Deletion of Bak1 alleviates microglial necroptosis and neuroinflammation after experimental subarachnoid hemorrhage

Microglial necroptosis exacerbates neurodegenerative diseases, central nervous system (CNS) injury, and demonstrates a proinflammatory process, but its contribution to subarachnoid hemorrhage (SAH) is poorly characterized. BCL‐2 homologous antagonist‐killer protein (Bak1), a critical regulatory molecule of endogenous apoptosis, can be involved in the pathologic process of necroptosis by regulating mitochondrial permeability. In this study, we revealed microglia undergo necroptosis after SAH in vivo and vitro. Western blot revealed that Bak1 was elevated at 24 h after SAH. Knocked down of Bak1 by adeno‐associated virus attenuates microglial necroptosis, alleviates neuroinflammation, and improves neurologic function after SAH in mice. Furthermore, oxyhemoglobin (10 μM) induced necroptosis in BV2 microglia, increasing Bak1 expression and mediating proinflammatory phenotype transformation, exacerbating oxidative stress and neuroinflammation. Abrogating BV2 Bak1 could reduce necroptosis by down‐regulating the expression of phosphorylated pseudokinase mixed lineage kinase domain‐like protein (p‐MLKL), then down‐regulating proinflammatory phenotype gene expression. RNA‐Seq showed that disrupting BV2 Bak1 down‐regulates multiple immune and inflammatory pathways and ameliorates cell injury by elevating thrombospondin 1 (THBS1) expression. In summary, we identified a critical regulatory role for Bak1 in microglial necroptosis and neuroinflammation after SAH. Bak1 is expected to be a potential target for the treatment strategy of SAH.


| INTRODUC TI ON
Subarachnoid hemorrhage (SAH) is a severe subtype of stroke produced mainly by intracranial artery rupture with high mortality and fatality rates. Despite massive efforts worldwide, challenges remain to relate to early management and long-term outcomes of SAH.
Early neuroinflammation following SAH is one of the top courses for poor prognosis, while clinical trials have failed to provide consistent evidence for the use of anti-inflammatory agents in SAH patients.
Thus, it is crucial to provide insight into its pathogenic mechanisms and develop innovative therapeutic strategies for SAH patients (Macdonald & Schweizer, 2017;Pang et al., 2019).
Cell necrosis promotes neurologic deficits and exacerbates brain injury following SAH by eliciting a severe proinflammatory response, as opposed to apoptosis, which is typically considered to be immunosilent. A detrimental role for programmed cell deaths during early brain injury following SAH, including ferroptosis, pyroptosis, and necroptosis, were recently reported Li et al., 2021;Xu, Hong, et al., 2021). These forms of cell death have morphologic features with necrosis, such as membrane rupture, organelle swelling, and release of cell contents; their molecular regulatory mechanisms are distinct and are intimately related to the inflammatory response (Yuan, Amin, & Ofengeim, 2019). Among them, necroptosis is triggered by the receptor-interacting serine/threonine-protein kinase 1 (RIPK1), receptor-interacting serine/threonine-protein kinase 3 (RIPK3), mixed lineage kinase domain-like protein (MLKL) cascade reaction, resulting in the release of damage-associated molecular patterns (DAMPs), and ultimately resulting in inflammation (Dionisio et al., 2020;Galluzzi et al., 2018).
BCL-2 homologous antagonist-killer protein (Bak1), a mitochondrial apoptosis effector molecule in the BCL-2 protein family, has recently emerged to play a vital role in regulating pyroptosis, and activation of Bak1 inhibits IAP proteins promoting caspase-8mediated activation of interleukin-1 beta (IL-1β) and inducing NLRP3 inflammasome-mediated caspase-1-dependent IL-1β maturation, promotes pyroptosis and aggravates inflammatory response (Vince et al., 2018). Additionally, it has been demonstrated that Bak1 plays a critical function in the regulation of mitochondrial outer membrane permeability (MOMP) downstream of necroptosis and enhances mitochondrial permeability transition pore (MPTP)-dependent necrosis (Karch et al., 2015). However, the precise mechanism by which Bak1 contributes to the inflammatory response in SAH necroptosis remains unknown, and additional investigation is required to elucidate.
In this study, we explored the regulatory mechanisms of Bak1 in the necroptosis and neuroinflammation of microglia in experimental subarachnoid hemorrhage. We found that the knockout of Bak1 downregulated the phosphorylated pseudokinase mixed lineage kinase domain-like protein (p-MLKL) content alleviated the necroptosis and proinflammatory phenotype of microglia, and reduced neuroinflammation. RNA-Seq verified the potential mechanism of Bak1 regulating necroptosis, and it was determined that Bak1 could reduce the inflammatory response and cell injury by elevating thrombospondin-1 (THBS1), providing a new concept and therapeutic strategy for the treatment of neuroinflammation after subarachnoid hemorrhage.

