HMGB1 Contributes to the Expression of P-Glycoprotein in Mouse Epileptic Brain through Toll-Like Receptor 4 and Receptor for Advanced Glycation End Products

The objective of the present study was to investigate the role of high-mobility group box-1 (HMGB1) in the seizure-induced P-glycoprotein (P-gp) overexpression and the underlying mechanism. Kainic acid (KA)-induced mouse seizure model was used for in vivo experiments. Male C57BL/6 mice were divided into four groups: normal saline control (NS) group, KA-induced epileptic seizure (EP) group, and EP group pretreated with HMGB1 (EP+HMGB1 group) or BoxA (HMGB1 antagonist, EP+BoxA group). Compared to the NS group, increased levels of HMGB1 and P-gp in the brain were observed in the EP group. Injection of HMGB1 before the induction of KA further increased the expression of P-gp while pre-treatment with BoxA abolished this up-regulation. Next, the regulatory role of HMGB1 and its potential involved signal pathways were investigated in mouse microvascular endothelial bEnd.3 cells in vitro. Cells were treated with HMGB1, HMGB1 plus lipopolysaccharide from Rhodobacter sphaeroides (LPS-RS) [toll-like receptor 4 (TLR4) antagonist], HMGB1 plus FPS-ZM1 [receptor for advanced glycation end products (RAGE) inhibitor], HMGB1 plus SN50 [nuclear factor-kappa B (NF-κB) inhibitor], or vehicle. Treatment with HMGB1 increased the expression levels of P-gp, TLR4, RAGE and the activation of NF-κB in bEnd.3 cells. These effects were inhibited by the pre-treatment with either LPS-RS or FPS-ZM1, and were abolished by the pre-treatment of SN50 or a combination treatment of both LPS-RS and FPS-ZM1. Luciferase reporter assays showed that exogenous expression of NF-κB p65 increased the promoter activity of multidrug resistance 1a (P-gp-encoding gene) in endothelial cells. These data indicate that HMGB1 contributes to the overexpression of P-gp in mouse epileptic brain tissues via activation of TLR4/RAGE receptors and the downstream transcription factor NF-κB in brain microvascular endothelial cells.


Introduction
Epilepsy is a chronic and devastating neurological disorder characterized by recurrent unprovoked seizures. A substantial proportion (~30%) of patients with epilepsy is refractory to carefully optimized pharmacological treatment [1]. The overexpression of P-glycoprotein (P-gp) induced by seizure activity [2,3] has been considered to play an important role in the development of drug-refractory epilepsy [4,5]. However, the precise mechanism underlying the seizure-induced overexpression of P-gp remains elusive [6]. P-gp is an efflux transporter protein encoded by multidrug resistance1 (MDR1) in humans and mdr1a (mainly expressed in brain vascular endothelium) and mdr1b (mainly expressed in brain parenchyma) in rodents [7,8]. It has been documented that the increased level and activity of P-gp on the blood-brain barrier (BBB) were associated with the inflammatory process in epileptic brain. Bauer et al. [9] reported that the level of expression of P-gp was increased by extracellular glutamate through N-methyl-D-aspartate (NMDA)/cyclooxygenase-2 (COX-2) pathway. Inflammatory mediator tumor necrosis factor-alpha (TNF-α) was also reported to enhance the activity of P-gp in BBB [10]. Recently, Yin et al. [11] reported that extracellular inflammatory molecule high-mobility group box-1 (HMGB1) may promote drug resistance by upregulating the expression of P-gp in human gastric adenocarcinoma cells. HMGB1-mediated inflammatory pathways have been verified to be activated in many seizure animal models and could initiate and expand inflammation in epileptic tissue [12][13][14]. The increase of HMGB1 in epileptic brain was observed between 1 h and 3 h after the onset of seizures [15], and the progressive up-regulation of P-gp often occurred at 3-24 h after kainic acid (KA)-induced seizures [16,17]. Taking together, we hypothesize that HMGB1 may be responsible for the upregulated expression of P-gp in the epileptic brain.
