Continuous evolution of dynamic, multi-state, and computational protein functionalities

Directed evolution is a powerful method for engineering proteins. Current approaches are best suited for refining and creating steady-state properties, e.g., the affinities of protein binders or brightness of fluorescent proteins. It is impractical or unfeasible to evolve dynamic or computational functionalities of proteins that can switch between multiple conformations. On the other hand, these kinds of switches are essential for studying the dynamics of biological systems, dissecting complex regulatory networks, and constructing synthetic biological circuits with tunable functions. Here, we develop a continuous, in vivo directed evolution approach for evolving proteins with multiple conformations. We assign the protein of interest to control the cellular process whose dynamic performance affects the fitness of the host cells. Hence, the cells that accumulate beneficial mutations in the gene of interest gain selective advantage and therefore lead to the enrichment of the gene-variants of interest in the population, resulting in a self-selecting system. Given that existing in vivo directed evolution methods do not allow simultaneous selection of on and off properties of dynamic proteins, the approach we introduce provides solutions to otherwise intractable protein optimization problems.