Metabolic network analysis of liver and adipose tissue

A biological system such as a cell is both structurally and functionally complex, as it is composed of many interacting molecules such as genes, proteins, and metabolites. When such a system experiences a perturbation, it can respond both locally and globally through a network of biochemical interactions. These often multi-layered responses cannot be understood by focusing on an individual molecule's behavior. Rather, mathematical frameworks are needed that integrate biochemical knowledge with quantitative experimental data to characterize the 'systems-level' behavior. This thesis presents novel approaches to model the enzyme-mediated connectivity of metabolic networks, and to integrate experimentally-derived data on the relative engagements of these connections.; These approaches were developed in three parts. The first part formulated an algorithm for top-down division of directed graphs with non-uniform edge weights. This algorithm enabled the rational partitioning of metabolic networks into natural groupings of metabolites, or 'modules.' A scoring system was formulated to determine which partitions likely produce biologically meaningful modules. The scoring was based on a projection of the partition results onto an inventory of feasible pathways defined by the elementary flux modes for the parent network. The second part applied the partition algorithm and scoring system to two distinct model systems, the rodent liver and fat cell. The analyses addressed how physiological stimuli elicit adaptations to the functional organization of the metabolic network. The analysis was subsequently expanded to include the responses of the liver network to a non-physiological stimulus, i.e. a drug chemical challenge. Finally, the third part addressed the issue of causal relations between network components through a probabilistic (Bayesian) framework. Metabolic flux data were used to identify conditional dependencies between reactions of the pyruvate sub-network in the liver and fat cell.; Through these studies, this thesis presents evidence that metabolic networks possess organizational flexibility, which enables the system to perform a variety of biochemical transformation functions using a finite set of reactions. Prospectively, the methods and findings of this thesis could be used to develop additional tools for systems biological studies on cellular metabolism, and thereby address significant questions in complex human diseases of multi-factorial origins.