Clonal Analysis of Alloreactive T Cell Responses Against the Closely Related B*2705 and B*2703 Subtypes

Transcription

1 Clonal Analysis of Alloreactive T Cell Responses Against the Closely Related B*2705 and B*2703 Subtypes Implications for HLA-B27 Association to Spondyloarthropathy' Daniel Lopez, Rosa Garcia-Hoyo, and JosC A. Lopez de Castro' Center of Molecular Biology "Sever0 Ochoa," Council of Scientific Investigations and Autonomous University of Madrid, Cantoblanco, Madrid, Spain Alloreactive responses against closely related HLA-B27 subtypes were compared at the clonal level by fine specificity analysis of anti-b*2705 and anti-b*2703 CTL clones from unrelated HLA-B27- individuals with target cells expressing B*2701 to B*2706, and other HLA Ags. T cell epitope sharing between B*2705 and other subtypes was B*2705 > B'2703 > B*2702 > B*2701 > B*2704 > B*2706, and correlated with their amino acid differences. This suggests that identical or similar peptides, or peptide motifs, can be presented by multiple HLA-B27 subtypes to T cells, a feature that may be critical for the similar linkage of several subtypes to spondyloarthropathies. Other cross-reactions were predominantly with HLA-B61 and HLA-B6O. Marked differences were observed in the nature and frequency of clonal reaction patterns among individuals. They correlated with structural features of the HLA-B Ags from each donor, suggesting that anti-hla-b27 T cell responses are partially determined by the HLA-B phenotype of the responder. Ability to respond to particular HLA-B27-associated epitopes may determine differences in disease susceptibility among HLA-B27+ individuals. The anti-b*2703 and anti-b*2705 responses in the same individual were different. A major feature of anti-b*2703 CTL was that a large majority cross-reacted with B*2705. This can be explained by the effect of the single amino acid change in B*2703 on peptide binding and suggests that the B*2703-bound peptide repertoire is mainly a subset of that bound to B*2705, with few peptides being specifically presented by B*2703 to T cells. Journal of Immunology, 1994, 152: C lass I MHC molecules constitutively bind and present at the cell surface a vast set of peptides derived from metabolic degradation of cellular proteins (1, 2). These complexes are recognized by CTL, and the diversity of alloreactive T cell responses, generated when lymphocytes are stimulated with cells from an MHC-mismatched individual, largely results from involvement of alloantigen-bound peptides in the allospecific T cell epitopes (3, 4). Differences in alloreactive CTL precursor frequencies and CTL responses among unrelated individuals have Received for publication December 15, Accepted for publication March 16, 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. Section 1734 solely to indicate this fact. ' This work was supported by Grant SAL 91/0408 from the Cornision Interministerial de Ciencia y Tecnologia (Plan Nacional de I+D). D.L. is a Fellow of the Fundacion Jimenez Diaz. The authors thank the Fundacion Ramon Areces for an institutional grant to the Centro de Biologia Molecular. ' Address correspondence and reprint requests to Dr. Jose A. Lopez de Castro, Centro de Biologia Molecular, Universidad Autonoma de Madrid, Facultad de Ciencias Biologicas, Cantoblanco, 28049, Madrid, Spain. been reported in humans (5-7), after more extensive studies in the mouse (8). Such differences are important because they can influence allograft survival and, because they reflect different TCR repertoires, susceptibility to HLA-associated disease in humans. The MHC phenotype of the individual must influence the nature of allospecific T cell responses (8) because of the role of MHC Ags in shaping T cell repertoires. Moreover, genetically unrelated HLA-matched individuals generate distinct responses (5). Other genetic and environmental factors, such as the Ab repertoire (9) or previous encounters with Ag, can modulate the TCR repertoire in each individual, influencing alloreactivity. There is strong evidence that HLA-B27 is directly involved in the pathogeny of ankylosing spondylitis and reactive arthritis. Classical population analyses showed that susceptibility to these diseases correlatesolely with HLA-B27 (10, 11). More recently, HLA-B27 transgenic rats have been shown to develop a multiorgan system disease with many of the features of human B27-associated spondyloarthropathies (12), which correlates with the level of HLA-B27 expression (13). Copyright by The American Association of Immunologists /94/$02.00

