Biochemical Systems Analysis of the Epithelial to Mesenchymal Transition

Cell development refers to the capacity of specific cells to undergo the functional and morphological changes that ultimately drive tissue and organ development. The epithelial to mesenchymal transition (EMT) is a developmental event where epithelial cells gain mesenchymal cell like properties. EMT is of importance during embryonic development, cancer progression and wound healing. Cell models that undergo EMT also provide experimental platforms where different genotype phenotype relationships can be explored on the basis of an identical genetic background. Gene expression and transcription factor analysis have therefore been fundamental to defining the molecular biology of EMT while metabolism and its role remains incompletely characterized.

Here we report an interdisciplinary effort towards defining the metabolism of EMT and how these compare with known regulatory features of EMT. Using a novel COBRA method we generated genome scale metabolic models (GEMs) based on human RECON X, constrained with RNA expression data intensities. Two models representing an in-house non-cancerous immortalized breast epithelial derived cell line, capable of undergoing EMT in vitro, alongside its mesenchymal like counterpart, were produced. Comparison of these highlighted metabolic differences that were however challenged when the models were further constrained with comprehensive metabolic data. The metabolic data ultimately served to define phenotypes that, interestingly, do not agree with previous reports on increased aerobic glycolysis following EMT. Follow up investigations in different cell lines support our observations of altered glycolysis, TCA cycle flux and lipid recycling and serves to define metabolism during EMT. We are currently pursuing these phenotypic alterations further via target gene knockdowns. Interestingly, when the GEMs were constrained with signaling constraints from a literature based EMT regulatory signaling network model, similar metabolic phenotypes were observed.

The results highlight pitfalls during context specific GEM reconstruction using multiple omics data types and suggest that more accurate metabolic phenotypes might be obtained by imposing regulatory constraints on metabolic GEMs as opposed to expression profiles of metabolic enzymes alone. Furthermore the results serve to better define metabolic phenotypes of EMT that have primarily been based on transcriptomic data of EMT cancer cell lines and therefore do not necessarily capture hallmarks of normal EMT.