Accurate models of substrate preferences of post-translational modification enzymes from a combination of mRNA display and deep learning
Promiscuous post-translational modification (PTM) enzymes often display non-obvious substrate preferences by acting on diverse yet well-defined sets of peptides and/or proteins. Thorough understanding of substrate fitness landscapes for promiscuous PTM enzymes is important because they play key roles in many areas of contemporary science, including natural product biosynthesis, molecular biology and biotechnology. Here, we report the development of an integrated platform for accurate profiling of substrate preferences for PTM enzymes. The platform features a combination of i) mRNA display with next generation sequencing as an ultrahigh throughput technique for data acquisition and ii) deep learning for data analysis. The high accuracy (>0.99 in each of two studies) and generalizability of the resulting deep learning models enables comprehensive analysis of enzymatic substrate preferences. The models can be utilized to quantify fitness across sequence space, map modification sites, and identify important amino acids in the substrate. To benchmark the platform, we perform substrate specificity profiling of a Ser dehydratase (LazBF) and a Cys/Ser cyclodehydratase (LazDEF), two enzymes from the lactazole biosynthesis pathway. In both studies, our results point to highly complex enzymatic preferences, which, particularly for LazBF, cannot be reduced to a set of simple rules. The ability of the constructed models to dissect and analyze such complexity suggests that the developed platform can facilitate the wider study of PTM enzymes.