Topological metabolic analysis: A robust optimization-based framework for metabolic process systems engineering

We introduce a new optimization-based modeling framework for the analysis of metabolic networks of arbitrary size and complexity. This framework derives from the state-space approaches for modeling of chemical process networks originally developed by Manousiouthakis and co-workers. Because a defining aspect of this approach is a comprehensive mathematical treatment of network topology, we name this new modeling technique topological metabolic analysis (TMA).;TMA defines a set of material-balance constraints describing the fundamental ways in which material (metabolites) may be distributed and transformed within a metabolic network whose reactions have known (or assumed) stoichiometries. The manner in which these balances are constructed not only permits quantitative modeling of reaction rates and the overall metabolite uptake and secretion rates, but also of the manner by which every network metabolite is distributed among and shared between each network reaction. This constraint model, like existing techniques, is consistently underdetermined and therefore must be paired with an objective function to identify a particular solution.;We therefore design a generalized quadratic "aggregate" objective function (AOF) offering a number of unique mathematical advantages. First, it does not require that any particular amount of experimental data be known in advance, enabling metabolic modeling even under circumstances where experimental data is scarce. Second, the convexity (or, more-precisely, semi-convexity) of the objective enables the identification and characterization of a potentially infinite number of topologically distinct network configurations, all of which are globally optimal solutions to the given modeling problem. Careful characterization of this family of solutions can offer insights into the robustness of the metabolic network.;We initially demonstrate the utility of our TMA modeling framework using a case study of bacterial metabolism (Actinobacillus succinogenes ). We show that our TMA framework, using only external metabolite uptake and secretion measurements, identifies modeling solutions that both offer useful biological insight and compare favorably against solutions obtained using much less-convenient 13C isotope tracing measurement.;We then show how TMA can be used to interrogate metabolic networks in ways not previously demonstrated using classical stoichiometric modeling methods. Employing a case-study of hybridoma metabolism, we examine the topology of a sample metabolic network, and show that multiple topologically-distinct network configurations (each of which is equivalently optimal in reproducing experimental observations) can be identified using TMA. We further prove that these alternate network configurations are not mathematically distinguishable using existing stoichiometric modeling approaches.;We then apply our TMA modeling framework to characterize the differential metabolism of three Chinese Hamster Ovary (CHO) clonal variant cell lines experimentally known to exhibit substantial differences in nutrient uptake and secretion, growth rate, biomass composition, and productivity of a recombinant humanized IgG antibody product. Since no suitable metabolic reconstruction for this species is yet available, we develop our own comprehensive metabolic reconstruction based upon available genomic data for Mus musculus. Using TMA, we accurately model the experimentally observed behavior of the three variant CHO cell lines..;Finally, we explore a unique application of our TMA framework; the rational design of an aerobic microbial fuel cell. Using a basic metabolic reconstruction describing the growth of a model microbe (Pseudomonas putida) within such a fuel cell, we evaluate a number of both aerobic and anaerobic operating conditions utilizing variety of substrates. We ultimately show that coulombically efficient aerobic operation of such a device can, in theory, be made possible through the inactivation of electron-transport chain complex. (Abstract shortened by UMI.)...