Andreas Drager, Ute Hofmann, Roland Keller, Stephanie Tscherneck, Benjamin Kan-del, Maria Thomas, Markus Klein, Ulrich M. Zanger and Andreas Zell1
Accepted Abstracts: Drug Design
Question: The cholesterol lowering drug atorvastatin (Lipitor) acts by inhibiting the enzyme HMG-CoA reductase, which plays a central role in the production of cholesterol in the liver. To investigate the effects of atorvastatin on the central carbon metabolism we used rat hepatocytes as a model system. In order to obtain a predictive model of the overall processes, the isotopomer model by Maier et al . (2008) had to be complemented with information from the recent structural liver model HepatoNet1 (Gille et al .2010, Mol Sys Biol) and further extended to also cover the bile acid metabolism. To facilitate the model building, simulation and parameter estimation, a specialized software workflow had to be developed. Methods: Primary rat hepatocytes were treated with different concentrations of atorvastatin. Extracellular and intracellular samples were collected over a time span of 28 h (Hofmann et al . 2008, Biotechnol Bioeng, Maier et al . 2008, Metab Eng). We constructed a sub-model of HepatoNet1 comprising the central carbon metabolism and the cholesterol synthesis (using CellDesigner) to complement the model by Maier et al. (2008). The structure of the model was validated by performing a mass balance check and flux bal-ance analysis (using FASIMU). A new version of SBMLsqueezer (Dräger et al . 2008, BMC Sys Biol) has been used to generate generalized mass action rate laws for all reactions within the model. The newly developed program SBMLsimulator was used to estimate the values of all uncertain quantities of the model. To this end, it uses differential evolution with the relative Euclidean distance as fitness function. The numerical calculation was done with Rosenbrock?s method. Results: The new model contains 84 reactive species participating in 61 reactions, located in the intra- and extracellular compartments. All measured steady-state fluxes could be approximated by the model in the flux balance analysis. The new version of SBMLsqueezer correclty derived all kinetic equations and auto-matically derived the units all parameters. The resulting ODE system contains 150 uncertain values (including kinetic constants and initial concentrations of 18 metabolites), which have successfully been fitted to the experimental data using SBMLsimulator, a fully-featured and efficient implementation for the modeling language SBML. Hence, the model reproduces the observed states of the system. Conclusion: With the help of our new model building and simulation pipeline a comprehensive model could be created that links together the central carbon metabolism and the cholesterol synthesis influenced by the drug atorvastatin. The result of this study is an optimized ODE system, which reproduces the ob-served effects of this important cholesterol lowering drug. This work demonstrates the usefulness, correct-ness and efficientcy of the newly developed software tools SBMLsqueezer and SBMLsimulator.
In 1999 Andreas finished his A-levels with honors and started his studies of bioinformatics at the Martin Luther University Halle-Wittenberg in 2000. He worked as an intern at the Max Planck Institute, Berlin, in 2001 and the University of Illinois at Chicago in 2004. After achieving his Diploma in 2006, he became a Ph.D student in the group of Prof. Zell at the University of Tuebingen. Since his Ph.D time he published or co-authored 19 scientific papers, gave talks on various international conferences, and spent three month at Keio University in Yokohama, Japan. Since January 2011 Andreas Drager works as a post-doc and junior group leader at the University of Tuebingen, Germany.