Human umbilical cord mesenchymal stem cell-derived exosomal miR-146a-5p reduces microglial-mediated neuroinflammation via suppression of the IRAK1/TRAF6 signaling pathway after ischemic stroke

1Ischemic stroke is a common cause of severe disability and death worldwide. To reduce primary injury due to acute ischemic stroke and limit infarct size, timely reperfusion therapy (thrombolysis or thrombectomy) is required. However, post-ischemic reperfusion itself causes damage and dysfunction in a process known as cerebral ischemia-reperfusion (I/R) injury. In this process, reperfusion triggers an inflammatory cascade, a key mechanism contributing to secondary neuronal damage and death following the initial ischemic episode. Overzealous up-regulation of endogenous neuroinflammatory processes leads to destruction of hypoxic tissue, induction of apoptosis, and initiation of an inflammatory cascade feed-forward loop that can enlarge the damaged region.

2Microglia, the major central nervous system resident innate immune cells, play an important role in modulating neuroinflammation. Cerebral IR-activated microglia undergo polarization into either “classically activated” M1 or “alternately activated” M2 phenotypes. While M1 microglia produce a large number of pro-inflammatory factors (e.g. IL-6, TNF-α, and IL-1β) that impede post-injury neural regeneration and produce poorer long­term neurological outcomes, M2 microglia are able to perform efferocytosis and produce numerous protective and trophic factors that promote neurogenesis. Decreasing microglial-mediated neuroinflammatory injury by targeting M1/M2 polarization may therefore have therapeutic potential in ischemic stroke.

3Our previous meta-analysis revealed that mesenchymal stem cell (MSC) transplantation after ischemic stroke significantly improves neurological deficits and quality of life. In addition, our previous original research demonstrated that MSCs are able to protect brain tissue, regulate inflammatory responses after traumatic brain injury and stroke, and can regulate microglia-mediated inflammatory responses. However, the mechanisms by which MSCs modulate microglia-mediated inflammation remain unclear. For the above reasons, elucidation of these mechanisms would be of significant benefit.

4Relative to human bone marrow-derived MSCs (BMSCs), human umbilical cord MSCs (hUMSCs) are more readily obtained, exhibit superior viability, are compatible with therapeutic methods featuring higher levels of patient acceptability and compliance, and are not susceptible to immune-mediated graft rejection. In addition, although hUMSCs are considered more primitive than bone marrow derived MSCs, they do not induce teratomas, but do exhibit immunomodulatory capabilities. Stroke occurs frequently at 45-65 years old, and there is an autologous bone marrow aging problem. In clinical research, it can avoid pain during bone marrow extraction and enhance volunteer compliance. We therefore selected hUMSCs for use in the current research.

5Although information regarding MSC therapeutic mechanisms of action is limited and fewer than 1% of transplanted MSCs reach and engraft at target sites (since most become entrapped in the pulmonary capillary bed due to their large size), their therapeutic effect is nonetheless frequently observed, currently mainly attributed to MSC-produced paracrine factors. We reasoned that transplanted MSC-derived exosomes (Exos) generated in vivo may be one factor contributing to the distal paracrine therapeutic effects of stem cell therapy.

6Small non-coding microRNAs (miRNAs) can act as inhibitors of mRNA transcription and protein translation in most cell types. Cell-derived exosomes contain numerous miRNAs that are able to act both locally and - via entering circulation - at distal sites. Exosomes are also internalized by neighboring or distal cells, thereby modulating the function of these recipient cells. A recent study demonstrated that MSC-derived exosome content member miR-146a decreases inflammation and enhances anti-sepsis therapeutic efficacy. Our sequencing data demonstrate that hUMSC-derived exosomes contain large amounts of miR-146a-5p. We thus hypothesized that hUMSC-derived exosomes (hUMSC-Exos), via provision of miR-146a-5p to microglia and consequent regulation of microglial gene expression, decrease microglia-mediated inflammation in the ischemic mouse brain. We injected wild-type and miR-146a-5p knockdown (miR-146a-5p k/d) hUMSC-Exos into ischemic mice to test this hypothesis as well as to explore mechanisms of potential therapeutic activity.

