Aerobic Glycolysis Induced by mTOR/HIF-1α Promotes Early Brain Injury after Subarachnoid Hemorrhage via Activating M1 Microglia

M1 microglial activation is crucial for the pathogenesis of early brain injury (EBI) following subarachnoid hemorrhage (SAH), and there is growing evidence that glucose metabolism is frequently involved in microglial activation. However, the molecular mechanism of glycolysis and its role in M1 microglial activation in context of EBI are not yet fully understood. In this study, �rstly, the relationship between aerobic glycolysis and M1 microglial activation as well as SAH-induced EBI was researched in vivo. Then, intervention on mammalian target of rapamycin (mTOR) was performed to investigate the effects on glycolysis-dependent M1 microglial activation and EBI, and its relationship with hypoxia-inducible factor-1α (HIF-1α) in vivo. Next, Hif-1α was inhibited to analyze its role in aerobic glycolysis, M1 microglial activation and EBI in vivo. Lastly, both in vivo and in vitro, mTOR inhibition and Hif-1α enhancement were administered simultaneously, and the combined effects were further con�rmed again. The results showed that aerobic glycolysis and M1 microglial polarization were increased after SAH, and glycolytic inhibition could attenuate M1 microglial activation and EBI. Inhibition of mTOR reduced glycolysis-dependent M1 microglial polarization and EBI severity by down-regulating HIF-1α expression, while enhancement had the opposite effects. Blockading HIF-1α had the similar effects as suppressing mTOR, while HIF-1α agonist worked against mTOR antagonist when administered simultaneously. In conclusion, the present study showed new evidence that aerobic glycolysis induced by mTOR/HIF-1α might promote EBI after SAH by activating M1 microglia. This �nding provided new insights for the treatment of EBI.

There is growing evidence that glucose metabolism is actively involved in microglial activation [10].
Activated M1 microglial cells always display a metabolic shift from oxidative phosphorylation towards aerobic glycolysis [11], and inhibiting aerobic glycolysis can attenuate M1 microglial activation in animal models of Parkinson's disease [12]. In addition, several studies showed that mammalian target of rapamycin (mTOR) may promote aerobic glycolysis by activating downstream hypoxia-inducible factor-1α (HIF-1α), which is bound with glycolytic enzymes including hexokinase 2 (HK2) and pyruvate kinaseisozymes M2 (PKM2) [13,14]. Up to date, acute microglial in ammation caused by glycolysis dependent on mTOR/HIF-1α pathway has been reported in Alzheimer's disease [15]. However, whether or not this mechanism is involved in the pathogenesis of SAH-induced EBI remains ambiguous. Thus, this study was conducted to explore the role of mTOR/HIF-1α-modulating aerobic glycolysis in the development of EBI by focusing on its relationship with M1 microglial polarization.

Experimental design
Four separate experiments were carried out for this investigation (Fig. 1). It has been reported that microglial activation reaches a peak at 24 h after SAH [16]. Therefore, all in vivo and in vitro testings were performed at 24 h after modeling.

Experiment 3
To demonstrate the role of Hif-1α on aerobic glycolysis, M1 microglial polarization and EBI, the inhibitor 2-methoxyestradiol (2-ME) was introduced. Rats were divided into Sham, SAH and SAH + 2-ME groups (n = 18 per group). Brain edema (n = 6) and BBB permeability (n = 6) were assessed. The expressions of HK2, PKM2 and iNOS were con rmed by western blot and RT-PCR, and the lactate concentration was also determined using a lactate assay kit (n = 6).

Experiment 4
To further study the effect of mTOR/Hif-1α pathway on glycolysis-dependent M1 microglial activation and EBI, mTOR antagonist rapamycin and Hif-1α agonist dimethyloxalylglycine (DMOG) were used together in vivo and in vitro. In vivo study, rats were randomly assigned to Sham, SAH, SAH + rapamycin and SAH + rapamycin + DMOG groups (n = 18 per group). Brain edema (n = 6) and BBB permeability (n = 6) were estimated. Western blot was conducted to observe the expressions of HK2, PKM2 and iNOS, and a lactate assay kit was used to test the lactate (n = 6). In vitro SAH models, the BV2 cells were divided into control, control + rapamycin, control + DMOG, Hb, Hb + rapamycin, Hb + DMOG and Hb + rapamycin + DMOG groups. Cell viability was tested. The expressions of HK2, PKM2 and iNOS were examined by western blot and RT-PCR. Extracellular acidi cation rate (ECAR) and glucose uptake assay were performed to observe the glycolytic process during the study. ECAR: Extracellular acidi cation rate Animal SAH model Adult male Sprague-Dawley rats weighing 250g to 300g were purchased from the Animal Centre of Shanxi Medical University. All experimental procedures were conducted according to the ARRIVE guidelines. The protocols for the animal experiments were performed according to the animal and ethics review committee of our institution. The endovascular perforation method was used on rats to establish SAH models as previously described ( Fig. 2A) [17]. Brie y, a blunt 4 − 0 nylon suture was placed into the external carotid artery (ECA) after intraperitoneal anesthesia. The lament was advanced from the internal carotid artery (ICA) to the right anterior cerebral artery (ACA) [18]. When resistance was encountered, the lament was further advanced 5 mm and then withdrawn immediately to perforate the ACA. For the Sham-operated animals, all procedures were the same except for the puncture of the ACA (Fig. 2B).

