Introduction

Epilepsy, a debilitating neurological disorder affecting more than 70 million people worldwide, is characterized by recurrent, spontaneous neuronal discharges caused by an imbalance between excitatory and inhibitory signaling1. The diverse etiologies of epilepsy, ranging from genetic predisposition to traumatic brain injury, underscore the complexity of its pathogenesis2. Although anti-seizure medications remain the mainstay of treatment, approximately one-third of patients are resistant to pharmacotherapy, and many experience serious adverse effects, including fatigue, hyperlipidemia, and osteoporosis3. Currently, no available therapy can halt disease progression or cure epilepsy, and affected individuals experience increased mortality, disability, and comorbidities including neurological, cognitive, and psychiatric disorders, placing a heavy burden on families and society4.

Neuroinflammation is common in epilepsy, often focal but sometimes widespread5. It can arise from seizure-induced neuronal injury with release of damage-associated molecular patterns (DAMPs), blood–brain barrier disruption with serum protein extravasation and infiltration of peripheral immune cells, CNS or systemic infection, or autoimmune or genetic disorders that activate immune pathways6. These events activate resident microglia and innate immune receptors, including Toll-like receptors (TLRs) and inflammasomes; the resulting cytokines and chemokines disrupt local physiology and recruit peripheral immune cells, thereby amplifying the inflammatory milieu7. Acute neuroinflammation contributes to repair, for example through microglial clearance of damaged cells. However, chronic inflammation causes sustained increases in cytokines (interleukin [IL]-1β, IL-6, and tumor necrosis factor [TNF]-α) and triggers apoptosis and pyroptosis. These alterations impair synaptic integrity and intrinsic neuronal excitability, lower the seizure threshold, and contribute to drug resistance8,9,10. Although the causes of neuroinflammation in epilepsy are heterogeneous, elucidation of inflammatory pathways mediated by microglia may identify targets for therapeutic intervention.

Interferon (IFN) signaling, a cornerstone of innate immunity, rapidly induces the expression of IFN-I and IFN-II and their downstream effectors to defend against bacteria and viruses pathogens11. Among these effectors, the IFN-induced transmembrane (IFITM) protein family, comprising IFITM1, IFITM2, IFITM3, IFITM5, and IFITM10, was initially identified as IFN-responsive genes in human neuroblastoma cells12. IFITM proteins localize to the plasma and endolysosomal membranes to inhibit RNA viral entry; IFITM3 also regulates cell adhesion, migration, apoptosis, differentiation, and neuroinflammatory signaling13,14.

The signal transducer and activator of transcription (STAT) family regulates various biological processes, including homeostasis, tumorigenesis, inflammation, and immunity15. STAT2, a key mediator of IFN signaling, promotes inflammatory cytokine production and mediates antiviral, anti-apoptotic, immunomodulatory, and anti-proliferative effects16. STAT2 has been shown to confers neuroprotection in multiple sclerosis, Alzheimer’s disease (AD) and encephalitis17. In addition, STAT2 modulates hippocampal GABA receptor expression following status epilepticus18.

MicroRNAs (miRNAs) function post-transcriptionally by binding to complementary sequences within the 3′-untranslated region (3′-UTR) of target mRNAs, thereby inhibiting translation or promoting mRNA degradation19. In mammals, the let-7 family comprises ten evolutionarily conserved members (let-7a, let-7b, let-7c, let-7d, let-7e, let-7f, let-7g, let-7i, miR-98, and miR-202)20. During biogenesis, precursor miRNAs may yield two functional strands (e.g., let-7g-5p and let-7g-3p)21. Members of the let-7 family regulate neuronal differentiation and maturation, dampen neuroinflammation and offer neuroprotection. Let-7g-5p has been proposed as a potential biomarker for AD and ischemic stroke22,23.

This study was designed to elucidate the molecular mechanisms underlying inflammation, apoptosis, and pyroptosis in epilepsy, with the objective of identifying potential therapeutic targets. Given that lipopolysaccharide (LPS) mimics bacterial components and acts as a TLR4 agonist, and that BV2 cells represent a well-established in vitro model of microglial activation24, experiments were conducted to investigate the molecular mechanisms of epilepsy-associated neuroinflammation in the context of TLR4/DAMP-like signaling and to validate the expression of associated molecules in animal models.

Materials and methods

Bioinformatics analysis

Expression data for mouse hippocampus in kainic acid induced mesial temporal lobe epilepsy (GSE73878) and BV2 microglia treated with LPS, luteolin, or both (GSE18740) were retrieved from the Gene Expression Omnibus (GEO). Differential expression analysis was assessed via the “limma” package. Genes with |log2FC| > 1 and adjusted P < 0.05 were deemed significant. The “VennDiagram” package was used to illustrate intersections among differentially expressed gene (DEG) sets.

