Fig. 2

Adaptive protein-coding substitutions clustered near the RNA binding loop in the C-terminal RNA recognition domain (RD) along the lineages leading from ancRLR to ancLGP2 (see Fig. 1). We show the consensus sequence alignment of the three human RLR RDs and the ancestral RLRs resurrected in this study, with residues colored by biochemical classification and sequence conservation. Stars above the alignment indicate significant support for protein-coding adaptation at specific residues along the branches separating ancLGP2 from ancMDA5/LGP2b (red), ancMDA5/LGP2b from ancMDA5/LGP2a (green) and ancMDA5/LGP2a from ancRLR (blue), respectively (see Fig. 1). Adaptive substitutions were inferred by Bayes-empirical-Bayes posterior probability >0.95 using the branch-sites test for positive selection (see Methods). The location of the RNA binding loop is indicated, and approximate secondary structural elements are shown below the alignment