Ligand protonation states and stereoisomers in virtual screening

Abstract

Ligand structure preparation is an essential step during thesetup of virtual screening (VS) experiments. In most casesonly the crystal structure of a protein-ligand complex isknown. Thus, the protonation of the ligand and the bind-ing site is largely unknown and often only limited infor-mation about hybridization and connectivity in theligand structure is provided by the pdb [1]. While manualpreparation of the ligand structures is the most accuratemethod, it is by far too time consuming for bigger data-sets. In VS experiments the large number of structures,which are often obtained from different sources, leads toadditional problems. A consistent treatment of all activeand inactive structures is needed to prevent a preferentialtreatment of some of the structures, which could lead toartificial enrichments. Additionally, changes in the proto-nation of the ligand (and the protein), when binding tothe active site, have to be considered. The large amount ofinactive ligand structures are usually taken from organicmolecule data banks like ZINC [2]. In this case, the hydro-gen atoms and hybridization are usually known but thedata banks often provide only one tautomeric form orstereoisomer for each molecule. Because different pro-tomers/stereoisomers can lead to huge differences inaffinity the other stereoisomers of selected ligands shouldbe tested too.To ensure an equal treatment of all structures for VS exper-iment we present an automated procedure called structureprotonation and recognition system (SPORES) for thesetup of VS datasets. It can be used to protonate structuresfrom pdb files and to generate different protonationstates, tautomers and stereoisomers. It is based on 3Dcoordinates only and does not use information about thebinding site for ligand preparation or information aboutactive ligands for the setup of the protein binding site. Theinfluence of ligand protonation and stereoisomers on thedocking results with PLANTS [3] and Gold [4] was firsttested on the well-defined ASTEX clean dataset [5]. After-ward several VS experiments on different target were con-ducted with PLANTS, in which the influence of ligandprotonation states and stereoisomers on the enrichmentwas tested.