A Selective Sweep on a Deleterious Mutation in CPT1A in Arctic Populations

Abstract

Arctic populations live in an environment characterized by extreme cold and the absence of plant foods for much of the year and are likely to have undergone genetic adaptations to these environmental conditions in the time they have been living there. Genome-wide selection scans based on genotype data from native Siberians have previously highlighted a 3 Mb chromosome 11 region containing 79 protein-coding genes as the strongest candidates for positive selection in Northeast Siberians. However, it was not possible to determine which of the genes might be driving the selection signal. Here, using whole-genome high-coverage sequence data, we identified the most likely causative variant as a nonsynonymous G>A transition (rs80356779; c.1436C>T [p.Pro479Leu] on the reverse strand) in CPT1A, a key regulator of mitochondrial long-chain fatty-acid oxidation. Remarkably, the derived allele is associated with hypoketotic hypoglycemia and high infant mortality yet occurs at high frequency in Canadian and Greenland Inuits and was also found at 68% frequency in our Northeast Siberian sample. We provide evidence of one of the strongest selective sweeps reported in humans; this sweep has driven this variant to high frequency in circum-Arctic populations within the last 6-23 ka despite associated deleterious consequences, possibly as a result of the selective advantage it originally provided to either a high-fat diet or a cold environment.

Figure 1

Localization of Positive-Selection Signals within a 3 Mb Region on Chromosome 11 (A) A high concentration of significant iHS (top) and Tajima’s D (bottom) signals was found in a 3 Mb region (chr11: 66–69 Mb) in Northeast Siberians. The results are shown in the context of the East Asian and European control populations. Pale gray lines highlight the boundaries of each 200 kb window in the region. The horizontal black dotted line marks the threshold of 1% significance. The overlapping 400 kb region (chr11: 68.2–68.6 Mb) of significant results from both tests is highlighted by a bold black rectangle. (B) The genome-wide selection scans show that the window containing CPT1A (chr11: 68.4–68.6 Mb), highlighted by the red dot in the three plots, is significant in the Northeast Siberian populations, but not in the control populations (Europeans and East Asians). The black dots are the significant data points in both iHS and Tajima’s D tests. Dotted lines show the significance thresholds of the respective tests.

Figure 2

Geographic Distribution and Network of the CPT1A c.1436C>T Mutation and Its Associated Haplotype (A) The c.1436C>T derived allele is defined by the rs80356779 G>A mutation and occurs with a frequency of 0.9, 0.875, and 0.5625 in Chukchi, Eskimo, and Koryaks, respectively, but is absent elsewhere (1000 Genomes Project, CG public data, and Personal Genome Project). We used three SNPs (rs10896365 A>G, rs80356779 G>A, and rs3794020 T>C) to define the haplotype GAT (red, see B). The haplotype ancestral to the c.1436C>T mutation (GGT) is shown in black. The white node represents all other haplotypes. Grey shading that encompasses both the white and black nodes refers to subjects for whom information was only available for rs80356779. The map shows the geographic distribution of these haplotypes. Haplotype data were drawn from modern-DNA (blue font) and ancient-DNA (red font) sources, including Chukchi (CEK), Eskimo, and Koryaks (present study); 1000 Genomes Project; Nunavut Inuit (NUN); Mal’ta (M’TA); Clovis (ANZ); Aleutian Islander (ALE); Early, Middle, and Late Dorset (DOR) (Table S10); Saqqaq (SQQ); and Greenland Inuit (GIN). (B) The Haplotype Median Joining Network was constructed from sequences of the Northeast Siberians and control populations (Europeans and East Asians) on the basis of 848 SNPs present in the 58,084 bp region (Table S6) surrounding c.1436C>T with the use of the Network 4.612 package. The circles are proportional to the frequency of the shared haplotype. On the basis of this network analysis, we chose two SNPs (rs10896365 A>G and rs3794020 T>C) that defined the haplotype (black nodes) ancestral to the c.1436C>T mutation (red nodes). For visualization, some branches in the figure are shortened (marked with break). The dashed lines represent likely recombination events.