de.mpg.escidoc.pubman.appbase.FacesBean
English
 
Help Guide Disclaimer Contact us Login
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT

Released

Poster

WPT7: Model Validation, Parameter Analysis and Experimental Design. Part I: Parameter estimation and parameter confidence intervalls. Part II: Key enzyme activities in central carbon metabolism of mammalian cells.

MPS-Authors
http://pubman.mpdl.mpg.de/cone/persons/resource/persons86195

Kremling,  A.
Systems Biology, Max Planck Institute for Dynamics of Complex Technical Systems, Max Planck Society;

http://pubman.mpdl.mpg.de/cone/persons/resource/persons86337

Janke,  R.
Bioprocess Engineering, Max Planck Institute for Dynamics of Complex Technical Systems, Max Planck Society;

http://pubman.mpdl.mpg.de/cone/persons/resource/persons86303

Genzel,  Y.
Bioprocess Engineering, Max Planck Institute for Dynamics of Complex Technical Systems, Max Planck Society;

http://pubman.mpdl.mpg.de/cone/persons/resource/persons86512

Wahl,  A.
Bioprocess Engineering, Max Planck Institute for Dynamics of Complex Technical Systems, Max Planck Society;

http://pubman.mpdl.mpg.de/cone/persons/resource/persons86448

Reichl,  U.
Otto-von-Guericke-Universität Magdeburg;
Bioprocess Engineering, Max Planck Institute for Dynamics of Complex Technical Systems, Max Planck Society;

Locator
There are no locators available
Fulltext (public)
There are no public fulltexts available
Supplementary Material (public)
There is no public supplementary material available
Citation

Kremling, A., Janke, R., Genzel, Y., Wahl, A., & Reichl, U. (2009). WPT7: Model Validation, Parameter Analysis and Experimental Design. Part I: Parameter estimation and parameter confidence intervalls. Part II: Key enzyme activities in central carbon metabolism of mammalian cells. Poster presented at Evaluation and Status Seminar FORSYS 2009, Heidelberg, Germany.


