Advances in Cancer Research Volume 90

Advances in CANCER RESEARCH Volume 90 This Page Intentionally Left Blank Advances in CANCER RESEARCH Volume 90 E...

Author: George F. Vande Woude (Editor) | George Klein (Editor)

34 downloads 848 Views 3MB 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

Advances in

CANCER RESEARCH Volume 90

This Page Intentionally Left Blank

Advances in

CANCER RESEARCH Volume 90

Edited by

George F. Vande Woude Van Andel Research Institute Grand Rapids, Michigan

George Klein Department of Tumor Biology Karolinska Insitutet Stockholm, Sweeden

This book is printed on acid-free paper. Copyright ß 2003, Elsevier Inc.

All Rights Reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the Publisher. The appearance of the code at the bottom of the first page of a chapter in this book indicates the Publisher’s consent that copies of the chapter may be made for personal or internal use of specific clients. This consent is given on the condition, however, that the copier pay the stated per copy fee through the Copyright Clearance Center, Inc. (222 Rosewood Drive, Danvers, Massachusetts 01923), for copying beyond that permitted by Sections 107 or 108 of the U.S. Copyright Law. This consent does not extend to other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works, or for resale. Copy fees for pre-2003 chapters are as shown on the title pages. If no fee code appears on the title page, the copy fee is the same as for current chapters. 0065-230X/2003 $35.00 Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone: (+44) 1865 843830, fax: (+44) 1865 853333, e-mail: [email protected]. You may also complete your request on-line via the Elsevier homepage (http://elsevier.com), by selecting ‘‘Customer Support’’ and then ‘‘Obtaining Permissions.’’

Elsevier Academic Press 525 B Street, Suite 1900, San Diego, California 92101-4495, USA 84 Theobald’s Road, London WC1X 8RR, UK http://www.academicpress.com International Standard Book Number: 0-12-006690-4 PRINTED IN THE UNITED STATES OF AMERICA 03 04 05 06 07 08 9 8 7 6 5 4 3 2 1

Contents

Contributors to Volume 90 ix

Microenvironmental Regulation of the Initiated Cell Harry Rubin I. Introduction 2 II. Initiation and Promotion of Tumor Development in the Skin of Animals Painted with Carcinogens 3 III. Inhibition of Initiated Keratinocytes in Culture by Normal Keratinocytes 5 IV. Inhibition of Neoplastic Progression in Mice or in Organotypic Culture by Normal Keratinocytes 8 V. Inhibition of Virally Transformed Malignant Keratinocytes by Dermal Fibroblasts in Skin Grafts and in Subcutaneous Injections 11 VI. Restriction of Mutant Keratinocytes to the Epidermal Proliferative Unit in Ultraviolet (UV) Carcinogenesis of the Skin 13 VII. Microenvironmental Suppression of Neoplastic Transformation Among Cultured Fibroblasts 17 VIII. Evaluation of the Evidence Concerning a Role for Proteases in Tumor Initiation 35 IX. Visualization of Microenvironmental Regulation in Spontaneous Transformation of Mouse Fibroblasts 37 X. Microenvironmental Effects on Mammary Neoplasia 40 XI. Further Developments on the Role of Proteases in Carcinogenesis 48 XII. Conclusions 51 References 56

Persistent Infection with Helicobacter Pylori and the Development of Gastric Cancer Staffan Normark, Christina Nilsson, Birgitta Henriques Normark, and Mathias W. Hornef I. Introduction—H. Pylori and Cancer Development II. Bacterial Colonization and Persistence 67 III. H. Pylori–Mediated Mucosal Inflammation 71

63

v

vi

Contents

IV. Effects of H. Pylori Infection on Epithelial Cell Turnover 76 V. H. Pylori–Mediated Promotion of Tumor Development 77 VI. Host Susceptibility Genes and Gastric Cancer Associated with H. Pylori Infection 81 VII. Concluding Remarks 82 References 83

High-Resolution Analysis of Genetic Events in Cancer Cells Using Bacterial Artificial Chromosome Arrays and Comparative Genome Hybridization John K. Cowell and Norma J. Nowak I. II. III. IV. V.

Introduction 92 Evolution of Molecular Cytogenetics 94 Development of BAC Resources 96 CGHa in the Analysis of Cancer 100 Summary 122 References 123

Natural Killer Cells and Cancer Jun Wu and Lewis L. Lanier I. II. III. IV. V. VI. VII.

Introduction 128 NK Cell Biology 130 NK Cell Effector Functions 136 Receptors Turning NK Cells ‘‘On’’ and ‘‘Off’’ 138 NK Cells in Tumor Immunosurveillance 145 Tumor Escape Mechanisms 146 Conclusion 148 References 148

Immunity to Cancer Through Immune Recognition of Altered Self: Studies with Melanoma Jose´ A. Guevera-Patin˜o, Mary Jo Turk, Jedd D. Wolchok, and Alan N. Houghton I. II. III. IV. V.

Introduction 158 Melanoma Antigens 159 Cellular Responses to Differentiation Antigens 160 Antibody-Mediated Tumor Immunity and Autoimmunity Clinical Applications 172 References 175

169

vii

Contents

Chemical Carcinogens as Foreign Bodies and Some Pitfalls Regarding Cancer Immune Surveillance Thomas Blankenstein and Zhihai Qin I. Introduction 180 II. Chemical Carcinogenesis 181 III. Increased Tumor Incidence by MCA in IFNR-KO Compared with IFNR-WT Mice 183 IV. No New Evidence that Supports T-Cell-Mediated Immune Surveillance of MCA-Induced or Spontaneous Tumors 185 V. The Protective Response in IFNR-WT Mice is Associated with Encapsuation of MCA 195 VI. Concluding Remarks 201 References 203

Epigenetic Theories of Cancer Initiation Lionel F. Jaffe I. II. III. IV.

Index

Introduction 210 Evidence that the Early Stages of Cancer are Epigenetic Mechanisms of Epigenesis 222 Concluding Remarks 226 References 226

231

212

This Page Intentionally Left Blank

Contributors

Numbers in parentheses indicate the pages on which the authors’ contributions begin.

Thomas Blankenstein, Max-Delbru¨ck-Centrum for Molecular Medicine, 13092 Berlin, Germany, and Institute of Immunology, Free University Berlin, 12200 Berlin, Germany (179) John K. Cowell, Roswell Park Cancer Institute, Department of Cancer Genetics, Buffalo, New York 14263 (91) Jose´ A. Guevera-Patin˜o, Memorial Sloan-Kettering Cancer Center and the Well Graduate School of Medical Sciences and Medical School of Cornell University, New York, New York 10021 (157) Mathias W. Hornef, Microbiology and Tumor Biology Center and Smittskyddinstitutet Karolinska Institutet, Stockholm, Sweden (63) Alan N. Houghton, Memorial Sloan-Kettering Cancer Center and the Well Graduate School of Medical Sciences and Medical School of Cornell University, New York, New York 10021 (157) Lionel F. Jaffe, Marine Biological Laboratory, Woods Holes, Massachusetts 02543, and OB/GYN Department, Brown University, Providence, Rhode Island 02905 (209) Lewis L. Lanier, Department of Microbiology and Immunology and the Cancer Research Institute, University of California, San Francisco, California (127) Christina Nilsson, Microbiology and Tumor Biology Center and Smittskyddinstitutet Karolinska Institutet, Stockholm, Sweden (63) Birgitta Henriques Normark, Microbiology and Tumor Biology Center and Smittskyddinstitutet Karolinska Institutet, Stockholm, Sweden (63) Staffan Normark, Microbiology and Tumor Biology Center and Smittskyddinstitutet Karolinska Institutet, Stockholm, Sweden (63) Norma J. Nowak, Roswell Park Cancer Institute, Department of Cancer Genetics, Buffalo, New York 14263 and Center of Excellence in Bioinformatics, State University of New York at Buffalo, Buffalo, New York 14203 (91) Zhihai Qin, Institute of Immunology, Free University Berlin, 12200 Berlin, Germany (179)

ix

x

Contributors

Harry Rubin, Department of Molecular and Cell Biology, Life Sciences Addition, University of California, Berkeley, California 94720 (1) Mary Jo Turk, Memorial Sloan-Kettering Cancer Center and the Well Graduate School of Medical Sciences and Medical School of Cornell University, New York, New York 10021 (157) Jedd D. Wolchok, Memorial Sloan-Kettering Cancer Center and the Well Graduate School of Medical Sciences and Medical School of Cornell University, New York, New York 10021 (157) Jun Wu, Shanghai Genomics, Inc., and Chinese National Genome Center, Shanghai, China (127)

Microenvironmental Regulation of the Initiated Cell Harry Rubin Department of Molecular and Cell Biology, Life Sciences Addition, University of California, Berkeley, CA 94720-3200

I. Introduction II. Initiation and Promotion of Tumor Development in the Skin of Animals Painted with Carcinogens III. Inhibition of Initiated Keratinocytes in Culture by Normal Keratinocytes IV. Inhibition of Neoplastic Progression in Mice or in Organotypic Culture by Normal Keratinocytes V. Inhibition of Virally Transformed Malignant Keratinocytes by Dermal Fibroblasts in Skin Grafts and in Subcutaneous Injections VI. Restriction of Mutant Keratinocytes to the Epidermal Proliferative Unit in Ultraviolet (UV) Carcinogenesis of the Skin VII. Microenvironmental Suppression of Neoplastic Transformation Among Cultured Fibroblasts A. Primary Chick Embryo Fibroblasts Infected with RSV B. Mammalian Fibroblast Inhibition of Hamster Fibroblasts Transformed by Polyoma Virus C. Influence of Normal Mammalian and Avian Fibroblasts on Proliferation of Polyoma Virus- and RSV-Infected Chicken and Mammalian Fibroblasts as well as Nonviral Mouse Sarcoma Cells D. Chemical Carcinogenesis of Mammalian Fibroblasts E. Microenvironmental Regulation of UV-Initiated, TPA-Promoted Mouse Fibroblasts VIII. Evaluation of the Evidence Concerning a Role for Proteases in Tumor Initiation IX. Visualization of Microenvironmental Regulation in Spontaneous Transformation of Mouse Fibroblasts X. Microenvironmental Effects on Mammary Neoplasia A. In Vivo Results B. In Vitro Results XI. Further Developments on the Role of Proteases in Carcinogenesis XII. Conclusions References

In the classical skin model of tumor initiation, keratinocytes treated once with carcinogen retain their normal appearance and growth behavior indefinitely unless promoted to growth into papillomas. Because many of the papillomas regress and may recur with further promotion, their cells can also be considered as initiated. The growth of initiated keratinocytes can be inhibited either in vitro or in vivo by close Advances in CANCER RESEARCH 0065-230X/03 $35.00

1

Copyright 2003, Elsevier Inc. All rights reserved

2

Harry Rubin association with an excess of normal keratinocytes, but it is enhanced by dermal fibroblasts. Chick embryo fibroblasts (CEF) in culture produce transformed foci after infection with Rous sarcoma virus (RSV) on a background of normal CEF in a medium containing 10% or less calf serum (CS), but they retain normal appearance and growth regulation in 10% fetal bovine serum (FBS) or 20% CS. Transformation of a carcinogentreated line of mouse embryo fibroblasts is prevented, and can be reversed, in high concentrations of FBS in the presence of an excess of normal cells. FBS has high, broadspectrum antiprotease activity. Increased protease production occurs in a variety of transformed cells and is correlated with progression in tumors. Protease treatment stimulates DNA synthesis and mitosis in confluent, contact-inhibited normal cell cultures. Synthetic inhibitors of proteases suppress transformation in carcinogen-treated cultures and inhibit tumor formation in animals. Several different classes of protease may be overexpressed in the same transformed cells. It is proposed that excessive protease production accounts for major features of neoplastic transformation of initiated cells, but that transformation can be held in check by protease inhibitors present in serum and released from surrounding cells. It would be informative to determine whether high concentrations of FBS would inhibit the neoplastic development of initiated keratinocytes. ß 2003 Elsevier Inc.

I. INTRODUCTION An impressive body of evidence indicates that most human tumors arise from single cells (Bedi et al., 1996; Fearon et al., 1987; Fialkow, 1976; Fujii et al., 1996; Ponte´n et al., 1997; Sidransky et al., 1992). Mice carrying a germline mutation in the APC gene, which is a homolog of the human APC gene, develop multiple intestinal adenomas. However, many of the adenomas have a polyclonal structure, even when very small (Merritt et al., 1997). Because human colorectal cancers are usually monoclonal (Fearon et al., 1987), one of the several clones that may originate the tumor presumably predominates during progression to the malignant state. The finding of monoclonality of most human tumors simply reinforces the natural tendency to concentrate analysis of tumor development on the tumor itself. This tendency was embedded in a common assumption that the tissue surrounding the tumor is normal and therefore likely to be polyclonal, which would complicate genetic analysis of the tissue. A growing number of observations lend weight to the idea that the development of tumors, particularly in their incipient stages when potential tumor cells are surrounded by apparently normal cells, is strongly influenced by those cells, which, along with humoral elements, constitute the microenvironment for tumor development. This idea is reinforced by increasing evidence that selection plays a major role in such common human tumors as cancers of the colon (Tomlinson and Bodmer, 1999; Tomlinson et al., 1996), lung (Rodin and Rodin, 2000, 2002), and skin (Zhang et al., 2001). The main purpose of this review is to evaluate the

Microenvironmental Regulation of the Initiated Cell

3

role of the microenvironment in neoplastic development in vitro and in vivo, with emphasis on the interplay of proteases and their inhibitors in the outcome. The microenvironment of cancer cells is widely assumed to consist only of the connective tissue stroma surrounding epithelial tumor cells (see Cunha and Matrisian, 2002). Little attention is usually paid to normal epithelial cells and humoral factors, which will be given serious consideration here.

II. INITIATION AND PROMOTION OF TUMOR DEVELOPMENT IN THE SKIN OF ANIMALS PAINTED WITH CARCINOGENS The systematic study of chemically induced tumorigenesis in animals had its origin in the discovery that repeated application of coal tar over many months to the skin of rabbits would produce tumors (Yamagiwa and Ichikawa, 1918). A major step forward came with the isolation from coal tar in the early 1930s of a strongly carcinogenic polycyclic aromatic hydrocarbon (PAH), now known as benzo(a)pyrene [B(a)P], and the synthesis of other carcinogenic PAHs (Cook et al., 1933, 1937). Carcinogenicity was mainly tested by repeated application to the skin of mice over a period of about 4 months. If the repeated applications were stopped at 2 months, no papillomas resulted (Hieger, 1936). The production of carcinomas required repeated applications for more than 4 months. The requirement for such long-term application indicated that something more was going on than the production of an irreversible lesion in a few cells. Valuable insight into the tumorigenic process was gained by the finding that it could be divided into two stages known as initiation and promotion. Rabbit ears were painted repeatedly with coal tar for several months until some papillomas arose (Rous and Kidd, 1941). The tarring was then stopped, and many of the papillomas disappeared over the next 8 months. When tarring was resumed, some of the papillomas reappeared, and they did so in much less time than it took for their original appearance. The disappearance and reappearance of papillomas warrants classifying their cells as initiated and dependent on their microenvironment for neoplastic expression. It was apparent that potential tumor cells had remained in a latent state for months but could be evoked to tumor production at any time. It was then found that the application of a noncarcinogen, turpentine, could bring the latent papilloma cells into proliferation and tumor formation, as could a prolonged period of epidermal proliferation induced by healing of a large wound (McKenzie and Rous, 1941). Numerous tumors appeared

4

Harry Rubin

on the healing surface of the wound on the tarred ears, but none appeared on a similar wound on the untreated control ears. A broader-scale, systematic study in mouse skin was developed independently, in which a particularly informative methodology was applied. A single subcarcinogenic painting with a PAH was used to initiate the epidermis and was followed by repeated application of a promoting agent (Berenblum and Shubik, 1947; Mottram, 1944). The early experiments used a complex plant derivative, croton oil, as the promoting agent. Once the active ingredient, TPA (12-Ø-tetradecanoyl phorbol-13-acetate) was isolated from croton oil (Hecker, 1968), it generally replaced croton oil as a promoting agent. The application of the promoting agent could begin immediately or be delayed for over 1 year after application of the initiating treatment, with no decrease in its effectiveness in producing papillomas (Van Duuren et al., 1975). Hence, it could be concluded that initiation by the PAHs took effect immediately and was a very stable change. In contrast, promotion required repeated application and was ineffective if too long an interval, for example, 1 month, separated each application (Pitot, 1986). Many of the papillomas induced by the initiation–promotion sequence regressed and rarely progressed to carcinomas, whereas those induced by repeated application of a PAH initiator regressed much less frequently and often progressed to carcinoma (Reddy and Fialkow, 1983; Shubik, 1950). The regression of mouse papillomas on cessation of promotional treatment indicates that their cells can, similar to those of rabbit papillomas, be classified as initiated. The initiation that resulted from treatment with PAHs had the characteristics of a mutation, and it was indeed later discovered that the diol epoxide metabolites of PAHs are highly mutagenic in animal cells and bacteria (Huberman et al., 1976; Newbold and Brookes, 1979; Wislocki et al., 1976; Wood et al., 1976). Because TPA is nonmutagenic in animal cells (Lankas et al., 1977) and its effects are reversible, it is believed to induce selective growth of the cells mutagenized by the initiator (Parkinson, 1985). Although initiation is a stable effect, there is no morphological evidence of altered cells until the promoter is applied. This indicates that the potential tumor cell is kept in check by its location in normal epidermis and that its tumorigenic potential is only brought out by promotion that alters its relation to the surrounding cells that make up its microenvironment. Indeed, as will be seen, there is evidence that occasional cells in some normal animal and human tissues carry endogenous genetic lesions that are only expressed as neoplasms when the regulatory capacity of the surrounding cells is disturbed. It is noteworthy that genetic abnormalities associated with cancer occur commonly in normal human tissues. For example, the use of nine highly informative microsatellite markers has revealed at least one genetic

Microenvironmental Regulation of the Initiated Cell

5

abnormality in normal breast ducts or terminal ductal-lobular units of 50% of the women examined (Larson et al., 1998). Half of the genetic markers used were in chromosomal regions commonly lost or mutated in breast cancer, but none of the tissue samples revealed any histological abnormality; the presence of such genetic lesions does not necessarily lead to cancer even many years later (Kasami et al., 1997). These results point to the need for additional genetic mutations for the development of affected clones into tumors, to a loosening of control by the regulatory microenvironment, or to both processes. Deliberate introduction of a mouse sarcoma virus into the epidermal cells of mice produced no tumors unless it was followed by TPA promotion (Brown et al., 1986). The promotion could be delayed at least 4 months, and tumors would appear within 4 weeks, which is sooner than they appear in the chemical initiation–promotion sequence. The tumors expressed viral oncogenes. The sarcoma virus induces tumors in connective tissue cells when injected subcutaneously without deliberate promotional treatment, although there is evidence that the trauma produced by the injection is a sufficient promoter for these cells (Sieweke and Bissell, 1994). It is not that sarcoma viruses cannot transform epithelial cells in general without promotion because they do so very well on the chorioallantoic membrane of the chicken embryo (Rubin, 1955). However, the epidermal cells of the adult animal are apparently more resistant to transformation by sarcoma viruses, and do require promotion.

III. INHIBITION OF INITIATED KERATINOCYTES IN CULTURE BY NORMAL KERATINOCYTES The role of the microenvironment in suppressing tumor formation was studied in vitro using the 308 line of keratinocytes derived from mouse skin initiated with 7,12-dimethylbenzanthracene (DMBA; Hennings et al., 1990). This line exhibited the initiated keratinocyte characteristic of continued proliferation in ‘‘high’’-calcium (1.2 mM) medium in which normal primary keratinocytes cease proliferation and terminally differentiate. In a valid culture model for initiated epidermis, the clonal expansion of initiated cells would most likely be inhibited by immediately adjacent normal keratinocytes, and the inhibition should be overcome by treatment with a tumor promoter. A small number (150) of the initiated cells were coplated with a large number (4 to 5 106) of the normal keratinocytes, or they were plated by themselves. The cultures were treated with TPA continuously, or they were treated with the dimethyl sulfoxide solvent for TPA.

6

Harry Rubin

The cultures containing only the initiated 308 cells without TPA treatment produced about 50 colonies at 14 days, out of 150 cells seeded, and the TPA raised the number of colonies to about 60. The TPA treatment also considerably increased the size of the colonies, indicating a direct stimulatory effect of TPA on multiplication of the initiated cells. The normal keratinocytes had to be added once a week because they underwent terminal differentiation in the high-calcium medium. No colonies of initiated cells appeared in coculture with normal keratinocytes, but the continuous addition of TPA to the cocultures produced about half the number of colonies as when the initiated cells were grown alone (Table I). The size of the colonies in the TPA-treated cocultures was smaller than those in TPA-treated cultures of the initiated cells alone. The results showed that TPA relieved the inhibitory effect of the normal keratinocytes on the initiated cells, but did not completely release it. Five other lines of initiated keratinocytes and papilloma cells were tested for sensitivity to inhibition of proliferation by normal keratinocytes, but the 308 line showed the maximal promotion by TPA and was used in most of the subsequent experiments (Hennings et al., 1990). The six lines displayed widely varying degrees of suppression by normal keratinocytes, but three of them showed no selective promotion of TPA on their cocultures. Inhibitors that block tumor promotion by TPA in vivo also inhibited colony formation

Table I

Normal mouse keratinocytes versus mouse epidermal tumor cells Neoplastic cell growth

Tumor cell Papilloma Papilloma Papilloma Papilloma Papilloma Papilloma Carcinoma

Excess nonneoplastic cells or conditioned medium None Normal keratinocytes Normal keratinocytes þ TPA treatment Fibroblasts Initiated keratinocytes Conditioned medium from keratinocytes Normal keratinocytes

Cell culture*

Skin grafts**

Transformed colonies

Tumor size

þþþ — þþ

þþþ þ

þþþ þþþ þþþ

þþþ þþþ ND

þþþ

þþþ

—, little or no neoplastic cell growth; , average about 90% reduction in size; þ, moderate reduction in neoplastic cell growth; þþ, slight reduction in neoplastic cell growth; þþþ, maximum neoplastic cell growth. ND, not determined. *, after Hennings et al., 1990. **, after Strickland et al., 1992.

Microenvironmental Regulation of the Initiated Cell

7

in TPA-treated cocultures, but not in cultures of the 308 cells alone. Confluent cultures of dermal fibroblasts affected neither the size nor the number of 308 colonies in cocultures. Conditioned medium from confluent cultures of normal keratinocytes failed to inhibit colony formation by the initiated cells, and conditioned medium from confluent dermal fibroblasts actually increased the size of the colonies. Therefore, the inhibition of growth of the initiated cells by keratinocytes in this cell culture system did not result from the relatively low concentration of extracellular factors when diluted in the medium, which had a volume many times greater than that of the cell. Inhibition, therefore, required close proximity of the cells to one another. As noted earlier, promotion by TPA in vivo is successful even if its application to normal skin is delayed for many months after it has been initiated by a PAH. When the treatment with TPA in cocultures was delayed for 21 days after their plating, the number and time of appearance of the 308 colonies was the same as when it was begun at day one (Hennings et al., 1990). There was no inhibition of colony formation in the cocultures in lowcalcium (0.05 mM) medium, although gap junctional communication among the 308 cells themselves and among primary keratinocytes was higher than it was in high-calcium medium. This indicated that gap junctional communication was not necessary for inhibition of 308 colony formation. Indeed, it was later found that there was no junctional communication, as measured by fluorescent dye transfer, between the 308 cells and the normal keratinocytes. The need for close proximity between the papilloma and normal cells to suppress colony formation by the former therefore indicates that material at the cell surface or at high concentration in its immediate vicinity was involved in the suppression. The informative paper by Hennings et al. (1990) established a number of basic parameters both about the suppression by normal keratinocytes of the growth of initiated epidermal cells and about the relation of tumor promotion to the interaction between these cells. Maximum suppression requires close apposition of the two cell types during the period of colony formation by the initiated cells. It occurs only in high-calcium medium in which differentiation of the normal keratinocytes and possibly of the initiated cells in contact with them takes place. Studies described below (Javaherian et al., 1998) support the idea that suppression is effected through induction of terminal differentiation of neoplastic cells by normal keratinocytes. The effect of promoters does not depend on direct stimulation of the initiated cells but on lifting their suppression by the normal keratinocytes. Confluent dermal fibroblasts cannot be substituted for keratinocytes in suppressing colony formation by the initiated cells. This indicates that dermal fibroblasts in vivo are not involved in maintaining the latent state of initiated cells, nor is it likely that other stromal elements are involved as normal keratinocytes alone are fully competent to maintain the latent state in cell culture.

8

Harry Rubin

IV. INHIBITION OF NEOPLASTIC PROGRESSION IN MICE OR IN ORGANOTYPIC CULTURE BY NORMAL KERATINOCYTES The suppression of neoplastic cell growth by normal keratinocytes was next studied by grafting mixtures made in cell culture to the skin of nude mice (Strickland et al., 1992). The cell mixtures were centrifuged and the pellets applied in an open-bottomed silicone chamber to a granulated graft bed. Grafting a mixture of primary keratinocytes with primary dermal fibroblasts forms normal skin. A mixture of the dermal fibroblasts with cells from a papilloma induced on mice by initiation with DMBA and promotion by TPA forms papillomas in the graft (Table I). The dermal fibroblasts more than doubled the size of the tumors formed by the papilloma cells alone. When an excess of primary keratinocytes is included with the mixture of papilloma cells and fibroblasts, there was a 90% reduction in the average volume of the papillomas. In contrast, inclusion of an excess of dermal fibroblasts with the usual mixture of papilloma cells and dermal fibroblasts had no effect on papilloma size. Replacing the normal keratinocytes with a line of carcinogen-initiated cells that did not differentiate when cultured in high-calcium medium, but formed normal-appearing skin in a graft, did not suppress tumor formation by the papilloma cell. Nor did keratinocytes transfected with TGF-, which stimulates their growth and increases their saturation density but does not transform them (Finzi et al., 1988), suppress tumor formation. This showed that the normal keratinocytes did not simply compete for attachment sites, three-dimensional space, or nutrients with the papilloma cells. The normal keratinocytes failed to suppress a variant of the papilloma cells that had progressed to malignancy by transfection with an activated oncogene. Repeated treatment with TPA of the grafts containing normal keratinocytes increased the percentage of grafts that developed papillomas and increased the size of the papillomas. The TPA treatment, however, did not restore the papilloma size to that obtained when the grafts did not contain normal keratinocytes. Nor did TPA increase the size of the tumors produced by the papilloma cells in the absence of added normal keratinocytes. This indicated that the TPA-induced increase in tumor size in the presence of the normal keratinocytes resulted from lifting the inhibition produced by the normal keratinocytes and not from stimulating the proliferation of the papilloma cells. The results of in vivo tumor production by cell mixtures agree in all major features with those obtained with colony formation in vitro (Hennings et al., 1990). Studies were undertaken in human cells of suppression by normal cells of the proliferation of keratinocytes with neoplastic potential (Javaherian et al., 1998). The neoplastic II-4 cell line was derived by transfecting

Microenvironmental Regulation of the Initiated Cell

9

spontaneously immortalized, aneuploid human keratinocytes (HaCaT) with an activated c-Harvey-ras oncogene. The II-4 cells display severe dysplasia in organotypic culture and low-grade malignant behavior after grafting on the dorsal surface of nude mice. A fourfold or greater excess of the normal human keratinocytes (NHK) resulted in elimination of the potentially neoplastic cells from the skin grafts. When the two cell types were grafted in equal numbers there was persistence of dysplastic foci at 4 weeks and invasion into the dermis at 8 weeks. Seven-day organotypic cultures of NHK alone showed a well-differentiated stratified epithelium, whereas the II-4 cells formed a poorly differentiated dysplastic epithelium. Mixtures exhibited distinct foci of altered cells in the context of normal cells, thereby simulating intraepithelial neoplasia. With a 12-fold excess of NHK, the individual, potentially neoplastic cells were in a suprabasal position and ceased to proliferate. If they rose to superficial layers surrounded by normal cells, they underwent terminal differentiation. With equal numbers of the two cell types, the neoplastic cells occurred in large clusters, where they continued to divide in suprabasal position but did not differentiate. It was concluded that normal tissue architecture suppresses early neoplastic progression of low-grade neoplastic human epidermal cells. If the latter were in clusters, they continued to behave like neoplastic cells, proliferating in suprabasal position and not differentiating. Apparently, a critical number of neoplastic cells in close proximity to one another are required for expansion and invasion. The effect of normal keratinocytes on neoplastic cell proliferation was compared with the effect induced by the spontaneously immortalized HaCaT line of keratinocytes (Vaccariello et al., 1999). As noted above, the spontaneously immortalized HaCaT cell line had served as the parent of the neoplastic II-4 line by transfecting it with the activated Harvey-ras oncogene. Unlike the NHK, the HaCaT cells proliferated in both the basal and suprabasal layers of organotypic cultures, but unlike their neoplastic II-4 derivative, they produced no tumors in grafts on mouse skin. An excess of NHK over II-4 cells in the organotypic cultures prevented proliferation of the low-grade neoplastic cells, but excess immortalized HaCT keratinocytes allowed proliferation of the II-4 cells into clusters. These effects could be expressed quantitatively by the percentage of II-4 cells in the original mixture and after 7 days’ incubation in organotypic culture (Table II). In a 12 : 1 ratio of NHK : II-4 cells, the percentage of II-4 cells remained constant over the 7-day period. In the same initial ratio of HaCaT : II-4 cells, the percentage of II-4 cells increased threefold. Hence, the immortalized cells constituted a permissive microenvironment for proliferation of the low-grade neoplastic cells. There was no induction by the HaCaT cells of terminal differentiation in the neoplastic cells. Cell-to-cell contact mediated by E-cadherin (Garlick, J.

10

Harry Rubin

Table II Comparison of the effect of NHK and HaCaT keratinocytes on proliferation of II-4 low-grade neoplastic keratinocytes

Ratio

Percentage II-4 in original mixture

Percentage II-4 after 7-day culture

12 : 1 (NHK : II-4) 12 : 1(HaCaT : II-4)

8.3 8.3

8.5 24.3

Abstracted from Vaccariello et al., 1999, Table 1.

personal communication) was apparently required for suppression of the II-4 by the NHK. Whether direct contact between apposed membranes is required, or only a highly localized concentration of extracellular matrix material, cannot be distinguished from this experiment but will be discussed later. Removing suprabasal layers from mixtures of the NHK and II-4 cells after a few days growth and then allowing 10 days of regrowth resulted in normal stratification and morphological differentiation of the NHK and elimination of the II-4 cells (Vaccariello et al., 1999). This showed that all the neoplastic cells had been displaced into the suprabasal layers by preferential attachment of the normal keratinocytes to the type I collagen matrix. The authors concluded that potentially malignant cells must be in a permissive microenvironment to allow progression to malignancy. In this case the low-grade malignant II-4 cells could not compete with normal cells for attachment to the extracellular matrix, thereby being displaced into the suprabasal layers, where they were subject to suppression by contact with the normal cells. Hence a normal stratified epithelium has the capacity to eliminate the small number of cells that have undergone the early stages of malignant change. (It should be noted that classically initiated cells in the mouse skin are not eliminated by the surrounding normal epidermis, but can be detected by promotion into tumors for at least a year after their initiation [Van Duuren et al., 1975]). The immortalized but nontumorigenic HaCaT cells, which represent an early stage of transformation, fail to suppress neoplastic growth of the II-4 cells, which are in a more advanced but still early stage of neoplastic progression. This indicates that carcinogenic treatment not only induces the low-grade malignant cells but alters their microenvironment to facilitate their growth and progression to higher grades of malignancy. The mechanism of tumor promotion was also studied in the organotypic human cell culture model (Karen et al., 1999). In a preliminary study using conventional monolayer culture methods, TPA markedly reduced the growth of NHK but had no effect on the growth of the II-4 low-grade malignant keratinocytes. When the II-4 cells were mixed with a suppressive excess of normal keratinocytes in organotypic culture, treatment with

Microenvironmental Regulation of the Initiated Cell

11

TPA inhibited the proliferation of the normal cells and thereby allowed selective expansion of tumor cell colonies. Thus TPA played a similar role in promoting neoplasia in conventional mouse epidermal cell culture (Hennings et al., 1990), in grafts of mouse cells on skin (Strickland et al., 1992), and in organotypic cultures of human epidermal cells (Karen et al., 1999).

V. INHIBITION OF VIRALLY TRANSFORMED MALIGNANT KERATINOCYTES BY DERMAL FIBROBLASTS IN SKIN GRAFTS AND IN SUBCUTANEOUS INJECTIONS A series of experiments was done in which primary keratinocytes from newborn mice were transformed to malignancy by infection with Harvey sarcoma virus (HaSV) plus helper virus and mixed with other cells to determine the effect of mixing on tumor growth (Dotto et al., 1988). The helper virus was included with the defective HaSV to ensure infection of all the keratinocytes. Murine leukemia virus alone was used to infect normal dermal fibroblasts or keratinocytes to block their infection by the HaSV when mixtures were made with the transformed cells. The transformed keratinocytes produced highly invasive carcinomas of grades I, II, and III in 100% of recipients when grafted alone to the skin of syngeneic mice. Grafting of the transformed keratinocytes mixed with a fourfold excess of dermal fibroblasts resulted in a drastic inhibition of malignant growth. No macroscopically visible tumors occurred in most of the mice grafted with the mixtures, and the few others exhibited a marked reduction in tumor size compared with those grafted with the transformed keratinocytes alone. When the mixture was reduced to a 1 : 0.4 ratio of transformed keratinocytes to dermal fibroblasts, there was no inhibition of tumor growth. Grafting an excess of normal keratinocytes with the transformed cells also failed to inhibit tumor development. These results were polar opposites to those described above with mixtures of early-stage neoplastic cells and normal keratinocytes or fibroblasts (Hennings et al., 1990; Javaherian et al., 1998; Strickland et al., 1992). In those cases initiated cells, papilloma cells, and early-stage malignant cells were inhibited only by normal keratinocytes and not at all by dermal fibroblasts—in fact, an excess of dermal fibroblasts was routinely added to initiated cells in grafts to increase the size of tumors—and the normal keratinocytes were ineffective at inhibiting the growth of more advanced malignant keratinocytes (Strickland et al., 1992).

12

Harry Rubin

It is also noteworthy that infection of keratinocytes with HaSV without helper virus resulted only in production of benign papillomas in grafts (Roop et al., 1986) rather than the invasive carcinomas produced by simultaneous infection with HaSV with helper virus (Dotto et al., 1988). In the absence of helper virus, only a small number of the keratinocytes would be infected, without a chance of reproducing new infectious virus to transform the rest of the keratinocytes. Without helper virus, the infected cells would be surrounded by uninfected keratinocytes and fibroblasts. When HaSV and helper virus were combined, sufficient time was allowed in culture to allow virus reproduction and secondary infection of all the keratinocytes before mixing with other cells and grafting. The results indicate that the proportion of infected keratinocytes is a strong determinant of the subsequent behavior of the cells that would determine whether they formed benign papillomas or highly malignant cancers. A similar conclusion was reached some 30 years earlier in the case of chicken embryo fibroblasts infected with different concentrations of Rous sarcoma virus (Rubin, 1960b) and will be discussed in a later section. The dermal fibroblasts that inhibit tumor formation by HaSV-transformed keratinocytes produce a diffusible inhibitory factor belonging to the transforming growth factor (TGF-) family (Missero et al., 1991a, 1991b). This factor suppresses the growth of the transformed keratinocytes in culture and was thought to be responsible for the inhibition of tumor formation by fibroblasts after subcutaneous injection into mice. The tumor suppression is associated with a striking induction of squamous cell differentiation (Ramon y Cajal et al., 1994). The suppression may be related to the observation that defects in TGF- signaling cooperate with a ras oncogene to cause aneuploidy and malignant transformation of mouse keratinocytes (Glick et al., 1999). The suppressive role of dermal fibroblasts on tumor formation by the HaSV-transformed keratinocytes may therefore be to counter the aneuploidy and transformation induced by the TGF- defect in cooperation with the ras oncogene of HaSV. The inhibition of the highly invasive carcinoma by dermal fibroblasts is of questionable significance for the failure of initiated keratinocytes to form tumors in mouse skin without repeated application of a promoter. The invasive carcinoma is, of course, a far more advanced neoplastic stage than initiated cells and one that requires no promotion to produce cancer. Attributing a suppressive role to dermal fibroblasts in classical epidermal carcinogenesis of the mouse is questionable, in any case, because the distribution of fibroblasts in the dermis is very sparse, and they occur at a distance from the epidermis. Their sparseness takes special significance because they must be added in four fold excess in close proximity to the carcinoma cells to suppress tumor formation (Dotto et al., 1988). Such ratios would not occur when a carcinoma invades the dermis. (In limited testing,

Microenvironmental Regulation of the Initiated Cell

13

there was no contact inhibition between normal epithelial and fibroblastic cells [Eagle and Levine, 1967]). This contrasts with the inhibition of earlystage neoplastic cells by keratinocytes, which are always, in the normal course of initiation of skin tumors, in great excess over the initiated cells and in direct contact with them. Hence, the experimental suppression of early-stage neoplastic keratinocytes by added normal keratinocytes is the more appropriate model for understanding the classic in vivo quiescence of initiated cells and their expression under treatment with promoters. However, the firmness of adhesion of keratinocytes to the basement membrane plays a role in suppression of proliferation of early-stage malignant cells by NHK. This was seen in organotypic cultures of mixed-cell populations in which the neoplastic cells were displaced by the strongly adhesive NHK to a suprabasal position, where they are more likely to differentiate than proliferate (Vaccariello et al., 1999). The shift in position is apparently related to a reduced level of 1 integrins in the neoplastic cell surface, as adhesiveness increases with the level of 1 integrins (Jones and Watt, 1993). Normal stem cells have the highest level of surface integrins and of adhesiveness to the basement membrane, which maintains them in the basal layer. The basement membrane plays an essentially passive role in this aspect of interaction between normal and neoplastic cells. A similar competitive adhesion between normal and neoplastic tracheal epithelium, which results in displacement and inhibition of the latter, was found when the mixture was added to that which had been denuded of their own epithelium (Terzaghi-Howe, 1987).

VI. RESTRICTION OF MUTANT KERATINOCYTES TO THE EPIDERMAL PROLIFERATIVE UNIT IN ULTRAVIOLET (UV) CARCINOGENESIS OF THE SKIN Basal cell carcinoma (BCC) of the skin is the most frequent cancer in the United States (Ziegler et al., 1993). It is believed to come from hair follicles and the tumor resembles the basal layer of the epidermis. It spreads mainly by local invasion and tends to remain diploid. The less common squamous cell carcinoma (SCC) is more cornified, has a greater tendency to metastasize, and usually becomes aneuploid. The major cause of these tumors is exposure to sunlight as their incidence correlates with outdoor exposure, sunny climates, and fair skin. One effect of sunlight is the induction of genetic damage, as indicated by the 2,000-fold increase of skin tumors in patients with xeroderma pigmentosum, who cannot repair UV-induced DNA photoproducts.

14

Harry Rubin

Because adults can reduce risk of precancers by using sunscreen, some of the sun-induced damage must occur during adult life (Brash, 1997), but most of the critical sunlight exposure occurs before age 18, as people who moved from England to Australia as children, but not those who moved as adults, acquire the high Australian skin cancer risk. Thus, some of the genetic lesions induced by sunlight are a half-century old. UV-induced mutations are located where one pyrimidine is next to the other, and they usually (70%) involve C ! T transitions. About 10% of the UV-induced mutations are CC ! TT, resulting from replacement of both cytosines, and are considered diagnostic for UV. The most important UV wavelength in skin tumor formation is UVB (280–315 nm). It was generally believed that about 50% of BCCs and over 90% of SCCs have mutations in the p53 gene: Many of these tumors have mutations in both alleles of the gene (Brash et al., 1996). However, meticulous microdissection of 11 BCCs and direct DNA sequencing revealed that all the tumors had p53 mutations (Ponte´n et al., 1997). There was a dominant clone in each tumor that was prone to genetic progression, with appearance of subclones that had a second and even third p53 mutation. All of the mutations seen in these tumors had a change in an amino acid of the p53 protein. Because it is likely that sunlight often mutates codons in the third position, leaving the amino acid unchanged, such synonymous mutations are evidently not tumorigenic. The sunlight-induced nonsynonymous mutations in p53 therefore must have been selected for tumor formation and have played a causal role in tumor development. Because they occur in actinic keratosis, which is a precursor for SCC, the p53 mutations are likely to be involved in an early stage of tumor development such as initiation. Although mutations in SCC and BCC occur in hotspots, they were spread almost evenly across the p53 genome in actinic keratosis, a precancerous lesion. This indicates that selection occurs between the precancer and cancer stages. Ultraviolet light, similar to PAHs, can act as both an initiator and a promoter. The promoter action of UV was directly demonstrated in mice in which DMBA was applied once to the skin as an initiator and was followed by repeated exposure of the skin to UV (Epstein and Epstein, 1962). Although a few mice treated with DMBA alone developed skin tumors, that number was greatly increased by twice-weekly exposure to UV light for 67 weeks. Mice exposed to UV light alone developed no tumors at all. Hence, UV as used in this experiment was acting strictly as a promoter on DMBA-initiated skin. Human skin contains clonal patches of p53-mutated keratinocytes (Jonason et al., 1996). The clones have 60–3,000 cells and involve as much as 4% of the epidermal volume. They are more frequent and larger in sunexposed skin than in sun-shielded skin. This indicates that in addition to

Microenvironmental Regulation of the Initiated Cell

15

being a tumorigenic mutagen, sunlight acts as a tumor promoter by favoring expansion of p53-mutated cells. The promotion activity of sunlight appears to contribute a greater number of mutant cells to the skin than does its initial mutagenic effect, because in chronically exposed skin the aggregate size of clones that are >0.05 mm2 (sizes that are clearly exposure dependent) exceeds that of the more numerous smaller clones. It seemed likely that clonal expansion was favored because p53 mutations conferred resistance to apoptosis resulting from UV radiation. Skin appears to possess a p53dependent response to DNA damage that aborts precancerous cells (Ziegler et al., 1994). A p53 mutation in a single cell presumably reduces the apoptotic response and allows selective clonal expansion of the mutated cell into actinic keratosis as the basis of tumor promotion. This model for the development of skin tumors in humans was tested by experiments in mice (Zhang et al., 2001). The size and number of p53mutated clones were compared in the skin of mice irradiated five times per week for varying periods of time and then terminated. The number and size of the clusters increased with the duration of UV exposure. There was a progressive increase in the percentage of large p53-mutated clusters and a corresponding decrease in the percentage of small clusters. The UVinduced expansion was clonal rather than an overlap of clusters with different p53 mutations. When the irradiation was terminated, the number of clones decreased rapidly, but the clonal areas remained constant, although cell number per clone increased. In contrast, continued irradiation resulted in a continuing increase in both the number and area of clones. This showed that clonal expansion is not independent of the microenvironment. Rather, it results from a local physiological change in the surroundings that requires continued presence of the UV carcinogen. The asymmetry of clonal geometry was consistent with the physiological model of clonal expansion. An unexpected result was the quantization of the clone sizes corresponding to multiples of 12–16 cells, which is similar to the number of nucleated cells in a mouse epidermal proliferative unit. This indicated that clonal expansion proceeded by successively colonizing adjacent stem-cell compartments, and that escape from the stem-cell compartment is the ratelimiting step in the process. In the absence of UV, clones were constrained in the stem-cell compartment despite increasing slightly in cell number per clone. Apparently the barriers to expansion are the flanking stem-cell compartments. The UV irradiation is thought to break down the barrier by killing the nonmutated epidermal cells, allowing the p53-mutated cells to breech the barrier. The model is consistent with the results in previous sections in which normal mouse or human epidermal cells suppress the growth of early-stage epidermal tumor cells (Hennings et al., 1990; Javaherian et al., 1998; Strickland et al., 1992). There is no indication that

16

Harry Rubin

regulation of these cells is determined by relatively distant dermal fibroblasts, which, in any case, are few and far between, and not arranged in stem cell compartments. Although the emphasis in the cited article (Zhang et al., 2001) is on a physiological change (e.g., apoptosis) in the microenvironment from chronic UV irradiation as the basis for expansion of the ‘‘imprisoned’’ p53-mutated clones in normal skin, the evidence does not exclude a contribution from accumulation of new mutations to expansion of the clones. This is especially true for agents such as UV that damage DNA and substantially increase the frequency of chromosome breakage and gene amplification (Windle et al., 1991; Windle and Wahl, 1992). Gene amplification is thought to underlie adaptive mutation and general hypermutability under selective conditions in bacteria (Andersson et al., 1998; Hendrickson et al., 2002), and it occurs at a high rate in neoplastic mammalian cells (Tlsty et al., 1989). Assuming that the primary effect of chronic UV is to relieve the suppression of clonal expansion among p53-mutated clones by damaging the suppressive normal keratinocytes in the microenvironment, the active proliferation of the p53 clones in the continuing exposure to UV would generate further high-frequency mutations such as gene amplification. The accumulation of mutations would then drive successive progression to dysplasia, actinic keratosis, BCC or SCC, and finally metastasis. The effects of UV have parallels in the classic initiation–promotion model in mouse skin induced by carcinogenic PAHs. A single application of a carcinogenic dose of PAH produces about 50,000 stable adducts to the DNA of every exposed epidermal cell within 24 hours (Melendez-Colon et al., 1999). Many of the epidermal cells are quickly killed or exhibit damage in the form of nuclear and cytoplasmic swelling (Cramer and Stowell, 1942). Most of the surviving cells are likely to have undergone several mutations (Mackay et al., 1992; Rubin, 2002). There is a long-term increase in sensitivity of the entire treated field to growth stimulation by promoters (Frankfurt and Raitcheva, 1973). Hence, it is probable that the entire cellular microenvironment becomes more permissive to tumor formation by the rare initiated cell in its midst (Yuspa et al., 1982). Not only does the altered microenvironment become more permissive for the unregulated multiplication of initiated cells, but it may become a breeding ground for ‘‘second primary tumors’’ (Braakhuis et al., 2003). This concept is referred to as field cancerization, first introduced to explain the development of multiple primary tumors in histologically abnormal tissue surrounding oral squamous carcinomas (Slaughter et al., 1953). It has since been extended to cancers in many other tissues and is supported by recent molecular findings (Braakhuis et al., 2003). It has important clinical consequences because such fields often remain after surgical removal of the

Microenvironmental Regulation of the Initiated Cell

17

primary tumor and have an increased likelihood for the development of new tumors.

VII. MICROENVIRONMENTAL SUPPRESSION OF NEOPLASTIC TRANSFORMATION AMONG CULTURED FIBROBLASTS At about the same time that initiation and promotion were recognized in chemical carcinogenesis of the epidermis in rabbits (Rous and Kidd, 1941) and mice (Berenblum, 1941), observations were made on Rous Sarcoma Virus (RSV) infection of mesenchymal tissue in chickens that indicate that that a similar situation exists under certain conditions in fibroblasts of that tissue. Intracelomic or intravenous injection of the original strain of RSV into chick embryos was followed by proliferation of the virus without eliciting tumors, but it did produce hemorrhagic lesions without evidence of neoplastic cells in serial sections of tissue around the lesions (Milford and Duran-Reynals, 1943). (Epitheliomas and adjacent sarcomas could be produced on the extra embryonic chorioallantoic membrane [Rubin, 1955]). The newly produced virus induced typical Rous sarcomas when injected into chicks and pullets. It was later found that injection into the chick embryo limb of either of two RSV strains of European origin led to virus replication and expression of the specific oncogene of RSV, but again no tumors were formed (Dolberg and Bissell, 1984). Dispersal of the virusproducing fibroblasts from the embryonic limb and their propagation in culture was followed by neoplastic transformation after a 24-hour delay. The results showed that the microenvironment of the infected cells determines whether neoplastic transformation occurs. In addition, it was established that wounding the RSV-infected tissue of the chickens is necessary for tumor formation to occur (Dolberg et al., 1985; Rous et al., 1912), which is reminiscent of the promotional role of wounding in chemical carcinogenesis of the rabbit (McKenzie and Rous, 1941) and mouse (Hennings and Boutwell, 1970). This view was reinforced by the finding that high concentrations of TPA could substitute for wounding in promoting sarcoma development in RSV-infected tissue (Dolberg et al., 1985). Although the lack of tumor development in RSV-infected embryos and the need for wounding to produce tumors in chickens were not originally subsumed under the rubric of initiation and promotion in fibroblasts, results obtained with RSV-infected and chemically treated fibroblasts in culture justify that designation and add insights about the role of the microenvironment and of selection in those processes.

18

Harry Rubin

A. Primary Chick Embryo Fibroblasts Infected with RSV The first indication that normal cells and proteins could suppress the growth of neoplastic cells came from studies of infection of chick embryo fibroblasts with the Bryan high-titer strain of RSV. The number of infectious particles of RSV in a preparation was determined by the number of transformed foci produced when serial dilutions were added to a culture of CEF that was overlaid with semisolid agar medium to prevent secondary foci from newly produced virus (Temin and Rubin, 1958). The number of foci increased directly with the concentration of virus that was added until they overlapped one another and became uncountable. To determine the number of infected cells after high concentrations of virus were added, the cells were dispersed, diluted, and seeded as ‘‘infective centers’’ on a background of uninfected chick embryo fibroblasts made less than 24 hours earlier (Rubin, 1960a). The proportion of focus-forming cells in a culture infected with concentrations of RSV exceeding the number of cells exposed did not exceed 10% of the population if the cells were reseeded within a few hours of infection. If reseeding was delayed for 30 hours or more after infection, when visible transformation had begun in the population, the proportion of cells forming foci on reseeding rose 5- to 10-fold, sometimes approaching 100%. The results indicated that the confluent background of uninfected cells was able to suppress focus formation by a high proportion of newly infected cells, but that suppression was ineffective once morphological transformation began in the infected cells. If the background culture of uninfected CEF had been growing for 48 hours or more and was densely confluent before adding the infective centers, no foci were formed even if the latter had been infected 2–3 days earlier and were already transformed. This result differs from others reported and is probably related to the use of 9.2% CS (plus 1.5% chicken serum) rather than lesser amounts of CS in the overlay, as will become apparent later. In a further series of experiments, the CEF cultures were exposed to a high enough concentration of RSV to infect most of them, or to a low enough concentration to give a countable number of well-separated transformed foci on a confluent background of the normal fibroblasts (Rubin, 1960b). Cultures infected with both concentrations of RSV were incubated in medium containing 5% CS (from 4–8-month-old calves) or 10% FBS (formerly called fetal calf serum). Most of the cells in cultures infected with the high concentration of RSV were transformed regardless of the serum type used. Easily countable numbers of multilayered transformed foci were produced in cultures infected with the low concentration of RSV and incubated in CS, but none were seen in those incubated in FBS, although the cells

Microenvironmental Regulation of the Initiated Cell

19

produced large amounts of RSV. Close inspection of the latter cultures revealed faint, discrete areas of monolayered cells that had a slightly higher population density than the normal cells in the surrounding monolayer, but that did not have the disorganized, multilayered arrangement characteristic of transformed foci (Fig. 1). These observations indicated that the morphological transformation and overgrowth characteristic of transformed cells were suppressed by the interaction of infected cells with components of FBS and, perhaps, with material associated with the large number of surrounding cells. To determine whether the difference in focus formation represented a qualitative or a quantitative difference between serum types, a series of RSV dilutions was used to infect cultures that were overlaid with a wide range of concentrations of either serum. The results are shown in Table III. Cultures infected with the highest concentration (1 : 20) of RSV displayed uncountable numbers of overlapping transformed foci, labeled ‘‘confluent,’’

Fig. 1 Foci of RSV-infected cells and their suppression. Cultures of CEF were infected with about 1,000 focus-forming units of the Bryan strain of RSV and overlaid with soft agar in medium containing (left) 5% CS or (right) 10% FBS. They were incubated for 7 days, and then fixed. Distinct, multilayered foci of transformed cells appeared in 5% CS, but only irregularly shaped, monolayered areas of normal-looking cells at a slightly higher population density than the surrounding uninfected cells were seen in 10% FBS, as outlined by the broken line. After Rubin, 1960b.

20

Harry Rubin

Table III The effect of serum type and concentration on chick embryo fibroblast transformation after infection with varying concentrations of Rous Sarcoma Virus No. transformed foci Serum concentration, % 2.5

5

10

15

20

Virus dilution 1: 1: 1: 1: 1: 1: 1: 1: 1: 1: 1: 1: 1: 1: 1:

20 1,000 10,000 20 1,000 10,000 20 1,000 10,000 20 1,000 10,000 20 1,000 10,000

Calf serum

Fetal bovine serum

Confluent 500 61 Confluent >>500 292 Confluent >>500 240 Confluent 44 (small) 0 Confluent 0 0

Confluent >500 91 Confluent 200 (very small) 0 Confluent 0 0 Confluent 0 0 Confluent 0 0

The 1:20 dilution of RSV contained about 0.3 focus-forming units per cell per infecting dose in 5% calf serum. Abstracted from Rubin 1960b, Table 2.

throughout the cell sheet in all concentrations of either serum type. The yield of discrete transformed foci in the lower concentrations of RSV (1 : 1000 and 1 : 10,000) in 2.5% serum was slightly higher in FBS than in CS. In 5% FBS, there were no transformed foci in a 1 : 10,000 dilution of RSV, whereas there were 200 multilayered foci in a 1 : 1,000 dilution, although they were markedly reduced in size compared with those in 5% CS. A roughly similar suppression was not achieved in CS until its concentration was raised to 15%. The number of transformed foci in 5% CS was the maximum seen with any serum concentration in the higher dilutions of RSV. In 10% or higher FBS no transformed foci were apparent in either of the higher dilutions of RSV. Such complete suppression of discrete focus formation by CS required raising its concentration to 20%. Hence, two to three times higher concentrations of CS were required than those of FBS to produce equivalent degrees of focus formation. However, no suppression of transformation was detected with either serum when the multiplicity of infection was high enough to initially infect a large fraction of the cells. To compare the growth rates and saturation densities of RSV-infected cells that do or do not undergo morphological transformation, the CEF were infected with a low dose of RSV in 5% CS and in 10% FBS, and the total number of cells and the number of focus-producing cells (infective

21

Microenvironmental Regulation of the Initiated Cell

centers) were measured daily. (See Rubin [1960b] for rates of virus release.) More cells attached to the culture dish in the FBS than in the CS (Fig. 2). The total number of cells increased for 3 and 4 days in FBS and CS, respectively, before the cells reached their postconfluence saturation density. The number of infective centers (cells from either CS or FBS that initiated focus formation when trypsinized and replated in nonsuppressive 5% CS) increased exponentially in both sera for 4 days, but there were more focus formers from cultures incubated in FBS than in CS when both were dispersed and assayed in 5% CS. Beyond 4 days, however, there was no further increase of infective centers in FBS, whereas the increase continued in CS at the original, exponential rate. RSV production and release into the medium continued in both sera throughout the experiment, although finally at a higher rate in CS than in FBS (Rubin, 1960b). The results showed that

TOTAL UNITS PER PLATE

106

TOTAL CELLS

105

104

103

102

INFECTIVE CENTERS

0

1

2

3 4 DAYS

5

6

Fig. 2 Growth curves of RSV-infected and uninfected CEF in media permissive and suppressive of transformed focus formation. Cultures were seeded with 5 105 cells in medium containing 5% CS or 10% FBS and infected with about 1,000 focus-forming units of RSV. They were then overlaid with soft agar with medium containing the same sera as before. At daily intervals the overlay was removed and the cells were detached and counted and then assayed for infected cells by their development into transformed foci in 5%. Cell attachment and RSV infection were higher in 10% FBS than in 5% CS in the original cultures, but transformed foci in those dishes only appeared in 5% CS. Note the plateaus in total cells after 3–4 days in both media and in number of RSV-infected cells in 10% FBS after 4 days. In contrast, the number of RSV-infected cells in 5% CS continued to increase throughout the experiment as multilayered foci were building up on the confluent, contact-inhibited sheet. After Rubin, 1960b.

22

Harry Rubin

the number of infective centers reached a plateau in FBS within a day after contact inhibition had suppressed further growth of the total number of cells. This indicated that the infected cells in FBS were almost as sensitive to contact inhibition as the surrounding normal cells, which would account for their failure to form multilayered transformed foci. In contrast, the cells in the CS concentration that was optimal for focus formation allowed continued growth of the infected cells to the end of the experiment and accounted for continued multilayering of cells in foci in the original culture for days after the surrounding normal cells had reached their saturation density. FBS also inhibits the spontaneous conversion of mouse embryo cells in culture to tumor-producing capacity in mice that otherwise occurs frequently in medium containing CS (Evans and Andresen, 1966; Sanford et al., 1972). The transformation-inhibiting components of FBS retained their activity after heating at 70 C for 30 minutes but were inactivated at 80 C and were nondialyzable, indicating that they were proteins (Rubin, 1960b). The total protein concentration of FBS is only about 65% as high as that of CS (Rubin and Xu, 1989), which indicated that the inhibitory proteins constituted a considerably larger fraction of total protein in FBS than in CS. Fetuin, an alpha globulin that constitutes approximately 45% of the protein in FBS (Harris, 1964), was inhibitory to focus formation when added to 5% calf serum. It also appears to inhibit spontaneous transformation of mouse embryo cells in culture for tumor-forming capacity in mice (Sanford et al., 1972). Fetuin was thought to be necessary for attachment and spreading of cells on the surface of culture dishes, and it was suggested that this function was related to its ability to inhibit the activity of the trypsin used in detaching the cells from the substratum and from each other (Fisher et al., 1958). Serum was indeed found to reverse trypsin damage to cells, just as soy bean trypsin inhibitor does (Hebb and Chu, 1960). Fetuin was not as strong a suppressor of transformation as moderate concentrations of FBS or high concentrations of CS, indicating that it was not the only protein that contributes to suppression. Major protease inhibitors in human plasma, and presumably serum of other mammals, are -globulins (Heimburger, 1975), as is fetuin. Recent genome search reveals that there are about 500 serine protease inhibitors, commonly known as serpins (Silverman et al., 2001), and there are no doubt many proteases of other types. An experiment was designed to explore the role of factors secreted by normal cells in the suppression of typical transformed foci and in imposing contact inhibition on the normal-appearing RSV-infected colonies (Rubin, 1960b). Medium containing 5% CS was incubated for 4 days on chick embryo cultures infected with high and low multiplicities of RSV or on empty dishes. The media were centrifuged at high speeds to pellet RSV, and the supernate was heated at 60 C for 1 hour to inactivate any remaining

Microenvironmental Regulation of the Initiated Cell

23

virus. The media were then overlaid on fresh cultures immediately after their infection with low concentrations of RSV. Medium conditioned on the heavily infected cultures supported the development of almost as many transformed foci as medium that had been incubated on empty dishes. In contrast, medium from the lightly infected cultures, which contained an overwhelming predominance of normal cells, suppressed focus formation more than 10-fold. The results indicated that normal chick embryo fibroblasts released substances, most likely proteins, that supplemented those in 5% CS to inhibit transformation, although not as completely as 10% FBS. However, the amount of protein released from the approximately 106 cells in the culture is far less than 1/100 the amount of serum protein in the medium, so the released protein may be far more efficient at suppressing focus formation. There was virtually no transformation-inhibitory activity in the medium of heavily infected, transformed cultures indicating either a marked reduction in synthesis or a degradation of the inhibitory material by the transformed cells. Reconsideration of these results in the light of later experiments indicates a possible identity of the transformation-suppressing material in medium conditioned by normal fibroblasts with the growth-supporting effect of such medium on small numbers of cells (Rubin, 1966b), in contrast to the growth-inhibitory effect of medium conditioned by RSV-transformed cells (Rubin, 1966a). Medium conditioned by epithelial cells from kidney or lung neither inhibits nor enhances the growth of small numbers of fibroblasts (Rubin, 1967), nor is there mutual contact inhibition between the cell types (Eagle and Levine, 1967). The growth-supporting material in normal fibroblast-conditioned medium is of high molecular weight and inactivated at much lower temperatures (Rubin, 1966a) than the focus-suppressing activity in FBS (Rubin, 1960b). It appears as though the material in serum that is reinforced by secreted material in medium conditioned by large numbers of normal chick embryo fibroblasts maintains the normal appearance and behavior of RSV-infected cells when the latter constitute a small fraction of the population. Part of the normalized behavior of the RSV-infected cells is to undergo contact inhibition shortly after the surrounding normal cells (microenvironment) undergo contact inhibition. The fact that the normalized RSV-infected cells retain their normal appearance and behavior even when they can be seen in contiguous contact with one another in localized groups of about 100 cells (also the number derived from the increase of infective centers) indicates that the dominant transformation-suppressive effect stems from substances in serum plus those released by the normal cells rather than by direct contact with those cells. Close proximity to the normal fibroblasts would increase the local concentration of large, temperature-sensitive, growth-enhancing molecules they release (Rein and Rubin, 1968). The growth-enhancing effect of close proximity is brought out by

24

Harry Rubin

seeding small numbers of cells at high density in restricted areas and by the sharply delimited distance of growth enhancement by subconfluent x-irradiated feeder cells on a very sparse group of living cells maintained on a gradient of distance from the feeder cells, with the optimal effect obtained by direct seeding on the same surface alongside growth enhancing cells (Rein and Rubin, 1968). If similar proximity effects were obtained in suppression of transformation by proteins released from normal cells supplementing those present in serum, they might simulate a contact requirement to suppress transformation. Alternatively, cell contact would supplement the soluble protein effects by virtue of the highest possible concentration of secreted products at the cell surface, or of the loosely bound structural components of the cell membrane itself. The conflation between close proximity and direct contact has a precedent in studies of embryonic induction. The requirement for contact was first adduced from the classic work of Spemann on lens induction in which reconstitution of the presumptive lens with the optic vesicle failed to induce lens (Spemann, 1938). The failure was attributed to a thin layer of mesenchyme that lay between the two parts. Many observations of this kind were made in different developing systems that seemed to confirm the conclusion that direct contact between interacting parts was necessary for induction (Grobstein, 1961). It was later found that interposition of thin cellulose ester membrane filters between several different interacting pairs of developing tissues did not interrupt the developmental reaction between them (Grobstein, 1961). The pore size of the membrane was sufficiently small (about 100 m) to exclude cytoplasmic penetration. The inductive activity declined noticeably at a distance of 30–40 m and was extinguished at 60–80 m. The limits of induction correlated with the penetration into the pores of labeled amino acids in proteins secreted by the inducing tissue. This indicated that inductive effects are closely associated with protein-rich material secreted by the cells into their immediate microenvironment. Materials have been seen by electron microscopy to be closely associated with but external to the plasma membrane of cells about to undergo induction. More mobile materials may be derived from the surface stuff, and these have a tendency to organize as particles large enough to be restricted by pores of 100 nm. Some inductive interactions have been found over distances that range up to tens of microns. An operational definition of contact or contiguity was developed to fit these inductive conditions, in which a cell is considered in contact with another body if they are separated by a narrow space occupied by a molecular population whose free mobility is restrained. This would be considered contact of a physiological rather than a physical kind. Agents such as tumor promoters that disturb junctional communication might be expected to lift the restraint on molecular mobility of intercellular molecules and to give the appearance of a requirement for

Microenvironmental Regulation of the Initiated Cell

25

membrane-to-membrane contact and junctional communication between normal and transformed cells for suppression of the neoplastic phenotype by normal cells. The suppression of RSV-induced transformation by conditioned medium from normal chick embryo cells and the failure of suppression of medium from transformed cells indicated that the latter might release material that degraded the suppressive activity of the former. This suggestion was reinforced by the growth-supporting activity for small numbers of cells by normal conditioned medium and by the growth-inhibiting effect of medium from RSV-transformed cultures, as well as the poor growth of pure cultures of RSV-transformed cells. One possibility that suggested itself was that the transformed cells released proteolytic enzymes. It was found that the addition of trypsin to the medium of a contact-inhibited chick embryo cell culture, in amounts that increased the refractility of the cells but that were too small to detach or visibly change their relations to one another, stimulated a round of DNA synthesis and a doubling of cell number (Sefton and Rubin, 1970). This lent concreteness to the hypothesis that the production of proteases by the transformed cells favored their excessive multiplication in forming foci on a confluent sheet of contact-inhibited normal cells. The release of proteases by RSV-transformed cells was first suggested in 1925, when it was found that such cells, unlike normal cells, lysed plasma clots in which they were growing (Fischer, 1946). This observation was later expanded by the finding that lysis of the clot by Rous sarcoma cells preceded and accounted for their change from a fibroblastic to a rounded morphology, which could be reversed by adding fresh, solidifying plasma (Doljanski and Tennenbaum, 1942). To support continued growth of the Rous sarcoma cells in culture, it was necessary to add normal cells at every transfer, indicating that the unmodulated activity of the proteolytic material was damaging to the cells. It is significant to note that a much more limited local liquefaction of the plasma clot occasionally occurred in cultures of normal chick embryo fibroblasts and was accompanied by the same changes from a spindle shape to a round shape seen in clot liquefaction by Rous sarcoma cells. A factor was detected in the medium of Rous sarcoma cells in culture that stimulates sustained overgrowth in crowded cultures of chick embryo cells, and it was inferred to be a protease (Rubin, 1970). It was then found that RSV-transformed chick embryo fibroblasts seeded on a film of insoluble fibrin lyse the film, but that normal chick embryo fibroblasts do not do so (Unkeless et al., 1973). FBS inhibits the fibrinolytic activity in contrast to chicken serum, which increases it, although serum from RSV-tumor-bearing chickens is inhibitory. The fibrin layer had to be in direct contact with the surface of the Rous sarcoma cells in the presence of FBS for fibrinolysis to occur, indicating a high concentration of fibrinolytic activity at the cell

26

Harry Rubin

surface and its immediate proximity. If the cells were incubated in the absence of serum, then chicken serum had to be added to produce fibrinolytic activity, which suggested that a cellular protease had to activate a zymogen in the serum. Similar findings were made in cultures of mammalian cells transformed by other RNA viruses and by DNA viruses or chemicals (Ossowski et al., 1973b). These findings are complemented by reports of the loss of a cell surface antigen with transformation of fibroblasts by RSV (Vaheri and Ruoslahti, 1974; Wartiovaara et al., 1974) and the decrease of cell surface fibronectins on tumor cells and of cell lines transformed by oncogenic viruses or carcinogens (Yamada and Olden, 1978). The serum zymogen was identified as plasminogen, and the cell factor was identified as a highly specific serine protease that activates plasminogen (Ossowski et al., 1973a). Both the transformed growth behavior of RSV-transformed cells as represented by colony formation in agar and their morphological appearance were closely associated with the onset of their fibrinolytic activity. TPA induces production of the serine protease plasminogen activator by normal and RSV-transformed chick embryo fibroblasts (Quigley, 1979). It is of interest that trypsin inhibitors that inhibit the lysis of fibrin by RSV-transformed chick embryo cells (Unkeless et al., 1973) also restore their morphology and adhesiveness to normal (Weber, 1975). However, the serine protease can catalytically alter cellular behavior in culture independent of its activation of plasminogen (Quigley, 1979). This indicates that the highly specific plasminogen activator is a reporter for the release of other proteases that degrade surface proteins. This point will be considered later when the large number of matrix metalloproteinases (MMPs) discovered in recent years is discussed. To recapitulate some major features of the last two sections, infection of the chick embryo with RSV at an early stage of development results in active reproduction of the virus and occasional production of hemorrhagic lesions (Milford and Duran-Reynals, 1943). The infected but nontransformed embryonic cells also express an active src-specific protein that underlies the transformation of cells in young and adult chickens (Dolberg and Bissell, 1984). When the infected cells are placed in culture, they are capable of expressing the transformed phenotype after a 24-hour delay (Dolberg and Bissell, 1984). The transformation may result from disorganization of the normal tissue architecture of the restraining environment, abetted by the inclusion of tryptose phosphate broth in the medium (Temin and Rubin, 1958). Transfection of the embryo with v-src oncogene obtained from RSV also fails to transform mesenchymal and epithelial tissues of the embryo, although some scattered epithelioid ‘‘neoplastic’’ cells were reported in the primitive endothelium (Stoker et al., 1990). The other cell types retain their normal differentiated function concurrently with expression of the oncogene. Hence, RSV-infected fibroblasts assume the

Microenvironmental Regulation of the Initiated Cell

27

characteristics of initiated cells in culture with FBS or high concentrations of CS, and in most tissues of the embryo. On a related note, Rous found soon after he demonstrated a causal virus in chicken tumors that local injury acted as a strong adjuvant in producing tumors when the relatively lowpotency RSV of those early days was injected into chickens (Rous et al., 1912). Inclusion of diatomaceous earth in the inoculum increased injury and tumor formation. Injury produced by the injection needle alone is sufficient for tumor production by high-potency RSV, but those tumors remain localized, even though the virus they produce is carried throughout the body in the circulation (Dolberg et al., 1985). A tumor can be induced at a distant site, however, if a wound is inflicted there or if a tumor promoter is administered in high enough concentrations to injure the tissue (Dolberg et al., 1985). TGF- is produced in the wound during inoculation of RSV and can be added to a minimal wound to induce tumors, whereas other growth factors are ineffective (Sieweke et al., 1990). Hence, there are a number of parallels between initiation and promotion in RSV infection of fibroblasts and chemical carcinogenesis of epidermis. In both cases, the nonneoplastic state of the initiated cell may be the result of the protective influence of the microenvironment, and the activation of promotion may represent persistent disruptive damage to that microenvironment. The results are consistent with inhibition of proteolytic activity at the surface of the RSV-infected cells by proteins disproportionately represented in FBS—with a minor contribution from proteins released by normal cells— that suppressed the transformed appearance and unregulated growth of the virus-producing cells.

B. Mammalian Fibroblast Inhibition of Hamster Fibroblasts Transformed by Polyoma Virus A clone of transformed hamster fibroblasts was isolated from an established line of hamster cells that had been infected by polyoma virus (Stoker, 1964). The transformed clone exhibited the random orientation and unrestricted growth characteristic of many mammalian tumor cells in culture when they are in contact with one another in discrete colonies on the surface of a bare culture dish, or when growing amid a low density of normal fibroblasts. They made no visible colonies or foci, however, when seeded on a contact-inhibited confluent layer of the normal mouse or hamster fibroblasts, either living or sterilized by x-irradiation. A confluent layer of x-irradiated polyoma-transformed hamster cells also prevented transformed colony formation, which is surprising because colonies of the living

28

Harry Rubin

polyoma-infected cells on the bare surface of the culture dish remained transformed although the cells were in continuous contact with one another throughout their growth into a colony, that is, contact with sterilized transformed cells was inhibitory to living transformed cells, but contact of the living transformed cells with each other was not. Seeding the transformed cells on a dish half covered with a confluent layer of normal fibroblasts resulted in transformed colonies only on the bare portion of the dish, which led to the conclusion that suppression required direct contact between the transformed cells and the contact-inhibited dense layer of normal cells. This conclusion was supported by a weakening of suppression when a thin layer (0.6 mm) of agar was interposed between the transformed and confluent normal cells. Neither of these findings rules out the possibility that the suppression was the result of high concentrations of material secreted at the surface of the cells and effective only at very short distances from the secreting cells. Polyoma-transformed cells marked with carbon took up the orientation of normal cells when seeded on a confluent layer of mouse embryo cells and failed to multiply there. When the number of transformed cells on the confluent mouse embryo cells was determined by trypsinizing and replating the cells at different times for colony formation on an empty dish, there was a population average of only a single doubling of the polyoma cells during their residence on the confluent sheet. Further experiments showed that confluence per se of the mouse embryo cells was not sufficient to inhibit growth of the polyoma-transformed cells, but proliferation of the confluent cells had to cease for inhibition of the contacting transformed cells to occur (Stoker et al., 1966). The inhibition by irradiated polyoma-transformed cells was unexplained, as was the fact that a confluent sheet of living uninfected cells of the established hamster fibroblast line killed polyoma cells seeded on top of them. The growth inhibition of the highly transformed polyoma-infected transformed cells by normal cells differed from that of the lack of inhibition of carcinomatous keratinocytes by excess normal keratinocytes (Hennings et al., 1990; Strickland et al., 1992). A significant feature of the inhibited growth of the transformed hamster cells is that certain substances, such as nucleotides, can pass directly from a dense sheet of normal mouse embryo fibroblasts to polyoma-transformed hamster cells (Stoker, 1967). This raises the possibility, but does not prove, that molecules concerned in growth regulation are transferred directly between contiguous cells. There remains the caveat that components of the extracellular matrix that are active only at short range are responsible for the inhibition of the transformed hamster cells.

Microenvironmental Regulation of the Initiated Cell

29

C. Influence of Normal Mammalian and Avian Fibroblasts on Proliferation of Polyoma Virus- and RSV-Infected Chicken and Mammalian Fibroblasts as Well As Nonviral Mouse Sarcoma Cells Transformed chicken cells and a variety of mammalian tumor cells were each seeded on low and high densities of each of a variety of normal avian and mammalian cells (Weiss, 1970a). The transformed cells included chicken and rat fibroblasts transformed by RSV, the latter producing no infectious virus; hamster fibroblasts transformed by polyoma virus; and cells from a mouse sarcoma maintained in vitro as a cell line. The normal fibroblasts were from the embryos of chickens, quail, geese, mice, and rats. The growth of RSV-transformed cells was not strongly inhibited by dense sheets of any of the normal cells. This result differs from the report of complete inhibition of RSV-transformed cells on 2–3-day-old dense sheets of chick embryo cells (Rubin, 1960a). The discrepancy can be explained by the use in the diverse cell experiment of 5% CS (Weiss, 1970a) which is optimal for focus formation by RSV-infected chicken cells (Rubin, 1960b), in contrast to the use in the latter case of 9.2% CS (plus 1.5% chicken serum), which is apparently inhibitory to focus formation by RSV-transformed chicken cells seeded on a confluent layer of normal chicken cells (Rubin, 1960b). The RSV-transformed rat fibroblasts multiplied exponentially on sparse and dense sheets of mouse embryo fibroblasts, although their growth rate was slowed about 25% on the dense sheets. The polyoma-transformed cells and sarcoma 180 were strongly inhibited by crowded mouse and rat cells, but not by any of the crowded avian cells. Seeding high densities of normal mouse embryo cells suspended in agar had no inhibitory effect on polyoma-transformed hamster cells or mouse sarcoma 180 cells seeded in suspension alongside, but not in contact with, the normal mouse cells, reinforcing the conclusion that contact was necessary for the growth inhibition. It seems evident, therefore, that RSV is a more powerful transforming agent than polyoma virus in mammalian cells. Another aspect of the work was that dense sheets of mouse embryo fibroblasts did moderately inhibit focus formation in chick embryo fibroblasts freshly infected with RSV. Further studies exhibited about a 2.5-fold inhibition of this early stage of infection (Weiss, 1970b), which was consistent with earlier evidence that cell division, and DNA synthesis in particular, is necessary to establish infection with RSV (Rubin and Temin, 1959; Temin, 1967). This inhibition is entirely different than the total inhibition of typical transformed foci by FBS or high concentrations of CS, which allow the establishment of infection in pure cultures of chick embryo cells while they are multiplying but prevent their transformation (Rubin, 1960b). The

30

Harry Rubin

inhibition by quiescent mouse embryo sheets blocks the formation of the RSV provirus in some of the adjoining chicken cells, but the effect of the serum treatment in low-multiplicity infection of chicken cells is to allow establishment of infection and virus production but block the transformation and impose contact inhibition by the surrounding noninfected cells.

