Abstract Seizures induce significant immune and metabolic stress in microglia, but the interaction between these processes remains unclear. This study, utilizing single-nucleus RNA sequencing data from temporal lobe epilepsy (TLE) patients, reveals that reactive oxygen species (ROS) stabilize hypoxia-inducible factor 1-alpha (HIF-1α), thereby inducing glycometabolic reprogramming in microglia and driving the development of a pro-inflammatory phenotype. To address this, a coordination acid-engineered Prussian Blue (PB@ZIF) nanosystem is developed, where Zn²⁺ sites in the zeolitic imidazolate framework (ZIF) lower the local pKa, thereby enhancing the reaction efficiency of PB with free radicals. In vivo experiments using a TLE model demonstrate that PB@ZIF is effectively internalized by microglia and significantly alleviates spontaneous recurrent seizures and seizure-related behaviors. PB@ZIF mitigates microglial inflammatory activation and reduces neuronal injury. Notably, PB@ZIF-induced ROS reduction enhances the enzymatic activity of prolyl hydroxylase domain enzymes, effectively inhibiting HIF-1α-driven glycometabolic reprogramming in microglia. This study identifies a molecular mechanism underlying the immune-metabolic interactions in epilepsy and proposes a promising therapeutic strategy regulating microglial metabolism to improve epilepsy management. Supplementary Information The online version contains supplementary material available at 10.1186/s12951-025-03408-9. Keywords: Temporal lobe epilepsy, Microglia, Metabolic reprogramming, Local pKa, Nanocatalyst Introduction Epilepsy is one of the most common and severe neurological disorders. Recent researches have highlighted the pivotal role of microglia, the resident immune cells of the central nervous system [[40]1], in the pathophysiology of epilepsy [[41]2, [42]3]. During seizures, microglia become activated and release various immune mediators that drive neuroinflammation and neuronal hyperexcitability, thereby contributing to the progression of epileptic disorders [[43]4, [44]5]. To investigate the changes in microglia during epilepsy, we conducted a comprehensive analysis of single-nucleus RNA sequencing (snRNA-seq) data obtained from the NCBI GEO database [[45]6]. Fresh brain samples from individuals with drug-resistant temporal lobe epilepsy (TLE) and age-matched controls (n = 4 per group) were analyzed. Differentially expressed cell-type markers were identified among distinct cell clusters through snRNA-seq analysis (Fig. [46]1a). Microglia were characterized by markers (APBB1IP, CX3CR1, CYBB, DOCK8, PLXDC2, and SFMBT2) (Figure S1), and differentially expressed genes (DEGs) of microglia were enriched. Gene Ontology (GO) enrichment analysis of these DEGs revealed significant microglial activation and neuroinflammatory responses (Fig. [47]1b). However, the key factors driving the immune activation of microglia remain insufficiently explored, hindering the identification of precise therapeutic targets and the development of effective treatments. Fig. 1. [48]Fig. 1 [49]Open in a new tab SnRNA-seq and transcriptome data from TLE patients and mouse models revealed immune and metabolic changes. (a) Unsupervised clustering of cell populations identified 7 distinct cell types in both epileptic patients and controls, with individual cells color-coded according to their respective clusters. (b) GO enrichment analysis of the identified DEGs showing the top 20 most enriched biological processes between epileptic patients and controls. (c) KEGG pathway enrichment analysis of the DEGs between epileptic patients and controls. (d) Schematic illustration of the proposed injury mechanism of epilepsy. (e) Heatmap displaying the differential gene expression between PILO-SE models and control mice. (f) GO enrichment analysis of DEGs identified between PILO-SE models and controls, highlighting the top 30 most significantly enriched pathways relative to controls. NADPH oxidase complex, identified as a major source of ROS during seizures, was among the top enriched pathways. (g) KEGG pathway enrichment analysis of the DEGs between PILO-SE models and controls Seizures induce metabolic stress due to excessive neuronal firing and heightened energy demand [[50]7, [51]8], resulting in elevated glycolytic flux and accelerated glycolysis [[52]9, [53]10]. Microglia may also adapt to these challenges through metabolic reprogramming. Evidence from Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis revealed significant enrichment of DEGs in microglia within pathways such as chemical carcinogenesis-reactive oxygen species (ROS), NOD-like receptor (NLR) signaling, and hypoxia-inducible factor (HIF) signaling (Fig. [54]1c). Among these, HIF-1α stands out as a key regulator of both immune responses and cellular metabolism, particularly under stress conditions. HIF-1α has been shown to promote the transcription of glycolytic enzymes [[55]11, [56]12], while also playing a critical role in activating the NLR protein 3 (NLRP3) inflammasome [[57]13, [58]14]. The stability of HIF-1α is tightly controlled by prolyl hydroxylase domain enzymes (PHDs), which mediate its ubiquitination and subsequent degradation [[59]15, [60]16]. The activity of PHDs depends on Fe (II) at its catalytic site [[61]17, [62]18]. However, ROS can inhibit this function by oxidizing Fe (II) to Fe (III), stabilizing HIF-1α and preventing its degradation [[63]19]. ROS-induced stabilization of HIF-1α is likely to drive glycometabolic reprogramming, linking metabolic shifts to immune responses. Thus, we propose that excessive ROS may impair the activity of PHDs, stabilize HIF-1α, and promote glycometabolic reprogramming in microglia. This process reshapes the metabolic activity and inflammatory phenotype of microglia during seizures, potentially contributing to the progression of epilepsy (Fig. [64]1d). Transcriptome sequencing of hippocampal tissues from the pilocarpine-induced status epilepticus (PILO-SE) model of TLE and control subjects showed consistent results with patient-derived data (Fig. [65]1e-g). Here, we developed a coordination acid-engineered Prussian Blue (PB@ZIF) system designed to be internalized by microglia, regulating glycometabolic reprogramming as a therapeutic strategy for TLE (Scheme [66]1). The Zn²⁺ sites in the Zeolitic Imidazolate Framework-8 (ZIF-8) effectively reduced the local pKa, enhancing the reaction efficiency of PB with ROS. Experimental results demonstrated that PB@ZIF efficiently scavenged ROS, restored microglial homeostasis, and alleviated neuronal damage caused by microglial activation in TLE models. Metabolomic analysis indicated that epilepsy induced a metabolic shift in the hippocampus from oxidative phosphorylation(OXPHOS) to glycolysis. Remarkably, PB@ZIF effectively reversed this glycometabolic reprogramming in microglia by enhancing PHDs’ activity and facilitating the degradation of HIF-1α. This study presents a molecular mechanism and therapeutic strategy for epilepsy treatment, underscoring the potential of regulating microglial metabolism to improve epilepsy management. Scheme 1. [67]Scheme 1 [68]Open in a new tab a. The sequencing data derived from TLE patients and animal models demonstrated pro-inflammatory activation and metabolic alterations, while indicating ROS as a pivotal regulatory target. b. The PB nanoparticles were assembled with ZIF-8 to form PB@ZIF composites. The Zn²⁺ coordination sites in ZIF-8 effectively reduced the local pKa, thereby enhancing the reaction efficiency between PB and ROS. c. Following internalization by microglia, the PB@ZIF catalyst effectively scavenged seizure-induced excessive intracellular ROS. This ROS elimination attenuated Fe (II) oxidation to Fe (III), thereby enhancing the activity of PHDs. The activated PHDs promoted hydroxylation of HIF-1α, which mitigated glycometabolic reprogramming while suppressing neuroinflammatory responses. Collectively, this cascade ameliorated pathological manifestations of TLE through metabolic regulation andanti-inflammatory mechanisms Methods Animals Male C57BL/6n mice, aged 6–8 weeks, were obtained from Charles River Co., Ltd. (Shanghai, China). The mice