Potential transcription factors (TFs) regulating the identified DEGs were predicted by querying the UCSC Genome Browser (https://genome.ucsc.edu/) and JASPAR database (http://jaspar.genereg.net/). Putative miRNAs predicted to target these TFs were obtained from TargetScan (http://www.targetscan.org/vert_71/).

Cell culture

BV2 microglial cells and HEK293T cells (Cellcook Biotech, Guangzhou, China) were maintained in high-glucose Dulbecco’s Modified Eagle Medium (DMEM; Gibco, USA) containing 10% heat inactivated fetal bovine serum and 100 U/mL penicillin–streptomycin (Beijing Solarbio). Cells were cultured at 37 °C in a humidified atmosphere containing 5% CO2.

Mouse microglia inflammation model

BV2 cells were stimulated with LPS at concentrations of 0.25, 0.5, 1, or 2 µg/mL and cultured for 6–24 h. Control BV2 cells were cultured with phosphate-buffered saline (PBS). The LPS concentration that induced a significant upregulation of IL-1β, IL-6, and TNF-α expression (P < 0.05) was designated as the optimal treatment condition for the in vitro model, supporting the robustness of this model in recapitulating seizure-related neuroinflammatory mechanisms.

Cell transfection

According to the specific experimental design, overexpression plasmids (IFITM3, IRF9, or STAT2; Table S1), siRNAs (si-IFITM3-1 to 3, si-IRF9-1 to 4, or si-STAT2-1 to 4; Table S2), miRNA mimics or inhibitors (let-7e-5p and let-7g-5p; Table S3), or their corresponding negative controls (Table S4) were mixed with Lipofectamine™ 2000 (Thermo Fisher Scientific, USA) in Opti-MEM™ and incubated at 20–25 °C for 20 min. Following dropwise delivery of the complexes, cells were incubated for 4–6 h at 37 °C and 5% CO2 before replacement with complete DMEM.

Animals

Male C57BL/6 mice (8 weeks old, weighing 20–25 g) were housed under specific pathogen-free (SPF) conditions (23 ± 1 °C temperature, humidity at 40–60%, alternating 12-h light/dark rhythm). All animals had free access to standard chow and water. The overall experimental design is illustrated in Fig. 1. All experimental procedures were conducted in an SPF facility and were approved by the Experimental Animal Ethics Committee of Xiangya Hospital, Central South University (Protocol No. 2025030673). This study was conducted in strict accordance with the U.S. National Institutes of Health Guide for the Care and Use of Laboratory Animals and is reported following the ARRIVE guidelines 2.0.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Schematic diagram of the in vivo experimental design.

Stereotaxic lentiviral injection and surgery

Lentiviral shRNA vectors targeting the murine IFITM3 gene (LV-sh-IFITM3) and a non-targeting control (LV-sh-NC) were procured from Genechem Biotechnology (Shanghai, China; Table S5). Mice (n = 48) were randomized into three groups of 16 each: (i) vehicle + LV-sh-NC, (ii) pentylenetetrazole (PTZ) + LV-sh-NC, and (iii) PTZ + LV-sh-IFITM3. Following deep anesthesia with isoflurane (induction, 3%; maintenance, 2%), mice were secured in a stereotaxic frame (Stoelting Co., Ltd., St. Louis, MO, USA). Subsequently, an 8 µL volume of lentiviral suspension was bilaterally injected into the dorsomedial caudate-putamen (AP -0.3 mm, ML ± 1.0 mm, DV -2.5 mm from the bregma) at a rate of 1 µL/min using a Hamilton syringe (Hamilton Company, Beltsville, MD, USA). The needle remained for 5 min post-infusion to facilitate diffusion, and was gradually withdrawn. Additionally, EEG recordings were obtained from three mice per group using a Quanlan AR4 four-channel amplifier (Shanghai Quanlan Biotech Co., Ltd., Shanghai, China). The skull was exposed through a midline scalp incision, and a stainless-steel screw electrode was inserted into the left dorsal hippocampal CA1 region via burr holes (AP − 1.5 mm, ML ± 1.8 mm, DV − 1.5 mm from the bregma). Four cortical screws were implanted anterior to the bregma to record bilateral cortical activity, while two screws placed posterior to the lambda acted as ground and reference. All electrodes were embedded in dental acrylic to form a watertight head cap, and animals were allowed to recover for 7 days.

PTZ-induced epilepsy and EEG recording

PTZ (Sigma, P6500) was dissolved in saline (0.01 g/mL) and injected intraperitoneally at 35 mg/kg every other day for 29 days to the treatment groups (ii and iii). Control animals (i) received equivalent volumes of saline. Seizure severity was scored using the extended Racine scale: 0, no abnormality; 1, mouth and facial movements; 2, head nodding; 3, forelimb clonus; 4, rearing; and 5, rearing and falling. Kindling was defined as the occurrence of three consecutive seizures of grade 4 or 5. EEG was recorded in accordance with the manufacturer’s instructions.