D. Chemical Carcinogenesis of Mammalian Fibroblasts The C3H 10T1/2 established line of mouse fibroblasts has been widely used as a target of transformation by chemical and physical carcinogens (Kennedy et al., 1980; Reznikoff et al., 1973). The cells in the transformation assay are usually seeded at a low density (1,000 per 60-mm dish) in 10% FBS, treated with chemical carcinogen, washed, and incubated in the same serum concentration for 5 or more weeks to establish a small number of transformed foci per culture. Switching the cells to 5% FBS at 8 days increases the number of foci two- to threefold (Bertram, 1977). In contrast, raising the FBS concentration to 15%–20% virtually abolishes the foci, although the number can be fully restored by lowering the serum to 5% 1 week or more before fixation and staining. The high concentrations of FBS also inhibited formation of colonies when the transformed cells were suspended in agar along with normal cells. The inhibition by high FBS concentration of transformed focus formation in this mammalian cell line after treatment with a chemical carcinogen bears several distinct similarities to the serum suppression of transformation observed in chick embryo cells infected with RSV (Rubin, 1960b). FBS was used in both cases, although a quantitative comparison with CS was included in the experiments with RSV-infected chick embryo cells. The range of serum concentrations of 2.5%–20% was the same, as was the concentration of 5% FBS or CS used in reversal. Reversal was measured in the chemically transformed mouse cells by counting foci in the original culture dish, but the RSV-infected chick embryo cells were trypsinized after the incubation in different concentrations and assayed for the number of focus-forming cells in a regular assay in 5% CS. The replating of the chick embryo cells allowed determination of the growth rate of RSVinfected cells throughout the experiment. No foci of altered cells were recorded in the chemically treated mouse cells in high serum concentrations, in contrast to the faint monolayered foci with a slightly increased population density noted in the RSV-infected chick cells maintained in high serum concentrations. The reseeding of the nontransformed RSV-infected chick embryo cells in a standard assay for focus formation in 5% CS allowed measurement of their growth rate and displayed their sensitivity to contact inhibition.

Microenvironmental Regulation of the Initiated Cell

31

It was noted that high concentrations of serum increased the saturation density of the normal-behaving mouse cells but had no effect on saturation density of the chemically transformed cells cultured separately, which reached their maximum saturation density in FBS as low as 2.5%. It was suggested that the serum-mediated growth inhibition of transformed cells in mixed cultures was the result of increased cell communication supposedly associated with increased crowding (Bertram, 1979). However, there was no increase in saturation density with increases in one of the 2 FBS samples beyond 5%, although there was moderate decrease in transformed foci with 10% FBS and a complete disappearance of foci in 15% FBS (Bertram, 1977). An explanation more in line with the results obtained in RSVtransformed cells is that the increased FBS concentration suppressed mouse fibroblast transformation by providing more of the multiple inhibitors in FBS for the diverse proteases at the cell surface that are involved in transforming the cells. A study was undertaken of the role of cyclic nucleotides in the suppression of transformed cell growth by normal cells. It was found that inhibitors of phosphodiesterase, which increased the concentration of cyclic AMP in the cells, inhibited the expression of transformation when added to cultures in 5% FBS 7 days after removal of carcinogen (Bertram, 1979). Adding cyclic AMP or its dibutyryl derivative potentiated the inhibition. Contact with nontransformed cells was required for the inhibition. The phosphodiesterase inhibitors also inhibited focus formation by established transformed cells seeded on a confluent layer of nontransformed cells in 5% FBS, but did not inhibit their growth when the transformed cells were seeded alone. It was assumed that the inhibitor stopped the multiplication of transformed cells in the presence of contact with nontransformed cells, but no attempt was made to rule out the possibility that the appearance and growth behavior of the transformed cells was normalized, as was the case of RSV-infected chick cells among normal chick cells in high serum concentration. It may be highly significant that an increase in cyclic AMP decreases the concentration of a key protease associated with transformation (Wilson and Reich, 1979). The relationship between inhibition of transformed cell growth by confluent nontransformed cells and junctional communication between them was studied in a wide variety of transformed fibroblasts and several lines of nontransformed cells (Mehta et al., 1986). Where heterologous junctional communication between the transformed and confluent normal cells was weak or nonexistent, there was no inhibition of colony formation of the transformed by the normal cells. Agents that increased cyclic AMP phosphorylation increased heterologous communication and decreased the size of the transformed colonies. The same agents had no effect on colony size when the transformed colonies were growing alone. Where heterologous communication was strong, the size of the transformed colonies was small but was

32

Harry Rubin

increased when communication was blocked by retinol or retinoic acid. It was concluded that cell-to-cell membrane channels between the transformed and normal cells are the likely conduits of the signals for this growth control. Junctional communication was measured in these experiments by transfer of the 443-da fluorescent dye Lucifer yellow, and it appeared that cyclic AMP was the regulatory molecule. As noted earlier, however, in another set of experiments with rat cells transformed by five different oncogenes, total inhibition of fluorescent dye transfer by an inhibition of gap-junctional communication did not prevent growth inhibition of transformed rat cells by an excess of non transformed rat cells (Martin et al., 1991). In a study of seven fibroblastoid and seven epithelial mammalian cell lines of normal or malignant origin, no relation was found between ionic coupling and tumor production (Hu¨lser and Webb, 1973). Ionic coupling was found only in the fibroblastoid lines and not in the epithelial lines; it was independent of tumor-forming capacity and was the same in medium with CS or FBS. These results, along with the absence of dye transfer between normal keratinocytes and the papilloma cells whose growth they suppressed (Strickland et al., 1992), argue against a primary role for junctional communication in neoplastic transformation or its suppression.

E. Microenvironmental Regulation of UV-Initiated, TPA-Promoted Mouse Fibroblasts Cells of the C3H 10T1/2 established line of mouse fibroblasts were exposed to either a single treatment with UV light, to continuous treatment with TPA, or to UV followed by TPA (Herschman and Brankow, 1986). Transformed foci were rarely seen in untreated cultures or in UV- or TPAtreated cultures. Those exposed first to UV irradiation and then to continuous TPA developed foci after 4–6 weeks. Ten foci each unavoidably mixed with surrounding normal cells were isolated from confluent sheets and grown to saturation density, where the previously transformed cells failed to produce foci. It was thought possible that the transformation of cells isolated from the foci was inhibited by the accompanying normal cells and that focus formation might be dependent on the continued presence of TPA. The combination of 200 cells from the 10 isolated populations with 1,800 cells from untreated C3H 10T1/2 cells produced no foci, but the addition of TPA to the combinations produced foci in three of 10 isolates, indicating that they contained initiated cells that required promotion for their expression. Clonal lines were established from the three positive populations, and these pure cultures expressed full transformation when grown to confluence

Microenvironmental Regulation of the Initiated Cell

33

both in the presence of and in the absence of TPA. A threefold excess of untreated cells fully prevented focus formation by the initiated populations, but treatment with TPA lifted the suppression even with the highest (ninefold) excess originally tested. There was partial suppression of focus formation even when the initiated cells themselves were in threefold excess over the normal cells, but this took the form of a reduction in size, but not number, of the foci. The addition of TPA increased the size of these foci. The results showed that a threefold excess of normal cells could completely block transformation by the initiated cells. Many independent isolates were made of foci produced by UV plus TPA, and all of these could be suppressed with an excess of untreated cells. All the suppressed mixtures could be promoted to focus formation with relatively low doses of TPA. The two-stage procedure of UV followed by TPA regularly produces cells that have the classical phenotype of initiated cells. TPA does not increase the size of the foci in a pure culture of initiated cells, but it does so in suppressed mixtures, which indicates that the action of TPA is to lift the suppression of initiated fibroblasts by untreated C3H 10T1/2 fibroblasts. In contrast to the UV-TPA-treated cells, a transformed clone derived from a culture treated with high doses of methylcholanthrene (MCA) produced foci without TPA despite the presence of a large excess of untreated C3H 10T1/2 cells, although TPA slightly increased the size of the foci. It is noteworthy that the MCA-transformed cells produced foci when grown with an excess of normal cells in the absence of TPA with twice the efficiency of the UV-TPA cells treated with TPA, indicating that the MCA-transformed cells had progressed beyond initiation. When the number of initiated cells seeded was kept constant (250 cells per 60-mm dish) and the number of untreated C3H 10T1/2 cells exceeded 2,000, the number and size of foci produced by adding TPA decreased with increasing numbers of untreated cells, and the countable foci reached zero with 32,000 untreated cells (Herschman and Brankow, 1987). Because there is a decrease in the number of divisions the initiated cells undergo to reach confluence with increases in the total number of cells seeded, the results indicated that the ability of TPA to reverse the growth inhibition imposed by the untreated cells depends on the colony size of the initiated cells at confluence. In the foregoing experiments, TPA was added on the day after seeding the mixed cultures and was maintained during medium changes throughout the experiment. TPA was therefore present during the entire growth and confluence periods of the mixed cultures. If TPA was added at various times after seeding the mixed cultures, the number and size of the foci decreased in the later additions of TPA as growth slowed down and the cultures approached confluence (Herschman and Brankow, 1987). If TPA was added 2 days after the number of cells reached a maximum, no foci were seen. The data

34

Harry Rubin

indicate that TPA is effective at eliciting full-focus formation in mixed cultures of the UV-initiated and untreated C3H 10T1/2 cells only if the initiated colonies at confluence are of adequate size and TPA is added during the growth period of the culture. A model was proposed in which the untreated C3H 10T1/2 cells generate and receive signals required for contact inhibition (Herschman and Brankow, 1987). It was suggested that the UV-initiated cells can receive but not generate the regulatory signal. Pure cultures of the UV-initiated cells in which no regulatory signal is generated would exhibit the transformed phenotype in which growth is not regulated at confluence. The situation has some striking similarities to that found in RSV-infected chick embryo cells in FBS or in high CS concentrations, where cultures consisting predominantly of RSV-infected fibroblasts are transformed, but no transformation is seen when they are surrounded by normal fibroblasts. Medium conditioned by residence for several days in high densities of normal chick fibroblasts enhances the growth of small numbers of the cells and suppresses focus formation in low concentrations of CS on mixed cultures. In contrast, undiluted medium conditioned by high densities of RSV-transformed cells is actually toxic to the growth of small numbers of cells, which is consistent with the poor growth of RSV-transformed chicken cells unless surrounded by normal fibroblasts. It would be of interest to determine whether the focus-suppressing effect of medium containing FBS or high concentrations of CS would be inactivated if conditioned for several days on pure cultures of RSV-transformed cells. It was proposed that TPA might overcome the focus-suppressing effect of untreated C3H 10T1/2 cells on UV-initiated cells by reducing the regulatory signal of the untreated cells (Herschman and Brankow, 1987). Because junctional communication was thought to mediate the suppression of a line of MCA-initiated C3H 10T1/2 cells by untreated cells (Mehta et al., 1986), and TPA blocks junctional communication in some cells (Yotti et al., 1979), the promotion of focus formation by TPA in mixed cultures might result from such a block. However, TPA does not block dye transfer in C3H 10T1/2 cells (Boreiko et al., 1987), which seems to rule out that possibility. An alternative consistent with the role of plasminogen activator in the expression of transformed growth behavior and morphology by RSVinfected fibroblasts (Ossowski et al., 1973a) is that TPA induces a protease in the C3H 10T1/2 mouse cells, as it does for plasminogen in chick, hamster, and rat embryo cells and human cervical carcinoma cells (Wigler and Weinstein, 1976). Of particular significance for the role of proteases in the classical skin model of initiation and promotion is the finding that tumorigenesis in mouse skin is inhibited by synthetic protease inhibitors (Hozumi et al., 1972; Troll et al., 1970). The protease activity of mouse skin was markedly increased by the application of promoters. The increase was

Microenvironmental Regulation of the Initiated Cell

35

prevented by the addition of the protease inhibitors, which also reduced leukocyte invasion of the tissue. Leukocyte infiltration probably contributes to the increase in protease activity and to promotion of tumorigenesis. These results therefore support a major role for proteases in promotion of the UV-TPA model in C3H 10T1/2 cells and in chemical carcinogenesis of mouse skin.

VIII. EVALUATION OF THE EVIDENCE CONCERNING A ROLE FOR PROTEASES IN TUMOR INITIATION The evidence for involvement of surface proteases in expression of the transformed phenotype recalled experiments that suggested a role for intracellular proteases at the initiation step. These experiments followed evidence that a high-frequency event is involved in the induction of chemically and radiation-induced transformation in cultured mouse cells (Fernandez et al., 1980; Kennedy et al., 1980; Mondal and Heidelberger, 1970). The protease inhibitor antipain added for only a single day following x-irradiation of C3H 10T1/2 mouse cells ‘‘completely suppressed’’ radiation transformation and indicated that a DNA repair process might be important in its action (Kennedy and Little, 1981). It should be noted, however, that the frequency of foci in the irradiated cultures not treated with antipain averaged less than one-half per culture, necessitating the use of many cultures to obtain significant results. Although unirradiated cultures not treated with antipain, that is, spontaneous transformants, had no foci, later studies by others using cells of the same origin but far larger numbers of cultures did find an average of about 0.1 spontaneously transformed focus per culture (Grisham et al., 1988). Because there were in fact occasional spontaneous transformants among the antipain-treated irradiated cultures (Kennedy and Little, 1981), it cannot be said the suppression was complete, but it appeared to be at a significant level because it was repeated in subsequent experiments (Kennedy, 1982). The suppression seemed to be effective even when antipain was added as late as 10 days and 13 cell divisions postirradiation (Kennedy, 1985). TPA treatment following the antipain exposure of x-irradiated cells did not lead to promotion, indicating that the protease inhibitor irreversibly reverted the initiated cells to their original uninitiated condition. Caution is advised in accepting this interpretation of the results unreservedly because later work by others concluded that carcinogen-induced transformation of these cells could not be significantly distinguished from spontaneous transformation when starting with small numbers of cells per culture (Grisham et al., 1988), such as those used in

36

Harry Rubin

the protease inhibitor experiments (Kennedy, 1982, 1985; Kennedy and Little, 1981). In addition, it had earlier been reported that antipain and other protease inhibitors inhibited chemical transformation of the same C3H 10T1/2 cells, but 5–6 weeks of treatment were required and the effect was reversible (Kuroki and Drevon, 1979). The requirement in this case for long-term treatment and its reversibility indicated that the observed effect was mainly on promotion rather than initiation. These results were obtained starting with cell numbers per culture that were large enough to allow a statistically significant distinction from spontaneous transformation (Grisham et al., 1988). Reversibility of the suppression of RSVtransformation of CEF and chemical transformation of mouse cells by FBS (Bertram, 1977; Rubin, 1960b) are further indications that protease inhibitors act on expression of the neoplastic phenotype. The effect of antipain on mutagenesis, chromosome aberrations, sister chromatid exchanges, and cell killing was studied in V79 Chinese hamster cells (Kinsella and Radman, 1980). The cells were treated with N-methylN1nitro-N-nitrosoquanidine (MNNG), a powerful mutagenic carcinogen and one of the first chemicals shown to transform human cells (Kakunaga, 1977). It induced a variety of chromosome aberrations in the V79 hamster cells, including breaks, tri- and quadriradials, and chromosome exchanges that were 10 times more frequent among nonhomologs than among homologs (Kinsella and Radman, 1980). The chromosome aberrations were markedly reduced in the presence of antipain during exposure of the cells to MNNG, whereas none of the other measured parameters was affected. The results indicate that MNNG-induced lesions cause chromosome rearrangements through an antipain-dependent process, presumably cellular proteases. It was suggested that chromosome aberrations represent a ratelimiting step in carcinogenesis, whereas mutagenesis, if required, is not sufficient to accomplish carcinogenesis. Chromosome aberrations could be the basis for both initiation, as in deletion mutagenesis, and promotion through chromosomal rearrangements, causing segregation to homozygosity or hemizygosity (Kinsella and Radman, 1980). A continuing nonclonal instability that persists for many cell generations occurs after x-irradiation, is characterized by chromosome aberrations that are associated with transformation (Kadhim et al., 1995; Marder and Morgan, 1993), and might be inhibited by protease inhibitors. A high proportion of cells that survive x-irradiation have a heritably reduced rate of growth (Nias et al., 1965; Sinclair, 1964). The dose of xirradiation to produce chromosome aberrations in mammalian cells is seven to eight orders of magnitude lower than the dose to produce mutations in bacteria and considerably less than the killing dose for the mammalian cells (Puck, 1995). There is evidence that the phenomenon of delayed reproductive death in cells surviving x-irradiation, which includes the slow-growth

Microenvironmental Regulation of the Initiated Cell

37

phenotype, is initiated by double-strand breaks in DNA (Chang and Little, 1992). The slow-growth phenotype is also induced by the mutagenic alkylating agent ethylmethane sulfonate, indicating double-strand breaks that would underlie chromosome exchanges. Other treatments that produce the slow-growth phenotype are methotrexate (Chow et al., 1998) and long-term confluence that does not directly damage DNA but that contributes to and selects for neoplastic transformation (Chow and Rubin, 2000; Rubin et al., 1995). Persistent genetic instability and transformation are produced in culture by other exposures that do not directly damage DNA, such as heat treatment and serum starvation (Boukamp et al., 1999; Hill et al., 1991; Li et al., 2001). The transmissible chromosome instability associated with the slowgrowth phenotype and inhibited by protease inhibitors provides a plausible foundation for understanding the reversal of the initiated phenotype by protease inhibitors as late as 13 generations after x-irradiation (Kennedy, 1985). It is noteworthy that high frequencies of initiation (1.5%–4.0%) have been found for in vitro carcinogenesis of the trachea and in vivo carcinogenesis of the thyroid and breast (Kennedy, 1994). Such high frequencies of carcinogenesis could be the result of an ongoing process similar to that described for x-irradiated C3H 10T1/2 cells and could be susceptible to inhibition by protease inhibitors. However, the reservations raised above about accepting the evidence for irreversible abrogation of the initiation process by brief application of protease inhibitors are valid until the results are confirmed in other laboratories.

IX. VISUALIZATION OF MICROENVIRONMENTAL REGULATION IN SPONTANEOUS TRANSFORMATION OF MOUSE FIBROBLASTS The NIH 3T3 line of mouse embryo fibroblasts was established by repeated selection of flat clones that had a high plating efficiency and a low saturation density (Jainchill et al., 1969). The cells are distinguished by the ease with which they undergo spontaneous neoplastic transformation when kept under the selective condition of confluence for more than 2 weeks (Rubin and Xu, 1989). This capacity for spontaneous transformation is reduced by repeated passages of the cells at low population density in high (10%) CS concentration but is increased by low density passage in selectively low concentrations (2%) of CS (Yao et al., 1990) or in 10% FBS, which is less stimulatory for these cells than an equal concentration of CS (Rubin, 1992). Because of the predilection of the originally established

38

Harry Rubin

NIH 3T3 cell line for spontaneous transformation, it can be considered an initiated line, with the caveat that maintenance of that state can be reversed by counter selection under conditions that maximize proliferation. The number of transformed cells in a heavily transformed culture is assayed by diluting the cells and mixing them with a large excess of untransformed cells, which form an orderly monolayered background for the display of the multilayered transformed foci. It was found (Rubin, 1994a) that most sublines of untransformed NIH 3T3 cells allow the continuous enlargement of foci with time when mixed with transformed cells, as seen with backgrounds of the 27M and 28M sublines (Fig. 3). In contrast,

Fig. 3 Disappearance of early appearing foci from spontaneously transformed NIH 3T3 mouse fibroblasts assayed on a suppressive background, and continuous expansion of the foci on a permissive background. 1,000 spontaneously transformed cells were seeded with 105 untransformed cells of the suppressive 173c subline or the permissive 27M or 28M sublines. Cultures were fixed and stained at 8, 10, and 14 days. After Rubin, 1994.

Microenvironmental Regulation of the Initiated Cell

39

however, the 173c subline, originating from the same frozen stock of NIH 3T3 cells as the permissive sublines, allowed early focus development by transformed cells up to 8 days, but the foci were regressing at 10 days and, except for a few small, dense foci, had almost entirely disappeared at 14 days. The results indicated that once the 173c background reached its maximum density at about 8 days, it could normalize cells in the small foci that had developed to that time. The 173c subline, when cultured alone, formed the same type of smooth monolayer as the permissive 27M and 28M sublines; it differed from those sublines, however, in the larger number of low-density passages undergone since it was thawed from the frozen stock, plus the fact that it had been refrozen for storage twice during the course of its passage and was itself somewhat more refractory to spontaneous transformation in repeated rounds of confluence. One of several possible explanations for its suppressive effect is that the 173c subline secreted stronger protease inhibitors than the permissive sublines. It was also noticed that the untransformed cells that formed the background for focus development by minority cells in the same culture became more permissive for foci in successive rounds of confluence (Chow and Rubin, 1999). This is apparent in a comparison of the large size of foci in the second round of confluence with their reduced size in the third round, when the cells were diluted with an excess of cells of the same subline that had not previously experienced selection at confluence (Fig. 4). This indicated that the same selective process that enhanced progression to focus formation in some cells of a population also selected for increased permissiveness of the other cells, which formed the confluent background for focus formation. In another experiment, it was found that progression toward focus formation occurred more quickly when the NIH 3T3 cells were kept at confluence in 10% CS than in 2% CS (Rubin, 1994b). This apparently occurred because the cells grew to five times higher density in 10% CS than in 2% CS, with increased opportunity for selection at confluence. Large foci were formed in the third round of confluence by two parallel lineages that had been held at confluence for 3 weeks in 10% CS in the first round, but the foci were greatly reduced in density and size if the cells were diluted in an excess of cells that had never been selected at confluence (Fig. 5). In fact, the light foci formed in the undiluted third assay of lineage 3 almost entirely disappeared when the responsible cells were diluted and surrounded by unselected cells. This experiment gave further evidence that the same selective procedure that drove progress to focus formation also increased the permissiveness of the non-focus- or weak-focus-forming background. The overall results indicated that the higher the saturation density of the NIH 3T3 cells that formed the microenvironmental background, the more permissive it is for focus formation by cells in the same population that had progressed the

40

Harry Rubin

Fig. 4 Serial assays for focus formation during spontaneous transformation of an easily transformed clone of the NIH 3T3 mouse fibroblast line. In the first assay, 105 cells were seeded in 2% CS medium. At 2 weeks, one culture was fixed and stained. Another was suspended for a second assay with 105 cells, and the procedure was repeated. Because there were large numbers of transformed cells in the very large foci of the second assay, only 500 cells from that assay were mixed with 105 standard, low-density passage cells to produce a countable number of foci in the third assay. Note two tiny foci of newly transforming cells in the first assays, the very large foci that ensued from them in the second assay against a background of untransformed cells from the same culture, and the greatly decreased size of irregularly shaped foci in the third assay, in which they are held in check by a confluent background of more suppressive untransformed cells from the standard, low-density passage. After Chow and Rubin, 1999.

farthest toward transformation. The possible role of proteases and protease inhibitors in spontaneous transformation has not been investigated.

X. MICROENVIRONMENTAL EFFECTS ON MAMMARY NEOPLASIA A. In Vivo Results Mammary tumor development in mice has commonly been used as an experimental model for human breast cancer. Viral, hormonal, and genetic factors interact in mouse mammary tumorigenesis (DeOme, 1965; Foulds, 1965). Mice infected with and expressing mammary tumor virus (MTV) develop a higher incidence of mammary tumors than mice in which the virus is unexpressed. In the normal course of mammary gland development, elongated ducts grow out in young virgins and branch dichotomously in fat pads, led by terminal buds (Faulkin and DeOme, 1960). During pregnancy, these proliferate to form lobuloalveolar (LA) structures, which return to ductal

Microenvironmental Regulation of the Initiated Cell

41

Fig. 5 Partial suppression of foci by spontaneously transformed mouse cells within the same third assay by plating them with an excess of untransformed cells from the standard, lowdensity passage. Two parallel lineages of cells, LN2 and LN3, went through first, second, and third serial assays with 105 cells, but the third assay was supplemented with a parallel assay by mixing 104 cells with 105 standard, low-density passage cells. As expected, the third assay of 104 cells gave 10-fold fewer foci than the third assay of 105 cells, but they were also smaller or lighter by virtue of a less-permissive cellular background that had not gone through a prior selection for growth at high density. Supplement to Rubin, 1994.

form between pregnancies. There are regulatory interactions between the buds that maintain the ducts at a constant distance from one another (Faulkin and DeOme, 1960). Mammary tumor development in MTV-expressing mice is preceded by the appearance of lobuloalveolar structures, which increase in number to form hyperplastic alveolar nodules (HAN). The HAN resemble the LA structures of pregnancy but occur in virgin as well as parous females and do not regress between pregnancies. They can be maintained by serial transplantation in mammary fat pads that have been freed of mammary glands (gland-free fat pads) in which epithelial cells proliferate to a much greater extent than in gland-intact fat pads (Table IV) (Faulkin and DeOme, 1960). Some of the HAN in the gland-free pads develop into tumors if given enough time, but those in the intact fat pads do not. If two HAN are placed in gland-free fat pads, their outgrowths approach each other, but they never intermingle. A similar growth-inhibitory effect between normal mammary gland elements is seen when several are transplanted into gland-free fat pads. There was limited growth regulation of tumors by normal duct

42 Table IV

Harry Rubin Growth of nodules (HAN) in gland-free versus intact mammary fat pads of mice

Host age

Fat pad

Time, weeks

No. takes

Percentage filling of fat pads

No. tumors

7 months

Gland-free intact

12

18

39 30

6

3 weeks

Gland-free intact

12 8

17 18

1 25 34

0 1

8

17

9 14

0

Abstracted from Faulkin and DeOme, 1960, Table 1.

elements. The results showed growth regulation between normal glands, between normal gland and HAN, between HAN, and to a limited extent between normal duct elements and tumors. Several aspects of growth regulation in tumor development came under further exploration. These included the role of tissue organization and both the effect of dissociated normal cells on HAN development into tumors in MTV-expressed mice (Balb/c f C3H) and the effect of cell population size on carcinogen-induced ductal dysplasia in MTV-unexpressed mice (Balb/c). Nodule transformed cells were recovered from 11%–18% of outgrowths when dissociated cells from 2–4-month-old virgin mice were injected into gland-free fat pads, whereas no nodular transformation was seen when intact pieces of mammary tissue from mice of about the same age were injected (DeOme et al., 1978). Nodular growths did not appear naturally in uninjected controls until they were about 9 months old. Hence, disruption of the normal architecture of the mammary gland greatly accelerated the appearance of HAN-like LA outgrowths. The injection of dissociated cells unmasked the presence of otherwise inapparent cells capable of nodule formation in MTV-expressing mice long before nodular growth could be detected in undisturbed tissue. The normal proximity relations of cells within the intact tissue therefore greatly delayed nodule-related changes, generally considered precursors to tumor formation. The question arose whether enzymatic dissociation of cells from transplanted nodule lines of MTV-expressing mice would enhance their development into tumors (Medina et al., 1978). Dissociation significantly increased the tumorigenicity of three different nodule lines compared with their transplantation as 1-mm3 pieces (Table V). The addition of large numbers of normal mammary cells from virgin, pregnant, or lactating mice decreased the tumorigenicity of the dissociated cells (Table VI) to a level equal to or less than that of the 1-mm3 pieces (Table V). Normal liver epithelial cells had no effect on the tumorigenic potential of the nodule cells. The growth of mammary tumor cells was not inhibited when mixed with an excess of

43

Microenvironmental Regulation of the Initiated Cell

Table V Tumor potential of enzymatically dissociated versus intact 1-mm3 pieces of mammary nodule lines Nodule line and injected material

Total tumors/ Total transplants

Percentage tumors

41/84 21/109

49 19

57/81 62/115

81 54

25/28 18/30

89 53

D1 Dissociated Intact D2 Dissociated Intact C4 Dissociated Intact Combined from Medina, 1978, Table 3.

Table VI

Effect of dissociated normal cells on tumor potential of dissociated nodule lines D1

and D2 Type of normal tissue D1 — Lactating D1 — Virgin D2 — Pregnant D2 — Lactating D2 — Virgin D2 — Liver

Ratio/Nodulea: Normal

Total tumors/ total transplants

Percentage with tumors

— 1:3

10/17 2/15

59 14

— 1:3

8/18 3/17

44 18

— 1:1

14/14 13/23

100 57

— 1:3

11/16 6/20

69 30

— 1:3

15/20 7/19

75 37

— 1:3

10/18 10/18

56 56

Abstracted from Medina, 1978, Table 5. a105 nodule cells were injected in each case. —, no normal tissue (only nodule cells).

44

Harry Rubin

normal mammary epithelial cells (not shown). The results showed, on the one hand, that dissociation of cells from HAN increases their tumorigenic potential just as it increases the capacity of normal cells to form nodules. This indicates the capacity of normal mammary tissue architecture to regulate the growth of previously expressed as well as unexpressed noduleforming cells. The results also showed more directly that normal cells inhibit the growth of nodule-forming but not tumorigenic cells. Mouse strains such as Balb/c that do not express MTV may have its genome integrated into their DNA, but they have a lower probability of developing HAN or mammary tumors than do mouse strains that express MTV. Treatment of Balb/c mice with whole-body -irradiation, or via gastric intubation of DMBA, increases the potential for expression of mammary ductal dysplasia in outgrowths derived from enzymatically dissociated mammary cells (Ethier and Ullrich, 1982). Serial transplantation in gland-free fat pads of mammary dysplasias induced by DMBA exhibit a high tumorigenic potential (Medina, 1979). Donor Balb/c mice were untreated or exposed to low doses of -irradiation or DMBA, their mammary epithelium was dissociated, and 105 or 104 enzymatically dissociated cells were injected into gland-free fat pads of syngeneic mice (Ethier and Ullrich, 1984). About 5% of the mice receiving 104 cells from untreated mice had outgrowths with ductal dysplasia at 10 weeks, and about twice as many of those with 104 carcinogen-treated cells displayed the same lesion (Table VII). Surprisingly, the injection of 105 cells was a less efficient inducer of ductal dysplasias at 10 weeks than 104 cells, regardless of treatment. The apparent reason for this inverse relation between cell number injected and lesions produced is that outgrowth from 105 cells had filled the fat pad at 10 weeks, resulting in cessation of their growth and disappearance of terminal buds. In contrast, fat pads injected with 104 cells had not yet been filled with outgrowth at 10 weeks and still had terminal buds, which are the structures that exhibit dysplasia. When the outgrowths were examined at 16 weeks, all of the fat pads were filled Table VII Frequency of ductal dysplasia in mammary outgrowths 10 weeks after injection of 105 and 104 cells Group

Cells injected

Percentage with ductal dysplasias

Controls Gamma-irradiation DMBA Controls Gamma-irradiation DMBA

105 105 105 104 104 104

0 3.2 0 5.3 11.5 11.9

DMBA, 7, 12-dimethyl benzanthracene. Abstracted from Ethier and Ullrich, 1984; Table 1.

45

Microenvironmental Regulation of the Initiated Cell

Table VIII Frequency of ductal dysplasia in second-generation outgrowths derived from injection of 104 cells derived from 16 week first-generation outgrowths Group and Time, weeks Controls 8 16 Gamma-irradiation 8 16 DMBA 8 16

Percentage with ductal dysplasia

5.0 0 30.8 ND 25.0 0

ND, not determined; DMBA, 7, 12-dimethylbenzanthracene. Abstracted from Ethier and Ullrich, 1984, Table 3.

regardless of the number of cells injected and there was no visible evidence of ductal dysplasia in any of the mice. When second-generation outgrowths were derived from the full, normallooking 16-week outgrowths, ductal dysplasias were reexpressed, but once again only within shorter-term still-developing outgrowths (Table VIII). This showed that a reproductively stable but phenotypically reversible alteration was present in some of the enzymatically dissociated cells. The results of these experiments indicated that the expression of abnormal ductal morphology in mammary epithelium of mice with unexpressed MTV depends on the proliferative state of the ducts, which is limited by the size of the fat pad. There is evidence that mammary tumor production in rats treated with certain carcinogens results from selection of preexisting mutations, possibly by increasing the permissiveness of the mammary microenvironment. Mutations in the Ha-ras 1 gene were found in small patches of mammary epithelium in young, untreated Fischer female rats (Cha et al., 1994, 1996). Injection of N-nitroso-N-methylurea (NMU) produced mammary tumors, more than 90% of which carried the specific preexisting Ha-ras 1 mutation, which was rarely seen in mammary tumors induced by DMBA. There was no increase in the number of ras-mutated patches in the NMUtreated rate, indicating that NMU acted by selection, possibly through increasing the permissiveness of the microenvironment.

B. In Vitro Results The previous section showed that disruption of the normal architecture of the mammary gland by enzymatic dissociation of the cells increases their capacity for neoplastic development when the cells are reinjected into the

46

Harry Rubin

gland-free mammary fat pad. It is common knowledge that cellular interactions within and between tissues are required during embryonic development. A standard piece of selected presumptive tissues from the embryo will undergo differentiation when placed in culture, but dividing that piece into many small but still multicellular fragments prevents differentiation (Grobstein and Zwilling, 1953). The fragments, however, retain the ability to differentiate, as they do so when they refuse into a single mass, which indicates the need for interaction between a minimum number, and perhaps types, of cells. Dissociating differentiated epithelial cells for monolayer culture usually results in their partial or total dedifferentiation (Davidson, 1964). This is certainly true for mouse mammary epithelium attached to a plastic or collagen surface, but they retain or restore their morphological and some of their biochemical differentiation on floating collagen gels. (Emerman and Pitelka, 1977; Emerman et al., 1979). The floating gel provides several unique features not found on a plastic substrate, including epithelial cell interaction with stromal elements, and substrate flexibility and contractility that permit cell shape change (Emerman et al., 1979). There is evidence that cells cultured on floating gels organize an endogenously synthesized basement membrane (Petersen et al., 1992; Streuli and Bissell, 1990). A reconstituted basement membrane was developed (Kleinman et al., 1986) that was more effective than the floating collagen gels in eliciting tissue-specific cellular morphology and increases in vectorial secretion and total amount of milk protein (Barcellos-Hoff et al., 1989). The reconstituted basement membrane was derived from the EnglebrethHolm-Swarm (EHS) mouse sarcoma, which lacks extracellular matrices other than basement membrane (Hassell et al., 1980). The reconstituted structures include laminin, type IV collagen, heparan sulfate, and proteoglycan, among other constituents (Kleinman et al., 1986). The expression of the milk protein -casein requires two components of the cellular response to basement membrane overlay (Roskelley et al., 1994). The first is rounding and clustering of the cells that can be physically mimicked by seeding the mammary cells on a nonadhesive substratum. The second is associated with 1 integrin clustering, which is itself morphology-independent but requires cell rounding and clustering for translation into a functional response. Because matrix structures produced by cultured fibroblasts provide substrates that are unique for fibroblasts (Kleinman et al., 1986), there is some question about the specificity of the EHS sarcoma matrix for mammary epithelium. This question is counterbalanced by the observation that the mammary cells form their own matrix in response to the EHS matrix (Petersen et al., 1992). All normal human breast tissue and mammary epithelial lines embedded as single cells in EHS matrix responded by growth regulation, polarized acinus formation with lumens, and deposition of endogenous collagen IV-

Microenvironmental Regulation of the Initiated Cell

47

containing basement membrane (Petersen et al., 1992). None of the carcinoma tissues or cell lines responded in any of these features to embedding in EHS matrix. Particularly striking was the absence of endogenous basement membrane formation in the malignant material. Studies were initiated with a normal-behaving cell line from a woman’s fibrocystic breast lesion and its malignant derivative obtained by the omission of epidermal growth factor (EGF) from the medium (Briand et al., 1987, 1996). Initially both lines were karyotypically identical, containing the same chromosome abnormalities, but the malignant line had three copies of a marker chromosome, 7p, which harbored the EGF receptor (Briand et al., 1996). A prominent feature of the malignant line was its failure to assemble adhesion molecules at the cell surface (Weaver et al., 1997). In addition, the acini formed by the nonmalignant mammary epithelial line had basally distributed integrins, consistent with their polarized phenotype, in contrast to the malignant derivative in which the integrins were randomly distributed and disorganized. The ratio of 1 to 4 integrins at the cell surface was 2.8-fold higher in the malignant than the nonmalignant line. Antibodies that blocked the function of 1 integrin caused the malignant cells to assume a morphology indistinguishable from that of the nonmalignant cells. Growth of the malignant cells was also arrested when they formed acini after the antibody treatment, and the adhesion molecules were normally distributed on the surface. Hence, blockage of function of 1 integrin in the malignant line restored a number of normal morphological and biochemical features. Although the changes in culture were reversible on removal of the antibody, injection of the antibody-treated cells into nude mice reduced the size and number of tumors formed (Weaver et al., 1997). This may have been caused by the continued action of the attached antibody to the cellular integrin for some time after injection. Treatment of the nonmalignant line with antibodies to the 6/4 integrin complex led to abrogation of normal morphogenesis and loss of growth control but had no effect on the malignant line (Weaver et al., 1997). These in vitro studies of mammary epithelium confirm the importance of tissue architecture in maintaining differentiated function of cells (Davidson, 1964; DePomerai et al., 1983; Grobstein and Zwilling, 1953). They show that microenvironmental cues such as removal of EGF are powerful factors in the induction of malignancy (Briand et al., 1996). They also indicate that seemingly small modifications of surface molecules such as the integrins may override the malignant or nonmalignant status of cells like those described here that differ in only one of their many genetic abnormalities (Weaver et al., 1997). The role of proteases in this system is not known, but it is conceivable that blocking antibody alters the sensitivity of integrins to proteolytic degradation or their ability to combine with other membrane components. However, the more physiological case of long-term

48

Harry Rubin

regulation of initiated mammary nodular epithelium in vivo requires a large population of normal mammary epithelial cells (Faulkin and DeOme, 1960; Medina et al., 1978).

XI. FURTHER DEVELOPMENTS ON THE ROLE OF PROTEASES IN CARCINOGENESIS The first suggestions about a role of proteases in carcinogenesis came with the observations of the lysis of plasma clots containing chicken cells transformed by infection with RSV (Fischer, 1946). This was followed by the observation that FBS, and to a lesser extent CS, in possible concert with proteins released from the surface of noninfected cells, suppressed RSV-transformation of chick embryo fibroblasts (Rubin, 1960b). It was later proposed that the suppression came from inhibition of proteolytic enzymes released from infected cells (Rubin, 1970). It was then found that RSV-infected cells released an enzyme that lysed fibrin (Unkeless et al., 1973). The enzyme was identified as a serine protease that activated plasminogen in serum and was correlated with transformation of the cells (Ossowski et al., 1973a); the activity of the enzyme was inhibited by FBS (Unkeless et al., 1973). It was also produced by fibroblasts transformed by other RNA as well as DNA viruses and by chemicals, rat mammary carcinoma cells, and several human tumor cell lines (Ossowski et al., 1973b). However, some nontransformed cell lines had high levels of fibrinolytic activity, and some transformed lines had no activity (Mott et al., 1974; Reich, 1975). This put a damper on the great enthusiasm that had been generated about the relation between the relation of plasminogen activator to transformation. It is now apparent that there are many surface-associated proteases in tumor cells that have nothing to do with activating plasminogen yet play an important role in neoplastic behavior. For example, plasminogen-activator appears to be necessary but not sufficient for invasion by an array of human squamous carcinoma cell lines, but release of an interstitial collagenase coincided with invasiveness (Ossowski, 1992). Each specific inhibitor of three different classes of protease induced partial inhibition of the degradation of ECM by RSV-transformed fibroblasts, and the combination of all three inhibitors had an additive effect (Fairbairn et al., 1985). RSV-transformed fibroblasts also release increased amounts of plasminogen activator (Unkeless et al., 1973, 1974) and gelatinase (Hamaguchi et al., 1995) that are correlated with transformation. Protease inhibitor studies indicate a role for thiol-, aspartic-, and metalloproteases with low pH optima in digesting proteins immediately surrounding mouse melanoma cells (Young and Spevacek, 1993). The evidence therefore points to

Microenvironmental Regulation of the Initiated Cell

49

the augmentation of multiple proteases as a common feature of neoplastic transformation. The number of possible candidates for involvement with transformation is constantly growing. For example, MMPs are a family of secreted and membrane proteases that require zinc or calcium and that can cleave components of the extracellular matrix (ECM). Their known number has increased from 15 in 1996 (Coussens and Werb, 1996) to 24 in 2002 (Egeblad and Werb, 2002). Other classes of protease are serine and thiol proteases. More than 500 proteases have been described, and the number keeps growing (Barrett et al., 1998). The relation of proteases and their inhibitors to neoplastic development is considered in greater detail in a more recent review (Rubin, 2003). The MMPs are associated with a number of normal and pathological functions, such as inflammation, cell migration during wound healing, embryo implantation, and ovulation (Coussens and Werb, 1996), but the main focus in tumor development has been their role in invasion and metastasis (Stetler-Stevenson et al., 1993) rather than initiation and promotion. The macromolecular structure of the mammalian basement membrane, the first barrier to invasion by epithelial tumors, is becoming ever more complex (Erickson and Couchman, 2000). The MMPs occur mainly as proenzymes that require activation in tumors to obtain the invasive phenotype. Activation alone may be insufficient for their effective function if there is an excess of inhibitors of their activity. It is the balance of active enzymes and their inhibitors that determine whether local matrix degradation occurs. It has become apparent, however, that the MMPs have functions in tumors other than promotion of invasion and have substrates other than ECM (Egeblad and Werb, 2002). There is evidence that they function in tumor development earlier than in the invasion stage. For example, it has been found that overexpression of a tissue inhibitor of MMPs (TIMP) in mice blocks the development of liver cancers through inhibition of events associated with tumor initiation by a tumor virus antigen, as well as blocking later stages of tumor development (Martin et al., 1996). Transgenic mice deficient in matrilysin, an MMP localized in the lumenal surface of intestinal cells, have a 60% reduction on intestinal tumors in a strain bred for 100% incidence of such tumors (Wilson et al., 1997). The lumenal localization of matrilysin indicates that it acts early in tumor development in a capacity independent of matrix degradation. The membrane-type MMPs (MT-MMPs) are localized at the cell surface and activated intracellularly by a group of transmembrane serine proteases (Coussens and Werb, 1996). Some MMPs, such as plasminogen activator, cleave propeptide domains of secreted pro-MMPs. There are several advantages to MMPs being bound at the cell surface, including the restriction of their activity to substrates in the vicinity of the cell or to adjacent matrix

50

Harry Rubin

components (Moscatelli and Rifkin, 1988). An example of such localization was found with plasminogen activator of RSV-transformed chick embryo fibroblasts (Unkeless et al., 1973). The degradation of labeled fibrin coating a coverslip only occurred when the covered face was placed downward in direct contact with the transformed cell monolayers; no fibrinolysis was observed when the fibrin-coated face was turned upward and away from the cells. Such localized behavior, however, only occurs when an inhibitory serum such as FBS is present; activity can be detected in medium containing only chicken serum. MMP activity is regulated by TIMPs (Coussens and Werb, 1996) that are presumably located in the ECM adjacent to the cells. It will be recalled that the inhibition of RSV-induced transformation by FBS was attributed to the inhibitory macromolecules in FBS, with the possible aid of material released from the cells (Rubin, 1960b). The latter is probably related to a macromolecular conditioning factor released from the cells that strongly enhances the proliferation of small numbers of cells (Rubin, 1966b). This conditioning factor exhibited a steep negative concentration gradient of activity within a short distance from the cell layer (Rein and Rubin, 1968). Various methods of inhibiting MMP activity have been tested for the treatment of cancer in Phase II and III trials (Egeblad and Werb, 2002). The methods include inhibition of MMP synthesis and blocking MMP activity by the use of TIMPs, peptidomimetic, and nonpeptidomimetic inhibitors. Although some tests have been successful in experimental animals, they have not been successful in treating human cancer, and at least one has been harmful, perhaps by blocking normal functions of MMPs. The discrepancy between positive results in treating relatively simple and uniform models of transplanted cancer in inbred animals and failures in the long-developing, heterogeneous cancers of diverse human population is unfortunately a common experience in chemotherapy. Some hope was held out for yet-to-be-tested use in early stages of cancer (Egeblad and Werb, 2002). Cathepsins are thiol or aspartic proteases that usually have a low optimal pH for activity and are mainly confined to lysosome and endocytic vesicles. They were among the first proteases found to increase in cancer (Maver et al., 1946; Sylve´n and Malmgren, 1957). Cathepsin B has more recently been detected in the plasma membrane of animal and human tumor cells (Rozhin et al., 1987). Cathepsin D is overexpressed in breast cancer cells and plays a major role in their progression (Rochefort et al., 1990). Cathepsin L is the major excreted protein of fibroblasts transformed spontaneously, chemically, or by infection with RNA or DNA tumor viruses (Denhardt et al., 1986; Gal and Gottesman, 1986; Gottesman, 1978). It constitutes up to 30% of the protein excreted from these cells and, as it has a broad substrate specificity, is likely to play a significant role in the

Microenvironmental Regulation of the Initiated Cell

51

invasion of surrounding tissue. Because the transformed fibroblasts also exhibit increases in plasminogen activator (Ossowski et al., 1973b; Unkeless et al., 1973, 1974) and in MMPs (Chen et al., 1991; Hamaguchi, et al., 1995) the concurrent use of a broad range of protease inhibitors for the treatment of cancer is suggested.

XII. CONCLUSIONS Epidermal cells of mice or rabbits are neoplastically initiated by a single treatment with carcinogens but do not develop into tumors unless a tumor promoter, such as TPA, is repeatedly applied to the skin. Because the tumor promoter is itself nonmutagenic and minimally carcinogenic to initiated skin, the papillomas produced by its application to initiated skin can be considered an overt manifestation of initiated cells. When an excess of normal keratinocytes is mixed with papilloma cells, either in cell culture (Hennings et al., 1990) or in skin grafts (Strickland et al., 1992), their proliferation is strongly inhibited. An excess of dermal fibroblasts is ineffective at inhibiting growth of initiated papilloma cells either in vitro or in vivo. In fact, the addition of an excess of fibroblasts to the papilloma cells doubles the size of the tumors formed in skin grafts, and the addition of a larger excess of fibroblasts has no further effect (Strickland et al., 1992). The addition of TPA stimulated the growth of the papilloma cells in mixtures with normal keratinocytes in culture and in grafts, but it did not stimulate growth of the papilloma cells in the absence of the inhibition by the normal cells. Keratinocytes that had some of the earliest properties of initiated cells, but could not form tumors, failed to inhibit the papilloma cells, and the normal keratinocytes did not inhibit fully malignant cells. Similar results to these were obtained with normal and neoplastic human epidermal cells in skin grafts on nude mice and in an organotypic culture system (Javaherian et al., 1998; Karen et al., 1999). It therefore can be concluded that initiated cells are kept in check by the surrounding normal keratinocytes, and that fibroblasts have no significant role in the process. The mechanism of initiated cell regulation by the normal keratinocytes remains unresolved. There is evidence that the normal and early neoplastic cells have to be in close proximity to one another for regulation of the latter to occur. Medium conditioned by normal mouse keratinocytes does not inhibit growth of papilloma cells (Hennings et al., 1990), and the suppressed growth of low-grade malignant human keratinocytes did not operate at low population densities (Vaccariello et al., 1999). The proximity relation raised the possibility that there had to be direct contact between the normal and initiated cells for inhibition of the latter, and that junctional communication

52

Harry Rubin

between the two cell types involving exchange of small molecules was necessary. However, there was no junctional communication between these cells as detected by the usual fluorescent dye transfer assay (Strickland et al., 1992), so it can be ruled out as the mechanism of inhibition in the case studied (Hennings et al., 1990). The regulatory role of junctional communication first became an issue in experiments on the regulation of initiated fibroblasts by their microenvironment. RSV is a powerful transforming agent for chick embryo fibroblasts in culture and in chickens, but it does not cause sarcomas in the chick embryo despite successful infection and virus production in connective tissue cells (Milford and Duran-Reynals, 1943). Transformed focus development in CEF culture is suppressed by moderate concentrations of FBS or high concentrations of CS, but only when the RSV-infected cells are a small minority of the total (Rubin, 1960b). Later evidence indicated that cultures heavily infected with RSV, and therefore transformed, released an overgrowth stimulatory factor for confluent cultures of CEF (Rubin, 1970), and that addition to CEF cultures of a very low dose of trypsin, one that did not retract much less detach the cells, stimulated DNA synthesis (Sefton and Rubin, 1970). These findings indicated that RSV-infected CEF released transforming proteases that were only effective in proximity to those cells. Transformation could be suppressed by proteins in relatively high concentration in FBS, with limited additional inhibition by conditioning factors released from the surface of normal cells, which are effective only at short range (Rein and Rubin, 1968). FBS is rich in fetuin, an -globulin that inhibits proteases (Fisher et al., 1958), and partially suppresses RSV-transformation when added to an otherwise permissive medium (Rubin, 1960b). The likelihood of a causal role of proteases in transformation is supported by findings that several proteases stimulate proliferation of confluent, contact-inhibited CEF (Sefton and Rubin, 1970; Zetter et al., 1976), as does the medium from CEF heavily transformed by RSV infection (Rubin, 1970). Proteolytic activity is also associated with epithelial (Jones and Ashwood-Smith, 1970), mesenchymal (Attardi et al., 1967), and nerve (Greene et al., 1968) growth factors. High concentrations of FBS also inhibited chemically induced transformation of an established line of mouse fibroblasts (Bertram, 1977). Once the cells had been transformed, their ability to form foci on plastic, or colonies in agar, was inhibited at high concentration of FBS, but only in the presence of a large excess of nontransformed cells. The inhibited expression of the transformed phenotype was attributed to the increased saturation density of the nontransformed cells in higher concentrations of FBS, with the assumption that the increased crowding enhanced membrane interactions between the cells and allowed the nontransformed cells to regulate the growth of the transformed cells. However, there was in fact no increase in saturation density with increases of high concentrations of one of the two sera

Microenvironmental Regulation of the Initiated Cell

53

used and only slight increases with the other, but there was strong inhibition of transformed focus formation in both cases (Bertram, 1977). Therefore it is likely that the high FBS concentrations exerted their suppression of focus formation by inhibiting pericellular proteases as they did in RSV infection of CEF (Rubin, 1960b). It was later reported that there is a strong correlation between the degree of junctional communication between several types of transformed mammalian cells and their cognate nontransformed cells, which was held responsible for inhibiting growth of the transformed cells (Mehta et al., 1986). Junctional communication between cells was probed by injection of a 443-da fluorescent tracer, and all the measurements were done in 10% CS, which allows expression of the transformed phenotype. Later experiments in a rat fibroblast line transformed by an array of different oncogenes showed that inhibition by cocultured normal fibroblasts depended on the type of transforming oncogene and the type of normal fibroblast but was independent of junctional communication between the transformed and normal cells (Martin et al., 1991). It therefore is unlikely for a variety of reasons that junctional communication is the mechanism that underlies the inhibition of transformed fibroblasts by normal fibroblasts. The role of short-range protease inhibitors released from the normal cells could be investigated, as it was in the RSV-infected CEF (Rein and Rubin, 1968; Rubin, 1960b). The effectiveness of FBS is not restricted to cell culture manifestations of transformation. FBS also prevents, or significantly delays, the conversion of kidney cells from 1-day-old mice and cells from 12–14-day-old embryos to production of malignant tumors on implantation into syngeneic mice (Evans and Andresen, 1966; Sanford et al., 1972). Cells growing in horse or calf sera underwent the conversion earlier than those in FBS (Parshad and Sanford, 1968), and the addition of FBS to those two sera delayed the conversion (Sanford et al., 1972). The addition of fetuin to horse serum also appeared to delay the conversion (Parshad and Sanford, 1968; Sanford et al., 1972). The latent period for the appearance of tumors after inoculation of cells into mice was longer for tumor-producing cells from FBS cultures than from those grown in other sera. The overall picture for protection against conversion to tumor production therefore parallels the suppression of morphological transformation and maintenance of susceptibility to contact inhibition in culture. Both the in vivo and in vitro tests are consistent with a role for protease inhibition in suppressing transformation. These results are reinforced by the findings that low-molecular weight specific protease inhibitors inhibit chemical and radiation transformation of mouse cells in culture (Kennedy and Little, 1981; Kuroki and Drevon, 1979). The protease inhibitors reduced saturation density of the transformed cells and suppressed their overgrowth (Kuroki and Drevon, 1979). There was marked heterogeneity in sensitivity to the protease inhibitors among transformed

54

Harry Rubin

cell lines, indicating that more than one protease species is involved in transformation (Kuroki and Drevon, 1979). Trypsin- and cathepsin-like activities were strongly elevated in mouse 3T3 fibroblast lines transformed by RSV and mouse sarcoma virus (Bosmann, 1972) and by DNA viruses (Bosmann, 1969). A protease known as plasminogen activator was released in large amounts from RSV-transformed CEF, and large quantities were also released from mammalian cells transformed by other RNA and DNA viruses as well as carcinoma cells (Ossowski et al., 1973b; Reich, 1975; Unkeless et al., 1973). However, plasminogenactivator release did not correlate with transformation in a group of cell lines derived from human tumors (Mott et al., 1974), and it was found in an untransformed line of Swiss 3T3 cells (Reich, 1975). Clear evidence for the release of more than one transformation-related protease was shown when invasiveness of an array of human carcinoma cell lines was correlated with an interstitial collagenase, but not with plasminogen activator released by the same cells (Ossowski, 1992). Single, specific inhibitors of several different classes of protease only partially inhibit digestion of ECM proteins by RSV-transformed fibroblasts, but they have an additive effect in combination with each other, indicating that at least three protease classes contribute to ECM digestion (Fairbairn et al., 1985). A similar conclusion was drawn from protease inhibitor effects on digestion of a fluorescent albumin substrate in the immediate region surrounding mouse melanoma cells cultured in agarose gels (Young and Spevacek, 1993). There are now 24 MMPs associated with degradation of ECM, so it is not surprising that the use of specific inhibitors of proteases in cancer chemotherapy has so far been unsuccessful (Egeblad and Werb, 2002). It might be more productive to apply an array of natural protease inhibitors, such as the -globulins, in therapeutic trials because they are likely to inhibit a wide spectrum of proteases. There are at least six protease inhibitors in human plasma, and they migrate together with 1- and 2-globulins, all of which are glycoproteins (Heimburger, 1975). Fetuin makes up a large fraction of the 1-globulins in serum from newborn calves, which has very little - and -globulin (Deutsch, 1954; Pedersen, 1944). The amount of fetuin in serum decreases with the age of the calves, but small amounts can still be demonstrated in the same protein fraction in adult cows. A similar situation exists in sera from sheep and horses; in contrast, human and rabbit fetuses have little fetuin but contain considerable amounts of globulin of higher molecular weight than fetuin (Pedersen, 1944). There is suggestive evidence of the presence of fetuin in the serum of chicken embryos, but it disappears after 18 days of development (Pedersen, 1944). The presence of fetuin in chicken embryo serum might contribute to the failure of tumor production by RSV infection of the embryo (Milford and Duran-Reynals, 1943), even though embryonic fibroblasts are readily transformed in cell culture containing CS

Microenvironmental Regulation of the Initiated Cell

55

(Temin and Rubin, 1958), and fibroblasts that are infected in the embryo but not transformed there undergo transformation when they are transferred to cell culture (Dolberg and Bissell, 1984). ECM proteins probably play a larger role than serum proteins in protease inhibition and suppression of transformation in vivo than in vitro because the ratio of cell volume to extracellular fluid space is vastly greater in the body than in cell culture. Primary cultures present a more reliable model than cell lines for testing the role of proteases and their microenvironmental inhibitors in tumor development (Reich, 1975) because the selective establishment of cell lines leads to subtle differences in transformability and in the suppressive capacity of the cellular microenvironment (Rubin, 1994a). RSV infection of CEF in culture provided an efficient and convincing way of demonstrating the role of proteases and their inhibitors because it results in transformation within 2–3 days, which can be easily quantitated, and had a history of lysing plasma clots. There was also an in vivo model of suppressed transformation of RSV-infected cells in the chick embryo, which lent significance to the in vitro results. Analogous methodology can be developed for testing the role of proteases and their inhibitors in the development of epithelial neoplasia. The results achieved thus far with fibroblasts indicate that promotion allows the expression of the neoplastic phenotype by stimulating the release of pericellular proteases or damaging the surrounding uninfected cells that provide the restraining antiproteases (Rubin, 2003). Progression would represent genetic changes in the initiated cells that enhance the release of proteases that drive the invasion of connective tissue and the dissemination of metastasis (Coussens and Werb, 1996; Stetler-Stevenson, et al., 1993). The model encourages further exploration with epithelial cells of the proteases (Barret, et al., 1998; Egeblad and Werb, 2002) and natural protease inhibitors (Heimburger, 1975; Silverman et al., 2001) involved in neoplastic development, preferably in a rapidly responding, rigorously controlled, quantifiable system like that developed with fibroblasts (Rubin, 1960b). Such studies would provide a firm foundation for the possible application of balanced combinations of natural protease inhibitors in treatment of cancer rather than specifically targeted inhibitors, which have thus far proved disappointing in cancer therapy (Coussens et al., 2002). It should be noted, however, that mutations that alter cell surface proteins directly, such as the APC gene in mouse intestinal neoplasia (Shoemaker et al., 1997) and colorectal cancer in humans (Kinzler and Vogelstein, 1996) may initiate transformation without excess protease release. Even in colorectal carcinogenesis, however, approximately 50% of benign adenomas express the MMP matrilysin in the mucosal epithelium and several MMPs in the stroma; the carcinomas express more matrilysin and additional MMPs in the stroma (Newell et al., 1994). Altering the cell surface and disrupting its interaction with neighboring cells may therefore be the key to progressive carcinogenesis (Rubin, 1985).

56

Harry Rubin

ACKNOWLEDGMENTS I am indebted to Dorothy Rubin for editing and transcribing the manuscript throughout its many drafts. Drs. Jonathan Garlick and Stuart Yuspa made helpful comments. Support for the present work came from the National Institutes of Health grant G13 LM07483-01.

REFERENCES Andersson, D. I., Slechta, E. S., and Roth, J. R. (1998). Science 282, 1133–1135. Attardi, D. G., Schlesinger, M. J., and Schlesinger, S. (1967). Science 156, 1253–1255. Barcellos-Hoff, M. H., Aggeler, J., Ram, T. G., and Bissell, M. J. (1989). Development 105, 223–235. Barrett, A. J., Rawlings, N. D., and Woessner, J. F. (1998). Handbook of Proteolytic Enzymes. Academic Press, San Diego, CA. Bedi, G. C., Westra, W. H., Gabrielson, E., Koch, W., and Sidransky, D. (1996). Cancer Res. 56, 2484–2487. Berenblum, I. (1941). Cancer Res. 1, 807–814. Berenblum, I., and Shubik, P. (1947). Brit. J. of Cancer 1, 383–391. Bertram, J. S. (1977). Cancer Res. 37, 514–523. Bertram, J. S. (1979). Cancer Res. 39, 3502–3508. Boreiko, C. J., Abernathy, D. J., and Stedman, D. B. (1987). Carcinogenesis 8, 321–325. Bosmann, H. B. (1969). Exp. Cell Res. 54, 217–221. Bosmann, H. B. (1972). Biochim. Biophys. Acta 264, 339–343. Boukamp, P., Popp, S., Bleuel, K., Tomakiidi, E., Bu¨rkle, A., and Fusenig, N. E. (1999). Oncogene 18, 5638–5645. Braakhuis, B. J. M., Tabor, M. P., Kummer, J. A., Leemans, C. R., and Brakenhoff, R. H. (2003). Cancer Res. 63, 1727–1730. Brash, D. E. (1997). Trends in Genetics 13, 410–414. Brash, D. E., Ziegler, A., Jonason, A. S., Simon, J. A., Kunala, S., and Leffell, D. J. (1996). J. Inv. Dermatol. Symp. Proc. 1, 136–142. Briand, P., Nielsen, K. V., Madsen, M. W., and Petersen, O. W. (1996). Cancer Res. 56, 2039–2044. Briand, P., Petersen, O. W., and Van Deurs, B. (1987). In Vitro Cell. Dev. Biol. 23, 181–188. Brown, K., Quintanilla, M., Ramsden, M., Kerr, I. B., Young, S., and Balmain, A. (1986). Cell 46, 447–456. Cha, R. S., Guerra, L., Thilly, W. G., and Zarbl, H. (1996). Carcinogenesis 17, 2519–2524. Cha, R. S., Thilly, W. G., and Zarbl, H. (1994). Proc. Natl. Acad. Sci. USA 91, 3749–3753. Chang, W. P., and Little, J. R. (1992). Radiat. Res. 131, 53–59. Chen, J.-M., Aimes, R. T., Ward, G. R., Youngleib, G. L., and Quigley, J. P. (1991). J. Biol. Chem. 266, 5113–5121. Chow, M., Koo, J. J., Ng, P., and Rubin, H. (1998). Mutation Res. 413, 251–264. Chow, M., and Rubin, H. (1999). Proc. Natl. Acad. Sci. USA 96, 2093–2098. Chow, M., and Rubin, H. (2000). Cancer Res. 60, 6510–6518. Cook, J. W., Haslewood, G. A. D., Hewett, C. L., Hieger, I., Kennaway, E. L., and Mayneord, W. N. (1937). Am. J. Cancer 29, 219–259. Cook, J. W., Hewett, C. L., and Hieger, I. (1933). J. Chem. Soc. (Lond). Coussens, L. M., Singleton, and Matrisian, L. M. (2002). Science 295, 2387–2392.

Microenvironmental Regulation of the Initiated Cell

57

Coussens, L. M., and Werb, Z. (1996). Chem. J. Biol. 3, 895–904. Cramer, W., and Stowell, R. E. (1942). J. Natl. Cancer Inst. 2, 379–402. Cunha, G. R., and Matrisian, L. M. (2002). Differentiation 70, 469–470. Davidson, E. (1964). Adv. Genetics 12, 143–280. Denhardt, D. T., Hamilton, R. T., Parfett, C. L. J., Edwards, D. R., St. Pierre, R., Waterhouse, P., and Nilsen-Hamilton, M. (1986). Cancer Res. 46, 4590–4593. DeOme, K., Miyamoto, M., Osborn, R., Guzman, R., and Lum, K. (1978). Cancer Res. 38, 2103–2111. DeOme, K. B. (1965). Cancer Res. 25, 1348–1351. DePomerai, D. I., Kotecha, F.-H., Fullick, C., Young, A., and Gali, M. A. H. (1983). J. Embryol. Exp. Morph 77, 201–220. Deutsch, H. F. (1954). J. Biol. Chem. 208, 669–678. Dolberg, D. S., and Bissell, M. J. (1984). Nature 309, 552–556. Dolberg, D. S., Hollingsworth, R., Hertle, M., and Bissell, M. J. (1985). Science 230, 676–678. Doljanski, L., and Tennenbaum, E. (1942). Cancer Res. 2, 776–785. Dotto, G. P., Weinberg, R. A., and Ariza, A. (1988). Proc. Natl. Acad. Sci. USA 85, 6389–6393. Eagle, H., and Levine, E. M. (1967). Nature 213, 1102–1106. Egeblad, M., and Werb, Z. (2002). Nature Reviews/Cancer 2, 161–174. Emerman, J. T., Burwen, D. R., and Pitelka, D. R. (1979). Tissue Cell 11, 109–119. Emerman, J. T., and Pitelka, D. R. (1977). In Vitro 13, 316–328. Epstein, J. H., and Epstein, W. L. (1962). J. Investig. Dermatol 39, 455–460. Erickson, A. C., and Couchman, J. R. (2000). J. Histochem. Cytochem. 48, 1291–1306. Ethier, S. P., and Ullrich, R. L. (1982). Cancer Res. 42, 1753–1760. Ethier, S. P., and Ullrich, R. L. (1984). Cancer Res. 44, 4523–4527. Evans, V. J., and Andresen, W. F. (1966). J. Natl. Cancer Inst. 37, 247–249. Fairbairn, S., Gilbert, R., Ojakian, G., Schwimmer, R., and Quigley, J. P. (1985). J. Cell Biol. 101, 1790–1798. Faulkin, L. J., Jr., and DeOme, K. B. (1960). J. Natl. Cancer Inst. 24, 953–968. Fearon, E. R., Hamilton, S. R., and Vogelstein, B. (1987). Science 238, 193–197. Fernandez, A., Mondal, S., and Heidelberger, C. (1980). Proc. Natl. Acad. Sci. USA 77, 7272–7276. Fialkow, P. J. (1976). Biochimica et Biophysica Acta 458, 283–321. Finzi, E., Kilkenny, A., Strickland, J. E., Balaschak, M., Bringman, T., Derynck, R., Aaronson, S., and Yuspa, S. H. (1988). Mol. Carcinogenesis 1, 7–12. Fischer, A. (1946). Biology of Tissue Cells. Cambridge University Press, Cambridge. Fisher, H. W., Puck, T. T., and Sato, G. (1958). Proc. Natl. Acad. Sci. USA 44, 4–10. Foulds, L. (1965). Cancer Res. 25, 1339–1347. Frankfurt, O. S., and Raitcheva, E. (1973). J. Natl. Cancer Inst. 51, 1861–1864. Fujii, H., Marsh, C., Cairns, P., Sidransky, D., and Gabrielson, E. (1996). Cancer Res. 56, 1493–1497. Gal, S., and Gottesman, M. (1986). Biochem. Biophys. Res. Comm. 139, 156–162. Glick, A., Popescu, N., Alexander, V., Ueno, H., Bottinger, E., and Yuspa, S. H. (1999). Proc. Natl. Acad. Sci. USA 96, 14949–14954. Gottesman, M. (1978). Proc. Natl. Acad. Sci. USA 75, 2767–2771. Greene, L. A., Shooter, E. M., and Varon, S. (1968). Biochemistry 60, 1383–1388. Grisham, J. W., Smith, G. J., Lee, L. W., Bentley, K. S., and Fatteh, M. V. (1988). Cancer Res. 48, 5977–5983. Grobstein, C. (1961). Exp. Cell Res. 8, 234–245. Grobstein, C., and Zwilling, E. (1953). J. Experimental Zool 122, 259–284. Hamaguchi, M., Yamagata, S., Thant, A. A., Xiao, H., Iwata, H., Mazaki, T., and Hanafusa, H. (1995). Oncogene 10, 1037–1043.

58

Harry Rubin

Harris, M. (1964). Cell Culture and Somatic Variation. Holt, Rinehart and Winston, New York. Hassell, J. R., Robey, P. G., Barrach, H.-J., Wilczek, J., Rennard, S. I., and Martin, G. R. (1980). Proc. Natl. Acad. Sci. USA 77, 4494–4498. Hebb, C. R., and Chu, M.-Y. W. (1960). Exp. Cell Res. 20, 153–157. Hecker, E. (1968). Cancer Res. 28, 2238–2349. Heimburger, N. (1975). In Proteases and Biological Control (D. B. Reich, Rifkin, and E. Shaw, eds.), Vol. 2, pp. 367–386. Cold Spring Harbor Conference of Cell Proliferation, Cold Spring Harbor. Hendrickson, H., Slechta, E. S., Bergthorsson, U., Andersson, D. I., and Roth, J. R. (2002). Proc. Natl. Acad. Sci. USA 99, 2164–2169. Hennings, H., and Boutwell, R. K. (1970). Cancer Res. 30, 312–320. Hennings, H., Robinson, V. A., Michael, D. M., Pettit, G. R., Jung, R., and Yuspa, S. H. (1990). Cancer Res. 50, 4794–4800. Herschman, H. R., and Brankow, D. W. (1986). Science 234, 1385–1388. Herschman, H. R., and Brankow, D. W. (1987). Carcinogenesis 8, 993–998. Hieger, I. (1936). Amer. J. Cancer 28, 522–529. Hill, M., Hillova, J., and Mariage-Samson, R. (1991). In Vitro Cell Dev. Biol. 27A, 270–272. Hozumi, M., Ogawa, M., Sugimura, T., Takeuchi, T., and Umezawa, H. (1972). Cancer Res. 32, 1725–1728. Huberman, E., Sachs, L., Yang, S. K., and Gelboin, H. V. (1976). Proc. Natl. Acad. Sci. USA 73, 607–611. Hu¨lser, D. F., and Webb, D. J. (1973). Expt. Cell Res. 80, 210–222. Jainchill, J. L., Aaronson, S. A., and Todaro, G. J. (1969). J. Virol 4, 549–553. Javaherian, A., Vaccariello, M., Fusening, N. E., and Garlick, J. A. (1998). Cancer Res. 58, 2200–2208. Jonason, A. S., Kunala, S., Price, G. J., Restifo, R. J., Spinelli, H. M., Persing, J. A., Leffell, D. J., Tarone, R. E., and Brash, D. E. (1996). Proc. Natl. Acad. Sci. USA 93, 14025–14029. Jones, P. H., and Watt, F. M. (1993). Cell 73, 713–724. Jones, R. O., and Ashwood-Smith, M. J. (1970). Cell 59, 161–163. Kadhim, M. A., Lorimore, S. A., Townsend, K. M. S., Goodhead, D. T., Buckle, V. J., and Wright, E. G. (1995). Int. J. Radiat. Biol. 67, 287–293. Kakunaga, T. (1977). The Transformation of Human Diploid Cells by Chemical Carcinogens. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Karen, J., Wang, Y., Javaherian, A., Vaccariello, M., Fusenig, N. E., and Garlick, J. A. (1999). Cancer Res. 59, 474–481. Kasami, M., Vnencak-Jones, C. L., Manning, S., Dupont, W. D., and Page, D. L. (1997). Am. J. Pathol. 150, 1925–1932. Kennedy, A. (1994). Cancer Res. (suppl) 54, 1999–2005s. Kennedy, A. R. (1982). Carcinogenesis 3, 1093–1095. Kennedy, A. R. (1985). Carcinogenesis 6, 1441–1445. Kennedy, A. R., Fox, M., Murphy, G., and Little, J. B. (1980). Proc. Natl. Acad. Sci. USA 77, 7262–7266. Kennedy, A. R., and Little, J. B. (1981). Cancer Res. 41, 2103–2108. Kinsella, A. R., and Radman, M. (1980). Proc. Natl. Acad. Sci. USA 77, 3544–3547. Kinzler, K. W., and Vogelstein, B. (1996). Cell 87, 159–170. Kleinman, H. K., McGarvey, M. L., Hassell, J. R., Star, V. L., Cannon, F. B., Laurie, G. W., and Martin, G. R. (1986). Biochemistry 25, 312–318. Kuroki, T., and Drevon, C. (1979). Cancer Res. 39, 2755–2761. Lankas, G. R., Baxter, C. S., and Christian, R. T. (1977). Mutation Res. 45, 153–156.

Microenvironmental Regulation of the Initiated Cell

59

Larson, P. S., de las Morenas, A., Cupples, L. A., Huang, K., and Rosenberg, C. L. (1998). Amer. J. Pathology 152, 1591–1598. Li, C.-Y., Little, J. B., Hu, K., Zhang, W., Dewhirst, M. W., and Huang, Q. (2001). Cancer Res. 61, 428–432. Mackay, W., Benasutti, M., Drouin, E., and Loechler, E. L. (1992). Carcinogenesis 13, 1415–1425. Marder, B. A., and Morgan, W. F. (1993). Molecular and Cellular Biology 13, 6667–6677. Martin, D. C., Ru¨ther, U., Sanchez-Sweatman, O. H., Orr, F. W., and Khokha, R. (1996). Oncogene 13, 569–576. Martin, W., Zempel, G., Hu¨lser, D., and Willecke, K. (1991). Cancer Res. 51, 5348–5354. Maver, M. E., Mider, G. B., Johnson, J. M., and Thompson, J. W. (1946). J. Natl. Cancer Ins. 2, 277–282. McKenzie, I., and Rous, P. (1941). J. Exp. Med. 73, 391–416. Medina, D. (1979). J. Natl. Cancer Inst. 62, 397–405. Medina, D., Shepherd, F., and Gropp, T. (1978). J. Natl. Cancer Inst. 60, 1121–1126. Mehta, P. P., Bertram, J. S., and Loewenstein, W. R. (1986). Cell 44, 187–196. Melendez-Colon, V. J., Luch, A., Seidel, A., and Baird, W. M. (1999). Carcinogenesis 20, 1885–1891. Merritt, A. J., Gould, K. A., and Dove, W. F. (1997). Proc. Natl. Acad. Sci. USA 94, 13927–13931. Milford, J. J., and Duran-Reynals, F. (1943). Cancer Res. 3, 578–584. Missero, C., Filvaroff, E., and Dotto, G. P. (1991a). Proc. Natl. Acad. Sci. USA 88, 3489–3493. Missero, C., Ramon Y Cajal, S., and Dotto, G. P. (1991b). Proc. Natl. Acad. Sci. USA 88, 9613–9617. Mondal, S., and Heidelberger, C. (1970). Proc. Natl. Acad. Sci. USA 65, 219–229. Moscatelli, D., and Rifkin, D. B. (1988). Biochem. e Biophys. Acta 948, 67–85. Mott, D. M., Fabisch, P. H., Sani, B. P., and Sorof, S. (1974). Biochem. Biophys. Res. Comm. 61, 621–627. Mottram, J. C. (1944). J. Path. Bact. 56, 181–187. Newbold, R. F., and Brookes, P. (1979). Nature 261, 52–54. Newell, K. J., Witty, J. R., Rodgers, W. H., and Matrisian, L. M. (1994). Mol. Carcinogenesis 10, 199–206. Nias, A. H., Gilbert, C. W., Lajtha, L. G., and Lange, C. S. (1965). Int. J. Rad. Biol. 9, 275–290. Ossowski, L. (1992). Cancer Res. 52, 6754–6760. Ossowski, L., Quigley, J. P., Kellerman, G. M., and Reich, E. (1973a). J. Exp. Medicine 138, 1056–1064. Ossowski, L., Unkeless, J. C., Tobia, A., Quigley, J. P., Rifkin, D. B., and Reich, E. (1973b). J. Exp. Medicine 137, 112–126. Parkinson, E. K. (1985). Brit. J. Cancer 52, 479–493. Parshad, R., and Sanford, K. K. (1968). J. Natl. Cancer Inst. 41, 767–779. Pedersen, K. O. (1944). Nature 154, 575. Petersen, O. W., Ronnov-Jessen, L., Howlett, A. R., and Bissell, M. J. (1992). Proc. Natl. Acad. Sci. USA 89, 9064–9068. Pitot, H. (1986). Fundamentals of Oncology. 3rd Edition. Marcel Dekker, New York. Ponte´n, F., Berg, C., Ahmadian, A., Ren, Z.-P., Nister, M., Lundenberg, J., Uhlen, M., and Ponte´n, J. (1997). Oncogene 15, 1059–1067. Puck, T. T. (1995). Mutation Res. 329, 173–181. Quigley, J. P. (1979). Cell 17, 131–141. Ramon y Cajal, S., Missero, C., Marchetti, E., and Dotto, G. P. (1994). Am. J. Path 145, 846–855. Reddy, A. L., and Fialkow, P. J. (1983). Nature 304, 69–71.

