Biochemical Reaction Networks from a Systems Biology Perspective

Abstract

Cells constantly sense and respond to cues via signaling networks. These networks operate in dynamic systems and often serve multiple purposes, translating multiple inputs into multiple outputs depending on context. Because experimentally monitoring every system component is not feasible, computational modeling has been used as a means to probe the network and make predictions that would otherwise not be possible. However, the process of translating biological knowledge into a model and then extracting information from the model remains challenging. This work addresses several of these challenging aspects with new and improved methods and applies them to a relatively simple biochemical system consisting of the enzyme COX-2 and two of its substrates, arachidonic acid and 2-arachidonoyl glycerol, which are turned over to prostaglandin (PG) and prostaglandin glycerol (PGG), respectively. The COX-2 Reaction Model (CORM) was successfully calibrated to experimental data and physiological constraints. Fitted kinetic rates from this calibration suggested that allostery plays a key role in modulating catalysis of both PG and PGG. Further analysis using a novel pathway flux method indicated that the degree to which allosterically-modulated pathways are used to produce PG and PGG varies significantly depending on the substrate levels. Sensitivity analysis of the system suggested that the system is more sensitive to perturbations that target particular pathways when in regions of substrate space where those pathways dominant. Finally, an information theoretical analysis uncovered the key role of both AA and 2-AG in information transfer within the network, particularly when they are strongly correlated. In that case, COX-2 can serve as an integrator, transferring more information via two correlated inputs than it could by two independent inputs.