A Data-Driven Approach to Protein Engineering of Aminotransferase ARO8

Project Summary/Abstract Biocatalytic reactions are critical to the production and discovery of modern pharmaceuticals. Directed evolution (DE) of enzymes enables the tuning of enzymatic selectivity to address the growing demand for complex natural product-derived therapeutics. These DE campaigns suffer from an enormous search space of possible mutations and slow initial optimization with unnatural substrates, which hinders the development of new biocatalytic pathways. Despite the importance of robust biocatalytic platforms, generalizable methods for predicting optimal mutations that enhance selectivity of an unnatural substrate remain elusive. To address these issues, this proposal applies data-driven reaction optimization methodology to the DE of aminotransferase ARO8. Recently, our collaborators at the Narayan group have discovered a novel C–C bond forming reaction of the PLP- dependent enzyme ARO8. Given the industrial importance of aminotransferases, a mechanistic understanding of this novel reactivity is critical for leveraging aminotransferases as flexible biocatalysts. The proposed data- driven workflow will reveal the subtle non-covalent interactions that govern this reactivity by computationally modeling multiple mutations in silico to predict mutations to maximize C–C bond formation. Furthermore, a multiple linear regression model will be developed to probe the fundamental residue-substrate interactions that promote C–C bond formation and further our understanding of aminotransferases. The studies in this proposal focus on ARO8. However, we anticipate that the proposed computational workflow and prediction methodology is transferable to other enzymes and substrates.