The NMDA receptor subunit GluN2D is a target for rapid antidepressant action

doi:10.21203/rs.3.rs-4807870/v1

Download PDF

Article

The NMDA receptor subunit GluN2D is a target for rapid antidepressant action

https://doi.org/10.21203/rs.3.rs-4807870/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Ketamine is the first glutamatergic agent in clinical use for major depression. The underlying mechanism and primary target of ketamine are unknown; further research is urgently needed to develop more specific interventions with fewer side effects and better treatment outcomes for severely affected patients. Ketamine is a noncompetitive antagonist of the glutamatergic N-methyl-D-aspartate (NMDA) receptor, a heterotetramer composed of two GluN1 and two GluN2 subunits. Here, we show that ketamine preferentially targets GluN2D-containing NMDA receptors on interneurons, and that selective GluN2D antagonism alone is sufficient to produce rapid antidepressant effects. We used ketamine, selective GluN2D inhibitors, GRIN2D-siRNA and chemogenetic approaches in hippocampal brain slices and in vivo in mice. We found that GluN2D antagonism inhibited NMDAR currents in interneurons but not in pyramidal cells. GluN2D-mediated recruitment of GABAergic interneurons powerfully controls feedback and feed-forward inhibitory circuits to moderate hippocampal network activity and synaptic plasticity. In a mouse model of depression, GluN2D inhibition recovered synaptic excitation-inhibition balance, reversed long-term potentiation deficits and restored synaptic and AMPAR density toward a naïve state. GluN2D antagonism could fully mimic the cellular and behavioral antidepressant actions of ketamine with fewer side effects in terms of motor coordination and anxiety. These findings identify novel and a highly specific target for drug treatment of major depression.

Biological sciences/Neuroscience/Diseases of the nervous system/Depression

Biological sciences/Neuroscience/Synaptic plasticity/Long-term potentiation

Biological sciences/Neuroscience/Ion channels in the nervous system

Major depressive disorder (MDD) is a widespread and devastating condition that leads to enormous individual suffering and imposes a serious socioeconomic burden¹. Despite its vast impact, current medical treatment options are limited. The delayed onset of the effects of classical antidepressants has been associated with increased suicidal ideation, which, in addition to the high percentage of patients who do not respond even after multiple treatment attempts, highlights the urgent need for novel and rapid-acting antidepressants. Subanesthetic doses of ketamine (KET) effectively reduce depressive symptoms within hours. An enantiomer of KET, S-ketamine, has been recently approved for treatment-resistant MDD. However, the rapid onset and comparatively large effect size of KET treatment are often accompanied by distressing adverse effects limiting its broad clinical use, including dissociative symptoms, anxiety, increased blood pressure and the abuse liability of the party drug KET ^2,3. Moreover, the exact mechanism of action of KET is poorly understood.

According to the “disinhibition hypothesis”, KET is thought to preferentially bind to NMDA receptors (NMDARs) on GABAergic interneurons, reducing the inhibitory input to pyramidal cells (PCs), which can increase glutamate release into the synaptic cleft or modify signal processing in the dendrites of postsynaptic neurons^4–7. Supporting the idea of increased glutamate release, increased extracellular levels of glutamate and enhanced glutamate cycling have been observed after KET administration⁸; however, studies have also reported no change or a decrease in glutamate release after KET administration⁹.

NMDARs are ligand-gated ion channels that are formed at most synapses by two ubiquitous GluN1 together with two GluN2(A-D) or GluN3(A-B) subunits as dimers of dimers or as coassemblies of different subunits¹⁰. The subunit composition determines the distinct biophysical, pharmacological and functional properties of the NMDAR subtypes in the CNS¹¹. The expression profiles of the four GluN2 subunits differ remarkably during development¹², in response to neuronal activity, between brain regions and cell types and in terms of their subcellular localization¹¹.

Here, we focused on the GluN2D subunit. In the embryonic brain, GluN2D subunits are highly expressed and decrease markedly during early postnatal development¹². GluN2D subunit-containing NMDARs display lower single-channel conductance, sensitivity to Mg²⁺ and Ca²⁺ permeability, open probability and slower deactivation than GluN2A/B-containing NMDARs^13–15. Notably, GluN2D subunits are predominantly expressed in hippocampal and cortical interneurons and are largely absent in principal neurons in adults^12,16,17. GluN2D is expressed in 60–80% of somatostatin-expressing (SOM-INs) and parvalbumin-expressing interneurons (PV-INs) in adult mice^18,19. Due to their unique expression pattern and electrophysiological properties, GluN2D-containing NMDARs on interneurons are expected to have a profound effect on hippocampal and cortical microcircuits, including the regulation of pyramidal neuron firing patterns, dendritic signal integration and synaptic plasticity^20–24. Importantly, in a recent unbiased transcriptomic study in humans, GRIN2D gene expression was upregulated in depressed patients who subsequently responded to KET compared with nonresponders, which could indicate an important role for GluN2D in the mechanism of action of KET²⁵.

Here, we clarify the molecular mechanisms underlying the preferential inhibition of NMDARs on interneurons by KET and the resulting modulation of excitation-inhibition (E/I) balance and synaptic long-term potentiation (LTP), which is a missing link between NMDAR inhibition, synaptic network restoration and rapid antidepressant effects. We identify the GluN2D NMDAR subunit as a promising target for more effective, tolerable and specific antidepressant strategies.

Differential inhibition of NMDA EPSCs in pyramidal cells and interneurons depends on the selectivity of antagonists for GluN2D

As a first step, NMDAR currents were evoked by extracellular stimulation in the hippocampal CA1 PCs and oriens-lacunosum/moleculare interneurons, which were identified by morphological and electrophysiological characteristics and fluorescence excitation, in SOM-Cre (SST tm2.1(cre)Zjh/J) td-Tomato mice (Fig. 1a). At a holding potential of + 40 mV in the presence of the AMPAR antagonist CNQX, the nonspecific NMDAR antagonist DAPV eliminated > 90% of the remaining stable current (Supplementary Fig. 1a-c). To pharmacologically dissect the impact of GluN2D subunits on isolated NMDAR currents, we used NAB-14 and (R,S)-ketamine.

NAB-14 (an N-aryl benzamide, NAB) is a GluN2C/GluN2D-selective negative allosteric modulator. As the GluN2C subunit is poorly expressed in adult mice, NAB-14 is selective for GluN2D in our experimental approach¹⁸. Site-directed mutagenesis experiments indicated that NAB-14 binds to a site between subunits in the transmembrane domain, specifically interacting with the GluN2D M1 helix²⁶. To gain insight into this interaction, we computationally docked NAB-14 against a published cryo-EM structure of GluN1a/GluN2D NMDARs (PDB ID: 7YFF)²⁷. The putative binding pose indicates that NAB-14 forms pi-stacking interactions with W586 and F590 via its indole and benzene moieties, respectively. It also forms an H-bond with W637 as well as hydrophobic interactions with M589, L594, I633, and M634 (Fig. 1b). In hippocampal brain slices, the addition of NAB-14 to the bath solution dose-dependently reduced the amplitude of NMDAR currents in interneurons but not in PCs (Fig. 1c,d and Supplementary Fig. 1d,e). Other studies have shown that NAB-14 has lower inhibitory potency on triheteromeric GluN1A/2B/2D NMDARs than on GluN1A/2D NMDARs²⁸.

KET binds to the phencyclidine-binding site within the channel pore of the NMDAR. This results in a use-dependent open-channel block that decreases the mean channel open time. Furthermore, an allosteric closed block might decrease the open frequency, causing trapping of KET in the channel pore²⁹. KET is rather unselective for GluN2 subunits; however, under physiological conditions with 1 mM Mg²⁺ added to the bath solution, KET shows 5- to 10-fold selectivity for GluN2C- and GluN2D-containing receptors over GluN2A and GluN2B^13,30. In our experiments, KET dose-dependently decreased the amplitude of NMDA EPSCs both in PCs and in INs (Fig. 1e and Supplementary Fig. 1f,g). However, the inhibition was more pronounced in INs, which suggests a preferential action of KET at GluN2D-containing NMDARs on INs.

Inhibition of GluN2D decreases the activity of feed-forward and feedback loops

PCs are embedded in a complex microcircuit containing different populations of interneurons, which form inhibitory feedback loops (FBLs) and feed-forward loops (FFLs, Fig. 2a). Both loops end in GABAergic synapses at different locations of the PC dendrites, which contribute to dendritic signal processing^20,21. GABAergic synapses of SOM-INs are located primarily at the distal dendrites of PCs and control synaptic NMDAR activation, burst firing and synaptic plasticity²². In contrast, GABAergic synapses of PV-INs are near the soma of PCs, generating shunting inhibition and controlling the timing of spiking output²³.

The FFL was selectively activated by subthreshold Schaffer collateral stimulation, i.e., in the absence of PC action potentials (APs). Under these conditions, the wash-in of 20 µM but not of 10 µM NAB-14 dose-dependently increased EPSP amplitudes in CA1-PCs (Fig. 2b). In contrast, low- and very high-dose KET did not significantly modify the EPSP amplitudes (Fig. 2c). During the very-low-frequency stimulation of Schaffer collaterals performed in these experiments, which evoked single APs in FFL-INs and allowed a reset of the NMDAR to a closed state between each stimulation event, the open channel NMDAR blocker KET²⁹ had no effect. In contrast, the negative allosteric modulator NAB-14 reduces NMDAR channel conductance independent of its opening state. The different mechanisms of NMDAR inhibition by the open channel blocker KET and the negative allosteric modulator NAB-14 might therefore explain the differential modulation of EPSP amplitudes by selective activation of the FFL in the absence of PC APs.

FBL-INs are highly dynamically recruited by excitatory PC activity and generate slow inhibitory postsynaptic currents (IPSCs) in PCs, effectively controlling PC NMDAR activation and synaptic plasticity^22,23,31. To activate FBL-INs, we repeatedly injected a depolarizing current into the soma of CA1-PCs, evoking a burst of APs. Subsequently, after each AP burst, an EPSP was induced by Schaffer collateral stimulation, which served as a readout for the effects of the modulation of the excitation/inhibition (E/I) balance on glutamatergic transmission in PCs (microcircuit-activating [MICA] protocol; Fig. 2d). After achieving a stable baseline for 10 min, NAB-14 was added to the bath solution, which increased the EPSP amplitudes and decay time constants compared to those of the control condition without NAB-14 (10 µM, Fig. 2d and Supplementary Fig. 2a,b). Similar effects were observed with KET (10 µM; Fig. 2e and Supplementary Fig. 2c). In contrast to the FFL-activating protocol (Fig. 2c), the MICA protocol induces a train of APs in PCs and INs so that the open channel blocker KET can effectively inhibit NMDAR currents in INs.

