Exosomal miR-149 from human umbilical cord mesenchymal stem cells attenuates spinal cord injury-induced blood-spinal cord barrier disruption by suppressing the ET-1/PI3K/Akt signaling pathway

Abstract

Background

Spinal cord injury (SCI) leads to persistent neurological deficits partly by disruption of the blood-spinal cord barrier (BSCB). Small extracellular vesicles (sEVs) from human umbilical cord mesenchymal stem cells (hUC-MSCs) can promote BSCB repair, but their active components remain unclear. This study examined whether miR-149 carried by hUC-MSC-derived sEVs (hUC-MSCs-sEVs) protects the BSCB after SCI by targeting endothelin-1 (ET-1).

Methods

Human brain microvascular endothelial cells (HBMECs) were subjected to oxygen–glucose deprivation/reoxygenation (OGD/R) to model barrier injury, and rats underwent a thoracic SCI. hUC-MSCs-sEVs were isolated and loaded with miR-149 mimics or inhibitors. Endothelial cell viability, paracellular permeability (FITC-dextran assay), and junction protein levels (ZO-1, Claudin-5, β-Catenin, Occludin) were measured by viability assays, Western blot, and immunofluorescence. ET-1 levels and PI3K/Akt pathway activation were measured by ELISA and Western blot. In SCI rats, sEVs (with or without the miR-149 inhibitor) were injected; motor function (BBB locomotor score), BSCB permeability (Evans blue/FITC-dextran leakage) and spinal cord histology were evaluated.

Results

hUC-MSCs-sEVs were internalized by HBMECs and significantly improved cell survival and barrier function after OGD/R. sEVs treatment restored tight and adherens junction proteins and suppressed OGD/R-induced ET-1 upregulation and PI3K/Akt activation. OGD/R reduced miR-149 expression, which was rescued by sEVs. sEVs loaded with miR-149 mimic further enhanced these protective effects, whereas a miR-149 inhibitor abolished them. Notably, co-administration of an ET-1 receptor antagonist reversed the barrier disruption caused by miR-149 inhibition. In vivo, hUC-MSCs-sEVs treatment improved locomotor recovery and reduced BSCB leakage and tissue damage, whereas miR-149 inhibition abolished these benefits.

Conclusions

hUC-MSC-derived exosomal miR-149 preserves BSCB integrity and promotes functional recovery after SCI by targeting ET-1 and inhibiting the PI3K/Akt pathway, thereby enhancing junctional protein expression. The miR-149/ET-1 axis may represent a promising therapeutic target for SCI.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s13287-025-04873-7.

Introduction

Spinal cord injury (SCI) is a severe disorder of the central nervous system that leads to persistent sensorimotor deficits [1, 2]. SCI involves two phases of injury: primary and secondary. Primary injury results from mechanical trauma, causing irreversible damage to neurons and axons. Secondary injury includes inflammation, oxidative stress, apoptosis, and disruption of the BSCB. These processes occur hours to weeks post-injury and further exacerbate neurological deficits [3]. Critically, this secondary phase involves a robust inflammatory cascade characterized by early release of pro-inflammatory cytokines (e.g., TNF-α, IL-1β, IL-6) from activated glial and immune cells [4, 5]. These inflammatory mediators not only inflict neural damage but also compromise the BSCB, leading to vasogenic edema as plasma proteins and fluid leak into the spinal cord parenchyma [4, 6–8]. Concurrently, astrocytes become reactive (astrogliosis) and proliferate to form a glial scar that helps contain the lesion and restore barrier function, although persistent scar formation can later inhibit regeneration [4, 9]. Astrogliosis can restrict inflammation and actively remove edema fluid via astrocytic aquaporins (e.g., AQP4), thereby limiting further tissue damage [10]. Thus, a dynamic interplay ensues between inflammation-driven injury (vasogenic edema) and astrocyte-mediated repair (glial scarring), both of which critically influence BSCB integrity and functional outcomes. Given the central role of neuroinflammation in SCI pathogenesis, immunomodulatory and anti-inflammatory strategies are being actively explored to limit secondary tissue damage [11]. For instance, high-dose glucocorticoids (e.g., methylprednisolone) have been used to suppress inflammation and stabilize the BSCB, thereby reducing vasogenic edema and secondary injury, although their benefit remains debated [12, 13]. Beyond corticosteroids, emerging evidence shows that novel cytokine inhibitors also offer therapeutic potential in SCI. For example, the JAK1/2 inhibitor tofacitinib significantly improved motor recovery and reduced microglial pro-inflammatory polarization via the JAK/STAT pathway in a rat SCI model [14]. Similarly, the anti-IL-6R antibody tocilizumab enhanced tight junction protein (ZO-1, CLDN5) expression and reduced BSCB leakage in a mouse SCI model, supporting vascular repair by cytokine blockade [15]. These studies highlight the importance of targeting specific inflammatory signalling axes (e.g., IL-6, JAK/STAT) in addition to generic anti-inflammatory treatment to protect the BSCB and recover spinal cord function. These anti-inflammatory strategies underscore the importance of targeting inflammation to preserve BSCB integrity and mitigate edema.

The BSCB, formed by tightly junctioned endothelial cells, functions as a robust barrier that effectively impedes paracellular and transcellular transport, preventing entry of external harmful substances into the central nervous system and thus preserving spinal cord homeostasis [16]. Evidence indicates that BSCB disruption following SCI elevates vascular permeability, facilitating infiltration of inflammatory mediators and immune cells into the spinal cord. This infiltration provokes ischemia, edema, and free radical release, establishing a deleterious cycle that culminates in irreversible neural tissue damage [17]. Regrettably, the precise mechanisms underlying BSCB disruption after SCI remain incompletely elucidated, with a paucity of targeted therapeutic interventions.

In recent years, stem cell therapy has gained prominence owing to its neuroprotective and regenerative capabilities. hUC-MSCs are particularly favored due to their abundant availability and minimal ethical concerns. Investigations have demonstrated that hUC-MSC transplantation enhances neurological outcomes post-SCI by secreting neurotrophic factors, modulating immune responses, and promoting angiogenesis [18]. Nonetheless, stem cell therapies are constrained by challenges including low in vivo survival rates, risks of immune rejection, and potential tumorigenicity, which hinder their broad implementation [19].

Small extracellular vesicles (sEVs) are natural nanoscale messengers encapsulating non-coding RNAs, mRNAs and functional proteins; operationally defined as vesicles with a predominant size distribution up to approximately 200 nm in diameter, in line with current EV research guidelines [20, 21]. These vesicles shuttle intercellular signals, influencing target cell fates [22]. Compared to their parent cells, stem cell-derived sEVs have greater stability, lower immunogenicity, and better barrier penetration, making them promising cell-free therapeutics [23]. Previous studies show that hUC-MSCs-sEVs demonstrate reparative efficacy for BSCB integrity following SCI in both in vitro and in vivo models [24, 25], with therapeutic outcomes intimately linked to their bioactive cargo, particularly miRNAs [26].

ET-1 is a potent vasoconstrictor that influences BSCB disruption and repair [27, 28]. Our previous research confirmed that hUC-MSCs-sEVs can repair BSCB damage after SCI by downregulating ET-1 expression [29]. However, the specific constituents within sEVs responsible for this effect and their precise targeting mechanisms remain undefined. Accordingly, this study, grounded in bioinformatics analyses, concentrates on the function and mechanisms of miR-149 within hUC-MSCs-sEVs in protecting BSCB integrity through targeted regulation of ET-1 at both cellular and animal levels.

