Proc. Natl. Acad. Sci. USA
Vol. 74, No. 11, pp. 5126-5130, November 1977 Immunology
Idiotypic regulation of the immune system by the induction of antibodies against anti-idiotypic antibodies (immunoregulation/idiotypic networks/anti-anti-idiotypic antibodies)
J. URBAIN, M. WIKLER, J. D. FRANSSEN, AND C. COLLIGNON Laboratory of Animal Physiology, Universit6 Libre de Bruxelles, 67, rue des Chevaux, 1640, Rhode-St-Genese, Belgique
Communicated by J. Brachet, August 24, 1977 ABSTRACT Anticarbohydrate antibodies (Abl) were isolated from a rabbit hyperimmunized with Micrococcus lysodeikticus and injected into allotype-matched rabbits in order
imagine that AbI and AbS could be idiotypically similar because both should be complementary to the idiotopes of Ab2. The experiments described in this paper were undertaken to answer the following questions: (i) is it possible to induce Ab3? (ii) if so, what are the properties of AbW, is AbS carrying idiotypic specificities similar to those of Abl, and does Ab3 react with the same antigen as Abl? and (iii) does immunization with Ab2 influence the immune repertoire expressed after injection with the original antigen?
to obtain specific anti-iodiotypic antibodies (Ab2). Ab2 was isolated by means of a Sepharose column coupled to the anticarbohydrate antibodies and was injected into two allotypematched rabbits. These latter rabbits produced specific antianti-idiotypic antibodies (Ab3) probably sharing idiotypic specificities with Abl. However, these Ab3 did not react with the antigenic carbohydrate moiety of bacteria. The two rabbits that had produced Ab3 were then immunized with M. lysodeikticus and synthesized anticarbohydrate antibodies (Abl') bearing idiotypic specificities similar to those of Abl. The immune repertoire which is effectively expressed in one individual depends not only on the antigenic stimulation but also on the previous idiotypic history of the individual. These data support the concept that the immune system is a functional idiotypic network.
MATERIALS AND METHODS Immunizations. Anticarbohydrate antibodies (anti-CHO) were obtained in rabbits, injected intravenously three times a week with Micrococcus lysodeikticus [bacteria (Mi)] during 10 weeks. Immunoadsorbents were prepared with lysozymedigested bacteria (Mi) cell walls coupled to Sepharose (21). The main fraction of anti-CHO antibodies from rabbit 2975 was eluted from the immunoadsorbent with 5% glucose in phosphate-buffered 0.15 M NaCl (Pi/NaCl), pH 7.5. Ab2 anti-idiotypic antisera were raised in allotype-matched rabbits. Allotypes tested were al, a2, a3, b4, b5, b6, b9, c7, c21, e14, dlI, and d12. Purified anti-CHO antibodies from rabbit 2975 (Abl) were polymerized with glutaraldehyde as described (22) and injected into six allotype-matched rabbits. Injections of 1 mg of AbI emulsified in complete Freund's adjuvant twice a week during 1 month was followed by the injection of 1 mg of AbI intravenously. Abl (1 mg) in incomplete Freund's adjuvant was then injected once a week. Ab2 from rabbit II (2238) were purified by means of a Sepharose column to which 200 mg of Ab had been coupled. Twenty milliliters of serum 2238 (Ab2) was passed through the column. After extensive washing, bound proteins were eluted with glycine-HCI, pH 2.2, and their idiotypic nature was checked by immunodiffusion and radioimmunoassay. The purified Ab2 were crosslinked with glutaraldehyde, emulsified with Freund's adjuvant, and injected into two allotype-matched rabbits [111(2281) and 111(5083)] according to the schedule used to elicit Ab2. After a rest period, rabbits III were immunized with bacteria (Mi). Anti-CHO antibodies from rabbits III (Abl') were isolated and injected into allotype-matched rabbits to elicit anti-idiotypic antibodies (Ab2') specific for Abl'. The experimental schedule used was the same as that used to elicit Ab2. The experimental scheme is depicted in Fig 1. Radioimmunoassay. Iodination was performed according to Hunter (23). The specific radioactivity of immunoglobulins
As a rule, in response to the same antigen, different individuals from the same species express different subsets of the total antibody repertoire of the species, with each subset bearing different idiotypic specificities (1-7) (these antibodies will be referred to as Abl). Idiotypic similarities were observed only in certain idiotypic systems (5, 8) and were only partial with two main kinds of exceptions: (i) some rabbit families or some inbred strains of mice immunized with certain antigens synthesize antibodies with very similar idiotypic specificities (9-11); and (ii) very similar idiotypic specificities are often found among different antibody subpopulations present in the serum of a single individual immunized with a single antigen
(12-20). The V domains of anti-idiotypic antibodies (which will be called Ab2) might themselves function as antigens and therefore elicit the synthesis of anti-anti-idiotypic antibodies (which will be referred to as AbS). These AbS might elicit the synthesis of Ab4, and so on. If the experimental rules that have been found at the Abi level (the AbI of different individuals from the same species express different idiotypic specificities) also apply to the Ab2 and AbS levels, different individuals injected with the same AbI might synthesize different Ab2 characterized by different idiotypic specificities and so on. This line of reasoning leads to a rapidly increasing diversity, and Ab4 and Ab5 will already encompass a large part of the immune repertoire. This is not necessarily so. The idiotypic diversity at the Ab2 level could be much more restricted than at the AbI level. In other words, anti-idiotypic antibodies synthesized by different animals in response to the same Abl may frequently express the same idiotype. On the other hand, if there is a close relationship between idiotypic specificities and active sites, it is possible to
Abbreviations: Abl, antibodies directed against bacteria (Mi); Ab2, anti-idiotypic antibodies; Ab3, anti-anti-idiotypic antibodies; Abl', antibacteria antibodies in rabbits that have previously synthesized Ab3; Ab2', anti-idiotypic antibodies against Abl'; anti-CHO, anti-carbohydrate antibodies; bacteria (Mi), Micrococcus Iysodeiktcus; Pi/NaCl, phosphate-buffered (pH 7.5) saline.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
5126
Proc. Natl. Acad. Sci. USA 74 (1977)
Immunology: Urbain et al.