| Animals
All experimental animal procedures involved in this study are ap-  the cages was controlled at 55-65%. The constant temperature inside the cages was maintained at 24-25°C. The animals were exposed to a 12 h cycle of light, were protected from light, and rested quietly at night. The animals all have free access to fresh feed and drinking water. The grouping and number of animals required for each experiment are shown in Figure S1. The study was not preregistered. Each mouse had an ear-tag and an internal ID. Mice were arbitrarily assigned to the experimental groups. No calculation was performed to predetermine the sample size.

| Animal experimental design
Experiment 1 ( Figure S1): A total of 37 mice were used, including 11 in Sham group and 26 in SAH group. The mortality rate in the Sham group was 0, while five mice humanely killed after successful establishment of the SAH model, and all died at 48 and 72 h. To examine the changes of Bak1, RIPK1, and RIPK3 during early brain injury in subarachnoid hemorrhage, Mice used for protein abundance assay were divided into four groups: Sham (n = 5), SAH 24 h (n = 5), SAH 48 h (n = 5), and SAH 72 h (n = 5), and the injured lateral hemispheres were harvested after modeling for subsequent validation. Immunofluorescence staining was used to determine the colocalization of MLKL with microglia in the 24 h group (Sham n = 3, 24 h n = 3), pyridinium iodide staining to detect cell necrosis and to detect co-localization with microglia in the 24 h group (Sham n = 3, 24 h n = 3). Experiment 2 ( Figure S1): Targeted knockdown of mice microglia Bak1 protein using pAAV-CD68p-Bak1-shRNA adenoassociated virus (shanghai genechem Co., Ltd. CN). Bak1 target sequence is 5'-GCTCT CAT CGG AGA TGA TATT-3′, NC negative control sequence is 5′-TTCTC CCC GA A CGT GTC ACGT-3′. In order to reduce the probability of peripherally infiltrating CD68+ macrophages contracting adeno-associated virus (because of CD68p's non-specific promoter), mice were injected with the adenoassociated virus (3 μl) in the lateral ventricle and then monitored for viral expression for 3 weeks. Total number of mice was 48; different C57BL/6J male mice groups were arbitrarily selected after infection for subsequent experimental validation. WB detection of RIPK1, RIPK3 expression in the ipsilateral cerebral hemisphere 24 h after SAH molding, NC group (n = 5), Bak1-KD (n = 5). ELISA for proinflammatory factor concentration (IL-6, IL-1β, and TNFα), NC group (n = 5), and Bak1-KD (n = 5). Co-localization of PI and microglia IBA1 were detected by immunofluorescence, NC group (n = 3), and Bak1-KD group (n = 3). Co-localization of THBS1 and microglia IBA1 were detected by immunofluorescence, NC group (n = 3), and Bak1-KD group (n = 3).
We excluded a total of 10 mice based on SAH grading scores. In Experiment 1, five mice died after successful establishment of the SAH model. We reselected new healthy mice to reestablish the subarachnoid hemorrhage model. In Experiment 2, a total of 13 mice in the SAH 24 h AAV-NC group were replaced, among them, there were two in the immunofluorescence group, one in the pyridinium iodide group, and total 10 in the WB and ELISA groups. Seven mice were substituted in the SAH AAV-Bak1-KD group, including one in the immunofluorescence group and total six in the WB and EILISA groups. The above was supplemented by recreating the subarachnoid hemorrhage model.
Power of test have been assessed for our data as previously described (Spay et al., 2018), Partial Eta Squared of intercept was 0.917 (>0.85), which meant group sample sizes of 5 could achieve 91.7% power to detect differences among each group.

