Endogenous 24S-hydroxycholesterol modulates NMDAR-mediated function in hippocampal slices

Slice preparation.

Hippocampal slices were prepared from postnatal day 18 (P18) to P32 wild-type (WT) or CYP46A1^−/− (KO) mice. WT C57Bl/6J or KO mice on a similar genetic background (Jackson Laboratories) were used in all studies. Slices for all studies were cohort and age matched to the extent possible. In accordance with protocols approved by the Washington University Animal Studies Committee, mice were anesthetized with isoflurane and decapitated. The brain was removed and glued onto a Leica VT1200 specimen holder. Sagittal (300 μm) slices were cut in ice-cold, modified artificial cerebrospinal fluid (ACSF) (in mM: 87 NaCl, 75 sucrose, 25 glucose, 25 NaHCO₃, 2.5 KCl, 1.25 NaH₂PO₄, equilibrated with 95% oxygen-5% CO₂ plus 0.5 CaCl₂, 3 MgCl₂; 320 mosmol). Slices were then incubated at 32–34°C for 30 min in choline-based ACSF (in mM: 92 choline chloride, 25 glucose, 30 NaHCO₃, 2.5 KCl, 1.2 NaH₂PO₄, 20 HEPES, 2 thiourea, 5 Na ascorbate, and 3 Na pyruvate, 2 CaCl₂ and 1 MgCl₂, equilibrated with 95% oxygen-5% CO₂; 300 mosmol), and subsequently stored at room temperature in regular ACSF (in mM: 125 NaCl, 25 glucose, 25 NaHCO₃, 2.5 KCl, 1.25 NaH₂PO₄, equilibrated with 95% oxygen-5% CO₂ plus 2.6 CaCl₂, 1.2 MgCl₂; 310 mosmol), allowing for at least 1 h recovery prior to experiments. Except for noted exceptions, drugs were obtained from Sigma (St. Louis, MO).

Analysis of endogenous 24S-HC level in mouse hippocampal tissues and slices.

Three mouse brains from each group of WT or CYP46A1 KO animals were harvested. Each brain was bisected into two hemispheres. Hippocampi from one of the hemispheres were frozen at −70°C immediately. The other hemisphere from the same brain was cut into 300-μm hippocampal slices and stored in ACSF at room temperature for 5–6 h before being frozen at −70°C. For measurement of 24S-HC level, the frozen tissue and slice samples were homogenized with 300 μl of PBS buffer. 24S-HC in 25 μl (50 μl of the sliced hippocampus tissue homogenate) of the whole hippocampus tissue homogenate was extracted with 200 μl of methanol. Deuterated 24S-HC (24S-HC d₇: 100 ng) was used as an internal standard and was added to the samples before extraction. The four-point calibration standards of 24S-HC, containing 24S-HC d₇, were also prepared for the absolute quantification. 24S-HC and 24S-HC d₇ were derivatized to N, N-dimethylglycinate (DMG) to increase the mass spectrometry (MS) sensitivity. Sample analysis was performed with a Shimadzu 20AD HPLC system, a LeapPAL autosampler coupled to a triple quadrupole mass spectrometer (API 4000) operated in multiple reaction monitoring (MRM) mode. The positive ion electrospray ionization (ESI) mode was used for detection of 24S-HC and deuterated 24S-HC d₇. The study samples were injected in duplicate for data averaging. Data processing was conducted with Analyst 1.5.1 (Applied Biosystems). Bicinchoninic acid (BCA) assay was performed on all the tissue samples for protein determination. The values reported for the hippocampal samples were normalized to protein content.

Whole cell patch-clamp recording.

