Advances in Cancer Research, Volume 36

ADVANCESINCANCER RESEARCH VOLUME 36 Contributors to This Volume Gianni Bonadonna Philip D. Lipetz Clark M. Edson J...

Author: Sidney Weinhouse | George Klein

10 downloads 1022 Views 19MB Size Report

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

DOWNLOAD PDF

ADVANCESINCANCER RESEARCH VOLUME 36

Contributors to This Volume Gianni Bonadonna

Philip D. Lipetz

Clark M. Edson

Janet D. Rowley

Maria E. Ferioli

Armando Santoro

Steinar Funderud

Giuseppe Scalabrino

Alan G. Galsky

Ralph E. Stephens

Kathi Geilinger

Joseph R. Testa

Tore Godal

David A. Thorley-Lawson Robert A. Weinberg

ADVANCES IN CANCER RESEARCH Edited by

GEORGE KLEIN Department of Tumor Biology Karolinska lnstitutet Stockholm, Sweden

SIDNEY WE INHOUSE Fels Research Institute Temple University School of Medicine Philadelphia, Pennsylvania

Volume 36-1982

ACADEMIC PRESS A Subsidiary of Harcourt Brace Jovanovich, Publishers

New York London Paris San Diego San Francisco SBo Paul0 Sydney Tokyo Toronto

COPYRIGHT @ 1982, BY ACADEMIC PRESS,INC. ALL RIGHTS RESERVED. NO PART O F THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM OR BY ANY MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMlSSION IN WRITING FROM THE PUBLISHER.

ACADEMIC PRESS,INC.

1 I1 Fifth Avenue, New York, New York 10003

United Kirigdoni Edition published by ACADEMIC PRESS, INC. ( L O N D O N ) LTD. 24/28 Oval Road, London N W I

7DX

LIBRARY OF CONGRESS CATALOG CARDNUMBER: 52-13360 ISBN 0-12-006636-X PRINTED IN THE UNlTED STATES OF AMERICA 82 83 64 85

9 8 7 6 5 4 3 2 1

CONTENTS CONTRlBUTORSTOVOLUME36

. . . .

ix

Polyamines in Mammalian Tumors: Part II GIUSEPPESCALABRINO AND MARIAE . FERIOLI I . Polyamine Biosynthesis and Concentrations in Different Lines of Cultured Neoplastic Cells . . . . . . . . . . . . . . . . . . . . I1. Polyamines in Human Oncology . . . . . . . . . . . . . . . . . . . . . 111. Diamine Oxidase Activity in Human and in Experimental Neoplasms . . . . . IV . Physiological and Pharmacological Inhibitors of Polyamine Biosynthesis in Neoplastic Tissues or Cells . . . . . . . . . . . . . . . . . V. Concluding Remarks and Speculations . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 20 56

62 86 88

Chromosome Abnormalities in Malignant Hematologic Diseases JANET D . ROWLEYA N D JOSEPH R . TESTA I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . . . Chronic Myelogenous Leukemia (CML) . . . . . . . . . . . . . Acute Nonlymphocytic Leukemia (ANLL) . . . . . . . . . . . . Acute Lymphocytic Leukemia (ALL) . . . . . . . . . . . . . . Polycythemia Vera . . . . . . . . . . . . . . . . . . . . . . Implications of Nonrandom Changes for Malignant Transformation . References . . . . . . . . . . . . . . . . . . . . . . . . .

I1. I11. IV . V. VI . VII .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

103 105 107 116 126 132 . . . . . 139 . . . . . 143

Oncogenes of Spontaneous and Chemically Induced Tumors ROBERTA . WEINBERG I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . I1. A Model of Cellular Oncogenes . . . . . . . . . . . . . . . . . I11. The Retrovirus-Associated Oncogenes . . . . . . . . . . . . . . IV . Oncogenes Present in Cells Transformed by Chemical Carcinogens .

. . . .

. . . 149 . . . 150 . . . . 150 . . . . 153

V . Multiplicity of Transforming Genes in 3-Methylcholanthrene-Transformed Cells 155 VI . Types ofTransformed Cells Yielding Focus-Induced DNA . . . . . . . . . . 156 V

vi

CONTENTS

VII . Multiplicity of Different Human Oncogenes . . . . . . . . . . . . . . . . VIII . Analogies between Virus- and Non-Virus-Induced Cellular Oncogenes . . . . . IX . The Process of Activation of Oncogenes . . . . . . . . . . . . . . . . . . X . The Role of Oncogenes in Carcinogenesis and Maintenance of Phenotype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI . The Proteins Encoded by Activated Oncogenes . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

157 158 159 160 161 162

Relationship of DNA Tertiary and Quaternary Structure to Carcinogenic Processes PHILIP

I. I1 . I11. IV .

D . LIPETZ. ALANG . GALSKY. A N D RALPHE . STEPHENS

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cancer and DNA Superstructure . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Note Added in Proof . . . . . . . . . . . . . . . . . . . . . . . . . .

165 167 189 202 204 210

Human B-Cell Neoplasms in Relation to Normal B-Cell Differentiation and Maturation Processes TOREGODALA N D STEINAR FUNDERUD I . Introduction . . . . I1 . The B-Cell System . 111. B-Cell Neoplasms . References . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

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

211 212 234 247

Evolution in the Treatment Strategy of Hodgkin’s Disease GIANNIBONADONNA A N D ARMANDO SANTORO I. I1 . 111. IV . V. VI . VII .

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . Radiation Therapy: From the Kilovoltage to the Megavoltage Era . . Chemotherapy: From Single Agents to Multiple Drug Treatment . . New Treatment Strategies . . . . . . . . . . . . . . . . . . . . Prognostic Factors Influencing Current Strategy . . . . . . . . . Morbidity Influencing Current Strategy . . . . . . . . . . . . . Conclusions: Toward the Total Conquest of Hodgkin’s Disease . . . References . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . 257 . . . . . 258 . . . . . 263 . . . . 267 . . . . . 277 . . . . . 283 . . . . . 287 . . . . 290

CONTENTS

vii

Epstein- Barr Virus Antigens-A Challenge to Modern Biochemistry DAVIDA . THORLEY.LAWSON. CLARKM . EDSON.A N D KATHIGEILINGER I . Introduction . . . . . . I1 . Transformation Antigens I11 . Early Antigens . . . . . IV . Late Antigens . . . . . V . Conclusions . . . . . . References . . . . . .

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

295 298 309 319 336 342

INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CONTENTS OF PREVIOUS VOLUMES. . . . . . . . . . . . . . . . . . . . . .

349 355

This Page Intentionally Left Blank

CONTRIBUTORS TO VOLUME 36 Numbers in parentheses indicate the pages on which the authors’ contributions begin.

GIANNI BONADONNA, Division of Medical Oncology, National Tumor Institute, Milan, Italy (257) CLARK M. EDSON,Sidney Farber Cancer Institute, Boston, Massachusetts 02115 (295) MARIAE . FERIOLI, Institute of General Pathology and C.N.R. Centre

for Research in Cell Pathology, University of Milan, 20133 Milan, Italy (1) STEINAR FUNDERUD, Laboratory for Immunology, Department of Pathology and The Norwegian Cancer Society, Norsk Hydro’s Institute for Cancer Research, The Norwegian Radium Hospital, Oslo, Norway (211) ALANG. GALSKY, Department of Biology, Bradley University, Peoria, Illinois 61625 (165) KATHIGEILINGER,’ Sidney Farber Cancer Institute, Boston, Massachusetts 02115 (295) TOREGODAL,Laboratory for Immunology, Department of Pathology and The Norwegian Cancer Society, Norsk Hydro’s Institute for Cancer Research, The Norwegian Radium Hospital, Oslo, Norway (211) PHILIPD. LIPETZ,Department of Radiology, The Ohio State University, Columbus, Ohio 43210 (165) JANET D . ROWLEY, Department of Medicine and The Franklin McLean Memorial Research Institute, The University of Chicago, Chicago, Illinois 60637 (103) ARMANDO SANTORO, Division of Medical Oncology, National Tumor Institute, Milan, Italy (257) GIUSEPPE SCALABRINO, lnstitute of General Pathology and C.N.R. Centre for Research in Cell Pathology, University of Milan, 20133 Milan, Italy (1) RALPHE . STEPHENS, Department of Radiology, The Ohio State University, Columbus, Ohio 43210 (165) ‘Present address: Department of Pathology, and Department of Medicine, Division of Geographic Medicine, Tufts University Medical School, Boston, Massachusetts 021 11. ix

X

CONTRIBUTORS TO VOLUME 36

JOSEPH R. TESTA,^ Department of Medicine and The Franklin McLean Memorial Research Institute, The University of Chicago, Chicago, Illinois 60637 ( 1 0 3 ) DAVIDA. T H O R L E Y - L A W S O N , ~Sidney Farber Cancer Institute, Boston, Massachusetts 02115 ( 2 9 5 ) ROBERTA. WEINBERG, Massachusetts Institute of Technology, Center for Cancer Research and Department of Biology, Cambridge, Massachusetts 02139 ( 1 4 9 )

'Present address: NCI-Baltimore Cancer Research Program, Baltimore, Maryland 21201. 3Present address: Department of Pathology, and Department of Medicine, Division of Geographic Medicine, Tufts University Medical School, Boston, Massachusetts 02111.

ADVANCES IN CANCER RESEARCH VOLUME 36

POLYAMINES IN MAMMALIAN TUMORS PART Ill Giuseppe Scalabrino and Maria E. Ferioli Institute of General Pathology and C N R Centre tor Research in Cell Pathology, University of Milan. Milan, Italy

Nil minus est hominis occupati quam vivere: nullius rei difficilior scientia est. Professores aliarum artium vulgo multique sunt, quasdam vero ex his pueri admodum ita percepisse visi sunt, u t etiam praecipere possent: vivere tota vita discendum est et, quod magis fortasse miraberis, tota vita discendum est mori. SENECA, “De Brevitate Vitae,” 7,3 L’ignorance qui estoit naturellement en nous, nous I’avons, par longue estude, confirmee e t averee. MONTAIGNE, “Essais,” L. 11, C. 12

I. Polyamine Biosynthesis and Concentrations in Different Lines of ........ Cultured Neoplastic Cells . . . A. Responses to Microenviron perature, PO,) and to the Presence of Different Exogenous Molecules (Amino Acids, Di- and Polyamines, Antipolyamine Antibodies) . . B. I n Relation to the Growth Rate and the Phase of the Cell Cycle . . . . . . . C . Two-way Relationships between Polyamines and Cyclic Nucleotides. Inducibility of the Two Polyamin D. Effects of Infection with Nononcogenic Vir E. Miscellaneous Effects of Polyamines . . . . . . 11. Polyamines in Human Oncology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............ A. Patterns of Polyamines in Human Neoplastic Tissues B. Levels of the Chief Polyamines a Urines of Normal Subjects and of C. Levels of the Chief Polyamines and Their Conjugated Forms in Blood, Plasma, Serum, Formed Blood Elements and Bone Marrow of Normal ................ Subjects and of Cancer Patients D. Levels of the Chief Polyamines ....................... Bloodand Urine . . . . . . . E. Levels of Activity of P Neoplastic Tissues in Relation to the Degree of Malignancy F. Metabolic Conjugation .............. Normals and in Cancer Patients 111. Diamine Oxidase Activity A. In Human Tumors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. In Experimental Tumors ................

2 2 11 12 18 18 20 24

38

49 53 55 56 57 60

I Part I of this review (see Volume 35 of this series) covered polyamines and their metabolism in normal tissues and in chemical, physical, and viral carcinogenesis.

1 ADVANCES IN CANCER RESEARCH, VOL. 36

Copyright 0 1982 by Academic Press, Inc. All rights of reproduction in any fonn reserved. ISBN 0-12-006636-X

2

GIUSEPPE SCALABRINO A N D MARrA E. FERIOLI

IV. Physiological and Pharmacological Inhibitors of Polyamine Biosynthesis in Neoplastic Tissues or Cells ............................... A. Physiological Inhibitors and Related Compounds ..................... B. Pharmacological Inhibitors .......................................... V. Concluding Remarks and Speculations ................................... References .............................................................

62 63 68 86

88

I. Polyarnine Biosynthesis and Concentrations in Different Lines of Cultured Neoplastic Cells

A. RESPONSES TO MICROENVIRONMENTAL CONDITIONS (OSMOLARITY, TEMPERATURE,

p o z ) AND TO THE PRESENCE OF

DIFFERENT EXOGENOUS MOLECULES(AMINOACIDS, POLYAMINES,

D I - AND

ANTIPOLYAMINE ANTIBODIES)

The need for expediency in experimental cancer studies has made the cultured neoplastic cell the principal tool for cancer research. There are obvious advantages in working with uniform cell lines that can be prepared as clean cell suspensions. However, the artificial conditions ofin vitro culture tend to change the characteristics of the cells. Several authors have studied different neoplastic cell lines growing in culture in order to delineate the metabolic pathways of the polyamines and to affect growth of these cells by selectively inhibiting polyamine synthesis or sequestering polyamines in order to define their roles (see also Section IV). It is particularly interesting to modify the culture conditions in order to clarify the influence of the environmental milieu on the activities of the polyamine biosynthetic enzymes by enhancing or decreasing the levels of these enzymes to see the role of polyamines in the cell growth process. To this purpose, several studies were carried out with rat hepatoma cells growing in culture. Studies of the effects of growth conditions on ornithine decarboxylase (ODC) activity presented evidence that dilution of high-density hepatoma cell cultures with fresh medium resulted in a very large and transient increase in enzyme activity, reaching a peak about 4 hr after dilution (Hogan, 1971; Hogan and Blackledge, 1972; Hogan e t d.,1973; Hogan and Murden, 1974). This increase was abolished by cycloheximide but not by actinomycin D, suggesting that ODC induction is controlled under these experimental conditions at the posttranscriptional level (Hogan, 1971; Hogan and Blackledge, 1972). The aforementioned increase of ODC activity appeared to be in part concomitant with a decrease in the rate of ODC degradation, i.e., with an increase in the half-life of ODC, hinting at a

POLYAMINES IN MAMMALIAN TUMOHS

3

causal relationship between the two phenomena (Hogan and Blackledge, 1972; Hogan et al., 1973; Hogan and Murden, 1974). The supplementation of high-density hepatoma cell cultures with glutamine or serum or nonessential amino acids, but not with essential amino acids, produced an ODC stimulation of severalfold, at least partially due to an increase of the half-life of ODC (Hogan et al., 1973, 1974; Hogan and Murden, 1974; Fong et ul., 1976). On the contrary, very high concentrations of pyridoxal do not affect the apparent halflife of ODC (Hogan and Murden, 1974). Other investigators followed ODC activity and putrescine levels through two generations of rat hepatoma cells cultured in the presence of serum (McCann et al., 1975). Biphasic ODC induction was noted during the first 24 hr. The intracellular putrescine concentration was found to correlate with rises in ODC activity (McCann et al., 1975). On the contrary, only one broad peak of ODC activity was observed over the same period in diluted hepatoma cell cultures without serum, with no parallel increase in the cellular putrescine content (McCann et al., 1975). Therefore, these authors concluded that only growing and dividing hepatoma cells have biphasic ODC induction that parallels increased putrescine levels, whereas a single peak of ODC stimulation can be achieved in nongrowing cells. Among the factors affecting the growth conditions of the cultured cells, the addition of fresh medium or serum to the culture has been demonstrated to be one of the most important for a variety of other cell lines. Induction of ODC activity, followed by a significant elevation of cellular putrescine concentration, in a rat glioma clone and in a mouse neuroblastoma cell clone when fresh medium was added to confluent cultures was reported (Bachrach, 1976c, 1977, 1980a; Bachrach e t al., 1978). In more detail, ODC activation following the addition of fresh serum was preceded by similar responses in both adenosine 3’ : 5’cyclic monophosphate (cyclic AMP, CAMP)-dependent and CAMPindependent protein kinases of glioma cells (Bachrach et al., 1978). However, in this case no difference in the half-life of ODC before and after the addition of fresh medium was observed (Bachrach, 1 9 7 6 ~ ) . Moreover, the induction of ODC activity appears to be specific for this enzyme, since the activities of other enzymes that decarboxylate other amino acids were not stimulated by the addition of fresh medium (Bachrach, 1 9 7 6 ~ ) This . study also suggests a correlation between growth rate and ODC activity in cultured glioma and neuroblastoma cells, since the enzyme activity was high when the cells were proliferating rapidly (Bachrach, 1 9 7 6 ~ )Interestingly . enough, the addition of serum to culture medium containing mouse neuroblastoma cells or

4

GIUSEPPE S C A L A B R I N O A N D MARIA E. FERIOLI

rat glioma cells also greatly increased formation of y-aminobutyric acid (GABA) from putrescine (Kremzner et al., 1975; Sobue and Nakajima, 1977). The S-adenosyl-L-methionine decarboxylase (SAMD) activity was also increased in glioma and neuroblastoma cells shortly after the addition of complete fresh medium (Bachrach, 1977, 1980a). As expected, the enhancements of the activities of the polyamine biosynthetic decarboxylases were found to be paralleled by increases in cellular concentrations of polyamines and of y-aminobutyric acid formed from putrescine (Bachrach, 1980a). In HeLa cells, in response to the addition of serum to quiescent cells not only ODC activity increased but also SAMD activity (Prouty, 197613; Maudsley et al., 1978). Under the same experimental conditions, putrescine and spermidine levels markedly increased as well (Maudsley et al., 1978). When labeled ornithine was added to the cells during the period of the serum stimulation and its uptake was measured, a marked and rapid increase in polyamine levels above that normally observed in resting cells was noted (Maudsley et al., 1978). It appeared that most of the cytosol ornithine was decarboxylated to yield putrescine, which in turn was quickly converted to spermidine (Maudsley et al., 1978). More or less analogous observations were made in KB cells (Pett and Ginsberg, 1968) and in hepatoma cells (Bondy and Canellakis, 1980). In HeLa cells growing in suspension culture, ODC activity was also found to be potently stimulated by the addition of glutamine to the medium; this stimulation was due, at least partly, to a decrease in the rate of decay of the ODC activity (Prouty, 197613). In cultures of L1210 mouse leukemic cells, of hepatoma H35 cells, of neuroblastoma cells (Chen et al., 1976a,b), of virally induced glioma-like hamster brain tumor cells (Hsu et al., 1977), and of Friend erythroleukemia cells (Tsiftsoglou and Kiriakidis, 1979; Gazitt and Friend, 1980), the addition of fresh medium plus serum also resulted in an increase in ODC activity. Additionally, when Friend leukemia cell cultures were stimulated to proliferate by dilution of stationaryphase cultures with fresh medium, both their nucleolar RNA synthesis rates and ODC levels were increased (Dehlinger and Litt, 1978).The addition of putrescine at the time of dilution with fresh medium blocked the increase in ODC levels, but did not prevent the increase in nucleolar RNA synthesis (Dehlinger and Litt, 1978).As observed by Ferioli et al. (1980) in postischemic liver repair, ODC induction in Friend leukemia cells can be dissociated from the stimulation of RNA synthesis.

POLYAMINES IN MAMMALIAN TUMORS

5

A proliferative stimulus for the cultured cells, such as the addition of serum, was followed by a dramatic increase in the rate of putrescine transport into both normal human fetal lung fibroblasts and the same cell line transformed by SV40 (Pohjanpelto, 1976). Conversely, the removal of serum resulted in a rapid decrease in the rate of putrescine transport. The magnitude of the increases or of the decreases in the rates of putrescine transport in these two cell lines in response to the addition or the removal of serum were nearly the same (Pohjanpelto, 1976). For studies of the effects of the addition of fresh serum, 12-0tetradecanoylphorbol-13-acetate(TPA), and/or a combination of the two on ODC activity of cultured malignant cells, the reader is referred to Section III,C,2,b, Part I, Vol. 35. Besides the previously mentioned glutamine, another a-amino acid with nonionic polar side chains, asparagine, is a powerful inducer of ODC activity in confluent neuroblastoma cells (Chen and Canellakis, 1977). Among the natural amino acids tested, asparagine led in ability to induce ODC, with L-glutamine second, half as effective as asparagine (Chen and Canellakis, 1977). This ODC induction was neither concomitant with nor followed by an increased rate of incorporation of precursors into DNA, RNA, or proteins (Chen and Canellakis, 1977). What is really astonishing is the finding that asparagine and glutamine play a “permissive” role in ODC induction by N 6 , 0 2 ’ dibutyryl cAMP or by prostaglandin E l (PGEl) plus S-isobutyl1-methylxanthine, since none of these three molecules alone stimulated ODC activity at all in confluent neuroblastoma cells in a medium devoid of asparagine or glutamine (Chen and Canellakis, 1977). The “stabilizing” effect of asparagine on ODC was demonstrated b y the very great lengthening of the enzyme’s half-life when asparagine was in the medium (Chen and Canellakis, 1977). Moreover, in mouse neuroblastoma cells induced to differentiate by any of several stimuli, the regulation of ODC induction by asparagine in undifferentiated and in differentiated cells was studied comparatively (Chen, 1979, 1980). The addition of asparagine to a salts-glucose medium elicited a maximal increase of ODC activity in undifferentiated cells with further addition of fetal calf serum or of N6,02’-dibutyryl cAMP not resulting in any additional increase (Chen, 1979, 1980). In contrast, the addition of asparagine alone caused a small increase in ODC activity in differentiated cells, and this increase was potentiated and reached a maximum after addition of fetal calf serum or of N 6 , 0 2 ’ dibutyryl cAMP (Chen, 1979, 1980).

6

GIUSEPPE SCALABRINO AND MARIA E . FERIOLI

In the presence of asparagine, confluent glioma cells exhibited an increase in ODC activity whereas the SAMD activity of the same cells under the same experimental conditions remained at the basal levels (Bachrach, 1980a). That asparagine stabilizes ODC, leading to a lengthening of the enzyme half-life and to an apparent increase in its activity, was confirmed (Bachrach, 1980a). These results are in good agreement with those of Chen and Canellakis (1977). Furthermore, the extents of ODC induction by asparagine were compared in normal and in transformed fibroblasts (Costa, 1979; Costa and Nye, 1979). Enhancement of ODC by this amino acid was much greater in cells transformed either by SV40 or by Ni& than in the counterpart normal cells (Costa, 1979; Costa and Nye, 1979). Despite the numerous investigations, the exact role of asparagine in influencing cellular polyamine metabolism remains to be elucidated, and further studies are needed to achieve a better understanding of the mechanisms by which this amino acid specifically affects ODC activity inside the cells. However, among the various inducers of ODC activity in cultured cells mentioned so far, GABA has been shown to be remarkably more effective in enhancing ODC activity, at least in cultured rat hepatoma cells, than the amino acids asparagine and glutamine (McCann et al., 197913). Like asparagine, GABA seems to have a direct stabilizing effect on ODC with a consequent slowing down of the enzyme’s turnover and a concomitant lengthening of its half-life (McCann et al., 1979b). The cellular putrescine levels after addition of GABA to the culture medium increased in parallel with the increases in ODC activity (McCann et al., 1979b). However, addition of GABA modified neither the cellular spermidine and spermine concentrations nor the SAMD activity of the tumor cells (McCann et al., 197913). Whether or not GABA has a general role in the comprehensive complex regulation of ODC activity in eukaryotic cells or, on the contrary, has only a limited role in particularly specialized cells, such as brain cells, with elevated GABA concentrations, remains a matter for speculation. Another environmental factor influencing basal ODC activity and polyamine contents in cultured cells is the osmolality of the surrounding medium. In HeLa cells the polyamine contents were found to be inversely related to the osmolality of the growth medium (Munro et al., 1975). A sudden increase in NaCl concentration of the medium causes a rapid fall in putrescine and spermidine concentrations. A sudden decrease in NaCl in the medium causes a rapid increase in putrescine (Munro et al., 1975). The levels of ODC activity in relation to external osmolality behaved like the polyamine contents and were, therefore, consistent with the changes in the polyamine levels (Munro

POLYAMINES IN MAMMALIAN TUMOHS

7

et al., 1975). The ODC soon declined when the external NaCl concentration rose and increased when the osmolality decreased. Interestingly enough, these variations in ODC activity were accompanied by similar variations in the half-life of the enzyme, since the half-life decreased when the enzyme activity decreased and increased when the enzyme activity increased (Munro et al., 1975). Nevertheless, suitable concentrations of intracellular cations are also important for regulating ODC activity, at least in L1210 mouse leukemic cells. It has been demonstrated that several ionophore antibiotics (which are compounds produced by microorganisms that specifically increase the permeability of the cell membrane to ions), such as valinomycin, nigericin, and monensin (valinomycin belongs to the group of the neutral ionophore carriers, and nigericin and monensin are carboxylic carriers) have the ability to strongly inhibit ODC activity, with only a slight decrease in protein synthesis (Chen and Kyriakides, 1977). The extracellular cations, in addition to regulating basal intracellular ODC activity, play essential roles in influencing ODC induction. Accordingly, the addition of suitable amounts of MgClz or NaCl or KCI completely or nearly inhibited the rises in ODC activity induced in cultured H35 or neuroblastoma cells by the addition of fresh medium with or without a serum supplement (Chen et al., 1976b). These cations, when present in the induction medium, prevented ODC enhancement (Chen et d.,197613). And, what is of more interest, if the L1210 cells have been grown for several generations in a medium containing a high Mg2+ concentration, the ODC induction produced by diluting these cells during the stationary phase with a medium of high Mg2+concentrations, i.e., with a medium theoretically inhibiting the ODC response, surprisingly reached the same levels as in the controls (Chen et al., 197613). This means that the leukemic cells are able to adapt progressively to new environmental conditions, at least in ODC induction. Furthermore, these cations did not significantly modify macromolecular synthesis when they were present in the medium at the same concentrations at which they prevented ODC induction by serum or fresh medium (Chen et al., 197613). Hitherto we have considered mainly those external factors able to induce ODC activity inside cells in culture. There are other factors that can cause the converse effects. The exposure of cultured hepatoma, W256 carcinosarcoma, leukemic, or neuroblastoma cells to different substances, chiefly to putrescine added to the culture medium, greatly decreased the ODC activity inside the cells (Clark and Fuller, 1976; Fong et al., 1976; Heller et al., 1976a,b, 1977a, 1978;

8


McCann et al., 1977a; Heller and Canellakis, 1980). As discussed in detail in Section I,B,l, Part I, Vol. 35, under such experimental conditions putrescine elicits the synthesis of the ODC antizyme. Among the neoplastic cell lines tested, the only negative reports in this respect are those of Clark and Fuller (1976), who did not detect this ODC inhibitor in cultures of polyoma-transformed 3T3 cells after exposure to serum and putrescine, and of Kudlow et al. (1980),who did not note the presence of the soluble inhibitor of ODC activity in cultured cells of a mouse adrenocortical tumor, but without any pretreatment whatever with polyamine. In addition to putrescine, spermidine, spermine, cadaverine, and other unphysiological polyamines can elicit the synthesis of ODC antizyme in a great variety of cultured cell lines, both neoplastic and normal (Heller et al., 1977a, 1978; McCann et al., 1977a). It must be recalled here once again that the concentrations of di- or polyamines in the medium that are required to stimulate the ODC antizyme are several orders of magnitude smaller than the amounts of di- and polyamines present inside cultured cells (Canellakis et al., 1978; Heller et al., 1978). The half-life of the ODC antizyme in various cultured neoplastic cell lines has been shown to vary roughly proportionally with the variations of the half-life of ODC measured under the same experimental conditions (Heller et al., 1976b). Last, it should also be noted that ODC antizyme is normally present in cultured hepatoma cells, not stimulated for synthesis of this ODC inhibitor (Heller et al., 1977b). Under these resting conditions the antizyme exists as an inactive protein bound to subcellular components from which it can be liberated b y treatment with very low concentrations of polyamines, noticeably of putrescine, at concentrations far less than those usually found inside the cells (Heller et al., 197713). The use of heat alone for treatment of cancer patients dates back to the late nineteenth century. However, there has been renewed interest, and considerable emphasis has been placed on using this old treatment for cancer patients, i.e., hyperthermia, either alone or in combination with other types of antineoplastic therapy, usually with radiation (Manning, 1979). It is now widely known that virtually every fundamental phase of cellular biochemistry (respiration, glycolysis, DNA, RNA, and protein synthesis) can be disrupted by sufficient exposure of mammalian cells to hyperthermia, resulting first in a loss of proliferative capacity and ultimately in cell death. In spite of this, the molecular mechanisms by which hyperthermia kills cells or causes prolonged cell cycles are still not fully understood. Nonetheless, some interesting studies have dealt with this topic, elucidating some aspects of the problem. In synchronous Chinese hamster ovary cultures, pro-


9

gressing through the cell cycle after exposure to 43°C for 1 hr during either the GI or the S phase, there is a remarkable leakage of polyamines, mainly spermidine and spermine, into the culture medium (Gerner and Russell, 1977; Gerner et al., 1980). Obviously, this has as natural consequence the depletion of intracellular spermidine and spermine (Gerner et al., 1980);this depletion was reversed when the temperature was reset at 37°C (Gerner and Russell, 1977). In contrast, the intracellular putrescine concentration was not affected b y exposure of the cells to heat shock (Gerner and Russell, 1977). It is reasonable to connect the depletions of intracellular spermidine and spermine, most probably due to membrane damage by heat, with the alterations in DNA synthesis observed in the same cell line under the same experimental conditions (Gerner and Russell, 1977). Again, polyamines have the property of potentiating the killing of the cells by heat. In fact, exposure of cultured Chinese hamster cells to hyperthermia plus a polyamine (cadaverine or putrescine or spermidine or spermine) in the growth medium resulted in dramatic, synergistic cell death, regardless of the order of the two treatments (Ben-Hur et al., 1978; Gerner et al., 1980). Spermine was the most effective polyamine for potentiating thermal cell killing, followed by spermidine, cadaverine, and putrescine, in order of effectiveness (Ben-Hur et al., 1978). When there was a long time interval between the two treatments, this synergism disappeared (Ben-Hur et al., 1978). This enhancement of thermal killing by polyamines is dependent on the time of exposure and on the concentration of the exogenous polyamines (Gerner et al., 1980).The minimal polyamine concentrations that enhance the thermal sensitivity of the cells were by far lower than those normally found intracellularly, strongly suggesting a membrane effect (Gerner et al., 1980). Moreover, prolonged hyperthermia caused an increase in the uptake of exogenous polyamines (with the exception of putrescine) added to the growth medium by the same cultured cell line (Ben-Hur and Riklis, 1978). Generally speaking, the polyamines that penetrated into the cells were metabolized into the same products at both the physiological and the high temperature, indicating that the enhancement by polyamines of cellular sensitivity to heat shock is due to these molecules as such, not to their metabolites (Ben-Hur and Riklis, 1978). Another line of evidence that the potentiation of hyperthermia-induced cytotoxicity by polyamines is specific for these polycations is that the effect was not obtained with mono- or divalent inorganic cations, such as KC1, CaCI2, and MgC12 at equimolar concentrations (Gerner et al., 1980). The importance of all the foregoing observations lies in the fact that neoplastic cells usually contain

10


larger amounts of polyamines, and this may be one reason for the wellknown higher sensitivity of cancer cells to heat. Hyperthermia also produced in cultured Chinese hamster fibroblasts a drastic decrease in ODC activity, which occurred rapidly and exponentially as a function of continued exposure to heat (Fuller et al., 1977; Ben-Hur and Riklis, 1979a; Gerner et al., 1980).When the temperature reverted to the physiological level, ODC activity recovered and returned to control levels, after overshooting (Ben-Hur and Riklis, 1979a). The activity of SAMD was affected in the same way as that of ODC b y hyperthermia (Fuller et al., 1977). Polyamines have been shown to potentiate the killing effect of heat on mammalian cells, and they can also amplify the synergism between heat and radiation. In fact, spermine enhanced the synergistic interaction between hyperthermia and y-radiation in cultured Chinese hamster fibroblasts (Ben-Hur and Riklis, 1979b).This property of spermine to further strengthen the radiosensitizing effect of heat on the cells resulted in (a) enhanced cell death in the presence of radiation plus heat plus tetraamine, as compared to radiation plus heat; and (b) a drastic inhibition of cell repair of radiation-induced sublethal damage (Ben-Hur and Riklis, 1979b). Unfortunately, the effects of exogenous polyamines on .cultured neoplastic cells exposed to hyperthermia have received little, if any, attention so far, and almost nothing is known about this topic. In addition to hyperthermia, severe hypoxia induces an arrest of the cell cycle, so that it has been, and might presently be, considered a potentially useful tool for cancer therapy. In this regard, there is an interesting report that Chinese hamster cells exposed to severe and prolonged hypoxia and then reoxygenated slowly reenter the cell cycle and progressively increase their protein and DNA syntheses, but, astoundingly, the ODC activity fails to increase when oxygen is supplied (Kehe and Harris, 1978). In this instance, as in hyperthermia, the effect of hypoxia on polyamine biosynthesis and metabolism in cultured neoplastic cells has never been tested. Cellular ODC activity ca'n be greatly inhibited by some mitotic poisons, such as colchicine and vinblastine, that disrupt the cellular microtubule system. Indeed, when these two drugs were added to the medium of cultured L1210 cells, the activation of ODC activity by the dilution of the cells was prevented (Chen et al., 1976a). In contrast, lumicolchicine, an isomer of colchicine without any effect on the microtubular system, did not inhibit the ODC rise under the same experimental conditions (Chen et a1., 1976a). Vinblastine and colchicine also blocked O D C induction in rat glioma cells by dibutyryl CAMP

11

POLYAMINES IN MA.MMALIAN TUMORS

(Gibbs et al., 1979, 1980). In this cell line, too, lumicolchicine had no effect on either basal or stimulated levels of ODC activity (Gibbs et al., 1979, 1980). Furthermore, the integrity of the cytoskeleton seems to b e of great importance for ODC stimulation by added serum, since cytochalasin B inhibited ODC induction in L1210 cells (Chen et al., 1976a; Gibbs et al., 1979, 1980). Last, using an immunological approach, Quash and his collaborators (1971, 1972, 1973, 1978) have demonstrated that antipolyamine antibodies are cytotoxic for baby hamster kidney cells transformed b y the polyoma virus and growing in cell culture, and that complement is a necessary factor for the cytolytic effect, indicating the involvement of the cell membrane in the phenomenon. Moreover, cytolysis was inhibited and the cells recovered if the antipolyamine antibodies were removed or if putrescine, but not spermidine or spermine, was added to the culture medium containing the antiserum. Finally, evidence was provided that cytolysis of BHK-transformed cells is caused by the interaction of antidiamine antibodies with putrescine-containing sites on the cell membrane. This stresses once again the importance of the cell membrane in regulating ODC activity inside the cell.

