Impact and Evolutionary Determinants of Neanderthal Introgression on Transcriptional and Post-Transcriptional Regulation

Abstract

Archaic admixture is increasingly recognized as an important source of diversity in modern humans, and Neanderthal haplotypes cover 1%-3% of the genome of present-day Eurasians. Recent work has shown that archaic introgression has contributed to human phenotypic diversity, mostly through the regulation of gene expression. Yet the mechanisms through which archaic variants alter gene expression and the forces driving the introgression landscape at regulatory regions remain elusive. Here, we explored the impact of archaic introgression on transcriptional and post-transcriptional regulation. We focused on promoters and enhancers across 127 different tissues as well as on microRNA (miRNA)-mediated regulation. Although miRNAs themselves harbor few archaic variants, we found that some of these variants may have a strong impact on miRNA-mediated gene regulation. Enhancers were by far the regulatory elements most affected by archaic introgression: up to one-third of the tissues we tested presented significant enrichments. Specifically, we found strong enrichments of archaic variants in adipose-related tissues and primary T cells, even after accounting for various genomic and evolutionary confounders such as recombination rate and background selection. Interestingly, we identified signatures of adaptive introgression at enhancers of some key regulators of adipogenesis, raising the interesting hypothesis of a possible adaptation of early Eurasians to colder climates. Collectively, this study sheds new light on the mechanisms through which archaic admixture has impacted gene regulation in Eurasians and, more generally, increases our understanding of the contribution of Neanderthals to the regulation of acquired immunity and adipose homeostasis in modern humans.

Keywords: Neanderthal; T-cells; adaptation; adipose tissue; archaic introgression; enhancers; gene regulation; immunity; miRNAs; promoters.

Figure 1

Enrichment of Neanderthal Variants in Regulatory Regions (A) Odds ratio depicting the excess or depletion of Neanderthal variants in coding regions and regulatory elements (promoters, enhancers, and miRNA binding sites) compared to the remainder of the genome. Enrichments are shown for three bins of minor-allele frequencies (MAFs), together with 95% bootstrap confidence intervals: ^∗p value < 0.05, ^∗∗p value < 0.01, ^∗∗∗p value < 0.001. (B) Relative density of aSNPs in promoters, enhancers, and miRNA binding sites in different MAF bins, with 95% bootstrap confidence intervals. (C and D) Comparison of density of conserved sites (GerpRS > 2) and mean B statistic of promoters, enhancers, and miRNA binding sites. (E) Percentage of alleles that are fixed in Neanderthal, absent from the African Yoruba from Nigeria (YRI), and introgressed at a MAF > 5% in Eurasians. For each type of region, box plots show the variability of the estimates based on 1,000 bootstrap resamples of 100 kb genomic windows. The dashed vertical line indicates the genome-wide average. (F) Total length of promoters, enhancers, and miRNA binding sites.

Figure 2

Effects of Archaic Introgression on miRNA-Mediated Regulation (A) Representation of the archaic (red) and modern (green) human alleles for the six miRNAs presenting a Neanderthal-introgressed variant in their mature sequence. The seed region of the miRNAs is shaded in gray. (B) Total number of genes bound by the archaic and/or modern human allele of each of the six miRNAs harboring a Neanderthal variant in their mature sequence. (C) Relationship between the number of targets of each miRNA and the number of common aSNPs in the corresponding miRNA binding sites. (D) Introgression of aSNPs altering the miRNABS at the ONECUT2 locus (MIM: 604894). Gene structure is shown in the upper panel, and miRNA binding sites that are altered by archaic introgression are highlighted in green. The middle panel represents the density of conserved sites (GerpRS > 2) in 1,000 bp windows, and the bottom panel represents the repartition and frequency of archaic alleles at the locus (blue for CEU, red for CHB). aSNPs that overlap miRNABS are represented with a darker shade, and aSNPs that disrupt a conserved site are marked with stars.

Figure 3

Effects of Archaic Introgression at Enhancers (A) Volcano plot illustrating the enrichment of common aSNPs in the enhancers of 127 different tissues from the Epigenomic Roadmap Consortium. Tissues with FDR < 5% (triangles) are significantly enriched. (B) Enrichments of common aSNPs in the enhancers of different immune tissues. Vertical bars indicate 95% confidence intervals computed by bootstrap analysis. (C) Enrichment of common aSNPs in the enhancers that are active in more than half of the investigated T cell subtypes (dark red, referred to as “core T cells”) and in enhancers that are active in each T cell subtype and are not part of core T cell enhancers (light red, referred to as “cell-type-specific enhancers”). Vertical bars indicate 95% confidence intervals computed by bootstrap. (B and C) Note that CD4⁺ T cells are separated on the basis of CD25 so that T_reg (CD25⁺), T_EM (CD25^Iow), and T_helper (CD25⁻) are distinguished from one another.

Figure 4

Factors Shaping Human-Neanderthal Divergence and Archaic Introgression at Enhancers (A) Comparison of the relative density of fixed human-Neanderthal differences and rate of introgression in the enhancers of the 127 tissues studied. The size of the circles is proportional to the relative density of common aSNPs in the enhancers of the corresponding tissue; a black circle is added when the relative density of common aSNPs is significantly higher in these enhancers (FDR < 5%) than in the rest of the genome. The density of each tissue category along the two axes is also presented. (B and C) Genome-wide correlations, using 100 kb windows, between either the rate of Neanderthal introgression (B) or the relative density of fixed human-Neanderthal differences (C) and neutral and selective forces. ^∗p value < 10⁻², ^∗∗p value < 10⁻¹⁰, and ^∗∗∗p value < 10⁻²⁰. For each correlation, horizontal lines indicate 95% confidence interval. (D) Observed values of rate of introgression and relative density of fixed differences and common aSNPs at the enhancers of core T cells, AdMSCs, and prefrontal cortex, with respect to expectations based on 100 kb windows matched for length of enhancers alone or for length of enhancers, percentage of GC, recombination rate, density of conserved sites, and mean B statistic of their enhancers (see Supplemental Data). n.s. = not significant; ^∗p value < 0.05, ^∗∗p value < 0.01, and ^∗∗∗p value < 10⁻³. Errors bars indicate 95% confidence intervals of the expected values obtained by resampling.

Figure 5

Manhattan Plots of Genes Interacting with Enhancers That Contain Archaic Variants (A and B) Genome-wide distribution of MAFs in CEU or CHB at aSNPs that overlap enhancers active in T Cells (core T cell enhancers). For each window of 1 Mb along the genome, only the aSNP with the highest MAF is shown. Point sizes reflect FPKM of the most expressed genes (max FPKM across T lymphocytes from Blueprint database⁵⁸) among genes interacting with the enhancer in T cells. (C and D) Similar plots for enhancers active in AdMSCs. Point sizes reflect the FPKM of the most expressed gene (max FPKM in GTEx tissues Adipose—Subcutaneous and Adipose—Visceral [Omentum]⁵⁹) among genes interacting with the enhancer in adipose tissue.