were housed in a specific-pathogen-free facility under controlled conditions of 24 ± 2 °C temperature and 60% ± 5% relative humidity, with a 12-h light–dark cycle. All experimental procedures were conducted in compliance with guidelines from the National Institutes of Health, and were approved by the Ethics Committee of Zhongshan Hospital. The PILO-SE model of TLE Mice were treated with pilocarpine to induce SE. The induction of SE was achieved via intraperitoneal (i.p.) injection of pilocarpine hydrochloride (300 mg/kg; Sigma-Aldrich). Thirty min prior to pilocarpine administration, mice received an i.p. injection of scopolamine methyl nitrate (5 mg/kg; Sigma-Aldrich) to mitigate peripheral cholinergic effects. Following the onset of SE, which was behaviorally confirmed by continuous seizure activity for at least 30 min, diazepam (10 mg/kg i.p.; Sigma-Aldrich) was administered 2 h later to terminate the seizures. The Control group, which received 0.9% sterile saline instead of pilocarpine, was treated with scopolamine and diazepam at the same doses as the SE-induced groups. Transcriptome sequencing analysis Transcriptome sequencing was performed by OE Biotech Co., Ltd (Shanghai, China). Hippocampal RNA was extracted from four mice per group two weeks after SE induction to construct cDNA libraries and conduct RNA-seq. DEGs were identified based on a significance threshold of p < 0.05 and a fold change ≥ 2 or ≤ 0.5. GO and KEGG pathway enrichment analyses were conducted on the DEGs from each group. Synthesis of PB@ZIF 1.11 g polyvinylpyrrolidone (PVP) was uniformly dispersed in 80 mL H[2]O through ultrasonic treatment. Then, 27.0 mg FeCl[3]·6H[2]O was introduced into the aqueous PVP solution, followed by an additional 10 min of ultrasonication. The resulting mixture was then heated to 60 °C and vigorously stirred at 12,000 rpm for 30 min. Afterwards, 42.2 mg K[4][Fe(CN)[6]] was accurately measured and dissolved in 20 mL H[2]O. After ultrasound for 5 min, this solution was gradually added to the reaction mixture at a controlled rate of 40 mL/h. Following the completion of the addition, stirring was maintained at 60°C and 12,000 rpm for 1 h. Upon completion of the reaction, the mixture was slowly cooled to room temperature. To purify the product, unreacted PVP and ionic impurities were eliminated through a centrifugation process repeated three times, and the final product of PB was dispersed in methanol. A 12 mL solution of PB, at a concentration of 0.25 mg/mL, was incorporated into a 78 mL methanol solution already containing 2.46 g of 2-methylimidazole. Following this, an additional 30 mL methanol solution containing 0.267 g of ZnNO[3]·H[2]O was added to the reaction. The entire mixture was then allowed to stand undisturbed at room temperature for a period of 2 h. To purify the product, any unreacted reactants were removed through a centrifugation process repeated three times, and the resulting material of PB@ZIF was subsequently dispersed in H[2]O. ABTS radical scavenging measurement The ABTS radical was generated by reacting 2,2’-azino-bis(3-ethylbenzthiazoline-6-sulfonate) (ABTS, 7 mM) with a potassium persulfate (K[2]S[2]O[8], 2.45 mM) solution under dark conditions for 16 h. This ABTS radical solution was then diluted with phosphate-buffered saline (PBS) to attain the desired absorbance at 734 nm. Subsequently, PB@ZIF, PB, ZIF, or varying concentrations of Edaravone (0, 5, 10, 15, 20, 25, 30 µM) were introduced to initiate the scavenging reaction, and the absorbance at 734 nm was recorded for each experimental group. DPPH radical scavenging measurement An ethanol solution containing 1,1-Diphenyl-2-picrylhydrazyl radical 2,2-Diphenyl-1-(2,4,6-trinitrophenyl) hydrazyl (DPPH, 0.25 mM) was prepared. Subsequently, PB@ZIF, PB, ZIF, or varying concentrations of Edaravone (0, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50 µM) were introduced to initiate the scavenging reaction, and the absorbance at 517 nm was recorded for each experimental group. Cyclic experiment Post-reaction with ABTS and DPPH, PB@ZIF was isolated through centrifugation, and the recovered material was reused for a subsequent catalytic cycle. This process was repeated for a total of five cycles. In the case of Edaravone, since it cannot be centrifuged, the reaction was directly added to ABTS and DPPH solution to verify the cyclic catalytic effect. Hydroxyl radicals (•OH) scavenging measurement The ·OH was produced by irradiation of TiO[2] (0.1 mg/mL) with ultraviolet light (365 nm). Coumarin and varying concentrations of PB@ZIF, PB, and PB&ZIF were then introduced into the reaction mixture. The fluorescence intensity within the range of 400 nm to 550 nm, excited at 340 nm, was measured to quantify the ·OH scavenging capacity. Importantly, the Fe content was kept consistent across all experimental groups to ensure comparability. NH[3]-TPD measurement A 100 mg sample of ZIF-8 underwent a programmed heating process at a rate of 10℃/min until it reached 250℃. Subsequently, the sample was subjected to a drying pretreatment under a continuous flow of He gas (30–50 mL/min) for a duration of 1 h. Following this, the temperature was reduced to 50℃, and a 10% NH[3]/He gas mixture (30–50 mL/min) was introduced for 1 h, allowing for adsorption saturation. To eliminate weakly adsorbed NH[3] molecules from the surface, the flow was switched to pure He (30–50 mL/min) and maintained for 1 h. Finally, under a He atmosphere, the sample was heated at a rate of 10℃/min up to 500℃ to initiate desorption, enabling the detection of the exfiltrated gas. Cell culture BV2 and HA-1800 cell lines were acquired from ScienCell (CA, USA), and N2a cell line was acquired from Zhong Qiao Xin Zhou Biotechnology (Shanghai, China). BV2 and N2a cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. HA-1800 cells were cultured in Astrocyte Medium (AM) containing 10% fetal bovine serum, 1% penicillin/streptomycin, and 1% astrocyte growth supplement. All cell lines were maintained at 37 °C in a 5% CO[2] atmosphere. Cell viability assay The cytotoxicity of PB and PB@ZIF in BV2 cells was assessed using the Cell Counting Kit-8 (CCK-8) assay (Beyotime). BV2 cells were seeded in 96-well plates at a density of 5000 cells per well and incubated for 24 h to allow adherence. The medium was then replaced with 100 µL of fresh complete DMEM medium containing varying concentrations of PB or PB@ZIF, and cells were incubated for an additional 24 h. Cell viability was determined following the CCK-8 assay protocol. Intracellular localization BV2 cells were inoculated into 6-well plates, stained with Hoechst 33,342 (blue) and Lyso Tracker (green), and then incubated with complete DMEM containing Cy5.5-labeled PB@ZIF (20 µg/mL, magenta) for 0, 1, 2, and 4 h, respectively. Finally, the cells were detected on a FV3000 confocal laser scanning microscope (Evident). Cell uptake BV2, HA-1800, and N2a cells were respectively inoculated in 6-well plates and stained by Hoechst 33,342 (blue), and then incubated with complete DMEM or AM containing Cy5.5-labeled PB@ZIF (20 µg/mL, magenta) for 4 h. The uptake of PB@ZIF was subsequently examined using an FV3000 confocal laser scanning microscope (Evident). Evaluation of intracellular ROS BV2 cells were divided into 5 experimental groups: (1) Control: Cells were cultured in complete DMEM medium with equal amount of vehicle control. (2) LPS: Cells were treated with lipopolysaccharide (LPS) at a concentration of 5 µg/mL and equal amount of vehicle control. (3) ZIF: Cells were treated with 5 µg/mL LPS and 6.4 µg/mL ZIF (The quantification of ZIF was calculated based on its content within PB@ZIF). (4) PB: Cells were treated with 5 µg/mL LPS and 20 µg/mL PB. (5) PB@ZIF: Cells were treated with 5 µg/mL LPS and 20 µg/mL of PB@ZIF. Intracellular ROS generation in BV2 cells was evaluated using APF (Maokang Biotechnology) and DCFH-DA (Beyotime) assay kits. Briefly, BV2 cells were seeded in 6-well plates. After cell adherence, BV2 cells were treated with or without 5 µg/mL LPS and different nano materials for 2 h, and then were incubated with the fluorescence probe (APF or DCFH-DA) at 37°C for 20 min. Samples were then assessed by