De Novo Enzyme Design by Customizing Minimal TIM Barrels

Abstract

Enzymes are the most efficient and selective catalysts known, and the ability to design an enzyme for a target reaction is a primary goal of protein engineering. Among natural enzymes, the TIM barrel represents the most prevalent and versatile protein fold, using extended loops and other structural elements to facilitate substrate binding, catalysis, and product release. While de novo TIM barrels have been successfully designed, their minimalistic architectures lack these essential structural elements. In this work, we present CANVAS, a computational framework for de novo enzyme design that introduces a structural lid into a minimal de novo TIM barrel to anchor catalytic residues and form an active-site pocket. To evaluate this workflow, we design TIM barrel enzymes with active sites for the Kemp elimination reaction. Four out of nine designs showed measurable activity, with the most active reaching a catalytic efficiency of 21,000 M⁻¹ s⁻¹ at pH 10. Crystal structures of this variant bound to a transition-state analogue confirmed the accuracy of the designed lid and identified a shift in the designed lid between bound and unbound forms, reminiscent of rearrangements in natural TIM-barrel enzymes. Using the X-ray structure of a lower-activity variant (19 M⁻¹ s⁻¹), we applied ensemble-based design to optimize its active site, increasing catalytic efficiency to 32,000 M⁻¹ s⁻¹. This optimized enzyme also demonstrated high regioselective preference for the target substrate. These results demonstrate that de novo TIM barrels can be customized with bespoke substrate binding pockets supporting efficient catalysis, establishing a platform for building de novo enzymes from minimal protein scaffolds. We propose strategies to push catalytic efficiency further by expanding CANVAS and combining it with complementary approaches, and we describe the next frontiers for de novo enzyme design.