M-Calpain Activation Facilitates Seizure Induced KCC2 Down Regulation

Potassium chloride co-transporter 2 (KCC2), a major chloride transporter that maintains GABAA receptor inhibition in mature mammalian neurons, is down-regulated in the hippocampus during epileptogenesis. Impaired KCC2 function accelerates or facilitates seizure onset. Calpain, with two main subtypes of m- and μ-calpain, is a Ca2+-dependent cysteine protease that mediates the nonlysosomal degradation of KCC2. Although recent studies have demonstrated that calpain inhibitors exert antiepileptic and neuroprotective effects in animal models of acute and chronic epilepsy, whether calpain activation affects seizure induction through KCC2 degradation remains unknown. Our results showed that: (1) Blockade of calpain by non-selective calpain inhibitor MDL-28170 prevented convulsant stimulation induced KCC2 downregulation, and reduced the incidence and the severity of pentylenetetrazole (PTZ) induced seizures. (2) m-calpain, but not μ-calpain, inhibitor mimicked MDL-28170 effect on preventing KCC2 downregulation. (3) Phosphorylation of m-calpain has been significantly enhanced during seizure onset, which was partly mediated by the calcium independent MAPK/ERK signaling pathway activation. (4) MAPK/ERK signaling blockade also had similar effect as total calpain blockade on both KCC2 downregulation and animal seizure induction. The results indicate that upregulated m-calpain activation by MAPK/ERK during convulsant stimulation down regulates both cytoplasm- and membrane KCC2, and in turn facilitates seizure induction. This finding may provide a foundation for the development of highly effective antiepileptic drugs targeting of m-calpain.

Potassium chloride co-transporter 2 (KCC2), a major chloride transporter that maintains GABA A receptor inhibition in mature mammalian neurons, is down-regulated in the hippocampus during epileptogenesis. Impaired KCC2 function accelerates or facilitates seizure onset. Calpain, with two main subtypes of m-and µ-calpain, is a Ca 2+ -dependent cysteine protease that mediates the nonlysosomal degradation of KCC2. Although recent studies have demonstrated that calpain inhibitors exert antiepileptic and neuroprotective effects in animal models of acute and chronic epilepsy, whether calpain activation affects seizure induction through KCC2 degradation remains unknown. Our results showed that: (1) Blockade of calpain by non-selective calpain inhibitor MDL-28170 prevented convulsant stimulation induced KCC2 downregulation, and reduced the incidence and the severity of pentylenetetrazole (PTZ) induced seizures.
(2) m-calpain, but not µ-calpain, inhibitor mimicked MDL-28170 effect on preventing KCC2 downregulation. (3) Phosphorylation of m-calpain has been significantly enhanced during seizure onset, which was partly mediated by the calcium independent MAPK/ERK signaling pathway activation. (4) MAPK/ERK signaling blockade also had similar effect as total calpain blockade on both KCC2 downregulation and animal seizure induction. The results indicate that upregulated m-calpain activation by MAPK/ERK during convulsant stimulation down regulates both cytoplasm-and membrane KCC2, and in turn facilitates seizure induction. This finding may provide a foundation for the development of highly effective antiepileptic drugs targeting of m-calpain.
Keywords: epilepsy, calpain, KCC2, MAPK/ERK, PTZ BACKGROUND The inhibitory effect of GABA A receptors has an important role in the maintenance of normal brain function and protection against epileptogenesis. These effects rely on a low intracellular Cl − ([Cl − ] i ) and high extracellular Cl − transmembrane gradient, which is established through Cl − transport by potassium chloride co-transporter 2 (KCC2) in mature neurons (Chamma et al., 2012;Mahadevan and Woodin, 2016;Wu et al., 2016). KCC2, the only KCC family member distributed in the central nervous system, mainly pumps Cl − out of neurons and maintains low [Cl − ] i (Kahle et al., 2008;Loscher et al., 2013). Numerous studies have shown that epileptic seizure stimuli significantly down-regulate KCC2 expression.
Calpain is a Ca 2+ -dependent cysteine protease that is ubiquitously expressed in mammals (Zimmerman and Schlaepfer, 1984;Ono et al., 2016). Despite its ubiquity and early discovery, calpain remains enigmatic. Two major isoforms of calpain exist, of which µ-calpain is activated by 3-50 µM Ca 2+ and m-calpain is activated by 400-800 µM Ca 2+ (Liu et al., 2008;Sorimachi et al., 2011a;Zanardelli et al., 2013), and both of them exist in mature neurons in central nervous system (Goll et al., 2003). However, in a specific neuron, the distribution of µ-calpain and m-calpain is considered to have regional specificity. The morphological evidence shows that µ-calpain mostly distributed at the synapse region, which facilitates its regulation on synaptic function through the effect to the cytoskeleton, the scaffold protein and the glutamate receptor (Liu et al., 2008). However, the ultrastructural localization of m-calpain in neurons is still unclear. Functional studies have shown that, in excitatory neurons, activation of synaptic NMDA receptors causes the activation of the µ-calpain (Xu et al., 2009), while the extra-synaptic NMDA receptor specifically activates m-calpain (Chen et al., 2007). Calpain activation has been reported to involve in many neurological disorders, such as traumatic brain injury (TBI), Alzheimer disease (AD), epilepsy, etc. (Sierra-Paredes et al., 1999;Araujo et al., 2008;Li et al., 2013;Lam et al., 2017;Nam et al., 2017). In terms of seizure, it is found that calpain was over-activated in KA induced seizure rats, and this effect could be blocked by calpain inhibitor MDL-28170 (Araujo et al., 2008). In another study in pilocarpine-induced status epilepticus (SE) model, calpain was also found to be abnormally activated, and MDL-28170 was proved to ameliorates seizure burden in that case (Lam et al., 2017). However, the specific mechanism underlying these effects remains unclear. One of the underlying mechanisms of calpain involvement in epilepsy might attribute to its regulation of KCC2 degradation, since KCC2 down regulation has been directly associated to seizure induction (Chen et al., 2017). KCC2 may follow two routes in clathrin-mediated KCC2 endocytosis. One route involves vesicle recycling, wherein KCC2 is reassembled and resumes its function in the plasma membrane. The other route involves degradation via a chain reaction mediated by calpain. Thus, the over-activation of calpain can diminish the intracellular KCC2 pool, eventually decreasing the expression of KCC2 on plasma membrane (Puskarjov et al., 2012(Puskarjov et al., , 2015Zhou et al., 2012;Chamma et al., 2013). These effects imply that calpain participates in epileptogenesis by regulating KCC2.
Meanwhile, calpain itself is subject to complex regulation. Recent studies have shown that in addition to regulation by intracellular Ca 2+ concentration ([Ca 2+ ] i ), m-calpain activity is also regulated through MAPK/ERK-mediated phosphorylation Liu et al., 2016). In Glading et al. (2004) provided a detailed discussion of MAKP/ERK-mediated m-calpain phosphorylation under physiological conditions and identified ser50 as the regulatory site of m-calpain. However, whether MAPK/ERK-mediated phosphorylation of m-calpain involves in seizure induction is still unknown.
Recent reports have shown that the contradictory effects exerted by nonselective calpain inhibitors may be attributed to the different or opposing roles of µ-calpain and m-calpain under certain pathological conditions Wang et al., 2016;Yin et al., 2016). Here, we report that blockade of calpain by calpain inhibitor MDL-28170 suppressed the seizure induction, as well as the KCC2 downregulation. We further demonstrated that m-calpain, but not µ-calpain, phosphorylation, which was partly mediated by the calcium independent MAPK/ERK signaling pathway activation, regulated KCC2 down regulation and hence the seizure onset.

