The sequence of our reconstructed Anc is

The sequence of our reconstructed βAnc is inferred based on probabilities, and the estimated probability of inferring the actual ancestral sequence is the product of the posterior probabilities for each site within the reconstructed sequence. Although the posterior probabilities (PP) for the majority of sites within βAnc are high (PP > 95% for more than 70% of sites; inset in Figure S1), the overall probability that βAnc represents the actual ancestral sequence is essentially zero. As pointed out by Wilson et al. (2015), these statistical considerations reflect the fact that, like modern proteins, ancestral proteins existed in large populations of organisms, and comprised a polymorphic ensemble of similar proteins that changed over time. Thus, βAnc can be viewed as a representative member of a group of ancestral β subunits that existed ∼480 million years ago, and likely shared similar amino IWR-1-endo mg sequences and functional properties. In any case, qualitatively the functional properties of reconstructed ancestral proteins appear to be robust to these statistical considerations (Eick et al., 2017). Reconstruction of ancestral sequences is based on a specific phylogeny, and inherent in any phylogeny is a degree of uncertainty. This uncertainty can influence the confidence in the reconstructed ancestral sequences, as well as the evolutionary conclusions that can be drawn from them. Our phylogeny is inferred from a single gene/protein, the AChR β subunit, and it is common for the topology of species trees based on a single gene to differ from those assembled from multiple genes (Rokas et al., 2003). One possible mechanism for this discrepancy is that in different species, orthologous genes (or proteins) may evolve under different selective pressures. This is particularly relevant for our AChR β subunit tree, and may explain the apparent discrepancy in the placement of Torpediniformes. Muscle-type AChRs in Torpediniformes have been repurposed within specialized electric organs, and have likely evolved under different selection criteria than typical muscle-type AChRs in other animals. An important finding of the present work is that our reconstructed βAnc is a viable protein, capable of functioning in hybrid human/ancestral AChRs, demonstrating that even in the face of the above statistical considerations and phylogenetic uncertainty, the ancestral sequence reconstruction approach provides a rational map (Lipovsek et al., 2014) for exploring AChR sequence space. For example, by studying ancestors along the path between βAnc and the human β subunit, we can identify successive ancestors that flank a functional change, such as the observed differences in single-channel conductance or kinetics. Comparing the sequences of these ancestors will allow us to narrow in on the amino acid substitutions that contribute to each functional difference. Clearly such a strategy will be limited by phylogenetic coverage, which affects the number of substitutions between successive ancestors along any evolutionary path. As a result, final identification of functionally relevant residues will require verification through site-directed mutagenesis. This approach has successfully identified residues involved in a variety of protein functions, including kinase drug sensitivity (Wilson et al., 2015), steroid hormone receptor specificity (Ortlund et al., 2007), and fluorescence in coral proteins (Field and Matz, 2010). Assuming our reconstructed βAnc faithfully represents β subunits that existed ∼480 million years ago, hybrid AChRs incorporating βAnc are essentially human AChRs in which the β subunit has been “frozen in time” and trapped in stasis for ∼480 million years. Thus, in these hybrid AChRs the functional differences between wild-type and βAnc-containing AChRs is a direct consequence of preventing ∼480 million years of AChR β subunit co-evolution. Our finding that βAnc-containing AChRs have a loss-of-function phenotype, with reduced open probability, is consistent with the hypothesis that co-evolution of human AChR subunits has occurred, and that this co-evolution is required in order for the subunits to maintain their ability to cooperate with one another. Future reconstruction of ancestral AChR α, δ, and ɛ subunits will allow for unprecedented insight into the evolution of AChR subunit cooperativity, as well as the amino acid origins of this enigmatic property.