60

Harry Rubin

Reich, E. (ed.) (1975). Plasminogen Activator: Secretion by Neoplastic Cells and Macrophages, CSH Conferences on Cell Proliferation. Rein, A., and Rubin, H. (1968). Exp. Cell Res. 49, 666–678. Reznikoff, C. A., Bertram, J. S., Brankow, D. W., and Heidelberger, C. (1973). Cancer Res. 33, 3239–3249. Rochefort, H., Capony, F., and Garcia, M. (1990). Cancer and Metastasis Reviews 9, 321–331. Rodin, S. N., and Rodin, A. S. (2000). Proc. Natl. Acad. Sci. USA 97, 12244–12249. Rodin, S. N., and Rodin, A. S. (2002). Mut. Res. 508, 1–19. Roop, D. R., Lowy, D. R., Tambourin, P. E., Strikland, J., Harper, J. R., Balaschak, M., Spangler, E. F., and Yuspa, S. H. (1986). Nature 323, 822–824. Roskelley, C. D., Desprez, P. Y., and Bissell, M. J. (1994). Proc. Natl. Acad. Sci. USA 91, 12378–12382. Rous, P., and Kidd, J. G. (1941). J. Exp. Med. 73, 365–390. Rous, P., Murphy, J. B., and Tytler, W. H. (1912). J. Am. Med. Assn. 58, 1751–1753. Rozhin, J., Robinson, D., Stevens, M. A., Lah, T. T., Honn, K. V., Ryan, R. E., and Sloane, B. F. (1987). Cancer Res. 47, 6620–6628. Rubin, H. (1955). Virology 1, 445–473. Rubin, H. (1960a). Virology 10, 29–49. Rubin, H. (1960b). Virology 12, 14–31. Rubin, H. (1966a). Exp. Cell Res. 41, 149–161. Rubin, H. (1966b). Exp. Cell Res. 41, 138–148. Rubin, H. (1967). The Harvey Lectures. 61, 117–143. Rubin, H. (1970). Science 167, 1271–1272. Rubin, H. (1985). Cancer Res. 45, 2935–2942. Rubin, H. (1992). Proc. Natl. Acad. Sci. USA 89, 977–981. Rubin, H. (1994a). Proc. Natl. Acad. Sci. USA 91, 1039–1043. Rubin, H. (1994b). Proc. Natl. Acad. Sci. USA 91, 12076–12080. Rubin, H. (2002). Oncogene 21, 7392–7411. Rubin, H. (2003). Carcinogenesis 24, 803–814. Rubin, H., and Temin, H. (1959). Virology 7, 75–91. Rubin, H., and Xu, X. (1989). Proc. Natl. Acad. Sci. USA 86, 1860–1864. Rubin, H., Yao, A., and Chow, M. (1995). Proc. Natl. Acad. Sci. USA 92, 4843–4847. Sanford, K. K., Jackson, J. L., Parshad, R., and Gantt, R. R. (1972). J. Natl. Cancer Inst. 49, 513–518. Sefton, B. M., and Rubin, H. (1970). Nature 227, 843–845. Shoemaker, A. R., Gould, K. A., Luongo, C., Moser, A. R., and Dove, W. F. (1997). Biochim. Biophys. Acta 1332, F25–F48. Shubik, P. (1950). Cancer Res. 10, 713–717. Sidransky, D., Frost, P., Von Eschenbach, A., Oyasu, R., Preisinger, A. C., and Vogelstein, B. (1992). N. Engl. J. Med. 326, 737–740. Sieweke, M. H., and Bissell, M. J. (1994). Crit. Rev. Oncogenesis 5, 297–311. Sieweke, M. H., Thompson, N. L., Sporn, M. B., and Bissell, M. J. (1990). Science 248, 1656–1660. Silverman, G. A., Bird, P. I., R. W., C., Church, F. C., Coughlin, P. B., Gettins, P. G. W., Irving, J. A., Lomas, D. A., Luke, C. J., Moyer, R. W., Pemberton, P. A., Remold-O’Donnell, E., Salveson, G. S., Travis, J., and Whisstock, J. C. (2001). J. Biol. Chem. 276, 33293–33296. Sinclair, W. K. (1964). Radiat. Res. 21, 584–611. Slaughter, D. P., Southwick, H. W., and Sinejkal, W. (1953). Cancer (Phila.) 6, 963–968. Spemann, H. (1938). Embryonic Development and Induction. Yale University Press, New Haven, CT.

Microenvironmental Regulation of the Initiated Cell

61

Stetler-Stevenson, W. G., Aznavoorian, S., and Liotta, L. (1993). Ann. Rev. Cell Biol. 9, 541–573. Stoker, A. W., Hatier, C., and Bissell, M. J. (1990). J. Cell Biology 111, 217–228. Stoker, M. (1964). Virology 24, 165–174. Stoker, M. (1967). J. Cell Sci. 2, 293–304. Stoker, M., Shearer, M., and O’Neill, C. (1966). J. Cell. Sci. 1, 297–310. Streuli, C. H., and Bissell, M. J. (1990). J. Cell Biol. 110, 1405–1415. Strickland, J. E., Ueda, M., Hennings, H., and Yuspa, S. H. (1992). Cancer Res. 52, 1439–1444. Sylve´n, B., and Malmgren, H. (1957). Acta Radiologica. Suppl. 154, 1–124. Temin, H., and Rubin, H. (1958). Virology 6, 669–688. Temin, H. M. (1967). J. Cell Physiol. 69, 53–64. Terzaghi-Howe, M. (1987). Carcinogenesis 8, 145–150. Tlsty, T., Margolin, B. H., and Lum, K. (1989). Proc. Natl. Acad. Sci. USA 86, 9441–9445. Tomlinson, I., and Bodmer, W. (1999). Nature Medicine 5, 11–12. Tomlinson, I. P. M., Novelli, M. R., and Bodmer, W. F. (1996). Proc. Natl. Acad. Sci. USA 93, 14800–14803. Troll, W., Klassen, A., and Janoff, A. (1970). Science 169, 1211–1213. Unkeless, J., Dano, K., Kellerman, G. M., and Reich, E. (1974). J. Biol. Chem. 249, 4295–4305. Unkeless, J. C., Tobia, A., Ossowski, L., Quigley, J. P., Rifkin, D. B., and Reich, E. (1973). J. Exp. Medicine 137, 85–111. Vaccariello, M., Javaherian, A., Wang, Y., Fusenig, N. E., and Garlick, J. A. (1999). J. Investig. Dermatol 113, 384–391. Vaheri, A., and Ruoslahti, E. (1974). Int. J. Cancer 13, 579–586. Van Duuren, B. L., Sivak, A., Katz, C., Seidman, I., and Melchionne, S. (1975). Cancer Res. 35, 502–505. Wartiovaara, J., Linder, E., Ruoslahti, E., and Vaheri, A. (1974). J. Exp. Medicine 140, 1522–1533. Weaver, V. M., Petersen, O. W., Wang, F., Larabell, C. A., Briand, P., Damsky, C., and Bissell, M. J. (1997). J. Cell Biol. 137, 231–245. Weber, M. J. (1975). Cell 5, 253–261. Weiss, R. A. (1970a). Exp. Cell Res. 63, 1–18. Weiss, R. A. (1970b). Int. J. Cancer 6, 333–345. Wigler, M., and Weinstein, I. B. (1976). Nature 259, 232–233. Wilson, C. L., Heppner, K. J., Labosky, P. A., Hogan, B. L. M., and Matrisian, L. M. (1997). Proc. Natl. Acad. Sci. USA 94, 1402–1407. Wilson, E. L., and Reich, E. (1979). Cancer Res. 39, 1579–1586. Windle, B., Draper, B. W., Yin, Y., O’Gorman, S., and Wahl, G. M. (1991). Genes Dev. 5, 160–174. Windle, B., and Wahl, G. M. (1992). Mutation Res. 276, 199–224. Wislocki, P. G., Wood, A. W., Chang, R. L., Levin, W., Yagi, H., Hernadez, O., Jerina, D. M., and Conney, A. H. (1976). Biochem. Biophys. Res. Comm. 68, 1006–10012. Wood, A. W., Wislocki, P. G., Chang, R. L., Levin, W., Lu, A. Y. H., Yagi, H., Hernandez, O., Jerina, D. M., and Conney, A. H. (1976). Cancer Res. 36, 3358–3366. Yamada, K. M., and Olden, K. (1978). Nature 275, 179–184. Yamagiwa, K., and Ichikawa, K. (1918). J. Cancer Res. 3, 1–29. Yao, A., Rubin, A. L., and Rubin, H. (1990). Cancer Res. 50, 5171–5176. Yotti, L. P., Chang, C. C., and Trosko, J. E. (1979). Science 206, 1089–1091. Young, P. R., and Spevacek, S. M. (1993). Biochimica t Biophysica Acta 1182, 69–74. Yuspa, S. H., Ben, T., Hennings, H., and Lichti, U. (1982). Cancer Res. 42, 2344–2349.

62

Harry Rubin

Zetter, B. R., Chen, L. B., and Buchanan, J. M. (1976). Cell 7, 407–412. Zhang, W., Remenyik, E., Zelterman, D., Brash, D. E., and Wikonkai, N. M. (2001). Proc. Natl. Acad. Sci. USA 98, 13948–13953. Ziegler, A., Jonason, A. S., Leffell, D. J., Simon, J. A., Sharma, H. W., Kimmelman, J., Remington, L., Jacks, T., and Brash, D. E. (1994). Nature 372, 22–29. Ziegler, A., Leffell, D. J., Kunala, S., Sharma, H. W., Gailani, M., Simon, J. A., Halperin, A. J., Baden, H. P., Shapiro, P. E., Bale, A. E., and Brash, D. E. (1993). Proc. Natl. Acad. USA 90, 4216–4220.

Persistent Infection with Helicobacter Pylori and the Development of Gastric Cancer Staffan Normark, Christina Nilsson, Birgitta Henriques Normark, and Mathias W. Hornef Microbiology and Tumor Biology Center and Smittskyddsinstitutet, Karolinska Institutet, Stockholm, Sweden

I. II. III. IV. V. VI. VII.

Introduction—H. Pylori and Cancer Development Bacterial Colonization and Persistence H. pylori–Mediated Mucosal Inflammation Effects of H. pylori Infection on Epithelial Cell Turnover H. pylori–Mediated Promotion of Tumor Development Host Susceptibility Genes and Gastric Cancer Associated with H. pylori Infection Concluding Remarks References

Gastric malignancies have been closely linked to infection of the gastric mucosa with Helicobacter pylori, but the individual factors involved in the multistage process of tumor development are still poorly understood. H. pylori evades the host defense system and causes persistent infection and chronic inflammation. Immune activation leads to DNA damage by the release of oxygen and nitrogen radicals. Ongoing tissue repair mechanisms and the secretion of cytokines and growth factors, as well as bacterial effector molecules, cause disturbances in the balance between epithelial cell proliferation and apoptosis, promote the accumulation of potential oncogenic mutations, and support neovascularization and tumor growth. In addition, H. pylori might hamper the development of an efficient antitumor immunity and provoke immune-mediated pathology. This review summarizes the recent progress in the understanding of the intimate bacteria– host relationship and the mechanisms by which H. pylori may promote the process of tumor development. ß 2003 Elsevier Inc.

I. INTRODUCTION—H. PYLORI AND CANCER DEVELOPMENT A functional relationship between persistent microbial infection, inflammation, and the development of cancer was suspected as early as 1863 by Rudolf Virchow (Balkwill and Mantovani, 2003). Whereas additional exogenous factors and genetic characteristics of the infecting microorganism Advances in CANCER RESEARCH 0065-230X/03 $35.00

63

Copyright 2003, Elsevier Inc. All rights reserved

64

S. Normark et al.

and the host are a matter of much debate, the role of microbial persistence and the activation of inflammation and innate host defense in the protracted and multistage process of cancer development are now widely accepted. A total of 1.2 million cases per year, or 15% of malignancies worldwide, can be attributed to infections (Kuper et al., 2000). This proportion is even higher regarding malignancies that arise along the gastrointestinal tract. Infection with H. pylori is associated with the development of gastric cancer. This organism was also the first bacterium to be classified as a class I human carcinogen. Only a very small fraction (approximately 1%–2%) of individuals infected with H. pylori develop gastric cancer. However, the overall prevalence of H. pylori infection ranges from 25%–30% in industrialized countries to 50%–90% in less developed parts of the world (Dunn et al., 1997; Parsonnet, 1995). As such, more than half of the world’s population is infected, creating a significant number of individuals with gastric cancer as a consequence of H. pylori infection. H. pylori is a spiral-shaped, flagellated, fastidious bacterium with remarkable adaptation to its human host. Although the mode of transmission is not entirely clear, person-to-person spread seems to account for the majority of cases, but published data indicate that food and possibly water also can be vehicles of transmission (Begue et al., 1998, McKeown et al., 1999, Ozturk et al., 1996; The et al., 1994). Infection with H. pylori leads to bacterial colonization and inflammation of the gastric mucosal surface. The absolute majority of infected individuals remain asymptomatic despite the fact that superficial gastritis is always found in histopathological examination. Some individuals exhibit a more pronounced inflammatory infiltration limited to the lower part of the stomach that is associated with increased acid production and the development of peptic ulcer disease. It has been estimated that about 10%–15% of H. pylori-infected human carriers will develop peptic ulcer disease. In other patients, the inflammatory reaction progresses over time from superficial nonatrophic gastritis to a more severe form with multifocal atrophy of the gastric mucosa and the development of intestinal metaplasia. Progressive development of intestinal metaplasia enhances the risk of the development of gastric adenocarcinoma (Lauwers, 2003). Although the incidence of distal gastric cancer has been declining during the last decades, this condition remains the second leading cause of cancer mortality worldwide. A marked geographical variation has been described, with particularly high incidence found in Japan, China, and South and Central America (Kelley and Duggan, 2003). In addition to adenocarcinoma, lymphomas arising from the mucosa-associated lymphoid tissue (MALT) of the stomach have been associated with H. pylori infection (Ernst and Gold, 2000).

Persistent H. Pylori Infection

65

Infection is usually persistent, and the same organism may prevail for decades within one individual. Because of the long lag period between primary infection and the onset of gastric cancer as well as the high genetic diversity, the organism may be dramatically altered with respect to the expression of virulence attributes. The high genetic variability among H. pylori isolates is caused by DNA-transfer events between bacteria and by a high spontaneous mutation frequency (Bjo¨rkholm et al., 2001). In fact, microdiversity within the same stomach may result in the development of bacterial populations at one location in the stomach that differ from bacteria located at other sites. Truly informative epidemiological studies that aim to link H. pylori infection with cancer development therefore require a prospective case-control design within a confined recruitment area in which the genetic background of the population, as well as the infecting microbes, is as homogenous as possible. Ideally, a number of different isolates of the infecting bacterium should be recovered from each individual at the onset of the study and genetically characterized. No such defined study has been reported to date. Different types of immunoassays have therefore been used to also detect past H. pylori infection. Using an immunoblot assay to detect antibodies directed against a virulence factor (Cag A) as a marker of past infection, 71% of patients in a Swedish population with noncardia adenocarcinomas were found to have been infected with H. pylori (Ekstrom et al., 2001). Although the infection prevalence is elevated in many populations with a high risk for cancer, some regions with a high infection prevalence of H. pylori show a very low incidence of gastric cancer. This so-called African enigma remains unexplained and illustrates a very important fact: Not all H. pylori infections increase the risk of gastric cancer (Holcombe, 1992). Even more striking, peptic ulcer disease associated with H. pylori infections seems to protect the patient against gastric adenocarcinoma (Hansson et al., 1996). The factors that determine the ultimate histopathological consequences of H. pylori infection such as increased acid production and peptic ulcer disease or atrophy and intestinal metaplasia are largely unknown. This variation in the disease rate can only in part be attributed to a different distribution of virulence-associated properties among bacterial strains in different parts of the world (Bravo et al., 2002). In addition to the genetic differences of the infecting microbe, polymorphisms of host susceptibility genes and environmental factors such as diet and smoking are all factors that may ultimately determine the disease outcome. This review aims to describe the possible mechanisms involved in the development of gastric cancer following persistent infection with H. pylori (Fig. 1). It illustrates the current understanding of bacterial virulence factors and their role in microbial pathogenesis. It also points out the mechanisms by which H. pylori induces persistent infection and how chronic inflammation

66 Fig. 1 Schematic illustration of the pathophysiological mechanisms that occur after Helicobacter pylori infection and might contribute to the development of gastric cancer. IL, interleukin; MALT, mucosa-associated lymphoid tissue.

Persistent H. Pylori Infection

67

might promote tumor development. Particular emphasis is given to the intimate relationship between this highly adapted pathogen and the complex and dynamic epithelial environment along the gastric mucosal surface. It finally discusses the role of genetic diversity of the human host as well as the bacterial pathogen that enhances the risk of tumor development and sheds light on this complex and multistage process. A better understanding of the mechanisms that create the balance between H. pylori and the human host on the one hand, and of the molecular basis of pathogen-induced chronic inflammation and tumor promotion on the other, might allow us to define strategies to intervene in the early steps of the development of microbial-induced carcinogenesis.

II. BACTERIAL COLONIZATION AND PERSISTENCE A number of microbial factors have been associated with the colonization of the gastric mucosa by H. pylori. The organism invariably produces the enzyme urease, which catalyzes the hydrolysis of urea to yield ammonia and carbonic acid to raise the acid pH present at the gastric mucosal surface (Karita et al., 1995). This provides an increase in the pH surrounding the bacterium and, therefore, a protection from the acidic environment in the stomach. Although urease is not required for in vitro growth, it is necessary for colonization of the gastric mucosa and represents a critical virulence determinant. Mutants unable to produce urease were unable to colonize gnotobiotic piglets (Eaton et al., 1991). Also, H. pylori is highly motile, as a result of the production of flagella, and may be able to penetrate the mucin layer and reach the epithelial surface guided by chemotactic gradients. The importance of bacterial motility for colonization was illustrated by the finding that nonmotile/nonflagellated H. pylori mutants were unable to colonize gnotobiotic piglets for more than 2 days (Eaton et al., 1996). Similar to many other mucosal pathogens, H. pylori firmly attaches to the epithelium to prevent mechanical removal from the mucosal surface. A number of adhesins have been described, most notably BabA and SabA. BabA mediates binding to fucosylated blood group antigens such as Lewis b (Gerhard et al., 1999), whereas SabA recognizes sialyl-dimeric-Lewis x glycosphingolipid (Mahdavi et al., 2002). The density of H. pylori correlated with gastric Lewis antigen expression, particularly the expression of Lewis b (Sheu et al., 2003). Interestingly, the normal gastric epithelium does not express sialylated glycoconjugates, but sialylated derivatives are produced as a consequence of inflammation. It has therefore been suggested that H. pylori attachment is a two-step process initially involving BabA. After the initiation of an inflammatory reaction and upregulation of sialylated

68

S. Normark et al.

glycoconjugates, SabA-mediated adhesion contributes as well. Because expression of sabA is found only in a fraction of infecting microorganisms, a gastric population of H. pylori likely contains some bacteria that are capable and others that are not capable of binding to inflammatory tissue—mechanisms that might contribute to bacterial persistence (Mahdavi et al., 2002). Other adhesive mechanisms have also been described for H. pylori, such as binding to sialylated, fucosylated, and sulphated epitopes on epithelial cells, mucins, and extracellular matrix receptors. Furthermore, it has been suggested that the O-antigen of the lipopolysaccharide (LPS) is involved in adhesion. More specifically, a role for the fucosylated Lewis antigens has been implicated, as mutants deficient in the expression of these antigens show reduced adhesion to tissue sections and cultured cells and reduced colonization in mice (Edwards et al., 2000; Logan et al., 2000; Moran et al., 2000). Homotypic interactions between fucosylated Lewis x antigens of the bacterial surface LPS and corresponding structures on epithelial surfaces might be involved in such adhesion (Taylor et al., 1998). However, there are conflicting data concerning the involvement of Lewis epitopes in adhesion and colonization (Lozniewski et al., 2003; Odenbreit et al., 2002; Takata et al., 2002). Most adhesion studies have been performed on fixed tissue sections or using soluble ligands in in vitro binding assays. Fixation of cells may disrupt certain types of surface structures, thereby altering the receptor availability for the bacteria. Also, the initial contact with the epithelial cell may influence the expression of adhesins by the pathogen or the expression of host receptors and thereby induce attachment. Bacterial adherence was significantly enhanced when viable cag-pathogenicity island (PAI)-positive (a DNA region of 40 kb that has been identified as playing a central role in microbial pathogenesis of H. pylori) strains were allowed to interact with the gastric cell line AGS as compared to fixed cells. Also, inhibitors of bacterial and eukaryotic protein synthesis significantly reduced adherence of cagPAL-positive but not cag-PAI-negative isolates, indicating that adhesion depends on the upregulation of one or more host cell receptors triggered by the bacterium and requires bacteria–cell interaction (Su et al., 1998; Zhang et al., 2002). In human biopsies H. pylori is preferentially associated with tight junctions. This specific localization seems to be dependent on CagA, encoded by the cag PAI. In cell-fractionation studies, CagA was found to be associated with the epithelial tight-junction scaffolding protein ZO-1 and the transmembrane protein junctional adhesion molecule. CagA may therefore act as a multidocking protein of junctional proteins, allowing H. pylori to adhere to tight junctions. At this site, bacteria might loosen up the tight junction structure, allowing leakage across the epithelial barrier that might have long-term pathophysiological implications for the gastric mucosa, but also providing nutrients for the organism (Amieva et al., 2003). Growth and persistence of H. pylori in the gastric mucosa requires

Persistent H. Pylori Infection

69

the availability of essential growth factors and nutrients such as, for example, iron. Fe3þ may be released from the inflamed mucosa and the opening of tight junction permeability. H. pylori is growing in an environment in which access to iron varies substantially. It has recently been shown that the Fur-regulated ferritin-iron storage protein Prf of H. pylori is required for adaptation to the changes in iron availability that occur in the gastric mucosa. (Frazier et al., 1993). Also, studies using the Mongolian gerbil model demonstrated the requirement of the expression of Prf for gastric colonization (Waidner et al., 2002). Future work will most likely demonstrate that a large variety of metabolic attributes contribute to the remarkable ability of H. pylori to cause lifelong persistence in the human stomach. The vacuolating cytotoxin of H. pylori, VacA, has also been shown to increase paracellular permeability between cells to small organic molecules and ions such as Fe3þ (Pelicic et al., 1999), potentially also contributing to the organism’s ability to obtain nutrients and, indirectly, to persistence. The formed vacuole is assumed to be a hybrid of late endosome and lysosome, and it was recently suggested that syntaxin 7 is involved in the vacuolization process (Suzuki et al., 2003). The interaction of VacA with cell trafficking is further illustrated by the inhibition of phagosome maturation in macrophages (Zheng and Jones, 2003). Another independent mechanism of VacA was recently described as leading to cellular detachment via interaction with the protein tyrosine phosphatase receptor type Z (Fujikawa et al., 2003). VacA is naturally polymorphic, with the two most diverse regions being the signal region (which can be type s1 or s2) and the midregion (m1 or m2). The VacA types s1/m1 and s1/m2 are more toxic and associated with peptic ulcer and gastric cancer, whereas VacA s2/m2 strains are associated with lower peptic ulcer and gastric cancer risk and are nontoxic (Figueiredo et al., 2002). However, in Mongolian gerbils, a vacA mutant, in contrast to a cagE mutant, remained able to generate severe gastritis and gastric ulcer (Ogura et al., 2000). Hence, it is not clear whether the epidemiological link between VacA type and severe disease is the result of the toxicity of the protein itself or of an association with other disease-associated properties in Helicobacter. Many mucosal pathogens have the capacity to modulate host-cell responses by delivering effector proteins directly into the cytosol of target cells via specialized translocation systems. A type IV secretion system, which allows the delivery of at least one bacterial protein, CagA, has been found on the cag PAI (Odenbreit et al., 2000). Translocation of CagA was identified because of its phosphorylation by the host tyrosine kinase c-Src (Selbach et al., 2002a), required for subsequent interaction with SHP-2 tyrosine phosphatase at the cell membrane (Yamazaki et al., 2003). This leads to activation of the phosphatase activity and cellular morphological

70

S. Normark et al.

changes in cultured cells (Backert et al., 2001). Phosphorylated CagA inhibits the catalytic activity of c-Src and leads to dephosphorylation of the actin-binding protein cortactin, required for rearrangement of the actin cytoskeleton (Selbach et al., 2003). The observed scattering (‘‘hummingbird’’) phenotype mimics the effect obtained by the hepatocyte growth factor (Segal et al., 1999). In fact, CagA might directly interact with the hepatocyte growth factor receptor c-Met, which has been associated with invasive tumor growth (Churin et al., 2003). Nonphosphorylated CagA that interacts with Grb-2 and activates the Ras/MEK/ERK pathway induces similar cellular phenotypes and cell proliferation (Mimuro et al., 2002). It therefore remains unclear to what extent CagA-phosphorylation contributes to this phenotype. Nevertheless, CagA-induced signaling events and disruption of the epithelial barrier function may explain why cag-PAIpositive isolates are associated with more severe gastro-duodenal pathology and the risk of adenocarcinoma (Amieva et al., 2003; Farinati et al., 2003; Parsonnet et al., 1997; Ponzetto et al., 1996). In addition, there are reports that patients from a population with high risk for gastric cancer exhibit higher relative frequencies of H. pylori strains with CagA-positive and vacA s1/m1 genotypes as compared to a population from a low-risk area (Bravo et al., 2002). Signature tagged mutagenesis of the H. pylori genome and subsequent infection of Mongolian gerbils has identified 47 genes in nine different functional groups essential for gastric colonization (Kavermann et al., 2003). These genes include previously known virulence factors such as urease production and motility as well as new candidate genes; for example, the ATPase ComB4, which is involved in the type IV transport system required for DNA uptake during transformation. This transport system is distinct from the secretion system encoded by the cag PAI (Kavermann et al., 2003). Thus, pathogenic H. pylori strains encode two functionally independent type IV transport systems: one for protein translocation encoded by the cag PAI and one for uptake of DNA by natural transformation (Hofreuter et al., 2001). Knowledge of the precise role of this second type IV secretion system in H. pylori–mediated pathology will certainly shed light on the mechanisms of persistent H. pylori infection. H. pylori is generally regarded as an extracellular pathogen, but the organism is able to enter cultured gastric epithelial cells, albeit at a low efficiency, and intracellular bacteria have been detected in gastric biopsies (Ko et al., 1999). A small number of bacterial cells with a transient intracellular habitat could serve as a seeder population, providing bacterial regeneration extracellularly in the gastric mucosa. Invasion into cultured cells proceeds by a zipper mechanism that is inhibited by Wortmannin, a potent inhibitor of phosphatidylinositol 3-kinase, as well as calphostin C, an inhibitor of protein kinase C (Kwok et al., 2002). It has also been suggested

Persistent H. Pylori Infection

71

that entry could involve 5- and 1-integrin receptors (Su et al., 1999). The intracellular localization of H. pylori could therefore represent an important mechanism for microbial persistence (Bjo¨rkholm et al., 2000). A low level of intracellular replication may have significant consequences on persistence and potentially also on the inflammatory response, as has recently been demonstrated for uropathogenic E. coli, an organism previously regarded as an extracellular pathogen (Schilling and Hultgren, 2002).

III. H. PYLORI–MEDIATED MUCOSAL INFLAMMATION Infection of the gastric mucosa by H. pylori inevitably causes local inflammation and infiltration by professional immune cells. The ability of the bacterial organism to circumvent host defense and avoid elimination leads to chronic infection and persistent inflammation. The critical role of chronic inflammation in tumor development is illustrated by the diminished risk among long-term users of nonsteroidal anti-inflammatory drugs such as aspirin. Although these drugs most obviously reduce the risk of colon cancer, a significant protection was also reported for gastric adenocarcinoma (Akre et al., 2001; Zaridze et al., 1999). In addition, pseudocancerous lesions found in chronic infections have been successfully treated by removal of the infectious agent with antibiotics (Lax and Thomas, 2002). Likewise, complete regression of low-grade MALT lymphoma was described after eradication of H. pylori (Wotherspoon et al., 1993). Hence, infectionmediated perturbation of the host signaling and the induction of resistance to normal cell cycle–control mechanisms can indeed mimic the effects seen in tumorgenesis. Although direct transformation by insertion of oncogenes into the host genome is seen in some DNA viruses, a role of cellular proliferation and inflammation has been demonstrated also to be essential for virus-related tumor development (Martins-Green et al., 1994). Even though the role of chronic inflammation in the process of tumor development is widely accepted, the underlying molecular and cellular mechanisms mediating this relationship remain unsolved. Bacteria-induced inflammation is mounted primarily by the innate immune system. This evolutionary conserved immune defense system is based on the detection of molecular structures exclusively expressed by microorganisms. The recognized structures are constitutively expressed, highly conserved within whole groups of microbes, and show little variation caused by essential function. They thereby allow detection of the whole spectrum of microorganisms with a limited number of receptor molecules. The recognized structures have been named pathogen-associated molecular patterns (PAMPs);

72

S. Normark et al.

however, they are not restricted to pathogens but are also found in environmental nonpathogenic and commensal microorganisms. The best-studied molecular pattern molecule is the glycolipid lipopolysaccharide (LPS), a potent immunostimulatory molecule and essential constituent of the outer cell membrane of all Gram-negative bacteria. LPS is recognized by Toll-like receptor (TLR) 4, a member of a family of 10 transmembrane immune recognition molecules. These recently discovered receptor molecules show striking structural and functional similarities between organisms as diverse as plants, insects, and mammals and play a central role in the transcriptional activation of innate and adaptive host defense mechanisms (Medzhitov et al., 2001). Most TLRs have been designated specific microbial ligands such as peptidoglycan (TLR2), (TLR2 and TLR1, or TLR2 and TLR6), double-stranded RNA (TLR3), flagellin (TLR5), and hypomethylated DNA (TLR9). Together, TLRs can mediate recognition of the whole spectrum of infectious agents ranging from viruses to Gram-positive and Gram-negative bacteria and to fungi and parasites. Whereas the extracellular domain provides ligand specificity, the highly conserved cytoplasmic domain of all known TLRs initiates an intracellular signaling cascade via the adapter protein MyD88, leading to the activation of transcription factors such as nuclear factor B (NF-B) or activator protein 1 (Li and Verma, 2002). In addition, a less well understood second signaling pathway (frequently referred to as MyD88 independent) that leads to activation of IRF3-dependent genes has been described for a subset of TLRs, such as TLR3 and TLR4 (Kopp and Medzhitov, 2003). In addition to TLRs, other molecules such as, for example, peptidoglycan recognition proteins (PGRPs) mediate cellular activation in response to the detection of microbial structures. However, the functional importance of PGRPs in the mammalian host is only beginning to be analyzed (Dziarski et al., 2003; Gelius et al., 2003). Also, integrins situated on the cell surface have been linked to bacteria-mediated cellular activation. Integrins anchor cells to the surrounding extracellular matrix and provide cell-to-cell communication. Interestingly, integrins have also been implicated in differentiation, cell-cycle regulation, and cell proliferation (Giancotti and Ruoslahti, 1999). Also, the cell cytosol seems to be equipped to detect the presence of microbial organisms. A family of related cytosolic molecules, the nucleotide-binding oligomerization domain (NOD) proteins, has been implicated in the induction of NF-B and caspase activation (Inohara and Nunez, 2003). NOD1 and NOD2 have recently been shown to mediate the recognition of specific bacterial components. The cytosolic NOD2 (also called Card15) has been shown to efficiently detect partial structures of bacterial peptidoglycan, reminiscent of pattern molecule recognition mechanisms in the fruit fly Drosophila melanogaster (Girardin et al., 2003; Inohara et al., 2003). Recently, polymorphisms of the human Nod2 gene

Persistent H. Pylori Infection

73

have been associated with susceptibility to Crohn’s disease, a form of chronic inflammatory bowel disease (Hampe et al., 2001; Hugot et al., 2001; Ogura et al., 2001). All three identified mutations give rise to proteins that are defective in peptidoglycan-mediated NF-B activation (Girardin et al., 2003; Inohara et al., 2003). Interestingly, Crohn’s disease is associated with an increased risk for cancer. These findings therefore link Crohn’s disease and the development of gastrointestinal malignancies with microbial pattern recognition and proinflammatory stimulation. The underlying mechanisms by which NOD2 variants increase the susceptibility to Crohn’s disease are at present not understood. Failure of the initiation of host defense mechanisms may contribute to the development of Crohn’s disease (Hisamatsu et al., 2003). Alternatively, the absence of efficient microbial recognition might impede normal differentiation of the intestinal mucosa and thereby lead to unintended immunostimmulation by the physiological intestinal microflora. An important question, therefore, is whether NOD proteins are also involved in the host response to H. pylori infection or the associated pathological consequences. Activation of proinflammatory transcription factors induces expression of various genes involved in the upregulation of local host defense mechanisms (e.g., antimicrobial peptides), recruitment of professional immune cells such as monocytes and macrophages (via chemokine attraction), neoangiogenesis, and cell proliferation (via growth factors and cytokines). Stimulation experiments using a gastric epithelial cell line have shown that physical interaction between H. pylori and epithelial cells is required to evoke cellular activation and the release of proinflammatory mediators (Aihara et al., 1997). Expression of a variety of adhesins, chemokines, and cytokines such as IL-8 or MIP1 is increased following stimulation with H. pylori. Chemokines are able to attract migrating inflammatory cells to sites of infection and most likely represent one explanation for the inflammation seen in vivo, as gastric explants exposed to H. pylori also respond by increased IL-8 production (Olfat et al., 2002). This effect might be enhanced by the fact that the IL-8 receptors, IL-8RA (CXCR1) and IL-8RB (CXCR2), are upregulated in H. pylori–infected human gastric biopsy samples (Ba¨ckhed et al., 2003a). As described above, the innate host response is activated through the detection of PAMPs. In Gram-negative organisms, LPS represents the major PAMP activating innate immune responses by interacting with TLR4. Interestingly, H. pylori LPS has been shown to be a poor activator of cytokine responses in human cells, most likely because of its acylation and phosphorylation pattern (Muotiala et al., 1992). However, LPS from cag-PAI-positive H. pylori, but not from cag-PAI-negative derivatives, was shown to stimulate TLR4-mediated innate immune responses from guinea pig gastric pit cells (Kawahara et al., 2001). However, the situation in the human gastric

74

S. Normark et al.

mucosa might be different. Whereas gastric epithelial cells from guinea pigs express TLR4, cultured primary human gastric epithelial cells have been shown to lack TLR4 expression and are also nonresponsive to E. coli LPS (Ba¨ckhed et al., 2003b). In addition, cag-PAI-positive H. pylori are able to induce cytokines and chemokines in the absence of TLR4 (Maeda et al., 2001; Smith et al., 2003). However, it has recently been suggested that H. pylori may induce TLR4 expression in gastric epithelial AGS cells by a yet-unidentified mechanism (Su et al., 2003). H. pylori–mediated induction of interleukin 8 (IL-8) secretion appears to be dependent on gene products emanating from the cag-PAI (Fischer et al., 2001). Induction of IL-8 secretion depends, largely on the same set of genes that were responsible for phenotypic changes and phosphorylation of CagA (all virB/D genes and several other genes of the cag PAI) but was independent of CagA and VirD4. Thus, CagA translocation and induction of IL-8 secretion are regulated by VirD4-CagA-dependent and VirD4-CagA-independent mechanisms, respectively (Selbach et al., 2002b). Nevertheless, upregulation of interferon gamma (IFN-) IL-1, IL-1, and IL-8 and a marked hyperplasia of secondary lymphoid follicles also occur in the gastric mucosa of cats when they are infected with cag-PAI-negative H. pylori (Straubinger et al., 2003). However, the responsiveness of the feline gastric mucosa to H. pylori in relation to the human mucosa has not been characterized. The molecular mechanism of cag-PAI-mediated inflammatory responses is not known, but recent work indicates that direct cell contact is required (Aihara et al., 1997). It is possible that parts of the type IV secretion apparatus itself are recognized as a PAMP. Alternatively, translocated effector proteins might be involved in cellular activation and recognized by a cytosolic receptor. The intracellular pathogen Shigella, which propagates in the host cytoplasm, was shown to activate cytokine production via Card15/NOD2. The ligand for Card15/NOD2 appears to be muramyl dipeptide, whereas Nod I recognizes diaminopimelic acid containing muropeptides (Giardin et al., 2003). Because H. pylori produces DAP, muropeptides derived from its peptidoglycan may be recognized by this intracellular recognition system. In addition to the recognition of bacterial ligands by the host innate immune system and the secretion of proinflammatory cytokines, H. pylori itself may produce chemoattractive mediators. The organism has been shown to produce a neutrophil-activating protein, HP-NAP, that activates polymorphonucelar cells (PMNs), monocytes, and mast cells—potentially through the MAPK pathway (Nishioka et al., 2003). In addition, a proinflammatory outer membrane protein, OipA, has recently been identified (Yamaoka et al., 2000). Thus, H. pylori may induce mucosal inflammation simultaneously via several mechanisms.

Persistent H. Pylori Infection

75

In addition to the activation of host defense effector mechanisms, activation of the innate immune system is a prerequisite for the activation of the adaptive immune system (Schnare et al., 2001). Even though H. pylori infection elicits specific antibodies, humoral responses are not capable of clearing the infection. A number of reasons may explain this finding. H. pylori, similar to human-specific mucosal pathogens, can undergo frequent phase and antigen variation (Appelmelk et al., 1998). This is particularly evident for expression of the outer cell membrane constituent LPS. H. pylori LPS of a given strain varies extensively both with regard to the number of carbohydrate repeats and with respect to the expression of Lewis x and Lewis y antigens on the LPS repeats (Wang et al., 2000). The variation of the LPS structure may be so extensive that virtually every isolated bacterium shows a different structural pattern even though the patient is colonized by one single strain. In addition, the vacuolating toxin VacA is known to interfere with the degradation of microbial antigens, an essential step in the process of antigen presentation and initiation of adaptive immune responses (Hornef et al., 2002; Molinari et al., 1998). Recent results also indicate specific impairment of the cell-mediated antibacterial immune response. Memory cells isolated from H. pylori–infected individuals showed less responsiveness toward bacterial antigen as compared to cells derived from noninfected controls. This defect was caused by specific regulatory CD4þ CD25 high T cells, as depletion of this cell population restored the functional response (Lundgren et al., 2003). Furthermore, cellular impairment may be aggravated by the cag-PAI-induced Fas ligand (FasL) expression on T lymphocytes and cellular depletion by apoptosis (Wang et al., 2001). In general, inflammation represents a self-limited program of consequent defined steps that ensure removal of the destructive agent and provide tissue repair. Infection and colonization of the gastric mucosal surface by H. pylori evokes a profound inflammatory reaction and host defense activation. Likewise, the recruitment and activation of professional phagocytes, the production of reactive nitrogen and oxygen intermediates, and the secretion of antimicrobial peptides may in fact significantly impede bacterial growth. For example, H. pylori infection was shown to induce human -defensin 2 (hBD-2) expression in the human gastric epithelium, and hBD-2 inhibited the growth of H. pylori in vitro (Hamanaka et al., 2001). Nevertheless, activation of the immune system fails to eradicate the infecting agent, leading to persistent infection and chronic inflammation. Bacterial strategies to circumvent recognition or removal by the immune system therefore seem to cause microbial persistence and ongoing inflammation, promoting tumor progression (Rosenblatt et al., 2001).

76

S. Normark et al.

IV. EFFECTS OF H. PYLORI INFECTION ON EPITHELIAL CELL TURNOVER The gastric epithelial surface forms glandlike tubular invaginations composed of a variety of specialized epithelial lineages. Mucus-producing pit cells and surface mucus cells migrate upward, whereas mucous neck cells move downward to the lower end and differentiate into pepsinogen-secreting zymogenic cells. The low pH is provided by parietal cells in the middle part of this glandlike structure. Enterochromaffin-like cells control the physiological function of the stomach by the secretion of regulatory mediators such as gastrin, histamin, or somatostatin. Perpetual regeneration is fuelled by a population of multipotent stem cells located in the central portion (isthmus) of these so-called gastric units (Karam et al., 1997). H. pylori infection induces alterations in epithelial cell migration, cell–cell contact, and cellular binding to extracellular matrix (Lim et al., 2003; Wroblewski et al., 2003). For example, H. pylori induces expression of the metalloproteinase 7, which was associated with enhanced spreading of gastric gland cells and may thereby contribute to disturbances of epithelial cell migration and the development of gastric neoplasia (Wroblewski et al., 2003). However, the variety of different cell populations and the continuous process of cellular differentiation along the gastric mucosal surface renders the interpretation of studies on single cell lines or isolated primary cells difficult to interpret. In mice, real-time quantitative reverse transcripts polymerase chain reaction (RT-PCR) studies of laser-capture microdissected cells retrieved from gastric epithelial progenitor cells have shown enriched expression of a selected set of genes (Mills et al., 2002). Similar studies using gastric tissue of mice infected with H. pylori are likely to give us more precise information on how H. pylori may affect the developmental pathway from a progenitor cell to a fully differentiated gastric epithelial cell. For example, infection of mice with H. pylori strain HP1 induced expression of 99 unique genes from parietal cells. Of these, only 16 genes had previously been associated with H. pylori infection (Mills et al., 2002). An important problem remains; to follow a H. pylori infection for longer than 2 weeks and to monitor how the expression profile of individual cell lineages are modulated over time. The low pH and enzymatically active content of the gastric lumen provide an explanation for the need of rapid epithelial cell renewal to ensure integrity of the mucosal surface. Cellular homeostasis is facilitated by cellular exfoliation and programmed cell death (Karam and LeBlond, 1993). Infection with H. pylori severely affects this delicate equilibrium and might thereby lower the threshold for tumor development. The bacteria-induced inflammatory reaction, but also specific bacterial gene products, alter the cellular homeostasis. VacA, CagE, and -glutamyl transpeptidase have been

Persistent H. Pylori Infection

77

reported to enhance apoptosis of gastric epithelial cells (Cover et al., 2003; Maeda et al., 2002; Shibayama et al., 2003). Whereas VacA and -glutamyl transpeptidase induced apoptosis as soluble proteins, the CagE-mediated effect was dependent on bacterial viability and direct cell contact consistent with delivery through a bacterial translocation system encoded by the cag PAI (Israel et al., 2001). The individual role of these bacterial virulence factors during gastric inflammation and atrophy is unclear. H. pylori– induced apoptosis of gastric epithelial cells seems to largely depend on the presence of the cag PAI and was proposed to be mediated via the mitochondrial pathway and Fas expression (Jones et al., 1999; Maeda et al., 2002). Whereas bacteria-induced Smad5 promotes apoptosis, peroxysome proliferator-activated receptor and extracellular signal-regulated kinase 1/2 may play a protective role (Choi et al., 2003; Gupta et al., 2001; Nagasako et al., 2003). In addition, antiapoptotic effects have been noted during H. pylori–mediated inflammation. This cag-PAI-dependent effect was mediated through activation of NF-B (Jones et al., 1999; Maeda et al., 2002). Interestingly, the kinetics of apoptotic and proliferative activity seem to deviate: Whereas the initial increase of the apoptosis rate declined early after H. pylori infection, the strongest proliferative effect was seen at a later time point (Peek 2002; Peek et al., 2000). Cellular proliferation was correlated with serum gastrin levels and enhanced expression of endothelial growth factor and EGF receptor in vivo (Peek et al., 2000; Yao et al., 2002). Thus, the combination of increased cell proliferation without enhanced rate of apoptosis may significantly contribute to the accumulation of tumor-promoting mutations and enhance the risk for gastric cancer.

V. H. PYLORI–MEDIATED PROMOTION OF TUMOR DEVELOPMENT Recruitment and tissue infiltration by inflammatory phagocytes has several consequences that have been associated with tumor promotion. The production of reactive nitrogen and oxygen species is enhanced, leading to the production of peroxynitrite, a recognized mutagenic agent that induces DNA damage (Maeda and Akaike, 1998). The degree of inflammationmediated DNA damage is illustrated by the finding that p53 mutations are found at similar frequencies in patients with chronic inflammatory bowel disease as compared to patients with tumors (Yamanishi et al., 2002). Infection of mice with H. pylori does not lead to the development of cancer, unless these animals are also exposed to chemical mutagens. Nevertheless, the murine infection has provided important clues regarding the cancerogenic

78

S. Normark et al.

potential role of H. pylori infection. For example, chronic infection of BALB/c mice has demonstrated that the mutation frequency of gastric cells is higher in mice infected with H. pylori, as compared with control mice. This higher mutation rate was associated with a high frequency of transversions that are known to result from oxidative damage, a consequence of the inflammatory reaction (Touati et al., 2003). In addition, the release of a variety of chemotactic factors, growth hormones, and cytokines leads to protracted growth stimulation. This promotes tumor formation by facilitating the acquisition of mutations in genes encoding the signaling and cell-cycle proteins that control proliferation. Angiogenic and lymphangiogenic growth factors and cytokines have been implicated in tumor progression (Richmond and Thomas, 1986; Schoppmann et al., 2002). A link to microbial infection was established by the finding that the indigenous bacterial flora play a key role in development of vascular network in the mouse intestine (Stappenbeck et al., 2002). Furthermore, bacterial LPS directly induced angiogenesis, vascular permeability, and tumor cell invasion (Pollet et al., 2003). Enhanced expression of migration inhibitory factor during inflammation leads to suppression of the transcriptional activity of p53. This facilitates enhanced proliferation, extended cellular life span, and reduced DNA repair (Hudson et al., 1999). In addition, infiltrating macrophages seem to promote invasive tumor growth and the spread of tumor cells to distant anatomical sites (Lin et al., 2001). In addition, H. pylori may also directly exert tumor-promoting activity. The cag PAI, and thus a functional type IV secretion system, has recently been shown to be involved in the increased transcription of the cell-cycle regulatory protein cyclin D1, as illustrated by the reduced ability of cagE and cag PAI mutants to induce production of cyclin D1 (Hirata et al., 2001). Upregulation of cyclin D1 has been associated with poor prognosis in lung and colon cancer (Keum et al., 1999). Furthermore, CagE promotes activation of cyclooxigenase 2, c-Jun amino-terminal kinase, and phospholipase A2, known to be associated with carcinogenesis (Nardone et al., 2001; Mitsuno et al., 2001; Romano et al., 1998). Although wild-type mice are resistant to H. pylori–induced carcinogenesis, gastric adenocarcinoma occurs regularly in a transgenic hypergastrinemic (INS-GAS) mouse model after H. pylori infection. Male, but not female, INS-GAS mice infected with H. pylori and fed on a high-salt diet develop atrophy, intestinal metaplasia, and dysplasia by 6 weeks and carcinoma 24 weeks after infection. Interestingly, infection of mice with CagE-deficient H. pylori delayed the progression to cancer (Fox et al., 2003b). Thus, the INS-GAS mouse model provides evidence that hypergastrinemia, high-salt diet, male sex, and H. pylori infection (particularly with cag-PAI-positive strains) promote

Persistent H. Pylori Infection

79

cancer development (Fox et al., 2003a). Hypergastrinemia is also seen in humans infected with H. pylori and is believed to contribute to altered acid secretion and loss of acid-producing parietal cells. A large number of epidemiological studies have been conducted with the aim of identifying H. pylori properties associated with subsequent cancer development. Most such studies found that the presence of the cag-PAI as such represents a tumor-promoting property, as do expression of the BabA and SabA adhesins as well as the s1/m1 type of vacuolating cytotoxin VacA. When coexpressed by the same H. pylori strain, cagA, vacA s1/m1, and babA work synergistically in enhancing inflammation. Individuals infected by strains coexpressing these three genes/alleles are believed to be at higher risk for intestinal metaplasia (Zambon et al., 2003). Hence, the degree of the inflammatory response to H. pylori infection may be a central factor for the development of gastric cancer. In most infected individuals, both cag-positive and cag-negative bacteria may coexist, and most likely the cag-negative bacterial strains are deletion derivatives of the cag-PAI-positive ancestors. Alternatively, the cag-island might be transferred from cag-positive to cag-negative bacteria via horizontal gene transfer. Thus, the ratio between cag-positive and cag-negative bacteria within the stomach of a given individual may be one determinant that affects the degree of inflammation. In addition to cell-cycle control mechanisms and growth arrest, the adaptive immune system exerts some protection against cells undergoing unrestricted proliferation. This antitumor immunity is mainly mediated by specific T lymphocytes and depends on the presence of mature antigenpresenting cells to present tumor antigens and induce lymphocyte proliferation and differentiation. H. pylori seems to have evolved mechanisms that interfere with the development of an effective antitumor immunity. In addition to the inhibition of efficient antigen presentation and T cell function (Hornef et al., 2002; Lundgren et al., 2003; Molinari et al., 1998; Wang et al., 2001), macrophages activated by the cecropin-like peptide Hp(220) from H. pylori specifically inhibited natural killer cells with antitumor properties (Betten et al., 2001). Finally, nitric oxide, in addition to its DNA-damaging action, was shown to impair T cell–mediated cytotoxicity and antitumor immune surveillance (Zhan and Xu, 2001). In addition to tumor-promoting mechanisms, inhibition of antitumor immunity might therefore contribute to the role of bacterial infection in oncogenesis. Moreover, H. pylori–induced activation of the adaptive immune system may also contribute to disease progression. Autoantibodies against Lewis carbohydrate epitopes induced by infection with H. pylori cross react with epitopes on acid-producing parietal cells. The concomitant loss of parietal cells could result in hyperproliferation of gastric stem cells and adenomatous

80

S. Normark et al.

lesions (Guruge et al., 1998). A Lewis b transgenic mouse model, allowing H. pylori BabA mediated adhesion to gastric epithelial cells, has been used to demonstrate parietal cell loss. This loss was associated with an enhanced production of antibodies against bacterial and parietal Lewis x glycans, suggesting indeed an autoimmune response as the cause for parietal cell disappearance (Guruge et al., 1998). Transgenic tox176 mice with an engineered ablatio of parietal cells are achlorhydric, lack parietal and zymogenic cells, and exhibit expression of sialylated glycoconjugates that act as receptors for H. pylori. In these mice there was an expansion of H. pylori colonization and a development of lymphoid aggregates in the formerly acid-protected parts of the gastric glandular epithelium (Syder et al., 2003). Persistent lymphocyte stimulation may further contribute to the development of MALT lymphoma, a gastric B cell lymphoma that occasionally develops in humans as a consequence of long-term H. pylori infection (Ahmad et al., 2003). The initial stages of lymphoma development are characterized by infiltration of reactive lymphocytes into the gastric tissue, the formation of lymphoid follicles, and clonal expansion. This process is also seen in H. pylori–infected mice. Microarray analysis and laser capture microdissection has allowed the identification of gene-expression patterns during these early stages of lymphoma development. On the appearance of lymphoepithelial lesions, an increased expression of genes previously associated with malignancy, including the laminin receptor-1 and the multidrug-resistance channel MDR-1, was observed. Transition to destructive tumor growth was accompanied by calgranulin A/Mpr-8 expression, previously associated with carcinogenesis (Mueller et al., 2003). In contrast to the mouse model, infection of Mongolian gerbils with H. pylori generates many of the pathophysiological characteristics that are also seen in humans. An acute antral gastritis develops after 4 weeks, concomitant with antral epithelial cell proliferation (Court et al., 2002), and in an early stage of infection, enhanced antral apoptosis is observed. Epithelial cell proliferation peaks later and is related to increased gastrin levels, indicating that enhanced epithelial cell proliferation in the H. pylori–colonized mucosa is mediated by a gastrin-dependent mechanism (Watanabe et al., 1998). Long-term colonization of Mongolian gerbils results in atrophic gastritis, intestinal metaplasia, and—in a significant portion of animals—the development of adenocarcinoma that develops in the antral region of the stomach. These tumors are composed of well-differentiated intestinal-type cells and arise in close relationship to intestinal metaplasia (Honda et al., 1998; Watanbe et al., 1998). The Mongolian gerbil model has also been used to monitor the enhancing effects of chemical mutagens and salt intake on H. pylori–induced carcinogenesis (Bergin et al., 2003; Shimizu et al., 1999).

Persistent H. Pylori Infection

81

VI. HOST SUSCEPTIBILITY GENES AND GASTRIC CANCER ASSOCIATED WITH H. PYLORI INFECTION In addition to genetic heterogeneity between bacterial strains associated with H. pylori–mediated disease, genetic diversity of the host population also has been suggested to explain the strikingly different outcome of gastric colonization with H. pylori in different individuals. A large Scandinavian twin study has revealed a significant contribution of heredity in gastric cancer (Lichtenstein et al., 2000). Development and progression of gastric cancer has been correlated with polymorphisms in different human cytokine genotypes associated with a more pronounced proinflammatory response. Proinflammatory polymorphisms within the genes encoding IL-1, IL-1 receptor antagonist (IL-1 RA), IL-10, tumor necrosis factor (TNF)-, and lymphotoxin (LTA)- as well as short alleles of the mucosa protecting mucin (MUC1) and certain HLA haplotypes have been linked to an increased risk of H. pylori–associated disease (El-Omar et al., 2003; El-Omar et al., 2000; Figueiredo et al., 2002; Lanas et al., 2001; Sakai et al., 1999; Vinhall et al., 2002; Wu et al., 2003). IL-1 and TNF-, as well as IL-10, represent central mediators of the inflammatory reaction and influence a large variety of other potent pro- and anti-inflammatory mediators; IL-1 is also involved in the regulation of gastric acid secretion. The glutathione S-transferase GSTT1 and GSTM1 that facilitate detoxification of carcinogens, as well as genes encoding proteins involved in the cellular antioxidant capacity and protection against DNA-damaging reactive oxygen species, have been linked to an increased gastric cancer risk (Gonzales et al., 2002; Rollinson et al., 2003). Carriage of multiple associated host polymorphisms or the combination of genetic risk factors of the host and of the bacterium potentiate the risk of gastric cancer (El-Omar et al., 2003; Figueireo et al., 2002). Notably, no association of a genetic proinflammatory profile was found with other upper gastrointestinal cancers such as cardia adenocarcinoma as well as esophageal squamous cell carcinoma and adenocarcinoma (El-Omar et al., 2003). Thus, the association of proinflammatory and protective polymorphisms with H. pylori–associated pathology supports the hypothesis of an infection-induced proinflammatory and destructive process in gastric tumor development. However, conflicting results for some attributed risk factors as well as small sample size in most studies performed preclude a definite conclusion (Gonzales et al., 2002). In addition, the contribution of individual polymorphisms during the chronic H. pylori–mediated inflammatory process has not been analyzed in vivo. Environmental risk factors for gastric cancer may in part impede direct comparison between different populations with different cultural and

82

S. Normark et al.

socioeconomic background and nutrition habits, thereby complicating interpretation of genetic host susceptibility factors. Population-based case–control studies have demonstrated a protective role of high intake of antioxidants in individuals colonized with H. pylori, providing an epidemiological support for a link between H. pylori–induced inflammation, increased rate of mutations through oxidative products, and gastric cancer (Ekstrom et al., 2000). Thus, dietary factors such as high fruit and vegetable intake with high concentrations of antioxidants such as ascorbic acid (Vitamin C), -tocopherol (vitamin E), or -carotenoids may provide protection from tumor development (Hansson et al., 1993; Lee et al., 2003). However, prospective studies on the value of prophylactic vitamin C supplementation have generated conflicting results (Blot et al., 1993; Sasazuki et al., 2003). In contrast, high-salt or nitrite diet have been associated with a moderately increased risk of gastric cancer (Ernst and Gold, 2000; Lee et al., 2003). Notably, H. pylori infection itself is inversely correlated with fruit and vegetable intake, and a positive association exists between the consumption of preserved meats and high salt or high nitrate intake. Not surprisingly, ionizing radiation, smoking (which itself is also positively associated with H. pylori infection), and pernicious anemia also have been associated with gastric cancer (Hansson et al., 1993). It is currently unclear to what extent the observed environmental risk factors reflect independent tumor-promoting factors or influence the process of H. pylori infection, colonization, and chronic inflammation of the gastric mucosal surface.

VII. CONCLUDING REMARKS Much has to be learnt before the process of H. pylori–associated oncogenesis will be completely understood. Bacterial virulence factors, predisposing genetic factors of the host, the process of microbial recognition, and perpetuation of the inflammatory response as well as environmental factors all seem to work together to cause the development of gastric malignancies. Although the prevalence of H. pylori infection in most parts of the world remains high, a significant decline has been noted in the population of the Western world. The disappearance of H. pylori is correlated with a dramatic decline in the incidence of gastric noncardia cancer over the last halfcentury. Factors including improved sanitation, widespread use of antibiotics, increased consumption of fruits and vegetables, and decreased intake of salt have largely been credited for the decline in gastric cancer. There is, however, an ongoing debate as to whether this decline bears only advantages or whether the colonization with H. pylori and subsequent low-grade inflammatory response evoked in most healthy individuals infected could

Persistent H. Pylori Infection

83

provide benefit for the host. Thus, the discussion about the role of H. pylori in human gastric disease and the molecular mechanisms of microbial pathogenesis and host immune defense activation will continue to attract considerable attention. Certainly this will enforce a better understanding of the intimate relationship between this highly adapted mircoorganism and the human host.

REFERENCES Ahmad, A., Govil, Y., and Frank, B. B. (2003). Am. J. Gastroenterol. 98, 975–986. Aihara, M., Tsuchimoto, D., Takizawa, H., Azuma, A., Wakebe, H., Ohmoto, Y., Imagawa, K., Kikuchi, M., Mukaida, N., and Matsushima, K. (1997). Infect. Immun. 65, 3218–3224. Akre, K., Ekstrom, A. M., Signorello, L. B., Hansson, L. E., and Nyren, O. (2001). Br. J. Cancer 84, 965–968. Amieva, M. R., Vogelmann, R., Covacci, A., Tompkins, L. S., Nelson, W. J., and Falkow, S. (2003). Science 300, 1430–1434. Appelmelk, B. J., Shiberu, B., Trinks, C., Tapsi, N., Zheng, P. Y., Verboom, T., Maaskant, J., Hokke, C. H., Schiphorst, W. E., Blanchard, D., Simoons-Smit, I. M., van den Eijnden, D. H., and Vandenbroucke-Grauls, C. (1998). Infect. Immun. 66, 70–76. Backert, S., Moese, S., Selback, M., Brinkmann, V., and Meyer, T. F. (2001). Mol. Microbiol. 42, 631–644. Ba¨ckhed, F., Torstensson, E., Seguin, D., Richter-Dahlfors, A., and Rokbi, B. (2003a). Infect. Immun. 71, 357–360. Ba¨ckhed, F., Rokbi, B., Torstensson, E., Zhao, Y., Nilsson, C., Seguin, D., Normark, S., Buchan, A. M., and Richter-Dahlfors, A. (2003b). J. Infect. Dis. 187, 829–836. Balkwill, F., and Mantovani, A. (2003). Lancet 357, 539–545. Begue, R. E., Gonzales, J. L., Correa-Gracian, H., and Tang, S. C. (1998). Am. J. Trop. Med. Hyg. 59, 637–640. Bergin, I. L., Sheppard, B. J., and Fox, J. G. (2003). Di. Dis. Sci. 48, 475–485. Betten, A., Bylund, J., Cristophe, T., Boulay, F., Romero, A., Hellstrand, K., and Dahlgren, C. (2001). J. Clin. Invest. 108, 1221–1228. Bjo¨rkholm, B., Sjolund, M., Falk, P. G., Berg, O. G., Engstrand, L., and Andersson, D. I. (2001). Proc. Natl. Acad. Sci. USA 98, 14607–14612. Bjo¨rkholm, B., Zhukhovitsky, V., Lofman, C., Hulten, K., Enroth, H., Block, M., Rigo, R., Falk, P., and Engstrand, L. (2000). Helicobacter 5, 148–154. Blot, W. J., Li, J. Y., Taylor, P. R., Guo, W., Dawsey, S., Wang, G. Q., Yang, C. S., Zheng, S. F., Gail, M., and Li, G. Y. (1993). J. Natl. Cancer Inst. 85, 1483–1492. Bravo, L. E., van Doom, L. J., Realpe, J. L., and Correa, P. (2002). Am. J. Gastroenterol. 97, 2839–2842. Choi, I. J., Kim, J. S., Kim, J. M., Jung, H. C., and Song, I. S. (2003). Infect. Immun. 71, 830–837. Churin, Y., Al-Ghoul, L., Kepp, O., Meyer, T. F., Birchmeier, W., and Naumann, M. (2003). J. Cell Biol. 161, 249–255. Court, M., Robinson, P. A., Dixon, M. F., and Crabtree, J. E. (2002). J. Infec. Dis. 186, 1348–1352. Cover, T. L., Krishna, U. S., Israel, D. A., and Peek, R. M., Jr. (2003). Cancer Res. 63, 951–957. Dunn, B. E., Cohen, H., and Blaser, M. J. (1997). Clin. Microbiol. Rev. 10, 720–741.

84

S. Normark et al.

Dziarski, R., Platt, K. A., Gelius, E., Steiner, H., and Gupta, D. (2003). Blood 102, 689–697. Eaton, K. A., Brooks, C. L., Morgan, D. R., and Krakowka, S. (1991). Infect. Immun. 59, 2470–2475. Eaton, K. A., Suerbaum, S., Josenhans, C., and Krakowka, S. (1996). Infect. Immun. 64, 2445–2448. Edwards, N. J., Monteiro, M. A., Faller, G., Walsh, E. J., Moran, A. F., Roberts, I. S., and High, N. J. (2000). Mol. Microbiol. 35(6), 1530–1539. Ekstrom, A. M., Serafini, M., Nyren, O., Hansson, L. E., Ye, W., and Wolk, A. (2000). Int. J. Cancer 87, 133–140. Ekstrom, A. M., Held, M., Hansson, L. E., Engstrand, L., and Nyren, O. (2001). Gastroenterology 121, 784–791. El-Omar, E. M., Carrington, M., Chow, W.-H., McColl, K. E. L., Bream, J. H., Young, H. A., Herrera, J., Lissowska, J., Yuan, C.-C., Rothman, N., Lanyon, G., Martin, M., Fraumeni, J. F., Jr., and Rabkin, C. S. (2000). Nature 404, 398–401. El-Omar, E. M., Rabkin, C. S., Gammon, M. D., Vaughan, T. L., Risch, H. A., Schoenberg, J. B., Stanford, J. L., Mayne, S. T., Goedert, J., Blot, W. J., Fraumeni, J. F., Jr., and Chow, W. H. (2003). Gastroenterology 124, 1193–1201. Ernst, P. B., and Gold, B. D. (2000). Annu. Rev. Microbiol. 54, 615–640. Farinati, F., Cardin, R., Russo, V. M., Busatto, G., Franco, M., and Rugge, M. (2003). Helicobacter 8, 227–234. Figueiredo, C., Machado, J. C., Pharoah, P., Seruca, R., Sousa, S., Carvalho, R., Capelinha, A. F., Quint, W., Caldas, C., van Doorn, L. J., Carneiro, F., and Sobrinho-Simones, M. (2002). J. Natl. Cancer Inst. 94, 1680–1687. Fischer, W., Puls, J., Buhrdorf, R., Gebert, B., Odenbreit, S., and Haas, R. (2001). Mol. Microbiol. 42, 1337–1348. Fox, J. G., Rogers, A. B., Ihrig, M., Taylor, N. S., Whary, M. T., Dockray, G., Varro, A., and Wang, T. C. (2003a). Cancer Res. 63, 942–950. Fox, J. G., Wang, T. C., Rogers, A. B., Poutahidis, T., Ge, Z., Taylor, N., Dangler, C. A., Israel, D. A., Krishna, U., Gaus, K., and Peek, R. M. (2003b). Gastroenterology 124, 1879–1890. Frazier, B. A., Pfeifer, J. D., Russell, D. G., Falk, P., Olsen, A. N., Hammar, M., Westblom, T. U., and Normark, S. (1993). J. Bacteriol. 175, 966–972. Fujikawa, A., Shirasaka, D., Yamamoto, S., Ota, H., Yahiro, K., Fukada, M., Shintani, T., Wada, A., Aoyama, N., Hirayama, T., Fukamachi, H., and Noda, M. (2003). Mice deficient in protein tyrosine. Nat. Genet. 33, 375–581. Gelius, E., Persson, C., Karlsson, J., and Steiner, H. (2003). Biochem. Biophys. Res. Commun. 306, 988–994. Gerhard, M., Lehn, N., Neumayer, N., Boren, T., Rad, R., Schepp, W., Miehlke, S., Classen, M., and Prinz, C. (1999). Proc. Natl. Acad. Sci. USA 96, 12778–12783. Giancotti, F. G., and Ruoslahti, E. (1999). Science 285, 1028–1032. Girardin, S. E., Boneca, I. G., Viala, J., Chamaillard, M., Labigne, A., Thomas, G., Philpott, D. J., and Sansonetti, P. J. (2003). J. Biol. Chem. 278, 8869–8872. Giardin, S. E., Travassas, L. H., Haue, M., Blanot, D., Boneca, I. G., Philpott, D. J., Sansonetti, P. J., and Mengin-Decreulx, D. (2003). J. Biol. Chem., in press. Gonzales, C. A., Sala, N., and Capella, G. (2002). Int. J. Cancer 100, 249–260. Gupta, R. A., Polk, D. B., Krishna, U., Israel, D. A., Yan, F., DuBois, R. N., and Peek, R. M. (2001). J. Biol. Chem. 276, 31059–31066. Guruge, J. L., Falk, P. G., Lorenz, R. G., Dans, M., Wirth, H. P., Blaser, M. J., Berg, D. E., and Gordon, J. I. (1998). Proc. Natl. Acad. Sci. USA 95, 3925–3930. Hamanaka, Y., Nakashima, M., Wada, A., Ito, M., Kurazono, H., Hojo, H., Nakahara, Y., Kohno, S., Hirayama, T., and Sekine, I. (2001). Gut 49, 481–487.

Persistent H. Pylori Infection

85

Hampe, J., Cuthbert, A., Croucher, P. J., Mirza, M. M., Mascheretti, S., Fisher, S., Frenzel, H., King, K., Hasselmeyer, A., MacPherson, A. J., Bridger, S., van Deventer, S., Forbes, A., Nikolaus, S., Lennard-Jones, J. E., Foelsch, U. R., Krawczak, M., Lewis, C., Schreiber, S., and Mathew, C. G. (2001). Lancet 357, 1925–1928. Hansson, L. E., Nyren, O., Bergstrom, R., Wolk, A., Lindgren, A., Baron, J., and Adami, H. O. (1993). Int. J. Cancer 55, 181–189. Hansson, L. E., Nyren, O., Hsing, A. W., Bergstro¨m, R., Josefsson, S., Chow, W.-H., Fraumeni, J. F., and Adami, H.-O. (1996). N. Engl. J. Med. 335, 242–249. Hirata, Y., Maeda, S., Mitsuno, Y., Akanuma, M., Yamaji, Y., Ogura, K., Yoshida, H., Shiratori, Y., and Omata, M. (2001). Infect. Immun. 69, 3965–3971. Hisamatsu, T., Suzuki, M., Reinecker, H. C., Nadeau, W. J., McCormick, B. A., and Podolsky, D. K. (2003). Gastroenterology 124, 993–1000. Hofreuter, D., Odenbreit, S., and Haas, R. (2001). Mol. Microbiol. 41, 379–391. Holcombe, C. (1992). Gut 33, 429–431. Honda, S., Fujioka, T., Tokieda, M., Satoh, R., Nishizono, A., and Nasu, M. (1998). Cancer Res. 58, 4255–4259. Hornef, M. W., Wick, M. J., Rhen, M., and Normark, S. (2002). Nat. Immunol. 3, 1033–1037. Hudson, J. D., Shoaibi, M. A., Maestro, R., Carnero, A., Hannon, G. J., and Beach, D. H. (1999). J. Exp. Med. 190, 1375–1382. Hugot, J. P., Chamaillard, M., Zouali, H., Lesage, S., Cezard, J. P., Belaiche, J., Almer, S., Tysk, C., O’Morain, C. A., Gassull, M., Binder, V., Finkel, Y., Cortot, A., Modigliani, R., LaurentPuig, P., Gower-Rousseau, C., Macry, J., Colombel, J. F., Sahbatou, M., and Thomas, G. (2001). Nature 411, 599–603. Inohara, I., and Nunez, G. (2003). Nat. Rev. 3, 371–381. Inohara, N., Ogura, Y., Fontalba, A., Gutierrez, O., Pons, F., Crespo, J., Fukase, K., Inamura, S., Kusumoto, S., Hashimoto, M., Foster, S. J., Moran, A. P., Fernandez-Luna, J. L., and Nunez, G. (2003). J. Biol. Chem. 278, 5509–5512. Israel, D. A., Salama, N., Arnold, C. N., Moss, S. F., Ando, T., Wirth, H. P., Tham, K. T., Canorlinga, M., Blaser, M. J., Falkow, S., and Peek, R. M., Jr. (2001). J. Clin. Invest. 107, 611–620. Jones, N. L., Day, A. S., Jennings, H. A., and Sherman, P. M. (1999). Infect. Immun. 67, 4237–4242. Karam, S. M., and Leblond, C. P. (1993). Anat. Rec. 236, 333–340. Karam, S. M., Li, Q., and Gordon, J. I. (1997). Am. J. Physiol. 272, G1209–G1220. Karita, M., Tsuda, M., and Nakazawa, T. (1995). J. Clin. Gastroenterol. 21, S160–S163. Kavermann, H., Burns, B. P., Angermuller, K., Odenbreit, S., Fischer, W., Melchers, K., and Haas, R. (2003). J. Exp. Med. 197, 813–822. Kawahara, T., Teshima, S., Oka, A., Sugiyama, T., Kishi, K., and Kokutan, K. (2001). Infect. Immun. 69, 4382–4389. Kelley, J. R., and Duggan, J. M. (2003). J. Clin. Epidemiol. 56, 1–9. Keum, J. S., Kong, G., Yang, S. C., Shin, D. H., Park, S. S., Lee, J. H., and Lee, J. D. (1999). Br. J. Cancer 81, 127–132. Ko, G. H., Kang, S. M., Kim, Y. K., Lee, J. H., Park, C. K., Youn, H. S., Baik, S. C., Cho, M. J., Lee, W. K., and Rhee, K. H. (1999). Helicobacter 4, 77–81. Kopp, E., and Medzhitov, R. (2003). Curr. Opin. Immunol. 15, 396–401. Kuper, H., Adami, H. O., and Trichopoulos, D. (2000). J. Intern. Med. 248, 171–183. Kwok, T., Backert, S., Schwarz, H., Berger, J., and Meyer, T. F. (2002). Infect. Immun. 70, 2108–2120. Lanas, A., Garcia-Gonzales, M. A., Santolaria, S., Crusius, J. B., Serrano, M. T., Benito, R., and Peno, A. S. (2001). Genes. Immun. 2, 415–421. Lauwers, G. Y. (2003). J. Clin. Gastroenterol. 36, S37–S43.

86

S. Normark et al.

Lax, A. J., and Thomas, W. (2002). Trends Microbiol. 10, 293–299. Lee, S. A., Kang, D., Shim, K. N., Choe, J. W., Hong, W. S., and Choi, H. (2003). J. Epidemiol. 13, 162–168. Li, Q., and Verma, I. M. (2002). Nat. Rev. Immunol. 2, 725–734. Lichtenstein, P., Holm, N. V., Verkasalo, P. K., Iliadou, A., Kaprio, J., Koskenvuo, M., Pukkala, E., Skytthe, A., and Hemminki, K. (2000). N. Engl. J. Med. 343, 78–85. Lim, J. W., Kim, H., and Kim, K. H. (2003). In. J. Biochem. Cell Biol. 35, 1284–1296. Lin, E. Y., Nguyen, A. V., Russell, R. G., and Pollard, J. W. (2001). J Exp. Med. 193, 727–740. Logan, S. M., Conlan, J. W., Monteiro, M. A., Wakarchuk, W. W., and Altman, E. (2000). Mol. Microbiol. 35(5), 1156–1167. Lozniewski, A., Haristoy, X., Rasko, D. A., Hatier, R., Plenat, F., Taylor, D. E., and AngioiDuprez, K. (2003). Infect. Immun. 71, 2902–2906. Lundgren, A., Suri-Payer, E., Enarsson, K., Svennerholm, A. M., and Lundin, B. S. (2003). Infect. Immun. 71, 1755–1762. Maeda, H., and Akaike, T. (1998). Biochemistry 63, 854–865. Maeda, S., Akanuma, M., Mitsuno, Y., Hirata, Y., Ogura, K., Yoshida, H., Shiratori, Y., and Omata, M. (2001). J. Biol. Chem. 276, 44856–44864. Maeda, S., Yoshida, H., Mitsuno, Y., Hirata, Y., Ogura, K., Shiratori, Y., and Omata, M. (2002). Mol. Pathol. 55, 286–293. Mahdavi, J., Sonden, B., Hurtig, M., Olfat, F. O., Forsberg, L., Roche, N., Angstrom, J., Larsson, T., Teneberg, S., Karlsson, K. A., Altraja, S., Wadstrom, T., Kersulyte, D., Berg, D. E., Dubois, A., Petersson, C., Magnusson, K. E., Norberg, T., Lindh, F., Lundskog, B. B., Arnqvist, A., Hammarstrom, L., and Boren, T. (2002). Science 297, 573–578. Martins-Green, M., Boudreau, N., and Bissell, M. J. (1994). Cancer Res. 54, 4334–4341. McKeown, I., Orr, P., Macdonald, S., Kabani, A., Brown, R., Coghlan, G., Dawood, M., Embil, J., Sargent, M., Smart, G., and Stein, C. N. (1999). Am. J. Gastroenterol. 94, 1823–1829. Medzhitov, R. (2001). Nat. Rev. Immunol. 1, 135–145. Mills, J. C., Andersson, N., Hong, C. V., Stappenbeck, T. S., and Gordon, J. I. (2002). Proc. Natl. Acad. Sci. USA 99, 14819–14824. Mimuro, H., Suzuki, T., Tanaka, J., Asahi, M., Haas, R., and Sasakawa, C. (2002). Mol. Cell 10, 745–755. Mitsuno, Y., Yoshida, H., Maeda, S., Ogura, K., Hirata, Y., Kawabe, T., Shiratori, Y., and Omata, M. (2001). Gut 49, 18–22. Molinari, M., Salio, M., Galli, C., Norais, N., Rappuoli, R., Lanzavecchia, A., and Montecucco, C. (1998). J. Exp. Med. 187, 135–140. Moran, A. P., Sturegard, E., Sjunnesson, H., Wadstrom, T., and Hynes, S. O. (2000). FEMS Immunol. Med. Microbiol. 29(4), 263–270. Mueller, A., O’Rourke, J., Grimm, J., Guillemin, K., Dixon, M. F., Lee, A., and Falkow, S. (2003). Proc. Natl. Acad. Sci USA 100, 1292–1297. Muotiala, A., Helander, I. M., Pyhala, L., Kosunen, T. U., and Moran, A. P. (1992). Infect. Immun. 60, 1714–1716. Nagasako, T., Sugiyama, T., Mizushima, T., Miura, Y., Kato, M., and Asaka, M. (2003). J. Biol. Chem. 278, 4821–4825. Nardone, G., Holicky, E. L., Uhl, J. R., Sabatino, L., Staibano, S., Rocco, A., Colantuoni, V., Manzo, B. A., Romano, M., Budillon, G., Cockerill, F. R., 3rd, and Miller, L. J. (2001). Infect. Immun. 69, 5857–5863. Nishioka, H., Baesso, I., Semenzato, G., Trentin, L., Rappuoli, R., Del Giudice, G., and Montecucco, C. (2003). Eur. J. Immunol. 33, 840–849. Odenbreit, S., Faller, G., and Haas, R. (2002). Int. J. Med. Microbiol. 292, 247–256.

