Altered glutamatergic tone reveals two distinct resting state networks at the cellular level in hippocampal sclerosis

Hippocampal sclerosis (HS), the most common subset of drug-resistant epilepsy (DRE), is associated with large-scale network abnormalities, even under resting state. We studied the excitatory postsynaptic currents (EPSCs) recorded from pyramidal neurons in resected samples under resting conditions from the hippocampal and anterior temporal lobe (ATL) obtained from patients with HS (n = 14) undergoing resective surgery. We observed higher frequency and amplitude of spontaneous EPSCs in both the samples compared to non-seizure control samples. Application of tetrodotoxin (TTX) reduced the frequency of spontaneous EPSCs by 49.6 ± 4.3% and 61.8 ± 6.2% in the hippocampal and ATL samples, respectively. The magnitude of reduction caused by TTX with respect to non-seizure controls was significantly higher in the ATL samples than in the hippocampal samples. The magnitude of the change in the expression of the NR2A subunit of the NMDA receptors also varied in these two regions. Thus, the mechanism of hyperexcitabilty mediated by glutamatergic network reorganization in the hippocampal region is different from that in the ATL region of patients with HS, suggesting two independent resting-state networks at the cellular level. Taken together, these findings will improve the understanding of the broadly distributed resting-state networks in HS.


Results
Histopathology and immunohistochemistry revealed typical hippocampal sclerosis but no neuronal loss and gliosis in ATL samples. The main pathological finding observed in excised tissue from patients with HS and treated with temporal lobectomy was confirmed by histopathology. Detailed histological examination of the tissues obtained from all the patients with HS (as mentioned in Table 1) were carried out. Typical aspects of a sclerotic hippocampus were observed in all the patients. We assessed for neuronal dropout leading to loss of normal laminar architecture, as highlighted on Neu N staining. Sections from the hippocampus showed dispersion of neurons as well as loss of neurons (Fig. 1a), as highlighted by NeuN staining (Fig. 1b). GFAP staining demonstrated reactive gliosis in the hippocampus (Fig. 1c). Sections from anterior temporal lobe (ATL) tissue showed the presence of ischemic neurons in few cases; however, no apparent neuronal loss was observed ( Fig. 1d and e). Reactive gliosis was absent in the ATL samples as GFAP staining did not show any positively stained reactive astrocytes (Fig. 1f). These results suggested that structural abnormalities like, loss of neurons and gliosis, was primarily present in the hippocampus of patients with HS.

Spontaneous postsynaptic activity in pyramidal neurons of slice preparations obtained from the hippocampal and ATL regions of patients with HS. Spontaneous EPSCs on pyramidal neurons
were recorded at −70 mV under whole-cell voltage clamps in samples obtained from non-seizure controls (cortical samples obtained from the periphery of tumour of patients undergoing surgery for glioma) as well as hippocampal and anterior temporal lobe samples obtained from patients with HS. These currents, which appeared as inward events, were completely blocked following 10 min superfusion of the slices with ACSF containing the AMPA receptor antagonist CNQX (10 µM) in an admixture with the NMDA receptor antagonist APV (50 µM) ( Fig. 2A). These results support the contention that these events are spontaneous EPSCs mediated by AMPA/ kainite and NMDA receptors. The frequency of EPSCs remains unaffected by bath application of GABA A receptor antagonist bicuculline (10 μM), suggesting that glutamatergic activity in the pyramidal neurons is not influenced by the basal GABAergic activity in the slices at −70 mV ( Fig. 2A). Similarly, spontaneous EPSCs were also recorded from slice preparations obtained from the hippocampal and anterior temporal regions of patients with HS ( Fig. 2A). Figure 2B shows that the normalized cumulative inter-event interval distributions in pyramidal neurons of the hippocampal (n = 14) and ATL samples (n = 14) were shifted to a higher interval in comparison with the non-seizure control (n = 20). The mean frequency of spontaneous EPSCs in the hippocampal samples was 0.99 ± 0.13 Hz, while in the case of samples obtained from the ATL, it was 1.39 ± 0.15 Hz. The mean amplitude of the spontaneous EPSCs in the hippocampal and ATL samples obtained from HS patients were significantly greater than the non-seizure controls (Table 2). However, the rise time and decay time constant (τ d ) in HS samples were not significantly different from those in non-seizure controls (Table 2), which may be because spontaneous EPSCs were mostly recorded from the soma of pyramidal neurons. Bath application of an admixture of CNQX (10 µM) with APV (50 µM) completely blocked the glutamatergic events in both the hippocampal and ATL samples from patients with HS. Together, these findings suggested enhanced glutamatergic activity in both the hippocampal and ATL samples of HS, while the magnitude of increase was higher in the ATL samples compared to the hippocampal samples.
