Epilepsy, a chronic neurological disorder characterized by recurrent seizures due to aberrant neuronal hyperexcitability, affects more than 70 million individuals globally, and approximately 30% of patients are resistant to currently available therapies (Espinosa et al., 2025). While neuronal dysfunction remains a central focus in the study of epilepsy, glial cells—astrocytes and microglia—have emerged as pivotal drivers of epileptogenesis, orchestrating neuroinflammation, synaptic remodelling, and metabolic imbalance (Hui Yin et al., 2013; McCormick and Contreras, 2001; Grillo, 2014). Activated astrocytes disrupt glutamate homeostasis due to impaired transporter function, leading to excitotoxic neuronal damage, whereas microglia are polarized toward the proinflammatory (M1) phenotype and release cytokines (e.g., IL-1β and TNF-α) that exacerbate blood–brain barrier leakage and synaptic hyperexcitability (Ghouli and Binder, 2025; Chen et al., 2017). This glia-centric “seizure-inflammation-neuronal injury” axis underscores the urgent need for multitargeted interventions to disrupt the cycle of pathological events (Scheid and Teich, 2007; Alsharafi et al., 2015; Hennebelle et al., 2014).

Within this axis, noncoding RNAs—microRNAs (miRNAs) and long noncoding RNAs (lncRNAs)—serve as master regulators of glial plasticity. miRNAs, small (~22 nt) posttranscriptional silencers (Martino et al., 2025), directly modulate glial gene expression. Despite increasing insights into the individual functions of miRNAs and lncRNAs, their synergistic regulation of glial crosstalk in epilepsy remains unknown. Previous studies have differentiated these mechanisms—focusing solely on miRNA-mediated neuronal excitability or lncRNA-driven blood–brain barrier (BBB) disruption—while neglecting the integrated glial network. This review is the first to delineate the cooperative roles of miRNAs and lncRNAs in glial-mediated epileptogenesis through three mechanisms: (1) Polarization control, i. e., bidirectional regulation of astrocyte A1/A2 and microglial M1/M2 polarization (e.g., miR-146a/Notch-1 vs. lncRNA SNHG5/NF-κB); (2) Metabolic rewiring, i. e., the modulation of glutamate uptake (miR-181c-5p/GLT-1) and oxidative stress (lncRNA UCA1/Nrf2) through miRNA–lncRNA networks; (3) Exosomal communication, i.e., propagation of inflammation through neuronal-glial communication via extracellular vesicles (EVs) carrying miR-155 or lncRNA ILF3-AS1.

By integrating the function of glia with that of noncoding RNAs, this work provides a transformative perspective for targeting the glial-RNA axis in drug-resistant epilepsy, bridging molecular discovery to therapeutic innovation.

As the most numerous glial cell population in the central nervous system (CNS), astrocytes are widely distributed in normal neural tissues and play several key physiological regulatory roles. In the brain, these cells not only participate in the regulation of neurohomeostasis by maintaining neuronal structural integrity and promoting functional activities, but also play important roles in the regulation of ionic homeostasis, energy metabolism, synaptic network assembly and neurotransmitter transmission (Murphy-Royal et al., 2015; Eberhard et al., 2025). Notably, this class of glial cells with immunomodulatory properties has shown significant pathological relevance in several human diseases, and in particular their involvement in epileptic pathomechanisms by mediating the neuroimmune response has been confirmed by several studies (Shi et al., 2025). In vitro experimental data have shown that astrocytes in the activated state have significant inflammatory factor-secreting properties, a finding that is consistent with the abnormally high expression of pro-inflammatory mediators, such as interleukin-1β (IL-1β) and tumour necrosis factor-α (TNF-α), in experimental epilepsy models and epileptogenic human brain tissue (Zhou et al., 2018). Recent research has shown that astrocyte dysfunction is closely associated with the pathological onset and progression of epilepsy, suggesting a central regulatory role in the pathogenesis of the disease (Shen et al., 2016). Some researchers have suggested that regulating the polarisation status of astrocytes may be a novel intervention strategy for the treatment of refractory epilepsy, and this therapeutic concept may also be applicable to other brain injury disorders involving glial cell polarisation (Zhang et al., 2022). Abnormally activated astrocytes in pathological states have been shown to exacerbate neuroinflammatory responses through the sustained release of pro-inflammatory cytokines. Experimental evidence suggests that this hyperactivated state is an important driver of inflammation-mediated neuronal degeneration (Devinsky et al., 2013). Moreover, it has been demonstrated that inflammatory over activation of astrocytes can result in impaired glutamate transport, which, in turn, can trigger abnormal neuronal firing and ultimately induce epileptiform seizures (Sanz and Garcia-Gimeno, 2020). More information regarding the mechanisms by which miRNAs regulate astrocyte function is shown in Figure 1 and Table 1.

