ADVANCES I N
Immunology
VOLUME 14
CONTRIBUTORS TO THIS VOLUME CHESTER A. ALPER ALAN E. BEER
R. E. BILLINGHAM L. HO...
26 downloads
1047 Views
21MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
ADVANCES I N
Immunology
VOLUME 14
CONTRIBUTORS TO THIS VOLUME CHESTER A. ALPER ALAN E. BEER
R. E. BILLINGHAM L. HOOD
GEORCEKLEIN J. Pam
FREDS. ROSEN
SIDNEY SHULMAN
ADVANCES IN
Immunology E D I T E D BY
F, J. DIXON
HENRY G. KUNKEL
Division of Experimenfal Pathology Scrippa Clinic and Rereorch Foundafion l a lolla, California
The Rockefeller University New Yo&, New York
VOLUME 1 4 1971
ACADEMIC PRESS, INC.
(29
New York San Francisco London
A Subsidiary of Harcourt Brace Jovanovich, Publishers
COPYRIGHT 0 1971, BY ACADEMIC PRESS, INC. ALL RIGHTS RESERVED NO PART OF THIS BOOK MAY BE REPRODUCED IN A N Y FORM, BY PHOTOSTAT, MICROFILM, RETRIEVAL SYSTEM,OR ANY OTHER MEANS, WITHOUT WRITTEN PERMISSION FROM THE PUBLISHERS.
ACADEMIC PRESS, INC.
111 Fifth Avenue, New York, New York 10003
United Kingdom Edition published by ACADEMIC PRESS, INC. (LONDON) LTD.
24/28 Oval Road, London NWl IDD
LIBRARY OF CONQRESS CATALOQ CARD
NUMBER: 61-17057
PRINTED IN THE UNITED STATES OF AMERICA
CONTENTS LIST OF CONTRIBUTORS .
. .
. PREFACE . . . . . CONTENTSOF PREVIOUSVOLUMES.
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
vii
iu xiii
I m munobiolog y of Mammalian Reproduction
.
ALAN E . BEER AND R. E BILLINGHAM
I. Introduction . . . . . . . . . . . . . I1. Essentials of Reproductive Biology . . . . . . . I11. The Uterus as a Craft Site and as a Route for Immunization . . . IV. Antigenic Status of Semen and Its Consequences . . . . . V. Choriocarcinoma . . . . . . . . . . . . VI . The Fetus Qua Homograft: Factors That May Contribute to . . . . . . . . . . . . Its Success . VII . Susceptibility of Pretrophoblastic Eggs to Transplantation Immunity . . . . . . . . . . . . . VIII. Histoincompatibility as a Determinant of Placental Size and Extent of Trophoblastic Invasion . . . . . . . . . IX . Organ-Specific Antigens of the Placenta . . . . . . . X. Maternal-Fetal Exchange of Cells . . . . . . . . XI. Natural Occurrence of Transplantation Disease . . . . . . XI1 Immunological Competence of the Placenta . . . . . . XI11. Concept of Immunological Inertia of Viviparity . . . . . XIV. Histocompatibility Gene Polymorphisms and Maternal-Fetal Interactions . . . . . . . . . . . . . References
.
2 4 6 15
22 26 39 41 43 49 66 67 69 70 76
Thyroid Antigens and Autoimmunity
SIDNEYSHULMAN
I. Introduction . . . . . . . . . . . . . I1. The Thyroid Gland: Structure. Function. and Malfunction . . . I11. Purification and Properties of Thyroid Proteins . . . . . . IV . Thyroid Antigens . . . . . . . . . . . . . V. Experimental Autoimmune Disease of the Thyroid: the Thyroid Gland as Source and Target . . . . . . . . . . . VI . Human Autoimmune Disease of the Thyroid . . . . . . VII. Features of the Autoimmune Response . . . . . . . . VIII. Chemical and Antigenic Structures of the Thyroglobulin Molecule IX . Concluding Remarks . . . . . . . . . . . References . . . . . . . . . . . . . V
85 87 93 107 114 132 142 155 170 173
vi
CONTENTS
Immunological Aspects
of Burkitt’s lymphoma
GEORGEKLEIN
I. Introduction . . . . . . . . . . . . . . . . . . I1. Humoral Antibody Studies . . . . . . I11. Studies on Cell-Mediated Immunity . . . . . . . . IV. One or Several EB Viruses? V. Imniunological Studies on Oncogenic Herpes Viruses in Animals VI. Implications . . . . . . . . . . . References . . . . . . . . . . .
.
.
.
.
I. Introduction . . . . . . . . . . . . I1. Hereditary Angioneurotic Edema . . . . . . . . I11. C4 Deficiency in Guinea Pigs . . . . . . . . IV. C2 Deficiency in Man . . . . . . . . . . V. Guinea Pigs Deficient in the “Third Component of Complement” VI . Genetic Structural Polymorphism in C3 . . . . . . VII . C3 Deficiency in Man . . . . . . . . . . . . . . . . . VIII . Congenital Hypercatabolism of C3 IX . C5 Deficiency in Mice . . . . . . . . . . X . C5 Dysfunction in Man . . . . . . . . . XI . C8 Deficiency in Rabbits . . . . . . . . . XI1. Miscellaneous . . . . . . . . . . . . References . . . . . . . . . . . .
. . . . . . . .
. . . . . .
. . . .
187 188 221 223 225 232 243
Genetic Aspects of the Complement System
CHESTERA . ALPERAND FREDS. ROSEX
.
. . . .
252 253 258 259 262 263 268 270 275 281 281 284 286
The Immune System: A Model for Differentiation in Higher Organisms
L . HOODAND J . F’RAHL
. . .
I Introduction . . . . . . . . . . . . . I1 Immunoglobulin Systems . . . . . . . . . . . . . . . . I11 Structural. Genetic. and Cellular Patterns . IV. A Genetic Mechanism for Differentiation: Two Genes + One Poly. . . . . . . . . . . . peptide Chain V Evolution of Immunoglobulin Variable and Constant Genes VI Theories of Antibody Diversity . . . . . . . . . VII Concluding Remarks . . . . . . . . . . . References Addendum . . . . . . . . . . . . .
. . .
AuTnoR INDEX .
. . .
. . . . . . . . . . . . .
. . . . . . . . . . . . . SUBJECTINDEX . . . . . . . . . . . . . .
291 296 298 305 311 314 344
345 351 353 374
LIST OF CONTRIBUTORS Numbers in parentheses indicate the pages on which the authors' contributions begin.
CHESTER A. ALPER, Blood Grouping Laboratory and Department
of
Medicine, Children's Hospital Medical Center, and Department of Pediatrics, Harvard Medical School, Boston, Massachusetts (251) ALAN E. BEER,* Departments of Medical Genetics and Obstetrics and Gynecology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania (1) R. E. BILLINGHAM,"Departments of Medical Genetics and Obstetrics and Gynecology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania ( 1 ) L. HOOD, Division of Biology, California Institute of Technology, Pasadena, California (291) GEORGEKLEIN, Department of Tumor Biology, Karolinska Institutet, Stockholm, Sweden (187)
J. PRAHL,Division of Biology, California Institute of Technology, Pasadina, California (291) FREDS. ROSEN,Blood Grouping Laboratory and Department of Medicine, Children's Hospital Medical Center, and Department of Pediatrics, Harvard Medical School, Boston, Massachusetts ( 251 )
SIDNEY SHULMAN,Department of Microbiology, New York Medical College, New York, New York (85)
" Present address: Department of Cell Biology, University of Texas Southwestern Medical School at Dallas, Dallas, Texas. vii
This Page Intentionally Left Blank
PREFACE The frontiers of immunology continue to expand into other areas of science, as the reviews in the present volume clearly indicate. Areas here involved include mammalian reproduction, autoimmune disease, oncology-virology, genetics, and cellular differentiation. In this widespread permeation, immunology may contribute conceptually or methodologically; it also benefits from observations revealing more and more biological phenomena which depend directly or indirectly on immunologic factors. Theoretically, the processes of mammalian conception and gestation might be considered among the most susceptible to immunologic interference; yet their relative invulnerability is evident in our current population crisis. The exceedingly complex sequence of events in these two processes, involving at least three genetically dissimilar individuals and multiplc potential immunologic reactions, is clearly analyzed in the first contribution to this volume, by Dr. Beer and Dr. Billingham. Of all the possible safeguards of the conceptus against immunologic injury, it is apparently the trophoblast with its ability to hide its transplantation antigens which plays the major role by preventing maternal sensitization to the fetus. Less important but still significant factors, such as the partial barrier to immunization offered by the decidua, the delayed appearance of major histocompatibility antigens in the embryo, and the immunosuppressive effect of the endocrine mix of pregnancy, are placed in perspective as contributors to the normal course of pregnancy. Also, the significance of both maternal and paternal sensitization to seminal antigens in preventing conception and the role of maternal sensitization by fetal blood cells in terminating pregnancy are evaluated. One of the most actively investigated of the numerous autoimmune states is that involving the thyroid. In the second article, Dr. Shulman presents a comprehensive review of this area, in which he has long been a leader, and relates thyroid structure and function to immunologic events. He discusses the physical, chemical, and immunologic characteristics of the various thyroid antigens and the methods best suited to their handling. The various types of humoral and cellular immune responses which may be induced by these antigens in animals are described in detail and related to the thyroid disorders which may accompany ix
X
PREFACE
them. Observations in man of spontaneous autoimmune responses to the several thyroid antigens and the various associated thyroid diseases are discussed from immunologic and pathologic points of view. The intriguing immunologic and virologic aspects of Burkitt’s lymphoma are considered by Dr. Klein in the third review. The humoral antibody and sensitized cell responses of patients with Burkitt’s lymphoma to tumor-associated antigens and the possible significance of these responses are presented, Intimately involved with at least some of the tumor antigens to which these immune responses are directed is the Epstein-Barr virus which is commonly associated with this disease. HOW these developments interrelate is not clear but it is possible that the tumor-associated antigens may give clues to the etiology of this disease or to the neoplastic behavior of the cells, while more information on the immune responses may provide possible approaches to therapy. As the immunologic and virologic study of human neoplasms increases, it is likely that a number of instances will be found with parallels to the situation in Burkitt’s lymphoma, and for this reason the pioneer work on this disease, much of which has been done by Dr. Klein, takes on special importance. In the fourth article, Dr. Alper and Dr. Rosen make an authoritative presentation of the genetic aspects of the complement system. The isolation and characterization of the complement components have initiated intensive study of the pathophysiology and genetics of this system. Numerous instances of deficiencies of one or another complement components or of inhibitors of this system have been discovered. Investigation of these deficiencies has provided a molecular explanation for the long recognized disease, hereditary angioneurotic edema, has defined new syndromes associated with some deficiencies, and has revealed relatively little ill effect of others. Equally important, these studies have contributed greatly to our understanding of the function of complement in reactions which involve host defense against infection and injury. In addition, understanding of the genetic polymorphism in the complement system provides a potent tool for the study of population genetics and cytogenetics and for the investigation of structure-function relationships of the complement proteins. Dr. Hood and Dr.Prahl provide, in the last article, a thoughtful and provocative treatment of the genetic basis of the Ig system, emphasizing its role as a model for differentiation in higher organisms. Based on an evaluation of the structural, genetic, cellular, and evolutionary aspects of the Ig system, they consider several possible explanations of antibody diversity and provide strong arguments for their preference for the germ line theory. This review not only puts in perspective the genetic and
PREFACE
xi
biochemical information on the Ig system but also relates this information to the more general biological problem of differentiation. As always, it is a pleasure to acknowledge the cooperation and assistance of the publishers, who have done much to ensure the quality of this series of volumes. FRANK J. DIXON HENRYG. KIJNKEL September 1971
This Page Intentionally Left Blank
Contents of Previous Volumes Volume 1 Transplantation Immunity and Tolerance
M. ~
E
K A.,
LENG~OVA, AND T. HRABA
Immunological Tolerance of Nonliving Antigens
RICHARDT. SMITH Functions of the Complement System
ABRAHAMG. OSLER In Vitro Studies of the Antibody Response
ABRAMB. STAVITSKY Duration of Immunity in Virug Diseases
J. H. HALE Fate and Biological Action of Antigen-Antibody Complexes WILLIAM
0.WEIGLE
Delayed Hypersensitivity to Simple Protein Antigens
P. G.H. GELLAND B. BENACERRAF The Antigenic Structure of Tumors
P. A. GORER AUTHORINDEX-SUBJECI: INDEX Volume 2 Immunologic Specificity and Molecular Structure FRED E;ARUSH Heterogeneity of y-Globulins JOHN
L. FAHEY
The Immunological Significance of the Thymus
J. F. A. P. MILLER, A. H. E. MARSHALL, AND R. G.WHITE
Cellular Genetics of Immune Responses
G. J. V. NOSSAL
Antibody Production by Transferred Cells
CHARLES G. COCHRANE AM) FRANK J. DXXON Phagocytosis
DERRICK ROWLEY xiii
XiV
CONTENTS OF PREVIOUS VOLUMES
Antigen-Antibody Reactions in Helminth Infections
E. J. L. SOULSBY Embryological Development of Antigens
REEDA. FLICKINGER
AUTHORINDEX-SUBJECT INDEX Volume 3 In Vitro Studies of the Mechanism of Anaphylaxis
K. FRANKAUSTENAND JOHN H. HUMPHREY
The Role of Humoral Antibody in the Homograft Reaction
CHANDLER A. STETSON Immune Adherence
D. S. NELSON
Reaginic Antibodies
D. R. STANWORTH Nature of Retained Antigen and Its Role in Immune Mechanisms
DANH. CAMPBELL AND JUSTINE S. GARVEY Blood Groups in Animals Other Than Man
W. H. STONEAND M. R. IRWIN Heterophile Antigens and Their Significance in the Host-Parasite Relationship
C. R. JE" AUTHORINDEX-SUBj ~ c INDEX r Volume 4 Ontogeny and Phylogeny of Adaptive Immunity
ROBERTA. GOODAND BENW. PAPERMASTER
Cellular Reactions in Infection
EMANUEL SUTERAND HANSRUEDY RAMSEIER Ultrastructure of Immunologic Processes JOSEPH
D. FELDMAN
Cell Wall Antigens of Gram-Positive Bacteria
MACLYNMCCARTYAND STEPHENI. MORSE Structure and Biological Activity of Immunoglobulins
SYDNEYCOHENAND RODNEYR. PORTER
CONTENTS OF PREVIOUS VOLUMES
Autoantibodies and Disease
H. G. KUNKELAND E. M. TAN Effect of Bacteria and Bacterial Products on Antibody Response
J. Mmoz AUTHOR INDEX-SUB J E C ~INDEX Volume 5 Natural Antibodies and the Immune Response
STEPHENV. BOYDEN Immunological Studies with Synthetic Polypeptides
MICHAELSELA Experimental Allergic Encephalomyelitis and Autoimmune Disease
PHILIPY. PATERSON The Immunology of Insulin
C.G. POPE Tissue-Specific Antigens
D. C. DUMONDE AUTHORINDEX-SUB J E C ~INDEX Volume 6 Experimental Glomerulonephritis: Immunological Events and Pathogenetic Mechanisms
EMILR. UNANUEAND FRANK J. DIXON Chemical Suppression of Adaptive Immunity
ANNE. GABRIELSON AND ROBERTA. GOOD Nucleic Acids as Antigens
OTTOJ. PLEscvl AND WEZWER BRAWN In Vifro Studies of Immunological Responses
of lymphoid Cells
RICHARDW. DWTON Developmental Aspects of Immunity JAROSLAV
STERZLAND ARTHUR M. SILVERSTEIN
Anti-antibodies
PHILIPG. H. GELLAND ANDREW S. KELUS
CongIutinin and Im munocong Iutinins
P. J. LACHMANN AUTHORINDEX-SUB JECT INDEX
xv
xvi
CONTENTS OF PREVIOUS VOLUMES
Volume 7 Structure and Biological Properties of Immunoglobulins
SYDNEYC o r n AND CESAR MILS~EIN
Genetics of Immunoglobulins in the Mouse MICHAELP o r n AND ROSELIEBERMAN Mimetic Relationships between Group A Streptococci and Mammalian Tissues JOHN
B. ZABRISm
lymphocytes and Transplantation Immunity DARCY B. WILSONAND R. E. BILLINCHAM Human Tissue Transplantation JOHN
P. MERRILL
AUTHORINDEX-SUBJECXINDEX Volume 8 Chemistry and Reaction Mechanisms of Complement
HANS J. MWR-EBERHARD
Regulatory Effect of Antibody on the Immune Response JONATHAN W.
Urn AND GORANMOLLER
The Mechanism of Immunological Paralysis D. W.DRESSER AND N.A. MITCHISON In Vitro Studies of Human Reaginic Allergy
ABRAHAM G. OSLER,LAWRENCE M. LICHTENSTEIN, AND DAVID A. LEW AUTHORINDEX-SUBJECTINDEX Volume 9
Secretory Immunoglobulins
THOMAS B. TOMASI, JR.,
AND JOHN
BIENENSTOCK
Immunologic Tissue Injury Mediated by Neutrophilic leukocytes
CHARXZS G. COCHRANE The Structure and Function of Monocytes and Macrophages
Zmm A. Corn The Immunology and Pathology of NZB Mice
J. B. HOWIE AND B. J. HELYER AUTHORINDEX-SUBJECXINDEX
CONTENTS OF PREVIOUS VOLUMES
xvii
Volume 10 Cell Selection by Antigen in the Immune Response
GREGORY W. SISKINDAND BARUJ BENACEPhylogeny of Immunoglobulins
HOWARD M. GREY Slow Reacting Substance of Anaphylaxis
ROBERTP. ORANGE AND K. FRANKAUSTEN
Some Relationships among Hemostasis, Fibrinolytic Phenomena, Immunity, and the Inflammatory Response
OSCAR D.
RATNOFF
Antigens of Virus-Induced Tumors
KARLHABEL Genetic and Antigenetic Aspects of Human Histocompatibility Systems
D. BERNARD AMOS AUTHORINDEX-SUBJECTINDEX Volume 11 Electron Microscopy of the Immunoglobulins
N. MICHAEL GREEN Genetic Control of Specific Immune Responses
HUGH0. MCDEVITT ANTI BARUJBENACERRAF The lesions in Cell Membranes Caused by Complement JOHN
H. HUMPHREY AND ROBERTR. DOURMASHKIN
Cytotoxic Effects of Lymphoid Cells In Vitro
PETERPERLMANN AM) GORANHOLM
Transfer kctor
H. S. LAWRENCE
Immunological Aspects of Malaria Infection
IVOR N. BROWN j ~ c INDEX r AUTHORINDEX-SUB Volume 12 The Search for Antibodies with Molecular Uniformity
RICHARDM. KRAUSE Structure and Function of yM Macroglobulins
HENRYMETZGER
xviii
CONTENTS OF PREVIOUS VOLUMES
Transplantation Antigens
R. A. REISFELDAND B. D. KAHAN The Role of Bone Marrow i n the Immune Response
NABIH I. ABDOUAND MAXWELLRICHTER Cell Interaction i n Antibody Synthesis
D. W. TALMAGE, J. RADOVICH, AND H. HEMMINGSEN The Role of lysosomes i n Immune Responses
GERALD WEISSMANN AND PETERDUKOR Molecular Size and Conformation of Immunoglobulins
KEITHJ. DOJUUNGTON AND CHARLES TANFORD AUTHORINDEX-SUB JECT INDEX Volume 13 Structure and Function of Human Immunoglobulin E
HANSBENNICH AND S. GUNNAR 0. JOHANSSON Individual Antigenic Specificity of Immunoglobulins
JOHNE. HOPPERAND ALFREDNISONOFF In Vitro Approaches to the Mechanism of Cell-Mediated Immune Reactions
BARRYR. BLOOM Immunological Phenomena in leprosy and Related Diseases
J. L. TURKAND A. D. M. BRYCESON Nature and Classification of Immediate-Type Allergic Reactions
ELMER L. BECKER
AUTHORINDEX-SUB JECT INDEX
ADVANCES IN
Immunology
VOLUME 1 4
This Page Intentionally Left Blank
lmmunobiology of Mammalian Reproduction' ALAN E. BEERZ AND R. E. BILLINGHAM2 Departments of Medico1 Genetics and Obstetrics and Gynecology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania
I. Introduction . . . . . . . . . . . 11. Essentials of Reproductive Biology . . . . . . Role of Hormones in Implantation . . . . . . 111. The Uterus as a Graft Site and as a Route for Immunization A. Endocrinological Determinants of the Fate of Genetically Compatible Intrauterine Crafts . . . . . . B. Nonimmunological Tissue Interactions between Grafts or Conceptuses and Their Implantation Sites in the Uterus . C. Consideration of the Uterus as an Immunologically . . . . . . . . . Privileged Site D. Influence of Genetically Alien Fetuses and Intrauterine Inocula of Homologous Cells 011 Maternal Reactivity . E. Graft-versus-Host Reactions in the Uterus . . . IV. Antigenic Status of Semen and Its Consequences . . . A. Autoantigenic Status of Spermatozoa . . . . . B. Isoantigenic Status of Spermatozoa . . . . . C. Local Antibody Production in the Uterus . . . . D. Reactivity of Sensitized Uterus to Antigenic Exposure . E. Somatic Fertilization . . . . . . . . V. Choriocarcinoma . . . . . . . . A. Evidence of Host Resistance . . . . . . B. Fetal-Maternal Isoantigenic Compatibility as a . . . . . . . Predisposing Factor . VI. The Fetus Qua Homograft: Factors That May Contribute to Its Success . . . . . . . . . A. Complete Separation of Maternal and Fetal Circulations . B. Antigenic Immaturity of the Fetus . . . . . C. Nonspecific Weakening of the Immunological Reactivity . . . . . of the Mother during Pregnancy D. Physiological Barrier between Mother and Fetus . . VII. Susceptibility of Pretrophoblastic Eggs to Transplantation Immunity . . . . . . . . . . . VIII. Histoincompatibility as a Determinant of Placental Size and Extent of Trophoblastic Invasion . . . , . .
.
.
. . . .
. . .
.
2 4 6 6
.
.
6
.
.
8
.
.
9
. . . . . . .
. . . . . . . . . .
11 14 15 15 18 21 21 21 22 23
. . . . . . .
24
.
.
. .
31 33
.
.
39
.
.
41
.
. . .
26 27 29
' This review is dedicated to the memory of Dr. D. R. S. Kirby in recognition of his signal contributions to our understanding of the immunological hazards of life before birth. a Present address: Department of Cell Biology, University of Texas Southwestern Medical School at Dallas, 5323 Harry Hines Blvd., Dallas, Texas 75235. 1
2
ALAN E. BEER AND R. E. BILLINGHAM
IX. Organ-Specific Antigens of the Placenta . . . . . A. Biological Activity of Heterologous Antiplacental Serum . B. Cross-Reactivity between Renal and Placental Antigens . C. Existence of a “Private,” Tissue-Specific Antigen . . . . . . . . . in Trophoblast D. Parturition as an Immunologically Mediated Process . E. Conclusions . . . . . . . . . . X. Maternal-Fetal Exchange of Cells . . . . . . A. Trophoblast Cells . . . . . . . . B. Blood Cells . . . . . . . . . . C. Malignant Cells . . . . . . . . . D. Consequences of Maternal Exposure to Fetal Antigens . E. Consequences of Maternal Exposure to Fetal Leukocyte . . . . . and Transplantation Antigens . F. Consequences of Fetal Exposure to Maternal Cells . . XI. Natural Occurrence of Transplantation Disease . . . XII. Immunological Competence of the Placenta . . . . XIII. Concept of Immunological Inertia of Viviparity . . . XIV. Histocompatibility Gene Polymorphisms and . . . . . . Maternal-Fetal Interactions . References . . . . . . . . . . . .
. . . . . . ,
. . . . . . . . .
.
.
.
43 44 45
.
46 48 49 49 50 51 53 53
. . .
57 62 66 67 69
. .
70 76
. . . . . . . . .
.
.
I. Introduction
Mammalian reproduction entails: ( 1) the repeated “inoculation” of female hosts, by one particular route, with several hundred million highly specialized, motile, short-lived cells of alien genetic origin suspended in a complex seminal fluid produced in organs peculiar to males; ( 2 ) the union of one of these spermatozoa with a free-floating ovum to produce a zygote with equal genetic endowments from both parents which, ( 3 ) after a free-living (larval) existence of approximately 5 days’ duration, becomes “beached on the prepared endometrial surface, and initiates ( 4 ) an extremely intimate “parasitic” type of grafthost or fetal-maternal relationship of finite duration. Depending upon the species concerned, the gestation period may be as short as 16 days, as in the Syrian hamster, or as long as 21 months, as in the Indian elephant. Furthermore, there is wide variation in the mean number of fetuses that develop concomitantly in the uterus of a female, ranging from 1 in man to 10 or more (which at term collectively may approach the maternal weight) in mice and rats and in the total number of litters a female may give birth to in her reproductive life-span. The fact that spermatozoa have long been shown to possess cytospecific antigens, in addition to their recently established expression of transplantation antigens, suggests one kind of maternal sensitization that might under natural and/ or experimental conditions interfere with the early stage of the reproductive process. Sensitization to the various
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
3
components of seminal plasma has also long been entertained as yet another possible immunological complication of fertility. Recognition of the “homograft” status of the fetuses resulting from matings between genetically dissimilar parents, in conjunction with such well-established empirical observations that ( a ) mothers give no evidence of progressive decline in fertility on repeated mating to the same genetically unrelated male, ( b ) homografts from offspring resulting from such a mating enjoy no special immunological dispensations when grafted to their mothers, and ( c ) mothers cannot be caused to reject embryos by presensitization against the alien tissue transplantation antigens of their consorts, has, over the past decade, stimulated efforts to elucidate the means by which Nature’s homografts are enabled to violate the laws of transplantation so effectively and consistently. From the viewpoint of the transplantation immunologist, the maternal-fetal relationship is much more complex than the relationship between a tissue or organ homograft and an immunologically mature host. This follows from the fact that pregnancy entails a parabiotic union between two different organisms in which not only is there intimate apposition and a comingling ( at the trophoblast-endometrial stromal interface) of relatively immobile tissue cells of dissimilar genetic constitutions but also, in some species, a bathing of fetal tissue components of the placenta with maternal blood as in the conventional host-graft relationship. In addition, there is evidence of a covert exchange of blood cellular elements of various types between the parabionts, and, in some species, a transfer of passive immunity from mother to fetus. The brilliant analysis of the etiology of Rh hemolytic disease of the newborn in man indicated that the risks of maternal isoimmunization through pregnancy are not of mere academic interest-there are very definite immunological hazards to becoming gestated. Interest was initiated in other systems of antigens that might contribute to fetal morbidity or mortality. Although this article is largely concerned with a review of maternalfetal relationships from the viewpoint of the transplantation immunologist, it does cover other aspects for completeness’ sake. No apology is needed for including a brief account of the salient features of reproductive biology from ovulation to nidation, for they amount to Nature’s carefully coordinated preparation of her grafts and the “beds” intended to receive them. Choriocarcinoma, a highly malignant tumor of trophoblastic origin which, like fetuses, consistently transcends histocompatibility barriers, is
4
ALAN E. BEER AND R. E. BILLINGHAM
considered in some detail for the light it sheds on the biological properties of normal trophoblast. The means by which young animals acquire their maternal endowment of antibodies will not be considered because this is the subject of the recent publication by the late Professor F. W. Rogers Brambell (1970). Finally, it is our opinion that although there have been several excellent review articles and symposium volumes dealing with some immunological facets of the fetal-maternal relationship, progress in the field has been sdciently rapid to justify yet another (Billingham, 1964; Billington, 1970; Boyd, 1959; Kirby, 1968a; Lanman, 1965; Medawar, 1953; Papiernik-Berhauer, 1966; Park, 1965; Scott, 1968; Simmons and Russell, 1967a; Wolstenholme and O’Connor, 1969). It. Essentials of Reproductive Biology
The successful implantation of a fertilized egg in the endometrium and its survival as a developing fetus to term are reflections of Nature’s highly successful solution of the problems attending the transplantation of one particular type of graft to one special type of bed-a solution that is obviously coeval with the origin of mammals. Since fetuses are transplants in every sense of the word (Billingham, 1964), analyses of the essential hormonal, vascular and immunogenetic parameters for the primary take or healing-in of the egg, its development into an organismic graft, and ultimate and consistent rejection (parturition) after a relatively constant survival time ( gestation period) can meaningfully be compared with those pertaining to the initial acceptance and continued functional well-being of the conventional grafts used in experimental biology and surgery. Mechanisms of the type that characterize immunological phenomena have been postulated by some authorities as indispensible components of two key events in the reproductive process: ( 1 ) a highly selective antigen-antibody-like stereochemical interaction between highly specific components of the plasma membranes of eggs and sperms which provides a plausible basis for the tissue specificity and species specificity of fertilization (Tyler, 1961a,b); (2) a local inflammatory response with which mononuclear leukocytic cells appear to be intimately associated, if not causally related, in the endometrium, having some features in common with a local delayed hypersensitivity reaction, at the site of implantation and which may be essential for nidation (Marcus and Shelesnyak, 1968). In most mammals, after fertilization in the upper reaches of the fallopian tube, the egg, invested and quarantined by the zona pellucida, spends approximately 4 to 5 days suspended in and nourished by the
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
5
fluid milieu secreted by the tuba1 and uterine epithelia. Fluid is imbibed through the zona pellucida, and the blastocoele of the fertilized, unimplanted egg distends and eventually collapses. The zona disappears, and fluid escapes to the endometrial environment ( Reynolds, 1969a,b). Local action of this fluid on the host tissue has been held responsible for the mobilization of maternally derived inflammatory and other blood cellular elements in the implantation site. This inflammatory reaction, reminiscent of the expression of homograft, tuberculin, and other types of cell-mediated, delayed-type hypersensitivities, is not inimical to the well-being of the zygote. On the contrary, its occurrence brings to the implantation area increased vascularity, macrophages, and othe;. cellular elements which probably help remove the tissue and cellular debris resulting from trophoblastic invasion. The end points of this quasi-immunological tissue response are the fully developed hypersecretory gestational endometrium (decidua) and a fibrinoid deposition in the vicinity of the trophoblast. From fertilization of the ovum to its implantation, the ovular and endometrial changes are closely coordinated and the success of implantation turns upon maintenance of this delicate synchrony. Like grafts, blastocysts are not fastidious in terms of their requirements for implanting and developing more or less normally to a relatively advanced stage. This is evidenced by their behavior when deliberately placed in such ectopic sites as the anterior chamber of the eye, the brain, the spleen, the cryptorchid testis, and the mesentery (Kirby, 1968a; McLaren, 1965a) and by the fact that they readily implant of their own accord and develop at least to an advanced stage in such deviant extrauterine sites as khe ovary, rectum, or pouch of Douglas in man. These organs, of course, are relatively insensitive to the hormones that affect the uterus and its epithelium. By contrast, the endometrium, which forms the natural recipient area or graft bed for fertilized eggs, behaves in a much more discriminating manner. Not only is the phase during the reproductive cycle when it will accept blastocyst “grafts” sharply delineated but, despite the relatively enormous surface area of the endometrium available, there seems to be a restricted distribution of potential recipient sites for blastocysts ( McLaren and Michie, 1959a; McLaren and Finn, 1967). Likewise, it is a well-established fact that the stage of maturation of the blastocyst is a critical determinant of its successful nidation. Utilizing blastocyst transfer techniques, it has been shown that ova younger than the appropriate stage of endometrial development fail to implant. The most favorable conditions in the reproductive cycle are when embryos are either in synchrony with, or 1 day ahead of, the
6
ALAN E. BEER AND R. E. BILLINGHAM
endometrium (Doyle et al., 1963). Not inappropriately has McLaren (1965a) depicted the uterus as a Procrustean bed. An obvious structural feature that may contribute significantly to the discriminatory properties of the uterus as a recipient site in non-Primate species is its uninterrupted investment, throughout the reproductive cycle, by a layer of epithelium. The successful establishment of a free graft on an intact, epithelialized surface is an event which, a primi, would seem improbable or, at best, difficult, to accomplish.
ROLEOF HORMONES IN IMPLANTATION The receptivity of the uterus is under hormonal control. It has long been known that progesterone, produced by the developing corpus luteum in the ovary, prepares the estrogen-primed endometrium in a general way for implantation, and that estrogen in some way triggers the process of implantation. In rats it has been shown that an estrogen surge on the fifth day of preimplantation pregnancy is essential for nidation to occur in the progesterone-primed uterus (Shelesnyak and Kraicer, 1963). This estrogen surge, of ovarian origin, appears to be induced by pituitary gonadotropins under hypothalamic control. If the pituitary is removed. implantation does not occur unless exogenous progesterone and estrogen are given. This complex endocrine control system appears to be concerned entirely with the preparation of the uterus as a graft site, having nothing to do with the blastocyst. Ill. The Uterus as a Graft Site and as a Route for Immunization
There are certain “privileged” sites in the body in which grafts of alien origin may acquire a blood supply and yet thrive for a long time, exempt from the rejection process (Billingham and Silvers, 1971). Moreover, such sites can be created by experimental procedures that deprive the intended graft site of its lymphatic drainage, i.e., interfere with the afferent arc of the immunological reflex (Barker and Billingham, 1968). A considerable weight of circumstantial evidence makes it difficult to believe that any kind of privileged status of the uterine milieu can account for the success of fetuses as homografts. Various studies have been carried out specifically to elucidate the properties of the uterus as a graft site for both solid tissue and cellular grafts of various types and of isologous and homologous origin. A. ENDOCRINOLOGICAL DETERMINANTS OF THE FATEOF GENETICALLYCOMPATIBLE INTRAUTERINE GRAFTS Beer and Billingham (1970) and Beer et al. (1971a) have developed simple techniques for placing short, cylindrical grafts of everted skin,
7
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
prepared from rats’ tails or monodisperse suspensions of viable epidermal cells in close and intimate contact with the completely intact and untraumatized endometrial surface, simulating in a rather crude way the implantation of a blastocyst. Studies with these grafts have evaluated the receptivity of the uterus as a graft site. The incidence of primary take or healing-in and indefinite survival of genetically compatible, unsutured skin grafts on the rat’s endometrium approached 100% provided that exogenous estrogen was given on the day of grafting. In nonestrogen-treated rats, however, the grafts failed to take, being extruded into the hosts’ vaginas within 48 hours. The results of various experiments indicated that the essential influence of the hormone was upon the uterus itself, rather than upon the skin graft, but the manner in which estrogen affects the uterine milieu to allow graft implantation on its fully epithelialized surface has yet to be clarified. In the absence of exogenous estrogen, skin grafts inserted into the uteri of female rats, mated 4-5 days previously, healed-in readily and did not interfere with the survival and continued development of the conceptuses in the affected horn. However, the time of grafting in relation to the antecedent mating did influence the site at which the skin graft implanted in relation to the implantation sites of the embryos. The various experimental findings summarized in Table I are consistent with the hypothesis that the normal, immediate, preimplantation hormonal milieu includes an estrogen surge that is essential for the primary healing-in of any intrauterine graft, whether it be a blastocyst or a “placebo” embryo in the form of a skin graft or a suspension of epidermal cells. When suspensions of isologous epidermal cells were injected into the uterine lumens of estrogen-treated female rats, numerous small foci TABLE I OF VARIOUS TREATMENTS OF HOST RATS ON THEIRACCEPTANCE OF INFLUENCE FREEINTRAUTERINE SKINGRAFTS Experimental group
No. of animals
1 2 3 4
50 24
5
24
6
12
Treatment 50 pg. Estrogen No estrogen Bilateral oophorectomy Bilateral oophorectomy, 50 fig. estrogen Day 4 or 5 of preimplantation pregnancy
No. and percentages of successful implantation 48/50 2/24 0/6 12/12
(96%) (8%) (0%)
24/24
(loo’%)
(100%)
8
ALAN E. BEER AND R. E. BILLINGHAM
of epidermal proliferation on the endometrial surface were demonstrable histologically by the twelfth postperative day. These plaques of squamous epidermis were distributed longitudinally along the endometrial surfaces, analogous to the attachment sites of conceptuses, thus, suggesting that the epidermal cells might only be capable of implanting on the uterine surface at the predetermined number of circumscribed sites that would normally receive blastocysts. As with free skin grafts, grafted monodisperse epidermal cells failed to become established in the uteri of nonestrogen-treated virgin rats. B. NONIMMUNOLOGICAL TISSUEINTERACTIONS BETWEEN GRAFTSOR CONCEPTUSES AND THEIRIMPLANTATION SITES IN THE UTERUS Although foreign bodies in the rats’ uterus stimulate a pseudodecidual response in the endometrium and a state of pseudopregnancy in the animal, successfully implanted skin grafts in this organ fail to evoke this response. Rats bearing established intrauterine skin grafts display normal estrus cycles and mating behavior. However, in animals that were pregnant while bearing established intrauterine skin grafts, decidual tissue developed beneath the latter and was indistinguishable from that in contact with the fetal trophoblast. Despite the close similarity between the decidual responses to fetal tissue and to skin grafts, respectively, established grafts of skin are completely exempt from parturition even when fetuses developing in the same or in the contralateral uterine horn are delivered. This suggests that the important parturitional role of the placenta must turn upon unique interrelations between the maternal decidua and the juxtaposed fetal trophoblast. From the maternal viewpoint, pregnancy represents an extremely delicate and intimate graft-host relationship in which the placenta, clearly the property of the fetus, is in contact with the maternal decidua in higher mammals with hemochorial placentas. The latter include a vast population of cells comprising the trophoblast, which as a consequence of its invasive and phagocytic activity becomes so intimately incorporated into maternal tissue that, in man and certain other species, it furnishes an endothelium-like lining for vascular channels through which the maternal blood passes. Furthermore, regularly in humans, and less frequently in a few other species including chinchillas (Billington and Weir, 1967), entire islands of trophoblast are known to become detached, enter the maternal blood, and be swept away by the circulation-a phenomenon known as trophoblastic deportation. Boyd ( 1959) has shown that in man “local patrols” of trophoblast cells migrate like salmon along vascular endothelial surfaces up into the mouths of the
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
9
spiral arterioles; this phenomenon was also observed in hamsters (Orsini, 1954). Unlike the situation pertaining to free tissue grafts, notably skin ( Haller and Billingham, 1967), in experimental surgery, there is no ingrowth and development of vascular primordia from the recipient area (maternal decidua of the uterus) into the placenta. Maternal blood is in direct contact with the trophoblast of the fetus.
C. CONSIDERATION OF THE UTERUS AS AN IMMUNOLOGICALLY PRIVILEGED SITE In all pregnancies resulting from heterospecific matings (i.e., those in which the male and female are genetically disparate, as in outbred populations or when the male and female are members of different inbred strains) one can envisage the female being confronted by an immunological dilemma in the form of a large homograft of alien placental tissue which, according to the laws of transplantation she should reject. The normal birth of living young is sufficient proof that the most pressing problems have been solved by Nature. Despite considerable evidence that the success of fetuses qua homografts is dependent upon an ineffective expression of transplantation antigens by the trophoblast ( see Section VI,D,2), whether any special immunological dispensations apply to the uterus, at least at the sites of implantation of conceptuses, is still equivocal. Schlesinger ( 1962) tested the possible privileged status of the uterus by implanting small test tumor grafts into the uterine horns of rats and mice. He found that if the tumor grafts were genetically compatible, they grew successfully, but tumor homografts survived for only a relatively short time in normal hosts and were rejected in an accelerated manner in the uteri of specifically presensitized hosts, irrespective of whether the latter were pregnant, pseudopregnant, or nonpregnant in one uterine horn. Unfortunately the design of this experiment failed to exclude the possibility that the test tumor homografts (or, effectively, placebo embryos) might have outgrown the limits of the physiological uterus, penetrating the myometrium, for example. Subsequently, Poppa et al. (1964) demonstrated that homografts of a normal, noninvasive tissue-parathyroid-transplanted to the uterus of pseudopregnant and nonpseudopregnant, parathyroidectomized rats were consistently rejected within 20 days. This indicates that, so far as homologous cells of nonembryonic tissue are concerned, transplantation immunity is both incitable and expressible in the uterine milieu. However, Kirby (1968a) objected to these findings as decisively refuting the hypothesis that at least the local implantation sites in the gravid uterus might have a privileged status, on the grounds that the parathy-
10
ALAN E. BEER AND R. E. BILLINGHAM
roid grafts studied by Poppa et al. (1964) failed to excite decidual reactions. Recently, Beer et al. (1971a) systematically analyzed the fate and consequences of introducing free solid tissue and monodisperse cellular homografts of various types into the lumen of the intact uteri of virgin and pregnant rats. When skin homografts from Lewis strain donors were inserted atraumatically into the uteri of estrogen-treated, virgin, Fischer female hosts they healed-in rapidly but were destroyed just as promptly as similar homografts transplanted orthotopically on control hosts, the median survival time of the grafts being about 11 days. However, when Lewis skin homografts were transplanted to the uteri of Fischer females during the preimplantation stage of pregnancy by males of their own strain, the grafts enjoyed a highly significant prolongation of survivalto the time of delivery in most instances (see Fig. 1). For example, grafts transplanted 4 days after successful mating gave evidence of viability when removed for histology just before the anticipated time of delivery-usually after they had been in residence for 17 days, However, in no case did pregnancy prolong the survival of similar skin homografts transplanted to the trunks of other hosts in the preimplantation stages of pregnancy. This prolongation of survival of intrauterine 100
,
I NON-PREGNANT
FISCHER FEMALES
WITH LEWIS INTRAUTERINE HOMOGRAFTS
---
*,
PREGNANT FISCHER FEMALES
WITH LEWIS INTRAUTERINE HOMOGRAFTS %
6
v)
I-
,
n.
I
1 " " 1 " " 1 " " ~ ' " ' 1 5
10
IS
' l " 1
20
25
30
DAYS A F T E R G R A F T I N G
FIG. 1. Survival times of Lewis skin homografts in the pregnant and nonpregnant uterus.
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
11
skin grafts in pregnant hosts may be due to the decidual tissue that develops beneath them interfering with the development of the host’s immunological response; i.e., it interferes with the afferent limb of the immunological reflex. That decidual tissue was incapable of affording any protection to grafts against a preexisting state of sensitivity was evidenced by the accelerated destruction of Lewis skin grafts in the uteri of pregnant Fischer females previously immunized against Lewis tissue antigens and by the fact that intrauterine skin homografts residing in the pregnant uterus of a tolerant host could be destroyed promptly during pregnancy by adoptive transfer of immunity, with sensitized lymphoid cells. Thus, at best in rats, decidual tissue can only play a very minor role in insuring the success of mammalian fetuses as homografts. In the rabbit, McLean and Scothorne (1968) have reported that, on the basis of dye injection experiments, although the myometrium is richly supplied with lymphatics, the only lymph vessels in the endometrium are located at the endometrial-myometrial junction. When skin homografts were implanted at various depths in the endometrium of pregnant and pseudopregnant does and recovered for histological study after they had been in residence for various periods, superficially implanted grafts (thought to lack a lymphatic drainage) fared much better than the more deeply located ones.
D. INFLUENCE OF GENETICALLY ALIENFETUSES AND INTRAUTERINE INOCULA OF HOMOLOGOUS CELLSON MATERNAL REACTJYITY As with homografts transplanted to most vascularized sites in the body, intrauterine exposure of female rats to skin homografts caused a striking enlargement and increase in weight (by a factor of about 3) of the draining, para-aortic node in the rat. That this regional lymphadenopathy was immunologically specific was indicated by the fact that genetically compatible grafts did not evoke it (see Fig. 2). A similar striking hypertrophy of the draining node was observed in the uteri of rats bearing genetically alien (ie., F, hybrid) fetuses, the enlargement being maximal by about the eighteenth day of gestation. Since no significant degree of regional lymphadenopathy was associated with pregnancies resulting from intrastrain matings, it seemed reasonable to attribute the node enlargement occurring with heterospecific pregnancies to stimulation by tissue antigens of fetal origin which, in some form, found their way into the draining maternal uterine lymphatics. Although this hitherto unobserved pregnancy-stimulated hypertrophy of draining uterine nodes (Nelson and Hall, 1964) parallels that obtained by intrauterine grafting of skin of similar genetic constitution, there is
12
ALAN E. BEER
AND R. E. BILLINGHAM
FIG.2. Influence of intrauterine grafts on weights of regional (para-aortic) nodes.
one clear-cut functional difference. For reasons as yet unknown, whereas intrauterine skin homografts are highly effective in eliciting transplantation immunity, alien fetuses are totally ineffective in this respect, though they may incite the formation of humoral antibodies on the part of the mother (see p. 57). Apart from its occurrence in the rat, enlargement of the lymph nodes
IMMUNOBIOLOCY OF MAMMALIAN REPRODUCTION
13
draining uterine horns during heterospecsc pregnancies has been found to occur in mice and hamsters and, on the basis of a few observations, in man. Although, in mice, pregnancy is associated with splenomegaly, Currie’s (1970) finding that there is no significant difference in spleen weight between intra- and interstrain pregnant mice at any stage of gestation suggests that immunological mechanisms are not involved. Comparative studies have shown that suspensions of homdlogous lymphoid cells introduced into the uterine lumen of virgin female rats are as effective in evoking sensitivity to homografts as similar inocula injected by the. conventional sensitizing routes. As few as 3OOO-4OOO viable Lewis node cells injected into one uterine horn of a mature virgin Fischer female host caused a three- to fourfold increase in weight of the regional para-aortic node. However, an inoculum of 500,000 cells is needed to evoke a state of transplantation immunity detectable in terms of the accelerated rejection of a second-set skin homograft. The immunological consequences of injecting spermatozoa directly into the lumen of the uterus are described in Section IV,B. If rats that have undergone primary sensitization against alien transplantation antigens by intrauterine inoculation with homologous lymphocytes are rechallenged a few weeks later by a similar inoculum in the same uterine horn, the injected organ undergoes a rapid enlargement and becomes conspicuously inflamed. This “recall flare,” which is reminiscent of that described by Rapaport and Converse (1957) at the rejection sites of prior skin homografts in men given subsequent grafts from the same donors, cannot be evoked in the uteri of rats presensitized by orthotopic skin homografts or by lymphoid cells injected by other than the intrauterine route. A provisional interpretation of this phenomenon is that it is owing to prompt reactivation of immunological memory cells which have persisted in the uterine endometrium and stroma after the initial intrauterine sensitization ( see Silverstein, 1964). Despite appropriate endocrinological preparation of the hosts, skin homografts introduced into the specifically presensitized uterine environment only transiently and feebly healed in, if at all. The facility with which transplantation immunity is expressed in the rat’s uterus parallels the expression of delayed-type hypersensitivity to dinitrochlorobenzene (DNCB) in the uteri of guinea pigs (Macher and Dorner, 1966). Observations, such as those just described may indicate that the intrauterine route of administration of leukocytes is a feasible means of contraception in man. Yet experimental findings in the rat indicate that this cannot be done. If, instead of challenging a presensitized uterus with a skin homograft or suspension of lymphoid cells from a donor
14
ALAN E. BEER AND R. E. BILLINGHAM
against whose tissue antigens the sensitivity was directed, the host female was mated with a male of that alien strain, F, hybrid zygotes were produced that confronted the mother with the foreign transplantation antigens against which her sensitivity was directed. Such zygotes implanted and developed normally, completely undaunted by the sensitivity specifically directed against them. Indeed, the bed afbrded by a locally sensitized uterus was found to display heightened receptivity as evidenced by the greater number of conceptuses it allowed to implant and subsequently sustained to term.
E. GRAFT-VERSUS-HOST REAC~IONSIN
UTERUS Since lymphoid cells deposited in the uterine lumen of a rat can easily traverse the intact endometrial epithelium and apparently gain access to the regional nodes, the capacity of the uterus to express local graft-versus-host reactions ( Billingham, 1968) has been explored by injecting lymphoid cell suspensions from parental strain donors into appropriate F, hybrid female rats. When 500,000 or more D A node cells were injected into the uteri of ( D A x Fischer) F, hybrid females there was enlargement, edema, and extravasation of fluid into the uterus within 48 hours, followed by a three- to fourfold increase in the weight of the draining lymph nodes. Uteri which had expressed this type of reactivity soon underwent irreversible atrophy and failed to support any subsequent pregnancies. There is evidence that mice may acquire the capacity to react against certain “strong” transplantation antigens even before birth ( Brent and Gowland, 1963; Howard and Michie, 1962). Our recent results substantiated this. When the draining para-aortic nodes of ((25’7 BL/6 x A) F, female mice bearing backcross fetuses, following mating with C57 males, were examined on the eighteenth day of gestation they were found to be significantly enlarged. In this particular genetic context, the mothers must necessarily have been genetically tolerant of the transplantation antigens of their offspring since, being F, hybrids, they had all the genetic determinants of transplantation antigens characterizing both the C57 and the A strains. However, all their F, progeny inherited a complete set of the antigenic determinants corresponding to the C57 strain and a variable number of A strain histocompatibility genes. Consequently, some of these fetuses should have been capable of reacting against their mothers. The finding of significant hypertrophy of the para-aortic nodes of the latter indicated that, at some time during gestation, a sufficient number of immunologically competent cells of fetal origin did gain access to maternal uterine tissue to interact with “native” cells in the draining node. THE
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
15
In the light of these various findings concerning the uterus as a graft site and its capacity to mediate various forms of transplantation immunity in both virgin and pregnant animals, suspicion falls heavily on the fetuses, and particularly upon their trophoblast since it is in the direct contact with maternal tissue, as responsible for exemption from rejection by both normal and specifically presensitized mothers. IV. Antigenic Status of Semen and Its Consequences
As a consequence of sexual activity the female reproductive tract is subjected to repeated inoculation with millions of spermatozoa-highly specialized, and usually immunogenetically alien, short-lived cellstogether with a minority of other cell types, including leukocytes, suspended in a complex, protein-containing seminal plasma secreted by specialized organs of the male. A tremendous amount of research and speculation have long focused upon possible immunological and other effects of this chronic, intermittent process which is much more frequent in man than in other species. S. Katsh (1969) pointed out that, in his “Descent of Man,” Darwin related profligacy of women to reduced fertility. Although as Hartman (1959) forthrightly stated, “much nonsense has been written on this subject,” its importance must not be underrated since, not only does it hold promise of explaining some cases of infertility, but it also affords one approach to the development of an immunological means of fertility control. If, as many authorities now believe, immunological tolerance represents an important developmental mechanism, any body constituents of an individual can be potentially autoantigenic if (1) the antigenic determinants concerned are not present in his immunologically competent cells and (2) he has not had the opportunity to become tolerant of them during early life (Brent and Medawar, 1959). It is scarcely surprising, therefore, that spermatozoa have been shown to be (or, more correctly, to contain) autoantigens since they develop at puberty in specialized organs in which they are isolated anatomically from the circulation long after the process of “self-recognition” normally occurs. A. AUTOANTIGENIC STATUSOF SPERMATOZOA Interest in the distinctive immunological properties of the components of male ejaculates and of the organs that produce them dates back to independent reports of Landsteiner ( 18!39), Metchnikoff (1900), and Metalnikoff (1900) that the serum of guinea pigs injected with the semen or homogenized testes from men, bulls, guinea pigs, or rabbits acquired the capacity to agglutinate and immobilize living spermatozoa of these various species. By 1961, according to Tyler (1961a),
16
ALAN E. BEER AND R. E. BILLINGHAM
upward of 150 independent experiments had been conducted in which guinea pigs, rabbits, rats, women, and other subjects were immunized with seminal materials of their own or of different species, primarily to study their antifertility effect. From a careful survey of this work, Tyler concluded that no reliable means had yet been discovered for the immunological control of fertility in any species. However, many of these experiments did show that antibodies could easily be produced which, if added to spermatozoa, destroyed their capacity to fertilize. Work in this field was greatly facilitated by the discoveries of Voisin et al. (1951) and Freund et al. (1953) that, in guinea pigs, spermatogenesis was suppressed following injection of homologous or autologous sperm or testicular extracts, and that the presence of complete Freund’s adjuvant was necessary to obtain consistent results. McLaren (1962, 1964, 1966) showed that, after a long course of intraperitoneal injections of isogenic or homologous spermatozoa in the absence of adjuvants, female mice might develop high titers of agglutinins equally effective against sperms from males of any murine genotype, i.e., the antigens were autoantigens. Although mating and ovulation occurred normally in such sensitized females, spermatozoa did not appear to reach the site of fertilization in adequate numbers, leading to a lowered rate of fertilization and reduction in litter size. Essentially similar results were obtained by Edwards (1964) who administered the antigenic material in Freund’s complete adjuvant and used a shorter immunization period. In some of her experiments McLaren (19ss) included pertussis vaccine in her immunization protocol, thereby obtaining higher spermagglutinin titers in the sera. Since this did not reduce the fertility of affected subjects she concluded that the level of sperm antibodies in the blood did not afford a meaningful indication of the degree of impairment of their fertility. A factor capable of agglutinating spermatozoa was consistently demonstrable in the serum of both very young and adult virgin female animals, and some workers have designated it as a naturally occurring sperm autoantibody (Torniov, 1970). A recent report by Boettcher et al. (1970) indicated that sperm-agglutinating activity in some human sera might not be due to immunoglobulins but to a lipoprotein-steroid conjugate, which could well explain the presence of sperm-agglutinating activity in the sera of pregnant and virgin women. However, under certain conditions, antigens or spermatozoa and seminal plasma might stimulate antibody production in the female and account for some cases of infertility in women. S. Katsh (19f39) reported that, in certain cases of idiopathic sterility, antibodies to seminal components were persistently detectable in serum and cervical mucus. How-
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
17
ever, attempts involving the use of prostitute volunteers to relate antibody response to the degree of coital exposure to seminal material have been unsuccessful (but see Schwimmer et al., 1967). These antibodies can be complement-fixing, agglutinating, immobilizing, or skinsensitizing. Various lines of evidence suggest that the rarely encountered high antibody titers result from normal or pathological conditions favoring deposition of large numbers of sperm into the uterus where they are rapidly absorbed. Experimental wounding of the reproductive tract prior to insemination facilitated the development of sensitization to sperm and, in cattle so treated, the presence of high titers of serum antispermatozoal antibody has been associated with sterility ( Bratanov, 1W9). In males, sperm-immobilizing and sperm-agglutinating autoantibodies may develop in the serum and appear in effective quantities in the ejaculates following inflammatory conditions of the testis or situations leading to occlusion of the vas deferens or the epididymis. Although it is conceivable that the demonstrable autoantibodies to spermatozoa can exert a prejudicial influence on an individual’s fertility, findings to date suggest that autoimmune mechanisms are only responsible for a small proportion of cases of male infertility. Various lines of evidence sustain the view that a cellular type of immunity may be the principal mechanism in both experimental and clinical autoallergic disorders involving spermatozoa in both sexes. For example, infertility in female mice immunized with sperm was associated with a rapid removal of spermatozoa from the uterus, and possibly with a rapid loss of sperm motility (Edwards, 1964); in guinea pigs, phagocytosis of spermatozoa in the uterus took place more actively in sensitized than in normal females (Maruta and Moyer, 1965). More telling is the consistent finding that the experimentally induced autoallergic disorders can only be transferred adoptively by viable lymphoid cells from sensitized subjects; serum transfers are ineffective. Furthermore, as in transplantation immunity, the level of humoral immunity may not parallel that of the cellular hypersensitivity with which it is associated. Human semen contains at least sixteen identifiable antigens, seven of which are present on spermatozoa, and the remainder are in the seminal plasma. Since four of the spermatozoa1 antigens are also present in the plasma, they probably represent secondarily acquired antigens, picked up through exposure of the sperms to secretions as they travel through the male reproductive tract. Needless to say, the presence of these “coating” antigens has impeded identification of the cytospecific sperm antigens.
18
ALAN E. BEER AND R. E. BILLINGHAM
Ultrasonic disintegration studies on mouse sperm (Henle et al., 1938) have revealed that three specific antigens reside in three distinct regions-one in the head, a second in the tail, and the third is assumed to be in the region of the acrosome, which shares properties common to both head and tail. Toullet et al. (1970) have recently fractionated guinea pig spermatozoa1 homogenates and studied them by a variety of immunological procedures which have revealed the existence of four different autoantigens, designated S, P, Z, and T, having distinctive locations. They are now trying to correlate the physicochemical properties and immunopathological properties. This work represents a great advance in the chemistry of sperm-specific antigens and holds promise of clarifying an important area of reproductive immunology. B. ISOANTICENIC STATUS OF SPERMATOZOA
Apart from the significance of the organ- or cytospec8c antigens associated with spermatozoa discussed above, another important question in the immunobiology of mammalian reproduction is the extent to which the genetic determinants of cellular isoantigens express themselves in the phenotypes of sperm. Do spermatozoa only express antigens corresponding to their haploid status or does their cytoplasm continue to express all the antigens corresponding to their diploid precursor spermatogonial cells? In more practical terms, does a sperm carrying the allelic determinant for blood group A express this antigen, and is it selected against by a female with the corresponding anti-A isoantibody? In species in which there is a histocompatibility locus on the Y chromosome, do Y-bearing sperm express this specificity? The answers to these questions are as yet unclear, but the possibilities that immunological forces of selection may be operating on sperm during their sojourn in the female reproductive tract is an intriguing one. Antigens of ABO blood group are expressed by sperm on the testimony of a variety of different techniques (Edwards et al., 1964; Gullbring, 1957; Landsteiner and Levine, 1926; Shahani and Southam, 1962). It is, however, important to realize that these cells can also absorb blood group substances from the seminal plasma of secretors. For example, group 0 sperm incubated with A secretor seminal plasma acquire the antigenic behavior of group A spermatozoa. Blood group antigens M, N, and Tja, although absent from seminal plasma, have also been identified on the sperm membrane (Edwards, 1964). The ability to detect Rh antigens on spermatozoa, and thus (hopefully) to be able to separate the cells from an Rh-positive Dd heterozygote into two populations, would be extremely useful clinically. Unfortunately, all attempts to detect these antigens on sperm have so far failed.
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
19
Whether transplantation antigens are present on spermatozoa is a question which has recently been answered affirmatively. Although Barth and Russell (1964), using a fluorescent antibody technique capable of detecting H-2 specificities on lymphoid cells, failed to detect these on sperm, Vojtifikovi and her associates (1969) suceeded in identifying some of the antigens determined by the H-2a allele on mouse spermatozoa by means of a hemagglutinin absorption technique. Application of an indirect fluorescent antibody procedure to sperm cell suspensions afforded direct confirmation that these cells expressed antigenic specificities determined by the H-2 locus. In addition, experiments in which mice received intraperitoneal injections of sperm followed, 5 to 21 days later, by a test skin homograft of similar alien genetic origin yielded suggestive evidence of the presence on sperms of antigens controlled by the linked minor H-3 and/or H-13 loci and by the H-Y (male specific) locus. The latter finding confirmed previous work of G. F. Katsh et al. (1964). As Vojti6kovA et al. (1969) pointed out, the fact that their sperm suspensions, obtained from the cauda epididymis and vas deferens, were always contaminated with about 10% of nonsperm cells might afford an alternative explanation for the results of the hemagglutinin absorption and grafting experiments. However, the results of the immunofluorescent procedure were unequivocal, since direct visualization of the antigens was entailed. With the aid of a cytotoxicity test, Goldberg et al. (1970) have also demonstrated unequivocally that H-2 antigens are present on mouse spermatozoa. With the aid of specific HL-A tissue typing sera and a microcytotoxicity test, Fellow and Dausset (1970) demonstrated that at least some antigens determined by the HL-A locus were present in high concentration on human sperm. When sperms from donors heterozygous for certain HL-A antigens were tested with monospecific antiserum, lysis of approximately one-half of the sperm population was observed, suggesting haploid expression of the HL-A antigens. In other experiments antigenically different types of sperm were detected in semen from a single donor. This interesting work opens the door for genetic engineering since pretreatment of sperm suspensions with an appropriate cytotoxic antibody could select gametes carrying hereditary disease. Encouraged by their evidence of the efficacy of the intrauterine route of administration of viable suspensions of lymphoid or epidermal cells for sensitization of animals against transplantation antigens, Beer et al. ( 1971a) studied the immunogenicity of homologous sperm administered by this route, By using different combinations of inbred strains of rats, mice, and hamsters, they found that washed homologous sperm injected directly into the host’s uterine lumen were as effective as similar num-
20
ALAN E. BEER AND R. E. BILLINGHAM
bers of lymph node cells in ( a ) stimulating hypertrophy of the draining para-aortic lymph node and ( b ) sensitizing the host with respect to skin homografts from the sperm donor strain transplanted 3 weeks later. It was also found that when C57 BL/6 female mice received 10 lo6 sperms from isogenic males (confronting them with the Y antigen), the majority rejected subsequent C57 BL/6 male skin isografts in an accelerated manner. Consonant with these findings were previous observations by Prehn (1960) and Lengerovi and Vojtiskovi (1963) that repeated mzting of C57 BL/6 females with males of the same strain, in the absence of pregnancy, might cause the females to become tolerant of male skin isografts. Preliminary findings of Beer and co-workers (1971b) indicated cross-reactivity between the Y antigens of rats and mice insofar as inoculation of Fischer rat sperm into the uteri of virgin C57 BL/6 female mice caused them to reject subsequent male skin isografts in an immune manner. In Syrian hamsters, which express transplantation immunity in the form of a delayed cutaneous hypersensitivity ( Ramseier and Billingham, 1966), if MHA strain hosts were injected with CB strain sperm via the intrauterine route, they subsequently responded to intracutaneous challenge with CB strain tissue antigen extract by intense direct hypersensitivity reactions. Careful appraisal of the proportion of contaminating nonsperm cells in the preparations used, in conjunction with the relatively low numbers of sperms required to immunize and the results of fractionation experiments, made it seem unlikely that contaminating cells were responsible for the apparent immunogenicity of the sperm cell preparations. This conclusion was reinforced by the finding that doses of spermatozoa found to be effective in eliciting transplantation immunity when injected into the uterus proved to be completely ineffective when administered by other routes. The fact that repeated normal matings did not elicit sensitization on the part of the female host hinted that either the numbers of spermatozoa gaining entrance into the uterus at any one time remained a subthreshold antigenic stimulus or these cells were cleared very rapidly from this organ. In rabbits killed from 2 to 36 hours postcoitus, Brackett (1971) found that the maximum number of spermatozoa recoverable from the uterine horns rarely exceeded 500,000. Similar numbers were recovered by Beer et al. (1971b) in rats. It is difficult to reconcile these low quantities in rabbits and rats with Austin’s (1957) figure of 50 x lo6 recoverable sperm from the uteri within 24 hours of mating in mice and rats,
x
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
C. LOCALANTIBODYPRODUCTIONIN
THE
21
UTERUS
Few attempts have been made to study the immunological activity of uterotubal, physiological secretions of subjects immunized primarily by this route. The uterine endometrium of most higher mammals is abundantly endowed with lymphatics and seems in all respects wellequipped to deliver either antigen or “antigen-primed,” i.e., peripherally sensitized, immunologically competent cells of hematologic origin to a draining lymph node. Diffuse deposits of lymphoid cells beneath the endometrium might be triggered to produce antibody locally. However, available evidence hints that local production of antibody at the uterine level against seminal antigens is minimal when serum antibody levels in the same individual are taken into consideration. D. REACTIVITY OF SENSITIZEDUTERUS TO ANTIGENICEXPOSURE We have seen that the uterus is able to express the cellular hypersensitivity responsible for homograft reactivity in a highly effective manner. Indeed, our studies on the recall flare type of reactivity in the uterus ( see Section II1,D) suggest that persistent immunological memory cells in this organ make a significant contribution to its reactivity when primary sensitization is achieved by this route. Anaphylaxis could be initiated in cattle by the intrauterine administration of antigen (Kerr and Robertson, 1953), and a woman was reported in anaphylactic shock following coitus, possibly because of sensitization to a glycoprotein present in the seminal plasma (Halpern et al., 1967). The various lines of evidence concerning the potential hyperreactivity of the uterus to various types of antigens hint that this kind of response may play a part in some cases of infertility.
E. SOMATICFERTILIZATION Terni and Maleci’s demonstration in 1937 that living rooster spermatozoa could penetrate living chick embryonic cells in vitro stimulated great interest in somatic fertilization and its possible immunological consequences. Austin (1959) found sperm heads in epithelial cells of the fallopian tube in several different species of rodents and concluded, on the basis of the frequency and reproducibility of these observations, that the phenomenon was real and probably of general occurrence. The potential significance of this process was highlighted by Reid and Blackwell’s (1965, 1967) evidence of incorporation of material from the labeled nuclei of living mouse sperm by peritoneal macrophages with which they had been maintained in vitro for 17 hours. Directly pertinent to
22
ALAN E. BEER AND €3. E. BILLINGHAM
this review is evidence that group A human spermatozoa cultured with type 0 HeLa cells were capable of passing the A antigen to the latter. The mechanism by which the cultured cells acquired their new antigenic properties has yet to be elucidated. If somatic fertilization is a relatively common event and if sperm are able to transmit, even transiently, genetically usable “information” to their host cells, then the apparent potent immunogenicity of homologous sperm when introduced into the uterine cavities of rodents may be partially explained. V. Choriocarcinorna
From the viewpoint of the transplantation immunologist, one of the most intriguing of all tumors is choriocarcinoma of gestational origin. They appear rarely in Caucasians but more frequently in inhabitants of the Middle East and Asia. These tumors are highly malignant and invasive derivatives of trophoblastic epithelium which follow at variable times after a conception that may have resulted in a normal birth, an abortion, or a hydatidiform mole (Bagshawe, 1969; Benirschke and Driscoll, 1967). They metastasize very readily via the veins, frequently involving many tissues and organs including the brain and lungs. What is so surprising about these tumors is that, despite their fetal origin and, therefore, homograft status, they are nearly always fatal in untreated women. Another anomaly of choriocarcinomas is their virtual absence in other species and the almost complete lack of success of attempts to produce them by experimental means (Benirschke and Driscoll, 1967). Rebognition of the genetically alien status of these tumors suggested the possibility of a simple immunotherapeutic approach, i.e., deliberate sensitization of affected women against the tissue antigens of their husbands, by means of skin grafts and/or injections of viable leukocytes, to arrest the growth and procure the destruction of the tumor cells. Following the pioneer work of Doniach et al. (1958), at least four independent groups of investigators subjected a total of 38 patients to this procedure, nearly always in conjunction with chemotherapy and sometimes surgery ( Bagshawe, 1969, 1970). Indeed, among the patients subjected to tissue antigen therapy, often administered repeatedly, chemotherapy alone failed to arrest the course of the disease. Unfortunately, no unequivocal successes have been obtained, and patients have succumbed to their tumors despite prompt rejection of their husband’s grafts, or in one case, a skin graft from a child isogenic with her mother’s tumor. The reactivity of some of these patients to grafts of their husbands’ skin was greatly inferior to their reactivity to concomitant grafts from
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
23
unrelated donors or to those of normal volunteers to skin grafts from unrelated donors. This aroused suspicions that exposure to the tumor, possibly in conjunction with chemotherapeutic anticancer drugs, might have resulted in specific alteration in their capacity to respond against the antigens concerned either as a consequence of the induction of immunological tolerance or of the development of isoantibodies having an “enhancing” influence ( see Section X,E ) . Discovery of antibodies reactive with their husbands’ leukocytes in the sera of some patients with gestational choriocarcinoma lent support to the latter notion. It need hardly be emphasized, however, that one cannot ascribe the formation of these antibodies to the presence of the tumor (see p. 57), although it might heighten their titers. Furthermore, in the absence of appropriate control data, i.e., the survival times of skin homografts from their husbands and unrelated donors on normal, multiparous women, no firm conclusions can be drawn from these interesting observations. A. EVIDENCE OF HOSTRESISTANCE Most authorities now agree that the high degree of success obtainable in the treatment of gestational choriocarcinoma depends in part upon its unusual susceptibility to folic acid antagonists, especially methotrexate. The effectiveness of chemotherapy in eradicating the malignant cells can be monitored by highly sensitive radioimmunoassay and other procedures for detecting human chorionic gonadotropin produced by them. However, various clinical findings, collectively appear to support the thesis that patients can, indeed, develop weak sensitivity against gestational choriocarcinoma that may act in concert with cytotoxic drug therapy to produce the long-term remission rate which is far more impressive than that for any other kind of carcinoma or sarcoma (Bagshawe, 1969). Salient among these findings are the following : 1. Despite claims to the contrary, leukocytic infiltration including mononuclear cells, has been observed in approximately 50% of patients studied. Moreover, evidence of significant correlation between the intensity of the infiltrating leukocytic response and the susceptibility of a particular tumor to chemotherapy exists (Bagshawe, 1970; Elston, 1969). 2. In contrast to the autochthonous choriocarcinomas that originate in the gonads of both males and females, spread of gestational choriocarcinomas by lymphatic routes is uncommon as is their presence in lymph nodes. 3. Well-documented instances of spontaneous regression of choriocarcinoma exist but probably below 5%of cases (Bagshawe, 1969). In
24
ALAN E. BEER AND R. E. BILLINGHAM
some of these a cellular response has been noted in surgically removed tumor material. 4. Despite a few notable exceptions, the results of chemotherapeutic treatment of the genetically compatible primary ovarian or testicular choriocarcinomas appear to be inferior to those obtained with the gestational (and alien) forms of the tumor. However, this observation loses some of its significance since tumors of gonadal origin probably include malignant cells of other types. 5. Women respond better to chemotherapy initiated soon after the onset of the disease (Hertz et al., 1961). This may result from some of these feebly antigenic tumors inducing tolerance in their hosts or possibly from enhancing antibodies.
B. FETAL-MATERNAL ISOANTIGENIC COMPATIBILITY AS A PREDISPOSING FAC~OR Repeatedly it has been suggested that choriocarcinoma is a consequence of absence of a maternal immune response to trophoblast ( a tissue normally of finite, intrinsically determined life-span of which the distinctive, functionally important properties include invasiveness ) resulting from chance fetal-maternal compatibility with respect to blood group and other antigens. Consanguineous marriages, alleged to be very frequent in areas where these tumors are common, have been cited as a predisposing factor (Ilya et al., 1967). Comparing the ABO and HL-A antigens of patients with those of their conceptuses or tumors might determine whether compatibility of the offending conceptus with respect to these two major histocompatibility systems (Amos, 1969) is an important predisposing factor for the development of choriocarcinoma. Unfortunately this is not possible, and probable tissue types of the offending conceptus must be deduced from data on the husband and other children when available. By immunofluorescent procedures, ABO erythrocyte antigens were not demonstrable in term trophoblasts (Thiede et al., 1965), although, according to Gross (1966), group A substance was present in the early trophoblast. Bagshawe ( 1970) has compiled and evaluated all available evidence concerning ABO compatibilities of patients and their husbands. This suggested that a very high proportion of the conceptuses from which tumors originated could have been ABO compatible with their mothers. Furthermore, the data suggested that the frequency of group 0 was abnormally high among husbands of patients with choriocarcinoma. An essentially similar picture seems to be emerging for demonstrable compatibilities between husband and wife with antigens of the HL-A series-the number of serologically detectable incompatibilities between
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
25
husbands and patients being less than expected for randomly selected pairs of individuals ( Bagshawe, 1970). In Denmark, Mogensen and his colleagues (1968; Mogensen and Kissmeyer-Nielsen, 1968) obtained evidence which suggested that the existence of HL-A incompatibility, especially with respect to the HL-A2 antigen, between the tumor and patient protected her frdm generalization of the tumor. The generalized tumor, which ran a much more severe clinical course, was probably highly, if not completely compatible with the patient’s strong transplantation antigens. This, of course, constituted indirect evidence of the antigenicity of some of these tumors, However, in studying 3 patients with choriocarcinoma, their husbands, and numerous children, Rudolph and Thomas (1970) used both mixed lymphocyte culture and cytotoxic tests and obtained evidence of major HL-A antigen incompatibilities in the tumors which were not detectable serologically. Additional findings along these lines would, of course, invalidate the thesis that survival and dissemination of postgestational choriocarcinoma depends upon a high degree of histocompatibility between the tumor and maternal host. If further typing data substantiate the provisional conclusion that choriocarcinoma is usually associated with conceptuses that are relatively compatible with their mothers, this would not imply a causal relationship. The most likely interpretation would be that, although malignant transformation of the trophoblast is fairly common in man, tumors that differ from their hosts with respect to strong histocompatibility genes might express the corresponding antigens to an extent sufficient to elicit an effective host response at an early stage, and so never become clinically recognizable. Although moral considerations preclude its performance, study of graft survival of choriocarcinoma in human volunteers would be useful. Possibly information could also be obtained from grafting experiments in subhuman Primates. The means by which choriocarcinoma is able to survive as a homograft will be discussed below (see Section V1,D). Thanks largely to studies on organ transplantation in rats, particularly when donor and host are similar at a major histocompatibility locus, skin grafts have provided an unduly pessimistic picture of the fate of homografts in general. The skin graft is probably one of the most exacting of all homografts, and to prolong its survival by immunological means, by immunosuppressive drugs, etc., is a much more formidable problem than to prolong the lives of heart or kidney grafts (Billingham and Silvers, 1971). Consequently, the continued acceptance of their choriocarcinomas by patients who have rejected skin homografts of
26
ALAN E. BEER AND R. E. BILLINGHAM
essentially similar genetic constitution from reasonably well-matched donors is in better accord with contemporary transplantation biology than seemed to be the case a few years ago. Of course, if it were found that choriocarcinoma would grow in practically any volunteer host, this reasoning would need revision. VI. The Fetus Qua Homograft: Factors That M a y Contribute to Its Success
As knowledge and understanding increased concerning the immunogenetically determined, restricted survival of tumor and normal tissue grafts surgically exchanged between members of outbred populations or of different inbred or isogenic strains, and of the peremptory rejection of second-set homografts, biologists became increasingly aware of the paradoxical success of fetuses as homografts. The classic principle that grafts from an F, hybrid donor were invariably rejected if transplanted to either of its parents obviously did not apply to a naturally transplanted F, hybrid fetus during its development in its mother’s uterus. What must surely represent the earliest experimental investigation of the fetal homograft problem was carried out unwittingly by Walter Heape in London in 1891 when he transferred two fertilized Angora rabbit eggs from a dce mated 32 hours previously, to the fallopian tube of a Belgian hare mated 3 hours beforehand with a buck of her own strain. The subsequent birth of a litter comprising two Angoras and four Belgian hares afforded formal proof that a fetus does not need to have any genetic endowment from its mother (in the form of a haploid set of chromosomes) in order to succeed as a homograft, i.e., there is no gene or antigen dosage effect. Subsequent workers have extended Heape’s contribution by establishing invulnerability to rejection of transferred homologous zygotes in a variety of species, including mice (McLaren and Michie, 1956), rats ( Nicholas, 1932), cattle ( Willett, 1953), and sheep (Averill and Rowson, 1958; Warwick and Berry, 1949). Repeatedly, and always with absolutely no glimmer of success, have transplanters sought to prejudice the fate of fetuses resulting from heterospecific matings in mice (Medawar and Sparrow, 1956; Mitchison, 1953; HaSkov6, 196l), rats (Woodruff, 1958), and rabbits (Heslop et al., 1954; Woodruff, 1958) by presensitization of the mothers against the alien, paternally inherited foreign transplantation antigens of their fetuses. Particularly forceful evidence of the futility of such attempts was the failure of Lanman and his associates (1962) to impair the successful development to term of blastocysts transferred to the uteri of pseudopregnant female rabbits hyperimmunized, by means of skin
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
27
homografts, against the tissue antigens of both parents of the transferred zygotes (the 3 rabbits involved in each such experiment were genetically disparate). However, there is one finding which differs from those summarized above. Breyere and Sprenger (1969), in a well-controlled series of experiments reported a 6-16% reduction in the number of offspring delivered by C57 BL female mice specifically hyperimmunized against DBA- or C3H-strain-specific tumors and subsequently mated heterospecifically to males of these strains. They suggested that the variance of their results with those of other workers might be owing to the presence of a “common” antigen in both the fetal and tumor tissues that was not present in normal adult tissue. Before the significance of this interesting finding can be evaluated, it would be necessary to know at what stage of gestation, elimination of the conceptuses occurred. Indeed, it is conceivable that the immunity was effective against the spermatozoa before fertilization (see work of Michie and Anderson described in Section XIV). To explain this seemingly unqualified success of fetuses as homografts has provided almost as great a challenge to biologists as to explain the significance of the complex multiple allelic systems of histocompatibility genes present in all groups of vertebrates and which are “responsible’’ for posing this central problem of reproductive immunobiology. As many armchair theorists as experimentalists have rallied to the cause. Some of the principal hypotheses or factors which, either singly or in combination, might account for the success of fetal organismic grafts will now be discussed in relation to the available evidence. One factor has already been considered and dismissed as a trivial contributor in nonsensitized mothers, i.e., that the uterus represents an immunologically privileged site ( Section II1,C).
A. COMPLETE SEPARATION OF MATERNAL AND FETALCIRCULATIONS Although the complexity and intimacy of the interface between maternal and fetal tissues display a wide range of variation among species-the basis of comparative placentology-fetal and maternal circulations in the placenta are always completely separate. Apart from its possible physiological significance, this has long been regarded as a crucially important protective factor from the immunological viewpoint ( Medawar, 1953). Breakdown of this vascular quarantine would clearly lead to sensitization of the mother against a multiplicity of isoantigens associated with both cellular and other components of the
28
ALAN E. BEER AND R. E. BILLINGHAM
fetal blood and also facilitate the passage of the resultant antibodies and immunologically competent cells into the fetal circulation. Apart from inducing a state of specific unresponsiveness, or immunological tolerance, regarding transplantation antigens, some of these cells would certainly react against their hosts causing runting or “transplantation” disease (Billingham, 1968). A model of disasterous free maternal-fetal vascular intercommunication is afforded by the phenomenon of parabiosis intoxication, best studied in situations where a parental strain individual is surgically parabiosed to its F, hybrid, which latter becomes intoxicated. Although the clinical picture in this homologous disease syndrome is usually complicated by a shunting of blood from the F, to parental strain partner, leading to anemia and polycythemia, respectively, graft-versus-host reactivity within the victim, mediated by transferred immunologically competent cells, both initiates and makes an important contribution to the disease. To explore the effects of establishing vascular connections between mother and fetus, Jackson ( 1967) ingeniously implanted maternal omentum beneath the skins of dog and sheep fetuses in utero. This resulted in deaths of the dog fetuses within 2 to 3 days and those of the sheep within 3 to 10 days. Pathological features in the dying fetuses included generalized edema, hemorrhages into major organs, and occasional though not conspicuous cell infiltrates. However, before these findings can be taken as indicative of the potential immunological hazard of maternal-fetal vascular parabiosis, it must be shown that the deaths observed were, indeed, due to the activity of maternal immunocytes. Obviously, if this is the case, the phenomenon should not occur if mother and fetus are genetically identical, which necessitates repetition of the work in a species in which inbred strains are available. In striking contrast to the evident dangers of maternal-fetal vascular intercommunication is the apparent harmlessness of synchorial vascular anastomoses that are consistently established between twin or multiple embryos in marmosetts, in the majority of instances of multiple births in cattle, and very rarely in man and sheep, irrespective of the zygosity of the embryos concerned. This situation facilitates an early prenatal exchange of blood cells, including hematopoietic stem cells of various types, leading to a persistent state of erythrocyte and leukocyte chimerism (Benirschke and Driscoll, 1967; Dain and Tucker, 1970). This, of course, is a result of each individual becoming immunologically tolerant of its twin’s transplantation antigens (Billingham et al., 1952, 1956; Billingham and Silvers, 1971). Although such animals are normally incapable of rejecting skin and other tissue homografts from their former parabiotic partners, in later life, they are fully capable of rejecting
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
29
grafts from either parent or from full siblings of separate birth (Billingham et al., 1952). There is one striking discrepancy in the influence of intrauterine synchorial vascular anastomoses between heterosexual twins in cattle and sheep, on the one hand, and in marmosetts and man, on the other hand. In both ungulate species, females born cotwin with males are normally sterile, whereas in the Primates they are phenotypically normal in all respects. Possible explanations of this disparity are discussed by Benirschke and Driscoll ( 1967). The biological significance of the synchorial vascular anastomoses in marmosetts, which regularly have fraternal twins, awaits explanation ( Hampton and Hampton, 1965). Obviously lack of vascular connections with their maternal hosts cannot explain either the inability of fetuses to elicit transplantation immunity or their resistance to it. This was evidenced when grafts of pure epidermis, or epidermal cell suspensions, transplanted to appropriate beds on the chests of rabbits and guinea pigs elicited a host immune response to which the grafts succumbed, despite the fact that this tissue was always avascular (Barker and Billingham, 1971; Billingham and Sparrow, 1954). B. ANTIGENICIMMATURITY OF THE FETUS Little’s (1924) ingenious suggestion that “the embryo has no definite physiologic characteristics which are individual enough to be recognized as foreign to the mother” was soon placed in jeopardy by early observations that minced embryonic tissue homografts implanted in adult hosts effectively elicited transplantation immunity to test grafts of tumors and other tissues of similar genetic origin. However, such findings were indecisive in that they failed to exclude the possibility of antigenic maturation on the part of the fetal grafts subsequent to their transplantation. More critical studies have minimized or obviated this possibility by ( 1 ) restriction of the time available for the embryonic cells to sensitize their hosts, (2) preclusion of further differentiation of embryonic cells by irradiation, ( 3 ) evaluation of the capacity of embryonic cellular inocula to induce tolerance in infant hosts whose age is such that their ability to become tolerant following inoculation with relatively small numbers of cells is almost at an end, ( 4 ) determination of the ability of embryonic cell preparations to absorb specific antibodies, or ( 5 ) direct visualization of the antigenic sites on the cells by fluorescent antibody procedures ( Davies, 1968; Billingham and Silvers, 1971; Palm et al., 1971). These studies have all indicated the presence of some transplantation antigens very early in ontogeny. The demonstration that mouse tuba1 eggs failed to develop if transferred beneath the renal
30
ALAN E. BEER AND R. E. BILLINGHAM
capsules of specifically prehyperimmunized instead of normal homologous hosts (Kirby et al., 1966; Simmons and Russell, 1966) hinted that some histocompatibility specificities were present in ova at the pretrophoblastic stage, However, the findings did not indicate whether the effective antigens were determined by the H-2 locus. Heyner et al. (1969) studied the influence of isoantibody and complement on the development of tuba1 mouse eggs to blastocysts in vitro. After careful removal of the zona pellucida with pronase, antisera containing both H-2 and non-H-2 antibodies in the presence of complement caused the eggs to degenerate. However, sera known to contain only specific H-2 antibodies failed to interfere with the formation of blastocysts, suggesting either the absence of H-2 determinant sites at this stage, or a distribution too sparse or sterically inappropriate so as to preclude a cytolytic effect. Subsequent work by Palm et al. (1971), using indirect immunofluorescence, confirmed the presence of weak antigens corresponding to so-called minor histocompatibility loci ( H-3 and H-6) diffusely distributed over the entire surface of blastomeres at the two-cell stage, but H-2 specificities were not detectable either at the two-cell or blastocyst stages. These weak antigenic determinants on the blastomeres reacted just as strongly with the specific antisera as did similar determinants on lymphocytes of the same genetic constitution. The presence of transplantation antigens on the surface of mouse egg cells at the morula stage was also established by Olds (1968) by means of an indirect fluorescent antibody procedure. In view of the antisera used, however, it was questionable whether the elements revealed were determined by the H-2 locus as she claimed. Thus, at least in the mouse, antigens determined by the major histocompatibility locus seemed to develop later in ontogeny than those determined by some of the minor loci. The absence of H-2 determinants on early egg stages may be contrasted with their presence on spermatozoa (see Section IV,B). As Palm et al. (1971) pointed out, it is conceivable that sperm may have a better opportunity to absorb H-2 substances from their fluid milieu during storage and maturation. Even in the perinatal period certain tissues have proved to be immunogenically inferior to those from more mature individuals. In mice, skin homografts from infant donors transplanted to adult H-2 compatible hosts may long outlive grafts from adult donors (Wachtel and Silvers, 1971). Essentially similar findings have been obtained in hamsters and rabbits ( Billingham and Silvers, 1964). Obviously, of much greater relevance to the behavior of the fetus as a homograft is the antigenic status of its extraembryonic tissues, par-
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
31
ticularly the placenta, for this is the site at which the confrontation between maternal and fetal tissues occurs over a considerable surface area. Several theories have been advanced to account for the apparent lack of antigenicity of the trophoblast which need only be briefly mentioned. Gordon’s ( 1960) premise that the syncytiotrophoblast (i.e., the outer layer of the trophoblast and the one in contact with maternal tissue) develops from maternal ovarian follicle cells rather than from the implanting fertilized egg is inconsistent with much embryological evidence. Galton’s ( 1960) suggestion, based on cytological evidence (presence or absence of Barr bodies) that the syncytiotrophoblast is haploid, being derived from the maternal set of chromosomes of the zygote has been refuted by evidence that this tissue arises by differentiation of the cytotrophoblast rather than by any unique mitotic activity. However, the success of homologous zygote transfers involving normal or presensitized surrogate mothers, such as the experiments of Heape (1891) and Lanman et al. (1962), utterly refutes the validity of any theories that attribute the success of fetuses as homografts to the absence of paternally inherited antigens in the syncytiotrophoblast. In general, grafting experiments conducted with placental tissue or cell suspensions prepared therefrom, have been too crude to be very informative. They have usually entailed grafting placental material from F, hybrid fetuses to normal host mice of the maternal strain, so that contaminating cells of maternal origin could make no contribution to any sensitivity evoked. It has been established by this approach that ( a ) paternally inherited transplantation antigens were present in the placenta, possibly associated with contaminating fetal leukocytes, or “passenger” cells (Billingham, 1971) and ( b ) F, hybrid placental grafts were vulnerable to rejection in specifically presensitized maternal strain hosts. At least some components of this composite organ were thus susceptible to rejection (Simmons and Russell, 1962; Uhr and Anderson, 1962). Kirby (1968a) suggested that HagkovCis (1963) claim that even relatively large doses of F, hybrid mouse placentas were practically ineffective as a source of paternally inherited transplantation antigens might have been attributable to heavy contamination of the material tested with maternal decidual tissue.
C. NONSPECIFIC WEAKENING OF THE IMMUNOLOGICAL REACTIVITYOF THE MOTHERDURING PREGNANCY During pregnancy there is increased production of certain hormones, notably adrenal corticosteroids of the cortisone family ( glucocorticoids) , which, when administered in abnormally high dosages under experimental conditions, bring about a transient lymphocytopenia
32
ALAN E. BEER AND R. E. BILLINGHAM
and involution of lymphoid tissue and have weak immunosuppressive properties ( Medawar, 1953, 1969). Evidence also suggests decreased resistance to certain infections and ameliorated symptoms of certain diseases of suspected autoimmune etiology, such as rheumatoid arthritis, during pregnancy in humans. On the basis of this kind of information, Medawar (1953) suggested that the immunological capacity of the mother might undergo some sort of physiological change that would protect the fetus against the dangers of immunizing its mother. Evidence favoring this hypothesis was Heslop et d ’ s (1954) finding that skin homografts transplanted to rabbits that were about 3 weeks pregnant survived nearly twice as long as grafts transplanted sooner or later in pregnancy or to nonpregnant animals. This observation was interpreted as indicating that the peak of enhanced steroid secretion occurred during the latter part of pregnancy. In humans, too, there was a hint that pregnancy might weaken the homograft reaction (Andresen and Monroe, 1962). In mice, extensive investigations have shown that homospecific ( or intrastrain) pregnancies did not perceptibly weaken a female’s capacity to react against H-2 incompatible skin homografts, though there was a trivial but, nevertheless, significant weakening of their capacity to react against skin grafts confronting them with only minor alien histocompatibility factors ( Medawar and Sparrow, 1956; Simmons et al., 196713). In cattle, with a gestation period of about 280 days, pregnancy had no influence on the development of homograft sensitivity (Billingham and Lampkin, 1957). Although, in both experimental and clinical contexts, certain glucocorticoid hormones have the important property of being able to erase immunological memory (Medawar, 1969), th’is requires the administration of high doses over prolonged periods. Neither in the rabbit nor in any other species is there any convincing evidence that pregnancy can weaken a preexisting state of homograft immunity. At best, therefore, the increased steroid production associated with pregnancy, and to which both fetus and its placenta probably contribute, can only be regarded as affording a weak ancillary mechanism for preventing the development of maternal isoimmunity during pregnancy in a few species. However, one additional possibility remains on probation. Since, in rabbits, cortisone could exercise its effect when applied topically on a skin homograft at dosages which had little activity when administered systemically (Billingham et al., 1951), some of the hormones secreted by the placenta might act locally and interfere with immunological transactions between maternal lymphocytes and fetal cells in this organ.
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
33
D. PHYSIOLOGICAL BARRIER BETWEEN MOTHERAND FETUS The most striking quality of the fetus as a homograft in utero is its apparent total refractoriness to a state of specific immunity or hyperimmunity on the part of its mother directed against its own transplantation antigens. This resistant property-a form of efferent inhibition of the immune response-is expressed most dramatically in species with hemochorial placentas, where maternal blood, which bears both antigen-sensitive cells and the cellular effectors of transplantation immunity, is continuously in contact with a large aggregate area of fetal trophoblast cells (estimated at 10 to 15 m.2 in humans). Only one hypothesis (Medawar, 1953) is capable of accounting for the survival of fetuses in both normal and sensitized mothers, i.e., they are surrounded by some kind of physical or anatomical barrier capable of ( a ) preventing the mother from developing a state of transplantation immunity to the alien histocompatibility antigens of her fetuses and ( b ) affording the latter complete protection against high levels of sensitivity procured experimentally by prior grafting with normal tissues of appropriate genetic origin. Simply because it represents the continuous, uninterrupted “frontier” component of the fetus, the trophoblast has long been under suspicion as fulfilling this role. As already indicated, the anatomical complexity of the maternal-fetal relationship in the placenta which is determined by invasiveness of the trophoblast, varies from species to species. It ranges from simple close apposition of the trophoblast to the intact endometrial epithelium of the ‘uterus, as in pigs and horses, to an erosive penetration of trophoblastic tissue through the endometrial epithelium, its subjacent connective tissue, and even the endothelium of maternal veins, so that fetal trophoblast cells covering the trophoblastic villi are in direct contact with, and actually bathed by, maternal blood as in the hemochorial placentas of humans, rodents, and rabbits.
1. Incrimination of Trophoblast As Quarantining Layer Circumstantial evidence pointed toward the trophoblast as the prime candidate for a protective or quarantining role, notably the discovery by Witebsky and his associates (Oettingen and Witebsky, 1928; Witebsky and Reich, 1932) that human placental villi are deficient in blood group antigens, recently reconfirmed by Thiede et al. (1965; Gross, 1966). Indeed, on this basis, Witebsky et al. put forward the idea that placenta could function as a barrier if its trophoblast cells were nonantigenic. In 1959, Bardawil and Toy sponsored another can-
34
ALAN E. BEER AND R. E. BILLINGHAM
didate-a local zone of fibrinoid substance (possibly a degeneration product), known as the layer of Nitabuch which, in man, normally separates maternal and fetal tissue where invading trophoblast meets decidual tissue. According to these workers this layer might behave “as an immunological no man’s land, walling the fetus off from chemical interaction with its host.” The success of choriocarcinoma as a foreign tumor has long been interpreted as reflecting nonisoantigenic properties of its normal tissue of origin-cytotrophoblast. Indeed, this is probably the only example of a tumor which has shed light on the biology of its normal tissue of origin. The development of human and other fetuses for long periods, and occasionally to term, in a variety of nonuterine ectopic sites, such as the fallopian tubes, the ileum, and rectum and peritoneum (see p. 5 ) affords compelling evidence that the immunological quarantining layer is of fetal origin and highly versatile in its capacity for functional deployment. Scrutiny of reports on ectopic pregnancies reveals no grounds for suspecting that immunological reactivity on the part of the mother plays a causal role in fetal death. 2. Transplantation Studies on Trophoblast The first discriminating analysis of the histocompatibility properties of mouse placental tissue was that of Simmons and Russell (1962) who studied the histocompatibility characteristics of mouse placental tissue at various stages of its development, In most of their experiments, grafts from F, hybrid embryos were transplanted to hosts of the maternal strain, since in this situation the accidental inclusion of maternal tissue fragments and cells could affect the hosts immunologically. Initially they confirmed the findings of others that grafts of placentas from lO#-day embryos elicited and succumbed to a typical host response, whether implanted intramuscularly or to host sites prepared in the integument. Then, recogrlizing the difficulty of interpreting results of this kind of experiment in which the graft included a diversity of cell types, they took advantage of the ease with which 7%-daymurine embryos were separable into trophoblastic precursors (the so-called ectoplacental cone) and the embryo moiety. When transplanted beneath the renal capsules of maternal strain hosts presensitized against paternal strain tissue antigens, embryonic grafts were totally destroyed within 7 days. By contrast, the trophoblastic homografts displayed marked proliferative activity on the part of their giant cells which, by virtue of their phagocytic and invasive properties, formed typical blood spaces.
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
35
Furthermore, there was little cellular response against these grafts on the part of the host. These trophoblastic grafts underwent a sort of nonspecific involution after about 12 to 13 days, just as they did in genetically compatible hosts. It was also noted that any trophoblastic cells accidentally included in embryonic grafts transplanted to presensitized hosts seemed to be completely unaffected by the cell-mediated destructive process that destroyed their nontrophoblastic neighbors. When fertilized eggs (at about the two- to eight-cell stages) were removed from the fallopian tubes and transplanted ectopically beneath the renal capsules or into the spleens of isogenic hosts, a significant proportion of them developed into small tumors of trophoblastic giant cells, displaying infiltrative properties and having a life-span of about 2 weeks. When F, hybrid ova were transplanted beneath the renal capsules of unimmunized maternal hosts, no evidence was obtained of curtailment of their life-span or of sensitization of the host to subsequent test skin homografts from the paternal strain. More impressive was the continued survival and transient growth of these pure trophoblastic grafts when transplanted to hosts presensitized by two consecutive skin grafts. Even more striking evidence of the ability of the trophoblast to override strong histocompatibility barriers was afforded by the birth of healthy interspecific hybrids after matings between horses and donkeys, sheep and goats, leopards and tigers, cattle, bison, and yak in different combinations, and bears of different species, etc. (Gray, 1954). In accord with such observations is the fact that, with certain species combinations, heterografts of trophoblastic tissue survived and displayed typical mitotic-invasive properties. Although rat eggs displayed only feeble development when inserted beneath the renal capsules of mice, upward of 25%of mouse eggs placed in the kidneys of rat hosts produced a flourishing trophoblast that invaded and phagocytized the surrounding tissue, just as it would have done in a host of its own species (Kirby, 1962). However, when a 6%-day mouse embryo was separated from its trophoblastic moiety and similarly transplanted, it incited a massive cellular response within the same time interval. Mouse trophoblast grafts also grew in the testes of rats and hamsters, a hamster trophoblast grew in mice, and in none of these combinations was there evidence of a host immunological response ( Billington, 1966). Simmons and Russell's (196%) finding that the growth of mouse ectoplacental cone grafts placed beneath the testicular capsules of rats was inhibited by prior grafting of the hosts with mouse skin indicated that the trophoblast expressed species-specific antigens.
36
ALAN E. BEER AND R. E. BILLINGHAM
3. Ultrastructural and Histochemical Studies on the Mouse Placenta-the Peritrophoblastic Fibrinoid Hypothesis On the basis of electron-microscopic and histochemical studies of murine placentas, Kirby et al. (1964; Bradbury et al., 1965) &rmed that the trophoblast represented an immunological buffer zone between mother and fetus and presented observations in favor of a thesis closely similar to that of Bardawil and Toy (1959). They believe that transplantation ( and probably species-specific) antigens are present in trophoblast cells but are probably unable to escape or express themselves effectively because each trophoblast cell is surrounded by a layer of amorphous, electron-dense fibrinoid material of mucopolysaccharide constitution ranging from 0.1 to 2 . 0 in ~ thickness. They draw attention to the close histochemical similarity that exists between the intercellular matrix of the hamster’s cheek pouch connective tissue, which appears to underlie the immunologically privileged status of this organ (Barker and Billingham, 1971), and placental fibrinoid material (Heyner, 1970). However, there is one important difference between the two situations -whereas hamster cheek pouch fibrinoid substance is unable to protect vascularized homografts from an extant state of sensitivity, the fibrinoid associated with trophoblast cells apparently does have this property. Apart from trophoblast, the only other tissue with cells able to withstand transplantation immunity is cartilage. Recent studies on chondrocytes isolated enzymatically from their matrix indicate that these cells have transplantation antigens and are susceptible to rejection by homologous hosts ( Billingham and Silvers, 1971; Heyner, 1970). However, the avascularity and physicochemical properties of the matrix in which these cells are normally “embedded,” and which they actually secrete, not only prevents them from sensitizing their hosts but also protects them from the familiar hazards of transplantation immunity. An important observation sustaining Kirby et aZ.’s (1964) thesis that peritrophoblastic fibrinoid material fulfills an immunological masking or concealing role is that its amount is related to the immunogenetic disparity between the mother and her fetus. Histocompatible fetuses, they claim, have less fibrinoid than F, hybrid fetuses resulting from zygotes transferred between females of different inbred strains. 4. Evidence of “Masked” Transplantation
Antigens in Trophoblast Observations sustaining the title thesis and indicating that trophoblast cells are not isoantigenically inert were presented by Currie and Bag-
IMMUNOBIOLOCY OF MAMMALIAN REPRODUCTION
37
shawe (1967; Currie, 1967). They found that growth of trophoblast in vitro in the presence of lymphocytes of either maternal or homologous origin caused it to undergo gross cytolysis. Postgestational choriocarcinoma cells suffered a similar fate on exposure to host lymphocytes in culture. However, trophoblast cells confronted by lymphocytes of their own genetic constitution in vitro were not damaged. The authors attributed this destructive action of lymphocytes upon homologus tropohoblast cells to the closely related phenomena of allogeneic inhibition or contact-induced cytotoxicity, said to result from intimate contact between unsensitized lymphoid cells and homologous target cells having different surface structural properties determined by histocompatibility genes (I. Hellstrom and Hellstrom, 1966; Moller and Moller, 1966). Currie and Bagshawe (1967) postulated, on the basis of a variety of observations, that the peritrophoblastic layer of fibrinoid, or sialomucin as they identified it, confers a negative charge on the cells by virtue of free carboxyl groups on sialic acid. Since lymphocytes likewise carry a negative charge, they suggested that in vivo the trophoblast escapes interaction with, or attack by, lymphocytes as a consequence of electrochemical repulsion of the latter. This attractive line of reasoning has been challenged as representing an oversimplification (B. M. Jones and Kemp, 1969; see below). What appeared to be unequivocal evidence of the presence of transplantation antigens on mouse trophoblast cells and of their normal masking b y pericellular sialomucin was subsequently presented by Currie and his co-workers (1968). They found that treatment of ectoplacental cone .cells from 7?;-day mouse embryos with neuraminidase in vitro enabled them to sensitize unrelated adult hosts against subsequent test skin homografts of their own genetic makeup. This observation is so important that it merits independent confirmation using other mouse strain combinations and extension to other species. Neuraminidase, which specifically removes sialic acid groups from sialomucins, was previously shown to be capable of revealing normally covert antigenic determinants on certain kinds of tumor cells. Lippman (1968) has reported that treatment of mouse tumor cells in vitro with various acid mucopolysaccharides may lead to the “suppression” of certain isoantigens that are normally present. B. M. Jones and Kemp ( 1969), in a recent critical review of the modus operandi of sialomucin in the “self-isolation” of fetal trophoblast, cited a variety of cogent biochemical and biophysical objections to the thesis that adhesive and nonadhesive properties of tissue cells are determined solely by the interplay of physical forces. This evidence also makes it hard to believe that attachment of lymphocytes to the sur-
38
ALAN E. BEER AND R. E. BILLINGHAM
face of the trophoblast is prevented by a sialomucin-associated electronegative barrier. According to them, cells are probably held together by chemical bonds at adhesive sites on their surfaces, and cells of similar type but of disparate genetic constitution can adhere to one another just as well as cells having similar genomes: This probably accounts for the firm initial adhesion of epithelial trophoblast cells to genetically different maternal endometrial cells during the early stages of development. As a consequence of some kind of informational exchange, possibly facilitated by the intercellular bridges that unite trophoblast and maternal cells, trophoblast may recognize its “genetic disaffinity” with the uterine epithelium and respond by secreting an intervening layer of sialomucin, thus quarantining itself. In addition, the authors suggest that this material renders the outer surface of trophoblast cells nonadhesive by masking the adhesive sites which had previously united trophoblastic to uterine epithelial cells. In the opinion of these reviewers, masking of these sites also prevents maternal lymphocytes from adhering to and interacting immunologically with trophoblast.
5. Sources of Concern about Morphological Identification of Trophoblast-Associated Protective Barrier Both the source of the trophoblast-associated sialomucin and its constancy of association with trophoblast in different species have yet to be established on an experimental basis. Reports of failure to detect pericellular coatings of sialomucin or fibrinoid substance on trophoblast cells by electron microscopy must not be neglected in this context. Simmons d d. (1967a) in a study of ectopic mouse trophoblast of blastocyst origin, growing beneath the renal capsules of both isogenic and homologous male hosts, were unable to find any difference between the compatible and the genetically incompatible grafts, both of which invaded the host tissue and in neither of which was there any electrondense fibrinoid material associated with the trophoblast cells, In an essentially similar study, Kirby and Malhotra (cited by Kirby et al., 1964) reported that the fibrinoid layer was conspicuously present around the cells. In the rat placenta, Martinek‘s ( 1970) recent ultrastructural studies indicated that at no time from midgestation to term did electron-dense fibrinoid material form an intact barrier between the fetal trophoblastic giant cells and maternal decidual cells. However, increased amounts of interfacial fibrinoid were observed as. the time of parturition approached. Significant amounts of viable trophoblast and decidua appeared to be intimately juxtaposed throughout the latter half of preg-
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
39
nancy. This finding and a similar one in the mouse (Martinek, 1971), in conjunction with some of the others considered in this section, cast doubt upon the general thesis that the thicker layer of fibrinoid described by Kirby et al. (1964) is the essential trophoblast-cell-associated immunological quarantining layer. It does, however, leave open the possibility that this role is fulfilled by an extremely thin and less easily detectable cell coating of mucopolysaccharide-or a glycocalyx. In the normal rabbit placenta, Tai and Halasz (1967) observed “prominent deposition of fibrinoid material in the intercellular area of the trophoblastic cell layer,” but they failed to reveal this material on trophoblastic microvilli. In an important series of comparative ultrastructural studies of several different kinds of hemochorial placentas, including that of man, Wynn (1967a,b, 1969) found material morphologically identical to the trophoblast cell-associated fibrinoid described by Kirby et al. in the mouse and, more recently, in man (see, also, Bradbury et al., 19f39) surrounding and apparently produced by the decidua. He interpreted his findings as suggesting a positive correlation of invasiveness of the trophoblast with ultrastructural complexity of the decidua, extent of necrosis of adjacent fetal and maternal tissues, and formation of noncellular barriers. He viewed the histologically demonstrable “fibrinoids” merely as effects of cellular interactions between trophoblast and endometrium rather than as primary factors in immunological protection of the alien trophoblastic cells. In epitheliochorial placentas, as found in cows and sows, microvilli of chorionic and endometrial epithelia intermingled without signi6cant necrosis and without deposition of fibrinoid. However, appropriate stains for electron microscopy did reveal an extracellular coat of mucopolysaccharides morphologically similar to those associated with epithelial microvilli of other tissues, such as the intestinal mucosa, which enjoyed no exemption from rejection. Of course, as Wynn was careful to point out, one might question whether in placentas of this type, where the trophoblast was not normally exposed to immunologically competent maternal lymphocytes, there was need for either trophoblastic or extratrophoblastic protection. VII. Susceptibility of Pretrophoblastic Eggs to Transplantation Immunity
Additional, cogent evidence concerning the competence of trophoblast to provide an effective immunological buffer zone has come from comparison of the fates of homologous 2%-day (postconception) fertilized mouse eggs and ectoplacental cones from 7-day homologous embryos transplanted to specifically hypersensitized hosts. Simmons and Russell (1966) found that the proportion of C3H
40
ALAN E. BEER AND R. E. BILLINGHAM
eggs transplanted beneath the renal capsules of C57 male mice, presensitized to various degrees against C3H strain tissue antigens, which developed into trophoblastic tissue was inversely related to the level of immunity in the sensitized host. For example, trophoblasts failed to develop in hosts presensitized by two consecutive skin grafts followed by 8 to 12 spleen cell injections. However, (C3H x C57) F, hybrid eggs proved to be less susceptible in similar hyperimmune hosts, probably because of a gene-dosage effect, the hybrid cells having fewer alien determinant sites on their membranes. It is important to note that whenever trophoblastic proliferation did occur in a hypersensitized host, its extent was not demonstrably curtailed. In marked contrast to the vulnerability of ectopically transplanted C3H fertilized egg homografts in hyperimmune C57 hosts, grafts of trophoblast from ectoplacental cones from 7-day C3H embryos proved to be completely refractory. On the basis of these important observations, Simmons and Russell (1966) concluded that transplantation antigens are present in pretrophoblastic embryos but the trophoblast represents a specialized form of embryonic cell that is either incapable of manufacturing, or expressing on its surface, antigens displayed by its immediate precursors. These findings have been confirmed and extended by Kirby et al. (1966), working with the C57+ C3H strain combination. They found that both two- to eight-cell tuba1 eggs and 3%-dayblastocysts failed to develop beneath the renal capsules of specifically hyperimmunized C3H male hosts, whereas ectoplacental cones developed normally. However, there was no impairment of the development of homologous blastocysts transplanted orthotopically to the uteri of pseudopregnant hyperimmune C3H females. Histological examination of the placentas of these embryos revealed no signs of any immunological reactivity on the part of their hosts. Although Kirby et al. (1966) ascribed the normal development of the blastocysts in the uteri of the hyperimmune animals to the immunological protective function of the surrounding decidual tissues, this interpretation seems improbable in the light of recent findings of Beer and Billingham (see p. 10). Another possibility is that, through some kind of interaction between decidual tissue and trophectoderm, the latter acquires its immunological protective properties prematurely. Finally, since in both Simmons and Russell's (l9f36) experiments and those of Kirby et al. (1966) the hyperimmunized hosts had very high titers of antibody, humoral immunity rather than cellular immunity might have been responsible for the inability of the ectopically im-
IMMUNOBIOLOGY OF M A M M A L I A N REPRODUCTION
41
planted zygotes to develop. Clearly there is a need for passive transfer experiments to resolve this latter question. Several investigators have suggested that the zona pellucida, which is impermeable to many proteins and to cells, affords immunological protection to fertilized eggs prior to the formation of the trophoblastic barrier in females who, for some reason, are hypersensitive with regard to their mates’ alien tissue antigens. Simmons and Russell (196%) attributed the occasional survival of ectopically transplanted mouse eggs in hyperimmunized hosts to protection afforded by persistence of the zona pellucida. The results of studies in which mouse blastocysts, with or without their zonas removed, were cultured in vitro in the presence of either immune serum (plus complement) or lymphoid cells from specifically sensitized mice of another inbred strain indicated that the zona (which contains sialic acid) could confer some measure of protection upon blastocysts against both cellular and humoral immunity (Heyner et al., 1969; James, 1969). By appropriate endocrinological procedures, Kirby ( 1969) deliberately prolonged the zona-free existence of transplanted H-2 locus-incompatible blastocysts in the uteri of hyperimmunized females. Despite their prolonged firm attachment to the wall of the host uterus in a zonafree state, there was no evidence that the survival rate of potentially susceptible zygotes differed significantly from that of control blastocysts, thus casting some doubt upon the immunological significance of the zona pellucida in vivo. Vlll. Histoincompatibility as a Determinant of Placental Size and Extent of Trophoblastic Invasion
Despite the invulnerability of fetuses in utero to a state of specific sensitivity directed against their alien tissue antigens, there is evidence that at least in some species the size of the placenta, and probably the extent of trophoblastic invasion, are affected by ( I ) the existence or otherwise of genetic disparity between an embryo and its mother and (2) by the immunological status of the latter with respect to the alien antigens of her fetuses, i.e., whether normal, specifically immune, or tolerant. In 1964 Billington made the interesting observation that (C57 BL X A,G) F, hybrid fetuses, differing from their C57 BL mothers at the important H-2 locus (as well as at other H loci), had significantly heavier placentas than did homozygous fetuses of either parental strain (see also McLaren, 1965b). Comparison of the size of placentas from intrastrain C57 matings with those which developed when fertilized C51
42
ALAN E. BEER AND R. E. BILLINGHAM
eggs from such matings were transferred to the uteri of H-2 locus-incompatible A2G surrogate mothers dismissed the obvious possibility that hybrid vigor was solely responsible. These observations have been confirmed by James (1965, 1967) and by Beer and Billingham (1971) in mice. The latter authors found that the principle also applied to Syrian hamsters but not to a variety of different strain combinations of rats tested. To evaluate the influence of fetal-maternal incompatibility on the invasive properties of trophoblast uncomplicated by a uterine milieu with its inevitable decidual response, Billington ( 1965) subsequently transplanted ectoplacental cones from 7X-day postcoital murine embryos to the testes of adult hosts. These ectopic grafts produced luxuriant growths of trophoblast, reflected in a measurable increase in weight of the organ, paralelling histological evidence of the degree of trophoblastic invasion. When the trophoblast was transplanted to H-2-incompatible hosts the extent of trophoblastic invasion proved to be greater than when transplanted to isogenic hosts. The experiments so far described merely established that fetalmaternal genetic disparity was a determinant of placental size and trophoblastic invasion in the mouse. A subsequent report by Kirby et al. (1966) that when ectoplacental cones were transplanted to the kidneys of specifically hyperimmunized hosts the trophoblast which developed “appeared to have invaded the host organ in an exceptionally vigorous manner” hinted that the phenomenon had an immunogenetic basis. Subsequent studies by James (1965, 1967) confirmed this interpretation and extended Billington’s work. He demonstrated that the immunological status of the mother with regard to the alien tissue antigens of her fetuses was an important determinant of placental size and the growth of the fetus. In C57 BL mothers, presensitized against A,G tissue antigens, (C57 BL x A2G) F, fetuses developed significantly larger placentas than similar fetuses born by normal mothers. Furthermore, the placentas of similar F, fetuses borne by mothers which had been rendered tolerant of the antigens of the A,G strain were significantly smaller than those of normal, untreated mothers. These important observations indicated that immunological reactivity per se on the part of the mother was in some way responsible for placental size. James’ histological studies suggested that the increased placental weights attributable to immunological factors might be due to incorporation of more decidual tissue in the placenta but left open the question whether more extensive trophoblastic invasion was involved. Koren et al. (196813) found that fertilized mouse eggs transplanted to kidneys of heavily irradiated homologous or isologous hosts gave
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
43
rise to more luxuriant and longer-lived growths of trophoblastic tissue than when transplanted to similar but unirradiated hosts. Transplantation antigens were clearly not involved, and the authors’ conclusion that impairment of the host’s capacity to respond to any kind of trophoblast-associated antigen was causally responsible for the enhanced growth of the grafts in the irradiated hosts was not substantiated. X-rayinduced changes in the properties of the host’s blood vessels or in the properties of its blood could equally well account for their observations. These findings pose many interesting questions. For example, ( a ) in the mouse, can genetic disparities not involving the H-2 locus cause placental enlargement? ( b ) Does the principle apply to species other than mice and hamsters? A report by Hancock et al. (1968) that invasion of goat uterine tissue by trophoblast from goat x sheep hybrid fetuses appeared to be more active than normal may be pertinent. Finally, ( c ) is the immunity involved a humoral or a cellular one? This latter question can easily be resolved by appropriately designed experiments involving transfer of antibodies or lymph node cells from sensitized mice. Since repeated heterospecific pregnancies in mice and other animals do lead to the appearance of isoantibodies in multiparous females, if humoral antibodies are determinants of placental size, then the placentas of the fetuses in later litters can be expected to be larger than those of earlier litters by the same mother. It is interesting to note that an increase in placental weight with birth order has been described in mice, rats, guinea pigs, and man (W. R. Jones, 1968) though there is as yet no evidence that immunological factors were involved. In an attempt to detect an immunological influence on placental development in man, W. R. Jones (1968) has carried out analyses of maternal ABO blood groups and placental weights from 3688 consecutive confinements, recognizing, of course, that the presence of these antigens on trophoblasts was in doubt and that there must have been many other histocompatibility differences between fetus and mother. Since blood group data were not available for the children, the expected proportions of ABO-incompatible pregnancies for the 0, A, B, and AB maternal groups were estimated from the gene frequencies in the population. The results obtained suggested that disparity between fetus and mother with respect to these antigens was associated with a relatively smaller placenta and vice versa, i.e., the situation in man appears to be exactly opposite to that of mice. IX. Organ-Specific Antigens of the Placenta
So far we have only considered the extent to which antigens determined by segregating histocompatibility genes are expressed by the
44
ALAN E. BEER AND R. E. BILLINCHAM
placenta and choriocarcinoma. The next question is whether this remarkable transitory organ possesses any antigens that are unique unto itself, i.e., organ-specific antibodies? The trophoblast is a highly specialized tissue with distinctive ingredients and products, including chorionic gonadotropin. It has some kind of immunological “self-quarantining” property and, apart from a transitory exposure to their own trophoblast during fetal life, animals (and then only females) are not reexposed to this tissue until adult life as a consequence of pregnancy. There is thus a prima facie case that this tissue, like brain, lens, and testis, may contain its own private autoantigens, i.e., specific groupings on certain protein or other molecules not represented in antibody-producing cells. The latter have had no opportunity to become tolerant or specifically unresponsive to trophoblast and consequently they may be capable of reacting immunologically against it (Brent and Medawar, 1959; Voisin, 1970). Although it must be conceded that an individual may have an opportunity to become tolerant of these postulated trophoblast-specific antigens in fetal life, once his placenta “life-line” is severed at birth, he would be deprived of the antigenic stimulus now considered to be necessary to maintain this antigen-induced state of nonreactivity ( Nossal, 1968). Interest in the possible organ-specific antigenicity of the placenta dates back to the beginning of the century when it was first suggested that the clinically important toxemias of pregnancy, a group of common diseases characterized by hypertension, edema, and proteinuria, might in some way be owing to maternal sensitization against placental antigens, secondarily leading to renal damage (Hellman and Eastman, 1966).
A. BIOLOGICAL A m v m OF HETEROLOGOUS ANTIPLACENTAL SERUM The principle that an antiserum raised in one species against placental homogenates, etc., from another species interrupts pregnancy in the species that provided the antigen was first established in guinea pigs and rabbits by Dobrowolski in 1903. Since then it has been established that placental degeneration and fetal death could be procured by administration of heterologous antiplacental serum in rats and mice and that antibodies to contaminating erythrocytes were not involved (Seegal and Loeb, 1940, 1946; Koren et al., 19Wa), suggesting the presence in this organ of at least one specific antigen. More important from the clinical viewpoint were demonstrations that these antiplacenta sera also had a striking nephrotoxic tffect (McCaughey, 1955; Bevans et al., 1955; Seegal and Loeb, 1946). Potent
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
45
nephrotoxic heterologous antikidney sera, on the other hand, have not been shown to damage the placentas of rats or dogs (Boss and Craig, 1963; Bevans et al., 1955), although Brent and Averich (1961) reported the development of fetal congenital anomalies in the offspring of mothers injected with nephrotoxic serum on the ninth day of gestation.
