Transplantation tolerance correlates with high levels of T- and B-lymphocyte activity

Transcription

1 Proc. Nati. Acad. Sci. USA Vol. 86, pp , January 1989 Immunology Transplantation tolerance correlates with high levels of T- and B-lymphocyte activity (neonatal tolerance/graft acceptance or rejection/chimerism/internal activity) A. BANDEIRA*, A. COUTINHO*, C. CARNAUDt, F. JACQUEMART*, AND L. FORNIt *Unitd d'immunobiologie, Institut Pasteur, 75724, Paris, France; tinstitut National de la Santd et de la Recherche Mddicale Unite U25, H6pital Necker, Paris 15, France; and tbasel Institute for Immunology, 45 Basel, Switzerland Communicated by Niels K. Jerne, September 26, 1988 (received for review July 15, 1988) ABSTRACT Mice tolerized (treated to make them tolerant) at birth to transplantation antigens by injection of semiallogeneic cells contain very high numbers of activated T and B lymphocytes in their spleen. Lymphoid hyperactivity correlates with the tolerant state: it is present only in animals accepting skin aliografts. Tolerized mice that reject the allogeneic skin graft have approximately the same numbers of total and activated lymphocytes as normal mice. The high level of lymphocyte activation in tolerant mice persists for up to 1 year of age, although it declines with age, and is markedly increased by a secondary aflograft. The magnitudes of both primary and secondary tolerant responses are significantly higher than the immunological response of a normal mouse rejecting the same type of aliograft. These observations contradict concepts of clonal deletion or anergy as the basis of neonatally induced transplantation tolerance and may contribute additional approaches to experimentation and control of transplantation reactions. Given that self-tolerance is ontogenically learned (1), the original observations on neonatally induced transplantation tolerance (2) have generally been interpreted according to Burnet's postulates of specific deletion ofautoreactive clones (3). Later, evidence for active suppression of persistent alloreactive lymphocytes in tolerant animals (4-6) has provided an alternative, yet clonal, interpretation of tolerance. Clonal deletion, achieved by elimination of specific cells (7-9), or clonal anergy after down-regulation of molecules that are necessary for the functional performance of tolerant cells (1) has been reported and suggested to represent the prototype mechanism of natural self-tolerance. Clearly, however, induced tolerance, even if neonatal, might follow different rules, particularly because the above processes are believed to occur intrathymically whereas neonatal transplantation tolerance is likely to be achieved in the periphery. For experimental systems that aim at intervening in organ transplantation, the latter type of tolerance is critical. We report here that adult mice tolerized (treated to make them tolerant) at birth to transplantation antigens maintain a high level of immune activity (involving both host and donor lymphocytes) that strictly correlates with the acceptance of an allogeneic skin graft. This observation of hyperactivity in tolerant immune systems is less compatible with theories of clonal anergy or suppression, even if suppression is seen as a dynamic state (11), than with models considering a positive definition of self, where tolerance shares many of the cellular characteristics associated with immunity. MATERIALS AND METHODS Mice. CBA/Ca, CBA/J, and C57BL/6J mice were bred in the animal facilities of the Institut Pasteur, Paris. The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C solely to indicate this fact. Antibodies. The following monoclonal antibodies were used: anti-i-ab B (12), anti-kb S (13) anti-l3t4 (mouse CD4) H (14), anti-lyt-2 (mouse CD8) (15), and anti-rat K MARK.1 (16). All antibodies were purified from culture supernatants or from ascites fluid by DEAE-cellulose chromatography (Servacell, Serva, Heidelberg) or by affinity chromatography on protein A-Sepharose CL-4B (Pharmacia). Sheep anti-mouse total immunoglobulin IgGs were affinity-purified on Sepharose-coupled mouse IgG. Rabbit anti-mouse u chain antibodies were raised by immunization with the myeloma protein MOPC-14-E (,u and A chain) and affinity-purified on Sepharose-coupled myeloma protein TEPC-183 (A and K chain). Induction of Neonatal Tolerance. CBA/Ca