microscope (OLYMPUS) or flow cytometry (Becton Dickinson). Malondialdehyde (MDA) assay The levels of MDA were measured with MDA Content Assay Kit (Solarbio). BV2 cells were cultured in 6-well plates with or without 5 µg/mL LPS and different nano materials for 24 h. Cell samples were then collected, and the assay was conducted according to the manufacturer’s protocol. JC-1 staining Mitochondrial membrane potential (MMP) in BV2 cells was assessed using the JC-1 staining kit (Beyotime). Cells were treated with or without 5 µg/mL LPS and different nanomaterials for 24 h, followed by incubation with JC-1 reagent for 20 min. After washing three times with JC-1 buffer, the nuclei were stained with Hoechst 33,342 (blue), and the cells were visualized and imaged using a FV3000 confocal laser scanning microscope (Evident). ELISA To assess the levels of interleukin-1β (IL-1β), IL-6, and Tumor Necrosis Factor-α (TNF-α) in BV2 cell supernatants via ELISA. BV2 cells were treated with or without 5 µg/mL LPS and different nanomaterials for 24 h. After incubation, supernatants were collected and centrifuged to remove cellular debris. The cytokine concentrations were measured using commercially available ELISA kits (YoBiBiotech), according to the manufacturer’s protocol. After antibody incubation and washing steps, absorbance was read using a microplate reader (Molecular Devices), with cytokine levels quantified by comparison to standard curves. In vivo biodistribution of PB@ZIF To monitor the accumulation of PB@ZIF in vivo, mice were intravenously (i.v.) injected with Cy5.5-labeled PB@ZIF (5 mg/kg). Mice were sacrificed at 24 h after PB@ZIF injection, and the heart, liver, spleen, lung, and kidneys were collected for fluorescence imaging (Vieworks). To dynamically observe the distribution and metabolism of PB@ZIF, alterations in the Cy5.5 fluorescence signal in the brains of the mice were monitored at 0, 1, 2, and 7 days post-injection. Drug administration and animal grouping After the SE induction, the mice were randomly assigned to four treatment groups: 1. Pilocarpine Group (Pilo): Mice in this group received 200 µL of 0.9% sterile saline solution (i.v.) as a vehicle control. 2. ZIF Group (ZIF): Mice were administered ZIF nanoparticles (1.6 mg/kg i.v.) dispersed in 200 µL of 0.9% sterile saline solution. 3. PB Group (PB): Mice were treated with PB nanoparticles (5 mg/kg i.v.) dispersed in 200 µL of 0.9% sterile saline solution. 4. PB@ZIF Group (PB@ZIF): Mice received PB@ZIF nanoparticles (5 mg/kg i.v.) dispersed in 200 µL of 0.9% sterile saline solution. The injections of saline, ZIF, PB, and PB@ZIF were performed via the tail vein 5 h after the onset of SE. The control group, which was administered 0.9% sterile saline instead of pilocarpine, received scopolamine and diazepam at doses identical to those used for the SE-induced groups. Moreover, the saline control treatment protocol was consistent with that implemented in the SE-induced groups. Seizures recording For electrodes implantation during in vivo electroencephalogram (EEG) recording, mice were anesthetized with isoflurane and fixed in a stereotaxic apparatus. Following a midline scalp incision, the skull was exposed, and recording electrodes were implanted over the frontoparietal cortex. Additionally, two electromyographic electrodes were inserted into the neck muscles for simultaneous recording of muscle activity. The electrodes were affixed with dental cement, and the scalp was sutured. Fourteen days post-surgery, continuous EEG recordings were conducted for 3 days using the Medusa EEG recording system (Bio-Signal Technologies). During the recording sessions, mice were allowed to move freely within the test cage, and their locomotor activity was tracked by infrared-sensitive cameras. EEG signals were analyzed to detect electrographic seizure events, characterized by high amplitude waveforms at least twice the baseline and durations of at least 10 s. Besides, behavioral seizures were recorded on video and evaluated using the Racine scale. The Racine scale [[69]20] was as follows: Stage 0: no response; Stage 1: facial clonus (blinking, moving, rhythmic chewing); Stage 2: stage 1 plus rhythmic nod; Stage 3: stage 2 plus forelimb myoclonus without upright hind limbs; Stage 4: stage 3 plus upright hind limbs; Stage 5: generalized tonic, a burst of seizure, and loss of control. Animals that died during the experiments were assigned stage 6. Only seizures that progressed to Stage 4 or 5 were included in the analysis. Y maze The Y maze test was used to evaluate short-term spatial memory in mice, which was linked to hippocampal function. The apparatus comprised three arms arranged at 120° angles. Mice were placed in one arm and allowed to explore freely for 10 min. Spontaneous alternation behavior, defined as successive non-repetitive entries into all three arms (e.g., ABC, BCA), was recorded using EthoVision XT 14 (Noldus). The percentage of alternation was calculated as: (total alternations × 100) / (total arm entries − 2), providing an index of working memory performance. Open field test The locomotor activity and the anxiety-like behavior of the mice were examined using the open field test, which was performed in a chamber with walls. Each mouse was given 10 min to explore the area and was tracked by the video recording system EthoVision XT 14 (Noldus). The distances travelled and time spent in the central area versus the periphery were measured to assess anxiety-like behavior and locomotion. Western blot analysis For protein analysis, mouse hippocampal tissues and cultured cells were lysed in ice-cold RIPA buffer supplemented with protease and phosphatase inhibitors (Sigma-Aldrich). The lysates were incubated on ice for 30 min, followed by centrifugation at 15,000 g for 20 min at 4 °C. The supernatants were collected, and protein concentrations were determined using a BCA Protein Assay Kit (Sigma-Aldrich). Protein samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto polyvinylidene fluoride membranes. The membranes were blocked with 5% non-fat milk in Tris-buffered saline with Tween 20 (TBST) for 2 h at room temperature, followed by overnight incubation at 4 °C with primary antibodies. The following primary antibodies were used in the present study: NLRP3 (1:1000, CST), caspase-1 (1:1000, CST), cleaved caspase-1 (1:1000, CST), IL-1β (1:1000, CST), cleaved IL-1β (1:1000, CST), HIF-1α (1:1000, Abcam), and β-actin (1:2000, CST). After washing with TBST, the membranes were incubated with HRP-conjugated secondary antibodies (Anti-Rabbit IgG (H + L) antibody, 1:5000, CST; Anti-Mouse IgG (H + L) antibody, 1:5000, CST) for 1 h at room temperature. The protein bands were visualized using an imaging system (Tanon) and quantified using ImageJ software (National Institutes of Health). Immunofluorescence staining Two weeks post-SE induction, mice were anesthetized by i.p. injection of sodium pentobarbital and perfused with PBS followed by 4% paraformaldehyde (PFA). The mouse brain samples were dissected and immobilized in 4% PFA overnight, then dehydrated with gradient sucrose solutions at 4 °C. The frozen brain sections, 20 μm thick, were prepared after embedding the brain tissues in an optimal cutting temperature compound. Slices were rinsed three times in PBS and then blocked by PBS containing 3% serum and 0.3% Triton X-100 for 1 h. Then the slices were incubated with primary antibodies overnight. The primary antibody information is as follows: anti-IBA1 (1:500, Servicebio), anti-NeuN (1:200, Abcam), anti-GFAP (1:200, Servicebio), anti-CD16 (1:150, BD Biosciences), and anti-Hif-1α (1:200, Abcam). Slices were washed with PBS three times and incubated with appropriate fluorescence-conjugated secondary antibodies at room temperature for 1 h in the dark. The secondary antibody information is as follows: Alexa Fluor^® 488 conjugated Goat Anti-Mouse IgG (H + L) (1:400, Servicebio), Cy3 conjugated Goat Anti-Mouse IgG (H + L) (1:300, Servicebio), FITC conjugated Goat Anti-Rat IgG (H + L) (1:200, Servicebio), Alexa Fluor^® 594 conjugated Goat Anti-Rabbit IgG (H + L) (1:400, Servicebio), and Alexa Fluor^® 488 conjugated Goat Anti-Rabbit IgG (H + L) (1:400, Servicebio). After that, the slices were rinsed three times with PBS and dried. Nuclei were visualized with 4,6-diamidino-2-phenylindole (DAPI, Servicebio). The slices were captured by a fluorescence microscopy (OLYMPUS). Terminal Deoxynucleotidyl transferase dUTP Nick end labeling (TUNEL) staining TUNEL assay was performed to detect apoptotic cells in brain sections. Brain tissues were sectioned as described above, and the sections were rinsed twice in PBS. The sections were then permeabilized with 1% Triton X-100 at 4 °C for 10 min. After washing in PBS, the sections were incubated with TUNEL reaction mixture according to the manufacturer’s protocol (Servicebio). Apoptotic cells were visualized using fluorescence microscope (OLYMPUS), and the images were analyzed to quantify apoptosis in brain regions. In vivo hematological evaluation of PB@ZIF C57BL/6n mice (6–8 weeks, male) were used to evaluate the toxicity of PB@ZIF in vivo. The mice were randomly divided into four groups: Control (0.9% sterile saline, 200 µL), ZIF (1.6 mg/kg), PB (5 mg/kg), and PB@ZIF (5 mg/kg) groups. All nanoparticle suspensions (ZIF, PB, and PB@ZIF) were prepared in 0.9% sterile saline, with an injection volume standardized to 200 µL per mouse, delivered through the tail vein. Following a two-week treatment period, blood samples were collected from the mice for a comprehensive hematological evaluation. This analysis included the measurement of key hematological parameters such as platelet count (PLT), white blood cell count (WBC), hemoglobin concentration (HGB), mean corpuscular hemoglobin concentration (MCHC), mean corpuscular volume (MCV), hematocrit (HCT), and red blood cell count (RBC). In addition, biochemical analyses were conducted through the assessment of alanine aminotransferase (ALT), aspartate transaminase (AST), creatine kinase MB (CKMB), and creatinine (CREA) levels. Hematoxylin and Eosin (HE) staining After two weeks of treatment with different nanoparticle suspensions and 0.9% sterile saline solution, major organs (brain, heart, liver, spleen, lung, and kidneys) of mice were harvested and fixed overnight at 4 °C in a 4% formaldehyde solution. Following dehydration and vitrification, the organs were embedded in paraffin and sectioned at a thickness of 5 μm. The sections were then deparaffinized using xylene, rehydrated through an ethanol gradient, and subsequently stained with HE. Visualization of the stained sections was performed using a fluorescence microscope (OLYMPUS). Metabolomic analysis Hippocampal tissues of mice were collected 14 days after pilocarpine-induced SE. The samples were weighed to ensure uniform weight across all specimens before the addition of the extraction solvent, consisting of methanol and water in a 4:1 ratio, containing internal standards. The extraction process involved grinding at -10 °C for 6 min, followed by ultrasonic extraction at 5 °C for 30 min. After extraction, the samples were precipitated at -20 °C for 30 min, centrifuged at 13,000 g and 4°C for 15 min. The supernatant was analyzed using LC-MS on a Thermo Fisher UHPLC-Q Exactive HF-X system equipped with an ACQUITY UPLC HSS T3 column (Waters). The analysis was conducted at Majorbio Bio-Pharm Technology (Shanghai, China). Chromatographic separation was performed using water/acetonitrile and acetonitrile/isopropanol as mobile phases. Electrospray ionization in both positive and negative modes was employed, generating total ion chromatograms. Quality control (QC) samples, prepared by pooling equal volumes from all extracts to ensure stability. Data processing was conducted using Progenesis QI v3.0, including baseline filtering, peak identification, retention time alignment, and metabolite identification using databases such as Human Metabolome Database (HMDB), METLIN, and Majorbio. In vitro epilepsy cell culture model An improved in vitro model of epilepsy was developed following previously described methods [[70]21, [71]22]. N2a cells were treated with 100 µM kainic acid (KA) for 1 h, followed by fresh complete DMEM medium replacement and a further 24-h incubation, after which the medium was collected. BV2 cells were then treated with 30% of the conditioned medium (KA/CM) and 70% fresh medium for 24 h to induce microglial activation. For the control group, N2a cells were treated with complete DMEM medium without KA under the same conditions, and the conditioned medium was used similarly for BV2 cells. Luciferase reporter assay To monitor prolyl hydroxylase-dependent degradation of HIF-1α, we constructed a reporter vector based on the pCDNA3.1(+) vector (Thermo Fisher). The HIF-1α’s oxygen-dependent degradation (ODD) was fused in-frame with Firefly Luciferase downstream of the transgene. BV2 cells were seeded in 24-well plates at a density sufficient to reach 50–70% confluency at the time of transfection. Cells were transfected with 1 µg of pCDNA3.1(+)-HIF-1α-ODD-Firefly Luciferase plasmid and 0.1 µg of pGMR-TK-Renilla Luciferase plasmid (Genomeditech) using Lipofectamine™ 3000 (Thermo Fisher) according to the manufacturer’s protocol. Renilla luminescence was used as a calibrator of transfection efficiency. Transfection complexes were prepared in Opti-MEM™, incubated for 15 min, and added to the complete culture DMEM of the cells. After transfection for 24 h, the cells were stimulated for 3 h with the in vitro epilepsy cell culture model as described above. The cells were then lysed with the Dual-Luciferase reporter assay system (Genomeditech). Firefly and Renilla Luciferase activities were sequentially measured using a GloMax^® 20/20 luminometer (Promega). Relative luciferase activity was calculated by normalizing Firefly Luciferase activity to Renilla Luciferase activity. Statistical analysis Statistical analyses were performed using GraphPad Prism 8. Data are presented as mean ± standard error of the mean (SEM). For comparisons among multiple groups, one-way analysis of variance (ANOVA) was used, followed by Tukey’s post hoc test for multiple comparisons. For comparisons between two groups, an unpaired two-tailed Student’s t-test was conducted. All experiments were repeated with at least three technical replicates to ensure reproducibility. Results Synthesis and characterization of PB@ZIF In order to investigate the pivotal role of microglia in epilepsy, we designed and synthesized PB@ZIF nanocatalyst that can be effectively phagocytosed by microglia to scavenge ROS. PB@ZIF consisted of PB as the core and ZIF-8 as the shell. PB was prepared through colloidal chemistry, using PVP as a stabilizing agent [[72]23]. Then, an in-situ self-assembly method was employed to coat PB with a ZIF-8 shell, resulting in the formation of PB@ZIF nanocatalyst (Fig. [73]2a) [[74]24]. Transmission electron microscopy (TEM) image revealed that PB@ZIF exhibited a cubic morphology, with an average size of approximately 70 nm. In high-resolution transmission electron microscopy (HRTEM) image and the selected area electron diffraction (SAED) mode, the exposed lattice spacing of PB in PB@ZIF nanocrystal was about 5 Å, corresponding to the (200) plane, which was consistent with the results of PB (Fig. [75]2b and Figure S4). Elemental mapping and line scanning further confirmed the uniform coverage of the ZIF-8 shell on PB nanoparticles (Fig. [76]2c, Figure S5). Dynamic Light Scattering (DLS) analysis indicated excellent size uniformity and dispersion of the nanocatalyst. Notably, the addition of the ZIF-8 shell increased the hydrated particle size and positively shifted the Zeta potential compared to PB (Figure S6). Powder X-ray Diffraction (XRD) patterns confirmed the crystalline structure of PB@ZIF, corresponding well to the individual components (Fig. [77]2d). Fourier Transform Infrared Spectroscopy (FTIR) analysis revealed a characteristic CN stretching peak at 2090 cm⁻¹, typical of the Fe^2+-CN-Fe^3+ bond in PB (Figure S7a) [[78]25]. Raman spectroscopy further confirmed the presence of PB and ZIF-8 characteristic peaks in PB@ZIF (Figure S7b). The Brunauer-Emmett-Teller (BET) surface area measurements indicated that PB@ZIF fell within the range of PB and ZIF-8, demonstrating the successful synthesis