Improving epilepsy management by targeting P2 × 7 receptor with ROS/electric responsive nanomicelles

Abstract

Background

The intricate pathogenesis of epilepsy, characterized by abnormal neuronal discharges and neuroinflammation, underscores the critical involvement of the adenosine triphosphate (ATP)-P2X purinoceptor 7 (P2 × 7) receptor pathway in inflammation activation. To address this, a reactive oxygen species (ROS)/electric-responsive d-α-tocopheryl polyethylene glycol 1000 succinate (TPGS)–ferrocene–poloxamer nanomicelle (TFP@A) was engineered to deliver the P2 × 7 receptor antagonist A 438,079, aiming to provide a targeted therapeutic strategy for epilepsy management.

Methods

The study meticulously designed and characterized TFP@A for precise drug delivery through various techniques including transmission electron microscopy (TEM), dynamic light scattering (DLS), and high-performance liquid chromatography (HPLC). Cellular uptake and blood-brain barrier (BBB) permeability were evaluated using fluorescein isothiocyanate (FITC)-labeled TFP@A in vitro and in a brain endothelial cell line (bEnd.3) cell BBB model. In vivo distribution and safety assessments were conducted in an epilepsy mouse model. The impact of TFP@A on epilepsy was investigated through seizure analysis, electroencephalogram (EEG) recordings, and inflammatory pathway assessment.

Results

TFP@A exhibited a robust drug release profile under ROS and electrical stimulation conditions. In vitro studies demonstrated its efficacy in scavenging ROS, reducing oxidative stress, and alleviating cell apoptosis in epilepsy models. Efficient cellular uptake, BBB penetration, and in vivo accumulation in the brain were observed. Notably, TFP@A effectively modulated the P2 × 7 receptor (P2 × 7R)-nucleotide-binding oligomerization domain-like receptor family pyrin domain-containing 3 (NLRP3) pathway, inhibiting inflammatory mediators and promoting anti-inflammatory responses.

Conclusion

TFP@A loaded with the P2 × 7 receptor antagonist showcases potential therapeutic benefits in suppressing NLRP3 inflammasome activation, mitigating microglial-neuron crosstalk, and ameliorating epilepsy symptoms, positioning it as a promising avenue for targeted epilepsy treatment.

Graphical Abstract

Introduction

Epilepsy is a chronic neurological disorder that significantly impacts patients’ quality of life. It is characterized by excessive, synchronized abnormal neuronal discharges in the brain, resulting in recurrent seizures and neurological dysfunction [1]. Globally, epilepsy affects approximately 50 million individuals, and with nearly one-third of patients exhibiting resistance to currently available antiepileptic drugs (AEDs), thereby posing a significant clinical challenge [2,3,4]. While traditional AEDs primarily target neuronal excitability, recent studies have highlighted the critical role of neuroinflammation in the onset and progression of epilepsy [5]. The complexity and diversity of epilepsy pathogenesis, particularly the vicious cycle between neuroinflammation and abnormal neuronal discharges, underscore the urgent need to develop novel therapeutic strategies.

The P2 × 7 receptor, an adenosine triphosphate (ATP)-gated ion channel, plays a central role in mediating neuroinflammatory responses within the central nervous system (CNS) [6]. Sustained activation of P2 × 7 receptors stimulates microglial cells to release large quantities of pro-inflammatory cytokines, including interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α), thereby exacerbating neuroinflammation through the activation of the NLRP3 inflammasome [7, 8]. In addition, dysregulated P2 × 7 receptor activity facilitates inflammatory crosstalk between neurons and microglia, contributing to neuronal injury, apoptosis, and broader neuropathological changes [9]. Mounting evidence indicates that inhibition of the P2 × 7 receptor not only attenuates neuroinflammation but also interrupts the vicious cycle linking abnormal neuronal discharges to immune activation [10, 11]. Thus, the P2 × 7 receptor represents a promising molecular target for therapeutic intervention in epilepsy.

In recent years, the development of nanotechnology has provided innovative solutions for drug delivery. Nanomicelles, self-assembled nanocarriers, offer excellent stability, high drug-loading capacity, and tunable drug release properties [12, 13]. Compared to traditional drug delivery systems (DDS), nanomicelles can be chemically modified to achieve disease microenvironment-specific responsive release, significantly enhancing therapeutic targeting and safety [14]. Oxidative stress is one of the key contributors to the pathogenesis of epilepsy, in which excessive reactive oxygen species (ROS) damage neurons and glial cells, aggravating neuronal dysfunction and inflammatory responses [15, 16]. In epilepsy treatment, lesion sites are often characterized by elevated levels of oxidative stress and abnormal neuronal electrical activity, presenting a compelling scenario for designing responsive nanocarrier systems [17]. Another key pathological feature of epilepsy is abnormal neuronal electrical activity, which not only serves as a direct manifestation of seizures but also acts as a major trigger for inflammatory responses [18, 19]. Therefore, dual-functional DDSs responsive to both ROS and electrical stimulation hold great promise for epilepsy treatment. By targeting these two critical pathological mechanisms simultaneously, such systems can enable precise treatment of epileptic foci while minimizing off-target effects on healthy tissues [20, 21].

This study designed A ROS/electric dual-responsive TPGS-Ferrocene-Poloxamer nanomicelle (TFP@A) to leverage its ROS and electrical stimulation-responsive properties for the targeted delivery of the P2 × 7 receptor antagonist A 438,079, aiming to explore its potential in epilepsy treatment. We systematically evaluated the physicochemical properties, blood-brain barrier (BBB) permeability, and brain-targeting delivery efficiency of TFP@A. In vitro and in vivo experiments validated its comprehensive therapeutic effects, including antioxidant, anti-inflammatory, and anti-epileptic activities. The study aimed to confirm the regulatory effects of TFP@A on epilepsy seizures, elucidate its inhibitory action on the P2 × 7 receptor-NLRP3 inflammasome pathway, and investigate its role in improving neuronal function and reducing neuroinflammation. This research not only offers a novel strategy for the precise treatment of epilepsy but also lays a foundation for developing drug delivery technologies targeting neuroinflammation. The findings hold promise for advancing the clinical translation of nanotechnology in treating neurological disorders and providing new therapeutic options for patients with refractory epilepsy.

Materials and methods

Ethical statement

C57BL/6J wild-type male mice (8 weeks old) and neonatal mice were purchased from Hunan SJA Laboratory Animal Co., Ltd. (Hunan, China). The mice were housed under specific pathogen-free (SPF) conditions in standard cages at a controlled temperature of 23 °C, humidity of 55%, and a 12-hour light/dark cycle. All animal experiments were approved by the Animal Ethics Committee of our institute and conducted in compliance with the guidelines of the International Association for the Study of Pain (IASP) Ethical Committee.

Synthesis and characterization of TFP@A

To synthesize vitamin E polyethylene glycol succinate–ferrocene conjugate (TPGS-Fc), 3.49 g of TPGS (57668, Merck KGaA) was dissolved in 40 mL of dichloromethane (DCM, 270997, Merck KGaA) in a clean flask under stirring. Subsequently, 0.576 g of EDCI (E7750, Merck KGaA) and 0.415 g of 4-dimethylaminopyridine (DMAP, 107700, Merck KGaA) were added. After 10 min of stirring, 0.506 g of ferrocene carboxylic acid (Fc-COOH, 46264, Merck KGaA), pre-dissolved in 10 mL of DCM, was introduced into the mixture. The reaction was maintained under continuous stirring at 2000 g and room temperature for 24 h. Upon completion, DCM was removed by rotary evaporation, and the residue was dissolved in 3 mL of anhydrous ethanol. The resulting solution was transferred into a dialysis bag (MWCO: 500–1000 Da) and dialyzed against 200 mL of anhydrous ethanol for 48 h. Ethanol was then removed by rotary evaporation, and the product was dried in an oven to yield TPGS-Fc. Successful conjugation was confirmed by Fourier-transform infrared (FT-IR) spectroscopy.

To prepare blank micelles (TFP), 50 mg of TPGS-Fc and 200 mg of poloxamer 407 (16758, Merck KGaA) were dissolved in 5 mL of methanol by sonication. The clear solution was rotary evaporated at 50 °C under 375 kPa to form a uniform thin film on the flask wall, followed by overnight drying in a fume hood. The film was then hydrated with purified water under stirring at 300 g and 50 °C for 1 h and cooled to room temperature. The final solution was filtered through a 0.22 μm microporous membrane to obtain TFP micelles.

For drug-loaded micelles (TFP@A), 10 mg of the P2 × 7 receptor antagonist A 438,079 (HY-15488, MedChemExpress, Shanghai, China) was co-dissolved with 50 mg of TPGS-Fc and 200 mg of poloxamer 407, and the preparation steps were carried out as described above.

Micelle morphology was characterized by transmission electron microscopy (TEM, H-7650, Hitachi, Japan), while particle size distribution and zeta potential were analyzed by dynamic light scattering (DLS, Nano-ZS90, Malvern, UK).

In vitro drug release

After preparing the drug-loaded or probe-loaded micelles, 3 mL of the micelle solution was placed into a beaker equipped with two platinum electrodes spaced 1 cm apart. A voltage of 2 mV was applied with the electrode distance set to 1 cm. Additionally, H₂O₂ was used to simulate ROS [22], with 50 µmol/L of H₂O₂ applied to stimulate the nanomicelles. Following stimulation, 1 mL of the micelle solution was taken for DLS analysis to measure the average particle size and micelle distribution, while another 1 mL was placed into a dialysis bag to evaluate in vitro release. The dialysis bag was immersed in a centrifuge tube containing 10 mL of phosphate-buffered saline (PBS) (pH 7.4) and incubated at 37 °C in a constant temperature shaker at 200 g for 30 min or 1 h. At 3, 15, 30, 60, 120 and 180 min, 1 mL of external PBS was replaced with fresh PBS. The concentration of A 438,079 was determined using high-performance liquid chromatography (HPLC) (e2695, WATERS) with the following parameters: Agilent TC-C18 column (4.6 mm × 250 mm, 5 μm); mobile phase: methanol and 0.1 mol/L sodium dihydrogen phosphate solution (55:45); column temperature: 50 °C; flow rate: 1.0 mL/min; detection wavelength: 230 nm; injection volume: 10 µL.

