Increased activity of brain cells controlling memory by boosting NMDA receptor function

All animal procedures were approved by the Emory University Institutional Animal Care and Use Committee (protocol: 201 700 124), fully certified by the Association for Assessment and Accreditation of Laboratory Animal Care International, and were performed in accordance with state and federal Animal Welfare Acts and the policies of the Public Health Service. C57BL/6J mice were purchased from Jackson Laboratory (catalogue no. 000664) and bred in animal facilities at Emory University. Grin2d knockout (GluN2D‐KO) mice were kindly provided by Dr S. Nakanishi at Kyoto University. GluN2D‐KO mouse genotyping was performed by TransnetYX using the following primers and reporters: GluN2D‐KO forward primer, GCCTTCTTGACGAGTTCTTCTGA; GluN2D‐KO reverse primer, CACGAGGAGCATGTAGAAGACA; Grin2d wild‐type (WT) forward primer, ATGTTGTCGATGATCCCC; and Grin2d WT reverse primer, CAACGACAAAATCGAGGTGATGAG. Mice were housed in ventilated cages with a 12 h–12 h dark–light cycle, at controlled temperature (22–23°C) with humidity at ∼60% and ad libitum access to food and water. For experimentation, animals were randomly selected from various breeder pairs or trios. GluN2D‐KO mice were kept separated from heavy foot traffic and on a 10 h–14 h (light–dark) cycle to improve poor breeding.

For live cell Ca²⁺‐imaging experiments, animal use was conducted in accordance with guidelines approved by the Administrative Panel on Laboratory Animal Care at University of Colorado, Anschutz School of Medicine, accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (00235) and approved under the animal protocol #0300. Timed pregnant Sprague–Dawley rat dams (typically, day 16) were obtained from Charles River Laboratories and housed in standard conditions (12 h–12 h light–dark cycle, with food and water ad libitum) until the litter was born.

(+)‐MK‐801 maleate, (−)‐bicuculline methiodide and NBQX were obtained from Tocris; d,l‐AP5 sodium salt was obtained from HelloBio (HB0252‐50mg); biocytin was obtained from Cayman; anti‐streptavidin Alexa Fluor™ 546 conjugate was obtained from ThermoFisher; and EU1622‐240 was synthesized according to published protocols (Chou, Epstein et al. 2024). All other chemicals were from Sigma. EU1622‐240 was dissolved in DMSO to a stock concentration of 10 mm, and working concentrations were made at 0.1, 0.3, 1, 3, and 6 µM. Vehicle treatments of DMSO were made to match the DMSO percentage in each final working solution. Drug stocks for experiments were made fresh on the first day that testing was performed. No particulate matter was observed in the solution, and there was no further evidence of drug precipitation during the electrophysiological experiments. The stocks were stored at −20°C.

Coronal brain slices (280–300 µm thick) were made from WT or GluN2D‐KO mice, postnatal day 17–22, of both sexes. Mice were killed by isoflurane overdose followed by decapitation and removal of the brain. The brain was rapidly placed into ice‐cold artificial cerebrospinal fluid (aCSF) composed of (mM): 230 sucrose, 24 NaHCO₃, 10 glucose, 2.5 KCl, 1.25 NaH₂PO₄, 3 sodium pyruvate, 5 sodium l‐ascorbate, 12 N‐acetyl‐cysteine, 10 MgCl₂ and 0.25 CaCl₂ saturated with 95% O₂–5% CO₂. The brain was then hemisected, glued to the removable stage and sectioned on a vibratome into 280‐ to 300‐µm‐thick slices (Leica VT1200S, Wetzlar, Germany). After sectioning, slices were transferred to a holding chamber with the same solution, except that Mg²⁺ was lowered to 1.5 mm. Slices were incubated at 32°C for 30 min in this solution before returning them to room temperature for ≥1 h recovery prior to use.

Slices were transferred to a submersion chamber, and neurons were identified via infrared light differential interference contrast (IR‐DIC) on an upright microscope (BX50WI; Olympus). Slices were continuously perfused at a rate of 4 ml/min with oxygenated standard recording aCSF containing (mM): 126 NaCl, 26 NaHCO₃, 10 glucose, 2.5 KCl, 1.25 NaH₂PO₄, 1.5 MgSO₄ and 1.5 CaCl₂ bubbled with 95% O₂–5% CO₂. The temperature of the external solution was heated to 30–32°C by a TC‐344C inline heater system (Warner Instruments). The recording chamber was flushed with ∼10 ml of 70% ethanol and ∼20 ml of diH₂O after each slice was recorded. Recording electrodes were made from borosilicate glass (1.50 mm outer diameter, 1.12 mm inner diameter; World Precision Instruments) via a micropipette puller (P‐1000; Sutter Instruments); pipette resistances were 6–8 MΩ. Recordings were made using a Multiclamp 700B amplifier (Molecular Devices), filtered at 2 kHz using an eight‐pole Bessel filter (−3 dB), and digitized at 20 kHz (Digidata 1550B) using Axon pClamp10 software.

