaped the tolerance induction mechanism. If the cell is capable of making an antibody response (or other type of immune response) against the tolerizing antigen they why is it spared during tolerogenesis, remaining poised and ready to be easily stimulated with the crossreacting antigen? Arguments have been put forth that since there is the need for more than one cell to produce antibody one could postulate that the crossactivity was in one of the cell populations and not in the other and only one of the two reacting cells was spared during toler-

CONTRASUPPRESSION

299

ogenesis. This type of argument does nothing more than confound the problem. One could postulate an infinite number of cells required to induce a B cell to make antibody with only one of them being crossreactive. One would still have to explain why the crossreactive cell, which must see epitopes on the tolerogen, was preferentially protected from tolerance. In one system tolerance to SRBC was broken by the injection of horse red blood cells (HRBC) and a source of normal (nontolerant) T cells (Gershon and Kondo, 1972). The antibodies produced under these conditions were not crossreactive, i.e., some could be absorbed only with SRBC, others only with HRBC. Similar findings in a different system were made by Benjamin and Weigle (1970). A key point in the above experiment is the requirement for both crossreactive antigen (HRBC) and normal T cells for tolerance to break. HRBC injected alone into SRBC-tolerant animals produced a normal anti-HRBC response but no antibodies to SRBC. Injection of T cells alone had no effect on the tolerant state. We suggest the following interpretation. Mice tolerant to SRBC are known to possess potent suppressor T cells specific for the antigen (Gershon and Kondo, 1971). These suppressor T cells have been shown to inactivate helper T cells (Green et al., 1981b). Once inactivated, helper cell function cannot be restored by contrasuppression. When HRBC were injected, antigen crossreactive contrasuppressor inducer cells were activated. When normal, nonsuppressed T cells were also provided, these could act as targets for the induced contrasuppression, allowing dominant immunity over tolerance. Tolerance to protein antigens is slightly different. Demonstrable suppressor cells do not persist in tolerant animals and the breaking of tolerance does not require an added source of naive T cells. We propose that a high, tolerizing dose of antigen causes a deletion of contrasuppressor cells (perhaps by activating level 2 suppressor cells, see Section V1,D). The suppressor circuit is also activated and has the dual effect of turning off helper T cells and feeding back on the inducer cells of the suppressor circuit itself (Green et al., 1981b). The result, then, is a wave of suppression which leaves inactive helper cells and no demonstrable suppressor cells. After a period of time, nonsuppressed naive T cells mature. At this point, addition of antigen may induce help, but this is instead inactivated by a wave of suppression (in the absence of contrasuppression). Indeed, suppressor cells have been induced in tolerant animals by otherwise immunogenic doses of antigen (Parks, 1981). With time the ability to induce suppression decreased until the tolerant state waned. Many investigators have shown that a state of hyperimmune responsiveness can exist following tolerance. This might represent the renewed activity of a contrasuppressor population.

300

DOUGLAS R . GREEN A N D RICHARD K . GERSHON

If a crossreactive antigen is used instead of the original this might more rapidly activate the contrasuppressor circuit to protect the naive helper cells, as discussed. The fact that a significant period of time must elapse before tolerance can be broken in this way supports this notion (Gershon and Kondo, 1972). The interplay between memory and tolerance is poorly understood, reflecting our knowledge of the regulation of these states. Nevertheless, immune memory is generated at the same time as some states of induced unresponsiveness, such as that produced by immune complexes (Tite and Playfair, 1978) (see Section VI, B). Contrasuppression and its regulation may well be the key to such phenomena. H. ADJUVANTS It was recognized very early on that under most circumstances it is difficult to induce a potent immune response to many foreign proteins. Thus, a special mechanism was looked for (and found) to help augment immune responses. The substances that helped make good immune responses became known as adjuvants. The most potent of all of those discovered was the “complete adjuvant” described by Jules Freund, which made his name a household word in immunology. There are two components which are responsible for the potency Freund’s complete adjuvant (CFA) enjoys. One of these is the oil that is also present in incomplete Freund’s adjuvant and which probably acts as a depot substance. It entraps the antigen and allows small amounts of it to be released over a prolonged period of time, a very efficient way to stimulate the immune response. However, the inclusion of mycobacterium tuberculosis in the adjuvant makes it much inore potent. In this particular section we would like to focus our attention on the role of the mycobacterial component of the CFA as this substance produces a very characteristic response: the production of large granulomas which contain many activated macrophages. We know that macrophage-like cells5 have been implicated as important cells in antigen presentation. It has been shown (Section VI, B) that certain subsets of “macrophage-like cells” induce a suppression resistant immune response. Britz and colleagues (1982) attempted to see if they could induce cells with fact, the antigen presenting cells are really not “inacrophage-like”in that they are either weakly phagocytic or not phagocytic at all. It would be better if a more appropriate word were to be found to describe this siil)set of cells which share characteristics, other than phagocytosis, with real phagocytic macrophages. However, present terminology generally refers to this class of antigen presenting cells as a part of the macrophage series. Although we do not think it is appropriate to call these cells macrophages, since no generally accepted alternative name for them has been found, we will use the term.

CONTHAS UPPHESSION

30 1

specialized antigen presenting qualities in peritoneal exudates. To do this they induced the peritoneal exudates with a number of different adjuvants. Of the materials they tested, they found that CFA could induce peritoneal exudate cells with a specialized ability to present antigen in such a way that a suppressor-resistaiit response was induced. The use of incomplete Freund’s adjuvant served as a good control because the nature of the cells in the exudates induced by these two types of adjuvants was very s h i l a r . The cells had the same morphologic appearance, were roughly the same size, and expressed approximately the same amount of Ia antigens on a population basis. Only the CFA-induced cells produced the effects described. Thus, Britz and colleagues found an interesting association: the best adjuvant known was also the best su1)stance for inducing cells i n peritoneal exudates that could induce suppressor cell-resistant immunity. The question of whether the CFA induced a phenotypic change in a precursor cell population or whether it simply increased influxing dendritic cells was not established. Nonetheless, although answering that question is a very important one in understanding tlie mechanism of antigen presentation, the findings of Britz and colleagues strongly correlates adjuvanticity with the ability to activate cells that are resistant to suppression. Given our present state of knowledge, it is riot inappropriate to equate the generation of suppressorresistant cells to activity of cells in the contrasuppressor circuit. Thus it is likely that one of the characteristics that give adjuvants their potent immunoaugmenting activity is that they recruit, activate, or otherwise bring into the imniiine response the effector cells of tlie contrasuppressor cell circuit.

VII. Immunological Consequences of Activation of the Contrasuppressor Circuit

Optimal immunization, which usually requires either the use of external adjuvants (see Section VI,H) or the use of antigens that have what Dresser has described as “inherent adjuvanticity” (Dresser, 1962), can lead to a state where animals produce copious amounts of very high-affinity antibody (note the need for adjnvanticity to produce hyperimmunity and the possi1)ility that this is only allowed by activation of contrasuppressor cells. If‘ the serum of‘ these hyperimmune mice is transferred to normal mice, the recipients become totally incapable of making an immune response to tlie specific antigen (Haughton and Nash, 1969). However, further iminunizatioii of the actively sensitized mice continues to push the titer of the antibody higher and make

302

DOUGLAS R . GREEN A N D RICHARD K . GERSHON

the affinity greater (Eisen and Siskind, 1964). Regardless of whether the suppressor effects of the serum transferred into the normal animals is due to antibody feedback and/or circulating suppressor factor (Iverson et al., 1983a), the results imply that the actively immunized mouse must contain a mechanism which makes it resistant to the suppressive material. We propose that the suppressor-resistant state in hyperimmune mice is a consequence of activation of the contrasuppressor system. Spleen cells from hyperimmunized mice can interfere with the activity of suppressor cells in intermediate cultures (as discussed in Section V). One major difference between the hyperimmune and the neonatal systems is that no antigen specificity was found in the neonatal system, whereas in the hyperimmune system the inhibition of suppression is antigen specific. This contrasuppression, however, does not seem to show the same crossreactivity that has been found in contrasuppressor induction (see Sections VI,A and VI, G). Whether this is due to different recognition specificities of contrasuppressor inducer cells and contrasuppressor effector cells or whether there are two contrasuppressor circuits, one crossreactive and the other antigen specific (see Addendum l), is not clear at the present time. However, as with the neonatal contrasuppressor effector cells, the contrasuppressor effector cells of hyperimmune animals expressed the 1-J+, Ly-1 phenotype and adhere to the Vicia oillosa lectin (Green and Gershon, 1982; unpublished observations). The presence of contrasuppressor cells in hyperimmune mice can help explain an apparent paradox. Iverson and colleagues (1983a)found that they can get copious amounts of suppressor factor from the the serum of mice that have been hyperimmunized with SRBC. They isolated the suppressor factor in the serum of these animals with antibodies that react with T cell idiotypes and/or isotypes. Two questions are raised by these findings. How do the mice continue to respond in the face of such high levels of suppressor factor? The answer to this question has been discussed above. It is most likely because contrasuppression counterbalances the effects of the suppressive material. The second question is more intriguing; that is, why does a hyperimmune animal make so much suppressor factor during immunization? Before considering this question one must return to a discussion of the mechanisms which regulate suppressor cell activity. One of these mechanisms is of a “negative feedback” type. Suppressor effector cells are dependent upon their partner inducer cells in order for them to continue to act as suppressor cells (Gershon, 1982). If the suppressor inducer cells are removed physically from suppressor effector populations, and no suppressor inducer cells are present in the assay populations, the activated suppressor cells will not produce any suppression. It has been shown that the suppressor inducer cell is itself a target of suppression; in fact it seems to be more sensitive to the

CONTRASUPPRESSION

303

activity of suppressor cells than are standard helper cells (Green et al., 1981b). It is likely that the suppressor inducer cell is also a target of the contrasuppressor cell. Thus, there can be the paradoxical situation where contrasuppressor cells cause the production of more (not less) suppression (a condition which has been referred to by J. Kemp as a “suppressor pump,” which “pumps” out large amounts of suppressor factors). The suppression, so produced, has no effect upon the contrasuppressed helper and suppressor inducer cells. This may be the situation in hyperimmune mice, which would explain the paradox noted above. The importance of Ly-2+ cells in generating secondary type immune responses has been directly demonstrated by Cantor and colleagues (Cantor and Boyse, 1977). They found that if they reconstituted B mice6 with antiLy-%treated T cells these mice could not make a secondary response, even though their primary responses were often of greater magnitude than the responses of mice reconstituted with unfractionated T cells. How many types of Ly-2+ cells are required to generate memory is not clear. One could push the argument and say the only Ly-2+ cells required are members of the contrasuppressor circuit that serve to protect the helper cells (and/or the B cells) from feedback suppression. This simplistic view is unlikely to be correct, but since there is a requirement for the presence of Ly-2 cells which contain the suppressor cells subset, there is the potential need for contrasuppressor cell protection of the process. +

B. ADOPTIVETRANSFER OF I M M U NRESPONSE E Immunity is often difficult to adoptively transfer to normal adult recipients. Celada (1966) made the observation that neonatal animals were much less resistant to the transfer of adoptive immunity than were adults. This coincided with Simonson’s finding (1962) showing that GVH reactions were also difficult to produce in normal adult mice but could be rather easily produced in neonatal animals. It has subsequently been shown that adoptive immunity can be easily transferred into adult mice if the recipient mice are treated with low dose X-irradiation or cyclophosphamide which temporarily destroys the recipients suppressor circuit (Eardley and Gershon, 1975).This open door to the transfer of adoptive immunity can easily be closed by the addition of relatively sinall numbers of suppressor cells to the treated recipients. In other situations adoptive immunity is not so difficult to transfer to untreated adult mice. One such situation goes back to the original experi“Mice that were thymectomized, lethally irradiated, and reconstituted with syngeneic bone marrow.

304

DOUGLAS R . GREEN A N D RICHARD K . GERSHON

ments of Billingham, Brent, and Medawar (Billinghain et al., 1963). They found that neonatally induced tolerance could be overcome by the addition of immune lymphocytes to the tolerant recipient. Nonimmunized lymphocytes however were very inefficient in breaking the state of immunological tolerance. Another type of immunity which is relatively easy to transfer to untreated recipients is the state of contact sensitivity present 4-5 days after optimal immunization with skin sensitizing reagents like picryl chloride (PCI) (Asherson and Ptak, 1968)(see Sections VI,B,1 and VI,D for the role of contrasuppressor cells in producing this type of immunity). Anti-I-J treatment or Vicia villosa fractionation of immune Ly-1 T cells before adoptive transfer into normal recipients was found to completely abrogate the adoptive transfer (Iverson et al., 198313). However, the 1-J or V . villosa adherent cells removed were not directly involved in the adoptive contact sensitivity. This was shown by transferring the I-J-depleted or V . villosa nonadherent immune cells to recipients that had their suppressor system quenched by pretreatment with low doses of cyclophosphamide. Under these circumstances the recipient mice made equally good adoptive immune responses whether or not the transferred cells had been treated with anti-I-J sera or were V . villosa nonadherent. In fact, their immune response was indistinguishable from the one produced in normal adults by untreated immune cells. The V. villosa adherent cells transferred no immunity on their own, regardless of the recipient’s treatment. These results strongly indicate that under most circumstances the transfer of immune cells activates a host suppressor system and in the absence of contrasuppression (mediated by immune I-J , V . villosa adherent cells) this suppressor system is dominant and immunity is not transferred. This conclusion leads to an important consideration in any investigation of immunological phenomena. It becomes imperative to ask whether any treatment that removes an immunological function is d u e to removal of the cell that performs that function or due instead to the elimination of a cell that protects the effector cells from suppression. This point is relevant, for example, to the interpretation of an experiment where it was shown that nylon adherent 1-J+, Ly-1 cells could amplify immune responses (Tada et al., 1978), although it was not clear that the amplifying cell could directly help B cells. One must consider the possibility that one component of the phenomenon was due to a contrasuppressor effector cell which operated to relieve suppression that was inherent in the system and thus gave the appearance of being a helper subset of T cells. Before the next consequence of activation of contrasuppression is discussed, we should consider why the adoptive transfer of immunity is much more easily achieved in young animals than in adults (as discussed above, see +

+

CONTRASUPPHESSION

305

Section VI,A). It is well known that neonatal mice have quite active suppressor cells. Thus, we would suggest that it is the “level 2” suppressor system which does not appear in significant numbers until animals are 2-3 weeks old that is responsible for blocking adoptive transfer by preventing the generation of contrasuppression in the recipient (see Sections VI,A and VI,F above).

C. MICROENVIHON M ENTAL IMMUNEREGULATION I . In the Intestines It is not difficult to conceive of situations where a potent immune response may be needed in a certain inicroenvironmerit in tlie body and where a comparable systemic immune response might have deleterious side effects. One example of such a microenvironment is tlie mucosal surface of the gut, which separates the body from the potential pathogens. At this barrier it would be highly advantageous to have a strong immune response directed against foreign antigens in the intestines. Although the protective immunity would be against the microorganisms in the gut, it would probably be difficult for the immune system to distinguish those antigens present on inicroorganisms from those present on food and/or its breakdown products. Naturally, there are several reasons why it would be disuduantageous to have the systemic immune system express the same level of iininunity that tlie gut would need to express in order to maintain a tight barrier. Most particularly one could see disasterous results stemming from the leakage of antigens, whether or not they were on pathogens, from the gut into the systemic system. If the systemic immune system responded to nonpathogenic material coming from the gut there would be an increased risk of systemic anaphalaxis and antigeii-antibody complex disease. Another important reason for preventing systemic immunization against microbial pathogens is that many of these pathogens carry heterophile antigens, i.e., the antigen on the pathogens are shared by host tissues. For example, there are crossreactions between antigens on some strains of streptococcus A with those present on heart tissue (Asherson, 1968). The potential autoimmune damage produced by systemic immune response to these heterophile antigens is obvious. Thus, in intestinal inucosa one could envision a very important role for microenvironmental immunoregulation, that is, high responsiveness at the intestinal barrier and decreased immunity elsewhere in the body. This teleological scenario actually describes what is known. To quote froin a recent editorial in Lancet (i, 702, 1981), there is “the paradox that orally encountered antigen can induce protective immunity and systemic toler-

306

DOUGLAS R . G R E E N A N D R I C H A R D K . CERSHON

ance.” The observation by Challacombe and Tomasi (1980) that oral immunization of mice with ovalbumin or streptococcus mutans leads to systemic tolerance with concurrent production of salivary antibodies is just one of the multiple observations on which the “paradox” discussed in the Lancet is built. The paradox could be made more orthodox if it could be shown that protective immunity at the intestinal barrier is maintained by local contrasuppressor cells that keep that microenvironment protected from systemic suppression. This notion has been experimentally tested by asking the question of whether one can find higher levels of endogenous contrasuppressor activity in the Peyer’s patches than in other peripheral lymphoid tissues. It has been found that the Peyer’s patches are particularly rich in I-J , Ly-2 contrasuppressor inducer cells (Green et al., 1982). Mixing cells from the Peyer’s patches with spleen cells in primary in vitro Mishell-Dutton cultures shows that the added Peyer’s patch cells either have no effect on the response of the splenic cells or augment their response. However, if contrasuppressor inducer cells are removed from the Peyer’s patches with an anti-I-J serum, or if their transducer cells are removed from the spleen cell culture with either an anti-I-J serum or an anti-Ly-2 serum, one can find extremely high levels of suppressor cells in the Peyer’s patches. Thus, under normal circumstances there are quite high levels of suppressor cell activity in the Peyer’s patches but this activity is masked by cells that express the contrasuppressor phenotype. If one were able to show that the suppressor cells migrated from the gut and then entered the peripheral lymphoid tissue (Mattingly and Waksman, 1978) while the contrasuppressor cells remained in the Peyer’s patches, one would then have a situation that would be optimal from the teleological point of view expressed above. This would also help explain the paradox of systemic suppression and local hyperimmunity against antigens entering the system from the intestines. +

2 . At Sites Where Interferon Is Released There are other places in the body where microenvironmental immunoregulation might occur but where the teleological explanations for it might require more imagination than in the intestinal situation just discussed. One such group of situations is on the battlefield where the immune system is encountering pathogens. For example, it is clear that T cells are important agents in ridding the host of virally infected cells. Thus, if there were some agent that augmented the T cell activity by removing suppression in the microenvironment where the T cells are attempting to destroy the infected cells that would certainly be advantageous to the host. One substance that is present at these sites is interferon. This agrees with observations that vesicular stomatitis virus (VSV) has the effect of greatly enhancing delayed hypersensitivity to xenogeneic red blood cells (Katsura et al., 1980) by interferring

CONTHASUPPHESSION

307

with suppressor cell activity in this system (Katsura, 1981). Interferon has been shown to interfere with suppressor cell activity in uiuo (Knopf et al., 1982). In addition, treatment of suppressor cells for DTH with interferon caused a loss of suppressive activity. It is not unlikely that this treatment activated contrasuppressor cells in the suppressor cell pool, although this possibility was not examined. Thus, interferon is another agent capable of inducing microenvironmental changes in immune regulation possibly related to contrasuppression.