In vitro safety evaluation

Hemocompatibility test: Red blood cells were isolated from C57BL/6J mice, washed three times with normal saline, and diluted to obtain a 2% suspension. The suspension was incubated at 37 °C for 30 min with various concentrations of TFP@A (3.125–100 µM), normal saline (negative control), or purified water (positive control). Samples were then centrifuged at 3000 g for 15 min, and the absorbance of the supernatant was measured at 415 nm to quantify hemoglobin release. Hemolysis percentage was calculated relative to the positive control (set as 100%).

Primary microglia isolation and identification: Primary microglial cells were isolated from the cerebral cortex of three neonatal mice. Brain tissues were enzymatically digested with trypsin, filtered, and suspended in DMEM supplemented with 10% fetal bovine serum (FBS; 10099158, Gibco, China) and 1% penicillin-streptomycin (C0222, Beyotime). Cells were seeded in poly-L-lysine–coated flasks (P1524, Merck KGaA, Germany) and cultured for 12–14 days. Microglia were separated by mild shaking and centrifugation (200 g, 1 h), and the supernatant was collected and replated in poly-D-lysine–coated flasks. CD68 immunofluorescence staining was used to confirm microglial identity in cultures aged 14–21 days.

Cell Culture and Drug Treatment: Mouse hippocampal neurons (HT22, CL-0697) and bEnd.3 cells (CL-0598) were obtained from Procell (Wuhan, China) and cultured in DMEM with 10% FBS and 1% penicillin-streptomycin. In drug treatment groups, neurons and microglia were incubated with serum-free DMEM containing A 438,079, TFP, or TFP@A (equivalent to 5 µM A 438079) for 72 h at 37 °C in a humidified incubator (5% CO₂). High-quality culture reagents and sterile plasticware were used throughout to ensure reproducibility.

Cytotoxicity and Apoptosis Assays: Cell viability was assessed using the CCK-8 assay after 48-hour treatment with gradient concentrations of TFP or TFP@A. For apoptosis analysis, neurons and microglia were seeded at 15% confluence in six-well plates and treated with Control, TFP, or TFP@A for 24 h. Cells were harvested, washed with PBS, and stained with Annexin V and propidium iodide for 10 min, followed by flow cytometry analysis to evaluate apoptosis.

In vitro uptake assay and BBB permeability assays

Cellular Uptake Assay: HT22 or primary microglial cells were seeded in six-well plates and incubated with TFP-FITC or TFP@A-FITC (1 mL) for 24 h once the cell density reached ~ 75%; control wells received no treatment. After incubation, cells were washed three times with PBS and stained with rhodamine working solution (02558, Merck KGaA, Germany) at 37 °C for 30 min to label the cytoplasm. Following additional PBS washes, cells were fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.1% Triton X-100 for 10 min. Nuclei were counterstained with DAPI (1 µg/mL) for 10 min at room temperature. After a final PBS wash, cells were trypsinized, centrifuged (1500 g, 5 min), and subjected to confocal laser scanning microscopy (CLSM, Olympus, Tokyo, Japan) or flow cytometry (Beckman, USA) for uptake analysis.

Preparation of FITC-Labeled Micelles: TFP@A-FITC and TFP-FITC were prepared by mixing 0.4 mL of FITC (2 mg/mL) and 0.4 mL of A 438,079 (20 mg/mL) in a 4 mL centrifuge tube, followed by incubation at 37 °C for 1 h. The resulting mixture was centrifuged at 3000 g for 1.5 h, and the collected product was freeze-dried to obtain labeled micelle powders.

bEnd.3 Uptake with Transferrin Pretreatment: To assess targeted uptake, bEnd.3 cells were preincubated with transferrin (10 µg/mL, 10652202001, Merck KGaA) for 30 min. Excess transferrin was removed by washing twice with serum-free medium prior to FITC-micelle exposure to avoid interference.

Transwell co-culture and BBB penetration assays

A Transwell system (0.4 μm pore size; Merck, Germany) was employed to model BBB penetration. bEnd.3 cells were cultured in the upper chamber and primary neurons in the lower chamber. After 15 days of bEnd.3 culture to allow monolayer formation, TFP-FITC or TFP@A-FITC (400 µg/mL) was added to the upper chamber, while HT22 cells in the lower chamber were stimulated with glutamate (50 mmol/L, G0355000, Merck KGaA). Fluorescence intensity in the lower chamber was measured at multiple time points using a spectrophotometer to calculate nanomicelle permeability.

Model Validation and Transport Kinetics: Transepithelial electrical resistance (TEER) and the apparent permeability coefficient (P_app) of FITC-labeled dextran (4 kDa) were measured to validate BBB model integrity. TEER was calculated using the formula: TEER = (R − R₀) × S. where RRR is the total resistance, R₀ is the resistance without cells, and SSS is the membrane area of the Transwell insert (0.33 cm²). Resistance values were measured using the ERS-2 resistance system (Millipore, Billerica, MA, USA). The content of A 438,079 was determined using HPLC. The apparent permeability coefficient P_app of A 438,079 was calculated with the formula: P_app = (dQ/dt)/(C₀ × S). Where dQ is the permeation rate (nmol/s), C₀ is the initial concentration in the upper chamber (nmol/mL), and SSS is the membrane area of the Transwell insert (0.33 cm²). Samples collected at 60 min were analyzed using DLS for particle size measurement and TEM for morphological observation.

Intracellular ROS detection and responsiveness

HT22 cells were stimulated with high-concentration Glu (50 mmol/L) for 24–48 h, followed by treatment with nanomicelles (400 µg/mL) for 24 h. Intracellular ROS levels were assessed using with 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA, 10 µmol/L, D6883, Merck KGaA, Germany) and fluorescence intensity was visualized by fluorescence microscopy (Olympus) or quantified by flow cytometry (Beckman) using the FITC channel.

For the ROS responsiveness assay, HT22 neurons were preincubated with glutathione (GSH) (100 µmol/L, Y0000517, Merck KGaA, Germany) to scavenge intracellular and extracellular ROS. Subsequently, the above-mentioned permeability experiment was repeated to assess the effect of GSH treatment on the release rate of A 438,079.

ATP test of P-gp

bEnd.3 cells were seeded in six-well plates at a density of 8 × 10⁵ cells per well and cultured overnight in DMEM containing 10% fetal bovine serum and 1% penicillin/streptomycin. TFP@A was diluted in DMEM containing 10% FBS and 1% penicillin/streptomycin to a final concentration of 10 µg⋅mL^− 1, and 1 mL was added to each well. After 3 h of incubation, 2 µg⋅mL^− 1 free FITC, TFP@A, or TFP@A + FITC was added to the wells and incubated for 30 min. Following incubation, the cells were washed twice with PBS and collected. Cell lysis buffer was added, and the lysate was centrifuged to collect the supernatant. ATP concentration was then measured using the Enhanced ATP Assay Kit, and the results were recorded with a microplate reader (Varioskan LUX, ThermoFisher Scientific).

Construction of 3D neuronal spheres and evaluation of nanomicelle penetration ability

To evaluate the BBB penetration ability of TFP@A, a three-dimensional neuronal sphere model was constructed using Matrigel. HT22 neurons were suspended in a high-glucose medium containing 2% Matrigel at a density of 1 × 10⁶ cells/mL. The cell suspension was dispensed into a 96-well low-adhesion culture plate, 100 µL per well, and incubated at 37 °C with 5% CO₂ for 48 h to form 3D neuronal spheres. TFP@A was covalently labeled with FITC to a final concentration of 400 µg/mL. The labeled TFP@A-FITC nanomicelle was added to the neuronal sphere medium and incubated at 37 °C for 4 h. After incubation, the neuronal spheres were fixed with 4% PFA for 30 min and permeabilized with 0.1% Triton X-100 for 10 min. DAPI was used to stain the nuclei for positional reference. Fluorescence images were captured using a CLSM with appropriate laser wavelengths and filters to detect the fluorescence signals of FITC and DAPI. The distribution of fluorescence intensity was analyzed using ImageJ software to evaluate the penetration depth and uniformity of the nanomicelles within the neuronal spheres.

In vivo biodistribution of TFP@A in mice

Synthesis of TFP/TFP@A-Cy5.5: A total of 2.5 mL of SBF solution was mixed with 0.5 mL of CaCl₂ solution dropwise, followed by the addition of 0.4 mL of TFP/TFP@A (20 mg⋅mL^− 1 and 0.4 mL of Cy5.5 (3 mg⋅mL^− 1. The pH was adjusted to 7.4 using Tris-HCl buffer, and the mixture was allowed to react at 37 °C for 1 h alongside a 0.1 OD control group. The final TFP@A-Cy5.5 product was collected by centrifugation at 3000 g for 1.5 h at 4 °C. The TFP@A-Cy5.5 was then freeze-dried using a lyophilizer. In vivo Study: Mice were randomly divided into three groups (n = 3 per group), and the hair on their backs was removed to prevent interference. Each group received a tail vein injection of 0.8 mg/kg of Cy5.5, TFP-Cy5.5, or TFP@A-Cy5.5. The animals were then examined using an in vivo imaging system (IVIS, MS Lumina XRMS, Revvity), with visualization performed using the red fluorescence of Cy5.5. Brain tissues were harvested 4 h post-injection for imaging to assess the biodistribution of TFP@A-Cy5.5.

Pharmacokinetics of TFP@A

Mice were administered A 438079 (5 mg/kg or TFP loaded with A 438079 (TFP@A) via tail vein injection. Following isoflurane anesthesia, blood samples were collected from the orbital sinus into heparinized centrifuge tubes (1.5 mL capacity). The samples were centrifuged at 4 °C (3000 g, 10 min), and the supernatant was transferred to clean centrifuge tubes. Methanol with isocyanate was added to extract A 438079 from the plasma. The mixture was centrifuged at 4 °C (14000 g, 15 min), and the supernatant was collected for analysis. Additionally, free A 438079 was separated using ultrafiltration centrifuge tubes with a molecular weight cut-off of 10 kDa. Plasma samples were added to the ultrafiltration tubes and centrifuged at 4 °C (12000 g, 30 min). The filtrate was collected to determine the concentration of free A 438079. The supernatant was analyzed using HPLC (e2695, WATERS). Furthermore, A 438079 concentrations in tissue samples from different brain regions, as well as the heart, liver, spleen, lungs, and kidneys, were quantified.