HMGB1, a nuclear chromatin protein, is ubiquitously expressed in all cells, and it obtains a new identity to act as a damage-associated molecular pattern (DAMP) when placed extracellularly [18]. During the pathogenesis of a number of inflammatory, autoimmune diseases and cancers, HMGB1 could play multiple roles and mediate processes ranging from inflammation to repair as well as drug resistance [19]. Toll-like receptor 4 (TLR4) and receptor for advanced glycation end products (RAGE) are both the best characterized receptors identified for HMGB1. Additionally, the two receptors are constitutively expressed by many cell types, and they can be rapidly upregulated upon interaction with their ligands. TLR4 is a member of TLRs, a group of innate immune system receptors that react to pathogen-associated molecular patterns and DAMPs, and mediate many cell responses including inflammation, innate and adaptive immune responses [20]. Activation of TLR4 by HMGB1 in neurons and astrocytes has been proposed as a critical event for decreasing seizure threshold and initiating brain inflammation [15]. RAGE, like TLR4, is a transmembrane receptor playing key roles in innate immunity activation and inflammatory processes [21]. Iori et al. [22] have suggested that RAGE induced in neurons, astrocytes and microvessels by epileptic activity contributes to hyperexcitability underlying seizures, as well as to the proictogenic effects of HMGB1. Nuclear factor-kappa B (NF-κB), a pivotal regulator of immune and inflammatory response, is one of the most important downstream transduction molecules in both the TLR4 and RAGE signaling pathway [20,21]. In cytosol, NF-κB presents as an inactive form due to the combination of inhibitory kappa B (IκB) to the active subunits P50 and P65 [23]. Once activated, NF-κB can translocate to the nucleus and bind to the promoter region of target genes, including previously reported rat mdr1 gene, therefore regulating the expression of these genes [24]. Emerging evidence suggests that NF-κB is activated during the neuropathological processes of seizure and epilepsy [25,26].
In the present study, the effect of HMGB1 on the expression of P-gp was investigated both in vivo using a KA-induced mouse seizure model and in vitro using a mouse microvascular endothelial cell line, bEnd.3. TLR4 antagonist, lipopolysaccharide from Rhodobacter sphaeroides (LPS-RS) [27], RAGE inhibitor, FPS-ZM1 [28], and NF-κB inhibitor, SN50 (amino acid sequence AAVALLPAVLLALLAPVQRKRQKLMP) [29], were used to investigate the involvement of different signal pathways. Our results indicate that HMGB1 increases the expression of P-gp in mouse epileptic brain tissues via activation of TLR4, RAGE, and the downstream NF-κB pathway in brain microvascular endothelial cells. These data provide an insight into the involvement of HMGB1 in seizure-induced overexpression of P-gp.

Animals and ethics statement
Adult male C57BL/6 mice weighing 20±2 g were purchased from Slack Shanghai Laboratory Animals (Shanghai, China) [License Number: 2013-0005]. Following arrival, animals were housed for 2 weeks in a room with a fixed 12 h light-dark cycle and a constant temperature (20±2°C) and humidity (50-60%). All procedures conformed to the international Guide for the Care and Use of Laboratory Animals [30] and were approved by the Biological Research Ethics Committee of Nanjing Medical University (Permit Number: 20110354). All surgery was performed under chloral hydrate anesthesia, and all efforts were made to minimize the suffering.

KA-induced mouse epileptic seizure model and treatment with HMGB1
Forty-seven mice were randomly divided into four groups, (1) normal saline control group (NS group, n = 11), (2) KA-induced epileptic seizure group (EP group, n = 12), (3) EP group pretreated with recombinant human HMGB1 (EP+HMGB1 group, n = 12), and (4) EP group pretreated with BoxA (a fragment of HMGB1 with antagonistic activity; EP+BoxA group, n = 12). For the mice in EP, EP+HMGB1, and EP+BoxA groups, seizure was induced by the stereotactic injection of KA (7 ng in 0.5 μL; Sigma, Louis, MO) into the models' left hippocampus area (1.8 mm posterior to the bregma, 1.5 mm left lateral from the midline, and 2.0 mm deep from the dura) according to the atlas of Franklin and Paxinos [31]. KA was dissolved in 0.1 M phosphate-buffered saline (PBS, pH 7.4). This dose of KA induces electroencephalographic epileptiform activity in the hippocampus in 100% of injected mice without any mortality [15,22,32]. For mice in the NS group, an equal amount of NS was microinjected. At 15 min prior to microinjection with KA, mice in the EP+HMGB1 and EP+BoxA groups were pretreated with HMGB1 (5.5 μg in 1 μL, dissolved in sterile 0.1 M PBS) or BoxA (2.5 μg in 1 μL, dissolved in sterile 0.1 M PBS; HMGBiotech, Milan, Italy), respectively, via intrahippocampal (i.h.) injection, while the NS and EP group mice received equivalent NS injection [15]. At 24 h after the onset of seizures, under deep anesthesia with chloral hydrate (10%), mice were perfused via the ascending aorta using 4% paraformaldehyde to fix brains for the immunohistochemistry (IHC) assay (NS group, n = 5; EP group, n = 6; EP+BoxA group, n = 6; EP+BoxA group, n = 6) or decapitated to dissect out hippocampal tissues for Western blotting assay (each group, n = 6).
