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.

Notably, only TLR8 displayed a significant, positive value for Fu's F S , indicating a lower than expected number of haplotypes, as would be predicted given a recent population bottleneck or strong balancing selection. However, the high r 2 that persists across nearly all adjacent variable sites strongly implies selection ( Table 2) . While previous studies have suggested that population bottlenecks may have occurred at the time of domestication and breed formation for modern cattle [3, 47] , these are expected to drive frequency distribution tests (D, F S ) toward more positive values because of the loss of rare genetic variation at all loci. In particular, the effects of bottlenecks are expected to be uniform and potentially dramatic for proximal, evolutionarily related Xlinked loci (TLR7, TLR8) performing similar functions (6, (11) (12) , especially given smaller effective population size (chromosomal) and female limited recombination. However, TLR7 possesses a fundamentally different frequency distribution trend (D = -0.19828 all cattle; D = -0.17037 B. t. taurus) as compared to TLR8 (TLR7#103 Kb from TLR8; Btau5.2), with no evidence for a significant deviation from a strictly neutral model (Table 4) . A regression based test also provided no evidence for the effects of a population bottleneck or selection operating on variation within TLR7 (P$0.05; see Figure 5 ). Therefore, it seems unlikely that historic bottlenecks are responsible for deviations from neutrality for bovine TLR8, and more likely that balancing selection is operating to preserve a limited number of functionally divergent haplotypes. Interestingly, the haplotypes observed for TLR8 were partitioned into two main functional groups, as classified by our AA modeling (Table 3 ) and median joining haplotype networks (Figure 3) . Specifically, haplotypes that fell into network nodes A, B, and C differed from haplotypes falling into nodes D, E, and F by eight nonsynonymous SNPs encoding AA substitutions (Table S2) , with at least two (S477N; K903T) that were predicted to impact protein Figure 3) . Additionally, the four most common haplotypes (nodes A, B, D, and E) differed only by one synonymous SNP (nodes A vs B; encoding S10S) and one putatively benign or tolerated nonsynonymous SNP (nodes D vs E; encoding S492N; see Table S2 ; Table 3 ). For these reasons, functional studies are now needed to comprehensively assess the dynamic range of ligand-induced TLR8 signaling in domestic cattle.