Sevoflurane preconditioning protects experimental ischemic stroke by enhancing anti‐inflammatory microglia/macrophages phenotype polarization through GSK‐3β/Nrf2 pathway

Abstract Aims Sevoflurane preconditioning (SPC) results in cerebral ischemic tolerance; however, the mechanism remains unclear. Promoting microglia/macrophages polarization from pro‐inflammatory state to anti‐inflammatory phenotype has been indicated as a potential treatment target against ischemic stroke. In this study, we aimed to assess the effect of SPC on microglia polarization after stroke and which signaling pathway was involved in this transition. Methods Mouse primary microglia with SPC were challenged by oxygen‐glucose deprivation (OGD) or lipopolysaccharide (LPS), and mice with SPC were subjected to middle cerebral artery occlusion (MCAO). Then, the mRNA and protein levels of pro‐inflammatory/anti‐inflammatory factors were analyzed. GSK‐3β phosphorylation and Nrf2 nuclear translocation were measured. The mRNA and protein expression of pro‐inflammatory/anti‐inflammatory factors, neurological scores, infarct volume, cellular apoptosis, the proportion of pro‐inflammatory/anti‐inflammatory microglia/macrophages, and the generation of super‐oxidants were examined after SPC or GSK‐3β inhibitor TDZD treatment with or without Nrf2 deficiency. Results Sevoflurane preconditioning promoted anti‐inflammatory and inhibited pro‐inflammatory microglia/macrophages phenotype both in vitro and in vivo. GSK‐3β phosphorylation at Ser9 was increased after SPC. Both SPC and TDZD administration enhanced Nrf2 nuclear translocation, reduced pro‐inflammatory microglia/macrophages markers expression, promoted anti‐inflammatory markers level, and elicited a neuroprotective effect. Nrf2 deficiency abolished the promoted anti‐inflammatory microglia/macrophages polarization and ischemic tolerance induced by TDZD treatment. The reduced percentage of pro‐inflammatory positive cells and super‐oxidants generation induced by SFC or TDZD was also reversed by Nrf2 knockdown. Conclusions Our results indicated that SPC exerts brain ischemic tolerance and promotes anti‐inflammatory microglia/macrophages polarization by GSK‐3β‐dependent Nrf2 activation, which provides a novel mechanism for SPC‐induced neuroprotection.


| INTRODUC TI ON
Cerebral ischemia/reperfusion (I/R) during cardio-cerebral surgery has serious adverse effects on patient prognosis. 1 Unfortunately, few therapies for the prevention and treatment of perioperative ischemic stroke have been clinically approved. Therefore, more therapeutic strategies for perioperative stroke are urgently needed. Sevoflurane preconditioning (SPC) results in tolerance against subsequent experimental cerebral I/R damage in vitro and in vivo. [2][3][4][5] However, the exact molecular and subcellular mechanisms underlying this volatile anesthetic's neuroprotective property are still unclear.
One of the most important pathophysiological features of I/R injury is the inflammatory responses in the central nervous system (CNS). 6 Although the exact mechanism is not completely understood, the activation of microglia/macrophages seems to be a characteristic process during I/R-induced excessive brain inflammation. 7 Microglia/ macrophages have particular properties suitable for mediating cellular inflammatory responses during ischemia. [8][9][10] Among these properties, the switching between the pro-inflammatory and the anti-inflammatory microglia phenotype is crucial in regulating inflammatory/anti-inflammatory gene expression and indenting excessive brain damage. [10][11][12] Promoting microglia and infiltrated macrophages toward anti-inflammatory phenotype, which subsequently increased neurogenesis and angiogenesis and remodeled neuronal circuity, could be a target of neuroprotective approaches, including preconditioning stimuli, to produce protective effects against cerebral ischemic damage.
As a multifunctional serine/threonine kinase, glycogen synthase kinase-3β (GSK-3β) is highly active in resting cells and usually inhibits multiple downstream pathways. GSK-3β drives a cascade of signaling pathways, including inflammatory responses in the brain. [13][14][15] This enzyme participates in the production of pro-inflammatory factors, while pharmacological or genetic inhibition of this kinase could act as a molecular brake to limit the brain inflammatory response. 15,16 Accumulating evidence has demonstrated the major role of GSK-3β inhibition in preventing neuronal death, including its feature in inducing the brain ischemic tolerance of SPC. 14,17,18 The phosphorylation of GSK-3β at Ser9, which indicates an inactive status, enables cells to become resistant to various pathophysiological injuries. 19 However, to our knowledge, the underlying mechanism of GSK-3β that governs the anti-inflammatory microglia/ macrophages polarization processes caused by SPC after ischemic stroke is still unknown.
In addition to Kelch-like ECH-associated protein 1 (Keap1), GSK-3β has also been identified as another upstream regulator of Nrf2. 20,21 GSK-3β could phosphorylate numerous Ser residues in the Neh6 domain of Nrf2, which overlap with an SCF/β-TrCP destruction motif to promote Keap1-independent Nrf2 degradation. 22,23 Nrf2 acts as a "master regulator" in response to oxidative electrophilic stress and chemical insults. It is also a major modulator factor associated with the shift of the anti-inflammatory/proinflammatory microglia/macrophages phenotype in response to cerebral I/R injury. 12,24,25 Recent studies have shown that GSK-3β inhibition-induced Nrf2 activation plays a crucial role in protecting organs from I/R injury. 18,22,26 However, whether SPC promoted microglia/macrophages toward anti-inflammatory phenotype and thereby induced ischemic tolerance through this signaling pathway is still unclear.
In the present study, we used oxygen-glucose deprivation Conclusions: Our results indicated that SPC exerts brain ischemic tolerance and promotes anti-inflammatory microglia/macrophages polarization by GSK-3β-dependent Nrf2 activation, which provides a novel mechanism for SPC-induced neuroprotection.

