A Global Experimental Analysis of Protein Function: a Case Study in the PDZ Domain

A complete understanding of the energetic architecture of a protein can be achieved only with a comprehensive description of the interaction of every amino acid with every other amino acid. Many efforts to understand the apparent complexity of protein function have attempted to address this problem with limited mutagenesis studies. A global computational description of amino acid interactions, Statistical Coupling Analysis has shown the existence of a contiguous subset of positions within a protein that displays significant co-evolution, termed protein sectors. Limited mutagenesis studies have shown sectors to be networks of higher-order interaction crucial for protein function; however, a theory of such global scope requires validation with a global experiment. Here, we design an assay system that measures the cellular function of a PDZ domain in a high-throughput and quantitative manner. We perform a comprehensive single amino acid mutagenesis experiment to show that most positions in the protein are robust to most mutations, and the set of positions that shows sensitivity to mutation is enriched for sector positions. Further, we perform a global pairwise epistasis experiment in which we measure the way in which every amino acid mutation in the PDZ domain feels the effect of a second mutation at a key specificity and affinity determining position in the peptide ligand of the PDZ domain. We find that those positions that show strong non-additivity in the context of the peptide mutation are all contained within the PDZ sector. Further, these sector positions that display strong non-additivity all display the property of rapidly changing specificity upon mutation. That is, any mutation at these sector positions has a negative functional effect in the context of the endogenous peptide. However, these positions appear to be spring-loaded for change since these same mutations enhance function in the context of an alternative peptide. We hypothesize that proteins are robust as shown by their insensitivity to general mutation. However, proteins are simultaneously fragile as shown by their sensitivity to specific mutagenesis at sector positions. This fragility, however, is strongly coupled to evolvability as shown by the enhancement of alternative function endowed by these endogenously detrimental mutations.