B. CROSS-REACTIVITY BETWEEN RENALAND PLACENTAL ANTIGENS Elucidation of the nature and distribution of the antigens apparently shared in common by placenta and kidney has been the subject of numerous studies using either crude homogenates or definable fractions prepared by differential centrifugation, etc. To evoke antibody formation it is virtually essential to administer the antigenic material in Freund's complete adjuvant. The specificity of the resultant antisera, after appropriate absorptions with various types of cell or homogenate, has been evaluated by in vivo organ toxicity tests, immunodiffusion, hemagglutination, application of direct or indirect fluorescent antibody procedures to freshly prepared sections of various tissues, and localization of passively administered antibodies in vivo by indirect fluorescent antibody or radioactively labeled antibody methods (see, for example, Boss, 1965; Boss and Craig, 1963; Curzen, 1968; Pressman and Korngold, 1957; Koren et al., 1968a, 1969; Steblay, 1962). The results of studies in a variety of species, including man, are consistent and can be summarized as follows: ( 1 ) antiplacental and renal antiglomerular basement membrane antibodies localize in vitro in what appears to be an identical manner in the basement membranes of glomeruli, tubules, capsules, intertubular capillaries, and certain extracellular sites in the media and adventitia of arteries; (2) both antisera display similar patterns of localization in placental tissue, on the basement membranes of the labyrinth and trophoblast, Reichert's membrane, and the yolk sac; ( 3 ) soluble antigens common to placenta and kidney are demonstrable by immunodiffusion but cannot be localized immunohistologically; and ( 4 ) common antigens are demonstrable in the mitochondria1 and microsomal fractions of the placental trophoblast and renal proximal tubule epithelium (there is strong evidence that the principal, if not the only, source of these antigens in the placenta is the trophoblast); and ( 5 ) , as evidenced by the findings of immunodifision and other studies, a mixture of antibodies is involved, indicating that kidney and placenta share more than one antigen in common. Attempts have been made to detect antibodies to placental tissue during pregnancy in humans. A circulating antibody to a placental
46
ALAN E. BEER AND R. E. BILLINCHAM
polysaccharide was described by Kaku (1953). Postulating that, if such antibodies are formed, they might easily be absorbed by trophoblastic tissue in the placenta, Hulka and his associates (1963; Hulka and Brinton, 1963) tested postpartum human serum in both normal and toxemic pregnancies and obtained evidence of specific binding of fluorescein-tagged maternal globulin by the syncytiotrophoblast. This finding was interpreted as indicative of the presence of antitrophoblast antibodies.
C. EXISTENCE OF A “PRIVATE,” TISSUE-SPECIFIC ANTIGENIN TROPHOBLAST Particularly forceful evidence of the abortifacient properties of antitrophoblast serum has been presented by Koren et al. (1968a). They showed that antiserum raised in rabbits against homogenates of whole mouse placentas or relatively pure preparations of trophoblast cells incorporated in Freund’s complete adjuvant and injected intravenously into pregnant mice on three consecutive occasions at a dosage level of 0.25 to 0.5 ml., consistently interrupted pregnancy. Histopathological studies revealed focal areas of hemorrhage and necrosis in the placentas, as well as significant lesions in their livers and kidneys, suggestive of an immunologically induced glomerular nephritis in the latter organ. Subsequent work, using an indirect fluorescent antibody technique to determine the fate of the passively transferred antitrophoblast antibody in duo, indicated a high degree of localization in trophoblast cells, but “sections of liver, kidney, and spleen were negative throughout the fluorescent staining procedure” (Koren et al., 1968a). However, in view of the complexity of the antigenic material used to raise the antisera and the fact that no absorptions were performed, the findings provided only a faint hint that a specific antitrophoblast antibody (not present in anti-kidney serum) was responsible for aborting the mice. Beer et al. (1971b) in an attempt to dissect further the story of organspecific antigens of the placenta, prepared antiserum by inoculating adult male rabbits with intact, viable trophoblast cells from an isogenic strain of rats incorporated with complete Freund’s adjuvant. This heterologous antiserum aborted pregnant rats of any strain at any stage of postimplantation pregnancy, but was completely ineffective when given to mice or hamsters during pregnancy. In contrast, rabbit antisera similarly prepared to rat fetal tissue, lymphoid, or epidermal cells had no deleterious effects on pregnancy in the rat. The activity of the antitrophoblast serum could be selectively removed by absorption with viable trophoblast but not by absorption
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
47
with other cell types, indicating that the trophoblast appeared to possess tissue-specific antigens and that antisera directed against these antigens could terminate gestation at any stage. Other evidence indicative of the possible existence of specific antitrophoblast antibodies in heterologous antitrophoblast serum capable of damaging trophoblast cells in vivo was Bagshawe’s (1970) observation that high dilutions of rabbit antisera against highly purified human chorionic gonadotropin were rapidly lethal to choriocarcinoma and normal trophoblast cells in uitro. Additional findings bearing upon the possible specific antigenicity of placental tissue have come from recent heterotopic transplantation studies carried out on ectoplacental cones and blastocysts, respectively, in mice. Hulka and Mohr (1968) compared the results of transplanting primary and secondary grafts of ectoplacental cone tissue, prepared from 7%-dayC3H mouse embryos, beneath the renal capsules of male C57 BL/6 host mice. Their observations indicated that prior exposure of the host to a single intrarenal homograft of trophoblastic tissue resulted in a lower proportion of takes and inhibition of growth when secondary challenge grafts of similar tissue were transplanted to the same hosts. Lymphocytic infiltration was observed in association with both primary and secondary grafts, being slightly more intense in the case of the latter. The authors interpreted their findings as indicative of sensitization of the hosts to trophoblast antigens, though the data shed no light on the nature of these antigens. Subsequently, Kirby (1968b) reported that repeated transfer of quartets of C3H blastocysts first beneath the renal capsules and later into the testes of C57 BL/6 male hosts resulted in a progressive diminution and, in some cases, the total inhibition of the ability of the blastocystic grafts to develop as evidenced by decreased size of the hemorrhagic swellings (due to trophoblastic invasions) at the implantation sites. Making the reasonable assumption that an immunological phenomenon was involved, conclusive evidence that the antigens responsible were not determined by histocompatibility genes but were very probably tissue-specific was provided by the demonstration that a single isogenic blastocyst transplanted to the testis of each of a group of C57 males, which had received four previous sets of C3H blastocysts, grew with significantly diminished vigor. Similar findings were obtained when C57 blastocysts were transplanted to C3H male hosts. In contrast to the suppression of trophoblastic activity caused by repeated exposure of host mice to homologous blastocysts, when they were finally challenged with skin homografts of the same alien genetic origin, the hosts’ capacity to reject them, was found to have been
48
ALAN E. BEER AND R. E. BILLINGHAM
weakened as evidenced by prolongation of graft survival. Skin homografts from an unrelated donor strain were rejected with normal promptitude, These observations indicated the capacity of homologous blastocysts in extrauterine sites to elicit a state of immunological unresponsiveness-probably enhancement ( see Section X,E ) . Kirby also noted a progressive diminution of the size of the hemorrhagic lesions at the implantation sites when C3H blastocysts were transplanted to isogenic, i.e., C3H hosts-an observation lending considerable direct support to the view that tissue-specific antigens are associated with trophoblast. However, in appraising the significance of both Hulka’s and Kirby’s interesting findings we must not overlook the possibility that the organ-specific antigens they appear to have demonstrated may only be expressed by trophoblast at an early stage in its development, i.e., they may not be demonstrable in the mature placenta. Furthermore, to define the possible biological significance of this phenomenon one additional experiment needs to be done, i.e., to study the development of consecutive ectopic blastocyst grafts in multiparous mice-. Animals that have been exposed to trophoblastic antigens as a consequence of normal pregnancies.
D. PARTURITIONAS
IMMUNOLOGICALLY MEDIATEDPROCESS Several authorities have considered the superficially intriguing possibiIity that separation of the placenta at parturition may be an immunological event. Thomas ( 1959) tentatively suggested that degenerative changes occurring in this organ as gestation proceeds may be caused by sensitization of the mother to a special organ-specific antigen, which may occur rather late in the maturation of this tissue. However, as Thomas pointed out, the consistent occurrence of parturition in inbred strains of rodents precludes the possible involvement of transplantation antigens in such a mechanism. Two obvious implications of this premise are that ( a ) successive pregnancies should be of somewhat shorter duration than the initial one in normal individuals because of maternal sensitization, and ( b ) at least initial pregnancies should be of abnormally long duration in patients suffering from immunological deficiency diseases or under chronic immunosuppressive therapy, such as recipients of renal homografts. So far as we are aware, there is no clinical evidence to sustain these predictions. Tyler (1961a) put forward the thesis that parturition is mediated by a graft-versus-host type of reactivity (Billingham, 1968) in which transplantation antigens are involved. Assuming that fetuses do not make all the transplantation antigens corresponding to their genotype during intrauterine life, in conjunction AN
IMMUNOBIOLOCY OF MAMMALIAN REPRODUCTION
49
with the knowledge that fetuses can react against some transplantation antigens before birth, he suggested that even in intrastrain pregnancies it might be possible for fetuses to “reject” their mothers. The remarkable constancy of the gestation period in various species, particularly in laboratory rodents and the normal delivery at “expected term of infants suffering from a variety of immunological deficiency diseases known to preclude or impair their engaging in reactivity against homografts makes this hypothesis virtually untenable.
E. CONCLUSIONS Although refractory to rejection as a homograft, the trophoblast does seem to express transplantation antigens in a form capable of inducing maternal unresponsiveness, and in some species fetal-maternal incompatibility with respect to these antigens leads to increased invasion and growth of this tissue. In addition, the trophoblast has organ-specific antigens which it shares in common with kidney, and good grounds now exist to suspect that it possesses a truly private antigen(s). The latter possibility clearly merits much more attention since, if it proves true, it may afford a more esthetic means of early therapeutic termination of pregnancy than the current, widespread practice of “salting out’’ or “pickling” arid also offer the basis of an effective immunological approach to choriocarcinoina therapy. A possible means of reducing the incidence of this tumor is implicit in Ober’s (1968) “tongue-in-cheek suggestion that restriction of choriocarcinoma to man is attributable to the fact that, unlike females of all other species, human females have abandoned the habit of eating their placentas after giving birth. X. Maternal-Fetal Exchange of Cells
In any graft-host relationship there are three qualitatively different ways in which the host may become “aware” of an alien solid tissue or organ graft and react immunologically (which includes to become tolerant) against it: ( I ) as a result of the apposition and subsequent union of relatively “fixed tissues of the host to those of the graft, (2) through escape of living cells, or ccllular degradation products, from the graft and their passage into blood vessels or lymphatic drainage channels in the host (in this context, contaminating donor leukocytes carried over in the vasculature and tissue spaces of the graft may play a significant role), and finally, ( 3 ) passage of host immunocompetent cells through the vmeulature of the graft and their return to the host via venous or lymphatic routes, providing an opportunity for peripheral sensitization.
50
ALAN E. BEER AND R. E. BILLINGHAM
Essentially similar opportunities occur for pregnant mammals to become immunologically aware of their immunogenetically alien fetuses, especially in the case of species having hemochorial placentas and, conversely, for fetuses to become immunologically cognizant of their mothers. Although in man the uterine endometrium and decidua have a rich lymphatic drainage system, maternal lymphatics have never been shown to penetrate the basal placental plate. This isolation of placenta from the host lymphatic system is, of course, atypical of the situation in nearly all kinds of grafts used in replacement surgery. A. TROPHOBLAST CELLS In some species, notably humans, trophoblastic giant cells occur in relatively large numbers in both the decidua basalis and myometrium both during and after pregnancy, and there is also evidence of migration of trophoblast cells into and along the linings of uterine arteries where they may persist into the postpartum period. These are normal, rather than abnormal, events, although their functional significance is not understood. However, it can be stated that there is no histological evidence of host cellular reactivity against these ectopic homografts of trophoblast. In normal human pregnancy, one of the most important sources of exposure of the mother to fetal cells is the chronic shedding of multicellular fragments or sprouts of syncytiotrophoblast from placental villi into the maternal venous system at a rate of 100,000 per day, from about the twenty-sixth day of gestation onward (Ilk&,1961, 1964). The majority of these trophoblastic elements, known to be highly susceptible to proteolytic enzymes in uitro, are probably destroyed enzymatically in the bloodstream. Nevertheless, the “survivors” are filtered out in the capillary bed of the lungs where they gradually disappear unaccompanied by any kind of demonstrable local host response-inflammatory or otherwise. The apparent inability of these ectopic trophoblastic grafts to proliferate and form benign metastases probably reflects their highly differentiated, end-cell status. Whether this normal physiological process of fetal-maternal deportation of trophoblastic elements has any functional significance has long been a subject of speculation. Its possible immunological significance will be discussed below. Deportation of trophoblast in the reverse direction was also demonstrated. Ilk6 ( 1961 ) identified trophoblast cells, histologically, in the umbilical veins of fetuses at various stages of gestation, and Salvaggio el al. (1960) found them in the cord blood obtained from fetuses at delivery.
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
51
B. BLOODCELLS
It is now generally recognized that the normal placenta is not a perfect impermeable barrier to the cellular elements of the blood; some covert transplacental exchange of these cells is a common, if not a normal physiological event in man, and probably in certain other mammalian species also.
I, E ythrocytes Since fetal-maternal passage of red cells was first postulated as the cause of Rh sensitization (Darrow, 1938; Levine and Stetson, 1939), many independent investigations, using different markers, have confirmed that fetal erythrocytes gain access to the maternal circulation. Fetal red cells are detectable in the maternal bloodstream as early as the eighth week of gestation (Zipursky et al., 1963), and the frequency of their exchange increases with the length of gestation. For example, Cohen et al. (1964; Cohen and Zuelzer, 1967) reported an incidence of 6.7%during the first, 15.8%in the second, and 28.9%in the third trimester. The last figure was consonant with Beer’s (1969) demonstration of fetal cells in the maternal circulation of more than 30%of Rh negative gravidas prior to the onset of labor. These findings suggest either that a progressive diminution in the integrity of the placental barrier develops as gestation progresses or that the incidence of leakages, probably due to small-scale hemorrhages at the level of the trophoblastic villi, is a function of the aggregate area of maternalfetal interface. Labor and delivery of the fetus and placenta further increase the incidence of fetal red cells in the maternal circulation to more than 50% of cases. Certain obstetrical procedures and manipulations bring this figure above 85%and increase the “dosage” of the fetal red cells transferred as well (Beer, 1969). Numerous estimates have been made of the amount of blood received by the mother from her fetus, using the “acid elution” technique for detecting fetal cells in smears of maternal blood (Kleihauer et al., 1957; Betke and Kleihauer, 1958). Most workers agree that 0.1 ml. of Rh-incompatible blood is an effective sensitizing dose, and corroborative evidence has been obtained in about 60%of all Rh-negative women having demonstrable fetal red cells in their blood after delivery. It has been estimated that 36%of postpartum patients whose blood smears were positive for fetal cells must have received more than 4.0 nil. of fetal blood, and 4%received at least 40 ml. (Beer, 1969).
52
ALAN E. BEER AND R. E. BILLINGHAM
Likewise, evidence is accumulating that passage of maternal erythrocytes into the fetal circulation is probably a normal physiological event in humans. Indeed, massive amounts of maternal blood may sometimes be transferred, as in cases of neonatal plethora. However, there is a dearth of factual data concerning the incidence and extent of cell traffic in this direction (Smith et al., 1961; Macris et al., 1958) which is not favored by the pressure differential.
2. Leukocytes
It is difficult to conceive how an exchange of red cells could take place without accompanying leukocytes and platelets, and confirmatory evidence has been forthcoming. Particularly cogent was the cytogenetic finding of Walknowska et al. (19SS) of forty-six XY cells in the circulation of mothers who subsequently gave birth to male fetuses (see also Turner et al., 1966). Recently Tuffrey et a,?. (1969a) presented cytogenetic evidence indicative of a considerable maternal-fetal leukocytic traffic in mice. They mated female CBA strain mice, homozygous for the TG chromosome marker with CBA (TOT,) males. Then, 2% days later, fertilized eggs from an unrelated strain of CFW mice lacking the To marker were transferred to their uterine horns. The resulting litters, comprising mice of CBA and CFW genotypes (distinguishable by their coat colors) were killed 40-60 days postpartum and various tissues including bone marrow and lymphoid tissues examined cytologically. According to these authors, 330% of alien (TGTG) labeled cells were present among the dividing cells examined, indicative of a fairly high level of chimerism with respect to components of the lymphohematopoietic tissue system in some of the subjects. Although Tuffrey et al. conceded that these cells might have been acquired from their CBA siblings in utero, they favored a maternal origin. In a second study (Tuffrey et al., 1969b), utilizing the same cytogenetic marker, and an experimental design that excluded the possibility of complication by exchange of tagged cells between fetuses, chimeras due to transplacental passage of maternal cells proved to be very infrequent (Billington et al., 1969). Careful attempts to confirm these results by other workers have so far failed. However, this does not exclude the possibility that, for genetic reasons, CFW mice gestated in CBA mothers may have been peculiarly susceptible to transgression of their trophoblastic frontiers by maternal leukocytes. Indeed, there is highly suggestive evidence that under some circumstances this actually happens in rats (see p. 65). In rabbits, Oehme et al. ( 1966) have presented evidence .of a maternal 9 fetal transmission of radioactively labeled leukocytes.
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
53
C. MALIGNANT CELLS Instances of transmission of malignant disease from affected women to their fetuses are exceedingly rare (Potter and Schoeneman, 1970). This is particularly surprising in the case of hematologic malignancies. In reviewing possible transmission of leukemia and allied diseases from mother to fetus, Diamondopoulos and Hertig (1963) found that in approximately 400 fetuses at risk transmission might have occurred in only 2 and possibly in 4 more. Even here, it was conceivable that the tumors in the progeny were initiated by an oncogenic agent rather than by a “cellular” graft or that by rare coincidence both mother and offspring were afflicted independently by the same type of tumor. Because malignant melanomas occasionally metastasize to the placenta, as in at least 6 well-documented cases, after crossing the placental barrier they may also give rise to widespread metastases in the child’s tissues, usually causing death within a few months of delivery though one complete regression is on record (Benirschke and Driscoll, 1967). From these clinical observations it is impossible to determine whether the infrequent “success” of these tumor homografts of maternal origin can be ascribed to induction of tolerance, enhancement, or simply to chance genetic compatibility. Evidence that human fetuses acquire immunological competence to react against some antigens long before birth (see p. 14) affords a possible partial explanation for the extreme rarity of maternal + fetal transmission of tumors, though none of these considerations provides a satisfactory explanation for the unequivpcal fact that melanohas are transmitted to fetuses more frequently than any other kind of tumor.
D. CONSEQUENCES OF MATERNAL EXPOSURE TO FETAL ANTIGENS 1 . Tolerance Induction in Adults Preliminary to consideration of the consequences of accidental leakage of fetal cells into the mother, it must be recognized that to become tolerant on inoculation with homologous cells is not a quality restricted to very young animals. Tolerance of tissue homografts is inducible in adults, though it usually requires their inoculation with massive dosages of antigen over prolonged periods. However, as with infant hosts, the magnitude of the individual inoculum and duration of the exposure period required to induce tolerance in adult, immunologically mature subjects depends upon the degree of genetic disparity-weak histoincompatibilities are much easier to overcome than strong ones (see Billingham and Silvers, 1971).
54
ALAN E. BEER AND R. E. BILLINGHAM
The induction of unresponsiveness in adult animals can be facilitated by treatment with a number of nonspecific agents or procedures, including immunosuppressive drugs, depletion of immunocompetent cells by irradiation, thoracic duct drainage, or treatment with heterologous antilymphocyte serum. Indeed the use of such agents, in conjunction with administration of antigen, results in tolerance under conditions in which the antigen alone would incite an immune response (or sensitization). In addition to the methods of tolerance induction in adult animals described above, a similar state of specific unresponsiveness can sometimes be induced in adult mice simply by prior exposure to homografts of ovary, testis, or neonatal skin and, in hamsters, by exposure to cheek pouch or neonatal skin, both of which are grossly inferior to ordinary skin in an immunogenic sense. The capacity of these tissues to weaken host reactivity is usually incomplete, i.e., they only extend the life-span of subsequent test grafts of adult skin of the same genetic origin but rarely prolong it indefinitely. This principle only applies in situations in which relatively minor histoincompatibilites prevail. As these procedures to induce unresponsiveness in adult hosts do not appear to involve systemic exposure to large amounts of antigen, chronic exposure to small dosages of antigen also may render a host unresponsive. In an important quantitative study of the induction of tolerance to bovine serum albumin in mice, Mitchison (1965) found that unresponsiveness could be induced by administration of high cumulative doses of aptigen ( > 5 mg.) or by very small dosages ( 10-40 pg.), i.e., high and low-zone tolerance. Billingham and Sparrow’s (1955) finding that highly significant prolongation of survival of skin homografts could be obtained in rabbits by prior intravenous injection of the host with relatively small numbers (10-20 X lo6) of viable dissociated epidermal cells or leukocytes might be an example of low-dosage tolerance induction. 2. Red Cell Antigens: Hemolytic Disease of the Newborn in Man According to Race and Sanger (1968), more than 30 inherited blood group antigens can incite maternal isoimmunization in man, though there is considerable intrinsic variation in the immunogenicity of these antigens. Their immunogenicity is influenced secondarily by compatibility or otherwise of the maternal environment with respect to the ABO blood group system, because of the presence of “natural” antibodies in the serum of individuals corresponding to those ABO group antigens which they lack. This situation is also responsible for the occasional
IMMUNOEIOLOGY OF MAMMALIAN REPRODUCTION
55
occurrence of hemolytic disease in the progeny of group 0 mothers bearing group A or B fetuses who, during gestation, are stimulated to produce immune anti-A or anti-B in addition to their naturally acquired quota of these antibodies. Hemolytic disease having this etiology affects only 1in 5 of the subjects at risk. Nearly 9!3%of all cases of hemolytic disease of the newborn that are not owing to ABO blood group incompatibility are caused by incompatibility with respect to the “D” antigen of the Rh blood group system. Unlike the ABO determinants which are present on many cell types, including leukocytes and epidermal cells, the Rh antigens are represented only on the cell membranes of erythrocytes. There have been many ingenious attempts to explain the observed variation in response of Rh-negative persons exposed to Rh antigens. Mitchison, Brambell, and Owen independently, suggested that nonreactivity to Rh antigens might arise from the gestation of Rh-negative fetuses in Rh-positive mothers, affording the offspring an opportunity to become tolerant of the antigens concerned (see Billingham et al., 1956). Thus an Rh-negative female child gestated in an Rh-positive mother should have an impaired capacity to form anti-D antibodies when in adult life she bears an Rh-positive fetus, in comparison with an Rh-negative female gestated by an Rh-negative mother. This ingenious hypothesis has languished for want of factual evidence in its support. Indeed, there is evidence that very rarely an Rh-negative female fetus may receive her first sensitizing dose of Rh-positive cells from her mother in utero. Such individuals comprise less than 0.5% of all Rhsensitized patients. A nonspecific weakening of the faculty of immunological unresponsiveness during pregnancy has also been invoked to help explain variation in response of Rh-negative persons to Rh antigens, but, again, it is devoid of factual support of any consequence. The most plausible explanation for the very low frequency of sensitization (0.5%)during the first pregnancy at risk turns upon the behavior of the Rh-incompatible but ABO-compatible erythrocytes in the maternal environment. The life-span of such cells is long-of the order of 100 to 200 days. They seem to be well tolerated and do not become immunogenic until they near the end of their life-span and are “tagged” for clearance by the maternal spleen where the foreign Rh antigens are recognized in the host’s lymphoid centers (Cohen and Zuelzer, 1967). The risk of subsequent sensitization of Rh-negative mothers bearing Rh-positive fetuses is now known to depend upon ( a ) the size of the transplacental hemorrhage of Rh-incompatible blood, the magnitude of the fetal-maternal bleed determining the incidence of isoimmuniza-
56
ALAN E. BEER AND R. E. BILLINGHAM
tion and ( b ) the specific Rh genotype of the fetus (Masouredis et al., 1960; Masouredis, 1962; Rochna and Hughes-Jones, 1965). There are approximately 12,000 D antigenic determinant sites on a D-positive erythrocyte. With the aid of 1311-labeledanti-D it has been confirmed that cDE cells react more strongly than CDe cells and that cDE/cDE cells agglutinate more strongly than any other cells with respect to the Rh system. The presence of C in the genome decreases the number of D antigen sites available for binding with the labeled globulin and probably reduces the immunogenicity of such a cell. Also, ABO incompatibility between mother and fetus affords a natural protective mechanism against the risk of Rh sensitization ( Nevanlinna and Vainio, 1956). For example, if a group 0, Rh-negative mother is pregnant with a group A, Rh-positive child, fetal erythrocytes gaining access to her circulation are lysed or opsonized by the naturally occurring anti-A antibodies in her serum, causing their removal in the liver by processes that are unlikely to lead to sensitization. A recent study of Woodrow and Donohoe (1968) indicated that anti-D antibodies were 8 times more likely to appear in the postnatal period following first pregnancies which were ABO compatible than when the first pregnancy was ABO incompatible. The discovery of this natural protection against the risk of hemolytic disease of the newborn stimulated the intensive investigations that eventually led to the administration of Rh immunoglobulin as a successful means of preventing primary Rh isoimmunization in patients at risk (see below).
3. Rh Prophyluxis-Circumvention of Maternal Sensitization In 1960 R. Finn, upon analysis of the protective role of ABO incompatibility against Rh sensitization, ingeniously suggested that similar protection might be afforded artificially by passive immunization of subjects at risk with anti-D antibody. His experimental studies established the feasibility of this idea. Quite independently, Gorman et al. (1964), exploring the application of the immunological principle that injection of antigen together with excess antibody failed to sensitize the host, established that excess anti-D antibody administered to Rh-negative male volunteers after a normally effective sensitizing dose of Rh-positive erythrocytes prevented primary immunization. Pertinent to this line of inquiry was the inability of Stern et al. (1956) to immunize group 0, Rh-negative male volunteers with group 0, Rhpositive cells coated in uitro with Rh antibodies, and the previous demonstration by Jandl et al. (1957), who administered red cells coated with anti-D antibody that were quickly cleared from the host’s circulation. Each of these independent studies played its part in setting the stage
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
57
for full-scale clinical tials in Rh-negative gravidas at many centers in the U. S. A. and abroad. Of over 2500 treated patients followed for a 6-month postinjection period, only %less than 1% 0.gav-e evidence of becoming sensitized to the Rh factor. The ultimate and most revealing evidence of the efficacy of anti-Rh antibody prophylaxis has been its ability to protect Rh-negative mothers through multiple Rhincompatible pregnancies.
E. CONSEQUENCES OF MATERNAL EXPOSURE TO FETAL AND TRANSPLANTATION ANTIGENS LEUKOCYTE The independent demonstration, by Payne and Rolfs (1958) and by Van Rood et al. (1959) that, in man, leukocyte antibodies are present in maternal serum following at least two pregnancies affords additional evidence of the occurrence and incidence of fetal-maternal transmission of leukocytes during gestation, since the antigens concerned are not present on erythrocytes nor, in all probability, are they present on trophoblast cells in an effective form (Seigler and Metzgar, 1970). Since multiparous women can only form antibodies against the leukocyte antigens transmitted to their fetuses from their husbands and absent in themselves, their antibodies are necessarily of limited specificity, sometimes capable of recognizing a single antigen. Such individuals are an invaluable source of sera for histocompatibility testing since most of the serologically detectable antigenic determinants referred to as “leukocyte antigens” are in fact histocompatibility antigens (Amos, 1969). These isoantibodies persist at relatively high titers in the sera of multiparous women for many years after their last pregnancy. In a recent retrospective study, Terasaki et al. (1970) have obtained data which suggest that women with HL-A antibodies have a significantly higher incidence of infants with congenital anomalies than those without these antibodies. The authors postulate that antibodies produced by mothers incompatible with the HL-A antigens of their fetuses may have an adverse influence on fetuses in subsequent pregnancies. In mice, too, it is well documented that antibodies, detected as isohemagglutinins and also corresponding to histocompatibility determinants inherited by fetuses from their fathers, appear in the sera of multiparous females ( Herzenberg and Gonzales, 1962; Goodlin and Herzenberg, 1964; Kaliss and Dagg, 1964). The fact that repeated matings of sterile females with unrelated males fail to incite the formation of these antibodies indicates that the fetuses are the source of the antigenic stimulus. The evidence reviewed so far indicates that ( I ) cells from fetuses do normally gain access to the maternal circulation and probably to the regional lymph nodes draining the uterus since these enlarge during
58
ALAN E. BEER AND R. E. BILLINGHAM
heterospecific pregnancy (Beer et al., 1971a); and ( 2 ) the mother is immunologically aware of and is, indeed, stimulated by the cellular antigens of her fetus. Long before maternal hemagglutinin responses to fetal transplantation antigens were discovered, grafting tests employing tumor or skin homografts from either the offspring, the father, or a member of the paternal inbred strain were performed to determine whether heterospecific pregnancies could alter the mother’s reactivity to the tissue antigens concerned. In cattle (Billingham and Lampkin, 1957), but not in sheep (Galton, 1965) suggestive evidence has been obtained that pregnancy sometimes sensitizes the dam to subsequent grafts from her offspring. In women there is some rather equivocal evidence that habitual spontaneous abortion may be associated with the development of sensitivity to grafts of their husbands’ skin (Bardawil et al., 1962). 1 . Parity-Induced Specific Weakening of Homograft Reactivity However, in mice exhaustive investigations have produced no evidence that repeated heterospecific pregnancies can curtail the survival of homografts from the paternal strain, with the exception of leukemic tumor grafts (see below). On the contrary, exactly the opposite may occur, leading to a long-lasting, specific weakening of a female’s capacity to reject paternal strain grafts (Breyere and Barrett, 1960a,b, 1962; Breyere, 1967). Tumor graft challenges are more impressive than those of skin grafts as the former are able to override weak degrees of immunological opposition. The extent of this parity-induced tolerance or unresponsiveness has been shown to be dependent upon ( a ) the degree of genetic disparity between the parents, most impressive when only nonmajor locus histocompatibility factors are involved, and ( b ) the parity of the female (Kaliss and Dagg, 1964). Host reactivity decreases up to a point with parity, but only when weak antigens are involved, such as the H-Y factor in C57 BL/6 mice, does it ever become complete (Billingham et al., 196%). According to the work of Breyere and Burhoe ( 1963) incompleteness of parity-induced tolerance probably reflects complete and permanent tolerance to some paternal strain antigens and unchanged or only partially suppressed reactivity to othersprobably the more important or stronger antigenic determinants involved (this is reminiscent of the well-known phenomenon of split tobra w e (Lustgraff et al., 1960). Evidence that, at least with one strain combination, the antigens responsible for maternal tolerance originate from the fetus before parturition has been presented by Porter and Breyere (1964). First, they showed that heterospecific matings of BALB/c females, whose uterine
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
59
horns had been ligated or ovaries had been transplanted subcutaneously, with DBA/2 males failed to weaken their reactivity to DBA/2 challenge grafts. Second, despite excision near term of the gravid uterine horns from BALB/c females pregnant by DBA/2 males, evidence of weakened reactivity was still obtainable. On the basis of their own observations and those of other workers that multiparity induced by males of an unrelated strain leads to both isoantibody formation and abrogation of reactivity to homografts of skin and sarcomas (but heightened reactivity to leukemic test grafts, when male and female differed at the H-2 locus), Kaliss and Dagg ( 1964) suggested that immunological enhancement-a highly specific “frustration of both the antigenic stimulus and the host’s cellular immune response by humoral antibody” (Kaliss and Rubinstein, 1968a; Snell, 1970)-rather than immunological tolerance, might be the phenomenon underlying this parity-induced weakening of homograft reactivity. The heightened reactivity of the multiparous mice to leukemic test grafts was a consequence of the vulnerability of leukemic cells and, indeed, normal cells of the lymphohematopoietic tissue system to the complement-dependent cytotoxic action of humoral antibodies. Conceded shortcomings of this enhancement theory were, first, inability to produce enhancement in normal mice by transfer of serum from multiparous donors and, second, absence of any correlation between the presence or absence of hemagglutinins in the multiparous females and their reactivity to sarcoma test homografts. Kaliss and Dagg (1964) felt that the latter might not be important since the antibodies responsible for enhancing activity might differ qualitatively from those responsible for the hemagglutinating activity. Subsequent work has done much to strengthen the attraction of this theory. It has been shown (Goodlin and Herzenberg, 1964; Kaliss and Rubinstein, 1968b) that there were marked cyclical undulations in the titers of isoagglutinins induced by multiparity which were not associated with the stage of pregnancy and which occurred in females that were nonpregnant at the time of testing. Furthermore, there was no evidence of any “anamnestic response” following a successive heterospecific pregnancy. Rubinstein and Kaliss’ (1964) evidence that pregnancy-induced hemagglutinins had a very short half-life after passive transfer is relevant here. In a well-controlled study with A,G female mice mated with CBA males, and using cells from a chemically induced CBA sarcoma as test grafts to detect altered immunological reactivity, Currie ( 1970) has demonstrated a feeble though definite degree of specific impairment of reactivity to paternal tissue antigens early in a first inter-
60
ALAN
E. BEER AND R. E. BILLINGHAM
strain pregnancy. More important is his observation that the reactivity of virgin A,G females to grafts of the tumor is significantly impaired by repeated injections of serum from A2G females multiparous by CBA males. From these findings he concludes that immunological enhancement is the mechanism responsible for the pregnancy-induced specific unresponsiveness to paternal antigens. However, the fact that the action of enhancing antibodies is almost certainly complex, having both afferent and efferent effects, in addition to a central inhibitory action on the immunological response machinery, must not be overlooked. It suggests that enhancing antibodies may also be able-to facilitate the induction of tolerance in a manner analogous to that of immunosuppressive drugs (see p. 31; Voisin and Kinsky, 1962).
2. Enhancement As a Possible Ancillary Protective Mechanism for the Fetus As a Homograft Recently I. Hellstrom and Hellstrom (1969) discovered that lymphocytes from tumor-bearing animals can inhibit the growth in vitro of neoplastic cells from the same animals and that sera from tumor-bearing animals frequently contain antibodies capable of binding specifically to the tumor cells, rendering them refractory to the influence of the lymphocytes. With the aid of their colony inhibition test, these investigators (K. E. Hellstrom et al., 1969) have shown that lymph node cells from BALB/c mice bearing antigenically alien (BALB/c x C3H)F, fetuses are capable of inhibiting the growth of C3H fetal target cells in vitro and that sera from the same pregnant mice can abrogate this inhibitory effect. If the specific protective factor in the serum proves to be an immunoglobulin, this phenomenon must almost certainly be a manifestation of the principle of immunological enhancement. On the basis of these in vitro findings, the authors suggest that a specific factor, possibly antibody, may be capable of conferring protection upon antigenically alien fetuses from the destructive effect of specifically sensitized maternal lymphocytes that may have penetrated the placenta. This “blocking” antibody, or enhancementmediated protective mechanism, may supplement or reinforce that afforded by the trophoblast-associated sialomucin material. The work described above admits of the possibility that, in females multiparous by unrelated males, there may be a state of cell-mediated transplantation immunity rendered incapable of expressing itself because of the concomitant presence of enhancing or “blocking” antibodies. Such a mechanism of homograft protection is not unique, for in rats it has
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
61
been shown that actively or passively acquired specific enhancing isoantibodies enable renal homografts to override Ag-B locus histoincompatibility barriers ( Winn, 1970; Billingham and Barker, 1969). Although there is suggestive evidence that a similar phenomenon applies in the rabbit, in man and dogs such antibodies appear to cause hyperacute rejections. S o r h ( 1967) has produced indirect evidence of an antibody-suppressed cellular immunity in multiparous mice after heterospecih matings. Lymphoid cells from females outcrossed to males of an H-2 locus incompatible strain (CBA or C57 females mated with A males) injected into neonatal progeny of such matings induced significantly more intense graft-versus-host reactions ( as evidenced by greater spleen indices; Simonsen, 1962) than similar numbers of lymphoid cells from virgin female donors of the same strain. A test which should be capable of shedding some light on the influence of parity, abortions, and trophoblastic tumors on the reactivity of women toward their husbands’ transplantation antigens determined by the HL-A locus is the so-called mixed leukocyte culture test (Wilson and Billingham, 1967; Bach, 1968). It is based on the fact that when immunocompetent cells (lymphocytes in peripheral blood leukocyte preparations for convenience) are exposed in vitro to similar cells from an HL-incompatible individual, some of them transform into large, basophilic, blastlike cells that take up tritiated thymidine and divide. Analysis has revealed that this is the outcome of A’s immunocompetent cells reacting against B’s alien HL-A antigens and vice versa. Although as described, the test is two-directional, it can be made unidirectional by prior incubation of the cells from one codonor with mitomycin C, which inhibits their capacity to proliferate while conserving their antigenic status. To date there has been one report suggestive of impaired reactivity of pregnant womens’ cells to the cellular antigens of their husbands, which was heightened by multiparity (Lewis et al., 1966). This lack of responsiveness appeared to be specific in that it was not demonstrable when pregnant women’s leukocytes were mixed with cells from unrelated males. Furthermore, there was no evidence of altered reactivity on the part of cells from women with tumors of placental origin. Halbrecht and Komlos (1968) claimed that there was an increase in the percentage of transformed leukocytes in mixed husband-wife leukocyte cultures in cases of abortion and hydatidiform moles. Unfortunately in neither of these studies were one-way reactions studied, which makes the findings difEcult to interpret.
62
ALAN E. BEER AND R. E. BILLINGHAM
F.
CONSEQUENCES
OF
FETALEXPOSURE TO MATERNAL CELLS
1. Maternally Induced Tolerance Ever since the principle of immunological tolerance was first worked out with tissue transplantation antigen systems in experimental animals, the possibility has been entertained that offspring might “naturally” become tolerant of their mother’s tissue antigens as a consequence of exposure of fetuses to maternal cells (leukocytes are the most favored candidates). Various studies have been performed to explore this intriguing possibility in mice, rabbits, guinea pigs, cattle, sheep, and even man (Billingham et al., 1956; Billingham and Lampkin, 1957; Galton, 1965; Peer et al., 1960). Where outbred animals had to be used, survival times of paternal and maternal skin homografts on their progeny were compared, but when inbred strains were available experiments of a more sophisticated design and capable of revealing feebler degrees of maternally induced tolerance were employed. Because of its long gestation period (about 60 days) and hemochorial type of placenta, the guinea pig seemed a favorable subject for investigation of this phenomenon. Billingham and Silvers ( 1965) reciprocally backcrossed F, hybrid progeny of isogenic strains Nos. 2 and 13 to strain No. 13 to produce two similar, genetically defined but heterogeneous populations of animals which differed only insofar as one group of animals had developed in an F, hybrid milieu, affording them an opportunity to incorporate F, cells bearing strain No. 2 antigens prenatally. To test for any alteration of immunological reactivity this might have caused, both groups of F, progeny were challenged, when 30 days old, with skin homografts from strain No. 2 donors. The survival time distributions of these two series of grafts were closely similar, indicating that maternally induced tolerance in guinea pigs must be a rare phenomenon if it occurs at all. Billingham et al. (1956) had previously obtained similar results in an application of this experimental approach to mice of strains A and CBA. However, using tumors as test grafts, Sanford ( 1963) obtained evidence of increased susceptibility of backcross progeny gestated in F, hybrid female mice, suggestive of maternally induced tolerance. Also pertinent is E. C. Jones and Krohn’s (1962) finding that when tolerant A-strain female mice, with ovaries replaced by functional grafts from CBA donors, were mated with CBA males, their CBA progeny gave no evidence of being tolerant of A-strain test skin grafts, despite gestation in an A-strain milieu. Billingham et al. (1965a) carried out an essentially similar experiment using rats of the Ag-B locus incompatible Lewis and BN strains. By mating (BN X Lewis)F, hybrid females,
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
63
with ovaries replaced by those from Lewis females, to Lewis males, Lewis fetuses were caused to be gestated in an F, hybrid milieu, thus exposing them to BN antigens. When 3 days old the offspring were challenged with BN skin homografts, which revealed weak, transient degrees of tolerance that would escape detection by test grafting when the subjects were grown. None of the animals gave evidence of being tolerant. Indeed, some rejected their test grafts more rapidly than Lewis infants gestated by Lewis mothers, suggesting that they had been specifically presensitized. This might have resulted from exposure to a very small dosage of F, cells at the time of parturition. Dkmant and associates (1966; Dkmant, 1968) have reported a slight but significant prolongation of survival of maternal compared with paternal skin grafts transplanted to the newborn progeny of rabbits that were fairly closely related genetically. However, when test grafting was delayed by as little as 3 days after birth the difference was only marginal. When both parents belonged to different “lines,” so that the genetic disparities involved were greater, no prolongation of maternal graft survival was demonstrable. Indeed, the findings suggested that some of the offspring might have been sensitized to maternal antigens. Neither in cattle nor in sheep was any evidence of maternally induced tolerance obtained, and, in man, Peer and his associates’ (1960) claim of children sometimes accepting skin homografts from their mothers for much longer periods than from their fathers has never received independent confirmation. Furthermore, there is no evidence that renal homografts in children from their mothers are more successful than from their fathers. The apparent rarity of immunological indications of the passage of cells from mothers into their fetuses is probably in part the consequence of an unfavorable blood pressure gradient, which may, therefore, be regarded as an ancillary protective mechanism.
2. Attempts to Increase Permeability of Placental Barrier to Cells Attempts have been made to increase the rate of transplacental cellular traffic between mother and fetus following the pioneer work of LengerovB ( 1957). This investigator irradiated the exteriorized gravid uterine horns of outbred rats on the fifteenth day of pregnancy, shielding the body of the mother. When the 4-week-old progeny were challenged with grafts of maternal skin the majority accepted these for more than 200 days, whereas grafts from mothers to nonirradiated offspring, or from unrelated females to previously irradiated hosts, were all destroyed within 30 days. This finding was attributed to irradiationfacilitated maternal induction of tolerance.
64
ALAN E. BEER AND R. E. BILLINGHAM
Subsequently, Ramseier and Brent (1966) locally irradiated the placentas of (DA x Lewis)F, hybrid fetal rats of 15 days’ gestation in the uteri of (DA x Lewis)F1 mothers previously mated with (DA X Lewis)F, males. When 60 days old, each survivor received a DA and a Lewis skin homograft. Control data were obtained by similarly test grafting F2hybrid rats whose placentas had not been irradiated. The life expectancy of DA grafts on the irradiated F, population was superior to that of similar grafts on the nonirradiated controls, consistent with the premise that local irradiation of the placenta had increased the maternal + fetal cell tr&c, resulting in tolerance in those instances where the genetic disparity between fetus and mother was not too great. However, the experimental results were asymmetrical in that no evidence was obtained of maternally induced tolerance to Lewis test grafts. In mice essentially similar attempts to weaken the placental barrier with X-rays, leading to tolerance of maternal skin grafts (Moulton et al., 1960) or to enhance the passage of Wr-labeled red cells from the maternal to the fetal circulation (Finegold and Michie, 1961) were unsuccessful. Two groups of investigators reported that the administration of hyaluronidase and/or histamine to pregnant female rabbits could partially or even completely abrogate the ability of a significant proportion of their offspring to reject grafts of maternal skin (Nathan et al., 1960; Najarian and Dixon, 1963). Indeed, Najarian and Dixon found that many of their treated does behaved as if they were partially or even completely tolerant of their off springs’ skin and that hyaluronidase treatment doubled the number of maternal erythrocytes that normally crossed the placenta during the last 2 weeks of pregnancy.
3. Apparent Transmission of Homograft Sensitivity from Mother to Fetus In 1965, Stastny presented well-controlled observations that offspring from Sprague-Dawley female rats, mated with males of the same outbred stock and sensitized to Lewis strain skin homografts during pregnancy, behaved as if they were sensitized when challenged with Lewis skin. Likewise, young Sprague-Dawley rats born of mothers which had been ( I ) rendered tolerant of Lewis grafts by neonatal inoculation of Lewis spleen cells and (2) subsequently injected during pregnancy with a large dose of viable Lewis epidermal cells also rejected Lewis skin grafts in an “immune” manner. It is difficult to reconcile these findings with those of Silvers and Billingham (1966). Working with the Ag-Bincompatible BN + Lewis strain combination, these investigators found that if Lewis rats were hyperunmunized against BN tissue cells and
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
65
then mated with Lewis males, there was significant impairment of the capacity of their progeny to react against BN skin grafts transplanted on the third day postpartum. It may be relevant that in Stastny’s work, Ag-B incompatibilities were probably not involved, and his regimen of sensitization and time of test grafting were quite different. However, Beer et al. (1971b) have recently confirmed and extended Stastny’s surprising and rather provocative observations, first with the same rat stocks as used by Stastny and, subsequently, with inbred Fischer rats exposed to Lewis strain tissues and cells. In Beer et aZ.’s analysis of this phenomenon, which they felt turned upon transmission of cells from mother to fetus, virgin Fischer females were mated with Fischer males. Then, at various stages of pregnancy, the females received a single intraperitoneal injection of a suspension of 100 x lo6 viable Lewis strain lymphoid cells. The Fischer progeny were challenged with Lewis skin homografts when 21 days old, The survival times of these grafts indicated that the capacity of many of the hosts to react against Lewis antigens had been significantly weakened as a consequence of inoculation of their mothers with Lewis cells 5-7 days before, or even only a few hours before their birth. Inoculation of the mothers after parturition had no influence on the subsequent immunological reactivity of their suckling young, The finding that in these experiments tolerance, rather than immunity, was usually the outcome, might well reflect the high degree of tolerance responsiveness of infant Fischer rats to Lewis tissue antigens, e.g., as few as 250,000 Lewis bone marrow cells injected intravenously into neonatal Fischer rats induced a high degree of tolerance in 90% of the subjects (Silvers and Billingham, 1969). In an attempt to make use of this tolerance responsiveness of Fischer rats to Lewis tissue antigens to elucidate maternal induction of tolerance, Beer et al. (1971b) carried out the following experiment. Young, virgin female adult Fischer rats were treated with cyclophosphamide and injected with Lewis bone marrow cells to render them highly chimeric with respect to their lymphohematopoietic tissue system and, of course, tolerant of Lewis skin grafts (Santos and Owens, 1968). They were then mated with Fischer males in the hope that sufficient Lewis leukocytes would cross the placentas of their fetuses to induce tolerance of Lewis tissue antigens. Despite the healthy appearance of the chimeric mothers, their reproductive performance was conspicuously subnormal, through frequent spontaneous abortions. Furthermore, of the offspring that seemed perfectly healthy at birth, approximately 50% died of a wasting syndrome by the twenty-fifth day and only about 30%survived in a healthy condition by the thirty-fifth day. All animals that were
66
ALAN E. BEER AND R. E. BILLINGHAM
healthy enough on the twentieth day postpartum were challenged with Lewis skin grafts, and those that survived gave evidence of being tolerant. The unlikely possibility that the cyclophosphamide treatment of their Fischer mothers was in some way responsible for this runting syndrome in their progeny was refuted by the demonstration that Fischer females, treated with cyclophosphamide and rehabilitated with Fischer marrow, gave birth to healthy litters following intrastrain matings. Although these findings leave many questions unanswered, they suggest that the runting syndrome encountered was due to graft-versushost reactivity by some transplacentally derived, Lewis, immunocompetent cells from the chimeric mothers. Essentially, the phenomenon may be the same as that described by Palm in a different experimental context (see Section XIV) , XI. Natural Occurrence of Transplantation Disease
It is well established that in circumstances where donor and host differ with respect to antigens determined by a major histocompatibility locus, such as the H-2 locus of the mouse or the Ag-B locus of the rat, inoculation of immunologically competent cells from adult donors into fetal or perinatal hosts may lead to graft-versus-host reactivity, culminating in overt, systemic homologous or runt disease ( Billingham, 1968). This may be acute and normally fatal or it may be subacute and run a chronic course with the possibility of eventual recovery. Even when donor and host do not differ at a major histocompatibility locus, subclinical levels of graft-versus-host reactivity may occur, which may be responsible for an increased incidence of tumors, especially lymphomas, etc. Soon after it had been shown that normal, adult, peripheral blood contained immunologically competent cells among its small lymphocyte moiety, attention was drawn to the possibility that runt disease might sometimes occur naturally if enough maternal leukocytes gained access to a genetically appropriate fetus. That human infants are susceptible to runt disease is evidenced by the untoward results of some therapeutic attempts to reconstitute immunological function in infants with thymic dysplasia and other immunological deficiency diseases by means of leukocyte or marrow cell homografts (Hathaway et al., 1965). Likewise in the early days of intrauterine transfusion of packed erythrocytes in the treatment of fetuses severely affected by Rh sensitization in the third trimester, when no attempt was made to render the blood free of leukocytes before trans-
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
67
fusion, at least one unequivocal case of fatal runt disease resulted (Naiman et al., 1966). In 1966, Kadowaki et al. (1966) described a male infant, apparently normal at birth who began to suffer from a variety of infections after a month and showed cessation of weight gain after 3 months. The skin was abnormal, displaying peculiar lymphohistiocytic infiltrations, the nodes were enlarged and, especially significant, there was XX/XY cell chimerism, apparently restricted to lymphoid cells in the blood. The child died after 16 months. Graft-versus-host reactivity by cells of maternal origin was held at least partially responsible for this disease, possible facilitated by primary thymic arrest. With appropriate symptoms present, graft-versus-host reactivity has repeatedly been invoked to account for otherwise inexplicable infant deaths. Lymphoid cell chimerism in affected infants must be regarded as an essential requirement, in addition to pathological lesions in the skin, lymphoid organs, etc., before a diagnosis of runt disease of natural (or, indeed, therapeutic) origin can reasonably be made. However, a variety of wasting diseases in man and experimental animals, closely similar to runt disease, are caused by other than graft-versus-host reactivity (Billingham, 1W8). XII. Immunological Competence of the Placenta
In considering the influence of naturally transmitted low dosages of maternal cells on the fetus, it is necessary to take into consideration evidence of the acquisition of competence to react against certain antigens long before birth. In man, when placental quarantine has been broached by a variety of infectious agents, including those of syphilis, toxoplasmosis, and certain viruses, histological and occasionally serological indications have been obtained of immunological reactivity of the fetus during the third and possibly during part of the second trimester ( Silverstein and Lukes, 1962). Fairly extensive studies on fetal sheep (Silverstein et al., 1967) revealed that they could reject with normal vigor skin homografts transplanted at midgestation, i.e., at about 75 days before birth and respond immunologically to inoculation with +X 174 bacteriophage as early as 40 days postconception, at a developmental stage before organized lymphoid tissue was demonstrable. However, they did not acquire the capacity to react against ovalbumin until 120 days of gestation and remained incapable of reacting against Salmonella or diphtheria until after birth. Thus, as Silverstein points out, there seems to be a stepwise development of immunological competence in early life, the ability to react against antigens of different types or classes being acquired at
68
ALAN E. BEER AND R. E. BILLINGHAM
widely different, apparently critical stages of gestation. There is no sudden, functional “switching on” of the immune apparatus as a whole. In rats and mice, although it is commonly believed that immunological competence is not acquired until after birth, several studies have shown that, at least with some strain combinations, the introduction of relatively small numbers of homologous cells into neonatal hosts may result in a weak, transient state of sensitivity (Brent and Gowland, 1963; Silvers and Billingham, 1966). Pertinent to this discussion is the fact that it is almost impossible to induce tolerance of A-strain tissues in C57 BL/6 mice by neonatal intravenous injection of high doses of A-strain cells. How soon before birth mice are capable of reacting against some transplantation antigens remains to be determined. Over the past decade, Dancis and his associates (Dancis et al., 1962, 1966, 1968) have presented exceedingly interesting findings indicative of potential immunological competence and hematopoietic function on the part of mouse placenta. Initially they found that carefully prepared placental cell suspensions from near-term C57 BL/6 mouse fetuses inoculated into neonatal BALB/c hosts were almost as effective as splenic cells from adult C57 donors in causing runt disease-a finding which Simmons and Russell (1964) were unable to confirm using a different mouse strain combination. Appropriate controls appeared to exclude the possibility that the deaths of the infant mice were caused by infection, toxic factors released from placental cells or the presence of contaminating immunocompetent cells. In subsequent studies, Dancis et al. ( 1968) reported that placental cells injected into ( a ) irradiated homologous adult mice caused an increased mortality, again suggestive of graft-versus-host reactivity, ( b ) isologous thymectomized newborn mice enhanced their capacity to synthesize antibodies against rat erythrocytes, and ( c ) lethally irradiated adult isologous hosts regularly “seeded” into the spleen producing nodules comprising erythroid and lymphoid cellular elements in approximately equal proportions. The use of cytologically marked cells showed unequivocally that these nodules developed from stem cells present in the placental cell inocula. The identity of the cell type present in the placenta has yet to be established. Dancis et al. obtained some evidence bearing upon this by carefully fetectomizing mice at 11 days gestation and removing the placentas to prepare cell suspensions 7 days later when they were almost entirely made up of trophoblast cells. They provisionally concluded that trophoblast cells are potentially pluripotential, while conceding the possibility that small numbers of adventitious cells of varied types, and from extraplacental sources in the fetus, may have secreted themselves in the trophoblast as early as the eleventh day of gestation.
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
69
Teleologically, one can envisage that the possession of premature immunological function within the placenta might well confer an important survival advantage upon mammals, protecting them not only against the hazards of contracting microbial infections afflicting their mothers, but also against the risk of surreptitious infiltration by maternal leukocytes which might cause runt disease. If the placenta does have a built-in immunological surveillance system, this might explain the great rarity with which maternal tumors are transmitted to fetuses. In conclusion we certainly agree with Benirschke and Driscoll ( 1967) that “the real role of the placenta as a reactant in infectious diseases and immunity and other defense mechanisms is just beginning to unfold.” XIII. Concept of Immunological Inertia of Viviparity
On the basis of the results of skin-grafting studies on outbred armadillos, dogs, sheep, but principally on rats, Anderson (1965, 1969, 1970) has postulated that a state of specific immunological hypoactivity or “immunological inertia” between mother and fetus occurs during pregnancy, having “special features which distinguished it from tolerance, paralysis or desensitization.” He interprets observations that skin grafts from very young offspring live significantly longer on their mothers than grafts from older offspring as indicative that the inertia of the mother toward her offspring’s tissue antigens is a temporary and gradually waning phenomenon. The finding that grafts from newborn rats transplanted to unrelated postpartum rats are promptly rejected is construed as evidence of the specificity of the phenomenon. However, the findings of others that the immunogenicity of infant rodents’ skin is demonstrably inferior to that of skin from older donors, especially where nonmajor locus histocompatibility differences are involved, suggests an alternative and more likely interpretation of the alleged inertia on the part of the mother ( Billingham and Silvers, 1964; Wachtel and Silvers, 1971). Furthermore, the evidence that the phenomenon is specific is totally unconvincing since, in outbred populations, one would expect grafts from offspring to survive on their own mothers better than grafts from unrelated donors of similar age. TO establish the validity of the concept of immunological inertia in outbred stocks requires demonstration that skin grafts from infants have a significantly greater expectation of survival on transplantation to their mothers than to their fathers, Obviously, the employment of inbred strains would facilitate the analysis. Furthermore, only uniparous females can be employed because of the risk of multiparity weakening the re-
70
ALAN E. BEER AND R. E. BILLINGHAM
activity of the mother through tolerance or enhancement (see Section X,E,l). XIV. Histocompatibility Gene Polymorphisms and Maternal-Fetal
Interactions
In his classic review of 1953, Medawar stated that “Although there are no factual grounds for supposing that antigenic diversity is anything but an unfortunate consequence of constitutional differences between individuals of a species, yet one is under some obligation to rack one’s brains for evidence of any good it might conceivably do. Only thus can antigenic polymorphism be made genetically respectable.” Over the last few years, as our knowledge of the genetics and immunology of transplantation has broadened and deepened, information has been forthcoming that in certain experimental animals, notably in mice and guinea pigs, very close associations exist between histocompatibility alleles at the major histocompatibility loci and capacity to react against certain synthetic antigens or susceptibility to viral oncogenesis ( Ellman et al., 1970; Snell, 1968). In man, too, unequivocal evidence is accumulating of association of susceptibility to certain malignant and other diseases with certain HL-A locus specificities (Pate1 et al., 1969; Walford et al., 1970; Zervas et al., 1970). Knowledge of the expression of transplantation antigens on the membranes of spermatozoa, on the cells of preimplantation embryos, and the homograft status of the latter in outbred populations, might lead one to suspect that some of the mechanisms which contribute to the maintenhnce of a stable genetic polymorphism of genes determining these cell membrane antigens (whatever their function may be) might operate ( I ) in the process of fertilization, which might be selective rather than random, ( 2 ) during the process of nidation, or (3) through interactions at the tissue or cellular level (including mutual cellular exchanges) between the fetus and its mother. In this section we shall review some of the evidence and concepts that bear upon this subject. When animals of two different inbred strains are mated, producing F, hybrid embryos which confront their mothers with alien histocompatibility factors, the litters tend to be larger and healthier than those produced by intrastrain matings due, so it is commonly asserted, to heterosis or hybrid vigor. However, such observations afford no proof that hybrids enjoy any particular advantage over genetically compatible offspring from conception to birth. To test this hypothesis entails comparison of the relative abilities of the two types of embryo to thrive in
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
71
the same maternal uterine milieu. The experimental approach requires animals of different inbred strains differing with respect to histocompatibility alleles that can be identified. Appropriate matings can then be set up to produce segregated populations of R, backcross or F, generations which can be tested to find out whether the H genotypes are present in the expected Mendelian ratios. For example, if we have two strains of mice which differ with respect to alleles at the H-2 locus, one being H-P/H-2" and the other H-2'c/H-2k,we would expect to find the genotypes H-2"/H-2", H-2a/H-2k,and H-2k/H-2kpresent in the ratio 1:2:1 in the F, generation. Alternatively, if we mate H-2"/H-2" mice with H-P/H-2" mice to produce a backcross population, we would expect genotypes H-PIH-2" and H-2a/H-2k in equal proportions. Significant deviations from these expected ratios would be indicative of the operation of some kind of prenatal selective factor. Experiments of this kind have been performed and, in some instances, the findings indicate that conceptuses that are disparate with their mothers at certain H loci enjoy a slight selective advantage over conceptuses that are compatible with their mothers at these loci. Hull ( 1969), following up a previous clue (Hull, 1964), carried out an experiment using mice of the C57 BLlO (H-3/H-3") strain and a congenic resistant strain B10. LP differing only with respect to a chromosomal segment bearing a different allele ( H-3b/H-3h)at the relatively minor H-3 locus. He found that in the progeny of H-3a/H-3b males backcrossed to H-3"/ H-3" or H-3b/H-3bfemales, significantly fewer than the expected 50% of the individuals were homozygous at this locus. Furthermore, this deficiency was only observed among the offspring of later (i.e. third and fourth) litters from H-3"/H-3" mothers and only among male offspring from H-3h/H-3b females. When reciprocal matings were set up, using F, hybrid mothers and homozygous fathers, the expected segregation ratios were obtained. Since there were no significant differences in weight at birth or at weaning between the homozygotes and the heterozygotes born of homozygous mothers, these findings suggest that if differential mortality is associated with this apparent "autoincompatibility," it may take place soon after conception and that after this stage the survivors develop normally. Hull's observation that, with one of the crosses tested, deficient segregation ratios were found only in third and fourth litters suggested that maternal sensitization to the alien antigen was an important factor in conferring a selective advantage upon the heterozygotes. Unfortunately, no experiments were performed with females specifically presensitized against the alien tissue antigen of their males before mating.
72
ALAN E. BEER AND R . E. BILLINCHAM
In rats, two independent groups of investigators have obtained evidence of a selective pressure( s ) that assured the survival of excess numbers of heterozygotes. Palm (1970) discovered an excess proportion of heterozygotes among the 2-month-old progeny of matings in which these animals were incompatible with their mothers at the Ag-B locus. When DA females were mated to (BN x DA)F, males, indications were strong that the abnormal segregation ratios resulted from a selective mortality among infant Ag-B homozygotes from a wasting disease, which affected twice as many Ag-B homozygotes as heterozygotes, male offspring being more susceptible than female. Under unfavorable environmental conditions all the individuals affected by this runting syndrome succumbed to it, but under other conditions recovery was possible. Evidence sustaining Palm’s tentative conclusion that this runting syndrome results from an immune response on the part of the mother against non-Ag-B antigens in the offspring includes the following: ( a ) the syndrome does not occur in reciprocal crosses, i.e., when (BN X DA)F, females are mated with DA males (under these circumstances, of course, all females are genetically tolerant of, and so incapable of reacting against, any of their offspring’s transplantation antigens) ; ( b ) onset of the disease is earlier and its severity increases with parity; and ( c ) the symptoms of this wasting disease, which included mild skin lesions and lymphoid tissue atrophy, are similar to those characterizing experimentally procured runt or homologous disease in this species (Billingham et al., 1962). As Palm points out, individuals that are heterozygous at the Ag-B locus, and so differ from their Ag-B homozygous mothers with respect to antigens determined by one Ag-B allele, likewise differ from their mothers at non-Ag-B histocompatibility loci, including one or more that render them susceptible to the postulated incompatibility reactions. Thus, at least in this particular experimental context, incompatibility with its mother at the Ag-B locus seems to confer protection of the fetus against the development and/ or consequences of immune reactivity on the part of the mother against antigens determined by other H loci. A much more striking example of heterozygote advantage had previously been discovered and partially elucidated by Michie and Anderson (1966) in the course of investigating the failure of seventy-two generations of brother x sister matings to produce an isohistogenic strain of Wistar rats. Despite this impressive history of inbreeding, about 50% of skin grafts exchanged between members of this strain were rejected within 2 to 3 weeks. Analysis revealed that the majority of the surviving rats were heterozygotes, resulting from an intense selection
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
73
against individuals that were homozygous for genes at an undefined H locus. These authors successfully selected and bred some of the few surviving progeny of one class of homozygote, producing an isohistogenic strain, but failed to recover the second class of homozygote. Since litter size data suggested that the selective elimination of homozygotes took place before implantation of the blastocysts, Michie and Anderson tentatively postulated that if the allelic histocompatibility genes concerned were g, and g,, and the survival value of heterozygous individuals was superior to that of either homozygote, this situation might result from selective fertilization, i.e., by g, sperm uniting preferentially with ge eggs, and g, sperm with g, eggs. Recent evidence that histocompatibility determinants were expressed by sperm in mice, rats, hamsters, and men (see Section IV,B) was consonant with this interpretation. In both Hull‘s work on mice and Palm’s on rats there was suggestive evidence that immunological reactivity on the part of homozygous mothers was in some way responsible for the maintenance of excessive numbers of heterozygotes among their progeny. In Michie and Anderson’s experiments there was no evidence of immunological reactivity on the part of the mothers being responsible for the abnormal segregation ratios. It is possible, however, that alleles were segregated at another histocompatibility locus ( i ) which their grafting tests did not reveal. In 1966, Clarke and Kirby postulated that some kind of immune interaction between females and their histoincompatible offspring might, contrary to expectation, actually favor the survival of such offspring and thus help to maintain the complex histocompatibility polymorphisms found in mammals. The essential basis of this thesis is the existing evidence that in mice both placental and fetal size are affected by antigenic differences (see Section VIII) suggesting that fetuses that are unlike their mothers tend to be larger at birth and have an increased chance of survival. In a recent review, and on the basis of somewhat tenuous evidence, Kirby (1970) has suggested that blastocysts that are genetically dissimilar to the mother implant more readily than blastocysts which are genetically similar. Although as yet there is no valid evidence that man has a Y-linked histocompatibility locus like that present in mice and rats ( Billingham and Silvers, 1971; LengerovL, 1970), assumption of its existence has led to some interesting speculations and interpretations of statistics pertaining to man. Kirby et a2. (1967) have presented an explanation of the human sex ratio of 0.5146 which turns upon the assumption that male zygotes must always be slightly more antigenic to their mothers than female zygotes by virtue of the superimposition of their Y-linked antigen upon
74
ALAN E. BEER AND R. E. BILLINCHAM
otherwise statistically equivalent, autosomally determined, paternal endowments of transplantation antigens. Since the results of sexing abortuses in humans indicate a considerably higher male sex ratio during pregnancy than at birth and yet, animal studies suggest a more or less equal number of male and female zygotes before implantation, these workers consider that more female blastocysts fail to be implanted than male blastocysts, the superior implantability of the latter revolving about possession of a Y antigen. If true, inbreeding should tend to lessen the differences at autosomal H loci between zygotes and their mothers and accentuate the importance of the Y antigen at implantation. This leads to the prediction that consanguineous marriages should produce a higher proportion of males than nonconsanguineous marriages. Reanalysis of previous data of one of these authors revealed that the sex ratio (i.e., 0.55) in one group of isolated, first-cousin marriages in a normal, outbreeding society was, indeed, significantly higher than the national average. Unfortunately, although essentially similar findings have been made independently by some investigators in various communities in the world, other studies which failed to produce corroborative evidence of disturbed sex ratios among the progeny of consanguineous marriages, particularly in “closed or isolated communities, greatly detract from the plausibility of this thesis. Recently, Kirby ( 1970) put forward another ingenious suggestion that the Y antigen in man may influence the sex ratio by interaction with antigens of the ABO blood group system, which are now recognized to be important determinants of histoincompatibility ( Dausset and Rapaport, 1968). Assuming that these antigens are expressed by the early blastocyst, if mother and zygote are compatible with respect to them (i.e., the mother has no ready-made isoantibody capable of reacting with the zygote), the Y-linked antigen of the latter should play a more important selective role in procuring successful implantation of male blastocysts than when zygotes are ABO incompatible with their mothers. Kirby cites sex ratio data in relation to blood group findings (Allan, 1959) that lend support to this reasoning. For example, AB mothers have significantly more male than female babies, and this high sex ratio is found among group 0 babies born of 0 mothers or of B babies born of B mothers. However, an irritating exception that cannot be explained is that the sex ratio in A babies born of A mothers is low. Finally, ex hypothesi, irrespective of the blood group of their mothers, 0 babies should have a high sex ratio, and this situation does appear to prevail in practice. Although maternal-fetal blood group incompatibilities have been under suspicion as a cause of toxemia of pregnancy for about 70 years,
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
7s
various studies appear to have exonerated the ABO and Rh blood group systems and, as we have already pointed out, the evidence is now very compelling that organ-specific antigens associated with the trophoblast probably play a leading role. Toivanen and Hirvanen (1970), in Finland, reported that the sex ratio in babies born to toxemic mothers is significantly elevated (i.e., 1.24). Furthermore, they have shown that this ratio increases with the severity of the disease as determined by daily urinary protein output or blood pressure. To explain these interesting findings the authors advanced the suggestion that paternally inherited histocompatibility antigens in the placenta may potentiate the immunogenicity of those antigens that it shares in common with the kidney and are believed to be responsible for toxemia. Male fetuses might be expected to be slightly more effective than female fetuses in initiating this disease on account of their immunogenic “edge” or advantage from the postulated possession of a Y-chromosome-dependent transplantation antigen. Pertinent to speculation about the biological significance of the Y antigen in man, if it is present, is McLaren’s (1962) analysis of the influence of the state of reactivity of C57 BL female mice to the Y antigen on their reproductive performance, when mated with males of their own strain, since on this genetic background the immunogenicity of this factor is not trivial. She found that the mean litter size and sex ratio of offspring were closely similar irrespective of whether their mothers were sensitized against the Y antigen, made tolerant of it (evidenced by inability to reject skin isografts from male donors), or received no prior treatment, i.e., were normal control mothers. It is worth emphasizing that the sex ratio in the progeny of these controls, as in the experimental series, was essentially equal. Ounsted and Ounsted (1970) have recently extended a prior suggestion that antigenic dissimilarity between human mothers and theif conceptuses may contribute to the enhancement of fetal growth rate, to explain the observed differences in growth rates between the fetuses of the two sexes. They cite birth weight data in support of this premise. In conclusion, the influence of isoantigenic disparities between preimplantation zygotes and their mothers on the segregation ratios of offspring at birth is a sul)ject that merits further investigation rather than speculation. Important unresolved questions are whether there is a Y-linked antigen in marl and whether it is expressed on sperms and early zygotes.
ACKNOWLEDGMENTS The expenses of some of the experimental work cited and the preparation of this article were defrayed in part by grants AI-07001 from the U. S. Public Health
76
ALAN E. BEER AND R. E. BILLINGHAM
Service, the Ford Foundation, and the Lalor Foundation. The authors are deeply indebted to their colleague, Dr. Willys K. Silvers, for advice. &WEIWNCES
Allan, T. M. (1959). Brit. Med. I. 1, 553-554. Amos, D. B. (1969). Advan. Zmmunol. 10, 251-297. Anderson, J. M. (1965). Nature (London) 206, 786. Anderson, J. M. (1969). Transplant. Proc. 1, 67-70. Anderson, J. M. (1970). Proc. Roy. SOC., Ser. B 176, 115-129. Andresen, R. H., and Monroe, C. W. (1962). Amer. J. Obstet. Gynecol. 84, 10961100. Austin, C. R. ( 1957). J. Endocrinol. 14, 335342. Austin, C. R. (1959). Nature (London) 183, 293. Averill, R. L. W., and Rowson, L. E. A. (1958). J. Endocrinol. 16, 32W36. Bach, F. H. (1968). Science 159, 1196-1198. Bagshawe, K. D. (1969). “Choriocarcinoma,” pp. 1-360. Williams & Wilkins, Baltimore, Maryland. Bagshawe, K. D. (1970). Zn “The Biology and Surgery of Tissue Transplantation” (E. J. Maxwell Anderson, ed.), pp. 139-152. Davis, Philadelphia, Pennsylvania. Bardawil, W. A., and Toy, B. L. (1959). Ann. N . Y. Acad. Sci. SO, 197-257. Bardawil, W. A., Mitchill, G. W., Jr., McKeogh, R. P., and Marchant, D. J. (1962). Amer. J. Obstet. Gynecol. 84, 1283-1299. Barker, C. F., and Billingham, R. E. (1968). J. Exp. Med. 128, 197-221. Barker, C. F., and Billingham, R. E. (1971). J. Exp. Med. 133, 620-639. Barth, R. F., and Russell, P. S. (1964). J. Zmmunol. 93, 13-19. Beer, A. E. (1969). Obstet. Gynecol. 34, 143-150. Beer, A. E., and Billingham, R. E. (1970). 1. Exp. Med. 132, 721-736. Beer, A. E., and Billingham, R. E. (1971). Unpublished data. Beer, A. E., Billingham, R. E., and Hoerr, R. A. (1971a). Transplant. Proc. 3, 809-611. Beer, A. E., Yang, S. L., and Billingham, R. E. (1971b). Unpublished data. Benirschke, K., and Driscoll, S. G. ( 1967). “The Pathology of the Human Placenta,” pp. 1 5 1 2 . Springer-Verlag, New York. Betke, K., and Kleihauer, E. (1958). Blut 4, 241. Bevans, M., Seegal, B. C., and Kaplan, R. (1955). J . Exp. Med. 10.2, 807-822. Billingham, R. E. (1964). N. Engl. J. Med. 270, 667-672 and 720-725. Billingham, R. E. ( 1968). Haruey Lect. 62, 21-78. Billingham, R. E. (1971). Cell. Zmmunol. 2, 1-12. Billingham, R. E., and Barker, C. F. (1969). P h t . Reconstr. Surg. 44, 2 0 3 0 . Billingham, R. E., and Lampkin, G. H. (1957). J. Embryol. Exp. Morphol. 5, 351-367. Billingham, R. E., and Silvers, W. K. (1964). Proc. Roy. SOC.,Ser. B 161, 168-190. Billingham, R. E., and Silvers, W. K. (1965). J. Exp. Zool. 160, 221-224. Billingham, R. E., and Silvers, W. K. (1971). “The Immunobiology of Tissue Transplantation.” Prentice-Hall, Englewood Cliffs, New Jersey. Billingham, R. E., and Sparrow, E. M. (1954). J. Exp. Biol. 31, 16-39. Billingham, R. E., and Sparrow, E. M. (1955). 1. Embryol. Exp. Morphol. 3, 265285. Billingham, R. E., Krohn, P. L., and Medawar, P. B. ( 1951). Brit. Med. J . 5 1049-1053.
IMMUNOBIOLOCY OF MAMMALIAN REPRODUCTION
77
Rillingham, R. E., Lampkin, G. H., Medawar, P. B., and Williams, H. L. (1952). Heredity 6, 201-212. Billingham, R. E., Brent, L., and Medawar, P. B. (1956). Phil. Trans. Roy. SOC. London, Ser. B 239, 357414. Billingham, R. E., Defendi, V., Silvers, W. K., and Steinmuller, D. (1962). J. Nut. Cancer Inst. 28, 365435. Billingham, R. E., Palm, J., and Silvers, W. K. (1965a). Science 147, 514516. Billingham, R. E., Silvers, W. K., and Wilson, D. B. (196513). Proc. Roy. SOC., Ser. B 163, 61-89. Billington, W. D. .( 1964). Nature (London) 202, 317. Billington, W. D. (1965). J . Reprod. Fert. 10, 343-352. Billington, W. D. (1966). Nature (London) 211, 988-989. Billington, W. D. (1970). In “Immunology and Development” (M. AdinoK, ed.), pp. 89-113. Lavenham Press, Ltd., Lavenham, Suffolk. Billington, W. D., and Weir, B. J. (1967). J. Reprod. Fert. 13, 593-595. Billington, W. D., Kirby, D. R. S., Owen, J. J. T., Ritter, M. A., Burtonshaw, M. D., Evans, E. P., Ford, C. E., Gauld, I. K., and McLaren, A. (1969). Nature (London) 224, 701. Boettcher, B., Hay, J., Kay, D. J., Baldo, B. A., and Roberts, T. K. (1970). Int. I. Fert. 15, 143. Boss, J. H. (1965). Amer. J. Obstet. Gynecol. 93, 574-582. Boss, J. H., and Craig, J. M. ( 1963). Amer. J . Pathol. 42, 443-454. Boyd, J. D. (1959). Ulster Med. 1. 28, 3 5 4 6 . Brackett, B. G. ( 1971). Personal communication. Bradbury, S., Billington, W. D., and Kirby, D. R. S. (1965). J. Roy. Microsc. SOC. [3] 84, 199-211. Bradbury, S., Billington, W. D., Kirby, D. R. S., and Williams, E. A. (1969). Amer. J. Obstet. Gynecol. 104, 416418. Brambell, F. W. R. (1970). “The Transmission of Passive Immunity from Mother to Young” (A. Neuberger and E. L. Tatum, eds.), pp. 1-385. North-Holland Publ., Amsterdam. Bratanov, K. ( 1969). In “Immunology and Reproduction” (R. G. Edwards, ed.), p. 178. International Publications, New York. Brent, L., and Gowland, G. (1963). Transplantation 1, 372-376. Brent, L., and Medawar, P. B. (1959). Recent Progr. Microbial., Symp. Int. Congr. Microbial., 7th, 1958 pp. 181-190. Brent, R. L., and Averich, E. (1961). Amer. J. Dis. Child. 102, 689-691. Breyere, E. J. ( 1967). Transplantation 5, 1504-1509. Breyere, E. J., and Barrett, M. K. (1960a). J . Nat. Cancer Inst. 24, 699-705. Breyere, E. J., and Barrett, M. K. (1960b). J. Nat. Cancer Inst. 25, 1405-1410. Breyere, E. J., and Barrett, M. K. (1962). J. Nut. Cancer Inst. 27, 409-417. Breyere, E. J., and Burhoe, S. 0. (1963). J. Nut. Cancer Inst. 31, 179-188. Breyere, E. J., and Sprenger, W. W. (1969). Transplant. Proc. 1, 71-75. Clarke, B., and Kirby, D. R. S. (1966). Nature (London) 211, 999-1000. Cohen, F., and Zuelzer, W. W. ( 1967). Blood 30, 796-804. Cohen, F., Zuelzer, W. W., Gustafson, D. G., and Evans, M. M. (1964). Blood 23, 621. Currie, G. A. (1967). J . Obstet. Gynaecol. Brit. Commonw. 74, 841-848. Currie, G. A. (1970). Proc. Roy. SOC. Med. 63, 61-64. Currie, G. A., and Bagshawe, K. D. (1967). Lancet 1, 708-710.
78
ALAN E. BEER AND R. E. BILLINGHAM
Currie, G. A., Van Doorninck, W., and Bagshawe, K. D. (1968). Nature (London) 219, 191-192. Curzen, P. (1968). J. Obstet. Gynuecol. Brit. Commonw. 75, 1128-1133. Dain, A. R., and Tucker, E. M. (1970). Proc. Roy. Soc., Ser. B 175, 183-200. Dancis, J., Samuels, B. D., and Douglas, G. W. (1962). Science 136, 382. Dancis, J., Douglas, G. W., and Fierer, J. (1966). Amer. J . Obstet. Gynecol. 94, 50-56. Dancis, J., Jansen, V., Gorstein, F., and Douglas, G. W. (1968). Amer. J. Obstet. Gynecol. 100, 1110-1121. Darrow, R. R. (1938). Arch. Pathol. 25, 378. Dausset, J., and Rapaport, F. T. (1968). In “Human Transplantation” (F. T. Rapaport and J. Dausset, eds.), pp. 383-393. Grune & Stratton, New York. Davies, D. A. L. (1968). In “Human Transplantation” (F. T. Rapaport and J. Dausset, eds.), pp. 619-634. Grune & Stratton, New York. Dkmant, P. ( 1968). Folia Biol. (Prague) 14, 9-20. Dkmant, P., Ivanyi, P., and EvaskovA, E. (1966). Ann. N . Y. Acad. Sci. 129, 234240. Diamondopoulos, G. T., and Hertig, A. T. (1963). Obstet. Gynecol. 21, 150-154. Dobrowolski, M. S. (1903). Bull. Int. Acad. Sci. Cracooie 5, 256-260. Doniach, I., Crookston, J. H., and Cope, T. I. (1958). J. Obstet. Gynaecol. Brit. Emp. 65, 553-556. Doyle, L. L., Gates, A. H., and Noyes, R. W. (1963). Fert. Steril. 14, 215. Edwards, R. G. (1964). Nature (London) 203, 50. Edwards, R. G., Ferguson, L. C., and Coombs, R. R. A. (1964). J. Reprod. Fert. 7, 153-161. Ellman, L., Green, I., and Martin, W. J. (1970). Lancet 1, 1104-1106. Elston, C. W. (1969). J. Patho2. 97, 261-268. Fellow, M., and Dausset, J. (1970). Nature (London) 225, 191. Finegold, M., and Michie, D. (1961). J. Embryol. Exp. Morphol. 9, 618-622. Finn, C. A., and McLaren, A. (1967). J. Reprod. Fert. 13, 259-267. Finn, R. (1960). Lancet 1, 526. Freund, J., Lipton, M. M., and Thompson, G. E. (1953). J. Erp. Med. 97, 711-725. Galton, M. (1960). Lancet I, 495-496. Galton, M. (1965). Transplantation 3, 39-43. Goldberg, E. H., Aoki, T., Boyse, E. A., and Bennett, D. (1970). Nature 228, 570-572. Goodlin, R. C., and Herzenberg, L. A. (1964). Transplantatton 2, 357461. Gordon, I. (1960). Nature (London) 185, 118. Gorrnan, J. G., Freda, V. J., and Pollack, W. (1964). Proc. Congr. Int. SOC.Hemutol., 9th, 1962 p. 545. Gray, A. P. (1954). “Mammalian Hybrids: Check List with Bibliography,” Tech. Commun. No. 10. Commonw. Bur. Anim. Breed. Genet., Edinburgh, Scotland. Gross, S. J. (1966). Amer. J. Obstet. Gynecol. 95, 1149-1159. Gullbring, B. (1957). Acfa Med. Scand. 159, 169-172. Halbrecht, I., and Komlos, L. (1968). Obstet. Gynecol. 31, 173-177. Haller, J. A., and Billingham, R. E. (1967). Ann. Surg. 166, 896-901. Halpern, B. N., Ky, T., and Robert, B. (1967). Immunology 12, 247-258. Hampton, J. K., and Hampton, S. H. (1965). Science 150, 915-917. Hancock, J. L., McGovern, P. T., and Stamp, J. T. (1968). J. Reprod. Fed. Suppf. 3, 29.
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
79
Hartman, C. G. (1959). Ann. N . Y. Acad. Sci. 83, 318-327. HaHkovi, V. ( 1961 ). Folia B i d . (Prague) 7, 322-331. HaHkovi, V. (1963). Folia Biol. (Prague) 9, 99-103. Hathaway, W. E., Githens, J. H., Blackburn, W. R., Fulginiti, V., and Kempe, C. H. (1965). N . Engl. J. Med. 273, 953-958. Heape, W. (1891). Proc. Roy. SOC. 48, 457-458. Hellman, L. M., and Eastman, N. J, (1966). “William’s Obstetrics,” 13th ed. Appleton, New York. Hellstrom, I., and Hellstriim, K. E. (1966). Ann. N. Y. A d . Sci. 129, 724-734. Hellstrom, I., and Hellstrom, K. E. (1969). Int. J. Cancer 4, 587-600. Hellstrom, K. E., Hellstrom, I., and Brawn, J. (1969). Nature (London) 224, 914-915. Henle, W., Henle, G., and Chambers, L. A. (1938). J. Exp. Med. 68, 335352. Hertz, R., Lewis, J., and Lipsett, M. B. ( 1961). Amer. J . Obstet. Gym’col. 82, 631440. Herzenberg, L., and Gonzales, B. (1962). Proc. Nut. Acad. Sci. U. S . 48, 570-573. Heslop, R. W., Krohn, P. L., and Sparrow, E. M. (1954). 1. Endocrind. 14, 240-256. Heyner, S. (1969). Transplantation 8, 666-677. Heyner, S. ( 1970). Transplantation 10, 278-280. Heyner, S., Brinster, R. L., and Palm, J. (1969). Nature (London) 222, 783-784. Howard, J. G., and Michie, D. (1962). Transplant. BuU. 29, 91-96. Hulka, J. F. (1968). Science 161, 696-698. Hulka, J. F., and Brinton, V. (1963). Amer. J . Obstet. Gynecol. 86, 130-134. Hulka, J. F., and Mohr, K. (1968). Science 161, 696-698. Hulka, J. F., Brinton, V., Schaaf, J., and Baney, C. (1963). Nature (London) 198, 501-502. Hull, P. (1964). Genetics 50, 563470. Hull, P. (1969). Heredity 24, 203-209. Ilkk, F. A. (1961). Schweiz. Med. Wochenschr. 91, 934445. Ilkk, F. A. (1964). Schweiz. Akad. Med. Wiss. 20, 62-72. Ilya, F. A., Williamson, S., and b a r , H. A. (1967). Cancer 20, 144-149. Jackson, B. T. (1967). Surgery 62, 232-237. James, D. A. ( 1965). Nature (London) 205, 613-614. James, D. A. (1967). 3. Reprod. Fed. 14, 265-275. James, D. A. (1969). Transplantation 8, 846851. Jandl, J. H., Jones, R. A., and Castle, W. B. (1957). J . Clin. Znuest. 36, 1428. Jones, B. M., and Kemp, R. B. (1969). Nature (London) 221, 829431. Jones, E. C., and Krohn, P. L. (1962). Nature (London) 195, 1064-1066. Jones, W. R. ( 1968). Nature (London) 218, 480. Kadowaki, J. I., Zuelzer, W. W., Brough, A. S., Thompson, R. I., Wooley, P. V., and Gruber, D. (1966). Lancet 2, 1152-1155. Kaku, M. ( 1953). J . Obstet. Gynaecol. Brit. Emp. 60, 148-156. Kaliss, N., and Dagg, M. (1964). Transplantation 2, 416-225. Kaliss, N., and Rubinstein, P. ( 1968a). In “Advances in Transplantation” (J. Dausset, J. Hamburger, and G. Mathk, eds.), pp. 531-537. Munksgaard, Copenhagen. Kaliss, N., and Rubinstein, P. (196813).Proc. SOC. Exp. B i d . Med. 128, 1214-1217. Katsh, G. F., Talmage, D. W., and Katsh, S. (1964). Science 143, 4142. Katsh, S. (1969). In “Ovum Implantation” (M. C. Shelesnyak and G. J. Marcus, eds.), pp. 309-344. Gordon & Breach, New York. Kerr, W. R., and Robertson, M. (1953). J. Hyg. 51, 405-415.
80
ALAN E. BEER AND R. E. BILLINGHAM
Kirby, D. R. S. (1962). Nature 194, 786. Kirby, D. R. S. (1968a). In “Human Transplantation” (F. T. Rapaport and J. Dausset, eds.), pp. 565-586. Grune & Stratton, New York. Kirby, D. R. S. (1968b). Transplantation 6, 1005-1009. Kirby, D. R. S. (1969). Fert. Steril. 20, 933-937. Kirby, D. R. S. (1970). Proc. Roy. SOC. Med. 63, 59-61. Kirby, D. R. S., Billington, W. D., Bradbury, S., and Goldstein, D. (1964). Nature (London) 204, 548-549. Kirby, D. R. S., Billington, W. D., and James, D. A. (1966). Transplantation 4, 713-718. Kirby, D. R. S., McWhirter, K. G., Teitlebaum, M. S., and Darlington, C. D. (1967). Lancet ii, 139-140. Kleihauer, E., Braun, H., and Betke, K. (1957). Klin. Wochenschr. 35, 637-638. Koren, Z., Abrams, G., and Behrman, S. J. (1968a). Amer. J . Obstet. Gynecol. 102, 340446. Koren, Z.,Abrams, G., and Behrman, S. J. (196813). Amer. J. Obstet. Gynecol. 100, 570575. Koren, Z.,Behrman, S. J., and Paine, P. J. (1969). Amer. J. Obstet. Gynecol. 104, 50-57. Landsteiner, K. ( 1899). Zentralbl. Bakteriol. 25, 546-549. Landsteiner, K., and Levine, P. (1926). J. Immunol. 12, 415-418. Lanman, J. T. (1965). J. Pediat. 66, 525540. Lanman, J. T., Dinerstein, J., and Fikrig, S. (1962). Ann. N. Y. Acad. Sci. 99, 706-7 16. Lengerovi, A. ( 1957). Folia Biol. (Prague) 3, 333-337. Lengerovi, A. ( 1970). In “Immunogenetics of Tissue Transplantation” (A. Neuberger and E. Tatum, eds. ), North-Holland Publ., Amsterdam. Lengerovh, A., and Vojtiskovi, M. (1963). Folia Biol. (Prague) 9, 72. Levine, P., and Celano, M. J. ( 1961). Vox Sang. 6, 720-723. Levine, P., and Stetson, R. E. (1939). J. Amer. Med. Ass. 113, 126. Lewis, J., Whang, J., Nagel, B., Oppenheim, J. J., and Perry, S. (1966). Amer. 1. Obstet. Gynecol. 96, 287-290. Lippman, M. (1968). Nature (London) 219, 33-36. Little, C. C. ( 1924). J . Cancer Res. 8, 75-95. Lustgraff, E. C., Fuson, R. B., and Eichwald, E. J. (1960). Transplant. Bull. 26, 145-150. McCaughey, W. T. E. (1955). J. Obstet. Gynaecol. Brit. Emp. 62, 863-869. Macher, E., and Dorner, V. (1966). J . Immunol. 97, 484-491. McLaren, A. (1962). Nature (London) 195, 1323-1324. McLaren, A. ( 1964). Nature (London) 201, 582585. McLaren, A. ( 1965a). In “Symposium on the Early Conceptus, Normal and Abnormal” (W. W. Park, ed.), p. 27. Williams & Wilkins, Baltimore, Maryland. McLaren, A. (1965b). J. Reprod. Fert. 9, 79-98, McLaren, A. (1966). Fert. SteriZ. 17, 492499. McLaren, A., and Finn, C. A. (1967). J. Reprod. Fert. 13, 259-267. McLaren, A., and Michie, D. (1956). J. Erp. Biol. 33, 394-416. McLaren, A., and Michie, D. (1959a). J. E m . Biol. 36, 40. McLaren, A., and Michie, D. (195913). Mem. SOC. Endocrinol. 6, 65. McLean, J. M., and Scothorne, P. J. (1968). J . Anat. 103, 402. Macris, M. T., Hellman, L. M., and Watson, R. J. (1958). Amer. J. Obstet. Gynecol. 76, 1214-1218.
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
81
Marcus, G. J., and Shelesnyak, M. C. (1968). Acta Endocn’nol. (Copenhagen) 57, 136-141.
Martinek, J. J. (1970). Anat. Rec. 166, 587-603. Martinek, J. J. (1971). Amer. J. Obstet. Gynecol. 109, 424-431. Maruta, H., and Moyer, D. L. (1965). Fed. Proc., Fed. Amer. SOC.Exp. Biol. 24, 450.
Masouredis, S. P. ( 1962 ). Transfusion (Philadelphia) 2, 363. Masouredis, S. P., Chi, C. A., and Ferguson, E. (1960). 1. Clin. Invest. 39, 1450. Medawar, P. B. (1953). Symp. SOC. Exp. Biol. 11, 320-338. Medawar, P. B. (1969). Proc. Roy. SOC.,Ser. B 174, 155-172. Medawar, P. B., and Sparrow, E. M. (1956). J. EndocrinoL 14, 240-256. Metalnikoff, S. (1900). Ann. Inst. Pasteur, Paris 14, 577-589. Metchnikoff, E. (1900). Ann. Inst. Pasteur, Paris 14, 1-12. Michie, D., and Anderson, N. F. (1966). Ann. N . Y. Acad. Sci. 129, 88-93. Mitchison, N. A. (1953). J. Genet. 51, 406-420. Mitchison, N. A. (1965). Proc. Roy. SOC.London B 161, 275-292. Mogensen, B., and Kissmeyer-Nielsen, F. (1968). Lancet 1, 721-725. Mogensen, B. Kissmeyer-Nielsen, F., and Hauge, M. ( 1968). Transplant. Proc. 1, 76-79. Moller, G., and Moller, E. (1966). Ann. N . Y. Acad. Sci. 129, 735-749. Moulton, M. A., Stimpfling, J., and Storer, J. B. (1960). Transplant. Bull. 26, 454455.
Naiman, J. L., Punnett, H. H., Destine, M. L., and Lischner, H. W. (1966). Lancet 2, 590. Najarian, J. S., and Dixon, F. J. (1963). Proc. SOC. E x p . Biol. Med. 112, 136-138. Nathan, P., Gonzales, E., and Miller, B. F. (1960). Nature (London) 188, 77-78. Nelson, J. H., and Hall, J. E. (1964). Amer. J. Obstet. Gynecol. 19, 482484. Nevanlinna, H. R., and Vainio, T. (1956). Vox Sang. 1, 26-36. Nicholas, J. S. (1932). Proc. SOC. E x p . Biol. Med. 30, 1111-1113. Nossal, G. J. V. (1968). In “Human Transplantation” ( F . T. Rapaport and J. Dausset, eds.), pp. 643-654. Grune & Stratton, New York. Ober, W. (1968). Obstet. Gynecol. 31, 8 6 H 6 9 . Oehme, J., Hundeshagen, H., and Eschenbach, C. (1966). Klin. Wochenschr. 44, 430-433.
Oettingen, Kj. V., and Witebsky, E. (1928). Muenchen. Med. Wochenschr. 75, 385-386.
Olds, P. J. ( 1968). Transplantation 6, 478-479. Orsini, M. (1954). Amer. J. Anat. 94, 273-332. Ounsted, C., and Ounsted, M. (1970). Lancet 2, 857-870. Palm, J. (1970). Transplant. Proc. 2, 162-173. Palm, J., Heyner, S., and Brinster, R. L. (1971). J . E x p . Med. 133, 1282-1293. Papiemik-Berhauer, E. (1966). Reu. Fr. Etud. Clin. Biol. 11, 239-247, Park, W. W., ed. (1965). “The Early Conceptus Normal and Abnormal.” Williams and Wilkins, Baltimore. Patel, R., Mickey, M. R., and Terasaki, P. I. (1969). Brit. Med. J. 2, 424-426. Payne, R., and Rolfs, M. B. (1958). 1. Clin. Inuest. 37, 1756-1763. Peer, L. A., Walia, I. S., and Pullen, R. (1960). Transplant. Bull. 26, 115-118. Poppa, G., Simmons, R. L., David, D. S., and Russell, P. S. (1964). Transplantation 2, 496-502. Porter, J. B., and Breyere, E. J. (1964). Transplantation 1, 247-250. Potter, J. F., and Schoeneman, M. (1970). Cancer 25, 380-388.
82
ALAN E. BEER AND R. E. BILLINGHAM
Prehn, R. T. (1960). J . Nut. Cancer Inst. 25, 883-886. Pressman, D., and Korngold, D. (1957). I. Immunol. 78, 75-78. Race, R. R., and Sanger, R. ( 1968). “Blood Groups in Man,” 5th ed. Blackwell, Oxford. Ramseier, H., and Billingham, R. E. (1966). J. Exp. Med. 123, 629-656. Ramseier, H., and Brent, R. L. (1966). Ann. N . Y. Acad. Sci. 129, 241-249. Rapaport, F. T., and Converse, J. M. (1957). Ann. N. Y. Acad. Sci. 64, 836-841. Reid, B. L. (1965). Exp. Cell Res. 40, 679483. Reid, B. L., and Blackwell, P. M. (1967). Anst. 1. Exp. Biol. Med. Sci. 45, 323-326. Reynolds, S. R. M. (1969a). Int. J. Gynecol. Obstet. 7, 43-55. Reynolds, S. R. M. (1969b). Int. I. Gynecol. Obstet. 7, 109-123. Rochna, E., and Hughes-Jones, N. C. (1965). Vox Sang. 10, 675-686. Rudolph, R. H., and Thomas, E. D. (1970). Amer. J. Obstet. Gynecol. 108, 11261129. Rubinstein, P., and Kaliss, N. (1964). Transplantation 2, 543-545. Salvaggio, A. T., Nigogosyan, G., and Mack, H. C. (1960). Amer. J. Obstet. Gynecol. 80, 1013-1021. Sanford, B. H. (1963). J. Nut. Cancer Inst. 31, 169-178. Santos, G. W., and Owens, A. H. (1968). In “Advances in Transplantation” ( J . Daussett, J. Hamburger, and G. Math6, eds. ), pp. 431436. Munksgaard, Copenhagen. Schlesinger, M. (1962). J. Not. Cancer Inst. 28, 927-945. Schwimmer, W. B., Ustay, K. A., and Behrman, S. J. (1967). Obstet. Gynecol. 30, 192-200. Scott, J. S. (1966). Brit. Med. J. 1, 1559-1567. Seegal, B. C., and Loeb, E. M. (1940). Proc. SOC. Exp. Biol. Med. 45, 248-252. Seegal, B. C., and Loeb, E. M. (1946). J. Exp. Med. 45, 211-221. Seigler, H. F., and Metzgar, R. S. (1970). Transplantation 9, 478-486. Shahani, S., and Southam, A. I-,. (1962). Amer. J. Obstet. Gynecol. 84, 660-666. Shelesnyak, M. C., and Kraicer, P. F. (1963). In “Delayed Implantation” (A. C. Enders, ed.), p. 265. Univ. of Chicago Press, Chicago, Silvers, W. K., and Billingham, R. E. (1966). J. Exp. Zool. 161, 413-423. Silvers, W. K., and Billingham, R. E. (1969). Transplantation 8, 167-178. Silverstein, A. M. (1964). In “Immunopathology of Uveitis” (A. E. Maumenee and A. M. Silverstein, eds.), pp. 209-217. Williams & Wilkins, Baltimore, Maryland. Silverstein, A. M., and Lukes, R. J. (1962). Lab. Invest. 11, 918-932. Silverstein, A. M., Parshall, C. J., and Prendergast, R. A. (1967). In “Ontogeny of Immunity” (R. T. Smith, R. A. Good, and P. A. Miescher, eds.), pp. 143147. Univ. of Florida Press, Gainesville. Simmons, R. L., and Russell, P. S. (1962). Ann. N . Y. Acad. Sci.99, 717-732. Simmons, R. L., and Russell, P. S. (1964). Transplantation 2, 431436. Simmons, R. L., and Russell, P. S. (1966). Ann. N. Y. Acad. Sci. 129, 35-45. Simmons, R. L., and Russell, P. S. (1967a). Aduan. Obstet. Gynecol. 1, 38. Simmons, R. L., and Russell, P. S. (196713). Transplantation 5, 85-88. Simmons, R. L., Cruse, V., and McKay, D. G. (1967a). Amer. J. Obstet. Gynecol. 97, 218-230. Simmons, R. L., Ozerkis, A. J., Butsch, D. W., and Russell, P. S. (1967b). Amer. J. Obstet. Gynecol. 99, 266-270. Simonsen, M. (1962). Progr. Allergy 6, 34-67,
IMMUNOBIOLOGY OF MAMMALIAN REPRODUCTION
83
Smith, K., Duhring, J. L., Green, J. W., Rochlin, D. B., and Blakemore, W. S. ( 1961). Obstet. Gynecol. 18, 673-676. Snell, G. D. (1968). Folia Biol. (Praha) 14, 335-358. Snell, G. D. (1970). Surg., Gynecol. Obstet. 130, 1109-1119. S o h , L. (1967). Nature (London) 213, 621-622. Stastny, P. (1965). J. Immunol. 95, 929-936. Steblay, R. W. (1962). J . Immunol. 88, 434442. Stem, K., Davidsohn, I., and Masaitis, L. (1956). Amer. J. Clin. Pathol. 26, 833843. Tai, C., and Halasz, N. A. (1967). Science 158, 125-126. Terasaki, P. I., Mickey, M. R., Yamazaki, J. N., and Vredovoe, D. (1970). Transplantation 9, 538-543. Terni, T., and Maleci, 0. (1937). Arch. Exp. Zellforsch. Besonders Geuebezuecht. 19, 165-170. Thiede, H. A., Choate, H. W., Gardiner, H. H., and Santay, H. (1965). J. Exp. Med. 121, 1039-1050. Thomas, L. ( 1959). In “Cellular and Humoral Aspects of the Hypersensitive States” (H. S. Lawrence, ed.), pp. 529532. Harper (Hoeber), New York. Toivanen, P., and Hirvanen, T. (1970). Science 170, 187-188. Torniov, A. (1970). Vet. Sci. 4, 83-90. Toullet, F., Voisin, G. A., and Nemirovsky, M. (1970). Ann. Inst. Pasteur, Paris 118, 513524. Tuffrey, M., Bishun, N. P., and Barnes, R. D. (1969a). Nature (London) 221, 10291030. Tuffrey, M., Bishun, N. P., and Barnes, R. D. (196913). Nature (London) 224, 701704. Turner, J. H., Wald, N., and Quinlivan, W. L. G. (1966). Amer. J. Obstet. Gynecol. 95, 831-833. Tyler, A. (1961a). J. Reprod. Fed. 2, 473-506. Tyler, A. (1961b). Clin. Obstet. Gynecol. 7, 829-843. Uhr, J, W., and Anderson, S. G. (1962). Nature (London) 194, 1292.1293. Van Rood, J. J., van Leeuwen, A,, and Eernisse, J. G. (1959). Vox Sang. 4, 427444. Voisin, G. A. (1970). Ann. Inst. Pasteur, Paris 118, 403419. Voisin, G. A., and Kinsky, R. (1962). Transplant. Ciba Found. Symp., 1961 pp. 286326. Voisin, G. A,, Delaunay, A., and Barber, M. (1951). Ann. Inst. Pasteur, Paris 81, 48-63. Vojtis’kovi, M. ( 1969). Nature (London) 222, 1293. Vojtis’kovi, M., PoliEkova, M., and Pokorni, Z. (1969). Folia Biol. (Prague) 15, 322-332. Wachtel, S. S., and Silvers, W. K. (1971). J . Exp. Med. 133, 921-937. Walford, R. L., Finkelstein, S., and Neerhout, R. (1970). Nature 225, 561-562. Walknowska, J., Coute, F. A., and Grumbach, M. M. (1969). Lancet 1, 1119-1192. Wanvick, B. L., and Berry, R. 0. (1949). J. Hered. 40, 297-303. Willett, E. L. (1953). Iowa State Coll. J . Sci. 28, 83-100. Wilson, D. B., and Billingham, R. E. (1967). Advan. Immunol. 7, 189-273. Winn, H. J. (1970). Transplant. PTOC.2, 83-91. Witebsky, E., and Reich, H. (1932). Klin. Wochenschr. 11, 1960-1961. Wolstenholme, G. E. W., and O’Conner, M., eds. (1969). “Ciba Foundation Symposium on Foetal Autonomy.” Churchill, London.
84
ALAN E. BEER AND R. E. BILLINGHAM
Woodrow, J. C., and Donohoe, W. T. A. (1968).Brit. Med. J. 4, 139. Woodruff, M. F. A. (1958).Proc. Roy. SOC. Ser. B 148, 68-75. Wynn, R. M. (1967a).Obstet. Gpecol. 29, 644-661. Wynn, R. M. (1967b).Amer. 1. Obstet. Gynecol. 97, 832-850. Wynn, R. M. (1969).Amer. 1. Obstet. GynecoZ. 103, 723-739. Zervas, J. D., Delamore, I. W., and Israels, M. C. G. (1970).Lancet 2, 634-635. Zipursky, A., Pollack, J., Neelands, P., Chown, B., and Israels L. G. (1963).Lancet 2, 489493.
Thyroid Antigens and Autoimmunity SIDNEY SHU LMAN Department of Microbiology. New York Medical College. New York. New York
I. Introduction . . . . . . . . . . . I1. The Thyroid Gland: Structure. Function. and Malfunction . A. Histological Structure and Iodine Storage . . . . B. Survey of Thyroid Disorders . . . . . . . 111 Purification and Properties of Thyroid Proteins . . . . A. Analysis of Protein Constituents in Tissue Extracts . . B. Fractionation Efforts and Results . . . . . . IV . Thyroid Antigens . . . . . . . . . . A. Tissue Specificity and Organ Specificity-a Semantic Problem . B. Antigenic Properties as Revealed by Heteroimmunization C . Isoimmunization and Autoimmunization and the Formation . . . . . . . . . of Autoantibodies . V. Experimental Autoimmune Disease of the Thyroid: the . . . . . Thyroid Gland as Source and Target . A . Induction of Autoantibodies and of Autoimmune Disease. . . . . . . . . Especially in the Rabbit B. Genetic Factors in Experimental Thyroiditis . . . . C . Delayed Hypersensitivity and Cellular Immune Responses D . Additional Animal Models . . . . . . . . VI. Human Autoimmune Disease of the Thyroid . . . . 4 Thyroiditis and Other Thyroid Diseases . . . . . B. Autoantigens . . . . . . . . . . . C . Genetic Factors in Human Thyroiditis . . . . . D . Serological Overlap with Diseases of Other Organs . . E . Autoimmunogenicity of Thyroglobulin . . . . . VII . Features of the Autoimmune Response . . . . . A . Distinctive Types of Antibody . . . . . . . B. Mechanisms in Autoimmunity . . . . . . . VIII . Chemical and Antigenic Structures of the Thyroglobulin Molecule . . A. Physicochemical and Biochemical Characterization B. Subunit Structures of the Molecule . . . . . . . . . . . C. Antigenic Structures of the Molecule IX . Concluding Remarks . . . . . . . . . . References . . . . . . . . . . . .
.
.
. . . . . . . . . .
85 87 87 90 93 93 99 107 107 109
.
112
. . . . . . . .
114
. . . .
.
. .
.
. .
. . .
114 121 124 128 132 133 135 138 140 141 142 142 151 155 155 159 166 170 173
.
I Introduction
The study of the thyroid antigens is interesting from a number of points of view. First. these antigens have served as a model system for the development of concepts and methods in the general area of tissue (or organ) specificity. Historically. the thyroid system of antigens pro85
86
SIDNEY SHULMAN
vided the first example wherein tissue specificity was considered to depend on protein substances. Earlier studies of tissue specificity, such as those of the lens and of the brain, had focused on lipid substances, although it was later realized that proteins were also involved in these examples ( Witebsky, 1929; Shulman, 1971b). Second, some of these antigens are related to the important phenomenon of autoimmunity to the thyroid gland. In this regard, the thyroid model was not the first to stimulate active interest and experimentation, although it did seem more closely related to definable human diseases than did some other systems, such as those involving brain or testis. The possibility of near-immediate clinical application gave encouragement to the exploration of autoantibodies to thyroid components and of the autoimmune responses under various conditions ( Doniach and Roitt, 1957, 1962, 1968; Witebsky and Rose, 1963). In addition to these primarily immunological points of interest, the important endocrinological characteristics of the thyroid attracted a great deal of attention-especially because disorders of thyroid function are significant in a variety of human diseases. It thus seemed important to study the properties of all the macromolecular constituents of the thyroid gland, by various means, including immunological ones. Biochemical aspects of the gland received much attention, especially as regards the manufacture and elaboration of the iodine-containing horm o n e ( ~ ) In . these areas there are now a number of links between the strictly biochemical or endocrinological mechanisms and properties and the inimunological or immunochemical ones. It will be our goal here to review the major points of knowledge concerning the specific antigens, especially the autoantigens, of the thyroid gland, as well as the known proteins that seem to be characteristic of this organ. Some of the antigenic activities have not yet been characterized as definite proteins, or any other kind of macromolecule, and, on the other hand, some of the characteristic proteins have not yet been shown to possess any significant antigenic properties. In general, however, the important antigens can be better understood on the basis of chemical analyses of their macromolecular nature. The properties of autoantibodies and of antoimmune disease, as related to this gland, will be discussed in some detail. The experimental disease will be considered from a number of points of view; the clinical manifestations and immunological aspects in humans with the autoimmune disease will be summarized more briefly, since some recent reviews have dealt with this topic at length. The terms autoallergic disease and autoallergy have been preferred by some writers, and there is much to be said for this point of view. However, current usage in the literature reveals that the great majority of reports still use the term
THYROID ANTIGENS AND AUTOIMMUNITY
87
autoimmune, and this review will do so. A careful discussion of these distinctions has been presented by Paterson ( 1966), which can be consulted for any further interest. Earlier reviews on the subject of thyroid autoimmunity have appeared, and the following can be considered for other aspects or points of view: C. A. Owen (1958), Roitt and Doniach (lW), Witebsky (1964), Rose et al. (1965a), Glynn and Holborow (1965), and Rose and Witebsky (1968). In addition, autoimmunity to thyroid has constituted a portion of several interesting reviews on autoimmunity in general, and these too can be consulted: Waksman ( 1959), Paterson ( 1959), Milgrom and Witebsky (1962b), Kunkel and Tan ( 1964), and McMaster (1958). The present discussison will give prominence to interpretations that can be correlated with chemical structures and principles as they relate to the autoantigenic units and to the mechanisms of development of the autoimmune responses. II. The Thyroid Gland: Structure, Function, and Malfunction
Only a few general comments will be given here, in order to provide an orientation for the discussion of the antigens of this gland. For greater depth and detail in these particular areas, several review monographs will be found to be useful (Means et al., 1963; Pitt-Rivers and Trotter, 1964; Hazard and Smith, 1964), as well as a number of research reports, some of which will be cited.
A. HISTOLOGICAL STRUCTURE AND IODINE STORAGE
1. Cellular Arrangement and Ultrmtructure The thyroid gland is an encapsulated organ, constructed from a parenchyma built up of numerous follicles. Each follicle is a spherical mass formed from a single layer of cells, the follicular epithelium, surrounding a central acellular region which is rich in protein, called the colloid. The size of each follicle varies somewhat, as discussed by Wissig (1964), depending on its location within the gland, but the range of diameters is between 15 and 100 p, as measured in the rat. The epithelial cells themselves range in shape between low cuboidal and columnar. Although there are said to be cells other than those of the follicular epithelium, it has been difficult and controversial to identify such cells, and it may generally be taken that the solid clumps of cells seen in tissue sections are most likely to represent follicles that have been sectioned at their border and thus cut only through their epithelium. The colloid is invariably found to fill the entire lumen of each follicle and has been described as a thin, clear, homogeneous fluid. It can be
88
SIDNEY SHULMAN
stained in a number of ways and is rich in protein as well as in carbohydrate. The colloid has also been considered to contain protease activity and ribonuclease activity. Smeds (1970) studied the colloid from single follicles. Sisson (1968) analyzed the hyaluronic acid content of thyroid. Ultrastructural studies reveal a basement membrane that forms a continuous mantle around the entire thyroid follicle. Although there had been controversy on this point, more recent examinations of thin sections in the electron microscope have visualized this structure for several species. Irvine and Muir (1963a) have discussed this membrane in connection with the ultrastructural changes seen in Hashimoto thyroiditis. The epithelial cells of the thyroid follicle are essentially similar to other cells that engage in protein secretion. They show microvilli along their apical border. There is a well-developed Golgi apparatus and an extensive ergastroplasmic network distributed throughout the cell. The ergastoplasmic vesicles are said to be polymorphic and dilated and, thus, differ in appearance from those that characterize pancreas, liver, and salivary secreting cells; rather, they resemble the vesicles found in the coagulating gland of the mouse (Brandes et al., 1959) and the albumin-secreting cells of the hen’s oviduct ( Hendler et al., 1957). Each vesicle is limited by a single membrane, with ribosomes attached to the outer surface. Free ribosomes also are found in the cytoplasmic structure between vesicles. The vesicles contain a slightly dense material which seems to resemble colloid, and, therefore, it may be that this is the pathway of secretion into the storage depot of the colloid, but this kind of detail is not yet well established. The amount and the exact distribution of the ergastoplasm are influenced by the circulating level of thyrotropic hormone and accompany changes in total cell number and cellular height. Mitochondria are present and secretory droplets are found. Many further details of structure and ultrastructure have been reported and discussed, and these may be of significance with respect to localization of iodinating activity and of hormone-synthesizing activity and release as well as (for our purposes) the formation and possible breakdown of various characteristic and thyroid-specific antigens. Helpful reviews have been prepared by Wissig ( 1964), Klinck (1964), and Heimann ( 1966), and several recent papers have provided considerable detail (Irvine and Muir, 1963b; Lupulescu et al., 1968; Klinck et al., 1970; NBve et al., 1970).
2. Iodine Trapping The most characteristic and distinctive feature of the thyroid gland is its interaction with iodine. This element is incorporated into the
THYROID ANTIGENS AND AUTOIMMUNITY
89
thyroid hormones, producing several iodinated amino acids and some unusual proteins that are iodoproteins. The chemical form in which thyroid tissue takes up the iodine is of necessity as iodide. The entry of iodide into the thyroid tissue occurs either by diffusion or by transport, but active transport is undoubtedly the important mechanism; this has been thoroughly discussed by Halmi (1964). Transport of iodide can, in fact, occur in some other tissues, including stomach, salivary glands, mammary glands, placenta, small intestine, skin, ciliary body, and choroid plexus in mammals and in certain occasional structures in other life forms. These activities may help to provide the developing fetus or newborn with useful amounts of iodide or may help to prolong the stay of iodide in the body. However, only in the thyroid tissue is the uptake of iodide the first step in a series of reactions leading to the formation of thyroid hormone-thyroid hormone is not synthesized in other tissues that exhibit iodide uptake. The concentration of iodide in the thyroid can exceed that in the surrounding blood plasma by several hundred times. Presumably, there is a carrier substance which is probably situated in the membrane of thyroid cells and which complexes with inorganic iodide, mediating the entry and active transport of iodide into the tissue. The chemical nature of this carrier has not been fully established; it is presumably not of antigenic interest. Subsequent steps require some form of organification of the iodide which usually involves an interaction with tyrosine groups. The organification process consists of the conversion of inorganic iodide into the organic iodine of 3,5-diiodotyrosine. Formation of this amino acid is an intermediate step in the subsequent formation of the principal thyroid hormone, thyroxine, which is known to be a tetraiodinated aromatic ether structure with an alanine side chain. Two other substances of importance are monoiodotyrosine and 3,5,3'-triiodothyronine. This latter substance has been shown to have biological activity even greater than that of thyroxine, but it is generally present in much smaller amounts than is thyroxine in thyroid tissue. It is generally considered to be a thyroid hormone, along with thyroxine itself. The detailed molecular dimensions and three-dimensional structures of these iodinated amino acids have been depicted elsewhere (Shulman, 1963). The sequence of events culminating in the formation of these iodoamino acids in the thyroid gland can be summarized as follows (PittRivers and Cavalieri, 1964) : 1. Inorganic iodide is transported into the gland from the circulation and is trapped. 2. The inorganic iodide is converted by some oxidative system to produce monoiodotyrosine and diiodotyrosine. This is thought actually to
90
SIDNEY SHULMAN
occur in the tyrosine residues of the polypeptide structure that is the precursor of the thyroglobulin molecule, a protein to be discussed below. 3. Thyroxine is formed by the coupling of two molecules of diiodotyrosine, with suitable splitting out of a portion of one molecule. This step is considered to occur within the polypeptide structure. 4. Triiodothyronine is formed by the coupling of one molecule each of mono- and diiodotyrosine. As a result of this process, the thyroid hormones, thyroxine and triiodothyronine, first occur as special amino acids directly involved in the polypeptide structure of the protein, thyroglobulin. It is felt that at some later time the thyroglobulin molecule and, perhaps, other iodoproteins are broken down into the constituent amino acids (or perhaps intermediate peptide forms ). The hormonal structures, the iodoamino acids, are then secreted into the circulation. The iodotyrosines are deiodinated by a thyroid enzyme, and the iodide that is liberated reenters the biosynthetic pathways or is lost into the circulation. The precise cellular location of the iodination process has been much discussed in recent years. This is also true of the location and mechanism of the completion of biosynthesis of the thyroglobulin molecule. Salabe et al. (1969) have studied bovine thyroid polyribosomes and attempted to identify certain biosynthetic fragments of thyroglobulin by immunochemical means. De Nayer and De Visscher (1970) discussed the covalent association of thyroglobulin subunits in thyroid polyribosomal systems. The attachment of the carbohydrate portions to the polypeptide has been reported by Herscovics (1969, 1970), clarifying some of the details regarding the sequence of attachment of monosaccharide units.
