Metabolomics as a Hypothesis Generator

The omics revolution has vastly expanded our capability to monitor and quantify molecular events in cells and tissues. Unfortunately, this technological advance has not (yet) been matched with conceptual advances in our ability to interpret and understand such data or to systematically generate hypotheses from them. In particular the rapidly developing metabolomics technologies generate data that are even harder to interpret mechanistically than transcript and protein abundances, simply because the direct connection to the genome is lost and many effects overlap. This is particularly problematic for metabolomics data in a cellular context, where multivariate statistics alone is typically not sufficient to obtain biologically meaningful hypotheses in fundamental science and metabolic engineering alike.

To temporally delineate overlapping influence on the metabolome, we developed technologies for broad coverage dynamic metabolomics at high-throughput. Based on previously developed high-throughput flow injection TOF mass spectrometry that enables to measure 400-800 metabolite in one sample per minute (1, 2), we developed automated, near real-time metabolome profiling directly from living cells at a resolution of 10 sec (3).  We thus effectively removed limitations on data generation in our lab, shifting the challenge even more to the data interpretation side. Here I will present two unpublished examples to illustrate recent advances in generating hypotheses on molecular functions from metabolomics data. In the first I will address the question how specific enzymes really are by in vitro metabolomics on about 1000 purified E. coli enzymes. We find a surprisingly large number of promiscuous enzymes and assess their evolutionary potential by FBA.  In the second example, I will exploit the potential of dynamic metabolomics data to systematically unravel bottlenecks in synthetic production pathways.

1. Fuhrer et al. 2011 Anal Chem 83: 7074;
2. Sevin & Sauer 2014, Nature Chem Biol 10: 266
3. Link H, Fuhrer T, Gerosa L, Zamboni N & U. Sauer. Nature Methods, 12: 1091-1079.