Identification of molecular targets in anti-cancer therapy using protein-ligand docking

The starting point of this work is a ligand set (natural compounds) which present an anti-cancer activity in experimental cellular assays, but it is unknown which specific enzymes they target. To determine these molecular targets we selected 30 enzymes used commonly as molecular targets in anti-cancer therapy and present in the different cell cycle stages, and studied the interactions between these two sets with molecular docking using the program eHiTS (electronic High Throughput Screening). In order to increase the accuracy of ligand ranking we used a modified score for each molecule by calculation of multiple active site correction (MASC) scores[1]. Additionally, we also verified the specificity of the compounds towards AURKA enzyme and a kinase family set, and we established that compounds present specificity 100 higher towards AURKA enzyme than for kinase family members tested. The screening results of a database of compounds deemed as putative non-binders spiked with some known binders produced enrichment factors between seven and ten times[2]. Computational and experimental data were correlated and we obtained correlation coefficients (R²) between 0.6081 and 0.9559. The results allowed us to predict the molecular targets for some of the compounds and these predictions could be verified by experimental enzymatic assays.

Acknowledgements: Financial support from Fundação para a Ciência Tecnologia (FCT), Post-Doc Grant SFRH/BPD/20716/2004 (Visvaldas Kairys).

References: 1. Vigers, G. et al. (2004)J. Med. Chem. 47:80–89.

2. Huang, N. et al. (2006)J. Med. Chem. 49:6789–6801.