HLA-DQ6-mediated protection in IDDM

HLA-DQ6-Mediated Protection in IDDM Carani B. Sanjeevi ABSTRACT: Insulin-dependent diabetes mellitus is positively associated with DQ8, DQ2, and DQ6 (DQB1*0604), and negatively associated with DQ6 (DQB1*0602), DQ6 (DQB1*0603), and DQ7 in Swedish caucasians. The protection conferred by DQ6 (DQB1*0602) is stronger in younger individuals and there is decrease in the effect of protection with increasing age. Three-dimensional modeling of the susceptible DQ6 (DQB1*0604) and protective DQ6 (DQB1*0602), which share the same DQA chain (DQA1*0102) but differ in the DQB chain at 6 residues, identifies residue 57 and 70 to be important for protection. Three-dimensional models of the DQ8 molecules were constructed from the coordinates of the DR1 crystal structure and other susceptibility and resistance molecules were made by homology model-

ABBREVIATIONS IDDM insulin-dependent diabetes mellitus GLY glycine ESP electrostatic potential VAL valine PCR polymerase chain reaction

INTRODUCTION The killing of the pancreatic islet ␤ cells that results in IDDM is most often thought to be due to an attack by lymphocytes and antibodies from the patient’s own immune system. This is the result of interaction of genetic susceptibility and environmental insult. IDDM is a polygenic disease and strongest association with IDDM are genes in the HLA region that are important for susceptibility and resistance to the disease [1]. IDDM is positively associated with HLA-DQ2, DQ8, and DQ6 (B1*0604) in Swedish caucasians [2– 4], while the fre-

From the Karolinska Institute, Molecular Immunogenetics Group, Department of Molecular Medicine (C.B.S.), Karolinska Hospital, Stockholm, Sweden. Address reprint requests to: Carani B. Sanjeevi, Karolinska Institute, Molecular Immunogenetics Group, Department of Molecular Medicine, Karolinska Hospital, CMM, L8:03, S-17176 Stockholm, Sweden; Tel: ⫹468-517 76254; Fax: ⫹46-8-517 76179; E-Mail: sanjeevi.carani@ molmed.ki.se. Received March 24, 1999; accepted August 31, 1999. Human Immunology 61, 148 –153 (2000) © American Society for Histocompatibility and Immunogenetics, 2000 Published by Elsevier Science Inc.

ing. The positively associated DQ molecules had weakly negative to significantly positive surface electrostatic potentials over the peptide binding and T cell recognition areas, whereas the negatively associated molecules had distinctly more negative areas over the relevant surface. This suggests that the variation in the physicochemical properties such as molecular electrostatic potentials among different DQ molecules are important. Human Immunology 61, 148 –153 (2000). © American Society for Histocompatibility and Immunogenetics, 2000. Published by Elsevier Science Inc. KEYWORDS: IDDM; HLA-DQ; genetic susceptibility; molecular modeling

ARG GAD65 ASP IA2

arginine glutamic acid decarboxylase 65 isoform aspartic acid tyrosine pyrophosphatase antigen

quency of DQ6 (B1*0602, B1*0603, and B1*0601) [5, 6] and DQ7 are decreased [2]. In the current update on the WHO nomenclature for HLA DR and DQ, several subtypes of DQ6 are identified [7]. Of the several subtypes, the subtype DQA1*0102-DQB1*0602 is the most protective. However, this subtype is low in frequency in the Asian population where DQ6 (DQA1*0103-DQB1*0601) is most frequent [8 –10]. In this population, this DQ is negatively associated with IDDM [11]. DQ6 (DQA1*0103-DQB1*0603) is also associated with protection in IDDM but the protective effect is not as strong as DQ6 (DQA1*0102DQB1*0602) [12]. The rest of the DQ6 subtypes are either positively associated (DQA1*0102-DQB1*0604) [4] or neutral in association (B*0605; B*0606; B*0607; B*0608, etc.) (Table 1). In spite of the fact that several DQ molecules are negatively associated with IDDM, a single copy of DQ6 (B*0602) is sufficient to give significant protection from IDDM [13]. 0198-8859/00/$–see front matter PII S0198-8859(99)00146-9