3. At Sites Where Histamine Is Released Another agent that might induce microenvironmental immune regulation “in the battlefield” is histamine. It has been known now for some time that histamine receptors, of the H2 type, are present on cells with suppressive activity. This has been shown by removing suppressor cells on histamine columns and blocking the removal with histamine 2 antagonists (Rocklin et al., 1978). There also are some data showing that histamine can directly induce suppressor cells and/or suppressor factors (Rocklin et al., 1979). In a recent series of studies Seigel and colleagues (1982) attempted to activate suppressor cells by pulsing immune cells with histamine. They had some occasional experimental successes that were quite dramatic but often got no obvious induction of suppressor cells or found paradoxical doseresponse curves. They therefore tested the proposition that contrasuppressor cells might also have histamine receptors and that these receptors might be of a different type than those found on suppressor cells. Therefore, they did a series of studies using histamine analogs as inducing agents. One of these analogs, dimaprit, has an agonistic effect on histamine type 2 receptors. The other one they used, PEA, has a much greater effect on histamine type 1 receptors. They found that the addition of PEA at 10W4 M could nullify the suppressive effects produced by the dimaprit pulse. They then pulsed separate aliquots of cells, one with dimaprit and other with PEA. The addition of the PEA pulsed cells to the suppressor cells induced by dimaprit completely inhibited the suppression. The cell responsible for this contrasuppressive activity in the PEA pulsed cells was I-J + . Thus, it appears that histamine has the capacity to significantly alter the immunoregulatory activities of more than one type of regulatory T cell. Since it is known that significant amounts of histamine are released at sites where delayed type hypersensitivity takes place (Askenase et al., 1980), one can easily envision a local form of “feedback” microenvironmental immunoregulation going on at the sites where histamine is released.

4 . In the Bone Marrow Another area to be covered where microenvironmental immunoregulation has been shown to be taking place is in the bone marrow. A number of

308

DOUGLAS R. G R E E N A N D RICHARD K . GERSHON

workers have described a very potent suppressor cell present in the bone marrow (Michaelson, 1982). This suppressor cell cannot be identified as any of the known types of immunoregulatory cells; it does not express T cell markers, macrophage qualities, or natural killer cell attributes. Interestingly, the presence of this suppressor cell is hard to demonstrate in the bone marrow of neonatal mice, unless the bone marrow cells are treated with an anit-Thy-1 serum (Michaelson, 1982). Under those circumstances neonatal mice can be shown to have bone marrow suppressor cells as active as those of normal adults. Thus, there is a T cell which, under certain conditions, masks or obscures the presence of a potent bone marrow suppressor cell. The Thy-1 + masking cell can be activated in adult mice by chronic blood letting leading to an anemic state or by stimulating osteoclastic activity with parathyroid hormones. It seems that one mode of microenvironmental regulation of hematopoiesis is brought about by activating a cell which overcomes a constitutive or endogenous state of suppression in the bone marrow. That is to say that the bone marrow suppressor cells seem to retain a constant level of activity independent of what stresses are put on the need for hematopoietic activity within the bone marrow, but its functional activity is regulated by a Thy-l+ contrasuppressor cell.

5 . At the Skin The possible role of contrasuppression in contact sensitivity has already been discussed in several sections, therefore we will only summarize the arguments here. Removal of the painted surface within 4 hr of sensitization results in tolerance rather than immunity (Macher and Chase, 1969). The skin, therefore, endows the system with a resistance to such tolerance induction when it is left intact to present the antigen. Indeed, Langerhans cells have been demonstrated to be capable of presenting antigen immunogenically in the face of tolerogenic stimuli (Section VI,B,5). It is suggested, therefore, that the Langerhans cells of the skin activate contrasuppressor cells to overcome suppression induced by antigen which seeps into the circulation. That contrasuppressor cells are activated by skin painting was demonstrated in transfer studies as discussed (Section VII, B). D. I M M U N RESPONSE E TO MALARIA During sublethal malaria infections in mice animals experience a period of nonspecific suppression (Jayawardena, 1981). Recovery from this suppression coincides with recovery from the infection. Subsequently, immunity can be transferred with Ly-1 T cells which, like the transfer of contact sensitivity, does not require treatment of recipients with X ray or cyclophosphamide (Jayawardena et ul., 1982; R. Mogil, personal communication).

309

CONTRASUPPHESSION

Preliminary evidence that contrasuppression is involved in antiinalarial iininunity coiiies from two sources. Recently, the T cells responsible for transfer of immunity have been found to be sensitive to treatment with antiI-J plus complement (H. Mogil, unpublished observations). Separation of immune cells on Viciu villosa lectin coated dishes produced two populations. The adherent cells were capable of protecting naive, untreate? animals. The nonadherent cells only protected animals which were pretreated with cyclophosphainide (H. Mogil, personal communication). Thus, as i n the transfer of contact sensitivity (Section VII, B) there may be two populations: an I-J - , V . villosa non-adherent helper cell, and an I-J , V. uillosu adherent contrasuppressor cell which protects the helper cell. Unlike the case of contact sensitivity (Section VII, B), the contrasuppressor cell in the inalaria system may have effects on its own, by protecting helper cells generated in the course of the infection. The second observation concerns noiispecific iminunoregulatory effects in uitro. While T cells from aniinals at the peak of a nonfatal infection can suppress in vitro response to SRBC, T cells fi-om aiiiinals in the recovery phase have potent contrasuppressive activity (Green, 1981). It is therefore possible that contrasuppression is responsible for recovery froin suppression and subsequent dominant immunity in nonfatal inalaria infections. What, then, determines whether this contrasuppression is induced? In the examination of two strains of Phsmotliunz yoelii (17X) (one nonfatal, the other a fatal strain derived from the first) it was found that while the fatal parasite predominantly infects mature red blood cells, the nonfatal parasite is found in reticulocytes. These parasitized reticulocytes, even if irradiated or fixed, can immunize aiiiinals against both strains. In addition, induction of reticulocytosis in animals with fatal strain infections allows protective immunity to develop. Again, these parasitized reticulocytes can be used as a vaccine to immunize against either strain (Jayawardena et al., 1983). Recently, parasitized reticulocytes have been used to educate T cells in vitro which, upon removal of red cells and transfer, protected animals from fatal infections (Mogil and Green, 1983). Thus, immunity depends in some way upon presentation of malaria antigens in reticulocytes as opposed to mature red cells. Tying in these observations with those discussed above, we may speculate that something allout reticulocytes allows the generation of contrasuppression which causes a dominant protective immunity. Reticulocytes are devoid of Class I1 MHC antigens, b u t possess low levels of Class I (K and D) antigens. In fact, elevated K and D antigen expression was observed on the parasitized reticulocytes (Jayawardena et nl., 1983) and implicated in the in uitro education (Green and Chue, 1983). The possible role of Class I anti+

310

DOUGLAS R. G R E E N A N D RICHARD K. GERSHON

gens in presentation of foreign antigen to the contrasuppressor circuit is discussed in Section VII1,E.

E. RECOVERYFROM THAUMA ASSOCIATEDSUPPRESSION Immunosuppression as a result of severe burn or surgical trauma has been well documented (Kupper et al., 1984; McLoughlin et al., 1979). A sequence of immunoregulatory events following thermal injury has been described in a murine model (Kupper et al., 1984). Five days following a 30% full thickness scald burn suppressor inducer cells (Ly-l+ ,2-) were found in the spleens. By 7 days postburn nonspecific suppressor effector cells were identified. Between 15 and 30 days postburn, I-J+ T cells appeared which possessed potent contrasuppressive activity. This appearance correlated roughly with a return to immunological responsiveness. Ninneman et al. (1979) described a serum factor in recovered burn patients which was capable of blocking trauma induced suppressor cell activity in uitro. It is tempting to equate this with the contrasuppressive activity found in the murine moldel.

F. IMMUNE REGULATIONI N OLD AGE Doria et al. (1982) described a suppressor T cell in the spleens of old mice which suppressed in uitro immune responses of young animals but had no effect upon the responses of old animals. This paradoxical situation suggested a contrasuppressive activity in old animals. When spleen cells from old mice (>20 months) were cultured for 5 days they produced cells capable of interferring with suppression produced by precultured adult cells. This contrasuppressive activity generated from cells of old animals was sensitive to treatment with anti-I-J C’ (Green, 1981).In addition, monoclonal anti-I-J antibodies which had previously been found to block contrasuppression but not suppression were seen to block this contrasuppressive activity when added to cultures directly (unpublished observations). One explanation for these effects may be that cells of the level 2 suppressor circuit responsible for the regulation of contrasuppression (see Section VI,A) are lost with age. Contrasuppressive effects may tend to dominate, leading to suppressor resistance and, perhaps, autoimmunity. Coincidence of suppressor cells and autoimmunity in old animals is documented (Makinodan and Kay, 1980). Such speculation leads us inevitably to a consideration of the potential role of contrasuppression in autoimmune phenomena.

+

G. AUTOIMMUNITY Defects in suppressor T cell function have been observed in autoimmune mice such as the NZB (Gershon et al., 1978a), and in several human autoim-

CONTRASUPPRESSION

311

mune diseases (reviewed by Smolen and Steinberg, 1982). Yet, autoimmunity can proceed in the face of potent suppressor cell activity, as in the autoimmunity which often accompanies malaria infections (Poels and van Niekerk, 1977) and old age (Makinodan and Kay, 1980). The potential role of contrasuppression in the immune state of each of these was considered above. This is also seen in the MRL model of systemic lupus erythematosis (Murphy and Roths, 1978). In this model, potent suppressor T cells were identified in the spleens which had no ability to affect the “protected” helper T cells in these animals (Gershon et al., 197813). While several lines of evidence suggest that the defects in the MRL mouse involve contrasuppression (Gershon, 1980) direct evidence is lacking. As mentioned above, NZB animals display a deficiency in suppressor cell function which may be related to their lupus-like syndrome (Gershon et al., 1978a). Thymocytes and pre-T cells from NZB animals, even in small numbers, have been shown to block tolerance induction in DBA lymphocytes when injected into irradiated (NZB x DBA/2)F, recipients (Laskin et al., 1983). Experiments are in progress to characterize the regulatory subsets responsible. Furthermore, the injected NZB cells displayed a remarkable loss of regulation, proliferating in their recipients to represent three times the number of DBA/2 cells in the total lymphocyte pool (Raveche et al., 1983). Correlations between lymphoproliferation, suppression (or tolerance) resistance, and contrasuppression recur (see below), but the relationships are not yet clear. A direct correlation of contrasuppression with autoimmunity was found in a murine model in which autoimmunity was induced in genetically normal animals (Smith et al., 1982, 1983a). Animals were thymectomized at 3 days of age, allowed to mature, then stimulated with polyclonal B cell activators (see Section V1,A). These animals became Coomb’s positive and displayed high levels of anti-single-stranded DNA and antithymocyte antibodies. In all cases, spleen cells from autoimmune (but not control) animals produced potent I-J contrasuppressor cells upon 5 day culture. One likely explanation for this phenomenon is that thymectomy occurred at a time following migration of contrasuppressor precursors to the spleen (Showron-Cendrzak et al., 1983) but preceding the production of level 2 suppressor cells, which serve to balance contrasuppression. Stimulation of B cells induces contrasuppression (see Section VI,B, 1)which cannot be homeostatically regulated, thereby leading to autoimmunity. Lymphoproliferation and lymphadenopathy in autoimmune disease may be under similar (but not the same) control. Janeway and colleagues (1981) found that even partial p-suppression greatly reduced the extent and delayed the occurrence of lymphadenopathy in the MRL murine model (in which the lymphadenopathy is predominantly +

312

DOUGLAS R . G R E E N A N D RICHARD K . G E R S H O N

composed of T cells and null cells (Gershon et al., 1978b). These animals, however, still possessed autoantibody. Alternatively, BXSB animals are autoimmune but display little lymphadenopathy (Murphy and Roths, 1978). Thymectomy of these animals on day 3 of life has no effect on the autoimmunity but removes regulatory constraints on proliferation such that massive lymphadenopathy occurs (Smith et al., 1983b). This lymphadenopathy predominantly involves Ig cells. The fact that proliferation and differentiation (or secretion) are under separable regulatory controls (see Addendum 1) opens an experimental route to determining the relationship between regulatory imbalances leading to autoimmunity and those leading to lymphoproliferative disorders. +

VIII. Contrasuppression and Tumor Immunity

A. THEORETICAL CONSIDERATIONS When a tumor grows within the body a massive bombardment of the system with tumor antigens can induce a potent general suppression of immunity (Fujimoto et al., 1976). Often the tumor must be removed before any immunity can be demonstrated. Such observations suggest that while there may be antigenic determinants on tumor cells which can serve as targets of immunity, iminunoregulatory modification might be necessary for such immunity to become manifest. Immune suppression induced by tumor challenge is probably the major stumbling block to effective immunity against many tumors. It is therefore likely that contrasuppression directed against such suppression would be a potent force in the effective engineering of an antitumor response. Alternatively, we can envision situations in which tumors of the lymphoid system may come under suppressor cell control and thereby be rendered benign. For example, Rohrer and Lynch (1979) demonstrated control of clone growth and secretion of the MOPC-315 myeloma by suppressor T cells. Similar effects have been observed by Abbas et al. (1980). Suppressor T cells appear in most people infected with Epstein-Barr virus (EBV) (Tosato et al., 1979), and such T cells have been shown to be capable of inhibiting in uitro EBV transformation of B cells (Thorley-Lawson, 1980). In some cases, therefore, failure to effect suppression of a proliferating cell may be a contributing cause of cancer. Thus, contrasuppression may be involved in the induction or expansion of the transformed cells. Possible examples of this in mouse and man will be considered in Section VIII,F.

CONT RAS UP P RE S S ION

313

B. CONTRASUPPHESSION I N TUMORRESISTANCE Contrasuppression may be implicated in natural resistance to certain oncogenic viruses. Meruelo et al. (1980) demonstrated that the transfer of resistance to AKR leukemia virus oncogenesis was dependent upon an I-J , Ly-1 T cell. In addition, resistant animals could be rendered sensitive to oncogenesis by injection of anti-Ly-1 or anti-I-J sera. Since the effector cell of contrasuppression is an I-J , Ly-1 T cell (Section IV,C), reinoval of this cell could result in loss of dominant immunity and account for these observations. Cells which interfere with suppressor cell function have been implicated in genetic resistance to Friend leukemia virus (FLV). Susceptibility to leukemogenesis was correlated with susceptibility to FLV-induced induction to immunodepression (Ceglowski and Friedman, 1969). Susceptibility to immunodepression, in turn, was correlated with ability of FLV to induce suppressor cells in uitro (Kumar and Bennett, 1976a). Resistance to suppressor cell induction by FLV was shown to be effected by a marrow-dependent cell (“M cell”). Reinoval of the M cell allowed induction of suppressor cells in resistant strains (Kumar and Bennett, 1976b). (Contrasuppressor cells have been identified in bone marrow, see Section VII,C,4). In addition to a contrasuppressive M cell, the FLV system also involves a “suppressor interferring cell” which exhibits striking contrasuppressive effects in uitro. Other characteristics of this system are considered in Section VII1,E. These observations support a role for contrasuppression in control of immunity to leukemia. In other tumor systems a phenomenon possibly related to contrasuppression has been observed. Often animals which are resistant to tumor transfer can be injected with high numbers of tumor cells to overcome the resistance. In many cases, very small numbers of tumor cells are also effective in producing tumors, “sneaking through” an otherwise active defense mechanism. Thus, very sinall and very large numbers of tumor cells can both escape iininune rejection, whereas intermediate numbers induce immune reactions and fail to produce tumors (Naor, 1979). This is reminiscent of the antigen dose-dependent induction of contrasuppression discussed in Section VI,D. We suggest that only the intermediate range of tumor cell numbers is effective in inducing contrasuppression to override the otherwise immunosuppressive influence of the tumor. +

+

I N TUMOR THERAPY C. CONTRASUPPRESSION

The above discussion suggests that an effective tumor therapy would be to activate contrasuppression in cases where suppression seems to dominate. Although contrasuppressor cells have not been used to directly modulate the iininune response in cancer, they have been implicated in several systems.