2 5558 THE ALLOREACTIVITY AGAINST B*2703 B*2705 AND SUBTYPES The mechanism of HLA-B27 association to disease is unknown. One hypothesis is that HLA-B27 is the restricting element for an arthritogenic peptide that would crossreact with a bacteria-derived Ag, triggering a cytotoxic T cell response upon infection (14). This model, which assumes a central role of T cells in the pathogeny of HLA- B27-associated disease, is based on the fact that the only established function of HLA-B27 is to present peptide Ags to CTL. Peptides from intracellular bacteria that are major triggering agents of reactive arthritis (15) can be presented by class I MHC Ags and induce class I-restricted cytotoxic T cell responses (16). A role of such responses in the pathogenesis of B27-associated disease is suggested by recent findings of B27-restricted, bacteria-specific CTL, as well as B27-specific autoreactive CTL, in the synovial fluid of patients with reactive arthritis and ankylosing spondylitis (17). Pathogenetic mechanisms must explain the association of multiple structurally and functionally distinct HLA-B27 subtypes to disease, and the fact that most HLA-B27+ individuals never develop B27-associated spondyloarthropathies. Of the seven known HLA-B27 subtypes, B"2701 to B*2707, those that are found with a frequency large enough for population studies, either in Caucasoids (B"2705 and B*2702) or in Orientals (B*2704), are equally linked to disease (18). However, they are structurally heterogeneous and easily distinguished by specific CTL (19, 20). If arthritogenic peptides are involved in the pathogeny of B27-associated disease, all these subtypes should be able to present such peptides to CTL. This implies that they should have some T cell epitopes in common. B*2703, which is a major subtype in West African blacks and is apparently restricted to this ethnic group, has not been found in ankylosing spondylitis patients, and it was suggested to be negatively associated to spondyloarthropathy (21). However, epidemiologic or other positive evidence for this is yet lacking. B"2703 differs from B*2705 in a single Tyr to His amino acid change at position 59 (22), which is unique among HLA Ags of known structure. Tyr59 participates in a conserved network of hydrogen bonds that stabilizes peptide binding by interacting with the amino-terminus of class I-bound peptides (23, 24). The change to His must disrupt this network, affecting peptide binding at its N-terminal end. All other subtypes have changes that affect peptide binding at more carboxyl-terminal positions (19). These studies underline the importance of assessing T cell epitope differences and similarities between HLA-B27 subtypes as well as the effect of subtype polymorphism on immunogenicity. In addition, individual differences among allospecific CTL responses to HLA-B27 are important in understanding the relationship of HLA-B27 to disease. That most HLA-B27+ individuals never develop B27-associated disease suggests that there may be additional pre- disposing factors. One such factor could be differences among individuals in their TCR repertoires, conditioning their capacity to respond to disease-associated antigenic determinants on HLA-B27. The aim of our study was to analyze alloreactive T cell responses against the B"2705 and B"2703 subtypes at the clonal level to assess T cell epitope sharing among HLA-B27 subtypes and with non-b27 class I Ags, differences among individuals in their capacity to respond to particular B27-specific T cell epitopes, and the differential features of T cell responses against the closely related B"2703 subtype. Materials and Methods Cell lines Human lymphoblastoid cell lines (LCL)', used as stimulators in mixed lymphocyte cultures and as target cells, were cultured in RPMI 1640 medium containing 25 mm HEPES buffer, 10% heat-inactivated FCS, and 2 mm L-glutamine (GIBCO BRL Laboratories, Paisley, UK) at 37 C and 5% CO,. The human class I-deficient HMy2.ClR cells, and the B*2705'-HMy2.C1R transfectant 1-38 were previously described (25, 26). They were cultured as LCL. mabs, flow cytornetry, and cytotoxicity assays The following mab were used: W6132 (anti-hla class I monomorphic determinant) (27), ME1 (anti-hla-b27 + B7 + B22) (28), and EDU-1 (anti-hla class I1 monomorphic determinant) (29). Phenotyping of T cells by flow cytometry was carried out using 1/300 dilutions of ascitic fluid or undiluted culture supernatants of the SPV-T3b (anti-cd3). HP216 (anti-cd4), and B9/4 (anti-cd8) mab, as previously described (30). "Cr-release cytotoxicity, cold target, and mab inhibition assays were previously described (30). The mab in these assays were used at dilution of ascitic fluid, except EDU-1. which was used at dilution. Isolation of HLA-927-specific CTL clones T cell clones were obtained by limiting dilution of peripheral blood mononuclear cells from three HLA-827- donors: DL, GM, and SR. They were stimulated in mixed lymphocyte culture (MLC) with irradiated (6.000 rads) human B*2705+ or B*2703+ LCL, as previously described (30). Responder-stimulator combinations used, and their HLA phenotypes, are detailed in Table I. Cells in wells growing below the statistical limit for clonality were initially screened for HLA-B27 reactivity by "Cr-release cytotoxicity at an unknown Eir ratio by one of two procedures: 1) using the stimulator LCL as target cells in the absence and in the presence of the ME1 mab; 2) anti-b*2705 CTL were sometimes screened with the B27+-CIR transfectant 1-38 and B27" ClR cells. T cells showing greater than 20% specific cytotoxicity against HLA-B27+ LCL, and inhibited at least 50% by MEl, or showing greater than 20% specific lysis of 1-3B, and at least double of the lysis of B27- C1R cells, were further expanded. T cell clones that showed specific reactivity for HLA-B27 upon more detailed analysis and which grew sufficiently for characterization of their fine specificity were used in this study. Most of the anti-827 CTL obtained were CD3 'CD8'CD4'. In those wells with a coexisting CD4+ population, this was often depleted by culturing the T cells in the presence of a different B*2705+ or B"2703+ LCL and anti- CD4 mab (11100 dilution of ascitic fluid) until a stable CD8'CD4- phenotype was obtained. Cloned T cells were maintained with 20 IU1rnl of ril-2 (a kind gift of Hoffmann-La Roche Inc., Nutley, NJ) and were restimulated weekly with the corresponding B*2705+ or B*2703' stimulator cells (Table I) as previously described (30). Abbreviation used in thls paper: LCL, lymphoblastoid cell lines.