K E Y W O R D S
anti-inflammatory polarity, glycogen synthesis kinase-3β, ischemic stroke, microglia/ macrophages phenotype shift, nuclear factor erythroid 2-related factor, sevoflurane preconditioning Male C57BL6j mice between 8 and 10 weeks old (25-30 g) were purchased from the Animal Laboratory of the Fourth Military Medical University, Xi'an, China . The mice were housed   individually and kept on a 12 h alternating light and dark cycle at   20-25°C and 60% humidity with freely available water and food for at least 1 week prior to treatment or surgery. The sample size was based on our previous study; however, the formal statistical power analysis was not used to guide the sample size of this study. 3 The number of animals used and their suffering were minimized in this study.
Two in vitro models were applied in this experiment: OGD and LPS stimulation. The primary cortical microglia received different treatments as follows: control, OGD, SPC + OGD, TDZD + OGD or control, LPS, and sevoflurane SPC + LPS (n = 6 per group). After different treatments were confirmed, mRNA levels were measured.
Additionally, to further examine the change of Nrf2 and GSK-3β directly in microglia, protein from control, OGD and SPC + OGD, and TDZD + OGD groups were collected and Nrf2 expressed in nuclear, GSK-3β phosphorylated at Ser-9 were analyzed by Western blot (n = 4 per group).

Experiment 2
Examination of anti-inflammatory microglia/macrophages polarization and GSK-3β phosphorylation after SPC and determination of the role of GSK-3β in the neuroprotective effect and the promotion of microglia/macrophages phenotype induced by SPC.
Mice were randomly allocated to three groups (n = 6 per group); (1) Control, (2) SPC + control, (3) I/R and SPC + I/R. The expression of pro-inflammatory and anti-inflammatory factors was analyzed seven days after reperfusion by enzyme-linked immunosorbent assays (ELISAs). Additionally, the expression of iNOS and arginase-1 in microglia was examined by immunofluorescence staining. The phosphorylation at Ser9 and total protein expression of GSK-3β were evaluated via Western blots 2 h after reperfusion (n = 4 per group).

Experiment 3
Verification of the downstream of GSK-3β that regulates the antiinflammatory microglia/macrophages phenotype polarization induced by SPC.
Mice were randomly divided into the control, control + SPC, control + TDZD, I/R + vehicle, SPC + I/R, and TDZD + I/R groups, and Nrf2 nuclear translocation was assessed using Western blot

| Cell culture of mouse primary cortical microglia
Culture of mouse primary cortical microglia was obtained from 24-h old C57BL/6 newborn pups. 29 Briefly, the entire brain of mouse was put in ice-pretreated D-hanks solution, and then, the meninges and other noncortical tissue were separated. The entire cortex was harvested and digested with 0.25% trypsin (Invitrogen) at 37°C for 7 min, followed by the supplement of DMEM/F12 (Invitrogen) in 10% FBS to stop the digestion. After fully dissociating the cortices with pipettes, the cell suspension was subsequently filtered with a 70μm-diameter mesh. Then, the cells were transferred to a 75 cm 2 poly-lysine (PLL, Sigma)-coated flask and incubated at 37°C with 5% CO 2 . About 50% of the culture media was replaced twice per week. After 10 days of culture, primary microglia were collected by shaking the flask for 2 h at 200 rpm and subsequently seeded onto PLL-precoated new plates for following experiments.

