Ursolic Acid Ameliorates Inflammation in Cerebral Ischemia and Reperfusion Injury Possibly via High Mobility Group Box 1/Toll-Like Receptor 4/NFκB Pathway

Toll-like receptors (TLRs) play key roles in cerebral ischemia and reperfusion injury by inducing the production of inflammatory mediators, such as interleukins (ILs) and tumor necrosis factor-alpha (TNF-α). According to recent studies, ursolic acid (UA) regulates TLR signaling and exhibits notable anti-inflammatory properties. In the present study, we explored the mechanism by which UA regulates inflammation in the rat middle cerebral artery occlusion and reperfusion (MCAO/R) model. The MCAO/R model was induced in male Sprague–Dawley rats (MCAO for 2 h, followed by reperfusion for 48 h). UA was administered intragastrically at 0.5, 24, and 47 h after reperfusion. The direct high mobility group box 1 (HMGB1) inhibitor glycyrrhizin (GL) was injected intravenously after 0.5 h of ischemia as a positive control. The degree of brain damage was estimated using the neurological deficit score, infarct volume, histopathological changes, and neuronal apoptosis. We assessed IL-1β, TNF-α, and IL-6 levels to evaluate post-ischemic inflammation. HMGB1 and TLR4 expression and phosphorylation of nuclear factor kappa-light-chain-enhancer of activated B cell (NFκB) were also examined to explore the underlying mechanism. UA (10 and 20 mg/kg) treatment significantly decreased the neurological deficit scores, infarct volume, apoptotic cells, and IL-1β, TNF-α, and IL-6 concentrations. The infarct area ratio was reduced by (33.07 ± 1.74), (27.05 ± 1.13), (27.49 ± 1.87), and (39.74 ± 2.14)% in the 10 and 20 mg/kg UA, GL, and control groups, respectively. Furthermore, UA (10 and 20 mg/kg) treatment significantly decreased HMGB1 release and the TLR4 level and inactivated NFκB signaling. Thus, the effects of intragastric administration of 20 mg/kg of UA and 10 mg/kg of GL were similar. We provide novel evidence that UA reduces inflammatory cytokine production to protect the brain from cerebral ischemia and reperfusion injury possibly through the HMGB1/TLR4/NFκB signaling pathway.

has been shown to occur after thrombolysis, exacerbating the reperfusion injury (4)(5)(6). Therefore, studies aiming to identify an effective adjunct to treatments for cerebral ischemia and reper fusion injury deserve more attention.
Tolllike receptor 4 (TLR4) plays a key role in cerebral ischemia and reperfusion injury by inducing the production of inflamma tory mediators, such as interleukins (ILs) and tumor necrosis factoralpha (TNFα) (7,8). TLR4 were initially identified as receptors for endogenous ligands known as damageassociated molecular patterns (DAMPs), particularly high mobility group box 1 (HMGB1), during brain injury. HMGB1 is a ubiquitous DNAbinding nuclear protein that is either passively released from necrotic cells or actively secreted in response to inflamma tory signals (9,10). In addition, overactive microglia and reactive astrocytes in the ischemic region can aggravate ischemic damage after activation of the TLR4 signaling pathways (11). Therefore, strategies that modulate postischemic TLR4 signaling in the brain may suppress inflammation induced by cerebral ischemia and provide new therapies for stroke.
Ursolic acid (UA: 3bhydroxyurs12ene28oic acid), a natu ral pentacyclic triterpenoid, has been reported to exhibit bio logical activities in the brain, including antioxidative, antitumor, antirheumatic, antiviral, and antiinflammatory effects (12). Furthermore, UA also inhibited microglial and astrocyte acti vation and decreased the levels of TNFα, IL1β, and IL6 in lipopolysaccharideinduced brain inflammation in mice with cognitive deficits (13). However, researchers have not deter mined whether UA protects against ischemia and reperfusion injury by antagonizing the HMGB1/TLR4 signaling pathway. In this study, we used glycyrrhizin (GL) as a positive control drug. GL is a direct HMGB1 inhibitor and the effective dose for treating cerebral ischemia and reperfusion injury has been established (14).
In the present study, we used the rat middle cerebral artery occlusion and reperfusion (MCAO/R) model with UA and GL to examine the mechanism by which UA regulates the inflammation response induced by ischemia and reperfusion. We investigated whether UA reduced inflammatory cytokine production to protect the brain from cerebral ischemia and reperfusion injury possibly though the HMGB1/TLR4/NFκB signaling pathway.