In vivo safety evaluation of TFP@A

To assess the safety of TFP@A, histological analysis of major organs, including the heart, liver, spleen, lungs, kidneys, and brain, was performed using hematoxylin and eosin (H&E) staining. Briefly, tissue samples from multiple organs were collected from mice and fixed in 4% PFA for 24 h. The tissues were then embedded in paraffin, sectioned, and stained with H&E (G1120, Solarbio, Beijing, China). Histological changes were observed under an inverted microscope. Serum levels of alanine aminotransferase (ALT, JN20465), aspartate aminotransferase (AST, JN20681), blood urea nitrogen (BUN, JN55961), and creatinine (CREA, JN7806) were measured using assay kits purchased from Shanghai Jining Industrial Co., Ltd. (Shanghai, China), following the manufacturer’s protocols.

Construction of epilepsy model

Pentylenetetrazol (PTZ)-Induced Acute Epilepsy Model: To monitor electroencephalographic (EEG) activity, mice were anesthetized with isoflurane and fixed in a stereotaxic apparatus for electrode implantation. Two miniature screws were inserted over the sensorimotor cortex for recording and one over the cerebellum as a ground, all secured with dental cement. After a 7-day recovery, EEG recordings were performed using the LabChart 8 system (ADInstruments). Mice were randomly assigned to four groups: tail vein injection of PBS (control), 5 mg/kg A 438079, untreated TFP micelles (TFP), or TFP micelles loaded with A 438079 (TFP@A, equivalent to 5 mg/kg. Two hours after injection, acute seizures were induced by intraperitoneal injection of 70 mg/kg PTZ (P6500, Sigma-Aldrich, Merck KGaA, Germany), and EEG recordings were conducted 30 min later. Seizure severity was scored using a modified Racine scale [23]: Stage 0, no response; Stage 1, immobility; Stage 2, rigidity; Stage 3, head nodding and circling; Stage 4, intermittent jerks; Stage 5, continuous jerking and clonic seizures; Stage 6, tonic-clonic seizures and rapid jumping. Mice that died during the experiment were assigned to Stage 6. EEG recordings were performed continuously for 3 days, 2 h each day.

Pilocarpine (Pilo)-Induced Epilepsy Model: Mice were anesthetized with isoflurane and secured in a stereotaxic frame. Bipolar electrodes were implanted into the right hippocampal CA3 region for EEG recording. An additional screw was used as a ground electrode, and two other screws were used to secure the assembly to the skull. After a 7-day recovery period, mice were randomly divided into four groups: tail vein injection of PBS, 5 mg/kg A 438,079, TFP micelles, or TFP@A. Thirty minutes after injection, scopolamine (1 mg/kg, S-099, Supelco, Merck KGaA, Germany) was administered intraperitoneally to reduce peripheral nervous responses. Sixty minutes later, Pilo (300 mg/kg, HY-B0726A, MCE, Shanghai, China) was injected intraperitoneally to induce status epilepticus (SE). Behavior was observed for 60 min post-injection, and EEG recordings were taken. Behavioral assessments were performed using the modified Racine scale as described above.

Kainic Acid (KA)-Induced Epilepsy Model: Mice were anesthetized with isoflurane and fixed in a stereotaxic frame. KA (0.5 mg/mL, 0.5 µL, HY-N2309, MCE, Shanghai, China) was injected into the dorsal hippocampal CA1 region (AP: −2.0 mm; ML: −1.3 mm; V: −1.6 mm). After a 2-month recovery period, bipolar electrodes were implanted into the hippocampal CA3 region (AP: −2.9 mm; ML: −3.1 mm; V: −3.1 mm) for EEG monitoring. After a 7-day recovery period, EEG recordings of freely moving mice were conducted for 8 h daily using the LabChart system. On the first day, saline was injected into the mice, and baseline EEG was recorded. Mice exhibiting exploratory epilepsy were selected for further experiments. On the second day, the selected mice received a tail vein injection of 5 mg/kg TFP@A or free A 438,079, followed by 8 h of EEG recording. On the third day, the mice received an intraperitoneal injection of saline, and post-baseline EEG was recorded. Generalized seizures (GS) were defined as regular spiking clusters with a duration ≥ 30 s, amplitude at least three times the baseline, and a baseline EEG peak frequency ≥ 2 Hz.

Ex vivo Fluorescence Imaging Model for Kindling: Mice were anesthetized with isoflurane and fixed in a stereotaxic frame before surgery. Two bipolar electrodes were implanted bilaterally into the CA3 region of the hippocampus. The bipolar electrode in the right CA3 hippocampus was used for stimulation to establish the hippocampal kindling model, while the electrode in the left CA3 was used to minimize surgical interference. A screw was placed over the sensorimotor cortex for EEG recording, and another screw over the cerebellum served as a ground electrode. After a 7-day recovery period, a stimulation sequence of 400 µA, 20 Hz, 1 ms pulse width was applied as a continuous train lasting 2 s. Mice were stimulated once every 30 min, 10 times per day until they reached Racine stages 1, 3, or 5. At each stage (1, 3, or 5), mice were administered a tail vein injection of 5 mg/kg TFP@A-Cy5.5. One hour after injection, epilepsy was induced, and the mice were euthanized to extract their brains. Whole brains and 1 mm thick coronal slices were imaged using an IVIS Spectrum IVIS (IVIS Lumina Series III). The samples were further sectioned into 25 μm thick cryostat slices. After washing the slices twice with PBS, the slides were dried and stained with DAPI for nuclear visualization.

A 438,079-Resistant Epilepsy Model: Mice were intraperitoneally injected with 300 mg/kg Pilo and intravenously administered 20 mg/k A 438,079 daily for 14 days to induce resistance to A 438,079. After successfully establishing the A 438,079-resistant epilepsy model, the animals were randomly divided into groups and given a single tail vein injection of the following: PBS, 5 mg/kg free A 438,079, or TFP@A. One hour later, Pilo (300 mg/kg) was injected intraperitoneally into all groups. To prevent peripheral cholinergic side effects, methyl scopolamine (1 mg/kg) was administered intraperitoneally 30 min before Pilo injection. The drug responses were evaluated using Racine scores during the 90 min following Pilo injection, and EEG recordings were monitored for 2 h daily over 3 consecutive days. If sustained seizures lasted for 90 min, the mice were treated with diazepam (10 mg/kg, D0940000, Merck KGaA, Germany) to terminate the seizures. Control group mice received saline injections instead of Pilo at all three stages, while other treatments remained identical to the experimental groups.

EEG recording

EEG data were collected from various epilepsy models as previously described. Briefly, mice were anesthetized and fixed in a stereotaxic apparatus, where bipolar electrodes were implanted in the CA3 region (AP: −2.9 mm; ML: −3.1 mm; V: −3.1 mm) for EEG recording. After a 7-day recovery period, baseline EEG was recorded for 30 min prior to epilepsy induction. Following the administration of PTZ, KA, or Pilo, EEG activity was continuously monitored for 3 days, 2 h each day. Random 10-minute EEG segments were recorded, and brain electrical power was analyzed across all groups. According to previous studies, epileptic seizure events were defined as high-amplitude discharges with peak frequencies exceeding 3 Hz and amplitudes three times the baseline.

Biochemical analysis

Hippocampal tissues were collected from mice following cardiac perfusion with saline and homogenized. The hippocampal homogenates were analyzed using the following assay kits: Caspase-1 Activity Assay Kit (C1101, Beyotime, Shanghai, China), IL-1β ELISA Kit (PI301, Beyotime), IL-10 ELISA Kit (PI522, Beyotime), Total Glutathione Assay Kit (S0052, Beyotime), Total Superoxide Dismutase (SOD) Assay Kit (S0109, Beyotime), and Lipid Peroxidation malondialdehyde (MDA) Assay Kit (S0131S, Beyotime).

Western blot analysis of protein expression

Total proteins were extracted from cells or tissues using RIPA buffer (25 mM Tris, pH 7.5, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X-100). Equal amounts of protein (20 µg) were separated by SDS-PAGE and transferred to nitrocellulose membranes. Membranes were blocked with 5% non-fat milk for 1 h and incubated overnight at 4 °C with primary antibodies (Table S1). After washing, membranes were incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. Protein bands were visualized using an enhanced chemiluminescence kit (WP20005, Thermo Fisher, USA) and imaged with the ChemiDoc XRS Plus system (Bio-Rad). Band intensities were quantified using ImageJ software, with Tubulin as the loading control. All experiments were independently repeated three times.

RT-qPCR for relative gene expression

Total RNA was extracted using Trizol reagent (15596026, Invitrogen, USA) according to the manufacturer’s instructions; RNA was reverse-transcribed into cDNA using the PrimeScript RT reagent Kit (RR047A, Takara, Japan) following the protocol; the synthesized cDNA was subjected to RT-qPCR using the Fast SYBR Green PCR Kit (11736059, Thermo Fisher Scientific (China) Co., Ltd., Shanghai, China), with each reaction performed in triplicate; GAPDH was used as the internal reference; relative gene expression was calculated using the 2^−ΔΔCt method; the experiment was repeated three times; the primer sequences used in this study are listed in Table S2, and all primers were synthesized by Takara.

Immunofluorescence staining

For cultured cells, samples were fixed with 4% paraformaldehyde (PFA) at room temperature for 15 min, followed by blocking with 3% bovine serum albumin (BSA) at 37 °C for 30 min. After overnight incubation with primary antibodies, cells were washed and incubated with secondary antibodies for 2 h at room temperature. Nuclei were counterstained with DAPI.

For tissue sections, mice were anesthetized and perfused with 4% PFA. Brain tissues were fixed, dehydrated in 30% sucrose, and embedded in OCT (45345, Merck KGaA, Darmstadt, Germany). Coronal cryosections (30 μm) were prepared and incubated with primary antibodies overnight, followed by secondary antibodies and DAPI staining. All stained samples were imaged using a fluorescence microscope (Olympus, Japan), and ImageJ software was used for quantification.

