Targeted delivery of the GPX4 activator via HUCMSC-derived exosomes inhibits ferroptosis in spinal cord injury

Abstract

Spinal cord injury (SCI) is a severe and multifaceted neurological trauma characterized by neuronal loss, axonal disruption, and limited regenerative potential, often causing lasting neurological deficits. Ferroptosis promotes neural cell death following SCI. The activation of glutathione peroxidase 4 (GPX4) to inhibit ferroptosis is a promising strategy to repair the injured spinal cord. The efficacy of existing treatments is limited due to their inability to penetrate the blood‒spinal cord barrier (BSCB). To address this challenge, we developed a multifunctional delivery system based on human umbilical cord mesenchymal stromal cell-derived exosomes (MSC-EXOs) targeting the spinal cord. These exosomes were modified with the transactivator of transcription peptide (TAT) and neuron-targeting peptide (RVG) to increase BSCB penetration and SCI-targeting specificity and were also being loaded with the GPX4 activator (GA) to enable ferroptosis inhibition. This multifunctional exosome system has the ability to reduce lipid peroxidation through the activation of GPX4, modulate iron homeostasis, mitigate inflammation, and effectively target the ferroptosis pathway. This targeted approach protects neuronal cells from ferroptosis, promotes axon regeneration and reduce neuroinflammation. This study introduces an innovative, exosome-based strategy to mitigate ferroptosis, underscoring its translational potential for SCI and expanding the possibilities for neuroregeneration.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-025-03755-7.

Introduction

Spinal cord injury (SCI) is a devastating condition of the central nervous system (CNS) characterized by complex and multifaceted pathophysiological processes that create significant barriers to recovery [1–3]. The initial trauma induces primary mechanical damage, leading to a cascade of secondary biochemical responses, which severely hinder functional restoration [4]. Among these mechanisms, secondary death of neural cells and ferroptosis play pivotal roles [5, 6]. Ferroptosis, driven by an overload of intracellular iron ions [7], is especially detrimental in the context of SCI, as accumulated iron catalyzes the generation of reactive oxygen species (ROS) via the Fenton reaction [8]. The development of therapies that directly regulate ferroptosis represents a promising approach for enhancing neuroprotection and achieving improved outcomes in SCI treatment.

Glutathione peroxidase 4 (GPX4) is a key lipid peroxidation scavenging enzyme that regulates ferroptosis in mammals [9]. Our previous study demonstrated that reduced GPX4 expression in the acute and subacute phases of SCI is implicated as a key driver of widespread neuronal ferroptosis. In contrast, the therapeutic activation of GPX4 promoted neuronal survival and enhanced functional recovery in models of SCI [10]. Enhancing the enzymatic activity of GPX4 is a promising strategy to inhibit ferroptosis and promote spinal cord injury repair [11]. The GPX4 activator (GA) mitigates oxidative damage by inhibiting ferroptosis and counteracts the excessive lipid peroxidation that drives ferroptosis [12]. Thus, GA-mediated GPX4 activation may reduce oxidative damage and mitigate the cellular dysfunction associated with SCI. To optimize treatment outcomes, there is an urgent need for targeted delivery systems that can not only enhance the bioavailability of GA but also specifically localize it within the spinal cord.

Exosomes are natural delivery systems due to their unique features, such as high biocompatibility, low immunogenicity, and intrinsic ability to carry proteins, RNAs, and small molecules. Human umbilical cord mesenchymal stromal cell (MSC)-derived exosomes (MSC-EXOs) have shown potential in promoting spinal cord injury repair via microenvironment modulation; additionally, they demonstrate potent drug delivery capabilities [13]. These properties highlight their translational potential for clinical applications in SCI. However, due to the abnormal immune-inflammatory microenvironment around the BSCB during the acute phase of injury [14], the dilution effect of blood, and the sequestration by the liver and spleen, there remains room for improvement in the targeting efficiency of exosomes to the spinal cord, particularly to the injury site. Enhancing the penetration ability of exosomes in complex microenvironments and conferring targeting specificity toward neurons are key strategies to address this issue.

Engineered exosomes can effectively achieve enrichment in the target tissue. The TAT peptide is a well-characterized cell-penetrating peptide [15]. The TAT peptide exhibits strong membrane-translocating capabilities, allowing efficient delivery of therapeutic molecules across cellular barriers [15, 16]. Its stability and adaptability across various cell types make it particularly suitable for applications requiring enhanced tissue penetration, such as SCI [17]. In addition, the RVG peptide further enhances neuron-specific targeting by binding to acetylcholine receptors on neuron, ensuring that the exosomes selectively accumulate in neural tissues [18]. CP05 is a peptide identified through phage display that specifically binds to CD63 and facilitates the loading and targeting of exosomes, thereby enabling their efficient use as delivery vehicles. Therefore, we used CP05 as an anchor to load RVG and TAT onto the surface of the exosomes [19]. By engineering MSC-EXOs with both TAT and RVG peptides, these exosomes enable dual functionality: robust cell penetration and precise neural targeting.

This study aimed to construct multifunctional targeted exosomes that can target and penetrate neuronal cells, delivering a GPX4 activator to regulate ferroptosis. These advancements could pave the way for broader clinical applications, positioning this multifunctional exosome therapy as a promising tool for promoting spinal cord repair.

Results

Engineering multifunctional exosomes for targeted ferroptosis regulation and enhanced SCI repair

A multifunctional exosome system based on human umbilical cord MSC-EXOs was engineered to enhance targeting and therapeutic efficacy in SCI, as illustrated in the graphical abstract. MSC-EXOs were modified with RVG-CP05, a neuronal targeting peptide, and TAT-CP05, a cell-penetrating peptide, to improve neuron-specific delivery and penetration capabilities, as confirmed by immunofluorescence analysis, demonstrating successful conjugation. GA was conjugated to engineered exosomes (MSC-EXO_R&T), enabling precise and efficient delivery to the SCI site. MSC-EXO_R&T-GA exhibited dual regulatory effects on ferroptosis by activating GPX4 to reduce lipid peroxidation and mediating NCOA4 to regulate iron metabolism. This regulation of ferroptosis reversed the ferroptotic microenvironment, modulated inflammatory responses, and promoted neural regeneration and remyelination.

In vivo experiments demonstrated that intravenous administration of MSC-EXO_R&T-GA to SCI mice resulted in increased targeting efficiency, reduced neuronal cell death, improved axonal regeneration, Remyelination, and showed superior motor and sensory recovery compared with those of SCI group, highlighting its therapeutic potential in SCI.

Characterization and functional assessment of the ability of the multifunctional MSC-EXO_R&T-GA to target neurons

First, we constructed a multifunctional exosome system by engineering MSC-EXOs with both TAT and RVG peptides and loading them with the GPX4 activator (GA). The morphology of the multifunctional exosomes was then investigated by transmission electron microscopy, revealing that MSC-EXOs, MSC-EXO_R&T, and MSC-EXO_R&T-GA exhibited similar bilayer membrane structures (Fig. 1a). Nanoparticle tracking analysis (NTA) demonstrated comparable particle size distributions [20] (Fig. 1b). MSC-EXOs, MSC-EXO_R&T, and MSC-EXO_R&T-GA all exhibited the presence of exosomal markers CD9 and CD63, along with the cell membrane marker TSG101, but lacked the organelle marker Calnexin (Supplementary Fig. 1a). The UV absorption spectrum of GA was analyzed, revealing a prominent absorption peak at approximately 300 nm (Supplementary Fig. 1b). The absorbance at 300 nm was subsequently measured for various concentrations of MSC-EXO_R&T-GA. A significant linear relationship between the absorbance and concentration was observed within the detection range, indicating the successful conjugation of GA to the exosomes (Supplementary Fig. 1c).

