[1] Idiotypic networks and nature of molecular mimicry: An overview

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

3

[1] I d i o t y p i c N e t w o r k s a n d N a t u r e o f M o l e c u l a r M i m i c r y : An Overview

By

H E I N Z K O H L E R , SRINIVAS KAVERI, THOMAS KIEBER-EMMONS, W . JOHN W . MORROW, SYBILLE M U L L E R , a n d SYAMAL RAYCHAUDHURI

Introduction In this first chapter we briefly recapitulate the history of the network hypothesis and proceed to discuss the major biological systems of idiotypic mimicry; we then attempt to incorporate recent immunochemical, structural, and biological evidence into a synopsis. This entails a synthesis of available data to propose conceptual adjustments of the network theory. These proposed changes are not semantic but address essential features of the network concept. Historical Perspectives of Network Concept The network hypothesis acted like a watershed in the area of immunology. It signaled the end of a period dominated by the ideas of Paul Ehrlich and McFarlane Burnett who saw the main objective of the immune system as the separation of self from nonself. This emphasis on self-recognition led to fundamental experimental research on autoimmunity and firmly established this aspect of immunology. It also stimulated interest in transplantation immunity and led to the first realization that tumor immunity might exist and be potentially beneficial for the cancer patient. The network hypothesis, however, appeared as something completely novel and without any experimental and conceptual roots. One could argue that it was created as an intellectually pleasing gedankenspiel (mind game) to deal with a growing number of paradoxical and unexpected findings. The satisfying aspect of the network theory was the introduction of logical connections between data that were otherwise difficult to explain. Because no precursors for this theory existed in the writings of eminent immunologists prior to the time the theory was presented, its proposers deserve full recognition as true pioneers.

Pre-Jerne Period The pre-network era (pre-Jerneian) is characterized by the experimental exploration of idiotypy. Before Oudin and Kunke! had made antiMETHODS IN ENZYMOLOGY,VOL. 178

Copyright© 1989by AcademicPress, Inc. All rightsof reproductionin any formreserved.

4

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

idiotypic reagents in 1963, the notion of anti-antibodies was not unfamiliar to immunologists.~ However, what was missing in the anti-antibody concept of this period was the idea that such antibodies would react with structures involved in the specificity of antibodies. Even the notion of " a u t o " anti-antibodies had surfaced without implying a biological role for such autoantibodies. The first notion that such autoanti-antibodies might play a biological role in the immune response was conceived by Lewis and colleagues. 2 The importance of Oudin's 3 and Kunkel's contribution 4 lies in the recognition that anti-antibodies would react with very unique "ailotypes" or "individual specific determinants" on myeloma proteins. Eventually Oudin's term "idiotype" prevailed and is used now to describe the unique antigenicity of antibodies. The potential role of anti-idiotypic antibodies dawned after the demonstration that anti-idiotypes were powerful suppressing agents of antibodies which express idiotypic specificities. 5,6 Anti-idiotypic suppression exhibited the hallmark of idiotypic specificity seen in anti-idiotype binding experiments. In 1971 Lewis and colleagues 2 postulated a biological role for antiantibodies as regulators of the immune response. This remarkable idea of autoanti-antibodies did not enter the consciousness of contemporary immunologists. It was not until 1974 that the conceptual potential had ripened in the laboratory atmosphere in Chicago to take odd results seriously and to interpret the findings in terms of a regulatory autoanti-idiotypic response. 7,8 These findings were made at the time the network concept was introduced by Lindenmann 9 and Jerne. 1° As an extension on the biological role of autoanti-idiotypes, Strayer and K6hler ~1 found spontaneous autoanti-idiotypic anti-Tl5 antibodies in sera of newborn mice before the dominant T15 ÷ idiotype appeared, and they speculated that this autoanti-idiotype might stimulate the development of the T15 idiotype. i F. Milgrorn and S. Dubinski, Nature (London) 179, 1351 (1957). 2 M. G. Lewis, T. M. Philips, K. B. Cook, and J. Blake, Nature (London) 232, 52 (1971). 3 j. Oudin and M. Michel, C. R. Hebd. Seances Acad. Sci. 257, 805 (1963). 4 H. G. Kunkel, M. Mannik, and R. C. William, Science 140, 617 (1963). 5 H. Cosenza and H. K6hler, Proc. Natl. Acad. Sci. U.S.A. 69, 2701 (1972). 6 D. A. Hart, A. L. Wang, L. L. Pawlak, and A. Nisonoff, J. Exp. Med. 146, 520 (1972). 7 L. Kluskens and H. KOhler, Proc. Natl. Acad. Sci. U.S.A. 71, 5083 (1974). 8 T. J. McKearn, F. P. Stuart, and F. Fitch, J. Immunol. 113, 1876 (1974). 9 j. LindenInann, Ann. Immunol. (Paris) 124C, 171 (1973). 10 N. K. Jerne, Ann. lmmunol. (Paris) 125C, 373 (1974). 11 D. S. Strayer and H. K6hler, Cell. Immunol. 25, 294 (1976).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

5

This notion was recently confirmed by the elegant experiments of Vakil and Kearney. 12

Work by Lindenmann and Jerne Lindenmann's description of the network preceded Jerne's by 1 year. However, Lindenmann credited the idea to a letter and seminar by Jerne. Nevertheless, what Lindenmann describes in 1973 is a fully developed network hypothesis. The major difference between him and Jerne lies in the greater emphasis on the biological importance of network interaction given by Jerne and the better conceptual and structural grip in Lindenmann's writing. For example, Jerne describes the biology of the immune response before he enters the network discussion. In contrast, Lindenmann gives a detailed nomenclature of different functional and structural sites on antibodies right at the onset of his paper. His most important contribution is the clear and crisp definition of "homobody," which Jerne has called the "Internal Image." Lindenmann describes the possible structural relationship of homobody and antibody in great detail. He sees the interaction of homobody and antibody as a three-dimensional complementarity of shapes.

Network Hypothesis In 1973 Jan Lindenmann and in 1974 Niels Jerne proposed theories which describe the immune system as a network of interacting antibodies and lymphocytes. The original network hypothesis as defined by Lindenmann and Jerne and expanded by Jerne in 198113 assigns different values to idiotype-anti-idiotype interactions. The parameters which are used to define the different kinds of anti-idiotypes are the following: (1) the location of the target idiotope to which anti-idiotype binds in relation to the antigen-binding site; and (2) the number of interconnections to other idiotopes in the idiotypic repertoire. The location of the target idiotope can be mapped using the relevant antigen or hapten as inhibitor in the binding of anti-idiotope to idiotope. If no inhibition is observed, the target idiotope is assumed to be distant from the binding site. If antigen inhibits, the target idiotope was believed to be in or near the binding site. Antigennoninhibitable anti-idiotypes were called Ab2a, antigen-inhibitable were called Ab2/3. In 1984 Bona and Kfhler 14 proposed another kind of anti12 M. Vakil and J. F. Kearney, Eur. J. Immunol. 16, 1151 (1986). 13 N. K. Jerne, J. Roland, and P. A. Cazenave, EMBO J. 1, 243 (1982). 14 C. A. Bona and H. KOhler, in "Monoclonal and Anti-Idiotypic Antibodies: Probes for Receptor Structure and Function" (J. C. Venter, C. M. Fraser, and J. Linstrom, eds.), Vol. 4, p. 141. Liss, New York, 1984.

6

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

idiotype; those antigen-inhibitable anti-idiotypes, which are inhibited because of steric interference, were classified as Ab2y. With respect to the degree of network contacts of a given idiotope, socalled parallel sets were recognized by Jerne. Parallel idiotype-antiidiotype sets originate from nonbinding site-related idiotopes (Ab2a defined). They are the conceptual precursors of Bona's regulatory idiotopes. J5 Regulatory idiotopes are defined by the following: (1) being immunogenic, i.e., can induce autoanti-idiotypic responses; (2) being recurrent; (3) being unrelated to a given binding site specificity; (4) being derived from the same variable gene families; and (5) being able to control the expression of antibodies of different specificities. Post-Jerne Period

Jerne's and Lindenmann's papers stimulated a great deal of theoretical thinking by immunologists and nonimmunologists. For example, Hoffmann ~6 and Richter ~7 proposed a mathematical interpretation of the immune network, and Adam and Weiler ~8used a mathematical approach to explain the immunological evolution of self-nonself discrimination. On the experimental side, a small number of idiotype laboratories, dominated by European investigators in this country and abroad, performed key experiments. Retrospectively it must be admitted that some of the experimental work, in particular that which depended on rabbit anti-idiotype sera, was flawed by a conceptual overinterpretation of data. However, the data improved when monoclonal reagents were used. The next conceptual leap in the network evolution occurred in the early 1980s. Interestingly, one of the driving new ideas originated from a practical suggestion to use anti-idiotypes as vaccines. J9-2~The first notion that anti-antibodies might resemble structures of the epitope to which the first antibody binds was made in 1951 by Najjar. 22 He had raised an antibody in rabbits against an antitoxin antibody. This anti-antibody inhib-

~5 C. A. Bona, "Regulatory Idiotypes: Modem Concepts in Immunology," Vol. 2. Wiley, New York, 1987. 16 G. W. Hoffmann, Eur. J. Immunol. 5, 638 (1975). i7 p. Richter, Eur. J. lmmunol. 5, 350 (1975). ~s G. Adam and E. Weiler, in "The Generation of Antibody Diversity: A New Look" (A. J. Cunningham, ed.), p. 1. Academic Press, London, 1976. 19 K. Eichmann and K. Rajewsky, Fur. J. lmmunol. 5, 661 (1975). 20 A. Nisonoff and E. Lamoyi, Clin. lmmunol. Immunopathol. 21, 397 (1981). 2~ I. M. Roitt, D. K. Male, G. Guamotta, L. D. De Carvhalo, A. Cooke, F. C. Hay, P. Lydard, Y. Thanalava, and J. Ivanyi, Lancet 1, 10411 (1981). 22 V. A. Najjar, Fed. Proc., Fed. Am. Soc. Exp. Biol. 10, 227 (1951).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

7

ited the enzymatic activity of the toxin. Najjar concluded that the antiantibody mimics the active site of the toxin sufficiently to inhibit the site by competing with substrate. This notion stimulated an entirely new branch of idiotype research and gave this formerly esoteric field a sense of importance, urgency, and funding. The other contribution was again by Jerne. 13He paid attention to the different biological roles of anti-idiotypes which he divided into a and/3 types (Ab2a and Ab2/3). From then on the "Internal Image" became Ab2/3 and the anti-idiotype (per Oudin) was Ab2a. The underlying key experiment was performed almost l0 years earlier by Nisonoff, 23 who had found that hapten could effectively interfere with the union of idiotype and anti-idiotype. These hapten-modifiable anti-idiotypes were now believed to resemble or mimic the shapes of external antigens. Through the 1970s and early 1980s T cells were always included in the network hypothesis. In particular, Eichmann and co-workers z4 showed data of idiotypic cross-reactions between idiotypes expressed by T ceils and antibodies. In 1982, however, evidence surfaced that T cells may recognize idiotypes differently than B cells. 25-:7 This finding does impose conceptual problems with the mechanism of idiotypic cross-reactivity in B and T cell responses. Biological Mimicry Systems: Anti-idiotypic Antibodies as Vaccines in Infectious Diseases The idea of anti-idiotypes as vaccines against infectious diseases has been derived from the successful preparation and characterization of antiidiotypic antibodies able to mimic bacterial and viral antigens. For example, McNamara et a l ) 8 showed in an assay measuring protection against Streptococcus pneumoniae infection by LDs0 determination in BALB/c mice that an anti-idiotypic monoclonal antibody (4C 1l) provided the same protection as the nominal antigen derived from bacteria, i.e., phosphorylcholine (PC). Other examples of vaccinelike antigen-mimicking antibodies being able to provide protection against bacterial, viral, and parasitological infectious diseases are described in the following sections.