For intrinsic and action potential firing properties, the intracellular solution contained (mM): 115 potassium gluconate, 0.6 EGTA, 2 MgCl₂, 2 Na₂ATP, 0.3 Na₂GTP, 10 HEPES, 5 sodium phosphocreatine, 8 KCl and 0.5% biocytin. After obtaining a whole‐cell configuration, cells were dialysed for 5 min in current‐clamp mode at I = 0 (without current injected). The liquid junction potential was not corrected for resting membrane potential (RMP), and all current‐clamp responses were automatically bridge balanced using the Multiclamp 700B clamp commander software. Input resistance was calculated by using the steady‐state voltage change in response to a −30 pA, 200 ms hyperpolarizing current injection period (Oginsky, Cui et al. 2014; Zhong, Cui et al. 2015) or using the slope of voltage deflections in response to 500 ms current injections from −120 to +12 pA, with Δ12 pA step change at an inter‐sweep interval of 2 s. The evoked action potential firing frequency was calculated in response to a 500 ms current injection every 2 s, starting at −40 pA and increasing by 20 pA until the cell displayed depolarization‐induced block of firing.

For voltage‐clamp recordings of evoked NMDAR‐mediated excitatory postsynaptic currents (EPSCs), the electrodes were filled with internal solution containing (mM): 105 caesium gluconate, 5 CsCl, 8 NaCl, 5 sodium phosphocreatine, 5 MgCl₂, 2 Na‐ATP, 0.3 Na‐GTP, 0.6 EGTA, 5 BAPTA, 40 HEPES and 5 QX314 (pH 7.3, adjusted with CsOH, ∼290 mOsmol/kg). Both 10 µM gabazine and 10 µM NBQX were added to the external solution to block GABA_A receptors and AMPA receptors, respectively. Cells were held at +40 mV, and NMDAR‐mediated EPSCs were recorded by injecting 50–120 µA of current for 0.1 ms using a monopolar platinum–iridium stimulating electrode (FHC) placed within the Schaffer collaterals at 0.03 Hz. Solutions containing drugs or vehicle were applied for 5 min, or as indicated. The NMDAR pore blocker MK‐801 (3 µM) was applied at the end of each recording to block; MK‐801 was more effective at blocking NMDARs in the presence of EU1622‐240 than competitive antagonists given the increased potency for glutamate of drug‐bound receptors (Ullman, Perszyk et al. 2024).

For voltage‐clamp recordings of spontaneous inhibitory postsynaptic current (sIPSC), the electrodes were filled with internal solution containing (mM): 110 caesium gluconate, 4 NaCl, 30 CsCl, 5 BAPTA, 5 HEPES, 0.5 CaCl₂, 2 MgCl₂, 2 Na‐ATP and 0.3 Na‐GTP. The aCSF for sIPSC recording contained (mM): 130 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 25 NaHCO₃, 1 MgCl₂, 2 CaCl₂ and 20 glucose. The sIPSCs were isolated by holding the CA1 pyramidal cell at the EPSC reversal potential, which was +10 mV for our solutions. The experimental recording protocol for EPSCs and sIPSC recording was 5 min of baseline recording, followed by wash‐in of 3 µM EU1622‐240 for 10 min. Recordings were finished by switching to aCSF that contained EU1622‐240 plus 10 µM gabazine to confirm that the currents we recorded were mediated by GABA_A receptors. The last 2 min of the recording for each condition were used for data analysis. Parallel control experiments were completed with the same recording protocol, perfusing 400 µM d,l‐APV in the aCSF to competitively antagonize GluN2D‐containing NMDARs throughout the entire recording period.

To record a compound excitatory postsynaptic potential (EPSP) and inhibitory postsynaptic potential (IPSP), a CA1 pyramidal cell was patched in current‐clamp mode with injection of current needed to set the RMP to −60 mV. The cell was rejected if the holding current was larger than ± 75 pA. Recording solution was the same aCSF as described above, but with 1.5 mm Mg²⁺ and 1.5 mm Ca²⁺. Electrical stimulation was applied to the Schaffer collaterals to evoke a compound postsynaptic potential, in which the depolarizing phase was classified as the EPSP and the hyperpolarizing phase as the IPSP. Electrical stimulation sites were set for two places: the distal CA3 region, ∼800 µm away from the patched CA1 pyramidal cell, or the local CA1 region, ∼200 µm away from the patched CA1 pyramidal cell (Dingledine, Roth et al. 1987). Stable recordings were made for ≥ 5 min of baseline, then 3 µM EU1622‐240 was applied for ≥ 10 min. The last 2 min of recording was used for data analysis and presentation of representative traces. Amplitudes of the EPSP and IPSP were measured, and the EPSP/IPSP amplitude ratio was calculated. Given the strong potentiating effect of EU1622‐240 on NMDARs, the stimulus strength was adjusted such that the initial postsynaptic potential depolarization during baseline was ∼2 mV to avoid generating action potentials during EU1622‐240 application.