Coefficient of variation (CV) analysis of the EPSP slopes before and after the wash-in of NAB-14 was consistent with a predominant mechanism at the dendrites of postsynaptic neurons (Supplementary Fig. 2d). PC excitability, as assessed by the PC input resistance (R_m) and the number of APs during the burst, did not significantly change in the presence of either NAB-14 or KET (Supplementary Fig. 2e-h). We therefore concluded that GluN2D inhibition of interneurons shapes the amplitude and time course of synaptic transmission in PCs.

Next, we assessed how GluN2D inhibition in interneurons modulates excitatory activity in hippocampal PCs. Using a protocol similar to that used in the MICA experiments, Widman and McMahon demonstrated that KET increases the probability of synaptically driven AP⁵. In our experiments, NAB-14 and KET caused a pronounced conversion of EPSPs to APs at a fixed extracellular stimulation intensity (Fig. 2f-g and Supplementary Fig. 2i,j).

Taken together, these results suggest that GluN2D-mediated recruitment of GABAergic interneurons powerfully controls feedback and feedforward inhibitory circuits to moderate hippocampal network activity.

Inhibition of GluN2D and SOM-IN activity increases hippocampal LTP

Since PV-INs modulate PC input at proximal dendrites, reduced PV-IN activity is concordant with an increase in EPSP amplitude and a conversion of subthreshold EPSPs to APs. SOM-INs preferentially activate α5-subunit-containing GABA_A receptors at CA1 PC dendrites and play a key role in dendritic computation and the shaping of LTP^22,24. Moreover, the suppression of SOM-IN activity by KET has resulted in increased synaptically evoked calcium transients in the apical dendritic spines of pyramidal neurons in the mPFC^6,32. Therefore, it is unclear whether GluN2D inhibition, especially by the partially selective antagonist KET, increases LTP induction in the hippocampus or potentially decreases LTP via direct inhibition of NMDA receptors in CA1 PCs^33,34.

We evaluated the effects of NAB-14 and KET on associative LTP induction in ex vivo hippocampal brain slices from mice. EPSPs evoked by Schaffer collateral stimulation were paired with postsynaptic APs in CA1 neurons. To avoid a putative ceiling effect of an optimized LTP induction protocol (Supplementary Fig. 3a,c)³⁵, we used a weak LTP protocol with 25 EPSP→AP pairings; this resulted in weak but significant LTP (weak-aLTP, Supplementary Fig. 3b,c). Bath application of 10 µM NAB-14 significantly increased LTP (Fig. 3a). A maximal difference between the inhibition of NMDA EPSCs in INs and PCs was found at a concentration of 10 µM or higher (monoexponential fit, Fig. 3b, c.f. Figure 1d). Lower concentrations of NAB-14 (1 and 5 µM) did not significantly affect either NMDA EPSCs in INs or weak-aLTP (Fig. 3a and Supplementary Fig. 3f). Bath application of 10 µM KET blocked weak-aLTP induction, whereas lower concentrations (1 and 5 µM) increased LTP (Fig. 3c and Supplementary Fig. 3g). This inverted U-shaped concentration‒response relationship between KET and weak-aLTP reached its maximum at 4.9 µM KET (calculated maximum of Gaussian fit; Fig. 3d). At this concentration, the difference between the inhibition of NMDAR excitatory postsynaptic currents (EPSCs) in the INs and PCs (c.f. Figure 1f) also reached its maximum (monoexponential fit, Fig. 3d). In addition, we tested the effect of KET on an optimized spike time-dependent LTP protocol. A total of 125 EPSP→AP pairings (aLTP) resulted in strong LTP (Supplementary Fig. 3a,c). Under these conditions, bath application of 10 µM KET had no effect on aLTP induction compared to the control (Supplementary Fig. 3d). However, when inhibition was blocked by the GABA_A antagonist picrotoxin (50 µM), the same concentration of KET inhibited aLTP (Supplementary Fig. 3e). Taken together, these results indicate that LTP is increased at concentrations of NAB-14 or KET at which predominantly NMDAR EPSCs on INs are inhibited but is blocked by high concentrations of KET, which inhibits NMDARs on PCs.

To assess the putative role of increased glutamate release in the modulation of LTP by NAB-14 or KET, we determined the paired-pulse ratio (PPR) before and after the induction of weak-aLTP. The PPR was unchanged in both the absence and presence of 10 µM NAB-14 or 5 or 10 µM KET (Supplementary Fig. 3f,g), indicating a postsynaptic effect. This suggests that an altered E/I balance and its computation in the dendrites of postsynaptic cells are sufficient for the modulation of synaptic plasticity and that a decisive role for a ‘glutamate burst’, as proposed for KET⁷, is unlikely.

To directly assess the role of SOM-INs in LTP, we performed chemogenetic designer receptor exclusively activated by designer drugs (DREADD) experiments. By stereotactic injection, adeno-associated viruses (AAVs) were inserted into the hippocampus, leading to the expression of Cre-dependent excitatory (Gq) or inhibitory (Gi) DREADDs (pAAV5-hSyn-DIO-hM3D(Gq)/hM4D(Gi)-mCherry) in SOM-Cre mice (Fig. 3e,f). Activation of SOM-Gq-DREADDs with clozapine-N-oxide inhibited weak-aLTP induction, whereas significant LTP could be induced with SOM-Gi-DREADD activation. LTP differed significantly between the two conditions (Fig. 3g). These results demonstrate that selective manipulation of SOM-IN activity modulates LTP induction in CA1 PCs.

Inhibition of GluN2D restores LTP in an animal model of depression

Behavioral stress has been shown to increase GABAergic inhibition and compromise the excitation-inhibition balance^36–38, which might contribute to the well-known stress-related disruption of hippocampal synaptic plasticity^39–41. SOM-INs might be specifically important for this process because they are strongly activated and increase inhibitory output during stress^36,42. We investigated the modulation of LTP by repeated stress and GluN2D inhibition in an animal chronic despair model (CDM) of depression^43–45. In CDM, animals were subjected to ten-minute forced swim sessions on five consecutive days to induce a depression-like state, followed by a rest period of two days, and different readout assessments were performed afterward (Fig. 4a). In this model, the behavioral despair phenotype remains stable for at least 4 weeks in the absence of further interventions; chronic but not acute administration of tricyclic and selective serotonin reuptake inhibitor (SSRI) antidepressants has been found to reverse the depression-like phenotype⁴⁴.

LTP induction was completely abolished in ex vivo brain slices from mice injected with the control solution used to dissolve NAB-14 and sacrificed two days after the CDM protocol. CDM-induced LTP blockade could be reversed in a dose-dependent manner after intraperitoneal injection of 5 or 10 mg/kg NAB-14 (Fig. 4b and Supplementary Fig. 4). KET injection equally restored LTP induction (Fig. 4c).

As a nonpharmacological method for interfering with GluN2D, we used a small interfering RNA (siRNA) to achieve posttranslational gene silencing of GRIN2D, the gene encoding GluN2D. We administered mouse GRIN2D-siRNA (50 nM/animal) by intrathecal injection using in vivo jetPEI as a nonviral delivery vector. Injections were performed the day after the termination of the CDM protocol; thereafter, the animals were left in their home cages for three more days (Supplementary Fig. 5a). Real-time PCR was used to verify that GRIN2D RNA expression was downregulated in the hippocampus and frontal cortex, and Western blotting was used to confirm the downregulation of GluN2D protein expression (Supplementary Fig. 5b). The CDM-induced decrease in LTP was fully reversed after siRNA treatment (Fig. 4d). Overall, the results of the CDM-LTP experiments showed that KET and GluN2D inhibition reversed LTP deficits in an animal model of depression.

GluN2D inhibition shifts synapses toward excitation

Functional alterations in plasticity are consolidated by morphological changes involving protein synthesis, especially increased spine density^46,47. Therefore, we assessed the modulation of synaptic morphology after CDM exposure and after GluN2D inhibition (Fig. 5a). Markers for both excitatory glutamatergic presynaptic boutons (vGlu1) in the CA1 stratum pyramidale and inhibitory GABAergic presynaptic boutons (GAD 65) in the stratum moleculare, as assessed by immunostaining, were numerically reduced after the CDM protocol. Compared with vector treatment, intraperitoneal injection of 10 mg/kg NAB-14 24 hrs before tissue harvesting significantly increased vGlu1 and GAD 65 immunostaining after CDM. KET treatment resulted in comparable but nonsignificant changes. (Fig. 5b,c). The numeric effects of the CDM protocol on the number of excitatory synapses were more pronounced (vGlu1 puncta after CDM/vGlu1 puncta naïve state) than the effects on inhibitory synapses (GAD 65 puncta after CDM/GAD 65 puncta naïve state), suggesting a shift toward inhibition. NAB-14 and partially KET treatment restored the CDM-induced changes to a naïve state (Fig. 5d). Western blotting was used to assess the levels of the postsynaptic density proteins PSD 95 and AMPARs (GluA1). Two hours after i.p. application, both PSD95 and AMPAR concentrations were significantly greater in NAB-14-treated and KET-treated animals than in vector-treated animals subjected to CDM (Fig. 5e-f and Supplementary Fig. 6a,b). These results support a predominant downregulation of excitatory but also inhibitory synapses after CDM, leading to a shift in the E/I balance toward inhibition in a depressed phenotype. GluN2D inhibition restored synapses and AMPAR density toward the naïve state.

GluN2D inhibition has rapid antidepressant activity

Next, we assessed the behavioral effects of GluN2D inhibition. In the CDM animal model of depression, the immobility time, conceptualized as a measure of behavioral despair, gradually increased during the daily swim sessions, reaching a plateau at days 4–5, and remained stable in the delayed swim session on day 8 (Supplementary Fig. 7). In previous experiments, we observed a persistent increase in immobility time for up to four weeks^44,45. Thereafter, the immobility time returns to baseline levels without further intervention. After a single intraperitoneal injection of either 5 or 10 mg/kg NAB-14 or 10 mg/kg KET and after the intrathecal application of GRIN2D siRNA, the escape behavior returned to baseline levels and significantly differed from that of the control group in which the respective vector was applied.

As an alternative behavioral readout after CDM, we performed the nose-poke sucrose preference test⁴⁸ (NP-SPT) using a progressive ratio reinforcement schedule to assess wanting-type anhedonia and reward behavior. Compared with naïve animals, CDM mice showed a significantly reduced sucrose preference. Reward behavior was restored by a single injection of NAB-14 or KET and by the application of GRIN2D siRNA but not by the application of a vector or scrambled siRNA (Fig. 6b).