Ethics Statement
All the animal experiments were approved by the Local Committees of the Use of the Laboratory Animals, Fudan University (Shanghai, China) and were carried out in accordance with the guidelines and regulations of National Natural Science Foundation of China animal research. In addition, institutional safety and biosecurity procedures were followed according to ''Laboratories-General requirements for biosafety (GB 19489-2008).''

Animal Preparation
Male SD rats were purchased from Shanghai Slack Experimental Animals Inc. The weight of the animal is 180-220 g, aged 5-6 weeks. Rats were housed in a regulated environment (22 ± 1 • C) with a 12 h light-dark cycle, and food and water were available ad libitum.

Hippocampal Slices Preparation
Rats were anesthetized by intraperitoneal injection of 1.25% pentobarbital sodium at a dose of 0.1 ml per 100 g weight. After being fully anesthetized, the rats were decapitated and their brains were separated and cooled in iced ACSF (in mmol/L: NaCl 124, KCl 3.3, KH 2 PO 4 1.2, NaHCO 3 26, CaCl 2 2.5, MgSO 4 2.4, Glucose 10) for 1 min, then trimmed on ice to expose the hippocampus as much as possible. The brain was then fixed on cryogenic vibration microtome for slicing. The brain was bathed in iced ACSF throughout the slicing. For immunoblotting the slice thickness was 350 µm and for immunohistochemistry the thickness was 100 µm. Slices were then transferred to an interface tissue chamber and superfused with ACSF at a temperature of 31-32 • C for 30 min. After that the slices were maintained at room temperature (RT) for 1 h to recover function. By then the slices were ready for further pharmacological treatments and sampling.