We also recorded the spontaneous inhibitory postsynaptic currents (IPSCs) at 0 mV from the pyramidal neurons of non-seizure control and the hippocampal and ATL samples of HS (Fig. 3A). The spontaneous IPSCs were completely blocked by superfusion of slices with ACSF containing bicuculline (10 μM). However, we did not observe a significant difference in the frequency of spontaneous inhibitory postsynaptic currents (IPSCs)

Suppression of spontaneous EPSCs in pyramidal neurons of slice preparations obtained from hippocampal and ATL samples by Na + -channel blocker tetrodotoxin (TTX).
To isolate action potential (AP)-independent from AP-dependent glutamatergic EPSCs, slice preparations obtained from non-seizure controls as well as the hippocampal and anterior temporal lobe samples obtained from patients with HS were superfused with TTX (200 nM)-containing ACSF for 10 mins. TTX reduced the mean frequency and the mean amplitude of glutamatergic events by 43.9 ± 2.2% and 50.9 ± 2.3%, respectively, in the non-epileptic control ( Fig. 4; Table 3). These results were consistent with TTX-induced block of the action potential-dependent glutamatergic transmission; only miniature EPSCs (mEPSCs) remained in the presence of TTX. In the hippocampal neurons of HS (n = 14), the frequency of spontaneous EPSCs in TTX-containing ACSF was reduced by 50.26 ± 2.1%, while in the ATL samples (n = 14) it was reduced by 61.8 ± 3.1% (Fig. 4). TTX reduced the mean amplitude of glutamatergic events by 24.7 ± 5.2% and 47.8 ± 7.6% in the hippocampal and anterior temporal lobe samples, respectively, compared to that in the case of the non-seizure control ( Table 3). Application of TTX significantly reduced the mean amplitude of spontaneous EPSCs in all the samples as summarized in Table 3. Furthermore, we did not observe any significant difference in the frequencies of mEPSCs between the non-seizure control and the hippocampal and ATL samples obtained from patients with HS (Table 3). In the hippocampal neurons, the reduction in the mean amplitude by TTX was higher compared to the ATL neurons of the HS patients (p = 0.02) as depicted by the mean amplitudes of mEPSCs ( Table 3). The cumulative distribution of mEPSC amplitudes was also skewed toward larger amplitudes in the ATL samples compared to the hippocampal samples (Fig. 5A). The rise time and decay time constant (τ d ) of the mEPSCs were not significantly affected by TTX (Table 3). Furthermore, TTX blocked the action potential-dependent EPSCs in all the samples, and we found that the contribution of AP-dependent EPSC frequency with respect to non-seizure controls was significantly higher in ATL samples compared to hippocampal samples (Fig. 5B). Our data indicated that the increased glutamatergic tone in the hippocampal samples is different from that in the case of ATL samples of HS.