Slices were transferred to a recording chamber and continuously perfused with oxygenated ACSF at 2 ml/min. Experiments examining tonic NMDAR activation upon glutamate transporter inhibition and LTP/OGD studies were performed at 30–32°C. Other recordings were performed at 23–25°C. For measuring dual-component EPSCs, picrotoxin (PTX, 100 μM) and d-serine (10 μM) were included in extracellular solutions to inhibit GABA-mediated neurotransmission and to enhance NMDAR-mediated currents. In some experiments, 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide (NBQX; Tocris, Bristol, UK) at 10 μM was included in extracellular solutions to isolate NMDAR EPSCs. For isolating AMPAR EPSCs or GABAR inhibitory postsynaptic currents (IPSCs), 100 μM PTX and 50 μM d-(−)-amino-5-phosphonopentanoic acid (d-APV; Tocris) or 10 μM NBQX and 50 μM d-APV were included in extracellular solutions, respectively. For measuring tonic NMDAR current, 100 μM PTX, 10 μM NBQX, and 10 μM d-serine were included in extracellular solutions. dl-threo-β-Benzyloxyaspartic acid (TBOA) (50 μM, Tocris) was bath-applied to evoke tonic NMDAR current. (+)-5-Methyl-10,11-dihydro-5H-dibenzo[a, d]cyclohepten-5,10-imine (MK-801; 40 μM, Tocris) was bath-applied to examine the time constant of NMDAR channel block. For measuring neuronal intrinsic excitability, experiments were performed in the presence of 100 μM PTX, 10 μM NBQX, and 50 μM d-APV to block synaptic transmission.

Somatic, whole cell patch-clamp recordings were performed using standard differential contrast interference microscopy under infrared illumination. In the experiments where PSCs or PSPs were evoked, a small cut was made to separate CA1 from CA3 area to prevent recurrent synaptic activity. CA1 pyramidal cells were identified on an upright Nikon Eclipse E600FN microscope and a QImaging camera controlled with QCapture (QImaging, Surrey, Canada). Somatic whole cell recordings were made with borosilicate patch pipettes (World Precision Instruments, Sarasota, FL; Sutter Instruments, Novato, CA), having open tip resistance of 3–7 MΩ. After a whole cell configuration was established, cells were allowed to fill with the intracellular solution for ∼5 min. Intracellular EPSCs or EPSPs were evoked by a stimulating electrode placed in striatum radiatum, ∼50 μm away from CA1 cell body layer and 50–100 μm away from the recorded cell. Stimulation was applied at 0.033–0.05 Hz, with 0.1-ms pulse width. Recordings were obtained using a MultiClamp 700B amplifier (Molecular Devices; Sunnyvale, CA) and pClamp 10.4 software (Molecular Devices).

For voltage-clamp experiments, the intracellular pipette solution contained (in mM) 120 cesium methanesulfonate, 20 HEPES, 10 EGTA, 2 MgATP, 0.3 Na₂GTP, and 5 QX-314; pH was adjusted with CsOH to pH 7.25 (290 mosmol). Cells were voltage-clamped at −30 mV, −70 mV, and 0 mV for evoked NMDAR EPSCs (including dual-component EPSCs), isolated AMPAR EPSCs, and GABA IPSCs, respectively. For measuring sEPSCs and sIPSCs, cells were voltage-clamped at −70 mV and 0 mV, respectively, and data were acquired in gap-free mode. For measuring tonic NMDAR current, cells were voltage-clamped at −30 mV.

For current-clamp experiments, the intracellular pipette solution contained (in mM) 140 K-gluconate, 4 MgCl₂, 10 HEPES, 0.4 EGTA, 4 MgATP, 0.3 Na₂GTP, and 10 phosphocreatine; pH was adjusted with KOH to pH 7.25 (290 mosmol). For measuring intrinsic excitability, CA1 neurons were initially maintained at −74 mV with constant current injection when necessary, and 500-ms current steps with 100 pA increment intervals were injected somatically, with a range of 0–800 pA. Current threshold was defined as the minimal injected current needed for firing a single spike. For measuring dual-component EPSPs and the resultant firing, a repetitive stimulation protocol containing 10 pulses at 50 Hz was applied to promote dual-component EPSPs. Neurons were initially maintained at −65 mV with constant current injection before application of stimulation. Stimulation intensity was adjusted to evoke 5–6 spikes (half of the pulse number) for each cell. Somatic bridge balance and pipette capacitance were adjusted using MultiClamp 700B Commander software.

Data were acquired at 100 kHz, filtered at 4 kHz using an 8-pole Bessel filter, and digitized using a DigiData1550 16-bit A/D converter (Molecular Devices). Somatic access resistance was monitored continuously, and cells with unstable access resistance (>20% change) were excluded from analysis. SGE-301 obtained from SAGE Therapeutics (Cambridge, MA) was dissolved at 1.25 mM in 100% DMSO to make concentrated stock solutions.