Hippocampal Slices Perfusion
Fresh hippocampal slices prepared from above were randomly divided into different groups at same numbers. Slices from one single preparation were seen as the same batch. Since our previous study, Chen et al. (2017) has reported that plasma membrane KCC2 (mKCC2) down-regulation in 0 Mg 2+ induced in vitro seizure model reached maximal at 2 h after 0 Mg 2+ ACSF incubation, we chose 2 h as the time scale of 0 Mg 2+ treatment in our current study. The time scale of BDNF treatment (#450-02, PeproTech, 200 ng/ml) was also set to 2 h according to a previous research (Rivera et al., 2002). In corresponding vehicle control or intervention groups, slices were also treated at same time with one of the following agents: MDL-28170: #ab145601, abcam, 50 µM; PD98059: #ab120234, abcam, 25 µM ; K252a: #420298, Calbiochem, 200 nM; Tautomycetin: #2305, Tocris, 20 nM; BAPTA-AM: # A1076, Sigma, 10 µM; Calpain Inhibitor I: # A6185-5MG, Sigma, 100 µM; Calpain Inhibitor IV: # 208724, Calbiochem, 200 µM. All drugs were dissolved in DMSO before being added to ACSF. The final concentration of DMSO was 0.1% in each treatment and bath incubation.

EEG Recording and Behavior Assays
Due to its low neurotoxicity and stability in inducing seizure, pentylenetetrazole-(PTZ, 50 mg/kg) induced seizure model was chosen in this study. Although PTZ has been widely accepted as a GABA A receptor antagonist, its actual mechanism in inducing seizure in in vivo animal model is not fully defined, since PTZ has been also reported to blockade of certain ion channels (Papp et al., 1987) and in vitro application of PTZ on hippocampal slices failed to evoke epileptiform burst activities as other GABA A receptor antagonist do (unpublished data). Behavioral seizures in freely moving rat combination with electroencephalograph (EEG) were recorded as described previously (Kong et al., 2010). In generally, male SD rat (180-220 g) were anesthetized with sodium pentobarbital (60 mg/kg, i.p.) and mounted in a stereotaxic apparatus with body temperature maintained at 37 • C. Two stainless steel screws (1 mm in diameter) were inserted through the skull with one screw serving as recording electrode above the hippocampus (AP −3.8 mm and ML 2.0 mm) and the other as reference electrode above the forehead. Screws were then connected to a connector-plug with wires for later connecting to recording leads. All electrodes were attached to a microconnector and fixed onto the skull with dental cement. After surgery, animals were allowed to recover for at least 5 days before the experiments. For experiment, rats were transferred to a plexiglas cage (25 × 25 × 40 cm) and habituated therein for at least 30 min, before intraperitoneal injection with either MDL-28170 (#ab145601, abcam, 50 mg/kg) or SL-327 (#HY-15437, MCE, 50 mg/kg) or equal volume vehicle (DMSO) in different groups as pre-treatment. Thirty minutes after that, PTZ (50 mg/kg) was injected intraperitoneally to induce seizure. Epileptic behavior and EEG were simultaneously recorded for 1 h after PTZ kindling, and then terminated by intraperitoneal injection of pentobarbital.
The EEG signals were sampled at rate of 2,500 Hz, analog inputs were amplified (1,000 times) and filtered (0.3-1 kHz) by using a NeuroLog System (Digitimer Ltd., Hearts, UK) and digitized with CED Micro 1401 (Cambridge Electronic Design, Cambridge, UK) and recorded in a personal computer using Spike two software (version 6.0, Cambridge Electronic Design, Cambridge, UK). Each recording lasted at least 1 h after PTZ injection. Classic Racine classification method was introduced to scale the PTZ-induced seizure severity: R1: chewing, blinking, facial or beard trembling twitching, stare, daze; R2: nodding, repeated scratch, circle around and wet dog shakes (WDS); R3: unilateral forelimb clonus, tail-erecting and back arching; R4: rearing with bilateral forelimb clonus; R5: rearing and falling (loss of postural control).

Plasma Membrane and Cytoplasm Protein Extraction and Sample Preparation
The plasma membrane and cytoplasm protein fractions were prepared followed by standard procedures provided by the Membrane Protein Extraction Kit purchased from Biovision (#K268-50, containing Homogenize Buffer, Protease Inhibitor Cocktail, Upper Phase Solution, and Lower Phase Solution). Hippocampal slices were dissociated to preserve only the hippocampus part under a dissecting microscope on ice, and then quickly homogenized in pre-cooled Homogenize Buffer containing 1/500 Protease Inhibitor Cocktail. The homogenate was centrifuged at 700× g for 10 min at 4 • C. Then the supernatant was transferred to a new vial and centrifuged at 10,000× g for 30 min at 4 • C. Collect both the pellet (the total membrane proteins) and the supernatant (the cytoplasm fraction). The total membrane proteins were further purified to get the plasma membrane proteins by affinity chromatography and density gradient centrifugation with agents provided in Membrane Protein Extraction Kit. The plasma membrane and cytoplasm fractions were then dissolved in 0.5% Triton X-100 in PBS, and were bathed in 45 • C with SDS sample buffer for 45 min for inactivation. In order to validate the effectiveness of membrane fraction and cytoplasm extraction, we examined the sodium-potassium adenosine triphosphatase (Na + /K + ATPase) which is considered as a specific marker of the plasma membrane in Western Blot by using its specific antibody. The result showed that under the same antibody concentration, the same exposure time, and the sample loading amount, a high intensity signal of sodium potassium ATPase was detected in the plasma membrane samples, while in the cytoplasm sample the signal is poor, which proves the effectiveness of our sample preparation (see Supplementary Figure S1 in supplementary data).