Materials and methods

Cell culture

The hUC-MSCs were purchased from Fuyuan Biotechnology (Fuyuan Biotechnology Co., Ltd. Shanghai, CHINA). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco, NY, USA) supplemented with 10% fetal bovine serum (FBS, Gibco, NY, USA) and 1% penicillin/streptomycin (Thermo Fisher Scientific, Waltham, MA, USA). HBMECs were purchased from Meisen Cell Technology (Meisen Cell Technology, Zhejiang, China). Cells were cultured in an endothelial cell medium (ScienCell Research Laboratories, San Diego, CA, USA) and incubated in a humidified atmosphere at 5% CO2 and 37 °C.

sEVs isolation and characterization

We adhered to the guidelines for the isolation, characterization, and functional analysis of EVs as stipulated in a consensus document published by the International Society of Extracellular Vesicles [30]. Extracellular vesicles were harvested from passage (P) 3 to P5 of the hUC-MSCs. For the production of conditioned medium, hUC-MSCs were seeded at a density of 5 × 10^5 cells per 75 cm² flask and allowed to grow to approximately 80% confluence. The culture medium was then replaced with DMEM supplemented with 10% exosome-depleted FBS (prepared by ultracentrifugation of standard FBS at 110,000 × g for 18 h) and 1% penicillin/streptomycin. The cells were conditioned in this medium for 48 h at 37 °C in a 5% CO₂ atmosphere. To isolate sEVs, the hUMSC-conditioned medium was first centrifuged at 500 × g for 10 min to remove cells. Subsequently, the supernatant was centrifuged at 10,000 × g for 30 min to eliminate apoptotic vesicles and other debris. The resulting liquid was then filtered through a 0.22 µm filter. The sEVs were then collected as a pellet using ultracentrifugation (Beckman Optima XPN, 45Ti) at 110,000 × g for 70 min. The sEVs pellet was resuspended in phosphate-buffered saline (PBS) for purification and subjected to another round of ultracentrifugation at 110,000 × g for 70 min to remove the contaminating proteins. Finally, the sEVs were resuspended in PBS. The Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) was used to assess the protein content of the sEVs. The hUC-MSCs-sEVs sample was stored at − 80 °C for further analysis. The size of sEVs was determined by nanoparticle tracking analysis (NTA) using ZetaView S/N 17–310 (Particle Metrix, Meerbusch, Germany) along with its associated software. Additionally, transmission electron microscopy (TEM; JEOL Ltd., Tokyo, Japan) was used to morphologically examine isolated sEVs. Western blot analysis was utilized to determine the levels of CD9 (Abcam, Cambridge, UK) and TSG101 (Abcam, Cambridge, UK) in sEVs.

Preparation of miRNA-loaded hUC-MSCs-sEVs

The protocol for loading sEVs with miRNAs was adapted from established methods [31–33]. Briefly, miR-149 mimic and miR-149 inhibitor (alongside their respective negative controls: NC mimic and NC inhibitor) were synthesized with cholesterol conjugated to the 5′ sense strand and modified with 2′-deoxy-2-fluoro-Cy3 (GenePharma, Shanghai, China). Cholesterol-conjugated miRNAs (100 nM) were incubated with 50 µg of hUC-MSCs-sEVs in 100 µL of phosphate-buffered saline (PBS) at 37 °C for 1 h to facilitate miRNA loading. Following incubation, the mixture was centrifuged at 100,000 × g for 70 min at 4 °C to pellet miRNA-loaded sEVs. The pellet was resuspended in fresh PBS, and the sEVs-miR-149 mimic, sEVs-miR-149 inhibitor, or control sEVs (sEVs-NC mimic, sEVs-NC inhibitor) were either used immediately for subsequent experiments or stored at − 80 °C. The loading efficiency of miRNAs into sEVs was confirmed by extracting RNA from each group of engineered sEVs and performing qPCR analysis for miR-149.

Experimental animals

The work has been reported in line with the ARRIVE guidelines 2.0. All animal experimental protocols conformed to the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health (NIH Publications No. 8023), and all procedures were approved by the Shanxi Provincial People’s Hospital Institutional Animal Care and Use Committee (Approval No. 2022-089). Adult female Sprague–Dawley rats weighing between 220 and 250 g were obtained from the Laboratory Animal Center of Shanxi Cancer Institute (animal production certificate # SCXK (Jin) 2017-0001; Shanxi, China). The entire experimental process was conducted at the same institution (animal usage certificate # SYXK (Jin) 2017-0003; Shanxi, China). Rats were housed in pathogen-free environments, with two to three animals per cage. They were provided with a standard commercial diet, had ad libitum access to water, and maintained under control humidity (40–60%) in a 12-hour light-dark cycle. At the experimental endpoints, euthanasia was performed by intraperitoneal injection of an overdose of sodium pentobarbital (150 mg/kg) at the Laboratory Animal Center of Shanxi Cancer Institute. Death was confirmed by the absence of spontaneous breathing and heartbeat, followed by cervical dislocation as a secondary physical method to ensure death, in accordance with AVMA guidelines for rodent euthanasia. The rats were randomly assigned to the following four groups: control (n = 20), SCI (n = 20), sEVs (n = 20), and sEVs-miR-149 inhibitor (n = 20) groups. All animal procedures were carried out by a skilled technician who was blinded to group allocation and unable to influence whether an animal would receive PBS, hUC-MSCs-sEVs, or sEVs-miR-149 inhibitor treatment.

Spinal cord injury and treatment

Rats were anesthetized with 1% sodium pentobarbital (3 mL/kg, i.p.), and a median dorsal incision was performed at the T10 segment. The surrounding tissues were carefully dissected to expose the T10 vertebral body, spinous process, and spinal cord. The muscles were dissected layer by layer while preserving the integrity of the dura mater. In the sham-operated group, the wound was sutured layer by layer after sterilization. A spinal cord injury model was established using the modified Allen’s method for the remaining three groups [34]. The spinal cord injury was created at the T10 level using a standardized force (10 g×5 cm), and the successful modeling was confirmed by the presence of congestion, edema, double hind limb convulsions, and spastic tail swing at the site of injury in the spinal cord.

Treatment

Rats subjected to SCI were randomly divided into several groups and administrated with hUC-MSCs-sEVs or sEVs-miR-149 inhibitor, or an equal volume of PBS (200 µL) through the tail-vein injection post-injury as we previously described [29]. Specifically, the hUC-MSCs-sEVs were administered at a dose of 200 µg of sEVs protein in 200 µL PBS per rat per injection. The injections were performed immediately after the induction of SCI (0 days post-injury, dpi) and subsequently at 1 dpi and 2 dpi. All rats were given intraperitoneal penicillin (200,000 U/d) for 3 days, while their bladder was manually massaged twice or thrice daily to facilitate urination until the urinary function was restored.

Behavioral tests

The Basso, Beattie, and Bresnahan (BBB) ratings were used to assess the functional impairments following SCI. Two independent examiners blinded to the experimental groups evaluated the BBB scores on an open-field scale. Evaluations were performed from 3 to 56 days after surgery to monitor the progression of functional recovery in the rats.

Western blot analysis

For western blot analysis, spinal cord tissue was mixed with RIPA lysate, lysed on ice for 30 min, and then centrifuged at 15 000 × g for 10 min at 4 °C. For protein analysis in vitro, HBMECs were lysed in RIPA buffer with protease and phosphatase inhibitors. The protein content of the supernatants was quantified using the PierceTM BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA, USA), and the supernatant was collected for subsequent protein analysis. A 10% gel was used to separate equivalent quantities of 20 µg of protein, which were then transferred onto a polyvinylidene difluoride (PVDF) membrane (Merck Millipore, Darmstadt, Germany). Following blocking with 5% nonfat milk in TBS with 0.05% Tween 20 for 1 h, the membrane was incubated with primary antibodies against ZO-1 (61–7300; Thermo Fisher Scientific), β-Catenin (ab32572, Abcam, Cambridge, UK), Occludin (ab216327, Abcam), Claudin-5 (352500; Santa Cruz Biotechnology, Inc., Dallas, TX, USA), ET-1 (ab117757, Abcam), PI3K (ab191606, Abcam), Akt (ab179463, Abcam), p-Akt (ab38449, Abcam) at 4 °C overnight (around 20 h). After three TBS-T washes, the membranes were incubated with the secondary antibodies for 1 h at room temperature. The protein bands were visualized using an automated gel imaging system (Bio-Rad ChemiDoc MP, Bio-Rad, Hercules, CA, USA), while the band densities were measured using ImageJ software. The relative density ratios normalized to the Sham or Control group were used to describe the findings.