B. SURVEY OF THYROID DISORDERS No extensive description of the medical or pathological aspects of abnormal functioning of the thyroid gland will be attempted here. For this goal, several specialized monographs (Means et at., 1963; Hazard and Smith, 1964; Pitt-Rivers and Trotter, 1964) can be profitably consulted, among other sources. It may be of value, nonetheless, to consider the major parameters of thyroid disease, since terminology in this area is very confusing to the nonspecialist. Indeed, the fine points of description and of diagnosis have led to disagreement among pathologists, so that at various times, different classifications of the spectrum of thyroid diseases have been published. There have been disagreements at times as to the most important criteria for establishing a certain diagnosis or as to the important points of similarity between related conditions. These diseases are generally grouped according to hyperor hypoactivity of this gland or according to a goitrous swelling, an
THYROID ANTIGENS AND AUTOIMMUNITY
91
idammatory condition ( thyroiditis) , or a tumorous state (carcinoma or adenoma) . The simplest breakdown of conditions woukl consist of ( I ) simple nonendemic goiter, (2) endemic goiter, ( 3 ) hyperthyroidism, ( 4 ) hypothyroidism, ( 5 ) nodules and malignancy, and (6) thyroiditis. The condition of simple nonendemic goiter has also been termed colloid goiter, simple struma, adolescent goiter, or nontoxic nodular goiter. It represents a parenchymal hypertrophy of the thyroid that is not dependent upon dietary inadequacy of iodine. Endemic goiter, on the other hand, is said to occur when there is a distinct enlargement of the thyroid gland that can be detected in more than 10%of the population, and assuming that other diagnoses are excluded. The state of hyperthyroidism would include thyrotoxicosis, toxic goiter, exophthalmic goiter, Graves’ disease, or Basedow’s disease. It is the general name for that group of signs and symptoms that arise when there is an excessive concentration of thyroid hormones in the blood. In all the forms of hyperthyroidism, there will be an increased rate of tissue metabolism and some disturbances of the neuromuscular system. Patients with Graves’ disease may also show certain abnormalities of the eye. The state of hypothyroidism represents an insufficiency of thyroid hormone in the blood. This condition may consist of several alternatives, including cretinism, juvenile myxedema, adult myxedema, and hypothyroidism without myxedema. Cretinism can include endemic cretinism, sporadic cretinism, metabolic cretinism, and the similar condition of Pendred’s syndrome. Hypothyroidism without myxedema may sometimes occur in the course of thyroiditis. The finding of thyroid nodules always raises the question of a possible condition of cancer. There are several forms of malignant disease of the thyroid. The condition of thyroiditis includes a number of pathological conditions of the thyroid gland. Among these, subacute or granulomatous thyroiditis is an inflammatory condition of the thyroid gland. Another, much rarer, condition is Riedel’s struma or ligneous thyroiditis. Of major interest is struma lymphomatosa or Hashimoto’s thyroiditis, which is a chronic disease that is manifested by progressive goiter that may occur with or without pressure symptoms. In histological terms, a characteristic atrophy of the parenchyma is seen, accompanied by a lymphocytic infiltration. This disease is much more commonly seen in women than men. It may come associated with other conditions, such as rheumatoid arthritis, Sjogren’s syndrome, or pernicious anemia. There are many other states of thyroid malfunction that could be listed as additional divisions or as subdivisions of those already men-
92
SIDNEY SHULMAN
tioned. A new classification of thyroid disease has appeared quite recently, based on the deliberations of the Committee on Nomenclature of the American Thyroid Association (Werner, 1!369). This report provided, in fact, both an abridged and a detailed version of the classification. The former version was suggested for simplicity and daily usefulness; the latter was provided for hospital diagnosis and for reports TABLE I CLASSIFICATION OF THYROID DISEASEO
I. Diseases primarily characterized by euthyroidism 1. Nontoxic diffuse goiter 2. Nontoxic uninodular goiter 3. Nontoxic multinodular goiter 4. Tumors Benign: adenoma or teratoma Malignant : carcinoma Primary Secondary Others 5. Acute thyroiditis Suppurative Nonsuppurative (subacute) 6. Chronic thyroiditis Lymphocytic (Hashimoto) Nonspecific Invasive fibrous (Riedel) Suppurative Nonsuppurative 7. Hemorrhage or infarction 8. Infiltration due to amyloid or hemochromatosis 9. Congenital anomaly 11. Diseases primarily characterized by hyperthyroidism 1. Toxic diffuse goiter (Graves’) 2. Toxic uninodular goiter 3. Toxic multinodular goiter 4. Nodular goiter with hyperthyroidism due to exogenous iodine 5. Exogenous thyroid hormone excess 6. Tumors 111. Diseases primarily characterized by hypothyroidism 1. Idiopathic myxedema 2. Cretinism Endemic Congenital goitrous 3. Thyrotropin deficiency 4. Thyroid-releasing factor deficiency due to hypothalamic injury or disease 5. Thyroid destruction 6. Congenital aplasia 0
Modified from Werner, 1969.
THYROID ANTIGENS AND AUTOIMMUNITY
93
in the literature. Table I shows a slightly modified version of the abridged tabulation from this report. 111. Purification and Properties of Thyroid Proteins
A. ANALYSIS OF PROTEIN CONSTITUENTS IN TISSUE EXTRACTS Study of the thyroid proteins began in the 1890s with several reports that are now only of archival interest, such as Gourlay (1894) and Hutchison (1896). The most memorable work of this period was that of Oswald (1899), who isolated the major protein substance from the thyroid and demonstrated its globulin properties. He gave it the name “thyroglobulin” ( Thyreoglobulin in German). He prepared the thyroglobulin from saline extracts of thyroid tissue by the use of the then very popular method of ammonium sulfate precipitation, collecting most of the protein between the levels of 26 and 44%of saturation. Some additional protein could be brought down by the addition of saturated salt to the supernatant fluid. Essentially all the iodine was found in the first product, and because this iodinated protein was so easily extracted in high yield, it was believed to come from the colloid of the gland. Because of the prevalent definition that a globulin is a protein which is precipitable by 50%saturation with ammonium sulfate, the name was appropriate. Heidelberger and Palmer (1933), using a different approach, applied a preliminary step of adjusting the pH to 5.0 in the cold, precipitating a product that showed the properties of nucleoprotein. After removal of this product and neutralization of the solution, sodium sulfate at half-saturation and at 35°C was used for salting-out the fraction of interest. This material was considered to be a purified thyroglobulin, although only a rather limited analytical comparison was made between the product and the original crude extract. The physicochemical properties of this hog thyroglobulin were then studied by Heidelberger and Pedersen (1935), who found the sedimentation coefficient to be 19.2s. From a measured partial specific volume of 0.72 ~ m . ~ / g mand . a diffusion coefficient of 2.39 x cm.zsecond-l, they obtained a molecular weight of 700,000. In addition, they employed equilibrium ultracentrifugation and obtained a molecular weight of 675,000. Thus were the first ideas established on the molecular weight of this thyroidal protein (and on its being a 19 S protein). Further details on physicochemical data will be given below. Subsequent studies on the major thyroid proteins seen in extracts were made by Shulman et al. (1955). They prepared saline extracts of hog thyroid glands, using mild procedures in order to avoid denaturation of any proteins. Analysis in the ultracentrifuge revealed the presence
94
SIDNEY SHULMAN
FIG. 1. Ultracentrifugal patterns of a hog thyroid extract. Upper line: protein concentration, 2.0%;photographs at 8, 26, and 44 minutes after rotor reached speed of 58,200 r.p.m. Lower line: protein concentration, 0.8%;photographs at 8, 26, and 34 minutes after rotor reached speed of 59,700 r.p.m. Direction of sedimentation is to the right. (From Shulman and Witebsky, 1960a.)
of five sedimenting components in these crude extracts. A set of characteristic pictures for hog thyroid extract is shown in Fig. 1. The sedimentation coefficients were extrapolated to infinite dilution and were then found to equal the following values: I, 29.5s; 11, 18.7s; 111, 11.3S; IV, 6.5 S; V, 4.2s. A composite graph of these sedimentation rates is shown in Fig. 2. A considerable degree of concentration dependence is seen for components I and 11, whereas the other coniponents show no such dependence. The 18.7s component was considered to be the same as the classic thyroglobulin, in part because of its predominant quantity in the total pattern and in part because of the identification of thyroglobulin as a 1 9 s component by Heidelberger and Pedersen (1935). The other four peaks represented new components. It was now observed that the 18.7s component constituted approximately 80% of the total protein in the mixture. The proportion of Component I (or F, for fast peak) was S10%of the total and that of Components IV plus V (or S, for slow peak) was 10-15%. Peak I11 could not always be seen, and its
95
THYROID ANTIGENS AND AUTOIMMUNITY
-
a
0 G
+ o ;n
?
0
-
Y
-
Concentration of Protein (grn./lOOml.)
FIG. 2. Sedimentation coefficient plotted against protein concentration for thirteen runs from three preparation of hog thyroid extract. (From Shulman et d., 1955.)
contribution was only, at most, about 1 or 2% of the total. The ratio of the amounts of peak V to peak IV was generally about 5. Table I1 shows the proportions in a typical set of extracts. Although it was suspected that some of the material composing peaks IV and V included serum globulin and albumin, respectively, it was shown quite definitely that serum proteins did not constitute the total composition of these two ultracentrifugal components. In this same study, by analysis of perfused glands, the presence of proteins of thyroidal origin in these boundaries TABLE I1 RELATIVEPROPORTIONS OF THE SEDIMENTING COMPONENTS OF HOGTHYROID CRUDE EXTRACT5 Proportions (%) Component T g (Peak 11) F (Peak I ) S (Peaks IV Total Peak IV Peak V V/IV a b
+ V)
4.46
Concentration of extract (gm./dl.): 2.23 0.89 0.45
84.0 2.1 13.6 99.7
81.9 4.3 13.6 99.8
81.2 5.6 13.6 100.4
79.7 6.6 13.6 99.9
2.2 11.4 5.2
-
-
-
-
From Shulman el al., 1955. Values in parentheses were obtained by extrapolation.
-
O.O@ (79.8) (6.7) (13.6)
96
SIDNEY SHULMAN
was clearly established. Each of these slower sedimentating peaks corresponds, in fact, to a mixture of proteins, and evidence of this polydispersity will be presented below. Components I and 111 have become generally known in the more recent literature as 27 and 12 S , respectively. As seen by means of filter paper electrophoresis (Shulman et al., 1955) or by moving boundary electrophoresis (Shulman and Witebsky, 1960a), one component predominated in the total extract, and this was seen to travel as a serum a-globulin. Figure 3 shows a typical pattern. One peak, B, was 924%of the total, and it had a mobility in barbital buffer, pH 8.5, ionic strength 0.10, of -4.8 x lc5cm.2V.-1second-1;this boundary clearly represented the thyroglobulin, along with additional minor components that were not resolved in electrophoresis. There were also three very small peaks (A, C, and D). Resolution was very difficult because the use of more concentrated preparations involved greater problems of light transmission through the darkly colored solutions. Witebsky et al. (1956) also reported that ultracentrifugal examination of the extracts of normal human thyroid tissue revealed patterns very similar to those that had been seen for the thyroid tissue from the hog species. The same five major peaks were observed with approxi-
FIG.3. Electrophoretic pattern (descending channel) of a hog thyroid extract. Protein concentration, 0.65%;barbital buffer, pH 8.60; ionic strength, 0.10; electric field strength, 6.8 volts/cm.; time, 180 minutes. (From Shulman and Witebsky, 1960a.)
97
THYROID ANTIGENS AND AUTOIMMUNITY
FIG.4. Ultracentrifugal patterns of an extract of normal human thyroid. Protein concentration: 1.6% Photographs at 0, 8, 16, 24, 40, and 48 minutes after attaining speed of 59,700 r.p.m. Direction of sedimentation is to the right. (From Witebsky et al., 1956.)
mately the same distributions, augmented by a sixth component which was still faster than the others; these were, therefore, labeled If, I, 11, 111, IV, and V. Figure 4 illustrates this kind of schlieren pattern. The sedimentation coefficients and percentages are indicated in Table 111. The main peak, component 11, again had the sedimentation characteristics of thyroglobulin. In an extract from a human cancer of the thyroid, on the other hand, the pattern was completely different and, thus, drastically different proportions of these components were indicated. In particular, the thyroglobulin component seemed to be reduced in quantity by a factor of at least 10. The data are listed in Table IV. We shall return below to this matter of abnormal proportions in some kinds of abnormal thyroid glands. The electrophoretic pattern of normal human thyroid extract revealed the presence of four components, one of which constituted 62% of the total and should, therefore, be identified with the thyroglobulin. In the cancer extract, a totally different pattern was seen, composed of nine components which were poorly resolved. The original report can be consulted for additional details and pictures ( Witebsky et al., 1956). Ultracentrifugal patterns for human thyroid extract were also shown by DeGroot and Carvalho ( 1960), by Lob0 et al. ( 1966), and by Thomson and Bissett (1969).
ULTRACENTRIFUGAL
TABLE I11 ANALYSISOF NORMAL HUMANTHYROID EXTRACT' Peak
Sedimentation parameter
If
I
I1
I11
IV
V
Coefficient, spa+ Proportion, yo
-70. 7.5
29.3 10.9
19.5 58.3
10.8 1.8
6.4 4.4
4.1 17.1
(I
From Witebsky et al., 1956.
98
SIDNEY SHULMAN
ULTRACENTRIFUGAL
TABLE IV ANALYSISO F MALIGNANT HUMANTHYROID EXTRACT^ Peak
Sedimentation parameter
CI
CII"
CII'
CII
CIII
CIV
78.0 2.8
-
-
17.4 4.9
6.0 18.3
3.8 74.0
68.0 6.0
-44. 3.2
-31. 2.6
18.5 4.3
6.1 9.4
4.3 74.5
Primary tumor: s!o.w
% Metas task : s2oo.w
% ~
~~
~
From Witebsky et al., 1956.
The proportions of components in extracts from human carcinomas or other tumors have attracted widespread interest. In general, it has been found that there is a very high proportion of a 4 s component and a relatively low proportion of the 18-20s component. There have been concomitant efforts to evaluate the degree to which these two components are identical, or at least highly similar, to those components of normal thyroid tissue which have the same sedimentation coefficients. The usual criterion has been the iodine content, although salting-out solubility studies have also been made. One interesting investigation is that of Stanley (1964a,b) who studied salting-out curves of the extracts prepared individually from twenty-one human thyroid glands, both normal and pathological. He showed that there was a good correlation between the presence of 19 S components and a fraction that salted out between 35 and 45% saturation of ammonium sulfate and that some of the 4 s material was salted out between 20 and 30%saturation with ammonium sulfate. This 4 S component cannot be considered equivalent to hog thyralbumin, to be described below, since the latter was found to be salted out at much higher levels of ammonium sulfate. Furthermore, the data on proportions of components indicated that in various pathological states the thyroid gland was not as able to synthesize thyroglobulin as in the normal state. Similar descriptions have been reported by Valenta et al. ( 1 9 6 8 ~ ) . The ultracentrifugal components of thyroid extracts from several other species have also been described, Rat thyroid tissue has been studied by several workers, and a pattern has been described which is very similar to that seen with the hog thyroid extract. Robbins et al. (1959a) and Wolff et al. ( 1959) published ultracentrifugal patterns which revealed sedimenting components that were 27, 18-20, 10, and 6 or 7s. These examinations were actually made with a transplantable
THYROID ANTIGENS AND AUTOIMMUNITY
99
thyroid tumor of the rat. Therefore, the proportions of these components were not similar to those of normal hog or human tissue, but they did, in fact, resemble human cancer extracts, described above. There was a conspicuous 4 S iodinated component which was termed “thyroid S-1 iodoprotein,” and which was, therefore, similar in properties to hog thyralbumin, to be described below. There have also been studies on the proteins of sheep thyroid tissue (Robbins et al., 1959b), and the patterns were reported to be similar to those for the human. Similar patterns and sedimentation coefficients were reported and studied in great detail by Edelhoch (1960) for thyroid proteins from cattle.
B. FRACTIONATTON EFFORTS AND RESULTS 1. Salting-out: Thyroglobulin and Thyralbumin Derrien and co-workers (1948, 1949) described the purification of hog thyroglobulin and the physical and chemical properties of this material. They employed several systematic procedures of salting-out, using several different kinds of salt, but ending with a preferred selection of phosphates at a total concentration of 3.5M as a stock solution. They found, by constructing the entire solubility curve in considerable detail over a wide range of salt concentrations, that approximately six components could be detected in the crude extract, although three of these were quite minor. The three major components were quite similar in their salting-out properties and were considered to be three fractions of thyroglobulin itself, because the content of iodine was identical in these constituents. It seemed quite reasonable to these workers to consider that the heterogeneity, as it might be called, of the thyroglobulin material in these three forms was probably an indication of diverse states of association of a single protein. This was considered true even though the preparations were homogeneous in ultracentrifugal and electrophoretic characterizations. They also emphasized that the iodine content of pure thyroglobulin is quite variable among. different preparations of this protein and that, therefore, the estimation of this property is no criterion of purity, It was proposed that thyroglobulin is probably a protein of rather constant structure and composition insofar as the noniodinated amino acids are concerned but that the degree of iodination in the synthesized protein may well vary with different physiological and nutritional conditions. They found that purified hog thyroglobulin had an extrapolated sedimentation coefficient of 19.4 S and a diffusion coefficient of 2.6 x lo-’ cm.2second-1. The partial specific volume was determined as being 0.72 ~ m . ~ / g mBy. combining these three values a sedimentation-diffusion molecular weight of 650,000 was determined.
100
SIDNEY SHULMAN
The formation of the thyroglobulin polypeptide and its iodination to give the protein product were affirmed by these workers to be two independent and successive processes, as had been earlier proposed by Harington (1944) and by Morton and Chaikoff (1943). Fractional precipitation studies, using ammonium sulfate, were reported over the years from 1956 to 1962 (Shulman, 1956; Shulman et al., 1957; Shulman and Witebsky, 1960a, 1962), and these led to the isolation of the major globulin and albumin fractions of hog thyroid extract. The optimal levels of precipitating salt were found to be 1.60-1.70 M for the preparation of thyroglobulin and 2.50-3.00 M for the preparation of a new fraction that was given the name “thyrall?umin.”Each fraction seemed quite highly purified according to ultracentrifugal analysis. The thyroglobulin was about 97% homogeneous and the thyralbumin was 100%homogeneous in ultracentrifugal analysis, the latter being the 4 S component or peak V of crude extract. The yield of thyralbumin was approximately 1%of the original protein and the yield of thyroglobulin was approximately 10 to 15%of the original total protein. Electrophoretic analysis of hog thyralbumin preparations revealed that the material is, in fact, quite heterogeneous, with a total of seven or eight migrating components; this type of preparation contains approximately 0.10% iodine, although the distribution is not yet known (Shulman, 1968, 1971b,d). It would, however, seem to correspond well to the iodoproteins of 4 s class that have been isolated from thyroid tissue by other investigators (Tata et al., 1956; Wolff et al., 1959; Robbins et d.,1959a; Beckers and De Visscher, 1M1, 1963; Ramagopal et al., 1965; Roitt et al., 1965). Some comparisons will be discussed below. Further studies on the purification of hog thyroglobulin were carried out by Ui and Tarutani (1961) and Ui et al. (1961), who showed that thyroglobulin purified by an ammonium sulfate procedure, using 1.55-1.75 M salt at low temperature (or 1.50-1.70 M at 2OoC), still contained a 28 S (or F ) component in addition to the major 19 S component, which was about 91% of the total. Chromatography on diethylaminoethyl (DEAE)-cellulose was also done, as will be mentioned below.
2. Diferential Centrifugation A new method of purification, based on several variations of preparative ultracentrifugation, was introduced by Edelhoch ( 1960). In a first procedure, the thyroglobulin of calf thyroid saline extract was precipitated by salting out with a 3.5M potassium phosphate buffer, pH 6.6. The material precipitating between 41 and 48% saturation was collected and then centrifuged for 260 minutes at 40,OOO r.p.m. at room
THYROID ANTIGENS AND AUTOIMMUNITY
101
temperature in the Spinco Model L ultracentrifuge. After this time the top portion, approximately 70%,of the centrifuged solution was pipetted off and the bottom portion was decanted as the product, leaving the packed pellet behind. This process was repeated several times and a product was obtained which showed a high degree of ultracentrifugal homogeneity. In an alternative procedure, the differential centrifugation technique was applied directly to the saline extract of the thyroid tissue.
3. Chromatography and Gel Filtration Thyroglobulin was prepared from sheep thyroid glands by Ingbar et al. (1959), using gradient elution chromatography on the anion exchanger, DEAE-cellulose. They observed a distribution of the eluted material which was essentially a single peak with an asymmetrical trailing region. Although the ultracentrifugal studies were carried out on rather dilute solutions, thus limiting the critical analysis of homogeneity of each pooled product, each material showed a 1 9 s component, and the trailing material showed an additional faster component. They concluded, especially from the iodine distribution, that the thyroglobulin is heterogeneous-even in the case of the thyroglobulin obtained from a single animal. Roche et al. (1960) examined the heterogeneity of hog thyroglobulin also by using DEAE-cellulose. The distribution of protein in the eluate and the concomitant distribution of radioactively labeled iodine led them to conclude that there was a heterogeneity of the thyroglobulin itself, in addition to the observed multiplicity of proteins that were seen in the total extract. Shulman and Stanley (1961), following Shulman and Witebsky ( 1960a), applied chromatographic methods to hog thyroid extract, using DEAE-cellulose with a gradient elution procedure. The concentration of the 1 9 s component in the best final product was 96% The content of iodinated amino acids, at various points in the effluent, showed a ratio of iodotyrosines to iodothyronines that steadily increased, ranging from 1.9 in the earliest eluted sample up to 8.8 in the last eluted sample. This again indicated a distribution of molecules differing in their iodine content. Ui and Tarutani (1961) and Ui et al. (1961) used DEAE-cellulose and stepwise elution with buffers of increasing ionic strength and a constant pH of 6.5. Some of the fractions showed a high degree of homogeneity of the 19 S material in the ultracentrifuge. The fraction that was eluted at an ionic strength of 0.22, following a level of 0.15, was found to have the highest purity of 1 9 s material and to be essentially homogeneous in the ultracentrifuge. In some of the fractions
102
SIDNEY SHULMAN
there was an enrichment of the 28 S component (which they called the Fcomponent), to levels as high as 62% of the protein. This fast component was accompanied by a still faster one, although sedimentation rates were not given. It was interpreted from the iodine analysis of the various mixtures that the 28 S component would itself be an iodoprotein. Chemical analyses of the various chromatographic fractions showed an increasing ratio of iodine to nitrogen in the sequential fractions eluted from the column, and this again gave support to the concept that there are a number of iodoproteins, differing in the degree of iodination, but apparently being very similar in sedimentation properties, belonging either to the 19 or 28 S families. Later studies by Shulman and Armenia (1963), on the purification of hog thyroid proteins, employed DEAE-cellulose in a stepwise elution technique, by which means the thyroid tissue extract was separated into five major fractions or elution peaks. Thyroglobulin was found in its most homogeneous form in peak 3 of the effluent pattern, illustrated in Fig. 5, and it was found possible to obtain this protein in an ultracentrifugal homogeneity of 99 to 100%and with a yield of 7 to 8%of the original protein total. About 35 to 40%of the material could be obtained with homogeneity above 95%. The homogeneity of this thyroglobulin preparation of peak 3 was also demonstrated by immunochem-
FIG. 5. Elution pattern from diethylaminoethyl-cellulose chromatography of a hog thyroid extract. Column size was 2.2 X 25 cm. (12 gm.). The sample was applied as 1.38 gm. in 0.0175 M sodium phosphate buffer, pH 7.5, to a column of 2.2 X 25 cm., and a stepwise gradient was employed, introducing the successive buffers, as indicated. An additional elution was performed .with 0.10 N NaOH. The short vertical lines indicate the tubes which were pooled to form the numbered peaks. (From Shulman and Armenia, 1963.)
THYROID ANTIGENS AND AUTOIMMUNITY
103
ical means. The crucial test was made with an antiserum to hog serum, in order to detect the serum contaminants in the thyroid preparations, as tested in gel diffusion precipitation. Such an antiserum could show several lines of precipitation when set against a sample of hog serum, and, although it also showed at least one line of precipitation when set against the crude extract of hog thyroid, it showed a complete absence of such lines of precipitation when tested against a sample of the highpurity thyroglobulin material. This indicated the essentially complete removal of at least the serum contaminants from the initial crude extract, at least down to 1/1O,OOO of their initial level. Investigations of the fractionation of thyroid proteins have developed in recent years on the basis of several gel filtration procedures. These studies have been conducted with the use of either of two media: one of these is dextran gel (more commonly known as Sephadex) and the other is granulated agar (or agarose). For example, Perelmutter et at. (1963) applied bovine thyroid extract to a column of Sephadex G-200 and achieved a good separation of the major components. They obtained a preparation of thyroglobulin that was described as 100%pure (Perelmutter and Stephenson, 1964). Similar methods were used by Mouriz and Stanbury ( 1967), Cheng et al. (1968), and Van Zyl et at. (1969) for preparation of pure thyroglobulin. Salvatore et al. (1964) also applied such gel filtration procedures to rat, guinea pig, cattle, and sheep materials, obtaining similar results, although few details were given concerning the slower-sedimenting proteins. They did, however, introduce a greatly improved procedure for the isolation of the 19 and 2 7 s materials, namely, the method of agar gel filtration, using thus a resin with a much higher range of exclusion values than can be obtained with Sephadex. Some important additional studies were reported by Salvatore and his colleagues (1965) and by Vecchio et al. (1966) on the purification and properties of the 2 7 s protein. By using human and bovine tissue, they showed that this component could be prepared in highly purified form by filtration through a granulated 5%agar gel or by the method of ultracentrifugation in a linear density gradient of sucrose. The sedimentation patterns for these two proteins are shown in Fig. 6, which is taken from their work. They could then demonstrate that this 27 S substance did contain iodine, establishing it definitely as another iodoprotein of the thyroid gland. In fact, the 2 7 s protein usually had a higher iodine content than did the 19 S (thyroglobulin) protein obtained from the same thyroid tissue; in the human preparations, for example, the levels were 1.0 and 0.7%, respectively. The heavier iodoprotein was now found to have an extrapolated sedimentation coefficient of 27.0 S, a diffusion coefficient of
104
SIDNEY SHULMAN
FIG. 8. Ultracentrifugal analysis of purified human 2 7 s iodoprotein and 19s thyroglobulin. Lower pattern (standard cell) ; 27 S iodoprotein preparation. Upper pattern ( wedge-window cell) : thyroglobulin purified from the same gland. Protein concentration, 0.7%.Photograph at 12 minutes after rotor reached speed of 52,840 r.p.m.; double sector cells. Direction of sedimentation is to the right. (From Salvatore et d.,1965.)
1.91 x 10-7cm.2second-1,and a calculated molecular weight of 1,200,000. It also showed similarity to the 1 9 s protein in amino acid composition and in certain antigenic determinants. For these various reasons, these investigators concluded that the 2 7 s protein is probably a dimeric form of thyroglobulin, although the exact relationship was not fully clear. Building on these observations, Tarutani et aZ. (1971a) studied the breakdown of the human 2 7 s thyroidal protein in the presence of sodium dodecyl sulfate. They interpreted their observations in terms
THYROID ANTIGENS AND AUTOIMMUNITY
105
of two forms of thyroglobulin, which differ in their stability toward dissociation, and considered the 2 7 s protein to be a mixture of forms made from combinations of dissociable or nondissociable thyroglobulin molecules paired with either the same or differing molecules. An extensive study, aimed at obtaining a full range of proteins, has been made of the separation of the proteins in human thyroid extracts, using gel filtration on Sephadex G-200, in which it was found that three major fractions could be eluted (Shulman et al., 1967). Samples from the first of these fractions could be pooled to give preparations containing thyroglobulin that was about 85% pure according to ultracentrifugal analysis; the separation of the 2 7 s component from the 1 9 s thyroglobulin was rather poor, and this was also true for the 1 2 s component. However, the 7 and 4 s comporients were well separated into the two slower-eluting fractions. It was then found that each of these two major fractions could be subdivided by means of zone electrophoresis on a paper curtain into two approximately equal fractions. It was, thus, discovered that the initial thyroid extract could be separated into two 7 and two 4 s subfractions. In addition, there was evidence for the existence of 12, 19, 27, and 34 S components. Hence, there were displayed a total of eight distinctive protein fractions, some of which seem to be homogeneous components, whereas others are still mixtures. Components 33 and 38 S are mentioned by Salvatore et al. (1965); thus, nine component fractions can be considered.
4. The Thyroid Family of Proteins When all the observations of the past decade relating to thyroid proteins are analyzed, it can be seen that there are several distinctive iodoproteins derived from thyroid tissue and there are additional proteins seemingly characteristic of the thyroid, although they apparently lack iodine. The iodoproteins include, in addition to the 19 and 27 S proteins, the 12 S protein and at least some portion of the 4 S group of proteins. Also, a 31 S protein (presumably the same as the 34 S mentioned earlier) has been described as being possibly iodinated (Robbins et al., 1966). Recent studies on highly purified human thyroid proteins have given extrapolated values of 27.7 and 32.3 S for the 27 S and next heavier proteins (D. J. Smith and Shulman, 1971b). The 1 2 s component has been accepted as being a molecule that corresponds to one-half of the thyroglobulin molecule, that is, it is considered as a subunit of the 1 9 s molecule, following the initial work of Edelhoch (1960). A number of reports have appeared on the 12 S fraction as a “short-life” precursor of the 19 S component during biosynthesis, as well as on descriptions of “stable” 1 2 s components; some of
106
SIDNEY SHULMAN
these distinctions seem to depend to a large extent on the animal species involved. Many of these relationships have been discussed by Roche et al. (1968), as part of an extensive examination and comparison of these iodoproteins in a variety of vertebrates. Interesting examples were found concerning subunits and their degrees of stability, as illustrated in the lamprey by Aloj et al. (1967) and in the guinea pig by Salvatore et al. (1967). With regard to the 4 S proteins, the two major subfractions reported above for human material can be compared to the electrophoretically polydisperse hog thyralbumin described above ( Section 111,BJ ) , as well as to the analogous protein, the S-1 iodoprotein, isolated from rat thyroid tumors by Robbins et al. ( 1959a) and Wolff et al. (1959). A 4 S thyroid protein was also isolated from normal rat thyroid extracts by Roitt et al. ( 1965), using sucrose density-gradient ultracentrifugation. In addition, a 4 s iodinated protein, similar to serum albumin in some properties, has been reported in the serum of humans that suffer from a variety of thyroid diseases. This protein resembles serum albumin in electrophoretic and solubility properties. One report described it to be immunologically identical to human serum albumin ( Stanbury and Janssen, 1962; Ramagopal et al., 1965), whereas a second report found that it was unreactive with antiserum to human serum albumin (Tata et al., 1956). It may have been that different members of the 4 S family of proteins were studied in these investigations. Many of the patients that have this 4 S iodoprotein in the serum also have an unusually large proportion of 4 S iodoprotein in their thyroid tissue. As mentioned above, electrophoretic studies of human thyroid cancer extracts revealed three to five poorly resolved components, all of which were associated with a 4 S ultracentrifugal component ( Witebsky et al., 1956). It was also found by Beckers and De Visscher (1961, 1963) that noniodinated proteins of the 4 s class occurred in certain human thyroid extracts; these were termed prethyroglobulin because of their electrophoretic mobility. Thyralbumin was also studied by Jonckheer (1963) and by Jonckheer et al. (1968). More recently, Jonckheer and his colleagues (Jonckheer and Karcher, 1971; Otten et al., 1971) have isolated a wellpurified preparation of this substance (termed by them “thyroid albumin”) from normal human thyroid glands and reported some of its chemical and physical properties. They concluded that this thyralbumin is an iodinated protein different from serum albumin and that it is synthesized within the thyroid tissue. Lissitzky and his colleagues (1964) have also described certain iodoproteins that were not thyroglobulin. They described an iodoprealbumin from both the thyroid tissue and blood serum of a patient with
THYROD ANTIGENS AND AUTOIMMUNITY
107
a eumetabolic goiter. They have also described an iodoalbumin from the tissue of congenital goiters and concluded that the initial source of this protein was the serum, from where it entered the thyroid tissue and was iodinated (Lissitzky et al., 1967, 1968a). The relationship of these components to the thyralbumin of hog or the two major 4 s fractions of human thyroid tissues has yet to be clarified. Stanley ( 1956) described solubility differences in the iodine-containing proteins from normal and abnormal thyroid glands. Abnormal human iodoproteins were also studied by Dowling et al. (1!36l), ToroGoyco and Matos ( 1965), Thomson and McGirr (1969), and Kivikangas et al. ( 1970). Iodoproteins, including iodoalbumins, have been described for a congenital goiter of sheep (Falconer et al., 1970; Furth et al., 1970). IV. Thyroid Antigens
A. TISSUE SPECIFICITY AND ORGAN SPECIFICITYA SEMANTICPROBLEM The term tissue specificity refers to the property of antigenic distinctiveness of one particular tissue, as compared to other tissues of the same body. A tissue-specific antigen, then, is an antigen that is characteristic for one particular tissue. From a historical standpoint, tissuespecific antigens were first found in the lens of the eye and in the brain, to be followed by those of the thyroid. These, and many other such antigens have been discussed in some detail by Dumonde (1966) and by Shulman (1971b). In the early studies, Fleisher and his collaborators (1920; Fleisher and Arnstein, 1921; Fleisher, 1922) worked with rabbit antisera to guinea pig liver, kidney, muscle, spleen, testis, and brain; they found that several antisera reacted best with the homologous organs but showed strong cross-reactions with other guinea pig organ extracts. By elaborate absorption procedures, they were able to identify three types of antigen: ( 1 ) species-specific antigens present in all organs of the animal, (2) tissue-specific antigens characteristic of the organ, and (3) antigens possessed in common by several different organs. Hektoen and Schulhof (1923) and Hektoen et al. (1927) identified thyroglobulin as a tissue-specific antigen that carries tissue specificity and also a limited species specificity. This kind of species specificity was further explored in the studies of Rose and Witebsky (1955); they showed, to take one example, that an antiserum against human thyroid extract, absorbed with bovine or porcine thyroid extract, would combine with thyroid extracts only from the human species. Unabsorbed, how-
108
SIDNEY SHULMAN
ever, the antiserum would show various species interactions, as will be further discussed below. With regard to terminology, we must recognize that the alternative term organ specificity has often been used to indicate an antigenic property that is characteristic of a particular organ of the body and is, in fact, characteristic of this organ regardless of the species. The most notable examples, classically, have been those of lens and brain. In both of these organs, one could demonstrate an antigenic similarity in the corresponding organ in many different species, and this was, therefore, characteristic of the organ rather than of the species. Some investigators had termed this organ specificity of the second order. There is another type of organ-specific antigen which is limited to the species. One example is that of the red blood cells of the horse, in that an alcoholic extract of horse red blood cells has a unique antigen that is not shared with any other tissue extract of the species nor with any red cell preparation from unrelated species (Landsteiner and van der Scheer, 1925). This was termed by some early workers, organ specificity of the first order or tissue specificity. It can be defined as the presence of an antigen which is characteristic of a particular organ within a single species (Fleisher et al., 1920; Fleisher and Arnstein, 1921; Fleisher, 1922; Landsteiner and van der Scheer, 1925). This property is, therefore, independent of the second dimension of organ specificity which measures the ability of the antiserum to react with corresponding organs from various animals of unrelated species. A good example of the contrast between the property of tissue specificity that is not restricted to the species and of tissue specificity that is restricted, is shown in the report of Witebsky and Milgrom (1962) on the characteristic autoantigen of adrenal tissue. These detailed terms, however, seem no longer to be used. Even though the early workers maintained some strict distinction in the use of the terms organ specificity and tissue specificity, it has grown more evident over the years that the terms are being used for rather interchangeable purposes and that there is no genuine usefulness in considering them to have distinctive meanings. Of these two terms, it is suggested that tissue specificity is preferable for the sake of precision and for a better clarity of definition. This is especially important for certain organs that contain two or more tissues. It should be emphasized that identification of tissue-specific antigens by serological procedures that were available at earlier times was usually possible only in those situations where the particular tissue extract contained strong tissue-specific antigens and relatively little contamination with serum proteins. Modern serological and immunochemical
THYROID ANTIGENS AND AUTOIMMUNITY
109
procedures, however, allow even the minor tissue-specific antigens in tissue extracts to be discovered. This may be illustrated by studies on adrenal specificity ( Milgrom and Witebsky, 1962a). Rabbits that were immunized with a suspension of bovine adrenal gland tissue developed antisera containing antibodies which combined, for example, in complement fixation tests, with extracts of all bovine tissues. One cduld reveal the adrenal specificity, in a direct observation, only by the use of double-diffusion gel precipitation tests. Certain antigens, each of which is characteristic for an organ, have been shown to be capable of eliciting the formation of autoantibodies. Usually this has been effected by means of immunization with tissue extracts obtained from the same species, that is, isoimmunization, and this has turned out to be a convenient procedure for producing tissuespecific antisera. As a result of immunization in this way, many antigens, and especially the serum proteins, cannot express themselves to any appreciable degree. Therefore immune sera are frequently obtained which are pure tissue-specific reagents, a goal that is reached with much more difficulty by immunization with tissue extracts from a foreign species.
B. ANTIGENIC PROPERTIES AS REVEALED BY HETEROIMMUNIZATION 1 . Species Distribution of Thyroid Antigens The earliest immunological studies on the thyroid were made by Hektoen and his colleagues. Hektoen and Shulhof (1923) studied thyroid extracts and also preparations of partially purified thyroglobulin. They prepared thyroglobulin from a variety of mammals, including bear, anteater, kangaroo, sloth, tapir, deer, fox, baboon, and zebra, in addition to the usual laboratory animals. They also prepared rabbit antisera to these thyroglobulin antigens, and studied the properties of specificity when tested with the various thyroglobulin preparations, making use of precipitation reactions (Hektoen et al., 1927). They observed that antisera that precipitated thyroglobulin did not precipitate the similarly prepared globulins from other organs of the body, and, on this basis, they considered thyroglobulin to be a tissue-specific antigen. The antisera to various mammalian thyroids gave cross-reactions with thyroglobulins isolated from a number of other mammals, but there was no cross-reaction with chicken thyroglobulin. The pattern of cross-reaction was rather puzzling; it was apparently not solely determined by the criterion of taxonomic closeness but seemed to occur rather randomly within the class of mammals.
110
SIDNEY SHULMAN
The work by Hektoen and his colleagues was confirmed, at least in principle, by Witebsky (1929) and by Adant and Spehl (1934). These workers emphasized that the immunological reactions obtained with heterologous preparations were very weak compared to the reactions with homologous preparations, as seen by both Precipitation and complement fixation methods. Somewhat later, quantitative precipitation methods were applied by Stokinger and Heidelberger (1937) in order to quantitate the degree of cross-reaction that could occur among the thyroglobulin antigens prepared from various mammals. More modern studies on tissue specificity were initiated in a series of investigations conducted by Witebsky and his collaborators. In the first report ( Witebsky et at., 1955), saline extracts of macerated thyroid glands from hogs, as well as other hog organs, were prepared as gently as possible, with the specific aim of avoiding denaturation of the proteins. Rabbits were immunized with these materials, using intravenous injection or intradermal injection with complete Freunds adjuvant. The intradermal adjuvant procedure was superior, because of the very small amount of extract needed for successful immunization and also because of the absence of the danger of anaphylactic reactions. Complement fixation and fluid precipitation tests were employed to detect antibodies in the resulting antisera. Although antibodies against hog serum were also elicited, these were usually of considerably lower titer than were the thyroid antibodies. In tests of specificity, the cross-reactions with other tissues disappeared on prolonged incubation at 37OC, whereas the reactions with the thyroid material were only slightly diminished. As an alternative, antisera absorbed with hog serum showed a lack of species-specific antibodies, but they continued to demonstrate potent levels of thyroid-specific antibodies. The cross-reactions of rabbit and dog antithyroid antisera with thyroid extracts of various animals were studied by Rose and Witebsky (1955). The absorption of an antiserum to human thyroid, for example, with hog thyroid extract eliminated the cross-reaction with hog thyroid and also cattle thyroid extracts, but it did not reduce the strong reaction with the homologous human thyroid extract. Cross-reactions between antithyroid antisera and thyroid extracts of various species were found to be thyroid-specific; in other words, no reactions were detected with other organs of the various animal species, including the homologous species. Several comparisons, taken from the work of Rose and Witebsky (1955), are shown in Table V. This summarizes the properties of rabbit antisera that were prepared against thyroid extracts of hog, cattle, man, and dog, and were then tested with saline extracts of thyroid glands
111
THYROID ANTIGENS AND AUTOIMMUNITY
TABLE V REACTIONS OF ANTITHYXOID ANTISERAWITH CRUDETHYROID EXTRACTS FROM VARIOUSSPECIES~.~ Crude saline extracts of thyroid glands from Rabbit antisera to thyroid extracts of Hog Hog Cattle Man
Dog
+ + f +
Cattle
Man
Dog Sheep Horse Cat
+ - + - + + - + + + r t + r t k + + + + + -
Guinea pig Rabbit
-
-
-
From Rose and Witebsky, 1955. Definite reaction in complement fixation and precipitation tests; (-) no definite reaction by complement fixation or precipitation tests; (k)disparity between different techniques or among different antisera. a
* (+)
from a large number of animals. It can be observed that reciprocal cross-reactions occurred between hog and cattle, hog and dog, cattle and dog, and dog and man. In addition, the antisera to cattle and dog gave cross-reactions with sheep and horse; and the antiserum to hog cross-reacted with horse. The various cross-reactions seemed to occur in a totally unpredictable manner. At this time, no reaction with rabbit thyroid was shown by any of these rabbit antisera. However, this observation was to be reversed in later studies on autoantibody formation. It was, thus, seen that antisera produced in rabbits by injection of crude extracts of thyroid glands will give cross-reactions with thyroid extracts of various other species in an inexplicable and irregular manner, in agreement with the results of Hektoen et al. (1927) who used isolated thyroglobulin preparations. Antisera produced by similar immunization of dogs contained thyroid-specific antibodies, but the cross-reactions with thyroid extracts of other species were considerably less prominent than in the case of rabbit antisera. The cross-reactions of thyroid antisera with thyroid extracts of other species were thyroid-specifi.,, nonetheless, fulfilling the definition of tissue specificity. 2. Diversity of Antibodies to Thyroid Antigens In later studies, Rose et al. (1960) clearly showed that rabbit antisera to various mammalian thyroid extracts contain, in fact, several kinds of specific thyroid antibodies. Two varieties could be easily distinguished, namely, those that reacted only with the homologous thyroid extract that was used for immunization and those that cross-reacted with thyroid extracts of other species. By means of double-diffusion techniques of agar precipitation, one of the two lines produced by the
112
SIDNEY SHULMAN
reactions of a hog thyroid antiserum with hog thyroid extract gave a reaction of partial identity with a single precipitation band produced by the same antiserum with the thyroid extracts of certain other species. It might be thought that two different antigens, both thyroid-specific, are involved. However, purified hog thyroglobulin sometimes gave two lines of precipitation in the gel diffusion experiments with the hog thyroid antiserum, even though, usually, one line was seen. Furthermore, absorption of the antiserum with thyroglobulin removed both thyroidspecific antibodies. In immunoelectrophoresis, only one line of precipitate formed, and this was located at the usual a-globulin site characteristic of thyroglobulin. From these and other findings, it was concluded that, most likely, the two thyroid-specific antigenic groups were on the same molecule, namely, thyroglobulin. The two types of thyroid-specific antibodies are, thus, directed against different antigenic sites on the thyroglobulin molecule. One of these groupings would be present on the thyroglobulin from many different species, whereas the other is to be found only on the homologous thyroglobulin molecule. Unfortunately, the detailed nature of these structures is still a mystery. Further discussion of molecular structure and the possibilities of clarification of the nature of the antigens is given later (Section VII1,C). 3. Substance Responsible for Thyroid Specificity Physicochemical methods were applied to the characterization of the major components of thyroid extracts, in order to establish the identity of the antigen or antigens responsible for the remarkable tissue specificity of this material. Thyroglobulin was identified as the actual component of the thyroid extract which is responsible for the property of tissue specificity (Shulman and Witebsky, 1962; Rose et al., 1962a). The salting-out procedures, using ammonium sulfate, which had been developed for the preparation of highly purified thyroglobulin and thyralbumin were utilized to prepare materials for serological testing of these two preparations. Most of the organ-specific thyroid antigen was, indeed, precipitated in the thyroglobulin fraction between the levels of 1.60 and 1.70 M ammonium sulfate, and very little species-specific antigen was precipitated in this fraction. The thyralbumin fraction, precipitated between 2.50 and 3.00 M ammonium sulfate, was found to contain little or no detectable thyroglobulin or thyroid-specific activity but, instead, did react with antiserum to hog serum. C. ISOIMMUNIZATION AND AUTOIMMUNIZATION AND THE FORMATION OF AUTOANTIBODIES An exciting development in the study of thyroid antigens was opened by the discovery that autoantibodies can be experimentally elicited
THYROID ANTIGENS AND AUTOIMMUNITY
113
against one or more components of thyroid material. This finding also made possible the investigation of experimental autoimmunity as related to the thyroid. Although it had been firmly believed that antibodies of this type could not, in fact, be stimulated, the intentional effort to produce them was successful, because of the utilization of Freund’s adjuvant and an intradermal immunization procedure ( Witebsky and Rose, 1956). Rabbits were immunized with saline extracts of pooled thyroid glands from other rabbits. The resultant antibodies could be demonstrated by complement fixation, precipitation, and tanned cell hemagglutination procedures. It was found that these antibodies were tissue-specific, exhibiting a very high degree, indeed, of specificity for thyroid, and that they were in fact autoantibodies. That they were autoantibodies and not isoantibodies was judged from the positive reactions seen with thyroid material taken as test antigen from any rabbit examined, including the rabbit that was the antibody-producing animal. It was also demonstrated that rabbits could be thyroidectomized or partially thyroidectomized and then immunized with preparations of their own individual thyroid glands, as a result of which they produced antibodies which would react with their own thyroid extracts and also with the thyroid extracts of all other rabbits tested. Rose and Witebsky (1956) then discovered that many of these rabbits developed severe histological changes in the thyroid glands, corresponding to partial or extensive replacement of the thyroid tissue by infiltrating mononuclear cells. These events then constituted an experimental autoimmune disease which will be discussed in Section V,A. For the moment, it should simply be pointed out that these lesions were found to be closely comparable to those observed in humans suffering from chronic thyroiditis, also called Hashimoto’s thyroiditis or struma lymphomatosa. It was subsequently discovered by the London group (Roitt e$ al., 1956) and also by the Buffalo group (Witebsky et al., 1957) that a large number of patients with chronic thyroiditis when tested did, in fact, exhibit the presence of circulating autoantibodies against human thyroglobulin or at least against some antigenic component of human thyroid extract. We can, therefore, speak of this human disorder also as an autoimmune disease, as will be further discussed below (Section VI). With regard to the human antibodies against thyroid components, it was important to prove that the antibody can react with the autologous antigen and not merely with thyroid preparations from other humans. The proof that this is an autoantibody, rather than an isoantibody, was provided in a study of thyroid material from a patient who had autoantibodies in her serum ( Witebsky et al., 1958).Further, it was possible to test the serum with ammonium sulfate-fractionated products from the thyroid and, thus, to show that the thyroglobulin fraction
114
SIDNEY SHULMAN
reacted quantitatively in a way that was very similar to the saline extract itself, This strongly suggested that the autoantigen was, in fact, thyroglobulin.
V.
Experimental Autoimmune Disease of the Thyroid: the Thyroid Gland as Source and Target
A. INDUCTION OF AUTOANTIBODIES AND ESPECIALLY IN THE RABBIT
OF
AUTOIMMUNE DISEASE,
1 . Varieties of lmrnunization Procedure and Source Material Immunization for the production of autoantibodies can be performed in a number of ways. We shall consider first the bodily origin of the antigenic material. Most logically, one can perform autoimmunization, using thyroidectomized or partially thyroidectomized rabbits and injecting them with extracts or homogenates of their own thyroid glands, with suitable adjuvant; an antibody response will result, as was shown by Witebsky and Rose (1956). Although of basic interest, this method is not very practical and is highly limited in the amount of antigenic material available. Most studies have involved the use of thyroid extract (or in some later work, purified thyroglobulin) derived from the pooled glands of other animals of the same species, that is, the process of isoimmunization. As mentioned above, such immunized rabbits develop antibody activity which reacts with rabbit thyroid extract (or with purified thyroglobulin ) in various kinds of test, including precipitation, tanned cell hemagglutination, complement fixation, and passive cutaneous anaphylaxis. These antibodies are highly specific for the thyroid tissue, as already mentioned, although they do show cross-reactions with the thyroid extracts of several other species, such as hog, dog, and horse (Rose and Witebsky, 1959); in this respect, they are merely reflecting the distribution of thyroid-specific antigens that had already been seen with heteroantibodies. Most of these cross-reacting antibodies can be removed by careful absorption work. The third method, theoretically, for producing autoantibodies would be that of heteroimmunization, and this has been found to be effective to some degree. Thyroid extracts from man, cattle, dog, and hog have been injected into rabbits, with adjuvant, and the resultant antibodies reacted with rabbit thyroid extract, as well as with the various heterologous and homologous preparations ( Witebsky and Rose, 1959). However, there were limitations in this response. First, absorption studies showed that only a small portion of the total antibody population could react with the rabbit antigen.
THYROID ANTIGENS AND AUTOIMMUNITY
115
Second, the antisera did not show complement fixation in tests with thyroid extract from rabbit, although they reacted well in precipitation and passive hemagglutination tests. These immunized rabbits did not show skin reactions to rabbit thyroglobulin, although they did react well against the injected foreign material (Rose and Witebsky, 1959). Furthermore, there was no development of thyroid lesions in these animals. In some ways, this picture resembles that of human autoimmune thyroiditis, namely, in the complement fixation defect, though in other ways it is even more restricted, since there is no response of cellular immunity. Differences in immunogenicity of small regions on the molecule of thyroglobulin may be of importance here. We should consider those aspects of immunization that are negative. Intravenous (or intradermal) injection of rabbits with thyroid extract of rabbit fails to elicit autoantibody production or any thyroid tissue damage. If bacterial lipopolysaccharide is also injected, the result is the same (Rose et al., 196213). Surgical trauma has also been tried, as well as administration of radioactive iodine, and these were both unsuccessful in stimulating an autoimmune reaction. Efforts at traumatizing the tissue by means of cryosurgery were also unsuccessful, although this procedure could produce an anamnestic response in an already injected animal (Ghayasuddin et al., 1969). The various kinds of adjuvant that can be used and their varied results will be discussed later. 2. Skin Tests and Histological Studies
It is important to emphasize that rabbits that were injected in such a way that they produced circulating autoantibody were frequently found to give positive skin test reactions to cutaneous injections of rabbit extract. This demonstration of delayed hypersensitivity has been confirmed in other animal species and in many kinds of experimental situations, and it will be of great importance in the discussion of mechanisms of pathogenicity. In a number of these animals, assays were made of the amount of antigen remaining in the injected animal’s own thyroid tissue. Examination of such tissue extracts showed that a considerable loss of antigen had occurred. This showed a logical correlation with the finding revealed by histological studies, namely, that much of the colloid had disappeared and that there had been extensive replacement with fibrous tissue in extreme cases or, at least, an inflammatory response. For the production of thyroiditis in these animals, it was generally found necessary to give repeated injections of the thyroid material and adjuvant. Witebsky et al. (1957) did describe occasional induction with
116
SIDNEY SHULMAN
FIG.7. Histological specimen of thyroid tissue from rabbit 876, showing only very slight abnormality. The somewhat eccentrically located cystic duct contains some desquamated epithelium and many lymphocytes. Infiltration by a few plasma cells and lymphocytes in the surrounding stroma is seen. (From Shulman and Witebsky, 1960a.)
a single injection. The lesions developed over a period of time to become more extensive. Some of these varieties of histological damage are illustrated in the work of Shulman and Witebsky (1960a), from which Figs. 7, 8, and 9 are taken. These sections show, for one of the rabbits, only a very slight alteration from normal, and for the other two animals, marked changes and damage. These changes include desquamated epithelium, the infiltration of lymphocytes and other cells, and the development of fibrosis. 3. Significance of Adjuvant In the work described above, the production of experimental thyroiditis (that is, tissue lesions in the thyroid) was invariable accomplished by the concomitant use of Freund's complete adjuvant in an emulsion with antigen, deposited in an intradermal depot. If incomplete Freund's adjuvant was used, on the other hand, the thyroiditis was seldom indueed, but autoantibody was formed, although at lower titers. Furthermore, if alum was used as an adjuvant, by injecting alum-precipitated thyroid extract, the rabbits generally developed good levels
THYROID ANTIGENS AND AUTOIMMUNITY
117
FIG.8. Histological specimen of thyroid tissue from rabbit 877, showing very distinct changes, with dense focal infiltration by lymphocytic and eosinophilic cells and by many disintegrating leukocytes. Numerous macrophages and desquamated epithelial cells are intermingled with leukocytes in the lumen of the follicles; there is also nuclear debris. (From Shulman and Witebsky, 1960a.)
of autoantibody, but they had only occasional positive skin tests and no occurrence of thyroid lesions. If killed tubercle bacilli were added to the alum-precipitated material, tbe skin reactivity was greatly increased and half of the animals showed typical thyroid lesions. Thus, it was shown by Rose and his colleagues that the response of autoantibody production and the response of tissue damage and delayed hypersensitivity could be dissociated from each other (Doebbler and Rose, 1961; Rose et al., 1962b, 1965a). In recent years, pertussis vaccine has been successfully used as an adjuvant, notably in producing thyroiditis in rats (Kalden et al., 1969a; Twarog and Rose, 1969). Sometimes, pertussis vaccine plus complete Freund's adjuvant was used (Paterson and Drobish, 1968). This approach is a new extension of the thorough studies of Levine and Wenk (1961, 1964, 1965), who used this adjuvant very effectively in the induction of experimental allergic encephalomyelitis, and of its use in the induction of experimental aspermatogenesis (Hargis et al., 1968). The role of Freund's complete adjuvant has attracted much speculation as to whether the components of acid-fast bacilli induce slight molecular changes in the thyroglobulin. However, no real evidence for
118
SIDNEY SHULMAN
FIG. 9. Histological specimen of thyroid tissue from rabbit 881, showing very marked changes with extensive fibrosis and cystic distention of follicles, which contain much nuclear debris and many macrophages, some with hemosiderin granules. Many small atrophic follicles without any colloid are seen. (From Shulman and Witebsky, 1960a. )
this conclusion has been collected. Weigle et al. (1969) have studied this question in some detail. They did not agree with the notion that the thyroglobulin antigen was modified in the process of simply emulsifying it with the suspension of tubercle bacilli or, for that matter, with mineral oil, nor even during in vitro storage in the adjuvant. The mycobacteria in the adjuvant have no direct effect on the thyroglobulin. However, their studies led them to believe that some alteration did occur in the molecule, rather soon after injection. They felt that the mycobacteria in the adjuvant attracted certain enzyme-liberating cells, largely neutrophiles, into the region of the depot of emulsion. The cells then produced a decrease in the local pH, within cells or in local areas in the granuloma, and there was a subsequent action of proteolytic enzymes. They felt that this acid environment promoted the swifter modification of the thyroglobulin antigen.
4. Efects of Chemically Modified Thyroglobulin In a very interesting series of studies, Weigle has explored the possibilities of inducing autoimmunity without the use of any adjuvant or
THYROID ANTIGENS AND AUTOIMMUNITY
119
at least without the use of the acid-fast bacilli, generally by means of using molecularly altered thyroglobulin. This approach was based on his earlier studies on immunological tolerance to bovine serum albumin, induced neonatally in the rabbit. He had shown that this tolerance could be terminated by injecting certain cross-reacting albumins ( Weigle, 1961), or it could be terminated following the injection of bovine serum albumin which had some structural alterations but which could still cross-react with the native antigen ( Weigle, 1962). On this basis, it was reasoned that the induction of autoimmunity could well involve some mechanisms similar to those involved in the termination of tolerance in the manner described. Rabbit thyroglobulin was altered by several chemical treatments, including the coupling through diazonium derivatives to sulfanilic acid or to arsanilic acid and sulfanilic acid, giving sulfanil-thyroglobulin or arsanil-sulfanil-thyroglobulin. Picrylthyroglobulin was prepared, using picryl chloride, and heat-modified thyroglobulin was also used. Both precipitating and hemagglutinating antibodies against native thyroglobulin were produced in rabbits that were injected with incomplete Freunds adjuvant, containing any of these four derivatives of thyroglobulin. In some of the rabbits, mild-tosevere thyroid lesions developed. In contrast, injection of native antigen in incomplete Freund's adjuvant resulted in little if any circulating antibody and very rarely any thyroiditis. Immunization was just as effective when arsanil-sulfanil-thyroglobulin was injected in the soluble form as when it was incorporated in the incomplete adjuvant. It was, thus, shown that autoimmunity, meaning both thyroiditis and antibody, could be induced in a number of cases, without any adjuvant or at least without any mycobacteria, provided one used a molecularly modified antigen ( Weigle, 1965a). Repeated periodic injections of aqueous preparations of thyroglobulin of a modified type resulted in a perpetuation of the production of circulating antibody and of thyroiditis ( Weigle and Nakamura, 1969). In another report, the effects of booster injections of native thyroglobulin were studied in those rabbits that had, at some earlier time, received the injections of altered thyroglobulin without adjuvant. After a latent period of 1 month following the injection of the modified antigen, rabbits responded to an injection of native antigen with an increase in antibody production and also in the severity and frequency of thyroid lesions (Weigle, 1965b). Some of the rabbits, in an alternative experiment, responded in booster fashion to their own antigen as released from an autotransplant of thyroid tissue, that is, a lobe of thyroid transplanted to the surface of a neck muscle, in this case. This bit of tissue suffers a central necrosis with loss of structure of the follicles and pre-
120
SIDNEY SHULMAN
sumed release of thyroglobulin. This liberated protein was apparently able to stimulate the immune response in a rabbit that was already sensitized. However, such an autotransplantation in nonsensitized rabbits failed to stimulate formation of antibody or lesions. It was further found that the effect of immunization with the altered antigen could be inhibited by injecting the native rabbit thyroglobulin, either prior to or simultaneously with the injections of the arsanil-sulfanil-thyroglobulin (Weigle, 1967). If was considered that this inhibition of termination of unresponsiveness might be a general phenomenon by which the body prevents autoimmunity from developing, especially to endogenous antigens. The possibility of inducing the several kinds of experimental autoimmunity would be attributed to the presence in the circulation of only a relatively low level of antigen from the respective organ. This reasoning seems to be essentially a different form of the sequestered antigen hypothesis of autoantibody formation, which was quite widely espoused at that time, but which has become less acceptable at the present time. The effects of autologous thyroglobulin that could be released in vivo from the thyroid glands was studied in rabbits previously immunized with either altered rabbit thyroglobulin or heterologous thyroglobulin. The release was effected by injection of an adequate dose of Na1311, resulting in disruption of the thyroid structure. The result of such release was a secondary response as seen in immune elimination of the 1311thyroglobulin and the production of antibody to native thyroglobulin ( Weigle and High, 1967a). The termination of unresponsiveness and the induction of autoimmunity were studied as consequences of antigenic competition with unrelated antigens ( Weigle and High, 196%). It was possible to inhibit the production of circulating antibody to thyroglobulin without inhibiting the development of thyroiditis, suggesting evidence either for an important role of cellular hypersensitivity in thyroiditis or for the role of a small amount of an immunoglobulin which is not affected by antigenic competition. McMaster and Kyriakos ( 1970) also studied antigenic competition in guinea pigs as a means of preventing autoimmunity and thyroiditis. The immunization of rabbits with native preparations of heterologous thyroglobulin without adjuvant resulted in the production of both antibody to rabbit thyroglobulin and thyroiditis ( Weigle and Nakamura, 1967). This was in contrast to previous studies ( Weigle, 1965a; Witebsky and Rose, 1959; Rose et al., 1962b) in which similar immunization, but using incomplete Freund's adjuvant, failed to give any thyroiditis, although antibody production occurred. The reasons for the difference were not clear. In terms of species comparisons, the injections of bovine
THYROID ANTIGENS AND AUTOIMMUNITY
121
and human antigens were able to give an effect, but the porcine antigen was rather ineffective. Other studies of Nakamura and Weigle (1967b,c, 1968b, 1970) have discussed the induction and maintenance of immunological unresponsiveness by various means. David and Holborow (1961) tried to induce immunological tolerance to thyroglobulin in neonatal rabbits, but this effort was unsuccessful.
B. GENETICFACTORS IN EXPERIMENTAL THYROIDITIS In recent studies, McMaster and his colleagues (1967) have indicated the importance of two additional factors in the mechanism of production of experimental thyroiditis. These factors are the immunizing dose and the genetic constitution of the experimental animal. Although it had been appreciated in the work of others that the dose of antigen and also the dose of mycobacteria would influence the frequency and severity of autoimmune disease in experimental animals, this study emphasized certain aspects of this relationship for autoimmune thyroiditis. It was found that both the dose of thyroid antigen and the dose of tubercle bacilli in the immunizing emulsion were important and that they appeared to act in an additive fashion, since large amounts of both are necessary to produce severe disease. If the dose of either one is decreased, the severity of the lesions is reduced and, ultimately, the incidence is decreased. These workers have also emphasized the importance of the genetic constitution of the experimental animal. They compared the results in two strains of animals, namely, the randombred Hartley strain and the inbred strain 13 guinea pigs. The Hartley strain developed thyroiditis more readily in response to low doses and developed it more severely in response to higher doses, in comparison to the strain 13 animals ( McMaster et al. 1965, 1967). At the same time, Munoz (1967) compared these same strains of guinea pig, although not in terms of thyroid antigen. He did find that the Hartley animals developed more intense reactions of delayed hypersensitivity than did the others, and both gave good responses of circulating antibody. Lerner et al. (1962, 1964) pointed out that the thyroiditis produced in strain 13 histocompatible guinea pigs after a single injection appeared in 5 dqys and persisted for more than 2 years. This corresponds to almost half the life-span of the animal. This prolonged state of disease was attributed largely to the use of a single injection, although other factors may have also been important, such as the emulsion viscosity, the use of Mycobacterium tuberculosis (rather than Mycobacterium butyricum) , the high dose of antigen, and other proposed possibilities. Genetic influence on this autoimmune disease has also been claimed
122
SIDNEY SHULMAN
in mice, in that the incidence of experimental thyroiditis is greater in the Swiss strain that in the black C57 strain ( McMaster et aZ., 1965). Other indications of genetic factors have been found in the spontaneous occurrence of thyroiditis in more-or-less inbred animals, particularly in dogs and chickens. The canine example is based on studies of beagles, using purebred colonies. Although there has been some evidence to suggest an inherited trait, the published reports do not fully define and characterize the genetic aspects. One large colony of about 1000 beagles has been described in which about 12%of the animals showed spontaneously arising thyroiditis ( Musser and Graham, 1968). In a noncolony group of beagles, about 3%had thyroiditis. Tucker (1962) had earlier reported that thyroiditis occurs in dogs, finding a 16% incidence in a smaller group of beagles. He also mentioned reports of two other studies to indicate similarly high incidence of thyroiditis in beagles. He did not claim any distinctive occurrence, compared to other breeds, although he cited several other studies that indicated a very low incidence of thyroiditis among dogs in general; the thyroid disease in these studies was not fully described, however. In more recent work, Beierwaltes and Nishiyama (1968) studied a total of 67 beagles, but no other breeds; these included animals from various degrees of inbreeding. Using tanned cells coated with canine thyroglobulin, they found hemagglutinating activity in a number of the serum samples. Almost all the dogs showed titers in this test of at least 1:16, with some as high as 1:512; there was no correlation with the occurrence or severity of histological damage. The lesions were carefully described, and they were considered indistinguishable from human Hashimoto’s disease. A subsequent study from this group (T.C. Evans et al., 1969) actually involved immunization efforts with this breed of dog. The results will be described in Section V,D. Mizejewski (1971) has studied this beagle colony further and has examined the next generation of dogs, finding a 73%incidence of thyroiditis. Using a variety of tests, thyroid antibodies were found in many of them. A random group of dogs, in contrast, showed no occurrence of thyroiditis. The other genetic study was undertaken with chickens. This model of spontaneous autoimmune disease was found in a selectively bred strain of White Leghorn chickens. Certain birds were found to show phenotypic characteristics that could be associated with hypothyroidism, such as excess subcutaneous fat, a somewhat smaller skeleton than normal, long silky feathers, and poor laying ability (Van Tienhoven and Cole, 1962; Cole, 1966). These characteristics were at first observed in less than 1%of a closed flock of White Leghorn chickens, the Cornell C-strain.
THYROID ANTIGENS AND AUTOIMMUNITY
123
By a process of selective breeding, the incidence of birds with these symptoms was increased to over 80%, and this new strain was termed the Obese Strain (0s). Histological study of the 0s thyroid gland revealed infiltration of lymphoid cells, resembling Hashimoto’s disease in man. Preliminary studies also revealed that many of these birds had circulating autoantibodies to thyroglobulin ( Cole et at., 1968). Antibodies were detected in most of the birds studied, either by passive hemagglutination or precipitation methods (Witebsky et al., 1969) or by fluorescent antibody methods (Wick el al., 1970d). By tanned cell hemagglutination, over 65%of these birds showed positive results, with titers from 1: 10 up to 1:5120. In contrast, other chickens (from various sources) showed no antibody activity (except for one serum out of several hundred) with these tanned cells coated with chicken thyroid extract. Tissue specificity was demonstrated by lack of reactivity in tests with other chicken organ extracts. The thyroid extracts of other avian species were positive, but mammalian sources provided only negative or weak reactions. Precipitation was demonstrable by using agar which contained 8% NaCI. This antibody was shown to be an autoantibody and to be distributed, by sucrose gradient fractionation, into both heavy and light antibody regions. By means of immunoelectrophoresis, only yM characteristics could be shown for the antibody. The colloid of the chicken thyroid was stained in immunofluorescence, with no staining of epithelial cytoplasm. Two kinds of antibody were proposed, depending on the test used. However, this may merely reflect (again) the production of antibodies against different antigenic sites on the thyroglobulin molecule. The pathological aspects of this disease were described in further detail by Kite et al. (1969). They reported the infiltration of the thyroid glands by both large and small lymphoid cells, although there was a preponderance of large mononuclear cells, many of which were plasma cells. The effects of bursectomy and of thymectomy have been studied by Wick et al. (1970a,b). Bursectomy, which was done on the day of hatching, or even more strikingly, in ovo on day 19 of incubation, resulted in a decrease in the incidence and in the severity of thyroiditis at the usual time after hatching. It was concluded that bursa-dependent lymphoid cells have a major function in the development of the spontaneous thyroiditis. In contrast, neonatal thymectomy caused a subsequent increase in the frequency and severity of the disease. This finding is in accord with other observations concerning the onset of autoimmune conditions after thymectomy. One such example is that of the earliest onset of spontaneous, autoimmune, hemolytic anemia in NZB mice after neonatal thymectomy (Helyer and Howie, 1963). Differences
124
SIDNEY SHULMAN
in regard to injection-induced disease, as contracted with spontaneous disease, will be mentioned later. The effects of radiation damage have also been presented and discussed by Wick et al. (1970~).
C. DELAYED HYPERSENSITIVITY AND CELLULAR IMMUNE RESPONSES The cellular phenomenon in lesion formation generally involves an infiltration of certain white blood cells. The typical tissue lesion in many of the autoimmune diseases involves an accumulation of mononuclear cells, as has been reviewed and emphasized by Waksman (1962). This is characteristic of experimental autoimmune thyroiditis where there is an extensive infiltration of lymphocytes and macrophages as the major cellular types. It would, therefore, be desirable to have a better understanding of the chemotactic influences that cause an invasion of these particular cells. An associated phenomenon in lesion formation is the attack that causes a loss of tissue cells. Some instances exist, such as in experimental aspermatogenesis, where inflammatory responses are not always seen and where there is, instead, a loss of the epithelium of the tubule (Shulman, 1971a). A similar effect is also seen in thyroiditis, as a matter of fact, since one aspect of the malfunctioning involves the occurrence of desquamated epithelium. There are, therefore, mechanisms that cause an invasion of inflammatory cells, and there are also mechanisms that are toxic to local tissue cells or at least loosen their attachment to the basement membrane. In general, the phenomena leading to tissue damage can be grouped into two major categories. One category consists of the influence of circulating autoantibody in its expression as a tissue-specific antibody. We would expect that this would involve a function of some cytotoxic antibody with an appropriate specificity for the target tissue, aided. by an interaction with complement and, probably, followed by a chemotactic attraction for certain lymphoid cells. The second category includes the mechanisms of cell-mediated immune responses, and this would involve essentially a mechanism of delayed hypersensitivity, directly attuned to the particular target tissue. In this situation, the circulating autoantibodies that are so easy to detect and so prominently present would simply represent a side product or at any rate a hallmark of the disease, rather than a pathogenic factor in itself. Although the antibody would not be a causative factor, it would still be considered to be of great importance in terms of recognition that an autoimmune process was in progress. The significance of delayed hypersensitivity in autoimmune disease has been discussed by Roitt and Doniach (1967a). Since the discovery of experimental autoimmune thyroiditis in 1956,
THYROD ANTIGENS AND AUTOIMMUNITY
125
many studies have been made on two associated problems in the effort to explore whether the circulating autoantibody could be implicated as a causative agent in the formation of lesions. In this sense, a number of reports were made on the correlation of the titer of circulating antibody with the degree of lesion damage. Although the reports have frequently been ambiguous to some extent, the general nature of the conclusions was that there was no convincing correlation of antibody titers with severity of lesions. For example, one can examine the report by Terplan et al. ( 1960) for typical observations along this line. Rose et al. (1962b) have also stated that there is no significant correlation between the titer in passive hemagglutination and the degree of thyroiditis. As an associated problem during these early years, from about 1958 to 1961,many efforts were made to transfer thyroiditis to normal animals by transfer of serum from immunized animals. These attempts failed consistently. In 1961 the first report appeared that indicated that thyroiditis could, in fact, be transferred by the transfer of lymphoid cells ( Felix-Davies and Waksman, 1961). This was a very brief and preliminary report, but it began a chain of studies which indicated ever more strongly that a form of delayed hypersensitivity was more likely to be the causative agent in thyroiditis than was the action of circulating autoantibody. Skin-sensitizing antibodies in experimental thyroiditis were described by Metzgar and Buckley (1967). It was claimed by McMaster et al. (1961) that there was a good correlation in guinea pig studies between the development of experimental thyroiditis and that of delayed hypersensitivity, although a role for circulating antibody also had to be admitted. During the first 5 days after isoimmunization of the guinea pigs, no antibody was detectable by the passive hemagglutination test, but there were signs of thyroiditis and also of mild delayed-type hypersensitivity reactions when skin tests with thyroglobulin were made. At 7 weeks, all the animals with thyroiditis showed delayed hypersensitivity reactions, but there was also a good correlation with antibody titer. In a group of guinea pigs that were immunized with thyroglobulin in incomplete Freund's adjuvant, on the other hand, there were low titers of circulating antibody, but with no development of either the tissue damage or the skin reactions. Rose and his colleagues (Doebbler and Rose, 1961; Rose et al., 1962b) showed clearly that one could dissociate the response of autoantibody formation from the response of tissue lesion development. They were able to stimulate a high titer of antibody production by means of immunization of rabbits with alum-precipitated antigen; under these conditions, there was no development of thyroiditis. However, if complete Freunds adjuvant was used, the immunization resulted in severe
126
SIDNEY SHULMAN
inflammatory changes of the thyroid characteristic of thyroiditis and also in the development of delayed skin hypersensitivity, as well as circulating autoantibodies. A quite different approach was used by Miescher et a1 (1961) in an effort to influence the antibody and the hypersensitivity responses. They showed a close relationship between the incidence of delayed hypersensitivity to thyroglobulin and the incidence of immune thyroiditis in guinea pigs that had been immunized with a picrylated preparation of guinea pig thyroglobulin, together with complete Freund’s adjuvant. This immunization procedure led to a considerably reduced antibody response, but left the delayed hypersensitivity response largely unchanged. On the other hand, they could not establish any correlation between the titer of circulating antibodies and either tissue damage or delayed hypersensitivity. They found, for example, that approximately 60%of the immunized animals developed thyroiditis after an interval of 17 days following the initial injection. In the same animals, approximately 70%developed delayed hypersensitivity to intradermal test injections of thyroglobulin. On the other hand, they found that circulating antibody against thyroglobulin, as measured by passive hemagglutination, was only elicited in either 5 or 30%of two groups of animals, depending on the dosage of the injected thyroglobulin. These two animal groups did not differ in the incidence of the tissue reactions. It was somewhat disconcerting that they did find, in a number of the animals, that positive skin tests were present although thyroiditis was absent. However, they felt that this finding did not contradict their thesis that delayed hypersensitivity to thyroglobulin is the pathogenetic pathway for experimental thyroiditis. This reasoning was based on the feeling that the first stage in this mechanism would be the development of delayed hypersensitivity and that thyroiditis would appear at a later stage. Later, Spiegelberg and Miescher ( 1963) evaluated the possible mechanisms for the pathogenesis of thyroiditis by means of studying the different consequences of two alternative antimetabolites-6-mercaptopurine and aminopterin-on the course of immunization. It had already been reported that both these drugs are active in inhibiting the development of delayed hypersensitivity in the guinea pig (Friedman et al., 1961; Hoyer and Condie, 1962) and that antibody formation in the guinea pig was suppressed only by aminopterin (Friedman et al., 1961; Genghof and Battisto, 1961). It appeared, therefore, that the mechanism of action would be quite different for these two compounds with regard to the immune responses in guinea pigs. Hence these workers administered various dosage levels of each of these antimetabolites during different stages of the immunization phase in the experimental
THYROID ANTIGENS AND AUTOIMMUNITY
127
animals. They found that both compounds depressed the formation of delayed hypersensitivity and the development of immune thyroiditis. They confirmed that the formation of circulating antibody was strongly depressed by aminopterin but not significantly influenced by 6-mercaptopurine. Furthermore, thyroiditis could be developed in animals that had been treated with aminopterin and in which antibody formation was, therefore, not demonstrable. They again showed that, in general, there was a good correlation between the animals that showed a positive skin test and thyroiditis. Braley and Freeman (1969) have recently made similar studies. A group of papers by Flax and his co-workers illustrated some histological features of the condition that they refer to as “experimental allergic thyroiditis” (Flax, 1W3; Flax et al., 1963; Flax and Billote, 1965). They reported a good correlation between the intensity of the delayed hypersensitivity and the severity of thyroiditis appearing in various individual members of the immunized groups of guinea pigs. On the other hand, it was not possible to show any correlation between the titers of circulating antibody to thyroglobulin and the degree of thyroid pathogenicity. It was concluded that a mechanism of delayed hypersensitivity to thyroid antigen was of importance in the pathogenesis of thyroiditis. There would still be some need to explore the identity of the antigenic groups that stimulate the two kinds of immune response, namely, that of circulating antibody and that of delayed hypersensitivity. In some of the studies, the sequential development of thyroid lesions after immunization with homologous thyroid material was examined in some detail. The lesions were described histologically with regard to the various types of cells that invade the tissues as the inflammatory response evolves. In later work (Flax and Billote, 1965), correlations were attempted between the morphological disturbances in the development of experimental thyroiditis and certain functional aspects of the thyroid follicles, as manifested by the uptake of iodine. These studies involved the use of radioautography along with other histological examinations. It was found that the reduced ability to take up iodine evolved at about the same rate as the infiltration with mononuclear cells. These findings suggest that cellular infiltration might very well be a causative factor in the damage to thyroid tissue, although no specific mechanism for the production of damage could be proposed, except to emphasize that a process of delayed-type hypersensitivity was inherent in the alterations. The study by Wasserman and PackalCn (1965) has also emphasized the importance of delayed hypersensitivity in the formation of experimental thyroiditis in guinea pigs. They confirmed the correlation be-
128
SIDNEY SHULMAN
tween skin sensitivity and thyroiditis that had been found by others in immunized animals. They also showed that the eventual regression of thyroiditis was accompanied by a decrease in skin reactivity, and they found that there was no correlation between delayed hypersensitivity and the circulating antibodies. They then attempted to apply the method of cell migration as an alternative manifestation of delayed hypersensitivity and the circulating antibodies. They then attempted to apply the method of cell migration as an alternative manifestation of delayed hypersensitivity, testing for leukocyte migration in the presence or absence of thyroglobulin or other antigen preparations. They found that the titers of serum antibody were correlated with the degree of leukocyte sensitivity. This finding suggested to them that there should be a reexamination of the possible role of humoral antibody in regard to cellular hypersensitivity in u i t ~ o . Many indications had thus accumulated for several years to suggest that delayed hypersensitivity was the causative mechanism in thyroiditis and in the manifestation of the typical lesion. The evidence seemed to indicate ever more convincingly that the humoral autoantibody was only some indication of the disease, a type of side product, and that it was not, in fact, a causative agent; therefore, it was not well correlated in its concentration with the severity of the disease. We shall, however, consider below (Section VI1,B) the data that indicate the other side of the argument-the data showing that the autoantibody by itself might well be involved as a causative factor in the formation of the lesions.
D. ADDITIONALANIMALMODELS Among experimental animals, the rabbit was the first model to be used, and it has continued to be of interest in current studies, especially with regard to the autoantibodies to the thyroglobulin antigen. Other animal species have been studied and compared, as will be surveyed below.
1. Dog Terplan et al. (1960) immunized a group of dogs with canine thyroid material in a manner similar to the procedure first developed with rabbits. They described the evolution of small lesions, seen after 1 month, into increasingly severe histological damage at various stages. Only a few of the animals developed circulating antibodies, and these were of quite low titer. Unfortunately, it was not clearly stated whether the dogs were bled more than once during a rather long course of immunization.
THYROID ANTIGENS AND AUTOIMMUNITY
129
A recent report by T. C. Evans et al. (1969) discusses renewed efforts to produce experimental thyroiditis in dogs and especially to produce a form that would closely resemble that seen in Hashimoto’s disease in man. A group of beagle dogs was used, several from their colony animals of inbred beagles and others from noncolony beagles. The colony animals were, thus, from a population that had been shown to have a rather high incidence of spontaneous thyroiditis, as discussed earlier ( SectionV,B). Four of the 6 noncolony dogs developed thyroiditis after some months of immunization, but unfortunately, no antibody studies were done. It was felt that the tissue damage occurred much sooner and with a greater intensity than in the study of Terplan and his colleagues (1960) cited above. The contributing role of a high intake of iodine in these animals was suggested as a possibility, but no evidence was given. These were other differences, including the number and timing of injections and the different (lesser) use of complete Freund’s adjuvant, as well as the fact of using beagles. Surprisingly, in this study, it was also found that the 3 inbred colony dogs failed to develop thyroiditis. No explanation could be given for this difference in response, which may be real, even though the number of animals is quite small. As with the spontaneous disease itself, one may be puzzled as to the parameters defining beagles of apparently different degrees of inbreeding. 2. Rat Jones and Roitt (1961) found a tissue response in rats as early as 10 days after a single injection. of homologous or heterologous thyroid extracts or thyroglobulin, in complete Freund’s adjuvant. In these studies, Wistar rats were used. The thyroid tissue suffered increasingly extensive invasion by lymphocytes and histiocytes, with a maximum in tissue damage at about 16 days. Regression then occurred, and by 4 weeks, the glands generally looked normal. Bjorklund (1964) used Sprague-Dawley rats and obtained thyroiditis after injecting rat thyroid extract in complete Freund’s adjuvant. Other reports have described experimental thyroiditis in rats (Metzgar and Grace, 1961; Imas et al., 1969; Kalden et al., 1969a,b). In fact, the occurrence of spontaneous thyroiditis in this species was reported by Hajdu and Rona (1969); further work on this kind of rat study will be of interest for consideration of possible genetic factors. Quite recently, the Lewis strain of rat has been used (Jankovib et al., 1969; Twarog and Rose, 1970; Paterson and Drobish, 1968). This will undoubtedly open a period of rewarding studies with this animal, which has been shown to be extraordinarily susceptible in general to experimental autoimmune disease (Levine and Wenk, 1961, 1964, 1965).
130
SIDNEY SHULMAN
Kalden et al. (1969a) produced thyroiditis in Wistar rats by a single intradermal injection of homologous thyroglobulin in Freund's complete adjuvant, along with an injection of pertussis vaccine. Severe lesions of the thyroid gland were seen as early as 8 days and autoantibody could be detected by the same time. The tissue response was completely similar to the results of Paterson and Drobish (1968). It made no significant difference in the production of thyroiditis and antibody whether the antigen and adjuvants were injected into the hind footpad or into the cervical lymph nodes. It was also shown that thyroiditis could be suppressed by the administration of the yG fraction of heterologous (horse) antilymphocytic serum, provided it was given immediately after the injection, and not later (Kalden et al., 1969b).
3. Guinea Pig Terplan et al. (1960) also showed that guinea pigs responded to suitable immunization with guinea pig thyroid extract and Freunds complete adjuvant, producing autoantibody and tissue damage. Many additional studies have been made with guinea pigs (Flax et al., 1963; Flax, 1963; Flax and Billote, 1965; Janltovid, 1962; Jankovid and Flax, 1963; Premachandra et al., 1963a,b; PackalCn et al., 1967; Salvin and Liauw, 1967). Some of this work, as well as other reports, have been described in some detail earlier (Sections V:B and C ) . Godal and KHresen (1967a,b) claimed that thyroiditis could be induced in normal guinea pigs by injection of serum 'from rabbits or from guinea pigs who had been immunized with guinea pig thyroglobulin. More recently, KHresen and Godal (1969a,b) reported the morphological details in the tissue of such animals. These successes would seem to contradict many earlier reports on the failure to transfer thyroiditis by passive transfer and even by cross-circulation efforts, as, for example, was cited for rats by Roitt et al. (1962) and recently reviewed by Rose and Witebsky (1968) for rabbits, dogs, and guinea pigs. The exact nature of the tissue damage must be carefully evaluated in this recent discovery and in future studies of passive transfer of this sort. Godal and KHresen (1967a,b), in fact, described the dominant infiltrating cell type as granulocytes, apparently eosinophiles, rather than lymphocytes and histiocytes. Comparable studies with rabbits will be discussed in detail later (Section VII,B,2). 4. Mouse
A brief report by Metzgar and Grace (1961) indicated that thyroiditis could be induced in mice, using DBA and C57 strains. More recently, Nakamura and Weigle (1968a) showed that thyroiditis could be in-
THYROID ANTIGENS AND AUTOIMMUNITY
131
duced in mice following injections of heterologous (bovine, equine, human) preparations of thyroglobulin. They studied several strains, including those that were normal and those that were deficient in complement. Multiple injections without adjuvant were effective in certain complement-intact and -deficient strains. The severity and the incidence varied from strain to strain. The Swiss Webster and A/JAX Btrains showed the highest responses of hemagglutinating antibody titer, in addition to lesion development. On the other hand, there was no correlation between antibody level and severity of lesions. Various other strains, including complement-intact and -deficient strains, such as B10-D2/ SN, DBA/lJ, and DBA/2J, produced autoantibody, but they did not respond with thyroid lesions. These results imply certain genetic factors in the induction of thyroiditis, and also a certain complexity in the two responses of tissue damage and of antibody production; differences in complement activity may be a factor in this effect of nonparallel responses, but this supposition has not been borne out in either of these studies. Many more data are needed in this regard and in regard to genetic parameters. Twarog and Rose (1968) also made studies with mice, usually injecting mouse thyroid extract in complete Freund's adjuvant. They did not obtain thyroiditis in any of the strains, which included C3H, F/HeHa, DBA/2, AKR, C57B1/6Ha, BS/VS, and BR/VR, as inbred strains, as well as Swiss mice. Only a few of the Swiss mice showed some minimal thyroiditis. Some animals did show the production of autoantibodies, however, but this varied among the strains. The C3H responded most intensely, for example, whereas the C57B1 responded poorly. 5. Chicken
An important additional model was discovered in birds by JankoviC. and Mitrovid ( 1963), using chickens for the development of thyroiditis by means of immunization. This type of autoimmunity should be carefully distinguished from the spontaneously arising autoimmunity in the 0 s chickens that was already described (Section V,B). One point of distinction can be emphasized ,as an important example. It was already described above that neonatal thymectomy of the 0s chickens resulted in an intensification of the spontaneous thyroiditis. However, in striking contrast, JankoviC. et al. (1965) found that a significant decrease in experimental allergic thyroiditis was the result in neonatally thymectomized chickens. Although quite different chickens were used, the major difference lies in the fact that one kind of thyroiditis arose spontaneously, whereas the other followed immunization; quite different mechanisms may be involved.
132
SIDNEY SHVLMAN
6. Monkey
Recently, the rhesus monkey has been found to be a very useful experimental animal (Kite et al., 1966; Doebbler and Rose, 1966; Rose et al., 1966; Andrada et al., 1968; Themann et al., 1968). The response in these animals, following immunization with monkey thyroid extract, may follow two quite different courses. In one group of animals, there may be a progressively increasing titer of antibody, as seen in passive hemagglutination, maintained for a Iong period of time. The tissue damage begins with slight patches of infiltration, but a progressive involvement ensues, ending with extensive chronic inflammation and fibrosis, and producing a small, firm, adherent thyroid gland. In other animals, there are antibody titers that reach a maximal level early but later decline, even with booster injections. The tissue damage seems to begin with a massive disorganization accompanied by extensive infiltration by lymphocytes and plasma cells (resembling the terminal condition in the first group of animals), which then later becomes milder and more resolved, until a final appearance may be virtually normal. Of special interest in studies with monkeys is the finding that cytotoxic antibody activity is induced in the antiserums and that an antigen additional to the thyroglobulin, which is the only autoantigen in other experimental animals, is operative. In these respects, the monkey model much more closely resembles human thyroiditis than do other animal models. It is more profitable to discuss this model system again at a later point (see section VI,B,2), after the discussion of thyroiditis in man.
VI.
Human Autoimmune Disease of the Thyroid
Although a few scattered reports throughout this century (cited by Doniach and Roitt, 1969) could be considered to have suggested that certain patients with thyroid disease did have antibodies to components of the thyroid tissue, the rational search for and interpretation of such findings could begin only after the immunization studies in experimental animals had led to a production of thyroid disease. Subsequently, the work of Roitt et al. (1956) first demonstrated that thyroglobulin antibodies occurred in the serum of patients with Hashimoto’s disease. By means of gel diffusion they found precipitins in sera from 7 of 9 patients with this condition ( generally termed then lymphadenoid goitre). Almost at the same time, Witebsky et al. (1957) followed up their work in rabbits by finding 12 of 18 patients to have such antibodies (the condition being generally termed then chronic thyroiditis or sometimes chronic nonspecific thyroiditis). They used primarily the tanned cell hemagglutination technique in this work.
THYROID ANTIGENS AND AUTOIMMWlTY
133
Numerous studies have appeared since that time, and many details of the phenomenon have been clarified. In addition, new tools for the study of autoimmunity have developed from such work. Newer problems of the relationship of this thyroid autoimmunity to a variety of thyroid diseases and to other diseases, and also to various genetic factors in man, have arisen, some of which have been explained. Since much of this area of study was well reviewed quite recently by Doniach and Roitt (1969), only a brief survey of some major points will be made here. An item to be omitted from the present discussion is the long-acting thyroid stimulator (LATS). Suffice it to say that this substance shows a number of biochemical and immunological properties of antibodyin fact, of immunoglobulin G-but the precise component of thyroid tissue that is the corresponding antigen needs further clarification. A number of reports can be cited for additional details (Adams and Kennedy, 1967, 1971; Pinchera et al., 1969; B. R. Smith, 1969; B. R. Smith et al., 1969; Doniach and Roitt, 1969; B. R. Smith and Munro, 1970; Solomon and Beall, 1970).
A. THYROIDITIS AND OTHERTHYROID DISEASES The autoimmune diseases of the thyroid are found among the several forms of lymphoid thyroiditis. These forms have been described as goitrous diffuse thyroiditis and as nongoitrous thyroiditis. The former variant includes Hashimoto’s disease, which shows a high incidence of thyroid antibody, and also De Quervain’s and Riedel’s (fibrous) thyroiditis, which are not associated with antibody, in general. The latter variant includes adult primary myxedema and atrophic thyroiditis; these are all forms of hypothyroidism, and all show an appreciable incidence of thyroid antibody activity. There have been clinical arguments to indicate that Hashimoto goiters may regress spontaneously and then terminate in an atrophy that is indistinguishable from primary myxedema. It has been stated that thyroid antibodies can be shown in 98%of patients with hypothyroidism of recent onset, although after some time, the percentage decreases (S. G. Owen and Smart, 1958; Goudie et al., 1959b). The actual incidence of thyroid autoantibody in patients is a rather complex consideration. It depends on the precise clinical condition that is studied and also on the test used, as well as the size of the population group. Generally, the tanned cell hemagglutination method has been used; thus, the thyroglobulin autoantibody is the one detected. In Hashimoto’s disease, antibody of significant titer is found almost universally; it is also found in a high percentage of patients with hypothyroidism. On the other hand, antithyroglobulin is rare in Graves’ disease; however, complement-fixing activity may occur in the sera from about 20%of patients with this disease. Possibly, this activity is related to con-
134
SIDNEY SHULMAN
current thyroiditis lesions in the gland. Of course, the incidence of antibodies in the normal population must be considered. A compilation of a number of published studies shows a mean incidence for thyroglobulin antibodies of 8.3%of 4.488 hospital cases (no thyroid diseases) and 7.6% of 2893 normal subjects (Doniach and Roitt, 1969). These authors also quote studies that give about 4‘ and 6%of 1430 normal subjects, for the incidence of microsomal and CA-2 antibodies ( see below), respectively. In terms of the general incidence of human thyroiditis itself, a matter which is of interest in comparison with the occurrence of spontaneous thyroiditis in other species, Mortensen et al. (1955) have reported a prevalence of 2.6%at autopsy in humans. A recent development of considerable interest in the study of immunological aspects of thyroid disease has been the accumulation of evidence that thyrotoxicosis is, indeed, a primary autoimmune disease (Buchanan et al., 1961). Clinical study had suggested that Graves’ disease may be related to primary myxedema and to Hashimoto’s disease. Immunological studies have supported this concept; this has been especially strengthened by the recent clarification of the nature of LATS, the outstanding component in the mechanism for stimulation of the thyroid gland in thyrotoxic disease. Further descriptions of these clinical situations and their interrelationships with each other and especially with immune factors have been presented in a number of reports (Doniach and Hudson, 1957; Stuart and Allan, 1958; Blizzard et al., 1959; Cline et al., 1959; Porter and Fennell, 1961; Irvine et al., 1962; Mellors et al., 1962; Paseyro et al., 1962; Hjort et al., 1963; Federlin et al., 1965; Godal and Berdal, 1967; Godal, 1967; Zavaleta and Stastny, 1967; Witebsky, 1968; Hjort, 1969; Eyquem, 1970) and in several recent reviews (Rose et al., 1965a; Irvine, 1964; Buchanan et al., 1965; Doniach, 1967; Doniach and Roitt, 1969).] The highly prevalent disease, endemic goiter, is believed to be the result of an iodine deficiency in the diet as the usual cause or at least the precipitating factor. However, other factors have also been suggested, among them the possibility of an autoimmune process. This possibility was first raised by the finding that approximately half of the patients with endemic goiter (but not those with sporadic goiter) showed increased titers to thyroglobulin, as was reported by Hofer and Schatz (1965) and by Soto et al. ( 1967). More recently, Werner et al. (1970) have reported that increased concentrations of immunoglobulin M, as measured by radial immunodiffusion methods, were found in the serum of about half of the patients with either endemic or sporadic nontoxic goiter. On the other hand, the levels of immunoglobulins G, A, and D were found to be normal. Several control groups of patients and of
THYROID ANTIGENS AND AUTOIMMUNITY
135
healthy subjects were also studied, and it was found that only 10%or fewer of these individuals showed an increased level of immunoglobulin M. Statistical analysis indicated that this was a highly significant difference that occurred between the incidence of elevated immunoglobulin M in the goitrous patients and that in the control populations. In this investigation, no specific antibody activity was explored, but the results seem highly suggestive of some sort of autoantibody phenomenon. B. AUTOANTIGENS Unlike the antisera obtained by experimental stimulation, the human sera which show moderate or even very high titers of antibody activity with thyroglobulin or with thyroid extracts in general, by means of tanned cell hemagglutination, do not show any activity at all or only very low levels of activity when tested by means of the complement fixation test. In fact, it was often stated that the human autoantibodies do not fix complement with human thyroid extract or human thyroglobulin, whereas the rabbit autoantibodies do fix complement with rabbit antigen. Some important clarification of this matter was shown a few years later, for in a number of cases, the human serum was, in fact, found to fix complement if the antigenic preparation was obtained from a thyrotoxic goiter rather than from normal thyroid material or from simple nontoxic goiters (Trotter et al., 1957; White, 1957). It was then discovered by Roitt and Doniach (1958, lWO), Holborow et al. (1959), and Belyavin and Trotter (1959) that these reactions of complement fixation were, in fact, due to a second antigen, which was associated with the microsomal fraction of the thyroid homogenate rather than with the thyroglobulin obtained in the soluble portion. The nonthyrotoxic thyroid homogenates simply did not contain a high enough content of microsomes to show a complement fixation reaction. A similar dichotomy could be observed by means of immunofluorescence techniques. This was nicely illustrated in experiments reported by Beutner and Witebsky (1962). It was found that three different types of antibody (two of them thyroid-specific) could be seen in the fluorescent antibody staining of human thyroid slices by means of human sera from thyroiditis patients. The antigens were thus characterized by means of the staining of ( a ) the colloid of the thyroid, ( b ) the cytoplasm of the cells of the thyroid epithelium, and ( c ) the nuclei of the thyroid. The reaction with nuclei does not really concern us, since this type of reaction is found to be non-tissue-specific; staining occurred in nuclei of all organs of the body. The other two reactions, however, indicated a distinction between antibodies against thyroglobulin in the colloid and antibodies against the microsomal antigen in the cellular cytoplasm. In various
136
SIDNEY SHULMAN
patients, the serum might contain one or another antibody or various combinations of them. Beutner and Witebsky (1963) also made studies of the specificity of this antigen, showing that an occasional positive serum reacts also with other tissue microsomes, such as those of brain or liver, but most sera are totally thyroid-specific. They also demonstrated some species cross-reactions for this antigen. Koffler and Friedman (1964.) and Koffler and Paronetto ( 1965a,b) studied the complementfixing activity against the colloid and the epithelial cells, and they described the localization of antibody in the thyroid tissue. It has been found that the microsomal antigen can also be studied in organ culture preparations ( Flanagan et al., 19S6). In studies reported by Balfour et al. (1961) and by Roitt et al. (1960), a third autoantigen was claimed to occur. This was also revealed by immunofluorescent studies and involved a staining of the colloid, but in a pattern which is distinct from that obtained with antibodies to thyroglobulin. This antigen has been termed a second colloid antigen and has been designed as the CA-2 antigen. 1. Thyroglobulin This antigen has already received much of our attention; it does not require more detail at this point. A more thorough discussion concerning this antigen, especially in regard to the human protein, is given later (Section VIII). It may suffice here to point out that the human autoantibodies to this particular antigen are detected quite directly by immunofluorescence, but a number of other procedures are used, varying in convenience for clinical utility and differing in sensitivity ( Rawstron and Farthing, 1962). The most sensitive procedure is the tanned red cell test; the Takatsy microtiter method is highly convenient for this purpose. Latex fixation methods can be used (J. R. Anderson et al., 1962), and bentonite agglutination is possible ( Ager et al., 1959). Gel precipitation methods are useful, although less sensitive. In addition to the usual methods (J. R. Anderson et al., 1960; Goudie et al., 1957), radial immunodiffusion has been suggested as a possibility (Feinberg et al., 1969), and a skin test has also been described (Buchanan et al., 1958).
2. Microsoma1 Antigen The antibody to human microsomes is responsible for the complement fixation activity and also for the cytotoxic activity against thyroid cells in culture. This will be further discussed below in connection with the studies in monkeys. Since there is some evidence that these two activities can be differentiated in various antisera, it is not yet clear whether exactly the same portion of the microsome is involved in
THYROID ANTIGENS AND AUTOIMMUNITY
137
stimulating both antibodies; if it is, then some other factors must be considered to explain different antibody activities. The complement-fixing activity in human thyroiditis serum has been the focus of numerous studies and applications (J. R. Anderson et al., 1959a,b; J. W. Anderson et al., 1967; Irvine, 1962, 1966; Nairn et al., 1963). Some respects of cytotoxicity will be discussed below ( Section VII,A,l ) . Roitt et al. (1964) treated the human microsomal fraction with enzymes and other reagents and could not separate the antigen from the membranes. It was destroyed by surface-active agents. They concluded that the complement-fixing antigen is an insoluble lipoprotein intimately bound to the structural elements of the vesicle membranes of the microsomal particle. Several investigators have recently described the lipids and lipid-protein complexes of thyroid (Okada et al., 1970; Posner and Ordonez, 1970; Scott and Trikojus, 1970). These findings may eventually be very helpful to the understanding of this complex cytoplasmic antigen. Experimental thyroiditis in the rhesus monkey has been used as a model to study some properties that result from the activity of the microsoma1 antigen. Kite et al. (1966) showed that monkey serum, resulting from immunization in the same way as had been done with other animals, contained the ability to fix complement with thyroid microsomes. These monkey antisera did not fix complement with thyroglobulin, thus behaving in a manner similar to the unusual behavior of human thyroiditis sera, although both kinds of primate sera did fix complement with crude thyroid suspensions. The complement fixation occurred also with monkey liver microsomes. The cross-reacting antibody could be removed by absorption with liver microsomes, leaving an antibody that is specific for microsomes of thyroid. The same situation is true for human thyroiditis sera in that most of them are only thyroid-specific, but a few will also react with other human tissue suspensions as was mentioned (Section V1,B). These antisera, resulting from repeated injections with monkey extract in complete Freund's adjuvant, contained autoantibodies that were cytotoxic for monkey and for human thyroid cells in culture. In contrast, no cytotoxicity was ever seen with monkey kidney or adrenal cells, The cytotoxic antibody was heat-stable. Complement was required for damage to occur to tissue cells. Some studies on the sedimentation rate and on the stability to reducing agents suggested that the cytotoxic and complement-fixing antibodies were not identical. Mixed agglutination antiglobulin tests were also used, and these were considered to be superior to the cytotoxicity test in sensitivity, with apparently the same (surface) antigen being detected by both methods. There may be some problem in that tissue-specific antigens remain available in cul-
138
SIDNEY SHULMAN
tured cells for only a short time (Goudie and McCallum, 1962, 1963; Ghose and Cerini, 1969). A related procedure of hemadsorption has been developed in a number of reports from Fagraeus and her colleagues. In the usual procedure, the test sera are applied to filter paper discs from which diffusion occurs through an agar layer to give reaction with a monolayer culture of antigenic cells, As a result of the subsequent application of indicator erythrocytes, the reacting antibodies are visualized as circular hemadsorption zones. In a recent study, Jonsson and Fagraeus (1969) described a ring-zone effect that was seen in a radial diffusion disc test, as a modification of the mixed hemadsorption procedure, in studies with human antithyroid sera and thyroid monolayer cultures. It was concluded that several antibodies of different specificities reacted with restricted antigenic areas carrying clusters of antigenic determinants. Fagraeus and Jonsson (1970) also studied the cell surface distribution of antigens by an immunofluorescence procedure. The action of cytotoxic antibody to thyroid cells has parallel interest in other experimental autoimmune diseases. One other example of some comparative relevance is that of cytotoxicity against sperm cells, although here there are semantic problems; the term spermotoxin was long ago applied to immobilizing activity, whereas a true cytotoxicity in terms of loss of vital staining ability has also been described in other studies. These points and a number of related items have been reviewed recently by Shulman ( 1971a,c).
3. Second Colloid Antigen (CA-2) Very little is known about this antigen except that it is distinct from thyroglobulin and from the thyroid protease. It constitutes less than 1% of the colloid and can only be detected by immunofluorescence. It may have some relationship to thyralbumin, but this is not yet clear (Doniach and Roitt, 1969). Hjort (1963b) studied sera from large numbers of patients with and without thyroid disease; this antibody was found frequently in patients with thyroid disease, but rarely in other patients. Mizejewski (1971) has reported an antigen of this sort to be present in many dogs with spontaneous thyroiditis. C. GENETIC FACTORS IN HUMANTHYROIDITIS Studies of familial patterns of thyroid disease have stimulated the concept that there are genetic factors in this autoimmunity. There has been controversy on the statistical sampling, however, and great caution seems necessary. The aggregation of Hashimoto’s disease, primary myxedema, and thyrotoxicosis in the same families has been noted in a
THYROID ANTIGENS AND AUTOIMMUNITY
139
number of reports, and a common autoimmune basis would be an appealing explanation of this, as well as an argument that strengthens the observation. The first report to suggest this concept was that of Hall et al. (1960), who found thyroid antibodies in 56%of siblings of patients with thyroiditis. Other family studies in this disease and in thyrotoxicosis have given further support to this concept (Buchanan et al., 1961; De Groot et al., 1962; Hall et al., 1964; Doniach et al., 1965; A. W. H. Evans et al., 1967). The survey of Masi et al. (1965) criticized certain earlier studies that were based on hospital-selected patients. It was pointed out that patients who are attending hospitals have the most severe forms of disease, whereas the milder cases may remain undetected. Social and occupational factors may cause a bias, also. False impressions about incidence can also result from the practice of giving greater study (and eventual publication) to cases that show an interesting combination of diseases. These problems were claimed to be especially likely to cause trouble in studies of thyroiditis. Roitt and Doniach (196%) made a reassessment of studies on familial aggregation, taking these objections into account. Even with these corrections, they found thyroid antibodies with a significantly greater incidence in the healthy relatives of Hashimoto patients than in similar randomly selected controls that were matched for age and sex. Studies on twins have also supported the genetic contention. In twins with autoimmune thyroid disease, a high degree of similarity has been found in the antibodies that were detected and in their titers. It has also been noted that families of patients with colloid goiters had almost the same incidence of thyrotoxicosis as those of juvenile thyroiditis probands. Additional arguments have been summarized by Doniach and Roitt (1969). It may well be that even in this kind of autoimmunity, belonging to the tissue-specific group of autoimmune diseases, there is a disturbance of the immunological response, in the sense that Irvine (1964) had discussed thyroid autoimmunity as being a disorder of immunological tolerance. This disturbance could perhaps combine with some local abnormality in the effected tissue to determine the tissue orientation of the immunological damage. Although not due to a genetic factor, the problem of cretinism as a possible result of autoimmunization has received some attention, in addition to the question of transplacental passage of thyroglobulin and of antithyroglobulin. Beienvaltes et al. (1959) were apparently the first to claim that cretinism sometimes results from maternal autoimmunization. This concept was confirmed and extended by Blizzard et al. (1960) and Chandler et al. (1962a). A high incidence of thyroid antibodies was
140
SIDNEY SHULMAN
found in the mothers of athyrotic cretins. It was thus indicated that antithyroid antibodies cross the placenta, although this crossing does not always result in abnormal children. Chandler et al. (1962b) tried to induce cretinism experimentally in the off spring of autoimmunized rabbits. This effort did not succeed, although the transplacental passage of the antibodies was demonstrated. Other studies on human thyroid autoantibodies in the serum and their significance and transplacental passage have been made by Sclare and Taylor (1961).
D. SEROLOGICAL OVERLAP WITH DISEASES OF OTHERORGANS Thyroid disease with autoimmune character quite often occurs in a concurrent situation with one or more autoimmune diseases of other organs. The earliest suggestion along this line and apparently the best supported contention is that of pernicious anemia; overlap with other autoimmune diseases also occurs. In addition, there are claims of a relationship with diabetes mellitus. We can also include here for convenience the associations that have been reported with certain chromosome defects. In pernicious anemia, the histological aspects of the stomach would suggest that immunological factors may be important in the pathogenesis. Autoantibodies against two quite different antigenic constituents of human gastric mucosa have been found. These antigens are the microsomal fraction of the parietal cell cytoplasm for one, and the intrinsic factor for the other. The relationship between thyroiditis and pernicious anemia and autoimmune gastritis has been quite striking, since about 30%of the patients with thyroiditis also have antibodies to gastric parietal cell, whereas 40 or 50% of the patients with pernicious anemia have thyroid antibodies (Irvine et al., 1965; Doniach et al., 1965; Irvine, 1966). Cruchaud and Juditz (1968) have made an analysis of these two kinds of antibodies in patients with these two disorders, A second major association with thyroid disease is disease of the adrenal. J. R. Anderson et al. (1957) showed that complement-fixing antibodies could be found which were directed against adrenal tissue in 2 of 10 patients with Addison’s disease. Antibodies to thyroid were also present in these patients, whereas tests with other organ extracts were all negative. In a number of subsequent reports from several laboratories, evidence has accumulated that a number of patients with this autoimmune phenomenon showed in addition antibody activity against thyroid, and there was an association of thyroiditis with Addison’s disease (Blizzard et al., 1963; Burke and Feldman, 1965; Irvine et al., 1967). A more thorough discussion of adrenal antigens, antibodies, and autoantibodies has been published elsewhere ( Shulman, 1971b).
THYROID ANTIGENS AND AUTOIMMUNITY
141
Other autoimmune diseases associated with autoimmune thyroid disease have included diseases of connective tissue (White et al., 1961; Hijmans et al., 1961; Becker et al., 1963; Doniach et al., 1963; Buchanan, 1965) and diabetes mellitus (Moore and Neilson, 1963; Irvine et al., 1970). Some development of thyroid autoantibodies seems to be associated with certain chromosome abnormalities, including Turner’s and Down’s syndromes (Williams et al. 1964; Fialkow et al., 1965; Ferguson-Smith et al., 1966; Doniach et al., 1968). The simultaneous occurrence of autoimmunity to several organs does seem to support the concept that the pathogenic process must primarily involve some malfunction of the immunological response mechanism. This proposal is one of the several alternatives among the theories of induction of autoimmunity; for example, it is broadly considered in the discussion of Irvine (1964), as mentioned above. The coexistence of several autoimmune disorders seems to give support to this view, but clearly, this cannot be the only mechanism, since there usually is, nonetheless, some limit to the number of organs affected. Other mechanisms, probably related to the particular antigens that are involved, must also be operative.
E. AUTOIMMUNOGENICITY OF THYROGLOBULIN It is beyond the scope of this review to discuss the diverse schemes that have been proposed to explain the fact that autoimmunity occurs spontaneously in man. The various theories have been examined recently in detail and with excellent perspective by Voisin (1970). The general framework of reference includes three types of mechanism. The first type considers that there is an absence of natural tolerance. This includes the “sequestered antigen” concept that certain endogenous substances are kept from contact with the immune response mechanism for a sufficiently long time, and so, tolerance does not really develop. The second type considers situations involving a breakdown in tolerance, caused by some change in, or of, an antigen. This would include either slight modifications of the native antigen or the entrance of an exogenous substance with antigenic similarity to a native antigen. The third type involves conditions resulting from some abnormal functioning of the immune system. For years, the thyroid example of autoimmunity was considered by many investigators to exemplify the sequestered antigen concept in that the thyroglobulin was sheltered in the colloid and had no contact normally with the immune system. This was based on the notion that thyroglobulin was absent from the circulating blood. This concept is
142
SIDNEY SHULMAN
no longer acceptable, and, therefore, we must discuss this one point in connection with the induction of thyroid autoimmunity in human disease. Actually, Hektoen and Schulhof ( 19%), using a specific antiserum, had shown that thyroglobulin is present in the human thyroid vein, or similarly, in certain lymphatics in goitrous dogs, although they found none in the blood at large. Similar findings were made by Dobyns and Hirsch (1956). Nonetheless, for many years, it had been generally thought that thyroglobulin did not reach the circulation under normal circumstances, and, therefore, it could be considered a sequestered antigen, More recently, more sensitive methods have been used to show that low concentrations of thyroglobulin are found in the blood of many normal individuals, including the cord blood of newborn babies. (Hjort, 1963a; Assem, 1964; Roitt and Torrigiani, 1967a,b). Hjort and Pedersen (1962) showed that, in fact, thyroglobulin circulates freely in the newborn. Gitlin and Biasucci (1969) have described the ontogeny of immunologically reactive thyroglobulin in the human conceptus. Ruebner et al. (1960) studied human fetal thyroglobulin, as did also Olin et al. (1970). Careful investigations made by Daniel et al. (1965, 1966, 1967a,b) showed that thyroglobulin drains from the thyroid in low concentration through the lymphatics in normal monkeys and rats. It was indicated that when hormone is needed, thyroglobulin reenters the thyroid cell from the colloid. It was proposed that not all the proteolysis of this protein is completed during the passage of this protein through the cell and that some undegraded protein is released along with the iodothyronines. More recently, Torrigiani et al. (1969) used a very sensitive immunoassay to measure levels of serum thyroglobulin in various healthy subjects and patients. The work of Weigle et a2. ( 1969) has already been discussed (Section V,A 3 ) , in which it was shown that some small but definite modification in the antigen molecule is caused in the experimental induction of thyroiditis by action of the adjuvant. Some similar activity may well be needed for the successful immunogenicity of this protein in the human. It might even be speculated that some abnormal event, such as a local infection, attracts enough of an inflammatory response locally that a change occurs in the pH of that tiny region, which can modify a portion of the thyroglobulin protein located there, and that this actually leads to the autoimmunogenic stimulus. VII. Features of the Autoimmune Response
A. DISTINCIWE TYPFS OF ANTIBODY The autoantibodies that form against thyroglobulin can be distinguished in various ways and can be classified according to two different
THYROID ANTIGENS AND AUTOIMMUNITY
143
criteria. One of these involves the nature of the antibody manifestation, that is, the types of activity exhibited by the antibody in reaction with the antigen. The other criterion involves the immunoglobulin nature of the antibody, or at least, its predominant form, Both of these approaches have been taken in studies of experimental autoimmunity, In human thyroiditis, a higher order of complexity is found, since there are not one, but three, autoantigenic components; different autoantibody types can be expected for each antigen. Finally, recent studies have shown that thyroglobulin as a molecule is autoantigenic in only a few of its structural details; hence, distinctions in specificity among autoantibodies and heteroantibodies to this molecule should be further sought.
1. Forms of Antibody Activity From the earliest studies of experimental autoimmunity, the methods of precipitation (fluid), complement fixation, and tanned cell hemagglutination were all applied to detecting thyroid autoantibodies (Witebsky and Rose, 1956). It soon became clear that there was a serious lack of correlation among these different forms of antibody activity when they were compared for a group of immunized animals or (as was more often done) for a group of thyroiditis patients. It was found in several studies that the passive hemagglutination test was apparently the most sensitive. Positive results were seen in about 3 times as many sera when tested by this method as when examined by the precipitation test (Rose et al., 1965a). At first it was thought that precipitation would not be positive with human sera, unless the hemagglutination titer was at least 1:250,000 (Roitt and Doniach, 1958), but later reports suggested that sera with titers as low as 1:1000 would also give precipitation ( Rose et al., 1965a). Complement fixation generally occurred in tests with human thyroid extract and human sera if the tanned cell hemagglutination titer with these extracts was high, but in truth there cannot be any correlation between complement fixation and agglutination, since the antibodies to thyroglobulin (which do give hemagglutination) rarely fix complement, if at all. In many reports (and even in some reviews) it has been implied that the precipitating and hemagglutinating activities should be grouped together, whereas the complement-fixing activity, being directed against a different antigen, is due to quite different antibody form. It is surely an oversimplification to consider discrepancies in degree of reactivity in precipitation and agglutination as merely a matter of sensitivity in antibody detection. This is indicated in the following rabbit study. The lack of correlation between precipitation and agglutination was demonstrated in a study by Shulman and Witebsky (1960a) of a group
144
SIDNEY SHULMAN
FIG. 10. Double-diffusion gel precipitation between rabbit thyroid (Thy) extract (central well) and sera of bleedings g from six immunized rabbits (peripheral wells). (From Shulman and Witebsky, 1960a).
of rabbits, in which 1 animal (rabbit 876) had serum that was persistently nonprecipitating, whereas the other 5 animals gave precipitating sera. The gel diffusion pattern for these six antisera is shown in Fig. 10, where all are tested simultaneously against ;I rabbit thyroid extract. Because bleedings were taken at a number of weekly intervals, it was possible to study each animal at lower and higher titers of agglutinating antibody and also to test each serum for precipitation. The data are duplicated in Table VI. It is clear that, even though the nonprecipitating rabbit had a generally lower titer than did the others, the antisera from the 5 other animals were able to show precipitation at an early time when their titers were much lower than that eventually achieved in rabbit 876. Another observation of interest about the precipitation behavior of these antibodies concerns what has been termed the doubled-line or split-Zine phenomenon. Using human thyroiditis serum in gel diffusion with purified human thyroglobulin, it sometimes happened that two
145
THYROID ANTIGENS AND AUTOIMMUNITY
TABLE VI TESTS : TITERSOF ALL BLEEDINGS FROM RABBITS PASSIVEHEMAGGLUTINATION IMMUNIZED WITH RABBIT THYROID EXTRACT' Trial bleeding First test a b C
d e f g h Second test h 1
j (final) 5
Rabbit
Rabbit
Rabbit
Rabbit
Week
876
877b
878b
87gb
0 1 2 3 4 6 8 11
0 27 81 243 243 2187 2187 729
11 15 21
7290
-
0 243 81p 243 243 2187 6561 6561
65,610 65,610 65,610
0 81 243p 810 2430 21,870 21,870 21,870
65,610 65,610 196,830
0 81 P
243 729 729 6561 6561 6561 65,610 65,610 196,830
Rabbit 88ob 0 243 243 729 2187 p 6561 6561 6561 65,610 65,610 -
Rabbit 881b
0 27 81 P 243 2187 19,683 19,683 19,683 65,610 65,610 196,830
From Shulman and Witebsky, 1960a. (p) First bleeding to show demonstrable precipitation in gel diffusion plate.
(or more) lines of precipitation were seen (Doniach and Roitt, 1957; Roitt et aZ., 1958; Korngold et al., 1959; Pressman et al., 1957). Although Pressman et al. (1957) considered that the appearance of multiple lines may have been an artifact and could be induced by the use of excess antigen or antibody, Korngold et al. (1959) considered the lines to be genuine, emphasizing in fact that they appeared in the region of equivalence. The same kind of observation was made by Shulman and Witebsky (1960a) for some of the antisera taken from the group of 6 rabbits that had been injected to produce autoantibodies. One of the rabbits (881) showed this effect more clearly than did the others, but the separation was not too distinct. To show it more clearly, the pairs of antigen and antiserum wells were spaced at various distances; at the optimum separation of wells, a clear separation of lines could be seen (Fig. 11).This pattern of lines is rather peculiar in that the ends of the precipitation arcs are fused, whereas the middle portions show a separation. Since an inverse kind of pattern occurs in the formation of the split lines that have been described from time to time under conditions of extreme excess of antigen or antibody, it was felt that this appearance could not be attributed to such an imbalance in the proportions. This effect was further studied by Goodman et d . (1964), using radioimmunoelectrophoresis with human or with guinea pig materials. They found that in immunoelectrophoresis of the extract, the
146
SIDNEY SHULMAN
FIG.11. Double-diffusion gel precipitation between rabbit thyroid (Thy) extract (lower wells) and antiserum 881 g (upper wells) with edge-to-edge separations of 1.0, 2.0, and 3.0 cm. Ab, antibody. (From Shulman and Witebsky, 1960a.)
two lines appeared as concentric arcs, both at the position of thyroglobulin mobility, Iodoprotein was present in each of these two arcs. The authors felt that the multiple lines (sometimes as many as three) were due to an immunochemical heterogeneity of the thyroglobulin (Goodman, 1965). It was studied again by Mates and Shulman (1967c), using human serum and several mixtures of 19s thyroglobulin with various proportions of the 27 or the 1 2 s components. Even with mixtures containing up to 28%of the 2 7 s component or up to 10% of the 1 2 s component, split lines did not appear as a result of making these mixtures. They concluded that the effect is due to a microheterogeneity of the thyroglobulin molecules. The antigenic heterogeneity of the thyroglobulin molecule might be related to the observation that only a few of the total collection of antigenic determinants on the molecule are able to elicity autoantibody formation. It was speculated, as the simplest hypothesis, that there are two major forms of this molecule, and that they differ in only a few of their autoantigenic sites; they thus give rise to the occasional phe-
THYROID ANTIGENS AND AUTOIMMUNITY
147
nomenon of split-line formation, depending perhaps on the degree of responsiveness of the antibody-forming system to each of these characteristic determinants. It was further suggested that the many groups which are identical on the two molecular forms are the same groups as the heteroantigenic ones, whereas the few groups which are distinctive between the two molecular forms are among the groups which are autoantigenic ( Mates and Shulman, 1967~). A very definite distinction of two forms of precipitating antibody was demonstrated in the reports by Roitt et al. (1958) and Shulman and Witebsky (1960b) in which quantitative precipitin curves were evaluated for human autoantibody reacting with human thyroglobulin. Although somewhat different techniques were used in these reports, it was found in both studies that there could be two kinds of curve, one resembling the typical rabbit-type precipitin curve, and the other resembling the classic horse-type flocculation curve, which shows a profound degree of inhibition by antibody excess. Precipitin curves from one of these studies are shown in Figs. 12 and 13. This matter has z
p
2.0
\
z n Q
1.0
0 .+ 0
[L
0.0 0.10
P 2 0.08 z 0.06 0.04
0.02
mg. A g N used
FIG. 12. Quantitative precipitin curve for human thyroiditis serum M.Th., for quantities of precipitate and of antibody (Ab) (by difference) obtained from 0.25 ml. of serum. Ag, antigen. (From Shulman and Witebsky, 1960b.)
