Reducing the Amount of M1 Microglia by Inhibiting CXCR4 and iNOS Exerts Neuroprotection in a Rat Model of Subarachnoid Hemorrhage


 Early inflammation is a significant factor in acute pathophysiological events of early brain injury (EBI) after subarachnoid hemorrhage (SAH). Although there have been numerous studies of neuroinflammation and SAH, the effect of M1 microglia on the progressions of neuroinflammation in SAH remains non-elucidated. CXCR4 is thought to be the critical regulator of the migration and recruitment of microglia, and early studies found that iNOS/NO• does represent an effective ferroptosis regulator and leads to the M1 microglia more resistant to the initiator of ferroptosis. Thus, we investigated the effect of AMD3100 (a highly selective antagonist of CXCR4) and L-NIL (an inhibitor of iNOS) on neuroinflammation in a rat SAH model. We found AMD3100 could suppress the migration of M1 microglia through the CXCL12/CXCR4 pathway. Treatment of AMD3100 could decrease the level of related inflammation factors and improved the prognosis within 24 h after SAH. Moreover, L-NIL could inhibit the expression of (i)NOS and promote the expression of ferroptosis-related proteins and the degree of lipid peroxidation. Importantly, the combination of AMD3100 and L-NIL could reduce the quantity of M1 microglia in the injured brain area and reduce the secretion of related inflammatory factors to improve the prognosis. To sum up, these data indicate that inhibiting CXCR4 and iNOS following SAH produces cerebral protection, and its anti-inflammation provides a potential therapeutic target for treating SAH.


Introduction
Subarachnoid hemorrhage (SAH) is a severe form of stroke accounting for high mortality and morbidity [1]. The main cause of SAH is rupture of an intracranial aneurysm, covering approximately Eighty-ve percent of SAH [2]. Despite the advances in disease management, the mortality rate rises to 30% in acute situations. In recent years, early brain injury (EBI), occurring within the rst 72 hours after aneurysmal SAH, is an essential contributor to the poor outcome after SAH. EBI contains numerous pathophysiological events: in ammation, brain edema, increased blood-brain barrier (BBB) permeability, apoptosis, and neurodegeneration [3]. Early in ammation is a signi cant factor in acute pathophysiological events. Within several hours after SAH, the in ammatory response is rapidly initiated, and various in ammatory cells, especially microglia, ock to the damaged cerebral parenchyma. Then, various proin ammatory cytokines are expressed from microglia and lead to severe damage to the brain [4]. Therefore, attenuating SAH-induced in ammation may have great bene ts in alleviating neurological damage and improving the prognosis.
Microglia cells, considered as innate immune cells of the central nervous system (CNS) widely, are one of the major cellular components that induce neuroin ammation. Microglia can be rapidly activated by pathological stimulation of the CNS. When the CNS is destroyed, microglia can be activated to remove apoptotic cells, damaged tissue and cell debris, tissue metabolites as well as contribute to repairing damaged nervous system. Whereas, continuously activated microglia secrete plenty of in ammatory mediators such as TNF-α, IL-1β, reactive oxygen species (ROS) and reactive nitrogen species (RNS), etc., thus impairing mitochondrial function, resulting in neuronal degeneration and necrosis, and further exacerbating CNS damage [5]. According to reports that microglia are classi ed into a classic pro-animals. Adult male Sprague-Dawley rats weighing between 300 and 350 g were purchased from the Animal Center of Chinese Academy of Sciences, Shanghai, China. Placed all animals in a humidi ed room with a constant temperature of (24 ± 2) °C, air humidity (50 ± 10%), and under a 12-h light/dark cycle, and free to food and water. All of the rats were placed under an aesthesia before the xationperfusion and exsanguination euthanasia procedures. Try our best to minimize the number of experimental animals and their suffering.

Rat SAH Model
As a previous study [24] described, 10% chloral hydrate (0.4 ml/kg) was injected intraperitoneally to anesthetized those rats and x all the rats in the stereotaxic apparatus. The experimental SAH model consisted of a needle with a rounded tip and a side hole stereotaxic insertion into the prechiasmatic cistern. The needle was at an angle of 45 degrees to the sagittal plane and was inserted 7.5 mm anterior to the bregma with the needle side hole facing the right side. The nal depth of the needle tip was 2-3 mm anterior to the chiasma (about 10-12 mm from the surface of the brain), then retracted 0.5 mm. If needed, use bone wax to plug the burr hole to prevent loss of CSF and blood from the midline vessels prior to needle insertion. Following this, 0.3 ml non-heparinized autologous arterial blood from the femoral artery was slowly delivered into the prechiasmatic cistern for 20 seconds under aseptic conditions using a syringe pump. For control, inject 0.3 ml saline into the sham group of rats by using the same method. To prevent dehydration of those rats, 20 ml of 0.9% saline were subcutaneously injected after the SAH procedures immediately. After recovering for about 45 min, return all rats to their cages, maintain the room temperature at 23 ± 1°C. Meanwhile, we monitored the heart rate and rectal temperature of the rats and maintained the rectal temperature at 37 ± 0.5°C during the experiments. Then at the indicated time point after experimental SAH, rats were euthanized. It was observed in this study that the inferior basal temporal lobe was stained by blood. Next, we collected the brain tissues surrounding the hemorrhage located in the temporal base and subjected them to analysis. We used the underlying temporal base brain tissues alone for western blot analysis and the total coronal sections containing temporal base brain tissues for immuno uorescence analysis, Prussian blue reaction, and FJB staining. The collected tissue area is shown in Fig. 1.