Fig. 1.

Characterization of MSC-EXO_R&T-GA. (a) Transmission electron microscopy (TEM) images of MSC-EXOs, MSC-EXO_R&T, and MSC-EXO_R&T-GA. (b) Nanoparticle tracking analysis (NTA) was used to assess the size distribution of the exosomes. (c) Immunofluorescence staining showing NeuN (red), RVG-CP05 (green), TAT-CP05 (orange), MSC-EXO (purple), and DAPI (blue). (d) Distribution of fluorescence intensity in body-wide tissues (HT22 cells). (e, f) Fluorescence intensity measurements comparing the neuronal targeting efficiency of different exosome groups (n = 5), showing NeuN (red), MSC-EXOs (purple), and DAPI (blue). (g) Targeted delivery of hUC-MSC-derived exosomes to the spinal cord via tail vein injection in a mouse model, demonstrating their ability to reach the injury site and penetrate cells for therapeutic effects. (h) The mice were sacrificed at 2 h post-injection, and the fluorescence in various tissues was imaged. Ex vivo fluorescence analysis of exosome accumulation in the spinal cord, with labeling for RVG-CP05 (green), TAT-CP05 (red), and DAPI (blue). (i) Distribution of fluorescence intensity in spinal cord tissues (SCI site) (j, k) Immunofluorescence staining was used to assess the distribution of the exosomes in the SCI site (n = 6), which revealed MSC-EXOs (orange) and DAPI (blue). (l, m) Distribution of fluorescence intensity in body-wide tissues. (n) Quantitative analysis of fluorescence intensity in spinal cord tissues (n = 3). The data are presented as the mean ± SD. P > 0.05 (not significant, ns); P ≤ 0.05 (significant, *); P ≤ 0.01 (highly significant, **); P ≤ 0.001 (very significant, ***); and P ≤ 0.0001 (extremely significant, ****). Unpaired Student’s t-test in (f). One-way ANOVA with a Tukey post-hoc test in (k), (n).

To evaluate whether TAT and RVG were successfully conjugated to MSC-EXOs, immunofluorescence experiments were performed. The results revealed that MSC-EXO_R&T-GA successfully entered the HT22 cell line, resulting in a clear distribution within the cytoplasm (Fig. 1c). The merged images revealed overlapping fluorescence signals from RVG-CP05, TAT-CP05, and MSC-EXOs, confirming the successful modification of the exosomes. The fluorescence intensity distribution profiles further confirmed multiple overlapping peaks of RVG-CP05, TAT-CP05, and MSC-EXO fluorescence signals, indicating the successful conjugation of TAT and RVG to MSC-EXOs (Fig. 1d). We also performed fluorescence resonance energy transfer (FRET) assay to verify the proximity of peptides present on the same exosome. As expected, stronger FRET signals were detected in exosomes modified with both CP05-RVG and CP05-TAT than when single-labeled MSC-EXO_RVG and MSC-EXO_TAT were mixed (Supplementary Fig. 1 d), demonstrating that CP05 can modify one exosomes with both TAT and RVG.

To validate the role of TAT in enhancing the penetration efficiency of MSC-EXOs, the HT22 cell line was treated with MSC-EXO_R and MSC-EXO_R&T and observed via confocal microscopy (Fig. 1e). Quantitative analysis of the fluorescence intensity revealed that the relative fluorescence intensity in the MSC-EXO_R&T-treated group was considerably greater than that in the MSC-EXO_R-treated group without TAT, indicating that the targeted exosomes exhibited greater cellular uptake efficiency and potential value for neuron-targeted delivery in SCI treatment (Fig. 1f).

The precise delivery of the exosomes to the SCI region can enhance treatment efficacy. For further histological analysis, exosomes were intravenously injected into SCI model mice 24 h postinjury (Fig. 1g). The results revealed a significant distribution of RVG-CP05 (green), TAT-CP05 (red), and MSC-EXO (blue) signals in the injured spinal cord tissue, confirming that the modified exosomes successfully targeted the SCI site (Fig. 1h). The fluorescence intensity distribution plot (Fig. 1i) further revealed overlapping peaks of the RVG-CP05, TAT-CP05, and MSC-EXO signals in the lesion region, confirming the successful conjugation of TAT and RVG on the exosome surface and their enhanced accumulation at the SCI site. TAT peptides can effectively transport various types of bioactive molecules, including peptides, DNA, RNA, and other small-molecule drugs into cells [15, 16].

Similarly, to examine the role of TAT peptide and RVG peptide in the targeted delivery of MSC-EXO to spinal cord injury sites, different types of exosomes were administered intravenously to SCI model mice. Subsequent immunofluorescence staining analysis of the injured spinal cord revealed that MSC-EXO_R exhibited significant fluorescent labeling, whereas MSC-EXO_R&T displayed even higher signal levels, indicating that TAT-modified exosomes exhibited superior tissue targeting and penetration efficiency (Fig. 1j and k).

To further validate these findings, IVIS imaging was performed to evaluate the biodistribution of Injury, MSC-EXO_R, MSC-EXO_R&T, and MSC-EXO_R&T-GA. DiR(DiIC18(7))-labeled exosomes were used to obtain high-resolution, high-sensitivity signals [21, 22]. IVIS imaging revealed greater fluorescence in the spinal cord lesions of the MSC-EXO_R&T and MSC-EXO_R&T-GA groups 24 h after tail vein injection, whereas the MSC-EXO_R group exhibited weaker fluorescence signals in the lesion area (Supplementary Fig. 1 d). Ex vivo fluorescence analysis also demonstrated that TAT-modified exosomes exhibited a stronger ability to target spinal cord tissue (Fig. 1 l-n).

Systemic safety of MSC-EXO_R&T-GA treatment in SCI

To evaluate the systemic safety of this treatment strategy, HE staining was performed on tissue from the heart, liver, spleen, lungs, and kidneys of mice intravenously injected with sham, injury, MSC-EXO_R, MSC-EXO_R&T, or MSC-EXO_R&T-GA for eight weeks postinjury. No morphological abnormalities were detected (Supplementary Fig. 1e). Serum biochemical indicators, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), total protein (TP), and urea, were analyzed to assess systemic toxicity following exosome treatment (Supplementary Fig. 1f-i). No significant differences in AST, ALT, TP, or urea levels were observed among the sham, injury, MSC-EXO_R, MSC-EXO_R&T, and MSC-EXO_R&T-GA groups, indicating that the treatment did not induce liver or kidney dysfunction.

MSC-EXO_R&T-GA inhibits ferroptosis in neurons and oligodendrocytes in vitro

To further evaluate the protective effects of MSC-EXO_R&T-GA against ferroptosis in neuronal and glial cells, an RSL3-induced ferroptosis model was established in HT22 neurons and OLN93 cell lines to analyze its regulation of cell viability, oxidative stress, and iron homeostasis (Fig. 2a). As the concentration of RSL3 increased, the viability of the HT22 cells significantly decreased (Fig. 2b). Treatment with MSC-EXO_R&T and MSC-EXO_R&T-GA significantly improved cell viability at a concentration of 2 µg/mL, with MSC-EXO_R&T-GA showing superior protective effects compared to MSC-EXO (Fig. 2c and d). The ROS levels were notably reduced in the MSC-EXO_R&T as well as MSC-EXO_R&T-GA groups, with the most pronounced reduction observed in the MSC-EXO_R&T-GA group (Fig. 2e and f). MDA levels, which were significantly elevated in the injury group after RSL3 treatment, were markedly reduced in the MSC-EXO_R&T and MSC-EXO_R&T-GA groups, with the MSC-EXO_R&T-GA group showing the most pronounced reduction, indicating effective mitigation of RSL3-induced oxidative stress (Fig. 2g).