Persistent H. Pylori Infection

87

Odenbreit, S., Puls, J., Sedlmaier, B., Gerland, E., Fischer, W., and Haas, R. (2000). Science 287, 1497–1500. Ogura, K., Maeda, S., Nakao, M., Watanabe, T., Tada, M., Kyutoku, T., Yoshida, H., Shiratori, Y., and Omata, M. (2000). J. Exp. Med. 192, 1601–1610. Ogura, Y., Bonen, D. K., Inohara, N., Nicolae, D. L., Chen, F. F., Ramos, R., Britton, H., Moran, T., Karaliuskas, R., Duerr, R. H., Achkar, J. P., Brant, S. R., Bayless, T. M., Kirschner, B. S., Hanauer, S. B., Nunez, G., and Cho, J. H. (2001). Nature 411, 603–606. Olfat, F. O., Naslund, E., Freedman, J., Boren, T., and Engstrand, L. (2002). J. Infect. Dis. 186, 423–427. Ozturk, H., Senocak, M. E., Uzunalimoglu, B., Hascelik, G., Buyukpamukcu, N., and Hicsonmez, A. (1996). Eur J. Pediatr. Surg. 6, 265–269. Parsonnet, J. (1995). Aliment Pharmacol. Ther. 9, (Suppl 2) 45–51. Parsonnet, J., Friedman, D. G., Orentreich, N., and Vogelman, I. I. (1997). Lancet 40, 297–301. Peek, R. M., Jr. (2002). J. Gastroenterol. 37, 10–16. Peek, R. M., Jr., Wirth, H. P., Moss, S. F., Yang, M., Abdalla, A. M., Tham, K. T., Zhang, T., Tang, L. H., Modlin, I. M., and Blaser, M. J. (2000). Gastroenterology 118, 48–59. Pelicic, V., Reyrat, J. M., Sartori, L., Pagliaccia, C., Rappuoli, R., Telford, J. L., Montecucco, C., and Papini, E. (1999). Microbiology 145, 2043–2050. Pollet, I., Opina, C. J., Zimmerman, C., Leong, K. G., Wong, F., and Karsan, A. (2003). Blood. 102, 1740–1742. Ponzetto, A., Soldati, T., and deGuili, M. (1996). Lancet 348, 758. Richmond, A., and Thomas, H. G. (1986). J. Cell Physiol. 129, 375–384. Rollinson, S., Levene, A. P., Mensah, F. K., Roddam, P. L., Allan, J. M., Diss, T. C., Roman, E., Jack, A., MacLennan, K., Dixon, M. F., and Morgan, G. J. (2003). Blood 102, 1007–1011. Romano, M., Ricci, V., Memoli, A., Tuccillo, C., Di Popolo, A., Sommi, P., Acquaviva, A. M., Del Vecchio Blanco, C., Bruni, C. B., and Zarrilli, R. (1998). J. Biol. Chem. 273, 28560–28563. Rosenblatt, K. A., Wicklund, K. G., and Stanford, J. L. (2001). Am. J. Epidemiol. 153, 1152–1158. Sakai, T., Aoyama, N., Satonaka, K., Shigeta, S., Yoshida, H., Shinoda, Y., Shirasaka, D., Miyamoto, M., Nose, Y., and Kasuga, M. (1999). J. Gastroenterol. 34, 24–27. Sasazuki, S., Sasaki, S., Tsubono, Y., Okubo, S., Hayashi, M., Kakizoe, T., and Tsugane, S. (2003). Cancer Sci. 94, 378–382. Schilling, J. D., and Hultgren, S. J. (2002). Int. J. Antimicrob. Agents 19, 457–460. Schnare, M., Barton, G. M., Holt, A. C., Takeda, K., Akira, S., and Medzhitov, R. (2001). Nat. Immunol. 2, 947–950. Schoppmann, S. F., Birner, P., Stockl, J., Kalt, R., Ullrich, R., Caucig, C., Kriehuber, E., Nagy, K., Alitalo, K., and Kerjaschki, D. (2002). Am. J. Pathol. 161, 947–956. Segal, E. D., Cha, J., Lo, J., Falkow, S., and Tompkins, L. S. (1999). Proc. Natl. Acad. Sci. USA 96, 14559–14564. Selbach, M., Moese, S., Hauck, C. R., Meyer, T. F., and Backert, S. (2002a). J. Biol. Chem. 277, 6775–6778. Selbach, M., Moese, S., Hurwitz, R., Hauck, C. R., Meyer, T. F., and Backert, S. (2003). EMBO J 22, 515–528. Selbach, M., Moese, S., Meyer, T. F., and Backert, S. (2002b). Infect. Immun. 70, 665–671. Sheu, B. S., Sheu, S. M., Yang, H. B., Huang, A. H., and Wu, J. J. (2003). Gut 52, 927–932. Shibayama, K., Kamachi, K., Nagata, N., Yagi, T., Nada, T., Doi, Y., Shibata, N., Yokoyama, K., Yamane, K., Kato, H., Iinuma, Y., and Arakawa, Y. (2003). Mol. Microbiol. 47, 443–451.

88

S. Normark et al.

Shimizu, N., Inada, K., Nakanishi, H., Tsukamoto, T., Ikehara, Y., Kaminishi, M., Kuramoto, S., Sugiyama, A., Katsuyama, T., and Tatematsu, M. (1999). Carcinogenesis 20, 669–676. Smith, M. F., Jr., Mitchell, A., Li, G., Ding, S., Fitzmaurice, A. M., Ryan, K., Crowe, S. E., and Goldberg, J. B. (2003). J. Biol. Chem. 278, 32552–32560. Stappenbeck, T. S., Hooper, L. V., and Gordon, J. I. (2002). Proc. Natl. Acad. Sci. USA 99(24), 15451–15455. Straubinger, R. K., Greiter, A., McDonough, S. P., Gerold, A., Scanziani, E., Soldati, S., Dailidiene, D., Dailide, G., Berg, D. E., and Simpson, K. W. (2003). Infect. Immun. 71, 2693–2703. Su, B., Ceponis, P. J., Lebel, S., Huynh, H., and Sherman, P. M. (2003). Infect. Immun. 71, 3496–3502. Su, B., Hellstrom, P. M., Rubio, C., Celik, J., Granstrom, M., and Normark, S. (1998). J. Infect. Dis. 178, 1379–1390. Su, B., Johansson, S., Fallman, M., Patarroyo, M., Granstrom, M., and Normark, S. (1999). Gastroenterology 117, 595–604. Suzuki, J., Ohnishi, H., Wada, A., Hirayama, T., Ohno, H., Ueda, N., Yasuda, H., Iiri, T., Wada, Y., Futai, M., and Mashima, H. (2003). J. Biol. Chem. 278, 25585–25590. Syder, A. J., Oh, J. D., Guruge, J. L., O’Donnell, D., Karlsson, M., Mills, J. C., Bjo¨rkholm, B. M., and Gordon, J. I. (2003). Proc. Natl. Acad. Sci. USA 100, 3467–3472. Takata, T., El-Omar, E., Camorlinga, M., Thompson, S. A., Minohara, Y., Ernst, P. B., and Blaser, M. J. (2002). Infect. Immun. 70(6), 3073–3079. Taylor, D. E., Rasko, D. A., Sherburne, R., Ho, C., and Jewell, L. D. (1998). Gastroenterology 115(5), 1113–1122. The, B. H., Lin, J. T., Pan, W. H., Lin, S. H., Wang, L. Y., Lee, T. K., and Chen, C. J. (1994). Anticancer Res. 14, 1389–1392. Touati, E., Michel, V., Thiberge, J. M., Wuscher, N., Huerre, M., and Labigne, A. (2003). Gastroenterology 124, 1408–1419. Vinhall, L. E., King, M., Novelli, M., Green, C. A., Daniels, G., Hilkens, J., Sarner, M., and Swallow, D. M. (2002). Gastroenterology 123, 41–49. Waidner, B., Greiner, S., Odenbreit, S., Kavermann, H., Velayudhan, J., Stahler, F., Guhl, J., Bisse, E., van Vliet, A. H., Andrews, S. C., Kusters, J. G., Kelly, D. J., Haas, R., Kist, M., and Bereswill, S. (2002). Infect. Immun. 70, 3923–3929. Wang, G., Ge, Z., Rasko, D. A., and Taylor, D. E. (2000). Mol. Microbiol. 36, 1187–1196. Wang, J., Brooks, E. G., Bamford, K. B., Denning, T. L., Pappo, J., and Ernst, P. B. (2001). J. Immunol. 197, 926–934. Watanabe, T., Tada, M., Nagai, H., Sasaki, S., and Nakao, M. (1998). Gastroenterology 115, 642–648. Wotherspoon, A. C., Doglioni, C., Diss, T. C., Pan, L., Moschini, A., de Boni, M., and Isaacson, P. G. (1993). Lancet 342, 575–577. Wroblewski, L. E., Noble, P. J., Pagliocca, A., Pritchard, D. M., Hart, C. A., Campbell, F., Dodson, A. R., Dockray, G. J., and Varro, A. (2003). J. Cell Sci. 116, 3017–3026. Wu, M. S., Wu, C. Y., Chen, C. J., Lin, M. T., Shun, C. T., and Lin, J. T. (2003). Int. J. Cancer 104, 617–623. Yamanishi, Y., Boyle, D. L., Rosengren, S., Green, D. R., Zvaifler, N. J., and Firestein, G. S. (2002). Proc. Natl. Acad. Sci. USA 99, 10025–10030. Yamaoka, Y., Kwon, D. H., and Graham, D. Y. (2000). Proc. Natl. Acad. Sci. USA 97, 7533–8753. Yamazaki, S., Yamakawa, A., Ito, Y., Ohtani, M., Higashi, H., Hatakeyama, M., and Azuma, T. (2003). J. Infect. Dis. 187, 334–337. Yao, Y. L., Xu, B., Song, Y. G., Zhang, W. D., Zhang, Y. L., and Zhang, Z. S. (2002). Di Yi Jun Yi Da Xue Xue Bao 22, 348–350.

Persistent H. Pylori Infection

89

Zambon, C. F., Navaglia, F., Basso, D., Rugge, M., and Plebani, M. (2003). J. Clin. Pathol. 56, 287–291. Zaridze, D., Borisova, E., Maximovitch, D., and Chkhikvadze, V. (1999). Int. J. Cancer 82, 473–476. Zhan, X. M., and Xu, Q. (2001). Melanoma Res. 11, 559–567. Zhang, Z. W., Dorrell, N., Wren, B. W., and Farthingt, M. J. (2002). J. Med. Microbiol. 51, 495–502. Zheng, P. Y., and Jones, N. L. (2003). Cell Microbiol. 5, 25–40.

This Page Intentionally Left Blank

High-Resolution Analysis of Genetic Events in Cancer Cells Using Bacterial Artificial Chromosome Arrays and Comparative Genome Hybridization John K. Cowell1 and Norma J. Nowak1,2 1

Roswell Park Cancer Institute, Department of Cancer Genetics, Elm and Carlton 2 Streets, Buffalo, New York 14263; and Center of Excellence in Bioinformatics, State University of New York at Buffalo, 901 Washington Street, Buffalo, New York 14203

I. Introduction II. Evolution of Molecular Cytogenetics III. Development of BAC Resources A. Development of CGH Array Technology B. Analysis of CGHa Arrays IV. CGHa in the Analysis of Cancer A. Constitutional Chromosome Abnormalities B. CGHa Analysis of Tumors C. Comparison of CGHa with Conventional Karyotyping D. CGHa Analysis of Primary Brain Tumors E. Verification of Genetic Changes Identified Using CGHa F. Definition of the Minimally Involved Regions from Overlapping NCAs G. Determination of the Gene Content on the NCAs H. Comparison between CGHa and Expression Array Analysis I. Comparison between CGHa and Loss of Heterozygosity J. Analysis of Archival Tissue Material V. Summary Acknowledgments References

Chromosome analysis of cancer cells has been one of the primary means of identifying key genetic events in the development of cancer. The relatively low resolution of metaphase chromosomes, however, only allows characterization of major genetic events that are defined at the megabase level. The development of the human genome-wide bacterial artificial chromosome (BAC) libraries that were used as templates for the human genome project made it possible to design microarrays containing these BACs that can theoretically span the genome uninterrupted. Competitive hybridization to these arrays using tumor and normal DNA samples reveals numerical chromosome abnormalities (deletions and amplifications) that can be accurately defined depending on the density of the arrays. At present, we are using arrays with 6,000 BACs, which provide an average resolution of less than 700 kb. Analysis of tumor DNA samples Advances in CANCER RESEARCH 0065-230X/03 $35.00

91

Copyright 2003, Elsevier Inc. All rights reserved

92

John K. Cowell and Norma J. Nowak using these arrays reveals small deletions and amplifications that were not detectable by chromosome analysis and provides a global view of these genetic changes in a single hybridization experiment in 24 hours. The extent of the genetic changes can then be determined precisely and the gene content of the affected regions established. These arrays have widespread application to the analysis of cancer patients and their tumors and can detect constitutional abnormalities as well. The availability of these highdensity arrays now provides the opportunity to classify tumors based on their genetic fingerprints, which will assist in staging, diagnosis, and even prediction of response to therapy. Importantly, subtle genetic changes that occur consistently in tumor cell types may eventually be used to stratify patients for clinical trials and to predict their response to custom therapies. ß 2003 Elsevier Inc.

I. INTRODUCTION One of the foundations of our understanding of the genetic events that give rise to cancer has come from the analysis of chromosome abnormalities in tumor cells. Chromosome translocations gave rise to the discovery of constitutively activated oncogenes (Rowley, 1998). Double minute chromosomes and homogeneously staining regions were the early visualization of gene amplification (Cowell, 1982), and microdeletions in chromosome regions from patients with specific cancer predisposition syndromes led to the discovery of tumor suppressor genes (Cowell, 2001). This whole field became possible with the development of technologies that adequately separate and view the individual chromosomes (Hsu, 1952; Tjio and Levan, 1956). This allowed them to be subgrouped based on their centromeric position; for example, telocentric (centromeres at the end) or metacentric (centromeres in the middle). This simple advance gave rise to countless studies of chromosomes in cancer cells, describing numerical changes such as trisomies or monosomies for individual chromosomes, as well as large changes in chromosome numbers defined as pseudodiploidy, aneuploidy, hyperdiploidy, and tetraploidy, depending on the extent of chromosome changes (Sandberg, 1990). Aneuploidy and hyperdiploidy, assessed either by chromosome number or DNA content measurement, are still used as a marker for tumor progression and, in some cases, may still be a valuable prognostic indicator. The ability to identify each individual chromosome came with the advent of chromosome banding techniques (Casperson et al., 1968; Seabright, 1971). Thus, a characteristic linear distribution of light and dark bands along the length of the chromosomes could be generated. To the expert, it was now possible to identify each of the human chromosomes based on their banding pattern and to unequivocally identify each of the 23 different ones. Sequential numerical labeling of the band pattern along the length of the chromosome arms also provided a means of communicating the nature of subchromosomal changes such as deletions, duplications, translocations,

CGHa and Cancer Diagnostics

93

and inversions. These advances began the extensive effort of characterizing cytogenetic changes in all types of tumor cells, which, in turn, formed the basis for much of the molecular cytogenetics era that followed. Many cancer syndromes had already been defined clinically by the coincidence of a limited set of phenotypes; for example, the association of mental retardation with retinoblastoma (Lele et al., 1963) and of aniridia with Wilms tumor (Riccardi et al., 1978). The ability to detect smaller and smaller chromosomal changes followed the development of new techniques to generate longer and longer chromosomes (Francke and Oliver, 1978; Yunis, 1976). The ability to unequivocally identify each of the normal human chromosomes had also opened up the possibility of analyzing the constitutional chromosomes, for example, in lymphocytes, from patients with cancer. It was subsequently found that individuals with some of these cancer syndromes (Riccardi et al., 1978; Yunis and Ramsay, 1978), carried inherited chromosome abnormalities. Importantly, these rearrangements not only pointed to the location of the genes responsible for the tumor predisposition but also provided a means of isolating them through positional cloning strategies. This proved to be true in many cases, and in fact, the cloning of many cancer predisposition genes has been facilitated by these cytogenetic observations (Cowell, 2001). As cytogenetic data accumulated, there became a need to catalog the various changes to identify consistencies between and within tumor types. These data provided a means to identify reliable markers that would not only aid in determining diagnosis and predict prognosis but also in highlighting regions within the genome that might contain genes important in the development of a particular malignancy. The task of doing this was adopted by Felix Mitelman and colleagues (Mitelman and Levan, 1981), who, after starting the catalog with an early report of approximately 1,900 cytogenetic changes in tumor cells in 1981, now presides over an incredible volume describing over 100,000 changes that is available for review online (Mitelman, 2000). These studies highlighted several phenomena in addition to the diversity of cytogenetic changes; First, that there were clearly consistencies in the regions of the chromosomes involved in abnormalities in specific tumor types. The consistency between tumors indicated that the rearrangements were important in tumorigenesis, especially where these represented the only cytogenetic change in the cells. Second, it was clear that the banding procedures could not always resolve the nature of every chromosome change, and so valuable genetic information was uninterpretable. These unidentified chromosomes were banished to the end of the karyotype and simply described as ‘‘marker’’ chromosomes to indicate their anonymous nature. Some of these marker chromosomes would turn out to be important, and their characterization depended on yet another development in technology that was being developed in the late 1980s, heralding the beginning of the era of ‘‘molecular cytogenetics.’’

94

John K. Cowell and Norma J. Nowak

II. EVOLUTION OF MOLECULAR CYTOGENETICS The final ability to map cloned DNA fragments to particular chromosomal regions became possible as a result of several technological advances that proved to be essential for the molecular cytogenetics revolution: the ability to label DNA probes with fluorochromes to high density; the ability to collect low levels of fluorescent light, using CCD cameras; and the construction of large insert genomic libraries. The ability to suitably compete out repetitive sequences within these large probes (Landergent, 1987; Pinkel et al., 1988), together with the increased fluorescence signal obtained from large insert clones, provided the opportunity to localize markers directly on chromosomes in metaphase spreads. Thus, mapping by fluorescence in situ hybridization (FISH) became the primary way of localizing individual clones (Baldini et al., 1992). This approach also identified chimeric clones, as signal could be seen on two different chromosomes, providing important information for gene cloning/mapping experiments. This technique soon resulted in large databases of mapped probes, which made the identification of candidate genes associated with particular cancers much faster. The application of chromosome ‘‘painting,’’ which used complex probes (cDNA libraries, flow-sorted chromosomes, Alu–polymerase chain reaction (PCR) products) derived from single chromosomes or chromosome arms to highlight specific chromosomes within a metaphase spread, overcame the need to describe the troublesome marker chromosomes, as their origin could now be unequivocally established. The ability to flow-sort chromosomes individually and then preferentially amplify unique sequences along their length led to the development of spectral karyotyping (SKY). With the probes for each chromosome labeled with a different combination of concentrations of five different dyes, it now became possible to perform karyotype analysis to identify the origin of every piece of the chromosomes even when they were involved in complex marker formation. Digital images are processed with custom software that generates the karyotype and allows identification of chromosome translocations in particular, as well as numerical changes, unequivocally. With this technology, marker chromosomes could be interpreted, subtle chromosome rearrangements could be identified, and the human chromosome complement of somatic cell hybrids could be established (Matsui et al., 2003). The limitation, however, was still the need for metaphase chromosome spreads, which, with their low resolution, could not define the exact position of breakpoints or identify subtle genetic changes in fine detail. Cytogenetics has clearly been the source for many hypotheses about cancer, but its major limitation is that it depends on sufficient numbers of dividing cells in the particular tumors to allow chromosome preparation.

CGHa and Cancer Diagnostics

95

For many slow-growing, solid tumors this is not always possible, especially when the sample size was limiting. In turn, FISH was mostly used to study single chromosomes or single loci on the chromosomes. FISH probes derived from the centromeres of the individual human chromosomes could be used to detect numerical chromosome changes (Kempski et al., 1995), and gene-specific probes could detect large chromosome deletions (Cowell et al., 1994). Despite this, the need to scale up to look at the whole chromosome complement in nondividing cells led to the development of comparative genome hybridization (CGH). In this procedure, tumor and normal DNA is competitively hybridized to normal metaphase chromosomes in a single hybridization reaction (Kallioniemi et al., 1992, 1994; Pinkel et al., 1986, 1988). Repeat sequences were competed out with Cot-1 DNA, and typically the tumor is labeled with a green fluorochrome and the normal DNA is labeled with a red fluorochrome. Copy number differences between tumor and normal cells are identified by measuring the ratio of green to red fluorescence along the length of normal chromosome spreads (Fig. 1). Thus, if a region is deleted in the tumor, only the normal red fluorescence is seen at that chromosome site. If a particular region of a chromosome is amplified in the tumor, then the signal is greener. Equal hybridization (yellow) represented no difference between samples (Getman et al., 1998). In general, however, this approach is not sensitive enough to identify submicroscopic changes, but it has the advantage of surveying the whole genome without the need for mitotic tumor cells and identifies consistent changes in tumors by pooling the images. Several Web sites for CGH data have been established such as ones at the National Center of Biotechnology Information (NCBI) and http://amba.charite.de. Despite the skills of observation and interpretation of cytogeneticists worldwide, all of these technologies were still limited by the level of resolution that could be achieved from conventional karyotype analysis.

Fig. 1 Examples of conventional comparative genome hybridization analysis of chromosomes from Wilms’ tumors. The upper curves in each case represent composite ratio profiles for the tumor/normal comparisons for each of the chromosomes shown below. Amplifications on chromosome 3 are identified by peaks which rise above a ratio of 1.2 (top horizontal line), deletions for chromosomes 5 and 13 show dips in the curve below a ratio of 0.8 (lower horizontal line). In each case the regions that are affected can only be related to the size of the metaphase chromosome region indicated.

96

John K. Cowell and Norma J. Nowak

Although estimates vary, the size of a conventional G-band at the 850-band resolution, for example, is in the range of 10–20 megabases. Thus, deletions of only hundreds of kilobases or less would not normally be detected using this approach. Although FISH would detect these rearrangements, the investigator would need to know roughly where to look in the first place. In a genomewide discovery mode, however, it was clearly impossible to screen the whole genome by FISH. With the creation of the BAC libraries that were used to provide the templates for sequencing the human genome, it became possible to design approaches that would map out the genomic landscape of genetic changes in cancers.

III. DEVELOPMENT OF BAC RESOURCES BACs are generated from partially digested human DNA that has been size selected to approximately 200 kb. These DNA molecules are then cloned into vectors that can be maintained as whole chromosomes in bacterial hosts. The inserts in these libraries were shown to be highly stable and, thus, not subject to the random structural changes leading to instability and deletions (Still et al., 1997) that were frequent in the yeast artificial chromosome (YAC) libraries that preceded them. BACs are also far more manageable in terms of their size and their ease of isolation and manipulation. The BAC libraries used extensively in the public human genome sequencing effort were generated at Roswell Park Cancer Institute, which provided over 70% of the clones currently available in databases (prefixed with RP11). In addition to being used for genome sequencing, selected BAC clones were identified as part of a large-scale mapping program (National Cancer Institute Extramural Cancer Chromosome Aberration Project) to identify a reference set of human BACs for the analysis of chromosome abnormalities using FISH. These BACs were assembled by screening the human RPCI-11 BAC library with 6,000 markers that had been mapped at 0.5–1-mb intervals across the genome, using the G3 and G4 radiation hybrid panels (Nowak et al., 2001). Importantly, all of the clones were subsequently analyzed using FISH to verify that they localize to only a single position in the karyotype, thereby reducing the possibility of cross-hybridization with other sequences during the analysis. From this effort, 6,116 sequenced or sequence-connected BACs were identified, which included 1,300 BACs containing known tumor suppressor genes, oncogenes, and genes associated with the cancer phenotype, as well as telomeric BACs. We have recently used a virtual screening strategy for interrogating the draft human sequence to select additional BAC clones that are being used to reach a higher level of resolution for array-based studies (see following). The extensive annotation of the human genome sequence

CGHa and Cancer Diagnostics

97

that has occurred over the last few years now provides a detailed description of the genes that are known, or predicted, on each BAC and within each inter-BAC interval. A set of interactive Web pages are available to view and search the complete set (http://genomics.roswellpark.Org/human/overview.html) of markers on these BAC clones. Another Web site, http:// www.ncbi.nlm.nih.Gov/genome/cyto/hbrc.shtml is also maintained at NCBI for this resource. This series of BACs provided the initial resource to create arrays of BACs spanning the entire length of the genome. Although BAC arrays are being used more extensively, the details of their construction have required considerable development to generate the current robust product. We have been actively involved in these technological developments, and some of the specific details that are important in understanding their application, and in being able to interpret the data, are given below.

A. Development of CGH Array Technology Conventional CGH uses metaphase chromosomes as immobilized targets to measure changes in genomic copy number between two samples. In contrast, array CGH uses discrete segments of DNA immobilized on a glass slide or solid support. These DNA segments may be PCR products from cDNA clones or long oligonucleotides representing genes (Hedenfalk et al., 2003). We have chosen PCR representations from the BAC clones as targets on the arrays in a procedure that is now generally referred to as ‘‘CGHa.’’ BACs provide distinct advantages, including significantly better signal-tonoise ratios in comparison to the other array platforms, as well as providing greater genomic coverage. This allows a true delineation of aberration borders and the DNA sequence content of the region. Genomewide scans of a tumor can be accomplished within 24 hours. We have developed the methodology described later to give a comprehensive picture of the cancer genome and to aid in the classification of malignancies. BACs are carried in the host cells as single-copy plasmids, and so DNA yields are typically low, even from large-scale cultures. Thus, direct spotting of whole BAC DNA at a concentration of 800–1,000 ng/mL, which is required for optimal hybridizations, has not been successful. A PCR-based approach, therefore, was developed that enabled even representation across the BAC-insert DNA, unlike previous attempts (Geschwind et al., 1998; Lucito et al., 2000). We have shown that independently prepared, PCRbased DNA representations from the BACs are highly reproducible and provide essentially identical ratios to those generated using plasmid DNA from the same BACs (Snidjers et al., 2001). The important considerations about preparing the ‘‘printing solutions’’ for the construction of these arrays are discussed later.

98

John K. Cowell and Norma J. Nowak

Purified individual BAC DNAs (Qiagen R.E.A.L. Prep Biorobot kit, Valencia, CA) are first digested with MseI to generate fragments between 70 bp and 2 kb. MseI was chosen as it will preferentially allow amplification of human DNA and not Escherichia coli, representing a significant improvement over other amplification schemes used for BAC array generation. Adapters, Mse-12 (TAACTAGCATGC), and Mse-21(50 aminolinker AGTGGGATTCCGCATGCTAGT) are then ligated to each digest. Two rounds of amplification are then required to produce enough product for array generation. In the first round, a double-stranded PCR product is generated that ideally ranges in size from 100 to 2,000 bp. To generate DNA for printing, a second amplification is performed using only the Mse-21 prime, which yields 10 mg of single-stranded DNA, with each fragment containing a 50 amino linker. This PCR product usually ranges in size from 200 to 2,000 bp. The advantage of this approach is that the initial PCR product acts as the template for the generation of all of the subsequent printing solutions. This procedure greatly reduces the frequency with which the ligated BAC DNA must be prepared, which is the most time-consuming part of the procedure. Another distinct advantage of BAC-based arrays for CGH is the overall size and complexity of the PCR products generated from the BAC clones. As PCR product size and complexity increases, the signal-to-noise ratio improves, resulting in increased accuracy in identifying true copy number changes. Other CGHa approaches have involved cDNA and Agilent arrays, but these studies have shown that the low signalto-noise ratios requires utilization of moving averages to identify actual regions of copy number change (Geschwind et al., 1998; Pollack et al., 1999, 2002). We have also investigated the relative merits of several different spotting solutions for printing PCR products and have found that 25% dimenthyl sulfoxide (DMSO) performs best in our hands. As a result, the purified PCR products are resuspended at a concentration of 800–1,000 ng/ml in 25% DMSO and robotically rearrayed into a 384-well format for printing. We have demonstrated that DNA prepared using this method outperforms other methods, including degenerate oligonucleotide primer PCR, inter-AluPCR, and amplification of BAC subclone mixtures. Arrays are printed by our group on silanated glass slides using a MicroGrid II TAS arrayer (Apogent Discoveries, Hudson, NH). The BAC DNA products have 150-mm diameter spots with 225-mm center-to-center spacing. Each DNA solution is printed in triplicate to create an array of, at present, 18,000 elements. The printed slides dry overnight, are UV-crosslinked, and are then hybridized without additional treatment, except for the prehybridization blocking step described below. Using 40 , 6-diamindo-2-phenylindole (DAPI) staining, we found no indication of DNA loss from the spots at any stage of the

CGHa and Cancer Diagnostics

99

procedure. We also include a variety of control elements with each print run to verify the quality of the print run and hybridizations. The hybridization probe consists of genomic DNA that is fluorescently labeled with Cy3 or Cy5-dCTP by random priming, using as little as 30–500 ng DNA. Where high–molecular weight DNA is limited, it is also possible to use a modified ligation-mediated PCR amplification scheme described above to prepare representation of the BAC inserts for spotting (Snidjers et al., 2001) or to apply whole-genome amplification with as little as 5–50 ng of DNA, using a modification of rolling-circle amplification (GenomiPhi, Amersham Biosciences, Piscataway, NJ). The control for each test sample consists of pooled samples, from either cytogenetically normal males or females, that are specifically sex-mismatched as described later. The labeled test and reference DNAs are combined with Cot-1 DNA and yeast tRNA and are coprecipitated with ethanol. The hybridization cocktail (Cowell et al., 2003) is incubated at 95 C for 5 minutes to denature the DNA, and the incubation is continued at 37 C for 60 minutes to block repetitive sequences. The arrays are then hybridized with pulsation in a GeneTac hybridization station (Genomic Solutions, Ann Arbor, MI) for 16 hours at 65 C. After hybridization, the slides are automatically washed with reducing concentrations of SSC and SDS, followed by one ethanol rinse and a quick centrifugation for drying. The arrays are then scanned immediately at a 10-mm resolution on an Affymetrix 428 scanner to generate high-resolution images for both Cy3 and Cy5 channels.

B. Analysis of CGHa Arrays Image analysis is the first step in analyzing CGH arrays and is performed identically to cDNA expression arrays. Imagene (BioDiscovery, Marina Del Rey, CA) software is used to identify the spots and to measure the fluorescent intensity at each element on the array. Optimal cDNA expression array analysis requires that the experiments are designed to incorporate a dye flip, addressing the possibility of dye bias and providing a replicate experiment. CGH arrays are far more robust as a result of significantly higher signal-tonoise ratios and are, thus, more reproducible than cDNA-expression arrays, allowing a single slide hybridization to accurately measure genomic copy number changes. The issue of dye bias in CGH arrays, however, should be addressed either by performing a dye-flip experiment or by applying an intensity dependent normalization. In our hands, any spot that has a signal-tobackground ratio of less than 6 is excluded as unreliable, as are any spots flagged by our image analysis software as ‘‘poor.’’ The data are normalized on the log scale by applying a lowess function to the data then normalizing on

100

John K. Cowell and Norma J. Nowak

the curve. Replicate data is combined by averaging the normalized ratio on the log scale for each replicate. Any BAC that does not have at least two of the three replicates considered useable is excluded. Mapping information from UCSC’s genome browser (http://genome.ucsc.edu/) is added to each of the BAC clones. The data are then imported to excel for visualization (Fig. 2). In all hybridizations in which the sex chromosome complement of the cells is known, a mismatch is performed; that is, if the sample is XX, then the control DNA is from an XY source. Thus, when XX tumor and XY normal are cohybridized to the array, there is a relative two-fold increase of X chromosome material, which should give a log2 ratio of þ1. Because there is no Y chromosome in the tumor, the ratio should, in theory, be 1(log20) for regions that are not pseudoautosomal or containing segments mapping to both the X and Y chromosomes (Fig. 3). In practice, the ratios we see for the sex mismatch, and by extension all other chromosomes, are smaller in magnitude than the theoretically expected values. This may be because of a number of reasons that are discussed later. The measured value of the X chromosome can then be used to estimate the amount of suppression. Structural chromosome changes identified using CGHa can be described in terms of their cytogenetic band nomenclature for ease of communication, but the relative location of the breakpoints can also be defined by the megabase position on the BAC along the length of the chromosome relative to the DNA genome sequence, from the telomere of the short arm, which is defined as 0. We have now used CGHa in an extensive analysis of genetic changes associated with tumor and normal cells, and examples of these applications are discussed later.

IV. CGHA IN THE ANALYSIS OF CANCER A. Constitutional Chromosome Abnormalities A number of different cancers, such as retinoblastoma (Rb), Wilms tumor (WT), and Beckwith-Wiedemann syndrome (BWS), are associated with inherited chromosome deletions and duplications (Cowell et al., 1989a, 1989b; Mannens et al., 1994). Because syndromes are almost invariably associated with other congenital abnormalities, these phenotypes can usually be used to predict the particular type of predisposition. However, where there is overlap between the associated phenotypes, such as BWS and Perlman’s syndrome (Grundy et al., 1992), the chromosome changes may become important as a diagnostic tool. In this respect, the BAC arrays potentially provide an unequivocal way of not only identifying the presence of

CGHa and Cancer Diagnostics

101

Fig. 2 Example of a total-genome bacterial artificial chromosome comparative genome hybridization ‘‘karyotype’’ from brain tumor 52. On a log-scale normal chromosome number is represented as 0. Bacterial artificial chromosomes associated with single-copy increases, as seen for chromosome 2 and 12, cluster around a median of approximately þ0.5. Single-copy decreases cluster around a median of 0.5, as for the proximal part of 3p and chromosome 13. These ratios are similar to that seen for the X chromosome because this hybridization was from the tumor of a male with an XX mismatch (see text). A typical profile for the Y chromosome for an XX–XY mismatch is shown. With a ratio of þ1, two extra copies of chromosome 7 are seen and a homozygous deletion of the 9p region containing the CDKN2A gene shows a ratio of 1.

102

John K. Cowell and Norma J. Nowak

Fig. 3 Analysis of the X chromosome from a typical sex mismatch hybridization with XX in the tumor and XY in the control DNA. On the linear scale, suppression of hybridization shows that the majority of bacterial artificial chromosomes (BACs) cluster around a median of approximately 0.6. Several BACs that show a ratio of nearer 1 (no change) were shown to correspond to the pseudoautosomal regions (PAR1/2) at the ends of the chromosomes. Other peaks contain genes (PRKX, AMELX, UTX, and PRS4X) that have homologs on the Y chromosome. In addition, BACs that contain areas of long interspersed repeats (IR), and in one case (XY) a BAC that was shown to hybridize by FISH to both the X and Y chromosomes also show ratios near to 1.

deletions and duplications but also determining their extent. This can be important, as the large deletions will encompass more genes, which will ultimately determine the various phenotypes. An example of this is seen in patients with aniridia, which is a rare hereditary disease resulting in the absence of irises. In the familial form, the phenotype segregates as an autosomal dominant disorder because of mutations in the PAX6 gene in 11p13 (Davis and Cowell, 1993). Importantly, sporadic cases of aniridia show a 50% risk to the development of (WT, a pediatric cancer of the kidney (Riccardi et al., 1978). In these patients, the cancer predisposition results from a deletion involving the 11p13 region containing both the WT1 and PAX6 genes. Patients with PAX6 gene mutations clearly represent the hereditary form of the disease and are not at increased risk of the development of WT. From a genetic counseling standpoint, however, a sporadic case of aniridia could either carry a deletion predisposing to WT or carry a de novo mutation in the PAX6 gene, and therefore be only at background risk to kidney cancer. Because the aniridia phenotype is obvious at birth, whereas the onset for WT often occurs between the ages of 4–5 years, the genetic diagnosis is important to ensure appropriate early screening of the kidneys. Being able to exclude the tumor risk in these patients, therefore, would involve either a mutation study of the PAX 6 gene, which is a complex and time-consuming procedure, or a cytogenetic analysis of the 11p13 region, which may be inadequate for small deletions. The analysis of constitutional deletions in these cases has the advantage over mutation studies of speed and accuracy for the diagnosis. To assess the utility of the CGHa approach in this situation we used DNA from patient GOS 157, which we had previously demonstrated carries a small deletion involving the 11p13 region (Cowell et al., 1989b). The CGHa profile is shown in Fig. 4, where the

CGHa and Cancer Diagnostics

103

Fig. 4 CGHa profiles from patients with constitutional chromosome abnormalities. DNA ratios are plotted on the linear scale with normal DNA content seen as 1. In the Wilms/aniridia patient GOS 157, a heterozygous deletion of the 11p13–14 region is seen compared with the cytogenetic change (right). Similarly, in the retinoblastoma patient GOS 115, a small deletion in the 13q14 region is seen by comparative genome hybridization (CGHa) compared with the subtle changes seen in G-banded chromosomes (right). CGHa from two patients, GOS 71 and GOS 107, with deletions of other regions of chromosome 13 show that they are clearly different and not overlapping with the retinoblastoma deletion. The amplification of the 11p15 region in the Beckwith-Wiedemann syndrome patient GOS 637 clearly demonstrates that the most telomeric BACs are not coamplified.

104

John K. Cowell and Norma J. Nowak

presence of a heterozygous deletion is clearly seen in the short arm of chromosome 11, spanning a distance between 26.62 and 44.6 Mbp (18 Mbp), which includes the PAX6 and WT1 genes. Clearly, therefore, CGHa has advantages over other diagnostic procedures that might eventually complement or even replace cytogenetic analysis for the detection of deletions and duplications. To assess the application of BAC arrays to other deletion syndromes, we analyzed a series of DNA samples from patients with Rb and mental retardation (Cowell, 1989; Cowell et al., 1989a), using CGHa. The results demonstrate that the heterozygous deletions in two patients (GOS 115 and GOS 191) were readily identified as the only abnormality in the sample. The deletion associated with patient GOS 191 was cytogenetically easily detectable (Cowell et al., 1989a), and this deletion was shown by CGHa to extend between BACs RP11-11k16 (32.04 Mbp) and RP11-37i864 (64.09 Mbp), which represents a distance of 32.05 Mbp. The deletion associated with patient GOS 115, however (Fig. 4), was cytogenetically more subtle (Cowell, 1989) involving only a subband deletion in 13q14.3. Our CGHa analysis demonstrates that, in fact, the GOS 115 deletion spans the region of 13q14 between BACs RP11-20k19 and RP11-37i6, which constitutes 9.5 Mbp and that includes the RB1 gene located at position 47.81– 47.99 Mbp on the long arm of chromosome 13. This analysis, therefore, also provides an approximate relationship between the DNA sequence and the appearance of deletions in metaphase chromosomes at the 850-band resolution (Fig. 4). To determine whether patients who had other 13q deletions that did not apparently affect the RB1 gene could be distinguished, we extended our CGHa analysis to patients who had been reported as having 13q-syndrome, which involves various partial deletions in the q22-qter region (Gutierrez et al., 2001; Luo et al., 2000). These patients, GOS 71 and GOS 107, have a well-defined set of clinical phenotypes, including mental retardation, in which the deletion was generally assumed to involve the terminal region of 13q. However, although cytogenetic analysis had suggested the same diagnosis, the clinical phenotypes were somewhat different. CGHa analysis (Fig. 4) demonstrated that GOS 107 showed the typical deletion involving the 13q34 region, but not including the telomere, confirming that, in fact, this is an interstitial deletion that was located between BACs RP11-86c3 and RP11-7b23, which spans a 12-Mbp region. In contrast, GOS 71 showed a much more proximal deletion involving the 13q12-13 region between BACs RP11-179a7 (33.2 Mbp) and RP11-269c23 (43.87 Mbp), a distance of 10.67 Mbp. Clearly, although there is some overlap in the clinical phenotype between these two patients, the deletions are very different, which accounts for the discrepancy in their actual clinical phenotype. Importantly, the deletion in GOS 71 does not include the RB1 gene, which is

CGHa and Cancer Diagnostics

105

located at 47.9 Mbp, and so formally excludes this patient as being at high risk for Rb. From the experiments described above, it is clear that CGHa can be used to quickly and efficiently identify heterozygous constitutional deletions and could be easily extended to any other chromosome deletion syndrome. Other genetic syndromes, however, are associated with extra copies of small chromosome regions. An example of this (Mannens et al., 1994) is the chromosome imbalance that is associated with the pediatric cancer predisposition syndrome, BWS. In these cases, it has been demonstrated that duplication in 11p15 results in three copies of the 11p15.5 region, which is responsible for the phenotype in some cases. BWS may sometimes be confused with the phenotypically similar Perlman’s syndrome (PS), which has a much higher frequency of cancer than BWS (Grundy et al., 1992), and for which no chromosome abnormality has yet been identified. To determine whether these types of chromosome aberration can also be detected using CGHa, we analyzed DNA from a patient that had been shown by extensive molecular and cytogenetic analysis to carry a nonreciprocal chromosome translocation t(5;11)(p15;p15), which resulted in the triplication of the 11p15 region (Grundy et al., 1998). The CGHa profile from this patient, GOS 637, is shown in Fig. 4 and demonstrates that the translocation event is more probably the result of an insertion of the 11p15.5 region spanning BACs RP11-120e20 (3.67 Mbp) and RP11-6k5 (20.37 Mbp), covering 16.77 Mbp, into the distal region of 5p15. On the linear scale, all of the BACs in this region show an intensity ratio of 1.5, consistent with the presence of an extra copy. The most distal BACs (0–3.67 Mbp), however, show a ratio closer to 1. The most telomeric BAC on the array, RP11-123f4, is clearly only present in a diploid complement, and although the adjacent series of BACs show a ratio of 1.2, this is still within the range of noise shown for the other BACs along the chromosome, indicating that this region is also present in only two copies. Thus, although the diagnosis of BWS is not at issue, this analysis provides valuable information about the extent, and hence the gene content, of the region involved. One area in which this approach would be particularly useful is in the direct analysis of amniotic fluid cells or chorionic villus samples for prenatal diagnosis of hereditary chromosome abnormality syndromes. The rapid turnaround time associated with CGHa presents clear advantages for the clinical management of these patients and has important implications for genetic counseling. The analysis of the constitutional chromosome abnormalities described earlier also provides some further insight into the issue of suppression of the hybridization ratios. Because the constitutional cells all carry the abnormality, the test DNA is homogeneous in terms of cellular DNA content. Thus, as described earlier, loss of a single copy of a chromosome should result in a suppression of the ratio to 0.5 on the linear scale. As shown in

106

John K. Cowell and Norma J. Nowak

Fig. 4, however, this ratio is more usually between 0.5 and 1. We feel that this suppression of the theoretical levels occurs because of nonspecific binding of the probe to the array, which will reduce the actual hybridization ratio. This suppression is also seen for the X chromosome (Fig. 3), in particular for a number of specific BACs along the length of this chromosome. When we analyzed the gene content and DNA sequence of several of these BACs, we found that where the ratios were significantly different from the predicted ones, they either mapped to the pseudoautosomal regions of the X and Y chromosomes or contained a high density of long interspersed element (LINE) repeats (Fig. 3). It is very likely that a high density of these LINE repeats would compete equally well, and nonspecifically, for the differently labeled probes, resulting in suppression of the magnitude of the real differences. Empirically, in any array, there are also outlier BACs, some of which are seen from slide to slide and are probably caused by biological phenomena such as repetitive elements. Other outliers are clearly random events, as they disappear after repeating the hybridization (see following). In our experience, the majority of these random spikes and dips involve only a single BAC. Thus, where the profile change involves two or more BACs, it is more likely to represent a true genetic change, as the random chance of two adjacent BACs giving the same variation is small. In any event, although CGHa is very accurate and informative, it is still sometimes advisable to verify the small changes using independent approaches (see following).

B. CGHa Analysis of Tumors Our survey of constitutional genetic changes using CGHa has demonstrated the ease with which specific deletions and duplications can be detected. The ability to perform these analyses, using relatively small amounts of DNA and from nondividing cells, has obvious practical advantages in the analysis of cancer cells. One complication in CGHa analysis of tumors, however, is the potential intrinsic genetic heterogeneity within the sample, which may affect the hybridization ratio for a particular abnormality. The second cause of hybridization ratio suppression will result from the presence of contaminating normal cells in the tumor sample. Ploidy is also a complicating factor, as the amount of DNA used in the hybridization from the tumor samples is constant. Thus, for tetraploid cells, because there is twice the amount of DNA per cell, only half the number of cells are used compared with diploid controls. Thus, if there are no changes in the karyotype, then diploid and tetraploid cells would give the same result; that is, a log ratio of 0. However, where there are changes, then the value of the reduction or increase can be interpreted to define the ploidy of

CGHa and Cancer Diagnostics

107

the cells to a large degree (Cowell et al., 2003). For example, if the tumor has a modal chromosome number of 80, a ratio of 0.2, therefore, indicates that there are three copies of the chromosome. The expected value for a chromosome with three copies would be log2 3/(80/23) ¼ 0.213 (3 is the copy number, 80 is the modal chromosome number, and 23 is the euploid number of chromosomes). In the examples described later, all of these principles are taken into account, and the analysis is usually relatively straightforward.

C. Comparison of CGHa with Conventional Karyotyping Application of BAC-array CGH technology for the analysis of genetic changes in human cancer offers the opportunity to use a genome scale approach to uncover new and subtle changes specific for tumor cell types and to develop highly specific diagnostic and prognostic tools for the analysis and classification of tumor samples. This technology requires relatively small amounts of DNA for the analysis. For example, normal samples such as buccal swabs and tumor samples such as biopsies, or DNA isolated from tissue sections, can now be extensively characterized at the molecular level. This flexibility provides a great advantage for examining precursor lesions to identify the earliest changes observed in the process of tumorigenesis. These analyses do not require preparation and analysis of chromosome spreads, which may not be possible anyway. The other advantage is that the resolution of the analysis is currently an average of 750 Kb but, in future generations of BAC arrays, this will improve, allowing the definition of the breakpoints possibly to within a single BAC (150–250 Kb). All of these analyses are achieved from a single hybridization experiment. With such a potentially powerful assay of genetic changes in cancer cells, it is important to investigate and understand the limitations of the approach, so that the data can be interpreted correctly. In an early study, therefore, we undertook a detailed comparison of CGHa with SKY, which is currently considered the most discriminating total-karyotype, cytogenetic approach, using a series of glioma cell lines (Cowell et al., 2003). In an analysis of four newly derived brain tumor cell lines the consensus SKY karyotypes were established, and numerical and structural changes in each case were defined within the resolution of this approach. SKY detected numerical chromosome changes unequivocally and was also able to detect and define most chromosome translocations. Because of the limited resolution of the processed DAPI banding, however, any small deletions or amplifications that have been difficult to identify, and the position of the breakpoints, can only be given in terms of a chromosome band. In some cases we could only describe marker chromosomes, where the origins of the chromosome material could not be determined accurately because

108

John K. Cowell and Norma J. Nowak

of their small size (Cowell et al., 2003). When DNA from these cell lines, at exactly the same low passage used for SKY analysis, was subjected to CGHa, it became immediately possible to identify all of the major changes and, in some cases, ones that could not be seen by SKY. Importantly, each CGHa analysis was repeated three times for all of the cell lines and essentially produced the same profile, demonstrating the reproducibility of the technique between and within print runs of the arrays. An example of the CGHa profile from cell line 52 is shown in Fig. 2. In most cases (Cowell et al., 2003), the loss of entire chromosomes could be easily detected because the signal ratio along the entire set of BACs for that chromosome lay along the 0.5 meridian, similar to that seen for the X chromosome in the same profile. Similarly, large and small deletions also show this reduced ratio. In these cases the deletion ‘‘breakpoints’’ could be defined as being between the points where the transition occurred from a ratio of 0 to a ratio of 0.5 (Fig. 5). Gains in genetic material appear as a relative increase in the ratio compared with that seen for the adjacent BACs. One of the few drawbacks of CGHa is that it only detects numerical chromosome changes where there is a net change in DNA content. Thus, any balanced reciprocal chromosome translocations will not be detected. Nonbalanced translocations, however, result in specific gains and losses of genetic material, and these have been remarkable easy to identify. In tumor 124, for example, the reciprocal t(12;18)(p12;q12) detected by SKY results in gains of material on 12p and loss of material on 18q. This is clearly seen on the CGHa profile (Fig. 5), with the added advantage that the position of the breakpoints on both chromosomes can be determined and defined to within a very short interval. This rapid positioning of chromosome translocation breakpoints has a major advantage when trying to identify genes whose expression may be influenced by the translocation. Thus, in the t(12;18) translocation in tumor 124 (Fig. 5) the breakpoint on 3p occurs between 22.55 and 26.26 Mbp, and the breakpoint on 18q occurs between 32.06 and 34.19 Mbp. Before the development of this technology, definition of the position of these breakpoints would have required extensive, sequential FISH analysis on a BAC-by-BAC basis, or through the characterization of the breakpoints using PCR and somatic cell hybrids using a marker-bymarker approach (Hawthorn and Cowell, 1995). CGHa, therefore, provides an instant and highly accurate assessment of the location of the breakpoints.

D. CGHa analysis of primary brain tumors Comparison between SKY and CGHa for the four glioma cell lines showed that they were remarkably similar (Cowell et al., 2003). To determine whether the same changes could be identified in the tumors from

CGHa and Cancer Diagnostics

109

Fig. 5 Examples of chromosome changes detected by comparative genome hybridization (CGH) in brain tumors. The 12;18 nonreciprocal translocation seen in brain tumor 124 is shown in the two profiles above. This rearrangement results in extra copies of the distal 12p region and loss of material from distal 18q. The arrows indicate the positions of the breakpoints on each chromosome. As an example of how the CGHa ratios are sometimes suppressed in tumor tissue, the same profile for chromosome 18 from the corresponding tumor sample is shown below (18T). In this case, although the deletion is still clearly visible, the ratio for the deleted region is approximately 0.3 in the tumor, compared with a ratio of 0.5 in the cell line.

which the cell lines were derived, DNA was isolated from the original tumors and used to analyze the same CGH arrays. Remarkably, the overall profiles for the tumors were virtually identical to those seen for the cell lines. The most notable feature in all the tumor samples is that the ratios depicting losses and gains were slightly suppressed compared with the cell lines (Fig. 5). We presume that this reflects the fact that there are normal cells contaminating the tumor sample, which modifies the extent of the numerical changes seen in the cell lines (see earlier). There were, however, some obvious differences between the cell lines and the tumors. First, amplifications were far more common in the tumors, which means that the amplified genes have been selected against during the development of the cell lines. Examples of these are shown in Fig. 6. Tumor 4, for example, shows an

110

John K. Cowell and Norma J. Nowak

apparent 20–30-fold amplification of a region in 7p12 that contains the EGFR gene and a glioblastoma amplified sequence (GBAS), whereas the cell line carries two almost complete copies of chromosome 7. Amplification on the short arm of chromosome 11 in 11p12 was seen in tumor 4 (Fig. 6),

Fig. 6 Examples of amplifications in brain tumors. In tumor cell line 4, a ratio of 2 on the linear scale indicates the presence of two extra copies for the region flanking the centromere on chromosome 7 (indicated by the vertical line) but only one extra copy of the distal 7p and distal 7q regions. The bacterial artificial chromosome with a ratio of 1 in distal 7p represents an artifact that is seen in the majority of comparative genome hybridization a profiles. In the tumor cells from the same patient (center), the majority of chromosome 7 shows a normal diploid content with the exception of a clear 14-fold amplification in the region of 7p adjacent to the centromere that contains the EGFR gene. In the same tumor, amplification at two discrete regions of the short arm of chromosome 11, one containing the MDK2 gene, are also clearly seen.

CGHa and Cancer Diagnostics

111

which, interestingly, contains the MDK midkine (neurite growth-promoting factor 2) gene. In addition, a second region of amplification was observed by a series of BACs in 11p14.1–11p13. Overall, from the comparison between CGHa and SKY, we are now confident that the CGHa arrays can provide a clear and accurate picture of the numerical chromosome abnormalities (NCAs) present in the cells. Because we are using total DNA, these genetic changes represent the consensus changes in the tumor sample, because minor variations will almost certainly be lost in the overall analysis. In some cases, CGHa was essential to resolve the position of breakpoints on chromosomes involved in marker chromosome formation. An example of this was seen in tumor 52, which has a tetraploid karyotype (Fig. 3). A homozygous deletion was observed in the 9p21.3–21.2 region in this tumor by CGHa that was not detected by SKY. However, there were clearly not four copies of 9q in the karyotype. More detailed analysis revealed that there are in fact two copies of an isochromosome 9q, which produces the tetraploid value in CGHa. The short arm of chromosome 9 is present as two small marker chromosomes, each of which carries the 9p13-21 deletion.

E. Verification of Genetic Changes Identified Using CGH As with most high-resolution genomewide analysis approaches, it is often prudent to confirm some of the initial observations, especially where the data appear to indicate the involvement of a novel marker or gene in tumorigenesis. To verify losses and gains it is possible to use PCR analysis, where homozygous losses will result in the absence of an amplification product (Fig. 7). Partial losses can be detected using quantitative PCR. Heterozygous deletions can also be detected either by FISH analysis of chromosomes from the tumors (Fig. 7) or through an interphase FISH analysis, if metaphase chromosomes are not available. As an example, evidence for a heterozygous deletion on the long arm of chromosome 12 in brain tumor 4 was also observed using CGHa, and using BAC RP11-230i13, which lies within this region (Fig. 7), only one signal was seen on 12q, confirming the heterozygous loss of this region. We have also used PCR analysis to confirm several of the amplifications seen by CGHa (see later). In one such application of CGHa we have been studying its use in the identification of amplification of the HER2 locus in breast cancer that has prognostic significance as well as stratifying patients for particular clinical trials. From the results shown in Fig. 8, it is clear that this amplicon on the long arm of chromosome 17 can readily be detected and the extent of the amplicon determined.

112

John K. Cowell and Norma J. Nowak

Fig. 7 Confirmation of deletions identified by comparative genome hybridization a (CGH) Two homozygous deletions were identified in pancreatic adenocarcinomas PL6 and SNU308 on chromosomes 18 and 12, respectively (right). Polymerase chain reaction amplification of an internal control region on chromosome 12 (band 1) is present in the tumor and control, whereas a region within the deletion (band 2) shows normal amplification in the control but not in the DNA from the tumor. Similarly, control (band 3) and deletion regions (band 4) on chromosome 12 for PL6 show complete absence of the tumor but not the control DNA sample. In the lower panel, fluorescence in situ hybridization (FISH) was used to verify a heterozygous deletion in brain tumor 4. CGHa indicated heterozygous deletion at the distal tip of 12q (arrow). FISH analysis (left) shows only hybridization signal at the tip of one copy of chromosome 12 (arrows).

F. Definition of the Minimally Involved Regions from Overlapping NCAs One of the important consequences of CGHa analysis is that it will detect the full range of large and small deletions and amplifications in cancer cells, which will then allow a compilation of the consensus region involved. This type of information has been used extensively in the past in attempts to clone genes important in tumorigenesis. As an example, the most frequent deletion seen in the brain tumor samples involves the 9p13-21 region harboring the CDKN2A (p16) gene. On the basis of the CGHa, we were able to define the consensus deletion (Fig. 9). Previously this sort of fine definition was only possible using somatic cell hybrids (Hawthorn and Cowell,

CGHa and Cancer Diagnostics

113

Fig. 8 Examples of HER2 amplification on the long arm of chromosome 17 in breast tumors. The scales in each case have been adjusted to demonstrate the variable levels of amplification (level indicated by Xr). In the example in the upper comparative genome hybridization a (CGH) profile, amplification of a more centromeric region is also seen. The location of the HER2 amplicon is clearly seen through the alignment of the profiles.

1995) or through a probe-by-probe FISH analysis. To confirm the CGHa observations for these deletions, we prepared DNA from the same individual BACs that were identified as deleted on the array and performed FISH

114

John K. Cowell and Norma J. Nowak

Fig. 9 Analysis of the consensus region of deletion in four brain tumors involving the CDKN2A/B locus on the short arm of chromosome 9. The maximum extent of the deletions is indicated by the arrows. In tumor (T) 52, two deletions were identified using fluorescence in situ hybridization involving a large region on one chromosome (arrows) and a smaller deletion on the other (horizontal bar). When the extent of all of the deletions is examined (below), it is possible to define the minimal region of overlap that contains only the four genes indicated.

CGHa and Cancer Diagnostics

115

analysis of the chromosomes from the individual cell lines. In all four cases we could confirm either a homozygous or heterozygous deletion as determined by CGHa; moreover, the complexity of the events could also be resolved. Thus, on the short arm of chromosome 9 in tumor 52 (Fig. 9), for example, we were able to show that the CDKN2A gene was completely absent using FISH, indicating a homozygous deletion, as indicated in the CGHa profile. BACs that flanked the CDKN2A gene showed only a single hybridization signal, demonstrating only heterozygous loss of this region. Thus, FISH analysis was able to confirm the CGHa observation and define the two independent deletion events that had taken place during the evolution of this tumor: one confined largely to the region containing the CDKN2A gene and another that was more extensive.

G. Determination of the Gene Content on the NCAs The analysis of specific chromosome deletions has frequently led to the discovery of the gene or genes involved in the associated phenotype. The ability to use CGHa to clearly determine the gene content within a deleted or amplified region and to compare these observations between patients provides a rapid way of selecting candidate genes for more detailed study. For amplifications, for example, the affected region is defined by the trailing end of the unamplified clone preceding the amplification and the leading end of the unamplified clone following the amplification. It is not sufficient to look for candidate genes only within the BAC that is amplified, as genes in the regions between the BACs, which may or may not be amplified, should also be considered. The mapping data for each BAC is found by querying the human genome sequence at http://genome.ucsc.edu. The 14 November 2002 build is currently being used to precisely position the BAC clones on the draft sequence. In Ensembl, the BAC clones can be aligned directly to the draft of human sequence using its BAC address or chromosome and nucleotide locations. Detailed information, including the known and predicted genes, additional BAC clones, SNPs, and sequences, can also be obtained for the affected regions. Thus, the 9p consensus deletion shown in Fig. 9 contains only the interferon gene cluster (IFNA) and the KIAA1354, MTAP, and CDKN2A/B genes. Similarly, a fragment of chromosome 11 from a pancreatic adenocarcinoma xenograft contains PDX1, CD44, SLC1A2, FJX1, and TRIM44. Interestingly, overexpression of CD44 has been implicated in pancreatic cancers (Ringel et al., 2001; Tsukada et al., 2001). This approach for characterizing breakpoints, therefore, has major advantages over conventional cytogenetics and PCR in that, in a single experiment, it is now possible to establish the minimum overlap of deletions in the same region from a single hybridization for each sample

116

John K. Cowell and Norma J. Nowak

and to accurately define the genes consistently involved in the deletion event. The analysis of the breakpoints associated with chromosome translocations is also simplified because the position of the breakpoint can also be defined to within 750 kb, which can make the cloning of that breakpoint a more straightforward effort.

H. Comparison Between CGHa and Expression Array Analysis The BAC arrays define physical loss and gain of genetic material and, by analysis of the gene content of that region, it is possible to determine which genes are lost or amplified. This information can be used to select candidate genes that are the drivers for the genetic change observed. Although it is possible to conduct an extensive gene-by-gene quantitative analysis of the affected region to determine which of these genes are expressed or lost in the tumor tissue, combining the analysis of the same sample using an expression array platform, such as the Affymetrix oligonucleotide arrays, can greatly streamline the choice of candidate genes. An example of this is given in Fig. 10, where analysis of primary lung tumor using CGHa identified a series of deletions on chromosome 7. In one case, eight genes were located in a 700-kb region, of which two genes were not present on the U95A chip used for expression analysis. By comparing the expression levels between tumor and normal samples from one patient, for example, only two genes showed a concomitant downregulation. In this way, it is possible to prioritize the genes in a larger region for more detailed analysis of their involvement in tumorigenesis.

I. Comparison Between CGHa and Loss of Heterozygosity One of the most widely investigated phenomena associated with the development of cancer involves so-called loss of heterozygosity (LOH). It was shown in 1983 (Cavenee et al., 1983) that regions of chromosomes that were heterozygous in the constitutional DNA from a patient could become homozygous at specific loci in their tumors. In 1971, Al Knudson had suggested a mathematically based two-hit hypothesis for the development of certain childhood cancers such as retinoblastoma. The essence of this theory, as it developed (Knudson, 1985), was that homozygous mutational inactivation of a critical cancer gene could give rise to tumorigenesis. These observations gave rise to the concept of tumor suppressor genes, as the presence of a normal copy of these genes protected the cells from tumorigenesis. Thus, in familial cases of cancer the first hit is inherited and this is present in

Fig. 10 Comparison of comparative genome hybridization a (CGHa) with Affymetrix gene expression analysis. A CGHa profile for chromosome 7 from a primary lung tumor is shown in the top panel. One region of deletion (arrow) between 98.2 and 98.9 Mbp was selected and the gene content (TRRAP – CYP3A) of that region determined from the University of California Santa Cruz Web site. In the histogram below are the-fold increases in expression for these genes in a comparison between the tumor and normal lung from the same patient. Two genes (KIAA632 and VIK) were not present on the U95 chip used. Other genes (TRRAP, ZFP95, and CYP3A) showed no change in expression levels. The PDAP and CPSF4 genes showed the reduced expression level expected from a deletion of the region.

117

118

John K. Cowell and Norma J. Nowak

every somatic cell, so that when the second hit occurs as a random event in a tumor precursor cell, tumorigenesis is initiated. In sporadic cases, both hits are random events in the same cell but achieve the same result: homozygous inactivation of the function of the critical tumor suppressor gene. In the work of Godbout et al. (1983) and Cavenee et al. (1983), it was presumed that the LOH in retinoblastoma tumors was the mechanism whereby a recessive mutation in one allele of the critical RB1 gene was ‘‘exposed’’ as a result of loss of the chromosome region containing the wild-type allele in the tumor cells. Cavenee and colleagues (1985) went on to formally demonstrate the mechanism in hereditary cases of Rb in which it was possible to track the chromosome region containing the mutant allele in families using standard linkage analysis. In these studies, it was shown that it was the chromosome region in 13q14 that was inherited from the affected parent (and so contains the gene defect) that was retained in the tumors, formally demonstrating the two-hit hypothesis. In addition to demonstrating the principle of LOH, it also became clear that the end point, loss of the normal allele to expose the mutation in the other, could be achieved by different mechanisms (Cavenee et al., 1983). First, whole-chromosome loss or interstitial deletions in Rb tumors accounted for 70% of the LOH events. Mitotic recombination between sister chromatids accounted for the other 30%. In other tumors, such as WT, we have shown that mitotic recombination rather than deletion is much more frequently the cause of LOH (Wadey et al., 1989). These early demonstrations that LOH analysis could pinpoint the genomic position of tumor suppressor genes and provide a mechanism for loss of function of these TSGs led to an explosion of studies in all cancers to define these events. In some cases, even in the absence of an understanding of the nature of the gene or genes that is involved, the observation alone has led to new approaches in classifying subgroups of patients on the basis of LOH profiles. One such example involves a subclass of brain tumors called anaplastic oligodendrogliomas (AO). Within this subgroup are patients who respond well to chemotherapy and others who do not. It was shown (Cairncross et al., 1998) that patients whose tumors showed LOH for markers on 1p had a good response to chemotherapy, compared with those who had tumors that retained heterozygosity. Empirically it was shown that the LOH at 1p was usually accompanied with LOH from 19q. In addition, loss of function of the CDKN2A/B genes was also associated with poor survival in this subclass of brain tumors. The majority of this analysis was performed using PCR at discrete loci along the length of 1p and 19q, although different laboratories have used different markers, making direct cross comparisons difficult. If the LOH in these cases is caused by loss of genetic material, rather than mitotic recombination, then CGHa should reveal this genetic change and provide a rapid alternative to the labor-intensive analysis, using microsatellite markers to establish chemosensitivity in these tumors.

Fig. 11 Analysis of the 1p and 19q deletions from two anaplastic oligodendrogliomas. From these two examples it is clear that the entire chromosome arms are deleted. These examples also show the reproducibility of hybridization between samples for bacterial artifical chromosomes that do not behave consistently along the length of the chromosome, possibly because of the presence of high concentration of repetitive sequences (see text). On chromosome 1p, for example, there are seven bacterial artificial chromosomes that show a ratio of 1 in both cases in which all of the other BACs for this chromosome arm show a ration of 0.6.

119

120

John K. Cowell and Norma J. Nowak

In an LOH study of low-grade oligodendrogliomas, AO, and mixed astrocytoma/oligodendroglioma, we previously defined tumors that did, and did not, show LOH for 1p and 19q (Chernova et al., 2003), using individual microsatellite markers. When we used these DNA samples to investigate the relationship between CGHa profiles and LOH (Fig. 11), it was clear that LOH resulted from loss of genetic material along the entire length of chromosome arms 1p and 19q. Thus, in these tumors LOH was caused by physical deletion of the chromosome arms rather than by mitotic recombination. In addition, these losses did not involve variable-length interstitial deletions. In fact, we were not able to identify a single tumor in this series in which there was loss of 1p without coincident loss of 19q. Thus, CGHa can be used as a realistic alternative for the analysis to predict chemoresponsiveness in this type of tumor. BAC array analysis of pancreatic adenocarcinoma has also determined amplification of a region in 17q containing the topoisomerase II alpha gene that may influence prognosis and choice of therapy. The clear advantage to this CGHa approach is that, in addition to defining the regions being used specifically to predict clinical outcome, all of the other information about genetic changes in the cells is also obtained in the same experiment, which might reveal more subtle associations between genetic changes and clinical parameters.

J. Analysis of Archival Tissue Material If specific CGHa genetic fingerprinting is to be used eventually as a diagnostic tool, then large numbers of tumors must be screened to identify consistent changes. Being able to take advantage of the archival material available in many pathology departments, therefore, requires that DNA derived from fixed tissues that have been embedded in paraffin can perform in a robust way in CGHa. To determine to what extent this archival material can be used for CGHa, we compared DNA isolated from a freshly frozen sample of breast tumor cells with the same material that was fixed and embedded in paraffin. When these DNAs were hybridized to the BAC arrays, there was a striking similarity in the overall profile (Fig. 12). From the hybridization ratios, however, it was clear that the signal produced from the fixed tissue was very weak, which resulted in the processing software rejecting many more BACs (see earlier) than in the fresh tissue; these BACs were excluded from the CGHa profile. This feature also significantly increased the signal-to-noise ratios. We presume that a major factor in this reduced quality of the data comes from a reduced efficiency either in labeling the probe or in its ability to hybridize to the target BACs along the length of the chromosome. Nonetheless, within this background noise, it is often relatively easy to determine some of the more obvious

CGHa and Cancer Diagnostics

121

Fig. 12 Comparison of comparative genome hybridization a profiles for chromosome 20 from chromosome 17 from a breast cancer tumor sample in which the DNA used as the probe was generated from fresh material (20T) or paraffin-embedded cells (20P). In general, the profiles are the same and the amplified region in distal 20q is seen in both cases. The number of bacterial artificial chromosomes that passed the quality control parameters (see text) in the 20P sample, however, was reduced. This effect is clearly seen on the proximal short arm of the chromosome (indicated by the bar).

genetic events (Fig. 12) that have taken place, but it is still not clear whether small heterozygous changes, if they happen to fall in a region of poor hybridization, could be lost in the noise. Fragmentation of the DNA as a result of formalin cross-linking may be a contributing factor to the noise, which prevents equal representation of the whole genome in the probe following labeling using random priming. It is clear, therefore, that the quality of the starting DNA is the most crucial aspect for the success of CGHa analysis. We have tried a number of approaches for labeling formalin-fixed archived material and have found that success is greatly dependent on the source of the material, organ type, storage conditions, and time elapsed since fixation. Thus, CGHa from archival material remains a challenge for the identification of small NCAs, but in general, the larger changes can be identified.

122

John K. Cowell and Norma J. Nowak

V. SUMMARY Chromosome changes in cancer have repeatedly led to the identification of genetic changes that drive the malignant phenotype and provide opportunities for targeted therapy. The amplification of HER2 in breast cancer led to the development of Herceptin, for example, and the observation of loss of genetic material from the short arm of chromosome 1 in subsets of brain tumors has provided a valuable means for stratifying patients for therapy. The availability of BAC arrays means that these types of abnormality can not only be identified but also now be put into context with total genome changes in tumors. The high resolution of the analysis also promises to be able to detect even smaller changes than is currently possible with chromosome-based approaches. Over the coming years it is very likely that the density of these arrays will approach the theoretical maximum of 30,000 clones, which will span the genome uninterrupted. At this point it should be possible to identify even the smallest deletions and amplifications in tumor cells, with the caveat that regions with high densities of repetitive elements may not be amenable to this type of analysis. Although CGHa may replace many of the applications of cytogenetics for the analysis of tumors, the drawback is the inability of this approach to characterize chromosome translocations and insertions/inversions to the same resolution as SKY, where there are no net changes in DNA content as a result of the rearrangement. This will be particularly important in the analysis of leukemias, where the classification of tumor types is very dependent on being able to characterize these translocations. It is possible, however, that as more-extensive CGHa becomes possible, the profile of genetic changes will provide new means of classifying tumors that are independent of existing ones. CGHa is particularly amenable to a high-throughput, objective analysis for routine characterization of samples, although at present, routine use of CGHa for molecular diagnostics is restricted because of the need for specialist instruments for preparation and analysis of the arrays, although these arrays are becoming more commonplace. The technically demanding approach for preparing the printing DNA solutions is also a restriction to more general use, and the cost of reagents, particularly the Cye dyes, makes the procedure relatively expensive. As alternative approaches of probe labeling and CGHa analysis are developed, however, these costs will inevitably be reduced. Although we have concentrated on the use of BAC arrays for the analysis of human tumors, this technology is by no means restricted, and obviously any BAC library (e.g., zebrafish or rat) can similarly be arrayed. For example, the mouse has been used extensively to develop models for specific cancers caused by overexpression of transgenes as well as to identify lowpenetrance tumor susceptibility genes (Hann and Balmain, 2001). The

CGHa and Cancer Diagnostics

123

sequence of the mouse genome is now generally available, and BAC libraries from the mouse are available (e.g., at Roswell Park Cancer Institute). The advantage of these modes is that large numbers of tumors can be produced relatively quickly for analysis, and on a relatively homogenous genetic background. Because the genetic changes that occur in human tumors are also seen in the development of mouse tumors, identifying consistent changes using BAC arrays provides important clues to the nature of genetic changes that can then be tested in human cancers.

REFERENCES Baldini, A., Ross, M., Nizetic, D., Vatcheva, R., Lindsay, E. A., Lehrach, H., and Siniscalco, M. (1992). Genomics 14, 181–184. Cairncross, J. G., Ueki, K., Zlatescu, M. C., Lisle, D. K., Finkelstein, D. M., Hammond, R. R., Silver, J. S., Stark, P. C., Macdonald, D. R., Ino, Y., Ramsay, D. A., and Louis, D. N. (1998). J. Natl. Cancer Inst. 90, 1473–1479. Casperson, T., Farber, S., Foley, G. E., Kudynowski, J., Modest, E. J., Simonsson, E., Wagh, U., and Zech, L. (1968). Exp. Cell Res. 49, 219–222. Cavenee, W. K., Dryja, T. P., Phillips, R. A., Benedict, W. F., Godbout, R., Gallie, B. L., Murphree, A. L., Strong, L. C., and White, R. L. (1983). Nature 305, 779–784. Cavenee, W. K., Hansen, M. F., Nordenskjold, M., Kock, E., Maumenee, I., Squire, J. A., Phillips, R. A., and Gallie, B. L. (1985). Science 228, 501–503. Chernova, O. B., Barnett, G. H., and Cowell, J. K. (2003). Br. J. Cancer 88, 1989–1893. Cowell, J. K. (1982). Ann. Rev. Genet. 16, 21–59. Cowell, J. K. (Ed.). (2001). Molecular Genetics of Cancer, 2nd ed. BIOS Publishing Corporation, Oxford. Cowell, J. K. (1989). Ophthalmol. Ped. Genet. 2, 75–88. Cowell, J. K., Jaju, R., and Kempski, H. (1994). J. Med. Genet. 31, 334–337. Cowell, J. K., Hungerford, J., Rutland, P., and Jay, M. (1989a). Ophthal. Ped. Genet. 110, 117–127. Cowell, J. K., Wadey, R. B., Buckle, B., and Pritchard, J. (1989b). Hum. Genet. 82, 123–126. Cowell, J. K., Matsui, S., Wang, J., LaDuca, J., Conroy, J., McQuaid, D., and Nowak, N. J. (2003). Cancer, Genetics, and Cytogenetics, in press. Davis, A., and Cowell, J. K. (1993). Hum. Mol. Genet. 2, 2093–2097. Francke, U., and Oliver, N. (1978). Hum. Genet. 45, 137–165. Geschwind, D. H., Gregg, J., Boone, K., Karrim, J., Pawlikowska-Haddal, A., Rao, E., Ellison, J., Ciccodicola, A., D’Urso, M., Woods, R., Rappold, G. A., Swerdloff, R., and Nelson, S. F. (1998). Dev. Genet. 23, 215–229. Getman, M. E., Houseal, T. W., Miler, G. A., Grundy, P., Cowell, J. K., and Landesc, G. M. (1998). Cytogenet. Cell Genet. 82, 284–290. Godbout, R., Dryja, T. P., Squire, J., Gallie, B. L., and Phillips, R. A. (1983). Nature 304, 451–453. Grundy, R. G., Aledo, R., and Cowell, J. K. (1998). Int. J. Mol. Med. 1, 801–808. Grundy, R. G., Pritchard, J., Baraitser, M., Risdon, A., and Robards, M. (1992). Eur. J. Pediatr. 151, 895–898.