A list of primary and secondary antibodies used is provided in Table S3.

Nissl staining

To assess neuronal survival, brain tissue sections (10 μm thick) were subjected to Nissl staining. Briefly, sections were stained at room temperature for 10 min using Nissl staining solution (G1436, Solarbio, Beijing, China). After dehydration, the sections were mounted with neutral balsam and observed under a light microscope (Leica, Germany). Representative images were captured, and the number of Nissl-positive cells in the hippocampal CA3 subregion was quantified using ImageJ software.

Timm staining

Brain coronal cryosections (10 μm thick) were subjected to Timm staining with 0.01 mol/L AgNO₃ solution. The Timm index was calculated as follows: Timm Index = Area of Timm Granules / Length of DG. All slides were observed using a scanning fluorescence microscope (Olympus).

ATP level measurement

Tissue homogenates of hippocampal samples were prepared by centrifugation at 4000 g for 10 min. ATP levels were determined using a commercial assay kit (A095-1-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Briefly, the reaction mixture, including 30 µL of brain tissue extract, 100 µL of substrate buffer I, 200 µL of substrate buffer II, and 30 µL of promoting buffer, was incubated at 37 °C for 30 min. Absorbance was then measured at 636 nm. ATP concentration was calculated based on a standard ATP solution.

Transcriptome sequencing

Hippocampal tissues were collected from three mice in each group (Normal, Pilo, and Pilo + TFP@A) for transcriptome sequencing analysis. Total RNA was extracted using TRIzol reagent (Cat. No. 15596026, ThermoFisher, USA), and the purity and concentration of the extracted RNA were assessed using a NanoDrop 2000 UV-Vis spectrophotometer (ThermoFisher, USA). RNA was reverse-transcribed into cDNA following the instructions of the PrimeScript RT reagent Kit (RR047A, Takara, Japan) for transcriptome sequencing. Differential expression analysis was conducted using the “limma” package in R. Genes with|log₂(FoldChange)| > 1, and significance p < 0.05 were identified as differentially expressed genes (DEGs).

Statistical methods

Statistical analyses were conducted using R version 4.2.1 with the integrated development environment RStudio and GraphPad Prism 8 (GraphPad Software, Inc.). All data were tested for normality and homogeneity of variance. Normally distributed data were expressed as mean ± standard deviation. Comparisons between two groups were performed using an independent sample ttt-test, while comparisons among multiple groups were analyzed using one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test. For repeated measurements at multiple time points, data were analyzed using repeated-measures ANOVA. A p < 0.05 was considered statistically significant.

Results

Synthesis and characterization of TFP@A loaded with a P2 × 7 receptor antagonist

Considering the vicious interplay between aberrant neuronal circuits and neuroglial inflammation, as well as the altered microenvironmental substance transport observed during epilepsy progression [17], we developed a nanomicelle-based DDS loaded with the P2 × 7 receptor antagonist A 438,079. The term “nanomicelle” is used here to emphasize the engineered nature of our system, which is designed for stimulus-responsive targeted delivery. This system aims to target epileptic plaques and restore the pathological brain microenvironment associated with seizures by countering excitotoxicity, inhibiting oxidative stress, and alleviating inflammatory responses.

To construct a micelle-based ROS/electric dual-responsive DDS (Fig. 1A), amphiphilic TPGS was covalently conjugated with carboxylated ferrocene (Fc) via esterification to form TPGS-Fc, and successful conjugation was verified by FT-IR spectroscopy (Figure S1A). Subsequently, TPGS-Fc was co-assembled with the amphiphilic block copolymer Poloxamer 407 to generate TFP-mixed micelles. TEM revealed that the micelles were spherical with an approximate diameter of 30 nm (Figure S1B). DLS analysis of the hydrodynamic size showed that the size of TFP mixed micelles decreased compared to TPGS-Fc micelles (Figure S1C). The TEM image of TFP micelles (Figure S1B) and the monomodal hydrodynamic diameter distribution (Figure S1D) demonstrated their nanoscale and uniform structure. Furthermore, the surface charge of TFP mixed micelles was lower compared to TPGS-Fc micelles (Figure S1E), confirming the successful preparation of TFP micelles.

Fig. 1

Degradation and drug release of TFP@A under ROS/electrical stimulation. Note: (A) Schematic illustration of ROS-responsive and drug release experiments of TFP@A in PBS (pH = 7.4); (B) Size distribution and PDI of TFP@A after incubation with 50 µmol/L H₂O₂; (C) Peak shift in DLS particle size distribution of TFP@A over time under H₂O₂ stimulation; (D) Drug release profile (A438079) of TFP@A in PBS (pH = 7.4) with and without H₂O₂; (E) H₂O₂ scavenging efficiency of TFP@A; (F) Schematic of TFP@A structural changes under electrical stimulation at 200 mV/m; (G) Size distribution and PDI of nanomicelles after different durations of 200 mV/m electrical stimulation; (H) Cumulative drug release rate of nanomicelles calculated for A438079. Experiments were performed in triplicate. ***p < 0.001

To investigate the ROS responsiveness and drug release characteristics of the nanomicelles, we used H₂O₂ to simulate ROS and stimulate the nanomicelles. As reported, the extracellular H₂O₂ concentration at epileptic foci can reach up to 50 µmol/L, so we selected 50 µmol/L H₂O₂ to stimulate the nanomicelles [24]. Our findings showed that the DLS particle size and polydispersity index (PDI) of the nanomicelles increased over time (Fig. 1B), likely due to changes in hydrophobicity that prevented the nanomicelles from maintaining their original structure. Additionally, we observed that the DLS size distribution of the nanomicelles became increasingly disordered over time, eventually showing small and large size peaks outside the main peak (Fig. 1C). We propose that the small size peaks represent nanomicelle fragments degraded by ROS, while the large size peaks are aggregates formed from these fragments.

Drug-loaded micelles were prepared by mixing 50 mg of TPGS-Fc, 200 mg of Poloxamer 407, and 10 mg of the P2 × 7 receptor antagonist A 438,079 (Figure S1F) to form TFP@A. The release of A 438,079 from TFP@A was analyzed using HPLC. We observed that under ROS stimulation, the nanomicelles released an increasing amount of A 438,079 over time, whereas in the absence of ROS, the release of the P2 × 7 receptor antagonist was negligible. After 48 h of ROS stimulation, A 438,079 release reached nearly 100% (Fig. 1D). Additionally, the residual H₂O₂ concentration in the solution was measured, and it was found that the nanomicelles significantly accelerated the clearance of H₂O₂ compared to PBS at pH 7.4, confirming the ROS-scavenging capability of the nanomicelles (Fig. 1E). Thus, the designed nanomicelles possess both ROS-responsive drug release capability and ROS-scavenging activity. While the initial depletion of H₂O₂ within 4 h is attributed to the ROS-scavenging properties of the nanomicelles, the sustained release of A 438,079 afterward is due to the structural transformation of the nanomicelles triggered by the early ROS exposure. These structural changes persist even after ROS levels decline, thereby continuing to promote drug release.

Next, we applied an electric field to the TFP@A nanomicelles to investigate their electric responsiveness (Fig. 1F). As reported, the electric field intensity at epileptic foci during seizures can reach 0.3 V/m, so we set the voltage to 2 mV with an electrode distance of 1 cm [25]. The results showed that the DLS particle size and PDI of the TFP@A nanomicelles increased over time under the electric field (Fig. 1G). Furthermore, after removing the electric field and leaving the nanomicelles overnight, both the particle size and PDI returned to their initial levels. This suggests that the disassembly induced by the electric stimulus is a reversible process, allowing the nanomicelles to recover. Additionally, we observed that the particle size and PDI increased with higher electric field intensity, indicating that the nanomicelles undergo electro-sensitive swelling (Figure S1G). We further combined electric and ROS stimulation to investigate the drug release from the nanomicelles. We observed that, compared to ROS stimulation alone, simultaneous electric and ROS stimulation promoted drug release, with more P2 × 7 receptor antagonists being released over time (Fig. 1H). This trend became more pronounced as the stimulation time increased. Additionally, we noted that at the same time point, the release ratio of the P2 × 7 receptor antagonist was highly consistent. These findings suggest that our nanomicelles possess notable electric responsiveness.

In summary, the results demonstrate that TFP micelles exhibit excellent responsiveness and controlled drug release capabilities under both ROS and electric stimulation.

Cellular evaluation of TFP@A

To assess the cellular safety of TFP@A, the CCK-8 assay was conducted to evaluate the biocompatibility of the nanomicelle in neurons and primary microglial cells. Isolated microglial cells were characterized using anti-CD68 staining (Figure S2A). For the CCK-8 assay, neurons and microglial cells were incubated with varying concentrations of A438079 or TFP@A loaded with different concentrations of A438079 for 48 h before the assay. When the concentration of A438079 was below 10 µg/mL, the cell viability of both neurons and microglial cells exceeded 80% (Figure S2B). Encapsulation of A438079 in nanomicelles allowed cell viability to remain above 50% even at A438079 concentrations as high as 600 µg/mL (Figure S2C). These results demonstrate that encapsulating A438079 within nanomicelles significantly reduces its cytotoxicity, enabling higher drug dosage while maintaining cellular safety.

To investigate the endocytosis of TFP@A by neurons and microglial cells, FITC-labeled TFP@A was prepared and incubated with these cells. Cellular fluorescence changes were monitored using CLSM. Incubation with TFP@A significantly increased FITC fluorescence signals in neurons and microglial cells compared to the Control group (Figure S2D-F and Figure S2H). Additionally, flow cytometry was used to quantitatively analyze the uptake of TFP@A by neurons and microglial cells (Figure S2G and Figure S2I). The results showed enhanced uptake of TFP@A in both neurons and microglial cells compared to the Control group.

