Evolutionary dynamics of human Toll-like receptors and their different contributions to host defense

Abstract

Infectious diseases have been paramount among the threats to health and survival throughout human evolutionary history. Natural selection is therefore expected to act strongly on host defense genes, particularly on innate immunity genes whose products mediate the direct interaction between the host and the microbial environment. In insects and mammals, the Toll-like receptors (TLRs) appear to play a major role in initiating innate immune responses against microbes. In humans, however, it has been speculated that the set of TLRs could be redundant for protective immunity. We investigated how natural selection has acted upon human TLRs, as an approach to assess their level of biological redundancy. We sequenced the ten human TLRs in a panel of 158 individuals from various populations worldwide and found that the intracellular TLRs -- activated by nucleic acids and particularly specialized in viral recognition -- have evolved under strong purifying selection, indicating their essential non-redundant role in host survival. Conversely, the selective constraints on the TLRs expressed on the cell surface -- activated by compounds other than nucleic acids -- have been much more relaxed, with higher rates of damaging nonsynonymous and stop mutations tolerated, suggesting their higher redundancy. Finally, we tested whether TLRs have experienced spatially-varying selection in human populations and found that the region encompassing TLR10-TLR1-TLR6 has been the target of recent positive selection among non-Africans. Our findings indicate that the different TLRs differ in their immunological redundancy, reflecting their distinct contributions to host defense. The insights gained in this study foster new hypotheses to be tested in clinical and epidemiological genetics of infectious disease.

Figure 1. Global genetic diversity of the TLR family in human populations.

Nucleotide diversity levels for the individual TLR genes in populations representing major ethnic groups. The expected diversity corresponds to the mean diversity levels observed for the 20 autosomal non-coding regions (“neutral” regions) in each geographical area.

Figure 2. Estimation of the intensity of natural selection acting on individual TLR genes.

(A) Strength of purifying selection acting on individual TLR genes, as measured by estimated ω values. Bars indicate 95% CIs and red circles indicate genes with ω estimates significantly lower than 1. (B) Strength of negative selection acting on individual TLR genes, as measured by the population selection coefficient γ. Bars indicate 95% CIs, and red circles indicate genes with γ estimates significantly lower than 0. (C) Joint estimates of the intensity of purifying (ω) and negative (γ) selection between intracellular nucleic acid sensors and cell-surface expressed TLRs.

Figure 3. Functional diversity is unevenly distributed between the two groups of TLRs.

(A) Distribution of the different classes of exonic polymorphisms (silent, stop, and the nonsynonymous variants considered to be benign, possibly damaging or probably damaging) between intracellular nucleic acid sensors and cell-surface expressed TLRs. (B) Proportion of individuals in the general population presenting a probably damaging or stop mutation in at least one TLR belonging from either of the two groups of TLRs. (C) Proportion of individuals in the general population presenting at least a probably damaging or a stop mutation for each individual TLR. The protein domains targeted by each of the nonsynonymous mutations identified are shown in Figure S9 for TLRs sensing nucleic acids, and in Figure S10 for the remaining TLRs.

Figure 4. Some TLR members present extreme levels of population differentiation.

(A) F _ST comparison for Africans vs Europeans (B) F _ST comparison for Africans vs East-Asians and (C) F _ST comparison for Europeans vs East-Asians. To account for possible differences in the allele frequency spectrum between our data and the HapMap genome-wide distribution , we compared F _ST as a function of heterozygosity. Green dots correspond to silent polymorphisms and red dots correspond to nonsynonymous mutations. The dashed lines represent the 95^th and 99^th percentiles of the HapMap genome-wide distribution (represented by the density area in blue). The significant values observed for TLR7 and TLR8 should be taken cautiously because these genes are located on the X chromosome and therefore, higher genetic drift may have exacerbated the levels of population differentiation observed.

Figure 5. Signatures of positive selection targeting the TLR10-TLR1-TLR6 gene cluster.

(A) Power of the DIND test with respect to other statistics as a function of the frequency of the selected allele when the selection coefficient (S = 2Ns) is set to be 100. DIND test in (B) Africans, (C) Europeans and (D) East-Asians. P values were obtained by comparing the iπ_A/iπ_D values for the TLRs against the expected iπ_A/iπ_D values obtained from 10,000 simulations considering both a model of constant population size and the demographic model of Voight et al.'s (Materials and Methods).

Figure 6. Functional analyses of the positively selected haplotype H34 in Europeans.

(A) Level of expression of TLR1 and TLR6 variants. HEK 293T cells were transiently transfected with 100 ng of the indicated construct and whole cell lysates were analyzed by immunoblot with anti-HA tag antibody. (B) The TLR1 602S variant results in a diminished NF-kB signalling. HEK 293T cells were transiently co-transfected with an NF-κB reporter plasmid, a Renilla luciferase plasmid for transfection control and with TLR2 alone or in combination with the indicated TLR1 and TLR6 variants. One day after transfection, cells were stimulated with the indicated agonist at 10 ng/ml for 4 h and luciferase activities were measured. The data shown are the mean±SD of three replicates of a representative of three independent experiments, expressed as the percent relative firefly luciferase activity (RLU) (normalized to Renilla luciferase activity). * p<0.001 (as determined by Student's t test) when comparing the different TLR1 allelic variants with respect to the ancestral TLR1 248S-602I form, and when comparing the TLR6 variants with the ancestral TLR6 249P form.