3 J DL Journal of Immunology 5559 Table I. Responder/sfimu/ator combinations used for obtaining anti-hla-627 CTL 6*2705+ Stimulator Cells Prlmary Secondary After Anti-8'2705 Responder" cloning MLC MLCh CTL Clones' DL 1-3B DM5 (11, DTB (1) R69 - R69 DRD (31, DAG (1) R69 R69 LC1 5 DLE (1) LC1 5 LG15 R69 DRF (8) GM R69 - R69 GRK (1 5) LC1 5 - LC1 5 GMZ (1) LC1 5 LC15 LC15 A2 (2) SR R69 - S69 R69 (41, SRF (2) SRY (4), SRB (1) R69 R69 R69 SRZ (l), SRC (3) R69 R69 LC1 5 SLC (8) R69 R15 R15 S15 (10) 8*2703+ Stimulator Cells Primary Secondary After Anti-B*2703 Responder cloning MLC MLC CTL Clones DL CH - - DCD (1) LAR - - DLL (11 CH LAR - DLH (5) LAR CH - DCJ (10) CH LAR DCI CHd (4) CH CM LAR LAR GLM (6) CH LAR CHd GCP (9) (HLA-A29, 31; 839, 44; DR2, 71, CM (HLA-AI, 24; 87, 8; DR 1. 31, and SR (HLA-A3, 29; 87, 44; DR2,7j. "Stimulator LCL used were R69 (HLA-A3, 24; B*2705, 7; DR3, 51, LC15 (HLA-A32; 6*2705; DRl), and R15 (HLA-A3; B*2705, 35) for anti-hla- B*2705 CTL, and CH (HLA-A31, 32; 6'2703, 18; DR8, 11) and LAR (HLA- A23, 31; 6'2703, 53) for anti-hla-b*2703 CTL. Sometimes the stimulator LCL was changed upon restimulation or cloning to select for growth of anti-627 CTL clones. ' Generic names of the T cell clones from each cloning experiment are followed in parentheses by the number of HLA-B27-specific CTL clones that grew in sufficient amounts for fine specificity analysis. "In these experiments a tertiary MLC with the same LCL was performed before clonlng. Results Anti-B*2705 and anti-b*2703 alloreactive CTL A total of 66 anti-b*2705 CTL clones from donors DL, GM, and SR, and 36 anti-b*2703 CTL clones from the two former donors, were analyzed for their fine specificity with a panel of target LCL expressing the B*2701 to B*2706 subtypes and numerous other HLA specificities. At least two LCL were used for each HLA-B27 subtype except B*2701, for which only one LCL was available. Fifteen anti-b*2705 and 21 anti-b*2703 CTL clones from donor DL, 18 anti-b*2705 and 15 anti-b*2703 CTL clones from GM, and 33 anti-b"2705 CTL clones from SR were characterized (Tables I1 to VI). Reactivity with each HLA-B27 subtype was confirmed by inhibition of cytotoxicity with the ME1 mab every time it could be tested, which was in all cases for anti-b*2703 CTL and in 90% of the positive cytotoxicities for anti-b*2705 CTL (data not shown). In most other cases, the possibility that other Ags were responsible for cytotoxicity of B27+ targets was ruled out from the panel analysis. In a previous study from our laboratory, 18 anti-b*2705 CTL clones from another donor, PA (HLA-A24, 33; B35, 39; DR6), were characterized (31). The results allow us to assess the degree of T cell epitope sharing among HLA-B27 subtypes and the cross-reactivity of anti-b27 CTL with other HLA Ags, and to compare anti-b27 T cell responses against different subtypes and from different individuals at a clonal level. Cross-reaction of anti-hla-bz7 CTL with HLA-B27 subtypes T cell cross-reactivity among B*2701 to B*2706 was established from the reaction patterns of 84 anti-b*2705 CTL clones from DL, GM, SR (Tables I1 to IV), and PA (31), and 36 anti-b*2703 CTL clones from the two former donors (Tables V and VI). The results are summarized in Table VII. Cross-reactivity with the six subtypes was observed with anti-b*2705 and anti-b"2703 CTL clones. Crossreactive patterns involved multiple combinations, and included from two to six subtypes (see below). Thus, HLA-B27 subtypes share allospecific T cell epitopes recognized by anti-b*2705 and anti-b*2703 CTL. There was a hierarchy in the cross-reactivity of B*2705 with other subtypes that correlated with subtype structure when assessed with the total of anti-b*2705 CTL clones: B*2703, with a single change at residue 59, was the most cross-reactive subtype; B*2702 and B*2701, with changes at residues 77 to 81 and 74 to 81, respectively, were similarly cross-reactive, but less than B*2703, with B*2705; B*2704, with changes at residues 77 and 152, and B*2706, with two additional changes at residues 114 and 116, were the antigenically most distant subtypes from B*2705. Among anti-b*2703 CTL, the most striking data was their high number (78%) cross-reacting with B*2705. This indicates that a large majority of the allospecific T cell epitopes in B*2703 are shared by B*2705. Crossreaction with other subtypes was found in much lower frequency. There were significant differences in the degree of subtype cross-reactivity among individuals. Crossreaction of anti-b*2705 CTL with B*2704 and B*2706 was not found in PA and DL, but was 33% and 28%, respectively in GM, and 12% and 6%, respectively, in SR. Crossreactions with B*2701, B*2702, and B*2703 were found in all four donors with a wide range of frequencies. Thus, subtype relatedness established with the total of anti-b*2705 CTL was not always reflected in individual responses. The major differences between DL and GM in their anti-b*2703 CTL were that cross-reaction with B*2702 was much more frequent in DL, and with B*2701 was only found in

4 5560 ALLOREACTIVITY AGAINST THE B*2705 AND B*2703 SUBTYPES

5 Journal of Immunology 5561

6 5562 ALLOREACTIVITY AGAINST THE B*2705 AND B*2703 SUBTYPES I\

7 Journal of Immunology

8 5564 ALLOREACTIVITY AGAINST THE t3*2705 AND 6*2703 SUBTYPES

9 Journal of Immunology 5565 Table VII. Crossreaction of anti-hla-627 CTL clones with HLA-B27 subtypes" Reactive ant1-hla-b*2705 CTL from Individual Donors 8*2703 8*2705 8*2701 8*2702 I' Figures represent the number (N) and percentage (YO) of CTL clones recognizing each subtype. Data from PA were previously reported (31) and are included here for comparison. The criteria for reactivity assignments are explained in Footnote a to Table II. B27-dlrected cytotoxicity was confirmed by inhibition of cytotoxicity with the ME1 rnab, except when insufficient numbers of CTL precluded performing the assay (see text). Cold target inhibition assays where performed, whenever possible, to confirm the monoclonal nature of the cross-reactions. "Only 17 CTL clones from this donor were tested with 6*2701 targets (Table 111). Table VIII. Cross-reaction of anti-6*2705 and anti-6*2703 CTL clones with non-hla-627 Ags" Donor PA DL GM SR TOTAL a-b*2703 o-b*2705 a-b*2705 a.b*2705 u.~*2703 a-8'2703 a-b*2705 a-8'2705 Cross-Reactive Antigen N (%I N (%) N (%) N (%,) N (%) N ("1") N (Y") N (7") B6l 0/15 (33) 5/15 (0) (7) 1/21 1/15 (19) 3/16 (5) 2/29 10/75 (7) (13) 2/36 (6) B60 1/13 (8) 0/14 (0) 0/21 (0) 0/16 (0)(3) 1/36 1/15 (15) 11/73 (33) 10/30 (7) NT B54 0/14 (0) 0/21 (0) 0/16 (0) 0/15 1/58 (0) (4) 1/28 (2) 0/36 (0) NT 855 1/13 (8) 1/20 (5) 0/16 (0) 0/15 (0) 2/283/57 (7) (5) 1/35 (3) NT 856 0/14 (0) 0/20 (0) 2/16 (13) 0/15 (0) 3/57 (4) 1/27 (5) 0/35 (0) 0/18 B7" (0) 0/15 (0) 1/210/15 (5) (0) 0/15 (0) 0/29 (0) 0/77 (0)(3) 1/36 N is the number of CTL reactive/total number of CTL tested with the particular Ag. The corresponding "A values are in parentheses. NT, not tested. Data concerning donor PA are from Ref. 31. Reactivity assignments were as explained in Footnote a lo Table II. Because CM and SR are HLA-B7+, cross-reaction with this Ag was not expected. this donor. In addition, within the very high cross-reaction with B*2705 in both donors, this was higher in DL. Donors DL and GM were also different in the crossreactions of their anti-b*2703 CTL with subtypes relative to their anti-b*2705 CTL (Table VII). Most anti-b*2703 CTL from DL (86%) cross-reacted with B"2705, whereas only 13% of his anti-b*2705 CTL cross-reacted with B*2703. Cross-reactions with other subtypes were similar in both responses from this donor, except for some crossreaction of anti-b*2703 CTL with B"2704 (19%) and B*2706 (5%), which was not observed in his anti-b*2705 response. In contrast, GM, whose anti-b*2705 CTL were frequently cross-reactive, showed little cross-reaction in his anti-b*2703 response except with B*2705. B*2703/ B*2705 cross-reactions were increased among anti- B*2703 relative to anti-b*2705 CTL in this donor, but not as much as in DL. These results indicate that CTL responses against structurally similar HLA-B27 Ags in a single individual are very distinct, and the way in which these responses differ varies greatly among individuals. Cross-reactions with non-b27 HLA antigens These cross-reactions are summarized in Table VIII. Anti- B*2705 CTL clones cross-reactive with HLA-B61 were found from all responders except DL, being most frequent in PA. HLA-B60 was recognized by 10 of 30 anti-b*2705 CTL from donor SR, none from DL or GM, and only 1 of