Fig. 2.

Evaluation of protective effects of MSC-EXO_R&T-GA against RSL3-induced ferroptosis in vitro. (a) Establishment of ferroptosis models in the HT22 and OLN93 cell lines using RSL3 treatment. (b) Cell viability analysis of HT22 cells treated with various concentrations of RSL3 (n = 3). (c, d) Viability of HT22 cells treated with MSC-EXO_R&T and MSC-EXO_R&T-GA in the presence of RSL3 (n = 3). (e) Measurement of ROS levels in HT22 cells across different treatment groups using the DCFH-DA assay. (f) Quantitative analysis of ROS fluorescence intensity in HT22 cells (n = 3). (g) MDA levels in HT22 cells treated with PBS, MSC-EXO_R&T, and MSC-EXO_R&T-GA (n = 3). (h) GPX activity assays were conducted in vitro to measure enzymatic activity (n = 3). (i) Intracellular iron levels in HT22 cells across different treatment groups (n = 3). (j, k) Expression levels of GPX4, FTH1, NCOA4, and 4-HNE in cells subjected to different treatments were assessed by Western blotting, with quantitative analysis (n = 3). (l) Cell viability analysis of OLN93 cells treated with various concentrations of RSL3 (n = 3). (m, n) Viability of OLN93 cells treated with MSC-EXO_R&T or MSC-EXO_R&T-GA in the presence of RSL3 (n = 3). (o) DCFH-DA assay showing the ROS levels in OLN93 cells across different treatment groups. (p) Quantitative analysis of ROS fluorescence intensity in OLN93 cells (n = 3). The data are presented as the mean ± SD. P > 0.05 (not significant, ns); P ≤ 0.05 (significant, *); P ≤ 0.01 (highly significant, **); P ≤ 0.001 (very significant, ***); and P ≤ 0.0001 (extremely significant, ****). All the adjusted p values were calculated by one-way ANOVA with a Tukey post-hoc test

In the RSL3-induced ferroptosis model in the HT22 cell line, MSC-EXO_R&T-GA significantly increased GPX activity compared with that in the other groups (Fig. 2h). Compared with those in the other groups, the levels of intracellular iron, which were markedly greater in the injury group, were significantly lower following MSC-EXO_R&T-GA treatment, further validating the role of exosomes in regulating iron homeostasis (Fig. 2i). Compared with the control, RSL3 treatment significantly increased NCOA4 expression while reducing GPX4 and FTH1 levels. Cotreatment with MSC-EXO_R&T and RSL3 effectively reversed these changes, restoring GPX4 and FTH1 levels while further enhancing NCOA4 expression. In contrast, the MSC-EXO_R&T-GA + RSL3 group presented an increased restorative effect (Fig. 2j and k). These results indicate that MSC-EXO_R&T-GA effectively protect neurons and oligodendrocyte cells from ferroptosis-induced damage by modulating oxidative stress and iron homeostasis.

Similarly, as the concentration of RSL3 increased, the viability of OLN93 cells significantly decreased (Fig. 2l). MSC-EXO_R&T and MSC-EXO_R&T-GA treatments significantly improved cell viability (Fig. 2m and n). ROS levels, which were markedly elevated in the RSL3-treated group, were significantly reduced following MSC-EXO_R&T and MSC-EXO_R&T-GA treatments, with MSC-EXO_R&T-GA showing the most pronounced reduction (Fig. 2o and p).

In summary, MSC-EXO_R&T-GA exhibited significant protective effects on both the HT22 and OLN93 cell models of RSL3-induced ferroptosis by alleviating ferroptosis-induced damage through the activation of GPX4.

MSC-EXO_R&T-GA promotes functional recovery after SCI

To evaluate the therapeutic efficacy in terms of functional and histological outcomes following SCI, multiple behavioral and histological assessments were performed (Fig. 3a). The Basso Mouse Scale (BMS) scores indicated a remarkable improvement in hindlimb motor function in the MSC-EXO_R&T and MSC-EXO_R&T-GA treatment groups compared with the injury group, with MSC-EXO_R&T-GA showing the most pronounced effect (Fig. 3b and c). CatWalk gait analysis revealed significant improvements in motor coordination and locomotor recovery in the MSC-EXO_R&T and MSC-EXO_R&T-GA groups compared with the Injury group. The MSC-EXO_R&T-GA group exhibited the highest scores for the stride sequence regularity index and maximum contact area, indicating enhanced walking performance (Fig. 3d-f). The results of the swim test supported these findings, as MSC-EXO_R&T-GA-treated mice presented markedly higher swim scores than did the injury and MSC-EXO-treated mice (Fig. 3j and k). Sensory recovery was assessed using the hot plate test. Compared with the Injury group, the MSC-EXO_R&T-GA-treated group presented significantly shorter reaction times, indicating improved sensory function (Fig. 3g). Autonomic function was evaluated through bladder diameter measurements, which revealed that MSC-EXO_R&T-GA significantly reduced bladder size, indicating enhanced recovery of autonomic regulation (Fig. 3h and i).

Fig. 3.

Behavioral, histological, and electrophysiological evaluation of the therapeutic efficacy of MSC-EXO_R&T-GA in SCI models. (a) Schematic representation of the experimental timeline for SCI model establishment and treatment evaluation. (b, c) BMS scores were used to assess hindlimb motor function recovery across different treatment groups (n = 6). (d) CatWalk gait analysis showing limb contact with the runway during forward movement in different groups of mice. (e, f) CatWalk gait analysis evaluating the maximum contact area and regularity index (n = 10). (g) Hot plate test assessing sensory recovery (n = 6). (h, i) Bladder diameter measurements evaluating autonomic function recovery (n = 3). (j, k) Swim test results for evaluating functional recovery (n = 6). (l, m) MR image showing lesion volume reduction in the spinal cord (n = 3). (n, o) Motor-evoked potential (MEP) recordings for measuring neuronal conductivity recovery (n = 6). The data are presented as the mean ± SD. P > 0.05 (not significant, ns); P ≤ 0.05 (significant, *); P ≤ 0.01 (highly significant, **); P ≤ 0.001 (very significant, ***); and P ≤ 0.0001 (extremely significant, ****). All the adjusted p values were calculated by one-way ANOVA with a Tukey post-hoc test

MRI revealed reduced lesion volumes in the MSC-EXO_R&T and MSC-EXO_R&T-GA groups, with the smallest lesion size observed in MSC-EXO_R&T-GA-treated mice (Fig. 3l and m). Electrophysiological analysis of motor-evoked potentials (MEPs) revealed marked recovery of neuronal conductivity in the MSC-EXO_R&T and MSC-EXO_R&T-GA groups, with the MSC-EXO_R&T-GA group achieving significantly greater amplitudes than the injury group did, suggesting effective neuronal repair (Fig. 3n and o).

In summary, MSC-EXO_R&T-GA treatment significantly enhanced motor coordination, sensory recovery, and autonomic function while reducing spinal cord lesion size and preserving tissue integrity, underscoring its therapeutic potential in SCI repair.

