Ferroptosis of brain microvascular endothelial cells contributes to hypoxia‐induced blood–brain barrier injury

Hypoxia is pivotal to the pathogeneses of myriad disorders, especially hypoxic cerebropathy. Much is known about the damage to the blood–brain barrier (BBB) in response to hypoxia. Studies have shown that endothelial cell death is closely linked to functional impairment of BBB. Mounting evidences have demonstrated that ferroptosis, a new pathway regulating cell death, is implicated in brain injury. However, whether ferroptosis is involved in hypoxia‐induced BBB disruption remains ambiguous. Here, we utilized in vivo zebrafish and in vitro bEnd.3 cells to explore the correlation between endothelial ferroptosis and hypoxia‐induced BBB damage. We found that hypoxic treatment for 45 min can induce BBB disruption by triggering down‐regulation of claudin‐5 (CLDN5) both in zebrafish cerebrovascluar endothelial cells and bEnd.3 cells. Besides, in vitro and in vivo studies revealed the cysteine/glutamate antiporter xCT (also known as solute carrier family 7 member 11; SLC7A11) decrease, glutathione peroxidase 4 (GPX4) and glutathione (GSH) reduction, 4‐Hydroxynonenal (4‐HNE) increasement, malondialdehyde (MDA) upregulation and reactive oxygen species (ROS) accumulation in hypoxia group. Further mechanism studies indicated that hypoxia‐induced BBB damage might associate with microvascular endothelial cellular ferroptosis, since hypoxic exposure significantly activated the expression of ferroptosis‐related genes (Ptgs2, Por, Lpcat3, Alox5, Alox12, Nfe2l2, and Ncoa4) and inhibited the expression of Slc7a11. Additionally, the application of 20 μM ferrostatin‐1 (Fer‐1), a ferroptosis inhibitor, could partially alleviate BBB disruption under hypoxia, suggesting that inhibition of ferroptosis might be a potential strategy for some neurological diseases with BBB defect.


| INTRODUCTION
Hypoxia is an important feature in a plethora of diseases, especially hypoxic cerebropathy. Hypoxia will cause a variety of pathological disorders during brain metabolism since the brain is excessively sensitive to hypoxic stimulation. Previous studies have demonstrated that cerebral hypoxia leads to neuronal degeneration and necrosis, which is often accompanied by behavioral changes and cognitive dysfunction. 1 The blood-brain barrier (BBB) is a natural structural and functional barrier between the blood circulation system and the central nervous system diseases (CNS), which strictly controls the influx and efflux of molecules. An intact BBB is essential for maintaining CNS homeostasis. Brain microvascular endothelial cells (BMECs) and tight junctions (TJs) between adjacent BMECs are the basis for the BBB formation. 2 BMECs are the first line of defense for peripheral substances to enter the CNS, so they are considered to be the core of regulating BBB. TJs between adjacent BMECs mainly consist of transmembrane proteins such as claudins, occludin, and junctional adhesion molecule, which are vital for the low paracellular permeability of the BBB by restricting the entry and exit of substance. 3 The disruption of TJs leads to BBB breakdown, which subsequently becomes the central element in the pathology of many CNS diseases including ischemic stroke. 4 Ferroptosis is an iron-dependent antioxidant system imbalance, characterized by an increase of intracellular iron and a decrease in antioxidant capacity, leading to lipid peroxide (LPO) accumulation. 5 The occurrence of ferroptosis is mainly regulated by various metabolic pathways, including lipid metabolism, amino acid and glutathian metabolism, iron metabolism, and so on. Amino acid metabolism pathway can regulate the synthesis of glutathione (GSH) and promote glutathione peroxidase 4 (GPX4) to play antioxidant protective role, which can remove reactive oxygen species (ROS), ultimately inhibiting ferroptosis. 6 Reduced GSH can cause impaired GPX4-mediated ROS clearance, resulting in ROS accumulation and ferroptosis. 7 In addition, recent researches have pointed to hypoxia as a regulator of ferroptosis. 8 Studies have shown that hypoxia-inducible factors (HIF), a hypoxic indicator, plays an important role in the regulation of hepcidin and iron metabolism homeostasis, which in turn affects ferroptosis. 9 These discoveries indicated that hypoxia may be a regulative factor of ferroptosis.
However, the crosstalk between hypoxia and ferroptosis remains to be clarified. Does ferroptosis contribute to the increased permeability of BBB after hypoxia? BMVECs are the central factors regulating BBB permeability, whether ferroptosis has an effect on the structure and function of BMVECs? All these problems need to be further explored. Here, we utilized in vivo zebrafish and in vitro bEnd.3 cells to explore the correlation between BMECs ferroptosis and hypoxia-induced BBB damage. We found that hypoxia-induced ferroptosis in BMECs and inhibition of ferroptosis by ferrostatin-1 (Fer-1) partially ameliorated hypoxia-induced BBB injury.

