Adult male C57BL/6J mice were obtained from the Jackson Laboratory (Bar Harbor, ME USA). The study was conducted in accordance with the ARRIVE and RIGOR guidelines for the use of experimental animals (31). Experimental protocols were approved by the Institutional Animal Care and Use Committee at Johns Hopkins University School of Medicine. Sample size calculations based on our previous (28) and pilot studies indicated that eight mice/group would provide at least 80% power for detecting a 20% decrease in lesion volume at α = 0.05 (two-sided). All efforts were made to ensure minimal pain or discomfort in animals. A total of 235 male C57BL/6J mice were used for in vivo experiments and 80 C57BL/6J pups from both sexes were used for OHSC experiments. Among them, 12 mice were excluded because they died before the end of the surgery or shortly thereafter. Animals were randomly assigned to different study groups by using the randomizer form at http://www.randomizer.org. Investigators blinded to the treatment groups evaluated outcomes in all mice and performed analyses.

Mice at 10–12 weeks of age were anesthetized by 1–3% isoflurane inhalation and ventilated with oxygen-enriched air (20% O₂:80% air). The right striatum of mice was injected with 0.5 μl of 0.075 U collagenase VII-S (MilliporeSigma, St. Louis, MO, USA) at 0.1 μl per minute. Injections were administered at 0.5 mm anterior and 2.2 mm lateral of the bregma, and 3.0 mm in depth, as previously described (32). Sham-operated mice received the same treatment, including needle insertion, but were not injected with collagenase. Mice that died before the end of the surgery or shortly thereafter were excluded.

HET0016, the 20-HETE stable mimetic N-(20-hydroxyeicosa-5(Z),14(Z)-dienoyl) glycine (20-5,14–HEDGE), and 20-HETE stable antagonist N-(20-hydroxyeicosa-6(Z),15(Z)-dienoyl) glycine (20-6,15–HEDGE) were synthesized in the laboratory of Professor John R. Falck (University of Texas Southwestern Medical Center) (33). Ferrostatin-1, a specific and potent inhibitor of ferroptosis, was obtained from Xcessbio Biosciences (#M60042-2s). Caspase-3 inhibitor VII (#13306) and necrostatin-1 (#11658) were obtained from Cayman Chemical.

Mice were randomly assigned to each group. HET0016 was administrated intraperitoneally (10 mg/kg per day) beginning at 2 h after surgery. Vehicle-treated mice received the same volume of phosphate-buffered saline containing 10% dimethyl sulfoxide (MilliporeSigma) and 10% Tween-80 (MilliporeSigma). The dose and frequency of HET0016 administration was based on our previous report and pilot studies (28).

Behavior tests including the neurological deficit score assessment and wire hanging test were conducted at day 1, 3, and 7 after ICH to assess neurological deficit. A 24-point scoring system that included six tests was applied to assess neurologic deficits of mice. The tests included body symmetry, gait, climbing, circling behavior, front limb symmetry, and compulsory circling (34). Each test was graded from 0 to 4, establishing a maximum deficit score of 24. Mice with a deficit score of more than 20 at 24 h after ICH were excluded from analysis.

For the wire hanging test, a metallic wire (1 mm ×55 cm) was stretched horizontally between two posts, 50 cm above the countertop, which was covered with a pillow to minimize injury caused by falls. The hind limbs of the mice were fixed with adhesive tape to prevent them from using all four paws. The mice were placed on the wire, and latency to fall was recorded. Mice had three trials with a maximum of 180 s per trial, and the longest latency was recorded. Mice that had a latency time <10 s before and after ICH were excluded from analysis.

The procedure for culturing OHSCs was similar to that described previously (4, 28). In brief, 7-day-old C57BL/6 pups ware rapidly decapitated. The two brain hemispheres were harvested and cut coronally into 350-μm-thick slices with a Mcllwain tissue chopper. Then, hippocampal slices were immediately separated and placed on a hydrophilic PTFE cell culture insert (PICM0RG50, MilliporeSigma) in a 6-well-plate. Culture medium consisted of 50% DMEM, 25% horse serum, and 25% Hanks' Balanced Salt Solution supplemented with 6.5 mg/mL D-glucose, 25 mM HEPES, 100 U/mL penicillin G potassium, and 100 mg/mL streptomycin sulfate. The slices were cultured in a 37°C incubator with 5% CO₂. The next day, the medium was changed to 70% DMEM and 5% horse serum. The culture medium was then changed every 2 to 3 days. At 12–14 days in vitro, cultured slices were incubated for 24 h in serum-free medium and then exposed to different treatments as designed.