Tissue collection and preparation

Mice were anesthetized with isoflurane (induction, 3%; maintenance, 2%). The anesthesia was carefully monitored to ensure the animals reached an adequate surgical plane, characterized by the absence of both the righting reflex and the paw-withdrawal reflex, before euthanasia by decapitation. Hippocampal and cortical tissues were quickly excised on ice and stored at − 80 °C. Tissues were thawed and homogenized in PBS with protease inhibitor cocktail (P8340; Sigma-Aldrich, St. Louis, MO, USA) for protein extraction. Homogenates were centrifuged at 12,000 × g for 15 min at 4 °C, and the supernatants were collected and stored at − 80 °C for subsequent analyses.

Flow cytometry

Cells were washed with PBS, resuspended in 1× binding buffer (2.5 mM Ca2+), and stained with 5 µL each of Annexin V-FITC and propidium iodide (PI) using an Annexin V FITC/PI kit (BD Biosciences, Franklin Lakes, NJ, USA). Samples then underwent a 15-minute incubation at 25 °C in darkness. Apoptotic events were detected on an Attune NxT flow cytometer (Thermo Fisher Scientific, Waltham, MA, USA), and the data were processed using the instrument-associated Attune NxT software.

Enzyme-linked immunosorbent assay (ELISA)

After overnight culture at 37 °C with 5% CO2 to 70–80% confluence, supernatants were harvested by centrifugation (2000 × g, 5 min, 4 °C). Concentrations of IL-6, IL-1β, and TNF-α were quantified by ELISA (Abcam) following the kit instructions. Absorbance at 450 nm was read (570 nm reference) on an xMark™ Microplate Spectrophotometer (Bio-Rad Laboratories, Hercules, CA, USA.) using Microplate Manager 6 software.

Dual-luciferase activity assay

A fragment of the IFITM3 promoter encompassing the predicted STAT2-binding sites was cloned upstream of firefly luciferase coding sequence in the pGL3-Promoter vector to produce the wild type reporter (IFITM3-pro-wt). Subsequently, site-directed mutagenesis was performed to create a STAT2 binding deficient variant (IFITM3-pro-mut). Separately, the STAT2 3′-UTR fragment harboring the let-7g-5p or let-7e-5p seed sequence was cloned downstream of the luciferase gene in the pMIR-REPORT vector to generate the wild type 3′-UTR reporter (STAT2-3′UTR-wt). Point mutations were introduced to abolish miRNA binding, yielding the mutant reporter (STAT2-3′UTR-mut). HEK293T or BV2 cells were co-transfection with the designated reporter constructs, along with either a STAT2 overexpression plasmid or let-7 mimics, and a Renilla luciferase vector for normalization. After 48 h, luminescence was measured with the Dual-Luciferase Reporter Assay System (Promega), and the firefly-to-Renilla ratio was calculated.

Reverse transcription quantitative real-time PCR (RT-qPCR)

Total RNA was isolated using TRIzol reagent (Invitrogen) and assessed by spectrophotometry. Complementary DNA was synthesized with the PrimeScript™ RT kit (Takara, Kyoto, Japan). The 7500 Real-Time PCR System (Applied Biosystems) was used to conduct RT-qPCR, employing GoTaq® qPCR Master Mix (Promega, A6001) for amplification. Target gene primers were obtained from Shenggong Bioengineering Co., Ltd. (China), with their corresponding sequences provided in Table S1. The reaction was performed under conditions in accordance with the manufacturer’s instructions: denaturation at 95 °C for 2 min, followed by 40 cycles of denaturation at 95 °C for 10 s and annealing at 60 °C for 1 min. Relative gene expression was quantified using the 2−ΔΔCt method.

Western blot

Total protein was isolated from the BV2 cells and from the supernatants of homogenized mouse tissues via a Whole Protein Extraction Kit (Abcam; Cat. No. ab270054), and was quantified by the bicinchoninic acid assay (Beyotime Biotechnology, Haimen, China). Separation was then performed on 10–15% sodium dodecyl sulfate-polyacrylamide gel (Bio-Rad, USA), and electrophoresis was carried out at 80 V for approximately 20 min in the stacking gel before increasing to 120 V for around 60 min in the resolving gel. Gels were equilibrated in ice-cold transfer buffer for 10 min and then transferred onto polyvinylidene fluoride membranes (Immobilon®-P, Merck Millipore, IPVH00010) at 100 V for 90 min at 4 °C. Membranes were exposed overnight at 4 °C to primary antibodies (1:1000; Abcam, Cambridge, MA, USA) targeting IFITM3, STAT2, Bax, Bcl-2, cleaved caspase-3 (c-caspase-3), caspase-3, cleaved caspase-9 (c-caspase-9), caspase-9, and the internal control GAPDH, followed by HRP-conjugated secondary antibodies (Boster Bioengineering) for 2 h. Enhanced chemiluminescence was used to visualize protein bands with the BioSpectrum imaging system (UVP, USA), and quantified with ImageJ software (NIH).

