Extracellular vesicles derived from CD73 modified human umbilical cord mesenchymal stem cells ameliorate inflammation after spinal cord injury

Umbilical cord tissue was harvested from informed consenting healthy mothers and approved by the hospital clinic ethics committee (institutional review board (IRB) No. 81701199). Under sterile conditions, the cord blood was washed and removed. Cords were chopped into 1 mm pieces and transferred into DMEM containing 10% FBS, 5%HS, 1% P/S(v/v), and collagenase II (1 g/L; Nanjing KeyGen Biotechnology Co., Ltd., Nanjing, China) at 37℃ in an incubator containing 5% CO₂. Centrifuged and washed with PBS, non-adherent cells were removed. Cells were then plated into a culture flask at a density of 1 × 10⁶/mL and the culture medium was replaced every 3 days. After 10 days, cells reached 90% confluency and appeared colonies of fibroblast-like cells, and then they were digested in trypsin and passed into a new plate. HucMSCs were cultured in stem cell culture medium (Cyagen, Guangzhou, China) for further expansion.

BV2 cells, and astrocytes were cultured under complete medium (DMEM, Gibco, Carlsbad, CA, USA) with 10% Fetal Bovine Serum (FBS, Gibco, Carlsbad, CA, USA) and 1% penicillin streptomycin antibiotic (HyClone). BV2 cells cultured in the above medium were defined as M0 macrophages. M0 macrophages were undergone with LPS/interleukin 4 (IL-4) for 8 h to activate M1/M2 phase [21].

To produce lentiviral vectors, the pCDH lentiviral plasmid was used to make the virus in a class II biosafety laboratory. The pCDH-GFP vector (Asia-Vector Biotechnology Co., Ltd, Shanghai, China) was constructed based on CD73-overexpressed plasmid, which was prepared from a synthetic gene including the CD73 sequence. The human embryonic kidney 293 T (HEK-293T) cell line was transfected with the lentivirus and packaging plasmids using Lipofectamine 3000 (Thermo Fisher Scientific, Waltham, MA) [22].

To determine the minimum lethal concentration, hucMSCs were treated with puromycin. At concentrations > 20 µl/ml for 6 days, almost 100% hucMSCs were undergone apoptosis; therefore 20 µl/ml was selected as the optimal puromycin screening concentration.

For transduction, hucMSCs were prepared in 12-well plates with 3–5 × 10⁵ per well, and then were cultured to 70–80% confluency. In the next day, the prepared hucMSCs were transduced with lentiviral viruses in the presence of 10 μg/ml polyberene (Sigma, Germany). The medium was changed every 12 h. In addition, 20 µl/ml puromycin was added at 24 h and changed every 2 days. Finally, fluorescent microscope and light microscope were used to check the activity of harvested stable cell lines of CD73+hucMSCs.

The isolation of EVs was performed using ultracentrifugation and size exclusion chromatography as described previously [23]. The hucMSCs and CD73+hucMSCs added with MesenCult Stimulatory Supplements (STEMCELL Technologies, Vancouver, Canada) were ultracentrifuged at 100,000 g for 18 h using ultracentrifugation (MICROCL, Thermo Fisher Scientific, Inc., USA) to remove cell debris. The supernatant was filtered through a 0.22 μm pore filter (Millipore), and then was centrifuged at 3000 g for 30 min, 10,000 g for 30 min to remove cells and debris. Therefore, the supernatant was concentrated volume by ultrafiltration (100 kD, Millipore, USA) at 2000 g for 8min [18]. And then, EVs obtained from ultracentrifugation were loaded onto qEV Size Exclusion column (Izon Science, Christchurch, New Zealand) according to the instruction manual [24], which allowed to separate EVs based on the size into 16 fractions. The 16 fractions were concentrated by a vacuum centrifuge. Subsequently, the collected particles were washed with phosphatebuffered saline (PBS) to resuspend EVs and ultracentrifuged at 100,000 g for 70 min twice. All centrifugation was operated at 4 °C, and finally the solution was stored at − 80 °C or used immediately for experiments.

