Extended linkage disequilibrium surrounding the hemoglobin E variant due to malarial selection

Abstract

The hemoglobin E variant (HbE; ( beta )26Glu-->Lys) is concentrated in parts of Southeast Asia where malaria is endemic, and HbE carrier status has been shown to confer some protection against Plasmodium falciparum malaria. To examine the effect of natural selection on the pattern of linkage disequilibrium (LD) and to infer the evolutionary history of the HbE variant, we analyzed biallelic markers surrounding the HbE variant in a Thai population. Pairwise LD analysis of HbE and 43 surrounding biallelic markers revealed LD of HbE extending beyond 100 kb, whereas no LD was observed between non-HbE variants and the same markers. The inferred haplotype network suggests a single origin of the HbE variant in the Thai population. Forward-in-time computer simulations under a variety of selection models indicate that the HbE variant arose 1,240-4,440 years ago. These results support the conjecture that the HbE mutation occurred recently, and the allele frequency has increased rapidly. Our study provides another clear demonstration that a high-resolution LD map across the human genome can detect recent variants that have been subjected to positive selection.

Figure 1

Biallelic markers used in this study and profile of LD around HbE. A, Location of nine genes in the genomic region studied. B, Location of HbE and 121 biallelic markers. C, Genotypes of HbE and 121 biallelic markers in 15 individuals. D, Pairwise |D′| between HbE and 43 biallelic markers in an enriched set of 64 individuals (see the “Material and Methods” section for details). The interval between regions I and II spans 2.0 kb.

Figure 2

Haplotype network for HbE and 10 biallelic markers in region I. The four major haplotypes with an estimated population frequency of >0.03 are shown as open circles, with the estimated haplotype frequency shown at the right. Each line segment between two circles indicates a single mutation event; small closed circles represent putative intermediate haplotypes.

Figure 3

Pairwise |D′| between HbE and 43 biallelic markers in a random set of 48 individuals (see the “Material and Methods” section for details).

Figure 4

Frequency distributions of the age of HbE obtained from the simulations under dominant- and overdominant-selection methods. The population size was assumed to have been constant (g=0) or to have been increased by a growth rate of 0.001/generation (g=0.001). The means and 95% credibility intervals are presented in table 2.