Mapping MHC-Encoded Susceptibility and Resistance in Primary Sclerosing Cholangitis: The Role of MICA Polymorphism

GASTROENTEROLOGY 2001;120:1475 1482 Mapping MHC-Encoded Susceptibility and Resistance in Primary Sclerosing Cholangitis: The Role of MICA Polymorphism SUZANNE NORRIS,* ELLI KONDEATIS,* ROBERT COLLINS,* JACK SATSANGI, MIKE CLARE,* ROGER CHAPMAN, HENRY STEPHENS,* PHILLIP HARRISON,* ROBERT VAUGHAN,* and PETER DONALDSON *Guy s, King s & Thomas (GKT) Combined Medical Schools, London; Gastroenterology Unit, John Radcliffe Hospital, Oxford; and Centre for Liver Research, University of Newcastle, Newcastle, England Background & Aims: Recent studies suggest that major histocompatibility complex encoded susceptibility to primary sclerosing cholangitis (PSC) maps to the HLA B TNFA region on chromosome 6p21.3. Methods: The present study uses a standard polymerase chain reaction protocol to investigate the 16 common alleles of the MICA locus as candidates in 2 patient populations (King s College Hospital, London, and John Radcliffe Hospital, Oxford). Results: The MICA*002 allele was found in 4 of 62 (6.4%) patients and none of 50 patients vs. 41 of 118 (35%) controls (pc 0.00018, odds ratio [OR] 0.12, and P 0.0000016, OR 0.0, respectively). Overall, the MICA*008 allele was more common in PSC (gene frequency 66% vs. 48% of controls, P 0.0023, OR 2.11). However, unlike MICA*002 in which the difference was a result of the absence of MICA*002 heterozygotes, the MICA*008 association may be caused by an increased frequency of MICA*008 homozygosity in patients (58% vs. 22%, pc 0.000015, OR 5.01 and 58% vs. 22%, P 0.0000056, OR 4.51, respectively). Though MICA*008 is found on the ancestral 8.1 haplotype, stratification analysis indicates that this association is independent of B8 and other HLA haplotypes associated with PSC. Conclusions: The MICA*002 allele has a strong dominant effect in reducing the risk of PSC, whereas the increased risk of disease associated with MICA*008 may be a recessive effect requiring 2 copies of the MICA*008 allele. Primary sclerosing cholangitis (PSC) is a chronic progressive inflammatory disease of the biliary system resulting in obliterative strictures of the intrahepatic and extrahepatic bile ducts and ultimately, cirrhosis. PSC is strongly associated with inflammatory bowel disease (IBD) with 70% to 100% of patients reported to have concomitant IBD, mostly ulcerative colitis (UC). 1 Although the etiology of PSC is unknown, a number of factors suggest that immune mechanisms are involved in the disease process: complement activation, high levels of circulating nonorgan autoantibodies and immunoglobulins, and lymphocytic infiltration in the portal tracts. 2 4 HLA genotyping studies suggest that there is a strong genetic component to this disease. Three different HLA haplotypes are associated with susceptibility to PSC: A1-B8-TNFA*2-DRB3*0101- DRB1*0301-DQB1*0201, DRB3*0101-DRB1*1301- DQB1*0603, and DRB1*1501-DQB1*0602, whereas DRB1*0401-DQB1*0302 may be associated with disease resistance. 5 10 Although there is a general consensus about these haplotypes, there is controversy concerning which allele (or alleles) within each haplotype may form the primary association. Some researchers favor the DRB1*0301 and DRB1*1301 alleles, whereas others believe that these HLA class II (DRB1) alleles are simply linked markers for a stronger association with the B8 and TNFA*2 alleles in the HLA class I/III regions of chromosome 6p21.3. 6,9 11 Whatever the outcome of this debate, the strongest associations so far described, DRB3*0101 and TNFA*2, account for only 53% and 58%, respectively, of patients. Linkage with other genes in the HLA region may also account for the observed associations, and a prime candidate is the major histocompatibility complex (MHC) class I chain related (MIC) gene family that maps between HLA B and TNFA approximately 46.4 kilobase and 141.2 kilobase centromeric to the HLA-B locus. 12 The MIC gene family consists of 5 members (MICA, MICB, MICC, MICD, MICE), of which only MICA and Abbreviations used in this paper: AAU, acute anterior uveitis; MHC, major histocompatibility complex; MIC, MHC class I chain-related; NEC, northern European Caucasoid; NK, natural killer; OR, odds ratio; PCR, polymerase chain reaction; PSC, primary sclerosing cholangitis. 2001 by the American Gastroenterological Association 0016-5085/01/$35.00 doi:10.1053/gast.2001.24041