Field EPSPs (fEPSPs), long-term potentiation (LTP), and oxygen-glucose deprivation (OGD) recordings.

The hippocampus was dissected from postnatal day 28 (P28)–P32 WT or KO mice under isoflurane anesthesia. Slices (400 μm) were cut from the septal hippocampus with a rotary slicer in ACSF containing (in mM) 124 NaCl, 5 KCl, 2 MgSO₄, 2 CaCl₂, 1.25 NaH₂PO₄, 22 NaHCO₃, 10 glucose, bubbled with 95% oxygen- 5% CO₂ at 4–6°C (296 mosmol). After recovering for at least 1 h at 30°C, hippocampal slices were transferred to a submerged recording chamber with continuous bath perfusion of ACSF at 2 ml/min at 30°C. Extracellular recordings were obtained from the apical dendritic layer of the CA1 region elicited with 0.1 ms constant-current pulses through a bipolar stimulating electrode placed in stratum radiatum. During an experiment, excitatory postsynaptic potentials (EPSPs) were monitored using a half-maximal stimulus based on a baseline input-output curve. For LTP studies, tetanic stimulation (100 Hz for 1 s) was delivered to a bipolar extracellular stimulating electrode placed in the Schaffer collateral pathway. fEPSPs were monitored for 60 min subsequent to the tetanus. As a measure of LTP, the slope of the fEPSP 50–60 min following tetanic stimulation was compared with the fEPSP slope for the 10 min immediately preceding tetanic stimulation. For OGD experiments, ASCF was replaced with identical fluid without glucose and gassed with 95% N₂-5% CO₂ for 20 min. An input-output curve was repeated 90 min following OGD for statistical comparisons of changes in EPSP slopes at half-maximal intensity. In some experiments, KO slices were treated with 1 μM bath SGE-301 for 20 min before OGD and during the 20 min OGD challenge. SGE-301 was then washed out after OGD challenge.

Data analysis.

Data analysis was performed in Clampfit 10.4 (Molecular Devices). For measurement of NMDAR/AMPAR, NMDAR current averaged from 10 traces was estimated as mean value at 50–55 ms following peak of dual-component EPSC (Hestrin et al. 1990), while AMPAR current averaged from 10 traces was calculated as EPSC peak amplitude following 5 min d-APV (50 μM) treatment. For measuring TBOA-induced tonic NMDAR current, the mean value of holding current 12 s immediately preceding TBOA application was calculated as the baseline, and the mean current at the end of a 6 min TBOA application was subtracted from the baseline. For estimating MK-801-induced decay of tonic NMDAR current, the decay tau of current was calculated by fitting the current trace to a single-exponential decay function. For calculating SGE-301 effects on synaptic currents, baseline EPSC or IPSC peak amplitude was determined from averaging 20 traces before SGE-301 bath application. After 6 min SGE-301 wash-in, EPSC or IPSC peak amplitude was averaged from the following 20 traces, and level of potentiation was calculated by normalizing this value to baseline. For comparing intrinsic excitability, spike firing threshold was determined as the minimal intensity of injected current needed for eliciting the first spike. Frequency of sEPSCs and sIPSCs was calculated using Mini analysis software (Synaptosoft; Fort Lee, NJ). For determining levels of OGD-induced synaptic depression, slope of fEPSP calculated at 90 min following OGD was normalized to fEPSP slope at the 10 min baseline.

Student's independent, two-tailed t-test was performed to determine statistical significance between the means of two groups. For data normalized to baseline (percentage changes) and for data sets that failed to pass the Shapiro-Wilk normality test, two-tailed nonparametric Wilcoxon tests were performed. Mann-Whitney tests were used for nonpaired comparison of drug effects between two groups, and a Kruskal-Wallace test with Dunn's post hoc analysis were used to compare nonparametric data with >2 groups (see text and figure legends). For detection of drug and genotype effects within an experiment, a two-way ANOVA and Bonferroni post hoc analysis were performed. Significance is described at the level of P ≤ 0.05, P ≤ 0.01, and P ≤ 0.001. Values are presented as means ± standard error of the mean (SE).

Ex vivo slices retain similar 24S-HC levels as fresh tissue.

