A total of 60 male Sprague-Dawley (SD) rats (aged 10–12 days; 24–28 g) were purchased from the Laboratory Animal Center, Fujian Medical University (Fuzhou, China). Rat pups were housed in the same standard laboratory polycarbonate cage with their mothers at a suitable temperature (21±2°C) and humidity (50±10%) with 12/12-h light/dark cycles to maintain their circadian rhythm. Maternal milk was used to feed the rat pups, while adult female rats had free access to standardized granular food (Beijing Keao Xieli Feed Co., Ltd.) and tap water. The 10–12 day old rat pups were randomly divided into the control, phenylhydrazine (PHZ) and deferoxamine (DFO) + PHZ groups, with 12 rats in each group.

To establish bilirubin encephalopathy secondary to hemolysis, 75 mg/kg PHZ (Sigma-Aldrich; Merck KGaA) was injected intraperitoneally once daily for 2 days, based on previous reports (26,27). Rats in the DFO + PHZ group were injected intraperitoneally with DFO (100 mg/kg; cat. no. D9533; Sigma-Aldrich; Merck KGaA) 1 h prior to administration of PHZ (28,29) whilst rats in the PHZ group received the same volume of saline. Control rats received the same volume of saline in parallel. Subsequently, 24 h after the last dosing, all animals were anesthetized intraperitoneally with 100 mg/kg pentobarbital sodium and then euthanized by decapitation to collect trunk blood and brain tissue. A total of 0.5 ml blood from each rat was collected and centrifuged at 1,500 × g at 4°C for 10 min and then preserved at −80°C for later biochemical analysis. Brain tissue was immediately collected, placed in liquid nitrogen for 5 min, and then stored at −80°C for future analysis. All treatments were performed with care and rat suffering was minimized. During the experiment, none of the rats reached any of the following humane endpoints: Inability to drink the maternal milk, labored breathing, inability to stand or no response to external stimuli. Death was verified by monitoring the cessation of breathing and heartbeat and pupil dilation. No animals were found dead before the end of the experiment.

The highly differentiated PC12 cell line, derived from a pheochromocytoma of the rat adrenal medulla, resembles neurons and is widely used in the study of neurological diseases in vitro (30,31). PC12 cells (Wuhan Boster Biological Technology, Ltd.) were cultured in Dulbecco's Modified Eagle Medium (cat. no. C11995500BT; Gibco; Thermo Fisher Scientific, Inc.), supplemented with 10% fetal bovine serum (cat. no. SV30207.03; HyClone; Cytiva) and 100 U/ml penicillin-streptomycin (cat. no. V900929; Sigma-Aldrich; Merck KGaA). Cells were cultured at 37°C as monolayers in a humidified atmosphere containing 5% CO₂. Based on MTT cell viability assay results (Fig. S1) and the level of superoxide dismutase (SOD), MDA and reactive oxygen species (ROS) in the PC12 cells exposed to 0, 200 and 400 µM ferric ammonium citrate (FAC; cat. no. F5879; Sigma-Aldrich; Merck KGaA) for 24 h (Fig. S2), 400 µM FAC treatment for 24 h was selected for subsequent experiments, to assess the mechanism of iron overload-induced neurotoxicity in vitro. PC12 cells were pretreated with 10 µM ferrostatin-1 (Fer-1; cat. no. SML0583; Sigma-Aldrich; Merck KGaA) or 100 µM DFO for 1 h at 37°C and exposed to 400 µM FAC.

Serum hemoglobin and hematocrit levels were assessed using an automated analyzer (Coulter LH 780; Beckman Coulter, Inc.). Another automated analyzer (Architect ci16200 Integrated System; Abbott Pharmaceutical Co. Ltd.) was used to measure the serum levels of total bilirubin, indirect bilirubin and direct bilirubin using a diazo reagent-based method (32). All procedures strictly followed the manufacturers' instructions.

Serum levels of S100B were assessed using the Rat S100B ELISA kit (cat. no. E-EL-R0868; Elabscience Biotechnology, Inc.), and those of serum NSE were assessed using the Rat NSE ELISA Kit (cat. no. E-EL-R0058c; Elabscience Biotechnology, Inc.) according to the manufacturer's instructions.

