Inhibition of BACE1 attenuates microglia-induced neuroinflammation after intracerebral hemorrhage by suppressing STAT3 activation

Elevated BACE1 expression in microglia induced by ICH

BACE1 protein levels and activity in microglia were shown to be increased in several neurological diseases [12, 17, 18]. Here, we investigated the location and expression of BACE1 in microglia following ICH-mediated acute injury. We found that activated microglia from mouse brains 3 d post-ICH showed strong colocalization of BACE1 with ionized calcium-binding adapter molecule 1 (IBA-1) (Figure 1A). We isolated microglia (CD45intCD11b+) in the striatum from mouse brains at different time points after ICH modeling (Figure 1B) and measured the temporal expression of microglial Bace1. The expression of BACE1 peaked rapidly at 1 day post-ICH and persisted until 3 days, while the mRNA level showed no significant difference with baseline at 7 days after ICH (Figure 1C). Consistently, Western blotting also indicated that the protein level of microglial BACE1 significantly increased at Day 1 and Day 3 post-ICH (Figure 1D, 1E). Hence, BACE1 was elevated in ICH-affected microglial cells at an early stage.

Figure 1. BACE1 increased in microglia after ICH. (A) Representative immunofluorescent images of BACE1 and the microglial marker IBA1 in 3-day ICH brain samples. (B) Gating strategy of microglia (Zombie-Ly6G-CD45intCD11b+) (CD45intCD11b+) sorting after ICH. (C) Time course of Bace1 gene expression in FACS-isolated microglia. (D) Representative blot images of BACE1 protein levels in FACS-isolated microglia. (E) Quantitative analysis of the relative change in BACE1 (percentage of sham group). Data are expressed as the means ± SDs; n = 54. ^*P < 0.05, ^**P < 0.01 vs. the sham group.

BACE1 is involved in microglia-mediated neural injury

To explore the role of BACE1 upregulation in microglia, we used MK-8931, a BACE1-specific inhibitor, to block the function of BACE1 [10]. Given that microglia are essential to ICH recovery via the inflammatory response, brain edema and hematoma clearance [19, 20], we wondered whether BACE1 would affect the above microglial functions and neurological impairments. We first measured the mNSS score and corner turn rate in the mice with ICH that received MK-8931 or vehicle at 1 and 3 days after ICH induction. The inhibition of BACE1 significantly reduced neurological deficits (Figure 2A, 2B). Moreover, we found that the brain water content following ICH at 1 day was lower in the MK-8931-treated mice (Figure 2C). Next, we compared the differences in hematoma clearance based on a previous study. At 3 days post-ICH, MK-8931 dramatically reduced the residual hematoma volume and hemoglobin levels compared to the vehicle, and even as early as 1 day after ICH, the residual blood was significantly decreased in the MK-8931-treated mice, although the residual hematoma volume was similar to that of the control group (Figure 2D–2G). In conclusion, BACE1 might promote neuroinflammation and inhibit phagocytosis by microglia, thereby affecting recovery from ICH injury.

Figure 2. MK-8931 attenuated ICH induced neurological deficits via BACE1 inhibition. (A, B) ICH was induced in mice by injection of autologous blood. Summarized results showing neurological assessment (mNSS score, corner-turning test) of the MK-8931- or vehicle-treated groups of mice. n = 8 per group. (C) Brain water content in groups of mice receiving the indicated treatments at Day 1 after ICH. n = 11 per group. (D) Representative coronal sections showing hematomas from the indicated treated mice. (E–G) Quantification of residual hemoglobin, hematoma volume, and hematoma index (volume × density). n = 8 per group. Data are presented as the mean ± SD. ^*P < 0.05, ^**P < 0.01.

BACE1 elevated neuroinflammation and impaired erythrophagocytosis of microglia

To explore the exact relationship between BACE1 and inflammation and phagocytosis, we isolated microglia in the striatum from ICH 3-day mouse brains and measured the levels of classical pro/anti-inflammatory cytokines. In the MK-8931-treated mice, the expression of the IL-6, IL-1β and iNOS genes was significantly decreased in sorted microglia beginning at 3 days post-ICH, whereas the anti-inflammatory factors Arg1 and IL-10 and the pro-hemoglobin/erythrocyte phagocytic molecules CD163 and CD36 were significantly upregulated (Figure 3A, 3B). We further used fluorescence-labeled RBCs to construct an autologous blood-brain hemorrhage model to confirm the phagocytic capacity of microglia. We found a higher percentage of microglia in the MK-8931-treated mice that underwent erythrophagocytosis than in the vehicle-treated mice (Figure 3C, 3D). Thus, inhibition of BACE1 promoted a shift from a proinflammatory phenotype to an anti-inflammatory, pro-clearing phenotype in microglia after cerebral hemorrhage.