MaTerials anD MeThODs animal Preparation and Drug administration
All experimental protocols involving animals were performed according to the guidelines of the National Institutes of Health Guide for the Care and Use of Laboratory Animals (Publication No. 8523, revised 1985), the UK Animals Scientific Procedures Act 1986 or the European Communities Council Directive of 24 November 1986 (86/609/EEC) and the "Guiding Principles in the Use of Animals in Toxicology," adopted by the Society of Toxicology in 1989. All procedures used in animal experiments were approved by the Institutional Animal Care and Use Committee of China Medical University.
According to previous studies clarifying the oral absorption rate and drug action time, UAadministered mice had a lethal dose 50 of 60 mg/kg and a rattomouse dosing ratio of 6.3/9.1. The final dose of UA was 5, 10, and 20 mg/kg. Since UA is insoluble in water, 0.1% Tween80 is used as a solubilizer and 0.1% Tween80 is used to dilute UA to 1 mg/ml, and the pH is adjusted to 7.4 to avoid the acid and alkali caused by the drug stimulate.
Fortyeight hours after reperfusion, 18 rats in each group were randomly divided into three groups by a researcher who was unaware of the neurological deficits in these rats. Six rats were decapitated to obtain fresh brain tissue samples for bio chemical analyses. The ischemic cortex, which was defined as the penumbra, was collected for ELISA and western blotting analyses based on methods modified from Jiang et al. (22). The brains of six rats were stained to determine the infarct volume; six rats were perfused with fixative for histological preparation and analysis of the brains. The brain samples from each animal were sectioned into three slices beginning 3 mm from the anterior tip of the frontal lobe in the coronal plane. The slices were 3, 4, and 3mm thick from front to back, respectively. The middle slices were embedded in paraffin and sliced into 5μm thick sections for Nissl staining, immunohistochemical staining, immunofluorescence staining, and doublelabeling using terminal deoxynucleotidyl transferasemediated dUTP nick end labeling (TUNEL) and neuronal nuclei (NeuN). To ensure that the positive cells were counted at the same coronal level, we collected ten 5μm thick coronal sections of the dorsal hippocampus (−3.3 to −4.5 mm from the bregma). The number of positive cells in each section was averaged from three non overlapping fields at the same site of the middle cerebral artery blood supply in the ischemic (right) cortex within the penumbral area based on methods modified from previous studies (23,24).
The success rate of model preparation in this experiment was 83.3%. No neurological impairment was observed in 3 of them, 3 with score of 4, 2 died in surgery, 14 with subarachnoid hemorrhage. Subarachnoid hemorrhage accounted for 77.8% of excluded factors, as the main excluded factor.

experimental Transient Middle cerebral artery Occlusion Model
Surgical procedures for MCAO/R were performed in rats using the intracranial suture method, as previously described (25). Briefly, a 5cm nylon monofilament (diameter, 0.26 mm) with a rounded tip coated with silicon (Guangzhou Jialing Biotechnology Company) was inserted into the right internal carotid artery to block the origin of the MCA (approximately 18 ± 2 mm) and maintained for 120 min. Rats in the sham group underwent the same surgical procedures without the insertion of a filament. The rectal temperature was maintained above 36.5°C during and after the surgery with a heating pad. Cerebral blood flow (CBF) was monitored throughout the entire operation. The success of the MCAO/R model was defined as a decrease in CBF by at least 80% during MCA occlusion and a return to 80% CBF after reperfusion.