SAH grading
The severity of SAH was evaluated by two blinded researchers according to a previous report [19]. As shown in Fig. 2C, after modeling at 24 h the basal cistern was divided into six segments. Each segment was scored from 0 to 3 based on the amount of blood clotting. A total score of the six segments ranging from 0 to 18 was obtained. To evaluate the homogeneity of non-mild SAH, rats scored below 7 were excluded and replaced with new ones.

Cell viability
Cell viability was estimated by a CCK-8 kit (Dojindo, Japan) [25]. In brief, BV-2 cells were plated in 96-well plates and conducted SAH stimulation as described above. Subsequently, 10 µL CCK-8 solution was added to each well and incubated for 2 h. Next, the OD value was read on a microplate reader (ThermoFisher, USA) at 450 nm. The results are reported as a percentage of viable cells, with 100% viability being considered for the control group.

Quantitative real-time PCR
Quantitative real-time PCR was conducted as previously described [27]. Total RNA was isolated from the right inferior basal temporal lobe and cultivated BV2 cells with the TRIzol reagent RNAiso Plus (Takara, Japan). The total RNA was then reverse transcribed to cDNA by the PrimeScript™ RT Master Mix (Takara, Japan). All protocols were performed according to the manufacturer's instructions. Quantitative RT-PCR was performed in the CFX96 Real-Time PCR Detection System with TB Green® Premix Ex Taq™ II (Takara, Japan) to quanti cation. The mRNA expression of β-actin was used as an internal control. The ndings were presented as fold changes as compared to the Sham group. The primer sequences used were listed in Table 1.

Neuroscore
Before death, a modi ed Garcia scoring system was used to evaluate the neurological de cits of animals [30]. The neurological de cits include spontaneous activity, symmetry of limbs, forepaw outstretching, climbing, body proprioception, and reaction to vibrissae, all of which were tested and scored from 0 to 3 for each segment. A higher score indicated better neurological function.

Brain edema
The dry-wet weight method was performed to examine brain water content after SAH. The right hemisphere was separated at 24 h after the operation and weighed immediately (wet weight). Then it was dried for 24 h at 105°C in an oven (dry weight). The severity of brain water was calculated as (wet weight − dry weight)/wet weight×100% [17,31].

BBB permeability
The assessment of BBB permeability was consistent with previous reports [10,17], Brie y, Evan's blue dye (50 mg/kg) was intraperitoneally injected at 24 h post SAH. After the dye was circulated for 3 h, the perforation sided inferior basal temporal lobe was removed and homogenized in PBS. Then centrifugated for 30 min at 15000 g. After that, 0.7 ml of the supernatant was added to an equal amount of trichloroacetic acid. Lastly, the amount of transudatory Evan's blue dye was detected by spectrofourophotometry after overnight incubation and centrifugation at 15000 g at 4°C for 30 min.
Measurements were conducted at an excitation wavelength of 615 nm.

Lactate concentration measurement
Lactate concentrations of the right basal inferior temporal lobe were determined by a luminometric Lactate-Glo assay kit (Promega) in accordance with the manufacturer's instructions [32]. The luminometry was measured with a microplate reader. Results were quanti ed using a standard curve.
ECAR assay BV cells were seeded in a 96-well plate with a density of 50,000 cells/well and treated as instructed. Cells were then administrated with ECAR reagents according to the manufacturer's recommendations (Abcam, American) [33]. The micro-plate reader system (Molecular Devices, American) was used to collect ECAR signals at 1.5 min intervals for about 120 min using excitation wavelengths of 380 nm and emission wavelengths of 615 nm respectively.
Glucose uptake assessment BV2 microglia cells in 96-well plates were tested using the bioluminescent glucose uptake assay. Each plate contained 2000 BV2 cells that were cultured with 1 mM 2-deoxyglucose (2-DG) for 10 min before being processed according to the manufacturer's instructions (Glucose Uptake-Glo Assay, Promega) [34].

