Introduction

Glutamatergic N-methyl-D-aspartate receptors (NMDARs) are increasingly appreciated for their role in cognitive integration [1,2,3,4]. Mutations that reduce expression or function of NMDARs are a direct cause of intellectual disability [5, 6]. Relatively little is known, however, about whether there is variability across cellular domains in the functional impact of NMDAR genetic compromise. This is a critical area of exploration because NMDARs in different subcellular compartments play distinct neurophysiological roles [2, 7, 8] and experience distinct regulatory environments that may permit differing degrees of homeostatic compensation [9,10,11,12,13,14]. Understanding the relative vulnerability of NMDAR subpopulations to genetic disruption is essential to appreciate mechanisms of cognitive compromise and to identify new treatment approaches.

NMDARs are high affinity ligand-gated channels that are also voltage-dependent and act as coincidence detectors, requiring both glutamate-binding and depolarization to open. It is increasingly appreciated that glutamate travels beyond the synapse and this spillover increases upon strong or repeated stimuli [15,16,17]. Glutamate spillover can activate extrasynaptic NMDARs which are the substrate for integrative phenomena called dendritic plateau potentials. Stimulation of extrasynaptic NMDARs during dendritic plateau potentials allows enhanced cortical output in response to strong, repetitive, or converging inputs [2, 7, 18,19,20] and is critical for cognition and plasticity [2,3,4, 21]. While the fundamental properties of dendritic plateau potentials have been studied extensively, there has been no examination of how they are affected during NMDAR disruption in neurodevelopmental disorders.

Here, we investigated Grin1 knockdown (Grin1KD) mice with a profound deficiency in NMDAR receptor expression and binding [22, 23] and severe cognitive deficits [24]. This mouse is a hypomorph with reduced transcript expression caused by genetic insertion of a Neo cassette in intron 19 of the Grin1 gene [22], resulting in substantially reduced GluN1 protein expression in the brain (<10% of wild-type levels) [24, 25]. Consistent with previous work in this mouse and in other models of developmental cognitive disruption [24, 26], neuronal membrane properties are unaltered in Grin1KD compared to wild type. Furthermore, low-intensity stimuli revealed that neither AMPA receptor (AMPAR) nor NMDAR synaptic currents differed significantly across genotypes. However, a sizable deficit in the Grin1KD NMDAR response was revealed by stronger, repetitive, or pharmacological stimulation. The magnitude of this functional deficit was consistent with deficits observed anatomically in previous receptor binding work. To probe the repercussions of this primarily extrasynaptic deficit in NMDARs, we examined dendritic plateau potentials and found that Grin1KD mice are severely impaired in this integrative domain. In the final experiment, we tested the possibility of restoring cognitively-critical synaptic integration in adulthood, building on recent work showing that adult intervention to increase Grin1 expression achieves meaningful cognitive restoration [24]. We determine that dendritic plateau potentials can indeed be rescued by adult intervention to increase Grin1 expression. Taken together, this work reveals that integrative NMDARs are disproportionately sensitive to genetic disruption but amenable to restoration upon intervention in adulthood.

Materials and methods

Animals

All experiments were approved by the University of Toronto Animal Care and Use Committee and followed Canadian Council on Animal Care guidelines. Mice were group-housed and kept on a 12-h light cycle with food and water access ad libitum Mice for the initial experiments were generated from intercross breeding of C57Bl/6J Grin1 heterozygotes with 129X1Sv/J Grin1 heterozygotes, producing an F1 generation of Grin1KD (Grin1neo/neo) and wild-type (WT) littermate siblings used for experiments [22, 24]. Adult male and female mice were used for experiments (sex-matched and age-matched; age: 102 ± 5 days), with recordings from 56 WT and Grin1KD mice.

For the adult genetic rescue experiments, we used an additional 14 WT, Grin1KD, and Grin1rescue mice of both sexes. The generation of the line permitting adult rescue with tamoxifen is described in greater detail [24]. Starting in adulthood at 84 ± 6 days, all three genotypes of mice for the rescue experiment were treated with tamoxifen chow (TD.140425, 500 mg/kg, Envigo) ad libitum for 14 days. Electrophysiology experiments were conducted upon 38 ± 5 days washout from tamoxifen (sex-matched and age-matched; age:135 ± 3 days). Another cohort of mice similarly treated in adulthood were used after 1 month of washout to measure cortical NMDAR subunit protein levels (details in Supplementary Methods).