The present study was performed using the recombinant human disulfide HMGB1 (with a disulfide bond at positions 23 and 45 and a free thiol at position 106) provided by HMGBiotech. The purified natural protein is free from LPS (Cambrex Limulus Amoebocyte Assay QCL-1000, <0.4 ng LPS per mg protein) and contains neither nucleic acids nor interleukin-1beta (IL-1β). Previous experiments demonstrated that HMGB1 (5.5 μg per mouse) i.h. in KA-induced epileptic mice could increase the number of seizures by 70% and the total time spent in seizures by 50% [22]. In the animal experiments of this study, the same amount of HMGB1 for injection was chosen.

Immunohistochemistry
IHC was performed as described earlier [33]. The brain tissues were paraffin embedded and cut. After being fixed in acetone, sections of the brain regions including the cerebral cortex, hippocampus CA1/CA3/CA4, and dentate gyrus were selected for the IHC study (at least two sections per mouse). Briefly, slides were incubated with 0.03% H 2 O 2 in PBS for 10 min at room temperature, then incubated in 5% normal goat serum in PBS for 15 min. Sections were then incubated with different primary antibodies: anti-HMGB1 (ab18256, 1:100; Abcam, HK, China), and anti-P-gp (#517310, 1:50; Calbiochem, Billerica, MA) in 3% bovine albumin serum in PBS at 4°C overnight. Immunoreactivity was detected using avidin-biotin-peroxidase technique, using 3,3ʹ-diaminobenzidine as the chromogen. Images were acquired using an upright optical microscope (DM4000; Leica Microsystems, Mannheim, Germany). For the representative section from each mouse, five nonoverlapping high-power fields (×400) from CA1 region were reviewed for cellular localization, and staining intensity of P-gp expression. For each field, the number of positively stained capillaries was counted and the percentage of P-gp positive cells was obtained. The average value of the five fields was calculated. Staining intensity was graded using a 4-point scale ：Ⅰ, 25% (rare); Ⅱ, 26%-50% (sparse); Ⅲ, 51%-75% (high), Ⅳ, >75% (all). Every mouse was assessed to corresponding grade according to the average percentage of P-gp positive cells.
For transfection, bEnd.3 cells were seeded at 3×10 4 /well in a 24-well plate. Co-transfections were carried out with 400 ng/well of reporter plasmids and 200 ng/well of NF-κB p65 expression vector or the empty vector using Lipofectamine™ 2000 (Invitrogen). The cells were harvested 72 h after transfection, and luciferase activity of transient transfectants was measured using Dual-Luciferase 1 Reporter Assay System (Promega) according to the luciferase assay protocol. To standardize the transfection efficiency, 8 ng/well of phRL-TK (Renilla luciferase) plasmid was co-transfected into the cells as an internal control. The luminescent intensity of firefly luciferase was normalized as a ratio to that of Renilla luciferase [39,40].

Statistical analysis
Quantitative data were expressed as mean±standard deviation from at least three independent experiments. Measurement data were analyzed using one-way analysis of variance followed by Fisher's protected least significant difference test (for data with equal variances) or Welch followed by Dunnett's T3 (for data that did not have equal variances) or independent samples t test; Ranked data were analyzed using Kruskal-Wallis test and Wilcoxon rank sum test. Analysis was performed with SPSS 13.0 (SPSS, Chicago, IL). Differences were considered statistically significant when P<0.05.

