Evolution and engineering of protein-protein interactions

Abstract

Protein-protein interactions are crucial elements in most biological processes. The gain and loss of interactions during evolution have important phenotypic consequences that are subject to selection. Therefore, in a crowded cellular environment, proteins must evolve mechanisms to maintain the correct interactions and avoid inappropriate ones. In this work, I leveraged high-throughput methods for the functional characterization of thousands of protein variants to characterize the sequence spaces associated with paralogous families of interacting proteins. Protein families are formed by gene duplication and divergence, a common source of evolutionary novelty. Family members maintain conserved structural and sequence elements, and yet must often form distinct protein-protein interactions. To probe the extent to which the requirement for interaction specificity constrains evolution, I focused on the twocomponent system family of bacterial signaling proteins. I tested protein variants with all possible single substitutions in the interacting domain of a model protein for their ability to interact with a cognate partner protein and with closely related non-cognate partners. I found that a large fraction of substitutions introduce non-specific interactions, suggesting that paralogs only evolve ‘marginal specificity’ that can easily be disrupted. Bioinformatic evidence indicates that the resulting crowded local sequence space has restricted the evolvability of two-component systems. I also characterized the effects of environmental context constraints, specifically temperature, on the sequence space relevant to two-component system function. This revealed generally conserved sequence-function landscapes across temperatures, with small numbers of variants showing either temperature sensitivity or resistance. Biochemical characterization of these variants challenges existing paradigms relating to the effects of temperature on evolution. Finally, I utilized insights into the evolution of protein-protein interaction specificity to inform the design of protein binders to toxin-antitoxin systems. These binders are selective in their interactions with toxin homologs, and inhibit toxin-antitoxin-mediated bacterial anti-phage defense activity, suggesting their potential use in clinical phage therapy applications. Taken together, these results shed light on the role of protein-protein interactions and their specificity in shaping evolution and suggest the utility of leveraging interaction specificity for engineering purposes.