System-level modeling and analysis of biological reaction networks in signal transduction and glycosylation
Novel high-throughput genomic and proteomic tools are allowing the integration of information from a range of biological assays into a single conceptual framework. This framework is often described as a network of biochemical reactions. Current strategies developed and applied to date for analyzing such biological reaction networks either are not well suited for transient processes, or computational expensive. We present a simple and systemic strategy for the analysis of complex reaction networks. Principal component analysis, based on an eigenvalueeigenvector analysis of the scaled sensitivity coefficient matrix, is applied to rank individual reactions in the network based on their effect on system output. When combined with flux analysis, sensitivity analysis allows model reduction or simplification. Using epidermal growth factor (EGF) mediated signaling and trafficking as an example of signal transduction, we demonstrate that sensitivity analysis results suggest that receptor internalization and endosomal signaling are important features regulating signal output only at low EGF dosages and at later times. Further, we extended this methodology to Tumor necrosis factor (TNF)-induced Nuclear Factor kappa B (NF-κB) signaling pathway. This pathway plays an important role in regulating cell function and human inflammatory ailments. We constructed a computational model for TNF induced signaling that couples TNF receptor activation complex formation with the IκB Kinase (IKK)-NF-κB module. Experimental data over a range of TNF dose fit the model. Sensitivity/Perturbation analysis suggests that the time-dependent change in phosphorylated IKK levels is a key parameter that regulates the period of NF-κB oscillation. Current computational tools focus on the simulation of cellular reaction networks. They place less emphasis on post-simulation analysis of data. To address this limitation, we developed a new software package called "IBRENA" which aims at the implementation of sensitivity analysis (both forward and adjoint methods) coupled with principal component analysis and singular value decomposition, and model reduction. The software features a graphical user interface that aids simulation and plotting of in silico results. Finally, we present a novel reverse engineering approach to better understand the concerted action of various glycosyltransferases in controlling the glycan structures on the mucinous protein P-selectin Glycoprotein Ligand-1 (PSGL-1). The binding of O-linked glycans decorated on the PSGL-1 to adhesion molecules belonging to the selectin family is involved in the initial step during leukocyte recruitment at sites of inflammation. In the computational model, glycosylation reactions were described as bimolecular chemical processes. Using defined starting material and glycosylation end products based on published literature, a library of reaction pathways were automatically generated. Hybrid genetic algorithm was then applied to fit the various glycosylation models to experimental data. Hierarchical clustering and principal component analysis were applied to cluster these models and to infer most probable reaction pathways and corresponding rate constants. Sensitivity analysis was performed to identify key glycosylating enzymes. Overall, this modeling approach reveals novel experimentally testable hypothesis.