In silico screening of commercially available libraries for CYP2D6 binding

Abstract

Compound databases are an important resource in high throughput screening and drug discovery, especially if compounds are scored for drug-like properties and subjected to a validated in silico analysis. In this study we have used a structure-based approach that combines homology modeling of CYP2D6 and docking to correctly position first codeine (as a positive control) in the CYP binding site, and then to dock 4,207 amines identified from a 94,426-member Sigma-Aldrich Screening collection. Our database provides 3D structures and molecular properties, including pKa values (Pipeline Pilot). The Web-based interface is easy to use to view all the calculated properties, the optimized structures (AM1 calculations), the docking pose in the CYP2D6 binding site, the docking scores (X-Score and Autodock) and scores for druglikeness, base on a comparison to known drugs. This chemoinformatic analysis and database can be used by investigators searching for amine-containing compounds, to screen in drug discovery efforts, especially to avoid those compounds that are likely to bind to CYP2D6 and to design protein inhibitors more effectively. Most of the current analysis of commercial compounds is done in terms of Lipinski properties, and do not specifically address likelihood of metabolism by CYPs. As a result of this study, we have found out that only a small fraction of compounds (<10%) are predicted to have extremely high affinity for CYP2D6. The docking results presented herein are predictions that are currently being subjected to experimental verification using fluorescence-based activity and NMR-based binding assays.