Exosomes from embryonic mesenchymal stem cells alleviate osteoarthritis through balancing synthesis and degradation of cartilage extracellular matrix

Background Mesenchymal stem cell therapy for osteoarthritis (OA) has been widely investigated, but the mechanisms are still unclear. Exosomes that serve as carriers of genetic information have been implicated in many diseases and are known to participate in many physiological processes. Here, we investigate the therapeutic potential of exosomes from human embryonic stem cell-induced mesenchymal stem cells (ESC-MSCs) in alleviating osteoarthritis (OA). Methods Exosomes were harvested from conditioned culture media of ESC-MSCs by a sequential centrifugation process. Primary mouse chondrocytes treated with interleukin 1 beta (IL-1β) were used as an in vitro model to evaluate the effects of the conditioned medium with or without exosomes and titrated doses of isolated exosomes for 48 hours, prior to immunocytochemistry or western blot analysis. Destabilization of the medial meniscus (DMM) surgery was performed on the knee joints of C57BL/6 J mice as an OA model. This was followed by intra-articular injection of either ESC-MSCs or their exosomes. Cartilage destruction and matrix degradation were evaluated with histological staining and OARSI scores at the post-surgery 8 weeks. Results We found that intra-articular injection of ESC-MSCs alleviated cartilage destruction and matrix degradation in the DMM model. Further in vitro studies illustrated that this effect was exerted through ESC-MSC-derived exosomes. These exosomes maintained the chondrocyte phenotype by increasing collagen type II synthesis and decreasing ADAMTS5 expression in the presence of IL-1β. Immunocytochemistry revealed colocalization of the exosomes and collagen type II-positive chondrocytes. Subsequent intra-articular injection of exosomes derived from ESC-MSCs successfully impeded cartilage destruction in the DMM model. Conclusions The exosomes from ESC-MSCs exert a beneficial therapeutic effect on OA by balancing the synthesis and degradation of chondrocyte extracellular matrix (ECM), which in turn provides a new target for OA drug and drug-delivery system development. Electronic supplementary material The online version of this article (doi:10.1186/s13287-017-0632-0) contains supplementary material, which is available to authorized users.


Background
Osteoarthritis (OA) is a common joint disease worldwide, which exerts a significant detrimental effect on the patients' life quality [1]. Pharmaceutical agents, particularly nonsteroidal anti-inflammatory drugs (NSAIDs) and paracetamol, have been widely applied in the clinic. However, their effects are largely limited to pain control with negligible effects on cartilage maintenance, which would ultimately lead to cartilage destruction [2,3]. Although surgical interventions such as total knee or hip arthroplasty could partially overcome pain and deformity in end-stage OA patients, this could also cause further complications such as increased risks of infection, thrombus formation and secondary surgery, particularly in elderly patients [4][5][6]. Hence, there is a dire need to develop new strategies to protect cartilage and attenuate OA development, particularly minimally invasive interventions at the early stages of OA.
Cell transplantation is an emerging therapeutic modality for OA treatment, and mainly involves the utilization of autologous chondrocytes and mesenchymal stem cells [7,8]. Although autologous chondrocytes as the main cell type in cartilage may provide a safe and efficacious solution, these have the inherent drawbacks of limited availability, de-differentiation and function loss during the process of in vitro expansion [8]. Therefore, the application of mesenchymal stem cells (MSCs) for OA therapy has attracted much more attention from scientific investigators [7,[9][10][11]. Clinical applications of adult MSCs are limited by donor source and individual physical conditions [12][13][14]. MSCs derived from pluripotent embryonic stem cells (ESC-MSCs) have become another potential and promising source of MSCs. The unlimited self-renewal capacity and pluripotency of ESCs [15] in turn ensures less inter-batch variability and a more consistent supply of ESC-MSCs. Furthermore, as a cell source for the treatment of various diseases, ESC-MSCs have also been reported to possess immunomodulatory properties, as well as regenerative potential, similar to adult MSCs [16]. However, there have not yet been any studies that have investigated their therapeutic effects on OA and probed the underlying mechanisms involved.
As one of the major pathways of extracellular signaling, exosomes have been implicated in many diseases and participate in many physiological processes. These are derived from fusion in cell to cell communication, as well as serve as carriers of genetic information [23]. Recently, exosomes derived from MSCs have received much attention due to a study showing that severe graft versus host disease (GVHD) can be treated with MSCs-derived extracellular vesicles [24], thus suggesting that the observed therapeutic effects of MSCs can in fact be reproduced by MSCs-derived exosome treatment [19,25].
In this study, we report the positive therapeutic effects of ESC-MSCs on osteoarthritis, and investigate the underlying mechanisms involved.

