Heterogeneous clinical and functional features of GRIN2D-related developmental and epileptic encephalopathy

Abstract

N-methyl d-aspartate receptors are ligand-gated ionotropic receptors mediating a slow, calcium-permeable component of excitatory synaptic transmission in the CNS. Variants in genes encoding NMDAR subunits have been associated with a spectrum of neurodevelopmental disorders. Here we report six novel GRIN2D variants and one previously-described disease-associated GRIN2D variant in two patients with developmental and epileptic encephalopathy. GRIN2D encodes for the GluN2D subunit protein; the GluN2D amino acids affected by the variants in this report are located in the pre-M1 helix, transmembrane domain M3, and the intracellular carboxyl terminal domain. Functional analysis in vitro reveals that all six variants decreased receptor surface expression, which may underline some shared clinical symptoms. In addition the GluN2D(Leu670Phe), (Ala675Thr) and (Ala678Asp) substitutions confer significantly enhanced agonist potency, and/or increased channel open probability, while the GluN2D(Ser573Phe), (Ser1271Phe) and (Arg1313Trp) substitutions result in a mild increase of agonist potency, reduced sensitivity to endogenous protons, and decreased channel open probability. The GluN2D(Ser573Phe), (Ala675Thr), and (Ala678Asp) substitutions significantly decrease current amplitude, consistent with reduced surface expression. The GluN2D(Leu670Phe) variant slows current response deactivation time course and increased charge transfer. GluN2D(Ala678Asp) transfection significantly decreased cell viability of rat cultured cortical neurons. In addition, we evaluated a set of FDA-approved NMDAR channel blockers to rescue functional changes of mutant receptors. This work suggests the complexity of the pathological mechanisms of GRIN2D-mediated developmental and epileptic encephalopathy, as well as the potential benefit of precision medicine.

Keywords: GluN; NMDA receptor; channelopathy; functional genomics; glutamate receptor.

Figure 1

EEG features of patients with GRIN2D variants. (A) Proband 2 (Val667Ile) shows multiple spike predominately in bilateral post-lobe (1 year 5 months of age). (B) Proband 6 (Ala678Asp) shows spike and spike-wave in bilateral Rolandic region and mid-line during awake periods (3 years 11 months of age). (C) Proband 7 (Ser1271Leu) shows hypsarrhythmia (3 years).

Figure 2

Brain MRI of patients with GRIN2D variants. (A) Proband 3 (Leu670Phe) showed cortical atrophy, global loss of white matter volume and enlargement of the lateral ventricles; compare MRI at 5 months old (left two panels) to 2 years old (right two panels). (B) Proband 4 (Leu670Phe) showed mild cortical atrophy. (C) Proband 6 (Ala678Asp) had normal MRI left: sagittal T₁. middle: coronal T₂. right: coronal T₂ at 3 years 5 months of age. (D) Proband 7 (Ser1271Leu) had normal MRI left: sagittal T₁, middle: coronal T₁, right: coronal T₂ at 14 months old.

Figure 3

Location of seven missense variants in GRIN2D/GluN2D. (A) Intolerance analysis of genetic variation across functional domains within GRIN2D/GluN2D. observed/expected mutant ratios (OE-ratio) below the 10th percentile are in red and indicate the regions under purifying selection (Traynelis et al., 2017). Residue Ser573 is located in pre-M1 helix, linker region between agonist binding domain S1 and transmembrane domain M1. Residues Val667, Leu670, Ala675, and Ala678 are located in transmembrane domain M3, one of the least tolerant regions. Residues Ser1271 and Arg1313 are located in intracellular CTD. The regions highlighted with a dashed box have insufficient data available. ATD = amino-terminal domain; S1, S2 comprise the ABD (agonist binding domain); M1, M2, M3, M4 comprise the transmembrane domain; CTD = carboxyl-terminal domain. (B) The residues of Ala675 and Ala678 are highly conserved across other vertebrate species and all other GluN subunits. The residues Ser573, Val667 and Leu670 are highly conserved across other vertebral species and all other GluN2 subunits, but not the GluN1 subunit. The residues Ser1271 and Arg1313 are conserved in most vertebral species evaluated, but not in other GluN subunits. (C) Homology model of the GluN1/GluN2D receptor built (Li et al., 2016) from the GluN1/GluN2B crystallographic data (PDB: 4PE5; Lee and Chung, 2014; Karakas and Furukawa, 2014) is shown as ribbon structure overlaid by space-filled representation. The GluN1 subunit is yellow and the GluN2D subunit is blue. The positions of p.Ser573Phe, Val667Ile, Leu670Phe, Ala675Thr and Ala678Asp are highlighted by red in the pre-M1 helix and M3 domain. (D) The tetrameric GluN1/GluN2D transmembrane region (ATD and ABD removed) viewed from the side. (E) Potential interaction between GluN2D pre-M1, GluN1-M3, and GluN2D-M4 top-down through the pore (left), side view of one GluN2D M3 domain (middle), and M3 domains of two GluN1 and two GluN2D viewed from the bottom through the pore (right). Wild-type residues Ser, Val, Leu and Ala are green, and mutant residues Phe, Ile, Thr, and Asp are shown as red.