148
SIDNEY SHULMAN
FIG. 13. Quantitative precipitin curve for human thyroiditis serum E.C., for quantities of precipitate and of antibody ( Ab) (by difference) obtained from 0.25 ml. of serum. Ag, antigen. (From Shulman and Witebsky, 1960b.)
been further pursued quite recently by Roitt et al. (1968), who have developed some stimulating ideas based on the nature of a few important determinant groups of the total number available on the antigen molecule. These ideas will be easier to discuss at a somewhat later point ( Section VIII,C,B). In the original study of the horse- and rabbit-type precipitin curves, it is perhaps of even greater fundamental interest to note the peculiar values which were determined for the molecular ratio of antibody/antigen in the precipitates (Roitt et al., 1958; Shulman and Witebsky, 1960b). In contrast to the earlier studies of Stokinger and Heidelberger (1937), in which it was shown that the precipitates formed by rabbit antibody and human thyroglobulin gave molecular ratios ranging between 4O:l and 60:1, it was now observed that with human autoantibodies to human thyroglobulin there were ratios ranging between
THYROID ANTIGENS AND AUTOIMMUNITY
149
2 : l and 6:l. These very much smaller values suggested that the autoantibodies are directed against more restricted portions of the thyroglobulin molecule than are the heteroantibodies. It was emphasized by Shulman and Witebsky (196Ob) that we must revise our concept that thyroglobulin can be autoantigen and, as a new hypothesis, state more precisely that certuin portions of the thyroglohulin molecule can be autoantigens. It was, thus, suggested from these observations that of the approximately fifty antigenic determinant groups which might exist on each thyroglobulin molecule, only about six of these would be immunogenic in the stimulation of autoantibody (Shulman and Witebsky, 196Ob; Shulman, 1968, 1969). It should, therefore, be possible to fragment the thyroglobulin molecule in such a way as to isolate molecular fragments which would only show interaction with heteroantibodies, in contrast to other fragments which would react either with autoantibodies alone or with both autoantibodies and heteroantibodies. Studies of this sort have been made and will be further discussed below (Section VIII,C,l). As yet another aspect of variation in the precipitation process, we may consider the important, but long neglected, clear-line phenomenon. Goudie et al. (1959a) showed that some thyroiditis sera not only fail to precipitate when tested in gel diffusion but actually show a clear zone in the agar. If such a serum is placed in a well adjacent to one containing a precipitating serum, however, then precipitation may be seen instead of the clear line. Goudie (1960) observed in immunoelectrophoresis that the antibody giving a clear-line arc had a higher mobility than that giving a conventional arc. In this regard, he considered the antibody to be similar to that studied by Kuhns (1954) in work on nonprecipitating skin-sensitizing diphtheria antitoxin of human origin. The clearline effect that Goudie discovered for thyroiditis sera was largely ignored until 1968, when it was incorporated by Roitt et al. (1968) into a precipitation hypothesis, which will be discussed later (Section VIII,C,2). Another quite different type of antibody activity, and one which is important for guiding us to some basic findings, is that of cytotoxicity. To examine the cytotoxic effect of the autoantibody, thyroid cells are studied in cell culture. A method for cultivation of these cells was developed by Pulvertaft et al. (1959b). The attack of thyroid autoantibodies on such cells has been described by Pulvertaft et al. (1959a, 1961), Irvine (1960a,b, 1962), Forbes et al. (1962), Chandler et al. (1962a), Rose et al. ( 1965b), and Kite et al. ( 1965). In the presence of complement, human antibody may destroy cells after 24 or 48 hours of incubation at 37"C, and this effect can be roughly titrated. The cytotoxic effect occurs only with (certain) human thyroiditis sera, but has not
150
SIDNEY SHULMAN
been demonstrable in rabbit antisera. This human antibody destroys rhesus monkey thyroid cells as well as human thyroid cells (Kite et al., 1965). It was also found that rhesus monkeys could be immunized to produce thyroid autoantibodies, and these sera included cytotoxic antibodies (Kite et al., 1965). The distinction between primate and rabbit autoantibody in this respect seems to parallel the fact that certain human patients show an autoantibody for complement fixation that reacts with a microsomal particulate antigen. This kind of microsomal antibody has not been detected in the rabbit, despite repeated efforts of immunization with thyroid homogenate (Rose et ul., 1965b). It might, thus, seem that the same antibody is revealed by cytotoxicity testing and by complement fixation. However, this is not so, as is shown by several findings. Chandler et al. (1962a) found that the immunofluorescence test for cellular antigen did correlate with cytotoxicity but not with complement fixation. Pulvertaft et ul. (1959a) and Forbes et al. (1962) showed that the microsomal antigen, as revealed by immunofluorescence, was related to the cytotoxicity antigen. Kite et al. ( 1965) showed an imperfect correlation between cytotoxicity titer and complement fixation titer, and they suggested that these antibodies are different. Further indications of the distinction were found in the time course of appearance of different antibody activities in the immunized monkeys. These animals first developed hemagglutinating antibody. Later, complement-fixing antibody and cytotoxic antibody developed at different times; however, the cytotoxic antibody has been reported to develop both later (Rose et al., 1965b) and earlier (Rose and Witebsky, 1969) than the other type of antibody. Other methods of antibody detection have also been used. The procedures of immunofluorescence have frequently been applied, as has been discussed above in several contexts. The method of passive cutaneous anaphylaxis of the guinea pig was successfully applied by Ovary et ul. (1958). Methods of mixed agglutination and of mixed hemadsorption have already been mentioned (see Section VI,B,2). Another method of interest is the electrophoretic migration of radiolabeled hormone (Premachandra et al., 1963a).
2. Immunoglobulin Forms of Anti-thyroglobulin Many reports have appeared on the molecular nature of the autoantibody, and much contradictory literature has developed on this question. It has been claimed by some that this human antibody is a 1 9 s molecule, whereas others have claimed it to be a 7 s molecule. Pressman et al. (1957) found a 1 9 s antibody in one serum, whereas Korngold et al. (1959) found 7 S antibody in three sera and 19 S in a fourth.
THYROID ANTIGENS AND AUTOIMMUNITY
151
Fahey and Goodman (1980) described 5 patients who had 7 S antibody and 2 others who had both forms of antibody. Shulman and Witebsky (1960b) reported for two sera that the antibody was mostly in a 7s form, but they felt that there probably were small amounts of 19s antibody. In the study of Goodman and his colleagues already mentioned (Section VII,A,l), it was found through the method of radioimmunoelectrophoresis that human antibody could also be a yA immunoglobulin, and so this antibody activity was shown to be in each of the three major immunoglobulin classes, namely, yG, yA, and yM. Torrigiani and Roitt (1963)studied several sera from patients, fractionating them into 7 and 19 S fractions, and recovered antibody activity in each case in the 7s fraction. These studies of human thyroiditis sera are confusing. With these sera, one must recognize that the primary immunizing stimulus was not known. Hence, it could not be ascertained with certainty how long after the primary stimulus these antibodies were formed. Furthermore, genetic differences might play a role among different individual patients. Torrigiani and Roitt (1963) also reported one preliminary experiment in the rat; they found that the autoantibody was mostly of 7s type. A detailed study in the rabbit was made by Ghayasuddin and Shulrnan (1967) and Shulman et al. (1968). In a group of rabbits immunized with rabbit thyroid extract and complete Freund's adjuvant, the sera collected at numerous intervals were evaluated by passive hemagglutination and gel digusion. Several late bleedings (at 53 to 75 days) were fractionated and the fractions tested; there were low but significant levels of 1 9 s autoantibody, in addition to the very high levels of 7s autoantibody. In another group of rabbits, early bleedings (at 7 to 14 days) showed significant, although low, amounts of 19s activity, with a higher level of 7s activity. Hence at the start, both immunoglobulin classes are formed with this activity; later, the 7s class is greatly predominant. Torrigiani et al. (1968), using a method of coprecipitation with radioactive thyroglobulin, found yG most often, yA in about half, and yM rarely, as the autoantibody in Hashimoto sera. B. MECHANISMS IN AUTOIMMUNITX
1. Antibodies oersus Cells Our earlier discussion on the role of delayed hypersensitivity (Section V,C) has reviewed many of the reports and arguments concerning the pathogenesis of experimental thyroiditis. The major argument is between the opposing views of whether the tissue damage is caused by circulating antibody or by cellular hypersensitivity. The previous
152
SIDNEY SHULMAN
discussion pointed out that an increasing weight of opinion had mounted to support the cellular factor as the important causative agent. The antibody was considered by most, if not all, workers to be essentially irrelevant in pathogenesis. Some of the older data in the literature had already suggested, on the other hand, that serum antibodies can be a possible trigger factor in the development of experimental thyroiditis (Roitt et al., 1962). It had been stated, for example, that even though passive transfer of antibodies to normal animals is ineffective, a pretreatment with Freund's adjuvant does allow injected antibody to cause thyroid lesions. It has now been reported that rats injected intraperitoneally with rabbit antirat thyroid serum 3 days after being injected intradermally with Freund's adjuvant did develop thyroiditis. However, this process could not be repeated using autoantibody, and it is considered doubtful whether one could make interpretations on the mechanism of autoimmune thyroiditis from experiments with heteroantibodies. It has also been emphasized that sera from humans with thyroiditis are often cytotoxic to thyroid cells in culture (Section VII,A,l), It was suggested that cytotoxic antibody can produce injury in the thyroid gland as a second step after the section of sensitized cells that produce an initial damage of the cell surface and permit the microsomal antigen to be reached (Roitt et al., 1962). However, other modes of action might be important for this kind of antibody, and the direct attack on the cell should not yet be ruled out. The more recent work of McMaster et al. (1967), with regard to the significance of the immunizing dose and of the genetic factors, has already been discussed (Section V,B). In that same investigation, some degree of doubt was cast on the concept that delayed hypersensitivity was the mechanism to be correlated with the appearance of thyroiditis. The experiments showed that delayed hypersensitivity developed in animals that had been given immunizing doses too small to produce any disease. The authors also pointed out that in certain earlier studies, delayed hypersensitivity in a group of experimental animals gradually diminished until it became undetectable within a year after immunization, in spite of the fact that the disease itself persisted throughout a second year. It may, then, be true that an association between delayed hypersensitivity and thyroiditis can be found only if the tissue damage results from more than one type of immune response factor.
2. Attempts at Passive Transfer In very recent studies the significance of circulating autoantibody has once more been emphasized, after many years of disinterest in this possibility. Let us briefly review the arguments that have been made
THYROID ANTIGENS AND AUTOIMMUNITY
153
for the lack of significance of autoantibody as a causative agent. The following five points have been considered in various reports, although not in combination. They summarize the major arguments that generally prevailed, at least until 1969. 1. The antibody titers do not correlate well with the severity of the lesions. 2. Passive transfer by transfer of serum from one animal to a normal animal has consistently failed to produce a transfer of lesions, 3. One can dissociate the formation of autoantibody from the induction of tissue lesions. This has been done in thyroiditis studies by the use of alum adjuvants, on the one hand, in contrast to Freunds complete adjuvant, on the other. It has also been accomplished in some cases by the use of incomplete Freund's adjuvant in contrast to the complete adjuvant. 4. It has often been reported that the severity of lesions does correlate well with the results of skin tests and, thus, with the phenomenon of delayed hypersensitivity. 5. Passive transfer of the disease has been produced by transfer of lymphoid cells and this has worked fairly well. For example, McMaster and Lerner (1967) transferred lymph node cells in histocompatible guinea pig (strain 13) and succeeded in producing thyroiditis in 28%of the recipient animals. Nakamura and Weigle (1967a) succeeded with a similar procedure in rabbits. Until quite recently, the evidence and opinions had accumulated overwhelmingly to support the concept that in thyroiditis and in most other autoimmune diseases the basic mechanism for producing tissue damage is related to delayed hypersensitivity. However, even though all the above arguments indicate strongly that serum antibodies are not the causative agents and, instead, indicate that lymphoid cells and a mechanism of delayed hypersensitivity are the causative factors of tissue damage, one must consider certain detracting statements to each of the above observations. 1. The antibody titer is a total of many classes and subclasses of immunoglobulins and these have different -and possibly opposed functions. Some of them may perhaps be protective, whereas others in the same antiserum may perhaps be cytotoxic, and others may be neither. 2. Perhaps the failure in previous studies to transfer the disease by the transfer of serum has been owing to the fact that this transfer has not been correctly done. 3. Autoantibodies can, indeed, be formed without the formation of lesions, but this does not remove the possibility that they may still be involved in the formation of lesions. 4. Some of the reported studies indicate that no better correlation is
154
SIDNEY SHULMAN
obtained between delayed hypersensitivity and lesion severity than between antibody titer and lesion severity. 5. Even though lymphoid cells have been successful in producing passive transfer of the disease, one would like to see an even higher rate of success. A figure of 28%is not high, in this regard. Nakamura and Weigle (1969) found that serum from immunized rabbits could be transferred in such a way as to induce thyroiditis in approximately 50%of recipient normal animals. Actually, three factors were of importance in this regard. First, it was necessary to use serum samples taken in a series of bleedings and to transfer these to recipient animals on the same time schedules. Second, it was necessary to thyroidectomize the donor animals before immunizing them. The thyroidectomy presumably prevents the soaking-up and removal of the significant autoantibody from the circulation. Third, it was necessary to use early bleedings and not late bleedings. The best bleedings were those taken during the first 2 weeks after the first injection, and the ineffectual bleedings were found at approximately 2 months after the injection. By these means, the transfer of disease was achieved in approximately 50% of recipient animals. It has been mentioned (Section VII,A,2) that Ghayasuddin and colleagues had investigated the sequence of formation of 19 and 7s autoantibodies to thyroglobulin in different bleedings from immunized rabbits. It was shown that 1 9 s antibody was present at early times, compared to a simultaneous higher level of 7s antibody. In late bleedings, in contrast, the 1 9 s form was in extremely low content in comparison to the other. We now see from the work of Nakamura and Weigle ( 1969) that these different immunoglobulin classes may harbor quite different activities in regard to cell injury, It must be concluded that very likely both factors-autoantibody and delayed hypersensitivity-must be involved in the induction of experimental thyroiditis. Studies of the phenomenon of experimental aspermatogenesis have also led to the conclusion that both of these mechanisms are involved. Ruddle and Waksman ( 1!368a,b,c) have investigated the mechanisms for production of tissue damage in delayed hypersensitivity. As an additional argument, it should be emphasized that the contribution of circulating antibodies may be falsely evaluated if only the total titer of antibody can be measured, for this total is a sum of many classes and subclasses of immunoglobulins which may have quite different functions. Even if a particular antibody population played a pathogenic role, comparison of the total titer with the lesion severity would obscure this activity. This would be made all the more critical if it just happened that the one pathogenic type of antibody were also
THYROID ANTIGENS AND AUTOIMMUNITY
155
avidly removed from the serum by the antigens in the target tissue, while all the other antibodies against the same antigen remained in circulation. VIII. Chemical and Antigenic Structures of the Thyroglobulin Molecule
A. PHYSICOCHEMICAL AND BIOCHEMICAL CHARACTERIZATION Thyroglobulin has been purified to a highly satisfactory degree in a number of preparations obtained from several animal species. Despite these achievements, there remains more difficulty than with most other purified proteins in probing the detailed structure of the molecule, for at least two reasons. One is the content of iodine and its variability. The other reason, related in part to the first one, is the microheterogeneity that is demonstrable in even a highly purified preparation of this protein, This heterogeneity is easily shown biochemically and is strongly indicated by some of the antigenic properties. It is, therefore, useful to have some fundamental criteria for defining the thyroglobulin molecule or group of molecules, and the criteria that are currently rather widely accepted are as follows (Robbins and Rall, 1962): 1. Thyroglobulin contains iodine. 2. Thyroglobulin has a sedimentation coefficient of 19 S. 3. Thyroglobulin has an electrophoretic mobility at pH 8.6 which resembles that of a serum a-globulin. 4. Thyroglobulin is salted out with ammonium sulfate at approximately 40%of saturation.
1. Evidence for Microheterogeneity Studies on heterogeneity within this type of apparently homogeneous material have been based most frequently on characteristics of the iodine incorporation, that is, on the varying contents and chemical activities of the iodinated amino acids in different subfractions of the material-subfractions which otherwise may seem to be identical protein materials. This type of analysis has been performed in a variety of ways and can be illustrated in reports by Pitt-Rivers (1963) and Bouchilloux et al. (1964). The first report described the release of radiolabeled monoand diiodotyrosines from autolyzing rat thyroid homogenate. At short times after the labeling, the rate of release of these 1311-iodotyrosines was greater than at later times. This indicated a heterogeneity of the molecules in terms of the biosynthesis of the iodotyrosines and their lability in the protein structure. The other report described the eluted fractions of thyroglobulin obtained by chromatography on DEAE-
156
SIDNEY SHULMAN
cellulose, from thyroid glands taken at various times after injection of radio-iodine. The first eluted fraction showed the highest specific activity of radioactive iodine, although it was the poorest in total iodine. These gradations in properties also indicated a heterogeneity in molecular structure. There had been earlier studies, already mentioned, which had demonstrated the concept of microheterogeneity of thyroglobulin, based on salting-out differences, or in other words, solubility differences at high salt levels (Derrien et aZ., 1948). Another approach to the study of this heterogeneity was described by M. J, Spiro (1961) and by Assem et al. (1965), in studies with starch gel electrophoresis as well as with various chromatographic separations of reasonably well-purified human thyroglobulin prepared by salting-out procedures. Their results indicated that several components were present in the thyroglobulin preparation. They believed that the three components seen in the starch gel electrophoresis should all be identified as thyroglobulin with regard to all other criteria. Ingbar et al. (1959) had been the first to use DEAE-cellulose to show a difference in iodine/nitrogen ratios among the different fractions of material that were all presumably purified thyroglobulin. Similar reports had been made by Ui et al. ( 196l), who also demonstrated on DEAE-cellulose that there were different degrees of iodination in the thyroglobulin fractions, and by Shulman and Stanley (1961), who also showed a chromatographic heterogeneity with regard to the composition of iodinated amino acids in the different subfractions. The heterogeneity of thyroglobulin was similarly studied by Robbins (1963) and Robbins et al. ( 1966), who explored the fractions obtained from a thyroglobulin preparation after being eluted from DEAE-cellulose and evaluated the ratios of the various iodinated amino acids. He found that certain ratios showed progressively changing values in the successive fractions. The heterogeneity of thyroglobulin can also, however, be indicated in terms of antigenic determinant groups of several kinds, and it is this type of analysis which is of greatest interest with regard to the immunochemistry of this molecule. These types of studies will be discussed at a later point (Section VII1,C). Before leaving this subject, we should refer to types of thyroglobulin which are really quite abnormal, in contrast to the presumably narrow range of properties that is the basis for heterogeneity. One such example is seen in the very low level of iodination of some preparations. A number of workers have isolated essentially noniodinated thyroglobulin from certain carcinomas and adenomas of the human thyroid gland. Similar products can be obtained from experimental animals after treatment
THYROID ANTIGENS AND AUTOIMMUNITY
157
with antithyroid drugs. These preparations generally do contain a very low level of iodine, but far below the normal value of about 0.5%; perhaps they should be termed hypoiodinuted. One of the early reports that described an unusually low iodine content in a human thyroglobulin preparation was that of Stanley (1964b), who studied the prqteins of normal and pathological portions of human thyroid glands containing discrete malignant nodules. He found that the iodine content of thyroglobulin from the malignant portions was lower than that from the normal portions, and, in one case, it was at a negligible level. Similar findings have been reported quite recently, giving more detailed information (Valenta et al., 196813). Here, it was again shown that thyroid carcinomas and adenomas contain thyroglobulin in amounts which are 0.1-0.001 of that found in normal thyroid tissue. This has been taken to indicate that there is a block in the iodination step. In addition there is a decreased synthesis of the thyroglobulin protein itself. There are strong indications that the varying levels of iodination are responsible for conformational and shape changes in this molecule. We would expect from this fact that there would be antigenic alterations as a result; however, no evidence has yet been obtained concerning the influence of these conformational changes on the antigenic groups or on their degree of heterogeneity in a given preparation. Nevertheless, such influences may well be anticipated, and studies should be designed in the future to study the antigenic properties as a function of the iodination of this molecule.
2. Molecular Size and Shape We can report on several of the more recent determinations of the physicochemical parameters of thyroglobulin. Only a brief survey will be given here; much more detailed discussions were presented by Edelhoch and Rall ( 1964) and Edelhoch ( 1965). Edelhoch (1960) obtained and studied a well-purified preparation of bovine thyroglobulin. The sedimentation coefficient of this protein was found to be 19.4s. This was in close agreement with the value that had been reported by O'Donnell et al. (1958) for hog thyroglobulin. A diffusion coefficient of 2.49 x lo-' was reported, slightly smaller than the value of 2.60 )( lo-' reported by ODonnell et al. (1958) and by Derrien et al. (1949) for hog thyroglobulin. By using a partial specific volume of 0.713, a molecular weight of 669,000 was then calculated from the Svedberg equation (Snell et al., 1965). This value for bovine thyroglobulin was in close agreement with the already reported values for hog thyroglobulin of 650,000 (Derrien et al., 1949; ODonnell et al., 1958). The molecular weight was also determined by means of light scattering, and a result close to 690,000 was obtained; this, however, is a weight-average
158
SIDNEY SHULMAN
molecular weight, which can be strongly influenced by small amounts of higher molecular weight material. The shape of the unaltered molecule has been explored in a number of studies. Many conclusions have been drawn from use of the sedimentation-diffusion method, From these measurements, a frictional ratio, f/fo, can be calculated; a value of 1.5 has generally resulted from this molecule ( Heidelberger and Pedersen, 1935; Edelhoch, 1960). One must then make some assumption as to a molecular model and the shape can be described on this basis. If one assumes that the form is a prolate ellipsoid of revolution, an axial ratio for this molecule of about 9 / 1 is obtained. This would be a maximum figure, assuming that the particle is not hydrated. It may be noted that judgments have been made from viscosity data also, and these have indicated a molecule with relatively slight asymmetry and/or hydration (Edelhoch, 1960). In recent studies, the method of fluorescence polarization was used, and the results suggested that the molecule behaves as a compact and rigid particle (Steiner and Edelhoch, 1961). Direct observation of the molecule has been made by electron microscopy. Jakoby et al. (1966) studied the crystallization and association of the molecule in this way. Bloth and Bergquist (1968) showed that the isolated molecules have the shape of a flexible helix with two turns, having a total length of 220 A. and maximal diameter of 110 A. Thus, the molecule has a much smaller axial ratio than was indicated by hydrodynamic measurements, but it is also much less compact. Figure 14 illustrates some of the quite striking electron-microscope visualizations from their study, along with the pictures of their structural models for the thyroglobulin molecule. Returning to physicocheniical studies, we must note that Edelhoch (1960) described a series of reversible changes of shape in the thyroglobulin molecule, as influenced by the pH and ionic strength of the solution. With increasing pH from 6.0 to 12.7, there are several changes, and at least four new molecular species are formed in a sequence of progressive disorganization of this molecule. Although the relative proportion of 1 9 s material falls, there is at first an increasing quantity of a 1 2 s boundary, and, at higher pH levels, there appear 15, 8, and 3s components. The earlier alteration is a type which involves the formation of a particle which has sedimentation properties similar to the aprotein of Lundgren and Williams (1939). The data were interpreted to indicate that this product-this 12 S component-results from the dissociation of the thyroglobulin molecule into two subunits. This halfmolecule has been discussed above ( Section III,B,4). In a more intensive alteration, produced by a pH above 9.5, two additional sedimenting
THYROID ANTIGENS AND AUTOIMMUNITY
159
FIG. 14. Electron micrographs of selected thyroglobulin molecules. It can be seen that the molecule has the shape of a flexible helix with two turns; the pictures can be compared with the pictures of models that were constructed and are shown below each micrograph. The scale lines represent 100 A. Magnification: X400,OOO. (From Bloth and Bergquist, 1968. Photograph kindly provided by Dr. Bloth.)
components appear, with sedimentation rates of 15 and 8s; these both seem to have lower molecular weights than thyroglobulin. At the highest pH levels, the only constituents that seem to remain are the 8 and 3 S materials. Except for the 3 S component, which must represent a disrupted fragment, these various forms were thought to be the result of rearrangement of subunit packing, with perhaps some splitting off of parts of the subunits.
B. SU~UNIT STRUCTURES OF
THE
MOLECVLE
1. Polypeptide Structure Relatively little work has yet been done with regard to the amino acid composition and the arrangement of the monomer units into the polypeptide chains of this large and complex molecule. Several reports have appeared on the total amino acid composition of thyroglobulin preparations, including those of Rolland et al. (1966) and M. J. Spiro (1970); some of the most recent data are tabulated in Table VII. There are approximately 5500 amino acids per molecule, representing 87-91% of the glycoprotein mass, and the remaining 8-10% is composed of carbohydrate (about 300 monosaccharide units) and iodine. Of the amino acids, about 20% are dicarboxylic amino acids, half of which are in the amide form, and about 10%are basic amino acids. It should be noted that the content of thyroxine is only about 3 or 4 residues per molecule, a fact that has been known and pondered on
160
SIDNEY SHULMAN
TABLE V I I AMINO ACID COMPOSITION OF THYROOLOBULINS FROM SEVERAL SPECIES (RESIDUES PER MOLECULE)"^^ Amino acid
ASP
Thr Ser Glu Pro GlY Ala Val Met Ile Leu TYr Phe LYS His Arg 3-cys Try MITO DITd T4
1 (%I
Man
Calf
Sheep
Pig
400 283 484 697 347 394 383 310 62 155 476 110 259 195 80 275 244 117 7 7
401 266 502 697 392 422 474 338 56 155 506 115 269 177 76 345 248 109 10 16 3 0.9
385 270 494 660 365 417 456 329 72 136 489
368 289 482 673 412 425 498 327 54 131 541 101 251 155 77 371 233 101 11 14 6 1.1
4 0.4
118
257 164 66 326 248 112 8 9 4
0.7
From M. J. Spiro (1970). Based on a molecular weight of 670,000. c (MIT) monoiodotyrosine. d (DIT) diiodotyrosine. 0
b
for some time. It seems remarkable that so few residues of this allimportant amino acid are created in the total wealth of residues per molecule. The possible contribution of thyroxine to antigenic properties of the molecule must be similarly infrequent. On the other hand, the mechanism of action of the thyroxine on the molecular level has intrigued many investigators; Gruenstein and Wynn ( 1970) have recently offered a scheme based on modification of cell membranes. The possibifity has been suggested that the thyroxine residue in the thyroglobulin may be an important determinant in the antigenic specificity of this protein (Churchill and Tapley, 1964). Proposals have been made regarding a role for this amino acid in the autoantibody response to thyroglobulin (Premachandra et al., 1963a; Margherita and Premachandra, 1969; Premachandra, 1970). Apparently the first attempt to study the physicochemical nature of
THYROID ANTIGENS AND AUTOIMMUNITY
161
molecular subunits appeared in the report by Pierce et al. (1965). These workers prepared highly purified human thyroglobulin, both from normal thyroid glands and from those of subjects with nontoxic colloid goiters. No significant differences in composition were found with regard to the noniodinated amino acids between the thyroglobulins from each of these two sources. Furthermore, the total amount of carbohydrate and the various types of sugar residues were found to be essentially identical. There was, however, a difference in the iodine content, inasmuch as the normal thyroglobulin contained 0.32% iodine and the goitrous thyroglobulin contained 0.07%iodine. Amino acid analysis was reported in some detail on these two preparations, as well as on preparations of hog and sheep thyroglobulin. An interesting point was the finding of approximately 180 half-cystine residues per molecular weight of 660,000. It was rather striking that all these half-cystine residues were apparently involved in disulfide linkages, since preparations that were treated with iodoacetate in 8 M urea, but without any chemical reduction, indicated less than 2 residues of S-carboxymethylcysteine per 660,000 molecular weight. There were, therefore, approximately ninety disulfide linkages holding the molecule in its configuration. When chemical reduction was performed with the use of mercaptoethanol, followed by alkylation with iodoacetate, the resultant product showed slow-sedimenting components in the ultracentrifuge. Although there were some problems of polydispersity, there was a frequently seen component with a corrected and extrapolated sedimentation coefficient of 3.2 to 3:4S. The molecular weight of this dissociated species was estimated to be approximately 110,000-125,000, and this indicated five or six subunits or polypeptide chains in the original thyroglobulin molecule. Although these peptide chains may be physically identical, that is, have very similar molecular weights, they are not necessarily identical chemically, and further studies will be required to clarify this point. It should be explained at this point that, although other investigators have concurred in finding essentially all the cysteine units in disulfide form, the actual number is now accepted as being about 11 or 12% higher than that reported earlier. The half-disulfide content has been given as 200 (Lissitzky et al., 1964) and as 202 (de Crombrugghe and Edelhoch, 1966; de Crombrugghe et al., 1966; Pitt-Rivers and Schwartz, 1967), and even as high as 208 (Rolland et al., 1966) and 240-250 (M. J. Spiro, 1970). A report by Pitt-Rivers and Schwartz ( 1967), mentioning the 202 half-cystine residue per molecule, described a careful study on the very small number of free sulfhydryl groups; they concluded that there were 5 such groups, only 2 of which are accessible without some denaturation.
162
SIDNEY SHULMAN
Other laboratories have also explored this question of the subunit structure of the thyroglobulin molecule. Several reports have been made by Edelhoch and colleagues (de Crombrugghe and Edelhoch, 1966; de Crombrugghe et aZ., 1966; Edelhoch and de Crombrugghe, 1966; Nissley et at., 1969) and by Lissitzky and colleagues (1965, 1968b). A subunit molecular weight of about 165,000 was found by the former group. They obtained a product from complete reduction of bovine thyroglobulin in 6 M guanidine or 8 M urea, and the molecular weight was estimated by a combination of sedimentation and viscosity data. The molecular properties of the reduced molecule were studied by a number of physical techniques, and a rigid, two-chain, highly permeable, spongelike structure was proposed. Reversibility of the reduction process was explored and considerable success was achieved. Recent studies of Edelhoch et at. have emphasized the formation, by mild reduction, of two sedimenting units; these were separated and further analyzed, although no new light was shed on the controversial question of the minimal subunit molecular weight. A rather painstaking approach to this problem has been taken by Lissitzky et aZ. (196813). On a fully reduced and alkylated sheep thyroglobulin, using a variety of methods for molecular weight determination, they obtained, under some conditions, values of 80 to 85,000, and under other conditions, 114,000 or 160,000. They concluded that the 80,000 value was the result to be most favored. They also felt that there were two different kinds of polypeptide chain of 80,000 molecular weight to be assembled, in suitable numbers, in the thyroglobulin molecule. Study of terminal amino acids has also been attempted, but has met frustrating difficulties with a number of partial residues being found, reminiscent of the early work on the immunoglobulins. This information is important for the better elucidation of the number of polypeptide chains, and it also will contribute to sequence studies. Dopheide and Trikojus (1964) found, for hog thyroglobulin, two residues of aspartic acid (plus asparagine) and one of glycine. Rawitch et al. (1W8) studied other sequence portions of the molecule. The most recent analysis is the thorough study of M. J. Spiro (1970). She found evidence for four chains in the thyroglobulin of each of four species. There were always two units of aspartic acid; a third terminal unit was leucine in sheep and calf, glycine in pig, and serine in man. Finally, in each species, a fourth unit is indicated by partial residues of several amino acids which give a total corresponding to approximately one unit, Other approaches have been taken to separating the chain units in the molecule, especially by nonenzymatic means. One method is that of
THYROID ANTIGENS AND AUTOIMMUNITY
163
treatment with succinic anhydride. Rolland and Lissitzky (1970) reported a dissociation of succinylated 199 S thyroglobulin (of sheep) into 1 2 s subunits, and the relationship of this transformation to the iodine content; some comparisons were made with a human goiter thyroglobulin. D. J. Smith and Shulman (1971a) described several slowersedimenting species among the succinylation products of normal and goitrous human thyroglobulin. At a moderate molar ratio of reagent to protein, both types of protein showed a minor 1 6 s peak and two slow species-a major 14 S peak and a minor 10 S peak-suggesting that there are two stages of molecular disruption.. The proportions changed at higher molar ratios. The goiter-derived, hence poorly iodinated, protein was more sensitive to dissociation than was the other protein, and it was more completely broken down to the 1 0 s component. Studies have also been reported on another method for the dissociation of the thyroglobulin molecule into subunit forms, namely, by means of sodium dodecyl sulfate. It appears that the degree of completeness of dissociation in this way is also related to the degree of iodination of the molecule. From this observation, it was concluded by Tarutani and Ui ( 1968, 1969a,b) that noniodinated thyroglobulin has structural differences from the iodinated form. Andreoli et al. (1969) discussed noncovalent subunit structure of human thyroglobulin, as indicated by the dissociating effects of alkaline pH and low ionic strength. From all these various investigations, it has been generally concluded that thyroglobulin consists of two types of molecules, that is, nondissociable and dissociable molecules, depending on the presence or (virtual) absence of iodine. These may yet turn out to be extreme cases, with intermediate degrees of dissociability for some species of the molecules. Nunez et al. (1965, 1966) described some data that indicate, on the basis of sedimentation analyses, a relationship of molecular conformation to iodination. This again corresponds to a heterogeneity in the molecular species, They found, as had others, that poorly iodinated or newly synthesized thyroglobulin would sediment 1 or 2 sedimentation coefficient units more slowly than did the so-called mature thyroglobulin, as judged from density gradient studies. Valenta et al. (1968a) also discussed this question, although emphasizing some negative correlations. Edelhoch et al. (1969), however, again provided some evidence for conformational changes as a function of iodine content. It is not yet known to what extent these differences affect the antigenic properties, especially with regard to autoantigenic or heteroantigenic forms. Many exciting new leads can be expected to arise from such explorations, clarifying the detailed architecture of the various antigenic units as chemically defined structural groups.
164
SIDNEY SHULMAN
2. Carbohydrate Structure It is known that thyroglohulin is a glycoprotein, containing in the carbohydrate portion, galactose, mannose, fucose, glucosamine, and sialic acid (Ujejski and Glegg, 1955; Wollman and Warren, 1961; Robbins, 1963; R. G. Spiro, 1963; R. G. Spiro and Spiro, 1963, 1965). The carbohydrate content varies slightly, ranging from 8 to 10%for human, ovine, bovine, and porcine types of thyroglobulin ( McQuillan and Trikojus, 1986). Analytic studies indicated the existence of two types of carbohydrate residue which could be separated according to their dialyzability properties, among others; each so-called type was a mixture of glycopeptides. These glycopeptides are released by the use of the proteolytic enzyme, pronase, obtained from Streptomyces griseus. Cheftel et al. (1964) described a nondialyzable glycopeptide fraction of sheep thyroglobulin, which contained all five sugar units, and which was estimated by ultracentrifugal examination to be fairly homogeneous and to have a molecular weight of approximately 3500. Electrophoretic studies, however, revealed a mixture of ten glycopeptides in this fraction. The dialyzable fraction, on the other hand, was estimated by them to have a molecular weight of about 1700. Unlike the larger peptide, it contained no sialic acid. A separation of this fraction into two or three glycopeptides was indicated. Narasimha Murthy et al. (1964) isolated and purified a major glycopeptide from sheep thyroglobulin. This peptide contained at least twelve amino acids, and the N-terminal residue was identified as aspartic acid. Therefore, this amino acid seems to be the connecting link of this oligosaccharide to the polypeptide framework of the thyroglobulin molecule. This glycopeptide also contained sialic acid, glucosamine, galactose, mannose, and fucose; it seemed to account for 60%of the carbohydrate of thyroglobulin and contained little or no iodine. The molecular weight of this highly purified glycopeptide component was then determined by the Archibald procedure and by the sedimentation equilibrium method, and the value was found to be 2400 (Narasimha Murthy et al., 1965). The molecular weight was also estimated by means of chemical analysis for aspartic acid; this result was 2600. The analysis of amino acids in the most highly purified preparation revealed the presence of only six amino acids, namely, aspartic acid, alanine, glycine, serine, threonine, and glutamic acid. The carbohydrate analysis gave results that indicated that with a molecular weight of 2600, there were present in 1 mole of the peptide, 2 residues of glucosamine, 5 of hexose, 0.5 of fucose, and 1 of sialic acid. If the total carbohydrate of thyroglobulin were to exist as a.single entity, the glycopeptide would have to have a molecular weight
THYROID ANTIGENS AND AUTOIMMUNITY
165
of approximately 50,000 since there is about 8% carbohydrate in a molecule of molecular weight about 650,000. Since the major glycopeptide that has been isolated has a molecular weight of only 2600, there would have to be approximately 20 units of this type distributed throughout the molecule, if all the carbohydrate units were identical. This point has been further developed by others, as will now be described. The carbohydrate compositions of bovine, ovine, porcine, and human thyroglobulins were also determined by R. G. Spiro and Spiro (1965). In each of these, the monosaccharide constitutents were galactose, mannose, N-acetylglucosamine, fucose, and sialic acid. The sialic acid was found to be present in the N-acetyl form except in the pig thyroglobulin, in which 10%occurs in the N-glycolyl form. Although the total content of carbohydrate is only 8-10%of the thyroglobulin weight, the large size of this protein molecule causes this relatively small percentage to represent a large number of sugar residues per molecule. Calculated on the basis of a molecular weight of 670,000, the analytical figures represent approximately 290 monosaccharide residues in the proteins from calf, sheep, and pig, and 350 residues in the human protein, per molecule of the glycoprotein. The higher percentage (10%) of carbohydrate in the human thyroglobulin corresponds to its greater content of mannose and glucosamine. In other studies on calf thyroglobulin, R. G. Spiro (1965) described two distinctive types of carbohydrate units which were quite different in size and composition. One unit had a molecular weight of 1050 and consisted of 5 residues of mannose and 1 residue of glucosamine. The other type of unit had a molecular weight of 3200 and consisted of sialic acid ( 2 residues per unit), fucose ( l ) ,galactose (4),mannose ( 3 ) , and glucosamine ( 5 ) . There were thought to be approximately nine of the smaller and fourteen of the larger carbohydrate unit in each molecule of bovine thyroglobulin. This total of approximately 23 units must be attached at twenty-three linkage points in the polypeptide chain. It was judged that at least a large number of these twenty-three amino acids were residues of aspartic acid. In very recent studies by Tarutani and Shulman (1970, 1971a), human thyroglobulin has been treated with neuraminidase, thus removing about 80 to 87%of the total sialic acid. The remainder of these groups were considered to be buried inside the molecule and, therefore, inaccessible. Thus, there seem to be two populations of sialic acid groups. This kind of desialized thyroglobulin was found to be more labile at acid pH than is the intact molecule, as judged especially from ultracentrifugal patterns. With regard to the total group of sialic acid residues in the native protein molecule, chromatography of a purified thyro-
166
SIDNEY SHULMAN
globulin preparation released a series of fractions with variable amounts of sialic acid, and this constituted a variability in parallel with the variation in iodine content. Thus, the heterogeneity in purified thyroglobulin seems to be related not only to the iodine content, but also to the sialic acid content (Tarutani and Shulman, 1971b). Additional studies were made to compare normal and goitrous thyroglobulin and also to explore the effects of galactose removal (Tarutani et al., 1971b). The sialic acid content was different in thyroglobulins prepared from normal and from several goitrous thyroids (Tarutani and Shulman, 1971~).
C. ANTIGENICSTRUCTURES OF THE MOLECULE
It was already mentioned that the thyroglobulin molecule seems to contain two kinds of antigenic determinants, which differ in their reactivity with autoantibody and heteroantibody. In addition, among the heteroantigenic groups, there must be groups characteristic of the homologous species and others that are similar or identical in structure to those on the molecules from other species. These groups could be classified into several categories, as judged from cross-reactions and the effects of absorption. Unfortunately, very little has been done to clarify the structures related to the latter set of criteria. We can, however, discuss some of the findings that refer to the distinctions between auto- and heteroantigenic determinants.
1. Antigenic Properties of Molecular Fragments Based on the hypothesis, described above (Section VII,A,l), that only a small fraction of the antigenic determinant groups on the thyroglobulin molecule are actually autoantigenic determinant groups, several studies have been carried out. In immunodiffusion experiments, Mates and Shulman (1967a) showed that the reaction of heteroantibody with a human thyroglobulin sample produced a spur of precipitate over the reaction of autoantibody with the same antigen; this spur formation would seem to be quite consonant with the hypothesis that has been presented above. This reaction is shown in Fig. 15. There were two rabbit antisera, serving as sources of heteroantibody; these had been well absorbed with normal human serum. This human thyroglobulin sample was 8 5 8 8 % thyroglobulin, along with small amounts of the 27 and 12 S proteins. These results provided an unusual observation of spur formation of two antisera tested against a single antigen, rather than the customary inverse relationship. The pattern seems to be quite compatible with the idea that a larger diversity of antibody specificities is found in the rabbit antiserum than is the case in the human autoimmune serum.
THYROID ANTIGENS AND AUTOIMMUNITY
167
FIG. 15. Double-diffusion precipitation of human thyroglobulin and its antibodies. In the central well is a human thyroiditis serum. The peripheral wells contain: ( 1 and 4 ) purified thyroglobulin; (2, 3, 5, and 6 ) rabbit antithyroid antisera, absorbed with human serum. (From Mates and Shulman, 1967a.)
The comparison of several thyroglobulin preparations indicated certain antigenic similarities in the molecules from different species. Rabbit antisera to human thyroglobulin precipitated with human, monkey, hog, and rabbit thyroid proteins, as well as with various molecular fragments of human thyroglobulin formed by the action of trypsin and with fragments formed by storage at low ionic strength. In contrast to this, a human thyroiditis antiserum, containing autoantibodies, gave no direct precipitation with any of these preparations. The thyroiditis serum precipitated only with human and monkey thyroid proteins but not with those of hog or rabbit. Further, in immunodiffusion, small spurs appeared at the junction of the human and monkey lines, indicating that the human protein contains a larger number of antigenic sites than does the monkey protein, when tested with autoantibody.
168
SIDNEY SHULMAN
Relatively large peptide fragments of human thyroglobulin were prepared by means of partial digestion with trypsin (Mates and Shulman, 1967a). The enzyme activity was stopped after a certain time by the addition of trypsin inhibitor, patterned in part after the report of Metzger et al. (1962). Smaller fragments were also prepared by allowing trypsin to act more completely. After some days of digestion, the reaction mixture was dialyzed to provide a product that was retained in the dialysis bag (dialyzate) and an alternative product in the dialysis bath (diffusate). During the early stages of digestion, both heteroantibody (rabbit) and autoantibody gave precipitation with the digest. After the first 4 hours the autoantibody no longer precipitated, but the heteroantibody continued to show precipitation until a much later state of digestion (52 hours). A number of such comparisons were made. In addition, if a digest failed to precipitate with an antiserum, it was tested as a possible inhibitor of precipitation with the undigested material. The comparisons are shown in Table VIII. These results suggest that there has been a separation of the two kinds of antigenic groups in these fragments. In subsequent studies by means of the passive hemagglutination method, Mates and Shulman (196%) showed that some of the fragments formed from human thyroglobulin after extensive trypsin hydrolysis are able to give some reaction with the human autoantibody, although the same TABLE VIII PRECIPITATION REACTIONS OF' HUMAN AUTOANTISERUM AND RABBITANTISERUM TO HUMAN THYROGLOBULIN (Tg)" Direct precipitation Teat antigen
Human autoantiserum
Thyroid extracts : Human Monkey Hog Rabbit Fragments of human Tg: Low-ionicstrength Trypsin digest, 8 hr 20 hr 52 hr
dialyzate diffusate From Mates and Shulman, 1967a.
* (n.d.) Not done.
+ +
Inhibition of precipitation of Rabbit autoantiserum heteroantiserum with human Tpb
+ + ++ + + + + +
+
n.d.
+ + n.d.
n.d.
THYROID ANTIGENS AND AUTOIMMUNITY
169
fragments were unable to affect the precipitation reactions of the same antiserum. In other studies, the antigenic groups on the 4 and 7 S tissue-specific thyroid proteins were found to resemble those on the 19 S thyroglobulin molecules ( Mates and Shulman, 1969). These comparisons gave further indication that these small molecules may contain subunits of the thyroglobulin molecule. The molecular splitting of human thyroglobulin has also been described by Rose and Stylos (1969; Stylos and Rose, 1969). Some studies were made of nonenzymatic splitting, using chemical reduction. The product was only little characterized physicochemically, but it did show a loss of precipitating reaction with autoantibody and not with heteroantibody. In other studies, digestions with trypsin and several other proteolytic enzymes were performed. It was stated that tryptic fragments incapable of precipitating with autoantibody were still able to precipitate with heteroantibody. Pepsin fragments behaved quite similarly. Papain fragments precipitated with both kinds of antibody, and chymotrypsin fragments precipitated with neither. No extensive characterization of the mixtures of peptides were performed, however, and it is not yet feasible to associate these antigenic activities with definite molecular units. Another group of such studies has been made on rabbit thyroglobulin, using rabbit autoantibodies and goat heteroantibodies for evaluation. An interrupted tryptic digest, containing 91% of a 1.2 S component, showed different patterns of reaction with the two types of antibody. Ghayasuddin and Shulman ( 1968a,b, 1970) separated these peptide fragments by gel filtration, obtaining eight pools; only pools 1 and 2 precipitated with heteroantibody, and only pool 1 precipitated with autoantibody. In more detailed terms, the pool 1 fraction showed four lines of precipitation in gel diffusion with heteroantibody, but only two lines with autoantibody. The lack of precipitation of some of the peptides with the autoantibody was taken to confirm the presence of none (or at most one) of the autoantigenic determinant groups on each of these peptides. At any rate, it seems clear that these fragments carried more heteroantigenic determinant groups than autoantigenic groups ( Shulman and Ghayasuddin, 1971). 2. Nonprecipitating and Precipitating Forms of Autoantibody Studies have been made by Roitt et al. (1968) on the interactions of human thyroglobulin with the different autoantibody forms. They continued the studies of the rabbit- and horse-type precipitating sera, and they revived interest in the clear-line sera. Their experiments indicated,
170
SIDNEY SHULMAN
first of all, that a precipitating antibody was able to react with more sites on the antigen molecule than could the antibody from a clearline serum. This restricted number of reactive sites on the antigen molecule may account for the formation of ncnprecipitating complexes which then somehow clarify the agar to produce the clear line. Second, they analyzed horse-type sera, and showed that a -yG fraction could itself give a precipitin curve similar to that seen for the whole serum. They ruled out the explanation that this phenomenon might be due to the presence of avid nonprecipitating antibodies, along with a smaller amount of precipitating but less avid antibody. They felt instead that the cause was to be found in a low value for the effective valence of the antigen. These investigators constructed a very interesting hypothesis on the basis of four autoreactive sites on the antigen molecule (although some other reports have claimed six) and the concept that the thyroglobulin molecule is symmetrical. They proposed that there are two pairs of these antigenic determinants, having corresponding antibodies AB and ab. They then made the following interpretations. ( I ) The rabbit-type sera would be thought to have comparable amounts of AB and ab; they thus form large complexes even in antibody excess. ( 2 ) Clear-line sera would contain only the AB or the ab, and so the antigen becomes divalent at most, and only linear chains of antigen and antibody can form or, in antibody excess, only small complexes can form, both being nonprecipitating. ( 3 ) The horse-type sera contain both AB and ab, but with a great preponderance of one of them. The description of anticipated complexes can explain the lack of precipitation in antibody excess and the presence of precipitation at higher antigen levels. In essence, this formulation holds that clear-line sera contain essentially only one or the other specific antibody, that horse-type sera contain a preponderance of one or the other, and that rabbit-type sera contain comparable amounts of both types. Through this concept, these investigators explain quite reasonably a number of precipitation phenomena that have been observed with thyroiditis sera. IX. Concluding Remarks
The thyroid antigens are a group of proteins, which can be characterized as having either a soluble form or a membrane-attached form. From a physicochemical viewpoint, the soluble variety includes at least nine distinctive proteins, five of which can be conveniently characterized as sedimentation boundaries in the analytical ultracentrifuge; some of the members of this large group seem to be structurally interrelated and hence may prove to be antigenically related. Several of them are iodo-
THYROID ANTIGENS AND AUTOIMMUNITY
171
proteins and glycoproteins. With respect to tissue specificity, the highly specific activity that is seen is based on the 19s thyroglobulin. This large and complex molecule, when examined from a variety of animal species, shows a complex and inexplicable pattern of antigenic similarities and differences, suggesting that larger or smaller portions of the molecule are essentially the same in certain groups of species and not in others. The corresponding structural units are yet to be elucidated. The fact that some reactions show more than one line of precipitation in gel diffusion may also be consonant with the concept that different portions of this same molecule may be reacting with antibodies of different specificities. These distinctions have been studied with heteroantibodies as reagents. A more incisive tool for investigation is available with the use of autoantibodies, and we will return to this point. The thyroid gland contains potent autoantigens, which can be activated experimentally or in a spontaneous fashion. Such activation leads in general to the production of autoantibodies in the circulation and to thyroiditis in terms of tissue lesions. In this sense, the thyroid is both the source and the target in a process of autosensitization. The major antigen is thyroglobulin, and most of the work, especially the chemical work, has been done with it, although the others may also be of great importance. The induced form of thyroid autoimmunity requires either the concurrent use of an appropriate adjuvant or the use of a chemically modified antigen. In either case, the molecular structure of this molecule apparently must be slightly altered in order to break the tolerance barrier. In any event, the structural change must be sufficiently slight so that there will still be reaction between the autoantibodies and the native antigen in the tissue. It seems probable, in fact, that an adequate degree of molecular alteration may be produced with some changes in only a few of the many antigenic determinant groups and that this renders the molecule autoimmunogenic, with most or even all of the autoantigenic groups remaining perfectly unchanged from their native structure. It is important that in several species, including of course the human, spontaneous thyroiditis occurs. As seen in the chicken and dog models, there are early suggestions in current study that some differences exist in the pathogenic mechanisms of spontaneous and induced thyroiditis; for example, where there is a high genetic predisposition to develop this condition, the effort to induce it seems to encounter additional barriers, compared to other animals of the same species. The thyroid autoimmunity of man differs strikingly from that of other animals (except for primates, apparently) in that several distinct autoantigens are activable. The time has arrived for stating that not only three, but four, autoantigenic components can be operative, since
172
SIDN'EY SHULMAN
the very recent work on LATS has definitely implicated it as an antibody molecule, and one that acts in an autologous fashion; the corresponding antigen is not yet known, but the best opinion at present favors a portion of the thyroid cell membrane for this role. One only waits for a good animal model of this autoantibody. The other three (primate) autoantigens are thyroglobulin ( CA-l), the microsomal particle, and the second colloid antigen (CA-2). There are thus two insoluble ones. With regard to the mechanism of pathogenicity in thyroiditis, two positions have been espoused at different times, namely, that the circulating antibody or the delayed hypersensitivity was the effective agent. Partisans have championed one mechanism or the other (the hypersensitivity, most of the time), but many experimental results that were not quite unambiguous were interpreted in too simple a fashion in order to support one or the other of these mechanisms; there is no doubt that some personal bias entered into these decisions, and, in fact, some investigators have made directly opposite interpretations at different periods of time. The ease of clarifying the mechanisms in thyroiditis has been illusory. With the recent findings that passive transfer with immune serum can be effective for stimulating thyroiditis in normal animals, and with new availability of an animal model for study of cytotoxic antibody, there will be an increased emphasis on the antibody factor, balancing the well-established emphasis on cellular hypersensitivity features. It is likely from all recent reports that both mechanisms can and do operate and that the real question may well be more a matter of which mechanism dominates in a given situation, depending on the species of animal, the condition of the antigen, the nature of any adjuvant, special genetic factors, and other parameters. To return to thyroglobulin as the best characterized autoantigen from the immunochemical standpoint, it is now seen that this molecule of molecular weight about 660,000 can be thought to have four polypeptide chains, although some studies persist in finding a larger number. The molecule is built from two half-molecules of equal size. However, recent data emphasize even more strongly that these two subunits are probably not identical, and so we cannot look for mirror-image sets of antigenic determinants, as is the case in the immunoglobulin G molecule. The most important point to observe is that this molecule has a large number of heteroantigenic determinants (perhaps fifty), but it has only a very limited number of autoantigenic determinants (perhaps four or five or six). Among the large number of other determinants must be found the groups that are sometimes common between the thyroglobulin molecules from different species and the groups that are
THYROID ANTIGENS AND AUTOIMMUNITY
173
distinctive. Among the autoreactive groups must be found the clues for the stimulation of autoantibody formation. This means that the study of structural requirements for the loss of tolerance should now be somewhat simplified by searching out these small specific structures. Until now, the studies of the specific autoantigenic groups have been concerned only with their interactions with humoral antibody. It seems clear that there should be some analogous difference in terms of cellular hypersensitivity and in the induction of tissue inflammation. A new line of investigation could be envisaged in terms of studying various molecular fragments, which may or may not bear these autoantigenic groups, in terms of their ability to engender a hypersensitivity reaction in a sensitized animal and also in terms of exploring their individual abilities to sensitize an animal. Studies of this sort may help to clarify some of the additional factors in the induction of autoimmunity.
ACKNOWLEDGMENTS The investigations done in the author’s laboratories during recent years have been largely supported by a research grant from The John A. Hartford Foundation, Inc. The author is highly indebted to his wife, Joanna, for her helpful criticism and editorial assistance. REFERENCES
Adams, D. D., and Kennedy, T. H. (1967). J. Clin. Endocrinol. Metub. 27, 173. Adams, D. D., and Kennedy, T. H. ( 1971). J . Clin. Endocrinol. Metab. 32, 47. Adant, M., and Spehl, P. (1934). C. R. SOC.Biol. 117, 230. Ager, J. A. M., Hutt, M. S . R., and Smith, G. (1959). Nature (London) 184, 478. Aloj, S., Salvatore, G., and Roche, J. (1967). J. Biol. Chem. 242, 3810. Anderson, J. R., Goudie, R. B., Gray, K. G., and Timbury, G. C. (1957). Lancet 1, 1123. Anderson, J. R., Goudie, R. B., and Gray, K. G. (1959a). Lancet 1, 644. Anderson, J. R., Goudie, R. B., and Gray, K. G. (1959b). Scot. Med. J . 4, 64. Anderson, J. R., Goudie, R. B., and Gray, K. G. (1960). Brit. J. Exp. Puthol. 41, 364. Anderson, J. R., Buchanan, W. W., Goudie, R. B., and Gray, K. G. (1962). J. Clin. Puthol. 15, 462. Anderson, J. W., McConahey, W. M., Alardn-Segovia, D., Emslander, R. F., and Wakim, K. G. (1967). J. CZin. Endocrinol. Metub. 27, 937. Andrada, J. A., Rose, N. R., and Kite, J. H., Jr. (1968). Clin. Erp. Immunol. 3, 133. Andreoli, M., Sena, I., Edelhoch, H., and Salvatore, G. (1969). Arch. Biochem. Biophys. 134, 242. Assem, E. S . K. (1964). Lancet 1, 139. Assem, E. S. K., Trotter, W. R., and Belyavin, G. (1965). Biochim. Biophys. Actu 100, 163. Balfour, B. M., Doniach, D., Roitt, I. M., and Couchman, K. G. (1961). Brit. J. Exp. Puthol. 42, 307.
174
SIDNEY SHULMAN
Becker, K. L., Ferguson, R. H., and McConahey, W. M. (1963). N. Engl. J . Med. 268, 277. Beckers, C., and De Visscher, M. (1961). Metab., Clin. Exp. 10, 695. Beckers, C., and De Visscher, M. (1963). J. Clin. EndocrinoZ. Metab. 23, 149. Beienvaltes, W. H., and Nishiyama, R. H. (1968). Endocrinology 83, 501. Beienvaltes, W. H., Dodson, V. N., and Wheeler, A. H. (1959). J . Clin. Endocrinol. Metab. 19, 179. Belyavin, G., and Trotter, W. R. (1959). Lancet 1, 648. Beutner, E. H., and Witebsky, E. (1962). J. Immunol. 88, 462. Beutner, E. H., and Witebsky, E. (1963). J. Immunol. 91, 204. Bjorkland, A. (1964). Lab. Invest. 13, 120. Blizzard, R. M., Hamwi, G. J., Skillman, T. G., and Wheeler, W. E. (1959). N. Engl. J . Med. 260, 112. Blizzard, R. M., Chandler, R. W., Landing, B. H., Pettit, M. D., and West, C. D. (1960). N . Engl. J. Med. 263, 327. Blizzard, R. M., Tomasi, T. B., and Christy, N. P. (1963). J. Clin. Endocrinol. Metab. 23, 1179. Bloth, B., and Bergquist, R. (1968). J . Exp. Med. 128, 1129. Bouchilloux, S., Rolland, M., Torresani, J., Roques, M., and Lissitzky, S. (1964). Biochim. Biophys. Acta 98, 15. Braley, H. C., and Freeman, M. J. (1969). Clin. Exp. Immunol. 4, 659. Brandes, D., Belt, W. D., and Bourne, G. H. (1959). Exp. Cell. Res. 16, 683. Buchanan, W. W. (1965). Geriatrics 20, 941. Buchanan, W. W., Anderson, J. R., Goudie, R. B., and Gray, K. G. (1958). Lancet 2, 928. Buchanan, W. W., Alexander, W. D., Crooks, J., Koutras, D. A., Wayne, E. J., Anderson, J. R., and Goudie, R. B. (1961). Brit. Med. J . 1, 843. Buchanan, W. W., Harden, R. M., and Clark, D. H. (1965). Brit. J. Surg. 52, 430. Burke, G., and Feldman, J. M. (1965). Amer. J. Med. 38, 470. Chandler, R. W., Blizzard, R. M., Hung, W., and Kyle, M. (1962a). N. Engl. J . Med. 267, 376. Chandler, R. W., Kyle, M. A., Hung, W., and Blizzard, R. M. (196213). Pediatrics 29, 961. Cheftel, C., Bouchilloux, S., and Lissitzky, S. (1964). C . R. Acad. Sci. 259, 1458. Cheng, H. F., Peterson, R. E., and Evans, T. C. (1968). Biochim. Biophys. Actu 168, 161. Churchill, W. H., and Tapley, D. F. (1964). Nature (London) 202, 29. Cline, M. J., Selenkow, H. A., and Brooke, M. S. (1959). N. Engl. J . Med. 260, 117. Cole, R. K. (1966). Genetics 53, 1021. Cole, R. K., Kite, J. H., Jr., and Witebsky, E. (1968). Science 160, 1357. Cruchaud, A., and Juditz, E. (1968). Clin. E x p . Immunol. 3, 771. Daniel, P. M., Pratt, 0. E., Roitt, I. M., and Tonigiani, G. (1965). J. Physiol. (London) 183, 33. Daniel, P. M., Plaskett, L. G., and Pratt, 0. E. (1966). Biochem. J. 100, 622. Daniel, P. M., Pratt, 0. E., Roitt, I. M., and Torrigiani, G. (1967a). Quart. J. Exp. Physiol. 52, 184. Daniel, P. M., Pratt, 0. E., Roitt, I. M., and Torrigiani, G. (1967b). Immunology 12, 489. David, J. R., and Holborow, E. J. (1961). Lancet 1, 83.
THYROID ANTIGENS AND AUTOIMMUNITY
175
de Crombrugghe, B., and Edelhoch, H. ( 1966). Biochemistry 5, 2238. de Crombrugghe, B., Pitt-Rivers, R., and Edelhoch, H. (1966). 3. Biol. Chem. 241, 2766. DeGroot, L. J., and Carvalho, E. (1960). J. Clin. Endocrinol. Metab. 20, 21. DeGroot, L. J., Hall, R., McDermott, W. V., Jr., and Davis, A. M. (1962). N. Engl. 3. Med. 267, 267. De Nayer, P., and De Visscher, M. (1970). FEBS Lett. 8, 137. Derrien, Y., Michel, R., and Roche, J. (1948). Biochim. Biophys. Acta 2, 454. Derrien, Y., Michel, R., Pedersen, K. O., and Roche, J. (1949). Biochim. Biophys. Acta 3, 436. Dobyns, B. M., and Hirsch, E. Z. (1956). J. Clin. Endocrinol. Metab. 16, 153. Doebbler, T., and Rose, N. R. ( 1961). Fed. Proc., Fed. Amer. Soc. Erp. Biol. 20, 39. Doebbler, T. K., and Rose, N. R. (1966). Clin. Erp. Immunol. 1, 159. Doniach, D. (1967). 3. Clin. Pathol. 20, 385. Doniach, D., and Hudson, R. V. (1957). Brit. Med. J. 1, 672. Doniach, D., and Roitt, I. M. (1957). J. Clin. Endocrinol. Metab. 17, 1293. Doniach, D., and Roitt, I. M. (1962). Annu. Reu. Med. 13, 213. Doniach, D., and Roitt, I. M. (1968). Medicine Today 2, 29. Doniach, D., and Roitt, I. M. (1969). In “Textbook of Immunopathology” (P. A. Miescher and H. J. Miiller-Eberhard, eds.), Vol. 2, pp. 516-533. Grune & Stratton, New York. Doniach, D., Roitt, I. M., and Taylor, K. B. (1963). Brit. Med. 3. 1, 1374. Doniach, D., Roitt, I. M., and Taylor, K. B. (1965). Ann. N . Y. Acad. Sci. 124, 605. Doniach, D., Roitt, I. M., and Polani, P. E. (1968). Proc. Roy. Soc. Med. 61, 278. Dopheide, T. A. A., and Trikojus, V. M. (1964). Nature (London) 201, 1128. Dowling, T. J., Ingbar, S. H., and Freinkel, N. ( 1961). 3. Clin. Endocrinol. Metab. 21, 1390. Dumonde, D. C. (1966). Aduan. Zmmunol. 5, 245. Edelhoch, H. (1960). 3. Biol. Chem. 235, 1326. Edelhoch, H. (1965). Recent Progr. H o r n . Res. 21, 1. Edelhoch, H., and de Crombrugghe, B. (1966). 3. Biol. Chem. 241, 4357. Edelhoch, H., and Rall, J. E. (1964). In “The Thyroid Gland (R. Pitt-Rivers and W. R. Trotter, eds.), Vol. 1, pp. 113-120. Butterworth, London. Edelhoch, H., Carlomagno, M. S., and Salvatore, G. ( 1969). Arch. Biochem. Biophys. 134, 264. Evans, A. W. H., Woodrow, J. C., McDougall, C. D. M., Chew, A. R., and Evans, R. W. (1967). Lancet 1, 636. Evans, T. C., Beienvaltes, W. H., and Nishiyama, R. H. (1969). Endocrinology 84, 641. Eyquem, A. ( 1970). Bull. Inst. Pasteur, Paris 68, 527. Fagraeus, A., and Jonsson, J. (1970). Immunology 18, 413. Fahey, J. L., and Goodman, H. C. (1960). 3. Clin. Invest. 39, 1259. Falconer, I. R., Roitt, I. M., Seamark, R. F., and Torrigiani, G. (1970). Biochem. 3. 117, 417. Federlin, K., Oppermann, W., and Pfeiffer, E. F. (1965). Deut. Med. Wochenschr. 90, 247. Feinberg, J. G., Hill, C. W., Doniach, D., and Roitt, I. M. (1969). Int. Arch. AG lergy Appl. Immunol. 35, 335. Felix-Davies, D., and Waksman, B. H. ( 1961). Arthritis Rheum. 4, 416.
176
SIDNEY SHULMAN
Ferguson-Smith, M. A., Anderson, J. R., Frgiland, A., and Gray, K. G. (1966). Lancet 2, 566. Fialkow, P. J., Uchida, I. A., Hecht, F., and Motulsky, A. G. (1965). Lancet 2, 868. Flanagan, T. D., Barron, A. L., Beutner, E. H., and Witebsky, E. (1966). J. Immunol. 97, 539. Flax, M. H. (1963). Lab. Inuest. 12, 199. Flax, M. H., and Billote, J. B. (1965). Ann. N . Y. Acad. Sci. 124, 234. Flax, M. H., Jankovi6, B. D., and Sell, S. (1963). Lab. Inuest. 12, 119. Fleisher, M. S. (1922). J. Immunol. 7, 51. Fleisher, M. S., and Amstein, N. (1921). J. Immunol. 6, 223. Fleisher, M. S., Hall, T. G., and Amstein, N. (1920). J. Immunol. 5, 437. Forbes, I. J., Roitt, I. M., Doniach, D., and Solomon, I. L. (1962). J. Clin. Znuest. 41, 996. Friedman, R. M., Buckler, C. E., and Baron, S. (1961). J. Exp. Med. 114, 173. Furth, E. D., Agrawal, R. B., and Propp, R. P. (1970). J. Clin. Endocn’nol. Metab. 31, 60. Genghof, D. S., and Battisto, J. R. (1961). Proc. SOC. Exp. Biol. Med. 107, 933. Ghayasuddin, M., and Shulman, S. (1967). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 26, 340. Ghayasuddin, M., and Shulman, S. (1968a). Abstr. Int. Congr. Biochem. 7th, 1967 p. 1024. Ghayasuddin, M., and Shulman, S. (196813). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 27, 275. Ghayasuddin, M., and Shulman, S. (1970). Immunology 19, 225. Ghayasuddin, M., Zappi, E., and Shulman, S. (1969). Cryobiology 6, 262. Ghose, T., and Cerini, M. (1969). Clin. Exp. Immunol. 5, 515. Gitlin, D., and Biasuccci, A. (1969). J. Clin. Endocrinol. Metub. 29, 849. Glynn, L. E., and Holborow, E. J. (1965). “Autoimmunity and Disease.” Davis, Philadelphia, Pennsylvania. Godal, T. ( 1967). Acta Pathol. Microbiol. Scad. 69, 205. Godal, T., and Berdal, P. ( 1967). Acta Pathol. Microbiol. Scand. 69, 63. Godal, T., and Kiresen, R. (1967a). Acta Pathol. Microbiol. S c a d 69, 332. Godal, T., and Kiresen, R. (1967b). Acta Pathol. Microbiol. Scand. 69, 343. Goodman, H. C. (1965). Ann. N. Y. Acad. Sci. 124, 231. Goodman, H. C., Exum, E. D., and Robbins, J. (1964). J. Zmmunol. 92, 843. Goudie, R. B. (1960). Immunology 3, 284. Goudie, R. B., and McCallum, H. M. (1962). Lancet 1, 348. Goudie, R. B., and McCallum, H. M. (1963). h c e t 2, 1035. Goudie, R. B., Anderson, J. R., Gray, K. G., Clark, D. H., Murray, I. P. C., and McNicol, G. P. ( 1957). Lancet 2, 976. Goudie, R. B., Anderson, J. R., and Gray, K. G. (1959a). Immunology 2, 309. Goudie, R. B., Anderson, J. R., and Gray, K. G. (1959b). J. Pathol. Bacteriol. 77, 389. Gourlay, F. (1894). J . Physiol. (London) 16, 23. Gruenstein, E., and Wynn, J. A. (1970). 1. Thew. Biol. 26, 343. Hajdu, A., and Rona, G. (1969). Expen’entiu 25, 1325. Hall, R., Owen, S. G., and Smart, G. A. (1960). Lancet 2, 187. Hall,R., Owen, S. G., and Smart, G. A. (1964). Lancet 2, 115.
THYROID ANTIGENS AND AUTOIMMUNITY
177
Halmi, N. S. ( 1964). In ”The Thyroid Gland” (R. Pitt-Rivers and W. R. Trotter, eds.), pp. 71-86. Butterworth, London. Hargis, B. J., Malkiel, S., and Berkelhammer, J. (1968). J. Zmmunol. 101, 374. Harington, C. R. (1944). J. Chem. SOC., London p. 193. Hazard, J. B., and Smith, D. E., eds. (1964). “The Thyroid.” Williams & Wilkins, Baltimore, Maryland. Heidelberger, M., and Palmer, W. W. (1933). J. Biol. Chem. 101, 433. Heidelberger, M., and Pedersen, K. 0. (1935). J . Gen. Physiol. 19, 95. Heimann, P. (1966). Acta Enducsinol. (Copenhagen) 53, Suppl. 110. Hektoen, L., and Schulhof, K. (1923). J. Amer. Med. Ass. 80, 386. Hektoen, L., and Schulhof, K. (1925). Proc. Nut. Acad. Sci. U. S . 11, 481. Hektoen, L., Fox, H., and Schulhof, K. (1927). J. Infec. D b. 40, 641. Helyer, B. J., and Howie, J. B. (1963). Lancet 2, 1026. Hendler, R. W., Dalton, A. J., and Glenner, G. G. (1957). J. Biophys. Biochem. Cytol. 3, 325. Herscovics, A. (1969). Biochem. J . 112, 709. Herscovics, A. (1970). Biochem. J. 117, 411. Hijmans, W., Doniach, D., Roitt, I. M., and Holborow, E. J. (1961). Brit. Med. J. 2, 909. Hjort, T. (1963a). Acta Med. S a n d . 174, 137. Hjort, T. (1963b). Acta Med. Scund. 174, 147. Hjort, T. (1969). Clin. Erp. Immunol. 5, 43. Hjort, T.,and Pedersen, G. T. (1962). Lancet 2, 259. Hjort, T., Jeppesen, F., Okholm, K., and Temler, J. (1963). Dun. Med. Bull. 10, 159. Hofer, R., and Schatz, H. (1965). Acta Endocrinol. Suppl. 100, 193. Holborow, E. J., Brown, P. C., Roitt, I. M., and Doniach, D. (1959). Brit. 1. E i p . Pathol. 40, 583. Hoyer, J., and Condie, R. M. (1962). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 21, 277. Hutchison, R. (1896). J . Physiol. (London) 20, 474. Imas, B., Eyquem, A., and Hermet, J. (1969). Ann. Znst. Pasteur, Paris 117, 860. Ingbar, S. H., Askonas, B. A., and Work, T. S. (1959). Endocrinology 64, 110. Irvine, W. J. (1960a). Scot. Med. J . 5, 511. Irvine, W. J. (1960b). J. Endocrinol. 20, 83. Irvine, W. J. (1962). Brit. Med. J. 1, 1444. Irvine, W. J. (1964). Quart. J. Erp. Physiol. 49, 324. Irvine, W. J. (1966). Clin. Em. Immunol. 1, 341. Irvine, W. J., and Muir, A. R. (1963a). Quort. 1. Exp. Physiol. 48, 13. Irvine, W. J., and Muir, A. R. (1963b). Exp. Cell Res. 29, 73. Irvine, W. J., MacGregor, A. G., and Stuart, A. E. (1962). Lancet 2, 843. Irvine, W. J., Davies, S. H., and Sumerling, M. D. (1965). Curr. Top. Thyroid Res., Proc. Int. Thyroid Conf., 5th, 1965 pp. 773-780. Irvine, W. J., Stewart, A. G., and Scarth, L. (1967). Clin. Exp. Immunol. 2, 31. Irvine, W. J., Clarke, B. F., Scarth, L., Cullen, D. R., and Duncan, L. P. J. (1970). Lancet 2, 163. Jakoby, W. B., Labaw, L., Edelhoch, H., Pastan, I., and Rall, J. E. (1966). Science 153, 1671. Jankovib, B. D. (1962). Int. Arch. Allergy Appl. Immunol. 21, 207.
178
SIDNEY SHULMAN
Jankovib, B. D., and Flax, M. H. (1963). J. lmmunol. 90, 178. Jankovib, B. D., and Mitrovib, K. ( 1963). Nature (London) 200, 186. Jankovii., B. D., IHvaneski, M., Popeskovik, L., and Mitrovik, K. (1965). Int. Arch. Allergy Appl. Immunol. 26, 18. Jankovii., B. D., Mitrovii., K., Popeskovii., L., and Milogevii., D. (1969). Clin Exp. Immunol. 5, 29. Jonckheer, M. H. (1963). Ann. Endocrin. 24, 756. Jonckheer, hl. H., and Karcher, D. M. (1971). J, Clin. Endocrinol. Metab. 32, 7. Jonckheer, M. H., Haelst, L. V., and Otten, J. (1968). Proc. Int. Congr. Endocrinol., 3rd, 1968 Int. Congr. Ser. No. 157, Abstr. 356. Jones, H. E. H., and Roitt, I. M. (1961). Brit. J . E x p . Pathol. 42, 546. Jonsson, J., and Fagraeus, A. ( 1969). Immunology 17, 387. Kalden, J. R., Williamson, W. G., and Irvine, W. J. (1969a). Clin. Exp. Immunol. 5, 549. Kalden, J. R., James, K., Williamson, W. G., and Irvine, W. J. (1969b). Clin. Exp. lmmutiol. 5, 597. Ktresen, R., and Godal, T. (1969a). Immunology 17, 847. Ktresen, R., and Godal, T. (196913). Immunology 17, 863. Kite, J . H., Jr., Rose, N. R., Kano, K., and Witebsky, E. (1965). Ann. N . Y. Acad. Sci. 124, 626. Kite, J. H., Jr., Argue, II., and Rose, N. R. (1966). Clin. Exp. Immunol. 1, 139. Kite, J. H., Jr., Wick, G., Twarog, B., and Witebsky, E. (1969). J. lmmunol. 103, 1331. Kivikangas, V., Lamberg, B. A., and MaenpG, J. (1970). Scand. J. Clin. Lab. Inuest. 25, 263. Klinck, G. H. (1964). In “The Thyroid” (J. B. Hazard and D. E. Smith, eds.), pp. 1-31. Williams & Wilkins, Baltimore, Maryland. Klinck, G. H., Oertel, J. E., and Winship, T. (1970). Lab. Inuest. 22, 2. Koffler, D., and Friedman, A. H. (1964). Lab. Invest. 13, 239. Koffler, D., and Paronetto, F. (1965a). Nature (London) 207, 773. Koffler, D., and Paronetto, F. ( 1965b). J. Immuiiol. 94, 329. Korngold, L., van Leeuwen, G., and Brener, J. L. (1959). 1. Lab. Clin. Med. 53, 517. Kuhns, W. J. ( 1954). J . E x p . Med. 99, 577. Kunkel, H. G., and Tan, E. M. (1964). Aduan. lmmunol. 4, 351. Landsteiner, K., and van der Scheer, J. (1925). J . Exp. Med. 42, 123. Lerner, E. M., 11, McMaster, P. R. B., and Exum, E. D. (1962). Fed. Proc., Fed. Amer. SOC. E x p . Biol. 21, 43. Lerner, E. M., 11, McMaster, P. R. B., and Exum, E. D. (1964). J. Erp. Med. 119, 327. Levine, S., and Wenk, E. J. (1961). Amer. I. Pathol. 39, 419. Levine, S., and Wenk, E. J. (1964). Science 146, 1681. Levine, S., and Wenk, E. J. (1965). Amer. J . Pathol. 47, 61. Lisitzky, S., Codaccioni, J. L., Cartouzou, G., and Mante, S. (1964). J . Clin. Endocrinol. Metab. 24, 305. Lissitzky, S., Rolland, M., and Bergot, J. (1965). Biochim. Biophys. Acta 111, 543. Lissitzky, S., Codaccioni, J. L., Bismuth, J., and Depieds, R. ( 1967). J . Clin. Endocrinol. Metab. 27, 185. Lissitzky, S., Bismuth, J., Codaccioni, J. L., and Cartouzou, G. (1968a). J . Clin. Endocrinol. Metab. 28, 1797.
THYROID ANTIGENS AND AUTOIMMUNITY
179
Lissitzky, S., Rolland, M., Reynaud, J., Savary, J., and Lasry, S. (1968b). Eur. J . Biochem. 4, 464. Lobo, L. C . C., Fridnian, J., Rebello, M. A., Hargreaves, F., and Figueiredo, J. G. (1966). Metub., Clin. Exp. 15, 330. Lundgren, H. P., and Williams, J. W. (1939). J. Phys. Chem. 43, 989. Lupulescu, A., Andreani, D., Monaco, F., and Andreoli, M. (1968). J. Clin. Endocrinol. Metub. 28, 1257. McMaster, P. R. B. ( 1958). N. Y. State J . Med. 58, 2980. McMaster, P. R. B., and Kyriakos, M. (1970). J. lmmunol. 105, 1201. McMaster, P. R. B., and Lerner, E. M., 11. (1967). J . Immunol. 99, 208. McMaster, P. R. B., Lerner, E. M., 11, and Exum, E. D. (1961). 1. Exp. Med. 113, 611. McMaster, P. R. B., Lerner, E. M., 11, and Mueller, P. S. (1965). Science 147, 157. McMaster, P. R. B., Lerner, E. M., 11, Kyriakos, M., and Mueller, P. S. (1967). J. Immunol. 99, 201. McQuillan, M. T., and Trikojus, V. M. (1966). In “Clycoproteins” (A. Gottschalk, ed. ), pp. 516531. Elsevier, Amsterdam. Margherita, S. S., and Premachandra, B. N. (1969). J. lmmunol. 102, 1511. Masi, A. T., Hartniann, W. H., Hahn, B., Abbey, H., and Shulman, L. E. (1965). Lancet 1, 123. Mates, G. P., and Shulman, S. (1967a). Immunochemistry 4, 319. Mates, G. P., and Shulman, S. (1967b). Immunology 14, 89. . 4, 475. Mates, G. P., and Shulman, S. ( 1 9 6 7 ~ )lmmunochemisty Mates, G. P., and Shulman, S. ( 1969). Immunochemistry 6, 35. Means, J. H., DeGroot, L. J., and Stanbury, J. B., eds. (1963). “The Thyroid and its Diseases,” 3rd ed. McGraw-Hill, New York. Mellors, R. C., Brzosko, W. M., and Sonkin, L. S. (1962). Amer. J . Pathol. 41, 425. Metzgar, R. S., and Buckley, R. H. (1967). Int. Arch. Allergy Appl. Immunol. 31, 174. Metzgar, R. S., and Grace, J. T. (1961). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 20, 39. Metzger, H., Sharp, C. C., and Edelhoch, H. (1962). Biochemistry 1, 205. Miescher, P., Gorstein, F., Benacemaf, B., and Cell, P. G. ( 1961). Proc. SOC. Exp. Biol. Med. 107, 12. Milgrom, F., and Witebsky, E. (1962a). lmmunology 5, 46. Milgrom, F., and Witebsky, E. (1962b). J. Amer. Med. Ass. 181, 706. Mizejewski, G. J. ( 1971). J. Reticuloendothel. SOC. (in press). Moore, J. M., and Neilson, J. M. (19f33). Lancet 2, 645. Mortensen, J. D., Woolner, L. B., and Bennett, W. A. (1955). J . Clin. Endocrinol. Metab. 15, 1270. Morton, M. E., and Chaikoff, I. L. (1943). J. Biol. Chem. 147, 1. Mouriz, J., and Stanbury, J. B. (1967). Can J. Biochem. 46, 51. Munoz, J. ( 1967). J. Immunol. 99, 31. Musser, E. A., and Graham, W. R. (1968). Lab. Anim. Care 18, 58. Nairn, R. C., Chose, T., Proteous, I. B., and Urquhart, J. A. (1963). J . Clin. Pathol. 16, 281. Nakamura, R. M., and Weigle, W. 0. (1967a). lnt. Arch. AUergy Appl. lmmunol. 32, 506. Nakamura, R. M., and Weigle, W. 0. (196713). J . Immunol. 98, 653. Nakamura, R. M., and Weigle, W. 0. ( 1 9 6 7 ~ )J. . lmmunol. 99, 357.
180
SIDNEY SHULMAN
Nakamura, R. M., and Weigle, W. 0. (1968a).Proc. SOC. Exp. Biol. Med. 129, 412. Nakamura, R. M., and Weigle, W. 0. (1968b). J. Immunol. 101, 876. Nakamura, R. M.,and Weigle, W. 0. (1969).J. Exp. Med. 130, 263. Nakamura, R. M., and Weigle, W. 0. (1970).Clin. Exp. Immunol. 7, 541. Narasimha Murthy, P. V., Raghupathy, E., Abraham, S., and Chaikoff, I. L. (1964). Biochem. Biophys. Res. Commun. 14, 514. Narasimha Murthy, P. V., Raghupathy, E., and Chaikoff, I. L. (1965).‘Biochemistry 4, 611. Nltve, P., Eagleton, G. B., and Wollman, S. H. (1970).J . Clin. Endocrlnol. Metab. 31, 38. Nissley, P., Cittanova, N., and Edelhoch, H. (1969).Biochemistry 8, 443. Nunez, J., Mauchamp, J., Macchia, V., and Roche, J. (1965).Bbchim. Biophvs. Acta 107, 247. Nunez, J., Mauchamp, J., Pommier, J., Cirkovi6, T., and Roche, J. (1966).Biochem. Biophys. Res. Commun. 23, 761. O’Donnell, I. J., Baldwin, R. L., and Williams, J. W. (1958’).Biochim. Biophys. Acta 28, 295. Okada, S., Kurata, Y., Konishi, K., and Matsuda, T. (1970).Int. Arch. Allergy Appl. Immunol. 39, 6. Olin, P., Vecchio, G., Ekholm, R., and Almqvist, S. (1970).Endocrinology 86, 1041. Oswald, A. (1899).Hoppe-Seyler’s Z. Physiol. Chem. 27, 14. Otten, J., Jonckheer, M., and Dumont, J. E. (1971).J. Clin. Endocrinol. Metab. 32, 18. Ovary, Z., Randall, H., Witebsky, E., Rose, N. R., Shulman, S., and Metzgar, R. (1958).Proc. SOC. Exp. Bwl. Med. 99, 397. Owen, C. A., Jr. (1958).J. Clin. Endocrinol. Metab. 18, 1015. Owen, S. G., and Smart, G. A. (1958).Lancet 2, 1034. Packalkn, T., Palmstierna, H. A. K., and Wasserman, J. (1967).Immunology 13, 119. Paseyro, P., Negrin, E., Moratb-Manaro, J., and Grosso, 0. F. (1962). J. Clin. Endocrinol. Metab. 23, 387. Paterson, P. Y. (1959). I n “Cellular and Humoral Aspects of the Hypersensitive States” (H. S. Lawrence, ed.), pp. 469-503. Harper (Hoeber), New York. Paterson, P. Y. (1966).Aduan. Immunol. 5, 131. Paterson, P. Y., and Drobish, D. G. (1968).I. Zmmunol. 101, 1098. Perelmutter, L.,and Stephenson, N. R. (1964).Can. J . Biochem. 42, 1169. Perelmutter, L.,Devlin, W., and Stephenson, N. R. (1963).Can. J. Biochem. Physiol. 41, 2493. Pierce, J. G., Rawitch, A. B., Brown, D. M., and Stanley, P. G. (1965).Biochim. Biophys. Acta 111, 247. Pinchera, A., Liberti, P., Martino, E., Fenzi, G. F., Grasso, L., Rovis, L., Baschieri, L.,and Doria, G. (1969).J. Clin. Endocrinol. Metab. 29, 231. Pitt-Rivers, R. (1963). Biochem. J. 87, 340. Pitt-Rivers, Rfi and Cavalieri, R. R. ( 1964).I n “The Thyroid Gland” (R. Pitt-Rivers and W. R. Trotter, eds.), Vol. 1, pp. 87-112. Buttenvorth, London. Pitt-Rivers, R., and Schwartz, H. L. (1967).Biochem. J. 105, 28c. Pitt-Rivers, R., and Trotter, W. R., eds. (1964). ‘The Thyroid Gland,” Vol. 1. Buttenvorth, London. Porter, D. D., and Fennell, R. H., Jr. (1961).N . Engl. J. Med. 265, 830. Posner, I., and Ordonez, L. (1970).Proc. SOC. Erp. Biol. Med. 134, 591. Premachandra, B. N. ( 1970). Endocrinology 86, 703.
THYROID ANTIGENS AND AUTOIMMUNITY
181
Premachandra, B. N., Ray, A. K., Hirata, Y., and Blumenthal, H. T. (lW3a). Endocrinology 73, 135. Premachandra, B. N., Ray, A. K., and Blumenthal, H. T. (196313). Endocrinology 73, 145. Pressman, D., James, A. W., Yagi, Y., Hiramoto, R., Woernley, D., and Maxwell, W. T. (1957). Proc. SOC. Exp. Biol. Med. 96, 773. Pulvertaft, R. J. V., Doniach, D., Roitt, I. M., and Hudson, R. V. (1959a). Lancet 2, 214. Pulvertaft, R. J. V., Davies, J. R., Weiss, L., and Wilkinson, J. H. (195913). J. Pathol. Bacteriol. 77, 19. Pulvertaft, R. J. V., Doniach, D., and Roitt, I. M. ( 1961). Brit. J. Exp. Pathol. 42, 496. Ramagopal, E., Spiro, M. J., and Stanbury, J. B. (1965). J. Clin. Endoctinol. Metab. 25, 742. Rawitch, A. B., Liao, T. H., and Pierce, J. G. (1968). Biochim. Biophys. A& 160, 360. Rawstron, J. R., and Farthing, C. P. (1962). J. Clin. Pathol. 15, 153. Robbins, J. (1963). J. Biol. Chem. 238, 182. Robbins, J., and Rall, J. E. (1962). Physiol. Reo. 40, 415. Robbins, J., Wolff, J., and Rall, J. E. (1959a). Endocrinology 64, 12. Robbins, J., Wolff, J., and Rall, J. E. (1959b). Endocrinology 64, 37. Robbins, J., Salvatore, G., Vecchio, G., and Ui, N. (1966). Biochim. Biophys. Acta 127, 101. Roche, J., Nunez, J., and Gruson, M. (1960). C. R. SOC. Biol. 154, 2194. Roche, J., Salvatore, G., Sena, L., Aloj, S., and Covelli, I. (1968). C a p . Biochem. Physiol. 27, 67. Roitt, I. M., and Doniach, D. (1958). Lancet 2, 1027. Roitt, I. M., and Doniach, D. (1960). Brit. Med. Bull. 16, 152. Roitt, I. M., and Doniach, D. (1967a). Brit. Med. Bull. 23, 66. Roitt, I. M., and Doniach, D. (1967b). Clin. Exp. Immunol. 2, 727. Roitt, I. M., and Torrigiani, G. (1967a). J. Clin. Pathol. 20, 391. Roitt, I. M., and Torrigiani, G. (196713). Endocrinology 81, 421. Roitt, I. M., Doniach, D., Campbell, P. N., and Hudson, R. V. (1956). Lancet 2, 820. Roitt, I. M., Campbell, P. N., and Doniach, D. (1958). Biochem. J. 59, 248. Roitt, I. M., Doniach, D., Wilson, E. G., and Couchman, K. (1960). Bull. SOC. Chim. Biol. 42, 1165. Roitt, I. M., Jones, H. E. H., and Doniach, D. (1962). In “Mechanism of Cell and Tissue Damage Produced by Immune Reactions” (P. Grabar and P. Miescher, eds.), pp. 174-183. Schwabe, Basel. Roitt, I. M., Ling, N. R., Doniach, D., and Couchman, K. G. (1964). Immunology 7, 375. Roitt, I. M., Jones, H. E. H., and Mills, G. L. (1965). Immunology 9, 281. Roitt, I. M., Torrigiani, G., and Doniach, D. (1968). Immunology 15, 681. Rolland, M., and Lissitzky, S. (1970). Biochim. Biophys. Acta 214, 282. Rolland, M., Bismuth, J., Fondarai, J., and Lissitzky, S. (1966). Acta Endocrinol. (Copenhagen) 53, 286. Rose, N. R., and Stylos, W. A. (1969). Clin. Exp. Immunol. 5, 129. Rose, N. R., and Witebsky, E. (1955). J. Immunol. 75, 282. Rose, N. R., and Witebsky, E. (1956). J. Immunol. 76, 417.
182
SIDNEY SHULMAN
Rose, N. R., and Witebsky, E. (1959). J. Immunol. 83, 34. Rose, N. R., and Witebsky, E. (1968). In “Textbook of Immunopathology” (P. A. Miescher and H. J. Miiller-Eberland, eds.), Vol. 1, pp. 150-163. Grune & Stratton, New York. Rose, N. R., Metzgar, R. S., and Isaacs, E. (1960). J. Immunol. 84, 649. Rose, N. R., Shulman, S., and Witebsky, E. (1962a). J. Immunol. 88, 229. Rose, N. R., Kite, J. H., Jr., and Doebbler, T. K. (1962b). In “Mechanism of Cell and Tissue Damage Produced by Immune Reactions” (P. Grabar and P. Miescher, eds. ), pp, 161-173. Schwabe, Basel. Rose, N. R., Witebsky, E., and Beienvaltes, W. H. (1965a). In “Immunological Diseases” ( M . Samter and H. L. Alexander, eds.), pp. 821-842. Little, Brown, Boston, Massachusetts. Rose, N. R., Kite, J. H., Jr., Doebbler, T. K., Spier, R., Skelton, F. R., and Witebsky, E. (1965b). Ann. N . Y. Acad. Sci. 124, 201. Rose, N. R., Skelton, F. R., Kite, J. H., Jr,, and Witebsky, E. (1966). Clin. Exp. Immunol. 1, 171. Ruddle, N. H., and Waksman, B. H. (1968a). J. Exp. Med. 128, 1237. Ruddle, N. H., and Waksman, B. H. (1968b). J. Exp. Med. 128, 1255. Ruddle, N. H., and Waksman, B. H. ( 1 9 6 8 ~ )J.. Exp. Med. 118, 1267. Ruebner, B., van Leeuwen, G., and Korngold, L. (1960). Lancet 1, 1108. Salabe, G., Kondo, Y., DeNayer, P., Robbins, J., and Rall, J. E. (1969). Endocrinology 85, 752. Salvatore, G., Salvatore, M., Cahnmann, H. J., and Robbins, J. (1964). J. Biol. Chem. 239, 3267. Salvatore, G., Vecchio, G., Salvatore, M., Cahnmann, H. J., and Robbins, J. (1965). J. Biol. Chem. 240, 2935. Salvatore, G., Aloj, S., Salvatore, M., and Edelhoch, H. (1967). J. Biol. Chem. 142, 5002. Salvin, S. B., and Liauw, H. L. (1967). J. Immunol. 98, 432. Sclare, G., and Taylor, G . ( 1961). J. Pathol. Bacteriol. 82, 29. Scott, T. W., and Trikojus, V. M. (1970). PTOC.Aust. Biochem. SOC.3, 83. Shulman, S . (1956). Fed. Proc., Fed. Amer. SOC. E x p . Biol. 15, 613. Shulman, S . ( 1963). “The Modeling of Bioniolecular Structures.” Ealing Press, Cambridge, Massachusetts. Shulman, S . (1968). Gunma Symp. Endocrinol. 5, 117. Shulman S . (1969). In “International Convocation on Immunology Buffalo, N. Y., 1968” ( N . R. Rose and F. Milgrom, eds.), pp. 313-322. Karger, Basel. Shulman, S. (1971a). Clin. E x p . Immunol. 9, 267. Shulman, S. ( 1971b). “Tissue Specificity.” Springer-Verlag, Berlin and New York (in press). . Crit. Reu. Clin. Lab. Sci. 2, 393. Shulman, S . ( 1 9 7 1 ~ )CRC Shulman, S. (1971d). In preparation. Shulman, S., and Armenia, J. P. (1963). 1. Biol. Chem. 238, 2723. Shulman, S., and Ghayasuddin, M. (1971). PTOC. Int. Thyroid Conf., 6th, 1970 (in press ) . Shulman, S., and Stanley, P. G. (1961). Aust. J. Biol. Sci. 14, 475. Shulman, S., and Witebsky, E. (1960a). Ann. N . Y. Acad. Sci. 86, 400. Shulman, S., and Witebsky, E. (1960b). J. Immunol. 85, 559. Shulman, S., and Witebsky, E. (1962). J. Immunol. 88, 221. Shulman, S., Rose, N. R., and Witebsky, E. (1955). J. Immunol. 75, 291.
THYROID ANTIGENS AND AUTOIMMUNITY
183
Shulman, S., Rose, N. R., and Witebsky, E. (1957). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 16, 433. Shulman, S., Mates, G., and Bronson, P. (1967). Biochim. Biophys. Actu 147, 208. Shulman, S., Ghayasuddin, M., and Witebsky, E. (1968). J. Immunol. 101, 1059. Sisson, J. C. (1968). J. Clin. Endocrinol. Metab. 28, 433. Smeds, S. (1970). Biochem. Biophys. Res. Commun. 34 1168. Smith, B. R. (1969). Biochim. Biophys. Acta 188, 89. Smith, B. R., and Munro, D. S. (1970). Biochim. Biophys. Actu 208, 285. Smith, B. R., Komngton, K. J., and Munro, D. S. (1969). Biochim. Biophys. Acta 192, 277. Smith, D. J., and Shulman, S. (1971a). Fed. Proc., Fed. Amer. SOC. Exp. Bid. 30, 1320. Smith, D. J., and Shulman, S. (1971b). In preparation. Snell, F. M., Shulman, S., Spencer, R. P., and Moos, C. (1965). ‘‘Biophysical Principles of Structure and Function.” Addison-Wesley, Reading, Massachusetts. Solomon, D. H., and Beall, G. N. (1970). Endocrinology 86, 191. Soto, J. R., Imas, B., Brunengo, A. M., and Goldberg, D. (1967). J. Clin. Endocrind. Metab. 27, 1581. Spiegelberg, H. L., and Miescher, P. A. (1963). J. Exp. Med. 118, 869. Spiro, M. J. (1961). J. Biol. Chem. 236, 2901. Spiro, M. J. (1970). J. Biol. Cbem. 245, 5820. Spiro, R. G. (1963). N. Engl. J. Med. 269, 566. Spiro, R. G. (1965). J. Biol. Chem. 240, 1603. Spiro, R. G., and Spiro, M. J. (1963). Fed. PTOC.,Fed. Amer. SOC. Exp. Biol. 22, 538. Spiro, R. G., and Spiro, M. J. (1965). J. Bid. Chem. 240, 997. Stanbury, J. B., and Janssen, M. A. (1962). J . Clin. Endocrinol. Metab. 12, 978. Stanley, P. G. (1956). Biochem. J. 63, 581. Stanley, P. G. (1964a). Actu Endocrinol. (Copenhagen) 45, 161. Stanley, P. G. (196413). Acta Endochnol. (Copenhagen) 46, 675. Steiner, R. F., and Edelhoch, H. (1961). J. Amer. Chem. SOC. 83, 1435. Stokinger, H. E., and Heidelberger, M. (1937). J. Exp. Med. 66, 251. Stuart, A. E., and Allan, W. S. A. (1958). Lancet 2, 47. Stylos, W. A,, and Rose, N. R. (1969). Clin. Exp. Immunol. 5, 285. Tarutani, O., and Shulman, S. (1970). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 29, 645. Tarutani, O., and Shulman, S. (1971a). Biochim. Biophys. Acta 229, 642. Tarutani, O., and Shulman, S. (1971b). Biochim. Biophys. Acta 236, 384. . preparation. Tarutani, O., and Shulman, S. ( 1 9 7 1 ~ )In Tarutani, O., and Ui, N. ( 1968). Biochem. Biophys. Res. Commun. 33, 733. Tarutani, O., and Ui, N. (1969a). Biochim. Biophys. Acta 181, 116. Tarutani, O., and Ui, N. ( 1969b). Biochim. Biophys. Acta 181, 136. Tarutani, O., Smith, D., and Shulman, S. (1971a). J. Biol. Chem. (submitted for publication). Tarutani, O., Muramatsu, T., Smith, D. J., and Shulman, S. (1971b). Endocrhonology (submitted for publication), Tata, J. R., Rall, J. E., and Rawson, R. W. (1956). J. Clin. Endocrinol. Metab. 16, 1554. Terplan, K. L., Witebsky, E., Rose, N. R., Paine, J. R., and Egan, R. W. (1960). Amer. J. Pathol. 36, 213.
184
SIDNEY SHULMAN
Themann, H., Andrada, J. A,, Rose, N. R., Andrada, E. C., and Witebsky, E. (1968). Clin. Exp. Immunol. 3, 491. Thomson, J. A., and Bissett, J. M. ( 1969). Acta Endocrinol. (Copenhagen) 62, 577. Thomson, J. A., and McGirr, E. M. (1969). J . Clin. Endocrinol. Metab. 29, 1259. Toro-Goyco, E., and Matos, M. (1965). J. Clin. Endocrinol. Metab. 25, 916. Torrigiani, G., and Roitt, I. M. (1963). Immunology 6, 73. Torrigiani, G., Roitt, I. M., and Doniach, D. (1968). Clin. Erp. Immunol. 3, 621. Torrigiani, G., Doniach, D., and Roitt, I. M. (1969). J. Clin. Endocrinol. Metab. 29, 305. Trotter, W. R., Belyavin, G., and Waddams, A. (1957). Proc. Roy. SOC. Med. 50, 961. Tucker, W. E. ( 1962). Amer. J. Clin. Pathol. 38, 70. Twarog, F. J., and Rose, N. R. (1968). J. Immunol. 101, 242. Twarog, F. J., and Rose, N. R. (1969). Proc. SOC. Exp. Biol. Med. 130, 434. Twarog, F. J., and Rose, N. R. (1970). J. Immunol. 104, 1467. Ui, N., and Tarutani, 0. (1961). J. Biochem. (Tokyo) 50, 508. Ui, N., Tarutani, O., Kondo, Y., and Tamura, H. (1961). Nature (London) 191, 1199. Ujejski, L., and Glegg, R. E. (1955). Can. J. Biochem. Physiol. 33, 199. Valenta, L., Roques, M., Torresani, J., Rolland, M., and Lissitzky, S. (1968a). Biochim. Biophys. A d a 168, 507. Valenta, L., Lissitzky, S., and Aquaron, R. (1968b). J. Clin. Endocrinol. Metab. 28, 437. Valenta, L., Kyncl, F., Niederle, B., and Jirousek, L. (1968~).J. Clin. Endocrinol. Metab. 28, 442. Van Tienhoven, A., and Cole, R. K. (1962). Anat. Rec. 142, 111. Van Zyl, A., Theron, C. N., and Van der Walt, B. (1969). Biochim. Biophys. Acta 171, 337. Vecchio, G., Edelhoch, H., Robbins, J., and Weathers, B. (1966). Biochemistly 5, 2617. Voisin, G. A. (1970). Ann. Inst. Pastew, Paris 118, 403. Waksman, B. H. (1959). Int. Arch. AUergy Appl. Immunol. 14, 1. Waksman, B. H. (1962). Medicine (Baltimore) 41, 93. Wasserman, J., and Packalh, T. (1965). Immunology 9, 1. Weigle, W. 0. (1961). 1. Erp. Med. 114, 111. Weigle, W. 0. (1962). 1. Erp. Med. 116, 913. Weigle, W. 0, (1965a). J. Erp. Med. 121, 289. Weigle, W. 0. (1965b). J. Exp. Med. 122, 1049. Weigle, W. 0. (1967). Immunology 13, 241. Weigle, W. O., and High, G. J. (1967a). J. Immunol. 94 1105. Weigle, W. O., and High, G. J. (196713). J . Immunol. 99, 392. Weigle, W. O., and Nakamura, R. M. (1967). J. Immunol. 99, 223. Weigle, W. O., and Nakamura, R. M. (1969). Clin. Exp. Immunol. 4, 645. Weigle, W. O., High, G. J., and Nakamura, R. M. (1969). J. Exp. Med. 130, 243. Werner, S. C. (1969). J . Clin. Endocrinol. Metab. 29, 860. Werner, S. C., Bora, S., Koutras, D. A., and Wahlberg, P. (1970). Science 170, 1201. White, R. G. (1957). Proc. Roy. SOC.Med. 50, 953. White, R. G., Bass, B. H., and Williams, E. (1961). Lancet 1, 368.
THYROID ANTIGENS AND AUTOIMMUNITY
185
Wick, G., Kite, J. H., Jr., Cole, R. K., and Witebsky, E. (1970a). J . Zmmunol. 104, 45. Wick, G., Kite, J. H., Jr., Witebsky, E. (1970b). J. Zmmunol. 104, 54. Wick, G., Kite, J. H., Jr., and Witebsky, E. ( 1 9 7 0 ~ )J.. Zmmunol. 104, 344. Wick, G., Witebsky, E., Kite, J. H., Jr., and Beutner E. H. (1970d). Clin. Exp. Immunol. 7 , 173. Williams, E. D., Engel, E., and Forbes, A. P. (1964). N. Engl. J. Med. 270, 805. Wissig, S. L. (1964). In “The Thyroid Gland” (R. Pitt-Rivers and W. R. Trotter, eds.), Vol. 1, pp. 32-70. Butterworth, London. Witebsky, E. ( 1929). Naturwissenschaften 17, 771. Witebsky, E. (1964). Zn “The Thyroid” (J. B. Hazard and D. E. Smith, eds.), pp. 143-151. Williams & Wilkins, Baltimore, Maryland. Witebsky, E. (1968). Amer. J. Clin. Pathol. 49, 301. Witebsky, E., and Milgrom, F. (1962). Immunology 5, 67. Witebsky, E., and Rose, N. R. (1956). J . Immunol. 76, 408. Witebsky, E., and Rose, N. R. (1959). J . Immunol. 83, 41. Witebsky, E., and Rose, N. R. (1963). N. Y. State J . Med. 63, 56. Witebsky, E., Rose, N. R., and Shulman, S. (1955). J. Immunol. 75, 269. Witebsky, E., Rose, N. R., and Shulman, S. (1956). Cancer Res. 16, 831. Witebsky, E., Rose, N. R., Terplan, K., Paine, J. R., and Egan, R. W. (1957). J. Amer. Med. Ass. 164, 1439. Witebsky, E., Rose, N. R., and Shulman, S. (1958). Lancet 1, 808. Witebsky, E., Kite, J. H., Jr., Wick, G., and Cole, R. K. (1969). J . Immunol. 103, 708. Wolff, J., Robbins, J., and Rall, J. E. (1959). Endocrinology 64, 1. Wollman, S. H., and Warren, L. (1961). Biochim. Biophys. Acta 47, 251. Zavaleta, A., and Stastny, P. (1967). Clin. Exp. Immunol. 2, 543.
This Page Intentionally Left Blank
Immunological Aspects of Burkitt’s Lymphoma GEORGE KLElN Deportmenf o f Tumor Biology, Korolinsko Insfifufst, Sfockholm, Sweden
I. Introduction . . . . . . . . . . . 11. Humoral Antibody Studies . . . . . . . . A. Immunofluorescent Tests . . . . . . . . B. Cytotoxic and Growth Inhibition Tests . . . . . C. Complement Fixation Tests . . . . . . . D. Immunoprecipitation . . . . . . . . . 111. Studies on Cell-Mediated Immunity . . . . . . A. Delayed Hypersensitivity in Viuo . . . . . . B. Mixed Lymphocyte Stimulation Tests . . . . . IV. One or Several EB Viruses? . . . . . . . . V. Immunological Studies on Oncogenic Herpes Viruses in Animals A. Marek’s Disease . . . . . . . . . . B. Luck6 Agent . . . . . . . . . . . VI. Implications . . . . . . . . . . . References . . . . . . . . . . .
. . .
. . . . . . .
.
. . . .
187 188 188 210 212 217 221 221 223 223 225 225 231 232 243
I. Introduction
The immunological aspects of Burkitt’s lymphoma (BL) can be considered under three main headings: ( I ) immune responses of BL patients against antigens associated with their tumor; (2) immunoglobulin production of BL cells; ( 3 ) effect of BL on the immune status of the host. This review concentrates on topic I . Some aspects of topic 2 are mentioned, whenever pertinent, but not in detail; sparse information on topic 3 is included when appropriate. The great upsurge of interest in the immunological aspects of BL can be attributed to several factors including the following: a. The postulated viral etiology of the disease (Burkitt, 1963) has led to a search for virus-associated and/ or virus-induced tumor-associated antigens. b. Clinical observations, including two documented cases of spontaneous regression (Burkitt and Kyalwazi, 1967), and 1520%of recorded patients who were long time survivors after chemotherapy including those receiving insufficient therapy (Burkitt, 196713; Clifford, 1966; Ngu, 1965), have been interpreted as suggesting that the .final outcome of the disease depends on the combined action of drug treatment and the im187
188
GEORGE KLEIN
mune response of the host and may be sometimes decided by the immune response alone. c. Progress in experimental tumor immunology showed that all virusinduced lymphomas studied contained characteristic, virally induced antigens, common for all neoplasms induced by the same virus ( G . Klein, 1966b, 1969; Old and Boyse, 1964; Pasternak, 1969; Sjogren, 1965) and lent increasing credence to the concept of immune surveillance (Allison and Law, 1968; Law, 1966; Penn et al., 1969), i.e., the continuous elimination of neoplastic cell clones in statu nascendi, by a process involving immune recognition. Indirectly, these developments favored problems a and b as realistic targets for investigation. Studies performed so far on immune responses of BL patients against tumor-associated antigens can be divided into a search for humoral antibodies and for cell-mediated reactions. Humoral antibody work can be conveniently subdivided according to the method of investigation, such as immunofluorescence ( intracellular and membrane, respectively), cytotoxicity, complement fixation, and immunoprecipitation. Studies on cell-mediated responses are less extensive and include rather preliminary experiments on delayed hypersensitivity and leukocyte-target cell interactions. The review of the tumor-associated immune responses is followed by a consideration of some implications of the immunological studies, in relation to the etiology and clinical behavior of BL. II. Humoral Antibody Studies
A. IMMUNOFLUORESCENT TESTS 1. Intracellular Antigens G. Henle and Henle (1966a) tested five stationary suspension culture lines derived from BL by indirect immunofluorescence (IF) on acetonefixed smear preparations. All seventeen BL sera tested stained brilliantly in a small proportion of the cells. In addition, many sera from American controls, including healthy donors and patients suffering from various diseases, gave positive reactions as well. The incidence of positive controls increased from about 30%in children to approximately 90%in adults. Direct IF staining of an essentially similar type was obtained by using fluorescein-conjugated 7-globulin pools. The reaction could be blocked by prior exposure of the fixed cell smears to human sera that gave positive reactions in the indirect test, but not by negative sera. All five Burkitt lines contained reactive cells, but in different proportions. The EB3 line
IMMUNOLOGICAL ASPECTS OF BUFIKITT’S LYMPHOMA
189
reacted best (S-lOa positive cells), whereas EB1, EB2, SL1, and AL1 showed decreasing proportions of reactive cells, in this order. Electron microscopy showed that the lines carried the herpeslike virus described by Epstein et al. (1964), now generally known under the name of EBV ( Epstein-Barr virus), with the same decreasing order of virus-containing cells. A number of control lines gave consistently negative results. Sera containing antibodies against herpes simplex, varicella or herpes zoster, and cytomegalovirus (CMV) were not positive in higher frequencies than expected for the age group. The fluorescein-conjugated human 7-globulin that gave positive direct staining with the Burkitt lines, did not stain control lines infected with CMV, varicella virus, simian virus 40 (SV 40), or reovirus Types 1, 2 and 3. Fluoresceinconjugated rabbit antibodies to herpes simplex virus (HSV) or the reoviruses gave also negative results against the lymphoma cells. Confirmatory results on a larger collection of sera were reported the same year by W. Henle et al. (1966). All thirty Burkitt sera tested were highly positive. Among 140 sera from healthy African control children, 54% were positive, but the fluorescence observed was often of low intensity. Thirty-five percent of American children and 85% of American adults were positive, regardless of the histories of the donors. It was concluded that the IF technique detected cells that harbored EBV, for the following reasons: 1. Burkitt lines free of detectable virus particles contained no stainable cells. 2. The percentage of virus-carrying cells detected by electron microscopy corresponded to the percentage of fluorescent cells in the same line. 3. When the virus became undetectable by electron microscopy in originally positive lines following serial cultivation, the proportion of stainable cells decreased to 0.1%or less. 4. A deoxyribonucleic acid ( DNA ) inhibitor, 5-methylamino-2’-deoxyuridine, capable of inhibiting the multiplication of HSV, also reduced the number of fluorescent cells. 5. Both stained and virus-carrying cells exhibited the same signs of cell degeneration. Further experiments (W. Henle et al., 1966) showed that in starved cultures, the proportion of virus-containing and fluorescing cells increased in parallel. Virus particles could be separated from EB3 cells and concentrated for electron-microscopic agglutination tests. Test sera were absorbed with Raji cells, a Burkitt line free from demonstrable EBV; some sera were also absorbed with HSV and with sheep red blood cells. Sera that were negative in the IF test against EBV-carrying Burkitt cells, including a number of American leukemic sera, did
190
GEORGE KLEIN
not aggregate the virus particles or coat them with antibody, Sera containing high concentrations of antibody against HSV, varicella, or CMV were equally negative. Burkitt sera or pooled American y-globulin, positive against EB3 cells, coated the isolated virus particles and agglutinated at least some viral preparations. The Raji line, mentioned above, was the subject of a special study by Epstein et al. (1966). Although this line was established from a BL in Nigeria and was similar in all respects to other Burkitt lines, at the time of its study between the sixteenth and twenty-eighth months in culture, it was free of virus particles by electron microscopy, and the IF test was also negative. The authors pointed out that the absence of demonstrable virus in this line was no evidence against its possible presence at some earlier stage, since EBV was found to disappear from some Burkitt lines during prolonged cultivation (Rabson et al., 1966). As discussed below in more detail, EBV-associated antigens have, in fact, been found in the Raji line by complement fixation and, very recently, by immunoprecipitation as well (Pope et al., 1971). Recently, it has been shown (zur Hausen and Schulte-Holthausen, 1970) that Raji cells contain DNA sequcnces that specifically hybridize with purified EBV-DNA, indicating that the viral genome, or parts of it, are present in some form. Raji cells can be superinfected with live EBV (Gergely et al., 1971b; W. Henle et al., 1970b; Horosziewicz et al., 1970), but the infection is largely abortive, perhaps due to the presence of viral repressors in the Raji line. When sera of healthy African children were tested by IF against EB3 cells, the positive reactions obtained with about half of the sera were, as a rule, relatively weak. No significant differences were found between children from areas with low and high frequency of BL (Levy and Henle, 1966). The slightly higher frequency of positives compared with American children (54 against 35%) was interpreted to indicate a more frequent and earlier exposure of the African children to the virus. The high titers of reaction in all Burkitt patients tested, significantly different from controls, was taken to indicate that the EBV may contribute to the etiology of this tumor. Further efforts to identify the EBV with other known members of the herpes group gave consistently negative results (G. Henle and Henle, 1966b), and it was concluded that the agent is a previously unknown virus of the herpes group. A direct study of the relationship between the IF reaction and the EBV particle was made by comparing the same cells by IF, radioautography and electron microscopy (zur Hausen et al., 1967). Individual fluorescent positive and negative cells from a Burkitt line (Jijoye) have
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
191
been sectioned and examined by electron microscopy. Satisfactory preparations were obtained from five fluorescent cells; they all contained numerous virus particles and showed various signs of degeneration. Six nonfluorescent cells were entirely devoid of EBV particles. An extension of the DNA label from the nucleus into the cytoplasm was the rule in fluorescent cells, whereas in nonstaining cells grains were restricted to the nucleus. Fluorescent cells also showed a much higher overall labeling than nonfluorescent cells, particularly when the comparison was made several days after X-irradiation. These findings are also in line with the expectation that EBV is a DNA virus, like other members of the herpes group. Using the P3HR-1 subline of the Jijoye (P3) Burkitt line, originally isolated by Pulvertaft (1964), Hinuma and Grace (1967) found rather extreme variations (from 1 to 75%)in the proportion of fluorescent positive cells at different times during serial propagation. There was good correlation between the number of fluorescent cells and the number of particle-containing cells at different times throughout this vast range of variation. As in the previous study of G. Henle and Henle (1966a), absorption of reactive sera with HSV did not influence their activity against P3HR-1 cells. A virus concentrate prepared from PSHR-1 cells removed all activity. Hinuma et al. (1967) also demonstrated a significant increase in the percentage of IF positive cells during starvation. This was paralleled by an increase in cell-associated virus particles and a decrease in the proportion of viable cells. Only about 10% of the released herpeslike virus particles were enveloped. The coating of EBV particles by immunoglobulins from fluorescent positive sera, described by W. Henle et al. (1966), was confirmed and extended by Mayyasi et al. (1967). All 19 African, Burkitt sera tested coated the virus; the activity could be absorbed by a pellet of purified virus. A similar coating reaction was obtained with 15 of 20 sera from children with acute lymphocytic leukemia, 4 of 5 children with Hodgkin’s disease, 6 of 6 children with different solid tumors, and 5 of 19 normal children. Adult sera were positive in high frequencies. Interestingly, the positive sera coated only the naked but not the enveloped particles. This might be interpreted to mean that the Henle I F reaction identified antigens expressed on the viral capsid but not on the viral envelope. Recently, this explanation was put forward by W. Henle et al. (1970b), on the basis of this evidence as well as the relatively late appearance of the antigen during the viral cycle. They have proposed the designation “viral capsid antigen” (VCA) for the antigen( s ) detected by their original IF test.
