Adult female C57BL/6 mice (8 weeks old) were obtained from the animal center at Central South University (Hunan, China). They were housed in a controlled specific pathogen-free (SPF) environment with a light/dark cycle. The animals have free access to food and water. All the procedures applied in this study were approved by the Animal Ethics Committee of Central South University and complied with relevant management regulations.

Neonatal mice (1 day after birth) were used for primary microglial isolation as previously described (Du et al., 2021). Briefly, mice were dissected to isolate the cerebral cortices and carefully peel off the meninges and blood vessels. Then, the brain tissues were cut using microscissors into pieces and digested with enzymes (8 U/ml papain and 125 U/ml DNase, Thermo Fisher Scientific, MA, United States) at 5% CO₂ and 37°C for 20 min. After the termination of the digestion, the cell suspension was filtered using a 40-μm cell strainer (Corning, NY, United States) and centrifuged at 200 g for 10 min. The cells were resuspended in a T25 culture flask with Dulbecco’s modified Eagle’s medium (DMEM, HyClone, United States) containing 10% fetal bovine serum (FBS, Gibco, United States) and 1% penicillin/streptomycin (Sigma, United States), and the supernatant was removed and washed twice with PBS the next day. On the fourth day, the culture flask was placed at 180 rpm (round per minute) for 30 min on a shaker, and the supernatant was transferred to a poly-d-lysine–coated 12-well plate. After 3 h of incubation, the supernatant was removed to obtain pure microglial cells and examined under an inverted microscope for morphological analysis.

Primary microglial cells were identified with immunocytofluorescence. The cells were fixed in 4% paraformaldehyde and rinsed in PBS. Primary microglia were incubated with primary antibodies against F4/80 and Iba-1 overnight at 4°C. Then, the slide was incubated with secondary antibodies. 4’, 6-diamidino-2-phenylindole (DAPI) was used for nuclear counterstaining. The antibodies used are listed in Supplementary Table S2.

For flow cytometry analysis, the isolated cells were immunolabeled with PerCP-Cy5.5-anti-CD11b antibody, FITC-anti F4/80 antibody, FITC-anti-CD31 antibody, APC-anti-O4 antibody, and PE-anti-GLAST antibody in flow cytometry staining buffer. For IBA-1 staining and analysis, cells were fixed in PBS containing 2% paraformaldehyde, permeabilized with 0.25% Triton X-100 and 1% BSA in PBS, and then immunolabeled with anti-IBA-1 antibody and Alexa Fluor 488-conjugated goat anti-rabbit IgG H&L (Alexa Fluor 488) in flow cytometry staining buffer. A flow cytometric analysis was performed on a BD FACSCanto II cytometer, and the data were analyzed using FlowJo software (FlowJo, LLC, Ashland, OR). The antibodies used are listed in Supplementary Table S2.

Microglia-Exos were acquired by gradient ultracentrifugation, according to a previous report (Peng et al., 2021). Briefly, microglia cells were cultured in complete medium containing EVs-free FBS for 48 h. The culture medium was centrifuged at 300×g for 10 min, 2000×g for 30 min, and 10,000×g for 30 min to sequentially remove cells, dead cells, and cellular debris. After filtering through a 0.22-μm filter (Millipore, Billerica, United States), the collected supernatant was ultra-centrifuged at 10,000×g for 2 h to completely remove supernatant, and then the exosome pellet was collected by resuspending in PBS. The protein content of exosomes was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, United States). The obtained exosomes were stored at −80°C for subsequent experiments.

Microglia cells were transfected with miR-125b-5p inhibitor or negative control (NC) inhibitor (Hanbio, China) at a concentration of 100 nM using Lipofectamine 3000 (Hanbio, China) after cell attachment. The sequences of miR-151-3p and negative control (NC) inhibitor were as follows: miR-151-3p-inhibitor (5′-CCU​CAA​GGA​GCC​UCA​GUC​UAG-3′); miR-151-3p-NC (5′-UCA​CAA​CCU​CCU​AGA​AAG​AGU​AGA-3′). 3 days post transfection, the total RNA from the microglia cells was extracted with TRIzol reagent (Invitrogen, United States) for further analysis.