The level of endogenous 24S-HC is dramatically reduced in mice genetically deficient in CYP46A1 (Russell et al. 2009). To confirm the reduction of endogenous 24S-HC in the CYP46A1 KO mice used in our study, we analyzed and compared 24S-HC level in WT and CYP46A1 KO mouse hippocampal slices. Data from LC-MS analysis showed that average 24S-HC level in freshly isolated hippocampus is significantly lower in KO (1.8 ± 0.2 ng/mg protein; N = 3) than in WT (385.0 ± 11.6 ng/mg protein; N = 3; Fig. 1A, P < 0.001), suggesting that, as expected (Lund et al. 2003; Meljon et al. 2014; Russell et al. 2009), endogenous 24S-HC level is greatly reduced in the KO mice. The estimated concentration of endogenous 24S-HC level in WT hippocampus was in the tens of micromolar, sufficient to modulate NMDAR function, assuming homogeneous distribution of the oxysterol (Paul et al. 2013). The results showed a slightly lower level of 24S-HC in WT slices than in intact WT tissue; however, 24S-HC was still much higher in WT slices (326.0 ± 16.2 ng/mg protein; N = 3) than in KO slices (2.0 ± 0.5 ng/mg protein; N = 3; Fig. 1A, P < 0.001), suggesting retention of high 24S-HC levels despite continuous perfusion of slices (Fig. 1A). This is consistent with strong retention of 24S-HC in cells (Paul et al. 2013) but could also imply ongoing synthesis. The observation that slices have nearly the same 24S-HC concentration as fresh tissue implies that slices are an appropriate model for exploring the potential role of endogenous 24S-HC in regulating NMDAR function.

NMDAR EPSCs, tonic current, and NMDAR-induced spiking are reduced in CYP46A1 KO slices.

24S-HC and its analogs potentiate NMDAR activity when applied exogenously (Paul et al. 2013). However, the role of endogenous 24S-HC in modulating NMDAR activity has not been established. To address this, we first examined effects of endogenous 24S-HC on NMDAR tone, measured in evoked, dual-component EPSCs. CYP46A1 KO slices exhibited a significantly smaller NMDAR-to-AMPAR EPSC ratio than WT slices (Fig. 1, B and C), suggesting that reducing endogenous 24S-HC in KO slices dampens NMDAR function. The results further suggest that basal NMDAR contribution to EPSCs reflects not only glutamate and glycine coagonist presence, but also the presence of the positive allosteric modulator, 24S-HC.

Although the effect of CYP46A1 KO on NMDAR EPSCs was significant, it was not a strong effect. To confirm that endogenous 24S-HC modulates the basal NMDAR current tone, we examined whether tonic NMDAR current induced by ambient glutamate was also reduced in CYP46A1 KO slices. We reasoned that 24S-HC's modulatory effects may be more evident at the lower, ambient glutamate concentrations than that at saturating, synaptic concentrations (Linsenbardt et al. 2014). Consistent with this idea, addition of the glutamate transporter blocker TBOA for 6 min, in the presence of blockers of GABARs and AMPARs, induced larger currents in WT neurons than in KO neurons (Fig. 2, A–C). The contribution of NMDARs to these currents was verified with MK-801, an activation-dependent blocker of NMDARs.

MK-801 blocking rate is proportional to channel open probability (Huettner and Bean 1988; Rosenmund et al. 1995). Consistent with our interpretation that larger TBOA-induced currents in WT slices represent channel openings with higher open probability, we found that MK-801-induced blocking rate was faster in WT slices than in KO slices (Fig. 2D).

Although NMDARs play a limited role in basal synaptic transmission at voltages near the resting membrane potential (approximately −60 to −70 mV), the NMDAR contribution is amplified when neurons receive repetitive excitatory synaptic drive, resulting in persistent depolarization, relief from physiological Mg²⁺ block of the NMDAR channel, and increased participation of NMDARs in spike generation (reviewed in Hunt and Castillo 2012). To address whether loss of CYP46A1 and the associated reduced NMDAR contribution to EPSCs (Fig. 1) affect the role of NMDARs in driving spike firing, we current-clamped WT or KO neurons and adopted a repetitive synaptic stimulation protocol to elicit NMDAR dependent spiking. Trains of 10 pulses at 50 Hz were delivered to promote dual-component EPSPs with stimulation intensity adjusted to evoke 5–6 spikes (approximately half of the pulse number) for each cell at baseline (Fig. 3, A–C). Input/output curves verified that WT and KO cells produced similar maximum firing rates with saturated levels of synaptic stimulation (8.7 ± 0.3 for WT, 8.7 ± 0.4 for KO, Fig. 3D). Therefore, the criterion of 5–6 spikes at baseline represented ∼60–70% of maximum for both groups. After baseline recording of 10 sweeps, 50 μM d-APV was washed-in to block NMDAR-mediated EPSPs (Fig. 3, A, B, and E), and the same stimulus protocol and intensity used at baseline was applied again to evoke spiking. d-APV more strongly reduced spiking in WT cells than in KO cells (Fig. 3, A, B, and E), suggesting more NMDAR contribution to EPSP-spike generation in WT animals. These results are consistent with the reduced NMDAR synaptic tone in KO animals (Figs. 1 and 2).

