Animals

The animal-based experimental protocols were approved by the Institutional Animal Care and Use Committee of Wuxi Clinical College of Anhui Medical University (Approval No. YXLL-2019-020). The procedures were in compliance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH) and adhered to the guidelines of Animal Research: Reporting of In Vivo Experiments, amendment of 1996, article 80 − 23. Male C57BL/6 mice (weighing 22–25 g; aged 8–10 weeks) were acquired from the Laboratory Animal Facility of Nantong University (Nantong, China). They were maintained in a controlled, pathogen-free environment at an ambient temperature of around 22 °C and under a 12-h light-dark cycle. The mice had unrestricted access to food and water and were utilized for in vivo experiments. Before the experiments, the animals were in a state of optimal health without prior medical interventions or exposure to any drug treatments.

SAH model in mice

We previously described a protocol [31] for generating the mouse SAH model using the endovascular perforation technique. Briefly, anesthesia was administered to the animals through an intraperitoneal route utilizing pentobarbital sodium (50 mg/kg). To ensure optimal surgical conditions, a heating pad was employed to maintain the operative temperature at 37 ± 0.5 °C. A median incision in the neck was performed to access the left external carotid artery (ECA) and internal carotid artery (ICA). Following this, the left ECA was ligated and severed, creating a residual stump measuring 3 mm. A sharpened 5–0 monofilament nylon suture was introduced into the left ICA via the stump of the ECA to perforate the artery at the bifurcation of the anterior and middle cerebral arteries. To penetrate the bifurcation of the anterior and middle cerebral artery, the particular approach involved encountering resistance and subsequently pushing it an additional 2 mm. After removing the suture, the ICA was repercussed to induce SAH. Furthermore, sham group mice experienced identical processes without any perforation of the artery.

Mortality and neurological deficits

Mortality was documented at 48 h after SAH. Additionally, the assessment of the severity of brain injuries was carried out using the neurological scoring system provided at the identical time interval [32]. In short, the assessment relied on an 18-point framework comprising six examinations that can be graded 0–3 or 1–3. These examinations encompassed spontaneous actions (0–3), symmetry of limb movements (0–3), body awareness (1–3), extension of forelimbs (0–3), climbing ability (1–3), and sensitivity to vibrissae touch (1–3). An impartial observer assessed all examinations without knowledge of the treatment conditions. The neurological score spans from 3 to 18, with a higher score denoting enhanced neurological functions [31].

Brain water content

The standard wet-dry method, as previously explained [6, 17, 31], was used to assess the water content of the brain. collected for analysis at the 48-hour mark following SAH, where mice were euthanized and their entire brains were obtained. The brains were swiftly extracted and immediately separated into the left and right cerebral hemispheres, brain stem, and cerebellum, followed by an instantaneous measurement to ascertain their wet weight. After that, the specimens were subjected to dehydration at a temperature of 105 °C for a duration of 24 h to obtain the weight when completely dry. The water content of the brain was determined by first deducting the dry weight from the wet weight, then dividing this difference by the wet weight, and finally multiplying the quotient by 100% to obtain the percentage.

Cultured cell line

HT22 cells from the hippocampus were acquired from the Bena Culture Collection located in China. In the incubation process, the cells were maintained at 37 °C in an environment comprising 5% CO2 using Dulbecco’s Modified Eagle’s Medium (DMEM). This medium was enriched with 10% FBS provided by Gibco (New Zealand) and 1% penicillin/streptomycin at 4 mg/ml from Sigma Aldrich. The tests were carried out with cell confluency of 60–80%. The authentication of HT22 cells was based on their visual morphology and susceptibility to toxicity triggered by erastin/hemin. Additionally, the cells utilized in the studies were restricted to a maximum of 25 passages. The plates utilized for the cultivation of cell lines were not coated.