Furthermore, we validated the dual efficacy of the drug-loaded nanomicelle in suppressing abnormal neuronal discharge and scavenging ROS. High concentrations of Glu are known to induce excitotoxicity in neurons and are commonly used in vitro to model excitotoxic damage to neuronal membranes [26, 27]. In our study, neurons were cultured in the lower compartment of Transwell inserts and subjected to excitotoxicity induced by 50 mmol/L Glu. Meanwhile, the upper compartment housed a fully established monolayer model of bEnd.3 cells (Fig. 2A), simulating the pathological microenvironment of the brain during epilepsy. TFP@A or free A438079 was added to the upper compartment, and the total concentration of A438079 in the lower compartment was measured to calculate permeability coefficients. Compared to normal neurons, excitotoxic neurons significantly enhanced the permeability of TFP@A (Fig. 2B). This suggests that brain injury may also increase the targeting efficiency of nanomicelles during epileptic seizures. Moreover, TFP@A demonstrated a markedly faster transmembrane permeation rate compared to free A438079 (Fig. 2C). The concentration of extracellular free A438079 was subsequently measured, revealing that TFP@A accelerated A438079 release under excitotoxic conditions. Pre-incubation with GSH, an endogenous antioxidant capable of scavenging intracellular and extracellular ROS, significantly inhibited this release, indicating that the nanomicelle exhibits ROS-responsiveness (Fig. 2D).

Fig. 2

Dual functionality of TFP@A in drug efflux and ROS scavenging. Note: (A) Schematic illustration of the experimental design in the Transwell system, demonstrating the construction of the bEnd.3 model and co-culture with neurons in vitro; (B) Changes in permeability coefficients of TFP@A under different conditions (Glu⁻, Glu⁺); (C) Permeability coefficients of A438079 and TFP@A under various treatment conditions; (D) Effect of Glu⁺ and GSH pretreatment on A438079 release; (E) DCFH-DA probe staining images of neurons under Glu stimulation, detecting intracellular ROS levels. Scale bar: 25 μm; (F) Flow cytometry analysis of DCFH-DA fluorescence intensity at different time points, illustrating changes in ROS levels; (G) Detection of HT22 cell apoptosis under various conditions using Annexin V/PI double staining, showing flow cytometry results and corresponding bar graphs of apoptosis rates. Cell experiments were performed in triplicate. *p < 0.05; ***p < 0.001

The ROS levels were further measured using the DCFH-DA probe. After 24 h of Glu stimulation, a significant increase in ROS production was observed (Fig. 2E). When HT22 cells were incubated with TFP@A under excitotoxic conditions for 24 h, TFP@A markedly reduced cellular ROS levels, demonstrating its efficiency in scavenging ROS. This trend was further confirmed by flow cytometry analysis (Fig. 2F). Considering that excitotoxicity induces cellular damage, an apoptosis assay kit was used to evaluate HT22 cell viability. Following 48 h of Glu stimulation, the apoptosis rate of HT22 cells significantly increased, whereas treatment with TFP@A notably reduced the apoptosis rate (Fig. 2G). In conclusion, TFP@A shows promising potential in treating epilepsy and related neurological disorders. Its ROS-responsive drug release and protective effects against oxidative stress highlight its suitability as a therapeutic carrier.

Evaluation of TFP@A BBB permeability and brain distribution

The BBB is a major obstacle limiting drug accumulation in epilepsy lesions. To evaluate the BBB permeability of TFP@A micelles, we first conducted in vitro studies using bEnd.3 cells to examine cellular uptake and BBB penetration. bEnd.3 cells were incubated with FITC-labeled TFP@A, and cellular uptake was assessed. As shown in Figure S3A, the mean fluorescence intensity (MFI) in bEnd.3 cells was significantly higher in the FITC-labeled TFP@A group compared to the free FITC group. To further investigate the BBB penetration ability of TFP@A micelles, we constructed a bEnd.3 monolayer model on Transwell inserts to simulate the BBB. The TEER of the monolayer was continuously monitored. After 15 days, the average TEER reached 200 Ω·cm², indicating the formation of a complete bEnd.3 monolayer suitable for simulating the BBB in vitro. Importantly, treatment with TFP@A did not significantly alter the TEER values of the bEnd.3 monolayer before or after the experiment (Figure S3B). This demonstrates that the integrity of the monolayer was well-maintained throughout the experiments, validating its suitability for in vitro BBB simulation.

FITC-labeled TFP@A was introduced into the upper compartment of the Transwell inserts. After 60 min, samples were collected from the lower compartment for TEM observation (Figure S3C) and DLS measurements (Figure S3D). TEM analysis confirmed that TFP@A nanomicelles maintained their intact structure after crossing the BBB. Additionally, fluorescence intensity within the lower chamber was recorded using fluorescence spectroscopy at different incubation times. The results showed a progressive increase in fluorescence after 1 and 4 h of incubation with TFP@A (Figure S3E), further validating the successful translocation of the nanomicelles across the BBB.

Receptors expressed on the BBB are known to mediate drug transport [28]. This study investigated the classical transferrin-mediated pathway for BBB crossing. bEnd.3 cells were pretreated with 10 µg/mL transferrin and subsequently incubated with FITC-labeled TFP@A (Fig. 3A). As shown in Figs. 3B-C, transferrin pretreatment reduced the fluorescence intensity of TFP@A within cells at 1 and 4 h. Notably, the fluorescence intensity in the TFP@A group was higher than in the transferrin-pretreated TFP@A group, indicating that TFP@A participates in the transferrin-mediated endocytosis pathway. In contrast, no significant difference was observed in the fluorescence intensity between transferrin-pretreated and untreated groups incubated with free FITC, suggesting a different uptake mechanism for FITC. These results suggest that TFP micelles are likely internalized by endothelial cells via transferrin-mediated endocytosis, thereby enhancing their BBB penetration efficiency. Additionally, three-dimensional neuronal spheroids were established using a matrix system to monitor the BBB penetration ability of TFP@A. Remarkably, in the TFP@A-treated group, fluorescence signals were distributed uniformly throughout the neuronal spheroids, whereas in the free FITC group, fluorescence was confined to the outer layer with no detectable FITC deposition in the inner regions. This observation underscores the enhanced BBB penetration ability of TFP@A (Fig. 3D).

Fig. 3

BBB penetration, in vivo distribution, and brain targeting of nanomicelles. Note: (A) Schematic illustration of the Transwell model for transferrin receptor-mediated transcytosis; (B) Workflow of TFP@A penetration experiments with or without endothelial cell pretreatment; (C) Effect of transferrin pretreatment on TFP@A fluorescence intensity, recording fluorescence signals at 1 and 4 h; (D) Fluorescence distribution of FITC and FITC-labeled TFP@A in 3D neuronal spheroids, including bright-field images and fluorescence images at different depths (50 μm), with quantitative analysis of fluorescence intensity. Scale bar: 10 μm; (E) Effect of TFP@A pretreatment on endothelial ATP consumption induced by FITC; (F) In vivo IVIS imaging of mice injected with Cy5.5 or Cy5.5-modified nanomicelles at different time points, with semi-quantitative analysis of Cy5.5 fluorescence signals in the head using IVIS software; (G) IVIS spectral imaging of the brain 4 h post-injection of Cy5.5 or Cy5.5-loaded nanomicelles, along with semi-quantitative analysis of Cy5.5 fluorescence in the brain using IVIS software; (H) Quantification of A438079 distribution in hippocampal tissues across groups at 4 h post-injection. *p < 0.05 for comparisons between groups. Cell experiments were performed at least three times, and animal experiments involved n = 3

Another factor contributing to insufficient brain delivery of AEDs is the P-gp efflux pump on the BBB, which actively transports penetrated drugs back into the bloodstream. It has been reported that TPGS-Fc may enhance drug accumulation by blocking the ATP-binding site of ATPase in P-gp [29]. To validate the inhibitory effect of TPGS-Fc on the efflux pump, we measured changes in intracellular ATP levels in endothelial cells. bEnd.3 cells were pretreated with TFP@A micelles for 3 h, and after removing the micelles, free FITC was added to the upper chamber to measure intracellular ATP levels (Fig. 3B). As shown in Fig. 3E, FITC treatment alone significantly decreased ATP levels in the absence of TFP@A. In contrast, treatment with TFP@A alone had no effect on ATP consumption. Interestingly, cells treated with both TFP@A micelles and FITC showed no changes in ATP concentration, indicating that TFP micelles may block the binding site between ATP and ATPase, thereby inhibiting the P-gp efflux pump. These findings suggest that TFP@A micelles enhance BBB permeability, a crucial feature for improving drug delivery in the treatment of CNS disorders.

We further evaluated the in vivo biodistribution of the nanomicelle. Prior to administration, a suspension of mouse red blood cells was prepared to assess the blood compatibility of the nanomicelle. Biomaterials are typically classified into three categories based on their hemolysis index: non-hemolytic (2%), mildly hemolytic (2–5%), and hemolytic (5%). We observed that even at a TFP@A concentration of 1 mg/mL, the hemolysis rate remained below 5% (Figure S4A), indicating good blood compatibility and suitability for intravenous administration. To investigate the pharmacokinetics, mice were intravenously injected with either free A438079 or TFP@A (both equivalent to 5 mg/kg of A438079), and the total A438079 content in plasma was measured at various time points. Compared to the elimination half-life (T1/2(β)) of free A438079, TFP@A significantly prolonged the plasma half-life (Figure S4B-C). This demonstrates the superiority of our TFP@A drug-loading strategy, which reduces the blood clearance rate, extends circulation time, and increases the likelihood of crossing the BBB. Additionally, we measured the concentration of unbound A438079 in plasma over time. At the same dose, TFP@A resulted in lower free A438079 levels in plasma within the first 4 h after administration compared to the free A438079 group (Figure S4D). These findings suggest that TFP@A effectively reduces unnecessary free A438079 in the bloodstream, potentially preventing unwanted pharmacological activity of A438079 in the peripheral circulatory system.