Series resistance was monitored throughout experiments and was typically 8−12 MΩ. For current‐clamp experiments, cells were briefly switched to voltage clamp and held at −60 mV while a series of 50 ms duration, 5 mV square hyperpolarizing potentials were applied to the cell. Current‐clamp recordings had series resistance measurements made at the beginning and at the end of each experiment, which usually lasted 5–10 min. All series resistances were measured off‐line by analysing the peak of the capacitive charging spike and applying Ohm's law. If the series resistance (R_s) changed by >25% during the experiment or exceeded 30 MΩ, the cell was excluded. For voltage‐clamp recordings of EPSCs, a −5 mV hyperpolarizing pulse was applied during each recording epoch before the evoked EPSC to determine R_s. Only stable recordings were accepted, defined as those with a drift in R_s of <25% over the course of the recordings.

Wild‐type postnatal day 20–25 C57BL/6J mice of both sexes were used for field potential recordings. Coronal brain slices were sectioned at 500 µm thickness using the same procedures described above. The protocol for recording and analysing field potentials corresponding to field EPSPs within the stratum radiatum close to the border of the stratum pyramidale was described by Lundbye, Toft et al. (2018). Hippocampal slices were transferred to an interface‐style recording chamber (BSC 1‐1, Scientific Systems Design Inc.) and bathed in the recording aCSF described above with 2.0 mm MgSO₄ and 3.0 mm CaCl₂, heated to 31–33°C. A monopolar platinum electrode was used to stimulate Schaffer collaterals at 0.03 Hz, and a recording electrode was placed in the upper one‐third of the stratum radiatum of CA1. Recording electrodes were made from borosilicate glass (1.50 mm outer diameter, 1.12 mm inner diameter; World Precision Instruments) via a micropipette puller (P‐1000; Sutter Instrument); pipette resistances were 4–6 MΩ, filled with recording aCSF. Signals were obtained by an A‐M Systems Model 1800 field recording amplifier using ×1000 gain, with high‐ and low‐pass filtering at 1 Hz and 5 kHz (−3 dB Bessel), respectively. Recordings were digitized at 20 kHz using NacGather acquisition software v.2.07 (Theta Burst, Irvine, CA, USA).

For input–output (I/O) curves, the stimulation level was ∼40–300 µA at 0.1 ms duration. Three sweeps were collected for each stimulation level during I/O curve recording. The slices that went through I/O curve recording were not used for assessment of LTP. For LTP recording, a brief preliminary I/O curve was recorded, with one sweep per stimulation level at 30, 60 and 100 µA stimulation levels, then the stimulation level was set to the level that yielded a response amplitude of ∼50% compared with that obtained in response to 100 µA stimulation. The slice was equilibrated in the recording chamber with the treatment solution for 20 min, resulting in a baseline change of <5%. Then the recording was started for 20 min of baseline, followed by a theta‐burst stimulation protocol (each protocol consisted of a train of 10 theta bursts, each containing four pulses at 100 Hz with an inter‐burst interval of 200 ms) to induce a form of LTP. The LTP recording protocol was: (i) 20 min low‐frequency stimulation at 0.03 Hz; (ii) theta‐burst stimulation; (iii) repeat periods (i) and (ii) three more times such that a total of four theta‐burst stimulation protcols were applied; and (iv) finished with a 20 min low‐frequency stimulation (Lundbye, Toft et al. 2018; Morgan & Teyler 2001). The effects of both vehicle (0.3% DMSO) and 3 µM EU1622‐240 were tested. To avoid contamination of remaining EU1622‐240 between each recording, perfusion lines, chambers and slice anchors were washed once with 50 ml of 10% DMSO, then twice with 50 ml of 70% ethanol, twice with 50 ml of deionized H₂O, then once with 50 ml of external recording solution after each slice.

Post hoc immunohistochemical staining for biocytin was performed to confirm that the proper cell type was recorded, used largely to confirm that recordings made from cells in stratumpyramidale were from pyramidal cells (i.e. contained a large apical dendrite with dendritic spines) and not from GABAergic interneurons (Pelkey, Chittajallu et al. 2017). After each recording, the patch pipette was withdrawn carefully from the patched cell to allow recovery for resealing of the cell membrane. The slice was subsequently transferred into 4% paraformaldehyde and stored at 4°C overnight. The next day, the slice was washed three times with PBS, then incubated in a 1.2% Triton‐X–PBS solution for 10 min. The slice was subsequently incubated in the following solution overnight at room temperature for biocytin staining: 10% normal goat serum (Abcam, ab7481), 0.3% Triton‐X (Fisher Scientific, BP151‐500), 2% bovine serum albumin (Jackson ImmunoResearch, 156 231), and anti‐streptavidin Alexa546 1:500 (Invitrogen, S11225) dissolved in 1× PBS. On the next day, the slice was washed with PBS three times, mounted onto a slide, and allowed to air dry for 4 h. Then, a coverslip was placed over the slide using Prolong Gold mounting medium (Invitrogen, P36934). Neurons were imaged at 20x magnification in 1‐µm‐thick optical sections on a spinning disc confocal microscope, and a maximum‐intensity projection was made from the z‐stack.

