- Email: [email protected]
Immunogenicity of B chain in insulin responder and nonresponder mice
CELLULAR
IMMUNOLOGY
130,129- 138 ( 1990)
lmmunogenicity of 6 Chain in Insulin Responder and Nonresponder Mice’ PETERE. JENSEN Department of Pathology and The Winship Cancer Center, Emory University School of Medicine, Atlanta, Georgia 30322 Received March 16, 1990; accepted April 22, 1990 The potential immunogenicity of insulin B chain in beef insulin low-responder H-2”” and high-responder H-2b*d mice was examined using lymph node proliferation assays. Oxidized B chain was immunogenic in H-2k+, but not H-2b,d, mice. The T cell population recognized a determinant in OX-B chain associated with I-Ak. These cells did not respond to intact insulin, suggesting that the B chain determinant was not available to I-Ak during immunologic processing of insulin. Responses were observed in H-2’ and H-2d, but not H-2b, after immunization with reduced and carboxyamidomethylated-insulin which contains equimolar A chain and B chain. These responses were I-A-restricted and heterogeneous, with reactivity to A chain and B chain determinants. In each case, little or no cross-reactivity was observed between RCAMinsulin and intact insulin. Furthermore, T cell populations induced in H-2k mice selectively recognized OX-B chain or RCAM-B chain, which differ in chemical modification of the thiols of Cys B7 and Cys B 19. Similarly, RCAM-BINS-immune T cells from H-2d did not react to OX-B chain. These results indicate that derivatization of the cysteine thiols, through disulfide bonds, oxidation, or carboxyamidomethylation, radically affects T cell recognition of insulin B chain. 0 1990 Academic PXSS. Inc.
INTRODUCTION Insulin serves as a useful model antigen for studying the genetic and cellular regulation of T cell-dependent immune responses. Closely related species variants with well defined amino acid sequence and tertiary structure are commercially available and the murine immune response to these variants is qualitatively regulated by class II genes of the major histocompatibility complex (1). Previous studies have indicated that, unlike other well studied model antigens, the determinants recognized by insulin-reactive T cells often require intact intrachain or interchain disulfide bonds (26). Naquet et al. (6) found that the interachain disulfide between Cys residues A6 and Al 1 and the interchain disulfide between A7 and B7 were required to maintain ’ This work was supported by the U.S. Public Health Service Grant CA-46667 from the National Cancer Institute, National Institutes of Health. 129 00088749190 $3.00 Copyright 0 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.
130
PETER E. JENSEN
antigenicity using BINS2-reactive T cell hybridomas restricted by I-Ad or I-Ab. Our own experience using I-Ad and I-Ad-restricted T cell hybridomas confirms this result. By contrast, other reports in the literature define T cell determinants confined to the B chain of insulin (7-9) suggesting that B chain peptides have access to class II molecules after disruption of disulfide bonds during immunological processing. To further explore potential T cell determinants present in the insulin molecule we have examined the immunogenicity of B chain derivatives of beef insulin in high responder H-2b,d and low responder H-2k*” mice. Using lymph node proliferation assays we find that B chain derivatives are immunogenic in high responder H-2d and low responder H-2k animals. However, very little cross-reactivity is observed between intact insulin and B chain using T cell populations induced by either antigen. Furthermore, B chain-responsive T cells are highly selective in their recognition of derivatives differing in covalent modification of the Cys B7 and Cys B 19 thiol groups. MATERIALS
AND METHODS
Mice. Female BALB/c BYJ(H-~~), B10.D2/nSnJ(H-2d), B10.BR/SgSnJ(H-2k), and B 1O.A/SgSnJ(H-2”) mice were obtained from The Jackson Laboratory (Bar Harbor, ME). These animals were maintained at least 1 week in the Emory University Animal Facilities before immunization at 8 to 11 weeks of age. Antigens and antibodies. BINS and PINS were purchased from Elanco Products Co. (Indianapolis, IN). The desoctapeptide of PINS (DOPINS) was purified by reverse-phase HPLC ( 10) after trypsin digestion ( 11). OX-BINS was produced by performic acid oxidation of BINS (12). OX-A chain and OX-B chain were purchased from Sigma Chemical Co. (St. Louis, MO) or purified from OX-BINS by reversephase HPLC using a C4 column with solvents A (60% CH,CN/O. 1% TFA) and B (0.1% TFA). The identity of the A chain and B chain fractions was confirmed by trypsin digestion. The purified fractions coeluted on reverse-phase HPLC with the peptides purchased from Sigma. OX-B chain (l-22) and (23-30) were purified from a trypsin digestion of OX-BINS using HPLC. RCAM-BINS was produced by reduction of BINS in buffer containing 8 M urea, 2 rrGt4 EDTA, 0.2 A4 Tris-HCl, pH 8.0, and a 20-fold molar excess of dithiothreitol at 37°C for 1 hr under nitrogen. The solution was cooled on ice and reacted with a slight excess of iodoacetamide over thiol groups for 1 hr prior to dialysis. RCAM-A chain and RCAM-B chain were purified by reverse phase HPLC. Purified protein derivative of tuberculin (PPD) (Connaugnt Medical Research Laboratory, Toronto, Ontario, Canada) was generously provided by Dr. J. W. Thomas (Houston, TX). The mAb 14-4-4 (anti-I-E) and 10-2-16 (antiI-Ak) were kindly provided by Dr. T. Hansen (St. Louis, MO) and used at l/200 dilution of ascites in culture. Lymph node proliferation assays. Mice were immunized in the foot pad with 50 pg antigen in CFA. After 9 to 11 days, draining popliteal and inguinal lymph nodes were removed, and single cell suspensions were prepared. Assays were set-up in flatbottom 96-well tissue culture plates using RPM1 1640 supplemented with 0.5% normal mouse serum, 5 X 10m5M 2-ME, 2 mM L-glutamine, 100 U/ml penicillin, and 100 U/ml streptomycin. Plates were incubated for 4 days at 37°C in 5% CO2 and ’ Abbreviations used: BINS, beef insulin; OX-BINS, oxidized beef insulin; PINS, pork insulin; RCAMBINS, reduced and carboxyamidomethylated BINS.