Differential regulation of mRNA and protein levels of NR2A and NR2B subunits of NMDA receptors in the hippocampal and ATL samples. Quantitative PCR was performed in triplicate to quantify the difference in the expression of NR2A and NR2B subunits in both the hippocampal (n = 10) and ATL (n = 10) samples resected from HS patients. As shown in Fig. 6A, a significant increase in the expression of NR2A was observed in both ATL (3.2-fold ± 0.087; p = 0.002) and hippocampal (2.08-fold ± 0.132; p = 0.003 according to one-way ANOVA followed by Dunnett post hoc test) tissues compared to non-seizure control, although the fold change was higher in the ATL tissues compared to the hippocampal tissues (p = 0.006 according to one-way ANOVA followed by Tukey post hoc test). No significant change was observed at the mRNA levels of the NR2B gene in both ATL (0.92-fold ± 0.135) and hippocampal (0.83-fold ± 0.146) tissues (Fig. 6A). Western blotting analyses of NR2A revealed significant upregulation of NR2A protein levels in both the ATL (p = 0.006) and the hippocampal samples (p = 0.043) compared to the control. Upregulation of NR2A in the ATL region was significantly higher than the hippocampal samples (p = 0.035). Again, no significant change was observed at the protein levels of the NR2B gene in both ATL and hippocampal tissues (Fig. 6B). Taken together, these results    Table 2. Kinetics of Spontaneous EPSCs. Characteristics of spontaneous EPSCs recorded from pyramidal neurons in resected samples from non-seizure control and the hippocampal and ATL samples from HS patients. The data are presented as the mean ± S.E.M. *p = 0.04; **p = 0.009 compared to non-seizure control according to one-way ANOVA followed by Tukey post hoc test.
suggest differentially increased expression of the glutamate NMDA receptor NR2A subunit both at the mRNA and protein levels in the hippocampal and ATL regions, whereas NR2B expression is unaltered in both the regions in patients with HS.

Ratio of frequency and peak amplitude of glutamatergic-GABAergic synaptic activity varied
in the hippocampal and ATL samples. Index ratios were obtained by dividing the average frequency and the mean peak amplitudes of glutamatergic events by that of GABAergic events in a subset of neurons where the glutamatergic and GABAergic synaptic activities could be recorded in the same cell. GABAergic inhibitory post-synaptic currents (IPSCs) were recorded at a holding potential of 0 mV (Fig. 3). Spontaneous IPSCs were observed as outward currents, which were abolished by the application of bicuculline (10 μM). This analysis included nine cells from the hippocampal region and eleven cells from the ATL region of HS patients. The ratio of frequency and peak amplitude of glutamatergic/GABAergic events were significantly higher in the hippocampal (n = 9) and the ATL samples (n = 9) obtained from the HS patients compared to the ratio in non-seizure controls (n = 12) as shown in Fig. 7. The average glutamatergic-GABAergic event frequency ratio in the hippocampal sample was lower (5.2 ± 0.7) than the ATL sample (9.8 ± 0.9; p = 0.006 according to one-way ANOVA followed by Tukey post hoc test) in these patients (Fig. 7A). The index ratio of glutamatergic-GABAergic event peak amplitude was also lower in the hippocampal samples (1.2 ± 0.16) compared to that in the ATL samples (1.9 ± 0.2; p = 0.04 according to one-way ANOVA followed by Tukey post hoc test) obtained from patients with HS. These data suggested that relative to inhibitory synaptic transmission, the excitatory synaptic transmission was higher in the ATL region compared to the hippocampal region of patients with HS.

Discussion
The present results disclose evidence that supports the concepts that i) two resting-state networks are present in patients with HS, one arising from the hippocampus and the other from the anterior temporal lobe; and ii) the spontaneous glutamatergic tone varies in resected brain specimens obtained from these regions, potentially mediated by two independent cellular mechanisms. These concepts can help clarify complicated network reorganizations as depicted in patients with HS at the cellular level. Under resting conditions, the frequency of the glutamatergic EPSCs recorded from pyramidal neurons in the hippocampal samples of HS patients were higher compared to the non-seizure controls 12 . Here, we showed that   the magnitude of increase in the glutamatergic transmission in the ATL samples was significantly higher than that in the case of the hippocampal samples under resting conditions (Figs 2 and 4). This finding indicated that under resting conditions, the mechanism that leads to abnormal excitatory activity in these two regions was different.
