All experiments were performed on C57BL/6J mice, in accordance with the Italian Animal Welfare legislation (D. L. 26/2014), implemented by the European Committee Council Directive. Experiments were approved by local veterinary authorities, the EBRI ethical committee and the Italian Ministry of Health (protocol F8BBD. N. HCD). All efforts were made to minimize animal suffering and to reduce the number of animals used.

Cortical slices were obtained from postnatal (P) day P30–P40 old animals (male and female), using a standard protocol (Schubert et al., 2001) and as reported in Giustizieri et al. (2023). Briefly, after being anesthetized with an intraperitoneal injection of a mixture of tiletamine/zolazepam (zoletyl, 80 mg/kg body weight) and xylazine (rompun, 10 mg/kg body weight), mice were decapitated. The brain was quickly removed from the skull and placed in ice-cold artificial cerebral-spinal fluid (ACSF) containing (in mM): sucrose 65, NaCl 85, glucose 10, KCl 2.5, NaH₂PO₄ 1.25, NaHCO₃ 25, CaCl₂ 0.5, MgCl₂ 7, saturated with 95% O₂ and 5% CO₂. Coronal slices (300 mm thick) were cut with a vibratome and stored at room temperature (22–24 °C) in a holding bath containing the same solution as above. After incubation for at least 1 h, an individual slice was transferred to a submerged recording chamber and continuously perfused at 33–34 °C with oxygenated ACSF of the following composition (in mM): NaCl 130, glucose 10, KCl 3.5, NaH₂PO₄ 1.2, NaHCO₃ 24, CaCl₂ 3, MgCl₂ 0.5, at a rate of 3 mL/min. For each experimental approach, data were collected from several slices (3–10) obtained at least from 3 different mice.

Electrophysiological recordings were obtained from layer 2/3 pyramidal neurons of the somato-sensory barrel cortex, visually identified with a 60 X water immersed objective, mounted on an upright Olympus CX23 microscope, equipped with differential interference contrast optics and an infrared video camera (Scientifica, UK). Membrane potential values were not corrected for liquid junction potentials. Whole cell patch clamp experiments were performed mainly in current-clamp configuration. Patch electrodes were pulled from borosilicate glass capillaries (WPI, Florida, USA). They had a resistance of 3–5 MΩ when filled with an intracellular solution containing (in mM): K-gluconate 135 mM, KCl₄, NaCl₂, HEPES 10 mM, EGTA 4 mM, MgATP 4 mM, NaGTP 2 mM; pH adjusted to 7.3 with KOH; 290 mOsm. Whole-cell current-clamp recordings were obtained using a Muticlamp 700B (Axon Instruments, Sunnyvale, CA, USA). The stability of the patch was checked by repetitively monitoring the input and series resistance during the experiments. Cells exhibiting >15% changes were excluded from the analysis.

Data were transferred to a computer hard disk after digitization with an A/D converter (Digidata 1550B, Molecular Devices). Data acquisition (digitized at 10 kHz and filtered at 1 kHz) was performed with pClamp 11.1 software (Molecular Devices, Sunnyvale, CA, USA). Input resistance and cells capacitance were measured online with the membrane test feature of the pClamp software.

Cortical slices were lysed on ice with RIPA buffer and protease inhibitors. An amount of 30 μg proteins was subjected to SDS PAGE on 4–12% denaturing gel and probed with the following antibodies: Gpx4 (1:1000, R&D Biotechne, Minneapolis, USA), 15-LOX (1:1000, Abcam, Bristol, UK), Gcl (1:6000, Proteintech, Rosemont, USA), xCT (1:1000, Abcam, Bristol, UK) and b-Actin (1:30,000, Sigma Aldrich) as loading control. Immunoreactive bands were detected using the Lite Ablot Extend Long Lasting Chemiluminescent substrate (Euroclone, Milan, Italy). Signals derived from appropriate HRP-conjugated secondary antibodies (Bethyl Laboratories, Montgomery, TX, United States) were captured by Chemi DocTM XRS 2015 (Bio-Rad Laboratories, Hercules, CA, United States) and densitometric analysis was performed using Image Lab software (Version 5.2.1, Bio- Rad Laboratories).

Glutathione levels were detected in mice cortical slices by an enzymatic re-cycling assay. Samples were de-proteinized with 5% (w/v) sulphosalycilic acid (SSA, Sigma-Aldrich, St. Louis, MO, USA) and the glutathione content was determined after dilution of the acid soluble fraction in Na-phosphate buffer containing EDTA (pH 7.5) with the ThioStar^® glutathione detection reagent (Arbor Assays, Michigan, MI, USA), using GSH as standard (Sigma Chemicals, St. Louis, MO, USA). The fluorescence was measured by an EnSpire^® Multimode Plate Reader (Perkin Elmer, Waltham, MA, USA). Protein concentration (mg/ml) was detected by the BCA method (ThermoFisher, Walthman, MA, USA) and GSH levels were expressed as nmol/mg prot.