and preservation of the structural integrity of both components (Figure S7c). Collectively, these results validated the successful fabrication of PB@ZIF nanocatalyst. Fig. 2. [79]Fig. 2 [80]Open in a new tab The local Lewis acid sites created by ZIF enhanced the catalytic activity of PB. (a) Schematic illustration of synthesizing PB@ZIF nanoparticles. (b) TEM image and HRTEM image with SAED patterns of PB@ZIF. (c) The elemental mapping of PB@ZIF. (d) XRD patterns of PB@ZIF, ZIF, and PB. (e) •OH scavenging of PB@ZIF, ZIF, and PB@ZIF (n = 3 independent samples). (f) UV-vis spectra of PB@ZIF, ZIF, and PB. (g) NH[3]-TPD of ZIF. (h) The protonation/deprotonation process of ZIF for H[2]O. (i) O-H bond lengths of H[2]O, Zn^2+-OH[2,] and Zn^2+-OH^−. (j) Zn-O and Zn-N bond lengths of Zn^2+-OH[2] and Zn^2+-OH^−. (k) Calculation of pKa changes caused by bond length. (l) Schematic diagram of Prussian blue reacting with free radicals PB@ZIF exhibited excellent free radical scavenging efficacy ABTS and DPPH radicals were used as typical free radicals to evaluate the catalytic scavenging activity of PB@ZIF. Upon adding PB@ZIF, a significant reduction in the characteristic intensities of both ABTS and DPPH (Figure S8a) was observed, confirming its strong free radical scavenging ability. In comparison, PB alone, used as a control, exhibited inferior scavenging performance. In addition, Edaravone, a clinical drug used in neurological disorders for scavenging free radicals [[81]26], was employed as a benchmark. It demonstrated a concentration-dependent scavenging effect on both ABTS and DPPH (Figure S8b) radicals. However, as shown in Figure S8c, Edaravone relied on its reaction with free radicals to function and was difficult to effectively remove after further addition of free radicals, limiting its effectiveness in long-term studies, such as those involving microglia function. In contrast, PB@ZIF retained its catalytic activity with minimal loss over five repeated cycles. As demonstrated in Figure S8d, PB@ZIF was separated via centrifugation after each reaction with ABTS and DPPH, and its scavenging activity remained stable across five cycles. This stability highlighted PB@ZIF’s potential for long-term free radical scavenging applications for extended studies. Given that •OH are among the most potent oxidizing free radicals [[82]27], we further investigated the scavenging ability of PB@ZIF for •OH. These radicals were generated by irradiating TiO[2] with UV light, and the content of •OH in different groups was characterized through fluorescence measurements [[83]28]. When coumarin molecules react with •OH, they yield 7-hydroxycoumarin, which exhibits a fluorescence peak at 460 nm [[84]29]. As anticipated, we observed a concentration-dependent decrease in the fluorescence intensity of the •OH characteristic peak with increasing concentrations of PB@ZIF. Notably, while PB also displayed a similar trend in fluorescence reduction, PB@ZIF demonstrated superior scavenging efficiency at equivalent Fe content (Figure S9). To ascertain whether the interaction between ZIF and •OH enhanced PB’s removal effect, we evaluated a physical mixture of PB and ZIF. However, this mixture showed no significant difference in performance compared to PB alone (Fig. [85]2e). The above experimental results indicated that the higher free radical scavenging efficiency of PB@ZIF than PB depended on the improvement of PB reactive activity in the presence of ZIF. To investigate the enhancement mechanism, we first monitored whether the valence states of different materials were distinct. Ultraviolet-visible absorption spectra (UV-vis) revealed distinct peaks at around 700 nm for both PB@ZIF and PB, indicating intermetallic charge transfer from Fe^2+ to Fe^3+ (Fig. [86]2f) [[87]30]. X-ray photoelectron spectroscopy (XPS) data (Figure S10a) further confirmed this, with the Fe 2p peak in PB@ZIF showing decomposition into three distinct peaks corresponding to Fe (II), Fe (III), and their satellite peaks. A higher content of Fe (II) compared to Fe (III) in both PB@ZIF and PB was attributed to rapid valence redistribution facilitated by intermetallic charge transfer during synthesis (Figure S10b) [[88]31]. Additionally, the presence of a Zn 2p peak confirmed the successful coating of ZIF on the PB surface (Figure S10c). This evidence suggested that the catalytic active centers of PB@ZIF and PB were largely consistent. ZIF known for its defect-rich structure, facilitates the adsorption of H₂O and CO₂, forming Lewis acid catalytic sites [[89]32]. Moreover, metal ions, acting as redox centers or Lewis acids, further modulated enzymatic catalysis [[90]33]. Consequently, the local catalytic environment is crucial, particularly in pH-dependent systems like PB, where pH shifts significantly influence its activity [[91]34–[92]36]. This raises the hypothesis that ZIF-8 may alter PB’s local catalytic microenvironment [[93]37, [94]38]. NH[3]-TPD analysis showed a high-temperature desorption peak between 200 °C and 500 °C, corresponding to both Lewis and Brønsted acid sites (Fig. [95]2g) [[96]39]. This indicated that Zn²⁺ played a role in modulating the protonation and deprotonation of H[2]O. Theoretical simulations (Fig. [97]2h) revealed that Zn^2+ interacts with H₂O by accepting lone pair electrons from oxygen, demonstrating Lewis acid activity and elongating the O-H bond (Fig. [98]2i). Following hydrogen ionization, the O-H bond length decreased, enhancing system stability. Additionally, the Zn-O and Zn-N bond lengths shortened after ionization (Fig. [99]2j), further stabilizing the structure. Calculations also indicated that the pKa of the system decreased from 14 to 10.57 due to Zn^2+ (Fig. [100]2k) [[101]40]. These findings suggested that ZIF created a localized acidic catalytic microenvironment, which accelerated the interaction between PB and free radicals, thereby enhancing the overall catalytic process (Fig. [102]2l). PB@ZIF protected BV2 cells against oxidative stress BV2 cells, a widely recognized microglial model [[103]41], were used in our in vitro experiments. The cytotoxicity of PB and PB@ZIF nanomaterials was assessed after a 24 h incubation. As shown in Figure S11 and 12, both PB and PB@ZIF demonstrated excellent biocompatibility at concentrations below 20 µg/mL. We then explored the interaction between PB@ZIF and BV2 cells. BV2 cells were incubated with CY5.5-labeled PB@ZIF for various time intervals, followed by nuclear and lysosome staining using Hoechst (blue) and Lysotracker (green). Results showed that PB@ZIF (magenta) began internalizing into the cells within 1 h of incubation, progressively increasing over time, with significant accumulation after 4 h (Fig. [104]3a). To extend these observations, we compared the uptake efficiency of CY5.5-labeled PB@ZIF across different cell types, including a neuronal cell line (N2a), an astrocyte cell line (HA-1800), and BV2 cells. As shown in Fig. [105]3b and Figure S13 BV2 cells exhibited significantly higher uptake of PB@ZIF compared to N2a and HA-1800 cells, indicating a preferential internalization by microglia. Fig. 3. [106]Fig. 3 [107]Open in a new tab PB@ZIF effectively maintained microglial homeostasis in vitro. (a) Intracellular tracking of CY5.5-labeled PB@ZIF in BV2 cells stained with LysoTracker and Hoechst. The white arrows indicated the intracellular localization of PB@ZIF. The scale bars are 10 μm. (b) The uptake of CY5.5-labeled PB@ZIF was assessed in BV2, HA-1800, and N2a cells. The nucleus was stained with Hoechst. The scale bars are 50 μm. (c) Intracellular ROS levels were detected using the DCFH-DA fluorescent probe and analyzed by flow cytometry. (d) Intracellular ·OH levels induced by LPS in BV2 cells, with or without different treatment, detected with APF fluorescence. The scale bars are 50 μm. (e, f) The corresponding quantified statistics of APF fluorescence intensity (e) and DCFH-DA fluorescence intensity (f). (g, h) Representative fluorescent images of JC-1 (g) and relative quantification of fluorescence ratio in BV2 cells (h). The scale bars are 10 μm. (i) The levels of cellular MDA in BV2 cells. (j-l) The concentrations of IL-1β (j), IL-6 (k), and TNF-α (l) in the BV2 cell culture supernatant were analyzed by ELISA. n = 4 per group, biological replicates. Data are represented with mean ± SEM. All statistical analyses were performed using one-way ANOVA with Tukey’s post hoc test for multiple comparisons LPS is commonly used to induce oxidative stress and inflammatory activation in BV2 cells, simulating pathological conditions [[108]42, [109]43]. To evaluate the effects of PB@ZIF on microglial activation, we quantified total ROS levels and the level of •OH using flow cytometry and fluorescence staining. Following a 2 h LPS exposure, flow cytometry confirmed a substantial increase in total ROS production within BV2 cells (Fig. [110]3c and f). Further, there was a marked increase in APF fluorescence following LPS stimulation, indicating a significant rise in •OH levels (Fig. [111]3d and e). The PB@ZIF treatment effectively reduced the increases in total ROS and •OH, exceeding the protective effects of PB. In contrast, ZIF alone exhibited no significant antioxidant activity, which was consistent with our prior chemical experiments. Since lipids are highly susceptible to oxidative damage from free radicals, we evaluated MDA levels, a marker of lipid peroxidation [[112]44]. Notably, PB@ZIF treatment significantly reduced MDA levels compared to LPS-stimulated cells (Fig. [113]3i). Additionally, higher ROS levels could lead to destruction of mitochondria [[114]45]. MMP depolarization is a hallmark of mitochondrial dysfunction [[115]46]. Our results showed that PB@ZIF restored MMP, helped maintain mitochondrial function (Fig. [116]3g and h). Moreover, we quantified the concentrations of pro-inflammatory cytokines in the supernatants of BV2 cells using ELISA. PB@ZIF treatment significantly suppressed LPS-induced secretion of pro-inflammatory cytokines, including IL-1β, IL-6, and TNF-α (Fig. [117]3j-l). Collectively, these results demonstrated that PB@ZIF functioned as a potent antioxidant to effectively reduce LPS-induced oxidative stress and maintain microglial homeostasis. PB@ZIF effectively alleviated seizure-related behaviors in TLE Given the remarkable protective effects of PB@ZIF on microglia in vitro, we further explored its therapeutic potential in a mouse model of TLE. Firstly, to evaluate the biodistribution of PB@ZIF, C57BL/6n mice received a single tail vein injection of Cy5.5-labeled PB@ZIF. In vivo fluorescence imaging revealed a pronounced enrichment of PB@ZIF within the brain, persisting for up to 7 days post-injection (Fig. [118]4a and Figure S14). One day following the injection, major organs (heart, liver, spleen, lung, and kidneys) were harvested for fluorescence imaging (Figure S15). We also assessed the colocalization of PB@ZIF with various cellular markers in the brain. The results indicated that CY5.5-labeled PB@ZIF exhibited a significantly higher co-localization with IBA1 (a microglial marker) (Fig. [119]4b) compared to NeuN (a neuronal marker) and GFAP (an astrocyte marker) (Figure S16 and 17). Collectively, these findings confirmed that PB@ZIF demonstrated an excellent ability to traverse the blood-brain barrier (BBB) and accumulate within the brain. Additionally, microglia exhibited enhanced phagocytic uptake of PB@ZIF, reflecting their specialized role as the resident immune cells of the central nervous system. Fig. 4. [120]Fig. 4 [121]Open in a new tab PB@ZIF showed significant therapeutic effects in TLE model. (a) In vivo fluorescent imaging of mice at different time points following tail vein injection of CY5.5-labeled PB@ZIF (n = 3 per group, biological replicates). (b) Uptake of CY5.5-labeled PB@ZIF in the microglia of mouse brain detected by immunofluorescence staining (n = 3 per group, biological replicates). The white arrows indicate the accumulations of PB@ZIF in the microglia. The scale bar is 20 μm. (c, d) SRS were assessed by in vivo EEG. Representative raw local field potential (LFP) and LFP spectrogram during SRS in Pilo group (c) and PB@ZIF group (d). Below are enlarged raw LFP segments corresponding to the red dotted boxes. The number and duration of seizures (EEG) were recorded for three continuous days (n = 3 per group, biological replicates). (e, f) Behavioral seizures were evaluated using the video recording for five consecutive days (n = 12 per group, biological replicates). (g) Representative images of the Y maze tests of the mice in each group. (h) Quantification of alternation ration in the Y maze tests (n = 10 per group, biological replicates). (i) Representative images of the open field tests of the mice in each group. (j) Quantification of time spent in the central area in the open field tests (n = 10 per group, biological replicates). Data are represented with mean ± SEM. All statistical analyses were performed using one-way ANOVA with Tukey’s post hoc test for multiple comparisons Then, we utilized a PILO-SE model of TLE to explore the in vivo effects of PB@ZIF. After PILO-SE in mice, a latent period (the time that lapses between the end of the PILO-SE and the occurrence of the first spontaneous seizure) will be observed, followed by the occurrence of spontaneous recurrent seizures (SRS) [[122]47]. This simulates the pathological process of human TLE and provides a robust platform for investigating therapeutic interventions. Accordingly, two weeks after successfully inducing the epilepsy model, we continuously monitored seizure-related behaviors using in vivo EEG recordings and video surveillance. PB@ZIF-treated mice exhibited a significant reduction in SRS frequency and duration compared to pilocarpine-treated controls (Fig. [123]4c-f). Memory deficits are common among patients with epilepsy, especially those with TLE, where seizure activity directly impacts memory-related brain structures [[124]48]. Additionally, depression and anxiety disorders are prevalent psychiatric comorbidities in people with epilepsy [[125]49]. Therefore, we performed a Y-maze test to evaluate memory function and an open field test to measure anxiety-related behavior, 28 days post-status epilepticus. The Y-maze test involved calculating the number of arm entry alternations to assess memory deficits. In Pilo group, spontaneous alternations were significantly reduced compared to Control group, indicating impaired memory; however, PB@ZIF treatment mitigated this impairment (Fig. [126]4g and h). Importantly, no significant differences were observed in the total number of arm entries and total distance traveled among the groups, indicating that the motor abilities of the mice remained unaffected (Figure S18a and b). In the open field test, Pilo group spent significantly less time in the central area compared to Control group, indicative of increased anxiety, while PB@ZIF treatment significantly alleviated this anxiety-like behavior (Fig. [127]4i and j). No significant differences were noted in the total distance and central distance between Pilo group and PB@ZIF group (Figure S19a and b). These results demonstrated that PB@ZIF effectively reduced seizure-related behaviors and ameliorated memory deficits and anxiety-like behavior in epileptic mice. PB@ZIF maintained the microglial homeostasis for epilepsy treatment The PILO-SE model is characterized by significant microglial activation and neuroinflammation in the mouse brain [[128]50], suggesting the impact of impaired microglial homeostasis. To further investigate the specific effects of PB@ZIF on microglial homeostasis, we evaluated microglial activation in PILO-SE model using immunofluorescence staining. We employed CD16, a marker of pro-inflammatory microglia [[129]51], to quantify the density of IBA1^+CD16^+ cells. The results revealed a significant reduction in the density of IBA1^+CD16^+ cells in the hippocampal CA1 region following PB@ZIF treatment compared to Pilo group, indicating an attenuation of pro-inflammatory microglial activation (Fig. [130]5a and b). Fig. 5. [131]Fig. 5 [132]Open in a new tab PB@ZIF improved microglial homeostasis in the pilocarpine-induced epilepsy model. (a, b) Double labeling of IBA1 (red) and CD16 (green) in the hippocampal CA1 region of mice from each group (scale bar, 20 μm). Representative fluorescence