10 5566 ALLOREACTIVITY AGAINST THE B*2705 AND B*2703 SUBTYPES Table IX. Reaction patterns of anti-b*2705 CTL clones with HLA-B27 subtypes" Reaction Pattern Donor' Number Subtypesb PA DL GM SR Total I 5 6 (33%) 94 (60%)(22%) 4 (27%) 23 (12%) II 5, 1 2 (11%) 2 1 (7%) (11%) 2 (6%) (8%) , (6%) (20%) 3 (6%) 6 (7%) IV 5, 3 2 (1 1 %) 1 (36%) (7%) 12 (28%) 5 20 (24%) V 5, 1,2 3 (9%) 3 (4%) VI 5,1,3 1 (3%) 1 (1%) 1, VI1 5, 1 (6%) 1 (1%) Vlll 5, 2, (33%) (6%) 8 (10%) IX 5, 2, 4 1 (3%) (1 1 %) X 5,4, 6 1 (3%) 1 (1%") XI XI1 1, 5, 2, 3 5, 1, 3,4 1 (6%) 3 1 (7%) 1 (6%) (9%) 5 (6%) 1 (ly01 Xlll 5, 2,4, 6 1 (6%ld 1 (1%) XIV 5, 1,2, 3,4 1 (3%) 1 (1%) xv 3, 5, 1, 2, 6 1 (6%) 1 (1%) XVI 4, 5, 1, 2, 6 1 (6%) 1 (1%") 3, XVll 1, 5, 6 1 (3%) 1 (1%), 2, XVlll 1, 5, (1 1 Yo) (2%) 2 Reaction Patterns Total CTL Ratio CTL differing only in their cross-reactions with non-b27 Ags are grouped together. Numbers indicate HLA-627 subtypes recognized in each reaction pattern group (1 means 6*2701, and so on). Figures are number and, in parentheses, percentages of CTL clones relative to the total analyzed. Data from donor PA are from Ref. 31 and are included for comparlson. "This reaction pattern is tentative because this CTL (122 CRK) was not tested with B*2701+ targets. 13 CTL from PA. These results suggest differences in the cross-reactivity of anti-b*2705 CTL with HLA-B61 and B60 among individuals. However, because 8 of the 10 CTL from SR cross-reacting with B60 showed identical reactivity patterns in Table IV, the possibility that these represent a highly proliferative clonotype in the bulk culture, which was picked up multiple times upon cloning, cannot be ruled out and must await molecular characterization of their TCR. Cold target inhibition involving B*2705+ and B60f or B61+ target cells was performed with 6 of 10 CTL cross-reacting with B60 and 2 CTL cross-reacting with B61 from donor SR. In all cases tested, mutual and complete inhibition was obtained (data not shown), demonstrating the monoclonal nature of the cross-reaction. Anti-B*2705 CTL crossreacting with HLA-B22 (B54, B55, or B56) were observed in donors DL, GM, and SR. CTL 2DLE lysed the B55+ LCL VEN (Table 111). Lysis was inhibited by anti-cd8 and ME1 mab and not by anti- CD4 and EDU-1 mab (data not shown). This LCL was also lysed by CTL 7S15 and 16S15 from SR (Table IV). Confirmation of B55 cross-reactivity by inhibition with mab was not obtained in these cases, but recognition of other HLA-A,B Ags was ruled out from the panel analysis. CTL 10OGRK and 122GRK, from GM (Table 111), and CTL 16S15 from SR (Table IV), lysed the B56+ LCL VOO. Crossreaction with HLA-B56 was confirmed by specific inhibition of lysis with the ME1 mab for CTL 122GRK and 16S15 (data not shown). CTL 36SRC, from donor SR (Table IV), lysed the B54+ LCL TTL, but the Ag responsible for this cross-reaction was not identified. The percentage of CTL cross-reacting with HLA-B22 was less than one-half of those cross-reacting with HLA-B60 or -B61 (Table VIII). Occasional cross-reactions with B60, B61, and 855 were also detected among anti-b*2703 CTL (Tables V and VI). Cross-reactions of B*2703 with B60 and B61 were confirmed by inhibition of cytotoxicity of B60+ and B61+ cells with the W6/32 mab, and by mutual inhibition of the cytotoxicity in cold target inhibition analysis (data not shown). Recognition of other class I Ags on these target cells was ruled out from the panel data. Cross-reaction with HLA-B7 was not observed among anti-b*2705 CTL. Only one anti-b*2703 CTL, from donor DL, cross-reacted weakly with this Ag. Cross-reactions of anti-b27 CTL with HLA-DR2, which were found in the two DR2- individuals examined, have been described in detail elsewhere (30, 32) and will not be discussed here. Diversity and frequency of allospecific clonal reacfion patterns The clonal reaction patterns of anti-b*2705 and anti- B*2703 CTL with HLA-B27 subtypes are summarized in Tables IX and X. The ratio between the number of reaction