23 B. W. Brient and A. Nisonoff, J. Exp. Med. 132, 951 (1970). 24 K. Eichmann, I. Falk, and K. Rajewsky, Eur. J. lmmunol. 8, 853 (1978). 25 K. Gleason and H. K6hler, J. Exp. Med. 156, 539 (1982). 26 p. M. Allen, G. R. Matsueda, E. Haber, and E. R. Unanue, J. Immunol. 135, 368 (1985). 27 R. Shimonkevitz, S. Colon, J. W. Kappler, P. Marrack, and H. M. Grey, Immunology 133, 2067 (1984). 2s M. K. McNamara, R. E. Ward, and H. Kfhler, Science 226, 1325 (1984).

8

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

Viral Network Epitopes

Kennedy et al. 29 have shown that anti-idiotypic antibodies against human antibodies specific for hepatitis B surface antigen (HBsAg) were protecting against hepatitis B virus infection in which the virus was bearing group-common antigenic determinants. Mice injected twice with antiidiotypic antibodies without subsequent virus challenge produced antibodies specific for HBsAg. Thus, anti-idiotypes can mimic the viral antigen and are thereby able to expand the precursors of anti-HB antibody-producing cells. Further evidence for the occurrence of an "internal image" or network epitope in the anti-idiotype was gathered from the earlier findings of an interspecies idiotype cross-reaction associated with antibodies to HBsAg, 3° and of antibodies in mice directed to HBsAg which serologically resembled the human idiotypic antibody to HBsAg produced by injecting anti-idiotype alone. 31,32Anti-idiotypic antisera generated to purified human anti-HBs- antibodies isolated from two different individuals recognized a common idiotype) 3 The use of anti-idiotypic antibodies mimicking the surface antigen of hepatitis B virus as a vaccine for a human disease has been demonstrated for hepatitis B virus in chimpanzees, being the relevant animal model for human infections) 4 Human antibodies to HBsAg served as the idiotype or first antibody (Abl) that was injected into rabbits to produce the antiidiotype or second antibody (Ab2). The rabbit anti-idiotype, or Ab2, was tested for its vaccine potential. Two chimpanzees were first immunized with the anti-idiotype and then challenged with infectious hepatitis B virus. Two control animals were either untreated or received a nonimmune rabbit IgG preparation. Both control chimpanzees developed clinical and serological signs of an active hepatitis B virus infection, whereas the two anti-idiotype-treated chimpanzees were protected from infection. Monoclonal mouse anti-idiotypic antibodies (Ab2) which mimic HBsAg have also been generated by Thanavala and Roitt (1986). 35 Polyclonal anti-HBsAg (Abl) sera were raised in four species of mammals. It 29 R. C. Kennedy, G. R. Dreesman, J. S. Butel, and R. E. Lanford, J. Exp. Med. 161, 1432 (1985). 30 R. C. Kennedy, I. Ionescu-Matin, Y. Sandoz, and G. Dreesman, Eur. J. Immunol. 13, 232 (1983). 31 R. C. Kennedy, K. Adler-Storthz,R. D. Henkel, Y. Sandoz, J. L. Melnick, and G. R. Dreesman, Science 221, 853 (1983). 32 R. C. Kennedy and G. R. Dreesman, J. Exp. Med. 159, 655 (1984). 33 R. C. Kennedy and G. R. Dreesman, J. Immunol. 1311,385 (1983). 34 R. C. Kennedy, J. W. Eichberg, R. E. Lanford, and G. R. Dreesman, Science 232, 220 (1986). 35 y . M. Thanavala and I. M. Roitt, Int. Reo. Immunol. 1, 27 (1986).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

9

is less likely that idiotypes unrelated to antigen binding would be identical in different species. These hyperimmune anti-HBsAg sera (Abl) were tested for possible binding to several hepatitis hybridomas using an indirect immunofluorescence technique. Two of six anti-idiotype-producing hybridomas showed immunofluorescence with hyperimmune anti-HBsAg sera from one goat, two rabbits, two swine, and five human subjects. Anti-idiotypic antibodies able to induce immunity against other viral agents have been characterized for the hemagglutinin of reoviruses, 36 Sendai v i r u s , 37 rabies virus glycoprotein, 38poliovirus type 11, 39 and Venezuelan equine encephalomyelitis virus. 4° This shows a wide range of application potential of antigen-mimicking anti-idiotypic antibodies in infectious diseases. It is unlikely that idiotype vaccines can fully substitute for conventional vaccines such as attenuated viruses, because of the ability of viral antigens to elicit both humoral and cellular immunity. ~5,41 However, idiotype vaccines may be more appropriate in infectious diseases caused by agents exhibiting a high degree of antigenic variation [e.g., influenza virus, trypanosoma, and human immunodeficiency virus (HIV)]. For example, Reale e t al. 42 have described antibodies which are specific for sequentially induced influenza virus variants, share idiotypes, and are encoded by similar VK and VFI genes. Therefore, the authors suggest the possibility that anti-idiotypic antibodies mimicking influenza hemagglutinin would protect not only against parental virus infection but also against infection by viral antigenic variants. AIDS (acquired immune deficiency syndrome) is a devastating disease resulting from infection by HIV of many cellular components vital for the maintenance of human homeostasis. The T4 antigen, or CD4 molecule, is the receptor for H I V . 43 AIDS can also be considered as an autoimmune disease, in which HIV mimics the "self" component and is able to bind to CD4-expressing helper-inducer T lymphocytes, thereby creating a cas-

J. R. Nepom, H. L. Weiner, M. A. Dichter, M. Tandieu, D. R. Spriggs, C. F. Gramm, L. M. Powers, B. N. Fields, and M. I. Greene, J. Exp. Med. 155, 155 (1982). 37 H. C. Ertl and R. W. Finberg, Proc. Natl. Acad. Sci. U.S.A. 81, 2850 (1984). 38 K. J. Regan, W. H. Wunner, T. Wiktor, and H. Koprowski, J. Virol. 48, 660 (1983). 39 F. G. C. M. UytdeHaag and A. D. M. E. Osterhaus, J. Immunol. 134, 1225 (1985). 4o j. T. Roehring, A. R. Hunt, and J. H. Mathews, High Technol. Route Virus Vaccines Abstracts, p. 32 (1984). 41 C. Bona and T. Moran, Ann. Immunol. (Paris) 136C, 29 (1985). 42 M. A. Reale, A. J. Mannheimer, T. H. Moran, G. Norton, C. A. Bona, and J. L. Soulman, J. Irnmunol. 137, 1352 (1986). 43 D. Klatzmann, E. Champagne, S. Chamaret, J. Gruest, D. Guetard, T. Hercend, J.-C. Gluckman, and L. Montagnier, Nature (London) 312, 767 (1984).

lO

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

cade of network errors leading to the destruction of the immune system. The gpl20 envelope protein of HIV is known to possess the CD4 binding domain by which HIV type I attaches to its target cell. 44,45 Problems in designing a vaccine and/or immunotherapy in AIDS arise from the primary target of HIV infection, i.e., T4 (CD4) receptor-bearing cells. Side effects may also be induced by a noninfectious HIV subunit or a CD4mimicking agent introduced as an HIV vaccine. CD4-expressing cells fulfill regulatory functions in every immune response, e.g., in expression of immunological memory by a vaccine or modulation of an immune response by immunotherapy. Any anti-HIV-targeted vaccine, based either on inactivated whole virus, viral recombinant subunit vaccine, or anti-idiotypic antibodies, must circumvent side effects such as immunosuppression or the induction of autoimmune responses. Approaches of anti-idiotype vaccines include anti-idiotypic antibodies related to viral epitopes and C D 4 . 46-48 Since the CD4 binding site appears to be conserved among a variety of different HIV isolates, the induction of anti-idiotypic antibodies bearing the internal image of the CD4 receptor appears to be a promising approach toward an idiotype vaccine in AIDS. It has been shown that certain monoclonal antibodies, i.e., anti-Leu3a and anti-OKT4a antibodies, block HIV infection of CD4-expressing cells and therefore recognize the binding epitope of HIV for the CD4 molecule .49 Dalgleish et al. 5° raised anti-idiotypes in mice immunized with antiLeu3a antibodies. The mouse sera containing polyclonal anti-idiotypes specifically recognized the envelope glycoproteins of HIV and the specific HIV binding site to CD4. These anti-idiotypic antibodies neutralized in vitro viral activity in three diverse isolates of HIV-1 and a single isolate of HIV-2. These results are important, considering that a vaccine must stimulate the production of a wide range of virus-neutralizing antibodies. 44 j. S. McDougal, M. S. Kennedy, J. M. Sligh, S. D. Cort, A. Mawle, and J. K. A. Nichols, Science 231, 382 (1986). 45 L. A. Lasky, G. Nakamura, J. H. Smith, C. Tennie, L. Shimosaki, E. Patzer, P. Berman, T. Gregory, and D. J. Capon, Cell (Cambridge, Mass.) 50, 975 (1987). E.-M. Zhou, T. C. Chanh, G. R. Dreesman, P. Kanda, and R. C. Kennedy, J. lmmunol. 139, 2950 (1987). 47 D. Zagury, J. Bernard, R. Cheynier, I. Desportes, R. Leonard, M. Fouchard, B. Reveil, D. Ittele, Z. Lurhuma, K. Mbayo, J. Wane, J.-J. Salaun, B. Goussard, L. Dechazal, A. Burny, P. Nara, and R. C. Gallo, Nature (London) 332, 728 (1988). 4s j. Homsy, K. Steimer, and R. Kaslow, Immunol. Today 8, 193 (1987). 49 Q. Sattentau, A. G. Dalgleish, R. A. Weiss, and P. C. L. Beverly, Science 234, 1120 (1986). A. G. Dalgleish, T. C. Chanh, B. J. Thomson, M. Malkovsky, and R. C. Kennedy, Lancet 2, 1047 (1987).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

