Multistate-CANVAS Enables the De Novo Design of Enzymes for Complex Multistep C-C Bond-forming Reactions

Abstract

The design of enzymes for reactions lacking natural biocatalysts is a major goal of computational protein design. While computational methods have enabled the creation of de novo enzymes for diverse chemical transformations, designing enzymes for complex multistep C-C bond-forming reactions remains a significant challenge. These reactions require a single active site to accommodate multiple substrates and evolving catalytic requirements throughout the reaction coordinate, making them difficult to represent using conventional single-state computational design approaches. To address this challenge, we developed Multistate-CANVAS (MsC), a computational workflow for the de novo design of enzymes for multistep C-C bond-forming reactions. As a proof of concept, MsC was applied to the Friedel-Crafts alkylation of 2-methylindole with trans-cinnamaldehyde, a multistep C-C bond-forming reaction. Experimental characterization demonstrated that multiple variants could be successfully expressed, purified, folded, and exhibited measurable catalytic activity. The most active variant, FCS_1, achieved a 254-fold enhancement in conversion relative to the uncatalyzed background reaction with a catalytic efficiency of 0.01 M⁻¹ s⁻¹. The successful generation of active Friedel-Crafts alkylases suggests the feasibility of extending computational enzyme design to complex multistep C-C bond-forming reactions. More broadly, MsC provides a framework for expanding the scope of de novo enzyme design toward increasingly challenging new-to-nature chemistry.