The mice cortical slices were washed twice in PBS, sonicated 10″ for 3 times and centrifuged at 1000 × g for 5 min. Cell lysates were precipitated with 75 μL of 10% 5-sulfosalicylic acid (SSA), left for 1 h at 4 °C and centrifuged at 20000 × g for 15 min. Supernatant (25 μL) was analysed by reverse-phase high-performance liquid chromatography and fluorescence detection (HPLC-FLD). The HPLC-system was an Agilent Technologies 1,260 Series equipped with a fluorescence detector operating at Excitation wavelength 385 nm and Emission wavelength 515 nm. Data obtained were analysed with the ChemStation program (Agilent Technologies Deutschland GmbH, Waldbronn, Germany). Method validation was performed based on the US Food and Drug Administration (FDA) guideline for industry bioanalytical method validation and European Medicines Agency (EMA) guideline. The validation of the assay was performed including selectivity, specificity, linearity and limit of quantification, accuracy and precision, matrix effects and recovery, and stability. Protein concentration (mg/ml) was detected by the BCA method (ThermoFisher, Walthman, MA, USA) and cysteine levels were expressed as nmol/mg prot.

Blood samples from four children with epilepsy were collected into EDTA Vacutainer Tubes (Becton Dickinson, Rutherford, NY) and leukocytes were isolated by adding 10% dextran. After 45 min at room temperature, the upper phase was centrifuged at 2600 × g (5 min) and the pellet washed with 0.9% NaCl and stored at −20 °C until RNA extraction. Blood from five healthy children, without history or clinical evidence of neurological, neuropsychological, oncological, and inflammatory diseases, was collected at the Department of Laboratory Medicine of the Bambino Gesù Children’s Hospital during routine blood tests and used as controls. All participants signed an informed consent, and the study was approved by the Ethics Committee of Bambino Gesù Children Hospital in Rome.

Total RNA was extracted from leukocytes using Total RNA Purification Plus Kit (Norgen, Biotek Corp., Thorold, ON, Canada), according to manufacturer’s protocol. RNA quantification was performed on a NanoDrop2000 Spectrophotometer (Thermo Scientific, Waltham, MA, USA). The purity of RNA was assessed by measuring the ratio of absorbance at 260 nm and 280 nm. An amount of 1 μg RNA was reverse transcribed with the SuperScriptTM First-Strand Synthesis system and random hexamers as primers (Life Technologies, Carlsbad, CA, USA). The expression levels of SLC7A11 were measured by qRT-PCR in an ABI PRISM 7500 Sequence Detection System (Life Technologies) using TaqMan Master Mix (ThermoFisher Scientific, Walthman, MA, USA). Data were analyzed using the 2^∆∆Ct method with TBP (TATA box binding protein) as housekeeping gene and expressed as fold change relative to controls.

Statistical analysis was performed using the GraphPad Prism 5.0 Software (San Diego, CA, United States). Statistically significant differences between two groups were analyzed using Student’s t-test. All data are shown as mean ± SD. Statistical significance was defined as *p < 0.05, **p < 0.01, ***p < 0.001 compared to untreated samples.

It is well known that RSL-3 specifically acts by inhibiting the activity of GPX4, an antioxidant enzyme that protects cells against membrane lipid peroxidation (Li et al., 2021). Thus, to validate the effect of the drug, we first analysed the expression of this enzyme in RSL-3-treated cortical slices. As expected, western blot analysis revealed a significant reduction of GPX4 expression in RSL-3-treated cortical slices, if compared with untreated ones (Figure 1A). A 36 and 47% reduction was found, respectively, after 15 min and 1 h exposure to RSL-3 (5 μM), confirming previous evidence of GPX4 degradation because of RSL-3 interaction. The enzyme 15-lipoxygenase (15-LOX), which contributes to the formation of lipid peroxides by catalysing the oxygen incorporation into carbon-hydrogen bonds of polyunsaturated fatty acids (PUFAs) (Yang et al., 2016; Shintoku et al., 2017), was consistently up-regulated after 15 min (4.5-fold) and 1 h (5.3-fold) RSL-3 administration (Figure 1B). Therefore, in cortical neurons, ferroptosis generates an imbalance between GPX4 and 15-LOX, with GPX4 no longer ensuring cell protection against lipid peroxidation and 15-LOX strongly contributing to the increase of lipid oxidation.