Expression and distribution of HMGB1 in the KA-induced mouse epileptic brain tissues
To investigate the expression level and changes in the distribution pattern of HMGB1 in postseizure mouse brain, KA-induced epileptic seizure model was used. The distribution of HMGB1 in the CA1 and CA3 regions from NS and EP group mice was observed using IHC. In the NS group, HMGB1 immunoreactivity was detected mostly in the nuclei of the pyramidal neurons. In the stratum radiatum (sr), stratum lacunosum (sl), and stratum moleculare (sm), decentralized glial cells with nuclear staining as well as neurons with both nuclear and cytoplasmic staining were observed; HMGB1-negative cells were also observed (Fig 1A and 1B). In the EP group, cytoplasmic staining of HMGB1 was substantially increased in the pyramidal neurons (Fig 1C and 1D). Besides of glial cells with nuclear staining, glial cells with both nuclear and cytoplasmic staining in sr, sl, and sm ( Fig 1D) of CA3 regions were also found. These results indicated the nuclear to cytoplasmic translocation of HMGB1 in neurons and glial cells, and predicted its subsequent release into the extracellular space.  3.2 Pre-treatment with HMGB1 enhanced KA-induced up-regulation of P-gp in the epileptic brain As detected by IHC, P-gp is mostly expressed in vascular endothelial cells in the hippocampal regions. Very few neurons and glial cells with P-gp staining were observed. In the NS group, only several P-gp immunopositive capillaries were detected in the CA1 region. Compared to the NS group, the numbers of P-gp-positive capillaries significantly increased in the EP group. This KA-induced P-gp up-regulation was further increased by the HMGB1 injection in the HMGB1+KA group and was abolished by the pre-treatment of BoxA (an antagonist to HMGB1) in the BoxA+KA group (Fig 2A). Result of the semi-quantitative assay for the IHC of P-gp expression was summarized in Table 2: EP group has more mice with stronger P-gp staining (been graded to the grade III and grade IV) as compared to the control NS group; the EP +HMGB1 group has even more mice been assessed in the grade III and grade IV, while the EP +BoxA group has less. Consistent to the above IHC findings, Western blotting detecting brain P-gp levels also showed an elevation of P-gp level in the EP group as compared to the NS group, and this up-regulation was enhanced by the pre-treatment of HMGB1 and was abolished by the BoxA pre-treatment (P<0.05, Fig 2B). Similarly, the brain HMGB1 level also increased in the EP group as compared to the NS group. This level further increased in the EP +HMGB1 group and attenuated in the EP+BoxA group. Collectively, these data show that HMGB1 is closely related to the level of expression of P-gp in the KA-induced epileptic brain. Fig 3, the incubation with HMGB1 does not decrease the viability of bEnd.3 cells significantly at any tested concentrations and time points, as compared to the control cells. The concentration of extracellular HMGB1 in epileptic brain remains unknown, but its local concentration is expected to be substantial, since a significant proportion of nuclear HMGB1 is released or secreted into a very limited extracellular space between the stressed adjoining cells. Andersson et al. [41] have found that concentrations of HMGB1 as low as 10 ng/mL significantly stimulated the release of TNF in monocytes, and 1 μg/mL of HMGB1 significantly activated the synthesis of many cytokines such as TNF, IL-1α, IL-1β, IL-1RA, IL-6, and IL-8. Yang et al. [42] have reported that HMGB1 ranging from 50 to 1000 ng/mL could markedly increase the adhesion of polymorphonuclear leukocytes to ECs from thoracic aorta and the expression of intercellular adhesion molecule-1 and E-selectin in ECs. Therefore, the dose range of HMGB1 from 10 to 500 ng/mL was kept in the following in vitro experiments.

HMGB1 upregulated the expression of P-gp, TLR4, and RAGE in bEnd.3 cells
To examine the effects of HMGB1 on P-gp, TLR4, and RAGE expression levels, bEnd.3 cells were treated with 10, 100, and 500 ng/mL HMGB1 for 4h (for detection of mRNA levels) and 16 h (for detection of protein levels), and mdr1a mRNA, TLR4 mRNA and P-gp, TLR4, and RAGE protein levels were detected using q-PCR and Western blotting, respectively. As shown in Fig 4A, treatment with HMGB1 at all tested concentrations significantly increased mdr1a mRNA levels in bEnd.3 cells as compared to cells treated with normal media (P<0.01, Fig 4A). TLR4 mRNA levels in cells treated with 10, 100 ng/mL HMGB1 were also significantly higher than that in cells treated without HMGB1 (P<0.05, Fig 4B). Interestingly, the maximal upregulating effect of HMGB1 on mdr1a and TLR4 mRNA levels was observed at the concentration of 100 ng/mL but not 500 ng/mL. However, HMGB1 increased P-gp TLR4, and RAGE protein levels in bEnd.3 cells in a concentration-related manner as detected using Western blotting (P<0.05, Fig 4C and 4D). The inconsistency of mRNA and protein expression changes of P-gp and TLR4 in the 500 ng/mL HMGB1 group may be related to the selected specific examination time point (the peak of mRNA level in the 500 ng/mL HMGB1 group may not occur at 4 h), or cell feedback inhibition.

