Three-point pharmacophore

Fig. 1. Schematic illustrating three- and four-point 3D pharmacophores. Three-point 3D pharmacophores encode three functional group types and the three distances separating them, and four-point 3D pharmacophores encode four functional group types and the six distances separating them. Functional group types commonly included are acids, bases, hydrophobes, H-bond acceptors, H-bond donors, and aromatic systems. Distances are assigned to bins (e.g., 2.5-4.0 A) to limit the individual 3D pharmacophore descriptors to a tractable number, and to aid in comparing the individual 3D pharmacophores.

The three-point pharmacophore models are shown in Fig. 7. The structures of the y0-alanine analogues and their antiepileptic activities against pilocarpine test are tabulated in Table 3. 

On the basis of the above results, a simple three-point pharmacophore model for the a1A-AR antagonists was proposed. This model was characterized by two hydro-phobic phenyl rings at an average distance of 4.43 and 6.41 A, respectively, from the essential protonated nitrogen atom. It was also inferred that, for the conformationally restricted compounds such as 3 (Scheme 8.1) and 45 (Scheme 8.2), the coplanarity of the protonated nitrogen atom and its substituents could contribute to the high affinity toward the a1A-AR [65]. 

The figure only shows the far left-hand side of the distribution, which is very long tailed (i.e, Zipfian) in nature. In the part shown, it can be seen that, for instance, there are 2118 of the three-point pharmacophores that are represented by only one molecule and 1396 of those that have only two molecules to represent them. Because of the Zipfian nature of the distribution, the figure is not the best way to convey the contents of the entire data set. This is better represented in Table 3.1. In this representation, the curve of Figure 3.2 is in effect summed in logarithmic portions from the right-hand... 

Fig. 15.25 Partial display of virtual screen of conformational ensemble of " 75 000 structure library based on multiple templates. A display of fitness scores to a three-point pharmacophore model provides initial signs of similarity of a library design to that model. Subsequent virtual screening against a more stringent four-point pharmacophore model further highlights potentially useful library designs.

Other approaches that could be extended to target class library design include work by McGregor and Muskal (13,19). These utilized PharmPrint 3D pharmacophore descriptors (three-point pharmacophores) and partial least... 

The DIVSEL program was developed by Pickett et al. for combinatorial reagent selection using three-point pharmacophores as the descriptor for similarity calculations [2], The algorithm starts by selecting the compound most dissimilar to the others in the set and then iteratively selects compounds most dissimilar to those already selected. DIVSEL was used to select a set of carboxylic acids from a collection of 1100 monocarboxylic acids for an amide library, based on the pharmacophoric diversity of the products. Eleven diverse amines were selected based on pharmacophoric diversity. A virtual library of 12100 amides was constructed from the 11 amines and 1100 carboxylic acids. The DIVSEL program used the pharmacophore fingerprints for the product virtual library to select a diverse set of the carboxylic acids. The products of 90 acids with the 11 amines selected with DIVSEL covered 85% of the three-point pharmacophores represented by the entire 12100 compound virtual library. 

The earliest discussion of a common pharmacophore in the tubulin literature was presented by Winkler and Axelsen [70], who employed the second strategy of simultaneous alignment while fitting only three points of S AR. Without a published 3D reference, we were not able to test their hypothesis. However, as pointed out by others [71], the predictions made based upon this earliest model were not supported by subsequent SAR studies. 

Three-point pharmacophores have traditionally been used for many applications but have recently been more and more replaced by four-point pharmacophores (Mason et al. 1999), which increases the complexity of the search but also the resolution of the pharmacophore analysis. This is the case because the additional point increases the total number of inter-point distances from three for a three-point pharmacophore to six for a four-point pharmacophore. Pharmacophore searching is further refined by assigning alternative features to each point (e.g., hydrogen bond acceptors, donors, or charged groups) and ranges to inter-point distances (rather than an exact distance). For example, five different features (e.g., atom types or groups) may be permitted for each point... 

These three pharmacophoric elements are shown as emboldened atoms within the generalized beta-blocker structure 60. The key atoms in 60 should be considered to be discretely displayed in three-dimensional (3D) space via the oxypropyl-connecting chain, so as to be able to dynamically occupy a specified residence within the beta-receptor pocket, rather than to be merely residing in a static manner on the rigid plane that becomes defined by three points. For example, reconsidering the previously noted historical development wherein a surprising enhance-... 

A similar approach based on pharmacophore keys is used in the ChemDiverse software [54]. Here, the key is based on three-point pharmacophores generated for seven features over 32 distances. This gives over 2 million theoretical combinations however, this number can be reduced by geometric and symmetry considerations. The key marks the presence or absence of the pharmacophores within the collection and because of its size it is normally used to represent a whole library of compounds, although in principle it can also be used to represent a single compound. 

F igure 5.2. Illustration of the creation of a pharmacophore key. As the conformation of a molecule changes, so do the distances between the pharmacophoric groups, shown as spheres. The two different three-point pharmacophores shown each set their own particular bit in the pharmacophore key. 

In an initial implementation from the authors (35), a set of 5916 three-point pharmacophore queries was generated and used to setirch a database. Compounds were charac-... 

---