Drug Administration
AMD3100, which was applied as CXCL12/CXCR4 antagonist extensively [25][26][27], was used in this study to restrain CXCR4. AMD3100 (Plerixafor 8HCl, JM3100 octahydrochloride) was obtained from Sigma-Aldrich (St. Louis, MO, USA), dissolved into DMSO (the nal concentration: 0.5%), then diluted to the nal injection concentration by 0.9% physiological saline. AMD3100 (1 mg/kg) was administered by intraperitoneal (i.p.) injection 1 h after SAH. The dosage of the time of administration and the dosage of AMD3100 were chosen based on previous studies [28] and have been proven to be effective. The vehicle (0.5% DMSO in saline) was administered to the rats 1 h after SAH.
L-N (6)-iminoethyl-lysine (L-NIL) (Sigma, St. Louis, MO, USA), an inducible (i) NOS inhibitor, was dissolved in DMSO (the nal concentration: 0.5%), then diluted to the nal injection concentration by 0.9% physiological saline. L-NIL (20 mg/kg, 0.3 mL i.p) was injected after 6 h post-trauma. The dosage of the time of administration and the dosage of L-NIL was chosen according to pharmacokinetic studies [29] and has been proven to be effective. The vehicle (0.5% DMSO in saline) was administered to the rats 15 min and 12 h after SAH.
Liproxstatin-1 (Lip-1), a potent and speci c inhibitor of ferroptosis, was purchased from TargetMol. It was dissolved in DMSO (the nal concentration: 0.5%), then diluted to the nal injection concentration by 0.9% physiological saline. The inhibitor was then administered to the rats at the doses of 5 mg/kg via intraperitoneal injection at 2 h before the induction of SAH. The dose of Lip-1 was chosen based on a previous study [30] and has been proven to be effective. The vehicle (0.5% DMSO in saline) was administered to the rat 2 h after SAH. The rats were anesthetized by 10% chloral hydrate (3.5 ml/kg) deeply at each speci c time point after SAH. Five rats per group were sacri ced, and then transcranial perfused using chilled PBS. The brain tissues were collected from the temporal lobe base, stored at − 80℃ immediately, and then used for Western blotting. Part 2. Probe the mechanism of M1 microglia migration in vitro and the effect of AM3100 on the prognosis of SAH rats. The mechanism of the migration of M1 microglia is probed by performing the transwell migration assay. BV2 microglia, treated with LPS for 24 h, polarized into M1 phenotype. M1 microglia was seeded in the chamber at appropriate densities. The experiment was performed with the following groups: (1) Vehicle (2) AMD3100 (3) CXCL12 (4) AMD3100 and CXCL12. The ability of migration of M1 microglia was quanti ed by viewing the cell number in four different elds at a microscope, at least in triplicate. Elisa assay and the neurobehavioral experiment were performed to evaluate the prognosis. For the Elisa assay, rats were divided into three groups: sham, SAH, and SAH + AMD3100. Collect serum from each rat to measure the expression of related in ammation factors at different time points, at least in triplicate. For the neurobehavioral experiment, the wire-grip test and the Morris water maze test have proceeded. The recovery of neurobehavioral was assessed by speci c indicators (each group set contained 6 samples; n = 6; Fig. 1c). Part 3. Probe the relation between ferroptosis and M1 microglia. OxyHb induced BV2 cells to mimic an in vitro model of SAH. The cells were divided into 6 groups randomly: Normal, OxyHb, OxyHb + DMSO, OxyHb + DMSO + L-NIL, OxyHb + DMSO + Lip-1, OxyHb + DMSO + L-NIL + Lip-1. Western blot, malondialdehyde (MDA) content and lipid peroxidation (Lipid-ROS) assay were evaluated at the speci c time after OxyHb was induced (Fig. 1c). Part 4. Probe the relation between the quantity of microglia and the degree of neuroin ammation, neuronal damage, behavioral de cits, and functional recovery after the early stage of SAH. Rats were divided into ve groups: Sham, SAH + Vehicle, SAH + AMD3100, SAH + L-NIL, SAH + ADM3100 + L-NIL, injected vehicles, or drugs by intraperitoneal injection. To observe the amount of M1 microglia, we performed the Western blot assay and the immuno uorescence staining (each group set contained 5 samples; n = 5). To detect the tendency of in ammatory factors such as IL-6, TNF-α, IL-1β at the speci c time point after SAH, we collected the serum of experimental rats performed for ELISA (each group set contained 5 samples; n = 5). To evaluate the degree of neuronal damage, we performed Nissl staining and Fluoro-Jade B (FJB) staining (each group set contained 5 samples; n = 5). In order to assess the degree of behavioral de cits and functional recovery of the early stage of SAH, the Wire-grip test and the Morris Water Maze test were evaluated at indicated time-points after SAH (each group set contained 6 samples; n = 6; Fig. 1c).

Western Blot Analysis
Western blot analysis was carried out by using standard procedures as previously described. The animals were anesthetized deeply via intraperitoneal injection using 10% chloral hydrate, and animals were sacri ced after anesthesia. We took out the tissues of the temporal-oor brain after perfusing the brain tissue transcardially with 150 ml of chilled PBS. Homogenize the samples by using RIPA lysis buffer and then sonicate them. We collected the supernatant after centrifuging the homogenates at 12000 × g at 4°C for 20 minutes. The protein concentrations were measured by using the A280 method with NanoDrop 2000C (Thermo Scienti c, USA). Equal quanti ed sample (30 µg) was loaded for 7.5% − 15% SDS-PAGE then electrophoresed with a semidry electrotransferred unit for transferred to polyvinylidene uoride membranes. At room temperature, blocking membranes were performed with 5% skim blocking grade milk for 2 hours. The diluted primary antibodies were incubated with membranes overnight at 4°C: CXCR4, CXCL12, iNOS, CD206, CD86, Tfr1, Tf, COX2, xCT, Fth, Ftl, GPX4, GAPDH, Tubulin. Membrane washing was performed three times with TBST after primary antibody incubation. The appropriate proportion of secondary antibodies was incubated with membranes (RT, 2h). The protein density of bands was captured by an ECL system (ChemiScope, Shanghai) after washing three times with TBST. Quanti cation of the Western blots was performed by using Image J software (NIH), and standardization was done with GAPDH or Tubulin.