or CBA/J mice were injected intravenously within 24 hr of birth with 2-25 x 16 spleen cells from adult (C57BL/6/J x CBA/Ca)F1 or (C57BL/6J x CBA/J)F1 mice. At 2 months of age, groups of animals were grafted simultaneously with syngeneic (CBA/ Ca or CBA/J) and allogeneic (C57BL/6J) skin. About 8-85% of the mice accepted both grafts and were considered tolerant; they will be termed TA (tolerized, graft accepted). The animals that rejected the allogeneic skin graft were considered nontolerant and will be termed TR (tolerized, graft rejected). Groups of four to six animals were sacrificed 2 or 5 weeks after the graft. A group of tolerant mice was kept for 8 months and challenged with a secondary allograft. Immunofluorescence. All monoclonal antibodies were coupled with biotin (17) and detected with fluorescein isothiocyanate (FITC)-labeled avidin (Sigma). Anti-total immunoglobulins and anti-s antibodies were directly labeled with FITC as described (18). For double-staining experiments, biotin-labeled anti-kb antibodies were detected with R- PE-streptavidin (Becton Dickinson). Anti-CD4 and anti-cd8 antibodies were either directly labeled with FITC or used unlabeled and detected by FITC-coupled mouse anti-rat K antibody MARK.1 (16). Spleen cell suspensions were stained on ice as described (18). Propidium iodide (5 I&g/ml) was added to the cell suspensions prior to analysis to exclude dead cells. Flow Cytometry. Stained cell suspensions were analyzed with the flow cytometers FACS analyzer I or FACScan (Becton Dickinson) for frequency of cells positive for the various markers. Then each population was gated and analyzed for either volume distribution or forward light scatter to assess the frequency of activated cells. Total numbers of activated cells were calculated on the basis of the flow cytometry data and of the total number of spleen cells. Assay for Immunoglobulin-Secreting Cells. Immunoglobulin-secreting cells were detected as plaque-forming cells (PFC) in the protein A plaque assay (19), using class-specific rabbit antibodies produced and tested as described (2). Abbreviations: FITC, fluorescein isothiocyanate; PFC, plaqueforming cell; MHC, major histocompatibility complex. To whom reprint requests should be addressed. 272

2 RESULTS Tolerant Mice Contain Very High Numbers of Activated Lymphocytes in Their Spleens. Fig. 1 shows the numbers of total and activated cells in the spleens of four to six mice in each experimental group. Neonatally tolerized mice, tested at 2-3 months of age, before any skin graft showed splenomegaly resulting from 3- to 4-fold higher numbers of lymphoid cells in the spleen. This was also true for tolerized animals accepting the allogeneic skin graft. The number of activated blast cells in the spleens of animals in these two groups was also higher, reaching values 5- to 1-fold higher than in controls (either nonmanipulated or accepting a syngeneic skin graft). It is thus clear that splenic lymphocyte activation precedes grafting. Interestingly, although normal mice that rejected an allogeneic graft also showed increased splenic lymphoid activity (normal/allogeneic skin), this response was transient and quantitatively less marked than the response of tolerant animals that accepted their grafts. Some of the neonatally tolerized mice (-2%) rejected the allogeneic graft. Analysis of these animals revealed surprisingly low splenic lymphoid activity, comparable to that of untreated control animals or of animals accepting a syngeneic graft. They thus differed from both tolerant mice and normal mice responding to a primary allogeneic graft. As shown below, their lymphoid activity resembles that of normal primed animals after a second specific challenge. We conclude, therefore, that acceptance of allogeneic tissues by tolerant individuals is associated with a high level of lymphoid activity, even higher than that required for rejection in normal mice. Lymphocyte Hyperactivity in Tolerant Animals Involves B, CD4+, and CD8+ Cells. Spleen cells from the same groups of animals were analyzed for activation of each major lymphocyte class, namely immunoglobulin-positive, CD4', and CD8' cells. As shown in Fig. 2, immune hyperactivity in tolerant animals was readily detected in all lymphocyte subsets. Again, there is good correlation between enhanced activity and state of tolerance, since