In the regulatory network of neuroinflammation, microglia act as core effector cells, and their phenotypic polarization process is precisely regulated by miRNAs and long chain non-coding RNAs (lncRNAs) (Terrone et al., 2017; Peng et al., 2019). These immunoreactive cells not only possess the function of secreting multiple cytokines/chemokines, but also sense the dynamic changes of inflammatory signaling molecules within the CNS (Hanisch, 2002). After brain injury of epileptic origin, these cells become a major source of pro-inflammatory cytokine secretion, and the inflammatory mediators they release not only enhance the neuronal excitability threshold, but also directly contribute to the development of abnormal discharge activity, which has been recognized as a key driver of epilepsy formation (Colonna and Butovsky, 2017). Microglia in the CNS are highly plastic and are capable of polarizing toward different phenotypes such as classically activated (M1) type and alternatively activated (M2) type (Ma et al., 2017). Among them, the M1 type exacerbates the neuroinflammatory process by releasing pro-inflammatory mediators (Kalkman and Feuerbach, 2016), whereas the M2 type secretes immunosuppressive factors, such as TGF-β and IL-10, and is involved in the regulation of glioma cell proliferation (Lan et al., 2017; Wesolowska et al., 2008). Studies have shown that in the hypoxic microenvironment of glioblastoma (GBM), tumor cells can induce tumor-associated macrophages to polarize toward the M2 type through the secretion of paracrine factors such as periostin (POSTN) and exosomes (Guo et al., 2016). In terms of miRNA regulatory mechanisms, miR-181c-5p effectively inhibited microglia overactivation in a sepsis model by targeting high mobility group protein B1 (HMGB1), while decreasing the levels of inflammatory factors, such as TNF-α and IL-1β, and decreasing hippocampal neuronal apoptosis (Li et al., 2021). These findings reveal the multidimensional regulatory properties of miRNAs in epilepsy pathology, and their specific mechanisms of action and potential as therapeutic targets will be systematically described below.

Mechanistically, long noncoding RNAs (lncRNAs) serve as competitive endogenous RNAs (ceRNAs) by sequestering shared miRNAs that would otherwise target mRNAs, thereby influencing disease pathogenesis through posttranscriptional regulation (Tan and Marques, 2016). Within the CNS, microglia and astrocytes demonstrate functional plasticity through polarized activation states: The pro-inflammatory M1/A1 phenotypes drive neurotoxicity via inflammatory mediator release and oxidative damage, while the M2/A2 phenotypes exhibit neuroprotective capacities through anti-inflammatory actions and synaptic maintenance (Kwon and Koh, 2020). Emerging research identifies lncRNAs as upstream epigenetic regulators of miRNAs, positioning them as potential diagnostic biomarkers and therapeutic targets for neurological disorders such as Alzheimer’s disease, ischemic cerebrovascular events, and demyelinating conditions (Khan et al., 2022). Despite these advances, the precise regulatory networks through which lncRNAs govern neuroinflammatory processes via glial cell modulation in epilepsy remain poorly characterized. There exists a critical need for comprehensive analysis of lncRNA-mediated regulatory circuits during epileptogenesis, particularly those involving glial activation and subsequent inflammatory pathway dysregulation. Current understanding of lncRNA-mediated control over microglial and astrocytic functionality is summarized in Table 2 and Figure 3, and how lncRNAs influence miRNAs is summarized in Figure 4, revealing promising avenues for future mechanistic investigations.