Enzyme-Linked Immunosorbent Assay (ELISA)

ET-1 concentrations in HBMEC culture supernatants were measured using a commercial ET-1 ELISA kit (R&D Systems, Minneapolis, MN, USA). The assay was performed according to the manufacturer’s instructions, with absorbance measured at 450 nm using a microplate reader (Bio-Rad 680, Hercules, CA, USA). ET-1 levels were calculated based on a standard curve.

Real-Time PCR

Spinal cord tissue samples were subjected to isopropanol precipitations after total RNA extraction using the RNAiso plus protocol (TaKaRa Bio Inc., Shiga, Japan). For RNA extraction from sEVs, approximately 50 µg of sEV protein was used as starting material. The extracted total RNA was used to synthesize first-strand cDNA. Quantitative reverse transcription-polymerase chain reaction (RT-PCR) was used to measure mRNA levels using SYBR Green fluorescent probes. The SYBR Green Master Mix (TaKaRa Bio Inc., Shiga, Japan) was added to each reverse-transcription product, and the reaction mixture was then subjected to amplification using a CFX96 Touch Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA).

The following primer pairs were used for amplification:

Prepro-ET-1: Forward: 5′-GTGAGAACGGCGGGGAGAAAC-3′.

Reverse: 5′- AATGATGTCCAGGTGGCAGAAGTAG − 3′.

GAPDH: Forward: 5′-CTCTGATTTGGTCGTATTGGG-3′.

Reverse: 5′-TGGAAGATGGTGATGGGATT-3′.

miR-149: Forward: 5′- GGCTCTGGCTCCGTGTCTT-3′.

Reverse: 5′- CAGTGCAGGGTCCGAGGTATT-3′.

U6: Forward: 5′-CTCGCTTCGGCAGCACA-3′.

Reverse: 5′-AACGCTTCACGAATTTGCGT-3′.

Serial dilutions of each amplicon were also amplified to generate standard curves for the quantification of the PCR products. The copy numbers of each PCR product, equal to 1 µg of total RNA, were used to calculate the quantity of mRNA. The prepro-ET-1 mRNA expression levels were normalized to GAPDH values. For miRNA quantification, U6 snRNA served as the endogenous control.

Evans Blue Dye assays

Evans Blue Dye Assays were performed to assess BSCB permeability. A 2% Evans blue saline solution (2 mL/kg) was administered into the tail vein of rats 7 days after SCI. After 2 h, the rats were anesthetized with 1% sodium pentobarbital (3 mL/kg), and saline was perfused through the heart until clear fluid began to flow from the right atrium. A 1 cm segment of the injured spinal cord, centered around the injury site, was carefully dissected, weighed, and homogenized in a 50% trichloroacetic acid solution. The homogenate was then centrifuged at 10,000 × g for 10 min, and the supernatant was collected. The absorbance of the sample was measured using a spectrophotometer (with an excitation wavelength of 620 nm and an emission wavelength of 680 nm). The established standard curve was used to determine the quantity of Evans dye present in the tissue (µg/g).

FITC-Dextran assays

The rats received an intravenous injection of 2% FITC-dextran (MW 70 kDa, 4 mg/kg; Sigma-Aldrich) solution in PBS via the tail vein 1 day after SCI. After 2 h, the rats were anesthetized with 1% sodium pentobarbital (3 mL/kg, i.p.), followed by perfusion with 0.9% normal saline. The FITC-dextran-damaged spinal cord tissues were weighed, homogenized in PBS, and centrifuged. The optical density of the supernatant was measured using a spectrophotometer at an excitation wavelength of 493 nm and an emission wavelength of 517 nm to assess the presence of FITC-dextran.

Oxygen–glucose deprivation/reoxygenation procedure

Oxygen–glucose deprivation/reoxygenation (OGD/R) procedures were conducted following previously established protocols [35]. Briefly, cultivated HBMECs were washed thrice with PBS and then transferred to serum-free DMEM without glucose (Gibco, Life Technologies, USA). Subsequently, the HBMECs were subjected to oxygen-glucose deprivation (OGD) by placing them in an anaerobic chamber containing 1% O2, 5% CO2, and 94% N2 at 37 °C for 6 h. After being exposed to OGD for 6 h, the HBMECs were washed once with PBS and then incubated under normal conditions (reoxygenation) for 24 h. For sEVs treatment experiments, a specific regimen was followed. At the beginning of the reoxygenation period, the culture medium was replaced with fresh endothelial cell medium containing hUC-MSCs-sEVs at a final concentration of 50 µg protein/mL. In the “OGD/R + sEVs + ET-1” group, exogenous ET-1 (100 nM) was added together with the sEVs at the start of reoxygenation. The cells were then incubated with the sEVs for the entire 24-hour reoxygenation period before being harvested for subsequent analyses.

Cell viability assay

Cell counting kit-8 (CCK-8) assay was used to assess cell viability. HBMECs were cultured in endothelial cell media and seeded in 96-well plates. After 1, 2, 3, 4, and 5 days of incubation, 10 µL of CCK-8 reagent (Dojindo, Japan) was added to the culture medium. A microplate reader (Bio-Rad 680, Hercules, USA) was then used to measure the absorbance of each well at 450 nm.

Paracellular permeability assay

HBMECs were seeded overnight in a 200-µL medium at a density of 1 × 105 cells/well on Transwell permeable supports (PET membrane 24-well cell culture inserts with 0.4-µm pore size; Corelle; Corning Life Sciences, Corning, NY, USA). Subsequently, the cells were subjected to OGD for 6 h, followed by reoxygenation for 22 h (OGD6h/R22h). The cells were then exposed to media containing FITC-dextran (1 mg/ml) for 2 h. The amount of FITC-dextran passing through the Transwell (in the lower chambers) was determined using an enzyme-labeled meter with an excitation wavelength of 493 nm and an emission wavelength of 517 nm.

Hematoxylin and Eosin (HE) staining

At 56 days post-SCI, spinal cord tissues were harvested from rats, fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned into 5 μm thick slices. Sections were deparaffinized, rehydrated, stained with hematoxylin and eosin, dehydrated, and mounted. Tissue morphology was observed under a light microscope.

Immunofluorescence assay

Cells or tissue sections were fixed with 4% paraformaldehyde and permeabilized with 0.3% Triton X-100, then blocked with 5% BSA, and finally incubated with corresponding primary antibodies at 4 ◦C overnight. Corresponding secondary antibodies and DAPI reagent were used to treat cells and sections the following day, and immunoactivity was visualized under a fluorescence microscope. To assess ZO-1 localization in vascular structures, co-immunostaining was performed using anti-ZO-1 (ab190085, Abcam) and anti-CD31 (ab222783, Abcam). To evaluate vessel–astrocyte interactions and astrocytic endfoot coverage in the injured spinal cord, co-immunostaining for CD31 (ab222783, Abcam) and GFAP (ab302644; Abcam) was performed on spinal cord sections. Images were acquired by confocal microscopy and fluorescence intensity was quantified using ImageJ software.

Bioinformatic analysis

The RNA-seq data were obtained from the Gene Expression Omnibus (GEO) database under the accession number GSE171441. This dataset originates from a rat model of spinal cord injury induced by contusion (severe injury) at the T9-T10 level. It includes transcriptomic profiles from spinal cord tissues at two critical post-injury phases: an acute phase (3 days post-injury) and a sub-chronic phase (35 days post-injury). Differentially expressed miRNAs (DEMs) between SCI and Sham groups were identified using the R package “DESeq2” [36] with thresholds of |log2 fold change| > 1 and adjusted p-value < 0.05. The intersection of DEMs from 3-day and 35-day post-injury time points was analyzed to identify common miRNAs. To predict miRNAs targeting EDN1, bioinformatic tools including TargetScan (v7.2, https://www.targetscan.org) and miRTarBase (v2023, https://mirtarbase.cuhk.edu.cn) were employed, and overlapping miRNAs from both databases were retained. Additionally, miRNA expression profiles in human umbilical cord mesenchymal stem cell-derived small extracellular vesicles (hUC-MSCs-sEVs) were extracted from GEO datasets GSE59814 and GSE212320, which characterize the miRNA cargo of sEVs derived from in vitro cultured hUC-MSCs under standard conditions. The final candidate miRNAs were determined by intersecting the SCI-associated DEMs, EDN1-targeting miRNAs, and hUC-MSCs-sEVs-derived miRNAs. Visualization of overlapping miRNAs was performed using a Venn diagram (InteractiVenn, http://www.interactivenn.net).