Power Spectrum Analysis
EEG data were exported to matlab (R2016a) as txt format and the recording channel was selected to generate a power spectrogram. The analysis script consisted of a fast Fourier transform using a Cosine-Bell data window with a window size of 1,024 data points. A window overlap of 87.5% was used to help to smooth the x-axis of the spectrogram. The power was expressed as µV 2 .

Data Analysis
For normally distributed data, unpaired student's t-tests was used for inter-group comparing, for not normally distributed data, we used Corresponding nonparametric analysis (i.e., Independent-Samples Mann-Whitney). Group data are expressed as mean ± SEM. Across different groups of data, statistically significant differences between means were determined using two-way ANOVA combined with Fisher's LSD test. Comparison within a group was carried out using a paired or unpaired t-test. The significance level was set at P < 0.05.

Calpain Inhibition Suppressed PTZ-Induced Acute Seizures
Calpain is involved in numerous neurological disorders, including seizure. Previous studies have demonstrated that calpain inhibitor MDL-28170 exerts suppressive effect on KA and pilocarpine induced seizure severity (Li et al., 2013;Lam et al., 2017). In current study, we further tested whether suppressing calpain activity with MDL-28170 would interfere the seizure activities in rats. We applied MDL-28170 to awake, freely moving rats 30 min prior to seizure induction with PTZ and simultaneously recorded seizure behavior and EEG patterns. Our result showed that MDL-28170 (50 mg/kg) significantly suppressed seizure behavior and epileptic activities. Seven out of 10 rats (70%) in the vehicle control group developed Racine-5 seizure behaviors after PTZ injection. Typical performances included twitching while standing and loss of balance. These behaviors were sometimes accompanied by systemic tonic convulsion. The EEG patterns of the rats showed highly synchronized cluster firing or burst firing ( Figure 1A). Although the other rats did not develop Racine-5 behavior, they demonstrated unilateral limb clonus (Racine 3), and their EEG components were mainly characterized by spikes and high-amplitude slow waves at approximately 3 Hz. By contrast, only 20% (2 out of 10) of the rats in the MDL-28170 pretreated group showed Racine-5 seizure behavior and burst firing after PTZ injection. The seizure incidence in the MDL-28170 pretreated group was significantly lower than that in the control group (Figures 1G-I, P < 0.05). The rats that failed to demonstrate Racine-5 behavior only showed some Racine-2 behaviors, such as nodding and WDS, and their EEG patterns mainly presented low-frequency oscillations ( Figure 1B). Time-frequency power analysis suggested that the firing patterns of these two groups differed after PTZ induction. In vehicle-PTZ animals, the intensity of high-frequency firing increased until HAFDs. This phase was followed by a defined silent phase, and regular spiking was later restored at approximately 10 Hz ( Figure 1C). By contrast, most of (80%) the MDL-28170pretreated animals exhibited low power but not HAFDs or a silent period during the recording ( Figure 1D). Meanwhile, the real-time behavioral score per minute of the MDL pretreated rats during the whole recording course was significantly lower than that of the control PTZ group (Figures 1E,F). These results suggested that calpain is involved in seizure induction and calpain inhibitor could reduce the intensity and possibility of seizure induction in an animal model of epilepsy.

Calpain Inhibitor MDL-28170 Prevented Convulsant-Stimulation-Evoked KCC2 Down-Regulation in the Hippocampus
Our previous work suggested that the decreased expression and function of KCC2 is necessary for epileptogenesis (Chen et al., 2017). The over-activation of calpain, a KCC2 decomposer, may contribute to the down-regulation of KCC2. Thus, we tested whether the rescuing effect of calpain inhibition on seizure susceptibility occurs through interference with KCC2 regulation. First, we performed immunostaining to examine KCC2 expression levels in hippocampal slices incubated with 0 Mg 2+ ACSF. We found that blocking calpain activity by MDL-28170 (50 µM) reversed 0 Mg 2+ -induced impairment of fluorescence intensity (Figure 2A). We further performed immunoblotting to quantify the expression level of mKCC2 in hippocampal slices from the in vivo PTZ model and in vitro 0 Mg 2+ model. Our results showed that mKCC2 was significantly down-regulated in the in vivo PTZ and in vitro 0 Mg 2+ models but was significantly reversed in the MDL-28170 pretreatment group (in vivo: PTZ: 78.5 ± 9.2%, n = 8; PTZ + MDL-28170: 109.8 ± 10.1%, n = 8. PTZ vs. Vehicle: P < 0.01; PTZ vs. (C) Spectrin breakdown product (mKCC2) expression from hippocampal slices. Slices were prepared from fresh brain and randomly divided into three different groups. KCC2 levels were normalized to control slices from the same preparation. (D) Spectrin breakdown product (SBDP)-145 level of 0 Mg 2+ and 0 Mg 2+ + MDL-28170 treated slices (total protein sample), normalized to control slices from the same preparation. * p < 0.05 compared with vehicle-PTZ group; * * p < 0.01 compared with vehicle-PTZ group; * * * p < 0.001 compared with vehicle-PTZ group; # p < 0.05 compared with 0 Mg 2+ group; ## p < 0.01 compared with 0 Mg 2+ group.