Mice and OA model
Thirty-two 2-month-old C57BL/6 J mice purchased from the animal center of Zhejiang University were utilized for all experiments. The mice were anesthetized with sodium pentobarbital (8 mg/ml, 1 ml/100 g) and randomly distributed into experimental and control groups. Among them, 12 mice were in the sham group, and 20 mice were utilized for the experimental OA model, which was induced by destabilization of the medial meniscus (DMM) surgery, following the transection of the medial meniscotibial ligament as previously described [26]. In the first animal section, DMM surgery was performed in the bilateral knee joints and five mice were allocated to each group. Three mice were allocated to each sham group. OA pathology developed gradually throughout 4 weeks before single intra-articular injection of 5 μL ESC-MSCs (1 × 10 6 /joint) or 5 μL PBS per joint with a microliter syringe (Hamilton Company, Reno, NV, USA, 1702) and 5 mm 30-gauge needles (Hamilton Company, 7803-05). In the second animal section, DMM surgery was performed in the bilateral knee joints and five mice were allocated to each group. Three mice were allocated to each sham group. OA pathology developed gradually throughout 4 weeks before multiple injections of 5 μL exosomes isolated from ESC-MSCs per joint or 5 μL PBS per joint, which was performed every 3 days for 4 weeks with a microliter syringe (Hamilton Company, 1702) and 5 mm 30-gauge needles (Hamilton Company, 7803-05).

Cell culture
MSCs were derived from the male H1 human ES cell line (ESC) that was obtained from WiCell Corporation. Madison, WI, USA (http://www.wicell.org), which was maintained in the undifferentiated state by culture in a feeder-free system. ESC-MSCs were obtained as described in our previous study [27]. Briefly, ESCs were detached from Matrigel (BD, Franklin Lakes, NJ, USA, Cat. #354234) coated plates by a 10-minute (min) incubation with 0.02% (w/v) EDTA (Sigma-Aldrich, St. Louis, MO, USA, Cat. #E8008) and seeded onto a gelatinized 10-cm plates. The cells were cultured in L-DMEM (Life Technologies, Carlsbad, CA, USA, Cat. #11885084) supplemented with 10% (v/v) fetal bovine serum (FBS; Life Technologies, Cat. #10099-141). The cells were then trypsinized (0.05% (w/v), Life Technologies, Cat. #15400-054) upon reaching 80% confluency and then plated at a density of 10 3 cells/cm 2 . The medium was pre-processed to delete the suspension in FBS with centrifugation at 110,000 × g overnight at 4°C, according to the published protocol of Thery et al. [28]. Cells between passage number 4 and 7 were utilized.

Chondrocyte isolation and culture
Primary chondrocytes were obtained from nine C57BL/6 mice within 2 days after birth. The articular cartilage derived from the terminal of tibia and femur was digested with 0.2% (w/v) type II collagenase [29,30], then expanded in medium containing F12 (Life Technologies, Cat. #88215) with 10% (w/v) FBS (Life Technologies, Cat. #10099-141), and 1% (w/v) penicillin/streptomycin (Life Technologies, Cat. #15140-122). Chondrocytes at the first passage were seeded into 24-well plates containing glass slides for immunofluorescence analysis and sixwell plates for western blot analysis. At 24 hours after cell seeding, we changed the medium into that with or without 2 ng/mL interleukin 1 beta (IL-1β) (PeproTech, Rocky Hill, NJ, USA) according to the experimental design. Forty-eight hours later, the cells were collected for analysis. Cells within 3 passages were utilized.