An exogenous analog demonstrates lack of saturation and specificity.

24S-HC appears to potentiate NMDAR activity at a unique binding site specific for oxysterols (Paul et al. 2013). To address whether the endogenous level of 24S-HC saturates this NMDAR binding site, the synthetic 24S-HC analog SGE-301 (10 μM) was bath-applied to slices after a stable 10 min baseline recording, and its effect on NMDAR function was examined. When overall drug effect across genotypes was evaluated, SGE-301 significantly potentiated NMDAR EPSCs, but there was no interaction between genotype and drug condition (Fig. 4, A and B). This suggests that SGE-301 potentiation was statistically indistinguishable in KO vs. WT slices and that the NMDAR allosteric site is therefore not saturated by endogenous 24S-HC despite estimated high levels of the oxysterol (Fig. 1A).

In hippocampal cultures, potentiation by 24S-HC and its analogs is selective for NMDARs, given that no potentiation of evoked AMPAR EPSCs or GABAR IPSCs was observed when the compounds were exogenously applied to WT neurons (Paul et al. 2013). However, selectivity has not been examined in situ or in the CYP46A1 null background, where saturation of putative sites by endogenous 24S-HC is not a concern. To address this, we studied effects of SGE-301 on AMPAR EPSCs and GABAR IPSCs in CYP46A1 KO slices, where potential effects of endogenous 24S-HC can be excluded. We found that, while significantly potentiating NMDAR EPSCs relative to baseline (Fig. 4, A and E), SGE-301 failed to potentiate AMPAR EPSCs (Fig. 4, C and E; Wilcoxon test; P > 0.2).

SGE-301 effects on GABAR IPSCs were more complex (Fig. 4, D and E). Overall, there was no significant change in IPSCs (Wilcoxon test, P > 0.05, N = 11). When one outlying cell exhibiting potentiation during SGE-301 presentation was excluded from analysis, we found a small but significant decrease (29.7 ± 4.4%, N = 10, Wilcoxon test, P < 0.01) in IPSC amplitude following 6 min application of SGE-301. However, analysis of the time course of IPSC depression in these cells revealed that IPSC depression was not clearly linked to SGE-301 application. Instead, IPSCs typically showed a gradual, time-dependent decrease, and this decrease did not differ from that observed in 7 KO cells treated with vehicle (DMSO) only. These IPSCs exhibited a 17.4 ± 8.7% depression over 6 min, not statistically different from the depression exhibited by SGE-301 (Mann-Whitney test, P > 0.2; data not shown). Thus we conclude that IPSCs exhibit time-dependent rundown but, like AMPAR EPSCs, show no evidence for SGE-301-related potentiation or inhibition.

Reducing endogenous 24S-HC does not affect neuronal intrinsic excitability or spontaneous transmission.

Increased neuronal intrinsic excitability has been reported in several GluN1 KO mouse models (Pagadala et al. 2013; Tatard-Leitman et al. 2015). This effect is apparently a homeostatic response to loss of the obligatory GluN1 subunit, without which NMDAR currents are eliminated. Other studies using sensory deprivation or pharmacological NMDAR block have also observed compensatory changes in intrinsic excitability and/or synaptic transmission (Fan et al. 2005; Turrigiano et al. 1994; Turrigiano and Nelson 2004; van Welie et al. 2004; Zhang and Linden 2003). It remains unclear whether the more subtle manipulation of altering NMDAR positive allosteric modulation also elicits homeostatic compensation from intrinsic conductances. To test this, we compared WT and KO responses to 500-ms current steps, ranging from 0 to 800 pA (Fig. 5, A and B). Neurons were current clamped to −74 mV prior to steps using a small bias current when needed. We found no significant difference in any of the parameters related to intrinsic excitability, including firing rates vs. injected current intensity, threshold of action potential firing, resting membrane potential with no injected bias current, or membrane resistance (Fig. 5, C–F). We conclude that loss of oxysterol positive allosteric modulation is not sufficiently severe to invoke homeostatic changes in excitability.

