Multiple sclerosis: major histocompatibility complexity and antigen presentation

Multiple sclerosis (MS), like many putative autoimmune diseases, has been known to be associated with the human leukocyte antigen (HLA) class II region for more than 3 decades. However, exactly how HLA class II alleles increase the risk of MS is not yet conclusively known. Recent work in large human cohorts has highlighted the fact that nearly all common HLA-DRB1 allelotypes are either positively or negatively associated with the disease, detracting from allele-specific antigen presentation as the sole mechanism of MHC associated disease susceptibility. Here, we put into context recent data on the HLA class II region in MS, including allelic heterogeneity, gene-environment interactions and epigenetics. It is clear that a complete understanding of the epistatic interactions and epigenetic features of this region will be crucial to comprehending disease pathogenesis.


Introduction
Multiple sclerosis (MS) as a disease has been recognized for well over a century, but relatively little is understood about its cause. MS is a putative autoimmune disorder of the central nervous system, characterized by inflammatory demyelination, varying degrees of axonal pathology and progressive neurological dysfunction. Risk factors asso ciated with the disease appear to exert effects many years before the clinical onset of MS, lending credence to the idea of a causal cascade in MS development. Genetic epidemiological studies point unequivocally to large genetic and environmental influences on susceptibility [1]. An association between MS and alleles of the major histo compatibility complex (MHC) was found in the 1970s, notably involving the class II human leukocyte antigen HLADR2 [2]. This was later finemapped to the extended haplotype HLA-DRB5*0101HLA-DRB1*1501 HLA-DQA1*0102HLA-DQB1*0602 [3] (to briefly explain HLA nomenclature, the first two digits of an allele describe its serological antigen (called an allelotype) while the third and fourth digits are used to list the allele subtypes. Alleles with different numbers in these first four digits must differ by at least one nonsynonymous nucleotide substitution). This extended haplotype confers a relative risk of approximately 3, but much larger effects are seen if haplotypic and diplotypic (both haplotypes in combination) information is taken into account, and the odds ratio for risk spanned by variation in the class II HLA region is thought to exceed 30.
Genomewide association studies have highlighted the fact that the HLA class II region exerts by far the strongest genetic effect on risk [4], but exactly how it alters the risk of developing MS is not yet fully understood. As HLA DRB1 alleles have different structural capacities for antigen presentation depending on their amino acid sequence, the MS MHC association has been used to support the concept that disease pathogenesis is the result of an autoimmune reaction, perhaps against myelinrelated antigens in the restricting context of HLA-DRB1*1501. However, it has become clear only very recently that it is now untenable that all MHC related disease risk is due to the DRB1*1501 allele, as was originally thought. This conclusion may be unwelcome for those who have made large investments in the transgenic animal models that depend on it, as these models are now clearly uninformative to truly understand disease pathogenesis.

Allelic heterogeneity
While MS is associated with the HLA-DRB1*1501 haplo type in Northern European populations [3], in other regions like the Mediterranean basin, such as Sardinia, association is predominantly seen with the HLA-DRB1*0301, HLA-DRB1*0405 and HLA-DRB1*1303 haplotypes [5]. HLA-DRB1*13 is also MSassociated in Israel [6], but in continental Italy HLA-DRB1*07 is the primary association [7]. A reexamination of the HLA associations in Northern European MS populations [811], using thousands of patients, uncovered many haplotypes (DRB1*03, *01, *10, *11, *14, *08) that were both positively and negatively associated with the disease. Haplotypes differed in their contribution to disease risk and either acted on their own

Multiple sclerosis: major histocompatibility complexity and antigen presentation
Sreeram V Ramagopalan* † and George C Ebers* † or had an effect in trans with another haplotype. Thus, every major allelotype of HLA-DRB1 is associated with MS (summarized in Table 1).
This conspicuous fact has drawn little attention. Animal models simply transgenic for HLA-DRB1*1501 seem increasingly irrelevant for the study of the human disease because of it [12]. Indeed, it has recently been shown that HLA-DRB1*1501 haplotypes can range from super susceptible to protective depending on other haplotypic features [13]. The HLA-DRB1 association with MS seems to be geographydependent and is probably one deter minant of the latitude gradient in MS incidence that is seen in temperate climes. It is worth considering that both disease and allele gradients could result from similar environ mental pressures. Although associations do reflect the frequency of specific alleles in different countries, the differences among countries cannot completely explain disease frequency. The influence of so many haplotypes on risk, not to mention the prominent interactions, brings into question the venerable belief that MHC associations are determined by structural capacity for antigen presentation.