Considering the long emission wavelength and excellent tissue penetration of near-infrared (NIR) fluorescence probes, we prepared nanomicelles physically encapsulating the classical NIR fluorescence probe Cy5.5. After intravenous injection of Cy5.5-labeled nanomicelles, all mice survived, and their distribution in vivo was monitored through imaging at different time intervals. Analysis of the average fluorescence intensity in the brain revealed that TFP@A exhibited stronger brain accumulation compared to the free probe (Fig. 3F). The highest brain fluorescence intensity was observed at the 4-hour mark, making this the optimal time point for sampling (Fig. 3F). At the 4-hour time point, the mice were perfused to eliminate interference from blood-borne fluorescence probes, and the brains were extracted and imaged directly. The results showed that the brain fluorescence intensity of the TFP@A group was significantly higher than that of the TFP group and the free probe group (Fig. 3G). Although Cy5.5-labeled nanomicelles effectively demonstrated the brain-targeting capability of TFP@A, they could not fully replicate the brain distribution of drug-loaded nanomicelles. To address this, the total amount of A438079 in different brain regions was quantified at the 4-hour time point using HPLC. Our analysis revealed no significant differences in TFP@A accumulation across the cortex, striatum, midbrain, and cerebellum. However, TFP@A exhibited markedly higher accumulation in all brain regions compared to free A438079, particularly in the hippocampus, a region highly sensitive to epilepsy (Fig. 3H and Figure S4E). Furthermore, at the 4-hour time point, we assessed TFP@A distribution in major organs, including the heart, lungs, liver, spleen, and kidneys, and found similar accumulation levels to those of free A438079 (Figure S4F). These results demonstrate the enhanced capacity of TFP@A to improve the distribution and accumulation of A438079 in brain lesions, particularly in the hippocampus, the most epilepsy-sensitive region. Additionally, in vivo, safety evaluations showed no significant pathological changes in organ tissues from any group based on H&E staining (Figure S5A). Serum biomarkers for liver and kidney function, including ALT, AST, BUN, and CREA, also showed no abnormalities across the groups (Figure S5B). Therefore, our nanomicelle formulation demonstrates sufficient safety in vivo.

Taken together, these findings suggest that drug-loaded nanomicelles can achieve the same concentration of A438079 in the brain as free drugs while reducing the required systemic dosage, thus minimizing peripheral side effects.

In vivo evaluation of TFP@A’s antiepileptic efficacy

The P2 × 7 receptor, an ATP-sensitive cation channel, has been shown to be effective in various epilepsy models [30, 31]. To validate the drug-loading capacity and therapeutic efficacy of TFP@A, we established acute, sustained, and spontaneous epilepsy models under TFP@A administration. An acute epilepsy model was first induced in mice via intraperitoneal injection of 70 mg/kg PTZ. The experimental design is outlined in Fig. 4A. TFP-mediated delivery of A438079 significantly reduced the frequency and duration of seizures, particularly in GS (Fig. 4B-D). Furthermore, TFP@A treatment markedly decreased seizure scores in epileptic mice (Figure S6A), although it did not significantly affect the latency to seizure onset in the PTZ-induced acute epilepsy model (Figure S6B). Importantly, all mice in the acute epilepsy model treated with TFP@A survived, providing further evidence of its antiepileptic efficacy and confirming its non-toxicity in vivo (Figure S6C). In addition to behavioral scoring, EEG analysis was performed to dynamically monitor epileptic activity using commonly employed metrics such as power (calculated by applying a fast Fourier transform to 10-minute segments of raw EEG signals) and spikes (transient, isolated bursts of high-amplitude activity occurring at less than one event per second). Representative EEG traces are shown in Figure S6D. Further analysis revealed no statistically significant differences in baseline power; however, TFP@A treatment significantly reduced both total power and the number of epileptic spikes, particularly on the first day of EEG recording (Fig. 4E). In summary, these findings demonstrate that TFP-mediated delivery of A438079 effectively exerts antiepileptic effects in the acute epilepsy mouse model.

Fig. 4

Antiepileptic effects of TFP@A-mediated A438079 delivery in various epilepsy mouse models. Note: (A) Experimental workflow illustrating the establishment of an acute epilepsy mouse model via intraperitoneal injection of PTZ, including baseline recording, drug administration, and seizure behavior monitoring; (B) Comparison of cumulative focal seizures (FS) and GS within 30 min post-treatment across different groups; (C) Heatmap of seizure scores representing frequency and severity of epilepsy in each group; (D) Effect of TFP@A on the number of seizures and seizure duration within 30 min; (E) Analysis of EEG parameters, including daily power, baseline power, total power, and spike frequency per 10 min; (F) Workflow of the Pilo-induced SE model, including behavioral and EEG analyses; (G) Comparison of SE and GS latencies across groups; (H) Incidence rates of GS in each group; (I) Survival rate analysis among groups; (J) Workflow of the KA-induced chronic epilepsy model, showing the formation of spontaneous epilepsy and the drug injection scheme; (K) Schematic representation of KA injection sites; (L) Comparison of GS frequency and duration before and after KA + TFP@A treatment; (M) Changes in GS frequency pre- and post-drug treatment in each group. Statistical significance levels are indicated as *p < 0.05, **p < 0.01, and ***p < 0.001. For survival analysis, n = 10 mice per group; for all other experiments, n = 6 mice per group

To investigate whether the enhanced antiepileptic efficacy of TFP@A is associated with electro-responsiveness in drug release, we utilized a hippocampal kindling model that mimics complex partial seizures [32]. Normal mice and epileptic mice with varying seizure severities (Figure S7A) were administered TFP@A-Cy5.5, and fluorescence intensities in brain slices were recorded. Cy5.5, a fluorescence dye with aggregation-caused quenching properties, emits fluorescence in a dispersed state but is quenched when aggregated, serving as an indicator of drug release. As shown in Figures S7B-C, ex vivo fluorescence imaging revealed stronger fluorescence intensity on the right side (epileptic focus) compared to the left side (non-epileptic hemisphere). In contrast, the normal group showed minimal bilateral differences, indicating that TFP micelles preferentially release Cy5.5 in epileptic foci while remaining largely inactive in non-epileptic regions. Notably, brains from mice with higher seizure stages exhibited greater fluorescence intensity. Representative fluorescence images of brain slices from stages 1, 3, and 5 are displayed in Figure S7D. Strong red signals were observed in the right-side slices, consistent with the ex vivo imaging results. Analysis of bilateral fluorescence intensity and power spectrum differences across different seizure stages further confirmed that TFP micelles release drug molecules in response to epilepsy-like discharges.

SE is a state of continuous epileptic seizures without recovery of consciousness [33]. As shown in Fig. 4F, the SE model was established through systemic injection of Pilo. Behavioral studies and subsequent EEG recordings were conducted to compare the therapeutic effects of free A438079 and TFP@A. Figure 4G demonstrates that TFP@A significantly prolonged the latency to SE onset and GS, indicating a preventive effect of TFP@A on epilepsy susceptibility. Furthermore, TFP micelle-mediated delivery of A438079 reduced the incidence of GS and improved survival rates (Fig. 4H-I). These findings were consistent with the power spectrum analysis of representative EEG recordings (Figure S7E). Collectively, these results highlight the effective delivery and electro-responsiveness of A438079 facilitated by TFP micelles.

To comprehensively assess the antiepileptic effects of TFP@A in temporal lobe epilepsy, a chronic recurrent spontaneous epilepsy model was established using the excitotoxic Glu analog KA. Following hippocampal injection of KA, mice entered a stable spontaneous epilepsy phase after 60 days (Fig. 4J-K). In the TFP@A-treated group, the number and duration of GS were significantly reduced by the second day of treatment, with antiepileptic effects lasting until the third day (Fig. 4L). In contrast, free A438079 at the same dosage showed minimal suppression of epileptic seizures (Fig. 4M). Representative EEG recordings from the hippocampus revealed that TFP@A treatment attenuated the severity of GS, whereas the free A438079 group showed no significant differences before, during, or after treatment (Figure S7F). These findings confirm that electro-responsive TFP micelles are well-suited for delivering A438079 to effectively suppress chronic and spontaneous epilepsy.

TFP@A improves drug resistance in a pilo-induced epilepsy model

Drug resistance significantly contributes to increased mortality and disability rates [34], severely impacting the quality of life of epilepsy patients. To investigate whether TFP@A can overcome drug resistance in epilepsy, a Pilo-induced A438079-resistant epilepsy model was established. This was achieved by administering 20 mg/kg A438079 daily for 14 consecutive days, following previously described protocols [35]. The experimental design is illustrated in Fig. 5A.

Fig. 5

TFP-mediated A438079 delivery reverses epilepsy drug resistance. Note: (A) Experimental timeline; (B) Seizure scores across different groups; (C) Cumulative seizure counts for FS and GS in each group; (D) Heatmap displaying individual seizure counts for each mouse in different groups; (E) Comparison of seizure latency, number of seizures within 90 min, and seizure duration across groups (n = 13); (F) Representative EEG traces from each group; (G) Analysis of EEG-related metrics, including daily power, baseline power, total power, and spike frequency per 10 min. Statistical significance levels are indicated as *p < 0.05, **p < 0.01, and ***p < 0.001. For survival experiments, n = 10 mice per group; for all other experiments, n = 6 mice per group

In summary, TFP@A treatment significantly reduced seizure scores in the A438079-resistant Pilo-induced epilepsy mouse model (Fig. 5B), particularly in GS (Fig. 5C). TFP@A administration also markedly decreased the frequency of GS (Fig. 5C), increased seizure latency, and reduced both the number of seizures (Fig. 5D) and their duration (Fig. 5E). Importantly, TFP@A effectively shortened seizure durations without affecting mouse body weight or survival (Figure S8). EEG analysis further demonstrated that TFP@A treatment significantly lowered daily power, total power, and the number of epileptic spikes per 10-minute interval, particularly on the first day of treatment, while baseline power remained unchanged (Fig. 5F-G). Collectively, these findings indicate that TFP@A is effective in overcoming A438079 resistance in refractory epilepsy.

Protective effects of TFP@A on neuronal damage and oxidative stress in epilepsy mouse models

While previous findings demonstrated that TFP@A maintains sufficient concentrations of A438079 in the brain for an extended period, we hypothesize that its therapeutic efficacy is also related to its pharmacological actions. TFP@A not only mitigates abnormal discharges but also releases A438079 to scavenge ROS, thereby suppressing oxidative stress and preventing neuronal damage caused by excitotoxicity (Fig. 6A). To further explore these protective effects, we conducted a detailed investigation of neuronal changes in epileptic lesions following treatment with TFP@A.