staining (a) and corresponding quantification (b). (c, d) Quantification of TUNEL-positive cells density (c) and representative images of TUNEL (red) and DAPI (blue) staining (scale bar, 50 μm) (d) in the mouse hippocampal CA1 region in each group. (e, f) Quantification of NeuN^+ cells density (e) and representative images of NeuN (green) and DAPI (blue) staining (scale bar, 50 μm) (f) in the mouse hippocampal CA1 region in each group. (g-l) Protein levels of NLRP3, caspase-1, cleaved caspase-1, IL-1β, cleaved IL-1β, and β-actin were assessed by western blot analysis (g), with corresponding quantitative data (h-l). n = 6 per group, biological replicates. Data are represented with mean ± SEM. All statistical analyses were performed using one-way ANOVA with Tukey’s post hoc test for multiple comparisons Besides, microglia-mediated inflammatory responses can lead to neuronal apoptosis. We assessed apoptosis by quantifying TUNEL -positive cells and calculating the density of NeuN-positive neurons. In comparison to Control group, Pilo group exhibited a substantial increase in TUNEL-positive cells (Fig. [133]5c and d) and a significant reduction in the density of NeuN^+ cells (Fig. [134]5e and f). However, PB@ZIF treatment effectively reduced apoptosis and increased neuronal density, highlighting its neuroprotective potential. To examine the underlying protective mechanisms of PB@ZIF, we focused on the NLRP3 inflammasome pathway. It functions as a key inflammatory pathway activated by enhanced glycolysis and operates as a downstream target of HIF-1α [[135]13, [136]14], as demonstrated enriched in our previous analysis of epileptic patients and animal models. Western blot analysis revealed that, compared to Control group, Pilo group exhibited markedly elevated expression levels of NLRP3, cleaved caspase-1, IL-1β, and cleaved IL-1β—key components of the pro-inflammatory signaling (Fig. [137]5g-l). Notably, PB@ZIF significantly downregulated the expression of NLRP3, cleaved caspase-1, and cleaved IL-1β. These results suggested that the effects of PB@ZIF on maintaining microglial homeostasis were mediated, at least in part, through inhibition of the NLRP3 inflammasome pathway. Given the critical importance of biosafety and toxicity in the development of nanomedicines, a thorough evaluation of the biosafety profile of PB@ZIF was conducted. Fourteen days after administering PB@ZIF to C57BL/6n mice, major organs (brain, heart, liver, spleen, lung, and kidneys) were subjected to histopathological examination using HE staining (Figure S20). The results revealed no observable morphological damage in any of the examined tissues, indicating a favorable safety profile. Additionally, we conducted a comprehensive analysis of blood parameters, including PLT, WBC, HGB, MCHC, MCV, HCT, and RBC. In addition, biochemical analyses were conducted through the assessment of ALT, AST, CKMB, and CREA levels (Figure S21). The results showed no significant differences between the PB@ZIF-treated group and the saline-treated control group, further confirming the absence of systemic toxicity. Collectively, these results verified the biosafety and excellent biocompatibility of PB@ZIF. PB@ZIF reduced the glycolytic reprogramming of microglia through the regulation of HIF-1α expression Since the onset of epilepsy is linked to alterations in the local brain metabolism, the role of metabolic regulation in maintaining microglial homeostasis has not been fully understood. Consequently, we performed a metabolomic analysis to profile the metabolic alterations in the hippocampus of mice between Pilo and Control groups. PCA analysis of QC samples exhibited tight clustering, indicating minimal technical variability and high data reliability (Figure S22a). Distinct metabolic differences were observed between brain samples from Pilo and Control groups (Figure S22a). OPLS-DA analysis further confirmed this separation (Figure S22b). The variable importance in projection (VIP) score identified 214 differential metabolites between Pilo and Control groups, with 132 upregulated and 82 downregulated as shown in the volcano plot (Fig. [138]6a). These metabolites were classified into 11 chemical categories based on HMDB classification (Fig. [139]6b), with their distribution and proportions shown. Fig. 6. [140]Fig. 6 [141]Open in a new tab PB@ZIF regulated HIF-1α expression for glycometablic reprogramming of microglia. (a) Volcano plot comparing metabolite levels between Control and Pilo groups. Significantly different metabolites were defined with p < 0.05, VIP > 1 and fold change > 1. n = 6 per group, biological replicates. (b) HMDB classification showing both the number and percentage of metabolites. (c) KEGG pathway enrichment analysis of differentially expressed metabolites after pilocarpine induction. (d) Heatmap illustrating the differentially expressed metabolites between Control and Pilo groups. (e-g) Differences in the levels of D-fructose-6-phosphate (e), glyceraldehyde-3-phosphate (f), and NAD^+(g) between Control and Pilo groups. Given the minimal variations in the levels of metabolites after normalization, a truncated Y-axis was used to emphasize intergroup differences. n = 6 per group, biological replicates. (h, l) HIF-1α (green) and IBA1 (red) double labeling in the mouse hippocampal CA1 region in each group. Representative fluorescence staining (scale bar, 10 μm) (h) and quantification (l) (n = 6 per group, biological replicates). (i, m) Protein levels of HIF-1α in BV2 cells at different time intervals after stimulation were assessed using western blot analysis (i) and quantitative data were presented in (m) (n = 4 per group, biological replicates). (j, n) The protein levels of HIF-1α in BV2 cells were determined by western blot analysis at 3 h after stimulation with or without PB@ZIF (j), and the quantitative data were shown (n) (n = 3 per group, biological replicates). (k) Schematic diagram of pCDNA3.1(+)-HIF-1α-ODD-Firefly Luciferase plasmid. (o) Quantitative data showing the effects of in vitro epileptic stimulation and PB@ZIF on PHDs’ activity in BV2 cells transfected with luciferase reporter (n = 3 per group, biological replicates). Data are represented with mean ± SEM. Statistical comparisons for (e-g) were conducted using an unpaired two-tailed Student’s t-test, while one-way ANOVA with Tukey’s post hoc test was applied for (l-o) These metabolites were further enriched by KEGG pathway, and the results showed that the energy metabolism pathway was significantly enriched in Pilo group compared with Control group (Fig. [142]6c). Within this pathway, the accumulation of important glycolytic intermediates, such as D-fructose-6-phosphate and glyceraldehyde-3-phosphate (Fig. [143]6e, f, and Figure S23), suggested an upregulation of glycolysis. While a concomitant reduction in β-nicotinamide adenine dinucleotide (NAD^+), a crucial intermediate in the tricarboxylic acid cycle, indicated a suppression of OXPHOS (Fig. [144]6g and Figure S24). Collectively, these findings indicated that epilepsy induced profound metabolic reprogramming in the hippocampus, characterized by impaired OXPHOS and increased glycolysis. Given that HIF-1α, the master transcriptional regulator of glycolysis, plays a critical role in this metabolic shift, it regulates the metabolic reprogramming of microglia [[145]52]. We further investigated the expression of HIF-1α in microglia. Consistent with metabolomic results, immunofluorescence staining of hippocampal tissue from PILO-SE mice revealed a marked increase in HIF-1α expression within microglia compared to control mice (Fig. [146]6h and l). This upregulation suggested an adaptive response aimed at supporting the enhanced glycolytic flux observed in the epileptic hippocampus. Importantly, treatment with PB@ZIF was shown to mitigate the upregulation of HIF-1α, indicating that PB@ZIF could modulate this critical metabolic pathway. To further substantiate the role of PB@ZIF on microglial glycolysis, we established an in vitro cell culture model of epilepsy. Conditioned medium from neuronal cells following KA treatment was collected to stimulate BV2 cells. HIF-1α demonstrated a temporal protein