| Subarachnoid hemorrhage model
Induction of SAH model by modified single-clamp puncture method . To minimize surgical pain in animals, we chose isoflurane and pentobarbital sodium for surgical anesthesia.
The primary process is: anesthesia was induced in experimental mice with 4-5% isoflurane, followed by intraperitoneal injection of 1% pentobarbital sodium (100 mg/kg) to create continuous anesthesia. The mice were placed supine on the operating table with a satisfactory level of anesthesia (no significant pain from skin incision). After skin preparation and disinfection, the skin of the neck was incised along the anterior midline of the neck using a sharp surgical blade (an incision of approximately 1 cm in length was sufficient). Micro forceps bluntly separated the subcutaneous fat, deep fascia, and muscle tissue, preserved the hyoid bone, pulled the hyoid bone to the side, exposed the right common carotid artery, separated the vessels and peripheral nerves, and then freed the right external carotid artery, down and on its serious side, separated the right internal carotid artery. After resistance is felt at the bifurcation of the anterior and middle cerebral arteries, the puncture wire is advanced 2 mm further through the vessel and immediately returned. We also observed whether the mice showed typical Cushing's reaction, which was used as an indirect criterion to judge the success of the SAH model. After the puncture is completed, the puncture wire is quickly withdrawn. The blind end of the external carotid artery vessel is ligated, the skin incision is sutured, and the skin is disinfected with dilute iodophor.
Mice in the sham group were performed the same surgical procedure. Still, instead of puncturing the vessel, the puncture line was immediately returned after resistance was felt at the bifurcation of the anterior and middle cerebral arteries. The principle of aseptic operation was followed during the operation and the mice were placed in a rewarming chamber at 25°C and observed and recorded every 15 min until they were revived from anesthesia and housed in separate cages by the group. After surgery, we replaced the mice with new soft bedding to make them more comfortable. The animal operations were carried out by qualified and experienced experimenters to guarantee that no unnecessary pain or distress was caused as a result of the procedures. We did not utilize any analgesic or anti-inflammatory medicines, which could have influenced the experimental outcomes because of their antiinflammatory properties (Stein & Kuchler, 2013). Additionally, animals were deeply anesthetized with 1% pentobarbital sodium to reduce the suffering before killing. The assessment of SAH grading score was determined after SAH by an investigator who was blinded to experimental groups (Sugawara et al., 2008). Only mice with scores higher than 8 were analyzed.

| Neurologic score
The neurologic function of mice was scored using the modified Garcia scale and the beam balance tests (Xie et al., 2020). The modified Garcia scale assessment included tests of spontaneous activity (0-3 scores), spontaneous movement of limbs (0-3 scores), forelimb extension (0-3 scores), climbing ability (1-3 scores), post-tentacle response (1-3 scores), and trunk contact (1-3 scores). Balance beam test, in which the ability of mice to walk on a circular wooden beam within 1 min, was tested (0-4 scores). To avoid any subjective bias on the experimental results, the neurologic scoring procedure was performed by two researchers who were unaware of the experimental design information. and necrostatin-1 were added to cell culture medium for induction for 30 min, followed by addition of oxy-Hb to stimulate microglia for modeling. All the reagents were persist throughout the experiment.

| Western blot
The harvested cells or animal tissues were lysed on ice with RIPA lysis solution for 30 min and then centrifuged at 12 000 g for 10 min at 4°C. The protein supernatant was added to the loading buffer and denatured at 100°C for 5 min. The samples were added to 10% SDS-PAGE gel for electrophoresis, after which the proteins were transferred to PVDF membrane, 5% skim milk powder was blocked for 1 h, and then overnight at 4°C with the following pri- was selected and incubated for 1 h; the membrane was then washed with TBST 10 min, three times. X-ray film and ImageJ software (NIH) were used for detecting and quantifying, respectively.

