Database Searching Using The Pharmacophore Model

To identify new classes of compounds, which may bind to the BZD site, the pharmacophore model in Figure 3.6 was used for database searching in CATALYST. Two different compound databases, the Maybridge database (Maybridge Chemical Company) with approximately 47,000 compounds and the Available Chemical Directory (ACD) database (MDL Information Systems, Inc.) with approximately 250, 000 compounds were searched. In order to search a database using a 3D-pharmacophore model, the compounds in the database, which are 2D structures has to be converted to 3D structures and a set of conformations for each compound has to be generated. The conformations can either be generated in advance or for some programs during the search. In CATALYST, a set of conformations for each compound in the database is generated in advance. When a pharmacophore model is used to search a database, compounds fitting all or some of the pharmacophore elements are identified. Such compounds are called "hits." In the present case, it is required that all pharmacophore elements must to some degree be fitted to give a hit. In the search procedure all compounds are given a fit value indicating how well they fit the pharmacophore elements of the model. After the search, a list of hits ranked according to their fit value is available.

As described in Sections 3.4.2 and 3.4.3, exclusion spheres and a vdW shape may be used in a pharmacophore model to represent the dimensions of the receptor binding cavity. This is of high importance for database searching in order to keep the number of hits to a manageable size.

The difference between an exclusion sphere and the vdW shape is that no ligand is allowed to touch an exclusion sphere without a severe penalty, whereas a ligand may be somewhat larger or smaller than the shape without a penalty. In the present case, ligands with vdW volumes 30% smaller or 10% larger than the volume of the shape are allowed.

When the two databases were searched using the pharmacophore model in Figure 3.6, 22 hits in the Maybridge database and 76 hits in the ACD database were obtained. (It should in this context be mentioned that these database searches are very fast, using less than 30 min per database.) Among the 98 hits, five compounds of the highest ranking hits with a significant diversity of the molecular structures were selected and purchased. The highest affinity of these compounds was the 4-quinolone derivative 3.14 (K = 122 nM) shown in Figure 3.8. As also shown in the figure, this compound is cf,

3.14

3.14

FIGURE 3.8 The most active hit from the database search and its fit to the pharmacophore model.

fitted to the pharmacophore model with its ethyl group corresponding to the fused benzene ring in the flavones and the CF3 group corresponding to the isopropyl ester in 3.4 (Figure 3.2). The structure of the compound is structurally much different from that of flavones and it is a novel compound in a medicinal chemistry sense as it represents a class of compounds that has not previously been tested for affinity for the BZD site of the GABAa receptor.

3.5.1 Postprocessing of Database Hits—An Essential Requirement

To avoid false positives, i.e., compounds that are ranked high in a database search, but are found to be of low affinity when tested, postprocessing of database hits is, in general, necessary before selection of compounds for synthesis or purchase. As an example, the second best hit in the database search is compound 3.15 fitting to the pharmacophore model with the hydroxyl group as a hydrogen-bond donor as shown in Figure 3.9. Even though the compound displays a good fit to the

Drug Pharmacophores

Global energy minimum 0.0 kJ/mol

Bioactive conformation +48 kJ/mol

FIGURE 3.9 Compound 3.15, a top ranking hit in the database search together with the calculated global energy minimum and the bioactive conformations.

Global energy minimum 0.0 kJ/mol

Bioactive conformation +48 kJ/mol

FIGURE 3.9 Compound 3.15, a top ranking hit in the database search together with the calculated global energy minimum and the bioactive conformations.

pharmacophore, the compound shows only weak affinity (Ki = 6400 nM). According to calculations, an intramolecular hydrogen bond between the hydroxyl group and the carbonyl group is present in the global energy minimum of the compound (Figure 3.9). In order to donate a hydrogen bond from the hydroxyl group as required by the pharmacophore model, the intramolecular hydrogen bond has to be broken giving a high conformational energy penalty (calculated to be 48 kJ/mol) and as a consequence of this the affinity is strongly decreased. Such conformational energy penalties may not be taken properly into account by database searching programs. It is also highly advisable to examine hydrogen bond distances to hydrogen bonding pharmacophore elements in the model to remove hits with too long or too short hydrogen bonds.

Was this article helpful?

0 0

Post a comment