For extracellular vesicle identification, the morphology of purified EVs was identified by transmission electron microscopy (TEM, JEM-2100F, Japan Electronics Co., Ltd.). The size distribution and concentration were examined by nanoparticle tracking analysis (NTA, ZetaView, Particle Metrix Inc., German).

EVs were repelleted by ultracentrifugation and resuspended in MES buffer (Sigma) to remove inorganic phosphate in PBS. And then, treated with 20 μM ATP or AMP (Sigma) in 500 μl of MES for 1 h, an indicated dose of 3, 10 or 30 μg/ml hucMSCs-EXs or CD73+hucMSC-EVs were added with or without 100 μM APCP (Santa Cruz Biotechnology, Santa Cruz, CA). The ATP Assay Kit (Beyotime, Shanghai, China) was used to test the concentrations of ATP, and SensoLyte MG Phosphate Assay Kit (Anaspec, USA) was applied to quantify phosphate produced by ATP or AMP. The Adenosine Assay Kit (BioVision, CA, USA) was employed for measuring concentrations of adenosine [25].

The cultured cells and each spinal cord slides were prepared for fluorescence microscopy by permeabilization for 5 min with 0.1% Triton- × 100, blocked with 5% BSA then incubated overnight with anti-iNOS (1:100; ab49999, Abcam Cambridge, MA, USA), anti-Arg1 (1:100; ab133543, Abcam Cambridge, MA, USA), anti-C3 (1:100; 21337–1-AP, ProteinTech, Manchester, UK), anti-S100A10 (1:100; 11250–1-AP, ProteinTech, Manchester, UK), or anti-GFAP (1:100; sc-33673, Santa Cruz Biotechnology, Santa Cruz, CA, USA). Cells were incubated with secondary antibody, Alexa Fluor 488 goat anti-mouse IgG (H + L) antibody/Alexa Fluor 594 horse anti-rabbit secondary antibody (all 1:1000; Jackson ImmunoResearch, West Grove, PA) for 1 h at room temperature. A DAPI solution was applied for 5 min for nuclear staining. Images were captured using a Leica TCA SP8 confocal laser scanning microscopy (Leica, Germany) [26].

Cells and EVs were lysed with ice-cold radioimmunoprecipitation assay buffer (Solarbio, Beijing, China) supplemented with a protease and phosphatase-inhibitor cocktail (Abcam, Cambridge, MA, USA), and the protein concentration was estimated using a BCA protein assay kit (Beyotime, P0010). And then, the protein samples were dissolved in SDS–polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. Blocked in locking buffer (TBST), it was incubated with primary antibodies (1:1000) at 4 °C overnight [27, 28]. Antibodies of CD73, CD9, CD63, CD81, TSG101, ALIX, calnexin, PKA, iNOS, and Arg1, were purchased from Abcam (Abcam, CA, USA). GAPDH antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). After washing with TBST for three times, the membranes were incubated with horseradish peroxidase-conjugated secondary antibody (1:5000) and bands were developed using enhanced chemiluminescence. Image analysis was performed using Image J (NIH Image J system, Bethesda, MD).

Total RNA was isolated using the TRIzol Reagent (Thermo Fisher Scientific, USA) following the manufacturer’s instructions. The RNA concentration was quantified by the NanoDrop ND-2000 (Thermo Fisher Scientific, USA). The SYBR Green reagent was used in qRT-PCR mRNA quantification. Corresponding primers are listed in Additional file: Table S1. 1

Intracellular cAMP was with the specific enzymeimmunoassay Biotrak (EIA) System (Amersham), following the manufacturer’s instructions. Briefly, 10⁵ BV2 cells were plated and incubated with 1 μM/mL LPS. Optical density was read at 450 nm and results were calculated [29].