| Propidium iodide staining
Intraperitoneal injection of pyridinium iodide (PI 10 mg/kg #ST511 Beyotime) 1 h before killing mice after SAH 24 h, brain tissue was obtained after cardiac perfusion with ice PBS and immediately em-

| Cell viability assays
Cells were seeded at a density of 5 × 10 3 cells/well in 96-well plates.
After overnight incubation, fresh medium containing oxyhemoglobin was added, and the plates were maintained for 0, 12, and 24 h.

| ELISA
TNFα ELISA kit, IL-6 ELISA kit, and IL-1β ELISA kit were obtained from NEOBIOSCIENCE CN (#EMC001b.96, #EMC102a.96, and # EMC004.96); subsequent testing follows manufacturer's protocols. The same number of BV2 cells were seeded into the culture well plates, and the cell supernatant was collected for detection after 24 h of hemoglobin stimulation. The mice's brain tissues were obtained from SAH molds for 24 h, lysed by RIPA, and BCAmeasured protein concentrations. Brain tissue lysates with the same protein content were taken for subsequent assays. The samples were added to the test wells, and the standard wells were set up simultaneously. After the operation, according to the kit's instructions, the OD450 values were measured, and the contents of IL-6, IL-1β, and TNFα were calculated according to the standard curve. The experimenters performing the ELISA were not aware of the experimental groups.

| RNA-Seq
Microglia were seeded in a 6-well plate, and after 12 h of com-

| Transcriptome analysis
The raw fastq files were trimmed using trim galore (version 1.18) to remove adaptor sequences and low-quality reads. Then, FastQC (version 0.11.9) was used for quality control. The remaining reads were aligned to the GRCm38 mouse genome using HISAT2 (v2.2.0) with default parameters and filtered with samtools (version 1.10, parameters used: samtools view-F 1804 -f 2 -q 30). Gene counts were calculated from the mapped reads using featureCounts (v2.0.1) with the Ensembl gene annotation (version mm10). Subsequently, TPM (Transcripts Per Kilobase of exon model per Million mapped reads) in each gene was calculated for subsequent analysis. Differential expressed genes (DEGs) was evaluated using DESeq2 package in R (version 4.2.0), using adjusted p < 0.05 and |log2FC| > 1 as cutoffs to define the DEGs.

| GSEA analysis
GSEA uses the expression matrix obtained by cluster profile package (version 4.05) in R software to carry out GSEA based on biological processes in gene ontology (GO-BP) and based on the Kyoto Encyclopedia of Gene and Genome (KEGG) signal pathway, respectively, to obtain the enrichment results of differentially expressed genes in GO functional annotation and KEGG pathway obtained by differential analysis between NC group and KD group. The gene sequence was arranged according to the descending order of multiple gene expression changes (foldchange, FC) in the Bak1-KD group compared with the NC group. The statistical method was Fisher's exact probability test (p < 0.05), and error detection rate (false discovery rate, FDR <0.05) was used to correct.

| Statistical analysis
GraphPad Prism 8 was used for statistical analysis. We used an unpaired two-tailed t-test to compare two groups. Multiple comparisons using a one-way ANOVA followed by Tukey's honest significant difference (Tukey-HSD) for post hoc means comparisons or two-way ANOVA followed by Fisher's LSD test as appropriate. Bar graphs were described as mean ± Standard error of the mean (SEM) of at least three independent experiments. Normality was assessed using Shapiro-Wilk normality test, statistically significant changes relative to a negative control were represented with p < 0.05. A test for outliers was not conducted on the data.