nontolerant mice showed nearly normal levels of B and T blasts. All lympho Immunology: Bandeira et A FIG. 1. Average numbers (+ 1 SD) of total cells (open bars) and activated blasts (solid bars) in the spleen of CBA/J mice tolerized at birth with semiallogenic (CBA/J x C57BL/6)F1 spleen cells. Numbers of blast cells were calculated on the basis of forward light scattering analysis. Cells from tolerized (bars 1-5) and normal (bars 6-9) animals are shown. Pairs of bars: 1, tolerized animals without a graft (TO); 2 and 3, tolerized animals grafted with allogeneic (C57BL/6) skin, graft accepted at 2 and 5 weeks (TA2w and TA5w, respectively); 4 and 5, tolerized animals grafted with allogeneic (C57BL/6) skin, graft rejected at 2 and 5 weeks (TR2w and TR5w, respectively); 6, control mice, not grafted; 7, normal mice grafted with syngeneic (CBA/J) skin, graft accepted at 2 weeks (Syng.2w); 8 and 9, normal mice, grafted with allogeneic (C57BL/6) skin, graft rejected at 2 and 5 weeks (Allo.2w and Allo.5w, respectively). 1 Proc. Natl. Acad. Sci. USA 86 (1989) 273 o -15 ~8-4- CD4+ ~~~~~~~~~Ig~ FIG. 2. Average numbers (+ 1 SD) of activated T and B blasts, calculated on the basis of the forward light scattering distribution of the cell populations bearing CD8, CD4, or immunoglobulin (Ig), their frequency, and the total spleen cell numbers. Bars are numbered as in Fig. 1. cyte subsets are also activated in normal animals rejecting an allograft. Interestingly, some qualitative differences exist between tolerant and normal mice challenged with an allograft at 2 weeks (compare bars 2 and 8, respectively, in Fig. 2), since proportionally more B cells were activated in the former resulting in a higher B/T-cell ratio (3.4 vs. 2.4). The hyperactivity of lymphocytes is not limited to blast transformation and it includes greatly enhanced effector activities. Thus, the number of splenic immunoglobulinsecreting PFCs in these animals was severalfold higher than in controls (Table 1). Since the isotypic pattern of the PFC response in tolerized animals was clearly T-cell dependent, many of the CD4' blasts must also be effector cells. Once again, nontolerant mice were different from both normal and tolerant animals in their response to a primary graft, as they hardly differed from normal unmanipulated controls. Lymphocyte Hyperactivity in Tolerant Animals Involves Both Host and Donor Lymphocytes. Previous observations in this experimental system have led to the proposal that chimerism (i.e., persistence of donor "tolerogenic" cells) is required to maintain the state of tolerance (21). In view of the present observations relating lymphocyte hyperactivity to tolerance, we investigated both the chimerism and the state of activation of donor cells in the two groups of animals tolerized at birth and accepting or rejecting the skin allograft. As shown in Table 2, donor cells were detected in both nontolerant and tolerant mice, although the frequency was 2-3 times lower in the nontolerant group than in the tolerant group. In addition, there was a >1-fold difference in the total numbers of activated chimeric (and host) cells between tolerant and nontolerant individuals (Fig. 3). The analysis of B and T cells in the donor population shows that tolerant mice contain proportionally more B and CD4' cells than nontolerant mice. Furthermore, the comparison of the state of activation of B, CD4', and CD8' in the host and donor compartment (Table 3) provides further indications that major differences between tolerant and nontolerant immune systems are found at the level of activation of all three lymphocyte classes of donor origin. Lymphocyte Activity in Tolerant Animals Shows Anamnestic Recall After a Second Graft. The results above indicate that tolerance is a pathway of immune responsiveness that may