For live cell Ca²⁺‐imaging experiments, dissociated hippocampal neurons were prepared from male and female postnatal day 0–2 Sprague–Dawley rat pups. Pups were rapidly decapitated and the hippocampi dissected and dissociated using papain (20 units/ml) digestion for 45–60 min at room temperature with gentle rocking. The cells were triturated five or six times through a fire‐polished pipette and plated on 18 mm glass coverslips coated with poly‐d‐lysine (Sigma, P6407) in minimal essential medium containing 10% FBS (Cytiva, SH30071.02HI) and penicillin–streptomycin, at a density of 100,pp0 cells per well (12‐well dish). After 24 h, the medium was switched to Neurobasal‐A medium (Invitrogen, 10888022) with B27 (Invitrogen, 17504044) and Glutamax (ThermoFisher, 35050061) with anti‐mitotic (10 µm 5‐fluorodeoxyuridine) added after 7 days in culture to block glial overgrowth. The neurons were maintained at 37°C in a humidified incubator with 5% CO₂. Hippocampal neurons were transfected between 14 and 16 days in vitro using lipofectamine 2000 (ThermoFisher, 11668027) according to the manufacturer's protocol. Microscopy experiments were performed within 24 h of transfection.

Live‐cell Ca²⁺ imaging was performed at 32°C on an Olympus IX‐71 equipped with a spinning disc scan head (Yokogawa). Images were acquired using an Olympus 60× Plan Apochromat 1.4 numerical aperture objective and collected on a 1024 × 1024 pixel Andor iXon EM‐CCD camera, with pixels binned 3 × 3. Data acquisition and analysis were performed with MetaMorph (Molecular Devices) and ImageJ software. To image NMDAR‐mediated quantal Ca²⁺ transients, neurons were transfected with GCaMP6s (Addgene plasmid #40753) at 12–14 days in vitro. Neurons were imaged 1 day later in Mg²⁺‐free extracellular solution containing (mM): 130 NaCl, 5 KCl, 10 HEPES, 30 glucose, 2 CaCl₂, 0.003 glycine and 0.002 TTX, pH 7.4. Single z‐plane images of the dendritic arbor were acquired at 15 Hz for 1–2 min to record baseline quantal Ca²⁺ transients. Then, vehicle (DMSO) or EU1622‐240 was added to the recording chamber (for final concentrations of 1, 3 or 9 µM), and the same cells were imaged 2–10 min after drug application. We performed a third round of imaging of the same cells after addition of 100 µM d,l‐APV and 20 µM MK‐801 (in the continued presence of EU‐1622‐240) to confirm that enhanced Ca²⁺ entry was NMDAR dependent.

To quantify the quantal Ca²⁺ transient amplitude, regions of interest were drawn around clearly resolved spines in the baseline (prior to EU1622‐240 addition, see Supporting_Video S1). The same regions of interest were transferred to the post‐drug (see Supporting_Video S1). The mean background‐subtracted GCaMP6s fluorescence within each region of interest was determined pre‐ and post‐drug. Only synapses displaying Ca²⁺ transient events both before and after EU1622‐240 addition were included in the analysis, allowing paired comparison of event amplitudes at each individual synapse, pre/post drug application.

Whole‐cell electrophysiology data were collected by Clampex 10.7 with MultiClamp 700B Commander. The sIPSC data were analysed offline using MiniAnalysis 6 (Synaptosoft) with an 8‐pA threshold for event detection, which was 5‐fold higher than our peak root mean square noise. For all whole‐cell current responses, all data were analysed offline using ChanneLab 2.210604 (Dr Stephen F. Traynelis) or pClamp 10.3 (Molecular Devices) on the average response per condition, except for charge transfer for evoked NMDAR‐mediated EPSCs, for which each trace was analysed individually. Rise time was measured as the 10%–90% fractional response of the rising phase of each current response. Peak amplitude was determined by finding the maximum of a moving search average of three to five data points. The deactivation time course for evoked NMDAR‐mediated EPSCs was measured by fitting a dual‐exponential function to the decay; the time constants are reported as weighted deactivation time constants (τ_w), calculated using the following equation (τ_w, weighted deacivation time constants; Amp_fast, amplitude of fast component of the EPSC; Amp_slow, amplitude of slow component of the EPSC; τ_fast, deacivation time constants of fast component of the EPSC; τ_slow, deacivation time constants of slow component of the EPSC): τw=AmpfastAmpfastAmpfast+AmpslowAmpfast+Ampslowτfast+AmpslowAmpslowAmpfast+AmpslowAmpfast+Ampslowτslow)

Charge transfer of current responses was calculated as the area under the curve (amplitude × τ) for each response. The EPSP/IPSP amplitudes were analysed offline by Clampfit 10.3. Traces with action potentials or postsynaptic potentials that were not able to be separated from the stimulation artefact were rejected. For input resistance, we analysed the response to −30 pA, 200 ms hyperpolarizing injected current to obtain the membrane voltage change in response to current injection. All other input resistance data were analysed by fitting a linear regression of the membrane potential response to step current injection. Input resistance was calculated according to Ohm's law. Slice field potential data were analysed offline by NACShow2 (Theta Burst Corp.) to obtain the slope of the rising phase of the field EPSP (fEPSP).