Figure 3. BACE1 correlated with microglial neuroinflammation and erythrophagocytosis. (A, B) Quantification of pro- or anti-inflammatory cytokine gene expression is shown in the bar graphs, with data from at least 5 mice at 3 days post-ICH. Data are shown as the relative change compared to the sham group, ^*P < 0.05 and ^**P < 0.01. (C) Flow cytometry histogram of fluorescently labeled erythrocytes in microglia/macrophages. (D) Quantification of the erythrophagocytosis population of microglia in the indicated treated mice at Day 3 after ICH. n = 5 per group. ^**P < 0.01.

BACE1 was detrimental to the transformation of the microglial protective phenotype in vitro

To further determine the role of BACE1 in microglial inflammatory and phagocytic phenotypes, we used hemoglobin (Hb)-treated primary microglia (PMG) as an in vitro model of ICH [21]. The BACE1 protein level was increased at 6 and 12 h after Hb stimulation (Supplementary Figure 1). Then, we analyzed cytokine gene expression in the MK-8931- or vehicle-treated PMG. Consistent with the in vivo results, proinflammatory cytokines (IL-6, IL-1β, iNOS) were significantly reduced by MK-8931, while anti-inflammatory/pro-phagocytosis molecules were dramatically increased (Figure 4A), and the results also validated by BACE1-targeted siRNA (Supplementary Figure 2A, 2B). The secreted IL-10 levels also increased significantly after MK-8931 treatment, while the secreted IL-1β levels decreased significantly in the PMG (Figure 4B). Thus, inhibition of BACE1 contributes to the conversion of microglia to an anti-inflammatory phenotype. We further measured the phagocytosis rate of the Hb-stimulated PMG with or without BACE1 inhibition by immunofluorescence and flow cytometry. We found that RBCs accumulated around the PMG, suggesting insufficient phagocytosis. In contrast, the number of intracellular fluorescence-labeled RBCs was significantly increased after the administration of MK-8931, and the aggregation of RBCs around the PMG disappeared (Figure 4C, 4D). Quantitative analysis of the proportion of PMG that underwent phagocytosis and their fluorescence intensity (indicating that one cell may involve multiple RBCs) revealed that that pharmacological inhibition of BACE1 by MK-8901 significantly enhanced the phagocytosis of PMG (Figure 4E, 4F) as well as the genetic inhibition by siRNA (Supplementary Figure 2C, 2D). Together, our results indicated that BACE1 inhibition contributes to a greater tendency of microglia to exhibit anti-inflammatory and pro-phagocytic phenotypes.

Figure 4. BACE1 impaired protective phenotype transition of microglia in vitro. (A) Quantification of cytokine gene expression is shown in the bar graphs, with data from at least 5 independent experiments. Data are shown as the relative change of the Ctrl group, ^*P < 0.05 and ^**P < 0.01. (B) Secreted IL-10 and IL-1β protein levels in the vehicle- and MK-8931-treated PMG measured by ELISAs. (C) Representative images of in vitro erythrophagocytosis of PMG, scale bar = 20 μm. (D) The percentage of erythrophagocytosis was calculated in the indicated treated PMG. (E, F) In vitro analysis of PMG erythrophagocytosis under Hb stimulation with/without MK-8901 treatment. The phagocytosis index was calculated as the median fluorescence intensity of the RBC-engulfed population, n = 3 per group. All data are presented as the mean ± SD. ^*P < 0.05, ^**P < 0.01.