Immunofluorescence staining

BV2 cells were fixed in 4% paraformaldehyde/sucrose, permeabilized with 0.3% Triton X-100, blocked in 10% goat serum for 1 h, and incubated overnight at 4 °C with primary antibodies diluted in 1% bovine serum albumin alongside isotype and non-primary controls. After PBS washes, the sections were incubated for 1 h in darkness with secondary antibodies. Then cells were mounted in antifade medium containing DAPI. The primary antibodies targeted STAT2 (rabbit, 1:100; Proteintech) and IFITM3 (mouse, 1:100; Zhongshan Golden Bridge, Beijing, China). Secondary goat anti-rabbit and anti-mouse IgGs (1:200; Zhongshan Golden Bridge) were conjugated to DyLight 488 and 549, respectively. Images were captured using a Panoramic scanner and analyzed using ImageJ software (version 8.0).

Statistical analysis

Data are presented as mean ± standard error of the mean (SME) and analyzed using GraphPad Prism version 10.1.2 for Windows (GraphPad Software, San Diego, California USA). For comparisons between two groups, an unpaired Student’s t-test was applied. One-way analysis of variance (ANOVA) was used to evaluate differences among more than two groups. In cases involving two independent variables, two-way ANOVA was employed to evaluate the main and interaction effects. All cell experiments were performed with three biological replicates. Statistical significance was defined as P < 0.05.

Results

Establishment of the microglia inflammation model

BV2 cells treated with 2 µg/mL LPS for 6 h induced significantly higher levels of IL-1β, IL-6, and TNF-α compared to other LPS concentrations, incubation durations, and the PBS control (Fig. S1), indicating that this treatment condition constitutes the most suitable in vitro model for studying epilepsy-related inflammatory responses.

IFITM3 overexpression in LPS-treated BV2 cells promotes inflammation

The GSE73878 and GSE18740 datasets were retrieved from the GEO repository. Differential expression analysis using an R package yielded 10 DEGs (Fig. 2A), of which the five most significantly altered were chosen for further analysis. Transcript levels of these candidate genes were measured via RT-qPCR in BV2 cells treated with PBS or LPS. LPS stimulation elicited a significantly increased expression of IFITM3 (Fig. 2B), implying that IFITM3 is involved in the inflammatory cascades triggered by microglial activation.

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Overexpression of IFITM3 in LPS-stimulated BV2 cells enhances inflammatory responses. (A) Venn diagram of the GSE73878 and GSE18740 datasets. (B) RT-qPCR analysis demonstrated that IFITM3 was significantly upregulated in the cellular inflammation model. (C) The knockdown efficiency of si-IFITM3-1 to -3 was measured by RT-qPCR. (D) RT-qPCR analyses and (E) western blot were used to detect the effect of IFITM3 overexpression. (F–H) Suppression of IFITM3 expression reduced the production of IL-1β, IL-6, and TNF-α in LPS-treated BV2 cells. (I–K) The production of IL-1β, IL-6, and TNF-α was enhanced following transfection of IFITM3 into BV2 cells treated with LPS. (L–N) The levels of IL-1β, IL-6, and TNF-α were measured using ELISA in LPS-treated BV2 cells transfected with si-IFITM3-2, si-IFITM3-3, IFITM3, or their corresponding negative controls.

IFITM3 promotes inflammation in LPS-treated BV2 cells

All three IFITM3-targeting siRNAs effectively reduced IFITM3 expression (Fig. 2C), while the IFITM3 overexpression plasmid produced a significant elevation of IFITM3 levels (Fig. 2D,E). BV2 cells treated with LPS were transfected with si-IFITM3, IFITM3 plasmid, or corresponding negative controls constructs. IFITM3 knockdown significantly reduced IL-1β, IL-6, and TNF-α expression (Fig. 2F–H), while IFITM3 overexpression produced a significant rise in these cytokines (Fig. 2I–K). These alterations were corroborated at the protein level by ELISA (Fig. 2L–N). These data suggested that IFITM3 silencing mitigates microglial inflammation responses, whereas overexpression of IFITM3 exacerbates inflammation.

IFITM3 promotes apoptosis and pyroptosis in LPS-treated BV2 cells

Flow cytometry and RT-qPCR analyses revealed that IFITM3 silencing significantly inhibited apoptosis and the expression of pyroptosis-associated markers (apoptosis-associated speck-like protein containing a caspase activation and recruitment domain [ASC] and IL-18), whereas IFITM3 overexpression produced a reciprocal increase (Fig. 3A–C). These results indicated that IFITM3 contributes to programmed cell death (PCD) and amplifies microglia-associated inflammatory responses.