IMMUNOGENICITY
131
OF B CHAIN
TABLE 1 Immunogenicity of BINS and OX-B Chain” Response in the presence of Strain Bl0.A BIO BlO.D2 BALB/c
Immunogen BINS OX-B BINS OX-B BINS OX-B BINS OX-B
chain chain chain chain
PPD
None
BINS
OX-B chain
17.4 + 1.6 25.9 +- 1.3 20.2 * 1.0 29.1 _+3.9 19.6 + 0.9 31.1 + 2.2 62.5 k 2.2 91.8k4.1
0.5 20.1 0.8~0.1 0.8 * 0.2 1.1 kO.3 0.9 +- 0.1 I.5 k 0.2 0.9 * 0.1 l.O_+O.I
2.2 _+0.3
1.1 kO.1 32.7 + 6.0 2.2 + 0.4 4.2 +- 0.6 3.4 +- 0.9 5.7 + 1.1 4.7 + 0.9 4.2 k 0.5
16.1 k 3.0 40.4 +- 2.2 43.2 f 5.3
L?Cell donors were immunized with 50 rg antigen 9 days prior to assay.Unpurified lymph node cells (4 105/well) were cultured with PPD (80 pg/ml), BINS (50 pg/ml), or OX-B chain (50 &ml) as described under Materials and Methods. Results represent mean + SD of cell-associated cpm X 1Om3. X
cultures were pulsed with 1 &i of [3H]thymidine for the final 18 hr. After harvesting onto glass fiber paper, [3H]thymidine incorporation into DNA was quantitated by liquid scintillation counting. The data represent the mean and SD of triplicate cultures. Secondary in vitro proliferation responses to insulin using this protocol were absolutely dependent upon CD4+ T cells and immunization of cell donors with appropriate antigen. Responses to high doses of insulin were not observed using cells from donors primed with irrelevant antigen (data not shown). Results were highly reproducible and only representative experiments are shown. Cell lines. The I-Ad-bearing B cell lymphoma, M 12.4 (13), was generously provided by Dr. R. Asofsky (Bethesda, MD) and maintained in RPM1 1640 supplemented with 5% FCS, 2 mA4 glutamine, and 5 X 10e5 M 2-ME, and 10% NCTC 109 (M-A Bioproducts). Cloned I-Ad-restricted, insulin specific, T cell hybridomas were derived as previously described and maintained in RPM1 1640 supplemented with 10% FCS, 2 mM glutamine, and 5 X lo-’ M 2-ME (RIO) (10). Pd-2.5 is cross-reactive with BINS and PINS. Bd-1.4 reacts with BINS, but not PINS and Pd-2.16 reacts with PINS, but not BINS. Culture conditions and lymphokine assay. Cultures with T cell hybridomas were set-up in 96-well flat-bottom tissue culture plates in a final volume of 0.2 ml RlO (10). Each well contained 1 X lo5 T cells, I X lo5 M 12.4, and antigen. Supernatants (100 ~1) were removed after 24 hr and subjected to one freeze/thaw cycle. Lymphokine production, reflecting T cell activation, was quantitated using the lymphokineresponsive cell line, HT-2. Supernatants were cultured for 40 hr with 1 X lo4 HT.2 in 0.2 ml. Each well was pulsed with 1 &i of [3H]thymidine during the final 16 hr. The results represent the mean +- SD of [3H]thymidine incorporation of triplicate cultures. RESULTS Immunogenicity of oxidized B chain. Mice of the H-2b,d haplotype are high responders to BINS, whereas, those of the H-2k,” are low responders (1). Cells from
132
PETER E. JENSEN
FIG. 1. Reactivity of PINS-immune H-2d lymphocytes. Proliferative responses of 4 X lo5 unpurified lymph node cells from BALB/c mice immunized with 50 r.g PINS were determined as described under Materials and Methods. The OX-A chain was derived from PINS. Separate experiments are represented in a and b.
BINS immunized animals of these haplotypes responded weakly or not at all to high doses of OX-B chain in lymph node proliferation assays. (Table 1). The immunogenicity of OX-B chain was assessed using lymphocytes from OX-B chain-primed mice. OX-B chain was poorly immunogenic in the H-2b,d BINS high-responder animals. Consistent secondary in vitro proliferation responses were observed using BINS lowresponder H-2k+ mice (Table 1 and unpublished results). Within the haplotypes examined, a reciprocal pattern of immunogenicity between OX-B chain and BINS was observed. This result contrasted with that of Rosenwasser et al. (7) who reported that the response to PINS in H-2d mice was focused on a determinant present in the C-terminus of the B chain. Their data suggested that PINS and OX-B chains were completely cross-reactive in proliferation assays involving cultured peritoneal T cells from this haplotype. PINS-immune lymph node T cells were not reactive to either OX-A chain or OX-B chain in our experiments (Fig. la). In addition, desoctapeptide was able to stimulate a response equivalent to that generated with intact PINS (Fig. lb), indicating that the terminal eight amino acids of insulin B chain do not play a critical role in this response. We obtained similar results using cloned T cell hybridomas (Fig. 2).
Antigen
(PM)
FIG. 2. Specificity of insulin-reactive, I-Ad-restricted, T cell hybridomas. T cell hybridomas (1 X 105/ well) were cultured for 24 hr with M 12.4 (5 X lO“/weIl) and various concentrations of PINS (0), BINS (O), OX-B chain (O), or OX-A chain (m). OX-A chain was derived from BINS. Results reflect lymphokine secretion as described under Materials and Methods.
IMMUNOGENICITY 40.
-
133
OF B CHAIN
PINS BINS B-chain
1
10 Antigen
100
@M)
FIG. 3. B chain-immune H-2k T cells do not respond to insulin. Lymph node lymphocytes (4 X 105/ well) from BIO.BR mice which had been immunized with 50 rg OX-B chain were cultured with various concentrations of OX-B chain or insulin, as described under Materials and Methods.
T cells specific for BINS, PINS, or cross-reactive to BINS and PINS did not respond to the isolated chains of oxidized insulin, even when mixed (not shown). These results suggested that the A chain loop structure, which is disrupted in oxidized insulin, is of major importance in the T cell response to insulin in H-2d mice ( 14). Immunogenicity of OX-B chain in H-2k. BINS is not immunogenic in BlO.BR. By contrast, we have observed that OX-B chain is fully immunogenic in mice of this haplotype. The T cells induced after immunization with OX-B chain do not respond to intact insulin (Fig. 3). Antibody blocking studies indicate that this response is largely restricted to I-Ak and focused upon residues in OX-B chain ( I-22) (Fig. 4). It is evident from these results that the determinant is able to bind I-Ak and form antigenic
a
cpm x IO-3
FIG. 4. Specificity of B chain-immune H-2” T cells. Proliferative responses of lymph node lymphocytes from Bl0.A donors which had been immunized with OX-B chain were determined as described under Materials and Methods. In each case, 33 pm Ag was used. B chain peptides were derived from OX-B chain. Antibodies reactive with 1-A’ or I-Ek were used at l/200 dilution of ascites in culture.