| Zebrafish husbandry
Care and breeding of zebrafish was carried out basically according to standard protocols. 10 Fertilized embryos were kept in embryo medium at 28.5°C with 0.03%1-phenyl-2-thiourea (PTU) (Sigma-Alrich, St. Louis, USA), which was added into the egg water at 12 h post fertilized (hpf) to suppress pigment formation. All animal experimentation was in accordance with "The legislation of Guangdong laboratory animal management regulations."

| Hypoxic brain injury model in zebrafish
To construct hypoxic brain injury model in zebrafish, wildtype AB and Tg (kdrl:eGFP) line was used where the green fluorescent protein-labeled vascular endothelial cells were present. This model was established according to a modified version of previously described methods. 11,12 At first, pure nitrogen (N 2 ) was perfused into hypoxia chamber made by 1000 mL clear glass bottle with two ports until the dissolved oxygen (DO) reached 1.5-1.7 mg/L. The DO value in embryo medium was monitored by an oximeter. Next, zebrafish, 4 days post fertilized (dpf) were transferred to the hypoxia chamber and placed in an incubator at 28.5°C for 45 min. During that time, in order to provide a closed environment, the apparatus was hermetically sealed.

| In vivo zebrafish BBB dye diffusion assay
The dye diffusion assay was performed to reflect the permeability of BBB in vivo. In brief, contrast agent was applied in transgenic Tg (kdrl:eGFP) zebrafish larvae (4 dpf) as previously described. 13 Initially, Rhodamine-Dextran (MW:10 kDa; 5 nL, 0.5 mg/mL) (R8881, Sigma-Aldrich, St. Louis, USA) was injected via the common cardinal veins (CCV) of the larvae. Immediately, the larvae were mounted in 1% low-melting point agarose (LMA) (16520100, Invitrogen, Carlsbad, USA) and imaged with a FV3000 confocal laser scanning microscope (Olympus, Tokyo, Japan). The relative fluorescence intensity of the dyes was measured to evaluate the leakage of BBB.

| TEER measurement
To analyze the tightness of BBB in vitro, the transendothelial electrical resistance (TEER) were estimated as previously reported. 14 In general, bEnd.3 cells were planted on transwell cell culture plates and grown until a confluent monolayer formed. The TEER values were measured using EVOM2 epithelial voltohmmeter (WPI, Sarasota, FL, USA) at series time points.

| MDA
The malondialdehyde (MDA) was detected according to the standard protocol of the MDA assay kit (BC0025, Salarbio). In brief, after building the hypoxia model, the head of zebrafish was removed, placed on ice, and washed with precooled phosphate buffer solution (PBS) three times. Then, about 0.1 g tissue was weighed, following by adding 1 mL extract before being centrifuged at 8000 g for 10 min at 4°C. The supernatant was collected. Next, MDA test working solution, samples, and reagent 3 were added respectively. The mixture was bathed in water at 100°C for 60 min before being cooled in ice bath, and then centrifuged at 10 000 g for 10 min at room temperature. Subsequently, 200 μL of supernatant was absorbed into 96-well plate and the absorbance of each sample was measured at 450 nm, 532 nm, and 600 nm. Finally, the content of MDA was calculated by the formula and analyzed using the GraphPad Prism 7.0.