HLA-DQ6 and IDDM

149

TABLE 1 DQB1 alleles of serologic type DQ6 and their association with IDDM

MOLECULAR ANALYSIS OF DQ ASSOCIATION WITH IDDM

DQ6-DQB1 alleles

Since the elucidation of the structure of the class II molecule [17], it has become possible to construct three dimensional models of the HLA-DQ and DR molecules, leading to increased understanding of the role of HLA in IDDM susceptibility. Three-dimensional models of the DQ8 molecules was constructed [16] by homology modeling techniques from the coordinates of the crystal structure of the DR1 molecule with influenza peptide [19] as there is 65% to 70% homology between the sequences of DRA and DQA, DRB and DQB. Surface electrostatic potentials of the solvent accessible surfaces in the peptide binding and T cell recognition areas in DQ8 and DQ2 showed mean ESP to be neutral to mildly negative and the DQ6 to be significantly negative. The positively associated DQ8 (DQA1*0301-DQB1*0302) and DQ2 (DQA1*0501-DQB1*0201) molecules and the negatively associated DQ6 (DQA1*0102DQB1*0602) molecule were constructed and surface electrostatic potential over the peptide binding and T cell recognition areas were compared. The positively associated DQ have weakly negative to markedly positive potential over the relevant surface, while negatively associated sequences display a distinctly more negative potential. The electrostatic potential over the peptide binding and T cell recognition areas of positively associated DQ2 (DQA1*0501-DQB1*0201) and negatively associated DQ6 (DQA1*0102-DQB1*0602) are shown in Figs. 1 and 2. The data indicated that the individual polymorphic positions in DQA1 and DQB1 loci did not explain fully the susceptibility or protection to IDDM. However, variation in physicochemical properties, such as molecular electrostatic potential, among different DQ molecules, such as molecular electrostatic potentials, was considered significant [20].

DQB1*0601 DQB1*0602 DQB1*0603 DQB1*0604 DQB1*0605 DQB1*0606 DQB1*0607 DQB1*0608 DQB1*0609 DQB1*0610 DQB1*0611 DQB1*0612 DQB1*0613 DQB1*0614

Association with IDDM Negative in Asians Negative Negative Positive in Swedish caucasians Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral Neutral

DR2 OR DQ6 IN IDDM PROTECTION DR2 haplotype is strongly negatively associated with IDDM [14, 15]. In a population based study from Sweden [16], DR2 was positive in 11/425 (3%) patients and 101/367(28%) controls. Analysis of the DR2 haplotypes in these 11 DR2 positive patients showed that, of the 5 subtypes (DRB1*1501, DRB1*1502, DRB1*1503, DRB1*1601, and DRB1*1602) analyzed, DRB1*1601DQA1*0102-DQB1*0502 was more frequent in the patient group when compared to the controls. DRB1*1501-DQA1*0102-DQB1*0602 was present only in two patients. In both the patients, the second haplotype was DR4. This finding is in contrast to that seen in Finnish patients where DRB1*1501DQA1*0102-DQB1*0602 was the most frequent haplotype among the DR2 positive patients [14]. DRB5*0101 was the DRB5 gene frequently seen in the DR2 haplotype but their association with IDDM was secondary to linkage disequilibrium with DRB1*1501 [16]. In addition, three-dimensional models of the peptide binding and the T cell recognition areas of the five subtypes of the DR2 molecules were constructed from the coordinates of the crystal structure of DR1 [17] by homology modeling techniques [16]. Molecular electrostatic potentials computed over the solvent accessible surfaces, show a significantly negative mean ESP over the peptide binding and T cell recognition areas and the range of these potentials varied very little between the DR2 molecules, supporting the idea that the resistance may not be associated with HLA-DR in agreement with the finding based on the use of different statistical methods [18]. These data support the idea that DQ rather than DR molecules confer decreased risk for IDDM.

EFFECT OF AGE ON THE PROTECTION CONFERRED BY DQ6 The protection conferred by DQ6 (B*0602) in IDDM is not applicable to all age groups [21]. The age dependence in HLA associated relative risks was assessed via likelihood ratio tests of regression parameters in separate logistic regression models for each HLA category. The risk for protection conferred by DQ6 (B*0602) was almost absolute at birth (relative risk for susceptibility ⫽ 0.03 or risk for protection being 33.3), but this risk was much less at age 35 (relative risk for susceptibility ⫽ 1.17 or the risk for protection being 0.85) [4]. This decrease in protection with increasing age is also seen in another disease associated HLA-DQ, DQ7

150

FIGURE 1 Color coded surface electrostatic potentials of DQ2 (DQA1*0501-DQB1*0201) positively associated with IDDM. (Reproduced with permission from the PhD thesis of Dr. C. B. Sanjeevi submitted to Karolinska Institute [20].)