Along with the GPX4/15-LOX imbalance, another important hallmark of ferroptosis is represented by the GSH content, for its essential role as a direct scavenger of ROS and as a cofactor for the GPX4 activity itself. As shown in Figure 2A, in mouse cortical slices, the GSH concentration was consistently decreased to 64 and 52% at 15 min and 1 h after RSL-3-treatment, respectively, thus deeply reducing the antioxidant protection of the brain. Furthermore, as the GSH deficiency may be caused not only by its increased consumption in the antioxidant protection, but also by a reduced de novo synthesis, we analysed the expression of the GSH synthesis rate-limiting enzyme, γ-glutamyl-cysteine ligase (GCL). In RSL-3-treated mouse cortical slices, a decrease of GCL protein content of 54 % and 45% was found at the two times of drug application, respectively (Figure 2B).

The GSH synthesis strongly depends on the availability of cysteine, one of the three aminoacids composing the tripeptide. To understand if the GSH synthesis could be affected by the cysteine neuronal concentration, we analysed by HPLC the cysteine levels in cortical slices undergoing RSL-3-mediated ferroptosis. Interestingly, we found 41% and 32% cysteine decreases after, respectively, 15 min and 1 h RSL-3 treatment (Figure 3). Although these values were not significantly different, the reduced expression of cysteine may contribute to slow down GSH synthesis, further weakening the antioxidant defence system of the brain.

The intracellular concentration of cysteine is largely dependent on the activity of the antiporter Xc⁻, which exchanges one molecule of glutamate outside the cell for one cysteine inside. This homeostasis is critical in central nervous system (CNS) to maintain balanced extracellular glutamate levels, to prevent excitotoxic effects, and to ensure adequate levels of cysteine for the GSH synthesis and the intracellular antioxidant response. To evaluate if low levels of cysteine in cortical slices could be due to a deficient expression of Xc⁻, we analysed SLC7A11, the catalytic subunit of system Xc⁻, in RSL-3 treated samples of cortical slices. Surprisingly, we found a significant 2.4- and 3.4-fold increase of the protein amount after 15 min and 1 h treatment, respectively (Figure 4), possibly reflecting a compensatory brain response to the GSH and cysteine decreases by the Xc⁻activation. Overall, these data provide evidence that the cysteine/GSH/GPX4 defence axis, responsible for the clearance of lipid peroxides, is down-regulated in cortical neurons undergoing ferroptosis-mediated epileptic activity, whereas 15-LOX, which promotes the formation of lipid peroxides, and Xc⁻, the modulator of excitotoxicity in brain, are up-regulated.

Based on our findings showing an imbalance between 15-LOX-catalyzed lipid peroxidation and GPX4-mediated lipid peroxides detoxification, we wondered if the main endogenous by-product of lipid peroxidation, 4-HNE, could elicit an epileptogenic response comparable to that promoted by RSL-3 in mouse cortical slices. We therefore performed electrophysiological patch clamp recordings, in current clamp mode, from layer 2/3 mouse cortical slices, before and after exposure for 15 min to 4-HNE (10 μM). As illustrated in Figure 5, the application of 4-HNE in the bath induced, within 15 min, an enhancement of spontaneous synaptic activity in cortical slices and the appearance of epileptogenic interictal bursts in all respects similar to those induced by RSL-3 (Giustizieri et al., 2023), that persisted after washing out of the drug. A burst was defined as a transient sequence of two or more action potentials occurring at high frequency (>20 Hz) (Zucca et al., 2017), often driven by a slow depolarizing after potential which allowed recruiting additional action potentials. These data demonstrate that the main by-product of lipid peroxidation can induce epileptogenic discharges in mouse cortical slices, thus leading to suggest that 4-HNE is a possible mediator between ferroptosis and epileptogenesis.

Having demonstrated the epileptogenic effect of 4-HNE on cortical slices, we wondered if, as in the case of RSL-3, 15-LOX and Xc⁻ could be affected by the treatment with 4-HNE. As reported in Figure 6, Western blot analyses showed a significant up-regulation of both 15-LOX (1.7-fold at 15 min; 2.2-fold at 1 h, Figure 6A) and Xc⁻ (1.5-fold at 15 min; 1.7-fold at 1 h, Figure 6B) after 4-HNE treatment, thus supporting a central role for these proteins in the 4-HNE-induced epileptic activity. Overall, our findings strengthen the ferroptosis as crucial contributor to the pathophysiology of epilepsy and open the way to the identification of new disease targets, with 4-HNE as a link between ferroptosis and epileptogenesis in mouse cortical slices.

Moving from these data, and from our previous study highlighting a significant increase of 4-HNE levels in plasma of children with epilepsy (Petrillo et al., 2021), we decided to evaluate the Xc⁻ expression in the blood of a little cohort of our patients, to understand if the system responsible for the Excitatory/Inhibitory balance in the brain could be modulated also in patients, such as in ferroptosis-induced mouse cortical slices. As reported in Figure 7, Xc⁻ is over-expressed in blood of 4 pediatric patients with epilepsy, thus providing evidence for this protein as a potential peripheral biomarker in epilepsy, able to mirror pathological changes occurring in the brain.