HMGB1 enhanced the activation of NF-κB in bEnd.3 cells
To identify the effect of HMGB1 on the activation of NF-κB, translocations to nuclei of NF-κB p65 in bEnd.3 cells that underwent 100 ng/mL HMGB1 stimulation for 0, 30, 45, and 60 min were observed using immunofluorescence method. Cells treated with 100 ng/mL HMGB1 for 60 min were harvested for the cytoplasmic and nuclear protein extraction and detection of p-NF-κB p65 and NF-κB p65 protein levels using Western blotting. As shown in Fig 5A, the expression of NF-κB p65 was only located in the cytoplasm in unchallenged bEnd.3 cells. After exposure to HMGB1, NF-κB p65 signal was observed also in the nuclei. The number of cells with nuclear NF-κB p65 signal increased gradually according to the extension of time of incubation with HMGB1. At 60 min after HMGB1 treatment, the translocation of NF-κB p65 was already quite obvious. In addition, results of Western blotting further confirmed that p-NF-κB p65 in the cytoplasm and NF-κB p65 in the nuclei fraction were increased in HMGB1-treated cells compared with the control (P<0.05, Fig 5B). These data suggested that the transcription factor NF-κB was activated in HMGB1-treated bEnd.3 cells.

Pre-treatment with LPS-RS, FPS-ZM1, or SN50 attenuated HMGB1-induced up-regulation of P-gp in bEnd.3 cells
To confirm the involvement of TLR4, RAGE, and NF-κB in HMGB1-induced up-regulation of P-gp, bEnd.3 cells were pretreated with either TLR4-specific antagonist LPS-RS (1 μg/mL, which is thought to form inactive complexes with TLR4/MD2), FPS-ZM1 (10 μg/mL, a highaffinity RAGE-specific inhibitor) or a joint treatment of LP-RS plus FPS-ZM1, or NF-κB nuclear translocation inhibitor SN50 (50 μg/mL) for 4 h before the incubation of HMGB1 (100 ng/mL). P-gp mRNA and protein levels were then observed. As shown in Fig 6A, the HMGB1-induced up-regulation of P-gp mRNA level in bEnd.3 cells was partially but  Fig 6A). The similar inhibiting effect of these inhibitors were also observed on the HMGB1-induced up-regulation of P-gp protein level (P<0.01, Fig 6B and 6C). Results indicated that both the two receptors (TLR4 and RAGE) and the transcription factor NF-κB play important roles in HMGB1-induced overexpression of P-gp.

TLR4 and RAGE were involved in HMGB1-induced activation of NF-κB in bEnd.3 cells
Both genetic and biochemical data support that TLR and RAGE signaling pathway could eventually lead to the activation of NF-κB. In the present study, to further validate the relationships between activation of NF-κB and activation of TLR4 or RAGE, bEnd.3 cells were pretreated with LPS-RS (1 μg/mL), FPS-ZM1 (10 μg/mL) or SN50 (50 μg/mL, as a positive control) for 4 h before being incubated with HMGB1 (100 ng/mL) for another 1 h. Nuclear NF-κB p65 levels and NF-κB p65 DNA-binding activity were detected using Western blotting and EMSA assay. Western blotting showed that HMGB1-induced NF-κB p65 translocation to nuclei was significantly reduced by the treatment of LPS-RS or FPS-ZM1 (P<0.01, Fig 7A), while SN50 nearly totally inhibited the translocation (P<0.01, Fig 7A). As indicated by the EMSA assay (Fig 7B), a significant increase in NF-κB p65 binding activity to DNA was observed upon stimulation with HMGB1, and this HMGB1-induced NF-κB activation was significantly attenuated by the pre-treatment of LPS-RS or SN50. 3.8 Exogenous expression of NF-κB p65 activated the mdr1a promoter in bEnd.3 cells and the predictive NF-κB binding site was essentially involved in this process.