| ROS
ROS fluorescence intensity was measured using 2′, 7′-Dichlorodihydrofluorescein diacetate (DCFH-DA) according to the manufacturer's protocol (ROS Assay Kit, S0033S, Beyotime). Briefly, the bEnd.3 cells were seeded and cultured in a 24-well cell culture plate. After treating the cells under hypoxia condition, the cells were incubated with DCFH-DA (10 mM, diluted in serum-free medium) and Hoechst (94403, Sigma-Aldrich; 1:1000; diluted in 5% BSA) for 20 min at 37°C in the dark. Then, the cells were washed three times with serum-free medium. Immediately after washing, images were captured by a fluorescent microscope (Leica, wetzlar, Germany) at 488 nm excitation and 525 nm emission wavelengths. The fluorescence intensity was quantitatively analyzed by the ImageJ software.

| Iron assay
Fe 2+ and total iron levels in zebrafish and cells were determined using the iron assay kit (ab83366, Abcam) according to the manufacturer's instructions. Briefly, samples were washed with cold PBS, homogenized in iron assay buffer on ice, and then the supernatant was collected. Next, iron buffer and iron reducer were added respectively, mixed, and incubated for 30 min to test Fe 2+ and total iron levels. Finally, iron probe was incubated for 1 h. Subsequently, the mixture was immediately measured at 593 nm on a colorimetric microplate reader.

| Real-time quantitative PCR analysis
Total RNA of zebrafish and bEnd.3 cells were extracted respectively using RNAiso Plus (9109, Takara, Japan) according to the manufacturer's protocols. RNA was reverse-transcribed with Hifair® III 1st Strand cDNA Synthesis SuperMix for qPCR (11141ES60, YEASEN, Shanghai, China). Quantitative PCR was performed in triplicate using Hief® qPCR SYBR Green Master Mix with Low Rox (11202ES08, YEASEN) on a real-time PCR detection system (ABI QuantStudio 6 Flex, Thermo Fisher). Sequences of primers for RT-qPCR are listed in Tables S1 and S2.

| GSH and GSSG measurements
The determination of total glutathione (T-GSH), reduced glutathione (GSH), and oxidized disulfide (GSSG) was performed using 5,5-dithiobis-(2-nitrobenzoic acid) (DTNB) method with kits (S0053, Beyotime). Briefly, the heads of fish were flash frozen with liquid nitrogen after being washed with cold PBS three times. Next, 0.3 g tissue was weighed, following by adding 300 μL protein removal reagent M solution and then homogenized before being centrifuged at 10 000 g for 10 min at 4°C. The supernatant was collected to determine T-GSH. GSSG was measured by 5-thio-2-nitrobenzoic acid (TNB) which was produced from the reaction of reduced GSH with DTNB. Absorbance was monitored continuously at 405 nm with a colorimetric microplate reader for 25 min and readings were recorded every 5 min. Standards (0.5, 1.0, 2.0, 5.0, 10.0, and 15.0 μM) of T-GSH and GSSG were also assayed. The concentration of GSH in the sample was obtained by subtracting GSSG from T-GSH. Experiments were independently repeated three times, and values shown were the mean ± SD.

| Statistical analysis
The data were presented as means ± standard deviation (mean ± SD). All the experiments were independently performed at least three times. All statistical analysis was carried out using GraphPad Prism 7.0 software. Data of the two groups were compared with each other with twotailed unpaired Student's t-test. Data of more than two groups were analyzed by one-way or two-way ANOVA followed by Tukey's test. p < .05, p < .01 and p < .001 were indicated by * or # , ** or ## , *** or ### respectively.