192
GEORGE KLEIN
In the study of Mayyasi et al. (1967), rabbit antisera produced against herpestype viruses obtained from cell lines derived from African lymphoma patients reacted against similar viruses derived from American or African lymphoma lines but not against HSV. This was interpreted to mean that the different lymphoma-derived agents were either similar or shared a common antigenic component. In another study with heterologous (nonhuman) serum, Epstein and Achong (1967) immunized rabbits with purified herpeslike virus derived from the EB3 strain. The resulting fluorescein-conjugated antiserum brilliantly stained a fraction of EB3 cells, but did not stain normal human lymphoid cells. Positive reactions were obtained with Burkitt lines derived from Uganda, Nigeria, and New Guinea, provided the cells carried EBV demonstrable by electron microscopy. There was good correspondence between the number of positive cells as judged by IF and electron microscopy, alkhough the incidence of positive cells in the different lines varied by a factor of 3. A virus-carrying strain of blast cells from an American case of chronic myeloid leukemia was also positive. Control cells of various kinds, including cells infected with HSV and varicella, gave negative results. In two later papers, Epstein and Achong (1968) reported fuither details of the reactions obtained with the sera of rabbits immunized with purified EBV. There was no detectable difference in the immunological specificity of the reactions obtained with the EB1, EB2, SLl, EB3, and GOR strains of Burkitt cells or a blast cell strain derived from an American patient with chronic myeloid leukemia (Jenkins/5630); they all appeared to carry an immunologically identical or at least closely related virus. The authors suggested that even the unidentified unusual herpestype viruses found in many other strains of leukemic cells of American patients and in seemingly normal hemopoietic cells, might represent the same agent. In a recent paper, Hampar et al. (1970) reported their findings with antisera prepared in rabbits by intravenous immunization with concentrated EBV, obtained from the supernatants of BL-derived EBV carrier lines. The rabbit antisera, as well as a human EBV-positive serum, were labeled with ferritin or fluorescein and tested against EBV carrier cells, They all reacted with EBV capsid antigens. The scarcity of enveloped particles did not allow any conclusion about possible reactivity with viral envelopes. There was no cross-reactivity with HSV capsid surface antigens. Starting from the observation that the proportion of EBV-producing cells tended to decline gradually when lines were maintained to permit vigorous growth (Rabson et al., 1966), W. Henle and Henle (1968a)
IMMUNOLOGICAL ASPECTS
OF BURKITT’S LYMPHOMA
193
looked for culture conditions that could reduce cellular replication to favor the virus. After preincubation at 37°C for 2 or more days, Eagle’s medium with 25%fetal calf serum was no longer able to support adequate growth of EB3 or other Burkitt cell lines. There was a rapid increase in the proportion of EBV-producing cells under these conditions, however, as indicated by IF and electron microscopy. Virus particles could be extracted from such cell populations in quantities sufficient for antibody coating and agglutination tests. A further analysis of the limiting factor indicated that arginine deficiency was mainly responsible for the rise in EBV-positive cells. Media containing no arginine, or only 14% of the usual amount, prevented cellular growth and increased the concentration of IF cells to a similar extent as the preincubated complete medium. Deprivation of the cells for other essential amino acids did not stimulate the synthesis of EBV antigens. It was assumed that arginine deficiency promoted EBV production indirectly, possibly by a reduction in intracellular inhibitors. Maximal (about tenfold) increase in IF cells occurred within 4 to 6 days after transfer of the cultures to the deficient medium. Subsequently the infected cells degenerated and partially disappeared. Recently, Weinberg and Becker (197Oa) analyzed the effect of arginine deprivation. They were particularly puzzled by the paradoxic situation in which herpes viruses required arginine but EBV production was, nevertheless, induced in arginine-deprived Burkitt cells. They found that arginine deprivation caused extensive cell death-about 80% of the cells died within 72 hours. Synthesis of EBV antigens occurred between 72 and 120 hours. They inferred that arginine became available for the synthesis of EBV structural proteins from the amino acid pool, due to the degradation of cellular proteins and that arginine deprivation must have affected regulatory processes in the Burkitt cells, leading to a stimulation of EBV replication. The response of EBV carrier cells to arginine deprivation differed from their response to removal of other essential amino acids. Arginine deficiency eliminated ribonucleic acid (RNA) and protein synthesis that normally follows medium change, but did not affect DNA synthesis. Removal of other amino acids either inhibited DNA, RNA, and protein synthesis as well, or decreased the rate of all three processes, without markedly affecting the periods of the cell cycle. Weinberg and Becker proposed that the unique ability of the EBV carrier cells to synthesize DNA in the absence of concomitant protein and RNA synthesis, in arginine-deficient medium, probably provided the conditions for the induction of EBV synthesis. They proposed, furthermore, that continued DNA synthesis in arginine-deprived cells led to a release of EBV-DNA from cellular control mechanisms,
194
GEORGE KLEIN
promoting independent replication of the viral genome, followed by the transcription and, finally translation to viral structural proteins. In a preliminary study of the DNA synthesized in arginine-deprived cells, they also found that in CsCl density gradients, a viral DNA band could be demonstrated after, but not before, arginine deprivation. An EBV-DNA band could also be demonstrated after exposing carrier cultures to very small doses ( 5 ppg./nil.) of niitomycin C (Weinberg and Becker, 1970b). a. Experimental Transmission of EBV. W. Henle et al. were the first to report experimental transmission of EBV ( 1967). Simultaneously, they noticed an important change in the growth behavior of the infected cells. X-Irradiated (3000-6000 r ) cells of EBV-carrying, young, male, BL-derived Jijoye line served as virus donors. They were admixed with fresh leukocyte suspensions from female infants. The mixtures were seeded on monolayers of human diploid female fibroblasts, unable to grow in suspension. In all seven experiments, the normal leukocytes started to grow in the suspension within 2 to 4 weeks after this mixed cultivation. In a similar experiment EBV-negative Raji cells did not induce growth. Separate cultures of the leukocytes or of the X-irradiated Burkitt cells failed to survive under the same conditions. On chromosomal analysis, the established culture lines all lacked the Y chromosome. They all showed positivc IF, and herpes-type particles could be demonstrated in the one line examined by electron microscopy. Fresh leukocytes before cultivation or several days after incubation were negative by IF. Several of the converted leukocytes were contributed by donors with EBV-negative sera. Closely similar transmission experiments, but with EBV-containing filtrates rather than X-irradiated cells as donors, with embryonic cells as recipients, and resulting EBV-carrying blastoid cell lines, have been reported by Pope et al. (1968). Transformation and proliferation of leukocytes were regularly obtained 2435 days after EBV inoculation and resulted in the establishment of blastoid cell lines. Sex chromosome markers confirmed that the established lines were of fetal origin. Transformation was demonstrated with two separate culture filtrates and with white cells, bone marrow, thymus, and spleen from 5 individual fetuses. Control filtrates, including culture medium of EBV antigennegative lines, were negative. The sensitivity and sedimentation characteristics of the transforming factor were compatible with EBV properties. It is particularly important that fetal tissues never showed spontaneous transformation, indicating that EBV was not transmitted vertically. Similar findings were obtained by Nilsson et a,?. (1971). Dunkel and Ziegel (1970) and Horosziewicz et al. (1970) have also reported the successful infection of blastoid cells of EBV antigen-free
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
195
lines with EBV concentrates, harvested from the culture supernatants of EBV-carrying, BL-derived lines. W. Henle and Henle (1970) and Gerber et al. (1969) have established permanent lines of lymphoblastoid cells, by adding EBV to buffy coat cells of healthy EBV-seronegative adults. Concentrated culturc fluids from the Raji line were ineflective. Recently, Pope et al. reported more extensive data on the “transformation” of fetal lymphocytes after EBV exposure (1971). Whereas no cell lines could be established from uninfected fetal lymphocytes, the cells of 16 different fetuses could all be transformed by EBV. With serial dilutions of EBV, the latent period before the onset of lymphocyte proliferation ranged from 22 to 88 days. If the fetal leukocytes were cultured up to 15 days before EBV exposure, transformation occurred after a similar incubation period following virus inoculation as in the cases with no preculturing, indicating that the latency before cell proliferation was virus-dependent and did not merely reflect the adaptation of the cells to culture. Under conditions that were suboptimal for cell line establishment, white cells from adults with low titers of anti-EBV antibodies failed to yield cell lines, unless inoculated with EBV. Durr et al. (1970) have compared the susceptibility of four different human lymphoblastoid cell lines, devoid of EBV particles and of IF antigens, to EBV infection in vitro. Cell responses ranged from an acute cytopathic effect ( C P E ) that became evident within 16 to 72 hours and was characterized by cell enlargement, polykaryocyte formation, and progressive degeneration, to a less severe CPE, with minimal cell death and degeneration. This was followed either by the establishment of a carrier culture or, morc frequently, the disappearance of the virus. Nonenveloped herpes-type particles were seen in the cell nuclei already 16 hours after infection and were followed by the appearance of enveloped virions in the cytoplasm. Using IF, nonproductive or abortive infection was indicated by the higher incidence of cells showing viral antigen than the presence of virus particles. Interestingly, two lines established from the peripheral blood of human donors without evidence of malignant disease showed the highest sensitivity, whereas the Burkitt-derived Raji line and even more, the acute leukemia-derived 6410 line showed a rcduced cell response to virus infection. Rather dramatic changes seen in the lymphoblastoid cells after exposure to EBV were frequently not reflected in an increased titer of infectious virus. In cells where the intranuclear replication of EBV was readily demonstrable, the ability to transmit the infection with supernatants or with extracted virus was only moderately successful, which might be related to the presence of a high proportion of defective virus progeny. This defectiveness, also evident by electron microscopy, was
196
GEORGE KLEIN
manifested by the presence of incomplete ( nonnucleated or nonenveloped ) virus particles. In preliminary experiments with sucrose gradient fractionated EBV, infectivity correlated with relative frequency of enveloped virus particles in the various fractions. The “normal” cell-derived lymphoblastoid cell lines, alqough more sensitive to primary infection with EBV than the two lines derived from the neoplastic conditions, were less ready to support a persistent infection with EBV. Serial transmission of EBV recovered from chronically infected 6410 cells was readily demonstrable, whereas the transmission of virus has been more difficult from NC-37, one of the normal cell-derived lines. b. Early Antigen. Study of EBV-infected blastoid cells has led to the discovery of another EBV-associated intracellular antigen system, detectable by IF and designated as “early antigen” (EA) (W. Henle et al., 1970b). The original anti-EBV test has been renamcd “anti-VCA test.” Not all anti-VCA-positive sera have antibodies to EAs. Sera with equally high anti-VCA titers may be anti-EA negative or positive at various levels. All anti-EA-positive sera contain anti-VCA as well, in equal, or, more frequently, higher titers ( W. Henle et al., 1970b). Anti-EA titers show important disease-related features, as discussed below. Heated and UV-irradiated virus, or virus mixed with anti-EBV-positive sera, failed to elicit the formation of EA. No neutralization was obtained with anti-EBV-negative sera. These findings were utilized to develop a virus neutralization assay (Pearson et al., 1970). A very similar and probably identical antigen system was recently described by Hinuma et at. (1971). They demonstrated a “new ( N ) antigen” in EBV-infected blastoid cells of the NC37 line. Their description of the antigen and the distribution of antibodies against it parallels the EA system of W. Henle et al. (1970b) so closely that the two systems are no doubt identical. Hinuma et al. suggested that the N antigen may be analogous to the T antigens of cells transformed by oncogenic DNA viruses. Recently, EA was studied in relation to host cell macromolecular synthesis by combined IF and radioautography, in EBV-infected cells, and in EBV carrier cultures as well. The relationship between EA and VCA was visualized by two-color IF tests (Gergely et at., 1971a,b,c). Another aspect of this work, the relationship between EA and membrane antigens, is described in the following section. The appearance of EA in EBV infected blastoid cell cultures was not prevented by DNA inhibitors, such as cytosine arabinoside (Ara C ) and iododeoxyuridine ( IUDR) , in doses that completely stopped the incorporation of tritiated thymidine ( Gergely et al., 1971~).Puromycin
IMMUNOLOGICAL A S P E ~ SOF BURKITT'S LYMPHOMA
197
prevented the appearance of EA. In EA-positive cells, DNA, RNA, and protein synthesis were inhibited, in comparison with EA-negative cells (Gergely et al., 1971~). In carrier cultures (Gergely et al., 1971a) both Ara C and IUDR led to an accumulation of EA-positive cells and inhibited or blocked the production of VCA. Reversion of Ara C-induced DNA inhibition by deoxycytidine led to the prompt appearance of VCApositive cells that reached a higher level than in the untreated control samples. These findings were interpreted to mean that EA is really an early antigen in the true sense, i.e., independent from viral DNA synthesis, whereas VCA probably represents a late viral product. Two-color I F tests showed that VCA-positive cells also contained EA, with a very small number of exceptional cells that may have been artifacts. In addition to the EA-positive-VCA-positive cells, a variable number of EA-positive-VCA-negative cells were present in different carrier lines. The frequency of the latter ranged from zero in some lines to a tenfold excess, in relation to the VCA-positive cells, in others. c. Recent Serological Surveys of EBV-Associated Antibodies. Understanding of EBV-associated serological reactions advanced considerably with the important discovery that EBV was related to at least one form of infectious mononucleosis (G. Henle et al., 1968; Niederman et al., 1968). The clue was obtained when a laboratory technician, who served as the donor of an EBV-negative control serum, developed infectious mononucleosis and subsequently became highly positive in the EBV test. Preillness sera were available from 24 patients, sampled 1 month to 5 years prior to the onset of infectious mononucleosis; they were all negative for EBV antibodies. In the course of the disease, anti-EBV antibodies'appeared in all of them and rose to titers between 40 and 640 within a few weeks. There was no correlation between EBV and heterophile antibody titers. The highest heterophile titers were noted during the first 4 or 5 weeks after the onset of the disease and, as a rule, they declined to low levels or disappeared shortly thereafter. Complete absorption of the heterophile antibodies with sheep erythrocytes did not diminish the EBV antibody level. The EBV titers were frequently high in the first serum available for testing after the onset of the disease; they often continued to increase to fourfold or higher during the course of the illness, reaching peak levels during the first 3 4 weeks. High levels werc maintained for several months after the disease and were still demonstrable at lower levels in sera tested 1 or several years later, long after heterophile antibodies disappeared. Acute phase and convalescent sera obtained from donors with other febrile illnesses showed no diseasc-related change in EBV antibodies. Niederman et al. (1968) pointed out that the EBV antibody patterns fit well with the
198
GEORGE KLEIN
suggestion ( Benyesh-Melnick et al., 1968) that infectious mononucleosis is a frequent and often unrecognized infection of childhood; the full clinical and serological features appear particularly when young adults are infected. In infectious lymphocytosis there was no rise of anti-EBV antibodies, in spite of a striking proliferation of lymphocytes (Blacklow and Kapikian, 1970). This indicates, on the one hand, that EBV is not involved in the etiology of infectious lymphocytosis and, on the other, that lymphocyte proliferation does not lead to an increase of anti-EBV antibodies per se. Beltran et al. (1971) titrated the anti-EBV level in acute and convalescent phase sera of patients with seven different viral illnesses, known to increase the frequency of lymphocytes in the peripheral blood. No significant differences were found between the antiEBV sera titers at either phase in a large number of cases. Anti-EBV titers did not rise in infectious mononucleosis caused by CMV, indicating that the lymphoid proliferation of mononucleosis alone is insufficient to induce elevated anti-EBV titers (Klemola d al., 1970). The postulated growth-stimulating effect of EBV on cells of the lymphopoietic system was strengthened by the finding (G. Henle et al., 1968; Diehl d al., 1968) that suspension cultures of blastoid cells grew more readily from peripheral white cells of patients with infectious mononucleosis than from other donors. This has also been reported by Benyesh-Melnick et al. (1968). By IF, EBV could be readily demonstrated in lines derived from monoucleosis patients, although several weeks of cultivation were required for its expression, as a rule. It was suggested that antibodies present in the original cell suspension first had to be diluted sufficiently implying that antibodies suppress the production of viral antigens. The agent and the antibodies directed against it appeared indistinguishable from the virus found in cell lines derived from African, European, American, or New Guinea cases of BL. Identification of the virus as a causative agent of infectious mononucleosis was not interpreted to mean that it could not be the etiological agent of BL as well, either alone or together with other factors. It was pointed out that all Burkitt sera tested (more than 60 until this time) showed high anti-EBV levels. Only about 50% of 188 control African sera were positive and, as a rule, at reduced titers. The most conclusive evidence concerning the relationship between EBV and infectious mononucleosis came from a prospective seroepidemiological study (Evans et al., 1968). Progress of 268 college students whose serum lacked demonstrable EBV antibody ( <1:10) was followed for 2 to 4 years. Infectious mononucleosis developed in 15% and was accompanied by seroconversion to anti-EBV positivity. Among 94 EBV-positive college students, none developed clinical disease. Even
IMMUNOLOGICAL ASPECI’S OF BURKITT’S LYMPHOMA
199
though it may be argued that not all “EBV negatives” are really negative, seronegativity must have a biological meaning, in at least a fraction of the cases, since these individuals lack protection against mononucleosis that is apparent in EBV-positive subjects. Henle et al. summarized the results of their anti-EBV tests on 139 BL patients and 489 African controls of various types ( G . Henle et al., 1969). The geometric mean titer (GMT) of anti-EBV in BL patients ( 1:326) was approximately 8 times higher than of the various control groups ( 1:37). Eighty-seven percent had high ( 2 1 :160) titers. There were no significant differences between control sera collected from areas with a high and a low incidence of BL, respectively. With the exception of a few, moribund cases, low ( <1: 80) anti-EBV titers were very rare among BL patients, and there were no histologically confirmed cases with negative ( <1 :10) titers. Occasionally, long-term regression cases tended to show falling titers after some years, but this was by no means the rule. Among the controls, 18%were negative (
200
GEORGE KLEIN
DNA synthesis) they are similar to EA, but in the latter respect (compatibility with cell multiplication) they are different. It is not known whether T antigens are released from the growing tumor cells by some kind of secretory process or whether they are liberated only from dead and dying cells. Release of EA may be related to an abortive virus cycle, perhaps induced in tumor cells on their way to necrosis, under circumstances where the cycle cannot proceed to completion. The frequent absence of anti-T and anti-EA antibodies after tumor removal is in sharp contrast to the membrane-reactive antibodies, which tend to remain at a high level, or even increase following rejection in experimental systems and, so far as this has been studied, probably in the BL system as well. This will be discussed in the following section. Further clarification of the relationship between the dynamics of antibody formation against different EBV-determined antigens and the clinical course of the “high EBV associated diseases,” such as BL and NPC, may be helpful in elucidating important virus-tumor-host relationships, particularly if compared to EBV positive sera from patients with other tumors that are not characterized by a regularly high EBV association. Two other diseases have received particular consideration in connection with EBV associated serological reactions. One of them, Hodgkin’s disease, shows an interesting relationship between anti-EBV titers and histopathology (Johansson et aZ., 1970; Levine et al., 1971). High-titered cases are particularly associated with the lymphocyte-poor, sarcomatous-type disease, whereas the lymphocyte-rich, paragranulomatous-type disease has low antibody levels, comparable to controls. This is discussed in more detail in the following section. An EBV association has also been claimed in relation to sarcoidosis, but the evidence is more controversial. According to one report (Hirshaut et al., 1970), all of 131 sarcoidosis patients had antibodies to EBV and 79%showed titers of 1:640 or higher. In another series (Wahren et al., 1971), 2 were negative (compared to 14%in the control group). There was a higher percentage of high anti-VCA titers than in the controls (35 versus 7%),but the geometric mean was only l:SO, as compared to 1:22 in the controls. Only 7 of 21 patients with antiVCA 2 1:160 had also anti-EA titers of 1:5-20. The patients also showed a higher incidence of positive reactions to HSV and CMV and a slight shift to higher titers.
2. Membrane Immunofluorescence Tumor associated transplantation antigens, localized on the cell surface ( G . Klein, 1967), have been demonstrated in most experimental
IMMUNOLOGICAL ASPECTS
OF BURKITT’S LYMPHOMA
201
tumors (G. Klein, 1966b, 1968; Old and Boyse, 1964; Pasternak, 1969; Sjogren, 1965). Distinctive surface antigens of certain types can be visualized by the membrane fluorescence technique, using living target cells (Irlin, 1967; E. Klein and Klein, 1964; Lherisson et al., 1967; Morton et d.,1968; Moller, 1961; Tevethia et al., 1968). Virus-induced lymphomas in mice are particularly suitable for the demonstration of virally determined surface antigens by these techniques (E. Klein and Klein, 1964; Pasternak, 1965). This has led to a search for analogous. membrane-expressed antigens on BL cells in biopsy and tissue culture lines. Living BL biopsy cells were exposed to the sera of Burkitt patients and of various other donors. Subsequently, the cells were washed and stained with fluorescein-conjugated, antihuman globulin reagents. Certain Burkitt sera gave a high incidence of positive reactions; others were less frequently positive and still others were negative (G. Klein et al., 1966, 1967b). The positive Burkitt sera showed no reactivity against various control cells, including normal lymph node cells or leukemic cells of Swedish patients. Normal bone marrow cells of the same BL subjects who donated positively reacting biopsies, were negative as well. There was a certain correlation between the durability of regression after chemotherapy and the frequency of positive IF tests against BL biopsy cells. In 3 cases where the responsiveness of a patient to chemotherapy changed in the course of the disease, there was a parallel change in serum reactivity. It did not appear likely that isoantibodies could explain the positive reactions, since they were seen in 5 of 6 cases where autochthonous and allogeneic BL cells could be compared. The finding that positive sera tended to react with almost all BL biopsy cells tested but not with autochthonous or allogeneic non-BL cells also spoke against isoantibodies. Another indication was the lack of any relationship between the reactivity of erythrocytes and BL cells from the same donor, when compared against the same battery of sera. A small number of positive reactions were also obtained with other sera including hospitalized African patients with nonmalignant diseases, hospitalized African patients with malignant diseases other than BL, and healthy blood relatives of the Burkitt patients. A small number of healthy Swedish blood donors yielded negative sera. Whereas it thus appeared unlikely that the reactivity detected in the majority of the Burkitt sera was due to isoantibodies, it was also quite obvious that this conclusion could not be generalized, and isoantibodies or other serum components capable of adhering to BL cell surfaces and reacting with the anti-immunoglobulin reagents used for the indirect test would be capable of giving false positive reactions. Isoantibodies could be expected to occur in patients who received blood
202
GEORGE MEIN
transfusions or in multiparous women. It was necessary to modify the approach to eliminate this potential source of error. Another difficulty, interfering with further studies on the specificity of the reaction, was the great variability of the biopsy preparations, particularly the inconsistent degree of immunoglobulin coating on the surface of the biopsy cells. This coating was detected by direct membrane IF with conjugated anti-immunoglobulin reagents. It could be of two basically different kinds-IgM and/or IgG-showing not only a difference in class specificity but also in behavior in relation to the course of the disease (E. Klein et al., 1968; G. Klein et al., 1969a). In the cases where the cell surface reacted with anti-IgM conjugates, reactivity was usually expressed on 100%or nearly 100%of the cells. When such cells were converted into established lines in uitro, their “IgM ring” was maintained during long-term propagation. The membrane-IgM reactive lines did not secrete IgM into the medium (van Furth et al., 1971). Preliminary characterization of the reactive substance in membrane fractionation experiments (E. Klein et al., 1970) indicates, at least so far as one cell of this type is concerned, that 7 S, IgM subunits with p- and K-chain specificities are integrated into the cell membrane. Conceivably, this is the neoplastic variety of a normal lymphoid cell that incorporates molecules of this type into its plasma membrane as part of its normal differentiation. Lymphoid cells of this type have been postulated to play an important role in immunological memory and/or delayed hypersensitivity (Singhal and Wigzell, 1971). The lymphoma cells may represent the neoplastic variant in the same way as myeloma cells project normal immunoglobulin secreting plasma cells into a magnified, neoplastic image. The phenomenon is not exclusive for BL; a Swedish case of chronic lymphatic leukemia has been found with the same cellular characteristics ( Johansson and Klein, 1970). Quantitative comparisons between different lymphoma lines of this type showed (E. Klein et al., 1970) that they carry different amounts of IgM K-reactive material on their membranes. Although an extensive search has been made on more than sixty biopsies and derived lines, so far no membrane-associated immunoglobulins other than IgM K were found. Whatever the nature of the membrane-associated immunoglobulins, it is important to note that this property behaved as a cell marker on repeated biopsies from the same patient. If present on the cells of a given tumor, it was maintained unchanged in the course of repeated biopsies; if absent, it remained absent. It was also maintained following successful heterografting of membrane IgM-positive BL cells to the rat (Levin et al., 1969). The IgG coat behaved quite differently. It was rarely pi-esent on untreated BL biopsy cells, but tended to appear if the
IMMUNOLOGICAL ASPECXS OF BURKITT’S LYMPHOMA
203
tumor persisted in spite of treatment. It accumulated following a recurrence that was presumably due to the selection of a tetraploid (immunoresistant?) BL cell variant (Clifford et al., 1968; G. Klein et al., 1969a). “Self-enhancement,” i.e., the accumulation of “blocking” antibodies that prevent the access of immune lymphoid cells (K. E. Hellstrom and Hellstrom, 1969) is an obvious possible explanation. Whether these considerations are realistic or not, the changing pattern of IgG coating with time and its failure to persist on derived in vitro lines (E. Klein et al., 1968; Nadkarni et al., 1969), indicates that it is owing to coating from the outside, unlike the membrane-associated IgM, which appears to represent a special type of production from the inside. The presence of preformed immunoglobulin on the cell surface may interfere with the indirect membrane IF reaction and, when present in subliminal degree, it probably explains some of the variability encountered when biopsy cells are used as targets. In order to avoid this variability, we started looking for more standardized target cells and turned to established culture lines. In addition, efforts were made to improve the system by selecting one representative, reactive Burkitt serum, derived from a patient in long-term regression after chemotherapy, and study very carefully the presence of isoantibodies. This approach has shown (G. Klein et al., 1967a) that the positive reference serum ( Mutua ) was free of detectable isoantibodies, as judged by lymphocytotoxicity, leukoagglutination, and membrane fluorescent tests, performed against cells derived from donors typed for isoantigen components of the HL-A system. Mutua’s serum reacted with thirteen of sixteen BL biopsy cells and with five of eight tissue culture cells of BL origin, but not with nine established suspension culture lines derived from various other lymphomas and leukemias or from normal white cells of the peripheral blood. It did not react against normal bone marrow cells of BL patients or allogeneic lymph node or leukemia biopsy materials. All reactivity was present in the 7S, IgG fraction, as shown by serum fractionation (Smith et al., 1967) and by the use of monospecific conjugates ( G . Klein et al., 1967a). Among the BL lines tested for membrane reactivity, Jijoye, B35M, SL1, and EB3 reacted with Mutua serum, whereas Ogun, Kudi, and Raji were negative. The four reactive lines were all known to carry EBV, whereas, of the three negative lines, two were EBV (i.e,, VCA)-negative, and the third carried EBV in a low proportion of the cells ( G . Klein et al., 1968a). It was, therefore, suspected that membrane reactivity might be associated with and, perhaps, determined by the
204
GEORGE KLEIN
presence of EBV. To examine this possibility, a study was carried out on newly established lines (Nadkarni et al., 1969) that were tested in parallel for EBV-related intracellular IF and membrane fluorescent reactivity (G. Klein et al., 1968a). The lines showed approximately the same sequence of reactivity with regard to both tests. There was no constant ratio between the proportions of reactive cells, although, as a rule, approximately 10 times more cells showed membrane fluorescence than EBV antigen. Several newly established cultures revealed significant degrees of membrane fluorescence from the beginning, i.e., as soon as sufficient numbers of viable cells became available for the first test. After the explantation of biopsy material, EBV-positive cells were rare initially but reached relatively high levels on further maintenance of the cultures. Occasionally, membrane fluorescence and EBV reactions appeared gradually and in parallel. Established lines that contained no or very few EBV-positive cells were, as a rule, negative in membrane fluorescent tests as well. Cell lines carrying EBV, established from a patient with infectious mononucleosis, showed membrane fluorescent reactivity of the same type as the BL-derived lines (G. Klein et al., 1968b). Biopsy preparations, in contrast to BL cell lines, were characterized by positive membrane fluorescence, while showing negative EBV reactivity. This suggests that the production of the viral nucleocapsid antigen is suppressed in the tumor cell in uivo. The suppressive factor could be antibody, but there are many other possibilities. Another curious observation was that repeated establishment of parallel lines from the same patient, derived from successive biopsies, led to lines with fairly similar EBV levels, whereas lines derived from different patients were quite different (Nadkarni et al., 1970). This suggests that the viral “load per cell, or the activatability of the virus, or both, are characteristic for the individual tumor. Since the membrane-associated IgM marker mentioned above and another study with G6PD isozyme markers (Fialkow et al., 1970) strongly indicate that the BL process has a clonal origin, this may reflect the virus-cell relationship that characterizes a particular clone. This is also suggested by the closely similar levels of EBV-DNA hybridizable cellular DNA in repeated biopsies taken from multiple tumors of the same BL patient (zur Hausen et al., 1970). The postulate that the membrane antigen (MA) is determined by the EBV was directly confirmed when it was found that MA could be induced to appear in EBV-negative lines by the admixture of heavily irradiated EBV-carrying cells (G. Klein et al., 1967c) or by infection with EBV concentrates (Gergely et al., 1971b; W. Henle and Henle,
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
205
1970; Horosziewicz et al., 1970). In the infected cells, MA appeared after 20 to 24 hours. Deoxyribonucleic acid inhibitors, such as Ara C or IUDR, did not prevent its appearance, whereas puromycin inhibited it completely (Gergely et al., 1971b). Similarly to EA, it behaved, in other words, like an early product of the viral genome, not requiring viral DNA synthesis. In contrast to the EA system, appearance of the MA did not inhibit host cell macromolecular synthesis, however (Gergely et al., 1 9 7 1 ~ )A . better parallel is thus shown with the membrane and T antigens found in experimental oncogenic DNA virus systems (Deichmann, 1969; Meyer et al., 1969; Rapp et al., 1965) than with EA. Although the relationship between EBV and MA was clarified by these studies, this applies only to the EBV carrier cultures in uitro, and it must be kept in mind that similar compelling evidence is lacking for a connection between MA detected on biopsy cells and the virus. Indications are strong that the biopsy cells probably express the same MA as the carrier cultures (Smith et al., 1967). The antigenic components entering the EBV determined membrane and the intracellular nucleocapsid complex, respectively, differ with regard to immunological specificity. By absorbing sera that reacted with the membrane and the intracellular EBV complex as well, with large numbers of intact, viable, MA-positive cells, it was possible to remove the membrane-reactive antibodies, with only a minor reduction in the anti-EBV titer (Pearson et al., 1969). Moreover, some sera could be found with antibodies against only the MA, or the VCA antigen, but not both. Although such “discordant” sera were in the minority, their existence was in line with the immunological distinctness of the two antigen types. Direct proof of a steric and antigenic difference was obtained by two-color I F tests on the same cells (G. Klein et al., 1971b). Further analysis of the two antigen systems revealed (Svedmyr et al., 1970, 1971) that the MA and the intracellular antigen detected by the Henle test must be regarded as antigen complexes, with a number of distinct subcomponents. Sera that contain antibodies against several subcomponents of the intracellular EBV complex also tend to carry, as a rule, several antibody components against various parts of the MA complex, but the relationship is not absolute and many combinations can be found. Patients with large, persisting tumors frequently had a larger number of serum antibody components against both antigen complexes than sera from healthy, EBV-positive individuals, sera from donors convalescing from infectious mononucleosis, or sera from BL patients whose tumors had gone to long-term regression following chemotherapy. The nature of MA, particularly its specification by the viral or the
206
GEORGE KLEIN
cellular genome remains to be clarified. Recently, evidence has accumulated suggesting that it may represent a viral envelope component. The ability of different sera to neutralize an artificial EBV infection of EBV-negative culture lines (such as Raji or 6410) could be related to their titer of membrane-reactive antibody and not to the antiEBV titer (Pearson et al., 1970). This was particularly apparent when a series of discordant sera were tested with great differences in their antiEBV and MA reactivity. In another type of test, the sera of rabbits immunized with EBV concentrates were able to block the MA reaction specifically (Bremberg et al., 1969). This indicated that MA was present in the immunizing material, either as a constituent or as a contaminant of the virus particles. In a third type of test (Gergely et al., 1971b), cultures with closely similar frequencies of EBV( VCA) -positive cells but with very different proportions of membrane-positive cells absorbed the virus-neutralizing activity of BL patients’ sera in direct relation to their MA reactivity. It has been shown that HSV induced the appearance of new MAS in infected cells (Roizman and Spring, 1967). At least some of these antigens, detectable by immune adherence, appeared to be independent of viral DNA synthesis and were, therefore, probably early products of the viral genome (Peter-Knecht et al., 1968). Furthermore, viral mutants with different envelope characteristics induced different membrane changes, and there was a close correspondence between the new membrane components and the corresponding viral envelope properties ( Roizman, 1971). This prompted the conclusion that the membrane changes are owing to the incorporation of viral envelope material. Presumably, the virus changes the cellular membrane in order to facilitate the process of its own envelopment. In view of the parallel between EBV neutralization and membrane-reactive antibody levels, it is conceivable that the EBV-associated MA also represents viral envelope material. Recent immunoferritin studies by Silvestre et al. (1971) confirm this assumption in showing the presence of common or cross-reactive antigens on viral envelopes and infected cell membranes, distinct from the viral capsid antigens. This is of interest, because, for HSV, a relationship has been demonstrated between the changed “social behavior” of infected cells and their membrane modification after exposure to different HSV mutants ( Roizman and Spring, 1967). Conceivably, EBV-induced membrane changes may also influence cellular interactions. In a nonlytic system this may have important consequences for growth regulation. Further exploration of this area may help to elucidate the relationship between EBV and the neoplastic diseases with which it is most regularly associated. a. Disease-Related Patterns of Membrane-Reactive Antibodies. For a study of disease-related distribution of antibodies reacting with EBV-
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
207
associated MA, indirect fluorescence is obviously impractical. The disadvantage is that all surface-attached immunoglobulins capable of reacting with the conjugate show up equally well, irrespective of site specificity. Isoantibodies are particularly prone to occur if the donor has received repeated blood transfusions. This is an important potential source of error. An extensive search for isoantibodies in every serum is obviously impractical. The method was, therefore, greatly improved when the indirect membrane test was replaced by a direct test, amenable to blocking experiments. As the main reference reagent, the highly reactive immunoglobulin of a BL patient in long-term regression (Mutua) was chosen, free of demonstrable isoantibodies. The Mutua conjugate reacted with the same target cell lines as the unconjugated Mutua serum in the indirect test (Goldstein et al., 1969). Test sera were compared for their ability to block the membrane staining with the Mutua conjugate (Goldstein et al., 1969; G . Klein et al., 1969b). This restricted the testable antibody population to only those antibodies that could interfere with the attachment of the reference conjugate. It is particularly important that the isoantibodies of repeatedly transfused patients were unable to block the direct reaction, in spite of the fact that they reacted strongly with the target cells in the indirect test. Most BL sera (Gunvkn et al., 1970) and NPC sera (de Schryver et at., 1969) showed high blocking activity. Normal African controls, Burkitt patients’ relatives, and African tonsillitis patients’ sera were mostly negative, although occasional positives were encountered. Patients with infectious mononucleosis showed a rise in membrane-reactive antibody level during their disease (G. Klein et al., 196813).Head and neck tumors other than BL or NPC were also largely negative but occasional highly positive sera have been encountered. The difference between the regularly high-reactive African or Chinese NPC and the predominantly lowreactive Indian hypoharyngeal and oropharyngeal carcinomas was particularly remarkable ( d e Schryver et d., 1969). One interesting question concerns the relationship between geographical locale and serological reactivity. Nasopharyngeal carcinomas are easier to evaluate in this respect, since they represent a clear pathological entity and are not readily confused with other conditions. The EBV associated serological reactivity of African, Swedish, French, Chinese, and American cases was uniformly high and appeared to be mainly characteristic for the anaplastic or poorly differentiated type (W. Henle et al., 1970a; Old et al., 1968; de Schryver et al., 1969). The evaluation of BL outside Africa presents a more difficult problem, because histopathology alone does not allow a sharp distinction against other lymphomas. The combined clinical and pathological picture has readily recognizable features in the high-endemic areas, but they are less characteristic in other regions and the classification becomes
208
GEORGE KLEIN
more arbitrary. Serological findings on Burkitt-like lymphomas outside Africa are partly in line with African BLs (Ahlstrom et al., 1967; G. Klein et al., 1970) and are partly different, i.e., have no distinctively high EBV-associated reactivity (defined as a high anti-EBV titer and/or a high level of membrane-reactive antibodies) and, thus, resemble ordinary lymphosaromas rather than “true” BLs. At present, this picture cannot be interpreted meaningfully since pathology cannot unequivocally guide any decision as to whether non-African cases are African Burkittlike or not, It is quite possible, of course, that the non-African Burkittlike cases are heterogeneous. Some of them may be “true Burkitts,” i.e., have the same etiology as the African cases, whereas others would be different and comparable to ordinary lymphosarcomas. Whether a classification can be built on the EBV associated serological pattern will, of course, depend on whether the relationship of EBV to BL is of an essential or of an accidental nature. Another approach to the study of disease-related EBV patterns is to follow the antibody titers horizontally, during the course of BL, NPC or other EBV associated diseases. For comparisons one may select seropositive individuals with neoplastic diseases that are not regularly associated with high anti-EBV titers. Some preliminary information is available. In BL, the indirect membrane test, performed with biopsy cell targets (G. Klein et al., 1966, 196%) indicated that the most reactive sera were derived from patients in Iong-term regression. Later, when the more specific and sensitive blocking of direct membrane fluorescence reblaced the indirect test and established culture lines were used instead of biopsies (Gunvkn et al., 1970; G. Klein et d.,1970), nearly all histologically confirmed African BL sera had high reactivity, i.e., showed a complete or nearly complete blocking of the reaction with the reference conjugate. The few exceptions have come from moribund patients. This monotonously uniform blocking activity, obtained with the undiluted sera, hides large quantitative differences, however. When compared by serial titration against the same reference conjugate, the blocking titer (taking a 0.5 blocking index as the end point) can vary between 1:2 and 1:400 (Gunv6n et al., 1970). In the individual patient, the titer may change considerably in the course of the disease, although it frequently remains within the same range. Most changes are restricted to relatively few dilution steps up or down, and the patients can be classified into groups of low, medium, and high reactivity. Preliminary findings indicate that the titer differences between patients and the horizontal changes within a patient are inffuenced by a number of factors. In the course of rapid and extensive tumor growth, antibody levels often fall, probably due to adsorption to the tumor cell
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
209
membranes. After chemotherapy and tumor regression, there is often an increase in titer. At first sight, this may seem paradoxical, in view of the immunosuppressive effect of chemotherapy. It is known, however, that chemical immunosuppression inhibits new primary antibody responses against antigens administered after the drug but is much less efficient against immune reactions already established before treatment. An increase in membrane-reactive antibody titers was particularly apparent in BL and NPC patients after local radiotherapy (Einhorn et al., 1970), including cases where therapy did not lead to complete tumor regression. This is of considerable interest, in view of the fact that X-irradiated tumor cells are relatively good immunogens in experimental systems (G. Klein et al., 1960). It may be also relevant that X-rays can induce the appearance of the MA on a large fraction of cells in certain EBV carrier cultures with a relatively low MA reactivity (Yata et al., 1970). In BL patients with recurring tumors, which continue to grow in spite of chemotherapy, membrane-reactive antibody levels that have decreased at or around the time of recurrence can rise again (e.g., G. Klein et al., 1969a). Subsequently, the lymphoma cells become coated with IgG, as a rule, if the tumor persists. It is conceivable that such cells represent immunoresistant variants ( Fenyo et al., 1968). This is supported, indirectly at least, by the history of 2 patients whose tumors recurred after several years of total regression and contained a high frequency of near-tetraploid cells (Clifford et al., 1968), in contrast to a large number of BL biopsies examined (Manolov, 1970), with a shorter clinical history, which were all in the near-diploid range. Tetraploid cells can frequently outgrow host responses that efficiently reject diploid cells of the same lineage (Hauschka et al., 1956). The immunoglobulin coat acquired by tumors that persist in spite of therapy presumably represents antibodies directed against cell surface antigens, since there was a good correlation between the immunoglobulin coating of biopsy cells, as assessed by direct membrane IF with conjugated anti-immunoglobulin reagents and the number of C1 molecules attached to the cells (Nishioka et al., 1968). This may be the equivalent of enhancing or blocking antibody in Hellstrom’s sense (K. E. Hellstrom and Hellstrom, 1969) and is not necessarily an alternative to the possibility that membrane-reactive antibodies may have a growth inhibitory action but rather another facet of the same complex picture. An antibody that is cytotoxic for or inhibits the growth of immunosensitive cells may enhance the growth of immunoresistant cells, i.e., protect them against the cell-mediated immune response. In addition, different antibodies no doubt differ; some can be
210
GEORGE KLEIN
cytotoxic and others enhancing toward the same target cell. In the course of chemotherapy that falls short of a total tumor kill, and the subsequent regrowth of the residual tumor with more antigen release and antibody binding, the immunosensitivity of the tumor cell population and the killing versus enhancing power of the antibody population must obviously change in a complex way. A multicompartmental experimental analysis of these events is not yet within reach. In addition to the changes in the membrane-reactive antibody levels brought about by the tumor itself (due to absorption, antigen release, effect of tumor growth on the immune response, etc.), the antibody titer may change for other, tunior-unrelated reasons, and this may, in turn, influence tumor growth. This possibility has been brought into focus by the history of a BL patient (G. Klein et al., 1969a) who was in total tumor regression for a period of 4% years and subsequently developed widespread abdominal metastases. Her niembrane-reactive antibody level fell abruptly, more than 6 months prior to recurrence, at a time when there was no reason to suspect the presence of any metastases. When the abdominal tumor recurrence became obvious half a year later, the membrane-reactive antibody level was still low, and the tumor cells were not yet coated with IgG. In the course of the subsequent 2 months, the serum antibody level increased again and the lymphoma cells became IgG-coated. The antibodies against EAs were at a moderately high level while the tumor was in regression, but increased to a very high level after the recurrence. At the same time immunoprecipitating antibodies appeared, reacting against the soluble antigen described by Old et al. (1968). This sequence of events does not lend much support to the possibility that the fall of the niembrane-reactive antibody level 6 months prior to recurrence was merely due to absorption by a cryptic tumor. In that case, a period of slow tumor growth would have occurred during the subsequent 6 months period, and the secondary increase in antibody level, as well as the immunoglobulin coating of the lymphoma cells should have taken place in the interim, appearing already at the time of clinical recurrence. Indirect as this reasoning is, it serves to raise the question whether a fall in membrane-reactive antibody levels may be the cause and not merely the consequence of tumor recurrence in facilitating the outgrowth of “dormant” neoplastic cells. B. CYTOTOXIC AND GROWTH INHIBITION TESTS Plasma from patients with Burkitt’s tumor in regression for more than 3 years after therapy were reported as having a certain antitumor activity when infused into untreated patients (Ngu, 1967a,b). Control
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
211
plasma had no similar effect in the few patients where it was tested. Parallel tissue culture studies suggested the presence of growth-inhibiting factors in some Burkitt sera and some nontumor-bearing adults as well (Osunkoya, 1967a). Although Burkitt patients had low serum IgM levels as a rule, the suppression of growth of BL cells in tissue culture was associated with increasing IgM values, whereas in normal healthy children such trends were not evident (Ngu et al., 1966). Recently, Fass et al. (1970b) infused 8 untreated BL patients with plasma from BL cases in remission or from normal volunteers. The donor plasma gave positive reactions in the anti-EBV (VCA), anti-MA (blocking of direct membrane I F ) , and cytotoxic tests (against BL biopsies), but there was no evidence for any antitumor activity in vivo. In contrast to the previous report, this study was performed under a controlled, double-blind protocol. The authors point out, however, that the donor plasma would have to possess very potent antitumor effects to produce noticeable reduction in large tumor masses. They suggest that infusion of remission plasma might yield more dramatic results if it were performed after chemotherapy, when only a few tumor cells remained. The risk of immunological enhancement may become considerable under such circumstances, and caution in the use of humoral antibody for immunotherapy in vivo may be prudent, until the mechanism of enhancement is better understood. As the experiments of Amos et al. (1970) illustrate, sera that kill certain target cells in vitro, in the presence of complement, may fail to inhibit the same cells in vivo and can occasionally enhance them. In the Burkitt system, the possibility of immunological enhancement is suggested by the observation of Clifford et al. (1967) that the infusion of remission sera was followed by increased tumor growth in 2 patients, although, as Fass et al. (1970b) point out, rapid increases in size occur so frequently in BL cases that immunological enhancement cannot be proven by these observations. Osunkoya (196713) tested the effect of Burkitt and other sera on the growth of an established BL cell line (OB3) in vitro, in relation to the indirect I F reaction obtained with the same sera against surface antigens. Sera from 37 BL patients (21 untreated and 16 in remission) showed a lower growth-promoting effect in vitro than the sera of 105 normal Nigerian blood donors, 14 parents of BL patients, 20 sick Nigerian children with nonmalignant diseases, 13 malignant lymphoma patients, 37 Negro and 42 Caucasian blood donors from New York, and 15 newly arrived and 18 resident American Peace Corps volunteers in Nigeria. Sera of the Burkitt group in remission showed an even lower growthpromoting effect than that from untreated BL cases. The three sera that gave the lowest growth indices had been derived from patients
212
GEORGE KLEIN
who had been symptom free during 2 or more years after treatment of BL. The membrane IF test gave a different pattern in that 57%of the Burkitt sera were positive against the OB3 cell, whereas 19%of the sera from children with various nonmalignant diseases and 20%of New York blood donors were positive. Of newly arrived American Peace Corps volunteers in Nigeria 7% (1 of 15) of the sera were positive, whereas from volunteers who had been resident in Nigeria for at least 2 years, 33%( 7 of 21) of the sera were positive. From Nigerian malignant lymphoma patients 43%, from parents of BL patients 64%, and from healthy Nigerian blood donors (in the age range 18-50) 41%of the sera were positive. There was no correlation between the two tests; sera with low growth indices in culture did not necessarily give a positive membrane IF reaction.
FIXATION TESTS C. COMPLEMENT Complement fixation ( C F ) tests with cell lines derived from BL and acute leukemia were first reported by Armstrong et al. (1966). They studied Burkitt lines of African origin (EB1, EB2, EB3, and SL1) and leukemia lines of American origin (6410, SKL1, and SKL2). The same lines were used as for the IF tests of G. Henle and Henle (1966a). For CF, cells were disintegrated by freezing and thawing on sonication. All but 1 of 13 sera from BL patients were positive against the EB1 and EB2 preparations, with titers ranging from 1:8 to 1:64. Of seventy-four sera from American children under 15 years of age, 1520%showed titers within the same range, regardless of their illness, Sera from 73 adults, including healthy donors and patients with various diseases, yielded positive results in 44 cases (60%),with occasional titers as high as 1:128 or 1:256. There was no correlation with antibody levels against HSV or CMV. A number of selected positive sera were also tested against antigens derived from leukemia lines. A comparison was made between the antigenic reactivity of EB1, EB2, EB3, and SL1 of Burkitt origin, 6P10, SKL1, and SKL2 of leukemia origin, and, in addition, primary leukocytes and HeLa cells. Some sera gave positive reactions with all antigens prepared from lymphoma and leukemia cell lines. Other sera failed to react with SL1, and SKLl but reacted with the others. None of the sera reacted with the HeLa antigens, and there was only one exceptional reaction against leukocyte antigens with serum derived from a patient who had received blood transfusions. Selected positive sera failed to react against antigens prepared from cultured neuroblastoma or Wilms’ tumor cells. The reactivity of some sera with all neoplastic leukemia
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
213
and lymphoma lines suggested the presence of common antigenic components. A comparison of the CF and IF tests (G. Henle and Henle, 1966a) showed certain similarities, but it also revealed some discrepancies. In both tests, the percentage of positive sera increased markedly with age. The two tests were in agreement for 72%of the sera studied. Most of the remaining sera gave positive indirect IF but no CF. The reverse situation was found in only a very few cases. All Burkitt sera gave positive IF, but 1 of 13 failed to fix complement. Another and presumably more important point of discrepancy was the fact that the 64-10 line of myeloid leukemia origin (Ikawata and Grace, 1964) contained no IF-positive cells but was, nevertheless, highly active in CF tests. It was pointed out, however, that the cell preparations contained more than one antigen; preliminary fractionation experiments suggested the presence of several active components. It was stressed that organ-specific cellular constituents, isoantigens, viral proteins, and tumor-specific components might all contribute. Some components might be related to a virus carrier state, whereas others could result from a nonpermissive interaction between the viral and the cellular genome, in analogy with experimental systems. In a later study, Gerber and Birch (1967) used a quantitative C F technique and partially purified viral antigens derived from the P3 (Jijoye) BL line, to detect antibodies in various sera. Disintegrated Jijoye cells were fractionated on a sucrose gradient and the fractions containing the highest concentration of EBV were localized by electron microscopy. For comparison, the EBV-free Raji line was fractionated in the same way and corresponding fractions were collected. Herpes simplex-infected BHK hamster cells served as another control. Twentyone of twenty-five African Burkitt sera gave positive reactions against the P3 antigen, with a GMT of 662 (range 305800). All but one gave low-titer reactions against HSV antigen (GMT 37). None of them reacted with the Raji cell fraction at the lowest serum dilution of 1:30. Among 22 American children with noninalignant diseases, 12 reacted against the P3 antigen, with titers ranging from 40 to 1174 (GMT 59). Six reacted with the HSV (GMT 25), none reacted with the Raji antigen. Four of 6 American BL cases and 7 of 7 American carcinomas of the postnasal space were positive against the P3 antigen (GMT 178 and 324, respectively), Among healthy American adults, 21 of 22 reacted with the P3 antigen (GMT 110). Therc was no relationship between the P3 and HSV antibodies in any of the serum categories. Antibodies to the P3 antigen started appearing in American children at 2 to 3 years
214
GEORGE MEIN
of age. Among nonhuman primates, there was a high incidence of positive sera against the P3 antigen in chimpanzees, cynomolgous, rhesus, and African green monkeys, but not in baboons. The GMT ranged from 40 to 348 in the different categories (chimpanzees being the highest). None of these sera reacted with the Raji antigen. A large number of other species, including rodents and all major domestic animals were negative against the P3 antigen. In ten selected human sera, a good correlation was obtained between the CF titer and the antibody coating of P3 virus particles (Mayyasi et al., 1967). This was interpreted to mean that the C F antibodies were directed against a viral component. Recently, the CF antigens prepared from BL-derived lines were separated into soluble and sedimentable components, and the corresponding antibody reactions were assessed separately by three different groups (Gerber and Deal, 1970; Pope et al., 1969a,b 1971; Vonka et al., 1970a). Vonka et al. described the presence of a soluble ( S ) CF antigen in EB3 and P3J, two EBV carrying BL lines (Vonka et al., 1969, 1970b). They suggested that this antigen was distinct from the EBV-associated antigens detected by the Henle test (and this corresponds to VCA in Henles’ new terminology) and also from the antigen(s) detected by the CF test performed with semipurified virus particles, since there was no relationship between EBV content and S antigen titer and also because the reactivity spectrum of human sera against viral and S antigens was different. They, nevertheless, concluded that the S antigen was associated with EBV infection, because no anti-S reactivity was ever detected in subjects lacking anti-EBV reactivity in the Henle test. It was also important that a blastoid cell line derived from a normal donor (AMC3O) and free of detectable EBV, contained an antigen that was closely related to, or identical with, the S antigen of the Burkitt lines. Closely similar findings were reported by Pope et al. (1969b). They have compared 158 human sera for the presence of CF antibodies and antibodies demonstrated by the Henle test, against two EBV carrier cultures (GOR and WIL, respectively). All 37 sera that were negative in the IF test were also negative in the CF test. All 83 sera with CF antibody also had I F antibody; 38 sera had IF antibody but were negative in the CF test. A selected assortment of human sera that were positive in the CF test, gave negative reactions against various other tissue antigens and established cell lines used as controls. There was no association with CF antibodies to HSV. Pope et al. also reported that 36 human sera tested in parallel against the two EBV carrier lines and the IF-negative Raji line gave closely similar CF reactions against all
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
215
three. The antigen responsible for this reaction was not sedimented during 2 hours at 60,OOOg. It may be pointed out that the earlier negative CF results of Gerber and Birch (1967) were obtained with sedimentable fractions of the Raji cell. The behavior of the Raji line is of considerable interest. Although derived from a case of BL, it contained no EBV particles and lacked all three antigen systems that could be demonstrated by IF (VCA, EA, and MA). It may be pointed out, however, that Raji cells contain DNA that hybridizes specifically with purified EBV-DNA (zur Hausen and Schulte-Holthausen, 1970). Very recently, Pope et al. ( 1971) also reported that an immunoprecipitating antigen could be extracted from the Raji line that gave a reaction of identity with one of the soluble antigens present in EBV carrier cultures (see next section), In a recent study, Vonka et al. (1970a) further characterized the S antigen. They could demonstrate it in three EBV carrier lines and in three blastoid cell lines that showed no evidence of EBV by electron microscopy or by the Henle test. The former were derived from 2 cases of BL and 1 infectious mononucleosis, respectively, whereas the latter three came from the peripheral blood of 3 healthy donors. Virtually identical titers were obtained with nineteen different human sera, tested against two S antigen preparations, derived from EBV carrying and EBV-free lines, respectively, and within a titer spectrum ranging from <10 (negative) to 320. It is particularly significant that among 10 sera that were negative with S antigen extracted from the EBV-carrying EB3 line, 9 were also negative for S antigen extracted from the EBV-free NC37 line, and 1 reacted at the level of 1:10. Antibodies against the S antigen could not be removed by absorption with EBV-positive or negative cell lines, nor did absorption with the S antigen block the Henle-type IF reaction. The S antigen could not be localized within the cell, either by routine IF or by anticomplementary IF. It was not identical with the MA, since the latter was only present in lines that showed a positive Henle reaction, whereas the S antigen was present in other lines as well. It may be added that, although no tests have been reported as yet concerning the question whether S antigen might be related to the recently discovered EA, discussed above, the same argument could be applied-EA-positive cells have not been found in EBV-negative lines, containing the soluble C F antigen, e.g., the Raji cell line. Vonka et al. (1970a,b) suggested that the S antigen was a nonstructural protein, coded by the EBV genome. The presence of the S antigen in some blastoid cell lines in the absence of other EBV associated antigens, would be either owing to the incorporation of an incomplete
216
GEORGE KLEIN
viral genome or to the presence of specific repressor(s). The S antigen would thus resemble the T antigen in the experimental, oncogenic DNA, virus systems. A parallel study of Gerber and Deal (1970) was based on the use of a viral antigen preparation, separated from the virus-rich supernatant fraction of the EBV carrying P3J line by sucrose-gradient centrifugation, and a soluble, virus-free extract prepared from concentrated culture fluids of the EBV carrier P3J and the EBV-free Raji line. Among 105 human sera, none reacted with the control antigen, a sedimentable fraction of Raji cells. Antibody GMTs against the viral antigen were 48 and 33 for 70 healthy American adults and 30 children, respectively, and 152 for 10 American cases of infectious mononucleosis. The titer was 980 in 10 African BL patients and 180 in 15 normal African controls. Corresponding titers against the soluble P3J antigen were 8 and 6 for the American controls, and 13 for American infectious mononucIeosis cases, 94 for the African BL, and 32 for their controls. None of 31 sera that were negative against the viral antigen, reacted with the P3J or with the Raji S antigen. Of the 54 sera that gave positive reactions with the P3J-S antigen, only 21 reacted with the Raji S antigen, This was different from the report of Vonka et al., discussed above, where the frequencies and titers of reactive sera were virtually identical against the S antigen preparations made from an EBV carrier and an EBV negative line, respectively. This difference could be purely quantitative, however, probably related to the fact that Vonka et al. extracted their S antigen from sonicated cells, whereas Gerber and Deal derived it from concentrated culture supernatants. A further subdivision of the antigens detectable by CF was recently presented by Walters and Pope (1971). Three components were identified in the EBV carrier WIL line. One was a virion component, harvested from sucrose-density gradients and resistant to heating at 56°C for 30 minutes. A second, less easily sedimentable CF component was heat-labile. The third antigen was heat-resistant and soluble. Only the latter component was present in the Raji line. The new, heat-labile component was detected in two further EBV carrier lines, In view of the fact that BL tumor cells in civo express MAS not present on the Raji cell but do not contain VCA and EA demonstrable by IF, as a rule, it would be important to know if they contain any of the antigens detectable by CF. At first, McCormick et al. (1969) reported that CF antigen could not be detected in BL biopsy material. More recently, Pope (1971) detected EBV-related CF antigen in two of three biopsies examined. The differences between the results of the two groups may be owing to quantitative factors or to technical details in
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
217
the preparation of the antigen. Alternatively, the demonstrated antibody coating of certain BL biopsy cells in vivo may interfere with the isolation of active CF antigenic components. The nature of the CF antigen detected in the two biopsies by Pope has not yet been characterized in relation to the various CF antigens identified in tissue culture systems.
D. IMMUNOPRECIPITATION A precipitating antibody, present in human sera, against an antigen isolated from EBV carrying BL cells in vitro, has been described by Old et al. ( 1966). These authors performed Ouchterlony immunodiffusion tests by preparing soluble antigen from frozen, thawed, and homogenized Jijoye cells, known to contain relatively large numbers of EBV particles after aging. One milliliter of the antigen preparation was derived from approximately 5 x lo9 cells. In a survey of sera for precipitating antibody, 31 out of 55 (56%) African Burkitt sera and 14 of 19 (74%) African carcinomas of the postnasal space gave positive reactions. This was the first demonstration of a serological similarity between BL and NPC with regard to reactivity against an EBV-associated antigen. Only 6 of 52 (12%)African sera from malignant diseases other than BL or carcinoma of the postnasal space were positive. The only American serum category tested that gave a high frequency of positive reactions was NPC (19 of 20 positives). Among 296 sera from healthy adult American donors or from patients with various other malignant or nonmalignant disease, only 24, i.e., 11%were positive: there was no particular disease-related pattern. Among 18 sera from patients with infectious mononucleosis, only 1 was positive. Six of 47 acute leukemia sera were positive. Many positive sera gave one sharply defined precipitin band. With some strong sera there were one or two additional bands. The few positive sera from healthy donors gave lines of identity with the Burkitt and the postnasal carcinoma sera. Reactive antigens could not be extracted from a large number of other human cell lines, leukemia cells and tonsillar tissue. Negative lines included the myeloblastic and lymphoblastic leukemias 6410, SKL2, and SKL6 and also the EBV-carrying Burkitt lines SL1 and EB3. There was no relation between the antigens prepared from Jijoye cells and two soluble HSV antigens that could be demonstrated by the same type of precipitin reaction, nor was there any correlation between the anti-HSV and the anti-Jijoye antigen reactivity of the various sera tested. A more extensive serum survey for precipitating antibody was reported in a subsequent publication (Oettgen et al., 1967). American carcinomas of the nasopharynx gave 27 of 33 (82%)positive reactions.
218
GEORGE KLEIN
Of 40 American lyniphosarcomas, 12 (30%)were positive. A total of 125 acute leukemias from the United States gave 25 positives (20%).A large number of control sera from patients with malignant or nonmalignant diseases gave a constant level of 5 to 13%positives in the various groups. The precipitating antibody was identified as a heat-stable IgG. A positive correlation was found between the immunoprecipitation test against the Jijoye antigen and the indirect IF reaction of the Henle type against EBV-carrying Jijoye cells. It appeared that the antibody threshold required for immunoprecipitation was higher than for the fluorescent reaction. The majority of sera that were positive in the IF test failed to precipitate, whereas all precipitating antisera gave a hightitered I F reaction against the Jijoye cells. The authors point out that IF and iniiiiunoprecipitation can be regarded as complementary methods -sensitivity is the main advantage of IF, whereas immunoprecipitation is unparalleled with regard to its power of resolution. Using the imniunoprecipitation test of Old et al., evidence has been obtained to indicate that antibodies against the soluble EBV associated antigens can be absent or reach undetectable levels in patients whose tumors have completely regressed, and may reappear on recurrence (G. Klein et al., 1969a, 1970). Although there were numerous exceptions to this when different patients were compared, a clear relationship appeared when the horizontal history of a BL patient was followed during long-term regression and subsequent recurrence ( G. Klein et al., 1969a). This may also explain why high anti-EBV titered NPC sera were more frequently precipitin positive than BL sera with comparably high titers. In NPC, the sera were mainly derived from patients with residual or progressively growing tumors, whereas collections of BL sera tested included progressor and regressor sera as well. The association between persisting tumors and precipitating antibody is reminiscent of the disease-related anti-EA pattern, already discussed. Nevertheless, it is clear that the soluble antigens detected by the test of Old et aE. are not identical with the EA system. In a comparison of the two tests on the same serum material (G. Henle et al., 1971), the majority of the sera were concordant in both tests, but a sufficient number of discordant sera could be identified as well, reacting in one test but not in the other. Since both complementary discordances were found to exist, it could be concluded that different antigenic specificities were involved. Both represented intracellular and probably nonstructural, early proteins, however. A third antigen of this kind corresponded to the CF S antigen, demonstrated in Raji and other EBVfree lines, as described in the previous section. The latter antigen might
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
219
be identical with one of the soluble antigenic components detected in the precipitin test. Pope et a2. (1971) recently showed by immunodiffusion that the soluble EBV-associated CF antigens of the virus-carrying WIL line and the virus-free Raji line contained at least one major and one minor component that gave lines of identity. Konn et al. (1969) compared the precipitin lines obtained with human sera and the patterns obtained with rabbit antisera against virus released by the HRlK subline of the P3J (Jijoye) BL line. In a material of 475 human sera, derived from 83 normal donors, 156 nonmalignant disease controls, and 236 malignancies of various types, no immunoprecipitation was obtained with 10 sera that were IF negative in the Henle test, 1 of 175 gave positive immunoprecipitation if the IF titer was less than 50, 25 of 172 if the titer was 100, 19 of 64 with titers between 200 and 300, 16 of 35 with titers between 400 and 500, and 18 of 19 with titers exceeding 500. Most of the high-titered sera were derived from cases of Hodgkin’s disease, lymphosarcoma, or chronic lymphatic leukemia, whereas other malignancies, such as acute lymphatic leukemia, acute myeloid leukemia, reticulum cell sarcoma, carcinomas, and melanoma gave lower titers and did not appear to differ significantly from the normal or nonmalignant disease controls, although the material was too small in several groups for a definite evaluation. When sera with identical IF titers, derived from donors with malignant or nonmalignant diseases, were compared, the proportion of immunodiffusion positive sera was consistently higher in the malignant disease groups. Twenty-one of the 79 ininiunoprecipitin positive human sera were tested together with the rabbit sera. All of them gave lines of identity with the rabbit antisera, against one of the antigens, designated as “a”. Eight sera yielded a “b” line as well, and only 2 yielded a “c” line. The latter was the only line that contained antibody-coated intact virions when examined by electron microscopy. The immunoprecipitating antibodies to the EBV antigens were removed by absorbing positive human sera with HRlK cells, but not with Raji cells. Konn et al. pointed out that the studies of Old et al. left open the question whether the immunoprecipitating antigens were virus-associated or represented some cellular constituent of virus-infected cells. The study of Konn et al. was interpreted to mean that the human sera gave precipitin lines with viral or virus-associated antigens, since the reactions were obtained with partially purified virus and gave reactions of identity with rabbit sera prepared against viral pellets. They neverthe-
220
GEORGE MEIN
less admitted that the soluble “a” and “ b components could be virally induced cellular antigens against which both the patients and the rabbits formed antibody. Using a microimmunodiffusion test, Fink et al. (1969) reported results that differed strikingly from the findings of Old et al. and Konn et al. In a comparison of immunodiffusion, indirect IF, CF, and antibody coating of virus particles, they found that the immunodiffusion test was most sensitive. It gave positive reactions with a number of sera that were negative by the other tests. Among the 121 sera positive in immunoprecipitation against the soluble antigen derived from the P3 cell line, 96 were positive and 25 negative by indirect IF, 100 positive and 21 negative by antibody coating, 98 positive and 23 negative by CF. None of the sera reacted with the extracts of two lines that were free of virus particles and negative in the Henle test, It is not clear whether the high sensitivity of the immunodiffusion test reported by Fink et al. is owing to the micromodification, to a difference in preparation of the antigen, or to other factors. Recently, Stevens et al. (1970) compared cellular antigen extracts from a number of EBV carrier and apparently EBV free cell cultures. These authors found immunologically indistinguishable antigens in sucrose-gradient purified virus prepared from supernatant culture fluids and from sonically disrupted cells of the P3J cell line, respectively, as well as in a soluble fraction, extracted from the HRlK clonal subline of the P3J cell and not sedimentable at 35,000 g after 1 hour. A maximum of four precipitin lines was found with the latter preparation and certain BL sera. The reactive antigenic components could be partially sedimented at 108,000 g from the supernatant, following the 35,000 g centrifugation step, while identical antigenic components still remained in the supernatant, even after the 108,OOOg centrifugation. This is consistent with partially disrupted and soluble virus subunits or viral proteins synthesized in excess. The possibility cannot be excluded, however, that these are virus-specific, but nonvirion antigens, or even that they are host cell proteins induced or increased by virus infection. Since the same antigens can be found on immunodiffusion with gradient purified virus, this is less likely, however. Normal host cell proteins could be associated with the virion, but they would most likely be in the envelope. Stevens et al. used a micromodified version of the immunodiffusion test, similar to the method of Fink et a2. They reported antibody distributions in various serum donor categories that resembled the results obtained with the Henle test and by CF and were thus similar to the findings of Fink et al. rather than of the Old group.
IMMUNOLOGICAL ASPECTS OF BURKITT’S
LYMPHOMA
221
In these studies, immunologically indistinguishable precipitin reactions were obtained with antigenic preparations derived for the Burkitt line P3J, its clonal subline HRlK, the EBV carrier lines F152 (of acute lymphatic leukemia origin), and F132 (of reticulum cell sarcoma origin). Certain quantitative differences were seen, however, with regard to the distribution of three precipitin lines (A, B, and C ) . With the BL derived EB3 line, only a single band (A) was seen, also present in the other four cell lines. Antigenic preparations from nine other cell lines and ten different biopsy specimens, including four BL biopsies did not yield any precipitin reactions with the sera tested, or, in some cases, gave reactions that did not show any identity with the four positive lines. It is particularly noteworthy that the F126 line, positive in the Henle test (24% IF-reactive cells), did not yield any immunoprecipitating antigen. The same was true for the Henle negative Raji line that was recently shown to share a precipitin line with an EBV carrier line (Pope et al., 1971). Stevens et al. interpreted their findings as indicating that the same virus type was associated with blastoid cell lines derived from African patients with BL and from American patients with leukemia and lymphoma, although they admitted that strain differences might exist that could not be demonstrated by this test. On the other hand, the precipitin lines obtained with the extracts of certain EBV free blastoid lines that gave no reactions of identity with EBV suggested the existence of other precipitating antigen-antibody systems. Absorption of antibody-positive reference sera with antigens contained in the growth medium, Forssman antigen or human EBV-free cell lines, failed to remove precipitin activity, whereas absorption with EBV containing tissue culture cells or extracted antigens was efficient. Ill. Studies on Cell-Mediated Immunity
A. DELAYED HYPERSENSITIVITY in V i m Only one report has been published on this subject so far. Fass et al. (1970a) skin-tested 12 patients with BL after initial biopsy. Only 1 patient, with a localized jaw tumor, gave a positive reaction to autologous tumor extract, and none reacted to autologous lymphocyte extract. Eight of these patients were retested after treatment, and 4 additional patients were tested for the first time when in clinical remission. Seven of the 12 had positive reactions to autologous tumor extract. All 7 remained in sustained clinical remissian for a median of 41 weeks after therapy. Four of the patients with negative reactions relapsed 14-20 weeks after therapy. It is noteworthy that the extracts were prepared by a modifi-
222
GEORGE KLEIN
cation of the method described by Davies (1966) and by Mann et al. ( 1968), developed to isolate cell membranes that carry transplantation antigens. They contained mostly surface membranes and some cell sap. Cutaneous reactivity could be elicited only by membranes that sedimented at 100,000 g during 60 minutes. Reaction was similar to isolated membranes and to whole cell extracts. The negative skin tests in patients with untreated tumor and during relapse raised the important question whether this response was owing to a state of general immunological anergy or to a more specific lack of reactivity to tumor-associated antigens. Fass et al. regarded anergy unlikely, since most of the patients gave positive delayed hypersensitivity (DHS) reactions to dinitrochlorobenzene (DNCB), even before therapy. Among 18 pretreatment patients ( 5 stage 1-11, 13 stage III-IV), 16 gave positive reactions; 8 of 9 BL patients in remission and 11 of 12 age-matched controls gave positive responses to DNCB as well. There was no difference in the quality of the response between the BL patients and the controls. Lymphocyte transformation by phytohemagglutinin (PHA) was tested in 22 BL pretreatment patients; 16 had positive lymphocyte transformation in the same range as the control subjects, whereas 6 showed impairment. Four of the latter were patients with disseminated disease in poor clinical condition. Apart from this, there was no apparent correlation among decreased lymphocyte transformation, DNCB reactivity, and prognosis. All patients in long-term remission showed lymphocyte transformation within the normal control range. Primary antibody response to the Vi antigen was greatly depressed in the BL pretreatment patients, compared to the remission patients and the controls. The IgG and IgM levels of BL pretreatment patients were within the control range, but the IgM concentrations were on the low side (Ziegler et al., 1970). Fass et a2. point out that the specificity of the skin reaction to autologous tumor extract must be investigated further, particularly in view of the fact that some control extracts have produced skin reactions in other human tumor studies (Hughes and Lytton, 1964; Stewart, 1969). In the work of Fass et aZ., 1 of 8 patients gave a positive reaction with normal peripheral lymphocyte extracts. Inoculation of autologous lymphocyte membrane extracts to normal volunteers showed that large amounts (derived from more than 10' cells per 0.1 ml.) could produce delayed skin reactions. Discussing the nature of the factors determining skin reactivity to tumor-associated antigens, Fass et al. considered the possibility that the immune response might become exhausted by the presence of a large tumor mass. If this were true, the negative skin reactions observed in 5
IMMUNOLOGICAL
ASPECTS OF BURKITT’S LYMPHOMA
223
patients in remission might be due to a residual undetected tumor. Alternatively, the low, cell-mediated, immune response of these patients might have contributed to the relapse of their tumors. In view of the correlation of the anti-EA titers with the probability of recurrence and, presumably, the presence of residual tumor ( W. Henle et al., 1970b), as already mentioned, it might be suggested that a study of the dnti-EA titers in such patients, if done in parallel with the DHS tests, might disclose further information about this important problem. The general .reactivity of BL patients in DHS tests was also studied by Stjernsward and Clifford (1970), with somewhat different results. Positive DNCB skin tests were found in 2 of 10 untreated BL patients, in 13 of 32 BL patients under chemotherapy, and in 5 of 7 patients in remission. The untreated group showed, furthermore, a certain reduction of lymphocyte stimulability by PHA, in comparison with the other two groups. B. MIXED LYMPHOCYTE STIMULATION TESTS
In mixed lymphocyte stimulation tests, involving the exposure of viable lymphocytes to autologous, mitomycin-treated BL biopsy cells, 3 of 16 BL patients showed a positive reaction (Stjernsward and Clifford, 1970). This relatively poor reactivity may be related to the infrequent skin response of untreated patients (Fass et al., 1970b) to autologous tumor extracts, since all mixed lymphocyte-tumor cell reactions were carried out with cells from tumor-bearing patients. Alternatively, the blocking of tumor-associated MAS by immunoglobulin attachment ( E. Klein et al., 1968) may have been responsible. IV. O n e or Several EB Viruses?
The question whether EBV is one virus, or a family of viruses is of considerable importance. Conceivably, a whole family of related viruses may exist, with closely similar morphological and antigenic characteristics, but different biological and oncogenic potential. There are many examples for this in the field of avian and murine leukosis. Another, potentially relevant point is that HSV mutants differ from each other with regard to the membrane changes they induce in lytically infected cells. Cells infected with the different mutants show different changes in their “social behavior” (Roizman, 1971). Since EBV is a nonlytic virus, it is conceivable that the membrane changes it induces may lead to various types of social misbehavior in different target cells. If coupled with maintained cellular viability and proliferative capacity, this may culminate in malignancy. Different subtypes of the virus may show preference for different target cells.
224
GEORGE KLEIN
A comparison of surface antigens on lymphoblastoid cells derived from BL and NPC patients, respectively, was performed by de Schryver et al. (1970). The reaction patterns obtained with a number of assorted reference conjugates, tested in combination with sera derived from NPC, BL, and infectious mononucleosis were indistinguishable wi!h the NPCand the BL-derived lines. Whereas these findings thus failed to demonstrate any difference, they merely showed that cross-reactive antigenic components are present. They did not exclude the possibility that, in addition, type-specific components may exist, which could be detected with more discriminative techniques. By using immunological methods of a different kind, Nishioka et al. ( 1971a,b) recently demonstrated a certain difference between BL cells (including biopsy and tissue culture lines as well) and blastoid cell lines derived from NPC, infectious mononucleosis, and normal peripheral white cells, respectively. Two kinds of cell membrane receptors have been identified on the surface of blastoid cells: ( a ) the immune adherence ( I A ) receptor, demonstrated by the adherence of sheep erythrocytes, sensitized with IgM antibody with added C4 and C3 (this receptor is actually combined with the C3 component of the complex), and ( b ) the “IgG receptor,” reacting with sheep red cells sensitized with IgG antibody. A similar IgG receptor was previously shown by Watkins (1964a,b, 1965) to appear on the surface of human monolayer cells, infected with HSV. The expression of the HSV-induced IgG receptor coincided with the appearance of viral antigens of the cell surface and was suggested to be an early function of the viral genome (Watkins, 1965). It did not react with free IgG nor with red cells coated with IgM antibody, but only with IgG-sensitized erythrocytes. Nishioka et al. (1971a,b) found that a considerable number of blastoid lines derived from NPC obtained from Taiwan, Hong Kong, and Nairobi showed the presence of the IA receptor and lacked the IgG receptor. To the contrary, all BL-derived cell lines (except one) and several BL biopsy specimens, lacked the IA receptor, but they all carried the IgG receptor. This included the EBV free Raji line. Two alternative possibilities were considered to explain the positive reaction of the BL cells with IgG-coated sheep red cells. It could be a cell marker, reflecting a property associated with the normal progenitor cell of the lymphoma or, alternatively, it could represent a virus-induced alteration of the cell membrane. Nishioka et al. attribute considerable significance to the fact that acetone or phenol treatment abolishes the reactivity of BL cells with sensitized sheep red cells, and normal IgG does not inhibit the reaction. The same behavior is also characteristic
IMMUNOLOGICAL ASPECTS OF BURKITT'S LYMPHOMA
225
for the HSV induced membrane receptor described by Watkins and is different from similar receptors occurring on the surface of certain normal cells. The Japanese workers, therefore, suggest that this may be a virus-induced receptor. If this interpretation is correct, the absence of a similar, virus-induced receptor on the surface of the NPC cell lines and the presence of another IA receptor raise a number of interesting questions. Conceivably, the NPC lines may harbor a different variant of the virus. Alternatively, they may harbor the same virus, but represent a different type of cell, unable to express the same membrane change as the BL cells. It should be possible to elucidate this important question and to distinguish between virally induced and cell-associated markers by appropriate crossinfection experiments. V. Immunological Studies on Oncogenic Herpes Viruses in Animals
The causative agents of at least two neoplastic diseases, Marek's neurolymphomatosis in the chicken (Churchill, 1968) and LuckB's carcinoma in the frog (Mizell et al., 1969) have been recently identified as herpes-type viruses. A simian lymphoma is also probably due to a herpestype virus ( Herpesvirus saimiri) ( Hunt et al., 1970). Immunological studies on these systems are still in a rather preliminary stage, but they are, nevertheless, of considerable interest, since a comparative study of virus-cell and virus-host relationships in the more easily accessible animal systems may serve as a model for corresponding studies on human tumor-associated, herpes-type viruses. A. MAREK'SDISEASE Purchase ( 1969) applied the indirect fluorescent-antibody method for the detection of Marek's disease (MD) antigen. Sera of hyperimmunized birds reacted with the nucleus and the cytoplasm of duck and chick embryo fibroblasts and chick kidney cells infected with MD. Uninoculated cultures were free of the antigen. Positive cells could be detected 24 hours after infection. No cytopathological changes were seen at this time. Seven days after infection, fluorescent areas coincided with the cytopathological areas. The kinetics of the infection indicated that one infectious unit produced one fluorescent or morphological focus. The same study confirmed the notorious cell association of this virus and demonstrated that the infection spreads from cell to cell. Eight different
226
GEORGE KLEIN
isolates of MD gave indistinguishable antigenic patterns. Electron microscopy showed ( Nazerian and Purchase, 1970) that all antigen-positive cells contained HV, and antigen-negative cells were virus-free. Recently, Chen and Purchase (1970) identified a surface antigen on chick kidney cells infected with MD. They observed a bright “ring” of fluorescence in infected chick kidney cells 24 hours after infection, when stained in the viable state in the ordinary membrane I F test. Normal cells surrounding the infected cells did not stain. Membrane-reactive antibody could be absorbed from the fluorescein-conjugated anti-MD chick globulin reagent by infected but not by normal chick kidney cells. Unlabeled anti-MD chicken serum blocked the membrane reaction, whereas normal sera had no blocking effect. There was no cross-reactivity with chick cells infected with infectious laryngotracheitis virus or Rous viruses of various types. The MA was different from the virion, since virus particles could not be found at the location of the antigen which could be demonstrated by ferritin-conjugated anti-MD chicken globulin as well. The authors suggest that the antigen is either a component of the cell membrane, modified by virus infection, or a new component made by the cell in response to virus infection. They consider it likely that the antigen is present on the envelope of the complete virion. In a combined fluorescent antibody and electron microscopy study, Nazerian and Purchase analyzed the duck and chick embryo fibroblast cultures infected with MD virus (Nazerian and Purchase, 1970). Two morphologically distinct IF antigens could be seen in the cytoplasma granular antigen in the perinuclear region, showing bright staining and lacking virions, and a diffuse antigen, present throughout the cytoplasm of the infected cell, less brightly stained and containing occasional naked virions. A diffuse nuclear I F antigen was occasionally seen in infected cells as well. Such cells contained many naked HV particles, randomly distributed throughout the nucleus. Although these studies represent an interesting start toward characterization of the virus cycle and virus-determined antigens, the immunological behavior of tumor tissue is less well known. It is of great interest, particularly in view of the fact that infectious virus cannot be extracted from the tumor cells, as a rule, but is abundantly present in an entirely unrelated tissue, the feather follicle epithelium (Calnek et al., 1970a). Mareks disease virus-associated antigens, detected by the fluorescent antibody test, are very infrequent in tumor cells. They are readily found in a number of normal epithelial tissues, as will be discussed in more detail below. It is, therefore, likely that MD virus can enter into a lytic and a nonlytic interaction with different cells in the same animal. Only the nonlytic interaction would lead to tumor formation. It would
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
227
be of great interest to compare infected and transformed cells from the antigenic point of view. A fluorescent antibody test, detecting antigens specific for MD was described by Kottaridis and Lugeinbuhl ( 1968). Antisera produced in rabbits against blood from infected birds gave a specific reaction with infected cell cultures and bone marrow smears. It did not stain control samples, nor did antiserum to cells from noninfected birds stain infected cells. Since antisera directed against the complement-fixing avian leukosis (COFAL) antigen, the group-specific component of the avian leukosis virus complex (Huebner et al., 1964), failed to show a reaction with infected cell cultures and bone marrow smears, the authors concluded that the etiological agent of MD differed from the viruses of the leukosis group. A precipitating antigen could be induced in chicken kidney cell cultures, infected with the herpes-type virus associated with MD (Chubb and Churchill, 1968). Sera from birds infected with MD gave a positive immunoprecipitation reaction. Five different MD isolates produced identical antigens. There was no antigenic relationship between the herpes-type virus associated with MD and two other avian herpestype viruses. Antigen production in culture proceeded in parallel with the CPE associated with the herpes-type virus found in cell cultures inoculated with MD tumor cells (Churchill and Biggs, 1967). Both antigen production and CPE were prevented by IUDR. No reactive antigen could be demonstrated in tumor homogenates. This was related to the fact that electron microscopy failed to reveal any virus particles ( Nazerian et al., 1968). In MD-infected flocks, many birds had precipitating antibodies, despite a low disease incidence. This shows the occurrence of subclinical infections. The highly contagious nature of MD is in strange contrast to the absence of virus particles from the tumor tissue, and the strictly cellassociated nature of the virus in derived cultures. This has led to a systematic search for viral antigens in the tissues of infected birds (Calnek and Hitchner, 1969). Immunofluorescent antigens were predominantly found in the medullary cells of the bursa Fabricius and in a variety of epithelial cells, especially the kidney tubules and, as already mentioned, the follicular epithelium of the feathers. Lymphoid cells of the neoplastic lesions were only infrequently stained. Antigen-positive cells were most frequent in the genetically most susceptible strain, in young birds, and after parenteral inoculation. The authors concluded that MD viral infection is cytocidal in the bursa Fabricius; productive in a number of epithelial tissues that showed fluorescence but no histological changes;
228
GEORGE KLEIN
and oncogenic, nonproductive. A similar range of effects has been described for other, oncogenic DNA viruses, such as polyoma, SV 40, and the oncogenic adenoviruses. Recently, Calnek et al. (1970b) examined the same infected chicken tissues for viral antigens, detected by IF and gel precipitation, for virus particles by electron microscopy, and for infectivity in vitro. All samples of the feather follicle epithelium were antigen positive by both methods, contained large numbers of both naked and enveloped virus particles and yielded infectious, cell-free virus. Other tissues contained IF antigens less consistently and extensively than the feather follicle epithelium, and nearly all virus particles were naked and intranuclear. Infectivity was cell-associated. The correlation among IF, gel precipitation, and electron-microscopicresults was very good. The fluorescent antibody test was most sensitive, since single infected cells could be detected. The apparent relationship between infectivity and the oceurrence of enveloped particles suggested that the envelope protected the virion in the extracellular environment and/ or helped it to infect susceptible cells. The cell-associated infectivity of tumor cells and of cell suspensions derived from other tissues than the feather follicle epithelium was fundamentally different from the effect of cell-free virus obtained from the feather follicles. The distribution of enveloped virus within the feather follicle itself was interesting. Although the germinal layers of the epithelium were infected, the appearance of the virus was similar to that in most other epithelial tissues-predominantly intranuclear and incomplete. Virus envelopment was mainly restricted to the keratinizing layers of the epithelium. This is strongly reminiscent of the Shope rabbit papilloma system (Noyes, 1959; Stone et al., 1959). Closely similar results were obtained by Purchase ( 1970), concerning the distribution of IF and precipitin antigens and recovery of infectious virus. Histopathological examination showed a close association between the antigens and cell degeneration and necrosis. It was, therefore, suggested that antigen production was associated with a cytolytic process. Antigens were found only in cells or birds infected with MD virus, and not in cells or birds infected with other agents, including lymphoid leukosis or infectious bursa1 agent. Infected tissue extracts containing two different MD virus strains gave a line of identity in precipitin tests, showing at least one antigen in common. Antibodies related to MD were also demonstrated by an indirect tannic acid hemagglutination test ( Eidson and Schmittle, 1969). Hemagglutination was obtained when tanned horse erythrocytes, treated with an antigen prepared from MD-infected duck embryo fibroblasts, were
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
229
tested against the sera of MD-infected birds. There was no hemagglutination with normal chicken serum in the presence of the MD tissue culture antigen nor with normal tissue culture antigen in the presence of a known positive serum. In MD injected birds, antibody levels showed a certain relationship to the course of the disease. Birds dying 9 weeks after inoculation had relatively low antibody titers (1:8-1:32). In room contact birds, not directly inoculated with the isolate, antibodies could be detected within 4 weeks and their titer rose from 32-64 at 9 weeks to 256512 by 12 weeks. Two room contact birds that died during the fifth and sixth weeks were negative. Three survivors that developed high antibody titers by 13 weeks (256512) showed no clinical signs of MD. When their kidneys were cultured, characteristic CPE was observed after 5 days, however. Two birds followed in parallel after intra-abdominal inoculation of a tumor suspension at 1 day of age, showed different disease courses and antibody levels. One showed no clinical signs between 6 and 12 weeks. Its antibody titer rose from 32 to 1024. The other bird that developed unilateral paralysis at 6 weeks, had an antibody titer of 6. It had ovarian and kidney tumors and nervous system involvement. The authors suggested that the antibody titer was directly related to the chicken’s ability to survive the disease. The availability of MD infected tissue cultures has stimulated disease prevention experiments. Kottaridis and Lugeinbuhl ( 1969) showed that l-day-old chicks could be protected, to a certain extent, against later infection by the inoculation of the MD agent propagated in chicken embryo fibroblasts. On later exposure to experimentally infected chicks, the incidence of the disease was reduced to approximately onethird compared to controls, inoculated with uninfected fibroblasts. An attenuated, cell-associated virus strain, designated HPRS-16, was established in chicken kidney cell cultures by Churchill et al. (1969). Infectivity was abolished by cell disruption. The transfer of the virus between monolayer cells was not inhibited by antiserum. When tested for pathogenicity in day-old chicks, neoplastic growth and f or histological lesions were regularly denionstrable until the twentieth passage of the virus, but pathogenicity was lost by the thirty-third passage and failed to return until the sixty-third passage, the highest number tested. The attenuated virus could protect against the development of MD upon subsequent challenge with a virulent strain of virus. Immunoprecipitation lines, obtained by confronting infected cell antigens or supernatant medium with the serum of infected chicks, showed a gradual weakening, in parallel with the decrease in patho-
230
GEORGE KLEIN
genicity. One line that was particularly strong in the original, highly infectious material (designated antigen A ) was completely lost by the thirtieth passage. Two other antigens ( B and C ) remained associated with the infected cell extracts during the entire series of passages. Four hypotheses were considered in trying to explain these findings: ( a) introduction of another cell-associated herpes-type virus from the outside, ( b ) heterogeneity of the original isolate, ( c ) an unidentified infectious agent, distinct from the herpes-type virus and responsible for MD, for antigen A, and for pathogenicity as well, and ( d ) mutation of the virus during in uitro passage. For a number of reasons, virus mutation during passage in culture has been considered most likely. This mutant would have been favored by its increased rate of spread within the cell culture, until it outgrew the parent virus. Okazaki et al. (1970) isolated a herpes virus from turkeys (HVT) that could protect chickens against virulent MD herpes virus. The turkey isolate was nonpathogenic and noncontagious. Vaccinated birds were completely protected against challenge with MD virus, regardless whether the HVT was propagated in duck embryo fibroblasts, chicken cells, or chicken embryo fibroblasts. The HVT did not spread from the vaccinated birds to contact birds, as judged by the absence of antibody formation. The lack of spread was attributed to low concentration or absence of virus in the feather follicle epithelium. The HVT showed a close serological relationship with the MDassociated herpes virus in chicks, but anti-HVT antibodies gave different staining patterns. All HVT-inoculated birds had circulating antibody, demonstrated by fluorescent antibody tests. When challenged with MD virus they developed antibody to the latter, in parallel with maintaining their antibody level against the vaccine HVT. Thus, the mechanism of their immunity, whether cellular or humoral, could not prevent multiplication of the virulent virus. This protection against tumor development is reminiscent of the “immunological reinforcement” demonstrated in other experimental tumor systems (for references, see G . Klein, 1969) where the second administration of an oncogenic virus or of irradiated tumor cells induced by the same virus, during the second part of the oncogenic latent period, reduced the incidence of tumors in neonatally inoculated animals. Immunological reinforcement is believed to act at the level of cell-mediated immunity, by increasing the recruitment of sensitized host cells, capable of checking the growth of virally induced, antigenically altered tumor cells before they can reach an irreversible colony size ( Deichmann, 1969). Recently, the epidemiology of Marek‘s disease herpes virus (MDHV) was studied in broiler flocks (Witter et at., 1970). Chicks derived from
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
231
infected flocks were free of infection at hatching, indicating the absence of vertical transmission. Infection could be detected directly at 4 weeks of age, by the presence of microscopic lesions and the isolation of MDHV. Acquired antibody appeared at 6 to 7 weeks. By 8 weeks, 9&9% of the birds carried isolatable MDHV and 78%had microscopic lesions. As it will have appeared from this brief survey of MD, there are some remarkable parallels between the antigenic behavior of MD virus, MD tumor tissue and derived cultures, on the one hand, and the corresponding EBV-associated, BL derived cell lines, on the other. This is further emphasized by the findings of Ahmed et d. (1970). Rarely, if ever, could MDHV be observed in tumor tissue by electron microscopy. Immunofluorescence tests on tumor tissue imprints prior to cultivation also failed to reveal MDHV antigen, although it could be readily demonstrated after cultivation. This is reminiscent of the behavior of EBV where viral antigens could not be demonstrated by IF in the biopsy material, as a rule, whereas derived cell cultures contained viral antigens and virus particles as well (Nadkarni et al., 1970). B. LUCKBAGENT Herpes-type virus had been identified in relationship to Luck6 renal adenocarcinoma of frogs (Fawcet, 1956; Mizell et aZ., 1968; Tweedell, 1967). The presence of detectable virus particles was related to the climate, as they appeared only in cold temperature. Tumors from prehibernating leopard frogs appeared to be virus-free. Virus particles can be regularly detected in all tumors of frogs emerging from hibernation ( McKjnnell, 1969). Preliminary serological studies on the Luck6 agent have been published recently ( Kirkwood et al., 1969). Rabbits were immunized with Luck6-associated HTV, consisting primarily of nonenveloped capsids. Antisera were absorbed with normal frog kidney extracts and gave a strong precipitation band with the Luck6 HTV. The band could be resolved into three distinct lines. Unrelated rabbit immune sera, including an antiserum to certain frog viruses, did not precipitate the Luck6 HTV, nor did the absorbed rabbit anti-Luck6 HTV sera react with extracts of normal frog tissues. Rabbit and frog immune sera gave reactions of identity against the Luck6 agent. Surprisingly, EBV preparations were also precipitated by rabbit anti-Luck6 HTV sera. Five unrelated hyperimmune rabbit sera did not react with EBV concentrates. Absorption of the rabbit anti-Luck6 HTV serum with EBV-carrying cell lines eliminated precipitating activity against both Luck6 HTV and EBV. It was concluded that EBV and the Luck6 virus share common antigens, presumably capsid components.
232
GEORGE KLEIN
Since the rabbit anti-Luck6 HTV serum also precipitated capsid preparations of HSV and of human and simian CMV, it is possible that herpes virus group antigens were responsible for this reaction. Fink et al. (1968) have also demonstrated the presence of an identical antigenic component in an EBV-carrying, BL derived cell line and the Luck6 system by the use of human, EBV-reactive sera. Vi. implications
Four main dilemmas arise from the pattern of findings on BL; they are interrelated but all have their specific aspects. They can be briefly stated as follows. a. The etiological dilemma. Can the occurrence of distinctive, tumorassociated antigens give any clues about the etiology of the disease? b. The problem of neoplastic behavior. Are the changes in the composition of the cell membrane, or other cellular organelles, as reflected by the appearance of new antigenic specificities, fundamentally involved in the neoplastic behavior of the cell? In other words, does the unresponsiveness of the cell to growth control depend on the change in composition or structure that is revealed by the immunological tests? c. The therapeutic problem. Can any of the immunological reactions now identified serve to measure the patient’s reactivity to his own tumor, in connection with various therapeutic procedures, including attempts at immunotherapy? d. Are there any approaches to prevention in sight? Concerning the etiological dilemma, it is a useful point of departure to state that all virally induced experimental tumors share the same antigen, as long as they are induced by the same virus, at least so far as the transplantation-type, membrane-associated antigens are concerned. The reverse, the assumption of a common viral etiology on the basis of common antigens found in tumors of unknown origin is not necessarily justified, however. It has been shown (MathC, 1967; Pasternak, 1965; Sjogren and Hellstrom, 1965; Stuck et al., 1964; Svet-Moldavsky et al., 1967) that virally induced new antigens can be made to appear by superinfecting normal cells or tumors of unrelated etiology with oncogenic and even with some nononcogenic viruses. The only differences between this secondary antigenic conversion and the primary event that occurs in direct relation to tumor induction are the lesser stability of the former, particularly in immune hosts (Sjogren, 1965) and a more irregular association between antigen and tumor, depending on the accidental nature of superinfection. As discussed above, high anti-EBV titers and high antibody levels against EBV associated membrane antigens and soluble antigens are
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
233
regularly associated with at least two neoplastic diseases: BL and NPC. For NPC, it is clear that this serological pattern is independent of geographical or ethnic origin. A similar situation may exist for BL, but the lack of reliable criteria by which the identity of the disease can be established outside the endemic areas and distinguished from ordinary lymphomas makes a similar evaluation of the non-African cases more difficult. It is important to stress that thc main difference between BL and NPC and other normal or neoplastic serum donor categories investigated is not EBV positivity nor the occurrence of high-titered reactions in occasional donors, since such donors may be found in most other categories as well, but the regular and consistent association of high-titered reactions according to a number of different, independent tests, measuring a variety of EBV associated antigens. Looking at it from this angle, BL and NPC are unique. One may question, however, whether this angle can be justified or, more specifically, what it implies. A starting point is the convincing demonstration that EBV is causally related to at least one form of infectious mononucleosis (G. Henle et al., 1968; Niederman et al., 1970). This form afflicts EBV seronegative adolescents as a rule, is frequently positive for heterophile antibodies, and is regularly accompanied by seroconversion to anti-EBV positivity. A prospective study (Niederman et al., 1970) suggests that seropositive individuals are protected from the disease. The serological screening of many different human populations also showed (G. Henle et al., 1969; W. Henle and Henle, 1970) that there is another, “early” seroconversion to anti-EBV positivity, culminating around 4 years of age and particularly frequent in low socioeconomic groups. This early infection does not lead to infectious mononucleosis or any other disease identity so far recognized. Viewed against this background, the relationship of EBV infection to BL and NPC may be considered in terms of the following alternatives. 1. The virus that causes infectious mononucleosis is also responsible for these two tumors; if this is true, intrinsic or extrinsic cofactors have to be postulated to explain the malignant conversion (the cofactor hypothesis). 2. Different virus subtypes are responsible for the different clinical entities (the multiple virus hypothesis). 3. The virus is a relatively harmless inhabitant of lymphoid tissues, although it may cause temporary proliferation (mononucleosis) under certain conditions. When lymphoid tissues proliferate for other reasons, e.g., in malignancies due to unrelated causes, the virus travels along as a passenger, with increased antigen production and high-titered anti-
234
GEORGE KLEIN
body formation as a result. This passenger liypothesis is the logical analog of the antigenic conversion of established tumors by etiologically unrelated viruses. In view of the high regularity of association, a requirement for a particular trophic relationship between EBV and the target (lymphoid) tissue may be added in the present case. The passenger hypothesis cannot be excluded at present, but it appears somewhat unlikely in view of the fact that lymphoproliferative diseases other than BL and anaplastic carcinomas other than NPC do not show a regular high-titered EBV association. This includes lymphoproliferative malignancies occurring in the same or closely adjacent anatomical areas, such as reticulum cell sarcoma and lymphosarcoma, and carcinomas that arise in or close to the tissues of the Waldeyer ring, such as the hypopharynx, oropharynx, the tonsil, base of the tongue, and soft palate. Carcinoma of the maxilla is a possible exception, but larger groups remain to be investigated. Hodgkin’s disease represents a very interesting case in itself. Recently, it has been found, that the sarcomatous form, i.e., the lymphocyte-poor type with the worst prognosis, shows a high anti-VCA and anti-MA reactivity, quite comparable to BL and NPC, whereas the lymphocyte-rich and relatively more benign paragranulomatous form is low-reactive in both tests and thus resembles the control material (Johansson et al., 1970). The granulomatous form is intermediate, both with regard to histological type and serological reactivity. This means, so far as the serology is concerned, that it represents a mixture of high- and low-reactive cases. Thus, some special role played by EBV in the etiology of Hodgkin’s sarconia cannot be excluded. In the present context, the inverse correlation with lymphocytic predominance would not be in line with a simple passenger hypothesis. None of this reasoning excludes the passenger hypothesis conclusively, of course, but quite a number of ad hoc assumptions have to be made to maintain it in face of all evidence. One would have to postulate some specific trophic relationship between the virus and the kind of lymphocyte that gives rise to BL and is particularly abundant in NPC. This would not apply to the lymphoid cells that proliferate in the various other malignancies, used as controls. No valid objection can be raised against such a hypothesis, but it appears rather far fetched, in view of the fact that EBV carrying blastoid cell lines can be isolated regularly from EBV-positive individuals, including donors with lymphoreticular malignancies of the “control” type, i.e., diseases that do not show a consistently high EBV positive serology. The sarcomatous form of Hodgkin’s disease is also very hard to explain. The ,possibility that EBV acts together with some cofactor in
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
235
causing neoplastic disease or, to phrase the same thesis differently, it acts by increasing the likelihood of neoplastic transformation brought about by other factors, has been recently proposed by Burkitt (1969) concerning the etiology of BL. In order to fit the geographic distribution of the disease with an ubiquitous virus, Burkitt proposed that an insecttransmitted cofactor is responsible for the malignant manifestation and specified it as chronic holoendemic malaria. This was based on the absence of BL from certain areas where malaria control has been enforced for some time and its presence in adjacent regions where malaria control was not regularly practiced. Interactions between viruses and other agents, capable of stimulating the proliferation of a target tissue may lead to malignant transformation in experimental systems where neither the virus nor the other agent is oncogenic per se (Southam et al., 1969). Since chronic malaria exerts a strong proliferative stimulus on the reticuloendothelial system, Burkitt’s modified theory is reasonable, although the same picture would result from the transmission of any etiological factor or cofactor mediated by the appropriate insect, and this includes other viruses. Some recent evidence concerning the frequency of the sickling trait in BL patients (Kafuko and Burkitt, 1970; Pike et al., 1970), indicates a possible role of malaria in the causation of the disease. Tumor patients had only about half of the incidence of AS-type hemoglobin, known to confer a relative immunity to malaria, in comparison with controls. This observation strengthens the postulated relationship between BL and malaria. It has also been found that infection of adult BALB/c mice with a murine plasmodium (P. berghei) increases the incidence of malignant lymphomas ( Wedderburn, 1970). The third possibility, the multiple virus hypothesis, implies the existence of closely related but biologically different EBVs with differences in their oncogenic power and their target tissue preference. In light of the information derived from experimental oncogenic viruses, this is a realistic alternative as well. As far as leukemia viruses of the RNA type are concerned, one recalls that prior to the discovery of the interference test for avian leukosis virus classification (Rubin, 1960), it was not possible to distinguish, morphologically or immunologically, between the viruses that were responsible for a number of quite different lympho- and myeloproliferative diseases or for fowl sarcoma. It is now known that the avian leukosis-sarcoma virus group has many closely related members; some induce solid tumors with highly distinctive properties, others are responsible for myeloid or erythromyeloid leukemia, or lymphomatosis, and still others cause no recognizable disease at all. A closely similar development can be noticed in the murine
236
GEORGE KLEIN
leukosis-sarcoma field. The Friend, Moloney, Rauscher, Gross, Kaplan, Rich, Graffi, Mazurenko, etc., agents are similar antigenically and indistinguishable by ultrastructure, but they induce distinct and characteristic clinical and pathological disease entities, specific for the viral agent (Rich, 1968). In the DNA field, a possibly relevant model is the series of HSV mutants, studied by Roizmann. Although this is not known to be an oncogenic system, it is important that different viral mutants induce different membrane changes in infected cells and, concomitantly, the cells are altered in their “social behavior” in ways that are characteristic for the virus mutant. Although a lytic virus obviously cannot transform its targets, the cellular changes are, nevertheless, concerned with intercellular relationship. Conceivably, other, nonlytic viruses of the same family might induce membrane changes compatible with cellular viability and reproductive integrity and a social behavior changed in the direction of disobedience to growth regulation, or, in other words, neoplasia. The oncogenic action of the agents responsible for MD, LuckB‘s carcinoma, and simian lymphoma, already discussed, show that at least some herpes-type viruses have a neoplastic potential. The immunological tests so far performed on EBV associated antigens, including those referred to in the previous sections, are not necessarily competent to reveal finer differences between closely related but biologically different agents with cross-reactive or overlapping antigenic components. Cross titration experiments, performed with infectious mononucleosis ( I M ) and BL sera or I M and BL-derived cell line smears, gave indistinguishable patterns ( W. Henle and Henle, 196813). A preliminary study of the membrane antigens on EBV positive lymphoblastoid cell lines derived from BL and NPC did not show any difference in reactivity patterns either (de Schryver et al., 1970), but this may simply reflect insufficient discriminating ability of the test. Further studies are needed to distinguish between these possibilities. In order to narrow down the passenger hypothesis, more extensive tests are desirable on tumor categories where occasional sera gave high EBV associated reactivity but only limited numbers of sera have been tested. It is also desirable to conduct nucleic acid hybridization studies on the corresponding tumors. More refined analytical methods are needed for the attempts to dissect different virus variants. The recent developments in the HSV field suggest that biochemical studies of viral envelopes and altered cell membranes may be particularly rewarding, As far as the seroepidemiology of EBV infections is concerned, it has to be emphasized that there were no significant differences between low and high BL endemic areas with regard to the distribution of antiEBV titers in relation to age (G. Henle et al., 1969); a comparison of
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
237
titers in normal populations is, therefore, unlikely to elucidate the possible role of this virus in BL. A prospective seroepidemiological study may be more rewarding. It may be relevant in this connection that BL is essentially a childhood disease, with a peak incidence between 4 and 7 years. This fact, together with the clinical and serological evidence indicating a relatively high antigenicity in the autochthonous host would speak for a short latent period during the oncogenic process. In experimental systems, highly antigenic tumors arise with brief latency, as a rule; or, in other words, highly antigenic tumors cannot escape rejection unless they grow rapidly after inception (G. Klein, 1969; Old and Boyse, 1964; Prehn, 1963). If this reasoning is essentially correct and the latent period of BL is relatively short, a prospective seroepidemiological study may be decisive. In the populations at risk within the high-endemic areas, there is a relatively small minority of EBV negative children as in other populations, and another, even smaller minority with high anti-EBV titers. The majority consists of relatively low-titered positives (G. Henle et al., 1969). It may be asked whether BL develops preferentially in one of the two minority groups or arises at random and irrespectively of antiEBV titer. Provided that a sufficiently large number of sera could be collected, stored, and appropriately identified, tests may become feasible within a few years’ time on predisease sera from individuals who subsequently develop BL. It might be objected that the sensitivity threshold of the anti-EBV test may be too high ( 1: l o ) , and a number of anti-EBVpositive sera may be classified as false negatives. Since relatively concentrated sera cannot be tested safely due to the nonspecific artifacts that tend to appear, this is probably true. It is also clear, however, that at least a substantial part of the anti-EBV-negative donors, as defined by the 1:10 threshold, must be negative in the biological sense, since the prospective study on infectious mononucleosis has clearly shown (Niederman et al., 1970) that the disease develops exclusively in this group, and seropositive individuals are protected. Concerning the relationship between EBV and NPC, the same types of hypotheses can be considered as for BL. The multiple virus hypothesis would imply an NPC-specific EBV variant. The cofactor hypothesis would lead to a consideration of both genetic and environmental factors, in light of the information on the incidence of the disease in migrant high-risk populations ( Muir and Shanmugaratnam, 1967). A prospective serological study of this question would be very difficult at the present time, since NPC, unlike BL, occurs over an extremely wide age range. The possible significance of the cell membrane changes reflected by
238
GEORGE KLEIN
the appearance of new antigens for the understanding of neoplastic cell behavior cannot be assessed as yet, but it may be pertinent to point out that cell membrane changes are among the most seriously considered parameters of neoplastic behavior at present. They are almost invariably found when comparable normal and transformed cells are studied in parallel. They may concern changes in behavior, such as contact inhibition ( Abercrombie, 1967), or altered expression of phytoagglutinin receptors (Burger, 1968; Inbar and Sachs, 1969) that may reflect a change in the synthesis of certain glycolipids (Hakamori and Murakami, 1968) and are perhaps linked to the appearance of new surface antigens (Mora et al., 1969). Membrane antigen changes have been demonstrated in all experimental tumors thoroughly studied ( Deichmann, 1969; K. E. Hellstrom and Hellstrom, 1969; G. Klein, 1966, 1969; Old and Boyse, 1964; Pasternak, 1969; Sjogren, 1965),and, although the details concerning antigenic strength and patterns of cross-reactivity vary from system to system, it is clear that they must reflect some remodelling of the membrane structure. Growth-regulating mechanisms, including both long-range, humoral and short-range, contactual signals must transmit their message to the target cell via receptors on the outer membrane. Nonlytic virus-cell interactions may result in the incorporation of virally determined (or virally derepressed) components into the membrane and render the appropriate receptors insensitive to regulation. If this is compatible with continued cell growth and division, it may trigger neoplastic development. Since infection with potentially oncogenic viruses and the concomitant surface antigenic changes are not limited to the oncogenic target tissue but can occur as well in cells that remain normal (i.e., subject to regulation), a tissue- or cell-type specificity must be added to explain transformation. Since different tissues must obey different types of growth regulation, this is not surprising. Virally determined surface antigens may be retained, while in vivo tumorigenic properties decrease or are lost from hybrids between malignant and normal cells ( G . Klein and Harris, 1971; G. Klein et al., 1971a; Wiener et al., 1971) and from “revertant” variants that can be selected from in vitro transformed cultures (Macpherson, 1970; Pollack et al., 1968; Rabinowitz and Sachs, 1970). Further studies on such systems will be interesting, not only to understanding neoplastic behavior and the possible role of MA changes, but also to elucidate normal growth responsiveness at the cell level. Meanwhile, the question whether EBV-associated MAS are essential for the neoplastic behavior of BL and NPC cells is not clear. So far as NPC is concerned, such antigens have been demonstrated on derived lymphoblastoid cell lines (de Schryver et al., 1970), but it is not known
IMMUNOLOGICAL ASPECXS OF BURKITT’S LYMPHOMA
239
whether they are present on the surface of the carcinoma cells. Established culture lines of BL cells carry EBV, as a rule, although at very different levels (G. Klein et al., 1968a; Nadkarni et al., 1970). Membrane antigen can only be demonstrated in lines with a relatively high EBV load (G. Klein et al., 1968a) and is subject to environmental fluctuations (Yata and Klein, 1969). There is at least one BL line (Raji) which contains no EBV antigen demonstrable by IF or virus particles (Epstein et al., 1966), although it carries DNA that hybridizes specifically with EBV-DNA, as already mentioned ( zur Hausen and Schulte-Holthausen, 1970), and an EBV associated soluble CF antigen as well. Since the Raji line can be superinfected with EBV (W. Henle et al., 1970b), the absence of virus production is presumably not due to repressors. If it carries genetic information derived from EBV, it is probably a defective viral genome, lacking the cistrons that specify the membrane, capsid, and early protein antigens. If there were any assurance that the Raji line represents a neoplastic cell, this would imply that MA is not required for neoplastic behavior. Since this question cannot be tested directly with a human cell, however, a conclusive answer is not available. Further studies on the presence of viral DNA and virus-specific mRNA in BLderived lines, in comparison with EBV-carrying blastoid cell lines of other origin may prove very informative. In this connection, it is interesting that infectious mononucleosis derived lines are reportedly more prone to lose their EBV than are BL-derived lines (W. Henle and Henle, 1970). Thus, whereas EBV is clearly helpful in inducing lymphoblastoid transformation and facilitates the establishment of stationary suspension cultures (Gerber et al., 1969; W. Henle and Henle, 1970; W. Henle et at., 1967; Pope et al., 1969a), it is clear that the expression of EBV determined antigens, as assessed by IF, is not required for their continued growth in uitro. The presence of the EBV genome in a cell line that lacks EBV particles and EBV antigens as detected by IF raises the question of the status of virally derived genetic information in these cells. Conceivably, a defective or a complete genome may be present, presumably integrated with host-cell DNA. This is also supported by the fact that the cloning of EBV carrier cultures, containing only a small proportion of EBV positive cells, in the presence of antiviral serum, leads regularly to isolation of EBV carrier clones. For example, Hinuma and Grace (1968) found that all of forty-seven clonally derived cell lines, established in semisolid agar from the EBV carrying P3 (Jijoye) line were EBV carriers. Zajac and Kohn (1970) established twenty-three sublines from the EB2 line, at a cloning efficiency of more than 40%. In spite of the fact that the parent culture only contained less than 1%EBV( VCA)-positive
240
GEORGE KLEIN
cells, all clonal sublines were VCA-positive, in 0.1 to 1%of the cells. Cross-infection between the clones was unlikely, because the cultures were maintained in the presence of an anti-EBV positive serum for several days before cloning and were cloned under antiviral antibody. Single cell clones, derived from two human leukocytic lines that carry EBV antigen were examined for the presence of EBV antigens by Miller et a2. (1970). One line had been derived from the peripheral blood of a leukemia patient, whereas the other was the BL-derived EB3 line. Cells were cloned in microdrops, under visual control. All twentyfive EB3 clones tested carried viral antigen, and the same was true for the ten clones derived from the leukemia patient. It was concluded that the EB3 genome is associated with many more cells, in carrier cultures, than those that demonstrate the EBV antigen. This was also supported by the finding that carrier cultures were not cured of EBV infection when maintained in 10%EBV antiserum for 3 months. It was considered likely that transmission of virus takes place by the passage of viral genetic material from parent cells to progeny. Further interesting findings were made by Maurer et al. (1970). They cloned several BL derived cell lines by two procedures-an agar method and a single-cell isolation procedure. Of the clones established from parental cell lines that contained a relatively high frequency of EBV-positive cells, 92100%were also EBV-positive, whereas only 7 3 3 % of the clones established from three lines with a very low incidence of EBV-positive cells ( less than 1%)were virus-positive. Recloning indicated that negative clones could give rise to virus-positive secondary clones. The presence of anti-EBV-positive immunoglobulin did not affect the incidence of EBV-positive clones. The authors suggested that all BL derived cells carried an inherited potential for EBV production and that the rate at which this potential was induced determined the viral positiveness of a particular cell line. This conclusion was also supported by the behavior of EBV-carrier lines transferred to arginine-deficient media and by the appearance of EBV antigens in explanted biopsy cells. W. Henle and Henle (1968a) showed that the transfer of established, EBV positive BL lines to arginine-free medium led to a five- to ten-fold increase of VCA-positive cells within 2 to 5 days. They postulated that the lack of arginine removed an intracellular block of viral replication, permitting complete expression of the viral genome, as long as enough arginine remained available for the synthesis of capsid proteins. The appearance of VCA-positive cells after the explantation of BL biopsies has been studied by Nadkarni et al. (1970). Definite VCA
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
241
positive cells were only seen in 6%of the biopsies and amounted to no more than a few cells among several hundred thousands in each positive biopsy. After explantation, VCA-positive cells appeared within 3 to 10 days, occasionally in substantial numbers (up to 5 % ) .It was particularly interesting that repeated biopsies from the same donor gave remarkably similar results. In time, all cultures which developed into lines showed the presence of EBV. Arginine-free medium failed to facilitate the emergence of EBV-antigen-producing cells in the young cultures, or to increase their number, unless 0.5%or more VCA positive cells were present in the parallel culture, propagated on full medium. The good reproducibility of the “EBV antigen load” in parallel lines derived from repeated biopsies of the same patient, often taken several weeks or months apart, suggested some kind of an equilibrium between virus and tumor cells, characteristic for the tumor (or, often, multiple tumors) of a given patient. This was further exemplified by recent studies showing the presence of DNA in the biopsies of BL and NPC tumor patients that could specifically hybridize with purified EBV-DNA. Zur Hausm and co-workers (1970) found that all thirteen BL and ten NPC biopsics studied contained DNA that hybridized specifically with purified EBV-DNA, whereas other kinds of tumors, arising in EBV seropositivc patients, contained no detectable hybridizable DNA. In thirteen Burkitt biopsies, the approximate number of EBV genome equivalents per cell varied between 2 and 26. In ten NPC biopsies, the number of EBV genome equivalents varied betwcen 1 and 19. Interestingly, 3 BL patients from whom double biopsies were taken showed closely similar genome equivalent values in their geometrically distant tumors (2-2, 7-8, and 21-26, respectively). Since BL is a uniclonal disease (Fialkow et al., 1970), this would mean that the EBV load is fairly constant and characteristic for a given clone. This is reminiscent of the different and characteristic numbers of SV 40 genome copies in different clones of SV 40-transformed cells ( Westphal and Dulbecco, 1968). Although the nucleic acid hybridization studies merely confirm that EBV is more closely associated with BL and NPC than with a number of other tumors, this evidence is at lcast consistent with the behavior of known oncogenic DNA virus systems. Turning now to the therapeutic problems, it seems clearly established that thc host immune response plays an important part in BL. This is indicated by the documented occurrence of spontaneous regression (Burkitt and Kyalwazi, 1967), by the substantial fraction of long-term survivors, sometimes after only mild chemotherapy ( Burkitt, 1967a,b;
242
GEORGE KLEIN
Clifford, 1966; Ngu, 1965), by the reactivity of the autochthonous host against its own tumor cells, indicated by the presence of humoral antibodies reacting with the surface of viable cells (G. Klein et al., 1966, 1967b), by the positive Cla fixation test (Nishioka et al., 1968), and by the transformation of host lymphocytes when confronted with mitomycintreated autochthonous lymphoma cells in the mixed lymphocyte-target cell interaction test (Stjernsward et al., 1968). In addition, the progressive accumulation of an IgG coating on the cell surface of tumors that persist in spite of therapy ( E . Klein, et al., 1968; G. Klein et al., 1969a) together with the tetraploid ( immunoresistant?) constitution of tumors that have recurred after long-term regression (Clifford et al., 1968) suggest that the dynamics of immunoselection may apply to this human system, in analogy with experimental tumors (Hauschka et al., 1956). Immunoresistance may be as important as drug resistance, if not more so, in frustrating therapy. The host response to an autochthonous tumor is no less complex than other immune responses against viable cells. Different effector components interact in ways so that rejection or its opposite, enhancement, dominate the eventual outcome. Humoral antibodies are cytotoxic in some situations, whereas in others they lack demonstrable growth inhibitory effects but, nevertheless, manage to attach and thereby prevent the access of host lymphoid cells (G. Klein, 1966a; Old and Boyse, 1964). Recent evidence indicates that such blocking antibodies may play an important role in counteracting the cell-mediated host response in experimental (I. Hellstrom and Hellstrom, 1970) as well as human (I. Hellstrom et al., 1969) tumors. The main therapeutic dilemma is what the proper stimuli are, specific or nonspecific, and how they are best administered to the immune system in order to achieve the objective, rejection, and avoid its opposite, enhancement. The rationale of introducing immune stimuli at a time when the tumor load confronting the host is minimal, i.e., after regression has been induced by chemotherapy, is obvious (Math&, 1969; Skipper, 1967), but the optimal form of stimulus and the best mode and timing of its administration are not. No a priori guidance can be given from experimental studies, because the same mode of administration, dosage, vehicle, etc., of the same preparation may favor rejection in one system and enhancement in another, and there is an immense variation depending on host species, tumor type, and individual characteristics of the tumor line. Ideally, it would be desirable to develop methods that allow the quantitative assessment of cell-bound immunity, the synergistic or antagonistic action of humoral antibodies in relation to it in each untreated patient, and to follow it subsequently during treatment. Although
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
243
this should be feasible, at least in principle, its practical application is still in the future. Meanwhile, an empirical approach, based on as much rational reasoning as the experimental models allow, may yield important information, as the work of Math6 and his group clearly indicates ( Math6, 1969). Obviously, the prevention approach will have to await further clarification of the relationship between seroconversion and tumor development, preferably from a prospective study. A discussion of this beyond the general statement that the ultimate goal of an immunological approach must be prevention rather than therapy appears premature at the present time.
REFERENCES Abercrombie, M. (1967). Nut. Cancer Inst., Monogr. 26, 249-277. Ahlstrom, C. G., Anderson, T., Klein, G., and Akerman, M. (1967). Int. J. Cancer 2, 583-585. Ahmed, M., Jensen, K. E., Slattery, S. M., Leech, J. B., and Schidlovsky, G. (1970). Avian Dis. 14, 349-363. Allison, A. C., and Law, L. W. (1968). Proc. SOC. E x p . Biol. Med. 127, 207-212. Amos, D. B., Kohen, I., and Klein, W. J. (1970). Transplant. Proc. 2, 68-75. Armstrong, D., Henle, G., and Henle, W. (1966). J. Bacteriol. 1, 1257-1262. Beltran, G., Northington, J. W., Leiderman, E., Mogabgab, W. J., and Stuckey, W. J. (1971). Int. J. Cancer (in press). Benyesh-Melnick, M., Fernbach, D. J., Dessy, J. I., and Lewis, R. T. (1968). 1. Nut. Cancer Inst. 40, 111-122. Blacklow, N. R., and Kapikian, A. Z. (1970). Nature (London) 226, 647-648. Bremberg, S., Klein, G., and Epstein, M. A. (1969). Int. J. Cancer 4, 761-766. Burger, M. M. (1968). Nature (London) 219, 499-500. Burkitt, D. (1963). Int. Reo. E x p . Puthol. 2, 67-138. Burkitt, D. (1967a). In “Treatment of Burkitt’s Tumours” J. H. Burchenal and D. P. Burkitt, eds.), UICC Monogr. Ser., Vol. 8, pp. 2-6. Springer-Verlag, Berlin and New York. Burkitt, D. (1967b). In “Treatment of Burkitt’s Tumours” (J. H. Burchenal, ed.), UICC Monogr. Ser. Vol. 8, pp. 94-101. Springer-Verlag, Berlin and New York. Burkitt, D. (1969). J. Nut. Cancer Inst. 42, 19-28. Burkitt, D., and Kyalwazi, S. K. (1967). Brit. 1. Cancer 21, 14-16. Calnek, B. W., and Hitchner, S. B. (1969). J. Nut. Cancer Inst. 43, 935-949. Calnek, B. W., Adldinger, H. K., and Khan, D. E. (1970a). Avian Dis. 14, 219-233. Calnek, B. W., Ubertini, T., and Adldinger, H. K. (1970b). J. Nut. Cancer Inst. 45, 341-351. Chen, J. H., and Purchase, H. G. (1970). Virology 40, 410-412. Chubb, R. C., and Churchill, A. E. (1968). Vet. Res. 83, 4-7. Churchill, A. E. ( 1968). J. Nut. Cancer Inst. 41, 939-950. Churchill, A. E., and Biggs, P. M. (1967). Nature (London) 215, 528530. Churchill, A. E., Chubb, R. C., and Baxendale, W. (1969). J. Gen. ViroZ 4, 557564. Clifford, P. (1966). East Afr. Med. J . 43, 179-199.