314

DOUGLAS H. G R E E N A N D RICHARD K . GERSHON

Hamaoka et al. (1977) described an immunization protocol which produced hapten-reactive T lymphocytes in the “absence” of suppressor cells. Using this protocol, Ptak et nl. (1983a,b) primed animals and collected 48 hr supernatants from the spleen cells. These supernatants contained an I-J hapten binding material with potent contrasuppressive activity. Hamaoka et al. (1980)further demonstrated that primed animals were capable of producing effective immunity to haptenated tumor cells. It is likely that the activation of contrasuppression which interferes with suppressor cell activity is responsible for the enhanced immune response against the haptenated tumor cells. This immunity, with time, became crossreactive such that after several immunizations resistance could be demonstrated for the same tumor cells without hapten (Hamaoka et al., 1980). This may be a reflection of the crossreactive nature of the contrasuppressor inducer cell (Section VI, G). Alternatively, antigen bridging (i.e., hapten-modified tumor cells) allowed the hapten specific contrasuppressor effector cells to render helper T cells (Specific for tumor antigens rather than hapten) resistant to host suppressor mechanisms. The resulting suppression-resistant immunity could then protect against unmodified tumor cells. The effective use of adjuvants in tumor therapy is well documented. The action of complete Fruend’s adjuvant in recruiting cells which present antigen to the contrasuppressor circuit was discussed (Section VI,H). Other adjuvants may have the ability to induce contrasuppression. Poly(A:U), like other adjuvants, can interfere with the induction of tolerance. Thymocytes treated with poly(A:U) produce 24 hr supernatants capable of blocking induction of tolerance to dHGG (Fessia et al., 1977). In addition, poly(A:U) accelerates the development of responsiveness in neonatal animals (Han and Johnson, 1976) and is capable of activating neonatal contrasuppressors in vitro in only 24 hr (Green, 1981). Cornybacterium parvum treatment has also been shown to interfere with tolerance induction. Serum from C . parvum-treated animals was capable of inhibiting induction of tolerance and expression of suppressor T cells for contact sensitivity (Knopf et al., 1980). The potential role of contrasuppression in some of the immunoregulatory effects of interferon was discussed in a previous section (Section VII,C,B). +

EVIDENCE FOR CONTRASUPPRESSION I N TUMOR IMMUNITY D. FURTHER Barrett and Deringer (1950) described a phenomenon in which C3H mammary tumors were permanently altered by a single passage in (C3H x BALB/c)F, animals. The passaged cells grew at unexpectedly high frequencies in (C3H x BALB/c) x BALB/c backcross animals (but not in any other mice). Klein and Klein (1957) demonstrated that this effect did not require direct cellular contact between the tumor cells and the first F, host (tumors

CONTRASUPPRESSION

315

were grown in implanted chambers). This remarkable adaptive differentiation of tumor cells has recently reappeared in a different guise. Methylcholanthrene-induced fibrosarcomas from BALB/c mice were grown in BALB/c and in F, animals. The tumors were observed to metastasize at a much higher frequency in the F, animals (Flood et al., 1984). Remarkably, serum from the F, hosts (but not parental hosts) was capable of blocking contrasuppression in vitro. This blocking acitivity was attributed to an antibody which reacts with a tumor surface antigen. Tumor cells passaged in BALB/c animals had no ability to absorb this antibody while those passaged in F, animals could completely remove it. Thus, this BALB/c tumor seemed to express a new antigenic determinant upon passage of the F, host. Antibodies produced against this determinant crossreact with a lymphocyte cellular interaction molecule required for the function of contrasuppressor cells. These F, animals show a higher incidence of metastasis, suggesting a deficiency in effective immune response to the tumor cells. It is tempting to equate this loss of immune function with the production of antibodies which block contrasuppression. The nature of the tumor antigens and the antibodies which show this interesting immunoregulatory activity have been discussed in more detail elsewhere (Flood et al., 1984). For our purposes, however, it still remains to be tested whether the F, antisera are capable of increasing metastatic frequency in BALB/c hosts.

E. CLASSI ANTIGENSI N CONTRASUPPRESSION A N D TUMOR IMMUNITY While it remains to be proven that contrasuppressor cells are needed for optimum tumor immunity, the evidence is compelling. In this section we will consider the activation of contrasuppression and hypothesize a role for antigen presentation in the context of Class I (rather than Class 11)antigens. In recent years it has become dogma that helper T cells recognize antigen in the context of Class I1 surface antigens for the initiation of immune responses. Class I antigens are generally viewed as targets for effector cell (CTL) function, such as in T cell killing of transformed or virally infected targets. With few exceptions, Zr gene effects mapping to Class I loci mediate responses to viral antigens (Zinkernagel et al., 1978) or minor histocompatibility antigens (Wettstein and Frelinger, 1977). It is becoming increasingly clear, however, that Class I antigen presentation in cell-mediated immunity may well involve activation of immunoregulatory subsets. Such regulation has implications for humoral immunity as well. Using H-2D region mutants, Stukart et al. (1982) demonstrated a role for the H-2D locus in regulating responses to Moloney leukemia virus, even when the effector cells were directed only at virus associated with K-end antigens. H-2D region control of immune responses has also been observed

316

DOUGLAS R . G R E E N A N D RICHARD K . GERSHON

for radiation leukemia virus-induced tumorigenesis (Meruelo et al., 1977). Friend virus-induced splenomegaly (Chesebro et al., 1974), T-lymphocyte proliferative autoimmune responses to thyroglobulin (Christadoss et al., 1978), antibody levels and cellular infiltration in autoimmune thyroiditis (Kong et al., 1979), and ability to induce suppression for contact sensitivity with DNFB (Moorhead, 1977) also map to the H-2D region. Antibody responses to equine myoglobulin are regulated by complementing genes in H-2D and I-A (Berzofsky et al., 1982). The presence of contrasuppressor inducer cells in the Peyers patch of naive animals has been discussed (Section VII,C, 1). This phenomenon is under genetic control by genes mapping to the H-2D region. Unlike those of other animals, Peyers patch T cells from animals with H-2D“ haplotype produce dominant suppression when added to normal spleen cells (Green and St. Martin, 1983). Analysis of this suppression suggests that it is of the level 2 type (see Section VI,F). FLV-infected T cells have been observed to be capable of suppressing concanavalin A-induced proliferation (Kumar and Bennett, 197613). If, however, the FLV-infected and Con A-stimulated T cells were genetically different in the H-2D region, “suppressor interferring cells” were induced which blocked the suppression (Kumar and Bennett, 1979). The generation of these suppressor interferring cells in the FLV-infected population could be prevented by irradiation of the cells. This fascinating observation directly points to activation of contrasuppression in a virally infected T cell population by allogeneic H-2D (or H-2D linked) antigens. Class I antigens have been shown to be important in induction of immunity in several tumor systems. SJL reticulosarcoma lines bearing H-2D antigens are capable of inducing immunity to lines which lack H-2D (and are otherwise nonimmunogenic) (Kuhn et al., 1982). Thus, H-2D antigens are not necessary as targets of immunity in this system, but rather as signals for dominant immune responses to other tumor antigens. This was also seen upon examination of progressor and regressor lines of a UV-induced sarcoma (Daynes et al., 1979). Again, the regressor line was found to be capable of inducing immunity to the progressor line. These lines differ by a tumor antigen mapping to the H-2D region of the MHC present only on cells of the regressor line. It is interesting that, in addition to the antitumor response, the presence of this antigen on the tumor cells induces a potent autoimmunity (H. Schrieber, personal communication). Recently, a system has been developed to analyze the activation of contrasuppression by antigen presenting cells in vitro. Results suggest that this antigen specific activation can be blocked by anti-Class I (especially H-2D) butnot by anti-Class I1 antibodies (Green and Chue, 1984). In light of the above observations, we purpose that antigen presentation in

CONTKASUPPHESSION

317

the context of Class I antigens, especially H-2D (or antigens closely linked to H-2D), may be important in initiation of contrasuppression. It may be relevant that dendritic cells (see Section VI,B,5) are high in H-2D antigen expression. A strategy for optimal tumor immunity inay be elevation of Class I antigen expression on tumor cells. This, in turn, inay activate contrasuppression and allow dominant immune responses over tumor-induced suppression.

F. CONTRASUPPRESSION I N ENHANCEMENT O F LYMPHOID TUMOH DEVELOPMENT It is well established that persistent activation of target cells b y their hormones can result in transformation and carcinogenesis. Regulatory factors are essentially hormones of the immune system, and we can propose that persistaiit activation of their targets can result in neoplasia. Signals which inhibit activation, such as suppressor cell factors, might then serve to prevent lymphoid transformation whereas activities like contrasuppression might, in some instances, enhance lymphoid tumorogenesis. For example, Lynes et al. (1978) have described a situation in which antigenic hyperimmunization causes the appearance of tumors of cells of the immune system. The fact that several B cell lymphomas produced in this way react with the immunizing antigen suggests direct involvement of the hyperiminunization protocol. Hyperiminunized animals have been shown to possess a potent antigen-specific contrasuppressive activity (see Section VI1,A). As mentioned in Section VII,D, malaria infections in mice produce a potent contrasuppression coincident with recovery. Such infections can enhance oncogenesis b y virus (Wedderburn, 1970). Whether there is any correlation of these effects is unknown, but suggests an exciting possibility. People infected with Epstein-Barr virus (EBV) exhibit potent suppressor T cell activity (Tosato et al., 1979), and such cells have been shown to be capable of inhibiting EBV transformation in oitro (Thorley-Lawson, 1980). Chronic infection with malaria, however, might induce a general contrasuppression which would interfere with this beneficial immunosuppression to allow expansion of the virus-transformed cells. This is a possible rationale for the assocaition of EBV-induced lymphomas in malarial regions (Miller, 1976). We suggested previously that the lyiiiplioproliferatioi~in the MKL mouse might be associated with contrasuppression (Section VII, G). While these proliferating cells are not tumor cells, they do transform spontaneously in uitro after which they produce tumors upon transfer (C. Reinisch, personal communication). A n understanding of the role of iininuiioregulatory T cells

318

D O U G L A S R . GREEN A N D RICHARD K . GERSHON

in the control of such lymphoid tumors will greatly increase our knowledge of lymphocyte regulation and the regulation of transformed cells in general. IX. Human Examples of Contrasuppression

Contrasuppression is too new a concept to have produced extensive work in human systems. Nevertheless, it is an indication of the prevalence of this immunoregulatory activity that a few examples have been found already and are being actively characterized. In this section we will review this work and discuss some other areas which may prove fruitful. Lehner (1982) employed the antigen dose phenomenon (Section VI, D) to produce regulatory T cell factors from human T cells. As expected, he found that at low and high doses of antigen (streptococcus mutans or sheep red blood cells) suppressor factors were produced in vitro, while at optimal doses helper factors were produced. When T8 cells were removed helper factors were produced at all antigen doses. This is not surprising, as the T8+ T population contains suppressor T cells (Reinherz et al., 1980). When T4+ cells were removed, however, a rather unexpected result was obtained: suppressor factors were produced at all antigen doses. Thus, at optimal antigen doses the helper cells in whole T populations are apparently protected from these suppressor factors. In addition, examination of the T8+ pool revealed an antigen binding cell which, upon removal, caused the whole T population to produce suppressor factor at all antigen doses. This T8+ antigen binding cell has also been found to be Ia+ and adhere to the Vicia villosu lectin (T. Lehner, personal communication). Thus, although T8 suppressor cells were observed to function at all antigen doses, a T8 antigen binding cell was protective at optimal doses (under conditions used, the suppressor cells failed to bind antigen). At present it is unclear whether this cell induces or effects the observed phenomenon (it might interact with the cells of the T4 population) but it seems very likely that this represents the human equivalent of a murine contrasuppressor cell. Immunoregulatory effects are often associated with human disease states. Elson et ul. (1981) examined regulatory cells in patients with Crohn’s disease, an inflammatory bowel disorder. Peripheral blood lymphocytes from patients with Crohn’s disease produced normal responses to pokeweed mitogen and displayed no suppressive effects when added to those of normal subjects. When B cells and T cells were purified by immunoabsorbent affinity chromatography, however, the purified T cells were potently suppressive. These “covert” suppressor cells, therefore, had their function interrupted by cells which were removed during purification. Two patients were also examined in which the suppression associated with Crohn’s disease was “overt,” that is, no cell purification was needed in order to observe +

+

+

+

CONTRASUPPHESSION

319

the suppression. This “overt” suppression accompanied hypogammaglobulinemia which waned with recovery from the disease. This indicates that the suppressor cells were capable of profound physiologic effects. Perhaps more importantly for our purpose it also suggests that the cells responsible for the contrasuppressive effects which made the suppression “covert” in most patients also protected these patients from the hypogammaglobulinemia in uiuo. The nature of these interferring cells is currently unknown, but their contrasuppressive activity makes them tantalizing targets for study. T cells from the synovial fluid of patients with rheumatoid arthritis display defects in suppressor cell activity. Unlike T cells from normal subjects, they do not suppress (but rather enhance) B cell responses to pokeweed mitogen (Chattopadhyay et al., 1979a) and cannot be induced by concanavalin A or antigen to suppress immune responses (Chattopadhyay et al., 1979b). Romain et al. (1982) examined the “augmenting” T cell from synovial fluid in these patients and found them to be I a + , T8+ cells like those described by Lehner (see above). Do these cells account for the “loss” of suppressor cell activity in these patients? It would be interesting to add these synovial T cells to suppressor cells induced from the more normal peripheral blood T cells of such patients to examine their ability to override suppression. Apparent loss of suppressor cell activity accompanies several other human diseases, including systemic lupus erythematosis (SLE) (Sakane et al., 1978), and “autoimmune” chronic active hepatitis (CAH) (Nouri-Aria et al., 1982). In light of the immunoregulatory activities discussed in this review, we may question whether suppressor cells have actually been “lost” or contrasuppression has been activated. In both SLE and autoimmune CAH, treatment of patients with low doses of corticosteroids can restore suppressor cell function (Hirschberg et al., 1980; Nouri-Aria et al., 1982). In one study (Nouri-Aria et al., 1982) T cells from autoimmune CAH patients were treated with prednisolone in uitro. These T cells then could be readily induced to become suppressive. No mixture experiments were performed to examine whether suppression or lack of suppression was the dominant effect. It is, however, quite possible that low doses of corticosteroids destroy contrasuppressor cells which obscure suppression. Facilitation of suppressor cell generation by low concentration of corticosteroids has been observed in mice (Cohen and Gershon, 1975) and humans (Hirschberg et al., 1980). In one experimental human study (Hirschberg et al., 1980) corticosteroid treatment not only enhanced the generation of suppressor cells, but yielded assay cells which were far more readily suppressed. These observations are compatible with the notion that corticosteroids can function at low concentrations to remove contrasuppressor cells. Thus, several disease states in humans may involve contrasuppressive

320

DOUGLAS H . GREEN A N D HICHARD K . GEHSHON

activity. Future studies will have to concentrate on cell surface phenotypes of human contrasuppressor cells to make identification easier. It is a simple matter to ask whether loss of suppressor cell activity is caused by the presence of other cells exerting a contrasuppressive effect. When such studies become more routine, we may quickly learn to recognize human contrasuppressor cells and identify their roles in human health and disease. X. Contrasuppression in the Future

In composing this review we undertook the difficult task of covering a subject about which little has been published. Contrasuppression is, however, a concept which we believe has a very wide application. For these reasons we have indulged in quite a bit of (hopefully) justified speculation. As the research in immunoregulation continues, many of these ideas will certainly be modified. A few important themes which have been stressed throughout this article, however, will probably persist and have already been independently verified. In this conclusion, then, we will reiterate some of these points.

Contrasuppression is an immunoregulatory T cell activity which is distinct from help and helper augmentation. This is not to say that contrasuppression cannot appear to behave like these activities in complex systems. However, contrasuppression functions to interfere with the interaction of suppressor cells and their targets. Therefore, the positive action of the contrasuppressor cells can only be observed in the presence of suppression. This balance of suppression and contrasuppression is important during every phase of an immune response. Contrasuppression protects helper cells. Helper T cells can act as a target of both suppression and contrasuppression and this has consequences for many helper cell studies. In the absence of contrasuppressor cells, immune T cells adoptively transfer contact sensitivity into cyclophosphamide-treated but not normal animals (Iverson et al., 198313). Any system, therefore, in which immune cells are capable of transferring immunity upon systemic injection is likely to involve a contrasuppressor cell which permits the immune population to evade the recipient’s suppressor mechanisms. Studies which seek to characterize the immune population without regard for these considerations are as likely to describe the contrasuppressor as helper or effector cell in the immune population. Reports of I-J+ helper or helper augmenting cells (Jayaraman et al., 1982; Tada et al., 1978) may be misleading for this reason (see Addendum 1). The balance of suppression and contrasuppression may control both the amplitude and the quality of the immune response. Suppressor cells have been described for virtually every type of immune response. If contrasup-

CX)NTHASUPPHESSION

32 1

pressor cells display a similar diversity, then fine tuning of the immune response may depend on activity of these opposing regulatory activities. (Evidence for such diversity in contrasuppression is discussed in Addendum 1.) An example of such “quality control’’ might be the regulation of the IgE response. This isotype, with its central role in allergy, is controlled by the interaction of T suppressor or inhibitory factors with T-enhancing or potentiating factors specific for IgE responses (Ishizaka et al., 1983; Katz, 1983). Complex regulatory T cell interactions are involved in the induction of these antagonistic activities. If, indeed, “enhancement” is contrasuppression (that is, its only function is to interfere with suppressive influences to allow, rather than induce, dominant immunity), then these systems are bound to shed light on issues of the involvement of immunoglobulin with the contrasuppressor T cell circuit. The probability that suppression and contrasuppression control not only the amplitude but the quality of an immune response underlines the potential importance of such interactions in both tumor rejection and autoimmunity. These represent the boundary between appropriate and inappropriate immune responses (in fact, tumor rejection could be viewed as a desirable “organ”-specific autoimmune phenomenon). As is well known, it is not only the existence of an immune response but the involvement of particular cell types or antibodies (or both) which determine its destructiveness. From this point of view, the future of research in contrasuppression, and irnmunoregulation in general, will determine our ability to perceive and manipulate the decision-making apparatus of the immune system. This claim should be no more, or less, astounding than that of a neurophysiologist who proposes to understand a behavior in terms of neuronal interactions and to manipulate that behavior by stimulation of particular cells. W e know that for some behaviors in “simple” systems such claims have been fulfilled. It is our ambition to do the same for the circuitry of the immune system, to understand and control its complex behavior. ACKNOWLEDGMENTS The concept of contrasuppression is the result of the combined work of many researchers who, in addition to the authors, include H. Cantor, B. Chue, I). I>. Eardley, T. A. Ferguson, P. M. Flood, M. Horowitz, C. A. Janeway, J. Kemp, J. D. Michaelson, R. J. Mogil, I). B. Murphy, W. Ptak, S. St. Martin, and K. Yamauchi. DRG received financial support from the Department of Surgery, Yale University and from the UNDP/World Bank/WHO special program for Research and Training in Tropical Diseases. The authors wish to thank Thomas A. Ferguson, Patrick M. Flood, and Rona J. Mogil for critical discussion of the manuscript as well as Astrid Swanson and Virginia Fowler for secretarial assistance. This article is dedicated to the memory of Richard Keev Gershon and to those who will continue his work and extend his ideas.