| Sevoflurane preconditioning in vitro
The procedure of sevoflurane preconditioning in vitro was based on a previous publication. 30 Briefly, the primary microglia were placed in an incubator chamber (Billups-Rothenberg, San Diego, CA), which was flushed for 5 min with 2.5% sevoflurane in the carrier gas of (95% air-5% CO 2 ), and then, the incubator chamber was sealed at 37°C for 1 h. An anesthetic gas analyzer was used to monitor the concentration of sevoflurane in the chamber.

| Oxygen and glucose deprivation
The OGD was performed as reported previously. 4 Primary microglia cells were plated in DMEM with 10% fetal bovine serum, streptomycin (100 μg/ml), and penicillin (100/units) at 37°C in 5% air. During OGD operation, the medium of culture was switched to serum-and glucose-free Dulbecco's modified Eagle's medium and placed in a modular incubator chamber, which was flushed with a mixture of 95% N 2 and 5% CO 2 at the rate of 3 L/min at room temperature for 30 min. Control cultures were incubated for the same period of time in a humidified atmosphere of 95% air and 5% CO 2 at 37°C. After 4-h challenge, microglial cells were removed from the anaerobic chamber, and the medium of culture was replaced by Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. The generation of reperfusion insult was confirmed by maintaining cells in a humidified 5% CO 2 incubator for an additional 24 h at 37°C.

| LPS stimulation in vitro
The administration of LPS was performed according to the previously reported paradigm. 31  following the manufacturer's instructions. LDH leakage was calculated as the percentage of LDH released into the medium out of total LDH activity (LDH in both medium and cells), which is LDH released (%) = (LDH activity in the medium/total LDH activity) × 100%.
Cultures without OGD/LPS treatment (control group) served as a baseline for LDH release.

| Sevoflurane preconditioning in vivo
Sevoflurane preconditioning of the mice was achieved by inhaling 97% O 2 containing 2.5 vol% sevoflurane 1 h a day for 5 consecutive days. Mice in the control group inhaled 97% O 2 without sevoflurane following the same schedule. After the final treatment and a 24-h washout period, mice were subjected to MCAO surgery.

| Arterial blood gas measurement
Five additional mice in each group were used to determine the amount of arterial blood gas. About 0.2 ml blood of each mouse was taken, respectively, from the femoral artery at the end of the subject's last exposure to sevoflurane or oxygen. Samples were analyzed immediately using the OMNI Modular System (Rapidlab 1260, Bayer HealthCare, Uxbridge, United Kingdom).

| Transient middle cerebral artery occlusion model
Cerebral I/R injury was induced by a transient middle cerebral artery model in mice as described previously. In brief, after an overnight fast, animals were anesthetized by 3% sevoflurane for induction and 2.5% for surgery. After the right carotid arteries were dissected out, an intraluminal 6-0 nylon monofilament with a round tip was inserted from the right common carotid artery to the right middle cerebral artery. Following 1 h of transient occlusion, the filament was withdrawn to allow reperfusion. The temporal temperature was maintained at 37 ± 0.5°C by a thermostatic blanket and a lamp. Sham-operated mice in the control group were subjected to the same surgical procedure but without inserting the filament.
A laser Doppler flowmeter (PeriFlux 5000; Perimed AB, Sweden) was placed on the skull's dorsal surface (caudal 2 mm and lateral 5 mm to bregma) before, during, and after the operation to quantitate the regional cerebral blood flow (rCBF). Mice were excluded from the final analysis if their rCBF did not fall below 20% of baseline during occlusion or recover over 80% during reperfusion. Moreover, if the MCAO procedure could not be completed within 10 min, the corresponding data were also discarded.

| Neurobehavioral evaluation and infarct size assessment
Neurobehavioral outcomes in the mice were assessed in accordance with the Garcia Score Scale by an observer blinded to the animal groups. 33 Data were expressed as median (interquartile range).
The infarct volume was assessed by TTC staining following standard procedures. Briefly, the mouse was decapitated after the last neurobehavioral test. The brain was rapidly removed and All primers used in this study were listed in Table 1.

