Patterns of nucleotide and haplotype diversity at ICAM-1 across global human populations with varying levels of malaria exposure

Abstract

Malaria is one of the strongest selective pressures in recent human evolution. African populations have been and continue to be at risk for malarial infections. However, few studies have re-sequenced malaria susceptibility loci across geographically and genetically diverse groups in Africa. We examined nucleotide diversity at Intercellular adhesion molecule-1 (ICAM-1), a malaria susceptibility candidate locus, in a number of human populations with a specific focus on diverse African ethnic groups. We used tests of neutrality to assess whether natural selection has impacted this locus and tested whether SNP variation at ICAM-1 is correlated with malaria endemicity. We observe differing patterns of nucleotide and haplotype variation in global populations and higher levels of diversity in Africa. Although we do not observe a deviation from neutrality based on the allele frequency distribution, we do observe several alleles at ICAM-1, including the ICAM-1 (Kilifi) allele, that are correlated with malaria endemicity. We show that the ICAM-1 (Kilifi) allele, which is common in Africa and Asia, exists on distinct haplotype backgrounds and is likely to have arisen more recently in Asia. Our results suggest that correlation analyses of allele frequencies and malaria endemicity may be useful for identifying candidate functional variants that play a role in malaria resistance and susceptibility.

Fig. 1

Representation of the ICAM-1 locus spanning ~15 kb. Exons are shown in blue boxes. The 5′ and 3′ UTRs are shown in maroon boxes. ICAM-1 was sequenced in three regions (I–III). The green lines represent silent (synonymous or non-coding) SNPs identified in this study, and the red lines represent non-synonymous SNPs identified in this study. The red asterisk (*) is placed over rs5491/ICAM-1^Kilifi

Fig. 2

Pairwise linkage disequilibrium (LD) in African (a), Asian (b), and European (c) populations. LD was measured using r². Gray shading represents values of 0 < r² < 1; r² = 0 is shown in white; r² = 1 is shown in black. The number in each square is the observed r²value (%). When r² = 1 no value is shown

Fig. 3

Median-joining haplotype network consisting of SNPs with a frequency ≥2 % in each continental region spanning the entire region that was sequenced. Haplotypes that were observed at least two times are shown. The proportion of each haplotype that was observed in a major geographic region is shown in orange (West Africa), gray (East Africa), blue (Europe), and green (Asia). Note that the African ICAM-1^Kilifi haplotypes cluster together in one part of the network (indicated by red circle). The arrows indicate the Asian specific ICAM-1^Kilifi haplotypes. Median vectors are indicated by red circles. This figure is not drawn to scale. The number on each branch indicates the mutation that separates each haplotype and corresponds to the numbers presented in Supplemental Table 2

Fig. 4

Frequency distribution of correlation coefficients (r²) calculated between a randomly chosen sample of genic (a) and intergenic (b) SNPs across the genome and malaria endemicity (see Supplemental Table 1 for population malaria endemicity values). Dashed line represents the top 5 % of each distribution. The four SNPs in the tails of both distributions include rs5491 (ICAM-1^Kilifi), the two SNPs that are linked to it (rs5030351 and rs5490), and rs5494