322

DOUGLAS R . GREEN A N D RICHARD K . GEHSHON

REFERENCES Abhas, A. K., Burokoff, S. J . , Gefter, M. L., and Greene, M . I. (1980).J . E x p . Med. 152, 969. Asherson, G. L. (1968). P r o g . Allergy 12, 192. Asherson, 6. L., and Ptak, W. (1968). Immunology 15, 405. Askenase, P. W., Hayden, B. J., and Gershon, R. K. (1975).J . E x p . Med. 141, 697. Askenase, P. W., Bursztajn, S., Gershon, M. D., and Gershon, R. K. (1980).J . Exp. Med. 152, 1358. Barret, M. K., and Deringer, M. K. (1950). J . Natl. Cancer Znst. 11, 51. Benjamin, D. C., and Weigle, W. 0. (1970).J . Zmmunol. 105, 1231. Berzofsky, J. A , , Buckenmeyer, G. K., and Hicks, G. (1982). J. Immunol. 128, 737. Billingham, R. E. , Silvers, W. K., and Wilson, D. B. (1963). J . E x p . Med. 118, 397. Braley-Mullen, H. (1980). J . Zmmunol. 125, 1849. Braley-Mullen, H. (1983). Fed. P r o c . , Fed. Am. Soc. E x p . B i d . 42, 3829 (Abstr.). Britz, J. S.,Askenase, P. W., Ptak, W., Steinman, R . M., and Gershon, R. K. (1982).J. E x p . Med. 155, 1344. Burrows, P. D., Gershon, R. K., Lawton, A. R . , and Mowry, R . W. (1976). Fed. P r o c . , Fed. Am. Soc, E x p . Biol. 35, 3615 (Abstr.). Cantor, H., and Boyse, E. A. (1975).J . E x p . Med. 141, 1376. Cantor, H., and Boyse, E. A. (1977). Cold Spring Harbor Symp. Quant. B i d . 41, 23. Ceglowski, W. S., and Friedman, H. (1969). Nature (London)224, 1318. Celada, F. (1966).J . E x p . Med. 124, 7. Challacomhe, S. J., and Thomasi, T. B. (1980).J . E x p . Med. 152, 1459. Chattopadhyay, C., Chattopadhyay, H., Natvig, J. B., Michaelson, T. E., and Mellbye, 0 . J . (1979a). Scand. J. Ztnniunol. 10, 309. Chattopadhyay, C., Chattopadhyay, H., Natvig, J. B., and Mellbye, 0 . J . (1979b). Scand. J. Immunol. 10, 479. Chesebro, B., Wehrly, K., and Stimpfling, J. (1974). J . E x p . Med. 140, 1457. Christadoss, P., Kong, Y. M., Elkehewy, M., Rose, N. R . , and David, C. S. (1978). In “Genetic Control of Autoimmune Disease” (N. R. Rose, P. E. Bigazzi, and N . L. Warner, eds.), p. 445. Elsevier, Amsterdam. Claman, H. N., Miller, S. D., Sy, M.-S., and Moorhead, J. W. (1980). Zmmunol. Reo. 50, 105. Cohen, P., and Gershon, R . K. (1975). Ann. N.Y. Acad. Sci. 249, 451. Daynes, R. A., Fernandez, P. A., and Woodward, J. G. (1979). Cell. Immunol. 45, 398. Doria, G., Mancini, C., and Adorini, L. (1982). Proc. Natl. Acad. Sci. U . S . A . 79, 3803. Dresser, D. W. (1962). lintnunology 5, 378. Eardley, D. D., and Gershon, R . K. (1975).J . E x p . Med. 142, 524-529. Eisen, H. N., and Siskind, G. W. (1964). Biochemistry 3, 996. Elson, C. O . , Graeff, A. S., James, S. P., and Strober, W. (1981). Gastroenterology 80, 1513. Fessia, S. L., Kong, Y. M., Bick, P. H., and Johnson, A. G. (1977). Eur. J . Zrnmunol. 8, 508. Flood, P. M., Yamauchi, K., and Gershon, R. K. (1982).J. E x p . Med. 156, 361-371. Flood, P. M., DeLeo, A. B., Green, D. R., Old, L. J., and Gershon, R. K. (1984). Submitted. Fresno, M., McVay-Boudreau, L., Nabel, G., and Cantor, H. (1982). J. Exp. Med. 153, 1260. Fujimoto, S., Greene, M. I., and Sehon, A. H. (1976).J . Zmmunol. 116, 791. Gershon, R. K. (1974a). In “The Immune System: Genes, Receptors, Signals” (E. Sercarz, A. Williamson, and C. F. Fox, eds.), pp. 471. Academic Press, New York. Gershon, R. K. (1974bj. In “Immunological Tolerance: Mechanisms and Potential Therapeutic Applications” (D. H. Katz and B. Benacerraf, eds.), p. 441. Academic Press, New York. Gershon, R. K. (1980).J . Allergy Clin. Immunol. 66, 18.

CONTRASUPPHESSION

323

Gershon, R. K. (1982). I n “Immunogenetics and Immune Regulation, Letture Schiapparelli” (B. Benacerraf, ed.), p. 61. Masson, Paris. Gershon, R. K., and Kondo, K . (1971). Ztntnunology 21, 903. Gershon, R. K., and Kotido, K. (1972). Science 175, 996. Gershon, R. K., Eardley, D. D., Naidorf, K. F., and Ptak, W. (1976). Cold Spring Harbur Symp. Quant. B i d . 45, 85. Gershon, R. K., Eardley, D. D., Naidorf, K., and Cantor, H. (1978a). Arthritis Rheum. 21, 180. Gershon, H. K . , Horowitz, M., Kemp, J. D., Murphy, D. B., and Murphy, D. (1978b). In “Genetic Control of Autoimmune Disease” (N. R. Hose, R. E. Bigazzi, and N. L. Warner, eds.), p. 223. Elsevier. Amsterdam. Gershon, R. K., Eardley, D. D., Dururn, S . , Green, D. R., Shen, F. W., Yamauchi, K., Cantor, H., and Murphy, D. B. (1981). J. Exp. Med. 153, 1533. Goidl, E. A , , and Siskind, G. W. (1974). J . Exp. Med. 140, 1285. Green, D. R. (1981). Ph.D. thesis, Yale University, New Haven. Green, D. R. (1984). Znt. Congr. Itnniunol. 5th. Kyoto. Green, D. R. and Chue, B. (1984). In preparation. Green, D. R., and Gershon, R. K. (1981). In “Immunoglobulin Idiotypes” (C. A. Janeway, E. E. Sercarz, H. Wigzell, and C. F. Fox, eds.), p. 893. Academic Press, New York. Green, D. R., and Gershon, R. K. (1982). Ann. N.Y. Acad. Sci. 392, 318. Green, D. R., and St. Martin, S. (1983). Ann. N.Y. Acad. Sci. 409, 284. Green, D. R., Eardley, D. D., Kimura, A., Murphy, D. B., Yamauchi, K., and Gershon, R. K. (1981a). Eur. J. Ztnmunol. 11, 973. Green, D. R., Gershon, R. K., and Eardley, D. D. (1981b). Proc. Natl. Acad. Sci. U . S . A . 78, 3819. Green, D. R., Gold, J . , St. Martin, S., Gershon, R., and Gershon, R. K. (1982). Proc. Natl. Acad. Sci. U.S.A. 79, 889. Green, D. R., Chue, B., arid Gershon, R. K. (1983). J. Mol. Cell. Zmmunol. 1 19. Green, M. I., and Benacerraf, B. (1980). Itnmunol. Rev. 50, 163. Hamaoka, T., Toshizawa, M., Yamamoto, H., Kuroki, M., and Kitagawa, M. (1977). J . Exp. Med. 146, 91. Hamaoka, T., Fujiwara, H., Teshima, K., Aoke, H., Yamamoto, H., and Kitagawa, M. (1980). J . Exp. Med. 149, 185. Han, I. H., and Johnson, A. G. (1976). J . Imtnunol. 117, 423. Haughton, G., and Nash, D. R. (1969). Transplant. Proc. 1, 616. Herzenberg, L. A,, Okumura, K., Cantor, H., Sato, V., Shen, F. W., Boyse, E. A , , and Herzenberg, L. A. (1976). J . Exp. Med. 144, 330. Heyman, B., Andrighetto, G . , and Wigzell, H. (1982). J . Exp. Med. 155, 994. Hirschberg, T., Randazzo, B., and Hirschberg, H. (1980). Scand. J . Zmmunol. 12, 33. Hodes, R. J., and Hatchcock, K. S. (1976). J . Ztnmunol. 116, 167. Huber, B., Devinsky, O., Gershon, R. K., and Cantor, H. (1976). J . Exp. Med. 143, 1534. Hurtenbach, U., Sachs, D. H., and Shearer, G . M. (1981). J . E x p . Med. 154, 1922. Ishizaka, K., Yodoi, J., Suemura, M., and Hirashiina, M. (1983). Zmmunol. Today 4, 192. Iverson, G. M., Eardley, D. D., Janeway, C. A,, and Gershon, R. K. (1983a). Proc. Natl. Acad. Sci. U.S.A. 80, 1435. Iverson, G . M., Ptak, W., Green, D. R., and Gershon, R. K. (1983b). I . E x p . Med. 158, 982. Janeway, C. A,, Sharrow, S. D., and Simpson, E. (1975). Nature (London) 253, 544. Janeway, C. A,, Jr., Broughton, B., Dzierzak, E . , Jones, B., Eardley, D. D., Durum, S., Yamauchi, K., Green, D. R., and Gershon, R. K. (1981). ICN-UCLA Symp. Mole. Cell. Biol. 10, 661.

324

DOUGLAS R. GREEN A N D RICHARD K. GERSHON

Jayaraman, S., Swierkosz, J. E., and Bellone, C. J. (1982).J. E x p . Med. 155, 641. Jayawardena, A. (1981). In “Parasitic Diseases” (J. M. Mansfield, ed.). Decker, New York. Jayawardena, A. N., Murphy, D. B., Janeway, C. A,, and Gershon, R. K . (1982).J. Zniniunol. 129, 377. Jayawardena, A. N., Mogil, R . , Murphy, D. B., Burger, D., and Gershon, R . K. (1983).Nature (London) 302, 623. Kapp, J. A,, Pierce, C. W., and Benacerraf, B. (1975).J. E x p . Med. 142, 50. Katsura, Y. (1981).J. Zmrnunol. 126, 2067. Katsura, Y., Takaoki, M., Kono, Y . , and Minato, N . (1980).J . Zrntnunol. 125, 1459. Katz, D. (1984).Allergy, in press. Klein, G., and Klein, E. (1957). Symp. Soc. E x p . Biol. 11, 305. Knopf, J . , Riechmann, R., and Mecher, E. (1980).J . Invest. Derinutol 77, 469. Knopf, J., Stremmer, R . , Neumann, C . , DeMaeyer, E., and Macher, E. (1982). Nature (London) 296, 757. Kong, Y. M., David, C. S . , Giraldo, A. A , , El-Rehewy, M., and Rose, N. R. (1979). J . Zmmunol. 123, 15. Kuhn, M., Dyer, D. A., Mayo, L., Jagdish Babu, K., Righthand, V. F., Lightbody, J., Beisel, K. W., and Lerman, S. P. (1982). Fed. Proc., Fed. Am. Soc. Exp. Biol. 41, 822. Kuniar, V., and Bennett, M. (1976a).J. E x p . Med. 143, 728. Kumar, V., and Bennett, M. (1976b).J. E x p . Med. 143, 713. Kuniar, V . , and Bennett, M. (1979). Proc. Natl. Acad. Sci. U . S . A . 76, 2415. Kupper, T. S., Green, D. R., Chaudry, I. H., and Baue, A. E. (1984). Surgery, in press. L’age-Stehr,J., Teichmann, H., Gershon, R. K., andCantro, H. (1980).E u r . ] . Zmnwnol. 10, 21. Laskin, C. A , , Smathers, P. A , , Lieberman, R . , and Steinberg, A. D. (1983). Fed. Proc., Fed. A m Soc. E x p . Biol. 42, 4607 (Abstr.). Lehner, T. (1982).J. Zmmunol. 129, 1936. Lehner, T. (1983). E u r . J. Zmrnunol. 13, 370. Lynes, M. A , , Lanier, L. L., Babcock, G . F., Wettstein, P. J., and Haughton, 6. (1978).J. Initnuno/. 121, 2352. McDougal, J. S., Shen, F. W., Cort, S. P., and Bard, J. (1980).J . Znimunol. 125, 1157. Macher, E., and Chase, M. W. (1969).J. Exp. Med. 129, 81. McLoughlin, G . A., Wu, A. V., et a / . (1979). Ann. Surg. 190, 297. Makinodan, T., and Kay, M. B. (1980). Ado. Znimunol. 29, 287. Mattingly, J. A , , and Waksman, B. H. (1978).J. Ztnmunol. 121, 1878. Mercer, E . H. (1981). “The Foundations of Biological Theory.” Wiley, New York. Meruelo, D., Lieberman, M., Gington, N . , Deak, B., and Mcllevitt, H. 0. (1977)./. E x p . Med. 146, 1079. Meruelo, D., Fleiger, N., Smith, D.,and McDevitt, H. 0. (1980). Proc. Natl. Acud. Sci. U . S . A . 77, 2178. Michaelson, J. D. (1982). Medical School thesis, Yale University, New Haven. Miller, 6. (1976). I n “Viral Infection of Humans’’ (A. S . Evans, ed.). Plenum, New York. Mogil, R. J., and Green, D. R. (1983). Znt. Congr. Zmn~unol.5th, Kyoto. Moller, G . , ed. (1983). ZnLniunol. Reo. 69. Moorhead, J. W. (1977).J . Imnunol. 119, 1773. Mosier, D. E., and Johnson, B. M. (1975).J . Exp. Med. 141, 216. Murgita, R. A., Goidl, E. A . , Kontianen, S . , Beverly, P. C. L., and Wigzell, H. (1978). Proc. Natl. Acad. Sci. U . S . A . 75, 2897. Murphy, E. 1). , and Roths, J. B. (1978). I n “Genetic Control of Autoimmune Disease” (Rose, Bigazzi, and Warner, eds.), p. 207. Elsevier, Amsterdam. Naor, D. (1979). Ado. Cancer Res. 29, 4Fj. Ninnemann, J. L., Fisher, J. C., and Wachtel, T. L. (1979).J. Zmmunol. 122, 1736.

