Functional assessment of triheteromeric NMDA receptors containing a human variant associated with epilepsy

Abstract

Key points: NMDA receptors are neurotransmitter-gated ion channels that are critically involved in brain cell communication Variations in genes encoding NMDA receptor subunits have been found in a range of neurodevelopmental disorders. We investigated a de novo genetic variant found in patients with epileptic encephalopathy that changes a residue located in the ion channel pore of the GluN2A NMDA receptor subunit. We found that this variant (GluN2A^N615K ) impairs physiologically important receptor properties: it markedly reduces Mg²⁺ blockade and channel conductance, even for receptors in which one GluN2A^N615K is co-assembled with one wild-type GluN2A subunit. Our findings are consistent with the GluN2A^N615K mutation being the primary cause of the severe neurodevelopmental disorder in carriers.

Abstract: NMDA receptors are ionotropic calcium-permeable glutamate receptors with a voltage-dependence mediated by blockade by Mg²⁺ . Their activation is important in signal transduction, as well as synapse formation and maintenance. Two unrelated individuals with epileptic encephalopathy carry a de novo variant in the gene encoding the GluN2A NMDA receptor subunit: a N615K missense variant in the M2 pore helix (GRIN2A^C1845A ). We hypothesized that this variant underlies the neurodevelopmental disorders in carriers and explored its functional consequences by electrophysiological analysis in heterologous systems. We focused on GluN2A^N615K co-expressed with wild-type GluN2 subunits in physiologically relevant triheteromeric NMDA receptors containing two GluN1 and two distinct GluN2 subunits, whereas previous studies have investigated the impact of the variant in diheteromeric NMDA receptors with two GluN1 and two identical GluN2 subunits. We found that GluN2A^N615K -containing triheteromers showed markedly reduced Mg²⁺ blockade, with a value intermediate between GluN2A^N615K diheteromers and wild-type NMDA receptors. Single-channel conductance was reduced by four-fold in GluN2A^N615K diheteromers, again with an intermediate value in GluN2A^N615K -containing triheteromers. Glutamate deactivation rates were unaffected. Furthermore, we expressed GluN2A^N615K in cultured primary mouse cortical neurons, observing a decrease in Mg²⁺ blockade and reduction in current density, confirming that the variant continues to have significant functional impact in neuronal systems. Our results demonstrate that the GluN2A^N615K variant has substantial effects on NMDA receptor properties fundamental to the roles of the receptor in synaptic plasticity, even when expressed alongside wild-type subunits. This work strengthens the evidence indicating that the GluN2A^N615K variant underlies the disabling neurodevelopmental phenotype in carriers.

Keywords: Epilepsy; Genetics; Human; electrophysiology; intellectual disability.

Figure 1

GluN2A^N615K reduces Mg²⁺ blockade in triheteromeric NMDA receptors A, representative whole‐cell, voltage clamp recordings from HEK293T cells expressing NMDA receptors containing wild‐type GluN2A diheteromers, GluN2A/2B triheteromers and GluN2A^N615K‐containing diheteromers and triheteromers, partnered with either GluN2A^WT or GluN2B^WT. All subunits, including wild‐type, had modified endoplasmic reticulum signals. The traces show responses to glutamate (1 mm) and blockade by Mg²⁺ (1 mm) in the continuous presence of glycine (100 μm), held at –60 mV. B, summary data showing percentage blockade by Mg²⁺ for each receptor composition. A one‐way ANOVA showed a significant effect of receptor subunit composition (F _4,64 = 346, P < 2 × 10^–16). Post hoc Bonferroni corrected t tests showed that GluN2A^N615K diheteromeric receptors had a significantly reduced Mg²⁺ blockade (9 ± 1%, n = 7) compared to GluN2A^WT diheteromers (107 ± 2%, n = 7, t _7.3 = 42, P = 5 × 10^–10). Triheteromeric receptors containing only one GluN2A^N615K subunit also showed a marked reduction in Mg²⁺ blockade, whether partnered with one GluN2A^WT subunit (53 ± 2%, n = 20), compared with GluN2A^WT diheteromers (t _19.9 = 16.5, P = 2 × 10^–12) or partnered with one GluN2B^WT subunit (36 ± 1%, n = 20), compared with GluN2A^WT/2B^WT (103 ± 1%, n = 15, t _32.6 = 36.9, P = 9 × 10^–16). The reduction in Mg²⁺ blockade was greater in GluN2A^N615K subunits partnered with GluN2B^WT than with GluN2A^WT (t _29.4 = 6.3, P = 3 × 10^–6). Some traces show greater than 100% blockade by Mg²⁺, reflecting a small contamination of the bath fluid (baseline exposure) by glutamate. C, summary data showing percentage blockade by Mg²⁺ for NMDA receptors with and without modified C‐termini. A two‐way ANOVA showed a significant influence of C‐terminus tag (F _1,35 = 8.0, P = 0.00785) and a significant effect of receptor subunit composition (F _2,35 = 2635, P < 2 × 10^–16), although there was no significant interaction (F _2,35 = 1.3, P = 0.28). However, post hoc Bonferroni corrected t tests showed no significant effect of modifying C‐termini with engineered tags for any subunit.