Colony-forming unit assay
ESC-MSCs were seeded at approximately 20 cells/cm 2 in triplicates and cultured for up to 21 days with regular medium replenishment every 3-4 days. The colonies were stained with 1% (w/v) crystal violet (Sigma-Aldrich, Cat. #3886) in methanol for 10 min.

Isolation of exosomes from ESC-MSCs medium
The exosomes were isolated from 600 mL of ESC-MSCs culture medium, which were collected when cells have attained around 60% confluency, and were finally resuspended in 600 μL PBS. The isolation and purification followed the multistep ultracentrifugation process as previously described [28]. Briefly, the conditioned medium containing exosomes (CM + Exo) was obtained by centrifugation at 2000 × g for 10 min at 4°C to remove dead cells and 10,000 × g for 30 min at 4°C to remove cell debris. This was followed by centrifugation at 110,000 × g for 90 min at 4°C to separate the exosomes from the CM + Exo. At the same time, the CM without exosomes (CM-Exo) was also collected. The exosomes were washed with PBS and subjected to a final centrifugation, prior to being stored in PBS at -80°C.
For tracking, 100 μL of exosomes were transferred to a 1.5 mL tube, and were incubated with 10 μM DiI (Beyotime, Shanghai, China, Cat. #C1036) at 37°C for 15 min, washed with PBS at 110,000 × g for 90 min at 4°C, and then stored in 100 μL PBS at -80°C.

Tissue histology and immunohistochemistry
Mouse joints were isolated and fixed in 4% buffered paraformaldehyde for 24 h, then decalcified in 10% (w/v) EDTA (pH 7.4) for 21 days (d) at 4°C before being embedded in paraffin. Sagittal joint sections at 6 μm thickness were processed for Safranin O and Fast Green staining. Cartilage destruction was evaluated using the Osteoarthritis Research Society International (OARSI) scoring system according to the percentage of the vertical clefts/erosion to the calcified cartilage [26]. All four parts of the joint: medial femoral condyle, lateral femoral condyle, medial tibial plateau and lateral tibial plateau presented in the figures were semi-quantitatively scored according to the recommended OA grading table (Table 1) supplied by Glasson et al. The OA grading of each joint is expressed as the maximum or summed score of the four quadrants, respectively. Immunohistochemical staining was performed using a standard protocol. After dewaxing, heat-induced antigen retrieval was performed in retrieval solution overnight at 64°C. The solution was composed of 0.1 M trisodium citrate (20.5 mL) and 0.1 M citric acid anhydrous (4.5 mL) in 225 mL distilled water. Sections were incubated overnight at 4°C with primary antibodies: rabbit anti-ADAMTS5 (1:100; Abcam; Cat. #ab41037), mouse anti-Col II (1:50; COL2A1, Santa Cruz Biotechnology, Cat. #sc-52658), rabbit anti-aggrecan neoepitope antibody (1:100; Novus Biologicals, Littleton, CO, USA, Cat. #NB100-74350SS). After washing off excess primary antibodies, these samples were incubated with secondary antibodies conjugated with HRP: HRP-labeled goat anti-mouse IgG (1:200; Beyotime, Cat. #A0216) and goat anti-rabbit IgG antibody, peroxidase-conjugated (1:600; EMD Millipore, Cat. #AP132P) was diluted in 1% (w/v) BSA solution and incubated the section for 1 h at room temperature (RT). DAB detection system (Solarbio, Cat. #DA1010) were used to visualized the section. The stained specimens were photographed digitally under a slide scanning machine (Pannoramic MIDI, 3DHISTECH Ltd., Budapest, Hungary).