1|

Chanh et al. 51 generated a monoclonal mouse anti-idiotypic antibody, HF1.7, and claimed that it mimics the CD4 receptor and binds HIV envelope antigens. As pointed out earlier, immunization of humans with anti-CD4 antibody preparations can be potentially harmful) 2,53 Anti-CD4 antibodies can block those epitopes on T cells which demonstrate helper activity) 4 However, in man one would immunize with low quantities of anti-CD4, such that an anti-idiotypic response is actively induced avoiding the immunosuppressive effects demonstrated in mice in which several milligrams of L3T4 antibody per kilogram body weight was used for passive immunotherapy)° Immunization of newly positive HIV patients with an intact immune system may induce an anti-idiotypic response to the antiCD4, which will represent a new neutralizing set of antibodies. Alternatively, in those patients with AIDS, anergic to neoantigens, and not responding to anti-CD4 antibody immunization, an anti-idiotype that mimics CD4 could be used for passive immunotherapy. Passive immunotherapy with such an antibody may be considered as a "CD4 sponge" which would bind all the HIV that is present) °,51 Thus, the idiotype mimicry of biological receptors is the underlying principle in using CD4 for designing an idiotype vaccine in AIDS. However, to date no evidence exists that CD4-mimicking anti-idiotypes stop the progression of AIDS in vivo. It has been pointed out by Dalgleish et al. 5° that the production of an anti-idiotype binding to and neutralizing HIV is dependent on the immunization schedule of animals with antiLeu3a, the use of a protein carrier such as keyhole limpet hemocyanin (KLH), or possibly the choice of adjuvant. Only little, if any, neutralizing activity against different HIV isolates could be found with antibodies (Abl) raised against HIV envelope glycoproteins. 55 However, there are recent observations that human monoclonal antibodies (Abl) against HIV antigens, particularly neutralizing antibodies against HIV glycoproteins, e.g., gp41, posses virus-neutralizing activities) 6 Efforts are underway to produce human monoclonal antibodies (Abl) against HIV gp 120 and gp 41 to be used not only for immu5~ T. C. Chanh, G. R. Dreesman, and R. C. Kennedy, Proc. Natl. Acad. Sci. U.S.A, 84, 3891 (1987). 52 p. del Guercio and M. Zanetti, lmmunol. Today 8, 204 (1987). 5s W. C. Koff and D. F. Hoth, Science 241, 426 (1988). 54 R; J. Benjamin and H. Waldman, Nature (London) 320, 449 (1986). 55 R. A. Weiss, P. R. Clapham, J. Weber, A. G. Dalgleish, P. Berman, and L. A. Lasky, Nature (London) 346, 572 (1986). 56 C. Degranges, J. Bernard, S. Sprecher, V. Boyer, S. Souche, B. Reveil, A. Bunny, and D. Zagury, Program Abstr., Int. Conf. AIDS, 4th Book 2, p. 375 (1988).

12

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

notherapy or vaccination, as shown by Gorny et al. 57 and Teeuwsen et a l . : 8 but also as antigen to generate anti-idiotypes related to HIV epitopes. Anti-idiotypes mimicking HIV epitopes could be used to screen sera from AIDS patients for idiotype-carrying anti-HIV antibodies as idiotypic markers in the immune response of patients to HIV. 59 It has been shown earlier that polyclonal rabbit anti-idiotypic antisera raised against a monoclonal antibody specific for a p25 gag region epitope detects a common interspecies idiotype associated with anti-HIV responses .60 Bacterial Network Epitopes

In several bacterial systems anti-idiotypic antibodies mimicking an essential bacterial antigen induce resistance in animals against lethal infectious diseases after challenge with the live bacteria. Stein and Soderstrom 61 demonstrated that an injection of a set of anti-idiotypes at birth, followed by immunization with polysaccharide vaccine, can protect mice against 10 LDs0 of Escherichia coli K13. Immune responses to polysaccharides have several peculiar characteristics, including delayed ontogenic development, isotype and idiotype restriction, and failure to generate memory cells. 62 Therefore, the results of Stein and Soderstrom 61 appear interesting, because their procedure circumvents the ontogenic delay of polysaccharide-specific clones by injection of anti-idiotype antibodies at birth. The anti-idiotype was directed against an IgM-Ab 1 specific for the capsular polysaccharide of E. coli K13. Antibodies produced in mice after immunization with S. pneumoniae are directed against phosphorylcholine (PC), which is the immunodominant determinant of the bacterial cell wall, although it is not a polysaccharide. Anti-PC antibodies in BALB/c mice express a dominant idiotype, TEPC 15 or T 15.63-65Trenkner and Riblet 66have shown that anti57 M. K. Gorny, V. Gianakakos, and S. Zolla-Pazner, Program Abstr., Int. Conf. AIDS, 4th Book 2, p. 78 (1988). 5a V. J. P. Teeuwsen, C. J. Siebelink, M. J. Stukcert, A. G. C. M. UytdeHaag, and A. D. M. E. Osterhaus, Program Abstr., Int. Conf. AIDS, 4th Book 2, p. 71 (1988). 59 S. Miiller, W. J. W. Morrow, L. Wing, and H. Kfhler, unpublished data (1988). 6o W. J. W. Morrow, I. Gaston, T. Anderson, K. Steimer, and M. S. McGrath, J. Virol (submitted for publication). 61 K. Stein and J. Soderstrom, J. Exp. Med. 160, 1001 (1984). 62 K. E. Stein, Curr. Top. Microbiol. lmmunol. 57 (1985). 63 W. Lee, H. Cosenza, and H. K6hler, Nature (London) 247, 55 (1974). 64 j. Quintans, M. R. Cohen, Z.-S. Quan, R. F. Dick, and B. Regueiro, Eur. J. lmmunol. 11, 236 (1981). 65 M. K. Wittner, M. A. Bach, and H. K6hler, J. Immunol. 128, 595 (1982). E. Trenkner and R. Riblet, J. Exp. Med. 142, 1121 (1975).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

13

T15 antibodies were able to induce in vitro an anti-PC plaque-forming cell (PFC) response in the absence of the antigen. McNamara et al. 28 have induced effective immunity and protection against a lethal dose of S. p n e u m o n i a e in mice using the monoclonal anti-idiotypic antibody 4C1 1. Interestingly, the monoclonal antibody in its unaltered form did not induce anti-streptococcal immunity nor any significant increase in the effective antibody idiotype. The internal idiotope antigen had to be coupled first to a large protein carrier (KLH) before it could induce protection. As pointed out by KOhler et al. ,67 this finding might indicate a general rule for internal images or network antigens which would predict that network antigens per se have poor immunogenicity. Strong network antigens would disturb the equilibrium between stimulation and suppression in the immune network. Parasites

Parasites cause chronic disease in man and other mammals. The life cycle of parasites involves a number of complex antigen systems and poses a great difficulty in identifying a universal antigen system which can be used to stimulate the immune system against the parasites. Furthermore, a continuous antigenic drift of parasite-associated surface antigens makes the use of these antigens more unsuitable for vaccine purposes. During cyclic infection, antibodies against variants are induced which can result in neutralization of homologous variants but not against the original variant. However, monoclonal anti-idiotype antibodies against protective antibodies generated against variant antigens can induce strong protective immunity against the original variant. Sacks et al. 68 first described the potential therapeutic use of anti-idiotype antibodies against antibodies to variant antigens present on Trypanosoma rhodesiense. Because proteinaceous antigens are poor protective immunogens, much attention has been given to the exploration of the role of carbohydrate in inducing tumor immunity against parasites. Surface carbohydrate antigens could be very effective in inducing protective immunity because they are not subject to antigenic drifts. However, carbohydrates per se are poor immunogens. Furthermore, DNA recombinant technology cannot be used for these antigens; anti-idiotype antibodies mimicking the carbohydrate epitope are effective in inducing a protective antibody response against carbohydrate in the E. coli and Trypanosoma cruzi systems.61,68 ~7 H. K6hler, S. Miiller, and C. Bona, Proc, Soc. Exp. Biol. Med. 178, 189 (1985). 6s D. L. Sacks, K. M. Esser, and A. Shcr, J. Exp. Med. 155, 1108 (1982).

14

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

The role of the idiotype-anti-idiotype cascade in parasitic diseases has been explored in several laboratories. The study of functional idiotypeanti-idiotype networks has produced interesting insights into the interaction of the networks and the parasite. Old and Kresina 69 studied network interactions in mice infected with Schistosoma japonicum. They found a cyclic appearance of Ab2 against the major cross-reactive idiotypes that are also present on antiegg antibody. Grzych et al. 7° studied the potential of an anti-idiotype vaccine against experimental schistosomiasis. They made anti-idiotype antibodies to a cytotoxic rat monoclonal antibody, IPLSmi, which binds to a 38K glycoprotein present on the schistosomulure target membrane. The glycanic nature of the target antigen excluded the possibility of using recombinant DNA technology for a nominal antigen for vaccine development. Immunization with monoclonal antiidiotype antibodies induced strong protective immunity against a challenge with the live infection. The Ab2-immunized mice showed a marked increase in serum Ab3 levels which was cytotoxic against parasites in the presence of eosinophils. Powell and Colley 71 demonstrated the presence of Ab2 PFC in mice exposed to human helminth cercariae. Mice challenged with such a dose developed a chronic Schistosoma mansoni infection and a cell-mediated granulomatous response to schistosome eggs. In a modified PFC assay, looking at anti-idiotype PFC expression recognizing idiotypes associated with antiegg antigens, a response was observed 2-3 weeks after the onset of egg production and continued throughout the infectious stage. This study emphasizes the existence of an idiotype-anti-idiotype network in parasitic infections. Such a network can be regulated so that beneficial protective immunity can be induced. Phillips et al. 72 studied a functional idiotype-anti-idiotype network that is germane to regulation in schistosomiasis. They studied interaction of antigen, idiotypes, and anti-idiotypes in the immune system and established correlations with resistance to parasitemia. What is interesting is that different monoclonal antibodies with the same epitopic specificity behave differently in terms of rendering protective immunity. These monoclonal Abls share many characteristics in an in vitro assay system, and antibody class or subclass cannot explain the in vivo effect. They also demonstrated a paratope-associated anti-idiotype against the protective 69 R. G. Old and T. F. Kresina, J. Clin. Invest. 76, 2338 (1985). 70 j. M. Grzych, M. Capron, P. H. Lambert, C. Dissous, S. Torres, and A. Capron, Nature (London) 316, 74 (1985). 7~ M. R. Powell and D. G. Colley, J. lmmunol. 134, 4140 (1985). 72 S. M. Phillips, E. G. Fox, N. G. Fathelbab, and D. Walker, J. lmmunol. 137, 2339 (1986).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

15

Abl in sera from actively infected mice. Therefore, it was concluded that idiotypic or functional components of antibody structure play a role in inducing resistance. Furthermore, the authors emphasized that epitopic specificity is inadequate in determining protective immunity in schistosomiasis and that an idiotypically directed regulatory network may be important in determining the fate of the immune response. Gorczynski 73 examined the protective immunity induction ability of anti-idiotype antibodies to Leishmania-specific B or T cells. In highly susceptible BALB/c mice they showed that immunization with anti-T cell receptor antibodies lead to exacerbation of Leishmania growth. They also demonstrated idiotypic (T cell idiotypes) cross-reactivity on T cells reactive or nonreactive to parasite antigens. These data demonstrate the existence of an idiotype-anti-idiotype regulatory network borne by T cells. It implies that anti-idiotype antibodies can be used to modulate the immune response to parasites.

