A Dynamic Flux-Driven Framework for Metabolic Enzyme Regulation Modeling

Thesis Abstract

Metabolic enzyme regulation emerges as a dynamic orchestration of flux-driven control points across cellular networks, yet current models inadequately capture real-time adaptivity of enzyme activities to fluctuating metabolite abundances and environmental perturbations. This study addresses the gap by developing a dynamic flux-driven framework that integrates regulatory mechanisms with kinetic enzyme constraints to predict context-specific metabolic responses. The aim is to formulate and validate a computational model that links intracellular flux distributions to regulatory enzyme states, enabling accurate forecasting of metabolic phenotypes under varying nutrient conditions and stressors. Specific objectives include (1) to synthesize a modular mathematical representation that couples flux balance with phosphorylation, allosteric modulation, and substrate channeling; (2) to calibrate the framework against time-resolved metabolomics and phosphoproteomics data; (3) to assess model robustness through sensitivity analysis and cross-condition validation; (4) to benchmark predictive performance against established metabolic models; and (5) to demonstrate translational relevance by simulating metabolic shifts in a representative mammalian cell line under hypoxic and nutrient-limiting conditions. The methodology adopts a multi-method research design combining systems biology modeling with empirical data integration. The population consists of HepG2 hepatocyte-like cells and U2OS osteosarcoma cells subjected to defined perturbations (hypoxia, glucose deprivation, and AICAR-induced AMPK activation). A sample size of n=30 biological replicates per condition, across three independent experiments, will be used to obtain time-series measurements. Data collection employs targeted metabolomics (LC-MS/MS for central carbon metabolites), untargeted metabolomics for pathway-level insights, and quantitative phosphoproteomics (SILAC-based workflows) to capture post-translational regulation signals. Additional enzyme activity assays will quantify specific catalytic steps. The instrumented data serve to parameterize and validate the dynamic flux-driven model. Analytical methods involve a hierarchical, modular modeling approach. First, a dynamic flux balance analysis (dFBA) layer defines time-dependent intracellular fluxes constrained by mass balance, energy state, and substrate availability. Second, a kinetic regulation layer introduces enzyme activity states governed by phosphorylation status and allosteric effectors, modeled via a system of ordinary differential equations with thermodynamically consistent rate laws. Third, parameter estimation utilizes Bayesian inference (MCMC) to quantify uncertainty in kinetic parameters and regulatory coefficients, with priors informed by literature and phosphoproteomics data. Model calibration proceeds in two stages (i) fitting steady-state and time-series fluxes to metabolomics data, and (ii) aligning enzyme regulation dynamics with phosphoproteomics trajectories. Model validation employs cross-condition predictive checks, leave-one-condition-out validation, and comparison with conventional stoichiometric models using root-mean-square error (RMSE) and concordance correlation coefficients (CCC). Sensitivity analysis via Sobol indices quantifies the influence of regulatory parameters on key outputs such as ATP yield, NADH/NAD+ balance, and lactate production. Expected findings include demonstration that incorporating dynamic regulatory states substantially improves the accuracy of flux predictions under perturbations, with improvements in RMSE by 12–28% and CCC gains of 0.05–0.15 over baseline dFBA models. The framework is anticipated to reveal condition-specific regulatory strategies, such as coordinated phosphorylation shifts that reallocate flux through glycolytic and oxidative pathways during hypoxia, and allosteric control patterns that stabilize energy charge under glucose limitation. The study contributes to knowledge by providing a validated, extensible framework that bridges metabolic flux modeling with regulatory signaling and post-translational modification data, enabling mechanistic interpretation of metabolic reprogramming in health and disease. The implications extend to drug targeting and metabolic engineering by identifying regulatory choke points and dynamic control nodes with high leverage on cellular phenotypes. The main conclusion is that a dynamic flux-driven regulatory model offers superior explanatory and predictive power for cellular metabolism compared to traditional static or purely kinetic models, and recommendations include extending the framework to multi-omics integration, applying it to different tissue types, and incorporating stochastic fluctuations for single-cell resolution analyses.