LncRNA X-inactive-specific transcript (XIST): The X-inactive specific transcript (XIST), a long noncoding RNA encoded on the X chromosome, has emerged as a key epigenetic regulator of inflammatory pathways across multiple disease states (Zhang Y. et al., 2019). Mechanistic studies reveal XIST functionally sequesters miR-29c-3p, a miRNA notably downregulated in epileptic rat models and demonstrating an inverse correlation with XIST levels (Zhang et al., 2021). Functional experiments demonstrate that miR-29c-3p overexpression suppresses proinflammatory cytokine secretion in BV2 microglial cells, effectively counteracting lipopolysaccharide (LPS)-induced inflammatory cascades (Wang R. et al., 2020). Of particular therapeutic interest, this XIST/miR-29c-3p axis has been implicated in controlling astrocytic polarization toward the neurotoxic A1 phenotype, positioning these molecules as promising therapeutic candidates for epilepsy intervention (Zhang et al., 2021; Table 3).

LncRNA Peg13: The 26S proteasome non-ATPase regulatory subunit 11 (Psmd11) serves as a pivotal coordinator of cellular stress adaptation and inflammatory modulation (Wang et al., 2018; Wei et al., 2019). Experimental evidence from rodent epilepsy models reveals that hippocampal Psmd11 expression undergoes coordinated regulation through lncRNA Peg13-mediated molecular interactions. Functioning as a competitive endogenous RNA, Peg13 binds and sequesters miR-490-3p, thereby alleviating its inhibitory effect on Psmd11 transcripts. This post-transcriptional derepression elevates Psmd11 levels, which in turn: (1) Inhibits Wnt/β-catenin signaling-mediated A1 astrocyte polarization. (2) Attenuates neuroinflammatory cascades.

The resultant suppression of epileptogenic processes highlights the Peg13/miR-490-3p/Psmd11 axis as both a disease-promoting network and a viable therapeutic target for seizure disorders (Hodges and Lugo, 2018). Pharmacological validation studies further confirm its dual role in epilepsy pathophysiology and treatment development (Feng et al., 2020; Wang et al., 2018).

LncRNA ILF3-AS1: Matrix metalloproteinases (MMPs) constitute pivotal mediators of epileptogenic processes through their proteolytic regulation of neurovascular integrity. Pathological upregulation of MMP activity post-injury drives epileptogenesis via tripartite mechanisms: (1) extracellular matrix degradation, (2) BBB compromise, and (3) neuroimmune activation, as evidenced by elevated MMP2/3/9 levels in both TLE patients and experimental models (Dubey et al., 2017; Korotkov et al., 2018; Rempe et al., 2018). This mechanistic understanding positions MMP inhibitors as promising disease-modifying therapies for refractory epilepsy (Dubey et al., 2017).

Emerging evidence implicates the long noncoding RNA ILF3-AS1 as a master regulator of MMP-mediated epileptogenic processes in TLE pathogenesis. Multi-omics analyses demonstrate consistent ILF3-AS1 upregulation coupled with miR-212 downregulation in hippocampal tissues and serum samples from TLE patients versus controls (Cai et al., 2020). Functional studies reveal ILF3-AS1 exacerbates neuroinflammation and matrix remodeling through dual epigenetic mechanisms: (1) Sponging miR-212 to derepress MMP3/9 expression; (2) Sequestering miR-504-3p to amplify HMGB1-mediated oxidative stress.

These synergistic effects promote neuronal hyperexcitability in both in vitro epileptiform models (magnesium-deprived hippocampal neurons) and in vivo TLE paradigms (Gao et al., 2024; Cai et al., 2020).

LncRNA UCA1: the long noncoding RNA UCA1 demonstrates antiepileptic effects through multimodal suppression of astrocyte activation and neuroinflammation. Its protective mechanisms involve three primary pathways: (1) miR-203/MEF2C/NF-κB Axis: Myocyte enhancer factor 2c (MEF2C), a MADS/MEF2 family transcription factor critical for epilepsy regulation, modulates both neuronal excitability/synaptic plasticity and neuroinflammatory processes via inflammatory mediators and signaling networks (Xu et al., 2015). UCA1 forms a regulatory circuit by suppressing miR-203 (which directly inhibits MEF2C) (Luo et al., 2014; Gong et al., 2018), thereby enhancing MEF2C expression. This upregulation attenuates NF-κB pathway activity, limiting A1 astrocyte activation and subsequent release of proconvulsive factors like IL-1β that promote neuronal hyperexcitability (Bellot-Saez et al., 2017; Yu et al., 2020). (2) miR-495/Nrf2 Pathway: UCA1 mitigates seizure-associated neuronal apoptosis and cerebral damage via modulation of the miR-495/Nrf2 signaling axis (Geng et al., 2018). (3) JAK/STAT3 Signaling: Through inhibition of JAK/STAT3 activation, UCA1 downregulates astrocytic glutamate transporters (e.g., GLAST), reducing glial hyperactivation and neuroinflammatory cascades (Raymond et al., 2011; Wang H. et al., 2020).