| Experimental subarachnoid hemorrhage induces necroptosis of microglia
Previous studies have reported that subarachnoid hemorrhage induces neuronal necroptosis and elevates the expression of RIPK1/ RIPK3 (Chen et al., 2017). However, the microglia necroptosis after subarachnoid hemorrhage is unclear. We established 24 , 48 , and 72 h mouse subarachnoid hemorrhage models during early brain injury. WB detected the expression of necroptosis-related proteins RIPK1/RIPK3, and similar to the reported results, RIPK1 was elevated 24 h after bleeding (Figure 1b), but the expression of RIPK3 was not significantly up-regulated ( Figure 1b). PI staining is commonly used to label necrotic cells, and PI-positive cells were found in the area of injury following subarachnoid hemorrhage. Immunofluorescence suggests co-localization of microglia marker IBA1 with PI, indicative of microglial cell necrosis (Figure 1c). MLKL is a functional executive protein of necroptosis, and immunofluorescence demonstrated that MLKL co-localized with IBA1 and NeuN after 24 h of subarachnoid hemorrhage, but not with GFAP in astrocytes. (Figure 1e). The sham group did not observe the phenomenon, suggesting that microglia and neuron suffered necroptosis. Neurologic function scores decreased in mice after subarachnoid hemorrhage (Figure 1d), suggesting that necroptosis of microglia may be associated with reduced neurologic function scores.

| In vivo knockdown microglia Bak1 protein attenuates necroptosis and neuroinflammation
It has been documented that Bak1 contributes to amplifying the biologic process of necroptosis (Rohde et al., 2017). We hypothesize that disrupting Bak1 may diminish microglia necrosis and its downstream inflammatory response in subarachnoid hemorrhage. The adeno-associated virus was used to precisely inhibit Bak1 protein production in mouse microglia; Following modeling, we chose brain tissue with similar levels of SAH hemorrhage between the NC control and Bak1 knockdown groups, reducing subsequent data disparities caused by varying degrees of bleeding. (Figure 2b).
The number of PI and IBA1 co-localized microglia in the Bak1-KD group mice was lower than in the NC group (p < 0.05) (Figure 2c

| Oxyhemoglobin stimulation induces necroptosis and oxidative stress in microglia
The role of microglial necroptosis in neuroinflammation following a subarachnoid hemorrhage is currently unknown. To investigate the occurrence of necroptosis, we stimulated BV2 microglia with oxyhemoglobin. PI staining revealed that microglia necrosis increased after 24 h
We utilized zVAD-FMK, a commonly used necroptosis inducer, to stimulate microglia together with oxyhemoglobin and discovered that it increased IL-6, IL-1β, and TNFα in microglia, whereas Nec-1, a necroptosis inhibitor, inhibited the expression of these inflamma-

F I G U R E 3
Oxyhemoglobin stimulation induces necroptosis and oxidative stress in microglia. (a) PI staining to detect microglial cell necrosis after BV2 microglia treated with 10 μM oxyhemoglobin, PI (red), nuclei: Hoechst 33342 signal (blue). Scale bar =100 μm. Magnification ×100. n = 3 (independent cell culture preparations) in each group. (b) RIPK1, RIPK3, and p-MLKL levels after BV2 microglia treated with 10 μM oxyhemoglobin. The vertical ordinates represent the relative densities of these proteins, n = 3 (independent cell culture preparations) in each group., **p < 0.01 versus the CON group, NS = no significant versus the CON group. (c) The intracellular ATP content was measured after 24 h of oxyhemoglobin stimulation, and the results were displayed as a ratio, n = 3 (independent cell culture preparations) in each group. **p < 0.01 versus the CON group. The cell viability after oxyhemoglobin stimulation was assessed using the CCK8 kit, and results were expressed as the ratio compared with the CON group, n = 3 (independent cell culture preparations) in each group. **p < 0.01 versus the CON group, n = 3 (independent cell culture preparations). (d) Fluorescence microscopy analysis of DCFH-DA fluorescence after 10 μM oxyhemoglobin-treated BV2 microglia; nuclei: Hoechst 33342 signal (blue); ROS: DCFH-DA signal (green). Scale bar =100 μm. Magnification ×100. n = 3 (independent cell culture preparations) in each group.