SAH model in HT22 cells

In the in vitro SAH experiments, cell death was initiated in hippocampal HT22 cells by subjecting them to hemin according to our previous study [17]. The dose-response ranged from 20 to 140 mmol/L, and this treatment lasted for 48 h. Hemin was dissolved using 0.1 M NaOH (Cat#H9039) provided by Sigma Aldrich. In this research, cells underwent treatment in the presence of designated compounds or sodium selenite (Sigma Aldrich) dissolved in ddH2O, using a concentration of 100 mmol/L hemin (LD50). Hemin was made ready and dissolved in a culture medium to a concentration of 100 mmol/L, then sterilized by filtering it through a 0.22-µm sterile filter. The assessment of cell viability was carried out 48 h following manipulation. Afterward, a warm 37 °C PBS wash was administered to the cells, and the evaluation of cell viability was conducted following the manufacturer’s recommendations, as specified by Promega. To replicate the conditions of SAH or ICH, neurons were stimulated with 100 mmol/L hemin for a duration of 48 h, as established in a previous investigation [17, 33].

Drug administration

OA (Chengdu MUST Bio-Technology Co., Ltd. Chengdu, China) was dissolved in dimethyl sulfoxide and stored at − 20 °C. To find the best dosage of OA treatment, HT22 cells were subjected to various concentrations, including 0.5 µM, 1 µM, 5 µM, and 10 µM (Fig. 1). In the end, we selected a concentration of 5 µM OA as the best choice for SAH experiments, as shown in Fig. 2b. The Supplementary Fig. S1 displays the structures of the OA. Concentrations of the drug were determined using the previously described method, with certain alterations, including an in vivo dosage of 5 mg/kg [34]. In the context of live SAH, OA was given through the tail vein intravenously at a dosage of 5 mg/kg after 2 h of SAH, which proved advantageous in monitoring post-SAH mortality. To carry out in vitro SAH experiments, cells were seeded into six-well plates and subjected to OA (10 µM) following 2 min of hemin-triggered neuronal damage. Both the sham group and the SAH + vehicle group were administered the same amount of the vehicle at corresponding time points, both in vitro and in vivo.

Fig. 1

Ferroptosis plays a vital role in SAH and hemin-induced neuronal injury in vitro.A: Cell viability results demonstrate that the addition of hemin augments neuronal damage dose-responsively, with 100 µM hemin chosen as the test concentration (n = 3). B: Schematic representation illustrates the experimental approach used to simulate SAH in mice and HT22 cells. C: Cell viability data reveal that hemin-induced cell damage mimics erastin-triggered cell damage (n = 5). D: Cell viability data show that hemin-triggered neuronal cell injury is not notably mitigated by inhibitors of autophagy (3‐MA) and caspase-dependent apoptosis (z‐VAD‐fmk), while fer-1 exhibits strong protection against such injury (n = 5). E: Representative views of Dead/Live staining (red arrows indicate dead cells, white arrows indicate live cells). Scale bar = 100 μm. F: Quantitation of MDA level increase in the in vitro SAH model (n = 3). G: Quantitation of GSH level decrease in the in vitro SAH model (n = 3). H: Quantitation of intracellular iron increase in the in vitro SAH model (n = 3). I: Quantitation of increases in ROS accumulation in the in vitro SAH model. J: Cell viability shows that si-GPX4 aggravates neuronal death caused by hemin (n = 5). K: Representative views of Dead/Live staining for increased neuronal death in response to GPX4 knockdown. Scale bar = 100 μm

Fig. 2

OA alleviates early brain injury and neuronal damage after SAH in vitro and in vivo.A: Schematic representation illustrates the experimental procedure for simulating SAH in mice and HT22 cells. B: Cell viability results show that OA addition dose-responsively diminished neuronal damage, with 5 µM OA selected as the test concentration (n = 5). C: Representative views of Dead/Live staining following OA treatment. Scale bar = 100 μm. D: Neurological scores of mice in the three groups following SAH (n = 15). E: Brain water content comparison among the three groups after SAH (n = 5; *p < 0.01 vs. Sham, #p < 0.05 vs. SAH; t-test; mean ± SD). F: TUNEL assay shows that OA attenuates neuronal death in the hippocampus following SAH. Scale bar = 50 μm