5127
Anti-CHO Abl (2975) Anti-Abl Ab2 (2238) Anti-Ab2 Ab3 (2281) Anti-bacteriia (Mi) Abl' (22'81)
Anti-Abl Ab2 (2239,2240,2716, 520)3,5271)
Anti-Ab2 Ab3 (5083) Anti-bacteria (Mi) Abl' (5083)
c 0
._ c
doC
Anti-Abl' Anti-Abl' Ab2' (2729) Ab2' (2722) FIG. 1. Immunization scheme of rabbits I, II, and III.
ranged between 106 and 5 X 106 cpm/tig. Binding curves of radiolabeled antibody were performed with 25 ng of antibody and increasing amounts of crosslinked anti-idiotypic sera. Crosslinking of antisera was performed with ethyl chloroformate (24). All dilutions were done in phosphate buffer containing 1% ovalbumin. The tubes were incubated overnight at room temperature and centrifuged at 13,000 rpm for 5 min. After centrifugation, the top half of the tube content was transferred to another tube and radioactivity in both tubes was assayed in a gamma scintillation counter (Beckman). Control tubes without solid antiserum were included in each experiment. Inhibition curves were performed as follows (25): a fixed amount of crosslinked antiserum, corresponding to 70% of the plateau (binding), the cold putative inhibitor in variable amounts, and a fixed amount of labeled antibody were mixed and the tubes were processed as described above. RESULTS AND DISCUSSION The M. lysodeikticus system is characterized by two major antigenic specificities, a glucose N-acetylaminomannuronic acid polymer and a peptidoglycan moiety (21). Anti-CHO antibodies were isolated from rabbit 2975 hyperimmunized with bacteria (Mi). This rabbit will be referred to as rabbit I and the purified anti-CHO antibody from this rabbit will be called Abl. Purified AbI were injected into six rabbits of matched allotypes to elicit the synthesis of Ab2. These rabbits will be designated 11(2238), II(2239), 11(2240), II(2716), 11(5203), and II(5271). The precipitating antisera obtained reacted with Abl in Ouchterlony immunodiffusion but not with the preimmune serum of rabbit I. These anti-idiotypic sera were tested in immunodiffusion with a panel of sera from 60 rabbits hyperimmunized with the same antigen. Only 1 of the 60 sera (rabbit 2972) showed a precipitation line with the different Ab2. This frequency of idiotypic heterologous crossreactions is of the same order of magnitude as has been found in other antigenic systems (5, 8, 16). The specificity of the Ab2 sera was further investigated by radioimmunoassay methods. Except for Abi' (see below), only the unlabeled homologous Abl is an efficient inhibitor of the reaction between radiolabeled Abl and insolubilized Ab2 sera (Fig. 2). The controls shown in the same figure demonstrate the idiotypic nature of the reaction. Even the anti-CHO antibody from rabbit 2972, which gave a precipitation line with Ab2, was unable to inhibit the reaction. The specificity of purified Ab2 from rabbit II(2238) was checked by immunodiffusion and radioimmunoassay (Fig. 3D). These purified Ab2 were capable of 85% binding to a solidphase anti-Ig serum.
Inhibitor, ,g
FIG. 2. Inhibition of the binding of radiolabeled antibody from rabbit I (Abl) to crosslinked homologous anti-idiotypic antiserum Ab2 (2238). Inhibitors are unlabeled, specifically purified, anti-CHO antibodies from rabbit I (Abl, 0) and rabbit III(2281) (Abi', A). Anti-idiotypic antiserum 2238 (M) was added in amounts ranging from 0.25 to 250 ,ul of whole serum. Symbols (O) at right top give inhibition values obtained with specifically purified antibacterial antibodies from seven different rabbits. Controls (0) (with preimmune sera from rabbits II and III and heterologous anti-idiotypic antiserum) are also shown. Antibody from rabbit 2972 is the only antibody (out of 60) that reacted with Ab2 (2238).
In order to elicit Ab3, Ab2 were injected into rabbits III(2281) and III(5083). Ten weeks after the start of immunization, sera from both rabbits showed a precipitation line with the immunizing Ab2 by Ouchterlony assay (Fig. 3 A-D). These AbS did not show a precipitation line with either the preimmune serum of rabbit 11(2238) and or with Ab2 directed against other anti-bacteria (Mi) antibodies or sera from other rabbits hyperimmunized with bacteria (Mi). A panel of 10 Ab2 and 12 immune sera was tested. The frequency of crossreactions at the AbS level did not seem to be larger than at the Ab2 level. It must be stressed that the two Ab3 from rabbits 11(2281) and III(5083) specifically recognized not only the immunizing Ab2 but also the other Ab2 specific for Abl (2975) (Fig. 3E). Unless one assumes that all the Ab2 raised by the same Abi share similar idiotopes recognized by the combining sites of both Ab3, these data seem to suggest that Abl and both AbS possess similar idiotypic specificities recognized by the combining sites of Ab2. The data of Fig. 4 indicate that both Ab3 inhibit the reaction between radiolabeled Fab'2 Abi and solid-phase Ab2 (2238). The two Ab3 were repeatedly passed through columns of Sepharose coupled with the carbohydrate of bacteria (Mi). This treatment did not remove more than 10% of the inhibitory activity of AbM. Therefore, the bulk of Ab3 did not bear a combining site with a sufficient affinity to recognize the carbohydrate determinants of bacteria (Mi). At this stage, different interpretations were possible: (i) the results of inhibition curves could be due to steric inhibitions because Abl and AbS are recognizing the same Ab2; and (ii) Abl and both AbS bear similar idiotypic specificities. The Ab3 of both rabbits would be an imperfect image of Abl because the Ab3 sera have no anticarbohydrate affinity. To obtain further information, the rabbits that had synthesized Ab3 were immunized with bacteria (Mi). The rationale
5128
Proc. Natl. Acad. Sci. USA 74 (1977)
Immunology: Urbain et al. B
A
antiserum Abi'
adsAbl' Ab 2972
Ab 1 0
A
Abl 0
0
0
B
Ab3 5083 0
0
Ab2
Ab2
0
Ab2978
Abi' *
0
Abl'2281
Ab1' 5083
0
0
Ab3 2281 0
Ab3 0 D
C
Ab1 0
Abl'
Ab3 *
.
Ab 2972
Abl 0
0
D
Ab3 5083 0
Ab2
Ab2 At)2978 * antiserum Atb1I
Ads Abl
Abl'2281 0
0
Ab 15083 0 Ab3 2281 0
F
E
Ab 1' 2281
Ab2' 2722
0
AD2 2238
Ab2 2239 Abi
0
0
Ab3
Ab3 2281
0
Ab 2
0
Abs' 2729
Ab2 5203 A 3 5083
Abi
Ab2 5271
Abi' 5083
0
0
FIG. 3. Ouchterlony assays of immune sera used. The center well in A, B, and C was filled with Ab2 (2238). The center well of D contained purified Ab2 which was used for immunizations. Wells marked "Abl" or "Abl"' were filled with anti-CHO purified by affinity chromatography. Wells marked "ads Abl"' contained the Abl' from rabbit 2281, which had been adsorbed with bacteria. Rabbit 2972 had been hyperimmunized with bacteria and gave the only serum (out of 60) that reacted with Ab2. Ab 2978 was a randomly chosen anti-CHO. The center well in E was filled with Ab3 (2281). The center well in F contained Ab2' (2729) raised against Abl' (2281).