B.

I N RELATIONTO CELL CYCLE

THE

GROWTHRATE

AND THE

PHASE OF

THE

Cultured cells of experimental neural neoplasias have been widely employed to investigate the connections between polyamine contents, combined or not with enzyme levels, and the growth rate of the cells. The activities of ODC and SAMD in a rat brain tumor cell line reached their maximum levels during the exponential growth phase and decreased as the growth curve reached a plateau (Heby et al., 1975b). The correlation coefficients obtained for the relationship of the enzyme activities to the specific growth rates were highly statistically significant (Hebyet al., 1975b).In studies of the correlations between cellular polyamine levels and the specific growth rate of the tumor cells, putrescine and spermine were not correlated, whereas spermidine and the spermidine : spermine ratio showed a direct positive linear correlation (Heby et al., 1975a,b). Parenthetically, it must be stressed that the rate of cell multiplication was maximal when the spermine content 1975b). Last, the compartmentalization of the was lowest (Heby et d., polyamines between nucleus and cytoplasm in this brain tumor cell line strongly indicates that spermidine and spermine act at the nuclear level, because the concentrations of these two polyamines were much

12

GIUSEPPE SCALARRINO AND MARIA E . FERIOLI

higher in the nucleus than in the cytoplasm (Heby, 1977).There were no significant changes in putrescine levels between the two cellular compartments, although the ODC activity was located mainly in the cytoplasm (Heby, 1977). In mouse neuroblastoma cells and in rat glioma cells, the spermidine : spermine ratio was found to decrease when growth was less rapid, and the putrescine content decreased as the cells entered the stationary phase (Kremzner, 1973; Kremzner et al., 1975; Sobue and Nakajima, 1977). However, the metabolism of the polyamines in these two kinds of neural neoplastic cell lines was found to be different, since in neuroblastoma cells the formation of GABA from putrescine was low during the logarithmic phase of cell growth and increased astoundingly during the stationary phase, whereas in glioma cells this metabolic conversion was always low throughout both phases (Sobue and Nakajima, 1977). There seems to be an inverse correlation between the rate of polyamine biosynthesis and the size of the polyamine pool in HeLa cells. The contents of polyamines were the highest during mitosis and the late GI phase, while at these times polyamine biosynthesis was minimal (Sunkara et al., 1979~). On the other hand, the polyamine contents were the lowest during early the GI and S phases, while the polyamine biosynthesis was maximal (Sunkara et al., 1979~).However, conclusions drawn from studies carried out with synchronized cell populations have to be drawn with caution, since it has been demonstrated in HeLa cells that the synchronization protocols, which yield large numbers of synchronized cells, can deeply affect both the basal cell content of polyamines and the polyamine accumulation during the cell cycle (Goyns, 1980). Polyamine biosyntheses and their levels in normal cultured cells in the different phases of the cell cycle have been reviewed briefly by Pardee et al. (1978)and exhaustively by Heby and Anderson (1980).

c. TWO-WAYRELATIONSHIPSBETWEEN POLYAMINES AND CYCLIC NUCLEOTIDES. INDUCIBILITY OF THE TWO POLYAMINE BIOSYNTHETIC DECARBOXYLASES As mentioned and discussed in Section I,E,l, Part I, Vol. 35, there are accumulated lines of evidence for CAMP as mediator of ODC induction in both in vitro and in vivo systems, with some arguments against it. We will now describe the experiments connected with this aspect carried out in cultured cells plus those experiments emphasizing the reverse aspect of the problem, i.e., the effects of polyamines on


13

biosynthesis and metabolism of the cyclic necleotides in some cultured cell lines. ODC activity has been induced by cAMP and by dexamethasone (which is thought to act without implicating cAMP as second messenger) in logarithmically growing hepatoma cells originated from a Morris rat hepatoma and maintained in suspension culture (Canellakis and Theoharides, 1976).The induced enzyme was characterized by its immunoprecipitation and heat-stability patterns and shown to be identical with the enzyme purified from untreated hepatoma cells (Canellakis and Theoharides, 1976). ODC induction by these two drugs is similar in that the rates of ODC synthesis are markedly enhanced in both responses over that in controls (Canellakis and Theoharides, 1976). However, these two types of ODC induction have been shown to differ from each other in several aspects, namely, in their time courses and in their responsiveness to different types of inhibitors, such as actinomycin D and polyamine (Canellakis and Theoharides, 1976). First, the time course of ODC induction after dexamethasone was much slower than that after CAMP, but the enzymic levels were steadily elevated for many more hours. Second, actinomycin D completely inhibited induction by the glucocorticoid but only partially inhibited induction by CAMP.Third, the reverse is true for the effects of spermine and spermidine, since either these two polyamines depressed the ODC levels in the presence of cAMP even below basal control levels, whereas they were without effect on dexamethasone induction (Theoharides and Canellakis, 1975; Canellakis and Theoharides, 1976). Therefore the control of ODC activity in cultured hepatoma cells implies two paths, one CAMP-dependent and one CAMP-independent. All these results are in substantial agreement with those obtained with the same two drugs on another cell line of cultured hepatoma, i.e., the Reuber H35, by Byus et al. (1976) at the same time. ODC activity was induced in the H35 cells not only by dexamethasone and CAMP, but also by a series of 8-substituted cAMP analogs (Byus et al., 1976). Addition of insulin to H35 cultured cells was not followed by any increase in enzyme activity (Byus et al., 1976). Furthermore, the inducibility of ODC activity in H35 hepatoma cells has also been shown to be dependent on the composition of the culture medium (Liu and Chen, 1979). In vitro incubation of slices of rat adrenocortical carcinoma in the presence of cAMP resulted in significant ODC induction (Richman et al., 1973). More recently, a genetic approach has been used to explore whether hormonal activation of ODC activity is mediated by cAMP

14


and CAMP-dependent protein kinase (Kudlow et al., 1978, 1980). These authors used a cell line of a mouse adrenocortical tumor and its mutant clones, defective in either the adenylate cyclase response to ACTH or the response of the CAMP-dependent protein kinase to the same hormone. Both ACTH and cAMP induced ODC activity in the intact cell line, whereas only cAMP was able to enhance the enzyme levels in the mutant clones defective in adenylate cyclase (Kudlow et al., 1978, 1980). In the mutant clones defective in CAMP-dependent protein kinase, the magnitude of ODC response to ACTH was greatly reduced, but not totally abolished in some instances (Kudlow et al., 1978,1980). Both cAMP and nerve growth factor (NGF) induced ODC activity in a clonal cell line that originated from a rat adrenal pheochromocytoma, although the two types of induction were shown to be not causally interdependent, since NGF added to the culture medium did not produce any significant increase in cellular cAMP levels, even in the presence of theophylline (Hatanaka et al., 1978).The addition of either insulin or epidermal growth factor (EGF) (which both share some structural analogies with NGF) or of N2,02’-dibutyrylcGMP caused only little or no stimulation of ODC activity in this type of cultured neoplastic cell (Hatanaka et al., 1978). A concurrent report confirmed that ODC activity can be induced in these pheochromocytoma cells b y NGF and that the phenomenon requires new protein synthesis (Greene and McGuire, 1978). Furthermore, it was demonstrated that other noteworthy biological effects of NGF in these responsive cells, i.e., the stimulation of cell survival and of neurite outgrowth, were not impeded by total suppression of the cellular ODC activity by treatment with 1,3-diaminopropane or 5-hexyne1,4-diamine (Greene and McGuire, 1978). In contrast with the earliest report of Hatanaka et al. (1978), it was shown that E G F induced ODC in the rat pheochromocytoma clone PC12 and that preincubation of these cells in the presence of NGF largely prevented the ODC response to addition of E G F (Huff and Guroff, 1979). Both E G F and insulin stimulated putrescine transport into KB cells, but only insulin significantly enhanced the ODC levels of this type of cultured neoplastic cell (Di Pasquale et al., 1978). Cultures of tumors of the central nervous system are a good experimental model for clarifying the links between cyclic nucleotides and polyamine biosynthetic decarboxylases. The ODC activity of a rat glioma clone was quickly stimulated by addition of norepinephrine or isoproterenol or 3-isobutyl-1-methylxanthine (IBMX) or dibutyryl cAMP (Bachrach, 1975). In a mouse neuroblastoma clone, ODC activity was induced by PGE, or adenosine, but preincubation with IBMX

POLYAMINES I N MAMMALIAN TUMORS

15

was absolutely necessary to obtain the stimulatory effect (Bachrach, 1975).Complementary to this observation is another provided by Bachrach and his co-workers (1979), which demonstrates that morphine (an opiate that inhibits adenylate cyclase activity in some neural cell lines) almost completely antagonized the stimulating effects of IBMX with or without PGEl on the ODC activity and of the combination of the two drugs on CAMP levels in neuroblastoma x glioma hybrid cells (Bachrach e t al., 1979). In this case morphine also inhibited the stimulation of the activity of CAMP-dependent protein kinase elicited b y PGEl plus IBMX in the same cell hybrids (Bachrach et al., 1979). The assumption that CAMP is involved in SAMD induction too, was made and verified in the same cultured neoplastic lines (Bachrach, 1977). In fact, the level of this second polyamine biosynthetic decarboxylase was eleveted in glioma cells by the phosphodiesterase inhibitor IBMX or by catecholamines, and in neuroblastoma cells by PGE, and IBMX (Bachrach, 1977). All the foregoing results strongly support the idea that inductions of both the polyamine biosynthetic decarboxylases in cell cultures derived from neoplasms of the central nervous system of the rat or the mouse are mediated by CAMP and probably imply a wide cascade of biochemical events. However, Chen and Canellakis (1977) demonstrated that the stimulation of ODC activity in cultured mouse neuroblastoma cells brought about by the addition of N6,0Z’-dibutyrylcAMP or by PGE, plus IBMX was completely dependent on the presence of optimal concentrations of asparagine in the medium. Even more interesting, these authors also demonstrated that ODC could be induced in this cell line also b y high concentrations of asparagir,,. without any CAMP (Chen and Canellakis, 1977). An attempt to reconcile these seemingly contradictory reports of the true role of cAMP in ODC induction in neoplastic neural cell lines has been made by Gibbs et al. (1979,1980),who showed that there are separate pathways of ODC induction in rat glioma cells, i.e., one involving CAMPmediation and one not, and that these two pathways have as a common biochemical feature an absolute requirement for Ca’+. In fact, both isoproterenol and dibutyryl CAMP induced ODC activity in these neural cells, but both these inductions were completely abolished by the presence in the culture medium of ethylene glycol bis(p-aminoethyl ether)-N,N,N N ’-tetraacetic acid (EGTA), a well-known and powerful Ca2+chelator (Gibbs e t al., 1979, 1980); EGTA alone reduced the basal level of ODC activity (Gibbs et d., 1979, 1980). On the other hand, EGTA was also able to prevent ODC induction in the same cellular line by fetal calf serum, which last had little, if any, effect upon intracellular CAMP content (Gibbs et al., I,

16


1979, 1980). The addition of calcium alone without serum did not modify ODC activity to any appreciable extent, and the specifity of Ca2+requirement for ODC induction has been demonstrated by the fact that this ion cannot be substituted for magnesium (Gibbs et al., 1979, 1980). The hypothesis that cAMP mediates the increase in ODC activity has also been tested in mouse S49 lymphoma cells (Insel and Fenno, 1977, 1978), with results opposite to those hitherto described. Incubation of cells of the wild-type of this lymphoma with dibutyryl CAMP, after an initial fleeting increase, profoundly and progressively decreased both ODC and SAMD activities to barely detectable levels (Insel and Fenno, 1977, 1978). Inhibition of ODC activity in S49 lymphoma cells is an early response to other agents that increase intracellular cAMP level, i.e., isoprotereno1, cholera toxin, or PGEl (Honeysett and Insel, 1980). These decreases in cellular ODC and SAMD levels were causally connected with a parallel decrease in the activity of CAMP-dependent protein kinase, since in “kinasenegative” mutant cells, that is, in a cell line totally lacking in CAMPdependent protein kinase activity and therefore in response to dibutyryl CAMP, no decrease in the levels of the two polyamine biosynthetic decarboxylases was observed after addition of this cyclic nucleotide to the culture medium (Insel and Fenno, 1977, 1978). On the contrary, the fall in ODC activity induced in the wild-type S49 cells by cAMP was not accompanied by a progressive decrease in protein synthesis, demonstrating that the two biochemical events are dissociable (Insel and Fenno, 1977, 1978). An analogous split between the decrease in the activities of the polyamine biosynthetic decarboxylases and the cell killing induced by cAMP is also possible, using the “CAMP-deathless” mutants, that is, mutants phenotypically resistant to the cytolysis induced by cAMP (Insel and Fenno, 1977, 1978; Kaiser et al., 1979). Furthermore, treatment with dibutyryl cAMP caused similar decreases in ODC activity in the ‘‘CAMP-deathless” mutants, regardless of the phase of the cell cycle at which the distinct cell populations were examined (Kaiser et al., 1979). Instead, Bachrach (1980b) found that ODC activity was induced by cAMP in cultures of S49 lymphoma cells, but not in a mutant line defective phenotypically in CAMP-dependent protein kinase activity. The aspect, which is converse and complementary to that hitherto analyzed, of the connections between polyamine biosynthesis and cyclic nucleotides, i.e., whether polyamines can modulate the metabolism of the cyclic nucleotides inside the eukaryotic cells, has been scarcely investigated. In spite of this, some evidence is emerging that pol yamines really can regulate the synthesis of the different cyclic


17

nucleotides. Indeed, spermine dramatically inhibited the activity of CAMP-dependent protein kinase activity in glioma cells, and the inhibition was shown not to be d u e to an interaction of the tetraamine with the regulatory subunit of the enzyme (Bachrach et al., 1978; Bachrach, 1980b). Addition of any of the three chief polyamines has been shown to cause a decrease in cAMP concentrations in cultured glioma or neuroblastoma or neuroblastoma x glioma hybrid cells, either unstimulated or stimulated with hormones or drugs (such as norepinephrine, isoproterenol, PGEI, adenosine, IBMX), which are well-known agents for inducing cAMP accumulation inside the cells (C16 e t al., 1979). It is of special interest that a decrease occurred when the exogenous polyamine was added even at low concentrations in the range of those found in physiological fluids (Cl6 e t aZ., 1979). Paradoxically, polyamines at higher concentrations caused a slight increase in the intracellular cAMP levels of the cultured neural neoplastic cell lines (C16 e t al., 1979). Last, a report shows that spermidine and spermine and, to a lesser extent, putrescine are effective inhibitors of the activity of specific cCMP phosphodiesterase obtained from leukemic L1210 cells (Bloch and Cheng, 1979). Friend erythroleukemia cells are a relatively pure population of virus-transformed mouse hematopoietic cells. These cells can be induced by a variety of chemical agents with different biological properties to differentiate to orthochromatic or polychromatic normoblasts and are a suitable experimental system for studying the biochemical events involved in cell differentiation. Interestingly enough, ODC activity can be rapidly induced in this cell line by some inducers of differentiation, such as dimethyl sulfoxide (Tsiftsoglou and Kyriakidis, 1979; Gazitt and Friend, 1980). The ODC induction was observed when the cell differentiation process was blocked or when the inducers were added to cell lysates (Tsiftsoglou and Kyriakidis, 1979). However, the cell differentiation process caused by the inducers in the Friend erythroleukemia cells appears to be not at all mandatory or causal for ODC induction, since some potent inducers, like actinomycin D or aminonucleoside of puromycin, do not stimulate ODC (Gazitt and Friend, 1980). Two final conclusions seem to us to be appropriate. First, the induction of O D C activity appears to be a common and easily observable phenomenon in cultured neoplastic cells, in striking contrast with what has been observed in uivo in neoplastic organs and in organs undergoing chemical carcinogenesis (Scalabrino e t al., 1978). Second, the mediation of CAMP claimed to be a general and possibly obligatory step in ODC induction appears, on the basis of the studies carried out with in uitro systems, to be so in some instances, but to be totally

18


extraneous in some others, and this is in quite good agreement with what has emerged from in vivo studies of ODC inducibility.

D. EFFECTSO F

INFECTION WITH

NONONCOGENICVIRUSES

Infection of HeLa cells with vaccinia virus brought about both quantitative and qualitative changes in ODC activity inside the cells, as was demonstrated by Hodgson and Williamson (1975). The ODC was induced and rose quickly following viral infection and, what is even more important, the mean K , value for ODC was significantly lower in infected than in uninfected cells (Hodgson and Williamson, 1975). On the contrary, at later postinfection times the biosynthesis of all the polyamines, including cadaverine, was greatly reduced, but not completely inhibited, in HeLa cells infected with vaccinia virus (Lanzer and Holowczak, 1975). Substantially the same thing was observed in KB cells infected with type 5 adenovirus, at long intervals after infection (Pett and Ginsberg, 1975). This time course for polyamine biosynthesis and concentration in neoplastic cells infected with nononcogenic viruses, with first an increase during early infection and then a decrease at late postinfection times, was also seen in Ehrlich ascites tumor cells infected with mengovirus (Egberts et al., 1977). By way of conclusion, let us tentatively compare the polyamine response of neoplastic cells infected by nononcogenic viruses with the responses observed in normal cells undergoing neoplastic transformation by oncogenic viruses. We can state that (a) there is an increase in cell polyamine biosynthesis immediately after the infection in both types of viral cell infection; (b)there is a clear dichotomy in late phases of postinfection time between the two types of viral cell infection, since the cell polyamine biosynthesis remains at high levels in the neoplastic viral transformation process (see Section IV, Part I, Vol. 35) and decreases progressively after nontransfonning infection of neoplastic cells by nononcogenic viruses.

E. MISCELLANEOUS EFFECTSO F POLYAMINES Most reports on the effects of polyamine addition to cultures of some neoplastic cell lines deal with protein synthesis and cell proliferation. Spermine stimulated poly(UG)-dependent phenylalanine incorporation in a subcellular protein-synthesizing preparation obtained from L1210 mouse ascites leukemic cells, and the stimulation was beyond

POLYAMINES IN MAMMALIAN TUhIOHS

19

that achieved with optimal magnesium concentrations, suggesting that spermine may act as more than merely a substitute for magnesium (Ochoa and Weinstein, 1964). All three chief polyamines (with spermine the most effective) had stirnulatory effects on tRNA methylases in extracts of L1210 cells (Hacker, 1973). Exogeneous spermidine and spermine stimulated the incorporation of orotic acid into RNA and considerably decreased the degradation of the newly synthesized RNA in Ehrlich ascites cells (Raina and Janne, 1968; Khawaja and Raina, 1970). The presence of spermine was essential for the translation in a cell-free system derived from wheat germ of tyrosine aminotransferase mRNA from hepatoma cells (Rether et al., 1978). Again, spermine could partially substitute for soluble factors present in dexamethasoneinduced hepatoma tissue culture that stimulate in a homologous cellfree system the translation of mRNA coding for tyrosine aminotransferase (Beck et al., 1978). In Walker 256 carcinosarcoma cells, putrescine and spermidine preserved the ultrastructural morphology of all nuclear structures, including the nucleolus (Busch et al., 1967). In HeLa cells, polyamines were present in abundant quantities in the chromosome cluster region (Goyns, 1979) and have been shown to stimulate the nuclear synthesis of the histone Hl-poly(ADP-ribose) complex (Byrne et al., 1978).The intercellular adhesiveness of HeLa cells harvested from densityinhibited suspension cultures was markedly enhanced by the addition of putrescine to the medium in which the cells were resuspended (Deman and Bruyneel, 1977). In contrast, the diamine did not modify the mutual adhesiveness of cells harvested from fast-growing cultures (Deman and Bruyneel, 1977). Polyamines have been found also to have some inhibitory effects on cultured neoplastic cells. Spermine depressed protein synthesis in Walker 256 carcinosarcoma cells (Goldstein, 1965). This tetraamine is distinctly cytotoxic for different hepatoma cell lines, and the effect was noticeably enhanced by the presence of fetal calf serum in the growth medium (Katsuta et al., 1975). Among the cytotoxic metabolites released from rat ascites hepatoma cells into culture fluid, some closely resembled spermine in chemical nature (Katsuta et al., 1974). Spermidine, putrescine, and cadaverine all inhibited replicative DNA synthesis in mouse ascites sarcoma cells (Seki et al., 1979).The addition of spermidine or spermine to the medium inhibited the growth of cultured human meningioma cells, whereas putrescine had a slight opposite effect (Duffy et al., 1971). Granulocytic chalone, but not the polyamines, inhibited [3H]TdR uptake in rat chloroleukemia cells in short-term cultures (Foa et al., 1979). This result favors the idea that

20


this activity of granulocytic chalone does not depend on its possible polyamine content. The reciprocal connections between synthesis of polyamines and that of 5’-methylthioadenosine (MTA) were evidenced in a human leukemic cell line lacking 5‘-methylthioadenosine phosphorylase (Kamatani and Carson, 1980). The addition of spermine or spermidine markedly depressed the synthesis of MTA, whereas the addition of MTA stimulated putrescine synthesis at low concentration or inhibited it at high concentrations (Kamatani and Carson, 1980). Exogenous MTA also depressed the intracellular levels of spermine in leukemic cells but not, very surprisingly, those of spermidine (Kamatani and Carson, 1980), although MTA is a well-known inhibitor of both spermidine and spermine synthetases (see Section I,D, Part I, Vol. 35). Nevertheless, these relationships between polyamine and MTA metabolisms in neoplastic cells await further experimental elucidation. Finally, spermine precipitated a cell-surface protein, fibronectin, from the culture medium into which it had been secreted by a human rhabdomyosarcoma cell line (Vuento et d.,1980). This observation raises the possibility that polyamines have a role in the deposition of fibronectin in vivo. This is of mounting interest, since many types of malignantly transformed cells, unlike the normal adherent cells, generally deposit small amounts of the surface fibronectin in the pericellular matrix, and this scarcity has been causally correlated with some malignant behavioral properties, which have been overcome by the addition of fibronectin to cultures of tumor cells. Whether elevated polyamine production and secretion (when present) and fibronectin scarcity are interconnected with other factors in determining the well-known poor adhesiveness of neoplastic cells, remains hypothetical, speculative but very attractive. II. Polyamines in Human Oncology

Theoretically, the need to diagnose and subsequently to locate a tumor as early as possible has always been considered to be a key goal for antineoplastic therapy. The expectation that this would be possible was kept alive by new findings, in both experimental and clinical oncology, which showed that certain kinds of neoplasias produce unusual metabolites. Many tumor-cell products, collectively called neoplastic markers,” or “tumor-associated markers” or “oncodevelopmental markers,” have been identified in blood, effusions, urine, and cerebrospinal fluid of tumor-bearing patients and in neoplastic tissue extracts. “


21

Measurements of these products have been employed extensively in clinical medicine for both initial diagnosis of neoplasia and monitoring tumor recurrence after different types of therapy. In fact, it was hoped that changes in the amounts of these “markers” in one or more of the different physiological fluids of cancer patients or in the neoplastic tissue could be used to reflect changes in the body’s tumor burden, since these products can be either secreted into the surrounding milieu or kept within the neoplastic cells. The most widely employed “markers” for neoplastic growth and tumor dedifferentiation or differentiation are listed in Table I. As new metheds for the measurement of tumor “markers” were introduced and became ever more sensitive, and clinical studies ever larger, the expectation that these “markers” would be of key importance or, at least, very useful for early diagnosis of neoplasia did not fully materialize. In fact, elevated levels of “ markers” were observed in a variety of nonneoplastic diseases, and, conversely, some neoplasms were not accompanied by any known marker.” Some “markers” have even been found in a percentage of normal healthy adults. In addition, some of these “markers” are also found in the earliest stages of human development, and this association has led to the widely accepted practice of referring to most neoplastic “markers” as oncofetal proteins and antigens (Sell, 1980). The divergence between expectation and practical results was explained after extensive studies of the biochemistry of cancer, which have taught as that tumor cells do not synthesize tumor-specific substances, i.e., substances never found in normal cells at any step in their differentiation (Wolf, 1979a). What is characteristic of tumor cells is that they either express certain normal gene information at the wrong time or in the wrong place or in the wrong amount, or completely fail to express some normal genes (Wolf, 1979a,b). Moreover, during recent decades, it has become ever clearer that neoplasia is not one single type of disease, but a group of very many diseases, each utterly different from the others from the clinical and biochemical points of view, with the only common features that they are lethal to the host and have a cell growth type that is invasive and can never b e stopped definitively. Some frequently found discrepancies between the amount of tumor “marker” present and the degree of growth of the neoplasm must be connected with phenotypic expression of these “markers,” which varies from cell to cell within the neoplasm. In fact, quantitative and qualitative variations in the production of “markers” may occur during the natural course of the malignancy. In other words, during the development of a tumor from preneoplasia to early neoplasia to advanced ‘I

MAIN DIFFERENT BIOCHEMICAL AND

TABLE I “MARKERS”OF NEOPLASTIC GROWTH AND TUMOR DEDIFFERENTIATION DIFFERENTIATION USED I N CLINICAL ONCOLOGY

IMMUNOLOGICAL OR

Products Acute-phase reactant proteins (APRPs): a,-Antitrypsin, a,-antichymotrypsin, ceruloplasmin, C-reactive protein, haptoglobins, fibrinogen Chromosomal abnormalities: Ph’, 13 q-, 14 q+ Cyclic nucleotides: CAMP, cGMP, ratio CAMP: cGMP Enzymes or isozymes: Leucine aminopeptidase, y-glutamyl transferase, copper oxidase, creatine kinase BB, histaminase (DAO), muramidase, galactosyl transferase 11, lysozyme, ribonuclease, arylsulfatase A, reverse transcriptase, terminal deoxynucleotidyl transferase (TdT), superoxide dismutase. Glycolytic isozymes: (a) glucose phosphate isomerase, (b) aldolase (shifting vs A form), (c) LDH (shifting in the isoenzyme pattern from LDH-1 to LDH-5 part of the isoenzyme spectrum). Phosphohydrolases: (a) acid phosphatase, (b) alkaline phosphatase, (c) 5’-nucleotidase Hormones, isohormones, fragments or catabolites of hormones: Ectopic production of hormones (paraneoplastic syndromes): ACTH, gonadotropins, ADH, PTH, ILA, TSH, erythropoietin, MSH, HGH, HPL, HCG, PL, PGA, PGE, CT; catechqlamines, metanephrine, vanilly1 mandelic acid; 5-HIAA, 5-HT, 5-HTP, bradykinin.

References Cooper and Stone (1979) Purtilo et al. (1978) Pardee et al. (1978); Pastan et al. (1975) Bodansky (1975); Bollum (1979); Fishman (1974); Fishman and Singer (1975); Goldberg (1979); Kaplan (1972); Oberley and Buettner (1979); Ruddon (1978); Schapira (1973, 1978); Uriel (1975, 1979); Weber (1977);Wolf (1979b); Yam (1974)

Hall (1974); Ode11 and Wolfsen (1975); Rees and Ratcliffe (1974); Ruddon (1978); Seyberth (1978); Sherwood and Could (1979); Wolf (1979b)

Immunoglobulins: Homogeneous (monoclonal) immunoglobulins (M components); Bence Jones proteins ( K or A light chains); abnormal or incomplete heavy chains: (a) y-chain subclasses (yl, yz. y3, y4) of IgG; (b) a-chain subclasses ( a l , a z )of IgA; (c) p-chains Miscellaneous proteins: Fetal hemoglobin, EDC1, milk casein. Placental and pregnancy proteins: (a) SP, pregnancy-associated a,-glycoprotein (aZPAG); (b) SP, pregnancy-specific &glycoprotein; ( c ) PPTPP8 (ubiquitous tissue) proteins; (d) PPs, placental protein five. Plasminogen activators Oncofetal proteins and antigens: a-FP, CEA, FSA, a2H-ferroprotein; pancreatic oncofetal antigen (POA), p-oncofetal antigen (BOFA), OFA, glial fibrillar acid protein (GFAP)

E3

w

Polyamines and their biosynthetic decarboxylases: Pubescine, spermidine, spermine, ornithine decarboxylase, S-adenosyl-Lmethionine decarboxylase Sterols: Desmosterol (cholesta-5,24-dien-3-P-ol, or 24-dehydrocholesterol)

Bodansky (1975); Solomon (1977); Waldenshom (1976)

Bohn (1980); Ruddon (1978); Rudman et al. (1976, 1977)

Fritsche and Mach (1975); Lehman (1979); Loewenstein and Zamcheck (1977); Martinet al. (1976); Ruoslahti and Seppala (1979); Seidenfeld and Marton (1979b); Sell and Becker (1978); Uriel (1975, 1979); Wikstrand and Bigner (1980) Bachrach (1976a); Janne et al. (1978); Milano et al. (1980); Russell (1977); Russell and Durie (1978); Savory and Shipe (1975); Scalabrino et al. (1980); Seidenfeld and Marton (1978, 1979b) Seidenfeld and Marton (197913);Wikstrand and Bigner (1980)

24

GIUSEPPE SCALARRINO AND MARIA E . FERIOLI

metastatic tumor, heterogeneous subvariant cell populations emerge within single clones. Heterogeneity of tumor cell populations may lead not only to the loss of some “marker(s)” but also to the emergence of new ones (Wolfe, 1978; Wolf, 1979a,b). Last, but not least, it is worth mentioning here that some naturally occurring cell labels, such as the glucose-6-phosphate dehydrogenase (G-6-PD) system and the surface-associated immunoglobulins, generally used for identifying different normal cell subpopulations, can also be employed to determine whether a given neoplasm has a single or multiple cell origin, providing an important clue to the initiating event (Fialkow, 1974). The polyamines, although they suffer from the same drawbacks listed above for the other neoplastic “markers,” are widely considered to b e clinically useful “markers” of neoplastic growth and for cancer diagnosis, and particularly for evaluation of the success or failure of an antineoplastic therapy. It is largely accepted to differentiate the different tumor cell “markers” into (a) those produced by dedifferentiation of neoplastic cells (e.g., carcinoembryonic antigen, a-fetoprotein, alkaline phosphatase isozyme) and (b) those produced as a result of overproduction by tumor cells or of the increased tumor cell multiplication (e.g., acid phosphatase, those hormones secreted by specific endocrine-gland neoplasms). Polyamines, for reasons discussed later, have to be included in the second group of neoplastic “markers.” The importance and the clinical significance of polyamines in human oncology have been well reviewed several times by Russell (1973a, 1977), Savory and Shipe (1975), Bachrach (1976b), Cohen (1977), Janne et al. (1978), Russell and Durie (1978), Seidenfeld and Marton (1978, 1979b), Buehler (1980), Durie (1980), and Milano et al. ( 1980). Therefore, we aim here to outline the current “state of the art” about the connections between polyamines and human cancer, together with some recently obtained advances, and to draw particular attention to the use of the levels of activity of the polyamine biosynthetic decarboxylases (PBD) as biochemical indicators of the growth rate, and consequently of the malignancy, of some types of human neoplasias. A. PATTERNS OF POLYAMINES I N HUMANNEOPLASTICTISSUES Hamalainen (1947) made the pioneering observations in this field, systematically screening spermine contents in a great number of organs obtained postmortem from patients who had died of different types of neoplasia. He found an elevated spermine content in the lung