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

IFITM3 promotes apoptosis and pyroptosis in LPS-treated BV2 cells. (A) The rate of BV2 cells undergoing pyroptosis was quantified by flow cytometry. (B,C) The relative expression levels of ASC and IL-18 were determined by RT-qPCR.

IFITM3 knockdown alleviates seizure severity and neuroinflammatory responses in PTZ-kindled mice

For in vivo validation of the function of IFITM3, a PTZ-kindled mouse model was established followed by lentiviral-mediated IFITM3 knockdown. Efficient silencing of IFITM3 was confirmed by Western blot and RT-qPCR (Fig. 4A,D, S2A). Behaviorally, the proportion of mice exhibiting severe seizures (score ≥ 4) was markedly decreased, with seizure scores predominantly shifting below grade 4 after LV-sh-IFITM3 treatment (Fig. 4B).

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

IFITM3 knockdown alleviates seizure severity and neuroinflammatory responses in PTZ-kindled mice. (A) Western blot analysis was conducted to assess the expression of IFITM3, verifying the knockdown efficiency of LV-sh-IFITM3 in BV2 cells treated with LPS (n = 3). (B) Severity of seizures was scored according to the Racine scale for the vehicle + LV-sh-NC, PTZ + LV-sh-NC, and PTZ + LV-sh-IFITM3 groups. No seizures were observed in the vehicle + LV-sh-NC group, yielding a score of 0 (n = 16). (C) Representative EEG traces recorded in vivo from each experimental group (n = 3). (D–I) mRNA levels of IFITM3, IL-1β, IL-6, TNF-α, IL-18, and ASC in mouse cortical tissue were determined by RT-qPCR (n = 16). (J,K) Protein levels of IFITM3, cleaved caspase-3, caspase-3, cleaved caspase9, caspase-9, BAX, and BCL2 were evaluated by western blotting in the cortex lysates (n = 3). Statistical analysis was conducted using two-way ANOVA followed by Bonferroni multiple comparison tests. Data are presented as mean ± standard deviation.

Consistently, EEG recordings showed that PTZ-induced interictal epileptiform discharges, characterized by high-frequency and high-amplitude spike activity, were reduced following IFITM3 knockdown (Fig. 4C).

Given the established link between neuroinflammation and seizure susceptibility, we next assessed inflammatory mediators. PTZ treatment significantly increased IL-1β, IL-6, and TNF-α levels, whereas IFITM3 silencing markedly reduced IL-1β and TNF-α expression (Fig. 4E–G, Fig. S2B–D). Furthermore, pyroptosis-related markers ASC and IL-18 were elevated in PTZ mice but were significantly suppressed after IFITM3 knockdown (Fig. 4H,I, Fig. S2E,F). Apoptotic signaling was also affected. PTZ exposure increased pro-apoptotic proteins (cleaved caspase-3, cleaved caspase-9, and Bax) while decreasing the anti-apoptotic protein Bcl-2; these alterations were reversed by LV-sh-IFITM3 treatment (Fig. 4J,K, Fig. S2G,H). Collectively, these data demonstrate that IFITM3 knockdown mitigates seizure severity and attenuates neuroinflammatory, pyroptotic, and apoptotic responses in PTZ-kindled mice.

STAT2 as a potential transcription factor for IFITM3

Based on transcriptional regulatory relationships and binding sites, potential TFs predicted to regulate IFITM3 were identified using the UCSC and JASPAR databases. Five candidate TFs were identified and subsequently evaluated in BV2 cells. Upon LPS treatment, IRF9 and STAT2 expression was significantly upregulated relative to PBS-treated controls (Fig. 5A).

Fig. 5
Fig. 5The alternative text for this image may have been generated using AI.
Full size image

STAT2 as a potential transcription factor for IFITM3. (A) The expression levels of transcription factors IRF9, STAT2, MAFF, TFAP2C, and PITX2 in BV2 cells were measured, with Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) serving as the internal control. The overexpression efficiency of (B) IRF9 or (C) STAT2 plasmids was measured. (D) The inhibition efficiency of four si-IRF9 constructs was assessed in BV2 cells using RT-qPCR. (E) The inhibition efficiency of four si-IFITM3 constructs was similarly evaluated in BV2 cells. IFITM3 expression was evaluated by RT-qPCR in BV2 cells transfected with (F) IRF9 or (G) STAT2. RT-qPCR was performed to quantify IFITM3 expression in BV2 cells transfected with (H) si-IRF9-1 or (I) si-STAT2-3. (J) HEK293T cells were co-transfected with the IFITM3 promoter and negative control, IRF9, or STAT2 plasmids. After 48 h, luciferase activity and corresponding protein expression were detected. (K) BV2 cells were co-transfected with STAT2 or negative control together with either the IFITM3 promoter or IFITM3-promoter-mut. The relative luciferase activity was measured. (L) IFITM3 protein levels were examined by western blot in BV2 cells transfected with si-NC, si-STAT2-3, si-STAT2-4, oe-NC or STAT2. (M) Representative immunofluorescence images (magnification ×400) demonstrate the expression of IFITM3. Blue: nuclear staining (DAPI); red: IFITM3 staining. Scale bar: 100 μm. STAT2 expression in the cortex (N) and hippocampus (O) of epileptic mice was assessed by RT-qPCR (n = 16).