11 Journal of Immunology 5567 Table X. Reaction patterns of anti-b*2703 CTL clones with HLA-527 subtypes Pattern Reaction Donor Number Subtypes DL CM TOTAl 2 (1 0%) 10 (48%) 1 (5%) 3 (1 4%) 1 (5%) 2 (1 0% 1 (5%) 1 (5%) 4 (27%) 6 (17%) 7 (47%) 17 (47%) 1 (3%) 1 (7%) 1 (3%) 1 (7%) 4 (11%) 1 (7%) 2 (6%) 2 (6%) 1 (3%) 1 (7%) 2 (6%) Reaction Patterns Total CTL Ratio 0, 38 0, 40 0, 25 See Footnotes a, b, and c to Table IX patterns and that of anti-b*2705 CTL clones from each donor was similar for PA, DL, and SR, and higher for GM (Table IX). This indicates that clonal diversity in the anti- B*2705 responses is similar in the three former donors and somewhat larger in GM, and suggests that additional reaction patterns could be expected only at low frequency if more CTL clones from these donors were analyzed. This ratio was similar in GM and DL for anti-b*2703 CTL clones (Table X). Global diversity among the anti-b*2705 or anti-b*2703 CTL, as estimated by the same ratio, was smaller than for individuals because the most frequent reaction patterns were found in multiple donors. Reaction patterns I and IV, including recognition of only the stimulator subtype and cross-reaction with only B*2703, accounted for 51% of the anti-b*2705 clonotypes (Table IX). Both were found in the four individuals and were in inverse ratio, so that the higher the percentage of reaction pattern I in each donor, the lower was that of reaction pattern IV. This was not observed with any other reaction pattern. Four other reaction patterns (11, 111, VIII, and XI), ranging in frequency from 6 to lo%, accounted for 31% of the clonotypes. They were found in three or four donors, except reaction pattern VIII, which was found only in two individuals. T cell epitope sharing between B*2705 and other subtypes was established from the global frequencies of individual reaction patterns (Table IX). First, 27% of the allospecific epitopes are present only in the stimulator subtype (reaction pattern I). Epitopes shared by B*2705 and only one other subtype accounted for 39% of the clonotypes (reaction patterns 11, 111, and IV), those involving B*2703 (IV) being by far the most frequent within this group. Epitopes shared by three subtypes (reaction patterns V to X) accounted for 18% of the clonotypes, the predominant pattern being that involving B*2705, B*2702, and B*2703 (VIII). Epitopes shared by more than three subtypes (reaction patterns XI to XVIII) accounted for 15% of the clonotypes. The most frequent pattern within this group (XI: 6%) included B*2705, 01, 02, and 03. Cross-reactive clonotypes involving B*2704 and/or B*2706 (VII, IX, X, and XI1 to XVIII) accounted for 13% of the total. Two were the main features of anti-b*2703 T cell responses (Table X). First, epitopes shared only by B*2703 and B*2705 (reaction pattern 11) were, by far, the most frequent ones. Second, relatively few CTL recognized B*2703 private determinants (reaction pattern I:17%), and they were more frequent in GM than in DL. Lack of crossreaction with B*2705 was confirmed by cold target competition analysis for all CTL clones in this group except CTL 74.8GCP. CTL 77GCP did not kill B*2705+ targets, but these efficiently inhibited its lysis of B*2703+ targets in cold target competition assays (Table VI), and was assigned reaction pattern 11. Reaction patterns involving cross-reaction with B*2705 plus other subtypes (V to IX) accounted for 31% of the anti-b*2703 CTL, further indicating the large T cell epitope sharing between B*2703 and B*2705. Differences among individuals in allospecific clonal reaction patterns In the anti-b*2705 response (Table IX), donors DL and PA, and donors GM and SR, shared some features that were different in the other pair. For instance, reaction pat- terns involving B*2704 or B*2706 were relatively frequent in GM and SR, but were not found in PA and DL. Reaction pattern IV was much less frequent in PA and DL than in GM or SR. Differences between DL and PA included the absence in DL and the high frequency (33%) in PA of CTL with reaction pattern VIII, and their frequencies of reaction patterns I and 111. Donor DL was the most narrow responder in this analysis, as 93% of his anti- B*2705 CTL clones cross-reacted with no other or with only one subtype (reaction patterns I to IV). These accounted for 61% of the CTL in each of the other three donors. Donor GM was the broadest responder because he showed the highest number of reaction patterns relative to CTL clones, and the highest percentage (33%) of CTL recognizing more than three subtypes. It may be argued that this results from limited in vitro clonal selection, as these CTL were cloned after primary MLR. However, of five broadly cross-reactive CTL in donor SR, two were obtained after primary (7.8SRY and 47S69), and three af- ter secondary MLR (2SRZ, 37SLG, and 21S15). Thus, broadly cross-reactive CTL are not necessarily selected against upon in vitro stimulation of bulk cultures. Among anti-b*2703 CTL, five of the nine observed reaction patterns, including the same percentage of the major one (11), were found in both responders. However, donorrelated differences were observed (Table X). First, the percentage of CTL against B*2703 private determinants (reaction pattern I) was nearly three times smaller in DL than in GM. Second, CTL cross-reactive with B*2702 or