For exosome identification, the purified exosomes were fixed with glutaraldehyde solution in PBS. After dropping on the copper net and staining with 2% phosphotungstic acid for 2 min, the exosomes were observed by transmission electron microscopy (FEI, United States). The diameters of exosomes were measured using a nanotracking analysis system. In addition, Western blotting was used to detect exosomal biomarkers, including CD9, CD63, and Tsg101.

For exosome uptake tracking, PKH67 and PKH26 kit (Sigma–Aldrich, San Louis, United States) fluorescence labeling of microglia-Exos was performed according to the instruction manual described. For in vitro uptake experiment, 50,000 primary neurons were seeded into a 24-well plate. 24 h later, 10 ug labeled exosomes were added into each well. After 48 h of incubation, neurons were fixed for in vitro exosome uptake experiments. For in vivo tracking, 200 μg exosomes in 100 ul saline were injected through the tail vein for three consecutive days after SCI. 3 days post injury, exosome-treated mice were terminated, and the spinal segment was harvested and dehydrated for sectioning and immunofluorescence analysis.

Primary cortical neurons were isolated from fetal mice (E14-16) as previously described (Guo Z. et al., 2021). Briefly, the cortical brains of the fetal mice were removed and placed into a cold PBS solution. Then, the meningeal tissues were carefully wiped off, and the brain tissues were transferred to a 15-ml centrifuge tube containing cold DMEM. They were gently blown upon using a 1-ml pipette for homogenization. After filtration using a 70-µm filter and centrifugation (1,000 rpm, 5 min), the supernatant was removed, and the cells were resuspended in a neurobasal medium (Cyagen, US) containing 2% B27 (Gibco, United States), 1% glutamine (Gibco, United States), and 1% penicillin-streptomycin (Sigma, United States). After cell adhesion (4 h), neurons were treated with myelin debris to simulate the in vivo microenvironment after spinal cord injury (SCI) to inhibit axon growth as previously described. 24 h after the neurons were seeded, 10 μg exosomes (exosomes/miR-151-3p^IN-exosomes) were incubated for 2 days with the cultured neurons for further Western blot, qRT-PCR, and neurite length measurement. The neurons in the control group were treated with the same volume of PBS. The average neurite length was calculated using ImageJ with a plugin of Ridge Detection by dividing the length of all neurites by the number of neurons.

The myelin debris was obtained following a previous research study (Rolfe et al., 2017). Briefly, after anesthesia, the mice were terminated by cervical dislocation. The brain tissues were harvested and cut into pieces in 0.32 M sucrose solution. Then, the tissues were homogenated by ultrasound, added into 0.83M sucrose solution, and centrifuged at 100,000 g at 4°C for 45 min. The crude myelin debris was homogenized again and resuspended into Tris-Cl buffer for centrifugation at 100,000 g at 4°C for 45 min. We discarded the supernatant and resuspended the myelin debris into sterile PBS for centrifugation at 220,000 g at 4 °C for 10 min. Then, we removed the supernatant, weighed the debris pellet, and suspended it to 100 mg/ml with PBS. Purified myelin debris was stored at −80°C for further experiments.

A modified Allen’s impact method was used to induce the spinal contusion SCI model as previously described (Li et al., 2021). Briefly, the mice were anesthetized with a ketamine/xylazine mixture (120 mg/3.3 mg/kg) solution. Then, a dorsal median incision was made at thoracic level 10 to expose the spinal cord carefully. After laminectomy at T10, a rod weighing 10 g was dropped onto the spinal cord at a height of 25 mm without creating any unnecessary pressure on the spinal cord using an NYU impactor; then the muscles and skin were sutured sequentially. Once the mice were awake, 1.5 ml (0.006 mg/ml) buprenorphine was applied subcutaneously for the first 24–48 h to relieve the pain caused by the surgery. The bladders should be manually expressed twice a day for the following week until the return of bladder function. For the sham group, the mice underwent laminectomy without establishment of the SCI model. For the treated groups, microglial cell-derived exosomes (200 μg exosome/miR-151-3p^IN-exosomes) were injected through the tail vein for three consecutive days after SCI establishment. The mice in the control group were treated with the same volume of saline through the same approach.

