HLA CLASS I1 SEQUENCE POLYMORPHISMS AND SUSCEPTIBILITY TO RHEUMATOID ARTHRITIS IN GREEKS

Transcription

1 749 HLA CLASS I1 SEQUENCE POLYMORPHISMS AND SUSCEPTIBILITY TO RHEUMATOID ARTHRITIS IN GREEKS The HLADRP SharedEpitope Hypothesis Accounts for the Disease in Only a Minority of Greek Patients KYRIAKI A. BOKI, GABRIEL S. PANAYI, ROBERT W. VAUGHAN, ALEXANDROS A. DROSOS, HARALAMPOS M. MOUTSOPOULOS, and JERRY S. LANCHBURY Objective. In Northern Europeans, rheumatoid arthritis (RA) is strongly associated with a relatively conserved pentapeptide sequence of HLADRP found notably in the HLADR4 subtypes Dw4 and Dw14 and in DRl. A previous serologic study of HLA class I1 polymorphism in a Greek population with RA failed to show significant associations with any antigen. Methuh. We characterized HLADRB polymorphism in Greek patients with RA and in control subjects by restriction fragment length polymorphism analysis. Allelic DRB subtypes were examined by polymerase chain reaction amplification and oligonucleotide hybridization. Resulls. DNA analysis in the RA patients showed that although individual HLADR allelic associations were weak, a relatively conserved HLADRf? motif was significantly associated with RA in this population of From the Molecular Immunogenetics Unit and the Rheumatology Unit, Division of Medicine, United Medical and Dental Schools, London, United Kingdom, and the Department of Internal Medicine, School of Medicine, University of Ioannina, Ioannina, Greece. Supported by Arthritis and Rheumatism Council project grant US. Dr. Boki is recipient of a EULAR Scholarship. Kyriaki A. Boki, MD: Molecular Immunogenetics Unit and Rheumatology Unit, Division of Medicine, United Medical and Dental Schools; Gabriel S. Panayi, MD, DSc: Rheumatology Unit, Division of Medicine, United Medical and Dental Schools; Robert W. Vaughan, MSc: Molecular Immunogenetics Unit, Division of Medicine, United Medical and Dental Schools; Alexandros A. Drosos, MD: Department of Internal Medicine, School of Medicine, University of Ioannina; Haralampos M. Moutsopoulos, MD: Department of Internal Medicine, School of Medicine, University of Ioannina; Jerry S. Lanchbury, PhD: Molecular Immunogenetics Unit and Rheumatology Unit, Division of Medicine, United Medical and Dental Schools. Address reprint requests to Jerry S. Lanchbury, PhD, Division of Medicine, United Medical and Dental Schools, 4th Floor Hunt's House, Guy's Hospital, London SEl 9RT, UK. Submitted for publication March 13, 1991; accepted in revised form January 2, Greek patients. The third hypervariable region amino acid sequences QRRAA, QKRAA, or RRRAA were found in the HLADRP1 of 43.5% of the RA patients versus 15.5% of the controls (uncorrected P = O.OOOO4). Conclusion. Sequences shown to influence susceptibility to RA in patients in the UK also play a role in patients in Greece. However, 57% of Greek patients lack the putative HLADRP motif, which suggests that considerable immunogenetic heterogeneity underlies disease susceptibility in this population. Although the pathogenesis of rheumatoid arthritis (RA) is unknown, findings from population and family studies indicate that both genetic and environmental factors are important in the development of the disease. The best characterized genetic component is the association of alleles of the major histocompatibility complex (MHC) with RA, in particular, the alloantigen HLADR4. The DR4 association with RA has been confirmed in many populations, including Caucasoids of Northern European descent, Japanese, and American blacks. However, in several ethnic groups, notably, Indians and Israeli Jews, RA is associated with DR1 rather than DR4, and an increased frequency of DRl is observed in Northern Europeans who have RA and are DR4 negative. HLADR4 was originally defined as a public serologic specificity associated with several cellularly defined HLADw variants in mixed leukocyte culture (MLC). HLADR molecules are composed of a nonpolymorphic a chain that is noncovalently linked to a polymorphic p chain. Recent studies have demonstrated considerable molecular complexity of the DR4Pl chain and the corresponding DR4B1 gene. Each classic MLC variant is associated with a unique Arthritis and Rheumatism, Vol. 35, No. 7 (July 1992)