Transmigration Assay
The ability of cell migration is analyzed by using a transwell system (aperture of 8 µm;24-well insert; Millipore, USA). Place the 200 µl cell solution. The concentration is 1.5 × 10 5 cells/mL and contains AMD3100 (2.5 µl /mL) or vehicle, in the upper chamber. Place the 600 µl medium containing CXCL-12 (200ng/ml, Beyotime Biotechnology, Shanghai) or vehicle in the lower chamber. In another independent experiment, the lower chamber was seeded with the astrocytes that were exposed to 10 µM OxyHb (Jinpin Chemical Technology, China) to mimic SAH condition as reported previously. After 16 hours of incubation at 37°C, cells were xed using 4% paraformaldehyde solution (RT, 1h). Then the non-migrate cells of the top of the upper chamber are removed by gently scrubbing with cotton swabs. The 0.1% crystal violet purchasing from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) is used to stain the migrated cells for 10 mins, and the count of cells is performed with Olympus LCX100 Imaging system (Olympus Corporation, Tokyo, Japan). All experimental results were performed independently, at least in triplicate.

Immuno uorescence Staining
Brains were removed and xed using 4% paraformaldehyde (24h), then brains were rapidly frozen and cut into 10-mm thick coronal slices. Frozen slices were air-dried and were rehydrated in PSB. Then slices were permeabilized with blocking buffer (PBS with 10% BSA and 0.3% Triton) for 2 h. Subsequently, these slices were incubated at 4°C overnight with the following primary antibodies: CXCR4/CXCL12/I-BA1. The appropriate proportion of secondary antibodies (Beyotime Biotechnology, Shanghai) was incubated with slices (RT, 1h after washing three times with PSB. To observe the cell density of total cells, nuclear staining of these slices was carried out with 4', 6-diamidino-2-phenylindol (DAPI, Beyotime Biotechnology, Shanghai), and then slices were observed using the uorescence microscope.

Perls' Prussian Blue Staining
The iron accumulation was detected with Perls' Prussian blue stain, as described previously [31]. After washing three times with PBS, the brain slices were permeabilized with PBS containing 0.3% Triton (RT, 5 mins). Subsequently, these slices were incubated with Perl's reagent (equal parts 5% hydrochloric acid and 5% potassium ferrocyanide) for 30 mins. Then slices were washed in PSB three times rightly again. To inactivated endogenous peroxidase activity, brain slices were covered with 0.3% hydrogen peroxide (H2O2) in methanol for 15 minutes and then were washed in PBS three times lastly. Then slices were incubated in 3, 3-diaminobenzidine (DAB) to develop( develop) immunoreactive signals and used hematoxylin to counterstaining.

Lipid Peroxidation Assay
The BV-2 cells were induced by LPS to acquired M1 polarization characteristics. The working solution of BODIRY 581/591 C11(10 µM) was used to co-incubate with the M1 microglia cells. The uorescence changed from red to green as lipid peroxidation increases.

Malondialdehyde (MDA) Content
The lipid peroxidation MDA assay kit (#S0131S, Beyotime Biotechnology, Shanghai, China) was used following the manufacturer's instructions to measure the microglia malondialdehyde (MDA) content. In brief, the cells were collected and homogenized with a cold lysis solution. The lysates were then centrifuged at 12,000 g for 10mins at 4°C to collect the supernatant. The MDA solution (200 µL) was added to each sample or supernatant (0.1 mL). The mixture was then heated at 100°C for 15 min and collected supernatant after centrifuged at 1000 g for 10 min after cooled to room temperature in the water bath. The supernatant (200 µL) was colorimetrically quanti ed (Thermo Scienti c Multiskan FC, OD = 532 nm). MDA levels were expressed as nmol/mg.

Nissl Staining
For Nissl staining, the coronal brain sections were xed in 4% formaldehyde and washed with PSBT 3 times. The Nissl staining solution purchased from Beyotime Biotechnology (Shanghai, China) was used to incubate with each brain section (10min, 50℃). The microscope (Nikon TE300; Nikon) was used to capture ve regions randomly for quanti cation to evaluate the peri-injury zone positively stained cells. Compared with normal neurons, we can observe shrunken and/or contained multiple vacuoles and darker color in nuclei stain.

Fluoro-Jade B (FJB) Staining
Whole brains were frozen and made as frozen sections (coronal sections of 10 µm thickness). Then the sections were xed with 4% paraformaldehyde for 24 h. Fluoro-Jade B (FJB) staining was performed based on the instructions provided by the manufacturer (Millipore, Germany). Firstly, the basic alcohol solution, which was made up of 1% sodium hydroxide and 70% ethanol, was used to immerse the coronal brain sections for 5 minutes. Then the sections were rinsed in 70% ethanol for 2minutes and were washed in distilled water for 2 minutes. Potassium permanganate solution (0.06%) was then used to immerse those tissues for 10 minutes. Subsequently, the specimens of slides were incubated in a Fluoro-Jade B solution (0.0004%) for 30 minutes after 2 minutes of distilled water rinse. The microscope (Nikon TE300, Japan) was used to observe the captured images. Measure and quantify the count of Fluoro-Jade B-positive cells in three sections per rat.
In mimic the experimental SAH in vitro, the cells were cultivated in Petri dishes until the density of cells reached 80%-90%, and then the medium was replaced with a serum-free medium. The cells were stimulated with oxyhemoglobin (OxyHb, Sangon Biotech) for 6 h. The appropriate duration (6h) of stimulation was obtained through the CCK8 assay.

Primary Culture
The primary astrocytes were obtained from the brain tissue of 1-to 2-day-old Sprague-Dawley rats as previously reported [33]. The rat pup's brain tissue was taken out and removed the blood vessels and meninges from the cortices. Wash the brain cortex with PBS and digest with 0.25% trypsin (37℃, 5min). Then complete DEME/F12 medium, containing 10 % FBS, 2 mM of L-Glutamine100 mM of nonessential amino acids, 1 mM of sodium pyruvate, 50 U/ml of penicillin/streptomycin was used to terminate trypsinization. Resuspend the cells with complete DMEM/F12 medium after centrifugation. Cell suspension was added in petri dishes. The oligodendrocytes and microglial cells were separated from astrocytes by gently shaken at 210rpm at 14 days after initial seeding. The astrocytes were seeded in a 24-well plate at a density of 5×10 5 /well. Then cells were stimulated with OxyHb for 6h and were prepared for transwell assay.
The cortex of 1-d-old neonatal rats was prepared to perform primary cultures of cortical neurons. We dissected the neonatal rats' cortices and cultured them in Neurobasal-A medium (Gibco) with 2% B27 (Gibco), 100 U/mL penicillin, and 100 µg/mL streptomycin (Gibco). Then the cell suspension was plated on a Poly-D-lysine-coated (0.1 mg/mL, Sigma) 96-well plate at the density of 1×10 4 cells/cm 2 for 6-8 days to perform the experiments as follows.