7In vitro, hUMSCs were expanded to the third and fifth generations (Figure 1A). Flow cytometry was used to characterize hUMSC surface phenotype (Figure 1B). The majority of hUMSCs were positive for expression of CD73 (100%), CD105 (99.96%), and CD90 (100%), and negative for expression of CD45 (0.06%), HLA-DR (0.50%), CD34 (0.06%), CD11b (0.06%), and CD9 (0.46%). Such results are representative of the known hUMSC phenotype, confirming hUMSC identity. After removal of dead cells and debris from the hUMSC-conditioned culture medium, secreted extracellular vesicles were harvested by differential centrifugation. Such vesicles displayed typical exosomal features, including morphology and size (30-150 nm) as detected by electron microscopy (Figure 1C) and NanoSight analysis (Figure 1D), and the presence of Exos markers CD9, Alix, and TSG101 as detected by western blot (Figure 1E). Therefore, they were designated hUMSC-Exos.

8We examined the in vivo effects of hUMSC-Exos in a murine model of ischemic stroke. Mice were randomly allocated to vehicle-only (phosphate-buffered saline (PBS)) or experimental (hUMSC-Exo in PBS) groups, and interventions were intravenously administered via the tail vein four hours post-reperfusion (Figure 2A). Seventy-two hours post-reperfusion, neurological function scores were examined. Lower Bederson scale and higher grip strength test scores (Figure 2C, P < 0.05) were observed in the experimental group. After experimental animal euthanasia, brain slices were obtained to assess infarct volume via 2,3,5-triphenyltetrazolium chloride (TTC) staining, demonstrating smaller infarct size in the experimental group (Figure 2B, p < 0.05). Additionally, the area of ischemic penumbra was smaller in the experimental group (Figure 2D). Furthermore, hematoxylin and eosin (H&E) staining demonstrated lower levels of tissue edema and cell edema and fewer contracted nuclei in the experimental group (Figure 2E). Immunohistochemical detection of IL-6 and NFκB in the ischemic penumbra demonstrated significantly lower expression of these proteins in the experimental group (Figure 2F, p < 0.05). Finally, infiltration of PKH26 (red)-labeled intravenous Exos into the site of brain injury was demonstrated by fluorescence microscopy (Figure 2G). Taken together, these results suggest that circulating hUMSC-Exos infiltrate the relevant anatomical site and are protective against I/R injury after ischemic stroke, in part by decreasing levels of local neuroinflammation.

9We next examined the effect of hUMSC-Exos on activated microglia in vivo. Treatment with hUMSC-Exos markedly decreased the presence of IBA-1+CD16+ cells at 72 h post-stroke (Figure 2H, p < 0.05) but markedly increased the presence of IBA-1+CD206+ cells at 72h post-stroke (Figure 2I, p < 0.05). Since IBA-1 is a marker of brain microglia, CD16 is an M1 marker, and CD206 is an M2 marker, this suggests that hUMSC-Exos treatment decreased and increased, respectively, the number of M1 and M2 microglia. Expression of pro-inflammatory cytokines IL-6, TNF-α, and IL-1β was also significantly decreased in the experimental group (Figure 2J, p < 0.01). Taken together, these results suggest that hUMSC-Exos may decrease microglia-mediated neuroinflammation after ischemic stroke in mice.

10First, we tested whether red fluorescent dye (PKH26)-labeled hUMSC-Exos are internalized during co-culture with BV2 microglia. After 6 h, microglia had efficiently internalized hUMSC-Exos as indicated by intracellular fluorescence (Figure 3A). To further validate the direct effects of hUMSC-Exo on activated microglia, the latter were cultured in serum-free medium for 6 h in a hypoxic incubator (to mimic oxygen-glucose deprivation (OGD)) prior to culture in conventional medium with or without hUMSC-Exos. After 24 h, IL-6, TNF-α, and IL-1β intracellular transcription and supernatant protein levels were determined by real-time polymerase chain reaction (RT-PCR) and enzyme-linked immunosorbent assay (ELISA), respectively. Treatment with hUMSC-Exos significantly decreased both IL-6, TNF-α, and IL-1β transcription and protein levels (Figure 3B, 3C, p < 0.01). Results suggest that hUMSC-Exos decrease microglial pro-inflammatory activity in vitro.

11In an attempt to elucidate at least one mechanism by which hUMSC-Exos modulate microglial activity, we investigated whether hUMSC-Exosomal miRNAs attenuate microglial pro-inflammatory activity in vitro. To demonstrate that miRNAs are a key functional component of these Exos, we conducted small interfering RNA (siRNA) knockdown of Drosha (an essential polymerase required for miRNA synthesis) in hUMSC (Figure 4A, 4B) in order to generate miRNA-depleted hUMSC-Exos (Figure 4C, 4D). Treatment of activated microglia with wild-type or Drosha-knockdown hUMSC-Exos demonstrated that miRNA depletion significantly weakened the anti-inflammatory effect of hUMSC-Exos (Figure 4E). Results suggest that hUMSC-Exosomal miRNAs contribute to microglial modulation.