3 274 Immunology: Bandeira et al. Proc. Natl. Acad. Sci. USA 86 (1989) Table 1. Immunoglobulin-secreting cells in the spleen of tolerant and normal mice challenged with an allograft Type of Type of Time after PFCs, no. X 13 per spleen mouse graft graft Response IgM IgG3 IgG1 IgG2b IgG2a TO ± ± 6 69 ± TA Allo. 2 weeks Accepted 114 ± ± ± ± ± 177 TA Allo. S weeks Accepted 1258 ± ± ± ± ± 227 TR Allo. 2 weeks Rejected ± 1 5 ± 1 3 ± 2 39± 1 TR Allo. 5 weeks Rejected ± 7 11 ± ± 2 N ± 2 5 ± N Syng.* 2 weeks Accepted N Allo. 2 weeks Rejected 685 ±39 9 ± 7 49 ± 17 16± ± 68 N Allo. S weeks Rejected ± 1 13 ± ± ± 15 TO, tolerized animal; TA, tolerized animals, graft accepted; TR, tolerized animals, graft rejected; N, normal animal; allo., allogeneic; syng., syngeneic. Data are presented as mean ± SD. *Only one mouse in this group was tested for this parameter. share other characteristics, such as memory, with conventional immunity leading to the elimination of the antigen. We assessed this possibility in tolerant or normal mice that had, respectively, accepted or rejected a first graft 8 months before by rechallenging with a second specific allograft. The lymphocyte activity in the spleens of these animals was analyzed as above (Fig. 4). At 8 months, the spleens of tolerant primed mice before a second graft contained roughly twice as many activated B and T cells as normal primed animals, simply because of the larger spleens of tolerant mice; the frequency of activated cells was the same in tolerant and normal primed mice. The animals in this set of experiments were considerably older than those analyzed above (1-12 months vs weeks) and the controls showed the typical age-related increase in lymphocyte activity and change of PFC isotype distribution. Two weeks after a secondary graft, a clear anamnestic tolerant response was observed-a marked increase in the numbers of activated B, CD4+, and CD8+ lymphocytes (Fig. 4) and PFCs (Table 4). The secondary graft also induced an increase of activated donor cells (from.9 to 2 x 16 B cells and from 1.4 to 5.8 x 16 T cells). Surprisingly, secondary graft rejection by primed normal mice did not result in an increase in splenic lymphocyte activity. DISCUSSION Neonatally induced tolerance to transplantation antigens in mice is associated with extensive activation of all major lymphocyte sets, B cells and CD4+ and CD8+ T cells. Splenomegaly is the macroscopic sign of tolerance in the combination of strains used and seems to be a common feature when transplantation tolerance is induced across major histocompatibility complex (MHC) class II antigens (B. Stockinger, personal communication). This splenomegaly, however, does not reflect host vs. graft disease occurring in some combinations of strains (22), as originally described Table 2. Frequencies and subset distribution of donor cells in the spleen of mice tolerized at birth and accepting or rejecting an allograft % of total Subset distribution, % Mice spleen cells IgM+ CD4+ CD8+ TO TA2w 1.7 ± TASw 5.6 ± TR2w TRSw 3.9 ± TO, tolerized animals; TA2w and TASw, tolerized animals with graft accepted at 2 and 5 weeks, respectively; TR2w and TRSw, tolerized animals with graft rejected at 2 and 5 weeks, respectively. Data for % total spleen cells are presented as mean ± SD. by Simpson et al. (23). In fact, it is not progressive but rather declines with age (the spleen of an old tolerant mouse is only twice the size of an age-matched control), is independent of lymph node enlargement, and is accompanied by persistent chimerism and by acceptance of allogeneic skin grafts; whereas host vs. graft reaction results in loss of chimerism and graft rejection (23). The increased spleen size is a consequence of the extensive activation of all lymphocyte subsets, resulting in as much as a 1-fold increase in absolute numbers of blasts compared to normal controls. The magnitude of a tolerant immune response is even greater than that of a conventional immune response to the same challenge-i.e., graft rejection by normal mice. The two pathways are also distinguishable by a relatively higher CD8' T-cell involvement in normal mice rejecting a graft and conversely by an increase in CD4' T-cell activation in tolerant animals accepting the graft. More striking is the difference in response of normal and tolerant mice to a secondary allograft: virtually no activation was found in normal mice as compared with an extensive expansion of all lymphocyte populations in tolerant mice. A tolerant state, therefore, shows characteristics of cellular activity that are usually associated with immunity, 1) U, t- xo x 6 (A u 1-1 CZ - < 6-2- I Host FIG. 3. Average numbers (+ 1 SD) of activated donor cells (bearing H-2b class I antigens) and host cells. These values were calculated by subtracting the numbers of activated donor cells from the total numbers of activated cells. Analysis and bars are as in Figs. 1 and 2.

