Development of a Dihydroquinoline-Pyrazoline GluN2C/2D-Selective Negative Allosteric Modulator of the N-Methyl-d-aspartate Receptor

Abstract

Subunit-selective inhibition of N-methyl-d-aspartate receptors (NMDARs) is a promising therapeutic strategy for several neurological disorders, including epilepsy, Alzheimer's and Parkinson's disease, depression, and acute brain injury. We previously described the dihydroquinoline-pyrazoline (DQP) analogue 2a (DQP-26) as a potent NMDAR negative allosteric modulator with selectivity for GluN2C/D over GluN2A/B. However, moderate (<100-fold) subunit selectivity, inadequate cell-membrane permeability, and poor brain penetration complicated the use of 2a as an in vivo probe. In an effort to improve selectivity and the pharmacokinetic profile of the series, we performed additional structure-activity relationship studies of the succinate side chain and investigated the use of prodrugs to mask the pendant carboxylic acid. These efforts led to discovery of the analogue (S)-(-)-2i, also referred to as (S)-(-)-DQP-997-74, which exhibits >100- and >300-fold selectivity for GluN2C- and GluN2D-containing NMDARs (IC₅₀ 0.069 and 0.035 μM, respectively) compared to GluN2A- and GluN2B-containing receptors (IC₅₀ 5.2 and 16 μM, respectively) and has no effects on AMPA, kainate, or GluN1/GluN3 receptors. Compound (S)-(-)-2i is 5-fold more potent than (S)-2a. In addition, compound 2i shows a time-dependent enhancement of inhibitory actions at GluN2C- and GluN2D-containing NMDARs in the presence of the agonist glutamate, which could attenuate hypersynchronous activity driven by high-frequency excitatory synaptic transmission. Consistent with this finding, compound 2i significantly reduced the number of epileptic events in a murine model of tuberous sclerosis complex (TSC)-induced epilepsy that is associated with upregulation of the GluN2C subunit. Thus, 2i represents a robust tool for the GluN2C/D target validation. Esterification of the succinate carboxylate improved brain penetration, suggesting a strategy for therapeutic development of this series for NMDAR-associated neurological conditions.

Keywords: NR2C; NR2D; blood–brain barrier; epilepsy; seizure; tuberous sclerosis complex.

Figure 1

Structure of the GluN1/GluN2D. (A) Model of the GluN1/GluN2D NMDAR (Strong et al.) with the intracellular C-terminal domain omitted. Known NMDAR modulator sites are shown. Green and orange chains are GluN1, while blue and magenta chains are GluN2 (see the Methods section). (B) Expanded ribbon structure shown for the homology model of the GluN1/GluN2D heteromeric tetrameric complex in panel (A). (C) Key residues (Gln701, Leu705) that are critical for DQP-1105 activity (Acker et al.) and QNZ-46 activity (Hansen and Traynelis) are shown on GluN1/GluN2D NMDA receptors.

Figure 2

Structures for GluN2C/D-selective NAMs: DQP-1105 (1) (Acker et al.), QNZ-46 (Mosley et al.; Hansen and Traynelis), UBP-1700 (Wang et al.), EU1794-4 (Katzman et al.; Perszyk et al.,), and NAB-14 (Swanger et al.).

Figure 3

SAR studies conducted on DQP-based GluN2C/D NAMs.

Figure 4

Agonist dependence of compound 2i (DQP-997-74). (A) Structure of compound 2i. (B) Whole-cell current recordings from HEK cells show GluN1/GluN2D responses to glutamate with preapplied compound 2i or control (vehicle, 0.1% DMSO). (C) Concentration–response curves show peak and steady-state responses normalized to control. (D) The current responses from panel (B) were expanded to show the concentration-dependent time course of relaxation following receptor activation during compound 2i preapplication. The plot shows a linear relationship between 1/τ_inhibition and compound 2i concentration. (E) The deactivation period following glutamate removal was expanded to show the slowing of receptor deactivation (τ_deactivation) by compound 2i. τ_deactivation for individual cells and the mean ± SEM are plotted, and data were compared by the mixed effect model, F(1.386, 13.40) = 27.45, p < 0.001. Asterisks indicate statistically significant differences, see the Results and Discussion section for p-values. (F) Whole-cell current recording of responses to 5 brief pressure-applied NMDA (500 μM) and glycine (250 μM) pulses at 1 s intervals. The responses in the absence or increase of concentrations of compound 2i are superimposed. (G) Concentration–response curves show current responses for the first and last pulse normalized to control. (H) Whole-cell current recordings from HEK cells show GluN1/GluN2D responses to glutamate with coapplication of compound 2i or control with glutamate. The mean fitted dissociation rate of compound 2i for all concentrations was 26.5 s (95% confidence interval; 17.4, 35.6; N = 15). (I) Plots of the concentration–response curves show the steady-state inhibition response compared to control. (J) The plot shows a linear relationship between a 1/τ_inhibition and compound 2i concentration and a concentration-independent dissociation rate for 2i.

