Over the past three decades, substantial progress has been made in understanding the genetic basis for diabetes. One of the important first steps that allowed this progress to occur was realizing that diabetes is heterogeneous, and, therefore, separation of clinically distinct forms of the disorder (i.e., type 1 vs. type 2 diabetes) improves the ability to detect genetic associations. The key points that allowed type 1 diabetes to be separated from type 2 diabetes included realization of the clinical differences (typically childhood onset, thin, ketosis-prone versus adult onset, obese, nonketosis prone); family and twin studies that demonstrated that the two forms of diabetes usually segregate separately and, while both demonstrate substantial monozygotic twin concordance, the concordance rate in type 2 diabetes is at least double that in type 1 diabetes; and appreciation that there is automimmune β-cell destruction in type 1 diabetes that does not occur in type 2 diabetes (rev. in 1). The confirmation that these clinical observations truly represented genetic differences came from the early studies of the human leukocyte antigen (HLA) region, which we now know plays an important role in type 1 but not in type 2 diabetes susceptibility.
HLA, the human form of the major histocompatibility complex (MHC), has indeed long been recognized as the major genetic region influencing risk for type 1 diabetes. The fact that it was the first genetic susceptibility region identified was, in part, serendipitous, as the emerging ability to distinguish a variety of HLA-A and -B serotypes made HLA one of the first highly polymorphic genetic regions that could be used in linkage and association studies. In retrospect, the autoimmune basis of type 1 diabetes made it a logical disorder to be linked to genes in the MHC, but, given the dearth of genetic markers available back in the 1970s, diabetes researchers would likely have tested the region, regardless of whether anything was known about its role in the immune system. Remember ABO and Kidd blood group testing (2,3)? Back in those days, we were desperate to test any polymorphic marker that we could lay our hands on.
The strong linkage disequilibrium (LD) that exists within the MHC has played a substantial role in allowing disease linkages and associations to be detected. The early studies that reported associations with HLA-B8 and -B15 used remarkably small numbers of cases and controls relative to the hundreds to thousands considered necessary in current studies (4–6). Had it not been for the strong LD, those initial studies would have been unlikely to yield positive results. Yet, at the same time, this LD has also created challenges in accomplishing the next step, that of identification of the specific genes that are responsible for disease susceptibility. The fact that researchers have spent >30 years trying to elucidate all of the loci responsible for type 1 diabetes susceptibility in the HLA region underscores the complexity of the task. While researchers attempting to localize the gene responsible for a linkage signal or association detected in other genomic segments have often struggled with the lack of an obvious candidate gene within the region, type 1 diabetes investigators have had the converse problem with HLA, i.e., too many candidate genes, all plausibly involved in autoimmune regulation/disregulation. When the initial HLA-B8 and -B15 associations were followed rapidly by identification of associations with HLA-DR3 and -DR4, interest in the class I HLA region was replaced by enthusiasm for the class II genes (7,8). Interest in the DR locus was then superceded by excitement over DQ and so forth (9). While the exact loci (and polymorphisms within loci) that account for HLA-linked susceptibility are still not clearly defined, the results of many reports show that multiple loci, including, at a minimum, DRB1, DQB1, DP, as well as HLA-A and HLA-B, all contribute (10–16).
Understanding the mechanisms by which loci in the HLA region result in diabetes susceptibility is made difficult by the tremendous genetic heterogeneity within this region. When initial HLA associations with type 1 diabetes were reported, each of the HLA loci had a handful of reported alleles. Current totals are approaching 1,000 for some of the most polymorphic HLA loci, such as HLA-B and -DRB1 (see http://www.anthonynolan.org.uk/HIG/index.html). Additional complexity is generated by the fact that for some loci (such as DQB1) polypeptide products from both chromosomes can form “trans-encoded” heterodimeric proteins that are thought to be major contributors to disease risk. Simply put, HLA region susceptibility to type 1 diabetes is extremely complicated. Thus, while great advances have occurred in the field over the past 30 years, much more remains to be learned. The article by Aly et al. (17) in this issue of Diabetes represents one more significant step toward unraveling the mysteries of the HLA region.
The research reported by Aly et al. (17) provides not only additional support for contributions by classical HLA loci, but also convincing evidence that at least one additional locus, in the vicinity of UBD/MASIL, is also involved. Thus, the stretch of chromosome 6p that contributes to type 1 diabetes susceptibility appears to be at least 4 Mb in length, twice as long as previously thought.
Aly et al. (17) state that “[t]here has been a renewed interest in” finding additional type 1 diabetes susceptibility loci within the HLA region. In fact, this interest has persisted for 30 years, as illustrated by the numerous type 1 diabetes association and linkage studies of other genes within the HLA region, including (among others) TNF, MICA, and TAP (18–23). As molecular technology has advanced, so has our ability to delve more deeply into the intricacies of the HLA region.
The work reported by Aly et al. (17) demonstrates several important points. First, the use of intensive single nucleotide polymorphism (SNP) genotyping across a region, even one demonstrating as much LD as the MHC, can help distinguish discrete loci that may each be contributing to overall disease susceptibility. Second, just because one or more putative causative genes have been identified within a region, one should not assume that all of the genes important in disease susceptibility in that region have been found. In fact, the magnitude of the evidence for linkage between this chromosome 6 region and type 1 diabetes, when contrasted to the more modest predisposition that can be accounted for by any specific gene association within the region, is one of the compelling arguments favoring the existence of at least two, and most likely several, genes that contribute to diabetes susceptibility in the region (24). Much about the human genome remains to be understood, and while in some cases two or more susceptibility genes may lie in close proximity by chance alone, in other cases an extended gene region may contain a cluster of interacting genes. This is not a new concept. Walter Bodmer eloquently discussed such clusters, using HLA as the emblematic example, in his 1980 Allen Award address (25). How and why such clusters of physically and pathophysiologically related genes developed, as well as what mechanisms have acted to perpetuate their persistence, however, will need to be the topic of ongoing investigations.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
See accompanying Original Article, p. 770.