A TUNEL (TdT-mediated dUTP Nick End Labeling) apoptosis detection kit (Cy3, Beyotime, China) was applied to detect apoptotic cells in spinal sections. Slices were selected near the central canal horizontally. After fixation and permeation, the sections were incubated with TUNEL solution for 1 h at room temperature. Then, the sections were washed 3 times with PBS and stained with DAPI. The region of interest (ROI) was selected from the anterior horn of the spinal cord at the 500 μm rostrally adjacent to the injured epicenter. Apoptotic cells were observed using a fluorescence microscope (Zeiss, Oberkochen, Germany). ImageJ was used to count the total cells and TUNEL-positive cell numbers. The apoptosis of neurons in different treatment groups was measured using an Annexin V (FITC)/propidium iodide (PI) detection kit (BD Biosciences, United States) and analyzed using a flow cytometer (BD-LSRII, United States) as previously described (Zhang et al., 2016).

Total RNA was extracted from microglial cells (3 days post transfection), neurons (3 days post exosomal treatment), and microglia-Exos by using TRIzol reagent (Invitrogen, United States) and reverse transcribed to complementary DNA (cDNA) using a PrimeScript™ RT Reagent kit (Promega, United States) based on the manufacturer’s instructions. Reverse transcription of microRNAs was performed using an miRNA First Strand cDNA Synthesis kit (Tailing Reaction, Sangon Biotech, China). The expression of targeted genes or microRNAs was measured using a GoTaq qPCR Master Mix kit (Promega, United States) and a quantitative PCR system (ABI, United States). GAPDH served as the internal reference of P53, and U6 was the internal reference of microRNAs. The relative quantitative expression was analyzed using the 2 ^(−ΔΔCt) method. The primers used are listed in Supplementary Table S1.

Total protein of cultured neurons (3 days post exosomal treatment) and microglia-Exos was extracted by using RIPA lysis buffer (Beyotime, China), and the protein concentration was detected using a BCA kit. The denatured proteins were separated by 10% SDS–PAGE and then electrophoretically transferred to PVDF membranes (Millipore) according to previously described methods. After blocking with 5% fat-free milk, the membranes were incubated at 4°C overnight with primary antibodies against CD9, CD63, Tsg101, p53, p21, CDK1, Bcl-2, bax, cleaved-caspase3, and β-actin. The membranes were washed with 1% TBST 3 times and incubated with HRP-labeled secondary antibody at room temperature for 90 min. Enhanced chemiluminescence reagent (ECL) was used to visualize immunoreactive bands. β-actin was used as an internal control. The antibodies used in this study are listed in Supplementary Table S2, including the catalog numbers and indicated dilutions.

After being anesthetized with ketamine/xylazine mixture (120 mg/3.3 mg/kg) solution, mice were perfused with saline via the left ventricle to allow rapid and sufficient draining of blood flow, followed by 10% formalin solution for fixation. After perfusing, a 10-mm spinal segment including the injury epicenter was harvested and dehydrated for sectioning. For hematoxylin-eosin (HE) staining, the spinal segment was embedded in paraffin, horizontally sectioned at a thickness of approximately 8 μm, and stained according to the instructions of the HE staining kit (Solarbio, China). Images were captured using an Olympus photomicroscope (magnification, ×400). The cavity area of HE staining was outlined manually and calculated using ImageJ.

For immunofluorescence staining, the spinal segment was embedded in OCT compound (Takara, United States) and horizontally cut into 16-µm slices. The slices were incubated overnight with primary antibodies, including anti-MAP2, anti–5-HT, and anti-NeuN. After washing with PBS, the slices were then incubated with secondary antibodies. 4′,6-Diamidino-2-phenylindole, dilactate (DAPI; 1 μg/ml, Thermo Fisher Scientific) was used for nuclear counterstaining. The region of interest (ROI) of MAP2 was selected from the anterior horn of the spinal cord at the 500 μm rostrally adjacent to the injured epicenter. Images were captured using an Axio Imager one microscope (Zeiss, Oberkochen, Germany). The antibodies used in this study are in Supplementary Table S2.