Tumor Antigens The development of cancer treatment modalities which take advantage of the immune system has been the object of numerous investigations. 74~3 The efficacy of immunotherapy in human cancers has demonstrated encouraging results in many studies, 79,81~3 but, in general, tumor immunotherapy has not been examined enough to assess clinical useful-

73 R. M. Gorczynski, J. Immunol. 139, 3070 (1987). 74 p. O. Livingston, E. J. Natoli, M. J. Calves, E. Stockert, H. F. Oettgen, and L. J. Old, Proc. Natl. Acad. Sci. U.S.A. 84, 2911 (1987). 75 R. Lathe, M. P. Kieny, P. Gerlinger, P, Clertant, I. Guizan, F. Cuzin, and P. Chambon, Nature (London) 326, 878 (1987). 76 M. K. Wallack, K. McNally, M. Michaelides, J. Bash, A. Bartolucci, H. Siegler, C. Balch, and H. Wanebo, Am. Surg. 52, 148 (1986). 77 G. G. Stevenson and F. K. Stevenson, Immunopathology6, 99 0983). 78 R. K. Oldham, K. A. Foon, A. C. Morgan, C. S. Woodhouse, R. W. Schroff, P. G. Abrams, M. Fer, C. S. Schoenberger, M. Farrell, E. Kimbell, and S. A. Sherwin, J. Clin. Oncol. 2, 1235 (1984). 79 A. N. Houghton, D. Mintzer, C. Cordon-Cardo, S. Welt, B. Fliegel, S. Vadhan, E. Carswell, M. R. Melamed, H. F. Oettgen, and L. J. Old, Proc. Natl. Acad. Sci. U.S.A. 82, 1242 (1985). 8o T. Tai, L. D. Cahan, T. Tsuchida, R. E. Saxton, R. F. Irie, and D. L. Morton, Int. J. Cancer 35, 607 (1985). 81 T. Tai, J. C. Paulson, L. D. Cahan, and R. F. Irie, Proc. Natl. Acad. Sci. U.S.A. 80, 5392 (1983). 82 T. C. Meeker, J. Lowder, D. G. Maloney, R. A. Miller, K. Thielamane, R. Warnke, and R. Levy, Blood 65, 1348 (1985). 83 R. A. Miller, D. G. Maloney, R. Warnke, and R. Levy, N. Engl. J. Med. 306, 517 (1982).

16

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[ll

ness. The active immunotherapy of cancer patients with tumor-derived material includes various risk factors; above all, tumor-associated antigens (TAA) are often found to be poorly immunogenic. Therefore, alternative approaches should be explored to enhance tumor immunogenicity. A common explanation for the absence of antitumor immunity is that the immune system has been tolerized by the tumor antigen. 84-87 If this were to be a correct conclusion, steps could be undertaken to break the existing antitumor tolerance. An effective method of breaking induced tolerance is to present the critical epitope in a different molecular environment to the tolerized host. 88 While this can be done easily with haptens and other small well-defined antigens, it is impossible for most tumor antigens because they are chemically ill-defined and difficult to purify. The network hypothesis of Niels Jerne offers an approach to transform epitope structures into idiotypic determinants expressed on the surface of antibodies. 1° According to this theory an internal idiotope image exists for each antigen, including self-antigen. Several such internal image antigens have been used as surrogate antigens to induce specific and protective immunity. 67,89-92For tumor immunotherapy using anti-idiotype (anti-Id) antibodies, experiments include the use of anti-Id antibodies which may or may not mimic TAA. 29,93,94 Various anti-idiotype approaches to manipulate the immune response against tumor growth are now briefly described. The utilization of the idiotype network in tumor immunotherapy opens new perspectives and may enrich the therapeutic armamentarium. Basically two approaches are feasible. The first takes advantage of the existence of internal antigen images in the idiotype repertoire. This approach has already been used successfully by several investigators, 28,93and it has W. H. McBride and S. E. M. Howie, Br. J. Cancer 53, 707 (1986). 85 M. I. Greene, Contemp. Top. 11, 81 (1980). H. D. Haubeck and E. Kolsch, Immunology 47, 503 (1982). s7 S. M. Howie and W. H. McBride, Eur. J. Immunol. 12, 671 (1982). W. O. Weigle, J. Exp. Med. 114, 111 (1961). 89 R. C. Kennedy, G. R. Dreesman, and H. Krhler, BioTechniques 3, 4040 (1985). J. A. Bluestone, O. Leo, S. L. Epstein, and D. H. Sachs, Immunol. Reo. 90, 1 (1986). 91 F. G. C. M. UytdeHaag, H. Bunschoten, K. Weijer, and A. D. M. E. Osterhaus, Immunol. Rev. 90, 93 (1986). 92 R. W. Finberg and H. C. J. Ertl, Immunol. Rev. 90, 129 (1986). 93 H. Koprowski, D. Herlyn, M. Lubeck, E. DeFreitas, and H. F. Sears, Proc. Natl. Acad. Sci. U.S.A. 81, 216 (1984). G. T. Nepom, K. A. Nelson, S. L. Holbeck, I. Hellstrrm, and K. E. HeUstrrm, Proc. Natl. Acad. Sci. U.S.A. 81, 2864 (1984).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

17

the advantage of not being genetically restricted. 95 Internal image idiotypes mimic the three-dimensional shapes of antigens and thus are effective across the species barrier. At the same time, antigen can be presented in a different molecular environment. The other method of using the idiotype network rests on the existence of so-called regulatory idiotypes, 25,96,97 which may also be linked to the regulatory network of anticancer responses. Discovering these linkages in antitumor responses would be important and could be the first step in using these regulatory idiotypes to control tumor growth by immunologic means. Although these ideas are at present speculative, they deserve to be explored. Forstrom e t al. 98 have shown in a mouse sarcoma system that what they believed to be anti-idiotype antibodies elicited a delayed-type hypersensitivity (DTH) reaction in a tumor-specific manner. They derived monoclonal antibody 4.72 from mice hyperimmunized with tumor cells. The monoclonal antibody 4.72 did not seem to recognize tumor cells, and therefore it was concluded that it must be an anti-idiotype antibody. What is interesting is that 4.72 induced tumor-specific DTH in an IgH-restricted fashion. Although this report lacks various immunochemical proofs of the anti-idiotypic nature of 4.72 and mechanistic aspects of 4.72-induced tumor immunity, it clearly shows the first demonstration of the potential role of anti-idiotype antibodies in tumor immunity. Evidence for a beneficial role of anti-idiotype antibodies in cancer patients came from a study by Koprowski et al. 93 Koprowski et al. demonstrated in a more indirect way that the presence of anti-idiotypic antibodies may be beneficial for tumor patients. Patients with metastatic or recurrent gastrointestinal adenocarcinoma were treated with the mouse monoclonal antitumor antibody 17-1A. The patients that developed antiidiotypic antibodies showed clinical improvement or long periods of remission from their disease. Koprowski et al. purified anti-idiotypic antibodies from patients who showed cross-reactivity and inhibited the binding of 17-1A to the tumor antigen. Therefore, it was concluded that the human immunoglobulin may bear an image of the tumor antigen, and the immune system may recognize this internal image differently from the nominal antigen which is beneficial to the patient. 95 H. Auchincloss, Jr., J. A. Bluestone, and D. H. Sachs, J. Exp. Med. 157, 1273 (1983). 96 j. Frelinger, A. Sign, A. Infante, and C. G. Fathman, lmmunol. Rev. 81, 22 (1984). 97 j. S. Kaye, J. Porcelli, J. Tite, B. Jones, and C. A. Janeway, J. Exp. Med. 158, 836 (1983). J. W. Forstrom, K. A. Nelson, G. T. Nepom, I. Hellstr6m, and K. E. Hellstr6m, Nature (London) 303, 627 (1983).