We conclude that escape and reward behavior is compromised in an animal model of depression. Inhibition of GluN2D after CDM normalized behavior to a level comparable to that of naïve animals; this indicates that GluN2D-mediated recruitment of inhibitory circuits powerfully controls depression-related behavior. GluN2D inhibition interferes with this process and has antidepressant activity in an animal model of depression.

No effect of NAB-14 on spatial memory, motor coordination, locomotion or anxiety

The activity of hippocampal SOM-INs is necessary not only depression-like behavior, but also for novelty-motivated spatial learning⁴⁹. To assess the putative disruption of episodic memory in nonstressed animals, we used an object location memory task (OLT, Fig. 7a). First, two identical objects were presented without pharmacological intervention. There was no difference in the exploration of either object. Immediately thereafter, saline, NAB-14 or KET was injected preceding a six-hour rest in the home cage, followed by a test session where one of the objects was presented at a different location. This newly located object was significantly more explored, with no difference between the treatment groups. We concluded that GluN2D inhibition does not impair novelty-motivated spatial memory. Therefore, GluN2D inhibition might preferentially interfere with the exaggerated recruitment of inhibitory circuits during repeated stress, while the unperturbed AMPA receptor-mediated recruitment of these interneurons is sufficient to support episodic memory processing in naïve animals.

Finally, we tested for acute behavioral alterations following GluN2D inhibition alone. We assessed motor coordination with the rotarod test. After a single KET injection, the latency to fall was significantly shorter than that after NAB-14 or vector injection (Fig. 7b). This might be due to increased oscillatory activity in the basal ganglia induced by KET⁵⁰. As GluN2D is not expressed in the striatum, these effects might be mediated by the action of KET on GluN2B receptors, which are abundantly expressed in striatal projection neurons⁵¹. Locomotion was unchanged after both NAB-14 and KET administration but decreased after administration of the benzodiazepine lorazepam, which was used as an active control (Fig. 7c). The time spent in the center in the open field test, which has been conceptualized as a measure of anxiety, was significantly decreased after KET but not after NAB-14 injection in comparison to that after vector injection (Fig. 7d). In terms of motor coordination and anxiety, the selective GluN2D antagonist NAB-14 produced fewer side effects than KET.

The GluN2D subunit of NMDARs is unique due to its selective expression on interneurons in adulthood, distinct electrophysiological properties and comparatively low overall expression levels. Selective GluN2D inhibition has profound behavioral consequences and might be a novel target for antidepressant action. Here, we describe a mechanistic sequence underlying the role of GluN2D-containing NMDARs in the stress response and the antidepressant action of GluN2D inhibition (Fig. 8).

Given the virtual absence of GluN2D expression in pyramidal neurons, the GluN2D inhibitor NAB-14 dose-dependently and selectively inhibits NMDAR currents in interneurons, whereas low-dose KET preferentially inhibits GluN2D subunit-containing NMDARs on GABAergic interneurons vs. NMDARs on PCs, depending on small dose-dependent differences in its ability to inhibit distinct NMDAR subunits. This might explain the small therapeutic window of KET for the treatment of MDD, as higher doses result in anesthesia but do not have antidepressant effects. Moreover, nonselective NMDAR inhibitors such as memantine do not exhibit antidepressant activity in rodent models or clinical trials⁵². We found that selective inhibition of GluN2D-containing NMDARs on interneurons by NAB-14 or in vivo application if GRIN2D siRNA is sufficient to mimic the cellular and behavioral effects of KET. The different time scales of functional GluN2D inhibition by KET, NAB-14 and siRNA result in comparable behavioral effects, so GluN2D inhibition might be regarded as a common initial signal that is translated to sustained plasticity changes further downstream.

At the microcircuitry level in the hippocampus, GluN2D antagonism leads to an inhibition of feedback and feed-forward loops that end in GABAergic synapses at different locations of the PC dendrites. The feed-forward loop is dominated by fast-spiking perisomatic PV-INs, which effectively control spike timing in PCs ^23,24,53,54. SOM-INs in the feedback loop predominantly target nonlinear α5-containing GABA_A receptors (α5GABA_ARs) that generate slow inhibitory postsynaptic currents in hippocampal CA1 PCs^22,55. Due to their strong outward rectification, α5GABA_ARs have a much larger conductance at depolarized membrane potentials, matching NMDAR properties, and therefore effectively control NMDA-dependent burst firing, dendritic integration and synaptic plasticity^{22–24,56,57}. Both silencing of SOM-INs and a negative allosteric modulator of α5GABA_ARs increased subthreshold EPSPs and spiking output^22–24^. Interestingly, inhibition of SOM-IN synapses via negative modulation of α5 subunit-containing GABA_A receptors counteracts stress-induced disruption of the hippocampal E/I balance³⁸, showing some antidepressant effects similar to those of KET. Burst firing is particularly effective at promoting synaptic plasticity⁵⁸. The activity of the feedback loop depends on AP firing in PCs as needed for LTP induction, which further emphasizes the predominant role of SOM-INs in the modulation of synaptic plasticity.

In our experiments, impaired activation of feed-forward and feedback loops by selective GluN2D inhibition shifted the E/I balance toward excitation, involving increased EPSP amplitudes and the conversion of EPSPs to APs. Moreover, selective GluN2D inhibition at doses equivalent to maximal NMDAR inhibition on interneurons is sufficient to increase hippocampal LTP. Conclusively, specific alterations in SOM-IN activity by inhibitory or excitatory DREADD receptors bidirectionally modulate LTP induction. It is highly probable that increased excitation and altered dendritic signal integration interact to augment LTP. Enhanced PC activity directly leads to an activity-dependent increase in protein synthesis, BDNF synthesis and release and mTOR signaling⁵⁹. In turn, BDNF is necessary for the early and late stages of LTP and its consolidation by protein synthesis⁶⁰.

LTP is commonly considered the initial step in learning and memory formation and is consolidated by the formation of dendritic spines^46,47. Behavioral stress as used in animal models of depression negatively modulates hippocampal synaptic plasticity, most likely through a corticosterone-dependent U-shaped relationship with mild stressors increasing and severe stressors impairing synaptic plasticity^61,62. We confirmed that severe and subchronic stress induced by the CDM protocol impaired LTP and synapse formation in the hippocampus.

Dysregulation of neuroplasticity is a key factor for both the pathogenesis and treatment of MDD^63,64. Many forms of neuroplasticity are exquisitely sensitive to stress⁶⁵. Psychosocial stressors, on the other hand, are highly relevant for the development of MDD in susceptible individuals⁶⁶. LTP and other forms of neuroplasticity were found to be blocked in animal models of depression⁴⁰ and human MDD patients⁶⁷ and can be effectively restored by antidepressants^41,67,68. The inhibition of LTP during sustained stress exposure might be adaptive to prevent an overload of negative valence learning and its neurobiological consequences. However, maladaptive downregulation of neuroplasticity could result in prolonged clinical depression, involving sustained shutdown of positive valence emotional learning.

In vivo GluN2D inhibition effectively reversed stress-induced impairments in LTP, synaptic morphology and density, as well as AMPAR expression, toward a naïve state. Restoration of impaired synaptic plasticity reestablishes neural adaptive capabilities and functionally reverses stress-induced network dysfunction. On a behavioral level, we found an impairment in escape and reward behavior after the application of the CDM protocol, and these effects were reversed by GluN2D inhibition. However, disruption of learning by GluN2D inhibition is limited to negative valence experience, which is associated with enhanced recruitment and enhanced inhibitory output of GluN2D-expressing interneurons^36,42. In contrast, we could not observe an effect during a neutral valence novelty-motivated learning task in nonstressed animals. This finding points to a central role for GluN2D and the recruitment of GluN2D-expressing interneurons in the stress response and the development of a depressed phenotype.

Immediate changes in network function caused by disinhibition of interneurons and an increase in excitatory activity, as shown by increased EPSP amplitudes and conversion of EPSPs to APs following GluN2D inhibition, might underlie the very rapid behavioral effects of KET, which occur within minutes to hours. Subsequently, restoration of impaired synaptic plasticity reestablishes neural adaptive capabilities and functionally repairs stress-induced network dysfunction. It is highly probable that the downstream effects of FFL and FBL inhibition interact and are mutually reinforcing, e.g., increased LTP inducibility and structural synaptic changes increase excitatory activity by upregulating AMPAR availability and vice versa. Functional alterations in plasticity are consolidated by morphological changes involving protein synthesis, especially increased spine density^59,69. This might explain the sustained clinical effects of KET for several days after a single application or much longer after repeated applications^3,7,69,70. Chronic application of conventional antidepressants for 2-3 weeks is needed to increase synaptic potentiation by more indirect mechanisms⁶⁹ or to preferentially inhibit long-term synaptic depression^43,67.

The behavioral consequences of pharmacological or siRNA-induced GluN2D antagonism in adult animals are in many respects opposite to those of constitutive GluN2D knockout⁷¹. GluN2D-KO mice displayed, e.g., increased immobility in the tail suspension task and reduced sucrose preference, in addition to reduced locomotion^71,72. GluN2D expression decreases markedly during development and becomes localized to specific brain regions and cell types^12,17. Tonic activation of GluN2D-containing NMDARs within the first weeks of postnatal development is thought to be critical for interneuron density and maturation and for GABAergic synapse density. Constitutive GluN2D mice are therefore expected to exhibit highly altered microcircuits in adulthood, which might explain these striking differences⁷³.

Additional targets might amplify the antidepressant mechanism of KET. Whereas the GluN2C subunit is virtually absent in adulthood, GluN2B is the second NMDAR subunit that is preferentially expressed in interneurons¹². S-ketamine has been shown to bind to GluN2B-containing NMDARs⁷⁴, and GluN2B knockdown in somatostatin-expressing interneurons (SOM-INs) has been shown to reduce interneuron activity and prevent some behavioral effects of KET^6,32. However, the GluN2B-selective antagonist Ro 25-6981 failed to elicit disinhibition of CA1 pyramidal neurons and was clearly differentiated from the disinhibiting effects of KET⁵. In a number of preclinical and clinical studies, negative allosteric modulators of GluN2B showed rapid antidepressant activity comparable to those of KET, but also significant disruption of cognitive function, amnestic and dissociative effects¹⁷.

