Defining the Role of the MHC in Autoimmunity: A Review and Pooled Analysis

The major histocompatibility complex (MHC) is one of the most extensively studied regions in the human genome because of the association of variants at this locus with autoimmune, infectious, and inflammatory diseases. However, identification of causal variants within the MHC for the majority of these diseases has remained difficult due to the great variability and extensive linkage disequilibrium (LD) that exists among alleles throughout this locus, coupled with inadequate study design whereby only a limited subset of about 20 from a total of approximately 250 genes have been studied in small cohorts of predominantly European origin. We have performed a review and pooled analysis of the past 30 years of research on the role of the MHC in six genetically complex disease traits – multiple sclerosis (MS), type 1 diabetes (T1D), systemic lupus erythematosus (SLE), ulcerative colitis (UC), Crohn's disease (CD), and rheumatoid arthritis (RA) – in order to consolidate and evaluate the current literature regarding MHC genetics in these common autoimmune and inflammatory diseases. We corroborate established MHC disease associations and identify predisposing variants that previously have not been appreciated. Furthermore, we find a number of interesting commonalities and differences across diseases that implicate both general and disease-specific pathogenetic mechanisms in autoimmunity.


Multiple Sclerosis
Multiple sclerosis (MS, MIM 126200) is a chronic inflammatory demyelinating disorder of the central nervous system (reviewed in [1,2]). The disease has a predilection for females and may manifest as relapsing and/or progressive forms. The only region of the genome that has shown consistent evidence of linkage and association with multiple sclerosis (MS) is the HLA locus (reviewed in [3]). The association of the HLA-DR2 haplotype with multiple sclerosis was first noted in 1972 [4] and remains one of the most reproduced findings in MHC genetics [5][6][7][8][9][10][11][12][13][14][15].
In all of the linkage scans performed to date, the HLA locus is the only one to have achieved genome-wide significance as an MS susceptibility locus both in a meta-analysis of linkage studies [10] and in a subsequent large highresolution linkage screen in MS [11]. An initial SNP-based association study of the MHC as well as more focused efforts dissecting the effects of the MHC in MS have recently confirmed the preeminent role of the DRB1*1501 (encompassed within the DR2 specificity) haplotype in MS in populations of northern European ancestry [12][13][14]. Indeed, populations with a high concentration of the DRB1*1501-DQB1*0602 haplotype, such as northern Scotland, exhibit some of the highest known prevalence rates of MS [16].
Due to strong LD within the MHC, it remains unclear whether the primary driver for the association of the HLA locus to MS is the DQB1*0602 allele or the DRB1*1501 allele (reviewed in [3]). However, evidence is mounting that, of these two alleles, DRB1*1501 appears to be the allele with stronger evidence of association both in African-Americans [17] and, possibly, in European populations (PLD, unpublished data). The long-range LD noted within the DRB1*1501-DQB1*0602 haplotype and its effect strength (OR = 2.7 for one copy of the DRB*1501 allele, [18]) hinders the interpretation of other associations in this region, even those several megabases away in the MHC class I region. Thus, it is uncertain whether MHC class I associations are truly independent of the DRB1*1501-DQB1*0602 haplotype effect [3], though some evidence suggests that the DRB1*1501-DQB1*0602 associated HLA-A3 allele may be an independent susceptibility signal in MS [19]. Investigators in the Netherlands have also reported an association with the microsatellite HLA C1_3_2*354 that is independent of DR2, although in strong LD with DR3 [20].
A more recent study of a different family-based cohort failed to demonstrate such independent HLA-A or -B effects [21]. On the other hand, another study that took both the DR15 and DR3 effects into account offers substantial evidence that the HLA C*05 allele may be independently associated to MS susceptibility [14]. Thus, the role of class I alleles remains open for now and awaits the interrogation of much larger subject samples in an effort to powerfully address the issue of DRB1-independent susceptibility loci.
Within the DRB1 gene itself, there appears to be evidence for allelic heterogeneity, particularly in non-European populations [22]. Nonetheless, the haplotype tagged by HLA DRB1*1501 remains a risk haplotype for MS in non-European populations, especially when one considers the typical disseminated forms of demyelinating disease captured by the MS diagnostic rubric [23][24][25]. Barcellos and colleagues [13] have recently explored the question of allelic heterogeneity in 1339 families of European ancestry with MS, and this robust analysis reports a dominant effect for DRB1*15, a recessive effect for DRB1*03, a protective effect for DRB1*14, as well as evidence for interaction between DRB1*15 and DRB1*08 in increasing susceptibility to MS. A separate study that shared some subjects with the latter study suggests that the HLA DRB1*0103 may also be a susceptibility allele, once the effects of DRB1*15 and DRB1*03 are taken into account [14].
A small study of Portuguese subjects recently also found an association to HLA DRB1*03 [26], as did earlier studies of Sardinian families [27].
Mediterranean populations such as the Sardinians also display associations to DR4 (reviewed in [1], and the DR4 effect has now been demonstrated in another Italian population consisting of families with MS and other autoimmune diseases [28]. Finally, as with the DRB1*14 allele described above, several HLA class II alleles (DR1, DR7, and DR11) have been implicated in protection against MS [29], while HLA-A*0201 has been shown to reduce DRB1*15-associated risk [19]. Such protective effects are notoriously difficult to distinguish from under representation of certain alleles in the face of the effects of DRB1*15, and therefore these findings await further validation.
Evidence for association of other DRB1 alleles comes from other human populations. In Japan, disseminated MS similar to that seen in Europeans may be associated with DRB1*1501, but the association of the DRB1*1501 haplotype with the opticospinal MS variant fails to reach significance (reviewed in [24]). Instead, opticospinal MS may be associated with DPB1*0501 and DPB1*0301 in the Japanese population [24,30]. An association with DR6 in Japanese has also been reported, but without replication [31]. In the Turkish population, an association has been suggested with DR4 ( [32]reviewed in [33]). The same haplotype is also reported to be associated with susceptibility to MS in Mexican Mestizos [34]. In African-Americans, MS association has also been demonstrated with DRB1*0301 and DRB1*1503 in addition to the expected DRB1*1501 [17].
Outside of MHC genes, polymorphisms in the immunologically relevant class III genes, TNF and NOTCH4, have been associated with MS, but this association was shown to be secondary to that of DRB1*1501 [35]. At least 10 small studies have shown no HLA-independent association with microsatellites and SNPs in linkage disequilibrium with the TNF gene (ibid.), though one as yet unreplicated larger study [36] suggests an association in a Spanish population. Recently, a small study of Tasmanian families has highlighted a haplotype containing a myelin oligodendrocyte glycoprotein (MOG) allele after conditioning for the effect of HLA-DRB1*15 [37]. Thus, the role of non-HLA genes also remains an area of active investigation that will benefit from much larger studies that comprehensively interrogate genetic variation within the extended MHC in MS. Two other ancestral haplotypes, DR3 and DR4, appear to play a role in MS susceptibility, although the effect of these haplotypes on disease is more modest than that of the DR2 haplotypes. Figure 1 [40,41]; and overlapping MOG peptides [42]. The data on DR4 alleles is limited, with the MBP 111-129 epitope being recognized in subjects with DRB1*0401 [43], but PLP and MOG epitopes were also recognized in one study where subjects with at least one DR4 allele were studied [44]. Further studies are required to fully elucidate and validate the allelic heterogeneity seen at DRB1, particularly in terms of the role of rare alleles such as DRB1*0403; recent efforts are offering increasing evidence supporting the role of the more common DRB1*03 allele in disease susceptibility [13,14].