Triphenyl tetrazolium chloride (TTC) was acquired from Solarbio (China; Cat: G3005). Beyotime supplied the hematoxylin and eosin Staining Kit (China; Cat: C0105). RSL3 was sourced from Selleck (USA; Cat: S8155). The Iron Assay Kit originated from Abcam (USA; Cat. ab83366). Solarbio provided malondialdehyde (MDA) (Cat: BC0025), glutathione (GSH) (Cat: BC1175), and superoxide dismutase (SOD) assay kits (China; Cat: BC0175). Cellular ROS detection using DCFH-DA employed kits from Cell Biolabs (USA; Cat: STA-342). Compound C was obtained from EMD Millipore (USA; Cat: 171260).

SI (>80% purity) was procured from Shaanxi Sciphar Natural Products Co., Ltd. (China). SI with >80% purity was selected based on standardized pharmaceutical grade requirements for preclinical studies. This purity level ensures consistent bioactive compound content while minimizing potential confounding effects from impurities. According to the manufacturer’s certificate of analysis, the SI extract primarily consists of genistein (35-45%), daidzein (30-40%), and glycitein (8-12%), with the remaining components comprising their respective glycoside forms (genistin, daidzin, glycitin). For 12 mg/mL SI solution preparation, 14.12 mg SI powder underwent dissolution in 1 mL ultrapure water, with subsequent storage at 4°C.

The Ethics Center of the Second Affiliated Hospital of Guangxi Medical University approved all protocols [Approval No. 2024-KY (1124)]. Sprague-Dawley male rats (250-280g; 6 weeks old) were obtained from Beijing Weitong Lihua Experimental Animal Technology (China). Male rats were specifically selected to eliminate estrogen cycle-related confounding variables that could interfere with the phytoestrogenic effects of SI. This approach allows for clearer assessment of SI’s direct AMPK-mediated neuroprotective mechanisms independent of endogenous hormonal fluctuations. Animals had unrestricted access to food and water. Ambient conditions maintained 20–24 °C temperature and 40–60% relative humidity throughout experiments. Prior to procedures, rats underwent one week of environmental acclimatization. All experimental procedures were conducted in strict accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals and the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. For all surgical procedures, rats were anesthetized with isoflurane delivered in oxygen. Anesthesia was induced with 3-4% isoflurane in an induction chamber and maintained at 1.5-2% isoflurane via a nose cone. Adequate depth of anesthesia was continuously monitored and confirmed by the absence of pedal withdrawal reflex and corneal reflex, along with a stable respiratory rate. At the designated endpoints, rats were euthanized by intraperitoneal injection of sodium pentobarbital (150 mg/kg), following the guidelines of the American Veterinary Medical Association (AVMA) for the Euthanasia of Animals. Death was confirmed by the cessation of heartbeat and respiration, followed by immediate decapitation for rapid brain extraction. Brain tissues and serum samples were then collected and stored at −80°C for subsequent analysis.

We created cerebral ischemia-reperfusion injury models using established MACO protocols (19). Sprague-Dawley rats first underwent isoflurane-induced inhalation anesthesia. Animals were then positioned supinely for midline cervical incision under microscopic guidance (Leica, Germany). The right common, external, and internal carotid arteries (CCA, ECA, ICA) were meticulously isolated from surrounding tissues. A minimal arteriotomy was created in the ECA using Vanes-style Spring scissors, permitting insertion of a silicone-coated 4–0 nylon monofilament (Doccol Corporation, USA; 403756PK10Re) with a rounded tip. This filament was threaded into the ICA, advancing 8–10 mm past the bifurcation to occlude the middle cerebral artery. Successful occlusion was confirmed by observing a >80% reduction in regional cerebral blood flow using laser Doppler flowmetry (Moor Instruments, UK) positioned over the ipsilateral parietal cortex. Following 6 h occlusion, filament withdrawal reinstated perfusion. The 6-hour occlusion duration was selected based on preliminary studies to produce consistent moderate-to-severe cerebral infarction (approximately 30-40% of hemisphere) suitable for detecting neuroprotective effects while minimizing mortality. Core temperature was regulated at 37.0 ± 0.5°C intraoperatively.