| Hypoxia treatment leads to damage of BBB both in vivo and in vitro
Zebrafish showed cardiac edema and brain hemorrhage under hypoxia ( Figure 1A). Based on our previous study, we considered the damage mentioned above was closely related to BBB impairment. 15 Hence, BBB integrity was accessed by cerebral angiography assay after hypoxia induction. Leakage of Rhodamine-Dextran (MV: 10 kDa) into the brain through BBB was found after hypoxia treatment ( Figure 1B). CLDN5, a member of the Claudin family of TJ transmembrane proteins, is the main component of TJ between endothelial cells. 16,17 Changes in CLDN5 epitomize the changing paracellular permeability of BBB. 18 Therefore, we detected the protein level of CLDN5 in zebrafish head. The result showed that CLDN5 significantly decreased under hypoxia ( Figure 1C). TEER assay showed that the barrier function decreased in a time-dependent manner under hypoxia in bEnd.3 cells ( Figure 1D). Moreover, CLDN5 protein expression in bEnd.3 cells was also significantly down-regulated after hypoxic treatment for 12 h, which was consistent with the results in vivo ( Figure 1E,F). Therefore, hypoxic treatment leads to BBB disruption both in vivo and in vitro.

| Hypoxia-induced BBB damage is related to oxidative stress
Previous studies have suggested that hypoxia exposure promotes oxidative damage and thereby further triggers inflammation. 19 The levels of malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE), a marker of LPO, were measured in zebrafish head under hypoxia. Interestingly, a markedly increased MDA and 4-HNE was found in zebrafish head after hypoxia treatment (Figure 2A,B). Oxidative stress (OS) is harmful to cells due to the excessive generation of highly ROS. 20 Subsequently, we wondered whether hypoxia will induce OS in vitro by measuring ROS in bEnd.3 cells. Interestingly, we found a higher ROS level in the hypoxic group ( Figure 2C). Consistent with the in vivo study, a significantly increased level of 4-HNE in bEnd.3 cells under hypoxia was observed ( Figure 2D,E). Together, these data indicated that hypoxia-induced BBB injury was connected with OS.

| Hypoxia induces the ferroptosis of brain microvascular endothelial cells
The key functions of CNS are maintained by the interaction of multiple cell types within the neurovascular unit (NVU), including endothelial cells, pericytes, astrocytes, microglia, and neurons. 21 Among them, BMECs are crucial for the integrity and function of BBB. Compared with pericytes and astrocytes, BMECs were more sensitive to hypoxia. 22 The death of BMECs will lead to BBB impairment. 23 However, previous studies mostly focused on alleviating the process of ferroptosis on neurons, 24,25 whether ferroptosis occurs in BMECs under hypoxic conditions is still unknown. It has been revealed that ferroptosis in bEnd.3 cells induced BBB disruption after mechanical stretch injury. 26 Nevertheless, our study focused on whether hypoxia induces the ferroptosis of BMVECs, and the subsequent effect of endothelial ferroptosis on BBB function. Firstly, we detected the transcription of ferroptosis-related genes in zebrafish head by RT-qPCR. We found that hypoxic brain injury led to the upregulation of ptgs2a, ptgs2b (the gene overexpressed in ferroptotic cells and regarded as a marker of ferroptosis), lpcat3, alox5, alox12, pora (the enzymes that play a positive regulatory role in LPO), ncoa4 (the gene related to iron metabolic pathway and involved in ferritinophagy), epas1b (known as hypoxia inducible factor 2a, hif-2a), and nfe2l2a (also named nuclear factor E2-related factor 2, nrf2, has antioxidant effect), while the transcription of slc7a11 (the gene encodes xCT) was down-regulated ( Figure 3A). Ferrous iron accumulation is one of the main characteristics of ferroptosis. 27 To ask for more evidences supporting that ferroptosis is responsible for the BBB impairment under short-term hypoxia, iron content in the head of zebrafish was measured. A significant increase of total Fe, Fe 2+ , and Fe 3+ was observed in the hypoxic group, indicating ferroptosis does occur under exposure of hypoxia ( Figure 3B). Additionally, GPX4 was reported to scavenge LPO and inhibit ferroptosis. 28 Interestingly, our result revealed a remarkable decrease in GPX4 and xCT due to hypoxia exposure ( Figure 3C).
Consistent with the in vivo data observed above, we discovered that hypoxia exposure significantly down-regulated the transcription of Slc7a11 while up-regulated ferroptosisrelated genes involving Ptgs2, Lpcat3, Ncoa4, Nfe2l2, Alox5, Alox12, and Por ( Figure 3D). Similarly, the protein expression of GPX4 ( Figure 3E,F) and xCT ( Figure 3E) were decreased in bEnd.3 cells with 12 h incubation of hypoxia. Taken together, ferroptosis of BMECs contributed to the BBB disruption under short-term hypoxia.

