TNFα-Mediated Necroptosis in BBB Endothelia as a Potential Mechanism of Increased Seizure Susceptibility in Mice Following Systemic Inammation

Systemic inammation is a potent contributor to increased seizure susceptibility. However, less is known about the effects of systemic inammation on blood-brain barrier (BBB) that affect neuron excitability. Necroptosis and inammation are intimately associated in various neurological diseases. We hypothesized that necroptosis is involved in the mechanism underlying sepsis-associated neuronal excitability in BBB components. Systemic inammation was induced by LPS. Seizure susceptibility of mice was measured by kainic acid intraperitoneal injection. Pharmacological inhibitors (C87 and GSK872) were used to block signaling of TNFα receptors and necroptosis. To identify the features of sepsis-associated response in the BBB and CNS, brain tissues of mice were obtained for assays of the necroptosis-related protein expression, and immunouorescence staining for morphological changes of endothelia and glia. Microdialysis assay was also used to evaluate the changes of extracellular potassium and glutamate levels in brain. Signicant ndings including induced increased seizure susceptibility and BBB endothelia necroptosis and leakage, Kir4.1 dysfunction, and microglia activation were observed in mice following LPS injection. Inhibition of TNFa receptor inhibitor C87 signicantly attenuated increased kainic acid-induced seizure susceptibility and endothelia necroptosis and microglia activation, and restored kir4.1 protein expression, compared with those in controls. GSK872 (a RIP3 inhibitor) treatment, like C87, had consistent effects on these changes following LPS. Our results showed that TNFα-mediated necroptosis in BBB endothelia damage contributes to the development of increased seizure susceptibility in mice after systemic inammation. Pharmacologic inhibition targeting this necroptosis pathway may provide a promising therapeutic approach to reduce sepsis-associated BBB dysfunction, astrocyte ion channel dysfunction, and subsequent neuronal excitability.


Introduction
Seizure is a common acute complication of sepsis and systemic in ammation [1][2][3]. Seizure is provoked by the hyper-excitability of circuits within the brain, due to an imbalance between neuronal excitatory and inhibitory activities. Sepsis leading to the excessive production of proin ammatory factors in the peripheral system and brain is believed to be an important factor in the pathogenesis of neuronal hyperexcitation, leading to an enhanced predisposition to seizure, and subsequent changes of neuroplasticity, which may evolve into a chronic seizure syndrome [3][4][5][6]. Our previous study showed that a single intraperitoneal (i.p.) injection of lipopolysaccharide (LPS) in mice causes increased susceptibility to pentylenetetrazole-induced seizures [7]. Similar observations have been made in postnatal and adult animals [3,6], including alterations to the number of neuronal receptors and changes to neuronal plasticity [8]. This evidence indicates that in ammatory processes initially from the peripheral system to brain are a common and crucial mechanism in the pathophysiology of seizures and epilepsy. However, the way in which unbalanced regulation of systemic in ammation contributes to seizure development is still unclear.
The blood-brain barrier (BBB) is a major part of the interface between the peripheral and central nervous systems (CNS). The BBB manages the exchange of blood and compounds between the brain and the circulation and maintains a stable CNS environment. It has long being proposed that dysfunction of the BBB may affect neuronal excitability or ring [3,9,10]. The BBB is composed of endothelial cells, tight junctions, basal lamina, and associated cells, including astrocytic end-feet and pericytes. The continuous non-fenestrated endothelial cells contribute in part to the restrictive permeability and control of leukocyte transmigration of the BBB via their inherent low pinocytic activity and their high concentration of e ux transporters [11]. A number of preclinical studies in vitro and in vivo have shown that in ammatory challenge results in an increase in the permeability of the BBB. The administration of cytokines, such as interleukin (IL)-1, tumor necrosis factor-α (TNFα), and IL-6, increases endothelial permeability [12,13].