behind the experiment was as follows. It has been shown by several laboratories (12-20) that it is not infrequent to find strong idiotypic similarities between different antibody subpopulations synthesized by a single rabbit. Furthermore, in many cases, antigen seems to induce not only specific antibodies but also the so-called nonspecific immunoglobulins (26, 27). Several workers have reported a sharing of idiotypic specificities between antibodies and nonspecific immunoglobulins (15, 17, 19, 28). This is easily explained in the context of network concepts, if one assumes that both antibodies and nonspecific immunoglobulins are under the control of a common set of antiidiotypic antibodies. In fact, recent data (16-20, 29-31) suggest that idiotypes are involved in regulatory phenomena and support the concept of a functional idiotypic network (35-38). Therefore, we thought that the injection of bacteria (Mi) into rabbits that had been primed with Ab2 and had synthesized AbS (rabbits III) would perhaps induce the synthesis of antiCHO bearing idiotypes similar to those of Abl and Ab3, provided that the immunized rabbits would possess the relevant genetic information for such antibodies. After injections of bacteria (Mi), rabbits III synthesized antibacterial antibodies and, interestingly enough, the bulk of the antibody was directed against the carbohydrate. These antiCHO were purified and will be referred to as Abl'. The antiCHO antibody of rabbit III(2281) showed a precipitation line with Ab2 and furthermore there was a line of identity between Abl and Abl' in Ouchterlony assay with Ab2 in the center well (Fig. 3C). The other antibody, Abl' (5083), did not give any appreciable reaction in this test. The Abl' were then used as unlabeled putative inhibitors of the reaction between radiolabeled Abi and solid Ab2. The results indicate that one of the
two
Abl' (2281) inhibited the reaction much better than did
any anti-CHO except AbI itself (Fig. 1). It must be recalled that even those crossreactive anti-CHO from unrelated rabbits that give a precipitation line with Ab2 were unable to inhibit the reaction. Fig. 5 shows that Abl can inhibit the reaction between iodinated Abl' and solid-phase Ab2. Abl is in fact a better inhibitor than Abl'. Finally, two Ab2 against Abi' were raised in allotype-matched rabbits (referred to as Ab2'). The specificity of these sera was checked by immunodiffusion and radioimmunoassay. They specifically recognized the two Abl', Abl,
and the two Ab3 from rabbits III (Fig. 3F). Fig. 6 shows that AbI (2975) and the two Abl' (2281 and 5083) were able to inhibit the reaction between radiolabeled Abi' (2281) and solid-phase Ab2'. It is clear that both rabbits III synthesized anti-CHO sharing idiotypic specificities with Abl. However, one of the two Abi' (2281) was much more idiotypically similar to Abl than the other Abl' (5083). Abl (2975) and Abl' (2281) were compared by isoelectric focusing (not shown). AbI and Abl' are both of restricted heterogeneity but showed nonidentical isoelectric spectra. One further point is worthing mentioning. All rabbits used in this study were heterozygous at the a locus: all rabbits were phenotypically a2a3. The anti-CHO Abl idiotype used to obtain Ab2 was mostly a2 (85% of radiolabeled Abl bound to a solid-phase anti-a2 serum). It is striking to note that Abi' (2281), which bore idiotypic specificities similar to those of Abl, was mainly or only of the a2 allotype (85% of Abl' bound to anti-a2 serum). Therefore, the similar idiotypes detected in AbI and Abl' may be linked to the a2 allotypes. A strong linkage between a allotypes and idiotypes has already been described (12).
Immunology: Urbain et al.
Proc. Natl. Acad. Sci. USA 74 (1977)
5129
0
500
75-
II 0.1
1 Inhibitor, sAg
10
u 100 Inhibitor, Mg
FIG. 4. Inhibition of the binding of I125-labeled Fab'2 fragments of anti-CHO from rabbit I (Abl) to homologous Ab2 2238. Inhibitors are unlabeled, specifically purified, anti-CHO from rabbit I (@); whole antisera to Ab2 were added in amounts ranging from 0.25 to 250 Ml; o antiserum Ab3 (5083), A, antiserum Ab3 (2281). Symbols at right top represent inhibition values obtained with specifically purified antibacterial antibodies from four different rabbits (0) and preimmune sera from rabbits 2238 and 2281 (0).
We draw the following conclusions: (i) it is possible to induce anti-anti-idiotypic antibodies (Ab3); (i) rabbits that have been primed by Ab2 and subsequently injected with the antigen synthesize specific antibodies idiotypically crossreactive with AbI (Abl and Abl' are strikingly similar because they recognize
250
C50 _
75
-
j.
~I
0.1
1
Inhibitor, Mg
l 10
100
FIG. 5. Inhibition of the binding of radiolabeled anti-CHO from rabbit III(2281) (Abl') to Ab2 2238. Insolubilized antiserum 2238 is capable of binding 85% of radiolabeled Ab3'. Inhibitors are unlabeled, specifically purified anti-CHO from rabbits III(2281) (Abl', A) and I (Abl, 0). Symbols at top right give inhibition values obtained with specifically purified antibacterial antibodies from four different rabbits (0) and preimmune sera from rabbits 2238 and 2281 (0). One antibody (0) gave a precipitation line with antiserum 2238 as shown in Fig. 3 (rabbit 2972).
FIG. 6. Inhibition of the binding of radiolabeled Fab fragments of anti-CHO from rabbit 2281 (Abl') to Ab2' (2729). Competitors were unlabeled, specifically purified, anti-CHO from rabbits I (Abl, 0), III(2281) (Abl', A) and III(5083) (Abl' 0). Symbols at top right represent inhibition values obtained with specifically purified antibacterial antibody from six different rabbits (0). Preimmune sera from rabbits I(2975), III(2281), III(5083), and IV(2729) also were tested (0). Insolubilized antiserum 2729 is capable of binding 82% of radiolabeled Abl'.
the same antigen and posses similar idiotypic specificities); and (tit) Abl and AbS might share idiotypic specificities. It could be argued that AbS (2281) and AbS (5083) have different idiotypes but similar combining sites which recognize idiotypes on Ab2. However, in that case, one would assume that all Ab2 and Ab2' have similar or identical idiotypes, which are recognized by the combining sites of Ab3. This seems unlikely and the data can be most simply explained by stating that Abi, Abl', and AbS are sharing idiotypic specificities. This is also more likely because, in several cases, antibodies and nonspecific immunoglobulins systhesized in the same rabbit sharing idiotypes have been described. AbS and Abl' could be said to be equivalent to antibodies and nonspecific immunoglobulins synthesized in the same animal and bearing very similar idiotypes. The final proof of a sharing of idiotypic specificities between AbI and AbS will depend on studies with anti-idiotypic antibodies raised against Ab3. Because heterologous idiotypic crossreactions are rare and only partial (5, 8), it seems likely that rabbit III, if it had been injected only with bacteria, would have synthesized antibacterial antibodies with idiotypic specificities unrelated to those of AbI (rabbit 2975). Priming with Ab2 (2238) strikingly influenced the available repertoire of rabbits III and favored the emergence of anti-CHO bearing idiotypes similar to those of AbI and of Ab3. In other words, the response of an animal to an antigen is not only due to antigenic selection but also depends on the previous idiotypic history of the animal. It seems therefore that idiotypes are involved in regulatory phenomena and that the immune system can be viewed as an idiotypic functional network (29). To a certain extent, the experiments reported here, for an outbred strain, are analogous to those of Eichmann and Rajewski (38) who used an inbred strain of mice. A/J mice immunized with streptococcus A synthesized specific antibodies of which a major proportion bore the A5A idiotype. C57BL
5130
Immunology:
Urbain et al.
