Selective sweep on human amylase genes postdates the split with Neanderthals

Abstract

Humans have more copies of amylase genes than other primates. It is still poorly understood, however, when the copy number expansion occurred and whether its spread was enhanced by selection. Here we assess amylase copy numbers in a global sample of 480 high coverage genomes and find that regions flanking the amylase locus show notable depression of genetic diversity both in African and non-African populations. Analysis of genetic variation in these regions supports the model of an early selective sweep in the human lineage after the split of humans from Neanderthals which led to the fixation of multiple copies of AMY1 in place of a single copy. We find evidence of multiple secondary losses of copy number with the highest frequency (52%) of a deletion of AMY2A and associated low copy number of AMY1 in Northeast Siberian populations whose diet has been low in starch content.

Figure 1. Distribution of estimates of maximum pairwise F_st per 200 kb non-overlapping segment among 12 human population groups.

‘AMY locus’ refers to data from three neighbouring 200 kb segments from the Chr1:103,800,000-104,400,000 region. The distribution of the maximum F_ST scores of the 200 kb regions is shown by bins of 0.05. The only 5% significant (red dotted line) F_ST estimate for the AMY locus comes from the African and Northeast Siberian comparison at the Chr1:104,000,000-104,200,000 segment.

Figure 2. Distribution of genetic diversity on chromosome 1 in human populations.

Presented on the y axis is the sum of derived allele frequency over all polymorphic loci in 1000 Genomes African data, estimated by 50,000 bp non-overlapping segments of chromosome 1, relative to the divergence of the human reference sequence from the ancestral sequence (determined by the 6 primate sequence consensus). Red dotted line indicates the 1% cut-off considering the empirical genome-wide distribution. Only those 50,000 bp segments which had >90% sites covered in human, Altai Neanderthal and Denisovan data were considered.

Figure 3. Phylogenetic tree and Bayesian Skyline Plot based on the analyses of a ~66 kb region downstream of the AMY locus with BEAST.

The analyses were restricted to bi-allelic SNP variants from a ~66 kb long high LD region (Chr1:104,303,310-104,369,301) which is downstream of (centromeric to) AMY1C gene (Figure S1). Coalescent time estimates are shown near branching points and assume mutation rate of 5.5 × 10⁻¹⁰ per bp per year. BSP–Bayesian Skyline Plot; thick brown line shows the median estimates of human effective population size (N_e) x generation time (y-axis) over time (x-axis) as estimated from the ~66 kb region data in the global sample of 480 individuals. The grey shaded area around the brown line shows the 95% higher posterior density intervals of the BSP estimates. A dotted blue arrow highlights the branch of the human phylogeny that has been the likely source of an introgression of a chunk of ~16 kb (Chr1:104,350,432-104,366,871) into the genome of the Altai Neanderthal.

Figure 4. Distributions of AMY1 and AMY2A copy number across the major continental groups.