Adult male ICR mice (postnatal 8 weeks) were approved by the comparative medicine center of Yangzhou University (Yangzhou, China) and used domestically for 1 week under standard conditions at 25 ± 2°C with a 12-h light/dark cycle and were allowed free access to water and standard chow. All experimental procedures were in accordance with the National Institutes of Health guidelines for the care and use of laboratory animals and the study protocol was approved based on ethical procedures and scientific care by the Yangzhou University-Institutional Animal Care and Use Committee (YIACUC-14-0015). All efforts were made to use the fewest animals possible and to minimize pain caused by the procedures.

A solution of KA and saline was administered into the unilateral ventricle (−1.0, −0.22, and −0.3 mm relative to bregma) of each mouse. Prior to surgery, all mice were anesthetized with 10% chloral hydrate (Aladdin, China) and placed into the stereotaxic apparatus. Subsequently, the animals were randomly divided into three groups (n = 14 each; 7 animals for histological analyses and 7 animals for biochemical analyses) as follows: normal saline group (control-group), 3 μg/kg kainic acid group (3 μg/kg KA-group), 10 μg/kg kainic acid group (10 μg/kg KA-group). Normal saline group (control-group), 3 μg/kg kainic acid group (3 μg/kg KA-group), 10 μg/kg kainic acid group were sacrificed after 3 days. Following the KA injection, the 10 μg/kg KA group was divided into four subgroups (n = 14 in each group; 7 animals used for histological analyses and 7 animals used for biochemical analyses) based on time: 6, 24, 48, and 72 h. After the injection, all animals were returned to their cages where seizure severity was assessed in 10 min intervals for up to 2 h using the modified Racine scale (30, 31): stage 0, normal behavior; stage 1, immobility; stage 2, forelimb and/or tail extension, rigid posture; stage 3, repetitive movements, head bobbing; stage 4, rearing and falling; stage 5, continuous rearing and falling; stage 6, tonic seizure or death. All the injected mice without stage 0 randomly were used in the later experiments.

The immunohistochemical analyses were performed as described previously (33). The sections were sequentially treated with 0.3% hydrogen peroxide (H₂O₂) in PBS for 20 min and 5% normal serum in 0.01 M PBS for 30 min. The sections were next incubated with diluted rabbit anti-NeuN (1:1,000, Cell Signaling Technology), rabbit anti-GFAP (1:500, Abcam), rabbit anti-ZO-1(1:500, Thermo Fisher Scientific), rabbit anti-GLUT-1 (1:500, Abcam), goat anti-IBA-1 (1:500, Abcam), and mouse anti-PECAM-1 (1:500, Bio-Techne), overnight at 4°C. Subsequently, the sections were exposed to biotinylated goat anti-rabbit, goat anti-mouse or rabbit anti-goat IgG (1:250, Vector, Burlingame, CA). And they were visualized with 3, 3'-diaminobenzidine tetrahydrochloride in 0.01M PBS, and then mounted on Adhesion Microscope slides. Following the dehydration procedure, the sections were mounted in neutral balsam (Solarbio; Beijing, China). To establish the specificity of immunostaining, a negative control experiment was performed with pre-immune serum rather than primary antibody. Negative controls are no immunoreactivity in all structures. Digital images of the hippocampal subregions were captured with an image-analyzing system equipped with a computer-based microscope (Nikon Corporation; Tokyo, Japan). Structures immunoreactive for NeuN, GFAP, Iba-1, PECAM-1, and/or ZO-1 were increased × 20^*10 for analysis and the figure of the NeuN immunohistochemistry was by enlarged × 4^*10. The staining intensities of the structures immunoreactive for GFAP, Iba-1, PECAM-1, and ZO-1 were evaluated using OD, as described above (34).

For Western blot analyses, the mice (n = 7 animals at each time) were sacrificed, the hippocampus was removed and dissected with a surgical blade, and the tissue samples were serially and transversely cut (400 μm) using a vibratome. The samples were preprocessed using a Total Protein Extraction Kit (KeyGEN; Nanjing, China), as previously described (35). Protein concentrations were determined using a Pierce BCA Protein Assay Kit (Thermo Scientific). Using 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), equal amounts of protein (30 μg) were separated and transferred to nitrocellulose membranes (Millipore; Bedford, MA, USA), which were stripped and used for incubating the antibodies. To reduce background staining, the membranes were incubated with 5% bovine serum albumin (BSA) in tris-buffered saline (TBS) containing 0.1% Tween 20 (TBS-T) for 60 min followed by incubation with rabbit anti-ZO-1 (1:1,000, Thermo Fisher Scientific), Claudin 5 (1:1,000, Abcam), LC3A/B (1:1,000, Cell Signaling Technology), Beclin-1 (1:1,000, Cell Signaling Technology), and β-actin (1:3,000, Arigo) overnight at 4°C. Subsequently, the membranes were exposed to secondary goat anti-rabbit IgG (Santa Cruz, USA) for 2 h at room temperature and SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific; Rockford, USA) was used for protein detection. The results of the Western blot analyses were assessed and densitometric analyses for the quantification of the bands were performed using Quantity One Analysis Software (Bio-Rad). These data were used to count the ROD; the ROD ratio was calibrated as a percentage with the control group designated as 100%. Each blot represents at least three similar independent experiments.