25

of a patient who died of lung carcinoma, in the uterus of a patient who died of uterine carcinoma, and in livers, spleens, and bone marrow of two patients who died of leukemia. These observations were subsequently extended to a variety of human malignancies by other authors. Including the more recent reports on the polyamine content of human tumors, it has become ever more evident that there is no general or unique pattern for the polyamine content of human neoplasias. In fact, brain tumor tissues (e.g., neurofibroma, meningioma, glioblastoma, astrocytoma, glioma) have as their particular biochemical feature very high putrescine concentrations in comparison with both gray and white areas of normal human brain (Kremzner, 1970, 1973; Kremzner et al., 1970).On the contrary, the levels of spermidine and spermine in the tumors studied did not greatly differ from levels observed in normal brain, the only exception being high concentrations of spermidine and spermine in astrocytoma and in glioma (Kremzner, 1970, 1973; Kremzner et al., 1970). Additionally, human tumor tissue in vitro and meningioma cells grown in culture actively incorporated [I4C]putrescine into spermidine and spermine, but showed low deaminating activity (Kremzner et al., 1972). The high levels of putrescine in several central nervous system-related tumor tissues have been confirmed (Harik et al., 1978; Harik and Sutton, 1979). Furthermore, it has also been demonstrated that the magnitude of the elevation of putrescine content in the astrocytoma groups is proportional to the degree of malignancy of the tumor as determined by conventional histopathological criteria. A variety of slowly growing and relatively benign intracranial or intraspinal tumors (such as meningioma, cerebellar hemangioblastoma, chordoma, neurofibroma, schwannoma) had low levels of putrescine that in many instances did not exceed the range in samples from normal brain tissue (Harik et al., 1978; Harik and Sutton, 1979). On the other hand, the tissue concentrations of spermidine and spermine varied broadly within normal cerebral cortical samples and within the various tumor types, with no obvious correlation with the degree of malignancy of the tumor (Harik et al., 1978; Harik and Sutton, 1979). Therefore, it can be tentatively concluded that, at least among the astrocytomas, the putrescine level may be a reliable biochemical “marker” not of the tumor per se, but of the degree of malignancy of the tumor. However, amazingly enough, high putrescine levels have been detected in papillary adenocarcinomas of the thyroid, which are the most clinically benign and extremely slow growing of all thyroid malignancies (Matsuzaki et al., 1978). Among the renal cell carcinomas, the concentration of spermidine in

26


the poorly differentiated types (grades 3 and 4) was significantly higher than in the well-differentiated types (grades 1 and 2), and in both it was always higher than in normal renal tissue (Matsuda et al., 1978). Thus, in renal carcinomas the concentration of spermidine correlates well with the degree of tumor malignancy ascertained by histopathological known criteria, particularly the nuclear atypia. In the same types of renal tumors, another supposed “marker” for the growth rate of neoplastic cells, i.e., the spermidine : spermine ratio (Russell, 1973b), progressively increases from the normal renal tissue to the poorly differentiated type of renal carcinoma, with the ratio for the well-differentiated type of tumor in the middle (Matsuda et al., 1978). This significant increase in the spermidine : spermine ratio in human renal adenocarcinomas was also confirmed by other authors (Dunzendorfer and Russell, 1978,1979,1980).The concentration of spermidine in neoplastic tissue was significantly higher than in the histologically normal areas of the same kidneys, while the spermine content of the tumor was generally lower than that of normal tissue (Dunzendorfer and Russell, 1978, 1979, 1980). On the contrary, thyroid adenocarcinomas have ratios of spermidine to spermine very close to those found in normal thyroids or those affected by other nonneoplastic diseases (Matsuzaki et al., 1978). The cellular content and secretion of polyamines, in relation to the cell cycle and proliferation kinetics, have been investigated in uitro with cultured cells of Burkitt’s lymphoma (Woo et al., 1979),which is a rapidly growing malignant tumor characterized by a high rate of cell proliferation combined with a high growth fraction. As for the time course of the intracellular contents of polyamines during the growth of Burkitt’s lymphoma cells, large amounts of spermidine and spermine and lower amounts of putrescine were observed during the lag and early exponential growth phases (Woo et al., 1979). This trend was reversed when the cultured cells entered the exponential growth phase and early plateau growth, since spermidine and spermine contents markedly decreased, while the putrescine content tripled (Woo e t al., 1979). The ratio of spermidine to putrescine and that of spermine to putrescine were significantly and positively correlated with both the labeling index and the specific growth rate, whereas there was no significant variation in the spermidine : spermine ratio throughout the growth period (Woo et al., 1979). As for the changes in the cellular polyamine content during the cell cycle, the cell fraction in GI showed a significantly high positive correlation with the intracellular content of putrescine, while negative correlations were calculated for spermidine and spermine (Woo et al., 1979).All these results

POLYAsfINES I N MAMILIALIAN TUMORS

27

seem to suggest that all the chief polyamines actively participate in the process of proliferation of Burkitt’s lymphoma cells. Last, a phospholipid-the so-called malignolipin-containing spermine has been found in some human malignant tumors [e.g., seminoma, gastric cancer, cancer of the colon, uterine cancer, breast cancer (Kosaki et al., 1958)] and in bloods of cancer patients (Kallistratos et al., 1970), but never in normal tissues. This phospholipid contains, in addition to spermine, choline, phosphoric acid, and fatty acids. Although this malignolipin was discovered several years ago, the function and the biological significance of such a compound in tumor development and in tumors of high malignancy remains to be determined and awaits more rigorous demonstration (Bachrach and Ben-Joseph, 1973). B. LEVELSO F THE CHIEF POLYAMINES A N D THEIR CONJUGATED FORMS I N URINES OF NORMAL SUBJECTS A N D OF CANCER PATIENTS Observation of the high polyamine contents in neoplastic tissues at once stimulated looking for increased quantities of these polycationic substances in the extracellular fluids of patients with malignancies. Unlike the neoplastic tissues, the human body fluids generally contain small quantities of polyamines, so that the quantification of these substances in these fluids requires highly sensitive methods. In the last decade noticeable improvements in the assay methods for polyamines have been achieved (Seiler, 1977, 1980). The assay methods most widely used at present for quantitative determinations of polyamines and their derivatives are thin-layer chromatography of dansylated polyamines, automated ion-exchange chromatography, high-pressure liquid chromatography, gas chromatography, and radioimmunological assay. All these methods are sensitive and accurate enough to detect very small amounts of polyamines in both physiological fluids and in biopsy material. Therefore, routine screening for polyamine levels in fluids of human beings with different types of pathologies, whether or not characterized by uncontrolled cell proliferation, is now possible. The data available in the literature on the daily urinary excretion of polyamines and of their conjugated forms by normal subjects are reported in Tables I1 and 111, where the values have been divided into groups according to the units of measurement used by the various authors to express their results. Those data reported by some authors as control levels, but taken from hospitalized patients with nonneoplastic diseases, have deliberately not been reported in Tables I1 and 111,

TABLE I1

NORMAL DAILYCONTENTS Unit mgl24-hr U

rmoll24-hr U

pmollkgl24-hr U

POLYAMINES IN HUMANURINE"

N

Putrescine

Spermidine

Spermine

2 50 50 10 5 12 8 6 42 NR 42 50 56 21 8 20 9

2.5 2.7 2 0.53 2.7 2 0.5 2.5 f 0.6 2.0 1.4 0.94 2.2 3.52 0.2-2.84 0.8-6.2 3.52 4.21 f 0.41 0.89 1.52 0.98 0.49 1.6 2 0.4 1.6 (9) 1.57 2 1.05 21.9 2 7.6 9.8 2.0 0.5 0.38 2 0.017 0.2 0.4 0.38 2 0.17 0.7

2.7 3.1 0.56 3.1 2 0.6 2.4 f 0.4 1.5 1.3 0.86 1.6 2.44 0.36-2.1 0.9-3.9 2.44 1.12 0.11 0.53 0.83 4.57 2 1.02 0.2 2 0.04 0.3 (7) 0.51 0.16 8.4 f 2.1 7.6 2 2.5 0.2 0.1 f 0.003 0.12 0.11 0.11 f 0.04 1.5

2.5 3.4 2 0.67 3.4 2 0.7 0.4 2 0.2 48

0 1 -1 2

5 1 2 -

6 0

22 4 9 -

6 1 0 -

35

7

0 0 1 1

8

46 52

" References: Reeves et al. (1972); Knight et ul. (1974);Hsu et ul. (1974); Tsuchimoto et al. (1974); Berger (1975); Westin (1976a,b); Hsu et al. (1977); Nowell and Finan (1978); Berger and Bernheim (1979); Van Den Berghe et a / . (1979a); D. Van Dyke

(personal communication); plus references listed in Table V. * Analyses done during a leukemic phase were excluded. 'All patients were pseudodiploid. These patients are from studies in which clinical details were sparse and the time of cytogenetic analysis (i.e., pre- or posttreatment) was not reported.

modal numbers. The abnormal cells from 72% (28/39) of the known untreated patients had hyperdiploid modal numbers, whereas abnormal cells from 78% (46/59) of the known treated patients were pseudodiploid.

C. NONRANDOMABNORMALITIES 1. Polycythemic Phase The pattern of chromosomal changes in PV is nonrandom. Figure 5 summarizes the karyotypic abnormalities seen during the polycythemic phase in 48 patients from Table VI who were studied with banding techniques, as well as changes in two other patients, who are not listed in Table VI because the modal numbers were not included in the original report. Five patients from Table VI who were studied with banding are excluded in Fig. 5: 4 were known to have developed myelofibrosis prior to the initial analysis; the time of analysis was unclear in a fifth patient who developed myelofibrosis. In the polycythemic phase, the nonrandom distribution of chromosomal changes is particularly evident in the gains, which usually involved chromosomes 8 and 9. Eleven patients had a gain of No. 9, and 9 patients had a gain of No. 8. Four of these patients showed gains of

136

JANET D. ROWLEY AND JOSEPH R. TESTA

FIG.5. Histogram of clonal karyotypic abnormalities seen in 50 untreated and treated polycythemia Vera patients. Analyses done during a transitional or a leukemic phase are omitted here. Each box represents a clonal abnormality seen in a single patient. “Untreated?’ patients are those for whom the time of cytogenetic examination (either pre- or posttreatment) was not specified by the authors in the original report.

both No. 8 and No. 9 (Westin et al., 1976; Testa et al., 1981).Clones containing both + 8 and + 9 are seldom observed in other hematologic diseases and may be unique to PV (Testa, 1980). Chromosomes No. 20 (17 patients) and No. 1 (10 patients) were the most frequently rearranged chromosomes seen in the polycythemic phase (Fig. 5). All but one of the rearrangements of No. 20 involved a 2Oq-. Thus, a 20q- was present in 32% (16/50)of the aneuploid PV patients summarized here. It had been thought initially that the 20qabnormality might have some diagnostic value in PV (Wurster-Hill et al., 1976); however, whereas a 20q- is a relatively frequent finding in PV, it has now been observed in patients with various other myeloid disorders as well (Testa et al., 1978b). Most (8/10 shown in Fig. 5) of the reported abnormalities of No. 1 in PV have consisted of trisomy of all or part (1q21-qter) of the long arm. Rowley (197%) and Gahrton e t al. (1978) have noted that trisomy of l q , especially of bands 1q25-32 and 1q23-25, can be found frequently in various hematologic diseases.

137

CHROMOSOME ABNORMALITIES IN MALIGNANT DISEASES

aLeukemic phose (I5cosesl Tront~l~onol (I1 cases) Unhealed Observed In polycylhemc phose 0s well

8 6

z

0 a

4

2

2

41 6

10 8 6 4

2 I

2

3 4

5

6

7

8

9 10 II 12 13 14 15 16 17 I8 19 20 21 22 X

Y

CHROMOSOME IDENTI FlCATlON

FIG.6. Histogram of clonal karyotypic abnormalities seen in 26 polycythemia Vera (PV) patients (25 treated) during a transitional or a leukemic phase. Abnormalities that were first observed earlier during the polycythemic phase in 5 patients (of 12 examined during that stage) are indicated by an asterisk (*). Unlike the pattern seen in the polycythemic phase, rearrangement of No. 5 and loss of No. 7 are quite common in the more advanced stages of PV.

Structural rearrangements were seen primarily in treated patients. For example, 14 of 17 patients with a rearranged No. 20 were treated prior to their first cytogenetic analysis. In contrast, most chromosome gains are found in untreated patients.

2. Transitional and Leukemic Phases There have been a number of PV patients with advanced disease who have been studied with banding techniques (Hoppin and Lewis, 1975; Stavem et al., 1975; Nowell et al., 1976; Westin e t al., 1976; Zech et al., 1976a; Weinfeld et al., 1977; Berger and Bernheim, 1979; Van Den Berghe et al., 1979a; Hagemeijer et al., 1981; Testa et al., 1981). Figure 6 summarizes the karyotypic abnormalities observed in 26 patients during the advanced stages of PV. Included here are abnormalities found during the leukemic phase or during a “transitional phase.” As used here, “transitional phase” refers to the development, during disease progression, of myelofibrosis, myelofibrosis with myeloid metaplasia, and/or increased granulocytic immaturity.

138


Whereas the cytogenetic pattern seen during the advanced disease stages shows some similarities (e.g., +8 and +9) to that found in the polycythemic phase, there are certain striking differences as well. For example, loss of No. 7 was never observed in the polycythemic phase, but was seen in 5 of 26 patients in transitional or leukemic phases. Rearrangements of No. 5 were the most frequent change seen in advanced disease stages, but seldom occurred in the less aggressive polycythemic phase. Rearrangements of No. 20 were less frequent in advanced disease and were found in only I patient who developed overt leukemia. Moreover, structural rearrangements of No. 12, which are rare in the stable phase (Fig. 5), are more frequent in the leukemic phase (Fig. 6). A 5q- may be a specific abnormality frequently associated with the terminal stage of PV (Testa et aZ., 1981). Nine of the 10 structural rearrangements of No. 5 shown in Fig. 6 were 5q- anomalies. In contrast, only one of three rearrangements of No. 5 observed in the polycythemic phase (Fig. 5) was a 5q-, and this was observed in a patient whose chromosomes were examined only very late in the disease course by Testa et al. (1981). A 5q- has been observed in the leukemic phase of five PV patients. Van Den Berghe et aZ. (1979a) reported on three patients with long-standing PV who first showed a 5q- late in the disease course;the abnormality accompanied the transition to myelofibrosis with myeloid metaplasia and the appearance of a preleukemic disorder. However, Testa et al. (1981)observed a 5q- in only two of eight PV patients with documented myelofibrosis. Thus, it appears that the 5q- anomaly may be associated more with disease progression in PV than with the actual development of myelofibrosis. Whereas the 5q- abnormality may be a marker of refractory anemia or, possibly, of preleukemia or of an early nonproliferative leukemia (Van Den Berghe et al., 1976), the presence of a 5q- plus other chromosome changes is nearly always associated with overt leukemia (Van Den Berghe et al., 1979a). In each of the 5q- patients summarized in Fig. 6, multiple chromosomal abnormalities were present. Thus, in PV currently available data indicate that a 5q-, in combination with other karyotypic changes, signals a terminal phase of the disease, which may involve anemia and other progressive changes, or transformation to myelofibrosis or overt leukemia. D. RELATIONSHIP OF TREATMENT TO ABNORMALKARYOTYPES

The reasons for the higher incidence of chromosome abnormalities in previously treated, as compared to untreated, PV patients are unknown. Westin (1976b) suggested that many of the treated patients


139

have had the disorder for a longer time than untreated ones prior to cytogenetic study, and data of Testa et al. (1981) support this suggestion. Furthermore, our study also showed that clonal abnormalities were seen primarily in patients who had had prior radiotherapy, and the incidence tended to correspond to the total prior dose of 32P.Similarly, Visfeldt et al. (1973) observed earlier that abnormal karyotypes occurred almost exclusively in their patients who were treated with 32P, and generally only after treatment with more than 2.5 mCi per year over more than 2 years. Follow-up studies done by Westin and Weinfeld (1978) showed that patients with initially normal karyotypes did not develop chromosome changes when they were treated with phlebotomies alone. In contrast, during therapy with 32Por chlorambucil, new abnormalities occurred in some patients whose karyotypes had been initially normal, as well as in others whose karyotypes were initially abnormal. Westin (1976b) noted that uncertainty exists as to whether these evolutionary changes are part of the natural history of the disease, or whether the myelosuppressive therapy induces the changes or accelerates a susceptibility to their spontaneous development. This question may eventually be resolved as a result of investigations such as the already ongoing study of the Polycythemia Vera Study Group (Wurster-Hill et al., 1976), in which cytogenetic analyses are done throughout the disease course on patients randomized for therapy with 32P,drugs, or phlebotomy.

VII. Implications of Nonrandom Changes for Malignant Transformation

The evidence presented demonstrates that nonrandom chromosome changes are closely associated with a variety of human hematologic disorders. Similar associations have been identified in other human The tumors and in animal tumors as well (reviewed in Rowley, 1980~). changes consist of gains or losses of part or all of certain specific chromosomes and of structural abnormalities, most frequently relatively consistent translocations, that are presumed to be reciprocal. The nonrandom translocation that we observe in malignant cells would represent those that provide a particular cell type with a selective advantage vis-a-vis the cells with a normal karyotype. There is very strong evidence that many cancers, CML and Burkitt lymphoma, for example, are of clonal origin. This means that a particular translocation in a single cell gives rise to the tumor or to the leukemia that ultimately overwhelms the host. Other rearrangements may be neutral, and the cells therefore will survive, but will not proliferate differentially; and still others may be lethal and thus would be eliminated.

140


In such a model, the chromosome change is fundamental to malignant transformation. Two questions are raised by these observations. First, how do such chromosome changes occur; and second, why do they occur? There is very little experimental evidence that is helpful in answering either of these fundamental questions. They clearly provide a focus for future research.

A. PRODUCTION O F CONSISTENT TRANSLOCATIONS

The mechanism for the production of specific, consistent reciprocal translocations is unknown. Chromosome breaks and rearrangements may occur continuously at random and with a low frequency, and only those with a selective advantage will be observed. Alternatively, certain chromosome regions may be especially vulnerable to breaks and therefore to rearrangements. Nonrandom breaks occur in certain human chromosomes exposed to various mutagenic agents. In the rat, Sugiyama (1971) showed that a particular region on No. 2 was broken when bone marrow cells from animals given DMBA were examined. In man, however, trisomy for l q is not necessarily related to fragile sites (Rowley, 1977b).Thus, a comparison of the break points seen in hematologic disorders that involve balanced reciprocal translocations with those leading to trisomy l q revealed a clear difference in preferential break points, depending on whether the rearrangement resulted in a balanced or an unbalanced aberration. Other possible explanations depend on either (a) chromosomal proximity, since translocations may occur more frequently when two chromosomes are close together; or (b) regions of homologous DNA that might pair preferentially and then be involved in rearrangements. Many of the affected human chromosomes, e.g., Nos. 1, 9, 14, 15, 21, and 22, are involved in nucleolar organization that would lead to a close physical association. All partial trisomies that result from a break in the centromere of No. 1involve translocations of l q to the nucleolar organizing region of other chromosomes, specifically Nos. 9, 13, 15, and 22 (Rowley, 1977b). In the mouse, chromosome No. 15 also contains ribosomal cistrons (rRNA) (Henderson et al., 1974). Sugiyama et al. (1978) noted that, in rat neoplasms, translocation trisomies, other markers, and aneuploidy frequently involve Nos. 1, 2, 13, and 19, which are chromosomes with late-replicating DNA, and Nos. 3, 11, and 12, which have rDNA and late-replicating DNA. They have suggested that nucleolus-associated late-replicating DNA rather than rDNA is involved in the origin of nonrandom chromosome abnormalities.


14 1

On the other hand, if chromosome proximity or homologous DNA sequences were the mechanism, this should lead to an increased frequency of rearrangements such as the 9;22 or 8;14 or 15;17 translocation, in patients with constitutional abnormalities, but this has not been observed. One of us (J. D. R.) wrote to all investigators who listed patients with these and other consistent translocations seen in leukemia and lymphoma in “The Repository of Chromosomal Variants and Anomalies in Man (Borgaonkar and Lillard, 1980).” Of the 39,971 patients with anomalies listed in this registry only one had one of these consistent translocations as a constitutional abnormality (Ferro and San Romin, 1981).It is possible that either or both of these mechanisms are subject to selection; a translocation might occur because the chromosomes are close together, but only certain specific rearrangements might have a proliferative advantage which results in neoplasia and thus allows them to be detected. One other possible mechanism that should be considered concerns transposable genetic elements that can cause large-scale rearrangements of adjacent DNA sequences. These consist of controlling elements that have been found in maize (McClintock, 1961) and in Drosophila (Green, 1973), and of insertion sequences in bacteria (Nevers and Saedler, 1977). Not only do these elements exert control over adjacent sequences, but the type of control, that is, an increase or a decrease in gene product, is related to their position and orientation in the gene locus. Whereas they can cause nonrandom chromosomal deletions adjacent to themselves, these controlling elements can also move to another chromosomal location, and they may transpose some of the adjacent chromosomal material with them. The evidence for the presence of transposable elements in mammalian cells is tenuous, but a more precisely defined gene map is required for the detection of such nonhomologous recombinations. B. FUNCTION OF NONRANDOM CHANGES Our ignorance of how nonrandom changes occur is matched by our ignorance as to w h y they occur. Two points should be emphasized; one concerns the genetic heterogeneity of the human population, and the second, the variety of cells involved in cancer. There is convincing evidence from animal experiments that the genetic constitution of an inbred strain of rats or mice plays a critical role in the frequency and type of neoplasms that develop. Some of the factors controlling the differential susceptibility of mice to leukemia not only have been identified, but also have been mapped to particular chromosomes, and their behavior as typical Mendelian genes has been demonstrated

142


(Lilly and Pincus, 1973; Rowe, 1973).These genes have been shown to be viral sequences that are integrated into particular sites on chromosomes; these sites vary for different inbred mouse strains and for different murine leukemia viruses. Thus, the sites in AKR and C3H mice are two different loci on chromosome 7 (Rowe, 1973), and that in Balb/c mice is on chromosome 5 (Rowe and Kozak, 1979; Ihle et al., 1979). Certain genetic traits in man predispose to cancer, especially leukemia and lymphoma, such as Bloom syndrome, Fanconi anemia, and ataxia-telangiectasia (German, 1972). How many gene loci are there in man that, in some way, control resistance or susceptibility to a particular cancer? We have no way of knowing at present. These genes may influence the types of chromosome changes that are present in malignant cells. The second factor affecting the karyotypic pattern relates to the different cells that are at risk of becoming malignant, and the varying states of maturation of these cells. The catalog of the nonrandom changes in various tumors maintained by Mitelman and Levan (1978) provides clear evidence that the same chromosomes, for example, Nos. 1 and 8, may be affected in a variety of tumors. On the other hand, some chromosomes seem to b e involved in neoplasia affecting a particular tissue; the involvement of No. 14 in lymphoid neoplasms and of No. 20 in myeloid-particularly red-cell-abnormalities might be suitable examples. All of the consistent translocations are relatively restricted to a particular cell lineage. Given the great genetic diversity, the number of different cell types that might become malignant, and the variety of carcinogens to which these cells are exposed, it is surprising that nonrandom karyotypic changes can be detected at all.

C. CONCLUSIONS The relatively consistent chromosome changes, especially specific translocations, that are closely associated with particular neoplasms provide convincing evidence for the fundamental role of these changes in the transformation of a normal cell to a malignant cell. In some tumors, these changes are too small to be detected, and the cells appear, with present techniques, to have a normal karyotype. When one considers the number of nonrandom changes that are seen in a cancer such as ANLL, it is clear that not just one gene, but rather a class of genes, is involved. Our knowledge of the human gene map has developed concurrently with our understanding of the consistent chromosome changes in neoplasia (McKusick and Ruddle, 1977). It is now possible to try to correlate the chromosomes that are affected with


143

the genes that they carry. Clearly, these efforts are preliminary, since relatively few genes have been mapped, and since some of the chromosomes that are most frequently abnormal have few genetic markers. In such a preliminary attempt, Rowley (1977b) observed that chromosomes carrying genes related to nucleic acid biosynthesis, and also the specific chromosome region associated with these genes, were frequently involved in rearrangements associated with hematologic cancers. In the future, we will be able to determine the break points in translocations very precisely, to measure the function of genes at these break points, and to compare the activity of these genes in cells with translocations with their activity in normal cells. Such information will be the basis for understanding how chromosome changes provide selected cells in certain individuals with a growth advantage that results in malignancy.

ACKNOWLEDGMENTS The results presented in this article were obtained during research supported in part by the Department of Energy, No. DE-AC02-80EV10360, and by grants supported by PHS Grants Nos. CA-16910, CA-19266, CA-23954, and CA-25568 awarded by the National Cancer Institute, DHHS.

REFERENCES Abe, S., Golomb, H. M., Rowley, J. D., Mitelman, F., and Sandberg, A. A. (1980).Cancer 45,84-90. Alimena, G., Brandt, L., Dallapiccola, B., Mitelman, F., and Nilsson, P. G. (1979). Cancer Genet. Cytogenet. 1, 79-85. Alimena, G., Dallapiccola, B., Gastaldi, R., Mandelli, F., Brandt, L., Mitelman, F., and Nilsson, P. G. (1981).Scnnd. J . Haemutol. (in press). Benedict, W. F., Lange, M., Greene, J., Derencsenyi, A,, and Alfi, 0. S. (1979). Blood 54, 818-823. Bennett, J. M., Catovsky, D., Daniel, M. T., Flandrin, G., Galton, D. A. G., Gralnick, H. R., and Sultan, C. [French-American-British (FAB) Co-operative Group] (1976). Br. J. Haematol. 33, 451-548. Bennett, J. M., Catovsky, D., Daniel, M . T., Flandrin, G., Galton, D. A. G., Gralnick, H. R., and Sultan, C. [French-American-British (FAB) Co-operative Group] (1980). Br. J . Haematol. 44, 169-170. Berger, R. (1975).Nouv. Presse Med. 4, 1972. Berger, R., and Bernheim, A. (1979). Cancer Genet. Cytogenet. 1, 1-8. Berger, R., Bernheim, A,, Daniel, M. T., Valensi, F., and Flandrin, G. (1979a). C.R. Hebd. Seances Acad. Sci. 288, 177- 179. Berger, R.,Bernheim, A,, Flandrin, G., Daniel, M. T., Schaison, G., Brouet, J. C., and Bernard, J. (1979b). N o w . Presse h f e d . 8, 181-183.

144


Berger, R., Bernheim, A., Weh, H. J., Flandrin, G., Daniel, M. T., Brouet, J. C., and Colbert, N. ( 1 9 7 9 ~ )Hum. . Genet. 53, 111-112. Berger, R., Bernheim, A., and Flandrin, G. (1980a). C. R. Hebd. Seances Acud. Sci. 290, 1557-1559. Berger, R., Bernheim, A,, Weh, H. J., Daniel, M. T., and Flandrin, G. (1980b).Leuk. Res. 4, 119-127. Bemstein, R., Mendelow, B., Pinto, M. R., Morcom, G., and Bezwoda, W. (1980a).Br. J . Haematol. 46, 311-314. Bemstein, R., Morcom, G., Pinto, M. R., Mendelow, B., Dukes, I., Penfold, G., and Bezwoda, W. (1980b). Cancer Genet. Cytogenet. 2,23-37. Bloomfield, C . D., Lingquist, L. L., Brunning, R. D., Yunis, J. J., and Coccia, P. F. (1978). Virchow’s Arch. B 29, 81-92. Borgaonkar, D. S., and Lillard, D. R. (1980).“The Repository of Chromosomal Variants and Anomalies in Man: An International Registry of Abnormal Karyotypes,” 7th listing June 1980. Boveri, R. (1914). “Zur Frage der Entstehung maligner Tumoren.” Fischer, Jena. Brandt, L., Mitelman, F., Panani, A., and Lenner, H. C. (1976).Scand. J . Haematol. 16, 32 1-325. Canellos, G. P., Whang-Peng, J., and DeVita, V. T. (1976). Am. J. Clin. Pathol. 65, 467-470. Carbonell, F., Kratt, E., and Neuhaus, K. (1980). Cancer Genet. Cytogenet. 2, 139-143. Caspersson, T., Gahrton, G., Lindsten, J., and Zech, L. (1970).E x p . Cell Res. 63, 238244. Catovsky, D. (1979).Br. J . Haernatol. 42, 493-498. Chessells, J. M., Hardisty, R. M., Rapson, N. T., and Greaves, M. F. (1977). h n c e t 2, 1307- 1309. Chessells, J. M., Janossy, G., Lawler, S. D., and Secker Walker, L. M. (1979). Br. J. Haematol. 41,25-41. Cimino, M. C., Rowley, J. D., Kinnealey, A., Variakojis, D., and Golomb, H. M. (1979). Cancer Res. 39, 227-238. Cunningham, I., Gee, T., Dowling, M., Chaganti, R., Bailey, R., Hopfan, S., Bowden, L., Tumbull, A., Knapper, W., and Clarkson, B. (1979).Blood 53,375-395. Engel, E., McGee, B. J., Flexner, J. M., Russell, M. T., and Myers, B. J. (1974).N. Engl.]. Med. 291, 154. Ferro, M. T., and San Roman, C. (1981).Cancer Genet. Cytogenet. 4, 89-91. Fialkow, P. J . (1974). N. Engl. J . Med. 291, 26-35. First International Workshop on Chromosomes in Leukaemia (1978). Br. J . Haematol. 39,311-316. Forman, E. N., Padre-Mendoza, T., Smith, P. S., Barker, B. E., and Farnes, P. (1977). Blood 49,549-558. Francke, U., Holmes, L. B., Atkins, L., and Riccardi, V. M. (1979). Cytogenet. Cell Genet. 24, 185-192. Gahrton, G., Lindsten, J., and Zech, L. (1974).Acta Med. Scand. 196,355-360. Gahrton, G., Friberg, K., Zech, L., and Lindsten, J. (1978).Lancet 1, 96-97. Geraedts, J. P. M., Mol, A., den Ottolander, G. I., van der Ploeg, M., and Pearson, P. L. (1977). Proc. Helsinki Chromosome Conf. p. 194. German, J. (1972). Prog. Med. Genet. 8,61-101. Geurts van Kessel, A. H. M., ten Brinke, H., Boere, W. A. M., den Boer, W. C., de Groot, P. G., Hagemeijer, A., Meera Khan, P., and Pearson, P. L. (1981).Cytogenet. Cell Genet. 30,83-91.


145

Gibbs, T. J., Wheeler, M. V., Bellingham, A. J., and Walker, S. (1977). Br. J. Haematol. 37,447-453. Golomb, H. M. (1980). Cancer Genet. Cytogenet. 1,249-256. Golomb, H. M., Vardiman, J. W., Rowley, J. D., Testa, J. R., and Mintz, U. (1978). N . Engl. J . Med. 299, 613-619. Golomb, H. M., Testa, J. R., Vardiman, J. W., Butler, A. E., and Rowley, J. D. (1979). Cancer Genet. Cytogenet. 1, 69-78. Golomb, H. M., Rowley, J. D., Vardiman, J. W., Testa, J. R., and Butler, A. (1980).Blood 55,253-259. Green, M. M. (1973). Genetics 73, Suppl., 187-194. Hagemeijer, A., Van Zanen, G. E., Smit, E. M. E., and Hahlen, K. (1979).Pediatr. Res. 13, 1247-1254. Hagemeijer, A., Stenfert Kroeze, W. F., and Abels, J. (1980).Cancer Genet. Cytogenet. 2, 317-326. Hagemeijer, A., Hahlen, K., and Abels, J. (1981). Cancer Genet. Cytogenet 3, 109-124. Henderson, A. S., Eicher, E. M., Yu, M. T., and Atwood, K. C. (1974). Chromosoma 49, 155-160. Hoppin, E. C., and Lewis, J. C. (1975).Ann. Intern. Med. 83, 820-823. Hossfeld, D. K., and Kohler, S. (1979).Br. J. Haematol. 41, 185-191. Hossfeld, D. K., Faltermeier, M. T., and Wendehorst, E. (1979). Blut 38, 377-382. Hossfeld, D. K., Higi, M., Kohler, S., Miller, A,, and Zschaber, R. (1980).Blut 40,27-32. Hsu, L. Y. F., Alter, A. V., and Hirschhorn, K. (1974).Clin. Genet. 6, 258-264. Hsu, L. Y. F., Pinchiaroli, D., Gilbert, H. S., Wittman, R., and Hirschhorn, K. (1977).Am. J . Hematol. 2, 375-383. Hustinx, T. W. J., Burghouts, J. T. M., Scheres, J. M. J. C., and Smits, A. P. T. (1980). Cancer 45, 285-288. Ihle, J. N., Joseph, D. R., and Domofor, J. J., Jr. (1979).Science 204, 71-73. International System for Human Cytogenetic Nomenclature (1978). Cytogenet. Cell Genet. 21,309-404. Ishihara, T., Kohno, S. I., and Kumatori, T. (1974). Br. J. Cancer 29, 340-342. Janossy, G., Greaves, M. F., and Sutherland, R. (1977).Leuk. Res. 1, 289-299. Kaiser-McCaw, B., Epstein, A. L., Kaplan, H. L., and Hecht, F. (1977).2nt.J. Cancer 19, 482-486. Kamada, N., Okada, K., Ito, T., Nakatsui, T., and Uchino, H. (1968).Lancet 1,364. Kamada, N., Okada, K., Oguma, N., Tanaka, R., Mikami, M., and Uchino, H. (1976). Cancer 37,2380-2387. Kaneko, Y., and Rowley, J. D. (1981).Proc. Am. SOC. Clin. Oncol. 22, 339. Kaneko, Y., and Rowley, J. D. (1982).Pediatr. Oncol. 2 (in press). Kaneko, Y., and Sakurai, M. (1980).Cancer Genet. Cytogenet. 2, 13-18. Kaneko, Y., Sakurai, M., and Hattori, M. (1978).Am. J. Hematol. 4, 273-280. Kaneko, Y., Rowley, J. D., Check, I., Variakojis, D., and Moohr, J. W. (1980).Blood 56, 782-785. Kaneko, Y., Hayashi, Y., and Sakurai, M. (1981).Cancer Genet. Cytogenet. 4. Kay, H. E. M., Lawler, S. D., and Millard, R. E . (1966). Br. J. Haematol. 12, 507-528. Kessous, A,, Corberand, J., Grozdea, J., and Colombies, P. (1975). Nouu. Reu. Fr. Hematol. 15, 73-82. Knight, L. A,, Davidson, W. M., and Cuddigan, B. J. (1974). Lancet 1,688. Kohno, S . I., and Sandberg, A. A. (1980). Cancer 46,2227-2237. Kondo, K., and Sasaki, M. (1979). Cancer Genet. Cytogenet. 1, 131-138. Lawler, S . D., Millard, R. E., and Kay, H. E. M. (1970). Eur. J. Cancer 6,223-233.

