Active site diversification of a non-canonical amino acid decarboxylase by merging substrate multiplexed screening with computationally guided recombination.

Recombination of active site mutations is a powerful strategy to alter enzyme activity. The vastness of sequence space, however, often limits screening-based engineering to single and double site libraries. Here, we explore focused recombination across five positions that enclose the active site of a tryptophan (Trp) decarboxylase. Our goal was to maximize the sequence diversity of enzymes that decarboxylate non-canonical amino acids (ncAAs) with minimal screening effort. We used substrate-multiplexed screening (SUMS) to distinguish recombinants that have impaired activity with all substrates from those that have altered specificity. In this way, we identified a larger fraction of active sequence space than could be found by single substrate screening alone. Wild-type primer doping during library assembly enabled the enrichment of double and triple mutants while simultaneously scanning five positions. A small screening effort, <200 measurements, was sufficient to train a logistic regression model that enriched active regions of the recombination space. This iterative strategy to library design resulted in TDC variants with distinct promiscuity profiles, and one variant displayed a nearly 500-fold increase in catalytic efficiency compared to wild-type TDC. These results illustrate how SUMS can be combined with iterative, deep recombination to generate a panel of catalytically diverse active site architectures.