2 750 BOKI ET AL DR4Pl primary sequence (1,2), and a number of new sequence variants have also been described. We and others have shown by molecular analysis that only a restricted group of DR4Pl sequences confer susceptibility to RA in Northern European and North American Caucasoids (35). The associations of DRl and DR4 with RA have been rationalized in terms of shared conserved HLA DRP epitopes, which are carried by DRI and the RAsusceptible subtypes of DR4 (6). There exist, however, populations such as the Greeks in whom no serologic HLADR association with RA has been demonstrated (7). To clarify whether the disease in this group is mediated by susceptibility sequtmces characterized previously, we examined HLA class I1 DNA polymorphisms among Greek patients with RA and compared them with those in healthy ethnically matched control subjects. PATIENTS AND METHODS Patients and controls. Ninetytwo adult Greek patients classified as having RA, according to the 1987 revised criteria of the American College of Rheumatology (4CR; formerly, the American Rheumatism Association) (8), were selected for study. The patients received medical care at the Department of Internal Medicine in Ioannina (Greece). Seventy patients were positive for serum rheumatoid factor, and 22 were seronegative; 75 were female and 17 male. Eightyfour healthy subjects with the same mean age and ethnic origin as the patients served as controls. Restriction fragment HLA class I1 typing. Blood samples from patients and controls were drawn into tubes containing EDTA and frozen at 20 C prior to the extra.ction of DNA. Isolation of high molecular weight genomic DNA and genotyping for DRB and DQA were performed according to established methods (9,lO). Definition of DQA polymorphism in this context is useful in distinguishing certain closely related DRw6 haplotypes. HLADR and DQ genotyping was performed by sequential hybridization of nylon membranes with a DRB probe (Sac IHind 111 complementary DNA [cdna] fragment of DRB [ 1 l]), and a DQA probe (Apa I cdna fragment of DQA [12]). The cdna probes were gifts of Dr. P. A. Peterson (Scripps Clinic, La Jolla, CA). DR4B1 polymerase chain reaction (PCR) amplification and analysis. Subtypes of HLADR4Pl were investigated at the nucleotidesequence level using a PCR amplification technique that is dependent on the presence of a DR4 template (13). PCR amplification of DNA from DRCpositive individuals yielded a 263basepair fragment of the second exon which corresponds to a portion of the first protein domain. The gel was Southern blotted onto nylon membrane according to standard methods (Amersham, Buckinghamshire, UK). Oligonucleotide probes used for the discrimination of DR4 subtypes have been described (13). In addition, a 17mer probe 102G (5 AGAGGAGTCCGTGCGCT) was designed to discriminate the serinefortyrosine substitution at position 37 that is present in the DwKT2 subtype. Hybridization and washing conditions, xray film exposure, and positive and negative controls for amplification and hybridization have been described previously (13). HLADRB gene amplification and analysis. Subtypes of HLADRI and the HLADRw6 complex were also examined by PCR amplification and oligonucleotide hybridization, using a primer set which amplifies a 239base pair fragment of the second exon of all DRB genes, as previously described (14). Oligonucleotide probes used in typing HLA DRB alleles in Greeks have been described previously (14). HLADRB subtypes were assigned by specific oligonucleotide hybridization patterns previously demonstrated on HLAhomozygous cell lines (14). Statistical analysis. Comparisons of genotype frequency between the patient and control groups were performed using a 1tailed Fisher s Ptest. Relative risks (RR) were calculated according to the WoolfHaldane method, and etiologic fractions (EF) according to standard methods. P values for HLA associations are presented uncorrected, since they are consistent with those previously reported