Genome-scale modeling identifies dynamic metabolic vulnerabilities during the epithelial to mesenchymal transition.

Epithelial-to-mesenchymal transition (EMT) is a conserved cellular process critical for embryogenesis, wound healing, and cancer metastasis. During EMT, cells undergo large-scale metabolic reprogramming that supports multiple functional phenotypes including migration, invasion, survival, chemo-resistance and stemness. However, the extent of metabolic network rewiring during EMT is unclear. In this work, using genome-scale metabolic modeling, we perform a meta-analysis of time-course transcriptomics, time-course proteomics, and single-cell transcriptomics EMT datasets from cell culture models stimulated with TGF-β. We uncovered temporal metabolic dependencies in glycolysis and glutamine metabolism, and experimentally validated isoform-specific dependency on Enolase3 for cell survival during EMT. We derived a prioritized list of metabolic dependencies based on model predictions, literature mining, and CRISPR-Cas9 essentiality screens. Notably, enolase and triose phosphate isomerase reaction fluxes significantly correlate with survival of lung adenocarcinoma patients. Our study illustrates how integration of heterogeneous datasets using a mechanistic computational model can uncover temporal and cell-state-specific metabolic dependencies.