124

John K. Cowell and Norma J. Nowak

Gutierrez, J., Sepulveda, W., Saez, R., Carstens, E., and Sanchez, J. (2001). Ultrasound Obstet. Gynecol. 17, 166–168. Hann, B., and Balmain, A. (2001). Curr. Opin. Cell Biol. 13, 778–784. Hawthorn, L. A., and Cowell, J. K. (1995). Genomics 27, 399–404. Hedenfalk, I., Ringner, M., Ben-Dor, A., Yakhini, Z., Chen, Y., Chebil, G., Ach, R., Loman, N., Olsson, H., Meltzer, P., Borg, A., and Trent, K. (2003). Proc. Natl. Acad. Sci. 100, 2532–2537. Hsu, T. C. (1952). J. Hered. 43, 167–172. Kallioniemi, A., Kallioniemi, O. P., Piper, J., Tanner, M., Stokke, T., Chen, L., Smith, H. S., Pinkel, D., Gray, J. W., and Waldman, F. M. (1994). Proc. Natl. Acad. Sci. USA 91, 2156–2160. Kallioniemi, A., Kallioniemi, O. P., Sudar, D., Rutovitz, D., Gray, J. W., Waldman, F., and Pinkel, D. (1992). Science 258, 818–821. Kempski, H., McDonald, D., Michalski, A. J., Roberts, T., Goldman, J. M., Cross, N. C. P., and Cowell, J. K. (1995). Genes Chroms. Cancer 12, 283–287. Knudson, A. G. (1971). Proc. Natl. Acad. Sci. USA 68, 820–823. Knudson, A. G., Jr (1985). Cancer Res. 45, 1437–1443. Landegent, J. E., Jansen in de Wal, N., Dirks, R. W., Baao, F., and van der Ploeg, M. (1987). Hum. Genet. 77, 366–370. Lele, K. P., Penrose, L. S., and Stallard, H. B. (1963). Ann. Hum. Genet. 27, 171–174. Lucito, R., West, J., Reiner, A., Alexander, J., Esposito, D., Mishra, B., Powers, S., Norton, L., and Wigler, M. (2000). Genome Res. 10, 1726–1736. Luo, J., Balkin, N., Stewart, J. F., Sarwark, J. F., Charrow, J., and Nye, J. S. (2000). Am. J. Med. Genet. 91, 227–230. Mannens, M., Hoovers, J. M. N., Redeker, E., Vergjaal, M., Feinberg, A. P., Little, P., Boavida, M., Coade, N., Steenman, M., Bliek, J., Niikawa, N., Tononki, H., Nakamura, Y., de Boer, E. G., Slater, R. M., John, R., Cowell, J. K., Junien, C., Henry, I., Tommerup, N., Weksberg, R., Pueschel, S. M., Leschot, N. J., and Westerveld, A. (1994). Eur. J. Hum. Genet. 41, 1–22. Matsui, S., Faitar, S. L., Rossi, M. R., and Cowell, J. K. (2003). Cancer Genet. Cytogenet. 142, 30–35. Mitelman, F. (2000). Catalog of chromosome aberrations in cancerhttp://www.wiley.com/ products/subject/life/mitelman/ Mitelman, F., and Levan, G. (1981). Hereditas 95, 79–139. Nowak, N., Cheung, V. G., Jang, W., Kirsch, I. R., Zhoa, S., Chen, X. N., Furey, T. S., Kim, U. J., Kuo, W. L., Olivier, M., Conroy, J., Kasprzyk, A., Massa, H., Yonescu, R, Sait, S., Thoreen, C, Snijders, A., Lemyre, E., Bailey, J. A., Bruzel, A., Burrill, W. D., Clegg, S. M., Collins, S., Dhami, P., Friedman, C., Han, C. S., Herrick, S., Lee, J., Ligon, A. H., Lowry, S., Morley, M., Narasimhan, S., Osoegawa, K., Peng, Z., Plajzer-Frick, I., Quade, B. J., Scott, D., Sirotkin, K., Thorpe, A. A., Gray, J. W., Hudson, J., Pinkel, D., Reid, T., Rowen, L., Shen-Ong, G. L., Strausberg, R. L., Birney, E., Callen, D. F., Cheng, J. F, Cox, D. R., Doggett, N. A., Carter, N. P., Eichier, E. E., Haussler, D., Korenberg, J. R., Morton, C. C., Albertson, D., Schuler, G., de Jong, P. J., and Trask, B. J. (2001). Nature 409, 953–958. Pinkel, D., Straume, T., and Gray, J. W. (1986). Proc. Natl. Acad. Sci. USA 83, 2934–2938. Pinkel, D., Landegent, J., Collins, C., Fuscoe, J., Segraves, R., Lucas, J., and Gray, J. (1988). Proc. Natl. Acad. Sci. USA 85, 9138–9142. Pollack, J. R., Perou, C. M., Alizadeh, A., Eisen, M. B., Pergamenschikov, A., Williams, C. F., Jeffrey, S. S., Botstein, D., and Brown, P. O. (1999). Nat. Genet. 23, 41. Pollack, J. R., Sorlie, T., Perous, C. M., Rees, C. A., Jeffrey, S. S., Lonning, P. E., Tibshirani, R., Botstein, D., Borresen-Dale, A.-L., and Brown, P. O. (2002). Proc. Natl. Acad. Sci. USA 99, 12963–12968. Riccardi, V. M., Sujansky, E., Smith, A. C., and Francke, U. (1978). Pediatrics 61, 604–610.

CGHa and Cancer Diagnostics

125

Ringel, J., Jesnowski, R., Schmidt, C., Ringel, J., Kohler, H. J., Rychly, J., Batra, S. K., and Loh, M. (2001). Teratog. Carcinog. Mutagen 21, 97–106. Rowley, J. D. (1998). Ann. Rev. Genet. 32, 495–519. Sandberg, A. A. (1990). The Chromosomes in Human Cancer and Leukemia 2nd ed. New York, Elsevier. Seabright, M. (1971). Lancet 2, 971–972. Snidjers, A., Nowak, N., Seagraves, R., Blackwood, S., Brown, N., Conroy, J., Hamilton, G., Hindle, A., Huey, B., Kimura, K., Law, S., Myambo, K., Palmer, J., Yistra, B., Yue, J., Gray, J., Jain, A., Pinkel, D., and Albertson, D. (2001). Nat. Genet. 29, 263–264. Still, I. H., Roberts, T., and Cowell, J. K. (1997). Annal Hum. Genet. 61, 15–24. Tjio, J. H., and Levan, A. (1956). Hereditas 42, 1–6. Tsukuda, H., Matsumura, Y., Inoue, T., Kobayashi, Y., Ookawa, K., Aoki, T., Nebiki, H., Yamazaki, O., Kondo, H., and Kakizoe, T. (2001). Anticancer Res. 21, 2175–2183. Wadey, R. B., Pal, N. P., Buckle, B., Yeomans, E., Pritchard, J., and Cowell, J. K. (1990). Oncogene 5, 901–907. Yunis, J. J. (1976). Science 191, 1268–1270. Yunis, J. J., and Ramsay, N. (1978). Am. J. Dis. Child 132, 161–163.

This Page Intentionally Left Blank

Natural Killer Cells and Cancer Jun Wu1 and Lewis L. Lanier2 1

Shanghai Genomics, Inc., and Chinese National Genome Center, Shanghai, 2 China and Department of Microbiology and Immunology and the Cancer Research Institute, University of California, San Francisco

I. Introduction II. NK Cell Biology A. NK Cells in Cancer B. NK Cells in Viral Immunity—A Link to Their Role in Tumor Immunity III. NK Cell Effector Functions A. Perforin/Granzyme-Mediated Cytotoxicity B. TRIAL/Fas Ligand-Mediated Apoptosis C. Interferon-y-mediated Effector Functions IV. Receptors Turning NK Cells ‘‘On’’ and ‘‘Off’’ A. Turning NK Cells Off—Inhibitory MHC Class I Receptors B. Activating NK Receptors V. NK Cells in Tumor Immunosurveillance VI. Tumor Escape Mechanisms VII. Conclusion References

Natural killer (NK) cells are lymphocytes that were first identified for their ability to kill tumor cells without deliberate immunization or activation. Subsequently, they were also found to be able to kill cells that are infected with certain viruses and to attack preferentially cells that lack expression of major histocompatibility complex (MHC) class I antigens. The recent discovery of novel NK receptors and their ligands has uncovered the molecular mechanisms that regulate NK activation and function. Several activating NK cell receptors and costimulatory molecules have been identified that permit these cells to recognize tumors and virus-infected cells. These are modulated by inhibitory receptors that sense the levels of MHC class I on prospective target cells to prevent unwanted destruction of healthy tissues. In vitro and in vivo, their cytotoxic ability can be enhanced by cytokines, such as interleukin (IL)-2, IL-12, IL-15 and interferon / (IFN-/). In animal studies, they have been shown to play a critical role in the control of tumor growth and metastasis and to provide innate immunity against infection with certain viruses. Following activation, NK cells release cytokines and chemokines that induce inflammatory responses; modulate monocyte, dendritic cells, and granulocyte growth and differentiation; and influence subsequent adaptive immune responses. The underlining mechanism of discriminating tumor cells and normal cells by NK cells has provided new insights into tumor immunosurveillance and has suggested new strategies for the treatment of human cancer. ß 2003 Elsevier Inc.

Advances in CANCER RESEARCH 0065-230X/03 $35.00

127

Copyright 2003, Elsevier Inc. All rights reserved

128

Jun Wu and Lewis L. Lanier

I. INTRODUCTION The ability of leukocytes, without deliberate immunization or activation, to kill certain tumors was first appreciated in the late 1960s and early 1970s, when investigators were attempting to generate tumor-antigenspecific cytotoxic T lymphocytes (CTL). Many labs working in the field experienced this ‘‘background’’ cytolytic activity in their in vitro cytotoxicity assays, and it soon became apparent that this was more than a technical artifact. Although there were numerous reports describing this activity, the first study to coin the term ‘‘natural’’ killer cells was an article by Kiessling, Klein, and Wigzell in 1975 (Kiessling et al., 1975). A distinguishing feature of NK cells is their ability to recognize and kill tumor cells that completely lack expression of MHC class I and II antigens. For a comprehensive review of the early developments in this field, the article by Trinchieri (1989) provides an excellent resource. Before discussing the role of NK cells in relationship to cancer, we will begin with a brief review of the cells responsible for ‘‘natural cytotoxicity,’’ their distinguishing characteristics, and their effector mechanisms. NK cells are a lineage of lymphocytes, distinct from B and T cells in that they do not require recombinase activity for development or for the generation of their receptors involved in tumor cell recognition. Morphologically, they have been described as ‘‘large granular lymphocytes’’ because of the predominant azurophilic granules in the cytoplasm and their somewhat larger size than either resting B or T cells (Timonen et al., 1979); however, smaller, agranular NK cells also exist (Smyth et al., 1995). They comprise 5%– 20% of peripheral blood lymphocytes, 5% of splenic lymphocytes, and 10% of hepatic lymphocytes, and they are present at lower frequencies in other hematopoietic tissues; for example, bone marrow, thymus, and lymph nodes. NK cells and T cells share many surface markers, likely because they arise from a common T/NK progenitor cell (Lian and Kumar, 2002; Spits et al., 1995) and mediate many of the same effector functions (e.g., cellmediated cytotoxicity and cytokine secretion). NK cell development requires IL-15 (Kennedy et al., 2000), and the cytolytic activity of NK cells against tumors can be substantially augmented by activation with IL-2, IL-12, IL15, and IFN-/. NK cells are unable to produce any of these cytokines, but are dependent on other cell types to provide these factors. The activation of NK cells by IFN or cytokines or by cognate interactions with tumor cells also induces NK cells to transcribe and secrete certain cytokines and chemokines, including IFN-, granulocyte–macrophage hematopoietic colony-stimulating factors, tumor necrosis factor- (TNF-), and others. Despite some similarities to T cells, NK cells are not dependent on the thymus for development, and they lack the T-cell antigen receptor and expression of the CD3 complex on the cell surface (Cooper et al., 2001;

NK Cells and Cancer

129

Lanier et al., 1992; Lian and Kumar, 2002; Spits et al., 1995). They develop in the bone marrow from NK-progenitors, and cytokines, such as stem cell factor, FLT3 ligand, and IL-7 promote NK cell development, although they are not absolutely required (Colucci et al., 2003; Lian and Kumar, 2002). IL-15, which is produced by the bone marrow stroma, is required for NK cell development and drives the final maturation process (Kennedy, et al., 2000; Mrozek et al., 1996). The most commonly used surface marker to identify human NK cells, CD56, is expressed on all NK cells and a small subpopulation of memory T cells, as well as on neural tissues (Lanier et al., 1989a). However, CD56 is not involved in NK activation and killing, and its function on NK cells is still unknown. Murine NK cells do not express CD56 (although CD56 is expressed in the brain tissues in mice), but are often identified by using the anti-NK1.1 monoclonal antibody (mAb; Koo and Peppard, 1984) that reacts with NKR-P1C, or DX5 mAb (Arase et al., 2001), which detects VLA-2 (2,1 integrin). Another useful NK marker is CD16A (FcRIII), a low-affinity Fc receptor for IgG, which is expressed on 90% of human NK cells and 50% of mouse NK cells (Trinchieri, 1989). CD16 enables NK cells to mediate antibody-dependent cell-mediated cytotoxicity against IgG-coated target cells, thereby conferring the antigen specificity of antibodies to NK cells. Major progresses have been made in recent years in identifying receptors that allow NK cells to discriminate between normal and tumor- or virusinfected cells. NK cells can be activated and subsequently lyse target cells that have lost MHC-class I or that express subnormal levels of MHCclass I molecules, a common event following transformation or viral infection. As predicted by the ‘‘missing self’’ hypothesis (Karre et al., 1986), NK cells are usually prevented from attacking cells expressing self MHC class I because they express surface receptors for MHC class I delivering signals that inhibit NK function (Karlhofer et al., 1992; Ravetch and Lanier, 2000). Studies of these MHC-specific inhibitory receptors have revealed an extraordinary complexity and specificity to prevent damage to normal healthy cells (Long, 1999; Moretta et al., 2001; Ravetch and Lanier, 2000). However, recent discovery of a group of activating surface receptors has solved the long-standing mystery of the ‘‘on’’ signal (Cerwenka and Lanier, 2001; Moretta et al., 2001). Unlike T and B cells that possess a single dominant antigen receptor, NK cells use a variety of activating receptors with specialized signaling machinery to recognize and kill unhealthy and abnormal cells. These activating receptors, as will be reviewed later, may determine the distinct roles of NK cells in various phases of the immune responses, including tumor surveillance. The final disposition of an NK cell response is tightly regulated by an intricate balance between the opposing signals from the activating versus inhibitory receptors.

130

Jun Wu and Lewis L. Lanier

As their name implies, a predominant effector function of NK cells is cellmediated cytotoxicity. This activity is predominantly mediated by the release of the contents of their cytoplasmic granules after encountering a ‘‘NK-sensitive’’ target cell. The process, called granule-mediated exocytosis, is directional, delivering the lethal payload directly into the interface between the NK cell and its target, thus preventing bystander killing (Henkart, 1994; Trapani and Smyth, 2002). The granules in NK cells, as in CTL, contain perforin (a pore-forming protein with homology to complement factor 9) and several granzymes (proteases that act to cleave caspases in the target cells and initiate apoptosis by both caspase-dependent and caspase-independent mechanisms). Although in most circumstances the perforin and granzyme-mediated pathway is responsible for their lytic function, NK cells can also express certain members of the TNF family, including membrane and secreted TNF, lymphotoxin, Fas ligand (FasL), and TNF-related-apoptosis-inducing ligand (TRAIL), that may also participate in the killing of certain sensitive target cells (reviewed in Trapani and Smyth, 2002). NK cells also are an important source of cytokines and chemokines during an immune response. The production of these factors can be induced by cognate interactions between NK cells and tumors or virus-infected cells, or they can be triggered as a bystander event in response to cytokines or interferons in the local environment. Although NK cells are able secrete numerous cytokines (e.g., IFN-, GM-CSF, G-CSF, M-CSF, TNF, IL-5, IL-10, IL-13, and others, Peritt et al., 1998) and chemokines (XCL1, CCL1, CCL3, CCL4, CCL5, CCL22, CXCL8, and others; Robertson, 2002), based on in vivo studies perhaps their most important function is their ability to rapidly produce IFN- early during an immune reaction (Biron et al., 1999). This pivotal cytokine is important in the activation of macrophages and in shaping the subsequent adaptive immune response. A critical role for IFN- in immune surveillance against cancer will be discussed later in this review.

II. NK CELL BIOLOGY A. NK Cells in Cancer NK cells were originally defined as fresh-isolated white blood cells that are capable of lysing certain tumor targets, such as K562, a tumor cell line derived from a patient with chronic myelogenous leukemia (Ortaldo et al., 1977), or YAC-1, a Moloney tumor virus-induced lymphoma from A/Sn mice (Kiessling et al., 1975). Activation of NK cells with IL-2 or IFN-/

NK Cells and Cancer

131

further enhances their cytolytic function, resulting in killing a broad array of other tumor targets that are not lysed by resting NK cells (Henney et al., 1981). IL-2 activated lymphocytes with the ability to kill tumor cells were initially designated ‘‘lymphokine-activated killer’’ (LAK) cells (Grimm et al., 1982); however, it was subsequently shown that almost all of this activity was mediated by activated NK cells (Phillips and Lanier, 1986). Numerous studies have also shown that NK cells preferentially kill MHC class I–deficient tumor cells in vitro and can reject class I–deficient tumors in vivo (Ljunggren and Karre, 1985). This may be relevant to human cancer, in that human tumors apparently frequently show loss of expression of MHC class I antigens on the cell surface (reviewed in Algarra et al., 2000). Thousands of published articles have provided unambiguous evidence that NK cells, particularly activated NK cells, possess the capacity to recognize and kill tumor cells in vitro and serve to prevent growth and metastasis of certain tumors in vivo. Several recent reviews provide an excellent overview of the literature on this subject (Herberman, 2002; Smyth et al., 2001c, 2002b). Although it is impossible to cite and discuss all publications dealing with a potential role for NK cells in antitumor activity, we will select a few representative studies to illustrate general concepts that have emerged from this large body of work done by many laboratories. In considering the evidence for the participation of NK cells in immunity against cancer in vivo, it is useful to consider separately the different situations in which this has been studied. These include NK cell protection against the growth and metastasis of transplantable tumors in rodents, NK cell protection against spontaneous or carcinogen-induced tumors in experimental animals, correlative studies indicating a role for NK cells in recognition of human tumors in vivo, and adoptive transfer of IL-2-activated human NK cells for the treatment of cancer. Scores of articles have shown that NK cells may play a role in protection against transplantable tumors. Depending on the experimental model, this has been done by depletion of NK cells before tumor transplantation to show more aggressive tumor growth or metastasis in the absence of NK cells. Another commonly used strategy is to administer cytokines known to booster NK cell effector functions, such as IL-2, IL-12, IL-15, and IFN/ (or IFN-inducing agents, and to demonstrate enhanced antitumor efficacy. In these experiments, cytokines have been given either systemically or introduced locally by transfection of transplantable tumors with the desired cytokine or chemokine gene in an expression vector. As a general rule in these types of experiments,) NK cells are more efficient at eliminating tumors that are MHC class I deficient or express low amounts of MHC class I; NK cells frequently are able to prevent outgrowth of small numbers of tumors, but this can be overwhelmed by using large tumor challenges; depending on the tumor, administration of factors that recruit or activate

132

Jun Wu and Lewis L. Lanier

NK cells may show therapeutic benefit; and NK cells do not have the capacity for ‘‘memory’’ and, therefore, on a secondary exposure to a tumor that has been eliminated by NK cells, there is no accelerated response or a larger pool of effector cells than on the first encounter. In the case of MHC class I– deficient tumors, for example, the well-studied RMA-S lymphoma, NK cells may be directly and exclusively responsible for tumor rejection (Karre et al., 1986), based on their ability to kill these tumors by a perforin-dependent cell-mediated cytolytic mechanism (van den Broek et al., 1995). However, in other circumstances NK cells may not be able to completely eliminate a tumor alone, but they may cooperate with T cells to achieve this end. An example of this cooperation was provided by studies of the mouse B16 melanoma transduced with the CD80 (B7-1) costimulatory molecule. In these experiments, C57BL/6 mice eliminated the CD80+ B16 tumors, but not the parental tumor, and host protection required both CD8+ T cells and NK cells. Antibody depletion of either CD8+ T cells or NK cells abrogated the tumor rejection (Wu et al., 1995). There is an emerging concept that although NK cells alone may be insufficient to eliminate tumors, they may kill some tumor cells, providing apoptotic tumor bodies to dendritic cells for priming of a subsequent T-cell response. In addition, IFN- or other cytokines produced by NK cells in response to tumors may activate local macrophages and dendritic cells, which reciprocally stimulate NK cells by cell–cell contact or soluble factors (Ferlazzo et al., 2002; Fernandez et al., 1999; Gerosa et al., 2002; Pawlowska et al., 2001; Peron et al., 1998; Piccioli et al., 2002). This is a rapidly developing area of investigation and indicates that NK cells may be more critical in facilitating an adaptive immune response than previously appreciated. Although IL-2 activated NK cells have the ability to kill many types of tumors in vitro, the more relevant question is, Does this happen in vivo? A role for NK cells in protection against tumor cells in vivo in experimental animal models has often relied on experiments in which participation by B and T cells has been excluded by the use of scid mice or RAG1 / or RAG2 / immunodeficient mice and by the depletion of NK cells by using anti-NK1.1 mAb or an antisera against asialoGM-1. Treatment with antiNK1.1 efficiently depletes NK cells, but only in certain mouse stains such as C57BL/6; however, NK1.1 is also present on a small subset of T cells, called ‘‘NK T cells’’ that may also mediate antitumor activity and functionally cooperate with NK cells in this task (reviewed in Smyth et al., 2002a). Depletion with anti-asialoGM-1 antisera is less satisfactory than anti-NK1.1 mAb in that CTL and some macrophages express this glycolipid. Unfortunately, there are no experimental animals that completely lack NK cells but retain normal B and T lymphocytes. NOD and beige mice are frequently used because they are ‘‘NK defective.’’ However, both these strains of mice have normal numbers of NK cells but show abnormalities in their cytolytic

NK Cells and Cancer

133

functions. In the case of beige mice, the problem lies with a defect in cytoplasmic granule formation that also affects CTL and macrophages. Nonetheless, by the combined use of T- and B-cell–deficient mice and depletion of NK cells using anti-NK1.1, it has been possible to confirm that NK cells are indeed responsible for eliminating transplantable tumor cells and preventing tumor metastasis in vivo (as reviewed recently in Smyth, et al., 2002b). In addition to the examples discussed previously, involving lymphoma and melanoma model systems, NK cells have been implicated in protection against tumors that invade the liver. This is particularly relevant given the abundance of NK cells resident in the normal liver. For example, a recent study showed that overproduction of IL-12 in a murine model of hepatic metastatic breast cancer resulted in increased long-term survival and significant tumor regression (Divino et al., 2000). Importantly, antibody depletion of NK cells, but not CD4+ or CD8+ T cells, abrogated such an effect. In addition, another study showed that mouse liver NK cells have the unique property of expressing a transmembrane protein called TRAIL, whereas T cells do not (Takeda et al., 2001). TRAIL has been shown to induce apoptosis in a variety of tumor cells, but not normal cells. In a murine model, constitutive expression of TRAIL on liver NK cells was shown to be crucial for controlling liver metastasis from tumors expressing a TRAIL receptor (Cretney et al., 2002; Takeda et al., 2001). TRAIL is constitutively expressed by immature human NK cells (Zamai et al., 1998) and is induced on most activated NK cells (Kashii et al., 1999). Interestingly, in humans NK cells comprise almost half of the resident lymphocyte population in the liver, and they have a distinct phenotype compared with peripheral blood NK cells (Hata et al., 1990). With respect to the role of NK cells in surveillance against de novo tumor formation, evidence is accumulating, but the jury is still out. The lack of an appropriate animal model that lacks NK cells, but has normal B and T cells, has hampered directly testing this hypothesis. Although specific depletion of NK cells using anti-NK1.1 mAb is adequate to prove a role for NK cells in immunity against transplantable tumors, in general it is not feasible to sustain depletion for the length of time necessary for the genesis of spontaneous tumors. Evidence that NK cell surveillance may participate in the prevention of cancer comes from the elegant experiments of Schreiber and colleagues, who examined spontaneous and chemical carcinogen-induced tumor formation in RAG2 / mice and RAG2 / mice that are unable to respond to interferons by a genetic deletion of the STAT1 signaling protein (Shankaran et al., 2001). Although RAG2 / mice developed more tumors than wildtype mice, implicating adaptive immunity, the informative result was obtained by comparing tumor occurrence in RAG2 / mice and the double-deficient RAG2 /2 STAT1 / mice. Here it was clear that the double-deficient mice developed more adenocarcinomas, showing that a component of the innate immune defense system was operational (but it still

134

Jun Wu and Lewis L. Lanier

needs to be established that NK cells are responsible). In addition, Smyth and colleagues have demonstrated that mice depleted with anti-asialoGM-1 antisera or anti-NK1.1 mAb showed a higher incidence of methylcholanthreneinduced sarcomas than untreated mice (Smyth et al., 2001b). Studies of perforin-deficient mice and IFN--deficient mice also have shown increased incidence of lymphomas and adenocarcinomas (Smyth et al., 2000b; Street et al., 2002) but have left open the question of whether NK cells, CTL, or both are involved in prevention of the primary tumors. Clear evidence exists that NK cells provide a barrier to the transplantation of these tumors arising in immunocompromised hosts into wild-type animals. Further studies introducing the disrupted perforin and IFN- genes onto a RAG / background will be needed to discriminate between effects mediated by the innate and adaptive immune responses. In addition, examining spontaneous and carcinogen-induced cancers in RAG / mice with disrupted IL-15 or IL-15 receptor- genes, which will eliminate NK cells, should prove informative. The close lineage relationship between NK cells and T cells has made the production of mice lacking NK cell, but not T cells, a challenge. In humans, activated human NK cells possess the capacity to secrete cytokines and kill tumor cells, including autologous tumors. However, a potential role for NK cells in human cancer in vivo is based on correlations. The most compelling evidence implicating NK cells in protection against tumors comes from the studies of Velardi and colleagues (Ruggeri et al., 2002). Their findings clearly indicate that allogeneic NK cells derived from the hematopoietic stem cell donor are able to mediate a graft-versus-leukemia effect in the recipient to prevent relapse of myeloid leukemia. Ex vivo IL-2 activated peripheral blood lymphocytes (also known as ‘LAK’ cells), together with high doses of IL-2, have been used therapeutically (Mule et al., 1984; Rosenberg et al., 1987), particularly in patients with melanoma and renal cell carcinoma. Despite the modest antitumor activity observed in these trials, the overall results were unsatisfactory because of IL-2-associated general toxicity. Most of the clinical trials that have been performed with adoptively transferred activated NK cells have been done in terminal patients with extensive tumor burdens. Therefore, therapeutic manipulation of NK cells for cancer treatment may be more realistic in an adjuvant setting with minimal residual disease.

B. NK Cells in Viral Immunity—A Link to Their Role in Tumor Immunity Although NK cells can recognize and kill tumors, this may represent a process that evolved primarily to provide antiviral immunity. Therefore, studies to understand how NK cells recognize and eliminate virus-infected

NK Cells and Cancer

135

cells likely will be directly applicable to cancer. The importance of NK cells in immunity against viruses is well documented in humans and experimental animals. For example, a human with a rare immunodeficiency syndrome having a complete absence of NK cells, but normal B and T lymphocytes, suffered from severe Herpesvirus infections and died at an early age from an undetermined viral infection (Biron et al., 1989). Another patient preferentially lacking NK cells developed vulvar and cervical carcinoma at an early age, possibly resulting from human papilloma virus infection (Ballas et al., 1990). Similarly, NK cells have been implicated directly in host resistance to mouse cytomegalovirus (MCMV). Antibody depletion of NK cells resulted in elevated levels of MCMV viral titer in susceptible mouse strains, whereas adoptive transfer of NK cells provided protection against MCMV infection (Bukowski et al., 1985). Certain strains of mice are resistant to MCMV (Scalzo et al., 1992) because they possess an activating NK cell receptor, called Ly49H (Daniels et al., 2001a, 2001b; Lee et al., 2001), that directly recognizes a glycoprotein encoded by MCMV (Arase et al., 2002; Smith et al., 2002). Activation of the NK cells in these mice provides host cell protection that is conferred by the killing of MCMV-infected cells and the production of IFN- (Tay and Welsh, 1997). Similar to certain tumors, some viruses including CMV downregulate MHC class I molecules on the infected cells, leading to impairment of CD8+ CTL effector function (reviewed in Tortorella et al., 2000). In principle, the virus-infected cells that downregulated expression of MHC class I may be more easily killed by NK cells because the inhibitory MHC receptors on NK cells were no longer engaged (reviewed in Cerwenka and Lanier, 2001). Another activating NK receptor, NKp46, which was initially discovered because of its role in NK cell killing of human tumors (Pessino et al., 1998), has claimed to recognize and kill influenza virus–infected cells (Mandelboim et al., 2001). Recent experiments have also implicated NK cells in Epstein-Barr virus (EBV) infection. CD244 is an activating NK receptor that signals by associating with a cytoplasmic adapter molecule (Tangye et al., 1999) called the signaling lymphocyte activation moleculeassociated protein (SAP; Sayos et al., 1998). Patients who have X-linked lymphoproliferative syndrome were found to have a defective SAP (Nichols et al., 1998; Sayos et al., 1998), and the NK cells from these patients could not be activated through CD244 (Benoit et al., 2000; Nakajima et al., 2000; Parolini et al., 2000; Tangye et al., 2000). Intriguingly, these patients are highly susceptible to EBV infection. They often develop fulminating and fatal infectious mononucleosis or B lymphomas, indicating that NK cells may play a role in controlling EBV infection. NK cells have also been implicated in immunity against another Herpesvirus—human herpesvirus 8 (HHV8)—the virus causing Kaposi’s sarcoma. HHV8 encodes viral proteins that block expression of MHC class I, but that also downregulate other cell

136

Jun Wu and Lewis L. Lanier

surface proteins, such as CD54 (ICAM-1), that potentiate NK cell recognition and activation (Coscoy and Ganem, 2000, 2001; Ishido et al., 2000). A recent study has reported that NK cell function in HIV-infected patients inversely correlates with HHV8 viremia and that response to highly active antiretroviral therapy resulted in enhanced NK cell activity and control of HHV8 (Sirianni et al., 2002). Collectively, these studies indicate that NK cells may actively participate in immunity against viruses that cause cancer, such as EBV and HHV8, and that the NK cell immune strategies and receptors that have evolved to primarily deal with virus infections are applicable to the control of transformed cells.

III. NK CELL EFFECTOR FUNCTIONS A. Perforin/Granzyme-Mediated Cytotoxicity Perforin is stored in cytoplasmic granules, and on activation, NK and T cells secrete these cytolytic granules (reviewed in Podack et al., 1991; Trapani and Smyth, 2002). Perforin monomers then insert into the plasma membrane of target cells and polymerize into pore-forming aggregates (Liu et al., 1995), which leads to osmotic lysis, granzyme entry, and killing of the target cells. Perforin and granzyme-mediated apoptosis is the principle pathway that NK cells use to kill tumors and virus-infected cells. Studies in perforin-deficient mice have demonstrated that perforin is crucial for NK cell–mediated cytotoxicity (Kagi et al., 1994; Lacorazza et al., 2002). In addition, the perforin-deficient mice have been observed to display increased susceptibility to many syngeneic, MHC class I–defective, chemical or virus-induced tumors (van den Broek et al., 1996, 1995). These mice were 10–100-fold less proficient than wild-type mice in preventing the metastasis of tumor cells to the lung (Smyth et al., 1999). Numerous studies have demonstrated that perforin-mediated NK cytotoxicity is important in immunosurveillance against spontaneous tumors (Smyth et al., 2000b; van den Broek et al., 1996). Perforin expression is constitutive in NK cells and is induced by IL-2 in T cells and NK cells (Podack et al., 1991). NK cells have also shown to be activated by IL-12 to kill various tumor cells in a perforin-dependent manner (Kodama et al., 1999). These studies provide the rationale for therapeutic applications of these cytokines in cancer treatment. Granzyme A and B are serine proteases present in the granules of NK cells and have been shown to induce target cell DNA fragmentation and apoptosis in vitro (reviewed in Trapani et al., 2000). Surprisingly, studies using mice deficient in either granzyme A, granzyme B, or both have shown that only perforin, not granzymes, is critical for tumor rejection and surveillance

NK Cells and Cancer

137

by CTL and NK cells (Davis et al., 2001). In contrast, a more recent study analyzing the potential of the granzyme-deficient and the perforin-deficient mice to control growth of NK-sensitive, syngeneic MHC class I–defective tumor cells in vivo has reported that a concerted action of perforin and the two granzymes is required for optimal NK cell–mediated tumor rejection in vivo (Pardo et al., 2002). Despite the controversial results, perforin is believed to predominantly mediate the cytotoxic function of NK cells in most circumstances.

B. TRAIL/Fas Ligand-Mediated Apoptosis TRAIL and FasL are members of the TNF family of cytokines that are expressed by effector lymphocytes and that serve as important mediators of apoptosis during immune responses. Soluble recombinant TRAIL can induce apoptosis of many transformed cells, but not normal cells. Strikingly, administration of soluble recombinant TRAIL in lab animals resulted in marked tumor regression without significant toxicity (Ashkenazi et al., 1999; Walczak et al., 1999; Wiley et al., 1995). In mice, a subset of liver NK cells constitutively express TRAIL, and they have been shown to prevent tumor metastasis in the liver (Cretney et al., 2002; Takeda et al., 2001). In perforin-deficient mice, antibody neutralization of TRAIL resulted in the enhancement of tumor metastasis (Smyth et al., 2001), indicating that TRAIL is an alternative cytotoxic pathway by which NK cells eliminate certain tumors in vivo. TRAIL is expressed constitutively on some NK cells and is highly induced on most NK cells after stimulation with IL-2, IFN, or IL-15 (Kashii et al., 1999; Kayagaki et al., 1999; Zamai et al., 1998). In humans, it remains to be determined whether TRAIL plays a role in NK-mediated antitumor activity. Certain human tumor cell lines and primary human epithelial tumors are susceptible to TRAIL-induced apoptosis ex vivo. A preliminary study has shown that TRAIL expression was upregulated on CD56þ NK cells in hepatocarcinoma patients responding to chemotherapy (Smyth et al., 2003). It has also been reported that TRAIL expression is lower in T cells from patients with advanced melanoma, and TRAIL-positive immune cells could be found in regressing primary melanomas and in metastases of patients responding to IFN therapy (Hersey and Zhang, 2001). However, given the significant heterogeneity in TRAIL sensitivity in various human cancers, one might expect that tumors may downregulate TRAIL receptor expression to escape from immune surveillance. The observation that recombinant soluble TRAIL kills normal hepatocytes also raises doubts about TRAIL as a promising anticancer agent (Lawrence et al., 2001).

138

Jun Wu and Lewis L. Lanier

C. Interferon-g-Mediated Effector Functions On activation, NK cells produce and secrete a number of cytokines, including IFN-, TNF-, and GM-CSF. It has been demonstrated that NK cells exert their antitumor activities not only by direct cytotoxicity but also by producing IFN- (reviewed in Trinchieri, 1995). Perforin-mediated cytotoxicity and IFN- function independently, and they appear to account for most of the NK-mediated protection from spontaneous tumor development and tumor metastasis in certain murine models (Street et al., 2001). IFN- is a pleiotropic cytokine that can act on both tumor cells and host immunity (reviewed in Boehm et al., 1997), and it has been shown to stimulate antigen presentation by upregulating MHC class I and II molecules on many cell types. IFN- directly inhibits proliferation of some tumor cells in vitro and indirectly inhibits tumor growth in vivo by inducing other factors, such as IP10, that suppress tumor angiogenesis (Angiolillo et al., 1995). IFN- also enhances NK cell cytotoxicity by upregulating the expression of adhesion molecules and by increasing the sensitivity of tumor cells to perforinand FasL-mediated cytotoxicity. In addition, the constitutive expression of TRAIL on the surface of murine liver NK cells is dependent on IFN-, which contributes to the natural antimetastatic activity of liver NK cells (Takeda et al., 2001). The IFN--induced TRAIL expression on NK cells has been implicated in the IL-12-mediated antimetastatic effects (Smyth et al., 2001a).

IV. RECEPTORS TURNING NK CELLS ‘‘ON’’ AND ‘‘OFF’’ Unlike T and B cells, NK cells do not require the specialized gene arrangement machinery to assemble their receptor genes. However, they certainly are capable of discriminating between normal and abnormal cells. It has been well documented that NK cells preferentially kill certain cells that lack MHC class I expression (Ljunggren and Karre, 1985). NK cells are also able to reject MHC-incompatible bone marrow grafts, in particular when the donor graft lacks MHC molecules of the host (reviewed in Yu et al., 1992). These observations led to the ‘‘missing self’’ hypothesis, stating that NK cells ignore potential target cells that express normal levels of autologous MHC class I molecules, whereas they attack cells that do not (Karre et al., 1986). This hypothesis is supported by numerous in vivo studies and provides a rather satisfying rationale for NK cell–mediated killing of transformed and virus-infected cells, as they often downregulate or lose MHC class I molecules. The molecular mechanism for the missing self

NK Cells and Cancer

139

hypothesis was uncovered when several MHC class I–specific receptors that turn NK cells ‘‘off’’ were discovered. However, the missing self hypothesis certainly cannot account for all NK cell–mediated cytotoxicity. For example, NK cells do not kill human erythrocytes, which do not express MHC class I, and normal resting cells with low levels of MHC class I are quite resistant to NK killing. Moreover, NK cells attack even MHC class I–positive tumor cells under certain conditions (Leiden et al., 1989; Litwin et al., 1993; Nishimura et al., 1988; Pena et al., 1990). A search for mechanisms to turn NK cells ‘‘on’’ led to the identification of a number of activating NK receptors, some are specific for ligands that are upregulated in tumor cells and stressed cells, others recognize ligands on normal cells. It is now believed that NK cell effector functions are regulated by a balance of inhibitory receptors specific for MHC class I and activating receptors specific for a diverse array of ligands.

A. Turning NK Cells Off—Inhibitory MHC Class I Receptors The first inhibitory MHC class I receptors discovered were the Ly49 receptors in rodents (Karlhofer et al., 1992; Yokoyama and Seaman, 1993). The mouse Ly49 receptors are encoded by a small family of genes and recognize polymorphic epitopes on classic H-2D and H-2K MHC class I molecules. Ly49 genes themselves are polymorphic and are regulated by alternative splicing and individual locus control, allowing for monoallelic expression of the Ly49 receptors (Held et al., 1995) on overlapping subsets of NK cells and memory (usually CD8þ) T cells. Ly49 receptors are type II transmembrane glycoproteins of the C-type lectin-like family. Except Ly49D and Ly49H, which are activating receptors, most Ly49 proteins in C57BL/6 mice contain a cytoplasmic signaling motif—an immunoreceptor tyrosine-based inhibitory motif [ITIM, (I/V/L/S)xYxx(L/I/V)] where x denotes any amino acid—allowing for inhibition of NK function (reviewed in Long et al., 1997; Ravetch and Lanier, 2000). Engagement of an inhibitory Ly49 receptor with its specific MHC class I ligand results in tyrosine phosphorylation of the ITIM and subsequent recruitment of the downstream SHP-1 or SHP-2 tyrosine phosphatases, which ultimately leads to inhibition of NK effector functions (Nakamura et al., 1997). The second family of inhibitory receptors identified was the killer cell immunoglobulin (Ig)-like receptors (KIR, also collectively called CD158; reviewed in Lanier, 1998b; Long, 1999; Vilches and Parham, 2002). Functional Ly49 genes are not present in humans, and the KIR genes apparently have taken over this biological function in primates. Similar to Ly49, the

140

Jun Wu and Lewis L. Lanier

KIR family comprises about 15 functional genes that are highly polymorphic. These receptors recognize polymorphic epitopes on human MHC class I molecules; namely human leukocyte antigen (HLA)-A, HLA-B, and HLA-C. KIR are expressed on overlapping subsets of NK cells and memory T cells. Only about half of the KIR molecules are inhibitory, and most possess two ITIM in their cytoplasmic domains. It has been shown that, on ligand binding, an inhibitory KIR predominantly uses SHP-1 to downregulate NK cell–mediated cytotoxicity and cytokine secretion (Burshtyn et al., 1996; Campbell et al., 1996; Fry et al., 1996; also, reviewed in Lanier, 2003; Ravetch and Lanier, 2000). Blocking inhibitory KIR permits human NK cells to kill tumors that may be otherwise protected by expression of MHC class I. The third inhibitory receptor family is functional in both human and rodents, and it consists of disulfide-bonded heterodimers of CD94 and NKG2A (Brooks et al., 1997; Lazetic et al., 1996; Vance et al., 1998). This receptor complex is expressed on about half of NK cells and a subset of memory CD8+ T cells. CD94 has a short cytoplasmic domain that lacks any signaling function, whereas NKG2A contains two ITIM in its cytoplasmic domain, allowing it to recruit SHP-1 and SHP-2 to exert the inhibitory effect (Houchins et al., 1997; Le Drean et al., 1998). The CD94/NKG2A heterodimer recognizes HLA-E in humans and the orthologous Qa-1b in mice (Borrego et al., 1998; Braud et al., 1998; Lee et al., 1998; Vance et al., 1998). It is interesting to note that the peptide-binding groove of HLA-E and Qa-1b is often occupied by nine–amino acid peptides generated from the leader segments of other MHC class I molecules, such as HLA-A, HLA-B, or HLA-C in humans and H-2D and H-2K in mice (Aldrich et al., 1994; Braud et al., 1997). Moreover, the stable surface expression of HLA-E and Qa-1b depends on these peptides and the transporter associated with antigen processing (TAP)-1. This unique mechanism may allow NK cell and T cells to monitor the general expression of MHC class I molecules on cells and tissues (reviewed in Lanier, 1998a). Antibody blockade studies have shown that a disruption of the inhibitory Ly49 and their MHC class I ligands on tumor cells resulted in enhancement of the antitumor responses in vitro and in vivo. Blockade of inhibitory Ly49 receptors has been shown to promote rejection of MHC class I–bearing mouse tumors in vivo (Koh et al., 2001). In addition, transfection of a gene encoding a binding peptide (Qdm) for Qa-1b into the TAP-deficient RMA-S tumor cells resulted in surface expression of Qa-1b and inhibition of NKmediated rejection of the tumors in vivo (Jia et al., 2000). These data implicate these inhibitory receptors in protection of tumors in vivo and indicate that blockade of inhibitory KIR or NKG2A receptors in humans may provide a therapeutic benefit in cancer.

NK Cells and Cancer

141

B. Activating NK Receptors For many years, the nature of the activating receptors on NK cells remained elusive. It was thought that adhesion molecules, such as LFA-1, and costimulatory molecules, such as CD2, might be sufficient to deliver the activating signals in NK cells in the absence of MHC class I (Hersey and Bolhuis, 1987). However, the recent identification and characterization of relatively NK cell–specific activating receptors have provided a molecular mechanism by which NK cells recognize and kill potential targets. Many of the activating NK receptors have short cytoplasmic domains that lack intrinsic signaling function; rather, they interact with small transmembrane signaling adapter proteins, such as Fc"RI, CD3, DAP-12, or DAP10, to exert their activating functions (reviewed in Cerwenka and Lanier, 2001; Lanier, 2003). The activating receptors can be broadly divided into those associated with immunoreceptor tyrosine-based activating motif (Reth, 1989; ITAM, YxxL/I x6–8 YxxL/I where x denotes any amino acid with 6–8 amino acids between the two YxxL/I elements) containing transmembrane adaptors and those that are not.

1. ACTIVATING RECEPTORS USING ITAM FOR SIGNALING Within the Ly49, KIR, and NKG2 gene complexes, certain loci have been shown to encode receptors that activate NK cells (Houchins et al., 1997; Mason et al., 1996; Moretta et al., 1995). These activating NK receptors include Ly49D and H in C57BL/6 mice, KIR2DS and KIR3DS in humans, and NKG2C in humans and mice. Although some of these receptors, like their inhibitory counterparts, may bind with low affinity to MHC class I, the relevance of this in vivo has not been established. Similar to the antigen receptors on T and B cells, these ligand-binding receptors themselves do not possess any signaling capability to activate NK cells. Rather, they associate with DAP12, a small transmembrane signaling adaptor containing a cytoplasmic activating ITAM sequence (Lanier et al., 1998a, 1998b; Smith et al., 1998). On stimulation, the tyrosine residues within the ITAM become phosphorylated, which results in the recruitment and activation of tyrosine kinases ZAP-70 and Syk. Subsequently, this turns on a signaling cascade, leading to NK cell activation, with downstream activation of MAP kinases and the induction of gene transcription. Other activating NK receptors functioning through ITAM-bearing adapter proteins include CD16 (Lanier et al., 1989b) and the ‘‘natural cytotoxicity receptors’’ (NCR), which include NKp30, NKp44, and NKp46 (reviewed in Moretta et al., 2001). NKp44 (Vitale et al., 1998) couples to DAP12, whereas CD16 (Lanier et al., 1989b), NKp30 (Pende et al.,

142

Jun Wu and Lewis L. Lanier

1999), and NKp46 (Pessino et al., 1998) associate with CD3 and Fc"RI. CD16 is a low affinity Fc receptor for IgG and is responsible for antibodydependent cellular cytotoxicity (Lanier et al., 1983; Perussia et al., 1983). There is evidence that at least some of these ITAM-containing receptor complexes are involved in tumor elimination by NK cells. NK cells from mice deficient in DAP12, CD3, or Fc"RI showed selective impairment in their ability to kill certain tumors (Tomasello et al., 2000). Moreover, mAbs against NKp44, NKp46, or NKp30 partially diminish NK cell killing of several human tumor cell lines, and in combination they can completely inhibit killing of some, but not all, tumors by NK cells (Moretta et al., 2000; Pende et al., 1999; Sivori et al., 1997; Vitale et al., 1998). The tumorassociated ligands for these receptors are unknown, and their identification should help us to understand their potential role in NK cell-mediated tumor immunity. NKp44 and NKp46 have been shown to bind to the haemagglutinins of influenza virus (Mandelboim et al., 2001), but the specificity and physiological relevance of this have not yet been validated. There is a report that CD16 may bind to ligands on tumors other than IgG, but the nature of these putative ligands has not been identified (Mandelboim et al., 1999). In mice, the activating isoforms of the NKR-P1 receptor (e.g., NKR-P1C; i.e., NK1.1) associate with Fc"RI (Arase et al., 1997). Although ligands for NKR-P1C are not known, there is evidence that NKR-P1C may participate in recognition of certain tumors (Ryan et al., 1995).

2. CD244 (2B4) AND NTB-A The cytoplasmic domains of CD244 (Tangye et al., 1999) and NTB-A (Bottino et al., 2001) contain a signaling motif (TxYxxV/I, where x is any amino acid) that, under normal conditions, binds to the cytoplasmic SAP adapter protein (Sayos et al., 1998; also, reviewed in Veillette and Latour, 2003). The ligand for CD244 is CD48 (Brown et al., 1998; Latchman et al., 1998), a broadly distributed cell-surface glycoprotein. The ligand for NTB-A remains unknown. Triggering of CD244 (Garni-Wagner et al., 1993; Mathew et al., 1993; Valiante and Trinchieri, 1993) and NTB-A (Bottino et al., 2001) on NK cells results in cytotoxicity and IFN- secretion. The physiological importance of both receptors is indicated in patients with a severe inherited immunodeficiency disorder called X-liked lymphoproliferative disease (XLP), which is caused by loss-of-function mutations in the SAP gene (Nichols et al., 1998; Sayos et al., 1998; Veillette and Latour, 2003). The XLP patients are otherwise healthy until they encounter EBV. On infection, they often develop fatal fulminant infectious mononucleosis or EBV-induced B lymphomas. It appears that the SAP-binding is required for CD244 and NTB-A receptors to activate NK cells, as both receptors have been found to be nonfunctional (or possibly inhibitory),

NK Cells and Cancer

143

rather than activating, in the SAP-deficient NK cells from the patients (Benoit et al., 2000; Nakajima et al., 2000; Parolini et al., 2000; Tangye et al., 2000). There is evidence that CD244 may cooperate functionally with the ITAM-bearing receptors on NK cells (Sivori et al., 2000) and augment integrin-dependent NK cell activation (Barber and Long, 2003) through its association with the cytoskeleton (Watzl and Long, 2003).

3. NKG2D AND ITS LIGANDS NKG2D is a C-type lectin-like receptor (Houchins et al., 1991) expressed on NK cells, þT cells, CD8þ þT cells, and activated macrophages (Baner et al., 1999; Jamieson et al., 2002). NKG2D associates with another small transmembrane signaling adaptor, DAP10, encoded by a gene that lies next to the DAP12 gene on human chromosome 19 (Wu et al., 1999). Cellsurface expression of NKG2D and DAP10 are dependent on each other, and triggering of the NKG2D receptor on NK cells results in cytotoxicity and augments cytokine secretion (Bauer et al., 1999; Wu et al., 1999). Unlike DAP12, which has an ITAM, DAP10 contains a cytoplasmic YxxM motif that recruits the p85 subunit of phosphatidylinositol 3-kinase (PI3-K) following receptor stimulation (Chang et al., 1999; Wu et al., 1999). Similar YxxM motifs are also present in the cytoplasmic tail of the T-cell costimulatory molecule CD28 (Pages et al., 1994; Prasad et al., 1994; Ward et al., 1993), suggesting a role of NKG2D in T-cell costimulation. The observation that NKG2D can substitute for CD28 in the activation of CMV-specific human þ T cells supports such a notion (Groh et al., 2001). Intriguingly, recently it has been observed that in mice deficient in DAP10, NKG2D was expressed and still functional in activated NK cells, but not T cells (Gilfillan et al., 2002). In contrast, NKG2D in DAP12-deficient mice appeared to be functional in T cells, but not fully active in NK cells (Diefenbach et al., 2002). This puzzle was solved when two alternatively spliced isoforms of NKG2D were discovered in mouse; NKG2D-L with a longer cytoplasmic tail and NKG2D-S with a shorter one (Diefenbach et al., 2002). NKG2D-L associates exclusively with DAP10, whereas mouse NKG2D-S is able to associate with either DAP10 or DAP12 (Diefenbach et al., 2002; Gilfillan et al., 2002). Resting NK cells only express NKG2DL, whereas NKG2D-S can be induced by IL-2 in vitro or by a type I interferon inducer in vivo (Diefenbach et al., 2002). In NK cells, DAP12 is required for NKG2D-dependent stimulation of cytokine production; however, the NKG2D-L/DAP10 complex is sufficient to initiate cellmediated cytotoxicity (Zompi et al., 2003). As yet, only a NKG2D-L protein associated with DAP10 has been identified in humans (Billadeau et al., 2003). The physiological role of such complex regulation of NKG2D and its promiscuous pairing with either DAP10 or DAP12 is yet to be determined.

144

Jun Wu and Lewis L. Lanier

NKG2D binds to a diverse array of ligands on cells that have undergone infection, transformation, or stress. In mouse, NKG2D binds to several cellular ligands, including the retinoic acid early-inducible-1 (RAE-1) proteins (Cerwenka et al., 2000; Diefenbach et al., 2000b), MULT-1 (Carayannopoulos et al., 2002; Diefenbach et al., 2003), and minor histocompatibility antigen H60 (Cerwenka et al., 2000; Diefenbach et al., 2000a). At least five RAE-1 genes have been identified in the mouse genome of different mouse strains. Similar to H60, RAE-1 genes are polymorphic (reviewed in Cerwenka and Lanier, 2003). Interestingly, H60 also serves as a precursor polypeptide for a peptide loaded into H-2Kb and is a potent minor transplantation alloantigen (Cerwenka et al., 2002; Choi et al., 2001; Malarkannan et al., 1998). RAE-1 proteins are not expressed by most normal healthy cells in adult mice, but are upregulated on most tumor cells (Cerwenka et al., 2001; Diefenbach et al., 2000b) and are induced by viral infection (Lodoen et al., 2003). By contrast, MULT-1 transcripts appear more broadly distributed, but whether MULT-1 proteins are expressed in these normal cells has not been evaluated (Carayannopoulos et al., 2002; Diefenbach et al., 2003). In humans, NKG2D binds with high affinity to two polymorphic MHC class I–related molecules, MICA and MICB, which are linked genetically to HLA-B (Bauer et al., 1999). MICA and MICB genes do not exist in mice. MICA/B proteins contain 1, 2, and 3 domains with homology to MHC class I, but they do not associate with 2-microglobulin or peptide (Li et al., 1999, 2001). MICA and MICB were originally found to be expressed in intestinal epithelia and were shown to be heat inducible, but later were discovered to be overexpressed in various primary tumors of epithelial origin (Groh et al., 1996, 1998, 1999). The human orthologs of the mouse RAE-1 genes are called UL16-binding proteins (ULBP). The first member of this family (ULBP-1) was discovered because of its ability to bind the UL16 protein encoded by human cytomegalovirus (Cosman et al., 2001). There are at least five functional ULBP genes in the family, generating many alternatively spliced isoforms and several pseudogenes (Radosavljevic and Bahram, 2003; Radosavljevic et al., 2002; Wu, 2003). ULBP proteins are expressed in either transmembrane, glycosylphosphatidylinositol (GPI)-linked, or soluble forms (Cosman et al., 1997; Wu, unpublished data). ULBP proteins have 1 and 2 domains with homology to MHC class I, but similar to MICA/B, they do not associate with 2-microglobulin or peptides (Radaev et al., 2001, 2002). The human ULBP proteins specifically bind to NKG2D (Cosman et al., 1997; Jan Chalupny et al., 2003). Similar to MICA and MICB, most ULBP proteins are absent or expressed at low levels in normal tissues, but are frequently overexpressed in many primary tumors, including colon, lung, stomach,

NK Cells and Cancer

145

and breast carcinomas (Wu, unpublished data). ULBP transcripts have been detected in many tissues (Cosman et al., 2001), indicating posttranslational control of expression. More in-depth studies on these NKG2D ligands and their association with primary tumors will help us to develop potential therapeutic vaccines against cancers.

V. NK CELLS IN TUMOR IMMUNOSURVEILLANCE Exposure to environmental stresses, such as pyrogens and inflammatory cytokines, may predispose cells to neoplastic transformation. Therefore, it is necessary for higher organisms to have sophisticated mechanisms to either repair such mutations or to recognize and eliminate the damaged cells that may result in tumor initiation. The concept of cancer immunosurveillance proposed by MacFarland Burnet and Lewis Thomas in the 1950s predicts that an unmanipulated immune system is capable of recognizing and eliminating nascent transformed cells. This hypothesis has remained controversial for many years and was out of favor until recently, largely on the basis of the observation that nude mice (lacking T cells) do not have a higher incidence of many cancers compared with wild-type mice (Lanier, 2001b). However, this T-cell-centric interpretation ignored the fact that nude mice have an intact innate immune system and functional B cells. Recent experiments in mice deficient in IFN-, perforin, or RAG-2 have provided strong support for the important roles of CTL and NK cells in cancer immunosurveillance (Engel et al., 1997; Kaplan et al., 1998; Shankaran et al., 2001; Smyth et al., 2000a,b, 2001b; van den Broek et al., 1996). Antibody depletion studies also revealed that NK cells are important in preventing spontaneous tumors induced by the chemical carcinogen methylcholanthrene (MCA; Smyth et al., 2001b). The recent discovery of the activating NK receptor NKG2D and its ligands on primary tumors has further provided a potential mechanism by which NK cells participate in tumor surveillance (Diefenbach and Raulet, 2001; Lanier, 2001a). It has been shown that transfection of the mouse NKG2D ligand RAE-1 or H60 into MHC class I–positive tumors resulted in NKG2D-mediated NK cell killing and macrophage activation in vitro (Cerwenka et al., 2001; Diefenbach et al., 2001). In addition, ectopic expression of RAE-1 or H60 in certain MHC class I–positive tumors rendered them susceptible to rejection by NK cells and CD8þ T cell in syngeneic mice, demonstrating that the NKG2D-mediated activation is able to overcome the MHC class I–mediated inhibitory signaling in responding NK cells. Therefore, it has been postulated that selective upregulation of the

146

Jun Wu and Lewis L. Lanier

NKG2D ligands on ‘‘stressed’’ cells may allow NK cells and T cells to shift from immune tolerance to activation, resulting in elimination of precancerous cells. Supporting this hypothesis, expression of RAE-1 was found to be induced by carcinogens in skin cells, and skin-associated -TcRþT cells killed these cancer cells in vitro via NKG2D (Girardi et al., 2001). Interestingly, in some cases mice that rejected RAE-1 or H60-expressing tumors were immune to rechallenge with the same tumors not expressing these NKG2D ligands (Diefenbach et al., 2001). Immunity to rechallenge depended solely on CD8þ T cells. Given the fact the cytoplasmic tails of NKG2D and CD28 both contain the consensus binding site for PI3K, these data indicate that NKG2D not only plays an important role in NK activation during innate immunity but also contributes to the subsequent adaptive immune responses against tumors. Although NKG2D is likely to participate in tumor immunosurveillance, it certainly cannot account for all innate immunity against primary tumors. Its ligands, MICA/B and ULBP, were upregulated only in some human tumor tissues (Groh et al., 1999; Wu, unpublished data). NK cells are also able to kill certain tumors lacking ligands for NKG2D (Pende et al., 2001), implicating additional ligands on tumor cells for other activating receptors. Further in vivo studies on NKG2D and stress-induced signals are required to ascertain the physiological role of NKG2D in immunosurveillance against cancer.

VI. TUMOR ESCAPE MECHANISMS The human NKG2D ligands, MICA/B and ULBP, and the mouse RAE1 ligands for NKG2D are frequently overexpressed by tumors. There is convincing evidence that these ligands can trigger NK cell attack, so how do these tumors avoid being eliminated by NK cells and still grow in the body? One explanation is that tumors may secrete or shed proteins that downmodulate NK and T cell function, thereby evading adequate immune responses. A recent study by Groh et al. (2002) has provided strong evidence to support such a mechanism. They found that NKG2D expression is significantly reduced on tumor-infiltrating and peripheral blood T cells and NK cells in individuals with carcinomas secreting MICA. The systemic downmodulation of NKG2D is associated with circulating soluble MICA presumably released from tumors, thereby resulting in severe impairment of NK and T-cell responses against these tumors. The GPI-linked ULBP proteins may also be shed from tumor cells, and at least one soluble ULBP has also been identified that is capable of binding to NKG2D and downmodulating its expression (Wu, unpublished data). Similar studies in the mouse

NK Cells and Cancer

147

have shown that when NK cells encounter RAE-1-bearing tumors, this results in modulation of NKG2D on the surface of the NK cells and causes a functional impairment in their ability to kill and secrete cytokines (Ogasawara et al., 2003). Therefore, chronic exposure of NK cells to tumors bearing or shedding NKG2D ligands may lead to impairment of their immune functions. This may explain earlier observations of impaired NK cell function in cancer patients. Finally, triggering of the NKG2D by its ligands on tumors might not be sufficient for optimal NK activation in vivo (Pende et al., 2001). Other adhesion, activating, or costimulatory receptors may be required for tumor rejection. Certainly, NKG2D has been shown to work in synergy with other activating receptors on T cells and NK cells (Groh et al., 2001b; Jamieson et al., 2002; Lodoen et al., 2003; Pende et al., 2001; Wu et al., 2000). In this regard, Costello and colleagues (2002) have observed low levels of expression of the NCR molecules on NK cells in patients with acute myeloid leukemia, indicating the possibility that tumor-associated ligands may have caused modulation of these activating receptors. Another explanation to account for tumor persistence and growth in the host may be lack of stimulatory cytokines or adequate NK and dendritic cell interactions. Tumors often develop in an environment in which lack of cytokines, such as IL-2, IL-12, IL-15, IL-18, and type I IFN, may lead to suboptimal NK activation. In addition, the cross-talk between dendritic cells and resting NK cells that induces NK activation may not occur in these circumstances (Gerosa et al., 2002). Some tumors also produce cytokines that are immunosuppressive, such as transforming growth factor or IL-10, which could inhibit NK cell or T cell functions. Because of selective pressure, loss of ligands for activating NK cell receptors may protect tumors from attack. Consistent with this notion, quantitative analyses of ULBP on primary tumor specimens have revealed rather low levels of expression of these molecules when compared with other cellular proteins (Wu, unpublished data). In addition, tumors may become resistant to the effector molecules of the immune system. For example, many fresh tumor isolates from patients with melanoma were TRAIL-resistant, which appeared to be associated with low TRAIL-receptor expression (Nguyen et al., 2001). It has been proposed that certain tumors may express FasL to eliminate infiltrating NK cells and T cells (Griffith et al., 1995; Hahne et al., 1996; Niehans et al., 1997; O’Connell et al., 1996; Strand et al., 1996). However, whether expression of FasL on tumors confers protection is a controversial topic (Walker et al., 1998). More in-depth knowledge of cell transformation and immune surveillance mechanisms will be required to understand the complex nature of the relationship between the immune system and tumors.

148

Jun Wu and Lewis L. Lanier

VII. CONCLUSION NK cells develop from bone marrow stem cells and are important cells in the innate immune system. They kill tumors predominantly through perforin-mediated pathways and by the secretion of cytokines, such as IFN-. A delicate balance of activating and inhibitory receptors on the cell surface of NK cells regulates their effector functions. Recent progress in understanding the structure and function of these receptors and their ligands has provided insights into NK-mediated tumor rejection, tumor immunosurveillance, tumor escape strategies, and potential therapeutic approaches. For example, our knowledge of the MHC class I inhibitory receptors may allow the design clinical trials using KIR mismatched NK cells or neutralizing antibodies against the inhibitory receptors to treat cancer patients. The tumor-associated NKG2D ligands may serve as useful diagnostic markers and provide ideal targets for antibody therapeutics and cancer vaccines.

ACKNOWLEDGMENTS JW is an investigator at Chinese National Human Genome Center and is supported by Chinese National ‘‘863’’ grant 2002AA214111 and Shanghai Municipal Government S&T Commission grants 024319108, 02SYC007, and PKJ2002-11. LLL is an American Cancer Society Research Professor and is supported by NIH grants CA89189, CA89294, and CA095137. We thank Mark Smyth for helpful discussions.

REFERENCES Aldrich, C. J., DeCloux, A., Woods, A. S., Cotter, R. J., Soloski, M. J., and Forman, J. (1994). Cell 79, 649–58. Algarra, I., Cabrera, T., and Garrido, F. (2000). Hum. Immunol. 61, 65–73. Angiolillo, A. L., Sgadari, C., Taub, D. D., Liao, F., Farber, J. M., Maheshwari, S., Kleinman, H. K., Reaman, G. H., and Tosato, G. (1995). J. Exp. Med. 182, 155–162. Arase, H., Mocarski, E. S., Campbell, A. E., Hill, A. B., and Lanier, L. L. (2002). Science 296, 1323–1326. Arase, H., Saito, T., Phillips, J. H., and Lanier, L. L. (2001). J. Immunol. 167, 1141–1144. Arase, N., Arase, H., Park, S. Y., Ohno, H., Ra, C., and Saito, T. (1997). J. Exp. Med. 186, 1957–1963. Ashkenazi, A., Pai, R. C., Fong, S., Leung, S., Lawrence, D. A., Marsters, S. A., Blackie, C., Chang, L., McMurtrey, A. E., Hebert, A., DeForge, L., Koumenis, I. L., Lewis, D., Harris, L.,

NK Cells and Cancer

149

Bussiere, J., Koeppen, H., Shahrokh, Z., and Schwall, R. H. (1999). J. Clin. Invest. 104, 155–162. Ballas, Z. K., Turner, J. M., Turner, D. A., Goetzman, E. A., and Kemp, J. D. (1990). J. Allergy Clin. Immunol. 85, 453–459. Barber, D. F., and Long, E. O. (2003). J. Immunol. 170, 294–299. Bauer, S., Groh, V., Wu, J., Steinle, A., Phillips, J. H., Lanier, L. L., and Spies, T. (1999). Science 285, 727–9. Benoit, L., Wang, X., Pabst, H. F., Dutz, J., and Tan, R. (2000). J. Immunol. 165, 3549–3553. Billadeau, D. D., Upshaw, J. L., Schoon, R. A., Dick, C. J., and Leibson, P. J. (2003). Nat. Immunol. 4, 557–564. Biron, C. A., Byron, K. S., and Sullivan, J. L. (1989). N. Engl. J. Med. 320, 1731–1735. Biron, C. A., Nguyen, K. B., Pien, G. C., Cousens, L. P., and Salazar-Mather, T. P. (1999). Annu. Rev. Immunol. 17, 189–220. Boehm, U., Klamp, T., Groot, M., and Howard, J. C. (1997). Annu. Rev. Immunol. 15, 749–795. Borrego, F., Ulbrecht, M., Weiss, E. H., Coligan, J. E., and Brooks, A. G. (1998). J. Exp. Med. 187, 813–818. Bottino, C., Falco, M., Parolini, S., Marcenaro, E., Augugliaro, R., Sivori, S., Landi, E., Biassoni, R., Notarangelo, L. D., Moretta, L., and Moretta, A. (2001). J. Exp. Med. 194, 235–246. Braud, V., Jones, E. Y., and McMichael, A. (1997). Eur. J. Immunol. 27, 1164–1169. Braud, V. M., Allan, D. S., O’Callaghan, C. A., Soderstrom, K., D’Andrea, A., Ogg, G. S., Lazetic, S., Young, N. T., Bell, J. I., Phillips, J. H., Lanier, L. L., and McMichael, A. J. (1998). Nature 391, 795–799. Brooks, A. G., Posch, P. E., Scorzelli, C. J., Borrego, F., and Coligan, J. E. (1997). J. Exp. Med. 185, 795–800. Brown, M. H., Boles, K., van der Merwe, P. A., Kumar, V., Mathew, P. A., and Barclay, A. N. (1998). J. Exp. Med. 188, 2083–2090. Bukowski, J. F., Warner, J. F., Dennert, G., and Welsh, R. M. (1985). J. Exp. Med. 161, 40–52. Burshtyn, D. N., Scharenberg, A. M., Wagtmann, N., Rajagopalan, S., Berrada, K., Yi, T., Kinet, J.-P., and Long, E. O. (1996). Immunity 4, 77–85. Campbell, K. S., Dessing, M., Lopez-Botet, M., Cella, M., and Colonna, M. (1996). J. Exp. Med. 184, 93–100. Carayannopoulos, L. N., Naidenko, O. V., Fremont, D. H., and Yokoyama, W. M. (2002). J. Immunol. 169, 4079–4083. Cerwenka, A., Bakker, A. B., McClanahan, T., Wagner, J., Wu, J., Phillips, J. H., and Lanier, L. L. (2000). Immunity 12, 721–727. Cerwenka, A., Baron, J. L., and Lanier, L. L. (2001). Proc. Natl. Acad. Sci. USA 98, 11521–11526. Cerwenka, A., and Lanier, L. L. (2001). Nat. Rev. Immunol. 1, 41–49. Cerwenka, A., and Lanier, L. L. (2003). Tissue Antigens 61, 335–343. Cerwenka, A., O’Callaghan, C. A., Hamerman, J. A., Yadav, R., Ajayi, W., Roopenian, D. C., Joyce, S., and Lanier, L. L. (2002). J. Immunol. 168, 3131–3134. Chang, C., Dietrich, J., Harpur, A. G., Lindquist, J. A., Haude, A., Loke, Y. W., King, A., Colonna, M., Trowsdale, J., and Wilson, M. J. (1999). J. Immunol. 163, 4651–4654. Choi, E. Y., Yoshimura, Y., Christianson, G. J., Sproule, T. J., Malarkannan, S., Shastri, N., Joyce, S., and Roopenian, D. C. (2001). J. Immunol. 166, 4370–4379. Colucci, F., Caligiuri, M. A., and Di Santo, J. P. (2003). Nat. Rev. Immunol. 3, 413–425. Cooper, M. A., Fehniger, T. A., and Caligiuri, M. A. (2001). Trends Immunol. 22, 633–640. Coscoy, L., and Ganem, D. (2000). Proc. Natl. Acad. Sci. USA 97, 8051–8056. Coscoy, L., and Ganem, D. (2001). J. Clin. Invest. 107, 1599–1606.

150

Jun Wu and Lewis L. Lanier

Cosman, D., Fanger, N., Borges, L., Kubin, M., Chin, W., Peterson, L., and Hsu, M. L. (1997). Immunity 7, 273–282. Cosman, D., Mullberg, J., Sutherland, C. L., Chin, W., Armitage, R., Fanslow, W., Kubin, M., and Chalupny, N. J. (2001). Immunity 14, 123–133. Costello, R. T., Sivori, S., Marcenaro, E., Lafage-Pochitaloff, M., Mozziconacci, M. J., Reviron, D., Gastaut, J. A., Pende, D., Olive, D., and Moretta, A. (2002). Blood 99, 3661–3667. Cretney, E., Takeda, K., Yagita, H., Glaccum, M., Peschon, J. J., and Smyth, M. J. (2002). J. Immunol. 168, 1356–1361. Daniels, K. A., Devora, G., Lai, W. C., O’Donnell, C. L., Bennett, M., and Welsh, R. M. (2001a). J. Exp. Med. 194, 29–44. Daniels, K. A., Devora, G., Lai, W. C., O’Donnell, C. L., Bennett, M., and Welsh, R. M. (2001b). J. Exp. Med. 194, 29–44. Davis, J. E., Smyth, M. J., and Trapani, J. A. (2001). Eur. J. Immunol. 31, 39–47. Diefenbach, A., Hsia, J. K., Hsiung, M. Y., and Raulet, D. H. (2003). Eur. J. Immunol. 33, 381–391. Diefenbach, A., Jamieson, A. M., Liu, S. D., Shastri, N., and Raulet, D. H. (2000a). Nat. Immunol. 1, 119–126. Diefenbach, A., Jamieson, A. M., Liu, S. D., Shastri, N., and Raulet, D. H. (2000b). Nat. Immunol. 1, 119–126. Diefenbach, A., Jensen, E. R., Jamieson, A. M., and Raulet, D. H. (2001). Nature 413, 165–171. Diefenbach, A., and Raulet, D. H. (2001). Immunol. Rev. 181, 170–184. Diefenbach, A., Tomasello, E., Lucas, M., Jamieson, A. M., Hsia, J. K., Vivier, E., and Raulet, D. H. (2002). Nat. Immunol. 3, 1142–1149. Divino, C. M., Chen, S. H., Yang, W., Thung, S., Brower, S. T., and Woo, S. L. (2000). Breast Cancer Res. Treat. 60, 129–134. Engel, A. M., Svane, I. M., Rygaard, J., and Werdelin, O. (1997). Scand. J. Immunol. 45, 463–470. Ferlazzo, G., Tsang, M. L., Moretta, L., Melioli, G., Steinman, R. M., and Munz, C. (2002). J. Exp. Med. 195, 343–351. Fernandez, N. C., Lozier, A., Flament, C., Ricciardi-Castagnoli, P., Bellet, D., Suter, M., Perricaudet, M., Tursz, T., Maraskovsky, E., and Zitvogel, L. (1999). Nat. Med. 5, 405–411. Fry, A., Lanier, L. L., and Weiss, A. (1996). J. Exp. Med. 184, 295–300. Garni-Wagner, B. A., Purohit, A., Mathew, P. A., Bennett, M., and Kumar, K. (1993). J. Immunol. 151, 60–70. Gerosa, F., Baldani-Guerra, B., Nisii, C., Marchesini, V., Carra, G., and Trinchieri, G. (2002). J. Exp. Med. 195, 327–333. Gilfillan, S., Ho, E. L., Cella, M., Yokoyama, W. M., and Colonna, M. (2002). Nat. Immunol. 3, 1150–1155. Girardi, M., Oppenheim, D. E., Steele, C. R., Lewis, J. M., Glusac, E., Filler, R., Hobby, P., Sutton, B., Tigelaar, R. E., and Hayday, A. C. (2001). Science 294, 605–609. Griffith, T. S., Brunner, T., Fletcher, S. M., Green, D. R., and Ferguson, T. A. (1995). Science 270, 1189–1192. Grimm, E. A., Mazumder, A., Zhang, H. Z., and Rosenberg, S. A. (1982). J. Exp. Med. 155, 1823–1841. Groh, V., Bahram, S., Bauer, S., Herman, A., Beauchamp, M., and Spies, T. (1996). Proc. Natl. Acad. Sci. USA 93, 12445–12450. Groh, V., Rhinehart, R., Randolph-Habecker, J., Topp, M. S., Riddell, S. R., and Spies, T. (2001a). Nat. Immunol. 2, 255–260.