Figure 2

GluN2A^N615K does not alter glutamate deactivation A and B, representative whole‐cell, voltage clamp recordings from HEK293T cells expressing NMDA receptors containing wild‐type GluN2A diheteromers, GluN2A/2B triheteromers and GluN2A^N615K containing diheteromers and triheteromers, partnered with either GluN2A^WT or GluN2B^WT. All subunits, including wild‐type, had modified endoplasmic reticulum signals. The traces show deactivation following 5 ms exposure to glutamate (1 mm) and glycine (100 μm) in the absence of Mg²⁺ when held at –60 mV. Responses are normalized to their peak amplitudes. C, summary data showing weighted time constants fitted to the glutamate deactivation curves. A one‐way ANOVA showed a significant effect of receptor subunit composition (F _4,64 = 14, P = 3 × 10^–8). Post hoc Bonferroni corrected t tests showed that GluN2A^WT/2B triheteromers had a slower deactivation time constant (77 ± 4 ms; n = 15), than GluN2A^WT diheteromers (45 ± 4 ms; n = 7; t ₁₇ = 6.2, P = 4 × 10^–5). GluN2A^WT diheteromers were indistinguishable from GluN2A^N615K diheteromers (41 ± 3; n = 7; t ₁₁ = 0.9, P ≥ 0.4), with GluN2A^N615K and GluN2A^N615K/2A (49 ± 2 ms, n = 20) also showing no difference (t _10.9 = 2.6, P = 0.1). GluN2A^N615K/2B triheteromers (79 ± 6, n = 20) showed a slower deactivation rate than GluN2A^N615K diheteromers (t _10.9 = 2.6, P = 5 × 10^–5)). D, summary data showing current density (peak amplitude/capacitance) for cells expressing diheteromeric and triheteromeric receptor combinations (for other experiments, only cells with a current response of between 100 and 1000 pA were used to allow accuracy without escape of unwanted subunit combinations but, here, all traces are included). A one‐way ANOVA showed no effect of receptor subunit composition (F _4,104 = 2.1, P = 0.08). E, summary data showing weighted time constants fitted to glutamate deactivation curves. Modifying the C‐terminus with engineered peptide tags designed to vary retention in the endoplasmic reticulum had no effect on glutamate deactivation in diheteromeric wild‐type receptors containing GluN2A^WT or GluN2B^WT, or in those containing GluN2A^N615K. A two‐way ANOVA showed no significant influence of C terminus tag (F _1,35 = 4.0, P = 0.054), a significant effect of NMDAR subunit composition (F _2,35 = 226, P < 2 × 10^–16) and no significant interaction (F _2,35 = 2.5, P = 0.1).

Figure 3

GluN2A^N615K reduces single‐channel conductance in triheteromeric NMDA receptors A–C, representative voltage clamp recordings made from cell‐attached patches from HEK293T cells expressing NMDA receptors containing wild‐type GluN2A diheteromers, GluN2A/2B triheteromers and GluN2A^N615K‐containing diheteromers and triheteromers, partnered with either GluN2A^WT or GluN2B^WT. All subunits, including wild‐type, had modified endoplasmic reticulum signals. The traces show single‐channel currents in the presence of glutamate (1 mm) and glycine (100 μm). The pipette potentials used for the traces illustrated in (A) to (C) were +120, +140 and +140 mV, respectively. ‘C’ = closed; ‘g1’ = first conductance fitted; ‘g2’ = second conductance fitted. A number of transitions between the two conductance states can be seen in (C). D, summary data showing single channel conductance for receptors with and without modified C‐termini. A two‐way ANOVA showed no significant influence of C‐terminus tag (F _1,62 = 0.6, P = 0.42), a significant effect of receptor subunit composition (F _2,62 = 278, P < 2 × 10^–16) and a significant interaction (F _2,62 = 5.7, P = 0.006). E, representative amplitude histograms showing fitted normal distributions, superimposed from three different patches. The means of the fitted distributions were used to calculate conductance. For the patch expressing a GluN2A^N615K‐containing triheteromer, two peaks can be seen: the prominent subconductance and the less frequent intermediate conductance. F, summary data showing conductances, calculated by plotting current amplitudes against pipette potential (at least three potentials within range + 40 to + 180 mV). A two‐way ANOVA showed a main effect of subunit (F _4,70 = 121, P < 2 × 10^–16) and of conductance level (g1 vs. g2; F _1,70 = 196, P < 2 × 10^–16), with no significant interaction. Post hoc Bonferroni corrected t tests showed that GluN2A^N615K diheteromeric receptors had a significantly lower conductance (18 ± 2 pS; n = 11) than GluN2A^WT diheteromers (69 ± 3 pS; n = 12; t _15.4 = 13.7, P = 2 × 10^–9). Triheteromeric receptors containing only one GluN2A^N615K subunit also showed a marked reduction in intermediate conductance (g2), whether partnered with one GluN2A^WT subunit (44 ± 1 pS; n = 6), compared with GluN2A^WT diheteromers (t _13.5 = 7.0, P = 3 × 10^–5) or with one GluN2B^WT subunit (44 ± 1 pS; n = 16), compared with GluN2A^WT/2B^WT (56 ± 2 pS; n = 10; t _18.4 = 4.2, P = 0.003). The reduction in conductance was no greater in GluN2A^N615K subunits partnered with GluN2B^WT than with GluN2A^WT (t _19.1 = 0.06, P > 0.9). G, summary data showing receptor mean open times calculated by fitting single exponentials to open time distributions from individual patches. Only the lower conductance openings were fitted for GluN2A^N615K/2A triheteromers, which showed briefer open times than GluN2A^N615K diheteromers (2A N615K/2A: 1.83 ± 0.07 ms; n = 6 vs. 2A N615K 2.42 ± 0.19 ms, n = 11; t _12.5 = 2.9, P = 0.014).