12To determine which hUMSC-Exosomal miRNAs may contribute to microglial polarization and attenuation of microglial pro-inflammatory activity, BV2 microglia were exposed to OGD with or without subsequent hUMSC-Exos treatment for 24 hours, and hUMSC-Exosomal small RNA expression analysis and deep sequencing were performed. The following formula was used to calculate corrected miRNA expression: reads per million (RPM) = (number of reads mapping to miRNA/number of clean reads) *106. To the best of our knowledge, this is the first study to report sequencing of hUMSC-Exos-derived miRNAs. The analysis revealed that several hundred miRNA species are present within hUMSC-Exos. Among the top ten miRNAs identified as significantly differentially expressed between hUMSC-Exos treated and untreated microglia (Figure 5A), miR-146a-5p was higher in the treatment group and is known to modulate inflammation. As demonstrated by PCR, treatment with hUMSC-Exos significantly increases BV2 microglial miR-146a-5p content (Figure 5B). Results suggest that hUMSC-Exosomal miR-146a-5p is internalized by microglia (or that hUMSC-Exos induce microglial miR-146a-5p expression) and may contribute to modulating inflammation. The in vitro experimental scheme is pictured (Figure 5C). Western blot analysis demonstrated decreased IL-6, TNF-α, and IL-1β, as well as IRAK1/TRAF6 signaling pathway member IRAK1, TRAF6, and NFκB (p65) protein expression after hUMSC-Exos treatment. Furthermore, expression of IL-6, TNF-α, and IL-1β was higher after treatment with miR-146a-5p knockdown hUMSC-Exos (Figure 5D–5F). Results suggest that hUMSC-Exosomal miR-145-5p contributes to modulating OGD-induced microglial pro-inflammatory activity via suppression of the IRAK1/TRAF6 signaling pathway.

13To investigate whether hUMSC-Exosomal miR-146a-5p exerts neuroprotective effects in the ischemic mouse brain via IRAK1/TRAF6 signaling pathway modulation, both wild-type and miR-146a-5p knockdown hUMSC-Exos were administered in a murine ischemic stroke model. Relative to treatment with miR-146a-5p knockdown hUMSC-Exos, treatment with wild-type hUMSC-Exos significantly reduced infarct volume three days post-ischemia (Figure 6A) and resulted in lower Bederson scale scores and higher grip strength test scores (Figure 6B, 6C). Results suggest that hUMSC-Exosomal miR-145a-5p may attenuate I/R damage and neural deficits after ischemic stroke in a murine model.

14Immunohistochemistry demonstrated that levels of IL-6 and NFκB in the ischemic penumbra were significantly lower in the group treated with wild-type hUMSC-Exos (Figure 6D, 6E, p < 0.01). Furthermore, immunofluorescent labeling demonstrated that significantly fewer microglia were activated in the group treated with wild-type hUMSC-Exos (Figure 6F, p < 0.01). Western blots demonstrated that expression of IRAK1/TRAF6 signaling pathway proteins IRAK1, TRAF6, and NFκB (p65) was significantly decreased in the group treated with wild-type hUMSC-Exos (Figure 6G, p < 0.001). Finally, levels of IL-6, TNF-α, and IL-1β were significantly reduced in the group treated with wild-type hUMSC-Exos (Figure 6H, p < 0.001). Results suggest that hUMSC-Exosomal miR-146a-5p may attenuate microglia-mediated neuroinflammation after ischemic stroke in a murine model.

15In vivo, OGD activates microglial-mediated inflammatory response. During the acute period after stroke, microglia secrete pro-inflammatory cytokines IL-6, TNF-α, and IL-1β, which can induce secondary cytotoxicity. Studies have proven that decreasing microglia-mediated neuroinflammation is beneficial during stroke recovery. Although MSC transplantation is neuroprotective after both traumatic brain injury and stroke, at least in part via modulating microglia-mediated neuroinflammation, mechanisms of the latter remain incompletely understood. However, since over 99% of transplanted MSCs become entrapped in the pulmonary vasculature without impeding therapeutic effect, mechanisms likely involve distally acting MSC-produced paracrine factors which may hold promise as cell-free therapies. Nearly all cell types secrete exosomes, which are important mediators of cellular communication.