We found that under resting conditions, there was no significant effect of GABA A receptor antagonist, bicuculline, on spontaneous EPSCs (Fig. 2), and there was no significant alteration of GABAergic synaptic transmission in both the hippocampal and ATL samples obtained from patients with HS (Fig. 3). This finding suggested that hyperexcitability in HS is regulated primarily by the prototypical glutamate inputs in the hippocampus and ATL. Two distinct types of spontaneous glutamatergic responses contribute to the excitatory transmission; one is the AP-independent spontaneous activity (mEPSCs), which reflects a quantal mechanism and is insensitive to the Na + channel blocker tetrodotoxin (TTX). The other is an AP-dependent spontaneous release, which is mediated by an AP invading the presynaptic terminals. Compared with the spontaneous mEPSCs, these events are sensitive to TTX and are characterized by a greater amplitude, as they reflect a multi-bouton glutamate release that is triggered by the presynaptic AP. The result that Na + blocker TTX reduced the frequency of spontaneous EPSCs in the hippocampal and ATL samples of HS patients as well as in non-seizure controls indicates an inhibition of the action potential-mediated events. The magnitude of contribution of the AP-dependent EPSC frequency to the increased glutamatergic activity on pyramidal neurons of ATL samples was higher than that in the case of the hippocampal samples (Fig. 5B). This finding indicated that under resting conditions, presynaptic glutamate release processes leading to increased excitatory activity in the hippocampus are different from those in the ATL region. The higher mean amplitude of mEPSCs in the ATL neurons compared to the hippocampal neurons of HS suggests that the increase in the release probability of glutamate in ATL neurons was higher compared to the hippocampal neurons in the HS samples. Because the normalized cumulative distribution is also suggestive of a  shift toward larger amplitudes of mEPSCs (Fig. 5A) in the ATL samples relative to the hippocampal samples, the role of postsynaptic glutamate receptors cannot be ruled out. In addition, we found that the relative contribution of excitatory synaptic transmission and inhibitory synaptic transmission significantly varied in these two regions, further confirming the fact that under a resting state, aberrant synaptic activity in these two regions were different from each other. Thus, we speculated that the glutamatergic network reorganization responsible for hyperexcitability in the hippocampus is different from that in the case of ATL in patients with HS. Even though we observed neuronal loss and reactive gliosis in the hippocampal samples and no apparent loss of neurons and reactive gliosis in the ATL samples (Fig. 1), it is not possible to correlate the severity of damage with the hyperexcitability pattern in these two regions. The possible relationship between inflammatory reactions and hyperexcitability in the hippocampus and ATL needs to be explored in future. Endogenous NMDA receptor activity plays a crucial role in the generation of enhanced excitatory activity in the hippocampal samples obtained from patients with HS 12 . Using RNA-seq analysis of the hippocampal tissue resected from patients with HS, we have previously identified hubs of genes linked to synaptic transmission and neuronal network modulation 18 . We reported significant alterations in the expression levels of transporters, receptors and molecules involved in calcium signalling 18 . However, in that study we could not detect mRNA level alterations of NR2A levels in the RNAseq analysis, which could be attributed to the small sample size. In the current study, we not only found differential upregulation of NR2A mRNA levels but also differentially increased In comparison to the ATL samples the index ratio of frequency was lower in the hippocampal samples. (B) Index ratio for peak amplitude was obtained by dividing the mean peak amplitude of glutamatergic events by that of GABAergic events in a subset of pyramidal neurons where the spontaneous EPSCs and IPSCs could be recorded in the same neuron of non-seizure control and the hippocampal and ATL samples of patients with HS. The index ratio of peak amplitude in the hippocampal samples was lower compared to the ATL samples. Graph and error bars represent mean and S.E.M., respectively, of data obtained from twelve non-seizure control samples from twelve patients, and nine hippocampal and nine ATL samples obtained from nine patients with HS. *p < 0.05; **p < 0.001 compared to non-seizure control according to one-way ANOVA followed by Dunnett post-hoc test. # p < 0.05; ## p < 0.01 according to one-way ANOVA followed by Tukey post hoc test.