| In vitro knockout of Bak1 alleviates microglia necroptosis and oxidative stress
Bak1 is a critical molecule in regulating mitochondrial MPTP in the BCL-2 family proteins associated with necroptosis mitochondrial damage (Tischner et al., 2012). WB data showed that Bak1 expression elevated 24 h after oxyhemoglobin stimulation in microglia, and that Bak1 expression increased further after zVAD-FMK induction ( Figure 5a). As Bak1 and MLKL can localize to mitochondria and mediate mitochondrial damage (Karch et al., 2015), it is unclear whether Bak1 interacts with RIPK3. We performed co-immunoprecipitation tests on RIPK3 protein and found that RIPK3 can pull down RIPK1 and Bak1 with oxyhemoglobin stimulation. The same results were obtained by CO-IP assay of Bak1 protein (Figure 5b). The results suggest that Bak1 may be involved in the biologic processes associated with necroptosis. We constructed a Bak1-KO microglia cell line, and PI assay revealed that deletion of Bak1 reduced the number of oxyhemoglobin-induced PI-positive cells in microglia (Figure 5c).
CCK8 assay reveals that Bak1 deletion increases cell survival that declines following oxyhemoglobin stimulation, while ATP assay implies that Bak1 deletion enhances mitochondrial function in microglia ( Figure 5d). Also, as a downstream damaging mechanism of necroptosis, reactive oxygen was significantly down-regulated in Bak1-deleted microglia (Figure 5e). The p-MLKL content was measured utilizing WB, and p-MLKL was found to be down-regulated in oxyhemoglobin+zVAD-FMK stimulation at the Bak1-KO cell group (p < 0.05), whereas RIPK1, RIPK3 were not significantly altered.
However, in the CON group, no change in p-MLKL was observed ( Figure 5f). These data suggest that deletion of Bak1 alleviates necroptosis and oxidative stress in BV2 cells.
Interestingly, when we evaluated the transcription of proinflammatory factors in Bak1-KO cells in the presence of zVAD-FMK, we found no difference in the production of IL-1β and TNFα compared to the DMSO group. In contrast, IL-6 expression was altered in the same way that it was in WT BV2 cells (p < 0.05) (Figure 6c), indicating that when Bak1 is deleted, it can inhibit the further elevation of necroptosis-induced IL-1β and TNFα. However, we did not observe the effect on IL6 expression. Overall, the knockout of microglia Bak1 attenuated the proinflammatory phenotype and neuroinflammation.

F I G U R E 6
Knockout of microglia Bak1 attenuates proinflammatory phenotype transformation and neuroinflammation. (a) RT-qPCR detection of Irg1, Gpr84 mRNA expression levels in CON and 24 h of 10 μM oxyhemoglobin+30 μM zVAD-FMK treatment with WT BV2 and Bak1-KO BV2. RT-qPCR results were normalized using β-action, and the fold change was calculated with the 2 −ΔΔCt method. Error bars represent mean ± SEM of n = 3 (independent cell culture preparations), and p values are calculated by two-way ANOVA followed by Fisher's LSD post hoc test. ****p < 0.001. Bak1-KO BV2 versus WT BV2. (b) ELISA detection of the supernatant concentration of IL-6, IL-1β, and TNFα in 10 μM oxyhemoglobin treatment with WT BV2 and Bak1-KO BV2; Error bars represent mean ± SEM of n = 3 (independent cell culture preparations). p values are calculated by unpaired two-tailed t-test. *p < 0.05, **p < 0.01. Bak1-KO BV2 versus WT BV2. (c) RT-qPCR detection of IL-6, IL-1β, and TNFα mRNA expression levels in response to 24 h of 10 μM oxyhemoglobin+30 μM zVAD-FMK treatment in Bak1-KO BV2. Error bars represent mean ± SEM of n = 3 (independent cell culture preparations), and p values are calculated by unpaired two-tailed t-test. *p < 0.05. NS = no significance. (d) RT-qPCR detection of Ccr5, Ccl5, Ccl2, Cxcl2, and Nos2 mRNA expression levels in CON and 24 h of 10 μM oxyhemoglobin treatment with WT BV2 and Bak1-KO BV2; p values are calculated by two-way ANOVA followed by Fisher's LSD post hoc test. Error bars represent mean ± SEM of n = 3 (independent cell culture preparations) in each group. **p < 0.01, ***p < 0.001, and ****p < 0.0001, NS = no significant.  Figure 8b). Based on the GSEA results of up-down differential genes at 12 h, a network of biological processes and genes in the top 30 GO was created. We discovered that multiple biological processes linked to necroptosis and tumor necrosis factor in the network were all connected to the gene THBS1, so we speculated that THBS1 played an essential role after Bak1-KD (Figure 7c).
In an oxyhemoglobin-stimulated model, WB demonstrates increased THBS1 expression after Bak1-KO. THBS1 expression was down-regulated after zVAD+oxyhemoglobin induction, whereas the necroptosis inhibitor Nec-1 increased THBS1, indicating that necroptosis down-regulates THBS1 (p < 0.05) (Figure 7d). At the same time, the immunofluorescence of mice indicated that the number of THBS1 and IBA1 co-localization cells was increased after interfering with microglia Bak1 protein (Figure 2g). We