MSC-EXO_R&T-GA enhances neuronal survival and reduces astrogliosis after SCI

To further evaluate the treatment effects of MSC-EXO_R&T-GA, histological and immunofluorescence analyses were performed at 8 weeks postinjury (wpi) following behavioral assessments to assess neuronal survival, astrocyte activity, and axonal regeneration. Immunofluorescence staining revealed a significant reduction in the number of neurons and markedly increased astrocyte activity in the injury group, particularly in the lesion core. In contrast, the MSC-EXO_R&T and MSC-EXO_R&T-GA groups exhibited increased neuronal survival and decreased astrocyte activity, with the MSC-EXO_R&T-GA group demonstrating the most pronounced effects (Fig. 4a). Quantitative analysis revealed that the number of NeuN⁺ cells in the MSC-EXO_R&T-GA group was markedly greater than that in the injury and MSC-EXO groups, indicating a prominent neuroprotective effect (Fig. 4b). Additionally, the GFAP fluorescence intensity was markedly elevated in the Injury group but notably reduced in the MSC-EXO_R&T-GA group, suggesting decreased astrocyte activation and reduced glial scar formation (Fig. 4c).

Fig. 4.

Immunofluorescence analyses of neuronal survival, astrocyte activity, and axonal regeneration in SCI models. (a) Immunofluorescence staining of spinal cord tissue showing NeuN⁺ neurons (red) and GFAP⁺ astrocytes (green) in the lesion core at 8 weeks postinjury (wpi). (b) Quantitative analysis of NeuN⁺ cell counts across treatment groups (n = 6). (c) Quantitative analysis of GFAP fluorescence intensity across treatment groups (n = 6). The data are presented as the mean ± SD. P > 0.05 (not significant, ns); P ≤ 0.05 (significant, *); P ≤ 0.01 (highly significant, **); P ≤ 0.001 (very significant, ***); and P ≤ 0.0001 (extremely significant, ****). All the adjusted p values were calculated by one-way ANOVA with a Tukey post-hoc test

NF200 is a widely used marker for neurofilaments, and its assessment provides critical insights into axonal integrity and regeneration following SCI [23, 24]. Further analysis of NF200-positive areas to evaluate axonal regeneration revealed that the NF200-positive area was significantly larger in the MSC-EXO_R&T-GA group than in the other groups, indicating its ability to significantly promote axonal regeneration (Supplementary Fig. 1j and k).

In summary, MSC-EXO_R&T-GA significantly protected neurons, inhibited astrocyte activation, and promoted axonal regeneration in the SCI model.

MSC-EXO_R&T-GA promotes oligodendrocyte survival in SCI

Oligodendrocytes play a critical role in myelin formation, and their survival and functional recovery are essential for neural repair [23]. To evaluate the effects of MSC-EXO_R&T-GA on oligodendrocyte survival and remyelination, immunofluorescence staining was performed using Oligo2 and myelin basic protein (MBP). The number of Oligo2⁺ cells in the injury group was markedly lower than the sham group, whereas the MSC-EXO_R&T and MSC-EXO_R&T-GA treatment groups demonstrated a significant increase in Oligo2⁺ cell numbers, with the MSC-EXO_R&T-GA group reaching levels comparable to the sham group, indicating stronger protective effects on oligodendrocytes (Fig. 5a and b). In the MBP staining results, the injury group presented significantly reduced MBP fluorescence intensity, indicating severe myelin damage. In contrast, the MSC-EXO_R&T and MSC-EXO_R&T-GA treatments significantly increased the MBP fluorescence intensity, with the MSC-EXO_R&T-GA treatment resulting in the greatest effect on myelin regeneration (Fig. 5c and d). Compared to the Injury group, a significant increase in NF200 intensity was observed near the injury site in the MSC-EXO_R&T-GA group (Fig. 5c and e). Masson’s trichrome staining further revealed reduced scar tissue and improved spinal cord tissue preservation in the MSC-EXO_R&T-GA group (Fig. 5f and g). LFB staining revealed that the MSC-EXO_R&T-GA group presented the greatest degree of myelin protection and repair after SCI, significantly reducing the damage area and improving myelin integrity (Fig. 5h). Different staining methods focused on different regions in the sagittal section of the spinal cord (Fig. 5i).

Fig. 5.

Immunofluorescence analysis of oligodendrocyte survival and myelin regeneration in SCI models. (a) Immunofluorescence staining of Oligo2⁺ (red) cells in spinal cord tissue at 8 weeks postinjury (wpi). (b) Quantitative analysis of Oligo2⁺ cell counts across treatment groups (n = 3). (c) Immunofluorescence staining of MBP (red) and NF200 (green) in spinal cord tissue to evaluate myelin regeneration. (d, e) Quantitative analysis of MBP and NF200 fluorescence intensity across treatment groups (n = 6). (f, g) Masson’s trichrome staining of spinal cord tissue was used to assess scar tissue formation and tissue preservation (n = 3). (h) LFB staining of spinal cord tissue to evaluate the extent of myelin damage and the protective effects of MSC-EXOs, MSC-EXO_R&T, and MSC-EXO_R&T-GA. (i) Schematic diagram showing the locations of the results obtained from different staining methods in the damaged spinal cord. The data are presented as the mean ± SD. P > 0.05 (not significant, ns); P ≤ 0.05 (significant, *); P ≤ 0.01 (highly significant, **); P ≤ 0.001 (very significant, ***); and P ≤ 0.0001 (extremely significant, ****). All the adjusted p values were calculated by one-way ANOVA with a Tukey post-hoc test

In summary, MSC-EXO_R&T-GA significantly promoted oligodendrocyte survival and enhanced myelin regeneration, providing strong support for neural repair in SCI.

MSC-EXO_R&T-GA suppresses ferroptosis via the dual regulation of ferritinophagy inhibition and GPX4 activation

To explore the potential mechanism by which MSC-EXO_R&T-GA regulate ferroptosis, RNA-seq was performed to identify differentially expressed genes and pathways at 7 days postinjury (Fig. 6a). Notably, this time point represents a critical node for inflammatory responses, cell death, and repair processes following spinal cord injury [25, 26]. Heatmap analysis revealed that MSC-EXO_R&T-GA treatment upregulated genes associated with iron homeostasis and antioxidative defense, such as FTH1, which is critical for iron storage [27], and Sqstm1, a regulator of oxidative stress [28], while downregulating ferroptosis-promoting genes such as ACSL4, which drives lipid peroxidation, compared with those in the injury group [29, 30] (Fig. 6b). GSEA demonstrated significant enrichment in the "iron ion transport" pathway, suggesting that MSC-EXO_R&T-GA is crucial for maintaining iron homeostasis (Fig. 6c). GO enrichment analysis further confirmed that the treatment was associated with terms such as "iron ion binding" and "cellular response to iron ions," which are key to ferroptosis regulation [31, 32] (Fig. 6d). KEGG pathway enrichment analysis revealed significant enrichment of the axon guidance, TNF signaling, and ferroptosis pathways in the exosome-treated group (P < 0.05). Axonal guidance suggests roles in neuronal network reconstruction, TNF signaling indicates reduced inflammation, and ferroptosis highlights iron metabolism and lipid peroxidation in neuronal death. Compared with Injury, exosomes have the potential to mitigate secondary SCI (Fig. 6e).

Fig. 6.