Transmission electron microscopy
Purified exosomes were fixed with 1% (w/v) glutaraldehyde in PBS (pH 7.4). After rinsing, a 20 μL drop of the suspension was loaded onto a formvar/carbon-coated grid, negatively stained with 3% (w/v) aqueous phosphotungstic acid for 1 min, and then imaged under transmission electron microscopy (HT7700, Hitachi, Tokyo, Japan). The diameter of the exosome was evaluated by the ImageJ software (ImageJ v2.1.4.7, National Institutes of Health), and we analyzed the diameter distribution and percentage values with the Excel software (Microsoft (Redmond, WA, USA) Excel for Mac version15.24).

Quantitative real-time polymerase chain reaction (qRT-PCR)
Total RNA was isolated from mouse primary chondrocytes by lysis in TRIzol (Takara, Shigo, Japan, Cat. #9109). The reverse transcription process applied Rever-Tra Ace qPCR RT Master Mix (Toyobo, Osaka, Japan, Cat. #FSQ-201). The qRT-PCR was performed utilizing Brilliant SYBR Green QPCR Master Mix (Takara, Cat. # RR420A) with a LightCycler apparatus (480II, Roche, Mannheim, Germany). The amplification efficiencies of primer pairs were validated to enable quantitative comparison of gene expression. All primer sequences (Invitrogen) were designed using Primer 5.0 software (see Additional file 1). Each qRT-PCR was performed three times on at least three different experimental replicates, and results were normalized to those obtained with the endogenous reference gene (Gapdh).

Statistical analysis
All data were expressed as mean ± SE unless otherwise stated. All experiments in vitro were repeated independently at least twice in addition to the triplicates applied in each experiment. Statistical results were analyzed and bar charts were constructed with GraphPad Prism version 5.0 (GraphPad Software, San Diego, CA, USA). Statistical results were considered significant when the p value was less than 0.05. Two-tailed Student's t test was used to compare two groups at the same time point.
One-way ANOVA including the Tukey-Kramer post hoc test was used to compare multiple groups at the same time point. Experimental data of the in vivo experiment was analyzed by the Mann-Whitney test with the SPSS software (IBM Corp., Armonk, NY, USA).

Establishment of ESC-MSCs
The human ESC is a cell line obtained from WiCell Corporation. The ESCs cultured on Matrigel yielded compact colonies with sharp borders (Fig. 1a). These cells, however, adopted a fibroblast-like morphology upon plastic adhesion. Compared with ESCs, these cells displayed spindle-shaped morphology in monolayer culture and were distributed sparsely (Fig. 1a). When these cells were seeded at densities as low as 200 cells per 9.5 cm 2 well, they developed into round and tight colonies (Fig. 1b). Flow cytometry analysis showed that more than 95% of these cells expressed the classical MSC markers including CD73, CD90, and CD105, while no cells expressed hematopoietic markers such as CD34 and CD45 (Fig. 1c). Upon osteogenic inducement, these cells synthesized matrix which were stained dark brown by Alizarin red, indicating calcium deposition. After adipogenic induction, these cells displayed round orange droplets within the cytoplasm upon oil red staining, which indicates formation of oil droplets. The cells were kept in aggregates before chondrogenic differentiation, resulting in formation of a pellet and synthesized matrix which were stained orange and red by Safranin O staining, thus indicating cartilage matrix formation (Fig. 1d).

Alleviation of osteoarthritis by ESC-MSCs
To evaluate the therapeutic effects of ESC-MSCs on osteoarthritis, we injected 5 μL of cell suspension (1 × 10 6 ) directly in the bilateral joints of five OA mice at 4 weeks after DMM surgery as the ESC-MSC group and injected 5 μL of PBS in the bilateral joints of five OA mice at 4 weeks after DMM surgery as the control group. We harvested the joint samples after the following 4 weeks (Fig. 2a). Histological evaluation revealed that the PBS-injected DMM mice have more sever fibrillation and erosion of the calcified cartilage. The maximal OARSI score was 3.8 ± 1.04, and the summed OARSI score was 10.8 ± 3.37. However, the ESC-MSC-injected DMM mice exhibited less fibrillation with vertical clefts only down to the layer immediately below the superficial layer, and some loss of surface lamina. The maximal OARSI score was 2.6 ± 0.45, and the summed OARSI score was 7.6 ± 1.97 (Fig. 2b). Hence, the OARSI evaluation in the ESC-MSCs group revealed significantly lower maximal and summed scores than the control group (p = 0.011 and 0.044 respectively) (Fig. 2c) (see Additional files 2 and 3). The immunohistochemistry results showed that the cartilage of the ESC-MSCs group exhibited much stronger Col II-specific staining, much weaker ADAMTS5-specific staining and aggrecan neoepitope-specific staining than the control group (Fig. 2d) (see Additional file 2).