Electrophysiological recordings

Prefrontal brain slices were prepared as previously described [27, 28] and as detailed in the Supplementary Methods. Layer 5 pyramidal neurons in the medial prefrontal cortex, including cingulate and prelimbic regions, were visually identified by their pyramidal shape and prominent apical dendrite using infrared differential inference contrast microscopy. Unless otherwise indicated, whole-cell patch clamp electrodes contained potassium-gluconate patch solution (120 K-gluconate, 5 KCl, 2 MgCl2, 4 K2-ATP, 0.4 Na2-GTP, 10 Na2-phosphocreatine, 10 HEPES buffer with the final pH adjusted to 7.3 with KOH, in mM). Normal ACSF (128 NaCl, 10 D-glucose, 26 NaHCO3, 2 CaCl2, 2 MgSO4, 3 KCl, 1.25 NaH2PO4, in mM) was used except for specific experiments where we specify reduced concentration of magnesium ions. All ACSF and pipette solutions used for the following experiments are listed in the Supplementary Methods. Intrinsic membrane properties and excitability were assessed in current-clamp. All recordings were compensated for the liquid junction potential (14 mV).

Evoked excitatory postsynaptic currents

AMPAR-mediated evoked excitatory postsynaptic currents (eEPSCs) were measured in voltage-clamp at a holding potential of −75 mV. A bipolar stimulating electrode (FHC) was located in layer 2/3 for apical dendrite stimulation with pyramidal neurons in layer 5 recorded ~250 µm away from the electrode. For basal dendrite stimulation, the stimulating electrode was placed in the basal dendritic field ~100 µm from the soma of the recorded layer 5 pyramidal neuron. For both apical and basal stimulation paradigms, single pulses of 40 µs duration were delivered at 0.1 Hz, increasing in 10 µA increments. The AMPAR-mediated eEPSCs were analyzed as an average of at least 3 traces with Clampfit (Molecular Devices).

Isolated NMDAR–mediated evoked postsynaptic currents (ePSCs) were measured in voltage-clamp at a holding potential of +60 mV using specialized patch solution (details in Supplementary Methods) to block voltage-gated potassium and sodium channels. These recordings were performed in the presence of modified ACSF (1 mM MgSO4), AMPAR antagonists CNQX (20 µM) or NBQX (20 µM), and GABA receptor antagonists picrotoxin (PTX, 50 µM) and CGP52432 (CGP, 1 µM). Stimulation in the apical or basal dendritic fields were delivered as above. The NMDA receptor–mediated ePSCs were analyzed as an average of 3 traces with Clampfit (Molecular Devices) and D-APV (50 μM) was applied to confirm NMDAR responses.

Enhancing glutamate spillover with high-frequency stimulation

We measured the ability of Grin1KD NMDARs to respond to increasing amounts of glutamate spillover during single pulse and low to high frequency [29, 30] train stimulation at mild and strong stimulus intensities.

Enhancing glutamate spillover pharmacologically

To recruit extrasynaptic NMDA receptors in a different approach, 20 Hz train of mild stimuli was delivered in the apical location. Glutamate spillover was then enhanced with the application of TBOA (30 µM) and LY341495 (1 µM) to block glial glutamate reuptake and mGluR2/3 presynaptic autoreceptors respectively [31, 32].

Pharmacological stimulation with NMDA application

Total synaptic and extrasynaptic NMDAR currents were measured by bath application of NMDA (20 µM, 30 s) in a different subset of brain slices. Voltage-clamp recordings were performed with potassium-gluconate patch solution in a modified ACSF to reduce magnesium blockade (details in Supplementary Methods) as neurons were held at –75 mV. The AMPAR antagonist CNQX (20 µM) was also included. The peak amplitude of the NMDA receptor current was compared to baseline current using Clampfit. In a subset of experiments, D-APV (50 μM) was applied to verify NMDAR mediation of the inward currents.

NMDAR-dependent dendritic plateau potentials

Plateau potentials were generated by stimulation of the basal dendritic field of layer 5 pyramidal neurons, with the stimulating electrode placed within ~100 µm radius of the cell body. Plateau potentials were recorded in current-clamp at an initial membrane potential of −75 mV. They were generated with 10 stimuli at 50 Hz at the minimal stimulus intensity to evoke glutamatergic EPSCs [7, 33]. PTX (20 μM) and CGP52432 (1 μM) were present to block GABA receptors in combination with AMPAR blockers CNQX (20 μM) or NBQX (20 μM) to isolate NMDAR plateau potentials. D-APV (50 μM) was applied to confirm NMDAR dependence of plateau potentials.

