Analysis and design of metabolic reaction networks

Abstract

Different mathematical methods can be used for the analysis of metabolic systems and the subsequent engineering of metabolism. The available experimental information dictates the most appropriate mathematical framework for such studies. Several approaches for metabolic system analysis and design are developed in this thesis. It is shown that for several model systems, a (log)linear model shows excellent agreement with the corresponding nonlinear model. The (log)linear model which is developed here describes the dynamical and steady-state responses of the logarithmic deviations of the metabolic variables and functions with respect to a change of the metabolic parameters around a corresponding reference state. The parameters of the (log)linear model are quantities easily estimated from experimental and theoretical tools developed within metabolic control analysis (MCA). A significant advantage of the newly developed (log)linear model is the linearity with respect to logarithms which makes computational analysis easier as compared to the correponding nonlinear model. A second approach introduces a novel, production-oriented optimization framework. Maximizing the performance of a metabolic reaction pathway is treated as a mixed-integer linear programming (MILP) formulation when a (log)linear model of the pathway is available and as a mixed-integer nonlinear programming (MINLP) formulation when a nonlinear model is available. The objective of the MILP and MINLP formulation is to identify changes in regulatory structure and strength, and in cellular content of pertinent enzymes, which should be implemented in order to optimize a particular metabolic process. A regulatory superstructure is proposed that contains all alternative regulatory structures that can be considered for a given pathway. The proposed approach is followed in order to find the optimal regulatory structure for maximization of phenylalanine selectivity in the microbial aromatic amino acid syn thesis pathway. The solution suggests that, from the 8 feedback inhibitory loops in the original regulatory structure of this pathway, inactivation of at least three loops and overexpression of three enzymes will increase phenylalanine selectivity by 42%. Moreover, novel regulatory structures with only two loops, none of which exists in the original pathway, could result in a selectivity of up to 95%.