Computational Tool for Fast in silico Evaluation of hERG K+ Channel Affinity

Abstract

The development of a novel comprehensive approach for the prediction of <i>h</i>ERG activity is herein presented. Software Phase has been used to derive a 3D-QSAR model, employing as alignment rule a common pharmacophore built on a subset of 22 highly active compounds (threshold <i>Ki</i>: 50 nM) against <i>h</i>ERG K⁺ channel. Five features comprised the pharmacophore: two aromatic rings (R₁ and R₂), one hydrogen-bond acceptor (A), one hydrophobic site (H), and one positive ionizable function (P). The sequential 3D-QSAR model developed with a set of 421 compounds (randomly divided in training and test set) yielded a test set (<i>Q</i>²) = 0.802 and proved to be predictive with respect to an external test set of 309 compounds that were not used to generate the model ([Formula: see text] = 0.860). Furthermore, the model was submitted to an <i>in silico</i> validation for assessing the reliability of the approach, by applying a decoys set, evaluating the Güner and Henry score (<i>GH</i>) and the Enrichment Factor (<i>EF</i>), and by using the ROC curve analysis. The outcome demonstrated the high predictive power of the inclusive 3D-QSAR model developed for the <i>h</i>ERG K⁺ channel blockers, confirming the fundamental validity of the chosen approach for obtaining a fast proprietary cardiotoxicity predictive tool to be employed for rationally designing compounds with reduced <i>h</i>ERG K⁺ channel activity at the early steps of the drug discovery trajectory.