All data are presented as a mean ± standard error of the mean (SEM). Differences in the means among the groups were statistically analyzed with one-way analysis of variance (ANOVA) tests followed by Duncan's post-hoc tests. P-values < 0.05 were considered to indicate statistical significance.

The seizure severity of mice was evaluated according to Racine's scale after KA injection (Figure 1). The scale in the control group were 0 that indicate no significantly seizure in mice, and in the 3 μg/kg KA-group, the scale were not significantly different with control group (about 1.5). The scores (about 4.5) in the 10 μg/kg KA-group were significantly higher than that in the control group (p < 0.05). These results indicate that the seizures in the 10 μg/kg KA group were more severe than those in the 3 μg/kg KA group and the control group.

Neuronal nuclei stained with NeuN were observed in the hippocampal region ipsilateral to the site of lesion (Figure 2 and Supplementary Figure 1). In the control group, many NeuN immunoreactive⁽⁺⁾ cells were observed in the all of regions in the hippocampus. The immunoreactivity of NeuN⁺ cells in the 3 μg/kg KA-group were almost equal with that in the control group (p > 0.05) (Figures 2A,B,D). After treatment for 72 h, the immunoreactivity and NeuN⁺ cells in the 10 μg/kg KA-group were significantly less than those in the control and 3 μg/kg KA-group, especially in the unilateral hippocampal CA3 region (p < 0.05) (Figures 2A–D). There were no other significant changes in the hippocampal neurons in any other region. Taken together, these findings suggest that the loss of neuronal cells caused by 10 μg/kg KA is more serious than the control group and 3 μg/kg KA-group.

In the control and 3 μg/kg KA-group, no FJ-B positive⁽⁺⁾ neurons were observed in all hippocampal subregion (Figures 3A,B). After 72 h treatment, in the 10 μg/kg KA-group, many FJ-B ⁽⁺⁾ neurons could be visualized in the unilateral hippocampal CA3 region (Figures 3C,D). The number of necrotic cells stained by FJB in the 10 μg/kg KA-group is more than other groups (p < 0.05). These data provide additional evidence that Neuronal death in the 10 μg/kg KA-group is more serious than the control group and 3 μg/kg KA-group.

To examine whether KA-induced neuronal injury in the hippocampus is likely associated with autophagic cell death, we test early chronological changes of Beclin-1 and LC3II levels. These results of western blot for the Beclin-1 and LC3II/LC3I expression levels in the hippocampus are shown in Figure 4. The expression of Beclin-1 was gradually increasing from control, 6 to 48 h after KA injection (p < 0.05) (Figures 4A,B). However, there was a small decrease at 72 h after the KA injection compared to 48 h after the KA injection. LC3II/LC3I levels in the hippocampus exhibited an increase at 6 h after the KA injection compared to the control group and its expression were peaked at 24 h after KA injection (p < 0.05); its expressions gradually decreased at 48 and 72 h after KA injection (p < 0.05) (Figures 4B,C). The present findings further indicate that KA-induced neuronal injury in the hippocampus was correlated with autophagic cell death.

Iba-1 is widely used as a specific marker for microglia in the CNS and many studies have reported microglial activation and subsequent upregulation of Iba-1 during CNS immune responses (36). In the present study, microglial activation was primarily detected in the hippocampal CA3 region (Figure 5 and Supplementary Figure 2). In the control group, Iba-1 immunoreactive microglia in the hippocampal CA3 region that exhibited characteristics of ramified or resting forms with fine processes in a web-like network. The degree of Iba-1 immunoreactivity and the number of activated microglia cells in the 10 μg/kg KA-group significantly increased in the hippocampal CA3 region compared to the control and 3 μg/kg KA-groups (p < 0.05) (Figures 5A,C). These data illustrate 10 μg/kg KA induced activated microglia in the hippocampal CA3 region.

The increased GFAP expression is widely used as a marker for astrogliosis. In the present study, we used the immunohistochemical analyses of GFAP to reveal differences in astrocyte activation between the control and KA-treated groups (Figure 6). In the control group, GFAP ⁽⁺⁾ astrocytes were distributed throughout the hippocampal CA3 region and showed a resting form with a small body and thin thread-like processes (Figure 6A). In the 3KA- and 10KA-groups, the activation and immunoreactivity of GFAP⁺ cells significantly increased compared to that in the control group (p < 0.05) (Figures 6B–D). We found that a certain dose of KA (10 μg/kg) injection could induce activated astrocytes in the hippocampal CA3 region.