We recently described a cholesterol-dependent binding site for KET and other antidepressants in the BDNF–TRKB receptor, which facilitates synaptic localization of TRKB and its activation by BDNF⁷⁵. Use-dependent trapping of KET by NMDARs in the lateral habenula has been proposed to explain the sustained antidepressant action of KET despite its short elimination half-life⁷⁶. Moreover, (2S,6S;2R,6R)-hydroxynorketamine (HNK), a major metabolite of KET, might exert an NMDAR-independent antidepressant effect through sustained activation of synaptic AMPARs⁷⁷. However, the exact binding site of HNK is unknown, and its antidepressant efficacy remains controversial⁷⁸.

In this set of experiments, GluN2D-containing NMDARs were identified as promising targets for rapid-acting antidepressant action via rescue of the E/I balance and disturbed synaptic plasticity. As these subunits are almost exclusively expressed on interneurons in adult life, these targets are highly specific, and targeting them might avoid the unwanted effects of less selective NMDAR antagonists, such as psychomimetic activity, anxiety, impaired motor coordination and anesthesia.

Animals and husbandry

Wild-type C57Bl6N mice were obtained from Janvier (Le Genest-Saint-Isle, France) and kept at the animal husbandry facilities of the Universities of Freiburg or Strasbourg. SOM-Cre (SST tm2.1(cre)Zjh/J) mice were provided by the Department of Biomedicine Basel and SOM-IRS-Cre/J from the Department of Physiology of the University of Freiburg and kept at the animal husbandry facility of the University of Freiburg. Adult animals (10-14 weeks) were used for all experiments. All procedures were previously approved by animal review boards in Germany (Regierungspräsidium Freiburg) or France (CREMEAS); the animal care use protocols followed national and international standards and were systematically documented. Prior power calculations for behavioral assessments were performed to reduce the number of experimental animals whenever possible.

Electrophysiology

Mice were preoxygenated in a 100% oxygen atmosphere for 5 min before cervical dislocation and decapitated according to national and institutional guidelines. Three hundred micrometer-thick transverse slices from the hippocampus were cut with a vibratome (VT1200, Leica Biosystems, Germany). Slices were prepared in artificial cerebrospinal fluid (aCSF) containing (in mM) 125 NaCl, 25 NaHCO3, 1.25 NaH2PO4, 2.5 KCl, 1 MgCl2, 27 glucose, and 2 CaCl2 (bubbled with carbogen (95% O2, 5% CO2)).

After 20 min of rest at 35 °C, the slices were kept at room temperature in aCSF, transferred to the recording chamber (volume of approximately 2-3 ml) and continuously superfused with aCSF (rate of approximately 5-10 ml/min-1). Differential interference contrast video microscopy (Zeiss Axioskop 2 FS plus, Zeiss Microscopy, Germany) was used to assess the location and morphology of CA1 pyramidal neurons. In addition to optical identification, neurons were classified according to their characteristic firing frequency adaptation to long depolarizing current pulses. Borosilicate glass tubes (2.0 mm outer diameter, 0.5 mm wall thickness; Hilgenberg, Germany) were used to pull the patch pipettes. The pipettes had an open resistance of 5-10 MΩ, and the series resistance (Rs) of 10-50 MΩ was compensated by a bridge balance. The patch pipettes were filled with two different internal solutions for current clamp (EPSP) and voltage clamp (current) measurements. The internal solution for the current clamp contained (in mM) 132 K-gluconate, 20 KCl, 2 MgCl2, 10 HEPES, 0.1 EGTA, 4 Na2ATP, and 0.3 NaGTP (pH adjusted to 7.2 with KOH). The internal solution for voltage clamp measurements contained 135 Cs-gluconate, 2 CsCl, 5 QX314, 10 HEPES, 10 EGTA, 2 MgCl2, 2 Na2ATP, and 2 TEA-Cl (pH adjusted to 7.2 with HCl). Osmolarity (280-300 mosmol/l) was controlled at the beginning of each experimental day with an osmometer (Osmomat, Gonotec, Germany).

A stimulation pipette (a patch pipette with an open resistance of 1-3 MΩ when filled with internal solution) was placed superficially in the stratum radiatum of the CA1 region approximately 30-50 µm from the PC layer. Subthreshold EPSPs (for LTP protocols and wash-in experiments: 2-7 mV; for the microcircuit-activating (MICA) protocol: 5-10 mV) were evoked by Schaffer collateral stimulation with voltage pulses of 10-80 V (frequency of 0.1 Hz, duration of 200 µs) using a stimulus isolator (Model 2100 isolated pulse stimulator, A-M Systems, USA). The resting membrane potentials were between -75 and -65 mV, and the holding potential was -70 mV for PCs and between -60 mV and -65 mV for interneurons, with a holding potential of -60 mV. Hyperpolarizing voltage test pulses (50 ms/-5 mV) were applied to assess the input (Rm) and series (Rs) resistance after every 10th EPSP. For aLTP and weak aLTP, EPSPs were combined with postsynaptic action potentials (APs) triggered by short (3 ms) 700 pA current injections via a patch-clamp electrode. EPC-10 amplifiers (HEKA, Germany) were used, and the signals were filtered at 5 kHz. Patchmaster NEXT software (version 1.2, HEKA, Germany) was used for data acquisition and initial analysis. All experiments were performed at room temperature.

Experiments were discarded if (i) Rs changed by more than 30% during the experiment, (ii) evidence of ictal discharges was observed, (iii) the membrane potential between the start and end of the experiment differed by more than 5 mV, or (iv) the neurons did not respond to a firing pattern control pulse at the beginning and end of the experiments. Two experienced raters who were blinded to the experimental group agreed to the exclusion of the experiments. The experiments were performed in parallel; the experimenters were blinded to the treatment groups. A maximum of two hippocampal slices per animal were used. All groups included at least 6 animals with equal sex ratios.

Identification of SOM-Ins: Slices from mice expressing tdTomato in SOM-INs (SOM-Cre (SST tm2.1(cre)Zjh/J)) or from wild-type mice were prepared as described above. Interneurons were visually identified by fluorescence imaging at an excitation wavelength of 553 nm and an emission wavelength of 580 nm by using a polychromator (Polychrome IV, Till Photonics, Germany), a fluorescence camera (Orca Flash 4.0, Hamamatsu, Japan) and the SmartLUX extension for Patchmaster NEXT software (HEKA, Germany). In slices from wild-type animals, SOM-INs were identified by their morphological characteristics, their location in the stratum oriens, and an initial membrane potential in the range of -60 mV.

aLTP: Five EPSPs and five postsynaptic APs were paired at 100 Hz with a 5 ms delay (AP after EPSP). Five of these bursts of synchronized EPSP→AP pairs were applied at theta frequency (5 Hz), followed by an interval of 10 s and 4 more theta blocks, resulting in 125 EPSP/AP pairings.

Weak-aLTP: Unlike aLTP, only one theta block was applied, resulting in 25 EPSP/AP pairings.

Wash-in experiments: EPSPs were elicited by Schaffer collateral stimulation at a frequency of 0.1 Hz. The mean baseline EPSP amplitude was calculated at 0-5 min. Substances were continuously applied to the bath solution after baseline recording for an additional 30 min. Mean EPSP amplitudes were calculated between 25 and 30 min after wash-in.

Voltage-clamp recordings of NMDAR currents: At a holding potential of -70 mV (PCs) or -60 mV (SOM-INs), the cells were depolarized to +40 mV for 1 s. Five hundred milliseconds after the beginning of depolarization, extracellular stimulation was applied by Schaffer collateral stimulation (PCs) or by extracellular stimulation (≤ 10 V) in the stratum oriens at a distance of approximately 200 µm from the clamped interneuron (SOM-IN). Ten peak current amplitudes with an interval of 30 s were averaged before and after the bath application of substances.

MICA: A 300 ms depolarizing current between +100 and +600 pA was injected via the patch pipette into a PC to elicit a train of 5-10 APs. After an interval of 600 ms, one EPSP (5-10 mV amplitude) was elicited by Schaffer collateral stimulation. This pattern was continuously repeated at a frequency of 0.2 Hz. Substances were added by bath application after a stable 10 min baseline recording for an additional 20 min.

siRNA

Two hours after the induction phase of the CDM protocol, the animals were anesthetized with isoflurane and placed on a heating pad. The lower back was shaved and disinfected, and the siRNA compound was slowly injected into the space between the L5 and L6 vertebrae. In some preparatory experiments, ink was injected to confirm the accuracy and reliability of the intrathecal injection technique (Supplementary Fig. 5a). The siRNA mixture contained (per animal) 0.06 μl in vivo-jetPEI solution (Polyplus-transfection, France), 50 nM siRNA, 5 μl glucose solution (10%) and 4.94 μl H2O, as described in a recently established protocol⁷⁹. Afterward, the animals were allowed to rest for three days in their cages before the outcomes were assessed. The details of the siRNAs used were as follows: Ambion In Vivo Ready, HPLC-IVR (Thermo Fisher Scientific, s201423, catalog # 4457310); chromosome location: Chr. 7: 45831883-45872689 on GRCm38; RefSeq: NM_008172.2; translated protein: NP_032198.2; target exons: 2 and 3; and siRNA location: 1197. For the control experiments shown in Fig. 4d and Fig. 6 (scrambled siRNA), we used Ambion In Vivo Negative Control #1 siRNA (Thermo Fisher Scientific, catalog # 4457289). In all other siRNA experiments, the control group received the treatment described above, but the siRNA was omitted.

DREADD and stereotactic surgery

Prior to surgery, 4 mg/kg carprofen and 0.1 mg/kg buprenorphine were injected subcutaneously. Mice were anesthetized with isoflurane (4% for sedation induction, 1.5% for maintenance) and placed on a stereotactic frame (51730UD, Stoelting, USA). Body temperature was maintained using a heating pad. A total of 300 nL of a viral suspension of either pAAV2/5-hSyn-DIO-hM4D(Gi)-mCherry (44362-AAV5, Addgene, USA) or pAAV2/5-hSyn-DIO-hM3D(Gq)-mCherry (44361-AAV5, Addgene, USA) was injected bilaterally into the hippocampus (-1.7 AP, +/-0.65 ML, -1.6 DV) of SOM-IRS-Cre/J mice using a 30G syringe (Hamilton Neuros syringe) and a microinjection syringe pump (UMP3T, WPI, USA) at a rate of 200 nL/min. After injection, the syringe was left in place for five minutes before removal. After surgery, the mice received 0.1 mg/kg buprenorphine subcutaneously and 0.06 mg/100 ml buprenorphine in the drinking water overnight. The mice were then allowed to rest for 3 weeks before further experimental procedures.