Transcriptomic and molecular analyses of ferroptosis regulation by MSC-EXO_R&T-GA in SCI models. (a) Exosome injections were administered intravenously on the 1 st and 3rd days after SCI, with tissue samples collected on the 7th day for transcriptome sequencing. (b) Heatmap of differentially expressed genes associated with iron homeostasis as well as ferroptosis regulation in the MSC-EXO_R&T-GA and Injury treatment groups. (c) GSEA showing enrichment in the "iron ion transport" pathway in MSC-EXO_R&T-GA-treated samples. (d) GO enrichment analysis bubble diagram (n = 3). (e) KEGG enrichment analysis bubble diagram (n = 3). (f) A protein–protein interaction (PPI) network was constructed using differentially expressed genes associated with ferroptosis regulation. The node size indicates the interaction degree, and the edge thickness represents the interaction strength, which visualizes key interactions within the network. (g, h) Expression levels of FTH1, NCOA4, and 4-HNE in spinal cord tissue subjected to various treatments were assessed by Western blotting, with subsequent quantitative analysis (n = 3). (i–l) Quantitative analysis of intracellular iron accumulation, MDA levels, GPX activity, and proinflammatory cytokines (TNF-α and IL-1β) was performed to assess the anti-inflammatory effects of MSC-EXO_R&T-GA (n = 3). (m) MSC-EXO_R&T-GA regulate ferroptosis by modulating ferritin degradation (via NCOA4), iron release, lipid ROS accumulation (via the Fenton reaction), and GPX4 inhibition, ultimately promoting oxidative stress-induced cell death. The data are presented as the mean ± SD. P > 0.05 (not significant, ns); P ≤ 0.05 (significant, *); P ≤ 0.01 (highly significant, **); P ≤ 0.001 (very significant, ***); and P ≤ 0.0001 (extremely significant, ****). All the adjusted p values were calculated by one-way ANOVA with a Tukey post-hoc test

PPI (protein‒protein interaction) network analysis identified FTH1 and NCOA4 as central hub genes, highlighting their pivotal roles in ferroptosis regulation. FTH1, which is essential for iron storage, and NCOA4, a key regulator of ferritinophagy, collaborate to maintain iron homeostasis. These interactions suggest that MSC-EXO_R&T-GA may regulate ferroptosis by targeting these key genes (Fig. 6f). To validate these findings, Western blot analysis was conducted on samples from the in vivo experiments. The ferroptosis defense protein GPX4 was restored by MSC-EXO_R&T-GA treatment, while the lipid peroxidation marker 4HNE was significantly reduced, indicating effective inhibition of ferroptosis. Additionally, the iron storage and release regulators FTH1 and NCOA4 were significantly upregulated, further supporting the results indicating modulation of iron homeostasis [33–35] (Fig. 6g and h).

In parallel, oxidative stress and inflammation markers were assessed using ELISA. MSC-EXO_R&T-GA treatment significantly reduced intracellular iron accumulation and decreased the levels of MDA, an indicator of oxidative stress, compared with those in the injury group (Fig. 6i and j). Moreover, the GPX activity of the spinal cord tissues collected on Day 7 postinjury was markedly greater in the MSC-EXO_R&T-GA treatment group than in the Injury and MSC-EXO groups (Fig. 6k). The proinflammatory cytokines TNF-α and IL-1β, which exacerbate ferroptosis-related inflammation, were significantly suppressed in the MSC-EXO_R&T-GA group, highlighting their anti-inflammatory potential [36, 37] (Fig. 6l and m).

In summary, MSC-EXO_R&T-GA protected against ferroptosis by modulating key pathways related to ferritinophagy, lipid peroxidation, and inflammation, highlighting its potential in mitigating ferroptosis in SCI (Fig. 6n).

Discussion

The formation of a ferroptotic microenvironment following SCI results in significant neuronal cell death. Targeted inhibition of ferroptosis represents a promising strategy for SCI repair. In this study, MSC-derived exosomes were conjugated with TAT and RVG peptides and loaded with a GPX4 activator. These engineered exosomes effectively target and accumulate in injured spinal cord tissue, enabling the efficient delivery of the GPX4 activator to suppress ferroptosis. Furthermore, exosomes can modulate iron metabolism, helping to regulate the ferroptotic microenvironment. This dual regulation of ferroptosis ultimately enhances nerve cell survival and facilitates spinal cord injury repair.

Our in vivo fluorescence imaging confirmed the improved targeting ability of MSC-EXOs, which were substantially enriched at the SCI site compared with those in the control group. This effective targeting ability is central to the potential of MSC-EXOs in SCI therapy, as it ensures that therapeutic agents are concentrated at the injury site, maximizing local efficacy while minimizing off-target effects. Modifications using the TAT and RVG peptides were crucial for achieving this targeted and penetrating effect. TAT is a well-known cell-penetrating peptide that facilitates translocation across cellular membranes, providing MSC-EXOs with the ability to reach intracellular targets within both neurons and glial cells at the injury site [15]. This feature is particularly important given that the repair of neuronal and glial cells requires the intracellular delivery of therapeutic agents. Moreover, RVG acts as a neuron-targeting peptide, binding specifically to acetylcholine receptors on neuronal cells [17], thereby increasing the accumulation of MSC-EXOs within neural tissues and promoting their selective localization to the site of SCI.

Together, TAT and RVG enable MSC-EXOs to effectively navigate cellular barriers and accumulate precisely where they are most needed, enhancing therapeutic delivery and reducing the risk of systemic distribution, which can lead to undesirable side effects. This highly targeted and localized delivery of MSC-EXOs, as confirmed by the imaging results, represents a promising step forward in overcoming the limitations of conventional treatments for SCI, which often involve drug dispersion and inadequate BSCB penetration. Our approach ensures that the therapeutic payload remains concentrated at the injury site, maximizing the neuroprotective and regenerative effects of exosome treatment. In SCI, dysregulated iron metabolism and oxidative stress are key contributors to ferroptosis [38], which exacerbates neural damage and hinders recovery [39]. After SCI, cellular iron homeostasis becomes critically disrupted. Typically, ferritin heavy chain 1 (FTH1) sequesters ferric iron (Fe³⁺) to maintain iron balance [40]; However, in response to injury, neuronal iron demand spikes, leading to nuclear receptor coactivator 4 (NCOA4)-mediated ferritinophagy [35]. This process releases Fe³⁺ from FTH1, converting it to ferrous iron (Fe²⁺) and increasing free Fe²⁺ levels. Through the Fenton reaction, Fe²⁺ generates ROS when it interacts with hydrogen peroxide (H₂O₂) [8], significantly increasing oxidative stress at the injury site. Elevated ROS drive lipid peroxidation, generating harmful byproducts such as 4-hydroxynonenal (4HNE) and malondialdehyde (MDA), which damage cell membranes, exacerbating oxidative stress. Increased ROS also upregulate proinflammatory cytokines, including TNF-α and IL-1β, perpetuating a cycle of inflammation as well as oxidative damage within injured tissue [41, 42].

Although GPX4, a critical antioxidant enzyme, plays a central role in detoxifying lipid peroxides [7, 10, 43], its protective capacity is often overwhelmed in the highly oxidative environment of SCI [44], which accelerates ferroptosis. These findings underscore the importance of therapies that target both iron homeostasis and oxidative stress reduction in SCI treatment. Modulating FTH1 expression to increase iron sequestration can mitigate the availability of free Fe²⁺ [45, 46], directly reducing ROS levels and consequently lowering lipid peroxidation. Increasing GPX4 activity to counter lipid peroxides would also serve as a robust defense against oxidative membrane damage and inflammation. After intervention with MSC-EXO_R&T-GA, the astrogliosis was significantly reduced compared with the control group, and this effect may be associated with the decreased levels of lipid peroxidation metabolites. Persistent oxidative stress post-injury leads to the accumulation of products such as ROS and 4HNE, which may induce astrocyte activation [47]. This activation further drives the formation of glial scar, which in turn hinder axonal regeneration and the recovery of neurological function [48]. Meanwhile, this study found that MSC-EXO_R&T-GA can promote axonal regeneration after injury. This may be due to the reduced ferroptosis, which improves the survival of oligodendrocytes. As key cells responsible for myelin synthesis in the central nervous system, the increased survival of oligodendrocytes can facilitate myelin regeneration. This not only provides structural support for axonal regeneration but also reduces the inhibitory effect of myelin debris on axons, ultimately promoting the regeneration of injured axons and the recovery of neurological function.