After incubation for 24 h in serum-free medium, the cultured slices were exposed to different reagents, including 10 μM Hb (H7379; MilliporeSigma), 10 μM HET0016, 10 μM 20-HETE antagonist (20-6,15-HEDGE), 5–20 μM 20-HETE mimetic (20-5,14-HEDGE), 25 μM caspase-3 inhibitor VII, 10 μM RIP-1–dependent necroptosis inhibitor necrostatin-1, and 10 μM ferrostatin-1 for 18 h. Slices were incubated with 5 μg/ml PI for 30 min before images were taken with an inverted fluorescence microscope (TE2000-E; Nikon, Japan). The PI fluorescence intensity was then measured by ImageJ software (NIH, Bethesda, MD). The fluorescence intensity before Hb or 20-5,14-HEDGE treatment was recorded as P0h, and that after the treatment was P18h. Maximum PI (Pmax) staining was taken as that induced by 100 μM N-methyl-D-aspartate. Cell death was calculated by the formula: (P18h–P0h)/(Pmax–P0h) × 100%.

The Image-iT lipid peroxidation kit (C10445; Life Technologies, Carlsbad, CA, USA) based on BODIPY 581/591 C11 reagent was used as an indicator of OHSC lipid peroxidation. After incubation with Hb and HET0016 or 20-6,15-HEDGE for 6 h, all treated slices were incubated with Image-iT Lipid Peroxidation Sensor at a final concentration of 10 μM for 30 min at 37°C. Some slices were observed by microscopy and others were lysed and emission fluorescence intensity detected at 590 and 510 nm on a microplate reader.

A lipid peroxidation colorimetric/fluorometric assay kit (K739; BioVision, Milpitas, CA, USA) was used to detect MDA levels in cell lysates of brain (4 mm coronal sections from the major hemorrhagic territory) 3 days after ICH according to the manufacturer's instructions. Briefly, MDA in the sample reacted with thiobarbituric acid (TBA) to generate an MDA-TBA adduct. The MDA-TBA adduct was collected and quantified colorimetrically with a microplate reader (OD 532 nm). The amount of MDA was calculated according to the standard curve and expressed as nmol/mg protein.

4-HNE is another byproduct of lipid peroxidation. For its measurement, we used western blot analysis and an ELISA kit (STA-838; Cell Biolabs, San Diego, CA, USA). The ELISA was carried out according to the manufacturer's instructions at day 3 after ICH. In brief, after coating an HNE conjugate on an ELISA plate, samples lysed from a 4 mm coronal brain section or HNE-BSA standards were added for incubation. Then an anti-HNE polyclonal antibody was added, followed by a horseradish peroxidase-conjugated secondary antibody. The quantity of HNE adduct in protein samples was determined by comparing its absorbance at 450 nm with that of a known HNE-BSA standard curve. The results were expressed as percentage relative to sham group.

GSH from brain tissue was measured with a GSH assay kit (DIGT-250; BioAssay Systems, Hayward, CA, USA) with the improved 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB) method at day 3 after ICH. The brain tissues were lysed in cold buffer containing 50 mM MES and 1 mM EDTA and the supernatant collected for assay. After deproteination, the samples reacted with DTNB to form a yellow product. The concentration of GSH was determined by comparing its optical density, measured at 412 nm, with that of calibrator and blank. The results were expressed as percentage relative to sham group.