Data analysis and figures were made with the assistance of GraphPad Prism 9.3.1 (GraphPad Software, LLC) and OriginPro 9.0.0 (OriginLab Corporation). Data are presented as the mean ± SD. Significance was determined via Student's t‐test or ANOVA as appropriate. An α‐value of p < 0.05 was accepted as statistically significant. Evoked EPSCs were analysed by Student's paired t‐test (two‐tailed); current‐clamp data were analysed by ordinary one‐way ANOVA with Dunnett's multiple comparisons test, and bicuculline‐treated intrinsic membrane properties were analysed by repeated‐measures ANOVA with Tukey's multiple comparisons test (details can be found in each table). When multiple tests were performed on the same recording, α‐values were corrected to prevent family‐wise errors using the Bonferroni post hoc correction method. All studies were designed such that an effect size of at least one was detected at ≥80% power.

Initially, we assessed the effects of EU1622‐240 on evoked synaptic NMDAR‐mediated EPSCs from both CA1 interneurons and CA1 pyramidal cells. We held neurons at +40 mV to remove voltage‐dependent Mg²⁺‐induced channel block, and we evoked EPSCs with Schaffer collateral stimulation. Figure 2 shows the effects of 3 µm EU1622‐240 on evoked EPSCs recorded from CA1 stratum radiatum interneurons (Fig. 2A) and pyramidal cells (Fig. 2B). For both CA1 interneurons and pyramidal cells, we observed a modest increase in EPSC amplitude and charge transfer (Fig. 2C–E; Table 1). Interneurons showed no detectable increase in the mean weighted time constant for deactivation (Fig. 2D1; Table 1), and no significant effect for decay time was observed for CA1 pyramidal cells (Fig. 2D2; Table 1).

We tested the impact of EU1622‐240 (labelled as 'EU240' in blue in Fig. 3) on neuronal function recorded under current clamp from hippocampal CA1 stratum radiatum interneurons and pyramidal cells (Fig. 3). We washed in sequentially increasing concentrations of EU1622‐240 of 0.1, 0.3, 1, 3 and 6 µm or matched vehicle (DMSO). Each drug was washed in for 5 min, with the first 1 min estimated to be the time needed to reach the bath (24 s) plus the time it takes to exchange the bath volume five times (38 s). Throughout the recording, we monitored instantaneous firing frequency and RMP. Figure 3A1 shows a typical recording from a CA1 stratum radiatum interneuron during a baseline period, followed by the application of increasing concentrations of EU1622‐240. There is a clear increase in the action potential firing frequency (Fig. 3A1) in this interneuron, in addition to a depolarization of RMP (Fig. 3A2) during drug application. Figure 3B shows that vehicle (0.01–0.6% DMSO) had no effect over the same recording period. Figure 4A–C summarizes the instantaneous firing frequency, RMP and input resistance for interneurons during the application of EU1622‐240. These data support the idea that EU1622‐240 can increase the excitability of interneurons when applied at concentrations of 3–6 µm. The concentration of 3–6 µm is likely to correspond to a 'within‐the‐slice' drug concentration of ≥1 µM, assuming the drug takes several minutes to penetrate the tissue. We also observed both statistically significant depolarization at 6 µM and reduction of input resistance at 3 µM (see Table 2). These results are consistent with an EU1622‐240‐induced increase in excitatory drive onto interneurons through potentiation of NMDARs that are likely to contain the GluN2D subunit, which showed the strongest level of enhancement by this PAM (Fritzemeier, Akins et al. 2025).

Figures 3C1 and C2 shows a typical recording from a CA1 pyramidal cell during a baseline period, followed by a recording during the application of increasing concentrations of EU1622‐240 (from 0.1 to 6 µm) by the same protocol as used for interneurons. Figure 3D1 and D2 shows the effects of vehicle (0.01–0.6% DMSO) during the same recording period. There is no obvious change in neuronal properties monitored during either EU1622‐240 or vehicle. Figure 4D and E summarizes the instantaneous firing frequency, RMP and input resistance during the application of EU1622‐240 or vehicle onto CA1 pyramidal cells. EU1622‐240 did not produce any detectable effects on these parameters in CA1 pyramidal cells. These data suggest that the larger effects on GluN2D‐containing receptors might underlie increased actions of EU1622‐240 on interneurons than on CA1 pyramidal cells, given that most hippocampal interneurons express GluN2D (Perszyk, DiRaddo et al. 2016; von Engelhardt, Bocklisch et al. 2015).

