A Study of Dynamic Responses of E. coli Metabolism upon Large Perturbations Using Large-Scale Kinetic Models

A Study of Dynamic Responses of E. coli Metabolism Upon Large Perturbations Using Large-Scale Kinetic Models

Milenko Tokic^1,2, Georgios Fengos^1,2, Ljubisa Miskovic^1,2, and Vassily Hatzimanikatis^1,2

¹Laboratory of Computational Systems Biotechnology (LCSB), Swiss Federal Institute of Technology (EPFL), CH-1015 Lausanne, Switzerland

²Swiss Institute of Bioinformatics (SIB), CH-1015 Switzerland

In recent years, constraint-based approaches have been widely used for analyzing cellular metabolism. These approaches use stoichiometric models for the characterization of the intracellular fluxes at steady state. Nonetheless, the stoichiometric models lack information about metabolic regulation and enzyme kinetics, and they are not able to capture the dynamic features of metabolic pathways. As a result, there are ongoing efforts to build large-scale and genome-scale kinetic models as they can predict the complex dynamic responses of metabolism to environmental and genetic perturbations. The development of kinetic models is mostly hindered by structural, e.g. unknown kinetic mechanisms, and quantitative, e.g. inconsistencies in the available kinetic data, uncertainties. To overcome these difficulties, we have developed the ORACLE (Optimization and Risk Analysis of Complex Living Entities) framework that uses Monte Carlo sampling techniques to build populations of kinetic models that are consistent with the observations while satisfying the stoichiometric and thermodynamic constraints^1,2. ORACLE was initially developed within the context of Metabolic Control Analysis (MCA) to compute control coefficients despite scarce information about kinetic properties of enzymes. Recently, we extended ORACLE capabilities beyond computing control coefficients to construct populations of a large class of nonlinear models of metabolism.

In this work, we used the ORACLE framework to build a population of large-scale nonlinear models of aerobically grown wild-type E. coli. We used these models to investigate the dynamic responses of E. coli metabolism to multiple gene deletion knockouts and we compared them with the experiments³. We further compared these predictions with the predictions obtained through MCA to assess to what extent the simpler MCA predictions are able to predict the responses of metabolism to large perturbations. These results demonstrate usefulness of large-scale kinetic models for quantitative and qualitative understanding of the global regulation of the cell.

1.            Miskovic, L. & Hatzimanikatis, V. Production of biofuels and biochemicals: in need of an ORACLE. Trends in biotechnology 28,391–7 (2010).

2.            Chakrabarti, A., Miskovic, L., Soh, K. C. & Hatzimanikatis, V. Towards kinetic modeling of genome-scale metabolic networks without sacrificing stoichiometric, thermodynamic and physiological constraints. Biotechnol. J. 8,1043–57 (2013).

3.            Fong, S. S., Nanchen, A., Palsson, B. O. & Sauer, U. Latent pathway activation and increased pathway capacity enable Escherichia coli adaptation to loss of key metabolic enzymes. J. Biol. Chem. 281, 8024–33 (2006).