Novel binding mode for negative allosteric NMDA receptor modulators

Abstract

NMDA-type ionotropic glutamate receptors mediate excitatory neurotransmission and synaptic plasticity, but aberrant signaling by these receptors is also implicated in brain disorders. Here, we present the binding site and the mechanism of action for UCM-101, a novel negative NMDA receptor modulator that produces full inhibition of NMDA receptor-mediated excitatory postsynaptic currents in hippocampal CA pyramidal neurons from juvenile mouse brain slices. UCM-101 has a 59-fold higher binding affinity at GluN1/2A compared with GluN1/2B receptors and inhibits diheteromeric GluN1/2A and triheteromeric GluN1/2A/2B receptors with IC50 values of 110 and 240 nM, respectively, in the presence of 1 µM glycine. The novel binding mode for UCM-101 is revealed in a high-resolution crystal structure of the GluN1/2A agonist binding domain heterodimer. UCM-101 and its analog TCN-213 inhibit NMDA receptors by negatively modulating co-agonist binding to the GluN1 subunit via an allosteric mechanism that is conserved with previously described GluN2A-selective antagonists, TCN-201 and MPX-004. Despite the shared mechanism of action, the structural determinants that mediate subunit selectivity for UCM-101 are distinct from those of TCN-201 and MPX-004. These findings provide detailed insights into the binding site and mechanism of action of a novel NMDA receptor modulator and open new avenues for the development of NMDA receptor ligands with therapeutic potential.

Figure 1.

Evaluation of negative allosteric modulators at NMDA receptor subtypes. (A) Chemical structures of negative allosteric modulators (NAMs), UCM-101, TCN-213, TCN-201, MPX-004, and MPX-007. (B) Representative two-electrode voltage-clamp recordings from recombinant GluN1/2A or GluN1/2B receptors activated by the indicated concentration of glycine in the continuous presence of 100 μM glutamate and inhibited by increasing concentrations of TCN-213 or UCM-101. (C) Concentration-response data for TCN-213, UCM-101, and TCN-201 at NMDA receptor subtypes activated by 1 μM glycine in the continuous presence of 100 μM glutamate. (D) Concentration-response data for TCN-213, UCM-101, and TCN-201 at NMDA receptor subtypes activated by a glycine concentration close to the glycine EC₅₀ in the continuous presence of 100 μM glutamate (1 μM glycine for GluN1/2A, 0.3 μM for GluN1/2B and GluN1/2C, and 0.1 μM for GluN1/2D). Dashed line indicates data for GluN1/2A from (C). See Tables 1 and 2 for NAM IC₅₀ and glycine EC₅₀ values.

Figure 2.

Negative allosteric modulation of triheteromeric GluN1/2A/2B receptors. Concentration-response data for TCN-213, UCM-101, and TCN-201 were determined using two-electrode voltage-clamp recordings of responses from recombinant GluN1/2A/2B receptors activated by 1 μM glycine in the continuous presence of 100 μM glutamate. Dashed lines indicate data for diheteromeric GluN1/2A and GluN1/2B from Fig. 1D. See Table 1 for NAM IC₅₀ values. See Fig. S1 for control experiments and details on the approach to selectively express triheteromeric GluN1/2A/2B receptors without confounding co-expression of diheteromeric GluN1/2A and GluN1/2B receptors.

Figure 3.

Negative allosteric modulation of NMDA receptor-mediated EPSCs. (A) Representative recordings of evoked NMDA receptor-mediated EPSCs from CA1 pyramidal neurons in juvenile (3–5 weeks old) mouse brain slices. The EPSCs were recorded at baseline (black) before the addition of 3 μM NAM (colored), followed by 100 μM AP5 (gray) at the end of the recordings to inhibit NMDA receptor-mediated currents. (B) The averaged data from multiple neuronal recordings show the time course of NMDA receptor-mediated EPSC peak amplitudes normalized to baseline. The bars indicate application of TCN-213 (N = 8), UCM-101 (N = 6), TCN-201 (N = 5), or MPX-004 (N = 6) as well as AP5. Data are shown as mean ± SEM. (C) Summary of EPSC peak amplitudes following NAM application normalized to baseline peak amplitudes. * denotes a significant difference (p < 0.05, one-way ANOVA with Tukey’s posttest). (D) Summary of the EPSC decay times (weighted deactivation time constant, τ_W) at baseline and following NAM treatment. ND indicates not determined. * denotes a significant difference (p < 0.05, ratio paired t-test).