Astrocytes also play a critical role in maintaining the homeostasis of the BBB with respect to neuroglial cells. Astrocyte end-feet contain several channel proteins, such as the inwardly rectifying potassium (Kir) channel subunit Kir4.1, one of the integral parts of the orthogonal arrays of particles, which are required to provide optimal BBB properties [11]. Kir 4.1 channels, speci cally expressed in astrocytes, play a key role in controlling spatial K + concentration and regulating extracellular glutamate concentration at tripartite synapses [14,15]. Astrocytes, via channels such as Kir4.1, directly affect neural excitability and have been implicated in the pathogenesis of seizures and the development of epilepsy [14].
Necroptosis is a regulated necrosis, different from apoptosis and necrosis, and is mediated by death receptors such as TNFα receptor 1, receptor-interacting protein kinase (RIP)1, and RIP3, which activate the phosphorylation of mixed lineage kinase domain-like (MLKL) protein, disturbing the integrity of the cells [16,17]. The necroptosis pathway has been reported to serve a vital role in many pathologies that involve in ammatory processes, including sepsis, in ammatory bowel disease, and neurodegenerative diseases [18]. The necroptosis mechanism has attracted considerable research attention because of its importance in endothelial damage and BBB leakage after stroke [19,20]. However, there have been only a few reports that necroptotic cell death signaling occurs in these BBB components following systemic in ammation.
Sepsis and system in ammation induce the excessive production of proin ammatory factors and rapidly corresponding neuroin ammation [7], which may lead to BBB damage and subsequent leakage [13]. Necroptosis and in ammation are intimately associated. Therefore, we hypothesized that necroptosis links systemic in ammation in ammatory mediators such as TNFα, changes in BBB components such as endothelial cell and astrocytes, and BBB dysfunction, contributing to alteration in susceptibility to seizure. By using an LPS-induced systemic in ammation mouse model, we found for role of TNFαmediated necroptosis in BBB changes, including endothelial damage and astrocytic Kir4.1 dysregulation, related to an increase in susceptibility to seizures in mice, following systemic in ammation.

Animals and the LPS-induced systemic in ammation model
Eight-to nine-week-old male C57BL/6J mice were purchased from the National Lab Animal Center (Taiwan). The animals were housed in a speci c pathogen-free room at 21°C under a 12-h light/12-h dark arti cial lighting cycle, with free access to feed. Age-and weight-matched animals were used for the present experiments. All procedures were approved by the Animal Care and Use Committee of Changhua Christian Hospital. The mouse model of systemic in ammation was induced by an i.p. injection of 4 mg/kg LPS (Escherichia coli, strain O111:B4, Calbiochem, San Diego, CA, USA), as previously described [7].
Determination of susceptibility to seizure using kainic acid To assess susceptibility to seizure, mice were injected i.p. with 3 or 20 mg/kg kainic acid (KA) 72 h after treatment with LPS (Sigma, St. Louis, MO, USA) or vehicle (normal saline). KA is a common proconvulsant agent used for the induction of seizures. Seizure activity was recorded on video during an observation period of 2 h after KA injection. Behavioral seizures were scored according to a previously de ned scale [21] as follows: stage 0, no response; stage 1, body position lowering and hypoactivity; stage 2, body automatic shaking, whisker twitching, and sudden muscle or tail contraction; stage 3, repetitive scratching, head bobbing, or circling; stage 4, forelimb clonus and rearing and falling; stage 5, repetitive behavior of stage 4; and stage 6, severe generalized tonic-clonic seizure. In the behavior test, we recorded the severity score according to the aforementioned criterion for each 5 min over a 120-min period.

Plasma levels of TNFα after peritoneal administration of LPS
To assess the acute effects of LPS on systemic in ammation with and without pretreatment with TNFα receptor inhibitor C87 and RIP3 kinase inhibitor GSK872, blood samples were obtained from the cheek of the mice 1 h after the i.p. injection of 4 mg/kg LPS. The plasma samples were stored at −80°C until they were assayed for TNFα concentrations using Duo set kits (R&D Systems, Minneapolis, MN, USA), according to the manufacturer's instructions.