In the control group, PECAM-1 immunoreactivity was detected in the hippocampal CA3 region (Figure 7A). Our result showed that its expression at 6 h after KA injection was non-significant change in the hippocampal CA3 region compared to that in the control group (p > 0.05) (Figure 7B). PECAM-1 immunoreactivity was peaky increased in the hippocampal CA3 region at 24 and 48 h after KA injection(p < 0.05) (Figures 7C,D,F). At 72 h after KA injection, its immunoreactivity had somewhat decreased compared to that at 48 h, however, it remained higher than that of the control group (p < 0.05) (Figures 7E,F). We noted a significant increase in PECAM-1 immunoreactivity in the hippocampal CA3 region beginning at 24 h after injection of 10 μg/kg KA (p < 0.05).

To examine whether TJ-associated proteins, such as ZO-1, are associated with neuronal injury in the hippocampal CA3 region, we detected the level of ZO-1 by immunohistochemistry (Figure 8 and Supplementary Figure 3). In the control group and 6 h after KA injection, weak ZO-1 immunoreactivity was detected in the hippocampal CA3 region (p < 0.05) (Figures 8A,B). At 24, 48, and 72 h after injection, their immunoreactivities were significantly higher than those in the control group in the hippocampal CA3 region, however, there are not exhibit differences in ZO-1 immunoreactivity in the hippocampal CA3 region among them (p < 0.05) (Figures 8C–F).

The Western blot results of the ZO-1 and Claudin-5 levels in the control group and KA-injected groups are presented in Figure 9. Claudin-5 is a 23 kilodalton four-transmembrane protein known to form TJs between ECs at the BBB (37). There was no obvious difference in the expression level of Claudin-5 between the control group and the other group at 6 h after injection (p > 0.05) (Figures 9A,B). However, the expressions of Claudin-5 at 24, 48, and 72 h after KA injection were significantly higher than that in the control-group (p < 0.05) (Figures 9A,B).

ZO-1 links prejunctional F-actin and other cytoskeletal components that are highly associated with cytoskeleton and cell barrier function (38). ZO-1 protein levels in the hippocampus gradually increased from 6 h to 72 after injection (p < 0.05) (Figures 9A,C). Based on the above data, we found that the early transient increases in the expressions ZO-1, and Claudin-5 may be closely related to neuronal injury.

The KA-induced epilepsy model has been widely used to investigate the mechanisms associated with epilepsy (40, 41). Some studies have shown that the neuronal injury caused by KA injections in the acute phase are the most serious (42) although autophagy has been induced in a variety of experimental epilepsy models, including KA- and pilocarpine-induced seizure models (43). Under normal conditions, autophagy is a salient cellular process that can either promote cell survival or cell death and is in dynamic equilibrium. Under pathological conditions, autophagosomes fuse with lysosomes to induce autophagic degradation, as in epilepsy, which can result in cell death via excessive autophagy. We observed significant chronological changes in neuronal death in the hippocampal CA3 region following an intracerebroventricular injection of 10 μg/kg KA as evidenced by NeuN immunohistochemistry and FJ-B staining. We also investigated whether neuronal death induced by KA injections would be associated with autophagic cell death. Beclin-1 and LC3II are commonly used markers of autophagy (44), and in the present study, the protein levels of Beclin-1 and LC3II/LC3I in the hippocampus gradually increased in a chronological manner until 48 h after the KA injection. When cells undergo autophagy, LC3I is converted into LC3II, which in turn is present on autophagosomal precursors and the autophagosomal membrane; thus, it is a specific marker of autophagosomes (13). Beclin-1 is an important regulatory gene for autophagy and has been extensively studied. This protein plays a role in the core complex with class III PI3K (45) and p150 (46) as a promoter of autophagy (47), which is affected in a variety of common neurodegenerative diseases and hereditary peripheral nervous system diseases (48). Overactivation of autophagy leads to cell death via mitochondrial degeneration and chromatin breakage (49) and KA-induced neuronal injury may be related to the activation of autophagy in the mouse hippocampus (50). The present findings further indicate that KA-induced neuronal injury in the hippocampus is likely associated with early chronological changes in Beclin-1 and LC3II levels, which are correlated with autophagic cell death.

