Editorial overview

Autoimmunity: insights from human genomics

Early efforts to localize and identify genes that contribute to the occurrence of human disease often used a limited subset of variants in candidate genes, either in case–control association analyses or family-based linkage studies. These studies typically focused on candidate genes selected based upon a perceived biological pathway, and variants that were selected on the basis of putative functional effects. At that time, however, our knowledge of human genomic variation and the nature of variants within specific genes were still lacking. This resulted in the publication of many findings with low statistical power and, often, with little potential for replication. As genotyping strategies and technologies evolved, there was a shift from candidate gene studies to genome-wide linkage scans in families. While linkage studies provided more coverage of the genome, they were often limited by the availability of large collections of families. Also, the primary focus of linkage scans was on multiple affected individuals in a family, thereby implicitly testing for a model where a handful of alleles had large effects on disease risk.

Recently, genome-wide association scans (GWAS) have profited from both increased genomic coverage and cost reduction in genotyping. As a result it has become possible to genotype large numbers of common variants in collections of large sample size. These GWAS studies have succeeded in identifying and replicating novel common genetic variation associated with the spectrum of common autoimmune diseases, including type 1 diabetes [1], ulcerative colitis [2], Crohn's disease [3], systemic lupus erythematosus (SLE) [4], alopecia areata [5], rheumatoid arthritis [6], and multiple sclerosis [7]. Unlike the family-based linkage and candidate gene studies, the individual allelic effect sizes detected by GWAS studies (excluding the MHC) are relatively small, consistent with the common variant hypothesis, with each allele accounting for only a small portion of genetic variance. Described by Voight and Cotsapas, in this issue, there has also been confirmation that many loci are shared across multiple autoimmune diseases [8]. Even though many loci have been identified by GWAS, the goal remains not only the discovery of individual loci or genes, but also the discovery of new biology. For the vast majority of GWAS findings, however, the functional variants remain yet to be determined; however, there has been progress recently in Crohn's disease (e.g. [9, 10]).

The findings in the GWAS era have recently been criticized, on the basis of the difficulty in pinpointing specific genes and causal variants, the limited amount of genetic risk burden that has been explained (missing heritability) and the incomplete genomic coverage. However, the use of GWAS technologies has provided the locations for further characterization of the genes and variants in a genomic region that statistically will contain genes and variants contributing to disease risk. Gene and causal variant discovery will continue as technical advances allow for denser genotyping and sequencing [11] as well as access to integrated health care systems that will permit rapid recruitment of patients from electronic medical records [12].

With these advances in technology, however, there are new challenges to understanding the genetic basis of autoimmunity and disease. To date, experimental strategies to understand the function of disease susceptibility alleles have been tailored toward those with highly penetrant effects that typically cause familial diseases. In contrast, the vast majority of discoveries in autoimmune diseases have resulted from GWAS studies of common variants [13]. Since these alleles are common they may play a role in influencing disease susceptibility in the general population — but with only a small effect on an individual. Common variant alleles represent mutational events that have survived the pressures of evolution, escaped purifying negative selection, and have therefore been able to propagate within a population [14]. Thus, within an individual, a common risk variant may have only modest and very specific influence on immune function. As Buckner and Nepom describe in their article, different variants for the same disease can influence disease progression at distinct steps in immune activation, expansion, and regulation. Their functional influences are likely to be cell-specific and subtle. Indeed, common risk variants may alter function differently in different cell-types (e.g. the same common variant can have disparate effects on gene expression in lymphoid and myeloid cell-types [15]). Traditionally, the evaluation of rare penetrant mutations has generated a standard classification that includes ‘loss-of-function’ or ‘gain-of-function’; however, with common variants, a more appropriate paradigm might be ‘altered function’.

Once an association is identified, determining which of many alleles may be functional and may account for the observed association with disease can be challenging. In many instances, due to linkage disequilibrium, an entire set of variants may demonstrate equivalent levels of association. Genetic evidence alone may not be able to identify which specific variant is contributing biologically to the association [13]. One rare instance where an observed SNP association was fine-mapped successfully to causal variants was the IRF5 locus in SLE. The initial association of SLE to a common SNP ultimately was explained as an exonic splice site variant, a 30-bp in-frame insertion/deletion, and a variant in a conserved polyA+ signal sequence that alters the length of the 3′ UTR and stability of IRF5 mRNAs [16]. For other loci, the path from GWAS to fine-mapping to detection of causal variation has yet to occur.

In cases where common variants have been shown to alter function, it is possible to associate disease with specific biological mechanisms. In this issue, Gardet and Xavier describe how common variants in Crohn's disease alter autophagy, normally thought to be a homeostatic response that ensures the recycling of cellular materials to promote cell survival. This unanticipated finding has established a new research direction for autophagy mucosal homeostasis. Even in the absence of molecular evidence of disease mechanism, it is possible to link common variants to specific disease pathways. In this issue, Bronson et al. present compelling evidence linking interferon abnormalities to SLE using genetics and functional genomics. These results are both influencing drug discovery and the mechanisms involved in autoimmune disease.

In order to make progress in understanding the genetic basis of autoimmunity, several approaches need to be pursued in parallel. The first is continued gene and causal variant discovery. As additional novel alleles are discovered they will continue to contribute to help identify novel pathways and potential therapeutic targets. A second approach would implement targeted DNA sequencing in regions of susceptibility. Sequencing will identify rare and highly penetrant alleles, some of which may be particularly useful in designing biomarkers and interventions. These rare variants, especially if they reside in coding regions of genes, will not only identify the gene as causal but may have more obvious function. This approach has already identified likely pathogenic genes in Crohn's disease and type 1 diabetes [17, 18, 19].

A third and potentially informative approach for understanding the role of common variants in autoimmunity is to broadly examine functional quantitative traits, for example expression quantitative traits, reviewed by Stranger and de Jaeger, and immune quantitative traits, described by Gregersen et al. Utilizing large populations with robust experimental and analytical approaches to examine expression and immune phenotypes could identify those variants that strongly influence immune responses. One obvious challenge is to identify those assays that are most valuable and relevant to look at from the perspective of autoimmunity and human disease. As massive parallel profiling becomes feasible, this approach may emerge as a powerful parallel method to augment genomic studies [20].

For human immunologists, the discoveries from genetics should be viewed as a great success, creating a platform for countless avenues of investigation into the immune system and how host genomic variation modifies this response. This is a particularly exciting time with the convergence of human genetic studies, human immunologic studies and an ever-increasing platform of reagents and model systems for probing immune biology. As Hu and Daly note, integrative approaches examining many alleles and their functions in concert might yield insight into disease mechanisms. Ermann and Glimcher write how, in certain key instances, traditional mouse models may present an opportunity to investigate specific allelic functions, and to connect to disease phenotypes. The integration of epigenetic data with genetic data, as reviewed by Leung et al. offers another promising avenue for progress, particularly in light of recent observation within data produced by the ENCODE consortium of the overlap between disease alleles and regulatory chromatin marks [21, 22]. The breadth and quantity of human risk alleles has given investigators many leads to follow on in order to understand the molecular underpinnings of autoimmune diseases.