Discussion
In the present study, the experiments were started by comparing the expression levels and distribution patterns of HMGB1 in the hippocampal tissues between the NS control mice and KA-induced epileptic mice. HMGB1 was found to be significantly upregulated in the post-seizure brain. Increased staining of perinuclear HMGB1 in neurons and emerged glial cells with both nuclear and perinuclear HMGB1 staining implied the nuclear-to-cytoplasmic translocation of HMGB1. Knowledge about the expression of HMGB1 in epilepsy is still highly limited. These results demonstrated that up-regulation and release of HMGB1 exist in the post-seizure mouse hippocampal tissues, being consistent with Maroso's study [15].
To examine the relationship between HMGB1 and overexpression of P-gp in post-seizure brain tissues, either intracephalic extracellular HMGB1 was increased by i.h. injection of recombinant HMGB1 or endogenous HMGB1 activity was inhibited by the injection of BoxA [41], preceding the induction of KA. Here we reported that, 24 h after the onset of KA-induced seizure, the expression of P-gp in the hippocampus area was significantly increased, and this up-regulation of P-gp was enhanced by the pre-treatment with HMGB1 and attenuated by the pre-treatment with BoxA, suggesting that HMGB1 plays an important role in the upregulated expression of P-gp in the post-seizure brain. Interestingly, it has been reported that HMGB1 could directly increase the gene transcription and expression of P-gp in tumor cells [11]. Whether HMGB1 signaling could directly upregulate P-gp expression in neuronal cells is still unclear and worthy further investigation.
Previous researches substantiated that seizures may induce overexpression of P-gp in four different cell types, i.e., EC, perivascular astrocytes, parenchymal astrocytes, and neurons [3,17,43,44]. However, in the IHC assay of this study, P-gp is mainly observed in vessels in epileptic brain tissues while astrocytic expression was invisible. This phenomenon may be related to the acetone fixation method used in the experiment, as Volk et al. have demonstrated that various fixation conditions and staining variables have a striking effect on the immunohistochemical localization of P-gp in rat brain [3]. And yet for all that, it is acknowledged that overexpression of P-gp in endothelial cells is the first line of defense for inhibiting the penetration of AEDS to parenchyma and contributes much to drug-resistance. Thus, the mouse brain microvascular endothelial cell line, bEnd.3, was used for the subsequent in vitro experiments to investigate the direct effect of HMGB1 on the expression of P-gp in vascular endothelial cells and to explore the signaling pathway involved. Omidi et al. [35] have demonstrated that this cell line possesses a range of relevant transporter systems (including P-gp) significant to the drug penetration of central nervous system and thus can be used as an appropriate cell model for expression modulation studies regarding selected transporters. In the present experiment, HMGB1 sitimulation significantly increased the expression of P-gp in bEnd.3 cells, and the effect was concentration-related. These data are similar to a recent study that showed the upregulating effect of extracellular HMGB1 on the expression levels of P-gp in gastric adenocarcinoma cells [11]. During seizures, in addition to the release of HMGB1 from excitotoxic binding site in mdr1a promoter. (B) Luciferase activity in transfected bEnd.3 cells was measured 72 h after the cotransfection of reporter plasmids and p65 expression vector. Overexpression of p65 in bEnd.3 cells led to a significant increase in mdr1a promoter activity as compared to its counterpart control. Mutation of the potential NF-κB binding site in mdr1a promoter abolished the promotor activation. Data were shown as mean±SD; n = 3. **P<0.01, compared with promotorless pGL3.0-Basic vector; # P<0.05, compared with pcDNA3.1 negative control. NF-κB, nuclear factor-kappa B.
doi:10.1371/journal.pone.0140918.g008 HMGB1 Regulates P-Glycoprotein Expression in Mouse Epileptic Brain neuronal cell death and induction of overexcitation, subsequently increased expression and secretion of HMGB1 by glial cells is expected to support a second and stable wave of extracellular HMGB1 [15]. Previous studies mostly focused on the pro-inflammatory and pro-convulsant effects of HMGB1 in the pathological process of epilepsy, while the present study firstly demonstrates the important role of HMGB1 in seizure-induced overexpression of P-gp.