Statistics

Statistical tests were performed in Prism 7 (Graphpad). Data are presented as mean ± SEM. Parametric statistical comparisons between responses from different groups of mice were determined using one-way ANOVA, or two-tailed unpaired t tests, and within-cell effects examined with two-tailed paired t tests. Where appropriate, interactions between genotype and other variables were assessed with two-way ANOVA or repeated-measure two-way ANOVA with post hoc Sidak-corrected t tests. Where 3 groups were treated with tamoxifen, the impact of adult intervention to rescue Grin1 expression was assessed with non-parametric Kruskal–Wallis test and Dunn’s post hoc tests due to the distribution of the data. Within cell pharmacological investigations for this dataset were therefore compared with a non-parametric paired test.

Results

Since Grin1KD mice have extremely low NMDAR protein expression (see Supplementary Table S1), we investigated the differential vulnerability of synaptic and extrasynaptic NMDARs to genetic disruption. We performed ex vivo electrophysiology in major output pyramidal neurons of prefrontal cortex from mice deficient in the obligate NMDAR subunit (Grin1KD) and their wild-type (WT) littermate controls (Fig. 1A). We found that neuronal properties, including resting membrane potential, input resistance, capacitance, spike amplitude, and rheobase did not differ significantly between the genotypes (Supplementary Table S2). The input-output relationship showed the expected effect of current (F3,123 = 307.6; p < 0.0001; Fig. 1B, C), but did not differ significantly between the genotypes (F1, 41 = 0.4525; p = 0.50), nor show an interaction F3,123 = 1.123; p = 0.34).

Fig. 1: Wild-type and Grin1KD have similar intrinsic excitability and postsynaptic AMPA and NMDA receptor responses.
figure 1

A Schematic of the prefrontal cortex with electrophysiological recording from layer 5 pyramidal neuron. B Example current-clamp traces from WT (left) and Grin1KD (right) in response to depolarizing current steps through the recording pipette. C Input-output graphs of spike frequency (Hz) in current-clamp for WT (n = 20) and Grin1KD (n = 25). D Schematic of recording pipette with extracellular stimulating electrode for assessment of postsynaptic currents. E Example voltage-clamp traces (Vh −75 mV) show inward AMPA receptor (AMPAR)-mediated electrically-evoked excitatory postsynaptic currents (eEPSC) in WT and Grin1KD. F Graph illustrates that WT (n = 15) and Grin1KD (n = 13) both show the expected relationship between stimulus strength eEPSC amplitude but no significant effect of genotype nor interaction for AMPAR eEPSCs. G Example voltage-clamp traces (Vh +60 mV) show outward NMDA receptor (NMDAR)-mediated evoked postsynaptic currents (ePSCs), isolated with AMPAR and GABA receptor blockade and recorded with pipette solution to internally block voltage-gated potassium and sodium channels. H Graph illustrates that both WT (n = 10) and Grin1KD (n = 15) show the expected relationship between stimulus strength and NMDAR ePSC amplitude, but no significant effect of genotype nor interaction for these ePSCs. Data are represented as mean ± SEM.

Full size image

Preserved synaptic glutamatergic responses in Grin1KD mice

To test AMPAR synaptic responses from stimulation in the apical dendritic field, we recorded from layer 5 pyramidal neurons at a holding potential of −75 mV and applied electrically-evoked stimulation in layer 2/3 (Fig. 1D). There was no significant difference between genotypes in the electrical stimulus required to elicit the minimal response (t27 = 0.3; p = 0.8), and response amplitudes were similar in both genotypes (Fig. 1E, F). We observed the expected effect of stimulus strength on response amplitude (F3,115 = 11.04; p < 0.0001), but not an effect of genotype (F1,115 = 2.354; p = 0.13), nor an interaction between genotype and stimulus strength (F3, 115 = 0.20; p = 0.9). These results show that AMPAR-mediated synaptic transmission in response to low-intensity stimulation is similar in WT and Grin1KD prefrontal cortex. Furthermore, miniature EPSCs measured in the presence of TTX did not differ in amplitude or frequency between WT and Grin1KD (Supplementary Fig. S1), confirming that AMPAR synaptic transmission is not altered in Grin1KD.