134
PETER E. JENSEN
+ -t+
0 I 1
OX-B chain OX-BINS RCAM-BINS
10 Antlgan (fl)
7 100
FIG. 5. Selective recognition of B chain derivatives. Lymph node lymphocytes (5 X 105/well) from OXB chain-immune B I0.A donors were cultured with various concentrations of antigen as described under Materials and Methods.
complexes available for T cell recognition, and that T cells with the appropriate specificity are available in the repertoire of these mice. However, the structural context of this determinant in the insulin molecule prevents it from being available for recognition. Importance of thiol modification for T cell recognition. Studies using A chain-loop specific T cells have indicated that the intrachain disulfide bond between A6 and Al 1 and the interchain disulfide between A7 and B7 are required to maintain antigenicity (6). This observation suggests that insulin-specific T cells recognize conformational determinants dependent upon intact disulfides (5). Such conformational constraints are not present in the OX-B chain (l-22) peptide recognized in association with IAk. We were therefore surprised to find that OX-B chain-immune T cells do not recognize RCAM-B chain (Fig. 5). This peptide differs from OX-B chain by a difference in the thiol modification of Cys B7 and Cys B19. One or both of these residues is essential for I-Ak or T cell receptor binding. RCAM-BINS like OX-BINS, is immunogenic in H-2k mice (Fig. 6) and the response is heterogeneous with a greater response to RCAM-B chain than RCAM-A chain (data not shown). This result indicates that RCAM-B chain is immunogenic and T cells are present in the repertoire which can recognize this peptide. However, these T cells are not cross-reactive with OX-B chain and react poorly with intact BINS (Fig. 6).
Antigen
WI)
FIG. 6. RCAM-BINS-immune lymphocytes. Unpurified lymph node lymphocytes (5 X 105/well) were cultured with various concentrations of RCAM-BINS (0), OX-BINS (O), or BINS (0) as described under Materials and Methods.
IMMUNOGENICITY
Antigen
OF B CHAIN
135
(NM)
FIG. 7. BINS-immune, I-Ad-restricted T cells do not recognize RCAM-BINS. (a) Lymphocytes from BINS-primed BALB/c were cultured with various concentrations of BINS (0) or RCAM-BINS (0) and proliferative responses were measured as described under Materials and Methods. (b) Pd-2.5 T cells (1 X 105) were cultured with M 12.4 (1 X 105) and BINS (Cl) or RCAM-BINS (0). Results reflect lymphokine secretion as described under Materials and Methods.
RCAM-BINS was immunogenic in H-2d mice (Fig. 6). This contrasted with our results using oxidized insulin (Table 1 and unpublished results). T cell from RCAMBINS-immune donors did not respond to OX-BINS and responded very weakly to intact BINS. The response was heterogeneous and directed toward both A chain and B chain determinants and largely restricted by I-Ad (data not shown). No response to RCAM-BINS was detected using heterogeneous insulin-immune lymphocytes or IAd-restricted T cell hybridomas (Fig. 7). DISCUSSION The potential immunogenicity of insulin B chain was explored using H-2b,k,d mice and preparations of insulin in which the disulfide bonds had been disrupted by oxidation or reduction and carboxyamidomethylation. In no case did insulin-immune T cells respond in vitro to B chain derivatives. However, B chains were immunogenic in H-2k*d animals. B chains from OX-BINS (Table 1) or RCAM-BINS (unpublished) were poorly immunogenic in H-2b mice. This result is consistent with published data which indicate that the I-Ab-restricted T cell response to insulin is focused upon a determinant involving A chain and an intact intrachain (loop) disulfide bond between A6 and A I 1. Keck (1) originally reported that H-2b mice respond to BINS but not PINS, and since these two species variants differ in sequence only at residues A8 and A 10, concluded that the determinant included the A chain loop. There is at least one report (4) of I-Ab-restricted BINS-specific T cells responding to A chain peptides with intact intrachain disulfide bonds, implying that the minimal immunodominant determinant is confined to the A chain. Other data provide evidence that the B chain residues, in addition to those of the A chain, are required (6). Our data is compatible with either a deficiency of functional B chain reactive T cells (hole in the repertoire) ( 15, 16) or a low intrinsic affinity between B chain peptides and I-Ab (17, 18). We favor the former interpretation because we have isolated a cloned T cell hybridoma which recognized a B chain determinant in association with I-Ab (manuscript in preparation). In addition, Delovitch and associates have provided evidence that B chain peptides can bind to I-Ab ( 19). We suggest that the B chain is poorly immunogenic in H-
136
PETER
E. JENSEN
2b because specific T cell clones are infrequent or prevented from expanding after encountering antigen, by dominant suppressor T cells (20). B chain, but not intact insulin, is immunogenic in H-2k+. The T cells induced by immunization with B chain are not cross-reactive for BINS. We have considered the possibility that regulatory cells present in the lymph node selectively inhibit the response to intact insulin, but not B chain, because of a suppressor determinant present on the A chain (2 1). However, preliminary data from our laboratory suggests that B chain reactive cloned T cells, like the heterogeneous population, do not recognize intact insulin. It seems likely that the relevant B chain determinant is not generated during processing of insulin. It is clear from several experimental systems that residues outside of the minimal T cell epitope can influence antigen processing and presentation and affect epitope immunodominance (22-26). One can envision a number of mechanisms whereby the structural context of a determinant may influence its accessibility to T cells. Structural features may influence rate or route of uptake by the antigen-presenting cell, alter sites of cleavage by endopeptidases, or provide alternative agretopes which compete for binding to Ia. Recently, Thomas et al. (27) have published data which indicate that T cell responses to cryptic B chain determinants are observed after immunization of strain 2 guinea pigs with B chain, but not intact insulin. These results are analogous to those reported here. RCAM-B chain was also found to be immunogenic in H-2d. The T cell population induced by immunization with intact insulin (BINS or PINS) did not overlap with that induced by RCAM-B chain. By contrast to the results using H-2k, no response was observed in H-2d animals immunized with OX-B chain. This result differs from that reported by Rosenwasser et al. (7) who found that PINS and OX-B chain were completely cross-reactive in H-2d mice using T cell proliferation assays. The authors concluded that the response was directed toward a determinant present on the Cterminal region of the B chain since insulin-primed T