Cell Viability Assay
The microglia cells were seeded into a 96-well plate at the density of 1x104 cells/well and incubated for 72h with or without the stimulation with OxyHb. The cells were incubated for 1hours in the dark at 37℃ after adding 10 µl Cell Counting Kit-8 (CCK8) reagents (Dojindo, Kumamoto, Japan) to per well. The absorbance of each well was measured at a wavelength of 450 nm by a microplate reader (Bio-Rad, Hercules, CA).

Enzyme-Linked Immunosorbent Assay (ELISA)
The content of the related factors of in ammation was detected using the corresponding ELISA kit (Beyotime Biotechnology, Shanghai). Brie y, the standard products and the tested samples were diluted in a particular proportion, then add them into the enzyme-labeled plate for 2 h at 37°C. Then we removed the solution and washed the enzyme-labeled plate 5 times with the washing buffer. The biotinylated antibody was added and incubated with each enzyme-labeled plate (RT, 1h). Remove the solution and wash the enzyme-labeled plate 5 times again. Add the horseradish peroxidase in plates to mark streptavidin at RT for 20 mins, protect from light. Remove the solution and wash the enzyme-labeled plate 5 times lastly. The color development was achieved with 3, 3, 5, 5-tetramethylbenzidine (TMB) after washing. Stopped solution was added to terminate the reaction. The OD of each plate was measured immediately at 450 nm in a microplate reader (Thermo, MA, USA) and was recorded.

Wire-grip Test and Morris Water Maze (MWM)
The grip strength of animals was assessed by using the wire-grip test. The test was performed as previously described [34,35]. A 40cm wire was suspended tightly between two upright bars at the height of 50cm above a table. Grasp the rats' tail and place rats on the wire so that rats grasp the wire with both front paws, and observe the following to evaluate the grip strength: the time length of rats remaining on the wire (0-60s), the way of holding on (paw and tail). The nal score of each rat about the wire grip test was obtained through an average of three trials.
To evaluate the impact of AMD3100 and L-NIL on the memory performance of rats after SAH, the Morris Water Maze (MWM) test was performed on days 8-16 post-SAH, as previously described. Three days before SAH, rats were trained in the standard MWM three times per day. Each rat was permitted 60s (cut-off time) to reach the hidden platform and stayed on it for 15 s during each trial. If a rat is unable to reach the platform in 60 s, it is permitted to be guided to it for staying for 15 s. The time of each rat to reach the platform (swimming escape latency) was recorded automatically, and the analysis was carried out by using video/compute tracking software.

Brain Water Content Measurement
The wet weight / dry weight method was conducted to evaluate brain edema [36]. The animals were euthanized and decapitated immediately at 24 hours after SAH. Brains were divided into four parts immediately. They respectively are cerebellum, brainstem and left and right cerebral hemispheres. Wet weight was obtained rstly. Then the brain specimens were dried in an oven at 100°C for 72 hours and then dry weight was obtained. The percentage of water content was calculated by using the following formula: ([wet weight -dry weight]/wet weight) ×100% [37].

Neurobehavioral
The 18-point Modi ed Garcia test and the beam balance test were used to evaluate neurobehavioral de cits blindly at 24 hours after SAH. 6 tests consist of Garcia Score [38] including spontaneous activity (in the cage for 5 min); the symmetry of all four limbs to move; symmetry of forelegs outstretching (outstretching while held by the tail); climbing wall of wire cage; body proprioception after trunk touch; and response to vibrissae stimulation. The score of each test ranged from 0 to 3. The total score is between 3 withe 18 (3 worst and 18 best). The beam balance was performed by putting rats on a cramped wooden beam to walk for 1 minute. The rats obtained 3 points for walking distance ≥ 20 cm; the rats obtained 2 points for walking distance < 20 cm; the rats obtained 1 point for walking distance < 10 cm, and the rats falling before the distance of walking < 10 cm obtained 0 points.

Statistical Analyses
Statistical analyses were performed using GraphPad Prism 8 Software (San Diego, CA, USA). The student's t-test (two-tailed, unpaired) was performed for experiments with only two groups. Statistical analyses were carried out using an ANOVA (one-way or two-way) to compare with more than two groups, followed by a post hoc Tukey test. All P values were two-sided, and statistical signi cance was set at the level of P < 0.05.

CXCR4 and CXCL12 Expression Was Upregulated after SAH
As introduction mention demonstrated that CXCR4 was expressed from rodent microglia. The pathway CXCR4/CXCL12 (CXCL12 is the ligand of CXCR4) is the critical factor of the migration and recruitment of microglia. In the SAH rats, we collected the brain tissues from the base of the temporal lobe for western blot to analyze the expression of CXCR4 at 1 h, 6 h, 12 h, 24 h, 48 h, 72 h after SAH. Results showed that the expression of CXCR4 increased gradually after SAH, and the peak elevation occurred at 12 h. Then, it presented a downward trend ( ** P < 0.01 and *** P < 0.001; Fig. 2b). Meanwhile, we offered another result within 72 h after SAH that compared with the sham group. The expression of CXCL12, the ligand of CXCR4, increased gradually, with the peak at 12 h, then it followed a decreasing trend ( ** P < 0.01 and *** P < 0.001; Fig. 2c).