Collectively, UCA1 counteracts epileptogenesis by targeting these interconnected pathways to suppress A1 astrocytosis, neuronal death, and neuroinflammation. These insights position UCA1 as a promising therapeutic target and underscore the potential of lncRNA-based strategies for epilepsy intervention.

miR-155hg: the long noncoding RNA miR-155hg serves as the precursor of miR-155, with its second exon directly encoding this miRNA. Both miR-155hg and miR-155 are critically involved in CNS disorders (Korotkov et al., 2020). miR-155hg is predominantly expressed in macrophages and microglia, where it plays a central role in maintaining brain homeostasis and modulating neuroinflammation. During convulsive status epilepticus (CSE), microglial phagocytic activity is essential for responding to neuronal injury and preserving brain homeostasis. Studies demonstrate that miR-155hg inhibits microglial phagocytic function, thereby delaying the clearance of damaged neurons and exacerbating hippocampal neuronal injury (Wang et al., 2024). Suppression of miR-155hg expression enhances microglial phagocytic activity, promotes neuronal repair, and reduces neuroinflammatory responses through regulation of miR-155 signaling, suggesting therapeutic potential for mitigating CSE progression (Wang et al., 2024).

LncRNA H19: functioning through immunoinflammatory and neuronal injury-associated pathways, H19 drives neuroglial hyperactivation via NF-κB-mediated inflammatory cascades (Han et al., 2017; Han et al., 2018b). In GBM tissues, this lncRNA promotes malignant phenotypes including glioma cell proliferation, migration, and invasion (Jiang et al., 2016). Within epileptic contexts, H19 exacerbates disease progression by dual mechanisms: (1) activating the JAK/STAT pathway to induce astrocytic GFAP expression and A1 polarization (Xu et al., 2011), and (2) enhancing microglial M1 polarization (evidenced by elevated CD86/iNOS) and astrocyte activation, which collectively amplify proinflammatory cytokine release (e.g., IL-1β, TNF-α). This creates a self-reinforcing cycle of gliosis, neuroinflammation, and hippocampal neuronal apoptosis, ultimately worsening seizure pathology (Han et al., 2018a; Vezzani and Friedman, 2011).

Experimental validation demonstrates that H19 knockdown suppresses NF-κB phosphorylation/nuclear translocation, reduces M1 marker expression, decreases neuronal apoptosis, and mitigates CSE-induced brain damage (Limanaqi et al., 2019). These findings establish H19 as a central regulator of TLE pathogenesis through glial phenotype modulation and NF-κB-dependent inflammation. Therapeutic targeting of H19 or its downstream effectors may disrupt this pathogenic network by restoring anti-inflammatory microenvironments and preserving neuronal integrity.

LncRNA Snhg5: the long noncoding RNA Snhg5, whose expression is dysregulated in inflammatory disorders, contributes to epileptogenesis through neuroinflammatory modulation (Shen et al., 2020). In vitro studies confirm that lipopolysaccharide (LPS) activates NF-κB signaling via p65 phosphorylation and nuclear translocation, facilitating proinflammatory gene transcription—consistent with established mechanisms (Sun et al., 2018b; Davari et al., 2020). Critically, Snhg5 silencing attenuates LPS-induced NF-κB pathway activation and exerts anticonvulsant effects by reprogramming microglial polarization. This dual modulation involves: (1) reducing proinflammatory M1 polarization (marked by CD86/iNOS downregulation) and associated cytokine release (IL-1β, TNF-α). (2) Enhancing anti-inflammatory M2 polarization (indicated by Arg-1/CD206 upregulation) and protective factor expression (e.g., IL-10). These shifts remodel the neuroinflammatory milieu, alleviating epilepsy-associated neuronal injury (Wang et al., 2022). Collectively, Snhg5 emerges as a pivotal regulator of microglial M1/M2 equilibrium via NF-κB signaling, highlighting lncRNA-targeted strategies to mitigate glia-driven neuroinflammation.