Small interfering RNA (siRNA) manipulation in vivo and in vitro

We previously described a protocol [6, 17] that served as the basis for the specific in vivo treatment method. In brief, anesthesia was administered to the animals via intraperitoneal injection. The mice were subjected to pentobarbital sodium at a dosage of 50 mg/kg and subsequently secured onto a stereotaxic apparatus manufactured by Narishige (Tokyo, Japan). This positioning allowed for the identification of the bregma location. Following the provided coordinates, a perforation was made in the left hemisphere, positioned 0.2 mm posteriorly, 1 mm laterally, and 2.2 mm below the horizontal plane defined by the bregma. A dedicated syringe from Hamilton(Reno, NV, USA) was employed to inject either GPX4-specific/Nrf2‐specific or scramble siRNA into the lateral ventricle. This injection was administered in a volume of two microliters. In this study, two categories of siRNAs obtained from Thermo Fisher Scientific were employed: one targeted mouse mRNA (Si-GPX4) or Nrf2-specific to inhibit its transcription, while the other was a random siRNA. To enhance the efficiency of silencing, the injection was administered at intervals of 12 and 24 h before SAH. In vitro, the siRNA was incubated in Opti-MEM (Thermo Fisher Scientific) [33]. The siRNA concentrations utilized for HT22 cells were following the manufacturer’s recommended guidelines. Before treatment or collection for message reduction verification, cells were subjected to siRNA treatment for a duration of 12–24 h.

RNA extraction and quantitative real-time PCR

Total RNA extraction from both cell cultures and hippocampal brain samples was carried out using TRIzol Reagent following the procedural guidelines provided by Gibco, Thermo Fisher Scientific (Waltham, MA, USA). The quantification of RNA was performed using a NanoDrop 2000 spectrophotometer (Thermo Scientific, Bremen, Germany). Afterwards, the RNA underwent reverse transcription. The process of transcription into complementary DNA (cDNA) was accomplished using the RevertAid First Strand cDNA Synthesis Kit (K1622; Thermo Fisher Scientific Inc., Rockford, IL). The determination of mRNA levels in each sample was conducted using qPCR with the SYBR Green Master Mix from Toyobo Co., Ltd. in Osaka, Japan. Normalization of the expression across all samples was achieved by referencing the β-actin gene expression. The qPCR thermocycling regimen commenced with an initial incubation at 45 °C for 2 min, followed by a step at 95 °C for 10 min. Subsequently, a total of 40 cycles were performed, involving denaturation at 95 °C for 15 s, annealing at 60 °C for 1 min, as well as extension at 72 °C for 1 min. The analysis of all samples was conducted three times. PCR amplification was the same as those described previously. Supplementary Table S1 includes the reverse transcription and qRT-PCR.

Liperfluo staining

After inducing neuronal damage with hemin or erastin, cells were exposed to OA (10 µM) after 2 h. After 48 h, the cells were treated with Liperfluo (10 µM) for 30 min at 37 °C, then harvested using trypsinization, and promptly examined under a fluorescence microscope [35].

Malondialdehyde (MDA), GSH, and iron assessment

To measure the amount of MDA in HT22 cells, a Lipid peroxidation Assay Kit (Ex/Em 532/553 nm, Ab118970, Abcam, Cambridge, UK) was conducted following the manufacturer’s guidelines [36]. The MDA and iron levels were measured using kits from Jianchen Bioengineer Institute, located in Nanjing, Jiangsu, China, following the manufacturer’s recommended protocols.

Flow cytometry detection of ROS

After the hemin and drug induction, the HT22 cells were promptly gathered and washed twice with PBS to quantify the accumulation of ROS. The cell suspension was processed and dyed with 10 mM DCFH-DA (Beyotime Biotechnology, Shanghai, China) and then kept at 37 ℃ for 30 min using the ROS Assay Kit following the instructions provided by Beyotime Biotechnology. The proportion of cells undergoing apoptosis in a sample was determined using flow cytometry. The experiments were conducted independently on three separate occasions.

