Non-targeted in vitro metabolomics for proteome-scale identification of novel enzymes in Escherichia coli

Many yet undiscovered enzymes still limit our understanding of metabolism. Here, we develop a high-throughput mass spectrometry 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, and accumulating or depleting metabolite ions are determined by nontargeted mass spectrometry. Combining chemometrics and database approaches, we established an automated pipeline for unbiased annotation of novel enzyme functions that was demonstrated to achieve 75% predictive accuracy. In a screen of all 1,275 functionally uncharacterized Escherichia coli proteins, this low false-positive rate discovery method identified 241 potential novel enzymes. Twelve of these were experimentally validated to catalyze 29 diverse metabolic reactions, including the long-elusive multispecific cytidine monophosphate nucleosidase PpnN (YgdH) and the two fumarases FumD (YdhZ) and FumE (YggD). Thus, we establish high-throughput nontargeted in vitro metabolomics and the corresponding hypothesis generation pipeline as an approach for proteome-scale enzyme discovery and assign enzymatic functions to about 20% of the so far uncharacterized E. coli proteins.