Rats were randomized into three cohorts: Sham, MCAO, and SI groups. SI group animals receiveddaily intragastric SI (120 mg/kg) for 21 preoperative days. The 120 mg/kg dose was selected based on our dose-response studies (Supplementary Figure S1), which demonstrated optimal neuroprotective effects with this dosage in cerebral ischemia models, while the 21-day pretreatment period was chosen to ensure adequate tissue accumulation and biological activity of SI compounds. Sham and MCAO counterparts were administered equivalent saline volumes via identical route. MCAO surgery commenced 30 min post-final dosing. Post-observation, euthanasia via sodium pentobarbital overdose preceded rapid brain extraction and serum collection, with immediate storage at −80 °C.

For ferroptosis induction experiments, RSL3 (10 mg/kg, i.p.) was administered 1 hour before MCAO surgery. For AMPK inhibition studies, Compound C (20 mg/kg, i.p.) was injected 30 minutes prior to MCAO. Control groups received equivalent volumes of vehicle (DMSO in saline, <1% final concentration).

All neurological assessments were performed at 24 hours after reperfusion. Rat neurological function was evaluated using Longa’s 0–4 scoring system (20): 0 indicating normal mobility; 4 representing comas with locomotion inability. Evaluators remained blinded to group assignments throughout. Cognitive impairment was additionally quantified through modified neurological severity scores (mNSS) (21). Longa’s scoring criteria were as follows: 0 = no neurological deficits; 1 = failure to extend left forepaw fully; 2 = circling to the left; 3 = falling to the left; 4 = inability to walk spontaneously with depressed consciousness. mNSS evaluates motor function, sensory function, reflex responses, and balance on a scale of 0-18, with higher scores indicating greater neurological impairment.

Brains were sectioned coronally at 2 mm thickness. Sections underwent 37°C TTC staining for 30 minutes, followed by 10-minute immersion-fixation in 10% paraformaldehyde. Infarct percentage was computed as: infarct rate = infarct area/total slice area × 100%. Brain water content was determined according to established methodology (22). After extraction, fresh tissue mass was recorded as wet weight (WW). Samples were then desiccated to constant mass (DW) in drying ovens. Edema percentage was calculated: [(WW - DW)/WW] × 100%.

Ferrous and total iron concentrations were measured using the Iron Assay Kit per manufacturer specifications. Optical density at 593 nm was determined spectrophotometrically. Iron measurements were performed in triplicate using freshly prepared standard solutions (0–100 mM FeSO₄). Instrument calibration was verified using certified reference materials before each batch analysis. Background absorbance was measured using tissue-free buffer controls.

SOD, MDA, and GSH activities were evaluated with corresponding assay kits following kit protocols. Resultant chromogenic reactions were quantified via microplate reader absorbance measurements. All biochemical assays were conducted in triplicate with concurrent analysis of kit-provided standards and quality controls. For MDA assays, 1,1,3,3-tetraethoxypropane standards (0–40 nmol/mL) were used for calibration. GSH measurements utilized reduced glutathione standards (0–60 mM), while SOD activity was calibrated against xanthine oxidase inhibition standards. Blank controls (assay buffer without tissue) and positive controls (oxidized samples) were included in each experimental run.

Cellular ROS levels were detected using the xylenol orange (XO) - based ferrous ion oxidation - xylenol orange (FOX) kit per manufacturer guidelines. Absorbance signals (at a characteristic wavelength of 560 nm for the XO - Fe³⁺ complex) were measured with a UV - visible spectrophotometer for quantitative analysis. ROS detection was performed in triplicate with H₂O₂ - treated positive controls and PBS - treated negative controls included in each experiment. The absorbance response for ROS detection was calibrated using known concentrations of H₂O₂ (0–200 mM). Background absorbance was measured in cells treated with vehicle alone (to eliminate interference from non - specific absorption of the vehicle or cellular components).

Western blot analysis targeted key ferroptosis regulatory proteins including glutathione peroxidase 4 (GPX4), acyl-CoA synthetase long-chain family members 3 and 4 (ACSL3, ACSL4), cystine/glutamate antiporter (xCT), and ferritin subunits (FTMT, FTH), as well as AMPK signaling pathway components (AMPK, phospho-AMPK). Brain tissue proteins were isolated using RIPA buffer containing protease inhibitors (PMS). After BCA quantification, we separated proteins on 10% SDS-PAGE gels and transferred them to PVDF membranes. Membranes were blocked with skim milk, then probed with appropriately diluted primary antibodies during overnight incubation at 4 °C. After washing, horseradish peroxidase-conjugated secondary antibodies were applied for 2 hours. Chemiluminescent visualization revealed target bands using Prime Western Blotting Reagent. Band intensities were quantified via ImageJ densitometry. Antibody specifications appear in Supplementary Table S1.