mice, immunized with the same antigen, expressed only as a minor component a crossreactive idiotype. Pretreatment of C57BL mice with guinea pig anti-idiotypic antiserum of the IgG1 class, followed by boosting with streptococcus A, was able to specifically expand this minor component in all mice. However, in those experiments (38), the idiotype whose expression was favored by anti-idiotypic antibodies was known to occur during a normal immunization. In previous experiments in our laboratory (17), irradiated rabbits grafted with allogeneic lymphoid cells from a rabbit hyperimmunized with tobacco mosaic virus, synthesized, after immune recovery and antigen injection, antibodies allotypically of recipient origin but idiotypically crossreactive with donor antibodies. Again, in some way, the presence of donor memory cells favored the emergence and selection of host B cells bearing receptors idiotypically crossreactive among all the B cells that develop during radiation recovery. Likewise, in the experiments reported here, priming with Ab2 favored the emergence and selection of clones whose receptors have idiotypes similar to those of Abl. Cazenave (39) independently obtained similar results using an entirely different antigenic system, the RNase system. Taken together, our results and those of Cazenave show that a large proportion of rabbits belonging to the same phenotype (group a) possess a closely related idiotypic repertoire, even though these rabbits express antibodies with different idiotypes when injected with the same antigen. The immune repertoire, effectively expressed in a rabbit, is not only a small part of the total species repertoire but is also only a small part of its own immune repertoire. Therefore, suppression must be dominant in the regulation of the immune system and idiotypy must stem from complex regulatory mechanisms of activation and suppression (19). We are indebted to Prof. J. Brachet, R. Jeener, J. Oudin, N. Jerne, and S. Rodkey for helpful discussions and comments on the manuscript. We thank our colleagues B. Mariame, 0. Leo, and C. Wuilmart for their criticisms and suggestions. The expert technical assistance of Mrs. G. Simon is gratefully acknowledged. The help of A. Rouvroy has been useful. This laboratory is supported by a contract between Universite Libre de Bruxelles and Euratom and by a grant from the Caisse Generale d'Epargne et de Retraite. J.U. is an established investigator (chercheur qualifie) of the Belgian Fonds National de la Recherche
Scientifique.
1. Oudin, J. & Michel, M. (1963) C. R. Hebd. Seances Acad. Sci. 257, 805-808. 2. Oudin, J. & Michel, M. (1969) J. Exp. Med. 130,595-617. 3. Oudin, J. & Michel, M. (1969) J. Exp. Med. 130,619-642. 4. Oudin, J. (1974) Ann. Immunol. (Paris) 125C, 309-337.
Proc. Natl. Acad. Sci. USA 74 (1977) 5. 6. 7. 8. 9.
Oudin, J. (1974) The Antigens 2, 277-374. Kelus, A. S. & Gell, P. G. F. (1968) J. Exp. Med. 127,215-234. Capra, J. D. & Kehoe, M. S. (1975) Adv. Immunol. 20, 1-40. Bordenave, G. (1973) Eur. J. Immunol. 3, 718-726. Eichmann, K. & Kindt, T. J. (1971) J. Exp. Med. 134, 531-
552. 10. Kuettner, M. 11. 12.
13. 14.
15. 16.
G., Wang, A. & Nisonoff, A. (1972) J. Exp. Med. 135,579-595. Eichmann, K. (1975) Immunogenetics, 2, 491-506. Sogn, S. A., Coligan, J. E. & Kindt, T. J. (1977) Fed. Proc. 36, 214-220. Cazenave, P. A. & Oudin, J. (1973) C. R. Hebd. Seances Acad. Sci. Se'r. D. 276,243-245. Thunberg, A. L. & Kindt, T. J. (1974) Eur. J. Immunol. 4, 478-483. Oudin, J. & Cazenave, P. A. (1971) Proc. Natl. Acad. Sci. USA 68,2616-2620. Urbain, J., Tasiaux, N., Leuwenkroon, R., Van Acker, A. & Mariame, B. (1975) Eur. J. Immunol. 5,570-575. Urbain, J. (1976) Ann. Immunol. (Paris) 127C, 357-374.
17. 18. Mariamei, B., Leo, O., Tasiaux, N., Urbain, J., Brezin, C. & Cazenave, P. A. (1977) Ann. Immunol. (Paris) 128C, 355-359. 19. Urbain, J. (1977) Ann. Immunol. (Paris) 128C, 445-455. 20. Aasted, B. & Kindt, T. J. (1976) Eur. J. Immunol. 6,727-732. 21. Wikler, M. (1975) Z. Immunol. Forsch. 144, 193-200. 22. Daugharty, H., Hopper, J. E., MacDonald, A. B. & Nisonoff, A. (1969). J. Exp. Med. 130, 1047-1062. 23. Hunter, R. (1970) Proc. Soc. Exp. Biol. Med. 133,989-992. 24. Avrameas, S. & Ternynck, T. (1967) J. Biol. Chem. 242, 1651-1659. 25. Tosi, R. M. (1973) Contemp. Top. Molecular Immunology, 79-98. 26. Urbain-Vansanten, G. (1970) Immunology 19,783-797. 27. Antoine, J. C. & Avrameas, S. (1976) Immunology 30, 537548. 28. Cazenave, P. A., Ternynck, T. & Avrameas, S. (1974) Proc. Natl. Acad. Sci. USA 71, 4500-4502. 29. Jerne, N. K. (1974) Ann. Immunol. (Paris) 125C, 373-389. 30. Jerne, N. K. (1974) "Clonal selection in a lymphocyte network" in Cellular Selection and Regulation in the Immune Response, ed. Edelman, G. M. (Raven Press, New York), pp. 39-48. 31. Jerne, N. K. (1976) Harvey Lect. 70, 93-110. 32. Lindenmann, J. (1973) Ann. Immunol. (Paris) 124C, 171184. 33. Hofmann, G. W. (1975) Eur. J. Immunol. 5,638-647. 34. Richter, P. H. (1975) Eur. J. Immunol. 5,350-354. 35. Rodkey, L. S. (1974) J. Exp. Med. 139,712-720. 36. Kluskens, L. & KUhler, H. (1974) Proc. Natl. Acad. Sci. USA 71, 5083-5087. 37. Cosenza, H. (1976) Eur. J. Immunol. 6, 114-116. 38. Eichmann, K. & Rajewsky, K. (1975) Eur. J. Immunol. 5, 661-666. 39. Cazenave, P.-A. (1977) Proc. Natl. Acad. Sci. USA 74, 51225125.