| The ferroptosis inhibitor Fer-1 alleviated hypoxia-induced BBB disruption
Subsequently, we hypothesized that the application of Fer-1, a ferroptosis inhibitor, could alleviate BBB damage induced by hypoxia. A serious diffusion through BBB was observed post-hypoxia. However, pre-injection of Fer-1 before hypoxia exposure could partly rescue the BBB disruption after hypoxia induction ( Figure 4A). To sought the in vitro evidence confirming the role of ferroptosis in BBB permeability, we measured the TEER value at 3, 6, 12, 24 h after hypoxia exposure in the presence of Fer-1 or not. Interestingly, a significant up-regulation of TEER value was observed in Fer-1 pretreated group ( Figure 4B). ROS induced by hypoxia could be significantly reduced by Fer-1 ( Figure 4C). SLC7A11 and solute carrier family 3 member 2 (SLC3A2) constitute system Xc-together on the cell membrane, which exports glutamate and imports cystine at a 1:1 ratio. Cystine is rapidly transformed into cysteine intracellularly, the rate-limiting amino acid for the synthesis of GSH. GPX4 converts GSH into oxidized glutathione (GSSG) and reduces LPO and cell damage. 29 Given that reduced GSH levels can trigger ferroptosis, 30 we measured GSH and GSH/GSSG ratio in zebrafish head after hypoxia. The GSH and GSH/GSSG ratio are significantly decreased in hypoxic group, which is rescued by pretreating the zebrafish with Fer-1 ( Figure 4D).
Moreover, the content of total Fe, Fe 2+ and Fe 3+ in Fer-1 pretreated bEnd.3 cells significantly declined compared with the hypoxic only group ( Figure 4E).
Ferroptosis-related genes involving Ptgs2, Lpcat3, Ncoa4, Nfe2l2, Alox5, Alox12, and Pora were significantly up-regulated after hypoxia, while the transcription of Slc7a11 was down-regulated. However, Fer-1 administration remarkably suppressed hypoxia-induced ferroptosis ( Figure 4F,G). Notably, hypoxia treatment down-regulated the expression of proteins GPX4 and CLDN5, accompanying by significant increases in intracellular 4-HNE. Nevertheless, the protein levels of GPX4 and CLDN5 were obviously up-regulated while 4-HNE down-regulated in Fer-1 pretreated group both in vivo and in vitro ( Figure 4H,I). Taken together, these results indicated that ferroptosis of BMECs induced by hypoxia is the cause of BBB dysfunction, which could be partially ameliorated by Fer-1.