SPSS 22.0 facilitated all statistical analyses. Continuous variables are presented as mean ± SD ( x¯ ± s). Between-group comparisons employed one-way ANOVA followed by Tukey’s HSD post-hoc test for pairwise comparisons when significant main effects were detected. Significance threshold was established at P < 0.05.

To evaluate SI’s therapeutic potential against cerebral ischemia-reperfusion injury, middle cerebral artery occlusion models were employed. TTC staining demonstrated substantial TTC-negative regions in MCAO group brains. SI administration significantly diminished cerebral infarct volumes (Figure 1A). Brain edema quantification revealed markedly elevated water content in ischemic rats, substantially reduced by SI treatment (Figure 1B). Neurological assessment indicated higher deficit scores and mNSS values in MCAO animals versus controls. SI intervention effectively ameliorated these pathological increases (Figures 1C, D).

Given ferroptosis involvement in CIRI pathogenesis, cerebral iron accumulation was quantified. MCAO rats exhibited significantly increased Fe²⁺ and total iron versus sham controls. SI treatment notably attenuated cerebral iron overload (Figures 2A, B). Oxidative stress analysis showed SI substantially decreased MDA while elevating GSH and SOD activities (Figures 2C, E). Additionally, TfR1 expression was examined to comprehensively evaluate iron homeostasis (Supplementary Figure S2). MCAO significantly increased TfR1 expression compared to Sham, indicating enhanced iron uptake under ischemic conditions. SI pretreatment markedly reduced TfR1 levels. Besides, SI suppressed ischemia-induced ROS generation (Figure 2F). The reliability of ROS detection was validated using PBS (negative control) and H₂O₂ (positive control) treatments, which showed expected baseline and elevated ROS levels, respectively (Supplementary Figure S3). Western blotting showed that SI decreased ACSL4 expression and increased GPX4, ACSL3, and xCT expression. Ferritin expression (FTMT, FTH) was also enhanced (Figure 2G). These results indicated AMPK’s involvement in SI-mediated iron metabolism regulation.

To ascertain whether ferroptosis inhibition underlies SI’s neuroprotection, the ferroptosis inducer RSL3 was co-administered. RSL3 exacerbated cerebral infarctions relative to MCAO controls (Figure 3A). Brain water content increased significantly in RSL3-treated animals (Figure 3B). Neurological impairment severity was greater in this group (Figures 3C, D). Critically, SI co-treatment reduced RSL3-induced exacerbation of tissue damage and functional deficits (Figure 3).

Subsequent investigation focused on SI’s molecular actions against CIRI. A marked reduction in the p-AMPK/AMPK ratio was observed within the brain tissue of MCAO model rats. Treatment with SI, however, substantially elevated this ratio (Figure 4). To further validate AMPK pathway specificity, we examined the phosphorylation status of AMPK downstream targets ACC and mTOR. As shown in Supplementary Figure S4, MCAO increased p-ACC and decreased p-mTOR levels, which were reversed by Compound C treatment. SI pretreatment modulated these downstream targets, and co-administration of Compound C with SI attenuated SI’s effects, confirming that SI’s protective effects are mediated through AMPK signaling.

To define the AMPK pathway’s role, compound C was administered to inhibit AMPK. Relative to MCAO controls, compound C elevated cerebral Fe²⁺ concentrations and total iron content (Figures 5A, B). Compound C also significantly increased MDA levels while markedly decreasing GSH and SOD levels compared to the MCAO group (Figures 5C-E). Furthermore, compound C notably elevated ROS generation in rat brain tissues relative to MCAO controls (Figure 5F). Additionally, compound C administration significantly upregulated ACSL4 expression, whereas it downregulated expression of GPX4, ACSL3, xCT, FTH, and FTMT (Figure 5G). Notably, SI co-administration effectively counteracted the pro-ferroptotic consequences induced by compound C (Figure 5).

The contribution of AMPK signaling to SI’s therapeutic impact on CIRI was further assessed. Administration of compound C significantly enlarged the cerebral infarct area and increased brain water content compared with the MCAO group (Figures 6A, B). Compound C also exacerbated neurological impairment in rats (Figures 6C, D). In contrast, SI treatment effectively reversed these detrimental outcomes (Figure 6). Collectively, these results confirm that SI alleviates CIRI by suppressing ferroptosis, an effect dependent on activation of the AMPK signaling pathway.