The use of computational methods to facilitate the drug discovery process is today well established and plays an important role in modern multidisciplinary drug discovery projects. A wide range of computational methods are used to find new active compounds and to optimize these compounds in order to produce new candidate drug molecules. The methods used depend on the available structural information of the target protein (or more generally the target biomacromolecule). If a three-dimensional (3D) structure of a target enzyme or receptor with a cocrystallized ligand is available from x-ray crystallography, a detailed knowledge of the nature of the ligand-binding site, the ligand-binding mode, and the interactions between the ligand and the receptor/enzyme can be obtained. On this basis, new ligands may computationally be "docked" into the binding site in order to study if they can effectively interact with the receptor. This can be performed by using sophisticated automated flexible docking and scoring computer programs. New and promising compounds identified by such computational experiments may then be synthesized and tested pharmacologically. This procedure is known as "structure-based drug design" and is discussed in Chapter 2.
However, many proteins of high interest as drug targets have so far resisted all attempts of crystallization and 3D-structure determination. This is, for instance, the case for most members of the large and important class of seven-transmembrane (7-TM) G-protein-coupled neurotransmitter receptors (see Chapter 12). In the absence of an experimentally determined 3D structure of the receptor, computational methodologies based on an analysis of the physicochemical and pharmacological properties of known ligands may be used for the design/discovery of new ligands. This computational procedure is called "ligand-based drug design" and is the subject of this chapter. The purpose of this chapter is to introduce and discuss some major methods used for ligand-based drug design and to exemplify the use of these methods by a case study in which the discovery of novel ligands for the benzodiazepine (BZD) site of the g-aminobutyric acid type A (GABAa) receptor have been successfully accomplished. For details of this case study the reader is referred to the references in the Further Readings section.
The most important and powerful method in ligand-based drug design is "pharmacophore modeling," which is used to develop a pharmacophore model describing the interactions between ligands and the target receptor from the ligand point of view. We will discuss and illustrate how a well-developed pharmacophore model can be used to search databases for new compounds, which fit the model and to optimize identified compounds by a pharmacophore-guided procedure.
A pharmacophore model does not give a quantitative prediction of receptor affinities. The main use of such a model is restricted to the prediction of candidate ligands as active or inactive. Such a classification may be fruitfully used in the selection of new molecules to be synthesized and pharmacologically tested in a drug discovery project. However, a pharmacophore model may additionally be used as a starting point for 3D quantitative structure-activity relationships (3D-QSAR) analysis. The 3D-QSAR combines pharmacophore models, molecular interaction fields and statistical chemometrics methods to give quantitative predictions of receptor affinities and in addition guidelines for ligand optimization. This methodology will be discussed in the last part of this chapter.
Was this article helpful?