Vol. 74, No. 11, pp. 5126-5130, November 1977 Immunology
Idiotypic regulation of the immune system by the induction of antibodies against anti-idiotypic antibodies (immunoregulation/idiotypic networks/anti-anti-idiotypic antibodies)
J. URBAIN, M. WIKLER, J. D. FRANSSEN, AND C. COLLIGNON Laboratory of Animal Physiology, Universit6 Libre de Bruxelles, 67, rue des Chevaux, 1640, Rhode-St-Genese, Belgique
Communicated by J. Brachet, August 24, 1977 ABSTRACT Anticarbohydrate antibodies (Abl) were isolated from a rabbit hyperimmunized with Micrococcus lysodeikticus and injected into allotype-matched rabbits in order
imagine that AbI and AbS could be idiotypically similar because both should be complementary to the idiotopes of Ab2. The experiments described in this paper were undertaken to answer the following questions: (i) is it possible to induce Ab3? (ii) if so, what are the properties of AbW, is AbS carrying idiotypic specificities similar to those of Abl, and does Ab3 react with the same antigen as Abl? and (iii) does immunization with Ab2 influence the immune repertoire expressed after injection with the original antigen?
to obtain specific anti-iodiotypic antibodies (Ab2). Ab2 was isolated by means of a Sepharose column coupled to the anticarbohydrate antibodies and was injected into two allotypematched rabbits. These latter rabbits produced specific antianti-idiotypic antibodies (Ab3) probably sharing idiotypic specificities with Abl. However, these Ab3 did not react with the antigenic carbohydrate moiety of bacteria. The two rabbits that had produced Ab3 were then immunized with M. lysodeikticus and synthesized anticarbohydrate antibodies (Abl') bearing idiotypic specificities similar to those of Abl. The immune repertoire which is effectively expressed in one individual depends not only on the antigenic stimulation but also on the previous idiotypic history of the individual. These data support the concept that the immune system is a functional idiotypic network.
MATERIALS AND METHODS Immunizations. Anticarbohydrate antibodies (anti-CHO) were obtained in rabbits, injected intravenously three times a week with Micrococcus lysodeikticus [bacteria (Mi)] during 10 weeks. Immunoadsorbents were prepared with lysozymedigested bacteria (Mi) cell walls coupled to Sepharose (21). The main fraction of anti-CHO antibodies from rabbit 2975 was eluted from the immunoadsorbent with 5% glucose in phosphate-buffered 0.15 M NaCl (Pi/NaCl), pH 7.5. Ab2 anti-idiotypic antisera were raised in allotype-matched rabbits. Allotypes tested were al, a2, a3, b4, b5, b6, b9, c7, c21, e14, dlI, and d12. Purified anti-CHO antibodies from rabbit 2975 (Abl) were polymerized with glutaraldehyde as described (22) and injected into six allotype-matched rabbits. Injections of 1 mg of AbI emulsified in complete Freund's adjuvant twice a week during 1 month was followed by the injection of 1 mg of AbI intravenously. Abl (1 mg) in incomplete Freund's adjuvant was then injected once a week. Ab2 from rabbit II (2238) were purified by means of a Sepharose column to which 200 mg of Ab had been coupled. Twenty milliliters of serum 2238 (Ab2) was passed through the column. After extensive washing, bound proteins were eluted with glycine-HCI, pH 2.2, and their idiotypic nature was checked by immunodiffusion and radioimmunoassay. The purified Ab2 were crosslinked with glutaraldehyde, emulsified with Freund's adjuvant, and injected into two allotype-matched rabbits [111(2281) and 111(5083)] according to the schedule used to elicit Ab2. After a rest period, rabbits III were immunized with bacteria (Mi). Anti-CHO antibodies from rabbits III (Abl') were isolated and injected into allotype-matched rabbits to elicit anti-idiotypic antibodies (Ab2') specific for Abl'. The experimental schedule used was the same as that used to elicit Ab2. The experimental scheme is depicted in Fig 1. Radioimmunoassay. Iodination was performed according to Hunter (23). The specific radioactivity of immunoglobulins
As a rule, in response to the same antigen, different individuals from the same species express different subsets of the total antibody repertoire of the species, with each subset bearing different idiotypic specificities (1-7) (these antibodies will be referred to as Abl). Idiotypic similarities were observed only in certain idiotypic systems (5, 8) and were only partial with two main kinds of exceptions: (i) some rabbit families or some inbred strains of mice immunized with certain antigens synthesize antibodies with very similar idiotypic specificities (9-11); and (ii) very similar idiotypic specificities are often found among different antibody subpopulations present in the serum of a single individual immunized with a single antigen
(12-20). The V domains of anti-idiotypic antibodies (which will be called Ab2) might themselves function as antigens and therefore elicit the synthesis of anti-anti-idiotypic antibodies (which will be referred to as AbS). These AbS might elicit the synthesis of Ab4, and so on. If the experimental rules that have been found at the Abi level (the AbI of different individuals from the same species express different idiotypic specificities) also apply to the Ab2 and AbS levels, different individuals injected with the same AbI might synthesize different Ab2 characterized by different idiotypic specificities and so on. This line of reasoning leads to a rapidly increasing diversity, and Ab4 and Ab5 will already encompass a large part of the immune repertoire. This is not necessarily so. The idiotypic diversity at the Ab2 level could be much more restricted than at the AbI level. In other words, anti-idiotypic antibodies synthesized by different animals in response to the same Abl may frequently express the same idiotype. On the other hand, if there is a close relationship between idiotypic specificities and active sites, it is possible to
Abbreviations: Abl, antibodies directed against bacteria (Mi); Ab2, anti-idiotypic antibodies; Ab3, anti-anti-idiotypic antibodies; Abl', antibacteria antibodies in rabbits that have previously synthesized Ab3; Ab2', anti-idiotypic antibodies against Abl'; anti-CHO, anti-carbohydrate antibodies; bacteria (Mi), Micrococcus Iysodeiktcus; Pi/NaCl, phosphate-buffered (pH 7.5) saline.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.
5126
Proc. Natl. Acad. Sci. USA 74 (1977)
Immunology: Urbain et al.