CONTRASUPPHESSION

325

Nouri-Aria, K. T., Hegarty, J. E., Alexander, G . J. M . , Etltlleston, A. L. W. F., and Williams, R. (1982). N . EtigI. J. M e d . 307, 1301. Nussensweig, M. C., Steinrnan. R. M . , Unkeless, J. C., Witmer, M. I)., Gutchinow, B., and Colin, Z.A. (1981).J . Ex)).M u d . 154, 168. Parks, 11. E. (1981). J. Suprunid. Strrrct. 5, 197 (Abstr.). Pearlinan, D. S. (1973). I n “lminuno~ogicDisorders in Itifants and Children” (E. R. Stiehin and V. A. Fulginiti, eds.). p. 333. Saunders, Philadelphia. Pennsylvania. Poels. L. B . , and van Niekerk, C. C. (1977). E x p . Purusitol. 42, 235. Polyani. M. (1968). Scicmce 160, 1308. Ptak, W., Naidorf, K.,Striyzewska, J., antl Gershon, R. K. (1979). Eur. J . Inimunol. 9, 495. I’tak, W., Hozycka, D.. Askenase, P. bt’., and Gerslion, R. K. (1980). J. E x p . Med. 151, 362. Ptak, W., Green, I). 11.. Drirrim, S. K.,Kimura, A , , Murphy, D. B., and Gershon, R. K. (1981). Eur. J. I)nmrno/. 11, 980. Ptak, W., Ptak, M . , Rosenstein, R. W., and Gershon, R. K. (1982). Proc. Natl. Accid. Sci. U . S . A . 79, 2375. Ptak, W., Bereta, M . , hlarcinkiewicz. J., Gershon, R. K . . and Green, I). R. (1984a). J. Iininunol., in press. Ptak, W., Bereta, M . . Ptak, hl.. Gershon. R. K.,antl Green, I). R. (19841)). J. I n r m c i n o l . , i n press. Rao. V. S., Bennett, J. A,, Siren, F. W., Gershon, R. K., and Mitchell, M. S. (1980). J , Irnniunol. 125, a. Raveche, E. S., Laskin. C. A , , Steintierg, A. 1). and Tjio, J . H. (1983). Fed. Proc., Fed. Am. Soc. E x p . Biol. 42, 6129 (Abstr.). Reitiherz, E. L., Kung, P. C.. Goldstein, 6 . . and Sclilossman, S . F. (1980). J. Ininiunol. 124, 1301. Rockliti, R. E . , Greineder, D., Littinan. B. H.. atid Melnion, K. L. (1978).Cell. In~mrrnol.37, 162. Rocklin, R. E . , Greineder, I). K . , and hlelnioti, K. L. (1979). Cell. Inrmund. 44, 404. Rohrer, J. W., and Lynch, R. 6. (1979). J . Iitununol. 123, 1083. Roinain, P. L., Burinester, G.R., Enlow, R. W., and Winchester, R. J. (1982). Rheunratol. I n t . 2, 121. Sakane, T., Steinlwrg, A. D.. and Green, I. (1978).Arthritis Rheum. 21, 657. Scliiff, C. (1980). Medical School thesis, Yale University, New Haven. Schwartz, M . , Waltenhrigh, C., Dorl‘. M . , Cesla, R., Sela, hl,, and Benacerraf, B. (1976). Proc. N u t / . Acud. Sci. C1.S.A. 73, 2862. Seigel, J. N . , Schwartz. A , , Askelraw, P. W., ;rnd Gershon, 11. K. (1982). Proc. Natl. Acud. Sci. U . S . A . 79, 5052. Sercarz, E. E., Yowell, R. L.. Turkin, I ) . . Miller, A , , Aratieo, B., and Adorini, L. (1978). Ininiunol. Rec. 39, 109. Shepherd, G . M. (1979). “The Synaptic Organization of the Brain.” Oxford Univ. Press, London and New York. Gershon, R . K . , and I’tak, W. (1983). Proc. N o t l . Acud. Skowron-Cendrzak, A , , Rybcynska, Z., Sci. U S . A . , 80, 5052. Simonson, M. (1962). Prog. Allerg!/ 6 , 349. Smith, H. R.. Green, I). R.,Raveche, E:. S.,Slnathers, P. A , , Gershon, R. K . , and Steinherg, A . I). (1982). J . ~ n i n i ~ r n o 129, /. 23:32. Smith. H . , Green, I). R.. Huveche, E. S., Sinathers, P. A , , (krshoii, R. K., and Steinherg, A. D. (1983a). C l i n . Ex”. Zninruiiol. 51, 579. l. Smith, H . R., Chusrd, T. M., Sinathers, P., and Steinl)erg, A. 11. (19831)).J. I ~ n m u n c ~130, 1200. Srnolen, J. S., and Steinl)erg, A . I). (1982). I n “The Pathophysiology of Human Imlnunohgic

326

DOUGLAS R. GREEN AND RICHARD K . GERSHON

Disorders” 0. J. Twomey, ed.), p. 173. Urban & Schwarzenberg, Baltimore, Maryland. Stukart, M. J., Vos, A,, Boes, J., Melvold, R. W., Bailey, D. W., and Melief, C. J. M . (1982).J . lmmunol. 128, 1360. Tada, T., and Okumura, K. (1979). Adu. Zmmunol. 28, 1. Tada, T., Takemori, T., Okumsera, K., Nonaka, M., and Takuhisa, T. (1978).J . E x p . Med. 147, 446. Taniguchi, M., Tada, T., and Tokuhisa, T. (1976).J. Exp. Med. 144, 20. Thorley-Lawson, D. A. (1980).J. lmmunol. 124, 745. Tite, J. P . , and Playfair, J. H. L. (1978). Immunology 34, 1097. Tosato, G . , Magrath, I., Koski, T., Dooley, N . , and Blaese, M. (1979). N . Engl. J . Med. 301, 1133. Wedderburn, N . (1970). Lancet 11, 1114. Weigle, W. 0. (1967). I n “Monographs in Microbiology.’’ World, New York. Wettstein, P. J . , and Frelinger, J. S. (1977).J. E x p . Med. 146, 1356. Wicker, L. S . , Benjamin, C. I]., Katz, M. E., Sercarz, E. E., and Miller, A. (1982). Fed. Proc., Fed. Am. Soc. E x p . B i d . 41, 199 (Abstr.). Yamauchi, K., Green, D. R., Eardley, D. D., Murphy, D. B., and Gershon, R. K. (1981).J . E x p . Med. 153, 1547. Yamauchi, K., Chao, N., Murphy, D. B., and Gershon, R. K. (198%).J. Erp. Med. 155, 655. Yamauchi, K., Taniguchi, M., Green, D., and Gershon, R. K. (1982b). Itnmunogenetics 16, 551. Zinkernagel, R . M., Althage, A., Cooper, S., Kreeb, G . , Klein, C. A , , Sefton, B., Flaherty, L., Stimpfling, J., Shreffler, D., and Klein, J. (1978).J. E x p . Med. 148, 592.

Addendum 1: Heterogeneity of Contrasuppressor T Cell Function and Specificity1

A dissection of carrier immunization induced regulation of a monoclonal B cell tumor (MOPC-315) has revealed that there are at least two types of contrasuppressor T cells whose major function is to modulate secretory differentiation, without affecting clone growth. These two cell types may be distinguished by comparing (1) the character of the suppressor cell counteracted, (2) the requirement for reexposure to the same or to crossreactive priming antigens, and (3) the presence or absence of binding specificity for the MOPC-315 paraprotein. INTRODUCTION:

THE MODULATION

OF SECHETORY

DIFFERENTIATION OF

M 0PC-315 Over the last several years we have been engaged in the dissection of the T cell-mediated control of the growth and differentiation of the TNP binding, IgA, A, plasmacytoma MOPC-315 (1-6). It has recently become appar‘By John D. Kemp (Department of Pathology, University of Iowa, Iowa City, Iowa 52242) and James W. Rohrer (Department of Microbiology and Immunology, University of South Alabama, Mobile, Alabama 36688). This work was supported by USPHS research Grant CA-28708 (J. W. R.),

CONTRASUPPRESSION

327

ent to us that the phenomenon of contrasuppression (7,8) is inextricably involved in the production of carrier immunization induced helper cell effects in vivo and we wish to focus here on cases that demonstrate the fundamental heterogeneity of the phenomenon. In particular, we have known for some time that immunization with a low dose of carrier antigen (4 x lo6 erythrocytes) results in carrier specific suppression of both growth and differentiation of the tumor (1).This can be seen either in situ in the immunized animals or when the immune T cells are adoptively transferred to a recipient who is then implanted with a diffusion chamber containing the target tumor cells and the appropriate hapten-carrier conjugate. After 7 days the chambers are retrieved and clone growth and secretory behavior assessed. The same studies also indicated that immunization with a high dose of carrier antigen (4 x lo8 erythrocytes) resulted in carrier-specific enhancement of the growth and differentiation of the tumor under the same experimental conditions of transfer, chamber implantation, and assay. We have also known for some time that the low dose immune suppressor cells could be “converted” into helper cells every bit as effective as the high dose immune cells if the donor animals were treated with corticosteroids or if the transferred cells were irradiated (4).We therefore reasoned that the suppressor T cells in the low dose immunization were somehow dominant over a group of helpers that were also present. We considered it paradoxical therefore when equivalent mixtures of helpers and suppressors failed to show clearcut suppressor dominance. We eventually considered the possibility that the presence of contrasuppressor cells might serve to distinguish the high dose immune helper population from the low dose immune suppressor populations. We were pleased to confirm that the V. villosa lectin adherence procedure published by Green et al. (8) allowed us to enrich for cells with apparent contrasuppressive effects from the high dose immune populations while simultaneously depleting cells with antigen-specific helper effects. This technical maneuver is critical to the performance of the experiments presented. As a further preface to the data presented about contrasuppressor cells we should point out that there are two distinct mechanisms by which low dose carrier immune cells suppress the secretory differentiation of the MOPC-315 cells. Specifically, it can be shown that cells which adhere to 315 protein-coated plates (but not to 460 protein-coated control plates) produce suppression by one mechanism while cells that do not adhere to 315 plates produce suppression by another mechanism (6). In particular, the former cells do not require a hapten-carrier bridge (but do require carrier reexposure) while the latter cells require not only carrier reexposure but also the presence of the appropriate hapten-carrier bridge. With these facts in mind, we are now nearly prepared to address the fact that high dose immune cell

328

DOUGLAS R . GHEEN A N D HICHARD K . GERSHON

populations appear to contain distinct contrasuppressor cells for each of these two mechanisms of suppression of secretory differentiation of the MOPC-315 cells. Only one concept needs to be further explained: that of carrier crossreactivity. In an early control experiment for specificity we accidentally rediscovered what had already been noted by Gershon and colleagues about contrasuppression, that is, the frequently demonstrable characteristic of antigen crossreactivity (9). Thus, we have noted that one of the two contrasuppressors to be described below has a finite range of crossreactivity (demonstrable on antigen rechallenge), i.e., mammalian erythrocytes such as sheep or rabbit exhibit crossreactivity while avian erythrocytes such as chicken and turkey crossreact with each other but not with mammalian erythrocytes (10).This stands in contrast to the strict carrier specificity of the other contrasuppressor to be discussed and the similar strict specificities of the helper and suppressor cells.

RESULTS: DEFINING HETEROGENEOUS CONTRASUPPRESSOR T CELLS The data in Table I show, in summary form, the observations from several experiments in protocols similar to those described in several prior publications (1, 2, 4-6). Thus, high dose immune populations contain helpers that, when rechallenged with the priming antigen, will double the tumor PFC frequency as compared to controls (groups A and B). If high dose primed cells are rechallenged with a different, albeit crossreactive, antigen, no help is observed (group C). When cells from either high dose population are passed over plates coated with V. villosa lectin it can be seen that the specifically eluted cells show no helper or suppressor activity (groups B1 and Cl). Furthermore, when low dose primed populations are rechallenged with the same antigen, they suppress differentiation (group D). As noted in a prior report, these suppressors may be divided into 315 protein nonadherent and adherent sets (6) (groups E and F). Various mixtures of the above cell populations show that 315 protein nonadherent suppressors may be neutralized by a V. villosa adherent cell when the latter is rechallenged with a crossreactive antigen (crossreactive contrasuppression) (group E Cl). In contrast, 315 adherent suppressors are not neutralized by a crossreactive contrasuppressor (group F + Cl). They are, however, neutralized by a V. villosa adherent immune cell when the latter is rechallenged with the specific priming antigen (specific contrasuppression) (group F B1). Note that the helper cell contained within the 315 adherent suppressor population is now revealed, and that the net result is the same as with the high dose immune cells. To conclude, we should say that a given high dose carrier immune population contains both types of contrasuppressors, and they may be tested for separately by the type of antigenic rechallenge. Although the data are not

+

+

329

CONTRASUPPRESSION

TABLE I HETEHOCENEOUS CONTRASUPPRESSOH T CELLSO

Immune T cell populations

MOPC-315 PFC rate (%)

A

Nonimmune controls

20 (2 1)

B

High dose priming (same antigen or rechallenge) High dose priming (crossreactive antigen on rechallenge)

40 (* 2)

C

20 (t 1)

B1

V. villosa adherent cells (VVA)

20 ( 2 1)

C'

(from group B) V. oillosa adherent cells (VVA) (from group C )

20 (* 1)

D E F

Low dose priming (same antigen on rechallenge) 315 protein nonadherent cells (from group D) 315 protein adherent cells (from group D)

E+C' 315 nonadhereiit suppressors and crossreactive contrasuppressor(s) F+CI 315 adherent suppressors and crossreactive contrasuppressor(s) F+B' 315 adherent suppressors and specific contrasuppressor(s)

Comment Baseline control; spontaneous tumor differentiation rate Augmentation by a specific helper N o augmentation; no specific helper

No effect. W A cells do not help or suppress Same as B1 when tested alone

5 (* 1)

Suppression

5 (* 1)

Suppression: mechanism is distinct from F6 Suppression: mechanism is distinct from Eb

5 (* 1)

20 (* 1)

Return to baseline: help is not seen

5 (* 1)

Adherent suppressors are unaffected

40 (22)

Adherent suppressors are neutralized: help in 315 adherent cells is revealed

a Immune cell populations were produced by priming mice with either 4 x lo8 (high dose) or 4 x 106 (low dose) erythrocytes. TNP-conjugated erythrocytes were placed in diffusion chambers along with MOPC-315 tumor cells and the chambers implated in mice which had been adoptive recipients of various combinations of immune cells (1,2,4-6). Chambers were retrieved after 7 days and the PFC percentage formation was determined. 6 315 nonadherent suppressors require a hapten-carrier bridge in order to function, whereas 315 adherent suppressors only require carrier reexposure (6).

discussed here, we have also determined that the crossreactive contrasuppressor adheres to a 315 protein-coated dish, while the specific contrasuppressor does not (Rohrer et n l . , manuscript in preparation). This is the exact opposite of the binding specificities displayed by their corresponding suppressor cells.

330

D O U G L A S R. G R E E N A N D RICHARD K . G E H S H O N

SUMMARY

The data discussed here touch upon several issues in the evolving story of T cell contrasuppression, the underlying theme being that of heterogeneity. First, there is the issue of function. We are considering here only those cells that affect the function of secretory differentiation. We have evidence that different contrasuppressor cells exist for clone growth, but have not yet studied them in the same depth as those for secretory differentiation. Second, there is the important issue of target cells. In this article by Green and Gershon it is pointed out that there is clear evidence that contrasuppressor effects can work by protecting helper cells from suppressor cell effects in vitro. On the other hand, direct additional inhibition of the suppressor cells themselves has not been excluded. The latter point is also true in our system. However, we must suppose for the sake of simplicity in many of our experiments that if suppressors are not the target of the contrasuppressor effects then the B cells themselves probably are. This is because the tumor cells engage in a spontaneous rate of growth and differentiation in the absence of help or suppression. When T cell-dependent, specifically triggered effects reduce this spontaneous behavior, then a suppressive effect must have been delivered directly to the B cells. This is a simplifying assumption which is attractive, but since the experiments are carried out in vivo and thus may be affected by some factors that we have not yet recognized, we are not confident on its “intuitive” appeal. A third issue revolves around triggering specificity. One of our contrasuppressors exhibits the phenomenon of carrier crossreactivity (CRCS) and is thus behaving in accord with expectations aroused by Green and Gershon in this review. The other cell is apparently quite carrier specific (SCS). The meaning of this is not at all clear, but its potential significance may somehow be related to a sort of “mirror image” relationship of the two cells. Thus, for example, in other experiments not discussed here, we have noted that the CRCS binds to 315 protein-coated plates (lo), but as noted here counteracts a suppressive effect which is generated by cells which do not adhere to these plates. In contrast to SCS does not bind to 315 plates and yet, as noted here, appears to counteract a suppressor effect generated by cells which do adhere to 315 plates. Moreover, the effects of the CRCS are partially mimicked by aLyb3 antiserum (11, 12) and are not detectable in anti-mu suppressed mice (13). The latter point is in agreement with studies by Green, Janeway, and Gershon. The SCS does not appear to exhibit either of these characteristics. We believe that our studies support several of the major prior observations about the phenomenon of T cell contrasuppression. We submit that they may also represent an early indication of potential complexity and

CONTHASUPPRESSION

33 1

heterogeneity. Thus, as with the study of helper and suppressor T cell subsets and circuitry, things may become more complex.

REFERENCESTO ADDENDUM1 1. 2. 3. 4. 5.

6. 7. 8. 9. 10. 11. 12. 13.

Rohrer, J. W., and Lynch, R. G. (1977)./. Immunol. 119, 2045. Rohrer, J. W., and Lynch, R. 6. (1978).J. Immunol. 121, 1066. Rohrer, J. W., and Lynch, R. 6. (1979).1. Immunol. 123, 1083. Lynch, R. G., Hohrer, J. W., Odermatt, B., Gebel, H. D., Autry, J. R., and Hoover, R. G . (1979). Immunol. Rec. 48, 45. Rohrer, J. W., Gershon, R. K., Lynch, R. G . , and Kemp, J. D. (1984). J. Mol. Cell. Immunol., in press. Rohrer, J. W., Gershon, R. K., and Kemp, J . D. (1982). In “B and T Cell Tumors’’ (E. S. Vitetta and C. F. Fox, eds.), p. 273. Academic Press, New York. Gershon, R. K., Eardley, D. D . , Durum, S., Green, D. R., Shen, F. W., Yamauchi, K., Cantor, H., and Murphy, D. B. (1981).J. E x p . Med. 153, 1533. Green, D. R . , Eardley, D. D., Kimura, A , , Murphy, D. B., Yamauchi, K., and Gershon, R. K. (1981). Eur. 1. Immvnol. 11, 973. Green, D. R . , Flood, P. M . , and Gershon, R. K. (1983). Annu. Reu. Immunol. VI, 439. Rohrer, J. W., and Kenip, J. D. (1984). In preparation. Kemp, J. D., Rohrer, J. W., and Huber, B. T. (1982). Inimunol. Reu. 69, 127. Rohrer, J. W., Huber, B. T., Cone, R. E., Gershon, R. K., and Kemp, J. D. (1984). Submitted for publication. Rohrer, J. W., and Janeway, C. A. (1984). In preparation.

Addendum 2: Relation of the Allogeneic Effect to Contrasuppression‘

The term “allogeneic effect” was coined by Katz et d.(1) to describe a phenomenon that was observed fortuitously during experiments aimed at defining certain characteristics of carrier-specific T helper (Th) cells and their function in hapten-specific antibody responses. In these experiments, it was found that an appropriately timed transfer of allogeneic lymphoid cells into hapten-carrier-primed guinea pigs bypassed the need for carrier-specific Th cells in the secondary response to the same hapten. This was indicated by the fact that an augmented secondary hapten-specific antibody response was elicited in such guinea pigs when challenged with a hapten-heterologous carrier conjugate (i.e., carrier different from the one used for priming). This phenomenon which has since been studied in different experimental systems was found to reflect the development of a T cell-dependent graftversus-host reaction (reviewed in 2, 3). Although one usually tends to associated the allogeneic effect with stimulatory or positive influences, it is impor‘By Amnon Altman and David H . Katz (Medical Biology Institute, La Jolla, California). This is publication number 23 of that Institute. This work was supported by Grants AI-19476 and CA-35299 from the United States Public Health Service.