Next, we selected 3 µM EU1622‐240 as a concentration to use to explore the actions of EU1622‐240 on neuronal properties. Before each recording epoch, a series of current injections was applied to assess the firing frequency for CA1 stratum radiatum interneurons and pyramidal cells. Figure 5 shows representative responses to increasing current injections for both CA1 stratum radiatum interneurons (Fig. 5A and B) and CA1 pyramidal cells (Fig. 5C and D). In the presence of EU1622‐240, injection of hyperpolarizing current failed to suppress firing (Fig. 5A2), whereas injection of depolarizing current produced more depolarization and spike firing in CA1 interneurons (Fig. 5B2) than in CA1 pyramidal cells (Fig. 5C2 and D2). Figure 5E–H plots the change in membrane potential for current injection for interneurons and CA1 pyramidal cells. These data suggest that EU1622‐240 decreases input resistance for CA1 interneurons [Fig. 5E; P = 0.0209, F(1, 20) = 6.288, two‐way ANOVA with Bonferroni's multiple comparisons test], given that the current injections produce less depolarization, a result consistent with an EU1622‐240‐induced increase in the open probability of NMDARs that contain the GluN2D subunit. There is no clear effect of EU1622‐240 on this same relationship determined for CA1 pyramidal cells [Fig. 5G, P = 0.092, F(1, 23) = 3.089, two‐way ANOVA with Bonferroni's multiple comparisons test]. This result is consistent with that described in the recordings shown in Fig. 4.

There is also an increase in action potential firing in response to depolarizing current injection for CA1 stratum radiatum interneurons [Fig. 5F; P = 0.0248, F(1, 10) = 6.963, two‐way ANOVA with Bonferroni's multiple comparisons test]. In CA1 pyramidal cells, there is a significant increase in depolarization‐induced firing frequency after EU1622‐240 treatment when compared with baseline [Fig. 5H; P = 0.028, F(1, 8) = 7.177, two‐way ANOVA with Bonferroni's multiple comparisons test]. These data show an increased response to depolarizing current injection, indicating that EU1622‐240 can increase the intrinsic excitability of both CA1 interneurons and pyramidal cells.

To confirm that the expression of GluN2D‐containing NMDARs is involved in the differential effects of EU1622‐240 on neuronal activity of CA1 interneurons and pyramidal cells, we repeated current‐clamp recordings with the same protocol on brain slices from Grin2D knockout (GluN2D‐KO) mice. Figure 6A shows a representative current‐clamp recording, with baseline, 10 min of 3 µM EU1622‐240 wash‐in, and co‐administration of 3 µM MK‐801 with 3 µM EU1622‐240. At 3 µM, EU1622‐240 did not alter input resistance, RMP and depolarization‐evoked firing frequency in CA1 interneurons from GluN2D‐KO mice (Table 3; Fig. 6B–F).

Besides GluN2D, EU1622‐240 also potentiates GluN2B, which is expressed on both CA1 stratum radiatum interneurons and pyramidal cells (Booker, Sumera et al. 2021; Hansen, Ogden et al. 2014). We hypothesized that there should be a change in the spontaneous excitatory signalling onto CA1 pyramidal cells via the effect of EU1622‐240 on NMDARs. To evaluate whether EU1622‐240 potentiated NMDAR function in response to spontaneous synaptic glutamate release, we imaged dissociated hippocampal neurons transfected with the genetically encoded Ca²⁺ indicator GCaMP6s (Chen, Wardill et al. 2013) in Mg²⁺‐free aCSF with TTX to block evoked neurotransmitter release (Fig. 7A). In these conditions, brief Ca²⁺ transients corresponding to NMDAR activation by spontaneous vesicle release events (quantal Ca²⁺ transients) can be visualized at individual dendritic spines (Reese & Kavalali 2016; Sinnen, Bowen et al. 2016) (Fig. 7A). Supporting Video S1 demonstrates the Ca²⁺ imaging of one neuron before EU1622‐240 treatment (left panel, pre‐EU1622‐240), with the Ca²⁺ imaging of the same neuron following administration of 3 µM EU1622‐240 (right panel, post‐EU1622‐240). This assay allows us to compare NMDAR function directly at the same synapses before and after EU1622‐240 application (Fig. 7B). As expected, we observed a robust increase in quantal Ca²⁺ transient amplitude following application of 3 µM EU1622‐240, with a similar increase in the integrated Ca²⁺ signal, which reflects quantal Ca²⁺ transient amplitude, kinetics and frequency over the duration of the imaging session (Fig. 7C–F). Similar results were obtained for 9 µM EU1622‐240, whereas 1 µM did not appear to potentiate NMDAR Ca²⁺ entry significantly (Fig. 7G). Blockade of NMDARs by d,l‐APV and MK‐801 confirms that the effect of EU1622‐240 on quantal Ca²⁺ transients was NMDAR dependent (Fig. 7B and C; Supporting Video S2).