Immuno uorescence images were used to analyze the staining density in the CA3 subregions using ImageJ software (NIH, Bethesda, MD, USA) [22]. The number of activated microglia and reactive astrocytes were counted using a stereological approach from three sections of the CA3 subregion in each hippocampus (n = 3 mice per group), employing an Olympus DP80® Dual CCD Microscope, as previously described [23]. Brie y, images were captured from the CA3 subregion at −2.18 to −2.54 mm from the bregma. The coordinates for the CA3 were taken from the 2.0-to 3.0-mm medial to lateral regions and the 2.0-to 3.2-mm dorsal to ventral regions. Activated microglia were distinguished from resting microglia by their amoeboid appearance and signi cant enlargement. Reactive astrocytes had a hypertrophic morphology distinct from that of resting astrocytes. A single experimenter, who was blind to each animal's treatment, performed the quanti cation of cells of interest using images taken at 200× magni cation in selected rectangular regions.

Brain interstitial uid microdialysis
Eight-to nine-week-old male C57BL/6J mice were used for brain microdialysis assays, using procedures modi ed from previous studies [24,25]. Mice were anesthetized with iso urane for stereotaxic surgery to place a guide cannula (Plastics One, Roanoke, VA, USA) into the right hippocampus (anterior-posterior, −2.3 mm; medial-lateral, 2.0 mm; and dorsal-ventral, −2.0 mm relative to the bregma). One day postoperatively, mice (n = 36) were treated with either vehicle (0.25% DMSO) or 2 mg/kg GSK872, and 1 h later treated with saline or 4 mg/kg LPS. Eighteen of these mice had CMA/12 probes (CMA/12 Elite, lengths 2 mm, CMA, Stockholm, Sweden) inserted 2 h before GSK872 treatment and were placed in a Plexiglas dialysis chamber. Then, the CMA/12 probes were immediately perfused with Ringer's solution (147 mM Na + , 2.2 mM Ca +2 , 4 mM K + , pH 7.0) at a ow rate of 2 µL/min, set to collect 60 µL of dialysate every 30 min for 6 h. Four days post-operatively, 18 more mice that had been treated with GSK872 and LPS had probes inserted into the placed guide cannulas to collect the dialysates once every 30 min for 2 h in each mouse. The extracellular K + concentrations of these dialysates were measured using a ame atomic absorption spectrometer (FAAS, Z6100 Hitachi, Japan). K + levels were calculated from a standard curve prepared from the standard solutions (Merck, Darmstadt, Germany). Glutamate concentrations were measured using an enzymatic colorimetric method using a microdialysis analyzer (CMA/600, Carnegie Medicin, Stockholm, Sweden).

Statistical analysis
The means of two groups were compared using Student's t-tests. The means of more than two groups were compared using one-way ANOVA, followed by Bonferroni post hoc tests. Differences in KA-induced seizure severity among treated mouse groups and extracellular potassium and glutamate changes among treated GSK872 mouse groups were assessed using two-way repeated measures ANOVA, adjusted using Bonferroni post hoc tests. Statistical analysis was performed using the GraphPad Prism version 7 software (GraphPad Software, San Diego, CA, USA, www.graphpad.com). All values are presented as mean ± standard error of the mean (SEM). Differences were considered signi cant at p < 0.05.