BV2 cells (15,000 cells/ml) were plated for 24 h, and then were incubated separately with 10⁸ MISSION® shRNA Lentiviral Transduction Particles for human A2b ADO receptors, as well as scrambled non-targeted shRNA (SHC002V) as control. The transfection was conducted according to manufacturer's recommendations, and then cells were selected by addition of puromycin (Sigma) at 2 µg/ml. The expression of A2bR was tested by quantitative PCR to verify its effect of knockdown [29].

Six-week-old male ICR mice (30–35 g) were purchased from SLAC Company (Shanghai, China). All mice were kept in the specific pathogen-free (SPF) laboratory, and the experimental procedures were approved by the Institutional Lab Animal Care and Use Committee.

Mice were anesthetized with pentobarbital sodium (50 mg/kg i.p.). Bilateral laminectomy of T8–T9 was performed to expose the spinal cord. A New York University Impactor was used to cause contusive SCI with a weight drop injury using a 10 g rod dropped at a height of 6.25 mm. After injury, the muscle and the skin were closed, and mice were placed in a temperature and humidity-controlled chamber. Manual bladder emptying was performed three times daily [30].

In the sham group (n = 20), mice received bilateral laminectomy without damage to the spinal cord and treated with saline injection in the wound. Besides, SCI mice were randomly divided into four groups, and 20 mice in each group were respectively treated with phosphate buffer saline (SCI group), or 20 μg hucMSC-EVs (SCI + hucMSC-EVs group), or 20 μg CD73 +hucMSC-EVs (SCI + CD73 +hucMSC-EVs group), or 20 ng rmCD73 (recombinant mouse CD73, Beijing Baiao Lai Bo Technology Co., Ltd., Beijing, China). For topical administration, 20 μg EVs were diluted to 50 μg/ml by phosphate buffer saline in a 1 ml syringe, which was injected into the wound at the spinal cord injury site of T9 exposed spinal cord once a day and for 10 days post injury [31].

To test the biodistribution of EVs, images were taken with an in vivo imaging system (IVIS) (IVIS Spectrum, PerkinElmer, USA) at 24 h post-injection and the fluorescence intensities were analyzed by Living Image Software. Briefly, 6 nude mice were purchased from SLAC Company (Shanghai, China). CD73+hucMSC-EVs were labeled with DiR for further administration. Three mice were treated with intraperitoneal injection, and 3 mice were injected in situ around the injured spinal cord. After 24 h, nude mice were anesthetized, and were analyzed by bioluminescence imaging in vivo. Anteroposterior and lateral graphs were taken to determine locations of labeled EVs.

Basso Mouse Scale (BMS) open-field score and Basso-Beattie-Bresnahan (BBB) score were used to evaluate the functional recovery of hind limb locomotion on 1, 3, 7, 14, 21, 28, 35 and 42 days post injury by two well-trained observers who were blinded to the experimental animal grouping. For these tests, mice were allowed to move freely in an open field for 4 min, and average locomotor scores were calculated and recorded [32, 33].

Mice were anaesthetized by isoflurane and oxygen, and performed by direct cisterna magna puncture with a glass capillary tube as described previously [34, 35]. And then, CSF was collected for ATP measurement using the ATP Assay Kit (Beyotime, Shanghai, China).

On day 42, animals were euthanized and were perfused with 4% paraformaldehyde through the left ventricle as previously described [36, 37].

The whole spinal cords were rapidly removed and embedded in paraffin. Longitudinal sections were produced at 5 μm from a 1 cm length of spinal cord centered on the injury epicenter. Slides of each spinal cord were stained with Haematoxylin and Eosin (H&E), Nissl staining, and TdT-mediated DUTP nick end labeling (TUNEL) staining.

To isolate microglia, spinal cord tissues were removed and minced with scissors in ice‐cold Dulbecco's Modified Eagle Medium (DMEM) (Invitrogen). Thereafter, a monocular suspension was formed by digesting with 0.25% trypsin (Invitrogen) in a water bath at 37 °C for 30 min, and then were co-stained for CD206 (an M2 microglia biomarker) and CD86 (an M1 microglia biomarker) for 45 min at room temperature following the manufacturer’s instructions [38]. The samples were detected using FACSAria III flow cytometer (BD Biosciences, San Jose, CA, USA) and then analyzed by FlowJo software v.7.6.1 (https://flowjo.com/↗).