16One important group of Exos cargo molecules is miRNAs (short non-coding RNAs that inhibit target gene expression by directly binding their mRNAs). Prior studies report that Exos play an important role in transmitting miRNAs between cells, via interstitial fluid and circulation delivering biologically active miRNAs to both neighboring and distant cells. For example, adipose tissue constitutes an important source of circulating exosomal miRNAs that serve as a previously unrecognized form of adipokine to regulate gene expression in distant tissues, adipose tissue-resident macrophage-derived exosomal miRNAs modulate insulin sensitivity, and exosomal transfer of miR-181b from cardiosphere-derived cells (CDCs) into macrophages reduces PKCδ transcription (a mechanism underlying the post-reperfusion cardioprotective effects of CDCs). The present study investigated the potential therapeutic role hUMSC-Exos in ameliorating I/R injury.

17Results suggest that in a murine model of ischemic stroke, quality-controlled hUMSC-Exos delivered intravenously four hours post-reperfusion are able to traverse the blood-brain barrier to access the site of ischemic injury and are then taken up by local microglia, in which exosome-derived miR-146a-5p inhibits IRAK1/TRAF6 signaling pathway-mediated NFκB activation and consequent M1 polarization and production of potent pro-inflammatory cytokines (instead favoring M2 polarization), ultimately resulting in decreased: I/R-induced tissue edema, cell death, extent of the ischemic infarct and penumbra, and functional motor deficits.

18This is consistent with prior studies which have demonstrated that immunofluorescently labeled Exos are detectable both extracellularly and intracellularly at sites of brain injury and are taken up by recipient cells. It has also previously been demonstrated that MSC-derived Exos exert powerful effects in the context of ischemic stroke, for example ameliorating inflammation-induced astrocyte alterations. Regarding mechanisms, pro-neurogenic effects of UMSC-Exos may be partially attributable to histone deacetylase 6 (HDAC6) inhibition by exosomal miR-26a. Furthermore, it is known that without miRNA-184 and -210, MSC-derived extracellular vesicles lose the ability to promote neurogenesis and angiogenesis. Although it is difficult to completely exclude the effects of other exosomal cargo molecule groups on microglia in the present study, miRNAs are considered a key functional element. The miRNA miR-146a-5p is a well-known anti-inflammatory molecule with a key role in inflammatory disorders. The receptor proteins IRAK1 and TRAF6 are abundant in the cytoplasm and nucleus of various cell types. They are largely involved in Toll-like receptor (TLR)-initiated pathways, leading to expression of pro-inflammatory mediators. Overexpression of IRAK1 and TRAF6 can activate NFκB, a transcription factor which is a key activator of pro-inflammatory gene expression programs. Through binding the 3′UTR of the mRNAs encoding IRAK1 and TRAF6, miR-146a-5p down-modulates inflammatory responses.

19Apart from recommending follow-up research into the therapeutic potential of cell-free miR-146a-5p in the context of I/R injury and inflammatory disorders, results raise a number of interesting theoretical questions. For example, it is unknown whether hUMSC-Exosomal cargo is also perhaps involved in regulating glial ion transporters (which are involved in the Warburg effect, glial activation, neuroinflammation, and neuronal damage during glioma). Additionally, might other miRNAs impact microglial function after ischemic stroke? Might hUMSC-Exos also impact the function of other local cell types (e.g. neurons, astrocytes, and/or oligodendrocytes) after ischemic stroke? Finally, by which mechanisms might such effects occur?

20In conclusion, hUMSC-Exos and or miR-146a-5p represent novel therapeutic options for the improvement of outcomes after ischemic stroke (Figure 7), warranting further investigation.

21Experimental animals (C57BL/6 mice aged 8 weeks, weighing 20-30 g each) were purchased from the Animal Experiment Center of Southern Medical University (Guangzhou, China), fed a standard laboratory diet with ad libitum access to food and water, and maintained at a temperature of 22 ± 1° C and a humidity of 65–70 % in a controlled room with a 12 h light–dark cycle. All experimental procedures were approved by the Southern Medical University Ethics Committee and performed in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals.