A total of 15 mg brain tissue from six rats in each group was placed into 200 µl ice-cold 10% radioimmunoprecipitation assay lysis buffer (RIPA; cat, no. P0013E; Beyotime Institute of Biotechnology). Brain samples were homogenized using a tissue homogenizer (cat. no. KZ-III-F; Wuhan Servicebio Technology Co., Ltd.) and centrifuged at 1,000 × g for 10 min at 4°C. The supernatant was then transferred into microcentrifuge tubes. For cell extracts, PC12 cells were seeded in the 6-well plate at a density of 50–60% for 24 h, and then exposed to FAC. A total of 200 µl RIPA lysis buffer were added to plate to homogenize the sample and centrifuged at 1,000 × g for 10 min at 4°C, waiting for next measurement.

SOD activity and total GSH/GSSG detection kits (cat. nos. A001-3-1 and A061-1-1; Nanjing Jiancheng Bioengineering Institute) were used to assess SOD and GSH/GSSG levels, respectively. Tissue iron detection kits (cat. no. A039-2-1; Nanjing Jiancheng Bioengineering Institute) were used to assess tissue iron levels. An MDA detection kit (cat. no. S0131S; Beyotime Institute of Biotechnology) was used to assess MDA levels. All assays were conducted in triplicate according to the manufacturers' protocols.

Proteomic analysis of separate triplicate brain tissue samples from the control and PHZ groups was performed by tandem mass tag (TMT) labeling (Cat. no. A44520; Thermo Fisher) (33) and liquid chromatography-tandem mass spectrometry (Q Exactive™ HF-X; Thermo Fisher) (34). The detection was conducted by multiple reaction monitoring (MRM) in negative ionization mode. The electrospray voltage was 2.1 kV. The m/z scan range was 350–1600 for the full scan, and intact peptides at a resolution of 120,000 were detected in the Orbitrap. We set the automatic gain control (AGC) at 3E6, and the fixed first mass was 100 m/z. The nitrogen gas temperature was at 300°C with a flow rate of 5 l/min and nebulizer pressure was 40 psi. The differentially abundant protein (DAP) screening parameters were adjusted to P<0.05 and log₂ fold change=0, KEGG enrichment analysis was performed using ClueGO (v2.5.6; http://apps.cytoscape.org/apps/cluego) (35). The databases used were KEGG_20.05.2019 (genome.jp/kegg/pathway.html) and Gene Ontology (GO) (https://www.geneontology.org/)_Biological Process-EBI-Uniprot-GO Annotation_20.05.2019 (https://david.ncifcrf.gov/home.jsp). The GO term ‘fusion mode’ was selected according to the P-value <0.05.

The number of differential protein sequences in the protein database (Maxquant: v1.6.15.0; http://www.maxquant.org/) screened according to a 1.2 expression difference multiplesin PHZ group vs. Control group, and then compared with those in the protein network interaction database of the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING; v. 11.0; http://cn.string-db.org/), with a confidence score of >0.7 (high confidence). The differential protein interaction was analyzed, and the R package (Dev-version: 0.4; christophergandrud.github.io/networkD3/) ‘networkD3’ function was used to visually display the differential protein interaction network.

In the FAC model, PC12 cells were seeded in 96-well plates for 24 h at a density of 1×10⁴ cells/well. Cells were then treated with different concentrations of FAC (0, 100, 200, 400 and 800 µM, 1 and 5 mM) for 3, 6, 12 and 24 h. The viability of PC12 cells was assessed using MTT (cat. no. F5655; Sigma-Aldrich; Merck KGaA). Briefly, after the aforementioned treatments, MTT (5 mg/ml) was added to the 96-well plate. The plate was incubated at 37°C for 4 h to allow the formazan crystals to form. The medium was then discarded and 100 µl dimethyl sulfoxide was added to dissolve the formazan. Absorbance was measured at a detection wavelength of 570 nm using a plate reader (Thermo Fisher Scientific, Inc.).