5127
Anti-CHO Abl (2975) Anti-Abl Ab2 (2238) Anti-Ab2 Ab3 (2281) Anti-bacteriia (Mi) Abl' (22'81)
Anti-Abl Ab2 (2239,2240,2716, 520)3,5271)
Anti-Ab2 Ab3 (5083) Anti-bacteria (Mi) Abl' (5083)
c 0
._ c
doC
Anti-Abl' Anti-Abl' Ab2' (2729) Ab2' (2722) FIG. 1. Immunization scheme of rabbits I, II, and III.
ranged between 106 and 5 X 106 cpm/tig. Binding curves of radiolabeled antibody were performed with 25 ng of antibody and increasing amounts of crosslinked anti-idiotypic sera. Crosslinking of antisera was performed with ethyl chloroformate (24). All dilutions were done in phosphate buffer containing 1% ovalbumin. The tubes were incubated overnight at room temperature and centrifuged at 13,000 rpm for 5 min. After centrifugation, the top half of the tube content was transferred to another tube and radioactivity in both tubes was assayed in a gamma scintillation counter (Beckman). Control tubes without solid antiserum were included in each experiment. Inhibition curves were performed as follows (25): a fixed amount of crosslinked antiserum, corresponding to 70% of the plateau (binding), the cold putative inhibitor in variable amounts, and a fixed amount of labeled antibody were mixed and the tubes were processed as described above. RESULTS AND DISCUSSION The M. lysodeikticus system is characterized by two major antigenic specificities, a glucose N-acetylaminomannuronic acid polymer and a peptidoglycan moiety (21). Anti-CHO antibodies were isolated from rabbit 2975 hyperimmunized with bacteria (Mi). This rabbit will be referred to as rabbit I and the purified anti-CHO antibody from this rabbit will be called Abl. Purified AbI were injected into six rabbits of matched allotypes to elicit the synthesis of Ab2. These rabbits will be designated 11(2238), II(2239), 11(2240), II(2716), 11(5203), and II(5271). The precipitating antisera obtained reacted with Abl in Ouchterlony immunodiffusion but not with the preimmune serum of rabbit I. These anti-idiotypic sera were tested in immunodiffusion with a panel of sera from 60 rabbits hyperimmunized with the same antigen. Only 1 of the 60 sera (rabbit 2972) showed a precipitation line with the different Ab2. This frequency of idiotypic heterologous crossreactions is of the same order of magnitude as has been found in other antigenic systems (5, 8, 16). The specificity of the Ab2 sera was further investigated by radioimmunoassay methods. Except for Abi' (see below), only the unlabeled homologous Abl is an efficient inhibitor of the reaction between radiolabeled Abl and insolubilized Ab2 sera (Fig. 2). The controls shown in the same figure demonstrate the idiotypic nature of the reaction. Even the anti-CHO antibody from rabbit 2972, which gave a precipitation line with Ab2, was unable to inhibit the reaction. The specificity of purified Ab2 from rabbit II(2238) was checked by immunodiffusion and radioimmunoassay (Fig. 3D). These purified Ab2 were capable of 85% binding to a solidphase anti-Ig serum.
Inhibitor, ,g
FIG. 2. Inhibition of the binding of radiolabeled antibody from rabbit I (Abl) to crosslinked homologous anti-idiotypic antiserum Ab2 (2238). Inhibitors are unlabeled, specifically purified, anti-CHO antibodies from rabbit I (Abl, 0) and rabbit III(2281) (Abi', A). Anti-idiotypic antiserum 2238 (M) was added in amounts ranging from 0.25 to 250 ,ul of whole serum. Symbols (O) at right top give inhibition values obtained with specifically purified antibacterial antibodies from seven different rabbits. Controls (0) (with preimmune sera from rabbits II and III and heterologous anti-idiotypic antiserum) are also shown. Antibody from rabbit 2972 is the only antibody (out of 60) that reacted with Ab2 (2238).
In order to elicit Ab3, Ab2 were injected into rabbits III(2281) and III(5083). Ten weeks after the start of immunization, sera from both rabbits showed a precipitation line with the immunizing Ab2 by Ouchterlony assay (Fig. 3 A-D). These AbS did not show a precipitation line with either the preimmune serum of rabbit 11(2238) and or with Ab2 directed against other anti-bacteria (Mi) antibodies or sera from other rabbits hyperimmunized with bacteria (Mi). A panel of 10 Ab2 and 12 immune sera was tested. The frequency of crossreactions at the AbS level did not seem to be larger than at the Ab2 level. It must be stressed that the two Ab3 from rabbits 11(2281) and III(5083) specifically recognized not only the immunizing Ab2 but also the other Ab2 specific for Abl (2975) (Fig. 3E). Unless one assumes that all the Ab2 raised by the same Abi share similar idiotopes recognized by the combining sites of both Ab3, these data seem to suggest that Abl and both AbS possess similar idiotypic specificities recognized by the combining sites of Ab2. The data of Fig. 4 indicate that both Ab3 inhibit the reaction between radiolabeled Fab'2 Abi and solid-phase Ab2 (2238). The two Ab3 were repeatedly passed through columns of Sepharose coupled with the carbohydrate of bacteria (Mi). This treatment did not remove more than 10% of the inhibitory activity of AbM. Therefore, the bulk of Ab3 did not bear a combining site with a sufficient affinity to recognize the carbohydrate determinants of bacteria (Mi). At this stage, different interpretations were possible: (i) the results of inhibition curves could be due to steric inhibitions because Abl and AbS are recognizing the same Ab2; and (ii) Abl and both AbS bear similar idiotypic specificities. The Ab3 of both rabbits would be an imperfect image of Abl because the Ab3 sera have no anticarbohydrate affinity. To obtain further information, the rabbits that had synthesized Ab3 were immunized with bacteria (Mi). The rationale
5128
Proc. Natl. Acad. Sci. USA 74 (1977)
Immunology: Urbain et al. B
A
antiserum Abi'
adsAbl' Ab 2972
Ab 1 0
A
Abl 0
0
0
B
Ab3 5083 0
0
Ab2
Ab2
0
Ab2978
Abi' *
0
Abl'2281
Ab1' 5083
0
0
Ab3 2281 0
Ab3 0 D
C
Ab1 0
Abl'
Ab3 *
.
Ab 2972
Abl 0
0
D
Ab3 5083 0
Ab2
Ab2 At)2978 * antiserum Atb1I
Ads Abl
Abl'2281 0
0
Ab 15083 0 Ab3 2281 0
F
E
Ab 1' 2281
Ab2' 2722
0
AD2 2238
Ab2 2239 Abi
0
0
Ab3
Ab3 2281
0
Ab 2
0
Abs' 2729
Ab2 5203 A 3 5083
Abi
Ab2 5271
Abi' 5083
0
0
FIG. 3. Ouchterlony assays of immune sera used. The center well in A, B, and C was filled with Ab2 (2238). The center well of D contained purified Ab2 which was used for immunizations. Wells marked "Abl" or "Abl"' were filled with anti-CHO purified by affinity chromatography. Wells marked "ads Abl"' contained the Abl' from rabbit 2281, which had been adsorbed with bacteria. Rabbit 2972 had been hyperimmunized with bacteria and gave the only serum (out of 60) that reacted with Ab2. Ab 2978 was a randomly chosen anti-CHO. The center well in E was filled with Ab3 (2281). The center well in F contained Ab2' (2729) raised against Abl' (2281).