LPS-induced systemic in ammation increased susceptibility to KA-induced seizure in mice
To explore the mechanisms underlying the onset and development of seizures following systemic in ammation, our established animal model was used with a single 4-mg/kg dose of LPS via intraperitoneal injection to induce systemic in ammation in 8-to 9-week-old male mice [7]. Seizure susceptibility was subsequently determined by scoring the severity and duration of the KA-induced seizure every 5 min for a 2-h period (Fig. 1A). The mortality rate among the mice was approximately 12% within 3 days after LPS injection (Fig. 1B). During a 2-h period after KA injection, one of 13 saline-treated mice administered 20 mg/kg KA and 2 out of 13 LPS-treated mice administered 3 mg/kg KA died from severe seizures. Among the groups treated with saline and either 3 or 20 mg/kg KA and the group treated with LPS and 3 mg/kg KA, two-way repeated measures ANOVA revealed that the main effect for these three groups yielded an F ratio value of F(2, 900) = 395.89, p < 0.0001 (Fig. 1C), indicating a signi cant difference in susceptibility to KA-induced seizure between these three groups. Bonferroni post hoc tests further revealed that there was a signi cant difference between the 3-mg/kg-KA-saline-treated group and 20-mg/kg-KA-saline-treated group (F(1, 600) = 860.18; p < 0.0001), but no difference between the 20mg/kg-KA-saline-treated group and 3-mg/kg-KA-and LPS-treated groups (F(1, 600) = 1.17, p = 0.285). These results indicated increased susceptibility to 3-mg/kg-KA in LPS-treated mice t was similar to that in saline-and 20-mg/kg-KA-treated mice (Fig. 1C). The latency to initial seizure onset that was de ned as seizure score stage 4 (i.e., tonic with or without clonic convulsion) or more after KA administration was signi cantly decreased in LPS-treated mice given 3 mg/kg KA, compared with that in saline-treated mice given 20 mg/kg KA (Fig. 1D). All saline-treated mice administered 3 mg/kg were not observed to reached tonic with or without clonic convulsion. We measured the duration of stage 4-6 seizure among the mice receiving LPS or vehicle control with KA. The seizure duration was 40.00 ± 5.31 min (mean ± SEM) for the saline-and 20-mg/kg-KA-treated groups and 57.69 ± 6.69 min for the LPS-and 3-mg/kg-KA-treated groups (Fig. 1E). The LPS-treated group had increased susceptibility to 3 mg/kg KA-induced seizures in the latency of initial seizure onset and seizure-behavior duration, compared with the 20-mg/kg-KA-treated mice without LPS injection.

LPS-induced systemic in ammation induced programmed necroptosis and Kir4.1 dysregulation in the hippocampus
We determined the role of several signaling pathways that could be involved in the mechanisms of neuronal hyper-excitability following systemic in ammation, including pathways involved in apoptosis, necroptosis, and ion channel changes [26,27]. Seventy-two hours after injection with vehicle (saline) or 4 mg/kg LPS, the mice were intraperitoneally administered vehicle (H 2 O) or 3 mg/kg KA. Two hours later, the hippocampus was obtained for the assessment of the expression of proteins involved in the apoptosis and necroptosis pathways, as well as astrocytic ion channels (n = 3 mice per group) using Western blotting. Among these four groups of saline-or LPS-treated mice administered vehicle or KA, we found the protein levels of JNK, Bax, and cCaspase 3, which are part of the apoptosis pathway, were signi cantly increased by KA treatment, but no difference was found in mice treated with LPS alone ( Fig.   2A-2D). The protein levels of phosphated RIP3 and phosphated MLKL, from the necroptosis pathway, were increased in LPS-treated mice, enhanced by KA treatment, compared with those in vehicle-treated mice ( Fig. 2A, 2E and 2F). The levels of phosphated MLKL and TNFα were markedly enhanced after treatment with 3 mg/kg KA ( Fig. 2A and 2G). The protein levels of Kir4.1, but not NKCC1, were signi cantly decreased in LPS-treated mice, compared with those in vehicle-treated mice ( Fig. 2H and 2I).
