DC SÉVIN¹, U SAUER<sup>1

1</sup>Institute of Molecular Systems Biology, ETH Zurich

Many yet undiscovered enzymes still limit our understanding of metabolism, and a growing body of literature suggests that many known enzymes might promiscuously catalyze additional reactions. Here, we present a high-throughput metabolomics approach to comprehensively profile proteome-scale protein collections for in vitro enzymatic activity. Overexpressed or purified proteins are incubated in a supplemented metabolome extract containing hundreds of biologically relevant candidate substrates. Accumulating or depleting metabolite ions are then identified using nontargeted metabolomics, and a combination of chemometrics and database approaches is used to assign chemical reactions.
Upon extensive method validation, a screen of 1,275 functionally uncharacterized Escherichia coli proteins revealed 241 potential novel enzymes. Moreover, a subsequent screen of all 1,054 known enzymes of E. coli identified previously unknown reactants for 495 enzymes. Intriguingly, many discovered reactions involved non-canonical or unknown metabolites, indicating that these reactions may be part of novel metabolic pathways. Using constraint-based modeling, we finally predict these reactions to enhance metabolic evolvability of E. coli in environments with flucutating nutrient availability by facilitating the evolution of novel catabolic pathways.
Thus, we establish high-throughput nontargeted in vitro metabolomics as an approach for proteome-scale enzyme discovery and offer a glimpse on the astonishing catalytic versatility of E. coli enzymes.