Molecular Modeling and the Design
of Dopamine D₂ Partial Agonists

Abstract

Traditional D₂ antagonist antipsychotics are efficacious for the positive symptoms assumed to be associated with hyperactive dopaminergic transmission in the mesolimbic brain region. Unfortunately, blockage of D₂receptors results in abnormal activity similar to what is seen in Parkinson's disease. It is believed that normalization of dopaminergic activity could occur through the use of a dopamine D₂ partial agonist. A successful D₂ partial agonist would be effective in treating the positive symptoms by selectively activating the inhibitory presynaptic D₂ autoreceptors located on the cell bodies and terminals of dopamine neurons. This would inhibit dopamine neuronal firing, synthesis, and release of dopamine via a feedback mechanism which decreases neuronal activity. Description of the dopamine pharmacophore and novel dopamine D₂partial agonists recently discovered is presented, and based on the SAR in this project, a predictive CoMFA model has been developed and is discussed in detail.

Introduction

Initially, we superposed active and inactive D2 agonists to examine the receptor-excluded vdW volume occupied by active D2 agonists. All compounds were superposed based on two pharmacophore groups (the phenolic oxygen and basic amine) and the centroid of the aromatic ring.

The first series of compounds (chroman analogs) were selected as targets as they occupied the receptor-excluded vdW volume map. In this series of compounds, however, the flexible side chain can reside in more than one local minima. A conformational analysis of the amino side chain was performed to ensure that the conformation which adhered to the receptor-excluded vdW volume was an energetically feasible conformation. Rigidification (synthetically) of the benzyl side chain confirmed the placement of the amino side chain relative to the chroman ring system.

Superposition of Chroman Analog and Known D₂ Agonist into Receptor-excluded vdW volume using Pharmacophoric Pieces

Preparation for CoMFA Analysis

In later series of D₂ partial agonists, it became obvious that we would need to refine our initial hypothesis regarding a receptor-excluded vdW volume. All compounds could still be superposed based on our original three point fitting method, and this led us to a refined hypothetical D₂ template for the D₂ pharmacophore.

An optimized alignment is crucial for the success of any CoMFA analysis and we used the refined hypothetical D₂ template as our alignment rule in this study. The basic amine was protonated in all calculations and AM1 charges were calculated. The -log(K_i) was used in the dependent column and CoMFA fields of 20 kcal/mol (steric) and 10kcal/mol (electrostatics) were utilized.

Partial Agonists Used in This Study

CoMFA Results

A crossvalidated R² of 0.7 using 5 components and a non-crossvalidated R² of 0.9 was achieved using the aforementioned alignment rule, charge, dependent and independent columns. The standard error of prediction was 0.5 log units.

The CoMFA fields depicted five main regions surrounding the superposed molecules where interactions with the receptor may be present. In the region around the carbonyl of the oxindole compounds, a potential hydrogen bond acceptor was evident as increased negative charge. A hydrogen bond donor from either the 3-OH (manifested in the aryl piperazines, aryl ethylamines or chroman analogs) or the NH of the angular indoles may be interacting with the D₂ receptor. In the same region of the hydrogen bond donor, the receptor was sensitive to increased bulk. In the case of the angular indoles and chroman analogs, there is a region in the receptor which appeared to be more amenable to the placement of a benzodioxane moiety as opposed to a chroman or benzoxazine ring. This was depicted as a region most suitable for negative charge. Finally, bulk (benzyl analogs) on the basic amine were favorable.

This CoMFA model was used to predict activities of other compounds and could be used to prioritize synthesis. For instance, the linear indole depicted in orange was predicted to have a K_i of > 500nM. We can readily see that this compound violates most of the favorable CoMFA fields. The NH vector is not pointing in a direction where a H bond donor may be present in the receptor. Rather it is pointing in a region optimized for a hydrogen bond acceptor. The indole portion of the compound is conferring bulk in a region which has already been shown to be sensitive to increased bulk. Finally a benzyl ring on the basic amine would most certainly aid in increasing affininty for the D₂ pharmacophore.

The Lead Compound

The compound that went "forward" in this program (not shown here) met the requirements depicted in the CoMFA analysis (it was not used to create the CoMFA model). The compound also satisfied "Pfizer's Rule of 5". We plotted the experimental logD(7.4) of known drugs on the market against frequency and found that this lead compound's calculated logD(7.4) was represented in the same bin as the majority of the drugs on the market.

Pfizer's Rule of 5 Poor absorption or permeation is more likely when:

> 5 H-bond donors > 500 MWT > 5 clogP > 10 N's and O's

References

1. Discovery of a Novel Scaffold Which Embraces the D₂ Agonist Pharmacophore. Structure-Activity Relationships of a Series of 2- (Aminomethyl)chromans, J. Med. Chem., 50, 4235, 1997.

2. Accessing the D₂ Pharmacophore Via the 7-Hydroxy-2- Aminomethyl Benzoxazine Moiety, Biorg. Med. Chem. Lett., 8, 295, 1998.

3. Discovery of 3-Hydroxy-Phenoxyethylamine and 3-Hydroxy-N1- Phenylpiperazine Dopaminergic Templates, Med. Chem. Research., in press.

4. Exploiting the 2-Methyl Chroman Scaffold. Synthesis and Evaluation of a Novel Series of Orally Active 2-(Aminomethyl)-3,4,7,9- Tetrahydro-2H-Pyrano[2,3-e]Indole and Indolone Derivaties, Tetrahedron, submitted.