4 Immunology: Bandeira et A Proc. Natl. Acad. Sci. USA 86 (1989) 275 Table 3. Numbers and state of activation of host and donor B and T cells in the spleen of mice tolerized at birth and accepting or rejecting an allograft Cells, no. x 1-6 per spleen IgM' cells CD4' cells CD8' cells Mice Host Donor Host Donor Host Donor TA2w 139 ± ± ± ± ± ±.4 ( ) (6.8 ± 7.1) (28.2 ± 7.6) (65.5 ± 7.5) ( ) (61.4 ± 15.7) TA5w ± ±.8 ( ) (55.6 ± 12.1) (18.4 ± 2.2) ( ) (8.4 ± 1.8) (56.3 ± 9.9) TR2w 42.5 ± ± ± (21.4 ± 5.2) ( ) (18.8 ± 5.5) ( ) ( ) (33. ± 2.3) TRSw * * 8.9 ± * (24.4 ± 6.8) (21.8) ( ) (21.6) (23.2 ± 4.6) (21.6) Control 42.8 ± ± ± 2.4 (24.3 ± 4.9) (22.4 ± 4.5) (14.8 ± 4.2) Numbers in parentheses are the percentage of activated cells. Data are mean + SD. Mice are as described in Table 2. *Only one mouse, having the highest number of donor cells, was tested for this parameter. such as an anamnestic enhanced secondary response. Other similarities include affinity thresholds (24) and restriction (25). The exuberant lymphoid activity of neonatally tolerized mice, rather than chimerism per se, correlates with the tolerant state as measured by the acceptance of the skin allograft. In fact, mice that were tolerized at birth with the same number of semiallogeneic cells but that rejected the skin graft have numbers of total and activated splenic lymphocyte close to those of normal mice. The activation level in these animals resembles that of normal mice rejecting a secondary graft, suggesting a priming effect of the semiallogeneic cells on the neonatal immune system, along the pathway of immunity. Although chimerism appears necessary for the maintenance of neonatal transplantation tolerance (21), certainly it is not sufficient, since we could detect donor cells also in %4 x 4- Tolerant UControl o~ ~ ~~~~~~~~c ~~~~~~~ ~~~3 P B C P B C FIG. 4. Secondary response to allograft of tolerant and normal mice. Average numbers (+ 1 SD) of activated CD8+, CD4+, and IgM+ cells (Left) and their frequency (Right) in the spleen of tolerant (open bars) and normal mice (stippled bars) 8 months after the primary challenge (bars P) and 2 weeks after the secondary challenge (bars B). The allograft was accepted by tolerant mice, still holding the primary graft. Control normal mice (bars C) were age-matched (1-12 months old). (Top) CD8+ cells. (Middle) CD4+ cells. (Bottom) Immunoglobulin-positive cells. nontolerant hosts. Rather, the major differences between tolerant and nontolerant mice were the absolute numbers of activated donor cells and their subset distribution. While in tolerant mice donor cells have a CD4'/CD8' cell ratio close to that found in normal spleen (-2.), in nontolerant animals the ratio drops to -.7. The most striking difference was seen in the survival of donor B cells in tolerant and nontolerant hosts, so that the B/T-cell ratio in the latter is 1 times lower than in the former. It appears, therefore, that the quality and functionality of donor cells rather than their numbers (and thus antigenicity) is relevant for tolerance, as suggested by experiments in rats (26). That such cells had to be activated was also indicated by the experiments of Silobrcic (27). Although the numbers of donor cells were considerably higher in tolerant than in nontolerant mice, most of the splenomegaly of tolerant animals was accounted for by host lymphocytes. The proportion of activated cells was not strikingly different in the two experimental groups; still tolerant mice had up to 4 times more activated cells than nontolerant and normal animals and had similarly higher numbers of immunoglobulin-secreting cells. A proportion of these are of donor origin, since immunoglobulins of donor allotype (Igb) were present at considerable levels in the serum of tolerant mice (data not shown). Our results suggest that the state of tolerance correlates with the global activity in the immune system, of which both host and donor cells participate. As suggested by the presence of tolerogen-specific CD4' lymphocytes in class II tolerant mice (28) and by syngeneic mixed lymphocyte reactions (29), natural or induced tolerance does not require (absolute) clonal deletion of (self) MHC class II reactive cells. In the amphibian system in addition to mixed lymphocyte reactions (3); anti-mhc antibodies are also present in tolerant animals (M. F. Flajnik and L. Du Pasquier, personal communication). Anti-idiotypic specificities to the autoreactive receptors have been proposed to provide suppressive Table 4. Immunoglobulin-secreting cells in the spleen of tolerant or normal mice challenged with a secondary allograft PFCs, no. X 1-3 per spleen Mice State IgM IgG3 IgG1 IgG2b IgG2a Primed (8 months) Tolerant Normal Boosted (2 weeks) Tolerant Normal Control Only one mouse per group was tested.