332

DOUGLAS R. GREEN A N D RICHARD K . GERSHON

tant to note that allogeneic effects cover the entire range of immunoregulatory influences and that, under appropriate conditions, they can exert profound suppression (2). The allogeneic effect concept was also extended to in vitro systems where it was demonstrated that allogeneic T cells or soluble products (i.e., lymphokines) from allogeneic mixed leukocyte cultures (MLC) can augment antibody and cytotoxic T lymphocyte (CTL) responses in culture (reviewed in 3). The possibility that a positive allogeneic effect is mediated, at least in part, via the induction of contrasuppressor T (Tcs) cells merits consideration since the end result of both activities is similar, i.e., reconstitution (or augmentation) of deficient or suboptimal immune responses. One of the basic tenets of the Tcs concept is that these cells augment immune responses by functionally eliminating suppressor T cell (Ts)-mediated suppression. More specifically, Tcs confers upon Th cells resistance to suppressor signals (4,5). According to this concept, therefore, Tcs cells will not be functionally detectable in systems lacking Ts cells. This has been shown in in vitro assays for the induction of anti-sheep red blood cells (SRBC) antibody responses (4, 5). In a typical assay, isolated subsets of T cells (selected on the basis of their Lyt surface markers) are tested for tbeir ability to reverse suppression of the antibody response mediated by antigenprimed, Lyt-2+ Ts populations. The effector Tcs cell detected in these assays is a Lyt-l+ T cell which can be separated from Lyt-l+ Th cells by several criteria (5). In those systems in which the allogeneic effect has been shown to augment immune responses, the mechanism(s) have not been elucidated in enough detail to allow us to conclude what subsets of T cells and cellular interactions are involved. Thus, in order to formally establish a role for Tcs cells in positive allogeneic effects, it will be necessary to carry out experiments with purified, well-defined T cell subsets and to ascertain whether the mechanism responsible for augmentation of a given response meets the criteria used to define the Tcs circuit. In the absence of direct proof, considerations of the involvement of Tcs cells in the allogeneic effect are speculative. However, certain similarities (as well as differences) between contrasuppression and positive allogeneic effects can be pointed out. There is indirect evidence to suggest that positive allogeneic effects may regulate immune responses in situations where Ts cells are likely to be the cause of the observed hyporeactivity. Perhaps the most relevant studies in this context are those aimed at studying the influence of the allogeneic effect on tumor-specific cell-mediated immune responses in vivo or in vitro. The dominant induction of Ts cells by tumorspecific antigens is one of the prevalent hypotheses trying to explain the failure of the immune system to inhibit the growth of autochthonous tumors.

CONTHASUPPRESSION

333

Tumor-specific Ts cells were shown to function in several experimental tumor systems (reviewed in 6). Allogeneic T cells or soluble products of an allogeneic MLC were shown to augment tumor-specific CTL responses in oitro (7-9)as well as to retard the growth of syngeneic tumors in cioo (10-13).We ourselves have shown that a cellular allogeneic effect can overcome suppression mediated by thymocytes of tumor-bearing mice in a system measuring the induction of fumor-specific CTL directed against the DBAl2-derived T cell leukemia line, L5178Y (Altman and Katz, unpublished observations). Ts cells were detected by others in the thymus of tumor-bearing mice, using similar assay systems (14,15). Thus, Tcs cells may operate in these systems although definite proof is still missing. In trying to compare (or contrast) the allogeneic effect with the Tcs concept, one has to consider the cellular targets for these phenomena. As pointed out before (4,5) the cellular target of the Tcs circuit is the Th cells which acquire resistance to suppressor signals. In the case of allogeneic effects, either T cells or B cells have been considered to constitute the target cell populations. In the case of in oitro induced CTL responses the target appears to be a T cell inasmuch as Th cells or CTL (and their precursors) participate in the response. In antibody responses the allogeneic effect could act on either the Th cell or the antibody-producing B cells. In fact, compelling evidence exists to support the direct action of the allogeneic effect on B lymphocytes. Katz et u1. demonstrated the ability of cellular allogeneic effects to convert normally tolerogenic signals provided by D N P conjugates of the copolymer D-ghtamic acid and D-lySine (DNP-DGL) into immunogeneic signals as evidenced by the development of potent antibody responses in guinea pigs or mice (reviewed in 2, 3). For reasons outlined earlier (16)very few or no D-GL-specific Th cells are likely to exist and it is a fair assumption, therefore, that under these conditions the allogeneic effect acts directly on B target cells. A more recent study also identified B lymphocytes as the cellular targets for the allogeneic effect (17). In this study it was shown, by using carrier-primed Th cells and haptenprimed B cells from partially histoincompatible congenic murine strains, that an allogeneic effect was detectable only when the T cells inducing the effect were allogeneic to the hapten-primed B cells but not when they were allogeneic to the carrier-primed Th cells (17).Thus, it appears unlikely that Tcs are involved in the allogeneic effects observed in these studies; however, this does not mean that Tcs could not participate in allogeneic effects studied in other systems. Although a positive allogeneic effect could be due to the induction of Tcs cells, thus resulting in functional inactivation of suppression, it could equally well act by expanding Th cells and thus tipping the balance between Ts and

334

DOUGLAS R . G R E E N A N D HICHARD K . GEHSHON

Th cells in favor of the latter. In this context, it is noteworthy that allogeneic cell interactions result in the production of various immunoregulatory lymphokines such as allogeneic effect factor (AEF) (reviewed in 3), T cell growth factor (TCGF), T cell-replacing factor (TRF), and others. These lymphokines provide differentiation and/or growth signals to cells of the immune system and they could act on either T or B cellular targets. It is not unlikely that some of the T cell-tropic lymphokines are “class-specific” in terms of their ability to activate preferentially T cell subsets of distinct functional attributes, i.e., Th, Ts, or Tcs cells. One major obstacle to the idea that allogeneic effects are associated with the induction of Tcs cells is the fact that in many, if not all, cases cellular allogeneic effects or their soluble products actually replace the normal requirement for Th cells. In contrast, demonstration of contrasuppression requires, by definition, Th cells (as well as Ts cells) since the Th cell is the cellular target of contrasuppression. Thus, according to the Tcs concept, it is expected that no contrasuppression will be demonstrable in systems lacking Th cells. This apparent contradiction is difficult to reconcile at present. In summary, the allogeneic effect is a complex phenomenon which may include the participation of different, independent cellular mechanisms. While contrasuppression is not a likely participant in some forms of allogeneic effects, it may be operational in others. However, formal proof for the participation of Tcs in allogeneic effects will require tedious and careful experimentation, using methods which were applied in the original discovery of the Tcs circuit. In the absence of direct evidence and in the face of some apparent similarities and differences between these two phenomena, the relation of the allogeneic effect to contrasuppression is an intriguing concept which merits detailed investigation. REFERENCES TO ADDENDUM2 1. Katz, D. H . , Pad, W. E . , Goidl, E. A., and Benacerraf, B. (1971).J. E x p . Med. 133, 169. 2. Katz, D. H . (1972). Transplant Reo. 12, 141. 3. Katz, 1).H.(1977). “Lymphocyte Differentiation, Recognition and Regulation.” Academic Press, New York. 4. Gershon, R . K . , Eardley, D. D., Duruni, S. K . , Green, D. R . , Shen, F. W . , Yamauchi, K . , Cantor, H . , and Murphy, I). B. (1981). J. E x p . Med. 153, 1533. 5. Green, D.R . , Eardley, D. D . , Kiinura, A . , Murphy, D. B . , Yamauchi, K . , and Gershon, R . K. (1981). E u r . J. Ztnmunol. 11, 973. 6. Naor, D. (1979). Ado. Cancer Res. 29, 45. 7. Sondel, P. M . , O’Brien, C . , Porter, L., Schlossman, S . T., and Chess, L. (1976). J . Iminitnol. 117, 2197. 8. Lee, S. K . , and Oliver, R . T. I). (1978). J E x p . M e d . 147, 912. 9. Fyfe. D. A , , and Finke, J. H . (1979). J . Itnmunol. 122, 1156. 10. Katz, D. H . , Ellman, L., Paul, W. E . , Green, I . , and Benacerraf, B. (1972). Cancer Res. 32, 133.

CONTRASUPPRESSION

335

11. Carnaud, C . , Markowicz, O . , and Trainin, N. (1974). Cell. Zminunol. 14, 87. 12. Osborne, D. P . , and Katz, D. H. (1977). J . Ztnmunol. 118, 1441. 13. Weiden, P. L., Sullivan, K . M . , Flournoy, N . , Storb, R . , andThomas, E. D. (1981). New Engl. J. Med. 304, 1529. 14. Fujiinoto, S., Greene, M. I . , and Sehon, A. H . (1976). J. Znimunol. 116, 791. 15. Takei, F., Levy, J. G . , and Kilbrun, D. G. (1977). J . Zmmunol. 118, 412. 16. Katz, D. H . , Paul, W. E . , and Benacerraf, B. (1972). J. Zmmunol. 110, 107. 17. Golding, H . , and Rittenberg, M. B. (1980). J. Zmmunol. 124, 1284.

This Page Intentionally Left Blank

A

Cell cycle, induced differentiation in erythroleukemia cells and, 160-162 Cell-mediated immunity, antigen-specific contrasuppressor factor in, 284 Cell surface, minimal changes in, biological models for, 32-36 Chromosomes, assignment of neoplasia to,

Adjuvants, contrasuppressor cells and, 30030 1 Age, immune regulation in, contrasuppressor circuit and, 310 Alloreactivity, immunological surveillance 1970 and, 4-5 Antigen dose, contrasuppressor cells and,

218-220

296-297 Antigenic determinants, special, contrasuppressor cells and, 293-295 Antigen presenting cells, contrasuppressor cell and, 289-293 Antigen profile, correlation with morphology, classification of B cell lymphomas and, 80-117 Autoimmunity, contrasuppressor circuit and,

310-312

B B cell, lymphomas and classifcation by immunophenotype, 76-79 by morphology and correlation of morphological types with antigen profile, 80-117

Contrasuppression allogeneic effect of, 331-334 definition, 282 future and, 320-321 human examples of, 318-320 tumor immunity and class I antigens in, 315-317 enhancement of lymphoid tumor development, 317-318 further evidence for, 314-315 theoretical considerations, 312 in tumor resistance, 313 in tumor therapy, 313-314 Contrasuppressor cells, conditions influencing generation andlor activation of adjuvants, 300-301 antigen dose, 296-297 crossreactivity of Ly-2 inducer cell, 298-

300 genetics, 297 level 2 suppressor cells, 297-298 ontogeny, 286-288 role of antigen presenting cell, 289-293 special antigenic determinants, 293-295 Contrasuppressor circuit defining a specific circuit and assigning a unique phenotype to its cellular and molecular members cellular interactions in cell-mediated immunity, 284 effector cells, 283-284 inducer cells, 282-283 transducer cells, 283

C Cancer, see also Leukemia; Neoplasia, Tumors; etc. susceptibility to assignment to oncogenes and regulatory genes, 220-228 MHC phenotype, 50-51 Cell(s) neoplastically transformed, occurrence in wild populations, 199-204 Cell circuits, definition cellular make-up, 281 techniques, 280-281

331

338

INDEX

functional activity of, 285-286 immunological consequences of activation of adoptive transfer of immune response, 303-305 autoimmunity, 310-312 hyperimmunity, 301-303 immune regulation in old age, 310 immune response to malaria, 308-310 microenvironmental immune regulation, 305-308 recovery from trauma associated suppression, 310 c-src oncogene, in the animal kingdom, 207-21 1

D Differentiation gene, Dffidependent characters, 252-256 general features, 251-252 modified tRNAS in Dijf-dependent differentiation, 256-261

E Effector cell, contrasuppressor circuit and, 283-284 Erythroleukemia cells differentiation cell cycle and induced differentiation, 160-162 globin gene replication, 162 gene expression, DNA structure and chro. matin structure during induced differentiation, 155-156 murine globin gene domains, 156 posttranscriptional regulation of globin gene expression, 157 role of chromatin structure in globin gene regulation, 157-160 transcriptional regulation of globin genes during development and differentiation, 156-157 terminal cell division change for which cells retain a memory, 153-154 characteristics of inducer-mediated differentiation, 151-152

commitment to, 152 multistep process and, 152-153 protein p 53 and, 154-155

G Genes, Class I, recognition and polymorphism, 17-19 Genetics contrasuppressor cells and, 297 Globin genes regulation in murine erythroleukemia cells domains, 156 posttranscriptional, 157 role of chromatin structure, 157-160 transcriptional, 156-157 replication in erythroleukemia cells, 162 Clycoproteins, Class 11, posttranslational associations and diversity, 19-21

H Host response, T cell surveillance and, 8-10 Hybrids, susceptibility to development of neoplasia, 211-213 Hyperimmunity, contrasuppressor circuit and, 301-303 Hyperplastic alveolar nodule, is it protoneoplastic?, 185-186 Hyperplastic outgrowth lines, of murine mammary tumors characterization of, 171-172 transplantable, development of, 170-171

I Immune regulation microenvironmental, contrasuppressor circuit and, 305-308 in old age, contrasuppressor circuit and, 310 Immune response adoptive transfer, contrasuppressor circuit and, 303-305 to malaria, contrasuppressor circuit and, 308-310 Immunological surveillance 1970, 3 alloreactivity and, 4-5

339

INDEX

criticisms of, 5 ininiunopotentiation and T cell subsets, 7-8 nude mice and natural surveillance, 6-7 early ideas about role of MHC and, 3-4 Immunophenotype, classification of B cell lymphomas and, 76-79 Immunosuppression, T cell surveillance and, 10-12 Inducer cell contrasuppressor circuit and, 282-283 crossreactivity of the Ly-2 contrasuppressor, 298-300 Zr genes, T cell repertoire and, 13-16

L Leukemia, see also Cancer; Neoplasia, Tumors; etc. chronic lymphocytic of B cell type, 80-85 chronic lymphocytic of T type, 129-131 hairy cell, 87-89 prolymphocytic, of B cell type, 85-86 prolymphocytic of T type, 131 T lymphoblastic, 128-129 Lymphoid tumors, development, cotitrasupression in enhancement of, 317318 Lymphoma(s) B cell, 73-76 classification by immunophenotype, 76-79 by morphology and correlation of morphological types with antigen profile, 80-117 with morphology of true histiocytic sarcoma, 106-108 centroblastic, imrnunoblastic lymphoma and, 102-106 centroblastic-centrocytic, 91-98 centrocytic, 98- 102 cutaneous T cell, mycosis fungoides and, 131-133 immunoblastic, centroblastic lymphoma and, 102-106 lymphoblastic, 108-117 lymphoplasmacytic/cytoid, 89-91 malignant distinction from other neoplasms, 69-71

division into Hodgkin’s, non-Hodgkin’s and true histiocyte sarcoma, 71-73 multilobated, of B cell type, 106 of plasmacytoid T cells, 140 pleomorphic T cell, 135-139 T cell, 117-120 of clear cell type, 140 clinically and morphologically defined, correlation with antigen profile, 120 peripheral T cell lymphoinas/leukemias, 127-140 prethymic and thymic (TdT-positive) lyrnphoblastic lymphonia/leukernia, 120-127 T ininiunoblastic, 139-140 T lymphoblastic, 128-129 T zone, 133-135

M Major histocompatibility complex determinants, T cell specificity for, 12-13 early ideas about role of, immunological surveillance and, 3-4 expression of antigens on tumor cells alien Class I, 44-45 Class I antigens and tumorigenicity, 36-42 cross-reactions with TSTA and antigens on tumor cells, 45-47 expression of Class I1 antigens on tumor cells, 47-50 mutations in Class I genes and T cell surveillance, 42-44 molecular nature of, 16 Class I genes: recognition and polymorphism, 17-19 Class 11 glycoproteins: posttranslational associations and diversity, 19-21 phenotype, susceptibility to cancer and, 50-51 Malaria, immune response to, contrasuppressor circuit and, 308-310 Mammary neoplasia, molecular biology of hyperplasias, 180-181 tumors, 178-180 Mammary tumors characterization of hyperplastic outgrowth lines, 171-172

340

INDEX

development of transplantable hyperplastic outgrowth lines, 170-171 murine, origin and evolution of, 181-185 restriction endonuclease mapping and general comments, 172-175 MuMTV, 175-178 Mammary tumor system, of mice, 168-170 Mammary tumorigenesis, role of MuMTV, 186-187 Murine mammary tumor virus, role in tumorigenesis, 186-187 Mycosis fungoides, cutaneous T cell lymphoma and, 131-133

N Neoplasia, see also Cancer; Leukemia; Tumors; etc. hybridization as a step toward, 213-218 insusceptibility of animals in wild populations to development of, 197-199 theoretical considerations on a general concept of common basis in different tissues, 261262 common basis in metazoa, 261 common basis of tumor etiology, 262 unified view of etiology, 263-268 Nude mice, natural surveillance and, 6-7

0 Oncogenes defective control in hybrids assignment of cancer susceptibility to oncogenes, 220-228 assignment of neoplasia to chromosomes, 218-220 hybridization as a step toward neoplasia, 213-218 susceptibility of hybrids to development of neoplasia, 211-213 tissue specificity or nonspecificity of oncogenes and regulatory genes, 228230 regulatory genes and compartment-specific regulatory genes, 249-251 differentiation gene, 25-261

oncogene, 230-242 pigment cell-specific regulatory gene, 242-249 retroviral, cellular homologs in wild populations, 207 ubiquity in purebred animals derived from wild populations biology and taxonomy of Xiphophorus, 194-197 cellular homologs of retroviral oncogenes in animals from wild populations, 204-207 competence for neoplastic transforrnation, 204-206 c-src in the animal kingdom, 207-211 in susceptibility of wild populations to neoplasia, 197-199 occurrence of neoplastically transformed cells in wild populations, 199-204 in Xiphophorus, historical background, 191-194 Ontogeny, contrasuppressor cells and, 286288

