Development of computational tools for metabolic flux analysis: parallel-labeling experiments and dynamic metabolic flux analysis

The central objective of metabolic engineering is the directed improvement of cellular properties of microbes through modification of specific biochemical reactions and regulatory mechanisms using modern molecular biology techniques. To better understand biological systems and improve predictions of desirable modifications, a variety of techniques have been developed which focus on the analysis of biological systems throughout various stages of phenotype expression. One technique of particular interest, fluxomics, focuses on the quantification of metabolic reaction rates (fluxes) under specific experimental conditions. Metabolic fluxes are a direct measure of the phenotype of a biological system and thus provide a wealth of information regarding how an organism responds to its experimental environment. One challenge of fluxomics is that flux profiles are not directly measureable but instead must be inferred from other measureable quantities using carefully designed experiments in combination with computational methods. In this work, I present novel computational methods for metabolic flux analysis (MFA). First, I present a method to estimate intracellular fluxes from dynamic external measurements at metabolic non-steady state. This method relies on describing intracellular fluxes as piecewise linear functions of time and allows dynamic flux profiles to be quickly estimated from typical experimental measurements. Additional development of this method allows the fluxes to be defined as biomass specific piecewise linear functions providing a higher order response which more accurately represents typical fermentation profiles. The dynamic MFA (DMFA) methodology along with a graphical user interface for its implementation are demonstrated using results from simulated batch and fed-batch fermentations. In addition to techniques for DMFA, this work also presents novel developments in metabolic flux analysis using 13C-labeling experiments. A computational method was developed to validate several assumptions of 13C-MFA. This methodology was demonstrated using parallel labeling experiments of 0 to 100% [U-13C]glucose and identified an important transport process in E. coli that has a measureable effect on flux estimates which has been neglected in the literature until now. Finally, the resulting fully validated reaction network was used to obtain the most precise intracellular flux estimates of E. coli to date through simultaneous analysis of six parallel experiments using single-labeled glucose tracers.