Inhibition of either TNFα receptor or RIP3 attenuated the increased susceptibility to KA-induced seizure in mice following LPS injection TNFα is rapidly released and is one of the most abundant mediators of in ammation in the peripheral blood after infection or exposure to LPS [7]. TNFα has been implicated in the pathogenesis of several in ammation-related diseases, such as vascular leaks, via different signaling pathways [28]. On the basis of the aforementioned evidence (Fig. 2), TNFα-dependent necroptosis appeared to be involved in the increase of LPS-induced susceptibility to seizure. Therefore, the TNFα receptor inhibitor, C87, and the RIP3 inhibitor, GSK872, were used to explore the role of TNFα-dependent necroptosis on LPS-associated susceptibility to seizure. Mice were injected intraperitoneally with 2 doses of CS87 (2 mg/kg, i.p.) at 1 and 24 h, or one dose of GSK872 (2 mg/kg, i.p.) at 1 h, before LPS injection (Fig. 3A). The mortality rate of the mice was approximately 12% within 3 days after the administration of LPS only, and no deaths were recorded in mice receiving either C87 or GSK872 with LPS treatment. Seventy-two hours after LPS injection, 3 mg/kg KA was administered to evaluate the susceptibility to seizure of these mice (n = 7-10 mice per group) by scoring once every 5 min for 2 h. During a 2-h period after KA treatment, one of the 10 LPS-treated mice given KA died because of severe seizures; no death was observed in the LPS-treated groups given C87 (n = 10), or GSK872 (n = 10), as well as in saline-treated group (n = 7). Two-way repeated measures ANOVA revealed that the main effect for these four groups yielded an F ratio of F(3, 825) = 192.65, p < 0.0001 (Fig. 3B). Bonferroni post-tests analysis further revealed that there was a signi cant difference between the C87-treated group (F(1, 450) = 273.49, p < 0.0001) and the GSK872treated group (F(1, 450) = 117.60, p < 0.0001) from vehicle-treated mice following LPS injection. These results indicated that either C87 treatment or GSK872 treatment attenuated the susceptibility to KAinduced seizure following LPS injection. There was a signi cant difference between C87-treated mice and GSK872-treated mice (F(1, 450) = 25.56, p < 0.0001), indicating that C87 treatment was better than GSK872 treatment for decreasing seizure susceptibility in the LPS-treated mice. The latency to initial seizure onset (Fig. 3C) and seizure duration during a 2-h period (Fig. 3D) after KA administration (stage 4 or more) were signi cantly attenuated by either C87 or GSK872 treatment in LPS-treated mice, compared with those in vehicle-treated mice. Half of the mice treated with C87 and GSK872 were not observed to have tonic with or without clonic seizure (i.e., stage 4).
Inhibition of TNFα receptor or RIP3-attenuated BBB leakage, endothelial necroptosis, and astrocytic Kir4.1 down-regulation in the hippocampus of mice treated with LPS To investigate the effect of the TNFα-dependent necroptosis signal pathway on the integrity of the BBB following LPS injection, we examined the effects of C87 and GSK872 treatment on BBB leakage and neuroin ammation. Mice were dosed with C87 at 24 and 1 h before LPS injection, or one dose of GSK872 at 1 h before LPS, or vehicle and then were sacri ced at 72 h after LPS treatment. Hippocampus sections were prepared for immunostaining with Iba-1 antibody for microglia, CD68 antibody for monocytes, GFAP antibody for astrocytes, and Kir4.1 antibody for channel proteins. Activated microglia were identi ed by their increased cell size and irregular shape. The percentage of activated microglia (Fig. 4A and 4B) and leaked monocytes (Fig. 4C) in the CA3 regions of the hippocampus was signi cantly increased in the LPS-treated group, which was attenuated by pretreatment with either C87 or GSK872. Reactive astrocytes indicate the presence of hypertrophic morphology. The percentage of active astrocytes was signi cantly higher in the LPS-treated mice than in the C87-or GSK-872-treated mice (Fig. 4D and 4E). The proportion of GFAP-positive cells co-localized with Kir4.1 was constant in the control and KA-treated mice but markedly decreased in the LPS-treated mice (Fig. 4F). For determining the MLKL activity of BBB endothelial cells, immunostaining was performed using CD31 antibody for the endothelial cells and p-MLKL antibody for the bioactivity (Fig. 4G). The proportion of CD31-positive cells co-localized with p-MLKL was lower in control mice, but markedly increased in LPS-treated mice (Fig. 4H). C87 and GSK872 pretreatment of LPS-treated mice reversed the phenomenon of enhanced p-MLKL-positive staining in BBB endothelial cells and decreased Kir4.1-positive staining in astrocytes.