22As described previously, mice were anesthetized via intraperitoneal injection (35 mg/kg sodium pentobarbital). Transient middle cerebral artery occlusion (MCAO) was produced by advancing a 4-0 nylon monofilament (0.23–0.25 mm) (Yushun Bio Technology Co. Ltd., China) via the left common carotid artery to occlude the middle cerebral artery for 60 minutes, prior to filament withdrawal (reperfusion). Success of blood-flow occlusion and restoration was verified using a laser Doppler flowmeter (Moor LAB, Moor Instruments, Devon, UK). During the MCAO procedure, head temperature was maintained at 36° C. At four hours post-reperfusion, 250 μL PBS with or without 50 μg hUMSC-Exos was injected into the tail vein (experimental and vehicle-only groups). Mice in a third (control) group received no injection.

23At 24 hours post-reperfusion, global neurological and motor function was assessed (Table 1) using a modified Bederson Scale and grip strength evaluation by a researcher blinded to group allocation. For grip strength evaluation, taut string (50 cm) was suspended between two vertical supports at a height of 40 cm. Each mouse was placed midway on the string and rated as shown in Table 1.

24At 72 hours post-reperfusion, mice were euthanized. Thereafter, the brain was removed for coronal sectioning. Infarct size was measured using 2% 2,3,5-triphenyltetrazolium chloride (TTC) staining in conjunction with microscopy. Infarct volume was evaluated by a blinded observer using ImageJ software version 1.61 (National Institutes of Health, Bethesda, MD, USA).

25Fresh brain tissues were fixed using 4% PFA (pH 7.4), gradually dehydrated, embedded in paraffin, cut into 4-μm-thick sections using a microtome, and stained with H&E to visualize cellular structures by microscopy.

26Four-micron-thick coronal brain sections were deparaffinized in xylene, rehydrated via an alcohol gradient, and washed with PBS (0.01 M, pH 7.4). Sections were blocked using 5% bovine serum albumin (BSA) for 60 min at room temperature prior to overnight incubation at 4° C with the following primary antibodies: goat anti-IBA-1 (1:200 dilution; ab48004, Abcam), rabbit anti-CD16 (1:200 dilution; ab252908, Abcam), and goat anti-CD206 (1:200 dilution, R&D Systems). Automated image analysis was performed using ImageJ software version 1.61 (National Institutes of Health, Bethesda, MD, USA). Cell numbers were calculated per square millimeter from three random microscopic fields (200 × magnification) on three sections (a total of nine images) (n = 6 animals per group). All counts were performed in a blinded manner.

27Human umbilical cords were provided by Guangzhou Saliai Stem Cell Science and Technology Co. Ltd. for hUMSC isolation. Thereafter, hUMSCs were cultured in complete Dulbecco’s modified Eagle’s medium (DMEM)/F12 (Gibco) supplemented with 10% fetal bovine serum (FBS) (Gibco, Australia), and cells between the 3rd and 5th generations were used for subsequent experiments. The BV2 murine microglial cell line was obtained from Xiehe Medical University (Beijing, China). Microglia were cultured in DMEM supplemented with 10% heat-inactivated FBS, 100 μg/mL streptomycin, and 100 U/mL penicillin (HyClone).

28After hUMSCs were cultured in DMEM/F12 medium supplemented with 10% (FBS), culture supernatants containing Exos were centrifuged at 2,000 x g for 20 min at "4° C, followed by 10, 000 x g for 30 min at 4° C, and the supernatant was passed through a 0.22 μm filter (Millipore), to pellet and exclude contaminating dead cells and debris. Thereafter, the filtrate was subjected to ultracentrifugation at 110,000 g for 70 min at 4° C to pellet Exos. Pellets were washed with PBS, followed by ultracentrifugation at 110,000 × g for 70 min at 4° C to re-pellet Exos, which were then resuspended in PBS. The bicinchoninic acid (BCA) Protein Assay kit (KeyGEN BioTECH) was used to estimate Exos concentration. For transmission electron microscopy (JEM-1200EX, JEOL Ltd.), 5-10 μl of each sample was added to a copper mesh and precipitated for 3 min. Remaining liquid was carefully pipetted from the filter paper edge. Thereafter, filter paper was rinsed with PBS and phosphotungstic acid was used for negative staining prior to drying at room temperature for 2 min and imaging (operating voltage: 80-120 kV).

29Exos were labeled using the red fluorescent membrane dye PKH67 (Sigma). Labeled Exos were washed by resuspension in 10 mL PBS, pelleted by ultracentrifugation as described above, and finally resuspended in 100 μl PBS. For cell treatment, a volume of suspension corresponding to 2 μg of Exos was added to 2×105 recipient cells prior to 24 hours of incubation.

30An anaerobic chamber containing 95% N2 and 5% CO2 was used in conjunction with deoxygenated glucose-free DMEM (Gibco) was used to simulate microglial OGD. After 6 hours, culture medium was replaced with maintenance medium, and cells were moved to a regular incubator to recover for 24 hours.