| DISCUSSION
We have demonstrated that ferroptosis is involved in hypoxia-induced BBB breakdown in our study. The major findings were as follows: (1) Hypoxia led to CLDN5 downregulation and consequently BBB leakage; (2) Hypoxia caused MDA upregulation, ROS accumulation and 4-HNE increasement under hypoxic condition; (3) Hypoxia activated most ferroptosis-related genes, induced iron accumulation, xCT and GPX4 deficiency; (4) Fer-1 partially ameliorated BBB disruption under hypoxia by inhibiting ferroptosis. These results indicated that ferroptosis inhibition might be a promising strategy to reduce BBB permeability after hypoxia exposure and might provide a potential therapeutic target for hypoxic brain injury. The brain is one of the main organs affected by hypoxia. BMVECs are the main component of BBB. BMECs exhibit greater sensitivity to oxygen deprivation than pericytes and astrocytes. 22 BMECs have a higher number of mitochondria, allowing the production of greater amounts of biological energy required to maintain BBB integrity. 31 Studies have shown that hypoxia in brain tissue can promote nitric oxide synthesis, calcium ion influx, inflammatory factors release, and hemodynamic changes, and ultimately lead to enhanced endocytosis of BMECs, which is the most important reason for BBB opening. 32 We found that hypoxia led to CLDN5 downregulation and consequently BBB leakage after hypoxia. The results were in consistent with our previous study which showed that the BBB under short-term hypoxia induction will be impaired by autophagy. 15 Imbalance of oxidative and antioxidant systems leads to tissue damage. OS is major predisposing factor for secondary brain injury. Hypoxia causes excessive ROS production in mitochondria to promote lipid oxidation, especially polyunsaturated fatty acids, also known as LPO. As expected, we have observed obvious OS in bEnd.3 cells, accompanied by significant increases in ROS, 4-HNE, etc ( Figure 2C-E). Our results were in consistent with a recent study which demonstrated that increased expression of MDA and 4-HNE might be connected with excessive ROS generation in hypoxic pulmonary arterial hypertension model rats. 33 Ferroptosis is a regulated cell death which is caused by ROS generation and iron overload. Ferroptosis is characterized by shrunken mitochondria, condensed mitochondrial membrane densities, and diminished or disappeared mitochondria crista. 34 Interestingly, mitochondria are known to be the primary source of ROS and the main venues of iron metabolism, thus it is intriguing to perceive that ferroptosis may be closely associated with OS. 35 Iron overload and loss of GSH are key mechanism involved in ferroptosis, and the production of intracellular GSH requires SLC7A11, which transports cystine (the precursor of GSH) into the cytosol. Here, we discovered that hypoxia exposure significantly decreased the mRNA level of Slc7a11, while upregulated other genes mentioned above ( Figures 3A,D and 4F,G). In our study, we observed a significant upregulation of Nfe2l2 (also named Nrf2) after hypoxia. However, some studies indicated that under ferroptosis conditions, the stability and activity of Nfe2l2 increased, which played an important role in preventing Mean ± SD, two-way ANOVA, normoxia vs. hypoxia (Total Fe), ***p < .0001; normoxia vs. hypoxia (Fe 2+ ), ***p = .0002; normoxia vs. hypoxia (Fe 3+ ), ***p < .0001; hypoxia vs. Fer-1 + hypoxia (Total Fe), ### p < .0001; hypoxia vs. Fer-1 + hypoxia (Fe 2+ ), ## p = .0032; hypoxia vs. Fer-1 + hypoxia (Fe 3+ ), ### p < .0001. (F) The mRNA expression levels of ferroptosis-related genes in the head of zebrafish. Mean ± SD, one-way ANOVA, * or # p < .05, ** or ## p < .01, *** or ### p < .001. N = 80 fish per group. (G) The mRNA expression levels of ferroptosis-related genes in bEnd.3 cells. Mean ± SD, one-way ANOVA, ## p < .01, *** or ### p < .001. N = 