Cite as: http://hdl.handle.net/11858/00-001M-0000-0013-91D7-7
Abstract
Part 1: The quantitative analysis of models for cellular systems requires the knowledge about kinetic parameters. In general, parameters are estimated based on time course data and are finally analyzed. In the first part of the project (Kremling) methods to improve parameter estimation itself as well as methods to improve the determination of parameter uncertainties are under development. To improve the estimation of parameter uncertainties two new methodsare applied. The second part of this project (Wahl, Reichl) focuses on the establishment of a model of the central metabolism of adherent mammalian cells (MDCK), used in commercial influenza vaccine manufacturing. As model validation relies on a comprehensive set of experimental data several assays are being established. Methods established in the first part of the project will be used for estimation of quality of all relevant model parameters and for experimental design for further model validation. The aim of the project is the improvement and the application of methods (a) to estimate kinetic parameters and (b) to determine confidence regions for kinetic parameters. For parameter estimation, the model is reformulated as a mixed integer model where the integer variables are used to discriminate different model variants. Therefore, it is possible to estimate the kinetic parameters as well as part of the underlying model structure. The method is developed by J. Banga in Vigo, Spain (cooperation partner in a SysMO project) and was applied here for two small networks in Escherichia coli: The potassium uptake system KdpFABC that is controlled by the two-component system KdpDE, and the lactose uptake system. For both systems new hypotheses could be generated by the procedure that currently are under investigation from an experimental point of view (WPB1, Bettenbrock). In previous publications it could be shown that the application of classical tools like the Fisher-Information-Matrix (FIM) leads to smaller values of the parameter confidence region in comparison to Monte-Carlo simulations (bootstrap approach). The correct prediction of the intervals is a prerequisite for planning new experiments. It also turns out that the direct application of the bootstrap procedure for experimental design is not possible due to the long computational time. During the last two years, two new approaches become important and were investigated to improve the calculation of the parameter confidence region: (i) The confidence regions are calculated with a statistical linearization. This method – also known as the sigma point method – was successfully applied in control theory and was now adapted for parameter estimation problems. The method leads to comparable results than the bootstrap approach but is clearly faster. With the method also the determination of confidence regions of the state variables could be improved. (ii) The second method explores the observation that the nonlinearity of the systems is mainly due to the parametric and not due to the intrinsic nonlinearity. This allows mapping all points of the solution locus to the corresponding tangent. Since the measurement uncertainties in this coordination system are represented by simple geometrical objects (circle in 2D, sphere in 3D), the intersection of the tangent with this geometry can be calculated simply. The respective intersection points represent the entire parameter confidence region. Currently, only simple examples are considered and compared with other methods. In the future, the methods under investigation will be refined and will be applied for experimental design studies in the various subprojects (WPB1, Bettenbrock, WPT7, part 2, Reichl). Literature: [1] Rehberg, M.. Model selection and parameter identification using mixed integer nonlinear programming (MINLP): a case study of the KdpD/KdpE system in Escherichia coli. Studienarbeit an der Otto-von-Guericke Universität Magdeburg, 2008 [2] Grüning, R.. Untersuchung der Antwortsignale bei Induktion des lac Operons mittels Lakotse, IPTG und TMG. Studienarbeit an der Otto-von-Guericke Universität Magdeburg, 2008 [3] R. Schenkendorf et al.. Optimal experimental design with the sigma point method. IET Systems Biology 3, 2009 Part 2: In order to apply a multi-scale approach to model growth behavior and metabolism of mammalian cells three important steps were of first relevance: 1) generation of a stoichiometric model for the central carbon metabolism, 2) set-up of analytical tools to measure fluxes, metabolite concentrations and enzyme activities, and 3) model validation and parameter analysis. Based on extensive sets of experimental data a stoichiometric model for an adherent MDCK cell line was established to determine intracellular flux distribution of the central energy and carbon metabolism under different growth conditions. The formulation of this model as well as theoretical considerations concerning flux distributions obtained for growth in different cultivation media was published (Wahl et al. 2008, Sidorenko et al. 2008). Furthermore, a high-throughput platform to determine key enzyme activities of the central carbon metabolism in plant cells has been adapted to monitor growth of adherent mammalian cells. The use of a set of enzymatic cycling assays provides high sensitivity to measure low amounts of product formed or substrates consumed. The platform allows the measurement of 18 enzymes in MDCK cells with protocols established at the MPI of Molecular Plant Physiology in Golm, and has been extended for 10 additional enzymes, relevant in mammalian cell metabolism. Currently, 28 different enzymes of the central carbon metabolism (glycolysis, gluconeogenesis, tricarboxylic acid cycle, glutaminolysis, and pentose phosphate pathway) can be measured in extracts of MDCK cells. From experiments of the stationary growth phase of MDCK cells grown under different cultivation conditions in 6-well plates, significant changes in maximum enzyme activities could be observed. The described differences in metabolism between two media, namely glutamine- and pyruvate-containing GMEM medium, (Wahl et al. 2008, Sidorenko et al. 2008) were confirmed by enzyme activity measurements. It could be seen that pyruvate dehydrogenase and glutamine synthetase were up-regulated. In contrast, the activities of ATP-citrate lyase, phosphoenolpyruvate carboxykinase and the glutaminolytic enzymes aspartate- and alanine transaminase were decreased in pyruvate-containing medium. Less pronounced activity changes were found for lactate- and malate dehydrogenase. Under all cultivation conditions the activities of glycolytic enzymes hexokinase and phosphofructokinase were relatively low, whereas the maximum enzyme activities of pyruvate kinase and lactate dehydrogenase were comparatively high. Additionally, the measurement of intracellular concentrations of corresponding metabolites has been set-up, resulting in extensive data sets for different process conditions (Ritter et al. 2008). Literature: Wahl, A.; Sidorenko, Y.; Dauner, M.; Genzel, Y. & Reichl, U. Metabolic model for an anchorage-dependent MDCK cell line. Characteristic Growth Phases and Minimum Substrate Consumption Flux Distribution. Biotechnol Bioeng, 2008. 101(1), 135-152. Sidorenko, Y.; Wahl, A.; Dauner, M.; Genzel, Y. & Reichl, U. Metabolic Flux Analysis of MDCK cell growth in glutamine-containing and glutamine-free medium. Biotechnology Progress, 2008. 24(2), 311-320. Ritter, JB.; Genzel, Y. & Reichl, U. Simultaneous extraction of several metabolites of energy metabolism and related substances in mammalian cells: optimization using experimental design. Anal Biochem. 2008. 373(2),349-69.