Functional virtual screening of estrogen receptor a modulators by FlexX-Pharm Istva ´n Vira ´g a , Tı ´mea Polga ´r b , Gyo ¨rgy M. Keseru ˝ b, * a Eo ¨tvo ¨s Lora ´nd University, Faculty of Science, Pa ´zma ´ny Pe ´ter se ´ta ´ny 1/a, H-1117 Budapest, Hungary b CADD&HTS, Gedeon Richter Ltd, P.O. Box 27, H-1475 Budapest, Hungary Received 13 December 2004; accepted 21 February 2005 Abstract A functional virtual screen was developed using FlexX-Pharm/ChemScore that separates ligands taking advantage of the different binding modes between estrogen receptor alpha agonists and antagonists. Increased performance of FlexX-Pharm over FlexX was also demonstrated on the estrogen receptor. q 2005 Elsevier B.V. All rights reserved. Keywords: Estrogen receptor; FlexX-Pharm; Virtual screening High-throughput docking and scoring methods can be applied to perform structure-based screening. Several docking programs that use various conceptual strategies have already been on the market. A collection of prominent algorithms has been compiled by van Leeuwen [1]. One of the most widely used programs is FlexX [2] that can be used for protein–ligand docking following an incremental construction algorithm while the receptor is kept rigid. A selected base fragment of the ligand is first placed into the active site using a pattern recognition technique, called pose clustering. Next, the remainder of the ligand is built up incrementally from other fragments. The conformational flexibility of the ligand is included by generating multiple conformations for each fragment. Placement of the ligand is scored by FlexX score on the basis of protein–ligand interactions that includes pairwise assignments of inter- action geometries. Finally, the binding energy is estimated by one of the scoring functions available in CScore (Dock [3a–c], Chem [3d], PMF [3e,f], Gold [3g,h] and FlexX [3i–k] scores), and placements are ranked. In FlexX-Pharm [4], which is an extended version of FlexX, a previously defined set of pharmacophore features in the active site constrains the docking calculation so that only solutions are produced that match the specified set of features. Herein, we tested the performance of FlexX-Pharm comparing with FlexX alone on ERa (estrogen receptor alpha), a target prototype in virtual screening. Quantitative predictions of binding affinities were first evaluated by a correlation study for 23 ERa antagonists. Next, a functional virtual screen was developed to differentiate agonists from antagonists. Crystal structures of the ERa ligand binding domain (LBD) were used to screen a benchmark library of known agonists and antagonists. ERa-LBD bound to the agonists (R,R)-5,11-cis-diethyl-5,6,11,12-tetrahydrocrysene-2,8-diol (THC), 17b-estradiol (E2) (PDB codes 1L2I [5] (1.95 A ˚ ) and 1GWR [6,7] (2.4 A ˚ ), respectively) and to antagonist 4-hydroxy-tamoxifen (TAM), (3ERT [8] (1.9 A ˚ )) were considered. For docking with FlexX and FlexX-Pharm active sites of each A-chain of three targets were defined as all the residues within the 6.5 A ˚ of the bound ligand and one conserved water molecule was added [9]. This structural water is adjacent to Glu353, Arg394 residues and the hydroxyl group of the ligand bound. For docking with FlexX-Pharm the most critical interactions between ligands and ERa i.e. H-bridges to Glu353, Arg394 and His524 were mapped in the crystal structures [10]. Hydrophobic contacts are mainly located at the A and D rings of E2 in 1L2I structure [7]. Antagonists bind at the same site as agonists within Journal of Molecular Structure: THEOCHEM 725 (2005) 239–242 www.elsevier.com/locate/theochem 0166-1280/$ - see front matter q 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.theochem.2005.02.083 * Corresponding author. Tel.: C36 431 4605; fax: C36 1 432 6002. E-mail address: gy.keseru@richter.hu (G.M. Keseru ˝).