evelopment of Behaviorally Selective Benzodiazapine Ligands Activating GABAA Receptors.

Our current approach to designing behaviorally selective GABAA benzodiazepine (BDZ) ligands includes the development of pharmacophores derived from both behavioral and binding data (Harris, DeLorey, He, Cook, and Loew, European Journal of Pharmacology, 401, 271-287 (2000)). We first develop initial 3D recognition pharmacophores for the overlapping binding region of agonists, inverse agonists, and antagonists at a particular behavioral endpoint. Accumulation of robust behavioral data drives the design process.  An initial 3D pharmacophore, shown in figure A below, is developed to ascertain if there are is a common 3D display of features  amongst the overlapping binding region of agonists/inverse agonists, and antagonists which are absent in compounds exhibing no effect at a particular behavioral endpoint. Such no-effect compounds are ligands which cross the blood-brain barrier but do not have significant binding affinities to the particular GABAA receptors associated with a particular behavior.

The initial pharmacohores consisting of a 3D distance matrix between the core recognition elements as depicted in figure A. Determination of the initial pharmacophore allows us to determine the initial bioactive conformations of each of the ligands binding to particular GABAA receptors associated with a particular behavioral endpoint and overlap the training set ligands at their pharmacophore points (see figure B). Such an initial pharmacophore or overlap rule may then be used to develop quantitative binding or activation pharmacophores via 3D-QSAR or multivariate discriminant analysis.

igures A-C below show a particular application of this approach. An initial pharmacophore was developed using behavioral data for 21 compounds which were BDZ sedation agonists, inverse agonists, or antagonists. Conformational libraries and stereochemical properties (computed from ab-initio or semiempirical quantum chemistry calculations) are used in conjunction with the program MOLMOD to ascertain if a particular 3D relationship exists between postulated recognition/activation moieties in the low energy conformers of the training set ligands which either elicit a behavioral response or are antagonists but is absent in compounds which elicit no effect. Should a common distance relationship be found, it constitutes a first stage pharmacohore or overlap rule (figure A) which may be used to overlap the training set ligands (B). The overlapped training set ligands are then used in a 3D QSAR analysis to determine if the pharmacophore is predictive of binding to a particular GABAA receptor subtype.

In the case of the  study depicted below, we determined that the sedation pharmacohore was predictive of binding to GABAA receptor subtypes containing  alpha-1 subunits. This result is in accord with mutational studies on the GABAA receptors in mice. The overlapped ligands in their MOLMOD determined bioactive conformations were used to perform Comparative Similarity Indice Analysis (COMSIA) analyses to prove the pharmacophore was indicative of binding to GABAA receptors containing the alpha1 subtype. Additional chemometric analyses (multivariate discriminant analysis) and 3D-QSAR analyses are underway to determine robust determinants of activation for this behavioral endpoint.


FIGURE A) SEDATION PHARMACOPHORE FOR THE OVERLAPPING BINDING REGION OF AGONISTS, INVERSE AGONISTS AND ANTAGONISTS


FIGURE B) SEDATION TRAINING SET LIGANDS OVERLAPPED AT THE PHARMACOPHORE POINTS IN FIGURE A) ABOVE.


FIGURE C) COMSIA ANALYSIS SHOWING THAT THE INITIAL PHARMACOPHORE AS ENCOMPASSED IN FIGURES A) AND B) ABOVE PREDICT BINDING TO GABAA RECEPTORS CONTAINING ALPHA-1 SUBTYPES.