Pronounced increases in IRF9 and STAT2 levels were achieved by transfection of their overexpression plasmids, whereas application of si-IRF9-1, si-STAT2-3, and si-STAT2-4 led to effective knockdown (Fig. 5B–E). When these agents or corresponding negative controls were introduced into LPS-treated BV2 cells, IFITM3 expression paralleled that of IRF9 and STAT2, indicating that both factors positively regulate IFITM3 (Fig. 5F–I).

IFITM3-pro-wt and IRF9, STAT2, or negative controls were co-transfected into HEK293T cells. STAT2 elicited a significantly greater enhancement of luciferase activity than IRF9 (Fig. 5J). In BV2 cells, STAT2 increased luciferase activity in the presence of IFITM3-pro-wt, but not in the presence of IFITM3-pro-mut (Fig. 5K). Western blotting and immunofluorescence staining confirmed that STAT2 overexpression upregulated IFITM3, whereas STAT2 knockdown downregulated IFITM3 (Fig. 5L,M). In PTZ-kindled mice, STAT2 expression was significantly elevated in both the cortex (Fig. 5N) and hippocampus (Fig. 5O). Notably, IFITM3 inhibition produced no significant change in STAT2 expression levels. These results demonstrated that STAT2 directly activates IFITM3 transcription and functions upstream of IFITM3 in this regulatory pathway.

STAT2 promotes inflammation, apoptosis, and pyroptosis through IFITM3

Compared to the control group, STAT2 overexpression significantly increased IL-1β, IL-6, and TNF-α levels, whereas introduction of si-IFITM3-3 resulted in a pronounced reduction of these proinflammatory cytokines. The STAT2-driven increase in cytokine levels was attenuated by co-transfection with si-IFITM3-3 (Fig. 6A–C). Cytokine production was reduced by STAT2 knockdown and was subsequently restored by overexpression of IFITM3 (Fig. 6D–F). Comparable trends were seen in apoptotic rates (Fig. 6G,H) and pyroptosis-associated markers ASC and IL-18 (Fig. 6I–L). These data indicated that STAT2 promotes inflammatory responses and PCD by transcriptionally activating IFITM3, consistent with its role as a transcriptional regulator in innate immune signaling.

Fig. 6
Fig. 6The alternative text for this image may have been generated using AI.
Full size image

STAT2 promotes inflammation, apoptosis, and pyroptosis through modulation of IFITM3. (A–C) Levels of IL-1β, IL-6, and TNF-α were quantified in BV2 cells transfected with STAT2, si-IFITM3-3, or the corresponding negative controls. The rescue experiments involved co-transfection of STAT2 with si-IFITM3-3. (D–F) The expression levels of IL-1β, IL-6, and TNF-α in BV2 cells transfected with si-STAT2-3, IFITM3, or the respective negative controls. The rescue experiments involved co-transfection of si-STAT2-3 with IFITM3. (G) The apoptosis rate of BV2 cells transfected with si-STAT2-3, IFITM3, or their negative controls. (H) Flow cytometry was performed to assess apoptosis rate in each treatment group. The expression levels of (I,K) ASC and (J,L) IL-18 in BV2 cells treated with different plasmids were detected by RT-qPCR analysis.

STAT2 as a potential target of let-7g-5p

Candidate miRNAs predicted to target STAT2 by TargetScan were identified, and screened in BV2 cells following exposure to LPS or PBS. Because miRNAs commonly repress target transcripts, candidates that displayed decreased expression after LPS stimulation were prioritized for downstream analysis. Among these candidates, let-7g-5p and let-7e-5p were selected for further study (Fig. 7A,B).

Fig. 7
Fig. 7The alternative text for this image may have been generated using AI.
Full size image

STAT2 as a potential target of let-7g-5p. (A) Let-7e-5p and (B) let-7g-5p were significantly downregulated in LPS-stimulated BV2 cells. (C,D) The relative overexpression efficiency of let-7e-5p mimic and let-7g-5p mimic was examined by RT-qPCR. (E,F) The relative knockdown efficiency of let-7e-5p inhibitor and let-7g-5p inhibitor was assessed using RT-qPCR. (G,H) STAT2 expression levels in BV2 cells transfected with let-7e-5p mimic or let-7g-5p mimic were measured by RT-qPCR. (I,J) STAT2 expression levels in BV2 cells transfected with let-7e-5p inhibitor or let-7g-5p inhibitor were evaluated by RT-qPCR. (K) Relative luciferase activities in HEK293T cells co-transfected with STAT2-3′UTR together with let-7e-5p mimic, let-7g-5p mimic, or negative controls. (L) A schematic diagram of the bases targeted by let-7g-5p in the STAT2 3′-UTR, as well as the base sequence of the mutated STAT2 3′-UTR. (M) Relative luciferase activities of STAT2-3′UTR and STAT2-3′UTR-mut co-transfected with let-7g-5p mimic or a negative control. Let-7g-5p expression in the cortex (N) and hippocampus (O) of epileptic mice was assessed by RT-qPCR (n = 16).