expression pattern following medium change that peaked at 3 h and subsequently decreased (Fig. [147]6i and m). Concurrent treatment with PB@ZIF significantly reduced the protein level of HIF-1α upregulated by conditioned medium treatment at 3 h (Fig. [148]6j and n). Under normoxic conditions, HIF-1α is highly unstable, with a half-life of a few minutes [[149]53]. Cells primarily enhance HIF-1α expression and function by increasing protein stability. PHDs hydroxylate specific proline residues in HIF-1α’s ODD domain, enabling binding to the von Hippel–Lindau (VHL) ubiquitin ligase complex, which ubiquitinates HIF-1α for proteasomal degradation [[150]54]. Besides, an increase in ROS has been shown to inhibit PHDs’ activity [[151]19]. To explore whether the regulation of HIF-1α by PB@ZIF was mediated by enhancing the activity of PHDs, we developed a dual-Luciferase reporter assay system plasmid containing the ODD domain of HIF-1α fused to the luciferase gene (Fig. [152]6k), as previously reported [[153]55]. This system enables the assessment of PHDs’ activity through luciferase expression stability. BV2 cells were transfected with the luciferase reporter. In vitro simulation of the epileptic environment enhanced luciferase expression indicating a reduction in PHDs enzymatic activity (Fig. [154]6o). However, PB@ZIF reduced the activity of luciferase, suggesting that PB@ZIF may enhance PHDs enzymatic activity by scavenging ROS. The results collectively demonstrated that PB@ZIF regulated the glycolytic reprogramming of microglia by enhancing the activity of PHDs and reducing the ubiquitin-mediated degradation of HIF-1α. Discussion This study revealed potential pathological mechanisms underlying TLE through the lens of neuroinflammation and metabolic reprogramming, and presented a novel therapeutic approach. Previous CITE-seq studies of resected brain tissues from epilepsy patients confirmed a pro-inflammatory microenvironment with extensive microglial activation and infiltration of immune cells [[155]2]. Consistent with this result, our current findings further emphasized the proinflammatory role of microglia in TLE. Furthermore, we identified potential metabolic alterations in microglia. In recent years, significant advancements have been made in immunometabolism, highlighting how metabolic pathways shape immune cell function [[156]56]. For instance, increased glycolysis is a key metabolic shift in immune cells during activation, providing the necessary ATP and intermediates for functions like phagocytosis and cytokine production [[157]56]. However, metabolic alterations in epilepsy remain insufficiently explored. Our metabolomic analysis revealed impaired OXPHOS and increased glycolysis in hippocampal tissues from PILO-SE models, suggesting a metabolic reprogramming post-epilepsy induction. HIF-1α is recognized as one of the central regulators of global metabolism in mammals [[158]57]. It plays a crucial role in metabolic reprogramming by facilitating the full expression of several key glycolytic genes [[159]58]. In other neurological disorders like Alzheimer’s disease, metabolic profiling reveals that amyloid-β (Aβ) exposure shifts metabolism from OXPHOS to glycolysis via the HIF-1α pathway and triggers acute microglial inflammation [[160]52]. However, reversing HIF-1α-induced metabolic impairment can enhance the microglial response to Aβ pathology [[161]59]. Our findings indicated that HIF-1α expression was upregulated in microglia following epilepsy induction, suggesting increased glycolysis of microglia during epilepsy. This finding highlights microglial metabolic alterations and expands on previous studies that reported upregulation of HIF-1α in neurons within hippocampal regions affected by seizures in TLE patients [[162]60]. HIF-1α is subjected to polyubiquitination by VHL ubiquitin ligase complex, leading to its subsequent proteasomal degradation [[163]61]. The highly specific molecular recognition of HIF-1α by VHL relies on PHDs, which mediate the post-translational hydroxylation of specific proline residues in HIF-1α [[164]54, [165]62]. PHDs belong to an evolutionarily conserved subfamily of dioxygenases that utilize oxygen and 2-oxoglutarate as co-substrates, with ferrous iron and ascorbate serving as essential cofactors [[166]63, [167]64]. ROS can inhibit PHDs by chelating and oxidizing PHDs bound Fe(II) to Fe(III) [[168]65, [169]66]. This suggests that regulating intracellular ROS could serve as a strategy to modulate the activity of PHDs. Therefore, in order to regulate PHDs and subsequently affect the levels of HIF-1α, we designed a nanocatalyst (PB@ZIF) with exceptional catalytic activity and stability, capable of effectively eliminating free radicals. Notably, compared with PB, the catalytic efficiency of PB@ZIF was enhanced by altering the local catalytic microenvironment, independent of changes in the valence state of elements. This finding implied that, in the realm of nanocatalytic medicine, apart from modulating the electronic and geometric structures of central metal atoms, employing non-catalytic centers to regulate reaction conditions can also influence catalytic performance. Based on PB@ZIF’s catalytic performance, we confirmed that it could significantly mitigate oxidative stress and enhance PHDs’ activity, thereby reducing HIF-1α protein levels in microglia. Furthermore, PB@ZIF effectively traversed the BBB and preferentially interacted with microglia. Consequently, it inhibited neuroinflammation and neuronal damage, and ameliorated seizures and behavioral impairments. These results suggested that epilepsy, characterized by impairment of metabolic homeostasis, also altered microglial metabolism, leading to their glycometabolic reprogramming. Glycometabolic reprogramming in microglia induced a damaging inflammatory response, promoting a vicious cycle of inflammation and seizure. We propose a novel approach by regulating glycolytic metabolism in microglia, which may provide a new perspective for epilepsy treatment. Our results demonstrate that nanocatalysts are an effective approach to regulating enzyme activity, though many challenges remain before achieving clinical application. Future research should focus on validating the efficacy of PB@ZIF in larger animal models and enhancing its delivery methods, such as modifying the compound to improve targeting and clinical applicability. Additionally, exploring the synergistic effects of PB@ZIF with other therapies, particularly those targeting neuroinflammation and neuroprotection, will be crucial for advancing its clinical application. We believe that PB@ZIF presents a promising avenue for epilepsy treatment while also offering a valuable perspective for advancing nanomedicine strategies aimed at regulating metabolic homeostasis. Conclusion This study investigates the pathological mechanisms underlying TLE, focusing on neuroinflammatory processes and metabolic reprogramming in microglia. The findings underscore the relevance of microglial neuroinflammation and metabolic reprogramming in TLE, driven by ROS-induced stabilization of HIF-1α. Comprehensive analyses, including snRNA-seq and metabolomic profiling, revealed significant metabolic alterations in microglia from TLE patients and animal models, specifically increased glycolysis and decreased OXPHOS. To regulate these mechanisms, we developed PB@ZIF, a nanocatalyst that combines PB with ZIF, designed to enhance catalytic efficiency by altering the pKa of the localized in-situ catalytic environment. PB@ZIF shows potential in scavenging free radicals, enhancing PHDs’ activity, and downregulating HIF-1α in microglia, which may help restore metabolic homeostasis and mitigate neuroinflammation. Its ability to cross the BBB allows it to accumulate significantly in microglia, offering a promising strategy to regulate both metabolic reprogramming and inflammatory processes in epilepsy. This research advances the understanding of TLE pathogenesis and introduces PB@ZIF as a potential nanocatalytic therapy. Electronic supplementary material Below is the link to the electronic supplementary material. [170]Supplementary Material^ (5.3MB, docx) Acknowledgements