Archaic Hominin Admixture Facilitated Adaptation to Out-of-Africa Environments

Abstract

As modern humans dispersed from Africa throughout the world, they encountered and interbred with archaic hominins, including Neanderthals and Denisovans [1, 2]. Although genome-scale maps of introgressed sequences have been constructed [3-6], considerable gaps in knowledge remain about the functional, phenotypic, and evolutionary significance of archaic hominin DNA that persists in present-day individuals. Here, we describe a comprehensive set of analyses that identified 126 high-frequency archaic haplotypes as putative targets of adaptive introgression in geographically diverse populations. These loci are enriched for immune-related genes (such as OAS1/2/3, TLR1/6/10, and TNFAIP3) and also encompass genes (including OCA2 and BNC2) that influence skin pigmentation phenotypes. Furthermore, we leveraged existing and novel large-scale gene expression datasets to show many positively selected archaic haplotypes act as expression quantitative trait loci (eQTLs), suggesting that modulation of transcript abundance was a common mechanism facilitating adaptive introgression. Our results demonstrate that hybridization between modern and archaic hominins provided an important reservoir of advantageous alleles that enabled adaptation to out-of-Africa environments.

Keywords: Denisovan; Hybridization; Neandertal; adaptive; admixture; human evolution; introgression.

Figure 1.. Genomic distribution and characteristics of high-frequency archaic haplotypes in geographically diverse populations.

A) Each dot represents the frequency and genomic position of an introgressed archaic haplotype. Loci above the grey lines correspond to putative targets of adaptive introgression (outliers in the ≥99^th percentile; FDR ≤ 50%). Outlier loci that had a significant phenotypic association (GWAS or eQTL) are highlighted in red. B) Relationship between the archaic haplotype frequency threshold for identifying adaptively introgressed loci and FDR for each population. Shaded regions delimit 95% confidence intervals. C) Venn diagram showing overlap of high frequency archaic haplotypes between populations. The inset pie chart shows how many of the Melanesian high frequency haplotypes are Neandertal, Denisovan, or Ambiguous. See also Figure S1 and Tables S1 and S2.

Figure 2.. Adaptive introgression of archaic sequence at the TNFAIP3 and OCA2 loci.

A. Schematic of the TNFAIP3 region is shown with vertical bars indicating introgressed SNPs. Black and red denote matches and mismatches to the Denisovan reference genome, respectively. Two missense SNPs are highlighted with stars. The track along the bottom depicts H3K27ac signal from seven ENCODE cell types [41]. B. Distributions of absolute genetic distance to the Denisovan reference genome for all haplotypes within the four populations studied, as well as Africans. The grey box indicates the portion of the distribution comprised of introgressed sequence. C. Schematic of the OCA2/HERC2 region. The purple box indicates the introgressed region, and the black boxes indicate previously identified positively selected and pigment associated regions in East Asians and Europeans. Below, a zoomed view of the introgressed region is shown with vertical bars indicating introgressed variants. Variants that overlap melanocyte regulatory elements are shown in red and GWAS study variants are indicated with a yellow star. D. Distributions of absolute genetic distance to Neandertal for all haplotypes within the four populations studied as well as Africans. The grey box indicates the portion of the distribution comprised of introgressed sequence. See also Figure S2.

Figure 3.. Introgression at the OAS locus and its impact on gene expression.

A schematic of the OAS1/2/3 region is shown with vertical bars indicate introgressed variants. Black and red denote matches and mismatches to the Neandertal reference genome, respectively. Below, gene expression for OAS1, OAS2, and OAS3 is shown stratified by the number of Neandertal haplotypes an individual has in fibroblasts and LCLs. See also Figure S3 and Table S3.

Figure 4.. Patterns of eQTL effects for the introgressed TLR1/6/10 haplotype across multiple cell types.

A. Schematic of the TLR1/6/10 region is shown with vertical bars indicating introgressed variants. Black and red denote matches and mismatches to the Neandertal reference genome, respectively. B. Gene expression for TLR6 in three different cell types is shown stratified by the number of Neandertal alleles each sample has. P values are indicated in each plot. The table below corresponds to each cell type, and check marks indicate characteristics of each cell type. C. Gene expression for TLR6 is shown for whole blood samples before and after stimulation with LPS. Expression is stratified by the number of Neandertal alleles each sample has. See also Figure S4.