Detecting recent natural selection at the human hemochromatosis locus (HFE) using allele age estimates

The high frequency of one hemochromatosis-associated mutation, HFE C282Y, its apparent recent origin, and the nature of this disease have led some to suggest a selective advantage for this mutation. To assess this claim, we surveyed nucleotide variability in the HFE locus for 60 chromosomes from a worldwide sample to determine common polymorphisms and haplotypes. Within a large Caucasian sample, we studied one aspect of linkage disequilibrium, the extent of haplotype sharing at linked sites, for alleles defined by polymorphisms in the HFE gene. We found that for most alleles at HFE the relationship between allele frequency and allele age estimated from the extent of haplotype sharing appears consistent with genetic drift. In contrast, the C282Y allele and one additional allele show a larger extent of haplotype sharing (and thus a younger age) than seems consistent with their high frequency resulting from drift, suggesting these haplotypes have recently risen in frequency due to positive selection. We develop a coalescent algorithm suitable for the simulation of the microsatellite haplotypes commonly used in estimating allele age. Samples produced under complex demographic models can be analyzed in order to produce null distributions of allele age for alleles of specified frequency. We simulate haplotype data under various demographic and recombination parameters and estimate allele ages for comparison to ages estimated for observed HFE alleles. The two alleles with extensive haplotype sharing show a significant deviation from a neutral constant population size model, and we test for models of population growth following a bottleneck that are consistent with all of the observed alleles. Our results indicate that all of the HFE alleles cannot easily be explained by any one model and that selection has affected multiple haplotypes.