in other RA populations. RESULTS Table 1 shows the frequencies of HLADR genotypes derived by restriction fragment length polymorphism (RFLP) analysis of the Greek patient and control groups. Although in most cases, it was possible to identify restriction fragments which corresponded to northern European HLADR genotypes, a number of band patterns could not be unambiguously assigned, and these were included in the unclassified group. Genotypes corresponding to HLADR4 and HLADRw 10 were significantly increased among the RA patient group: 22.8% of RA patients were positive for DR4 compared with 11.9% of controls (P = 0.04, RR 2.1, EF 0.12), whereas 6.5% of the patients carried the HLADRw 10 genotype compared with none of the controls (P = 0.02, RR 12.7, EF 0.06). HLADRl was increased, although not significantly, in the patient group (21.7%, versus 11.9% in controls; RR 2.0, EF 0.11). Two HLA genotypes were markedly reduced in frequency among the RA patients. DRwll was significantly less frequent among the RA group (36.9%, versus 52.4% in controls; P = 0.03), and DRw16 was present in 35.7% of the controls and only 23.9% of the patients, a difference which did not reach statistical significance (P = 0.06). The frequency of the unclassified group was similar in both groups. To examine the nature of the DR4 association with RA in Greece, the PCR was used to selectively

3 HLADRP AND RA SUSCEPTIBILITY IN GREEKS 75 1 Table 1. HLADR genotype associations with rheumatoid arthritis (RA) in Greeks* ~~~ ~~ ~ ~ ~ Control RA patients subjects Fisher's Relative Etiologic (n = 92) (n = 84) P risk fraction ~~ DR 1 DR2 DRwlS DRwl6 DR3 DR4 DRwll DRwl2 DRw6 DR7/9 DRw8 DRwlO Unclassified 20 (21.7) 40 (43.5) 12 (13.0) 22 (23.9) 20 (21.7) 21 (22.8) 34 (36.9) l(1.1) 8 (8.7) 9 (9.8) 5 (5.4) 6 (6.5) 8 (8.7) 10 (11.9) 40 (47.6) 11 (13.1) 30 (35.7) 13 (15.5) 10 (11.9) 44 (52.4) I (1.2) 13 (15.5) 12 (14.3) 3 (3.6) 0 (0.0) 7 (8.3) (0.06) (0.06) I.o * Values are the number (%) positive, except where noted otherwise. Unclassified genotypes were those which could not be unambiguously assigned. = not significant. amplify DR4 sequences from those individuals who were positive for the HLADR4 RFLP genotype. DNA from 21 RA patients and 10 control subjects was amplified as described in Patients and Methods. Figure 1 shows results obtained after DR4 amplification and hybridization with a panel of oligonucleotide probes which, in this system, detect DR4 subtypes. Patient and normal control DNA was amplified alongside homozygous typingcell DNA, which represented 6 different DR4, Dw specificities. All oligonucleotide reaction patterns could be interpreted in terms of the previously described variants of DR4. Figure 1. DNA analysis of HLADR4 subtypes in Greek patients with rheumatoid arthritis (RA) and in healthy control subjects. Representative Southern blots of polymerase chain reactionamplified DRBl products using DR4BIspecific primer 337C and universal DRB primer 336C. This primer pair amplifies a 263basepair fragment from DR4BI second exons only. Duplicate filters were hybridized with oligonucleotides detecting Dwspecific third hypervariable region sequences and with oligonucleotides informative for the position86 dimorphism. Probes are shown at the left; Dwassociated specificities are shown at the right. Control homozygous typing cells of known HLADR4, Dw types (lanes l4), control subjects (lanes 712), and RA patients (lanes 1320). together with DR and inferred DR4, Dw types, are as follows: lane I, MTF DR4, Dw14.1 (0404); lane 2, Priess DR4, Dw4 (0401); lane 3, JHAF DR4, Dw13.2 (0407); lane 4, YAR DR4, DwlO (0402); lane 5, SJAH DR4, Dw15 (0405); lane 6, KT2 DR4, DwKT2 (0406); lane 7, DR4,wl I, DwlO (0402); lane 8, DR4,wl I, Dwl5 (0405); lane 9, DR4,w16, Dw13.1 (0403); lane 10, DR4,7, Dw15 (0405); lane 11, DR1,4, Dw13.1 (0403); lane 12, DR4,w15, Dwl5 (0405); lane 13, DR4,wll. Dw4(0401); lane 14, DR4, Dw13.1 (0403); lane 15, DR4,wll, Dw13.1 (0403); lane 16, DR4,wl1, Dw13.1 (0403); lane 17, DR4,7, Dw13.1 (0403); lane 18, DR3,4, Dw14.1 (0404); lane 19, DR3.4, Dw4 (0401); lane 20, DR4,w14(Dw9), Dwl5 (0405).