Biotin-labeled miR-151-3p (Sangon Biotech, China) was constructed for fluorescent miRNA in situ hybridization as previously described (Guo M. et al., 2021). Briefly, formalin-fixed paraffin-embedded sections were cut into 8-µm thick sections. The slices were dewaxed with xylene and rehydrated in an ethanol gradient. Then, the slices were treated with proteinase K (20 μg/ml, Ambion, United States), followed by prehybridization for 1 h at 37°C. Then, the prehybridization buffer was removed and incubated with the miR-151-3p-biotin probe (sequence: 5′-biotin-CCUCAAGGAGCCUCAGUCUAG-biotin-3′) at 60°C. After overnight incubation, the slices were washed with 2 × SSC, 1 × SSC, and 0.5 × SSC. After blocking for 1 h at room temperature with blocking buffer (3% BSA in 0.1% PBST), the slices were incubated with Cy3-conjugated streptavidin (Jackson ImmunoResearch, United States) for 1 h. Anti-NeuN antibody was used to stain the nuclei of neurons, followed by the corresponding secondary antibody. DAPI was used to label the cell nucleus. The region of interest (ROI) was selected from the anterior horn of the spinal cord at the 500 μm rostrally adjacent to the injured epicenter in the horizontal sections. For quantification, three sections from different animals in each group were chosen to evaluate the relative immunofluorescent intensity ratio of miR-151-3p/Neun using ImageJ.

The p53 plasmid and mmu-miR-151-3p mimic were constructed by Hanbio Co., China. Briefly, wild-type 3′UTR of Trp53 containing the mmu-miR-151-3p binding site and the mutated binding sequence were cloned into pSI-Check2 plasmids (Hanbio Co., China), referred to as Trp53-WT and Trp53-MUT, respectively. A human renal epithelial cell line (293T cells) was cotransfected with 3′UTR luciferase reporter constructs (3′UTR- Trp53-wild type, 3′UTR- Trp53-mutant) and miRNAs (miR-151-3p-negative control (NC) or miR-151-3p-mimic). After 2 days of transfection, a Dual-Luciferase Reporter Gene Assay Kit (Beyotime, China) was used to detect the binding relationship between Trp53 and miR-151-3p based on the instructions. The luciferase activity of each group was measured using a microplate reader. The sequence of the miR-151-3p mimic was 5′ CUA​GAC​UGA​GGC​UCC​UUG​AGG3’. The sequence of miR-151-3p-NC is 5′UCA​CAA​CCU​CCU​AGA​AAG​AGU​AGA3’.

The BMS scoring system was applied to evaluate the locomotor function as previously described (Basso et al., 2006). Two well-trained investigators who were blinded to treatment assessed hindlimb movements. The BMS score ranges from 0 to 9 (0 represents complete paralysis, while nine indicates normal mobility). A thermal paw stimulation apparatus (IITC, United States) was used for thermal sensory function assessment. When the mice felt pain and withdrew their paw, the reaction time was recorded as the withdrawal threshold. The experiment was repeated 3 times with an interval of 10 min, and the average withdrawal threshold was taken as the final result. The locomotor and sensory function was assessed at 1, 3, 7, 14, 21, 28, 35, 42, and 56 days post injury.

Motor evoked potential (MEP) and latent period were used to assess neurological connectivity as previously described 56 days post-injury after treatment (Schlag et al., 2001). After the mice were anesthetized, we exposed the bregma of mice and used a stereotaxic device for positioning. Stimulating electrodes were placed on the surface of the skull at the point of 1 mm caudal, 0.5 mm lateral to the bregma and −4 mm caudal, 0.6 mm lateral to the bregma, separately. Recording electrodes were inserted into the tibialis anterior muscle of the contralateral hindlimb. Subcutaneous tissue between the stimulating and recording electrodes was used to insert the reference electrode. Electrical stimulation was performed at 3 V voltage and 333 Hz and repeated every 2 seconds. The MEP was calculated by summing the voltage values of the peak and the trough. The potential period was measured from the initiation time of the first response point to the wave starting to change drastically.

We analyzed the data using GraphPad Prism 8 software. A p value <0.05 indicates statistical significance. Data were shown as the mean ± standard deviations (SDs). Statistical comparisons of two groups were conducted using a two-tailed unpaired Student’s t-test. Differences among multiple groups were analyzed with one-way ANOVA followed by Tukey’s post hoc tests. The analysis of BMS scores and tactile sensory tests were analyzed with repeated-measures two-way ANOVA at different time points.