This study demonstrates that MSC-EXO_R&T-GA significantly promotes motor functional recovery after SCI through the targeted inhibition of neuronal ferroptosis. However, the inherent complex biological functions of the exosome carrier itself must be carefully considered. MSC-EXOs possess well-documented intrinsic immunomodulatory and neuroprotective properties [49, 50]. These functions may independently contribute to the neural repair process. The observed improvement in motor function in both the MSC-EXO and MSC-EXO_R&T groups compared to the injury group demonstrates the protective role that the MSC-EXOs can exert in SCI. Critically, the MSC-EXO_R&T-GA group exhibited superior motor functional recovery and neuronal protection compared to both the MSC-EXO and MSC-EXO_R&T control groups. This rigorous comparative analysis provides strong evidence that the superior functional recovery achieved by the drug-loaded system primarily stems from the specific therapeutic intervention of its ferroptosis inhibitor on this key pathological mechanism.

MSC-EXOs hold significant promise for clinical translation because of their unique biocompatibility, low immunogenicity, and high drug-loading capacity [13]. These properties make it an ideal candidate for clinical applications, particularly in neuroregenerative medicine, where minimally invasive and highly targeted treatments are in high demand [51]. The multifunctional exosome-based strategy developed in this study, which combines cell-penetrating and neuron-targeting peptide modifications, represents a cutting-edge approach to precisely target the SCI region. The natural compatibility of MSC-EXOs with the human body reduces the risk of adverse immune responses, making them promising vehicles for delivering therapeutic agents in the clinical setting. Furthermore, the enhanced targeting and treatment effects observed in this study demonstrate the potential of MSC-EXOs as a viable and effective strategy for treating SCI. By modulating iron homeostasis and reducing inflammation, MSC-EXOs address two key pathological drivers of SCI, supporting neurorepair and functional recovery. These results underscore the potential of MSC-EXOs as versatile tools for SCI therapy, offering a practical and innovative approach that could be adapted for other neurological injuries.

As shown by GFAP IF, the lesion area in the MSC-EXO group appeared larger than that in the Injury group in certain images. This discrepancy may be attributed to the inherent variability in lesion size across different tissue regions within the same injury model, which is commonly observed in spinal cord injury studies and reflects the complex and heterogeneous nature of SCI lesions. Importantly, this difference is not related to tissue processing or sectioning, but rather represents biological variability within the model. To further clarify this point, we examined both neuronal counts and GFAP fluorescence intensity in the MSC-EXO and Injury groups. The results demonstrated that the MSC-EXO group exhibited a greater number of neurons and lower GFAP fluorescence intensity compared with the Injury group, indicating better tissue recovery in the MSC-EXO group. These conclusions are consistent with previously published studies [52–54]. Collectively, these results support the neuroprotective role of MSC-EXOs and suggest that the apparent variability in lesion size does not contradict the overall reparative effects observed in our study.

Despite these significant findings, further exploration is needed. First, while the targeting efficiency of MSC-EXOs in acute SCI models has been confirmed, it remains to be seen whether this efficiency can be sustained in chronic SCI models, where the injury environment is more complex and treatment windows are extended [55, 56]. Thus, the temporal dynamics of in vivo degradation of MSC-EXOs require further investigation. Additionally, optimizing the dosage and route of administration could further enhance therapeutic efficacy. While intravenous administration was effective in our study, alternative routes or modified dosages may yield improved outcomes and reduce the need for repeated administration. Another limitation is the need to investigate the long-term effects of multifunctional targeted exosome therapy. Although short-term improvements were observed, understanding the durability of these effects over extended periods is essential for assessing the potential of MSC-EXOs as long-term treatments for SCI. Long-term studies could help determine the safety and sustained efficacy of MSC-EXOs, ensuring that their benefits continue without adverse effects or diminishing returns.

Future studies should also explore the likelihood of combining MSC-EXOs with other therapeutic agents to achieve synergistic effects. For example, delivering anti-inflammatory drugs, neurotrophic factors, or other neuroprotective compounds alongside MSC-EXOs could amplify regenerative benefits, further improving recovery outcomes for SCI patients. Additionally, investigating the compatibility of MSC-EXOs with existing SCI therapies, such as physical rehabilitation and pharmacological treatments, may reveal opportunities for integrative approaches that maximize patient recovery.

In conclusion, the development and application of multifunctional targeted exosomes, as demonstrated in this study, highlight the potential of MSC-EXOs as a transformative treatment method for SCI. The combination of targeted delivery, dual regulatory action of ferroptosis, and clinical compatibility positions MSC-EXOs as promising therapeutic candidates with broad applicability. This study suggests that continued innovation in exosome-based therapy could significantly improve outcomes for SCI patients and lay a foundation for broader clinical applications in neuroregeneration.

Materials and methods

Preparation of peptides

The peptides were synthesized by Chinapeptide Co. Ltd. and stored as lyophilized powders at −80 °C. To prepare the exosomes, these peptides were dissolved in PBS. The peptide sequences used were as follows:

RVG-CP05, YTIWMPENPRPGTPCDIFTNSRGKRASNGGGGSCRHSQMTVTSRL

TAT-CP05, YGRKKRRQRRRGGGSCRHSQMTVTSRL

Isolation, functional modification, and characterization of MSC-EXO_R&T-GA

The supernatants obtained from human umbilical cord MSC cultures underwent sequential centrifugation to eliminate cells and debris. The process followed a stepwise protocol: initial centrifugation at 300 × g for 10 min, subsequent centrifugation at 2,000 × g for 20 min, and, ultimately, centrifugation at 10,000 × g for 30 min. After passing through a 0.22-μm sterile membrane (Millipore), the resulting supernatant was ultracentrifuged for 70 min at 130,000 × g (Optima L-100 XP, Beckman Coulter). A second ultracentrifugation step was performed on the pellet at the same parameters after the pellet was reconstituted in 1 mL of PBS. The BCA test was used to measure the amount of protein.

MSC-EXOs were incubated with RVG-CP05 and TAT-CP05, either individually or together, at 4 °C for 6 h. To remove unbound peptides, the mixtures were passed through 100-Kd filter tubes (Millipore) and washed five times with PBS. Finally, PBS was used to resuspend the complexes, yielding MSC-EXO_R and MSC-EXO_R&T.

Exosomes were conjugated to GA through a DIC/NHS-mediated amide reaction. First, MSC-EXO_R&T were treated with MES buffer (pH 6.0) and dispersed ultrasonically to activate carboxyl groups. DIC and NHS in DMSO activated these groups, forming NHS esters. After 30 min, an amino-containing GA solution was added, and the pH was adjusted to 7.4. NHS esters formed amide bonds with GA amino groups during a 1-h incubation. Excess byproducts were removed by ultrafiltration and ultracentrifugation.Successful conjugation was validated using UV absorption spectroscopy and fluorescence resonance energy transfer (FRET) assay. Prepared fresh for immediate use, avoiding repeated freeze–thaw cycles.

The morphology of the MSC-EXOs, MSC-EXO_R&T, and MSC-EXO_R&T-GA was examined by TEM, and their size and distribution were assessed using NTA. EV-specific markers were verified by Western blot analysis.

Cell culture

The HT22 and OLN93 cell lines were cultured in Dulbecco’s modified Eagle medium/nutrient mixture F-12 (Cat# 11,330,057; Gibco, Carlsbad, CA, USA), supplemented with 10% fetal bovine serum (Cat# 16,000,044; Gibco) and 1% penicillin–streptomycin (Cat# 15,140,122; Gibco) at 37 °C and 5% CO₂. The culture medium was changed every 2 days. To circumvent the oxidative stress effects of glutamate on the HT22 cell line, all medium used in vitro experiments were devoid of glutamate.