18

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

In another report, Nepom e t al. 94 had generated polyclonal anti-idiotypic antibodies against the anti-human melanoma antibody 8.2. The polyclonal antisera contained antibodies against binding site-related idiotopes. The mice could be sensitized against melanoma-associated antibodies by immunizing them with site-specific polyclonal rabbit anti-idiotypic antibodies. Kennedy e t al. 29 observed that immunization with anti-idiotype directed against the monoclonal antibody recognizing a distinct amino and carboxy terminal of T antigen could induce tumor-specific immunity against SV40-transformed cells. Although the authors could not find the presence of Ab3 in the sera of the anti-idiotype-immunized mice, they could not rule out the possibility of anti-idiotype-induced perturbation of the T cell regulatory network. Herlyn e t al. 99 demonstrated that polyclonal anti-idiotype antibodies against a monoclonal antibody to gastric carcinoma induced an Ab3 response in mice and rabbits. A monoclonal Ab3 generated from polyclonal anti-idiotype-immunized mice showed very similar reactivity to that of the original Abl; i.e., Ab3 bound to the same tumor antigen with similar avidity as Abl. Abl and Ab3 sequence determination will prove whether these similarities in their reactivity to the tumor antigen are due to their sequence homology throughout the chain or only at the complementaritydetermining region (CDR). The authors concluded that anti-idiotype antibodies bear an image of the tumor antigen, and therefore Ab2 may have potential for modulating the immune response against the tumor. The first detailed correlative study regarding the fine specificity of anti-idiotype antibodies and their biological effect was that of Raychaudhuri e t al. ~°°,~°~ In this study the tumor-specific immune response induced by irradiated tumor cells (L1210/GZL) and by anti-idiotype antibodies was analyzed. The anti-idiotype antibodies (Ab2) were made against the paratope of a monoclonal antitumor antibody (11C1) that recognizes a TAA which cross-reacts with the mouse mammary tumor virusencoded envelope glycoprotein gp52. Two Ab2s, 2F10 and 3A4, induced idiotypes expressed by the monoclonal antitumor antibodies l lC1 and 2B2. Cytotoxic T cells, generated by immunization with irradiated tumor cells, lyse 2F10 and 3A4 hybridoma cells. Furthermore, immunization with Ab2 induces tumor-specific cytotoxic T lymphocytes. The frequency 99 D. Herlyn, A. H. Ross, and H. Koprowski, Science 232, 100 (1986). S. Raychaudhuri, Y. Saeki, J.-J. Chen, H. Iribe, H. Fuji, and H. Krhler, J. lmmunol. 139, 271 (1987). 101 S. Raychaudhuri, Y. Saeki, and H. Krhler, J. Immunol. 139, 3902 (1987). too

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

19

of tumor-reactive cytotoxic T lymphocytes was found to be similar in mice immunized with Ab2 or with irradiated tumor cells when examined at the precursor level. However, only 2F10 induces protective immunity against the growth of L 1210/GZL tumor cells. The depletion of a L3T4 + T cell population from 2FI0 immune mice was found to increase the effectiveness of transferred T cells to induce inhibition of tumor growth. The inability of 3A4 to induce antitumor immunity could be correlated with the presence of a population of Lyt2 + regulatory T cells. Collectively, these results ~°°,l°l demonstrate the existence of a regulatory network controlling the expression of effective tumor immunity. Selection of binding site-related Ab2, however, may not be a sufficient criterion for the development of an idiotype vaccine. A better understanding of the regulatory interactions induced by anti-idiotypes is needed for the design of effective antitumor immunotherapy. Bhattacharya-Chatterjee et al. ~°2,1°3 generated anti-idiotypic monoclonal antibodies (Ab2), binding to a hybridoma SN2 (Abl), which recognizes a glycoprotein, gp37, expressed by human leukemic T cells. To characterize these anti-idiotopes further, they were used to immunize mice and rabbits. Several murine antianti-idiotype monoclonal antibodies (Ab3), mostly of IgMr isotype, were obtained. Monoclonal Ab3 and sera from rabbits immunized with Ab2 contained antibodies that bind to gp37 antigen and leukemic MOLT-4 and JM cells. Also, monoclonal Ab3 and immune sera from rabbits competed with Abl for binding of iodinated Abl to Ab2, indicating that Ab3 in mice and rabbits shares idiotopes with Abl (SN2). Furthermore, both the murine monoclonal Ab3 and rabbit polyclonal Ab3 immunoprecipitated the same gp37 antigen as SN2 (Abl). The production of antigen-specific Ab3 (Abl') in mice and rabbits in the absence of any exposure to gp37 indicates that these Ab2 may indeed carry the internal image of the gp37 antigen. Such anti-idiotopes (Ab2fl) may be useful as antigen substitute for the induction of therapeutic immunity in T cell leukemia patients. Molecular Mimicry and Autoimmunity The concept of molecular mimicry has important ramifications for autoimmunity, both in the etiopathogenesis of these disorders and in their immunoregulation. io2 M. Bhattacharya-Chatterjee, M. W. Pride, B. K. Seon, and H. K6hler, J. lmmunol. 139, 1354 (1987). xo3M. Bhattacharya-Chatterjee, S. K. Chatterjee, S. Vasile, B. K. Seon, and H. KOhler, J. Immunol. 141, 1398 (1988).

20

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

Origins of Disease. Autoimmune diseases are of unknown etiology, but genetic, hormonal, and environmental factors have all been postulated to play roles in triggering immune dysfunction. Although it would seem that the events leading to an autoimmune state are multifactorial, the notion that a microorganism can trigger such a disorder by immunologic cross-reactivity or molecular mimicry to the host has been long considered an attractive and feasible explanation (see Refs. 104-106 for review). Although this aspect of molecular mimicry is not exactly within the scope of this chapter, it cannot easily be dismissed, particularly as a corollary of this idea has been proposed by Plotz, ~°7 who suggested that autoimmune diseases may arise from anti-idiotypic reactions through antiviral antibodies; the anti-idiotypes then induce pathogenic sequelae in the host tissues. Indeed, this hypothesis has been extended to the case of HIV infection. Ziegler and Stites ~°8have hypothesized that an autoimmune reaction could result from antigenic similarities between the CD4-binding receptor site on the gpl20 envelope of the virus and HLA Class II molecules (HLA-DR is the natural ligand of the CD4 molecule as it interacts with the process of presenting antigen). Interestingly, del Guercio and Zanetti 52 have suggested that autoimmune reactions may result as a consequence of the formation of anti-idiotypic antibodies to the gpl20 receptor site and that the resulting anti-idiotypic antibodies have specificity for CD4-bearing cells. It should be emphasized, however, that despite the intriguing nature of these possibilities, the anti-idiotype-mediated pathogenesis of autoimmune disease remains to be proved experimentally. Irnmunoregulation with Anti-idiotypes. The regulation of immunological networks has been a goal of experimentalists for over a decade. In particular, the possibility of modulating autoimmune conditions, where network aberrations are evident, is a particularly attractive concept. This approach has been attempted for several autoimmune disease states, although in most the anti-idiotypic reagents used in these studies have not been characterized in terms of molecular mimicry, and thus the mechanisms by which they function remain to be elucidated. A discussion of these immunoregulatory experiments is beyond the scope of this review. However, diseases for which anti-idiotype antibodies have been used therapeutically and which appear to involve mechanisms that include molecular mimicry are listed in Table I. 1o4 S. H. Zwillich and P. E. Lipsky, Rheum. Dis. Clin. North Am; 13, 339 (1987). 1o5 T. Dryberg and M. B. A. Oldstone, Curr. Top. Microbiol. lmmunol. 130, 25 (1986). to6 M. B. A. Oldstone, Cell (Cambridge, Mass.) 50, 819 (1987). to7 p. H. Plotz, Lancet 2, 824 (1983). 1o8j. L. Ziegler and D. P. Stites, Clin. lmmunol, lmmunopathol. 41, 305 (1986).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

21

TABLE I DISORDERS IN WHICH ANTI-IDIOTYPE ANTIBODIES HAVE BEEN USED TO REGULATE AUTOANTIBODY PRODUCTION AND CLINICAL PROGRESSIONa

Systemic lupus erythematosus b,c,d Sj6gren's syndrome c Rheumatoid arthritis c Thyroiditis a Myasthenia gravis d Idiopathic thrombocytopenic purpura • Autoimmune neutropenia e a Further review of some of the experiments can be found in W. J. W. Morrow and D. A. Isenberg, "Autoimmune Rheumatic Disease," Blackwell Scientific Publications, Oxford, 1987. b In vitro experiments conducted in murine systems. c Experiments conducted on human cells in vitro. a In vivo

experiments conducted in rodent systems. • Human therapy.

Molecular Mimicry and Receptors Owing to the e n o r m o u s variability in the antibody population, it m a y be possible that a m o n g all the different antibodies a few antibodies which are p r o d u c e d m a y recognize a h o r m o n e in a fashion that mimics the way a physiological r e c e p t o r recognizes the hormone. I f this is the case, the combining sites of such " r e c e p t o r - l i k e " antibodies m a y display structural features which m a y be c o m m o n with the hormone-binding part of the receptor. C o n v e r s e l y , a minority of a second set of antibodies raised against the combining sites (the idiotopes) of the receptor-like antibodies m a y , in their variable regions, be similar to those structures of the horm o n e that bind to the physiological receptor. A m o n g such groups of antibodies, s o m e would p r o b a b l y interact with the h o r m o n e receptor in a hormonelike manner. This second set of antibodies, t e r m e d Ab2/3, would be the "internal i m a g e " of the original antigen. 1°'13 It is precisely this theoretical concept that p r o m p t e d Sege and Peterson, 1°9 w h o for the first time d e m o n s t r a t e d that anti-idiotypic antibodies to9 K. Sege and P. A. Peterson,

Proc. Natl. Acad. Sci. U.S.A.

75, 2443 (1978).

22

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

raised against antihormone (anti-insulin) antibodies were capable of interacting with the (insulin) receptors and evoking physiological responses. Ever since, substantial data have been accumulated in this field which explore the interactions among ligand, physiological receptors, Abl, and Ab2/3 at various levels. Analysis of shared amino acid sequences between antigen and anti-idiotypic antibody may reveal striking similarities. Computer-generated models to localize these determinants may reveal the regions potentially responsible for the molecular mimicry.~l° Similarities in electrostatic interactions could be responsible for some molecular mimicry, especially in the case of anti-idiotypes which mimic carbohydrate residues.IH. 112 Receptors for different endocrine hormones, neurotransmitters, neuropeptides, growth factors, etc. have been subjected to investigation using anti-idiotypic antibodies as probes. The concept of molecular mimicry has particularly drawn the attention of receptor biochemists, as these anti-idiotype molecules serve as useful tools in purification and characterization of receptors that are scantily expressed on the surface of cells. Interactions of idiotypic ligands with receptors for catecholamine hormones, 113-115thyroid-stimulating hormones, ~16acetylcholine, 117 insulin, t~8 lymphokines, 119 etc. have been considered as a means for structural and functional characterization of these receptors. Another interesting application of idiotypic mimicry of receptor-ligand interactions lies in drug screening and drug designing. The complementary nature of ligand-receptor and idiotype-anti-idiotype structures could be exploited to develop anti-idiotypic antibodies which mimic the structure and binding properties of cellular receptors and their ligands. These antibody receptor surrogates (Abls) could be employed to screen large numbers of candidate drugs for their potential to bind to cellular receptors and could further be tested for agonistic or antagonistic properties. Also, the structures deduced from molecular modeling of amino acid ,0 C. Bruck, M. S. Co, M. Slaoui, G. N. Gaulton, T. Smith, B. N. Fields, J. I. Mullins, and M. I. Greene, Proc. Natl. Acad. Sci. U.S.A. 83, 6578 (1986). Ill F. Erlanger, lmmunol. Today 6, 10 (1985). ,2 I. M. Roitt, Y. M. Thanavala, D. K. Male, and F. C. Hay, lmmunol. Today 6, 265 (1985). H3 A. B. Schreiber, P. O. Couraud, C. Andre, B. Vray, and A. D. Strosberg, Proc. Natl. Acad. Sci. U.S.A. 77, 7385 (1980). N4 G. J. Homcy, S. G. Rockson, and E. Haber, J. Clin. Invest. 69, 1147 (1982). ,5 j. G. Guillet, S. V. Kaveri, O. Durieu, C. Delavier, J. Hoebeke, and A. D. Strosberg, Proc. Natl. Acad. Sci. U.S.A. 82, 1781 (1985). ii6 N. R. Farid and T. C. Y. Lo, Endocr. Rev. 6, 1 (1985). H7 W. L. Cleveland, N. H. Wasserman, R. Sarangarayan, A. S. Penn, and B. F. Erlanger, Nature (London) 305, 56 (1983). 118 D. Elias, R. Maron, I. R. Cohen, and Y. Schecter, J. Biol. Chem. 259, 6416 (1984). N9 p. L. Osheroff, T. R. Chiang, and D. Manousos, J. Immunol. 135, 306 (1985).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