Differentially expressed genes (DEGs) between the SCI group (including 3-day and 35-day post-injury time points) and Sham group were identified using the R package “DESeq2” with thresholds of |log2 fold change| > 1 and adjusted p-value < 0.05. For functional exploration of miR-149, its potential target genes were predicted via TargetScan and miRTarBase, and overlapping genes from both databases were selected. These predicted targets were further intersected with the SCI-associated DEGs (from the first step) to identify conserved regulatory mechanisms. To visualize the intersection of target genes predicted by TargetScan (v7.2), miRTarBase (v2023), and adhesion junction (AJs)/tight junction (TJs)-related genes, an UpSet diagram was generated using the R package “UpSetR”. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed using the R package “clusterProfiler” [37] with a significance threshold of p-value < 0.05. Protein-protein interaction (PPI) networks were constructed using the STRING database (v12.0, https://string-db.org) and visualized in Cytoscape (v3.9.1) to identify hub genes.

Statistical analysis

All the experiments were performed three times at least. All data are shown as mean ± standard deviation, and statistical analysis was performed in GraphPad Prism (version 8.0, GraphPad Software Inc., USA). One-way ANOVA followed by Tukey’s post hoc analysis was used for multiple comparisons. P-value < 0.05 was considered statistically significant.

Results

hUC-MSCs-sEVs increase the expression of junction proteins in HBMECs after OGD/R

The successful isolation of the hUC-MSCs-sEVs was confirmed by TEM, which revealed their characteristic cup-shaped morphology (Fig. 1A). The hUC-MSCs-sEVs isolates were further identified using NTA, showing that particles with a diameter of 100 to 140 nm were the predominant populations (Fig. 1B), which falls within the typical size range reported for mesenchymal stem cell-derived sEV [20, 21]. Western blot analysis of the sEVs lysates demonstrated significant positive bands for CD63 and TSG101, indicating the presence of exosomal markers, while GAPDH was employed as a loading control (Fig. 1C). In addition, PKH26-labeled sEVs were detected in the perinuclear region of the cytoplasm of HBMECs, indicating that sEVs were successfully internalized by HBMECs (Fig. 1D).

Fig. 1.

Characterization of hUC-MSCs-sEVs and their effects on HBMECs after OGD/R. A TEM image showing the characteristic cup-shaped morphology of isolated sEVs. B NTA analysis of sEV size distribution (predominant 100–140 nm). C Western blot confirming exosomal markers (CD63, TSG101) with GAPDH control. D Internalization of PKH26-labeled sEVs (red) in perinuclear region of HBMECs (DAPI, blue). Scale bar = 20 μm. E CCK-8 assay: cell viability in Control, OGD/R, and OGD/R + sEVs groups. **p < 0.01 vs. Control; ****p < 0.0001 vs. Control; #p < 0.05 vs. OGD/R; ###p < 0.001 vs. OGD/R. F FITC-dextran permeability: OGD/R increased flux, attenuated by sEVs. **p < 0.01; ***p < 0.001. G–H Western blot (G) and quantification (H) of junction proteins (Claudin-5, Occludin, β-Catenin, ZO-1). *p < 0.05; **p < 0.01; ***p < 0.001

To detect the effects of hUC-MSCs-sEVs on OGD/R-injured HBMECs, we conducted a series of experiments, including Cell Viability Assay, Paracellular Permeability Assay, and Western blot analysis. hUC-MSCs-sEVs significantly enhanced cell viability compared to the OGD/R group (Fig. 1E). Moreover, to investigate the impact of OGD/R on the integrity of HBMECs, FITC-dextran permeability assays revealed that OGD/R significantly increased endothelial barrier permeability, while hUC-MSCs-sEVs addition significantly attenuated this effect (Fig. 1F). The expressions of junctional proteins, including Claudin-5, Occludin, β-Catenin, and ZO-1, were significantly decreased in HBMECs subjected to OGD/R. However, the presence of hUC-MSCs-sEVs notably reversed this decrease in expression (Fig. 1G, H).

hUC-MSCs-sEVs increase the expression of junction proteins in HBMECs after OGD/R by downregulation of ET-1

In our previous study, we demonstrated a significant increase in ET-1 levels in the spinal cord tissue of SCI rats, suggesting that excessive ET-1 production following SCI contributes to vasoconstriction, which is closely associated with spinal cord ischemia and hypoxia [29]. To investigate the role of ET-1 in cellular endothelial dysfunction, HBMECs were subjected to OGD/R and treated with hUC-MSCs-sEVs. Our results revealed a significant increase in ET-1 expression in OGD/R-treated HBMECs. However, the administration of hUC-MSCs-sEVs effectively reduced the OGD/R-induced upregulation of ET-1 expression (Fig. 2A, B). These results indicate that hUC-MSCs-sEVs mitigate OGD/R-induced ET-1 overexpression, a critical mediator of vascular pathology.

Fig. 2.

hUC-MSCs-sEVs restore endothelial barrier function by suppressing ET-1. A RT-qPCR analysis of prepro-ET-1 mRNA expression in HBMECs. *p < 0.05; ***p < 0.001. B ET-1 peptide levels measured by ELISA. *p < 0.05; **p < 0.01. C CCK-8 assay: Cell viability in Control, OGD/R, OGD/R + sEVs, and OGD/R + sEVs + ET-1 groups. **p < 0.01 vs. Control; ****p < 0.0001 vs. Control; #p < 0.05 vs. OGD/R; ###p < 0.001 vs. OGD/R; $p < 0.05 vs. OGD/R + sEVs; $$p < 0.01 vs. OGD/R + sEVs. D FITC-dextran permeability assay in HBMEC monolayers. *p < 0.05; **p < 0.01; ****p < 0.0001. E–F Western blot (E) and quantification (F) of junction proteins (Claudin-5, Occludin, β-Catenin, ZO-1) in HBMECs. Data are mean ± SD; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

To further dissect the interplay between ET-1 and endothelial barrier integrity, we evaluated the effects of hUC-MSCs-sEVs and exogenous ET-1 on key functional parameters. Cell Viability Assays demonstrated that hUC-MSCs-sEVs significantly improved survival in OGD/R-injured HBMECs, whereas exogenous ET-1 alone suppressed cell viability (Fig. 2C). FITC-dextran permeability assays further confirmed that OGD/R increased paracellular permeability, which was ameliorated by hUC-MSCs-sEVs. Strikingly, ET-1 treatment exacerbated endothelial barrier leakage, even in the presence of sEVs (Fig. 2D). Similarly, OGD/R injury markedly reduced the expression of junctional proteins (Claudin-5, Occludin, β-Catenin, and ZO-1) in HBMECs, as shown by Western blot. While hUC-MSCs-sEVs treatment restored these protein levels, co-administration of exogenous ET-1 (100 nM) partially reversed the therapeutic effects of sEVs, suggesting ET-1 antagonizes barrier repair (Fig. 2E, F). Collectively, these findings highlight that hUC-MSCs-sEVs restore junctional protein expression, enhance cell survival, and reduce pathological permeability by counteracting ET-1-driven endothelial dysfunction.

Bioinformatic analysis predicted the miR-149/ET-1/PI3K/Akt axis

Bioinformatic analysis identified 17 differentially expressed miRNAs (DEMs) between SCI and Sham groups (|log2 fold change| > 1, adjusted p < 0.05) from the GSE171441 dataset (Fig. 3A, B). Among these, miR-149 emerged as a candidate through further intersection with EDN1-targeting miRNAs (predicted by TargetScan and miRTarBase) and hUC-MSCs-sEVs-derived miRNAs (GSE59814 and GSE212320), as visualized by Venn diagram (Fig. 3C).

Fig. 3.