At day 1, 3, or 7 after ICH, mice were euthanized by deep anesthesia with isoflurane and perfused transcardially before brains were collected. Each brain was post-fixed in 4% paraformaldehyde overnight and then transferred to 30% sucrose. Then, 25 μm-thick coronal brain sections spaced 150 μm apart were cut with a cryostat microtome (Leica, Wetzlar, Germany). Sections were mounted on slides and left to dry overnight at room temperature. Lastly, the slides were stained with cresyl violet (for neurons, MilliporeSigma) and luxol fast blue (for myelin). Coronal images were digitized with a microscope (Nikon Eclipse 90i) and analyzed with ImageJ software. The lesion volume was calculated as the sum of the damaged area multiplied by the distance between the sections (150 μm). The total lesion volume was also corrected to account for brain edema: corrected lesion volume = volume of non-hemorrhagic hemisphere–(volume of hemorrhagic hemisphere–lesion volume).

Iron staining was performed with OHSCs18 h after exposure to Hb with or without HET0016 and at day 3 and 7 after ICH. We applied 3,3′-diaminobenzidine (DAB)-enhanced Perls' staining (HT20; MilliporeSigma) to detect iron deposition according to the manufacturer's instructions with minor modification. OHSCs and brain slices were fixed and washed before being incubated with working solution, which was prepared by mixing equal volumes of potassium ferrocyanide solution and hydrochloric acid solution. Then the samples were blocked in 0.3% hydrogen peroxide and incubated with DAB (SK-4100; Vector Laboratories, Burlingame, CA, USA) for 3 min. Next, the nuclei were stained in working pararosaniline solution. After the slices were dehydrated and mounted on slides, images were taken under a microscope. Images were captured from five optical fields (20 × 10 magnification) in each of three sections per animal. Iron-positive cells were counted with ImageJ software.

After incubation for 24 h in serum-free medium, the cultured slices were exposed to Hb with or without HET0016 for 18 h. Then, 20-HETE was detected by using a 20-HETE ELISA kit (20H1; Detroit R&D, Detroit, MI, USA). To prepare OHSCs for the detection, we added a solution containing a final concentration of 0.1 mM triphenylphosphine onto each insert, scraped the OHSCs, and transferred the sample into 1.5 ml tubes. After saponification, the whole homogenate was acidified with acetic acid to a pH of ~3–4. Then the samples were extracted with an equal amount of ethyl acetate three times to pool the organic phase. After the organic phase was dried with nitrogen gas, the remaining residue was dissolved in N,N-dimethyl-formamide. Finally, this solution was diluted with sample dilution buffer, and 20-HETE was detected by ELISA as indicated by the manufacturer. The results were determined by comparing its absorbance at 450 nm with that of a known 20-HETE standard curve.

We extracted total protein from OHSCs 18 h after applying different treatments and from 4 mm coronal sections of the major hemorrhagic territory at day 1 and day 3 after ICH by using T-PER tissue protein extraction reagent (78510; Thermo Fisher Scientific, Waltham, MA) supplemented with protease inhibitor cocktail and phenylmethylsulfonyl fluoride. Total protein was quantified by bicinchoninic acid assay (23227; Thermo Scientific), and equal amounts were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis before transfer onto polyvinylidene fluoride membranes. Membranes were incubated with the following antibodies overnight: GPX4 (1:1,000, ab125066; Abcam, Cambridge, UK), HNE (1:1000, 393207; MilliporeSigma), cyclooxygenase (COX)-2 (1:500, 35-8200; Invitrogen, Carlsbad, CA, USA), SAPK/c-Jun NH2-terminal kinase (JNK; 1:1000, 9252; Cell Signaling Technology, Danvers, MA, USA), phospho-SAPK/JNK (T183/Y185; 1:1000, 4668; Cell Signaling Technology), p44/42 MAPK extracellular signal- regulated kinase (ERK)1/2 (1:1000, 4695; Cell Signaling Technology), phospho-p44/42 MAPK_ERK1/2 (T202/Y204; 1:1000, 4370; Cell Signaling Technology), p38 MAPK (1:1000, 8690; Cell Signaling Technology), phospho-p38 MAPK (T180/Y182; 1:1000, 4511; Cell Signaling Technology), vinculin (1:1000, 4650; Cell Signaling Technology), β-actin (1:2000, sc-47778; Santa Cruz Biotechnology, Dallas, TX). Then appropriate horseradish peroxidase-conjugated secondary antibodies were applied to detect bands under an ImageQuant ECL imager (GE Healthcare, Chicago, IL, USA) or iBright™ CL1000 imaging system (Themo Fisher Scientific). No blot was stripped and re-used for antibody probing in this study. Bands were analyzed by ImageJ software. All data were normalized to the corresponding loading control and expressed as fold change to the sham group in mice or control group in OHSCs.

