Two months old male C57BL/6 mice (20–25 g) were used in the study. Animals were purchased from the institute of zoology, Chinese academy of sciences. Animals were housed in cages with wood shaving bedding and were maintained on a 12:12 light:dark cycle in the Tianjin Medical university of China-approved animal facility. Animals were given enough food and water throughout the studies unless otherwise noted. All experiments were approved by the Animal Care and Use Committee of Tianjin medical university, in compliance with National Institutes of Health guidelines.

Transient MCAO was adopted in our study, mainly according to the method of Rousselet et al. (2012). Briefly, 20–25 g male mice were anesthetized with isoflurane. During the whole surgery body temperature of the mice is maintained constant by a heating pad (Rousselet et al., 2012). A monofilament suture (about 9–10 mm is coated silicon, Doccol Corporation) was used to occlude the right MCA. Heart rate, rectal temperature, and blood gases were monitored during the MCAO surgery. All the mice were subjected a 1 h MCAO. For sham group, all procedures of operation were same except that the MCAO suture was not inserted into MCA. For drug treatment experiments, drugs were then performed via intraperitoneal injection 2 h after the onset of MCAO (GLYX-13, Tocris, 10 mg/kg, D-serine, Sigma, 50 mg/kg, Ro256981 Tocris 5 mg/kg, NVP-AAM077 Tocris 2.4 mg/kg).

Cerebral blood flow was measured by Laser Doppler Flowmetry (Perimed, PeriFlux5010, Sweden) with a straight laser Doppler probe (Perimed, Probe 418, Sweden) as previously described (Iwai et al., 2004). The probe was placed in the territory supplied by the MCA (AP-1.0 mm, L5.0 mm from bregma). Relative cerebral blood flow was measured before MCAO (baseline) and during occlusion. Changes of cerebral blood flow were expressed as percentage of the baseline value. Only those mice which ipsilateral regional cerebral blood flow ≤20% of baseline were used for further study.

Twenty-four hours after reperfusion, neurological score was assessed. Neurological score contributed to evaluate the success of transient MCAO and to estimate the degree of neurological deficits and injury severity. Neurological assessments were scored as previously described (Jiang et al., 2005). 0:

Normal.

Twenty-four hours after reperfusion, the animals were decapitated and the brains were removed into a mouse brain matrix for 1 mm section. Then sections were immersed in 1% triphenyltetrazolium chloride (TTC, Sigma) in PBS and incubated about 15 min at 37°C. The area of unstained brain was identified as the infarct area. The infarct area was measured using the image pro plus software 6.0. The infarct volume was reckoned by multiplying summed infarct area from each section by the slice thickness. Take the edema after ischemia into consideration, the infarct volume were expressed as percentage of hemisphere volume (Yang et al., 1998).

The animals were sacrificed 24 h after reperfusion. The brain was fixed using 4% paraformaldehyde and then was embedded in paraffin, and was cut into coronal sections of 5 μm thickness for histological evaluation.

Western blot was done as previously described (Zhao et al., 2013). Briefly, equal amounts of prepared sample protein (ipsilateral hippocampus n = 6) were separated by 10% SDS-PAGE and then were transferred to polyvinylidene fluoride (PVDF) membranes. The membranes were blocked by 5% BSA or skimmed milk for 1 h, then were incubated with primary antibodies (mouse anti-NMDAR2A 1:500, mouse anti-NMDAR2B 1:500 BD Transduction Laboratories; rabbit anti-NMDAR2A phospho Y1325 1:1000, rabbit Phospho-NMDAR2B Tyr1472 1:1000 Cell Signaling; β-actin 1:2000 Santa Cruz) at 4°C overnight. β-actin was used as a loading control. Next the membranes were incubated with anti-rabbit, anti-mouse secondary antibodies. To visualize the reaction, the enhanced chemiluminescence (ECL) detection reagents (Beyotime Biotechnology) were used. The intensity of blots was quantified with image J software.

The preparation of brain slices were performed as previously described (Shen et al., 2007; Yao et al., 2012). Briefly, 4 week old male C57BL/6 mice were sacrificed rapidly. The whole brain were removed and 320 μm thickness coronal slices were cut by a vibratome in ice, ACSF bubbled continuously with 95% O₂/5% CO_2. The brain slices were incubated in a chamber with above ACSF at 32°C for 1 h recovery.