Furthermore, the Bio-Plex system (Bio- Rad, CA) with a 23-plex cytokine array kit (Bio-Rad) was used for cytokine analysis of spinal cord samples.

Similarly, the levels of cytokines were also measured by ELISA in spinal cord samples, including IL-1β (cat. no. P1303), IL-6 (cat. no.P1328), TNF-α (cat. no. PT516), and IL-10 (cat. no. PI525), using protocols supplied by the manufacturer (Beyotime Biotech., Jiangsu, China).

Under light microscope, hucMSCs derived from umbilical cord tissue and the modified CD73+hucMSCs by lentiviral transduction appeared similar growth characteristics (Fig. 1A). The hucMSC-EVs and CD73+hucMSC-EVs displayed classic type of spherical morphology under the transmission electron microscopy (Fig. 1B). NTA showed that the diameters of hucMSC-EVs and CD73+hucMCS-EVs were 103.41 ± 42.03 and 106.51 ± 53.99 nm, respectively (Fig. 1C). The protein concentration in the EVs was 2.5 μg/μl detected by BCA kit. In addition, western blot revealed that the surface markers (CD9, CD63, CD81, TSG101, and ALIX) were all positive in hucMSC-EVs and CD73 +hucMSC-EVs, but calnexin was negative in hucMSC-EVs and CD73+hucMSC-EVs. They are acknowledged biomarkers of testing EVs. Moreover, the expression of CD73 was significantly higher in CD73+hucMSCs and CD73+hucMCS-EVs (Fig. 1D).

In the detection of the ATP and AMP hydrolytic activities, it showed that hucMCS-EVs and CD73+hucMCS-EVs could dose-dependently hydrolyze ATP and AMP, which was significantly mitigated by the CD73-specific inhibitor APCP (Fig. 1E and F). Moreover, hydrolytic effects of CD73+ hucMCS-EVs performed significantly stronger than that of hucMCS-EVs.

To reduce the release of neuroinflammatory factors, the M1/M2 balance of microglia has far-reaching significance. In this study, BV2 cells from mice were isolated and cultured for 7 days, and then were induced with 1 μg/ml LPS or 5 μg/ml IL-4 for the polarization to M1 or M2 cells, respectively. And then, iNOS and Arginase 1 were selected as the surface marker of M1/M2 cells (Fig. 2). Finally, it showed that hucMSC-EVs, CD73+hucMSC-EVs, and CD73 could reduce the expression of iNOS induced by LPS. CD73+hucMSC-EVs decreased it the most, while its effect could inhibited by APCP. On the other hand, it also showed that CD73+hucMSC-EVs significantly increased the expression of Arginase 1, which was reversed by APCP.

To verify the downstream target of overexpression of CD73, we used SCH58261↗ and MRS1706 as inhibitors of A_2a and A_2b receptors, respectively. A₂Rs are known to mediate inflammation in microglia, and A₂Rs are Gs-linked protein and increase cAMP. In Fig. 3A, 30 μg/mL CD73+hucMCS-EVs significantly increased the level of cAMP. And it was significantly reduced by 1 μM MRS1706 (A_2bR inhibitor) but not 1 μM SCH58261↗ (A_2aR inhibitor). For downstream targets of cAMP, PKA also increased after treating with CD73+hucMCS-EVs, and it was significantly prevented by MRS1706 but not SCH58261↗ (Fig. 3B). In addition, Arginase 1 and iNOS were often used as markers of microglia subtypes, since arginase 1 could effectively downregulate nitric oxide production caused by iNOS and activate anti-inflammatory function. When treated with LPS, CD73+hucMCS-EVs significantly decreased the level of iNOS; and when treated with IL-4, CD73+hucMCS-EVs significantly increased the level of Arginase 1. Interestingly, MRS1706 performed a dose dependent way on inhibiting the effect of CD73+hucMCS-EVs (Fig. 4C, D). In contrast, SCH58261↗ showed little effect on CD73+hucMCS-EVs (Fig. 4E, F). Furthermore, the expression of M1 mRNA markers (TNF-α, IL-1β, iNOS, and CD86), and the expression of M2 mRNA markers (Arginase, IL-10, and CD206) were detected. It also showed similar outcomes to the western blot results (Fig. 4G, H). As a result, CD73 + hucMCS-EVs might enhance M2 polarization and moderate M1 polarization via A_2b adenosine receptor activation, but not A_2a adenosine receptor.