Primary microglial cells were isolated from the cerebral cortex of neonatal mice (1 day post birth) and showed a long spindle shape (Figure 1A). Immunocytochemistry demonstrated that the cells expressed the microglial markers IBA-1 and F4/80 (Figure 1B). To further identify the purity of isolated microglia, we used flow cytometry with the microglia positive markers F4/80, IBA-1, and CD11b and the negative markers glast, CD31, and O4 to exclude astrocytes, endothelial cells, and oligodendrocytes separately, as depicted in Figure 1C. These results confirmed the phenotypic characterization of primary microglia. Exosomes were isolated from microglial cell culture medium by ultracentrifugation (Figure 1D) and identified using TEM, nanoparticle tracking analysis (NTA), and Western blot analysis. As shown in Figure 1E, TEM images showed that microglia-Exos had a typical cup-shaped morphology. The diameter of the particles was in the range of 35–145 nm, which was further validated by nanotracking analysis (Figure 1F). Western blot analysis indicated that microglia-Exos expressed the exosomal markers CD9, CD63, and Tsg101 in isolated exosomes (Figure 1G).

To investigate the effect of Microglia-Exos on functional healing after SCI, we labeled Microglia-Exos with PKH26 and injected them via the tail vein for three consecutive days after injury (Figure 2A). Microglia-Exo–treated mice exhibited improved locomotor function recovery compared with the control group 2 weeks after injury until the end of observation, as shown by BMS scores (Figure 2B). In addition, as shown in Figure 2C, better sensory function recovery was observed in the microglia-Exo–treated group from 7 days after injury to the end of the experiment than in the control groups. In addition, we performed neuro-electrophysiology analysis to evaluate the neurological connectivity of the spinal cord to the hindlimb muscles at 56 days after injury among different treatment groups. As depicted in Figure 2D and Figure 2E, mice treated with microglia-Exos showed a higher motor evoked potential (MEP) and shorter latent period than those in the control group, indicating that microglia-Exos could promote nerve conduction capacity after SCI. The cavity area of the spinal cord in hematoxylin-eosin (HE)-stained sections was smaller in the microglia-Exo–treated group than in the control group (Figures 2F,G). Collectively, microglia-Exo treatment reduced the injured size of the cavity and promoted locomotor and sensory function recovery after spinal cord injury.

To explore the underlying mechanism of microglia-Exos on neurological functional recovery, we labeled microglia-Exos with PKH67 (Sigma, green) and added them to the neuron culture medium. As shown in Figure 3A, exosomes could be taken up by neurons (stained by Tuj-1). Myelin debris was one of the key factors leading to the inhibitory microenvironment for axon growth and neuronal apoptosis (Sosa et al., 2015; Wu et al., 2021). Here, we used purified myelin debris to simulate the inhibitory microenvironment in vivo after SCI. Myelin debris treatment increased neuronal apoptosis using Annexin/PI staining, and this effect was reversed after microglia-Exo treatment (Figures 3B,C). Western blot analysis also indicated that neuronal apoptosis was induced by myelin debris treatment, accompanied by increased levels of the apoptosis-related proteins Bax and cleaved-caspase-3 and decreased expression of the antiapoptotic protein Bcl2 (Figures 3F,G). Myelin debris inhibited cortical neuron axon regrowth. However, the decrease in axon regrowth with the administration of myelin debris was markedly blocked by the addition of microglia-Exos (Figures 3D,E). p53 has been reported to mediate cell apoptosis in CNS injury (Hong et al., 2010). The p53-p21 pathway induced by DNA damage could inhibit cyclin-dependent kinase (CDK) activities, thus resulting in cell apoptosis (Karimian et al., 2016). We measured the protein levels of the p53/p21/CDK1 signaling pathway by Western blotting. Myelin debris treatment increased the expression levels of p53 and p21 while decreasing cyclin-dependent kinase 1 (CDK1) in neurons, while this effect was attenuated when cells were treated with microglia-Exos (Figures 3F,G). To investigate whether exosomes affect neuronal apoptosis and axon growth through the p53/p21 signaling pathway, we used Aaptamine (MedChemExpress, US) at a concentration of 30 μm to simulate p53. Aaptamine is an alkaloid isolated from sponge aaptos, which is a competitive antagonist of α-adrenergic receptors to activate the P21 promoter independently of the p53 pathway (Aoki et al., 2006). As shown in Supplementary Figure S2A,B, compared to the microglia-Exo–treated group, expression level of p21 and apoptosis-related proteins Bax and cleaved-caspase-3 in neurons was increased, while expression of CDK1 and antiapoptotic protein Bcl2 was decreased in the microglia-Exos + Aaptamine–treated group. In Supplementary Figure S2C,D of Annexin/PI staining and Supplementary Figure S2E,F of Tuj-1 staining, neurons in the microglia-Exos + Aaptamine–treated group showed an increased apoptosis cell rate and decreased axon regrowth compared to the microglia-Exo–treated group. These results indicated that microglia-Exos can enhance axon growth and inhibit neuronal apoptosis by suppressing the apoptosis-related p53/p21/CDK1 pathway.