Immunohistochemistry

Brains were extracted after the intracardiac perfusion of mice with 20 mL of ice-cold phosphate-buffered saline (PBS; 8.1 mM Na2HPO4, 138 mM NaCl, 2.7 mM KCl and 1.47 mM KH2PO4 [pH 7.4]), followed by 15 mL of 4% paraformaldehyde (PFA) in PBS solution. The brains were postfixed in 4% PFA/PBS for 6 hours at 4 °C and then cryoprotected for 48 hours in a 30% sucrose/PBS solution at 4 °C. After sectioning with a cryostat, 30 μm-thick sections of the hippocampus were washed and blocked with 4% normal goat serum in PBS for 1 hour at room temperature. The primary antibodies were diluted in PBS, 0.3% Triton X-100, and 1% goat serum and incubated overnight at 4 °C under agitation. The sections were then washed three times for five minutes in PBS supplemented with 1% Triton X-100 and incubated for one hour at room temperature with secondary antibodies diluted in PBS supplemented with 1% Triton X-100 and 1% goat serum. The sections were then incubated for 15 minutes at room temperature with nuclear 1 mg/mL 4,6-diamidino-2-phenylindole (DAPI, Thermo Fisher Scientific, USA, 1:1000) in PBS containing 0.2% Triton. After washing, the sections were mounted with Mowiol-DABCO (803456, Merck, Germany).

The following primary antibodies were used: guinea pig polyclonal anti-VGLUT1 (AB5905, 1:5000; Millipore, USA, lot #3878831) and mouse anti-GAD65 (ab26113, 1:500, Abcam, United Kingdom, lot #GR3308495-2). The following secondary antibodies were used: goat polyclonal anti-guinea pig Alexa Fluor 594 (Invitrogen, USA, A-11076, 2 μg/mL, lot #2540867) and goat polyclonal anti-mouse Alexa Fluor 488 (Invitrogen, USA, A-32723, 2 μg/mL; lot #XA336883). All images were taken using a Zeiss Celldiscoverer 7 microscope equipped with a confocal LSM 900 with AiryScan 2. The objective used was a Plan-Apochromat 50 x/1.2 W, N. A 1.2, H²O immersion. The laser intensity and gain were set to occupy the full dynamic range of the detector.

GluN2D immunoblotting: Hippocampal tissue was dissociated in ice-cold RIPA buffer (30 mM Tris base (pH 7.4), 150 mM NaCl, and 1% Triton X-100) containing protease and phosphatase inhibitors. Debris was removed by centrifugation (13,000 × g at 4 °C, 15 min). Protein quantification was performed by the BCA method. Proteins (50 μg) were resolved on a 10% polyacrylamide gel under denaturing conditions and transferred onto a polyvinylidene difluoride membrane. The membranes were blocked with Tris-buffered saline (10 mM Tris and 200 mM NaCl, pH 7.4) containing 5% nonfat dry milk. The blots were incubated overnight with GluN2D primary antibody from Millipore Sigma (Germany, MAB5578, lot #3352415; 1:500). After incubation with the appropriate IRDye 800 CW-conjugated anti-rabbit secondary antibody, the proteins were visualized with an Odyssey Imaging System (LI-COR Biotechnology, USA). Tubulin (Abcam, United Kingdom, ab11321; 1:15 000) was used as a loading control, and ImageJ (https://imagej.nih.gov/ij/, NIH, USA) was used to calculate the relative density of the bands.

PSD95 and GluA1 immunoblotting: Dissected brain regions or whole acute slices were mechanically homogenized in either homogenization buffer (320 mM sucrose, 4 mM HEPES [pH 7.4], 2 mM EDTA) or immunoprecipitation (IP) buffer (50 mM Tris HCl [pH 7.4], 120 mM NaCl, 5 mM EDTA, 0.5% Triton X-100) and centrifuged at 800 × g for 10 min at 4 °C to generate total (S1) and nuclear fractions (P1). All buffers contained a phosphatase and protease inhibitor cocktail (Sigma‒Aldrich, USA). Synaptic (P2) and cytosolic (S2) fractions were obtained by centrifugation of S1 at 10 000 × g for 20 min at 4 °C. Washed synaptosomal pellets (P2) were either processed for PSD isolation or directly lysed in lysis buffer (50 mM Tris-HCl [pH 6.8], 1.3% SDS, 6.5%, glycerol, 100 μM sodium orthovanadate) containing phosphatase and protease inhibitor cocktail (Sigma‒Aldrich, USA), boiled for 10 min at 95 °C and processed for SDS‒PAGE and Western blotting. For PSD isolation, P2 was resuspended in 4 mM HEPES [pH 7.4] and 2 mM EDTA, incubated for 30 min at 4 °C and centrifuged at 25 000 × g for 20 min at 4 °C. The obtained pellets were resuspended in homogenization buffer, placed on top of a sucrose gradient (0.8 M, 1.0 M and 1.2 M sucrose with 4 mM HEPES [pH 7.4] and 2 mM EDTA) and centrifuged at 150 000 × g for 2 h at 4 °C. The fraction collected at the interface between 1.0 M and 1.2 M was pelleted by centrifugation at 150 000 × g for 30 min at 4 °C, lysed in lysis buffer and processed for SDS‒PAGE and Western blotting. The protein concentrations of the different fractions were determined using a BCA assay kit (Thermo Fisher Scientific, USA) according to the manufacturer’s instructions. DTT (10 mM) and bromophenol blue were added to total (S1, mixed 1:1 with lysis buffer) or synaptosomal (P2) lysates, boiled for 10 min at 95 °C, separated by SDS‒PAGE on 7.5%-12% acrylamide gels and transferred to polyvinylidene difluoride (PVDF) membranes (Merck Millipore, USA). The membranes were blocked with 5% nonfat dry milk in TBS-T (1% Tween 20 in Tris-buffered saline (TBS)) or Roti-Block (Carl Roth, Germany) and then incubated with the following primary antibodies diluted in TBS: mouse anti-GluA1-NT (Millipore, USA, MAB2263, 1:2000, #3847897), rabbit anti-PSD95 (Cell Signaling, USA, 2507, 1:2000, lot #4), and mouse anti-GAPDH (Abcam, United Kingdom, ab8245, 1:1000, lot #1035914-6). After three washes, the membranes were incubated with the following horseradish peroxidase-conjugated secondary antibodies diluted in TBS-T: sheep anti-mouse (GE Healthcare, USA, NA931, 1:20000, lot #17556470) and donkey anti-rabbit (GE Healthcare, USA, NA9340, 1:25000) for 1 h at room temperature. In some experiments, the membrane was washed and stripped for further reincubation with another antibody. In that case, the membrane was incubated two times for 15 min in a glycin stripping solution (glycin 15g, SDS 1g, Tween20 10mL, 800mL dionized water), washed three times ten minutes in TBS-T, and then blocked and incubated following the regular procedure. The washed membranes were developed with a ChemiDoc MP imaging system (Bio-Rad) using an enhanced chemiluminescence detection kit (Thermo Fisher Scientific, USA). Band intensity was quantified by densitometry with ImageJ 1.47v software (National Institute of Health, USA) and normalized to the appropriate loading control.

Real-time PCR

RNA was isolated from powdered frozen hippocampal samples using a NucleoSpin RNA kit (Macherey Nagel, Germany), and cDNA was prepared using Oligo d(T) primers and Ready-To-Go You-Prime First-Strand Beads (GE Healthcare, Germany). Real-time PCR was performed using a Takyon No Rox SYBR MasterMix dTTP Blue Kit (Eurogentec, Belgium) and a LightCycler 480 (Roche, Switzerland). The reference genes glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and 40S ribosomal protein S18 (RPS18) were used as internal controls. The following primer pairs (sequences provided as 5′-3′) were used:

GluN2D (fwd: CTGTGTGGGTGATGATGTTCGT, rev: GTGAAGGTAGAGCCTCCGGG);

GAPDH (fwd: ACAACTTTGGTATCGTGGAAGG, rev: GCCATCACGCCACAGTTTC); RPS18 (fwd: GCGGCGGAAAATAGCCTTTG, rev: GATCACACGTTCCACCTCATC).

Amplification was performed with an initial denaturation of 45 cycles of 95 °C for 10 s, followed by 45 cycles of 60 °C for 15 s and 72 °C for 15 s. A melting curve was obtained at the end of cycling to verify the amplification of a single PCR product. The expression of the GRIN2D/GluN2D gene relative to a normalization factor (the geometric mean of the levels of two reference genes) was calculated using the 2−ΔCt method as previously described⁸⁰.

Behavioral procedures

Prior to all behavioral experiments, the mice were handled daily for 1 min over a time course of 5 days. Tunnel handling was used for all animal placement.

Chronic Despair Model (CDM): Mice were forced to swim in a glass cylinder (Ø 26 cm, 60 cm high) filled to a depth of 25 cm with 25 °C water for 10 min on 5 consecutive days. The immobility time was measured in each session. After swimming, the animals were gently dried and kept in a cage under a heating light for an additional ten minutes. Afterward, the mice were allowed to rest for two days in their home cages. Mice were treated after the resting period depending on the experimental group. For the measurement of escape behavior, an additional swim session was performed. For intraperitoneal injections, (±)-ketamine was diluted in NaCl (0.9%) and NAB-14 was dissolved in PEG 400 (40%), dimethylacetazolamide (10%) and 50% glucose (5%). Lorazepam was purchased as an injection solution (Tavor Pro injection, 2 mg, Pfizer, Germany). A maximum volume of 0.2 ml was injected per mouse; the volume was individually adjusted to the weight of the animal to achieve the intended dose. As a control, the respective vehicle was administered.

Behavioral readout: After the intervention, the animals were either subjected to an additional swim session (test day) to measure immobility time or subjected to the nosepoke sucrose preference test (NP-SPT) using the IntelliCage system (TSE Systems, Germany). For assessment of immobility time, the mice were videotaped, and two independent experienced raters who were blinded to the experimental groupings analyzed the videos. The mean values determined by both raters were used for further analysis. Immobility time was defined as the cumulative time that the animals remained stationary, during which only movements of the tail or forepaws necessary to keep the head above the water surface were made, the animals did not travel any distance and merely passively floated, no directed movement of the front paws was observed, and the body was mostly oriented parallel to the walls of the cylinder. A detailed description of this protocol and its analysis can be found here⁴⁵.