| Western blot
At 30 min and 2 h after reperfusion, mice from each group (n = 5) were euthanized. Brain tissues corresponding to the ischemic penumbra were harvested as previously described. The tissues were homogenized in ice-cold RIPA lysis buffer containing 1% phenylmethanesulfonylfluoride (Beyotime, Nantong, China).
To examine Nrf2 nuclear translocation, nuclear protein was exacted with a Nuclear Extraction Kit (Pierce Biotechnology, USA).

| SPC promoted mouse primary microglia polarization into the anti-inflammatory microglia/ macrophages phenotype, increased Nrf2 nuclear expression and GSK-3β phosphorylation against OGD
As illustrated in Figure 1A, As shown in Figure 1I, Nrf2 protein content in the nucleus was increased in the SPC and TDZD-treated groups, but not in the SPC-treated control group, as compared to the I/R group (p = 0.037 and 0.018, respectively). The level of phosphorylated GSK-3β in the OGD groups was decreased compared with that of the control group, but this reduction was reversed in the SPC and TDZD-treated group ( Figure 1J, p = 0.0074 and 0.0053, respectively).

| SPC reduced the pro-inflammatory factors, increased the anti-inflammatory cytokines, and protected microglia culture against LPS-induced injury
Moreover, LPS stimulation was also performed in this study to further determine the effect of SPC on polarizing microglia toward the anti-inflammatory phenotype. The microglial culture was severely damaged by LPS stimulation, which was evident from the increased LDH release (

| SPC shifts microglia/macrophages polarization toward anti-inflammatory phenotype in the Ischemic
Hemisphere 7 days after reperfusion.
Changes in physiologic parameters at the end of the preconditioning operation and various intervals of I/R are summarized in Table S1.
No significant differences in pH values, temporal temperatures, or partial pressures of carbon dioxide (P CO2 ) were detected among the groups. As shown in Figure S1, SPC did not alter the regional cerebral blood flow.
As presented in Figure 3A-C, the TNFα, IL-1β, and iNOS mRNA levels were increased on day 7 after I/R, while this elevation was reversed by sevoflurane pretreatment (p = 0.0202, 0.006, and <0.001, respectively). As shown in Figure 3D As shown in Figure 3G,I, the number of Arg1 positive microglia in ischemic penumbra increased in the SPC group but not in the I/R group (p = 0.0079). In line with this, the number of iNOS positive microglia increased in the I/R group, while SPC treatment reversed this increase (p = 0.032, Figure 3H,J).

| SPC increased the phosphorylation of GSK-3β, and supplementation with a GSK-3β inhibitor reduced cerebral I/R injury
As shown in Figure 4A, GSK-3β phosphorylation was determined by Western blot analysis. The abundance of phosphorylated GSK-3β in the I/R groups was reduced compared with that of the control group (p = 0.0218), but this reduction was ameliorated in the SPC group (SPC vs. I/R, p = 0.0206). SPC did not alter GSK-3β phosphorylation levels in control mice. Neither I/R nor SPC treatment affected the total GSK-3β protein expression.
To further verify the role of GSK-3β in I/R tolerance, we used a GSK-3β inhibitor, TDZD, in this study. As demonstrated in Figure S2, GSK-3β phosphorylation was measured by Western blots. The phosphorylation of GSK-3β at Ser9 was decreased in the vehicle-treated I/R groups compared with the control group (p = 0.0210), but this reduction was ameliorated by TDZD supplementation (TDZD + I/R vs. I/R + vehicle, p = 0.0045). Moreover, as shown in Figure S3, the TDZD + SPC group did not further enhance the phosphorylation of GSK-3β at Ser9.
As shown in Figure