100 fish per group. (H) The protein levels of GPX4, CLDN5 and 4-HNE in zebrafish head were detected by immunoblotting assay. Mean ± SD, one-way ANOVA, normoxia vs. hypoxia (GPX4), ***p = .0002; hypoxia vs. Fer-1 + hypoxia (GPX4), ## p = .0017; normoxia vs. hypoxia (CLDN5), ***p < .0001; hypoxia vs. Fer-1 + hypoxia (CLDN5), # p = .0208; normoxia vs. hypoxia (4-HNE), **p = .0091; hypoxia vs. Fer-1 + hypoxia (4-HNE), ## p = .0027. N = 100 fish per group. (I) Protein expression levels of GPX4, CLDN5 and 4-HNE in bEnd.3 cells were analyzed by immunoblotting analysis. Mean ± SD, one-way ANOVA, normoxia vs. hypoxia (GPX4), ***p < .0001; hypoxia vs. Fer-1 + hypoxia (GPX4), ## p = .0018; normoxia vs. hypoxia (CLDN5), ***p = .0001; hypoxia vs. Fer-1 + hypoxia (CLDN5), ### p = .0010; normoxia vs. hypoxia (4-HNE), *p = .0182; hypoxia vs. Fer-1 + hypoxia (4-HNE), # p = .0108. N = 100 fish per group.
LPO and had an anti-ferroptosis role. 36 Nrf2 regulated the transcription of components of the GSH. 37 We speculated there was a compensatory transcriptional upregulation of Nfe2l2 by Nrf2/SLC7A11/GPX4 axis. GPX4, the major LPO scavenger, can reduce phospholipid hydroperoxide by taking advantage of GSH, thus playing a central role in inhibiting ferroptosis. Previous study on neonatal rats indicated that hypoxic-ischemic brain damage reduced GPX4 expression over the first 3 days after hypoxia exposure. 38 Consistently, we found that GPX4 markedly decreased under hypoxia in the study ( Figure 3C,E,F).
There are a variety of ferroptosis inhibitors, such as iron chelator, to suppress Fenton reaction 39 ; antioxidants like ferropstain-1 or liproxstain-1, to eliminate ROS 40 ; lipoxygenase inhibitors, like ACSL4 inhibitors 41 and so on. In our study, we observed the accumulation of oxidative products including ROS, MDA, and 4-HNE after hypoxic treatment. Therefore, we preferred Fer-1 for intervention. Fer-1 administration ameliorated BBB breakdown by reducing CLDN5 loss, decreasing iron content and improving GPX4 and GSH during hypoxia exposure ( Figure 4D-I). What we found in the study was actually in accord with some studies, including researches on intracerebral hemorrhage, 38 subarachnoid hemorrhage 42 and traumatic brain injury. 43 Taken together, these results suggested that ferroptosis plays an essential role in BBB damage under short-term hypoxia induction.
Although we elucidated that ferroptosis was involved in hypoxia-induced BBB damage, its specific mechanisms like how ferroptosis triggers BBB damage need to be explored in further studies. On the other hand, there is hitherto no gold standard for detecting ferroptosis and more complicated correlation between ferroptosis and the BBB breakdown remains to be clarified. Our previous research indicated that the hypoxic zebrafish brain is more sensitive than other organs, basing on the blood circulation in cerebrovasculature and peripheral vasculatures, the heart beating, and the circulation of red blood cells in the cerebrovasculature. 15 In this study, in order to investigate the BBB permeability impairment, we mainly applied shortterm hypoxia treatment (45 min) to generate zebrafish hypoxia models. The results showed a significant correlation of endothelial ferroptosis and the BBB impairment in hypoxia-induced zebrafish larvae (Figures 2-4). Undoubtedly, we believe that a long-term hypoxia (longer than 2 h) treatment will surely cause multi-organ injury.
In summary, our study investigated whether ferroptosis gets involved in hypoxia-induced BBB injury utilizing in vitro and in vivo models. The obtained results demonstrated that hypoxia-induced ferroptosis in BMECs and inhibition of ferroptosis by Fer-1 alleviated hypoxiainduced BBB disruption, which provides the possibility that inhibition of ferroptosis could be a novel potential target for treatment of CNS diseases correlated with BBB breakdown.