4 752 BOKI ET AL Table 2. DR4 and DRI subtype associations with rheumatoid arthritis (RA) in Greeks* QKRAA and Dw4 Dw14.1 Dw13.1 DwlO DwlS Dwl Dw20 QKRAA and QRRAA and s QRRAA RRRAA RA patients (n = 92) 5 (5.4) 2 (2.2) 5 (5.4) 0 (0.0) 9 (9.8) 16 (17.4) 3 (3.3) 36 (39.1) 40 (43.5) Control subjects (n = 84) 0 (0.0) 0 (0.0) 5 (6.0) 2 (2.4) 3 (3.6) 8 (9.5) 2 (2.4) 13 (15.5) 13 (15.5) Fisher's P 0.04 (0.09) o.oooo4 Relative risk Etiologic fraction * Values are the number (76) positive, except where noted otherwise. = not significant. The frequencies of DR4 sequencedefined subtypes and corresponding inferred Dw types are given in Table 2. The Dw4 (0401) and Dw14.1 (0404) afleles were absent from the Greek control population (n = 84), but present at low frequency in the RA group, such that Dw4 was weakly but significantly associated with RA (P = 0.04, RR 10.6, EF 0.05). Dwl5 (0405) was more common among patients than controls (9.8% versus 3.6%), but this increase failed to reach statistical significance (P = 0.09, RR 2.7, EF 0.07). Dw13.1 (0403) was present at approximately equal frequency in both groups, whereas DwlO (0402) was absent from the RA patient group but present at a frequency of 2.4% among the controls. The DR4 subtypes Dw14.2 (0408), Dw13.2 (0407), and DwKT2 (0406) were not observed in either population. HLADRI is also heterogeneous with two variants that are distinguished by amino acid sequence polymorphisms in the DRIP1 chain. The two variants Dwl (DRB1*0101) and Dw20 (DRB1*0102) differ in sequence at amino acids 85 and 86. It was possible that the nonsignificant association of DRl with RA in this population represented a bias toward one or the ather DRI variant. In order to clarify the position of DR1 variants in Greek patients with RA, we examined DRlassociated polymorphisms, including those at positions 85 and 86, in the RA patients and the controls. Individuals positive for an HLADRI RFLP genotype were subtyped by amplification of gencimic DNA, followed by hybridization with a range of oligonucleotide probes for DRB. Results of DRIrelated oligonucleotide analysis are summarized in Table 2. Dwl was the predominant subtype in the control group, with a genotype frequency of 9.5% compared with 2.4% for Dw20. Among the RA patients, IlwI appeared to be selectively enriched compared with Dw20 (17.4% versus 3.3%), although the numbers were too few to reach statistical significance. Ampli fied DNA from one individual positive for DR1 associated restriction fragments failed to hybridize with oligonucleotide probes directed to DRl third hypervariable region sequences. The HLADRPl chains of DR1 and the Dw4. Dw14, and Dw15 variants of DR4 share a conserved third hypervariable region pentapeptide sequence, with a single lysine for arginine substitution at position 71 in the Dw4 chain. We analyzed the weak associations of DRI and subtypes of DR4 with RA in Greek subjects in terms of this conserved sequence. Table 2 shows that the QKRAA and QRRAA (singleletter code) amino acid sequences were common to 39.1% of the RA patients and 15.5% of the healthy controls (P = , RR 3.4, EF 0.28). The other significant positive association with RA in this Greek population was that of DRw10. The DRPI chain of this antigen carries the sequence RRRAA at the homologous position of amino acids When the QKRAA, QRRAA, and RRRAA sequences of the third hypervariable region were analyzed together, 43.5% of the RA patients were included, compared with 15.5% of the controls, which was highly statistically significant (P = ). The QRRAA sequence is also found in the third hypervariable region of a DRw6 variant called DRB I * 1402 (associated with the Dw 16 T celldefined specificity). It was of interest to establish whether this allele played any role in susceptibility to RA in Greek patients. It is often difficult to distinguish RFLPs associated with DR3 and those of the DRw6 complex, especially when heterozygous combinations of these alleles exist. We therefore amplified DNA from patients and controls who were positive for DR3 and DRw6 RFLPs, and screened by hybridization with an oligonucleotide panel designed to discriminate subtypes associated with these genotypes. The DRBI"1402 was not detected in either the patient or