146


Lawler, S. D., Lopp, D. S., and Wiltshaw, E. (1974). Br. J. Haematol. 27,247-252. Lawler, S. D., O’Malley, F., and Lobb, D. S. (1976). Scand. J. Haematol. 17, 17-28. Lawler, S. D., Summersgill, B., Clink, H. M., and McElwain, T. J . (1980). Br. J. Haematol. 44,395-405. Lilly, F., and Pincus, T. (1973).Adu. Cancer Res. 17, 231-277. Lindgren, V., and Rowley, J. D. (1977). Nature (London) 266,744-745. McClintock, B. (1961). Am. Nat. 95, 265-277. McKusick, V. A., and Ruddle, F. H. (1977). Science 196,390-405. Martin, P. J., Najfeld, V., Hansen, J. A., Penfold, G. K., Jacobson, R. J., and Fialkow, P. J. (1980).Nature (London) 2 8 7 , 4 9 4 0 . Mayall, B. H., Carrano, A. V., Moore, D. H., 11, and Rowley, J. D. (1977).Cancer Res. 37, 3590- 3593. Millard, R. E., Lawler, S. D., Kay, H. E. M., and Cameron, C. B. (1968).Br.J.Haematol. 14,363-374. Miller, D. R., Leikin, S., Albo, V., Vitale, L., Sather, H., Coccia, P., Nesbit, M., Karon, M., and Hammond, D. (1980). Cancer Treat. Rep. 64,381-392. Mitelman, F. (1974). Hereditas 76, 315-316. Mitelman, F., and Levan, G. (1978).Hereditas 89, 207-232. Mitelman, F., Brandt, L., and Nilsson, P. G. (1978). Blood 52, 1229-1237. Mitelman, F., Anvret-Andersson, M., Brandt, L., Catovsky, D., Klein, G., Manolov, G., Manolova, Y.,Mark-Vendel, E., and Nilsson, P. G. (1979).Znt. J. Cancer 24, 27-33. Miyoshi, I., Hiraki, S., Kimura, I., Miyamoto, K., and Sato, J. (1979).Experientia 35,742. Morse, H., Hays, T., Peakman, D., Rose, B., and Robinson, A. (1979). Cancer 44, 164170. Nevers, P., and Saedler, H. (1977).Nature (London) 268,109-115. Nilsson, P. G., Brandt, L., and Mitelman, F. (1977).Leuk. Res. 1, 31-34. Nowell, P., and Finan, J. (1978). Cancer 42, 2254-2261. Nowell, P. C., and Hungerford, D. A. (1960). Science 132, 1197. Nowell, P. C., Jensen, J., Gardner, F., Murphy, S., Chaganti, R. S . K., and German, J. (1976). Cancer 38, 1873-1881. Olah, E., Kiss, A., and Jak6, J. (1980). Znt. J. Cancer 26, 37-45. O’Riordan, M. L., Robinson, J. A., Buckton, K. E., and Evans, H. J. (1971). Nature (London)230,167-168. Oshimura, M., Freeman, A. I., and Sandberg, A. A. (1977a). Cancer 40, 1143-1148. Oshimura, M., Freeman, A. I., and Sandberg, A. A. (1977b). Cancer 40, 1161-1172. Padre-Mendoza, T., Farnes, P., Barker, B. E., Smith, P. S., and Forman, E. N. (1979. Br. J. Haematol. 41, 43-48. Paris Conference (1972). Birth Defects, Orig. Artic. Ser. 8, NO. 7. Petit, P., and Van Den Berghe, H. (1979).Ann. Genet. 22, 103-105. Pierre, R. V., and Hoagland, H. C. (1972). Cancer 30,889-894. Prieto, F., Badia, L., Mayans, J., Gomis, F., and Marty, M. L. (1978).Sangre 23,484-488. Prigogina, E. L., Fleischman, E. W., Volkova, M. A., and Frenkel, M. A. (1978). Hum. Genet. 41, 143-156. Prigogina, E. L., Fleischman, E. W., Puchkova, G. P., Kulagina, 0. E., Majakova, S. A., Balakirev, S. A., Frenkel, M. A., Khvatova, N. V., and Peterson, I. S. (1979). Hum. Genet. 53, 5-16. Reeves, B. R., Lobb, D. S., and Lawler, S. D. (1972). Humangenetik 14, 159-161. Rowe, W. P. (1973). Cancer Res. 33,3061-3068. Rowe, W. P., and Kozak, C. A. (1979).Science 204, 69-71. Rowley, J. D. (1973a).Nature (London) 243,290-293. Rowley, J. D. (1973b).Ann. Genet. 16, 109-112.


147

Rowley, J. D. (1977a).I n “Population Cytogenetics” (E. B. Hook and I. H. Porter, eds.), pp. 189-216. Academic Press, New York. Rowley, J. D. (1977b). Proc. Natl. Acad. Sci. U.S.A. 74, 5729-5733. Rowley, J. D. (1980a).Br. J . Haematol. 44, 339-346. Rowley, J . D. (1980b). Clin. Haematol. 9, 55-86. Rowley, J. D. ( 1 9 8 0 ~ )Cancer . Genet. Cytogenet. 2, 175-198. Rowley, J. D., and Fukuhara, S. (1980). Seniin. Oncol. 7, 255-266. Rowley, J . D., Golomb, H. M., and Dougherty, C. (1977). Lancet 1, 549-550. Rowley, J. D., Golomh, H. M., and Vardiman, J. W. (1981a).Blood 58, 759-767. Rowley, J . D., Variakojis, D., Kaneko, Y., and Cimino, M. (1981b). Hum. Genet. 58, 166- 167. Sakurai, M., and Sandberg, A. A. (1973).Blood 41, 93-104. Sakurai, M., and Sandberg, A. A. (1976a). Cancer 37,285-299. Sakurai, M., and Sandberg, A. A. (1976b). Cancer 38, 762-769. Sakurai, M., Hayata, I., and Sandberg, A. A. (1976).Cancer Res. 36, 313-318. Sandberg, A. A. (1980a). “The Chromosomes in Human Cancer and Leukemia.” ElseviedNorth-Holland, New York. Sandberg, A. A. (1980b). Cancer 46, 2221-2226. Sandberg, A. A,, and Sakurai, M. (1973).Lancet 1, 375. Secker Walker, L. M., Lawler, S. D., and Hardisty, R. M. (1978). Br. Med. J. 2, 15291530. Second International Workshop on Chromosomes in Leukemia (1980). Cancer Genet. Cytogenet. 2, 89-113. Shabtai, F., Weiss, S., van der Lijn, E., Lewinski, U., Djaldetti, M., and Halbrecht, I. (1978).Hum. Genet. 41, 281-287. Shabtai, F., Lewinski, U . H., Har-Zahav, L., Gaffer, U., Halbrecht, I., and Djaldetti, M. (1979).A m . /. Clin. Pathol. 72, 1018-1024. Shiraishi, Y., Hayata, I., Sakurai, M., and Sandberg, A. A. (1975).Cancer 36, 199-202. Slater, R. M., Philip, P., Badsberg, E., Behrendt, H., Hansen, N. E., and van Heerde, P. (1979).Znt. J . Cancer 23, 639-647. Sonta, S., and Sandberg, A. A. (1978). Cancer 41, 153-163. Sonta, S., Oshimura, M., and Sandberg, A. A. (1976).Blood 48,697-705. Stavem, P., van der Hagen, C. B., Vogt, E., andSandness, K.(1975). Clin. Genet. 7, 227-23 1. Stoll, C., and Oberling, F. (1979).Leuk. Res. 3, 61-66. Sugiyama, T. (1971).JNCZ, /. N o t l . Cutlcer Z I I , Y ~ .47, 1267-1276. Sugiyama, T., Uenaka, H., Ueda, N., Fukuhara, S., and Maeda, S. (1978).JNCI,J . N u t l . Cancer I n s t . 60, 153-160. Tanzer, J., Najean, Y., Frocrain, C., and Bernheim, A. (1977).N. EngZ.1. Med. 296,571. Teerenhovi, L., Borgstrom, G. H., Mitelman, F., Brandt, L., Vuopio, P., Timonen, T., Almqvist, A,, and de la Chapelle, A. (1978).Lancet 2, 797. Testa, J . R. (1980). Cancer Genet. Cytogenet. 1,207-215. Testa, J . R., and Rowley, J. D. (1980).Cancer Genet. Cytogenet. 1, 239-247. Testa, J . R., and Rowley, J . D. (1981).111 “The Leukemic Cell” (D. Catovsky, ed.), pp. 184-202. Churchill-Livingstone, Edinburgh and London. Testa, J. R., Golomb H. M., Rowley, J. D., Vardiman, J. W., and Sweet, D. L. (1978a). Blood 52,272-280. Testa, J. R., Kinnealey, A., Rowley, J. D., Golde, D. W., and Potter, D. (1978b).Blood 52, 868-877. Testa, J. R., Mintz, U., Rowley, J . D., Vardiman, J. W., and Golomb, H. M. (1979).Cancer Res. 39,3619-3627.

148

JANET D . ROWLEY AND JOSEPH R. TESTA

Testa, J . R., Kanofsky, J. R., Rowley, J. D., Baron, J. M., and Vardiman, J. W. (1981).Am. J. Hematol. 11, 29-45. Third International Workshop on Chromosomes in Leukemia (1981). Cancer Genet. Cytogenet. 4,95-142. Trujillo, J. M., Cork, A., Ahearn, M. J., Youness, E. L., and McCredie, K. B. (1979).Blood 53,695-706. Tsuchimoto, T., Buhler, E. M., Stalder, C. R., Mayr, A. C., and Ohrecht, J. P. (1974). Lancet 1, 566. Van Den Berghe, H. (1973). Lancet 2, 1030. Van Den Berghe, H., David, C., Michaux, J. L., Sokal, G., and Verwilghen, R. (1976). Blood 48,624-625. Van Den Berghe, H., David, G., Broeckaert-Van Orshoven, A., Louwagie, A., and Verwilghen, R. (1978).N. Engl. J. Med. 299, 360. Van Den Berghe, H., Broeckaert-Van Orshoven, A., Louwagie, A., Venvilghen, R., Michaux, J. L., and Sokal, G. (1979a).Cancer Genet. Cytogenet. 1, 157-167. Van Den Berghe, H., Louwagie, A., Broeckaert-Van Orshoven, A., David, G., Verwilghen, R., Michaux, J. L., and Sokal, G. (1979h). Cancer 43, 558-562. Van Den Berghe, H., David, C., Broeckaert-Van Orshoven, A., Louwagie, A., Verwilghen, R., Casteels-Van Daele, M., Eggermont, E., and Eeckels, R. (197%). Hum. Genet. 46, 173-180. Van Den Berghe, H., Parloir, C., Goseye, S., Englehienne, V., Comu, G., and Sokal, G. (1979d). Cancer Genet. Cytogenet. 1,9-14. Verma, R . S., and Dosik, H. (1980). B r . J . Haematol. 45, 215-222. Visfeldt, J. (1971).Actu Pathol. Microbiol. Scand., Sect. A 79,513-523. Visfeldt, J., Franz&n,S., Nielsen, A,, and Trihukait, B. (1973).Actu Pathol. Microbiol. Scand., Sect. A 81, 195-203. Vogler, L. B., Crist, W. M., Vinson, A. S., Brattain, M. G., and Coleman, M. S. (1979). B2ood 54,1164-1170. Weinfeld, A., Westin, J., Ridell, B., and Swolin, B. (1977). Scand. J. Haematol. 19, 255-272. Westin, J. (1976a). Thesis, from the Section of Oncological Haematology, Medical Department 11, Sahlgren’s Hospital, Gotehorg, Sweden. Westin, J. (1976b).Scand. J. Haematol. 17, 197-204. Westin, J., and Weinfeld, A. (1978). Znt. Congr. Hematol., 17th, 1978 Abstract, p. 971. Westin, J., Wahlstrom, J., and Swolin, B. (1976). Scand. J . Haematol. 17, 183-196. Whang-Peng, J., Canellos, G. P., Carbone, P. P., and Tjio, J. H. (1968). Blood 32, 755766. Whang-Peng, J., Knutsen, T., Ziegler, J., and Leventhal, B. (1976).Med. Pediatr. Oncol. 2,333-351. Whang-Peng, J., Henderson, E. S., Knutsen, T., Freireich, E. J., and Cart, J. J. (1970). Blood 36,448-457. Wurster-Hill, D., Whang-Peng, J., McIntyre, 0.R., Hsu, L. Y.F., Hirschhorn, K., Modan, B., Pisciotta, A. V., Pierre, R., Balcerzak, S. P., Weinfeld, A., and Murphy, S. (1976). Semin. Hernatol. 13, 13-32. Yunis, J . J., and Rainsay, N. (1978).Am. J . Dis. Child. 132, 161-163. Yunis, J . J., Bloomfield, C. D., and Ensrud, K. (1981).N. Engl. J. Med. 305, 135-139. Zech, L., Cahrton, C., Killander, D., Franzbn, S., and Haglund, U. (1976a). Blood 48, 687-696. Zech, L., Hoglund, U., Nilsson, K., and Klein, G. (1976h). Znt. J . Cancer 17,47-56.

ONCOGENES OF SPONTANEOUS AND CHEMICALLY INDUCED TUMORS Robert A. Weinberg Massachusetts institute of Technology. Center for Cancer Research and Department of Biology, Cambridge. Massachusetts

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

149

11. A Model of Cellular Oncogenes

111. The Retrovirus-Associated Oncogenes .................................. IV. Oncogenes Present in Cells Transformed by Chemical Carcinogens V. Multiplicity of Transforming Genes in 3-MethylcholanthreneTransformed Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Types of Transformed Cells Yielding Focus-Induced D N A . . ............. VII. Multiplicity of Different Human Oncogenes VIII. Analogies between Virus- and Non-Virus-Ind IX. The Process of Activation of Oncogenes . . . . X. The Role of Oncogenes in Carcinogenesis and Maintenance of Phenotype. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI. The Proteins Encoded by Activated Oncogenes ......................... References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

150

155 156

160 161 162

I. Introduction

The molecular basis of nonviral tumorigenesis is poorly understood, in large part because pertinent experimental approaches have been lacking over the past several decades. Indirect experiments have suggested that DNA damage is a central step in transformation, largely because DNA-damaging agents tend to be carcinogens (McCann and Ames, 1976). However, this realization of the importance of DNA damage in no way facilitates an understanding of which DNA sequences must be damaged in order to induce transformation. Moreover, one still is ignorant of the importance of any such sequence changes compared with a number of nongenetic alterations whose role in inducing transformation may greatly overshadow that of the altered DNA. In the present review, we shall discuss how the process of gene transfer, also termed transfection, makes possible the resolution of some of these issues. The introduction of this experimental approach into the investigation of the molecular basis of oncogenic transformation has already yielded important insights into several central issues in this field, 149 ADVANCES IN CANCER RESEARCH. VOL. 36

Copyright 0 1982 by Academic Press, Inc. All rights of reproduction in any form reserved. ISBN 0-12-006636-X

150

ROBERT A. WEINBERG

II. A Model of Cellular Oncogenes

The present discussion is based on a model that invokes a cellular oncogene as being centrally important in the process of oncogenic transformation (Comings, 1973; Todaro and Huebner, 1972). According to this model, such an oncogene derives from a normally innocuous cellular DNA sequence whose inappropriate activation during oncogenesis confers a novel transforming potency on the altered gene. This paradigm of oncogenesis contains two elements important for the present discussion. First, the oncogenic sequence arises as an alteration of preexisting cellular information. As such, the resulting oncogenic sequence is of endogenous cellular origin and is therefore not a foreign sequence imposed on the cellular genome from outside the cell. Second, the DNA sequences encoding transformation comprise a discrete unit of function and may represent an allelic variant of a normal cellular gene. This in turn implies that the activation of critical cellular genes, not the global activation of large portions of the genome, is important for oncogenic transformation. Ill. The Retrovirus-Associated Oncogenes

A partial vindication of this model has come from tumor virology. Among four classes of oncogenic viruses that have been intensively studied, it would appear that three-adenoviruses, papovaviruses, and herpesviruses-have evolved specific viral genetic sequences that they use to transform virus-infected cells. In contrast, many members of the fourth class, the retroviruses, induce transformation by expropriating cellular genetic sequences whose presence in the retrovirus genomes confers a transforming potential on the viral genome. Having incorporated a cellular gene into the viral genome, the virus liberates the cellular gene from control mechanisms that governed its expression in the normal cell chromosome and places the cellular gene under viral regulation. Moreover, having become allied with the viral genome, the cellular gene acquires the genetic mobility normally associated with infectious virus. This intimate relationship between retrovirus-associated oncogenes and normal cellular genes has been most thoroughly documented in the case of avian (Rous) sarcoma virus (ASV). The ASV viral gene inducing transformation has been termed the src and encodes a protein of 60,000 daltons (Brugge and Erikson, 1977). Mutations in this src gene affect the transforming competence of the ASV genome. It is clear that the DNAs of normal, uninfected cells carry sequences that are

ONCOGENES OF NONVIRAL TUMORS

151

strongly homologous, if not identical to the viral src gene (Stehelin et al., 1976). Perhaps most relevant is the fact that a homolog of the src protein is present in normal, uninfected cells (Collett et al., 1979; Oppermann et al., 1979; Sefton et al., 1980). The model of retroviruses acting as transducing agents of cellular genes has been extended to as many as 10 other acutely oncogenic retroviruses (Coffin et d.,1981; Klein, 1980), each of whose genomes appears to carry a different cellular sequence. These retrovirus-associated cellular sequences are termed generically onc genes. With most of these viruses, the presence of a normal cellular gene within the viral genome is indicated only by nucleic acid sequence homology studies (Frankel and Fischinger, 1977; Stehelin et al., 1976). However, more direct demonstrations of the functional and structural affinities of the viral oncogenes and their cellular counterparts have been forthcoming. For example, mutants of ASV whose onc genes have been largely lost through genetic deletion can be shown to reacquire both the missing sequences and oncogenic potency by passage through normal, hitherto uninfected cells (Karess et al., 1979).The restoration of function depends on genetic recombination between the deleted ASV genome and a bloc of normal cellular sequences (Karess and Hanafusa, 1981). This shows that the missing src sequences preexisted in the cell prior to infection by the ASV deletion mutant. A totally different experiment using the onc gene of Moloney murine sarcoma virus (MSV) leads to supporting conclusions for this model. Molecular cloning has allowed the isolation of the normal cellular sequences whose homolog is found as the onc sequence of the MSV genome. This cellular sequence would appear to be very similar if not identical to its viral counterpart, yet the cloned DNA of the cellular sequence is biologically inactive when applied to cells in culture using transfection. However, when the cellular gene is juxtaposed to a viral transcriptional promoter, then DNA carrying the fused sequence exhibits strong transforming competence (Blair et al., 1981). This experimental manipulation recapitulates the outlines of a genetic recombination event between viral and cellular sequences which led originally to creation of the hybrid MSV genome. It indicates that at least one cellular onc gene, that related to M-MSV, possesses an intrinsic transforming competence whose expression awaits appropriate activation, in this case, an activation achieved by an alliance with a viral transcriptional regulator. A third example of affiliation of retrovirus and cellular oncogene derives from work on avian leukosis virus (ALV)-induced tumors (Hayward et al., 1981). This virus appears to carry no transforming

152

ROBERT A. WEINBERG

genes in the viral genome, and its ability to induce a variety of tumors, especially leukoses, has therefore been puzzling. Analysis of the DNAs of a series of avian lymphomas has revealed that the ALV genome, or a fragment thereof, is frequently seen to be integrated next to a specific cellular sequence. This cellular sequence had, fortuitously, been studied previously, since it is identical to the sequence expropriated from the cellular genome during the recombination events that led to the formation of avian myelocytomatosis virus. It appears that the ALV genome, by integrating next to this cellular oncogene, is able to activate the gene transcriptionally. In this case, the alliance between retrovirus and cellular oncogene exists only within the chromosomal DNA of a tumor cell, not within a transmissible genome of an infectious virus particle. It remains to be proved that this ALV-induced gene-oncogene activation is the sole cause of transformation of the avian lymphocyte precursors. Others, using the gene transfer procedure, have detected a transforming gene in these tumors that is unlinked to the ALV genome and would appear to represent a second, independently acting, transforming sequence (Cooper and Neiman,

1980). In principle, activation of cellular oncogenes b y retroviruses might be achieved by two different strategies. The cellular oncogene could suffer alterations in structural sequences encoding a transforming protein, which might in turn confer novel properties on the normal cellular protein. Alternatively, the new gene may differ from the old in its degree of expression. Thus, the novel oncogene may depend on enhanced dosage of a normal cellular protein to elicit a phenotype which low level expression could not achieve. The retrovirus models examined to date favor this latter alternative of increased expression of a structurally normal sequence. For example, cells transformed by ASV or Harvey sarcoma virus (Coffin et al., 1981; Hughes et al., 1979; Oppermann et al., 1979; Sefton et al., 1980) exhibit 50- to 100fold higher levels of the respective onc proteins than one found in normal, uninfected cells. It would appear that the virtually identical, cellular onc proteins, present in low level in normal cells, have no oncogenic effects on cellular phenotype. It is important to note that the several cellular onc genes that have been characterized share properties with many other types of cellular genes studied over the past several years. These onc genes are present in low or single copy number in the genome and have no apparent affiliation with any viral sequence in the normal, uninfected cell (Hughes et ul., 1979). They are conserved evolutionarily like cellular genetic sequences, and they would appear to be as central to normal


153

cellular function as would any other gene encoding required cellular enzymic functions. IV. Oncogenes Present in Cells Transformed by Chemical Carcinogens

By providing proof of the existence of cellular onc genes, the retrovirus model has given great impetus to the search for transforming genes present in cells transformed by nonviral agents. Rather than invoking exogenous activation of cellular oncogenes via viral regulation, one may presume that an equivalent activation can be achieved by rearrangement of blocs of normal cellular sequences after carcinogen-induced DNA damage. This as yet unproved model is embedded in much of the discussion that follows. Such a model need not imply that precisely the same genes that are expropriated and activated by retroviruses will also be activated after nonviral insults to cellular DNA. Thus, nucleic acid sequence onc probes, developed from chimeric retrovirus genomes, may not be useful reagents to identify chemically activated cellular oncogenes. Rather, an alternative experimental approach is required to illuminate these genes. Gene transfer represents a direct strategy to detect transforming genes in these transformed cells, and the use of the procedure in no way depends upon reagents developed from retrovirus genomes. This same gene transfer procedure was used with great benefit in resolving a variety of problems surrounding the transforming genes of tumor viruses (Graham, 1977). The gene transfer or “transfection” procedure depends technically on the coprecipitation of DNA with calcium phosphate crystals. These crystals settle onto monolayers of cultured cells and are taken up by the cells via a poorly understood mechanism. A small but significant proportion of the applied DNA succeeds in entering a cell intact and in being expressed stably in a recipient cell and its descendants (Graham and van der Eb, 1973). One exploitation of this transfection procedure has occurred in the area of tumor virology. The DNAs of a variety of tumor viruses are able to induce foci of transformed cells on fibroblast monolayers (Andersson, 1980;Graham, 1977;Graham and van der Eb, 1973).These monolayers usually display these transformed foci several weeks after exposure to the donor DNA. These foci arise as a direct consequence of the uptake and fixation of the transforming genes carried by the transfected viral genomes. Introduction of DNAs of 3-methylcholanthrene-transformedmouse fibroblasts into untransformed, recipient fibroblasts also results in

154

ROBERT A. WEINBERG

transformation of the recipient cells (Shih et al., 1979). Use of DNA from untransformed donor cells does not induce foci of transformation in a recipient cell monolayer. This result provides direct proof that the DNA of a 3-methylcholanthrene-transformed cell is structurally altered with respect to the DNA of an untransformed control cell. This alteration relates directly to an important phenotype, since it is likely that the oncogenic information passed by DNA transfer is partly responsible for the transformation phenotype in the donor cell from which DNA has been prepared. The results of this experiment provide direct proof that DNA alteration occurs in chemically transformed cells in regions of the cellular genome that can elicit a transformation phenotype. As described below, many analogous experiments have been performed, using different types of tumor cells, that further support this conclusion. However, all such experiments give no insight into the mechanisms whereby a mutagen, interacting with cellular DNA, is able to induce activation of cellular oncogenic sequences. The biologically active sequences in the transforming DNAs behave very much like discrete segments of nucleotides. These segments may eventually be defined as genes by conventional genetic criteria. These transforming sequences are transferable from cell to cell by serial passage of DNA. Such serial passaging from donor to recipient uses DNA of the transformed recipient cell as donor DNA in a subsequent cycle of transfection. These repeated manipulations d o not result in the diminution of the transforming activity of the DNAs. Therefore, the transforming sequences do not behave like a group of unlinked, cooperating genes whose association is readily disrupted and diluted b y gene transfer (Shih et al., 1979, 1981). Rather, a transforming sequence behaves like a discrete, compact bloc of information, and is thus reminiscent of a gene. Moreover, the biological activity of a given DNA preparation may be destroyed by certain sequence-specific endonucleases (restriction enzymes) and may be left unaffected by other endonucleases (Krontiris and Cooper, 1981; Shilo and Weinberg, 1981). This means that within a given chemically transformed cell, a discrete, definable segment of DNA carries the transforming potential that is observed upon cell-to-cell transfer of DNA. One can rule out an activation of large numbers of scattered gene blocs whose concerted actions are required for specifying the transformation trait. The transforming activities of these DNAs are not associated with readily detectable retrovirus genomes. Repeated attempts at demonstrating transmissible type C retroviruses in association with the donor cells and derived recipients have yielded no trace of titrable virus


155

(Krontiris and Cooper, 1981; Shih et al., 1979, 1981).This has led to the tentative conclusion that the observed transforming genes represent activated versions of normal cellular sequences and are not associated with retrovirus genomes. Such a conclusion will be totally credible only when transforming sequences have been isolated and analyzed in structural detail. V. Multiplicity of Transforming Genes in 3-MethylcholanthreneTransformed Cells

These experimental manipulations allow one to provide preliminary answers concerning the multiplicity of transforming genes that are targets of activation upon treatment with 3-methylcholanthrene. Since the mammalian genome may carry several tens of thousands of genes, it becomes possible that any one of several hundred of these genes will serve as a suitable precursor of the activated transforming genes seen in the 3-methylcholanthrene-transformed cells. Experiments have been performed with the DNAs of four independently transformed mouse fibroblast lines, each of whose transformation derived from exposure to 3-methylcholanthrene. These experiments were designed to measure whether the same gene, or four different transforming genes, was activated in the four transformed lines (Shilo and Weinberg, 1981). As probes for the structures of the different genes, the transforming DNAs were treated with site-specific endonucleases (restriction enzymes), which recognize specific hexanucleotide sequences at the site of cleavage of DNA. Each of the DNAs was treated with one of a series of restriction enzymes and then tested by transfection for retention or loss of biological activity. For example, a transforming gene carried entirely within an EcoRI endonuclease-generated DNA fragment should have its activity unaffected by treatment with this enzyme. Conversely, a transforming gene whose sequence contains the hexanucleotide recognition cleavage site of this enzyme will be split and inactivated after EcoRI endonuclease treatment. Transforming genes associated with different nucleotide sequences should exhibit differing patterns of resistance and inactivation to restriction enzymes, depending upon the nucleotide sequences carried by the gene. This type of experiment yields a specific signature of each gene tested, since the pattern of restriction enzyme sites within a gene is a reflection of the structure of that gene alone and is not shared by other genes of the chromosomal DNA. Because the presence of such a cleavage site depends solely on statistical happenstance, each gene will have a different array of restriction enzyme sites in its DNA.

156

ROBERT A. WEINBERG

The results indicate that all four 3-methylcholanthrene-induced transforming sequences behave identically to one another by these tests. Thus, all four genes are resistant to BamHI endonuclease cleavage, but are inactivated by EcoRI endonuclease. Use of three other endonucleases extends this pattern of identical behavior among the four DNAs. In the case of these four independently transformed mouse fibroblast lines, one concludes that the same cellular sequence was repeatedly activated to yield a transforming gene (Shilo and Weinberg, 1981). The result is perhaps unexpected when considering the multitude of cellular genes, the activation of any one of which might lead to creation of a potent oncogene. This result may pertain only to the small number of 3-methylcholanthrene-transformed mouse fibroblasts tested in these experiments. The same target cells, transformed by other oncogenic agents, might well carry an alternative set of transforming sequences. Perhaps more importantly, nonfibroblastic transformants, such as carcinoma cells, may carry totally unrelated active oncogenes in their DNA (see below). VI. Types of Transformed Cells Yielding Focus-Induced DNA

The utility of these types of experiments is not limited to the DNAs of 3-methylcholanthrene-transformed mouse cells. Rather, a variety of different types of transformed cells yields DNAs that are active when introduced into mouse fibroblasts. This group includes cell lines of rat neuroblastoinas and gliomas (Shih et al., 1981), human neuroblastomas (A. Cassill and R. A. Weinberg, unpublished observations), mouse, rabbit, and human bladder carcinomas (Krontiris and Cooper, 1981; Shih et al., 1981), mouse and human lung carcinomas (C. Shih and R. A. Weinberg, unpublished observations; Shih e t al., 1981), human leukemia (Murray et al., 1981), and colon carcinoma (Murray et al., 1981). It is apparent that the NIH3T3 mouse fibroblasts used as recipients in these gene transfers serve as sensitive indicators of transforming genes from a variety of sources. Since human carcinoma and leukemia DNAs are active in these transfection-focus assays, it is clear that these genes are able to act across tissue and species barriers. Also, it seems that tumors originating via a number of carcinogenic stimuli, including “spontaneous” tumors of unknown etiology, carry transforming genes detectable in these assays. Of more than passing interest is the fact that such experiments can be performed to study transforming genes present in a variety of frequently occurring human neoplasms.


157

The successful transfer of these various tumor-transforming genes will shortly make possible the isolation and detailed characterization of these genes. These advances should not obscure the fact, however, that the DNAs of most types of human and nonhuman tumors are inactive upon transfection (Krontiris and Cooper, 1981; Shih et al., 1981). It is possible that these nontransferable genes represent a distinct class of sequences whose nature will differ markedly from those discussed here. Alternatively, their unsuccessful transfer may only reflect a technical shortcoming in the transfection-focus assay system.