GSK872 suppressed RIP3-mediated necroptosis and restored Kir4.1 protein expression in mice within 3 days after LPS injection
We examined the effects of GSK872 on dynamic changes in RIP3-mediated necroptosis and Kir4.1 protein within 3 days after induction by LPS. Mice were pretreated with GSK872 (2 mg/kg, i.p.) or vehicle (0.25% DMSO) 1 h before LPS was given (Fig. 5A). The protein levels of the hippocampus were measured at 6, 48, and 72 h after LPS injection (n = 3 per group for each time point). The protein levels of p-RIP3 and p-MLKL were signi cantly higher (Fig. 5B-5D) and Kir4.1 protein levels were signi cantly lower than those given vehicle ( Fig. 5B and 5E) at these time points within 3 days after LPS treatment. GSK872 treatment improved the RIP3-mediated necroptosis and the parallel decrease in expression of Kir4.1ion channel proteins.

GSK872 attenuated increased levels of extracellular potassium and glutamate in the hippocampus within 3 days after LPS injection
Microdialysis was used for continuous measurement of free and unbound analyte levels in the extracellular uid in mice treated with GSK872 and LPS. One day after the cannula was implanted in the hippocampus, the mice were treated with GSK872 (2 mg/kg, i.p.) or vehicle (0.25% DMSO, i.p.) (n = 6 per group) 1 h before the administration of either saline or 4 mg/kg LPS (Fig. 6A). The mean of the rst three samples immediately before LPS administration was de ned as the basal levels (100%) of extracellular potassium and glutamate in the hippocampus. Within 5 h after GSK872 and LPS injections, we found that the concentrations of extracellular potassium and glutamate had increased ( Fig. 6B and 6C). Twoway repeated measures ANOVA revealed that the main effect for these three groups yielded an F ratio of F(2, 165) = 28.66, p < 0.0001 on potassium levels (Fig. 5B) and F(2, 165) = 24.31, p < 0.0001 on glutamate levels (Fig. 6C). Bonferroni post-test analysis further revealed that there was a signi cant difference in potassium (F(1, 110) = 25.87, p < 0.0001) and glutamate levels (F(1, 110) = 12.59, p = 0.0006) in the GSK872-treated group compared with that in the vehicle-treated group following LPS injection. These ndings indicated that GSK872 treatment attenuated the changes in the levels of these extracellular molecules induced by LPS. Microdialysis experiments were also performed in treated mice (n = 6 per group) 72 h after LPS injection, with samples collected once every 30 min for 2 h from each mouse after 1 h stabilization. The mean levels of extracellular potassium (Fig. 6D) and glutamate (Fig. 6E) were signi cantly higher in the LPS-treated group than in the vehicle-treated group and GSK872-treated group. These results indicated that GSK872 treatment attenuated LPS-induced aggravated potassium and glutamate changes within 3 days after LPS injection.

Discussion
In this study, we consistently observed that sepsis and systemic in ammation is linked to increased seizure susceptibility into corresponding neuro-in ammation such as the activation of microglia and astroglia in the brain [3][4][5]7]. We also demonstrated that the TNFα-dependent necroptosis signal pathway governed the changes to the integrity of the BBB, such as endothelial cell damage and astrocytic ion channel Kir4.1 dysregulation following systemic in ammation, leading to neuronal hyper-excitability and the induction of seizures.
