P67: Directed evolution of a computationally designed esterase

The burgeoning field of computational enzyme design has shown great promise for the design of novel protein catalysts, but achieving native-like activity remains a difficult problem. High-throughput screening and directed evolution has been used to improve a number of enzymes very successfully.  However, this powerful approach has not yet been used to improve de novo computationally designed enzymes, with most efforts consisting of low to medium throughput screens of mutants.

Enzymes active for the hydrolysis of non-natural ester compounds were designed onto inactive scaffold proteins using the Rosetta molecular modeling and design suite. Mutation screening was used to identify more active variants of these enzymes, with multiple variants identified with 200-400% increases in activity.

Starting from one of these improved enzymes, we devised a directed evolution approach to further improve activity.  The mechanism for hydrolysis of the ester substrate passes through a covalent acyl-enzyme intermediate, and our selection uses this to allow selection for increased enzyme activity. Using a tagged version of the enzyme’s substrate as a probe, we selected for formation of the acyl-enzyme intermediate by yeast display and fluorescence activated cell sorting from libraries of variants.

Multiple serial and parallel selections were performed, with acylation rates improving over the course of the selections. We also saw improved enzyme kinetics for these variants when testing with the original untagged substrate.  High throughput screening and directed evolution holds promise to direct improvements to computational enzyme design methods and help bring de novo designed enzymes to native-like levels of activity.