VII. Multiplicity of Different Human Oncogenes

The experiments on 3-methylcholanthrene-transformedmouse cells suggested that the same oncogene was activated during four different fibroblast transforming events. The possibility remained that other, as yet unstudied, fibroblasts would carry other unrelated, active oncogenes. Additionally, transformation of different tissue types might also depend on different distinct oncogenes being activated. In fact, work on transforming genes of a human colon carcinoma, a bladder carcinoma, and a myelogenous leukemia cell line indicates three different structures for these three genes (Murray et al., 1981). These analyses depend upon transferring the respective human genes through two cycles of transfection in mouse cells, until the transforming gene represents virtually the only human DNA in the final transformed recipient. The structural outline of each gene could then be analyzed, since all human genes are embedded in complex arrays of highly repeated DNA sequences (Houck et al., 1979). Since the structure of these repeat sequences is species-specific, these sequences can be readily detected and analyzed in mouse cells using nucleic acid sequence probes specific for highly repeated human DNA. This type of analysis, which can be readily extended to other human transforming genes, indicates that each of the human genes present in the mouse cells is embedded in a distinct matrix of repeated sequence blocs in the human genome. This in turn provides strong support for the conclusion that one is studying three unique, different genes. Such work establishes the principle that a number of different human oncogenes exist. Following the result with the 3-methylcholanthrene-induced fibroblast lines, one might speculate that each type of tumor carries a characteristic tissue-specific oncogene, and that all tumors of a given type bear the same activated

158

ROBERT A . WEINBERG

oncogene. This hypothesis remains to be vindicated by further work analyzing the DNAs of an extensive series of different types of tumors. Of related interest is the behavior of mouse fibroblasts that have acquired these various transforming genes. The appearance of these transfected cells in culture is quite similar and is independent of the gene that they have acquired. Also, these transfected cells all grow into fibrosarcomas when seeded into young mice. For example, the transforming gene of a carcinoma converts an untransformed fibroblast into the precursor of a fibrosarcoma. Thus, the transfer of these genes results only in the passage of a transformation phenotype, not in the additional cotransfer of other, tissue-specific characteristics. A series of these genes from different types of tumors would appear to act very similarly. Although some of these oncogenes are associated with different sequences, and likely encode different gene products, the consequences of the presence of these active genes are similar if not identical. Such a pattern of a common, convergent phenotype induced by many genetically distinct oncogenes has been observed when studying the onc sequences of retroviruses (Klein, 1980). VIII. Analogies between Virus- and Non-Virus-Induced Cellular Oncogenes

The various data on nonvirally activated oncogenes suggest a series of analogies or parallel properties between the two groups of transforming sequences. The first, and most salient, feature concerns the structural features of the sequences of both groups. In both cases, the oncogenic sequences behave as though they were discrete blocs of sequences and they segregate upon gene transfer as though they were definable elements that could be termed genes. The virus- and nonvirus-induced oncogenes each constitute groups of genes. The size of each of these groups is not yet discernible. With the virus-associated oncogenes, the gene group contains at least eight members (Coffin et al., 1981). A further, and as yet speculative, analogy stems from the behavior of these two groups of genes. Both groups of oncogenes would seem to derive from the normally innocuous cellular sequences with no viral affiities, which have become subverted by virus or mutagenmediated genetic rearrangements. This conclusion is increasingly well substantiated for the retrovirus onc genes and still rests on only indirect data related to the nonvirally activated genes. Finally, one can mention a striking parallel between ALV-induced chicken leukosis and 3-methylcholanthrene-inducedmouse fibroblast transformation. In a series of avian leukosis tumors, the virus is seen


159

repeatedly to be integrated next to the same cellular oncogene, even though the ALV genome is capable of integrating at a virtually infinitely large number of sites in the cellular genome (Hayward et al., 1981). In the case of the 3-methylcholanthrene-transformedfibroblasts, analysis of four different transformants appears to reveal repeated activation of the same cellular gene, even though the responsible carcinogen is likely capable of affecting millions of different sites on the cell genome (Shilo and Weinberg, 1981). In both cases, one suggests that the oncogenic pathway within a given tissue or cell type repeatedly leads to the activation of a specific gene. The parallel properties of the virus-associated and non-virusassociated oncogenes may indicate that the two groups overlap or are congruent with one another. Thus, each retrovirus-associated onc gene may eventually be shown to have a counterpart in a certain type of spontaneous tumor that carries the same activated sequence. Alternatively, the two sets of oncogenes may represent mutually exclusive, nonoverlapping groups. In any case, it is already apparent that the virus-associated onc genes represent an extremely useful experimental model for the as yet unexplored genes of non-virus-induced tumors. IX. The Process of Activation of Oncogenes

The repeated activation of the same mouse fibroblast cellular oncogene after 3-methylcholanthrene carcinogenesis raises an important question. It is clear that the carcinogen can interact with a multitude of sites on the cellular genome. The work on the human tumor genes indicates that several, and likely many, latent cellular oncogenes exist in the genome awaiting appropriate activation. Why then does a nontargeted agent (a carcinogen), when interacting with a multitarget genome, repeatedly activate the same transforming sequences? It is apparent that the process of oncogene activation cannot follow classic models of mutagenesis. Rather, tissue- or cell-specific factors must predispose certain genes to be activated and protect other genes from activation. One such predisposing factor might be the state of expression of a gene in the normal cell prior to the carcinogenic insult. Other paradoxes have been reported that also make it unlikely that classic models of mutagenesis will explain the activation of these genes (Kennedy et d., 1980; Reznikoff et al., 1973).The most recent of these paradoxes derives from in vitro transformation of C3H 1OT1/2 fibroblasts by X-rays. These mouse fibroblasts are of the same cell line that yielded the 3-methylcholanthrene-transformedlines discussed above. The work on X-ray carcinogenesis makes it clear that there is no appar-

160

ROBERT A. WEINBERG

ent proportionality between the number of C3H10T1/2 cells exposed initially to the carcinogen and the number of foci of transformants that appear in descendant cultures. This and earlier work by others (Armitage and Doll, 1957; Nordling, 1953;Whitternore, 1978)leave one with unorthodox hypotheses on the mechanisms of carcinogenesis, all of which are incompatible with a simple target theory of gene activation. One such hypothesis would argue that it is unnecessary for the carcinogen directly to alter the target oncogene in a treated cell in order for that oncogene to become activated in a descendant of that cell. Rather, the carcinogen may induce a metabolic state in the cell, whose presence is required for a second event-the subsequent activation of an oncogene such as those described here (Kennedy et al., 1980). The isolation and detailed nucleotide sequence of an activated oncogene will be forthcoming over the next several years. The sequence information may shed no light on the mechanism by which any novel sequence arrangements of the oncogene were achieved. It is quite possible that the novel nucleotide sequence arises many weeks after the mutagenic activities of a carcinogen. In such cases, the mechanisms that intervene between the initial carcinogenic stimulus and the final creation of novel sequences may not be illuminated by studies on isolated activated oncogenes. X. The Role of Oncogenes in Carcinogenesis and Maintenance of Phenotype

A problem raised by the experiments described here concerns the multiplicity of alterations that are responsible for the achievement of a transformed phenotype. It would be simplistic and likely incorrect to assume that the activation of a cellular oncogene is the only necessary alteration occurring during the conversion of a normal cell into a tumor cell. Rather, a variety of other genetic and epigenetic controls may also need to be perturbed in order to fully realize the oncogenic phenotype. These other controls, whose nature is obscure, will likely be seen to cooperate with the oncogene in eliciting the tumor phenotype. Transformation would appear to be a multistep process (Armitage and Doll, 1957; Nordling, 1953; Whitternore, 1978), and the activation of the oncogene likely represents only one of several necessary alterations that make possible the final transformed phenotype. Such a postulated multistep process would seem to be inconsistent with the results of the gene transfer experiments described here. In these experiments, it appears that a single genetic determinant is able to effect the conversion of the recipient cell from a normal, untrans-


161

formed fibroblast into a highly tumorigenic fibrosarcoma cell. These existing experiments would suggest that this single genetic factor has allowed the recipient cell to achieve the total conversion from normalcy to malignancy. The resolution of this dilemma may be provided by detailed study of the characteristics of the recipient cells, which in this case are an established line of mouse fibroblasts that have been immortalized for tissue culture. It is possible that these recipients have already undergone many of the necessary predisposing alterations that normally occur during carcinogenesis in vivo. These cells may therefore be poised for transformation and highly responsive to the introduced active oncogene. These predisposing alterations may therefore eliminate the need for all but the final alteration, which in this case is provided by the transfected oncogene. XI. The Proteins Encoded by Activated Oncogenes

The use of molecular cloning techniques will make possible the isolation of a series of different activated oncogenes. However, structural analysis of the various molecular clones will provide few insights into the mechanisms used by these genes to convert a normal cell into a tumor cell. Analogy with tumor viruses would suggest that these oncogenes specify transformation proteins whose continuous presence is required to maintain the oncogenic phenotype. Such transforming proteins play a central regulatory role in the oncogenic phenotype, and therefore contrast with a myriad of other proteins whose levels are altered only as secondary consequences of oncogenic conversion. These other, secondarily regulated proteins have been extensively catalogued over the past decades. Study of the complex catalog of “transformation specific” proteins of tumor cells has provided relatively few insights into the central mechanism of oncogenic conversion. The exploitation of gene transfer should make possible the detection of a variety of centrally acting transforming proteins. This detection will come via two types of experiments. Some of these oncogenes will be isolated as molecular clones, whose detailed structural features will be resolved by nucleotide sequence analysis. Using techniques pioneered by Walter et al. (1980),the proteins encoded by these genes may then be isolated in a relatively straightforward fashion (Lerner et al., 1981; Walter et al., 1980). An alternative path for the isolation of transforming proteins is less dependent on sophisticated molecular technology and is already being

162

ROBERT A . WEINBERG

followed. This approach depends upon the fact that the transfection procedure allows one to create a mouse fibroblast line that contains the transforming gene as the only foreign tumor gene in an otherwise totally murine genetic background. Such a cell line can be expanded into a mass culture and used to seed a tumor in a young mouse. The serum of a tumored animal may then be analyzed for reactivity to any xenogeneic protein encoded by the introduced transforming gene. The immunogenicity of such a protein is not predictable. One is aided, however, by the fact that the foreign protein may be of foreign species origin and may contain antigenic determinants to which the immune system of the host mouse is not tolerant. This strategy has already been attempted for the detection of any protein encoded by the transforming gene present in the DNA of ethylnitrosurea-induced rat neuroblastomas and glioblastomas. This DNA readily induces transformation of mouse fibroblasts (Shih et al., 1981), and these fibroblasts all contain a phosphoprotein of 185,000 daltons mass. The protein is precipitated specifically by the immune sera of animals carrying neuroblastoma DNA-induced tumors (Padhy et aZ., 1982). Mouse fibroblasts transformed by a variety of other tumor DNAs do not contain the protein, nor do a variety of types of tumor cell lines, with the exception of rat neuroblastomas and glioblastomas (Padhy et aZ., 1982). The behavior of the 185,000 dalton phosphoprotein satisfies many criteria of a protein whose structure is directly encoded by the neuroblastoma-glioma transforming gene. A conservative interpretation, which is justified by currently available data, is that the protein is induced specifically by this transforming gene and by no other, transferable, transforming gene. A rigorous proof of the genetic origins of this protein depends upon detailed structural analysis of the protein, which is not yet at hand. This example is cited to illustrate an experimental strategy that will make possible the detection of a variety of tumor-specific transforming proteins over the next several years. These proteins will be of central importance in understanding the intracellular metabolic alterations that initiate and maintain the oncogenic phenotype. Antisera to these proteins should as well serve as useful reagents in the detection and diagnosis of a variety of specific tumors.

REFERENCES Andersson, P. (1980). Ado. Cancer Res. 33, 109-172. Armitage, P.,and Doll, R. (1957). B r . J . Cancer 11, 161-169. Blair, D.G.,Oskarsson, M., Wood, T. G., McClements, W. L., Fischinger, P. J., and Van de Woude, G . G . (1981). Science 212,941-943.


163

Brugge, J. S., and Erikson, R. L. (1977).Nature (London) 269,346-348. Coffin, J. M., Varmus, H. E., Bishop, J. M., Essex, M., Hardy, W. D. Jr., Martin, G. S., Rosenberg, N. E., Scolnick, E. M., Weinberg, R. A., and Vogt, P. K. (1981).J . Virol. 40,953-957. Collett, M. S., Erikson, E., Purchio, A. F., Brugge, J. S., and Erikson, R. L. (1979).Proc. Natl. Acad. Sci. U.S.A. 76, 3159-3163. Comings, D. E . (1973). Proc. Natl. Acad. Sci. U.S.A. 70, 3324-3328. Cooper, G., and Neiman, P. E. (1980).Nature (London) 287,656-659. Frankel, A. E., and Fischinger, P. J. (1977).J.Virol. 21, 153-160. Graham, F. L. (1977).A d o . Cancer Res. 25, 1-46. Graham, F. L., and van der Eb, A. J. (1973).Virology 52, 456-467. Hayward, W. S., Neel, B. G., and Asbin, S. M. (1981). Nature (London) 290,475480. Houck, C. M., Rinehart, F. P., and Schmid, C. W. (1979).J.Mol. Biol. 132, 289-306. Hughes, S. H., Payvar, F., Spector, D., Schimke, R. T., Robinson, H. L., Payne, G. S., Bishop, J. M., and Varmus, H. E. (1979).Cell 18,347-359. Karess, R. E., and Hanafusa, H. (1981). Cell 24, 155-164. Karess, R. E., Hayward, W. S., and Hanafusa, H. (1979).Proc. Natl. Acad. Sci. U.S.A. 76, 3 154- 3158. Kennedy, A. R., Fox, M., Murphy, G., and Little, J. B. (1980). Proc. Natl. Acad. Sci. U.S.A. 77,7262-7266. Klein, G., ed. (1980). “Viral Oncology.” Raven, New York. Krontiris, T., and Cooper, G. M. (1981). Proc. Natl. Acad. Sci. U.S.A. 78, 1181-1184. Lemer, R. A,, Green, N., Alexander, H., Liu, F. T., Sutcliffe, J. G., and Shinnick, T. M. (1981). Proc. Natl. Acad. Sci. U S A . 78,3403-3407. McCann, J., and Ames, B. M. (1976). Proc. Natl. Acad. Sci. U.S.A. 73,950-954. Murray, M., Shilo, B., Shih, C., Cowing, D., Hsu, H. W., and Weinberg, R. A. (1981).Cell 25,355361. Nordling, C. 0 . (1953). Br. J . Cancer 7, 68-72. Oppermann, H., Levinson, A., Varmus, H., Levintow, L., and Bishop, J. M. (1979).Proc. Natl. Acad. Sci. U.S.A. 76, 1804-1808. Padhy, L. C., Shih, C., and Weinberg, R. A. (1982). Cell (in press). Reznikoff, C. A., Brankow, D. W., and Heidelberger, C. (1973).Cancer Res. 33,32313238. Sefton, B. M., Hunter, T., and Beemon, K. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 2059-2063. Shih, C., Shilo, B., Goldfarb, M. P., Dannenberg, A., and Weinberg, R. A. (1979).Proc. Natl. Acad. Sci. U.S.A. 76, 5714-5718. Shih, C., Padhy, L. C., Murray, M., and Weinberg, R. A. (1981).Nature (London) 290, 261-264. Shilo, B., and Weinberg, R. A. (1981). Nature (London) 289, 607-609. Stehelin, D., Varmus, H. E., Bishop, J. M., and Vogt, P. K. (1976).Nature (London) 260, 170-173. Todaro, G . J., and Huebner, R. J. (1972). Proc. Natl. Acad. Sci. U.S.A. 69, 1009-1015. Walter, G., Scheidtmann, K. H., Carbone, A., Laudano, A. P., and Doolittle, R. F. (1980). Proc. Natl. Acad. Sci. U.S.A. 77, 5197-5200. Whittemore, A. S. (1978).Ado. Cancer Res. 2 7 , 5 5 4 8 .


RELATIONSHIP OF DNA TERTIARY AND QUATERNARY STRUCTURE TO CARCINOGENIC PROCESSES Philip D. Lipetz, Alan G. Galsky,' and Ralph E. Stephens Department of Radiology. The Ohio State University. Columbus, Ohio

I. Introduction 11. Background

.................................. 165 ................. ................... 167

A. The DNA Superstructure B. Nucleoid Measurement of DNA Strand Breaks ....................... C. Possible Chromatin Origins of Nucleoid DNA Supercoiling . . . . . . . . . . . D. Physiological Relevance of Decreased Nucleoid DNA Supercoiling . E. Novobiocin and Nalidixic Acid as Probes of DNA Supercoiling . . . . . ......... F. Significance of Prokaryotic DNA Supercoiling G. Significance of Eukaryotic DNA S 111. Cancer and DNA Superstructure . . . . ......................... A. Crown Gall Tumorigenesis .................... B. Chronic Lymphocytic Leukemia ..................................... C. TPA Alterations of DNA Superstructure in Normal Human Cells .......

167 169 173 175 178 185 187 189 189 193 197 . . . . . 199 IV. Conclusion . . . . . 202 References ............................ . . . . . . . . . . 204 Note Added in Proof ................................................... 210

I. Introduction

Eukaryotic DNA possesses at least four subchromosomal levels of organization. Cancer studies have focused upon the primary (nucleotide sequence) and secondary (double-stranded helix) levels of DNA organization. This chapter reviews (a) DNA superstructure as studied in nucleoids (permeabilized and protein-depleted nuclei); and (b)possible correlations between modification of the tertiary (DNA supercoiling) and quarternary (supercoiled domains) levels of DNA organization and carcinogenic processes. When the eukaryotic genome is isolated as a nucleoid, the DNA is supercoiled (DNA negative superhelicity) (Cook and Brazell, 1975; Benjayati and Worcel, 1976; Lipetz, 1981). Most eukaryotic DNA supercoiling results from the wrapping of DNA around histone core Department of Biology, Bradley University, Peoria, Illinois 61625.


Copyright 0 1982 by Academic Press, Inc. All rights of reproduction in any form reserved. ISBN 0-12006636-X

166

PHILIP D. LIPETZ ET AL.

particles to form nucleosomes (McGhee and Felsenfeld, 1980). DNA supercoiling may also represent conformational stress induced by the condensation of nucleosomes into 25-30 nm chromatin fibers (Worcel et al., 1981).Since there is no compelling evidence to necessitate the existence of a eukaryotic topoisomerase that induces generalized DNA supercoiling in a manner analogous to the prokaryotic gyrase (Denhardt, 1979; Champoux, 1978), physiologically relevant alterations of DNA supercoiling may represent a probe of chromatin structure. A causal relationship between alterations of DNA superstructure and carcinogenesis has not been demonstrated, although the following suggestive correlations have been noted.

1. Agrobacterium tumefaciens-induced tumorigenesis may be modulated by DNA supercoiling (Lipetz et al., 1981a). 2. Lymphocytes isolated from chronic lymphocytic leukemia (CLL) patients have abnormally high DNA supercoiling (Yew and Johnson, 1979a; Lipetz et al., 1981b), and differentiation of such cells (Totterman et al., 1980) is accompanied by renormalized DNA supercoiling (Lipetz et al., 1981b). 3. Treatment with the carcinogenic promoter 12-O-tetradecanoylphorbol-13-acetate (TPA) alters normal and CLL human lymphocyte and human fibroblast DNA supercoiling and the size of normal human lymphocyte DNA domains (Lipetz et al., 1981b). 4. Many carcinogens alter DNA supercoiling (Drinkwater et al., 1978; Lipetz, 1981; Lipetz et al., 1982).

The purpose of this review is to stimulate research; therefore, we will report these intriguing correlations before causality is fully established. Interpretation of nucleoid results has been traditionally retarded by two uncertainties: (a) the origin of nucleoid DNA supercoiling; and (b) the relevance of nucleoid DNA supercoiling, since DNA strand breaks result in decreases in nucleoid DNA supercoiling without corresponding decreases in in vivo DNA supercoiling. Herein w e resolve the latter problem and discuss proposed solutions to the former problem. We also review evidence that the nucleoid technique can be used to quantitate extremely low levels of DNA strand breaks (less than one per 4.4x 10" daltons of DNA). This article is not intended to be an exhaustive review of DNA superstructure and topoisomerase enzymology, but rather to impart enough information so that the reader can form an appreciation of the possible significance of DNA superstructure to the cancer problem.

RELATIONSHIP OF DNA STRUCTURE TO CARCINOGENESIS

167

It. Background

A. THE DNA SUPERSTRUCTURE O F NUCLEOIDS The DNA superstructure is conveniently studied in permeabilized and protein-depleted nuclei called nucleoids (Cook and Brazell, 1975). Gentle lysis of cells in the presence of nonionic detergents and high salt results in nucleoids that contain nuclear DNA and small quantities of bound protein and RNA (Cook and Brazell, 1976b, 1978). Nucleoids are very similar to nuclear matrix structures consisting of membrane and attached DNA (Nelkin et al., 1980). Nucleoids possess DNA supercoiling (DNA negative superhelicity) (Benyajati and Worcel, 1976; Cook and Brazell, 1975, 197613) and are partitioned into multiple supercoiled DNA domains (Benyajati and Worcel, 1976; Cook and Brazell, 1975, 1976b, 1978). Nucleoid DNA superstructure is indirectly demonstrated by techniques originally developed by Vinograd to analyze supercoiled, closed circular, double-stranded viral DNAs (for a complete review, see Bauer and Vinograd, 1974). Eukaryotic DNA supercoiling can be quantitated by sedimentation of nucleoids in ethidium bromide (EB) containing neutral sucrose gradients (Benyajati and Worcel, 1976; Cook and Brazell, 1975; Lipetz et al., 1982). The intercalation of EB into DNA tends to introduce DNA-positive superhelicity into a closed circular molecule whose DNA twist is held constant by high ionic concentration. As the EB concentration increases, these tendencies counter the preexisting negative superhelicity and force nucleoid DNA into a more extended conformation resulting in a decreased sedimentation rate. When the average preexisting negative superhelicity is countered b y equal but opposing forces of EB intercalation, then nucleoid DNA is maximally extended and the sedimentation rate of the nucleoids is minimal. Higher concentrations of EB induce positive superhelicity, and nucleoid DNA again assumes a compact conformation with a corresponding increase in the rate of sedimentation. The EB concentration required to induce minimal sedimentation is proportional to the preexisting average DNA negative superhelicity (DNA supercoiling). This quantitation of DNA supercoiling reflects the conformation of DNA from which most constraints imposed by chromosomal proteins have been removed by high salt. High salt also increases DNA supercoiling, and so the observed DNA supercoiling and any alterations due to strand breaks in DNA supercoiling will be magnified. When the NaCl concentration is lowered from the generally used concentration

168


of 2 M to a more physiologically relevant concentration of 0.2 M, nucleoid DNA supercoiling is halved (Cook and Brazell, 1978). Preliminary reports indicated that DNA supercoiling was similar to HeLa, chicken, insect, and amphibian nucleoids (Cook and Brazell, 1976a). Ethidium bromide titrations of HeLa nucleoids indicate that there is one supercoil every 90-180 base pairs (bp) (Cook and Brazell, 1977). Electron micrographs of HeLa nucleoids indicate the presence of a supercoil every 200-750 b p (Mullinger and Johnson, 1979); however, this latter methodology is subject to artifacts of mounting. At the time these demonstrations were performed the eukaryotic nucleoid technique was in the early stages of development, and therefore small variations would have been beyond the range of experimental error. As increasing DNA strand breaks are introduced, the rate of nucleoid sedimentation in neutral sucrose gradients decreases (Cook and Brazell, 1975). One DNA strand break completely relaxes the supercoiling of covalently closed circular DNA (cccDNA). One DNA strand break however, does not completely relax the supercoiling of the entire eukaryotic nucleoid genome. Therefore, there must be limits that prevent a DNA strand break from acting as a relaxation swivel for the entire genome. Cook and Brazell (1976b) proposed that eukaryotic nucleoid DNA is organized into multiple domains and that each DNA domain is constrained by nonhistonal protein and RNA in such a manner that the DNA supercoiling of only one domain can be relaxed by a single DNA strand break. (The terms “DNA loops” and “chromosome folds” are used by some authors in preference to the term “DNA domains .”) Cook (1974) originally hypothesized that a DNA domain corresponded to a replicon or chromomere unit. Vogelstein et al. (1980) have noted that DNA replication appears to start at the point where a DNA loop is attached to the nuclear matrix and then continues throughout the entire DNA domain. However, early evidence suggests that nucleoid DNA domains are too large to be single replicon units. DNA domain size appears to be approximately lo9 daltons of DNA. Cook and Brazell (1975) utilized the induction of DNA strand breaks by ionizing radiation, the assumption of single-hit kinetics, and nucleoid sedimentation in neutral sucrose gradients to estimate HeLa DNA domain size to be approximately lo9 daltons. Previously published sedimentation patterns of nucleoids exposed to ionizing radiation can be similarly analyzed to obtain DNA domain sizes of 1.8 x lo9 daltons DNA for human dermal fibroblasts (Lipetz et al., 1982) and 3.8 X log daltons DNA for rat spleen cells (lymphocytes) (Egg e t al.,


169

1977). We have also used this methodology to calculate a DNA domain size of approximately 8.4 x los daltons for normal human lymphocytes. Other estimates of nucleoid DNA domain size have resulted in much lower values. Cook and Brazell (1978) more recently used isolated HeLa nucleoids to obtain an estimate of 1.8 x 10' daltons. Mullinger and Johnson (1979) used electron microscopy of isolated nucleoids to calculate that a supercoiled domain of HeLa is 0.4 to 3 X lo* daltons. Benyajati and Worcel (1976) utilized alkaline sucrose sedimentation of DNase I-generated DNA fragments to estimate a Drosophila DNA domain size of 5.7 x lo7 daltons. Pinon and Salt (1977) used a similar technique to estimate a DNA domain size of 1.5 x 10' daltons in yeast. One possible resolution of these differences in DNA domain size involves the extreme sensitivity to mechanical disruption of nucleoid DNA (Lipetz et al., 1982). We have noted that handling (or isolating) nucleoids results in sufficient DNA strand breaks to account for the differences between results obtained with previously isolated nucleoids, and those results obtained with nucleoids created by lysis on top of the analytical gradient (as in the preceding paragraph). Further experimental variation could have been introduced either by alkaline sucrose gradients cleaving DNA at alkaline-labile sites (Brash and Hart, 1978) or by nonrandom access to nucleoid DNA by DNase I under the conditions utilized. Thus, we conclude that an average DNA domain size is approximately los daltons of DNA in length.

B. NUCLEOIDMEASUREMENTOF DNA STRAND BREAKS The nucleoid technique can be used qualitatively (Cook and Brazell, 1976c; Weniger, 1979) and quantitatively (Lipetz et al., 1982) to measure the relative increase in DNA strand breaks when control and treated samples are compared. Single-hit kinetics are used to calculate the number of evenly distributed DNA strand breaks that must be introduced to relax the DNA supercoiling of all domains and thereby results in minimal nucleoid sedimentation. The difference between the number of strand breaks required to completely relax a treated sample and the number required to relax a control is the number of preexisting DNA strand breaks in the treated sample. Thus, the nucleoid technique quantitates the relative increase in DNA strand breaks between two samples; it does not quantitate the absolute number of DNA strand breaks.

170


The sensitivity of the nucleoid technique as a measurement of DNA strand breaks can be determined by the minimum number of DNA strand breaks that induce a reproducibly detectable decrease in nucleoid sedimentation. Increments as low as that induced by 30 rad X ray [one DNA strand break per 2.2 x 10" daltons of DNA (Brash, 1979; Dean et al., 1969)] in normal human dermal fibroblasts (Fig. 1A) (Lipetz et al., 1982) and less than 17 rad X ray in normal human lymphocytes (greater than one DNA strand break per 4.4 X 10'" dal-

"0°

1

A

0 25

50 X RAY ( R I D )

75

100

FIG. 1. Nucleoid sensitivity to X-ray induction of DNA strand breaks. The minimum number of DNA strand breaks that can be resolved by the nucleoid technique was determined by inducing DNA strand breaks with X-irradiation. FS1, normal human dermal fibroblasts (foreskin), PDLp(A) and normal human mixed lymphocytes (B) were One DNA exposed to X ray and examined as previously reported (Lipetz et al., 1981~). strand break per 2.2 x 10"' daltons DNA (30rad X ray) is resolvable in FS1 cells, and less than one DNA strand break per 4.4 x 10'O daltons DNA (less than 17 rad X ray) is resolvable with mixed normal human dermal fibroblasts. Error bars indicate standard error of the mean. Our nucleoid sedimentation technique has been described elsewhere (Lipetz et al., 1982). It is essentially the method of Cook and Brazell (1976a) with several important variations: (a) the presence of DNA in individual gradient fractions is detected by postsedimentation labeling of DNA with the fluorescent dye 4',6-diamidino-Bphenylindole (DAPI); (b) gradients are fractionated with a Buchler miniscus following fractionator, thus avoiding the artifacts first described by Weniger (1979) that are created when gradients are fractionated from the bottom of the tube; (c) the lysing solution is prepared fresh daily because old lysing solution results in erratic sedimentation behavior; (d) gradients containing freshly lysed nucleoids are handled extremely gently because the method is sensitive to the extremely low numbers of DNA strand breaks induced by mechanical disruption. The migration of human dermal fibroblast DNA was examined by ultraviolet techniques as detailed in Cook and Brazell (1975).


171

B

17

34

50

67

84

X R A Y (RAD)

FIG.1B.

tons) (Fig. 1B) can be detected. The greater sensitivity of lymphocyte nucleoids to DNA strand breaks may be due to their unusually large DNA domain size acting to amplify the effect of a single DNA strand break by resulting in the relaxation of the DNA supercoiling of a greater length of DNA. In our system, the nucleoid technique approaches minimal nucleoid sedimentation at an X-ray dose (500 rad) that is at the lower limit of accurate detection by alkaline sucrose gradient techniques; indicating that the nucleoid technique should be considered only when low levels of DNA strand breaks are anticipated. The nucleoid technique is perhaps the most sensitive of the four commonly used methods of quantitating DNA strand breakage; the other three are alkaline sucrose gradients, alkaline elution, and Sarma gradients (Brash and Hart, 1978).Yew and Johnson (1979b) utilized a variant of the nucleoid technique to measure strand breaks induced by the endonuclease recognition step of DNA excision repair and demon-

172


strated that successful detection utilizing alkaline sucrose gradients required at least one order of magnitude more UV irradiation to stimulate measurable DNA repair than did the nucleoid technique. Alkaline elution techniques apparently rely upon a strand break acting as a focus from which the DNA molecule can unwind, denature, and assume a conformation capable of passing through a membrane filter (Kohn et al., 1977). More recent calculations indicate that alkaline elution techniques may not be quite as sensitive as originally thought and may be capable of resolving only approximately the number of DNA strand breaks induced by 300 rad X ray (Brash and Hart, 1978). Sarma-type gradients are alkaline sucrose gradients in which incomplete lysis of the nuclei occurs; thus, these partially lysed nuclei retain many of the features of DNA superstructure that are ordinarily removed in alkaline sucrose gradients (Zubroff and Sarma, 1976; Cox et al., 1973). We suggest that the latter two techniques rely on hydrodynamic and biophysical principles similar to the nucleoid technique but, unlike the nucleoid technique, do not fully exploit the basis of this similarity, namely, DNA tertiary and quaternary structure. Another advantage of the nucleoid technique over the three other techniques is that it utilizes neutral sucrose gradients. Therefore, what is measured is more likely to have been a DNA strand break in vivo than are DNA strand breaks measured under alkaline conditions [such alkaline conditions may have induced the breaking of alkali labile bonds, such as apurinic sites and phosphotriesters, during lysis and sedimentation (Brash and Hart, 1978)l. Examination of DNA from tissue samples or mitotically inactive cells requires techniques that do not rely on the use of radiolabel. Cook and Brazell (1975, 197613) used ultraviolet (UV) absorption to detect the location of nucleoid DNA in gradients; however, the Triton X-100 detergent used to lyse nucleoids also absorbs UV. Therefore, to avoid masking information contained in the detergent-containing upper portion of the gradient, it was proposed that nucleoids be lysed and isolated separately from the analytical procedure (Cook and Brazell, 1977), a procedure likely to introduce DNA strand breaks and reduce the sensitivity of the method (Lipetz et al., 1982). A wide variety of fluorescent indicators of DNA were examined by Lipetz e t al. (1982) as possible alternatives to UV detection of nucleoid DNA in nonradiolabeled cells. Ethidium bromide was rejected as a fluorescent agent due to formation of fluorescent complexes with non-DNA components and a lower DNA sensitivity. 4’,6-Diamidino-2phenylindole 2 HC1 (DAPI) was found to be well suited to this application since it does not significantly interact with Triton X-100 or non-DNA cellular components (Brunk et al., 1979; Kapuscinski and

-


173

Skoczylas, 1977) and is a sensitive indicator of double-stranded DNA (Lipetz et d., 1982). The sensitivity of DAPI is such that only 15,000 human dermal nucleoids (100 ng of DNA) are required to produce a peak signal of 2.5 times background fluctuation (Lipetz et al., 1982). The nucleoid technique combined with DAPI fluorescent labeling of centrifuged nucleoid DNA has been used as a technique to examine the in vivo response of rat liver cells to carcinogen exposure (Lipetz et al., 1982). The advantages of a technique that can quantitate extremely low levels of DNA damage as well as DNA supercoiling from whole animal tissue samples is obvious. Low levels of non-strandbreak forms of DNA damage can be detected by treating isolated nuclei with repair endonucleases that induce a DNA strand break at the DNA damage site (Brash, 1979) and then lysing the nuclei on top of the analytical nucleoid gradient. It is hoped that this technique will allow in vivo studies of carcinogen-induced DNA pathologies at physiological levels of carcinogen exposure. Similarly, the nucleoid technique will detect extremely low levels of DNA repair. For example, Yew and Johnson (1979b) used nucleoids, combined with the inhibition of the DNA strand-break-rejoining step of DNA excision repair, to detect the repair of the damage induced by 0.01joules/m2 UV in unstimulated lymphocytes but had to increase the UV exposure to 0.1 joules/m2 in order to detect damage by alkaline sucrose gradients. One advantage of such a technique is that it does not require hydroxyurea to inhibit scheduled DNA synthesis and thus avoids possible artifactual inhibition of unscheduled excision-repair DNA synthesis b y hydroxyurea (Collins and Johnson, 1979). C. POSSIBLECHROMATINORIGINSOF NUCLEOID DNA SUPERCOILING