5 276 Immunology: Bandeira et al. circuits necessary for the maintenance of tolerance (31). In any event, however, tolerance is far from being a suppressed state. As an alternative to clonal inactivation, transplantation tolerance and immunity may differ by the nature of effector functions of activated cells. Different CD4' effector types (inflammatory vs. helper cells) (32) might be induced in the two situations and participate in establishing the same kind of dichotomy (suppression vs. cytotoxicity) in the CD8' population. Interestingly, suppression in transferable transplantation tolerance is mediated by a CD4' cell (11). Selftolerance in normal individuals is likely to be, at least in part, achieved by similar mechanisms. Thus, normal and antigenfree mice contain high numbers of activated B and T cells (33, 34). Importantly, CD8' blasts isolated from normal mice are efficient suppressors of B-cell response but lack cytotoxic ability, while helper cells but not inflammatory CD4' cells are naturally activated. Differential expression of effector functions, in turn, is not necessarily a genetic property of a lymphocyte subset but may result from the context in which the cell is activated, particularly from the level of its interaction (connectivity)' with other lymphocytes. Thus, a distinctive feature of internally activated B and T cells in normal mice appears to be a high degree of idiotypic connectivity involving B cells plus antibodies and T-cell receptors (35, 36). This functional idiotypic network is established early in life (37-39), maintained by continuous mutual selection (4) and positively selecting autoreactive clones, particularly those producing autoantibodies (41, 42), as in transplantation tolerance (43). We can then consider that neonatally induced transplantation tolerance is established by the complementation of host and donor functional networks and maintained by self-referential lymphoid activity. The arising immune system of the neonate exposed to the established self-defining network -of the donor (reflecting the donor genotype and determined primarily, but not exclusively, by its MHC, immunoglobulin, and T-cell receptor genes) will include in its repertoire specificities and interactions that would not be selected for on the basis of its own self. For the resulting unique functional network of the tolerant animal, self includes donor and recipient, through interactions involving multiple connected specificities, where the alloreactive repertoire to host and donor MHC antigens might even be a minority among all the anti-self specificities. Emergent properties of this immune network, and not specific clonal characteristics, may determine the global behaviors described as tolerance or immunity. Obviously, an (a x b)fj mouse does not show the hyperactivated characteristics of the tolerant animal. Thus, normal induction of self-tolerance must use at least some pathways that are different from the induced tolerance studied here. It would be interesting to induce neonatal tolerance with tissues (fetal liver or T- cell-depleted bone marrow), where no obvious mature network can preexist, to gain additional information on the behavior of the immune system in the establishment of tolerance. The study of dispersed (nonclonal) properties of an organized system is confronted with obvious difficulties. However, the results presented here may suggest other pathways of thought and provide methods of control of transplantation reactions that have thus far not been indicated by clonal (deletion or suppression) models. We thank Drs. Louis Du Pasquier and Gennaro De Libero for critical reading of the manuscript and Ms. Shirley Kirschbaum for secretarial work. Part of this work was done with the support of Proc. Natl. Acad. Sci. USA 86 (1989) Conseil National de la Recherche Scientifique and Institut National de la Santd et de la Recherche M6dicale. The Basel Institute for Immunology was founded and is supported by Hoffmann-La Roche, Basel, Switzerland. 