NK Cells and Cancer

151

Groh, V., Rhinehart, R., Secrist, H., Bauer, S., Grabstein, K. H., and Spies, T. (1999). Proc. Natl. Acad. Sci. USA 96, 6879–6884. Groh, V., Steinle, A., Bauer, S., and Spies, T. (1998). Science 279, 1737–1740. Groh, V., Wu, J., Yee, C., and Spies, T. (2002). Nature 419, 734–738. Hahne, M., Rimoldi, D., Schroter, M., Romero, P., Schreier, M., French, L. E., Schneider, P., Bornand, T., Fontana, A., Lienard, D., Cerottini, J., and Tschopp, J. (1996). Science 274, 1363–1366. Hata, K., Zhang, X. R., Iwatsuki, S., Van Thiel, D. H., Herberman, R. B., and Whiteside, T. L. (1990). Clin. Immunol. Immunopathol. 56, 401–419. Held, W., Roland, J., and Raulet, D. H. (1995). Nature 376, 355–358. Henkart, P. A. (1994). Immunity 1, 343–346. Henney, C. S., Kuribayashi, K., Kern, D. E., and Gillis, S. (1981). Nature 291, 335–338. Herberman, R. B. (2002). Semin. Oncol. 29, 27–30. Hersey, P., and Bolhuis, R. (1987). Immunol. Today 8, 233–239. Hersey, P., and Zhang, X. D. (2001). Nat. Rev. Cancer 1, 142–150. Houchins, J. P., Lanier, L. L., Niemi, E., Phillips, J. H., and Ryan, J. C. (1997). J. Immunol. 158, 3603–3609. Houchins, J. P., Yabe, T., McSherry, C., and Bach, F. H. (1991). J. Exp. Med. 173, 1017–1020. Ishido, S., Choi, J. K., Lee, B. S., Wang, C., DeMaria, M., Johnson, R. P., Cohen, G. B., and Jung, J. U. (2000). Immunity 13, 365–374. Jamieson, A. M., Diefenbach, A., McMahon, C. W., Xiong, N., Carlyle, J. R., and Raulet, D. H. (2002). Immunity 17, 19–29. Jan Chalupny, N., Sutherland, C. L., Lawrence, W. A., Rein-Weston, A., and Cosman, D. (2003). Biochem. Biophys. Res. Commun. 305, 129–135. Jia, S. H., Kurepa, Z., Bai, A., and Forman, J. (2000). J. Immunol. 165, 6142–6147. Kagi, D., Ledermann, B., Burki, K., Seiler, P., Odermatt, B., Olsen, K. J., Podack, E. R., Zinkernagel, R. M., and Hengartner, H. (1994). Nature 369, 31–37. Kaplan, D. H., Shankaran, V., Dighe, A. S., Stockert, E., Aguet, M., Old, L. J., and Schreiber, R. D. (1998). Proc. Nat. Acad. Sci. USA 95, 7556–7561. Karlhofer, F. M., Ribaudo, R. K., and Yokoyama, W. M. (1992). Nature 358, 66–70. Karre, K., Ljunggren, H. G., Piontek, G., and Kiessling, R. (1986). Nature 319, 675–678. Kashii, Y., Giorda, R., Herberman, R. B., Whiteside, T. L., and Vujanovic, N. L. (1999). J. Immunol. 163, 5358–5366. Kayagaki, N., Yamaguchi, N., Nakayama, M., Takeda, K., Akiba, H., Tsutsui, H., Okamura, H., Nakanishi, K., Okumura, K., and Yagita, H. (1999). J. Immunol. 163, 1906–1913. Kennedy, M. K., Glaccum, M., Brown, S. N., Butz, E. A., Viney, J. L., Embers, M., Matsuki, N., Charrier, K., Sedger, L., Willis, C. R., Brasel, K., Morrissey, P. J., Stocking, K., Schuh, J. C., Joyce, S., and Peschon, J. J. (2000). J. Exp. Med. 191, 771–780. Kiessling, R., Klein, E., and Wigzell, H. (1975). Eur. J. Immunol. 5, 112–117. Kodama, T., Takeda, K., Shimozato, O., Hayakawa, Y., Atsuta, M., Kobayashi, K., Ito, M., Yagita, H., and Okumura, K. (1999). Eur. J. Immunol. 29, 1390–1396. Koh, C. Y., Blazar, B. R., George, T., Welniak, L. A., Capitini, C. M., Raziuddin, A., Murphy, W. J., and Bennett, M. (2001). Blood 97, 3132–3132. Koo, G. C., and Peppard, J. R. (1984). Hybridoma 3, 301–303. Lacorazza, H. D., Miyazaki, Y., Di Cristofano, A., Deblasio, A., Hedvat, C., Zhang, J., Cordon-Cardo, C., Mao, S., Pandolfi, P. P., and Nimer, S. D. (2002). Immunity 17, 437–439. Lanier, L. L. (1998a). Cell 92, 705–707. Lanier, L. L. (1998b). Annu. Rev. Immunol. 16, 359–393. Lanier, L. L. (2001a). Nat. Immunol. 2, 23–27. Lanier, L. L. (2001b). Nat. Med. 7, 1178–80. Lanier, L. L. (2003). Curr. Opin. Immunol. 15, 308–314.

152

Jun Wu and Lewis L. Lanier

Lanier, L. L., Corliss, B., Wu, J., and Phillips, J. H. (1998a). Immunity 8, 693–701. Lanier, L. L., Corliss, B. C., Wu, J., Leong, C., and Phillips, J. H. (1998b). Nature 391, 703–707. Lanier, L. L., Le, A. M., Phillips, J. H., Warner, N. L., and Babcock, G. F. (1983). J. Immunol. 131, 1789–1796. Lanier, L. L., Spits, H., and Phillips, J. H. (1992). Immunol. Today 13, 392–395. Lanier, L. L., Testi, R., Bindl, J., and Phillips, J. H. (1989a). J. Exp. Med. 169, 2233–2238. Lanier, L. L., Yu, G., and Phillips, J. H. (1989b). Nature 342, 803–805. Latchman, Y., McKay, P. F., and Reiser, H. (1998). J. Immunol. 161, 5809–5812. Lawrence, D., Shahrokh, Z., Marsters, S., Achilles, K., Shih, D., Mounho, B., Hillan, K., Totpal, K., DeForge, L., Schow, P., Hooley, J., Sherwood, S., Pai, R., Leung, S., Khan, L., Gliniak, B., Bussiere, J., Smith, C. A., Strom, S. S., Kelley, S., Fox, J. A., Thomas, D., and Ashkenazi, A. (2001). Nat. Med. 7, 383–385. Lazetic, S., Chang, C., Houchins, J. P., Lanier, L. L., and Phillips, J. H. (1996). J. Immunol. 157, 4741–4745. Le Drean, E., Vely, F., Olcese, L., Cambiaggi, A., Guia, S., Krystal, G., Gervois, N., Moretta, A., Jotereau, F., and Vivier, E. (1998). Eur. J. Immunol. 28, 264–276. Lee, N., Liano, M., Carretero, M., Ishitani, A., Navarro, F., Lopez-Botet, M., and Geraghty, D. E. (1998). Proc. Natl. Acad. Sci. USA 95, 5199–5204. Lee, S. H., Girard, S., Macina, D., Busa, M., Zafer, A., Belouchi, A., Gros, P., and Vidal, S. M. (2001). Nat. Genet 28, 42–45. Leiden, J. M., Karpinski, B. A., Gottschalk, L., and Kornbluth, J. (1989). J. Immunol. 142, 2140–2147. Li, P., Morris, D. L., Willcox, B. E., Steinle, A., Spies, T., and Strong, R. K. (2001). Nat. Immunol. 2, 443–451. Li, P., Willie, S. T., Bauer, S., Morris, D. L., Spies, T., and Strong, R. K. (1999). Immunity 10, 577–584. Lian, R. H., and Kumar, V. (2002). Semin. Immunol. 14, 453–460. Litwin, V., Gumperz, J., Parham, P., Phillips, J. H., and Lanier, L. L. (1993). J. Exp. Med. 178, 1321–1336. Liu, C. C., Persechini, P. M., and Young, J. D. (1995). Immunol. Rev. 146, 145–175. Ljunggren, H. G., and Karre, K. (1985). J. Exp. Med. 162, 1745–1759. Lodoen, M., Ogasawara, K., Hamerman, J. A., Arase, H., Houchins, J. P., Mocarski, E. S., and Lanier, L. L. (2003). J. Exp. Med. 197, 1245–1253. Long, E. O. (1999). Annu. Rev. Immunol. 17, 875–904. Long, E. O., Burshtyn, D. N., Clark, W. P., Peruzzi, M., Rajagopalan, S., Rojo, S., Wagtmann, N., and Winter, C. C. (1997). Immunol. Rev. 155, 135–44. Malarkannan, S., Shih, P. P., Eden, P. A., Horng, T., Zuberi, A. R., Christianson, G., Roopenian, D., and Shastri, N. (1998). J. Immunol. 161, 3501–3509. Mandelboim, O., Lieberman, N., Lev, M., Paul, L., Arnon, T. I., Bushkin, Y., Davis, D. M., Strominger, J. L., Yewdell, J. W., and Porgador, A. (2001). Nature 409, 1055–1060. Mandelboim, O., Malik, P., Davis, D. M., Jo, C. H., Boyson, J. E., and Strominger, J. L. (1999). Proc. Natl. Acad. Sci. USA 96, 5640–5644. Mason, L. H., Anderson, S. K., Yokoyama, W. M., Smith, H. R. C., Winkler-Pickett, R., and Ortaldo, J. R. (1996). J. Exp. Med. 184, 2119–2128. Mathew, P. A., Garni-Wagner, B. A., Land, K., Takashima, A., Stoneman, E., Bennett, M., and Kumar, V. (1993). J. Immunol. 151, 5328–5337. Moretta, A., Biassoni, R., Bottino, C., Mingari, M. C., and Moretta, L. (2000). Immunol. Today 21, 228–234. Moretta, A., Bottino, C., Vitale, M., Pende, D., Cantoni, C., Mingari, M. C., Biassoni, R., and Moretta, L. (2001). Annu. Rev. Immunol. 19, 197–223.

NK Cells and Cancer

153

Moretta, L., Sivori, S., Vitale, M., Pende, D., Morelli, L., Augugliaro, R., Bottino, C., and Moretta, L. (1995). J. Exp. Med. 182, 875–884. Mrozek, E., Anderson, P., and Caligiuri, M. A. (1996). Blood 87, 2632–2640. Mule, J. J., Shu, S., Schwarz, S. L., and Rosenberg, S. A. (1984). Science 225, 1487–1489. Nakajima, H., Cella, M., Bouchon, A., Grierson, H. L., Lewis, J., Duckett, C. S., Cohen, J. I., and Colonna, M. (2000). Eur. J. Immunol. 30, 3309–3318. Nakamura, M. C., Niemi, E. C., Fisher, M. J., Shultz, L. D., Seaman, W. E., and Ryan, J. C. (1997). Exp. Med. 185, 673–684. Nguyen, T., Zhang, X. D., and Hersey, P. (2001). Clin. Cancer Res. 7, 966s–973s. Nichols, K. E., Harkin, D. P., Levitz, S., Krainer, M., Kolquist, K. A., Genovese, C., Bernard, A., Ferguson, M., Zuo, L., Snyder, E., Buckler, A. J., Wise, C., Ashley, I., Lovett, M., Valentine, M. B., Look, A. T., Gerald, W., Housman, D. E., and Haber, D. A. (1998). Proc. Natl. Acad. Sci. USA 95, 13765–13770. Niehans, G. A., Brunner, T., Frizelle, S. P., Liston, J. C., Salerno, C. T., Knapp, D. J., Green, D. R., and Kratzke, R. A. (1997). Cancer Res. 57, 1007–1012. Nishimura, M. I., Stroynowski, I., Hood, L., and Ostrand-Rosenberg, S. (1988). J. Immunol. 141, 4403–4409. O’Connell, J., O’Sullivan, G. C., Collins, J. K., and Shanahan, F. (1996). J. Exp. Med. 184, 1075–1082. Ogasawara, K., Hamerman, J. A., Hsin, H., Chikuma, S., Bour-Jordan, H., Chen, T., Pertel, T., Carnaud, C., Bluestone, J. A., and Lanier, L. L. (2003). Immunity 18, 41–51. Ortaldo, J. R., Oldham, R. K., Cannon, G. C., and Herberman, R. B. (1977). J. Natl. Cancer Inst. 59, 77–82. Pages, F., Ragueneau, M., Rottapel, R., Truneh, A., Nunes, J., Imbert, J., and Olive, D. (1994). Nature 369, 3327–3329. Pardo, J., Balkow, S., Anel, A., and Simon, M. M. (2002). Eur. J. Immunol. 32, 2881–2887. Parolini, S., Bottino, C., Falco, M., Augugliaro, R., Giliani, S., Franceschini, R., Ochs, H. D., Wolf, H., Bonnefoy, J. Y., Biassoni, R., Moretta, L., Notarangelo, L. D., and Moretta, A. (2000). J. Exp. Med. 192, 337–346. Pawlowska, A. B., Hashino, S., McKenna, H., Weigel, B. J., Taylor, P. A., and Blazar, B. R. (2001). Blood 97, 1474–1482. Pena, J., Alonso, C., Solana, R., Serrano, R., Carracedo, J., and Ramirez, R. (1990). Eur. J. Immunol. 20, 2445–2448. Pende, D., Cantoni, C., Rivera, P., Vitale, M., Castriconi, R., Marcenaro, S., Nanni, M., Biassoni, R., Bottino, C., Moretta, A., and Moretta, L. (2001). Eur. J. Immunol. 31, 1076–1086. Pende, D., Parolini, S., Pessino, A., Sivori, S., Augugliaro, R., Morelli, L., Marcenaro, E., Accame, L., Malaspina, A., Biassoni, R., Bottino, C., Moretta, L., and Moretta, A. (1999). J. Exp. Med. 190, 1505–1516. Peritt, D., Robertson, S., Gri, G., Showe, L., Aste-Amezaga, M., and Trinchieri, G. (1998). J. Immunol. 161, 5821–5824. Peron, J. M., Esche, C., Subbotin, V. M., Maliszewski, C., Lotze, M. T., and Shurin, M. R. (1998). J. Immunol. 161, 6164–6170. Perussia, B., Starr, S., Abraham, S., Fanning, V., and Trinchieri, G. (1983). J. Immunol. 130, 2133–2141. Pessino, A., Sivori, S., Bottino, C., Malaspina, A., Morelli, L., Moretta, L., Biassoni, R., and Moretta, A. (1998). J. Exp. Med. 188, 953–960. Phillips, J. H., and Lanier, L. L. (1986). J. Exp. Med. 164, 814–825. Piccioli, D., Sbrana, S., Melandri, E., and Valiante, N. M. (2002). J. Exp. Med. 195, 335–341. Podack, E. R., Hengartner, H., and Lichtenheld, M. G. (1991). Annu. Rev. Immunol. 9, 129–157.

154

Jun Wu and Lewis L. Lanier

Prasad, K. V., Cai, Y. C., Raab, M., Duckworth, B., Cantley, L., Shoelson, S. E., and Rudd, C. E. (1994). Proc. Natl. Acad. Sci. USA 91, 2834–2838. Radaev, S., Kattah, M., Zou, Z., Colonna, M., and Sun, P. D. (2002). J. Immunol. 169, 6279–6285. Radaev, S., Rostro, B., Brooks, A. G., Colonna, M., and Sun, P. D. (2001). Immunity 15, 1039–1049. Radosavljevic, M., and Bahram, S. (2003). Immunogenetics 55, 1–9. Radosavljevic, M., Cuillerier, B., Wilson, M. J., Clement, O., Wicker, S., Gilfillan, S., Beck, S., Trowsdale, J., and Bahram, S. (2002). Genomics 79, 114–123. Ravetch, J. V., and Lanier, L. L. (2000). Science 290, 84–89. Reth, M. (1989). Nature 338, 383–384. Robertson, M. J. (2002). J. Leukoc. Biol. 71, 173–183. Rosenberg, S. A., Lotze, M. T., Muul, L. M., Chang, A. E., Avis, F. P., Leitman, S., Linehan, W. M., Robertson, C. N., Lee, R. E., Rubin, J. T. et al. (1987). N. Engl. J. Med. 316, 889–897. Ruggeri, L., Capanni, M., Urbani, E., Perruccio, K., Shlomchik, W. D., Tosti, A., Posati, S., Rogaia, D., Frassoni, F., Aversa, F., Martelli, M. F., and Velardi, A. (2002). Science 295, 2097–2100. Ryan, J. C., Niemi, E. C., Nakamura, M. C., and Seaman, W. E. (1995). J. Exp. Med. 181, 1911–1915. Sayos, J., Wu, C., Morra, M., Wang, N., Zhang, X., Allen, D., van Schaik, S., Notarangelo, L., Geha, R., Roncarolo, M. G., Oettgen, H., De Vries, J. E., Aversa, G., and Terhorst, C. (1998). Nature 395, 462–469. Scalzo, A. A., Fitzgerald, N. A., Wallace, C. R., Gibbons, A. E., Smart, Y. C., Burton, R. C., and Shellam, G. R. (1992). J. Immunol. 149, 581–589. Shankaran, V., Ikeda, H., Bruce, A. T., White, J. M., Swanson, P. E., Old, L. J., and Schreiber, R. D. (2001). Nature 410, 1107–1111. Sirianni, M. C., Vincenzi, L., Topino, S., Giovannetti, A., Mazzetta, F., Libi, F., Scaramuzzi, D., Andreoni, M., Pinter, E., Baccarini, S., Rezza, G., Monini, P., and Ensoli, B. (2002). Eur. J. Immunol. 32, 2711–2720. Sivori, S., Parolini, S., Falco, M., Marcenaro, E., Biassoni, R., Bottino, C., Moretta, L., and Moretta, A. (2000). Eur. J. Immunol. 30, 787–793. Sivori, S., Vitale, M., Morelli, L., Sanseverino, L., Augugliaro, R., Bottino, C., Moretta, L., and Moretta, A. (1997). J. Exp. Med. 186, 1129–1136. Smith, H. R., Heusel, J. W., Mehta, I. K., Kim, S., Dorner, B. G., Naidenko, O. V., Iizuka, K., Furukawa, H., Beckman, D. L., Pingel, J. T., Scalzo, A. A., Fremont, D. H., and Yokoyama, W. M. (2002). Proc. Natl. Acad. Sci. USA 99, 8826–8831. Smith, K. M., Wu, J., Bakker, A. B., Phillips, J. H., and Lanier, L. L. (1998). J. Immunol. 161, 7–10. Smyth, M. J., Browne, K. A., Kinnear, B. F., Trapani, J. A., and Warren, H. S. (1995). J. Leukoc. Biol. 57, 88–93. Smyth, M. J., Cretney, E., Takeda, K., Wiltrout, R. H., Sedger, L. M., Kayagaki, N., Yagita, H., and Okumura, K. (2001a). J. Exp. Med. 193, 661–670. Smyth, M. J., Crowe, N. Y., and Godfrey, D. I. (2001b). Int. Immunol. 13, 459–463. Smyth, M. J., Crowe, N. Y., Hayakawa, Y., Takeda, K., Yagita, H., and Godfrey, D. I. (2002a). Curr. Opin. Immunol. 14, 165–171. Smyth, M. J., Godfrey, D. I., and Trapani, J. A. (2001c). Nat. Immunol. 2, 293–299. Smyth, M. J., Hayakawa, Y., Takeda, K., and Yagita, H. (2002b). Nat. Rev. Cancer 2, 850–861. Smyth, M. J., Takeda, K., Hayakawa, Y., Peschon, J. J., van den Brink, M. R., and Yagita, H. (2003). Immunity 18, 1–6.

NK Cells and Cancer

155

Smyth, M. J., Thia, K. Y., Cretney, E., Kelly, J. M., Snook, M. B., Forbes, C. A., and Scalzo, A. A. (1999). J. Immunol. 162, 6658–6662. Smyth, M. J., Thia, K. Y., Street, S. E., Cretney, E., Trapani, J. A., Taniguchi, M., Kawano, T., Pelikan, S. B., Crowe, N. Y., and Godfrey, D. I. (2000a). J. Exp. Med. 191, 661–668. Smyth, M. J., Thia, K. Y., Street, S. E., MacGregor, D., Godfrey, D. I., and Trapani, J. A. (2000b). J. Exp. Med. 192, 755–760. Spits, H., Lanier, L. L., and Phillips, J. H. (1995). Blood 85, 2654–2670. Strand, S., Hofmann, W. J., Hug, H., Muller, M., Otto, G., Strand, D., Mariani, S. M., Stremmel, W., Krammer, P. H., and Galle, P. R. (1996). Nat. Med. 2, 1361–1366. Street, S. E., Cretney, E., and Smyth, M. J. (2001). Blood 97, 192–197. Street, S. E., Trapani, J. A., MacGregor, D., and Smyth, M. J. (2002). J. Exp. Med. 196, 129–134. Takeda, K., Hayakawa, Y., Smyth, M. J., Kayagaki, N., Yamaguchi, N., Kakuta, S., Iwakura, Y., Yagita, H., and Okumura, K. (2001). Nat. Med. 7, 94–100. Tangye, S. G., Lazetic, S., Woollatt, E., Sutherland, G. R., Lanier, L. L., and Phillips, J. H. (1999). J. Immunol. 162, 6981–6985. Tangye, S. G., Phillips, J. H., Lanier, L. L., and Nichols, K. E. (2000). J. Immunol. 165, 2932–2936. Tay, C. H., and Welsh, R. M. (1997). J. Virol. 71, 267–275. Timonen, T., Saksela, E., Ranki, A., and Hayry, P. (1979). Cell Immunol. 48, 133–148. Tomasello, E., Desmoulins, P. O., Chemin, K., Guia, S., Cremer, H., Ortaldo, J., Love, P., Kaiserlian, D., and Vivier, E. (2000). Immunity 13, 355–364. Tortorella, D., Gewurz, B. E., Furman, M. H., Schust, D. J., and Ploegh, H. L. (2000). Annu. Rev. Immunol. 18, 861–926. Trapani, J. A., Davis, J., Sutton, V. R., and Smyth, M. J. (2000). Curr. Opin. Immunol. 12, 323–329. Trapani, J. A., and Smyth, M. J. (2002). Nat. Rev. Immunol. 2, 735–747. Trinchieri, G. (1989). Adv. Immunol. 47, 187–376. Trinchieri, G. (1995). Semin. Immunol. 7, 83–88. Valiante, N. M., and Trinchieri, G. (1993). J. Exp. Med. 178, 1397–1406. van den Broek, M. E., Kagi, D., Ossendorp, F., Toes, R., Vamvakas, S., Lutz, W. K., Melief, C. J., Zinkernagel, R. M., and Hengartner, H. (1996). J. Exp. Med. 184, 1781–1790. van den Broek, M. F., Kagi, D., Zinkernagel, R. M., and Hengartner, H. (1995). Eur. J. Immunol. 25, 3514–3516. Vance, R. E., Kraft, J. R., Altman, J. D., Jensen, P. E., and Raulet, D. H. (1998). J. Exp. Med. 188, 1841–1848. Veillette, A., and Latour, S. (2003). Curr. Opin. Immunol. 15, 277–285. Vilches, C., and Parham, P. (2002). Annu. Rev. Immunol. 20, 217–251. Vitale, M., Bottino, C., Sivori, S., Sanseverino, L., Castriconi, R., Marcenaro, E., Augugliaro, R., Moretta, L., and Moretta, A. (1998). J. Exp. Med. 187, 2065–2072. Walczak, H., Miller, R. E., Ariail, K., Gliniak, B., Griffith, T. S., Kubin, M., Chin, W., Jones, J., Woodward, A., Le, T., Smith, C., Smolak, P., Goodwin, R. G., Rauch, C. T., Schuh, J. C., and Lynch, D. H. (1999). Nat. Med. 5, 157–163. Walker, P. R., Saas, P., and Dietrich, P. Y. (1998). Curr. Opin. Immunol. 10, 564–572. Ward, S. G., Westwick, J., Hall, N. D., and Samson, D. M. (1993). Eur. J. Immunol. 23, 2572–2577. Watzl, C., and Long, E. O. (2003). J. Exp. Med. 197, 77–85. Wiley, S., Schooley, K., Smolak, P. J., Din, W. S., Huang, C.-P., Nicholl, J. K., Sutherland, G. R., Smith, T. D., Rauch, C., Smith, C. A., and Goodwin, R. G. (1995). Immunity 3, 673–682. Wu, J., Cherwinski, H., Spies, T., Phillips, J. H., and Lanier, L. L. (2000). J. Exp. Med. 192, 1059–1068.

156

Jun Wu and Lewis L. Lanier

Wu, J., Song, Y., Bakker, A. B., Bauer, S., Spies, T., Lanier, L. L., and Phillips, J. H. (1999). Science 285, 730–732. Wu, T.-C., Huang, A. Y. C., Jaffe, E. M., Levitsky, H. I., and Pardoll, D. M. (1995). J. Exp. Med. 182, 1415–1421. Yokoyama, W. M., and Seaman, W. E. (1993). Annu. Rev. Immunol. 11, 613–635. Yu, Y. Y. L., Kumar, V., and Bennett, M. (1992). Ann. Rev. Immunol. 10, 189–214. Zamai, L., Ahmad, M., Bennett, I. M., Azzoni, L., Alnemri, E. S., and Perussia, B. (1998). J. Exp. Med. 188, 2375–2380. Zompi, S., Hamerman, J. A., Ogasawara, K., Schweighoffer, E., Tybulewicz, V. L., Santo, J. P., Lanier, L. L., and Colucci, F. (2003). Nat. Immunol. 4, 565–572.

Immunity to Cancer Through Immune Recognition of Altered Self: Studies with Melanoma Jose´ A. Guevara-Patin˜o, Mary Jo Turk, Jedd D. Wolchok, and Alan N. Houghton Memorial Sloan-Kettering Cancer Center and the Weill Graduate School of Medical Sciences and Medical School of Cornell University, 1275 York Avenue, New York, NY 10021

I. Introduction II. Melanoma Antigens A. Germ Cell-Cancer Antigens B. Cancer Antigens Derived from Genetic Mutations and Atypical Gene Products C. Melanoma Differentiation Antigens III. Cellular Responses to Differentiation Antigens A. Cellular Immune Responses B. Heteroclitic Peptides C. Xenogeneic Immunization D. Altered Peptide Ligands IV. Antibody-Mediated Tumor Immunity and Autoimmunity A. Antibody-Mediated Immunity in Melanoma V. Clinical Applications A. Heteroclitic Peptides B. Xenogeneic DNA Vaccines C. Heat Shock Protein Peptides References

The adaptive immune system is capable of recognizing cancer through T- and B-cell receptors. However, priming adaptive immunity against self antigens is potentially a difficult task. Presentation of altered self to the immune system is a strategy to elicit immunity against poorly immunogenic antigens. We have shown that immunization with conserved paralogues of tumor antigens can induce adaptive immunity against self antigens expressed by cancer. Remarkably, cancer immunity elicited by closely related paralogues can generate distinct adaptive immune responses, either antibody or T-cell dependent. Cancer immunity induced by xenogeneic immunization follows multiple and alternative pathways. The effector phase of tumor immunity can be mediated by cytotoxic T cells or macrophages and perhaps natural killer cells for antibody-dependent immunity. Helper CD4þ T cells are typically, but not always, required to generate immunity. Autoimmunity is frequently observed following immunization. Cancer

Advances in CANCER RESEARCH 0065-230X/03 $35.00

157

Copyright 2003, Elsevier Inc. All rights reserved

158

Jose´ A. Guevara-Patin˜o et al.

immunity and autoimmunity use overlapping mechanisms, and therefore they are difficult to uncouple, but distinct pathways can be discerned that open the eventual possibility of uncoupling tumor immunity from autoimmunity. Studies examining the molecular basis for immunogenicity of conserved paralogues are facilitating the development of new strategies to rationally design vaccines that trigger adaptive immune responses to cancer. ß 2003 Elsevier Inc.

I. INTRODUCTION Melanoma is a malignant tumor arising from melanocytes in epithelium. This malignancy metastasizes readily and responds poorly to any form of therapy once it has metastasized. Melanoma is the most deadly form of skin cancer, and in the year 2000, almost 50,000 cases of malignant melanoma were predicted to occur in the United States in 2003. The risk of melanoma increases with age, but it frequently affects young, otherwise healthy people and represents the number one cause of cancer death in women aged 20–30 years (Agarwala and Kirkwood, 2002). The identification of an effective therapy against melanoma still represents a major endeavor. In this review, we have compiled some of the most relevant findings that may help us understand immune responses against melanoma antigens, and we discuss new immunotherapeutic strategies against melanoma. Cancer vaccine development faces the fundamental difficulty that cancer cells arise from the host’s self tissues. Because most tumor antigens are products of normal or altered cellular genes, they are typically not efficient at initiating an immune response. Thus, a crucial problem in cancer immunotherapy is how to efficiently prime immune cells to respond to poorly immunogenic tumor antigens. Immune recognition and rejection of cancer cells have been extensively documented in experimental models, and immune recognition of cancer antigens has been characterized in healthy humans and individuals with cancer. Recent experiments in mouse models have resurrected the notion of immune surveillance, with immunological pressures shaping and editing tumors. In immune surveillance, incipient de novo malignancies are attacked by innate and adaptive immune responses, but immune editing selects cancer cell populations with the ability to evade and survive the recognition by the immune system. Despite these obstacles, approaches for triggering effective tumor immunity have been developed. Essential to the success of these strategies has been the detailed understanding of the biological mechanism involved in tumor immunity. We briefly describe the types of melanoma antigens seen by the immune system, characterize the mechanism underlying tumor immunity and autoimmunity, and finally describe our current clinical strategies in the area of active immune therapy.

Inducing Immuniy to Melanoma

159

II. MELANOMA ANTIGENS A. Germ Cell-Cancer Antigens These antigens are normally present in male germ cells, which lack major histocompatibility complex (MHC) surface expression and are also expressed by certain tumors (these antigens are also called cancer-testes antigens or tumor-specific shared antigens). Although germ cell antigens are silenced in healthy somatic cells, they are reexpressed in certain malignancies, probably through changes in DNA methylation. Among these antigens, MAGE-1 was identified as the first human gene product on cancer cells recognized by CD8þ T cells (Chaux et al., 1999a, 1999b; Van den Eynde and van der Bruggen, 1997). Following this finding, members of related and unrelated families of germ cell antigens have been identified, including BAGE, GAGE, and MAGE families (Boel et al., 1995; De Backer et al., 1999). Identification of NY-ESO-1 as the target of humoral and cellular components of the immune system has been notable. Autoantibodies have been frequently detected in patients with NY-ESO-1þ tumors (Chen et al., 2000; Jager et al., 2000), and moreover, antibody titers against NY-ESO-1 directly correlate with tumor load. Importantly, it has been demonstrated that CD8þ and CD4þ T cells recognize NY-ESO-1-derived MHC class I and II epitopes (Jager et al., 1998).

B. Cancer Antigens Derived from Genetic Mutations and Atypical Gene Products Mutations are major contributors to the pathogenesis of cancer. Immunogenicity of mutated gene products can be the result of alteration of protein trafficking or stability, or of enhanced visibility of otherwise poorly immunogenic epitopes to T or B cells. The immune recognition of unique antigens on chemically induced tumors in mice is probably largely the result of nucleotide point mutations leading to individual amino acid substitutions (Lurquin et al., 1989; Van den Eynde et al., 1999). The immune system can also see gene products from alternative reading frames, pseudogenes (Moreau-Aubry et al., 2000), and antisense strands of DNA (Van den Eynde et al., 1999). Mutations in the genes p53, p16/INK4, CDK4, and -catenin have been implicated in the pathogenesis of selected malignancies. Similarly, these mutations have also been described as targets of T-cell recognition in individual patients. In arguably the best example of immune recognition of a point mutation implicated in cancer pathogenesis, CD8þ T cells from a patient with metastatic melanoma treated at Memorial Sloan-Kettering

160

Jose´ A. Guevara-Patin˜o et al.

Cancer Center (MSKCC) were shown to recognize multiple epitopes expressed by autologous melanoma cells, including one produced by a single point mutation in the CDK4 protein (Houghton, 1994). This mutation created an anchor residue that facilitated binding of a CDK4 peptide to a MHC molecule, permitting recognition by autologous T cells. CDK4 is an essential kinase that regulates progression through the cell cycle, and the mutation found in this melanoma patient led to loss of an inhibitory site regulated by the cell cycle inhibitor p16/INK4a. Mutations in both CDK4 and p16/INK4a have been implicated in familial forms of melanoma.

C. Melanoma Differentiation Antigens Melanoma differentiation antigens are expressed concomitantly by healthy melanocytes and by melanoma. Among these antigens are the coat color proteins, which include a panel of membrane proteins present at different stages of melanocyte differentiation (Houghton et al., 1982), such as tyrosinase-related protein (Tyrp)-1 and Tyrp-2 (gp75/Tyrp-1 and Tyrp-2/ DCT), gp100, and tyrosinase. Immune recognition of differentiation antigens has been extensively studied in patients with melanoma. The initial identification of gp75 as a target of an antibody response in a patient with melanoma was described approximately a decade ago (Vijayasaradhi et al., 1990). This discovery was followed by the identification of tyrosinase, Tyrp2, and gp100 as targets of CD8þ T cells in melanoma patients. Anecdotally, the presence of CD8þ T cells against a gp100 epitope correlated with longterm survival in patients with melanoma (Bakker et al., 1994; Kawakami et al., 1994). Differentiation antigens are commonly recognized by T cells in patients with melanoma. In general, these T cells are found at a much higher frequency in patients than are T cells against germ cell antigens. It is harder to compare the frequencies of T cells recognizing mutations because these have been characterized in only a handful of patients.

III. CELLULAR RESPONSES TO DIFFERENTIATION ANTIGENS A. Cellular Immune Responses The protective immune response against tumors and invading intracellular pathogens relies in part on the vigorous attack of cytotoxic T lymphocytes (CTLs). This response occurs on recognition by the T-cell receptor (TCR) of short peptides (8–10 amino acids) bound to self MHC class

Inducing Immuniy to Melanoma

161

I molecules on the surface of target cells. These short peptides are the products of proteolytic processing of tumor or pathogen-derived proteins. On primary immunization, naive CD8þ T cells proliferate in the draining lymph nodes, generating a large number of effector T cells that migrate to target tissues to eliminate tumor cells. However, to differentiate into effector CTLs, naive CD8þ T cells have to concurrently receive at least two activation signals. The first signal (peptide:MHC) is transmitted through the TCR-CD3 complex, and the second (costimulatory) signal is delivered by professional antigen presenting cells (APCs; Bretscher and Cohn, 1970). In the thymus, a strong peptide:MHC signal induces negative selection of immature thymocytes, regardless of the second signal (Nossal, 1994). In the periphery, the same strong peptide:MHC signal induces immunity through naive CD8þ T-cell differentiation or anergy, depending on the presence or absence of the second signal (Bretscher and Cohn, 1970). However, differentiated CD8þ T cells can lyse their target cells in the presence of a weak peptide:MHC without the second signal, and this distinction can be exploited in vaccine development, as discussed below. To prevent immune responses against host-derived protein products, T cells are educated during thymic development by a mechanism of negative selection (Nossal, 1994). Intracellular proteolytic processing of self-derived antigens generates a large variety of peptides that can interact with the MHC class I complex. These peptides are classified according to their ability to bind to the MHC class I molecule. Good binders are capable of stabilizing the MHC class I complex, but they typically tolerize the CD8þ T cells that recognize them. In contrast, poor binders partially bind to the MHC class I complex, leading to ineffective MHC class I complex stabilization, which allows low-avidity CD8þ T cells to escape negative selection. In summary, partially stabilized MHC class I complexes do not tolerize CD8þ T cells and do not elicit T-cell responses. As a consequence, the immune system remains blind to these shrouded ‘‘weaker’’ antigens. Interestingly, we have shown that these antigenic, but poorly immunogenic, peptides can be used as vaccine targets in cancer.

B. Heteroclitic Peptides The presence of subliminal epitopes on the surface of cancer cells can be used in the development of CTL-based vaccines. The potency and the specificity of the CTL response can be enhanced by selectively replacing certain amino acid residues that are necessary for optimal binding to the MHC class I complex. These optimized peptide variants, called heteroclitic peptides or MHC anchor-modified ligands, can effectively prime naive CD8þ T cells and can induce a cross-reactive response against the native peptide

162

Jose´ A. Guevara-Patin˜o et al.

(Fig. 1). This strategy was successfully demonstrated for the first time in a tumor model using a melanoma differentiation antigen, gp75 (Dyall et al., 1998). Following amino acid analysis using the canonical H2-Kb binding motif, an epitope that exhibited relatively poor binding, TWH (TWHRYHLL), was identified. Further structural analysis of the peptide sequence showed the presence of bulky side chains on two amino acids, W and H, at positions 2 and 3. Because these amino acids were predicted to be buried in the H2-Kb binding groove, and based on the presumption that their bulky chains may interfere with MHC class I binding, W and H were substituted by A and Y, respectively. To demonstrate changes in the overall peptide affinity, a TAP-2-deficient RMA-s cell assay was used. The optimized peptide variant called TAY (TAYRYHLL) was shown to bind to H2-Kb with a much higher affinity than the wild-type counterpart (Fig. 2). Furthermore, a standard 4-hour 51Cr-release assay was used to show that TAY is a heteroclitic immunogen for the native gp75 melanoma peptide, TWH. We observed that enriched CTLs purified from mice immunized with TAY (heteroclitic peptide) could recognize and lyse B16 melanoma cells

Fig. 1 Model of naive CD8þ T cell activation by heteroclitic peptides. (A) Naive CD8þ T cells are not activated by sub-optimal peptide:MHC class I complexes. (B) Naive CD8þ T cells can be primed by heteroclitic peptide:MHC class I complexes and differentiate as CTLs. (C) CTLs against heteroclitic peptides can cross-react and lyse tumor cells expressing wild-type peptide:MHC class I complexes.

163

% Maximal stabilization

Inducing Immuniy to Melanoma

120 90 60 30 0 5

6

7

8

9 10 11

−log [peptide (M)] Fig. 2 The heteroclitic peptide TAY binds better to Kb than the wild-type TWH peptide. An RMA-S stabilization assay (Dyall et al., 1996) was carried out using the following peptides: positive control, SSI (SSIEFARL) (^), negative control, the Dd-binding peptide HIV-10 (RGPGRAFVTI) (&), TWH (*) and its heteroclitic variant TAY (~). In this type of assay, the percentage of maximal stabilization provides a direct correlate of peptide binding. Reproduced from the Journal of Experimental Medicine, 1998, vol. 188, p. 1557 (Dyall et al., 1998), with copyright permission from The Rockefeller University Press.

expressing only the native TWH form of gp75, but not its gp75-negative variant, B78.H1 cells. In addition, lysis of B78.H1 melanoma was only observed following cell coating with TWH. The biological relevance of this approach was further demonstrated when mice immunized with the heteroclitic peptide variant were protected against challenge with syngeneic tumor, B16 melanoma (Fig. 3). The results obtained in this study established the principle for tumor immunity that priming naive CD8þT cells to recognize and destroy tumor cells in vivo can be achieved by rationally optimizing antigenic, not nonimmunogenic, cancer epitopes to achieve the best MHC:peptide:TCR interactions. It is important to emphasize that the methodical selection of MHC class I target epitopes and the design of heteroclitic peptides are complex and time-consuming tasks. To improve this procedure, our laboratory has developed a computer algorithm (EpitOptimizer) that facilitates the design and discovery of heteroclitic peptides for enhanced MHC class I and class II binding (Houghton and Guevara, unpublished data). Studies carried out in our laboratory evaluated the basis for immunity induced by active immunization of mice with xenogeneic human gp100. This study demonstrated that the presence of a single heteroclitic epitope in human gp100 created by three amino acid substitutions at predicted anchor residues was necessary and sufficient to induce protective tumor immunity

164

Jose´ A. Guevara-Patin˜o et al.

TWH 0/10

300 200

Tumor size mm2

100 0 0

5

10

15

20

25

10

15

20

25

TAY 10/10

300 200 100 0 0

5

Days post-challenge Fig. 3 In vivo efficacy of a heteroclitic anti-melanoma vaccine. Immunization with the heteroclitic peptide TAY prevents B16 tumor growth. Groups of 10 mice were vaccinated with either TWH or TAY in Titermax adjuvant (Dyall et al., 1995). Seven days later, mice were challenged with 105 B16 melanoma cells. Tumor growth was monitored for >90 days. Reproduced from the Journal of Experimental Medicine, 1998, vol. 188, p. 1559 (Dyall et al., 1998), with copyright permission from The Rockefeller University Press.

against melanoma B16 challenge (Gold et al., 2003). Tumor rejection was assessed following immunization with either human or mouse gp100. Although C57BL/6 mice immunized with human gp100 DNA were protected against syngeneic melanoma challenge through cytotoxic T-cell responses, no tumor protection was observed in mice immunized with mouse gp100 DNA. Furthermore, the relevance of a H2-Db-restricted peptide (KVPRNQDWL) was evidenced by the lack of tumor protection in mice who received tolerizing doses of human gp100 H2-Db-restricted peptide, following immunization with human gp100 DNA. The protective properties of human gp100 were abrogated by substituting the human gp100-H2-Dbrestricted peptide with the mouse gp100 peptide (EGSRNQDWL). In this model of xenogeneic DNA immunization, it was shown that a single heteroclitic epitope is sufficient to induce protective tumor immunity. The results in these studies further support a strategy to immunize against tumor antigens using heteroclitic epitopes of differentiation antigens (Gold et al., 2003).

Inducing Immuniy to Melanoma

165

C. Xenogeneic Immunization Cellular and humoral immune responses induced by xenogeneic immunization against melanoma differentiation antigens have been the focus of our laboratory for the last 8 years. Human differentiation antigens, including the paralogues TYRP-1/gp75, Tyrp-2/DCT, tyrosinase, and gp100 are highly conserved at the protein level (80%–85%) compared with their mouse counterpart. Although immunization with syngeneic mouse homologues of these genes confers no detectable tumor immunity or autoimmunity in mouse models, xenogeneic immunization with paralogues, human tyrosinase, TYRP-1/gp75, Tyrp-2/DCT (Fig. 4), or gp100, induces tumor immunity and autoimmunity manifested as coat depigmentation. Other altered forms of protein may also be effective immunogens. A study by Naftzger et al. (1996) showed that immunization with lysed insect cells expressing an insoluble immature form of mouse gp75 overcame immune tolerance to this self antigen through cross-reactive antibody-mediated immunity (Naftzger et al., 1996). Most important, lysed human melanoma cells administered with adjuvant induced the strongest antibody response

Fig. 4 Xenogeneic DNA immunization of C57BL/6 mice induces autoimmune depigmentation. Mice immunized with hTyrp-2 developed vitiligo, which manifested as appearance of patches of white fur. Picture shows a control mouse (top) and a mouse immunized with hTyrp-2 (bottom). (See Color Insert.)

166

Jose´ A. Guevara-Patin˜o et al.

against mouse melanoma cells (see section on Antibody-Mediated Immunity to Melanoma). An extensive study to characterize the immune response induced by xenogeneic immunization with human differentiation antigens showed that tumor immunity and autoimmunity were the common result of immunization with human TYRP-1/gp75, Tyrp-2/DCT, gp100, and tyrosinase. Furthermore, these studies demonstrated that the cellular and humoral requirements for tumor protection and autoimmunity are dependent on minor differences in the immunogen. Immunization of C57BL/6 mice with human TYRP-1/gp75 elicits a potent antibody-mediated antitumor response and autoimmunity, whereas the antitumor immunity and autoimmunity induced by immunization with human tyrosinase, Tyrp-2/DCT, and gp100 in murine models requires the presence of CD8þ T cells (Bowne et al., 1999; Hawkins et al., 2000; Weber et al., 1998). The immune response to Tyrp-2/DCT was CD8þ T-cell dependent and required CD4þ T-cell help at the priming phase, but played little or no role during the effector phase. In contrast, the immune response to gp100 was partially CD4þ T-cell independent. In a mouse model, immunization with human Tyrp-2/DCT DNA (Bowne et al., 1999) prevented tumor growth by >90% following challenge with autologous B16 melanoma cells, and it decreased lung metastasis recurrences after surgical resection of tumors (Hawkins et al., 2002). Further characterization of immunity induced by human Tyrp-2/DCT showed that tumor immunity could still occur in the absence of perforin or Fas ligand (Fig. 5). Tumor immunity in this model was dependent on both CD8þ and Perforin −/− Number of surface lung metastases

gld/gld 100

800

75 400

50 25

0

No treat

hTRP-2

0

No treat

hTRP-2

Fig. 5 Tumor immunity mediated by Tyrp-2 (TRP-2) is CD8þ T-cell dependent and perforin and Fas ligand independent. Groups of 9 or 10 perforin deficient (pfp-/-) and Fas liganddeficient (gld/gld) mice were immunized cutaneously with hTyrp-2 DNA and challenged intravenously with syngeneic B16 melanoma cells. The numbers of lung tumor metastases were quantified in treated and untreated mice. Significant tumor protection was observed in pfp–/– (P ¼ 0.0002) and gld/gld (P < 0.0001) mice. Reproduced from the Journal of Experimental Medicine, 1999, vol. 190, p. 1720 (Bowne et al., 1999), with copyright permission from The Rockefeller University Press.

167

Inducing Immuniy to Melanoma

CD4þ T cells, whereas autoimmunity was not observed in perforin-deficient mice (Fig. 6). In addition, autoimmunity induced by human Tyrp-2/DCT was not dependent on B or natural killer (NK) cells (Bowne et al., 1999). Neither tumor immunity nor autoimmunity was observed in IFN--deficient mice immunized with human Tyrp-2/DCT (Wolchok et al., 2001). Significantly, repletion of mice with exogenous recombinant IFN- at the priming and effector phases led to reconstitution of tumor immunity and autoimmunity. In contrast, IL-4 had no effect on tumor immunity, and paradoxically, its absence accelerated autoimmunity in mice immunized with xenogeneic human Tyrp-2/DCT. Xenogeneic immunization with human gp75 induced tumor immunity and autoimmunity, which was shown to be mediated by autoantibodies. The effect of IFN- was strikingly different, with higher-antibody titers to mouse gp75 occurring in IFN- deficient mice immunized with human gp75 compared with wild-type mice (Wolchok et al., 2001). Similar to Tyrp-2/DCT, immunization of mice with human gp100 results in induction of autoreactive CTLs, which lead to autoimmunity and considerable tumor immunity against tumor B16 melanoma cells (Hawkins et al., 2000). In this study, tumor immunity was distinctively mediated by CD8þ T cells in a CD4þ T-cell-independent manner. This was further confirmed in mice lacking the MHC class II complex (A. N. Houghton, unpublished data). As observed

Depigmented quadrants

4

3

2

1

0

wt (h)

wt (m)

m −/−

b 2 −/−

MHC II −/−

NKdepl

pfp −/−

gld/gld

Fig. 6 Human Tyrp-2-mediated autoimmunity is CD8þ T-cell and perforin dependent. Groups of 12–15 mice were immunized with hTyrp-2 or mTyrp-2 and challenged with syngeneic B16 melanoma cells. Abdominal fur coat depigmentation was quantified and expressed as depigmentation quadrants. Each dot represents a separate mouse. Groups include C57BL/6 (wt) mice treated with hTyrp-2 (h) or m Tyrp-2 (m) DNA. In addition, immunoglobulin deficient (–/–), 2 microglobulin-deficient (2–/–), MHC II-deficient (MHC II–/–), NK-depleted, perforin-deficient (pfp–/–), and fasl-deficient (gld/gld–/–) mice were immunized with hTyrp-2. Reproduced from the Journal of Experimental Medicine, 1999, vol. 190, p. 1720 (Bowne et al., 1999), with copyright permission from The Rockefeller University Press.

168

Jose´ A. Guevara-Patin˜o et al.

Requirements for xenogeneic DNA immunization hTyrp-1 Immunization phase B cells CD4+ T cell Effector phase Tumor Immunity Antibodies CD4+ T cell Macrophages Autoimmunity Antibodies Macrophages Complement

hTyrp-2 Immunization phase CD8+ T cell CD4+ T cell Effector phase Tumor Immunity CD8+ T cell Perforin/Fas indep Autoimmunity CD8+ T cell Perforin INFγ IL4

hgp 100 Immunization phase CD8+ T cell CD4+ T cell indep Effector phase Tumor Immunity CD8+ T cell

Autoimmunity CD8+ T cell

Fig. 7 Alternative pathways to tumor immunity and autoimmunity. Xenogeneic DNA immunization against gp100, DCT and gp75 melanoma antigens induces tumor immunity and autoimmunity by multiple overlapping mechanisms. Immunological pathways and immune components are indicated.

in xenogeneic immunizations with human Tyrp-2/DCT DNA, the induction of tumor protection by human gp100 was abrogated in MHC class I–deficient mice. Results obtained in these studies indicated the existence of antigen-specific, alternative immunological pathways to generate tumor immunity and autoimmunity. Animal models of melanoma have unveiled some of the intricacies of immunity to cancer and demonstrated the presence of overlapping, but alternative, mechanistic pathways for two common ends; tumor protection and autoimmunity (Fig. 7). The significant role of both humoral and cellular arms of the immune system in tumor clearance has also been confirmed. These studies facilitate the development of new strategies for active immunization that can lead to tumor immunity without undesirable autoimmunity.

D. Altered Peptide Ligands The potential application of selective manipulation of the MHC:peptide:TCR interactions can also be positively influenced by direct modification of the amino acids that are indirectly involved in the peptide:TCR contact (Evavold et al., 1993). These peptide analogs, named altered peptide

Inducing Immuniy to Melanoma

169

ligands, can be used to augment CTL immune responses. Manipulation of the peptide:TCR contact residues can have either T-cell agonistic or antagonistic effects. Agonist altered peptide ligands can be classified into partial agonists or superagonists, according to their effect on CTLs. This functional difference is based on the ability of the peptide to induce either some or all T-cell functional and proliferative activities, respectively. In contrast, use of antagonist-altered peptide ligands can lead to downregulation or functional inactivation of CTLs.

IV. ANTIBODY-MEDIATED TUMOR IMMUNITY AND AUTOIMMUNITY A. Antibody-Mediated Immunity in Melanoma Experiments with the melanoma differentiation antigen gp75 are an ideal platform for exploring the mechanism of humoral immunity, because antibodies to gp75 induce both tumor rejection and autoimmunity. The 80% homologous human and mouse gp75 proteins are recognized by the mouse IgG2a monoclonal antibody TA99 (Vijayasaradhi et al., 1991; Vijayasaradhi and Houghton, 1991). Cellular localization of gp75 is mainly melanosomal, but 2% of the newly synthesized protein resides on the cell surface, where it is accessible to antibodies (Takechi et al., 1996). To model antibody-mediated immune responses to gp75, two strategies have been used; active and passive immunizations. In the process of active immunization, the animal is immunized to elicit the production of antibodies or cellular immunity against gp75. Passive immunization refers to the transfer of anti-gp75 antibodies into the animal. Examination of both strategies reveals different mechanisms responsible for immunity to tumor and normal melanocytes in the skin.

1. ACTIVE IMMUNIZATION To generate an antibody-mediated immune response against a differentiation antigen, the immune system must first break tolerance or overcome ignorance to the antigen. In the case of gp75, an immune response is not elicited in mice given syngeneic protein, even in conjunction with various adjuvants. However, we have shown that an antigen can be presented to the host as an ‘‘altered self’’ to break immune tolerance (Naftzger et al., 1996). For example, immunizing mice with lysed Sf9 insect cells that were infected with baculovirus expressing mouse gp75 elicited antibodymediated rejection of B16 tumors and induced autoimmune depigmentation

170

Jose´ A. Guevara-Patin˜o et al.

of coat (Naftzger et al., 1996). Although Sf9 cells produce an insoluble early-processed form of gp75, which is similar to nascent gp75 following translation in the endoplasmic reticulum (both contain only unprocessed high-mannose carbohydrate chains), immunized mice produced antibodies recognizing the mature mouse protein after it had been processed through the Golgi apparatus/secretory pathway. Thus, ‘‘altered’’ mouse gp75 could effectively induce antibody-dependent immunity to wild-type mouse gp75. Presentation of ‘‘altered self’’ in the form of the related xenogeneic human paralogue is another strategy that we have shown to break apparent tolerance to differentiation antigens in mice. Human gp75 protein administered in Freund’s adjuvant elicited an antibody response against mouse gp75 that was potent enough to protect mice challenged with B16 tumor cells (Naftzger et al., 1996). This work was later followed by the design and characterization of a DNA vaccine encoding human gp75, which also elicited autoantibodies against mouse gp75 with concurrent tumor immunity and autoimmunity (Weber et al., 1998). It is notable that after two immunizations with human gp75 DNA, immunization with mouse gp75 DNA induced substantially higher levels of autoantibodies against mouse gp75, compared with three immunizations with only human gp75 DNA (Weber et al., 1998). This important observation indicates that, rather than eliciting anti-human gp75 antibodies that cross-react with mouse gp75, immunization with human DNA directly broke tolerance to the mouse protein. Overall, it is likely that polymorphisms in the human gp75 orthologue acquired during evolution provide novel MHC class II–restricted peptide epitopes for presentation on APCs. Several candidate cryptic peptides at Asn-linked glycosylation sites have been identified (Weber et al., 1998). Mechanistically, the pathways leading from vaccination with human gp75 DNA to tumor immunity and to autoimmunity appear to be overlapping, but distinct. Depletion and genetic experiments demonstrated that CD4þ T cells, NK cells, and FcyR expression are all required for immunity against B16 tumors (Weber et al., 1998). In contrast, autoimmune depigmentation occurs in the absence of FcyR or CD4 expression and in mice depleted of or deficient in CD4þ, CD8þ, or NK cells (Weber et al., 1998). In addition, investigation of the Sf9/gp75 vaccine in FcyR-deficient mice showed that tumor immunity, not autoimmunity, requires the activation of Fc receptors (Clynes et al., 1998).

2. PASSIVE IMMUNIZATION Studies reported by Hara et al. (1995) demonstrated that passive transfer of antibodies against gp75-induced tumor rejection in mice bearing B16 tumors. The TA99 antibody, which itself has no inhibitory effect on the growth of B16 cells (Takechi et al., 1996), dramatically decreased lung

171

Inducing Immuniy to Melanoma

metastases when given as late as 7 days after intravenous B16 tumor cell challenge (Fig. 8). Tumor rejection was shown to be partially dependent on CD4þ T cells, with NK cells providing weak natural immunity against B16 melanoma (Takechi et al., 1996). Further investigation of this mechanism showed that TA99 conferred no protection against B16 lung metastases in FcRdeficient mice; however, it prevented tumors from growing in FcRII-deficient mice (Clynes et al., 1998, 2000). Because FcRII is an inhibitory receptor present on monocytes and macrophages, not on NK cells, the antitumor effect of TA99 is presumably mediated by antibodies through activation of Fc receptors on macrophages, which we have confirmed in adoptive transfer experiments (Clynes et al., 2000; Trcka et al., 2002; Wolchok et al., 2001). Autoimmune depigmentation also occurred in C57BL/6 black mice that were passively immunized with TA99 antibody. Compared with tumor rejection, a fivefold larger amount of antibody was required to achieve hypopigmentation, and autoimmune melanocyte destruction was not found to be NK or CD4þ T cell–dependent (Hara et al., 1995). Although TA99 induced depigmentation in mice deficient in FcR or the C3 component of complement, mice lacking both FcR and C3 did not develop vitiligo. Interestingly, replacement of either C3 or a macrophage population with functional FcR

B

UPC10

TA99

0

UPC10

Day 7

0

50

Day 4

50

100

Day 2

100

150

Day 0

150

No. surface Metastatic Lung Nodules

No. surface Metastatic Lung Nodules

A

TA99

Fig. 8 Adoptive transfer of monoclonal antibody anti-mgp75 conferred protection against B16 tumor challenge. (A) Following adoptive transfer with either the monoclonal antibody TA99 or an isotype-matched, irrelevant monoclonal antibody, UPC10, C57BL/6 mice were intravenously challenged with B16 melanoma. (B) The number of surface lung metastases decreased in a dose dependent manner in mice that were transferred with TA99. Tumor protection was expressed as the number of surface metastatic lung nodules. Reproduced from the Journal of Experimental Medicine, 1995, vol. 182, p. 1611 (Hara et al., 1995), with copyright permission from The Rockefeller University Press.

172

Jose´ A. Guevara-Patin˜o et al.

reconstituted the autoimmune response (Trcka et al., 2002). These findings both demonstrate that complement and activated tissue macrophages are involved in two redundant pathways of autoimmunity against gp75 and confirm that autoimmunity and tumor immunity to gp75 proceed by different mechanisms.

V. CLINICAL APPLICATIONS A. Heteroclitic Peptides For many decades, various strategies for cancer immunization have been proposed and tested in clinical trials. Although many studies have immunized patients with dead or attenuated tumor cells (Berd et al., 1991; Bystryn et al., 1992), this method does not allow an accurate quantification of the immune response to specific antigens. Peptide vaccines have more recently been used to immunize patients with melanomas against specified epitopes on tumor cells (Brichard et al., 1993). An important discovery, which helped improve the strategies for peptide immunization, was that specific peptides derived from differentiation antigens are presented to CD8þ T cells by MHC class I molecules on melanoma cells. Although peptides are relatively simple to synthesize, they elicit immune reactions only when injected together with an immunologic adjuvant. Lewis et al. (2000) evaluated the immune response against the tyrosinase peptide 368-376 (YMD) in HLA-A2.1-positive healthy individuals and patients with melanoma. The peptide was administered to patients (not healthy individuals) in combination with the saponin immune adjuvant QS-21, and the immune responses reflected by CD8þ T-cell frequencies were measured in IFN- ELISPOT assays. The frequency of CD8þ T cells recognizing the tyrosinase peptide was significantly augmented in two of the nine immunized patients with metastatic melanoma following immunization: fourfold higher in one patient, and sevenfold higher in the other patient. In addition, these two patients survived beyond 2 years with stage IV (distant) metastases. It is important to note that the frequency of CD8þ T cells recognizing the tyrosinase peptide in healthy individuals was similar to that seen in the patients with melanoma before immunization. This observation supports the hypothesis that the CD8þ T cells against tyrosinase were part of the immune repertoire and were not induced by the presence of the tumor. This clinical trial also demonstrated that an expansion of CD8þ T cells specific for a differentiation antigen can be generated in vivo by immunization with peptide plus the immunologic adjuvant QS-21, at least in a subset of patients with melanoma.

Inducing Immuniy to Melanoma

173

Another strategy for cancer immunization is based on the use of heteroclitic peptides as a method for enhancing the weak immunogenicity of peptide vaccines. Schaed et al. (2002) conducted a randomized clinical trial in which patients were vaccinated with both the tyrosinase 368–376(370D) and gp100 209–217(210M) peptides using one of three different adjuvants: incomplete Freund’s adjuvant (IFA), QS-21, or granulocyte macrophage colony stimulating factor (GM-CSF) (a growth factor for dendritic cells, major antigen-presenting cells) for priming T-cell responses. The gp100 peptide used in this study was modified at amino acid 210 to increase the binding affinity to HLA-A2.1 (Parkhurst et al., 1996) and to enhance induction of T cells against the native peptide expressed by melanoma. Significantly increased numbers of CD8þ T cells against tyrosinase 370D were found in four of nine (44%) and four of eight (50%) patients immunized using QS-21 and GM-CSF, respectively, compared with zero of nine patients immunized using IFA. Therefore, this study shows that QS-21 and GM-CSF were superior to IFA as adjuvants for peptide vaccines. Overall, clinical trials with peptide vaccines confirm that self peptides are characterized by low immunogenicity, which is an obstacle in the development of effective cancer immunotherapies. In addition, use of peptide vaccines is also restricted to those patients with the MHC haplotype able to bind the peptides used for immunization. Specifically, most peptide vaccines have been restricted to HLA-A2.1 patients and although this is a common haplotype in patients with melanoma, more than 50% of the patients remain ineligible.

B. Xenogeneic DNA Vaccines A promising alternative approach to whole-protein or peptide immunization is vaccination with DNA encoding the antigens. Preparation of DNA vaccines involves fewer manufacturing challenges than full-length proteins, and DNA is relatively simpler to purify in large amounts compared with proteins. Furthermore, this strategy bypasses MHC restriction, allows the presentation of multiple potential epitopes, and does not require injection in combination with an immunologic adjuvant. The injected full-length cDNA is transcribed and translated, resulting in the production of the whole protein with all its potential epitopes, thus eliminating the haplotype limitation. In addition, bacterial plasmid DNA contains unmethylated CpG motifs that are immunostimulatory and have an adjuvant effect comparable to that of complete Freund’s adjuvant (Hemmi et al., 2000; Wagner, 2001). Studies in mice demonstrated that syngeneic immunization with gp75 does not induce an immune response or tumor immunity, whereas immunization with xenogeneic DNA broke tolerance to gp75. On the basis of this observation, a phase I clinical trial with human tyrosinase DNA was

174

Jose´ A. Guevara-Patin˜o et al.

performed in dogs with melanoma (Bergman et al., 2003). Of the nine treated dogs, three had either stable disease or remained with no evidence of disease, and one had complete regression of lung metastases that has persisted for 12 months. Whereas the usual survival of dogs with mucosal or metastatic melanoma is less than 80 days, the median survival of dogs immunized with tyrosinase DNA was 389 days. These encouraging studies performed by our group in collaboration with the Donaldson-Atwood Cancer Clinic at the Animal Medical Center in New York stimulated the development of phase I clinical trials to test the efficacy and safety of xenogeneic immunization in humans. At MSKCC, two separate studies are under way. The first study has a randomized crossover design and evaluates safety and antitumor response in patients with American Joint Committe on Cancer stage IIB, IIC, III, or IV melanoma immunized with mouse (xenogeneic) or human tyrosinase DNA. Alternative schedules of immunization with mouse tyrosinase DNA followed by human DNA, and human DNA followed by mouse DNA, are compared. A second study tests safety and efficacy of mouse gp75 DNA in patients with AJCC stage III and IV melanoma.

C. Heat Shock Protein Peptides Heat shock proteins (HSPs) are a family of highly conserved proteins with essential functions in protein folding and trafficking. They also play an important role as chaperones of antigenic peptides, which are delivered to antigen-presenting cells. HSPs somehow facilitate the entry of peptides into the class I MHC antigen presentation pathway and further enhance the generation of an immune response. Processing and MHC class I presentation of the HSP-associated antigen can occur through either a cytosolic or an endosomal pathway (Castellino et al., 2000). HSP70 derived from tissues and cells can elicit CTL responses against peptides bound to it, depending on the affinity of these peptides for HSP70 (Flynn et al., 1991). Because the affinity of peptides for HSP is highly variable, the number of peptides available to induce a CTL response by HSP immunization is limited. Complexes consisting of hybrid peptides, which contain a high-affinity ligand for the peptide-binding site of HSP70 joined to T-cell epitopes by a glycine–serine–glycine linker, and HSP70 were injected in mice (Moroi et al., 2000). These complexes effectively primed specific CTL responses, which were more potent than the ones produced by T-cell peptide epitopes alone with HSP70. Furthermore, mice immunized with HSP70 and hybrid peptide rejected tumors expressing antigen with greater efficacy than those immunized with peptide epitope plus HSP70. This preclinical study also showed that induction of CTL responses occurred independent of CD4þ T

Inducing Immuniy to Melanoma

175

cells, indicating that immunization directly primed APCs to elicit CD8þ cytotoxic T-cell responses without T-cell help. Javelin (Mojave Therapeutics, Inc., Hawthorne, NY), which is an 8-amino acid motif with high affinity for HSP70 (Musselli et al., 2002), was used in a phase I clinical trial in patients with melanoma at MSKCC. Two fusion peptides containing JavelinE and either tyrosinase 370D or gp100 209M were noncovalently bound to recombinant HSP70, and these complexes were then administered to a group of 27 patients with AJCC stage III and IV melanoma. The immune responses induced by these HSP-peptide complexes were measured by ELISPOT assays in 15 evaluable patients. Preliminary results showed that three out of 11 evaluated patients demonstrated an increased T-cell response to one or more of the tested peptides (tyrosinase 370D, gp100 209M, and the native gp100 sequence 209). Although preliminary, these results support the use of HSP–peptide complexes in priming the immune system to fight tumors. Thus, future clinical trials will evaluate the safety and efficacy of this new immunotherapeutic strategy for cancer.

ACKNOWLEDGMENTS The authors wish to thank Swim Across America, Mr. William H. Goodwin, Mrs. Alice Goodwin, and the Commonwealth Cancer Foundation for Research and Experimental Therapeutics Cancer Center of MSKCC, the Quentin J. Kennedy Family Foundation, T. J. Martell Foundation, and grants from the National Cancer Institute. M. J. T. was also supported by the Sloan-Kettering Institute ‘‘Immunology Tanining Grant’’ and J. D. W. by the New York City Council Speaker’s Fund for Biomedical Research and the Etta Weinheim Fund.

REFERENCES Agarwala, S. A., and Kirkwood, J. (2002). Melanoma: Epidemiology. In ‘‘Encyclopedia of Cancer’’ (J. R. Bertino and P. Bort, eds.), pp. 153–166. Academic Press, San Diego. Bakker, A. B., Schreurs, M. W., de, B. A., Kawakami, Y., Rosenberg, S. A., Adema, G. J., and Figdor, C. G. (1994). J. Exp. Med. 179, 1005–1009. Berd, D., Murphy, G., Maguire, H. C., Jr., and Mastrangelo, M. J. (1991). Cancer Res. 51, 2731–2734. Bergman, P. J., McKnight, J., Novosad, A., Charney, S., Farrelly, J., Craft, D., Sadelain, M., Wulderk, M., Jeffers, Y., Hohenhaus, A. E., Segal, N., Gregor, P. D., Engelhorn, M. E., Riviere, I., and Houghton, A. N. (2003). Clin. Cancer Res. 9, 1284–1290. Boel, P., Wildmann, C., Sensi, M. L., Brasseur, R., Renauld, J. C., Coulie, P., Boon, T., and van der Bruggen, P. (1995). Immunity 2, 167–175. Bowne, W. B., Srinivasan, R., Wolchok, J. D., Hawkins, W. G., Blachere, N. E., Dyall, R., Lewis, J. J., and Houghton, A. N. (1999). J. Exp. Med. 190, 1717–1722. Bretscher, P., and Cohn, M. (1970). Science 169, 1042–1049.

176

Jose´ A. Guevara-Patin˜o et al.

Brichard, V., Van Pel, A., Wolfel, T., Wolfel, C., De Plaen, E., Lethe, B., Coulie, P., and Boon, T. (1993). J. Exp. Med. 178, 489–495. Bystryn, J. C., Henn, M., Li, J., and Shroba, S. (1992). Cancer Res. 52, 5948–5953. Castellino, F., Boucher, P. E., Eichelberg, K., Mayhew, M., Rothman, J. E., Houghton, A. N., and Germain, R. N. (2000). J. Exp. Med. 191, 1957–1964. Chaux, P., Luiten, R., Demotte, N., Vantomme, V., Stroobant, V., Traversari, C., Russo, V., Schultz, E., Cornelis, G. R., Boon, T., and van der Bruggen, P. (1999a). Immunol. 163, 2928–2936. Chaux, P., Vantomme, V., Stroobant, V., Thielemans, K., Corthals, J, Luiten, R., Eggermont, A. M., Boon, T., and van der Bruggen, P. (1999b). J. Exp. Med. 189, 767–778. Chen, J. L., Dunbar, P. R., Gileadi, U., Jager, E., Gnjatic, S., Nagata, Y., Stockert, E., Panicali, D. L., Chen, Y. T., Knuth, A., Old, L. J., and Cerundolo, V. (2000). J. Immunol. 165, 948–955. Clynes, R., Takechi, Y., Moroi, Y., Houghton, A., and Ravetch, J. V. (1998). Proc. Natl. Acad. Sci. USA 95, 652–656. Clynes, R. A., Towers, T. L., Presta, L. G., and Ravetch, J. V. (2000). Nature Med. 6, 443–446. De Backer, O., Arden, K. C., Boretti, M., Vantomme, V., De Smet, C., Czekay, S., Viars, C. S., De Plaen, E., Brasseur, F., Chomez, P., van den Eynde, B., Boon, T., and van der Bruggen, P. (1999). Cancer Res. 59, 3157–3165. Dyall, R., Vasovic, L. V., Molano, A., and Nikolic-Zugic, J. (1995). Int. Immunol. 7, 1205–1212. Dyall, R., Fremont, D. H., Jameson, S. C., and Nikolic-Zugic, J. (1996). J. Exp. Med. 184, 253–258. Dyall, R., Bowne, W. B., Weber, L. W., LeMaoult, J., Szabo, P., Moroi, Y., Piskun, G., Lewis, J. J., Houghton, A. N., and Nikolic-Zugic, J. (1998). J. Exp. Med. 188, 1553–1561. Evavold, B. D., Sloan-Lancaster, J., and Allen, P. M. (1993). Immunol. Today 14, 602–609. Flynn, G. C., Pohl, J., Flocco, M. T., and Rothman, J. E. (1991). Nature 353, 726–730. Gold, J. S., Ferrone, C. R., Guevara-Patino, J. A., Hawkins, W. G., Dyall, R., Engelhorn, M. E., Wolchok, J. D., Lewis, J. J., and Houghton, A. N. (2003). J. Immunol. 170, 5188–5194. Hara, I., Takechi, Y., and Houghton, A. N. (1995). J. Exp. Med. 182, 1609–1614. Hawkins, W. G., Gold, J. S., Dyall, R., Wolchok, J. D., Hoos, A., Bowne, W. B., Srinivasan, R., Houghton, A. N., and Lewis, J. J. (2000). Surgery 128, 273–280. Hawkins, W. G., Gold, J. S., Blachere, N. E., Bowne, W. B., Hoos, A., Lewis, J. J., and Houghton, A. N. (2002). J. Surg. Res. 102, 137–143. Hemmi, H., Takeuchi, O., Kawai, T., Kaisho, T., Sato, S., Sanjo, H., Matsumoto, M., Hoshino, K., Wagner, H., Takeda, K., and Akira, S. (2000). Nature 408, 740–745. Houghton, A. N., Eisinger, M., Albino, A. P., Cairncross, J. G., and Old, L. J. (1982). J. Exp. Med. 156, 1755–1766. Houghton, A. N. (1994). J. Exp. Med. 180, 1–4. Jager, E., Chen, Y. T., Drijfhout, J. W., Karbach, J., Ringhoffer, M., Jager, D., Arand, M., Wada, H., Noguchi, Y., Stockert, E., Old, L. J., and Knuth, A. (1998). J. Exp. Med. 187, 265–270. Jager, E., Jager, D., Karbach, J., Chen, Y. T., Ritter, G., Nagata, Y., Gnjatic, S., Stockert, E., Arand, M., Old, L. J., and Knuth, A. (2000). J. Exp. Med. 191, 625–630. Kawakami, Y., Eliyahh, S., Delgado, C. H., Robbins, P. F., Sakaguchi, K., Appella, E., Yannelli, J. R., Adema, G. J., Miki, T., and Rosenberg, S. A. (1994). Proc. Natl. Acad. Sci. USA 91, 6458–6462. Lewis, J. J., Janetzki, S., Schaed, S., Panageas, K. S., Wang, S., Williams, L., Meyers, M., Butterworth, L., Livingston, P. O., Chapman, P. B., and Houghton, A. N. (2000). Int. J. Cancer 87, 391–398. Lurquin, C., Pel, A. V., Mariame, B., Plaen, E. D., Szikora, J., Janssens, C., Reddehase, M. J., Lejeune, J., and Boon, T. (1989). Cell 58, 293–303.

Inducing Immuniy to Melanoma

177

Moreau-Aubry, A., Le Guiner, S., Labarriere, N., Gesnel, M. C., Jotereau, F., and Breathnach, R. (2000). J. Exp. Med 191, 1617–1624. Moroi, Y., Mayhew, M., Trcka, J., Hoe, M. H., Takechi, Y., Hartl, F. U., Rothman, J. E., and Houghton, A. N. (2000). Proc. Natl. Acad. Sci. USA 97, 3485–3490. Musselli, C. M., Mayhew, M., Warren, N., Reid, M., Houghton, A. N., and Livingston, P. O. (2002). Javelin (TM)-fusion peptides bound to HSP-70: A novel vaccine for melanoma [abstract]. AACR Annual Meeting, San Francisco, CA.. Naftzger, C., Takechi, Y., Kohda, H., Hara, I., Vijayasaradhi, S., and Houghton, A. N. (1996). Proc. Natl. Acad. Sci. USA 93, 14809–14814. Nossal, G. J. (1994). Cell 76, 229–239. Parkhurst, M. R., Salgaller, M. L., Southwood, S., Robbins, P. F., Sette, A., Rosenberg, S. A., and Kawakami, Y. (1996). J. Immunol. 157, 2539–2548. Schaed, S. G., Klimek, V. M., Panageas, K. S., Musselli, C. M., Butterworth, L., Hwu, W. J., Livingston, P. O., Williams, L. L., Lewis, J. J., Houghton, A. N., and Chapman, P. B. (2002). Clin. Cancer Res. 8, 967–972. Takechi, Y., Hara, I., Naftzger, C., Xu, Y., and Houghton, A. N. (1996). Clin. Cancer Res. 2, 1837–1842. Trcka, J., Moroi, Y., Clynes, R. A., Goldberg, S. M., Bergtold, A., Perales, M. A., Ma, M., Ferrone, C. R., Carroll, M. C., Ravetch, J. V., and Houghton, A. N. (2002). Immunity 16, 861–868. Van den Eynde, B. J., and van der Bruggen, P. (1997). Curr. Opin. Immunol. 9, 684–693. Van den Eynde, B. J., Gaugler, B., Probst-Kepper, M., Michaux, L., Devuyst, O., Lorge, F., Weynants, P., and Boon, T. (1999). J. Exp. Med. 190, 1793–1800. Vijayasaradhi, S., Bouchard, B., and Houghton, A. N. (1990). J. Exp. Med. 171, 1375–1380. Vijayasaradhi, S., Doskoch, P. M., and Houghton, A. N. (1991). Exp. Cell Res. 196, 233–240. Vijayasaradhi, S., and Houghton, A. N. (1991). Int. J. Cancer. 47, 298–303. Wagner, H. (2001). Immunity 14, 499–502. Weber, L. W., Bowne, W. B., Wolchok, J. D., Srinivasan, R., Qin, J., Moroi, Y., Clynes, R., Song, P., Lewis, J. J., and Houghton, A. N. (1998). J. Clin. Invest. 102, 1258–1264. Wolchok, J. D., Srinivasan, R., Perales, M.-A., Houghton, A. N., Bowne, W., and Blachere, N. (2001). Cancer Immunity. 1, 9–19.