analysis of neurological Deficits
A fivepoint scale of neurologic deficit scores was used to evalu ate neurological behavior. The neurological deficits were scored 48 h after reperfusion by other investigators who were blinded to the experimental groups (n = 18 in each group). The scoring criteria for neurological deficits have been described previously by Longa et al. (25) and Bederson et al. (26,27).

infarct Volume Measurements
Infarct volume was assessed 48 h after reperfusion (n = 6 per group) with 2,3,5triphenyltetrazolium chloride (TTC, Sigma), as previously described in detail (28,29). The stained slices were photographed and quantified using ImagePro Plus 6.0. Lesion volumes were corrected using the following formula to compen sate for the effect of postischemic edema on the volume of the injury (26,30):

nissl staining
Sections were deparaffinized and then incubated with a 1% cresyl violet (Sigma) solution for Nissl staining. Images were captured using a light microscope (at 400× magnification). In the Nissl stained sections, only intact neurons were counted.

Double-labeling Using TUnel and neun
A TUNEL assay was performed according to the manufacturer's instructions (Roche Molecular Biochemicals, Inc., Mannheim, Germany). Sections were incubated with rabbit antiNeuN antibody (Cell Signaling Technology, Danvers, MA, USA) in PBS/0.2% TX100 and then incubated with the TUNEL reaction mixture to verify the neuronal identity of the TUNELpositive cells. Finally, 4′,6diamidino2phenylindole (DAPI) was added. The total number of TUNELpositive neurons was counted by an investigator who was blinded to the study protocol.

immunohistochemical staining of hMgB1 and Tlr4
Immunohistochemical staining of HMGB1 and TLR4 was per formed using paraffinembedded brain samples from each animal (n = 6 per group), which were sectioned and deparaffinized. The sections were incubated with an antiHMGB1 monoclonal anti body (diluted 1:400, Cell Signaling Technology, Danvers, MA, USA) and an antiTLR4 monoclonal antibody (diluted 1:100, Abcam PLC, Cambridge, UK). Binding was detected using the streptavidinperoxidase kit (Maixin, Fuzhou, China). The posi tive cells were identified, counted, and analyzed in the sections with the ImageJ software.

immunofluorescence staining of iba-1 and gFaP
Immunofluorescence staining of the microglial marker Iba1 and the astrocytic marker GFAP were performed using paraffin embedded brain samples of rats (n = 6 per group) that had been sectioned and deparaffinized. Sections were incubated with primary antibodies (goat antiIba1, 1:100, Abcam, Cambridge, UK, or rabbit antiGFAP, 1:200, Abcam, Cambridge, UK) and then with secondary antibodies labeled with fluorescent dyes (rabbit antigoat, 1:200, Santa Cruz Biotechnology, CA, USA, or mouse antirabbit, 1:200, Santa Cruz Biotechnology, CA, USA). Photomicrographs were quantified performed by converting the images to gray scale, inverting their color, and quantifying the staining intensity in each field with ImageJ software.
Measurement of the il-1β, TnF-α, il-6, and Plasma hMgB1 levels by elisa The IL1β, IL6, and TNFα levels in the ischemic cortex and the HMGB1 levels in the plasma samples were determined using ELISA kits (USCN Life Science Inc., Wuhan, China) according to the manufacturer's instructions.

statistical analysis
All data are expressed as mean ± SD and analyzed with oneway analysis of variance using SPSS20.0. P < 0.05 was defined as statistically significant. The neurological deficit scores among the different groups were compared using the Kruskal-Wallis test. When the Kruskal-Wallis test showed a significant difference, the Dunn's multiple comparisons test was applied. Given the simple size of six animals per group, actual power was performed with the G*Power 3.1.9.2 software at 5% significance level. We got a power greater than 0.9.