244
GEORGE KLEIN
Clifford, P., Singh, S., Stjernsward, J., and Klein, G. (1967). Cancer Res. 27, 25782615. Clifford, P., Gripenberg, N., Klein, E., Fenyo, E. M., and Manolov, G. (1968). Lancet 2, 517518. Davies, D. A. L. (1966). Immunology 11, 115-125. Deichmam, G. I. (1969). Aduan. Cancer Res. 12, 101-137. de Schryver, A., Friberg, S., Klein, G., Henle, W., Henle, G., de Th6, G. Clifford, P., and Ho, H. C. (1969). Clin. E x p . Immunol. 5, 44-59. de Schryver, A., Klein, G., and de Thk, G. (1970). Clin. Exp. Immunol. 7, 161-171. Diehl, V., Henle, G., Henk, W., and Kohn, G. (1968). J. Virol. 2, 663-669. Dunkel, V. C., and Zeigel, R. F. (1970). J. Nut. Cancer Inst. 44, 133-144. Durr, F. E., Monroe, J. H., Schmitter, R., Traul, K. A., and Hirshaut, Y. (1970). Int. J . Cancer 6, 436-449. Eidson, S. C., and Schmittle, S. C. (1969). Auian Dis. 13, 774-782. Einhorn, N., Klein, G., and Clifford, P. (1970). Cancer 26, 1013-1022. Epstein, M. A,, and Achong, B. G. (1967). Cancer Res. 27, 248CL2493. Epstein, M. A,, and Achong, B. G. (1968). 1. Nut. Cancer Inst. 40, 593-607. Epstein, M. A., Achong, B. G., and Barr, Y. M. (1964). Lancet 1, 702-703. Epstein, M. A., Achong, B. G., Bur, Y. M., Zajac, B., Henle, G. and Henle, W. ( 1966). J. Nut. Cancer Inst. 37, 547-559. Evans, A. C., Niederman, J. C., and McCollum, R. W. (1968). N . Engl. J. Med. 279, 1121-1127. Fass, L., Herberman, R. B., and Ziegler, J. (1970a). N . Engl. J. Med. 282, 776-780. Fass, L.,Herberman, R. B., Ziegler, J., and Morrow, R. H. (1970b). J. Nut. Cancer Inst. 44, 145-149. Fawcet, D. W. (1956). J. Biophys. Biochem. Cytol. 2, 725-742. Fenyo, E. M., Klein, E., Klein, G., and Swiech, K. (1968). J. Nat. Cancer Inst. 40, 69-89. Fialkow, P. J., Klein, G., Gartler, S. M., and Clifford, P. (1970). Lancet 2, 384386. Fink, M. A., King, G. S., and Mizell, M. (1968). J. Nut. Cancer Inst. 41, 1477-1478. Fink, M. A., Isibul, L. R., and Waggoner, D. E. (1969). J . Nat. Cancer Inst. 42, 623-631. Gerber, P., and Birch, S. M. (1967). PTOC. Nut. Acad. Sci. 58, 478484. Gerber, P., and Deal, D. R. (1970). Proc. SOC. Exp. Biol. Med. 134, 748-751. Gerber, P., Whang-Peng, J., and Monroe, J. H. (1969). Proc. Nut. Acad. Sci. U. S . 63, 740-747. Gergely, L., Klein, G., and Ernberg, I. (1971a). Int. J. Cancer, 7, 293-302. Gergely, L., Klein, G., and Emberg, I. (1971b). Virology 45, 10-21. Gergely, L., Klein, G., and Ernberg, I. ( 1 9 7 1 ~ ) Virology . 45, 22-29. Goldstein, G., Klein, G., Pearson, G., and Clifford, P. (1969). Cancer Res. 29, 749-752. Gunvh, P., Klein, G., Henle, G., Henle, W., and Clifford, P. (1970). Nature (London) 228, 1053-1056. Hakamori, S. I., and Murakami, W. T. (1968). Proc. Nut. Acad. Sci. U. S. 59, 254-261. Hampar, B., Gerber, P., Hsu, K. C., Martos, L. M., Walker, J. L., Sigiienza, R. F., and Wells, G. A. (1970). J. Nut. Cancer Inst. 45, 75-85. Hauschka, T. S., Kvedar, B. J., Grinnel, S. T., and Amos, D. B. (1956). Ann. N . Y. Acad. Sci. 63, 683-705. Hellstrom, I., and Hellstrtim, K. E. (1970). Int. 1. Cancer 5, 195-201.
IMMUNOLOGICAL ASPECXS
OF BURKITT’S LYMPHOMA
245
Hellstrom, I., Hellstrom, K. E., Evans, C. A., Heppner, G. H., Pierce, E. G., and Yang, J. P. S. (1969). Proc. Nut. Acad. Sci. U. S . 62, 362368. Hellstrom, K. E., and Hellstrijm, I. (1969). Aduan. Cancer Res. 12, 167-224. Henle, G., and Henle, W. (1966a). J. Bacteriol. 91, 1248-1256. Henle, G., and Henle, W. (1966b). Trans. N . Y. Acad. Sci. [Z]29, 71-79. Henle, C., Henle, W., and Diehl, V. (1968). Proc. Nut. Acad. Sci. U. S . 59, 94-101. Henle, G., Henle, W., Clifford, P., Diehl, V., Kafuko, G. W., Kirya, B. G., Klein, G., Morrow, R. H., Munube, G. M. R., Pike, P., Tukei, P. M., and Ziegler, J. L. (1969). J. Nut. Cancer Inst. 43, 1147-1158. Henle, C., Henle, W., Hein, G., Gunvh, P., Clifford, P., Morrow, R. H., and Ziegler, J. L. (1971). J. Nut. Cancer Inst. 46, 861-871. Henle, W., and Henle, G. (1968a). J. ViroZ. 2, 18%191. Henle, W., and Henle, G. (1968a). J . Virol. 2, 182-191. Henle, W., and Henle, G. (196813). Perspec. Virol. 6, 105-117. Henle, W., and Henle, G. (1970). In “Comparative Leukemia Research” (R. M. Dutcher, ed.), pp. 706-713. Karger, Basel. Henle, W., Hummeler, K., and Henle, G. (1966). 1. Bacteriol. 92, 269-274. Henle, W., Diehl, V., Kohn, G., zur Hausen, H., and Henle, G. (1967). Science 157, 1064-1065. Henle, W., Henle, G., Burtin, P., Cachin, Y., Clifford, P., de Schryver, A., de Th6, G., Diehl, V., Ho, H. C., and Kieip, G. (k970a). J. Nut. Cancer Inst. 44, 225-231. Henle, W., Henle, G., Zajac, B. A., Pearson, G., Waubke, R., and Scriba, M. (1970b). Science 169, 188-190. Hinuma, Y., and Grace, J. ( 1967). Proc. SOC. Exp. BioE. Med. 124, 107-111. Hinuma, Y., and Grace, J. (1968). Cancer 22, 1089-1095. Hinuma, Y., Conn, M., Yamaguchi, J., Wudarski, D. J., Blakeslee, J. R., and Grace, J. T. (1967). J. ViroZ. 1, 1045-1051. Hinuma, Y., Sairenji, T., and Ohta-Hatano, R. (1971). Proc. Jap. Acad. 46, 989-993. Hirshaut, Y., Glade, P., Ctavo, L., Viera, B. D., Ainbender, E., Dvorak B., and Siltzbach, L. E. (1970). N . Engl. J. Med. 283, 502-506. Horosziewicz, J. S., Dunkel, V. C., Avila, L., and Grace, J. T. (1970). In “Comparative Leukemia Research” ( R . M. Dutcher, ed.), pp. 72%738. Karger, Basel. Huebner, R. J., Armstrong, D., Okuyan, M., Sarma, P. S., and Turner, H. C. (1964). Proc. Nut. Acad. Sci. U . S . 51, 74%750. Hughes, L. E., and Lytton, B. (1964). Brit. Med. J. 1, 209-212. Hunt, R. D., Melendez, L. V., King, N. W., Gilmore, C. E., Daniel, M. D., Williamson, M. E., and Jones, T. C. (1970). J. Nut. Cancer Inst. 44, 447466. Ikawata, S., and Grace, J. T. (1964). N . Y. State J . Med. 64, 2279-2282. Inbar, M., and Sachs, L. (1969). Nature (London) 223, 710-712. Irlin, I. S. ( 1967). ViroZogy 32, 725-728. Johansson, B., and Klein, E. (1970). Clin. Exp. ImmunoZ. 6, 421-428. Johansson, B., Klein, G., Henle, W., and Henle, G. (1970). lnt. J. Cancer 6, 450462. Kafuko, G. W., and Burkitt, D. P. (1970). Int. J. Cancer 6, 1-9. Kirkwood, J. M., Geering, G., Old, L. J., Mizell, M., and Wallace, J. (1969). Recent Results Cancer Res., Spec. Suppl., pp. 365-367. Klein, E., and Klein, C. (1964). J. Nut. Cancer Inst. 32, 547568. Klein, E., Klein, G., Nadkami, J. S., Nadkarni, J. J., Wigzell, H., and Clifford, P. (1968). Cancer Res. 28, 1300-1310.
246
GEORGE KLEIN
Klein, E., Eskeland, T., Inoue, M., Strom, R., and Johansson, B. (1970). Exp. Cell Res. 62, 133-148. Klein, G. (1966a). In “Viruses Inducing Cancer, Implications for Therapy” ( W . J. Burdette, ed.), pp. 323-349. Univ. of Utah Press, Salt Lake City, Utah. Klein, G. (I966b). Annu. Rev. Microbiol. 20, 223-252. Klein, G. ( 1967). In “Specificity of Cell Surfaces” (B. Davis and N. L. Warren, eds.), pp. 165-180. Prentice-Hall, Englewood Cliffs, New Jersey. Klein, G. ( 1968). Cancer Res. 28, 625-635. Klein, G. (1969). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 28, 1739-1753. Klein, G., and Harris, H. (1971). To be published. Klein, G., Sjogren, H. O., Klein, E., and Hellstriim, K. E. (1960). Cancer Res. 20, 1561-1572. Klein, G., Clifford, P., Klein, E., and Stjernsward, J. (1966). Proc. Nat. Acad. Sci. U. S. 55, 1628-1635. Klein, G., Clifford, P., Klein, E., Smith, R. T., Minowada, J., Kourilsky, F. M., and Burchenal, J. H. (1967a). J. Nat. Cancer Inst. 39, 1027-1044. Klein, G., Clifford, P., Klein, E., and Stjernsward, J. (196713). In “Treatment of Burkitt’s Tumour” (J. H. Burchenal and D. P. Burkitt, eds.), UICC Monogr. Ser., Vol. 8, pp. 209-232. Springer-Verlag, Berlin and New York. Res. 27, 2510-2520. Klein, G., Klein, E., and Clifford, P. ( 1 9 6 7 ~ ) Cancer . Klein, G., Pearson, G., Nadkarni, J. S., Nadkarni, J. J., Klein, E., Henle, G., Henle, W., and Clifford P. (1968a). J. E x p . Med. 128, 1011-1020. Klein, G., Pearson, G., Henle, G., Henle, W., Diehl, V., and Niederman, J. C. ( 1968b). J. Exp. Med. 128, 1021-1030. Klein, G . , Clifford, P., Henle, G., Henle, W., Old, L. J., and Geering, L. (1969a). Int. J. Cancer 5, 185-194. Klein, G., Pearson, G., Goldstein, G., Henle, W., Henle, G., and Clifford, P. (196913). J. Exp. Med. 129, 697-706. Klein, G., Geering, G., Old, L. J., Henle, G., Henle, W., and Clifford, P. (1970). Int. J. Cancer 5, 185-194. Klein, G., Bregula, U., Wiener, F., and Harris, H. (1971a). 1. Cell. Sci. 8, 659. Klein, G., Gergely, L., and Goldstein, G. (1971b). Clin. E x p . Immunol. 8, 593. Kleniola, E., von Essen, R., Henle, G., and Henle, W. (1970). I. Infec. Dis. 121, 608-614. Konn, M., Yohn, D. S., Hinuma, Y., Yamaguchi, J., and Grace, J. T. (1969). Cancer 23, 990-997. Kottaridis, S. D., and Lugeinbuhl, R. E. (1968). Atiian Dis. 12, 383-393. Kottaridis, S. D., and Lugeinbuhl, R. E. (1969). Nature (London) 221, 1258-1259. Law, L. W. (1966). Cancer Res. 26, 551571. Levin, A. G., Friberg, S., and Klein, E. (1969). Nature (London) 222, 997-998. Levine, P. H., Ablashi, D. V., Berard, C. W., Carbone, P. P., Waggoner, D. E., and Malan, L. ( 1971). Cancer 27, 416-420. Levy, J. A., and Henle, G. (1966). J. Bacteriol. 91, 27.5276. Lherisson, A. M., Meyer, C., and Bonneau, H. (1967). Bull. Cancer 54, 419-422. McCormick, K. J., Stenback, W. A., Trentin, J. J., Klein, C., Nadkarni, J. S., Nadkami, J. J., and Clifford, P. (1969). J. Virol. 3, 525-527. McKinnell, R. G. (1969). Recent Results Cancer Res. Spec. Suppl. pp. 254-260. Macpherson, I. (1970). Adtian. Cancer Res. 13, 169-215. Mann, D. L., Rogentine, G. N., and Fahey, J. L. (1968). Nature (London) 217, 1180-1181. Manolov, G. ( 1970). Personal communication.
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
247
MathB, G. (1967). C . R. Acad. Sci. 264, 27022704. MathB, G. (1969). Brit. Med. J. 4, 7-10. Maurer, B. A., Imamura, T., and Wilbert, S. M. (1970). Cancer Res. 30, 2870-2875. Mayyasi, S. A., Schidlovsky, G., Bulfone, L. M., and Buschesk, F. T. (1967). Cancer Res. 27, 2020-2024. Meyer, G., Birg, F., and Bonneau, H. (1969). C. R. Acad. Sci. 268, 28482849. Miller, M. H., Stitt, D., and Miller, G. (1970). J. Virol. 6, 699-701. Mizell, M., Stackpole, C. W., and Halpern, S. (1968). Proc. SOC. E x p . Biol. Med. 127, 96-98. Mizell, M., Toplin, I., and Isaacs, J. J. (1969). Science 165, 1134-1137. Moller, G. (1961). j . E x p . Med. 114, 415-434. Mora, P. T., Brady, R. O., Bradley, R. M., and McFarland, V. W. (1969). Proc. Nut. Acad. Sci. U.S . 63, 1290-1296. Morton, D. L., Malmgren, R. A., Holmes, E. C., and Ketcham, A. S. (1968). Surgery 64, 233-240. Muir, C. S., and Shanmugaratnam, K. (1967). In “Cancer of the Nasopharynx,” UICC Monogr. Ser., pp. 13-17. Munksgaard, Copenhagen. Nadkarni, J. S., Nadkarni, J. J., Clifford, P., Manolov, G., Fenyo, E. M., and Klein, E. (1969). Cancer 23, 64-79. Nadkarni, J. S., Nadkarni, J. J., Klein, G., Henle, W., Henle, G., and Clifford, P. (1970). lnt. J. Cancer 6, 10-17. Nazerian, K., and Purchase, H. G. (1970). J. Virol. 5, 79-90. Nazerian, K., Solomon, J. J., Witter, R. L., and Burmester, B. R. (1968). Proc. SOC. Exp. Biol. Med. 127, 177-182. Ngu, V. A. (1965). Brit. J. Cancer 19, 101-107. Ngu, V. A. (1967a). In “Treatment of Burkitt’s Tumour” ( J . H. Burchenal and D. P. Burkitt, eds.), UICC Mongr. Ser., Vol. 8, pp. 71-76. Springer-Verlag, Berlin and New York. Ngu, V. A. (1967b). Brit. Med. J. 1, 345447. Ngu, V. A,, McFarlane, H., Osunkoya, B. O., and Udeozo, I. 0. K. (1966). Lancet 2, 414-416. Niedernian, J. C., McCollum, R. W., Henle, G., and Henle, W. (1968). J. Arner. Med. Ass. 203, 205-209. Niedernian, J. C., Evans, A. C., Subrahnianyan, L., and McCollum, R. W. (1970). N . Engl. J. Med. 282, 361465. Nilsson, K., Klein, G., Henle, W., and Henle, G. (1971). lnt. J. Cancer (in press). Nishioka, K. (1971). Aduan. Cancer Res. 14 (in press). Nishioka, K., Tachihana, T., Klein, G., and Clifford, P. (1968). Gann Monogr. 7, 49-59. Nishioka, K., Tachibana, T., Sekine, T., Inoue, M., Kirayama, T., Yoshida, T., Takada, M., Katamura, A., Sugano, H., and Wang, C. H. (1971a). Gann Monogr. LO ( in press ) . Nishioka, K., Tachibana, T., Kirayama, T., de ThB, G., Klein, G., Takada, M., and Kawamura, A., Jr. (1971b). Proc. lnt. S y m p . Princess Takamutsu, Cancer Res. Fund, l s t , 1971 (in press). Noyes, W. F. (1959). J. Exp. Med. 109, 423428. Oettgen, H. F., Aoki, T., Geering, C., Boyse, E. A., and Old, L. J. (1967). Cancer Res. 27, 2532-2534. Okazaki, W., Purchase, H. G., and Burmester, B. €3. (1970). Avian Dis. 14, 413429.
248
GEORGE KLEIN
Old, L. J., and Boyse, E. A. (1964). Annu. Reo. Med. 15, 167-186. Old, L. J., Boyse, E. A., Oettgen, H. F., de Harven, E., Geering, G., Williamson, B., and Clifford P. (1966). Proc. Nut. Acad. Sci. U. S. 56, 1699-1704 Old, L. J., Boyse, E. A,, Geering, G., and Oettgen, H. F. (1968). Cancer Res. 28, 1288-1299. Osunkoya, B. 0. ( 1967a). In “Treatment of Burkitt’s Tumour” (J. H. Burchenal and D. P. Burkitt, eds.), UICC Monogr. Ser., Vol. 8, pp. 233-247. Springer-Verlag, Berlin and New York. Osunkoya, B. 0. (1967b). Brit. J. Cancer 21, 302-311. Pasternak, G. (1965). J. Nut. Cancer Inst. 34, 71-83. Pasternak, G. I. (1969). Adoan. Cancer Res. 12, 1-99. Pearson, G., Klein, G., Henle, G., Henle, W., and Clifford, P. (1969). J. Exp. Med. 129, 707-718. Pearson, G., Dewey, S., Klein, G., Henle, G., and Henle, W. (1970). J. Nut. Cancer lnst. 45, 989-997. Penn, I., Brettschneider, L., and Starzl, T. E. (1969). Transplant. PTOC. 1, 106-111. Peter-Knecht, W., Bitter-Suermann, D., and Falke, D. (1968). Z. Med. Mikrobiol. Immunol. 154, 234-244. Pike, M. C. Morrow, R. H., Kisuuli, A., and Mafigiri, J. (1970). Brit. J. Preu. SOC. Med. 24, 3 9 4 1 . Pollack, R. E., Green, H., and Todaro, G. J. (1968). Proc. Nut. Acad. Sci. U. S . 60, 12CL.133. Pope, J. H. ( 1971 ). Personal communication. Pope, J. H., Horne, M. K., and Scott, W. (1968). Int. J. Cancer 3, 857-866. Pope, J. H., Horne, M. K., and Scott, W. (1969a). Int. J. Cancer 4, 255-259. Pope, J. H., Home, M. K., and Wetters, E. J. (1969b). Nature (London) 222, 186-187. Pope, J. H., Scott, W., Reedman, B. M., and Walters, M. K. (1971). Proc. Int. Symp. Princess Takamatsu, Cancer Res. Fund, lst, 1971 (in press). Prehn, R. T. (1963). In “Conceptual Advances in Immunology and Oncology,” 16th Annu. Symp., M. D. Anderson Hospital, pp. 475485. Harper (Hoeber), New York. Pulvertaft, J. V. (1964). Lancet 1, 238-240. Purchase, H. G. ( 1969). J. Virol. 3, 557365. Purchase, H. G. (1970). Cancer Res. 30, 1898-1908. Rabinowitz, Z., and Sachs, L. (1970). Virology 40, 193-198. Rabson, A. S., O’Conor, G. T., Baron, S., Whang, J. J., and Legallais, F. Y. (1966). Int. J. Cancer 1, 89-106. Rapp, F., Beutel, J. S., Feldman, L. A., Kitahara, T., and Melnick, J. L. (1965). J . Exp. Med. 121, 935-944. Rich, M. A. (1968). In “Experimental Leukemia” (M. A. Rich, ed.), pp. 15-40. North-Holland Pub!., Amsterdam. Roizman, B. (1971). Proc. Int. Symp. Appl. Med. Virol., 1971 (in press). Roizman, B., and Spring. S. B. (1967). In “Proceedings of the Conference on Cross Reacting Antigens and Neo-antigens” (J. J. Trentin, ed.), pp. 85-97. Williams & Wilkins, Baltimore, Maryland. Rubin, H. (1960). Proc. Nut. Acad. Sci. U.S. 46, 1105-1119, Silvestre, D., Kourilsky, F. M., Klein, G., Yata, J., Neauport-Sautes, C., and Levy, J. P. ( 1971). Int. J. Cancer (in press). Singhal, S . K., and Wigzell, H. (1971). Progr. Allergy 15 (in press). SjGgren, H. 0. (1965). Progr. Exp. Tumor Res. 6, 289322.
IMMUNOLOGICAL ASPECTS OF BURKITT’S LYMPHOMA
249
Sjogren, H. O., and Hellstrom, I. (1965). E x p . Cell Res. 40, 208-211. Skipper, H. E. (1967). Cancer Res. 27, 2636-2645. Smith, R. T., Klein, G., Klein, E., and Clifford, P. (1967). I n “Advances in Transplantation” ( J . Dausset, J. Hamburger, and G. Math& eds.), Vol. 2, pp. 484493. Munksgaard, Copenhagen. Southam, C. M., Tanaka, S., Arata, T., Simkovic, D., Miura, M., and Peptiopules, S. F. (1969). Progr. E x p . Tumor Res. 11, 194-212. Stevens, D. A,, Pry, T. W., Blackham, E. A., and Manaker, R. A. (1970). Int. J. Cancer 5, 229-234. Stewart, T. H. M. (1969). Cancer 23, 1368-1379. Stjernswiird, J., and Cliflord, P. (1970). In “Immunity and Tolerance in Oncogenesis” (L. Severi, ed.), pp. 749-758. Div. Cancer Res., IV Perugia Quadrennial Int. Conf. Cancer, Perugia. Stjernsward, J., Clifford, P., Singh, S., and Svedmyr, E. (1968). East Afr. Med. J . 45, 484497. Stone, R. S., Shope, R. E., and Moore, D. H. (1959). J. Exp. Med. 110, 543-546. Stuck, B., Old, L. J., and Boyse, E. A. (1964). Nature (London) 202, 1016-1018. Svedmyr, A., Demissie, A., Klein, G., and Clifford, P. (1970). 1. Nut. Cancer Inst. 44, 595-610. Svedmyr, A., Demissie, A., Klein, G., Gergely, L., and Clifford, P. (1971). Ann. N . Y. Acud. Sci. 177, 241-249. Svet-Moldavsky, G . J., Mkheidze, D. M., and Liozner, A. L. (1967). J . Nut. Cancer Inst. 38, 933-938. Tevethia, S. S., Couvillion, L. A,, and Rapp, F. (1968). J. Immunol. 100, 358362. Tweedell, K. S. (1967). Cancer Res. 27, 2042-2052. van F~irth,R., Gerter, H., Nadkarni, J. S., Nadkarni, J. J., Klein, E., and Clifford, P. ( 1971 ) . Immunology ( in press ). Vonka, V., Porter, D. D., McCombs, R. M., and Benyesh-Melnick, M. (1969). Bacten‘ol. Proc. 69, 154-154. Vonka, V., Benyesh-Melnick, M., Lewis, R. T., and Wimberly, I. (1970a). Arch. Gesumte Virusforsch. 31, 113-124. Vonka, V., Benyesh-Melnick, M., and McCombs, R. M. (1970b). J. Nut. Cancer Inst. 44, 865-872, Wahren, B., Espmark, A., Lundback, H., Henle, G., and Henle, W. ( 1971). Manuscript in preparation. Walters, M. K., and Pope, J. H. ( 1971 ). I t i t . J. Cancer (in press). Watkins, J. F. (1964a). Nature (London) 202, 1364-1365. Watkins, J. F. (1964b). Virology 23, 436438. Watkins, J. F. (1965). Virology 26, 746-753. Wedderburn, N. (1970). Lancet 2, 1114-1116. Weinberg, A,, and Becker, Y. (1970a). Exp. Cell Res. 60, 470-474. Weinberg, A., and Becker, Y. ( 1970b). Personal communication. Westphal, H., and Dulbecco, R. (1968). Proc. Nut. A d . Sci. U. S . 59, 1158-1165. Wiener, F., Klein, G., and Harris, H. (1971). J. Cell. Sci. 8, 681-692. Witter, R. L., Moulthrup, J. I., Burgoyne, G. H., and Connell, H. C. (1970). Avian Dis. 14, 25S267. Yata, J., and Klein, G. (1969). Int. J. Cancer 4, 767-775. Yata, J., Klein, G., Hewetson, J., and Gergely, L. (1970). Int. J. Cancer 5, 394-403. Zajac, B. A., and Kohn, G. (1970). J. Nut. Cancer Inst. 45, 399406. Ziegler, J. L., Cohen, M. H., Morrow, R. H., Kyalwazi, S. K., and Carbone, P. P. (1970). Cancer 25, 734-739.
250
GEORGE KLEIN
zur Hausen, H., and Schulte-Holthausen, H. ( 1970). Nature (London) 227, 245248. zur Hausen, H., Henle, W., Hunimeler, K., Diehl, V., and Henle, G. (1967). J . Virol. 1, 830-837. zur Hausen, H., Schulte-Holthausen, H., Klein, G., Henle, W., Henle, G., Clifford, P., and Santesson L. (1970). Nature (London) 228, 10551058.
Genetic Aspects of the Complement System1
.
CHESTER A ALPER AND FRED S . ROSEN* Blood Grouping Laboratory and Department o f Medicine. Children’s Hospital Medical Center. and Department o f Pediatrics. Harvard Medical School. Boston. Massochusetfs
I . Introduction . . . . . . . . . . . I1. Hereditary Angioneurotic Edema . . . . . . . A. Clinical Aspects . . . . . . . . . . B . Characteristics of C1 Inhibitor . . . . . . . C . Genetics . . . . . . . . . . . D. Pathogenesis . . . . . . . . . . . I11. C4 Deficiency in Guinea Pigs . . . . . . . . IV C2 Deficiency in Man . . . . . . . . . V. Guinea Pigs Deficient in the “Third Component of Complement” A . Characteristics of the Defect . . . . . . . B . Applications of the Model . . . . . . . . VI . Genetic Structural Polymorphism in C3 . . . . . A . C3 Polymorphism in Man . . . . . . . . B C3 Polymorphism in the Rhesus Monkey (Mucaca mulutta) . VII C3 Deficiency in Man . . . . . . . . . VIII . Congenital Hypercatabolism of C3 . . . . . . . A Characterization of the Defect . . . . . . . B . Hypothesis on the Pathogenesis . . . . . . . IX . C5 Deficiency in Mice . . . . . . . . . A. Characterization of the Defect . . . . . . . B Applications of the Model . . . . . . . . X . C5 Dysfunction in Man . . . . . . . . . XI . C6 Deficiency in Rabbits . . . . . . . . . A. Characterization of the Defect . . . . . . . . . . . . . . B . Applications of the Model . XI1 Miscellaneous . . . . . . . . . . . A. Polymorphism of Human C4 . . . . . . . B Polymorphism of Guinea Pig C2 . . . . . . C . Inherited Structural Polymorphism of Human GlycineRich P-Glycoprotein . . . . . . . . . D . Deficiency of Clq in Agammaglobulinemia . . . . References . . . . . . . . . . . .
.
.
.
.
.
.
.
. . . . . . . . .
.
. . . . . . . . .
. . . . . . . . .
.
. .
252
253 253 254 255
255 258 259 262 262 263 263 263 268 268 270 270 274
275 275 277 281 281 281 283 284 284
285 285 285 286
.
‘The original observations cited in this review were aided by U . S Public Health Service Grants AM 13855 and A1 05877 and by a grant from the John A . Hartford Foundation . ‘Recipient of a career development award (1.K3.AM.19, 650) from the U S Public Health Service.
. .
251
252
CHESTER A. ALPER AND FRED S . ROSEN
I. Introduction
During the past decade, many of the complement components and complement-related proteins were isolated and characterized for the first time. It then became possible to investigate the physiology and pathophysiology of these proteins in health and disease, including genetic variations, by techniques previously established for proteins in general. In this way, a beginning has been made in delineating genetic factors affecting the complement system. Much of the genetic variability thus far encountered among the proteins of the complement system consists of structural heterogeneity which for the most part is harmless. Inherited deficiency states have been observed, but these too are largely without clear-cut ill effects on their hosts. The only known genetic abnormalities of the complement system with dire consequences for their bearers are those that affect inhibitors of the system or, in one instance, involves a possible structural abnormality interfering with the function of one of the primary complement components. In some instances, new methods of investigation have revealed abnormalities of the complement system in long-recognized inherited diseases, as in the case of hereditary angioneurotic edema, a disorder described over 80 years ago. In other cases, new syndromes have been recognized and defined by modern techniques. The presence of genetic variation in the complement system contributes to several areas of scientific and practical endeavor. Genetic polymorphism in this system can be added to the long list already accumulated from the study of other proteins to aid in the investigation of population genetics. Such polymorphisms can also be expected to be included in procedures related to identification of individuals in medicolegal matters as in paternity testing, With the rapid progress now occurring in cytogenetics, we can expect genetic polymorphisms in the complement system to play a role in chromosome cartography and the respective genetic loci themselves to be localized on specific chromosomes within the next few years. Structural genetic polymorphisms will certainly aid in the investigation of structure-function relationships in the complement system. Such polymorphisms may also provide hints about the structure of the complement proteins themselves, at least in so far as subunit organization is concerned, much as they have done in the case of isoenzymes. From an immunological and complementological point of view, the most important aspect of genetic variation in the complement system has been and will be the use of inherited deficiency and dysfunctional
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
253
states in defining the role of complement in reactions which play a role in host defense against infection and injury. In this sense, these “natural experiments” provide us with superb controls for in vitro systems and the schemata derived therefrom. The questions that ultimately matter, after all, are not those that ask “What is the complement system capable of doing?” but rather “What does it do in the living animal?” It is the aim of this review not only to summarize present knowledge in the field but also to describe experiments which utilize genetic structural variants or deficiency states in the study of complement function in general. II. Hereditary Angioneurotic Edema
Hereditary angioneurotic edema (HANE) is a genetically transmitted defect in the biosynthesis of the C1 esterase inhibitor. Several hundred individuals with this biochemical defect have been identified since its description by Donaldson and Evans (1963). A. CLINICAL ASPECTS Although the disease was recognized as a distinct entity by Graves and Quincke in the middle of the nineteenth century, Osler’s description (1888) attests to the astuteness of his clinical prowess. Affected individuals are prone to sudden, unheralded attacks of circumscribed subcutaneous edema. The swelling, which may be severe enough to cause remarkable disfigurement of the affected part, evolves very quickly and usually subsides within 72 hours. There is no discoloration, redness, pain, or itching associated with the edema. Despite its undistinguished appearance, involvement of the mucous membranes of the hypopharynx and larynx may result in untimely death. A rash, reminiscent of erythema marginatum, is recognized by some affected patients, and may occur without associated swelling. All affected individuals have recurrent attacks of abdominal pain due to collections of edema fluid in the intestinal wall. The colic is severe and spasmodic. Bilious vomiting or copious, watery diarrhea may ensue, but in any case, the hemoconcentration is noteworthy; the hematocrit rises above 50%. Abdominal attacks are often not accompanied by cutaneous swelling. Although trauma is clearly relatcd to attacks of angioedema, more frequently the cause of any individual bout of swelling is obscure. Patients frequently associate menses, violent exercise, extremes of temperature, and psychic trauma with onset of attacks of angioedema. The frequency and severity of attacks typically exacerbate at adolescence and
254
CHESTER A. ALPER AND FRED S. ROSEN
subside in the fifth decade of life (Donaldson and Rosen, 1966; Landerman, 1962). Hereditary angioneurotic edema is transmitted as an autosomal dominant characteristic. There is considerable confusion about the inheritance of the disease in the older literature, prior to the ascertainment of cases by laboratory means. Since the age of onset and the severity of symptoms are extremely variable, the earlier reports of the disease “skipping a generation” merit no further consideration. The mutation frequency of the gene may be fairly high (approximately 1/100,000 births). It is remarkable that affected individuals, who are heterozygous for the defect by virtue of its dominant inheritance, have very little or no C1 esterase inhibitor activity ( Donaldson and Evans, 1963), the biosynthetic error which causes the disease.
B. CHARACTERISTICS OF C1 INHIBITOR Purification of C1 led to an appreciation of the fact that it circulates in blood as a proenzyme (Lepow et al., 1956). Conversion of C1 to its enzymatic form, as well as the activity of the enzyme vis-A-vis its natural or synthetic substrates, was inhibited by a heat-labile serum protein, designated C1 esterase inhibitor, which has been purified from normal human serum. It was labile to heating at 60°C and to pH changes below 6 where it became irreversibly inactivated. It stoichiometrically inhibited C1 esterase activity (Pensky et al., 1961). The molecular weight is not yet precisely known. It migrated as an a,-globulin on electrophoresis. Of the glycoproteins of serum, it is the most carbohydrate-rich, containing 12%hexose, 13%hexosamine, 17% sialic acid, and less than 1%fucose. Isolation was reported independently by Schultze et al. (1962) who, unaware of its biological activity, termed it a,-neuraminoglycoprotein ( a?-NGP). Pensky and Schwick ( 1969) have recently proved the identity of wNGP and C1 esterase inhibitor. Although a,-NGP was at first considered to be nonantigenic, antisera have subsequently been raised in rabbit and goat and are obviously useful in the immunochemical estimation of the C1 esterase inhibitor (Rosen et al., 1965; A.-B. Laurel1 et al., 1969) and in phylogenetic studies ( Donaldson and Pensky, 1970). The C1 esterase inhibitor in serum can be assessed on a functional basis in addition to immunochernical estimation. Its ability to inhibit hydrolysis of N-acetyl tyrosine ethyl ester by C1 esterase (CE) formed the basis for the definition of one unit of C1 esterase inhibitor as that serum volume which will inhibit 10 units of CG. One unit of CG will hydrolyze an amount of N-acetyl tyrosine ethyl ester equivalent to a net titration of 0.01 ml. of 0.05 N NaOH in 15 minutes at 37OC (L. R. Levy and Lepow, 1959). By this assay, the normal serum content of C1
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
255
esterase inhibitor was 6.0 & 1.8 units (Donaldson, 1966). A colorimetric assay has been developed using N-a-acetyl-L-lysine methyl ester as a more sensitive substrate than the tyrosine ester ( Harpel, 1970). The C1 esterase inhibitor prevented the formation of EAC14 from EAC4 and EAC142 from EAC14 (Lepow and Leon, 1962) in direct proportion to the concentration of the C1 esterase inhibitor present' ( Leon and Lepow, 1962). These observations led to an assay based on the stoichionietric inhibition of guinea pig C i in the formation of EACl4 from EAC4 (Gigli et al., 1968).
C. GENETICS By immunochemical means, it was estimated that the serum of affected individuals contained a mean of 17.5%of the normal concentration (Rosen et al., 1971). The normal serum concentration has been variously reported to be 18 f 5 mg. % (Rosen et al., 1971) and 23.5 3 mg. % (W. Becker et aZ., 1970). Fifteen percent of all affected kindred were unusual in that sera of HANE patients in these kindred contained a biologically inactive form of C1 esterase inhibitor which was immunochemically identical to the normal. Of 11 such families identified thus far, the affected sera in 8 contained normal amounts of C1 esterase inhibitor, whereas in the other 3, the C1 esterase inhibitor content was 3-4 times normal. About half the excess C1 esterase inhibitor in such sera was bound noncovalently to albumin (Rosen et al., 1971). In the sera containing normal or elevated concentrations of C1 esterase inhibitor, the entire population of immunochemically discernible molecules was abnormal in its electrophoretic mobility (Fig. 1 ) . Further variation has been noted among these abnormal proteins in their capacity to bind C1 esterase. In two kindred, the abnormal protein blocked esterolysis of the N-acetyl tyrosine ethyl ester substrate but not the inactivation of C4 (Rosen et al., 1971). By immunofluorescent techniques, the hepatic parenchymal cell has been identified as the site of synthesis of C1 esterase inhibitor. Three to four percent of normal hepatic cells fluoresced with monospecific antiserum to C1 esterase inhibitor. No fluorescent cells were found in liver biopsies of 2 HANE patients with the common form of the disease, i.e., low serum concentration of antigenic C1 esterase inhibitor (Johnson et al., 1971).
*
D. PATHOGENESIS The plasma of patients with HANE was unusual in that it contained measurable C1 esterase ( CG ) activity, particularly during attacks of angioedema ( Donaldson and Rosen, 1964; Siboo and Laurell, 1965).
256
CHESTER A. ALPER AND FRED S. ROSEN
FIG. 1. Patterns of C1 inhibitor in serum after electrophoresis in agarose gel at pH 8.6 and immunofixation with specific antibody. The anode was a t the top. The serum at the far right was from a normal subject, that at the far left was from a patient with hereditary angioneurotic edema and low C1 inhibitor antigen concentration. The central four sera were from patients with normal or elevated C1 inhibitor antigen concentration. The more anodal of the two C1 inhibitor bands in the second pattern from the right is C1 inhibitor-albumin complex.
Blood was collected carefully in siliconized glassware and mixed immediately with ethylenediaminetetraacetate ( EDTA ), so as to prevent activation of Cls extracorporeally. As might be anticipated, in viuo C1 activation in HANE patients resulted in depletion of C4 and C2. Titers of these components fluctuated with the symptomatic state of the patient, and C2 titers were sometimes normal in the intervals between attacks of angioedema (Donaldson and Rosen, 1964; Austen and Sheffer, 1965; A.-B. Laurel1 et al., 1966; Pickering et aZ., 1968; Ruddy and Austen, 1967). However, C4 titers were consistently depressed, becoming more so during symptomatic periods. Catabolic studies of isotopically labeled C4 revealed a two- to fourfold increase in the rate of C4 destruction in uivo (Carpenter et at., 1969). The consumption of C4 and C2 in the fluid phase, or plasma, appeared to be largely ineffectual for serum levels of C3, and other late-acting components were within the normal range; the rate of C3 catabolism, using isotopically labeled
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
257
purified C3, was normal or minimally increased ( Alper and Rosen, 1967; Carpenter et al., 1969). The question remains as to whether these rapid, dynamic changes in complement interactions in vivo have, in fact, anything to do with the pathogenesis of angioedema. The plasma of HANE patients, particularly during an attack of angioedema, provoked an exaggerated response when injected intradermally into the patient’s own skin but not when injected into normal volunteers (Landerman et al., 1960). Highly purified CG (Haines and Lepow, 1964) produced angioedema when injected in guinea pig (Ratnoff and Lepow, 1963) or human skin (Klemperer et al., 1968). The lesion involved the formation of large interendothelial cell gaps in the postcapillaiy venule, both in man and guinea pig, with extravasation of vascular contents and also modest degranulation of mast cells (WillmsKretschmer et al., 1970), similar to the change observed in biopsy specimens of angioedema lesions (Sheffer et al., 1971). Although the lesion could be prevented with antihistamine in the guinea pig (Ratnoff and Lepow, 1963), it was not altered in man by prior ingestion of antihistamine (Klemperer et al., 1968). Guinea pigs deficient in C4 were much less responsive than normal guinea pigs to intradermal injection of C& (Frank and Rosen, 1971), and C2-deficient people were unresponsive to 8 times the dose of C G which yielded a maximal response in normal controls (Klemperer et al., 1968). Injection of C& into patients with HANE provoked an attack of angioedema. However, for about 96 hours following a symptomatic period, HANE patients were unresponsive to CG injection; they may have, in fact, developed true tachyphylaxis. The evidence obtained thus far indicates that a vasoactive peptide may be generated from the interaction of CG, C4, and C2 (Klemperer et al., 1969). The central dilemma which remains to be resolved concerns the mechanisms involved in the intermittent in vivo activation of C1 in the plasma of HANE patients. They are obviously liable to uninterrupted C1 activation, but in reality this occurs at relatively infrequent intervals. Activation of C1 could be effected in vitro by plasmin (Ratnoff and Naff, 1967) and kallikrein (Gigli et al., 1968). The addition of urokinase (Donaldson, 1968a) or activated Hageman factor (Donaldson, 1968b) to HANE plasma resultcd in rapid C1 activation, presumably via the fibrinolytic mechanism in the former case and of the kinin system in the latter. The activation of C1 was blocked by soybean trypsin inhibitor, which did not inhibit the esteratic activity of activated C1 ( C i ) . The C1 esterase inhibitor also inhibited plasmin, kallikrein, the globulin permeability factor PF/dil, and the C l r subcomponent of C1 (Kagen and Becker, 1963; Ratnoff et al., 1969). Thus, the broad spectrum
258
CHESTER A. ALPER AND FRED S. ROSEN
of inhibitory properties of this protein in the interlocking directorate of C1 activation increased the liability of C1 activation in HANE plasma. The variable therapeutic efficacy of t-aminocaproic acid and a cycloaminohexane analog, tranexamic acid, may be explained by their effects in blocking plasmin activation of C1 (Lundh et al., 1968). As previously mentioned, plasma from HANE patients, during an attack of angioedema, enhanced vascular permeability upon intradermal injection into HANE patients (Landerman et al., 1960) or guinea pigs (Donaldson et al., 1965). This was not true of plasma obtained during an interval between attacks. If such plasma, obtained during an asymptomatic period, was incubated in siliconized Lusteroid containers in the presence of EDTA or citrate, permeability activity could be generated at 37°C in 3 hours (Donaldson et al., 1969). The generation was blocked by prior addition of C1 esterase inhibitor or soybean trypsin inhibitor to the plasma. The permeability factor generated was dialyzable, was not blocked by antihistamine, and had kininlike activity in that it contracted estrous rat uterus in addition to promoting vascular permeability. The activity was destroyed by carboxypeptidase and chymotrypsin (Donaldson et ale, 1969). However, it differs from bradykinin in that the vasopermeability peptide was also trypsin-sensitive and did not lower blood pressure nor cause pain and redness on intradermal injection. Its amino acid composition differed from bradykinin; it contained no aromatic amino acids (Donaldson et al., 1970). The presumption that this peptide was derived from C2 was supported by the observation that anti-C4 and anti-C2 blocked its generation, whereas anti-C3 did not (Donaldson et al., 1970). It thus appears that the ultimate mediation of symptoms in angioedema is perpetrated by a highly vasoactive kininlike peptide, probably derived from C2, and not by bradykinin as was originally postulated (Landerman et al., 1960; E. L. Becker and Kagen, 1964; Burdon et al., 1965). Bradykinin levels, ascertained by radioimmune assay, were not consistently elevated in angioedema ( Talamo et al., 1969). However, kallikrein and fibrinolytic mechanisms may be intimately involved in the activation of C1 and, thus, in the initiation of the sequence of biochemical events which lead to angioedema formation ( Donaldson, 1970). Ill. C4 Deficiency in Guinea Pigs
As a result of screening guinea pig sera for possible allotypic antibody following immunization with normal guinea pig serum, a single, male, NIH guinea pig was found to have made antibody to guinea pig C4. With the antiserum, five other C4-deficient guinea pigs were found among 250 animals screened. Mating of the propositus with a normal
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
259
female yielded a litter of four babies with intermediate C4 titers. Subsequent mating of the propositus with a presumably heterozygous daughter resulted in a litter of 2 deficient male offspring and 1 heterozygous female. A third mating of the propositus with an unrelated C4deficient female produced 4 CCdeficient offspring. Thus, C4 deficiency was transmitted as an autosomal recessive characteristic with easy detectability of the heterozygous state (Ellman et al., 1970). The affected animals were healthy. Serum from CCdeficient guinea pigs had no hemolytic activity and no detectable hemolytic C4. Addition of purified guinea pig C4 to their serum resulted in restoration of hemolytic activity (Frank et al., 1971). Sera of an earlier reported strain of complement-deficient guinea pigs, discovered in Vermont in 1919, could not be restored to normal hemolytic activity by zymosan-treated serum ( C’3-depleted), whereas sera of the present NIH strain could be. It appears that the defects are, in fact, different although this cannot be conclusively proved because of the lamentable demise of the Vermont strain (cf. Section V,B). Serum titers of other complement components in C4-deficient guinea pigs were normal with the exception of C2 which was consistently low (about half of the normal level). Antibody to guinea pig C4, from a deficient guinea pig, reacted only with guinea pig C4 but not with rabbit, mouse, cat, and goat C4, nor with human C4, except for that of a single individual (Frank et al., 1971). Heterozygous guinea pigs had 83m of the normal C4 level and were thus similar to persons heterozygous for C2 deficiency in that their sera contained less than half the normal C4 content (Frank et al., 1971). Both tuberculin type and contact sensitivity were intact in C4deficient guinea pigs. However, they had a subtle impairment of antibody formation. When given ovalbumin or bovine serum albumin (BSA) (1 pg.), there was little or no antibody production by C4-deficient animals compared with randomly bred controls (Ellman et al., 1971). Nonimmune inflammation, direct and reverse passive Arthus reactions, and passive cutaneous anaphylaxis were normal in C4-deficient guinea pigs (Ellman et al., 1971). Clearance of “Cr-labeled autologous or isologous red cells coated with rabbit antiguinea pig erythrocyte serum was markedly impaired in deficient animals. IV. C2 Deficiency in M a n
Silverstein (1960) reported the absence of C2 in a healthy adult male. Although the first-degree relatives were studied, it was not appreciated at the time that 3 of them were heterozygous for the deficiency (Klemperer et al., 1967). Klemperer et al. (1966) reported a kindred in which
260
CHESTER A. ALPER AND FRED S. ROSEN
3 siblings’ sera lacked hemolytic C2. Total hemolytic activity was also undetectable in these three sera. No measurable SAC142 could be formed when EAC14 were incubated in C2-deficient serum. However, the deficient sera did not inhibit the formation of SAC142 when mixed with normal serum or purified C2. These sera thus did not contain an inhibitor of C2 activity. The three affected individuals in this kindred, now in the sixth decade of life, are clinically well and exhibit no undue susceptibility to infection; in this regard they resemble the case originally reported by Silverstein, Hypersensitivity of the delayed and immediate types were intact. When the immune adherence and bactericidal activities of C2-deficient sera were tested with Brucella abortus and Salmonella typhosa in the presence of excess antibody, it was found that they contained 515% of normal activity. Although the C2 deficiency was profound when measured in the hemolytic assay, which requires fixation of all nine complement components, detectable C2 could be appreciated when an assay system requiring fixation of only the first four reacting components of complement was used, e.g., immune adherence or enhancement of bacterial phagocytosis (Klemperer et al., 1966, 1967; Gewurz et al., 1966; Johnston et al., 1969). Bacteriolysis of gram-negative strains by C2-deficient sera was also more efficient when the assays were performed without an excess of antibodies to the bacteria (Klemperer et al., 1967). Seventeen heterozygous relatives of the propositi in the 2 aforementioned kindred have been detected (Klemperer et al., 1966, 1967). The sera of the heterozygotes contained slightly less than half-normal C2 titers, and total serum hemolytic complement activity was slightly depressed. Their sera also yielded subnormal values in certain immune adherence and bactericidal assays. Polley (1968) and Klemperer (1969) have obtained rabbit antibodies to purified human C2. With these antisera, no detectable C2 could be quantitated in the homozygous deficient individuals. Klemperer ( 1969) has found the seventeen heterozygous sera to contain between 30 and 60%of the normal level (Fig. 2). Cooper et al. (1968) have reported a third kindred. The serum of a homozygous C2-deficient female child lacked detectable hemolytic whole complement and C2 activities. Her serum, however, had a respectable immune adherence titer. The mother, maternal grandmother, and halfbrother of the proposita were heterozygous for the deficiency. Cooper et al. (1968) were able to assess the amount of C2 in the serum of the proposita by measuring the uptake of l””I1abeled C 3 onto EAC142. When EAC14 with >3000 C4 molecules/cell was incubated in normal serum a maximal C2 input of thirty effective C2 molecules resulted in
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
261
FIG. 2. Electroimmunodiffusion for estimation of C2 concentration. Specific anti-C2 was incorporated into agarose gel and electrophoresis was carried out with the anode at the top. From left to right: the first and fourth wells contained sera from homozygous C2-deficient persons; the fifth, sixth, and seventh wells sera from heterozygous CZdeficient persons; and the remaining wells contained normal sera.
the uptake of 4 x lo5 C3 molecules from normal serum. By this sensitive method, it was found that the homozygous C2-deficient serum contained between 1 and 4% of the normal C2 level which resulted in uptake of approximately 1.2 x lo4 C3 molecules/cell. A fourth kindred has been described by Ruddy et al. (1970). The propositus had an affected brother and sister, 3 heterozygous siblings, and 3 normal siblings. Heterozygosity was shown in both parents of the propositus for the first time. A plot of C2 protein concentration against C2 activity in this kindred revealed that 1 pg. of C2 was equivalent to 4 x 1 O ' O effective C2 sites. It thus required 130 molecules of C2 to produce an effective site. The cord blood of the last-born homozygote contained no C2, indicating failure of transplacental passage of C2. Although members of this large kindred dwell in less than optimal sanitary conditions in the vicinity of Mexico City, they are in good health.
262
CHESTER A. ALPER AND FRED S. ROSEN
V. Guinea Pigs Deficient in the “Third Component of Complement”
A. CHARACTERISTICS OF THE DEFECT Some 20 years after the discovery and definition of complement, a strain of complement-deficient guinea pigs was found in the Veterinary Laboratory of the Vermont State Agricultural Experiment Station in Burlington (Moore, 1919; Coca, 1920; Ecker, 1921). Although the deficient animals were at first thought to be more susceptible to death following inoculation with Vibrio cholerae suis and to changes in temperature than normal guinea pigs, later reports questioned these increased susceptibilities (Hyde, 1923) and finally the impression was that they were hardy as normal guinea pigs (Hyde, 1932). In a naturally occurring laboratory streptococcal infection, almost an entire colony of 500 guinea pigs died, yet the complement-deficient guinea pigs fared no worse than the normal animals. It was promptly ascertained that the complement-deficient guinea pigs made antibody in a normal fashion (Moore, 1919) and that their serum lacked opsonic activity despite the presence of antibody. At the time, complement was known to consist of at least two components separable by dialysis of fresh serum against water-an insoluble fraction ( euglobulin ) termed “midpiece” and a soluble fraction ( pseudoglobulin ) designated “end piece.” Experiments a few years prior to the discovery of the complement-deficient guinea pigs had demonstrated “a third component” of complement susceptible to destruction by incubation of whole serum with cobra venom or yeast. It was soon found that heatinactivated human or normal guinea pig serum, even in small amounts, restored hemolytic complement activity to the deficient guinea pig serum (Coca, 1920; Hyde, 1924) and that this restoring capacity of normal serum was abolished by prior incubation with cobra venom, yeast, or bacteria. It was therefore concluded that the missing complement factor was the “third component” (C3, C5, C6, C7, C8, or C9, in modern terms). As measured in CH,,, units, deficient guinea pig serum had about 1% of the hemolytic complement activity of normal guinea pig serum (Hyde, 1923). The complement deficiency was inherited as a simple Mendelian recessive trait (Hyde, 1923) in breeding experiments. This was, then, “the first record of the inheritance of a single non-sex linked character affecting the blood serum of a mammal” (Hyde, 1932). The discovery of the pattern of inheritance was credited to Raymond Downing (Hyde, 1932) who, it seems, did not publish. As little as 1%( v / v ) of fresh normal guinea pig serum restored full
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
263
hemolytic activity to the deficient guinea pig serum (Hyde, 1923), and small amounts of human, dog, cat, or rabbit were also effective. On the other hand, ox, pig, duck, pigeon, sheep, frog, goat, or mouse serum was ineffective. Following the intravenous injection into the deficient animals of fresh or heated normal guinea pig or human serum, there was “instantaneous” restoration of complement activity which persisted for 3 days.
B. APPLICATIONS OF THE MODEL The complement-deficient guinea pig was used as a model to show that the missing complement component did not cross the placenta in either direction (Hyde, 1923; Hyde, 1932). The serum of newborn deficient animals born of hybrid mothers with complement activity was deficient at birth, whereas when hybrid offspring were carried by deficient mothers the maternal serum remained deficient. Although the deficient animals sustained the same sort of anaphylactic shock as normal guinea pigs, it was found that on the intravenous administration of 0.5 ml. or more of heterophile (rabbit antichicken red cell) antibody, normal guinea pigs regularly died, whereas the complement-deficient guinea pigs lived ( Hyde, 1927). Also, whereas the subcutaneous injection of such an antiserum produced a local reaction in normal guinea pigs, none was seen in the deficient animals. Unfortunately, this valuable strain of complement-deficient guinea pigs has been lost and we shall probably never know the exact nature of their defect, although the excellent studies of the early workers made it likdy that the deficiency was, indeed, in one of the later-acting components. Our own prejudice, which cannot be tested at this late date, is that these guinea pigs were deficient in C3 since enhancement of phagocytosis and ( probably) local anaphylatoxin generation were impaired. VI. Genetic Structural Polymorphism in C3
A. C3 POLYMORPHISM IN MAN In the course of analyzing a large number of patients’ sera in agar gel electrophoresis, Wieme (1965) noted three patterns in which the slow p area showed two bands instead of the usual one. In two of the sera, the unusual band was of more rapid mobility and, in one, it was slower. Both bands were labile on storage, suggesting that they were both C3. The unusual bands were not seen on electrophoresis of sera from relatives of affected persons. Ropartz and co-workers (1965) reported briefly on a possible polymorphism in human C3. They found that 0.65%of normal human sera
264
CHESTER A. ALPER AND FRED S. ROSEN
contained an agglutinator for tanned red cells coated with purified C3. Addition of purified C3 to the agglutinating serum resulted in the loss of agglutinating ability. On examining sixteen such agglutinators and several different lots of purified C3, it was noted that the agglutinators differed in specificity. By using one agglutinator and one preparation of C3, these workers tested sera from 329 individuals and found that 11.25% inhibited the agglutination. The remaining sera either did not inhibit the agglutination or did so only partially. Although the authors speculated that their observations might be consistent with genetic polymorphism in C3, family studies have never been published, and no further work appears to have been done with this system. In a brief report (Alper and Propp, 1967), a slow electrophoretic variant of C3 was described in a single serum. Both bands in this serum reacted with antiserum to C3. Although no clearly double C3 pattern was seen in samples from the relatives of the proposita, in the serum of a maternal aunt the C3 pattern was slightly broad and blurred cathodally. It was later shown (Alper and Propp, 1968) that this variability was inherited. Wieme and co-workers (Wieme and Demeulenaere, 1967; Wieme et al., 1968) described slow C3 variants in individuals in each of three generations in one and in two generations of a second Flemish family. They also showed that both bands in affected individuals reacted with anti-C3. Their electrophoretic system did not allow accurate comparison of mobilities, so that it is not known whether the variants in the two kindred were of the same or different mobilities. The same group of investigators found a high frequency of variants, both rapid and slow, in s6ra from the Bantus of Rwanda (Wieme and Segers, 1968), such that the combined variant frequency was 4%. Since C3 is a relatively large molecule (molecular weight 240,000), it is apparent that only fairly large charge differences in variants are detectable by ordinary electrophoretic methods. By using higher voltage and prolonging the separation, the nature of the extensive genetic polymorphism in human C3 was defined. Use was also made of the fact that the mobility of C3 is slowed by the addition of Ca2+to the electrophoresis buffer to separate the C3 area from the other visible protein bands of the /3 region, transferrin and ,&lipoprotein (C.-B. Laurel1 et al., 1956). Using this system, seven C3 variants were initially identified (Alper and Propp, 1968). Unexpectedly, two of these were found to represent a common polymorphism in Caucasians and to a lesser extent, in American Negroes. The most common form of C3 in all populations was designated C3S (for slow) and the less common faster form was named C3 F. Because the absolute electrophoretic mobilities of all C3 variants are quite sensitive to Ca2+concentration, and mobilities could,
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
265
therefore, not be related to the mobility of some other serum protein such as transferrin, it was decided to designate rare variants in relation to the fastest and slowest of these then available, which were arbitrarily named F, and S,, respectively. Other rare variants were designated according to their electrophoretic positions relative to C3 S and either C3 F, or C3 S, in barbital buffer (ionic strength 0.05, p H 8.6, with 0.0018 M calcium lactate) (Fig. 3). Thus, C3 migrates 80% of the distance from C3 S to C3 F,, C3 So,, migrates 60% of the distance from C3S to C3S1, and so on. Azen and co-workers (1969; Azen and Smithies, 1968) independently observed the common polymorphism in Caucasians using high-voltage starch gel electrophoresis with an Mg"-containing buffer. They also found three rare variants which, on exchange of samples, corresponded to C3 So.c,C3 S,, and C3 Rose and Geserick (1969) independently described a polymorphism in aged serum which they designated Pt.
FIG.3. Prolonged agarose gel electrophoresis of fresh sera in barbital buffer, pH 8.6, ionic strength 0.05 with 0.0018 M calcium lactate. All sera contained C3 S and, in addition, the following C3 variants, reading from left to right: Fu, K, FM, Fa.,, F a . 6 , F, ( S ) , Sa.0, Si.
266
CHESTER A. ALPER AND FRED S . ROSEN
They showed that there were two common autosomal codominant alleles controlling the system (Rose and Geserick, 1969; Geserick et al., 1970a). It later became apparent that Pt was C3c (Geserick et al., 1970b) so that the basic polymorphism was in C3. Both family studies and analyses of random unrelated populations indicated that all of the C3 bands are controlled by autosomal codominant alleles (Alper and Propp, 1968; Azen and Smithies, 1968; Azen et al., 1969; Teisberg, 1970). In general, in serum from heterozygotes, the two major C3 bands were of approximately equal intensity. In all sera, each major band was accompanied by a corresponding minor anodal C3 band. The minor bands comprised 10-2W of the total C3 protein. The present list of identified forms of C3 in our laboratory is comprised of C3 F,.,, C3 Fl, C3 F,).,,,C3 Fo.s,C3 Fo.n, C3 F, C3 S, C3 So.B,and C3 S,. Teisberg (1971) has found at least four additional rare variants: C3 F,.,, C3 So.4, C3 So.8,and C3 So,9.All the variants were of similar size and do not appear to differ antigenically (Alper and Propp, 1968). Gene frequencies for the common C3 alleles are given in Table I. The remaining variant alleles occur with a frequency of less than 0.01 in most populations reported to date. As can be seen in Table I, C 3 F occurs not at all or in very low frequency in Oriental populations. The incidence for this allele in American Negroes was about one third that of American Caucasians, which is compatible with the hypothesis that C 3 F is purely a Caucasian gene. Sera from African Negroes have not yet been examined in any number for the presence of C 3 F but one would expect the incidence to be very low. The C3 activity in various genetic variants has been examined grossly by subjecting sera to preparative electrophoresis in agarose gel and testing fractions for their ability to increase the hemolytic complement titer of serum from patients with progressive glomerulonephritis speTABLE I C3 GENE FREQUENCIES IN VARIOUS POPULATIONS Population North American Caucasian
Norwegian German North American Negro North American Oriental
No.
S
F
References
349 0 . 7 7 0.22 Alper and Propp, 1968; Alper, 1970 123 0.77 0.21 Azen and Smithies, 1968; Asen et al., 1969 400 0.80 0 . 1 9 Teisberg, 1970 226 0 . 7 8 0 . 1 9 Hose and Geserick, 1969 113 0.92 0.07 Alper and Propp, 1968; Alper, 1970 41 0 . 9 6 0.04 Azen et al., 1969 68 0 . 9 9 - Alper and Propp, 1968; Alper, 1970
267
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
ad 0.040 1
I
5
10
(+I
0.240-(-)
ci Q
E g
h
k
-
- 5.0 - 4.0 - 3.0
0.080 -
- 2.0
0.040 -
-
0.200 -
0.1600.120
B
3 C,
1.0
1
S€G#€ffT NU#=R FIG. 4. Functional and immunochemical estimations of C3 in serum by preparative agarose gel electrophoresis. Two sera were examined. The upper contained C3 FIS and the lower, C3 SS.
cifically deficient in C3 (Klemperer et aZ., 1965). It had been shown previously that the addition of purified C3 to such serum resulted in a linear increase in CH,, to normal. Figure 4 shows the results of such an analysis using C3 F,S and C3 SS sera. The C3 hemolytic activity roughly paralleled C3 protein, including the minor gene-specific component. Similar results were obtained with C3 F,F, C3 F0.& C3 Fo.& C3 So.& and C3 FS,. The electrophoretic slowing of C3 in the presence of Ca2+has been alluded to earlier. Although the mechanism has not been elucidated, it appears likely that binding of Ca2+ (or Mg2+) to the C3 molecule is involved, and the reduction in mobility reflects a loss in net negative charge attendant upon this binding. In any event, the electrophoretic mobilities of all the C3 variants relative to each other were preserved independently of Ca2+concentration with two exceptions. The C3 S, had a mobility between C3S0., and C3S in the absence of Ca2+and in the
268
CHESTER A. ALPER AND FRED S. ROSEN
presence of 0.0018 M EDTA (Alper, 1970). A variant with the mobility and C3 S, in O.OOO9 M of C3 S, in 0.0018 M Ca2+migrated between C3 Ca2+(Teisberg, 1971). These observations suggest that in C3 S, and in this new slow variant, structural alteration occurs at a point on the C3 molecule such that divalent cation binding is affected. The genetic polymorphism of human C3 has been used to study physiological aspects of C3. Maternal and fetal C3 types were compared and found to differ in approximately one-third of the pairs (Propp and Alper, 1968; Azen and Smithies, 1968). Since no evidence of admixture was seen in such pairs, transplacental passage of C3 in either direction probably does not occur or is minimal. In a case of orthotopic hepatic transplantation, C3 types in the serum of a donor and recipient were compared (Alper et al., 19f39a). The recipient’s serum prior to transplantation was C3 FS,.,, whereas the donor’s was C3SS. A sample obtained from the recipient 20 hours after transplantation showed a mixture of types, but all subsequent samples were C3SS until his death 46 days after operation. These observations indicated that the liver is the primary, if not sole, site of synthesis of C3 in vivo.
B. C3 POLYMORPHISM IN THE RHESUS MONKEY (Macaca mulatta) Genetic polymorphism in C3 has also been found in the rhesus monkey (Alper et aZ., 1971a). In eighty-one sera from presumably unrelated animals, evidence was obtained for three alleles, designated rhesus C3s, C3F, and C3F~,with gene frequencies of 0.66, 0.33, and 0.01. In the small number of rhesus families available for study, the variants were inherited as autosomal codominant traits. Minor gene products, analogous to those in human C3, were observed in the rhesus sera, although they migrated cathodal to the major C3 bands. VII. C3 Deficiency in M a n
Subjects with approximately 50% of the normal C3 level were found in three generations of one kindred (Alper et al., 1969b). On typing for the genetic variants described above, it was found that all affected persons were apparent homozygotes at the C3 locus. Examination of the family tree in Fig. 5 suggests the genetic mechanism in affected individuals. Consider the propositus, 111-3, whose serum formed a single band with the mobility of C3 F on electrophoresis. His father’s serum was C3 FS and his mother, who also had half-normal C3 concentration, had C3 with the mobility of C3 S. The two possible mechanisms whereby half-normal concentration could have been inherited are either ( 1 ) that two structural alleles were
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
269
Hub family -
AT+
I
F-
1
3
II FS
S-
F-\
S-
FS
s-
ss
S-
S-
m FS
FS
SS
SS
FS
FIG. 5. Genealogy of the Hub family with partial C3 deficiency. Individuals indicated by half-blackened symbols had approximately half-normal C3 concentrations in serum; C3 types are shown below each symbol. Symbols without C3 types are untested individuals. (From Alper et al., 1969b.)
each producing protein at a half-normal rate, i.e., % (C3 SS) or % (C3 FF) or (2) that one structural gene was producing at a normal rate and the allelic gene was producing no significant protein. Since the mother of the propositus had no C3F gene to contribute, the second mechanism was operating. The same arguments can be applied to explain the inheritance of half-normal C3 levels by the mother of the propositus and by her brother, 11-4, from their father, 1-3. Thus, in this family, there was a nonexpressed or “silent” gene, C3-, which, like the structural variant alleles, was inherited in autosomal codominant fashion. Homozygotes for C3- with little or no detectable serum C3 are predicted by these findings but have not yet been found. The serum of these individuals with inherited partial C3 deficiency have CH,, contents in the normal or slightly subnormal range. The enhancement of the phagocytosis of antibody-sensitized pneumococci by normal peripheral blood leukocytes i n vitro was half-normal (Johnston et al., 1969), since this function is quite C3-sensitive. Nevertheless, these persons are asymptomatic and have no increased susceptibility to infection. In the serum of a healthy young man, C3 formed two bands on electrophoresis which corresponded to C3 F and C3 S (Alper and Rosen,
270
CHESTER A. ALPER AND FRED S . ROSEN
I t
fS 146
ss
fS I32
I57 2
I
m fS
I35
ss
144
FIG.6. Genealogy of the Boe family with a hypomolphic C3 variant, C3 f.
1971). However, in contrast to the usual C3 FS pattern in which C3 F and C3S occur in approximately equal concentrations, the C 3 F in this man’s serum was approximately one-half that of C3S. Similar C3 patterns were found in serum from this man’s mother and from one of his two sons, as seen in Fig. 6. Total serum C3 concentrations and hemolytic complement ( CH,, ) titrations were normal in affected heterozygous persons. Metabolic studies with isotopically labeled purified C3 FF and C3 SS revealed normal fractional catabolic rates in the propositus of this family and in C 3 F F and C3SS subjects. These studies suggested, but did not prove, that the variant C3F was hyposynthetic compared with C3s. Because of its hypomorphic and possibly hyposynthetic character, the variant allele was designated C3‘. Individuals homozygous for C3‘ would be expected to have subnormal C3 concentrations. VIII. Congenital Hypercatabolism of C3
A. CHARACTERIZATION OF THE DEFECT A 26-year-old man with Klinefelter’s syndrome and increased susceptibility to pyogenic infection since infancy manifested multiple abnormalities affecting C3. It appears likely that his basic defect is a genetic deficiency of a protease inhibitor. For this reason as well as the relevance of the findings in this patient to the biological functions of the complement system, studies of his abnormalities will be discussed in detail.
GENETIC ASPECXS OF THE COMPLEMENT SYSTEM
271
The infections which he has sustained have included pneumonia on four occasions, p hemolytic streptococcal and meningococcal septicemia, sinusitis, mastoiditis, otitis media, and a diphtheritic postauricular abscess ( Alper et al., 1970a). His only other symptom has been urticaria on exposure to warm or cold water. All aspects of immunity were found to be intact except for those mediated by the complement system and the latter were essentially absent. His serum was devoid of bactericidal activity for smooth gramnegative bacilli, of enhancement of the phagocytosis of antibody-sensitized Type I1 pneumococci by normal peripheral blood leukocytes, and of the elicitation of chemotaxis for neutrophiles in the presence of antigen-antibody aggregates. His serum hemolytic complement ( CH,, ) was 15 units/ml. (normal range: 32-45 units/ml.). Assay for individual complement components indicated that all were intact by functional or immunochemical tests or both, except for C3. Total serum C3 by immunochemical assay was 28 mg./100 ml. but about three-fourths of this was in the form of inactive conversion product, C3b, and only 7-8 mg./100 ml. had the mobility of native C3. The addition of C3 to his serum in uitro failed to restore any of the complement-mediated functions. Although there was no accelerated conversion of C3 to C3b in his serum or plasma in uitro, when purified 1251-labeledC3 was administered intravenously, the fractional catabolic rate was 10%of the plasma pool per hour (Alper et al., 1970b) or approximately 5 times normal (Alper and Rosen, 1967), but the rate of synthesis of C3 was normal. The rapid catabolism of C3 was ascribable to the rapid conversion of normal C3 in uiuo, since 404: of the labeled material was shown to be in the form of C3b 2 hours after administration and converted C3 has a fractional catabolic rate several-fold that of C3 ( Alper and Rosen, 1967). Indirect evidence for accelerated conversion of C3 in uiuo in this patient was a marked elevation of urinary histamine which was further increased by his taking a warm shower. The increased histamine excretion was attributed to in uitro elaboration of C3a, which is capable of degranulating both rat (Dias da Silva and Lepow, 1967) and human (Lepow et al., 1970) mast cells. The C3a inactivator was present in the patient’s serum. Because C3 is known to be highly susceptible to cleavage by a wide variety of proteolytic enzymes in addition to E A C E or C z , evidence for somc uninhibited proteolytic activity in the patient’s serum for gelatin, fibrin, and several synthetic substrates was sought, but not found. Moreover, a wide variety of protease inhibitors were measured
272
CHESTER A. ALPER AND FRED S. ROSEN
immunochemically or functionally in his serum, and were found to be normal, including a,-macroglobulin, al-antitrypsin, a,-antichymotrypsin, inter-a-trypsin inhibitor, and C1 inhibitor. Recently, what is presumed to be an uninhibited proteolytic activity has, indeed, been found in the patient’s serum. Glycine-rich pglycoprotein (GBG) (Boenisch and Alper, 1970b) is a 6 . 2 s protein of normal serum which converts very slowly on storage at 37°C into two 4 S products, glycine-rich y-glycoprotein (GGG) (Boenisch and Alper, 1970a), probably identical to p,-glycoprotein I1 ( H . Haupt and Heide, 1965), and an 0 1 ~product, GAG. The patient’s serum contained no material reactive with an antiserum to either GGG or GBG. When purified GBG was added to his serum and incubated at 37°C for 10 minutes, there was complete conversion to GGG and GAG, whereas no conversion was noted in a control serum similarly treated. The presumed proteolytic activity required Mg“ and could be inhibited by small amounts of normal serum. The link to C3 conversion remains obscure, however, in that even if GBG was added to the patient’s serum, no enhanced conversion of C3 occurred in vitro. On investigating the abnormalities in complement-mediated functions, it was found that these were normalized by small amounts of normal, C2-deficient, C3-deficient, C4- and C2-deficient, C5-deficient, C6-deficient, and “complement-fixed serum. This restoring ability was found to be heat-labile (52”C, 30 minutes) and to reside in the 5-6 S p pseudoglobulin fraction. This fraction of serum alone restored bactericidal and chemotactic activity, but C3 was also required for the enhancement of phagocytosis of antibody-sensitized pneumococci and for hemolytic activity. No enhancement of these activities was observed when purified GBG was added to the patient’s serum. Because of the physicochemical similarities of the “restoring fraction” of normal human serum and the C3 proactivator ( Muller-Eberhard et al., 1966), the activity of the latter in his serum was investigated. Although purified cobra factor (MullerEberhard et al., 1966) induced virtually complete conversion of C3 in normal serum, it caused no change in C3 in the patient’s serum (Alper et al., 1970b). Furthermore, although a 9 S radioactive peak on Sephadex G-200 gel filtration corresponding to cobra factor-proactivator complex could be demonstrated in normal serum with added lZ5I-labeledcobra factor, no such peak was seen in the patient’s serum. It thus appeared that the patient’s serum lacked C3 proactivator activity. The possibility that GBG and the C3 proactivator were the same was made unlikely by two observations. Purified GBG, C3, and cobra factor mixed and incubated at 37°C in the presence of Mg?+failed to result in either complex
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
273
formation or C 3 conversion. I n addition, when cobra venom or purified cobra factor was added to fresh serum, no change in size of GBG was noted on gel filtration of the mixture compared with that of GBG in untreated fresh serum. Similarly, GBG had no C3 inactivator activity, did not react with an antiserum to the C3 inactivator provided by Lachmann, and anti-GBG did not react with Lachmann’s purified C3 inactivator. In addition to these deficiencies, it was noted that the patient’s red cells were coated with C3 (as judged by agglutinability with anti-C3) in the absence of any demonstrable antierythrocyte antibodies or coating by other complement components (Abramson et al., 1969, 1971). The patient’s red cells lost their reactivity with anti-C3 on incubation with normal serum, purified C3 inactivator, or serum from a variety of patients with C3-coated red cells in association with immunohemolytic anemias but not on incubation in his own serum. The C3 inactivator was detected in his serum neither by titration for conglutinogen-activating factor (Lachmann and Muller-Eberhard, 1968) nor by reaction in agar gel with an antiserum specific for the C3 inactivator. Although incubation of the patient’s red cells with the C3 inactivator in purified form in buffer or in the patient’s serum or native in normal serum removed reactivity with anti-C3, reactivity with anti-arD (West et al., 1966; Bokisch et al., 1969) persisted. These findings suggested that C3b was attached to the patient’s red cells on the C3d portion of the molecule and that the C3 inactivator cleaved C3b from the red cell at a site similar to that attacked by trypsin on prolonged incubation with C3 (Bokisch et al., 1969). It could be shown that C3b, whether in the patient’s serum or produced by the action of trypsin or C m on purified C3, was altered by the C 3 inactivator in the fluid phase in such a fashion that its electrophoretic mobility coincided with that of C3c. However, when this altered fragment was compared with C3c produced by aging normal serum, it was found that the C3 inactivator-altered C3b was of approximately the same size as C3b ( 9 S) rather than that of C3c ( 7 S). In addition, whereas C3c produced by aging normal serum of the same genetic type as the patient’s ( C 3 SS) consisted of multiple bands on prolonged agarose gel electrophoresis, the C3 inactivator-altered C3b formed a single band. Since on aging of the patient’s serum there was no conversion of C3b to C ~ Cit, appears likely that thc C3 inactivator plays a role in the normal aging changes of C3b on storage, probably in conjunction with another substance (or substances) also deficient in the patient’s serum. Properdin activity, as measured by RP titration and the lysis of a properdin-sensitive strain of Shigella, was absent from the patient’s
274
CHESTER A. ALPER AND FRED S . ROSEN
serum. Nevertheless, using antiproperdin antiserum prepared by Pensky (Pensky et al., 1968), properdin antigen was shown to be present in normal concentration and was of normal mobility in the patient’s serum on inimunoelectrophoresis. Since properdin activity requires the presence of two protein cofactors, it is possible that one or the other or both of these proteins is deficient in the patient’s serum. The effects of an infusion of 500 ml. of normal plasma on the patient’s deficiencies were studied. Bactericidal activity, hemolytic complement activity, chemotactic activity, and enhancement of phagocytosis were all partially or completely restored, in some instances for as long as 17 days. Changes in C3 consisted of disappearance of C3b (after conversion of C3b in vivo to material with the electrophoretic mobility of C3c) and a rise in C3 from 8 to 70 mg.1100 ml. The changes in C3 persisted for 17 days, after which the original patterns were re-established. The GBG became detectable at 5.6 hours after the infusion, reached a peak concentration of 8%of that in normal pooled serum at 16 hours, and was again no longer detectable after 25 hours. The C3 inactivator activity was restored and the patient’s red cells lost their reactivity with anti-C3, and C3 proactivator activity was partially restored in that lT-labeled C3 added to his serum in vitro in the presence of cobra factor was promptly converted, but there was no formation of 9 s cobra factor-C3 proactivator complex. A second metabolic study with l?zI-labeled C3 performed during the normalization period revealed a diminution in the fractional catabolic rate for C3 and, once again, a normal synthesis rate. No i n vivo conversion of labeled C3 was observed. Further evidence that the rise in C3 was the result of C3 synthesis by the patient and was unrelated to the C3 in the normal plasma infusion was obtained by genetic typing. The donor plasma contained C3 FS, whereas at the height of the rise, the patient’s plasma was C3 S S .
PATHOGENESIS Despite the multiplicity of deficiencies in this patient, it seems likely that he has a single genetic defect, most reasonably of GBGase inhibitor. The other defects including hypercatabolism of C3, are probably secondary in some way to the uninhibited effects of GBGase. A working hypothesis of the interrelationships is presented in the scheme in Fig. 7. No attempt is made to explain the deficiency of the C3 inactivator, since the patient’s serum contains no demonstrable enzymatic activity against this protein. However, as in the case of C3, for which no aceelerated conversion can be shown in vitro, there may be one or more proteins in the chain reaction beginning with GBGase which are depleted i n vivo. B. HYPOTHESIS ON
THE
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
7
Y'
/
275
*+=J
1 '
C3a inactivator
Histamine
FIG.7. Hypothetical schema of the interrelationships of the defects in congenital hypercatabolism of C3. Deficient proteins or protein fragments are outlined. Rbcred blood cells; GBG-glycine-rich P-glycoprotein; GGG-glycine-rich y-glycoprotein; CAG-glycine-rich a-glycoprotein.
The recent observations (Goodkofsky et al., 1971) that purified preparations of GBG contained properdin factor B activity (Blum et al., 1959) and that the patient's serum was devoid of this activity provide an explanation for the absence of properdin activity in his serum. They also indicate that the patient's primary defect may be in the properdin system. Since deficiencies of C4 in the guinea pig and C2 in man are not associated with increased susceptibility to infection, it may be that this alternate pathway to C3 activation is crucial in resistance to infection. Disturbances in the latter, as exemplified thus far only in this patient, may prove to be of greater clinical import than those involving primary deficiencies of complement components. The pathogenetic mechanisms which result from this patient's primary defects may be analogous to those observed in patients with hereditary angioneurotic edema. A defective inhibitor results in both instances in spontaneous activation of an enzyme such that an unbridled attack on natural substrates proceeds in oioo, with a resulting depletion of these substrates and their attendant functions. IX. C5 Deficiency in Mice
A. CHARACTERIZATION OF THE DEFECT When complement in mouse serum was first studied, it was thought that the species did not possess a complete hemolytic system. As soon as the proper assay systems were developed (Borsos and Cooper, 1961), it became apparent that there were, indeed, inbred strains of mice whose serum lacked hemolytic complement, whereas the serum of other
276
CHESTER A. ALPER AND FREJJ S. ROSEN
inbred strains had this activity (Rosenberg and Tachibana, 1962; Terry et al., 1964). Thus, the hereditary basis for the complement deficiency was almost immediately suspected. This hypothesis was borne out in formal genetic analysis (Erickson et al., 1963; Herzenberg et al., 1963) by which it was shown in backcross breeding experiments that the presence or absence of serum complement hemolytic activity was controlled at a single segregating locus, Hc'. The positive allele, Hcl, was dominant, so that complement-deficient mice were homozygous for HcO. It was recognized that many strains of mice were complementdeficient as were many animals from noninbred strains (Tachibana et al., 1963). On immunizing complement-deficient mice with serum from mice with normal complement, a precipitating antibody was obtained which reacted with a p-globulin in normal mouse serum but failed to react with any protein in the serum of complement-deficient mice (Erickson et al., 1963, 1964). It was, therefore, thought that a single gene controlled both the production of the serum antigen and the missing complement component. Immunization of mice bearing hemolytic complement with serum from complement-deficient mice failed to produce antiserum to any serum protein, suggesting that complement-deficient animals truly lacked the protein antigen and that the Hco gene produced no detectable product. Antigen MuBl (Cinader and Dubiski, 1963) was found as the result of a deliberate search for genetic polymorphisms in mouse serum proteins. Antisera were raised in mice of one inbred strain against serum from another inbred strain, and the resulting antiserum reacted with sera from 61%of the inbred strains of mice tested (Cinader et al,, 1964). The reactive antigen was designated MuB1. No antibody to a corresponding antigen could be raised on immunization of animals who had the antigen to sera of those who lacked it. These results were interpreted as indicating that mice deficient in MuBl had no protein corresponding to MuB1. Anti-MuB1 produced in the mouse reacted widely with an analogous protein in the sera of almost all mammalian species tested. These findings suggested that deficient mice either lacked MuBl entirely or had so few circulating molecules that immunological tolerance was not induced. This interpretation was supported by the observation that antiserum to mouse MuBl produced in the rabbit did not react (after absorption) with MuB1-deficient mouse serum but reacted with normal mouse serum and with analogous proteins only in the sera of species closely related to the mouse, i.e., rat and hamster. The presence or absence of MuBl correlated well with the presence or absence of a complete complement system in the mouse, suggesting that the two phenomena were reflections of the same defect. The MuBl was present in
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
277
the serum of complement-deficient rabbits, indicating that the defect in rabbits was different from that in the inbred strains of mice. The physicochemical characteristics of hc’ and MuBl were identical: both were euglobulins, heat-labile in whole serum, and insensitive to ammonia or hydrazine (Erickson et al., 1964; Cinader et al., 1964; Tachibana and Rosenberg, 1964). In addition, 0.1 M EDTA or 0.1 M 2-mercaptoethanol had no effect on hc’ (Tachibana and Rosenberg, 1964), and the approximate sedimentation constant of MuBl was 7 S (Cinader et al., 1964). Further evidence for a relationship between hc’, MuB1, and complement was provided by the observation that the ,8globulin fraction of normal mouse serum restored hemolytic activity to Hco/Hcomice (Erickson et al., 1964). The final link was closed when it was shown that mouse anti-MuB1 reacted with purified human C5, that antiserum to human C5 prepared in mice lacking MuBl reacted only with sera from mice with MuBl and a complete complement system, and that anti-MuB1 and anti-C5 reacted in immunological identity with human C5 and normal mouse serum (Nilsson and Muller-Eberhard, 1965). It was further demonstrated that purified C5 restored complement activity to complement-deficient mouse serum, and anti-MuB1 inhibited C5 activity in human serum (Nilsson and Muller-Eberhard, 1967). The genetic aspects were considerably clarified by the demonstration that heterozygous C5-deficient mice had almost exactly 50% the serum concentration of age- and sex-matched homozygote normal animals ( Dubiski and Cinader, 1966). OF THE MODEL B. APPLICATIONS
Although C5-deficient mice appeared to be as hardy as mice with a complete complement system (Caren et al., 1965), some small but apparently definite differences have been shown in several immunological functions. For example, small differences were detected in the T2 phageneutralizing activity of C5-deficient and normal mouse sera (Okada and Rosenberg, 1964). Infection with various numbers of Corynebacterium kutscheri, a common mouse pathogen producing pseudotuberculosis, revealed that at doses of 5.1 and 1.5x loGbacteria, the differences in survival of deficient and normal mice were significant ( P < 0.02 and <0.001, respectively) ( Caren et al., 1965; Caren and Rosenberg, 1966). However, when activation of natural infection with the same organism was induced by the administration of hydrocortisone (which depresses serum complement in the mouse), survival rates were equal. Interestingly, in the same study it was found that C5-deficient animals appeared to have a survival advantage over normal mice until weaning at 3 weeks of age. This observation remains unexplained. When guinea pig erythro-
278
CHESTER A. ALPER AND FRED S. ROSEN
cytes, for which mice have natural antibodies, were injected intravenously, both C5-deficient and normal mice showed the same pattern of hepatic sequestration. Similarly, the rates of clearance in vivo of Salmonella typhimurium were identical in normal and C5-deficient mice (Stiffel et al., 1964). In the presence of added antibody, in vivo “opsonization” was the same in both sets of mice. In addition, decomplementation i n vivo by antigen-antibody complexes resulted in decreased opsonization to the same extent in C5-deficient and normal mice although reticulo-endothelial uptake of S . typhimurium was the same in the decomplemented mice as in normal mice. In testing the clearance in vivo of intravenously administered complement-sensitive and complement-resistant strains of Escherichia coli in C5-deficient and normal mice, it was noted that although overall bloodstream clearance of both strains was the same for the two groups of mice and both strains of bacteria, the C5-deficient mice killed the complement-sensitive strain more slowly ( Glynn and Medhurst, 1967). In an investigation of the requirements for maximal enhancement of the phagocytosis of antibody-sensitized pneumococci, it was shown that C1, C4, C2, and C3 were required as well as a dialyzable factor and a noncomplement 5-6 S ,8 pseudoglobulin (Johnston et al., 1969). The C5deficient mouse serum enhanced phagocytosis as well as serum from mice with normal complemcnt. These findings are in distinction to those of a subsequent study (Shin et al., 1969), in which it was found that C5deficient mouse scruin enhanced pneumococcal phagocytosis slightly but measurably less than nornial mouse serum. The addition of purified C5 to the deficient serum normalized the enhancement of phagocytosis. The difference in results between this and the previous study may be the result of other differences in the complement systems of the supposedly coisogenic strains used (BlO.D2/Sn “old” and “new” lines). First of all, there may be genetic heterogeneity in the new line mice in that new line mice regularly accepted grafts from old line mice (lacking C5), but the reverse was not always true (Caren and Rosenberg, 1965). These genetic differences did not appear to involve the H-2 locus, since the old line mice did not show second-set reactions. Second, there is suggestive evidence for other hereditary differences in inbred (and noninbred) strains of mice responsible for wide differences in hemolytic complement unrelated to C5 deficiency ( Rosenberg and Tachibana, 1968). Third, when different C5-deficient inbred strains of mice (other than B10.D2), for example BALB/c, were examined for serum enhancement of phagocytosis of pneumococci, it was found to be the same as in normal mice, even if age and sex were matched (Johnston et al., 1971).
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
279
These considerations may also play a role in the observed lower LDB0 for BlO.D2/Sn old line mice of a modestly virulent pneumococcus compared with the LD,, for B10.D2/Sn new line of the same bacteria (Shin et al., 1969). Alternatively, defective chemotaxis may explain the increased susceptibility of the old line mice. A possible role for complement, and specifically C5 or later-acting components, in passive cutaneous anaphylaxis was suggested by experiments with ribonuclease and antiribonuclease ( Ben-Efraim and Cinader, 1964). If antigen was injected into the skin and antibody was subsequently injected intravenously, increased vascular permeability was noted in both C5-deficient and normal mice. If, however, antibody was injected first and several hours elapsed before the injection of antigen, it was possible to show reactions in normal mice but not in C5-deficient mice. A normal reaction could be elicited in C5-deficient mice under the same circumstances by the prior administration of normal mouse or guinea pig serum but not of C5-deficient serum or heated guinea pig serum. Chemotactic activity for leukocytes could be produced in normal mouse serum after incubation with antigen-antibody precipitates, aggregated IgG, or zymosan. No such activity was generated on similar treatments of C5-deficient mouse serum, suggesting that C5 was necessary for this function (Ward et al., 1965). There appears to be no doubt that C5-deficient mice reject histoincompatible skin grafts normally (Caren and Rosenberg, 1965; Crisler and Frank, 1965). Similarly, the active Arthus reaction was the same in C5-deficient mice as in normal mice (Crisler and Frank, 1965). Antibody against mouse immunoglobulin allotypes suppressed these allotypic specificities in the serum of C5-deficient as well as normal mice (Cinader and Dubiski, 1968), indicating that complement-mediated lysis of immunoglobulin-producing cells was not involved in the mechanism of suppression. Antilymphocyte serum had equal potency in decreasing the number of plaque-forming cells in the spleen and hemolysin titers in the serum of C5-deficient as in normal mice (Barth and Carrol, 1970; Cinader et nl., 1971), so that complement-mediated lysis of lymphocytes was probably not involved in the mechanism of immunosuppression by antilymphocyte serum. Evidence that antilymphocyte serum may also contain graft-rejecting properties and that the late-acting complement components mediate this effect was obtained in recent studies of normal and C5-deficient mice (Cinader et al., 1971). It was observed that skin allografts survived longer on antilymphocyte serum-treated C5-deficient mice than on normal mice similarly treated-a .difference obliterated
280
CHESTER A. ALPER AND FRED S. ROSEN
by the administration of normal mouse serum to the deficient animals. A percentage of grafts from antilymphocyte serum-treated normal mice when transferred to untreated syngeneic normal mice underwent rejection, whereas no rejection was observed when C5-deficient animals served as donors and recipients in the same kind of experiment. Studies of nephrotoxic serum nephritis (Unanue et al., 1967; Lindberg and Rosenberg, 1968) showed similar histopathological changes in C5-deficient and normal mice and, although there may have been small strain differences in proteinuria and survival, the overall disease was similar in both kinds of mice. The lupuslike disease of NZB mice (Howie and Helyer, 1965) occurs despite C5 deficiency in this strain. Using C5-deficient niouse serum as a reagent to detect C5, this component was found to be present at 11 days of gestation in the mouse (Tachibana and Rosenberg, 1966). No maternal-fetal transfer of C5 was observed. The C5-deficient mouse has been used to obtain evidence for synthesis of C5 by bone marrow (Phillips et al., 1969). Bone marrow cells from inbred mice with C5 were injected intravenously into C5-deficient inbred mice. No prior treatment of recipients was necessary for B10.D2 new line into B10.D2 old line transfers since, as discussed earlier, these lines are purportedly coisogenic. However, A mice ( C5-deficient ) were lethally irradiated prior to receiving bone marrow from BGAF, (normal complement) animals since these lines are allogeneic. During the first 4 weeks after infusion, many of the recipients developed serum hemolytic complement activity which was greater in the case of A mice than BlbD2 old line mice. It was shown that in vitro culture of the recipients’ splenic tissues but not livers incorporated “C-labeled amino acids into C5 as judged by radioimmunoelectrophoresis. This suggested that the infused marrow was responsible for C5 synthesis and had lodged in the recipients’ spleens. Component C5 as antigen could not be detected in the recipients’ sera. The application of cell-hybridization techniques promises to yield much information with regard to cellular genetic mechanisms. A beginning has already been made in the study of the C5-deficient mouse (N. L. Levy and Ladda, 1971). Since dense cultures of normal mouse spleen (but not kidney) released material (probably C5) into the supernatant medium capable of restoring hemolytic activity to C5-deficient mouse serum, hybrids were produced using spleen cells from B10-D2/Sn new line (normal C5) and old line (C5-deficient) mice, kidney cells from new line mice, and adult chicken erythrocytes. A suspension of hybrid cells was injected intravenously into old line mice and recipient sera were serially analyzed for hemolytic complement activity. The injection of two kinds of hybrid cells resulted in restoration of complement activity
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
281
which reached a maximum at about 6 days and disappeared after 3 weeks: new line kidney cells: old line spleen cells and chicken erythrocytes: old line spleen cells. When Sendai virus, added to promote cell fusion, was omitted from these mixtures, no restoration was observed. Although exact details and mechanisms are yet to be resolved, it is clear that the hybrid synthesizes gene products not produced by either parental cell line alone. X. C5 Dysfunction in M a n
Recently, Miller and his colleagues (Miller et al., 1968; Miller and Nilsson, 1970) have described a syndrome in two unrelated infants, 1 male and 1 female, with failure to thrive, eczematoid dermatitis, and sepsis due to staphylococci and gram-negative enteric bacilli. A defect in the phagocytosis of yeast particles has been demonstrated to reside in failure of the sera of the propositi to opsonize the yeast. The serum defect was corrected by normal serum, and frequent infusions of normal plasma rendered the patients free of infections. After it was found that C5-deficient mouse serum was unique in that it failed to restore the patients’ sera, purified human C5 was found to be effective. Total hemolytic complement and immunochemically estimated C5 were normal in the propositi. However, when C5 was isolated from affected relatives, it was shown to have only 10%the hemolytic activity of normal C5. It may also prove true that C5a from the dysfunctional C5 is ineffective as a chemotactic stimulus. No altered electrophoretic mobility has been discerned in the purified dysfunctional C5 ( Nilsson, 1970). The transmission of the genetic defect in C5 dysfunction is unclear. Paternal and maternal relatives of the propositi are affected but are healthy in adult life. I t may be that the dysfunctional C5 is inherited as a codominant allele of normal C5 and, like dysfibrinogenemia, varying proportions of normal and abnormal C5 arc encountered among the various individuals in a single kindred. XI. C6 Deficiency in Rabbits
A. CHARACTERIZATION OF THE DEFECT
An inborn defect of the complement system in rabbits was discovered 10 years ago in Germany (U. Rother and Rother, 1961). Subsequently, rabbits with a similar deficiency have been found in Mexico (Biro and Ortega, 1966) and England (Lachmann, 1970). In the original studies of these rabbits ( U . Rother and Rother, 196l), 3 affected aniwals were found among 200 examined and these 3 were littermates. It is estimated,
282
CHESTER A. ALI’ER AND FRED S . ROSEN
on the other hand, that as many as 5%of the rabbits offered for sale in the market at Xochimilco are complement-deficient ( Biro and Ortega, 1966). In the early studies ( U . Rother and Rother, 1961) it was recognized that the sera of affected animals contained no anticomplementary or inhibitory activity for normal rabbit serum, that the properdin system was intact, and the hemolytic complement activity could be restored by the addition of purified “C’3” (latc-acting components). With the denionstration ( K . Rother et al., 1966a) that purified human C6 restored complement activity to coniplement-deficient rabbit serum, it was establishcd that these animals had a specific deficiency of C6. Moreover, when C6 was measurcd functionally, it was found to be absent (Nelson and Biro, 196S), although the C6 inactivator was present in normal amounts in the deficient rabbit sera. When partially purified rabbit C6 was injected into deficient animals, an antiserum was produced that inhibited the hemolytic activity of C6 (K. Rother et d., 1966a). Subsequently, it was shown (Lachmann, 1970) that a precipitating antiserum could be raised in this fashion and that this antiserum reacted with a wide variety of mammalian sera in addition to normal rabbit serum but failed to react with C6-deficient rabbit serum. Chicken antiserum prepared to rabbit C6 showed a much more narrow range of reactivity, producing a precipitin arc in gel diffusion with serum from man, guinea pig, horse, dog, and possibly goat and cow but not from mouse, rat, sheep, pig, or cat. It thus appears that, as is true of C5-deficient mice, C6-deficient rabbits have essentially no C6 protein. In genetic studies (Lachmann, 1970) of the English strain, the propositus, a “Himalayan” buck, was originally crossed with 5 does, one of which proved to be heterozygous deficient, and over 200 offspring in three generations were tested. The C6 deficiency was shown to be inherited as a single autosomal recessive trait. The antiserum was used to measure C6 protein in the sera of normal and heterozygous deficient rabbits, and it was observed that the heterozygous deficient rabbits had a mean concentration of 40%(range 2572%) of that in normal rabbit serum (range 70-160%). NO linkage between C6 deficiency and y-globulin allotypes or the VallLeu ambiguity in hemoglobin in rabbits was found (Hunter and Munro, 1969). This same antiserum failed to detect any C6 protein in the serum of the Mexican strain of rabbits, so that neither produced a dysfunctional molecule. With a gel diffusion method in which 1%EAC43( Antrypol) was incorporated into agarose gel, it was possible rapidly to screen a large number of rabbit sera for C6 deficiency ( Lachmann, 1970).
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
B. APPLICATIONS OF
THE
283
MODEL
The C6-deficient rabbits or their serum have been used to study the role of complement in many immunological reactions in vivo and i,nvitro. Whereas in normal rabbit serum incubated with antigen-antibody complexes or aggregated IgG, lysolecithin was elaborated, none was found under similar conditions when CG-deficient rabbit serum was used (I. Haupt et al., 1963). This suggests that late-acting complement components are necessary for lysolecithin formation-a reaction implicated in the mechanism of complement-mediated cell lysis. If limited antibody was used in eliciting a passive Arthus reaction, none was observed in C6-deficient rabbits (K. Rother et al., 1964a). If partially purified “C’3” was administered, this reactivity was restored. On the other hand, the active Arthus reaction was normal in deficient animals. In a subsequent study (Biro, 1966), it was observed that both the active and passive Arthus reactions were incomplete in C6-deficient rabbits in that the hemorrhagic necrosis characteristic of the normal reaction was absent and leukocytic infiltration was less. With respect to the status of delayed hypersensitivity, in one study it was found to be possibly absent (Volk et al., 1964), whereas in another it was completely intact ( Biro, 1966) in C6-deficient animals. A similar apparent discrepancy was obtained with respect to allograft rejection. This function was observed to be impaired (Volk et al., 1964; U. Rother et al., 1967) and entirely normal (Biro, 1966) in separate studies of C6-deficient rabbits. Rabbit serum deficient in C6 was devoid of bactericidal activity for Salmonella typlai (K. Rother et al., 1964b), but this defect was corrected by the addition of normal rabbit serum heated at 56°C or of partially purified “C’3.” Restoration of this activity by purified C6 has not been reported. The clearance of intravenously administered heat-killed S . typhi, with or without prior incubation of the organisms with antibody, was identical in C6-deficient and normal rabbits (K. Rother and Rother, 19f35). The enhancement of phagocytosis in vitro of antibody-sensitized erythrocytes (K. Rother et al., 1966a) or pneumococci (Johnston et al., 1969) was normal in CG-deficient rabbit serum. These in vitro findings support the observation that C6-deficient rabbits are as hardy under laboratory conditions as are heterozygous deficient animals ( Lachmann, 1970) . Nephrotoxic serum nephritis induced with antibody s&cient to produce only second-stage disease was identical in C6-deficient and normal rabbits (K. Rother et al., 1966b, 1967). This observation suggests that
284
CHESTER A. ALPER AND FRED S. ROSEN
in the development of this experimental renal disease the components C 6 9 are not required. In both normal and C6-deficient rabbits, intravenously administered aggregate-free IgG was slowly cleared without splenic sequestration and induced immunological tolerance. Aggregated IgG given similarly was cleared rapidly by the spleen and resulted in antibody formation in both normal and deficient animals (Biro and Garcia, 1965). Chemotaxis for leukocytes could be generated in normal rabbit serum but not in C6-deficient rabbit serum, on incubation with antigen-antibody aggregates, zymosan, or aggregated IgG (Ward et al., 1965). Addition of purified C6 to the deficient serum restored chemotactic activity generation. These findings supported the hypothesis that both C5 and C6 are necessary for the generation of chemotaxis in rabbit serum. Anaphylatoxin generation and histamine release from rabbit blood cells were similar in normal and C6-deficient rabbit serum, suggesting that only complement components through C5 are necessary for this function (Giertz et al., 1964). The disappearance half-time (final slope T % )of rabbit C6 was about 30 hours as determined from the infusion of normal rabbit plasma into a C6-deficient rabbit (Biro and Ortega, 1966) and the measurement of C6 by restoration of hemolytic activity to the deficient serum. Rabbit serum deficient in C6 serves as an ideal reagent for the detection of C6 functional activity and has been extensively used as such. For example, this material was highly useful in the characterization and isolation (K. Rother et al., 1966a; Biro and Ortega, 1966; Lachmann, 1970) of both rabbit and human C6. It was used to detect C6 synthesis in vitm by a minimal deviation hepatoma culture consisting only of hepatocytes and by cultures of normal liver tissue, of macrophages, and possibly of splenic tissue ( K . Rother et al., 1968; U. Rother et al., 1968). XII. Miscellaneous
A. POLYMORPHISM OF HUMANC4 Inherited structural polymorphism in other complement components has been sought but not yet definitely established. Using antigen-antibody crossed electrophoresis, differences in C4 patterns among individual sera were found (Rosenfeld et al., 1969). These patterns were constant in samples taken from any individual over a prolonged period. Ten distinct constellations of electrophoretic C4 patterns were recognized but these could not be explained on a genetic basis in family studies. Nevertheless, in fifty-two maternal-cord serum pairs, nine were found to be discordant (Bach et al., 1971), suggesting that C4 does not cross the
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
285
placenta. Despite resolution of C4 into two bands on acid-urea starch gel electrophoretic analysis (Azen and Cooper, 1971)) no genetic basis for electrophoretic differences could be demonstrated.
B. POLYMORPHISM OF GUINEA PIG C2 On isoelectric focusing of human and guinea pig C2 in serum and pseudoglobulin, it was observed that C2 was bimodal ( P I 5.6 and 5.2) in six guinea pig sera, whereas in two guinea pig sera and in all human sera examined C2 was iinimodal and had a pZ of 5.6 (Colten et al., 1970).
POLYMORPHISM OF HUMAN C. INHERITED STRUC~URAL GLYCINE-RICH /3-GLYCOPROTEIN Glycine-rich /3-glycoprotein ( Bocnisch and Alper, 1970b) is probably complement-related for reasons cited earlier and because in whole serum it undergoes rapid fragmentation on incubation with antigen-antibody aggregates, zyniosan, or endotoxin ( Alper et al., 1971b). This protein was highly polymorphic in whole human serum, forming at least five bands in all sera tested. On studying sera from families as well as from unrelated persons, it was found that the polymorphism could be explained on the basis of a four-allele autosomal codominant system. The four genes were designated GbS,GbF, GbS1, and GbF1. Genes GbS and GbF were common in all populations tested, whereas GbF1 has thus far only been found in Negroes and GbS1in Caucasians. The electrophoretic appearance of the protein and the nature of the polymorphism suggested that GBG was a tetramer consisting of two kinds of subunits in an unequal ratio controlled by nonallelic loci. The polymorphism appeared to reside in a third moiety controlled by a third locus (Alper et al., 1971b). D. DEFICIENCY OF C l q
IN
AGAMMAGLOBULINEMIA
Miiller-Eberhard and Kunkel initially reported ( 1961) that some agammaglobulinemic sera were low in Clq. Subsequently, Kohler and Miiller-Eberhard (1968) were able to correlate yG levels with C l q titers in the sera of a variety of patients with immunoglobulin abnormalities. This correlation has not been confirmed by Gewurz et al. (1968) in a study of an equally large number of patients with defects in y-globulin synthesis. OConnell et al. (1966, 1967) reported absence of C l q from the serum of an infant with severe combined immunodeficiency (Swisstype or lymphopenic agammaglobulinemia) . Jacobs et al. ( 1968) reported an additional case with C l q deficiency. Gewurz et d.(19sS) have found 3 infants with the autosomal recessive form to be deficient in Clq, whereas 2 patients with the X-linked recessive form had normal Clq.
286
CHESTER A. ALPER AND FRED S . ROSEN
The implication that a defective or incomplete C1 molecule is synthesized by these infants is borne out by the findings of normal Cls levels functionally ( O’Connell et al., 1967) and inimunochemically (Stroud et al., 1970).
REFERENCES Abramson, N., Alper, C. A., Rosen, F. S., and Jandl, J. H. (1969). Clin. Res. 17, 318 (abstr.). Abramson, N., Alper, C. A., Lachniann, P. J., Rosen, F. S., and Jandl, J. H. (1971). J. Immunol. 107, 19. Alper, C. A. (1970). Protides Biol. Fluids, Proc. Colloq. 17, 295. Alper, C. A,, and Propp, R. P. ( 1967). Clin. Res. 15, 291 (abstr.). Alper, C. A., and Propp, R. P. (1968). J. Clin. Inoest. 47, 2181. Alper, C. A., and Rosen, F. S. (1967). J. Clin. Inuest. 46, 2031. Alper, C. A., and Rosen, F. S. (1971). J. Clin. Inoest. 50, 324. Alper, C. A., Johnson, A. M., Birtch, A. G., and Moore, F. D. (1969a). Science 163, 286. Alper, C. A., Propp, R. P., Klemperer, M. R., and Rosen, F. S. (196913). J. Clin. Invest. 48, 553. Alper, C. A., Abramson, N., Johnston, R. B., Jr., Jandl, J. H., and Rosen, F. S. (1970a). N . Engl. J. A4ed. 282, 349. Alper, C. A., Abramson, N., Johnston, R. B., Jr., Jandl, J. H., and Rosen, F. S. (1970b). J. Clin. Znoest. 49, 1975. Alper, C. A., Robin, N. I., and Refetoff, S. (1971a). J. Immunol. 107, 96. Alper, C. A., Boenisch, T., and Watson, L. (1971b). J. Immunol. 107, 323 (abstr.). Austen, K. F., and Sheffer, A. L. (1965). N . Engl. J. Med. 272, 649. Azen, E. A,, and Cooper, N. R. ( 1971). Clin. Res. 19, 437 (abstr.). Azen, E. A., and Smithies, 0. (1968). Science 162, 905. Azen, E. A., Smithies, O., and Hiller, 0. (1969). Biochem. Genet. 3, 215. Bach, S., Ruddy, S., MacLaren, J. A,, and Austen, K. F. (1971). Submitted for publication. Barth, R. F., and Carroll, G. F. (1970). J. Immunol. 104, 522. Becker, E. L., and Kagen, L. (1964). Ann. N . Y. Acad. Sci. 116, 866. Becker, W., Schwick, H. G., and Storiko, K. (1970). Clin. Chem. 15, 649. Ben-Efraim, S., and Cinader, B. (1964). J. E x p . Med. 120, 925. Biro, C. E. (1966). Immunology 10, 563. Biro, C. E., and Garcia, G. (1965). Immunology 8, 411. Biro, C. E., and Ortega, M. L. (1966). Arch. Znst. Cardiol. Mex. 36, 166. Blum, L., Pillemer, L., and Lepow, I. H. (1959). Z . Immunitaetsforsch. Exp. Ther. 118, 34. Boenisch, T., and Alper, C. A. (1970a). Biochim. Biophys. Acta 214, 135. Boenisch, T., and Alper, C. A. (1970b). Biochim. Biophys. Acta 221, 529. Bokisch, V. A., Miiller-Eberhard, H. J., and Cochrane, C. G. (1969). J. Exp. Med. 129, 1109. Borsos, T., and Cooper, M. (1961). Proc. Soc. E x p . Biol. Med. 107, 227. Burdon, K. L., Queng, J. T., Thomas, 0. C., and McCovern, J. P. (1965). J. Allergy 36, 546. Caren, L. D., and Rosenberg, L. T. (1965). Immunology 9, 359. Caren, L. D., and Rosenberg, L. T. (1966). J. E x p . Med. 124, 689.
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
287
Caren, L. D., MacKnight, J. F., and Rosenberg, L. T. (1965). Fed. Proc., Fed. Amer. Soc. Exp. Biol. 24, 698 (abstr.). Carpenter, C. B., Ruddy, S., Shehadeh, I. H., Miiller-Eberhard, H. J., Merrill, J. P., and Austen, K. F. (1969). J. Clin. Invest. 48, 1495. Cinader, B., and Dubiski, S. (1963). Nature (London) 200, 781. Cinader, B., and Dubiski, S. (1968). J. Immunol. 101, 1236. Cinader, B., Dubiski, S., and Wardlaw, A. C. (1964). J. E x p . Med. 120, 897. Cinader, B., Jeejeebhoy, H. F., Koh, S. W., and Rabbat, A. C. (1971). I. E x p . Med. 133, 81. Coca, A. F. (1920). Proc. SOC.Exp. Biol. Med. 18, 71. Colten, H. R., Borsos, T., and Rapp, H. J. (1970). Immunology 18, 467. Cooper, N . R., ten Bensel, R., and Kohler, P. F. (1968). J. lmmunol. 101, 1176. Crisler, C., and Frank, M. M. (1965). Fed. Proc., Fed. Amer. SOC. E x p . Biol. 24, 447 ( abstr. ) . Dias da Silva, W., and Lepow, I. H. (19G7). J. Exp. Med. 125, 921. Donaldson, V. H. (1966). J. Lab. Clin. Med. 68, 369. Donaldson, V. H. (1968a). J. Appl. Physiol. 25, 416. Donaldson, V. H. (196813). 1. Exp. Med. 127, 411. Donaldson, V. H. (1970). Ser. Haematol. 3, 39. Donaldson, V. H., and Evans, R. R. ( 19G3). Amer. I. Med. 35, 37. Donaldson, V. H., and Pensky, J. (1970). J. Immunol. 104, 1388. Donaldson, V. H., and Rosen, F. S. (1964). J . Clin. Invest. 43, 2204. Donaldson, V. H., and Rosen, F. S. (1966). Pediatrics 37, 1017. Donaldson, V. H., Ratnoff, 0. D., and Rosen, F. S. (1965). J. Lab. Clin. Med. 66, 867 ( abstr. ). Donaldson, V. H., Ratnoff, 0. D., Dias da Silva, W., and Rosen, F. S. (1969). J. Clin. Invest. 48, 642. Donaldson, V. H., Merler, E., Rosen, F. S., Willms-Kretschmer, K., and Lepow, I. H. (1970). J. Lab. Clin. Med. 76, 986 (ahstr.). Dubiski, S., and Cinader, B. (1966). Proc. SOC. Exp. Biol. Med. 122, 775. Ecker, E. E. (1921). J . Infec. Dis. 29, 611. Ellman, L., Green, I., and Frank, M. (1970). Science 170, 74. Ellman, L., Green, I., Judge, F., and Frank, M. (1971). J. Exp. Med. 134, 162. Erickson, R. P., Tacliibana, D. K., Herzenberg, L. A., and Rosenberg, L. T. (1963). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 22, 612 (abstr.). Erickson, R. P., Tachibana, D. K., Herzenberg, L. A., and Rosenberg, L. T. (1964). 1. lmmunol. 92, 611. Frank, M. M., and Rosen, F. S. (1971). Unpublished observations. Frank, M. M., May, J., Gaither, T., and Ellman, L. (1971). J . Exp. Med. 134, 176. Geserick, G., Rose, M., Blaszczyk, J., and Iwanow, R. (1970a). 2. ZmmunitaetsAllergieforsch. 140, 50. Geserick, G., Rose, M., and Schnitzler, S. ( 1970b). 2. Immunitaets- Allergieforsch. 140, 54. Gewurz, H., Pickering, R. J., Muschel, L. H., Mergenhagen, S. E., and Good, R. A. ( 1966). Lancet 2, 356. Gewurz, H., Pickering, R. J., Christian, C. L., Snyderman, R., Mergenhagen, S. E., and Good, R. A. (1968). Clin. Exp. Immunol. 3, 437. Giertz, H., Reutter, W., and Schmutzler, W. (1964). Nature (London) 202, 1346. Gigli, I., Ruddy, S., and Austen, K. F. (1968). J . lmmunol. 100, 1154. Glynn, A. A., and Medhurst, F. A. (1967). Nature (London) 213, 608. Coodkofsky, I., Alper, C. A., and Lepow, I. H. (1971). Unpublished observations.
288
CHESTER A. ALPER AND FRED S. ROSEN
Haines, A. L., and Lepow, I. H. (1964). J. lmmunol. 92, 456. Harpel, P. C. (1970). 1. lmmitnol. 104, 1024. Haupt, H., and Heide, K. (1965). Clin. Chim. Acta 12, 419. Haupt, I., Fischer, H., Rother, U., and Rother, K. (1963). Nature (London) 200, 686. Herzenberg, L. A., Tachibana, D. K., Herzenberg, L. A,, and Rdsenberg, L. T. (1963). Genetics 48, 711. Howie, J. B., and Helyer, B. J. (1965). Ann. N . Y. Acad. Sci. 124, 167. Hunter, T., and Munro, A. J. (1969). Nature (London) 223, 1270. Hyde, R. R: ( 1923). J. lmmunol. 8, 267. Hyde, R. R. ( 1924). Amer. J. H y g . 4, 62. Hyde, R. R. (1927). Amer. J. Hyg. 7, 619. Hyde, R. R. (1932). kmer. J. H y g . 15, 824. Jacobs, J. C., de Capon, A., McCivray, E., Morse, H. H., Schullinger, J. N., Blanc, W. A. Heird, W. C., Miller, 0. J., Rossen, R. D., and Walzer, R. A. (1968). Lancet 1, 499. Johnson, A. M., Alper, C. A,, Rosen, F. S., and Craig, J. M. (1971). Science 173, 553. Johnston, R. B., Jr., Klemperer, M. R., Alper, C. A., and Rosen, F. S. (1969). J. Exp. Med. 129, 1275. Johnston, R. B., Jr., Anderson, P., Rosen, F. S., and Smith, D. H. (1971). Submitted for publication. Kagen, L. J., and Becker, E. L. (1963). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 22, 613 (abstr.). Klemperer, M. R. (1969). J. lmmunol. 102, 168. Klemperer, M. R., Gotoff, S. P., Alper, C. A., Levin, A. S., and Rosen, F. S. (1965). Pediatrics 35, 765. Klemperer, M. R., Woodworth, H. C., Rosen, F. S., and Austen, K. F. (1966). J. Clin. Invest. 45, 880. Klemperer, M. R., Austen, K. F., and Rosen, F. S. (1967). J. lmmunol. 98, 72. Klemperer, M. R., Donaldson, V. H., and Rosen, F. S. (1968). J. Clin. Invest. 47, 604. Klemperer, M. R., Rosen, F. S., and Donaldson, V. H. (1969). J. Clin. Invest. 48, 44a ( abstr. ). Kohler, P. T., and Miiller-Eberhard, H. J. (1968). Science 183, 474. Lachmann, P. J. (1970). Protides Biol. Fluids, Proc. Colloq. 17, 301. Lachmann, P. J., and Miiller-Eberhard, H. J. (1968). J. lmmunol. 100, 691. Landerman, N. S. (1962). J. Allergy 33, 133. Landerman, N. S., Webster, M. E., Becker, E. L., and Ratcliffe, H. E. (1960). J. Allergy 33, 330. Laurel], A.-B., Lundh, B., Malmquist, J., and Siboo, R. (1966). Clin. Exp. lmmunol. 1, 13. Laurel], A.-B., Lindegren, J., Malmros, J., and MHrtensson, H. (1969). Scand. J . 'Clin. Lab. Invest. 24, 221. Laurell, C.-B., Laurell, S., and Skoog, N. (1956). Clin. Chem. 2, 99. Leon, M. A., and Lepow, I. H. (1962). Immunology 5, 235. Lepow, I. H., and Leon, M. A. (1962). Immunology 5, 222. Lepow, I. H., Ratnoff, 0. D., Rosen, F. S., and Pillemer, L. (1956). PTOC. SOC. Exp. Biol. Med. 92, 32. Lepow, I. H., Willms-Kretschmer, K., Patrick, R. A., and Rosen, F. S. (1970). AmeT. J . Puthol. 61, 13.