Structure-function relationships
Different HLA-DRB1 alleles encode proteins with different binding affinities for diseaserelated peptides, as determined by their protein sequence. This has plausibly been considered to influence the composition of T cell repertoires, ultimately resulting in HLA-DRB1 alleles restricting disease risk. However, our analysis [14] has shown that no sequence variant of HLA-DRB1 can fully explain the risk attributable to all diseaseassociated alleles across the globe. One explanation could be that disease causing peptides vary by geography, but the similar disease pathology worldwide would not support this. In the Canadian melting pot of immigrants, MHC associations have remained true to region of origin and give no support to the notion that any geographic specificity of antigenic peptides is relevant (SVR and GCE, unpublished observations).

Environment
Another plausible hypothesis is that the environment of each geographical region interacts with liable HLA-DRB1 haplotypes. In a given population such interaction could influence the likelihood of presenting disease peptides with a timing and tissue localization that will have an impact on MS susceptibility. This makes the assumption that the associations of MHC class II molecules in MS result entirely from roles in specific and restricted antigen presentation to T cells, a dogma that now warrants reconsideration [15].
Environmental factors with convincing evidence for some involvement in MS pathogenesis include sunshine/vitamin D, EpsteinBarr virus (EBV) and smoking [1618]. Twin concordance varies by place of birth, strongly hinting that geneenvironment interactions will be important in MS [19].
There are several ways in which the environment could interact with the MHC. Recent studies have localized a functional vitamin D response element (VDRE) to the promoter region of HLA-DRB1 and this VDRE is always present on HLA-DRB1*15 haplotypes [20]. Although this interaction may have a key role in the increased risk of MS indicated by this haplotype in Northern Europe, it cannot explain why different HLA-DRB1*15 haplotypes confer different risks [13]. More recently, a second interaction has been identified involving the curious monthofbirth effect in MS. This has been linked to the same HLA-DRB1 allele [21].
No studies have yet examined the role of smokingHLA interactions in MS. Investigations of antiEBV antibody levels or symptomatic infection with EBV, HLA-DRB1*15 and the risk of MS have shown that HLA-DRB1*15 may act synergistically with antiEBV antibodies or infectious mononucleosis to increase MS risk [22,23]. The biological nature of this statistical interaction needs to be elucidated, but again it must be remembered that HLA-DRB1*15 is not the only MS risk allele.

Epistasis or haplotype effects
Although other risk components are present on HLA-DRB1 haplotypes in the class II region, and HLA-DQ molecules undoubtedly have a role [5,24], there is no single HLA-DQ element common to all diseaseassociated haplotypes. It does, however, seem that there are combinations of HLA DQB1, HLADQA1 and HLADRB1 that are required to confer risk of MS [25], and investigation of alleles present at HLA-DQ have shed light on haplotypic associations of HLA-DRB1*13 and HLA-DRB1*04 in MS [25]. These haplotypic effects may reflect the effects of selection for functions that are epistatic in nature. HLA class I haplotype tagging can differentiate the risk conferred by different HLA-DRB1*15 haplotypes (despite all having the same alleles of DQ) [13], further indicating that there is more in the MHC than HLA-DQ and HLA-DRB1 in determining MS risk. HLA class I may be an epistatic partner of HLA-DRB1, but given that several class I alleles differentiate HLA-DRB1*15 haplotypes [13] and that HLA class I associations in MS have been conflicting (HLAA, B and C have all been implicated [2628]), it is unlikely that HLA class I has a major role in MS, and the more reliable haplotype transmission data imply it is not an independent contributor to risk [29].

Epigenetics
A missing link seems to be the epigenetic modification of class II region genes. The genetic epidemiology of MS had clearly implied a major epigenetic effect, with mothers more likely to be the common parent in affected half siblings [18] and to be the intervening parent when affected auntniece pairs are studied [30]. This effect has now been localized to the MHC itself [31]. DNA  It may be that MHC disease associations are driven less by allelespecific antigen presentation and more by the propensity of specific haplotypes to undergo strategic epigenetic modifications. The role of DNA methy lation in the process of tissuespecific expression might plausibly relate to the establishment of immunological tolerance, but there is no direct evidence to support such a notion.

Conclusions
The notion of HLA-DRB1*1501 as the one disease allele in MS is rapidly yielding to a more complex view. An orchestra of class II genes, their interactions and their regulatory components have now been shown to be important. The epigenetic pattern within the MHC laid down by differential methylation warrants consideration as the master conductor of MHC diplotypeassociated disease risk.

Competing interests
The authors declare that they have no competing interests.

Authors' contributions
SVR and GCE conceived the idea of the commentary and wrote the manuscript.