Nose-poke sucrose preference test (SPT): The IntelliCage system (TSE Systems, Germany), which allows automatic analysis of the spontaneous and exploratory behavior and drinking preference of rodents, was used. Radiofrequency identification transponders were implanted in the mice. IntelliCage comprises a common space in the middle and four measuring corners. Up to 16 group-housed mice can be given free access to food in the center, while water is supplied in the corners (by two zipper bottles each) behind electronically managed doors. The corners can be visited by only one mouse at a time. Using IntelliCage Plus software, the system records i) the number and duration of corner visits, ii) the number of nosepokes (approaches) on the doors, and iii) the number of licks on the drinking bottles. Initially, the mice were adapted to the IntelliCage for five days and provided free access to food and water in all corners. Then, for three days, the animals were habituated to sucrose (each corner contained one bottle with 1% sucrose solution and one with water; the contents of all bottles were exchanged every day) and subjected to the nosepoke protocol. The mice had to perform a nosepoke to open the doors and access both water and sucrose. Next, an NP-SPT paradigm⁴⁸ in which gradually increasing effort (number of nosepokes) was needed to access the sucrose-containing bottles for a short period (24 h) was used for the measurement of sucrose preference. In this protocol, each door opened in response to a nosepoke and closed after the animals drank from the sucrose-containing bottle for 5 s, while the mice were allowed to consume water from the water-filled bottles for an unlimited amount of time. The number of nosepokes needed to access the sucrose-containing bottle progressively increased (1, 2, 3, 4, 5, 6, 7) after each of the 10 licking sessions. This test was used to assess the drive for obtaining a reward; a decreased sucrose preference indicates reduced reward behavior. For every side, the number of licks was measured, and the mean sucrose preference was calculated as the percentage of the licks of the sucrose solution-containing bottles compared to the total number of licks. “post 1” and “post 2” represent two consecutive 24 h test periods.

Open field test (OFT): The test was performed in a square arena (50 × 50 cm) surrounded by a 35 cm high wall made of gray PVC. The mice were placed in the center of the field and allowed to move freely. The mice were video tracked for 10 min, and the time spent in the center (16% of the entire area of the box) was compared to the time spent outside the center (time in the center/time outside the center). Tracking and analysis were performed with EthoVision XT software (Noldus, Netherlands).

Object location memory task (OLT): For habituation, the mice were placed in the center of an empty square arena (50 × 50 cm) surrounded by a 35 cm high wall of the gray PVC and maintained there for 10 min. On the following day, the mice were placed in the empty arena for 10 min for a second habituation phase. Immediately thereafter, the mice were removed, and two slightly different objects were introduced into the arena (for location, see Fig. 7a). The animals were then placed back into the area and allowed to rest there for 10 min. Directly after the exploration of the two objects, the mice were injected (KET/NAB-14/saline) and placed back into their home cages. Six hours later, the mice were placed in the arena where the location of one of the two objects was changed (see Fig. 7a for the location of the object) and remained there for another 10 min. During all the time in the arena, the mice were video tracked using EthoVision XT software (Noldus, Netherlands).

Rotarod: A ROTA ROD LE8205 (PanLab, Harvard Apparatus, Spain) was used for the experiments. This instrument automatically detects the latency to fall and enables simultaneous assessment of up to five animals. Before testing, the animals were exposed to three training sessions at 5 rpm for 1-3 min with a 10 min interval between each session. Animals that did not stay on the rod rotating at 5 rpm for at least 60 sec in the third training session were excluded from further testing. For testing, the animals were set on a rod at 4 rpm. The rod was set to accelerate from 4 to 40 rpm in 300 sec, and the latency to fall was measured. This procedure was repeated at 5, 20, 30 and 50 min postinjection, and mean values were calculated for each treatment condition.

Locomotion: The test was performed in a square arena (50 × 50 cm) surrounded by a 35 cm high wall made of gray PVC. The mice were placed in the center of the field and allowed to move freely. Behavior was recorded for 10 min, and the total distance traveled was analyzed with EthoVision XT software (Noldus, Netherlands).

Quantification and statistical analysis

All given values are the mean ± SEM, and the error bars represent the SEMs in the figures; n represents the number of experiments. For the statistical analyses, GraphPad Prism version 8.3.0 (GraphPad Software, USA) was used. To test for significance within experimental groups (e.g., baseline vs. effect), we used a paired t test or a Wilcoxon signed rank test. To compare two experimental groups, either an unpaired t test or the Mann–Whitney test was performed. To compare several time points in one experimental series, repeated-measures ANOVA followed by Dunnett’s correction was used. To test for differences among several treatment groups, ordinary one-way ANOVA followed by Dunnett correction was used. To test for differences between groups with two variables, repeated-measures ANOVA within factors was conducted, and for estimation of effect sizes, ETA squared values were calculated. In cases of sphericity violation, Greenhouse‒Geisser adjustment was applied. Outliers were excluded using ROUT methods set at 1%. All the statistical tests were two-tailed, if applicable. Significance levels are depicted by asterisks in the figures: *p<0.05, **p<0.01, ***p<0.001, ****p<0.001. The exact p and F values as well as the applied tests for each group are listed in Supplementary Table 1.

Sex aspects

All experiments were performed with an equal distribution of male and female mice. For all experiments, a separate exploratory analysis focusing on the potential effects of sex on the experiments was performed, but no relevant differences were observed.

Molecular modeling

The Schrödinger Software Suite 2021-4 (Schrödinger LLC, NY, USA) was used for modeling studies. The SMILES of NAB-14 was converted to a valid 3D Lewis structure using LigPrep (Schrödinger LLC, NY, USA) and was geometrically optimized with the OPLS4 force field⁸¹. The cryo-EM image of the heterotetrameric GluN1a/GluN2D NMDAR (PDB ID: 7YFF) structure was prepared for docking with the Protein PrepWizard⁸². This step involves adding missing hydrogen atoms and adjusting the ionization state of polar amino acids at neutral pH and, finally, energetically minimizing the structure using the OPLS4 force field. The geometric center of residue C590 (located on the GluN2D-TMD M1 helix) was considered the grid centroid, and flexible docking was carried out using Glide with the SP scoring function⁸³.

Analysis of patch-clamp whole-cell data

Rs and Rm: Whole-cell currents in response to 50 ms/-5 mV hyperpolarizing pulses were analyzed. Rs was calculated by dividing -5 mV by the maximal current amplitude (Is). Rs represents the electrical resistance between the pipette and cell soma in whole-cell measurements. Rm was calculated by dividing -5 mV by the stable current amplitude (Im) after the initial current peak and represents the input resistance of the patched cell.

Analysis of LTP: In the LTP figures, each dot represents six consecutive averaged maximum EPSP amplitudes ± SEM. The horizontal bars at 25-30 min indicate the significant differences between baseline and postinduction EPSP amplitudes within each experimental group. The mean amplitude of 30 consecutive EPSPs was calculated before the induction protocol to obtain the baseline amplitude. Twenty-five minutes after LTP induction, the mean amplitude of 30 consecutive EPSPs was calculated and compared to the baseline value from the same experiment with a two-tailed t test. Changes in EPSP amplitudes are expressed as percentages of the baseline measurements and were analyzed for each experimental group to quantify the degree of LTP.

Paired-pulse ratio (PPR): In most experiments, every 10th EPSP was replaced by two consecutive EPSPs (EPSP 1/2) with an interval of 50 ms, and their maximal amplitudes were measured. The PPR was then calculated as (EPSP2 (mV))/(EPSP 1 (mV)), and the mean PPR values were compared between baseline values before induction of the weak aLTP and those 25-30 min after induction. A change in the PPR commonly indicates a change in the probability of presynaptic transmitter release, whereas no change indicates a postsynaptic mechanism^84–86.

MICA-EPSP amplitude: The maximum EPSP amplitudes from min 0-10 were averaged and compared to the average maximum EPSP amplitudes from min 20-30.

MICA-AP-conversion: A 10-minute baseline was recorded during continuous MICA protocol stimulation at 0.2 Hz with EPSPs adjusted close to AP-threshold levels (ca. 15-25 mV). Thereafter, test substances were added to the bath solution, and recording continued for 20 minutes. AP spikes were counted and calculated as the mean number of APs over periods of 10 minutes.

MICA-Decay Tau: The trace of each EPSP was fitted using the following formula:

y(x) = RMP + Amp_1 * {1 - exp(-(x-t0_1)/Tau_on)} * exp(-(x-t0_1)/Tau_decay)

where RMP is the resting membrane potential (mV) and AMP_1 (mV) is the maximum EPSP amplitude. Tau_on (ms) is the onset tau of the EPSP, and Tau decay (ms) is the decay. For each cell, the mean EPSP-decay tau time constants were calculated and averaged for each experimental group. Fitmaster NEXT (HEKA, Germany) was used for fitting.

Fitting (Fig. 3): For the NMDAR current experiments and weak-aLTP experiments with NAB-14, a nonlinear regression was fitted to the measured x/y values, and the IC50 values were calculated (x = NMDAR current/amount of LTP, y = KET/NAB-14 concentration). The model used was as follows:

IC50 = ICF/(F/(100-F))^(1/Hill slope)

Y = Bottom + (Top-Bottom)/(1+(IC50/X)^HillSlope)

(F = 50)

For weak LTP experiments with KET, a Gaussian distribution model was fitted to the measured x/y values (x = amount of LTP, Y = KET concentration). The model used was as follows:

Y = Amplitude*exp(-0.5*((X-Mean)/SD)^2)

CV analysis: The slope of the EPSP rise at 20% to 40% of its maximal amplitude, at which point an approximately linear increase in its voltage could be assumed, was fitted with Fitmaster software (HEKA, Germany). We analyzed 20 EPSPs before and 20 EPSPs 25-30 min after NAB-14 wash-in in the MICA protocol for each experiment. The coefficient of variation (CV) is the standard deviation of the EPSP slopes divided by the mean. The inverse square of the CV of the postwash-in slopes was divided by the inverse square of the prewash-in slopes and plotted against the corresponding normalized slopes. Pre- and postwash-in measurements from a single experiment are connected by a line. In a standard quantal model for synaptic transmission, synaptic responses are affected either by presynaptic changes in the number of release sites and/or the probability of release or by modifications of the postsynaptic response to a single vehicle. A change in the ratio of CV^-2 reflects a presynaptic action, whereas horizontal lines in the CV^-2 plot indicate a change in postsynaptic responsiveness^85,87,88.

Morphological analysis

All quantifications were performed by a blinded researcher. For GAD65 quantification, all images were normalized using the quantile-based normalization plugin Fiji (https://www.longair.net/edinburgh/imagej/quantile-normalization/). The intensity distribution of the images was normalized using 256 quantiles for each staining. The synaptic boutons were extracted from the background using the 3D Weka Segmentation plugin (https://imagej.net/Trainable_Weka_Segmentation) after manual selection of signal and background samples for training. The Fiji built-in plugin 3D object counter was then used to count and measure every object (cluster of GAD65-positive signal).