1. Owen, R. D. (1945) Science 12, Billingham, R. E., Brent, L. & Medawar, P. B. (1953) Nature (London) 172, Btirnet, F. M. (1959) The Clonal Selection Theory ofacquired Immunity (Cambridge Univ. Press, Cambridge, England). 4. Dorsch, S. & Roser, B. J. (1975) Nature (London) 258, Streilein, J. W. & Gruchalla, R. S. (1981) Immunogenetics 12, Stockinger, B. (1984) Proc. Natl. Acad. Sci. USA 81, Kappler, J. W., Roehm, N. & Marrack, P. (1987) Cell 49, Kappler, J. W., Staerz, U., White, J. & Marrack, P. (1988) Nature (London) 332, McDonald, H. R., Schneider, R., Lees, R. K., Howe, R. C., Acha- Orbea, H., Festenstein, I., Zihkernagel, R. & Hengartner, H. (1988) Nature (London) 332, Kisielow, P., Blutmann, H., Staerz, U. D., Steinmetz, M. & von Boehmer, H. (1988) Nature (London) 333, Roser, B. J., Herbert, J. & Godden, U. (1983) Transplant. Proc. 15, Lemke, H., Hammerling, G. J. & Hammerling, U. (1979) Immunol. Rev. 47, Ozato, K. & Sachs, D. H. (1981) J. Immunol. 126, Pierres, A., Naquet, P., van Agthoven, A., Bekkhouca, F., Danizot, F., Mishal, I., Schmitt-Verhulst, A.-M. & Pierres, M. (1984) J. Immunol. 132, Ledbetter, J. A. & Herzenberg, L. A. (1979) Immunol. Rev. 47, Bazin, H., Xhurdebise, L.-M., Burtonboy, G., Lebacq, A.-M., De- Clercq, L. & Cormont, F. (1984) J. Immunol. Methods 66, Heitzmann, H. & Richards, F. M. (1974) Proc. Natl. Acad. Sci. USA 71, Forni, L. & de Petris, S. (1984) Methods Enzymol. 18, Gronovicz, E., Coutinho, A. & Melchers, F. (1976) Eur. J. Immunol. 6, Bernabe, R. R., Tuneskog, M., Forni, L., Martinez-A., C., Holmberg, D., Ivars, F. & Coutinho, A. (1981) in Immunological Methods, eds. Lefkovits, I. & Pernis, B. (Academic, London), Vol. 2, pp Lubaroff, D. M. & Silvers, W. K. (1973) J. Immunol. 111, Tateno, M., Kondo, N., Itoh, T. & Yoshiki, T. (1985) Clin. Exp. Immunol. 62, Simpson, E., O'Hopp, S., Harrison, M., Mosier, D., Melief, K. & Cantor, H. (1974) Immunology 27, Matzinger, P. (1986) EMBO Workshop on Tolerance: The Tolerance Workshop (Roche, Basel), Vol. 1, pp Matzinger, P., Zamoyska, R. & Waldmann, H. (1984) Nature (London) 36, Dorsch, S. & Roser, B. (1982) Transplantation 33, Silobrcic, V. (1971) Eur. J. Immunol. 1, Mohler, K. M. & Streilein, J. W. (1987) J. Immunol. 139, Glimcher, L. M., Longo, D. L., Green, I. & Schwartz, R. (1981) J. Exp. Med. 154, Flajnik, M. F., Du Pasquier, L. & Cohen, N. (1985) Eur. J. Immunol. 15, Dorsch, S. & Roser, B. (1982) Transplantation 33, Hossmann, T. R., Cherwinski, H., Bond, M. W., Giedlin, M. A. & Coffman, L. (1986) J. Immunol. 136, Pereira, P., Larsson, E.-L., Forni, L., Bandeira, A. & Coutinho, A. (1985) Proc. Natl. Acad. Sci. USA 82, Pereira, B., Forni, L., Larsson, E.-L., Cooper, M. D., Heusser, C. & Coutinho, A. (1986) Eur. J. Immunol. 16, Holmberg, D., Freitas, A., Portnoi, D., Jacquemart, F., Avrameas, S. & Coutinho, A. (1986) Immunol. Rev. 93, Martinez-A., C., Pereira, P., Bernabe, R., Bandeira, A., Larsson, E.-L., Cazenave, P.-A. & Coutinho, A. (1984) Proc. Natl. Acad. Sci. USA 81, Holmberg, D., Forsgren, S., Ivars, F. & Coutinho, A. (1984) Eur. J. Immunol. 14, Kearney, J. F. & Vakil, M. (1986) Ann. Immunol. (Paris) 137C, Martinez-A., C., Bernabe, R. R., Pereira, P., Cazenave, P.-A. & Coutinho, A. (1985) Nature (London) 317, Coutinho, A., Marquez, C., Araujo, P. M. F., Pereira, P., Toribio, M. L., Marcos, M. A. A. & Martinez-A., C. (1987) Eur. J. Immunol. 17, Dighiero, G., Lymberi, P., Holmberg, D., Lundquist, I., Coutinho, A. & Avrameas, S. (1985) J. Immunol. 134, Portnoi, D., Freitas, A., Bandeira, A., Holmberg, D. & Coutinho, A. (1986) J. Exp. Med. 164, Lusuy, S., Merino, J., Engers, H., Izui, S. & Lambert, P. H. (1986) J. Imniunol. 136,