1476 NORRIS ET AL. GASTROENTEROLOGY Vol. 120, No. 6 MICB encode expressed proteins. 13,14 MICA and MICB have been reported to be stress-induced antigens that are recognized by and capable of inducing immune responses involving T cells and natural killer (NK) cells 15 independent of conventional MHC class I antigen processing. 16,17 The tissue distribution of the MICA gene product seems to be limited to gastrointestinal and thymic epithelial cells, and expression may be induced by heat shock. 16 Although the precise role of MIC molecules is not fully understood, it has been postulated that they may function as stress-induced self antigens and expression of MIC may signal cell infection or transformation. This hypothesis is further supported by the expression of MIC molecules by tumors and tumor cell lines. 18 There is considerable genetic variation in MICA genes with several polymorphisms reported in the extracellular and transmembrane domains. 19 Thus, like classical HLA alleles, MICA polymorphisms may have significant functional impact on antigen binding and recognition both in normal immune responses and in the development of autoimmune and inflammatory disease. This theory is supported by recent reports linking genetic variation in the MICA transmembrane domain with susceptibility to MHC-associated diseases. 20 The aim of the present study was to examine the relationship between MICA polymorphisms and susceptibility to PSC and, if possible, confirm or refute the hypothesis that MHC-encoded genetic susceptibility to PSC lies closer to the B8-TNF region than to DRB1. Materials and Methods Patients and Controls Two independently collected series of PSC patients were studied, the first from King s College Hospital (first set), and the second from John Radcliffe Hospital, Oxford (second set). All of the patients and controls in this study were British and of northern European Caucasoid (NEC) descent. In each series, the diagnosis of PSC was based on accepted endoscopic retrograde cholangiopancreatography and histologic findings. The first study population consisted of 62 NEC adult patients. Forty-four patients (71%) were male, the median age at presentation was 43 years (range, 15 to 78 years), and concurrent IBD was present in 77% (47 had UC and 1 had Crohn s disease). Six patients (9.7%) developed hepatobiliary malignancy, and a further 4 (6.4%) developed gastrointestinal cancers. The second population consisted of 50 adult NEC PSC patients from Oxford. The mean age at presentation was 42 years (range, 16 to 77 years), 34 were male (68%), and 41 (82%) had concurrent IBD (37 UC, 4 Crohn s disease). Control DNA samples consisted of 118 geographically and racially matched NEC healthcare workers. MICA Polymorphisms Genomic DNA was extracted from whole blood using a standard salting out procedure. Polymerase chain reaction (PCR) amplification was performed in 10 L reaction mixes as previously published. 21 Briefly, 25 ng DNA was amplified in a1 commercial buffer (Advanced Biotechnologies, Epsom, Surrey, UK) containing 0.581 mmol/l MgCl 2 (Advanced Biotechnologies); 0.2 mmol/l deoxynucleoside triphosphates buffer (Advanced Biotechnologies); 0.1 mmol/l stock of sense and antisense human growth hormone (internal control primers that generate a 480 base pair product); 0.5 mol/l stock MICA primers (Tables 1 and 2, as previously described 21 ); and 0.4 units Taq polymerase (Advanced Biotechnologies). The cycling conditions for amplification were 96 C for 5 minutes, followed by 30 cycles of 94 C for 20 seconds, 65 C for 50 seconds, and 72 C for 30 seconds, before cooling to 4 C. PCR products were identified by gel electrophoresis in 1.0% (wt/ vol) agarose gels containing 1 mg/l ethidium bromide at 200 V for 30 minutes with appropriate commercially available molecular weight markers. HLA and Tumor Necrosis Factor Genotyping All patients and controls were previously typed for HLA DRB1, DRB3, DRB4, DQA1, and DQB1 alleles using standard PCR protocols either as previously published 8 or using a modification of the method of Bunce et al., 22 and HLA A and B typing was performed by microcytotoxicity and/or PCR genotyping. In addition, the TNFA-308 and TNFA-238 polymorphisms were determined as previously described. 