5 HLADRP AND RA SUSCEPTIBILITY IN GREEKS 753 the control group, which suggests that it plays no role in susceptibility to RA in Greeks. DISCUSSION It appears that in specific racial or ethnic groups, either DR4 or DRl, or both, mediate susceptibility to RA. The Greek population is of particular interest since it has been reported that there is no association of any HLA class I1 antigen with RA (7). Our findings show that previously characterized RAassociated alleles mediate susceptibility in a minority of Greek patients, since both HLADRl and DR4 were present at increased frequency among the patients. Although the difference for DR1 did not reach statistical significance, the etiologic fractions for both genotypes were virtually identical (0.11 and 0.12, respectively), suggesting that DRl and DR4 are equally weighted in terms of disease predisposition in Greeks. This contrasts with the situation in Northern Europeans, in whom DR4 is by far the stronger association (3). The other major positive association among the Greek patients was with DRwlO, which was absent among the control group but accounted for 6.5% of patients, with a relative risk of This specificity is relatively rare among Northern European control subjects but has been implicated in RA susceptibility (15). Although 11 DRPl sequence variants belonging to the DR4 family have been officially recognized, only the Dw4 (0401) and Dw14.1 (0404) alleles exist at frequencies greater than 10% among healthy Northern Europeans, and it is these 2 variants which are largely responsible for the DR4 association with RA in that population. Analysis of the sequencedefined DR4P1 alleles among Greek controls showed the Dw4 and Dw14.1 alleles to be absent from the sample tested. The commonest DR4 allele among Greek controls was Dw13.1 (0403), followed by Dw15 (0405) and DwlO (0402). The frequency distribution among the Greek patient group, however, was somewhat different. The Dw15 allele was relatively enriched among the patients, and both Dw4 and Dw14.1 were present, with the frequency of Dw4 reaching statistical significance. There was no selection for Dw13.1, and DwlO was absent from the RA group. These observations, although based on relatively small frequency changes of DR4 subtypes, are consistent with previous findings in Northern Europeans and provide further evidence that RA is mediated mainly by the Dw4, Dw14.1, and Dwl5 subtypes of DR4. HLADRP alleles that have a positive association with RA, namely, DRI, DR4 Dw4, Dw14, and Dw15 share an area of relatively conserved sequence in their third hypervariable region. The pentapeptide sequence QRRAA or QKRAA (singleletter amino acid code) at positions 7074 appears to be the MHC component that is primarily associated with RA (39, which confirms the original hypothesis reported by Gregersen et a1 (6). RA in Greece is associated with the QRRANQKRAA motif, which is present in 39.1% of patients and 15.5% of controls. HLADRw10 is also significantly increased among Greek patients. The third hypervariable region sequence of the HLADRP chain that corresponds to DRw10 at this point is RRRAA. Thus, the strongest association