Phosphor-Serine M-Calpain Regulated Cellular KCC2 Level During Seizure Induction
The enhancement in calpain activity after seizure induction with 0 Mg 2+ may be attributed to increases in calpain expression levels. Therefore, we examined the expression levels of both m-calpain and µ-calpain. Surprisingly, neither m-nor µ-calpain altered during seizure induction with 0 Mg 2+ (m-calpain: 106.5 ± 9.5%, n = 3, µ-calpain: 103.4 ± 8.8%, n = 3; 0 Mg 2+ vs. Vehicle: p > 0.5; Figure 3C). This result suggested that enhanced calpain activation after seizure induction with 0 Mg 2+ was not resulted from changing of either m-or µ-calpain expression. Given that calpain expression levels remained unchanged, we focused on alteration of calpain function and activity. As previously reported, calpain can be activated through the phosphorylation of its Ser50 site (Glading et al., 2004;Zadran et al., 2010). We therefore examined the serine phosphorylation levels of both m-calpain and µ-calpain after epileptic induction with 0 Mg 2+ . After brain homogenate extracts were incubated with the main m-calpain or µ-calpain antibody to form immune complexes then subjected to immunoprecipitation and immunoblotting with an anti-serine phosphorylation antibody, we found that the phosphorylation rate of µ-calpain was not significantly changed, whereas that of m-calpain was significantly increased (µ-calpain: 95.8 ± 10.5%, n = 3, P > 0.05; m-calpain: 154.6 ± 5.2%, n = 3, P < 0.01; Figure 3D).
These results suggested that m-calpain, but not µ-calpain, is over-activated as a result of its serine site phosphorylation by the activation of TrkB-MAPK/ERK signaling pathway, which in turn plays a major role in the regulation of KCC2 expression during seizure induction.
Next, we examined the changes in m-calpain phosphorylation levels after BAPTA-AM treatment in both 0 Mg 2+ and BDNF induced model. The results demonstrated that the increased phosphorylation level of m-calpain induced by 0 Mg 2+ significantly decreased when co-treated with BAPTA-AM (0 Mg 2+ : 170.3 ± 7.6%, n = 3; 0 Mg 2+ + BAPTA: 137.5 ± 7.1%, n = 3. 0 Mg 2+ + BAPTA vs. 0 Mg 2+ : P < 0.05; Figure 4C), but still significantly higher than that of the control group (P < 0.05). In contrast, although BDNF also induced the increase of m-calpain phosphorylation level, BAPTA-AM treatment failed the affect the BDNF induced m-calpain phosphorylation (BDNF: 175.2 ± 9.4%, n = 3; BDNF + BAPTA: 157.0 ± 11.8%, n = 3. BDNF vs. Vehicle: P < 0.05; BDNF + BAPTA vs. Vehicle: P < 0.05; BDNF vs. BDNF + BAPTA: P > 0.05; Figure 4C). Putting together, these results demonstrated that, in 0 Mg 2+ model, both the mKCC2 level and the m-calpain phosphor-serine level were significantly affected by the cleavage of intracellular Ca 2+ , while neither KCC2 level nor phosphorylation of m-calpain induced by BDNF were affected by BAPTA. The data suggested that BDNF-TrkB-MAPK/ERK signaling pathway in regulation of m-calpain phosphorylation and in turn regulation of cellular KCC2 level is likely independent of [Ca 2+ ] i . Since m-calpain phosphorylation and cellular KCC2 regulation are likely more complicated with multiple mechanism, our data also suggested that, at least with some parts, this process is dependent on [Ca 2+ ] i level during seizure induction in 0 Mg 2+ epilepsy model.

Blocking the MAPK/ERK Signaling Pathway Alleviated Acute Seizures in vivo
Similar to direct calpain inhibition by MDL-28170, MAPK/ERK blockade also reversed the reduction of KCC2 during seizure stimulation. Thus, we next tested whether blockade of the MAPK/ERK pathway would also have similar effects as MDL-28170 on seizure induction in vivo. Given that PD98059 cannot penetrate the blood-brain barrier, we applied the brain penetrable ERK1 inhibitor SL-327 (Carr et al., 2009) to abolish the serine-site phosphorylation of m-calpain and recorded PTZ induced seizure behavior and the EEG patterns in rats. Our results showed that the proportion of rats with Racine-5 seizure behaviors and HDFAs in the SL-327 pretreatment group was significantly lower than that in the Vehicle-PTZ group, as described in the previous section (Figures 1F,G, 5F,G; P < 0.05). Two out of 10 rats in the SL-327 pretreatment group presented Racine-5 seizure behaviors and EEG patterns with bursts and HDFAs. Eight of the remaining rats developed only Racine-3 behaviors (Figure 5E), and their EEG patterns were mainly characterized by slow waves mixed with 0-10 Hz spikes ( Figure 5A). The fact that SL-327 mimicked the anti-seizure effect of complete calpain blockade suggested that the activation of m-calpain phosphorylation by the MAPK/ERK signaling pathway is the dominant mechanism of calpain involvement in epilepsy in the mechanism of calpain involved in epilepsy.