Many studies have reported that HMGB1 exerts its effects through various receptors, among which, RAGE and TLR4 may be the most representative ones. Upon activation, TLR4 initiates either MyD88 or TRIF dependent pathways and leads to the activation of transcription factor NF-κB, AP-1, or IRF3 [20]. The TLR4/NF-κB signaling pathway has been shown to mediate regulation of P-gp in renal proximal tubule epithelial cells [45]. RAGE engagement by ligands activates various signaling pathways, and NF-κB is the first signal transduction molecule found to be activated. The stimulation of RAGE is able to activate the mitogen-activated protein kinase signaling cascades which converge in IκB kinase complex to inhibit IκB, thereby release and activate NF-κB [21]. The present study supports the critical roles of TLR4, RAGE and NF-κB in mediating the upregulating effect of HMGB1 on P-gp in endothelial bEnd.3 cells. TLR4 and RAGE expression in HMGB1-treated bEnd.3 cells displayed the same pattern of dynamic changes as P-gp. Moreover, TLR4-specific antagonist and RAGE inhibitor could significantly attenuate the effect of HMGB1 on the expression of P-gp, and combined blockage of both TLR4 and RAGE could further decrease this effect. In addition, HMGB1 induced a significant activation of NF-κB p65 in bEnd.3 cells, and this effect could be abolished by the pretreatment of NF-κB inhibitor SN50. Inhibition of NF-κB could remarkably reduce the overexpression of P-gp in bEnd.3 cells induced by HMGB1, which are in consistent with our previous reports [25,33]. On the other hand, HMGB1-mediated activation of NF-κB could be significantly inhibited by TLR4 antagonist or RAGE inhibitor. Biological software analysis found there is a potential NF-κB p65 binding site in the promoter sequence of mdr1a, and luciferase reporter assay discovered overexpression of p65 in bEnd.3 cells could lead to a significant increase in mdr1a promoter activity through the NF-κB binding site. These results suggested that, in bEnd.3 cells, HMGB1 interacted with TLR4 and RAGE, which in turn activated the nuclear translocation of NF-κB, and the activated NF-κB promoted the genetic expression of P-gp directly through the specific binding site (Fig 9).
It is noteworthy that there are some limitations in the present study: 1) P-gp levels in the in vivo experiments may be a consequence of modifications in seizures and may be mediated by some other pathways activated by seizures independently on HMGB1 signaling. For instance, Maroso et al [15] have demonstrated that HMGB1 increased seizure activities in KA-induced mouse model, and seizure-activated pathway-glutamate/COX-2 was shown to be involved in the up-regulation of P-gp in brain capillaries [9,46]. Thus, there is the possibility that HMGB1 may upregulate P-gp expression through influencing seizure-related pathways such as Glutamate/COX-2 and so on; 2) immortalized brain microvascular cell line, bEnd.3, may not display the fully differentiated phenotype shown by its in vivo counterpart. For example, it was reported to display a less P-gp efflux functionality than the in-vivo microvascular endothelial cells [35]; 3) besides overexpression of P-gp in the primary endothelial barrier of the BBB, the increased P-gp expression in perivascular astrocytes in response to seizures may act as a "second line of defense" for inhibiting the penetration of AEDS to parenchyma [3]. In addition, it has been reported that HMGB1 could also activate TLR4 and RAGE in astrocytes. For this reason, the effect of HMGB1 on the P-gp expression in astrocytes also needs further investigation.
In conclusion, this study indicated that HMGB1 could increase the expression of P-gp in mouse brain endothelial cells via TLR4/NF-κB and RAGE/NF-κB signaling pathways and contributes to the seizure-induced overexpression of P-gp in brain tissues. These experimental findings provided a novel insight into the mechanisms underlying the overexpression of P-gp induced by seizure activities. Pharmacological interventions targeting HMGB1, TLR4, RAGE, and NF-κB may be effective for drug-resistant epilepsy in the future. Currently, various approaches have been explored in order to efficiently inhibit HMGB1 in epileptic brain, including clinical drug glycyrrhizin. In addition, some molecules which directly interact with RAGE or TLR4 as competitive antagonists of HMGB1, such as Box A, S100P-derived small peptides, may also be very promising for the future clinical application.