Increased spontaneous basal excitatory synaptic transmission has been reported in GluN1 KO mouse models (Pagadala et al. 2013; Tatard-Leitman et al. 2015), possibly arising as a result of altered intrinsic excitability or representing a separate homeostatic effect on transmission. To elucidate whether changing 24S-HC levels alters basal spontaneous synaptic transmission, we measured frequency of spontaneous EPSCs (sEPSCs) and IPSCs (sIPSCs) in individual neurons, and we compared the frequency ratio of sEPSCs to sIPSCs (E/I ratio) in WT vs. KO slices. sEPSCs and sIPSCs were recorded at holding potentials of −70 mV and 0 mV, respectively (Fig. 6, A and B). We observed no difference of sEPSC and sIPSC frequency and E/I ratio between WT and CYP46A1 KO neurons (Fig. 6C). Furthermore, no significant change of sEPSC amplitude was observed between WT and KO neurons (data not shown). Taken together, we find that loss of positive allosteric modulation is not sufficiently severe to invoke compensatory changes in intrinsic conductances or in spontaneous synaptic transmission.

Reducing endogenous 24S-HC does not affect LTP induction.

Impaired NMDAR-dependent LTP in CYP46A1 KO mice has been demonstrated previously in 5- to 8-wk-old mice (Kotti et al. 2006; Kotti et al. 2008; Russell et al. 2009). We compared LTP in slices prepared from 30-day-old WT and KO animals. Field EPSPs (fEPSPs) were evoked at 50% maximal level to obtain a stable 10 min baseline recording. Slices were then stimulated at 100 Hz for 1 s to induce LTP, after which the fEPSP was monitored for an additional 60 min. Surprisingly, KO slices exhibited LTP indistinguishable from LTP in WT slices (Fig. 7A), suggesting that endogenous 24S-HC is not necessary for LTP that is induced by tetanic stimulation. Thus it appears that basal NMDAR function in 30-day-old KO slices is sufficient for normal LTP induction.

CYP46A1 KO slices exhibit higher resistance to OGD-induced synaptic depression.

Despite our results showing intact LTP in CYP46A1 KO slices, we found that tonic NMDAR currents, perhaps generated largely by extrasynaptic receptors, are strongly affected by 24S-HC deletion (Fig. 2). Such conditions may be met by pathophysiological overstimulation of NMDARs during ischemia in situ or by oxygen-glucose deprivation in vitro (Soria et al. 2014). Given that NMDAR overactivation participates in ischemia-mediated excitotoxicity and synaptic depression (Weilinger et al. 2013), and a form of NMDAR-dependent synaptic depression can be induced in response to oxygen deprivation (Izumi et al. 1998), we hypothesized that 24S-HC reduction may alter OGD-induced synaptic depression or its recovery. To test this hypothesis, we examined synaptic depression caused by 20 min OGD in hippocampal slices. Field EPSPs (fEPSPs) were evoked at 50% maximal level, and fEPSP slopes were compared before and after OGD in WT vs. KO slices. OGD resulted in >80% depression of fEPSPs from both WT and KO slices (30′). Ninety minutes following OGD wash-out (120′), recovery of fEPSPs was significantly greater in KO slices compared with WT slices (Fig. 7B), suggesting that reduction of 24S-HC protects neurotransmission from OGD-induced synaptic depression. To test if 24S-HC is likely responsible for the persistent synaptic depression in WT slices, we treated KO slices with 1 μM SGE-301 before and during OGD. Similar to WT slices, KO slices treated with SGE-301 exhibited fEPSP depression that failed to recover to the level observed from untreated KO slices after OGD wash-out (Fig. 7B). Thus acute application of SGE-301 on KO slices restores persistent OGD-induced synaptic depression observed in WT slices.