DNA supercoiling primarily results from interactions between DNA and histone proteins (Crick and Klug, 1975; Sobell et al., 1976). The wrapping of approximately 140 base pairs (bp) of DNA in a nucleosomal particle (DNA plus an octomer of histones H2A, H2B, H3, and H4) can result in supercoiling of the DNA (Crick and Klug, 1975; Sobell et al., 1976; Felsenfeld, 1978; Champoux, 1978; Weintraub et al., 1976). The wrapping of DNA in a nucleosomal core particle introduces at least 1% DNA supercoil (McGhee and Felsenfeld, 1980). It is not clear whether the DNA supercoiling that results from the wrapping of DNA around nucleosomal core particles can vary and exert a regulatory influence. Such DNA supercoiling is bound in a nucleosome particle and therefore is constrained to a limited range of conformations (Champoux, 1978). However, if transcribing DNA re-

174


gions are either not nucleosome associated (Foe, 1978; Scheer, 1978; Grainger and Ogle, 1978) or are in altered nucleosome conformations (Grainger and Ogle, 1978; Butler et aZ., 1978; Johnson et al., 1978, 1979), then organization of the nucleosomal induced DNA supercoiling may be a much more physiologically significant phenomenon. Furthermore, the conformational constraints placed upon nucleosomal induced DNA supercoiling will be at least partially relaxed during passage of the DNA replication fork (see below). The DNA of each nucleosome is wrapped around the histone octomer at least 1% times; thus, one would predict the introduction of 1% DNA supercoils per nucleosome. However, the average change in DNA supercoiling induced by the reconstitution of nucleosomes onto SV40 DNA is only 1% DNA supercoils induced per nucleosome (Shure and Vinograd, 1976). Similarly, Cook and Braze11 (1977) observed an interval of 90-180 bp between the center of HeLa superhelical coils: since there are 188 b p between HeLa nucleosomes (Compton et al., 1976), only 1 to lY2 DNA supercoils are observed per human nucleosome. Thus, it appears that the observed DNA supercoiling cannot b e solely and simply the result of DNA wrapping around histones to form nucleosomal core particles, since some process must reduce the theoretical average DNA supercoiling. One possible solution to this apparent conflict is if the nucleosomeassociated DNA varies by less than 5% in its twist when compared with internucleosomal linker region DNA (Prunell et al., 1979; Rhodes and Klug, 1980; Crick, 1976). This solution is possible since, if the DNA twist does vary in nucleosome-associated DNA, it will be observed as a change that will be interpreted as a change in DNA supercoiling. The quantity measured in electrophoresis of SV40 DNA or nucleoid sedimentation is a change in the topological linking number ( L ) ,defined as the sum of DNA twist ( T )and writhing ( W )where W represents conformational deformations including DNA supercoiling, and is usually negative (DNA negative superhelicity) (Crick, 1976; Fuller, 1975). In the nucleoid technique and in electrophoresis T is assumed to be held constant by ions, and thus dL = dW. When in vivo DNA supercoiling (-W) is made more negative (increased DNA supercoiling) by nucleosomal interactions, an in vivo increases in DNA twist will result in less of a decrease in the quantity, L , which is measured in vitro than would be predicted by the alteration in DNA supercoiling alone. Prunell et al. (1979) approached this problem by demonstrating that the periodicity of DNase I digestion of DNA is approximately 10.4 bp in both nucleosomal and linker-region DNA. If DNase I digestion is assumed to reflect DNA twist, then this suggests that DNA twist does not vary in nucleosomes. However, DNA twist


175

may in fact vary in nucleosomes, but DNA-histone interactions may restrict the accessibility of DNA to DNase I digestion so that no difference in digestion pattern is visible (Prune11 et al., 1979; Rhodes and Klug, 1980; Girardet and Lawrence, 1979). Assuming that chromatin-associated DNA maintains a constant twist, Worcel et al. (1981) have proposed a model of nucleosome stacking that is mathematically consistent with both sets of conflicting data. The histone, H1, condensed 25-30 nm fiber of chromatin (Gorka and Lawrence, 1979; Matsumoto et aZ., 1980; McGhee and Felsenfeld, 1980) is proposed to consist of nucleosomes stacked with a specific arrangement of linker-region DNA such that the d L is approximately one per nucleosome. One prediction of this model is that adjacent nucleosomes are arranged in an opposing manner so as to minimize dL per nucleosome. This prediction appears to be consistent with Burgoyne and Skinner’s (1981) observation that the DNase I digestion pattern of chicken erthrocyte nuclei reflects a dinucleotide pattern where every other nucleosome is resistant to digestion. The nucleosome stacking pattern proposed by Worcel et aZ. would result in d L of one per nucleosome; however, there may be a dL of approximately 1% per nucleosome (Compton et al., 1976; Cook and Brazell, 1977; Shure and Vinograd, 1976; Lipetz, 1981). Thus, if the nucleosome stacking model is applicable, chromatin may be composed in a variety of nucleosome stacking patterns and DNA supercoiling could be both decreased and increased by varying the ratio of these types of stacking. Also theoretically possible are models in which L is altered either b y regions of positive DNA superhelicity or altered conformation to Z form DNA (Benham, 1980a); however, the transformation to Z form appears to require specific DNA sequences (Wang et al., 1979) and so is unlikely to be a generalized phenomenon. While a modified version of the nucleosomal stacking model of Worcel et d . (1981) is appealing, it is not yet possible unequivocally to determine which, if any, model is correct. However, in any of the above models, physiologically relevant alteration of nucleoid DNA supercoiling is a probe of chromatin structure, although the exact chromatin structure reflected is dependent upon which model(s) is correct. O F DECREASED NUCLEOID D. PHYSIOLOGICAL RELEVANCE DNA SUPERCOILING

One problem that has previously impeded the study of nucleoid DNA supercoiling has been that alterations of eukaryotic nucleoid DNA supercoiling do not necessarily indicate that correspondingly

176


significant in vivo alterations of DNA supercoiling and chromatin structure have occurred. Decreased DNA supercoiling may be observed either in nucleoids with intact DNA molecules or in nucleoids with DNA molecules containing DNA strand breaks (Cook and Brazell, 1975; Lipetz, 1981). Nucleoids are prepared in high salt, which removes most chromosomal constraints, thereby allowing a single DNA strand break to relax the DNA supercoiling of an entire DNA domain (Cook and Brazell, 1975; Lipetz, 1981; Lipetz et al., 1982). In contrast, DNA supercoils are constrained in vivo by association with nucleosomal core particles and chromosomal proteins such that a DNA strand break does not appear to be able to relax DNA supercoiling (Sinden et al., 1980). Sinden et al. (1980)have attempted to observe the relaxation of linker region DNA after gamma irradiation. Their results indicate either that there is no relaxation of linker region DNA or that the length of DNA relaxed per gamma-induced DNA strand break is less than 5000 bp. [Interpretation of these experiments may be further complicated by the fact that gamma irradiation induces DNA-histone cross-links that may alter the relaxation characteristics of chromatin-associated DNA (Mee and Adelstein, 198l)l. It is possible to discriminate between decreases in nucleoid DNA supercoiling induced by DNA strand breaks (which probably do not represent large alterations in in vivo chromatin structure) and decreases in DNA supercoiling of intact DNA (which probably do represent large alterations ofin viuo chromatin structure). Nucleoid DNA is organized into domains whose D N A supercoiling is independently modulated in vitro (Cook and Brazell, 1975; Benjayati and Worcel, 1976; Lipetz, 1981). A DNA strand break will result in the loss of a single domain’s, and only that domain’s, DNA supercoiling. The average D N A supercoiling of the unaffected domains will remain unchanged. Thus, both unaltered nucleoids and nucleoids containing DNA strand breaks will exhibit minimum sedimentation at the same ethidium bromide (EB) concentration. In contrast, nucleoids with altered DNA supercoiling as a result of mechanisms involving intact DNA will exhibit minimal sedimentation at a different E B concentration than will control nucleoids. Verification of this analysis was demonstrated by the radical change in sedimentation pattern induced by 75 rad X ray [one strand break per 9 X lo9daltons DNA (Brash, 1979; Dean et al., 1969; Lipetz et al., 1982)l (Fig. 2). Figure 2 demonstrates that the EB concentration at which minimum nucleoid sedimentation is obtained does not change despite the fact that DNA strand breaks have decreased the average DNA supercoiling per DNA domain (including both broken and un-

RELATIONSHIP OF DNA STRUCTURE T O CARCINOGENESIS

0

2 4 6 0 ETHlDlUM BROMIDE ( ~ t Q / m l )

177

10

FIG.2. Nucleoid sedimentation in ethidium bromide (EB) gradients after 75 rad X ray. DNA strand breaks will relax nucleoid DNA supercoiling, but probably not in uiuo DNA supercoiling. Use of the nucleoid technique to measure DNA supercoiling requires that reduced nucleoid DNA supercoiling due to DNA strand breaks be differentiatable from altered DNA supercoiling with intact DNA, which probably represents altered in uiuo chromatin structure. FS1 cells were irradiated with 75 rad X ray while on ice and then converted to nucleoids and analyzed as per Lipetz et al. (1982) except that samples were centrifuged for 100 min at 10,000 rpm. The results indicate that in the presence of preexisting DNA strand breaks (one per 9 x lo8daltons of DNA), nucleoids continue to exhibit minimum nucleoid sedimentation at the same EB concentration as do nucleoids derived from cells without induced DNA strand breaks. Thus, a change in the EB concentration at which nucleoids derived from treated cells exhibit minimum sedimentation indicates that the DNA supercoiling has been altered in a way that probably reflects in uiuo chromatin structural alterations. Error bars indicate standard error of the mean.

broken domains). When decreased nucleoid DNA supercoiling represents a change in DNA conformation that will be reflected in in uivo chromatin structures, then the EB concentration at which minimum nucleoid sedimentation is observed will be decreased. Conversely, when DNA strand breaks have resulted in a decrease in the average DNA supercoiling per nucleoid domain, which probably is not indicative of large-scale in v i m alterations of chromatin structure, then the E B concentration at which minimum nucleoid sedimentation is achieved will remain the same so long as a majority of the domains are unaffected. It should be noted that the above discussion applies mainly to the DNA supercoiling associated with the 10 nm chromatin fiber; if stacking of nucleosomes into the 25-30 nm fiber also modulates DNA

178


supercoiling, then it is relatively easy to construct models in which a linker-region DNA strand break alters the nucleosome stacking of a nucleosome oligomere. Such an alteration in nucleosome stacking would probably not be reflected in the experiments of Sinden et al. (1980). Thus, we are aware of models in which a DNA strand break might alter local regions of DNA supercoiling i n vivo; however, at the present time there is no evidence to support such a model.

E. NOVOBIOCINAND NALIDIXICACID AS

PROBES OF

DNA SUPERCOILING In prokaryotes, novobiocin and nalidixic acid have become accepted agents for decreasing DNA supercoiling (Cozzarelli, 1980). Because they also reduce eukaryotic DNA supercoiling, much effort has centered upon the metabolic alterations induced by these antibiotics in eukaryotes. Many eukaryotic topoisomerases have been isolated, and their modulation of chromatin structure may be the target of these antibiotics; however, aukaryotic topoisomerases are beyond the scope of this review. Novobiocin and nalidixic acid will be considered as preliminary model systems with which to study the significance of DNA supercoiling. In Section II1,C we shall propose that the carcinogenic promoter 12-O-tetradecanoyl-phorbol-13-acetate(TPA) is another such probe. Nalidixic acid and novobiocin have been repeatedly proposed as systems with which to study eukaryotic DNA supercoiling. Mattem and Painter (1979b) demonstrated that Chinese hamster ovary (CHO) cell nucleoid DNA supercoiling was decreased by novobiocin treatment. Mattern and Painter (1979a,b) also demonstrated that novobiocin treatment decreased initiation of scheduled replicative DNA synthesis and that similar inhibition could be induced by directly decreasing CHO DNA supercoiling with ethidium bromide. Since novobiocin also decreases prokaryotic DNA supercoiling by inhibiting the activity of subunit B of prokaryotic gyrase (topoisomerase 11) (Gellert et d.,1976a,b), Mattern and Painter (197913) suggested that such an enzyme might be present in eukaryotes. Collins and Johnson (1979) confirmed novobiocin’s activity and further demonstrated that the endonuclease recognition step of excision repair of UV-induced DNA damage was also inhibited b y novobiocin treatment. Lipetz et al. (1980a) and Mattern and Scudiero (1981)confirmed novobiocin inhibition of scheduled and repair DNA synthesis and demonstrated that similar inhibitions were induced by nalidixic acid. Since nalidixic acid also inhibits subunit A of prokaryotic gyrase (Sugino et al., 1977; Gel-

179

RELATIONSHIP OF DNA STRUCTURE TO CARCINOGEKESIS

lert et al., 1977), the total evidence appeared to be consistent with the existence of an eukaryotic topoisomerase analogous to gyrase. The significance of these demonstrations has been questioned, however, since direct efforts to detect gyrase-like activity in nuclear protein preparations have failed (Champoux, 1978). There appears to be little reason to require a gyrase, since nucleosomes are capable of inducing more eukaryotic DNA supercoiling than is observed in chromatin. Therefore, the problem would appear to be relaxation, not introduction, of eukaryotic DNA supercoiling. Finally, Edenberg (1980) demonstrated that while novobiocin did inhibit SV40 DNA supercoiling, coumermycin, another inhibitor of subunit B of prokaryotic gyrase (Gellert et aZ., 1976b), did not. This finding is inconsistent with the existence of a gyrase that would be truly analogous to the prokaryotic gyrase. Edenberg (1980) explained novobiocin inhibition of scheduled DNA synthesis by demonstrating and reviewing evidence that novobiocin inhibited the activity of eukaryotic DNA polymerases involved in scheduled DNA synthesis (Sung, 1974). Nalidixic acid also can inhibit the DNA polymerases involved in B

0

0

2

4

6

8

ETHlDlUM BROMIDE (&g/ml)

1

10

0

0

2

4

6

8

1

ETHIDIUM BROMIDE ( m / m l )

FIG. 3. Nalidixic acid treatment reduces FS1 DNA supercoiling. FS1 cells were exposed to 7 hr of either 2000 pg/ml nalidixic acid (A) or 200 p g m l nalidixic acid (B). Both concentrations induce similar decreases in FS1 DNA supercoiling, and both decreases are physiologically relevant in that they are not due to DNA strand breaks. The difference in the sedimentation pattern of A and B results from different sedimentation ) that conditions; both experiments were via the protocols of Lipetz et al. ( 1 9 8 1 ~except samples in A were centrifuged for 4% hr at 5000 rpm in a Beckman SW27 rotor, and samples in B were centrifuged for 100 min at 10,000 rpm. Error bars indicate standard error of the mean.

1

0

180


scheduled DNA replication (Nakayama and Sugino, 1980; Poulson et al., 1974). Novobiocin inhibition of DNA supercoiling might also be explained as acting via inhibition of topoisomerase I (Nakayama and Sugino, 1980; Burrington and Morgan, 1978) resulting in decreased nucleosome formation (Germond et al., 1979; Ruiz-Carrillo et al., 1979; Nelson et al., 1979, 1981; Stein e t al., 1979). In contrast, nalidixic acid does not inhibit topoisomerase I (Nakayama and Sugino, 1980; Burrington and Morgan, 1978). As possible probes of the significance of eukaryotic DNA supercoiling, nalidixic activity was compared with novobiocin activity (Fig. 4). The results of Mattern and Painter (1979b) can be reinterpreted to FIG. 4. Nalidixic acid and novobiocin inhibition of scheduled and unscheduled DNA synthesis in normal human dermal fibroblasts. Scheduled DNA synthesis of FS1 cells was inhibited by both nalidixic acid (A) and novobiocin (B) addition at time zero. Twelve glass coverslips (11 x 22 mm) were placed in each Lab-Tek 100 x 15 mm plastic petri dish. Each coverslip was individually seeded with 5 x 103 FS1 cells by beading 250 p1 of monodispersed cell suspension onto the coverslip. The cell suspension was retained on the coverslip by its greater affinity for glass and the relative hydrophobic nature of the plastic. Cell growth does not extend either onto the petri dish or onto the reverse side of the coverslip. This procedure allows precise control of the number of cells on each coverslip. Cells were allowed to attach and acclimate for 24 hr in Eagle’s minimum essential medium (MEM) supplemented with 2 mM glutamine, 1 mM pyruvate, 100 U/ml penicillin, 100 pg/ml streptomycin, 100 pg/ml Fungizone, and 10% fetal bovine serum (Flow). The cultures were then washed with phosphate-buffered saline (PBS); 10 ml of new medium (containing 3% fetal bovine serum) were added to each plate, and the cultures were incubated for an additional 24 hr. The plates were then divided into groups for experimental protocols, and one plate from each group was W-irradiated as a control. Each group was treated with the proper concentration of nalidixic acid or novobiocin in Eagle’s MEM (3% fetal bovine serum) supplemented with 2 pCi/ml L3H]Tdr(25 Ci/mmol, Amersham). At each time point (0,1.5,3,6 hr) three coverslips were removed from each dish, rinsed with PBS, and fixed in Carnoy’s solution (ETOH and glacial acetic acid, 1:3). The coverslips were dried and placed in scintillation vials containing PC5 solution (New England Nuclear). Incorporation of radioactive label, [3H]Tdr,into FS1 was quantitated in a Beckman LS-8000 scintillation counter. Nalidixic acid (C) and novobiocin (D) inhibition of unscheduled DNA synthesis (presumptive excision DNA repair) was demonstrated via a modified version of the above protocol with the addition of either nalidixic acid or novobiocin immediately after UV irradiation. Scheduled DNA synthesis was inhibited by adding 3 mM hydroxyurea to the medium whenever 3% fetal bovine serum was utilized in the above protocol. FS1 not pretreated with hydroxyurea or UV irradiation are indicated as “no pretreatment” and demonstrate the inhibition of scheduled DNA synthesis by hydroxyurea. Such inhibition of scheduled DNA synthesis was also confirmed by autoradiography. Uptake of radiolabel after UV irradiation is considered to be proportional to repair DNA synthesis. Cells washed with PBS were exposed to 10 joules/m2 of UV radiation (primarily 254 nm) from a General Electric germicidal lamp. Dosimetry was determined using a Latejet meter.

A

10.000 NALIOIXIC AGIO FS I CELLS

1000

500 300

2OC I00 50

2c 10 o.uq/ml 201u)/ml mOpg/rnl 2000,q/ml

L

I

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

30

15

60

TREATMENT (HR)

B

10,000

W W I N FSI CELLS

1000 500

300

ax I00

5c

2C

7-----

IC

/" /"o,uq/ml 20,uq/ml 200,uq/ml 20001u)/ml

I

I 5

----_--------30

TREATMENT (HR)

FIGS.4A and B.

60

C

NALlDlXlC ACID FSI CELLS

.

I

I 5

30

60

TREATMENT IHR)

2 I

I .5

30 TREATMENT IHR)

FIGS.4C and D (see legend p. 180).

60


183

show that novobiocin can induce physiologically relevant decreases of DNA supercoiling. After 6 hr of treatment with 2000 pg/ml novobiocin scheduled DNA synthesis was almost completely inhibited; 200 pg/ml induced approximately an 80% decrease; and 20 pg/ml, a 41% decrease (Fig. 4B). Unscheduled (excision repair) synthesis was decreased by approximately 98% by 6 hr of treatment with 2000 pg/ml novobiocin; 200 pg/ml induced approximately a 90% decrease; and 20 pg/ml induced approximately a 63% decrease (Fig. 4D). Nalidixic acid was examined (Lipetz et al., 1980a) to determine whether it could induce physiologically relevant decreases in DNA supercoiling in FS1 cells (normal human dermal fibroblasts). Nalidixic acid treatment (2000 and 200 pg/ml for 7 hr) decreases DNA supercoiling (Fig. 3). Treatment with 2000 or 200 pg/ml nalidixic acid results in minimum nucleoid sedimentation at 3-4 pglml EB, while minimum control sedimentation is 5 pg/ml. Clearly, these decreases in DNA supercoiling represent physiologically relevant mechanisms, not DNA strand breaks. Treatment with 20 pg/ml nalidixic acid does not result in a decrease in DNA supercoiling. In all reported FS1 DNA supercoiling experiments the cells were subjected to nalidixic acid exposure for 7 hr; 3-hr exposures altered nucleoid sedimentation but induced alterations that were not as clearly beyond the range of experimental error. This requirement for an extended exposure could reflect either low nalidixic acid activity or cell-cycle-dependent activity of the nalidixic acid target, which is discussed below. After 6 hr of treatment, 2000 pg/ml nalidixic acid induced approximately an 87% inhibition of FS1 scheduled DNA synthesis; 200 pg/ml induced an inhibition of approximately 26%; and 20 pg/ml induced approximately a 22% inhibition (Fig. 4A). It should be noted that while 2000 and 200 pg/ml nalidixic acid induced a similar response in decreasing DNA supercoiling, DNA synthesis was inhibited to different extents by these two concentrations. Thus, it appears that the inhibition of scheduled DNA synthesis b y nalidixic acid is a complex phenomenon involving more targets of inhibition than just DNA supercoiling. It is not known to what extent these results reflect direct inhibition of DNA polymerases. Unscheduled (excision repair) synthesis was decreased with 6 hr of treatment with 2000 pglml nalidixic acid to approximately 79% of the untreated value; 200 pg/ml induced a decrease of approximately 34%; and 20 pg/ml induced a decrease of approximately 26% (Fig. 4C). Nalidixic acid inhibition of DNA repair showed a definite dose-time response; at 1.5 hr of treatment neither 20 pg/ml nor 200 pg/m1 inhibited DNA repair; at 3 hr of treatment 20 puglml did not inhibit repair;

184


and at 6 hr both doses inhibited DNA repair (Fig. 4C). The observation that inhibition of DNA repair and DNA supercoiling both require extended exposure for maximal inhibition is consistent with DNA supercoiling directly modulating DNA excision repair. Neither nalidixic acid nor novobiocin are known to inhibit excision-repair polymerases (Poulson et al., 1974; Sung, 1974; Edenberg, 1980). It should be noted that nalidixic acid exposure does not inhibit excision repair in Escherichia coli (Simon et al., 1974), and therefore it is unlikely that nalidixic acid is acting via inhibition of a topoisomerase truly analogous to prokaryotic gyrase. The observation that both 2000 and 200 pglml nalidixic acid induced similar decreases in DNA supercoiling may suggest that the long-exposure requirement is due to cell-cycle dependence rather than low nalidixic acid activity. We examined the activity of 2000 and 200 pglml nalidixic acid upon confluent, contact-inhibited FS1 cells that were not undergoing cellular division or DNA synthesis [Go cells (Pinon, 1978)l. In five trials, we were unable to detect any significant decrease in the supercoiling of nalidixic acid-treated cells resting in Go.These results are consistent with the target of nalidixic acid activity being cell-cycle dependent. The requirement for extended exposure in a population of dividing cells could reflect the fact that nucleoid gradients quantitate the average DNA supercoiling of all nucleoids. Hence, a sufficient proportion of treated cells must have entered a sensitive phase of the cell cycle in order that a decreased nucleoid migration can be manifest. The nalidixic acid target that modulates DNA supercoiling appears to require cell cycle progression. In DNA synthesis, nucleosome structure and stacking are restored shortly after DNA replication fork passage (DePamphilis and Wasserman, 1980). Interference with this process would result in altered DNA supercoiling. Histones H3 and H4 are transferred from preexisting chromatin to newly replicated chromatin (Jackson and Chalkley, 1981)and may act to introduce DNA supercoiling in the newly replicated DNA (Nelson et al., 1981; BinaStein and Simpson, 1977; Camerini-Otero e t al., 1976). Replication fork-induced DNA twisting creates regions of positive superhelicity (Champoux, 1978; Drlica et al., 1980). One of the functions of prokaryotic gyrase is to act locally to relax this replication fork-induced positive DNA superhelicity (Drlica et al., 1980). It is not known what eukaryotic topoisomerases perform this function. Both type I and type I1 eukaryotic topoisomerases can relax DNA positive superhelicity (Champoux, 1978; Hsieh and Brutlag, 1980; Liu et al., 1980). Failure

RELATIONSHIP OF DNA STRUCTURE TO CARCINOCENESIS

185

to relieve local positive superhelical supercoiling could result in decreased nucleoid DNA supercoiling and may interfere with chromatin restoration after replication fork passage. If nalidixic acid and/or novobiocin act by inducing local regions of positive DNA superhelicity, then such alterations would probably be confined to a relatively small portion of the in vivo DNA domain and so probably could not serve as a probe of the generalized significance of DNA supercoiling. It is not clear what are the target(s) of nalidixic acid or novobiocin. Inhibition of scheduled DNA synthesis by these antibiotics is a complex process that does not clearly involve DNA supercoiling. There appears to be a stronger correlation between inhibition of DNA excision repair and inhibition of DNA supercoiling. This relationship is detailed in Section II1,D.

F. SIGNIFICANCE OF PROKARYOTIC DNA SUPERCOILING Since no significant body of evidence implicates or excludes DNA supercoiling as a primary regulator of eukaryotic DNA supercoiling, we are forced to rely on the prokaryotic literature in order to examine such a correlation. Prokaryotic DNA supercoiling is primarily regulated by topoisomerase I relaxation of DNA supercoiling and topoisomerase I1 (gyrase) introduction of DNA supercoiling (Cozzarelli, 1980). Although prokaryotic DNA supercoiling is partially stabilized b y histone-like proteins (Pettijohn and Pfenninger, 1980; RouviereYaniv and Gros, 1975; Varshavsky et al., 1977; Rouviere-Yaniv, 1977, 1979; Griffith, 1976), there are no known prokaryotic systems that stabilize and introduce DNA supercoiling in a manner truly analogous to eukaryotic chromosomal proteins. [For a review of the origins of prokaryotic DNA supercoiling, see Denhardt (1979) and Cozzarelli (1980).] Prokaryotic DNA supercoiling has been shown to regulate DNA metabolism and gene expression (Denhardt, 1979). Such demonstrations are possible because it is relatively easy to decrease prokaryotic DNA supercoiling. There are significant differences between the mechanisms of prokaryotic and eukaryotic gene regulation (Lewin, 1980; Marx, 1981). We will consider the prokaryotic literature regarding modulation of gene expression while bearing in mind the possible limited applicability of such results to eukaryotes. Examination of bacterial and viral DNAs of varying supercoiling has revealed that supercoiling may modulate (a) gene expression as measured by patterns of protein synthesis (Yang et al., 1979; Smith et al.,

186


1978; DeWyngaert and Hinkle, 1979) and patterns of RNA synthesis (Smith et al., 1978; DeWyngaert and Hinkle, 1979; Botcham et al., 1973; Botchan, 1976); (b) RNA polymerase binding to DNA, both quantitatively and qualitatively (Richardson, 1975; Wang, 1974); (c) DNA replication (DeWyngaert and Hinkle, 1979; Itoh and Tomizawa, 1977; Crumplin and Smith, 1976; Marians et al., 1977; Staudenbauer, 1976; Pietsky et al., 1972); and (d) DNA recombination including recombinative integration of virus into host DNA (Holloman et al., 1975; Holloman and Radding, 1976; Radding, 1978; Abremski and Gottesman, 1979; Kikuchi and Nash, 1979). Direct evidence suggests that the expression of some prokaryotic and bacteriophage genes is regulated by DNA supercoiling. Decreasing DNA supercoiling inhibits the transcription of late, but not of early, genes in T7 bacteriophage (DeWyngaert and Hinkle, 1979). Decreasing DNA supercoiling alters patterns of protein synthesis from E . coli, ColEl plasmid, and phage genomes (Yang et al., 1979; Smith et al., 1978). Decreasing DNA supercoiling modulates the activity of lactose, maltose, and tryptophanase, but not threonine and trytophan in E . coli (Sanzey, 1979). There is a degree of correlation between those genes that can be modulated by decreased DNA supercoiling and sensitivity to catabolic repression (Sanzey, 1979; Shuman and Schwartz, 1975). Smith (1981) has reviewed evidence that supX mutations that alter gene expression of Salmonella may be topoisomerase I-difficient mutants and that such mutants may alter gene expression by increasing DNA supercoiling. DNA supercoiling may modulate prokaryotic and viral gene expression by altering the denaturation of promoter regions, thereby creating single-stranded regions that favor RNA polymerase binding (Botchan, 1976; Benham, 1979; Hsieh and Wang, 1975; Vollenweider et al., 1979). The initiator region of many promoters is A-T rich and thus more susceptible to denaturation than non-A-T-rich regions (Benham, 1979, 1980b; Botchan et al., 1973; Botchan, 1976; Vollenweider et al., 1978; Hossenlopp et al., 1974). Increased DNA supercoiling creates stress that denatures such regions (Botchan, 1976; Benham, 1979; Hsieh and Wang, 1975; Vollenweider et al., 1979; Brack et al., 1975; Delius et aZ., 1972; Beerman and Lebowitz, 1973; Dean and Lebowitz, 1971). It has been demonstrated that supercoiling stress controls the utilization of promoters for RNA polymerase binding (Levine and Rupp, 1978; Richardson, 1975; Wang, 1974). A comparison of the 429, A, M13, and SV40 viral partial denaturation maps and RNA polymerase binding sites indicates that A-T-rich sites coincide with some, but not


187

all, of the in uitro polymerase binding sites (Sogo et al., 1979; Botchan, 1976; Vollenweider and Szybalsky, 1978; Dasgupta et al., 1977; Wasylyk et al., 1979). Denaturation maps of the 4x174 system can be similarly interpreted as predicting promoter regions that are not suggested b y gene mapping (Funnel1 and Inman, 1979); and RNA polymerase binding experiments appear to have supported these predictions (Rassert and Spencer, 1978). Thus, it appears plausible that DNA supercoiling may act differentially to control prokaryotic gene expression by altering RNA polymerase binding.

G . SIGNIFICANCE OF EUKARYOTIC DNA SUPERSTRUCTURE Cook (1973) gave momentum to the study of eukaryotic DNA superstructure with his proposal that DNA superstructure controlled differentiation. Although conceived before subchromosomal eukaryotic DNA superstructure had been elucidated, the model was a brilliant speculation. Akrigg and Cook (1980)have since shown that abnormally increasing HeLa nucleoid DNA supercoiling results in a dramatic increase in in uitro transcription by wheat germ RNA polymerase. However, other than in our demonstrations that TPA, a known modulator of eukaryotic gene expression, alters DNA superstructure, there is no direct evidence involving chromatin to support this hypothesis. Studies are underway in our laboratory to further examine Cook’s hypothesis. Two studies indicate that specific genes may have well defined locations with the loop of DNA represented by a DNA domain. Nucleoids are digested to various degrees with either nucleases or restriction endonucleases, and the DNA that remains attached to the nuclear matrix is separated from the digested fragments. Nuclear matrix attached DNA is transferred to a filter and hybridized against known probes. Nelkin et al. (1980) reported that SV40 genes are preferentially located near nuclear matrix attachment sites of SV40-transformed 3T3 cells. Also, Cook and Braze11 (1980) reported that a-,but not /3- or y-globin genes are preferentially located near the nuclear matrix attachment sites in HeLa. It should be cautioned that it has not yet been fully established that association of a gene with the nuclear matrix attachment sites indicates that it is preferentially associated with such sites; however, the observation that certain globin genes are not associated with the nuclear matrix tends to support such a hypothesis. As discussed by Nelkin et al. (1980), the association between the

188


nuclear matrix and DNA domains may be different for transcribing and nontranscribing domains (Faiferman and Pogo, 1975).The nuclear matrix may provide support for transcriptional activities in a manner similar to what occurs in DNA replication in prokaryotes (Nelkin et al., 1980; Pardoll et al., 1980).Support for this hypothesis includes (a) the association of hnRNA and snRNA with the nuclear matrix (Miller et al., 1978a,b; Herman et al., 1978); (b) the specific binding of steroids to the nuclear matrix of target tissues (Barrack et al., 1977; Barrack and Coffey, 1980); (c) the preferential association of transcribed SV40 genes with the nuclear matrix of SV40-transformed 3T3 cells, while nontranscribed globin genes are not so associated (Nelkin et al., 1980); (d)alteration of DNA domain size in brain cells during mouse fetal and neonatal stages of development (P. Lipetz, 1981;unpublished observations); and (e)the alteration of lymphocyte DNA domain size under the stimulus of TPA, a known modifier of lymphocyte gene expression (Lipetz et al., 1981b) (see below). Transcriptionally active genomes are preferentially sensitive to DNase I digestion (Weintraub and Groudine, 1976; Garel e t al., 1977). Such sequences remain DNase I sensitive after transcription has been shut off, and DNase I sensitivity is independent of transcription rate (Weintraub and Groudine, 1976; Garel e t al., 1977; Younget al., 1978). DNase I sensitivity is increased when DNA supercoiling is increased (Campbell and Jackson, 1980). This may either imply increased DNA supercoiling in eukaryotic transcriptionally active regions, or else the increased DNase I sensitivity may reflect other structures. Weisbrod et al. (1980) suggest that HMG 14 and 17 (NHCP proteins) binding to chromatin may, in part, account for increased DNase I sensitivity. They hypothesize that such binding may alter basic nucleosome structure. Sandeen et al. (1980)have shown that the major sites of HMG 14 and 17 interaction is near the ends of the nucleosomal core DNA. Such interactions might alter DNA-histone interactions and thus might alter local supercoiling. While there exist conflicting reports as to whether histones and/or nucleosome conformation along the DNA is altered in transcriptionally active regions of the genome (Franke e t al., 1976; Johnson et al., 1978; Foe, 1978; Scheer, 1978; Grainger and Ogle, 1978; Butler et al., 1978), such alterations might also modify local supercoiling stress upon the DNA. These findings are compatible with, but fall considerably short of proving, the hypothesis that an alteration of higher order structures (including DNA superstructure) reflect a permissive condition in which other components (including protein binding) can modify the rate of transcription.