23

sequences of ligand and receptor "internal image" antibodies may be used as a starting point for peptide drug design and synthesis. In order to analyze the relationship between the immune system and the hormone-receptor system, the first and critical step is the selection of an appropriate antihormone antibody which can be used to obtain antiidiotypic antibodies. It appears that one prerequisite for the induction of Ab2fl anti-idiotypic antibodies with reactivity toward the receptor to a ligand is that antiligand antibodies display binding properties similar to those observed for the receptor, which indeed has been shown in many hormonal systems. 12°-122 Monoclonal anti-idiotypic antibodies may present certain advantages over their polyclonal counterparts. Based on molecular characterization of idiotype specificity of antibodies to neuroleptic peptides, Linthicum has pointed out certain basic principles in order to produce receptor-like antibodies. It is suggested that the hapten must be somewhat rigid to prevent the formation of antibodies which bind an inactive drug conformer and that the ligand must be conjugated to the carrier molecule in such a way that regions believed to be involved in receptor binding are fully exposed for antibody recognition. J23The nature of interaction of certain receptors with anti-idiotypes is briefly summarized below. Anti-idiotypic Antibodies to Insulin Receptors Sege and Peterson j°9 were the first to demonstrate that the anti-idiotypic antibodies raised against antibodies to insulin could reproduce certain biological actions of the hormone itself upon binding to the insulin receptors of rat thymocytes. These antibodies stimulated the uptake of aaminoisobutyric acid by the cells, thus mimicking the effects of insulin, and also inhibited the binding of 125I-labeled insulin to the receptors of thymocytes in a dose-dependent manner. These results were later confirmed by Schechter et a l . , TM who observed that mice immunized with insulin developed antibodies of two specificities: anti-insulin antibodies and anti-idiotypes to specific mouse insulin antibodies behaving as antibodies to insulin receptor. These observations suggest that the antirecep-

120 W. A. Marasco, H. J. Showell, R. J. Freer, and E. L. Becker, Proc. Natl. Acad. Sci. U.S.A. 80, 4084 (1982). 12~ T. Meo, C. Gramsch, R. Inan, V. HoUt, E. Weber, A. Herz, and G. Rieuthmiiller, Proc. Natl. Acad. Sci. U.S.A. 80, 4084 (1983). ~22 S. Chamat, J. Hoebeke, and A. D. Strosberg, J. lmmunol. 133, 1547 (1984). ~z3 M. B. Bolger, M. A. Sherman, and D. S. Linthicum, in "Anti-Idiotypes, Receptors and Molecular Mimicry" (D. S. Linthicum and N. R. Farid, eds.), p. 93. Springer-Verlag, Berlin and New York, 1988. ~24y . Schecter, D. Elias, R. Maron, and 1. R. Cohen, J. Biol. Chem. 259, 6411 (1984).

24

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

tor antibodies were developed in the mice as a part of an idiotype-antiidiotype network in which the specific idiotypic antibody mimicked the structure of insulin receptor. Anti-idiotypic Antibodies to Nicotinic Acetylcholine Receptors Wasserman et al. 125prepared antibodies against a Bis-Q derivative, an agonist of nicotinic acetylcholine receptor, in rabbits. These antibodies have agonist binding properties similar to that with the receptor; in other words, there is a correlation between the affinity of the analogs for the antibodies and their biological activity. These results indicate that in the case of anticholinergic antibodies, the antibodies present a structural analogy with the acetylcholine receptor only when it is in its active state. Furthermore, by immunizing mice with Bis-Q coupled to bovine serum albumin (BSA), Cleveland et a1.117 obtained a population of splenocytes which secreted anti-Bis-Q antibodies. The fact that they obtained from the same mice anti-idiotypic antibodies directed against nicotinic acetylcholine receptor suggests the existence of a certain degree of homology between anti-Bis-Q antibodies and the receptor. Anti-idiotypic antibodies to acetylcholine receptors were obtained by immunizing rabbits with antibodies to Bis-Q, a cholinergic antagonist. 125 These antibodies recognized the acetylcholine receptors of Torpedo, rat, and eel. The antibody binding to the receptor was inhibited by the ligand Bis-Q. Two of the immunized rabbits developed signs of muscular weakness resembling the symptoms of myasthenia gravis. These findings prompted the authors to propose that a similar mechanism may play a role in the etiology of at least some autoimmune disorders in which antibodies to various other receptors may be involved. Anti-idiotypic Antibodies to Substance P Receptors Polyclonal antibodies raised against substance P were characterized for their specificity toward substance P and about 30 substance P-related peptides. 126For each compound the authors observed a close correlation between the capacity to bind to antisubstance P antibodies and biological activity, indicating that the combining sites of antisubstance P antibodies and of substance P receptors may be similar. Anti-idiotypic antibodies to substance P receptors were obtained by ~25N. H. Wasserman, A. S. Penn, P. I. Freimuth, N. Treptow, S. Wentzel, W. L. Cleveland, and B. F. Erlanger, Proc. Natl. Acad. Sci. U.S.A. 79, 4810 (1982). 126j. y . Couraud, E. Escher, D. Regoli, V. Imhoff, B. Rossignol, and P. PradeUes, J. Biol. Chem. 2611, 9461 (1985).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

25

immunizing rabbits with antisubstance P antibodies. These anti-idiotypic antibodies inhibited the spasmogenic action of substance P on the guinea pig ileum. They specifically enhanced phospholipid turnover in rat parotid gland cells, a physiological function mediated through an activation of substance P receptors, in a similar way to substance P. In other words, these anti-idiotypic antibodies behaved either as agonists or as antagonists for substance P receptors depending on the biological test? 26

Anti-idiotypic Antibodies to Angiotensin H Receptors Couraud 127 has obtained a panel of monoclonal antibodies against angiotensin II, an effector hormone of the renin-angiotensin system. The antibody specificity was analyzed by using a range of 20 structural analogs of the hormone. From the binding studies and the putative three-dimensional structures of the tested peptides, three families of antibodies were distinguished which recognized overlapping epitopes, suggesting that conservation of the native conformation of the angiotensin II molecule in the analogs is essential for retaining a high affinity to the antibodies. With one of the monoclonal antibodies the affinities of the angiotensin II analogs were correlated with their biological activities but not with their affinity to the receptor, which could be interpreted as mimicry of the active conformation of the receptor site by the antibody-binding site. One of the antiangiotensin antibodies was found to mimic the receptor-binding site in an activated state, and a second one presented no similarity with the receptor. Nevertheless, antisera raised against both these Abls crossreacted with the angiotensin II receptor.128

Anti-idiotypic Antibodies to fl-Adrenergic Receptors Both polyclonal and monoclonal antibodies have been produced in the laboratory of S t r o s b e r g 122,129 by immunizing mice alprenolol (a fl-adrenergic antagonist) coupled to the carrier protein BSA. These high-affinity antibodies specifically recognized other fl-adrenergic ligands with an affinity higher to antagonists than to agonists. One of these monoclonal antibodies, on account of its exclusive physicochemical properties, was the ideal choice as a probable model of receptor to raise anti-idiotypic antibodies. Schreiber et al. 113were the first to prepare anti-idiotypic antibodies to fl-adrenergic receptors (fl-AR). They first produced antibodies against 127p. O. Couraud, J. Immunol. 136, 3365 (1986). 128p. O. Couraud, J. Immunol. 138, 1164 (1987). 129j. Hobeke, G. Vauquelin, and A. D. Strosberg, Biochem. Pharmacol. 27, 1527 (1978).

26

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

alprenolol, a potent adrenergic antagonist, in rabbits. The antialprenolol antibodies were capable of binding to other antagonists and had less avidity to catecholamine agonists. They then raised anti-idiotypic antibodies in rabbits by immunizing with antialprenolol immunoglobulins. These anti-idiotypic antibodies bound specifically to turkey erythrocytes, which possess/31-AR. They competed with [3H]dihydroalprenolol for binding on the fl-AR on purified turkey erythrocyte membranes and also elicited specific physiological responses upon binding to receptors, stimulated the basal adenylate cyclase activity, and enhanced adenylate cyclase activation by catecholamine. Production of anti-idiotypic antibodies directed against/3-AR was also reported later by Homcy et al. 114 However, as immunogen they used affinity-purified antialprenolol immunoglobulins (eluted with another antagonist), as compared to the DEAE-purified immunoglobulins used by Schreiber et al., to raise anti-idiotypic antibodies. A difference in the characteristics of their antibodies was that they behaved as competitive inhibitors, determined both by saturation binding analysis and by isoproterenol-mediated adenylate cyclase activation. Schreiber et al. attributed this discrepancy in the results to the impurity of the immunogen and the immunization schedule. The significance of the immunization schedule in the anti-idiotypic response was pointed out later by Couraud et al.13° They investigated the transient nature of the antireceptor anti-idiotypic response, which was shown to be cyclical and corresponded to the production of antibodies that could bind to/3-AR and stimulate adenylate cyclase. Disappearance of the antireceptor antibodies in the serum coincided with appearance of adrenergic ligand-binding activity, corresponding to the development of the autologous anti-anti-idiotypic antibodies (Ab3) directed against the induced anti-idiotypic molecules. The results of Couraud indicate that the rapid and cyclic disappearance of antireceptor antibodies (Ab2) is due to effects mediated by anti-idiotypic antibodies directed against Ab2. It was shown that a fraction of these anti-anti-idiotypic antibodies (Ab3) were able to bind alprenolol, the original antigen, and therefore could be considered as "Ab l-like" antibodies. Comparison of the Ab I and "Ab l-like" antibodies revealed differences in affinities but similarities in the ability to bind the anti-idiotypic antibodies and to the ligand-coupled affinity gel. Anti-idiotypic Antibodies to Thyroid-Stimulating Hormone Receptor

Rabbits immunized with rat anti-human thyroid-stimulating hormone (TSH) antibodies developed anti-idiotypic antibodies which recognized J30 p. O. Couraud, B. Z. Lu, and A. D. Strosberg, J. Exp. Med. 157, 1369 (1983).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

27

TSH receptors and inhibited the binding of bovine TSH onto the receptors of thyroid tissue membranes. These antibodies were also capable of eliciting a physiological response in that they stimulated adenylate cyclase activity in the presence of guanosine 5'-[/3, a-imido] triphosphate [Gpp(NH)p]. 131The anti-idiotypic antibodies which recognize TSH receptors have been used to examine the structure, biosynthesis, and turnover of the receptor as well as the interaction of the hormone subunits with that receptor. The role of autoanti-idiotypic antibodies which bind to TSH receptors leading to the development of Graves' disease has been investigated. 131,132 Structural Principles Underlying Molecular Mimicry The structural basis of anti-Id surrogacy remains largely unknown and provides one of the most interesting areas of structure-function relations yet to be explored. Short of direct crystallographic evidence on Id-anti-Id complexes, structure-function relationships for idiotypy have drawn on results from analysis of antibody structure and antigen-antibody complexes, 133-136 sequence analysis of anti-idiotypes compared with antigens, 110,137and inhibition experiments involving synthetic VH and VL peptides that compete with antigen for antigen-binding sites.~38,139 Here, we briefly summarize some of the more salient features of such studies.

