Systematic discovery of enzyme promiscuity in Escherichia coli using in vitro metabolomics

Metabolic enzymes have traditionally been regarded as highly specific catalysts; however, many can catalyze multiple reactions. To systematically investigate the prevalence of such enzyme promiscuity, we used nontargeted metabolomics to measure dynamic metabolite ion profiles in in vitro assays with 667 successfully purified Escherichia coli enzymes in a natural intracellular metabolome extract. Notably, nearly half of these enzymes elicited significant changes in ion traces. Using a machine learning-derived multivariate classifier at a false-discovery rate of 33%, we identified unexpected changes in 135 putatively annotated metabolite ion traces, indicating the presence of so far unknown promiscuous activities in 11% of the tested enzymes, most of which have yet to be recognized for their ability to catalyze multiple reactions. Notably, we found that nucleotide-related substrates or cofactors were enriched among the newly identified reactants. For 11 promiscuous enzymes, we successfully reconstructed 22 complete reaction stoichiometries, four of which were validated experimentally. Key findings include the nucleoside phosphorylase DeoA, for which we expanded the substrate range to include pyrimidines relevant to carbon and energy utilization, and the N-acetylmannosamine kinase (NanK), which displayed both cofactor and sugar substrate promiscuity. Additionally, CobC, a putative adenosylcobalamin/α-ribazole phosphatase, catalyzes flavin mononucleotide dephosphorylation, suggesting a generalist role in vitamin biosynthesis pathways. Beyond specific examples, the results suggest that metabolism harbors a wealth of underexplored catalytic flexibility, relevant for functional annotation, evolution, and genome-scale metabolic models.