Statistics
All data was expressed as mean ± standard deviation (SD). Multiple comparisons were performed with one-way ANOVA followed by Tukey's post hoc test. Differences in mortality between groups were analyzed with χ 2 tests or Fisher's exact test. SPSS (version 22.0) was used in statistical analysis, and P < 0.05 was considered statistically signi cant.

Animal mortality
None of the rats died in the Sham group. In the rst experiment, the mortality was 27.3% (9 of 33) in the SAH group and 25.0% (8 of 32) in the SAH + 2-DG group. In the second experiment, the mortality was 25.0% (8 of 32) in the SAH group, 14.3% (7 of 31) in the SAH + rapamycin group, and 29.4% (10 of 34) in the SAH + 3-MA group. In the third experiment, the mortality was 25.0% (6 of 24) in the SAH group and 21.7% (5 of 23) in the SAH + 2-ME group. In the fourth experiment, the mortality was 25.0% (6 of 24) in the SAH group, 18.1% (4 of 22) in the SAH + rapamycin group, and 21.7% (5 of 23) in the SAH + rapamycin + DMOG group. There was no statistical signi cance in mortality among the modeling groups in each set of the experiment (Fig. 3).

Severity of SAH models
The SAH grading method (Fig. 2C) was performed to estimate the severity of SAH. There was no difference among any of the SAH-operated groups for each set of the experiment (Fig. 4).

Aerobic glycolytic inhibition ameliorated M1 microglial activation and EBI
In the rst experiment, the glycolytic key enzymes (HK2, PKM2), the metabolite lactate and M1 microglial markers (iNOS, TNF-α, IL-1β) were found to be signi cantly higher at 24 h after SAH. Nonetheless, the glycolytic inhibitor 2-DG alleviated this trend (Fig. 5A-E). Iba-1 staining further con rmed the M1 microglial polarization under SAH conditions. Immuno uorescence results showed that the number of Iba-1-positive M1 microglia increased at 24 h after SAH but decreased obviously after treatment with 2-DG (Fig. 5F). In addition, the severity of EBI, evidenced by brain water content, BBB destruction and neurological de cits, was aggravated remarkably at 24 h post SAH, while intervention of 2-DG appeared to mitigate this trend (Fig. 5G). All these results suggested that aerobic glycolysis and M1 microglial polarization were increased after SAH, and the initiation of glycolysis could participate in M1 microglial polarization and subsequent EBI formation. mTOR regulated glycolysis-dependent M1 microglial polarization and EBI In the second experiment, levels of key glycolytic enzymes (HK2, PKM2), M1 microglial marker (iNOS) and lactate were signi cantly increased at 24 h after SAH, whereas all of which were further decreased by down-regulating mTOR with rapamycin and increased more by up-regulating mTOR with 3MA ( Fig. 6A-C). HK2 and PKM2 were co-located with Iba-1 respectively to detect the number of activated M1 microglia dependent on glycolysis. Immuno uorescence results exhibited that HK2-Iba-1 and PKM2-Iba-1 doublepositive cells were enhanced under SAH conditions. And they were declined in the SAH + rapamycin group and further rised in the SAH + 3-MA group as compared to the SAH group ( Fig. 6D and E). In addition, mTOR inhibition further attenuated the severity of EBI, as measured by brain water content, BBB destruction and neuroscores, while enhancement had the contrary effect (Fig. 6F). Taken together, mTOR played a critical role in regulating the activation of glycolysis-dependent M1 microglia and the severity of EBI.
Hif-1α acted as a downstream factor of mTOR In the second experiment, the levels of Hif-1α in SAH-operated groups were remarkably enhanced at 24 h after surgery. Compared with the SAH group, the down-regulation of mTOR signi cantly decreased the expression of Hif-1α, whereas up-regulation had the opposite effects (Fig. 7A). In the third experiment, at 24 h after SAH, blocking Hif-1α with 2-ME dramatically suppressed the levels of HK2, PKM2, iNOS, and lactate, as well as reduced EBI severity which was shown by brain edema, BBB disruption, and neurological de cits (Fig. 7B-E). All these results validated that Hif-1α might act as a downstream factor of mTOR to regulate aerobic glycolysis, M1 microglial activation and EBI severity.
Combined effects of mTOR/Hif-1α pathway intervention on glycolysis-dependent M1 microglial polarization and EBI In the fourth experiment, inhibiting mTOR with rapamycin declined the levels of HK2, PKM2, iNOS and lactate, as well as attenuated the severity of EBI, as displayed by brain edema, BBB disruption and neurological impairments, whereas this trend was ameliorated by activating Hif-1α by DMOG at 24 h after SAH ( Fig. 8A-C). The cultured BV2 cells were co-incubated with Hb to simulate the SAH insult in vitro. The cell viability assay showed that neither rapamycin nor DMOG could cause signi cant damage to cells cultured in vitro (Fig. 8D). Consistent with the in vivo ndings, data from western blot and RT-PCR showed that Hb intervention dramatically raised the levels of HK2, PKM2 and iNOS, all of which were reduced by rapamycin, but further administration of DMOG seemed to mitigate this trend ( Fig. 8E and F). The ECAR assay showed that glycolysis was enhanced at 24 h after Hb treatment, displaying an increased ECAR value. After inhibiting the expression of mTOR, increased glycolysis was largely diminished, but downregulation of glycolysis was remarkably interfered by activating Hif-1α (Fig. 8G). In addition, luminescence analysis using glucose analog 2-DG revealed rapamycin signi cantly inhibited the glucose uptake, whereas the introduction of DMOG ameliorated this trend (Fig. 8H). The evidence above revealed that the mTOR/Hif-1α pathway promoted EBI by activating glycolysis-dependent M1 microglia.