To isolate NMDAR synaptic responses, we next recorded evoked currents at a holding potential of +60 mV in the presence of AMPAR and GABAA receptor antagonists, using recording pipette solution designed to block voltage-gated potassium and sodium channels. Again, there was no genotype difference in the minimal current required to elicit a response (t24 = 0.4; p = 0.71), nor in response amplitudes across an increasing range of stimuli (Fig. 1G, H). We observed the expected effect stimulus strength on response amplitude (F3,87 = 12.53; p < 0.0001), but no effect of genotype (F1,87 = 0.1926; p = 0.66), nor an interaction between genotype and stimulus strength (F3,87 = 0.1485; p = 0.93). There was no effect of sex nor an interaction between sex and genotype on the amplitude of NMDAR ePSCs (Interaction: F1,54 = 0.081, p = 0.78; Sex: F1,54 = 1.838, p = 0.181). Consistent with the intended NMDAR-mediation of these ePSCs, the evoked currents were strongly suppressed by the selective NMDAR antagonist, D-APV (50 µM): t(10) = 6.1, p = 0.0001). The GluN2B selective antagonist ifenprodil also caused significant suppression of NMDAR ePSCs (F1,25 = 5.879, p = 0.023) in both WT and Grin1KD without any genotype effect (F1,25 = 0.003, p = 0.957). These results demonstrate that the amplitudes of isolated NMDAR currents are similar between genotypes in response to low-intensity stimulation. This unexpected finding was surprising because of the prominent differences in the expression of the obligate subunit and NMDAR binding between the genotypes in previous reports [22, 23].

Deficient extrasynaptic NMDAR responses in Grin1KD mice

We hypothesized that deficits are more prominent in the extrasynaptic NMDAR subpopulation, which can be recruited by stronger electrical stimulation to increase glutamate spillover [34, 35]. Therefore, we delivered stronger single stimuli (80 µA) (Fig. 2A, B). In contrast to the relatively-homogenous effects of low-intensity stimulation, stronger stimuli elicited a significant and substantial difference in NMDAR ePSC amplitude between genotypes (WT: 599 ± 105 pA, n = 9; Grin1KD: 339 ± 62 pA, n = 16; t23 = 2.29; p = 0.032). This result prompted a detailed characterization of extrasynaptic NMDAR in Grin1KD mice using multiple approaches.

Fig. 2: Extrasynaptic NMDARs are not recruited in Grin1KD mice during glutamate spillover.
figure 2

A Example voltage-clamp traces show outward NMDAR ePSCs evoked by strong stimulation. B Scatter graph shows the amplitude of NMDAR ePSCs at strong stimulus strength is significantly reduced in Grin1KD (*p < 0.05). C Voltage-clamp traces show NMDAR-mediated outward currents during AMPAR blockade in WT (left) and Grin1KD (right) evoked by a high-frequency stimulus train (100 Hz, 10 pulses) at mild and stronger stimulus intensities. D Bar graph shows the ratio of NMDAR response area at mild and stronger stimulus intensities at single pulse (1P), 20 Hz, and 100 Hz train stimulation. There is a significant effect of genotype, with Grin1KD showing deficient NMDAR responses compared to WT at 100 Hz stimulation (*p < 0.05). E Voltage-clamp traces show NMDAR-mediated outward currents during AMPAR blockade in WT (above) and Grin1KD (below) evoked by a stimulus train (20 Hz, 10 pulses) under baseline conditions and with the addition of TBOA and LY341495 to enhance glutamate spillover (red line). The dotted line illustrates the consistency of the first evoked postsynaptic current. NMDAR responses isolated with AMPAR and GABA receptor blockade. F The bar graph shows the significant potentiation of the peak amplitude in the stimulus train under conditions of enhanced glutamate spillover for WT (n = 4) but not Grin1KD (n = 6); significant interaction of genotype and spillover condition (***p < 0.001). G Voltage-clamp traces show bath application of NMDA to pharmacologically stimulate NMDAR in WT (left) and Grin1KD (right). H The scatter graph shows the peak amplitude of pharmacologically-elicited inward NMDA currents is significantly lower in Grin1KD (n = 21) compared to WT (n = 23) (****p ≤ 0.0001). Data represented as mean ± SEM.

Full size image

To physiologically evoke glutamate spillover and recruit extrasynaptic NMDARs, we used repetitive stimulus trains delivered in layer 2/3. High-frequency stimulation (100 Hz) is particularly known to stimulate glutamate spillover and activate extrasynaptic NMDARs [29, 30]. We measured the ability of Grin1KD NMDARs to proportionately respond to increasing amounts of glutamate spillover during 100 Hz train stimulation at mild vs. stronger stimulus intensities by quantifying the area of the response. Grin1KD neurons showed deficient integration at 100 Hz stimulation compared to WT and were unable to proportionately increase their responses at stronger stimulus strengths (Fig. 2C, D, effect of genotype: F1,36 = 5.478, p = 0.025). The train NMDAR response evoked by strong 100 Hz stimulation is significantly reduced in Grin1KD (Sidak’s post hoc test: t36 = 2.804, p = 0.024). Single pulse or 20 Hz stimulation that release far lesser glutamate compared to 100 Hz, did not reveal a major deficit in Grin1KD (1 pulse: t36 = 0.434, p = 0.962; 20 Hz: t36 = 0.813, p = 0.807).