9 All of the HLA and TNF data presented here are for use in extended haplotype and stratification analysis only. Statistical Analysis Allele and genotype frequencies were compared using 2 and Fisher exact probability tests as appropriate. All analyses were performed using the Epistat statistical software (CDC, Atlanta, GA). In the absence of preliminary data, no assumptions could be made about the relationship between MICA alleles and PSC; therefore, comparisons of both allele frequencies (i.e., counting chromosomes; Table 1) and phenotype frequencies (i.e., counting individuals; Table 2) were performed to detect both dominant (heterozygote) susceptibility and recessive (homozygote) susceptibility effects. Throughout the analysis, the King s series was evaluated as the first set and the Oxford set as the second series. The probability values obtained were corrected for multiple testing (Bonferroni s correction) using a correction factor of 16 for the first set (King s) only (i.e., the total number of alleles MICA alleles tested). No correction was used for the second (confirmatory) set (Oxford). Stratification analysis was performed using an abridged version of the method described by Svejgaard and Ryder. 23

May 2001 MICA ALLELES AND PSC 1477 Table 1. MICA Primer Sequences Sense primers Code Exon/posn MICA allele homology TAACCTCACggTgCTgTCCT 1 2 40 001, 003 010, 012, 016 CCTCACggTgCTgTCCg 2 2 40 002, 011, 013, 014, 015 GgAACTACggCgATATCTAg 3 3 517 003, 004, 006, 008 010, 013, 014, 016 TCAgCCCTTCCTgCgCTA 4 a 2 107 004, 005, 006, 008 010, 016 AggAACTACggCgATATCTAA 5 a 3 517 001, 002, 005, 007, 011, 012, 015 TgTgCAgTCAgggTTTCTCg 7 a 269 002, 007, 011, 013, 014, 015 ACggCgATATCTAgAATCCg 8 a 3 523 001, 002, 005, 007, 008, 011, 012 015 CAGAgCCCCACAgTCTTCC 10 2 17 010 TTTCTTgAAggAAgATgCCg 11 3 451 011 CCCAgCATTTCTACTACgATA 14 3 340 014, 015 Antisense primers TggggCATTgTCCATTCCTT 1 3 373 001 CTCAggACTACgCCggATTT 2 a 3 517 001, 002, 005, 007, 011, 012, 015 GggCACAgggTgAgTgCC 3 a 4 802 003 GgCCAgCgTCCgTACCTC 4 3 542 003, 004, 014 AACCTCTgCTCCTCTCCTC 5 4 752 005, 008, 010, 013, 016 CCTgTTCTCCTCAggACTAT 6 3 526 006 CTCTggAggACTggggCAT 7 3 385 001, 002, 007, 011, 012, 014, 015 CCTgACgCCAggTCAgTA 8 4 638 008, 010, 016 TCCATTCCTCAgTCTCCAC 9 3 364 003, 004, 006, 009 TggATTCTAgATATCgCCgT 10 a 3 504 001 016 CTgCATgCATAgCgTgATAgA 12 3 467 012 TCTggAggACTggggCAC 13 a 3 385 003 006, 008, 009, 010, 013, 016 TCAggACTACgCCggATTC 14 3 517 003, 004, 006, 008 010, 013, 014, 016 GTgTCgTggCTCAAAgATAg 16 4 661 016 GgCCAgCgTCCgTACCTg 17 3 542 001, 002, 005 014, 016 CCTgTTCTCCTCAggACTAC 18 3 526 001 005, 007 016 a Primers 4, 5, 7, 8, 2, 3, 10, and 13 also share homology with alleles encoded by the MICB locus. 4,10,12 All primers were selected to have strand melting temperatures within the range of 58 C to 62 C, and allele and group-specific nucleotides on the terminal 3 base. The given positions of primers are based on the scheme of Fodil et al. 19 Table 2. Forward and Reverse Primer Combinations for MICA 001 016 Alleles Mix Primer codes MICA alleles recognized Amplicon size (bp) 1 1 1 001 646 2 2 2 002, 011, 015 777 3 3 3 003 909 4 4 4 004 744 5 5 5 005 861 6 4 6 006 730 7 1 2 001, 005, 007, 012 780 8 8 8 008 739 9 4 9 004, 006, 009 568 10 10 10 010 799 11 11 2 011 106 12 1 12 012 741 13 2 13 013 653 14 14 14 014 217 15 14 2 015 218 16 3 16 016 770 17 7 7 002, 007, 011, 014, 015 628 18 1 10 001, 003, 004, 005, 006, 007, 008, 009, 010, 012, 016 777 19 1 17 001, 005, 006, 007, 008, 009, 010, 012, 016 813 20 1 18 001, 003, 004, 007, 008, 009, 010, 012, 016 799 NOTE. Alleles in bold type indicate target MICA allele for each primer combination.