P Pigment cell, regulatory genes and, 242-249 Protein p53, erythroleukemia cell terminal division and, 154-155

R Restriction endonuclease, mapping of MuMTV, 175-178

S Suppression, trauma associated, recovery from, 310 Suppressor cells, level 2, contrasuppressor cells and, 297-298

T T cell contrasuppressor, heterogeneity of function and specifcity, 326-331 effectors and tumor-specific transplantation antigens, 21-22

34 1

INDEX

nature of tumor-specific transplantation antigens, 27-31 specificity for viruses, 22-27 importance of surveillance immunosuppression and transplantation, 10-12 viruses, tumors and host reponse, 8-10 lymphomas, 117-120 clinically and morphologically defined, correlation with antigen profile, 120 peripheral T cell lymphomas/leukemias, 127-140 prethymic and thymic (TdT-positive) lymphoblastic lymphomalleukemia, 120-127 MHC-restricted recognition repertoire and Zr genes, 13-16 specificity for MHC determinants, 1213 responses to SV 40 TSTA and single minor H antigens, 32-36 subsets, immunopotentiation and, 7-8 surveillance, general concepts, 51-53 Tissues, specificity or nonspecificity of oncogenes and regulatory genes, 228-230 Transducer cell, contrasuppressor circuit and, 283 Transformation, neoplastic, competence for, 204-206 Transplantation, T cell surveillance and, 1012 Transplantation antigens historical aspects and early speculations concerning, 2-3 alloreactivity and immunological surveillance 1970, 4-5 early ideas about role of MHC and immunological surveillance, 3-4 immunological surveillance 1970, 3 tumor-specific, nature of, 27-31

Tu gene, completely deregulated oncogenic effect in pigment cell system, 232-234 c-src and, 236-242 dosage effect, 235-236 indispensable and accessory copies of, 234-235 general features of, 230-232 Tumor(s), see also Cancer; Leukemia, Neoplasia; etc. T cell surveillance and, 8-10 Tumor cells, expression of MHC antigens on alien Class I, 44-45 Class I MHC antigens and tumorigenicity, 36-42 cross-reactions with TSTA Class I and alloantigens, 45-47 expression of Class I1 MHC antigens on tumor cells, 47-50 mutations in Class I MHC genes and T cell surveillance, 42-44 Tumor immunity, contrasupression and class I antigens in, 315-317 enhancement of lymphoid tumor development, 317-318 further evidence for, 314-315 theoretical considerations, 312 in tumor resistance, 313 in tumor therapy, 313-314

v Viruses T cell specificity for, 22-27 T cell surveillance and, 8-10

X Xiphophorus, biology and taxonomy of, 194197

This Page Intentionally Left Blank

CONTENTS OF PREVIOUS VOLUMES

Volume 1

Ionizing Hatliations and Cancer

Electronic Configuration and Carcinogetiesia C. A . CiJU~SOIl Epidermal Carcinogenesis E . V. Catcdry The Milk Agent i n the Origin of Maminary Tumors in Mice 1.. Dmochotcski Hormonal Aspects of Experitnental Tu morigenesis

Survival and Preservation of Tiiniors in the Frozen State

Austin M . Rrrtes

jaiiies Craigie

Energy and Nitrogen h1etal)olisin in Cancer

Leonard 1). Fennitiger attd G . Burroughs Mider Some Aspects of the Clinical Use of Nitrogen Mustards

Culciri T. K l o p p

u t i t l Jeantie C. Rotelnun Genetic Studies in Experitiieiital Cancer L. w. Laic The Role of Viruses in the Production of Cancer C. Oberlitig and A t . Gtteriti Experimental Cancer Chemotherapy

T . U . Gardner Properties of the Agent of R C J INo. I~ 1 Sarcoma R. I . C . Harris Applications of Hatlioiwtopea to Studirs of Carcinogenesis and Tuinor Metahlisin Charles lleidelberger The Carcinogenic Aniiiioazo Dyes Jnnies A . Miller u t i d E/i;abet/z C . hfillrr The Chemistry of Cytotoxic Alkylating Agents M . C . J . Ross Nutrition in Relation to Cancer Albert Tantietil~uuni and llerhet? Silcerstotie Plasma Proteins in Cancer Richard]. W i n d e r

C . Chester Stock AC'TIIOH INI>EX-SUHJIC:T INl)t?X

Volume 3 Etiology of Lung Cancer

Richard Doll The Experimental I>evelopment and Metal)olistn of Thyroid Gland Tutnors

A L ' T I I O H INIIEX-SVHJK(:T I N I I E S

llarold P. hforris Electronic Structure and Carcinogenic Activity and Aromatic Molecules: New Ileveloprnents A . Pu/ltnan and B. P u h u n Some Aspects o f Carcinogenesis

Volume 2 The Reactions of Carcinogens with Macromolecules

P . Rondoni

Peter Alexander

Pulmonary Tumors i n Experimental Animals

Chemical Constitutioti arid Carcinogenic Activity G . M . Badger Carcinogenesis and Tunior Pathogenesis

Michael B. Shimkin Oxidative Metabolism o f Neoplastic Tissues

Sidney Weinhouse

I . Berenb/uni

ALITIIOR INDEX-SUBJE(:T I N D E X

343

344

CONTENTS OF PREVIOUS VOLUMES

Volume 4 Advances in Chemotherapy of Cancer in Man Sidney Farber, Rudolf Toch, Edward Manning Sears, and Donald Pinkel The Use of Myleran and Similar Agents in Chronic Leukemias D. A. G. Galton The Employment of Methods of Inhibition Analysis in the Normal and TumorBearing Mammalian Organism Abraham Godin Some Recent Work on Tumor Immunity P. A . Corer Inductive Tissue Interaction in Development Clifford Grobstein Lipids in Cancer Frances L. Haven and W . R. Bloor The Relation between Carcinogenic Activity and the Physical and Chemical Properities of Angular Benzacridines A. Lacassagne, N . P. Buu Hot, R. Daudel, and F . Zajdela The Hormonal Genesis of Mammary Cancer 0. Miihlbock AUTllOR I N D E X - S U B J E C T I N D E X

Volume 5

of 2-Fluorenamine and Related Compounds Elizabeth K. Weisburger andJohn H . Weisburger AUTHOR I N D E X - S U B J E C T I N D E X

Volume 6 Blood Enzymes in Cancer and Other Diseases Oscar Boda nsk y The Plant Tumor Problem Armin C . Braun and Henry N . Wood Cancer Chemotherapy by Perfiusion Oscar Creech, J r . and Edward T. Krernentz Viral Etiology of Mouse Leukemia Ludwick Gross Radiation Chimeras P. C. Koller, A. J . S . Daoies, and Sheila M . A . Doak Etiology and Pathogenesis of Mouse Leukemia J . F . A. P . Miller Antagonists of Purine and Pyrimidine Metabolites and Folic Acid G . M . Tirnrnis Behavior of Liver Enzymes in Hepatocarcinogenesis George Weher AUTIIOH INDEX-SUBJE(:T I N D E X

Tumor-Host Relations R. W . Begg Primary Carcinoma of the Liver Charles Berman Protein Synthesis with Special Reference to Growth Processes both Normal and Abnormal P. N. Campbell The Newer Concept of Cancer Toxin War0 Nakahara and Fumiko Fukuoka Chemically Induced Tumors of Fowls P. R. Peacock Anemia in Cancer Vincent E . Price and Robert E . Greenfield Specific Tumor Antigens L. A. Zilber Chemistry, Carcinogenicity, and Metabolism

Volume 7 Avian Virus Growths and Their Etiologic Agents J . W . Beard Mechanisms of Resistance to Anticancer Agents R. W . Brockman Cross Resistance and Collateral Sensitivity Studies in Cancer Chemotherapy Dorris J . Hutchison Cytogenic Studies in Chronic Myeloid Leukemia W . M . Court Brown and lshbel M . Tough Ethionine Carcinogenesis Emmanuel Farber

CONTENTS OF PREVIOUS VOLUMES

Atmospheric Factors in Pathogenesis of Lung Cancer Paul Kotin and Hans L. Falk Progress with Some Tumor Viruses of Chickens and Mammals: The Problem of Passenger Viruses G . Negroni AUTllOR INDEX-SUBJECT INDEX

Volume 8 The Structure of Tumor Viruses and Its Bearing on Their Relation to Viruses in General A. F . Howatson Nuclear Proteins of Neoplastic Cells Harris Busch and William 1. Steele Nucleolar Chromosomes: Structures, Interactions, and Perspectives M . 1. Kopac and Gladys M . Mazeyko Carcinogenesis Related to Foods Contaminated by Processing and Fungal Metabolites H . F. Kraybill and M . B. Shimkin Experimental Tobacco Carcinogenesis Ernest L. Wynder and Dietrich Hoffman AUTllOH INDEX-SUBJECT INDEX

Volume 9 Urinary Enzymes and Their Diagnostic Value in Human Cancer Richard Stambaugh and Sidney Weinhouse The Relation of the Immune Reaction to Cancer Louis V . Caso Amino Acid Transport in Tumor Cells R. M . lohnstone and P. G . Scholefeld Studies on the Development, Biochemistry, and Biology of Experimental Hepatomas Harold P. Morris Biochemistry of Normal and Leukemic Leucocytes, Thrombocytes, and Bone Marrow Cells I . F . Seitz AUTHOR INDEX-SUBJECT I N D E X

345

Volume 10 Carcinogens, Enzyme Induction, and Gene Action H . V . Gelboin In Vitro Studies on Proteins Synthesis by Malignant Cells A. Clark G r i f f n The Enzymatic Pattern of Neoplastic Tissue W . Eugene Knox Carcinogenic Nitroso Compounds P. N . Magee a n d ] . M . Barnes The Sulfhydryl Group and Carcinogenesis I. S. Harrington The Treatment of Plasma Cell Myeloma Daniel E . Bergsagel, K . M . G r i f f t h , A. Haut, and W . 1. Stuckley, ]r. AUTHOR INDEX-SUBJECT I N D E X

Volume 11 The Carcinogenic Action and Metabolism or Urethran and N-Hydroxyurethran Sidney S . Miruish Runting Syndromes, Autoimmunity, and Neoplasia D. Keast Viral-Induced Enzymes and the Problem of Viral Oncogenesis Saul Kit The Growth-Regulating Activity of Polyanions: A Theoretical Discussion of Their Place in the Intercellular Environment and Their Role in Cell Physiology William Regelson Molecular Geometry and Carcinogenic Activity of Aromatic Compounds. New Perspectives Ioseph C . Arcos and Mary F . Argus CUMULATIVE I N D E X

Volume 12 Antigens Induced by the Mouse Leukemia Viruses G . Pasternak Immunological Aspects of Carcinogenesis by

346

CONTENTS OF PREVIOUS VOLUMES

Deoxyribonucleic Acid Tumor Viruses G . 1. Deichmun Replication of Oncogenic Viruses in VirusInduced Tumor Cells-Their Persistence and Interaction with Other Viruses H . Hanafusa Cellular Immunity against Tumor Antigens Karl Erik Hellstrom and lngegerd Hellstrom Perspectives in the Epidemiology of Leukemia lroing L. Kessler and Abraham M . Lilienfeld AUTIIOR INDEX-SUBJECT INDEX

The Investigation of Oncogenic Viral Genomes in Transformed Cells by Nucleic Acid Hybridization

Ernest Winocour Viral Genome and Oncogenic Transformation: Nuclear and Plasma Membrane Events

George Meyer Passive Immunotherapy of Leukemia and Other Cancer

Roland Motta Humoral Regulators in the Development and Progression of Leukemia

Donald Metcalf Complement and Tumor Immunology

Kusuya Nishioka

Volume 13 The Role of Immunoblasts in Host Resistance and Immunotherapy of Primary Sarcomata P . Alexander and /. G . Hall Evidence for the Viral Etiology of Leukemia in the Domestic Mammals

Oswald larrett The Function of the Delayed Sensitivity Reaction as Revealed in the Graft Reaction Culture Haim Ginsburg Epigenetic Processes and Their Relevance to the Study of Neoplasia Gajanan V . Sherbet The Characteristics of Animal Cells Transformed in Vitro lan Macpherson Role of Cell Association in Virus Infection and Virus Rescue J . Sooboda and I . Noianek Cancer of the Urinary Tract D . B . Cluyson and E . I€. Cooper Aspects of the EB Virus M . A. Epstein A U T H O R INDEX-SUBJECT I N D E X

Alpha-Fetoprotein in Ontogenesis and Its Association with Malignant Tumors G . 1. Abeleo Low Dose Radiation Cancers in Man

Alice Stewart AUTI IOH IN 1) EX-S U BJECT INDEX

Volume 15 Oncogenicity and Cell Transformation b y Papovavirus SV40: The Role of the Viral Genome 1. S. Butel. S . S . Teoethia, and J. L.

Melnick Nasopharyngeal Carcinoma (NPC)

J. H . C . H o Transcriptional Regulation in Eukaryotic Cells A . J . MacGillioray, 1. Paul, and G .

Threlfall Atypical Transfer RNA's and Their Origin i n Neoplastic Cells

Ernest Borek and Syloia]. Kerr Use of Genetic Markers to Study Cellular Origin and Development of Tumors in Human Females

Philip J . Fialkow Electron Spin Resonance Studies of Carcinogenesis

Volume 14 Active Immunotherapy

Georges Mathd

Harold M . Swartz Some Biochemical Aspects of the Relationship between the Tumor and the Host

V . S . Shapot

CONTENTS OF PREVIOUS VOLUMES

Nuclear Proteins and the Cell Cycle

Gary Stein and Renato Baserga

Mammary Neoplasia in Mice

S . Nandi and Charles M . McGrath

AUTIIOR I N D E X - S U B J E C T I N D E X

AUTHOR INDEX-SLIBJECT I N D E X

Volume 16

Volume 18

Polysaccharides in Cancer

Immunological Aspects of Chemical Carcinogenesis

Vijai N . Nigam and Antonio Cantero Antitumor Effects of Interferon

lon Gresser Transformation by Polyoma Virus and Simian Virus 40

Joe Sainbrook Molecular Repair, Wound Healing, and Carcinogenesis: Tumor Production a Possible Overhealing'?

Sir Alexander Haddou: The Expression of Normal Histocoinpatibility Antigens in Tumor Cells

Alena Lengerova 1,3-Bis(2-ChloroethyI)-l-Nitrosourea (BCNU) and Other Nitrosoureas i n Cancer Treatment: A Review

Stephen K. Carter, Frank M . Schabel, J r . , h w r e n c e E . Broder, and Thoinas P . Johnston AUTHOR I N D E X - S U B J E C T I N D E X

347

R. W . Baldwin Isozymes and Cancer

Fanny Schapira Physiological and Biochemical Reviews of Sex Differences and Carcinogenesis with Particular Reference to the Liver

Yee Chu Toh Immunodeficiency and Cancer

John H . Kersey, Beatrice D. Spector, and Robert A . Good Recent Observations Related to the Chemotherapy and Immunology of Gestational Choriocarcinoma

K . D. Bagshave Glycolipids of Tumor Cell Membrane

Sen-itiroh Hakomori Chemical Oncogenesis in Culture

Charles Heidelberger AUTIIOR I N D E X - S U B J E C T I N D E X

Volume 19 Volume 17 Comparative Aspects of Mammary Tumors Polysaccharides in Cancer: Clycoproteins and Glycolipids Vijai N. Nigam and Antonio Cantero Some Aspects of the Epidemiology and Etiology of Esophageal Cancer with Particular Emphasis on the Transkei, South Africa

Gerald P. Warwick and John S . Harington Genetic Control of Murine Viral Leukemogenesis

Frank Lilly and Theodore Pincus Mareks Disease: A Neoplastic Disease of Chickens Caused by a Herpesvirus

K. Nazerian Mutation and Human Cancer

Alfred G. Knudson, J r .

J . M . Hamilton The Cellular and Molecular Biology of RNA Tumor Viruses, Especially Avian Leukosis-Sarcoma Viruses, and Their Relatives

Howard M . Temin Cancer, Differentiation, and Embryonic Antigens: Some Central Problems J . H . Coggin, J r . and N . G. Anderson Simian Herpesviruses and Neoplasia

Fredrich W . Deinhardt, Lawrence A . Falk, and Lauren G . Wolfe Cell-Mediated Immunity to Tumor Cells

Ronald B. H e r b e m n Herpesviruses and Cancer

Fred Rapp Cyclic AMP and the Transformation of

348

CONTENTS OF PREVIOUS VOLUMES

Fibroblasts

Ira Pastan and George S . lohnson Tumor Angiogenesis

Judah F o l k m n SUBJECT INDEX

Cases Mutant, Mammalian Cells in Culture

G . B . Clements The Role of DNA Repair and Somatic Mutation in Carcinogenesis

James E . Trosko and Ernest H. Y. Chu SUBJECT INDEX

Volume 20 Tumor Cell Surfaces: General Alterations Detected by Agglutinins

Annette M . C . Rapin and M a x M . Burger Principles of Immunological Tolerance and Immunocyte Receptor Blockade

G . 1. V. Nossal The Role of Macrophages in Defense against Neoplastic Disease

Michael H . Levy and E . Frederick Wheelock Epoxides in Polycyclic Aromatic Hydrocarbon Metabolism and Carcinogenesis

P. Sims and P. L. Grover Virion and Tumor Cell Antigens of C-Type RNA Tumor Viruses

Heinz Bauer Addendum to “Molecular Repair, Wound Healing, and Carcinogenesis: Tumor Production a Possible Overhealing?’