protein levels in the hippocampal and the ATL samples. NMDARs with different subunit combinations differed in their electrophysiological properties as well as their sensitivities to modulation by intracellular messengers 19 . NR2A and NR2B subunits prevailed in the adult cortex and the hippocampus, and previous studies have proposed that an increase in the NR2A/NR2B ratio in response to enhanced neural activity may not only contribute to maturational differences in the human hippocampal NMDAR function but may also play a role in the pathophysiology of neurodevelopmental disorders 20,21 . In this study, we examined the contribution of these two subunit-containing NMDARs to the hyperexcitability observed in the ATL and the hippocampal samples. Our findings suggest differential upregulation of the mRNA as well as protein levels of NR2A between the ATL and the hippocampal samples, while no significant change was observed in the NR2B levels (Fig. 6). This finding again highlights the fact that the mechanism of NMDA receptor-mediated hyperexcitability varies between the hippocampal and ATL regions of patients with HS. Previous reports on rat models of limbic epilepsy are also suggestive of a contribution of NR2A-containing NMDA receptors in epileptogenesis and the development of mossy fiber sprouting, whereas NR2B-containing NMDA receptors were not shown to contribute in epileptogenesis and mossy fiber sprouting 13 . We speculated that the differential expression of NR2A subunit-containing NMDARs might contribute to the distinct hyperexcitability pattern in the ATL and the hippocampal regions of patients with HS.
In conclusion, this is the first direct evidence that large-scale networks exist at the cellular level in patients with HS under resting conditions, further confirming the conjecture that HS is a distributed network disorder and not a focal disorder. We have shown that under resting state independent of spiking activities, patients with HS showed two aberrant networks: one emanating from the hippocampus and the other from the anterior temporal lobe. Our study suggests that the excitatory synaptic connections in these two regions varied significantly, thereby causing a difference in the glutamatergic tone. This finding indicated that the reinforced glutamatergic synaptic connectivity to form a network in the hippocampus was different from that in the ATL in patients with HS. Thus, an anterior temporal lobectomy along with amygdalo-hippocampectomy is likely to have a better outcome than selective amygdalo-hippocampectomy, which spares the anterior temporal lobe. This study also provided some interesting information on the complexity of networks that may exist at the cellular level in HS, which may partly explain the reduction in the overall seizure-free outcome on long-term follow-up in patients undergoing surgery for HS.
The main limitation of this study was the inability to sample extra-temporal tissues during surgery due to ethical considerations. In addition, the possibility of finding altered glutamatergic synaptic mechanisms in the extra-temporal regions cannot be ruled out. At this stage, it also cannot be ascertained whether the networks in ATL were initially dependent or linked to those in the hippocampus and later became independent due to kindling or were de novo in nature. Another limitation of this study was the use of margins of tumors from patients with gliomas without seizures as non-seizure controls. Ideally, the non-seizure control could be samples resected from the temporal lobe structures of patients with similar ages with non-epilepsy pathologies such as glioma. However, it is ethically not possible to obtain region-specific tissues, and most of the temporal lobe tumors are associated with seizures. The temporal lobe samples obtained from autopsy patients were also not suitable to perform cellular electrophysiological studies. Thus, region-specific alterations in glutamatergic tone observed in the current study needs to be confirmed in a relevant animal model of HS. Future studies are also needed to differentiate the role of NMDA receptors and AMPA receptors in the hippocampus and ATL of patients with HS. In addition, further investigations to differentiate the involvement of NR2A and NR2B-containing NMDARs should also be considered.