Effect of ESC-MSCs culture medium deficient in exosomes on cartilage matrix maintenance
To investigate whether exosomes or cytokines exert a critical effect on the alleviation of OA by ESC-MSCs, we harvested the conditioned medium (CM + Exo), the conditioned medium without exosomes (CM-Exo) and the exosomes from the ESC-MSCs culture medium (Fig. 3a). Under TEM, the isolated exosomes were round lipid bilayers-vesicles and approximately 30-200 nm in diameter (Fig. 3b). The diameters of isolated exosomes ranged from 38 nm to 169 nm and the mean value was 78 nm (Fig. 3c). The concentration of the isolated exosomes is 176.2 μg/mL (see Additional file 4). The western blot results confirmed the identity of these isolated exosomes by showing thick CD63 and CD9 bands, while there was a weak CD9 band and no CD63 band in the supernatant of cultured medium (Fig. 3d). Initially, we observed the effects of CM-Exo in maintaining the cartilage matrix. In vitro, immunocytochemistry showed that IL-1β could obviously inhibit the synthesis of Col II and increase the expression of ADAMTS5. However, this effect could be reversed by CM + Exo but not CM-Exo (Fig. 4a). Furthermore, the quantification results showed that the percentage of Col II-positive cells significantly increased from 22.08% ± 13.6% in the IL-1β group to 57.80% ± 7.9% in the CM + Exo group (p = 0.0068), which was similar to the control group (55.48% ± 20.7%). By contrast, the percentage of Col IIpositive cells in the CM-Exo group was much lower at 27.27% ± 9.2%, which was similar to the IL-1β group (Fig. 4b). The ADAMTS5 immunofluorescence intensity was greatly upregulated after IL-1β treatment (0.044 ± 0.003) compared with the control group (0.02 ± 0.007). The CM + Exo treatment can downregulate the ADAMTS5 immunofluorescence intensity to 0.022 ± 0.004, while the CM-Exo can only downregulate the intensity to 0.028 ± 0.002 (Fig. 4b). These results were doubly confirmed by Western blot analysis (Fig. 4c). The CM + Exo group displayed an obviously thicker Col II band than the CM-Exo group, which was similar to that of the control group. Meanwhile, the CM + Exo group had an obviously lighter ADAMTS5 band than the CM-Exo group, which was similar to that of the control group.

Effects of exosomes derived from ESC-MSCs on cartilage matrix maintenance
To further confirm the effects of exosomes on cartilage matrix maintenance, the isolated exosomes were directly added into the chondrocytes with IL-1β treatment. Immunocytochemistry showed that the Col II expression in the exosomes group obviously increased, while the ADAMTS5 expression in this group obviously decreased (Fig. 5a). The quantification results confirmed that there were 50.18% ± 8.8% of Col II-positive cells in the   c Expression of Col II and ADAMTS5 by the primary chondrocytes with different treatments. COL II collagen type II, IL-1β, interleukin 1 beta exosomes group, which was significantly higher than that in the IL-1β group (22.08% ± 13.6%) (p = 0.0323) (Fig. 5b). The immunofluorescence intensity of ADAMTS5 was upregulated from 0.024 ± 0.007 to 0.044 ± 0.003 after IL-1β treatment; but upon addition of exosomes, the intensity was downregulated to 0.027 ± 0.004 (Fig. 5b). Meanwhile, gene expression showed the similar effects of exosome treatment (see Additional file 5). Meanwhile, the DiI-labelled exosomes can be detected within the Col II-expressing chondrocytes (Fig. 5c), thus indicating the colocalization between exosomes and chondrocytes. The western blot results also showed gradually denser Col II bands and gradually fainter ADAMTS5 bands of chondrocytes with increasing dosages of exosomes. (Fig. 5d).