Sir Alexander Haddow

Volume 22 Renal Carcinogenesis

1. M . Hamilton Toxicity of Antineoplastic Agents in Man: Chromosomal Aberrations, Antifertility Effects, Congenital Malformations, and Carcinogenic Potential Susan M . Sieber and Richard 11.

Adamson Interrelationships among RNA Tumor Viruses and Host Cells Raymond V. Gilden Proteolytic Enzymes, Cell Surface Changes, and Viral Transformation

Richard Roblin, Iih-Nan Chou, and Paul H . Black Immunodepression and Malignancy

Osias Stutinan SUBJECT INDEX

SUBIECT INDEX

Volume 23 Volume 21 Lung Tumors in Mice: Application to Carcinogenesis Bioassay

Michael B. Shimkin and Gary D. Stoner Cell Death in Normal and Malignant Tissues

E. H . Cooper, A . 1. Bedford, and T . E . Kenny The Histocompatibility-Linked Immune Response Genes

Baruj Benacerraf and David H. Katz Horizontally and Vertically Transmitted Oncornaviruses of Cats

M . Essex Epithelial Cells: Growth in Culture of Normal and Neoplastic Forms

Keef A . Rafferty, ]r. Selection of Biochemically Variant, in Some

The Genetic Aspects of Human Cancer W . E . Heston The Structure and Function of Intercellular Junctions in Cancer Ronald S. Weinstein, Frederick B. Merk,

andloseph Alroy Genetics of Adenoviruses

Harold S . Ginsberg and C . S . H . Young Molecular Biology of the Carcinogen, 4Nitroquinoline 1-Oxide

Minako Nagao and Takashi Sugimura Epstein-Barr Virus and Nonhuman Primates: Natural and Experimental Infection

A. Frank, W . A . Andiman, and G . Miller Tumor Progression and Homeostasis

Richmon; T. Prehn

CONTENTS OF PREVIOUS VOLUMES

Genetic Transformation of Animal Cells with Viral DNA or RNA Tumor Viruses

349

SUBJECT INDEX

Formation and Metabolism of Alkylated Nucleosides: Possible Role in Carcinogenesis by Nitroso Compounds and Alkylating Agents

Volume 24

Immunosuppression and the Rule of Suppressive Factors in Cancer

Miroslav Hill and Jana Hillooa

Anthony E. P e g

The Murine Sarcoma Virus-Induced Tumor: Exception or General Model in Tumor Immunology? J. P. Levy a n d ] . C . Leclerc Organization of the Genomes of Polyoma Virus and SV40 Mike Fried and Beverly E . Cr@n &-Microglobulin and the Major Histocompatibility Complex Per A . Peterson, Lars Rask, and Lars

Ostberg Chromosomal Abnormalities and Their Specificity in Human Neoplasms: An Assessment of Recent Observations by Banding Techniques

Joachim Mark Temperature-Sensitive Mutations in Animal Cells

Claudio Basilico Current Concepts of the Biology of Human Cutaneous Malignant Melanoma

Wallace H. Clark, Jr., Michael J. Mastrangelo, Ann M. Ainsworth, David Berd, Robert E . Bellet, and Evelina A . Bernardino SUBJECT INDEX

Volume 25 Biological Activity of Tumor Virus DNA

F. L. Graham Malignancy and Transformation: Expression in Somatic Cell Hybrids and Variants Harvey L. Ozer and Krishna K. Jha Tumor-Bound Immunoglobulins: In Situ Expressions of Humoral Immunity

Isaac P. Witz The A h Locus and the Metabolism of Chemical Carcinogens and Other Foreign Compounds Snorri S . Thorgeirsson and Daniel W .

Nebert

lsao Kamo and H e m n Friedman Passive Immunotherapy of Cancer in Animals and Man

Steoen A . Rosenberg and U'illiam D. Terry SUBJECT INDEX

Volume 26 The Epidemiology of Large-Bowel Cancer

Peloyo Correa and William Haenszel Interaction between Viral and Genetic Factors in Murine Mammary Cancer 1. Hilgers and P. Bentoelzen Inhibitors of Chemical Carcinogenesis

Lee W . Wattenberg Latent Characteristics of Selected Herpesviruses

Jack G . Stevens Antitumor Activity of Corynebacterium

parvum Luka Milas and Martin T . Scott SUBJECT INDEX

Volume 27 Translational Products of Type-C RNA Tumor Viruses John R. Stephenson. Sushikumar G. Deware, and Fred H . Reynolds, Jr. Quantitative Theories of Oncogenesis

Alice S. Whittemore Gestational Trophoblastic Disease: Origin of Choriocarcinoma, Invasive Mole and Choriocarcinoma Associated with Hydatidiform Mole, and Some Immunologic Aspects J. 1. Brewer, E. E. Torok, B . D. Kahan, C . R. Stanhope, and B . Holpern The Choice of Animal Tumors for Experimental Studies of Cancer Therapy Harold B. Hewitt

350

CONTENTS OF PREVIOUS V O L U M E S

Mass Spectrometry in Cancer Research

John Roboz Marrow Transplantation in the Treatnient of Acute Leukemia E . Donna// Thomas, C . Dean Buckner,

Alexander Feler, Paul E . Neiinan. and Rainer Storb Susceptibility of Human Population G r o u p to Colon Cancer

Martin Lipkin Natural Cell-Mediated Immunity

Ronald B. Herbernlon and Howard T . Holden SL'BJE(:T INDEX

Volume 28 Cancer: Somatic-Genetic Considerations

F . M . Burnet Tumors Arising in Organ Transplant Recipients

Suppressor Cells: Perinitters and Promoters of Malignancy'?

David Naor Retrodifferentiation and the Fetal I'atterns of Gene Expression in Cancer jos6 Uriel The Role of Glutathione and Glutathione STransferases in the Meta1)olisin of Chemical Carcinogens and Other Electrophilic Agents L. F. Chusseulitf a-Fetoprotein in Cancer and Fetal Developmeii t Erkki Ruos/ahti and Markktr S c p p d i i Maniinary Tumor Viruses Dan I f . Moore. Cut& A . IAJII~, Akhif H . Vaidya, Joel R . Sltcffiirltl. Ariioltl S . Dion. and Etieiinr Y, Imfargrtc~s Role of Selenium iii the Chernol)reveiition of Cancer

A. Clark Griffin SLIHJE(:T IN l>k:X

lsruel Penn Structure and Morphogenesis of Type-C Retroviruses

Ronald C . Montelaro mid l h n i P. Bolognesi BCG in Tumor Iminunotherapy Robert W . Baldtciii and Mulcoliii V . P i m i i t The Biology of Cancer Invasion antl Metastasis Isaiah I. Fidler, D C J I I ~h ~ f .U Gersteii. S uric1 lan R . Hurl Bovine Leukemia Virus Involvenient i n Enzootic Bovine Leiikosis A . Burny, F . Rex, t l . Chantrentie. Y. Cleuter, 1). Ilekegel, 1. Gh!ystlacl. R .

Volume 30 Acute Phase Reactant Proteins in Cancrr E . 11. Cooper antl Ioaii S t o w Induction of Leukemia in Mice I)? IrratliLition and Radiation Leukemia \'irus Variants Nechaitia llurana-Chera and Alpha Peletl On the Multiform Relationships Iwtween the Tumor and the Host

v. s. s h U l ) O t

Role of Hvclruine in Carciiiogenesis Ioseph Bulb Experimental Intestinal Cancer Kesearch Kettinann. M . Leclercy. I . I ~ ~ I I I I ~ I I , with Special Reference to Hunian M . Maminerickx, und 11. Portetelk Pdth0k)gy Molecular Mechanisms of Steroid Hormone Kazyinir hl. Pozharisski. Alesei 1. Action Likhnachec. Vuleri F . Kliit1ushccski. Stephen I. Higgins and Ulrich Gehring and J U C ( J ~D. Shuposhnikot: The Molecular Biology of Lyniphotropic SUHJECT I N D E X Herpesviruses

Volume 29 Influence of the Major Histocomi'dtibilit) Complex on T-cell Activation

J . F . A. P. Miller

Bill Sugden. Christopher R. Kintner. and Willie Mark Viral Xenogenization of Intact Tumor Cells Hiroshi KohayaShi Virus Augmentation of the Antigenicity of Tumor Cell Extracts

Faye C . Airstin and Clicirlc~sW. Boonu INDEX

Volume 31 The Epidemiology of Ixiikeinin Michael ‘4/derso n The Role of the Major Hist,)compatil)ilit\ C,ene Complex i n Murine Cytotoxic T Cell Ilesponses //ernionn \\’ugner, K1ari.s Pfi;ent~iuic~r.untl A4artin Riillinghoff The Seqiiential Analysis of Cancer I~eveloplnent hmu n i t r / F a rbor (i nd R m s Caitic~ron Genetic Control of Natural Cytotouicity ilnd Hyl)rid Ilesistance Edlcurd A . Clark unrl Richard C. llarnron 1)evelopment ol Human Breast CancelSqftoii R . M’el/iiigs

Tumor Cell Multiplication Michael 6. Totiey Ectopic Hormone Production Viewed as an Abnormality in Regulation of Gene Expression Hiroo lmirrci The Hole of Viruses i n Human Tnmors //arald zur IIairsen The Oncogenic Frinction of hlanimalian Sarcoma Viruses P o d Andersson Hecent Progress i n Research on Esophageal Cancer i n China Li Mingxin (Li Min-//sin). Li Ping, arid Li Baorottg iLi Pao-Jitng) Mass Transport in Tumors: Characterization and Applications to Chemothrrapy Rukesli K. l a i n , Jonas ,if. \Veiss/Jrod. cind James Wei INDEX

INI>I:X

Volume 34

Volume 32

The Transformation of Cell Growth and Transmogrification of DNA Synthesis h y Simian Virus 40 Robert 6. Martin Immunologic h.Iechanisins in UV Radiation Carcinogenesis Margaret L. Kripke The Tuinor Ilormant State E . Federick \YIwelock. Kent J . Weinhold. and Jctdith Letiich Marker Chromosome 14q- i n Human Cancer and Leukemia Felix Mite/tnuii Structural Diversity among Retroviral <;ene Products: A Molecular Approach to the Study of Biological Function through Structural Varialdity James W. Cuutsrli. John / I . Elcler. Fred C.Jensen, and Richard A . Ixrner Teratocdrcinolnas and Other Neoplasms as Ikvelopmental Defects in Gene Expression Beatrice Mintz mid Roger A . Fleischnian Immune Deficiency Predisposing to Epstein-Barr Virus-Induced Lymphoproliferative Diseases: The X-Linked

Tumor I’romoters and the hlecliaiiism of Tumor Promotion /A&I /Iiui~iond,T h t ~ ~ t l (C. i . ~ O’Brien. c i n d Willitiin M . Buird Shedding from the Cell Surface of‘ Norind and Cancer Cells Pair/ I / . Block Tumor Antigens on Neoplitslns Induced Chemical Carcinogens antl I)\ D N A antl RNA-Containing \‘iruses: Proptarties of the So\ul)iIizetl Antigens Lloyd W.Laic. Micliud J . Rogcjrs. c i n d Ettore A p / J d / U Nutrition antl Its Helntiollship to Cancer Bandarii S . R d d y , Lconard A . C o h c ~ . 6. D a d , Z I K i q . Peter //ill. J i ~ l i i iI / . Weis/iurgyr. crntl Ernst L. \&‘yncIvr Ilril>KX

Volume 33 The Cultivation of Animal Cells i n the Chemostat. Application to the Study of

352

CONTENTS OF PREVIOUS VOLUMES

Lymphoproliferative Syndrome as a Model

Daoid T . Purtilo

Normal B-Cell Differentiation and Maturation Processes

Tore Godal and Steinar Funderud

INDEX

Evolution in the Treatment Strategy of Hodgkin’s Disease

Volume 35

Epstein-Barr Virus Antigens-A to Modern Biochemistry

Gianni Bonadonna and Annando Santoro

Polyoma T Antigens

Walter Eckhart Transformation Induced by Herpes Simplex Virus: A Potentially Novel Type of Virus-Cell Interaction

Berge Hampar Arachidonic Acid Transformation and Tumor Production

Lawrence Leoine The Shope Papilloma-Carcinoma Complex of Rabbits: A Model System of Neoplastic Progression and Spontaneous Regression ]ohn W . Kreider and Gerald L. Bartlett Regulation of SV40 Gene Expression A dolf Graessnia n , M o n ika G raessinun n ,

and Christian Mueller Polyamines in Mammalian Tumors, Part I

Giuseppe Scalabrino and Maria E . Feriolo Criteria for Analyzing Interactions between Biologically Active Agents

Morris C . Berenbautn INDEX

Challenge

Daoid A . Thorley-Luwson, Clark M . E d son, and Kathi Geilinger INDEX

Volume 37 Retroviruses and Cancer Genes

1. Michael Bishop Cancer, Genes, and I>evelopnient: The Drosophila Case

Elisabeth Gateff Transformation-Associated Tunior Antigens

Arnold 1. Leoine Pericellular Matrix in Malignant Transformation

Kari Alitalo and Antti Vaheri Radiation Oncogenesis in Cell Culture

Carmia Borek Mhc Restriction and l r Genes

Jan Klein and Zoltan A . Nagy Phenotypic and Cytogenetic Characteristics of Human B-Lymphoid Cell Lines and Their Relevance for the Etiology of Burkitt’s Lymphoma

Kenneth Nilsson and George Klein

Volume 26 Polyamines in Mammalian Tumors, Part I1 Giuseppe Scalabrino and Maria 6. Ferioli Chromosome Abnormalities in Malignant Hematologic Diseases

lanet D. Rowley und]oseph R. Testu Oncogenes of Spontaneous and Chemically Induced Tumors

Roberi A . Weinberg Relationship of DNA Tertiary and Quaternary Structure of Carcinogenic Processes

Philip D. Lipetz, Alan G . Galsky, and Ralph E. Stephens Human 9-Cell Neoplasms in Relation to

Translocations Involving l g Locus-Carrying Chromosomes: A Model for Genetic Transposition in Carcinogenesis

George Klein and Gilbert Lenoir INDEX

Volume 38 The SJLlJ Spontaneous Reticulum Cell Sarcoma: New Insights in the Fields of Neoantigens, Host-Tumor Interactions, and Regulation of Tumor Growth

Benjamin Bonaoida The Initiation of DNA Excision-Repair

CONTENTS OF PREVIOUS VOLUMES

George W. Teebor and Krystyna Frenkel Steroid Hormone Receptors in Human Breast Cancer George W. Sledge, Jr, and Williain L. McGuire Relation between Steroid Metabolisni of the Host and Genesis of Cancers of the Breast, Uterine Cervix, and Endometrium Mitsuo Kodanlo and Toshiko Kodum Fundamentals of Chemotherapy of Myeloid Leukemia by Induction of Leukemia Cell Differentiation Motoo Hozumi The in Vitro Generation of Effector Lymphocytes and Their Employment in Tumor Immunotherapy Eli Kedar and David W. Weiss Cell Surface Glycolipids and. Glycoproteins in Malignant Transformation G. Yogeeswaran INDEX

Volume 39 Neoplastic Development in Airway Epithelium P. Nettesheim and A. Marchok Concomitant Tumor Immunity and the Resistance to a Second Tumor Challenge E. Gorelik Antigenic Tumor Cell Variants Obtained with Mutagens Thierry Boon Chromosomes and Cancer in the Mouse: Studies in Tumors, Established Cell Lines, and Cell Hybrids Dorothy A. Miller and Orlando J . Miller Polyomarvirus: An Overview of Its Unique Properties Beverly E. Griffin and Stephen M. Dilworth The Pathogenesis of Oncogenic Avian Retroviruses Paula 1. Enrietto andJohn A . Wyke Adjuvant chemotherapy for Common Solid Tumors David A. Berstock and Michael Baum INDEX

353

Volume 40 5-Methylcytosine, Gene Regulation, and Cancer Arthur D. Riggs and Peter A. Jones Immunobiology of Infection with Human Cytomegalovirus H. Kirchner Genetics of Resistance to Virus-Induced Leukemias Daniel Meruelo and Richard Bach Breast Carcinoma Etiological Factors Dan H. Moore, Dan H. Moore 11, and Cathleen T. Moore Treatment of Actue Leukemia-Advances in Chemotherapy, Immunotherapy, and Bone Marrow Transplantation Costa Gahrton The Forty-Year-Old Mutation Theory of Luria and Delbriick and Its Pertinence to Cancer Chemotherapy Howard E. Skipper Carcinogenesis and Aging Vladimrr N. Anisimoa INDEX

Volume 41 The Epidemiology of Diet and Cancer Tim Byers and Saxon Graham Molecular Aspects of Immunoglobulin Expression by Human B Cell Leukemias and Lymphomas John Gordon Mouse Mammary Tumor Virus: Transcriptional Control and Involvement in Tumorigenesis Nancy E. Hynes, Bernd Groner, and Rob Michalides Dominant Susceptibility to Cancer in Man David Harnden, John Morten, and Terry Featherstone Multiple Myeloma, Waldenstrom’s Macroglobulinemia, and Benign Monoclonal Gammopathy: Characteristics of the B Cell Clone, Immunoregulatory Cell Populations and Clinical Implications Hdkan Mellstedt, Goran Holm, and Magnus Bjorkholm

354

CONTENTS OF PREVIOUS VOLUMES

Idiotype Network Interactions in Tumor Immunity

Hans Schreiber Chromosomal Location of Immunoglobulin Genes: Partial Mapping of These Genes

in the Rabbit and Comparison with Ig Genes Carrying Chromosomes of Man and Mouse

Leandro Medrano and Bernard Dutrillaux