Bioinformatics identification of miR-149/ET-1/PI3K/Akt axis. List (A) and Volcano plot (B) of 17 DEMs in SCI vs. Sham (GSE171441). C Venn diagram: miR-149 intersection among DEMs, ET-1-targeted miRNAs, and hUC-MSCs-sEVs miRNAs. D Overlap of miR-149 targets from TargetScan and miRTarBase (n = 3476). E Intersection of miR-149 targets with SCI-related DEGs (e.g., EDN1). F miR-149 targets shared with adhesion junction (AJ, 51 genes) and tight junction (TJ, 39 genes) genes. GO (G) and KEGG (H) enrichment of miR-149 targets (p < 0.05)

For functional validation, miR-149’s potential target genes were predicted using TargetScan and miRTarBase, yielding 3476 overlapping candidates (Fig. 3D). Intersection of these targets with SCI-associated differentially expressed genes (DEGs) identified 317 genes (e.g., EDN1), highlighting miR-149’s regulatory association with SCI-related signaling pathways (Fig. 3E). Notably, intersection of miR-149 targets with adhesion junction (AJs) and tight junction (TJs)-related genes uncovered 51, 39 shared candidates respectively (e.g., CLDN5, OCLN), implicating miR-149/ET-1 axis in endothelial barrier dysfunction post-SCI (Fig. 3F). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of these genes revealed significant enrichment (p < 0.05) in pathways related to vascular permeability and inflammatory response, with EDN1 central to these processes (Fig. 3G, H).

Collectively, these results demonstrate that miR-149 directly targets ET-1 and modulates downstream pathways critical to vascular pathology, establishing the miR-149/ET-1 axis as a key mechanistic node in SCI-induced endothelial dysfunction.

hUC-MSCs-sEVs increase the expression of junction proteins in HBMECs after OGD/R by delivering miR-149 in vitro

To directly confirm the presence of miR-149 in hUC-MSCs-sEVs and validate our engineering approach, we first performed qPCR analysis on RNA extracted from the engineered sEV preparations. The results confirmed that miR-149 was detectable in all sEV groups. Importantly, sEVs loaded with miR-149 mimic showed significantly higher miR-149 levels compared to the sEVs-NC mimic group, while sEVs loaded with miR-149 inhibitor showed significantly lower miR-149 levels compared to the sEVs-NC inhibitor group (Fig. 4A). This demonstrates successful modulation of miR-149 cargo in the engineered sEVs.

Fig. 4.

Exosomal miR-149 enhances barrier function in vitro. A RT-qPCR analysis of miR-149 levels in engineered hUC-MSCs-sEVs. B RT-qPCR: intracellular miR-149 expression in HBMECs under Control, OGD/R, and OGD/R + sEVs conditions. C RT-qPCR: intracellular miR-149 levels in HBMECs after treatment with engineered sEVs. D Cell viability and E FITC-dextran permeability assays. F Western blot and G quantification of junction proteins. *p < 0.05; **p < 0.01; ***p < 0.001

To evaluate the effects of OGD/R and sEVs derived from hUC-MSCs on miR-149 expression in HBMECs, we measured miR-149 levels using RT-qPCR across three experimental groups: Control, OGD/R, and OGD/R + sEVs. The results demonstrated that miR-149 expression was significantly decreased in the OGD/R group compared to the Control group (p < 0.05). In contrast, in the OGD/R + sEVs group, treatment with hUC-MSCs-sEVs markedly elevated miR-149 levels compared to the OGD/R group (p < 0.05) (Fig. 4A). These findings suggest that OGD/R suppresses miR-149 expression, whereas hUC-MSCs-sEVs effectively restore its levels.

To investigate the effect of miR-149 in sEVs on endothelial barrier integrity, hUC-MSCs-sEVs was isolated and decorated with miR-149 mimic or inhibitor then co-cultured with HBMECs. RT-qPCR revealed that sEVs-miR-149 mimic upregulated miR-149 expression, while the sEVs-miR-149 inhibitor reduced its expression (Fig. 4B). CCK-8 and FITC-dextran permeability assays were conducted to evaluate cell viability and paracellular permeability. The results indicated that cell viability and the barrier function of HBMECs were enhanced following co-culture with sEVs-miR149 mimic compared with sEVs-NC mimic and reduced with sEVs-miR-149 inhibitor relative to sEVs-NC inhibitor (Fig. 4C, D).

Furthermore, Western blot analysis revealed that the protein expression of ZO-1, β-Catenin, Occludin, and Claudin-5 was higher when HBMECs were cultured with sEVs-miR-149 mimic and reduced with sEVs -miR-149 inhibitor (Fig. 4E, F). Together, the results indicate that sEVs containing miR-149 enhance cell survival, reduce pathological permeability, and restore junctional protein expression.

MiR-149 increases adhesion junctions and tight junctions by inhibiting ET-1 expression

We speculate that miR-149 specifically targets ET-1 in hUC-MSCs, thus affecting the barrier function of endothelial cells. This is consistent with a previous study, which showed that the upregulation of miR-149-5p reduced the high glucose-induced dysfunction in the HUVECs by significantly decreasing the levels of ET-1, vWF, and ICAM-1 and increasing the level of NO and the expression of eNOS [38]. To verify our hypothesis, hUC-MSCs were treated with plasmids of miR-149 mimic and inhibitor. RT-qPCR and ELISA analysis indicated that where the level of miR-149 was high, ET-1 expression was at a low level, namely in hUC-MSCs co-cultured with miR-149 mimic. The converse was observed in hUC-MSCs treated with miR-149 inhibitor (Fig. 5A, B). In summary, miR-149 inhibited the expression of ET-1 in hUC-MSCs.

Fig. 5.

miR-149 suppresses ET-1 to restore junction proteins. ET-1 levels by RT-qPCR (A) and ELISA (B) in miR-149-modified hUC-MSCs. *p < 0.05; **p < 0.01; ***p < 0.001. Cell viability (C) and permeability (D) in HBMECs with Bosentan rescue. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Junction protein expression (E) and quantification (F). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

To further investigate whether miR-149 in small extracellular vesicles affects connexin expression in HBMECs by targeting ET-1, researchers treated OGD/R-injured HBMECs with hUC-MSCs-sEVs containing or lacking miR-149 inhibitors, in combination with the ET-1 receptor antagonist Bosentan. CCK-8 and FITC-dextran permeability assays were conducted to evaluate cell viability and paracellular permeability in HBMECs. Cell Viability Assays demonstrated that the OGD/R + sEVs group showed significantly enhanced cell viability compared to the OGD/R group (p < 0.05), whereas this improvement was reversed in the OGD/R + sEVs-miR-149 inhibitor group (p < 0.05). Intriguingly, the addition of the ET-1 receptor antagonist Bosentan (OGD/R + sEVs-miR-149 inhibitor + Bosentan) rescued the decline, with cell viability significantly elevated compared to the OGD/R + sEVs-miR-149 inhibitor group (p < 0.05; Fig. 5C). FITC-dextran permeability assays further confirmed that OGD/R + sEVs treatment significantly reduced paracellular permeability compared to the OGD/R group (p < 0.05), while miR-149 inhibition (OGD/R + sEVs-miR-149 inhibitor) partially abolished the protective effect of sEVs, leading to increased permeability (p < 0.05 vs. OGD/R + sEVs). Notably, co-treatment with Bosentan (OGD/R + sEVs-miR-149 inhibitor + Bosentan) restored barrier function, significantly reducing permeability compared to the OGD/R + sEVs-miR-149 inhibitor group (p < 0.05; Fig. 5D). These findings collectively indicate that sEVs deliver miR-149 to suppress ET-1 signaling, thereby improving cell survival and preserving endothelial barrier integrity.

Western blot analysis demonstrated dynamic changes in the expression of junctional proteins (ZO-1, β-Catenin, Occludin, and Claudin-5) across experimental groups (Fig. 5E, F). Compared to the Control group, the OGD/R group exhibited a significant decrease in the expression of all four proteins (p < 0.05), indicating that OGD/R injury disrupted endothelial barrier integrity. However, compared to the OGD/R group, the sEVs-NC inhibitor group (equivalent to the sEVs treatment group) showed a significant increase in the expression of all four proteins (p < 0.05), indicating that hUC-MSCs-sEVs restored endothelial barrier integrity. But this protective effect was markedly attenuated in the sEVs-miR-149 inhibitor group, with protein levels reduced compared to the sEVs-NC inhibitor group (p < 0.05), suggesting that miR-149 depletion compromised sEVs’ therapeutic efficacy. Strikingly, the addition of an ETA/ETB non-selective antagonist in the sEVs-miR-149 inhibitor + Bosentan group reversed this decline, significantly elevating junctional protein expression compared to the sEVs-miR-149 inhibitor group (p < 0.05). The stepwise rescue effect of the ET-1 receptor antagonist group further confirms that miR-149 plays an endothelial barrier protection role by targeting and regulating ET-1 to maintain the stability of tight junctions and adherens junctions in HBMECs.