12 5568 ALLOREACTIVITY AGAINST THE B*2705 AND B*2703 SUBTYPES B*2701 (V, VII, and VIII) were absent or rare in GM, but amounted to 29% of those from DL. Thus, in the anti- B*2703 response, DL was a broader responder than GM, because of his lower reactivity against private determinants, and his higher frequency of cross-reactive clonotypes involving three or more subtypes: reaction patterns V to IX, 38% in DL vs 20% in GM. Discussion Several key issues concerning HLA-B27-specific T cell responses were addressed in our study. First, the degree of T cell epitope sharing among HLA-B27 subtypes. This is important because it defines the role of subtype polymorphism in modulating HLA-B27 antigenicity, and because antigenic relatedness among HLA-B27 subtypes may explain their similar linkage to ankylosing spondylitis (18) if T cells are involved in the pathogeny of this disease (14, 17). A second issue was defining individual differences in the anti-hla-b27 response and correlating them with the HLA phenotype of the responder. Such differences reflect the capacity of individual T cell repertoires to respond to a given T cell epitope, and this might be related to individual differences in susceptibility to disease. A third issue was to compare allospecific T cell responses against B*2705 and B*2703 in the same individuals, to establish the role of the single amino acid change between these two subtypes, which is located in a key position for peptide binding, in shaping HLA-B27 antigenicity. This also provides an immunologic basis for assessing putative differences in the linkage of these two subtypes to disease, as suggested in a previous study (21). The experimental strategy was to compare anti-b*2705 CTL responses from four unrelated individuals, as well as anti-b*2703 responses in two of them, at the clonal level. Studies in the mouse have analyzed fine specificity repertoires in anti-h-2kh responses, as well as genetic and environmental factors influencing their outcome (8). In humans, individual differences in CTL precursor frequency against B*2702, and other class I Ags, have been reported (6), but the nature of the epitopes involved was not addressed. An advantage of analyzing CTL reaction patterns to compare alloreactive responses is that it allows a far greater characterization of clonal fine specificities than CTL precursor analysis. A limitation is the relatively low number of CTL available for comparison. Nevertheless, the similar ratio between the number of clonal reaction patterns and that of CTL clones analyzed in most of the individuals in this study suggests that the most frequent clonotypes from each donor were picked up upon cloning. From the analysis of anti-b*2705 CTL, the most crossreactive subtype, B*2703, shared only about 50% of the allospecific epitopes recognized on B"2705. Cross-reactive patterns excluding B*2703 were not frequent, although together amounted to a sizeable fraction (17 to 28%) of the response in each individual. The antigenically most distant subtypes, B*2704 and B*2706, shared with B* % and 9% of the epitopes, respectively. Although most alloreactive T cells appear to recognize peptides (3,4), T cell cross-reactions do not necessarily imply recognition of identical peptides. Only some of the peptidic amino acid residues interact with the TCR, and peptides putatively involved in subtype cross-reactivity need only to share motifs amenable to recognition by a same TCR. As far as allospecific T cell epitope sharing reflects recognition of identical or similar peptides, or peptide motifs, our data indicate that these can be efficiently presented by B*2705 and other subtypes to the same CTL. Because the target cells used in this analysis were from individuals of different genetic backgrounds, the peptide repertoires of the different cell lines are probably not identical. Therefore, the data are actually a minimum estimate of subtype cross-reactivity at the T cell level. They are in agreement with a comparison of endogenous peptide pools, indicating that the peptide repertoires presented by the B"2701 to B*2706 subtypes were partially overlapping (33). Putative arthritogenic peptides may be among those carrying motifs that can be presented by multiple subtypes to T cells. The most striking feature of anti-b*2703 CTL was their high percentage cross-reacting with B*2705 in each of the two individuals examined, indicating that most of the allospecific epitopes of B*2703 are also present in B"2705. Cross-reactivity between these two subtypes was much higher than with anti-b*2705 CTL from the same responders, suggesting that there are many more B*2705 allospecific epitopes absent from B*2703 than the reverse situation. A likely explanation is in the effect of the amino acid change between these two subtypes on peptide binding. Tyr59 contributes to stabilize the amino-terminal group of B*2705-bound peptides (24). The change to His59 in B*2703 must disrupt this interaction, presumably decreasing the contribution of the peptidic N-terminus to the binding of many peptides. Therefore, those B*2705- bound peptides for which the N-terminus contributes a substantial part of their binding energy will not bind to B*2703 or will do so much less efficiently. Other peptides may bind in a way that they will not be properly presented by this subtype to T cells. A relatively low number (22%) of anti-b*2703 CTL clones did not cross-react with B*2705, indicating that few new epitopes are created by introducing His59. This could happen by three mecha- nisms. First, it could happen by a direct effect on interaction with the TCR. This is questionable because residue 59 is not very exposed in B*2705 (24), although it cannot be ruled out. Second, some peptides that are unable to bind to B*2705 may bind to B*2703. Finally, a same peptide may adopt slightly different conformations on B*2703 and B*2705. This does not require subtype-specific binding and is likely to be a frequent mechanism for creating epitopes in B*2703 that are absent in B*2705. Thus, our data suggest that B*2703 binds a peptide repertoire that is

13 Journal of Immunology 5569 mainly a subset of the B*2705-bound peptides. Some peptides might bind to B*2703 and not to B*2705, but many would bind to B*2703 and B*2705 with different conformations. The most frequent cross-reaction of anti-b*2705 CTL with non-hla Ags was with HLA-B40 (B60 and B61). In two of three donors it predominantly involved B61, which is more homologous than B60 to HLA-B27 (35). Donor SR showed more cross-reaction with B60 than with B61, contrary to PA and GM. Because GM and SR share HLA-B7 and differ at the B locus by having B8 and B44 respectively, one possibility is that the influence of these Ags on TCR selection could be related to the observed differences in B40 cross-reactivity. In contrast to B8, and to the HLA-B Ags from donor PA, B pocket residues of HLA-B44 are identical, except at position 9, to those in B60 and B61 (36). Because the B pocket of a given class I HLA Ag binds a conserved peptidic motif, it is conceivable that the B44 and B40-bound peptide repertoires may be partially overlapping. This could influence B44-mediated positive selection of B27/B40 cross-reactive T cells in donor SR. Moreover, B44 shares Arg97 with B60, this residue being Ser in B61 and Asn in B27. Because this is the only residue among those different between B61 and B60 that is also different between B61 and B27 (3.9, and is located in a position that can influence peptide presentation, identity at this position between B44 and B60 might introduce some bias in TCR selection toward B27/B60 cross-reactivity in SR. Increased susceptibility to ankylosing spondylitis when HLA-B60 is present in HLA-B27+ individuals (34) could be related to the antigenic similarities between these two specificities. The lower cross-reactivity with HLA-B22 than with B40 must be related to the nature of the differences between HLA-B27 and each of these Ags. For instance, B22 differs from B27 and B61 by lacking acidic residues at positions 63 and 163. These changes interfere with the presence of amino-terminal Arg, which is frequent among B27-bound peptides (1, 33, 37), and this could affect T cell cross-reactivity. In addition, there are more differences in pocket B residues in B22 than in B40 relative to B27. This could also impair cross-reactivity, but the high similarity between B22 and B7 in the B pocket (38) might have favored positive selection of B27/B22 cross-reactive TCR in the HLA-B7+ GM and SR donors similarly as proposed for the crossreaction with B60 in SR. Differences among individuals in their anti-b*2705 and anti-b*2703 responses were also found in 1) cross-reactivity with subtypes, 2) frequency of individual reaction patterns, and 3) frequency of clonotypic patterns in the anti-b*2703 relative to the anti-b*2705 response in the same donor. Although other factors can induce individual differences among allospecific T cell responses, the HLA phenotype of the responder must influence their outcome because of the role of HLA Ags in shaping T cell repertories (39). Such influence was suggested in this study by the observation that the two narrower responders, DL and PA, shared HLA-B39, whereas GM and SR shared HLA-B7. Attempts to relate structural features of their HLA class I Ags to the clonotypic reaction patterns showed a striking correlation between the nature of residue 152 in the HLA-B Ags from the responders and the cross-reactions observed. Polymorphism at this position is restricted in class I HLA Ags to either Val or Glu. B*2705, 01,02, and 03 have Val at this position, whereas B*2704 and B*2706 have Glu. HLA-B Ags in PA (HLA-B35, 39) and DL (HLA-B39,44) have Va1152. The only cross-reactions observed with anti-b*2705 CTL from these donors involved class I Ags having Va1152, such as the B*2701 to B*2703 subtypes, B60, B61, and B55. Cross-reaction with Ags having Glu152 were not observed. In previous studies, the precursor frequency of anti-b*2702 CTL cross-reacting with B*2704 and B*2706 in two independent B*2705+, B35+ individuals (both with Va1152) was extremely low (6). Allo-CTL precursor frequencies in these two individuals against B56 (with Va1152) were also clearly higher than against B7 (with Glu152) (5). Furthermore, donor PA responded well to B*2705, but poorly against B*2704 (40). In that same study, other responders with HLA-B Ags having only Val152 also gave a poor anti-b*2704 response, and the only CTL clone characterized recognized an epitope that was largely independent of residue 152 (40). Taken together, the data suggest that the influence of residue 152 in T cell repertoire selection is such that cross-reaction across a change at this position is greatly impaired. Such impairment is not absolute, as shown by some cross-reaction with B*2704, B*2706, and B7 among anti-b*2703 CTL. Cross-reactivity with B"2704, B*2706, and other HLA Ags having Glu152 would be expected in GM and SR because both donors have HLA-B Ags with Glu152 (HLA-B7) and Val152 (B8 in GM, and B44 in SR), thus allowing for CTL crossreacting with Ags both matched and mismatched at this position. The apparently important role of residue 152 in shaping T cell repertories may be related to the very drastic nature of the change and to its location in a position influencing peptide presentation. Moreover, peptide-induced conformational changes in the side chain of Glu152, as shown in H-2Kb (41), can influence accessibility to direct TCR interaction, providing an additional influence of this residue on repertoire selection. The results in this study may be relevant to the mechanism of HLA-B27 association to spondyloarthropathy. T cell epitope sharing among B27 subtypes provides an immunologic basis for the equal association of various subtypes to disease. To the extent to which allospecific CTL responses reflect peptide recognition, our data suggest that most of the peptides bound to B*2703 are also presented by B*2705, although a significant fraction of B*2705- bound peptides are not bound or properly presented to T cells by B*2703. It seems that the peptide repertoire