cells did not react in vitro to desoctapeptide. Our data indicate that the response of heterogeneous I-Ad-restricted lymphocytes and cloned T cell hybridomas from insulin-immune mice is focused on a determinant which requires intact loop or interchain disulfide bonds and in which the terminal eight amino acids of the B chain play no essential role. Whether the discrepancy between our results and those of Rosenwasser et al. (7) is a consequence of differences in our immunization schedules, assay systems, or antigen preparations remains to be determined. However, evidence for A chain loop specificity among insulin-reactive H-2d T cells has been provided by a number of laboratories ($6, 14) and is further supported by the differential reactivity to PINS vs BINS among cloned I-Ad-restricted T cell hybridomas (Fig. 2 and unpublished data). As discussed for H-2k, it is possible that residues present in the intact insulin molecule hinder processing and presentation of potential B chain determinants which associate with I-Ad. Indeed, one must consider the possibility that disulfide bonds remain intact during immunologic processing of insulin in mouse antigen-presenting cells, generating only peptides which contain residues from both chains. It is likely that the conformation available for T cell recognition of such peptides would differ generally from that of peptides derived from isolated B chain. This possibility is weakened by the observed cross-reactivity between B chain and insulin using T cells from guinea pigs (2, 8) and humans (28). These results indicate that B chain determinants can be generated by guinea pig and human antigen-presenting cells during immunologic processing of insulin. In addition, these are reports of cross-reactivity between
IMMUNOGENICITY
OF B CHAIN
137
insulin and B chain using mouse T cells (5, 7, 9). It is therefore unlikely that our results can be attributed to a general inability of mammalian antigen-presenting cells to generate isolated B chain determinants from intact insulin which would be available for binding to class II glycoproteins. It is interesting that while RCAM-B chain was immunogenic, no response was observed after immunization of H-2d mice with OX-B chain or OX-BINS. Similarly, T cells from H-2k mice, after immunization with OX-B chain, were not cross-reactive for RCAM-B chain and vice versa. These results imply that the response to B chain in H-2d and H-2k is critically dependent upon thiol modifications present in Cys B7 or Cys B 19. One or both of these residues may be present in the T cell receptor contact site or may influence this epitope through interaction with the class II molecule. The fate of disulfide bonds during immunologic processing of proteins is unknown, Transhydrogenases capable of catalizing disulfide interchange and reduction of insulin and other substrates are well described (29,30). It is therefore possible that insulin B chain peptides bearing free thiol groups or mixed disulfides may be generated during immunologic processing of insulin, and that T cells may recognize these peptides without cross-reactivity to OX-B chain or RCAM-B chain. The lack of observed cross-reactivity between intact insulin and B chain is compatible with this hypothesis. Current efforts in our laboratory are directed toward further defining the potential role of thiol modification in T cell determinants and antigen processing. Studies using selected T cell hybridomas demonstrate that the antigenicity or ovalbumin, lysozyme, and other model protein antigen is preserved after disruption of disulfide bonds and covalent modification of thiols. Conformational constraints enduced by disulfide bonds and the chemical identity of cysteine side chains do not play a critical role in these T cell determinants. However, there is at least one published example of a T cell determinant requiring intact disulfide bonds within an antigen other than insulin (3 1). It will therefore be of interest to determine whether disulfide bonds are generally preserved during antigen processing, and if not, whether thiol groups remain free or chemically modified. ACKNOWLEDGMENTS I thank Joseph Moore for technical assistance, Dr. A. Ahmed Ansari for advice, and Janice Bell for preparation of this manuscript.
REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. I I.
Keck, K., Nature (London) 254,78, 1975. Barcinski, M. A., and Rosenthal, A. S., J. Exp. Med. 145,726, 1977. Spaeth, E., and Rude, E., Eur. J. Immunol. 13,756, 1983. Ishii, N., Klein, J., and Nagy, Z. A., Eur. J. Immunol. 13,658, 1983. Glimcher, L. H., Schroer, J. A., Chan, C., and Shevach, E. M., J. Zmmunol. 131,2868, 1983. Naquet, P., Ellis, J., Sin& B., Hodges, R. S., and Delovitch, T. L., J. Zmmunol. 139,3955, 1987. Rosenwasser, L. J., Barcinski, M. A., Schwartz, R. H., and Rosenthal, A. S., J. Immunol. 123, 471, 1979. Thomas, J. W., Danho, W., Bullesbach, E., Fohles, J., and Rosenthal, A. S., J. Immunol. 126, 1095, 1981. Huber, B. T., and Hochman, P. S., Eur. J. Immunol. 14, 1106, 1984. Jensen, P. E., J. Immunol. 141,2.545, 1988. Tager, H., Thomas, N., Assoian, R., Rubenstein, A., Seakow, M., Olefsky, J., and Kaiser, E. T., Proc. Natl. Acad. Sci. USA 77,3 18 1, 1980.
138
PETER E. JENSEN
12. Sanger, F., Biochem. J. 44, 126, 1949. 13. Kim, K. J., Kannelopoulos-Langevin, C., Merwin, R. M., Sachs, D. H., and Asofsky, R., J. Immunol. 122,549, 1979. 14. Cohen, I. R., and Talmon, J., Eur. J. Immunol. 10,284, 1980. 15. Shwartz, R. H., &and. J. Immunol. 7,3, 1978. 16. Jensen, P. E., Pierce, C. W., and Kapp, J. A., J. Exp. Med. 160, 1012, 1984. 17. Shevach, E. M., and Rosenthal, A. S., J. Exp. Med. 138, 1213, 1973. 18. Benacerraf, B., J. Immunol. 120, 1809, 1978. 19. Delovitch, T. L., Semple, J. W., Naquet, P., Bernard, N. F., Ellis, J., Champagne, P., and Phillips, M. L., Immunol. Rev. 106, 195, 1988. 20. Jensen, P. E., and Kapp, J. A., J. Mol. Cell. Immunol. 2, 133, 1985. 2 1. Yowell, R. L., Araneo, B. A., Miller, A., and Sercarz, E. E., Nature (London) 279,70, 1979. 22. Shastri, N., Miller, A., and Sercarz, E. E., J. Mol. Cell. Immunol. 1,369, 1984. 23. Shastri, N., Miller, A., and Sercarz, E. E., J. Immunol. 136,371, 1986. 24. Gammon, G., Shastri, N., Cogswell, J., Wilbur, S., Sadegh-Nasseri, S., Krzych, U., Miller, A., and Sercarz, E., Immunol. Rev. 98,53, 1987. 25. Werdelin, O., Annu. Inst. Pasteur Immunol. 137D, 323, 1986. 26. Brett, S. J., Cease, K. B., and Berzofsky, J. A., J. Exp. Med. 168,357, 1988. 27. Thomas, J. W., George-Gattner, H., and Danho, W., Eur. J. Immunol. 19,557, 1989. 28. Naquet, P., Ellis, J., Tibensky, D., Kenshole, A., Singh, B., Hodges, R., and Delovitch, T., J. Immunol. 140,2569,1988. 29. Varandani, P. T., Biochim. Biophys. Acta. 286, 126, 1972. 30. Carmichael, D. F., Morin, J. E., and Dixon, J. E., J. Biol. Chem. 252,7163, 1977. 31. Regnier-Vigouroux, A., El Ayeb, M., Defendini, M., Granier, C., and Pierres, M., J. Immunol. 140, 1069, 1988.