189

Ill. Cancer and DNA Superstructure

A. CROWNGALLTUMOFUGENESIS The most straightforward correlation between altered DNA supercoiling and tumorigenesis is in the crown gall tumor system (Lipetz et al., 1981a). Crown galls are plant neoplasms arising from the integration of T DNA sequences from the Ti plasmid of Agrobacterium tumefaciens into dicotyledonous plant cell DNA (Watson et al., 1975; Thommashow et al., 1980; Chilton et al., 1977; Zambryski et al., 1980; Yadau et al., 1980; Willmitzer et al., 1980).The Ti plasmid system has been studied extensively since it is also a natural vector for introducing DNA sequences into plant cells (Klapwijki et al., 1978; Bomhoff et al., 1976; Montoya et al., 1977; Hernalsteens et al., 1980). There are obvious analogies between the integration of a cccDNA plasmid into a eukaryotic genome to induce neoplastic transformation and the integration of a cccDNA virus (such as SV40) into a eukaryotic genome to induce neoplastic transformation. However, unlike the viral systems, probes exist that will both increase and decrease the DNA supercoiling of the Ti plasmids and so allow the relationship between tumorigenesis and DNA supercoiling to be examined. Ti plasmid DNA supercoiling can be manipulated (Lipetz et al., 1981a). Novobiocin and nalidixic acid are antibiotics that decrease DNA supercoiling by inhibiting topoisomerase I1 introduction of DNA supercoiling (Cozzarelli, 1980; Gellert et al., 1976b, 1977; Sugino et al., 1977). Lipetz et al. (1980b, 1981) have demonstrated that under physiological cation conditions, physiological concentrations of polyamines (spermidine and spermine) inhibit the in vitro DNA superhelical relaxing activity of Micrococcus luteus. Similarly, spermidine inhibits the activity of purified A. tumefaciens and E . coli DNA superhelical relaxing enzymes. Spermidine also enhances in vitro M . luteus and E . coli topoisomerase I1 activity (Kung and Wang, 1977; Gellert et al., 1976a). Therefore, it is possible that spermidine may exert coordinate control over opposing enzyme activities to maximize prokaryotic DNA supercoiling. Alteration of Ti plasmid DNA supercoiling was verified by agarose gel electrophoresis. Assuming a constant molecular weight, the electrophoretic mobility of covalently closed circular (ccc) DNA in an agarose gel can be proportional (nonlinearly) to DNA supercoiling: the more negatively superhelical the molecule, the faster the migration (Shure and Vinograd, 1976; Keller, 1975) (see legend of Fig. 5). The

190


1

0.I

1.0 10.0 SPERMIDINE (mMl NALlDlXlC ACID or NOVOBIOCIN bg/ml)

FIG. 5. Ti plasmid agarose gel electrophoretic mobility after treatment. Spermidine increased DNA supercoiling, and either nalidixic acid or novobiocin decreased Ti plasmid DNA supercoiling. Agrobacterium tumefaciens were cultured as per Favus et al. (1977) and incubated with spermidine, nalidixic acid, or novobiocin for 8 hr. Ti plasmids were extracted and subjected to electrophoresis in 0.8% agarose; the gels were ethidium bromide stained and photographed as per Birnboin and Doly (1979). The resulting photographs were scanned on an Ortec 4310 densitometer, and the distance migrated was measured. PM2 RFI bacteriophage controls were included in all runs to demonstrate that the Ti band had migrated proportionately to its molecular weight. Preincubation of PM2 RFI with spermidine did not alter migration. Molecules whose DNA supercoiling is above a certain range form a single fast migrating band during electrophoresis; the superhelicity of these molecules may be altered without affecting their electrophoretic migration (Shure and Vinograd, 1976: Keller, 1975). Ti plasmids exhibit a distribution about a migration value (not a sharp band) indicating that Ti plasmid negative superhelicity is within the range where electrophoretic mobility is nonlinearly proportional to DNA supercoiling. With covalently closed circular DNA molecules of less molecular weight, agarose gel electrophoresis of molecules within the superhelical range where electrophoresis is nonlinearly proportional to superhelicity should result in the appearance of distinct bands varying in superhelicity; this does not occur with Ti plasmids because the DNA molecule is large (90to 120 x 108 daltons) and possesses too much superhelicity for DNAs varying in linking number, L , by only one to form distinct bands in the narrow separation agarose gel electrophoresis induces in such large molecules. The error bars indicate SEM.


191

O' 1

Y

Y

n

OJ

0.1

1.0

10.0

PERMIDINE (mM) NALlDlXlC ACID or NOVOBIOCIN @g/ml)

EZG. 6. Rate of crown gall tumorigenesis after treatment. Spermidine pretreatment increased the rate of crown gall tumorigenesis, whereas either nalidixic acid or novobiovin decreased the formation of crown gall tumors. Spermidine- , nalidixic acid-, or novobiocin-treated cells were cultured on potato disks to determine the rate of tumor formation (Favus et al., 1977).All experimental points were the subject of at least two independent experiments of at least 15 disks, and the standard error of the mean of all points was less than 10%.

agarose gel electrophoretic mobility of selected linearized controls was examined to ensure constant molecular weight, therefore, migration is assumed to be proportional to DNA supercoiling. Figure 5 demonstrates that both novobiocin and nalidixic acid decrease the agarose gel migration rate of Ti plasmids isolated from B6 A . tumefaciens. When A . tumefaciens were incubated with spermidine the agarose gel migration rate of isolated Ti plasmid increased (Fig. 5 ) , indicating an increase in DNA supercoiling. Agrobacterium tumefaciens from the aforementioned cultures was tested for virulence on potato disks by the method of Favus et al. (1977) (Fig. 6). Cultures grown in novobiocin or nalidixic acid possess decreased tumorigenic capacity, whereas those grown in spermidine exhibit increased tumorigenic capacity (Fig. 6). The rate of tumor growth was constant in all experimental protocols, and the removal of antibiotic or spermidine before infecting potato disks with the bacteria did not alter results. Thus, it was a bacterial, not a plant, process that

192


LJNEAR CORRELATION COEFFICIENTS" Treatment

r

Spermidine Nalidixic acid Novobiocin Combined

0.996 0.981 0.878 0.916

Significance (P)

IgG3 > IgGl > IgG2b > IgG2a, which correlates directly with the 5‘ to 3’ heavy-chain constant-region gene order in the mouse. This finding may suggest that isotype switching in the absence of T cells is a probabilistic event directly related to the distance of a particular constant-region gene from the 5‘ end of the C, domains involved in gene switching (see Section II,B,5). However, when such mice were immunized in the presence of T cells, substantially higher levels of IgG2 antibody were produced, showing that T cells even in T-independent systems can promote class switching, as has also been reported in other systems (Brayley-Mullen, 1974). With regard to affinity maturation, analyses of the V region genes of IgM- as compared to IgA- or IgG-expressing cells have shown that more variations are found in VH gene segments in IgA- and IgG- as compared to IgM-producing cells (Gearhart et al., 1981).This intriguing finding suggests that affinity maturation not only is the result of a selection of B-cell clones committed at the pre-B-cell level to high affinity antibody production as suggested by the studies of Julius and Herzenberg ( 1974), but that genetic changes (“mutations”) may take place in the V region of mature B cells leading to antibodies of higher affinity.

3. Zdiotypes, Networks, and Suppression In the interaction between cells, soluble factors, and antibodies of the immune system, antigen obviously plays an essential role. However, as was first postulated by Jerne (1974), idiotypes may play important roles. As suitable test systems have become developed, idiotypes have in fact been shown to be capable of “mimicking” antigen in almost every aspect. This applies to post-receptor repertoire expression, as mentioned in Section II,C, generation of T-cell help (Gleason et al., 1981), and particularly suppression, where the documentation now is extensive (see below). As envisioned by Jerne (1974), idiotype anti-idiotype reactions put the immune system into networks of interlinked circuits by which an antigen or an idiotype not only will induce an immune response toward itself, but also elicit network perturbations. Whether idiotypic networks have distinct roles to play in B-cell development, particu-

234

TORE CODAL AND STEINAR FUNDERUD

larly in the development of early B cells to virgin B cells, remains unclear. But the fact that both idiotypes (Bona et al., 1981) and antiidiotype antibodies (Hiernaux et al., 1981) may induce expression of silent clones clearly suggests this as a possibility. T-cell-mediated suppression of the B-cell system represents a very complex circuit involving different T cells and factors. These factors may carry anti-antigenic or anti-idiotypic determinants. Suppression may be specific or, under certain conditions, nonspecific (see Germain and Benacerraf, 1981). An essential consideration from our point of view is that T-suppressor cells have T-helper cells as target. Thus, according to Gershon (1980),there is no evidence that suppressor cells can act directly on B cells with the following exception. Lynch et al. (1979) have provided evidence that T-suppressor factors may prevent the secretion of Ig from myeloma cells. This finding again underlines the need for separate studies of different aspects of B-cell development. Ill. B-Cell Neoplasms

B-cell neoplasms represent a very heterogeneous group of diseases ranging from rapidly progressive ones such as ALL to very slowly progressing conditions found among CLL patients. It comprises stem cell neoplasms as well as cells in an end stage of differentiation (myelomas). However, in most of the conditions the precise relationship to normal B-cell development remains unclear. This is to a large extent due to a lack of knowledge about normal B-cell development. Moreover, precise prognostic factors are also lacking. Thus, both at the basic as well as at the clinical level much information is lacking. Nevertheless, most of the neoplasms have some features in common, such as monoclonality and maturation arrest. There are, however, some informative exceptions to these general features, which therefore require some discussion. A. GENERALFEATURES AND THEIREXCEPTIONS

1. Monoclonality To an immunologist diagnosing B-cell neoplasia expressing Ig, the single most important feature is light-chain isotype exclusion. Although this is not a formal proof of monoclonality, there is extensive evidence for monoclonality in B-cell tumors (see Fialkow, 1976). In an examination of more than 200 lymph node biopsies we have not yet

B-CELLLYMPHOMAS

AND

B-CELLDEVELOPMENT

235

seen a single case with a monoclonal B-cell staining pattern without histopathological or clinical evidence of malignant lymphoma. On the other hand, a lack of light-chain isotype exclusion may be found, but in such cases it is difficult to determine whether the cell suspension is representative of the neoplastic population. Using immunochemistry with reagents purified b y immunoadsorbents, a much lower occurrence of so-called bitypic cases of lymphoma has been found with the immunoperoxidase technique (Landaas et al., 1981) than was previously reported by Taylor (1978a). However, from a pathogenic point of view, monoclonality appears to be a fairly late event. Thus, there are several interesting conditions that suggest that the neoplastic process is initiated by a polyclonal phase followed by a selective process resulting in a monoclonal neoplastic population. Examples of these follow. a. Epstein-Burr Virus Znfection in Immunodeficient Subjects. Epstein-Barr virus, (EBV) can cause a spectrum of clinical manifestations in patients with immune defects against EBV (Purtilo, 1980). These include polyclonal B-cell lymphomas directly associated with primary infection with EBV (Robinson et al., 1980). b. Angioimmunoblastic Lymphadenopathy. This condition is characterized by hypergammaglobulinemia, fever, weight loss, occasionally a rash, and general lymphadenopathy. Histopathologically the condition is characterized by polyclonal immunoblastic and plasmacell proliferation and vascular proliferation (Rappaport and Moran, 1975). This condition is apparently triggered b y a hypersensitivity reaction (Schulz and Yunis, 1975) and may progress to malignant lymphoma (Lukes and Tindle, 1975). c. Autoimmune Conditions. Certain autoimmune conditions-most notably Sjogren’s syndrome, in which lymphoid cell infiltration of salivary and lacrimal glands is found-may progress to a more general involvement called “pseudolymphoma,” and terminate in malignant lymphoma (see Tala1 et al., 1980).This transition is often associated with a shift from polyclonal to monoclonal B-cell populations (Zulman et al., 1978). These clinical conditions, as well as other experimental data as outlined by Klein (1979), suggest that the pathogenesis of lymphomas is a multifactorial process that may often be initiated by a polyclonal B-cell proliferative stage, and that monoclonality represents a relatively late event, due to a selective process, in the case of EBV apparently mediated by immune-regulatory mechanisms, of a tumorigenic (“autonomous’’) clone of cells (Klein, 1979). Such cells often carry characteristic chromosomal changes (Klein, 1979).

236

TORE GODAL AND STEINAH FUNDEHUD

2. Maturation Arrest This phenomenon is quite striking in most cases of non-Hodgkin lymphomas and CLL. For example, in B-cell lymphomas expressing sIgM + sIgD the proportion of cells expressing sIgM is very similar to those expressing sIgD (Godal et al., 1981b), suggesting that both receptors are expressed simultaneously on the great majority of neoplastic B cells. On the other hand, there is clear evidence that the maturation arrest may be incomplete, and various maneuvers to release cells from maturation arrest in vitro have been reported. a. Zncomplete Maturation Arrest in Vivo. This is found in a proportion of B-cell lymphomas in which different stages of B cells from small lymphocytes to plasma cells may be found. Such cases belong to the lymphoplasmacytoid groups (see Lennert, 1978). Another case in point is diffuse lymphocytic lymphoma, in which Ig-containing immunoblastic cells may be found in so-called “maturation zones” contrasting the monotonous diffuse infiltration of small lymphocytes in such cases (Landaas et al., 1981).The presence of monoclonal Ig in a proportion of CLL with identical idiotype to that found on the CLL cells represents a third example (e.g., Fu et al., 1978). With regard to surface markers, the lymphoplasmacytoid group is heterogeneous (Godal et al., 1981b).This indicates that B-cell maturation toward plasma cells can take place from different subsets of B cells (see Section 111,D). b . In Vitro-Promoted Maturation and Diflerentiation. Maturation, and in some cases differentiation, of B-cell neoplastic populations in vitro has been reported. Based on the leads given by studies on murine erythroleukemia (see Marks and Riilcind, 1978), a large number of agents have been tried. Some of those found successful are described below. i. T-cell factors. Fu et al. (1978) have reported induction of plasma cells with allogeneic T cells in a high proportion of CLL cells synthesizing IgM of the same idiotype as sIgM present on CLL cells. DNA synthesis was not required for this induction. This maturation was found in two cases of CLL with monoclonal components in serum. T cells of these patients were found to be defective in providing help to normal B cells. ii. Phorbol ester and anti-immunoglobulin. Totterman et al. (1980, 1981a,b) have reported maturation toward plasma cells with cytoplasmic Ig synthesis in a high proportion of CLL cells by 12-O-tetradecanoyl phorbol-1Sacetate (TPA). Also in this case, the maturation was not associated with cell proliferation.

B-CELLLYMPHOMAS AND B-CELLDEVELOPMENT

237

Independently, we (Godal et al., 1981a,c) have found that TPA + anti-Ig can induce morphological alterations and proliferation in a proportion of B-cell lymphomas. In some of these, increased synthesis of intracellular Ig has also been detected. One case of sIgM + sIgD nodular lymphoma has been extensively studied. In this particular lymphoma, anti-IgD + TPA have been found to induce proliferation only, while anti-IgM + TPA induced both proliferation and Ig synthesis (Ruud et al., 1981).This case illustrates that differentiation (i.e., a bifurcation process) can be achieved in vitro. This case also suggests that sIgD may lead to a “readout” different from that of sIgM on the same cell. Some experimental studies are relevant to these findings. In a murine B-cell leukemia (BCL,), Isakson et al. (1981)have reported induction of proliferation with Sepharose-coupled anti-IgM or antiIgD. By adding T-cell factors, secretion of Ig could also be achieved with both anti-p and anti-6. Although their findings and ours are not strictly compatible since they did not need to use TPA, it is possible that the differences with regard to anti-8 may be due to different subsets. Thus, BCLl carries sIgM/sIgD in a high ratio and lacks CR receptors, whereas our particular lymphoma carries twice as much sIgD as sIgM and also carries CR. Both the findings of Totterman et ul. (1981a) and ours show that different subsets of B cells may be triggered to increased Ig synthesis and morphological transformation toward plasma cells, providing further support for our previous conclusion that plasma cell maturation may take place from different subsets of B cells. iii. Epstein-Burr virus (EBV)and other B-cell mitogens. EBV may be used to convert EBV-negative lymphoma cell lines such as Ramos. By comparing markers on the converted sublines as compared to the original Ramos, Spira et al. (1981) found that the converted sublines acquired sIgD. Transformation of CLL cells with EBV has been reported in a few cases, in which evidence of maturation also was found with regard to Ig synthesis and secretion (Hurley et al., 1978; Karande et al., 1980). Robert (1979) has reported both induction of DNA synthesis and Ig synthesis and secretion in CLL cells with mitogens, such as phytohemagglutinin, pokeweed mitogen, dextran sulfate, lipopolysaccharide, and anti-&-microglobulin and EBV. However, his findings with regard to proliferation are at variance with those of a number of other investigators, including our own (for references, see Godal et al., 1978), who have reported a lack of mitogen responsiveness in CLL cells. The only mitogen we found to which some CLL cells responded was the Ca2+ionophore A23187, suggesting together with other data

238

TORE GODAL AND STEINAR FUNDERUD

that there is a cell membrane-associated block to mitogens in CLL cells. With regard to Robkrt’s report on Ig synthesis and secretion, the proportion of cells showing light-chain-specific plaques (reversed plaque assay), was in general small (31 months, respectively. Also the 5-year survival of patients with no evidence of disease was significantly in favor of those treated with alternating chemotherapy (84%) compared to MOPP alone (54%). Because of the limited number of patients at risk for 5 years, the difference in total survival was not significant. It is important to emphasize that in both treatment groups the survival curves leveled off after the second year from starting treatment. The conclusion from this study was that the cyclical delivery of two noncross-resistant combinations offers a promise for higher CR rate, longer duration of RFS, and possibly higher cure rate of advanced Hodgkin’s disease compared to the continuous administration of MOPP alone. A prospective randomized study similar to that carried out in Milan is in progress at the NCI (26). It employs alternating cycles of MOPP with cycles of a combination of streptozotocin, CCNU, adriamycin, and bleomycin (SCAB). Only a limited number of patients with stages 11, 111, and IV have so far been entered into the trial, and therefore the results are premature (120). V. Prognostic Factors Influencing Current Strategy

A. AGE, HISTOLOGY, A N D SYMPTOMS Although impressive advances have been made in the treatment of Hodgkin’s disease, in some instances prognosis remains bleak. Improvement could possibly reside in the accurate identification of pre-

2.78

GIANNI BONADONNA AND ARMANDO SANTORO

treatment variables affecting the therapeutic response. A number of factors including age, sex, histopathologic type, anatomical extent of disease, presence or absence of systemic symptoms, treatment modality, and response to therapy are strictly correlated to the prognosis of Hodgkin’s disease. It is noteworthy that a more effective prognostic index could be constructed by considering multiple prognostic factors simultaneously (64).In fact, frequently the association of more prognostic factors in the same group of patients reduces the value of each one separately. It is also important to emphasize that modern aggressive treatment with cyclical chemotherapy or chemotherapy plus radiotherapy, has a tendency to blur the influence of classical prognostic factors such as sex, age, histologic type, and pathologic stage. The overall survival in the pediatric age group is now slightly superior to that of young adults, and in general the prognosis for patients older than 40 years remains less favorable than that for younger patients (64).At present, however, there are no treatment approaches that specifically take into consideration the age factor for adults with Hodgkin’s disease. On the contrary, as will be discussed later, a modification in the treatment strategy is being applied to children under the age of 10 to minimize morbidity from radiation therapy. Results with MOPP chemotherapy (31) and combined treatment modality identified age as an important prognostic factor. However, the observation made by the Yale group (35)that chemotherapy plus low-dose radiotherapy produced the best results in patients under 40 years of age was not confirmed by the Milan group (9) utilizing a similar strategic approach. In patients subjected to modern megavoltage irradiation, the prognosis of mixed cellularity, once almost as poor as that of lymphocyte depletion, has markedly improved, with an RFS that plateaued at 50% at about 10 years. When the analysis was further subdivided according to stage, there was no longer any significant difference in either RFS or survival between mixed cellularity and nodular sclerosing lymphoma in patients with stage I, 11, and I11 disease (64).Thus, at present, in patients who are candidates for primary irradiation there is no particular strategic approach based on histology. In patients subjected to MOPP chemotherapy (31), nodular sclerosing Hodgkin’s disease had significantly shorter RFS than mixed cellularity or lymphocytedepleted lymphoma. Since there is preliminary evidence that such a difference was not observed after ABVD alone or combined with radiotherapy or alternated with MOPP (9,98),future trials should also consider the use of this new combination regimen in the presence of nodular sclerosis histology.

TREATMENT EVOLUTION I N HODGKIN’S DISEASE

279

TABLE VI PROGNOSTIC FACTORS IN HODGKIN’S DISEASE

Factor Age greater or lesser than 40 years Histologic subgroup, particularly nodular sclerosis Disease extent at diagnosis (stage) Systemic “B” symptoms Extensive disease: mediastinal adenopathy :thoracic ratio > 0.33; abdominal adenopathy > 5 cm; multiple extranodal “E” sites; extensive spleen involvement; multiple visceral involvement Prior chemotherapy Complete pathologic remission Relapse-free survival > 2-3 years

Clinical relevance

++

++

++ +++ +++

+++ +++ +++

The presence of systemic symptoms remains strictly correlated with an unfavorable prognosis (1, 64) in patients treated with MOPP chemotherapy, either alone or combined with irradiation. Thus, systemic symptoms represent the most important single prognostic factor in practically each subset of patients. However, some results with ABVD combined with radiotherapy (9, 100) failed to indicate a difference in the CR rate, RFS, and total survival between patients with “A” and “B” symptoms. Therefore, also in this prognostic subgroup ABVD chemotherapy should be considered in the design of future strategies. Studies (96) have identified several new unfavorable prognostic factors that seem to acquire a growing importance in treatment planning. These are (a) bulky disease either in the mediastinum (masdthoracic ratio >0.33) or in the abdomen (mass > cm 5; extensive spleen involvement >5 nodules); and (b) multiple extranodal (E) involvement (Table VI). B. MEDIASTINALBULKYDISEASE The presence of large mediastinal involvement (greater than onethird of the chest diameter) in patients with pathologically staged supradiaphragmatic Hodgkin’s disease has been associated with an increased risk of relapse as compared to patients with lesser or no mediastinal disease. About 50-74% of such patients initially treated with radiation therapy alone relapse compared to 5-27% with small mediastinal masses (40,53,73,76,77,79,87,111, 117).The majority of

280


relapses are intrathoracic, as recurrences within the treated volume, recurrences at the margin of the treated field, and diffuse pulmonary relapses. On this clear evidence, several institutions are evaluating the role of chemotherapy (MOPP) associated with radiotherapy. The preliminary results obtained with chemotherapy plus radiotherapy strongly suggest the use of a combined modality approach for this subset with poor prognosis. Another controversial point is the sequence of administration of chemotherapy and radiotherapy. The increased risk of complications, especially pericarditis and pneumonitis, because of the large radiation fields needed if radiotherapy is initially utilized, suggests that chemotherapy should precede irradiation to secure bulk reduction and to facilitate subsequent radical radiotherapy (79, 100, 117).The feasibility and utility of low-dose lung irradiation as an initial part of the supradiaphragmatic field was examined by Lee et al. (74).Fifteen patients were treated with 10-20 Gy to the lung as part of the extended-field radiotherapy, and results were compared with those in 20 similar patients who were treated only with total nodal radiotherapy. With a minimum follow-up of 24 months, only 13% of patients who received lung irradiation have recurred (only one in the lung) compared to 79% who were treated without lung irradiation, nine of whom relapsed in the intrathoracic region. Thus, lung irradiation appears feasible and should be further evaluated with adjuvant chemotherapy in the treatment strategy of patients with Hodgkin’s disease and large mediastinal masses. C. LIMITEDEXTRANODAL DISEASE

The Ann Arbor classification (108) assumes that the prognosis in patients with localized extranodal (E) disease is similar to that of comparable patients with the same stage disease without extranodal spread. The introduction of the “E” subgrouping has had an important influence on subsequent therapeutic approaches employed by many centers for these patients. The implication has been that patients with E stage could be treated adequately with similar radiation techniques used for comparable patients with nodal disease (108). Confusion may arise, however, with respect to the E classification (86).The excessive use of the E stage sometimes may have led to conceptual errors in the management of patients. In fact, there is always some subjective element in deciding whether a patient should be placed in stage IV or E category. Following Musshoffs observation (108), E designation should be restricted to patients whose disease appears to be curable


28 1

with radiotherapy alone. Levi and Wiernik (75) in reporting the results of a comparative analysis of patients with stage IIA-IIIA and IIEAIIIEA treated concurrently with extended field irradiation alone or limited-field irradiation followed by MOPP, found a wide difference in the 5-year relapse rate for patients treated with radiotherapy alone (stage IIA-IIIA, 29%;IIEA-IIIEA, 82%).By contrast, in patients receiving adjuvant MOPP after irradiation the relapse rate was 6% and 14%, respectively. Most relapses were observed within the lung parenchyma. Statistical analysis revealed a significantly shorter remission duration and survival for patients with E stage after irradiation alone, but there was no difference between the two treatment groups treated with RT plus MOPP. The authors concluded that the use of extendedfield irradiation alone has been inadequate for patients with E Hodgkin’s disease and suggested a combined approach for these patients. The Stanford group (113) analyzed patients with E disease and failed to confirm the results of Levi and Wiernik. The difference in results obtained by the two groups could be related in part to different patient selection, and maybe Levi and Wiernik classified as E stage some patients who actually had stage IV disease (86).Nevertheless, the optimal treatment strategy for stage E Hodgkin’s disease remains controversial, and definitive conclusions cannot as yet be drawn.

D. STAGEIIIA The optimal treatment strategy for PS IIIA with or without splenic involvement remains controversial. Total nodal irradiation has resulted in a definite improvement in both RFS and survival compared to less aggressive forms of radiotherapy (64).Nevertheless, a significant proportion, ranging from 30 to 65% of patients treated with aggressive irradiation alone, continue to relapse, especially in extranodal sites (47,90).Mauch et al. (78) utilizing total nodal irradiation alone in stage IIIA reported RFS and overall survival rates that were inferior to those obtained in stage IIIB with the same irradiation followed b y MOPP, and they suggested the use of a combined therapy program in stage IIIA. In addition, as mentioned before, the updated results from Stanford (96) confirmed a significant improvement in RFS and probably also in total survival for patients receiving total lymphoid radiotherapy plus six cycles of MOP(P) compared to those treated with irradiation alone. However, some of the treatment complications, such as sterility and second neoplasms, following combined treatment may suggest caution before adopting intensive radiotherapy followed b y

282

CIANNI BONADONNA AND ARMAND0 SANTORO

adjuvant MOPP for all patients with stage IIIA. The use of primary irradiation alone may also be justified because of the high salvage rate (about 50%) with MOPP chemotherapy administered at the time of first relapse after radiotherapy. However, the observation of Canellos et al. (15) and Valagussaet al. (115,116),that patients receiving MOPP chemotherapy as salvage treatment for relapse after extensive radiotherapy have a considerably higher risk of developing second acute leukemia compared to patients receiving extensive radiotherapy and MOPP as first-line treatment, would contraindicate the aforementioned strategy. The use of MOPP chemotherapy alone in the treatment of IIIA has also been investigated. However, with the exception of the NCI study (31), the results of chemotherapy alone seem to be inferior to those obtained with radiotherapy alone (13). New information concerning optimal treatment strategy has been provided by research in staging. In fact, specific patterns of intraabdominal involvement were reported to correlate with response to therapy, patterns of relapse and survival. As reported by the Chicago group (25, 50), detailed surgical staging allowed subdivision of patients with PS I11 into two “anatomic substages”: PS 1111(involvement limited to those lymphatic structures in the upper abdomen that accompany the celiac-axis group of arteries, i.e., spleen, splenic hilar nodes, celiac nodes, and/or portal nodes); and PS IIIz (involvement of lower abdominal nodes, i.e., para-aortic, iliac, or mesenteric nodes with or without involvement of nodes belonging to 1111groups). The 5-year RFS and total survival rates after total nodal irradiation were definitely higher for substage 111, than for substage II12. The findings were improved by the addition of chemotherapy only in PS IIL, not in 1111. The large number of PS IIIA patients studied gave strong support to the contention that the anatomic substage is the major prognostic indicator for patients in PS IIIA. However, the different prognosis of patients with PS 1111 vs 1112 was not confirmed b y the Stanford series (54). Analysis of a large number of potential prognostic factors among patients with PS IIIA treated at Stanford showed that the most important adverse prognostic factors were extensive splenic involvement (five or more nodules visible on the cut section of the spleen), bulky abdominal masses ( 2 5 cm), and presence of five or more sites of involvement. Thus, PS IIIA Hodgkin’s disease appears to include several groups with different prognostic factors. Probably, the optimal management remains to be further elucidated and alternating chemotherapy with non-cross-resistant regimens in conjunction with moderate doses of radiotherapy appears to be worth testing.