Our previous study showed increased susceptibility to seizure induced by using the proconvulsant pentylenetetrazole in mice following LPS injection [7]. This phenomenon was further observed that the susceptibility to low-dose (3 mg/kg, i.p.) KA-induced seizure in mice with LPS injection was similar but more severe in initial latency of seizure onset and tonic-clonic seizure duration than those in mice injected with a high dose (20 mg/kg, i.p.) of KA with saline injection (Fig. 1). These results indicate that systemic in ammation increased neuronal excitability, which in turn reduced the threshold at which seizures are initiated by proconvulsants, or induced the onset of seizures. Besides sepsis-induced neuroin ammation and the subsequent production in the brain of proconvulsive cytokines such as TNFα and IL-1β, which may be involved in neuronal activity changes [7], the results of this study suggest that dysregulation of the ion channel Kir4.1 of astrocytes may play an important role in the underlying mechanism for changes in seizure threshold in LPS-treated mice. The inwardly rectifying K + channels, Kir4.1, are enriched on the processes of astrocytes surrounding the synapses and blood vessels of the BBB but not in neurons and oligodendrocytes in the brain [29]. In neuronal excitation, the astrocytic Kir4.1 channels play a major role in extracellular potassium (K + ) buffering to maintain the homeostasis of the neuronal microenvironment [30]. Diminished Kir4.1 buffering capabilities, such as pharmacological or genetic inhibition and downregulation, may induce membrane hypo-polarization coupled to reduced glutamate clearance in astrocytes, leading to neuronal hyper-excitability [30,31]. Increasing evidence strongly suggests that astrocytic Kir4.1 channels are involved in the development of seizure and epilepsy (epileptogenesis) [14,32,33]. The present study found for the rst time that down-regulation of Kir4.1 in astrocytes combined with increase of extracellular potassium and glutamate levels may be involved in the LPS-induced decreased seizure threshold in mice. Both in vitro and in vivo studies suggest that primary mediators of the in ammatory response, such IL-1β, could have an effect on the down-regulation of Kir4.1 transcription and protein expression in astrocytes [34,35]. The present study further demonstrated that Kir4.1 dysregulation in astrocytes produced by systemic in ammation could be restored by the inhibition of the TNFα-mediated necroptosis signaling pathway on BBB endothelial cells.
The BBB helps to regulate the reciprocal periphery-to-brain exchange of molecules and immune cells and maintain a tightly stable microenvironment for the CNS. Dysfunction of the BBB destroys hemostasis, leading to the pathological development of seizure disorders and other neurological disorders [36,37]. During systemic in ammation, the components of the BBB could be changed at histological and/or molecular levels [13]. Endothelial cells are key components of the BBB, and damage to these cells during systemic in ammation may contribute to barrier dysfunction, whereas the other BBB components, including astrocytes, pericytes, and microglia/macrophages, seem to play a little role in the LPS-mediated disruption of the BBB [38]. In the present study, pretreatment with TNFα receptor inhibitor C87 abolished BBB endothelial necroptosis with a parallel change in monocyte leakage and corresponding neuroin ammation, and Kir4.1 dysregulation. By excluding the possibility that C87 treatment might alleviate LPS-induced peripheral TNF-α levels (see Additional le 1), these results strongly indicate that proin ammatory TNFα is a key factor involved in endothelial cell damage in the BBB and changes in the permeability of the BBB in the event of systemic in ammation.
Upon binding to TNFα receptor 1, TNFα triggers a range of signal pathways to regulate bioactivity in various cell types and tissues. TNFα may therefore impact the BBB endothelial cells and other components [28]. Both in vitro and in vivo studies show that TNFα may induce endothelial cell injury and death via the apoptosis and necroptosis pathways [39,40] and disrupt endothelial tight junction barriers via targeting different pathways, such as the NF-κB pathway [41,42], increasing endothelial leakage. The present study found that both inhibition of TNFα receptor by C87 and RIP3 necroptosis by GSK872 signi cantly attenuated the increased seizure susceptibility in LPS-treated mice (Fig. 3), although there were signi cant differences between these two treatments. This phenomenon could be explained by the possibility that TNFα may more broadly target the components of the BBB. RIP3 inhibition might also affect apoptosis in some situations, due to the possibility of cross-talk between necroptosis and apoptosis [43]. Given that our data showed that LPS did not increase the level of cCasp3 in mouse brains 72 h after injection ( Fig. 2A and 2D), the contribution of LPS-triggered apoptotic death of BBB endothelium following necroptosis to the disruption of BBB integrity, if any, could be not signi cant. Therefore, among the signaling pathways involving TNFα, our data suggest that endothelial necroptosis plays a critical role in changes of seizure susceptibility in the context of TNFα-induced in ammatory events, including astrocytic ionic Kir4.1 channels and BBB dysfunction.
