HLA-DR and -DQ
The highest risk of T1DM is conferred by heterozygosity for the DRB1*0301-DQA1*0501-DQB1*0201 and DRB1*04-DQA1* 0301-DQB1*0302 haplotypes, referred to as the DR3.DQ2/DR4.DQ8 genotype. This allelic combination is carried by 30–40% of individuals with T1DM, but only around 2.5% of the general population . A recent meta-analysis of multiple ethnic groups suggested that this translates into an OR value greater than 16, an unusually large odds ratio for a complex disease . This is consistent with an earlier study which estimated that the risk of developing T1DM was between 1 in 15 and 1 in 25 among those with the DR3.DQ2/DR4.DQ8 genotype, compared with 1 in 300 in the general population . High risk is also conferred by the DR3.DQ2/DR3.DQ2 and DR4.DQ8/DR4.DQ8 homozygous genotypes (OR = 6.32 and OR = 5.68, respectively, from meta-analysis) . Approximately 95% of subjects with T1DM are positive for DR3.DQ2 and/or DR4.DQ8 compared with 40–50% of the general population . It is noteworthy that heterozygosity for these haplotypes confers a much greater risk than would be expected from the combined OR values observed for each homozygous genotype, suggesting that the alleles on each haplotype interact synergistically to promote disease development. This may be due to the formation of a highly diabetogenic DQ molecule, encoded in trans by DQA1*0501 on the DR3 haplotype and DQB1*0302 on the DR4 haplotype. This potential synergism supports the conclusion that the HLA genotype of an individual (the combination of alleles inherited from both parents) is a better indicator of disease risk than the presence or absence of specific risk alleles.
Although the DR3.DQ2 and DR4.DQ8 haplotypes are the strongest determinants of T1DM risk, there is a complex hierarchy of allelic, haplotypic, and genotypic effects of the DR/DQ loci, which encompass the spectrum from highly susceptible, through intermediate risk, to very protective markers. The haplotypes most strongly associated with T1DM (predisposing and protective) are summarized in Table 30.1. As shown, the risk conferred by DRB1*04 haplotypes varies considerably according to the specific DRB1 allele present. DRB1*0405 and *0401 are associated with the highest risk, while DRB1*0403 confers protection against T1DM . Indeed the protective influence of DRB1*0403 is able to override DQA1*0301- DQB1*0302-associated susceptibility, even among individuals carrying the high risk DR3.DQ2/DR4.DQ8 genotype . The DRB1*1501-DQA1*0102-DQB1*0602 (DR15.DQ6) haplotype is particularly interesting as it confers strong protection against T1DM, occurring in less than 1% of individuals with the disease compared with 20% of the general white European population.The protective effect is dominant over susceptibility conferred by other alleles; individuals carrying the DR15.DQ6 haplotype in combination with the predisposing DR3.DQ2 or DR4.DQ8 haplotype do not usually develop diabetes . The analysis of the transmission of DR/DQ alleles from unaffected parents to diabetic offspring suggests that protection is likely to be conferred by DQ6 (DQA1*0102-DQB1*0602) rather than DRB1*1501. Rare haplotypes carrying DRB1*1501 in the absence of DQ6 are transmitted significantly more frequently than the DQ6 allelic combination alone . DQ6 can prevent progression to overt T1DM even after the onset of islet autoimmunity, suggesting that the molecule encoded by these alleles may have an immunomodulatory role . The protective effect is not absolute, however, as a small number of individuals with T1DM have been shown to be positive for DQA1*0102-DQB1*0602 [21,22]. It is possible that protection from disease is attenuated in these subjects by the effects of other loci [23–25]. Other DR/DQ haplotypes also exert strong protection from T1DM (Table 30.1), but are much less common in the background population than DR15.DQ6.
It is difficult to dissect the relative contributions of the DR and DQ genes to disease susceptibility using purely genetic means as the tight LD between the loci hampers attempts to dissociate the effect of a given allele from the effects of other alleles with which it is coinherited. Studies of HLA-transgenic animals, however, have suggested that both DR and DQ molecules play a role in the pathogenesis of T1DM. B10 mice expressing human DR3 (DRB1*03) or DQ8 (DQA1*0301-DQB1*0302), singly or in combination, developed spontaneous T-cell reactivity to glutamic acid decarboxylase (GAD65), a β-cell protein implicated in the autoimmune disease process, while insulitis occurred in DR3/DQ8 double transgenic animals [26,27]. Expression of DQ8 or DR4 (DRB1*0401) in C57BL/6 mice was unable to induce insulitis, but when these animals were crossed with transgenic mice expressing the T cell co-stimulatory molecule, B7.1, on their pancreatic β cells, the resulting double transgenic offspring developed spontaneous diabetes [28,29]. Similarly, the coexpression of B7.1 was necessary for diabetes to develop in mice transgenic for DR3 and/or DQ8 . Interestingly, in this study, the DR3 and DQ8 molecules appeared to be equally permissive for islet autoimmunity. In contrast, Kudva and coworkers reported greater insulitis in the presence of DQ8 alone, with the coexpression of DR3 apparently decreasing the frequency and severity of autoreactivity in double transgenic animals . A similar observation was made in DR4/DQ8 transgenic mice; DQ8 appeared to be the more diabetogenic molecule, with its effect being downregulated by the coexpression of DR4 . These studies suggest a primary role for DQ in the development of autoimmunity, with DR playing a regulatory role. Significantly, transgenic mice expressing the DQ6 molecule did not develop autoimmunity to GAD65, even in the presence of DR3 or DQ8 ; this is consistent with the dominant protective effect reported for DQ6.
The influence of a particular HLA molecule on diabetes risk is largely determined by its three-dimensional structure, which governs the way in which it interacts with autoantigenic peptides and the receptors of autoreactive T cells.
The antigen-binding sites of diabetogenic molecules, such as DQ2, DQ8, and I-Ag7 (the diabetes-predisposing MHC molecule found in nonobese diabetic [NOD] mice), share similar chemical and geometric properties that dictate the shape of the critical pockets into which the amino acid side chains of antigen peptides are anchored during presentation. Likewise there is marked structural similarity between the Diabetes-protective molecules, such as DQ6, DR4 (DRB1*0403), and murine I-Ab . The structural features differ markedly between the two groups, however, leading to differences in antigen selectivity, binding affinity, and molecular stability. It is unclear how this translates into effects on disease risk, but it has been hypothesized that diabetogenic molecules may bind poorly to autoantigenic epitopes, leading to ineffective tolerance induction and subsequent autoimmunity. In contrast, the protective molecules bind well to self antigens, promoting efficient thymic deletion of autoreactive T cells. The potential “loss of function” mechanism for the predisposing molecules is supported by the observation that transgenic overexpression of the diabetogenic allele, I-Ag7, in NOD mice protects from diabetes, presumably by compensating for the low affinity of the encoded molecule for autoantigens .