(DQA1*0501-DQB1*0301). The peak risk for this DQ was at age 18. For the positively associated genotype DQ8/DQ2, the maximum risk for the development of IDDM was at birth (relative risk ⬎ 60), but this risk for susceptibility dropped at age 35 (relative risk ⬍ 20). For the DQ6 (DQA1*0102-DQB1*0604), IDDM risk increased with increasing age (neutral at birth to positive at age 35) [4]. EFFECT OF SEX ON THE PROTECTION CONFERRED BY DQ6 After puberty, the incidence of IDDM in males exceeds that in females [22, 23]. The male preponderance in IDDM is paradoxical since other organ specific auto immune disease are more common among women. In Swedish caucasians with IDDM, the negatively associated DQ6 (B*0602) is more negatively associated in

C. B. Sanjeevi

FIGURE 2 Color coded surface electrostatic potentials of DQ6 (DQA1*0102-DQB1*0602) negatively associated with IDDM. Both molecules are in the same orientation, with DQA chain on the left and DQB chain on the right, viewed from the top and the peptide binding groove in the middle. The electrostatic potential color codes range from yellow (strongly negative potential that is ⬍ ⫺200.0 kT/e) to green (negative) to cyan (mildly negative) to blue (weakly positive) to red (strongly positive potential that is ⬎ ⫹77.0 kT/e). This qualitative picture demonstrates that the solvent-accessible surface of the peptide binding and the T cell recognition areas are positively associated for DQ2 and negatively associated for DQ6 molecule. (Reproduced with permission from the PhD thesis of Dr. C. B. Sanjeevi submitted to Karolinska Institute [20].)

IDDM females than in IDDM males. The positively associated DQ6 (B*0604) gives higher risk for IDDM in females than in males [4]. SUSCEPTIBILITY DQ6 AND PROTECTIVE DQ6 DQ6 molecules (B1*0603 and B1*0601) share the same DQ␣ chain (A1*0103) and since both are negatively

HLA-DQ6 and IDDM

associated, identifying the disease associated and disease non-associated peptide may not be valuable. However, DQ6 (B1*0602) is negatively associated with IDDM, but DQ6 (B1*0604) shows positive association [6]. Both molecules share the same DQ␣ chain but differ in the ␤ chain by 6 residues at position 9, 30, 57, 70, 8, and 87. By molecular modeling, The protective effect of DQ6 (B*0602) was identified to be due to changes at Asp (D) at ␤57 and Gly (G) at ␤70 [5]. The substitution at position 57␤ from Asp in DQB1*0602 to Val in DQB1*0604 is in a key anchor pocket, while the substitution at position 70␤ from Gly in DQB1*0602 to Arg in DQB1*0604 affects T cell recognition. When D57␤ is present we predict that a salt bridge is formed with R79␣. For the D57V substitution, the salt bridge is broken and R79␣ probably adopts an alternate conformation to increase solvent exposure, leading to significant increase in anchor pocket size and hydrophobicity. It is proposed that change in the size and hydrophobicity of this anchor pocket will alter the preferred anchor residues for this pocket and thus lead to altered peptide binding selectivity or affinities. The G70R substitution introduces a large, basic residue at the top of the helical segment of the ␤-chain in a location where it is likely to form hydrogen bonds with peptide ligand backbone and interact directly with T cell receptors during complex formation. It seems highly probable that this nonconservative substitution could alter T cell recognition and/or binding affinity substantially. When Swedish patients (n ⫽ 425) with IDDM were compared to controls (n ⫽ 367), D57/G70 was present in 24/425 (6%) patients compared to 160/367 (44%) controls (OR 0.08; Pc ⬍ 0.01). When the genotypic combination of D57/G70-D57/G70 was analyzed 0/425 patients had this combination when compared to 31/367 (8%) controls (OR 0.01; Pc ⬍ 0.01) suggesting the important role these two residues play in protection from IDDM [5]. The DQ2 (DQA1*0501-DQB1*0201) and DQ6 (DQA1*0103-DQB1*0603) is positively and negatively associated with diabetes, respectively, is also positively and negatively associated with Myasthenia gravis, respectively [24, 25]. Molecular modeling studies of these 2 molecules showed the differences in the peptide binding properties to be localized to peptide-anchor pockets P7 and P9. It is speculated that this differences might be important for the binding of disease associated peptides. IDDM IN DQ6 POSITIVE PATIENTS Baisch et al., have shown that the protection conferred by DQ6 (B*0602) is dominant. In spite of the protective effect of the DQ6 (B*0602) molecule, several IDDM patients have been reported to carry DQ6 (B*0602). The