with RA in Greeks is apparent when data are analyzed in terms of a QRRANQKRAA/RRRAA motif (RR 4.0, EF 0.33). This region of the HLADRPI first domain is known to influence T cell responses (1,2), and there is support for the concept that several of the alleles in question share functional epitopes (16). Our results show that RA in Greece is associated with the same HLADRP alleles which confer susceptibility in Northern European Caucasoids. However, whereas 83% of Northern European patients carry the HLADRP motif (3), it is found in only 43.5% of Greek patients. Thus, most Greek patients with RA lack putative HLA epitopes that are believed to play an important susceptibility role in populations in northern Europe, the United States, and Japan. How can we explain the observation that HLAmediated susceptibility appears to be relevant to only a minority of Greek patients? Findings from pedigree, twin, and siblingpair analyses suggest that a large proportion of the genetic component predisposing to RA maps outside the HLA region, and that HLA susceptibility alleles are neither necessary nor sufficient for development of the disease (17). It is clear that in northern Europe and the US, where most population studies have been carried out, 1520% of RA patients lack the wellcharacterized HLAsusceptibility alleles. It is possible, therefore, that the majority of Greek patients with RA correspond to this group. Unfortunately, it is not within the scope of this population study to address the question of whether the overall genetic component determining RA is less in Greece than elsewhere or whether nonhlasusceptibility loci play a greater role in Greek populations with RA. The suggestion that HLADR4 serves as a marker for clinical severity of RA rather than overt

6 754 BOKI ET AL susceptibility is a controversial notion, but might explain the HLA distribution differences between Northern and Southern Europeans if the popuhtions differed clinically. Although both an association and a lack of association of DR4 with severity of articular disease have been demonstrated, consistently high frequencies of DR4 have been associated with extraarticular manifestations such as rheumatoid vasculitis and Felty s syndrome (18,19). We have recently shown that Felty s syndrome in the UK is highly associated with the Dw4 and Dw14.1 subtypes of DR4, which are rare among Greek control subjects (20). In light of these observations, it is interesting to consider that Felty s syndrome is unknown as a clinical entity in Greece (unpublished observations) and may indicate significant HLArelated clinical differences between Greeks and Northern Europeans. These possibilities are presently under investigation. The definition of a universal paradigm describing the HLA component of susceptibility to RA has been a potentially valuable advance, since it provides a rationale for the development of specific immunotherapeutic approaches such as MHCblocking peptides. We have shown that HLADRP alleles bearing third hypervariable region sequences known to.influence susceptibility to RA in the UK also play a role in RA susceptibility in Greece. The majority of Greek patients, however, lack these sequences, and therefore will not express conserved HLA epitopes. This finding will have important implications for the use of allelespecific peptide therapies in Greece and in countries with populations of similar genetic background. ACKNOWLEDGMENT We are grateful to Dr. B. P. Wordsworth for critical review of the manuscript. REFERENCES Cairns JS, Curtsinger JM, Dahl CA, Freeman S,,4lter BJ, Bach FH: Sequence polymorphism of HLA DRPI alleles relating to Tcellrecognised determinants. Nature 317:166168, 1985 