Fig. 6

Effects of TFP@A on neuronal damage, glial response, and oxidative stress in the epileptic microenvironment. Note: (A) Schematic illustration of TFP@A’s mechanism, highlighting its anti-oxidative stress properties and electrical responsiveness for A438079 release to mitigate neuronal damage induced by epilepsy; (B-C) Nissl staining of the CA3 region in the hippocampus across different treatment groups, showing neuronal damage following epilepsy (scale bar: 100 μm); (D-E) Timm staining of the DG region in the hippocampus, comparing mossy fiber distribution and mossy fiber sprouting (MFS) across groups (scale bar: 25 μm, 100 μm); (F-G) NeuN immunofluorescence staining showing changes in NeuN-positive cells, a neuronal-specific marker, in the hippocampus across treatment groups (scale bar: 50 μm); (H) Quantitative analysis of MDA, SOD activity, and GSH concentration in hippocampal tissues across groups; (I) ATP levels in hippocampal tissues following epilepsy and different treatments. Statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001. N = 6 per group

We observed that Cy5.5-labeled TFP@A effectively colocalized with neurons (Figure S9A). Nissl staining was performed on frozen brain tissue sections, as Nissl bodies in normal neurons are easily stained blue by toluidine blue. Given the heightened sensitivity and vulnerability of pyramidal neurons during SE, we focused on staining the CA3 region (Fig. 6B-C). After SE onset, extensive neuronal degeneration and loss were observed in the CA3 region, making staining with Nissl solution difficult. Treatment with free A438079 showed some improvement, but vacuolar degeneration persisted, and neuronal arrangement remained irregular. In contrast, TFP@A treatment restored Nissl body morphology to nearly normal levels, with neurons showing more regular alignment and greater structural integrity. Additionally, in normal brains, the axons of granule neurons in the dentate gyrus (DG)—known as mossy fibers—project exclusively toward the CA3 region in a fixed direction. When neurons are damaged by epilepsy, surviving neurons attempt to rebuild synapses, and regenerated mossy fibers often project in disordered or even reversed directions. Since mossy fibers are rich in zinc, which can be detected by Timm sulfide silver staining, we performed Timm staining on mossy fibers in the DG region of the hippocampus. Following SE onset, a significant increase in mossy fibers was observed within the DG region and an abnormal presence of mossy fibers outside the DG area. Treatment with TFP@A reduced the presence of mossy fibers within the DG region (Fig. 6D-E). Additionally, NeuN, a neuron-specific biomarker, was analyzed in Pilo-induced epilepsy mice using immunofluorescence. Nissl staining revealed that TFP@A treatment significantly increased the percentage of NeuN-positive cells in the hippocampus, indicating improved neuronal survival and integrity (Fig. 6F-G).

We further evaluated oxidative stress and ATP levels in the epileptic microenvironment. Hippocampal tissues were extracted from each group of mice to measure common oxidative stress indicators, including SOD, GSH, and MDA. Following acute seizures, SOD activity and GSH concentrations in the hippocampus were significantly reduced, while MDA levels were elevated, indicating the occurrence of oxidative stress in the lesion tissue (Fig. 6H). Compared to free A438079, TFP@A treatment significantly alleviated oxidative stress. Additionally, we observed that ATP levels in the brain tissues of epileptic mice decreased following TFP@A treatment (Fig. 6I).

In summary, TFP@A effectively mitigates neuronal damage and microenvironmental dysregulation in epilepsy mouse models by scavenging oxidative stress and protecting neurons.

TFP@A promotes an anti-inflammatory microglial phenotype in epileptic mice

The purinergic P2 × 7 receptor (P2 × 7R), an ATP-sensitive ion channel, is highly expressed in microglia and plays a critical role in microglial activation, proliferation, cytokine release, and the progression of epilepsy [36]. Following epileptic seizures, neuronal damage is induced by abnormal discharges and excitotoxicity on one hand, and on the other, damaged neurons release inflammatory factors that activate glial cells. This activation leads to gliosis, oxidative stress, chronic inflammation, and further excitotoxicity, forming a pathological feedback loop that exacerbates the microenvironment and perpetuates neuronal damage (Fig. 7A). To investigate microglial activation, we used Iba-1 as a marker for microglia and performed immunofluorescence staining on brain tissue sections. After SE onset, a significant number of activated microglia were observed in the hippocampus. However, TFP@A treatment effectively suppressed microglial activation compared to the Pilo or free A438079 groups (Fig. 7B). We observed colocalization of Cy5.5-labeled TFP@A with microglia (Figure S9B). In the hippocampal tissues of Pilo-induced epilepsy mice, P2 × 7R expression was significantly increased in the Pilo group compared to the Normal group, whereas the colocalized expression of P2 × 7R and Iba-1 was significantly reduced in the Pilo + TFP@A group (Figure S9C-E).

Fig. 7

Effects of TFP@A on microglial activation and inflammatory phenotypes in Pilo-induced epilepsy mice. Note: (A) Schematic illustration of neuronal damage and microglial activation following epilepsy; (B) Iba-1 immunofluorescence staining showing microglial activation in the hippocampus across treatment groups (scale bar: 25 μm); (C) Immunofluorescence staining of CD86 (pro-inflammatory) and CD206 (anti-inflammatory) expression in microglia within the hippocampus across groups (scale bar: 25 μm); (D) Western blot analysis of CD86 and CD206 protein expression in the hippocampal tissues of different groups; (E) ELISA quantification of IL-10 (anti-inflammatory) and IL-1β (pro-inflammatory) levels in hippocampal tissues across groups. Statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001. N = 6 per group

Next, we examined the inflammatory phenotype of microglia. Compared to the Normal group, the Pilo group showed increased expression of both anti-inflammatory genes (Mrc1 (Cd206), Arg1) and pro-inflammatory genes (Cd86, Il-6). However, in the Pilo + TFP@A group, anti-inflammatory gene expression was further enhanced, while pro-inflammatory gene expression was significantly reduced compared to the Pilo or A438079 groups (Figure S9F). This indicates that TFP@A facilitates an anti-inflammatory response during SE. To extend these findings, we performed double staining for the microglial marker Iba-1 alongside the anti-inflammatory marker CD206 and the pro-inflammatory marker CD86. In the Pilo group, the number and protein expression levels of both CD86- and CD206-positive cells were significantly increased compared to the Normal group. However, in the Pilo + TFP@A group, CD86-positive cell numbers and protein expression were markedly reduced, whereas CD206-positive cell numbers and protein expression were significantly increased (Fig. 7C-D). These results suggest that TFP@A promotes a more anti-inflammatory microglial phenotype in the epileptic microenvironment.

Further, ELISA analysis revealed that compared to the Normal group, the Pilo group exhibited elevated levels of both the anti-inflammatory cytokine IL-10 and the pro-inflammatory cytokine IL-1β. However, in the Pilo + TFP@A group, hippocampal levels of IL-10 were significantly increased, while IL-1β levels were markedly reduced compared to the Pilo group (Fig. 7E).

These findings indicate that TFP@A effectively suppresses the pro-inflammatory microglial phenotype and enhances anti-inflammatory responses in the Pilo-induced epilepsy mouse model. This mitigates the vicious cycle of neuroinflammation, highlighting its potential as a therapeutic strategy for epilepsy.

TFP@A inhibits inflammatory damage by suppressing NLRP3/caspase-1/IL-1β axis activation

To further elucidate the mechanism underlying the therapeutic effects of TFP@A in epilepsy, transcriptomic sequencing was performed on hippocampal tissues from Normal, Pilo, and Pilo + TFP@A groups (Fig. 8A). Volcano plots revealed significant DEGs among the groups (Figure S10A-B). A Venn diagram identified 85 core genes shared between the Pilo vs. Normal and Pilo + TFP@A vs. Pilo comparisons (Figure S10C). Protein interaction analysis of these 85 genes highlighted Il1b (IL-1β) as a key gene (Fig. 8B). Previous studies have established that the NLRP3-caspase-1-IL-1β axis is a major pathway mediating the transformation of oxidative stress into inflammatory damage in epilepsy [37, 38]. Additionally, ATP-P2 × 7R signaling has been shown to activate the NLRP3 inflammasome and trigger the release of inflammatory molecules [39]. In our study, Pilo + TFP@A treatment reversed the Pilo-induced upregulation of Il1b and Nlrp3 (Figure S10D), suggesting that TFP@A therapy may inhibit the activation of the NLRP3 inflammasome pathway.

Fig. 8

TFP@A inhibits NLRP3 inflammasome activation in the Pilo-induced epilepsy model. Note: (A) Schematic workflow of transcriptome sequencing for hippocampal tissues from Normal, Pilo, and Pilo + TFP@A groups (N = 3); (B) Top 10 significant genes from protein-protein interaction analysis of DEGs between Pilo vs. Normal and Pilo + TFP@A vs. Pilo groups; (C-D) Immunofluorescence staining showing colocalization of Iba-1 with NLRP3 and HMGB1 in the hippocampus of each group (scale bar: 25 μm); (E) Western blot analysis of NLRP3 pathway-related proteins, including HMGB1, NLRP3, and IL-1β, in hippocampal tissues across groups; (F) Caspase-1 enzymatic activity in hippocampal tissues of each group. Statistical significance is indicated as **p < 0.01 and ***p < 0.001. N = 6 per group

We further analyzed the colocalization of NLRP3 and HMGB1 with Iba-1, a marker of microglia. As shown in Fig. 8C, the expression of HMGB1/Iba-1 and NLRP3/Iba-1 was significantly increased in the Pilo group compared to the Normal group. Notably, TFP@A treatment markedly suppressed the expression of HMGB1/Iba-1 and NLRP3/Iba-1 in the Pilo group (Fig. 8D). Western blot analysis of proteins related to the HMGB1-NLRP3 pathway revealed that the expression levels of HMGB1, NLRP3, and IL-1β were significantly elevated in the Pilo group compared to the Normal group. However, TFP@A treatment significantly reduced the expression of these proteins compared to the Pilo group (Fig. 8E). Furthermore, we measured caspase-1 enzymatic activity using specific assay kits. Pilo-induced epilepsy markedly increased caspase-1 activity, and while free A438079 showed limited inhibition of inflammatory factors, TFP@A significantly reduced caspase-1 activity and associated inflammatory factors (Fig. 8F).