1478 NORRIS ET AL. GASTROENTEROLOGY Vol. 120, No. 6 Table 3. MICA Allele Frequencies Allele frequency (%) Probability Allele Controls n 236 PSC King s n 124 PSC Oxford n 100 pc (first set) P (second set) *001 0 1 0 NS *002 18 3.2 0 1. pc 0.0013, OR 0.15 2. P 0.0000065, OR 0.0 *004 10 5.6 0 NS *006 2.5 0 0 NS *007 5.1 5 12 NS *008 48 66 66 1. pc 0.015, OR 2.13 2. P 0.0023, OR 2.11 *009 5.5 1.6 0 NS *010 5.1 4 9 NS *012 3.4 1.6 3 NS *013 0 0.8 0 NS *016 0.8 1.6 0 NS Unassigned 2.5 11.3 10 Results Comparison of Allele and Genotype Distribution Overall, comparing patients and controls, there was an increased frequency of TNFA*2 (63% of patients vs. 32% of controls) and the HLA alleles B8 (40% vs. 21%), DRB1*0301 (36% vs. 21%), DRB1*1301 (21% vs. 15%), DRB1*1501 (27% vs. 21%), and lower frequency of DRB1*04 (13% vs. 42%) in patients compared with controls. These data are shown for comparison only and were used for analysis of extended haplotypes and stratification. Comparing the first patient set (King s PSC) and controls, there was a lower frequency of the MICA*002 allele (3.2% vs. 18%, pc 0.0013, odds ratio [OR] 0.15) and an increased frequency of the MICA*008 allele (66% vs. 48%, pc 0.015, OR 2.13) in patients (Table 3). These differences were confirmed in the second set for MICA*002 (0% vs. 18%, P 0.0000065, OR 0.0) and MICA*008 (66% vs. 48%, P 0.0023, OR 2.11). Further analysis indicated that the difference in MICA*002 allele frequency was a result of mostly a lower incidence of heterozygotes for this allele (6.4% vs. 35% of controls; pc 0.00018, OR 0.12), whereas the variation in MICA*008 allele frequency was caused entirely by an increased frequency of homozygous MICA*008 patients (58% vs. 22%, pc 0.000015, OR 5.01; Table 4). These observations were confirmed in the second set for MICA*002 (0% vs. 35%, P 0.0000016, OR 0.0) and for MICA*008 (58% vs. Table 4. MICA Alleles Distribution (Phenotypes) Number of individuals positive (%) MICA allele Controls n 118 PSC King s n 62 PSC Oxford n 50 pc (first set) P (second set) *001 0 1 (1.6) 0 NS *002 41 (35) 4 (6.4) 0 1. pc 0.00018, OR 0.12 2. P 0.0000016, OR 0.0 *004 20 (17) 7 (11) 0 NS *006 6 (5) 0 0 NS *007 12 (10) 4 (6.5) 9 (18) NS *008 87 (74) 46 (74) 37 (74) NS *009 11 (9) 2 (3.2) 0 NS *010 11 (9) 4 (6.4) 6 (12) NS *012 8 (7) 2 (3.2) 3 (6) NS *013 0 1 (1.6) 0 NS *016 2 (2) 2 (3.2) 0 NS Unassigned 6 7 (11.3) 5 (10) NS *008/*008 26 (22) 36 (58) 29 (58) 1. pc 0.000015, OR 5.01 2. P 0.0000056, OR 4.51

May 2001 MICA ALLELES AND PSC 1479 Table 5. MICA* and HLA DRB1*, DQB1* Genotypes of Patients With MICA*002 Patient ID MICA* genotype DRB1* DQB1* 36 002/008 1301, 14 0503, 0603 64 002/008 0101, 0301 0501, 0201 99 002/008 1301, 1501 0603, 0602 118 002/008 0301, 1301 0201, 0603 22%, P 0.0000056, OR 4.51). Further analysis indicated that the presence of MICA*002 was not associated with a particular MHC haplotype including DRB1*0401-DQB1*0302 (Table 5). Analysis of MICA*008 Extended Haplotypes Overall, MICA*008 was found on 48% of HLA haplotypes in the normal controls. The allele seems to be most strongly linked with 2 haplotypes, (1) the ancestral 8.1 haplotype (A1-B8-Cw*0701-TNFA*2- DRB3*0101-DRB1*0301-DQA1*0501-DQB1*0201), and (2) the B7-Cw*0701-TNFA*1- DRB5*0101- DRB1*1501-DQA1*0102-DQB1*0602 haplotype. One explanation for the association between PSC and MICA*008 may be linkage with the 2 haplotypes above. For this reason, we have performed stratification analysis (Table 6) for both B7, B8, and the combination of B7 or B8 with MICA*008. The stratification analysis clearly indicates that the MICA*008 association is independent of both B8 and B7 (B8-ve/MICA*008 ve, pc 0.021, B7-ve/MICA*008 ve, pc 0.00024, and for B7-ve or B8-ve/MICA*008 ve, pc 0.016). In contrast, the association with B8 was not independent of MICA*008 (pc NS). Although there was a slight increase in B8 homozygosity in patients (11% vs. 3%), this was not sufficient to account for the 36% increase in MICA*008 homozygosity in both patient sets vs. controls. Furthermore, the total number of B8 homozygotes, B7/B8 heterozygotes, and B7 homozygotes was also insufficient to account for this increase (17% patients vs. 8.5% controls, P 0.065). Further stratification analysis for MICA*008 and the DRB1*0301 and DRB1*1501 alleles was not thought necessary because the MICA maps closer to the B locus on chromosome 6p21.3, and the linkage with B7 and B8 is stronger than for DRB1*0301 or DRB1*1501. Thus, any additional analysis would not be informative. Correlation With Clinical Phenotypes There was no correlation between MICA*002 or MICA*008 and sex, presence or absence of concurrent IBD, age at presentation, disease duration (survival), and development of hepatobiliary malignancy. In addition, there was no increase in MICA*007 in PSC patients with IBD in contrast with a recent report. 24 Discussion The present study includes 2 original observations. First, we have identified a highly significant association between PSC and MICA*008 allele in 2 independently collected sets of NEC PSC patients. This association is caused entirely by an increased frequency of homozygous MICA*008 patients in both sets (58% compared with 22%), and is independent of linkage with B8 and B7. Because both the B7 and B8 haplotypes usually also carry the DRB alleles DRB3*0101-DRB1*0301 and DRB5*0101-DRB1*1501, the current observation that possessing 2 copies of MICA*008 is associated with a fivefold increased risk of disease may also account for our Table 6. Stratification Analysis for HLA B7 and B8 Versus MICA*008 present absent Alleles Number Analysis present absent Patients N 112 Controls N 236 Probability (pc P 9) Odds ratio B8 MICA*008 27 20 pc 0.000033 4.74 5 4 pc NS 55 89 pc 0.021 2.17 35 123 B7 MICA*008 15 28 pc NS 0 2 pc NS 67 83 pc 0.00024 2.84 35 123 B7 or B8 MICA*008 42 48 pc 0.00054 3.1 5 6 pc NS 40 59 pc 0.016 2.38 35 123