Whole cell patch clamp recording were performed. The pipettes solution containing (in mmol/L): 130 CsMeSO₃, 8 NaCl, 10 HEPES, 5 QX314, 4 Mg-ATP, 0.3 Na-GTP, 0.2 EGTA, meanwhile PH was adjusted to 7.2 by CsOH and osmolarity was maintained to 275–290 mOsm (Massey et al., 2004). The pyramidal neurons of CA1 region were visualized with an upright microscope (BX51-WI, Olympus, Japan) and an infrared video camera (710 M, DVC, Pleasant Hill, CA, USA). Recording were conducted with a 2 kHz Bessel filter at a 10 kHz sampling frequency using Axon patch 200B patch-clamp amplifier (Molecular Devices, Foster City, CA, USA) and Digidata 1440 interface (Molecular Devices, Foster City, CA, USA). Series resistance was monitored during the experiments and cells that showed >20% change were removed from subsequent analysis. If the series resistance was greater than 30 MΩ, recordings were also discarded from analysis (Dennis et al., 2011). Synaptic responses were induced at 0.05 Hz by a tungsten bipolar electrode (FHC, USA) which was placed in the stratum radiatum (for Schaffer collateral stimulation); meanwhile the recording pipette was located in the pyramidal cell layer of CA1 (Dennis et al., 2011). Evoked Excitatory postsynaptic current (eEPSC) were recorded at −70 mV holding potential in ACSF perfusion containing bicuculline methiodide (10 μmol/L, sigma) to block GABA_A receptor-mediated inhibitory synaptic currents (Yao et al., 2012). NMDAR mediated EPSCs were recorded in Mg²⁺-free ACSF and clamped at +40 mV in the presence of NBQX (5 μmol/L) and bicuculline methiodide (10 μmol/L) to block AMPAR and GABA_AR respectively.

OGD was produced by perfusing slices with ACSF saturated with 95% N₂/5% CO₂ and replacing the 10 mmol/L glucose with 10 mmol/L sucrose (Dennis et al., 2011).

Results were presented as mean ± SEM. Mann-Whitney U-test was used in analysis of Neurological scores. One-way analysis of variance (ANOVA) with a post hoc analysis (Newman-Keuls) for multiple comparisons was applied to determine the statistical differences between three or more groups. Differences were considered significant when p < 0.05.

There were no significant difference in heart rate, the pH of blood, PO₂, PCO₂ between vehicle and GLYX-13 treated group during MCAO surgery (Table 1).

We evaluated neurological score 24 h after reperfusion. The neurological function were significantly improved in GLYX-13 treatment group in comparison with the vehicle group (Figure 1; GLYX-13 1.8 ± 0.2 vs. Vehicle 3.2 ± 0.2, n = 12, ***p < 0.001).

We explore the possible neuroprotective effects of GLYX-13 in vivo MCAO model. The cerebral infarct volume was evaluated by TTC staining 24 h after reperfusion. Normal brain tissue was stained red, while the infarction remained unstained (white). Compared with the vehicle group, the infarct volume exhibited remarkable decrease in the GLYX-13 group (Figure 2; GLYX-13 17 ± 3.4% vs. Vehicle 45.0 ± 3.1%, ^###p < 0.001).

HE staining was performed for histological evaluation within hippocampus (Figure 3). FJC staining was also performed to further assess the neuronal injury. FJC positive neurons were with strong green fluorescent signal. Notably, there was a dramatic decrease in the number of FJC positive neurons in GLYX-13 treated group in dentate gyrus (DG; GLYX-13 12.4 ± 4.2/10⁻²mm² vs. Vehicle 195 ± 14.2/10⁻²mm², n = 5, ^###p < 0.001) and CA3 region (GLYX-13 27.2 ± 7.6/10⁻²mm² vs. Vehicle 99.4 ± 9.7/10⁻² mm², n = 5, ^###p < 0.001), moreover, there was still a decrease in CA1 region in GLYX-13 treated group (GLYX-13 101.3 ± 7.4/10⁻² mm² vs. Vehicle 118.3 ± 8.3/10⁻² mm², n = 5, ^#p < 0.05; Figure 4).

NR2A and NR2B containing NMDARs are linked to different intracellular cascades and play the opposite roles in neuronal survival and death during ischemia, which is the core composition of subtype hypothesis of NMDARs. Thus activation of the synaptically localized NR2A-containing NMDARs promotes neuronal survival and exerts a neuroprotective potential. In contrast, activation of extra-synaptic NR2B containing NMDARs mainly result in excitotoxicity (Lai et al., 2011). It was reported that phosphorylated NR2A (Tyr1325) and phosphorylated NR2B (Tyr1472) were most relevant site to synaptic activity and excitotoxicity (Wu et al., 2013). Therefore, we focused on the expression of NR2A and NR2B and their phosphorylation at site Tyr1325 and Tyr1472 respectively after ischemia. Western blot analysis showed the level of p-NR2B (Tyr1472) in hippocampal tissue exhibited a marked increase after reperfusion 4 h (Vehicle 0.6488 ± 0.04057 vs. Sham 0.2221 ± 0.02572, **p < 0.01, n = 3 independent experiments), whereas the expression of p-NR2A (Tyr1325) reduced compared with sham group (Vehicle 0.4299 ± 0.02689 vs. Sham 0.6264 ± 0.02831, *p < 0.05, n = 3 independent experiments). No significant difference was found in NR2A and NR2B between vehicle and sham after reperfusion 4 h. Notably, in this study we found that GLYX-13 alleviated excitotoxicity through decreasing the tyrosine phosphorylation of NR2B subunit at site Tyr1472 after reperfusion 4 h (GLYX-13 0.3708 ± 0.06223 vs. vehicle 0.6488 ± 0.04057, ^#p < 0.05, n = 3 independent experiments), which site was reported closely related to neuronal death before Li et al. (2017). Meanwhile, GLYX-13 treated group had a greatly higher level of p-NR2A expression than vehicle (GLYX-13 0.6190 ± 0.02101 vs. vehicle 0.4299 ± 0.02689, ^#p < 0.05). Moreover the ratio of pY1472 NR2B/total NR2B and pY1325 NR2A/total NR2A were also restored by GLYX-13 treatment (Figures 5, 6). Furthermore, GLYX-13 ameliorated the loss of NR2A (GLYX-13 0.8613 ± 0.08200 vs. vehicle 0.5096 ± 0.1002, ^#p < 0.05, n = 3 independent experiments) 24 h after reperfusion (Figure 7). However, GLYX-13 had no significant effect on the expression of NR2B 24 h after reperfusion (Figure 8).