To further explore the underlying mechanism in vivo, we injected PKH26-labeled exosomes through the tail vein after SCI. Fluorescence images of frozen spinal sections showed that PKH26-labeled exosomes mainly accumulated in neurons after SCI. The neurons in the injured spinal cord took up more microglia-Exos than those in the sham group (Figures 4A,B). As shown in Figure 4C, the apoptosis rate was markedly increased after SCI but decreased after microglia-Exos administration, as shown by TUNEL staining (Figures 4C,D). SCI usually causes neuronal loss, which is related to neuronal apoptosis. We sequentially counted the NeuN⁺ cells in areas of 0–500 μm, 500–1,000 μm, 1,000–1,500 μm, 1,500–2000 μm, and 2000–2,500 μm roaster from the epicenter. Compared to the sham group, the number of NeuN⁺ cells decreased after injury but was ameliorated after microglia-Exo treatment (Figures 4E,F). These results showed that microglia-Exos have a neuroprotective effect and could protect neurons from apoptosis induced by SCI. To investigate whether microglia-Exos could promote axon regrowth in vivo, we performed fluorescence staining with MAP2 and 5-hydroxytryptamine (5-HT), separately. We found that microglia-Exos could increase the MAP2-positive area (Figures 4G,H) and the 5-HT–positive area (Supplementary Figure S3A,B) compared with that in the control group. Collectively, these data suggested that microglia-Exos could protect against neuron loss and facilitate axon regrowth by suppressing neuronal apoptosis.

To explore the key components of microglia-Exo cargo that mediate the neuroprotective effects of microglia-Exos on neurons, we comprehensively analyzed the miRDB, TargetScan, and microT databases. We found 169 potential microRNAs that could bind to p53 (Figure 5A). By comparing the microRNA expression profiles of microglia-Exos in public databases, the top 23 highly expressed microRNAs were selected and are listed in Figure 5B. We chose the top eight expressed microRNAs for our research of interest and validated them using qRT-PCR. Among them, the relative expression of miR-151-3p was highest in microglia-Exo cargo (Figure 5C). To further determine whether microglia-Exos could deliver miR-151-3p to neurons, we cultured neurons by adding microglia-Exos and performed qRT-PCR to measure the expression level of miR-151-3p in neurons. We found that the expression level of miR-151-3p in neurons was markedly decreased with added myelin debris and was increased under treatment with microglia-Exos (Figure 5D). Similar to the real-time PCR results, the FISH experiment for miR-151-3p detection in the in vivo SCI model showed the same trend. As shown in Figure 5E, miR-151-3p was mainly expressed in neurons, and it was downregulated in the SCI groups. Microglia-Exo administration significantly increased miR-151-3p in neurons compared with the control groups. Collectively, these results suggested that microglia-Exos could deliver miR-151-3p to neurons and increase the expression of miR-151-3p in neurons.

