Development of model-driven approaches for metabolic flux analysis and anticancer drug discovery

Abstract

[eng] Metabolism is a hallmark of life and underlies most biological processes in both health and disease. For instance, dysregulation of liver metabolism underlies multifactorial disorders such as diabetes or obesity. Similarly, cancer progression involves a reprogramming of metabolism to support unchecked proliferation, metastatic spread and other facets of the cancer phenotype. Hence, the study of metabolism is of great biomedical interest. The metabolic phenotype emerges from the complex interactions of metabolites, enzymes, and the signaling cascades regulating their expression and thus must be studied following a holistic approach. With this aim, Systems Biology formulates the interactions between the molecular components of metabolism as a set of mathematical expressions, termed metabolic models, and uses them as a framework to integrate multiple layers of data (e.g., transcriptomics, proteomics and metabolomics) and simulate the emergent metabolic phenotype. The Systems Biology toolbox for the analysis of metabolism consists of several complementary model-based approaches, each with its strengths and limitations. For instance, constraint-based modeling can predict flux distributions at a genome-scale, whereas kinetic modeling and 13C metabolic flux analysis (13C MFA) can more accurately model central carbon metabolism. As part of this Ph.D. thesis, we have expanded this toolbox through the development of new model-based approaches for computing both detailed metabolic maps of central carbon metabolism and genome-scale flux maps. With this aim, we developed HepatoDyn, a highly detailed kinetic model of hepatocyte metabolism capable of dynamic 13C MFA and used it to characterize the negative effects of fructose in hepatic metabolic function. Similarly, we also developed Iso2Flux, a novel steady-state 13C MFA software, and parsimonious 13C MFA, a new 13C MFA algorithm that can integrate transcriptomics to trace flux through large metabolic networks. Even more, we developed r2MTA a constraint-based modeling algorithm to robustly identify the optimal interventions to induce a transition towards a therapeutically desirable metabolic state. Finally, we also developed a workflow for integrating transcriptomics, metabolomics, gene dependencies, and 13C MFA to predict genome-scale flux maps. Furthermore, we apply the systems biology toolbox, using both newly developed and existing tools, to the genome-scale analysis of the molecular drivers underlying cancer stem cells and metastasis in prostate and colorectal cancer, respectively. We identify putative therapeutic interventions against both phenotypes paving the way for a new generation of anticancer drugs.