Evolution of the Bovine TLR Gene Family and Member Associations with Mycobacterium avium Subspecies paratuberculosis Infection

Fisher, Colleen A.; Bhattarai, Eric K.; Osterstock, Jason B.; Dowd, Scot E.; Seabury, Paul M.; Vikram, Meenu; Whitlock, Robert H.; Schukken, Ynte H.; Schnabel, Robert D.; Taylor, Jeremy F.; Womack, James E.; Seabury, Christopher M.

Across all adjacent variable sites within the bovine TLR gene family, we observed higher levels of LD (r 2 ) in B. t. taurus cattle (0.32) than in the combined sample (0.26) of Bos t. taurus, Bos t. indicus, and composite breeds ( Table 2 ). This is generally consistent with previous studies of bovine subspecific divergence, haplotype structure, and LD across short to moderate physical distances [3, 47] , including our previous study on bovine TLR haplotype structure [30] . However, in this study intragenic estimates of r 2 increased for several loci upon pooling (all cattle), including TLR4, TLR8, and TLR10, which was not predicted given previously reported trends in LD [3, 30, 47] . We previously found that r 2 values were enhanced after pooling only for TLR7 and TLR8 [30] . This result indicates that phase-relationships have been preserved across bovine subspecies and specialized breeds for these loci, perhaps due to selection (Table 4) , and is only apparent at high genotyping densities. Moreover, this observation may represent a signature of selection on some individual variable sites, with detectable levels of intragenic selection only becoming apparent (Table 4 ) with increasing numbers of variable sites subject to selection, and/or uniformly higher selection coefficients. For all genes except TLR2 (Network 1 only), TLR3 (Network 1 only), TLR5, TLR8, and TLR9, one or two predominant haplotypes were predicted for the majority of the cattle investigated (Figures 2,3 ,4, Figure S1 ; Table S4 ). Moreover, significantly positive values for Tajima's D were detected for genomic regions encoding TLR3 and TLR8 (Table 4) despite correction for multiple testing, and for TLR3, the addition of best haplotype pairs for sires with phase probabilities,0.90 produced very similar test statistics (D) for B. t. taurus cattle, indicating that D is not falsely inflated by the absence of rare alleles within the sires that could not be stringently phased. Additionally, a regression based test also demonstrated that TLR3 and TLR8 possess significantly more diversity than do all other TLR loci (P#0.05; Figure 5 ). Significantly positive values for Tajima's D are often interpreted as evidence for a recent population bottleneck, or for some form of balancing selection [48] [49] [50] , with D being the most powerful test in its class [51] , but may also indicate violations of the mutation-drift equilibrium assumption or random sample requirement. Worthy of discussion is the fact that variation within TLR3 displayed the second highest average r 2 values between adjacent variable sites ( Table 2) , which in conjunction with a large, significantly positive D statistic for taurine cattle (Table 4) suggests that this gene is under selection. However, unlike TLR8, high r 2 ($0.50 for 10/13 SNPs in TLR8) did not persist across the majority of all adjacent variable sites in TLR3, and therefore, it is relatively unsurprising that our analysis of TLR3 revealed no evidence for a deficiency of total discrete haplotypes in B. t. taurus cattle (i.e., F S was not significant).