Computational modeling of methionine cycle-based metabolism and DNA methylation and the implications for anti-cancer drug response prediction

// Mengying Zhang 1 , Christian Saad 4 , Lien Le 1 , Kathrin Halfter 1 , Bernhard Bauer 4 , Ulrich R. Mansmann 1, 2 and Jian Li 1, 2, 3 1 Institute for Medical Informatics, Biometry and Epidemiology, Ludwig-Maximilians University of Munchen, Munich, Germany 2 German Cancer Consortium (DKTK), Heidelberg, Germany 3 German Cancer Research Center (DKFZ), Heidelberg, Germany 4 Department of Computational Science, University of Augsburg, Augsburg, Germany Correspondence to: Jian Li, email: lijian@ibe.med.uni-muenchen.de Keywords: metabolism; methylation; chemotherapy; molecular modelling; treatment prediction Received: July 06, 2017     Accepted: July 29, 2017     Epub: February 21, 2018     Published: April 27, 2018 ABSTRACT The relationship between metabolism and methylation is considered to be an important aspect of cancer development and drug efficacy. However, it remains poorly defined how to apply this aspect to improve preclinical disease characterization and clinical treatment outcome. Using available molecular information from Kyoto Encyclopedia of Genes and Genomes (KEGG) and literature, we constructed a large-scale knowledge-based metabolic in silico model. For the purpose of model validation, we applied data from the Cancer Cell Line Encyclopedia (CCLE) to investigate computationally the impact of metabolism on chemotherapy efficacy. In our model, different metabolic components such as MAT2A, ATP6V0E1, NNMT involved in methionine cycle correlate with biologically measured chemotherapy outcome (IC50) that are in agreement with findings of independent studies. These proteins are potentially also involved in cellular methylation processes. In addition, several components such as 3,4-dihydoxymandelate, PAPSS2, UPP1 from metabolic pathways involved in the production of purine and pyrimidine correlate with IC50. This study clearly demonstrates that complex computational approaches can reflect findings of biological experiments. This demonstrates their high potential to grasp complex issues within systems medicine such as response prediction, biomarker identification using available data resources.