For VGLUT1 quantification, the raw images were first deconvoluted using Huygens software (Scientific Volume Imaging, The Netherlands). VGLUT1 density and volume were then quantified using built-in Spot detection analysis with Imaris 10.0 software (Oxford Instruments, United Kingdom).

Animals and husbandry

Data availability

All data are available in the manuscript or the Supplementary information. All materials are commercially available.

Acknowledgments

We thank Daniel Zell, Magdalena Weidner-Büchele and Johanna Diemer for preparatory experiments. Funding: Berta-Ottenstein-Program for Clinician Scientists, Faculty of Medicine, University of Freiburg, Germany (S.V.). Medical Research Foundation, France (FRM) grant AJE201912009450 (T.S.). University of Strasbourg Institute of Advanced Study (USIAS) grant 2020-035 (T.S.). The German Research Foundation (DFG) grants SE2666/2-1 (T.S., C.N.) and NO370/7-1 (C.N.). Centre National de la Recherche Scientifique, France, grant CNRS UPR3212 (T.S.). GO-Bio-initial grants INNOVADE 16LW0223 and 16LW0455 (C.N.) were obtained from the German Federal Ministry for Education and Research (BMBF).

Author contributions

Conceptualization, S.Ve., N.G., M.V., T.S., S.G., J.B., C.N.; Methodology, S.Ve., M.V., N.G., D.S., T.S., F.H., S.G., P.P., C.N.; Investigation, S.Ve., M.V., M.C.P., M.E., A.L., J.L., L.M.W., F.E., J.E., J.M., D.W., A.T., E.W., L.M.B., S.Vo., F.H., J.B., F.H., C.V., D.J., P.L., S.B., L.S. S.Z., P.P., D.S. G.S., J.B., G.L., C.D.V., E.G., N.G., T.S.; Formal analysis, S.Ve., M.C.P., P.P., M.V., S.G., A.M., N.G., T.S., C.N.; Visualization, S.Ve., A.L., A.M., S.G.,C.N.; Funding acquisition, S.Ve., K.D., T. S., C.N.; Project administration, S. Ve., T.S., M.V., C.N.; Supervision, S.Ve., N.G., K.D., S.G., T.S., M.V., C.N.; Writing – original draft, S.Ve., C.N.; Writing – review & editing, S.Ve., M.C.P., K.D., T.S., J.B., C.N.

Ethics declarations

Competing interests

C.N. and K.D. received lecture fees and advisory board honoraria from Johnson & Johnson, the manufacturer of Esketamine. C.N. received research support as a principal investigator in clinical trials sponsored by Johnson & Johnson. C.N. received honoraria as a member of a DMC board by Novartis. C.N. and S.Ve. are named coinventors on a patent filed by the University of Freiburg for the use of GluN2D inhibitors in the treatment of depression. T.S. received honoraria consulting Primetime Life Sciences, LLC. All other authors declare that they have no competing interests.