The Amount of M1 Microglia Increased and May Peak 24 h after SAH.
Previous studies showed that M1 polarization microglia could express marks (CD86, iNOS, and IL-1β) and release pro-in ammation factors (TNF-α and IL-6) [39,40]. And CD206 and IL-10 are the special marker and expression factors respectively of M2 polarization microglia [10,41,42]. We divided the animals into seven groups according to the speci c time point after the SAH model: sham, 1 h, 6 h, 12 h, 24 h, 48 h, 72h. Western blot was performed using the brain tissues to observe the expression levels of CD86, iNOS, and CD206. The result showed the expression of CD86 and iNOS had a rising tendency, and the peak elevation occurred at 24 h. Within 72 h after SAH, the expression level of CD206 of the experiment group had a gradually increasing trend compared with the sham group ( * P < 0.05, ** P < 0.01 and *** P < 0.001; Fig. 2). Our results suggest that the amount of M1 microglia may peak 24 h after SAH.
CXCR4/CXCL12 Was the Critical Pathway of the Migration of M1 Microglia.
We then performed the transwell experiment in vitro to explore the exact role of the CXCL12/CXCR4 pathway in the migration of M1 microglia. It is known that LPS can activate microglia polarize to M1. Firstly, we used LPS to induce BV2 microglia for 24 h for polarizing cells into the M1 phenotype. The merged image of the immunostaining for Iba-1 and iNOS carried out that LPS induced most of the BV2 cells to M1 (Fig. 3a). Additionally, the protein level of the biomarkers (CD86 and iNOS) of M1 microglia was detected by Western blot assay. The expression of CD86 and iNOS increased after LPS induced (Fig.  3b). The above results validate that the BV-2 microglia acquired M1 polarization characteristics by LPS induced.
To imitate the SAH model, the M1 microglia (LPS-induced BV-2 microglia) was exposed under oxyhemoglobin for 6 h. In this study, AMD3100 was the highly selective antagonist of CXCR4.AMD3100 or vehicle was added into the upper chamber, and CXCL12 or vehicle was added into the lower chamber. Then transwell assay was conducted on the four groups. When both chambers were added vehicle, only a small amount of M1 microglia could be migrated to the opposite surface of the transwell insert, while less M1 microglia migrated to the insert surface when AMD3100 was added to the upper chamber.
Conversely, when we added CXCL12 into the lower chamber, more M1 microglia migrate to the surface. In the nal group, CXCL12 was added into the lower chamber. Meanwhile, the AMD3100 was added into the upper chamber. We observed the migration of M1 microglia induced by CXCL12 was inhibited by AMD3100 ( ** P < 0.001; Fig. 3c).
Previous studies found that the expression of CXCL12 increased after the brain injury [43]. The results of the previous assay that CXCL12 had an increasing trend after SAH also con rmed this point. In the CNS, CXCL12 was produced mainly by activated astrocytes [44][45][46]. Thus, we cultured and seeded the primary astrocytes in the lower chambers, which were induced by oxyhemoglobin to imitate the in vitro SAH environment. The result of the transwell assay showed that in comparison with the control treatment group, the oxyhemoglobin-induced astrocytes could induce the migration of M1 microglia, and the migration phenomenon was signi cantly inhibited by AMD3100 ( ** P < 0.001; Fig. 3d).
AMD3100 (an antagonist of CXCR4) Alleviated SAH-induced In ammation and Improved the Outcome.
Elisa assay was performed to explore the expression level of related in ammation factors after AMD3100 treatment. We divided rats into three groups: sham, SAH, and SAH + AMD3100. Serum samples of each rat were collected for ELISA at 12 h, 24 h, 48 h, and 72 h after SAH modeling. The result showed that AMD3100 did not affect the sham group. Importantly, we found that the expression of in ammatory factor (IL-6, IL-1β, TNF-), secreted by M1 microglia, had a decreased trend after the AMD3100 treatment compared with the vehicle group. However, we found the expression of the above factors had not signi cantly different from the vehicle group when SAH rats were treated by AMD3100 later than 24 h ( * P < 0.05, ** P < 0.01 and *** P < 0.001; Fig. 4).
To explore the effect of AMD3100 on the prognosis of SAH rats,we performed a series of behavioral experiments. In wire-grip test, we found that the rats of treatment group had a better score than the SAH group ( # P < 0.05, ## P < 0.01 and ### P < 0.01; ** P < 0.01 and *** P < 0.001; Fig. 4d). We then performed the Morris water maze test when the muscle strength of the experimental group was not signi cantly different from that of the sham group. The SAH rats with AMD3100 treated had a shorter latency as the Fig. 4 shows ( # P < 0.05 and ## P < 0.01; ** P < 0.01 and *** P < 0.001; Fig. 4e). The above experiment proves that AMD3100 could alleviate the SAH-induced in ammation and improve the outcome.

