Missing Links in the Genotype-Phenotype Path: Metabolic Network Connectivity Together with Protein Levels As Predictors of Metabolite Changes

Reorganization of intra-cellular metabolite levels is a requirement for adaptation and evolution of organisms. Interfering with metabolite levels can generate opportunities to target cancer and design biotechnologically important ‘cell-factories’. Metabolic adaptation occurs through complex and barely understood mechanisms, which include metabolic feed-back regulation, transcriptional regulation and signal transduction. A primary regulatory mechanism used by cells to control the activity of their metabolic network is the alteration of enzyme levels through changes in the expression of enzyme-coding genes. Here we demonstrate an analytical approach with the potential to predict metabolite levels from enzyme protein data. We analysed >100 deletion mutants of protein kinases in Saccharomyces cerevisiae using a newly developed high-throughput mass spectrometry platform based on SWATH technology, generating over 450 proteomes.  By integrating the topology of yeast metabolic network with proteome data and metabolite measurements using linear models we could identify the strongest variables affecting variation in metabolite levels. For multiple metabolites, including energy charge metabolites (ATP, ADP, AMP), we could well explain (up to R² = 0.7) metabolite level variation using protein expression of enzymes involved in their production and consumption reactions. Overall, this study demonstrates an example of combining high-throughput proteomics and metabolomics technologies with a priori biological knowledge of metabolic networks. This data-driven approach has implications for decision-making in health and biotechnological applications.