resUlTs effect of Ua on neurological Deficits in rats With McaO/r
After 48 h of reperfusion, neurological deficit scores were significantly increased in the control group ( Figure 1B). The UAtreated group (10 and 20 mg/kg) and the GLtreated group displayed significant improvements in their general condi tion and in neurological deficits compared with the control group ( Figure 1B). Moreover, rats treated with 20 mg/kg UA displayed lower median neurological deficit scores than rats treated with GL.   (Figures 1A,C). A significant difference in the infarct volumes was observed between the 10 and 20 mg/kg UAtreated groups.

effect of Ua on histological changes in the Brain of rats With McaO/r
Brains were examined histologically by Nissl staining to deter mine the neuroprotective effects of UA. UA (10 and 20 mg/kg) and GL significantly alleviated the damage in the rat brains.   UA and GL increased the number of intact neurons, and the number of injured neurons with cell shrinkage decreased compared with that in the control group (Figure 2). The number of intact neurons increased significantly as the UA concentration increased.

effect of Ua on neuronal apoptosis in the Brain of rats With McaO/r
Numerous TUNELpositive neurons were observed in the ischemic region compared with those in the sham group. A sig nificant reduction of TUNELpositive neurons was observed in the UA (10 and 20 mg/kg) and GLtreated groups compared with the control group (Figure 3).

effect of Ua on levels of il-1β, TnF-α, and il-6 inflammatory cytokines in rats With McaO/r
The concentrations of IL1β, TNFα, and IL6 in the ischemic cortex of the control group were significantly higher than those in the sham group after 48 h of reperfusion. The concentrations of IL1β, TNFα, and IL6 were significantly reduced after UA (10 and 20 mg/kg) and GL treatment compared with those in the control group (Figure 4). The 20 mg/kg UA treatment had a stronger effect on IL1β and IL6 than the 10 mg/kg UA and GL treatments.

effect of Ua on hMgB1 and Tlr4 in rats With McaO/r
Immunohistochemistry and western blotting analyses were performed to confirm HMGB1 and TLR4 expression in the rat brains. Immunohistochemistry staining of nuclear HMGB1 was observed in the cerebral cortex ( Figure 5A). However, markedly increased HMGB1 staining was observed in the extracellular space in the control group. The number of nuclear HMGB1 positive cells significantly increased after UA (10 and 20 mg/kg) and GL treatment, and this increase was accompanied by a decrease in extracellular HMGB1 staining ( Figure 5B). We also analyzed the brain and plasma HMGB1 levels to meas ure HMGB1 release. The brain and plasma HMGB1 levels were significantly higher for the control than for the sham group after 48 h of reperfusion. Conversely, the brain and plasma HMGB1 levels were significantly reduced after UA (10 and 20 mg/kg) and GL treatment compared with the levels in the control group ( Figure 5). The difference between the 20 mg/kg UA treatment and 10 mg/kg UA treatment was significant. We also observed that UA treatment (10 and 20 mg/kg) and GL treatment decreased the percentage of the percentage of TLR4 positive cells (Figure 6). UA (10 and 20 mg/kg) and GL reduced TLR4immunoreactivity. The semiquantitative immunohisto chemical analyses showed the same results as the western blot analyses. Thus, the UA and GL treatments significantly changed the TLR4 protein level (Figure 6).

effect of Ua on the activation of Microglia and astrocytes During McaO/r in rats
Iba1 and GFAP are specific markers for activated microglia and astrocytes, respectively. Iba1positive microglia and GFAPpositive astrocytes were mostly located around the penumbra of the ipsilateral hemisphere in the ischemic brain. MCAO/R significantly increased the Iba1 and GFAP expres sion in the ischemic region compared with the sham group. The administration of UA (10 and 20 mg/kg) and GL significantly reduced the Iba1 expression, whereas UA did not change the GFAP expression (Figures 7 and 8).

effect of Ua on the nFκB signaling Pathway in rats With McaO/r
We examined the effect of UA on the NFκB pathway to identify the mechanism by which UA regulates inflammatory cytokines. Western blot analyses revealed significantly higher levels of phosphoIκB and phosphoNFκB p65 in the control group than in the sham group. The UA (10 and 20 mg/kg) and GL treat ments significantly decreased the levels of the phosphoIκB and phosphoNFκB p65 proteins. No obvious differences in the levels of total NFκB p65 and total IκB proteins were observed among the experimental groups. We confirmed that UA and GL inhib ited NFκB activation, because the cytosolic and nuclear fractions showed decreased NFκB p65 translocation from the cytosol to the nucleus in the UA and GLtreated rats compared with those in the untreated rats (Figure 9).