1480 NORRIS ET AL. GASTROENTEROLOGY Vol. 120, No. 6 earlier suggestion that homozygosity for the amino acid leucine at position 38 of the DR -polypeptide is the major HLA-encoded determinant of disease susceptibility in PSC. 8 Secondly, we have identified a significantly lower frequency of the MICA*002 allele in patients compared with controls. None of the patients in the second set expressed this allele. This suggests that the MICA*002 allele is strongly associated with protection from disease and because one copy of the allele is sufficient to prevent PSC, the resistance allele may be dominant. The very low frequency of MICA*002 in PSC prevents further exploration of the haplotype; however, none of the 4MICA*002-positive patients were DRB1*0401- DQB1*0302 positive, and the MICA*002 allele was found to be normally distributed in controls with and without DR4. Thus, of 50 DR4-positive controls, 15 (30%) had MICA*002 compared with 26 of 68 (38%) DR4-negative controls, indicating that DR4 and MICA*002 are not part of the same conserved haplotype. Though none of the second patient set had MICA*002, the finding of 4 patients with this allele in the first set indicates that protection from PSC associated with MICA*002 is not complete. A recent report found that the MICA*007 allele is associated with an increased risk of IBD. 24 In the present study, there was no correlation between MICA*007 and PSC. MICA*007 was found in 4 of 62 and 9 of 50 patients, respectively, compared with 12 of 118 controls. In addition, although the number of patients in the present study with hepatobiliary or gastrointestinal cancers was small, no one MICA allele was exclusively associated with the presence of either hepatobiliary or other tumors. The MICA locus has received considerable attention as a candidate locus for MHC-encoded disease susceptibility in immune-mediated diseases especially, though not exclusively, diseases with strong documented HLA class I associations. Thus, recent studies in insulin-dependent diabetes mellitus have identified MIC-A5 as an independent marker of disease risk, 25,26 which seems to be particularly important in determining age of onset. 27 A significant association between MICA*5 and resistance to multibacillary leprosy in southern China has also been reported. 28 Studies in Behçets disease have produced mixed results, 29 34 but a clear message has emerged from studies of acute anterior uveitis (AAU) and ankylosing spondylitis, both normally associated with HLA B27. These investigations indicate that MICA or a closely linked gene may be involved in the development of disease in AAU, 35 but not in ankylosing spondylitis in which the primary susceptibility allele appears to be B27. 36 39 The same MICA allele, MICA*4, which is associated with AAU, has been linked with a lower risk of coronory aneurysms in children with Kawasaki disease after aspirin and gamma-globulin treatment. 40 In a series of investigations of HLA, encoded susceptibility to Takayasu s arteritis has been found to map to 2 different HLA class I haplotypes on one of which susceptibility may be closer to MICA than HLA B. 41,42 In contrast, susceptibility to Beurger s disease seems to map more closely to HLA B than to MICA. 41 Notably, none of the above studies found associations with MICA*002 or MI- CA*008, the 2 alleles with which we are currently concerned. MIC molecules, recently identified as ligands for T cells and NK cells, are related to MHC class I molecules but seem to adopt a different configuration with no evidence of peptide binding. However, unlike HLA class I, stable expression of MICA does not require 2-microglobulin or other antigen processing molecules, and the molecule does not have a CD8 binding site. In normal tissue, MICA is almost exclusively expressed in gastrointestinal and thymic epithelia, and is not expressed by normal liver tissue or freshly isolated mononuclear cells. 16 Heat shock induction of nonproliferating cell lines results in large increases in MICA messenger RNA and protein expression, 16 suggesting that expression of MICA is regulated by cell stress. MICA expression has been reported on carcinoma of lung, breast, and colon, and current hypotheses suggest that MICA may act as a tumor marker for recognition by infiltrating T cells. 18 Although the current study did not seek to examine the functional impact of MICA in PSC, but simply attempted to map MHC-encoded disease susceptibility/ resistance, it is tempting to speculate that 2 copies of the MICA*008 allele could result in increased surface expression of the MIC protein. T cells and NK cells, commonly found in the normal human liver, 43,44 have been reported to be increased in number in PSC livers. 45 Heat shock induced expression of MICA in biliary epithelium in individuals susceptible for the development of PSC could provide the necessary stress signal to intrahepatic T cells and/or NK cells with subsequent cytotoxic T lymphocyte activation and cytokine secretion. The catalyst for heat shock induction in PSC may be an infectious agent, although there is only speculative evidence to support an infectious etiology for PSC. Alternatively, the absence of MICA may provide a permissive environment for the development of autoimmune disease through failure to down-regulate inappropriate immune responses to self.