behind the experiment was as follows. It has been shown by several laboratories (12-20) that it is not infrequent to find strong idiotypic similarities between different antibody subpopulations synthesized by a single rabbit. Furthermore, in many cases, antigen seems to induce not only specific antibodies but also the so-called nonspecific immunoglobulins (26, 27). Several workers have reported a sharing of idiotypic specificities between antibodies and nonspecific immunoglobulins (15, 17, 19, 28). This is easily explained in the context of network concepts, if one assumes that both antibodies and nonspecific immunoglobulins are under the control of a common set of antiidiotypic antibodies. In fact, recent data (16-20, 29-31) suggest that idiotypes are involved in regulatory phenomena and support the concept of a functional idiotypic network (35-38). Therefore, we thought that the injection of bacteria (Mi) into rabbits that had been primed with Ab2 and had synthesized AbS (rabbits III) would perhaps induce the synthesis of antiCHO bearing idiotypes similar to those of Abl and Ab3, provided that the immunized rabbits would possess the relevant genetic information for such antibodies. After injections of bacteria (Mi), rabbits III synthesized antibacterial antibodies and, interestingly enough, the bulk of the antibody was directed against the carbohydrate. These antiCHO were purified and will be referred to as Abl'. The antiCHO antibody of rabbit III(2281) showed a precipitation line with Ab2 and furthermore there was a line of identity between Abl and Abl' in Ouchterlony assay with Ab2 in the center well (Fig. 3C). The other antibody, Abl' (5083), did not give any appreciable reaction in this test. The Abl' were then used as unlabeled putative inhibitors of the reaction between radiolabeled Abi and solid Ab2. The results indicate that one of the
two
Abl' (2281) inhibited the reaction much better than did
any anti-CHO except AbI itself (Fig. 1). It must be recalled that even those crossreactive anti-CHO from unrelated rabbits that give a precipitation line with Ab2 were unable to inhibit the reaction. Fig. 5 shows that Abl can inhibit the reaction between iodinated Abl' and solid-phase Ab2. Abl is in fact a better inhibitor than Abl'. Finally, two Ab2 against Abi' were raised in allotype-matched rabbits (referred to as Ab2'). The specificity of these sera was checked by immunodiffusion and radioimmunoassay. They specifically recognized the two Abl', Abl,
and the two Ab3 from rabbits III (Fig. 3F). Fig. 6 shows that AbI (2975) and the two Abl' (2281 and 5083) were able to inhibit the reaction between radiolabeled Abi' (2281) and solid-phase Ab2'. It is clear that both rabbits III synthesized anti-CHO sharing idiotypic specificities with Abl. However, one of the two Abi' (2281) was much more idiotypically similar to Abl than the other Abl' (5083). Abl (2975) and Abl' (2281) were compared by isoelectric focusing (not shown). AbI and Abl' are both of restricted heterogeneity but showed nonidentical isoelectric spectra. One further point is worthing mentioning. All rabbits used in this study were heterozygous at the a locus: all rabbits were phenotypically a2a3. The anti-CHO Abl idiotype used to obtain Ab2 was mostly a2 (85% of radiolabeled Abl bound to a solid-phase anti-a2 serum). It is striking to note that Abi' (2281), which bore idiotypic specificities similar to those of Abl, was mainly or only of the a2 allotype (85% of Abl' bound to anti-a2 serum). Therefore, the similar idiotypes detected in AbI and Abl' may be linked to the a2 allotypes. A strong linkage between a allotypes and idiotypes has already been described (12).
Immunology: Urbain et al.
Proc. Natl. Acad. Sci. USA 74 (1977)
5129
0
500
75-
II 0.1
1 Inhibitor, sAg
10
u 100 Inhibitor, Mg
FIG. 4. Inhibition of the binding of I125-labeled Fab'2 fragments of anti-CHO from rabbit I (Abl) to homologous Ab2 2238. Inhibitors are unlabeled, specifically purified, anti-CHO from rabbit I (@); whole antisera to Ab2 were added in amounts ranging from 0.25 to 250 Ml; o antiserum Ab3 (5083), A, antiserum Ab3 (2281). Symbols at right top represent inhibition values obtained with specifically purified antibacterial antibodies from four different rabbits (0) and preimmune sera from rabbits 2238 and 2281 (0).
We draw the following conclusions: (i) it is possible to induce anti-anti-idiotypic antibodies (Ab3); (i) rabbits that have been primed by Ab2 and subsequently injected with the antigen synthesize specific antibodies idiotypically crossreactive with AbI (Abl and Abl' are strikingly similar because they recognize
250
C50 _
75
-
j.
~I
0.1
1
Inhibitor, Mg
l 10
100
FIG. 5. Inhibition of the binding of radiolabeled anti-CHO from rabbit III(2281) (Abl') to Ab2 2238. Insolubilized antiserum 2238 is capable of binding 85% of radiolabeled Ab3'. Inhibitors are unlabeled, specifically purified anti-CHO from rabbits III(2281) (Abl', A) and I (Abl, 0). Symbols at top right give inhibition values obtained with specifically purified antibacterial antibodies from four different rabbits (0) and preimmune sera from rabbits 2238 and 2281 (0). One antibody (0) gave a precipitation line with antiserum 2238 as shown in Fig. 3 (rabbit 2972).
FIG. 6. Inhibition of the binding of radiolabeled Fab fragments of anti-CHO from rabbit 2281 (Abl') to Ab2' (2729). Competitors were unlabeled, specifically purified, anti-CHO from rabbits I (Abl, 0), III(2281) (Abl', A) and III(5083) (Abl' 0). Symbols at top right represent inhibition values obtained with specifically purified antibacterial antibody from six different rabbits (0). Preimmune sera from rabbits I(2975), III(2281), III(5083), and IV(2729) also were tested (0). Insolubilized antiserum 2729 is capable of binding 82% of radiolabeled Abl'.
the same antigen and posses similar idiotypic specificities); and (tit) Abl and AbS might share idiotypic specificities. It could be argued that AbS (2281) and AbS (5083) have different idiotypes but similar combining sites which recognize idiotypes on Ab2. However, in that case, one would assume that all Ab2 and Ab2' have similar or identical idiotypes, which are recognized by the combining sites of Ab3. This seems unlikely and the data can be most simply explained by stating that Abi, Abl', and AbS are sharing idiotypic specificities. This is also more likely because, in several cases, antibodies and nonspecific immunoglobulins systhesized in the same rabbit sharing idiotypes have been described. AbS and Abl' could be said to be equivalent to antibodies and nonspecific immunoglobulins synthesized in the same animal and bearing very similar idiotypes. The final proof of a sharing of idiotypic specificities between AbI and AbS will depend on studies with anti-idiotypic antibodies raised against Ab3. Because heterologous idiotypic crossreactions are rare and only partial (5, 8), it seems likely that rabbit III, if it had been injected only with bacteria, would have synthesized antibacterial antibodies with idiotypic specificities unrelated to those of AbI (rabbit 2975). Priming with Ab2 (2238) strikingly influenced the available repertoire of rabbits III and favored the emergence of anti-CHO bearing idiotypes similar to those of AbI and of Ab3. In other words, the response of an animal to an antigen is not only due to antigenic selection but also depends on the previous idiotypic history of the animal. It seems therefore that idiotypes are involved in regulatory phenomena and that the immune system can be viewed as an idiotypic functional network (29). To a certain extent, the experiments reported here, for an outbred strain, are analogous to those of Eichmann and Rajewski (38) who used an inbred strain of mice. A/J mice immunized with streptococcus A synthesized specific antibodies of which a major proportion bore the A5A idiotype. C57BL
5130
Immunology:
Urbain et al.