Figure 4

The mutant GluN2D receptors change NMDAR pharmacological and biophysical properties, and reduce cell surface expression. (A–C) Composite concentration-response curves for glutamate (A, in the presence of 30–100 μM glycine), glycine (B; in the presence of 10 μM glutamate), and d-serine (C; in the presence of 10–30 μM glutamate) at V_HOLD −40 mV. Smooth curves are Equation 1 fitted to the data. (D) Composite concentration-response curves for Mg²⁺ (in the presence of 100 μM glutamate and 100 μM glycine) at holding potential −60 mV. Smooth curve is Equation 2 fitted to the data. (E) Summary of proton sensitivity, evaluated by current ratio at pH 6.8 to pH 7.6 at holding potential −40 mV. (F) Summary of calculated channel open probability (P_OPEN) evaluated by the degree of MTSEA potentiation at holding potential −40 mV (see ‘Materials and methods’ section; ND = determined). **P < 0.01, ***P < 0.001 compared to wild-type GluN2D, one-way ANOVA with Dunnett’s multiple comparison test. (G) Representative whole cell voltage clamp current recordings (normalized) are shown in response to application (1.5-s duration) of 10 μM glutamate (10 μM glycine was in all solutions) from wild-type GluN2D- (grey) and GluN2D-L670F- (black) transfected HEK293 cells. (H and I) Summary of current amplitudes and weighted deactivation time course. *P < 0.05, **P < 0.01, one way ANOVA, with Dunnett’s multiple comparisons. Fitted parameters are given in Table 2. (J) Representative plots of nitrocefin absorbance (optical density, OD) versus time are shown for HEK293 cells expressing wild-type or mutant GluN2D. β-lac-GluN1 was present in all conditions except control cells. (K) The slopes of OD versus time were averaged (n = 4–6 independent experiments) and graphed as percentages of wild-type for the ratio of surface/total. Data in all composite concentration-response curves (A–D) are mean ± SEM. Data in all bar graphs (E, F, H, I and K) are mean ± 95% CI (confidence interval). Data were analysed by one-way ANOVA with Dunnett’s multiple comparison test compared to wild-type (surface/total ratio, *P < 0.05, **P < 0.01, ***P < 0.001).

Figure 5

The GluN2D-A678D variant induces neurotoxicity. Transfection of GluN2D(Ala678Asp) into cultured cortical neurons reduces cell viability, which can be prevented by the low affinity NMDAR antagonist memantine. (A) The mean cell viability determined by a Luciferase assay is shown as a per cent of control group (Vector) (% control) with wild-type GluN2D (–), 83%; wild-type GluN2D, 86%; GluN2D-A678D, 55%; GluN2D-A678D + mem, 77%. mem = 50 μM memantine; *P < 0.05; one-way ANOVA with Dunnett’s multiple comparison test (bar graph is mean ± 95% CI). (B) Dendritic bleb analysis indicated no significant changes between wild-type and GluN2D-A678D. (C) Confocal images display morphological features of cultured rat cortical neurons transfected with GFP-N1 and either wild-type GluN2D (left) and GluN2D-A678D (right), respectively. Note the presence of cellular debris in the mutant-transfected cells (asterisk), indicative of toxicity and as quantified in A. Scale bars = 20 μm. Experiments were repeated in four independent culture dates.

Figure 6

Effects of FDA-approved NMDAR channel blockers on wild-type and mutant GluN1/GluN2D receptors. Composite concentration-response curves of FDA-approved NMDAR antagonists were evaluated by TEVC recordings from Xenopus oocytes in the presence of 100 μM glutamate and 100 μM glycine at holding potential of −40 mV. (A) memantine, (B) dextromethorphan, (C) dextrorphan, and (D) ketamine. Data are mean ± SEM. Smooth curves are Equation 2 fitted to the data.