GENETIC ASPECTS OF THE COMPLEMENT SYSTEM
289
Levy, L. R., and Lepow, I. H. (1959). Proc. SOC.E x p . Biol. M e d . 101, 608. Levy, N. L., and Ladda, R. L. (1971). Nature (London) 229, 51. Lindberg, L. H., and Rosenberg, L. T. (1968). J. Immunol. 100, 34. Lundh, B., Laurell, A.-B., Wetterqvist, H., White, T., and Cranerus, G. (1968). Clin. E x p . Immunol. 3, 733. Miller, M. E., and Nilsson, U. R. (1970). N . Engl. J. Med. 282, 354. Miller, M. E., Seals, J., Kaye, R., and Levitsky, L. C. (1968). Lancet 2, GO. Moore, H. D. (1919). J. Irnmunol. 4, 425. Muller-Eberhard, H. J., and Kunkel, H. G. (1961). Proc. SOC. E x p . Biol. Med. 106, 291. Miiller-Eberhard, H. J., Nilsson, U. R., Dalmasso, A. P., Polley, M. J., and Calcott, M. A. (1966). Arch. Pathol. 82, 205. Nelson, R. A., Jr., and Biro, C. E. (1968). Irnmrcnologr~14, 527. Nilsson, U. R. ( 1970). Fed. Proc., Fed. A’mcr. SOC. E x p . B i d . 29, 433 (abstr.). Nilsson, U. R., and Miiller-Eberhard, H. J. (1965). Fed. Proc., Fed. Amer. SOC. E x p . Biol. 24, 620 (abstr.). Nilsson, U. R., and Mhller-Eberhard, H. J. (1967). J. E x p . Med. 125, 1. O’Connell, E. J., Enriquez, P., Linman, J. W., Gleich, G. J., and McDuffie, F. C. (1966). J. Pediat. 69, 681. O’Connell, E. J., Enriquez, P., Linnian, J. W., Cleich, G. J., and McDuffie, F. C. (1967). J. Lab. Clin. A4ed. 70, 715. Okada, N., and Rosenberg, L. T. (1964). Bacteriol. Proc. p. 141. Osler, W. (1888). Amer. J. Merl. Sci. 95, 363. Pensky, J., and Schwick, H. G. (1969). Science 163, 698. Pensky, J., Levy, L. R., and Lepow, I. 11. (1961). J . Biol. Clzem. 236, 1674. Pensky, J., Hinz, C . F., Jr., Todd, E . W., Wedgwood, R. J., Boyer, J. T., and Lepow, I. H. (1968). J. Immunol. 100, 142. Phillips, M. E., Rother, U. A., Rother, K. O., and Thorbecke, G. J. (1969). Immunology 17, 315. Pickering, R. J., Cewurz, H., Kelly, J. R., and Good, R. A. (1968). Clin. E x p . Immunol. 3, 423. Polley, M. J. (19%). Science 161, 1149. Propp, R. P., and Alper, C. A. (1968). Science 162, 672. Ratnoff, 0. D., and Lepow, I. H. (1963). J. E x p . Med. 118, 681. Ratnoff, 0. D., and Naff, G. B. (1967). J. E x p . Med. 125, 337. Ratnoff, 0. D., Pensky, J., Ogston, D., and Naff, C. B. (1969). J. E x p . Med. 129, 315. Ropartz, C., Rivat, L., Rousseau, P.-Y., Debeaux, P., and Steinbuch, M. (1965). Proc. Congr. Int. SOC. Blood Transf.,loth, 1964 (abstr.), p. 383. Rose, M., and Geserick, G. (1969). Acta Biol. Med. Ger. 23, 351. Rosen, F. S., Charache, P., Pensky, J., and Donaldson, V. (1965). Science 148, 957. Rosen, F. S . , Alper, C. A., Pensky, J., Klempercr, hl. R., and Donaldson, V. H. ( 1971). J. Clin. Invest. (in press). Rosenberg, L. T., and Tachibana, D. K. (1962). J. Immunol. 89, 861. Rosenberg, L. T., and Tachibana, D. K. ( 19G8). J. Immunol. 101, 817 (abstr.). Rosenfeld, S. I., Ruddy, S., and Austen, K. F. (1969). J. Clin. Invest. 48, 2283. Rother, K., and Rother, U. (1965). Proc. Soc. Exp. Biol. Med. 119, 1055. Rother, K., Rother, U., and Schindera, F. (1964a). 2. Immrcnitaetsforsch. E x p . Ther. 126, 473. Rother, K., Rother, U., Petersen, K. F., Genisa, D., and Mitze, F. (1964b). J. Immunol. 93, 319.
290
CHESTER A. ALPER AND FRED S. ROSEN
Rother, K., Rother, U., Muller-Eberhard, H. J., and Nilsson, U. R. (1966a). J. E x p . Med. 124, 773. Rother, K., Vassalli, P., Rother, U., and McCluskey, R. T. (1966b). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 25, 309 (abstr.). Rother, K., Rother, U., Vassalli, P., and McCluskey, R. T. (1967). J . lmmunol. 98, 965. Rother, K. O., Rother, U. A., Phillips, M. E., Goetze, O., and Thorbecke, G. J. (1968). J. lmmunol. 101, 814 (abstr.). Rother, U., and Rother, K. (1961). 2. Immunitaetsforsch. E x p . Ther. 121, 224. Rother, U., Ballantyne, D. L., Jr., Cohen, C., and Rother, K. (1967). J. E x p . Med. 126, 565. Rother, U., Thorbecke, G. J., Stecher, V. J., Hurlimann, J., and Rother, K. (1968). Immunology 14, 649. Ruddy, S., and Austen, K. F. (1967). J. lmmunol. 99, 1162. Ruddy, S., Klemperer, M. R., Rosen, F. S., Austen, K. F., and Kumate, J. (1970). Immunology 18, 943. Schultze, H. E., Heide, K., and Haupt, H. (1962). Naturwissenschaften 49, 133. Sheffer, A. L., Craig, J. M., Willms-Kretschmer, K., Austen, K. F., and Rosen, F. S. (1971). J. Allergy (in press). Shin, H. S., Smith, M. R., and Wood, W. B., Jr. (1969). J. E x p . Med. 130, 1229. Siboo, R., and Laurell, A.-B. (1965). Acta Pathol. Microbiol. Scand. 65, 413. Silverstein, A. M. (1960). Blood 16, 1338. Stiffel, C., Biozzi, G., Monton, D., Bouthillier, Y., and Decreusefond, C. (1964). J . Immunol. 93, 246. Stroud, R. M., Nagaki, K., Pickering, R. J., Cewurz, H., Good, R. A., and Cooper, M. D. (1970). Clin. E x p . Immunol. 7, 133. Tachibana, D. K., and Rosenberg, L. T. (1964). Folia Microbiol. (Prague) 9, 129. Tachibana, D. K., and Rosenberg, L. T. (1966). J. lmmunol. 97, 213. Tachibana, D. K., Ulrich, M., and Rosenberg, L. T. (1963). J. Immunol. 91, 230. Talamo, R. C., Haber, E., and Austen, K. F. (1969). J. Lab. Clin. Med. 74, 816. Teisberg, P. ( 1970). Vox Sang. 19, 47. Teisberg, P. ( 1971 ). Personal communication. Terry, W. D., Borsos, T., and Rapp, H. J. ( 1964). J. lmmunol. 92, 576. Unanue, E. R., Mardiney, M. R., and Dixon, F. J. (1967). J. Immunol. 98, 609. Volk, H., Mauersberger, D., Rother, K., and Rother, U. (1964). Ann. N. Y. Acad. Sci. 120, 26. Ward, P. A., Cochrane, C. G., and Miiller-Eberhard, H. J. (1965). J. E x p . Med. 122, 327. West, C. D., Davis, N. C., Forristal, J., Herbert, J., and Spitzer, R. (1966). J. lmmunol. 96, 650. Wieme, R. J. (1965). “Agar Gel Electrophoresis,” p. 238. Elsevier, Amsterdam. Wieme, R. J., and Demeulenaere, L. (1967). Nature (London) 214, 1042. Wieme, R. J., and Segers, J. (1968). Nature (London) 220, 176. Wieme, R. J., Demeulenaere, L., and Segers, J. (1968). Protides Biol. Fluids, Proc. Colloq. 15, 499. Willms-Kretschmer, K., Donaldson, V. H., Klemperer, M. R., and Rosen, F. S. (1970). Unpublished observations,
The Immune System: A Model for Differentiation in Higher Organisms1
. H O O D AND J . PRAHL
1
Division of Biology. California lnsfitufe o f Technology. Pasadena. California
I . Introduction . . . . . . . . . . . I1 Immunoglobulin Systems . . . . . . . . 111. Structural. Genetic. and Cellular Patterns . . . . . A General Immunoglobulin Structure . . . . . B . Variable and Constant Regions . . . . . . C. Immunoglobulin Families . . . . . . . D . One Cell+One Antibody . . . . . . . E . variable-Region Subgroups . . . . . . . F . Rabbit Allotypy . . . . . . . . . IV . A Genetic Mechanism for Differentiation: Two Genes -+ One Polypeptide Chain . . . . . . . . . A One Gene or Two? . . . . . . . . B. Level of Joining . . . . . . . . . C . A Mechanism for Differentiation . . . . . V. Evolution of Immunoglobulin Variable and Constant Genes . A . Homology Units and Gene Duplication . . . . B . The Domain Hypothesis . . . . . . . C . Evolution of Immunoglobulins from Membrane Molecules VI . Theories of Antibody Diversity . . . . . . A Mechanisms for Information Storage . . . . . B . Variable-Region Patterns . . . . . . . C. Evolution of Immunoglobulin Variable Genes . . . D . Other Multiple Gene Systems . . . . . . E . Evolutionary Mechanisms for Multiple Gene Systems . F. Problems of Control . . . . . . . . G. Summary: Immoglobulin Patterns and Theories of Antibody Diversity . . . . . . . . H . The Germ Line Theory . . . . . . . VII . Concluding Remarks . . . . . . . . . References . . . . . . . . . . . Addendum . . . . . . . . . . .
.
.
.
.
. . . . . . . . .
. . . . . . . . . . . . . . . . . . .
.
.
. . . .
. . .
.
.
. . .
. . . . . .
. . . .
. .
. . .
.
291 296 298 298 299 300 301 301 304 305 305 308 310 311 311 313 313 314 314 315 323 329 331 335 339 342 344 345 351
.
I Introduction
The vertebrate immune system shares certain general properties with other complex systems in higher organisms and. accordingly. may serve as an ideal model for studying aspects of differentiation and evolution
.
'This work was supported in part by U S. Public Health Service Grant GM 06965 and by National Science Foundation Grant GB 27605.
291
292
L. HOOD AND J. PRAHL
in eukaryotes. We shall discuss the general patterns which have emerged from genetic, molecular, and cellular studies, making a special effort not to get lost in the welter of details for which the immune system is justly famous. Where possible we shall attempt to discuss various alternative explanations of these patterns and indicate what, we feel, may be the most reasonable alternatives, with the reservation that in view of the current state of flux in mammalian genetics, one’s “reasonable alternatives” may change in relatively short periods of time. The immune system of vertebrates can respond to an immense number of antigenic stimuli through the elaboration of specific antibody molecules ( Landsteiner, 1947). This functional diversity of antibody molecules is determined by a corresponding diversity in their amino acid sequence (Haber, 1968). Each separate antibody sequence must, correspondingly, be encoded by a separate antibody gene. The specificity and range of the immune response raise two questions which are central to modern immunology and, in our view, central to the problem of differentiation in higher organisms: (1) How is this information stored?2 ( 2 ) How is this information selectively expressed? Let us consider in more detail the implications of each of these questions. There are two general possibilities for information storage in the immune system and in other complex systems. First, all of the information (i.e., antibody genes) may preexist in the germ line; a separate antibody gene may be present in the germ line for each antibody polypeptide chain the organism can elaborate (Szilard, 1960; Dreyer and Bennett, 1965; Hood and Talmage, 1970). Since all of the information is contained in the germ line, proposals of this general nature are designated germ line theories. Second, information may be generated from a limited number of germ line genes through some type of mutational or recombinational process which operates during somatic differentiation (Lederberg, 1959; Edelman and Gally, 1967; Smithies, 1967; Cohn, 1968). In this view, each individual starts with a few basic antibody genes and elaborates during his lifetime the immense array of different antibody genes. Since antibody diversity is generated during somatic differentiation, such theories are given the general designation somatic theories. As we shall establish subsequently, the analysis of various immunoglobulin patterns has placed important constraints on each of these two general models of antibody diversity. The question of how specific information is expressed in a given cell
‘ The term information storage is compatible with the storage of all variable region information in the germ line (germ line theory) or with the storage of one or a few basic variable genes of which the inforination content is modified by mutation or recombination during somatic differentiation (somatic theory).
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
293
is, of course, a paradigm of differentiation. The genome of each vertebrate cell may contain all of the library of information that is required to construct a new organism (Gurdon and Woodland, 1970). Yet, in the differentiated individual cell, only a miniscule subset of this information is expressed. The immune system of vertebrates appears capable of generating a large variety of different antibody molecules (perhaps a million). How then does the immune system respond to a particular antigenic stimulus with the elaboration of specific antibody molecules? In the immune system this question of the expression of specific information can be considered at two separate levels. First, each antibodyproducing cell appears to synthesize just a single type of antibody molecule (see Section III,D), thus expressing only one unit of information. Similar types of information restriction are also operant in other complex differentiating systems. The critical question which we shall consider later is, How does a single immunocyte restrict itself to the synthesis of a single type of antibody molecule? Second, there must be a mechanism for amplification of the immune response so as to expand the population of cells making antibody that is complementary to a given antigenic stimulus. Presumably this amplification is achieved by the simple process of antigenic selection. That is, antigen enters into the vertebrate immune system where it circulates until it interacts with those cells bearing a complementary receptor unit ( antibody-like molecule) at their surface. The antigen, upon interaction with this receptor, triggers a cellular response initiating a twofold process of cellular proliferation and synthesis of the specific antibody molecules. This general phenomenon has been designated the clonal selection hypothesis (Burnet, 1959). Thus in the immune system there are mechanisms for the expression of one unit of information (one antibody molecule) per cell and for the amplification of cells expressing appropriate information which are available for analysis at the molecular and cellular levels. Before we consider in detail the specific patterns of the immune system which afford insights into various genetic and evolutionary mechanisms, a bricf historical analysis on the evolution of thinking about the immune system provides a useful perspective. One can reasonably divide thc intellectual analysis of the immunc system into four general and successively ovcrlapping periods. Initially investigators were very much concerned with explaining the specificity of antibody molecules ( 1890-present ) . Then followed an awareness of some of the intriguing biological properties of the immune system including immunological memory and the failure to react with self-antigcns ( 1949-present). The third period of analysis evolved as the result of the availability of the detailed amino acid sequence information from immunoglobulin mole-
294
L. HOOD AND J. PRAHL
cules and might be properly designated the era of molecular immunology (196Ppresent). Each of the first three modes of thought was, in part, characterized by the feeling that the immune system was unique and presented problems quite different from those encountered in other complex systems. The fourth period of analysis suggested that the immune system is an excellent model system for the study of differentiation in higher organisms (1967-present; see Dreyer et d.,1967; Cohn, 1970; Edelman, 1971b) and that problems of information storage and expression are common to many other systems. However, no other complex system is currently so readily accessible to detailed analyses both at the cellular and molecular levels. In historical terms, Paul Ehrlich (1900) was among the first to consider in detail the problem of immunological specificity. H e envisioned immunocytes as pluripotential, possessing diverse “side-chain groups” which extended from the surface of the cell into the environment. When antigen was introduced into the organism, it interacted with complementary side-chain groups and caused them to be released from the cell. This release triggered the synthesis of more of that particular side group in the corresponding cell. This proposal was the first selective theory of antibody diversity and assumed that the role of the antigen was merely to select and stimulate the synthesis of preformed antibody molecules. This selective view, however, was overshadowed in the twenties and thirties by Landsteiner’s observations ( see Landsteiner, 1947) which demonstrated that the immune system had an exquisite specificity and a broad range of responses. It seemed unreasonable to suggest that the vertebrate organism could have such an enormous library of preformed antibody molecules as appeared to be required by the selective theory. Rather it seemed more likely that the antigen must instruct the immune system and in some fashion provide directly the information required for specific antibody synthesis. This view led to the instructionistic theories of Horowitz (Breinl and Horowitz, 1930), Alexander (1932), and others which appeared in the early thirties. Linus Pauling (1940; also see Karush, 1962) reduced the instructionistic interpretations to molecular terms by hypothesizing that the antigen served as a template which permitted the antibody polypeptide chain to fold about it in a complementary fashion, thus providing direct instruction at the polypeptide level. It appeared at that time that the problem of antibody specificity had been adequately resolved. However, the instructionistic theory left unexplained certain of the general biological properties of the immune response. For example, the problem of immunological memory was disturbing for the instructionists, in that it implied that antigen must persist in immune system
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
295
for relatively long periods of time. How antigen could persist in the face of a declining immune response was not obvious. In order to avoid this difficulty, Burnet (1956) suggested that antigen could induce permanent changes in the deoxyribonucleic acid (DNA) of the organism ( antibody genes in contemporary terms ), which thereafter synthesized antibody complementary to this antigen ( also see Schweet and Owen, 1957). This was antigenic instruction at the level of DNA. Jerne (1955) set forth the first modern selectionistic theory of antibody diversity, postulating that antibodies of all potential specificities were contained in the serum of each individual. An antigen, upon interaction with its complementary antibody, formed an antigen-antibody complex which was then recognized by the immunocyte. By an unspecified mechanism, this complex stimulated the synthesis of the specific antibody molecule in immunocytes. Jerne further sought to explain tolerance, that is, the failure to react against self, by suggesting that antibodies that reacted against self were purged from the vertebrate organism at or shortly after birth. This selectionistic approach was further expanded by Burnet ( 1957, 1959) and Talmage (1957) who suggested that the “unit of selection” was the immunocyte itself rather than the humoral antibody molecule. The immune system of the vertebrate organism was envisaged as the composite of some 1011-1012immunocytes, each with the potential of expressing a unit of information (i.e., synthesizing a single antibody molecule). The antigen served merely as a selective agent to interact at the cell surface with complementary receptor molecules and to initiate antiljody synthesis and proliferation of the complementary clones of antibody-producing cells. Antigenic instruction at the level of the polypeptide chain became unlikely with the advent of studies by Anfinsen and others (Epstein et al., 1963) which indicated that the primary amino acid sequence of a protein determines its three-dimensional configuration. Haber and Whitney extended this concept to antibody molecules by demonstrating that they could be completely unfolded and refolded in the absence of antigen and that nonspecific immunoglobulin would not yield complementary antibody if refolded in the presence of antigen (Haber, 1964; Whitney and Tanford, 1965). This concept has been supported by the amino acid sequence studies of immunoglobulins. Since it seemed unlikely that antigen could instruct protein or genomic DNA, the selectionistic theory of antibody diversity has been accepted as dogma from the early 1960s to the present. Prior to 1965, however, little could be said about the problems of
296
L. HOOD AND J. PRAHL
information storage and the specific expression of information. Recent amino acid sequence analysis of antibody molecules has contributed the most detailed insights into these problems. The strategy has been to compare a series of antibody niolecules at the level of protein sequences and nucleic acid sequences, looking for patterns which might impose constraints on the various genetic and evolutionary mechanisms proposed. Other patterns, of course, have also emerged from sophisticated modern studies combining serologic, cell biology, and genetic analyses. However, before we begin with the detailed discussion of these patterns, let us first consider the general systems from which this information has been derived. I I. I m mu noglobu lin Systems
Four general types of systems have been utilized in the genetic and structural analysis of immunoglobulins: normal immunoglobulins, myeloma proteins, heterogeneous antibodies, and restricted antibodies. Normal immunoglobulins are those which are isolated from the serum of normal animals that have not undergone any special immunization procedures. These imn~unoglobulinsare extremely heterogeneous by chemical criteria and generally contain no major types of antibody specificities ( Fahey, 1962; Fleischman, 1966). Presumably they reflect the past immunization history of the animal from which they are taken. The information derived from the use of normal immunoglobulins is severely limited by their extensive heterogeneity. Nature can expand a single molecular species of immunoglobulin through the neoplastic disease known as myelomatosis or multiple mycloma. Myelomatosis has been observed in a number of species (see Capra and Hurvitz, 1970), although it has been most thoroughly examined in mouse (Potter, 1967) and man (Snapper and Kahn, 1971). Although myelomatosis usually is a spontaneous process, it can be induced through the intraperitonal injection of mineral oil in two strains of inbred mice (NZB and BALB/c) (Potter, 1967; Warner, 1970). This “cancer,” involving the differentiated ininiunocyte, is characterized by the loss of the control mechanisms which normally govern cellular proliferation, generally leading to the repopulation of the lymphoid tissue with the progeny of a single clone (biclonal myelomas are rare). When the malignant clone has been derived from an inimunocyte secreting antibody of one molecular type, this single component can eventually comprise up to 9% of the normal serum immunoglobulin. Thus homogeneous immunoglobulins can readily be obtained which by genetic, chemical, and physical criteria appear to be representative of the normal immunoglobulin population ( Fahey, 1962; Kunkel, 1965; Cohen and Mil-
THE I M M U N E SYSTEM: A MODEL FOH DIFFERENTIATION
297
stein, 1967). In addition, myeloma proteins and macroglobulins have been found which appear capable of binding activity when tested against a varicty of antigens. Binding activities against y-globulin, red cells, streptococcal products, and lipoproteins have been known for a number of years (Kritzman et al., 1961). More recently a number of simpler chemical antigens have been added to the list. These include nucleic acid bases (Schubert et al., 1968), various derivatives of nitrophenol (Eisen et aZ., 1968; Schubert et al., 1968), and certain types of polysaccharidc (Potter and Leon, 1968; Zettervall, 1968; Cohn et al., 1969). Two aspects of this complementarity, however, are worth noting. First, the affinity of binding is generally quite low (Eisen et al., 1968; Parker and Osterland, 1970; Terry et al., 1970); second, a larger proportion of the population of myeloma proteins demonstrates binding activities against some of these simple antigens than would be statistically anticipated (Cohn et al., 1969). The first point suggests the “true” specificity of most of these proteins has not yet been found, if, indeed, they are all antibodies. One protein, MOPC 315, has affinity constants of lo7 and 5 x 10; liters mole-’, respectively, for dinitrophenyl ( D N P ) moiety and menadione, and is quite a respectable antibody molecule (Eisen et al., 1970). The second point raises the possibility that the myeloma process selects nonrandomly a subset of the normal immunocyte population. This point will be considered subsequently. Altei-natively, apparent binding activity may represent interactions of a more generalized nature unrelated to antibody activity (Parker and Osterland, 1970; Glazer, 1968). A third system studied has been that of heterogeneous antibody immunoglobulin which arises when the organism is exposed to an antigenic stimulus. Although the antibodies share a common specificity, they still appear to be relatively heterogeneous by chemical and genetic criteria (Cebra et al., 1968; Lark et aZ., 1965; Fleischman, 1966). Usually the affinity of the population of induced antibodies for the antigen is broad, with dissociation constants ranging from to lo-’” liters mole-’, although this distribution is strongly dependent upon the antigen employed and the mode and course of its administration (Eisen and Siskind, 1964; Siskind and Benacerraf, 1968). Just as the myeloma process expands out a single molecular species from a normal immunoglobulin population, special immunization procedures and antigens can be used to evoke a restricted or homogeneous antibody response in various animals. It has been noted that some antigens and special immunization procedures may lead to the appearance of predominantly one cIass of immunoglobulin response (restricted antibodies). This was found to be a characteristic of various carbohydrate
298
L. HOOD AND J. PRAHL
antibodies in man such as anti-A and anti-levan (Allen et d.,1964; Kunkel, 1965). Reccntly, it has been demonstrated in certain families of rabbits that a variety of polysaccharide antigens are capable of inducing antibody responses which are greatly rcstricted in their heterogeneity. On occasion they induce the production of a sillgle molecular species of antibody molecule ( Krause, 1970; Haber, 1970). Chemical verification of the restriction has been obtained by the purification and partial sequence analysis of several of the antibodies obtained, yielding unique primary amino acid sequences (Hood d al., 1970a; Jaton et al., 1970). This ability to elicit restricted responses also has been reported with other antigens (Nisonoff et al., 1970) and in other species (i.e., nurse shark; Clem and Leslic, 1971) thus enlarging the potential of the technique. One might envisage the use of the “restricted response” in a variety of vertebrate species to gcnerate homogeneous proteins for genetic and evolutionary studies. The gcnetic, structural, and ccll biology patterns that have emerged using each of these systems will be summarized in the next section. Ill. Structural, Genetic, a n d Cellular Patterns
A. GENERALIMMUNOGLOBULIN STRUCTURE The basic structure of the immunoglobulin molecule as illustrated in Fig. 1 is composed of two identical light ( L ) polypeptide chains (ca.
7
I W I I
L-chain H-choin
.OGY UNITS V-REGIONS C-REGIONS
Disulfides
Fc Domoin
FIG. 1. A model depicting the basic structure of the immunoglobulin (IgC) molecule. The hinge region of the heavy chnin is accessible to proteolytic enzymes which can be used to cleave the molecule into its respective Fah and Fc domains. The homology units are -110 residues in length with a centrally placed disulfide briJge of about 60 residues (see text). (Reprinted from Smith et al., 1971, with permission. )
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
299
22,500 daltons) and two identical heavy ( H ) chains (ca. 53,000-70,000 daltons ), linked by disulfide bonds and noncovalent interactions. The molecule folds into three compact domains which are connected by a hinge region in the heavy chain (Edelman and Gall, 1969). These three domains can be clearly delineated by electron microscopy (Valentine and Green, 1967; N. M. Green, 1969) and by X-ray crystallographic studies (Davies et al., 1971). A variety of proteolytic enzymes preferentially cleave the native molecule in or near this hinge region to give a single Fc and two Fab fragments (Cohen and Porter, 1964; Givol and DeLorenzo, 1968). Each of these domains appears to be concerned with discrete functions of the immunoglobulin molecule. The Fab fragments retain the antigen-binding properties of the whole antibody molecule, whereas the Fc fragment is concerned with the variety of general effector functions, such as complement fixation, placental transfer, and skin sensitization ( Cohen and Porter, 1964). The four-chain structure depicted in Fig. 1 is the most common type of immunoglobulin molecule found in higher organisms (IgG class) and is the basic subunit structure for all immunoglobulin classes. B. VARIABLEAND CONSTANT REGIONS
All immunoglobulin polypeptide chains can be divided into an amino terminal portion, the variable ( V ) region; and a carboxy terminal portion, the constant ( C ) region (Lennox and Cohn, 1967). The V regions for both light (V,) and heavy (V,) chains are generally about 110 amino acid residues in length, whereas the C regions appear to be 110 residues in length for light (C,) chains and 3-4 times this size for various heavy ( CJI)chains (Edelman et al., 1969; Putnam et al., 1971). The V and C regions were initially defined by sequence variation (or the absence of variation). For example, if the amino acid sequences from ten IgG molecules (see Fig. 1) were aligned to maximize sequence homology, the VL regions would show marked sequence variations as would the VH regions. The CL regions for a given chain type (or class) would be identical as would the C,, regions, apart from minor genetic variants. The V, regions generally differ from one another by 15 to 40 amino acid residues, whereas the V,, regions can differ even more (Smith et al., 1971; Milstein and Pink, 1970). The VL and VII regions show striking amino acid sequence homologies with one another as do the CL,CH1, Clr2, and C,,3 homology units depicted in Fig. 1 (Hill et al., 1966; Edelman et al., 1969). The diversity present in the V region of immunoglobulin chains probably encodes the differing antigenic specificities. The V and C regions of a given immunoglobulin chain seem to be encoded
300
L. HOOD AND J. PRAHL
by separate germ line genes, a V gene and a C gene, which are joined at some level of protein synthesis (see Section IV,A). C. IMMUNOGLOBULIN FAMILIES
Three major families of immunoglobulin-the heavy chain family and two light families, h and K-are defined by various criteria, such as amino acid sequence homology and genetic linkage studies (Smith et al., 1971). In man there may be as few as one ( K ) or as many as ten ( H ) C-region members of a given family, each encoded by a separate g e m line gene (Table I ) . Genetic markers are present on the CH regions of certain members of the H-chain family in various species which indicate that the C-gene members of this family are genetically linked (see Natvig and Kunkel, 1968; Gally and Edelman, 1970). In rabbit and man the C, genes are not linked to the ClI genes (Dray et al., 1963; Kelus and Gell, 1963; Steinberg, 1966). A genetic marker on rabbit X chains also indicates that they are unlinked to either of the other two immunoglobulin families ( K or H ) ( Gilman-Sachs et al., 1969). Variable-region markers appear to be present on rabbit VII and VK regions which indicate that the variable genes in each family are linked to the corresponding C genes (Reisfeld et al., 1965; Small et al., 1965, 1966; Mandy and Todd, 1968, 1970; Landucci-Tosi et al., 1970; Hood et al., 1971a). This seems reasonable as each immunoglobulin family also has a set of V regions which can only be associated with the corresponding C-region members of the same family. Thus the general picture is that there are three immunoglobulin families in mammals ( X , K,and H ) which are unlinked to one another in the mammalian genome. The V and C genes of a given family appear to be linked. Although genetic markers do exist for each of the immunoglobulin families, no linkage has been established with other genetic markers in the mammalian genome. TABLE I CONSTANT REGIONGENESOF HUMANIMMUNOGLOBULINS
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
301
D. ONE CELL+ ONEANTIBODY Antibody-producing cells appear to be committed to the synthesis of a single molecular type of imrn~nog!obulin.~Immunofluorescent techniques have been used to demonstrate that the cytoplasm of most lymphocytes contains a single C,, region and a single CL region (Cebra et al., 1966; Pernis et al., 1965). Thus, it appears that each immunoglobulin-producing cell activates one Crr and one CL germ line gene. The analysis is somewhat more complicated with regard to antibody V regions. The only reasonable criterion available for the study of heterogeneity in the V regions from single cells is that of specificity. The use of two distinct antigens (and antigens bearing two distinct antigenic determinants ) has not revealed immunocytes capable of double response (I. Green et al., 1967a,b). Multiple myeloma, the clonal expansion of a single immunocyte, generally results in the elaboration of a single species of immunoglobulin molecule (one V, and one VkI). Thus it would appear each immunocyte also expresses just a single type of V region for light and for heavy chains. The immunocyte also demonstrates the property of allelic exclusion; that is, in an individual heterozygous for CL (or C I I ) genetic markers, individual antibody-producing cells express just a single allelic product (Cebra et al., 1966). All other autosomal genes studied to date, such as the hemoglobins, express both allelic products in a single cell (Rosenberg, 1970). The only other documented example of allelic exclusion is that of the female X chromosome (Lyon, 1968; Eicher, 1970). In female somatic cells, one of the two X chromosomes condenses into a Barr body and is genetically repressed. In this case the maternal and paternal chromosomes are randomly excluded (repressed). The same result appears to be achieved in the immune system, although it is not known if the mechanism of allelic exclusion is dependent on total or partial chromosomal inactivation.
E. VARIABLE-REGION SUBGROUPS If the amino acid sequences of forty V Kregions of humans are compared, the individual V, regions can be assigned to one of three distinct subgroups (Table 11). Individual V K regions in one subgroup can be distinguished from those in the other subgroups by two general criteria: ( 1 ) linkage groups of amino acid residues (underlined in Table 11) and ( 2 ) sequence gaps, that is, deletions or insertions (Hood et al., 1967; A lymphocyte may switch from the synthesis of p to y molecules as will be discussed later (Section IV,A).
TABLE I1 COMPARISON OF VARIABLE-REGION SEQUENCES OF HUMAN K CHAINS~.~ Amino terminal position
20
10
30
40
VKISubgroup ROY DI QMTQSP&SLSASXGQRYTITCQASQllJS------I F L N W Y Q Q K P AG D I Q M T Q S P S S L S A S ~ G I J I t ~ T I T C Q A S Q D 1 _ N - - - -- - H Y L N W Y Q Q G P EV B I - - - - - T W L A W Y Z Z K P
BJ OY HBJ4 DAV FIN KER TRA CON LUX BEL PAIJL
~MTZSPSTLSASVGBRV_TITCRASZSIBDVQMTQSPSSLSASGG~I~_IITITCQASQ_DIIN- - - K Y D I G l I T Z S P S S L S A S _V G B I t V T I _ T C R A S Z T I S - - S W L B W Y Z(Z K P )
r
D D D D D D D B D
2 tr
-
I I I I I I I I I
-
- - - - - -
& M T Q S P S T L ( S A S ~ G B R ~ T I T ) C I ~ A- - SB WQ LBAVW SY Q- -E -L -P
&MTQSP~TLSTV~GQR~TI~CDASQ~I~B----
GMTQSPS
- - S W L I WYQQYP T L S T V ~ G Q R ~ T CI D A S Q B I B - - - - - s WLI WYQQYP s LSA SXGQRI _- K - - - - - - D F s L S A s ~ G Q I ~ T~C T I
p i TI~CQASQBI G M T Q S PS GMTQSPSSLSAS~GDI~~TIT &MTQS
QL T Q S F S S L S A S V G D R Y T I T Z L T Z S P S S L S A S _ V G _ D R V T I T C ZA S Z B L S - - - - - - K < -M T Q S P g T L S A _ S V_G D-R f T I r C R A S Q S L S - - - - - - S S L A W Y Z Z K P
VKIISubgroup Ti E I V L T Q S P G T L S L S P G E R A T C S C R A S Q S V S - - - - - N S F L A W Y Q Q K p E(I V L ) T Q S P G _ T L S ~ S P G E R A T & S C R A S Q S V R -- - - - N N Y L A W Y Q Q I L P Flt4 Z I - S P G Z R A A & S C R A S Q S L S - - - - - G N Y L A W Y Q Q K P B6
VETZSPGTLSL
3
0 u P
2z
r
E I V_L T Q S P G -T L S & S P G D R A T & S C R A S Q -
RAD CAS SMI DIL
I V L T Q S P G T L SL S P G D R A T L S E I V L T Q S P A T L S L S P G E R A T L S E I V L T Q s P G T L S - s P G nit A -
NIG GRA
K I VI,TQSPATLSLSPGEIEAT&S E M V M T Q S P A T L S M S P G E R A T & S
V S - - - - S N S Y L A W Y Q Q K P
E
i
TLSc R ASQ s L s
-
- - - -
s
KS L
s wYzz
KP
2m 2
a 3 J
c VKIIISubgroup CUM E D TEW D MIL D MAN D BATES D
1 I I I I
V i L I T Q T P & S L _ P-V T P G E _ P A S I S C I ) L S S Q S L L A S G D G N T Y L N W Y L Q V S I T Q S P & S L ~ ~ T P G E ~ A & I S C R s S Q - - H ( G B ) S - - - - F L N W Y I , VL TQSP&SLPYTPGEPAEISCR8S&NLLZS-BGBY L D W Y __ L Z V M T Q S P L S L P V T P G E P A S I S CR YL Z
K A Q K P K P
Key to one-letter amino acid code:
A
..
V M T Q S P L S L P V _ T P G E P A S I S G R ~ S Q ( S ) L L H ( S ) BK GP B B - Y L B
The data were previously reviewed by Hood and Talmage (1970), Hilschmann et al. (1970), Edelman and Gall (1969), and Capra and Kunkel (1970). 6 Division of the proteins into subgroups I, 11, and I11 is made on the basis of sequence homology (see text). Subgroup-specific residues are underlined and deletions are indicated by dashes.
Ala Asx c cys D Asp E Glu F Phe B
v,
G H I
Gly His Ile K Lys L Leu M Met
N Asn P Pro Q
Gln
R Arg S Ser T Thr
V W Y 2
Val Trp Tyr Glx
9
r
::
E
304
L. HOOD AND J . PRAHL
Milstein, 1967; Niall and Edman, 1967). On the average, a V region of one subgroup differs from those of a second subgroup by approximately 40% of its amino acid sequence and by at least one gap (Hood and Talmage, 1970). The proposal which has emerged from the existence of subgroups is that each subgroup is encoded by at least one germ line gene, for it is difficult to imagine a genetic mechanism whereby a single V gene could generate three sets of proteins so different from one another (Hood et al., 1967; Hood and Ein, 1968). Furthermore, the three VK genes in man are not alleles at a single genetic locus, as normal individuals have V, regions from each of the three subgroups (Milstein et al., 196913; Grant and Hood, 1971a). Thus the V, subgroups appear to be encoded by at least three germ line genes. Similar data indicate that there appear to be at least three VKgenes in rabbits (Hood et al., 1970a), nine VK genes in mice (Hood et al., 1970b), and four Vlr (Kaplan et al., 1971; Kohler et al., 1970; Cunningham et al., 1969; Press and Hogg, 1969), and five VA (Hood and Ein, 1968) genes in man (also see Smith et al., 1971). In our view the concept of V-region subgroups is misleading in that it tends to obscure certain genetic implications inherent in the V-region sequence data by forcing all V sequences into distinct and frequently arbitrary sets (subgroups).4 The genetic significance of the subgroups will be considered in detail when genealogical patterns are discussed.
F. RABBIT ALLOTYPY Segregating genetic markers on proteins within individuals of a single species which are detected by serologic techniques are defined as allotypes ( allotypy) ( Oudin, 1956). The Mendelian behavior of such genetic markers in mouse (Potter and Lieberman, 1967), man (Steinberg, 1966), and rabbit (Kelus and Gell, 1967) has been studied. Of the three, the rabbit appears the most interesting because it is the only species which has V-region markers. Rabbit allotypy (genetic markers) will be discussed in detail be‘To most immunologists each subgroup represents at least one germ line gene. The arbitrary nature of this concept is obvious, as those who believe in the germ line theory would contend that each distinct V-region sequence is a different subgroup (germ line gene), whereas those who believe in the somatic theory contend that subgroups are represented by nodal sequences on the genealogical tree and must be placed at some level before F in Fig. 3. The only way this level can be determined for somatic theories is to agree on the maximum number of parallel mutations and gaps which can be fixed during the somatic differentiation of the individual. No such agreement has been reached. Therefore we feel that “subgroup” is a vague and confusing term the use of which should be avoided unless it is precisely defined.
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
305
cause of its importance in past discussions of antibody diversity (Cohn, 1968; Milstein and M L I I ~ ~1970; O , Gally and Edelman, 1970; Smith et al., 1971). At least five groups of specificities (sets of alleles) have been identified (Prahl and Todd, 1971); that is, group a (al-3) (Dray et al., 1962), b (b4-6,9) (Stemke, 1964; Dubiski and Miller, 1967), c (c7,21) (Mage et ul., 1968; Vice and Gilman-Sachs, 1969), d (A11,12)(Mandy and Todd, 1968, 1970) and e (A14,15) (Dubiski, 1969). Groups a, d, and e are found on the H chain, and groups b and c on the L chain, The group b markers are found on the K chains (Appella et al., 1968), and the c group on rabbit chains ( Gilman-Sachs et al., 1969). Although group d and e specificities are restricted to the C, gene of rabbit (Mandy and Todd, 1969;Dubiski, 1969),those of group a are associated with the N-terminal half of the y (Marrack et al., 1962),the (Feinstein, 1963),the p (Todd and Inman, 1967),and the E (Kindt and Todd, 1969) chains. This implies that the group a specificities are present on the V,, region because it is shared by all classes of H chain (Todd, 1966). The data in support of this conclusion and the genetic significance of VHregion markers will be discussed later. Within each group the specificities behave as alleles of a single locus. The markers of group a appear to be genetically linked to both group d and e specificities in pedigree studies (Zullo et al., 1968; Mandy and Todd, 1970; Landucci-Tosi et al., 1970), although linkage between d and e has not yet been established. Sequence and serologic data also suggest that the V, and C, genes of rabbit are also linked (Hood et al., 1971a; Appella et al., 1969). Thus the V and C genes of each immunoglobulin family appear to be genetically linked to one another. (Y
IV. A Genetic Mechanism for Differentiation: Two Genes -+O n e Polypeptide Chain
A. ONE GENEOR Two? Perhaps the most provocative proposition that has emerged from the genetic and sequence data of the immune system is the supposition that each antibody polypeptide chain is encoded by two separate genes ( a V and a C gene) and that these genes (or their products) are joined at some level of protein synthesis to produce a single polypeptide chain (Dreyer and Bennett, 1965; Hood et ul., 1967; Dreyer et al., 1967). If this mechanism operates at the level of DNA, that is, if genes are actually translocated in the vertebrate genome, then V- and C-gene joining affords a specific mechanism whereby one cell (an immunocyte) can become committed to the synthesis of a single molecule (only that V gene which is joined to its corresponding C gene is expressed). Thus
306
L. HOOD AND J . PRAHL
differentiation in the immune system and hypothetically in other complex systems may occur through the translocation of V and/or C genes. Before considering the general implications of this joining mechanism in more detail, we shall ( I ) review the arguments that support the two gene + one polypeptide chain hypothesis and ( 2 ) consider the level in protein synthesis at which joining occurs. The logic supporting this supposition is illustrated in the human K chain and may be surnniarized as follows: ( I ) there are multiple V, germ line genes (three or more); ( 2 ) there is a single C, germ line gene with two allelic forms; ( 3 ) any V, region may be associated with either allelic form of the C, region. If there are niultiple germ line V, genes and a single C, gene, it follows that there must be a joining mechanism for uniting V and C genes (or their products) (Hood et ul., 1967). We shall briefly review the evidence for each of these points. First, the hypothesis that there are multiple germ line V, genes follows from the existence of the V, subgroups (see Table I1 and Section III,E on V-region subgroups). No reasonable genetic mechanism has been proposed whereby a single V, gene could producc three sets of proteins (V, subgroups) which differ by 240% of their amino acid sequence and by precise sequence gaps. Such a mechanism would need to be programmed in a precise fashion or it would require intense selective pressures. Both of these alternatives are unattractive, as will be discussed later. Rather, it seems reasonable to suggest that at least one germ line gene is required for each V, subgroup. Recent sequence data have escalated the number of V, subgroups (V, genes) in man to six (Smith et al., 1971), and a detailed analysis of genealogical V-region pattern suggests that this number will increase even more as additional data accumulate (see Section VI,B,2). Thus it seems reasonable to conclude that there are multiple V, germ line genes in man, although the question of how many is still subject to debate. Second, the supposition that the C, gene is encoded by a single germ line gene is based both on arguments that ( I ) there is one C, gene and (2) multiple C, genes would require unusual evolutionary mechanisms. There is a single amino acid substitution that occurs in the C, region of nian which segregates in family studies in a classic Mendelian fashion (for references, see Terry et ul., 1969). This suggests that the C, region is encoded by a single structural gene with two allelic forms. The alternative to a single C, gene is to propose that there are, for example, six identical C, genes joined in the germ line to each of the proposed V, germ line genes. Such an explanation might be compatible with Mendelian behavior (the C, genes would be closely linked); however, it is difficult to see how multiple C, genes could evolve to become
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
307
species-specific. For example, the multiple C, genes must evolve such that ( 1) the multiple C, genes of a single species are all identical (or nearly identical) and ( 2 ) the C, genes of different evolutionary lines are quite different (40%difference between human and mouse C, genes). Furthermore, it is difficult to see how Mendelian markers might arise simultaneously on six linked C, genes (in man) and, perhaps even more d a c u l t to understand, why meiotic recombination has not scrambled these markers to make Mendelian behavior impossible (for a more complete discussion, see Hood et al., 1967; Hood and Ein, 1968). These evolutionary arguments are less impressive in view of the unusual evolutionary mechanisms required by species specificity and rabbit V-region allotypes (see Section V1,E). Nevertheless, if we were to have six germ line VKCK genes, it would seem necessary to postulate different evolutionary mechanisms for the V and C halves of these genes ( V diverge and C evolve in parallel). Thus it is attractive to obviate these difficulties by suggesting that the C, region in man is encoded by a single germ line gene. Third, it appears that V, regions from each of the major subgroups are associated with either allelic form of the C, region (Milstein et al., 1969a). Thus, presumably any V, gene can be joined to either corresponding C, allele. Since there appear to be multiple germ line V genes and a single C , gene, it follows that there must be a joining mechanism for uniting V and C genes or their products. Another line of evidence in the heavy chain family supports the supposition that there are separate V,, and CIr genes; namely, it appears that the same VII region may be joined with two different CII regions. In the course of immunization with many antigens, antibodies of the IgM class are produced initially, and later specific antibodies of the IgG class are synthesized. Furthermore, during a transition period in early primary response, single antibody-producing cells appear to produce both IgG and IgM molecules (Nossal et al., 1965, 1971). Pernis et d. ( 1971) have also demonstrated by immunofluorescent techniques that the same immunocyte can have IgG molecules in the cytoplasm and IgM molecules on the ccll membrane. Thus one cell appears to be capable of shifting from the synthesis of p chains to y chains. Since antibody specificity is maintained during the maturation of the immune response from IgM to IgG, it is reasonable to postulate that the same V,, region shifts from a C, to a C, region. More direct support for this supposition has come from the independent study of 2 human patients who had a biclonal myeloma response with the simultaneous production of serum IgM and IgG myeloma proteins from two cell populations ( Wang et al., 1970; Penn et al., 1970). Presumably the myeloma-
308
L. HOOD AND J. PRAHL
tous change occurred in an immunocyte in transition ( p to y ) and, accordingly, generated two cell populations. In the study of Wang et al. ( 1969), the light chains isolated from each immunoglobulin class were identical in their amino terminal 25 residues and had identical electrophoretic mobilities. However, mobility differences in the light chains were observed in the study of Penn et al., although the light chains had identical idiotypic specificity, Isolated heavy chains from the yM and yG proteins also showed very similar ( possibly identical) idiotypic specificity. The VH regions from each heavy chain ( y and p ) were identical for 27 residues at their amino terminus ( Wang et al., 1970). If the V H regions of the y and p chains prove to be identical, this would support the suggestion that a given Va region can be initially attached to the C, region and later joined to a Cy region, explaining the shift in specificity from the IgM to the IgG class during the development of the immune response. Such an observation supports the hypothesis that two separate genes encode the antibody heavy chain-a V, gene and a CH gene (see similar arguments for the singularity of the C, gene in the preceding section). Furthermore, the VH gene may be retranslocated from one C, to a second CII gene within the heavy chain family. Thus the joining event may not be irreversible and, accordingly, a translocation mechanism for differentiation need not be irreversible. The importance of absolutely confirming the complete identity of the VH region from the y and p chains is obvious.
B. LEVELOF
JOINING
If there is a separate V and a separate C gene for each immunoglobulin polypeptide chain, joining can occur at any level of protein synthesis [DNA, ribonucleic acid (RNA), or protein]. There is, as far as we know, no compelling evidence as to the level of joining. A number of general observations have placed certain constraints on the joining mechanism. 1. Population studies and suppression studies (Zullo et al., 1968; Mandy and Todd, 1970; Landucci-Tosi et al., 1970) on the heavy chain markers for rabbit V and C regions also indicate that the V and C genes are closely linked to one another. Rabbit V Kand CK genes also appear to be linked (Hood et al., 1971a). A V region from one immoglobulin family has never been observed joined to a C region from a second family. Thus the joining mechanism appears to unite only those V and C genes that are closely linked (i.e., in the same family). Perhaps the joining mechanism requires this close linkage, 2. Although the V, and C,, genes appear to be linked, one can still ask whether or not they are transcribed in a cis fashion with regard to
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
309
the parental chromosomes. Since separate genes encode the VH and C , regions (see Section IV) and since markers are available for the VH ( a group) and C, ( d and e group) genes, one can determine in appropriately heterozygous individuals whether a maternal V,, gene can be associated with a paternal C,, gene and vice versa (Prahl et al., 1970; Kindt et al., 1970; Landucci-Tosi et al., 1970). The f, offspring of homozygotic parents ( a l , A l l / a l , A l l X a3,A12/a3,A12) typed al,All/a3,A12, but few if any molecules appear to be al,A12 or a3,All. Suppression experiments also support this cis linkage of V, and Ca regions. In rabbits, it is possible to suppress the expression of one allele by a variety of techniques (David and Todd, 1969; Mage and Dray, 1966). In a double heterozygote if the maternal V, allele is suppressed, then the corresponding maternal C, allele is also suppressed and again VH and C , linkage and cis translation is suggested. For example, Landucci-Tosi et al. (1970) observed suppression of both a1 and A14 in the offspring of genotype al,A14/a2,A15 born of an a2,A15 mother immunized with al,A15 antigen. Thus the joining mechanism must operate in a cis fashion with regard to the parental chromosomes. That is, V regions from the maternal chromosome must always be joined with C regions from that same chromosome and likewise with paternal V and C genes. This observation seems to suggest that joining occurs at the DNA level as joining at the RNA or protein level might lead to molecules of mixed parentage. The absence of mixed molecules could, however, also be explained by the independent phenomena of allelic exclusion; that is, in a given antibody-producing cell one chromosome (maternal or paternal) is repressed. 3. Heavy chain disease proteins have deletions which can span the carboxy terminal portion of the V,, region and the amino terminal portion of the Clrregion (Prahl, 1968; Frangioni and Milstein, 1969; Franklin and Frangione, 1971). Two genetic events might account for these deletions, First, the joining mechanism may have united the V,, and C,, genes at the DNA level followed by a subsequent deletion. This interpretation would support joining at the DNA level (Terry and Ohms, 1970). Second, a germ line deletion may have occurred which, because of the close linkage of V,, and C,, genes, fused two formerly separate germ line genes. This event may have occurred in the individual from which the heavy chain disease protein was obtained or earlier during the evolution of the species. The latter possibility suggests that heavy chain disease proteins may represent germ line genes in certain populations, a supposition which is testable. If true, it would be interesting to tcst whether or not additional V,, gene diversity has also been deleted in these populations. Thus the presence of H-chain disease proteins sug-
310
L. HOOD AND J. PRAHL
gests that ( a ) joining of V and C genes occurs at the DNA level or ( b ) the human VH and CH genes are closely linked or ( c ) both a and b are true. Since the VH and CH genes appear to be linked (at least in the rabbit), the existence of heavy chain disease proteins of the a (Seligmann et ul., 1969), p (Ballard et al., 1970), and y (Franklin et al., 1964; Osserman and Takatsuki, 1964) types suggests that these C H genes are linked. 4. Pulse-labeling experiments of the type used by Dintzis (1961) to establish that the polypeptide chains of hemoglobin have a single growing point, also suggest that immunoglobulin chains have a single growing point starting at the N terminus and proceeding to the carboxy terminus of the chain (Fleischmann, 1967; Knopf et al., 1967). Unfortunately, it is possible to make models that involve two polypeptide fragments for each chain giving pulse-labeling data similar to that obtained. 5. In their studies on immunoglobulin assembly in mouse myeloma tumor lines, Schubert and Cohn (1970) report the presence of a light chain fragment of approximately one-half the molecular weight of the intact chain with kinetics of turnover not consistent with its being a degradation product. Although there are technical complications, these observations raise the possibility that joining may occur at the peptide level. 6. In human K chains, it is interesting to note that the 15 carboxy terminal residues in the V, region (“switch region”) do not fit into the subgroups given in Table 11. The switch region sequence is scrambled (see Section VI,B ), perhaps reflecting the translocational mechanism which may be used to join V and C genes at the DNA level. C. A MECHANISMFOR DIFFERENTIATION Although no compelling evidence argues for joining at the DNA level, the general implications of such a mechanism are fascinating and have led to three mechanisms for the somatic, intrachromosomal, DNA level joining of V and C genes. The translocation model proposes that one of a tandem array of V genes is moved next to a C gene in another part of the chromosome by a crossing-over event similar to that which integrates the phage X genome into a specific part of the Escherichia coli genome (Dreyer and Bennett, 1965; Gally and Edelman, 1970). The copy splice model suggests that a copy splice event occurs which operates via a programmed mechanism and special organelle ( Dreyer et al., 1967). The other mechanism, the lateral array model, envisions a branched network of DNA forming a lateral array of V genes, all of which are simultaneously adjacent to the C genes, which themselves form a branched lateral array (Smithies, 1970; Smithies et al., 1971). This model obviates
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
311
the problem of translocating genes by introducing new rules for replication and transcription. For each of these theories of joining there is a corresponding mechanism by which a cell might become committed to the synthesis of a single molecular species of immunoglobulin. Dreyer et al. (1967; Dreyer and Bennett, 1965) and Gally and Edelman (1970) have suggested that the translocation or copy splice event commits the cell to the synthesis only of the joined V-C pair. Smithies (1970), on the other hand, proposes that protein switches are either randomly or under physiological control “set” in one of two opposite configurations at each bifurcation in the branching network of DNA, thus specifying which of the many possible pathways through the network is to be expressed in that cell. Either joining mechanism is quite distinct from the phenomenon of allelic exclusion. Such a joining mechanism may play a general role in differentiation, namely, it may provide a mechanism whereby the cells in other complex systems can become committed to a single “unit” of information. In this regard it is interesting to recall that the joining mechanism may not be irreversible in that one Va gene may be joined with two CH genes. Thus a cell could conceivably change its commitment with a joining mechanism or, presumably, dediff erentiate. A programmed joining mechanism would permit the orderly read out (expression in separate differentiated cells) of all of the V-gene information. It will be interesting to see if the specific molecules in other complex systems have Vand C-gene correlates. V. Evolution of Immunoglobulin Variable and Constant Genes
A. HOMOLOGY UNITSAND GENEDUPLICATION
The immune system provides a graphic example of the various mechanisms employed in gene evolution, that is, mutation, deletion and insertion, and gene duplication of the discrete (separate) and contiguous (fused) types. The basic homology unit in the immunoglobulin system is a gene that encodes a polypeptide of about 110 amino acids with a symmetrically placed disulfide bridge spanning 2 6 0 residues (Hill et al., 1966; Singer and Doolittle, 1966). A hypothetical scheme for the evolution of immunoglobulin V and C genes is given in Fig. 2. An early gene duplication gave rise to the primordial V and C genes. The V-gene library was expanded through discrete duplications into three V-region families: Vh, VK,and V,. The size of these V-gene families will be considered later. Discrete duplications of the primordial C gene gave rise to the CA
L. HOOD AND J. PRAHL
312
/I I Ivg I' IIII. -I I I.
1
-
Translocation
K
4
Evolution
c
c- Dillerentiation --c
FIG.2. Hypothetical scheme for the evolution and differentiation of the mammalian immune system. The immunoglobulins presumably evolved by gene duplication of a primordial gene of 110 codons in length (see text). Joining of a V region to an appropriate C region probably commits the immunocyte to the synthesis of a single Vl. and Vlr regions (see text).
and C, genes, whereas contiguous duplications gave rise to C, genes which consist of 3 or 4 homology units the size of the original primordial gene (see Fig. 1 for a description of these homology units) (Edelman et al., 1969; Putnam et al., 1971). The V regions form one set of homology units and the C regions a second set (see Fig. 1) (Edelman et al., 1969). The only obvious relationship between V and' C homology units is a similarity in length and the presence of a single disulfide bridge which
THE IMhfUNE SYSTEM: A MODEL FOR DIFFERENTIATION
313
spans about 60 residues (Edelman et al., 1969). This lack of homology stresses the early divergence of primordial V and C genes.
B. THE DOMAINHYPOTHESIS Edelman ( 1971a) has proposed thc domain hypothesis which suggests that each homology unit has a discrcte function(s); for example, the V regions are involved in antigen binding. The CL and C, 1 homology units may position and join the light and heavy chains to one another (Milstein, 1966). The C,,2 unit seems to be implemented in effector functions such as complement fixation and passive cutaneous anaphylaxis ( Kehoe et al., 1969; Utsumi, 1969; Prahl, 1967). Perhaps the C,,3 homology units position the heavy chains (Inman and Nisonoff, 1966; Prahl, 1967; Charlwood and Utsumi, 1969). Thus different classes of immunoglobulin molecules (c.g., IgG, IgA, and IgM) appear to have evolved such that their homology units (especially C,,2) have come to assume different functions. Evidcnce for the domain hypothesis is indirect, but more sophisticated methods are available for the fragmentation of immunoglobulin molcculcs ( Solomon and McLaughlin, 1969) and should permit detailed tests of the hypothesis. High-resolution X-ray crystallographic data should also be revealing in this regard (see Davies et al., 1971). OF IMMUNOGLOBULINS FROM C. EVOLUTION MEMBRANE MOLECULES
The evolution of the immune response has required the development of at least four separate components: ( I ) a library of V (and C ) genes, (2) a joining mechanism, ( 3 ) a means for initiating cell proliferation once the membrane-bound antibody receptor molecule has combined with antigen (amplification of the response), and (4) a mechanism for synthesizing and secreting antibody molecules ( humoral response). The size of the V-gene library and a consideration of the need for special mutational and selectional mechanisms will be considered shortly. The possibility that the library of V genes evolved from membraneplaced receptor molecules is attractive (Dreyer et al., 1967). As one ascends the evolutionary scale from single cell organisms to the complex Metazoa, more and more cell receptor molecules are required for a variety of different functions such as cell recognition, scavenging of debris, and hormonal triggering. Thus libraries of V-like genes may have evolved to carry out diverse receptor functions. Presumably the joining mechanism evolved to join these V-like cell surface receptor genes to their corresponding C genes. Thus a single C gene could serve as the handle whereby receptor molecules of differing specificities could be
314
L . IIOOD AND
1. PRAIIL
appropriately positioned in the membrane. Perhaps an event such as chromosome doubling (Ohno, 1970) provided a library of such receptor V gencs for thc evolution of the immune systcin. Initially there may have been a very limited number of C genes of which the primary function was positioning the attached V genes on the membrane. Perhaps the critical function which differentiated the early immune rcsponse from other cell surface receptor functions was the ability to initiate cell division aftcr the antibody receptor molecule had combined with its corresponding antigen-thus producing more cells with identical antibody receptor molccules ( cellular immune response). Presumably this event occurred just prior to or at thc divergence of vertebrates and invertebrates as it is uncertain if a bona fide immune response exists in invertebrates. Later certain immunocytes, upon antigenic stimulation, acquired the property of synthesizing and secreting antibody molecules (humoral immunc response). This view of the evolution of the immune response suggests that ( 1 ) the existence of V-gene libraries and the joining mechanism antedated the emergence of the immune system and that both may be general mechanisms in the differentiation of other complex systems in higher organisms and ( 2 ) the cellular immune response arose before the humoral response. A series of provocative questions follow. Is there evidence for cellular immunity in invertebrates (see E. L. Cooper, 1968)? Are there V and C correlates for other cell membrane receptor systems? If so, do any show even a distant evolutionary relationship to their immunoglobulin counterparts? Is there evidence for a joining mechanism in other systems? VI. Theories of Antibody Diversity
A. MECHANISMS FOR INFORMATION STORAGE Information can be stored in the mammalian genome by two general mechanisms. First, each unit of information (antibody V region) may be stored as a corresponding germ line gene which arose during the evolution of the species (the germ line theory). Alternatively, information may be stored as a limited number of germ line genes which are expanded by a mutational or recombinational process during somatic differentiation of the individual (somatic theories), Each of these mechanisms had attractive features prior to the availability of detailed Vregion sequence information. The somatic theories were appealing because they permitted an economy of germ line genes which molecular biologists had come to expect through the study of phage and bacterial genetics. With a single V gene for each immunoglobulin family, phenomena such as rabbit heavy chain allotypy and species-specific residues
TIIE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
315
( apparent V-region genetic markers) were easily explained. Finally, it was easy to imagine how naturhl selection might operate on a single V gene, but rather difficult to imagine how it might work in a multigene system. It was felt that the many “silent” genes of each generation could mutate, unchecked by selective pressures, and lose their respective functions (Smithies, 1967; Cohn, 1968; Edelman and Gally, 1967; Hilschmann, 1967). The germ line theory was attractive because it did not require an ad hoc mechanism for explaining V-region diversity. Phylogenetic evolution through gene duplication, mutation, and selection in the germ line had produced gene diversity in other systems such as the hemoglobins and cytochronies ( Braunitzer, 1965; Margoliash et al., 1968) and, accordingly, could also generate antibody diversity. Each of the somatic theories required ad Aoc assumptions for a special mutational mechanism or intense somatic selective pressures to avoid prohibitive cell wastage (Hood and Talmage, 1970). Furthermore, the germ line theory predicted the need for a joining mechanism (for uniting V and C regions ) which is accepted by most contemporary immunologists ( Dreyer and Bennett, 1965; Hood et al., 1966; Dreyer et d.,1967). The joining mechanism is an ad hoc addition to all somatic theories. Finally, the presence of a complete V-gene library in the germ line permits one to envision a programmed mechanism for expressing in an orderly fashion (and not by random mutation or recombination) all potential antibody specificities through differentiated irnmunocytes ( Dreyer et al., 1967). The somatic and germ line theories will be considered in terms of three important patterns which have emerged from amino acid sequence data of immunoglobulin V regions-the genealogical tree, species specificity, and rabbit allotypes. We feel that only the genealogical tree speaks in a direct way to the problem of antibody diversity and information storage. Species specificity and rabbit allotypes require ad hoc evolutionary mechanisms for all multigene theories, even those with a relatively limited number of germ line V genes. As we shall show, somatic and germ line theories alike are now considered to be multigene at least to the extent of 2 1 0 V genes per immunoglobulin family. Accordingly, the patterns of species specificity and rabbit allotypes do not distinguish among these alternative theories.
B. VARIABLE-REGION PATTERNS 1. General Observations
The simplest approach to the analysis of amino acid sequence data from immunoglobulin V regions is to align the sequences so as to max-
316
L. HOOD AND J. PRAHL
imize amino acid homology and to inspect them visually. A set of partial V, sequences from man is given in Table 11. Two general patterns are evident. (I) Each V, sequence can be assigned to one of three subgroups on the basis of linked amino acid residues (which are underlined) and sequence gaps (see Section 111,E). Variable regions from the same subgroup are considerably more closely related ( 1525%difference) than are V regions from differing subgroups ( 4 0 %difference). ( 2 ) The V regions from individual subgroups show three areas of hypervariability (Wu and Kabat, 1970; Capra, 1971).5 One such area extends from positions 30 to 38 in Table 11. We feel this hypervariability merely reflects the amino acid sequence diversity of the antigen-binding site and that it does not place (obvious) constraints on genetic mechanisms. However, a detailed analysis of the genetic meaning of subgroups is possible using a more sophisticated procedure of data analysis.
2. Genealogical Pattern All immunologists would agree that V-region diversity should be explained by the simplest possible genetic mechanism that is compatible with the experimental data. Given a set of proteins ( V regions) how does one determine the nature of the genetic mechanism which produced them? The genealogical approach assumes that contemporary proteins can be related through a genealogical tree (Fig. 3 ) which permits the determination of both immediate and distant ancestral or nodal sequences (levels F back to A in Fig. 3 ) (Fitch and Margoliash, 1967). Hence one can determine the minimum number of mutational events required to construct a set of proteins ( V regions) from a single ancestral gene (primitive VL gene at level A ) . Evaluation of the alternative genetic models for antibody diversity is possible in terms of the genetic events required to produce the various V-region families. This genealogical approach is independent of the time scale and, accordingly, may be compatible with somatic or germ line mechanisms. It does, however, Hypervariable regions are characterized by extensive sequence diversity and are located at residue positions in the early 30s ( I ) , 50s (11), and 90s (111) for human V K regions. Hypervariable regions I and I11 also have residue deletions and insertions. The two V, region half-cystines (positions 23 and 88) juxtapose hypervariable regions I and I11 and raise the possibility that they may join together to form a part of the antigen-binding site. Hypervariable regions with similar positioning have also been noted in the heavy chain (Capra, 1971). Furthermore, affinity labeling has been achieved on peptides from hypervariable region I (Goetzl and Metzger, 1971; Franek and Novotny, 1969) and possibly from region I11 (Singer and Thorpe, 1968; Thorpe and Singer, 1969). See Wu and Kabat (1970) for a thorough analysis of this concept.
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
I
I A
, I I ‘ 8 ’
m
, , I ,II , m , I P , P
4
317
Level -
FIG. 3. Hypothetical human light chain genealogical tree. The tree is constructed from a set of proteins, such as human VI, regions, by generating a series of ancestral or nodal sequence (levels E, D, C, etc.) using the minimum possible number of “genetic events” ( base substitutions, sequence insertions or deletions, and gene duplications) (Fitch and Margoliash, 1967). The numbers on branches between levels represent the number of base substitutions required in going from one level to the next. The brackets represent deletions which removed codons at the numbered positions. The genetic significance of the tree can best be understood by noting the genetic events required to go from the primordial VL gene (level A ) to contemporary genes (level F ) . First a gene duplication of the VL must occur to produce two genes which through genetic events evolve to become the primordial Vn and VA genes (level B ) . For example, from the primordial VL gene (level A), it is possible to derive the primordial V r gene by five base substitutions at certain codon positions (for example, at 12, 28, 56, 72, and 103). The primordial VA gene could also be generated by five different base substitutions (for example, at 15, 19, 58, 77, and 92) and by the deletion of codons at positions 94-95 and 9. From these primordial genes the ancestral sequences at successively higher levels (C, D, E ) can be derived until contemporary sequences are reached (level F ) . V, indicates all the contemporary K sequences, and VA the contemporary sequences; V,I, VXII, and V X mindicate major branches on the K genealogical tree which have been designated SUbgTOUpS. The branch Vnrl has been expanded to give some indication of the actual diversity which might exist at level F (sequences 1-6). Other portions of the tree show only a few representative sequences but, in fact, should probably be expanded in a similar fashion. The genetic events responsible for generating this genealogical pattern could occur, in part, during somatic differentiation (somatic theory) or entirely during the evolution of the species (germ line theory) (see text).
318
L. HOOD AND J. PRAHL
permit visualization of the genetic events that are required of either mechanism. Figure 3 is a hypothetical genealogical tree for human V, and Vx regions (genes) which has been idealized to clarify the genetic meaning of such a pattern. Given that such a genealogical tree can be constructed for V, genes, the implications can best be illustrated by considering the general features of this tree. The primordial V, gene (level A) has a particular nucleotide sequence (hypothetically of 2 3 3 0 bases, see Section V,A) which can generate all of the genes at level F by fewer “genetic events” than would be required of any other primordial V, sequence. Three kinds of genetic events occur: (1)single base substitutions, (2) nucleotide insertions or deletions, and ( 3 ) gene duplication. All are illustrated in Fig. 3. The branch numbers (generally 5) indicate the number of nucleotide substitutions which occur in a particular gene in going from a lower to a higher level. Gene duplication occurs at each successive level. How is diversity (level F ) generated from the ancestral V, gene? Gene duplication occurs first ( Fig. 3 ) . One daughter gene undergoes five single base substitutions at various codons throughout the gene to become the primordial V, gene (level B ) , whereas the second daughter gene undergoes five base substitutions and two nucleotide deletions which remove codons at positions 94-95 and 9 to become the primordial VAgene (level B). All of the genes above level B in the h branch of the genealogical tree will share the corresponding gaps and base substitutions (apart from subsequent random back mutations). Similar genetic events occur in moving from level B to C on the K branch. The primordial V, gene duplicates, and five base substitutions plus a codon deletion at position 35 are fixed in the daughter gene VKA,whereas five different nucleotide changes occur in daughter gene VKB.Once again all of the genes on the VKAbranch will share these base substitutions and the deletion at position 35. As successive levels of the genealogical tree are passed, additional gene duplication and mutation leads to increasing gene diversity until level F is reached which corresponds to actual protein (gene) sequences. Major branches on this genealogical tree are designated VKI,VKII,VKIII,V A ~etc., , and they correspond to the V-region subgroups which have already been described. The potential diversity of the tree is indicated only in the left-most branch (proteins 1-6) due to space limitations. Presumably the particular structure. of a genealogical tree is dictated by the forces of natural selection (or somatic selection). In practice, of course, the genealogical tree is constructed by starting with a known set of sequences and working from levels F to A to derive the nodal sequences at each branch point (Fitch and Margoliash, 1967). How is this genealogic pattern related to theories of antibody di-
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
319
versity? The germ line theory contends that the terminal-most twigs (level F ) represent germ line genes, whereas somatic theories would place the boundary between germ line genes and somatic mutational events at an earlier level (somewhere below level F). Genealogical analysis also permits us to derive the most probable sequence for those genes that are germ line under a somatic model. Leaving aside the detailed justification, let us assume that a somatic mechanism requires three germ line V, genes, one for each of the branches labeled VKI, V,,,, and VK],,.The most probable sequences for these genes would be the nodal sequences at V K A l , and VKA2, and VKB (levels D, D, and C, respectively) because each gene can generate one of the three groups of V regions with a minimum number of genetic events. Thus, under a somatic theory the nodal sequences in the genealogical tree at some level represent germ line genes. The critical question with regard to mechanisms of antibody diversity is, What level in the genealogical tree represents germ line genes? Suppose only the primordial V , gene is in the germ line (level A ) . Then 25 somatic base substitutions plus various deletions must occur in each immunocyte line to generate serum light chains with VAand V, regions. This appears to be an excessive number of mutations required during the somatic differentiation of each individual as each mutation must be fixed by somatic selection and the mutant immunocyte clone expanded before the next somatic mutation can occur (unless a programmed hypermutational mechanism is postulated; see Section VI,G ). More importantly, intense selective forces must operate during the lifetime of the individual in order to generate, for example, proteins 1-6 (level F ) in the VKIA branch of the genealogical tree. The V, genes in six distinct immunocyte lines encoding the corresponding six V regions (1-6 in Fig. 3) must undergo twenty-two identical mutations and two identical deletions (at codon positions 35 and 3 0 3 4 ) . Somatic selective forces of such magnitude and efficiency are difficult to conceive (see Section V1,G). How then does one decide the germ line level for a genealogical tree? The primordial V, and VA genes are germ line because inferential evidence suggests that they are located in two different regions of the mammalian genome.B If the germ line level in Fig. 3 is B (primordial V, and VA genes), then immunocytes which are to synthesize VKIpro-
' There are genetic markers on rabbit and K chains which are unlinked (Gilman-Sachs et d.,1969). The rabbit V, genes appear to be linked to the corresponding C, gene (Hood et al., 1971a). Presumably the same is true of the rabbit VA and Cx genes. Thus, irrespective of where the genetic markers are located, the VAand V, genes of rabbit, and presumably of other mammals, appear to be unlinked and accordingly must be separate germ line genes.
320
L. HOOD AND J . PRAHL
teins must undergo ten distinct base substitutions and a single deletion at identical positions. In order to avoid excessive parallel mutation and/or unreasonably intense somatic selection, most immunologists would agree that the V, regions are encoded by at least three germ line genes. This is a more explicit presentation of the subgroup argument' mentioned in Section II1,E. Are the proteins from the VKIAand VKIBbranches (Fig. 3) also encoded by separate germ line genes by the same criteria? Again many immunologists would probably agree that they are ( Weigert et al., 1970; Cohn, 1971a,b; Hilschmann et al., 1970; Milstein and Pink, 1970; Edelman and Gally, 1971). Thus, how far out on the genealogical tree do the nodal points represent germ line genes? How much parallel mutation can be generated by selection during somatic differentiation? These questions, it seems to us, are unanswerable, but one should opt for the simplest genetic mechanism that is compatible with the data. How well do the actual data fit with the hypothetical genealogical tree given in Fig. 3? In Fig. 4 is given the genealogical tree for the complete (or nearly complete) sequences of twelve human and two BALB/c mouse (MBJ) V, regions. The division of the genealogical tree into the major branches (subgroups; genes 1-23, gene 4, and gene 5 ) is apparent. More than ten base substitutions (21.8, 10.5, and 25.6) separate the ancestral V, gene from the first nodal ancestors of each major branch, as do sequence gaps which are not indicated in this figure. The left-most branch of this genealogical tree shows branching which suggests the presence of at least three additional germ line genes (gene 1, gene 2, and gene 3-see legend to Fig. 4 for the justification of these assignments). The mouse V, sequences branch from the genealogical tree at distinctly different nodal points from their human counterparts and the significance of this species specificity will be considered subsequently. Thus the general features of the idealized and the actual genealogical trees are very similar. What is the genetic implication of identical (or nearly identical) Vregion sequences? Capra and Kunkel (1970) have reported that two VKIregions from unrelated humans are identical over the amino terminal 40 residues which have been examined. Furthermore, two pairs of potentially identical V, sequences have been detected among the myeloma proteins from the inbred BALB/c mouse (Hood et al., 1970b). Thus identical V regions may occur in separate individuals. Each pair of identical V sequences suggests the presence of one additional germ line gene. This argument is the same as that previously discussed for delineating germ line genes on the genealogical tree, namely, the reasonable desire to avoid excessive parallel ( identical) mutations. Similar
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
321
FIG.4. The actual genealogical tree for the V K regions of mouse and man. This genealogical tree is based on the residue-by-residue analysis of Smith et al. (1971) (see text) and is reprinted here with permission. The average mutational difference from each nodal point to the resultant sequence is shown along each branch. The rule for enumerating germ line genes is based on the following reasoning: if a divergence A definitely precedes another divergence B in the genealogy, then A must have occurred presomutically, giving rise to two different germ line genes. Otherwise the two or more proteins emanating from divergence B must have independently accumulated those shared mutations which placed divergence A prior to divergence B. Divergences of which the order is uncertain are enclosed in a dashed line. Human genes inferable from these studies are shown by a bracket.
reasoning suggests that V sequences differing by a few residues ( 21-4) represent at least one additional germ line gene (the nearest nodal sequence). Thus identical and nearly identical V sequences will add to the library of germ line V genes required. Most proponents of contemporary somatic theories feel that there are at least ten to fifty V K (or V,) genes in the human genome because of the obvious genealogical patterns and the desire to avoid excessive parallel mutation during somatic differentiation ( Cohn, 1970; Gally and Edelman, 1970; Jerne, 1971). This would place the germ line boundary somewhere between level E and F in Fig. 3. A critical question can be asked at this point. Does the detailed branching pattern extend from these conceded germ line genes to the terminal-most twigs of the genealogical tree (as shown for proteins 1-6
322
L. HOOD AND J. PRAHL
in Fig. 3 ) ? Such a branching pattern implies that the genes are subject to intense selective pressures even at the outermost twigs of the genealogical tree. The alternative possibility would be that V, regions differing, for example, by just a few residues might eminate randomly and without evidence of selection (branching) from a single nodal sequence. This question has been provisionally answered in one system (mouse myeloma chains), and it appears that selection extends to the outermost twigs of the genealogical tree. Mouse VA regions are very closely related in amino acid sequence. Six of ten VAregions studied were identical to one another, and the other four differed from these by one to four base substitutions (Weigert et al., 1970). Each of the seven mutations observed fell into one of the three hypervariable regions (see Section VI,B,l), implying that selective pressures (germ line or somatic) must operate to restrict the sequence alternatives that can be expressed phenotypically. The question arises as to whether it is more reasonable to have selection operate in the soma during the somatic differentiation of each individual or in the germ line during the evolution of the species. The somatic and germ line theories make different predictions concerning the sequences of mouse VA regions. Germ line theories contend that mouse VA regions are encoded by a smaller number of germ line genes (perhaps 5-20) and that repeats of all the variant proteins (those differing from the six identities) should be seen in subsequent analyses (Smith et al., 1971). On the other hand, somatic theories suggest that a large number of variants should be produced. Subsequent sequence analysis of mouse X chains should prove interesting in this regard. The types of amino acid (nucleotide) interchanges seen in each family of V regions are similar to those seen in large sets of evolutionarily related proteins such as hemoglobulins and cytochromes (Hood and Talmage, 1970). This is demonstrated ( 1 ) by the predominance of single base substitutions, ( 2 ) by the random nature of transversional and transitional basc changcs, and ( 3 ) by the striking tendency of G (guanine) to mutate more frequently than is expected. Each of these properties is demonstrated by other evolutionarily related sets of proteins (Nolan and Margoliash, 1968). This similarity between antibody V regions and other sets of proteins is also supported by the presence of the genealogical patterns themselves. Thus the genetic mechanism responsible for antibody diversity produces a set of proteins ( V regions) the variation of which resembles that produced in the proteins of other systems ( hemoglobins and cytochromes ) by phylogenetic evolution. As the amount of V-region data increases, the structure of the genealogical tree becomes more apparent ( Fig. 4), and additional
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
323
germ line genes must be postulated to avoid excessive parallel mutation. This is certainly true of human K and chains, where the analysis of 12vK,~ V Aand , 4VlI regions suggests that there are at least 5vK, ~ V A , and 2V, germ line genes, respectively (Smith et al,, 1971; see Fig. 4). The number of hypothetical germ line genes as opposed to available V-region amino acid sequences is as large as could be reasonably expected and will certainly continue to increase as more sequence data become available. Presumably a similar pattern of multiple germ line V genes will emerge as data are gathered on the immunoglobulin families of other species. Thus there appears to be a rapidly escalating number of germ line V genes emerging from genealogical analysis. Finally, the genealogical pattern is not consistent with a somatic recombinational mechanism unless special ad hoc assumptions are made. Recombination would scramble the differences among proteins on different branches of the genealogical tree and lead to a braided structure. This constraint will be examined in detail later (see Section V1,G). In summary, the genealogical pattern suggests that ( 1) there are multiple V genes in most immunoglobulin families (certainly at least ten and perhaps thousands), ( 2 ) the genealogical structure is branched and ordered even at the most-terminal twigs, ( 3 ) the types of amino acid (nucleotide) interchanges seen in each family of V regions are similar to those seen in large sets of evolutionary related proteins such as hemoglobins and cytochromes, and ( 4 ) there is no evidence for frequent somatic recombination. Thus all theories appear to be multigene, somatic, and germ line alike, at least to the extent of ten or more V genes per immunoglobulin family, and, accordingly, all must postulate unusual mechanisms to explain the evolution of immunoglobulin V genes.
C. EVOLUTION OF IMMUNOGLOBULIN VARIABLEGENES The immunoglobulin V regions of one species can be distinguished from those of a second species by two criteria-distinct genealogical patterns (trees) and species-specific residues. Since the V regions of immunoglobulin chains are encoded by multiple germ line genes, each of these patterns poses evolutionary problems which are also seen in two other systems-mouse V-region genetic markers and rabbit allotypy. 1. Species Specificity of Variable Regions The genealogical pattern of the two available complete VKsequences of the BALB/c mouse is distinct from those of their human counterparts, although it is obvious they diverged from a common ancestor (see Fig. 4 ) . Furthermore, the V-region genealogical tree constructed from the amino terminal 23 residues of twenty-two mouse V, regions, when com-
Amino terminal position Chains6 Rabbit K (9) Human
K
0
1
2
Ala Ala Ile Asp Val Phe (41) Asp Ile Glu Glu Val Lys Met
3
4
5
6
7
8
9
Val Met Thr Glx Thr Pro Ma Glx Val Ser Leu Gln Met Thr Gln Ser Pro Ser Val Leu Thr Ala Leu Leu GlY Ile Thr
10
12
13
Ser Val Ser Thr Glx Ser Leu Ser Thr Pro Phe
Glx Ala Val Ala Leu Val Met
11
14
15
16
17
18
19
20
Pro Val Gly Gly Thr Val Thr Ala Ser Val Gly Asp Arg Ala Thr Thr Pro Arg Glu Pro Val Ser Leu Ile Ala
Asp Ile Val Met Thr Glu Ser Pro Ala Glu Val Gln Val Thr Thr Ser Gly Thr Thr Ile Asx Glx Thr Leu Leu Leu Asx
{
U
+
3 Y
ASX
Mouse K (21)
P
Ser Leu Ser Val Ala Ala Gly Glu Arg Val Thr Thr Ala Ala Ser Leu Ser Lys Lys Ala Ser Phe Pro Met Thr Ile Asp Glu Tyr Val Gln Pro Ser Pro
Residue alternatives for the rabbit have been taken from Hood et al. (1970a), for the human from Hood and Talmage (1970), and for t,he mouse from Hood et al. (1970b). The chains have been aligned to maximize sequence homology. In parentheses are shown the number of proteins upon which the alternative residues have been based.
?
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
325
pared against similar amino terminal portions of more than thirty human V, regions, suggests that independent V, genealogical patterns exist in each species (Hood et al., 1970b; see Fig. 2 of Smith et al., 1971). Although limited data are available on all but the human chains, distinct genealogical patterns appear to characterize human (Smith et al., 1971) and mouse (Weigert et al., 1970) A chains, and the heavy chains of man (Smith et al., 1971), mouse (Bourgois and Fougereau, 1970; Hood et al., 1970c), and rabbit (Wilkinson, 1969a; Smith et al., 1971). Whereas there is no problem in explaining why the genealogical patterns of separate species are different, explaining why the V genes on a given branch of the genealogical tree are so similar presents a paradox. For example, each human V, region differs from each mouse V, region by about 40% of its amino acid sequence. This is the same degree of divergence noted between human and mouse C , regions, indicating that the V and C genes of different species are diverging at approximately the same rate. Yet human V, regions on the same genealogical branch may differ by only 15%of their sequence (Hood and Talmage, 1970). How then can the V, genes of man be diverging at one rate from other human V genes and at a second rate from the V genes of other species? This observation suggests that (1) gene duplication has occurred subsequent to the divergence of the species or (2) gene correction occurs to keep multiple germ line genes similar. This paradox is even more clearly expressed in the phenomenon of species specificity. The amino acid alternatives for the amino terminal 23 residues of the K chains of rabbit, human, and mouse illustrate the second criterion that distinguishes immunoglobulin chains from differing species-“species-specific residues” (Table 111). The rabbit chains contain valine and glycine in positions 11 and 17, respectively, whereas their human and mouse counterparts have leucine and primarily aspart‘ic and glutamic acid in the corresponding positions. Since the rabbit V, regions were obtained from restricted antibody responses to the streptococcal antigen (Hood et al., 1970a), heterogeneous populations of normal antibody light chains were also examined on the automatic protein sequencer (Hood et al., 1971a). The same species-specific residues were present in a majority of the heterogeneous chains. Since the homogeneous rabbit V, regions can be assigned to at least three distinct branches on the rabbit genealogical tree (primarily by the presence of sequence insertions or deletions at the amino terminus), at least three germ line genes probably encode rabbit V, regions (Hood et al., 1970a). Most, if not all, of the rabbit V, germ line genes share these species-specific codons. Species-specific residues also distinguish mouse and human V, regions. Three mouse v, regions from different branches of the genealogical
326
L. HOOD AND J. PRAHL
tree (different V, genes) can be compared over 60%of their amino acid sequence with twelve human VK sequences, and at least two positions appear to be species-specific in the mouse chains (see Smith et al., 1971; Hood et al., 1 9 7 1 ~ ) .Species-specific residues are also present in the chains of pig, human, mouse, and bird (Grant and Hood, 1971b) and the heavy chains of human, mouse, and rabbit (see the compilation of data in Smith et al., 1971). The genetic significance of speciesspecific residues is similar to that discussed for species-specific genealogical patterns and will be discussed after a description of two related phenomena-mouse V-region genetic markers and rabbit V-region allotypes. 2. hlouse Variuble Region Genetic Markers Edelman and Gottlieb (1970) examined the half-cystine peptides obtained by tryptic digestion of the light chains from seventeen inbred strains of mice and identified a V-region peptide in three strains not seen in the other fourteen. Preliminary breeding experiments indicated that this difference segregated in a Mendelian fashion. They suggested that this V-region marker must be encoded by one (or a few) germ line V genes (using arguments similar to those discussed for the singularity of the CK gene in man; see Section IV). Unfortunately, normal mouse V, regions are extremely heterogeneous (Hood et al., 1970b), and it is difficult to know whether this observation represents a qualitative or a quantitative difference among the positive and negative strains. This peptide ( V gene) may be expressed in the fourteen negative strains at levels below the limits of detection in the method used. If so, this peptide merely reflects differences in the control (or selective) mechanism which expresses VKgenes in the positive and negative strains, If the difference should prove qualitative, it suggests that the V genes of closely related evolutionary lines (inbred strains of mice) can diverge rapidly from one another. Indeed, other preliminary experiments suggest that V-region strainspecific differences may be present in differing inbred strains of mice. Hood and Warner (1970) have examined the V regions (amino terminal 20 residues) of three myeloma K chains from the inbred NZB strain. Each of these sequences differs from those of more than 30 BALB/c K chains. Although more data must be gathered before concluding that the genealogical patterns of immunoglobulin V genes differ among the inbred strains of mice, these data are provocative. Again the genetic implications of these observations will be discussed after a consideration of rabbit V-region allotypes.
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
327
3. Rabbit Variable Region Allotypy
The genetic significance of the group a specificities (a1-3) is unclear at this time. As indicated earlier (see Section III,F), these markers behave as alleles at a single genetic locus and are found on the Fd portion of all rabbit heavy chains-y, a, p, and E . This implies that the group a markers are present on the variable region, as the V, regions are probably shared by all classes of heavy chain (Todd, 1966). Amino acid analyses of the Fd piece and of peptides derived from the Fd piece (by chemical cleavage with cyanogen bromide) of antibody and normal rabbit IgG revealed amino acid variations which correlated with the group a specificities (Koshland, 1967; Prahl and Porter, 1968). Wilkinson (1969a) demonstrated that normal (heterogeneous) a1 and a3 y chains have “allotype-related residues” at the amino terminus (positions 1 3 4 ) which correlate with the corresponding allotypes. Allotype-related sequence variation apparently extends at least to position 94 in the V, region of rabbit y chains (Mole et al., 1971). Preliminary data suggest that similar allotype-related variations are present in the VII regions of IgA (Wilkinson, 196913) and IgM (Koshland et al., 1969). The existence of serologic markers behaving in a Mendelian fashion which correlate with differing V, sequences suggests that the VI, regions of rabbits are encoded by a single structural gene with three alleles. There are, however, certain observations which render this simple interpretation unlikely. Extensive amino acid differences distinguish these reputed alleles (eleven differences in the amino terminal 34 residues of a1 and ~3 chains ) and, accordingly, distinguish them from other established allelic systems (the normal p and sickle cell p gene of human hemoglobins are alleles that differ by 1 residue in 146). In addition, rabbits of the a3 allotype have two sets of V,, sequences (Wilkinson, 1969a), indicating that there are at least two major branches on the V,, genealogical tree and, accordingly, at least two V, genes. It remains to be determined whether both of these major sequences exhibit allotypic specificity. Also, numerous technical difficulties underlie the analyses of V,, regions ( Wilkinson, 1969a). First, the heavy chains examined were derived from normal or antibody IgG and therefore are heterogeneous. Second, the VrIregions of rabbit have a blocked a-amino group and cannot be examined by the routine Edman degradation procedure. Third, this has necessitated the fractionation of V,, peptides which, in most cases, were obtained in low yield (1530%in Wilkinson’s study, 1969a). Thus it is uncertain whether the major sequence alternatives truly represent the entire VI1population and whether minor sequence alternatives were overlooked.
328
L. HOOD AND J. PRAHL
The use of restricted rabbit antibodies will be extremely useful in clarifying this problem ( Krause, 1970; Haber, 1970). For example, Fleischman ( 1971) reported the sequence of the amino terminal 64 residues of a restricted a2 VFIregion of the y class. Of the eleven amino acid differences which appeared to be allotype-related in comparing the amino terminal 34 residues of heterogeneous a1 and a3 y chains, six were present in the a2 sequence. Thus the preliminary observations made on heterogeneous rabbit heavy chains may be misleading, and great caution should be exercised in drawing conclusions about genetic or evolutionary mechanisms from the rabbit group a specificities at this time. A correlation among four light chain serologic markers (b4, b5, b6 and b9) which segregate in a Mendelian fashion and distinct VK-region amino acid profiles has also been noted in rabbits (Hood et al., 1971a; Chersi et al., 1971). The V sequences from b4 rabbits can be divided into three major branches on the V, genealogical tree (see Section V1,B) (Hood et al., 1970a; Smith et al., 1971). Homogeneous light chains from each of these major branches have the b4 allotype marker (Kindt, 1971). Variable regions homologous to those seen in each of the three major branches of the V, genealogical tree in b4 rabbits appear to be present in rabbits homozygous for the b5, the b6, and the b9 markers. The amino acid alternatives in rabbit antibody light chains (heterogeneous ) of the same allotype are very similar to one another in their amino terminal 20 residues, whereas the residue profiles from light chains of differing allotypes can readily be distinguished from one another (Hood et al., 1971a). Since b4 light chains from each of the three branches of the V, genealogical tree express the allotype marker to the same extent (Kindt, 1971) and since amino acid differences exist around certain of the C , half-cystine residues in b4 and b5 light chains (Appella et al., 1969), the group b specificities appear to correlate with C region residue differences. Why then are the V-region amino acid profiles characteristic of each light chain allotype? There are two general possibilities. First, rabbits from all four allotypes may have the same pool of V, genes (see Loeb, 1968; Rivat et al., 1970). A special control mechanism may lead to the expression of different subsets of the V, genes for each allotype. For example, let us suppose that the joining mechanism for V, and C, has four allelic alternatives which can join distinct subsets of the V, library to each of the four C, alleles. Presumably such a mechanism would involve distinct recognition sites for each allotype although these recognition sites need not be present in the resulting polypeptide chains (Dreyer et al., 1967). Alternately, the V, genes may be distinct for rab-
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
329
bits of each allotype. Thus the genome from rabbits of the b4 allotype would have VKgenes which are different from those found in b5 rabbits, If true, we must ask ( 1 ) how V-region differences associated with allotype can evolve in multiple VKgenes and ( 2 ) why these multiple genes do not recombine in the germ line with VK genes of other allotypes to scramble the allotype-related differences. The first question is similar to that raised by the evolution of species-specific residues in multiple genes and will be discussed shortly. The second requires a mechanism for preventing meiotic recombination (see Smith et aZ., 1971, for a more thorough discussion of this point). Two points emerge from a consideration of rabbit V-region allotyperelated diiferences. First, although neither light nor the heavy chain systems have been adequately characterized, there are striking similarities between the sequence and genetic observations. In both, V-region residue alternatives seem to correlate with allotype; in both, there probably are multiple germ line V genes; finally, in both, there is uncertainty as to the precise location of the serologic (allelic) marker. Thus similar mechanisms may well be operant in producing antibody diversity and allotypy in both the light and heavy chains of rabbits. Although one can postulate vague and complex genetic mechanisms to explain allotypy (see above), the final solution may well be one that has not yet been considered. It is our feeling that no firm genetic conclusions can be drawn from rabbit allotypy at this time. In summary, multiple germ line V genes of each immunoglobulin family (+lO-lOOOs, depending on one’s bias) evolve such that speciesspecific residues (codons) permit one to distinguish most, if not all, of the V genes in one evolutionary line (e.g., mouse) from those in a second evolutionary line ( e.g., rabbit) (see Fig. 4 and Table 111).There are indications that evolution in a multigene system may occur rapidly enough to generate VK-regiondifferences in inbred strains of mice (Vregion genetic markers) or even, possibly, in different populations of rabbits (rabbit V-region allotypy ) . All theories of antibody diversity require multiple germ line V genes, and, accordingly, evolutionary mechanisms must be postulated to explain these phenomena. We shall now discuss other multigene systems also exhibiting species specificity. D. OTHERMULTIPLEGENESYSTEMS With DNA-DNA hybridization techniques, it has been demonstrated that families of repeated sequences are present in all higher organisms ( eukaryotes) ( Britten and Kohne, 1968; Britten, 1!369). These families contain 50-2 million related sequences and comprise from 20 to 80% of the total nuclear DNA. Certain of these repeated sequences must be
330
L. HOOD AND J . PRAHL
transcribed and perform some function in eukaryotes since RNA complementary to some of the repeated DNA sequences has been observed in every cell type studied. Certain of these multigene families can diverge rapidly from one another, suggesting at the DNA level a phenomenon analogous to the species specificity of genealogical trees observed at the protein level. Hennig and Walker (1970) have compared the DNA from 16 members of two rodent families and conclude that even in closely related families the banding profiles in cesium chloride and the reassociation rates of the DNA exhibit significant differences. Thus individual families of genes must be able to change rapidly such that, for example, those of the rat (Rattus norvegicus) can readily be distinguished from those of the mouse (Mus musculus). At present three multigene families with known functions have been investigated. The transfer RNAs (tRNA) of higher organisms are represented by repetitive genes. Each of the tRNAs of Drosophik appears to be encoded by a minimum of fifteen genes (Ritossa et al., 1966). Kedes and Birnstiel (1971) have obtained data from the sea urchin consistent with the interpretation that the genes for the histones are reiterated ( 400-fold repetitive ) , closely clustered, and potentially separable from the bulk of the nuclear DNA. Finally, the number of ribosomal genes has been estimated to range from 100 to 400 per haploid set of chromosomes in species as varied as yeast (Retel and Plauta, 1968), Drosophila ( Ritossa and Spiegelman, 1965), frogs (Wallace and Birnstiel, 1966), and man (Attardi et al., 1965). The ribosomal system has been studied in detail because of the GC7 content of these genes is high, enabling their separation from the bulk of the nuclear DNA. In the frog, three genes are tandemly linked as a “single ribosomal unit”the 18 S, the 28 S, and the spacer gene (Brown and Weber, 1968). Five hundred germ line copies of this single ribosomal unit are clustered at a single complex locus. At present it is a matter of controversy whether these 500 germ line copies are identical or merely very similar within a species. However, it is known that the ribosomal genes from species widely separated on the phylogenetic tree can differ markedly in size (Perry et al., 1970), GC content (McCai-thy, 1969), and base sequence (Pinder et al., 1969). Thus, the ribosomal system is a functional, clustered, multigene system which can evolve in a species-specific fashion. If all of the ribosomal genes are identical to one another in a given species, the mechanism responsible for evolving species specificity in ribosomal genes must be different from that for immunoglobulin V genes which show enormous diversity. Mouse satellite DNA is, however, an example of a multigene system with nonidentical genes that can evolve
’ ( G C ) Guanine-cytosine.
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
331
in a species-specific fashion. The mouse (Mus musculus) genome has a satellite DNA which can be isolated from chromosomal DNA because of its unusual GC content (Walker et d., 1969). This satellite comprises 10%of the total nuclear DNA and represents approximately 10 million copies of similar but nonidentical nucleotide sequences less than 10 residues in length (Southern, 1970). Hybridization studies demonstrate the satellite DNAs of mouse (Mus musculus) and the guinea pig (Labia porcallus) share little if any sequence homology, nor could homologous sequences be found in the genomes of the rat (Rattus m r vegicus), the European wood mouse ( Adodemas sylvaticus), or two strains of North American deer mouse (Peromyscus municulatus and P . pohonotus) (Walker et al., 1969). The satellite genes are scattered throughout the mammalian genome around each centromere. There is no evidence that satellite DNA is ever transcribed. Nevertheless, this multiple gene system with nonidentical copies can evolve species specificity in relatively short periods of evolutionary time (the divergence of these rodents probably occurred in the past 1-5 million years). In summary, there are documented examples of multigene systems, apart from the immunoglobulin V genes, which evolve in a speciesspecific fashion. The question of whether these other multigene systems are adequate analogies for the immunoglobulin V-gene system is a fair one, but they do establish unequivocally that there are evolutionary mechanisms for generating species specificity in sets of germ line genes which may be nearly identical (ribosomes) and in sets of genes that are nonidentical ( immunoglobulin V genes and satellite DNAs). Furthermore, many as yet unidentified families of repeating sequences also exhibit species specificity. How can multiple germ line genes evolve so as to be species specific?
E. EVOLUTIONARY MECHANISMS FOR MULTIPLE GENE SYSTEMS The problem of species specificity is, simply, How can multiple germ line genes share codons at a few (immunoglobulin V genes) or many positions ( ribosomal genes ) in one evolutionary line completely different from their counterparts in a second evolutionary line? Theoretically this could be achieved by random mutation followed by intense selection (Fig. 5A), by a gene conversion or coevolution process (Fig. SB), or by the rapid and continual duplication and deletion of the germ line gene population (Fig. 5C). Let us consider each of these possibilities. Contemporary multiple genes may have evolved directly from their primordial ancestors ( without significant gene duplication or deletion ) with species specificity generated by random mutation and fixed in the population by natural selection (Fig. SA). This model is unattractive
332
L. HOOD AND J. PRAHL
FIG. 5. Models accounting for the generation of species-specific residues in a multigene system. The @ and * represent germ line genes bearing differences which are to become species-specific. ( A ) Random mutation changes the ancestral qi to rabbit * genes which are then fixed by selection. ( B ) The ancestral @ genes are converted to rabbit ' genes by the process of pairing and copy repair. ( C ) The ancestral genes are expanded by replication and incorporated into the genome of the line leading to contemporary species, such that ' genes are found in rabbit and 9 genes in man. Later expansions of ' and qi genes may give rise to V-region subgroups. Thus, species specificity arises because of the expansion of different germ line genes in the two lines. (Reprinted from Hood et al., 1970a, with permission.)
because it is difficult to imagine selective pressures which could fix identical mutations in each of 500 ribosomal genes. A family of contemporary genes may have evolved as the result of a gene correction (coevolution) process which resulted in parallel evolution ( again without significant gene duplication or deletion). Different evolutionary lines would follow independent paths and species
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTLATION
333
specificity would ensue (Fig. 5B). Two models have been proposed differing as to whether the gene family is comprised of identical members, master-skve model (Callan, 1967), or of similar but nonidentical members, democratic gene conversion model (Edelman and Gally, 1971). Callan (1967) proposed a master-slave model in which multiple gene copies (slaves) are routinely corrected against a single (master) copy at each cell division. Hence the entire multigene system would always be identical to the master gene (also see Whitehouse, 1967; Thomas, 1971 ) . This mechanism cannot, however, apply to antibody V regions, as they are not all identical. Edelman and Gally (1971) have proposed a variant of the master-slave model which hypothesizes that favorable mutations can spread through a population of similar genes by “democratic gene conversion.” In this model (Fig. 5B) any of the multiple genes that has incurred a mutation can spread it to neighboring genes by a gene conversion or copy repair mechanism (Fogel and Mortimer, 1969). The democratic gene conversion model is unattractive because two ad hoc assumptions must be made to explain, respectively, the genealogical pattern and species-specific residues. First, if any one V gene in the genealogical tree could convert (correct) all other genes, it would not be possible to maintain distinct genealogical branches. Thus democratic gene conversion postulates that there are distinct sets (subgroups) of V genes (at least one subgroup for each major branch on the genealogical tree). It is suggested that there are 10-100 V genes per subgroup and that gene conversion can occur only among the V genes of a given set (Gally and Edelman, 1970). The V genes in a subgroup would, therefore, serve as a net to trap favorable mutations, and gene conversion could spread these favorable mutations to other members of that set. In this fashion the genes of each subgroup could coevolve; the coevolution of all of the subgroups in various species should be independent and would, of course, lead to genealogical patterns that are species-specific. Such a postulate escalates enormously the number of germ line genes (10-100 per subgroup). A second ad hoc assumption must be evoked to explain species-specific residues which are common to all of the V genes of a given immunoglobulin family. For example, all human VK regions have leucine, whereas all rabbit V, regions have valine at position 11. Since multiple V-region subgroups are required by the gene conversion model, each of these subgroups must have independently evolved identical species-specific residues. As the number of V-region branches ( subgroups ) increases, this ad hoc assumption becomes increasingly unattractive. The mechanism of democratic gene conversion must add,
MATERNAL CHROMOSOME
PATERNAL CHROMOSOME
EXPANDED CHROMOSOME
I
HOMOLOGOUS BUT UNEOUAL CROSSING OVER
+ CONTRACTED CHROMOSOME
FIG.6. Gene duplication and deletion (expansion and contraction) of homologous genes on chromosomes . by homologous but unequal crossing over. (Reprinted from Smith et al., 1971, with permission).
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
335
therefore, the twin ad hoc assumptions of parallel mutation and selection for species specificity in all V-region subgroups and gene conversion within, but not between, subgroups to preserve the genealogical pattern. Species specificity in multigene systems may also ensue as a result of gene duplication and gene deletion (gene expansion and contraction). The gene expansion-contraction model postulates that species specificity arises by the duplication and deletion of different genes in different evolutionary lines (Fig. 5C). Two mechanisms have been proposed for this process. First, unequul but homologous crossing over appears to be a common event, as exemplified by the bar locus of Drosophila (Peterson and Laughnan, 1963). This mechanism, depicted in Fig. 6, leads to chromosomes with extra genes (expansion) and chromosomes with deleted genes ( contraction). Homologous but unequal crossing over need not be cataclysmic, for with frequent (perhaps as frequent as nucleotide mutation) expansion and contraction of even 5% of the multiple gene population, new genes may replace the old in relatively short periods of evolutionary time. Thus to evolve by gene expansion and contraction, at no time need individual organisms lose more than a few percent of their multiple genes. Second, species specificity may originate in rather sudden events of excessive replication of particular sequences, saltutory replication ( Britten and Kohne, 1968).The rapid appearance of various satellite DNAs suggests that they may have originated by such a mechanism. Possible models for such a mechanism are discussed by Britten and Davidson ( 1971). Any mechanism that postulates frequent gene duplication (or deletion) must acknowledge the powerful but unknown control mechanism( s ) which regulates the levels of genomic DNA in higher organisms. The haploid DNA contents of various mammalian species differ by less than 10%(Atkin et al., 1M). The selective pressures to fix duplications of useful genes appear to be balanced by corresponding pressures to fix the deletions of useless ones.
F. PROBLEMS OF CONTROL Control in the immune system operates at many levels, and we shall indicate some of the more interesting problems. 1. Differing h-to-K Ratios
The X / K ratio8 varies widely in various mammals from species which express a predominance of one chain type ( A or K ) to those which have intermediate levels (compare horse, man, and rat in Fig. 7) (Hood
' The X / K ratios were determined by quantitating the ratio of certain characteristic C-region peptides for each chain type (Hood et al., 1967).
w w
0)
Rabbit
Mouse
Lagornorphr
Rat
Rodents
Guinea Rhesur Baboon Human Whale
Primates
Cetaceae
Dog
Cat
Carnivores
Mink
Pig
Bovine
Goat
cArtiodactylr-(
Sheep
Mule
Horse
Perirsodacwls
I
I
Mammals
FIG.7. Distribution of and K light chains among various mammals, based on data of Hood et al. (1967, 1971b). The vertical bar on the left above each species indicates proportion of light chains in K class; right-hand (shaded) bar indicates A chains. (Reprinted from Hood et al., 1971b, with permission).
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
337
et al., 1967, 1971b). Generally those animals that are closely related on the phylogenetic tree share similar A / K ratios (compare sheep, cow, and horse in Fig. 7). These ratios are stable among individual members of a given species (Grant and Hood, 1971a) unless special procedures are used to suppress one chain type (Appella et al., 1968). Differing serum light chain ratios could reflect (1) the A / K ratio of germ line V genes, ( 2 ) control mechanisms which can preferentially express one chain type or (3) the selection of particular A / K ratios by various environmental factors. Man expresses X / K chains in a 1:2 ratio (Fahey, 1963), and comparable degrees of V, region diversity are noted in the myeloma proteins of both light chain types (Hood and Talmage, 1970). In contrast, where a minor chain type ( < a few % of the serum light chains) has been examined with myeloma proteins (mouse X chains), a very restricted degree of heterogeneity was observed ( Weigert et al., 1970). Thus the A / K ratio may, in part, reflect differences in the ratios of germ line VA and V, genes (see Section VI,B, 2). 2. Light and Heavy Chain Pairing Are light and heavy chains randomly associated with one another or are there certain factors that permit only certain subsets of light chains to join with certain other subsets of heavy chains? Light and heavy chain reassociation experiments have indicated a preferential degree of reassociation by homologous as opposed to heterologous light and heavy chains from myeloma proteins (Mannik, 1967). Recent experiments on rabbit immunoglobulins place in question this concept of preferential association ( Bjork and Tanford, 1971). Furthermore, since the light and heavy chain families are genetically unlinked in the mammalian genome, it is difhult to conceive of control mechanisms to regulate the association process. Random association of light and heavy chains with subsequent selection for the “best fitting pairs” at the cellular level is an attractively simple model although the actual association process may be more complex.
3. Random Association of Constant and Variable Regions Can any V region associate with any C region of the same immunoglobulin family? Although the data are totally inadequate to answer this question, Bennett (1968) made the observation that certain V, sequences seem to be associated only with Cp regions. Sufficient VH data now exist to show this association is not due to chance selection ( Kaplan et al., 1971). Perhaps these proteins represent a recently evolved immunoglobulin family in which at least one Cp and a subset of V, genes were translocated away from the
338
L. HOOD AND J. PRAHL
major CH locus. Thus the translocated C , could only join with those V, genes that were also translocated. Alternatively, it may suggest that V and C genes of a given family may be intermingled. 4. Selection in the Myeloma System
Selection appears to occur for certain V and, on occasion, for certain C regions of myeloma proteins. Selection of V, regions in man has been suggested by the observation that certain residue alternatives are found in the light chains from normal individuals at 2 to 5 times the level they are observed in the myeloma population (Grant and Hood, 1971a). This suggests that chains from at least one major branch on the light chain genealogical tree comprise up to 20% of the normal serum light chains, whereas they are infrequently (less <4%) seen in the myeloma population. Furthermore, the myelomas induced with mineral oil in BALB/c mice are often of the IgA type (60%),whereas the normal serum level of IgA is much lower (Potter, 1970). The selection for the C, regions may, in part, reflect the fact that the myelomas are artificially induced in the peritoneal cavity which has a correspondingly high population of IgA-producing cells. In any case, it appears that V- and C-region selection can occur in the myeloma process and that any generalizations from the myeloma system must be tempered by the realization that only a subset (of unknown size) of the normal immunoglobulin population is expressed.
5. Differentiation of the Immune Response to Antigens Studies carried out in the fetal lamb by Silverstein and Prendergast (1971) on the maturation of the immune response are provocative. Silverstein and his co-workers have demonstrated that immunological competence to all antigens does not arise simultaneously in the developing fetus-rather, there is a stepwise maturation of immunological competence to different antigens at different stages of development (Silverstein et al., 1963; Sterzl and Silverstein, 1967). Among the antigens tested, antibody formation occurs first to bacteriophage +X174, later in gestation to ferritin and hemocyanin, and even later to ovalbumin. The immune response continues to mature even after term. The precision with which the fetus develops competence to a given antigen at a given stage of gestation is remarkable. For example, the fetal lamb developed competence to ovalbumin at 120 to 125 days of gestation (150 days gestational period) in some 60 animals tested (Silverstein and Prendergast, 1971). Thus the ability to respond to a series of antigens appears to be carefully controlled and temporally reproducible sequence within a given species.
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
339
6. Genetic Control of the Immune Response
The control of the immune response to certain antigens (high or low response) seems to be linked to the histocompatibility locus of the mouse.9 We shall not discuss these and similar observations concerned with the genetic control of the immune response as it appears this phenomenon is not directly related to the genetic control of antibody V genes (Mozes et al., 1969). This subject has recently been reviewed by McDevitt and Benacerraf (1969) and Milstein and Munro (1970).
G. SUMMARY:IMMUNOGLOBULIN PATTERNS AND THEORIES OF ANTIBODY DIVERSITY Immunoglobulin patterns from genetic, structural, and cellular studies have removed much of the previous simplicity of the somatic theories. The V-region genealogical patterns suggest that all theories have multiple germ line V genes and, indeed, the number seems to be increasing in proportion to the amount of sequence data available. Thus all somatic theories must postulate multiple germ line V genes-most advocates for this point of view concede 10-50 germ line V genes per immunoglobulin family (Cohn, 1971a,b; Gally and Edelman, 1970; Jerne, 1971).If this is so, then the somatic theories lose their most powerful raison d'dtre, namely their ability to explain the phenomena of species specificity (and rabbit V-region allotypy) as the evolution of one (or even two) germ line V genes. Thus, somatic and germ line theories alike, as we have previously indicated, are multigene and must postulate unusual evolutionary mechanisms for species specificity and related phenomena. Accordingly, we fail to see any compelling reason to make the ad hoc assumption that antibody V-gene information can be expanded during somatic differentiation by a mutational or recombinational mechanism. Indeed, upon closer examination, the genealogical pattern is incompatible 'This and other related observations have led Jerne to propose a new somatic theory of antibody diversity (Jerne, 1971). He postulates that the germ line carries a set of V genes (-10-20) which are structural genes for antibodies directed against the histocompatibility antigens of the species. Those cells that express genes making antibodies against the individual's own histocompatibility antigens are stimulated during embryonic development and subsequently are suppressed during the remainder of somatic differentiation. Mutant clones emerge from the suppressed cells by somatic mutational events. These mutant clones no longer combine with the histocompatibility antigens of the individual and, accordingly, they are no longer suppressed. These mutants proliferate and generate the functional antibodies observed in the adult animal. Thus Jerne has very elegantly provided the powerful selective force which is essential to most somatic theories (see Section V1,G).
340
L. HOOD AND J. PRAHL
with either popular form of the somatic theory unless additional ad hoc assumptions are made. Random somatic mutation should generate a pattern in which the V sequences eminate in a random and unstructured fashion from the germ line V gene (nodal point) in a genealogical pattern. Since the actual V-region sequences show a high degree of restriction even at the terminal-most twigs of the genealogical tree (mouse VA regions), two possible ad hoc explanations can render the somatic mutation theory compatible with the ordered genealogical structure. 1. There may be a special (programmed) hypermutational mechanism (Brenner and Milstein, 1966) which only operates on limited areas of the V gene (i.e., the hypervariable regions) during somatic differentiation. This suggestion is unattractive in that ( a ) the kinds of base substitutions observed in the V regions are similar to those observed in other sets of evolutionary proteins by a variety of criteria and, accordingly, offer no support for an unusual mutational mechanism (Hood and Talmage, 1970), ( b ) it is an ad hoc mechanism for which there is no precedent, and ( c ) it is difficult to imagine how such a mechanism could operate on V and not C genes without an additional ad hoc proposal. 2. Powerful selective forces could act preferentially to expand clones of cells in which advantageous V-gene mutations had occurred by ordinary somatic mutation.1° Thus somatic selection must be of sufficient magnitude to expand over the parental immunocyte clone, a mutant immunocyte clone which changes 1 amino acid residue in 220 ( ~ 1 1 for 0 VL and the same for V,). To explain the mouse VA gene with “four” mutations, for example, a somatic mutational theory must posit that mutant clone 1has sufficient selective advantage to overgrow the parental clone and expand to such a size that a second favorable mutation might occur. Then mutant clone 2 must have sufficient selective advantage to overgrow mutant clone 1, and so on, through four discrete expansions. It is difficult to imagine somatic selective forces which can operate so effectively during such a brief time scale-the lifetime of the individual. Powerful selective forces can operate during the evolution of the species (Nolan and Margoliash, 1%8), and it is reasonable to ask about utilizing evolutionary selective forces in conjunction with gene duplication and mutation to generate antibody V-region diversity. It is again an ad hoc postulate to suggest that powerful selective forces can shape the entire information content of the immune system during the differentiation of the individual. lo These selective forces have included antigen (Cohn, 1968), light-heavy chain pairing (Cohn, 1968), and self-antigens (Jerne, 1971). This latter theory is particularly clever in its use of negative selection pressures to generate mutant clones of immunocytes (see footnote 9 ) .
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
341
A second case in which a genealogical pattern would not be expected is that of somatic recombination. If V, genes from differing branches of the V, genealogical tree could recombine, the resulting structure would be one of braided and not discrete branches. For example, in Fig. 3, if the genes encoding V, regions in branches VKI and VKIIIdid recombine such that half of the resulting V gene came from each parent, it would be impossible to place the corresponding V region on the V, genealogical tree because the base substitutions and deletions would be derived from the two distinct genealogical branches. It would be no more correct to place the recombinant protein on the VKl than on the VKIIlbranch. Smith et al. (1971) uscd a new method to search for recombinant proteins. Genealogical trecs were generated at each residue position for V regions from each of the major branches (subgroups) of h, K , and H. The power of this analysis can be illustrated with the following example (see Fig. 3 ) . A recombinant protein might start in K branch IA for residues 1-6, switch to K branch I1 for residues 7-9, return again to branch IA for residues 10-68, switch back to I11 for the next few residues, etc. The generation of genealogical trees at each residue position permits one to detect these multiple and scattered recombinational events, This method revealed a single questionable case of recombination in the analysis of 21 complete (or nearly complete) V sequences. If somatic recombination is responsible for antibody diversity, one would expect that a majority of the V regions should be recombinants. Finally, as we discussed earlier, two additional ad hoc assumptions must be made to explain V-region genealogical patterns and species specificity (see Section VI,E,2). In addition, it is difficult to see why gene conversion would not correct ( eliminate) the hypervariable regions ( see Smithies, 1971, for additional conceptual difficulties with democratic gene conversion). In our opinion the total lack of evidence for a recombinational mechanism in addition to all the necessary ad Aoc assumptions makes this model the least attractive of the contemporary theories of antibody diversity. Thus the escalation of the number of germ line V genes and the loss of the elegant explanation for species specificity (one V gene) make the somatic theories less attractive than the altcrnativc, germ line theory. W e shall list briefly the merits of this theory and then consider in the next section certain of its interesting implications. For those who wish a more balanced discussion of the somatic theories, sce Cohn (1968, 1971a,b), Gally and Edelman ( 1970), Edelman (1971), Milstein and Munro ( 1970), and Milstein and Pink (1970). The germ line theory was initially attractive because studies on cytochromes and hemoglobins had established that phylogenetic evolution
342
L. HOOD AND J. PRAHL
can generate protein diversity-even among multiple genes in the same organism (consider the diversity of the human hemoglobin family). The V-region genealogical pattern supports an evolutionary model for diversity, and as the V-region sequence data accumulate, the number of required germ line genes escalates ever upward. A more detailed analysis of the variation occurring among V regions shows that it is similar to that seen in other evolutionarily related sets of proteins. Thus no ad hoc mechanism is required to generate antibody diversity. The multigene nature of the germ line hypothesis is also compatible with recent observations on the multigene families of eukaryotes (Britten and Kohne, 1968; Britten and Davidson, 1971). Multigene families with established functions, such as the ribosomal genes, share certain properties with antibody V genes, namely, both are multigene, closely linked, and exhibit species specificity (see Sections VI,C and D ) . Unusual evolutionary mechanisms must be postulated in both cases to explain species specificity (see Section V1,E). We feel the existence of other multigene systems with properties similar to those of the immune system constitutes a compelling precedent for the germ line nature of antibody diversity. Let us consider in more detail certain of the general implications of a germ line theory.
H. THE GERMLINE THEORY The multigene nature of the germ line hypothesis is also compatible codes each V-region sequence and that these genes arose by ordinary chemical evolution, namely, gene duplication followed by mutation and selection in the germ line. The essential feature of the germ line theory is that antibody molecules should reflect faithfully the germ line genes from which they are derived and not that every immunoglobulin family should have 1000 or even 10,000 V genes. Lower limits can be placed on the number of germ line V genes required for a given immunoglobulin family by examining randomly selected V regions for repeats (identities). Statistical calculations based on the nonidentity of fifty-two randomly chosen human K chains (Quattrocchi et al., 1969) suggest that there are more than 425 VK genes at the 95%confidence level (Goodfliesch, personal communication, quoted in Smith et al., 1971)." In mouse x chains the number of sequences is limited, and the germ line theory " I f the proteins were randomly selected from a V-region pool, one could use the number of proteins screened to obtain the first identical pair to calculate the number of different V sequences in the pool at the 95% confidence level. If each V region is encoded by a separate germ line gene, this would give a rough estimate of the V-gene pool size. Genetic polymorphism in most species may lead to an overestimate of the V-region pool.