To investigate whether there is an increase in inhibitory tone and whether the increased inhibitory signal might affect intrinsic membrane properties, we repeated the current‐clamp experiment in CA1 pyramidal cells in the presence of bicuculline to block GABA_A receptor‐mediated signalling. We detected both depolarization of RMP and increased spontaneous firing in CA1 pyramidal cells with co‐administration of bicuculline and EU1622‐240 (Fig. 8A–D). The bicuculline treatment alone did not affect the basal firing frequency of CA1 pyramidal cells even with a moderate but significantly depolarized RMP (Fig. 8E). Co‐application of EU1622‐240 with bicuculline produced a clear depolarization of CA1 pyramidal cells, which led to an increased firing frequency (see Table 4). One potential interpretation is that the increased tonic inhibition produced by enhanced NMDAR drive onto interneurons masks a depolarizing effect of EU1622‐240 on CA1 pyramidal cells, which occurs via potentiation of GluN2B‐containing NMDARs on pyramidal cells.

The increased excitability and increased NMDAR‐mediated synaptic drive onto CA1 interneurons should increase the frequency of IPSCs in CA1 pyramidal cells. To evaluate this hypothesis, sIPSC recordings were performed on CA1 pyramidal cells at a holding potential of +10 mV under voltage clamp. Figure 9A1 shows representative traces before (upper trace) and during application of 3 µM EU1622‐240 (lower trace). Figure 9A2 shows that EU1622‐240 shifted the sIPSC inter‐event interval significantly to the left in comparison to baseline (P < 0.0001, two‐sample Kolmogorov–Smirnov test).

EU1622‐240 significantly increased IPSC frequency to 1.8 ± 0.56‐fold of control (n = 8; Fig. 9B and C; Student's unpaired t test, P = 0.000169 < 0.025). In contrast, there was no effect of EU1622‐240 on IPSC frequency when d,l‐APV was present to block NMDARs (treatment/control was 0.76 ± 0.48‐fold, n = 14, P = 0.34). There was no detectable difference in sIPSC amplitude in EU1622‐240 in the absence (0.91 ± 0.21‐fold of baseline, Student's paired t test, P = 0.086) or in the presence of d,l‐APV (1.0 ± 0.22‐fold of baseline, P = 0.33). When parallel experiments were performed with 400 µM d,l‐APV present in the aCSF, d,l‐APV itself did not detectably alter mean sIPSC frequency (control 8.3 ± 4.1 Hz, d,l‐APV 9.0 ± 6.0 Hz; Student's paired t test, P = 0.307; Fig. 9B). The un‐normalized sIPSC amplitude and frequency are shown in Fig. 9D and E. These data suggest that tonically active NMDARs do not impact spontaneous IPSC frequency in resting conditions. However, EU1622‐240 can increase spontaneous IPSC frequency in an NMDAR‐dependent manner, consistent with EU1622‐240 enhancement of NMDAR function on interneurons.

If EU1622‐240 selectively enhances inhibitory signalling onto CA1 pyramidal cells, we should be able to detect a shift in the excitatory–inhibitory balance in CA1 pyramidal cells. To test this idea, we recorded EPSPs and IPSPs from CA1 pyramidal cells in response to Schaffer collateral afferent stimulation. This should produce a compound postsynaptic potential that consists of a glutamate‐mediated EPSP followed by a GABA‐mediated IPSP. We used the amplitude of this evoked EPSP and IPSP to assess the excitatory–inhibitory balance in CA1 pyramidal cells before and during 3 µM EU1622‐240 treatment. Two types of stimulation positions were used: stimulating at the CA3 region (distal stimulus) and at the CA1 region (local stimulus). Figure 10A shows representative recordings from the distal stimulus with baseline recording before EU1622‐240 application (black), and after 10 min of EU1622‐240 treatment. We observed a significant reduction of evoked EPSP amplitude (baseline 2.1 ± 1.3 mV, EU1622‐240 1.3 ± 0.88 mV, P = 0.0044 < 0.025), an increase in the evoked IPSP amplitude (baseline 0.93 ± 0.71 mV, EU1622‐240 2.4 ± 1.8 mV, P = 0.0037), hence a significantly decreased EPSP/IPSP ratio (baseline 3.7 ± 3.4‐fold, EU1622‐240 0.77 ± 0.91‐fold, P = 0.0075 < 0.0125, n = 10, Student's paired t test with Bonferroni correction; Fig. 10B–D). Figure 10E shows representative recordings from a local stimulus with baseline recording before EU1622‐240 application (black) and after 10 min of EU1622‐240 treatment. Interestingly, in contrast to distal stimulation, we observed a significant increase of evoked EPSP amplitude (baseline 5.29 ± 1.99 mV, EU1622‐240 7.45 ± 2.15 mV, P = 0.00041), an increase in the evoked IPSP (baseline 2.0 ± 0.41 mV, EU1622‐240 3.0 ± 0.53 mV, P = 0.0038 < 0.025), with no significant change of EPSP/IPSP ratio for local stimulation (baseline 2.8 ± 1.4‐fold, EU1622‐240 2.5 ± 0.53‐fold, P = 0.410, n = 8, Student's paired t test with Bonferroni correction; Fig. 10F–H). These data are consistent with enhanced recruitment of feedforward inhibition for distal stimulation, whereas distal excitatory drive is diminished in comparison to proximal stimulation in acutely prepared slices (Dingledine, Roth et al. 1987).