PI-stained OHSCs were fixed in 4% paraformaldehyde in 6-well-plates for 2 h. Then the slices were permeabilized by 0.3% triton for 30 min and blocked with 10% goat serum for 2 h at room temperature. Next, different antibodies were added to each insert, including anti-neuronal nuclei (NeuN) antibody (1:500, 12943S; Cell Signaling Technology), anti-glial fibrillary acidic protein (GFAP; 1:500, 13-0300; Life Technologies), and anti-ionized calcium binding adaptor molecule 1 (Iba1) antibody (1:500, 019-19741; Wako, Osaka, Japan), and incubated overnight at 4°C. After appropriate secondary antibody incubation, the nuclei were labeled by DAPI (H-1200; Vector Laboratories). Pictures were taken with an inverted microscope (Nikon Eclipse 90i). We performed immunofluorescence to assess GPX4 level at day 3 after ICH and the cultured slices were exposed to Hb with or without HET0016 for 18 h before GPX4 staining. The procedure to stain GPX4 (1:200, ab125066; Abcam) in OHSCs and brain sections was similar to that described above but without PI staining. The specificity of the primary antibodies in immunostaining was assessed by performing the experiments without the primary antibodies.

Data are presented as mean ± SD or median ± confidence interval. The Mann–Whitney U-test was applied to compare ranked data from neurological scores or nonparametric data between two groups. For other normally distributed data from two-group comparisons, we applied a two-tailed Student's t-test. Differences among multiple groups were analyzed by one-way ANOVA followed by Bonferroni or Dunn's post-hoc test. All analyses were carried out with GraphPad Prism 7 software. p < 0.05 was considered statistically significant.

As a pre-animal experimental phase for physiologic and pathologic brain research, OHSCs offer outcomes that are relatively closer to those of whole-animal studies than outcomes obtained from cell culture in vitro (35). In addition, OHSCs have been successfully used to investigate the pathophysiology of intracerebral hemorrhage (4). Therefore, we chose Hb-treated OHSCs as an ex vivo ICH model. First, HET0016, a 20-HETE–specific synthesis inhibitor, and 20-6,15-HEDGE, a stable antagonist of 20-HETE were applied. As shown in Figures 1A,B, both 20-HETE inhibitors significantly reduced Hb-induced cell death in OHSCs, in agreement with previous work (28). We extended the previous work by showing that Hb can induce 20-HETE production in OHSCs, whereas HET0016 application significantly reduced it (Figure 1C). Furthermore, exposure to Hb induced more than double amounts of iron deposition relative to that in control slices, but treatment with either 20-HETE inhibitor lessened the iron deposition (Figure 1D).

To further explore how 20-HETE inhibition affects ferroptosis, we measured lipid peroxidation markers in slices exposed to Hb with or without 20-HETE inhibitors. We detected the changes in lipid peroxidation with the C11 BODIPY reagent measured at 6 h and by assessing 4-HNE levels at 18 h. Both 20-6,15-HEDGE and HET0016 attenuated Hb-induced lipid peroxidation markers (Figures 2A,B), indicating that 20-HETE participates in Hb-induced lipid peroxidation in slice cultures.

Next, we examined the effects of 20-HETE inhibition on the protein expression of GPX4, a central regulator that serves to inhibit ferroptosis. Western blotting showed that in OHSCs exposed to Hb for 18 h, both 20-HETE inhibitors significantly increased GPX4 expression (Figure 2C). Furthermore, images of GPX4 immunostaining in OHSCs confirmed this result (Figure 2D). These findings are consistent with the hypothesis that inhibition of 20-HETE decreases Hb-induced ferroptosis in brain slice cultures.