Discussion
In the present study, we explored the mechanism of M1 microglial activation during EBI after SAH, and we made the following signi cant ndings: (1) Aerobic glycolysis could promote M1 microglial polarization and EBI after SAH. (2) mTOR had a vital impact on the severity of EBI by regulating glycolysis-dependent M1 microglial activation. (3) Hif-1α, acting as a downstream factor of mTOR, participated in the activation of M1 microglia and consequent EBI formation. To the best of our knowledge, this is the rst study to elucidate the role of mTOR/Hif-1α-mediated aerobic glycolysis in development of EBI by activating M1 microglia (Fig. 9).
Increasing evidence indicates that the effect of metabolic reprogramming in the regulation of the in ammatory response. Researches on macrophages validate that the activation of M1-type is often accompanied by a shift in cells from oxidative phosphorylation to aerobic glycolysis for producing energy [35]. More recently, the link between aerobic glycolysis and polarization has been considered in microglia. Several studies have shown that glycolytic inhibition can attenuate M1 microglial activation in Parkinson's disease and perioperative neurocognitive disorders [12,36].However, the relationship between aerobic glycolysis and M1 microglial polarization remains unclear in SAH context. In the present study, our results validated that initiation of aerobic glycolysis could lead to M1 microglial polarization and further promote EBI.
Recently, the connection between mTOR and aerobic glycolysis has received more attention. mTOR is a ubiquitous serine/threonine kinase of the phosphatidylinositol 3-kinase-related kinase family and regulates many important physiological functions, such as cell metabolism [37]. A previous study revealed that mTOR played a crucial role in the regulation of glycolysis in microglia to shape their distinct functions under various states [38]. Another research demonstrated that inhibiting mTOR induced the microglial polarization to the M2 phenotype to exhibit neuroprotection in SAH-induced EBI [16]. In this study, we further investigated the relationship between mTOR and M1 microglia activation during EBI. Interestingly, we found that after SAH, mTOR inhibition could decrease M1 microglial activation to reduce the EBI severity, possibly through inhibiting aerobic glycolysis initiation. Collectively, previous and present studies demonstrated that mTOR displayed a critical role in development of EBI via regulating glycolysisdependent microglial polarization.
It is reported that HIF-1α is a crucial factor in mTOR pathway-mediated diseases, including central nervous system diseases such as Alzheimer's disease and glioblastoma [15,39]. HIF-1α is also reported to play an important role in glycolysis by interacting with glycolysis-related genes [40]. A study has shown that inhibiting mTOR can lead to a decrease of HIF-1α levels [41] and therefore decline the expression of HIF-1α-dependent glycolytic and in ammatory genes in tumors [42]. And another in vitro study exhibited that HIF-1α expression was involved in M1 microglial activation [43]. In this study, we discovered a positive link between the expression of mTOR and Hif-1α after SAH, and inhibiting Hif-1α with 2-ME suppressed the glycolysis-dependent M1 microglial activation and the severity of EBI. In addition, although mTOR antagonist rapamycin obviously reduced the glycolysis-dependent M1 microglial polarization and EBI severity, the administration of Hif-1α agonist DMOG seemed to attenuate this trend.
To sum up, all these results validated that HIF-1α may act as a downstream factor of mTOR to promote the glycolysis-dependent M1 microglial activation and EBI formation.
To date, there are few speci c and potent drugs targeting the pathological processes in SAH-induced EBI, and this provides a great challenge for investigators. In this study, it was shown that intervention on the mTOR/HIF-1α pathway and aerobic glycolysis with chemical agents could subsequently affect microglial M1 activation, further in uencing the severity of EBI. This nding identi ed the mTOR/HIF-1α pathway and aerobic glycolysis as potential therapeutic targets of EBI and had signi cant implications for improving the prognosis of SAH patients. In the future, more pharmacological experiments should be conducted to validate this point.
Several limitations in the current study should not be overlooked. Firstly, the most di cult point is to control the amount of bleeding and the severity of SAH in establishing animal models as previously reported [29,31]. Nevertheless, in the present study, no signi cant differences in SAH grading scores were observed among SAH groups, implying similar bleeding volume and brain damage. Secondly, it is well known that microglia adopt two opposing phenotypes in response to different environments, including M1 and M2. However, we just focused on the relationship between aerobic glycolysis and M1 microglial activation in the current study. Therefore, further investigation should be conducted to explore the connection between glucose metabolism and M2 microglia after SAH. Finally, we concentrated on the role of mTOR/HIF-1α in regulating microglia-mediated neuroin ammation following SAH-induced EBI.
However, it is unknown whether this pathway promotes neuroin ammatory damage via modulating other immune cells, such as T lymphocytes or astrocytes. Thus, more efforts are required to further explore its role in other types of immune cells after SAH in the future.

