A survey of pairwise epistasis supports an outside-in hierarchy of clade-specifying and function-defining residues in PSD95 PDZ3

Abstract

Deep mutational scanning enables data-driven models of protein structure and function. Here, we adapted Saturated Programmable Insertion Engineering as an economical and programmable deep mutational scanning technique. We validate this approach with an existing single mutant dataset in the PSD95 PDZ3 domain, and further characterize most pairwise double mutants to study how a mutation’s phenotype depends on mutations at other sites, a phenomenon called epistasis. We observe wide-spread proximal negative epistasis, which we attribute to mutations affecting thermodynamic stability, and strong long-range positive epistasis, which is enriched in an evolutionarily conserved and function-defining network of ‘sector’ and clade-specifying residues. Conditional neutrality of mutations in clade-specifying residues compensates for deleterious mutations in sector positions. This suggests an outside-in hierarchy of interactions through which positive epistasis between clade-specifying residues and the PDZ sector facilitated the evolutionary expansion and specialization of PDZ domains.