| DISCUSS ION
In this study, we explored some of the mechanisms by which Inflammation may play a significant role in both the acute and chronic phases of neural injury associated with SAH, according to emerging evidence (Chaudhry et al., 2019). As the earliest events in pathophysiology, molecules from extravasated blood and damaged brain appear to be the primary initiators of the inflammatory cascade, accompanied by the activation and infiltration of immune cells at the site of injury (Schneider et al., 2018); this vicious cycle of inflammation almost certainly contributes to nearly every mechanism of SAH, including apoptotic or necroptosis Yuan, Yu, et al., 2019). Necroptosis is a proinflammatory programmed death with morphology similar to cellular necrosis and can be induced by exogenous ligands and intracellular stress, resulting in caspaseindependent cell death . Following subarachnoid hemorrhage, necroptosis accelerates early brain injury, exacerbates disruption of the blood-brain barrier, increases brain edema, exacerbates neuroinflammation, and damages synapses, resulting in neurologic deficits Xu, Cai, et al., 2021;Yang et al., 2018;Yuan, Yu, et al., 2019 spreading inflammation is a key contributor to secondary brain injury after SAH (Heinz et al., 2021). Consistent with our finding in vivo, oxyhemoglobin-stimulated BV2 microglia increased p-MLKL and the number of PI-positive microglia. Thus, blood components may operate as a predisposing factor for microglial necroptosis. Previous study has reported that released chemokines from microglia can induce infiltration of peripheral immune cells into the damaged area. The chemokines released from microglia after proinflammatory transformation can induce infiltration of peripheral immune cells into the damaged area (Shields et al., 2020). Additionally, necroptosis often accompanies definite oxidative stress, further aggravating early brain injury after SAH (Li et al., 2020;Shindo et al., 2013). In our results, the necroptosis inducer zVAD-Fmk coupled with oxyhemoglobin enhanced the reactive oxygen species concentration, proinflammatory factors IL-6, IL-1β, and TNFα, in comparison to WT BV2 cells treated with oxyhemoglobin. These findings strongly suggest necroptosis of microglia following SAH promotes microglia proinflammatory phenotype transformation, aggravating early neuroinflammation and oxidative stress, which is similar to chronic inflammatory neurological diseases (Mifflin et al., 2021).
Recently, a novel concept of necroptosis has been proposed, meaning a combination of apoptosis and necrosis (Karch et al., 2015 RNA-Seq data provided us with new insights. We found that a broad range of proinflammatory biological processes, oxidative stress, and biologic processes related to mitochondrial membrane potential were altered after Bak1 depletion, consistent with the results obtained in vitro. While some of the biologic processes related to ligand-receptor signaling and regulation of cellular junctions are also altered, we speculate that disturbance of Bak1 can affect the expression of some downstream exocytotic proteins. THBS1, a glycoprotein that regulates macrophage migration and extracellular matrix alterations, was increased after the Bak1 knockout. Recently, THBS1 was reported as a key regulator of the microglial inflammatory response (Ju et al., 2020;Xiao et al., 2018). Although some studies show that it can serve as a prognostic biomarker (Shen et al., 2015), but its specific role in intracranial function was unreported. Deficiency of THBS1 gene in TBI may exacerbate neurologic impairment in mice, suggesting its important role in neurologic protection (Cheng et al., 2017). This evidence suggests its potential as a therapeutic target for SAH. Our data further suggested that the necroptosis inducer zVAD-FMK down-regulated THBS1 expression. Interestingly, after knocking down THBS1, the expression of proinflammatory factors was partially reversed. This appears to explain the altered biologic processes and receptor-related signals observed in GSEA analysis, suggesting that THBS1 is a downstream molecule of Bak1 regulation of necroptosis-mediated neuroinflammation. Based on these results, it may have therapeutic value in SAH. Although we discovered that the THBS1-related receptors CD47, CD36, and Tgfb1 transcriptional changes in the depletion of Bak1 (data not shown), the particular target of THBS1 in controlling necroptosis remains unknown, and more in-depth regulatory mechanisms need to be further investigated.
F I G U R E 7 GSEA (Gene set enrichment analysis) analysis of altered GO-BP (Gene Ontology-Biological Processes) following Bak1 interference and identification of the effect of Bak1 on the high contributing gene Thbs1. (a) DEGs heatmap in microglia exposed to 10 μM oxyhemoglobin stimulation for NC and Bak1-KD at CON, 12 and 24 h time points. K-means splits the significantly altered genes into two clusters. (b) GO-BP of Bak1-KD versus NC using GSEA analysis: The graph shows the 12-h oxyhemoglobin stimulation and the 24-h oxyhemoglobin stimulation. Red indicates up-regulated biological processes, while blue indicates down-regulated. Color bar: -log10 (p-value). NES: Normalized enrichment score. (c) THBS1 is involved in the GO biologic processes network. The yellow circles represent the altered biologic processes caused by Bak1-KD. The green circles represent genes involved in biologic processes, and the size of the circles indicates the number of biologic processes. (d) THBS1 levels in Bak1-KO, zVAD-FMK, and Nec-1 after 24 h of oxyhemoglobin stimulation. The vertical ordinates represent the relative densities of the protein, n = 3 (independent cell culture preparations) in each group. The unpaired two-tailed t-test was used for comparison between the two groups. *p < 0.05. (e) The intracellular ATP content was measured after 24 h of oxyhemoglobin stimulation, and the results were displayed as a ratio. **p < 0.01, ***p < 0.001, the cell viability after oxyhemoglobin stimulation was assessed using the CCK8 kit, and results were expressed as the ratio of CCK8. *p < 0.05, **p < 0.01 NS = no significant, versus NC group, n = 3 (independent cell culture preparations) in each group. (f) RT-qPCR detection of IL-6, TNFα, and IL-1β mRNA expression levels in response to 24 h of 10 μM oxyhemoglobin treatment with BV2 NC, Bak1-KO NC, and Bak1-KO + THBS1-KD, n = 3 (independent cell culture preparations) in each group. p values are calculated by t-test. **p < 0.05, ***p < 0.01, and ****p < 0.0001. NS = no significant.