May 2001 MICA ALLELES AND PSC 1481 Although we have shown a significant association between PSC and MICA*008 homozygosity, there are a number of caveats. First, this allele does not account for 100% of all PSC patients, and MICA*002 was found in 4 patients. Secondly, our results do not explain the increased frequency of DRB3*0101-DRB1*1301- DQB1*0603 haplotype that does not carry the MICA*008 allele. Thirdly, both the MICA*002 and MICA*008 associations are independent of the HLA associations above. Thus, the model of MHC-encoded genetic susceptibility and resistance to PSC is likely to be more complex than previously reported, and HLA haplotypes may carry multiple susceptibility and resistance alleles that may act in concert or independently. Finally, the present study was confined to a single ethnic population, British NEC, and future studies will need to investigate other populations to determine whether MICA is truly responsible for MHC-encoded disease susceptibility and resistance. In addition, the correlation between specific MICA alleles and MIC function needs to be investigated. References 1. Aadland E, Schrupmf E, Fausa O, Elgjo K, Heilo A, Aakhus T, Gjone E. Primary sclerosing cholangitis: a longterm followup study. Scand J Gastroenterol 1987;22:655 664. 2. Chapman RWG, Cottone M, Selby WS, Shepherd HA, Jewell DP. Serum autoantibodies, ulcerative colitis and primary sclerosing cholangitis. Gut 1986;27:86 91. 3. Schrumpf E, Fausa O, Forre O, Dobloug JH, Ritland S, Thorsby E. HLA antigens and immunoregulatory T cells in ulcerative colitis asociated with hepatobiliary disease. Scand J Gastroenterol 1982;17:187 191. 4. Brinch T, Teisberg P, Schrumpf E, Akessom I. The in vivo metabolism of C3 in hepatobiliary disease associated with ulcerative colitis. Scand J Gastroenterol 1982;17:523 527. 5. Mehal WZ, Dennis Lo YM, Wordsworth BP, Neuberger JM, Hubscher SC, Fleming KA, Chapman RW. HLA DR4 is a marker for rapid disease progression in primary sclerosing cholangitis. Gastroenterology 1994;106:160 167. 6. Olerup O, Olsson R, Hultcrantz R, Broome U. HLA DR and HLA DQ are not markers for rapid disease progression in primary sclerosing cholangitis. Gastroenterology 1995;108:870 878. 7. Donaldson PT, Farrant MJ, Wilkinson ML, Hayllar K, Portman BC, Williams R. Dual association of HLA DR2 and DR3 with primary sclerosing cholangitis. Hepatology 1991;13:129 133. 8. Farrant MJ, Doherty DG, Donaldson PT, Vaughan RW, Hayllar KM, Welsh KI, Eddleston AL, Williams R. Amino acid substition at position 38 of the DRb polypeptide confer susceptibility to and protection from primary sclerosing cholangitis. Hepatology 1992; 16:390 395. 9. Bernal W, Moloney M, Underhill J, Donaldson PT. Association of tumour necrosis factor polymorphism with primary sclerosing cholangitis. J Hepatol 1999;30:237 241. 10. Spurkland A, Saarinen S, Boberg KM, Mitchell S, Broome U, Caballeria L, Ciusani E, Chapman R, Ercilla G, Fausa O, Knutsen I, Pares A, Rosina F, Olerup O, Thorsby E, Schrumpf E. HLA class II haplotypes in primary sclerosing cholangitis patients from five European populations. Tissue Antigens 1999;53:459 469. 11. Donaldson PT, Manns MP. Immunogenetics of liver disease. In: Bircher J, Benhamou JP, McIntyre N, Rizzetto M, Rodes J, eds. Oxford textbook of clinical hepatology. Oxford, England: Oxford University Press, 1999:173 188. 