151

DNA sequence analysis of DQ6 (B*0602) positive IDDM patients [26] suggested that these DQ6 (B*0602) molecules originally typed by PCR based typing technique, carried unique sequences that were similar to DQB1*0602 and DQB1*0603, but not exactly identical to that of DQB1*0602. The variations from the DQB1*0602 sequence in these IDDM patients were located at codons 9, 38, 59, and 62. All these variant DQ6 molecules carried the same DQA1*0102 chain. Molecular modeling of these variant DQ6 sequences present in IDDM patients showed differences in the surface contour of the peptide binding groove from normal DQA1*0102-DQB1*0602 sequence. The surface contour of the pockets were modified due to these variations [26]. Even though typing techniques indentified the patients to be positive for DQA1*0102DQB1*0602, sequence analysis suggested variations, which could be explained as reasons for these molecules to lose their protective effect. This finding assumes importance in the prediction strategies adopted in different screening programs for identifying prediabetic IDDMs, where auto antibody positive and DQ6 (DQA1*0102-DQB1*0602) positive individuals are suggested to be non-progressors to IDDM [27–29]. It would be wise to reach such a conclusion after sequence analysis of the auto antibody positive and DQ6 (DQA1*0102-DQB1*0602) positive individuals. DQ6 MOTIFS AND IDDM Naturally processed DQB1*0602 peptides have not yet been described in the literature. Peptide binding motifs for IDDM have been identified by indirect method [30, 31]. Peptides were constructed from the IDDM auto antigen, insulin B1-15. The peptide binding motif for DQB1*0602 was defined by examining the effect of single amino acid substitutions in insulin B(5-15) on the binding to DQB1*0602. Arginine substitutions were chosen to map the primary anchors which were identified at 6, 8, 9, 11, and 14 (relative to positions P1, P3, P4, P6, and P9) as important for binding. Amino acids 11 and 14 (P6 and P9) exhibited highest specificity. Ettinger and Kwok [30] showed amino acid 11 (P6) preferred large aliphatic amino acids while amino acid 14 (P9) preferred smaller aliphatic and hydroxyl amino acids. The authors selected peptide 11 amino acid long from GAD65, IA2 and proinsulin (auto antigens in IDDM), which contained the peptide binding motif. 19 out of 24 peptides (79%) were able to bind to DQ6 (DQB1*0602). This suggests that DQ6 (DQB1*0602) might be peptide presenters to T-cells that may be important for IDDM protection.

152

MECHANISM OF DQ6-MEDIATED PROTECTION The mechanisms by which HLA-DQ6 molecules control the development of diabetes and protect auto immune reactions to specific ␤ cell antigens are not understood. Several mechanisms are suggested for the protective role by DQ6. The peptide binding hypothesis of Nepom [32, 33] suggest that protective alleles bind key auto antigenic peptides with higher affinities than susceptible alleles and protective outcome ensues. Another mechanism which might involve peptide binding and protection from diabetes is the possibility that the protective HLA allele is an efficient restriction element for negative selection of auto reactive T cells [34]. There is no convincing evidence to discriminate between these mechanisms for the dominantly protective HLA alleles in IDDM.

C. B. Sanjeevi

7.

8.

9.

10.

ACKNOWLEDGMENTS

The studies in the authors laboratory was supported by Swedish Medical Research Council, Karolinska Institute, Barn diabetes fond, Swedish Diabetes Association, Swedish Physicians Association, Novo Nordisk Fond and Åke Wiberg Stiftelse. Author wishes to thank Prof. T. P. Lybrand for collaboration and fruitful discussions about molecular modeling. CBS is a fellow of Swedish Medical Research Council.

11.

REFERENCES

12.