After PC12 cells were seeded in the 35 mm confocal dish (cat. no. FCFC016; Beyotime Institute of Biotechnology; China) at a density of 50 %, the cells were treated with FAC as aforementioned for 24 h, and then were rinsed with phosphate-buffered saline (PBS) three times. To visualize neuronal lipid oxidation, 10 µM BODIPY 581/591 C11 (cat. no. D3861; Thermo Fisher Scientific, Inc.) was added to each well and incubated at 37°C for 0.5 h and stained with 1 µg/ml DAPI for 5 min at 37°C. Cells were rinsed with PBS three times and imaged using a confocal laser scanning microscope (SP5; Leica Microsystems GmbH) (36). The argon laser (excitation, 488 nm; emission, 500–535 nm) was used to detect oxidized lipids, whereas the white light laser (excitation, 561 nm; emission, 573–613 nm) was used to detect non-oxidized lipids.

PC12 cells were seeded in the 6-well plates for 24 h at a density of 2×10⁵ cells/well and then were treated with FAC for 24 h as aforementioned. Subsequently, cells were stained with the ROS indicator 2′,7′-dichlorodihydrofluorescein diacetate (cat. no. S0033S; Beyotime Institute of Biotechnology) for 30 min at 37°C. Subsequently, the cells were washed with PBS three times and digested with 0.25 % trypsin (cat. no. C0201S; Beyotime Institute of Biotechnology) at 37°C for 2 min. Finally, labelled cells were detected with a flow cytometer (LSRFortessa™ X-20; BD Biosciences) and analyzed by Flow v10.9 (BD Biosciences) (37).

FerroOrange (cat. no. F374; Dojindo Laboratories, Inc.) was used to detect intracellular Fe²⁺. PC12 cells were seeded in the 12-well plates for 24 h at a density of 2×10⁵ cells/well and then were treated with 400 µM FAC as aforementioned., They were rinsed with PBS and stained with 1 µM FerroOrange at 37°C for 0.5 h. The cells were rinsed with PBS three times and imaged under a fluorescence microscope (38).

PC12 cells were seeded in a 10 cm dish for 24 h at a density of 1×10⁶ cells/well and then exposed to FAC as aforementioned. Subsequently, cells were fixed in 2.5% glutaraldehyde (diluted in 0.1 µM PBS; pH 7.4) at 25°C for 24 h and then post-fixed in 1% osmium tetroxide (dissolved in 0.1 µM PBS; pH 7.4) at 25°C for 60 min. After dehydration with ethanol (50, 70, 80, 90 and 100%), samples were embedded by resin (cat. no. 45347; Merck) with different condition (37°C for 12 h; 45°C for 12 h; 60°C for 12 h) and ultrathin sectioning (50 nm). Then, the samples were stained with uranyl acetate at room temperature for 30 min and stained with lead citrate at room temperature for 8 min. Digital images were captured using TEM (FEI Tecnai G2 F30; Thermo Fisher Scientific, Inc.) (39).

Proteins were extracted from brain tissue using RIPA lysis buffer (cat. no. P0013E; Beyotime Institute of Biotechnology) supplemented with phenylmethylsulfonyl fluoride and then denatured at 95°C in 5X protein loading buffer (cat. no. P0286; Beyotime Institute of Biotechnology) for 10 min. The concentration of protein in sample were measured by Enhanced BCA Protein Assay Kit (cat. no. P0009; Beyotime Institute of Biotechnology) according to the manufacturer's instructions. An equal amount of protein samples (20 µg/lane) were added to per lane and then were separated on 10% gels by SDS-PAGE. Proteins were then transferred to 0.2 µm PVDF membranes and blocked with 5% non-fat milk for 60 min at room temperature. PVDF membranes were incubated overnight at 4°C with primary antibodies against the following: Rabbit acyl-coenzyme A synthetase long-chain family member 4 (ACSL4; 1:1,000; cat. no. A16848; ABclonal Biotech Co., Ltd.); ferritin heavy chain (FTH; 1:1,000; cat. no. A19544; ABclonal Biotech Co., Ltd.); transferrin receptor 1 (TFRC; 1:1,000; cat. no. A5865; ABclonal Biotech Co., Ltd.); ferroportin 1 (FPN1; 1:1,000; cat. no. A14884; ABclonal Biotech Co., Ltd.); divalent metal transporter 1 (DMT1; 1:1,000; cat. no. A10231; ABclonal Biotech Co., Ltd.); iron regulatory protein 2 (IRP2; 1:1,000; cat. no. A6382; ABclonal Biotech Co., Ltd.); GAPDH (1:5,000; cat. no. A19056; ABclonal Biotech Co., Ltd.); xCT (1:1,000; cat. no. ab175186; Abcam); GSH peroxidase 4 (GPX4; 1:1,000; cat. no. ab125066; Abcam) and ferroptosis suppressor protein 1 (FSP1; 1:1,000; cat. no. AVARP09054_P050; Aviva Systems Biology, Corp). After washing by TBS buffer with 0.1% Tween-20 (cat. no. ST1727; Beyotime Institute of Biotechnology), PVDF membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (1:8,000; cat. no. 31460; Thermo Fisher) at 37°C for 60 min. Enhanced chemiluminescence detection was performed (cat. no. 1705061; BIO-RAD) (40) and ImageJ software (version 1.53; National Institutes of Health) was used to semi-quantify the protein bands.