12. Ando H, Mizuki N, Ota M, Yamazaki M, Ohno S, Goto K, Miyata Y, Wakisaka K, Bahram S, Inoko H. Allelic variants of the human MHC class I chain-related B gene (MICB). Immunogenetics 1997; 46:499 508. 13. Bahram S, Bresnahan M, Geraghty DE, Spies T. A second lineage of mammalian major histocompatibility complex class I genes. Proc Natl Acad Sci U S A 1994;91:6259 6263. 14. Bahram S, Spies T. Nucleotide sequence of a human MHC class I MICB cdna. Immunogenetics 1996;43:230 233. 15. Bauer S, Groh V, Wu J, Steinle A, Phillips JH, Lanier LL, Spies T. Activation of NK cells and T cells by NKG2D, a receptor for stress-inducible MICA. Science 1999;285:727 729. 16. Groh V, Bahram S, Bauer S, Herman A, Beauchamp M, Spies T. Cell stress-regulated human major histocompatibility complex class I gene-regulated in gastrointestinal epithelium. Proc Natl Acad Sci U S A 1996;93:12445 12450. 17. Groh V, Steinle A, Bauer S, Spies T. Recognition of stressinduced MHC molecules by intestinal epithelial T cells. Science 1998;279:1737 1740. 18. Groh V, Rhineheart R, Secrist H, Bauer S, Grabstein KH, Spies T. Broad tumor-associated expression and recognition by tumorderived T cells of MICA and MICB. Proc Natl Acad Sci USA 1999;96:6879 6884. 19. Fodil N, Laloux L, Wanner V, Pellet P, Hauptmann G, Mizuki N, Inoko H, Spies T, Theodorou I, Bahram S. Allelic repertoire of the human MHC class I MICA gene. Immunogenetics 1996;44:351 357. 20. Mizuki N, Ota M, Kimura M, Ohno S, Ando H, Katsuyama Y, Yamazaki M, Watanabe K, Goto K, Nakamura S, Bahram S, Inoko H. Triplet repeat polymorphism in the transmembrane region of the MICA gene: a strong association of six GCT repetitions with Behcet disease. Proc Natl Acad Sci U S A 1997;94:1298 1303. 21. Stephens HAF, Vaughan RW, Collins R, Kondeatis E, Theron J, Payne A. Towards a molecular phototyping system for alelic variants of MICA, encoded by polymorphisms in exons 2, 3 and 4 of MHC class I chain-related genes. Tissue Antigens 1999;53: 167 174. 22. Bunce M, O Neill M, Barnardo MCNM, Browning MJ, Morris PJ, Welsh KI. Phototyping: comprehensive DNA typing for HLA-A, B, C, DRB1, DRB3, DRB4, DRB5 & DQB1 by PCR with 144 primer mixes utilising sequence-specific primers (PCR-SSP). Tissue Antigens 1995;46:355 367. 23. Svejgaard A, Ryder LP. HLA and disease associations: detecting the strongest association. Tissue Antigens 1994;43:18 27. 24. Orchard TR, Vaughan R, Simmonds JD, Welsh KI, Jewell DP. Susceptibility to ulcerative colitis (UC) but not Crohn s disease is associated with the MHC class I like gene-a (MIC-A): a two stage study (abstr). Gut 1999;44(suppl 1):W201. 25. Gambelunghe G, Ghaderi M, Cosentino A, Falorni A, Brunetti P, Falorni A, Sanjeevi CB. Association of MHC class I chain-related A (MIC-A) gene polymorphism with type 1 diabetes. Diabetologia 2000;43:507 514. 26. Lee YJ, Huang FY, Wang Ch, Lo FS, Tsan KW, Hsu CH, Huang CY, Chang SC, Chang JG. Polymorphism in the transmembrane region of the MICA gene and type 1 diabetes. J Pediatr Endocrinol Metab 2000;13:489 496. 27. Kawabata Y, Ikegami H, Kawaguchi Y, Fujisawa T, Hotta M, Ueda H, Shintani M, Nojima K, Ono M, Nishino M, Taniguchi H, Noso S, Yamada K, Babaya N, Ogihara T. Age related association of MHC class I chain related gene A (MICA) with type 1 (insulin dependent) diabetes mellitus. Hum Immunol 2000;61:624 629. 28. Wang LM, Kimura A, Satoh M, Mineshita S. HLA linked with leprosy in southern China: HLA-linked resistance alleles to leprosy. Int J Lepr Other Mycobact Dis 1999;67:403 408.