Idiotope Topography Protein antigenicity is clearly a surface property. Monoclonal mapping of protein surfaces indicates that the entire surface of a protein is poten13~ N. R. Farid, in "Anti-Idiotypes, Receptors and Molecular Mimicry" (D. S. Linthicum and N. R. Farid, eds.), p. 61. Springer-Verlag, Berlin and New York, 1988. 132L. D. Kohn, W. A. Velente, P. Laccetti, C. Marcocci, M. De Luca, P. Ealey, N. J. Marshall, and E. F. Groliman, in "Monoclonal and Anti-Idiotypic Antibodies: Probes for Receptor Structure and Function" (J. C. Venter, C. M. Fraser, and J. Linstrom, eds.), Vol. 4, p. 85. Liss, New York, 1984. 133T. Kieber-Emmons and H. K6hler, lmmunol. Rev. 90, 29 (1986). 134T. Kieber-Emmons, E. Getzoff, and H. K6hler, Int. Reo. lmmunol. 2, 339 (1987). 135 p. M. Coleman, G. M. Air, R. G. Webster, J. N. Vargehese, A. T. Baker, M. R. Leutz, P. A. Tullock, and W. G. Laver, Immunol. Today 8, 323 (1987). ~36C. Bruck, M. S. Co, M. Slaoui, G. N. Gaulton, T. Smith, B. N. Fields, J. I. Mullins, and M. I. Greene, Proc. Natl. Acad. Sci. U.S.A. 83, 6578 (1986). m G. Mazza, P. Oilier, G. Somm6, D. Moinier, J. Rocca-Serra, J. Van Rietschoten, J. Th~ze, and M. Fougereau, Ann. lmmunol. (Paris) D6D, 259 (1985). t3s C.-Y. Kang, T. K. Brunck, T. Kieber-Emmons, J. E. Blalock, and H. KOhler, Science 240, 1034 (1988). m W. V. Williams, H. R. Guy, D. H. Rubin, F. Roby, J. N. Myers, T. Kieber-Emmons, D. B. Weiner, and M. I. Greene, Proc. Natl. Acad. Sci. U.S.A. 85, 6488 (1988).

28

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

tially antigenic, with epitope domains being associated with highly variable residue positions. This relationship between sequence variation and surface epitope expression has led to a statistical approach to identify variable domains that are restricted to the protein s u r f a c e J 4° The potential antigenic variability and the expression of idiotopes by immunoglobulins has been demonstrated by the application of this approach. 133,134It was shown that particular domains which are homologous on the surface of immunoglobulins are involved in idiotope expression. These domains have been referred to as idiotope-determining regions (IDR). Essential findings were the role of framework regions, lending to the interacting complementarity of Id-anti-Id recognition and a theoretical view of the antibody molecule as having multiple interacting surfaces, with no distinction being made between paratopes (antigen-binding sites) and idiotopes. 133 The identified overlap between idiotopes and paratopes implies that idiotopes can participate as binding sites; the dual roles of a binding determinant or being bound by another antibody may be expressed simultaneously by the same topographic site. Consequently, it was proposed that sites on immunoglobulins cannot be easily characterized functionally, highlighting the arbitrariness of anti-Id classification with respect to a and/3 terminology. IDR positions play a dominant role in determining the idiotypic topography of an antibody. Depending on the site of the anti-Id recognition site, idiotope expression may overlap IDR sites on both light and heavy chains, as well as constant regions. In addition, the analysis of surface variation predicts a broad nature in IDR domains. This broadness suggests a fine structure in idiotope recognition which has been experimentally observed. In other words, the diversity in sequence within a hypervariable segment alone will not generate the topographic diversity required for antiprotein specificity. Differences in specificity require hypervariable regions of differing length as well as some ability for local mobility within the hypervariable loops, leading to a notion of inducible complementarity in antibody binding.

Mimicry of Contact Residues Since antibody-antibody interactions are modulated by their large surfaces, the complete description for an idiotype may entail contact points which are close in space but remote in sequence. The summation of individual atomic interactions between idiotype and anti-Id is the molecular grammar of the system. Of interest is an understanding of the extent to ~40T. Kieber-Emmons and H. K6hler, Proc. Natl. Acad. Sci. U.S.A. 83, 2521 (1986).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

29

which a molecular grammar is shared between an antigen and a surrogate anti-Id and how the immune system perceives this grammar. Possible ways in which anti-Id antibodies can potentially mimic protein antigen have been summarized. 112Sequence homology with a protein antigen has been suggested for the mimicking capabilities of some monoclonal anti-Ids, n° The possibility also exists that anti-Id antibodies can regulate the immune response without bearing a structural or stereomimicry relationship with a nominal antigen. Perhaps such is the case for carbohydrate and hapten antigens. 28a41a42 These Ab2s effectively immunoregulate via biological mimicry of a nominal antigen. In this case, the difference in therapeutic quality between Ab2s acting as immunoregulators for a particular disease will depend on the affinity of the Id-anti-Id interaction, the frequency of idiotype-positive antigen-positive (Id ÷ Ag ÷) versus Id ÷ Ag- lymphocytes, and the availability of the right T helper cells. The idea of a sharing of molecular grammar between antigen and antiId does not imply that sequence identities need to be found between protein antigens and anti-Ids. The manner in which an antibody binds an antigen can be quite different from the way the corresponding anti-Id may bind. These extended interactions bring into question the requirements for fidelity or faithfulness of an antibody in mimicking antigen binding. There are several components which contribute to the degree of mimicking fidelity. These components relate to the sharing of a configuration of the essential "mimicking" functional groups between an antigen and antiId antibody, complementary contact residues between two antibodies which lend to the overall association constant for a particular complex formation, and the requirement for the stabilization of a particular structural environment for full antigenic mimicry. These aspects have been highlighted in considering a VH peptide that inhibits the self-binding of antiphosphorylcholine antibodies T15 and MCPC603.138 This peptide spans the second hypervariable domain and a portion of Fr3 of the respective heavy chains. While this peptide segment is conserved in M167, M167 does not self-bind, nor does it inhibit the self-binding of TI5 and M603.138,143 In general terms, the phenomenon of molecular mimicry also does not imply that the tertiary folding pattern of an antigen and anti-Id need be the same. The identification of relatively homologous sequence regions be~41 D. L. Sacks, L. V. Kirchoff, S. Hieny, and A. Sher, J. lmmunol. 135, 4155 (1985). 142H.-L. Cheng, A. K. Sood, R. E. Ward, T. Kieber-Emmons, and H. K6hler, Mol. Immunol. 25, 33 (1988). ~43C.-Y. Kang and H. KOhler, J. Exp. Med. 163, 787 (1986).

30

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

tween nominal antigen and anti-Id does not ensure the complete structural identification of chemical-structural equivalence since different threedimensional environments can influence the folding patterns of related sequences. However, what seems to be important in terms of structural mimicry is the superpositioning of a similar putative functional binding site. This has been observed in the reovirus system. H° In this system, a 17-residue stretch of Fr2 and L2 of a monoclonal antireceptor antibody exhibited sequence homology with a putative site in the sigma 1 cell attachment protein of reovirus type 3. This peptide has since been characterized structurally and functionally as the putative anti-Id-hemagglutinin mimicking site. 139 The superpositioning of similar reactive groups was also recently illustrated via sequence analysis of an anti-Id that mimics a rabbit allotype, which showed a reverse sequence homology between anti-Id and antigen.144 A synthetic peptide of this reverse sequence inhibited competitive inhibition with the antigen. 144

Biological and Molecular Mimicry We have previously suggested that the biological activities of anti-Ids may not equate with stereochemical or stereomolecular mimicry.142 For example, while we have identified anti-Ids that are functionally similar in that they compete with gp52 antigen binding to 1 IC 1, they are biologically dissimilar. For the purpose of making therapeutic idiotypic agents, the distinction between biological and chemical mimicry needs to be considered. Biological mimicry of stimulation by antigen occurs when the target idiotope is associated with the given paratope with high frequency in the responding precursor B cell repertoire. For the practical purpose of producing an effective idiotype vaccine, molecular mimicry and biological mimicry of antigens are equally valid concepts. For the understanding of the molecular basis of stimulatory idiotope-anti-Id interactions at the structural level the primary sequence structure is nonetheless needed. Computer modeling of contact sites between antigen and antibody as well as between anti-Id and idiotype will help to explain the three-dimensional basis for molecular and biological mimicry of antigens. Discussion and Synopsis

Theory The Jerneian network concept is preoccupied with attempts to describe the structural basis of how idiotypes and anti-idiotypes interact. i44 V. H. Van Cleave, C. W. Naeve, and D. W. Metzger, J. Exp. Med. 167, 1841 (1988).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