Figure 4.

Crystal structure of the GluN1/2A ABD heterodimer with DCKA, glutamate, and UCM-101. (A) Views of the GluN1/2A ABD heterodimer structure with ligands shown in spheres. The competitive GluN1 antagonist DCKA stabilizes the ABD in an open cleft conformation, while the GluN2 agonist glutamate stabilizes the ABD in a closed cleft conformation. UCM-101 binds an allosteric site at the heterodimer interface between the orthosteric sites in GluN1 and GluN2 subunits. See Table S3 for data collection and refinement statistics. (B) Views of UCM-101 with the electron density shown as mesh. (C) The allosteric binding site for UCM-101. Dashed lines indicate hydrogen bonding interactions between residues and two water molecules. (D) Ligand-protein interactions for UCM-101 in the allosteric binding site. Hydrogen bonding interactions are shown as dashed lines with distances in angstrom (Å). (E) Structural alignment of DCKA/glutamate-bound GluN1/2A ABD heterodimer structures with UCM-101 (green) or MPX-007 (grey; PDB ID: 5JTY). UCM-101 occupies the allosteric site in a more extended binding mode compared to MPX-007 that adopts a U-shaped conformation. The binding sites for UCM-101 and MPX-007 are overlapping, but distinct, with a UCM-subsite uniquely engaged by UCM-101 and a TCN/MPX-subsite only engaged by MPX-007. (F) Representative two-electrode voltage-clamp recordings from mutant GluN1-R755A/2A receptors activated by glycine in the continuous presence of 300 μM glutamate, and concentration-response data for TCN-213 and UCM-101 at wild type and mutant GluN1/2A receptors activated by glycine close to the EC₅₀ value (GluN1/2A, 1 μM; GluN1-Y535A/2A and GluN1-R755A/2A, 3 μM). Dashed line indicates data for wild type GluN1/2A from Figure 1C. See Tables 1 and 2 for NAM IC₅₀ and glycine EC₅₀ values.

Figure 5.

Mutagenesis of residues in the UCM-subsite. (A) The ligand and residues in the vicinity of the ethyl group on UCM-101 with the electron density shown as mesh (2F_o-F_c map contoured at 1σ). The GluN2A I755F mutation is modeled and shown with the predicted distance to the ethyl group. (B) Alignment of key GluN2 residues in the UCM-subsite of the allosteric binding site with non-conserved residues in red. Amino acid numbering is according to the full-length GluN2A subunit. (C) Representative two-electrode voltage-clamp recording from mutant GluN1/2A-I755F receptors activated by glycine in the continuous presence of 300 μM glutamate and inhibited by increasing TCN-213 concentrations. (D) Concentration-response data for TCN-213 and UCM-101 at wild type and mutant GluN1/2A receptors activated by glycine close to the EC₅₀ value (1 μM). Dashed line indicates data for wild type GluN1/2A from Figure 1C. (E) Fold-change in IC₅₀ relative to wild type GluN1/2A. * denotes significantly different from wild type; # denotes significantly different IC₅₀ shifts between TCN-213 and UCM-101 (p < 0.05, two-way ANOVA with Tukey’s posttest). See Tables 1 and 2 for NAM IC₅₀ and glycine EC₅₀ values.

Figure 6.