283

VI. Morbidity Influencing Current Strategy

A. MORBIDITYFROM SURGICAL STAGING During the 1960s and 1970s, various efforts were made to stage Hodgkin’s disease properly. The major intent was to make proper decisions regarding therapy by improving the selection between patients suitable for curative radiotherapy and those who are candidates for a systemic treatment program with or without radiotherapy. Over the years, lymphography, needle marrow biopsy, laparotomy, and laparoscopy have become important steps in the clinical accuracy of staging both in adults and children. The systemic use of available procedures to determine the extent of disease led first to the Rye (1965)and then to the Ann Arbor (1970) international staging classification for Hodgkin’s disease (107, 108). A major advance was the introduction (46, 66) of staging laparotomy, which has provided an unparalleled contribution in staging accuracy and knowledge of the natural history of disease. Today, the role of laparotomy is under critical reevaluation as a routine staging procedure. It should be performed only if management decisions depend on the identification of occult abdominal lymphoma, particularly of a positive spleen. Since a combined treatment approach is presently being applied for many patients with intermediate stages of Hodgkin’s disease, laparotomy is becoming less important as a routine staging procedure. To detect patients with stage IV disease in the liver, laparoscopy with multiple hepatic biopsies can substitute for laparotomy in the large majority of patients (5, 6, 28). Staging laparotomy remains at present a necessary procedure in clinical stage IA and IIA with no bulky mediastinal mass, as the 5- and 10-year survival rates of patients without occult disease below the diaphragm and treated with subtotal or total nodal irradiation alone approach 90%. As in adults, staging laparotomy in children with Hodgkin’s disease modifies the clinical stage in a considerable number of patients, particularly because in the pediatric age group there is a high incidence of occult splenic involvement (64).Complications from surgical staging are not high, but they are definitely less after laparoscopy compared to laparotomy. In the sequential laparoscopy-laparotomy study carried out in Milan on 146 consecutive patients (6), laparoscopy was associated with less morbidity (3%)compared to staging laparotomy (35%).The morbidity from laparotomy is of particular importance in children under 5 years of age because of the specific hazard of overwhelming postsplenectomy infection (57). This complication as well

284

GIANNI RONADONNA AND ARMANDO SANTORO

as the need of a combined modality approach in children to decrease radiation morbidity, has led the large majority of pediatric oncologists to abandon routine staging laparotomy in children, particularly in those under the age of 5. Current results with combined modality therapy in children not subjected to laparotomy has yielded a 5-year survival in the range of 90% (56). B. MORBIDITYFROM RADIOTHERAPY

The most important complications from radical radiation therapy (18, 110) that can influence the current treatment strategy in adult patients with Hodgkin’s disease are (a) radiation pneumonitis and fibrosis; (b)radiation pericarditis and carditis; (c) radiation nephritis; (d) radiation myelitis; (e) prolonged myelosuppression; (fj sterility. The aforementioned complications are primarily due to high-dose, large treatment volume techniques, and they are more likely to be expressed with the increased number of long survivors. Therefore, prevention is the best treatment and may be accomplished by appropriate shielding of vital organs and/or by keeping the dose below the risk threshold. I n particular, the incidence of symptomatic radiation pneumonitis (1530%) can be decreased in patients with massive mediastinal and/or hilar involvement by starting treatment with one to three courses of combination chemotherapy, which can substantially reduce the irradiated volume, and, in complete responders, also the dose of radiotherapy. In patients with symptomatic pneumonitis, steroid therapy is indicated. The radiation-induced heart disease, particularly pericarditis (13-15%), depends on the total dose, dose fraction, relative weighting of the anterior to posterior, volume of the heart included, and whether one or two fields are treated each day (110).Since it has been suggested that a lower dose, in the order of 30-35 Gy, should suffice for the eradication of microscopic disease in apparently uninvolved lymph node chains (64), a decreased morbidity and risk of late complications should be observed by delivering the aforementioned dosage to the mediastinal area in the absence of radiological or radioisotopic signs of Hodgkin’s disease. It should be recalled that acute steroid withdrawal, either during mantle irradiation or during the first and fourth courses of MOPP, may activate occult radiationinduced heart disease (64). Although after the administration of ABVD in previously irradiated patients, no episodes of symptomatic heart disease were documented (8), it is important to remember that there is at least an additive, if not synergistic, effect between conventional


285

doses of radiotherapy and cumulative doses of adriamycin exceeding 400 mg/m’. To limit radiation nephritis, appropriate shielding of a portion of the kidneys and keeping the dose to both organs below 24 Gy, particularly in the presence of negative lymphography and laparotomy, usually minimizes or prevents renal damage. Similar technical precautions are applied to limit radiation myelitis and sterility in man (110). Attempts to preserve ovarian function were initiated at Stanford in 1964 (64) through oophoropexy at the time of staging laparotomy. The procedure gained popularity, since many patients have been noted to retain menstrual ability and became pregnant with the delivery of normal children. The irradiation of pelvic nodes is often followed by prolonged myelosuppression, especially thrombocytopenia, and this side effect markedly increases in elderly patients as well as in those subjected to combined chemotherapy and radiotherapy. For this reason, many radiation therapists altered the treatment strategy in patients with pathologic supradiaphragmatic IA and IIA after the report of Goodman et al. (51). These investigators noticed that, utilizing subtotal nodal rather than total nodal irradiation, the failure rate in pelvic lymph nodes was less than 10%. Subtotal nodal irradiation is also often applied in a combined modality setting when only the para-aortic nodes appear to b e involved b y Hodgkin’s disease. In children, extensive high-dose radiation therapy has produced a number of delayed complications, namely growth retardation, particularly in the shoulders and clavicles following mantle field irradiation. Growth retardation is manifested primarily as decreased sitting height and is most severe in children treated when less than 6 years of age or when they are 12-13 years of age (64, 85). To overcome this type of morbidity, reduction in field extension as well as in the dose and combining low-dose radiotherapy with adjuvant chemotherapy (MOPP or ABVD) has represented the most important evolution in the treatment strategy for young children. Current results (32, 37, 56, 64) have indicated that in both clinical and pathologic stages 1-111 the relapse rate was low with bimodal treatment and was not associated to important acute and delayed side effects from radiotherapy and chemotherapy. c . LATE COMPLICATIONS FROM CHEMOTHERAPY The toxic effects of chemotherapy that are relevant to the treatment strategy of Hodgkin’s disease are as follows: (a) cardiotoxicity; (b) lung fibrosis; (c) infertility; (d) second neoplasms. Cardiotoxicity may be

286

GIANNI BONADONNA AND ARMANDO SANTOHO

secondary to adriamycin therapy, and the risk becomes very high when

the cumulative dose exceeds 550 mg/mz. Patients with prior radiotherapy to the mediastinurn may manifest congestive heart failure after cumulative doses higher than 400-450 mg/mz. Therefore, all adriamycin-containing regimens should be administered with caution. The Milan experience with ABVD has been so far successful also from the toxicologic point of view, for no patient out of the 400 treated so far with this regimen has exhibited signs and symptoms compatible with cardiomyopathy. It should be pointed out that in all studies the cumulative dose of adriamycin has almost never exceeded 350 mg/m' (8). However, a longer follow-up analysis is required to define better the risk of heart damage following the administration of adriamycin with and without irradiation. Pulmonary toxicity may be observed after treatment with bleomycin and BCNU or CCNU. Also from this point of view, ABVD appears to be a safe combination (8) provided the total dose of bleomycin does not exceed 200-250 mg/m' and its administration is withheld in patients with chronic lung disease or overt pulmonary postirradiation fibrosis. Male infertility and second neoplasms are of particular concern, especially because many patients with Hodgkin's disease are young and the potential for cure is high. The NCI group (101, 102) has noted a high incidence of male infertility following MOPP chemotherapy, and this has been attributed to the toxic effect on spermatogenesis by alkylating agents and procarbazine. However, in 25-40% of patients treated with MOPP or MOPP-like combinations, spermatogenesis may return after about 2 years from completion of treatment. The Milan group (8, 100) has reported that azoospermia occurred in 100% of patients treated with MOPP and in only 15%of patients given ABVD. A similar difference was also noticed in the comparative incidence of prolonged amenorrhea. These observations are important, also in that the therapeutic activity of ABVD appears to be equivalent to that of MOPP, at least in the randomized studies so far carried out in Milan (8). Thus, treatment strategy for young patients who desire to have children could be modified by selecting ABVD rather than MOPP. Another alternative could be a more careful analysis of toxicological effects of MOPP alternated with ABVD (98). By administering MOPP every 2 months, and therefore reducing the total dose of mechlorethamine and procarbazine, it is conceivable that their action on reproductive organs will be decreased. Second neoplasms, particularly acute nonlymphocytic leukemia and non-Hodgkin's lymphomas, are now being reported with more frequency than in the past in patients subjected to intensive prolonged chemotherapy with MOPP or

287

TREATMENT EVOLUTION I N HODGKIN'S DISEASE

CASE

SERIES

TABLE VII SHOWING THE RISK OF ACUTE LEUKEMIA AFTER DIFFERENT, TYPESOF THERAPY FOR HODGKIN'S DISEASE

Author Baccarani et al. (3)

Coleman et al. (20)

Valagussa et ul. (1 15)

a

Treatment

Patients

Risk (%)

Time (years)

Radiotherapy Chemotherapy" Combined" Radiotherapy Chemotherapy" Radiotherapy-gold t chemotherapy" Combined" Adjuvant Salvage Radiotherapy Cornbined" Adjuvant Salvage Combined Radiotherapy + ABVD

117 152 344 417 37

0 2.0 2.04 0 17.6

7 7 7 9 9

65

7.7

9

453 124 272

5.1 2.8 0

9 9 10

430 176

1.5 6.1

10 10

0

10

84

All regimens including alkylating agents and/or procarbazine.

MOPP-derived regimens (3, 14, 15, 19-21, 112, 115, 116). Within 10 years from diagnosis of Hodgkin's disease, the risk was considerable for patients treated with combined extensive radiotherapy -chemotherapy and was even higher, at least in some published series (15,115, 116), after salvage chemotherapy (Table VII). In contrast, the updated results with ABVD (115) confirmed that this regimen, either when administered alone or when combined with extensive irradiation, appeared to be devoid of carcinogenic activity. Also this observation may change the future therapeutic strategy, and the practical considerations on this subject are identical to those made about infertility. VII. Conclusions: Toward the Total Conquest of Hodgkin's Disease

The evolution of treatment strategy for Hodgkin's disease indicates that improvement in results is being achieved in all stages at major research centers. Current achievements are based on solid advances and represent the results of brilliant leadership in the rational development of therapeutic programs. The fact that today no stage of the disease is beyond cure after the initial treatment program, and even

288


after recurrence following primary irradiation, represents a dramatic improvement in the overall prognosis of a disease that only about 20 years ago was considered to be almost universally fatal. Future strategies should refine available treatment programs, because given subgroups with unfavorable prognostic signs will probably require a more aggressive approach whereas in other subgroups the risk of treatment complications should be minimized. First of all, since the reason why some patients with widespread lymphoma are not cured by drugs is due to primary cell resistance, more attention should be given to the potential of alternating treatments. At present the only approach that we can see to increase the cure rate of advanced Hodgkin’s disease is by cyclical (or sequential) use of non-cross-resistant combinations in the manner now being investigated in the MOPP-ABVD (8,98) and the MOPP-SCAB (26, 120) protocols. By this we do not mean that these particular combinations, or the ratios of doses in them, or the timing of the treatment in each cycle is the optimum, but the effort to minimize failures due to the overgrowth of MOPP-resistant tumor cells seems a step in the right direction. Thus, in the presence of stages IIIB and IV or at relapse following primary irradiation, MOPP monthly alternated with ABVD (minimum six cycles to achieve CR and then two additional cycles as consolidation therapy) appears to be the treatment of choice. Further to improve present results, the research group in Milan is now exploring the therapeutic effect of three non-cross-resistant combinations, i.e ., MOPP, ABVD, and C E P (CCNU, VP-16, and prednimustine). In fact, in a pilot study being carried out in patients resistant to both MOPP and ABVD, CEP was able to induce objective tumor response in about 60% and CR in about 40% at the expense of minimal toxicity. If ongoing trials continue to show an advantage of cyclical delivery of two or three non-cross-resistant combination of drugs, then the Hodgkin’s disease model will have taught us another important lesson, one that should be taken into consideration in future design of treatment for other neoplastic diseases that are highly responsive, moderately responsive, and not so responsive to available chemotherapy (103).The unanswered question in stages IIIB and IV is how to use irradiation, and potentially improve the results, without compromising the chemotherapy program in these patients. In other words, the relative merits of a single drug combination plus low-dose radiotherapy versus multiple non-cross-resistant combinations with and without irradiation remains to be clearly demonstrated. The greatest disagreement in terms of optimal treatment strategy concerns stage IIIA as to whether irradiation should be the primary


289

treatment, or whether chemotherapy should be added to total nodal irradiation, or even whether combination chemotherapy should be used without irradiation. Considering the cure rate after primary radiation therapy (64) and the most consistent finding of a survival benefit for combined modality therapy for patients with PS IIIA (96, loo), we believe the new clinical studies should evaluate combined therapy utilizing non-cross-resistant combinations. Treatment should start with chemotherapy and should be alternated with irradiation. Most probably, by making drug therapy more effective than MOPP, also the survival benefit could become more evident than was observed in the past experience. The same strategic approach should be explored further in patients with PS IIA-IIB with massive mediastinal involvement and with lymphocytic depletion histology, as well as in young children, to limit excessive morbidity from high-dose radiotherapy. As far as PS IA and IIA is concerned, surgical staging with laparotomy followed by extensive-field radiotherapy (probably subtotal nodal irradiation) represents the established treatment method. In fact, this strategic approach is followed by a high cure rate after about 2 months of treatment and a moderate morbidity if treatment is carried out b y experienced radiation therapists. Furthermore, most patients showing relapse can now be saved with effective chemotherapy. Nearly one-third of the patients with Hodgkin’s disease die without evidence of lymphoma at autopsy. Infection, mainly bacterial, remains the most common cause of death, but a significant number of patients die of complications of therapy, both benign and malignant, including patients with hematologic or de novo lymphoid neoplasms. Therefore, new treatment programs should give more attention to the costs and morbidity of primary and salvage therapies. Most of the problems, particularly those related to prolonged aggressive chemotherapy (change in physical appearance, fear of treatment programs, loss of libido, dislocation from home, etc.) are difficult to express as actuarial curves and p values, but should become important considerations in the design and selection of new treatment programs (96). In conclusion, the total conquest of Hodgkin’s disease does not appear to be a too distant goal. To achieve this goal new treatment studies are needed for high-risk groups as well as more consideration to overt and relatively occult treatment morbidity. To accomplish this program, patients with Hodgkin’s disease should continue to b e referred to major cancer research centers where efforts in accurate diagnosis, proper staging, discipline of controlled trials, and identification of complications will remain the essential ingredients of treatment approach. The laudable goal of total conquest of Hodgkin’s disease

290


will probably not be represented b y the delivery of combination chemotherapy to all patients in private offices or in community medical centers. Rather, it will be the judicious balance of refined diagnostic, predictive, and therapeutic methods for the various prognostic subsets.

REFERENCES Aisenberg, A. C., Linggood, R. M., and Lew, R. A. (1979).Am.J.Med. 67,921-928. Andrieu, J. M . , Montagnon, B., Asselain, B. et al. (1980). Cancer 46, 2126-2130. Baccarani, M., Bosi, A., and Papa, G. (1980). Cancer 46, 1735-1740. Banfi, A., Bonadonna, G., Carnevali, G., et al. (1968). Eur. J . Cancer 4, 319-324. Beretta, G., Spinelli, P., Rilke, F., et al. (1976).Cancer Treat. Rep. 60, 1231-1237. Bonadonna, G., Beretta, G., Castellani, R., et al. (1977). In “Recent Advances in Cancer Treatment” (H. J. Tagnon and M. J. Staquet, eds.), pp. 55-67. Raven, New York. 7. Bonadonna, G., and Santoro, A. (1979).“Current Diagnosis and Treatment of Malignant Lymphomas,” Cancer Clin. ser. Bristol-Myers Company International Division, New York. 8. Bonadonna, G., and Santoro, A. (1982).Cancer Treat Rev. (in press). 9. Bonadonna, G., Santoro, A., Zucali, R., et al. (1979).Cancer Clin. Trials 2,217-226. 10. Bonadonna, G., Uslenghi, C., and Zucali, R. (1975). Eur. J . Cancer 11,251-266. 11. Bonadonna, G., Zucali, R., De Lena, M., et al. (1977). Cancer Treat. Rep. 61, 769-777. 12. Bonadonna, G., Zucali, R., Monfardini, S., et al. (1975). Cancer 36, 252-259. 13. British National Lymphoma Investigation (1976). Lancet 2, 991-995. 14. Brody, R. S., and Schottenfeld, D. (1980). Semin. Oncol. 7, 187-201. 15. Canellos, G. P., De Vita, V. T., Arseneau, J. C., et al. (1975). Lancet 1, 947-949. 16. Canellos, G. P., Young, R. C., and De Vita, V. T. (1972).Clin. Pharmacol. Ther. 13, 750- 754. 17. Carbone, P. P., Kaplan, H. S., Musshoff, K., et al. (1971).Cancer Res. 31, 18601861. 18. Carmel, R. J., and Kaplan, H. S. (1976). Cancer 37, 2813-2825. 19. Coleman, N. C., Williams, C. J., Flint, A., et al. (1977). N . Engl. J . Med. 297, 1249-1252. 20. Coleman, N. C., Burke, J. S., Varghese, A., et a l . (1982). In “Malignant Lymphomas: Etiology, Immunology, Pathology, Treatment” (H. S. Kaplan and S. A. Rosenberg, eds.), Vol. 3. Academic Press, New York (in press). 21. Coltman, C. A., Jr. (1980). Semin. Oncol. 7,155-173. 22. Coltman, C. A., Jr., Hall, W., Montague, E., et al. (1977).In “Adjuvant Therapy of Cancer” (S. E. Salmon and S. E. Jones, eds.), pp. 529-536. North-Holland Publ., Amsterdam. 23. Coltman, C. A., Jr., Myers, J. W., Montague, E., e t al. (1982).In “Malignant Lymphomas: Etiology, Immunology, Pathology, Treatment” (H. S. Kaplan and S. A. Rosenberg, eds.), Vol. 3. Academic Press, New York (in press). 24. Craft, C. B. (1940). Bull. StaflMeet. Hosp. Unio. Minn. 11,391-409. 25. Desser, R. K., Golomb, H. M., Ultmann, J. E., e t al. (1977). Blood 49,883-893. 26. De Vita, V. T., Jr. (1979). Int. J. Radiat. Oncol., Biol. Phys. 5, 1855-1867. 27. De Vita, V. T., Jr. (1981). Cancer 47, 1-13.

1. 2. 3. 4. 5. 6.


29 1

28. De Vita, V. T., Bagley, C. M., Goodell, B., et al. (1971).Cancer Res. 31,1746-1750. 29. De Vita, V. T., Lewis, B. J., Rozencweig, M., et al. (1978).Cancer 42, 979-990. 30. De Vita, V. T., Jr., Serpick, A. A., and Carbone, P. P. (1970).Ann. Intern. Med. 73, 881-895. 31. De Vita, V. T., Jr., Simon, R. M., Hubbard, S. M . , e t al. (1980).Ann.Intern. Med. 92, 587-595. 32. Donaldson, S. S., Glatstein, E., Rosenberg, S. A,, e t al. (1976).Cancer 37, 24362447. 33. Easson, E. C. (1966).Cancer 19,345-350. 34. Easson, E. C., and Russel, M. H. (1963).Br. Med. J. 1, 1704-1707. 35. Farber, L. R., Prosnitz, L. R., Cadman, E. C., et al. (1980).Cancer 46,1509-1517. 36. Fisher, R. I., De Vita, V. T., Hubbard, S. P., et al. (1979).Ann. Intern. Med. 90, 761-763. 37. Fossati-Bellani, F., Musumeci, R., Doci, R., et al. (1981).Proc. Int. Conf. Malignant Lymphomas, September 5-12, 1981, Lugano. 38. Frei, E., 111, D e Vita, V. T., Moxley, J. H., 111, et al. (1966). Cancer Res. 26, 1284-1289. 39. Frei, E., 111, Luce, J. K., Gamble, J. F., et al. (1973).Ann.Intern. Med. 79,376-382. 40. Fuller, L. M., Madoc-Jones, H., Hagemeister, F. B., et al. (1980).I n t . J . Radiat. Oncol., Biol. Phys. 6, 799-808. 41. Gilbert, R. (1925).J. Radiol. Electrol. 9, 509-514. 42. Gilbert, R. (1939).A m . J. Roentgenol. Radium Ther. 41, 198-241. 43. Gilbert, R., and Babaiantz, L. (1931).Actu Radiol. 12, 523-529. 44. Gilman, A., and Philips, F. S. (1946).Science 103,409-415. 45. Glatstein, E. (1977).Cancer 39, 834-842. 46. Glatstein, E., Guernsey, J. M., Rosenberg, S. A., e t al. (1969).Cancer 24,709-718. 47. Click, J. H. (1978).Int. J . Radiat. Oncol., Biol. Phys. 4, 909-911. 48. Goldie, J. H., and Coldman, A. J. (1979).Cancer Treat. Rep. 63, 1727-1733. 49. Goldsmith, M. A., and Carter, S. K. (1974).Cancer 33, 1-8. 50. Golomb, H. M., Sweet, D. L., Ultmann, J. E., et al. (1980). Semin. Oncol. 7, 136-143. 51. Goodman, R. L., Piro, A. J., and Hellman, S . (1976).Cancer 37,2834-2839. 52. Hoppe, R. T. (1980).Semin. Oncol. 7, 144-154. 53. Hoppe, R. T., Coleman, C. N., Kaplan, H. S., et a1. (1980).Proc. Am. SOC. Clin. Oncol. 21, 471 (abstr.). 54. Hoppe, R. T., Rosenberg, S. A., Kaplan, H. S., et al. (1980).Cancer 46, 1240-1246. 55. International Symposium on Hodgkin’s Disease (1973).Natl. Cancer Inst. Monogr. 36, 1-581. 56. Jenkin, D., Freedman, M., McClure, P., et al. (1979).Cancer 44, 80-86. 57. Jenkin, R. D., and Berry, M. P. (1980).Semin. Oncol. 7, 202-211. 58. Kaplan, H. S. (1962).Radiology 78, 553-561. 59. Kaplan, H. S. (1966).Cancer Res. 26, 1221-1224. 60. Kaplan, H. S. (1968).Cancer 22, 1131-1136. 61. Kaplan, H. S. (1970).Haroey Lect. 64, 215-259. 62. Kaplan, H. S. (1976).Cancer Res. 36, 3863-3878. 63. Kaplan, H. S. (1979).Semin. Oncol. 6, 479-489. 64. Kaplan, H. S. (1980).“Hodgkin’s Disease.” Harvard Univ. Press, Cambridge, Massachusetts. 65. Kaplan, H. S. (1980).Cancer 45, 2439-2474. 66. Kaplan, H. S., Dorfman, R. F., Nelsen, T. S., et al. (1973). Natl. Cancer Inst. Monogr. 36,291-301.

292

GIANNI BONADONNA A N D AHMANDO SANTOHO

Kaplan, H. S., Goodenow, R. S., and Gartner, S. (1979). Cancer 43, 1-24. Kaplan, H. S., and Rosenberg, S. A. (1966). Cancer Res. 26, 1268-1276. Kaye, S. B., Juttner, C. A,, Smith, I. E., et al. (1979). B r . ] . Cancer 39, 168-174. Kinmonth, J. D. (1952). Clin. Sci. 11, 13-20. Kuhn, L. E., De Vita, V. T., Young, R. C., et al. (1976).Znt. J. Radiat. Oncol., Biol. Phys. 1,619-626. 72. Lacher, M. J., and Durant, J. R. (1965).Ann. Intern. Med. 62, 468-476. 73. Lee, C. K. K., Bloomfield, C. D., Goldman, A. I., et al. (1980).Cancer 46, 24032409. 74. Lee, C. K. K., Bloomfield, C. D., Goldman, A. I., et al. (1981).Znt.J. Radiat. Oncol., Biol. Phys. 7, 151-154. 75. Levi, J. A., and Wiernik, P. H. (1977).Am. J. Med. 63,365-372. 76. Levi, J. A., Wiemik, P. H., and O’Connel, M. (1977). Znt. J. Radiat. Oncol., Biol. Phys. 2,853-862. 77. Mauch, P., Goodman, R., and Hellman, S. (1978). Cancer 42, 1039-1045. 78. Mauch, P., Goodman, R., Rosenthal, D. S., et al. (1979). Cancer 43, 1255-1261. 79. Mauch, P., and Hellman, S. (1980).Znt. J. Radiat. Oncol., Biol. Phys. 6,947-949. 80. Nissen, N. I., Pajak, T. F., Glidewell, O., et al. (1979).Cancer 43, 31-40. 81. Pene, F., Henry-Amar, M., Le Bourgeois, J. P.,et al. (1980).Cancer 46,2131-2141. 82. Peters, M. V. (1950). Am. J. Roentgenol. Radium Ther. 63, 299-311. 83. Peters, M. V. (1966). Cancer Res. 26, 1232-1243. 84. Peters, M. V., and Middlemiss, K. C. H. (1958).Am. J. Roentgenol., Radium Ther. Nucl. Med. 79, 114-121. 85. Probert, J. C., and Parker, B. R. (1975).Radiology 114, 155-162. 86. Prosnitz, L. R. (1977). Znt. J. Radiat. Oncol., Biol. Phys. 2, 1039. 87. Prosnitz, L. R., Curtis, A. M., Knowlton, A. H., et 01. (1980).Znt. J . Radiat. Oncol., Biol. Phys. 6, 809-813. 88. Prosnitz, L. R., Farber, L. R., Fischer, J. J., et al. (1973). Radiology 107, 187-193. 89. Prosnitz, L. R., Farber, L. R., Fischer, J. J., et al. (1976).Cancer 37,2826-2837. 90. Prosnitz, L. R., Montalvo, R. L., Fischer, D. B., et al. (1978).Znt. J. Radiat. Oncol., Biol. Phys. 4, 781-787. 91. Rhoads, C. P. (1946).JAMA,J. Am. Med. Assoc. 131,656-658. 92. Rosenberg, S. A. (1971).Cancer Res. 31, 1737-1740. 93. Rosenberg, S. A., and Kaplan, H. S. (1966). Cancer Res. 26, 1225-1231. 94. Rosenberg, S. A., and Kaplan, H. S. (1975). Cancer 35, 55-63. 95. Rosenberg, S. A., Kaplan, H. S., Glatstein, E. J.,etal. (1978).Cancer 42,991-1000. 96. Rosenberg, S. A,, Kaplan, H. S., Hoppe, R. T., et al. (1981).In “Adjuvant Therapy of Cancer 111” (S. E. Jones and S. E. Salmon, eds.), pp. 65-761. Grune & Stratton, New York. 97. Santoro, A., and Bonadonna, G. (1979).Cancer Chemother. Pharmacol. 2,101-105. 98. Santoro, A., Bonadonna, G., Bonfante, V., et al. (1982).N . Engl. J. Med. (in press). 99. Santoro, A.,Bonfante,V.,and Bonadonna,G.(1982).Ann. Intern. Med. Febr. Issue. 100. Santoro, A., Bonfante, V., Bonadonna, G., et al. (1981). In “Adjuvant Therapy of Cancer 111” (S. E. Jones and S. E. Salmon, eds.), pp. 85-91. Grune & Stratton, New York. 101. Sherins, R. J., and De Vita, V. T. (1973).Ann. Intern. Med. 79,216-220. 102. Sherins, R. J., Olweny, C. L. M., and Ziegler, J. L. (1978). N . Engl. J. Med. 299, 12- 16. 103. Skipper, H. E. (1980). “On the Remarkable Progress That Has Been Made in Treatment of Hodgkin’s Disease, Booklet 13. Southern Research Institute, Birmingham, Alabama. 67. 68. 69. 70. 71.

TREATMENT EVOLUTION IN HODGKIN’S DISEASE

293

104. Skipper, H. E., Schabel, F. N., Jr., and Wilcox, W. S. (1964).Cancer Chemother. Rep. 35,3-111. 105. Straus, D . J., Myers, J., Passe, J., et al. (1980).Cancer 46, 233-240. 106. Symposium International sur “La radiothbrapie de la maladie de Hodgkin” (1966). N o w . Reu. Fr. Hematol. 6, 5-176. 107. Symposium on “Obstacles to the control of Hodgkin’s disease” (1966).Cancer Res. 26,1045-1311. 108. Symposium on “Staging in Hodgkin’s disease” (1971).Cancer Res. 31,1707-1870. 109. Teillet, F., Bayle, C., Asselain, B., et al. (1981).Proc. Int. Conf. Malignant Lymphomas, September 5-12, 1981, Lugano. 110. Thar, T. L., and Million, R. R. (1980).Semin. Oncol. 7, 174-183. 111. Thar, T. L., Million, R. R.,Hausner, R. J., et al. (1979).Cancer 43, 1101-1105. 112. Toland, D. M., Coltman, C. A., Jr., and Moon, T. E. (1978).Cancer Clin. Trials 1, 27-33. 113. Torti, F. M., Portlock, C. S., Rosenberg, S. A., et al. (1981).A m . / . Med. 70,487-492. 114. Tubiana, M., Henry-Amar, M., Hayat, M., et al. (1979).Eur. J . Cancer 15,645-657. 115. Valagussa, P., Santoro, A., Fossati-Bellani, F., et al. (1981).Proc. A m . Assoc. Cancer Res. 22, 197 (abstr.). 116. Valagussa, P.,Santoro, A., Kenda, R.,et al. (1980).Br. Med. 1. 280, 216-219. 117. Velentjas, E., Barrett, A., McElwain, T. J., et al. (1980).Eur. /. Cancer 16, 10651068. 118. Weller, S. A,, Glatstein, E., Kaplan, H. S., et al. (1976).Cancer 37, 2840-2846. 119. Wiemik, P. H., Gustafson, J., Schimpff, S. C.,et al. (1979).Am.J.Med. 67,183-193. 120. Wiernik, P. H., Longo, D., Duffey, P. L., et al. (1981).Proc. A m . Assoc. Cancer Res. 22, 159 (abstr.). 121. Young, R. C., Canellos, G. P., Chabner, P. A., et al. (1973).Lancet 1, 1339-1343. 122. Young, R. C., Canellos, G. P., Chabner, B. A.,et al. (1978).Cancer 42,1001-1007. 123. Zubrod, C. G. (1979).Semin. Oncol. 6,490-505.


EPSTEIN-BARR VIRUS ANTIGENS-A CHALLENGE TO MODERN BlOCH EM I STRY

David A. Thorley-Lawson,l Clark M. Edson, and Kathi Geilinger' Sidney Farber Cancer Institute. Boston, Massachusetts

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. What Is Known about the Antigens of Epstein-Barn Virus? B. What Questions Remain To Be Answered about the Antigens o Epstein-Barr Virus? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Transformation Antigens . ........... A. Epstein-Barr Virus Nuclear Antigen (EBNA) . . . . . . . . . . . . . . . . . . . . . . . . . Surface of EBV-Transformed Cells (LYDMA) 111. Early Antigens

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

A. Intracellular Early Antigens.. . . . B. Early Membrane Antigens ........... ........... IV. Late Antigens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Introduction B. Membrane Antigens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Conclusions . . . . . . . . . . . . . A. Molecular Biology of the Antigens and the Polypeptides . . . . . . . . . . . . . . . B. Immune Responses to the Viral Antigens.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . ............ ..............

295 295 297 298 298 304 309 309 318 319 319 32 1 335 336 336 339 342

I. Introduction

A. W H A T IS KNOWN ABOUT T H E ANTIGENS OF EPSTEIN-BARR VIRUS?2

Epstein-Barr virus (EBV) has two fascinating features in its biology. The first is its ability to transform human B lymphocytes in uitro (W. Henle et al., 1967; Pope et al., 1969; Miller et al., 1969; Gerber et al., 1969).The ability of EBV to convert small, normal, resting B lymphocytes into exponentially proliferating, transformed lymphoblasts is one ofthe most dramatic events that may be witnessed in tissue culture. I Present address: Department of Pathology, and Department of Medicine, Division of Geographic Medicine, Tufts University Medical School, Boston, Massachusetts. For a comprehensive overview of Epstein-Barr virus, see Epstein and Achong (1979).


Copyright 0 1982 by Academic Press, Inc. All rights of reproduction in any form reserved. ISBN 0-12-006636-X

296

DAVID A. THORLEY-LAWSON E T AL.

The second is its association with human disease, in particular malignant disease. The central question of importance for EBV studies is derived from the synthesis of these two observations. To what extent does the ability of the virus to transform human lymphocytes play a critical role in human disease? Is EBV truly a human tumor virus? The weight of the evidence indicates that this transformation is either malignant or premalignant (Shope et ul., 1973; Giovanella et ul., 1979). This leads to the questions of what is the mechanism of transformation; what is the nature of the immune responses that control transformation in viuo; and what is different about these processes in individuals who succumb to the EBV-associated diseases-nasopharyngeal carcinoma (NPC) (Old et al., 1966; W. Henle e t al., 1970b),Burkitt’s lymphoma (BL) (Levy and Henle, 1966; G. Henle e t al., 1969; de-Tht5 et al., 1979), and fatal infectious mononucleosis (fatal I M ) (Bar et al., 1974; Purtillo et al., 1975; Virelizier et al., 1978; Crawford et al., 1979; Robinson e t al., 1980b)? The virus was discovered originally as a typical herpesvirus in cultured tissue of BL biopsies (Epstein et al., 1964) and was shown to be distinct from other human herpesviruses on a serological basis (G. Henle and Henle, 1966a,b). Until recently, the biology of EBV has been studied primarily in terms of serologically defined antigens that can be grouped into three classes depending on the stage of the virus lytic cycle.

1. Transformation Antigens EBV-transformed lymphocytes have the capacity to grow indefinitely in culture and possess three markers for the presence of EBV: (a) multiple copies of the viral genome (zur Hausen and SchulteHolthausen, 1970; zur Hausen et al., 1970; Nonoyama and Pagano, 1971, 1973); (b) expression in their nucleus of a characteristic, serologically defined antigen termed the EB nuclear antigen (EBNA) (Reedman and Klein, 1973); (c) expression, on the plasma membranes, of antigens that elicit cellular immune responses, described under the general title of lymphocyte-determined membrane antigens (LYDMA) (Svedmyr and Jondal, 1975; Misko et al., 1980; Thorley-Lawson, 1981). The transformation antigens are expressed in all EBV-infected cells so far studied. 2. Early Antigens At any one time, a small number of EBV-transformed cells will spontaneously enter a viral lytic cycle. The absolute number of these cells in a culture depends on several factors, such as (a) the cell type [cell lines derived by in vitro transformation or spontaneous outgrowth

297

EPSTEIN-BARR VIRUS ANTIGENS

from the peripheral blood of IM patients have a low number,

Advances in Cancer Research, Volume 36

Advances in Cancer Research

Advances in Cancer Research Volume 51

Advances in Cancer Research Volume 43

Advances in Cancer Research, Volume 91

Advances in Cancer Research Volume 42

Advances in Cancer Research Volume 41

Advances in Cancer Research Volume 78

Advances in Cancer Research, Volume 106

Advances in Cancer Research, Volume 84

Advances in Cancer Research, Volume 20

Advances in Cancer Research Volume 76

Advances in Cancer Research Volume 65

Advances in Cancer Research Volume 40

Advances in Cancer Research Volume 63

Advances in Cancer Research Volume 35

Advances in Cancer Research Volume 75

Advances in Cancer Research Volume 49

Advances in Cancer Research Volume 32

Advances in Cancer Research Volume 82

Advances in Cancer Research, Volume 57

Advances in Cancer Research, Volume 5

Advances in Cancer Research Volume 70

Advances in Cancer Research Volume 60

Advances in Cancer Research Volume 47

Advances in Cancer Research, Volume 30

Advances in Cancer Research Volume 52

Advances in Cancer Research Volume 54

Advances in Cancer Research Volume 99

Advances in Cancer Research, Volume 101

Advances in Cancer Research, Volume 109

Advances in Cancer Research, Volume 36