Protein kinases represent interesting targets for drug discovery. However, development of selective inhibitors for these targets is a challenging task due to high physicochemical similarity of the ATP binding site. In this talk, we present results from a systematic analysis of interactions between ligand and the ATP binding site residues and describe role of these interactions on compound selectivity. This analysis is performed using a large dataset consisting of over 1000 compounds from our in-house collection for which crystal structures in complex with kinase targets as well as panel screening data are available. Using a consistent numbering scheme to represent the ATP binding site residues across all kinase targets, protein-ligand interactions are analyzed using contact surface areas between the bound ligand and each binding site residues. Role of individual amino acid contacts in imparting compound selectivity and promiscuity has been assessed by analyzing the trends between contact surface areas and selectivity. Further insights into the structural basis of selectivity is obtained by modeling this dataset using machine learning techniques such as Random Forest and Support Vector Machine and using the contact surface areas as the input features. We find that polar interactions between ligand and the binding site residues provide little gain in the selectivity. Whereas, lipophilic contacts between ligand and several key amino acids in the ATP pocket significantly improve compounds selectivity. In addition, selectivity is found to be driven, to a large extent, by the unique binding site conformations that a target is able to adopt in complex with the bound ligand.

Selectivity determinants for protein kinase inhibitors from a systematic analysis and modeling of large-scale structural and panel screening data

243rd National Meeting (2012)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.