To further probe extrasynaptic NMDAR involvement, we examined whether pharmacologically boosting glutamate spillover with TBOA and suppression of autoinhibition with LY341495 [31, 32] even at lower frequency stimulation (20 Hz) would reveal a deficit in Grin1KD (Fig. 2E, F). In wild-type mice, repetitive stimulation led to summation of postsynaptic responses, yielding a higher peak response compared to the first input, with further potentiation of peak response caused by glutamate spillover in the presence of TBOA. In Grin1KD, by contrast, pharmacologically boosting spillover did not increase the peak response, leading to a significant interaction between the genotype and spillover conditions (F2,16 = 11.37; p = 0.0008). Repetitive stimulation in the presence of TBOA significantly potentiated the peak response compared to the first stimulus in WT (Sidak’s post hoc test, p = 0.0001) but not in Grin1KD (p = 0.2). These results suggest a lack of extrasynaptic NMDARs in Grin1KD available to be recruited by glutamate spillover.

To reach an even broader group of extrasynaptic receptors, we activated NMDARs using direct pharmacological manipulation with the agonist NMDA. For these experiments, we bath-applied NMDA to the prefrontal slice in the presence of AMPAR antagonist CNQX and low-Mg2+ to permit NMDAR activation at a holding potential of −75 mV. As anticipated [24], pharmacological NMDAR currents were substantially and significantly reduced in Grin1KD mice compared to their littermates (WT: 87 ± 5 pA, n = 23; Grin1KD1: 24 ± 2 pA, n = 21; t42 = 10.6, p = 0.0001; Fig. 2G, H). These pharmacologically-elicited inward currents were suppressed by D-APV (50 µM; WT: n = 5, t4 = 6.2, p = 0.003; Grin1KD mice: n = 7, t6 = 3.5, p = 0.01). Of note, the 3-fold genotype difference in the response to bath NMDA mirrors the difference in NMDAR binding observed in prefrontal cortex in Grin1KD compared to wild-type controls [24].

Stronger, repetitive, and pharmacological stimulations that recruit extrasynaptic NMDARs all unmask genotype differences between the wild-type littermates and Grin1KD mice, consistent with the interpretation that Grin1KD mice have a specific and disproportionate deficit in extrasynaptic NMDARs.

Impaired dendritic plateau potentials in Grin1KD mice: further evidence for extrasynaptic NMDAR disruption

Dendritic plateau potentials can be evoked by spillover of glutamate onto extrasynaptic NMDARs under conditions of high-frequency repetitive stimulation of inputs to basal dendrites [7, 33]. This integrative phenomenon depends on the recruitment of extrasynaptic NMDARs (Fig. 3A) and would be vulnerable if this population were compromised (Fig. 3B). Dendritic plateau potentials are considered an important cognitive substrate to link multiple streams of incoming information and generate burst firing [16, 19, 20, 36], an output signal predicted to exert stronger downstream consequences [37, 38]. Deficient extrasynaptic NMDARs are predicted to have profound consequences for such signaling [7, 33].

Fig. 3: Deficits in extrasynaptic NMDA receptors disrupt integrative basal dendrite plateau potentials in Grin1KD.
figure 3

Schematics depict hypothesized differences in extrasynaptic NMDA receptors (NMDARs) between (A) WT and (B) Grin1KD. The initial stimulus (1.) yields glutamate spillover that permits priming of extrasynaptic NMDARs during the inter-stimulus interval (2.) making them available to be activated immediately by depolarization from the next stimulus (3.). This form of integration is sufficient to yield a dendritic plateau potential in response to repeated mild stimulation and is typically measured in current-clamp. Note that this is a simple model and extrasynaptic NMDARs could also include a population within spines but with different dynamics or biochemical modifications [51,52,53, 78, 79]. C Inset: Schematic of layer 5 pyramidal cell recording with stimulation in the basal dendritic field. Averaged current-clamp recordings of excitatory responses to repeated minimal stimulation (50 Hz, 10 pulses, 30–40 µA) in WT (black, n = 8) and Grin1KD (gray, n = 8). NMDAR-mediated dendritic plateaus isolated with AMPA and GABA receptor blockade. D Graph of peak plateau amplitude illustrates that basal dendrite integration is substantially reduced in Grin1KD mice compared to WT (****p < 0.0001). Data represented as mean ± SEM.