LncRNA NEAT1: the long noncoding RNA NEAT1, a critical regulator of paraspeckle formation in mammalian nuclei, plays a multifaceted role in glial-mediated neuroinflammation and neurological disorders through its functional dysregulation (Katsel et al., 2019). In oligodendroglial cells, NEAT1 maintains neural homeostasis by orchestrating myelin-related gene expression, with its aberrant activity linked to schizophrenia pathogenesis (Katsel et al., 2019). Within epileptic contexts, NEAT1 exacerbates neuroinflammatory responses and glial activation via two principal mechanisms: (1) Inflammasome activation: NEAT1 enhances NLRP3/NLRC4 inflammasome activity, triggering pyroptotic cell death and proinflammatory mediator release (IL-6, COX-2, TNF-α), which collectively induce oxidative stress and mitochondrial dysfunction (Zhang P. et al., 2019); (2) microglial polarization: by sequestering miR-129-5p to derepress Notch signaling, NEAT1 drives microglial M1 polarization, amplifying IL-1β secretion and aggravating seizure-associated neuroinflammation and neuronal injury (Wan and Yang, 2020). These coordinated actions position NEAT1 as a promising therapeutic target for epilepsy, given its dual regulatory capacity over glial functions (myelination regulation, phenotypic polarization) and inflammatory network modulation.

LncRNA SNHG1: small nucleolar RNA host gene 1 (SNHG1), a chromosome 11-derived lncRNA critical for 18S rRNA processing (Huang L. et al., 2018), demonstrates neuroprotective properties in cerebral ischemic pathologies (Zhang et al., 2018). Mechanistically, SNHG1 overexpression modulates the transcription factor YY1 (implicated in acute ischemic hippocampal injury) (Wan et al., 2019) through miR-186-5p sequestration, achieving dual immunoregulatory outcomes: (1) Suppression of proinflammatory mediators (TNF-α, IL-1β). (2) Enhancement of anti-inflammatory IL-10 production via microglial phenotype regulation. These effects collectively mitigate hypoxia-induced neuronal damage while preserving M2 microglial function. Notably, nanoengineered delivery systems for SNHG1 mimics present a novel therapeutic strategy for ischemia-related epilepsy (Zhang et al., 2018).

The emerging evidence collectively underscores that lncRNAs orchestrate epileptogenic processes through convergent mechanisms: (1) ceRNA-mediated miRNA sequestration (e.g., XIST/miR-29c-3p, Peg13/miR-490-3p, ILF3-AS1/miR-212), which derepresses downstream targets to drive glial polarization; (2) inflammatory pathway amplification via NF-κB (H19, Snhg5), JAK/STAT (H19), and inflammasome activation (NEAT1); and (3) dual regulation of neuronal excitability and neurovascular integrity through MMPs (ILF3-AS1), glutamate transporters (UCA1), and oxidative stress responses (UCA1/miR-495/Nrf2). Crucially, these lncRNAs form interconnected axes—such as NEAT1/miR-129-5p/Notch in microglial M1 polarization and H19/NF-κB in astrocytic A1 conversion—that create self-reinforcing cycles of gliosis and neuroinflammation. Their consistent dysregulation in epileptic tissues (e.g., ILF3-AS1↑ in TLE hippocampi, miR-155hg↑ in CSE) positions them not only as biomarkers but as master regulators of the “glial-RNA axis.” Therapeutic targeting of these nodes—via antisense oligonucleotides against H19/NEAT1 or engineered exosomes delivering UCA1/SNHG1 mimics—represents a promising paradigm for multitargeted intervention in drug-resistant epilepsy.

Neuroinflammation constitutes a pivotal pathological mechanism in epilepsy, where inflammatory cascades frequently drive disease progression and seizure recurrence. Targeting neuroinflammatory pathways represents a promising disease-modifying strategy, supported by (1) demonstrated efficacy of anti-inflammatory drugs in drug-resistant epilepsy (Vezzani et al., 2019), (2) preclinical evidence showing reduced seizure frequency and enhanced neuroprotection via anti-inflammatory interventions (Terrone et al., 2020), and (3) regulatory potential of noncoding RNAs (miRNAs/lncRNAs) in inflammatory modulation. However, therapeutic application of free miRNAs faces challenges including (1) rapid degradation by serum nucleases and lysosomal pathways, (2) limited BBB permeability, and (3) poor target specificity.