IMMUNOLOGY
130,129- 138 ( 1990)
lmmunogenicity of 6 Chain in Insulin Responder and Nonresponder Mice’ PETERE. JENSEN Department of Pathology and The Winship Cancer Center, Emory University School of Medicine, Atlanta, Georgia 30322 Received March 16, 1990; accepted April 22, 1990 The potential immunogenicity of insulin B chain in beef insulin low-responder H-2”” and high-responder H-2b*d mice was examined using lymph node proliferation assays. Oxidized B chain was immunogenic in H-2k+, but not H-2b,d, mice. The T cell population recognized a determinant in OX-B chain associated with I-Ak. These cells did not respond to intact insulin, suggesting that the B chain determinant was not available to I-Ak during immunologic processing of insulin. Responses were observed in H-2’ and H-2d, but not H-2b, after immunization with reduced and carboxyamidomethylated-insulin which contains equimolar A chain and B chain. These responses were I-A-restricted and heterogeneous, with reactivity to A chain and B chain determinants. In each case, little or no cross-reactivity was observed between RCAMinsulin and intact insulin. Furthermore, T cell populations induced in H-2k mice selectively recognized OX-B chain or RCAM-B chain, which differ in chemical modification of the thiols of Cys B7 and Cys B 19. Similarly, RCAM-BINS-immune T cells from H-2d did not react to OX-B chain. These results indicate that derivatization of the cysteine thiols, through disulfide bonds, oxidation, or carboxyamidomethylation, radically affects T cell recognition of insulin B chain. 0 1990 Academic PXSS. Inc.
INTRODUCTION Insulin serves as a useful model antigen for studying the genetic and cellular regulation of T cell-dependent immune responses. Closely related species variants with well defined amino acid sequence and tertiary structure are commercially available and the murine immune response to these variants is qualitatively regulated by class II genes of the major histocompatibility complex (1). Previous studies have indicated that, unlike other well studied model antigens, the determinants recognized by insulin-reactive T cells often require intact intrachain or interchain disulfide bonds (26). Naquet et al. (6) found that the interachain disulfide between Cys residues A6 and Al 1 and the interchain disulfide between A7 and B7 were required to maintain ’ This work was supported by the U.S. Public Health Service Grant CA-46667 from the National Cancer Institute, National Institutes of Health. 129 00088749190 $3.00 Copyright 0 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.
130
PETER E. JENSEN
antigenicity using BINS2-reactive T cell hybridomas restricted by I-Ad or I-Ab. Our own experience using I-Ad and I-Ad-restricted T cell hybridomas confirms this result. By contrast, other reports in the literature define T cell determinants confined to the B chain of insulin (7-9) suggesting that B chain peptides have access to class II molecules after disruption of disulfide bonds during immunological processing. To further explore potential T cell determinants present in the insulin molecule we have examined the immunogenicity of B chain derivatives of beef insulin in high responder H-2b,d and low responder H-2k*” mice. Using lymph node proliferation assays we find that B chain derivatives are immunogenic in high responder H-2d and low responder H-2k animals. However, very little cross-reactivity is observed between intact insulin and B chain using T cell populations induced by either antigen. Furthermore, B chain-responsive T cells are highly selective in their recognition of derivatives differing in covalent modification of the Cys B7 and Cys B 19 thiol groups. MATERIALS
AND METHODS
Mice. Female BALB/c BYJ(H-~~), B10.D2/nSnJ(H-2d), B10.BR/SgSnJ(H-2k), and B 1O.A/SgSnJ(H-2”) mice were obtained from The Jackson Laboratory (Bar Harbor, ME). These animals were maintained at least 1 week in the Emory University Animal Facilities before immunization at 8 to 11 weeks of age. Antigens and antibodies. BINS and PINS were purchased from Elanco Products Co. (Indianapolis, IN). The desoctapeptide of PINS (DOPINS) was purified by reverse-phase HPLC ( 10) after trypsin digestion ( 11). OX-BINS was produced by performic acid oxidation of BINS (12). OX-A chain and OX-B chain were purchased from Sigma Chemical Co. (St. Louis, MO) or purified from OX-BINS by reversephase HPLC using a C4 column with solvents A (60% CH,CN/O. 1% TFA) and B (0.1% TFA). The identity of the A chain and B chain fractions was confirmed by trypsin digestion. The purified fractions coeluted on reverse-phase HPLC with the peptides purchased from Sigma. OX-B chain (l-22) and (23-30) were purified from a trypsin digestion of OX-BINS using HPLC. RCAM-BINS was produced by reduction of BINS in buffer containing 8 M urea, 2 rrGt4 EDTA, 0.2 A4 Tris-HCl, pH 8.0, and a 20-fold molar excess of dithiothreitol at 37°C for 1 hr under nitrogen. The solution was cooled on ice and reacted with a slight excess of iodoacetamide over thiol groups for 1 hr prior to dialysis. RCAM-A chain and RCAM-B chain were purified by reverse phase HPLC. Purified protein derivative of tuberculin (PPD) (Connaugnt Medical Research Laboratory, Toronto, Ontario, Canada) was generously provided by Dr. J. W. Thomas (Houston, TX). The mAb 14-4-4 (anti-I-E) and 10-2-16 (antiI-Ak) were kindly provided by Dr. T. Hansen (St. Louis, MO) and used at l/200 dilution of ascites in culture. Lymph node proliferation assays. Mice were immunized in the foot pad with 50 pg antigen in CFA. After 9 to 11 days, draining popliteal and inguinal lymph nodes were removed, and single cell suspensions were prepared. Assays were set-up in flatbottom 96-well tissue culture plates using RPM1 1640 supplemented with 0.5% normal mouse serum, 5 X 10m5M 2-ME, 2 mM L-glutamine, 100 U/ml penicillin, and 100 U/ml streptomycin. Plates were incubated for 4 days at 37°C in 5% CO2 and ’ Abbreviations used: BINS, beef insulin; OX-BINS, oxidized beef insulin; PINS, pork insulin; RCAMBINS, reduced and carboxyamidomethylated BINS.