Gregersen PK, Shen M, Song QL, Merryman P, Degar S, Seki T, Maccari J, Goldberg D, Murphy H, Schwenzer J, Wang CY, Winchester RJ, Nepom GT, Silvzr J: Molecular diversity of HLADR4 haplotypes. Proc Natl Acad Sci USA 83: , 1986 Wordsworth BP, Lanchbury JSS, Sakkas LI, Welsh KI, Panayi GS, Bell JI: HLADR4 subtype frequencies in rheumatoid arthritis indicate that DRBl is the major susceptibility locus within the HLA class I1 region. Proc Natl Acad Sci USA 86: , Nepom GT, Byers P, Seyfried C, Healey LA, Wilske KR, Stage D, Nepom BS: HLA genes associated with rheumatoid arthritis: identification of susceptibility alleles using specific oligonucleotide probes. Arthritis Rheum 32:1521, Gao X, Olsen NJ, Pincus T, Stastny P: HLADR alleles with naturally occurring amino acid substitutions and risk for development of rheumatoid arthritis. Arthritis Rheum 33:939946, Gregersen PK, Silver J, Winchester RJ: The shared epitope hypothesis: an approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis. Arthritis Rheum 30: , Papasteriades CA, Kappou ID, Skopouli FN, Barla MN, Fostiropoulos GA, Moutsopoulos HM: Lack of HLAantigen association in Greek rheumatoid arthritis patients. Rheumatol Int 5:201203, Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper, Healey LA, Kaplan SR, Liang MH, Luthra HS, Medsger TA Jr, Mitchell DM, Neustadt DH, Pinals RS, Schaller JG, Sharp JT, Wilder RL, Hunder GG: The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 31:315324, Bidwell JL, Jarrold EA: HLADR allogenotyping using exonspecific cdna probes and application of rapid minigel methods. Mol Immunol 23: , Bidwell JL, Bidwell EA, Laundy GJ, Klouda PT, Bradley BA: Allogenotypes defined by short DQa and DQp cdna probes correlate with and define splits of HLADQ serological specificities. Mol Irnmunol24: , Gustafsson K, Wiman K, Emmoth E, Larhammar D, Bohme J, HyldigNielsen JJ, Rome H, Peterson PA, Rask L: Mutations and selection in the generation of class I1 histocompatibility antigen polymorphism. EMBO J 3~ , Scheming L, Larhammar D, Bill P, Wiman K, Jonsson AK, Rask L, Peterson PA: Both a and p chains of HLADC class I1 histocompatibility antigens display extensive polymorphism in their aminoterminal domains. EMBO J 3:447452, Lanchbury JSS, Hall MA, Welsh KI, Panayi GS: Sequence analysis of HLADRB1 subtypes: additional first domain variability is detected by oligonucleotide hybridization and nucleotide sequencing. Hum Immunol 27: , Vaughan RW, Lanchbury JSS, Marsh SGE, Hall MA, Bodmer JG, Welsh KI: The application of oligonucleotide probes to HLA class I1 typing of the DRB subregion. Tissue Antigens 36:149155, Sanchez B, Moreno I, Magarino R, Garzon M, Gonzalez MF, Garcia A, NunezRoldan A: HLADRwIO confers

7 HLADRp AND RA SUSCEPTIBILITY IN GREEKS 755 the highest susceptibility to rheumatoid arthritis in a Spanish population. Tissue Antigens 36: , Goronzy J, Weyand CM, Fathman CG: Shared T cell recognition sites on human histocompatibility leukocyte antigen class I1 molecules of patients with seropositive rheumatoid arthritis. J Clin Invest 77: , Deighton CM, Walker DJ: The familial nature of rheumatoid arthritis. Ann Rheum Dis 50:6265, Ollier W, Venables PJW, Mumford PA, Maini RN, Awad J, Jaraquemada D, D Amaro J, Festenstein H: HLA antigen associations with extraarticular rheumatoid arthritis. Tissue Antigens , Westedt ML, Breedveld FC, Schreuder GMT, D Amaro J, Cats A, DeVries RRP: Immunogenetic heterogeneity of rheumatoid arthritis. Ann Rheum Dis , Lanchbury JSS, Jaeger EEM, Sansom DM, Hall MA, Wordsworth BP, Stedeford J, Bell JI, Panayi GS: Strong primary selection for the Dw4 subtype of DR4 accounts for the HLADQw7 association with Felty s syndrome. Hum Immunol32:5664, 1991