Exosomes (30–100 nm vesicles) address these limitations through (1) endogenous phospholipid bilayer protection against enzymatic degradation, (2) bidirectional BBB penetrability without structural modification, and (3) natural tropism via membrane-bound ligands (Shyam et al., 2025). Clinically, exosomal biomarkers exhibit diagnostic value through (1) elevated miR-134 levels in TLE serum, exacerbating neuronal hyperexcitability via LIMK1 inhibition (Jimenez-Mateos et al., 2012), (2) upregulated miR-146a/miR-155 correlating with neuroinflammation severity and Electroencephalogram abnormalities (Omran et al., 2012; Liu et al., 2022), and (3) downregulated forebrain miR-346/miR-331-3p levels (Gitaí et al., 2020). Therapeutically, exosome advantages over synthetic nanoparticles include (1) reduced immunogenicity and macrophage clearance, (2) prolonged circulatory half-life, and (3) versatile cargo-loading capacity for proteins/nucleic acids (Xian et al., 2019; Ou et al., 2020). These dual diagnostic-therapeutic properties position exosomes as transformative tools for precision epilepsy management.

Inhibiting the expression of pathogenic miRNAs is a promising therapeutic strategy for example. For example, the delivery of miR-134 antisense oligonucleotides (antagomirs) via exosomes significantly reduces spontaneous seizures and protects hippocampal neurons from damage in an epilepsy mouse model (Jimenez-Mateos et al., 2015b). Similarly, exosome-mediated delivery of inhibitors of miR-155 inhibits microglial activation and attenuates epilepsy-associated inflammation (Liu et al., 2022). In addition, measuring miR-155 levels in serum exosomes may be used to diagnose epilepsy in early stages to assess epilepsy severity (Liu et al., 2022). Levetiracetam also inhibits M2 polarization of microglia by blocking abnormal neuronal activation and reducing miR-200c-3p levels in exosomes, which inhibits the protumorigenic effects of M2 microglia (Guo et al., 2024). Exosomes, as natural therapeutic vectors with an excellent ability to cross the BBB, can be used to efficiently deliver therapeutic miRNAs. miR-142-3p, miR-223-3p and miR-21-5p levels were found to be significantly increased in epileptogenic TSC lesions and to contain nucleic acid motifs that activate toll-like receptors (TLR7/8), allowing them to activate a neuroinflammatory response. These results provide new evidence for the role of exosomes and noncoding RNA cargo in the neuroinflammatory cascade in epilepsy and may help advance the development of novel biomarkers and therapeutics for refractory epilepsy (Cukovic et al., 2021). Liu et al. (2024) demonstrated that exosomes loaded with miR-129-5p significantly attenuate neurodegeneration in a mouse model of persistent epilepsy and reduce neuronal damage in the CA3 region in the epileptic brain by inhibiting HMGB1/TLR4-mediated neuroinflammation. miR-23b-3p interacts with the 3’-UTRs of STAT1 and GlyR1 to inhibit inflammatory factor expression in M1 microglia, and miR-23b-3p -loaded exosomes derived from adipose-derived stem cells (ADSCs) alleviate KA-induced inflammation in mice with epilepsy (Yang et al., 2024). Although exosomal miRNAs show great potential in treating epilepsy, there are still many related challenges, including the need to standardize of exosome sources, improve target delivery efficiency, and assess long-term safety. Future studies involving the use of multiomics techniques to screen specific miRNAs and optimize exosome engineering strategies are needed to facilitate clinical translation.

Although important progress has been made in understanding the mechanisms by which miRNAs and lncRNAs regulate glial cell function and neuroinflammation in epilepsy, the clinical translation of related therapies still faces several challenges. First, current evidence on glial ncRNA mechanisms relies predominantly on mouse models, limiting the translatability of glial findings to humans. Expanding investigations to diverse animal systems is imperative to validate pathophysiological relevance. Second, miRNAs and lncRNAs act through a complex multi-target network, and the inhibition of glial inflammation may also inhibit the phagocytosis of glial cells, which is not conducive to the control of epilepsy. Therefore, the mechanism of action of these molecules should be further explored (Huang et al., 2015). Third, although exosomes can penetrate the BBB, they have low yields and unstable drug delivery efficiencies, and free miRNAs are easily degraded in body fluids. The use of exosomes for therapeutic delivery remains a bottleneck (Wortzel et al., 2019; Liu et al., 2024).