In silico prediction of Heterocyclic Aromatic Amines metabolism susceptible to form DNA adducts in humans

Abstract

Heterocyclic Aromatic Amines (HAAs) are environmental and food contaminants that are classified as probable or possible carcinogens by the International Agency for Research on Cancer. Thirty different HAAs have been identified. However the metabolism of only three of them have been fully characterized in human hepatocytes: AαC (2-amino-9H-pyrido[2,3-b]indole), MeIQx (2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline) and PhIP (2-amino-1-methyl-6-phenyl-imidazo[4,5-b]pyridine). In this study, we use an integrative approach to accurately predict the biotransformation of 30 HAAs into DNA reactive and non DNA reactive compounds. We first build predicted metabolites networks by iterating a knowledge-based expert system of prediction of metabolic reactions based on fingerprint similarities. Next, we combine several methods for predicting Sites Of Metabolism (SOM) in order to reduce the metabolite reaction graphs and to predict the metabolites reactive with DNA. We validate the method by comparing the experimental versus predicted data for the known AαC, MeIQx and PhIP metabolism. 28 of the 30 experimentally determined metabolites are well predicted and 9 of the 10 metabolites known to form DNA adducts are predicted with a high probability to be reactive with DNA. Applying our approach to the 27 unknown HAAs, we generate maps for the metabolic biotransformation of each HAA, including new metabolites with a high-predicted DNA reactivity, which can be further explored through an user-friendly and interactive web interface.