Structure-activity relationships and pharmacophore model of a noncompetitive pyrazoline containing class of GluN2C/GluN2D selective antagonists

Abstract

Here we describe the synthesis and structure-activity relationship for a class of pyrazoline-containing dihydroquinolone negative allosteric modulators of the NMDA receptor that show strong subunit selectivity for GluN2C- and GluN2D-containing receptors over GluN2A- and GluN2B-containing receptors. Several members of this class inhibit NMDA receptor responses in the nanomolar range and are more than 50-fold selective over GluN1/GluN2A and GluN1/GluN2B NMDA receptors, as well as AMPA, kainate, GABA, glycine, nicotinic, serotonin, and purinergic receptors. Analysis of the purified enantiomers of one of the more potent and selective compounds shows that the S-enantiomer is both more potent and more selective than the R-enantiomer. The S-enantiomer had an IC50 of 0.17-0.22 μM at GluN2D- and GluN2C-containing receptors, respectively, and showed over 70-fold selectivity over other NMDA receptor subunits. The subunit selectivity of this class of compounds should be useful in defining the role of GluN2C- and GluN2D-containing receptors in specific brain circuits in both physiological and pathophysiological conditions.

Figure 1. Previously reported best in class compound and representative structure for SAR

A. The structure of the previously reported best in class compound, DQP-1105 B. The structure of a general analogue with numbered substituents is shown.

Figure 2. Evaluation of substituent effects for A- and B-ring modifications

A. The σ substituent constants of the para-substituted A-ring analogs vs. activity show a correlation for GluN2C- and GluN2D-containing receptors, when the R₁ position of the C-ring is substituted with chloro. (GluN2D r²=0.82, p < 0.05 Pearson two-tailed correlation analysis; GluN2C r²=0.84, p < 0.05 Pearson two-tailed correlation analysis; Compounds 2–5 and 8–10). B. The analysis of the para-substituents on the B-ring as a function of activity at GluN2C- and GluN2D-containing receptors appears parabolic with respect to the σ substituent constants, with an optimal value close to that of the chloro- and bromo-substitutions (Compounds 29–33). C. The analysis of the para-substituents at GluN2A-containing receptors shows a similar parabolic relationship as observed at the GluN2C- and GluN2D-containing receptors when the activity is plotted as a function of the σ substituent constants (Compounds 29–33). D. The analysis of the substituent effects appears parabolic with respect to the π substituent constants for B-ring meta-substituted compounds at GluN2A-containing receptors, suggesting an optimal hydrophobicity close to that of the chloro-substitution. Substituent constants were obtained from the same source (Compounds 34–39).

Figure 3. Separation of enantiomers

A. The enantiomers of the final compound, 26, could be separated using reverse phase chiral chromatography (see Methods). B. The crystal structure of the inactive enantiomer, 70 (Table 6) was solved using X-ray diffraction and has the R configuration.

Figure 4. Improvements in selectivity and potency

A. The potency of the racemic compounds at GluN2D-containing receptors was improved 10-fold over the previous members in the class. B. The potency of the S-enantiomer of compound 26, compound 69, is two-fold more potent than the racemic mixture at GluN2D-containing receptors while the potency at GluN2A- and GluN2B-containing receptors is unaffected, making it more selective for GluN2C- and GluN2D-containing receptors. C. The potency of the R-enantiomer of compound 26, compound 70, at GluN2C- and GluN2D-containing receptors is diminished as compared to the racemate 26, making it less selective over GluN2A- and GluN2B-containing receptors. D. Bar graph showing the fold-selectivity improvements attained through SAR. Data for compounds 997 and DQP-1105 (Panels A and D) are previously published and shown here for comparison.

Figure 5. Pharmacophore model and electrostatic potential maps of para-B-ring modifications

A. The para-substitution of the A-ring shows correlation between the σ substituent constants and activity at GluN2C/D-containing receptors, when R₁ is a chlorine. The length and configuration of the acyl-chain is flexible, with the trans-configuration improving potency; B-ring modification shows an optimal para-sigma coefficient close to that of chloro- and bromo-substitutions for GluN2A-, GluN2C- and GluN2D-containing receptors, suggesting a conserved nature of the binding interaction at each of the three receptors. The C-ring substitutions explored are consistent with this portion of the molecule interacting with a hydrophobic pocket and could allow for improvements is selectivity. B. The electrostatic potential maps of the para-B-ring modifications evaluated are shown. Only the Cl-and Br-substituents show significant electron deficiency at the termini of the substituents, suggesting a potential halogen bond could be responsible for the improved potency of these compounds.

Scheme 1. Synthesis of dihydro-quinolone-pyrazoline derivatives

(a) Anhydrous THF, Triphosgene (warning, triphosgene is toxic, see Methods), reflux. (b) EtOH, Weinreb’s HCl salt, reflux. (c) Anhydrous THF, n-Butyllithium, −78 °C. (d) Ethylacetoacetate, DMF, 4Å molecular sieves, 180 °C, µW. (e) 4:3 EtOH:H2O (0.05 M), 0 °C to r.t. (f) hydrazine monohydrate, EtOH, 110 °C, µW. (g) Anhydrous THF, 4Å molecular sieves, 165 °C, µW.

Scheme 2. Modifications to the acyl-chain

(a) HCl, MeOH. (b) BH₃-Me₂S, Anhydrous THF, 0 °C. (c) EDCI, DMAP, NH₃ in dioxane (0.5M), THF. (d) EDCI, DMAP, (E)-4-methoxy-4-oxobut-2-enoic acid. (e) Anhydrous THF, (E)-methyl 4-chloro-4-oxobut-2-enoate, 4Å molecular sieves, 165 °C, µW. (f) methyl 4-oxobutanoate, BH₃-Me₂S, THF. (g) EDCI, DMAP, 4-fluorobutanoic acid, DCM (h) NaOH, EtOH:H₂O.