HLA DNA sequence variation among human populations: molecular signatures of demographic and selective events

Abstract

Molecular differences between HLA alleles vary up to 57 nucleotides within the peptide binding coding region of human Major Histocompatibility Complex (MHC) genes, but it is still unclear whether this variation results from a stochastic process or from selective constraints related to functional differences among HLA molecules. Although HLA alleles are generally treated as equidistant molecular units in population genetic studies, DNA sequence diversity among populations is also crucial to interpret the observed HLA polymorphism. In this study, we used a large dataset of 2,062 DNA sequences defined for the different HLA alleles to analyze nucleotide diversity of seven HLA genes in 23,500 individuals of about 200 populations spread worldwide. We first analyzed the HLA molecular structure and diversity of these populations in relation to geographic variation and we further investigated possible departures from selective neutrality through Tajima's tests and mismatch distributions. All results were compared to those obtained by classical approaches applied to HLA allele frequencies.Our study shows that the global patterns of HLA nucleotide diversity among populations are significantly correlated to geography, although in some specific cases the molecular information reveals unexpected genetic relationships. At all loci except HLA-DPB1, populations have accumulated a high proportion of very divergent alleles, suggesting an advantage of heterozygotes expressing molecularly distant HLA molecules (asymmetric overdominant selection model). However, both different intensities of selection and unequal levels of gene conversion may explain the heterogeneous mismatch distributions observed among the loci. Also, distinctive patterns of sequence divergence observed at the HLA-DPB1 locus suggest current neutrality but old selective pressures on this gene. We conclude that HLA DNA sequences advantageously complement HLA allele frequencies as a source of data used to explore the genetic history of human populations, and that their analysis allows a more thorough investigation of human MHC molecular evolution.

Figure 1. HLA genetic diversity within populations.

A) heterozygosity (h) and B) nucleotide diversity (π_n) within populations grouped according to their geographic location, at each locus under study. Only the regions represented by populations at every locus are illustrated on the graphs. See File S2 for detailed values (means and standard deviations). Sub-Saharan Africa (SAF), Europe (EUR), Southwest Asia (SWA), Northeast Asia (NEA), Southeast Asia (SEA), , Pacific (PAC), Australia (AUS), North America (NAM), and South America (SAM). See Supporting Information S1 for the list of populations included in each region.

Figure 2. Distributions of pairwise differences between HLA sequences (mismatch distributions).

Average mismatch distributions within populations grouped according to their geographical location, at each HLA locus under study. Histograms at the top-right indicate the proportion of allele pairs with 0, 1–10, 11–20, 21–30, 31–40 and more than 40 nucleotides of divergence within each group of populations.

Figure 3. Tests of selective neutrality.

Percentages of significant outcomes for Tajima's D statistic and Ewens-Watterson test after Bonferroni's correction, at each locus under study (ordered according to their location on chromosome 6, from telomere to centromere).

Figure 4. Summary of the comparisons of Reynolds genetic distance matrices between the molecular and allelic approaches.

Percentage of populations exhibiting a majority of higher Reynolds genetic distances computed with the molecular approach compared to Reynolds genetic distances computed with the allelic approach (see “Statistical analyses” in Materials and Methods for the definition of the two approaches) within geographic regions, at each locus under study. The detailed comparisons of Reynolds matrices are given in File S4. Percentages were only computed for regions including at least 5 populations (n≥5).

Figure 5. Plot comparisons of Reynolds genetic distance matrices in two populations.

Plot comparisons of Reynolds genetic distance matrices computed with either the molecular (X-axis) or the allelic (Y-axis) approach (see “Statistical analyses” in Materials and Methods for the definition of the two approaches) in Samoans for HLA-B (A) and in Lebanese Arabs for HLA-DRB1 (B). The red squares represent the genetic distances between the single out population and the other populations available in the databank at this locus. The blue circles represent the genetic distances between each pair of populations at this locus, excluding the single out population. The black line on the plot indicates the equality of distance between both approaches. The detailed comparisons of Reynolds matrices are given in File S4.