A₂bR was knockdown by permanently transfected A_2bR shRNA. Treated with LPS and CD73+hucMCS-EVs, the concentration of cAMP in shRNA A_2bR group was significantly decreased when compared with shRNA scrambled group (Fig. 4A). The standardized expression level of PKA was also significantly weakened in A_2bR shRNA group when treated with CD73+hucMCS-EVs (Fig. 4B). In addition for M1/M2 subtypes, CD73+hucMCS-EVs downregulated the level of iNOS induced by LPS in both shRNA scrambled group and shRNA A_2bR group, but the level of iNOS in shRNA A_2bR group was significant higher than that in shRNA scrambled group. And the level of Arginase 1 was significantly lower in shRNA A_2bR group than that in shRNA scrambled group when treated with CD73+hucMCS-EVs and IL-4 (Fig. 4C and D). Similarly, the mRNAs of M1/M2 markers was observed in A_2bR knockdown cells, and the results were concurred with the western blot findings (Fig. 4E and F). As a result, it confirmed that A_2bR was important in the effects of CD73+hucMCS-EVs.

Meanwhile, cAMP signals proceed via intracellular activation of PKA, and the PKA inhibitor H-89 is used to determine this downstream target. It showed that the level of iNOS was increased after LPS induction. And CD73+hucMCS-EVs decreased the iNOS expression, which was inhibited by H-89. It also showed that H-89 decreased the upregulating effect of CD73+hucMCS-EVs on the level of Arginase 1 (Fig. 4G). Thus, A_2bR was identified as the receptor of adenosine produced by CD73+hucMCS-EVs, and downstream biomarkers might be activated via A_2bR/cAMP/PKA signaling pathway.

Prior to exploring the effects of EVs, in vivo biodistribution of EVs was investigated by IVIS imaging analysis. The fluorescence signal of DiR-labeled EVs remained well around the wound 24 h after subcutaneous injection (Fig. 5A). However, 24 h after intraperitoneal injection, it presented intense signals in the abdominal cavity and mainly in liver and kidneys, but it showed almost none fluorescence signals around the spinal cord. (Fig. 5B). Therefore, EVs were directly injected into the wound in the following study.

We established a mouse spinal cord injury model according to Allen's method. Saline, hucMSC-EVs, CD73+hucMSC-EVs, or recombinant mouse CD73 was respectively treated for 10 days after the injury. Firstly, we recorded the body weight of each mouse. There was a significant decrease in the body weight of all groups on day 3 after the injury, and mice in SCI + CD73+hucMCS-EVs group gradually recovered with a significant increase in body weight, when compared with other groups suffered from SCI (Additional file 2: Fig. S1A). And then, to compare the different levels of motor function of the five groups of mice, we recorded BMS scores and BBB scores at different times at 1, 3, 7, 14, 21, 28, 35 and 42 days after operation. The exercise scores were calculated according to the scoring criteria, and Excel was used to record and plot the plot lines of different groups with BMS scores and BBB scores. It showed that the SCI + CD73+hucMCS-EVs group had a significant improvement on post-SCI motor dysfunction than SCI group. The SCI + CD73+hucMCS-EVs group also showed a better result than the SCI + hucMCS-EVs group and the SCI + CD73 group (Fig. 5C, D). In addition, cerebrospinal fluid of mice was extracted to detect ATP concentration under pathological conditions on the corresponding 1, 3, 7, 14, 21, 28, 35 and 42 days. The concentration of sham group was basically unchanged. ATP level in SCI + CD73+hucMCS-EVs group was significantly increased than in SCI group. And ATP levels in SCI + +hucMCS-EVs group were moderately increased. Interestingly, the ATP level upgraded within 1–7 days, and slowed down after day 14 due to the withdrawn of treatment on the day 10 (Fig. 5E).

