Introgression of Neandertal- and Denisovan-like Haplotypes Contributes to Adaptive Variation in Human Toll-like Receptors

Abstract

Pathogens and the diseases they cause have been among the most important selective forces experienced by humans during their evolutionary history. Although adaptive alleles generally arise by mutation, introgression can also be a valuable source of beneficial alleles. Archaic humans, who lived in Europe and Western Asia for more than 200,000 years, were probably well adapted to this environment and its local pathogens. It is therefore conceivable that modern humans entering Europe and Western Asia who admixed with them obtained a substantial immune advantage from the introgression of archaic alleles. Here we document a cluster of three Toll-like receptors (TLR6-TLR1-TLR10) in modern humans that carries three distinct archaic haplotypes, indicating repeated introgression from archaic humans. Two of these haplotypes are most similar to the Neandertal genome, and the third haplotype is most similar to the Denisovan genome. The Toll-like receptors are key components of innate immunity and provide an important first line of immune defense against bacteria, fungi, and parasites. The unusually high allele frequencies and unexpected levels of population differentiation indicate that there has been local positive selection on multiple haplotypes at this locus. We show that the introgressed alleles have clear functional effects in modern humans; archaic-like alleles underlie differences in the expression of the TLR genes and are associated with increased [corrected] microbial resistance and increased allergic disease in large cohorts. This provides strong evidence for recurrent adaptive introgression at the TLR6-TLR1-TLR10 locus, resulting in differences in disease phenotypes in modern humans.

Figure 1

The Introgressed Region Encompassing the Genes TLR10, TLR1, and TLR6 on Chromosome 4 The genetic length of the region (chr4: 38,760,338–38,905,731; hg19) based on three recombination maps ranges from 0.04 cM to 0.1 cM (HapMap, 0.1 cM; deCode, 0.04 cM; and Hinch African American map, 0.08 cM). (i) The gene structures for TLR10, TLR1, and TLR6 are displayed. (ii) Predicted Neandertal haplotypes from the Neandertal ancestry map by Vernot et al. for Asians (green) and Europeans (blue). (iii) Neandertal introgression posterior probabilities across polymorphic positions for Asians (blue solid lines) and Europeans (green solid lines) from the Neandertal ancestry map by Sankararaman et al. The chromosomal average introgression posterior probabilities are given as green and blue dashed lines. (iv) Sharing of the Neandertal allele across archaic-like SNPs within seven core haplotypes: the Neandertal and Denisovan sequences and the seven modern human core haplotypes. The frequencies of Neandertal and Yoruba alleles for the archaic-like SNPs within each core haplotype are colored in red and black, respectively. (v) Significantly associated SNPs with Helicobacter seroprevalence or allergic disease overlapping archaic-like SNPs are displayed.,

Figure 2

Divergence of the Identified Core Haplotypes at TLR10, TLR1, and TLR6 (A) Neighbor-joining tree on the inferred sequences of seven modern human core haplotypes (III–IX), and the Neandertal (I), Denisovan (II), orangutan, and chimpanzee genome sequences. Bootstrap values (1,000 replicates) for the topology are provided in blue squares at each node. The pie charts show the frequency in the four continental population groups of each modern human core haplotype. Next to each pie chart, the frequency of haplotypes among 1000 Genomes individuals assigned to the corresponding core haplotype is displayed. For core haplotype-defining positions in each branch, see Table S2. (B) Pair-wise average nucleotide distance between haplotypes in core haplotypes.

Figure 3

Genotype-Dependent Expression of the TLR Genes The normalized expression in EBV-transformed lymphocytes (individuals in black cubes and entire distribution shown as boxplot; y axis) across GTEx individuals with different genotypes at shared archaic-like SNP positions (x axis: NT, Neandertal allele; AF, African allele) for each TLR gene. Differential expression p values between genotypes are displayed in the upper part of each plot. Median and quartiles are shown and whiskers show the range of expression values.

Figure 4

Clustering of the Seven Core Haplotypes Based on Allele Sharing with GWAS SNPs Principal-component analysis based on the sequence distance between modern human core haplotypes III–IX for the GWAS SNPs reported to be significantly associated with allergic disease (A) and Helicobacter pylori seroprevalence (B).

Figure 5

Geographic Distribution of the Neandertal-like TLR Haplotypes World map showing the frequencies of Neandertal-like core haplotypes in the 1000 Genomes dataset (A) and the Simons Genome Diversity Panel (B). In (B), the size of each pie is proportional to the number of individuals within a population. Core haplotypes (III, orange; IV, green; non-archaic core haplotypes V, VI, VIII, IX, blue) are colored.