These results demonstrate that TFP@A effectively inhibits the activation of the NLRP3 inflammasome pathway in Pilo-induced epilepsy, providing a protective anti-inflammatory effect.

Discussion

This study introduces a novel therapeutic approach based on TFP@A, marking the first application of such a design in epilepsy treatment. Unlike conventional DDS, which often target a single mechanism or rely on passive release [40, 41], TFP@A achieves precise drug release in response to multiple stimuli. Traditional epilepsy treatments primarily focus on broadly suppressing neuronal excitability, a strategy that often results in significant side effects and fails to address the underlying causes of seizures. Recent studies on the role of the P2 × 7 receptor in the pathophysiology of epilepsy have highlighted it as a promising target for anti-inflammatory therapy [42]. By incorporating a dual-responsive design, this study successfully achieved combined regulation of neuroinflammation and abnormal neural activity, representing a significant innovation in both therapeutic strategy and technical implementation.

The P2 × 7 receptor, an ATP-gated ion channel, plays a critical role in neuroinflammation [6, 43]. The results of this study demonstrate that targeting the P2 × 7 receptor effectively inhibits downstream activation of the NLRP3 inflammasome and reduces the release of pro-inflammatory factors such as IL-1β. These findings align with previous reports that excessive activation of the P2 × 7 receptor exacerbates neuroinflammation and, through inflammatory crosstalk, intensifies neuronal damage [44, 45]. However, traditional studies have predominantly employed conventional antagonists to directly inhibit the P2 × 7 receptor. The efficacy of such interventions is often hindered by limitations in drug targeting and BBB permeability [46, 47]. This study addresses these challenges by leveraging TFP@A for targeted delivery of A438079. This approach not only overcomes the BBB restrictions but also enhances drug accumulation at the lesion site, significantly improving therapeutic outcomes. Notably, significant differences exist in blood-brain barrier (BBB) permeability between mice and humans. Therefore, future studies will include pharmacokinetic and pharmacodynamic evaluations in human-derived models to determine whether TFP@A can effectively cross the human BBB and achieve therapeutically relevant concentrations in the brain.

The dual ROS- and electrical-stimulation-responsive design of TFP@A enables precise drug release in the specific microenvironment of epilepsy lesions. Compared with previously reported single-responsive or non-responsive nanosystems, the nanomicelle developed in this study demonstrates superior targeting and release efficiency in pathological conditions characterized by elevated ROS levels and abnormal electrical activity. In in vitro models, TFP@A significantly reduced neuronal ROS levels and minimized apoptosis. Moreover, in vivo experiments revealed that TFP@A not only efficiently crossed the BBB but also specifically accumulated in the brain, releasing its payload to reduce the frequency and duration of epileptic seizures. These findings highlight the unique advantages of this system in epilepsy treatment, demonstrating its potential as a highly targeted and effective therapeutic strategy [48,49,50,51,52,53].

Neuroinflammation and oxidative stress are two key pathological processes in epilepsy, with complex interactions between them [54, 55]. Previous studies have often focused on interventions targeting a single mechanism, such as directly scavenging ROS or inhibiting the release of pro-inflammatory factors. However, these approaches frequently encounter limitations due to the complexity of the mechanisms and the diversity of targets involved [56,57,58,59,60]. In this study, a TFP@A was employed to simultaneously target ROS and the P2 × 7-NLRP3 pathway, achieving comprehensive regulation of both neuroinflammation and oxidative stress. Specifically, TFP@A reduced oxidative stress-induced neuronal damage by scavenging ROS while also suppressing P2 × 7 receptor activation to block the release of inflammatory factors. This multi-targeted regulatory strategy effectively disrupted the pathological feedback loop between neurons and microglia, offering a novel perspective for both the mechanistic study and therapeutic management of epilepsy. The potential impact of TFP@A on neuronal plasticity and synaptic remodeling warrants further investigation. Whether TFP@A can modulate cognitive function or memory in epilepsy models by regulating neuroplasticity and synaptic remodeling will be a key focus in future research. We plan to assess changes in neuronal plasticity and synaptic remodeling following TFP@A treatment and their potential effects on cognition and memory in epileptic models.

In vitro experiments demonstrated that TFP@A significantly reduced oxidative stress levels induced by H₂O₂ and enhanced neuronal survival. Cellular uptake studies further revealed that TFP@A is efficiently internalized by both neurons and microglia, thereby improving drug delivery efficiency. Based on reported electric field intensities during epileptic discharges in both animal models and clinical observations, which can reach up to 0.3 V/m, we applied an electric voltage of 2 mV with a 1 cm electrode spacing to simulate these conditions [25]. However, we acknowledge that replicating these conditions in vivo presents challenges. In vivo, TFP@A exhibited excellent BBB penetration, and NIR imaging confirmed its targeted accumulation in the brain. This stands in stark contrast to previous studies in which certain nanosystems faced challenges with low brain-targeting efficiency. Moreover, EEG monitoring showed that TFP@A significantly improved power spectrum parameters associated with epileptic activity, indicating its remarkable efficacy in reducing seizure frequency and suppressing neuronal hyperexcitability.

In this study, we employed multiple epilepsy models to comprehensively evaluate the short-term efficacy and safety of TFP@A. In a PTZ-induced acute seizure model, TFP@A demonstrated significant anti-seizure activity without short-term toxicity (Fig. 4A–E, Figure S6). In the hippocampal kindling model, we observed that the enhanced anti-epileptic effect of TFP@A correlated with electrically responsive drug release (Figure S7A–D). In the status epilepticus (SE) model, TFP@A significantly prolonged the latency to SE and generalized seizures (GS), suggesting a preventive effect against seizure susceptibility (Fig. 4F–I, Figure S7E). Furthermore, in the KA-induced chronic recurrent spontaneous seizure model, electrically responsive TFP micelles effectively delivered A 438,079 to suppress spontaneous seizures (Fig. 4J–M, Figure S7F). To investigate drug resistance, we utilized a pilocarpine (Pilo)-induced A 438,079-resistant epilepsy model, in which TFP@A overcame drug resistance and reduced seizure severity (Fig. 5, Figure S8). Although TFP@A consistently reduced seizure frequency and severity across acute, chronic, and drug-resistant models, the current study primarily provides preliminary insights. Detailed exploration of model-specific efficacy and underlying mechanisms will be addressed in future research.

Safety is a critical factor for the clinical translation of nanomedicine systems. This study verified the excellent biocompatibility of TFP@A through serum biochemical analysis and histopathological evaluation. H&E staining revealed no significant histological abnormalities in major organs, and biochemical markers such as ALT, AST, BUN, and CREA remained within normal ranges. These findings indicate that TFP@A exhibits high short-term safety. Furthermore, compared to traditional AEDs, TFP@A demonstrated significant advantages in reducing non-specific side effects while enhancing therapeutic efficacy. Nevertheless, further investigations are needed to assess its long-term safety and optimize large-scale production processes to facilitate its progression toward clinical application [61].

Despite the significant achievements of this study, several limitations need to be addressed. First, the research primarily relied on an epilepsy mouse model, which, while replicating certain pathological features of epilepsy, differs from human epilepsy in terms of pathophysiology and drug responses. We recognize the importance of assessing the long-term therapeutic potential of TFP@A in chronic epilepsy models. Future studies will incorporate long-term administration protocols to evaluate sustained effects on microglial behavior, neuronal health, and seizure control. This will also determine whether repeated administration maintains efficacy or induces adaptive changes in microglial responses. Secondly, the long-term safety and chronic delivery efficacy of TFP@A nanomicelles remain to be evaluated, particularly the immune responses and potential long-term toxicity following repeated administration, as well as their effects on BBB integrity. Additionally, the complexity of preparing dual-responsive nanocarrier systems may pose challenges for clinical translation, particularly regarding production costs and quality control. Finally, the applicability of this system to other neurological diseases has not been extensively explored, limiting its potential for broader therapeutic applications. Future research should address these limitations to facilitate the clinical development and wider applicability of this innovative technology.

Future investigations could focus on the following aspects: (i) further optimization of the composition and structural design of the nanomicelles to enhance their responsiveness and therapeutic efficacy under various pathological conditions; (ii) conducting large-scale animal studies and comprehensive preclinical evaluations to validate the reproducibility and long-term safety of the therapeutic effects, and to assess their potential value in chronic epilepsy management; (iii) initiating clinical trials to evaluate the translational potential of TFP@A nanomicelles in human epilepsy therapy; (iv) developing combination therapeutic strategies by integrating gene therapy or other small molecule drugs to enhance overall treatment efficacy; (v) exploring the applicability of this technology in other CNS disorders, such as Alzheimer’s disease, Parkinson’s disease, and post-stroke inflammation, thereby broadening its clinical utility. Ultimately, this platform may provide a solid scientific and technological foundation for precision medicine in the treatment of CNS diseases.

Conclusion

This study designed and developed a TFP@A with dual responsiveness to ROS and electrical stimulation, achieving precise delivery of the P2 × 7 receptor in epilepsy treatment. This approach significantly enhanced drug release efficiency and therapeutic outcomes at lesion sites. For the first time, the feasibility of a dual ROS and electrical stimulation-responsive mechanism in epilepsy therapy was valid ated. The study also elucidated the critical role of the P2 × 7-NLRP3 inflammatory pathway in epilepsy pathogenesis, providing new theoretical insights for the combined regulation of neuroinflammation and oxidative stress.

The preclinical experimental results demonstrated that TFP@A significantly reduced the frequency and duration of epileptic seizures, alleviated neuronal damage, and minimized the risk of systemic side effects through targeted drug delivery. This technology offers an innovative and precise therapeutic strategy for epilepsy patients, paving the way for personalized interventions in refractory epilepsy. Additionally, the TFP@A nanomicelle system holds potential for application in other CNS diseases involving inflammation and oxidative stress, such as Alzheimer’s and Parkinson’s diseases, and is anticipated to become a new platform for precision medicine in neurological disorders.