WHO | Depression and Other Common Mental Disorders. WHO http://www.who.int/mental_health/management/depression/prevalence_global_health_estimates/en/ (2019).
Berman, R. M. et al. Antidepressant effects of ketamine in depressed patients. Biol. Psychiatry 47, 351–354 (2000).
Johnston, J. N., Kadriu, B., Kraus, C., Henter, I. D. & Zarate, C. A. Ketamine in neuropsychiatric disorders: an update. Neuropsychopharmacology (2023) doi:10.1038/s41386-023-01632-1.
Zanos, P. & Gould, T. D. Mechanisms of ketamine action as an antidepressant. Mol. Psychiatry 23, 801–811 (2018).
Widman, A. J. & McMahon, L. L. Disinhibition of CA1 pyramidal cells by low-dose ketamine and other antagonists with rapid antidepressant efficacy. Proc. Natl. Acad. Sci. U.S.A. 115, E3007–E3016 (2018).
Gerhard, D. M. et al. GABA interneurons are the cellular trigger for ketamine’s rapid antidepressant actions. J Clin Invest 130, 1336–1349 (2020).
Duman, R. S., Sanacora, G. & Krystal, J. H. Altered Connectivity in Depression: GABA and Glutamate Neurotransmitter Deficits and Reversal by Novel Treatments. Neuron 102, 75–90 (2019).
Chowdhury, G. M. I. et al. Transiently increased glutamate cycling in rat PFC is associated with rapid onset of antidepressant-like effects. Mol Psychiatry 22, 120–126 (2017).
Lazarevic, V., Yang, Y., Flais, I. & Svenningsson, P. Ketamine decreases neuronally released glutamate via retrograde stimulation of presynaptic adenosine A1 receptors. Mol Psychiatry (2021) doi:10.1038/s41380-021-01246-3.
Salussolia, C. L., Prodromou, M. L., Borker, P. & Wollmuth, L. P. Arrangement of subunits in functional NMDA receptors. J Neurosci 31, 11295–11304 (2011).
Paoletti, P., Bellone, C. & Zhou, Q. NMDA receptor subunit diversity: impact on receptor properties, synaptic plasticity and disease. Nat Rev Neurosci 14, 383–400 (2013).
Monyer, H., Burnashev, N., Laurie, D. J., Sakmann, B. & Seeburg, P. H. Developmental and regional expression in the rat brain and functional properties of four NMDA receptors. Neuron 12, 529–540 (1994).
Traynelis, S. F. et al. Glutamate receptor ion channels: structure, regulation, and function. Pharmacol. Rev. 62, 405–496 (2010).
Wyllie, D. J. A., Livesey, M. R. & Hardingham, G. E. Influence of GluN2 subunit identity on NMDA receptor function. Neuropharmacology 74, 4–17 (2013).
Siegler Retchless, B., Gao, W. & Johnson, J. W. A single GluN2 subunit residue controls NMDA receptor channel properties via intersubunit interaction. Nat Neurosci 15, 406–413, S1-2 (2012).
Thompson, C. L., Drewery, D. L., Atkins, H. D., Stephenson, F. A. & Chazot, P. L. Immunohistochemical localization of N-methyl-D-aspartate receptor subunits in the adult murine hippocampal formation: evidence for a unique role of the NR2D subunit. Brain Res Mol Brain Res 102, 55–61 (2002).
Hanson, J. E. et al. Therapeutic potential of N-methyl-D-aspartate receptor modulators in psychiatry. Neuropsychopharmacol. 49, 51–66 (2024).
Perszyk, R. E. et al. GluN2D-Containing N-methyl-d-Aspartate Receptors Mediate Synaptic Transmission in Hippocampal Interneurons and Regulate Interneuron Activity. Mol. Pharmacol. 90, 689–702 (2016).
von Engelhardt, J. et al. GluN2D-containing NMDA receptors-mediate synaptic currents in hippocampal interneurons and pyramidal cells in juvenile mice. Front Cell Neurosci 9, (2015).
Tzilivaki, A. et al. Hippocampal GABAergic interneurons and memory. Neuron 111, 3154–3175 (2023).
Pelkey, K. A. et al. Hippocampal GABAergic Inhibitory Interneurons. Physiol Rev 97, 1619–1747 (2017).
Schulz, J. M., Knoflach, F., Hernandez, M.-C. & Bischofberger, J. Dendrite-targeting interneurons control synaptic NMDA-receptor activation via nonlinear α5-GABA A receptors. Nat Commun 9, 1–16 (2018).
Royer, S. et al. Control of timing, rate and bursts of hippocampal place cells by dendritic and somatic inhibition. Nat Neurosci 15, 769–775 (2012).
Lovett-Barron, M. et al. Regulation of neuronal input transformations by tunable dendritic inhibition. Nat Neurosci 15, 423–430, S1-3 (2012).
Cathomas, F. et al. Whole blood transcriptional signatures associated with rapid antidepressant response to ketamine in patients with treatment resistant depression. Transl Psychiatry 12, 12 (2022).
Swanger, S. A. et al. A Novel Negative Allosteric Modulator Selective for GluN2C/2D-Containing NMDA Receptors Inhibits Synaptic Transmission in Hippocampal Interneurons. ACS Chem Neurosci 9, 306–319 (2018).
Zhang, J. et al. Distinct structure and gating mechanism in diverse NMDA receptors with GluN2C and GluN2D subunits. Nat Struct Mol Biol 30, 629–639 (2023).
Yi, F., Bhattacharya, S., Thompson, C. M., Traynelis, S. F. & Hansen, K. B. Functional and pharmacological properties of triheteromeric GluN1/2B/2D NMDA receptors. The Journal of Physiology 597, 5495–5514 (2019).
Orser, B. A., Pennefather, P. S. & MacDonald, J. F. Multiple mechanisms of ketamine blockade of N-methyl-D-aspartate receptors. Anesthesiology 86, 903–917 (1997).
Kotermanski, S. E. & Johnson, J. W. Mg2+ imparts NMDA receptor subtype selectivity to the Alzheimer’s drug memantine. J. Neurosci. 29, 2774–2779 (2009).
Kapfer, C., Glickfeld, L. L., Atallah, B. V. & Scanziani, M. Supralinear increase of recurrent inhibition during sparse activity in the somatosensory cortex. Nat Neurosci 10, 743–753 (2007).
Ali, F. et al. Ketamine disinhibits dendrites and enhances calcium signals in prefrontal dendritic spines. Nat Commun 11, 72 (2020).
Ribeiro, P. O., Tomé, A. R., Silva, H. B., Cunha, R. A. & Antunes, L. M. Clinically relevant concentrations of ketamine mainly affect long-term potentiation rather than basal excitatory synaptic transmission and do not change paired-pulse facilitation in mouse hippocampal slices. Brain Res. 1560, 10–17 (2014).
Citri, A. & Malenka, R. C. Synaptic plasticity: multiple forms, functions, and mechanisms. Neuropsychopharmacology 33, 18–41 (2008).
Normann, C. et al. Associative long-term depression in the hippocampus is dependent on postsynaptic N-type Ca2+ channels. J Neurosci 20, 8290–8297 (2000).
Joffe, M. E. et al. Acute restraint stress redirects prefrontal cortex circuit function through mGlu5 receptor plasticity on somatostatin-expressing interneurons. Neuron 110, 1068-1083.e5 (2022).
Hu, W., Zhang, M., Czéh, B., Flügge, G. & Zhang, W. Stress impairs GABAergic network function in the hippocampus by activating nongenomic glucocorticoid receptors and affecting the integrity of the parvalbumin-expressing neuronal network. Neuropsychopharmacology 35, 1693–1707 (2010).
Troppoli, T. A. et al. Negative Allosteric Modulation of Gamma-Aminobutyric Acid A Receptors at α5 Subunit-Containing Benzodiazepine Sites Reverses Stress-Induced Anhedonia and Weakened Synaptic Function in Mice. Biol Psychiatry 92, 216–226 (2022).
Xu, L., Anwyl, R. & Rowan, M. J. Behavioural stress facilitates the induction of long-term depression in the hippocampus. Nature 387, 497–500 (1997).
Alfarez, D. N., Joels, M. & Krugers, H. J. Chronic unpredictable stress impairs long-term potentiation in rat hippocampal CA1 area and dentate gyrus in vitro. Eur J Neurosci 17, 1928–34 (2003).
Holderbach, R., Clark, K., Moreau, J. L., Bischofberger, J. & Normann, C. Enhanced long-term synaptic depression in an animal model of depression. Biol Psychiatry 62, 92–100 (2007).
Lovett-Barron, M. et al. Dendritic inhibition in the hippocampus supports fear learning. Science 343, 857–863 (2014).
Normann, C. et al. Antidepressants Rescue Stress-Induced Disruption of Synaptic Plasticity via Serotonin Transporter-Independent Inhibition of L-Type Calcium Channels. Biol. Psychiatry 84, 55–64 (2018).
Serchov, T. et al. Increased Signaling via Adenosine A1 Receptors, Sleep Deprivation, Imipramine, and Ketamine Inhibit Depressive-like Behavior via Induction of Homer1a. Neuron 87, 549–562 (2015).
Vestring, S., Serchov, T. & Normann, C. Animal Models of Depression - Chronic Despair Model (CDM). JoVE (Journal of Visualized Experiments) e62579 (2021) doi:10.3791/62579.
Toni, N., Buchs, P. A., Nikonenko, I., Bron, C. R. & Muller, D. LTP promotes formation of multiple spine synapses between a single axon terminal and a dendrite. Nature 402, 421–5 (1999).
Engert, F. & Bonhoeffer, T. Dendritic spine changes associated with hippocampal long-term synaptic plasticity. Nature 399, 66–70 (1999).
Alboni, S. et al. Fluoxetine effects on molecular, cellular and behavioral endophenotypes of depression are driven by the living environment. Mol. Psychiatry 22, 552–561 (2017).
Honoré, E. & Lacaille, J.-C. Object location learning in mice requires hippocampal somatostatin interneuron activity and is facilitated by mTORC1-mediated long-term potentiation of their excitatory synapses. Mol Brain 15, 101 (2022).
Nicolás, M. J. et al. Ketamine-induced oscillations in the motor circuit of the rat basal ganglia. PLoS One 6, e21814 (2011).
Küppenbender, K. D. et al. Expression of NMDA receptor subunit mRNAs in neurochemically identified projection and interneurons in the human striatum. J Comp Neurol 419, 407–421 (2000).
Gideons, E. S., Kavalali, E. T. & Monteggia, L. M. Mechanisms underlying differential effectiveness of memantine and ketamine in rapid antidepressant responses. Proc Natl Acad Sci U S A 111, 8649–8654 (2014).
Isaacson, J. S. & Scanziani, M. How Inhibition Shapes Cortical Activity. Neuron 72, 231–243 (2011).
Hu, H., Gan, J. & Jonas, P. Interneurons. Fast-spiking, parvalbumin⁺ GABAergic interneurons: from cellular design to microcircuit function. Science 345, 1255263 (2014).
Martin, L. J. et al. α5GABAA Receptor Activity Sets the Threshold for Long-Term Potentiation and Constrains Hippocampus-Dependent Memory. J. Neurosci. 30, 5269–5282 (2010).
Gentet, L. J. et al. Unique functional properties of somatostatin-expressing GABAergic neurons in mouse barrel cortex. Nat Neurosci 15, 607–612 (2012).
Major, G., Larkum, M. E. & Schiller, J. Active properties of neocortical pyramidal neuron dendrites. Annu Rev Neurosci 36, 1–24 (2013).
Takahashi, H. & Magee, J. C. Pathway interactions and synaptic plasticity in the dendritic tuft regions of CA1 pyramidal neurons. Neuron 62, 102–111 (2009).
Castrén, E. & Antila, H. Neuronal plasticity and neurotrophic factors in drug responses. Mol Psychiatry 22, 1085–1095 (2017).
Minichiello, L. TrkB signalling pathways in LTP and learning. Nat Rev Neurosci 10, 850–860 (2009).
Sandi, C. & Pinelo-Nava, M. T. Stress and Memory: Behavioral Effects and Neurobiological Mechanisms. Neural Plast 2007, 78970 (2007).
Joëls, M. Corticosteroid effects in the brain: U-shape it. Trends Pharmacol Sci 27, 244–250 (2006).
Tartt, A. N., Mariani, M. B., Hen, R., Mann, J. J. & Boldrini, M. Dysregulation of adult hippocampal neuroplasticity in major depression: pathogenesis and therapeutic implications. Mol Psychiatry 27, 2689–2699 (2022).
Castren, E. Is mood chemistry? Nat Rev Neurosci 6, 241–6 (2005).
Pittenger, C. & Duman, R. S. Stress, Depression, and Neuroplasticity: A Convergence of Mechanisms. Neuropsychopharmacol 33, 88–109 (2008).
Kendler, K. S., Karkowski, L. M. & Prescott, C. A. Causal relationship between stressful life events and the onset of major depression. Am J Psychiatry 156, 837–41 (1999).
Normann, C., Schmitz, D., Fürmaier, A., Döing, C. & Bach, M. Long-term plasticity of visually evoked potentials in humans is altered in major depression. Biol. Psychiatry 62, 373–380 (2007).
Maya Vetencourt, J. F. et al. The antidepressant fluoxetine restores plasticity in the adult visual cortex. Science 320, 385–8 (2008).
Moda-Sava, R. N. et al. Sustained rescue of prefrontal circuit dysfunction by antidepressant-induced spine formation. 13 (2019).
Reif, A. et al. Esketamine Nasal Spray versus Quetiapine for Treatment-Resistant Depression. N Engl J Med 389, 1298–1309 (2023).
Doyle, M. A., Brown, J. A. & Winder, D. G. GluN2D expression is regulated by restraint stress and supports active stress coping bouts. Neuropharmacology 225, 109377 (2023).
Yamamoto, H. et al. Loss of GluN2D subunit results in social recognition deficit, social stress, 5-HT2C receptor dysfunction, and anhedonia in mice. Neuropharmacology 112, 188–197 (2017).
Hanson, E. et al. Tonic Activation of GluN2C/GluN2D-Containing NMDA Receptors by Ambient Glutamate Facilitates Cortical Interneuron Maturation. J Neurosci 39, 3611–3626 (2019).
Zhang, Y. et al. Structural basis of ketamine action on human NMDA receptors. Nature 596, 301–305 (2021).
Casarotto, P. C. et al. Antidepressant drugs act by directly binding to TRKB neurotrophin receptors. Cell 184, 1299-1313.e19 (2021).
Ma, S. et al. Sustained antidepressant effect of ketamine through NMDAR trapping in the LHb. Nature 622, 802–809 (2023).
Zanos, P. et al. NMDAR inhibition-independent antidepressant actions of ketamine metabolites. Nature 533, 481–486 (2016).
Abdallah, C. G. (2R,6R)-Hydroxynorketamine (HNK) plasma level predicts poor antidepressant response: is this the end of the HNK pipeline? Neuropsychopharmacology 45, 1245–1246 (2020).
Njoo, C., Heinl, C. & Kuner, R. In Vivo SiRNA Transfection and Gene Knockdown in Spinal Cord via Rapid Noninvasive Lumbar Intrathecal Injections in Mice. J Vis Exp (2014) doi:10.3791/51229.
Schmittgen, T. D. & Livak, K. J. Analyzing real-time PCR data by the comparative CT method. Nat Protoc 3, 1101–1108 (2008).
Lu, C. et al. OPLS4: Improving Force Field Accuracy on Challenging Regimes of Chemical Space. J. Chem. Theory Comput. 17, 4291–4300 (2021).
Madhavi Sastry, G., Adzhigirey, M., Day, T., Annabhimoju, R. & Sherman, W. Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments. J Comput Aided Mol Des 27, 221–234 (2013).
Friesner, R. A. et al. Glide: A New Approach for Rapid, Accurate Docking and Scoring. 1. Method and Assessment of Docking Accuracy. J. Med. Chem. 47, 1739–1749 (2004).
Bender, V. A., Bender, K. J., Brasier, D. J. & Feldman, D. E. Two coincidence detectors for spike timing-dependent plasticity in somatosensory cortex. J Neurosci 26, 4166–77 (2006).
Niehusmann, P. et al. Coincidence detection and stress modulation of spike time-dependent long-term depression in the hippocampus. J Neurosci 30, 6225–35 (2010).
Jackman, S. L. & Regehr, W. G. The Mechanisms and Functions of Synaptic Facilitation. Neuron 94, 447–464 (2017).
Ohshima, M., Itami, C. & Kimura, F. The α2A -adrenoceptor suppresses excitatory synaptic transmission to both excitatory and inhibitory neurons in layer 4 barrel cortex. J Physiol 595, 6923–6937 (2017).
Malinow, R. & Tsien, R. W. Presynaptic enhancement shown by whole-cell recordings of long-term potentiation in hippocampal slices. Nature 346, 177–180 (1990).

Yes there is potential Competing Interest. C.N. and K.D. received lecture fees and advisory board honoraria from Johnson & Johnson, the manufacturer of Esketamine. C.N. received research support as a principal investigator in clinical trials sponsored by Johnson & Johnson. C.N. received honoraria as a member of a DMC board by Novartis. C.N. and S.Ve. are named coinventors on a patent filed by the University of Freiburg for the use of GluN2D inhibitors in the treatment of depression. T.S. received honoraria consulting Primetime Life Sciences, LLC. All other authors declare that they have no competing interests.

TableS1.xlsx
Supplementary table
SupplementaryinformationVestring.docx
Supplementary information

Download PDF

Version 1

posted

You are reading this latest preprint version

The NMDA receptor subunit GluN2D is a target for rapid antidepressant action

Status:

Version 1

Abstract

Figures

Introduction

Results

Inhibition of GluN2D decreases the activity of feed-forward and feedback loops

Inhibition of GluN2D and SOM-IN activity increases hippocampal LTP

Inhibition of GluN2D restores LTP in an animal model of depression

GluN2D inhibition shifts synapses toward excitation

GluN2D inhibition has rapid antidepressant activity

No effect of NAB-14 on spatial memory, motor coordination, locomotion or anxiety

Discussion

Methods

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1