mice, immunized with the same antigen, expressed only as a minor component a crossreactive idiotype. Pretreatment of C57BL mice with guinea pig anti-idiotypic antiserum of the IgG1 class, followed by boosting with streptococcus A, was able to specifically expand this minor component in all mice. However, in those experiments (38), the idiotype whose expression was favored by anti-idiotypic antibodies was known to occur during a normal immunization. In previous experiments in our laboratory (17), irradiated rabbits grafted with allogeneic lymphoid cells from a rabbit hyperimmunized with tobacco mosaic virus, synthesized, after immune recovery and antigen injection, antibodies allotypically of recipient origin but idiotypically crossreactive with donor antibodies. Again, in some way, the presence of donor memory cells favored the emergence and selection of host B cells bearing receptors idiotypically crossreactive among all the B cells that develop during radiation recovery. Likewise, in the experiments reported here, priming with Ab2 favored the emergence and selection of clones whose receptors have idiotypes similar to those of Abl. Cazenave (39) independently obtained similar results using an entirely different antigenic system, the RNase system. Taken together, our results and those of Cazenave show that a large proportion of rabbits belonging to the same phenotype (group a) possess a closely related idiotypic repertoire, even though these rabbits express antibodies with different idiotypes when injected with the same antigen. The immune repertoire, effectively expressed in a rabbit, is not only a small part of the total species repertoire but is also only a small part of its own immune repertoire. Therefore, suppression must be dominant in the regulation of the immune system and idiotypy must stem from complex regulatory mechanisms of activation and suppression (19). We are indebted to Prof. J. Brachet, R. Jeener, J. Oudin, N. Jerne, and S. Rodkey for helpful discussions and comments on the manuscript. We thank our colleagues B. Mariame, 0. Leo, and C. Wuilmart for their criticisms and suggestions. The expert technical assistance of Mrs. G. Simon is gratefully acknowledged. The help of A. Rouvroy has been useful. This laboratory is supported by a contract between Universite Libre de Bruxelles and Euratom and by a grant from the Caisse Generale d'Epargne et de Retraite. J.U. is an established investigator (chercheur qualifie) of the Belgian Fonds National de la Recherche
Scientifique.
1. Oudin, J. & Michel, M. (1963) C. R. Hebd. Seances Acad. Sci. 257, 805-808. 2. Oudin, J. & Michel, M. (1969) J. Exp. Med. 130,595-617. 3. Oudin, J. & Michel, M. (1969) J. Exp. Med. 130,619-642. 4. Oudin, J. (1974) Ann. Immunol. (Paris) 125C, 309-337.
Proc. Natl. Acad. Sci. USA 74 (1977) 5. 6. 7. 8. 9.
Oudin, J. (1974) The Antigens 2, 277-374. Kelus, A. S. & Gell, P. G. F. (1968) J. Exp. Med. 127,215-234. Capra, J. D. & Kehoe, M. S. (1975) Adv. Immunol. 20, 1-40. Bordenave, G. (1973) Eur. J. Immunol. 3, 718-726. Eichmann, K. & Kindt, T. J. (1971) J. Exp. Med. 134, 531-
552. 10. Kuettner, M. 11. 12.
13. 14.
15. 16.
G., Wang, A. & Nisonoff, A. (1972) J. Exp. Med. 135,579-595. Eichmann, K. (1975) Immunogenetics, 2, 491-506. Sogn, S. A., Coligan, J. E. & Kindt, T. J. (1977) Fed. Proc. 36, 214-220. Cazenave, P. A. & Oudin, J. (1973) C. R. Hebd. Seances Acad. Sci. Se'r. D. 276,243-245. Thunberg, A. L. & Kindt, T. J. (1974) Eur. J. Immunol. 4, 478-483. Oudin, J. & Cazenave, P. A. (1971) Proc. Natl. Acad. Sci. USA 68,2616-2620. Urbain, J., Tasiaux, N., Leuwenkroon, R., Van Acker, A. & Mariame, B. (1975) Eur. J. Immunol. 5,570-575. Urbain, J. (1976) Ann. Immunol. (Paris) 127C, 357-374.
17. 18. Mariamei, B., Leo, O., Tasiaux, N., Urbain, J., Brezin, C. & Cazenave, P. A. (1977) Ann. Immunol. (Paris) 128C, 355-359. 19. Urbain, J. (1977) Ann. Immunol. (Paris) 128C, 445-455. 20. Aasted, B. & Kindt, T. J. (1976) Eur. J. Immunol. 6,727-732. 21. Wikler, M. (1975) Z. Immunol. Forsch. 144, 193-200. 22. Daugharty, H., Hopper, J. E., MacDonald, A. B. & Nisonoff, A. (1969). J. Exp. Med. 130, 1047-1062. 23. Hunter, R. (1970) Proc. Soc. Exp. Biol. Med. 133,989-992. 24. Avrameas, S. & Ternynck, T. (1967) J. Biol. Chem. 242, 1651-1659. 25. Tosi, R. M. (1973) Contemp. Top. Molecular Immunology, 79-98. 26. Urbain-Vansanten, G. (1970) Immunology 19,783-797. 27. Antoine, J. C. & Avrameas, S. (1976) Immunology 30, 537548. 28. Cazenave, P. A., Ternynck, T. & Avrameas, S. (1974) Proc. Natl. Acad. Sci. USA 71, 4500-4502. 29. Jerne, N. K. (1974) Ann. Immunol. (Paris) 125C, 373-389. 30. Jerne, N. K. (1974) "Clonal selection in a lymphocyte network" in Cellular Selection and Regulation in the Immune Response, ed. Edelman, G. M. (Raven Press, New York), pp. 39-48. 31. Jerne, N. K. (1976) Harvey Lect. 70, 93-110. 32. Lindenmann, J. (1973) Ann. Immunol. (Paris) 124C, 171184. 33. Hofmann, G. W. (1975) Eur. J. Immunol. 5,638-647. 34. Richter, P. H. (1975) Eur. J. Immunol. 5,350-354. 35. Rodkey, L. S. (1974) J. Exp. Med. 139,712-720. 36. Kluskens, L. & KUhler, H. (1974) Proc. Natl. Acad. Sci. USA 71, 5083-5087. 37. Cosenza, H. (1976) Eur. J. Immunol. 6, 114-116. 38. Eichmann, K. & Rajewsky, K. (1975) Eur. J. Immunol. 5, 661-666. 39. Cazenave, P.-A. (1977) Proc. Natl. Acad. Sci. USA 74, 51225125.