IMMUNOGENICITY
131
OF B CHAIN
TABLE 1 Immunogenicity of BINS and OX-B Chain” Response in the presence of Strain Bl0.A BIO BlO.D2 BALB/c
Immunogen BINS OX-B BINS OX-B BINS OX-B BINS OX-B
chain chain chain chain
PPD
None
BINS
OX-B chain
17.4 + 1.6 25.9 +- 1.3 20.2 * 1.0 29.1 _+3.9 19.6 + 0.9 31.1 + 2.2 62.5 k 2.2 91.8k4.1
0.5 20.1 0.8~0.1 0.8 * 0.2 1.1 kO.3 0.9 +- 0.1 I.5 k 0.2 0.9 * 0.1 l.O_+O.I
2.2 _+0.3
1.1 kO.1 32.7 + 6.0 2.2 + 0.4 4.2 +- 0.6 3.4 +- 0.9 5.7 + 1.1 4.7 + 0.9 4.2 k 0.5
16.1 k 3.0 40.4 +- 2.2 43.2 f 5.3
L?Cell donors were immunized with 50 rg antigen 9 days prior to assay.Unpurified lymph node cells (4 105/well) were cultured with PPD (80 pg/ml), BINS (50 pg/ml), or OX-B chain (50 &ml) as described under Materials and Methods. Results represent mean + SD of cell-associated cpm X 1Om3. X
cultures were pulsed with 1 &i of [3H]thymidine for the final 18 hr. After harvesting onto glass fiber paper, [3H]thymidine incorporation into DNA was quantitated by liquid scintillation counting. The data represent the mean and SD of triplicate cultures. Secondary in vitro proliferation responses to insulin using this protocol were absolutely dependent upon CD4+ T cells and immunization of cell donors with appropriate antigen. Responses to high doses of insulin were not observed using cells from donors primed with irrelevant antigen (data not shown). Results were highly reproducible and only representative experiments are shown. Cell lines. The I-Ad-bearing B cell lymphoma, M 12.4 (13), was generously provided by Dr. R. Asofsky (Bethesda, MD) and maintained in RPM1 1640 supplemented with 5% FCS, 2 mA4 glutamine, and 5 X 10e5 M 2-ME, and 10% NCTC 109 (M-A Bioproducts). Cloned I-Ad-restricted, insulin specific, T cell hybridomas were derived as previously described and maintained in RPM1 1640 supplemented with 10% FCS, 2 mM glutamine, and 5 X lo-’ M 2-ME (RIO) (10). Pd-2.5 is cross-reactive with BINS and PINS. Bd-1.4 reacts with BINS, but not PINS and Pd-2.16 reacts with PINS, but not BINS. Culture conditions and lymphokine assay. Cultures with T cell hybridomas were set-up in 96-well flat-bottom tissue culture plates in a final volume of 0.2 ml RlO (10). Each well contained 1 X lo5 T cells, I X lo5 M 12.4, and antigen. Supernatants (100 ~1) were removed after 24 hr and subjected to one freeze/thaw cycle. Lymphokine production, reflecting T cell activation, was quantitated using the lymphokineresponsive cell line, HT-2. Supernatants were cultured for 40 hr with 1 X lo4 HT.2 in 0.2 ml. Each well was pulsed with 1 &i of [3H]thymidine during the final 16 hr. The results represent the mean +- SD of [3H]thymidine incorporation of triplicate cultures. RESULTS Immunogenicity of oxidized B chain. Mice of the H-2b,d haplotype are high responders to BINS, whereas, those of the H-2k,” are low responders (1). Cells from
132
PETER E. JENSEN
FIG. 1. Reactivity of PINS-immune H-2d lymphocytes. Proliferative responses of 4 X lo5 unpurified lymph node cells from BALB/c mice immunized with 50 r.g PINS were determined as described under Materials and Methods. The OX-A chain was derived from PINS. Separate experiments are represented in a and b.
BINS immunized animals of these haplotypes responded weakly or not at all to high doses of OX-B chain in lymph node proliferation assays. (Table 1). The immunogenicity of OX-B chain was assessed using lymphocytes from OX-B chain-primed mice. OX-B chain was poorly immunogenic in the H-2b,d BINS high-responder animals. Consistent secondary in vitro proliferation responses were observed using BINS lowresponder H-2k+ mice (Table 1 and unpublished results). Within the haplotypes examined, a reciprocal pattern of immunogenicity between OX-B chain and BINS was observed. This result contrasted with that of Rosenwasser et al. (7) who reported that the response to PINS in H-2d mice was focused on a determinant present in the C-terminus of the B chain. Their data suggested that PINS and OX-B chains were completely cross-reactive in proliferation assays involving cultured peritoneal T cells from this haplotype. PINS-immune lymph node T cells were not reactive to either OX-A chain or OX-B chain in our experiments (Fig. la). In addition, desoctapeptide was able to stimulate a response equivalent to that generated with intact PINS (Fig. lb), indicating that the terminal eight amino acids of insulin B chain do not play a critical role in this response. We obtained similar results using cloned T cell hybridomas (Fig. 2).
Antigen
(PM)
FIG. 2. Specificity of insulin-reactive, I-Ad-restricted, T cell hybridomas. T cell hybridomas (1 X 105/ well) were cultured for 24 hr with M 12.4 (5 X lO“/weIl) and various concentrations of PINS (0), BINS (O), OX-B chain (O), or OX-A chain (m). OX-A chain was derived from BINS. Results reflect lymphokine secretion as described under Materials and Methods.
IMMUNOGENICITY 40.
-
133
OF B CHAIN
PINS BINS B-chain
1
10 Antigen
100
@M)
FIG. 3. B chain-immune H-2k T cells do not respond to insulin. Lymph node lymphocytes (4 X 105/ well) from BIO.BR mice which had been immunized with 50 rg OX-B chain were cultured with various concentrations of OX-B chain or insulin, as described under Materials and Methods.
T cells specific for BINS, PINS, or cross-reactive to BINS and PINS did not respond to the isolated chains of oxidized insulin, even when mixed (not shown). These results suggested that the A chain loop structure, which is disrupted in oxidized insulin, is of major importance in the T cell response to insulin in H-2d mice ( 14). Immunogenicity of OX-B chain in H-2k. BINS is not immunogenic in BlO.BR. By contrast, we have observed that OX-B chain is fully immunogenic in mice of this haplotype. The T cells induced after immunization with OX-B chain do not respond to intact insulin (Fig. 3). Antibody blocking studies indicate that this response is largely restricted to I-Ak and focused upon residues in OX-B chain ( I-22) (Fig. 4). It is evident from these results that the determinant is able to bind I-Ak and form antigenic
a
cpm x IO-3
FIG. 4. Specificity of B chain-immune H-2” T cells. Proliferative responses of lymph node lymphocytes from Bl0.A donors which had been immunized with OX-B chain were determined as described under Materials and Methods. In each case, 33 pm Ag was used. B chain peptides were derived from OX-B chain. Antibodies reactive with 1-A’ or I-Ek were used at l/200 dilution of ascites in culture.