Inhibition of Inducible NO Synthase/NO• Promoted Ferroptosis in M1 Microglia.
M1 microglia is a major microglia-type in the ischemic boundary zone within the rst 72 h post-stroke [47]. We performed Perls' Prussian blue staining to observe the ferroptosis in the tissues of the temporaloor brain. The outcomes showed no obvious higher level of non-heme iron in the temporal-oor brain within the rst 48 h after the SAH model. Then the level increased signi cantly at 72 h and showed a gradual upward trend (Fig. 5a). This outcome was in line with that the M1-polarized microglia that were highly resistant to ferroptosis, and M2-polarized microglia were highly sensitive to ferroptosis stimulation.
As we stated before, three programs were included in the cell death program of ferroptosis: iron metabolism, lipid peroxidation, and thiol regulation-the related proteins of these programs like Tfr1, Tf, Fth, Ftl, xCT, GPX4. Moreover, COX2 is a well-recognized biomarker of ferroptosis [48]. BV2 cells were used for the Vitro experiment, and We examined the expression level of the protein associated with ferroptosis by performing the western blot assay. L-NIL and Lip-1 were used to treat, respectively. Our western blot result revealed that compared with the normal group, the expression of Tfr1, Tf, Fth, Ftl, and Cox2 presented an increasing tendency after induced OxyHb while the expression of GPX4 and xCT with a trend toward decrease ( # P < 0.05, ## P < 0.01, and ### P < 0.001; Fig. 6). Compared with the vehicle group, protein expression of Tfr1, Tf, Fth, Ftl, and Cox2 showed a rising trend with the L-NIL treated, the GPX4 and xCT expression decreased. The result of Lip-1 treated group showed the opposite trend ( * P < 0.05, ** P < 0.01, and *** P < 0.001; Fig. 6).
As mentioned above, lipid peroxidation is associated with ferroptosis. C11-bodipy581 = 591 is very suitable for ratio uorescence imaging of reactive oxygen species in living cell membranes. Figure 5b showed that OxyHb promoted lipid peroxide formation with the red uorescence shifted to green uorescence of the BODIPY, indicating the lipid peroxidation rises after OxyHb induced. We also found that L-NIL facilitated the signi cant upregulation of lipid peroxidation following OxyHb induced. Signi cantly, treating L-NIL-treated microglia cells with Lip-1could reduce BODIRY-C11 oxidation following OxyHb induced (Fig. 5b).
To sum up, these data suggest that OxyHb resulted in ferroptosis of M1 Microglia while L-NIL promotes this cell death program.
Reduced Quantity of M1 Microglia Alleviated In ammation after Early Brain Injury.
According to previous studies, the expression levels of Cd86 and iNOS ref lected the M1 polarization of microglia and the count of M1 microglia correspondingly. To validate the effect of AMD3100 and L-NIL on the count of M1 microglia, we performed the following experiments. The rats were categorized into 5 groups: (1) Sham, (2) Vehicle, (3) Vehicle + AMD3100, (4) Vehicle + L-NIL and (5) Vehicle + AMD3100 + L-NIL. Then after 24 h's successful modeling, rats were euthanized. The brain tissues were used for Western blot. The result showed CD86 and iNOS had a signi cant rise tendency after SAH. The expression of CD86 and iNOS had a decreasing tendency when AMD3100, L-NIL, or both two drugs were administered ( # P < 0.05, ## P < 0.01 and ### P < 0.01; ** P < 0.01 and *** P < 0.001; Fig. 7). Hence, these data indicated that AMD3100 and L-NIL treatment could decrease the count of M1 microglia after SAH.
To further investigate the effect of AMD3100 or L-NIL on the amount of M1 microglia in the temporaloor brain after SAH, we detected M1 marker (iNOS) co-expression with microglia marker Iba-1 at 24 h after SAH. We found the amount of M1 microglia (Merge /Iba-1+) increased signi cantly at 24 h after SAH. Importantly, the amount of M1 microglia decreased with AMD3100 and/or L-NIL treatment ( * P < 0.05, ** P < 0.01 and *** P < 0.001; Fig. 7e). The result further con rmed that the AMD3100 and L-NIL treatment could decrease the amount of M1 microglia after SAH.
The in ammation factors in the sera of rats were detected using the speci c ELISA kits to con rm the effect of AMD3100 and L-NIL on reducing the in ammation. TNF-, IL-6, and IL-1β are the classic in ammation factors. Figure 8 showed that the level of in ammation factors of the vehicle group presented a signi cant upward tendency compared with the sham group. The expression level of in ammation factor compared with vehicle group had a decreasing trend after treatment with AMD3100 and L-NIL ( ** P < 0.01 and *** P < 0.001; Fig. 8). These results further demonstrated that AMD3100 and L-NIL could alleviate in ammation after SAH.

Both AMD3100 and L-NIL Attenuated Neuronal Damage in SAH-induced Injury Brain Area
To further investigate the role of AMD3100 and L-NIL in neuronal damage following SAH. We conducted a series of experiments: Nissl staining and Fluoro-Jade B (FJB) staining. In Nissl staining, compared with the living neurons in the sham group, the cell bodies of damaged neurons exhibited cytoplasmic shrunken or nuclear fragments following SAH. After AMD3100 and L-NIL treatment, these histopathological changes were alleviated (Fig. 8f). We also performed FJB staining to determine neuronal degeneration after SAH. We found that SAH resulted in the number of degenerating neurons increased in damaged brain areas. The quantity of degenerating neurons of AMD3100 and L-NIL treatment group reduced obviously compared with the vehicle SAH group ( ** P < 0.01 and *** P < 0.001; Fig. 8d). The above results showed both AMD3100 and L-NIL could prevent neuronal damage in the SAHinduced injury brain area.

Both AMD3100 and L-NIL Attenuated Neurological Dysfunctions and Brain Edema after SAH
To evaluate the effect of AMD3100 and L-NIL on neurobehavioral functions and brain edema after SAH.
In neurobehavioral functions test, there were no signi cant differences between each group before SAH.
Compared with the sham group, the neurological score of the animals in the vehicle group signi cantly decreased. We also got the observation that the neurobehavioral was markedly improved when the post-SAH animals were treated with AMD3100 and L-NIL. The result of the neurological score was given in the gure below ( * P < 0.05, ** P < 0.01; Fig. 9a).
In the brain edema test, both the animals of sham and experiment group were euthanasia 24 h after SAH.
We found the brain edema of 24 h post-SAH animals increased signi cantly by 0.58% in the left hemisphere ( ## P < 0.01), 0.54% in the right hemisphere ( ## P < 0.01), and 0.30% ( ## P < 0.01) in the cerebellum compared with the animals of sham groups, while no change was observed in the brain stem.
Our result also showed that the brain water content in the left and right hemisphere decreased after the To explore the effect of AMD3100 and L-NIL on SAH-induced neurological outcome, we performed two experiments: wire-grip test and Morris Water Maze (MWM). Animals used for experiment analysis were the same. Animals were divided into ve groups: Sham, Vehicle + SAH, AMD3100 + SAH, L-NIL + SAH, AMD3100 + L-NIL + SAH. The wire-grip test was performed on days 1-7 rstly, and then we tested the Morris Water Maze (MWM) test on days 8-16 post-SAH (n = 7; Fig. 9c). Regarding the wire-grip test, the post-SAH animals had a signi cantly declined tendency on days1-7 compared with those in the sham group. But the motor performance of the animals treated with AMD3100 and L-NIL was ameliorated compared with those in the vehicle-treated SAH group (For Vehicle group: ** P < 0.01, *** P < 0.001 vs Sham group; for AMD3100 group: # P < 0.05, ## P < 0.01, and ### P < 0.001 vs Vehicle SAH group; for L-NIL group: && P < 0.01, &&& P < 0.001 vs Vehicle SAH group. for AMD3100 and L-NIL group: $ P < 0.05, $$ P < 0.01, and $$$ P < 0.001vs Vehicle SAH group; Fig. 9d). In the MWM test, the time of the animal found the platform was recorded. The shorter time means better learning and memory function. We observed that the post-SAH animals had an increased latency compared with the animals in the sham group on days 8-16. The animals treated with AMD3100 and L-NIL both displayed a shorter escape latency signi cantly than the vehicle-treated SAH animals (Fig. 9e).