31

This desire to understand idiotypic structures has its roots in the prevailing emphasis on antibody structure in the 1970s. When Jerne and Lindenmann developed the concept of an idiotypic network, the antibody structure had just been solved 145,146and the focus was on sequencing antibodies with known activities; furthermore, the first three-dimensional immunoglobulin structures were determined. 147,~48 Jerne's distinction between main idiotypic circuits and parallel sets was derived from structural consideration on the complementarities of paratopes (binding sites) and idiotopes associated with different antibodies. Much intellectual satisfaction could be obtained from developing network schemes. Using idiotypic complementarities it was easy and logical to draw cartoons of open-ended networks or circular network cascades. 149 Retrospectively, one could argue that the biological consequences of idiotypic network connectivity were overlooked or not seriously considered. Certainly this can be said about the concept of the so-called internal image. It was conceived as a three-dimensional shape mimicry creating an idiotypic structure which resembles the shape of an external antigen. The next step in the sequel of network thinking was to propose that such internal images could be used as vaccines.~9-2~ Shortly after these ideas had been proposed, anti-idiotypes were screened for paratopic specificity and used to induce specific immune resporlses. 28,68,~5°The successful induction of protective immune responses with a special class of anti-idiotypes was considered experimental proof for the internal image concept and also indirectly for the "chemical" view of the network idea. More recently, however, a more complex picture has begun to appear, as more experimental data from anti-idiotype induction became available. What is emerging from these studies is evidence that the biological results cannot be correlated with the chemical classification of anti-idiotypes into Ab2~,/3, or 3/)4 Other issues concern the classification of internal image anti-idiotypes, which deals with the question of affinity of the idiotypic 145G. M. Edelman, B. A. Cunningham, W. E. Gall, P. D. Gottlieb, U. Rutishauser, and M. J. Waxdal, Proc. Natl. Acad. Sci. U.S.A. 63, 78 (1969). ~46F. W. Putnam, A. Shimizu, C. Paul, T. Shinoda, and H. K6hler, Ann. N.Y. Acad. Sci. 190, 83 (1971). i47 R. J. Polljak, L. M. Amzel, H. P. Avey, B. L. Chen, R. P. Phizackerley, and F. San, Nature (London) 245, 165 (1973). 148 E. A. Padlan, D. Segal, T. F. Spande, and D. R. Davies, Proc. Natl. Acad. Sci. U.S.A. 69, 3689 (1972). ~49j. UrbaJn, M. Wikler, J. D. Franssen, and C. Collignon, Proc. Natl. Acad. Sci. U.S.A. 74, 5126 (1977). ~50R. C. Kennedy, J. L. Melnick, and D. R. Dreesman, Science 223, 930 (1984).

32

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1l

complex versus the antigen-antibody complex.151 Looking back to early experiments on autoanti-idiotypic antibodies, 7 internal images appear not to be found among such autoanti-idiotypic antibodies. Furthermore, when the Ab3 response induced by Ab2s was analyzed, the Abl antibodies expressed only the cross-reactive major idiotype characteristics for the normal response. 152Thus, it appears that the so-called internal image antiidiotypes do not represent a major portion of the network circuit and are not effective as expected in repertoire selection. The biological behavior of anti-idiotypes is best interpreted in terms of a network modulator or regulator. This definition comes close to Bona's term of regulatory idiotope which plays important roles in the response. Molecular Mimicry versus Network Mimicry

The concept of idiotypic mimicry is an integral part of the network hypothesis. Because the problem of self-nonself discrimination does not exist with idiotypes and anti-idiotypes, the network hypothesis supersedes the clonal selection theory which had been created to deal with the self-nonself issue. However, the original concept of the network hypothesis had been based on structural complementarity of idiotype and antiidiotype; the view of what constitutes an internal image was biased in its interpretation as a three-dimensional similarity between nominal antigen and an idiotope expressed by the so-called paratope of Ab2/3. In the following we present a biological interpretation of idiotypic mimicry and highlight the differences between the new network concept and the original Jerneian concept. The existence of classic internal images is explained with the statistical necessity of a vast idiotypic repertoire which contains shapes found on external antigens. However, it is clear that even simple monoepitopic hapten antigens induce a specific immune response only by involving complex network reactions of idiotypic help and suppression. 153 Why should an anti-idiotype engage with the network in less complicated ways than a hapten like phosphorylcholine? It is more reasonable to emphasize a statistical necessity derived from the complex connectivity in the network, which predicts that network interaction induced by nominal antigen and idiotopes can be similar and overlap. The result can be either induction or suppression (tolerance) of a specific response through either nomi15~ F. J. Stevens, J. Jwo, W. Carperos, H. K6hler, and M. Schiffer, J. Immunol. 137, 1937 (1986). 152J.-H. Huang, R. E. Ward, and H. Kfhler, Immunology 63, 1 (1987). 153H. K6hler, Transplant. Rev. 27, 24 (1975).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

33

nal antigen or idiotope. In other words, the network connectivity creates a statistical necessity that external antigen and network antigen (idiotope) trigger the same regulatory interaction, which produces a specific immune response. A corollary of this is that all idiotopes are regulatory. Regulatory idiotopes, as defined by Bona, 15 are probably not a special class of idiotopes. It is interesting to note that one of Bona's criteria, membership to the same Vn or VL family, has recently been eliminated (C. A. Bona, personal communication).

New Mimicry Concept Based on Network Connectivity In Table II the most important differences between the Jerneian network mimicry concept and the one discussed here are listed. A detailed description of the revised network hypothesis has been given elsewhere. 154 The need to revise the Jerneian network concept and particularly the concept of "internal image idiotopes" arises from a number of recent experimental data. First, the detailed understanding of how an antibody makes contact with a protein antigen ~55 demonstrates that an overall shape complementarity between the antibody binding site and the antigen contact surface may not be needed. This notion is supported by the lysozyme-antilysozyme complex in which the contacting surfaces are fiat. Thus, by extrapolation to idiotype-anti-idiotype complexes, three-dimensional shape mimicry of antigen and idiotopes is not required. The similarity between antigen-antibody and anti-idiotype-antibody contacts needs only to relate to a similar number of amino acid side-chain contacts producing similar overall affinities of the two complexes. The requirement for a three-dimensional mimicry between antigen and anti-idiotype probably applies only to a small number of situations. Second, as alluded to already, the most important factor in determining effective biological mimicry of antigens by idiotopes is the affinity of the complexes. If, for example, the affinity of the idiotope-anti-idiotope complex is magnitudes higher than the affinity of the idiotype to a given antigen or hapten, then inhibition of the idiotypic complex by antigen may not be experimentally achievable. Thus, by inference of the original Jerneian criteria, such anti-idiotypes could be classified as Ab2a and would not be further pursued as surrogate antigens in biological testing. 154 H. K/~hler, T. Kieber-Emmons, S. Srinivasan, S. Kaveri, W. J. W. Morrow, S. Miiller, C.-Y. Kang, and S. Raychaudhuri, Clin. Immunol. Immunopathol. (in press). 155A. G. Amil, R. A. Manizinzzu, S. E. V. Phillips, and R. J. PoUjak, Science 233, 747 (1986).

34

IDIOTYPES, ANTI-IDIOTYPES, AND MOLECULAR MIMICRY

[1]

TABLE II COMPARISON OF JERNEIAN AND NEW ANTIGEN MIMICRY CONCEPTS

Jerneian concept

New concept

Statistical necessity from the idiotype repertoire predicts existence of idiotypic mimicries of external antigens

Statistical necessity from network connectivity predicts that the same regulatory interactions are triggered by external antigen and (anti)-idiotypes to produce a specific response Similarity of side-chain and backbone contacts in antigen and (anti)-Id binding All idiotopes are regulatory

Shape mimicry of external antigen by anti-Id Existence of regulatory idiotopes as special class Ab2 c~-fl distinction based on inhibition of Id-anti-Id binding by antigen Screening of internal image anti-Id by inhibition studies and response in third party species Internal image anti-Id concept

Network antigen defined by affinity and number of network contacts (frequency of idiotope target precursors) Screening of anti-Id expressing network antigens by idiotype "typing and test immunization in relevant species Biological network antigen (epitope) concept

Third, to demonstrate this point, experiments have been reported in which an Ab2ot-classified anti-idiotype induced a significant and specific immune r e s p o n s e ) 52'156 In o n e c a s e , 157 the Ab2ct-induced response was not significantly different than the Ab2/3-induced response. Thus, it is clear that the original criterion of antigen inhibition for selecting the internal image anti-idiotype may be misleading and is based on an erroneous conceptual basis. In support of this notion, Raychaudhuri and colleagues ~58have recently shown that only one out of several anti-idiotypes, which were typed as Ab2/3, did induce protection against tumor growth. These findings underscore the importance of the nature and specificities of network interactions induced by immunization with anti-idiotypes. It has become evident that immunochemical screening is a poor predictor for the biological responses induced by anti-idiotype immunization. The new concept has important ramifications for the methods of screening for so-called internal image idiotopes, or network epitopes or antigens, as defined here. Immunochemical screening of anti-idiotypes identifies potential candidates for effective network antigens. The critical 156M. R. Schick, G. R. Dreesmann, and R, C. Kennedy, J. Immunol. 138, 3419 (1982). 157J.-H. Huang, R. E. Ward, and H. Krhler, J. Immunol. 132, 770 (1986). ~8 S. Raychaudhuri, Y. Saeki, H. Fuji, and H. Krhler, J. Imrnunol. 137, 1743 (1986).

[1]

IDIOTYPIC NETWORKS AND MOLECULAR MIMICRY

35

screen has to be done biologically in the relevant target species. Biological effectiveness of network antigens depends on a number of different parameters, such as affinity to connecting idiotopes and frequency of B or T cell idiotopes in the repertoire with which the network antigen can interact. Experimentally obtained anti-idiotypes provide points of entry into a complex regulatory network. In the absence of a complete description of the connectivity of network segments responsible for a given response, the effective entries can only be found by biological screening. The emphasis on network connection in discussing idiotypic mimicries is not meant to negate the fact that the union of idiotypes and anti-idiotypes follows biochemical rules not dissimilar from those controlling protein-protein interactions. Thus, knowledge obtained from studies of protein complexing should be used to understand idiotypic complexes. In their most reduced form, protein-protein interactions are governed by affinities of their amino acid side chains accessible on the surface. The three-dimensional relationship of protein contacting surfaces is secondary to the potential of interacting side chains. The interacting surfaces can restrict side-chain contacts but not create them. Again, more data on the three-dimensional complexes of idiotypes and anti-idiotypes, as well as of antibody-protein complexes, are needed to better define the rules of idiotypic network interaction and the biological and molecular basis of idiotypic mimicry. For the time being, biological and molecular mimicry in the immune network coexist as different entities because of an incomplete understanding of the complexity of network connectivities and the lack of molecular details in protein-protein binding. It is hoped that this volume will promote bridging of the gaps between biological and chemical mimicry in the immune system. Acknowledgments The authors wish to thank Liz Cart for typing and assisting with the editing of the manuscript.