De novo design of photoenzymes with substrate-recognition-encoded enantioselectivity

Abstract

Photoenzymes enable stereoselective excited-state transformations beyond the chemical space accessible to natural enzymes, yet their development remains largely dependent on repurposing few privileged protein scaffolds and extensive directed evolution. Although de novo protein design enables tailor-made catalytic architectures, its prevailing transition-state stabilization paradigm is not readily transferable to photoenzymes. Here we establish a substrate-recognition-guided strategy for de novo photoenzyme design, in which the computational construction of substrate-tailored catalytic environments determines stereoselectivity and the genetic incorporation of abiotic thioxanthone photosensitizer endows excited-state reactivity. This approach generated customized photoenzymes for the deracemization of 4-alkenylcyclohexanols that exhibit nanomolar substrate affinity, catalytic efficiencies exceeding 1.5 × 105 M-1 s-1, and up to 99% enantiomeric excess. Crystal structure and kinetic analyses validate the designed model, revealing substrate-recognition-dependent enantioselective energy transfer within the tailored active site. This work establishes substrate recognition as a general design principle for extending de novo enzyme design to excited-state biocatalysis.