| CON CLUS ION
Collectively, deletion of microglial Bak1 reduces oxyhemoglobininduced necroptosis and proinflammatory phenotype transformation, as well as ameliorates necroptosis and neuroinflammation in mice with experimental SAH. We also demonstrated that THBS1 could be a potential downstream molecule of Bak1-related microglial necroptosis. Bak1 is expected to be a potential molecular target for SAH treatment.

AUTH O R CO NTR I B UTI O N S
QXC designed the study and prepared the manuscript. QXC, TQK, and ZLH performed the experiments. QXC, and ZLH completed bioinformatics analysis, QXC, and TQK contributed equally to this work.
KCH, XYK, and ZLF were involved in experiment performance and data collection. YSG, PJH, and JY were responsible for the supervision of the entire project and were involved in the study design, data interpretation, manuscript preparation, and funding. All authors read and approved the final manuscript.

ACK N OWLED G M ENTS
We thank all the members of the Laboratory of Neurological Diseases and Brain Function, The Affiliated Hospital of Southwest Medical University, Luzhou, China. The graphical abstract created with BioRe nder.com.
All experiments were conducted in compliance with the ARRIVE guidelines.

CO N FLI C T O F I NTE R E S T
No potential conflict of interest was reported by the authors.

O PE N R E S E A RCH BA D G E S
This article has earned an Open Data badge for making publicly available the digitally-shareable data necessary to reproduce the reported results.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets used and/or analyzed during this current study are available from the corresponding author on reasonable request.