Materials and Methods
Subjects & clinical evaluation. Patients with HS underwent phased pre-surgical assessment, and the pathology in each patient was demonstrated by documenting convergent data on MRI, video EEG (vEEG), Fluoro-2-deoxyglucose positron emission tomography (FDG-PET) evaluations and electrocorticography (ECoG) and confirmed by histopathological examinations in accordance with the recent World Health Organization classification. Fourteen patients with HS who underwent surgery from September 2014 to December 2015 were included in this study ( Table 1). Six of the patients were female. The inclusion criteria included MRI-vEEG-ECoG concordance, absence of any obvious of structural changes/pathologies in the extra-hippocampal regions (e.g., gliosis, atrophy), and absence of dual pathology (e.g., cortical dysplasia). The mean age of the patients was 23.6 ± 10.2 years. The duration of seizures averaged at 10.5 ± 8.3 months. The seizure onset was 13.1 ± 8.1 years. The average number of AEDs/patient was 3.1 ± 0.7. Surgery included a standard anterior temporal lobectomy along with amygdalo-hippocampectomy. The structures resected included anterior temporal lobe, amygdala, hippocampus up to the level of superior colliculus, uncus, para-hippocampal gyrus, enterorhinal cortex, and subiculum. Prior to resection, electrocorticography (EcoG) was performed by placing a 4 × 5 grid over the lateral surface of temporal lobe, and grading was performed under anesthesia conditions as previously described 22,23 (Table 1). All these patients were seizure-free post-operatively (Class I Engel outcome) 24 . Part of the excised tissue from the hippocampus and ATL was fixed in 10% neutral buffered formalin, processed and embedded in paraffin for immunostaining. Haematoxylin and eosin stained sections of the hippocampal and ATL were examined. Immunohistochemistry was performed on the formalin-fixed paraffin-embedded tissue using primary antibodies against NeuN (neuronal nuclear protein; 1:100; Dako, Denmark) and glial fibrillary acidic protein (GFAP; 1:800; Dako, Denmark), as described previously 25 . The experiments described in this study were performed per the guidelines of the institutional ethics committee (IEC), All India Institute of Medical Sciences, New Delhi and human ethics committee, National Brain Research Centre, Manesar. Informed consent was obtained from all patients and their family members prior to the start of the study. Research material for this study consisted of tissues from the hippocampus and anterior temporal region, which was a part of the planned surgical resection (hence had no additional risk for the patient). Patients with HS were treated with polytherapy and anti-epileptic drugs (AEDs) that included valproic acid, carbamazepine, phenytoin and levetiracetam. Non-seizure controls included cortical tissues obtained from the margins of tumors during surgical resection (again a part of the planned surgical resection) in patients with gliomas without seizures ( Table 1). The experiments performed in the study were approved by IEC, All India Institute of Medical Sciences, New Delhi and human ethics committee, National Brain Research Centre, Manesar. Cellular electrophysiology. Hippocampal and anterior temporal lobe samples resected during surgery were immediately transferred to well carbogenated, ice-cold artificial cerebrospinal fluid (ACSF). ACSF consisted of CaCl 2 , 2 mM; NaHCO 3 , 25 mM; NaH 2 PO 4 , 1.25 mM; NaCl, 125 mM; KCL, 2.5 mM; MgCl 2 , 1 mM; and Glucose, 25 mM. Samples were further processed for cellular electrophysiological experiments as per the protocol reported earlier 12 . Briefly, samples were placed in well carbogenated, ice-cold ACSF within 2 minutes of resection. Within 15 minutes of resection, slices (350-µm thick) were prepared using a vibrating blade microtome. Slices were transferred to the recording chamber and perfused with carbogenated ACSF at the rate of 2 ml/min. Whole-cell recordings were obtained from the soma of visually identified pyramidal neurons in slice preparations according to the standard patch-clamp technique using an Axopatch 200B amplifier (Molecular Devices, USA). We have specifically chosen neurons which had a pyramid-like soma and a single thick tapering apical dendrite, as visualised using IR-DIC videomicroscopy, for our studies. Patch pipettes were filled with internal solution containing HEPES, 10 mM; Cs-methanesulfonate, 130 mM; EGTA, 10 mM; CsCl, 10 mM; MgCl 2 , 2 mM (pH adjusted to 7.3 with CsOH; 340 mOsm). Pipette resistance ranged between 3 and 5 MΩ. To record spontaneous EPSCs from neurons under whole-cell configuration at a holding potential of −70 mV, 5 mM QX-314 was added in the internal pipette solution. Spontaneous IPSCs were recorded at 0 mV. All experiments were performed at room temperature (~22 °C). Experiments on non-seizure control samples were performed in the same manner as in the case of HS samples.