This Page Intentionally Left Blank

Chemical Carcinogens as Foreign Bodies and Some Pitfalls Regarding Cancer Immune Surveillance Thomas Blankenstein1,2 and Zhihai Qin2 1 Max-Delbru¨ck-Centrum for Molecular Medicine, 13092 Berlin, Germany 2

Institute of Immunology, Free University Berlin, 12200 Berlin, Germany

I. II. III. IV.

Introduction Chemical Carcinogenesis Increased Tumor Incidence by MCA in IFNR-KO Compared to IFNR-WT Mice No New Evidence that Supports T Cell-Mediated Immune Surveillance of MCA-Induced or Spontaneous Tumors A. Cancer Immune Surveillance B. Some Current Knowledge of Cancer Biology, Immune Regulation and Inflammation that is Difficult to Reconcile with Immune Surveillance C. Is Adaptive Tumor Immunity Involved in Control of MCA-Induced Tumors? V. The Protective Response in IFNR-WT Mice is Associated with Encapsulation of MCA A. Tissue Damage, Tissue Repair Response and Fibrosis at the Site of MCA B. Local Production of IFN Leads to Encapsulation of Transplanted Tumor Cells C. MCA-Diffusion and DNA-Damage is Inhibited in an IFNR-Dependent Fashion D. Encapsulated MCA Persists in Long-Term Tumor-Free Mice VI. Concluding Remarks References

Interferon--receptor (IFNR)-deficient mice are more susceptible to tumor induction by methylcholanthrene (MCA) in comparison to control littermates. The cellular source of IFN is not known, but the absence of T cells does not significantly increase the incidence of MCA-induced tumors. However, it appears that the presence of T cells in combination with unknown, perhaps environmental, factors can decrease MCAinduced tumor incidence, indicating that IFN of unknown origin contributes to the protective response. The current knowledge of cancer biology, immune regulation, and tumor-promoting effects of inflammation are difficult to reconcile with the concept of immune surveillance against non-virus-associated cancer. Analysis of the primary MCAtreated mouse indicates, as one protective mechanism, a tissue repair response against MCA-induced damage, in the course of which MCA is encapsulated and persists for long time in tumor-free mice, termed foreign-body reaction. The protection from DNA damage could simultaneously diminish tissue injury and malignant transformation. We argue that inhibition of MCA-induced carcinogenesis is mechanistically different from tumor transplantation immunity and that a longer latency in MCAtreated mice is unlikely due to T cell-mediated tumor recognition and selection of Advances in CANCER RESEARCH 0065-230X/03 $35.00

179

Copyright 2003, Elsevier Inc. All rights reserved

180

Thomas Blankenstein and Zhihai Qin

less immunogenic variants. We discuss that the IFNR-dependent mechanism against MCA is unrelated to the original concept of T cell-mediated immune surveillance and that the increased spontaneous tumor incidence observed in some immune-deficient mice is likely to be explained by opportunistic infection and tumor-promoting chronic inflammation. ß 2003 Elsevier Inc.

I. INTRODUCTION This review deals with complex mouse tumor models and the dominant hypothesis in tumor immunology—immune surveillance—both together a complicated issue. The experimental tumor models have become more and more sophisticated and better. Nevertheless, probably the two most important notions related to immune surveillance, based on primary tumors induced by chemical carcinogens or oncogenic viruses in normal or spontaneous immune-deficient mice, date back 25 years: that the absence of T cells has no effect on the incidence of spontaneous or chemically induced tumors (Stutman, 1975), and that, in contrast to the evidence from spontaneous tumors, at least some virus-induced tumors are under effective immune surveillance (Klein and Klein, 1977). The availability of mouse inbred lines and improved cell culture conditions allowed tumor transplantation experiments. These much faster experiments brought important discoveries, for example, the demonstration of tumor-specific transplantation antigens and a large number of possible tumor-immune cell interactions, and they certainly should have a firm place in tumor immunology in the future. However, tumor transplantation experiments have also widely been overinterpreted in so-called proof-of-concept experiments to test immune therapy strategies against primary cancer. Transplantable tumor models cannot test adequately the concept of immune surveillance, as pointed out by Schreiber (1993). Spontaneous tumors are too rare in mice, and, as their potential tumor antigens are not known, whether the clonal tumor evolution is influenced by T cells cannot be analyzed. Transgenic mice bearing an oncogene expressed by a tissue-specific promotor have delivered important information about the mechanism of malignant transformation; however, as often the whole organ is transformed and the mice have usually developed at least some form of tolerance for the oncogene as target antigen, they are not suitable to prove or disprove immune surveillance. Knock-out (KO) mice belong to the most important additives to immunological research of the last 10 years; large numbers of mice with deficiency in defined immune genes are already available. A large part of the results discussed in this review deals with immune-deficient KO mice. When working with immune-deficient mice, regardless of whether they were generated by KO technology or arose spontaneously, one has to keep in mind that developmental defects are difficult to

Chemical Carcinogens as Foreign Bodies

181

exclude and that the lifelong immune suppression can lead to secondary phenomena induced by environmental factors. We discuss that this can be problematic with regard to interpretation of tumor development. KO mice have clearly revealed, however, one immune component—IFN—for inhibition of MCA carcinogenesis. Future studies based on somatic oncogene activation by the Cre-Lox recombination system will perhaps allow to better analyze the role of T cells during spontaneous tumor development.

II. CHEMICAL CARCINOGENESIS Chemical carcinogens, a large number of which have been described, contribute to cancer in humans (Fraumeni et al., 1993; Yupsa and Shields, 2001). One class are polycyclic aromatic hydrocarbons (PAHs) to which MCA and dimethylbenz()anthracene (DMBA) belong. PAHs are contained, for example, in industrial pollution, fossil fuels, tobacco smoke, and diet. They are genotoxic, meaning that they cause DNA damage. The efficacy of chemical carcinogens to induce tumors in mice correlates with DNA-adduct formation that is thought to be the basis for introducing mutations in genes. The accumulation of a series of somatic mutations in specific genes is required for malignant transformation. Some investigators estimate up to 10 mutational hits that have to occur for a tumor to develop and progress (DePinho, 2000). Frequently, mutations in the same genes can be found in different PAH-induced tumors; for example, the ras oncogene or the p53 tumor suppressor genes (Halevy et al., 1991; Watanabe et al., 1999). Because there is no reason to assume that PAHs induce gene locus–specific mutations, it is likely that only cells with the right mutations finally become apparent, whereas the majority of cells with randomly induced mutations in any transformation-irrelevant genes never become apparent. The carcinogenicity of PAHs is concentration-dependent. PAHs are thought to require metabolic activation to be genotoxic; for example, by cytochrome P450 enzymes, a polymorphic multigene family (Pelkonen and Nebert, 1982). However, the processes of toxification and detoxification of chemical carcinogens in vivo are still poorly understood. Two experimental models, full carcinogenesis and two-stage carcinogenesis, are most often employed to analyze chemical carcinogenesis. During two-stage carcinogenesis, the shaved skin of mice is painted with a single dose of DMBA, followed some time later by repeated paintings of the same site of the skin with reagents such as phorbolesters (the most often used is 12-O-tetradecanoylphorbol-13-acetate [TPA]). At certain doses, DMBA will not induce tumors, but only if applied in combination with TPA. TPA or a variety of other treatments such as wounding that can replace TPA and

182

Thomas Blankenstein and Zhihai Qin

that have in common the ability to induce an inflammatory response have not been shown to be mutagenic. In fact, some of the earliest experimental tumor models have employed skin carcinogenesis (Deelman, 1926; MacKenzie and Rous, 1941; Rous and Kidd, 1941). From these studies, the initiation–promotion–progression model emerged. The initial treatment with DMBA or PAH-containing tar that was used before purified PAHs were available is believed to induce first mutations that persist for long periods of time (van Duuren et al., 1975) but the cells remain silent (initiation). During TPA treatment papillomas are induced (promotion). Many papillomas disappear if TPA treatment is stopped, but some will develop into carcinomas (progression). Detailed reviews of two-stage carcinogenesis as a model for tumor-promoting effects of inflammation can be found elsewhere (DiGiovanni, 1992; Iversen, 1995; Schreiber and Rowley, 1999). During full carcinogenesis, a single injection of MCA, for example, intramuscularly or subcutaneously, in an appropriate dose will induce tumors, usually fibrosarcomas at the injection site. Mice susceptible to MCA can be resistant to DMBA/TPA and vice versa, indicating different mechanisms leading to tumor development (DiGiovanni, 1992). For example, 129SvEv mice are relatively resistant to MCA but susceptible to DMBA/TPA (unpublished observation). Different strains of mice vary dramatically in susceptibility to MCA. For example, C57Bl/6 and BALB/c mice injected with 800 g MCA mixed in sesame oil (MCA is water-insoluble) develop tumors in 100% of the cases until around week 20, whereas 129SvEv mice treated in the same way developed tumors after a much longer latency period, and a significant number of mice remained tumor-free (Fig. 1). The reason for these strain-specific differences is not known. Recently, it became apparent that tumor incidence by the same dose of MCA varied substantially between the same inbred strain of mice in experiments by different investigators. For example, 100 g MCA induced tumors in 80%–100% of C57Bl/6 mice in some studies (Takeda et al., 2002; van den Broek et al., 1996; unpublished data), but below 40% (Smyth et al., 2000, 2001; Street et al., 2001) or only 20% tumors in others (Cretney et al., 2002). After injection of 25 g MCA into C57Bl/6 mice, tumor incidence varied from 0% (Takeda et al., 2002), 20% (van den Broek et al., 1996), around 25% (Smyth et al., 2000, 2001; Street et al., 2001) and between 80% and 100% ( unpublished observation). Similarly, 100 g MCA induced tumors in 80%–100% BALB/c mice in some studies (Noguchi et al., 1996; Qin et al., 2002), but only below 40% in others (Smyth et al., 2001). The observation time in most of these studies was 25–30 weeks. Again, the reason for these discrepancies is not known, but genetic variation or, more important, environmental factors could be responsible, as discussed below. This becomes obvious by the observation showing that rats injected intraperitoneally with an endogenous nontransforming rat retrovirus develop significantly fewer

183

Chemical Carcinogens as Foreign Bodies

100% 129/Sv/Ev (IFN R+/−) 129/SV/Ev (IFN R−/−)

Tumor free mice

80%

BALB/c C57Bl/6

60% 40% 20% 0% 0

10

20

30

40 Weeks

50

60

70

80

Fig. 1 Interferon- receptor (IFNR) expression and other genetic factors determine susceptibility to methylcholenthrene (MCA). Female IFNR-competent C57Bl/6 (open triangles), BALB/c (open squares), 129Sv/Ev (open circle) mice and IFNR-deficient 129Sv/Ev (closed circles) were intramuscularly injected with 800 g MCA resolved in 0.1 mL sesame oil, and tumor development was observed. The IFNR-competent 129Sv/Ev mice were heterozygous control littermates. The two 129Sv/Ev congenic lines were injected in parallel; C57B1/6 and BALB/c data are from different experiments. As can be seen, IFNR responsiveness decreases tumor incidence when the two 129Sv/Ev congenic lines are compared. Comparison of the three different IFNR-competent inbred lines shows that other unknown genetic factors are of at least equal importance in determining susceptibility to MCA. The data with 129Sv/Ev mice are reproduced with permission from Qin et al. (2002).

tumors after subsequent subcutaneous injection of MCA in comparison to rats that did not receive the retrovirus (Fish et al., 1981). The above– mentioned intrastrain differences are problematic if gene-deficient mice on a given genetic background are compared in susceptibility to MCA to control mice of the same genetic background that are not directly matched; for example, by the use of control littermates.

III. INCREASED TUMOR INCIDENCE BY MCA IN IFNgR-KO COMPARED WITH IFNgR-WT MICE Starting in 1995, we observed that mice with an experimentally introduced genetic defect in the IFNR, termed IFNR-KO mice (Huang et al., 1993), developed faster and more frequent tumors after injection of MCA

184

Thomas Blankenstein and Zhihai Qin

than did a control group of mice that could express the IFNR, termed IFNR wild-type (WT) mice (Fig. 1; Qin et al., 2002). Mice were of the MCA-resistant 129Sv/Ev strain. The control group consisted of control littermates, either IFNR-heterozygous or homozygous-WT mice that showed similar resistance to MCA (Qin et al., 2002). The observation that 129Sv/Ev IFNR-KO mice developed tumors after a longer latency than the (IFNR-WT) BALB/c and C57Bl/6 mice injected with the same dose of MCA (Fig. 1) indicates that susceptibility or resistance to MCA is determined by multiple factors, one of which is IFNR. C57Bl/6 IFN-KO mice are also more susceptible to MCA compared to control mice. This reveals, however, at lower amounts of MCA (unpublished observation). The IFNR is a cytokine receptor expressed by most cells of the body. There is one known ligand—IFN—a key cytokine during many immune responses and produced mainly by different T cell subsets (CD4þ and CD8þ T cells, and T-cell-receptorþ T cells, NKT cells, or natural killer (NK) cells (Boehm et al., 1997). However, other cell types such as B cells or macrophages are also able to produce IFN under certain experimental conditions (Harris et al., 2000; Ibe et al., 2001; Munder et al., 1998). Usually, cells must be activated to express IFN. There can be many activating stimuli; pathogens such as bacteria and viruses are probably the most typical or physiologic ones. IFN has many biological effects (pleiotrophic cytokine) depending on the milieu and the cell type on which it acts. Because a signal transmitted into cells by the IFNR requires engagement with its ligand, one has to assume that IFN must be available to protect from MCA-induced carcinogenesis in IFNR-WT mice. However, the question of the source of IFN that would give insight into the protective mechanism against MCA-induced tumors is controversially discussed. Suffice to say here that the source of IFN is not yet known, nor is it known where, when, and in response to which stimulus IFN is produced. It is important to note that IFNR-KO mice do not have increased susceptibility to tumor induction by DMBA/TPA compared with IFNR-WT mice (unpublished observation). Furthermore, the absence of the IFNR in mice additionally defective in one or both copies of the p53 tumor suppressor gene does not change tumor incidence, latency, or spectrum in comparison with IFNRcompetent mice (Qin et al., 2002). At the time when we observed increased susceptibility to MCA-induced carcinogenesis in IFNR-KO mice, we had two alternatives. We could analyze the mechanism that controls MCA carcinogenesis by tumor transplantation experiments in the hope that rejection of transplanted tumor cells involves the same mechanism that also controls MCA-induced tumor development or we could analyze the local events at the site of MCA. It is well established that rejection of transplanted tumor cells occurs in an antigen-specific fashion and requires T cells, CD4þ, and/or CD8þ; their timely activation; rejection antigens of MCA fibrosarcomas

Chemical Carcinogens as Foreign Bodies

185

that they usually express and that are usually unique for each tumor (Schreiber, 1993); and in most models, IFN/IFNR (Blankenstein and Qin, 2003). Meanwhile, it is also known that IFN-producing T cells, again either CD4þ or CD8þ, if they arrive at the tumor site rapidly after tumor cell injection (e.g., because the mice were previously immunized with irradiated cells of that tumor), often inhibit tumor-induced angiogenesis by IFN acting on non–bone marrow–derived tumor stroma cells. In essence, T cells in an antigen-specific and MHC-restricted fashion prevent establishing the tumor stroma. We argued that this mechanism prevents rapid tumor burden of the transplanted cells, giving direct killing mechanisms the chance to eliminate residual tumor cells (Blankenstein and Qin, 2003; Qin and Blankenstein, 2000; Qin et al., 2003; Schu¨ler and Blankenstein, 2003). Because we had no indication that MCA carcinogenesis is controlled by a similar mechanism, we analyzed the local events at the site of injected MCA and, in parallel, asked whether MCA-induced tumor development is also increased in other immune-deficient mice. While these experiments were and still are in progress, a series of studies was published that described increased tumor incidence induced by MCA or occurring spontaneously in KO mice with different immune deficiencies. Because many of these KO mice lacked one or more different T-cell subsets or, in general, cell types that can produce IFN, it was concluded from tumor transplantation experiments that such cells, as they produce IFN only on stimulation, must be active and locally present participants in the protective response against MCA-induced tumor development (Dunn et al., 2002). This assumption let to a revival of the hypothesis of immune surveillance as a general mechanism against chemical carcinogen-induced and spontaneous (non–virus associated) cancer. Therefore, we will first critically discuss whether this revival is justified and then will describe at least one mechanism that we think contributes to the protective response against MCA-induced tumor development.

IV. NO NEW EVIDENCE THAT SUPPORTS T-CELL-MEDIATED IMMUNE SURVEILLANCE OF MCA-INDUCED OR SPONTANEOUS TUMORS A. Cancer Immune Surveillance First proposed by Ehrlich (1909), the concept of immune surveillance against cancer was developed half a century ago by Thomas (1959) and, in particular, Burnet (1964, 1970, 1971, 1973, 1978). At the time when Burnet proposed his theory, the existence but no function of T cells (thymus-dependent cells) was known. Being aware of the fact that malignant

186

Thomas Blankenstein and Zhihai Qin

transformation requires somatic mutations in genes that potentially create new antigen epitopes and appear on the surface of the tumor cell, Burnet thought that circulating T cells recognize the foreign (as opposed to self) structure on the frequently arising tumor cells and eliminate them. Burnet believed that T cells evolved because cancer is contagious and that the histocompatibility barrier prevents horizontal cancer cell transmission; for example, from the mother to the newborn, not yet immune-competent, offspring (Burnet, 1964). Today, there is little doubt that the T-cell system evolved to protect against the universe of pathogens; for example, from bacterial and viral infections. When Stutman (1974, 1979) showed that nude mice, which do not have a functional thymus and lack thymus-dependent T cells because of a spontaneous mutation, do not develop tumors, either spontaneous or MCA-induced, faster or more frequently than their control littermates, immune surveillance as a broadly applicable mechanism was seen critically (Rygaard and Povlsen, 1976; Stutman, 1975). Not surprisingly, nude mice are, however, very susceptible to viral infections, and there is convincing evidence that immune surveillance, the spontaneous and successful control of tumors, is effective at least against some virus-associated tumors (Klein and Klein, 1977).

B. Some Current Knowledge of Cancer Biology, Immune Regulation, and Inflammation that is Difficult to Reconcile with Immune Surveillance In the last 50 years, our knowledge of cancer biology, the mechanisms, how immune responses are induced and regulated, and the role of inflammation for tumor development has dramatically increased. We briefly mention only a few examples that are relevant to the question of cancer immune surveillance.

1. CELL-INTRINSIC CONTROL MECHANISMS A large number of genes, termed tumor suppressor genes, have been detected whose products control the cell cycle, ensure that DNA damage is repaired before the cell divides, or in general, maintain genomic integrity or induce apoptosis of damaged cells (Vogelstein and Kinzler, 1992). After a limited number of cell divisions, most cells appear to become senescent and, at least in vitro, cannot further divide, except they have accumulated already inactivating mutations in critical tumor suppressor genes like p53, retinoblastoma, INK4a and so forth. Usually, both alleles of a tumor suppressor gene must be inactivated, for example, by gene loss, inactivating point mutations, or aberrant methylation patterns, indicating that the probability of passing senescence checkpoints is exceedingly low.

Chemical Carcinogens as Foreign Bodies

187

2. PROBABILITY OF MALIGNANT TRANSFORMATION The mutation frequency in vitro has been estimated to 2 10 7 per gene per cell division. Assuming that between four and 10 rate-limiting, stochastic events are necessary for tumor initiation and progression, it has been argued that the spontaneous mutation rate cannot explain the frequent occurrence of malignant transformation (DePinho, 2000). Therefore, mutations in genes whose products maintain genome stability are necessary for a dramatic increase of the mutation rate, termed ‘‘mutator phenotype’’ (Lengauer et al., 1998; Loeb, 1991). The multiple hurdles that a normal cell has to overcome to grow as invasive carcinoma have been comprehensibly reviewed by Hanahan and Weinberg (2000).

3. CANCER IS A DISEASE OF AGE The frequency of invasive cancer increases at the age of 40–59 years and further increases dramatically at the age of 60–79 years (DePinho, 2000). At least the age between 40–59 years is well before the function of the immune system declines. The immune surveillance hypothesis was based on the assumption that tumor cells occur continuously throughout life, but that later in life immune function decreases and tumors are not efficiently eliminated anymore. Beyond the age of 80 years when immune function most likely can be expected to decline, the incidence of cancer does not further increase (DePinho, 2000). In recent years, it was shown that the frequency of somatic mutations increases with age in mice and humans, as does the spectrum of somatic mutations, at least in some organs that have been analyzed (Dolle et al., 2000; Martin et al., 1996). Very informative is a group of rare genetic diseases, termed progeroid syndromes (Martin and Oshima, 2000). Individuals with such a predisposition show the typical features of senescence. Werner syndrome patients, for example, usually die before the age of 50 from myocardial infarction as a result of artheriosclerosis (also a disease of age) or cancer. The defective gene, a helicase, has been implicated in DNA replication, DNA repair, transcription, and recombination (Martin and Oshima, 2000). Obviously, most forms of cancer correlate better with aging and increased frequency of mutations than with immune status.

4. TUMOR SPECTRUM IN YOUNG AND OLD INDIVIDUALS DIFFERS In young individuals, epithelial carcinomas are rare (9% of all malignancies) and other types, for example, lymphomas, leukemias and sarcomas, are prevalent (DePinho, 2000). In contrast, in adult individuals, the majority

188

Thomas Blankenstein and Zhihai Qin

of malignancies are epithelial carcinomas (83.6%). While it is difficult to see how the immune system should control early in life predominantly epithelial carcinomas but later in life predominantly lymphomas, leukemia, and sarcomas, there is a striking correlation between telomere attrition, (re)activation of telomerase, and genomic instability in epithelial carcinomas (DePinho, 2000). During replication, chromosome ends (telomeres) are shortened in the absence of telomerase (a specialized reverse transcriptase maintaining telomere length and function) leading to cellular senescence after a certain number of cell divisions. Telomerase-KO mice that are heterozygous for mutant p53 have a high incidence of developing epithelial carcinomas (Artandi et al., 2000). p53-heterozygous control mice with intact telomere function develop mainly lymphomas and sarcomas but rarely epithelial carcinomas. These interesting observations indicate that the vast majority of human cancers—epithelial carcinomas—are primarily the result of aging and a decrease in the function of cell-intrinsic control mechanisms that maintain genomic integrity.

5. HOW, WHERE, AND WHEN ARE T CELLS ACTIVATED? Naive antigen-inexperienced T cells usually circulate in the blood and lymphoid system and do not have the requirements to migrate into peripheral tissues. The antigen has to be brought to the T cells—the T cells do not come to the antigen. This task is best performed by dendritic cells (DCs)— specialized antigen presenting cells. They reside in peripheral tissues, for example, Langerhans cells in the skin, where they collect antigens, migrate to the draining lymph node, and present peptides of the captured and processed antigens by MHC class I and class II molecules to the circulating CD8þ and CD4þ T cells, respectively. If the T cells recognize the antigen specifically, they remain in contact with the DCs and become activated (Banchereau and Steinman, 1998). However, at that time, DCs must have undergone maturation steps and be activated, for example, they must express certain T-cell costimulatory molecules like those of the B7 family. If naive T cells recognize antigen in the absence of costimulation, they often become anergic instead of being activated. Molecules that induce activation of DCs through so-called pattern recognition receptors (Medzhitov and Janeway, 2000) are often derived from bacteria, viruses, or fungi; for example, lipopolysaccharide, bacterial DNA, or double-stranded viral RNA. It is currently not known which capacity a spontaneously arising tumor has to activate DC. It seems more important, however, to ask when one can assume that a spontaneous tumor, provided it expresses already-immunogenic antigen or antigens in the initial phase of malignant transformation (which is an open question), is recognized by the immune system and, in particular, by T cells. T-cell activation requires a certain

Chemical Carcinogens as Foreign Bodies

189

amount of antigen. It seems likely that, to capture a sufficient amount of antigen by DCs, substantial tumor cell death has to occur. At some points during progression, tumors often contain areas of necrosis. However, we argue that compared with the capacity of the immune system to reject transplanted tumor cells, certainly a much easier task (Willimsky and Blankenstein, 2000), the activation of T cells in primary tumor hosts is a too-late event. To induce tumor immunity against reasonable immunogenic transplanted tumor cells, for example, a MCA-induced fibrosarcoma, it requires a substantial amount of nonreplicating (irradiated or mitomycin C-treated) tumor cells in the absence of any adjuvant (which includes all the little cell culture artifacts like mycoplasm, fetal calf serum, etc.) in the range of 106 tumor cells. If within a primary nontransplanted tumor 106 tumor cells (simultaneously) die, one can assume that the tumor size is already orders of magnitude higher, has established at least some form of stroma, and is vascularized. Transplanted tumors of comparable size are usually not rejected any more; in other words, the tumor sneaks through because the T cells that are needed for tumor rejection are induced—if at all—too late. Even if the tumor expresses strong antigens from the very beginning and if T cells are effectively activated by the primary tumor, the unfortunate relationship between tumor size necessary for providing a sufficient amount of putative antigen and time point of recognition by T cells is problematic to reconcile with the idea of immune surveillance. In this regard, it is interesting to ask what distinguishes virus-associated from non–virus associated tumors. Certainly, evolutionary forces on effective antiviral responses and the involvement and early presence of strong viral antigens favor the control of virus-associated tumors; for example, Epstein-Barr virus–associated lymphomas (Klein and Klein, 1977). However, we like to argue that the relative proportion between antigen amount and tumor size makes an important difference. Whereas for non–virus associated tumors the amount of a given (cellular) antigen is directly proportional to tumor load, for virus-associated tumors this must not be the case; that is, sufficient amount of viral antigen resulting from viral replication is generated to induce T cells when tumor load is still minimal.

6. INFLAMMATION THAT MIGHT BE EXPECTED TO BE INVOLVED IN IMMUNE SURVEILLANCE OFTEN SUPPORTS TUMOR PROGRESSION One could argue that until effector T cells have been generated, innate effector-like NK cells, monocytes/macrophages, and so forth control the initially small tumor. However, we are not aware of any convincing example (perhaps with one exception; see following) that innate effectors are

190

Thomas Blankenstein and Zhihai Qin

actively involved in control of a primary tumor. In contrast, there is strong evidence from a variety of different primary tumor models, two-stage chemical carcinogenesis, and oncogene-transgenic mice that an inflammatory response supports tumor development or, vice versa, that defects in the inflammatory response or defects in products of inflammatory responses impair tumor development (Coussens and Werb, 2001, 2002, Coussens et al., 2000. Lin et al., 2001; Moore et al., 1999; Oshima et al., 1996; Schreiber and Rowley, 1999). Collectively, the above (in part theoretical) considerations make it difficult to find supportive evidence for the existence of immune surveillance against non–virus associated tumors, even though the absence of evidence does not prove the absence of its existence and the almost unlimited variability of tumors makes any firm conclusion impossible at this time. It is more likely, however, that the majority of malignancies, namely, epithelial carcinomas, rarely occur in childhood because of effective cell-intrinsic control mechanisms, instead of being controlled by the immune system. Whether some leukemias represent an exception remains an open question. Some suggestive observations are worth mentioning. A hallmark of chronic myeloid leukemia (CML) is a chromosomal translocation between chromosomes 9 and 22, resulting in expression of the chimeric bcr-abl oncogene. The fusion region creates a potential tumor-specific peptide epitope that can bind to some human HLA molecules (HLA-B8 and HLA-A3). Within patients with CML, HLA-B8 and HLA-A3 alleles are underrepresented (Posthuma et al., 1999). The tumor-specific bcr-abl fusion peptide is presented by CML cells from HLA-A3-positive patients, and antigen-specific cytotoxic T lymphocytes can be detected in such patients (Clark et al., 2001). Furthermore, by reverse transcription polymerase chain reaction bcr-abl transcripts were detected in 22 of 73 healthy adult individuals (Biernaux et al., 1995). Whether these findings have a causal relationship will be important to analyze, as they could demonstrate a clear example for the existence of immune surveillance against a non–virus associated tumor.

C. Is Adaptive Tumor Immunity Involved in Control of MCA-Induced Tumors? Basically, Stutman (1974, 1975) answered this question already in T-celldeficient nude mice. These mice are so immune compromized that they accept normal or tumor allografts and xenografts. In the absence of infection, nude mice have a normal lifespan and do not develop spontaneous tumors more frequently than T-cell-competent control mice. Tumor incidence and latency following MCA injection was not different between nude

Chemical Carcinogens as Foreign Bodies

191

and control CBA/H mice. Outzen et al. (1975) confirmed these results with nude mice on a BALB/c genetic background. The unaltered tumor incidence by MCA in nude mice compared with control mice on the one hand and the inability to reject transplanted tumor cells or develop tumor immunity of nude mice on the other hand also indicated that rejection of transplanted tumor cells is mechanistically unrelated to any (at that time questionable) mechanism inhibiting MCA-induced tumor development. Recently, a series of publications appeared showing that perforin-KO (lacking the major cytolytic mechanism by which T and NK cells kill target cells; Smyth et al., 2000; Street et al., 2001; vanden Broek et al., 1996), IFNR-KO (Kaplan et al., 1998), IFN-KO (Street et al., 2001), Stat1-KO (insensitive to IFN; Kaplan et al., 1998), RAG2-KO (lacking all T, B and NKT cells; Shankaran et al., 2001), severe combined immunodeficiency (SCID) (lacking all T, B, and NKT cells; Smyth et al., 2001), T-cell-receptor (TCR) þ T cell-KO (Girardi et al., 2001), TCR þ T cell-KO (Girardi et al., 2001), and T-cell-receptor J281-KO (lacking a subset of NKT cells; Smyth et al., 2000, 2001) all had an increased tumor incidence following injection of MCA compared with control mice. These data were taken as evidence for the existence of immune surveillance; the discrepancy with the experiments in nude mice was explained by the existence of residual T cells in nude mice (Dunn et al., 2002). However, no clear in vivo function has been shown for the very few T cells in these mice (Hu¨nig and Bevan, 1980; Ikehara et al., 1984; Lawetzky and Hu¨nig, 1988; Maleckar and Sherman, 1987). The T-cell progenitors in the bone marrow of nude mice have in addition a maturation defect (Chatterjea-Matthes et al., 2003), raising doubts that the nude mouse significantly differs from the above-mentioned mice, several of which are less immune-deficient than nude mice, with regard to susceptibility to MCA. Our own experiments show that we cannot confirm significantly increased tumor incidence by MCA in the absence of T cells. RAG1-KO mice that lack all T cells and that phenotypically are indistinguishable from RAG2-KO mice develop MCA-induced tumors with similar incidence compared to control littermates (unpublished observation). The latency period of tumors in RAG1-KO mice is nonsignificantly delayed for a few weeks. This is in the range of variability between two individual experiments. SCID and nude mice develop MCA-induced tumors also with the same incidence and latency. In some but not all experiments, the average tumor latency differed a few weeks between T-cell-competent and T-cell-deficient mice; this was not necessarily linked to the genotype of the mice. Similarly, perforin-KO and perforin-WT mice developed tumors with similar incidence and kinetics in our hands (unpublished observation). On the basis of data from three different T-cell-deficient mice (RAG1-KO, SCID, nude) and mice that lack the major lytic pathway by which T cells and NK cells eliminate target cells (perforin-KO mice), we conclude that the absence of T cells does

192

Thomas Blankenstein and Zhihai Qin

not increase the risk of a mouse injected with MCA of developing a tumor. In contrast to IFNR-KO (Qin et al., 2002) or IFN-KO mice (unpublished observation), mice that lack some but not all cell types able to produce IFN do not have an increased risk of developing MCA-induced tumors. How can these discrepancies be explained? We have no definite answer; however, we noted that one difference between our experiments and those mentioned above appeared to be the type of control animals used. We either used littermates as controls bred in our facility or obtained KO and WT mice together from the same breeder. In most of the studies (in some, sufficient information was not given) demonstrating increased tumor incidence in the above-mentioned KO mice lacking some or all T cells and perforin, the KO mice were bred locally and the WT mice were obtained from another source. Why could it be important to include control littermates? It has been shown that a local T-cell response induces widespread IFN/ IFNR-dependent MHC expression in nonaffected organs (Halloran et al., 1992; Takei et al., 2000). Furthermore, when germ-free mice are transferred to a conventional colony, MHC expression in remote organs is upregulated (Cockfield et al., 1990). Even though specific-pathogen-free colonies were used in most of the above-mentioned experiments with MCA in KO mice, there are likely environmental differences in flora, food, air, and water that could lead to an innate or adaptive response influencing also the IFN levels in these mice. Therefore, the use of nonlittermates and, in particular, the recent transfer of one experimental group (e.g., the WT group) into a new environment poses the risk that IFN is increased in the control mice unrelated to the experimental treatment. In addition, genetic drift in breeding colonies and unknown genetic or epigenetic changes introduced by the embryonic stem cells (Humpherys et al., 2001) are difficult to exclude. A possible explanation for the decreased tumor incidence by MCA in T-cell-competent compared with T-cell-deficient (and perforin-KO) mice in some but not other experiments is that the exposure of WT mice to a new environment increases the steady-state level of IFN that indirectly contributes to the protective response against MCA-induced tumor development. This, then, could also explain why, regardless of which cellular source able to produce IFN (all T cells, TCR þ T cells, TCR þ T cells, NKT cells, or NK cells) is lacking in the respective KO mice, mice have increased tumor incidence compared with such WT mice that are not optimally matched. This assumption is supported by the observation that rats injected with a retrovirus or mice injected with IL-12, both inducers of IFN, are relatively resistant to MCA-induced tumors (Fish et al., 1981; Noguchi et al., 1996). Other reasons for the different results cannot be excluded, but it is apparent that adaptive immunity to the tumor is not involved in inhibition of MCAinduced tumor development, as noted by Stutman, who included control littermates in his experiments (Stutman, 1974). If our explanation turns

Chemical Carcinogens as Foreign Bodies

193

out to be true, the question remains of whether IFN that is clearly involved in the protective response against MCA-induced tumors is completely provided indirectly in response to other stimuli or is also produced in a MCAtreatment-specific fashion, and if yes, by which cells. In this regard, it is interesting to note that the basal constitutive MHC class I expression in nontreated IFN-KO or IFNR-KO mice is reduced compared with control mice, indicating that IFN is constitutively present in low amounts in naive mice (Goes et al., 1995; Takei et al., 2000). Two further arguments have been brought up to support a role of immune surveillance against MCA-induced tumors: First, longer latency of MCAinduced tumors to appear in RAG2-KO compared with RAG2-KO mice was explained by T-cell recognition and selection of less-immunogenic variants, as tumors derived from RAG2-WT mice could be less efficiently transplanted into RAG2-WT mice compared with those derived from RAG2-WT mice (Shankaran et al., 2001). This contrasts experiments in SCID mice. Tumor incidence and kinetics induced by MCA do not significantly differ between SCID-CB17 and WT-CB17 mice, but tumors from SCID mice grow less frequently on transplantation to WT mice than do the tumors from WT mice (Engel et al., 1997). Conversely, MCA-induced tumors grow, on the average, after a longer latency in IFNR-WT compared with IFNR-KO mice, yet tumors of both types of mice could be transplanted with a comparable efficiency (Qin et al., 2002). In this case, tumor fragments were transplanted. Because, in addition, we have also seen that the decreased tumor incidence and longer latency in RAG2-WT compared with RAG2-KO mice observed by some investigators (Shankaran et al., 2001) appears to be the result of other, perhaps environmental, factors but not a T cell response to the tumor, it becomes clear that latency in MCA-treated animals is unlikely to be a result of T-cell-mediated tumor cell recognition and selection of lessimmunogenic tumor cell variants. This is compatible with our observation that at the site of MCA, T cells can be detected only in very low numbers during the critical phase of tumor development (between weeks 3 and 20) (Qin et al., 2002). It is important to note that it has been known for a long time that mice with progressively growing immunogenic MCA-induced tumors can reject the same tumor after surgical excision and transplantation into the same (autochthonous) host (Klein et al., 1960). This required immunization with irradiated cells of the same tumor and tumor rejection was comparable between the autochthonous host and control mice, therefore indicating that the primary host had not developed a significant form of tumor immunity. It is furthermore interesting to note that Prehn demonstrated a while ago that the latency period of MCA-induced tumors does not strictly correlate with their immunogenicity (Prehn and Bartlett, 1987). He even concluded from his experiments that weak immune stimulation by the tumor promotes tumor growth (Prehn, 1994) that is in

194

Thomas Blankenstein and Zhihai Qin

contrast to immune surveillance but in line with a tumor-promoting effect of inflammation. Why MCA-induced tumors from T cell-competent mice can be less efficiently transplanted into WT mice than those from T cell-deficient mice is unclear, but T-cell recognition of the tumor in the primary host and selection of less immunogenic variants can hardly explain this phenomenon. Second, to support a role of T cells and IFN in control of MCA-induced tumors, RAG2-KO and IFN-KO mice were observed and shown, in contrast to WT mice, to develop spontaneous tumors (mainly gastrointestinal carcinomas and lymphomas) with a high frequency (Shankaran et al., 2001; Street et al., 2002). These studies were taken as evidence to support, in principle, the original concept of cancer immune surveillance (Shankaran et al., 2001). We think that it is very problematic to compare spontaneous tumors in lifelong immune-suppressed aged mice with an age-matched immune-competent control group of mice, as the lifelong different immune status may result in a very different microbial load and composition in these mice. This, in turn, could result in persistent infection as a result of incomplete pathogen elimination and chronic inflammatory responses that promote tumor development in the immune-deficient mice. In fact, recently it was shown that the spontaneous tumor development in IFN-KO mice could be completely prevented by adding antibiotics into the drinking water (Enzler et al., 2003). Similarly, spontaneous tumor development in aged RAG2-KO mice was not confirmed in another study (Erdman et al., 2003). However, when the mice were infected with Helicobacter hepaticus, they developed an inflammatory bowel disease and subsequent colon carcinomas. Helicobacter pylori infection is one of the best pieces of evidence that chronic inflammation contributes to cancer in humans (Blaser and Smith, 1999). It appears, therefore, that spontaneous tumors in immune-deficient KO mice are the result of persistent infections and imbalanced inflammatory responses. The data cannot be taken as evidence that in WT mice a tumor developed similarly to the KO mice and was subsequently recognized and eliminated by T cells. Discrepancies in the literature with regard to spontaneous or MCAinduced tumors are not new. We cite from Stutman (1975), published 28 years ago in this journal: ‘‘Significant variations in spontaneous tumor incidence or tumor growth can be observed in mice depending on type of cage, isolation versus grouping, well tolerated endemic infections, or dietary influences. The variation produced by these factors is indeed of the same magnitude as some of the differences between groups and between experiments observed both in the positive and negative experiments discussed.’’

Chemical Carcinogens as Foreign Bodies

195

V. THE PROTECTIVE RESPONSE IN IFNgR-WT MICE IS ASSOCIATED WITH ENCAPSULATION OF MCA A. Tissue Damage, Tissue Repair Response, and Fibrosis at the Site of MCA We have seen that IFN/IFNR has a protective role against MCAinduced tumors, but not against DMBA/TPA-induced tumors or tumors that are caused by p53 deficiency, and furthermore, that spontaneous tumors do not occur more frequent in IFN-KO mice, if infection in the mice is controlled by antibiotics. We have also seen that T cells do not seem to play a direct role during the protective response against MCA-induced tumors but that they, perhaps in response to other stimuli, can provide IFN that, by a poorly understood mechanism, could contribute to protection from MCA carcinogenesis. This indicated that the clue for the mechanism of the protective response was locally at the site of MCA. In the majority of cases, tumors appear at the site at which MCA was injected— very frequently fibrosarcomas. We undertook an analysis in which we injected MCA into the muscle, performed serial tissue sections of that muscle at different time points after MCA injection (between week 3 and 20; the critical period before tumors become apparent) to detect the injected MCA within the tissue, and then analyzed the local events by (immuno)histology (Qin et al., 2002). Regardless of the time point, MCA was always detected on some of the serial sections in both IFNR-WT and IFNR-KO mice. An example is shown in Fig. 2a. It shows coal-like crystalline structures within the muscle tissue. This is interesting, as MCA is water insoluble and therefore was mixed in sesame oil before injection. Probably, waterinsoluble material of certain size cannot be removed by the organism. To verify that these structures were indeed MCA, we made use of the fact that MCA is a polycyclic aromate and can be visualized under the fluorescence microscope as a result of its autofluorescence. This is shown in Fig. 2b. The muscle tissue around the MCA was completely destroyed, and a strong reactive infiltrate was detected (Fig. 2a). It appears that MCA primarily induces tissue damage. Consecutive sections often also contained MCA and were analyzed for the type of inflammatory cells that accumulated at the site of MCA. Granulocytes (Gr1þ) and monocytes/macrophages (Mac1þ) occurred in large numbers at all time points at the site of MCA. Most dramatically, however, was the accumulation of fibroblasts, detected by monoclonal antibody (mAb) ER-TR7, which detects fibroblasts and extracellular matrix (ECM) components. On non-MCA-containing sections this mAb yielded a regular faint staining, completely lining virtually all muscle

196

Thomas Blankenstein and Zhihai Qin

Fig. 2 Tissue damage at the site of Methylcholenthrene (MCA). The left site shows a section of a muscle tissue injected with MCA 6 weeks before. MCA can be seen as coal-like crystals. Between weeks 3 and 20 MCA was always detected on some of the consecutive muscle sections in both interferon--receptor-wild type mice and interferon--receptor knock-out mice. Note that the muscle tissue is completely destroyed at the site of MCA and a strong reactive infiltrate can be recognized. On the right site the same section under a fluorescence microscope is shown. The autofluorescence is the result of the fact that MCA is a polycyclic aromatic hydrocarbon. Reproduced with permission from Qin et al. (2002). (See Color Insert.)

fibers (Fig. 3a). On sections with MCA, the dramatic fibroblast accumulation around MCA resembled a microscopic fibrosis (Fig. 3b), similar in IFNR-WT and IFNR-KO mice. Fibroblasts seemed to be in closest contact to MCA, so that it is likely that they were the cell type most exposed to the mutagenic activity of MCA. It is easy to understand, therefore, that MCA usually induces fibrosarcomas. On some of the sections derived from IFNR-KO mice at later time points (20 weeks after MCA injection), small tumors were detected. These tumors had typically downregulated the fibroblast marker. It is known that tumors derived from fibroblasts often downregulate fibroblast-specific markers (Hynes, 1990; Kopp et al., 1995; Schreier et al., 1988). Together, it seems that MCA primarily induces tissue damage and, comparably rarely, tumors. The reactive infiltrate is typical for a tissue repair (or wound-healing) response (Chettibi and Ferguson, 1999; Kovacs, 1991). Compatible with a tissue repair response, T cells (CD4þ, CD8þ and þ T cells were analyzed) occurred in only low numbers at the site of MCA, if all sections from the different time points were taken together (Qin et al., 2002). Because only the tumor but not the tissue damage is macroscopically visible, one could erroneously conclude that the protective response in the presence of IFNR is directed against tumor development. Because of the abundance and early appearance of tissue damage at the site of MCA and the typical features of a tissue repair response, it is more likely,

Chemical Carcinogens as Foreign Bodies

197

Fig. 3 Fibroblast/extracellular matrix accumulation at the site of methylcholenthrene (MCA) as a feature of a tissue-repair response. The left site shows a muscle section of a mouse that was not injected with MCA. The right site shows a muscle section of a mouse injected 10 weeks ago intramuscularly and was confirmed before staining to contain MCA. The sections are stained with the mAb ER-TR7 that detects fibroblasts and extracellular matrix. Reproduced with permission from Qin et al. (2002). (See Color Insert.)

however, that the protective response is directed against tissue damage and concomitantly protects from tumor development.

B. Local Production of IFNg Leads to Encapsulation of Transplanted Tumor Cells Because of the difficulty in detecting IFN-producing cells at the site of MCA, the effects of chronic local IFN production on tissue structure was analyzed (Qin et al., 2002). Therefore, tumor cells were engineered to secrete IFN and injected into T-cell-deficient mice. Because of IFN secretion, the growth of the cells is strongly delayed, but without T cells, mice usually cannot completely reject such cells, even though they continue to produce IFN (Hock et al., 1993). After several weeks of in vivo growth, the IFN-producing tumor cells were completely surrounded by a collagenous capsule that separated the tumor from neighboring tissue. Whether this is a direct effect of IFN is not clear. Because other studies showed that IFN, for example, produced by tumor-specific T cells, inhibits angiogenesis that occurs shortly after injection of a tumor cell suspension (Blankenstein and Qin, 2003; Qin and Blankenstein, 2000; Qin et al., 2003), it is possible that ECM deposition surrounding the tumor is a secondary event caused by chronic blood deprivation. How the encapsulation of transplanted IFNproducing cells relates to the mechanism that inhibits MCA-induced

198

Thomas Blankenstein and Zhihai Qin

carcinogenesis is currently not known because, among other reasons, a significant number of IFN-producing cells has not been detected so far at the site of MCA.

C. MCA Diffusion and DNA Damage is Inhibited in an IFNgR-Dependent Fashion Two experiments indicated that the protective effect of IFNR acted in a very early phase of malignant transformation. We noticed, in a very simple assay, that the MCA/sesame oil emulsion is retained for weeks if injected subcutaneously into IFNR-WT mice. In IFNR-KO mice, the emulsion diffused more rapidly (Qin et al., 2002). This phenomenon is reminiscent of the observation that patients with IFNR deficiency became apparent because of disseminated mycobacterial infection (Jouanguy et al., 1996; Newport et al., 1996). Disseminated tuberculosis correlating with aberrant granuloma formation has also been observed in IFN-KO mice (Cooper et al., 1993; Flynn et al., 1993). Interestingly, the type of solvent influences tumor incidence and location of tumor development. Mice injected with the same dose of MCA, mixed either in benzene or in olive oil, developed tumors in 100% or 50% of the cases, respectively (Reddy and Fialkow, 1981). The above-mentioned discrepancies concerning MCA-induced tumor incidence in different KO mice are not the result of the solvent, as oil was used in most of the studies. Assuming that MCA induces mutations in a random fashion and that the majority of mutations do not create a growth advantage but decrease cell survival (as indicated by the dramatic tissue damage; Fig. 1), we analyzed whether the extent of DNA damage in cells at the site of MCA differed between IFNR-WT and IFNR-KO mice. By TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nickend labeling) staining that detects DNA breaks, it was determined that the cells surrounding MCA contained more DNA damage in IFNR-KO mice compared with IFNR-WT mice (Qin et al., 2002). Possible explanations will be discussed below. This experiment is interesting, because it indicates that the protective response in IFNR-WT mice, by reducing DNA damage, may inhibit cell death and malignant transformation at the same time.

D. Encapsulated MCA Persists in Long-Term Tumor-Free Mice The observations that the presence of IFNR decreased the diffusion of the MCA/oil emulsion from the injection site, that in IFNR-WT mice less DNA damage was detected in cells at the site of MCA in comparison to

Chemical Carcinogens as Foreign Bodies

199

IFNR-KO mice, and that the local release of IFN induced a collagenous capsule enclosing the IFN-producing cells led us to analyze whether in long-term tumor-free mice encapsulated MCA could be detected. In almost all mice that had been injected with MCA 65–75 weeks ago and that had not developed a tumor, MCA crystals were detected on some of the consecutive muscle sections (Qin et al., 2002). Staining with a fibroblast/ECMspecific mAb revealed that MCA was encapsulated by ECM embedded within fibrotic tissue (Fig. 4). Obviously, parts of the MCA can persist at the site of injection for very long times, perhaps forever, within microscopic scars. This response has typical features of a foreign-body reaction. The foreign-body reaction is characterized by encapsulation of foreign material that the organism cannot remove or degrade. Phylogenetically, it belongs to the oldest defense mechanisms together with phagocytosis and cytotoxicity and predates adaptive immunity. Encapsulation is a major defense mechanism in insects; for example, against larger parasites that cannot be phagocytozed (Gillespie and Kanost, 1997; Hoffmann, 1995). In mollusks, the foreign-body reaction can best be seen in form of the pearl, encapsulated sand corns naturally, or artificially implanted in a way that they cannot be excreted (Cooper, 1976). Even the simplest multicellular organisms, porifera, can form a barrier by deposition of a fibrous wall at the interphase to an

Fig. 4 Encapsulated methylcholenthrene persists within fibrotic area in long-term tumor-free mice. Shown is the staining with mAb ER-TR7 of a muscle section of a tumor-free interferon-R-wild-type mouse injected more than a year ago with MCA. The inlet figure shows a piece of encapsulated MCA at higher magnification. Reproduced with permission from Qin et al. (2002). (See Color Insert.)

200

Thomas Blankenstein and Zhihai Qin

allograft (Smith and Hildemann, 1984). It appears that the foreign-body reaction belongs to the oldest defense mechanisms for keeping tissue integrity. In humans, it is typically associated with a pathological situation (Rigdon, 1975) that is most dramatically seen after exposure to asbestos (Wagner, 1997). It is not clear whether the foreign-body reaction to MCA that is observed in the course of a tissue-damage response is of physiological relevance, as usually a large amount of MCA causing tumor development within a few months in most of the mice has been applied. However, some clinical phenomena are worth mentioning. PAHs are ubiquitous environmental pollutants and are taken up and can accumulate in the lung; (Matsumoto and Kashimoto; 1985; Chkubo et al., 1988; Seto et al., 1993; Sun et al., 1984; Tomingas et al., 1976). It is possible that naturally inhaled PAHs, despite their usually lower concentration, also induce tissue damage, and because it is not clear whether inhaled PAHs can be completely excreted, one should not exclude the possibility that part of the inhaled PAH is encapsulated. If this is the case, several questions arise: In which metabolic form do PAHs persist in the body, and are they still mutagenic? Does the encapsulation response inactivate PAH, and are the PAH, reactivated by degrading the capsule? It is also not known whether the MCA that persists in an encapsulated form in long-term tumor-free mice has still mutagenic activity that can be reactivated; for example by degrading the ECM. At least in vitro, PAH solutions are very stable (Vaessen et al., 1988). PAH in tobacco is the major cause of smoking-induced lung cancer. Former smokers retain a substantial risk of developing lung cancer for several years after cessation from smoking (Enstrom and Heath, 1999). Although this is often explained by infection and tumor-promoting inflammatory response comparable to two-stage carcinogenesis in mice, an interesting additional possibility is that during inflammation-induced tissue remodeling, the ECM is degraded and PAHs are set free to induce (additional) mutations. However, this is only speculation at this time. The microscopic scars containing encapsulated MCA in aged tumor-free mice are reminiscent of the description of scar cancer (Narbenkrebs) reported by Friedrich (1939) and Ro¨ßle (1943). These pathologists noticed that lung carcinomas often grew at sites of scars and suggested that the scars predisposed to carcinomas. On the basis of the type of collagen produced, it was shown that scars within the carcinoma had an immature phenotype, whereas those located at some distance from the carcinoma had a more mature phenotype (Madri and Carter, 1984). Even though the relevance of the mouse experiments for this clinical observation is not clear, it is possible that the scars reflected an ineffective or too slow, but otherwise protective, response.

Chemical Carcinogens as Foreign Bodies

201

VI. CONCLUDING REMARKS We have shown that IFN/IFNR prolongs the latency and decreases the incidence of MCA-induced tumors. This is so far a unique example for the spontaneous control of primary, non–virus associated, non–infection induced tumors in mice. Spontaneous tumors, tumors induced because of p53-deficiency, or tumors induced by DMBA/TPA do not occur more frequently in IFNR-KO mice compared with IFNR-WT mice. It has been a widely held misconception that the increased MCA-induced tumor incidence in the absence of IFN/IFNR is related to Burnet’s immune surveillance. Indeed, the absence of T cells does not increase the incidence of MCA-induced tumors. However, it appears that the presence of T cells in combination with environmental factors can decrease MCA-induced tumor incidence. This experimental artifact is interesting, as it could point to some of the enigmatic sources of IFN. A generally heightened immune reactivity induced by environmental antigens could nonspecifically elevate protective IFN levels. How systemic IFN that has a very short half-life in the blood can perform this task is completely unknown. The data discussed also indicate that an innate inflammatory response, in most systems a protumorigenic event, can be protective in certain experimental situations; for example, the carcinogen-induced tissue repair response. The finding of increased tumor incidence induced by MCA in IFNR-KO mice has raised more questions than have been answered. Several alternatives that may occur or that have been shown to occur at the site of MCA are summarized in Fig. 5. They focus on fibroblasts, as they occur abundantly at the site of MCA-induced tissue damage; they are exposed to the genotoxicity of MCA; they are the cell type that is most often transformed; and they are the predominant source of ECM. First, MCA induces random mutations—in rare instances the right combination that predisposes to malignancy. Once a malignant clone arises, it will grow within a short period of time. There is no indication of an adaptive or innate immune mechanism that selectively recognizes cells harboring transforming mutations and hampers tumor progression. Second, regardless of the type of mutation, cells with DNA damage respond to IFN from an unknown source with increased DNA repair, perhaps because they are recognized as stressed. Even though this possibility is not so far supported by data, it should be mentioned, as mouse embryonic fibroblasts derived from IRF1-KO mice acquire more mutations after cisplatin exposure than those from IRF1-WT mice (Nozawa et al., 1999). IRF-1 is a transcription factor essential for IFN-induced responses. Third, most abundant at the site of MCA is cell death, next to the inflammatory response typical for tissue repair. The straightforward explanation is that MCA induces mutations in genes whose products are

202

Thomas Blankenstein and Zhihai Qin

Fig. 5 Possible local events at the site of methylcholanthrene (MCA). Fibroblasts (F) accumulate at the site of MCA during tissue repair and are exposed to the random mutagenic activity of MCA (upper-left part; asterisks in the enlarged cell indicate DNA damage). 1. MCA hits the ‘‘right’’ genes in a cell; for example, a base substitution in the ras oncogene or an inactivating mutation in the p53 tumor suppressor. This is a rare event, as it requires a very large amount of the chemical carcinogen to reproducibly induce tumors. Cells with the ‘‘right’’ mutations will grow as tumor. No immunologic growth restrictive mechanism, innate or adaptive, that impairs growth of a (pre)malignant clone has been demonstrated so far. For simplicity, malignant transformation is shown as a single step but can be expected to occur as a step-wise accumulation of mutations. The absence of interferon /receptor (IFNR)/IFNR increases the tumor incidence. 2. Following DNA damage by MCA, cells able to respond to IFN from an unidentified source have increased DNA repair activity and mutations are not fixed. The question mark indicates that no experimental evidence exists for this possibility. 3. MCA-induced DNA damage leads to abundant cell death. Cell-intrinsic control mechanisms, for example, induction of apoptosis, may be responsible. A role for IFN/IFNR has not been demonstrated so far. 4. During tissue repair, MCA is encapsulated and persists in mice for long time with no apparent genotoxicity. This response appears to be more effective in the presence of IFN/IFNR. In which metabolic form MCA persists in mice and whether it is still mutagenic without the capsule is not known. Reducing DNA damage would simultaneously inhibit tissue damage and malignant transformation. ECM, extracellular matrix. (See Color Insert.)

Chemical Carcinogens as Foreign Bodies

203

essential for survival. In addition to DNA damage-induced cell-intrinsic death mechanisms, IFN could contribute to elimination of damaged cells. In some models, IFN has been shown to induce apoptosis. However, no experimental evidence exists for this possibility. The more pronounced TUNEL staining at the site of MCA in IFNR-KO mice compared with IFNR-WT mice does not support this assumption. Fourth, MCA or MCA-caused tissue damage induces a tissue-repair response in the course of which MCA crystals that cannot be removed or degraded are encapsulated as foreign bodies leaving behind microscopic scars. Whether MCA persists in tumor-free mice in a genotoxic form and can be reactivated is not known. The evidence for a direct contribution of IFN during the foreign-body reaction is so far only indirect, shown by reduced diffusion of the MCA emulsion in IFNR-WT mice compared with IFNR-KO mice and the observation that transplanted IFN-producing cells are encapsulated. The MCA-encapsulation response might differ from the second and third alternatives in one important aspect. It is not the damaged cells after exposure to MCA that are the target of the protective response, but MCA itself, to reduce cell damage, thereby simultaneously reducing tissue damage and malignant transformation. Additional mechanisms and their relative contribution to protection from MCA-induced carcinogenesis may exist but are yet to be discovered.

ACKNOWLEDGMENTS The work cited in this review is supported by grants from the Deutsche Krebshilfe, Dr. Mildred Scheel-Stiftung, e.V., the Deutsche Forschungsgemeinschaft, and the Bundesministerium fu¨r Bildung und Forschung. We are grateful to Hans Schreiber, Chicago, for his continuous critique and for drawing our attention to unconsidered viewpoints and important older literature.

REFERENCES Artandi, S. E., Chang, S., Lee, S. L., Alson, S., Gottlieb, G. J., Chin, L., and DePinho, R. A. (2000). Nature 406, 641–645. Banchereau, J., and Steinman, R. M. (1998). Nature 392, 245–252. Biernaux, C., Loos, M., Sels, A., Huez, G., and Stryckmans, P. (1995). Blood 86, 3118–3122. Blankenstein, T., and Qin, Z. (2003). Curr. Op. Immunol. 15, 148–154. Blaser, M. J., and Smith, P. D. (1999). In: Inflammation: Basic Principles and Clinical Correlates. 3rd ed., 1107–1116. Lippincott Williams & Wilkins, Philadelphia. Boehm, U., Klamp, T., Groot, M., and Howard, J. C. (1997). Annu. Rev. Immunol. 15, 749–795.

204

Thomas Blankenstein and Zhihai Qin

Burnet, F. M. (1964). Brit. Med. Bull. 20, 154–158. Burnet, F. M. (1970). Progr. Exp. Tumor Res. 13, 1–27. Burnet, F. M. (1971). Transplant. Rev. 7, 3–25. Burnet, F. M. (1973). Aust. N. Z. J. Med. 3, 71–77. Burnet, F. M. (1978). Adv. Cancer Res. 28, 1–29. Chatterjea-Matthes, D., Garcia-Ojeda, M. E., Dejbakhsh-Jones, S., Jerabek, L., Manz, M. G., Weissman, I. L., and Strober, S. (2003). J. Immunol. 171, 1207–1215. Chettibi, S. (1999). In. Inflammation: Basic Principles and Clinical Correlates. 3rd edn., pp. 865–881. Lippincott Williams & Wilkins, Philadelphia. Clark, R. E., Dodi, I. A., Hill, S. C., Lill, J. R., Aubert, G., Macintyre, A. R., Rojas, J., Bourdon, A., Bonner, P. L., Wang, L., Christmas, S. E., Travers, P. J., Creaser, C. S., Rees, R. C., and Madrigal, J. A. (2001). Blood 98, 2887–2893. Cockfield, S. M., Urmson, J., Pleasants, J. R., and Halloran, P. F. (1990). J. Immunol. 144, 2967–2974. Cooper, A. M., Dalton, D. K., Stewart, T. A., Griffin, J. P., Russell, D. G., and Orme, I. M. (1993). J. Exp. Med. 178, 2243–2247. Cooper, E. L. (1976). Isr. J. Med. Sci. 12, 479–494. Coussens, L. M., Tinkle, C. L., Hanahan, D., and Werb, Z. (2000). Cell 103, 481–490. Coussens, L. M., and Werb, Z. (2001). J. Exp. Med. 193, F23–F26. Coussens, L. M., and Werb, Z. (2002). Nature 420, 860–867. Cretney, E., Takeda, K., Yagita, H., Glaccum, M., Peschon, J. J., and Smyth, M. J. (2002). J. Immunol. 168, 1356–1361. Deelman, H. T. (1926). Z. Krebsforschung 24, 86–98. DePinho, R. A. (2000). Nature 408, 248–254. DiGiovanni, J. (1992). Pharmac. Ther. 54, 63–128. Dolle, M. E., Snyder, W. K., Gossen, J. A., Lohman, P. H., and Vijg, J. (2000). Proc. Natl. Acad. Sci. USA 97, 8403–8408. Dunn, G. P., Bruce, A. T., Ikeda, H., Old, L. J., and Schreiber, R. D. (2002). Nat. Immunol. 3, 991–998. Ehrlich, P. (1909). Ned. Tijdschr. Geneeskd. 5, 273–290. Engel, A. M., Svane, I. M., Rygaard, J., and Werdelin, O. (1997). Scand. J. Immunol. 45, 463–470. Enstrom, J. E., and Heath, C. W., Jr. (1999). Epidemiology 10, 500–512. Enzler, T., Gillessen, S., Manis, J. P., Ferguson, D., Fleming, J., Alt, F. W., Mihm, M., and Dranoff, G. (2003). J. Exp. Med. 197, 1213–1219. Erdman, S. E., Poutahidis, T., Tomczak, M., Rogers, A. B., Cormier, K., Plank, B., Horwitz, B. H., and Fox, J. G. (2003). Am. J. Pathol. 162, 691–702. Fish, D. C., Demarais, J. T., Djurickovic, D. B., and Huebner, R. J. (1981). Proc. Natl. Acad. Sci. USA 78, 2526–2527. Flynn, J. L., Chan, J., Triebold, K. J., Dalton, D. K., Stewart, T. A., and Bloom, B. R. (1993). J. Exp. Med. 178, 2249–2254. Fraumeni, J. F., Jr., Devesa, S. S., Hoover, R. N., and Kinlen, L. J. (1993). In. Cancer: Principles and Practice of Oncology. 4th edn., J. B. Lippincott Co., Philadelphia. Friedrich, G. (1939). Virchows Arch. Pathol. Anat. 304, 230–247. Gillespie, J. P., and Kanost, M. R. (1997). Annu. Rev. Entomol. 42, 611–643. Girardi, M., Oppenheim, D. E., Steele, C. R., Lewis, J. M., Glusac, E., Filler, R., Hobby, P., Sutton, B., Tigelaar, R. E., and Hayday, A. C. (2001). Science 294, 605–609. Goes, N., Sims, T., Urmson, J., Vincent, D., Ramassar, V., and Halloran, P. F. (1995). J. Immunol. 155, 4559–4566. Halevy, O., Rodel, J., Peled, A., and Oren, M. (1991). Oncogene 6, 1593–1600.

Chemical Carcinogens as Foreign Bodies

205

Halloran, P. F., Autenried, P., Ramassar, V., Urmson, J., and Cockfield, S. (1992). J. Immunol. 148, 3837–3846. Hanahan, D., and Weinberg, R. A. (2000). Cell 100, 57–70. Harris, D. P., Haynes, L., Sayles, P. C., Duso, D. K., Eaton, S. M., Lepak, N. M., Johnson, L. L., Swain, S. L., and Lund, F. E. (2000). Nat. Immunol. 1, 475–482. Hock, H., Dorsch, M., Kunzendorf, U., Qin, Z., Diamantstein, T., and Blankenstein, T. (1993). Proc. Natl. Acad. Sci. USA 90, 2774–2778. Hoffmann, J. A. (1995). Curr. Op. Immunol. 7, 4–10. Huang, S., Hendriks, W., Althage, A., Hemmi, S., Bluethmann, H., Kamijo, R., Vilcek, J., Zinkernagel, R. M., and Aguet, M. (1993). Science 259, 1742–1745. Humpherys, D., Eggan, K., Akutsu, H., Hochedlinger, K., Rideout, W. M., 3rd, Biniszkiewicz, D., Yanagimachi, R., and Jaenisch, R. (2001). Science 293, 95–97. Hu¨nig, T., and Bevan, M. J. (1980). J. Exp. Med. 152, 688–702. Hynes, R. O. (1990). Fibronectins. Springer, New York. Ibe, S., Qin, Z., Schuler, T., Preiss, S., and Blankenstein, T. (2001). J. Exp. Med. 194, 1549–1559. Ikehara, S., Pahwa, R. N., Fernandes, G., Hansen, C. T., and Good, R. A. (1984). Proc. Natl. Acad. Sci. USA 81, 886–888. Iversen, O. H. (1995). Cri. Rev. Oncog. 6, 357–405. Kaplan, D. H., Shankaran, V., Dighe, A. S., Stockert, E., Aguet, M., Old, L. J., and Schreiber, R. D. (1998). Proc. Natl. Acad. Sci. USA 95, 7556–7561. Jouanguy, E., Altare, F., Lamhamedi, S., Revy, P., Emile, J. F., Newport, M., Levin, M., Blanche, S., Seboun, E., Fischer, A., and Casanova, J. L. (1996). N. Engl. J. Med. 335, 1956–1961. Klein, G., Sjo¨gren, H. O., Klein, E., and Hellstro¨m, K. E. (1960). Cancer Res. 20, 1561–1572. Klein, G., and Klein, E. (1977). Proc. Natl. Acad. Sci. USA 74, 2121–2125. Kopp, M. U., Croci, L. P., and Trueb, B. (1995). Int. J. Cancer 60, 275–279. Kovacs, E. J. (1991). Immunol. Today 12, 17–23. Lawetzky, A., and Hu¨nig, T. (1988). Eur. J. Immunol. 18, 409–416. Lengauer, C., Kinzler, K. W., and Vogelstein, B. (1998). Nature 396, 643–649. Lin, E. Y., Nguyen, A. V., Russell, R. G., and Pollard, J. W. (2001). J. Exp. Med. 193, 727–740. Loeb, L. A. (1991). Cancer Res. 51, 3075–3079. MacKenzie, I., and Rous, P. (1941). J. Exp. Med. 73, 391–415. Madri, J. A., and Carter, D. (1984). Hum. Pathol. 15, 625–631. Maleckar, J. R., and Sherman, L. A. (1987). J. Immunol. 138, 3873–3876. Martin, G. M., Ogburn, C. E., Colgin, L. M., Gown, A. M., Edland, S. D., and Monnat, R. J., Jr. (1996). Hum. Mol. Genet. 5, 215–221. Martin, G. M., and Oshima, J. (2000). Nature 408, 263–266. Matsumoto, H., and Kashimoto, T. (1985). Bull. Environ Contam. Toxicol. 34, 17–23. Medzhitov, R., and Janeway, C., Jr. (2000). Immunol. Rev. 173, 89–97. Moore, R. J., Owens, D. M., Stamp, G., Arnott, C., Burke, F., East, N., Holdsworth, H., Turner, L., Rollins, B., Pasparakis, M., Kollias, G., and Balkwill, F. (1999). Nat. Med. 5, 828–831. Munder, M., Mallo, M., Eichmann, K., and Modolell, M. (1998). J. Exp. Med. 187, 2103–2108. Newport, M. J., Huxley, C. M., Huston, S., Hawrylowicz, C. M., Oostra, B. A., Williamson, R., and Levin, M. (1996). N. Engl. J. Med. 335, 1941–1949. Noguchi, Y., Jungbluth, A., Richards, E. C., and Old, L. J. (1996). Proc. Natl. Acad. Sci. USA 93, 11798–11801. Nozawa, H., Oda, E., Nakao, K., Ishihara, M., Ueda, S., Yokochi, T., Ogasawara, K., Nakatsuru, Y., Shimizu, S., Ohira, Y., Hioki, K., Aizawa, S., Ishikawa, T., Katsuki, M., Muto, T., Taniguchi, T., and Tanaka, N. (1999). Genes Dev. 13, 1240–1245.

206

Thomas Blankenstein and Zhihai Qin

Ohkubo, T., Seto, H., Ohsawa, M., Suzuki, T., and Kanoh, T. (1988). Ann. Rep. Tokyo Metr. Res. Lab. Public Health 39, 193–197. Oshima, M., Dinchuk, J. E., Kargman, S. L., Oshima, H., Hancock, B., Kwong, E., Trzaskos, J. M., Evans, J. F., and Taketo, M. M. (1996). Cell 87, 803–809. Outzen, H. C., Custer, R. P., Eaton, G. J., and Prehn, R. T. (1975). J. Reticuloendothel. Soc. 17, 1–9. Pelkonen, O., and Nebert, D. W. (1982). Pharmacol. Rev. 34, 189–222. Posthuma, E. F., Falkenburg, J. H., Apperley, J. F., Gratwohl, A., Roosnek, E., Hertenstein, B., Schipper, R. F., Schreuder, G. M., D’Amaro, J., Oudshoorn, M., van Biezen, J. H., Hermans, J., Willemze, R., and Niederwieser, D. (1999). Blood 93, 3863–3865. Prehn, R. T. (1994). Cancer Res. 54, 908–914. Prehn, R. T., and Bartlett, G. L. (1987). Int. J. Cancer 39, 106–110. Qin, Z., and Blankenstein, T. (2000). Immunity 12, 677–686. Qin, Z., Kim, H. J., Hemme, J., and Blankenstein, T. (2002). J. Exp. Med. 195, 1479–1490. Qin, Z., Schwartzkopff, J., Pradera, F., Kammerto¨ns, T., Seliger, B., Pircher, H., and Blankenstein, T. (2003). Cancer Res. 63, 4095–4100. Reddy, A. L., and Fialkow, P. J. (1981). Int. J. Cancer 27, 501–504. Rigdon, R. H. (1975). CRC Crit. Rev. Toxicol. 3, 435–476. Ro¨ßle, R. (1943). Schweizerische Med. Wochenschrift 39, 1200–1203. Rous, P., and Kidd, J. D. (1941). J. Exp. Med. 73, 365–389. Rygaard, J., and Povlsen, C. O. (1976). Transplant. Rev. 28, 43–61. Schreiber, H. (1993). In. Fundamental Immunology. 3rd edn., pp. 1143–1178. Raven Press, New York. Schreiber, H. and Rowley, D. A. (1999). In. Inflammation: Basic Principles and Clinical Corelates, pp. 1117–1129. Lippincott Williams & Wilkins, Philadelphia. Schreier, T., Friis, R. R., Winterhalter, K. H., and Trueb, B. (1988). Biochem. J. 253, 381–386. Schu¨ler, T., and Blankenstein, T. (2003). J. Immunol. 170, 4427–4431. Seto, H., Ohkubo, T., Kanoh, T., Koike, M., Nakamura, K., and Kawahara, Y. (1993). Arch. Environ. Contam. Toxicol. 24, 498–503. Shankaran, V., Ikeda, H., Bruce, A. T., White, J. M., Swanson, P. E., Old, L. J., and Schreiber, R. D. (2001). Nature 410, 1107–1111. Smith, L. C., and Hildemann, W. H. (1984). J. Immunol. 133, 2351–2355. Smyth, M. J., Thia, K. Y., Street, S. E., Cretney, E., Trapani, J. A., Taniguchi, M., Kawano, T., Pelikan, S. B., Crowe, N. Y., and Godfrey, D. I. (2000). J. Exp. Med. 191, 661–668. Smyth, M. J., Crowe, N. Y., and Godfrey, D. I. (2001). Int. Immunol. 13, 459–463. Street, S. E., Cretney, E., and Smyth, M. J. (2001). Blood 97, 192–197. Street, S. E., Trapani, J. A., MacGregor, D., and Smyth, M. J. (2002). J. Exp. Med. 196, 129–134. Stutman, O. (1974). Science 183, 534–536. Stutman, O. (1975). Adv. Cancer Res. 22, 261–422. Stutman, O. (1979). J. Natl. Cancer Inst. 62, 353–358. Sun, J. D., Wolff, R. K., Kanapilly, G. M., and McClellan, R. O. (1984). Toxicol. Appl. Pharmacol. 73, 48–59. Takeda, K., Smyth, M. J., Cretney, E., Hayakawa, Y., Kayagaki, N., Yagita, H., and Okumura, K. (2002). J. Exp. Med. 195, 161–169. Takei, Y., Sims, T. N., Urmson, J., and Halloran, P. F. (2000). J. Am. Soc. Nephrol. 11, 250–261. Thomas, L. (1959). Discussion of cellular and humoral aspects of the hypersensitive states. pp. 529–532. Hoeber-Harper, New York. Tomingas, R., Pott, F., and Dehnen, W. (1976). Cancer Lett. 1, 189–195.

Chemical Carcinogens as Foreign Bodies

207

Vaessen, H. A., van de Kamp, C. G., and Jekel, A. A. (1988). Z. Lebensm. Unters. Forsch. 186, 308–310. van den Broek, M. E., Kagi, D., Ossendorp, F., Toes, R., Vamvakas, S., Lutz, W. K., Melief, C. J., Zinkernagel, R. M., and Hengartner, H. (1996). J. Exp. Med. 184, 1781–1790. Van Duuren, B. L., Sivak, A., Katz, C., Seidman, I., and Melchionne, S. (1975). Cancer Res. 35, 502–505. Vogelstein, B., and Kinzler, K. W. (1992). Cell 70, 523–526. Wagner, J. C. (1997). Am. J. Public Health 87, 687–688. Watanabe, H., Shimokado, K., Asahara, T., Dohi, K., and Niwa, O. (1999). Jpn. J. Cancer Res. 90, 40–47. Willimsky, G., and Blankenstein, T. (2000). Cancer Res. 60, 685–692. Yuspa, S., and Shields, P. (2001). In. Cancer, Principles and Practice of Oncology. 6th Edn., pp. 179–193. Lippincott Williams & Wilkins, Philadelphia.

This Page Intentionally Left Blank

Epigenetic Theories of Cancer Initiation Lionel F. Jaffe Marine Biological Laboratory, Woods Hole, Massachusetts 02543, and OB/GYN Department, Brown University, Providence, RI, 02905

I. Introduction II. Evidence that the Early Stages of Cancer are Epigenetic A. Arguments from the Frequency, Reversibility, and Character of cell change B. Arguments from the Long Delays between an Oncogenic Insult and Polymerase Chain Reaction–Demonstrable Mutations C. Evidence from the Effects of Highly Localized, Ionizing Radiation and from ‘‘Bystander’’ Effects of Ionizing Radiation D. Evidence from Normal Development E. Prehn’s Epigenetic Explanation of the High Frequency of Oncogenic Mutations in Tumors F. Evidence from the High Speed of Proto-Oncogene Activation G. Evidence from Fused Cells or Cell Parts H. Evidence from the Nonmutagenicity of Nitrogen Mustards III. Mechanisms of Epigenesis A. DNA Methylation B. The Histone Code C. Tissue Disorganization IV. Concluding Remarks References

I argue that carcinogenic insults injure many cells rather than mutate a few. This results from evidence that such insults convert too many cells to a precancerous state and that too many of the converted cells then revert to plausibly involve mutation and its repair; from evidence that the delays between such insults and chemically demonstrable mutations are long enough to easily allow nonmutational mechanisms to work; from evidence that even ionizing radiation first acts on the cytoplasm and mainly affects cells unhit by it; from the fact that such insults induce proto-oncogene expression far too quickly to do so by mutation; and from the fact that fusions of various cells and cell parts show that the tumorous or nontumorous nature of the product depends on its cytoplasmic rather than its nuclear component. I further argue that reduced DNA methylation, modifications of the histone code, and tissue disorganization are the three main mechanisms of epigenetic cancer initiation. Hypomethylation would result from DNA excision repair. Moreover, a methyl-deficient diet is carcinogenic and demethylation is also known to be carcinogenic via the histone code. Finally, I strongly argue for tissue disorganization as a mechanism of cancer Advances in CANCER RESEARCH 0065-230X/03 $35.00

209

Copyright 2003, Elsevier Inc. All rights reserved

210

Lionel F. Jaffe

initiation. This results from evidence that skin carcinogens disrupt the dermal/epidermal connection and from the fact that tumorigens swiftly disrupt gap junctions, as well as from evidence that such disruption is tumorigenic. ß 2003 Elsevier Inc.