134
PETER E. JENSEN
+ -t+
0 I 1
OX-B chain OX-BINS RCAM-BINS
10 Antlgan (fl)
7 100
FIG. 5. Selective recognition of B chain derivatives. Lymph node lymphocytes (5 X 105/well) from OXB chain-immune B I0.A donors were cultured with various concentrations of antigen as described under Materials and Methods.
complexes available for T cell recognition, and that T cells with the appropriate specificity are available in the repertoire of these mice. However, the structural context of this determinant in the insulin molecule prevents it from being available for recognition. Importance of thiol modification for T cell recognition. Studies using A chain-loop specific T cells have indicated that the intrachain disulfide bond between A6 and Al 1 and the interchain disulfide between A7 and B7 are required to maintain antigenicity (6). This observation suggests that insulin-specific T cells recognize conformational determinants dependent upon intact disulfides (5). Such conformational constraints are not present in the OX-B chain (l-22) peptide recognized in association with IAk. We were therefore surprised to find that OX-B chain-immune T cells do not recognize RCAM-B chain (Fig. 5). This peptide differs from OX-B chain by a difference in the thiol modification of Cys B7 and Cys B19. One or both of these residues is essential for I-Ak or T cell receptor binding. RCAM-BINS like OX-BINS, is immunogenic in H-2k mice (Fig. 6) and the response is heterogeneous with a greater response to RCAM-B chain than RCAM-A chain (data not shown). This result indicates that RCAM-B chain is immunogenic and T cells are present in the repertoire which can recognize this peptide. However, these T cells are not cross-reactive with OX-B chain and react poorly with intact BINS (Fig. 6).
Antigen
WI)
FIG. 6. RCAM-BINS-immune lymphocytes. Unpurified lymph node lymphocytes (5 X 105/well) were cultured with various concentrations of RCAM-BINS (0), OX-BINS (O), or BINS (0) as described under Materials and Methods.
IMMUNOGENICITY
Antigen
OF B CHAIN
135
(NM)
FIG. 7. BINS-immune, I-Ad-restricted T cells do not recognize RCAM-BINS. (a) Lymphocytes from BINS-primed BALB/c were cultured with various concentrations of BINS (0) or RCAM-BINS (0) and proliferative responses were measured as described under Materials and Methods. (b) Pd-2.5 T cells (1 X 105) were cultured with M 12.4 (1 X 105) and BINS (Cl) or RCAM-BINS (0). Results reflect lymphokine secretion as described under Materials and Methods.
RCAM-BINS was immunogenic in H-2d mice (Fig. 6). This contrasted with our results using oxidized insulin (Table 1 and unpublished results). T cell from RCAMBINS-immune donors did not respond to OX-BINS and responded very weakly to intact BINS. The response was heterogeneous and directed toward both A chain and B chain determinants and largely restricted by I-Ad (data not shown). No response to RCAM-BINS was detected using heterogeneous insulin-immune lymphocytes or IAd-restricted T cell hybridomas (Fig. 7). DISCUSSION The potential immunogenicity of insulin B chain was explored using H-2b,k,d mice and preparations of insulin in which the disulfide bonds had been disrupted by oxidation or reduction and carboxyamidomethylation. In no case did insulin-immune T cells respond in vitro to B chain derivatives. However, B chains were immunogenic in H-2k*d animals. B chains from OX-BINS (Table 1) or RCAM-BINS (unpublished) were poorly immunogenic in H-2b mice. This result is consistent with published data which indicate that the I-Ab-restricted T cell response to insulin is focused upon a determinant involving A chain and an intact intrachain (loop) disulfide bond between A6 and A I 1. Keck (1) originally reported that H-2b mice respond to BINS but not PINS, and since these two species variants differ in sequence only at residues A8 and A 10, concluded that the determinant included the A chain loop. There is at least one report (4) of I-Ab-restricted BINS-specific T cells responding to A chain peptides with intact intrachain disulfide bonds, implying that the minimal immunodominant determinant is confined to the A chain. Other data provide evidence that the B chain residues, in addition to those of the A chain, are required (6). Our data is compatible with either a deficiency of functional B chain reactive T cells (hole in the repertoire) ( 15, 16) or a low intrinsic affinity between B chain peptides and I-Ab (17, 18). We favor the former interpretation because we have isolated a cloned T cell hybridoma which recognized a B chain determinant in association with I-Ab (manuscript in preparation). In addition, Delovitch and associates have provided evidence that B chain peptides can bind to I-Ab ( 19). We suggest that the B chain is poorly immunogenic in H-
136
PETER
E. JENSEN
2b because specific T cell clones are infrequent or prevented from expanding after encountering antigen, by dominant suppressor T cells (20). B chain, but not intact insulin, is immunogenic in H-2k+. The T cells induced by immunization with B chain are not cross-reactive for BINS. We have considered the possibility that regulatory cells present in the lymph node selectively inhibit the response to intact insulin, but not B chain, because of a suppressor determinant present on the A chain (2 1). However, preliminary data from our laboratory suggests that B chain reactive cloned T cells, like the heterogeneous population, do not recognize intact insulin. It seems likely that the relevant B chain determinant is not generated during processing of insulin. It is clear from several experimental systems that residues outside of the minimal T cell epitope can influence antigen processing and presentation and affect epitope immunodominance (22-26). One can envision a number of mechanisms whereby the structural context of a determinant may influence its accessibility to T cells. Structural features may influence rate or route of uptake by the antigen-presenting cell, alter sites of cleavage by endopeptidases, or provide alternative agretopes which compete for binding to Ia. Recently, Thomas et al. (27) have published data which indicate that T cell responses to cryptic B chain determinants are observed after immunization of strain 2 guinea pigs with B chain, but not intact insulin. These results are analogous to those reported here. RCAM-B chain was also found to be immunogenic in H-2d. The T cell population induced by immunization with intact insulin (BINS or PINS) did not overlap with that induced by RCAM-B chain. By contrast to the results using H-2k, no response was observed in H-2d animals immunized with OX-B chain. This result differs from that reported by Rosenwasser et al. (7) who found that PINS and OX-B chain were completely cross-reactive in H-2d mice using T cell proliferation assays. The authors concluded that the response was directed toward a determinant present on the Cterminal region of the B chain since insulin-primed T