March 5, 2003:

"Estimation of Ligand-Receptor Binding Energies"

Professor Stefan Balaz
Department of Pharmaceutical Sciences
North Dakota State University Understanding of specific binding of low-molecular-weight ligands (up to 1,000 Da) to macromolecules is important for several disciplines and areas including the design of new drugs and other bioactive compounds. Binding affinities can be estimated whether the structure of the binding site is known or not. In the latter case, experimental binding energies and ligand structures are used to infer the structure of the binding site and to construct a spatial quantitative structure-activity relationship (3D-QSAR). Contributions from my laboratory to the development of both receptor-based and ligand-based methods will be reviewed.

Ligands are often present in the receptor surroundings as several molecular species, e.g. molecules ionized to various degrees or tautomers that bind to receptors with different affinities. Moreover, each molecular species can bind in the receptor cavity simultaneously in different orientations or conformations. Both cases are referred to as 'multiple binding modes'. Usually, the overall ligand-receptor association constant Ki is only determined and the binding affinities of individual bound species are not analyzed.

Affinity estimation methods frequently consider one mode per ligand. Inclusion of multiple modes substantially expands complexity of the problem. In general, for L ligands considered in M binding modes each, the number of models to be constructed and compared is ML. To illustrate the magnitude of this number, a systematic analysis of only two binding modes in a small set of 30 ligands needs 230 QSAR models that would be created in about 34 years by a fast QSAR method requiring only 1 second per model. Apparently, an exhaustive one-by-one examination of possible modes for all ligands is practically impossible for any real-world problem. The feasible solutions to this problem will be discussed.