Full size image

To examine basal dendrite plateau potentials in both genotypes, we recorded from layer 5 pyramidal neurons while electrically stimulating inputs in the basal field. AMPAR eEPSCs evoked by basal dendritic stimulation were similar between wild-type and Grin1KD mice, had the expected effect of stimulus strength (F2,66 = 12.7; p = 0.0001), but no effect of genotype (F1, 66 = 0.148, p = 0.7) nor interaction between genotype and stimulus strength (F2,66 = 0.127, p = 0.88, data not shown). Next, we recorded NMDAR plateau potentials in current-clamp in response to trains of stimuli (50 Hz, 10 pulses) in the presence of AMPA and GABA receptor blockade and observed a marked genotype difference (Fig. 3C, D). While wild-type neurons showed clear NMDAR plateau potentials (peak amplitude: 2.15 ± 0.27 mV, n = 8), the train of stimuli did not elicit dendritic plateau potentials in Grin1KD neurons (0.48 ± 0.10 mV, n = 8; t14 = 5.8, p < 0.0001; Fig. 3C, D). This genotype effect was seen in both sexes with no interaction between genotype and sex (Interaction: F1,12 = 0.143, p = 0.711; Sex: F1,12 = 0.061, p = 0.81). Plateau potentials in wild-type neurons could be eliminated by the NMDAR antagonist APV (significant genotype × D-APV interaction: F1,7 = 7.53, p = 0.029; peak amplitude at baseline vs. APV in WT: t7 = 4.12, p = 0.009, Sidak’s post hoc test, data not shown). Grin1KD prefrontal pyramidal neurons have a significant deficit in dendritic plateau potentials compared to those recorded in brain slices from wild-type littermate mice. This measure confirms a profound physiological impact of insensitivity to glutamate spillover in Grin1KD.

Electrophysiological examination of consequences of adult Grin1 rescue

To identify whether a genetic intervention in adulthood could restore crucial aspects of NMDAR function in Grin1KD mice, we tested a tamoxifen-induced Cre-based approach to rescue Grin1 expression in adult mice (as described in Supplementary Methods). This has previously been shown to increase prefrontal NMDAR radioligand binding and reverse key behavioral deficits [24]. These mice are referred to as Grin1rescue mice. In order to ensure equivalent comparison, all 3 genotypes (WT, Grin1KD, Grin1rescue) were treated with tamoxifen at the same age and for the same time course. Western blot of synaptoneurosomal protein extracts confirmed that NMDAR subunit protein levels are significantly reduced in the Grin1KD and increased in Grin1rescue upon tamoxifen treatment (Supplementary Fig S2). Intrinsic electrophysiological properties of prefrontal layer 5 pyramidal neurons including the resting membrane potential, input resistance, capacitance, and action potential amplitude were not significantly different across the tamoxifen-treated, littermate wild-type, Grin1KD and Grin1rescue mice (Supplementary Table S3). Consistent with Fig. 1, synaptic NMDAR ePSCs evoked by minimal stimuli were not significantly different in amplitude across the three genotypes (Supplementary Fig. S3). Furthermore NMDAR ePSC kinetics were also similar (tau decay slow: F2,23 = 1.74, p = 0.197) suggesting that tamoxifen treatment to rescue NMDARs did not result in lasting changes in NMDAR molecular composition [39, 40].

Adult intervention rescues dendritic plateau potentials in prefrontal cortex

To identify whether an adult intervention to boost Grin1 expression can restore dendritic plateau potentials in mice after a lifelong deficit, we examined NMDAR plateau potentials in the three groups of tamoxifen-treated mice. Under these conditions, Grin1KD mice again showed significantly smaller NMDAR plateau potentials compared to wild-type mice, but there was a striking increase in the amplitude of the NMDAR plateau potentials in the Grin1rescue mice compared to the Grin1KD (Fig. 4B). The distribution of the data prompted non-parametric analysis (Kruskal–Wallis test = 11.30, p = 0.003; Dunn’s post hoc tests: WT vs. Grin1KD, Z = 3.18, p = 0.004; Grin1KD vs. Grin1rescue, Z = 2.55, p = 0.032; but no significant difference WT vs. Grin1rescue, Z = 0.76, p = 0.99). Correspondingly, dendritic plateau potentials were significantly suppressed by D-APV in both WT and Grin1rescue mice (Wilcoxon matched-pairs signed rank test: p = 0.016, n = 7, data not shown).

