Recent Selection on a Class I ADH Locus Distinguishes Southwest Asian Populations Including Ashkenazi Jews

Abstract

The derived human alcohol dehydrogenase (ADH)1B*48His allele of the ADH1B Arg48His polymorphism (rs1229984) has been identified as one component of an East Asian specific core haplotype that underwent recent positive selection. Our study has been extended to Southwest Asia and additional markers in East Asia. F_st values (Sewall Wright's fixation index) and long-range haplotype analyses identify a strong signature of selection not only in East Asian but also in Southwest Asian populations. However, except for the ADH2B*48His allele, different core haplotypes occur in Southwest Asia compared to East Asia and the extended haplotypes also differ. Thus, the ADH1B*48His allele, as part of a core haplotype of 10 kb, has undergone recent rapid increases in frequency independently in the two regions after divergence of the respective populations. Emergence of agriculture may be the common factor underlying the evident selection.

Keywords: alcohol dehydrogenase; genetic selection; population genetics.

Figure 1

All 42 populations used in this study. Populations are geographically categorized and colored into eight groups: Africa (BIA, MBU, YOR, IBO, HAS, CGA, MAS, AAM, and ETJ), Southwest Asia (YMJ, DRU, and SAM), Europe (ASH, ADY, CHV, RUA, RUV, FIN, DAN, IRI, and EAM), Siberia (KMZ, KTY, and YAK), Pacific Islands (NAS and MCR), East Asia (CBD, CHS, CHT, HKA, KOR, JPN, AMI, and ATL), North America (NPA, SWA, PMM, and MAY,) and South America (QUE, TIC, SUR, and KAR). The three-character population symbols and corresponding full names are listed in pairs.

Figure 2

The schematic representation of regions of high linkage disequilibrium (LD) across the human alcohol dehydrogenase (ADH) clusters. HAPLOT and the default r² algorithm were used. All 118 genetic markers (including one insert-deletion site treated as a single nucleotide polymorphism (SNP)) are listed at the bottom. The locations of two short tandem repeat polymorphisms (STRP) loci are indicated at the bottom by the black triangles. The arrow above each gene symbol indicates the downstream direction; the region is plotted from centromere to telomere. Generally speaking, populations of the same group have similar LD structure with some variation. Highlighted within the blue box, Southwest Asian populations show LD patterns similar to those of European and African populations.

Figure 3

The heterozygosity (HET) and F_st (Sewall Wright’s fixation index) of each genetic marker. Most markers have moderate heterozygosity with only three below 0.05. The 5-site moving average curve of heterozygosity suggests a level of 0.2 to 0.4. The two highest F_st values marked by the empty diamond symbols are ADH1B Arg48His/rs1229984 and rs3811801. The 5-site moving average curve of F_st values, drawn as a solid line, peaks around those two SNPs. The F_st of ADH1B Arg48His is 3.61 SDs (standard deviation) above the mean, while that of rs3811801 is 3.38 SDs above the mean. Three nearby SNPs, #55 = rs1042026 (F_st = 0.401), #57 = rs2066701 (the RsaI restriction site) (F_st = 0.388), and #58 = rs2075633 (F_st= 0.391), have F_st values > 2.5 SD above the mean.

Figure 4

The core haplotype pattern based on seven SNPs (see Table 1) is shown for 42 populations. Haplotype frequencies are highly variable across all geographic regions with Africans displaying the most variation while Native Americans have the least variation. The haplotype TCGAGGC is observed at very common frequencies in Southwest Asians, Ashkenazi, and Ethiopians. In East Asia and the Pacific region this haplotype has comparable frequencies. Another haplotype, TCGAAGT, is predominant in East Asian populations with frequency estimates ranging from 45 to 70% in Han, Koreans, and Japanese. The haplotype TCGA (from TCGAAGT) has shown a signature of recent positive selection in East Asian populations in earlier studies [13,14,15].

Figure 5

Extended haplotype homozygosity (EHH) and Relative EHH (REHH) plots in Southwest Asian and East Asian populations. The core haplotype position is defined to be zero. The left side of the core is toward downstream direction, while the right side of the core is toward upstream direction. (a) EHH and REHH results for the core haplotype TCGAGGC in Southwest Asian populations. Population abbreviations as well as corresponding sample size and the frequency of the selected core haplotype are listed in the middle. All five populations show that, in the downstream direction, the EHH extends over 250 kb from the core haplotype at a level above 0.6. In the upstream direction, however, the EHH only extends around 80 kb at a level above 0.6. The REHH plot reveals a V shape, which means REHH increases continuously from the core to either direction. This is a typical sign of potential recent selection on the target core. (b) REHH values from Southwest Asia and E Asia against those from simulated haplotypes. On the left are the REHH values sampled from 253 kb downstream of the core against simulated reference points. Clearly, the REHH of ETJ, DRU, and YMJ is well above the 95th percentile, and that of ASH rides on the 95th percentile curve. Only SAM falls between 75th and 95th percentile. However, the REHH values of East Asian populations are mostly below 95th percentile except JPN (above the line) and KOR (on the line). On the right are the REHH values sampled from 149 kb upstream of the core against simulated reference points. All Southwest Asian populations have REHH above 95th percentile. In East Asia, JPN, KOR, CHS, and CBD have REHH above 95th percentile, while HKA and CHT have REHH on the 95th percentile line. However, the REHH of AMI and ATL fall below 75th percentile.

Figure 6

The allele profiles of SNPs in the flanking regions of the core haplotype under selection. Two different colors in each bar represent the relative percentage of two different alleles. Populations of the same geographic location tend to share similar allele profiles, while populations in two different areas apparently differ in allele profiles except occasional similarity.

Figure 7

Schematic of STRP evolution flanking the two haplotypes with evidence of selection. The 25 STRP haplotypes for H5 and H7 in the populations studied are shown; color highlighting indicates their frequency counts. See Table S2 for more details. H5 = TCGAGGC in Figure 4; H7 = TCGAAGT in Figure 4. The core haplotypes correspond to H5 and H7 in Li et al. [41]; the shorter core in this study does not allow H6 to be distinguished. Given the assumed mutation rates at the STRPs, these are consistent with recent selection on the chromosomes and compatible with selection associated with the origins of agriculture.

Figure 8

The REHH of three new confirmatory populations, Ethiopian Jews (ETJ2), Ashkenazi (ASH2), and Palestinian Arabs (PAL). At lower SNP density but approximately similar distance from the core, all populations show REHH increases over distance.