cells did not react in vitro to desoctapeptide. Our data indicate that the response of heterogeneous I-Ad-restricted lymphocytes and cloned T cell hybridomas from insulin-immune mice is focused on a determinant which requires intact loop or interchain disulfide bonds and in which the terminal eight amino acids of the B chain play no essential role. Whether the discrepancy between our results and those of Rosenwasser et al. (7) is a consequence of differences in our immunization schedules, assay systems, or antigen preparations remains to be determined. However, evidence for A chain loop specificity among insulin-reactive H-2d T cells has been provided by a number of laboratories ($6, 14) and is further supported by the differential reactivity to PINS vs BINS among cloned I-Ad-restricted T cell hybridomas (Fig. 2 and unpublished data). As discussed for H-2k, it is possible that residues present in the intact insulin molecule hinder processing and presentation of potential B chain determinants which associate with I-Ad. Indeed, one must consider the possibility that disulfide bonds remain intact during immunologic processing of insulin in mouse antigen-presenting cells, generating only peptides which contain residues from both chains. It is likely that the conformation available for T cell recognition of such peptides would differ generally from that of peptides derived from isolated B chain. This possibility is weakened by the observed cross-reactivity between B chain and insulin using T cells from guinea pigs (2, 8) and humans (28). These results indicate that B chain determinants can be generated by guinea pig and human antigen-presenting cells during immunologic processing of insulin. In addition, these are reports of cross-reactivity between
IMMUNOGENICITY
OF B CHAIN
137
insulin and B chain using mouse T cells (5, 7, 9). It is therefore unlikely that our results can be attributed to a general inability of mammalian antigen-presenting cells to generate isolated B chain determinants from intact insulin which would be available for binding to class II glycoproteins. It is interesting that while RCAM-B chain was immunogenic, no response was observed after immunization of H-2d mice with OX-B chain or OX-BINS. Similarly, T cells from H-2k mice, after immunization with OX-B chain, were not cross-reactive for RCAM-B chain and vice versa. These results imply that the response to B chain in H-2d and H-2k is critically dependent upon thiol modifications present in Cys B7 or Cys B 19. One or both of these residues may be present in the T cell receptor contact site or may influence this epitope through interaction with the class II molecule. The fate of disulfide bonds during immunologic processing of proteins is unknown, Transhydrogenases capable of catalizing disulfide interchange and reduction of insulin and other substrates are well described (29,30). It is therefore possible that insulin B chain peptides bearing free thiol groups or mixed disulfides may be generated during immunologic processing of insulin, and that T cells may recognize these peptides without cross-reactivity to OX-B chain or RCAM-B chain. The lack of observed cross-reactivity between intact insulin and B chain is compatible with this hypothesis. Current efforts in our laboratory are directed toward further defining the potential role of thiol modification in T cell determinants and antigen processing. Studies using selected T cell hybridomas demonstrate that the antigenicity or ovalbumin, lysozyme, and other model protein antigen is preserved after disruption of disulfide bonds and covalent modification of thiols. Conformational constraints enduced by disulfide bonds and the chemical identity of cysteine side chains do not play a critical role in these T cell determinants. However, there is at least one published example of a T cell determinant requiring intact disulfide bonds within an antigen other than insulin (3 1). It will therefore be of interest to determine whether disulfide bonds are generally preserved during antigen processing, and if not, whether thiol groups remain free or chemically modified. ACKNOWLEDGMENTS I thank Joseph Moore for technical assistance, Dr. A. Ahmed Ansari for advice, and Janice Bell for preparation of this manuscript.
REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. I I.
Keck, K., Nature (London) 254,78, 1975. Barcinski, M. A., and Rosenthal, A. S., J. Exp. Med. 145,726, 1977. Spaeth, E., and Rude, E., Eur. J. Immunol. 13,756, 1983. Ishii, N., Klein, J., and Nagy, Z. A., Eur. J. Immunol. 13,658, 1983. Glimcher, L. H., Schroer, J. A., Chan, C., and Shevach, E. M., J. Zmmunol. 131,2868, 1983. Naquet, P., Ellis, J., Sin& B., Hodges, R. S., and Delovitch, T. L., J. Zmmunol. 139,3955, 1987. Rosenwasser, L. J., Barcinski, M. A., Schwartz, R. H., and Rosenthal, A. S., J. Immunol. 123, 471, 1979. Thomas, J. W., Danho, W., Bullesbach, E., Fohles, J., and Rosenthal, A. S., J. Immunol. 126, 1095, 1981. Huber, B. T., and Hochman, P. S., Eur. J. Immunol. 14, 1106, 1984. Jensen, P. E., J. Immunol. 141,2.545, 1988. Tager, H., Thomas, N., Assoian, R., Rubenstein, A., Seakow, M., Olefsky, J., and Kaiser, E. T., Proc. Natl. Acad. Sci. USA 77,3 18 1, 1980.
138
PETER E. JENSEN
12. Sanger, F., Biochem. J. 44, 126, 1949. 13. Kim, K. J., Kannelopoulos-Langevin, C., Merwin, R. M., Sachs, D. H., and Asofsky, R., J. Immunol. 122,549, 1979. 14. Cohen, I. R., and Talmon, J., Eur. J. Immunol. 10,284, 1980. 15. Shwartz, R. H., &and. J. Immunol. 7,3, 1978. 16. Jensen, P. E., Pierce, C. W., and Kapp, J. A., J. Exp. Med. 160, 1012, 1984. 17. Shevach, E. M., and Rosenthal, A. S., J. Exp. Med. 138, 1213, 1973. 18. Benacerraf, B., J. Immunol. 120, 1809, 1978. 19. Delovitch, T. L., Semple, J. W., Naquet, P., Bernard, N. F., Ellis, J., Champagne, P., and Phillips, M. L., Immunol. Rev. 106, 195, 1988. 20. Jensen, P. E., and Kapp, J. A., J. Mol. Cell. Immunol. 2, 133, 1985. 2 1. Yowell, R. L., Araneo, B. A., Miller, A., and Sercarz, E. E., Nature (London) 279,70, 1979. 22. Shastri, N., Miller, A., and Sercarz, E. E., J. Mol. Cell. Immunol. 1,369, 1984. 23. Shastri, N., Miller, A., and Sercarz, E. E., J. Immunol. 136,371, 1986. 24. Gammon, G., Shastri, N., Cogswell, J., Wilbur, S., Sadegh-Nasseri, S., Krzych, U., Miller, A., and Sercarz, E., Immunol. Rev. 98,53, 1987. 25. Werdelin, O., Annu. Inst. Pasteur Immunol. 137D, 323, 1986. 26. Brett, S. J., Cease, K. B., and Berzofsky, J. A., J. Exp. Med. 168,357, 1988. 27. Thomas, J. W., George-Gattner, H., and Danho, W., Eur. J. Immunol. 19,557, 1989. 28. Naquet, P., Ellis, J., Tibensky, D., Kenshole, A., Singh, B., Hodges, R., and Delovitch, T., J. Immunol. 140,2569,1988. 29. Varandani, P. T., Biochim. Biophys. Acta. 286, 126, 1972. 30. Carmichael, D. F., Morin, J. E., and Dixon, J. E., J. Biol. Chem. 252,7163, 1977. 31. Regnier-Vigouroux, A., El Ayeb, M., Defendini, M., Granier, C., and Pierres, M., J. Immunol. 140, 1069, 1988.