BACE1 mediates microglial phenotype by activating STAT3 signaling

To explore the mechanisms by which BACE-1 inhibits the beneficial phenotype of microglia, we noted that STAT3 was closely associated with the microglial polarization shift and microglia-dependent regulation of neuroinflammation and phagocytosis in stroke [13, 22]. In addition, STAT3 was reported to be regulated by BACE1. We assessed the expression levels of STAT3 and phosphorylated STAT3 in the in vitro ICH model after MK-8931 and vehicle treatment and found that Hb stimulation significantly promoted STAT3 phosphorylation, whereas MK-8931 treatment significantly reduced this alteration (Figure 5A). We therefore treated PMG with recombinant mouse BACE1 protein (rmBACE1) and/or the STAT3-specific inhibitor niclosamide. The results showed that BACE1 significantly promotes STAT3 phosphorylation, while this effect is blocked by niclosamide (Figure 5B). We further determined the role of STAT3 in PMG function. We indeed observed that BACE1 promoted the expression of a high level of proinflammatory genes in the PMG and that treatment with niclosamide effectively reduced the level of proinflammatory factor genes (IL-6, IL-1β, iNOS). In addition, niclosamide significantly upregulated the expression of anti-inflammatory (Arg-1, IL-10) and pro-phagocytic genes (CD36) (Figure 5C, 5D). Flow cytometry indicated that rmBACE1 strongly reduced the phagocytosis ratio of RBCs by PMG, but niclosamide treatment effectively reversed the inhibitory effect of rmBACE1 (Figure 5E, 5F). Finally, we overexpressed microglia STAT3 levels under MK-8931 treatment and the results showed that the regulation of both microglia inflammation and phagocytosis by MK-8931 was significantly blocked (Supplementary Figure 3A–3D). Thus, BACE1-mediated STAT3 activation modulates neuroinflammation and erythrophagocytosis of microglia.

Figure 5. BACE1 induced microglia pro-inflammatory phenotype by STAT3 signaling activation. (A) The levels of phosphorylated and total STAT3 relative to the internal control actin are shown in the indicated treated PMG. n = 5 per group. (B) Primary microglia were treated with rmBACE1 protein with/without niclosamide treatment. Phosphorylated and total STAT3 were determined 12 h after treatment. n = 5 per group. (C, D) Quantification of pro- or anti-inflammatory cytokine gene expression is shown in the bar graphs, with data from at least 5 indicated treated PMGs. Data are shown as the relative change compared to the vehicle group. (E, F) In vitro analysis of the PMG erythrophagocytosis rate under rmBACE1 stimulation with/without niclosamide treatment, n = 3 per group. All data are presented as the mean ± SD. ^*P < 0.05, ^**P < 0.01.

STAT3 inhibition ameliorated neurological impairment after ICH

To investigate the effect of STAT3 inhibition on neurological function in ICH, we employed the corner turn test and mNSS score (Figure 6A). Administration of niclosamide resulted in improved neurological function and reduced brain water content at 1 day post ICH (Figure 6B, 6C). Similarly, the mNSS score also demonstrated that niclosamide significantly attenuated neurological impairments at both 1 and 3 days after ICH. We hypothesized that STAT3 inhibition would reduce BACE1-induced neuroinflammation by potentially impeding signal transduction activities. Specifically, microglia sorted from the mice with ICH 3 days after clonidine treatment showed a significant decrease in proinflammatory factor gene levels and a further increase in the expression of anti-inflammatory factor and pro-erythrophagocytosis molecule genes (Figure 6D, 6E). Finally, we used an in vivo RBC phagocytosis analysis system to determine the role of niclosamide in hematoma clearance. The results revealed that microglia of the mice with ICH 3 days under niclosamide treatment displayed a further elevated phagocytic ratio of involved erythrocytes (Figure 6F, 6G). In summary, inhibition of microglial STAT3 blocked the effect of BACE1 and had a positive effect on neuroprotection in ICH.

Figure 6. STAT3 inhibition facilitated neurological recovery after ICH. (A, B) Summarized results showing neurological assessment (mNSS score, corner-turning test) of MK-8931- or niclosamide-treated groups of mice at 1 and 3 days post-ICH. n = 8 per group. (C) Brain water content in groups of mice receiving the indicated treatments at Day 1 after ICH. n = 11 per group. (D, E) Analysis of the gene expression of inflammatory cytokines in sorted ICH 3-day microglia under the indicated treatment, n = 5 per group. Data are shown as the relative change compared to the sham group. (F) Flow cytometry histogram of in vivo RBC engulfment in microglia/macrophages. (G) Quantification of microglial phagocytosis in the indicated treated mice at Day 3 after ICH. n = 5 per group. All data are presented as the mean ± SD. ^*P < 0.05, ^**P < 0.01.