Figure 5

Mouse plasma and brain pharmacokinetic profiles of 2i (DQP-997-74). C57Bl/6 mice were administered (A) 10 mg/kg of 2i IP using 50:50 PEG400/H₂O as a vehicle or (B) 5 mg/kg 2i IV in 5% N-methyl-2-pyrrolidine, 5% Solutol HS-15, and 90% saline. The concentration of 2i was followed in the plasma and brain. Data are the mean concentration ± SEM at each time point; SEM shown when larger than symbol. BLQ indicated below the level of quantification.

Figure 6

Antiepileptic effect of IP administration of 2i (DQP-997-74) in vivo. (A) Representative intracortical EEG recordings of spontaneous electrographic seizures in layer 2/3 (top) and layer 4 (bottom) in a head-restrained P14 Tsc1^± mouse. Recordings were performed prior to the administration of 14 mg/kg of 2i as indicated by the red arrow. (B) Representative intracortical EEG recordings of spontaneous seizures in layer 2/3 (top) and layer 4 in head-restrained P14 Tsc1^± mouse. The administration of 28 mg/kg of 2i is indicated by the red arrow. Note the diminution of the seizure frequency after IP injection. Black boxes show extended traces of a seizure recorded before (left) and a seizure-free trace recorded after the injection of 2i (right). (C) Mean and SEM of seizure frequency, duration, and amplitude over the 2 h EEG recording before IP administration of 7 mg/kg (n = 3), 14 mg/kg (n = 4), and 28 mg/kg (n = 5) of 2i and the 2–2.5 h of EEG recording postadministration.

Figure 7

Mouse plasma and brain pharmacokinetic profiles of 2i (DQP-997-74) produced by prodrugs. C57Bl/6 mice were administered (A) 5 mg/kg 2l IV in 5% N-methyl-2-pyrrolidine, 5% Solutol HS-15, and 90% saline. (B–D) Concentration of 2i in plasma and brain following IV administration of prodrugs 2m (B), 2o (C), and 2p (D) (5 mg/kg; same vehicle as (A)). Shaded area is the brain 2i level 1 h after IV administration, from Figure 5. Data represent the mean concentration ± SEM at each time point; SEM shown when larger than symbol.

Scheme 1. Synthesis of DQP Intermediates and Derivatives with Modified Acyl Side Chains

Reagents and conditions: (a) ethyl acetoacetate, 4 Å mol sieves, THF, μwave 160 °C, 30 min, 78%; (b) 4-chlorobenzaldehyde, KOH, 4:3 EtOH/H₂O, 0 °C to rt, overnight, 93%; (c) 65% H₄N₂·H₂O, EtOH, 110 °C, 2 h, 90%; (d) glutaric anhydride, 4 Å MS, THF, reflux, 5 h, 63%; (e) Boc-Glu-OtBu, HBTU, TEA, DMF, rt, overnight; (f) 1:1 TFA: DCM, 70 °C, μwave, 2 min, 29% over two steps; (g) carboxylic acid, HBTU, TEA, DMF, 85 °C, μwave, 20 min, 45–59%;

Scheme 2. Synthesis of a DQP Analogue with Tetrazole Bioisostere

Reagents and conditions: (a) 1 N NaOH, MeOH, rt, 16 h; (b) 6, Et₃N, T3P, DMF, 0 °C–rt, 16 h, 64%.

Scheme 3. Synthesis of Fluorinated DQP Analogues and Prodrugs

Reagents and conditions: (a) TFAA, i-PrOAc, rt to 50 °C, 1.5 h; (b) 6, THF, 0 °C, 1 h, 89% over two steps; (c) TFAA, THF, 4 Å MS, 0 °C to rt, 1 h, 86%; (d) 2g, oxalyl chloride, DCM, DMF, rt; (e) alcohol or amine, DCM, 0 °C, 41–73% over two steps.