Abstract
The function of dynamic proteins is determined by the stability of distinct conformational states and the energy barriers that separate these states. For most dynamic proteins, the molecular details of the energy barriers are not known, implying a fundamental limit to the ability of protein design methods to engineer beneficial mutations without disrupting activity. We hypothesized that designing mutations that are compatible with structurally distinct equilibrium conformations may enable a reliable stability design. We focus on periplasmic binding proteins (PBPs), a superfamily of dynamic proteins that change conformation from open to closed states in response to binding their small-molecule ligands. We find that the evolutionary constrained space of allowed mutations computed for one conformation is incompatible with that for the other. Therefore, putative conformational hinge points and interface residues were additionally constrained, and incompatible mutations were filtered out. Starting from four different PBPs, we designed a total of 16 stabilized variants with 7-28 mutations each. Our results show that design based on a single conformation with evolutionary constraints is not sufficient to maintain a wild-type-like binding affinity. Conversely, using a subset of mutations compatible with both conformations and structural constraints reliably enhances thermal stability while mitigating trade-offs in ligand binding. Our work demonstrates a straightforward method for the one-shot stabilization of dynamic proteins, which is critically required to generate robust starting points for thermostable and responsive biosensors.