Discussion
Early studies have demonstrated that brain damage following stroke, including SAH, is strongly associated with in ammation [49][50][51]. An extensive data of clinical have linked early in ammation in SAH to poor neurological outcomes [52]. The neuroin ammatory response that SAH resulted in including the phenotype of various CNS cells (astrocytes, neurons, and microglia) from endogenous and the in ux of several leukocytes from the periphery (neutrophils, macrophages, and T-cells). Microglia, as the resident immune cells in the CNS, has versatile effectors. Microglia may polarize into two phenotypes under ischemic conditions: M1 phenotype, which could secrete proin ammatory cytokines and contribute to tissue damage, and M2 phenotype, which could phagocytose debris to repair tissue [53]. Selective regulation of different phenotypes microglia quantity and function could be an effective strategy to improve outcomes after SAH.
This study found that the microglia, particularly the M1 phenotype microglia, increased rapidly several hours after SAH. The primary sources of M1 microglia include the activation of resident microglia and chemotaxis from surrounding brain tissue [54]. The M2 phenotype microglia increased slowly in the beginning and then increased signi cantly at 24 h after SAH. Thus, the measures to reduce the quantity of M1 microglia in the early phase of SAH may be effective in alleviating the in ammation injury subsequently.
Our study showed that the expression of CXC motif chemokine ligand 12 (CXCL12) increased post-SAH. Moreover, we performed the transwell assay in vitro, and the assay demonstrated that the migration of M1 microglia could be induced by CXCL12 in vitro, and the AMD3100, the antagonist of CXCR4, could inhibit the tendency of migration effectively. This experiment suggested that the M1 microglia migration was mediated by CXCL12/CXCR4 signaling pathway. Then, in vivo experiment, rats were injected with AMD3100 at different timepoints after SAH, and we detected the expression level of the in ammation factor of M1 phenotype microglia to con rm whether disturbing CXCL12/CXCR4 pathway could cause inhibit M1 phenotype microglia recruitment to the brain injury area and reduce the in ammation factor. We found that the expression of the speci c in ammation secretion (IL-6, IL-1β, TNF-) of M1microglia had a decreasing tendency after AMD3100 treatment within 24 h after SAH. However, when SAH rats were treated by AMD3100 later than 24 h, M1 phenotype microglia had been abundantly recruited into the brain injury area, and we found the expression of the above factors had not signi cantly different from the vehicle group (Fig. 3). The bene t of AMD3100 could not be achieved. As previous studies mentioned, AMD3100 could inhibit the migration of M2 microglia after ischemia, and the effect of the inhibition increased 48 h after ischemia. And the previous studies even found that the enhanced neurogenesis and behavioral recovery were abrogated by consecutive infusing for 14 days with AMD3100 [55][56][57][58]. We further performed neurobehavioral experiments. Rats were divided into sham group, SAH group, and treatment group to perform wire-grip test and morris water maze test on consecutive seven days. The result showed that improvements in neurobehavior in the treatment group compared to the SAH group.
Consequently, our study may support the idea that inhibiting the CXCRL12/CXCR4 pathway reduces M1 microglia recruitment and in ammation in the early brain injury to improve the prognosis and highlights the importance of an effective therapeutic time window for this pathway.
Based on the above results, we naturally wondered whether further reduce the amount of M1 microglia in the early brain injury could further alleviate in ammation and improve prognosis. Kapralov Alexandr A et al. [23] found that iNOS/NO• is a potent regulator of ferroptotic death. This feature leads the M1 polarized microglia to have a high resistance to ferroptosis. Conversely, the M2 polarized microglia, which lacks iNOS, displayed a heightened sensitivity to ferroptotic death. Notably, this sensitivity from pro-ferroptosis stimulation was abolished entirely when exogenously NO• donors were added. Iron metabolism, lipid peroxidation, and thiol regulation have been considered fundamental parts of ferroptosis. In iron metabolism, the free Fe3 + in peripheral blood combined with transferrin (Tf) is extracellular, and the This function is essential for the cell to avoid oxidant injuries and is vital for cell survival [59]. GSH is an indispensable cofactor for GPX4 to operate. The cysteine is transported into the cell by cystineglutamate transporter (system XC _ ), and meanwhile, the glutamate is carried out from the cell. The cystine in the cytosol transformed to cysteine (Cys), and the latter plays an essential role in GSH synthesis [60]. The content of Cys in intracellular decrease due to blocked transport of cystine resulted in the synthesis of GSH reduced. Then the loss of GPX4 activity led to the accumulation of lipid ROS resulting in ferroptosis (Fig. 10b) [16].
Therefore, we performed in vitro experiments to detect the ferroptosis-associated factors of M1 microglia.
In vitro experiment, the protein expression (Tf, Tfr1, Fth, Ftl), associated with iron metabolism, increased after SAH modeling. This trend was inhibited by a recognized inhibitor (Lip-1) of ferroptosis. Notably, the L-NIL, the inducible (i) NOS inhibitor, reversed this inhibition and even promoted the expression. For the regulatory pathways of GPX4. The expression level of GPX4 and xCT decreased after SAH was induced. L-NIL further reduces the expression of xCT and GPX4. And the decrease in xCT caused the decline in the transport of cystine which ultimately results in a signi cant reduction of GPX4. Ferroptosis occurred with the substantial accumulation of lipid ROS due to the decrease of GPX4. We also con rmed that the ferroptosis marker Ptgs2/COX2 (prostaglandin-endoperoxide synthase 2) [61] increased after SAH, and the L-NIL led to this tendency more signi cantly. It is well known that lipid peroxidation is an essential process of ferroptosis. The content of Malondialdehyde (MDA), a lipid peroxidation marker, is measured to assess the degree of lipid peroxidation [62]. The result showed that L-NIL promoted the degree of lipid peroxidation of M1 microglia while the degree of lipid peroxidation of M1 microglia was alleviated when treated with LIP-1. More importantly, we found that the results after two drugs were treated together were similar to the consequence treated with Lip-1 lonely. Similar effects were observed in C11-BODIPY581/591 (an alternative to detecting lipid peroxide formation[63]). The above results indicate that the L-NIL, the inducible (i) NOS inhibitor, may promote the ferroptosis of M1 microglia and thus reduces its amount. We even venture to guess that LIP-1 can resist the facilitation of L-NIL, but the exact mechanics will need to be explored further in the future.
We naturally wondered whether promoting ferroptosis in M1 microglia based on based on inhibiting M1 microglia migration would lead to a better prognosis. The results showed that the number of M1 microglia after being treated with both AMD3100 and L-NIL is less than using one agent only, and the in ammation factors are signi cantly decreased. Notably, we also observed better outcomes in the group treated with both agents.
The conclusion is that reducing the quantity of M1 microglia in the early brain injury (EBI) may produce cerebral protection in a rat SAH model. The conclusion is based on three key ndings. First, the number of microglia was reduced by inhibiting migration via the CXCL12/CXCR4 pathway. Second, the L-NIL increased the susceptibility of microglia to ferroptosis via inhibiting iNOS, which, in turn, reduced the number of M1 microglia. Finally, a decrease in the number of M1 microglia resulted in a signi cant reduction in in ammatory cytokines. Our study found the idea that the combination of the inhibitor AMD3100 of CXCR4 and the inhibitor L-NIL of iNOS may ameliorate the early brain injury of subarachnoid hemorrhage and could provide a new approach for the future treatment of subarachnoid hemorrhage. However, we did not explain the speci c pathways by which L-NIL promotes ferroptosis in M1 microglia, which is also a de ciency of our study. In the future, we will further explore the mechanisms by which L-NIL promotes ferroptosis in M1 microglia. W.Q., and Y.C., T.R. and W.P. performed the Nissl staining, FJB staining, Perl's staining, the behavioral experiment, and the brain edema test and analyzed the data.
All the authors listed in the manuscript have agreed upon and reviewed the manuscript, and provided feedback.