Exosome uptake and endocytosis

Exosomes were labeled with the lipophilic dye DiO (3,3’-dioctadecyloxacarbocyanine perchlorate, Invitrogen), while RVG-CP05 and TAT-CP05 were tagged with FITC and Rhodamine B, respectively. MSC-EXO_R and MSC-EXO_R&T were prepared following standard protocols. Purified exosomes were resuspended in PBS at 100 µg/mL. A 10 µM DiO stock solution in ethanol was diluted in PBS to a final concentration of 1 µM. Equal volumes of DiO working solution and the exosome suspension were gently mixed and incubated at 37 °C for 30 min, enabling the dye to be incorporated into the exosome lipid bilayer. To remove any unbound dye, ultracentrifugation was performed at 130,000 × g for 70 min at 4 °C (Beckman Coulter, USA). The resulting pellet was washed with PBS and centrifuged again to remove any residual dye, after which the labeled exosomes were resuspended in PBS. Fluorescence microscopy confirmed successful labeling.

For the endocytosis assay, 10 μg of DiO-labeled MSC-EXO_R or MSC-EXO_R&T were added to the cell culture medium. Following incubation, the cells were fixed and stained with a rabbit anti-NeuN antibody (1:500, Abcam, ab104225), followed by treatment with a secondary antibody. Exosome uptake was visualized using an LSM900 inverted confocal microscope (Germany, ZEISS).

Spinal cord injury

This research utilized female C57BL/6 mice (8 weeks old, 20 g body weight) from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Guangdong, China). To induce the spinal contusion model, the mice were anesthetized with isoflurane (RWD, R510-22), and a midline incision was made at the T10 vertebra. The surrounding muscles and connective tissues of the thoracic spine were carefully dissected with minimal disruption to adjacent structures, exposing the T9–T11 spinous processes. Laminectomy was performed at T10 to expose the spinal cord, and precise incisions were made along the vertebral arch to remove the lamina. To stabilize the spine, a fixation device was used, and moderate spinal cord contusion was induced via the NYU Impact-III device (WM Keck, USA) by applying a free-fall force of 5 g from a 12.5 mm height. Postinjury, the wound was sutured in layers. The bladder was manually expressed for two weeks following the injury. All the animal experiments were approved by the Medical Ethics Committee and the Animal Welfare Ethics Committee of Tianjin Medical University General Hospital (IRB2021-KY-317, IRB2021-DWFL-354).

Tissue distribution

To assess the biodistribution of MSC-EXO_R&T-GA in an SCI mouse model, DiR dye (Life Technologies) was used for exosome labeling. DiR-labeled MSC-EXOs (30 μg), MSC-EXO_R&T (DiR-labeled MSC-EXOs incubated with RVG-CP05 and TAT-CP05 (30 μg each), MSC-EXO_R&T-GA, or MSC-EXO_R were injected intravenously into SCI mice. Two hours after injection, 50 mL of cold PBS was used to perfuse the mice to clear the circulating exosomes. Tissue samples, including the spinal cord, quadriceps, triceps, liver, spleen, heart, lungs, kidneys, and brain, were collected. Imaging was performed using an IVIS Spectrum system (PerkinElmer), and signal intensity was quantified using Living Image software. Regions of interest (ROIs) were analyzed to determine the signal distribution.

For spinal cord tissue analysis, MSC-EXOs were labeled with DiO, while RVG-CP05 and TAT-CP05 were labeled with FITC and rhodamine B. Multifunctional targeted exosomes (MSC-EXOs, MSC-EXO_R&T, and MSC-EXO_R&T-GA, 30 μg each) were injected into SCI mice or C57BL/6 control mice. Spinal cord tissues were fixed with 4% paraformaldehyde, dehydrated with graded sucrose solutions, and embedded for sectioning. Sections (10 μm thick) were visualized using a Zeiss 900 inverted confocal microscope (Germany) for analysis of MSC-EXO colocalization.

In vivo experimental design

To assess the therapeutic efficacy of these multifunctional exosomes, groups of mice were established for intravenous injection. These included the sham, equal-volume Injury, MSC-EXO, MSC-EXO_R&T, and MSC-EXO_R&T-GA groups. Exosomes were administered at a dose of 1 mg/kg (dissolved in 200 μL of PBS) via tail vein injection on Days 1, 3, 7, and 14 post SCI.

BMS score

To evaluate hindlimb motor function recovery, open field BMS scores were recorded for the mice prior to SCI [57], as well as on Day 1, Day 7, and weekly until Week 8. Scoring was conducted independently by two observers. The BMS ranges from 0 to 9, reflecting the degree of hindlimb functional recovery. A score of 0 indicates no voluntary hindlimb movement, whereas a score of 9 represents fully normal hindlimb function.

CatWalk analysis

The CatWalk XT system (version 10.6, Noldus, Wageningen, Netherlands) and its software were used to objectively assess gait, locomotion, and body coordination in the mice [58]. The system consists of a transparent glass runway, with a camera positioned beneath the glass runway to capture and record the mouse paw print data in real time. A red LED background above the platform enhances the visibility of the mouse’s body outlines and movement paths. All the experiments were conducted in a quiet, dimly illuminated environment. Prior to testing, the mice received training, which required at least three consecutive spontaneous crossings of the illuminated walkway. During the experiment, the camera-acquired paw print data were analyzed automatically using CatWalk XT software to evaluate gait and coordination.

Electrophysiological assessment

Electrophysiological testing was performed to assess the recovery of descending motor conduction pathways in 8 weeks post SCI. The mice were deeply anesthetized using pentobarbital, and a single electrical stimulus of 5 mA was applied to the motor cortex to elicit motor-evoked potentials (MEPs). The amplitude of the recorded MEPs were measured for further analysis using an electrophysiological system (YRKJ-G2008, Zhuhai Yirui Technology Co., Ltd.).

Louisville swim scale

The swim test was conducted to evaluate motor function recovery in mice after SCI [59, 60]. Each mouse was placed individually in a water tank maintained at 25 °C for 1 min, and their swimming behavior was observed. Hindlimb coordination, movement, and posture during swimming were evaluated on a scale from 0 to 3 by two independent, blinded observers. Each mouse performed three trials, and the average score was used for further analysis.

Hot-plate test

The hot-plate test was performed to assess sensory function recovery in the mice following SCI. Each mouse was placed on a heated metal plate set at 52 ± 0.5 °C. The time required for the mice to exhibit a nociceptive response, such as shaking, licking, or withdrawal of the hindlimbs, was recorded as the reaction time. A maximum time limit of 30 s was applied to avoid tissue damage. Each mouse experienced three trials with a 10 min interval between each trial. The average reaction time was calculated and analyzed.

Iron content detection

The tissue iron content was determined by tissue iron content assay Kit(Cat# BC4355; Beijing Solarbio Science & Technology Co., Ltd., China), with the iron concentration measured by a spectrophotometer. At the beginning of the experiment, approximately 0.1 g of tissue sample was weighed, and 1 mL of extraction solution was added for homogenization in an ice bath. The mixture was then centrifuged at 4,000 × g for 10 min at 4 °C, and the supernatant was collected for analysis. A visible spectrophotometer was preheated for 30 min, with the wavelength set to 520 nm, and calibrated to zero with distilled water. During sample preparation, the blank tube, standard tube, and sample tube were filled with distilled water, standard solution (0.125 mM Fe³⁺), and sample, respectively. Reagents one and two were added, followed by mixing and heating in a boiling water bath for 5 min. Chloroform was then added, and the mixture was shaken thoroughly. After centrifugation, the upper inorganic phase was collected, and its absorbance was measured. The tissue iron content was calculated by determining the change in absorbance (ΔA) on the basis of the difference in absorbance between the standard solution and the sample, along with the sample weight and protein concentration.

