Unravelling major histocompatibility complex diversity in the Soay sheep of St Kilda

The major histocompatibility complex (MHC) is one of the most variable regions in the vertebrate genome. Many genes within the MHC play important roles in the development of an immune response, including the response to pathogens, by presenting pathogen fragments to T cells. Pathogen-mediated balancing selection is thought to be important in maintaining the high levels of allelic variation at these loci, though the precise mechanism remains unclear. The number of studies of MHC diversity in non-model organisms has increased dramatically in recent years as genotype data have become cheaper and easier to generate; however, key limitations in many such studies remain a lack of high quality MHC genotypes and associated phenotype data. Many studies focus on a single MHC locus, assuming that one locus will represent the full range of variation within each MHC haplotype. Alternatively, the products of different loci may co-amplify, preventing locus-specific genotypes and hence heterozygosity being accurately determined. Non-model systems are also often limited by small sample sizes and limited recording of associated host and pathogen measures, which, combined with high levels of allelic variation at MHC loci, can limit statistical power. Finally, few MHC studies control for the general effect of relatedness in explaining host traits before testing for MHC effects. With so many methodological impediments, it is challenging to identify robust associations between MHC variation and host phenotypes, such as parasite burden or fitness, and to draw conclusions about the mechanisms underpinning the maintenance of diversity at MHC loci. In this thesis, I address these problems by developing a SNP-based haplotyping system for a population of unmanaged Soay sheep (Ovis aries) on Hirta, St. Kilda, for which data is available on pedigree, phenotypic traits and fitness and its components over a 30- year study period. The ovine MHC consists of four classes of loci, within which loci are tightly clustered and show reduced recombination rates compared to the genome average. Although the mammalian MHC is usually highly variable, one would expect that the number of haplotypes within an MHC class in an island population of sheep with no immigration to be limited. The class IIa region of the ovine MHC includes the classical class II loci which are typically thought to be involved in the presentation of peptides derived from extracellular pathogens, including gastrointestinal helminths, in sheep and other mammals. In chapters 2 to 4, I describe the characterisation of class IIa haplotypic diversity in the Soay sheep using direct Sanger sequencing of PCR amplified fragments, which, in combination with cloning, revealed eight distinct haplotypes. With this knowledge of haplotypic diversity, and genotypes for a sample of Soay sheep typed on the Ovine Infinium HD chip (approximately 600K SNPs), I developed a panel of 13 SNPs which could be used to impute the class IIa haplotypes. This panel was genotyped by KASP (Kompetitive Allele Specific PCR) in 6034 samples and used to impute the class IIa haplotypes. After quality control measures, class IIa haplotypes were successfully imputed for 5349 individuals. Evidence of balancing selection was identified using the Ewens-Watterson test at different life history stages and within the standing population each year between 1985 and 2012, showing that allele frequencies were more even than would be expected under neutrality. However, there was no evidence of deviation from Hardy-Weinberg equilibrium identified at different life stages or in the standing population in any year. In chapter 5, I investigate associations between the MHC class IIa haplotypes and individual-level data on host phenotypes - body weight, plasma immunoglobulin levels (measured as anti-Teladorsagia circumcincta third larval stage IgA, IgE and IgG) and strongyle faecal egg counts (FEC). Associations were tested within mixed effects models which were used to account for repeated measures and control for fixed effects known to affect the response variables, as well as within an animal model framework to account for relatedness between individuals. Haplotype heterozygosity was unrelated to any of the traits investigated, suggesting a general heterozygote advantage is unlikely to be operating within the Soay sheep. Six of the eight class IIa haplotypes were associated with multiple traits in different age-sex classes, although many of these associations were removed after inclusion within animal models. The evidence of balancing selection and associations between class IIa haplotypes and phenotypes related to health offers a promising glimpse into the evolutionary mechanisms which may be operating to maintain diversity within this region.