Thereafter, let-7g-5p and let-7e-5p mimics, their inhibitors, and negative controls were transfected into LPS-treated BV2 cells. Transfection of mimics produced substantial increases in the corresponding miRNA levels, whereas inhibitor transfection resulted in effective reductions (Fig. 7C–F). Overexpression of either let-7g-5p or let-7e-5p reduced STAT2 expression (Fig. 7G,H), while inhibition of these miRNAs produced an increase in STAT2 (Fig. 7I,J).

HEK293T cells were co-transfected with STAT2-3′UTR-wt together with either let-7g-5p mimic, let-7e-5p mimic, or negative controls. A more pronounced suppression of luciferase activity was observed with the let-7g-5p mimic compared to let-7e-5p (Fig. 7K). BV2 cells were co-transfected with STAT2-3′UTR-wt/mut and let-7g-5p mimic or negative controls. Let-7g-5p mimic failed to reduce luciferase activity of the STAT2-3′UTR-mut, whereas it robustly suppressed the wild-type reporter, indicating target site specificity (Fig. 7M). In PTZ-kindled mice, levels of let-7g-5p were markedly reduced in both the cortex (Fig. 7N) and hippocampus (Fig. 7O). These results support that let-7g-5p interacts directly with the STAT2-3′UTR to inhibit STAT2 expression.

Let-7g-5p inhibits inflammation, apoptosis, and pyroptosis by targeting STAT2

Let-7g-5p mimic in LPS-treated BV2 cells significantly reduced levels of IL-1β, IL-6, and TNF-α, and concurrently decreased apoptosis rate and pyroptosis-associated markers (ASC, IL-18), while reintroducing STAT2 completely reversed these effects (Fig. S3A–C,G,I,J). Reciprocally, let-7g-5p inhibitor increased cytokine secretion, apoptosis, and pyroptosis, which were all blocked by STAT2 knockdown (Fig. S3D–F,H,K,L). These findings confirmed that let-7g-5p inhibits inflammatory cytokine production, apoptosis, and pyroptosis by targeting STAT2.

Let-7g-5p inhibits IFITM3 expression by targeting STAT2

In LPS-treated BV2 cells, let-7g-5p inhibitor increased IFITM3 and STAT2 expression, while let-7g-5p mimic decreased their expression (Fig. S4A–D). Considering that STAT2 activates IFITM3 transcription, this suggests that let-7g-5p targets STAT2 to restrain IFITM3 expression, attenuating downstream inflammation, apoptosis, and pyroptosis, thereby providing neuroprotection in epilepsy.

Discussion

Epilepsy remains a clinically formidable neurological disorder with limited therapeutic options to control seizures or halt progression. Emerging evidence implicates neuroinflammatory processes in the pathogenesis of epilepsy25. Microglial activation and cytokine release disrupt neuronal excitability, accelerate neurodegeneration, impair blood–brain barrier (BBB) integrity26, and increase seizure susceptibility27.

In our study, IFITM3 functioned as a pro-inflammatory effector in the epilepsy model, promoting the secretion of IL-1β, IL-6, and TNF-α, potentiating apoptosis, and elevating pyroptosis-associated markers (ASC, IL-18). Mechanistically, STAT2 activated IFITM3 transcription by binding to the gene promoter, while let-7g-5p targeted STAT2 to inhibit IFITM3 expression and provide neuroprotection.

Pro-inflammatory cytokines induce IFITM3 expression, which enhances γ-secretase activity in neurons and astrocytes, driving excessive Aβ generation; in turn, Aβ activates microglia cGAS–STING–IFITM3 signaling to elevate neuroinflammation and AD risk28,29,30. Silencing IFITM3 attenuates Aβ accumulation, neuropathology, and cognitive impairments31,32. Likewise, ischemia selectively upregulates IFITM3 expression in microglia, precipitating glial activation, pro-inflammatory cytokine release, and neuronal apoptosis, which can be attenuated by IFITM3 silencing33,34. Consistent with previous descriptions of a pro-inflammatory function35,36, increased IFITM3 levels were detected in epilepsy models. IFITM3 not only promotes inflammation but also influences microglial phenotypes. It also shapes adaptive immunity by driving CD4+ T-helper cell differentiation and modulating IL-6/IL-10 signaling12,37. A polymorphism in IFITM3 correlates with Human Cytomegalovirus (HCMV) persistence and progression in Rasmussen’s encephalitis38, highlighting the need to balance anti-inflammatory interventions with the preservation of intrinsic antiviral defenses. Collectively, the interplay among IFITM3, neuroinflammation, and epilepsy-related cognitive impairment, as well as IFITM3’s involvement in diverse cell types, merit continued investigation.