Immunofluorescence staining revealed that, compared to the OGD/R group, the expression of ZO-1 and Claudin-5 was significantly increased in the OGD/R + sEVs group (p < 0.05). The protective effect in the OGD/R + sEVs-miR-149 inhibitor group was markedly attenuated, with fluorescence intensities of ZO-1 and Claudin-5 lower than those in the OGD/R + sEVs group (p < 0.05). In the OGD/R + sEVs-miR-149 inhibitor + Bosentan group, the addition of the ETA/ETB non-selective antagonist reversed this decline, with ZO-1 and Claudin-5 expression significantly elevated compared to the OGD/R + sEVs-miR-149 inhibitor group (p < 0.05), and fluorescence intensity restored to levels comparable to the OGD/R + sEVs group. Confocal microscopy showed that in the OGD/R group, the distribution of ZO-1 and Claudin-5 at cell junctions was disrupted, whereas in the OGD/R + sEVs group, these proteins were redistributed to the cell membrane, exhibiting a continuous linear pattern. In the OGD/R + sEVs-miR-149 inhibitor group, the distribution became discontinuous again, while the Bosentan group restored continuity (Fig. 6A). For quantitative analysis, ImageJ software was used to measure the mean fluorescence intensities of ZO-1 and Claudin-5; results were consistent with Western blot findings, demonstrating that miR-149 maintains the expression and proper distribution of junctional proteins by inhibiting ET-1 (Fig. 6B).

Fig. 6.

Immunofluorescence of junction protein redistribution. A Confocal images: ZO-1/Claudin-5 (red) distribution with DAPI (blue). Scale bar = 50 μm. B Quantification of ZO-1 and Claudin-5 fluorescence intensity. *p < 0.05 vs. OGD/R; #p < 0.05 vs. OGD/R + sEVs; $p < 0.05 vs. OGD/R + sEVs-miR-149 inhibitor. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Exosomal miR-149 enhances endothelial barrier function by suppressing the ET-1/PI3K/Akt signaling pathway

To investigate the role of miR-149 in sEVs derived from hUC-MSCs on OGD/R-injured HBMECs, we examined the expression of ET-1, phosphoinositide 3-kinase (PI3K), total Akt (t-Akt), and phosphorylated Akt (p-Akt) using Western blot analysis. The experiment included four groups: OGD/R, OGD/R + sEVs, OGD/R + sEVs-miR-149 inhibitor, and OGD/R + sEVs-miR-149 inhibitor + Bosentan (an ETA/ETB non-selective antagonist).

The results showed that treatment with OGD/R + sEVs significantly reduced the expression of ET-1, PI3K, and p-Akt compared to the OGD/R group (p < 0.05). In contrast, the OGD/R + sEVs-miR-149 inhibitor group exhibited a significant increase in ET-1, PI3K, and p-Akt levels compared to the OGD/R + sEVs group (p < 0.05), indicating a reversal of the suppressive effect. Notably, co-treatment with Bosentan in the OGD/R + sEVs-miR-149 inhibitor + Bosentan group restored the inhibitory effect, significantly lowering ET-1, PI3K, and p-Akt expression compared to the OGD/R + sEVs-miR-149 inhibitor group (p < 0.05). The expression of t-Akt remained unchanged across all groups, suggesting that the observed changes in p-Akt were due to alterations in phosphorylation rather than total protein levels (Fig. 7A, B).

Fig. 7.

Exosomal miR-149 inhibits ET-1/PI3K/Akt signaling. Western blot (A) and quantification (B) of ET-1, PI3K, t-Akt, p-Akt. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

These findings demonstrate that exosomal miR-149 from hUC-MSCs-sEVs suppresses the activation of the ET-1/PI3K/Akt signaling pathway in OGD/R-injured HBMECs, thereby enhancing endothelial barrier function. The reversal of this effect by miR-149 inhibition and its rescue by Bosentan treatment confirm the specificity of this regulatory mechanism.

Exosomal miR-149 attenuate SCI-induced BSCB disruption in vivo

To investigate the in vivo effects of exosomal miR-149 on SCI-induced BSCB disruption, we employed a rat model of SCI and evaluated motor function, BSCB permeability, and histological changes across four experimental groups: Sham, SCI, SCI + sEVs, and SCI + sEVs + miR-149 inhibitor (Fig. 8A).

Fig. 8.

Exosomal miR-149 preserves BSCB integrity in vivo. A Experimental timeline. B BBB scores from 3–56 days post-SCI. *p < 0.05 vs. SCI; **p < 0.01 vs. SCI; #p < 0.05 vs. sEVs; ##p < 0.01 vs. sEVs. Evans Blue extravasation (C) and quantification (D) at 7 dpi. E FITC-dextran permeability at 7 dpi. *p < 0.05; ***p < 0.001; ****p < 0.0001. F HE staining of spinal cords at 56 dpi. Scale bar = 1 mm

Motor function was measured using the BBB scale from 3 to 56 days post-SCI, which demonstrated that SCI rats exhibited significantly impaired motor function compared to the Sham group (p < 0.05). However, SCI + sEVs treatment markedly improved BBB scores (p < 0.05 vs. SCI), indicating partial functional recovery. This improvement was significantly attenuated in the SCI + sEVs + miR-149 inhibitor group (p < 0.05 vs. SCI + sEVs), underscoring miR-149’s critical role in sEVs-mediated repair (Fig. 8B).

BSCB integrity was assessed through Evans Blue Dye extravasation at 7 days post-SCI. The Sham group showed minimal dye extravasation, reflecting an intact BSCB. In the SCI group, dye extravasation was significantly elevated compared to the Sham group (p < 0.05), indicating pronounced BSCB disruption. In the SCI + sEVs group, sEVs treatment significantly reduced dye extravasation compared to the SCI group (p < 0.05), demonstrating a protective effect on BSCB permeability. However, the addition of the miR-149 inhibitor in the SCI + sEVs + miR-149 inhibitor group markedly increased dye extravasation compared to the SCI + sEVs group (p < 0.05), revealing that the protective effect of sEVs on BSCB integrity is substantially weakened when miR-149 is inhibited (Fig. 8C, D).

To further validate the permeability of the BSCB, we conducted a FITC-dextran (70 kDa) permeability assay, with results consistent with those of Evans Blue, showing that the permeability in the SCI group was significantly higher than in the Sham group (**p < 0.05**), indicating marked BSCB disruption. The permeability in the SCI + sEVs group was significantly lower than in the SCI group (**p < 0.05**), demonstrating the protective effect of sEVs on BSCB permeability. However, in the SCI + sEVs + miR-149 inhibitor group, the addition of the miR-149 inhibitor significantly increased permeability (**p < 0.05**), indicating that inhibiting miR-149 significantly attenuated the protective effect of sEVs on BSCB integrity (Fig. 8E).

Spinal cord tissue morphology was examined using Hematoxylin and Eosin (HE) staining at 56 days post-SCI. In the Sham group, spinal cord tissue exhibited normal architecture, with uniform cell distribution, dense structure, round nuclei, clear nucleoli, and no evident necrosis. The SCI group, however, displayed severe tissue damage, including cell lysis and disappearance, local vacuole formation, necrosis, and hemorrhage. Treatment with sEVs in the SCI + sEVs group significantly alleviated spinal cord tissue damage compared to the SCI group, showing improved cellular structure and reduced necrosis. Conversely, the SCI + sEVs + miR-149 inhibitor group exhibited significantly worsened tissue damage compared to the SCI + sEVs group, with increased cell lysis, necrosis, and structural disruption, underscoring the pivotal role of miR-149 in the histological protection conferred by sEVs (Fig. 8F).