I. INTRODUCTION Familial or inherited cancers account for only about 5% of human cancers (Balmain, 2001); whereas almost all of the other 95% seem to be initiated by chronic rather than single insults. The only established exceptions are the cancers induced by single atomic explosions at Hiroshima, at Nagasaki, and—more recently—at Chernobyl. Yet most investigators, as well as the general public, have long believed that environmental insults act by more or less directly inducing mutations in a small number of founder cells Indeed, the mutational theory of cancer initiation has been generally presented as a well-established truth (Farber, 1996; Hanahan and Weinberg, 2000; Ohlsson et al., 2003; Weinberg, 1989). Indeed, to my knowledge, the last paper to actually argue the case for a mutational origin of cancer was that of Mariano Barbacid (1986). Nevertheless, many investigators have long argued that oncogenic insults injure many cells rather than mutate a few founder cells (Farber and Rubin, 1991; Kennedy, 1985; MacCleod, 1996; Prehn, 1994; Rous, 1959; Sonnenschein and Soto, 2000). Nobelist Peyton Rous’s argument against a mutational origin of cancer remains effective. As he put it, ‘‘[the relationship of carcinogens] to the neoplastic state may be likened to that of ignition to combustion: a fire can be kindled in any one of numerous ways but, with that done, its decisive share in events is ended.’’ One of his still cogent points is that few, if any, spontaneous somatic mutations seem ever to have been observed in mice, and no valid ‘‘mosaics’’ indicative of somatic mutation have ever been reported for human skin. He also points out that urethane (which can induce germinal mutations in Drosophila) induces lung tumors in the offspring of pregnant mice, yet no signs of somatic mutations have been found in serial sections of their lungs. Moreover, a still remarkable and well-confirmed 1906 paper reported that a certain dye (scharlach R), when dissolved in olive oil and injected under the skin of rabbits, induces an apparent malignancy; however, the apparent carcinoma disappears as the dye’s effect wears off. Thus—as in myriad instances since—an apparent tumor proved to be entirely reversible (Fischer, 1906). Moreover, Holliday (1979) proposed an epigenetic theory of carcinogenesis as early as 1979 that is epigenetic and mediated by DNA methylation; whereas Dennis’s recent commentary in Nature magazine suggests a sea change toward the epigenetic view of cancer’s origins. In any case, it is the disagreements between the mutationists and the epigeneticists that continues to generate the broadest conceptual issues in cancer research

Epigenetic Theories of Cancer

211

(Fig. 1). In considering these questions, I will focus on those human behaviors and those environmental insults that seem to generate most human cancers: Aflatoxin injestion, smoking, sunlight, and exposure to a variety of industrial pollutants such as asbestos and benzene (Weston and Harris,

Fig. 1 Mutational (A) and epigenetic (B) models for the initiation of ‘‘sporadic’’, that is, nongenetic carcinomas or epithelial cancers, in man. (Similar concepts apply to sarcomas, but these are harder to illustrate). A. In mutational models, chronic insults produce two to three mutations in individual cells within particular tissues, and these eventually initiate tumors (Knudson, 1977; Miller, 1977; Tomlinson et al., 2001). (B) In epigenetic models, chronic insults repeatedly injure and transiently excite many cells in particular tissues. These excited cells undergo epigenetic responses, and eventually this group of cells becomes a tumor. Mutations are secondary events (Farber and Rubin, 1991; Jaffe, 1982; Kennedy, 1985; Sonnenschein and Soto, 2000).

212

Lionel F. Jaffe

2000). Experiments with other insults will be considered insofar as they contribute to our theoretical understanding of carcinogenesis in man.

II. EVIDENCE THAT THE EARLY STAGES OF CANCER ARE EPIGENETIC A. Arguments from the Frequency, Reversibility, and Character of Cell Change Many investigators have long argued that both the conversion of healthy cells to a precancerous state and their later reversion occur in too high a fraction of these cells to plausibly involve specific DNA damage or its repair. Good reviews of such arguments can be found in Farber and Rubin (1991) and in Prehn (1994). As one example of evidence based on frequency of transformation, when cells from a line of mouse prostatic cells were treated with 4–10 M 3methylcholanthrene, 100% of the cells yielded malignant clones, as compared with 2% of vehicle-treated controls (Mondal and Heidelberger, 1970). A paragraph in a paper from the same laboratory a decade later ‘‘considered’’ this remarkable report and adds that a study of a mouse cell line treated with the same carcinogen at the same concentration show a radically lower frequency of transformed cells. However, it was a different cell line and (unlike the earlier study) ‘‘no replating’’ was done (Fernandez et al., 1980, p. 7274). As a second example, even when the oncogenic insult consisted of Xrays—an insult that is widely supposed to induce cancers by directly inducing mutations—a study using a line of cultured mouse cells showed that almost all of the exposed cells had been affected, for they bore progeny with an increased chance of oncogenic transformation (Kennedy et al., 1980). As a third example, when hamster cheek pouches were treated with dimethylbenz[a]anthracene (using a well-established procedure to induce carcinomas), virtually all treated cells responded with a large, long-lasting increase in proteolytic activity, whether or not tumors arose (Messadi et al., 1986). As a fourth example, individual cells in a fibroblastic mouse cell line were treated with 5-aza-20 -dexycytidine (which induces DNA hypomethylation). Ten percent of these individual cells formed colonies in which most or all of the cells had a cancerous appearance, yet the the ‘‘pretransformed’’ cells initially looked just like ‘‘nonpretransformed’’ ones (Rainier and Feinberg, 1988).

Epigenetic Theories of Cancer

213

As a fifth example, cells in a fibroblastic cell line were grown in such a way that the frequency and circumstances of ‘‘spontaneous’’ transformation could be determined. 100% of the cells were eventually transformed. Moreover, most of these events occurred in nondividing cells (Chow et al., 1994). As a probable sixth example, when mice were injected with a tobacco-specific carcinogen, both DNA methyltransferase and DNA methylation levels doubled within populations of lung alveolar cells by a week after treatment began, and tumors later arose in about 30% of the mice (Belinsky et al., 1996). High frequencies of reversion from a neoplastic to an apparently normal phenotype have been reported in at least four diverse systems. First, mouse teratocarcinoma cells have been reported to revert to a wide variety of differentiated cells after growth in vivo as an ascites tumor for 8 years and almost 200 transplant generations. Indeed, they revert to such a wide variety of cells that they are called totipotent. Moreover, the authors of this report concluded that these ‘‘results . . . furnish an unequivocal example in animals of a nonmutational basis for transformation to malignancy and of reversal to normalcy’’ (Mintz and Illmensee, 1975), whereas an earlier review likewise ‘‘proposed that the irreversibility of the malignant change is incorrect’’ (Pierce, 1967). Moreover, later papers on teratocarcinoma cells do not challenge this important conclusion (Lehtonen et al., 1989; Papaioannou, 1993). Second, one would not expect to be able to induce the repair of mutations induced by X-rays in large numbers of cells simply by treating them with protease inhibitors. Yet cells from the same 10T 1/2 line that can be largely transformed by X-rays can be entirely returned to their preirradiated, untransformed state simply by treatment with micromolar concentrations of the protease inhibitor antipain as late as 13 cell divisions after irradiation. The ease of inducing this reversal is underscored by the fact that another protease inhibitor, chymostatin, suppresses such X-ray transsformation at concentrations as low as 2 pM. Apparently, X-ray transformation of such cells acts by somehow inducing an epigenetic change that introduces some critical protein or proteins into the plasma membrane or into the extracellular matrix. Third, four spontaneously transformed cell lines that were derived from embryonic mouse fibroblasts showed high frequencies of spontaneous reversion to a nontumorigenic phenotype when repeatedly cloned in vitro (Lavrovsky et al., 1992). Fourth, rats that are fed a remarkable variety of poisons eventually develop liver cancers that are long preceded by the development of histologically and biochemically characteristic ‘‘hepatocyte nodules,’’ very few of which go on to form carcinomas. The great majority spontaneously revert by redifferentiation of all or almost all of the precancerous nodules’ cells into ones of normal appearance and organization (Farber and Rubin,

214

Lionel F. Jaffe

1991; Takematsu et al., 1983). Furthermore, the character of late-arising genetic changes induced by X-rays or by alpha rays is very different from the character of induced directly by X-rays. A study of changes in the HPRT gene in CHO cells showed that the large majority of late-arising genetic changes are point mutations and thus resemble spontaneous genetic changes, whereas the large majority of genetic changes produced directly by X-rays consists of deletions (Little, 2000; Little et al., 1997).

B. Arguments from the Long Delays Between an Oncogenic Insult and Polymerase Chain Reaction–Demonstrable Mutations The shortest delays are about 1–2 weeks, as is shown in Table I. Elsewhere in the murine literature, demonstrations of DNA damage are months to years after the oncogenic insult or insults occur or start. The authors of cases 2 and 3 attributed these delays to the time taken for the pair of cell divisions needed to ‘‘fix’’ mutations. Moreover, in case 2, the authors estimated the time required for two cell divisions as 7 days, which would have fully explained the observed delay. However, subsequent evidence indicates that the time needed for two cell divisions was only about 2 days. To consider this evidence, one must first show that the insults applied—skin-applied carcinogens in case 1 and ultraviolet irradiation in case 3—would have killed numerous epidermal cells and thereby induced hidden wounds in the insulted mouse skins. This, as evidence discussed later shows that cells within a wounded epidermis have doubling times of about 1 day rather than the 3.5 days cited by Nelson et al. for unperturbed skin. When UV was the insult, Table I

The Shortest Delays Between Oncogenic Insults and Polymerase Chain Reaction Demonstrable Mutations in Mice or Rats Case Insult (and Tissue Examined by PCR for what) 1 Topical urethane or dimethylbenzanthracene application (epidermis analyzed for Ha-ras) 2 Intraperitoneal urethane injection (lungs analyzed for K-ras) 3 Ultraviolet irradiation (skin analyzed for p53)

Delay (days) 6 <13 <7

Ref. (1) (2,3) (4)

Note. Kumar et al. (1990) reported the appearance of Hras and of Kras mutations in mouse mammary glands less than 13 days after nitrosomethylurea injection. However, Cha, Thilly, and Zarbel reinvestigated this claim in 1994 and obtained compelling evidence that these mutations exist at about the same frequency in mammary glands before nitosomethylurea injection. (1) Nelson et al. (1992). (2) Ichikawa et al. (1996). (3) Yano and Yuasa (1997). (4) Ananthaswamy et al. (1999).

Epigenetic Theories of Cancer

215

application of even a single UV insult of the sort that had been (repeatedly) applied by Ananthaswamy et al. (1999) was later seen to kill most of the epidermal cells in the irradiated skin (Ouhtit et al., 2000). When topically applied carcinogens were the insult, such direct evidence of cell death is not available; however, close consideration of the method used to apply carcinogens by Nelson et al. leaves little doubt that massive cell killing occurred Now let us consider the evidence for reduced doubling times of epidermal cells that are freed of external restraints by wounds that call for swift repair by newly generated cells (as well as by cell migration) or that are freed of such restraint by isolation from the epidermis. With regard to wounding effects, a remarkable paper by Song et al. (2002) provides striking evidence of accellerated cell division rates. Song et al. studied the division of keratinocytes within rat corneas after these tissues had been grossly wounded by cutting out a 3.5-mm disc from the corneal epithelium. Because the corneal epidermis is transparent, they could observe these divisions within the living tissue. The divisions induced by wounding tended to be perpendicular to the epidermal surface rather than parallel to it, as they are in the unperturbed epidermis. More important, their peak doubling times were radically reduced. Indeed, it can be estimated that they were reduced to about 1 day. A 1-day doubling time also held for a line of mouse-derived cells that as they generated formed a skin-like, keratinized, squamous epithelium in vitro (Rheinwald and Green, 1975). Because men live so much longer than mice, one would expect correspondingly longer delay periods in man. This expectation is confirmed by estimates of the delays between the atomic bombing of Hiroshima and the development of leukemias or of lethal solid tumors in the irradiated survivors. These are found to be years to decades, respectively (Pierce et al., 1996). Another interesting case that illustrates the extremely long delays between ultimately oncogenic insult and demonstrable mutation in man is provided by a recent study of patients with chronic ulcerative colitis. The cumulative risk of colon cancer in such patients lies between 5% and 10% after 20 years and between 10% and 30% after 30 years (Azarschab et al., 2002); yet the epigenetic phenomenon of of cytosine methylation is observed far earlier and before demonstrable mutation (Schumacher et al., 1999).

C. Evidence from the Effects of Highly Localized, Ionizing Radiation and from ‘‘Bystander’’ Effects of Ionizing Radiation Using a recently developed alpha particle microbeam, Wu et al. (1999) have studied the mutagenic effects of of cytoplasmic compared with nuclear irradiation of human–hamster hybrid cells. An equilethal dose level of

216

Lionel F. Jaffe

microbeams that only traversed the cytoplasm induced sevenfold more mutants than those that also traversed the nucleus. Moreover, 90% of the mutations that were induced by purely cytoplasmic irradiation are similar to spontaneous ones and are radically different from those induced by nuclear irradiation. Thus, spontaneous mutations were seen as small alterations of a gene coding for a cell surface antigen; whereas those induced by nuclear irradiation were seen as five other mutations, as assessed by PCR. These remarkable results indicate that the mutagenic effects—and presumably the carcinogenic ones—of alpha-particle irradiation are indirect consequences of cytoplasmic damage rather than direct consequences of cytoplasmic damage, and rather than direct effects of nuclear damage. ‘‘Bystander’’ effects are somehow transmitted or propagated to unirradiated cells from cells that are irradiated with -particles or neurons. Surely they cannot act via direct mutagenesis. Moreover, epidemiological evidence shows that -particle irradiation via inhaled radon and its decay products is an important source of lung cancer in man (Lubin and Boice, 1997). For these reasons, bystander effects have excited much interest and have been recently reviewed by Grosovsky (1999) and by Little (2000). The existence of bystander effects was first indicated by the observations of Nakazawa and Little (1992), who irradiated nearly confluent monolayers of Chinese hamster ovary cells with doses of -particles so low as to cross only 1% of the exposed cell population, yet sister chromatid exchanges were induced in about 40% of the cell population, indicating the transmission or propagation of the insult’s consequences to unirradiated cells. Moreover, calculations by Deshpande et al. (1996) indicate that this transmission had occurred over the remarkable distance of about 20 cell diameters. Other efforts to pursue Nakazawa and Little’s discovery have yielded additional indicators of such transmitted or propagated effects but have been hindered by the very limited ability to do such studies with recently isolated cells or whole living animals. An approach to these needs has been recently reported by Watson et al. (2000). They irradiated freshly isolated bone marrow cells with neutrons, mixed these irradiated cells with appropriately labeled unirradiated ones and transplanted the mixture back into mice. Observations made 3–12 months later showed a small but unmistakeable transmission of genetic instability into unirradiated cells within the mice, as shown by various cytogenetic aberrations. Speculation on the mechanisms of intercellular transmission in bystander effects has involved roles for both secreted transmitters and transmitters that pass through gap junctions. Because cells that are subjected to ionizing radiation are known to increase their free calcium levels (Todd and Mikkelsen, 1994), as increased [Ca2+]i induces secretion and also initiates fast calcium waves that can be propagated via gap junctions. For these reasons, I would propose that bystander effects are initiated by increases

Epigenetic Theories of Cancer

217

in free intracellular calcium and spread by fast calcium waves. The evidence behind this thinking will be detailed later. For now, let me note old reports that X-rays and gamma-rays can initiate the development of sea urchin eggs. Moreover, even enucleate halves of sea urchin eggs can be induced to develop by X-rays (Bohn, 1903; Giese, 1947; Harvey, 1956). Because egg development is now known to be started by a fast transcellular calcium wave or waves (L. A. Jaffe, 1993; L. F. Jaffe, 1991), parthenogenesis via X-rays and gamma-rays immediately lends some credence to my proposal.

D. Evidence from the Speed of X-Ray Effects on Proto-Oncogene Expression In an effort to study the mechanism of radiotherapy for brain tumors, Hong et al. (1997) exposed the midbrains of whole mice to X-rays and observed the speed of protooncogene expression. Within 15 minutes after irradiation, C-fos and junB expression began to rise. Such speeds are not compatible with a mutagenic mechanism. Moreover, anesthesia with pentobarbital delayed and attenuated the increases in RNA expression which suggests a calcium-mediated mechanism.

E. Evidence from Normal Development With the exception of the antibody-generating B lymphocytes, healthy mammalian cells are known to be genetically equivalent (Kalthoff, 1996). Why, then, should one suppose that carcinogenesis starts with genetic rather than epigenetic change?

F. Prehn’s Epigenetic Explanation of the High Frequency of Oncogenic Mutations in Tumors A 1986 paper by Barbacid argued that the high frequency of oncogenic mutations in tumors indicates that oncogenesis starts with such mutations. Thus about 15% of established human tumor cells or of the DNAs isolated from unmanipulated human tumor biopsies contain DNA capable of ‘‘transforming’’ suitable recipient rodent cells. However, a convincing epigenetic explanation of the high frequency of such mutations was advanced by Prehn (1994): They are more frequent because they are expressed and are therefore more effectively repaired by nucleotide excision. This differential repair explantion, in turn, was based on the work of Phillip Hanawalt et al. Thus, in the 1985–1986 studies cited by Prehn, repair was three to five times

218

Lionel F. Jaffe

faster in the active gene than in the genome as a whole. Intensive study of DNA repair mechanisms continues. See, for example, de Laat’s 1999 review. However, none of it seems to affect the validity of Prehn’s explanation.

G. Evidence from the High Speed of Proto-Oncogene Activation Proto-oncogene expression is ultimately oncogenic (Bakin and Curran, 1999; Grigoriadis et al., 1995), yet it can be induced so swiftly and by means so unlikely to directly produce mutations as to argue for an epigenetic origin of cancer. Table II documents evidence for the speed with which proto-oncogene expression can be induced in rat and mouse cells in vitro and also shows the nature of the inducers. It only takes various inducers minutes to hours to speed proto-oncogene expression—speeds that would seem to preclude a mechanism that includes mutations as intermediates. Moreover, a 1984 paper by Campisi et al. reported direct evidence that serum induction of expression by the proto-oncogene c-myc in an immortalized line of murine fibroblasts does not rearrange or amplify the c-myc genes. For simplicity, Table II includes little about c-myc expression, but it is generally a bit slower than c-jun expression. A fine, detailed review of much of this can be found in Morgan and Curran (1991), and an excellent introduction to proto-oncogenes can be found in Hesketh (1994). Of the score of inducers of proto-oncogene expression listed in Table II, only ultraviolet, X-rays, phorbol ester, and some estrogens have been reported to have mutagenic activity—a point that further argues for epigenetic action of these fast inducers of proto-oncogene expression.

H. Evidence from Fused Cells or Cell Parts Such experiments provide rather direct evidence of epigenetic mechanisms of tumorigenesis. Guided by the reviews of Shay (1983) and Stern et al. (1999), I will consider the original evidence obtained from hybrids, or fused whole cells; from cybrids, or the products of fusing cytoplasts with whole cells; and from recons, made by fusing cytoplasts with karyoplasts. The main original literature in this area consists of seven papers from 1977 to 1998. These provides substantial further support for an epigenetic rather than a mutagenic mechanism of tumorigenesis. First, Jonasson and Harris (1977) reported a study of hybrid clones formed by the fusion of diploid human fibroblasts and lymphocytes with malignant mouse melanoma cells. Their tumorigenicity was radically suppressed despite procedures that practically eliminated action by the chromosomes of

Table II Induction of Proto-Oncogene Expression within Mammalian Cells in vitro Gene c-myc

Cellsa

Delay

c-myc c-fos

PC 12 PC 12 Human HL 60 PC 12 Swiss 3T3 PC 12

Con A PDGF LPS PDGF PDGF PDGF FCS PDGF, FGF, or EGF FCS EGF NGF NGF, EGF, or FGF Phorbol ester High Kþ Phorbol ester Nicotine PGF, FGF High Kþ

c-fos, c-jun c-fos

Swiss 3T3 Whole brain Whole brain Balb 3T3

PGF, FGF Epileptogenicc Epileptogenicsc Serum

<1 h <15 m 15 m <20 m

c-fos p21ras

Whole brain Human lymphoblasts

Epileptogenic Antibody

<15 m <1 m

c-fos c-fos

T lymphocytes Fibroblasts B lymphocytes Balb 3T3 3 T3 3 T3 3 T3 3 T3

Inducerb

MEF A-431 carcinoma PC 12 PC 12

1–3 h 1–3 h 1–3 h <3 h 10 m 10 m 5m 1h 15 m 1h 30 m 5–10 m 30 m 30 m 15 m <5 m <3 h 45 m

Reference Kelly et al. (1983)

Cochran et al. (1984) Greenberg and Ziff (1984) Kruijer et al. (1984) Mu¨ller et al. (1984)

Bravo et al. (1985) Curran and Morgan (1985) Greenberg et al. (1985)

219

Mu¨ller et al. (1985) Greenberg et al. (1986) Kaibuchi et al. (1986) Morgan and Curran (1986) Tsuda et al. (1985) Morgan et al. (1987) Saffen et al. (1988) Caggana and Kennedy (1989) Sonnenberg et al. (1989) Downward et al. (1990) (continues )

220 Table II (continues) Gene

Cellsa

Inducerb

Delay

c-myc p21ras c-fos, c-jun Ha-ras etc. c-fos c-fos, c-jun NF1 c-fos NF1 c-fos, junB c-srcd Ki-ras c-fos, c-jun c-fos, c-jun ect-2

Whole colon Human T lymphocytes 3T3 He La S3 Cavy endometrial Hepatoma cells Schwann cell line L1-3c-fos Human melanocytes Whole brain Human monocytes Whole gut C 3H 10T1/2 Whole uteri C3H/10T1/2

X-rays Antibody Thapsigargin 254-nm UV Estrogens Dioxin Forskolin IPTG PMA or FGF X-rays Antibody Ethanol injury Anisomycin Diethylstilbestrol Nickel compounds

1–7 d <3 m <30 m <5 m <3 h >2.5 h <1 h 30 m ? <15 m <3 h <3 h <15 m <1 h ?

Reference St. Clair et al. (1990) Graves et al. (1991) Scho¨nthal et al. (1991) Devary et al. (1992) Pellerin et al. (1992) Puga et al. (1992) Gutmann et al. (1993) Miao and Curran (1994) Griesser et al. (1997) Hong et al. (1997) Higuchi et al. (1999) Jones et al. (1999) Thomson et al. (1999) Yamashita et al. (2001) Landolph et al. (2002)

Note. Enucleated cells are fully responsive to 254 nm radiation, confirming that their target was not nuclear in Devary et al’s laboratory experiments (Devary et al., 1993); however, sunlight contains negligible amounts of such short-wave length radiation. Nevertheless, action spectra for the induction of squamous cell carcinomas as studied with human cells indicate that DNA is the direct target of sunlight. Moreover, dimerizations between adjacent pyrimidines are the most common form of damage produced by sunlight (Cleaver and Mitchell, 2000). However, a theoretical analysis indicates that the repair of such dimers demethylates DNA and thereby alters gene expression so as to ultimately yield skin cancers by an epigenetic route (Holliday, 1979). Except where noted as human, all cells are from rats or mice. aThree T3 cells are ?; C 3h 10T1/2 cells are mouse fibroblasts; MEF cells are tertiary mouse embryo fibroblasts; PC12 cells are rat phaeochromocytoma cells; HL 60 cells are human promyelocytic precursor cells; L1-3c-fos cells are ones in which gene expression is controlled by an inert sugar analog, IPTG. bCon A means concanavillin A; EGF, epidermal growth factor; FCS, fetal calf serum; FGF, fibroblast growth factor; LPS, lipopolysaccharide; NGF, nerve growth factor; PDGF, platelet-derived growth factor; PMA, phorbol ester. cEpileptic seizures were induced by intraperitoneal injection of the convulsant, Metrazole (pentylenetetrazole). dThe c-src gene was the first cellular proto-oncogene detected (Stehelin et al., 1976).

Epigenetic Theories of Cancer

221

the nontumorigenic human cells. The authors inferred that tumorigenicity was suppressed by extrachromosomal elements and speculated that these were centrosomes. However, an epigenetic change now seems far more likely. Second, Howell and Sager (1978) reported a study of cybrids formed between tumorigenic and nontumorigenic Chinese hamster cells or of similar mouse cell pairs. Suppression of tumor-forming ability was cytoplasmically transmitted in the hamster cells, though not in the mouse ones. Third, Koura et al. (1982) reported a study of recons and of cybrids formed by fusing highly tumorigenic mouse melanoma cells and cytoplasts from nontumorigenic rat myoblastic cells. Tumorigenicity was suppressed in all of the recons and cybrids, although it did reappear in some clones after prolonged cultivation of the cells. Moreover, recons become highly tumorigenic after treatment with a so-called tumor promoter. In discussing this last finding the authors write that ‘‘it is known that tumor promoters are not significantly mutagenic when used alone.’’ Fourth and fifth, Israel and Schaeffer (1987, 1988) reported two studies of cybrids and of recons between normal and ‘‘transformed’’ rat liver epithelial cells. In the recons—cells in which ‘‘essentially all’’ of the cytoplasm comes from normal cells—tumorigenicity vanished, whereas in cybrids—cells in which less of the cytoplasm comes from normal cells—tumorogenicity was nevertheless substantially reduced. Sixth, Shay and Werbin (1988) reported a study of a recons between a karyoplast of tumorigenic mouse fibroblasts and the cytoplasts of similar fibroblast derived from nontumorigenic cells. None of 10 surviving clones derived in this way proved to be tumorigenic. The authors conclude that, ‘‘These findings offer support for the presence of cytoplasmic factors in nontumorigenic mouse cells that suppress benzo(a)pyrene epoxide-induced tumorigenicity.’’ Finally, Jox et al. (1998) reported a study of hybrids clones formed by fusing cells from a highly tumorigenic murine lymphoma cell line with cells from a nontumorigenic murine lymphoblastoid line. None of the resulting hybrids proved to be tumorigenc.

I. Evidence from the Nonmutagenicity of Nitrogen Mustards It was long ago pointed out that nitrogen mustard gas—well known to be a potent mutagen—nevertheless entirely fails to initiate skin cancers in mice when tested with a classical initiator/promoter experiment (Berenblum and Shubik, 1949). A reasonable, epigenetic explanation of this old mystery is

222

Lionel F. Jaffe

now provided by modern work that shows that nitrogen mustard effectively alkylates DNA and RNA (Gray et al., 1991).

III. MECHANISMS OF EPIGENESIS A. DNA Methylation Any discussion of the role of DNA methylation as an epigenetic mechanism of carcinogenesis must start with Robin Holliday’s remarkable 1979 paper. It argues cogently that hypomethylation will usually result from DNA excision repair and that such hypomethylation will be carcinogenic. One basis for Holliday’s second point is the enormously higher (perhaps 106- to 109-fold higher) probability of the spontaneous origin of a proliferating cancer cell in mice than in men—a remarkable difference pointed out by Peto (1977) and confirmed by Holliday (1996). Because there appears to be little difference in the mutability of rodent cells and of human cells, this difference is far easier to explain by an epigenetic than a mutational mechanism. Another basis for it is that it provides an explanation for the long-established but mysterious fact (reviewed in this journal by Farber [1963]) that dietary ethionine yields liver cancers in mice. There is no reason to believe that this analogue of methionine can be mutagenic; however, it is known to replace S-adenosyl methionine with S-adenosyl ethionine. This, in turn, impairs DNA methylation (Swann et al., 1975) and should thereby lead to hypomethylation. Riggs and Jones (in a 1983 review in this journal) pursued evidence for Holliday’s model. They particularly supported his model with further evidence that various carcinogens inhibit DNA methylation in various living cells. These include ethionine, N-methyl-N-nitroso urea, 5-azacytidine, and benzo(a)pyrene. As they put it so well in their conclusion, ‘‘methylation changes can easily masquerade as mutations.’’ Jones and Buckley (in a 1990 review in this journal) provided further support for Holliday’s model. In particular, they document further evidence that agents that inhibit DNA methylation can transform cells. Thus, they note confirmation by Rainier and Feinberg (1988) that 5-azacytidine application transforms cells in the 10T1/2 line. Further support for the hypomethylation hypothesis was provided by the paper of Wainfan and Poirier (1992). It reported the carcinogenic effects of feeding mice a so-called methyl-deficient diet; that is, one that lacks methionine and several other nutrients that are present in normal diets and thereby interfere with methylation. Mice developed DNA hypomethylation in their livers after only 1 week on such a diet, and likewise developed increased expression of various proto-oncogenes and eventually developed liver

Epigenetic Theories of Cancer

223

tumors. Moreover, an effective new test of the hypomethylation hypthesis has been recently published by Gaudet et al. (2003). They reported their creation of mice with substantially reduced levels of the methyltransferase enzyme that normally maintains DNA methylation and showed that DNA methylation levels were indeed substantially reduced. Remarkably, 80% of these hypomethylated mice developed thymus gland lymphomas at 4–8 months of age—a finding that again clearly supports the hypomethylation hypothesis. Altogether, the evidence for DNA hypomethylation as a mechanism of epigenetic carcinogenesis seems very strong to me.

B. The Histone Code As Peters et al. (2001) put it, ‘‘Recently, a ‘histone code’, hypothesis has been suggested (Strahl and Allis, 2000), which predicts that different modifications (e.g., acetylation, phosphorylation, methylation) of histone N termini are interdependent and represent an evolutionarily conserved mechanism that can induce and stabilize functionally distinct chromosomeal subdomains (Jenuwein and Allis, 2001).’’ Peters et al. (2001) also report the creation of mice deficient in a particular histone methyltransferase, and they used antibodies to indicate reduced methylation of key histones in the heterochromatin of these mice. Interestingly, about a third of such mice develop B-cell lymphomas after about a year, cancers that apparently arose as a result of chromosome instabilities. These observations support the possibility of carcinogenesis initially induced by epigenetic changes in histone methylation and, thus, through the histone code. Additional support for this possibility is provided by numerous observations of a line of mouse fibroblasts (C3H 10 T1/2 cells) in Louis Mahadevan’s laboratory. Thus these include evidence for rapid histone modifications (e.g., histone 3 phosphorylation in 5 minutes) on the addition of substances that include a carcinogenic phorbol ester (Mahadevan et al., 1991). Moreover, they also report rapid activation of the proto-oncogenes c-fos and c-jun (Thomson et al., 2001). Altogether, the evidence for epigenetic carcinogenesis via the histone code seems to be substantial.

C. Tissue Disorganization Sonnenschein and Soto (1999, 2000) have proposed that carcinogens need not directly affect the parenchyma or characteristic tissues of an organ. Rather, they directly affect its stroma or connective tissue so as to disrupt

224

Lionel F. Jaffe

the stroma-parenchyma connection and thereby induce cancer in the parenchyma. I would assume that the ‘‘transformed’’ stroma induces parenchymal cancer by an initially epigenetic mechanism. So, ‘‘tissue disorganization’’ may be considered to be a distinct mechanism of epigenetic carcinogenesis. Strong evidence for this theory comes from old studies—particularly those of J. W. Orr’s group from 1934 to 1972—on the mechanisms of chemical carcinogenesis in the skins of mice and men. Orr and Spencer (1972) reviewed the compelling evidence that a single application of methylcholanthrene can directly affect the dermis (which is the skin’s stroma). If the skin is later likewise treated with croton oil, then much later—when the carcinogens are gone—the dermis somehow frequently induces tumors within sheets of laterally regenerated epidermis (which is the skin’s parenchyma) or within sheets of transplanted epidermis that have never seen the carcinogens and have come to overlie the ‘‘transformed’’ stroma. Moreover, if a sheet of carcinogen-treated epidermis is transplanted to an untreated site on the other side of the body, it does not develop tumors even when subsequently treated with croton oil, whereas whole-skin grafts do (Billingham et al., 1951; Marchant and Orr, 1955; Orr, 1955). Strong modern evidence for the tissue disorganization theory is provided by data showing that tumorigenic substances disrupt gap junctions and that such disruption is tumorigenic, whereas the artificial maintenance of gap junction communication is antioncogenic. Table III summarizes the main available data, which show the disruption of gap junctions by tumorigenic substrances. In the most effective studies, such disruption as been shown to start in only 10–20 minutes in vitro and in only 3–7 days in vivo. An especially revealing study of such disruption in a dish was provided by Fitzgerald et al. (1983). Not only did they show the high speed of disruption, they also showed that it could be fully reversed in 100 minutes just by washing out the phorbol. Two particularly revealing studies of such disruption in vivo were those of Krutoskikh et al. (1995) on rat liver slices and of Mally and Chipman (2002) on a variety of organs within whole rats. In the Krutoskikh study, rats were chronically force fed phenobarbital, polychlorinated biphenyls, dichlorodiphenyl-trichloroethane (DDT), or clofibrate and gap junction communication assayed in liver slices As early as 1 week after the force feeding of carcinogens was begun, such communication was found to have been substantially disrupted—a process associated both with fewer gap junctions and with the appearance of connexin 32 (a major gap junction protein) in the cytoplasm rather than the cell membrane. Mally and Chipman introduced one of six ‘‘nongenotoxic’’ oncogens—namely, dioxin, methapyrilene, hexachlorobenzene, chloroform, p-dichlorobenzene, and d-limonene—into the stomachs of rats. With the understandable exception of d-limonene, they all acted to reduce gap

225

Epigenetic Theories of Cancer

Table III

The Disruption of Gap Junctions by Tumorigenic Compounds (shown in Chronological Order) Carcinogen

System

When Seen

Reference

Phorbol ester Phorbol ester Phorbol ester Phorbol ester Phorbol ester Initiator/promoter ras protein DDT Initiator/promoter Initiator/promoter Dioxin Barbiturates Phorbol Barbiturates DDT Six (see text!)

Hamster cells Mouse epidermal cells Mouse epidermal cells Mouse and hamster cells Mouse epidermal cells Whole rat liver Liver cells Whole rat liver Whole rat’s liver Whole rat’s liver Rat hepatocytes Rat hepatocytes Rat liver line Rat liver line Rat liver line Whole rats

1 wk 4h 10 min 20 min 10 min 4 wk 1d 1 wk 14 mon 1 wk <4 h <1 h <2 h <2 h <2 h 3

Yotti et al. (1979) Murray and Fitzgerald (1979) Fitzgerald et al. (1983) Enamoto et al. (1984) Gainer and Murray (1985) Krutoskikh et al. (1991) de Feijter et al. (1992) Tateno et al. (1994) Neveau et al. (1994) Krutoskikh et al. (1995) Baker et al. (1995) Ren and Ruch (1996) Ren et al. (1998) Ren et al. (1998) Ren et al. (1998) Mally and Chipman (2002)

junction frequencies (as evaluated via connexin-32) in their respective target organs, which included the liver, kidney, and thyroid. By far the strongest evidence for the reverse—that gap junction disruption is tumorigenic—lies in the report of Temme et al. (1997). It reports the study of mice that were genetically deficient in connexin-32. Male and female mice with this deficiency showed 25 times or eight times higher rates of spontaneous liver tumor formation, respectively. Some of the considerable evidence for the well-known proposition that gap junction maintenance or restitution suppresses tumorigenicity is reviewed in Mehta et al. (1999), in Sulkowski et al. (1999), and in Yamasaki et al. (1999). None of this suppression is quite complete. Nevertheless, some impressive examples of it have been reported. See, for example, Rose et al. (1993), Hirschi et al. (1993), and Proulx et al. (1997). Thus, Rose et al. (1993) transferred a connexin-43 gene into a line of tumorigenic mouse cells so as to increase cell–cell communication. These cells were then unilaterally injected into nude mice. After 11 weeks, gross tumors appeared on the sides of the mice that had been injected with connexin-transfected cells but not on the control sides. However, some very limited escape from tumor suppression was seen at 15 weeks.

226

Lionel F. Jaffe

IV. CONCLUDING REMARKS Seen from the perspective of a developmental biologist, the evidence for an epigenetic or at least nonmutational origin of almost all cancers seems compelling. If the mutationists could demonstrate the production of tumorigenic mutations soon after tumorigenic insults—perhaps by improved polymerase chain reaction—then those of the epigenetic persuasion would have to change our views; however, I obviously consider that to be unlikely. This article argues for epigenetic mechanisms of cancer initiation that are based on DNA demethylation, on changes in the pattern of histone adducts (the histone code) that likewise feature demethylation, and above all, on forms of tissue disorganization that feature gap junction disruption. A companion paper will argue that the chronic injuries that are well known to induce most human cancers act to bring about such broad changes via chronic increases in cell calcium.

ACKNOWLEDGMENTS I thank the director of the Marine Biological Laboratory, for his steadfast support despite the absence of external funding, and the whole MBL library staff, for their excellent help on a very demanding project.

REFERENCES Ananthaswamy, H. N., Ouhtit, A., Evans, R. L., Gorny, A., Khaskina, P., Sands, A. T., and Conti, C. C. (1999). Oncogene 18, 7395–7398. Azarschab, P., Porschen, R., Gregor, M., Blin, N., and Holzmann, K. (2002). Genes Chromosomes Cancer 35, 121–126. Baker, T. K., Kwiatowski, P. K., Madhukar, B. V., and Klaunig, J. E. (1995). Carcinogenesis 16, 2321–2326. Bakin, A. V., and Curran, T. (1999). Science 283, 387–390. Balmain, A. (2001). Nat. Rev. Cancer 1, 77–82. Barbacid, M. (1986). Carcinogenesis 7, 1037–1042. Belinsky, S. A., Nikula, K. J., Baylin, S. B., and Issa, J.-P. J. (1996). Proc. Natl. Acad. Sci. USA 93, 4045–4050. Berenblum, I., and Shubik, P. (1949). Br. J. Cancer 3, 109–118. Billingham, R. E., Orr, J. W., and Woodhouse, D. L. (1951). Br. J. Cancer 5, 417–. Bohn, G. (1903). C.R. Acad. Sci. Paris 136, 1085–1086. Bohr, V. A., Smith, C. A., Okumoto, D. S., and Hanawalt, P. C. (1985). Cell 40, 359–369. Bravo, R., Burckhardt, J., Curran, T., and Mu¨ller, R. (1985). EMBO J. 4, 1193–1197. Caggana, M., and Kennedy, A. R. (1989). Carcinogenesis 10, 2145–2148. Campisi, J., Gray, H. E., Pardee, A. B., Dean, M., and Sonenshein. (1984). Cell 36, 241–247.

Epigenetic Theories of Cancer

227

Chow, M., Yao, A., and Rubin, H. (1994). Proc. Natl. Acad. Sci. USA 91, 599–603. Cleaver, J. E., and Mitchell, D. L. (2000). In Cancer Medicine 5th ed. B. C. Decker Hamilton. Cochran, B. H., Zullo, J., Verma, I. M., and Stiles, C. D. (1984). Science 226, 1080–1082. Curran, T., and Morgan, J. I. (1985). Science 229, 1265–1268. de Feijter, A. W., Trosko, J. E., Krizman, D. B., Lebovitz, R. M., and Lieberman, M. W. (1992). Mol. Carcinog. 5, 205–212. de Laat, W. L., Jaspers, N. G. C., and Hoeijmakers, J. H. J. (1999). Genes Dev. 13, 768–785. Dennis, C. (2003). Nature 421, 686–688. Deshpande, A., Goodwin, E. H., Bailey, S. M., Marrone, B. L., and Lehnert, B. E. (1996). Radiat. Res. 145, 260–267. Devary, Y., Gottlieb, R. A., Smeal, T., and Karin, M. (1992). Cell 71, 1081–1091. Devary, Y., Rosette, C., DiDonato, and Karin, M. (1993). Science 261, 1442–1446. Downward, J., Graves, J. D., Warne, P. H., Rayter, S., and Cantrell, D. A. (1990). Nature 346, 719–723. Enamoto, T., Martel, N., Kanno, Y., and Yamasaki, H. (1984). J. Cell. Physiol. 121, 323–333. Farber, E. (1963). Adv. Cancer Res. 7, 383–474. Farber, E. (1996). Adv. Cancer Res. 70, 21–48. Farber, E., and Rubin, H. (1991). Cancer Res. 51, 2751–2761. Fernandez, A., Mondal, S., and Heidelberger, C. (1980). Proc. Natl. Acad. Sci. USA 77, 7272–7276. Fischer, B. (1906). Munch. Med. Watch. 53, 2041. Fitzgerald, D. J., Knowles, S. E., Ballard, J., and Murray, A. W. (1983). Cancer Res. 43, 3614–3618. Gainer, H. St. C., and Murray, A. W. (1985). Biochem. Biophys. Res. Commun. 126, 1109–1113. Gaudet, F., Hodgson, G., Eden, A., Jackson-Grusby, L., Dausman, J., Gray, J. W., Leonhardt, H., and Jaenisch, R. (2003). Science 300, 489–492. Giese, A. C. (1947). Quart. Rev. Biol. 22, 253–282. Graves, J. D., Downward, J., Rayter, S., Warne, P., Tutt, A. L., Glennie, M., and Cantrell, D. A. (1991). J. Immunol. 146, 3709–3712. Gray, P. J., Cullinane, C., and Phillips, D. R. (1991). Biochemistry 30, 8036–8040. Greenberg, M. E., Greene, L. A., and Ziff, E. B. (1985). J. Biol. Chem. 260, 14101–14110. Greenberg, M. E., and Ziff, E. B. (1984). Nature 311, 433–438. Greenberg, M. E., Ziff, E. B., and Greene, L. A. (1986). Science 234, 80–83. Griesser, J., Kaufmann, D., Maier, B., Mailhammer, R., Kuehl, P., and Krone, W. (1997). J. Invest. Dermatol. 108, 275–280. Grigoriadis, A. E., Wang, Z.-Q., and Wagner, E. F. (1995). Trends Genet. 11, 436–441. Grosovsky, A. J. (1999). Proc. Natl. Acad. Sci. USA 96, 5346–5347. Gutmann, D. H., Tennekoon, G. I., Cole, J. L., Collins, F. S., and Rutkowski, J. L. (1993). J. Neurosc. Res. 36, 216–223. Hanahan, D., and Weinberg, R. A. (2000). Cell 100, 57–70. Harvey, E. B. (1956). The American Arbacia and Other Sea Urchins. Princeton University Press, Princeton, NJ. Hesketh, R. (1994). The Oncogene Handbook. Academic Press, London. Higuchi, Y., Ito, M., Tajima, M., Higuchi, S., Miyamoto, N., Nishio, M., Kawano, M., Kusagawa, S., Tsurudome, M., Sudo, A., Katou, K., Uchida, A., and Ito, Y. (1999). Bone 25, 17–24. Holliday, R. (1979). Br. J. Cancer 40, 513–522. Holliday, R. (1996). Cancer Surv. 28, 103–115. Hong, J.-H., Chiang, C.-S., Sun, J.-R., Withers, H. R., and McBride, W. H. (1997). Mol. Brain Res. 46, 223–228.

228

Lionel F. Jaffe

Howell, A. N., and Sager, R. (1978). Proc. Natl. Acad. Sci. USA 75, 2358–2362. Ichikawa, T., Yano, Y., Uchida, M., Otani, S., Hagiwara, K., and Yano, T. (1996). Cancer Lett. 107, 165–170. Israel, B. A., and Schaeffer, W. I. (1987). In Vitro Cell. Dev. Biol. 24, 627–632. Israel, B. A., and Schaeffer, W. I. (1988). In Vitro Cell. Dev. Biol. 24, 487–490. Jaffe, L. A., ed. (1993) Dev. Biol. 158, 9–78. Jaffe, L. F. (1982). In Ions Cell Proliferation and Cancer, pp. 302–310, Academic Press, NY. Jaffe, L. F. (1991). Proc. Natl. Acad. Sci. USA 88, 9883–9887. Jenuwein, T., and Allis, C. D. (2001). Science 293, 1074–1080. Jonasson, J., and Harris, H. (1977). J. Cell Sci. 34, 255–263. Jones, M. K., Itani, R. M., Wang, H., Tomikawa, M., Sarfeh, I. J., Szabo, S., and Tarnawski, S. S. (1999). Am. J. Physiol. 276, G1345–G1355. Jones, P. A., and Buckley, D. (1990). Adv. Cancer Res. 54, 1–22. Jox, A., Taquia, E., Vockerodt, M., Draube, A., Pawlita, M., Mo¨ller, P., Bullerdiek, J., Diehl, V., and Wolf, J. (1998). Cancer Res. 58, 4930–4939. Kaibuchi, K., Tsuda, T., Kikuchi, A., Tanimoto, T., Yamashita, T., and Takai, Y. (1986). J. Biol. Chem. 261, 1187–1192. Kalthoff, K. (1996) Analysis of Biological Development. McGraw-Hill, New York. Kelly, K., Cochran, B. H., Stilesd, C. D., and Leder, P. (1983). Cell 35, 603–10. Kennedy, A. R., Fox, M., Murphy, G., and Little, J. B. (1980). Proc. Natl. Acad. Sci. USA 77, 7262–7266. Kennedy, A. (1985). Carcinogenesis (Lond.) 6, 1441–1445. Knudson, A. G. (1997) In Origins of Human Cancer, pp. 45 52. Cold Spring Harbor Press, Cold Spring Harbor, NY. Koura, M., Isaka, H., Yoshida, M. C., Tosu, M., and Sekiguchi, T. (1982). Gann 73, 574–580. Kruijer, W., Cooper, J. A., Hunter, T., and Verma, I. M. (1984). Nature 312, 711–716. Kumar, R., Sukumar, S., and Barbacid, M. (1990). Science 248, 1101–1104. Krutovskikh, V. A., Nesnil, M., Mazzoleni, G., and Yamasaki, H. (1995). Lab. Invest. 72, 571–577. Krutovskikh, V. A., Oyamada, M., and Yamasaki, H. (1991). Carcinogenesis 12, 1701–1706. Landolph, J. R., Verma, A., Ramnath, J., and Clemens, F. (2002). Environ. Health Perspect. 10(Suppl 5), 845–850. Lavrovsky, V. A., Guvakova, M. A., and Lavrovsky. (1992). Eur. J. Cancer 28, 17–21. Lehtonen, E., Laasonen, A., and Tienari, J. (1989). Int. J. Dev. Biol. 33, 105–151. Little, J. B., Nagasawa, H., Pfenning, T., and Vetrous, H. (1997). Radiat. Res. 148, 299–307. Little, J. B. (2000). Carcinogenesis 21, 397–404. Lubin, J. H., and Boice, J. D., Jr. (1997). J. Natl. Cancer Inst. 89, 49–57. MacLeod, M. C. (1996). Mol. Carcinog. 15, 241–250. Mahadevan, L. C., Willis, A. C., and Barratt, M. J. (1991). Cell 65, 775–783. Mally, A., and Chipman, J. K. (2002). Toxicology 180, 233–248. Marchant, J., and Orr, J. W. (1955). Brit J. Cancer 9, 128–136. Mehta, P. P., Perez-Stable, C., Nadji, M., Mian, M., Asotra, K., and Roos, B. A. (1999). Dev. Genet. 24, 91–110. Mellon, I., Bohr, V. A., Smith, C. A., and Hanawalt, P. C. (1986). Proc. Natl. Acad. Sci. USA 83, 8878–8882. Messadi, D. V., Billings, P., Shklar, G., and Kennedy, A. R. (1986). J. Natl. Cancer Inst. 76, 447–452. Miao, G. G., and Curran, T. (1994). Mol. Cell. Biol. 14, 4295–4310. Miller, R. W. (1977). In Origins of Human Cancer, 381–388. Cold Spring Harbor Press Cold Spring Harbor, NY. Mintz, B., and Illmensee, K. (1975). Proc. Natl. Acad. Sci. USA 72, 3585–3589.

Epigenetic Theories of Cancer

229

Mondal, S., and Heidelberger, C. (1970). Proc. Natl. Acad. Sci. USA 65, 219–225. Morgan, J. I., Cohen, D. R., Hempstead, J. L., and Curran, T. (1987). Science 237, 192–197. Morgan, J. I., and Curran, T. (1986). Nature 322, 552–555. Morgan, J. I., and Curran, T. (1991). Annu. Rev. Neurosci. 14, 421–451. Mu¨ller, R., Bravo, R., Burckhardt, J., and Curran, T. (1984). Nature 312, 716–720. Mu¨ller, R., Curran, T., Mu¨ller, D., and Guilbert, L. (1985). Nature 314, 546–548. Murray, A. W., and Fitzergerald, D. J. (1979). Biochem. Biophys. Res. Commun. 91, 395–340. Nagasawa, H., and Little, J. B. (1992). Cancer Res. 52, 6394–6396. Nelson, M. A., Futscher, B. W., Kinsella, T., Wymer, J., and Bowden, G. T. (1992). Proc. Natl. Acad. Sci. USA 89, 6398–6402. Neveu, M. J., Hully, J. R., Babcock, K. L., Hertzberg, E. L., Nicholson, B. J., Paul, D. L., and Pitot, H. C. (1994). J. Cell Sci. 107, 83–95. Ohlsson, R., Kanduri, C., Whitehead, J., Pfeifer, S., Lobanenkov, V., and Feinberg, A. P. (2003). Adv. Cancer Res. 88, 146–168. Orr, J. W. (1955). Br. J. Cancer 9, 623–630. Orr, J. W., and Spencer, A. T. (1972) In Tissue Interactions in Carcinogenesis, pp. 291–304. Academic Press, London. Ouhtit, A., Muller, H. K., Davis, D. W., Ullrich, S. E., McConkey, D., and Ananthaswamy, H. N. (2000). Am. J. Pathol. 156, 201–207. Papaioannou, V. E. (1993). Int. J. Dev. Biol. 37, 33–37. Pellerin, I., Vuillermoz, C., Jouvenot, M., Royez, M., Ordener, C., Marechal, G., and Adessi, G. (1992). Endocrinology 131, 1094–1100. Peters, A. H. F. M., O’Carroll, D., Scherthan, H., Mechtler, K., Sauer, S., Scho¨fer, C., Weipoltshammer, K., Pagani, M., Lachner, M., Kohlmaier, A., Opravil, S., Doyle, M., Sibilia, M., and Jenuwein, T. (2001). Cell 107, 323–337. Peto, R. (1997) In Origins of Human Cancer, pp. 1403. Cold Spring Harbor Press Cold Spring Harbor, NY. Pierce, G. B. (1967). Curr. Topics Dev. Biol. 2, 223–246. Pierce, D. A., Shimizu, Y., Preston, D. L., Vaeth, M., and Mabuchi, K. (1996). Radiat. Res. 146, 1–27. Prehn, R. T. (1994). Cancers beget mutations versus mutations beget cancers. Cancer Res. 54, 5296–5300. Proulx, A. A., Lin, Z. X., and Naus, C. C. (1997). Cell Growth Differ. 8, 533–540. Puga, A., Nebert, D. W., and Carrier, F. (1992). DNA Cell Biol. 11, 269–281. Rainier, S., and Feinberg, A. P. (1988). Proc. Natl. Acad. Sci. USA 85, 6384–6388. Ren, P., Mehta, P. P., and Ruch, R. J. (1998). Carcinogenesis 19, 169–175. Ren, P., and Ruch, R. J. (1996). Carcinogenesis 17, 2119–2124. Rheinwald, J. G., and Greeen, H. (1975). Cell 6, 331–343. Riggs, A. D., and Jones, P. A. (1983). Adv. Cancer Res. 40, 1–30. Rose, B., Mehta, P. P., and Loewenstein, W. R. (1993). Carcinogenesis 14, 1073–1075. Rous, P. (1959). Nature 183, 1357–1361. Saffen, D. W., Cole, A. J., Worley, P. F., Christy, B. A., Ryder, K., and Baraban, J. M. (1988). Proc. Natl. Acad. Sci. USA 85, 795–799. Scho¨nthal, A., Sugarman, J., Brown, J. H., Hanley, M. R., and Feramisco, J. R. (1991). Proc. Natl. Acad. Sci. USA 88, 706–710. Schumacher, C., Becker, I., Oswald, S., Atkinson, M. J., Nekarda, H., Becker, K. F., Mueller, J., Siewert, J. R., and Hofler, H. (1999). Virchows Arch. 434, 489–495. Shay, J. W. (1983). Mol. Cell. Biochem. 57, 17–26. Shay, J. W., and Werbin, H. (1988). Cancer Res. 48, 830–833. Song, B., Zhao, M., Forrester, J. V., and McCaig, C. D. (2002). Proc. Nat. Acad. Sci. USA 99, 13577–13582.

230

Lionel F. Jaffe

Sonnenberg, J. L., Macgregor-Leon, P. F., Curran, T., and Morgan, J. I. (1989). Neuron 3, 359–365. Sonnenschein, C., and Soto, A. M. (1999). The Society of Cells. Springer, New York. Sonnenschein, C., and Soto, A. M. (2000). Mol. Carcinogenesis 29, 205–211. St. Clair, W. H., Billings, P. C., and Kennedy, A. R. (1990). Cancer Lett. 52, 145–152. Stehelin, D., Varmus, H. E., and Bishop, J. M. (1976). Nature 260, 170–173. Stern, R. G., Milestone, B. N., and Getenby, R. A. (1999). Med. Hypotheses 52, 367–372. Strahl, B. D., and Allis, C. D. (2000). Nature 403, 41–45. Sulkowski, S., Sulkowska, M., and Skrzydlewska (1999). Pol. J. Pathol. 50, 227–233. Takamatsu, M., Nagamine, Y., and Farber, E. (1983). Cancer Research 43, 5049–5058. Tateno, C., Ito, S., Tanaka, M., Oyamada, M., and Yoshitake, A. (1994). Carcinogenesis 15, 517–521. Temme, A., Buchmann, A., Gabriel, H.-D., Nelles, E., Schwarz, M., and Willecke, K. (1997). Curr. Biol. 7, 713–716. Thomson, S., Clayton, A. H. L., and Mahadeven, L. C. (2001). Mol. Cell 8, 1231–1241. Thomson, S., Mahadevan, L. C., and Clayton, A. L. (1999). Semin. Cell Dev. Biol. 10, 205–214. Tomlinson, I. P., Roylance, R., and Houlston, R. S. (2001). J. Med. Genet. 38, 81–85. Todd, D. G., and Mikkelsen, R. B. (1994). Cancer Res. 54, 5224–5230. Tsuda, T., Kaibuchi, K., Kawahara, Y., Fukuzaki, H., and Takai, Y. (1985). FEBS Lett. 191, 205–210. Wainfan, E., and Poirier, L. A. (1992). Cancer Res. 52, 2071s–2077s. Watson, G. E., Lorimore, S. A., Macdonald, D. A., and Wright, E. G. (2000). Cancer Res. 60, 5608–5611. Weinberg, R. A. (1989). Oncogenes and the Molecular Origins of Cancer. Cold Spring Harbor Press, Cold Spring Harbor, NY. Weston, A., and Harris, C. C. (2000) Chemical carcinogenesis. In Cancer Medicine 5th ed., B. C. Decker, Hamilton, London. Wu, L.-J., Randers-Pehrson, G., Xu, A., Waldren, C. A., Geard, C. R., Yu, Z.-L., and Hei, T. H. (1999). Proc. Natl. Acad. Sci. USA 96, 4959–4964. Yamasaki, H., and Naus, C. C. G. (1996). Carcinogenesis. 17, 1199–1213. Yamashita, S., Takayanagi, A., and Shimizu, N. (2001). Histol. Histopathol. 16, 131–140. Yano, T., and Yuasa, M. (1997). Anal. Biochem. 244, 187–189. Yotti, L. P., Chang, C. C., and Trosko, J. E. (1979). Science 206, 1089–1091.

Index

A Actinic keratosis, 14 Affymetrix oligonucleotide arrays, 116 African enigma, 65 Alveolar nodules, hyperplastic mammary tumor virus and, 41 Anaplastic oligodendrogliomas (AO), 118 Aneuploid human keratinocytes (HaCaT), 9–10, 10t Aniridia, 102 Anisomycin, 220t APC gene, mutation of, 2 Archival tissue material analysis comparative genome hybridization array technology and, 120–21 Avian fibroblasts, 29–30

B BACs. See Bacterial artificial chromosomes Bacterial artificial chromosomes (BACs) resource development of, 96–97 Bacterial colonization of Helicobacter pylori, 67–71 Balb/c, 44 Basal cell carcinoma (BCC), cause of, 13 Basement membrane, 13, 46 BCC. See Basal cell carcinoma Beckwith-Wiedemann syndrome (BWS), 101, 105 Benzo(a)pyrene (B(a)P), 3 B(a)P. See Benzo(a)pyrene Brain tumors comparative genome hybridization array technology and, 109–11, 110f Breast cancer, 5, 40–48 Englebreth-Holm-Swarm and, 46 hyperplastic alveolar nodules and, 41–42 mammary gland development and, 41 mammary tumor virus and, 42, 44

N-nitroso-N-methylurea and, 45 pregnancy and, 41 tissue organization and, 42 BWS. See Beckwith-Wiedemann syndrome

C cag-Pathogenicity island (PAI)-positive strains, 68, 78, 79 Cancer age and, 186 breast, 5, 40–48. See also Breast cancer cell change character of, 212–14 cell parts and, 218, 221 cervical, 34 colon, 2 colorectal, 2 comparative genome hybridization array technology and, 101 demonstrable mutations, 214–15, 214t epithelial, 211f frequency of, 212–14 fused cells and, 218, 221 gastric. See Gastric cancer genetic abnormalities and, 4–5 ionizing radiation and, 215–17 kidney, 102 lung, 2 malignant transformation and, 185–86 natural killer cells in, 130–34 nongenetic, 211f oral, 16 reversibility of, 212–14 skin, 2 sunscreen and, 14 vaccine development, 158 Cancer immune surveillance, 185–86 Carcinogenesis, proteases in, 48–51 Carcinogens, tumors and, 3–5 Card15/NOD2, 74

231

232 Cathepsin D, 50 CCL1, 130 CCL3, 130 CCL4, 130 CCL5, 130 CCL22, 130 CD4, 133, 166, 167, 171, 185, 196 CD8, 132, 133, 135, 139, 145, 159, 160, 161, 166, 167, 172, 185, 196 CD28, 143 CD44, 115 CD54, 136 CD56, 129 CD80, 132 CD158, 139 CD244, 135, 142 CD16A, 129 CDK4, 159, 160 CDKN2A/B, 115, 118 Cell-intrinsic control mechanisms, 186 Cells activation of T, 188–89 C3H 10T1/2, 32–37 dendritic, 188 endocrine, 66f endothelial, 66f fused cancer, 216 gastric epithelial, 66f gastric stem, 66f liver epithelial, 42 mucus, 66f natural killer. See Natural killer cells parietal, 66f polymorphonuclear, 74 transplanted tumor, 197–98 V79 Chinese hamster, 36 Cervical cancer, 34 C-fos, 219t, 220t CGH. See Comparative genome hybridization C3H 10T1/2 cells, 32–37 c-Harvey-ras oncogene, 9 Chemical carcinogens, 181–83, 183f Chromosome aberrations, 36 Chronic myeloid leukemia (CML), 190 C-jun, 219t, 220t c-Jun amnio-terminal kinase, 78 CML. See Chronic myeloid leukemia C-myc, 219t, 220t Coal tar isolation from, 3

Index papillomas and, 3 tumors and, 3 Colorectal cancer monoclonal, 2 polyclonal, 2 Comparative genome hybridization (CGH), 95, 95f Comparative genome hybridization (CGH) array technology archival tissue material analysis and, 120–21, 121f brain tumor analysis and, 109–11 cancer analysis and, 101 conventional karyotyping and, 107–9, 108f development of, 97–101, 100f, 101f genetic changes identified using, 111, 112f heterozygosity loss and, 116, 118, 120 tumor analysis and, 106 Constitutional chromosome abnormalities, 101–6, 103f Beckwith-Wiedemann syndrome, 101 Perlman’s syndrome, 102, 105 retinoblastoma, 101 Wilms tumor, 101 Crohn’s disease, 73 Croton oil, 4 C-srcd, 220t CTLs. See Cytotoxic T lymphocytes CXCL8, 130 CXCR1, 73 CXCR2, 73 Cyclic nucleotides, 31 Cyclooxigenase 2, 78 Cytotoxic T lymphocytes (CTLs), 128, 160 Cytotoxin, vacuolating of Helicobacter pylori, 69

D DAP10, 43 DAP12, 43 DAP. See Diaminopimelic acid DCs. See Dendritic cells Dendritic cells (DCs), 188 Diaminopimelic acid (DAP), 74 7,12-dimethylbenzanthracene (DMBA), 5, 181, 214t Dioxin, 220t DMBA. See 7,12-dimethylbenzanthracene DNA methylation, 222–23

233

Index Drosophila melanogaster, 72 Ductal dysplasia, frequency of, 44, 44t, 45t

E E-cadherin, 9 Ect-2, 220t EGF. See Epidermal growth factor EHS. See Englebreth-Holm-Swarm Endocrine cell, 66f Endothelial cell, 66f Englebreth-Holm-Swarm (EHS), 46, 47 Epidermal growth factor (EGF), 47 Epigenesis, mechanisms of DNA methylation, 222–23 histone code, 223 tissue disorganization, 223–25, 225t Epithelial cancers, 211f Escherichia coli, 98 Estrogen, 220t Extracellular matrix (ECM), 48–50, 195, 197f

F Fas ligand (FasL), 75, 130 mediated apoptosis, 137 FasL. See Fas ligand Fibroblasts avian, 29–30 mammalian, 29–30 neolatic transformation among, 17 12-Ø-tetradecanoyl phorbol-13-acetate promotion of, 32–35 Rous Sarcoma Virus and, 17 transformation of, 37–40, 38f, 40f Fibrosis, methylcholanthrene and, 195–97 Field cancerization, 16 FISH. See Fluorescence in situ hybridization FJX1, 15 Fluorescence in situ hybridization (FISH), 94, 95 Forskolin, 220t Fruit fly, 72

G Gastric Gastric Gastric Gastric

adenocarcinoma, 64, 66f, 78 cancer. See Helicobacter pylori epithelial cell, 66f stem cell, 66f

GBAS. See Glioblastoma amplified sequence Gene amplification, 16 Germ cell-cancer antigens, 159 Glioblastoma amplified sequence (GBAS), 109 Glycolipid lipopolysaccharide (LPS), 72 Glycosylphosphatidylinositol (GPI) linked, 144 GPI. See Glycosylphosphatidylinositol linked Grafting, 8–9, 11 12-Ø-tetradecanoyl phorbol-13-acetate and, 8

H HaCaT. See Aneuploid human keratinocytes HAN. See Hyperplastic alveolar nodules Harvey sarcoma virus (HaSV), 11–12 HaSV. See Harvey sarcoma virus Heat shock proteins (HSPs) peptides, 174–75 Helicobacter hepaticus, 194 Helicobacter pylori, 66f, 194 bacterial colonization of, 67–71 environmental risk factors and, 81–82 epithelial cell turnover and, 76–77 geography and, 64 host susceptibility genes and, 81–82 infection with, 64–65, 67 intestinal metaplasia and, 64 mediated mucosal inflammation, 71–75 Mongolian gerbil and, 80 mutagenesis of, 70 pathophysiology of, 64, 66f peptic ulcer disease and, 64 persistence of, 67–71 tumor development and, 77–80 vacuolating cytotoxin of, 69 Heteroclitic peptides, 161–64, 162f, 163f, 164f, 172–73 Heterozygosity, loss of comparative genome hybridization array technology and, 116 HHV8. See Human herpesvirus 8 Histone code, 223 HLA-A3, 188 HLA-B8, 188 HSPs. See Heat shock proteins peptides Human herpesvirus 8 (HHV8), 135 Hypergastremia, 78–79 Hyperplastic alveolar nodules (HAN), 41 growth of, 41–42, 42t

234

I IFA. See Incomplete Freund’s adjuvant IFNA. See Interferon gene cluster IL-2. See Interleukin-2 IL-4. See Interleukin-4 IL-5. See Interleukin-5 IL-7. See Interleukin-7 IL-8. See Interleukin-8 IL-10. See Interleukin-10 IL-12. See Interleukin-12 IL-13. See Interleukin-13 IL-15. See Interleukin-15 Immunization active, 169–70 passive, 170–72, 171f Incomplete Freund’s adjuvant (IFA), 173 INF-. See Interferon- Injections, subcutaneous malignant keratinocytes in, 11–13 INS-GAS. See Transgenic Hypergastremia Interferon gene cluster (IFNA), 115 Interferon- (INF-) mediated effector functions, 138 transplanted tumor cells and, 197–98 Interleukin-2 (IL-2), 131, 134, 137 Interleukin-4 (IL-4), 167 Interleukin-5 (IL-5), 130 Interleukin-7 (IL-7), 129 Interleukin-8 (IL-8), 73, 74 Interleukin-10 (IL-10), 130 Interleukin-12 (IL-12), 131, 133 Interleukin-13 (IL-13), 130 Interleukin-15 (IL-15), 129, 131, 134, 137 Interleukin-1 receptor antagonist (IL-1 RA), 81 Intestinal metaplasia, 66f Helicobacter pylori and, 64

J JunB, 218t

K Kaposi’s sarcoma, 135 Karyotyping, conventional comparative genome hybridization array technology and, 107–9 Kenogeneic immunization, 165–68, 165f, 166f, 167f, 168f

Index Keratinocytes initiated, 5–7 neoplastic progression by, 8–11 subcutaneous injections and, 11–13 ultraviolet carcinogenesis and, 13–17 virally transformed malignant, 11–13 KIAA1354, 115 Kidney, cancer of, 102 Knock-out (KO) mice, 178–79 KO. See Knock-out mice

L LA. See Lobuloalveolar structures, 41 LAK. See Lymphokine-activated killer Lewis carbohydrate epitopes, 79 Liver, epithelial cells of, 42 Lobuloalveolar (LA) structures, 41 Loss of heterozygous (LOH), comparative genome hybridization and, 116, 118, 120 LPS. See Glycolipid lipopolysaccharide LTA. See Lymphocytoxin- Lung cancer, 2 Ly49, 39 Lymphocyte, 66f Lymphocytoxin (LTA)-, 81 Lymphokine-activated killer (LAK), 131

M Macrophage, 66f Major histocompatibility complex (MHC), 159 class I receptors, 139–40 MALT. See Mucosa-associated lymphoid tissue Mammalian fibroblasts, 29–30 chemical carcinogenesis of, 30–32 Mammary neoplasia in vitro results, 45–48 in vivo results, 40–45 Mammary tumor virus (MTV), 43t, 44t breast cancer and, 42, 44 hyperplastic alveolar nodules and, 41 infection of, 41 Matrix metalloproteinases (MMPs), 26, 49–51 function of, 49 membrane-type, 49–50 occurrence of, 49

235

Index MCA. See Methylcholanthrene Melanoma antibody-mediated immunity in, 169–72 differentiation antigens, 160 Melanoma antigens atypical gene products, 159–60 genetic mutations, 159–60 germ cell-center, 159 Membrane-type matrix metalloproteinases (MT-MMPs), 49–50 Memorial Sloan-Kettering Cancer Center (MSKCC), 160 Mental retardation, 102, 104 Methylcholanthrene (MCA), 33 fibrosis and, 195–97 local events and, 202f tissue damage and, 195–97, 197f tissue repair response and, 195–97 MHC. See Major histocompatibility complex MMPs. See Matrix metalloproteinases MNNG. See N-methyl-N1nitro-Nnitrosoquanidine Molecular cytogenetics, evolution of, 94–96 Moloney tumor virus-induced lymphoma, 130 Mongolian gerbil, Helicobacter pylori and, 80 Monoclonal antibody (mAb), 195 Monoclonal colorectal cancer, 2 MSKCC. See Memorial Sloan-Kettering Cancer Center MT-MMPs. See Membrane-type matrix metalloproteinases MTV. See Mammary tumor virus MUC1. See Mucosa protecting mucin Mucosa protecting mucin (MUC1), 81 Mucosa-associated lymphoid tissue (MALT), 64, 71 Mucus cell, 66f Murine leukemia virus, 11–12 Mutagenesis antipain on, 36 of Helicobacter pylori, 70 Mutations of APC gene, 2 cancer, 214–15 genetic melanoma antigens, 159–60 keratinocytes and, 13–17 oncogenic, 217–18

polycyclic aromatic hydrocarbon and, 4 UV-induced, 14

N Natural cytotoxicity receptors (NCRs), 142 Natural killer (NK) cells biology of, 130 in cancer, 130–34 effector functions, 136–38 receptor activation of, 141–45 tumor escape mechanisms and, 146–47 in tumor immunosurveillance, 145–46 turning off of, 139–40 in viral immunity, 134–36 NCAs. See Numerical chromosome abnormalities NCRs. See Natural cytotoxicity receptors Neolatic transformation, among fibroblasts, 17 Neoplasia, mammary, 40–48. See also Mammary neoplasia NF1, 220t NHK. See Normal human keratinocytes Nickel compounds, 220t NIH 3T3, 37–40, 38f, 40f Nitrogen mustards, nonmutagenicity of, 221–22 NK. See Natural killer cells NKG2D, 143–45 N-methyl-N1nitro-N-nitrosoquanidine (MNNG), 36 NMU. See N-nitroso-N-methylurea N-nitroso-N-methylurea (NMU), 45 NOD. See Nucleotide-binding oligomerization domain Nongenetic cancer, 211f Normal human keratinocytes (NHK), 9–11, 10t, 13 12-Ø-tetradecanoyl phorbol-13-acetate and, 10–11 Nucleotide-binding oligomerization domain (NOD), 72, 73 Numerical chromosome abnormalities (NCAs), 110 gene content and, 115 overlapping of, 111–15, 112f, 113f, 114f NY-ESO-1, 59

236

P p53, 14, 15, 16, 159 PAHs. See Polycyclic aromatic hydrocarbons PAI. See cag-Pathogenicity island-positive strains, 68 PAMPs. See Pathogen-associated molecular patterns Papillomas coal tar and, 3 polycyclic aromatic hydrocarbon and, 4 Parietal cell, 66f Pathogen-associated molecular patterns (PAMPs), 71–72, 73 PAX6 gene, 102 PDX1, 15 Peptic ulcer disease, Helicobacter pylori and, 64 Peptide ligands, altered, 168–69 Peptidoglycan recognition proteins (PGRPs), 72 Perforin/granzyme-mediated cytotoxicity, 136–37 Perlman’s syndrome, 102, 105 PGRPs. See Peptidoglycan recognition proteins Phorbol ester, 219t, 225t Phospholipase A2, 78 p16/INK4, 159 PMNs. See Polymorphonuclear cells Polyclonal colorectal cancer, 2 Polycyclic aromatic hydrocarbons (PAHs), 3, 4, 181 12-Ø-Tetradecanoyl phorbol-13-acetate and, 7 Polymorphonuclear cells (PMNs), 74 Polymorphonuclear leukocyte, 66f Polyoma virus, 27–28 P21ras, 219t, 220t Pregnancy breast cancer and, 41 Proteases in carcinogenesis, 48–51 tumor initiation and, 35–37 Proto-oncogene activation, high speed of, 218, 219t–220t Proto-oncogene expression, X-ray speed effects on, 217

R Radiation, ionizing, 215–17 Rb. See Retinoblastoma

Index Retinoblastoma (Rb), 101, 102, 104 Rous Sarcoma Virus (RSV), 12, 29–30 Bryan high-iter strain of, 18 fibroblasts and, 17 growth curves of, 21f 12-Ø-tetradecanoyl phorbol-13-acetate and, 17 proteases release by, 25 serum type effect on, 20t suppression of, 19f, 25

S SCC. See Squamous cell carcinoma Shigella, 74 Sialylated glycoconjugates, 67–68 Sister chromatid exchanges, 36 Skin cancer, 2 SKY. See Specific karyotyping SLC1A2, 15 Specific karyotyping (SKY), 94 comparative genome hybridization and, 107–9, 108f Squamous cell carcinoma (SCC), cause of, 13 Sunlight, 13–15 Sunscreen, 14

T T-cell receptor (TCR), 160 T-cells, activation of, 188–89 TCR. See T-cell receptor 12-Ø-Tetradecanoyl phorbol-13-acetate (TPA), 4, 181–82 fibroblasts promotion by, 32–35 grafts and, 8 normal human keratinocytes and, 10–11 polycyclic aromatic hydrocarbon and, 7 Rous Sarcoma Virus and, 17 stimulatory effect of, 6, 6t ultraviolet, 32–35 TGF-. See Transforming growth factor- TGF-. See Transforming growth factor- Thapsigargin, 220t Tissue, disorganization of epigenesis and, 223–25 TLRs. See Toll-like receptors TNF-. See Tumor necrosis factor- TNF-related apoptosis-inducing ligand (TRAIL), 130 mediated apoptosis, 137

237

Index Toll-like receptors (TLRs) 1, 72 2, 72 3, 72 4, 72, 73, 74 5, 72 6, 72 9, 72 TPA. See 12-Ø-Tetradecanoyl phorbol-13-acetate TRAIL. See TNF-related apoptosis-inducing ligand Transforming growth factor-(TGF-), 8 Transforming growth factor- (TGF-), 12 Transgenic Hypergastremia (INS-GAS), 78 TRIM44, 15 Tumor necrosis factor- (TNF-), 81 Tumors. See also Cancer brain, 109–11, 10f carcinogens and, 3–5 coal tar and, 3 comparative genome hybridization array technology and, 106 development of, 2–3 escape mechanisms of, 146–47 Helicobacter pylori and, 77–80 immunosurveillance of, 145–46 initiation of, proteases and, 35–37 mammary virus, 41, 42, 43t, 44, 44t oncogenic mutation in, 217–18 proteases and, 35–37 rise of, 2 transplanted cells of, 197–98 Wilms, 101 Turpentine, 3 TWHRYHLL, 162 Tyrosinase-related protein (Tyrp)-1, 160 Tyrp-1. See Tyrosinase-related protein-1

U UL16-binding proteins (ULBPs), 144 ULBPs. See UL16-binding proteins Ultraviolet (UV) carcinogenesis mutant keratinocytes and, 13–17

Ultraviolet 12-Ø-tetradecanoyl phorbol-13-acetate (UV-TPA), 32–35 Urethane, topical, 214t UVB, 14 UV-induced mutations, 14 UV-TPA. See Ultraviolet 12-Ø-tetradecanoyl phorbol-13-acetate

V V79 Chinese hamster cells, 36 VacA, See Vacuolating cytotoxin Vaccines cancer and, 158 xenogeneic DNA, 173–74 Vacuolating cytotoxin (VacA), of Helicobacter pylori, 69, 70 Viral immunity, natural killer cells and, 134–36 Virchow, Rudolf, 63

W Wild type (WT) mice, 184–85 Wilms tumor (WT), 101 Wortmannin, 70 WT. See Wild type mice WT. See Wilms tumor

X XCL1, 130 Xenogeneic DNA vaccines, 173–74 Xeroderma pigmentosum, 13 X-linked lymphoproliferative disease (XLP), 142 XLP. See X-linked lymphoproliferative disease X-ray, speed of, on proto-oncogene expression, 217

Y YAC. See Yeast artificial chromosome Yeast artificial chromosome (YAC), 96

This Page Intentionally Left Blank