DISCUSSION
Our research indicates that, during the seizure onset, the over-activation of m-calpain, but not µ-calpain, enhances KCC2 degradation, and in turn, results in the functional FIGURE 5 | SL-327 intervention improved the incoming seizure induced by PTZ in freely moving rats. (A) Typical EEG recording traces from SL-327 pre-treated rats. The lower traces are an enlarged drawing from the dashed boxes of upper traces. (B) EEG power from the recording of above trace was plotted against time, from PTZ administration to the following 30 min. The intensity of high frequency firing has been increasing in the first 10 min, but never reached to HAFDs during 30 min analysis in comparison with PTZ control rat described in Figure 1A. (C) Rate of animal with bursts and HAFDs from the two treating groups. (D,E) The static behavioral seizure scores over the 30 min period from the PTZ alone and PTZ + SL-327 groups illustrated in panel. (F) Proportion of rats with different seizure score. R3 and R5 represent the seizure score of Racine3 and Racine5, respectively. (G) The average of the maximal behavior scores for rats from either PTZ or SL-327 group. * p < 0.05 compared with PTZ group.
down-regulation of KCC2 on the plasma membrane. Enhanced m-calpain activation during seizure induction is due to the over phosphorylation by the MAPK/ERK signaling pathway activity. Inhibition of either calpain or MAPK/ERK activities significantly suppressed seizure induction in vivo. Since our previous study (Chen et al., 2017) and studies of other researchers (Huang et al., 2012;Baek et al., 2016) have already shown that KCC2 is a key protein in the onset of seizure, and its functional down-regulation plays a decisive role in the formation and development of epileptiform neuronal bursting discharges and subsequent epilepsy, our current study may provide a possible mechanism that enhanced m-calpain activation during convulsant stimulation facilitates seizure induction by increasing KCC2 down-regulation.
Our in vivo epileptic model study found that the nonselective calpain blocker MDL-28170 significantly inhibited the onset of seizures in PTZ rat model, which is well in agree with previous reported that inhibition of calpain by MDL-28170 could reduce seizure probability and severity in both kainate acid and pilocarpine induced seizure animal models (Sierra-Paredes et al., 1999;Araujo et al., 2008;Nam et al., 2017). PTZ induced seizure animal model is one of the widely used animal models for epilepsy study and drug screening. Our own previous studies have found that PTZ induced seizure behaviors and epileptiform bursting neuronal activities are relatively simple in its form and easy to be characterized (Qian et al., 2011;Liu et al., 2013). Thus, classic PTZ rat model has been used in this study. However, although PTZ has been widely accepted as a GABA A receptor antagonist, the mechanism of PTZ peripheral injection to induce epilepsy is very complicated, and to our knowledge, it is not well defined so far. PTZ has been reported acts not only as a competitive inhibitor for GABA receptors, but also has been suggested to induce epileptic seizures through other mechanisms. A 1987 study found that PTZ acts at calcium channels, and it causes calcium channels to lose selectivity and conduct sodium ions as well (Papp et al., 1987). One other study pointed out the involvement of cAMP and its downstream effects upon PTZ activity (Hosseini-Zare et al., 2011). In addition, our own unpublished data showed that PTZ, not like other selective GABA A receptor antagonists, failed to evoked epileptiform burst discharges in vitro in hippocampal slices, and was also not able to induce seizure behavior by intracerebroventricular (i.c.v.) injection (data not presented). Thus, we chose PTZ rat model to study the mechanism underlying KCC2 down regulation and GABA receptor dysfunction in current study. Indeed, we also found in current study that, similar as in other seizure animal models, PTZ induced seizure behaviors are closed correlated to the down regulation of the mKCC2 (Huang et al., 2012;Chen et al., 2017). In this study, we found that calpain inhibitor MDL-28170 could also prevent either in vivo PTZ induced or in vitro 0 Mg 2+ induced KCC2 down regulation. Thus, the antiepileptic effect of MDL-28170 may be attributed to the regulation of KCC2, since KCC2 down regulation during convulsant stimulation has been closely associated to the seizure onset (Huang et al., 2012;Baek et al., 2016;Chen et al., 2017). Furthermore, this antiepileptic action of calpain inhibition is likely resulted from blockade of m-calpain rather than from modulation of µ-calpain, since selective m-calpain selective inhibitor Calpain Inhibitor IV (Griscavage et al., 1996;Gellerman et al., 1997;Rosenberger et al., 2005), but not µ-calpain selective inhibitor Calpain Inhibitor I (Griscavage et al., 1996;Gellerman et al., 1997), prevented 0 Mg 2+ induced KCC2 down regulation (see Figures 3A,B,D).