These findings collectively demonstrate that exosomal miR-149 from hUC-MSCs attenuates SCI-induced BSCB disruption in vivo by enhancing motor function, reducing BSCB permeability, and ameliorating histological damage. The significant reversal of these protective effects upon miR-149 inhibition highlights its essential contribution to the therapeutic efficacy of sEVs in SCI treatment.

Exosomal miR-149 restores junctional protein expression in the injured spinal cord

Following the demonstration in subsection 3.7 that exosomal miR-149 attenuates SCI-induced BSCB disruption, we next explored its effects on the expression of tight junction (TJ) and adhesion junction (AJ) proteins critical for BSCB integrity. These proteins—ZO-1, β-Catenin, Occludin, and Claudin-5—were evaluated at 7 days post-injury across four experimental groups: Sham, SCI, SCI + sEVs, and SCI + sEVs + miR-149 inhibitor. Protein expression was quantified using Western blot analysis, and ZO-1 localization in vascular structures was assessed via co-immunostaining with the endothelial marker CD31.

Western blot analysis revealed a significant reduction in the expression of ZO-1, β-Catenin, Occludin, and Claudin-5 in the SCI group compared to the Sham group (p < 0.05), indicative of junctional complex disruption following SCI. In the SCI + sEVs group, treatment with sEVs significantly increased the levels of these proteins compared to the SCI group (p < 0.05), suggesting a restorative effect on TJ and AJ integrity. However, this enhancement was notably diminished in the SCI + sEVs + miR-149 inhibitor group, where expression levels of all four proteins were significantly lower than in the SCI + sEVs group (p < 0.05). These results highlight miR-149 as a crucial component of sEVs-mediated junction protein upregulation post-SCI (Fig. 9A, B).

Fig. 9.

Exosomal miR-149 restores junction proteins in injured spinal cord. Western blot (A) and quantification (B) of ZO-1, β-Catenin, Occludin, Claudin-5 at 7 dpi. CD31/ZO-1 co-immunostaining (C) and fluorescence quantification (D). Scale bar = 20 μm. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Co-immunostaining of spinal cord sections at day 7 post-injury further confirmed these findings. In the Sham group, robust ZO-1 fluorescence was observed along CD31-positive endothelial cells, reflecting intact TJs. The SCI group displayed a marked reduction in ZO-1 fluorescence intensity within vascular structures of the injured spinal cord (p < 0.05 vs. Sham), consistent with compromised BSCB integrity. Treatment with sEVs in the SCI + sEVs group significantly elevated ZO-1 fluorescence intensity compared to the SCI group (p < 0.05), indicating improved TJ reformation. In contrast, the SCI + sEVs + miR-149 inhibitor group exhibited a pronounced decrease in ZO-1 fluorescence compared to the SCI + sEVs group (p < 0.05), underscoring the dependence of sEVs’ protective effects on miR-149 (Fig. 9C, D).

Collectively, these data demonstrate that exosomal miR-149 enhances the expression of TJ and AJ proteins in the spinal cord following SCI, contributing to the stabilization of the BSCB. The reversal of these effects upon miR-149 inhibition emphasizes its pivotal role in promoting junction protein recovery and maintaining endothelial barrier function in vivo.

Exosomal miR-149 restores vessel–glia interactions

To assess vessel–glia interactions and astrocytic endfoot coverage of microvessels following SCI, we performed CD31–GFAP double immunofluorescence on spinal cord sections at day 7 post-injury across the four experimental groups (Sham, SCI, SCI + sEVs, SCI + sEVs + miR-149 inhibitor). Images were acquired by confocal microscopy and analyzed quantitatively for perivascular GFAP fluorescence intensity.

Qualitatively, Sham animals displayed clear astrocytic endfeet in close contact with CD31⁺ endothelial cells, with organized GFAP processes encircling microvessels. In contrast, SCI animals exhibited disrupted astrocyte morphology with retracted and disorganized GFAP processes and markedly reduced perivascular GFAP signal adjacent to CD31⁺ vessels. Importantly, treatment with hUC-MSCs-sEVs (SCI + sEVs) substantially restored close astrocyte–vessel contact and perivascular coverage: GFAP processes appeared more organized and closely associated with CD31⁺ endothelial tubes, and perivascular GFAP intensity was visibly increased compared to the SCI group. Conversely, the SCI + sEVs + miR-149 inhibitor group showed a pronounced attenuation of this restoration, with less astrocytic coverage of vessels than the SCI + sEVs group.

Quantitative analysis confirmed these observations. Perivascular GFAP fluorescence intensity was significantly decreased in the SCI group versus Sham (p < 0.05). SCI + sEVs animals exhibited a significant increase in perivascular GFAP intensity relative to SCI (p < 0.05), whereas the addition of the miR-149 inhibitor significantly reduced this increase (p < 0.05 vs. SCI + sEVs).

Collectively, these data indicate that exosomal miR-149 promotes re-establishment of close astrocyte–endothelial contacts and perivascular astrocytic coverage after SCI, which likely contributes to the recovery of junctional proteins and barrier function observed in this study (Fig. 10)

Fig. 10.

CD31–GFAP co-immunofluorescence reveals restored vessel–astrocyte interactions after sEVs treatment. A Representative confocal immunofluorescence images of spinal cord sections at day 7 post-injury showing CD31 (green) and GFAP (red) staining in the four groups: Sham, SCI, SCI + sEVs, and SCI + sEVs + miR-149 inhibitor. Scale bar = 20 μm. B Quantification of perivascular GFAP fluorescence intensity adjacent to CD31⁺ microvessels. *p < 0.05; **p < 0.01; ***p < 0.001

Discussion

Spinal cord injury is a devastating traumatic disorder of the central nervous system, characterized pathophysiologically by irreversible neuronal destruction and axonal severance, which lead to profound neurological deficits, often presenting as chronic paralysis, and ultimately causing a precipitous decline in patients’ quality of life [2]. The pathological progression of SCI encompasses primary mechanical injury and secondary injury, the latter involving complex cascades such as inflammation, disruption of the BSCB, and neuronal apoptosis [3]. The integrity of the BSCB is essential for maintaining spinal cord microenvironment homeostasis; its disruption exacerbates inflammatory factor infiltration and tissue damage [39]. Current therapeutic strategies for SCI face substantial challenges, as conventional pharmacological and surgical interventions offer limited efficacy against established structural neural tissue damage, highlighting the urgent need for novel approaches. In recent years, hUC-MSCs-sEVs have attracted attention due to their low immunogenicity and capacity to deliver bioactive molecules [25]. Studies suggest that miRNAs within sEVs possess potential in neuroprotection and tissue repair, although their specific mechanisms in SCI remain undefined. miR-149, a key miRNA, exhibits regulatory functions in various diseases, yet its role in SCI repair remains unclear.

To elucidate the role of miR-149 in SCI, we screened 17 DEMs associated with SCI from the GSE171441 dataset and identified miR-149 as a key candidate through intersection with EDN1-targeting miRNAs and those derived from hUC-MSCs-sEVs. Analysis showed that miR-149 target genes intersect with 317 SCI-related differentially expressed genes (DEGs), including 51 and 39 genes related to adherens junctions (AJs) and tight junctions (TJs), respectively, such as CLDN5 and OCLN, which are vital for BSCB barrier function [40]. GO and KEGG pathway enrichment indicated that miR-149 target genes are primarily involved in processes such as vascular permeability, inflammation, and apoptosis. EDN1 (endothelin-1), a core target, is closely linked to elevated expression with vascular injury post-SCI [41]. Our in vivo experiments further validated elevated EDN1 expression post-SCI, aligning with bioinformatics predictions. Bioinformatics analysis provided a theoretical foundation for this study, revealing that miR-149 may alleviate endothelial cell damage by downregulating ET-1 and inhibiting excessive PI3K/Akt pathway activation. This finding aligns with prior studies, such as those showing that upregulation of miR-149-5p significantly reduces ET-1, vWF, and ICAM-1 levels while increasing NO and eNOS expression, thereby mitigating high-glucose-induced dysfunction in HUVECs [38]; however, its function in SCI is reported here for the first time. Compared to other miRNA studies, this research focuses more on vascular protective mechanisms in the BSCB, expanding the application scope of miR-149. The systematic bioinformatics analysis not only enhances the scientific rigor of the study but also provides clear directions for subsequent experimental design.