TNFα production peaked in the peripheral circulating blood at 1 h after LPS injection [7] and in the brain at approximately 30 h [44]. Evans Blue-measured BBB leakage was observed early, within 6 h after LPS injection (see Additional le 2), which corresponded to the nding of increased RIP3-mediated necroptosis (Fig. 5) and extracellular K + and glutamate within 4 h after LPS injection (Fig. 6). The results support the contention that systemic TNFα is an early and key peripheral proin ammatory factor causing BBB disruption during systemic in ammation but could not completely exclude the effects of the TNFα derived from microglia [19]. In the CNS, the levels of TNFα consistently returned to the basal levels by 72 h after LPS (Fig. 2G). KA treatment could not raise the brain levels of TNFα but rapidly increased TNFα levels (Fig. 2G) [45,46] and enhanced the activity of MLKL in necroptosis in mice 72 h after LPS injection (Fig. 2F). Although the detailed mechanisms of this phenomenon remain incompletely understood, severe seizures may induce brain injury via a process of necroptosis involving MLKL [47]. These results suggest that a seizure threshold low enough to easily induce severe seizure in LPS-treated mice and increased levels of TNFα after KA treatment may play a role in additional MLKL-executed necroptosis in the brain (Fig. 2F).
Necroptosis is a type of programmed cell death with necrosis and is involved in a variety of biological processes, including in ammation and immune responses. Accumulating evidence suggests that the TNFα-mediated necroptotic pathway may be a potential therapeutic target in the treatment of in ammatory diseases [48,49]. Over the past years, several types of inhibitors targeting the kinase activity of necroptosis, such as RIP3, have been reported [49,50]. In the present study, GSK872, an RIP3 inhibitor, reduced the phosphorylation of MLKL and the programmed necrosis of BBB endothelial cells, supporting the suggestion that necroptosis could be a therapeutic target in efforts to prevent BBB damage from systemic in ammation. RIP3 is indispensable in the TNFα-stimulated necroptosis pathway and can also promote non-necroptotic pathways such as in ammation activation and cytokine IL-1β production through stimulation of Toll-like receptors [48]. Therefore, the effects of RIP3 inhibition on improvement of seizure susceptibility related to systemic in ammation could be partially due to nonnecroptotic anti-in ammation, even though MLKL activation in BBB endothelial cells was signi cantly attenuated in our observations. It will be important to further determine which RIP3 function drives the signaling pathways, such as those of in ammation and necroptosis, in each disease condition. In the future, studies exploiting more speci c inhibitors of RIP3 and MLKL kinases may provide crucial insight into the prevention of necroptosis-associated BBB damage and neuroin ammation following sepsis.

Conclusions
Our results showed that TNFα-mediated necroptosis induced BBB endothelial damage and BBB leakage, leading to neuroin ammation and astrocyte Kir4.1 dysregulation and contributing to the development of increased susceptibility to seizure induced by KA in mice, following LPS-induced systemic in ammation.
Pharmacologic inhibition targeting elements of the necroptosis pathway, such as TNFα receptor and RIP3 kinases, reduces BBB endothelial cell damage and BBB dysfunction. This evidence may indicate a promising therapeutic approach to reduce sepsis-associated BBB dysfunction, astrocyte ion channel dysfunction, and subsequent neuronal excitability.

Declarations
Availability of data and materials The datasets supporting the conclusions of this article are included within the article and its Additional les. All material used in this manuscript will be made available to researchers subject to con dentiality.   Bonferroni post hoc test among the groups; *p < 0.05, **p < 0.01, and ***p < 0.001.