Furthermore, we selected the spinal cord tissue of the mouse model segment for HE staining, Nissl staining, and TUNEL staining on day 42 post injury. The SCI group presented that the Nissl body neuron revealed to be disordered, swelling, and vacuolar structure. HucMCS-EVs reduced the injury, and CD73+hucMCS-EVs performed better (Fig. 5F, G). In addition, TUNEL-positive cells were obviously reduced in SCI + CD73+hucMCS-EVs group, when compared to other groups (Fig. 5H).

To determine the ratio of M2 and M1 microglia in spinal cord samples of mice, immunofluorescence was used for mice executed on day 7 after SCI. Antibodies of Arg-1 and iNOS were applied for immunofluorescence in accordance with antibodies used in vitro. CD73+hucMCS-EVs significantly reduced iNOS positive cells and improved Arg-1 positive cells in vivo (Fig. 6A, B, and Additional fiile 2: Fig. S1B). In addition, flow cytometry was also employed to test CD206 and CD86, biomarkers of M2 and M1 microglia, respectively. And it showed that the calculated M2:M1 was significantly increased in SCI + CD73+hucMCS-EVs group than other groups (Fig. 6D, E).

Additionally, 7 days after treatments, cytokines in mice spine samples were quantitatively analyzed with bio-plex system. After the cytokine content of each group was standardized, a heatmap was established for each group (Fig. 7A). Pro-inflammatory cytokines, such as IL-1β, IL-6, TNF-α, MCP-1, IFN-α, and MIP-1β, were significantly reduced in SCI+CD73+hucMCS-EVs group. Anti-inflammatory cytokines, such as IL-4 and IL-10, were significantly increased in SCI+CD73+hucMCS-EVs group (Fig. 7B–H). Similarly, the releases of IL-6, IL-1β, TNF-α, and IL-10 in spinal cord tissues of SCI mice were also detected by ELISA. Compared with the SCI group, the levels of TNF-α, IL-6, and IL-1β were increased in SCI+CD73+hucMCS-EVs group (p < 0.05), and the important anti-inflammatory cytokine, IL-10 (p < 0.05) was markedly decreased (Fig. 7I–L). Collectively, CD73+hucMCS-EVs upregulated the polarization of M2:M1 and alleviated inflammatory response in the repair of spinal cord injury.

After SCI, microglia might release molecules that trigger astrocyte proliferation and astrocyte scar formation [39]. To determine the activation astrocytes, we tested GFAP, which represents the astroglial scar and is expressed throughout the entire injury site. It showed a significant expression of GFAP-positive cells in SCI group, and GFAP-positive cells were significantly decreased in CD73+hucMCS-EVs group when compared with other mice after SCI. Furthermore, to examine the subtypes of astrocytes, which are defined as A1 phenotype (inflammatory astrocyte) and A2 phenotype (tissue repair astrocyte), C3 and S100A10 were used as biomarkers, respectively [40]. As a result, C3-positive cells were significantly increased in SCI group but significantly inhibited in CD73+hucMCS-EVs group (Additional file 3: Fig. 2A). However, there was no significant difference in S100A10-positive cells among groups (Additional file 3: Fig. 2B).

Umbilical cord tissue was harvested from informed consenting healthy mothers and approved by the hospital clinic ethics committee (institutional review board (IRB) No. 81701199). All mice were kept in the specific pathogen-free (SPF) laboratory, and the experimental procedures were approved by the Institutional Lab Animal Care and Use Committee.

Written informed consent for publication was obtained from all participants.

The authors do not have any possible conflicts of interests.

All data are fully available without restriction.