Figure 4

GluN2A^N615K reduces Mg²⁺ blockade and current density in cultured neurons A–C, representative whole‐cell, voltage clamp recordings made from DIV 9 primary mouse cortical neurons transiently transfected with an inert control: β globin (A), GluN2A^WT (B) or GluN2A^N615K (C). Traces show the response to saturating NMDA (150 μm) and inhibition by Mg²⁺ (1 mm), before and after a 1 min application of ifenprodil (3 μm) (a selective GluN2B negative allosteric modulator). D, summary data showing percentage inhibition by ifenprodil. A one‐way ANOVA showed a significant effect of transfected subunit (F _2,42 = 46, P = 2.8 × 10^–11), with post hoc Bonferroni corrected t tests indicating that neurons transfected with GluN2A^WT showed lower ifenprodil sensitivity than control transfection cells (WT: 32 ± 3; n = 15; globin 75 ± 2; n = 15; t _21.9 = 10.7, P = 1.0 × 10^–9), as did neurons transfected with GluN2A^N615K (43 ± 4; n = 15; t _20.0 = 7.2, P = 1.9 × 10^–6) and with no difference between GluN2A and GluN2A^N615K (t _27.3 = 1.9, P = 0.19). These results confirm that the GluN2A^N615K subunits have been successfully trafficked to the surface in neurons. E, summary data showing Mg²⁺ blockade in the presence and absence of ifenprodil. Squares show blockade without ifenprodil; triangles show blockade with ifenprodil. Solid shapes represent means; hollow shapes are individual cells. A two‐way ANOVA showed a significant effect of transfected subunit (F _2,42 = 111, P = 2 × 10^–16) and the presence of ifenprodil (F _1,42 = 31, P = 1.9 × 10^–6), with a significant interaction (F _2,42 = 90, P = 6.7 × 10^–16). Post hoc Bonferroni corrected t tests showed lower Mg²⁺ blockade in the absence of ifenprodil in neurons transfected with GluN2A^N615K compared to GluN2A^WT (WT: 93 ± 1%; n = 15; N615K 52 ± 5%; n = 15; t _16.7 = 8.2, P = 9.7 × 10^–7), although there was no difference between GluN2A^WT and control (globin: 97 ± 1%; n = 15; t _23.3 = 2.2, P = 0.13). In the presence of ifenprodil, the reduction in Mg²⁺ blockade associated with GluN2A^N615K vs. wild‐type was even more marked (WT: 95 ± 1%; n = 15; N615K 31 ± 5%; n = 15; t _15.4 = 12.1, P = 8.4 × 10^–9). F, summary data showing current density in the presence and absence of ifenprodil. As with Mg²⁺ blockade, we found that current density was reduced in neurons transfected with GluN2A^N615K, and this effect was more pronounced in the presence of ifenprodil. A two‐way ANOVA showed a significant effect of transfected subunit (F _2,42 = 9.4, P = 0.0004) and the presence of ifenprodil (F _1,42 = 178, P = 2 × 10^–16), with a significant interaction (F _2,42 = 9.2, P = 0.0005). Post hoc Bonferroni corrected t tests showed lower current density in the absence of ifenprodil in neurons transfected with GluN2A^N615K compared to GluN2A^WT (WT: 58 ± 5 pA pF^–1; n = 15; N615K 40 ± 4 pA pF^–1; n = 15; t _27.3 = 2.7, P = 0.037), although there was no difference between GluN2A^WT and control (globin: 45 ± 4 pA pF^–1; n = 15; t _27.4 = 1.9, P = 0.22). In the presence of ifenprodil, the reduction in current density associated with GluN2A^N615K vs. wild‐type was even more marked (WT: 38 ± 4 pA pF^–1; n = 15; N615K 23 ± 3 pA pF^–1; n = 15) (t _27.8 = 3.3, P = 0.009).