In vitro, we employed an OGD/R model to simulate post-SCI endothelial cell injury, finding that OGD/R significantly decreased miR-149 expression in HBMECs, whereas hUC-MSCs-sEVs treatment markedly restored miR-149 levels, accompanied by improved cell viability and barrier function. Application of sEVs-miR-149 mimic enhanced these protective effects, while sEVs-miR-149 inhibitor attenuated sEVs’ actions, indicating that miR-149 is a key mediator of sEVs’ function. Western blotting and immunofluorescence analyses revealed that sEVs-miR-149 mimic upregulated ZO-1, β-Catenin, Occludin, and Claudin-5 expression and enhanced their membrane localization, whereas the miR-149 inhibitor reversed these changes.

Importantly, CD31–GFAP co-immunofluorescence results provide evidence that sEVs treatment enhances vessel–astrocyte apposition after SCI: sEVs increased perivascular GFAP signal, whereas inhibition of miR-149 attenuated these effects. These data suggest that the therapeutic effects of exosomal miR-149 extend beyond endothelial cells and involve re-establishment of neurovascular unit (NVU) interactions, particularly astrocyte–endothelial coupling, which likely supports the recovery of junctional proteins and barrier function observed in this study.

The miR-149 agonist reduced ET-1 levels, whereas the inhibitor upregulated ET-1 expression. sEVs-miR-149 mimic significantly improved barrier function in HBMECs post-OGD/R, while the miR-149 inhibitor weakened this effect, and the ET-1 receptor antagonist Bosentan reversed the negative impact of the inhibitor, demonstrating that miR-149 exerts protection by inhibiting ET-1. Furthermore, Western blotting showed that sEVs treatment decreased ET-1, PI3K, and p-Akt expression, with the miR-149 inhibitor reversing this inhibition, while Bosentan rescued the inhibitor’s effects. This suggests that miR-149 regulates endothelial cell function via the ET-1/PI3K/Akt axis.

These results align with existing research, where miRNAs in hUC-MSCs-sEVs play a pivotal role in neuroprotection [25]. This study further reveals that miR-149 maintains endothelial barrier integrity by inhibiting ET-1, thereby reducing inflammatory factor release and oxidative stress damage. Upregulation of junctional proteins is closely associated with BSCB function recovery [39], and miR-149 may indirectly influence their expression by modulating ET-1-mediated inflammatory responses. This mechanism aligns with miR-149’s role in blood-brain barrier protection [42], where it exerts protective effects in cerebral ischemia and blood-brain barrier injury by upregulating tight junction proteins. The present study extends its function to the SCI domain for the first time, focusing on BSCB protection, clarifying ET-1’s upstream regulation, and suggesting broad applicability in nervous system injuries.

In vivo experiments further validated miR-149’s role in SCI repair. Assessment of locomotor activity using the BBB scale indicated that the SCI + sEVs group exhibited significantly higher scores compared to the SCI group, alongside reduced BSCB permeability and attenuated spinal cord tissue damage, whereas the miR-149 inhibitor reversed these improvements. However, we acknowledge that the BBB scale primarily assesses gross locomotor function, and a more comprehensive evaluation using additional sensorimotor tests (e.g., grid walking, CatWalk gait analysis) in future studies will be crucial to fully characterize the extent of functional recovery and further clarify the therapeutic potential of miR-149. Western blotting and co-immunofluorescence analyses demonstrated that sEVs treatment upregulated ZO-1, β-catenin, occludin, and claudin-5 expression, enhancing ZO-1 and CD31 co-localization, while the miR-149 inhibitor diminished these effects. This indicates that hUC-MSCs-sEVs alleviate SCI-induced BSCB disruption and promote junctional protein expression by delivering miR-149.

ET-1 has been reported to exacerbate injury in SCI by inducing vasoconstriction and inflammation [28]. This study found that sEVs treatment significantly reduced ET-1 expression in spinal cord tissue post-SCI, with the miR-149 inhibitor reversing this effect, suggesting that miR-149 protects the BSCB by targeting ET-1. Inhibition of the PI3K/Akt pathway and upregulation of junctional protein expression align with previous studies [43], further supporting the conclusion that miR-149 enhances endothelial barrier function via the ET-1/PI3K/Akt axis. These results are consistent with the role of hUC-MSCs-sEVs in SCI repair [24], providing direct evidence for miR-149’s therapeutic potential in vivo. Moreover, the rescue effect of Bosentan in vivo experiments suggests potential synergy with miR-149, offering new insights into ET-1-targeted therapies [44].

Although this study advances understanding of miR-149’s function and mechanisms, limitations include: (1) Limited cell types: Focus primarily on endothelial cells; while the CD31–GFAP data indicate involvement of astrocyte–endothelial interactions, we did not evaluate pericytes or perform mechanistic assays in astrocytes—future studies employing co-culture systems and targeted astrocytic manipulations are warranted. (2) Immune cell infiltration: We did not quantify macrophage or peripheral immune cell infiltration in the present study; given the role of BSCB integrity in regulating leukocyte entry, assessment of Iba1/CD68+ cells will be an important aim for subsequent investigations. (3) Incomplete ET-1 receptor analysis: Expression changes in ET-1 receptors (ETaR and ETbR) were not assessed; future work should clarify their specific roles in miR-149 regulation. (4) Restricted target genes: Emphasis only on ET-1-related mechanisms, neglecting other miR-149 targets; RNA sequencing could explore additional pathways. (5) Behavioral assessment scope: Our functional analysis relied primarily on the BBB locomotor rating scale. While this established measure confirmed general motor improvement, a more detailed characterization of sensorimotor recovery using additional specialized tests (e.g., assessing fine motor control, sensory function, or gait dynamics) in future work will be critical to fully elucidate the therapeutic potential of miR-149 in SCI recovery. (6) Limited clinical translation: Reliance on cellular and animal models without human data; preclinical trials are needed for efficacy validation. (7) Small sample size: Limited experimental samples may affect statistical reliability; larger cohorts are required in future.

Conclusions

This study, through bioinformatics and in vitro/in vivo experiments, elucidates that exosomal miR-149 derived from hUC-MSCs improves endothelial barrier function and preserves BSCB integrity following SCI, primarily through regulation of the ET-1/PI3K/Akt signaling pathway. While the observed recovery in BBB scores is promising, future investigations employing more comprehensive sensorimotor assessments are warranted to fully characterize the functional benefits. Nonetheless, our findings unveil a significant therapeutic dimension of hUC-MSCs-sEVs and establish miR-149 as a promising molecular target for SCI treatment.

Author contributions

Conceptualization, CH X, XM G, and L S; Methodology, CH X; Investigation, CH X, XC Q, WX W, ZW G, X C, XH Y, H W, JS J, HY F, and H Z; Software, CH X; Formal analysis, CH X; Writing—original draft, CH X; Writing—review & editing, XM G and L S; Funding acquisition, XM G and L S; Resources, XM G and L S; Supervision, XM G and L S. The authors read and approved the final manuscript.

Funding

This work was supported by by the Natural Science Foundation of Shanxi Province (202403021211177); the Fundamental Research Program of Shanxi Province (202303021211217); and the Research and Innovation TeamProject for Scientific Breakthroughs at Shanxi Bethune Hospital (2024ZHANCHI10).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This article does not include any studies involving human participants conducted by any of the authors. The hUC-MSCs were purchased from Fuyuan Biotechnology (Fuyuan Biotechnology Co., Ltd. Shanghai, China), and HBMECs were purchased from Meisen Cell Technology (Meisen Cell Technology, Zhejiang, China). The original sources have confirmed that there was initial ethical approval for collection of these human cells, and that the donors had signed informed consent. All animal experiments were performed in accordance with the ARRIVE guidelines and approved by the Shanxi Provincial People’s Hospital Institutional Animal Care and Use Committee (Title: BMSC-Exos Repair Blood-Spinal Cord Barrier Disruption after Spinal Cord Injury through Regulation of ET1/ETR; Approval No. 89/2022; Date: March 29, 2022).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.