1482 NORRIS ET AL. GASTROENTEROLOGY Vol. 120, No. 6 29. Wallace GR, Verity DH, Delamaine LJ, Ohno S, Inoko H, Ota M, Mizuki N, Yabuki K, Kondeatis E, Stephens HA, Madanat W, Kanawati CA, Stanford MR, Vaughan RW. MIC-A allele profiles and HLA class I associations in Behcet s disease. Immunogenetics 1999;49:613 617. 30. Gonzalez-Escribano MF, Rodriguez R, Valenzuela A, Garcia A, Nunez-Roldan A. Complex associations between HLA-DRB1 genes and female rheumatoid arthritis: results from a prospective study. Hum Immunol 1999;60:1259 1265. 31. Kimura T, Goto K, Yabuki K, Mizuki N, Tamiya G, Sato M, Kimura M, Inoko H, Ohno S. Microsatellite polymorphism within the MICB gene among Japanese patients with Behcet s disease. Hum Immunol 1998;59:500 502. 32. Mizuki N, Inoko H, Ohno S. Molecular genetics (HLA) in Behcet s disease. Yonsei Med J 1997;38:333 349. 33. Mizuki N, Ota M, Katsuyama Y, Yabuki K, Ando H, Goto K, Nakamura S, Bahram S, Ohno S, Inoko H. Association analysis between MIC-A and HLA-B alleles in Japanese patients with Behcet s disease. Arthritis Rheum 1999;42:1961 1966. 34. Yabuki K, Mizuki N, Ota M, Katsuyama Y, Palimeris G, Stavropoulos C, Koumantaki Y, Spyropoulou M, Giziaki E, Kaklamani V, Kaklamani E, Inoko H, Ohno S. Association of MICA gene and HLA-B*5101 with Behcet s disease. Invest Ophthalmol Vis Sci 1999;40:1821 1926. 35. Goto K, Ota M, Maksymowych WP, Mizuki N, Yabuki K, Katsuyama Y, Kimura M, Inoko H, Ohno S. Association between MICA gene A4 allele and acute anterior uveitis in white patients with and without HLA-B27. Am J Ophthalmol 1998;126:436 441. 36. Goto K, Ota M, Ohno S, Mizuki N, Ando H, Katsuyama Y, Maksymowych WP, Kimura M, Bahram S, Inoko H. MICA gene and ankylosing spondylitis: linkage analysis via a transmembraneencoded triplet repeat polymorphism. Tissue Antigens 1997;49: 503 507. 37. Martinez-Borra, Gonzalez S, Lopez-Vasquez A, Gelaz MA, Armas JB, Kanga U, Mehra NK, Lopez-Larrea C. HLA-B27 alone rather than B27-related class I haplotypes contributes to ankylosing spondylitis susceptibility. Hum Immunol 2000;60:131 139. 38. Tsuchiya N, Shiota M, Moriyama S, Ogawa A, Komatsu-Wakui M, Mitsui H, Geraghty DE, Tokunaga K. MICA allele typing of B27 positive Japanese patients with seronegative spondylarthropathies and healthy individuals: differential linkage disequilibrium with HLA-B27 subtypes. Arthritis Rheum 1998;41:68 73. 39. Yabuki K, Ota M, Goto K, Kimura T, Nomura E, Ohno S, Mizuki N, Katsuyama Y, Maksymowych WP, Bahram S, Kimura M, Inoko H. Triplet repeat polymorphism in the MICA gene in HLA-B27 positive and negative caucasian patients with ankylosing spondylitis. Hum Immunol 1999;60:83 86. 40. Huang Y, Lee YJ, Chen MR, Hsu CH, Lin SP, Sung TC, Chang SC, Chang JG. Polymorphism of transmembrane region of MICA gene in Kawasaki disease. Exp Clin Immunogenet 2000;17:130 137. 41. Kimura A, Kobayashi Y, Takahashi M, Ohbuchi N, Kitamura H, Nakamura T, Satoh M, Sasaoka T, Hiroi S, Arimura T, Akai J, Aberajinai W, Yasukochi Y, Numano F. MICA gene polymorphism in Takayasu s artertitis and Buergers s disease. Int J Cardiol 1998;66(suppl):S107 S113. 42. Kimura A, Ota M, Katsuyama Y, Ohbuchi N, Takahashi M, Kobayashi Y, Inoko H, Numano F. Mapping of the HLA-linked genes controlling the susceptibility to Takayasu s arteritis. Int J Cardiol 2000;75(suppl):S105 S110. 43. Norris S, Collins C, Doherty DG, Smith F, McEntee G, Traynor O, Hegarty JE, O Farrelly C. Resident human hepatic lymphocytes are phenotypically different from circulating lymphocytes. J Hepatol 1998;28:84 90. 44. Norris S, Doherty DG, McEntee G, Traynor O, Hegarty JE, O Farrelly C. Natural T cells in the human liver: cytotoxic lymphocytes with dual T cell and natural killer cell phenotype and function are phenotypically heterogenous and include V 24-J Q and T cell receptor bearing cells. Hum Immunol 1999;60:20 31. 45. Martins EBG, Graham AK, Chapman RW, Fleming KA. Elevation of T lymphocytes in peripheral blood and livers of patients with primary sclerosing cholangitis and other autoimmune liver diseases. Hepatology 1996;23:988 993. Received October 2, 2000. Accepted January 10, 2001. Address requests for reprints to: Peter T. Donaldson, Ph.D., Centre for Liver Research, School of Clinical Medical Sciences, 4th Floor William Leech Building, University of Newcastle, Newcastle NE2 4HH. e-mail: p.t.donaldson@newcastle.ac.uk; fax: 0044-191-2220723.