| GSK-3β Inhibition promoted microglia/ macrophages shift to the anti-inflammatory phenotype in the ischemic hemisphere 7 days after reperfusion
As shown in Figure 5A-C, the mRNA expression of the proinflammatory cytokines was significantly elevated in the ischemic penumbra but was reduced when mice received SPC. The TDZD treatment also reduced the elevated mRNA expression of these pro-inflammatory cytokines (TDZD + I/R vs. I/R: p = 0.0206, <0.001, 0.0157, respectively). As shown in Figure 5D-F, the mRNA levels of the anti-inflammatory cytokines in the SPC group were higher than those in the I/R group. TDZD also increased the level of these anti-inflammatory factors (CD206, YM1/2, arginase-1: p < 0.001). Scale bar = 10 μm. *p < 0.05, **p < 0.01, ***p < 0.001 versus the control group; #p < 0.05, ##p < 0.01, ###p < 0.001 versus the I/R group. One-way ANOVA with Tukey's post hoc test was used for statistical analysis. I/R, ischemia/reperfusion; IL-1β, Interleukin-1β; iNOS, inducible nitric oxide synthase; Sevo, sevoflurane preconditioning; TNFα, Tumor necrosis factorα

| Knockdown of Nrf2 reversed antiinflammatory microglia/macrophages polarization and abolished the inhibition of reactive oxygen species generation produced by TDZD treatment
Representative images of flow cytometry used to examine the ratio of pro-inflammatory/anti-inflammatory positive microglia/macrophages are shown in Figure 8A. In Figure 8B, activated pro-inflammatory microglia/macrophages (CD86 + ) intensely accumulated in the ischemic penumbra compared with the ipsilateral hemisphere of the control group (p < 0.001). Meanwhile, SPC and TDZD treatment significantly reduced this accumulation (p < 0.001 and p = 0.0003, respectively).
In Figure 8C, the percentage of anti-inflammatory -positive microglia/ macrophages (CD206 + ) was not affected by I/R surgery, but both SPC and TDZD treatment increased this ratio compared with that of the I/R group (Sevo + I/R: p < 0.001, TDZD + I/R: p = 0.0013). As expected, supplementation with AAV-Nrf2 reversed the accumulation of antiinflammatory positive microglia (p = 0.016). No significant difference was detected between the TDZD and TDZD + AAV-GFP groups regarding the percentage of pro-inflammatory and anti-inflammatory positive microglia/macrophages. Figure 8D, the generation of reactive oxygen spe-  also prevented pro-inflammatory cytokines mRNA expression and enhanced the expression of anti-inflammatory cytokines genes in these in vitro models, which was consistent with other studies using different protective stimuli. 12,47 Another major finding of this study is that SPC engages antiinflammatory microglia/macrophages phenotype polarization possibly in a GSK-3β phosphorylation-dependent manner. In addition to regulating physiological processes such as glucose metabolism, cellular development, and differentiation, the phosphorylation of GSK-3β is also a target that prevents ischemic insult. In the CNS, GSK-3β was expressed in both neurons and microglia, which affected the microglial/macrophages activation by modulating a cascade of signals. 48-51 SPC increased the phosphorylation of GSK-3β in ischemic penumbra and also in primary microglia culture after OGD. We also employed TDZD, a GSK-3β

As indicated in
inhibitor at Ser9 that previously used by other groups, and found that both SPC and TDZD elicited a neuroprotective effect by improving anti-inflammatory marker gene expression and reducing pro-inflammatory marker mRNA expression. 52,53 These results indicated that SPC-mediated improvement of anti-inflammatory microglia/macrophages phenotype polarization was GSK-3β inactivation dependent.

| Limitations
Some limitations of the present study should be noted. First, we did not clarify the role of cell-specific expression of Nrf2, especially in microglia, in the regulation of inflammasome or other inflammatory response produced by SPC. 55 Moreover, microglia/macrophages appear to be heterogeneous with diverse functional phenotypes that range from immuno-enhanced phenotypes to anti-inflammatory phenotypes. 56 Considering the unique roles of microglia/macrophages, the promotion of neurogenesis and angiogenesis produced by the activation of the immune-suppressive microglia phenotype may play a vital part in stimulating neuroprotective mechanisms to protect neurons from injury. 12,16,47,57 However, whether these complex phenotypes of microglia/macrophages could be directly or indirectly affected by SPC, and whether this polarization of phenotypes was influenced by sex or sex related signaling, still needs to be studied in the future. 58

| CON CLUS IONS
In summary, this study demonstrated that GSK-3β phosphorylationmediated Nrf2 activation is involved in SPC-induced cerebral ischemic tolerance by shifting microglia/macrophages toward antiinflammatory phenotype after cerebral I/R. These investigations may reveal a potential mechanism of SPC-induced neuroprotection.

ACK N OWLED G M ENT
We thank American Journal Experts (AJE) for assisting in the preparation of this manuscript.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable requests.