1. Thorsby E, Ronningen KS: Role of HLA genes in predisposition to develop insulin-dependent diabetes mellitus. Ann Med 24(6):523, 1992. 2. Sanjeevi CB, Lybrand T, DeWeese C, Landin-Olsson M, Kockum I, Dahlqvist G, Sundkvist G, Stenger D, Lernmark Å: Polymorphic amino acid variations in HLA-DQ are associated with systematic physical property changes and occurrence of insulin-dependent diabetes mellitus. Diabetes 44:125, 1995. 3. Sanjeevi CB, Kockum I, Lernmark Å: The role of major histocompatibility complex in insulin-dependent diabetes. Curr Opin Endocrinol Diabetes 2:3, 1995. 4. Graham J, Kockum I, Sanjeevi CB, Landin-Olsson M, Nystro¨m L, Sundkvist G, Arnqvist H, Blohme G, Lithner ¨ stman J, Lernmark F, Littorin B, Schersten B, Wibell L, O Å, Breslow N, Dahlquist G, and the Swedish Childhood Diabetes Registry Study Group: Negative association between IDDM and HLA DQB1*0602-DQA1*0102 is attenuated with age-at-onset. Eur J Immunogenet 26:117, 1999. 5. Sanjeevi CB, DeWeese C, Landin-Olsson M, Kockum I, Dahlquist G, Lernmark Å, Lybrand T: Analysis of critical residues of HLA-DQ6 molecules in insulin-dependent diabetes mellitus. Tissue Antigens 50:61, 1997. 6. Sanjeevi CB, Landin-Olsson M, Kockum I, Dahlqvist G, Lernmark Å: Effects of the second haplotype on the asso-

13.

14.

15.

16.

17.

ciation with childhood insulin-dependent diabetes mellitus. Tissue Antigens 45:148, 1995. Bodmer J, Marsh S, Albert E, Bodmer WF, Bontrop RE, Charron D, Dupont B, Erlich HA, Fauchet R, Mach B, Mayr WR, Parham P, Sasazuki T, Schreuder GM, Strominger JL, Svejgaard A, Terasaki PI: Nomenclature for factors of the HLA system, 1996. Tissue Antigens 49:297, 1997. Sanjeevi CB, Olerup O: Restriction fragment length polymorphism analysis of HLA-DRB, DQA and DQB genes in South Indians and Swedish Caucasians. Transplant Proc 24:1687, 1992. Mehra NK, Verduijn W, Taneja V, Drabbels J, Singh SP, Giphart MJ: Analysis of HLA-DR2-associated polymorphisms by oligonucleotide hybridization in an Asian Indian population. Hum Immunol 32(4):246, 1991. Mehra NK, Bouwens AG, Naipal A, Rajalingam R, Grubic Z, Taneja V, Tilanus MG, Giphart MJ: Asian Indian HLA-DR2-, DR4-, and DR52-related DR-DQ genotypes analyzed by polymerase chain reaction based nonradioactive oligonucleotide typing. Unique haplotypes and a novel DR4 subtype. Hum Immunol 39(3):202, 1994. Serjeantson S, Ranford P, Kirk R, Kohonen-corish M, Mohan V, Ramachandran A, Snehalatha C, Viswanathan M: HLA-DR and -DQ DNA genotyping in insulin-dependent diabetes patients in South India. Dis Markers 5:101, 1987. Rønningen KS, Spurkland A, Tait BD, Drummond B, Lopez-Larrea C, Baranda FS, Thorsby E: HLA class II associations in insulin-dependent diabetes mellitus among Blacks, Caucasoids, and Japanese. Oxford, Oxford University Press, 1991. Baisch JM, Weeks T, Giles R, Hoover M, Stastny P, Capra JD: Analysis of HLA-DQ genotypes and susceptibility in insulin-dependent diabetes mellitus. N Engl J Med 322:1836, 1990. Reijonen H, Ilonen J, Åkerblom H, Knip M, Dosch H-M: Multilocus analysis of HLA class II genes in DR2 positive IDDM haplotypes in Finland. Tissue Antigens 43:1, 1994. Zeliszewski D, Tiercy J-M, Boitard C, Gu X-F, Loche M, Krishnamoorthy R, Simonney N, Elion J, Bach J-F, Mach B, Sterkers G: Extensive study of DRB, DQA and DQB gene polymorphism in 23 DR2-positive insulin-dependent diabetes mellitus patients. Hum Immunol 33:140, 1992. Sanjeevi CB, Lybrand TP, Landin-Olsson M, Kockum I, Dahlquist G, Hagopian WA, Palmer JP, Lernmark Å: Analysis of antibody markers, DRB1, DRB5, DQA1 and DQB1 genes and modeling of DR2 molecules in DR2 positive patients with insulin-dependent diabetes mellitus. Tissue Antigens 44:110, 1994. Brown JH, Jardetzky TS, Gorga JC, Stern LJ, Urban RG, Stominger JL, Wiley DC: Three-dimentional structure of

HLA-DQ6 and IDDM

18.