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
343
proposes a correspondingly limited number of different VA genes (perhaps as few as 5-20, see Section VI,B,2). How large must the V-gene library be? Two factors operate to reduce the number of V genes required. First, if any light chain can associate with any heavy chain (see Section VI,F,2), the potential number of antibody molecules which can be generated from a library of nV, and mVrr genes is n x m. If only 10% of the light-heavy chain associations are functional, 100,000 antibody molecules could be generated from 1000 V, and 1000 V, genes (1000 x 1000 x 0.1 = 100,000). Thus the association of two polypeptide chains in the immunoglobulin molecule amplifies enormously the number of antibody molecules a given number of V genes can produce. Second, a single antibody molecule can probably cross-react with a variety of different antigens sharing certain three-dimensional features ( antigenic determinants), For example, myeloma antibody MOPC 315 reacts with the apparently unrelated molecules, DNP and menadione, with affinity constants of lo7 and 5 x lo5 liters mole-', respectively (Eisen et al., 1970). The multiplicity of reactions that a single antibody molecule can undergo correspondingly reduces the total number of antibody molecules required to cover the antigenic spectrum of a given environment. Vertebrate responses to chemicals an animal (species) has never seen can be simply explained by shared antigenic sites between the new molecule and some antigen the animal has seen in its past evolutionary history. Thus vertebrates may only need 100,000 antibody molecules for immunity, and these could probably be encoded by 2000 (or fewer) germ line V genes. What percent of the vertebrate genome will encode antibody genes according to germ line theory? The amount of haploid DNA in a human germ cell is sufficient to encode lo7 genes the size of a V gene (see Hood and Talmage, 1970). Thus 2000 V genes would require 0.02%of the haploid DNA-not an unreasonable requirement for a system as vital as immunity. How does selection work in a multigene system if the genes share a common function (ribosomes) or if they share a series of diverse but overlapping functions (antibodies)? In the frog, it is difficult to believe that selective pressures can act to eliminate one nonfunctional mutation out of 500 successfully functioning ribosomal genes. In fact, frogs can lose half of their ribosomal genes and still develop normally (Brown and Gurdon, 1964). Thus, selection for multigene systems probably occurs at the level of the whole organism and not at the level of individual germ line genes. In the casc of immunity, the organism can meet pathogenic stimuli (plague bacillus) with a variety of different re-
344
L. HOOD AND J. PRAHL
sponses because of the extensive cross-reactive nature of antibody niolecules. It does not matter whether a particular antiplague V gene is present so long as the host can mount an effective immune response using other V genes, If homologous but unequal crossing over is a frequent event in multiple gene systems (see Section VI,E,3), then individuals should be available with many different combinations of V genes which can, accordingly, be selected by the environment. Thus selection, acting on the products of gene expansion and contraction, may enable organisms to have a rapid and flexible response to their ever-changing environments. Successful organisms will pass their successful V-gene chromosomes on to progeny and the unsuccessful will not. In this light, selection probably acts in response to the overall effectiveness of the immune system and not at the level of the individual antibody V genes. Finally, if information for immunity is encoded in the germ line, perhaps the entire library of V genes is read out in a programmed fashion during the maturation of the immune response (Dreyer and Gray, 1968). The existence of such a programmed readout mechanism is suggested by Silverstein’s observations on the precise and reproducible temporal sequence of immunological maturation to a given spectrum of antigens (see Section III,F,5). This concept is attractive because the generation of information for immunity is analogous to the programmed readout of a computer tape-each antibody V gene (“bit” of information) is “readout” and its expression through the differentiated immunocyte is not left to the chance vagaries of random somatic mutation or reconibination. For speculation on the nature of this programmed mechanism, see Dreyer et al. (1967), Dreyer and Gray (1968), and Dreyer (1971). VII. Concluding Remarks
The immune system has provided profound insights into two of the most basic questions of development: ( I ) How is information stored in the vertebrate genome? ( 2 ) How is this information usefully expressed? Information for the immune system seems, in part, to be encoded as germ line genes because of the V-region genealogical pattern. There do not appear to be any compelling reasons for justifying ad hoc somatic mutational or recombinational mechanisms and, indeed, the genealogical pattern places additional ad hoc constraints on the somatic mechanisms. Antibody polypeptide chains seem to bc encoded by two genes, a V and a C, which are united at some level of protein synthesis by a joining mechanism. If this joining occurs at the level of DNA, then, perhaps, the translocation of V and C genes provides a specific mecha-
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
345
nism for the commitment of each immunocyte to the synthesis of one “bit” of information-one VL and one V, gene. Antibody molecules may have evolved from other types of cell surface receptor molecules. Thus the general strategy of the immune system ( a library of germ line V genes; separate C genes; a joining mechanism) may be employed by other complex systems in higher organisms. In this regard, the repeated DNA sequences of eukaryotes presents a compelling precedent for the existence of other multigene systems which share certain features (species specificity) with antibody V regions. Certainly unusual evolutionary mechanisms must be posited to account for the evolution of these multigene systems. The immune system will continue to be an exciting and fruitful model for the study of information storage, expression, and evolution in higher organisms. ACKNOWLEDGMENT We wish to thank Mrs. Marijo Valenciana for her patience and efforts in the preparation of the manuscript.
REFERENCES Alexander, J. (1932). Protoplasma 14, 296. Allen, J. C., Kunkel, H. G,, and Kabat, E. A. (1964). J. E x p . Med. 119, 453. Appella, E., Mage, R. G., Dubiski, S., and Reisfeld, R. A. (1968). Proc. Nut. Acad. Sci. U . S. 60, 975. Appella, E., Rejnek, J., and Reisfeld, R. A. (1969). J. Mol. Biol. 41, 473. Atkin, N. B., Mattinson, G., Becak, W., and Ohno, S. (1965). Chromosoma 17, 1. Attardi, G., Huang, P-C., and Kabat, S. (1965). Proc. Nut. Acad. Sci. U.S. 54, 185. Ballard, H. S., Hamilton, L. M., Marcus, A. J., and Illes, C. H. (1970). N . Engl. J. Med. 282, 1060. Bennett, J. C. (1968). Biochemistry 7, 3340. Bjork, I., and Tanford, C. (1971). Biochemistry 10, 1289. Bourgois, M., and Fougereau, M. (1970). FEBS Lett. 8, 265. Braunitzer, G. (1965). J. Cell Physiol. 67, Suppl. 1, 1. Breinl, R., and Haurowitz, F. (1930). Z. Phys. Chem. 192, 45. Brenner, S., and Milstein, S. (1966). Nature (London) 211, 242. Britten, R. J. (1969). Carnegie Inst. Wash., Yearb. 68, 376. Britten, R. J., and Davidson, E. (1971). Quart. Reu. Biol. ( i n press). Britten, R. J., and Kohne, D. E. (1968). Science 161, 592. Brown, D. D., and Gurdon, J. B. (1964). Proc. Nut. Acad. Sci. U . S. 51, 139. Brown, D. D., and Weber, C. S. (1968). J. Mol. Biol. 34, 661. Burnet, F. M. (1956). “Enzyme, Antigen and Virus.” University Press, Cambridge. Burnet, F. M. ( 1957). Aust. J. Sci. 20, 67. Burnet, F. M. (1959). “The Clonal Selection Theory of Acquired Immunity.” Cambridge Univ. Press, London and New York. Callan, J. D. ( 1967). J . Cell Sci. 2, 1. Capra, J. D. (1971). Nature (London) ( i n press). Capra, J. D., and Hurvitz, A. I. (1970). J. Immunol. 105, 949.
346
L. HOOD AND J . I’RAHL
Capra, J. D., and Kunkel, H. G. (1970). PTOC.Nut. A d . Sci. U . S . 67, 87. Cebra, J. J., Colberg, J. E., and Dray, S. (1966). J. E x p . Med. 123, 547. Cebra, J. J., Givol, D., and Porter, R. R. (1968). Biochem. J. 107, 69. Charlwood, P. A., and Utsunii, S. (1969). Biochem. J. 112,357. Chersi, A., Mage, R. G., and Appella, E. (1971). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 30, 1307 (abstr. ). Clem, L. W., and Leslie, G. A. (1971). Proc. Nut. Acad. Sci. U . S. 68, 139. Cohen, S., and Milstein, C. (1967). Aduan. Zmmunol. 7, 1. Cohen, S., and Porter, R. R. (1964). Aduan. Immunol. 4, 287. Cohn, M. ( 1968). In “Nucleic Acids in Immunology” (D. N. Plescia and W. Braun, eds. ), p. 671. Springer Publ., New York. Cohn, M. (1970). Contr. Process Multicell. Organisms, Ciba Found. Symp. pp. 255303. Cohn, M. (1971a). Cell. Immunol. 1, 461. Cohn, M. (1971b). Ann. N . Y. Acad. Sci. (in press). Cohn, M., Notani, G., and Rice, J. A. (1969). Immunochemistry 6, 111. Cooper, E. L. (1968). Transplantation 6, 322. Cunningham, B. A., Plunim, M. N., Rutishauser, U., and Edelman, G. M. (1969). Proc. Nat. Acad. Sci. U . S. 64,997. David, G. S., and Todd, C. W. (1969). PTOC. Nut. Acad. Sci. U . S. 62, 860 Davies, D., Sarma, R., Silverton, E., and Segal, D. ( 1971). Ann. N . Y. Acad. Sci. (in press). Dintzis, H. M. ( 1961). PTOC.Nut. Acad. Sci. U . S. 47, 247. Dray, S., Dubiski, S., Kelus, A., Lennox, E. S., and Oudin, J. (1962). Nature (London) 195, 785. Dray, S., Young, G. O., and Gerard, L. (1963). J. Immunol. 81, 142. Dreyer, W. J. (1971). In “Developmental Aspects of Antibody Formation and Structure” (J. Sterzl and H. Rhai, eds.), 2nd ed., Vol. 1, p. 919. Academic Press, New York. Dreyer, W. J., and Bennett, J. C. (1965). Proc. Nut. Acad. Sci. U. S. 54, 864. Dreyer, W. J., and Gray, W. R. (1968). In “Nucleic Acids in Immunology” (D. N. Plescia and W. Braun, eds.), p. 614. Springer Publ., New York. Dreyer, W. J., Gray, W. R., and Hood, L. (1967). Cold Spring Harbor Symp. Quant. Biol. 32, 353. Dubiski, S. (1969). I. Immunol. 103, 120. Dubiski, S., and Miller, P. J. (1967). Nature (London) 214, 696. Edelman, G. M. (1971a). In “The Neurosciences: Second Study Program” (F. 0. Schmitt, ed.), p. 885. Rockefeller Univ. Press, New York. Edelman, G. M. (1971b). Ann. N . Y. Acad. Sci. (in press). Edelman, G. M., and Gall, W. E. (1969). Annu. Reu. Biochem. 38, 415. Edelman, G. M., and Gally, J. A. (1967). Proc. Nut. Acad. Sci. U. S. 57, 353. Edelman, G. M., and Gally, J. A. (1971). In “The Neurosciences: Second Study Program” (F. 0. Schmitt, ed.), p. 962. Rockefeller Univ. Press, New York. Edelman, G. M., and Gottlieb, P. D. (1970). Proc. Nut. Acad. Sci. U . S. 67, 1192. Edelman, G. M., Cunningham, B. A., Gall, W. E., Gottlieb, P. D., Rutishauser, U., and Waxdal, M. (1969). PTOC.Nut. Acad. Sci. U . S. 63, 78. Ehrlich, P. (1900). Proc. Roy. SOC., Ser. B 66, 424. Eicher, E. M. (1970). Aduan. Genet. 15, 175.
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
347
Eisen, H. N., and Siskind, G. W. (1964). Biochemistry 3, 996. Eisen, H. N., Simms, E. S., and Potter, M. (1968). Biochemisry 7, 4126. Eisen, H. N., Michaelides, M. C., Underdown, B. J., Schulenberg, E. P., and Simms, E. S. (1970). Fed. Proc., Fed. Amer. Soc. E x p . Biol. 29, 78. Epstein, C. J., Goldberger, R. F., and Anfinsen, C. B. (1963). Cold Spring Harbor Symp. Quant. Biol. 28, 439. Fahey, J. L. (1962). Adoan. Immunol. 2, 41. Fahey, J. L. (1963). J. Immunol. 91, 438. Feinstein, A. (1963). Nature (London) 199, 1197. Fitch, W. M., and Margoliash, E. (1967). Science 155, 279. Fleischman, J. B. (1966). Annu. Reu. Bwchem. 35, 835. Fleischman, J. B. (1967). Biochemistry 6, 1311. Fleischman, J. B. (1971). Biochemistry 10, 2753. Fogel, S., and Mortimer, R. K. (1969). Proc. Nut. Acad. Sci. U . S . 62, 96. Franek, F., and Novotny, J. (1969). Eur. J. Biochem. 11, 165. Franklin, E. C., and Frangione, B. (1971). Proc. Nut. Acad. Sci. U . S. 68, 187. Franklin, E. C., Lowenstein, J., Bigelow, B., and Meltzer, M. (1964). Amer. J. Med. 37, 332. Frangione, B., and Milstein, C. (1969). Nature (London) 224, 597. Gally, J. A,, and Edelman, G. M. (1970). Nature (London) 227, 341. Gilman-Sachs, A., Mage, R. G., Young, G. O., Alexander, C., and Dray, S. (1969). J . Immunol. 103, 1159. Givol, D., and DeLorenzo, F. (1968). J. Biol. Chem. 243, 1886. Glazer, A. N. (1968). Proc. Nat. Acad. Sci. U. S . 59, 996. Goetzl, E., and Metzger, H. (1970). Biochemistry 9, 3862. Grant, J. A., and Hood, L. (1971a). lmmunochemistry 8, 63. Grant, J. A., and Hood, L. (1971b). Biochemistry 10, 3123. Green, I., Vassalli, P., and Benacerraf, B. (1967a). J. Exp. Med. 125, 527. Green, I., Vassalli, P., Nussenzweig, V., and Benacerraf, B. (1967b). J. Exp. Med. 125, 511. Green, N. M. (1969). Aduan. Immunol. 11, 1. Gurdon, J. B., and Woodland, H. R. (1970). Curr. Top. DeoeZop. Biol. 5, 39. Haher, E. ( 1964). Proc. Nut. Acad. Sci. U. S . 52, 1099. Haber, E. (1968). Annu. Reo. Biochem. 37, 497. Haher, E. (1970). Fed. Proc., Fed. Amer. SOC. Exp. Biol. 29, 66. Hennig, W., and Walker, P. M. B. (1970). Nature (London) 225, 915. Hill, R. L., Delaney, R., Fellows, R. E., Jr., and Lebovitz, H. E. (1966). Proc. Nut. Acad. Sci. U . S . 56, 1762. Hilschmann, N. (1967). Gamma Globulins, Proc. Nobel Symp. 3rd, 1967 p. 33. Hilschmann, N., Ponstingl, H., Baczko, K., Braun, D., Hess, M., Suter, L., Barnikol, H. U., and Watanabe, S. (1970). Protides Biol. Fluids, Proc. ColZoq. 17, 53. Hood, L., and Ein, D. (1968). Nature (London) 220, 764. Hood, L., and Talmage, D. W. (1970). Science 168, 325. Hood, L., and Warner, N. ( 1970). Unpublished data. Hodd, L., Gray, W. R., and Dreyer, W. R. (1966). Proc. Nut. Acad. Sci. U. S . 55, 826. Hood, L., Gray, W. R., Sanders, B. G., and Dreyer, W. R. (1967). Cold Spring Harbor Symp. Quant. Biol. 32, 133.
348
L. HOOD AND J. PRAHL
Hood, L., Eichmann, K., Lackland, H., Krause, R. M., and Ohms, J. (1970a). Nature (London) 228, 1040. Hood, L., Potter, M., and McKean, D. J. (1970b). Science 170, 1207. Hood, L., Potter, M., and McKean, D. J. ( 1 9 7 0 ~ ) Unpublished . data. Hood, L., Waterfield, M., Morris, J., and Todd, C. (1971a). Ann. N . Y. Acad. Sci. (in press). Hood, L., Grant, J. A., and Sox, H. C., Jr. (1971b). In “Developmental Aspects of Antibody Formation and Structure” (J. Sterzl and I. Riha, e d s . ) , 2nd ed., Vol. 1, p. 283. Academic Press, New York. Hood, L., Waterfield, M., Potter, M., McKean, D. J., and McIntire, R. ( 1 9 7 1 ~ ) . Unpublished data. Inman, F. P., and Nisonoff, A. (1966). 1. Biol. Chem. 241, 322. Jaton, J. C., Waterfield, M. D., Margolies, M. N., and Haber, E. (1970). Proc. Nut. Acad. Sci. U. S . 66, 959. Jerne, N. (1955). Proc. Not. Acad. Sci. U. S . 41, 849. Jerne, N. (1971). Eur. J. lmmunol. 1, 1. Kaplan, A., Hood, L., Terry, W., and Metzger, H. ( 1971). lmmunochemistry ( in press ) , Karush, F. (1962). J. Pediat. 60, 103. Kedes, L. H., and Birnstiel, M. J. (1971). Nature (London) 230, 165. Kehoe, J. M., Fougereau, M., and Bourgois, A. (1969). Nature 224, 1212. Kelus, A. S., and Cell, P. G . H. ( 1963). Genet. Today, Proc. l n t . Congr. Ilth, 1963 p. 194. Kelus, A. S., and Cell, P. G. H. (1967). Progr. Allergy 11, 141. Kindt, T. J. ( 1971 ). Personal communication. Kindt, T. J., and Todd, C. W. (1969). 1. E x p . Med. 130, 859. Kindt, T. J., Mandy, W. J., and Todd, C. W. (1970). Biochemistry 9, 2028. Knopf, P. M., Parkhouse, R. M. E., and Lennox, E. S. (1967). Proc. Nut. Acad. Sci. U. S. 58, 2288. Kohler, H., Shimizu, A., Paul, C., Moore, V., and Putnam, F. W. (1970). Nature (London) 227, 1318. Koshland, M. E. (1967). Cold Spring Harbor Symp. Quant. Biol. 32, 119. Koshland, M. E., Davis, J. J., and Fujita, N. J. (1969). Proc. Nut. Acad. Sci. U. S. 63, 1275. Krause, R. M. ( 1970). Fed. Proc., Fed. Amer. SOC. Erp. Biol. 29, 59. Kritzman, I. J., Kunkel, H. G., McCarthy, B. J., and Mellors, R. C. (1961). 1. Lab. Clin. Med. 57, 905. Kunkel, H. G. (1965). Haruey L e d . 59, 219. Landsteiner, K. ( 1947). “The Specificity of Serological Reactions,” rev. ed. Harvard Univ. Press, Cambridge, Massachusetts. Landucci Tosi, S., Mage, R. G., and Dubiski, S. (1970). 1. Immunol. 104, 641. Lark, C. A., Eisen, H. N., and Dray, S. (1965). J. lmmunol. 95, 404. Lederberg, J. (1959). Science 129, 1649. Lennox, E. S., and Cohn, M. (1967). Annu. Reu. Biochem. 36,365. Loeb, N. (1968). Aust. I. Erp. Biol. Med. Sci. 46, 397. Lyon, M. F. (1968). Annu. Reu. Genet. 2, 31. McCarthy, B. J. ( 1969). In “Handbook of Molecular Cytology” ( A . Lima-de-Faria, ed. ), p. 13. North-Holland Publ., Amsterdam. McDevitt, H. O., and Benacerraf, B. (1969). Aduan. lmmunol. 11, 31. Mage, R. G., Young, G . O., and Reisfeld, R. (1968). 1. lmmunol. 101, 617.
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
349
Mage, R. G., and Dray, S. (1966). Nature (London) 212, 699. Mandy, W. J., and Todd, C. W. (1968). Vox Sang. 14, 264. Mandy, W. J., and Todd, C. W. (1969). Immunochemistry 6, 811. Mandy, W. J., and Todd, C. W. (1970). Biochem. Genet. 4, 59. Mannik, M. (1967). Biochemistry 6, 134. Margoliash, E., Fitch, W. M., and Dickerson, R. E. (1968). Brookhauen Symp. Biol. 21, 259. Marrack, J. R., Richards, C. B., and Kelus, A. S . (1962). Protides Biol. Fluids, Proc. Colloq. 9, 200. Milstein, C. (1966). Biochem. J . 101, 338. Milstein, C. (1967). Nature (London) 216, 330. Milstein, C., and Munro, A. J. (1970). Annu. Reu. Microbiol. 24, 335. Milstein, C., and Pink, J. R. L. (1970). Progr. Biophys. Mol. Biol. 21, 209. Milstein, C., Milstein, C. P., and Jarvis, J. M. (1969a). J . Mol. B i d . 46, 599. Milstein, C., Milstein, C. P., and Feinstein, A. (1969b). Nature (London) 221, 151. Mole, L. E., Jackson, S. A., Porter, R. R., and Wilkinson, J. M. (1971). Biochem. J . 124 ( in press ) . Mozes, E., McDevitt, H. O., Jaton, J. C., and Sela, M. (1969). J. E x p . Med. 130, 1263. Natvig, J. B., and Kunkel, H. G. (1968). Ser. Haematolo. 1, 66. Niall, H., and Edman, P. (1967). Nature (London) 216, 262. Nisonoff, A., MacDonald, A. B., Hopper, J. E., and Daugharty, H. (1970). Fed. Proc., Fed. Amer. SOC. Erp. Biol. 29, 72. Nolan, C., and hlargoliash, E. (1968). Annu. Reu. Biochem. 37, 727. Nossal, G. J. V., Ada, G. L., and Austin, C. M. (1965). J. Exp. Med. 121, 945. Nossal, G. J. V., Warner, N. L., and Lewis, H. (1971). Cell. Immunol. 2, 41. Ohno, S. ( 1970). “Evolution by Gene Duplication.” Springer-Verlag, Berlin and New York. Osserman, E. F., and Takatsuki, K. (1964). Amer. J . Med. 37, 351. Oudin, J. (1956). C . R. Acad. Sci. 242, 2606. Parker, C. W., and Osterland, C. K. (1970). Biochemistry 9, 1074. Pauling, L. (1940). J . Amer. Chem. SOC.62, 2643. Penn, G. M., Kunkel, H. G., and Grey, H. M. (1970). Proc. SOC.E x p . Biol. Med. 135, 660. Pernis, B., Forni, L., and Amante, L. (1971 ). Ann. N . Y. Acad. Sci. (in press). Pernis, B., Chiappino, G., Kelus, A. S., and Cell, P. G. H. (1965). J . E x p . Med. 122, 853. Perry, E., Cheng, T. Y., Freed, J. J., Greenberg, J. R., Kelly, D. E., and Tartof, K. P. (1970). Proc. Nat. Acad. Sci. U . S . 65, 609 Peterson, H. M., and Laughnan, J. R. (1963). Proc. Nut. Acad. Sci. U . S. 50, 126. Pinder, J. C., Gould, H. J., and Smith, I. (1969). J . Mol. B i d . 40, 289. Potter, M. (1967). Methods Cancer Res. 2, 105. Potter, M., and Leon, M. (1968). Science 162, 369. Potter, M., and Lieberman, R. (1967). Aduan. Immunol. 7, 91. Potter, M. (1970). Fed. Proc., Fed. Amer. SOC.E x p . Biol. 29, 85. Prahl, J. W. (1967). Biochem. J. 104, 647. Prahl, J. W. (1968). Nature (London) 215, 1386. Prahl, J. W., and Porter, R. R. (1968). Biochem. J . 107, 753. Prahl, J. W., and Todd, C. W. (1971). Ann. N . Y. Acad. Sci. (in press).
350
L. HOOD AND J. PRAHL
Prahl, J. W., Mandy, W. J., David, G. S., Steward, M. W., and Todd, C W. (1970). Protides Biol. Fluids, Proc. Colloq. 17, 125. Press, E. M., and Hogg, N. M. (1969). Nature (London) 223, 807. Putnam, F., Shimizu, A., Paul, C., Shinoda, T., and Kohler, H. (1971). N . Y. Acud. Sci. (in press). Quattrocchi, R., Cioli, D., and Baglioni, C. (1969). J. E x p . Med. 130, 401. Reisfeld, R. A., Dray, S., and Nisonoff, A. (1965). lmmunochemistry 2, 155. Retel, J., and Planta, R. J. ( 1968). Biochim. Biophys. Acta 169, 416. Ritossa, F. M., and Spiegelman, S. (1965). Proc. Nut. Acad. Sci. U . S. 53, 737. Ritossa, F. M., Atwood, K., Lindsley, D., and Spiegelman, S. (1966). Nut. Cancer Inst., Monog. 23, 449. Rivat, L., Gilbert, D., and Ropartz, C. (1970). Protides Biol. Fluids, Proc. Colloq. 17, 233. Rosenberg, M. (1970). Proc. Nut. Acud. Sci. U . S. 67, 32. Schubert, D. A., and Cohn, M. (1970). J. Mol. Biol. 53, 305. Schubert, D. A., Jobe, A., and Cohn, M. (1968). Nature (London) 220, 882. Schweet, R. S., and Owen, R. S. (1957). 1. Cell. Comp. Physiol. 50, Suppl. 1, 199. Selignlann, M., Mikaesco, E., and Hurez, D. (1969). J. Clin. Inoest. 48, 2374. Silverstein, A. M., and Prendergast, R. A. (1971). In “Developmental Aspects of Antibody Formation and Structure” (J. Sterzl and H. Riah, eds.), 2nd ed., Vol. 1, p. 69. Academic Press, New York. Silverstein, A. M., Uhr, J. W., Kraner, K. L., and Lukes, R. J. (1963). J. Exp. Med. 117, 799. Singer, S. J., and Doolittle, R. F. (1966). Science 153, 13. Singer, S. J., and Thorpe, N. (1968). Proc. Nut. Acad. Sci. U . S . 60, 1371. Siskind, G. W., and Benacerraf, B. (1968). Aduan. Immunol. 10, 1. Small, P. A,, Jr., Reisfeld, R. A., and Dray, S. (1965). J. Mol. Biol. 11, 713. Small, P. A., Jr., Reisfeld, R. A,, and Dray, S. (1966). J Mol. Biol. 16, 333. Smith, G. P., Hood, L., and Fitch, W. M. (1971). Annu. Reu. Biochem. 40, 969. Smithies, 0. ( 1967). Science 157, 267. Smithies, 0. (1970). Science 169, 882. Smithies, 0. ( 1971 ). In “Developmental Aspects of Antibody Formation and Structure” ( J . Sterzl and H. Riah, eds.), 2nd ed., Vol. 1, p. 1001. Academic Press, New York. Smithies, O., Gibson, D. M., Fanning, E. M., Percy, M. E., Parr, D. M., and Connell, G. E. ( 1971). Science 172, 574. Snapper, I., and Kahn, A. I. ( 1971 ). “Myelomatosis.” Univ. Park Press, Baltimore, Maryland. Solomon, A., and McLaughlin, C. L. (1969). J. Biol. Chem. 244, 3393. Southern, F. M. (1970). Nature (London) 227, 794. Steinberg, A. G. ( 1966). In “Symposium in Inimunogenetics” (T. J. Greenwalt, ed. ), p. 75. Lippincott, Philadelphia, Pennsylvania. Stemke, G. W. (1964). Science 145, 403. Sterzl, J., and Silverstein, A. M. (1967). Aduan. Immunol. 6, 337. Szilard, L. (1960). Proc. Nut. Acad. Sci. U . S . 46, 293. Talmage, D. (1957). Annu. Reu. Med. 8, 239. Terry, W. D., and Ohms, J. (1970). Proc. Nut. Acad. Sci. U . S. 66, 558. Terry, W. D., Hood, L., and Steinberg, A. G. (1969). Proc. Nut. Acad. Sci. U . S . 63, 71.
THE IMMUNE SYSTEM: A MODEL FOR DIFFERENTIATION
351
Terry, W. D., Boyd, M. M., Rea, J. S., and Stein, R. (1970). J . Immunol. 104, 256. Thomas, C. A., Jr. ( 1971 ). In “The Neurosciences: Second Study Program” ( F . 0. Schmitt, ed.), p. 973. Rockefeller Univ. Press, New York. Thorpe, N. O., and Singer, S. J. (1969). Biochemistry 8, 4523. Todd, C. W. (1966). J. Cell. Physiol. 67, Suppl., 95. Todd, C. W., and Inman, F. P. (1967). Immunochemistry 4, 107. Utsumi, S. (1969). Biochem. J. 112, 343. Valentine, R. C., and Green, N. M. (1967). J. Mol. Biol. 27, 615. Vice, J. L., and Cihnan-Sachs, A. (1969). Fed. Proc., Fed. Amer. SOC. Elrp. Biol. 28, 436. Walker, P. M. B., Flamm, W. G., and McLaren, A. (1969). “Handbook of Molecular Cytology” ( A . Lima-de-Faria, ed. ), p. 53. North-Holland, Publ., Amsterdam. Wallace, H., and Birnstiel, M. J. (1966). Biochim. Biophys. Acta 114, 296. Wang, A. C., Wang, I. Y. F., McCormick, N., and Fudenberg, H. H. (1969). Immunochemistry 6, 451. Wang, A. C., Wilson, S. K., Hopper, J. E., Fudenberg, H. H., and Nisonoff, A. (1970). Proc. Nat. Acud. Sci. U. S. 66, 337. Warner, N. ( 1970). Personal communication. Weigert, M. G., Cesari, I. M., Yonkovich, S. J., and Cohn, M. (1970). Nature (London) 228, 1045. Whitehouse, H. L. K. (1967). Nature (London) 215, 371. Whitney, P. L., and Tanford, C. (1965). Proc. Nat. Acad. Sci. U . S. 53, 524. Wilkinson, J. M. (1969a). Biochem. J. 112, 173. Wilkinson, J. M. (196913). Nature (London) 223, 616. Wu, T. T., and Kabat, E. A. (1970). J. E x p . Med. 132, 211. Zettervall, 0. (1968). A d a Med. S c a d , S u p p l . 5, 492. Zullo, D. M., Todd, C. W., and Mandy, W. J. (1968). Proc. Can. Fed. Biol. SOC. Abstr. 11, 111.
ADDENDUM The 5 S ribosomal system of Xenopus laevis is supposedly a multigene system in which each of the 24,000 genes has a constant portion and a variable portion (D. Brown, personal communication). If this is so, some of the evolutionary arguments presented in this paper for separate V and C genes are incorrect-at least in the case of 5 S ribosomal genes.
This Page Intentionally Left Blank
Author Index Numbers in italics refer to the pages on which the complete references are listed. Andrada, J. A., 132, 173, 184 A Andreani, D., 88, 179 Abbey, H., 139, 179 Andreoli, M., 88, 163, 173, 179 Abercrombie, M., 2.38, 243 Andresen, R. H., 32, 76 Ablashi, D. V., 200, 246 Anfinsen, C. B., 295, 347 Abraham, S., 164, 180 Aoki, T., 19, 78, 217, 247 Abrams, G., 42, 80 Appella, E., 305, 328, 337, 345, 346 Abramson, N., 271, 273, 286 Achong, B. G., 189, 190, 192, 239, 244 Aquaron, R., 157, 184 Arata, T., 233, 249 Ada, G. L., 307,349 Argue, H., 132, 137, 178 Adams, D. D., 133, 173 Armenia, J. P., 10.2, 182 Adant, M., 110, 173 Armstrong, D., 212, 227, 243, 245 Adldinger, H. K., 226, 228, 243 Arnstein, N., 107, 108, 176 Ager, J . A. M., 136, 173 Askonas, B. A., 101, 156, 177 Agrawal, R. B., 107, 176 Assem, E. S . K., 142, 156, 173 Ahlstrom, C. G., 208, 243 Atkin, N. B., 375, 345 Ahmed, M., 231, 243 Attardi, G., 330, 345 Ainbender, E., 200, 245 Atwood, K., 330, 350 Akerman, M., 208, 243 Austen, K. F., 255, 256, 257, 258, 259, Alardn-Segovia, D., 137, 173 260, 261, 284, 286, 287, 288, 289, Alexander, C., 300, 305, 319, 347 290 Alexander, J., 294, 345 Austin, C. M., 307, 349 Alexander, W. D., 134, 139, 174 Austin, C. R., 20, 21, 76 Allan, T. M., 74, 76 Averich, E., 45, 77 Allan, W. S. A., 134, I83 Averill, R. L. W., 26, 76 Allen, J. C., 298, 345 Avila, L., 190, 194, 205, 245 Allison, A. C., 188, 243 Azar, H. A,, 24, 79 Almqvist, S., 142, 180 Azen, E . A., 256, 265, 266, 268, 285, 286 Aloj, S., 106, 173, 181, 182 Alper, C . A., 255, 257, ,260, 264, 266, B 267, 268, 269, 270, 271, 272, 275, 278, 283, 285, 286, 288, 289 Bach, F. H., 61, 76 Amante, L., 307, 349 Bach, S., 284, 286 Amos, D. B., 24, 57, 76, 209, 211, 242, Baczko, K., 303, 320, 347 243, 244 Baglioni, C., 342, 350 Anderson, J. M., 69, 76 Bagshawe, K. D., 22, 23, 25, 36, 37, 47, Anderson, J. R., 133, 134, 136, 137, 139, 76, 77, 78 140, 141, 149, 173, 174, 176 Baldo, B. A., 16, 77 Anderson, J. W., 137, 173 Baldwin, R. L., 157, 180 Anderson, N. F., 72, 81 Balfour, B. M . , 136, 173 Anderson, P., 278, 288 Ballantyne, D. L., Jr., 283, 290 Anderson, S. G . , 31, 83 Ballard, H. S., 310, 345 Andersson, T., 208, 243 Baney, C., 4G, 79 Andrada, E. C., 132, I84 Barber, M., 16, 83 353
354
AUTHOR INDEX
Bardawil, W. A., 33, 36, 58, 76 Barker, C. F., 6, 29, 36, 61, 76 Barnes, R. D., 52, 83 Barnikol, H. U., 303, 3 2 4 347 Baron, S., 126, 176, 190, 192, 248 Barr, Y. M., 189, 190, 239, 244 Barrett, hl. K., 58, 77 Barron, A. L., 136, 176 Barth, R. F., 19, 51, 76, 279, 286 Baschieri, L., 133, 180 Bass, B. H., 141, 184 Battisto, J. R., 126, 176 Baxendale, W., 229, 243 Beall, G. N . , 133, 183 Becak, W., 375, 345 Becker, E. L., 257, 258, 286, 288 Becker, K . L., 141, 174 Becker, W., 255, 286 Becker, Y., 193, 194, 249 Beckers, C., 100, 106, 174 Beer, A. E., 6, 10, 19, 20, 42, 46, 58, 65, 76 Behrman, S. J., 17, 42, 44, 45, 46, 80, 82 Beierwaltes, W. H., 87, 117, 122, 129, 134, 139, 143, 149, 174, 175, 182 Belt, W. D., 88, 174 Beltran, G., 198, 243 Belyavin, G., 135, 156, 173, 174, 184 Benacerraf, B., 126, 179, 297, 301, 339, 347, 348, 350 Ben-Efraiin, S., 279, 286 Benirschke, K., 22, 28, 29, 53, 69, 76 Bennett, D., 19, 78 Bennett, J. C., 292, 305, 310, 311, 315, 337, 345, 346 Bennett, W . A,, 134, 179 Benyesh-Melnick, M., 198, 214, 215, 243, 249 Berard, C. W., 200, 246 Berdol, P., 134, 176 Bergot, J., 162, 178 Bergquist, R., 158, 159, 174 Berkelhanmier, J., 117, 177 Berry, R. O., 26, 83 Betke, K., 51, 76, 80 Beutel, J. S., 199, 205, 248 Beutner, E. H., 123, 135, 136, 174, 176, 185 Bevans, M., 44, 45, 76
Biasucci, A., 142, 176 Bigelow, B., 310, 347 Biggs, P. M., 227, 243 Billingham, R. E., 4, 6, 9, 10, 14, 19, 20, 25, 28, 29, 30, 31, 32, 36, 42, 46, 48, 53, 54, 55, 58, 61, 62, 64, 65, 66, 67, 68, 69, 72, 73, 76, 77, 78, 82, 83 Billington, W. D., 4, 8, 30, 35, 36, 38, 39, 40, 41, 42, 52, 77, 80 Billote, J. B., 127, 130, 176 Biozzi, G., 278, 290 Birch, S. M., 213, 215, 244 Birg, F., 205, 247 Birnstiel, M. J., 330, 348, 351 Biro, C. E., 281, 282, 283, 284, 286, 289 Birtch, A. G., 268, 269, 286 Bishun, N. P., 52, 83 Bismuth, J., 107, 159, 161, 178, 181 Bissett, J. M., 97, 184 Bitter-Suermann, D., 206, 248 Bjork, I., 337, 345 Bjorkland, A., 129, 174 Blackburn, W. R., 66, 79 Blackham, E. A., 220, 249 Blacklow, N. R., 198, 243 Blackwell, P. M., 21, 82 Blakemore, W. S., 52, 83 Blakeslee, J. R., 191, 245 Blanc, W. A., 285, 288 Blnszczyk, J., 266, 287 Blizzard, R. M. , 134, 139, 140, 149, 150, 174 Bloth, B., 158, 159, 174 Blum, L., 275, 286 Blumenthal, H. T., 130, 150, 160, 181 Boenisch, T., 272, 285, 286 Boettcher, B., 16, 77 Bokisch, V. A,, 273, 236 Bonneau, H., 201, 205, 246, 247 Bora, S., 134, 184 Borsos, T., 275, 276, 285, 286, 287, 290 Boss, J. H., 45, 77 Bouchilloux, S., 155, 164, 174 Bourgois, A., 313, 348 Bourgois, M., 325, 345 Bourne, G. H., 88, 174 Bouthillier, Y.,278, 290 Boyd, J. D., 4, 8, 77 Boyd, M. M., 297,351 Boyer, J. T., 274, 289
AUTHOR INDEX
355
Boyse, E. A., 19, 78, 188, 201, 207, 210, Burtonshaw, M. D., 52, 77 217, 232, 237, 238, 242, 247, 248, Buschesk, F. T., 191, 192, 214, 247 249 Butsch, D. W., 32, 82 Brackett, B. G., 20, 77 C Bradbury, S., 36, 38, 39, 77, 80 Bradley, R. M., 238, 247 Cachin, Y., 199, 207, 245 Brady, R. O., 238,247 Cahnmann, H. J., 103, 104, 105, 182 Braley, H . C., 127, 174 Calcott, M. A., 272, 289 Brambell, F. W . R., 4, 77 Callan, J. D., 333, 345 Brandes, D., 88, 174 Calnek, B. W., 226, 227, 228, 243 Bratanov, K., 17, 77 Campbell, P. N., 113, 132, 145, 147, 148, 181 Braun, D., 303, 320, 347 Braun, H., 51, 80 Capra, J. D., 296, 303, 316, 320, 346, Braunitzer, G., 315, 345 346 Brawn, J., 60, 79 Carbone, P. P., 200, 222, 246, 249 Bregula, U., 238, 246 Caren, L. D., 277, 278, 279, 286, 287 Breinl, R., 294, 345 Carlomagno, M . S., 163, 175 Breniberg, S., 206, 243 Carpenter, C. B., 256, 257, 287 Brener, J. L., 145, 150, 178 Carroll, G. F., 279, 286 Brenner, S., 340, 345 Cartouzou, G., 106, 107, 161, 178 Brent, L., 14, 15, 44, 55, 62, 68, 77 Carvalho, E., 97, 175 Brent, R. L., 45, 64, 77, 82 Castle, W. B., 56, 79 Brettschneider, L., 188, 248 Cavalieri, R. R., 89, 90, 180 Breyere, E. J., 27, 58, 77, 81 Cebra, J. J., 297, 301, 346 Brinster, R. L., 29, 30, 79, 81 Celano, M. J., 80 Brinton, V., 46, 79 Cerini, M., 138, 176 Britten, R, J., 329, 335, 342, 345 Cesari, I. M., 320, 322, 325, 337, 351 Bronson, P., 105, 183 Chaikoff, I. L., 100, 164, 179, 180 Brooke, M. S., 134, 174 Chambers, L. A., 18, 79 Brough, A. S., 67, 79 Chandler, R. W., 139, 140, 149, 150, 174 Brown, D. D., 330, 343, 345 Charache, P., 254, 289 Brown, D. M., 161, 180 Charlwood, P. A., 313, 346 Brown, P. C., 135, 177 Cheftel, C., 164, 174 Brunengo, A. M . , 134, 183 Chen, J. H., 226, 243 Brzosko, W. M., 134, 179 Cheng, H. F., 103, 174 Buchanan, W. W., 134, 136, 139, 141, Cheng, T. Y., 330, 349 173, 174 Chersi, A., 328, 346 Buckler, C. E., 126, 176 Chew, A. R., 139, 175 Buckley, R. H., 125, 179 Chi, C. A,, 56, 81 Bulfone, L. M., 191, 192, 214, 247 Chiappino, G., 301, 349 Burchenal, J. H., 203, 246 Choate, H. W., 24, 33, 83 Burdon, K. L., 258, 286 Chown, B., 51, 84 Burger, M. M., 238, 243 Christian, C. L., 285, 287 Burgoyne, G. H., 230, 249 Christy, N. P., 140, 174 Burhoe, S. O., 58, 77 Chubb, R. C., 227, 229, 243 Burke, G., 140, 174 Churchill, A. E., 225, 227, 229, 243 Burkitt, D. P., 187, 235, 241, 243, 245 Churchill, W. H., 160, 174 Cinader, B., 276, 277, 279, 286, 287 Burmester, B. R., 227, 230, 247 Burnet, F. M., 293, 295, 345 Cioli, D., 342, 350 Cirkovii., T., 163, 180 Burtin, P., 199, 207, 245
356
AUTHOR INDEX
Cittanova, N., 162, 180 Clark, D. H., 134, 136, 141, 174, 176 Clarke, B., 73, 77 Clarke, B. F., 141, 177 Clem, L. W., 298,346 Clifford, P., 187, 199, 201, 202, u)3, 204, 205, 207, 208, 209, 210, 211, 216, 217, 218, 223, 231, 233, 236, 237, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250 Cline, M. J,, 134, 174 Coca, A. F., 262, 287 Cochrane, C. G., 273, 279, 284, 286, 290 Codaccioni, J . L., 106, 107, 161, 178 Cohen, C., 283, 290 Cohen, F., 51, 55, 77 Cohen, M. H., 222, 249 Cohen, S., 296, 297, 299, 346 Cohn, M., 292, 294, 297, 299, 305, 310, 315, 320, 321, 322, 325, 337, 339, 340, 341, 346, 348, 350, 351 Colberg, J. E., 301, 346 Cole, R. K., 122, 123, 174, 184, 185 Colten, H. R., 285, 287 Condie, R. M., 126, 177 Conn, M., 191, 245 Connell, G. E., 310, 350 Connell, H. C., 230, 249 Converse, J. M., 13, 82 Coombs, R. R. A., 18, 78 Cooper, E. L., 314, 346 Cooper, M., 275, 286 Cooper, M. D., 286, 290 Cooper, N. R., 260, 285, 286, 287 Cope, T. I., 22, 78 Couchman, K. G., 136, 137, 173, 181 Coute, F . A., 52, 83 Couvillion, L. A., 201, 249 Covelli, I., 106, 181 Craig, J. M., 45, 77, 255, 257, 288, 290 Crisler, C., 279, 287 Crooks, J., 134, 139, 174 Crookston, J. H., 22, 78 Cruchaud, A., 140, 174 Cruse, V., 38, 82 Ctavo, L., 200, 245 Cullen, D. R., 141, 177 Cunningham, B. A., 299, 304, 312, 313, 346
Currie, C.A., 13, 36, 37, 59, 77, 78 Curzen, P., 45, 78
D Dagg, M., 57, 58, 59, 79 Dain, A. R., 28, 78 Dalinasso, A. P., 272, 289 Dalton, A. J., 88, 177 Dancis, J., 68,78 Daniel, M . D., 225, 245 Daniel, P. M., 142, 174 Darlington, C. D., 73, 80 Darrow, R., R., 51, 78 Daugharty, H., 298, 349 Dausset, J., 19, 74, 78 David, D. S., 9, 10, 81 David, G . S., 309, 346, 350 David, J . R., 121, 174 Davidsohn, I., 56, 83 Davidson, E., 335, 342, 345 Davies, D., 299, 313, 346 Davies, D. A. L., 29, 78, 222, 244 Davies, J. R., 149, 181 Davies, S. H., 140, 177 Davis, A. M., 139, 175 Davis, J. J., 327, 348 Davis, N. C., 273, 290 Deal, D. R., 214, 216, 244 Debeaux, P., 263, 289 de Capoa, A., 285, 288 Decreusefond, C., 278, 290 de Crombrugghe, B., 161, 162, 175 Defendi, V., 72, 77 DeGroot, L. J., 87, 90, 97, 139, 175, 179 de Harven, E., 217, 248 Deichmann, G. I., 205, 230, 238, 244 Delamore, I. W., 70, 84 Delaney, R., 299, 311, 347 Delaunay, A., 16, 83 DeLorenzo, F., 299, 347 DBmant, P., 63, 78 Demeulenaere, L., 264, 290 Deniissie, A., 205, 249 De Nayer, P., 90, 175,182 Depieds, R., 107, 178 Derrien, Y.,99, 156, 157, 175 de Schryver, A., 199, 207, 224, 236, 238, 244, 245 Dessy, J. I., 198, 243 Destine, M. L., 67, 81
357
AUTHOR INDEX
de ThB, G., 199, 207, 224, 236,238, 244, 245, 247 De Visscher, M., 90, 100, 106, 174, 175 Devlin, W., 103, 180 Dewey, S., 196, 206, 248 Diamondopoulos, G. T., 53, 78 Dias da Silva, W., 258, 271, 287 Dickerson, R. E., 315, 349 Diehl, V., 190, 194, 197, 198, 199, 203, 204, 207, 233, 236, 237, 239, 244, 245, 246, 250 Dinerstein, J., 26, 31, 80 Dintzis, H. M., 310, 346 Dixon, F. J., 64,81, 280, 290 Dobrowolski, M . S., 44, 78 Dobyns, B. M., 142, 175 Dodson, V. N., 139, 174 Doebbler, T. K., 112, 115, 117, 120, 125, 132, 143, 149, 150, 175, 182 E r n e r , V., 13, 80 Donaldson, V. H., 253, 254, 255, 256, 257, 258, 287, 288, 289, 290 Doniach, D., 86, 87, 113, 124, 130, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 145, 147, 148, 149, 150, 151, 152, 169, 173, 175, 176, 177, 181, 184 Doniach, I., 22, 78 Donohoe, W. T . A., 56, 84 Doolittle, R. F., 311, 350 Dopheide, T. A. A., 162, 175 Doria, C., 133, 180 Douglas, G . W., 68, 78 Dowling, T . J., 107, 175 Doyle, L. L., 6, 78 Dray, S., 297, 300, 301, 305, 309, 319, 346, 347, 348, 349, 350 Dreyer, W. J., 292, 294, 305, 310, 311, 313, 315, 328, 344, 346 Dreyer, W. R., 301, 304, 305, 306, 307, 315, 335, 336, 337, 347 Driscoll, S. C.,22, 28, 29, 53, 69, 76 Drobish, D. C., 117, 129, 130, 180 Dubiski, S., 276, 277, 279, 287, 300, 305, 308, 309, 337, 345, 346, 348 Duhring, J. L., 52, 83 Dulbecco, R., 241, 249 Dumonde, D. C . , 107, 175 Dumont, J. E., 106, 180 Duncan, L. P. J., 141, 177
Dunkel, V. C., 190, 194, 205, 244, 245 Durr, F . E., 195, 244 Dvorak, B., 200, 245
E Eagleton, G. B., 88, 180 Eastman, N. J., 44, 79 Ecker, E. E., 262, 287 Edelhoch, H., 99, 100, 103, 105, 106, 157, 158, 161, 162, 163, 168, 173, 175, 179, 180, 182, 183, 184 Edelman, G . M., 292, 294, 299, 300, 303, 304, 305, 310, 311, 312, 313, 314, 315, 320, 321, 326, 333, 334, 341, 346, 347 Edman, P., 304, 349 Edwards, R. G., 16, 17, 18, 78 Eernisse, J. C., 57, 83 Egan, R. W., 113, 115, 125, 128, 129, 130, 132, 183, 185 Ehrlich, P., 294, 346 Eicher, E. M . , 301, 346 Eichmann, K., 298, 304, 324, 325, 328, 332, 336, 348 Eichwald, E. J., 58, 80 Eidson, S . C., 228, 244 Ein, D., 304, 307, 347 Einhorn, N., 209, 244 Eisen, H. N., 297, 343, 347, 348 Ekholm, R . , 142, 180 Ellman, L., 70, 78, 259, 287 Elston, C. W., 23, 78 Emslander, R. F., 137, 173 Engel, E., 141, 185 Enriquez, P., 285, 286, 289 Epstein, C . J., 295, 347 Epstein, M. A., 189, 190, 192, 206, 239, 243, 244 Erickson, R. P., 276, 277, 287 Ernberg, I., 190, 196, 197, 204, 205, 206, 244 Eschenbach, C., 52, 81 Eskeland, T . , 202, 246 Espmark, A., 200, 249 Evans, A. C., 198, 233, 237, 244, 247 Evans, A. W . H . , 139, 175 Evans, C . A., 242, 245 Evans, E. P., 52, 77 Evans, M. M., 51, 77
358
AUTHOR INDEX
Evans, R. R., 253, 254, 287 Evans, R. W., 139, 175 Evans, T. C., 103, 122, 129, 174, 175 EvaskovP, E., 63, 78 Exum, E. D., 121, 125, 145, 176, 178, 179 Eyquem, A., 129, 134, 175, 177
F Fagraeus, A., 138, 175, 178 Fahey, J. L., 151, 175, 222, 246, 296, 337, 347 Falconer, I. R., 107, 175 Falke, D., 206, 248 Fanning, E. M., 310, 350 Farthing, C. P., 136, 181 Fass, L., 211, 221, 223, 244 Fawcet, D. W., 231, 244 Feclerlin, K., 134, 175 Feinberg, J . G., 136, 175 Feinstein, A., 304, 305, 347, 349 Feldman, J. M., 140, 174 Feldman, L. A., 199, 205, 248 Felix-Davies, D., 125, 175 Fellous, M., 19, 78 Fellows, R. E., Jr., 299, 311, 347 Fennel], R. H., Jr., 134, 180 Fenyo, E. M., 203, 204, 209, 242, 244, 247 Fenzi, C. F., 133, 180 Ferguson, E., 56, 81 Ferguson, L. C., 18, 78 Ferguson, R. H . , 141, 174 Ferguson-Smith, M. A., 141, 176 Fernbach, D. J., 198, 243 Fialkow, P. J., 141, 176, 204, 241, 244 Fierer, J., 68, 78 Figueiredo, J. G., 97, 179 Fikrig, S., 26, 31, 80 Finegold, M., 64,78 Fink, M . A., 220, 232,244 Finkelstein, S., 70, 83 Finn, C. A,, 5, 78, 80 Finn, R., 56, 78 Fischer, H., 283, 288 Fitch, W. M., 298, 299, 300, 305, 306, 315, 316, 317, 318, 321, 322, 323, 325, 326, 328, 329, 334, 341, 342, 347, 349, 350 Flamm, W. G., 331, 351 Flanagan, T. D., 136, 176
Flax, M. H., 127, 130, 176, 178 Fleischman, J. B., 296, 297, 310, 328, 347 Fleisher, M. S., 107, 108, 176 Fogel, S., 333, 347 Fondarai, J., 159, 161, 181 Forbes, A. P., 141, 185 Forbes, I. J., 149, 150, 176 Ford, C. E., 52, 77 Forni, L., 307, 349 Forristal, J., 273, 290 Fougereau, M., 313, 325, 345, 348 Fox, H., 107, 109, 111, 177 Franek, F., 316, 347 Frangione, B., 309, 328, 347 Frank, M., 257, 259, 279, 287 Franklin, E. C., 309, 310, 347 Freda, V. J., 56, 78 Freed, J. J., 330, 349 Freeman, M. J., 127, 174 Freinkel, N., 107, 175 Freund, J., 16, 78 Friberg, S., 202, 207, 244, 246 Fridman, J., 97, 179 Friedman, A. H., 136, 178 Friedman, R. M., 126, 176 Fr@land,A., 141, 176 Fudenberg, H. H., 307, 308, 351 Fujita, N . J., 327, 348 Fulginiti, V., 66, 79 Furth, E. D., 107, 176 Fuson, R. B., 58, 80
G Gaither, T., 287 Gall, W . E., 299, 303, 312, 313, 346 Gally, J. A., 292, 300, 305, 310, 311, 315, 320, 321, 333, 334, 339, 341, 346, 347 Galton, M., 31, 58, 62, 78 Garcia, G., 284, 286 Gardiner, H. H., 22, 33, 83 Gartler, S. M . , 204, 241, 244 Gates, A. H., 6, 78 Gauld, I. K., 52, 77 Geering, G. 207,.u)8, 210, 217, 231, 245, 246, 247, 248 Geering, L., 202, 203, 209, 210, 218, 242, 246
359
AUTHOR INDEX
Gell, P. G. H., 126, 179, 300, 301, 304, 348, 349 Gemsa, D., 283, 289 Genghof, D. S., 126, 176 Gerard, L., 300, 346 Gerber, P., 192, 195, 213, 214, 215, 216, 239, 244 Gergely, L., 190, 196, 197, 204, 205, 206, 209, 244, 246, 249 Gerter, H., 2U2, 249 Geserick, G., 265, 266, 287, 289 Gewurz, H., 256, 260, 285, 286, 287, 289, 290 Ghayasuddin, M., 115, 151, 169, 176, 182, 183 Chose, T., 137, 138, 176, 179 Gibson, D. M., 310, 350 Giertz, H., 284, 287 Gigli, I., 255, 257, 287 Gilbert, D., 328, 350 Gilman-Sachs, A., 300, 305, 319, 347, 351 Gilmore, C. E., 225, 245 Githens, J. H., 66, 79 Gitlin, D., 142, 176 Givol, D., 297, 299, 346, 347 Glade, P., 200, 245 Glazer, A. N., 297, 347 Glegg, R. E., 164,184 Gleich, G. J., 285, 286, 289 Glehner, G. G., 88, 177 Glynn, A. A., 278, 287 Glynn, L. E., 87, 176 Godal, T., 130, 134, 176, 178 Goetze, O., 284, 290 Goetzl, E., 316, 347 Goldberg, D., 134, 183 Goldberg, E. H., 19, 78 Goldberger, R. F., 295, 347 Goldstein, D., 36, 38, 39, 80 Goldstein, G., 205, 207, 244, 246 Gonzales, B., 57, 79 Gonzales, E., 64, 81 Good, R. A,, 256, 260, 285, 286, 287, 289, 290 Goodkofsky, I . , 275, 287 Goodlin, R. C., 57, 59, 78 Goodman, H. C . , 145, 146, 151, 175, 176 Gordon, I., 31, 78 Corninn, J. G., 56, 78
Gorstein, F., 68, 78, 126, 179 Gotoff, S. P., 267, 288 Gottlieb, P. D., 299, 312, 313, 326, 346 Goudie, R. B., 133, 134, 136, 137, 138, 139, 140, 149, 173, 174, 176 Gould, H. J., 330, 349 Gourlay, F., 93, 176 Gowland, G., 14, 68, 77 Grace, J. T., 129, 130, 179, 190, 191, 194, 205, 213, 219, 239, 245, 246 Graham, W. R., 122, 179 Granerus, G., 258, 289 Grant, J. A,, 304, 326, 336, 337, 338, 347, 348 Grasso, L., 133, 180 Gray, A. P., 35, 78 Gray, K. G., 133, 136, 137, 140, 141, 149, 173, 174, 176 Gray, W. R., 294, 301, 304, 305, 306, 307, 310, 311, 313, 315, 328, 335, 336, 337, 344, 346, 347 Green, H., 238, 248 Green, I., 70, 78, 259, 287, 301, 347 Green, J. W., 52, 83 Green, N. M., 299, 347, 351 Greenberg, J . R., 330, 349 Grey, H. M., 307,349 Grinnel, S. T., 209, 242, 244 Gripenberg, N., 203, 209, 242, 244 Gross, S. J., 24, 33, 78 Grosso, 0. F., 134, 180 Gruber, D., 67, 79 Gruenstein, E., 160, 176 Grumbach, M. M., 52, 83 Gruson, M., 101, 181 Gullbring, B., 18, 78 G u n v h , P., 199, 207, 208, 218, 244, 245 Gurdon, J. B., 293, 343, 345, 347 Gustafson, D . G., 51, 77
H Haber, E., 258, 290, 292, 295, 298, 328, 347, 348 Haelst, L. V., 106, 178 Hahn, B., 139, 179 Haines, A. L., 257, 288 Hajdu, A., 129, 176 Hakamori, S . I., 238, 244 Halasz, N. A., 39, 83
360
AUTHOR INDEX
207, 208, 209, 210, 212, 218, 231, Halbrecht, I., 61, 78 234, 239, 240, 241, 242, 243, 244, Hall, J. E., 11, 81 245, 246, 247, 248, 249, 250 Hall, R., 139, 175, 176 Henle, W., 18, 79, 188, 189, 190, 191, Hall, T . G., 107, 108, 176 192, 194, 195, 196, 197, 198, 199, Haller, J. A., 9, 78 200, 202, 203, 204, 205, 206, 207, Halmi, N. S., 89, 177 208, 209, 210, 212, 213, 216, 218, Halpern, B. N., 21, 78 223, 233, 234, 236, 237, 239, 240, Halpern, S., 231, 247 241, 242, 243, 244, 245, 246, 247, Hamilton, L. M., 310, 345 248, 249, 250 Hampar, B., 192, 244 Hennig, W., 330, 347 Hampton, J. K., 29, 78 Heppner, G. H., 242, 245 Hampton, S. H., 29, 78 Herberman, R. B., 211, 221, 223, 244 Hamwi, G. J., 134, 174 Herbert, J., 273, 290 Hancock, J. L., 43, 78 Hermet, J., 129, 177 Handl, J. H., 56, 79 Herscovics, A,, 90, 177 Harden, R. M., 134, 141, 174 Hertig, A. T., 53, 78 Hargis, B. J., 117, 177 Hertz, R., 24, 79 Hargreaves, F., 97, 179 Herzenberg, L. A., 57, 59, 78, 79, 276, Harington, C. R., 100, 177 277, 287, 288 Harpel, P. C., 255, 288 Heslop, R. W., 26, 32, 79 Harris, H., 238, 246, 249 Hess, M., 303, 320, 347 Hartman, C. G., 15, 79 Hewetson, J., 209, 249 Hartmann, W. H., 139, 179 Heyner, S., 29, 30, 36, 41, 79, 81 HaEkovA, V., 31, 79 High, C . J., 118, 120, 142, 184 Hathaway, W. E., 66, 79 Hijmans, W., 141, 177 Hauge, M., 25, 81 Hill, C. W., 136, 175 Haupt, H., 254, 272, 288, 290 Hill, R. L., 299, 311, 347 Haupt, I., 283, 288 Hiller, O., 265, 266, 286 Haurowitz, F., 294, 345 Hilschmann, N., 303, 315, 320, 347 Hauschka, T. S., 209, 242, 244 Hinuma, Y., 191, 196, 219, 239, 245, 246 Hay, J., 16, 77 Hinz, C. F., Jr., 274, 289 Hazard, J. B., 87, 90, 177 Hiramoto, R., 145, 150, 181 Heape, W., 26, 31, 79 Hirata, Y., 130, 150, 160, 181 Hecht, F., 141, 176 Hirsch, E. Z., 142, 175 Heide, K., 254, 272, 288, 290 Heidelberger, M., 93, 94, 110, 148, 158, Hirshaut, Y., 195, 200, 244, 245 Hirvanen, T., 75, 83 177, 183 Hitchner, S. B., 227, 243 Heimann, P., 88, 177 Hjort, T., 134, 138, 142, 177 Heird, W. C., 285, 288 Ho, H. C., 199, 207, 244, 245 Hektoen, L., 107, 109, 111, 142, 177 Hofer, R., 134, 177 Hellman, L. M., 44, 52, 79, 80 Hellstriim, I., 37, 60, 79, 203, 209, 232, Hoerr, R. A., 6, 10, 19, 58, 76 Hogg, N. M . , 304, 350 238, 242, 244, 245, 249 Hellstrom, K. E., 37, 60, 79, 203, 209, Holborow, E. J., 87, 121, 135, 141, 174, 176, 177 238, 242, 244, 245, 246 Holnies, E. C., 201, 247 Helyer, B. J., 123, 177, 280, 288 Hood, L., 292, 294, 298, 299, 300, 301, Hendler, R. W., 88, 177 303, 304, 305, 306, 307, 308, 310, Henk, W., 198, 244 311, 313, 315, 319, 320, 321, 322, Henle, C., 18, 79, 190, 194, 196, 197, 323, 324, 325, 326, 328, 329, 332, 198, 200, 202, 203, 204, 205, 206,
AUTHOR INDEX
361
334, 335, 336, 337, 338, 341, 342, J 344, 346, 347, 348, 350 Jackson, B. T., 28, 79 Hopper, J. E., 298, 307, 308, 349, 351 Jackson, S. A., 327, 349 Home, M . K., 194, 214, 239, 248 Jacobs, J. C., 285, 288 Horosziewicz, J. S., 190, 194, 205, 245 Jdioby, W. B., 158, 177 Howard, J. G., 14, 79 J a m s , D. A., 30, 40, 41, 42, 79, 80 Howie, J. B., 123, 177, 280, 288 James, A. W., 145, 150, 181 Hoyer, J., 126, 177 Jandl, J. H., 56, 79, 271, 272, 273, 286 Hsu, K. C., 192, 244 Jankovii., B. D., 127, 129, 130, 131, 176, Huang, P.-C., 330, 345 177, 178 Hudson, R. V., 113, 132, 134, 149, 150, Jansen, V . , 68, 78 175, 181 Janssen, M. A., 106, 183 Huebner, R. J., 227, 245 Jawis, J. M., 304, 307, 349 Hughes, L. E., 222, 245 J a b , J. C., 298, 339, 348, 349 Hughes-Jones, N. C., 56, 82 Jeejeebhoy, H. F., 279, 287 Hulka, J. F., 46, 47, 79 Jensen, K. E., 231, 243 Hull, P., 71, 79 Jeppesen, F., 134, 177 Hummeler, K., 189, 190, 191, 245, 250 Jerne, N., 295, 321, 339, 340, 348 Hundeshagen, H., 52, 81 Jirousek, L., 98, 184 Hung, W., 139, 140, 149, 150, 174 Jobe, A., 297, 350 Hunt, R. D., 225, 245 Johansson, B., 200, 202, 234,245, 246 Hunter, T., 282, 288 Johnson, A. M., 255, 268, 269, 286, 288 Hurez, D., 310, 350 Johnston, R. B., Jr., 260, 269, 271, 272, Hurlimann, J., 284, 290 278, 283, 286, 288 Hurvitz, A. I., 296, 345 Jonckheer, M. H., 106, 178, 180 Hutchison, R., 93, 177 Jones, B. M., 37, 79 Hutt, M. S. R., 136, 173 Jones, E. C., 62, 79 Hyde, R. R., 262, 263, 288 Jones, H. E . H., 100, 106, 129, 130, 152, 178, 181 I Jones, R. A., 56, 79 Ikawata, S., 213, 245 Jones, T. C., 225, 245 Ilk&,F. A., 50, 79 Jones, W. R., 43, 79 Illes, C. H., 310, 345 Jonsson, J., 138, 175, 178 Ilya, F. A,, 24, 79 Judge, F., 259, 287 Imamura, T., 240, 247 Juditz, E., 140, 174 Imas, B., 129, 134, 177, 183 Inbar, M., 238, 245 K Ingbar, S . H., 101, 107, 156, 175, 177 Kabat, E. A., 298, 316, 345, 351 Inman, F. P., 305, 313, 348, 351 Kabat, S., 330, 345 Inoue, M., 202, 224, 246, 247 Kadowaki, J. I., 67, 79 Irlin, I . S., 201, 245 Irvine, W. J., 88, 117, 129, 130, 134, Kafuko, G. W., 199, 233, 235, 236, 237, 245 137, 139, 140, 141, 149, 177, 178 Kagen, L. J., 257, 258, 286, 288 Isaacs, E., 111, 182 Kahn, A. I., 296, 350 Isaacs, J. J., 225, 247 Kaku, M., 46, 79 Isibul, L. R., 220, 244 Kalden, J. R., 117, 129, 130, 178 Israels, M. C . G., 51, 70, 84 Kaliss, N., 57, 58, 59, 79, 82 IHvaneski, M., 131, 178 Kano, K., 149. 150. 178 Ivanyi, P., 63, 78 Kapikian, A. Z., 148,243 Iwanow. R.. 266. 287 I
362
AUTHOR INDEX
Kaplan, A., 304, 337, 348 Kaplan, R., 44, 45, 76 Karcher, D. M., 106, 178 KPresen, R., 130, 176, 178 Karush, F., 294, 348 Katamura, A., 224, 247 Katsh, S., 15, 16, 19, 79 Kawamura, A,, Jr., 224, 247 Kay, D. J., 16, 77 Kaye, R., 281, 289 Kedes, L. H., 330, 348 Kehoe, J. M., 313, 348 Kelly, D . E., 330, 349 Kelly, J. R., 256, 289 Kelus, A. S., 300, 301, 304, 305, 346, 348, 349 Kemp, R. B., 37, 79 Kempe, C . H., 66, 79 Kennedy, T . H., 133, 173 Kerr, W . R., 21, 79 Ketcham, A. S., 201, 247 Khan, D. E., 226, 243 Kindt, T . J., 305, 309, 328, 348 King, G. S., 232, 244 King, N. W., 22.5, 245 Kinsky, R., 60, 83 Kirayama, T., 224, 247 Kirby, D. R. S., 4, 5, 9, 30, 31, 35, 36, 38, 39, 40, 41, 42, 47, 52, 73, 74, 77, 80 Kirkwood, J. M., 231, 245 Kirya, B. G., 199, 233, 236, 237, 245 Kissmeyer-Nielsen, F., 25, 81 Kisuuli, A., 235, 248 Kitahara, T., 199, 205, 248 Kite, J. H., Jr., 112, 115, 117, 120, 123, 124, 125, 132, 137, 143, 149, 150, 173, 174, 178, 182, 185 Kivikangas, V., 107, 178 Kleihauer, E., 51, 76, 80 Klein, E., 201, 202, 203, 204, 205, 209, 223, 239, 242, 244, 245, 246, 247, 249 Klein, G., 188, 189, 190, 191, 192, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 20.1, 205, 206, 207, 208, 209, 210, 211, 212, 213, 216, 218, 223, 224, 230, 231, 233, 234, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250
Klein, W. J., 211, 243 Kleniola, E., 198, 246 Klemperer, M . R., 255, 257, 259, 260, 261, 267, 268, 269, 278, 283, 286, 288, 289, 290 Klinck, G . H., 88, 178 Knopf, P. M. , 310, 348 Kohler, H . , 299, 304, 312, 348, 350 Koffler, D., 136, 178 Koh, S. W., 279, 287 Kohen, I., 211, 243 Kohler, P. F., 260, 285, 287, 288 Kohn, G., 194, 198, 239, 244, 245, 249 Kohne, D. E., 329, 335, 342, 345 Komlos, L., 61, 78 Kondo, Y., 90, 100, 101, 156, 182, 184 Konishi, K., 137, 180 Konn, M., 219, 246 Koren, Z., 42, 44, 45, 46, 80 Korngold, D., 45, 82 Korngold, L., 142, 145, 150, 178, 182 Korrington, K. J., 133, 183 Koshland, M. E., 327, 348 Kottaridis, S . D., 227, 229, 246 Kourilsky, F . M., 203, 206, 246, 248 Koutras, D. A., 134, 139, 174, 184 Kraicer, P. F., 6, 82 Kraner, K. L., 338, 350 Krause, R. M., 298, 304, 324, 325, 328, 332, 336, 348 Kritzman, I . J., 297, 348 Krohn, P. L., 26, 32, 62, 76, 79 Kuhns, W . J., 149, 178 Kumate, J., 261, 290 Kunkel, H . G., 87, 178, 285, 289, 296, 297, 298, 300, 303, 307, 320, 345, 346, 348, 349 Kurata, Y., 137, 180 Kvedar, B. J., 209, 242, 244 Ky, T., 21, 78 Kyalwazi, S. K., 187, 222, 241, 243, 249 Kyle, M., 139, 140, 149, 150, 174 Kyncl, F., 98, 184 Kyriakos, M. , 120, 121, 152, 179
1 Labaw, L., 158, 177 Lachmann, P. J., 273, 281, 282, 283, 2.84, 286, 288
363
AUTHOR INDEX
Lackland, H., 298, 304, 324, 325, 328, 332, 336, 348 Ladda, R. L., 280, 289 Lamberg, B. A., 107, 178 Lampkin, C. H., 28, 29, 32, 58, 62, 76, 77 Landerman, N. S., 254, 257, 258, 288 Landing, B. H., 139, 174 Landsteiner, K., 15, 18, 80, 108, 178, 292, 294, 348 Landucci Tosi, S., 300, 305, 308, 309, 348 Lanman, J. T., 4, 26, 31, 80 Lark, C. A., 297, 348 Lasny, S., 162, 179 Laughnan, J. R., 335, 349 Laurell, A,-B., 254, 255, 256, 258, 288, 289, 290 Laurrell, C.-B., 264,288 Laurell, S., 264, 288 Law, L. W., 188, 243, 246 Lebovitz, H. E., 299, 311, 347 Lederberg, J., 292, 348 Leech, J. B., 231, 243 Legallais, F. Y., 190, 192, 248 Leiderman, E., 198, 243 LengerovL, A,, 20, 63, 73, 80 Lennox, E. S., 299, 305, 310, 346, 348 Leon, M. A., 255, 288, 297, 349 Lepow, I. H., 254, 255, 257, 258, 271, 274, 275, 286, 287, 288, 289 Lerner, E. M., 11, 121, 122, 125, 152, 153, 178, 179 Leslie, G. A,, 298, 346 Levin, A. G., 202, 246 Levin, A. S., 267, 288 Levine, P., 18, 51, 80 Levine, P. H., 2430,246 Levine, S., 117, 129, 178 Levitsky, L. C., 281, 289 Levy, J. A., 190, 246 Levy, J. P., 206, 248 Levy, L. R., 254, 289 Levy, N. L., 280, 289 Lewis, H., 307, 349 Lewis, J., 24, 61, 79, 80 Lewis, R. T., 198, 214, 215, 243, 249 Lherisson, A. M., 201, 246 Liao, T. H., 162, 181 Liauw, H. L., 130, 182
Liberti, P., 133, 180 Lieberman, R., 304, 349 Lindberg, L. H., 280, 289 Lindegren, J., 254, 256, 288 Lindsley, D., 330,350 Ling, N. R., 137, 181 Linman, J. W., 285, 286, 289 Liozner, A. L., 232, 249 Lippman, M., 37, 80 Lipsett, M. B., 24, 79 Lipton, M. M., 16, 78 Lischner, H. W., 67, 81 Lissitzky, S., 106, 107, 155, 157, 159, 161, 162, 163, 164, 174, 178, 181, 184 Little, C. C., 29, 80 Lobo, L. C. C., 97, 179 Loeb, E. M., 44, 82 Loeb, N., 328, 348 Lowenstein, J., 310, 347 Lugeinbuhl, R. E., 227, 229, 246 Lukes, R. J., 67, 82, 338, 350 Lundback, H., Zoo, 249 Lundgren, H. P., 158, 179 Lundh, B., 258, 288, 289 Lupulescu, A., 88, 179 Lustgraff, E. C., 58, 80 Lyon, M. F., 301, 348 Lytton, B., 222, 245
M McCallum, H. M., 138, 176 McCarthy, B. J., 297, 330, 348 McCaughey, W. T. E., 44, 80 Macchia, V., 163, 180 McCluskey, R. T., 283, 290 McCollum, R. W., 197, 198, 233, 237, 244, 247 McCombs, R. M., 214, 215, 249 McConahey, W. M., 137, 141, 173, 174 McCormick, K. J., 216, 246 McCormick, N., 308, 351 McDermott, W. V., Jr., 139, 175 McDevitt, H. O., 339, 348, 349 MacDonald, A. B., 298, 349 McDougall, C. D. M., 139, 175 McDuffie, F. C., 285, 286, 289 McFarland, V. W., 238, 247 McFarlane, H., 211, 247 McGirr, E. M., 107, 184
364
AUTHOR INDEX
Margolies, M. N., 298, 348 Marrack, J. R., 305, 349 MBrtensson, H., 254, 256, 288 Martin, W. J., 70, 78 Martinek, J. J., 38, 39, 81 Martino, E., 133, 180 Martos, L. M., 192, 244 Maruta, H., 17, 81 Masaitis, L., 56, 83 Masi, A. T., 139, 179 348 Masouredis, S. P., 56, 81 McKeogh, R. P., 58, 76 Mates, G. P., 105, 146, 147, 166, 167, McKinnell, R. G., 231, 246 168, 169, 179, 183 MacKnight, J. F., 277, 287 McLaren, A,, 4, 5, 6, 16, 41, 52, 75, 77, Math6, G., 232, 242, 243, 247 Matos, M., 107, 184 78, 331, 351 Matsuda, T., 137, 180 MacLaren, J. A., 284, 286 Mattinson, G., 375, 345 McLaughlin, C. L., 313, 350 Mauchamp, J., 163,180 McLean, J. M., 11, 80 McMaster, P. R. B., 87, 120, 121, 122, Mauersberger, D., 283, 290 Maurer, B. A., 240, 247 125, 152, 153, 178, 179 Maxwell, W . T., 145, 150, 181 McNicol, G. P., 136, 176 May, J., 287 Macpherson, I., 238, 246 Mayyasi, S. A., 191, 192, 214, 247 McQuillan, M. T., 164, 179 .Means, J. H., 87, 90, 179 Mac&, M. T., 52, 80 Medawar, P. B., 4, 15, 26, 27, 28, 29, McWhirter, K. G., 73, 80 32, 44, 55, 62, 70, 76, 77 Maenpi%, J., 107, 178 Medhurst, F. A., 278, 287 Mafigiri, J., 235, 248 Mage, R. G., 300, 305, 308, 309, 319, Melendez, L. V., 225, 245 328, 337, 345, 346, 347, 348, 349 Mellors, R. C., 134, 179, 297, 348 Melnick, J. L., 199, 205, 248 Malan, L., 200, 246 Meltzer, M., 310, 347 Maleci, O., 21, 83 Mergenhagen, S. E., 260, 285, 287 Malkiel. S.. 117. 177 Merler, E., 258, 287 Malmg;en,’R. A., 201, 247 Merrill, J. P., 256, 257, 287 Malmquist, J., 288 Metalnikoff, S., 15, 81 Malmros, J., 254, 256, 288 Metchnikoff, E., 15, 81 Manaker, R. A., 220, 249 Mandy, W. J., 300, 305, 308, 309, 348, Metzgar, R. S., 57, 82, 111, 125, 129, 130, 150, 179, 180, 182 349, 350, 351 Mann, D. L., 222,246 Metzger, H., 168, 179, 304, 316, 337, Mannik, M., 337, 349 347, 348 Manolov, G., 203, 2.04,209, 242, 244, Meyer, G., 201, 205, 246, 247 246, 247 Michaelides, M. C., 297, 343, 347 Mante, S., 106, 161, 178 Michel, R., 99, 156, 157, 175 Marchant, D. J., 58, 76 Michie, D., 5, 14, 64, 72, 78, 79, 80 Marcus, A. J., 310, 345 Mickey, M. R., 57, 70, 81, 83 Marcus, G. J., 4, 81 Miescher, P. A., 126, 179, 183 Mardiney, M. R., 230, 290 Mikaesco, E., 310, 350 Margherita, S. S., 160, 179 Margoliash, E., 315, 316, 317, 318, 322, Milgrom, F., 87, 108, 109, 179 Miller, B. F., 64, 81 340, 347, 349
McGivray, E., 285, 288 McGovern, J. P., 258, 286 McGovern, P. T., 43, 78 MacGregor, A. G., 134, 177 Macher, E., 13, 80 McIntire, R., 326, 348 Mack, H. C., 50, 82 McKay, D. G., 38,82 McKean, D. J., 304, 320, 324, 325, 326,
365
AUTHOR INDEX
Miller, G., 240, 247 Miller, M. E., 281, 289 Miller, M. H., 240, 247 Miller, 0. J., 285,288 Miller, P. J., 305, 346 Mills, G. L., 100, 106, 181 Milos’evib, D., 129, 178 Milstein, C., 296, 297, 299, 304, 305, 307, 309, 313, 320, 339, 341, 346, 347, 349 Milstein, C. P., 304, 307, 349 Milstein, S., 340, 345 Minowada, J., 203, 246 Mitchill, G. W . , Jr., 58, 76 Mitchison, N. A., 26, 54, 81 Mitrovih, K., 129, 131, 178 Mitze, F., 283, 289 Miura, M., 235,249 Mizejewski, G. J., 122, 138, 179 Mizell, M., 225, 231, 232, 244, 245, 247 Mkheidze, D. M., 232, 249 Moller, E., 37, 81 Moller, G., 37, 81, 201, 247 Mogabgab, W . J.. 198, 243 Mogensen, B., 25, 81 Mohr, K., 47, 79 Mole, L. E., 327, 349 Monaco, F., 88, 179 Monroe, C. W., 32, 76 Monroe, J. H., 195, 239, 244 Monton, D., 278, 290 Moore, D. H., 228, 249 Moore, F. D., 268, 269, 286 Moore, H. D., 262, 289 Moore, J. M., 141, 179 Moore, V., 304, 348 Moos, C., 157, 183 Mora, P. T., 238, 247 Morat6-Manaro, J., 134, 180 Morris, J., 3C9, 305, 308, 319, 325, 328, 337, 348 Morrow, R. H., 199, 211, 218, 222, 223, 233, 235, 2.36, 237, 244, 245, 248, 249 Morse, H. H., 285, 288 Mortensen, J. D., 134, 179 Mortimer, R. K., 333, 347 Morton, D. L., ‘201, 247 Morton, M. E., 100, 179
Motulsky, A. G., 141, 176 Moulthrup, J. I., 230, 249 Moulton, M. A,, 64,81 Mouriz, J., 103, 179 Moyer, D. L., 17, 81 Mozes, E., 339, 349 Mueller, P. S., 121, 122, 152, 179 Muller-Eberhard, H. J., 256, 257, 272, 273, 277, 279, 282, 283, 284, 285, 286, 288, 289, 290 Muir, C. S., 237, 247 Muir, A. R., 88, 177 Munoz, J., 121, 179 Munro, A. J., 282, 288, 305, 339, 341, 349 Munro, D. S., 133, 183 Munube, G. M. R., 199, 233, 236, 237, 245 Murakami, W. T., 238, 244 Muramatsu, T., 166, 183 Murray, I. P. C., 136, 176 Muschel, L. H., 260, 287 Musser, E. A., 122, 179
N Nadkami, J. J.. 202, 203, 204, 205, 216, 223, 231, 239, 240, 242, 245, 246, 247, 249 Nadkarni, J. S., 202, 203, 204,205, 216, 223, 231, 239, 240, 242, 245, 246, 247, 249 Naff, G. B., 257, 289 Nagaki, K., 286, 290 Nagel, B., 61, 80 Naiman, J. L., 67, 81 Naim, R. C., 137, 179 Najarian, J. S., 64, 81 Nakamura, R. M., 118, 119, 120, 121, 130, 142, 153, 154, 179, 180, 184 Narasimha Murthy, P. V., 164, 180 Nathan, P., 64, 81 Natvig, J. B., 300, 349 Nazerian, K., 226, 227, 247 Neauport-Sautes, C., 206, 248 Neelands, P., 51, 84 Neerhout, R., 70, 83 Negrin, E., 134, 180 Neilson, J. M., 141, 179 Nelson, J. H., 11, 81
366
AUTHOR INDEX
Nelson, R. A., Jr., 282, 289 Nemirovsky, M., 18, 83 Nevanlinna, H. R., 56, 81 NBve, P., 88, 180 Ngu, V. A., 187, 210, 211, 242, 247 Niall, H., 304, 349 Nicholas, J. S., 26, 81 Niederle, B., 98, 184 Niederman, J. C., 197, 198, 203, 204, 207, 233, 237, 244, 246, 247 Nigogosyan, G., 50, 82 Nilsson, K., 194, 247 Nilsson, U. R., 272, 277, 281, 282, 283, 284, 289, 290 Nishioka, K., 209, 224, 242, 247 Nishiyama, R. H., 122, 129, 174, 175 Nisonoff, A., 298, 300, 307, 308, 313, 348, 349, 350, 351 Nissley, P., 162, 180 Nolan, C., 322, 340, 349 Northington, J. W., 198, 243 Nossal, G. J . V., 44, 81, 307, 349 Notani, G., 297, 346 Novotny, J., 316, 347 Noyes, R. W., 6, 78 Noyes, W. F., 228, 247 Nunez, J., 101, 163, 180, I81 Nussenzweig, V., 301, 347
0 Ober, W., 44, 49, 81 O’Connell, E. J., 285, 286, 289 O’Conner, M., 4, 83 O’Conor, G. T., 190, 192, 248 O’Donnell, I. J., 157, 180 Oehme, J., 52, 81 Oertel, J. E., 88, 178 Oettgen, H. F., 207, 210, 217, 247, 248 Oettingen, Kj. V., 33, 81 Ogston, D., 257, 289 Ohms, J., 298, 304, 309, 324, 325, 328, 332, 336, 348, 350 Ohno, S., 314, 375, 345, 349 Ohta-Hatano, R., 196, 245 Okada, N., 277, 289 Okada, S., 137, 180 Okazaki, W., 230, 247 Okholm, K., 134, 177 Okuyan, M., 227, 245
Old, L. J., 188, 201, 202, 203, 207, 208, 209, 210, 217, 218, 231, 232, 237, 238, 242, 245, 246, 247, 248, 249 Olds, P. J., 30, 81 Olin, P., 142, 180 Oppenheim, J. J., 61, 80 Oppermann, W., 134, 175 Ordonez, L., 137, 180 Orsini, M., 9, 81 Ortega, M. L., 281, 282, 284,286 Osler, W., 253, 289 Osserman, E. F., 310, 349 Osterland, C. K., 297, 349 Osunkoya, B. O., 211, 247, 248 Oswald, A,, 93, 180 Otten, J., 106, 178, 180 Oudin, J., 304, 305, 346, 349 Ounsted, C., 75, 81 Ounsted, M., 75, 81 Ovary, Z., 150, 180 Owen, C. A., Jr., 87, 180 Owen, J. J . T., 52, 77 Owen, R. S., 295, 350 Owen, S. G., 133, 139, 176, 180 Owens, A. H., 65, 82 Ozerkis, A. J.. 32, 82
P PackalBn, T., 127, 130, 180, 184 Paine, J. R., 113, 115, 125, 128, 129, 130, 132, 183, 185 Paine, P. J., 45, 80 Palm, J., 29, 30, 62, 72, 77, 81 Palmer, W . W., 93, 177 Palmstierna, H. A. K., 130, 180 Papiernik-Berhauer, E., 4, 81 Park, W. W., 4, 81 Parker, C. W., 297, 349 Parkhouse, R. M . E., 310, 348 Paronetto, F., 136, 178 Parr, D. M., 310, 350 Parshall, C. J., 67, 82 Paseyro, P., 134, 180 Pasternak, G. I., 188, 201, 232, 238, 248 Pate], R., 70, 81 Paterson, P. Y., 87; 117, 129, 130, 180 Patrick, R. A., 271, 288 Paul, C., 299, 304, 312, 348, 350 Pauling, L., 294, 349 Payne, R., 57, 81
367
AUTHOR INDEX
Pearson, G., 190, 191, 196, 203, 204, 205, 206, 207, 223, 239, 244, 245, 246, 248 Pedersen, G. T., 142, 177 Pedersen, K. O., 93, 94, 99, 157, 158, 175, 177 Peer, L. A,, 62, 63, 81 Penn, G. M., 307,349 Penn, I., 188, 248 Pensky, J., 254, 255 257, 274, 287, 289 Peptiopules, S. F., 235, 249 Percy, M. E., 310, 350 Perelmutter, L., 103, 180 Pernis, B., 301, 307, 349 Perry, E., 330, 349 Perry, S., 61, 80 Peter-Knecht, W., 206, 248 Petersen, K. F., 283, 289 Peterson, H. M., 335, 349 Peterson, R. E., 103, 174 Pettit, M. D., 139, 174 Pfeiffer, E. F., 134, 175 Phillips, M. E., 280, 284, 289, 290 Pickering, R. J., 256, 260, 285, 286, 287, 289, 290 Pierce, E. G., 242, 245 Pierce, J. G., 161, 162, 180, 181 Pike, M . C., 235, 248 Pike, P., 199, 233, 236, 237, 245 Pillemer, L., 254, 275, 286, 288 Pinchera, A., 133, 180 Pinder, J. C., 330, 349 Pink, J. R. L., 299, 320, 341, 349 Pitt-Rivers, R., 87, 89, 90, 155, 161, 162, 175, 180 Planta, R. J., 330, 350 Plaskett, L. G., 142, 174 Pluinm, M. N., 304, 346 PokornP, Z., 19, 83 PolhEkova, M., 19, 83 Polani, P. E., 141, 175 Pollack, J., 51, 84 Pollack, R. E., 238, 248 Pollack, W., 56, 78 Polley, M. J., 260, 272, 289 Pommier, J., 163, 180 Ponstingl, H., 303, 320, 347 Pope, J. H., 190, 194, 195, 214, 215, 216, 219, 221, 239, 248, 249 Popeskovik, L., 129, 131, 178
Poppa, G., 9, 10, 81 Porter, D. D., 134, 180, 214, 249 Porter, J. B., 58, 81 Porter, R. R., 297, 299, 327, 346, 349 Posner, I., 137, 180 Potter, J. F., 53, 81 Potter, M., 296, 297, 304, 320, 324, 325, 326, 338, 347, 348, 349 Prahl, J. W., 305, 309, 313, 327, 349, 350 Pratt, 0. E., 142, 174 Prehn, R. T., 20, 82, 237, 248 Premachandra, B. N., 130, 150, 160, 179, 180, 181 Prendergast, R. A., 67, 82, 338, 350 Press, E. M., 306, 350 Pressman, D., 45, 82, 145, 150, 181 Propp, R. P., 107, 176, 257, 264, 266, 268, 269, 286, 289 Proteous, I. B., 137, 179 Pry, T. W., 220, 249 Pullen, R., 62, 63, 81 Pulvertaft, R. J . V., 149, 150, 181, 191, 248 Punnett, H. H., 67, 81 Purchase, H . G., 225, 226, 228, 230, 243, 247, 248 Putnam, F. W., 299, 304, 312, 348, 350
Q Quattrocchi, R., 342, 350 Queng, J. T., 258, 286 Quinlivan, W. L. G., 52, 83
R Rabbat, A. G., 279, 287 Rabinowitz, Z., 238, 248 Rabson, A. S., 190, 192, 248 Race, R. R., 54, 82 Raghupathy, E., 164, 180 Rall, J. E., 90, 98, 99, 100, 106, 155, 157, 175, 181, 182, 183, 185 Ramagopal, E., 100, 106,181 Ramseier, H., 20, 64, 82 Randall, H., 150, 180 Rapaport, F. T., 13, 74, 78 Rapp, F., 199, 201, 205, 248, 249 Rapp, H . J., 276, 285, 287, 290 Ratcliffe, H . E, 257, 258, 288
368
AUTHOR INDEX
Ratnoff, 0. D., 254, 257, 258, 287, 289 Rawitch, A. B., 161, 162, 180, 181 Rawson, R. W., 100, 106, 183 Rawstron, J. R., 136, 181 Ray, A. K., 130, 150, 160, 181 Rea, J. S., 297, 351 Rebello, M. A., 97, 179 Reedman, B. M., 190, 195, 214, 219, 221, 248 Refetoff, S., 268, 286 Reich, H., 33, 83 Reid, B. L., 21, 82 Reisfeld, R. A., 300, 305, 328, 337, 348, 350 Rejnek, J., 305, 328, 345 Reynolds, S. R. M., 5, 82 Retel, J., 330, 350 Reutter, W., 284, 287 Reynaud, J., 162, 179 Rice, J. A., 297, 346 Rich, M. A., 236, 248 Richards, C. B., 305, 349 Ritossa, F. M., 330, 350 Ritter, M. A., 52, 77 Rivat, L., 263, 289, 328, 350 Robbins, J., 90, 98, 99, 100, 103, 105, 106, 145, 155, 156, 164, 181,182,184, 185 Robert, B., 21, 78 Roberts, T. K., 16, 77 Robertson, M., 21, 79 Robin, N. I., 268, 286 Roche, J., 99, 101, 106, 156, 157, 173, 175, 180, 181 Rochlin, D. B., 52, 83 Rochna, E., 56, 82 Rogentine, G . N., 222, 246 Roitt, I. M., 86, 87, 100, 106, 107, 124, 129, 130, 132, 133, 134, 136, 137, 138, 139, 140, 141, 143, 145, 147, 148, 149, 150, 152, 169, 173, 174, 175, 176, 178, 181, 184 Roizman, B., 206, 223, 248 Rolfs, M. B., 57, 81 Rolland, M., 155, 159, 161, 162, 174, 178, 179, 181, 184 Rona, G., 129, 176 Ropartz, C., 263, 289, 328, 350
288,
215,
345,
104, 176,
163,
113, 135, 142, 151, 177,
163,
Roques, M., 155, 163, 174, 184 Rose, M., 265, 2.66, 287, 289 Rose, N. R., 86, 87, 93, 95, 96, 97, 98, 100, 107, 110, 111, 112, 113, 114, 115, 117, 120, 125, 12.8, 129, 130, 131, 132, 134, 137, 143, 149, 150, 169, 173, 175, 178, 180, 181, 182, 183, 184, 185 Rosen, F. S., 254, 255, 256, 257, 258, 259, 260, 261, 267, 268, 269, 270, 271, 272, 273, 278, 283, 286, 287, 288, 289, 290 Rosenberg, L. T., 276, 277, 278, 279, 280, 286, 287, 288, 289, 290 Rosenberg, M., 301, 350 Rosenfeld, S. I., 284,289 Rossen, R. D., 285, 288 Rother, K. O., 280, 281, 2.82, 283, 284, 288, 289, 290 Rother, U. A., 280, 281, 282, 283, 284, 288, 289, 290 Rousseau, P.-Y., 263, 289 Rovis, L., 133, 180 Rowson, L. E. A., 26, 76 Rubenstein, P., 59, 79, 82 Rubin, H., 235, 248 Ruddle, N. H., 154, 182 Ruddy, S., 255, 256, 257, 261, 284, 286, 287, 289, 290 Ruebner, B., 142, 182 Russell, P. S., 4, 9, 10, 19, 30,31, 32, 34, 35, 39, 40, 41, 51, 68, 76, 81, 82 Rutishauser, U., 299, 304, 312, 313, 346
s Sachs, L., 238, 245, 248 Sairenji, T., 196, 245 Salabe, G., 90, 182 Salvaggio, A. T., 50, 82 Salvatore, G., 103, 104, 105, 106, 156, 163, 173, 175, 181, 182 Salvatore, M., 103, 104, 105, 106, 182 Salvin, S. B., 130, 182 Samuels, B. D., 68, 78 Sanders, B. G., 301, 304, 305, 306, 307, 335, 336, 337, 347 Sanford, B. H., 62, 82 Sanger, R., 54, 82 Santay, H., 24, 33, 83 Santesson, L., 204, 241, 250
AUTHOR INDEX
Santos, G. W., 65, 82 Sarma, P. S., 227, 245 Sarma, R., 299, 313, 346 Savary, J., 162, 179 Scarth, L., 140, 141, 177 Schaaf, J., 46, 79 Schatz, H., 134, 177 Schidlovsky, G., 191, 192, 214, 231, 243, 247 Schindera, F., 283, 289 Schlesinger, M., 9, 82 Schmitter, R., 195, 244 Schmittle, S. C., 228, 244 Schmutzler, W., 284, 287 Schnitzler, S., 266, 287 Schoeneman, M., 53, 81 Schubert, D. A., 297, 310, 350 Schulenberg, E. P., 297, 343, 347 Scliulhof, K., 107, 109, 111, 142, 177 Schullinger, J. N . , 285, 288 Schulte-Holthausen, H., 190, 204, 215, 239, 2.11, 250 Schultze, H. E., 254, 290 Schwartz, H. L., 161, 180 Schweet, R. S., 295, 350 Schwick, H . G., 254, 255, 286, 289 Schwimmer, W . B., 17, 82 Sclare, G., 140, 182 Scothorne, P. J., 11, 80 Scott, J . S., 4, 82 Scott, T . W., 137, 182 Scott, W., 194, 214, 239, 248 Scriba, M., 1W, 191, 196, 223, 239, 245 Seals, J., 281, 289 Seamark, R. F., 107, 175 Seegal, B. C., 44, 45, 76, 82 Segal, D., 299, 313, 346 Segers, J., 264, 290 Seigler, H. F., 57, 82 Sekine, T., 224, 247 Sela, M . , 339, 349 Selenkow, H . A,, 134, 174 Seligmann, M., 310, 350 Sell, S., 127, 130, 176 Sena, I., 106, 163, 173, 181 Shahani, S., 18, 82 Shanmugaratnam, K., 237, 247 Sharp, C. C., 168, 179 Sheffer, A. L., 256, 257, 286, 290 Shehadeh, I . H., 256, 257, 287
369
Shelesnyak, M. C., 4, 6, 81, 82 Shimizu, A., 299, 304, 312, 348, 350 Shin, H . S., 278, 279, 290 Shinoda, T., 299, 312, 350 Shope, R. E., 228, 249 Shulman, L. E., 139, 179 Shulman, S., 86, 89, 93, 94, 95, 96, 97, 98, 100, 101, 102, 104, 105, 106, 107, 112, 113, 115, 116, 117, 118, 124, 125, 138, 140, 143, 144, 145, 146, 147, 148, 149, 150, 151, 156, 157, 163, 165, 166, 167, 168, 169, 176, 179, 180, 182, 183 Siboo, R., 255, 288, 290 Sigiienza, R. F., 192, 244 Siltzbach, L. E., 200, 245 Silvers, W . K., 6, 25, 28, 29, 30, 36, 53, 58, 62, 64, 65, 68, 69, 72, 73, 76, 77, 82, 83 Silverstein, A. M . , 13, 67, 82, 259, 290, 338, 350 Silverton, E., 299, 313, 346 Silvestre, D., 206, 248 Simkovic, D., 235, 249 Simmons, R. L., 4, 9, 10, 30, 31, 32, 34, 35, 38, 39, 40, 41, 68, 81, 82 Simnis, E. S., 297, 343, 347 Sinionsen, M., 61, 82 Singer, S. J., 311, 316, 350, 351 Singh, S., 211, 242, 244, 249 Singhal, S. K., 202, 248 Siskind, G. W., 297, 350 Sisson, J. C., 88, 183 Sjogren, H . O., 188, 201, 209, 232, 238, 246, 248, 249 Skelton, F. R., 132, 143, 149, 150, 182 Skillman, T. G., 134, 174 Skipper, H . E., 242, 249 Skoog, N., 264, 288 Skittery, S. M., 231, 243 Small, P. A., Jr., 300, 350 Smart, G. A., 133, 139, 176, 180 Smeds, S., 88, 183 Smith, H. R., 133, 183 Smith, D., 104, 183 Smith, D. E., 87, 90, 177 Smith, D. H., 278, 288 Smith, D. J., 105, 163, 166, 183 Smith, G., 136, 173 Smith, G. P., 298, 299, 300, 305, 306,
370
AUTHOR INDEX
321, 322, 323, 325, 326, 328, 329, 334, 341, 342, 350 Smith, I., 330, 349 Smith, K., 52, 83 Smith, M. R., 278, 279, 290 Smith, R. T., 203, 205, 246, 249 Smithies, O., 265, 266, 268, 286, 292, 310, 311, 315, 341, 350 Snapper, I., 296, 350 Snell, F. M., 157, 183 Snell, C. D., 59, 70, 83 Snyderman, R., 285, 287 S o r h , L., 61, 83 Solomon, A., 313, 350 Solomon, D. H., 133, 183 Solomon, I. L., 149, 150, 176 Solomon, J. J., 227, 247 Sonkin, L. S., 134, 179 Soto, J. R., 134, 183 Southam, A. L., 18, 82 Southam, C. M., 235, 249 Southern, F. M., 331, 350 Sox, H. C., Jr., 336, 337, 348 Sparrow, E. M., 26, 29, 32, 54, 76, 79, 81 Spehl, P., 110, 173 Spencer, R. P., 157, 183 Spiegelberg, H. L., 126, 183 Spiegelman, S., 330, 350 Spier, R., 143, 149, 150, 182 Spiro, M. J., 100, 106, 156, 159, 160, 161, 162, 164, 165, 181, 183 Spiro, R. C., 164, 165, 183 Spiber, R., 273, 290 Sprenger, W. W., 27, 77 Spring, S. B., 206, 248 Stackpole, C. W . , 231, 247 Stamp, J. T., 43, 78 Stanbury, J. B., 87, 90, 100, 103, 106, 179, 181,183 Stanley, P. G., 98, 101, 107, 156, 157, 161, 180, 182, 183 Starzl, T . E., 188, 248 Stastny, P., 64, 83, 134, 185 Steblay, R. W., 45, 83 Stecher, V. J., 284, 290 Stein, R., 297, 351 Steinberg, A. G., 300, 304, 306, 350 Steinbuch, M., 263, 289 Steiner, R. F., 158, 183
Steinmuller, D., 72, 77 Stenike, G. W., 305, 350 Stenback, W. A., 216, 246 Stephenson, N. R., 103, 180 Stern, K., 56, 83 Sterzl, J., 338, 350 Stetson, R. E., 51, 80 Stevens, D. A., 220, 249 Steward, M. W., 309, 350 Stewart, A. C., 140, 177 Stewart, T. H . M., 222, 249 Stiffel, C., 278, 290 Stimpfling, J., 64,81 Stitt, D., 240, 247 Stjernsward, J., 201, 211, 242, 244, 246, 249 Storiko, K., 255, 286 Stokinger, H. E., 110, 148, 183 Stone, R. S., 228, 249 Storer, J. B., 64, 81 Strom, R., 202, 246 Stroud, R. M., 286, 290 Stuart, A. E., 134, 177, 183 Stuckey, W . J., 198, 243 Stuck, B., 232, 249 Stylos, W. A., 169, 181, 183 Subrahmanyan, L., 233, 237, 247 Sugano, H., 224, 247 Sumerling, M. D., 140, 177 Suter, L., 303, 320, 347 Svedmyr, A., 205, 249 Svedmyr, E., 242, 249 Svet-Moldavsky, G. J., 232, 249 Swiech, K., 209,244 Szilard, L., 292, 350
T Tachibana, D. K., 276, 277, 278, 280, 287, 289, 290 Tachibana, T., 209, 224, 242, 247 Tai, C., 39, 83 Takada, M., 224,247 Takatsuki, K., 310, 349 Talamo, R. C., 258, 290 Talmage, D. W., 19, 79, 292, 295, 303, 304, 315, 322, 324, 325, 337, 340, 343, 347, 350 Tamura, H., 100, 101, 156, 184 Tan, E. M., 87, 178
AUTHOR INDEX
371
Tanaka, S., 235, 249 Trentin, J. J., 216, 246 Tanford, C., 295, 337, 345, 351 Trickojus, V. M., 137, 162, 164, 175, 179. 182 Tapley, D. F., 160, 174 Tartof, K. P., 330, 349 Trotter, ‘W. R., 87, 135, 156, 173, 174, Tarutani, O., 100, 101, 104, 156, 163, 180, 184 Tucker, E. M., 28, 78 165, 166, 183, 184 Tata, J. R., 100, 106, 183 Tucker, W. E., 122, 184 Taylor, G., 140, 182 Tuffrey, M., 52, 83 Tukei, P. M., 199, 233, 236, 237 245 Taylor, K. B., 139, 140, 141, 175 Teisberg, P., 266, 268, 290 Turner, H. C., 227, 245 Teitlebaum, M. S., 73, 80 Turner, J. H., 52, 83 Temler, J., 134, 177 Twarog, B., 123, 178 Twarog, F. J., 117, 129, 131, 184 ten Bensel, R., 260, 287 Terasaki, P. I., 57, 70, 81, 83 Tweedell, K. S., 231, 249 Terni, T., 21, 83 Tyler, A., 4, 15, 48, 83 Terplan, K. L., 113, 115, 125, 128, 129, 130, 132, 183, 185 U Terry, W. D., 276, 290, 297, 304, 306, Ubertini, T., 228, 243 309, 337, 348, 350, 351 Uchida, I. A., 141, 176 Tevethia, S. S., 201, 249 Udeozo, I. 0. K., 211, 247 Themann, H., 132,184 Uhr, J. W., 31, 83, 338, 350 Theron, C. N., 103, 184 Ui, N., 100, 101, 105, 156, 163, 181, Thiede, H. A., 24, 33, 83 183, 184 Thomas, C. A., Jr., 333, 351 Ujejski, L., 164, 184 Thomas, L., 48, 83 Ulrich, M., 276, 290 Thomas, 0. C., 258,286 Unanue, E. R., 280, 290 Thompson, G. E., 16, 78 Underdown, B. J., 297, 343, 347 Thompson, R. I., 67, 79 Urquhart, J. A., 137, 179 Thomson, J. A., 97, 107, 184 Ustay, K. A., 17, 82 Thorbecke. G. J., 280, 284, 289, 290 Utsumi, S., 313, 346, 351 Thorpe, N:, 316, 350 Thorpe, N. O., 316, 351 Timbury, G. C., 140, 173 V Todaro, G. J., 238, 248 Vainio, T., 56, 81 Todd, C. W., 300, 305, 308, 309, 319, 325, 328, 337, 346, 348, 349, 350, Valenta, L., 98, 157, 163, 184 Valentine, R. C., 299, 351 351 van der Scheer, J., 108, 178 Todd, E. W., 274, 289 Van der Walt, B., 103,184 Toivanen, P., 75, 83 van Doorninck, W., 37, 78 Tomasi, T. B., 140, 174 van Furth, R., 202, 249 Toplin, I., 225, 247 van Leeuwen, A., 57, 83 Tornivov, A., 16, 83 van Leeuwen, G., 142, 145, 150, 178,182 Toro-Goyco, E., 107, 184 van Rood, J. J., 57, 83 Torresani, J., 155, 163, 174, 184 Van Tienhoven, A., 122, 184 Torrigiani, G., 107, 142, 151, 174, 175, Van Zyl, A,, 103, 184 181, 184 Vassalli, P., 283, 290, 301, 347 Toullet, F., 18, 83 Vecchio, G., 103, 104, 105, 142, 158, Toy, B. L., 33, 36, 76 180, 181, 182, 184 Vice, J. L., 305, 351 Traul, K. A,, 195, 244
372
AUTHOR INDEX
Viera, B. D., u)o,245 Voisin, G. A., 16, 18, 44, e0, 83, 141, 184 Volk, H., 283, 290 Vojtskovi, M., 19, 2.0, 80, 83 von Essen, R., 198, 246 Vonka, V., 214,215, 249 Vredovoe, D., 57, 83
W Wachtel, S. S., 30, 69, 83 Waddams, A., 135, 184 Waggoner, D. E., 200, 220, 244, 246 Wahlberg, P., 134, 184 Wahren, B., 200, 249 Wakim, K. C., 137, 173 Waksman, B. H., 87, 124, 125, 154, 175, 182, 184 Wald, N., 52, 83 Walford, R. L., 70, 83 Walia, I. S., 62, 63, 81 Walker, J. L., 192, 244 Walker, P. M . B., 330, 331, 347, 351 Walknowska, J., 52, 83 Wallace, H., 330, 351 Wallace, J., 231, 245 Walters, M. K., 190, 195, 214, 215, 216, 219, 221, 248, 249 Walzer, R. A., 285, 288 Wang, A. C., 307, 308, 351 Wang, C. H., 224,247 Wang, I . Y. F., 308, 351 Ward, P. A., 279, 284, 290 Wardlaw, A. C., 276, 277, 287 Warner, N., 296, 326, 347, 351 Warner, N . L., 307, 349 Warren, L., 164, 185 Wanvick, B. L., 26, 83 Wasserman, J., 127, 130, 180, 184 Watanabe, S., 303, 320, 347 Waterfield, M., 298, 300, 305, 308, 319, 325, 326, 328, 337,348 Watkins, J. F., 224, 249 Watson, L., 285, 286 Watson, R. J., 52, 80 Waubke, R., 190, 191, 196, 223, 239, 245 Waxdal, M., 299, 312, 313, 346 Wayne, E. J., 134, 139, 174
Weathers, B., 103, 184 Weber, C. S., 330, 345 Webster, M. E., 257, 258, 288 Wedderburn, N., 235, 249 Wedgwood, R. J., 274, 289 Weigert, M. G., 320, 322, 325, 337, 351 Weigle, W. O., 118, 119, 120, 121, 130, 142, 153, 154, 179, 180, 184 Weinberg, A., 193, 194, 249 Weir, B. J., 8, 77 Weiss, L., 149, 181 Wells, G. A., 192, 244 Wenk, E. J., 117, 129, 178 Werner, S. C., 92, 134,184 West, C. D., 139, 174, 273, 290 Westphal, H., 241, 249 Wetters, E. J., 214, 248 Wetterqvist, H., 258, 289 Whang, J. J., 61, 80, 190, 192, 248 Whang-Peng, J., 195, 239, 244 Wheeler, A. H., 139, 174 Wheeler, W. E., 134, 174 White, R. C., 135, 141, 184 White, T., 258, 289 Whitehouse, H. L. K., 334, 351 Whitney, P. L., 295, 351 Wick, C., 123, 124, 178, 185 Wieme, R. J., 263, 264, 290 Wiener, F., 238, 246, 249 Wigzell, H., 202, 205, 223, 242, 245, 248 Wilbert, S. M., 240, 247 Wilkinson, J. H., 149, 181 Wilkinson, J. M., 325, 327, 349, 351 Willett, E . L., 26, 83 Williams, E., 141, 184 Williams, E. A,, 39, 77 Williams, E. D., 141, 185 Williams, H. L., 28, 29, 77 Williams, J. W., 157, 158, 179, 180 Williamson, B., 217, 248 Williamson, M. E., 225, 245 Williamson, S., 24, 79 Williamson, W. C., 117, 129, 130, 178 Willms-Kretschmer, K., 257, 258, 271, 287, 290 Wilson, D. B., 58, 61, 77, 83 Wilson, E. G., 136, 181 Wilson, S. K., 307, 308, 351 Wimberly, I., 214, 215, 249
373
AUTHOR INDEX
Winn, H. J., 61, 83 Winship, T., 88, 178 Wissig, S. L., 87, 88, 185 Witebsky, E., 33, 81, 83, 86, 87, 93, 94, 95, 96, 97, 98, 100, 101, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 120, 123, 124, 125, 128, 129, 130, 132, 134, 135, 136, 143, 144, 145, 146, 147, 148, 149, 150, 151, 174, 176, 178, 179, 182, 183, 184, 185 Witter, R. L., 227, 230, 247, 249 Woernley, D., 145, 150, 181 Wolff, J., 98, 99, 100, 106, 181, 185 Wollman, S. H., 88, 164, 180, 185 Wolstenholme, G. E. W., 4, 83 Wood, W. B., Jr., 278, 279, 290 Woodland, H. R., 293, 347 Woodrow, J. C., 56, 84, 139, 175 Woodruff, M. F. A., 26, 84 Woodworth, H. C., 259, 260, 288 Wooley, P. V., 67, 79 Woolner, L. B., 134, 179 Work, T. S., 101, 156, 177 Wu, T. T., 316, 351 Wudarski, D. J., 191, 245 Wynn, R. M . , 39, 84 Wynn, J. A,, 160, 176
Y Yagi, Y., 145, 150, 181 Yamaguchi, J., 191, 219, 245, 246 Yamazaki, J. N., 57, 83 Yang, J. P. S., 242, 245 Yang, S. L., 20, 46, 65, 76 Yata, J., 206, 209, 239, 248, 249 Yohn, D. S., 219, 246 Yonkovich, S. J., 320, 322, 325, 337, 351 Yoshida, T., 224, 247 Young, G. O., 300, 305, 319, 346, 347, 348
Z Zajac, B. A., 190, 191, 196, 223, 239, 244, 245, 249 Zappi, E., 115, 176 Zavaleta, A., 134, 185 Zeigel, R. F., 194, 244 Zervas, J. D., 70, 84 Zettervall, O., 297, 351 Ziegler, J. L., 199, 211, 218, 221, 222, 223, 233, 236, 237, 244, 245, 249 Zipursky, A., 51, 84 Zuelzer, W. W., 51, 55, 67, 77, 79 Zullo, D. M., 305, 308, 351 Zur Hausen, H., 190, 194, 204, 215, 239, 241, 245, 250
Subject Index A Agammaglobulinemia, deficiency of C l q in, 285-286 Allotypy, rabbit immunoglobulins and,
304-305 Antibody autoimmune response, 142-151 production in uterus, 21 theories of diversity evolutionary mechanisms for multiple gene systems, 331435 evolution of immunoglobulin variable genes, 323-329 germ line and, 342-344 immunoglobulin patterns and, 339-
342
delayed hypersensitivity and cellular immune responses, 124-128 genetic factors in experimental thyroiditis, 121-124 human thyroid gland, 132-133 autoantigens, 135-138 autoimmunogenicity of thyroglobulin, 141-142 genetic factors in, 138-140 serological overlap with other diseases, 140-141 thyroiditis and other thyroid diseases, 133-135 Autoimmune response, distinctive types of antibody, 142-151 Autoimmunity, mechanisms, 151-155
mechanisms for information storage,
0
314-315 other multiple gene systems, 329-
Blood cells, maternal-fetal exchange, 51-
52
33 1 problems of control, 3 3 5 3 3 9 variable-region patterns, 315-323 Antigens fetal, maternal exposure to, 53-57 “private,” tissue-specific of trophoblast,
Burkitt’s lymphoma cell mediated immunity delayed hypersensitivity in vivo,
221-223 mixed lymphocyte stimulation tests,
223
4648
humoral antibody studies complement fixation tests, 212-217 cytotoxic and growth inhibition tests, 210-212 imniunofluorescent tests, 188-210 immunoprecipitation, 217-221 implications, 232-243 one or several Epstein-Barr viruses,
renal, cross-reactivity with placental,
45-46 thyroid gland heteroinimunization and, 109-112 isoininiunization and autoimmunization, 112-114 tissue and organ specificity, 107-109 transplantation, maternal exposure to,
223-225
57-61 Antiplacental serum, heterologous, biological activity, 44-45 Autoantibodies, induction of, 114-121 Autoantigens, autoimmune disease of thyroid and, 135-138 Autoimmune disease thyroid gland additional animal models, 128-132 autoantibodies and disease in rabbit,
C Cell( s ) maternal-fetal exchange, 49-SO blood cells, 5 1 5 2 fetal exposure to maternal cells, 62-
114-121
66 malignant cells, 53 maternal exposure to fetal antigens,
53-57
374
SUBJECT INDEX
maternal exposure to fetal leukocyte and transplantation antigens, 5761 trophoblast cells, 50 Choriocarcinoma, 22-23 evidence of host resistance, 23-24 fetal-maternal isoantigenic compatibility and, 24-26 Complement C1 esterase inhibitor, characteristics of, 254-255 C l q deficiency, agammaglobulinemia and, 285-286 C2 deficiency in man, 259-261 C2 polymorphism in guinea pig, 285 C3 deficiency in man, 268-270 C3 hypercatabolism characterization of defect, 270-274 hypothesis on pathogenesis, 274-275 C3 polymorphism human, 262-268 rhesus monkey, 268 C4 deficiency in guinea pig, 258-259 C4 polymorphism in human, 284-285 C5 deficiency in mice applications of model, 277-281 characterization of defect, 275-277 C5 disfunction in man, 281 C6 deficiency in rabbits applications of model, 283-284 characterization of defect, 281-282 third component deficiency application of model, 263 characteristics of defect, 262-263 Complement fixation, Burkitt’s lymphoma, 212-217 Cytotoxic tests, Burkitt’s lymphoma, 21&212
D Delayed hypersensitivity Burkitt’s lymphoma and, 221-223 thyroid autoimmune disease and, 124128 Differentiation genetic mechanism, 310-311 level of joining, 308-310 one gene or two -+ one polypeptide chain, 3 0 5 3 0 9
375
Domain hypothesis, immunoglobulin evolution and, 313
E Eggs pretrophoblastic, transplantation immunity and, 3941 Epstein-Barr viruses, one or several, 223225
F Fertilization, somatic, 21-22 Fetuses antigenic immaturity, 29-31 exposure to maternal cells, 62-66 genetically alien, maternal reactivity and, 11-14 qua homografts, 26-27 antigenic immaturity of fetus, 29-31 immunological reactivity of mother, 31-32 physiological barrier between mother and fetus, 33-39 separation of maternal and fetal circulation, 27-29
G P-Glycoprotein glycine-rich, inherited structural polymorphism in human, 285 Graft-versus-host reaction, uterus and,
14-15 Growth inhibition tests, Burkitt’s lymphoma, 210-212 Guinea pig C2 polymorphism in, 285 C4 deficiency in, 258-259 deficiency of “third component of complement” applications of model, 263 characteristics of defect, 262-263
H Hereditary angioneurotic edema characteristics of GI inhibitor, 254255 clinical aspects, 253-254 genetics, 255 pathogenesis, 255-258
376
SUBJECT INDEX
Herpes virus oncogenic Luck6 agent, 231-232 Marek’s disease, 225-231 Histocompatibility gene polymorphisnis, maternal-fetal interactions and, 70-75 Histoincompatibility, placental size and extent of trophoblastic invasion, 41-
43 Hormones, role in implantation, 6 Human C2 deficiency in, 259-261 C3 deficiency in, 268-270 C3 polymorphism in, 263-268 C4 polymorphism in, 284-285 C5 disfunction in, 281 inherited structural polymorphism of glycine-rich P-glycoprotein, 285
I Immune response cellular, thyroid autoimmune disease and, 124-128 Immune system, model for differentiation, 291-296 Immunofluorescent tests, Burkitt’s lymphoma, 188-210 Immunoglobulins evolution of variable and constant genes domain hypothesis, 313 homology units and gene duplication, 311-313 membrane molecules and, 313-314 families, 300 general structure, 298-299 one cell+one antibody, 301 patterns, theories of antibody and,
1 Leukocyte( s ) , antigens, maternal exposure to, 57-61 Luck6 agent, immunological studies,
231-232
M Malignant cells, maternal-fetal exchange,
53 Marek‘s disease, immunological studies,
225-231 Membrane( s ) molecules, evolution of immunoglobulins from, 313314 Mixed lymphocyte stimulation tests, Burkitt’s lymphoma, 223 Mouse complement C5 deficiency applications of model, 277-281 characterization of defect, 275-277
P Parturition, immunological mediation,
48-49 Placenta immunological competence, 67-69 organ specific antigens, 43-44, 49 activity of heterologous antiplacental serum, 44-45 cross-reactivity with renal antigens,
45-46 parturition and, 48-49 trophoblast “private,” tissue-specific antigen, 4 6 4 8 size, histocompatibility and, 41-43 Protein( s ) thyroid gland analysis in tissue extracts, 93-99 fractionation, 99-107
339-342 rabbit allotypy, 304-305 systems, 296-298 variable and constant regions, 299-300 variable genes, evolution of, 323-329 variable-region subgroups, 301-304 Immunoprecipitation, Burkitt’s lymphoma, 217-221 Iodine storage, thyroid gland, 88-90
R Rabbit allotypy, immunoglobulins and, 304-
305 autoimmune disease of thyroid in,
114-121 C6 deficiency applications of model, 283-284 characterization of defect, 281-282
377
SUBJECT INDEX
Reproduction biology, essentials of, 4-6 role of hormones in implantation, 6 Rhesus monkey, C3 polymorphism in, 268
S Spermatozoa autoantigenic status, 15-18 isoantigenic status, 18-20
T Thyroglobulin antigenic structure, 166-170 autoimmunogenicity, 141-142 physicochemical and biochemical characters, 155-159 subunit structure, 159-166 Thyroid gland antigens heteroimmunization and, 109-112 isoimmunization and autoimmunization, 112-114 tissue and organ specificity, 107-109 disorders, survey of, 90-93 experimental autoimmune disease additional animal models, 128-132 autoantibodies and disease in rabbit, 114-121 delayed bypersensitivity and cellular immune responses, 124-128 genetic factors in experimental thyroiditis, 121-124 histological structure and iodine storage, 87-90 human autoimmune disease, 132-133 autoantigens, 135-138 autoimmunogenicity of thyroglobulin, 141-142 genetic factors in, 138-140
B
c a D E F 6
9 O 1 2
H 3 1 4 1 5
serological overlap with other diseases, 140-141 thyroiditis and other thyroid diseases, 133-138 proteins analysis in tissue extracts, 93-99 fractionation, 99-107 Thyroiditis experimental, genetic factors in, 121124 human autoimmune disease, 133-138 genetic factors, 138-140 Transplantation disease, natural occurrence, 66-67 Trophoblast cells, exchange of, 50 “private,” tissue-specific antigens, 4648
U Uterus genetically compatible grafts, endocrinological determinants of fate, 6-8 graft-versus-host reactions, 14-15 as immunologically privileged site, 911 local antibody production in, 21 nonimmunological interaction between grafts or conceptuses and implantation sites, 8-9 reactivity, genetically alien fetuses and inocula of foreign cells, 11-14 sensitized, reactivity to antigenic exposure, 21
V Viviparity, concept of immunological inertia, 69-70
This Page Intentionally Left Blank