Conclusions
This study investigated the underlying molecular mechanism and glycolytic metabolism of M1 microglial polarization in the development of SAH-induced EBI. Our data indicated that the mTOR/HIF-1α pathway participates in aerobic glycolysis and further activates M1 microglia, contributing to the in ammatory response in EBI post SAH. More importantly, the present ndings support the notion that the mTOR/HIF-1α pathway and aerobic glycolysis might be novel therapeutic targets to alleviate the severity of EBI after SAH.

Declarations Ethical Approval and Consent to participate
This study was conducted in accordance with the ARRIVE guidelines and approved by the Animal and Ethics Review Committee of the Second Hospital A liated to Shanxi Medical University of China.

Human and Animal Ethics
All experimental procedures were conducted according to the ARRIVE guidelines. The protocols for the animal experiments were performed according to the animal and ethics review committee of our institution.

Consent for publication
Not applicable Availability of supporting data The authors con rm that the ndings of this study are supported by the data therein. The data are available from the corresponding author. Representative pictures of SAH-operated and Sham-operated brains, and SAH grading scores. The right basal inferior temporal lobe taken for assays are shown in the schematic representation in SAH-operated group (A) and Sham-operated group (B). The basal brain surface was divided into six regions for SAH grading (C).   IL-1β (E). Double-immuno uorescence analysis was performed with antibody for microglial M1 marker (Iba-1, red), and nuclei (DAPI, blue). Rectangular regions were magni ed. Magni cation, ×40, scale bar, 50 μm (F). EBI severity was evaluated by brain edema, BBB disruption and neuroscore (G). n = 6 for each group. Data are represented as mean ± SD. * P < 0.05, ** P < 0.01 vs Sham group. # P < 0.05, ## P < 0.01 vs SAH group. group. Data are represented as mean ± SD. * P < 0.05, ** P < 0.01 vs Sham group. # P < 0.05, ## P < 0.01 vs SAH group.

Figure 7
The effects of Hif-1α on microglial M1 polarization and EBI through acting as a downstream factor of mTOR. Representative pictures and quantitative analysis of Hif-1α levels in different groups (A).
Representative bands and relative densitometric analysis of HK2, PKM2, and iNOS 24 h after SAH (B). Quanti cation of the mRNA levels of HK2, PKM2, and iNOS (C). Quanti cation for lactate levels of every group (D). EBI severity was assessed by brain water content, BBB disruption and neurological impairments (E). n = 6 for each group. Data are represented as mean ± SD. * P < 0.05, ** P < 0.01 vs Sham group. # P < 0.05, ## P < 0.01 vs SAH group.