Fig. 4: Adult genetic intervention to boost Grin1 expression restores dendritic plateau potentials.
figure 4

A Grin1rescue schematic illustrates strategy for enhancing Grin1 expression and increasing NMDAR density in adulthood (adapted from Mielnik et al. [24]). All mice are treated with tamoxifen in adulthood but only in Grin1rescue will this treatment trigger Cre expression and lead to the excision of the Neo cassette to increase Grin1 mRNA, NMDAR radioligand binding, and cognitive performance significantly [24]. B Averaged current-clamp recordings of responses to repeated mild stimulation (50 Hz, 10 pulses, 40 µA) in the three genotypes of mice all treated with tamoxifen in adulthood: WT (black, n = 17), Grin1KD (gray, n = 18), and Grin1rescue (red, n = 21). C Graph illustrates that basal dendrite integration is greatly reduced in Grin1KD compared to WT and is restored in the Grin1rescue (**p < 0.01, *p < 0.05). Data represented as mean ± SEM.

Full size image

Here we show that increasing expression of the obligate NMDAR subunit in adulthood is sufficient to restore dendritic plateau potentials, consistent with the significant behavioral improvement observed previously [24]. These findings suggest that the boost in Grin1 expression results in an increase in functional extrasynaptic NMDARs, as illustrated in the working model in Fig. 5. This work demonstrates the potential for adult treatments to restore NMDAR function critical for signal integration.

Fig. 5: Working model schematics for prefrontal synapses across the three genotypes.
figure 5

In wild-type mice (WT), prefrontal neurons have both synaptic and extrasynaptic NMDARs. In Grin1KD mice, there is relative preservation of synaptic NMDARs and disproportionate compromise of extrasynaptic NMDARs. In Grin1rescue mice, adult manipulation to boost Grin1 expression is successful and sufficient to restore extrasynaptic NMDARs needed for dendritic integration of repetitive mild stimuli.

Full size image

Discussion

Our data reveal that developmental deficiency in the obligate Grin1 subunit leads to a profound bias in NMDAR function in the prefrontal cortex. The subpopulation of synaptic NMDARs recruited by mild stimulation shows markedly greater functional preservation than the extrasynaptic receptors recruited by stronger, repetitive, or pharmacological stimuli. To probe the physiological implications of this uneven pattern of NMDAR disruption, we examined dendritic plateau potentials and identified striking deficits in this integrative phenomenon in Grin1KD mice. Lastly, we discovered that genetic rescue of Grin1 expression restores this form of integrative neurophysiology in the mature brain. Our work suggests that, in mice with NMDAR insufficiency, the window for functional improvement remains open into adulthood.

Broader relevance of this model of NMDAR insufficiency

The Grin1KD mouse has been used as a model to study aspects of schizophrenia, autism spectrum disorder, and most recently as a general model for variants in Grin1 that cause GRIN disorder [41, 42]. Grin1KD mice most closely model Grin1 haploinsufficiency, since they have a genetic modification causing a dramatic reduction in the amount of GluN1 protein and NMDAR without a change in amino acid sequence or in the biophysical properties of the receptor. The Grin1KD mouse expresses low levels of the obligate NMDAR subunit and greatly reduced cortical NMDARs, as measured by radioligand binding [22,23,24]. Understanding the cellular electrophysiological consequences of this substantial deficit is relevant beyond GRIN disorder, since perturbed NMDAR levels are also a key contributing factor to the symptoms of other neurodevelopmental disorders, including those arising from variants in DLG3, SHANK3, and FMRP [5, 43,44,45,46,47]. Our investigation of Grin1KD mice suggest that patients with reduced NMDARs are likely to have a functional deficit in extrasynaptic NMDAR, with a relative preservation of synaptic receptors. Given the historical focus on synaptic NMDAR for neural communication and extrasynaptic receptors for excitotoxicity, it is remarkable that the profound cognitive impairments of Grin1KD mice could be attributed to extrasynaptic deficits. It is also striking that rescue experiments in adulthood, which improve executive function and sensory integration [24], appear to boost functioning of this extrasynaptic population to restore dendritic plateau potentials, a measure of integrative neurophysiology. Future work needs to examine whether adult rescue of Grin1 is sufficient to restore extrasynaptic NMDAR function across brain circuits or whether earlier age at rescue might be required for proper rescue in specific brain circuits. This combination of findings urges greater attention to extrasynaptic NMDARs in developmental disorders and their treatment.