MDA detection

The MDA detection kit was used to perform the assay.(Cat# BC0025; Beijing Solarbio Science & Technology Co., Ltd., China). First, tissue samples of approximately 0.1 g were homogenized with 1 mL of extraction solution in an ice bath. For the cell or bacterial samples, ultrasound disruption was followed by centrifugation. The supernatant from the processed samples was used for subsequent analysis. The MDA detection working solution was prepared by dissolving reagent two in reagent one, thoroughly mixing, and then the mixture was stored. Next, the visible spectrophotometer or microplate reader was preheated, with the wavelength set to 532 nm, and calibrated to zero using distilled water. In accordance with the instructions of the kit, the sample, MDA detection working solution, and reagent three were added to the measurement tube, mixed thoroughly, and then incubated at 100 °C in a water bath for 60 min. After incubation, the samples were cooled to room temperature and centrifuged to remove impurities, and the supernatant was collected for absorbance measurement in a cuvette. Finally, the MDA content was calculated on the basis of the difference in absorbance (ΔA), specifically the difference between ΔA532 and ΔA600. The MDA content was then calculated according to the tissue weight, sample protein concentration, or cell count to ensure the accuracy of the experiment.

Perfusion, tissue processing, and sectioning

Fully anesthetized mice underwent a midline incision, to fully expose the heart by opening the chest and abdominal cavity. Cardiac perfusion was first performed with precooled PBS and continued until the drainage fluid was clear, indicating that any residual blood had been effectively washed out. Concurrently, the liver tissue gradually turned pale. Perfusion was then continued with precooled 4% paraformaldehyde.

The spinal cord was immersed overnight in a 4% paraformaldehyde solution for fixation. Following fixation, the spinal cord samples were subjected to dehydration using 30% sucrose solution. The dehydrated samples were embedded in optimal cutting temperature (OCT) compound and sectioned into 10-μm-thick slices via a cryostat (Leica CM3050S, Germany).

Immunofluorescence

Following TBST washes, the tissue sections were incubated in QuickBlock™ solution (P0260, Beyotime) for blocking. The sections were subsequently incubated with primary antibodies diluted to their working concentrations. The primary antibodies used included goat anti-GFAP (1:500, Abcam, ab53554), rabbit anti-NeuN (1:500, Abcam, ab104225), rabbit anti-Oligo2 (1:500, Abcam, ab254043), mouse anti-NF200 (1:400, Servicebio, GB12144-100), and rabbit anti-MBP (1:400, Servicebio, GB11226-100) antibodies.

After thorough washing with TBST, the sections were incubated for 2 h with appropriate secondary antibodies conjugated to Alexa Fluor 488 or Cy3. Additional TBST washes were performed to eliminate any unbound reagents, and the tissue sections were mounted with DAPI-stained glass slides. Imaging was conducted using a panoramic imaging system (Vectra Polaris, USA). Areas positive for GFAP, NeuN, Oligo2, NF200, and MBP were quantified via ImageJ software, and threshold analysis was applied to exclude background signals.

RNA-Seq analysis

TRIzol reagent (Invitrogen, USA) was used to extract total RNA from spinal cord tissues or cultured cells in accordance with the manufacturer’s instructions. An Agilent 2100 Bioanalyzer (Agilent Technologies, USA) and a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA) were used to assess the quality and amount of RNA. For library development, only samples with RNA integrity numbers (RINs) greater than 7.0 were utilized.

The NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs, USA) was used to create RNA-seq libraries. Oligo(dT) beads were used to enrich the mRNAs, which were then broken up and converted into cDNA. To produce 150 bp paired-end reads, the libraries were indexed, amplified by PCR, and sequenced using the Illumina NovaSeq 6000 platform.

The raw sequences were processed using FastQC and Trimmomatic for adapter trimming to guarantee data quality. Gene expression levels were measured using featureCounts after the cleaned reads were matched to the mouse reference genome (GRCm39) using STAR. DESeq2 was used to identify differentially expressed genes (DEGs) with an adjusted P value < 0.05 and a |log2(fold change)| threshold > 1.2. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were used to perform functional enrichment analysis of the DEGs, and ClusterProfiler in R was used for visualization.

UV–Vis calibration of GA in MSC-EXO_R&T-GA

MSC-EXO_R&T-GA, MSC-EXO_R&T and GA were dissolved in PBS and analyzed by ultraviolet–visible (UV–Vis) spectroscopy between 200 and 400 nm, revealing an absorption maximum for GA at 300 nm.

Serial dilutions of MSC-EXO_R&T-GA (1, 2, 4, 6, 8, and 10 μg/mL) were analyzed by UV–Vis spectrophotometry at 300 nm. A linear calibration curve (Y vs X) was generated, with the regression equation Y = 0.003926X + 0.7629 (R² = 0.9882).

Statistical analysis

The data are expressed as the mean ± SD. For comparisons among three or more groups, one-way analysis of variance (one-way ANOVA) was performed, followed by Tukey’s post-hoc test; while statistical comparisons between two groups were conducted using Student’s t-test. Details of the specific statistical analyses are offered in the figure legends. Significance thresholds were defined as follows: P > 0.05 (not significant, ns); P ≤ 0.05 (significant, *); P ≤ 0.01 (highly significant, **); P ≤ 0.001 (very significant, ***); and P ≤ 0.0001 (extremely significant, ****).

Sample sizes were determined on the basis of prior power analyses to ensure sufficient statistical power. All experimental procedures and analyses adhered to ethical guidelines to ensure reproducibility and consistency throughout the study.

Author contributions

B.F., X.G., X.C., and X.L. conceived and designed the study, performed the majority of the experiments, analyzed the data, and wrote the initial manuscript draft. P.W. and Y.R. contributed to the data collection, statistical analysis, and figure preparation. B.H. and J.L. assisted with experimental validation and literature review. N.R., H.D., and Y.R. (Yiming Ren) conducted the technical investigations and supported the data interpretation. Z.S., C.S., F.W., Y.H. provided critical revisions, resource acquisition, and methodological guidance. T.L., G.N., X.Y., and S.F. (corresponding authors) supervised the entire project, secured funding, and finalized the manuscript. All the authors reviewed and approved the final manuscript.

Funding

This work was supported by the National Key Research and Development Project of Stem Cell and Transformation Research (Grant No. 2019YFA0112100), the National Natural Science Foundation of China (Grant No. 82472414), and the Tianjin Key Medical Discipline (Specialty) Construct Project (Grant No. TJYXZDXK-027A). Additionally, it was funded by the Young Scientists Fund of the National Natural Science Foundation of China (Grant No. 82102563) and the Tianjin Natural Science Foundation (Grant No. 21JCQNJC01300; Grant No. 21JCYBJC00790; Grant No. 21JCYBJC01780). This study was supported by the Tianjin Health Research Project (Grant No. TJWJ2023QN002).

Data availability

The RNAseq datasets analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Medical Ethics Committee and the Animal Welfare Ethics Committee of Tianjin Medical University General Hospital. Medical Ethics Committee Approval Number: IRB2021-KY-317. Animal Welfare Ethics Committee Approval Number: IRB2021-DWFL-354. This study did not involve the use of human data or tissues; thus, the requirement for informed consent was not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.