14 5570 ALLOREACTIVITY AGAINST THE B*2705 AND B*2703 SUBTYPES shared by B*2705 and B*2703, which are the antigenically closest subtypes, is large enough for putative arthritogenic peptides being presented by both subtypes. However, if a negative association of B*2703 to disease was confirmed by additional research, this could imply that the arthritogenic peptide(s) belong to those that cannot be properly presented by this subtype. Additional studies must identify structural features determining differential peptide binding to B*2705 and B*2703. Great differences among responders were demonstrated by multiple criteria and were shown to correlate with HLA-B phenotype, suggesting that they are caused by distinctly shaped TCR repertoires. Individual differences in recognition of HLA- B27-associated T cell epitopes might influence differential susceptibility of HLA-B27+ individuals to disease. Acknowledgments The authors thank J. Martinez (Hospital 12 de Octubre, Madrid) and J. L. Vicario (Centro de Transfusiones de la Comunidad Autdnoma de Madrid) for cell typing and blood reagents, respectively. We also acknowledge the help of our colleagues D. Jaraquemada, P. Aparicio, B. Galocha, D. Barber, G. Marques, J. M. Muelas, and S. Rojo. The Blood Bank laboratories from the Hospital 12 de Octubre and Fundacidn JimCnez Diaz, Madrid, supplied human sera. References 1. Jardetzky, T. S., W. S. Lane, R. A. Robinson, D. R. Madden, and D. C. Wiley Identification of self peptides bound to purified HLA-B27. Nature Hunt, D. F., R. A. Henderson, J. Shabanowitz, K. Sakaguchi, H. Michel, N. Sevelir, A. L. Cox, E. Appella, and V. H. Engelhard Characterization of peptides bound to the class I MHC molecule HLA-A2.1 by mass spectrometry. Science 255: Heath, W. R., K. P. Kane, M. F. Mescher, and L. A. Sherman Alloreactive T cells discriminate among diverse sets of endogenous peptides. Proc. Natl. Acad. Sci. USA 88: Rotzsche, O., K. Falk, S. Faath, and H. G. Rammensee On the nature of peptides involved in T cell alloreactivity. J. Exp. Med. I 74: Breur-Vriesendorp, B. S., and P. Ivanyi Individual differences in the T-lymphocyte response in man to public HLA determinants. Cell. Immunol. 103: Breur-Vriesendorp, B. S., J. Vingerhoed, W. P. Schaasberg, and P. Ivanyi Variations in the T-cell repertoire against HLA antigens in humans. Hum. Immunol. 27:l. 7. Breur-Vriesendorp, B. S., J. Vingerhoed, E. van Twuyver, L. P. de Waal, and P. Ivanyi Frequency analysis of HLA-specific cytotoxic T lymphocyte precursors in humans. Transplantation 51: Sherman, L. A,, and J. R. Maleckar Genetic and environmental regulation of the cytolytic T-lymphocyte receptor repertoire specific for alloantigen. Immunol. Rev. 101: Pereira, P., A. Bandeira, A. Coutinho, M.A. Marcos, M. Toribio, and C. Martinez- Alonso V-region connectivity in T cell repertoires. Annu. Rev. Immunol. 7: Brewerton, D. A,, M. Caffrey, F. D. Hart, D. C. 0. James, A. Nichols, and R. D. Sturrock Ankylosing spondylitis and HL-A27. Lancet i: Schlosstein, L., P. I. Terasaki, R. Bluestone, and C. M. Pearson High association of HL-A antigen,, W27, with ankylosing spondylitis. N. Engl. J. Med. 288: Hammer, R. E., S. D. Maika, J. A. Richardson, J. P. Tang, and J. D. Taurog Spontaneous Inflammatory disease in transgenic rats expressing HLA-B27 and human &m: An animal model of HLA- B27 associated human disorders. Cell Taurog J. D., S. D. Maika, W. A. Simmons, M. Breban, and R. E. Hammer Susceptibility to inflammatory disease in HLA-B27 transgenic rat lines correlates with the level of B27 expression. J. Immunol. 150: Benjamin, R. J., and P. Parham Guilt by association: HLA- B27 and ankylosing spondylitis. Immunol. Today 11: Kingsley G., and J. Sieper Current perspectives in reactive arthritis. Immunol. Today 14: Agganval A,, S. Kumar, R. Jaffe, D. Hone, M. Gross, and J. Sadoff Oral Salmonella: malaria circumsporozoite recombinants induce specific CD8+ cytotoxic T cells. J. Exp. Med. 172: Hermann, E., D. T. Y. Yu, and K. H. Meyer HLA-B27-restricted CD8 T cells derived from synovial fluids of patients with reactive arthritis and ankylosing spondylitis. Lancet 342: Breur-Vriesendorp, B. S., A. J. Dekker-Saeys, and P. Ivinyi Distribution of HLA-B27 subtypes in patients with ankylosing spondylitis: the disease is associated with a common determinant of the various B27 molecules. Ann. Rheum. Dis. 46t Lopez de Castro, J. A HLA-B27 and HLA-A2 subtypes: structure, evolution and function. Immunol. Today 10: Choo, S. Y., L. A. Fan, and J. A. Hansen A novel HLA-B27 allele maps 827 allospecificity to the region around position 70 in the a1 domain. J. Immunol. 147: Hill, A. V., C. E. Allsopp, D. Kwiatkowski, N.M. Anstey, B. M. Greenwood, and A. J. McMichael HLA class I typing by PCR: HLA-B27 and an African B27 subtype. Lancet 337: Rojo, S., P. Aparicio, J. A. Hansen, S. Y. Choo, and J. A. Lopez de Castro Structural analysis of an HLA-B27 functional variant, B27d, detected in American Blacks. J. Immunol. 139: Saper, M. A., P. J. Bjorkmann, and D. C. Wiley Refined structure of the human histocompatibility antigen HLA-A2 at 2.6A resolution. J. Mol. Biol. 219: Madden, D. R., J. C. Gorgd, J. L. Strominger, and D. C. Wiley The three-dimensional structure of HLA-B27 at 2.1 A resolution suggests a general mechanism for tight peptide binding to MHC. Cell 70: Storkus, W. J., D. N. Howell, R. D. Salter, J. R. Dawson, and P. Cresswell NK susceptibility varies inversely with target cell class I HLA antigen expression. J. Immunol. 138: Calvo, V., S. Rojo, D. Mpez, B. Galocha, and J. A. Mpez de Castro Structure and diversity of HLA-B27-specific T cell epitopes. Analysis with site-directed mutants mimicking HLA-B27 subtype polymorphism. J. Immunol. 144: Barnstable, C. J., W. F. Bodmer, G. Brown, G. Galfre, C. Milstein, and A. F. Williams Production of monoclonal antibodies to Group A erythrocytes, HLA and other human cell surface antigens. New tools for genetic analysis. Cell 14: Ellis, S. A,, C. Taylor, and A. M. McMichael Recognition of HLA-B27 and related antigens by a monoclonal antibody. Hum. Immunol Colombani, J., V. Lepage, and J. Kalil HLA monoclonal antibody registry: second listing. Tissue Antigens 22: Aparicio, P., D. Jaraquemada, and J. A. Mpez de Castro Alloreactive cytolytic T cell clones with dual recognition of HLA- B27 and HLA-DR2 antigens.. Selective involvement of CD8 in their class I-directed cytotoxicity. J. Exp. Med. 165: Aparicio, P., D. Jaraquemada, S. Rojo, and J. A. Ldpez de Castro Clonal heterogeneity of HLA-B27 cellular allorecognition. Delineation of immunodominant sites. Eur. J. Immunol. 18: Mpez, D., D. F. Barber, J. A. Villadangos, and J. A. L6pez de Castro Cross-reactive T cell clones from unrelated individuals reveal similarities in peptide presentation between HLA-B27 and HLA-DR2. J. Immunol. 150:2675.