New perspectives on extrasynaptic NMDARs and their integrative role

Extrasynaptic NMDARs, located perisynaptically [10], or non-synaptically on dendritic shafts [10], used to be predominantly described in terms of pathology and their role in activating excitotoxic cell death pathways [48]. However, this view is shifting as growing preclinical research demonstrates the physiological conditions under which extrasynaptic NMDARs are recruited [15,16,17]. This recent body of work points to their role in normal brain function via generation of dendritic plateau potentials [3, 4, 49]. Extrasynaptic receptors bind the small amount of glutamate that escapes the synapse, to become “primed” and ready for rapid activation by subsequent depolarizing input(s). NMDARs on small dendritic branches are thus positioned to detect the activation of multiple synapses close together in space and time. Such temporal and spatial integration is required to generate dendritic plateau potentials [2, 7, 18, 19]. These NMDAR-mediated integration events trigger burst firing [7, 19], a robust neuronal response [37, 38], thought to be essential for behavior-evoked network activity [4, 20, 50]. Our results indicate that developmental disorders with reduced NMDARs are likely to have compromised neurophysiological integration resulting from disrupted extrasynaptic NMDAR population. Intriguingly, an adult intervention yielding an increase in Grin1 expression and NMDAR radioligand binding [24] (to ~60% of wild-type), restores the neurophysiological phenomenon of dendritic plateau potentials. This integrative recovery is consistent with the marked improvement of cognitive performance observed after treatment in adulthood [24].

Subcompartment-specific NMDAR alterations: potential mechanisms and caveats

Microscopic examination of synaptic substructure has shown that NMDARs localized within dendritic spines are organized in highly heterogenous and dynamic nanodomains [51,52,53] that are functionally segregated [54, 55]. By examining function, we show that despite substantial reduction in GluN1 protein level, there is remarkable preservation of synaptic but not extrasynaptic NMDARs in Grin1KD. Protein levels or localization alone cannot explain this disparity. Reduction and preservation of NMDAR function in extrasynaptic vs. synaptic regions respectively may involve a reorganization of nanodomain architecture or stability [56, 57]. Disparate functional consequences across NMDAR populations have been observed in response to different perturbations [58,59,60,61,62]. Research in cell systems demonstrates that NMDARs move between synaptic and extrasynaptic compartments upon pharmacological manipulation [58,59,60,61], or exposure to antibodies from people with anti-NMDAR encephalitis [63]. Receptor trafficking, however, is not the only path to achieve divergent functional outcomes for synaptic and extrasynaptic NMDAR populations. Multiple mechanisms for functional NMDAR enhancement display compartmental specificity, including post-translational modification pathways [64, 65], co-agonism [66,67,68], and mechanisms of receptor desensitization [69,70,71]. The functional preservation of synaptic NMDAR responses in Grin1KD mice may therefore be caused by multiple complex mechanisms, and not necessarily reflect wild-type levels of receptor density in this compartment [24]. Future work also needs to examine such mechanisms that operate to restore extrasynaptic NMDAR function after adult genetic rescue.

While NMDARs are the focus of a large body of work in models of neurodevelopmental disorders, many characterizations use relatively strong stimuli under conditions where “synaptic” measures may inadvertently include a broader population. Here, we pursued carefully calibrated electrical stimulation under several conditions to isolate synaptic NMDARs from their extrasynaptic counterparts. Our strategy was adopted due to the inherent challenges in separating these contributions with pharmacological tools [72, 73]. This problem is particularly difficult to overcome in the prefrontal cortex, where synaptic and extrasynaptic NMDARs show a high degree of overlap in molecular composition and pharmacological affinities [74, 75], complicating specific manipulations. Differentiating synaptic and extrasynaptic NMDAR populations remains challenging, but it will be an increasingly-important focus for future work into the mechanisms of cognitive compromise arising from NMDAR insufficiency.

Clinical relevance and future implications

Current treatments for cognitive disability arising from genetic disruption of NMDARs focus on supportive therapies because it is assumed that lifting cognitive restrictions hard-wired by abnormal brain development is impossible. However, this assumption has recently been challenged. Promising preclinical data [24, 76, 77] suggest the potential for cognitive improvement, even when intervention is delayed until adulthood. If adult treatments are to be seriously pursued, it is essential to appreciate what neural components are functionally compromised and what may be preserved. Here, we address a critical knowledge gap about the specific cellular and circuit mechanisms by which genetic NMDAR disruption impairs cognitive function. We demonstrate that two important NMDAR subpopulations do not suffer equal consequences from genetic disruption of the obligate subunit Grin1. Extrasynaptic NMDARs are disproportionately compromised with resulting disruption of the integrative capacity required for the generation of dendritic plateau potentials. This deficit, strikingly, proves amenable to rescue by intervention in adulthood. Developing effective treatments for the cognitive impairments caused by NMDAR disruption requires the identification of the most efficient targets. Our discovery underscores the need for research into additional approaches to safely enhance extrasynaptic NMDAR functioning. Overall, our findings suggest that deficient integrative mechanisms are amenable to improvement, even with adult intervention.