We also investigated the effect of 3 µM EU1622‐240 on the local evoked field potentials in mouse hippocampal slices. Given the ability of EU1622‐240 to enhance NMDAR function, we tested whether this modulator could influence cellular models of plasticity, such as LTP. We tested EU1622‐240 on brain slices produced from WT mice. We measured the slope of the evoked fEPSP recorded in the CA1 stratum radiatum to determine whether EU1622‐240‐induced potentiation of NMDAR can alter the fEPSP. Figure 11 shows the effect of EU1622‐240 on the slope of the fEPSP produced in the CA1 region with electric stimulation of Schaffer collateral fibres. We recorded evoked fEPSPs from slices treated with either vehicle (0.3% DMSO, labelled as baseline) or 3 µM EU1622‐240 applied for 60 min. Figure 11A shows the slope of the field response with 60 min baseline (last 10 min shown, black trace, BL: aCSF with 0.3% DMSO) followed by 60 min of 3 µM EU1622‐240 application (blue trace, EU240: EU1622‐240). Figure 11B shows full field potential response expanded traces for 10 min baseline (black trace, aCSF with 0.3% DMSO) followed by 20 and 30 min of 3 µM EU1622‐240 application (coloured traces). The baseline fEPSP slope was 1.3 ± 1.3 mV/ms. Following the wash‐in of 3 µM EU1622‐240, the fEPSP slope was (in millivolts per millisecond): 1.5 ± 1.4 (10 min), 1.9 ± 1.8 (20 min), 2.4 ± 2.1 (30 min), 2.6 ± 2.4 (40 min), 2.7 ± 2.4 (50 min), 2.7 ± 2.7 (60 min) and 3.7 ± 3.0 (65 min) after EU1622‐240 administration. At 3 µM, EU1622‐240 significantly increased the fold‐change of field amplitude (repeated‐measures one‐way ANOVA with Dunnett's multiple comparison test [F(1.178, 8.247) = 6.653, P = 0.0285; Fig. 11C]. Representative currents and data points were collected by averaging the last 2 min of each treatment period.

Given the ability of EU1622‐240 to produce a long‐lasting increase in fEPSP slope, we predicted that it would occlude LTP if it engaged similar mechanisms to LTP. We therefore measured the effect of 3 µM EU1622‐240 on LTP induced by four sets of theta‐burst stimulation. We observed that 3 µM 1622‐240 disrupted potentiation of the slope of fEPSP in response to theta‐burst stimulation, when compared with the fEPSP slope on the fourth theta burst at 100 min administration (Fig. 11D and E; P = 0.0992, Student's unpaired t test).

This project focused on the effects of an NMDAR positive allosteric modulator, EU1622‐240, on hippocampal neurons within the postnatal day 17–22 developmental window because this time period represents a foundational period in circuit formation that is essential for lifelong learning, memory and cognitive function. These data will thus help us to understand the role of EU1622‐240 potentially in targeting both early‐onset neuropathological conditions (schizophrenia and epilepsy). Although these results might also be relevant for later‐onset disorders that also involve perturbation of NMDAR function (Alzheimer's disease and age‐related cognitive decline), additional studies will be required to evaluate potential actions of this class of NMDAR PAMs on conditions involving older ages.

The novel class of NMDAR PAM exemplified by the tool compounds EU1622‐240 shows a preferential potentiating effect onto GluN2D‐containing CA1 interneurons over GluN2A/GluN2B‐expressing CA1 pyramidal cells, which could have useful actions in modulating circuit function in a variety of pathological conditions. There were some actions at excitatory synapses, however, and EU1622‐240 had an ability to induce a form of enhanced potentiation that occluded LTP. This suggests that this series of modulators might have therapeutically relevant effects on NMDARs that mitigate clinical problems, including anxiety, treatment‐resistant depression or cognitive dysfunction (Hanson, Yuan et al. 2024). Furthermore, enhancing GABA signalling can effectively reduce seizure‐like activity in the mouse hippocampus (Liddiard, Suryavanshi et al. 2024). The dysfunction in the excitation–inhibition balance is apparent in rodent models of schizophrenia (Gawande, KK et al. 2023; Lisman, Coyle et al. 2008), with a reduced inhibitory signalling level. Thus, PAMs such as EU1622‐240 might be effective, via their enhancement of inhibitory tone, for disorders involving diminished circuit inhibition.