To investigate the function of exosomal miR-151-3p in mediating the neuroprotective effect of microglia-Exos on neuronal and axonal growth, we constructed a miR-151-3p inhibitor (miR-151-3p^IN) and transfected it into microglia. Microglia cells transfected with the miR-151-3p inhibitor (miR-151-3p^IN microglia) or the miR-151-3p negative control (miR-151-3p^NC microglia) showed no difference in morphology (Supplementary Figure S1A). TEM, nanotracking analysis, and Western blot were used to identify the exosomes isolated from miR-151-3p^IN microglia and miR-151-3p^NC microglia, and we found no difference in exosome diameters and surface markers between the two groups (Supplementary Figure S1B,D). We then extracted RNA from microglial cells and microglial exosomes for qRT–PCR to verify the inhibitory effect of the miR-151-3p inhibitor. As shown in Figure 6A and Figure 6B, the expression of miR-151-3p in microglia and microglia-Exos was obviously decreased 3 days after transfection with the miR-151-3p inhibitor. The 3′UTR of p53 shared a complementary sequence of miR-151-3p (Figure 6C). The dual luciferase reporter assay was used to examine the binding relationship between miR-151-3p and p53. As presented in Figure 6D, miR-151-3p reduced luciferase activity in cells carrying wild-type p53 plasmids when compared with the miR-151-3p negative control (NC). Furthermore, no change in luciferase activity was seen in cells carrying mutant p53 plasmids. These results indicated that miR-151-3p could directly bind to p53.

To investigate whether exosomal miR-151-3p was involved in the downregulation of the p53/p21 signaling pathway and apoptosis-related proteins in neurons, we added microglia-Exos or miR-151-3p inhibitor-transfected microglia-derived exosomes (miR-151-3p^IN−Exos). As shown in Figure 6E, qRT–PCR data indicated that miR-151-3p^IN−Exos could upregulate p53 gene expression in neurons, which was downregulated in the microglia-Exos treatment groups. Similar to the qRT–PCR results, Western blot analysis showed that miR-151-3p^IN−Exos obviously abolished the effect of microglia-Exos at the protein level of p53. (Figures 6F,G). In addition to p53, the miR-151-3p^IN-Exos promoted the protein expression of p21 and p53, upregulated the expression of apoptosis-related proteins, such as Bax and cleaved caspase-3, and downregulated the expression of the antiapoptotic protein Bcl2 (Figures 6F,G). Flow cytometry for neuronal apoptosis demonstrated that miR-151-3p^IN−Exos abolished the effect of microglia-Exos on the inhibition of neuronal apoptosis (Figures 6H,I). Immunofluorescence staining indicated that miR-151-3p^IN−Exos abolished the ability of microglia-Exos to promote axon growth (Figures 6J,K). Taken together, these results demonstrated that microglial exosomal miR-124-3p inhibited neuronal apoptosis and promoted axonal growth by suppressing p53 by directly targeting its 3′-UTR and thereby modulating the p53/p21/CDK1 signaling pathway.

To explore the protective effect of exosomal miR-151-3p on the neuroprotection of microglia-Exos in the injured spinal cord, we injected microglia-Exos or miR-151-3p^IN-microglia-Exos via the tail vein into SCI mice. The expression of miR-151-3p in neurons was significantly lower in the miR-151-3p^IN-Exo–treated group (Figures 7A,B) and was associated with more apoptotic cells and neuronal loss than the microglia-Exo–treated group (Figures 7C–F). For axon visualization, the mice treated with miR-151-3p^IN-Exos showed a smaller MAP2-positive area (Figures 7G,H) and 5-HT–positive area (Supplementary Figure S3C,D) than the microglia-Exos (NC-Exos)-treated group. To determine whether miR-151-3p was involved in microglia-Exo–mediated neurological function healing after SCI, we conducted BMS score and neuro-electrophysiology analyses. After SCI, the BMS score in each group increased gradually, indicating that mice had a certain self-recovery ability. Compared with the microglia-Exo–treated (NC-Exos) group, the recovery of the miR-151-3p^IN-Exo–treated group was relatively worse from the second week until the endpoint of observation (Figure 8A). The results of neuro-electrophysiology analysis showed similar results, and miR-151-3p^IN-Exos abolished the beneficial effect of microglia-Exos on promoting neurological functional healing after SCI and demonstrated a decreased MEP and increased latent period when miR-151-3p was downregulated in microglia-derived exosomes (Figures 8B,C). Histological analysis of HE-stained spinal cord sections showed a larger cavity area in miR-151-3p^IN-Exo–treated mice than in microglia-Exo–treated mice (Figures 8D,E). Therefore, these results indicated that exosomal miR-124-3p mediates microglia-Exos to promote neurological functional healing effects after SCI.