Funding
This study was supported by the National Nature Science Foundation of China (Nos. 81571121 and 81971163).

Data Availability
All data generated or analyzed during this study are included in this article and its supplementary information les.

Ethics Statement
Our study was approved by the Ethical Committee of Soochow University.

Consent to Participate
Not applicable.

Consent for Publication
Not applicable.   Disturbing CXCL12/CXCR4 pathway in the rst 24h after SAH could inhibit the in ammation responses and improve the outcome. Serum was collected from the SAH rats at speci c time after model: 12h, 24h, 48h, and 72h. The expression of in ammation factors at speci c time points after AMD3100 or vehicle treatment was measured. The activity levels of the in ammatory cytokines :(a) TNF-α, (b) IL-6, (c) IL-1β were measured using ELISA kits. *P < 0.05, **P < 0.01, ***P < 0.001 versus Vehicle, each group set contained 6 samples. The SAH rats was used to performed wire-grip test and morris water maze test on the condition with and without AMD3100 or vehicle treatment.   Quantity of M1 microglia was decreased by the treatment of AMD3100 and L-NIL. Rats were sacri ced 1 day after SAH. (a)The brain of Sham, SAH+Vehicle, SAH+AMD3100, SAH+L-NIL, and SAH+AMD3100+L-NIL groups were obtained. Western blot for iNOS and CD86 in temporal-oor brain tissue was performed; each group set contained 5 samples (n=5). The expression level of iNOS in temporal-oor brain tissue was detected under the condition with or without treatment of AMD3100 or L-NIL. Data are expressed as the ratio of iNOS/Tubulin (b). The expression level of CD86 in temporal-oor brain tissue was detected under the condition with or without treatment of AMD3100 or L-NIL. Data are expressed as the ratio of CD86/Tubulin (c). #P < 0.05, ##P < 0.01, and ###P <0.001 versus Sham group; *P < 0.05 and **P < 0.01 versus Vehicle group. Rats were sacri ced 24h after the SAH model. The brain slices of Sham, SAH+Vehicle, SAH+AMD3100, SAH+L-NIL, and SAH+AMD3100+L-NIL groups were obtained. (d) Immuno uorescent double staining was performed for the brain slices: iNOS (green) and Iba-1 (red) in the presence or absence of AMD3100 or L-NIL. The nuclei were stained with DAPI. Double positive cells indicated by arrows are shown at higher magni cation in the left-bottom panels. (e) Data were expressed as the percent of Merge/Iba-1; *P < 0.05 and **P < 0.01 versus Vehicle group; the scale bars represent 100 μm.

Figure 8
Both AMD3100 and L-NIL alleviated in ammation and rescued the neuronal damage after SAH. Serum was collected at 24 h after the SAH model with or without AMD3100 or L-NIL treatment. The activity levels of the in ammatory cytokines: (a) IL-1β, (b) IL-6, and (c) TNF-α were measured using ELISA kits. *P < 0.05, **P < 0.01, and ***P < 0.001 versus Vehicle group, each group set contained 6 samples (n=6). The brain sections were obtained from Sham, SAH+Vehicle, SAH+AMD3100, SAH+L-NIL, and Page 34/35 SAH+AMD3100+L-NIL groups respectively at 24h after SAH to performed FJB (d) and Nissl (f) staining.
(e) The data represented the quantity of FJB+ cells per square millimeter; **P < 0.01 and ***P < 0.001 versus Vehicle group; the scale bars represent 100 μm and 20 μm, respectively.

Figure 10
The mechanism of M1 microglia decreasing after SAH. (a) The schematic of M1 microglia migration. (b) The schematic of M1 microglia ferroptosis.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.