ICH mouse model

All animal experiments were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Animal Experimentation Ethics Committee of Shandong Provincial Hospital Affiliated with Shandong First Medical University. Male C57BL/6 mice (weight 25–28 g) were obtained from Shanghai Model Organisms (Shanghai, China). All animals were housed in pathogen-free conditions and had free access to food and water.

The ICH model was established in mice as described previously [21]. Mice were anesthetized with 3% isoflurane for induction and 1.5% isoflurane for maintenance. Tail was sterilized with 70% ethanol and then incised with a sterilized surgical blade. Next, autologous tail arterial blood was collected in a sterilized film. A total of 25 μL of whole autologous blood or the same volume of saline was collected by 50 μl Hamilton syringe injected into the striatum (0.5 mm anterior and 2.5 mm lateral to bregma, at a depth of 3.5 mm) at a rate of 2.5 μL/min after a midline scalp incision. The needle was withdrawn after a 5-minute pause. During surgery and anesthetic resuscitation, the animal’s body temperature was maintained 37 ± 0.5°C with an electric blanket.

Administration of chemicals and regents

MK-8931 was purchased from MedChemExpress, USA (HY-16759). For in vivo experiments, MK-8931 was orally administered at a dose of 10 mg/kg per day until the mice were sacrificed [17]. For in vitro experiments, primary microglia treated with 100 nM MK-8931 [10] or vehicle were maintained in DMEM under 30 μM hemoglobin (Hb) exposure for 12 h [35].

Niclosamide was purchased from MedChemExpress (HY-B0497). For in vivo experiments, Niclosamide was intraperitoneally administered at a dose of 20 mg/kg per day until the mice were sacrificed [36]. For in vitro experiments, primary microglia treated with 5 μM Niclosamide or vehicle were maintained in DMEM under 5 ug/ml recombinant mouse BACE1 protein (R&D Systems, USA, 2976-AS) exposure for 12 h [37, 38].

Behavioral assessment

Neurological dysfunction was assessed by a double-blind procedure on Days 1 and 3 after establishment of the ICH autologous blood model. The modified neurological severity score (mNSS) corner test was performed as previously described [21, 39]. Briefly, the mNSS score evaluates neurological function from a combination of motor, sensory, reflex and balance tests, with a score out of 15, with higher scores indicating more severe neurological deficits. In the corner test, mice are placed into a corner at an angle of 30° and required to turn to the left or right to exit the corner. Unilateral neurological impairment was determined by rating the percentage of right turns.

Microglial isolation and fluorescence-activated cell sorting (FACS)

Single-cell suspensions of brain tissue were prepared as described previously [40, 41]. Briefly, mice were transcardially perfused with ice-cold medium A (HBSS, 15 mM HEPES, 0.05% glucose, and 1:500 DNase I) to remove blood cells after euthanized by isoflurane overdose. The tissue was mechanically ground with a Dounce homogenizer and then centrifuged at 340 g for 5 min. The supernatant was discarded, and the cells were resuspended in 30% Percoll (GE Healthcare, USA). The myelin sheath was then removed by centrifugation at 900 × g for 20 min. The sediment was collected and resuspended with CD45-AF700 (1:200; BioLegend, USA) Ly6G-BV421 (1:200; Biolegend), CD11b-PE/Cy7 (1:200; BioLegend) and Zombie NIR (1 μl/test; BioLegend) antibodies for 20 min at 4°C in the dark. All samples were analyzed with a CytoFLEXLX analyzer (Beckman, USA), or microglia was obtained by sorting with a BD Aria SORP system (BD, USA). All results were analyzed with FlowJo 10.8.1.

Quantitative real-time PCR

Total RNA isolation (RNeasy Kit, Qiagen, USA) and cDNA reverse transcription were performed according to the manufacturer’s instructions (Vazyme, China). The primers were designed and provided by Tsingke Biotechnology Co., Ltd, China. Quantitative real-time PCR was performed according to the manufacturer’s instructions (Vazyme, China). The reaction conditions were as follows: first denaturation (95°C, 10 min), 40 denaturation cycles (95°C, 15 s), and extension (60°C, 40 s). Beta-actin was used as an internal control, and the relative levels of mRNA expression were calculated using the 2^−ΔΔCT formula. All data were shown as relative change to Sham or Vehicle group. The primers used for quantitative PCR are listed in Table 1 in the Appendix.