19.

20.

21.

22.

23.

24.

25.

the human class II histocompatibility antigen HLA-DR1. Nature 364:33, 1993. Kockum I, Wassmuth R, Holmberg E, Michelsen B, Lernmark Å: HLA-DQ primarily confers protection and HLA-DR susceptibility in Type 1 (insulin-dependent) diabetes studied in population-based affected families and controls. Am J Hum Genet 53:150, 1993. Stern LJ, Brown JH, Jardetsky TS, Gorga JC, Urban RG, Stominger JL, Wiley DC: Crystal structure of the human class II MHC protein HLA-DR1 complexed with an influenza virus peptide. Nature 368:215, 1994. Sanjeevi CB: Molecular mechanism of HLA class II association with insulin-dependent diabetes mellitus. Thesis submitted to Karolinska Institute, Stockholm, Sweden. 49 –55, 1994. Caillat-Zucman S, Garchon HJ, Timsit J, Assan R, Boitard C, Dkilali-Saiah I, Bougueres P, Bach JF: Age dependent HLA genetic heterogeniety of Type 1 insulindependent diabetes mellitus. J Clin Invest 90(6):2242, 1992. Nystrom L, Dahlquist G, Ostman J, Wall S, Arnqvist H, Blohme G, Lithner F, Littorin B, Schersten B, Wibell L: Risk of developing insulin-dependent diabetes mellitus (IDDM) before 35 years of age: indications of climatological determinants for age at onset. Int J Epidemiol 21:352, 1992. Blom L, Persson LA, Dahlquist G: Growth velocity and development of insulin-dependent diabetes mellitus. Diabetologia 35:528, 1992. Hjelmstrom P, Giscombe R, Lefvert AK, Pirskanen R, Kockum I, Landin OM, Sanjeevi CB: Different HLA-DQ are positively and negatively associated in Swedish patients with myasthenia gravis. Autoimmunity 22(1):59, 1995. Hjelmstrom P, DeWeese SC, Penzotti JE, Lybrand TP, Sanjeevi CB: Structural differences between HLA-DQ molecules associated with myasthenia gravis characterized by molecular modeling. J Neuroimmunol 85(1):102, 1998.

153

26. Hoover ML, Marta RT: Molecular modelling of HLA-DQ suggests a mechanism of resistance in type 1 diabetes. Scand J Immunol 45(2):193, 1997. 27. Pugliese A, Gianani R, Moromisato R, Awdeh ZL, Alper CA, Erlich HA, Jackson RA, Eisenbarth GS: HLADQB1*0602 is associated with dominant protection from diabetes even among islet cell antibody-positive firstdegree relatives of patients with IDDM. Diabetes 44(6): 608, 1995. 28. Gianani R, Verge CF, Moromisato GR, Yu L, Zhang YJ, Pugliese A, Eisenbarth GS: Limited loss of tolerance to islet autoantigens in ICA⫹ first degree relatives of patients with type I diabetes expressing the HLA dqb1*0602 allele. J Autoimmun 9(3):423, 1996. 29. Honeyman MC, Harrison LC, Drummond B, Colman PG, Tait BD: Analysis of families at risk for insulin-dependent diabetes mellitus reveals that HLA antigens influence progression to clinical disease. Mol Med 1(5):576, 1995. 30. Ettinger RA, Kwok WW: A peptide binding motif for HLA-DQA1*0102/DQB1*0602, the class II MHC molecule associated with dominant protection in insulindependent diabetes mellitus. J Immunol 160(5):2365, 1998. 31. Ettinger RA, Liu AW, Nepom GT, Kwok WW: Exceptional stability of the HLA-DQA1*0102/DQB1*0602 alpha beta protein dimer, the class II MHC molecule associated with protection from insulin-dependent diabetes mellitus. J Immunol 161(11):6439, 1998. 32. Nepom GT: A unified hypothesis for the complex genetics of HLA associations. Diabetes 39:1153, 1990. 33. Nepom GT, Kwok WW: Molecular basis for HLA-DQ associations with IDDM. Diabetes 47(8):1177, 1998. 34. Schmidt D, Verdaguer J, Averill N, Santamaria P: A mechanism for the major histocompatibility complexlinked resistance to autoimmunity. J Exp Med 186(7): 1059 –1075, 1997.