To more directly observe whether 20-HETE can induce ferroptosis, we applied the stable 20-HETE mimetic 20-5,14-HEDGE to OHSCs and assessed cell death. PI staining showed that 20-5,14-HEDGE significantly and dose-dependently induced cell death (Figure 3A). Moreover, the dead cells consisted mainly of neurons, as co-staining of PI and cell markers indicated predominant colocalization with NeuN (Figure 3B). Figure 3C shows that 20-HETE-induced cell death was significantly ameliorated by the ferroptosis-specific inhibitor ferrostatin-1. To determine how the magnitude of protection with ferrostatin-1 compared to other cell death pathway inhibitors, we treated 20-5,14-HEDGE–exposed OHSCs with the caspase-3 inhibitor VII and the RIP-1–dependent necroptosis inhibitor necrostatin-1. Necrostatin-1 had a protective effect comparable to that of ferrostatin-1, whereas the caspase-3 inhibitor exerted a small, non-significant effect (Figure 3D). The combination of all three inhibitors had the greatest protective effect. These data indicate that ferroptosis is one of the major contributors to 20-HETE–induced cell death, although cross-talk between cell death signaling pathways likely exists.

To corroborate our in vitro findings in vivo, we investigated whether treatment with HET0016 reverses ICH-induced ferroptosis in a collagenase-induced ICH model. Previous work demonstrated that the brain concentration of 20-HETE increased after ICH and that administration of HET0016 blocked this increase and reduced the lesion size at 3 days (28). Here, we first confirmed that HET0016 protected mice with ICH, but we prolonged the time course to 7 days after surgery. Mice treated with HET0016 exhibited improved neurologic deficit scores (Figure 4A), prolonged latency to fall in the wire-hanging test (Figure 4B), and smaller lesion volumes (Figure 4C) after ICH relative to those in the vehicle-treated group. These collective data indicate that HET0016 protected the brain for at least 7 days post-ICH. In addition, our team previously reported by that HET0016 administration can alleviate early neurodegeneration, as assessed by Fluoro-Jade B staining and the TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) method at 3 days after ICH (28). Here, because ferroptosis is driven by the amount of mobile iron, we measured the effect of HET0016 on ICH-induced iron overload in brain sections by DAB-enhanced Perls' staining. The number of iron-positive cells was counted per 200 × field and compared. As shown in Figure 4D, treatment with HET0016 significantly reduced the iron overload at 3 and 7 days post-ICH.

Next we investigated the effect of HET0016 on the accumulation of lipid peroxides 3 days after ICH. Treatment with HET0016 significantly blunted ICH-induced increases in the levels of the lipid peroxidation markers MDA and 4-HNE, and attenuated the ICH-induced decreases in GSH, which acts as a cofactor of GPX4 to inhibit ferroptosis (Figures 5A–C). However, the expression of COX-2, another enzyme that uses arachidonic acid as a substrate and can metabolize 20-HETE, increased after ICH, but this increase was not affected by HET0016 treatment (Figure 5D).

We also investigated changes in the ferroptosis signaling pathway. GPX4 protein was significantly increased in coronal sections at 1 day after ICH, compared to that in sham mice, but returned to baseline levels at 3 days (Figure 6A). However, with HET0016 treatment, GPX4 levels remained elevated at 3 days of recovery (Figure 6B). This elevation was confirmed by immunofluorescence staining (Figure 6C). These in vivo data are consistent with our finding that HET0016 inhibited Hb-induced ferroptosis in OHSCs. Furthermore, previous research has shown that MAPK signaling is associated with ferroptosis induced by erastin (36, 37) and by oxygen-glucose deprivation and reoxygenation (38). In addition, 20-HETE was reported to affect the MAPK pathway (25, 39), which mainly includes ERK, p38 MAPK, and c-Jun NH2-terminal kinase (JNK). We explored components of the MAPK pathway 3 days after ICH and found enhanced ERK1/2, p38 MAPK, and JNK phosphorylation in the vehicle-treated ICH group. These increases were attenuated significantly after HET0016 treatment (Figure 6D).