Semi-explicit Solvation Improves Ligand Binding Site Design in an Allosteric Protein

Abstract

A protein's function is closely derived from its structure; the ability to understand and precisely design structures can thus help create proteins with specific functions. An often-overlooked aspect of protein structural design is the placement of individual water molecules, which are normally approximated as bulk solvent. Previous work has shown that specifying key water molecules improves structural predictions. In particular, designing binding sites in proteins could benefit from modeling waters because the binding affinity and resulting response of proteins to their ligands are influenced by water-mediated hydrogen bonds. Here, we applied solvation to the design of the allosteric transcriptional repressor LacI, with the design goal of inverting the inducibility of both an inducing and anti-inducing ligand. We used a statistical semi-explicit solvation model to solvate LacI; this elucidated specific hydrogen bonds we used to inform our computational redesign of the LacI binding site. We experimentally screened the top six designs to measure their response to various ligands and observed a variety of altered responses compared to wild type; a future challenge is to pinpoint the interactions responsible for the observed allosteric effects. Altogether, our results indicate that considering semi-explicit waters yields greater insight into ligand-protein interactions and can improve current protein design protocols.