31Total RNA was extracted from cells and brain tissues using Trizol reagent (Life Technologies) and from Exos using an Exosome RNA Purification Kit (EZB-exo-RN1), both according to the manufacturer’s instructions. Reverse transcription was performed using the PrimeScript RT reagent Kit (RR037A, Takara Bio Inc., Shiga, Japan). Real-time PCR was conducted using SYBR Green PCR Master Mix (Applied TaKaRa, Otsu, Shiga, Japan) in conjunction with an Applied Biosystems 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Sequences of all primers are provided (Table 2).

32Concentrations (pg/mL) of IL-6, TNF-α, and IL-1β in BV2 microglia culture supernatants and damaged cerebral tissue were determined via ELISA (kits from R&D Systems, Minneapolis, MN, USA), performed as per manufacturer instructions. Briefly, standards and samples were added to a 96-well ELISA plate pre-coated with biotinylated anti-IL-6, anti-TNF-α, and anti-IL-1β. Unbound substances were washed away, and enzyme-linked polyclonal antibodies specific for IL-6, TNF-α, and IL-1β were added to the corresponding wells. Plates were incubated for 2 hours, washed four times, and enzyme substrate was added prior to 30 min incubation. Color development was terminated using stop solution and absorbance at 450 nm was determined using a microplate reader. The concentration of each sample was calculated from a standard curve prepared using the cytokine standards. Each experiment was performed in triplicate.

33In order to decrease miRNA synthesis, siRNA-Drosha (Ribobio) was transfected into recipient hUMSC using riboFECTTMCP Reagent (Ribobio). After 24 hours, transfection efficiencies were evaluated via qPCR or western blots.

34Tissues, cells, and Exos were lysed using RIPA buffer (KeyGEN BioTECH), followed by protein quantitation using a Bradford Protein Assay kit (KeyGEN BioTECH). Briefly, lysates were subjected to SDS–PAGE and transferred onto PVDF membranes (Millipore). Membranes were incubated overnight at 4° C with primary antibodies specific for the following proteins, as required by each experiment: TSG101 (1:1000 dilution; ab125011, Abcam), CD9 (1:1000 dilution; ab92726, Abcam), ALIX (1:1000 dilution; ab117600, Abcam), IRAK1 (1:1000 dilution; #511166, ZEN BIO), TRAF6 (1:1000 dilution; #380803, ZEN BIO), NFκB (p65) (1:1000 dilution; ab76302, Abcam), GAPDH (1:5000 dilution; 60004-1-Ig, Proteintech), IL-6 (1:1000 dilution; ab6672, Abcam), TNF-α (1:1000 dilution; Ab8348, Abcam), IL-1β (1:1000 dilution; 12242S, CST), and Drosha (1:1000 dilution; ab183732, Abcam). Thereafter, membranes were incubated with the relevant horseradish peroxidase (HRP)-conjugated secondary antibody to visualize protein spots using an enhanced chemiluminescence (ECL) kit (Thermo Scientific). Housekeeping protein GAPDH was selected as an internal control. Each experiment was conducted in triplicate.

35The miRNA content of hUMSC-Exos (n = 3) total RNA was profiled via small RNA deep sequencing analysis (Illumina). Library preparation and miRNA sequencing were performed by Ribobio (Guangzhou, China). Briefly, total RNA samples were fractionated and only small RNAs (18-30 nucleotides in length) were used for library preparation. After amplification by PCR, products were sequenced using the Illumina HiSeq 2500 platform.

36Lentiviruses expressing miR-146a-5p inhibitor were used to transduce hUMSCs at a multiplicity of infection (MOI) of 200 particles/cell. The procedure was performed in 24-well plates in DMEM (HyClone), in a 5% CO2 incubator at 37° C for three days. Successful transduction was confirmed by assessing hUMSC and hUMSC-Exos miR-146a-5p content via RT-PCR. No-load shRNA lentivirus was used as a control.

37All data are expressed as mean ± standard error (SE). Differences between two groups were analyzed by Student’s t-test (two-tailed), while differences between multiple groups were analyzed by one-way ANOVA in conjunction with the Bonferroni/Dunn post hoc test. A p-value < 0.05 was considered statistically significant. All statistical analyses were carried out using GraphPad Prism 8 software.

38All experimental procedures were approved by the Ethics Committee of Southern Medical University and performed in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals.