Hydrophobicity is an important factor in molecular recognition [1-4] and the accurate prediction of the binding modes of nonpolar molecules to proteins in aqueous solvent is useful for ligand docking and drug design [5]. We have recently developed an approach to calculate and visualize the hydrophobicity at the surface of a protein (hydrophobicity maps) [6]. It is based on the evaluation of the nonpolar energy and electrostatic desolvation of the receptor with a continuum model. These energy contributions determine the binding modes of a nonpolar compound to the hydrophobic surface regions of a re

* To whom correspondence should be addressed. E-mail: [email protected]

ceptor. Electrostatic interactions between ligand and receptor do not play a significant role in the case of a nonpolar ligand.

Computer programs for structure-based ligand design are useful tools for de novo design and lead modification [7-10]. The combinatorial strategy chosen for structure-based ligand design consists of three parts: the docking ofmolecular fragments, the connection ofthe docked fragments by combinatorial principles to generate candidate ligands, and the estimation of (relative) binding affinities [11]. The docking approach implemented in the program SEED determines optimal positions and orientations of small to medium-size molecular fragments in the binding site of an enzyme or receptor [12]. SEED docks polar fragments so that at least one hydrogen bond with optimal distance to a protein polar group is made. The hydrophobicity maps are used for the docking of apolar fragments. Our numerical continuum electrostatic methodology [13,14] and ad hoc look-up tables are employed to efficiently evaluate the protein and fragment desolvation upon binding and the screened electrostatic interaction. The fragments are then connected in a combinatorial way by the program CCLD [15]. For the final evaluation of candidate ligands, which is not discussed in this article, one could use a multi-layer scoring system that utilizes more than one binding affinity estimation method [16].

A recent review on computational approaches for drug design contains a detailed discussion of QSAR methods, structure-based docking and design programs, as well as a complete recapitulation of the available techniques to estimate (relative) binding affinities, i.e., from knowledge-based scoring functions to molecular dynamics-based free energy techniques [17]. The reader, interested in implicit solvation models is referred to two recent and comprehensive reviews [18,19],

0 0

Post a comment