The structure of an ancient genotype-phenotype map shaped the functional evolution of a protein family

Some phenotypes are more likely to be produced by mutation than others, but the causal role of these propensities in the evolution of extant phenotypic diversity remains unclear. There are two major challenges: it is difficult to separate the effect of the genotype-phenotype (GP) map from that of natural selection in causing natural patterns of diversity, and most extant phenotypes evolved long ago in species whose GP maps cannot be recovered. Using reconstructed ancestral transcription factors as a model to address this problem, we created libraries containing all possible amino acid combinations at historically variable sites in the proteins DNA binding interface (the genotypes) and measured their capacity to bind specifically to response elements containing all possible combinations of nucleotides at historically variable sites in the DNA (the phenotypes). The ancestral proteins we used existed during an ancient phylogenetic interval when a new phenotype--specificity for a new response element--evolved. We found that the two ancestral GP maps were strongly anisotropic (the distribution of phenotypes encoded by genotypes is highly nonuniform) and heterogeneous (the phenotypes accessible around each genotype vary dramatically among genotypes), but the extent and direction of these properties differed between the maps. In both cases, these properties steered evolution toward the lineage-specific phenotypes that evolved during history. Our findings establish that ancient properties of the GP relationship were causal factors in the evolutionary process that produced the present-day patterns of functional conservation and diversity in this protein family.