Archives
To reconstruct the amino acid sequence of an ancestral
To reconstruct the amino Sulfo-NHS-LC-Biotin sequence of an ancestral β subunit a molecular phylogeny is required, which is inferred from a β subunit sequence alignment. To build this alignment we searched the NCBI database for proteins similar in sequence to the human muscle-type β subunit precursor protein, which contains a ∼20-amino-acid signal peptide that is eventually cleaved from the mature protein. This search returned 116 sequences that were unambiguous orthologs of the human muscle-type β subunit. These sequences were aligned, and those with contiguous stretches of amino acid insertions or deletions, or those sequences with unknown residues, were removed. The remaining 65 sequences were realigned and a best-fit model of amino acid evolution was identified (Abascal et al., 2005) to facilitate maximum-likelihood phylogenetic analysis (Guindon et al., 2010). The resulting β subunit phylogenetic tree recapitulates many of the accepted relationships between taxa, and serves as a roadmap for reconstructing strategic β subunit ancestors (Figures 1B and S1). Inherent to any molecular phylogeny is a degree of uncertainty, and although our β subunit phylogenetic tree reproduces many of the hypothesized relationships between species, a notable exception is the placement of Torpediniformes within Euteleostomi, a clade whose species all shared a common ancestor roughly 430 million years ago. The accepted phylogeny places Torpediniformes outside of Euteleostomi, indicating that Torpediniformes and species within Euteleostomi diverged from a common ancestor around 480 million years ago (Hedges et al., 2006, Hinchliff et al., 2015). Confidence in maximum likelihood phylogenies can be estimated using a number of statistical tests (Anisimova et al., 2011), and this discrepancy is also reflected in a lower confidence in our tree topology in this region (Figure S1). Although simulations have shown that ancestral sequence reconstructions are robust to phylogenetic uncertainty (Hanson-Smith et al., 2010), we wondered whether uncertainty in this region of our tree would present a roadblock, precluding our ability to reconstruct viable ancestral β subunit sequences. To address this concern, we chose to reconstruct the β subunit ancestral to both humans and Torpediniformes based on our β subunit molecular phylogeny. An ancestor shared by humans and Torpediniformes also has historical and practical significance. The first sequence information for any pLGIC subunit was derived from N-terminal sequencing of the AChR α subunit from Torpedo marmorata (Devillers-Thiéry et al., 1979), and subsequently Torpedo californica (Hunkapiller et al., 1979). From a practical perspective, the most recent common ancestor between humans and Torpediniformes existed ∼480 million years ago (Hedges et al., 2006), with the human β subunit sharing only ∼56% sequence identity with its Torpedo counterparts. Human and Torpedo subunits are thus sufficiently diverged from one another to represent a rigorous test of our ability to reconstruct viable ancestors, and test their compatibility with extant subunits. Sequence alignment reveals that the reconstructed ancestral β subunit (βAnc) shares ∼72% and ∼66% identity with the human and Torpedo subunits, respectively (Figure 1E). Mapping the substitutions onto the structure of an AChR β subunit reveals that Eubacteria are delocalized throughout the protein (Figure 1C). In general, the solvent exposed intra- and extracellular domains display more mutations in comparison with the transmembrane pore, with the highest mutational density centered within the cytoplasmic region between the M3 and MA helices, for which structural information is absent. As expected, regions previously identified as functionally important, including those at the interface between the extracellular ligand-binding and transmembrane domains, are highly conserved in βAnc. Nevertheless, the reconstructed ancestral subunit exhibits mutations in these regions, hinting that ancestral reconstructions have the potential to uncover functionally relevant combinations of residues.