Ferroxidase assay

Ferroxidase activity quantification was performed using a ferroxidase assay following a protocol previously outlined [16, 17]. In this procedure, the brain sample was introduced into a reaction system comprising PIPES (200 mM, pH 6.5) and apo-TF (50 µM), resulting in a final volume of 200 µL with the specified concentrations. Before incorporation into the assay, ferrous ammonium sulfate (100 µM), functioning as the assay substrate, was dissolved using N2-purged ddH2O. Subsequently, absorbance measurements were monitored using a PowerWave HT microplate spectrophotometer (BioTek, Burlington, VT, USA) under the following conditions: readings at 30-s intervals for 4 min with continuous agitation, at wavelengths of 310 and 460 nm, all maintained at 24 °C. After 4 min, Ferene S (500 µM) was introduced and briefly stirred, followed by an immediate measurement of absorbance (590 nm). During this investigation, which primarily examined enzyme rates, the reactions were under scrutiny for 3 min, characterized by the ferroxidase reaction sustaining a consistent linear phase. Blank reactions were performed using sample buffers containing all components of the reaction mix except for the sample. For transferrin loading (Ferric Gain), the extinction coefficients of Fe were 2.28 mM-1 Fe3 + cm-1, while for Ferrous Loss assays, it was 37.3 mM-1 Fe2 + cm-1.

Determination of cytokines

Blood specimens were obtained 48 h after surgery/anesthesia. Subsequently, these samples underwent a 15-minute centrifugation at 3,000 rpm before being stored at -80 °C until the testing phase. The concentrations of IL-1β, IL-10, IL-6, and tumor necrosis factor-alpha (TNF-α) in the serum were tested via enzyme-linked immunosorbent assay (ELISA) kits provided by R&D Systems (USA).

Cell viability detection

The assessment of viability of cells was carried out using the WST-1 technique and a kit following the guidelines provided by Jianchen Bioengineer Institute (Nanjing, Jiangsu, China). The viability assessment for the HT22 cells was conducted 48 h after the initiation of the hemin treatment. Then, the medium was replaced with fresh medium containing 10 µl WST-1 reagent. Relative cell numbers were calculated by measuring the absorbance of each well at 450 nm.

Dead/Live assay for HT22 cells

A previously outlined protocol of a Dead/Live assay [17] was utilized for HT22 cell analysis. The HT22 cells were subjected to two washes and subsequent suspension in PBS. PI (4 µM), along with calcein-AM (3 µM), was introduced to the samples, which were then kept in a light-restricted environment at ambient temperature for 20 min. Following the PBS rinsing and suspension of cells, we subjected them to examination using a Leica fluorescence microscope system (DMIL 4000B). The distinction between dead and live cells was made based on the red fluorescence emitted as a result of bound PI for dead cells and the green fluorescence generated by calcein-AM for live cells. To determine cell viability, we computed the percentage in relation to the value observed in the control cultures. This experimental procedure was conducted independently in triplicate.

TUNEL assay

The assessment of neuronal death in the brain cortex involved the application of a TUNEL assay, following a protocol as detailed in previous references [6, 17]. In brief, sections of mouse brain samples were subjected to a paraformaldehyde solution (4%) for 30 min at ambient temperature. Subsequent to this step, the sections were immersed in a 0.2% H2O2-containing methanol solution for another 30 min. Then, each sample received 50 µL of TUNEL reaction mixture, followed by a light-restricted incubation with humidity for 1 h at 37 °C. Afterward, the slides were subjected to a light-restricted staining process with DAPI for a duration of 5 min at ambient temperature to label the cell nuclei. Following this step, the nucleus visualization was achieved using Nikon laser confocal microscopy (A1, Tokyo, Japan).