Many studies have demonstrated that BBB leakage in experimental epileptic mice is coincident with morphological changes in BBB-related cells, including astrocytes and ECs (51). Furthermore, BMVEC injuries induced by diseases of the CNS, including cerebral ischemia and epilepsy, occur during the initial phase of BBB disruption and result in a poor prognosis for patients. In both the pilocarpine- and KA-induced epilepsy models, BBB damage is observed in all regions of the hippocampus and brain (52). In addition, BBB damage is closely related to the extent and duration of seizures and may promote the occurrence of epilepsy (53). Serum levels of albumin are significantly elevated after BBB damage and can induce epilepsy by activating inflammatory cytokines, decreasing the epilepsy threshold, and causing gap junction coupling (54, 55). We found that a certain dose of KA (10 μg/kg) activated astrocytes and microglia in the hippocampal CA3 region. Previous studies have shown that KA-induced epilepsy models are characterized by the activation of astrocytes and microglia (56) and that the proliferation and activation of glial cells in hippocampus are closely related to epilepsy induced-neuronal death (57). Astrocytes are crucial for maintaining cell architecture and homeostasis under normal brain conditions. In fact, some studies have proposed that “reactive astrocytosis” involves morphological changes and functional changes in astrocytes in response to brain injury (58, 59). Astrocytes are an important component of the BBB and reactive astrogliosis is highly associated with BBB failure (60), which is a pathophysiological process that results from neuronal injury during epilepsy (61, 62). One possible cause of BBB leakage in experimental epilepsy models may be related to morphological changes in astrocytes that can damage the functional BBB structure (55, 63). Excessive microglial activation may also contribute to the pathophysiological process of epilepsy (64, 65) because pathophysiological changes in the brain that are induced by activated microglia, such as inflammation, can compromise barrier function and kill neurons in the BBB (66). Thus, it is likely that the astrocytic and microglial activations observed in the present KA-induced experimental epilepsy model were related to neuronal death and BBB failure.

PECAM-1, which plays a vital role in the migration of leukocytes through ECs, is a common marker of microvascular ECs (67). We noted a significant increase in PECAM-1 immunoreactivity in the hippocampal CA3 region beginning at 24 h after the injection of 10 μg/kg KA. Significant increases in PECAM-1 expression have been observed in a spinal cord ischemia model, focal cerebral ischemia model, and other peripheral ischemia models and may be due to neutrophil migration (68). PECAM-1 plays an important role in maintaining the integrity of the EC connection and modulating its connection stability (20). PECAM-1 inhibits the activation of circulating platelets and supports the integrity of EC-cell junctions, which protects the vascular bed against some types of stimulation, and also appears to play a significant role in these interrelated processes (20, 69). Furthermore, PECAM-1 endows the vascular endothelium with the ability to maintain vascular integrity (70). Under pathological conditions, early and transient increases in PECAM-1 may be associated with the response to stimuli such as ischemia and inflammation (71). Similarly, the transient endogenous increase in antioxidants, including superoxide dismutase (SOD), after ischemia reperfusion protects against oxygen free radicals induced by cerebral ischemia (72). In TLE models, there is an overexpression of vascular endothelial factor (VEGF) and early BBB failure during seizures, which is followed by gradual increases in angiogenesis that may be involved in TLE-induced neuronal death (73). Thus, it is possible to speculate that the increased early PECAM-1 expression observed in the hippocampal CA3 region after the KA injection in the present study was related to the epilepsy-induced neuronal death.

TJs are composed of claudins and TJ-associated marvel proteins (TAMPs), such as occludin and tricellulin, that are present to varying degrees in different EC beds, particularly those that require the tight regulation of vascular permeability, including the BBB (74). We observed significant increases in the immunoreactivity of ZO-1 and Claudin-5 in the hippocampal CA3 region at 24 h after the 10 μg/kg KA injection. The ZO-1 and Claudin-5 proteins are indispensable components of the BBB system (75, 76). In addition, the expression of ZO-1 maintains normal functioning in TJs and improves paracellular transport between ECs and endothelial monolayers (77, 78). Dysfunction in TJ-associated proteins occurs in a number of CNS diseases. For example, neurodegenerative diseases such as stroke and Alzheimer's disease are closely related to changes in TJ-related proteins (79–81), and experimental studies have shown that the expressions of ZO-1 and Claudin-5 exhibit long-term decreases during seizures (73, 82). Our results indicate that transient early increases in the expressions of TJ-associated proteins, such as ZO-1 and Claudin-5, and changes in PECAM-1 expression may be associated with neuronal injury in the hippocampal CA3 region following injections of KA.

Our findings indicate that KA-induced neuronal death in the hippocampal CA3 region is related to autophagy. In addition, the overexpression of astrocyte activation and early transient increases in the expressions PECAM-1, ZO-1, and Claudin-5 may be closely related to neuronal injury (Figure 10). However, we only assessed changes in autophagy, the BBB, and neuronal injury during the acute phase of epilepsy. Further studies are necessary to verify the roles of autophagy and BBB damage in neuronal cell death using drugs. These results will provide novel therapeutic targets for the treatment of epilepsy.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.