Structural determinants of subunit-selectivity. (A) Alignment of GluN2 residues in the allosteric binding site with non-conserved residues in red. Amino acid numbering is according to the full-length GluN2A subunit. (B-D) Concentration response data for UCM-101 and TCN-201 at wild-type and mutant GluN1/2A receptors activated by glycine close to the EC₅₀ value (GluN1/2A, GluN1/2A-V529I, and GluN1/2A-T797S, 1 μM; GluN1/2A-V783F, GluN1/2A-V783L, GluN1/2A-M788I, and GluN1/2A-M788T, 0.5 μM; GluN1/2A-E789Q, 3 μM). Dashed line indicates data for wild type GluN1/2A from Figure 1C. (E) Fold-change in IC₅₀ relative to wild type GluN1/2A. * denotes significantly different from wild type; # denotes significantly different IC₅₀ shifts between UCM-101 and TCN-201 (p < 0.05, two-way ANOVA with Tukey’s posttest). See Tables 1 and 2 for NAM IC₅₀ and glycine EC₅₀ values.

Figure 7.

Mechanism of negative allosteric modulation by UCM-101 and TCN-213. (A) Representative two-electrode voltage-clamp recordings of UCM-101 inhibition of GluN1/2A responses activated by low glutamate (10 μM) and glycine (1 μM) or high glutamate (100 μM) and glycine (30 μM). (B) Concentration response data for TCN-213 and UCM-101 at GluN1/2A receptors activated by different combinations of glutamate and glycine concentrations. Dashed line indicates data for wild type GluN1/2A from Figure 1C. (C) Cartoon representation of the GluN1-CC/2A heterodimer with an engineered disulfide bond in the GluN1 ABD. The engineered disulfide bond stabilizes the closed cleft, active GluN1 ABD conformation and receptors with formed disulfide bonds do not require glycine binding and can be activated by glutamate alone. (D) Representative recordings of GluN1-CC/2A receptors in the continuous presence of 4 mM DTT. Glutamate (300 μM) alone is sufficient to activate these receptors since the disulfide bonds continuously break and reform, even in the continuous presence of DTT. However, UCM-101 and TCN-213 stabilize inhibits the glutamate response by stabilizing the open cleft, inactive GluN1 ABD conformation and preventing disulfide bonds from reforming. (E) Bar graph summarizing responses without (control) and with DTT application (+ DTT) and with vehicle (0.1% DMSO), the competitive GluN1 antagonist 7CKA (10 μM), or the negative allosteric modulators, UCM-101 (10 μM) or TCN-213 (10 μM). The responses are normalized to the response to glutamate alone applied at the beginning of the recording. The allosteric modulators mimic the functional effects of stabilizing the open cleft, inactive GluN1 ABD conformation with 7CKA. * denotes significantly different from control (p < 0.05, two-way ANOVA with Tukey’s posttest).

Figure 8.

Negative allosteric modulator binding affinity and strength of allosteric modulation. (A) Representative recordings of glycine concentration-response data in the presence and absence of UCM-101. (B) Equilibrium model describing agonist binding affinity (i.e. the inverse of agonist binding dissociation constant, 1/K_A), agonist efficacy (equilibrium constant, E), modulator binding affinity (i.e. inverse of modulator binding dissociation constant, 1/K_B), and strength of allosteric interaction (i.e. the allosteric binding interaction constant, α). The equation describes the dose ratio, where agonist EC₅₀ is measured in the absence of modulator and agonist EC₅₀’ is measured in the presence of a modulator concentration [B]. (C) Schild plot for UCM-101 (red) and TCN-213 (blue) at GluN1/2A and UCM-101 at GluN1/2B (black). The calculated values for log(DR-1) are plotted over data simulated using the best fit K_B and α values from global non-linear regression of the data shown in (D-F) (i.e. the calculated log(DR-1) data are not used for model fitting of the equation). (D-F) Glycine concentration-response data in the absence and presence of increasing concentrations of TCN-213 or UCM-101 at GluN1/2A or UCM-101 at GluN1/2B. The best fit K_B and α values are stated above the graphs. * indicates that the α values for UCM-101 at GluN1/2B was fixed during global non-linear regression fitting. See Materials and Methods for details on data analysis.