Double immunofluorescence staining

Cell and neuronal counting were conducted through double immunofluorescence staining of GPX4, a protein associated with ferroptosis, and NeuN, a marker specific to neuronal staining. Detailed procedures for this method have been extensively described in our previous studies [17, 31]. In short, following a 24-hour brain fixation with a 4% solution of formaldehyde at 4 °C, the brains underwent dehydration in a solution of 30% sucrose. The brain specimens were sliced into sections, each 10 μm thick, and were used for further analysis. These sections were subjected to an overnight incubation at 4 °C with such primary antibodies (all from Abcam, except for anti-GPX4 from Affinity) as rabbit anti-NeuN polyclonal antibody (pAb; dilution 1:200, ab128886), rabbit anti-GPX4 monoclonal antibody (mAb; 1:200, ab125066), rabbit anti-AIFM2/FSP1 antibody (dilution 1:1000, DF8636), and rabbit anti-Iba1 (mAb; dilution 1:100, ab178847). Next, the brain sections underwent three rounds of PBS washing and were subsequently treated with a goat pAb secondary antibody (Abcam) targeting rabbit IgG - H&L (dilution 1:200, goat pAb, ab150077) for a duration of 2 h at ambient temperature. Next, the sections underwent an incubation with DAPI (4’,6-diamidino-2-phenylindole) for 5 min. They were then examined and analyzed using Nikon laser confocal microscopy.

Western blot analysis

The Western blot assays were conducted following the methods described earlier [17, 31]. Protein extraction was performed on the HT22 sample. Next, 50 mg of protein lysates underwent separation through sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Following this, proteins that were separated were transferred onto Immobilon nitrocellulose membranes provided by Millipore (Boston, MA, USA). These membranes underwent a blocking process in 5% skim milk-containing TBS-T, followed by a 1-h incubation with Tween 20 (0.1%) at ambient temperature. Afterward, the membranes were subjected to overnight incubation at 4 °C with the specified primary antibodies: anti-GPX4 (mAb; dilution 1:1000, rabbit mAb, ab125066; Abcam), anti-Nrf2 (dilution 1:1000, rabbit pAb, ab92946; Abcam), rabbit anti-AIFM2/FSP1 (dilution 1:1000, Affinity, DF8636), rabbit anti-Heme Oxygenase 1 (dilution 1:2000, rabbit pAb, ab13243; Abcam), anti-FTH1 (dilution 1:1000, rabbit mAb, ab18878; Abcam), rabbit anti-CD206 (dilution 1:200, DF4149, Affinity), rabbit anti-CD32 (dilution 1:500, DF6402, Affinity), anti-SLC7A11 (1:2000, rabbit mAb, ab175186; Abcam), anti-SLC3A2 (rat pAb, dilution 1:1000, #PA5-96401, Thermo Fisher Scientific), anti-NQO1 (dilution 1:2000, rabbit pAb, ab34173; Abcam), anti-GAPDH (dilution 1:1000, mouse mAb, ab8245; Abcam). Following three TBST washes, membranes were exposed to HRP-conjugated secondary antibodies (dilution 1:5000) specific to rabbit or mouse IgG. The incubation occurred at ambient temperature for 1.5 h. The protein band visualization and quantification were carried out utilizing a Bio-Rad imaging system (Hercules, CA, USA) and the ImageJ method.

Statistical analysis

Each group was subjected to testing in over three replicate experiments, and the findings were depicted as the mean values along with their respective standard deviations (SDs). Statistical assessments were executed utilizing IBM SPSS version 19 (Armonk, NY, USA). When comparing two groups, the Student’s t-test was employed, while in the case of comparing two independent variables, a one-way analysis of variance (ANOVA) was carried out, followed by a subsequent Bonferroni post-hoc test. For the analysis of data that did not conform to normal distribution or exhibited nonhomogeneous variance, the Kruskal‒Wallis test was applied, followed by a subsequent Dunn’s post-hoc test. Data were deemed significant at a P-value of less than 0.05.