RBC phagocytosis in vivo

Mice whole blood was collected by cardiac puncture after anesthetized. Whole blood was washed twice in PBS and then RBCs were isolated from the pellet after Ficoll gradient centrifugation in SepMate tubes (STEMCELL Technologies). Isolated RBCs were treated with a fluorescently labeled probe (PKH-26 Red Fluorescent Cell Linker Kit, Sigma-Aldrich, USA) or CellTracker CFSE dye (Invitrogen, USA) to generate fluorescent RBCs, and fluorescently labeled RBCs were mixed with plasma from the same donor mouse source. Then, 25 μl of reconstituted blood was injected into the basal ganglia of drug-pretreated mice in the same manner as the autologous injection ICH mouse model. After cold PBS perfusion, we carefully separated the tissue from the border of the hematoma with microsurgical forceps at a distance of 1 mm under the microscope. The phagocytosis of erythrocytes by microglia (identified as Zombie-Ly6G-CD45intCD11b+PKH26+) in the area surrounding the hematoma was analyzed by flow cytometry at 3 days post-ICH [42].

RBC phagocytosis in vitro

The cell culture system was established as described previously [42]. Fluorescently labeled erythrocytes were added to primary microglia and incubated for 2 h at 37°C. Erythrocyte phagocytosis was observed using a STELLARIS 5 confocal microscope (Leica). The phagocytosis rate was calculated as the ratio of the number of microglia that phagocytosed erythrocytes to the total number of cells and is shown as a relative change from the control group.

ELISA

Cell culture media were collected, and the secreted interleukin-10 (IL-10) and interleukin-1β (IL-1β) protein concentrations were measured by ELISAs. The culture medium was centrifuged at 4°C for 15 min at 12,000 g to collect the supernatant, which was then assayed and quantified with IL-10 and IL-1β ELISA kits (R&D Systems, USA) according to the manufacturer’s instructions.

Immunofluorescence staining

Mice were euthanized and perfused with 4% paraformaldehyde (PFA), and whole brains were carefully removed and fixed by immersion in PFA, followed by dehydration with 30% sucrose. The brains were OCT-embedded to obtain 8-μm-thick frozen sections. For the staining procedure, brain slices were incubated in a blocking solution (10% donkey serum, 0.3% Triton 100 in PBS solution) at room temperature for 1 h. After PBS washes, the sections were incubated with primary antibodies overnight at 4°C. The following primary antibodies were used: Iba-1 (ab283346; Abcam) and BACE1 (ab183612; Abcam). After PBS washes, sections were incubated with fluorescent secondary antibody antibodies: donkey anti-rat Alexa Fluor 488 (A48269TR; Invitrogen) and donkey anti-rabbit Alexa Fluor 594 (A32754; Invitrogen). Finally, the brain slices were washed 3 times with PBS and then incubated with DAPI Fluoroshield mounting medium in the dark, followed by photography using confocal microscopy. All the images were taken at the peri-hematoma area.

Western blot

The standard protocol was performed as previously described [21]. Briefly, cells were lysed on ice for 20 min with RIPA buffer containing protease inhibitor and phosphatase inhibitor. After buffer was loaded, all samples were subjected to SDS-PAGE and blotted onto PVDF membranes. After blocking with 5% fetal bovine serum (BSA) in TBST, membranes were incubated with primary antibodies overnight at 4°C. The primary antibodies were rabbit-anti-BACE1 antibody (ab183612, Abcam), rabbit-anti-STAT3 antibody (A19566, Abclonal), and rabbit-anti-phospho-STAT3-S727 antibody (ab267373, Abcam). After three washes with TBST, the membranes were incubated with HRP-conjugated anti-rabbit secondary antibodies in 5% BSA for 2 hours at room temperature. The signals on the membranes were developed with ECL HRP substrate, and images were acquired with a molecular imager (ChemiDoc MP, Bio-Rad) and analyzed with ImageJ.

Statistics

All data are shown as the mean ± SD. Statistical differences between two groups were tested by two-tailed Student’s t test or Wilcoxon’s signed-rank test, and one-way ANOVA was used to compare multiple groups. Two-way ANOVA accompanied by a Bonferroni post hoc test was used for multiple comparisons. Significance was defined as P < 0.05. All results were analyzed by investigators blinded to the experimental group using GraphPad Prism8 software.