Genomic Biochemical Engineering | | Cell Culture Engineering | Metabolic Pathway Engineering | Liver Cell Self Assembly | Analysis of Bioreaction Network | Stem Cell Culture Engineering | Image Processing in Fluorescence Microscopy | Bioartificial Liver | MAIN RESEARCH PAGE

Harnessing the capability of biological systems for biotechnological applications used to entail the understanding and manipulation of one of a few metabolic pathways. The availability of the genomic tools in the past few years has completely changed the landscape of biochemical engineering research. The strategy of examining one or a few genes at a time to unmask biological mechanisms is giving way to new tools that generate immensely large datasets. Genomics offers the engineering and mathematical sciences a new frontier, and with it the challenges it sets that require new concepts and new tools. Employing DNA microarrays, or enzyme arrays, we now have access to the information on thousands or even tens of thousands of genes whose expression profiles eventually give rise to the cellular physiology. The expression of regulatory genes, the modulation of enzyme concentration through transcriptional and translation regulation, and the fluxer through biochemical reactions form multiple layers of complex reaction networks that are interconnected in multiple dimensions. It is the interaction of those reaction networks that is manifested as "physiology". The immensity of the reaction network conferred by the examination of even a small segment of the genome awes that of any existing large-scale reaction network. Assessing the performance of the cellular system through the evaluation of the underlying reaction network is a tremendous challenge. The number of equations involved, the scarcity of experimental data for the parameters involved, and the non-linearity of the system, all make the task of evaluating the system dynamics extremely difficult. We employ a systematic non-linear model reduction approach based on time-scale analysis to capture the dominant dynamics of biochemical reaction networks with a minimal number of kinetic parameters.

The model system we have chosen to evaluate is the metabolism of mammalian cells in culture. Mammalian cells have a vast capacity to metabolize nutrients differently; some metabolic states give higher productivity than others. Modeling cell metabolism will provide insight into a better control of cell metabolism in culture. To obtain a reduced order description of mammalian cell metabolism, we are applying our adaptation of the reduction methods developed by Vora and Daoutidis for general non-isothermal reaction systems which exploits the connection between singularly perturbed models with non-explicit separation of fast and slow states and differential algebraic systems of high index. This method allows the systematic identification of independent quasi-steady state constraints for the fast reactions and the derivation of a non-stiff model of the slow dynamics. When applied to isothermal reaction systems, the method leads itself naturally to a change of variables that separates the fast and slow state variables.

We consider metabolic networks with fast and slow reactions and address the derivation of non-stiff nonlinear models of the slow dynamics of such networks. We begin with a scaling, which brings the kinetic model of the network in a singularly perturbed form; we then employ singular perturbation arguments to identify reaction equilibrium or complete conversion constraints for stoichiometrically independent fast reversible and irreversible reactions. Furthermore, we derive nonlinear ODE models of the slow dynamics, which do not contain large reaction rate constants. We also determine an explicit change of variables that yields linear combinations of metabolite concentrations whose dynamics evolve only in the slow time scale. The method in applied to two systems: central carbon metabolism in human erythrocytes and energy metabolism in Saccharomyces cerevisiae. The results show that the simplification of the model does not imply a significant loss of information in the dynamics of the system.

Publications:

1. Gerdtzen, Z. P., Daoutidis, P. & Hu, W.-S., Nonlinear model reduction of metabolic networks using time-scale analysis, in: American Control Conference 2867-2872 (Anchorage, Alaska, May 8-10, 2002).

2. Gerdtzen, Z. P., Daoutidis, P. & Hu, W.-S., Non-linear model reduction for metabolic networks with multiple time-scales, in: Conference on Decision and Control (Maui, Hawai (submitted), 2003).

3. Gerdtzen, Z. P., Daoutidis, P. & Hu, W.-S, 2003, Nonlinear reduction for kinetic models of metabolic reaction networks, Met. Eng (submitted).

Team Member: Ziomara Gerdtzen

Collaborator: Prodromos Daoutidis