Prediction on the risk population of idiosyncratic adverse reactions based on molecular docking with mutant proteins
Hongbo Xie, Diheng Zeng, Xiujie Chen, Diwei Huo, Lei Liu, Denan Zhang, Qing Jin, Kehui Ke, Ming Hu · Oncotarget · 2017
// Hongbo Xie 1, * , Diheng Zeng 1, * , Xiujie Chen 1, * , Diwei Huo 2, * , Lei Liu 1 , Denan Zhang 1 , Qing Jin 1 , Kehui Ke 1 and Ming Hu 1 1 Department of Pharmacogenomics, College of Bioinformatics Science and Technology, Harbin Medical University, Harbin, Heilongjiang 150081, PR China 2 The 2nd Affiliated Hospital of Harbin Medical University, Harbin, Heilongjiang 150081, PR China * These authors contributed equally to this work Correspondence to: Xiujie Chen, email: [email protected] Denan Zhang, email: [email protected] Keywords: drug-induced liver injury, personalized medicine, homology-modeling, molecular simulation, risk population prediction Received: July 06, 2017 Accepted: September 20, 2017 Published: October 05, 2017 ABSTRACT Idiosyncratic adverse drug reactions are drug reactions that occur rarely and unpredictably among the population. These reactions often occur after a drug is marketed, which means that they are strongly related to the genotype of the population. The prediction of such adverse reactions is a major challenge because of the lack of appropriate test models during the drug development process. In this study, we chose withdrawn drugs because the reasons why they were withdrawn and from which countries or regions is easily obtained. We selected Dilevalol and its chiral drug (Labetalol) as the investigatory drugs, as they have been withdrawn from a European market (Britain) because of serious hepatotoxicity. First, we searched for and obtained the Dilevalol-induced- liver-injury related protein, multidrug resistance protein 1 (MDR1), from the Comparative Toxicogenomics Database (CTD). Then, we searched and extracted 477 non-synonymous single nucleotide polymorphisms (nsSNP) on MDR1 in the dbSNP database. Second, we used the VarMod tool to predict the functional changes of MDR1 induced by these nsSNPs, from which we extracted the nsSNPs that significantly change the functions of this protein. Third, we built the three-dimensional structures of those variant proteins and used AutoDock to perform a docking study, choosing the best model to determine the sites of nsSNPs. Finally, we used the data from the 1000 Genomes Project to verify the dominant population distribution of the risk SNP. We applied the same strategy to the post-marketing drug-induced liver injury drugs to further test the feasibility of our method.