Alleviation of osteoarthritis by exosomes from ESC-MSCs
Finally, the effects of ESC-MSC-derived exosomes in alleviating osteoarthritis was evaluated in the DMM mouse model. The injection was started 4 weeks after DMM surgery. Five microliters exosomes or 5 μL PBS were injected into bilateral knee joints of mice in exosomes group and PBS group respectively, every 3 days for 4 weeks (Fig. 6a). The SO staining results showed that the fibrillations had extended to some degree to the calcified cartilage in the PBS group. The maximal OARSI score was 3.5 ± 0.76, and the summed OARSI score was 9.4 ± 2.49. By contrast, the exosomes group exhibited milder OA pathology such as roughened articular surface fibrillations below the superficial layer and some loss of lamina. The maximal OARSI score was 2.7 ± 0.66, and the summed OARSI score was 7.3 ± 1.75 (Fig. 6b) (see Additional files 2 and 6). Both the maximal and total OARSI scores were significantly lower in the exosomes group than the control (Fig. 6c). The immunohistochemistry results showed that the cartilage of the exosomes group displayed much stronger Col II-specific staining, much weaker ADAMTS5-specific staining and aggrecan neoepitope-specific staining than the control group (Fig. 6d) (please refer to Additional file 2).

Discussion
In this study, we successfully established MSCs from the H1 human ES cell line for OA treatment. These ESC-MSCs fulfilled the minimal criteria of MSCs as defined by the International Society for Cellular Therapy [33], which encompass the plastic adherence property, fibroblastic morphology, molecular profile phenotype, proliferative capacity (see Additional file 4A), and trilineage differentiation potential of these cells. Our results showed that intra-articular injection of ESC-MSCs significantly impeded cartilage destruction in a DMM OA mice model. According to the immunohistochemistry results, the ESC-MSCs can increase Col II (the major cartilage matrix component) expression and decrease ADAMTS5 (aggrecan-degrading enzyme 5) expression in the cartilage matrix.
It is well documented that the secretome of MSCs mediates many of the beneficial therapeutic effects of MSCs on various diseases, and includes a diverse array of cytokines, exosomes, and nuclei acids. Among these, cytokines are the most well studied, but accumulating scientific evidence suggests the important roles of other components. In the past few years, the exosomes derived from MSCs have been reported to have a therapeutic role in various diseases [34,35]. Hence, we investigated whether the exosomes from ESC-MSCs can alleviate the OA process. To distinguish the activity of exosomes and other soluble components that maintain the chondrocyte phenotype in an inflammatory environment, we harvested the CM + Exo (containing both exosomes and other soluble components), CM-Exo (excluding exosomes) and exosomes by serial centrifugation (Fig. 3a). The following results showed that the CM + Exo decreased ADAMTS5 expression and maintained Collagen type II synthesis by primary chondrocytes under IL-1β treatment in vitro. These effects were almost identical to that of exosomes that play the same role in a dose-dependent pattern. However, the CM-Exo group exhibited lower capacity in maintaining the chondrocyte phenotype. This thus indicates that the exosomes may be one of the major effectors of OA treatment mediated by ESC-MSCs. Additionally, upon tracking the location of exosomes, the exosomes can be found to be colocalized with Col II-expressing chondrocytes, suggesting that these may fuse with chondrocytes and exert a direct effect on these cells.
To confirm the regenerative effects of exosomes in vivo, the exosomes were directly injected into the knee joints of the DMM-induced OA model. The results also confirmed that the exosomes derived from ESC-MSCs effectively prevented the progress of cartilage destruction in an OA model.