DiscUssiOn
Ursolic acid, a pentacyclic terpenoid, exhibits extraordinary neuroprotective properties with antiinflammatory effects during early brain injury for SAH (31,32). In our study, UA protected against cerebral ischemia and reperfusion injury by improving neurological deficits and reducing cerebral infarct volumes when administered i.g. at doses of 10 and 20 mg/kg.  This finding provides further evidence that UA could be an effective therapeutic agent for cerebral ischemia (17,33). Our findings provide new insights into the potential effects of UA on brain ischemia. Cerebral ischemia and reperfusion injury result from the rapid and explosive reoxygenation induced by the inflamma tory response, which is tightly associated with inflammatory mediators such as IL1β, TNFα, and IL6. In fact, inflamma tory cytokines are involved in aggravating brain infarction both in humans and in experimental stroke models (4,5,(34)(35)(36). Therefore, the regulation of any one of these factors may contribute to reducing ischemic injury. We observed that UA reduces the levels of IL1β, TNFα, and IL6 to modulate ischemic pathology. These cytokine levels were clearly elevated after MCAO/R, and UA significantly inhibited these increases. Moreover, the UAinduced reduction of these cytokines paral leled the reduction of ischemic volume. According to more recent findings, administration of the inhibitor of the IL1 receptor improved the prognosis in terms of the size of the neurological deficit and the survival rate (37). Mice injected with a neutralizing antiTNFα antibody after the induction of stroke exhibited a marked decrease in both infarct volumes and mortality (38). Therefore, UA may be an effective therapy for brain infarction due to its ability to reduce the levels of inflam matory cytokines.
Tolllike receptors are a family of pattern recognition receptors that represent key elements in the initiation and progression of inflammatory cytokine production in response to ischemia and reperfusion injury (39). TLR4 is expressed primarily in microglia and astrocyte in the central nervous system and can be activated by DAMPs (such as HMGB1) to induce downstream signals that lead to cytokine production and thus initiation of an inflammatory response after cerebral ischemia and reperfusion injury (40). In our study, the HMGB1 and TLR4 protein expression levels in the ischemic tissue were reduced and HMGB1 translocation was inhibited after UA treat ment. Moreover, HMGB1 and penumbral neuronal apoptosis and death presented the same trends, indicating that UA might have a protective effect against MCAO/reperfusionmediated HMGB1 release from neural cells, resulting in TLR4 activa tion. We also observed that UA reduced the expression of Iba1, associated with evidence of microglial activation. Previous stud ies have shown that TLR4 is a key signaling pathway involved in ischemic penumbral microglial activation, which may be involved in the pathological cerebral conditions by upregulat ing NFκB (41,42). In the present study, we examined the levels of NFκB pathway components. The phosphoNFκB p65 and phosphoIκB levels were partially decreased, suggesting that the NFκB signal pathway was inactivated by the inflammatory response after UA treatment. Similar to the results obtained in the present study, TNFα, IL6, and IL1β can be inactivated by a HMGB1 antibody or HMGB1 inhibitor after MCAO/R (43)(44)(45). Other reports have shown that NFκB pathway activa tion is responsible for TLR4induced target gene expression after hypoxic treatment in microglia (46,47). Based on these results, the UAinduced reduction in IL1β, TNFα, and IL6 production after MCAO/R may be related to the inhibition of the HMGB1/TLR4/NFκB pathway in microglia. Subsequently, UA attenuated ischemia and reperfusioninduced neuronal apoptosis and death. UA was initially described that modulated potentially of HMGB1/TLR4/NFκBmediated inflammation and ameliorated cerebral ischemia and reperfusion injury in the present study.
