Deep-learning-based design of synthetic orthologs of SH3 signaling domains.

Evolution-based deep generative models represent an exciting direction in understanding and designing proteins. An open question is whether such models can learn specialized functional constraints that control fitness in specific biological contexts. Here, we examine the ability of generative models to produce synthetic versions of Src-homology 3 (SH3) domains that mediate signaling in the Sho1 osmotic stress response pathway of yeast. We show that a variational autoencoder (VAE) model produces artificial sequences that experimentally recapitulate the function of natural SH3 domains. More generally, the model organizes all fungal SH3 domains such that locality in the model latent space (but not simply locality in sequence space) enriches the design of synthetic orthologs and exposes non-obvious amino acid constraints distributed near and far from the SH3 ligand-binding site. The ability of generative models to design ortholog-like functions in vivo opens new avenues for engineering protein function in specific cellular contexts and environments.
Competing Interests: Declaration of interests R.R. and A.L.F. are co-founders and consultants of Evozyne, Inc. and co-authors of US Patent Application 17/642,582 and International Patent Application PCT/US2020/050466. A.L.F. is also co-author of US Patent Application 16/887,710, US Provisional Patent Applications 63/314,898, 63/479,378, and 63/521,617 and International Patent Application PCT/US2020/035206.
(Copyright © 2024 Elsevier Inc. All rights reserved.)