15 Journal of Immunology Rojo, S., F. Garcia, J. A. Villadangos, and J. A. Mpez de Castro Changes in the repertoire of peptides bound to HLA-B27 subtypes and to site-specific mutants inside and outside pocket B. J. Exp. Med. I77: Robinson, W. P., S. M. van der Linden, M. A. Khan, H. U. Rentsch, A. Cats, A. Russell, and G. Thomson HLA-Bw60 increases susceptibility to ankylosing spondylitis in HLA-B27+ patients. Arthritis Rheum. 32tl Aparicio, P., S. Rojo, D. Jaraquemada, and J. A. Lbpez de Castro Fine specificity of HLA-B27 cellular allorecognition. HLA- B27f is a functional variant distinguishable by cytolytic T cell clones. J. Immunol. 139t Buxton, S. E., R. J. Benjamin, C. Clayberger, P. Parham, and A. M. Krensky Anchoring pockets in human histocompatibility complex leukocyte antigen (HLA) class I molecules: Analysis of the conserved B ( 45 ) pocket of HLA-B27. J. Exp. Med. 175: Rotzschke, O., K. Falk, S. Stevanovic, V. Gnau, G. Jung, and H. G. Rammensee Dominant arornatic/aliphatic C-terminal anchor in HLA-B*2702 and B*2705 peptide motifs. Immunogenetics 39: Hildebrand, W. H., J. A. Madrigal, A.M. Little, and P. Parham HLA-Bw22: a family of molecules with identity of HLA-B7 in the a1 -helix. J. Immunol. 148: Akolkar, P. N., B. Gulwani-Akolkar, R. Pergolizzi, R. D. Bigler, and J. Silver Influence of HLA genes on T cell receptor V segment frequencies and expression levels in peripheral blood lymphocytes. J. Immunol. 150: Jaraquemada, D., B. Galocha, P. Aparicio, S. Rojo,V. Calvo, and J. A. Upez de Castro Modulation on immunogenicity by HLA-B27 subtype polymorphism. Eur. J. Immunol. 18: Frernont, D. H., M. Matsumura, E. A. Stura, P. A. Peterson, and I. A. Wilson Crystal structures of two viral peptides in complex with murine MHC class I H-2Kh. Science 257: Vilches, C., R. de Pablo, and M. Kreisler Nucleotide sequence of HLA-B*2706. Immunogenetics 39: Lbpez, D., S. Rojo, V. Calvo, and J. A. L6pez de Castro Peptide-presenting similarities among functionally distant HLA-B27 subtypes revealed by alloreactive T lymphocytes of unusual specificity. J. Immunol. 148:996.