Furthermore, our results showed that UA and GL had similar effects on HMGB1/TLR4/NFκB expression and the reduction of the neurological deficit scores, infarct volume, and apoptosis in penumbral neurons. We used GL, which is the most studied smallmolecule inhibitor of HMGB1, as a positive control drug (19,48). GL has been reported to function as an HMGB1 inhibitor by binding directly to HMGB1 through interactions The target protein bands were densitometrically analyzed and normalized to GAPDH. Western blot analysis revealed significantly increased expression of phospho-NFκB p65 and pIκB in the control group compared with the sham group (**P < 0.01). UA (10 and 20 mg/kg) and GL decreased the phospho-NFκB p65 and phospho-IκB expression levels and the translocation of NFκB to the nucleus as determined by immunoblotting of the nuclear and cytosolic fractions (c-F) ( ## P < 0.01). A significant difference was observed between the low-dose UA (L-UA) group and the high-dose UA (H-UA) group ( ▲ P < 0.05) (**P < 0.01: compared with the sham group; ## P < 0.01: compared with the control group; # P < 0.05: compared with the control group; ▲ P < 0.05: compared with the H-UA group).
with the two shallow concave surfaces formed by the two arms of both HMG boxes in a wide number of HMGB1involved diseases (14,19,49,50). According to our results, the effect of intragastric administration of 20 mg/kg of UA was similar to the effect of intravenous administration of 10 mg/kg of GL on brain ischemia prior to reperfusion. Interestingly, the UA treatment had no effect on astrocytes and a stronger effect on IL1β and IL6 during the acute stage of ischemic stroke. Considering that reactive astrocytes in the penumbra of the unaffected area are isolated from the toxic environment of the lesion during the recovery process (51,52), UA may also benefit patients at risk for ischemic stroke at 2 weeks into recovery stage. Further work is necessary to clarify this point. In addition, clinical studies have utilized liposomes as a drug delivery system to overcome the poor solubility of UA and enhance the bioavailability of this drug (53,54). These findings above support the possibility and safety of using UA orally to treat cerebral ischemia and reperfu sion injury.
However, this study provided only suggestive data. Thus, the mechanism by which UA affected HMGB1 release from the core that led to TLR4mediated signal transduction was unclear. Further mechanistic studies investigating how UA inhibits of HMGB1/TLR4/NFκB activation are required.
In conclusion, UA attenuated inflammatory cytokine produc tion to protect the brain against cerebral ischemia and reperfu sion injury in a rat model possibly through HMGB1/TLR4/NFκB signaling pathway activation. Based on these findings, UA may be useful as a potential effective adjunct to therapy for ischemic brain injury prior to reperfusion.

eThics sTaTeMenT
All experimental protocols involving animals were performed according to the guidelines of the National Institutes of Health Guide for the Care and Use of Laboratory Animals (Publication No. 8523, revised 1985), the UK Animals Scientific Procedures Act 1986 or the European Communities Council Directive of 24 November 1986 (86/609/EEC). All agreements for animal experiment were approved by the Institutional Animal Care and Use Committee of China medical university and the "Guiding Principles in the Use of Animals in Toxicology, " adopted by the Society of Toxicology in 1989.

aUThOr cOnTriBUTiOns
YZW helped establish the animal model, collect and analyze samples, and write the manuscript. YZW and LL helped perform the Western blotting, immunohistochemistry experiments and review the manuscript. YZW and SMD helped design the study, establish the animal model, perform data analysis, and write the manuscript. FL helped with drafting the work or revising it critically for important intellectual content. ZYH contributed to study planning, data analysis and review of the manuscript.

acKnOWleDgMenTs
The authors have no conflict of interest or disclosures regarding the results presented in this manuscript. This work was supported by the National Natural Science Foundation of China (81070913).