Ancient haplotypes of the HLA Class II region

Tens to hundreds of functionally distinct alleles have been described for the genes encoding the antigen-presenting molecules HLA-DQA1, -DQB1, and -DRB1 in the class II region of HLA (Bontrop et al. 1999; Marsh et al. 2000). Although there is detailed knowledge of coding-region variation in these genes, the data on haplotype evolution are fragmentary. Such information will be essential for understanding the molecular evolution of HLA and also for carrying out fine-structure mapping of associations between HLA haplotypes and the many human diseases that are influenced by an individual's HLA genotype (Price et al. 1999). Existing data on noncoding variation are either too anecdotal or too localized to particular gene segments to provide an overview of haplotype variation across any major segment of HLA. The few long-range studies that are available are limited to pairwise comparisons of arbitrarily chosen haplotypes (Guillaudeux et al. 1998; Horton et al. 1998; Gaudieri et al. 2000; Satta and Takahata 2000; Stewart et al. 2004); in contrast, short-range studies examine multiple haplotypes only in the immediate vicinity of particular polymorphic exons (McGinnis et al. 1994; Bergstrom et al. 1998; Kotsch and Blasczyk 2000). Despite their limited scope, these data clearly reveal patterns and levels of haplotype variation that have not been seen elsewhere in the human genome (International SNP Map Working Group 2001). At the high end, pairwise divergences reach 5%-10% across tens of thousands of base pairs, while at the low end they collapse to background levels of 0.1% or less. The lack of systematic, long-range sequences of many haplotypes is largely due to the technical difficulty of acquiring haplotype-resolved sequences across tens of thousands of base pairs from the genomes of many individuals. For our analysis, we developed an efficient fosmid-based recloning system to obtain the sequences of 20 human haplotypes, as well as single sequences from a chimpanzee and a gorilla, across an average of 106 kbp of the class II region. Most of the human sequences, which were obtained from a geographically diverse panel of 10 individuals, include all of the DQB1, DQA1, and DRB1 genes. We believe that these data represent the first haplotype-resolved resequencing study of this breadth and depth for any locus in any organism.