(34a) Integrating Transcription Factor Dynamics and Metabolic Network Activity in Differentiating Adipocytes

Adiposity, and hence obesity, occurs through both increases in fat cell number (hyperplasia) and size (hypertrophy). The increase in cell number involves the recruitment, proliferation, and differentiation of locally resident precursor cells into mature adipocytes (adipogenesis). Terminally differentiated adipocytes do not undergo further mitotic division. However, genetic or environment conditions can bring about significant hypertrophic growth through progressive lipid loading. During adipogenesis, a handful of key transcription factors such as PPAR-γ and C/EBP-α act as master switches for a complex cascade of gene expression events to convert undifferentiated precursor cells into metabolically active adipocytes. Isolated observations on gene expression patterns or transcription factor levels are unlikely to fully characterize the regulatory events underlying the metabolic phenotype changes, because these data contain only limited information on the dynamics of interactions between the regulatory (transcription factor, TF) and metabolic networks. In this study, we develop an integrated model relating the adipocyte TF levels to the activities of major metabolic pathways. To obtain dynamic profiles of the regulatory network during adipogenesis, we generated a stably transduced murine preadopocyte line as a reporter system for each of the following TFs: PPAR-γ, C/EBP, SREBP, CREB, FOXO1, and NFAT. Following induction of adipogenic differentiation using a standard chemical cocktail, TF activities were monitored for 21 days at timed intervals (ranging from 2 to 12 hrs). Parallel experiments in batch cultures profiled the time-dependent changes in the exchange rates of major primary metabolites released or consumed by the adipocyte. The combined TF and metabolite data were used to train a set of coupled mass action rate law (MRL) expressions describing the time evolution of TF and intracellular metabolite concentrations. Model parameters were obtained by non-linear optimization with the objective function minimizing the sum squared error between measured and calculated TF and metabolite levels. Optimization was performed using the Genetic Algorithms and Directed Search (GADS) toolbox in MATLAB. Simulation results using the trained model suggested the presence of a dynamic feedback loop between lipogenic enzymes, their synthesis products (fatty acids) and several of the TFs, including PPAR-γ. On-going work tests the impact of specific perturbations to the metabolic network that reduce lipid loading. Prospectively, the described methodology could be applied to identify novel molecular targets for pharmacological therapy of excessive adiposity.