Data are presented as mean ± standard error of the mean. All analyses were conducted using GraphPad Prism 8.0 (GraphPad Software; Dotmatics). Significance of differences between two groups was analyzed using two-way independent-sample t-test. For three or more groups, one-way analysis of variance followed by Tukey's post-hoc test was used to determine significance. P<0.05 was considered to indicate a statistically significant difference.

To establish a hemolytic hyperbilirubinemia model of neonatal SD rats, PHZ was used to increase erythrocyte turnover via hemolysis (Fig. S3A). In the PHZ group, the appearance of the neonatal rats and their brains was more jaundiced than that of the control group (Fig. S3B and C). PHZ group had significantly decreased hematocrit and hemoglobin levels compared with the control group (Fig. S3D and E). Moreover, the serum levels of three types of bilirubin (total, direct and indirect bilirubin) significantly increased in the PHZ group compared with that of the control group (Fig. S3F-H), indicating that the hemolytic hyperbilirubinemia model was established successfully.

Subsequently, changes in the brain injury markers NSE and S100B in the PHZ group were assessed. Compared with those in the control group, serum S100B and NSE levels were significantly increased in the PHZ group (Fig. 1A and B). Simultaneously, the concentration of iron significantly increased in the brain tissue of the PHZ group compared with that of the control group (Fig. 1C). Disturbances in oxidative balance caused by hemolytic hyperbilirubinemia were then assessed. SOD, which serves an important function in the first line of antioxidant defense, significantly decreased in the brains of the PHZ group compared with that of the control group (Fig. 1G). Compared with the control group, the GSH and GSH/GSSG levels significantly decreased in the PHZ group. However, there was no difference in the level of GSSG between the two groups (Fig. 1D-F). Similarly, the level of lipid peroxidation, as assessed by detecting the levels of MDA, significantly increased in the PHZ group compared with that of the control group (Fig. 1H). Overall, these results suggested that hemolytic hyperbilirubinemia could disrupt the redox balance in rat brains, resulting in brain injury. However, the specific mechanism of action remained unclear. Therefore, mass spectrometry analysis was performed to evaluate the potential molecular mechanisms of hemolytic hyperbilirubinemia.

Proteomic analysis of brain tissue samples from the control and PHZ groups were performed in triplicate using TMT labelling to evaluate whether ferroptosis is involved in HHIBD. KEGG analysis using ClueGO software demonstrated the enrichment of ferroptosis-related proteins in the PHZ group compared with those in the control group. When the DAPs screening parameters were adjusted to P<0.05 and log₂ fold change=0, compared with the control, a total of 26 ferroptosis-associated DAPs were demonstrated to be enriched in the PHZ group (Fig. 2A). These DAPs were also involved in other metabolic pathways, including mineral absorption, glutathione metabolism, and fatty acid biosynthesis. The STRING database was then used to determine the protein-protein interaction network (Fig. 2B; confidence score >0.7) among the screened ferroptosis-associated DAPs.

