Decoding the substrate specificity landscape of a promiscuous enzyme through multi-substrate mutational scanning

Substrate specificity is a defining feature of enzyme function, but its molecular underpinnings remain difficult to decode and engineer. Here, we leverage enzyme proximity sequencing (EP-Seq) to systematically map how single-point and combinatorial mutations reshape the substrate preferences of D-amino acid oxidase (DAOx) from Rhodotorula gracilis, a model promiscuous enzyme. We generate ~40,000 sequence-phenotype pairs, enabling us to profile the activities of ~6,500 unique DAOx variants against five D-amino acid substrates with distinct physicochemical properties. Our analysis reveals that substrate-specific mutations are distributed throughout the enzyme structure. Mutations near the active site drive strong specificity shifts but also incur catalytic penalties, while distal mutations subtly rewire intramolecular contacts in order to modulate specificity with minimal loss of activity. We identify and validate positional hotspots that act allosterically to influence specificity, and characterize key variants that acquire exclusive substrate specificity or exhibit up to 230-fold changes in substrate preference. Combining mutations with complementary effects further sharpens substrate discrimination, enabling rational design of highly selective biocatalysts. This work establishes a powerful framework for decoding enzyme specificity and provides foundational datasets to advance AI-guided enzyme engineering.