Type I IFNs (IFN-α/β) exert antiviral effects by forming the IFN-stimulated gene factor 3 (ISGF3) complex, composed of STAT1, STAT2, and IRF9, which binds to IFN-stimulated response elements in the promoters of IFITM1/2/339. In our epilepsy models, upregulated STAT2 directly activates IFITM3 transcription, promoting inflammation and PCD. In chronic epilepsy, network transcriptomic analyses reveal a biphasic JAK/STAT activation, with an early transient induction after epileptogenic insult and a later resurgent activation coinciding with spontaneous seizures. Suppressing this second phase alleviates seizures, rescues cognitive deficits, and mitigates neuropathology40. Although this pattern was observed for JAK/STAT signaling, STAT2, as a downstream effector of interferon-mediated JAK signaling, may also exhibit stage-dependent effects, with early activation potentially neuroprotective and sustained activation promoting neuroinflammation and seizure perpetuation. Beyond temporal dynamics, STAT2 function is influenced by cell type, signal intensity, and inflammatory context. Moderate activation in astrocytes may support interferon-mediated homeostasis and neuronal resilience, whereas strong or sustained activation in microglia or neurons amplifies inflammatory gene expression. In AD, STAT2 can suppress pro-apoptotic signals via mTOR pathway modulation, reducing Bax and cleaved caspase-3 while increasing Bcl-217, whereas PSEN1 mutations impair STAT2 signaling, leading to inflammatory imbalance and increased pro-inflammatory susceptibility41. These findings highlight the context- and stage-dependent duality of STAT2, integrating interferon signaling, apoptosis regulation, and immune responses in neurological disorders.

In our experimental models of epilepsy, let-7g-5p exerts neuroprotective effects through direct targeting of STAT2, resulting in downregulation of IFITM3, and consequent attenuation of neuroinflammation, apoptosis, and pyroptosis. This aligns with the accumulating evidence that let-7 family members broadly regulate neuroimmune responses within CNS disorders42. Restoration of let-7a/c/g/i limits immune cell infiltration and cytokine release, and preserves BBB integrity43,44. In neurodegenerative diseases, let-7 miRNAs modulate microglial activation and autophagy45,46. Within epileptic contexts, let-7b targets STAT3 to inhibit glial activation and seizures47,48, and reduced let-7a-5p levels are associated with heightened inflammatory states in hippocampal sclerosis49. Conversely, increased let-7i-5p levels after status epilepticus has been proposed as a candidate biomarker of epileptogenesis50. Our findings expand this landscape by identifying let-7g-5p as a critical upstream suppressor of STAT2-mediated inflammatory signaling and underscoring its promise as a therapeutic target in epilepsy.

Several limitations should be acknowledged in this study. Although expression of related molecules was validated in animal experiments, neuroinflammation associated with epilepsy generates distinct cytokine and interferon profiles that differentially affect neuronal excitability and epileptogenesis. LPS mainly activates MyD88-dependent NF-κB signaling and induces pro-inflammatory cytokines (IL-1β, IL-6 and TNF-α), whereas viral stimuli such as poly(I: C), a TLR3 agonist, preferentially engage TRIF/IRF3 pathways and type I interferons (IFN-α/β)51,52. Consequently, employing viral stimuli in vitro and quantifying the levels of IFN-α, IFN-β and IFN-γ would likely yield novel insights. Moreover, the LPS-induced BV2 cell model cannot fully recapitulate the complex, chronic neuroinflammation observed in clinical epilepsy, and the limited sample size together with the lack of patient-derived specimens restrict the generalizability of our findings. Future studies should validate these results in larger cohorts using human brain tissue, cerebrospinal fluid, and peripheral blood from patients with epilepsy, as well as more advanced models such as human iPSC-derived microglia or brain organoids, to enhance clinical translatability. Finally, only male C57BL/6 mice were used to minimize variability from the estrous cycle and hormonal fluctuations53, and future studies including female mice are needed to assess potential sex-specific effects.

Conclusions

The study identified the let-7g-5p/STAT2/IFITM3 axis as a critical regulator of neuroinflammation, apoptosis, and pyroptosis in epilepsy models. IFITM3 reinforces the pro-inflammatory microenvironment and increases seizure susceptibility, potentially contributing to seizure recurrence and drug resistance. Neuroprotective effects of let-7g-5p were mediated via targeting STAT2, which resulted in suppressed transcription of IFITM3. Targeting this pathway may offer a promising therapeutic strategy for seizure control and drug-resistant epilepsy. Further research is warranted to explore its clinical potential.