Data-Driven Screening to Infer Metabolic Modulators of the Cancer Epigenome

Abstract

AbstractMetabolites such as acetyl-CoA and citrate play an important moonlighting role by influencing the levels of histone post-translational modifications (PTMs) and regulating gene expression. This cross talk between metabolism and epigenome impacts numerous biological processes including development and tumorigenesis. However, the extent of moonlighting activities of cellular metabolites in modulating the epigenome is unknown. We developed a data-driven screen to discover moonlighting metabolites by constructing a histone PTM-metabolite interaction network using global chromatin profiles, metabolomics, and epigenetic drug sensitivity data from over 600 cancer cell lines. Our ensemble statistical learning approach uncovered metabolites that are predictive of histone PTM levels and epigenetic drug sensitivity. We experimentally validated synergistic and antagonistic interactions between histone deacetylase and demethylase inhibitors with the metabolites kynurenic acid, pantothenate, and 1-methylnicotinamide. We apply our approach to track metabolic-epigenetic interactions during the epithelial-mesenchymal transition. Overall, our data-driven approach reveals extensive metabolic-epigenetic interactions than previously thought, with implications for reversing aberrant epigenetic alterations and enhancing epigenetic therapies.