To evaluate the hypothesis that ferroptosis has a role in HHIBD, ferroptosis-associated proteins were assessed using western blotting. As it was previously demonstrated that the iron concentration in the brain tissue of the PHZ group significantly increased compared with that of the control group (Fig. 1C), protein-controlled iron import and export were assessed. The levels of the iron transporters TFRC (Fig. 3A and B) and DMT1 (Fig. 3A and C), which represent important components of ferroptosis that regulate iron levels, significantly increased in the PHZ group compared with that of the control group, whilst FPN1 levels did not significantly change (Fig. S4A and B). The results also demonstrated that levels of the iron storage protein FTH significantly increased in the PHZ group compared with that of the control group (Fig. 3D and F). Furthermore, the levels of IRP2, which is involved in the regulation of iron metabolism, significantly increased in the brain in the PHZ group compared with that of the control group (Fig. 3D and E). ACSL4 is a specific biomarker and driver of ferroptosis (41), while FSP1 has been identified as key molecules in independent pathways associated with ferroptosis inhibition (42). ACSL4 levels significantly increased and FSP1 levels significantly decreased in the PHZ group compared with that of the control group (Fig. 3G-I). The level of GPX4, an inhibitor of ferroptosis that converts lipid hydroperoxides into non-toxic lipid alcohols, was also demonstrated to have significantly increased after the onset of hemolytic hyperbilirubinemia in the PHZ group compared with that of the control group (Fig. S4A and C). By contrast, the level of xCT, an important antioxidant defense molecule, did not change significantly (Fig. S4A and D).

DFO and its function as an iron chelator and specific inhibitor of ferroptosis (28,29), was used to further assess the function of ferroptosis in HHIBD. Compared with that in the PHZ group, DFO did not significantly influence the levels of hematocrit, hemoglobin and bilirubin in the DFO + PHZ group (Fig. S5). However, DFO pretreatment significantly reduced the levels of NSE and S100B, compared with that of the PHZ only group, indicating that brain injury was alleviated (Fig. 4A and B). DFO mitigated the imbalanced iron levels in the brain (Fig. 4C), thereby partially alleviating the redox imbalance compared with the PHZ group. The levels of GSH and SOD were increased following treatment with DFO (Fig. 4D-G), whereas MDA levels were significantly reduced, compared with that of the PHZ only group (Fig. 4H). There was no difference in the level of GSSG between the DFO + PHZ groups and the PHZ group (Fig. 4E).

The effect of DFO on the proteins related to ferroptosis was then evaluated. The protein levels of DMT1 and TFRC, which regulate iron transport in and out of cells, were decreased following DFO treatment compared with those of the PHZ group (Fig. 5A-C), whereas the FPN1 level was not significantly altered (Fig. S6A and B). Similarly, DFO restored the levels of IRP2 and FTH proteins compared with those of the PHZ group (Fig. 5D-F). Moreover, the altered levels of the key proteins in the ferroptosis pathway, ACSL4 and FSP1, were decreased following DFO treatment (Fig. 5G-I). However, the protein level of xCT did not significantly change following DFO pretreatment (Fig. S6A and D). Furthermore, compared with the PHZ group, the GPX4 level decreased in the DFO + PHZ group (Fig. S6A and C).

Pretreatment with Fer-1 or DFO was demonstrated to antagonize the reduction in cell proliferation induced by FAC (Fig. 6A). The results also demonstrated that FAC exposure significantly increased MDA and reduced SOD activity levels and induced excessive generation of ROS compared with those in control group (Fig. 6B-E). Furthermore, FAC treatment also increased the Fe²⁺ content in PC12 cells, as indicated by significantly higher FerroOrange signals compared with those observed in control cells (Fig. 6F and G). Changes in the mitochondria were also demonstrated, with FAC treatment resulting in a reduction in mitochondrial volume and increased membrane density (Fig. 6H). Furthermore, the level of lipid peroxidation assessed using confocal laser scanning microscopy demonstrated that FAC significantly increased lipid peroxidation compared with that in control cells (Fig. 6I and J). However, pretreatment with 10 µM Fer-1 and/or 100 µM DFO effectively attenuated the aforementioned damage induced by FAC treatment.