We adopted whole-cell patch clamp to record eEPSC in CA1 neurons of hippocampal slice when cells were clamped at −70 mV. OGD induced a significant increase of eEPSC (OGD 190.1 ± 3.9% of control, n = 5, **p < 0.01, one-way ANOVA; Figure 9). However this pathological potentiation could be abolished by selective NMDAR antagonist AP5, which implied that i-LTP required the participation of NMDARs (Figure 11). Notably, the application of 10 μmol/L GLYX-13 ameliorated pathological potentiation of eEPSC (GLYX-13 116.1 ± 3.0% vs. OGD 190.1 ± 3.9%, n = 5, ^##p < 0.01, Figure 9). To detect whether the inhibition of i-LTP by GLYX-13 was because of binding to the glycine co-agonist site on the NMDARs, 100 μmol/L full glycine site agonist D-Serine was used, which concentration was confirmed to fully stimulate the glycine site (Zhang et al., 2008). The application of 100 μmol/L D-serine completely occluded the effect of GLYX-13 (D-serine 193.7 ± 3.1% vs. GLYX-13 116.1 ± 3.0%, n = 5, p < 0.01, Figure 9). The above results suggested that GLYX-13 exerts inhibitory effect on pathological synaptic plasticity by binding the glycine co-agonist site of NMDARs.

Considering the results of western blot analysis, we have found that GLYX-13 could influence the expression of NR2A and NR2B and the phosphorylation of above subunits. Accordingly, we hypothesized that GLYX-13 probably inhibit the pathological potentiation of eEPSC by modulating the subunit composition of NMDARs. Therefore, we carried out a series of experimental protocols and use full of the selective NR2B antagonist Ro256981 (20 nmol/L) to investigate the functional NR2A and NR2B-containing NMDAR mediated eEPSC (NR2A/NR2B-eEPSC) amplitude under various conditions (Figure 10A). We observed that OGD induced a significant increase of the amplitude of NR2B-eEPSC (OGD 277.4 ± 9.1% of control) with the enhancement of total NMDAR-eEPSC (OGD 162.8 ± 8.0% of control, n = 5, **p < 0.01; Figure 10B). However, we only found a slight but not significant increase in NR2A-eEPSC. Notably, simultaneous giving 10 μmol/L GLYX-13 during OGD effectively suppressed the pathological potentiation of NR2B-eEPSC (GLYX-13 91.3 ± 8.3% vs. OGD 162.8 ± 8.0%, n = 5, ^##p < 0.01). Moreover, 100 μmol/L full glycine site co-agonist D-serine entirely reversed the inhibiting effect on NR2B-eEPSC produced by GLYX-13 (D-serine 278.3 ± 7.9% vs. GLYX-13 91.3 ± 8.3%, n = 5, **p < 0.01; Figure 10B). Consequently, GLYX-13 restored the NR2B subunit-mediated EPSC back to normal level.

Above results of electrophysiological study suggested that GLYX-13 exerts neuro-protective effects in vitro OGD model, to a great extent which mimics the acute situation after stroke attack. However, it is necessary to verify whether modulation of NMDAR NR2 subunits would also be effective to alleviate cerebral ischemic injury in vivo. TTC staining showed that treatment of GLYX-13 + Ro-2569819 (selective NR2B antagonist 5 mg/kg) after MCAO failed to present additional neuro-protective effect compared to GLYX-13 intervention alone (G + R 17.45 ± 1.442% vs. GLYX-13 17 ± 3.468% n = 6, p > 0.05); by contrast treatment of GLYX-13 + NVP-AAM077 (selective NR2A antagonist 2.4 mg/kg) completely reversed the amelioration of cerebral infarction elicited by GLYX-13 (G + N 46.79 ± 4.620% vs. GLYX-13 17 ± 3.468%; Figure 12). Above results suggested that compared with the principal role of NR2B subunits in vitro acute OGD model, the dysfunction of NR2A subunits are likely more obvious in vivo a few hours even a few days after stroke.