Calpain is a Ca 2+ -dependent cysteine protease that is ubiquitously expressed in mammals (Zimmerman and Schlaepfer, 1984;Ono et al., 2016), and two major subtypes of calpain have different sensitivity to the Ca 2+ concentration with µ-calpain at about 3-50 µM Ca 2+ and, distinguishably, m-calpain at around 400-800 µM, respectively (Liu et al., 2008;Sorimachi et al., 2011a;Zanardelli et al., 2013). It is reported that, at normal condition, intracellular Ca 2+ concentration is at around 50-300 nM and extracellular Ca 2+ concentration is approximately at around 2 mM (Maravall et al., 2000), and pathological stimulation, such as convulsant, may trigger the Ca 2+ influx and the release of Ca 2+ from the endoplasmic reticulum into cytoplasm, to a level of hundreds of nM (Ono et al., 2016). Thus, convulsant stimulation induced intracellular Ca 2+ concentration ([Ca 2+ ] i ) increase to a level which may be sufficient to activate µ-calpain, but may not enough to activate m-calpain. Since our pharmacological results, by using selective inhibitors for either m-or µ-calpain, indicated, m-calpain, but not µ-calpain, is likely to involve in convulsant stimulation induced KCC2 down regulation, it suggests that increases of [Ca 2+ ] i might only contribute minimal effect on calpain mediated KCC2 down regulation in in vitro 0 Mg 2+ model or in vivo PTZ model used in our current study. Meanwhile, seizure stimuli such as 0 Mg 2+ or PTZ would also causes the release of BDNF (Wang et al., 2009;Liu et al., 2013), which activates the MAPK/ERK pathway, in turn, could lead to the phosphorylation of the downstream m-calpain serine site Liu et al., 2016). Our results indeed showed that K252a, a BDNF receptor TrkB receptor blocker, and PD98059, a MAPK/ERK inhibitor, both fully reversed enhanced m-calpain phosphorylation during 0 Mg 2+ stimulation. Thus, our results suggest that MAPK/ERK-mediated m-calpain phosphorylation activation may be responsible for excessive calpain activation to modulate KCC2 expression under convulsant stimulation conditions and causes seizure induction.
The results from our current study indicate that, during seizure induction, activation of m-calpain, but not µ-calpain, promotes KCC2 down regulation. Since KCC2 down regulation is essential for seizure onset (Huang et al., 2012;Baek et al., 2016;Chen et al., 2017), the result that µ-calpain has little effect on KCC2 regulation might suggest that µ-calpain has relatively little contribution to seizure development. In fact, recent studies on different pathological models have revealed that the two main calpain subtypes have different functions in response to pathological stimuli. For example, µ-calpain and m-calpain play different and even opposing roles in TBI, stroke, and neurodegenerative diseases (Puskarjov et al., 2015;Etehadi Moghadam et al., 2018;Wang et al., 2018b). Generally speaking, m-calpain activation promotes neuronal apoptosis and aggravates neurotoxicity, whereas µ-calpain activation provides neuroprotective effects and relieves pathological stimulation Wang et al., 2018a). In present study, we have proved that m-calpain is the dominant participant in seizure onset, by regulating of KCC2 level. Besides m-calpain and µcalpain, whom also called traditional calpain, the other calpain subtypes, which are called non-traditional calpain (Sorimachi et al., 2011b), have been paid little attention of their role in pathological condition. Whether these non-traditional calpain subtypes also contribute to the seizure onset needs to be studied in future.
KCC2 has a high turnover rate at about 50% renewal rates during 10 min period (Lee et al., 2007;Zhao et al., 2008). Under physiological conditions, normal calpain activation sustains the dynamic balance of KCC2 transportation and degradation. By contrast, under pathological conditions, such as seizure onset, calpain is excessively activated to scavenge cKCC2. The over-activation of calpain degrades KCC2 and in turn diminishes the intracellular KCC2 pool, an effect that eventually decreases KCC2 expression in the plasma membrane (Puskarjov et al., 2012(Puskarjov et al., , 2015Zhou et al., 2012;Chamma et al., 2013). The stability of KCC2 on the plasma membrane is affected by the phosphorylation of its ser940 site, which can be dephosphorylated by PP1 (Lee et al., 2007(Lee et al., , 2011Silayeva et al., 2015). In our current study, after the internalization of mKCC2 was prevented by blocking the dephosphorylation of the ser940 site of membrane KCC2 with TMC, our results, however, showed that although the down-regulation of mKCC2 was rescued, the cKCC2 level was still at low level (see Figure 4). It suggested that calpain might still remain in over activation status in functioning of cleavage of KCC2. Indeed, our result further demonstrated that KCC2 down-regulation was completely restored after MAPK/ERK blockade by PD98059. These results indicated that internalized KCC2 will be further degraded by calpain activation during seizure induction. The phosphoractivation of m-calpain may function as a quick response to accumulated cKCC2.