Previous studies have shown that the MSCs can be a candidate for osteoarthritis therapy [7,11,36]. Most research focus on adult MSCs, which may vary between donors' physical conditions. Here we attempted to establish MSCs from a promising and consistent new source (ESCs). And these ESC-MSCs displayed similar characteristics to adult MSCs with therapeutic potential in OA. Recent studies have shown that paracrine mechanisms, particularly exosomes are responsible for the therapeutic effects of MSCs mediated OA treatment [37][38][39]. Tao et al. [38] demonstrated that exosomes derived from human synovial MSCs can promote chondrocyte proliferation and migration. However, only when the exosomes were derived from miR-140-5p overexpressing MSCs, could these exert a regenerative effect in OA treatment. Zhu et al. [39] compared exosomes secreted by induced pluripotent stem cell-derived MSCs and synovial membrane-derived MSCs for OA treatment in a collagenase-induced mouse model. Both studies demonstrated that the exosomes from MSCs could somehow prevent OA progression. And it appears that the exosomes from induced pluripotent stem cell-derived MSCs may be more efficient than the exosomes from synovial membrane-derived MSCs for OA treatment. In this study, we utilized ESC, which is also a pluripotent cell line, as the source of MSCs. Consistent with the published results, the exosomes derived from ESC-MSCs effectively maintained chondrocyte phenotype both in vitro and in vivo. Previously, Zhang et al. reported that the exosomes from ESC-MSCs can enhance cartilage repair in a cartilage defect model [22]. However, there is no direct evidence that has demonstrated the regenerative effects of the exosomes from these cells on OA. In this experiment, we used a DMM mouse model, which is currently the preferred animal OA model resembling the progressive OA pathology. Different from a simple cartilage defect, OA is considered a multifactorial cartilage lesion disease within a chronic inflammatory microenvironment. In our model, the cartilage lesion was caused by accumulated damage on the central weight-bearing area of the medial femoral condyle and the medial tibial plateau. This model is also sufficiently sensitive to allow investigation of new therapeutic strategies that confer significant protection against OA progression. In this model, exosomes by themselves were demonstrated to be able to prevent lesion progression and exert a similar regenerative effect as ESC-MSCs. Further in vitro analysis revealed that this effect was mediated by direct contact between exosomes and chondrocytes and involves balancing the synthesis of extracellular matrix protein Collagen type II with expression of the matrix degradation enzyme ADAMTS5.
Although this study showed that ESC-MSC-derived exosomes play a significant role in preventing OA progression similar to ESC-MSCs, there are still some limitations of our results. We set the two animal experiments sequentially not simultaneously. Meanwhile, the frequency of injection was also different between the ESC-MSC and exosomes groups. Hence, we cannot compare the therapeutic efficacy of exosomes with ESC-MSCs directly, as we cannot exclude other factors that may work together with exosomes to facilitate ESC-MSC-mediated OA treatment. In addition, we only set a single time point and a single dosage in the animal experiment. It would be more clinically relevant to add more time points of intervention in our experiments to explore whether there is a therapeutic window in OA disease. Furthermore, the effective dosage of exosomes needs to be optimized. To date, the active ingredients in exosomes are unknown. As the exosome is a complex carrier which contains lipids, DNA, mRNA, miRNA, LncRNA, and even various proteins, they exhibit therapeutic function through immunomodulation, bioenergetics or biochemical effects in different disease models [19,21,24,34]. Hence, the underlying mechanisms of their therapeutic effects on osteoarthritis need to be further investigated in more detail.
In conclusion, the exosomes are one of the effectors produced by ESC-MSCs to alleviate OA by balancing the synthesis and degradation of cartilage matrix. The study of exosomes in OA therapy has just begun, and our results suggest that exosomes from ESC-MSCs have great potential for OA treatment, which merits further investigation.

Conclusions
In this study, we confirmed that ESC-MSCs can alleviate OA in a DMM mouse model. The following in vitro