EPSCs and IPSCs were analyzed using pCLAMP 10.0 software (Molecular Devices, Sunnyvale, USA). Visual inspection of each recording was performed to select single events with a sharp rising phase and an exponential decay for kinetic analysis of the EPSC events. The frequency, peak amplitude, rise time (10-90%), and decay-time constant (τ d ) of the synaptic events were measured using the clampfit module of pCLAMP 10.0 software. EPSCs and IPSCs showing double-and multiple-peaks were excluded from analysis of kinetic properties but were utilized for the calculation of frequency of events as multiple events. The data were expressed as the mean ± S.E.M. of results obtained from various patients. Statistical tests included one-way ANOVA in Sigma plot 12.0 (Systat Software, Inc., Chicago, IL) to determine any significance in the data. Furthermore, the cumulative distributions of events in non-seizure control versus the hippocampal and ATL groups were compared using the Kolmogorov-Smirnov test (K-S test). For this, events from different neurons in each group were pooled together and then subjected to the K-S test using the Clampfit module of the pCLAMP10.0 software.

Molecular biology studies. Tissue preparation.
Approximately 5-mm-thick hippocampal and ATL samples resected during surgery from patients with HS were stored in RNAlater stabilization reagent (Qiagen) in order to prevent degradation of RNA for RNA extraction and snap frozen for western blotting analyses. All samples were stored at −70 °C until further processing. The samples with RNAlater were stored at 4 °C overnight and subsequently frozen and kept at −70 °C.
Western blotting analysis. Frozen brain tissues were homogenized on ice in RIPA buffer (20 mM Tris-HCl, 150 mM NaCl, 1 mM Na2 EDTA, 1 mM EGTA, 1% NP-40, 1% sodium deoxycholate, pH 7.5), supplemented with protease inhibitor cocktail and phosphatase inhibitors. Protein concentrations were measured using a bicinchoninic acid Protein Assay Kit (Pierce, Rockford, IL, USA) and a microplate reader at 570 nm (microplate reader (iMark ™ Microplate Absorbance Reader, Bio-Rad, USA), applied at concentrations of 25 µg protein per lane on 10% polyacrylamide gels along with precision plus protein dual color standards (Bio-Rad, USA) and transferred onto polyvinylidene fluoride membranes using standard procedures. Because NR2A (~177 kDa)/ NR2B (~166 kDa) and GAPDH (~37 kDa) have large molecular weight differences, the same membrane was cut into two halves from the center and used for further development with the respective antibodies. GAPDH was used as a loading control. Briefly, the membranes were incubated overnight in primary antibody against NR2A (sc-390094; Santa Cruz Biotechnology, Inc.), NR2B (ab28373; Abcam) and GAPDH (sc-365062; Santa Cruz Biotechnology, Inc.) in 1:500 dilution in 5% skimmed milk in Tris buffered saline (200 mmol/L Tris-HCl, 1·37 mol/L sodium chloride, pH 7·6) with 0·1% Tween 20. Membranes were blotted using anti-mouse IgG-HRP (sc-2005) and anti-rabbit IgG-HRP (sc-2030) (Santa Cruz Biotechnology, Inc.) and images were detected with a FluorChem M imager (Cell BioSciences, Santa Clara, CA, USA) using enhanced chemilluminescence with Pierce Super Signal (Pierce Biotechnology; Rockford, IL, USA). Densitometry analysis of the desired bands in western blots was performed using Quantity One software (Bio-Rad, USA). All densitometries are expressed in arbitrary units (AU). The signal intensity of the loading control (GAPDH) was used for normalization. Statistical analysis was performed using one-way analysis of variance (ANOVA), post hoc Dunnett for comparison with non-seizure control, and post hoc Tukey for comparison between ATL with H samples using Sigma Plot 12.0 software. The data are presented as the mean ± SEM. A p-value of <0.05 was considered statistically significant.