As discussed above, it is likely during convulsant stimulation, that the enhanced m-calpain activation is due to MAPK/ERK phosphorylation of serine-site of m-calpain. We further tested, in vivo, whether inhibition of MAPK/ERK pathway could also suppress seizure induction, as well as interference of KCC2 down regulation. Indeed, SL-327, a brain penetrable ERK inhibitor (Carr et al., 2009), significantly blocked PTZ evoked KCC2 reduction and simultaneously prevented PTZ induced seizure behaviors in rats (see Figures 3, 5). Thus, our results that inhibition of m-calpain, as well as its up-stream regulator MAPK/ERK signaling pathway could block both KCC2 down regulation and seizure induction by convulsant stimulation, indicated MAPK/ERK-m-calpain pathway is one of the important regulatory pathways, by regulating the KCC2 expression level, hence influence the GABA inhibitory efficiency, to modulate seizure onset.
Since calpain is a Ca 2+ -dependent cysteine protease, whether intracellular Ca 2+ concentration change during convulsant stimulation would affect m-calpain over activation is an important issue. Our BAPTA experiment results demonstrated that, in 0 Mg 2+ model, both the KCC2 level and the m-calpain phosphor-serine level were significantly affected by the cleavage of intracellular Ca 2+ , while neither KCC2 level nor phosphorylation of m-calpain induced by BDNF were affected by BAPTA. These data suggested that, during seizure induction, over release of BDNF triggered TrkB-MAPK/ERK signaling pathway in regulation of m-calpain phosphorylation and in turn regulation of cellular KCC2 level is likely independent of [Ca 2+ ] i . However, since m-calpain phosphorylation and cellular KCC2 regulation are likely more complicated regulated with multiple mechanisms, our data from 0 Mg 2+ model also suggested that, at least with some parts, this process is dependent on [Ca 2+ ] i level during seizure induction in 0 Mg 2+ or many other epilepsy models.
Although our current study showed that both non-selective calpain inhibitor MDL-28170 and MAPK/ERK inhibitor SL-327 all significantly inhibited acute epileptic seizures, it is worth paying attention to whether these targets are suitable for epileptic seizure treatment. Since m-calpain and µ-calpain have different or even opposite function (Puskarjov et al., 2015;Etehadi Moghadam et al., 2018;Wang et al., 2018b), the non-selective inhibition of calpain by using non-selective drugs such as MDL-28170 may cause serious side effects. The study in the TBI model on µ-calpain knockout mice showed that although m-calpain was activated in both wild type (WT) and KO mice, the proportion of cell death in µ-calpain KO mice was significantly higher than that of WT, suggesting that the activation of µ-calpain may have certain neuroprotective effects (Wang et al., 2018b). Other studies have also shown that expression of endogenous calpain inhibitor calpastatin to reduce the activation of calpain failed to prevent neuronal apoptosis (Schoch et al., 2013). On the other side, MAPK/ERK signaling pathway involves in the regulation of many proteins, of which calpain is only one of them. Functionally, MAPK signaling pathway is associated with apoptosis, differentiation, migration, proliferation, et cetera (Anderson and Tolkovsky, 1999;Ortega and Alcántara, 2010;Huo et al., 2015;Sun et al., 2015). In CNS, activation of the MAPK/ERK signaling pathway can help the cell survive by promoting neuron anti-oxidation and anti-glutamic toxicity (Zhao et al., 2013;Ortuño-Sahagún et al., 2014), while inhibition of MAPK/ERK signaling pathway induces apoptosis and inhibits cell proliferation (Roy et al., 2010;Anne et al., 2013). Thus, simple inhibition of MAPK/ERK pathway by using their inhibitors such as SL-327 or PD98059 may cause broad effect in the central nervous system. Taken together, suppressing of m-calpain, rather than inhibition of upstream MAPK/ERK activation or broad calpain inhibition, during seizure induction might be a suitable target for future anti-epileptic drug development.
In conclusion, our current study discovered that, during convulsant stimulation, m-calpain, but not µ-calpain, was over activated, which was partly regulated by calcium independent MAPK/ERK signaling pathway. The enhanced m-calpain activation, in turn, down regulated KCC2 expression, and hence facilitated seizure induction (Figure 6). Our results suggest, in conjunction with many other facts, that m-calpain might be a suitable target for anti-epileptic drug development.

AUTHOR CONTRIBUTIONS
LW is the main writer of this manuscript, he carried out in vivo recording, WB and IP experiments, and helped with the experiments design. LR contributed to experiments design and analysis of EEG data. LC carried out the immunostaining part. GW contributed to in vitro experiments and result analysis. XL assisted in the manuscript writing. BW helped with the WB experiments and YW guided the design and implementation of this research.