Characterization and miRNA Profiling of Extracellular Vesicles from Human Osteoarthritic Subchondral Bone Multipotential Stromal Cells (MSCs)

Osteoarthritis (OA) is a heterogeneous disease in which the cross-talk between the cells from different tissues within the joint is affected as the disease progresses. Extracellular vesicles (EVs) are known to have a crucial role in cell-cell communication by means of cargo transfer. Subchondral bone (SB) resident cells and its microenvironment are increasingly recognised to have a major role in OA pathogenesis. The aim of this study was to investigate the EV production from OA SB mesenchymal stromal cells (MSCs) and their possible influence on OA chondrocytes. Small EVs were isolated from OA-MSCs, characterized and cocultured with chondrocytes for viability and gene expression analysis, and compared to small EVs from MSCs of healthy donors (H-EVs). OA-EVs enhanced viability of chondrocytes and the expression of chondrogenesis-related genes, although the effect was marginally lower compared to that of the H-EVs. miRNA profiling followed by unsupervised hierarchical clustering analysis revealed distinct microRNA sets in OA-EVs as compared to their parental MSCs or H-EVs. Pathway analysis of OA-EV miRNAs showed the enrichment of miRNAs implicated in chondrogenesis, stem cells, or other pathways related to cartilage and OA. In conclusion, OA SB MSCs were capable of producing EVs that could support chondrocyte viability and chondrogenic gene expression and contained microRNAs implicated in chondrogenesis support. These EVs could therefore mediate the cross-talk between the SB and cartilage in OA potentially modulating chondrocyte viability and endogenous cartilage regeneration.


Introduction
Osteoarthritis (OA) is a heterogeneous disease that affects synovial joints, which starts from injury or as a molecular alteration and results in structural changes, such as loss of cartilage, bone sclerosis or osteophyte formation, pain, and other clinical symptoms [1]. OA is one of the most prevalent joint diseases and frequently causes disability, with high incidence worldwide due to population aging [2][3][4].
Early OA changes in articular cartilage include regional proteoglycan loss, chondrocyte clustering, collagen disorganization, and tissue fibrillation [5]. While the catabolic/anabolic imbalance in bone is associated to overloading microdamages, in cartilage it seems to be associated with persistent, low-grade inflammation [6]. It begins in the synovial tissue where it is characterized by cytokine release and immune cell infiltration and is commonly associated with pain in OA [7,8]. With OA progression, subchondral bone (SB) gains more influence on OA chondrocytes as osteochondral junction becomes more porous [5,6,[9][10][11][12][13]. Bone-targeting treatments have a positive effect on OA chondrocytes [6] while cocultures of OA subchondral bone osteoblasts [10] or osteoclasts [14] with OA chondrocytes have a negative effect on chondrocytes' viability or anabolic activity. The bone-cartilage crosstalk is therefore emerging as a novel therapeutic target for OA [13,15].
In OA, different types of SB cells enter normally the nonvascular calcified cartilage through the cracks and fissures or along the newly formed vessels, all contributing to cartilage destruction "from below" [12,[16][17][18]. These SB cells include osteo/chondroclasts that dissolve the calcified cartilage matrix [18,19], neurons that exacerbate joint pain [13,20,21], and multipotential stromal cells (MSCs) that may be involved in both, cartilage formation and cartilage degradation [12,18]. Besides the cells themselves, SB-cartilage communication in OA may be mediated via soluble proteins [22], for example, VEGF or IL6, which are increased in damaged SB areas in OA [23], and or via extracellular vesicles (EVs); the influence of which on OA chondrocyte homeostasis remains poorly understood [24].
EVs are nonreplicative bilipid-layered particles that are released naturally from most of the cells, carrying peptides, miRNAs, lipids, and other molecules in their cargo, which can be transmitted to other cells. They can be broadly subcategorised based on their biogenesis, size, and markers into exosomes, microvesicles, and apoptotic bodies [25][26][27]. EVs have a crucial role in cell-to-cell communication, allowing the cargo transfer through endocytic internalization or direct fusion to provoke a biological response [28][29][30][31].
The aim of this study was to investigate EV production from SB MSCs, the cells previously shown to contribute to disease pathogenesis in OA [18,38,39]. We hypothesised that SB MSC-EVs may have an influence on OA chondrocytes' viability and their anabolic gene expression and that they contain specific sets of miRNAs that are different from their parental MSCs. 2.2. Isolation and Culture of OA Chondrocytes. Articular cartilage was harvested from tibiofemoral surfaces, and chondrocytes were isolated as previously described [18]. Briefly, the cartilage was harvested and minced using a scalpel and digested with collagenase overnight. Chondrocytes were expanded in a high-glucose Dulbecco's Modified Eagle Medium (DMEM; MilliporeSigma, USA) supplemented with glutamine, 10% fetal bovine serum (FBS; ThermoFisher Scientific, USA), and 1% penicillin/streptomycin (P/S; Milli-poreSigma, USA). Media were changed twice a week, and subculture was performed when chondrocytes reached 80% confluence, and passaged to passage 2 (p2).

Materials and Methods
2.3. Isolation and Culture of MSCs. Osteoarthritic MSCs were obtained from the subchondral bone of medial femoral condyles after removal of cartilage, as previously described [18]. Medial condyles were chosen as they commonly display a more prominent OA phenotype compared to lateral condyles [18,40]. Bone was weighted and mechanically minced into small fragments with a rongeur and digested with collagenase, as previously described [18]. MSC cultures were established in StemMACs™ MSC expansion media (Miltenyi Biotec, Germany) then transferred to DMEM supplemented with 5% human platelet lysate (PL; PLTMax, MilliporeSigma), 100 IU/ml penicillin, 100ug/ml streptomycin, 2,500 IU/ml heparin and 2 mM L-glutamine (all from MilliporeSigma) (5%PL/DMEM).
Healthy control MSCs were cultured from surplus cells (wash-outs) of hematopoietic stem cell transplantation bags, as previously described [41]. Briefly, bone marrow mononuclear cells (MNCs) were isolated by density gradient centrifugation and cultured in 5%PL/DMEM. Both types of MSCs were previously characterized [18,41] ( Supplementary Figures 1 and 2) according to the criteria set by the International Society of Cellular Therapy (ISCT) [42] and expanded in 5%PL/DMEM to passage 3 before changing the media to EV-depleted 5%PL/DMEM, for EV isolation. EV-depleted MSC media were prepared by 18-hour ultracentrifugation of 10%PL/DMEM at 100,000 × g followed by 1 : 1 dilution in DMEM.

EV Isolation from OA and Control
MSCs. EVs were isolated from both osteoarthritic (OA-EVs) and control healthy (H-EVs) p3 MSCs, as previously described [41]. In brief, when the cells reached 50% confluence, they were washed twice with phosphate buffered saline (PBS, MilliporeSigma) and cultured in a 5%PL/DMEM-EV-depleted medium, for a further duration of 48 hours prior to harvesting the conditioned media and MSC-EV isolation. In addition to collecting the conditioned medium, the number of MSCs in the flask was counted and an aliquot of cells was frozen as pellets in a QIAzol lysis reagent (Qiagen, Germany) for RNA isolation. To obtain MSC-EVs, the conditioned medium was centrifuged at 400 × g for 5 min, 2,000 × g for 20 min at 4°C, then transferred to ultracentrifuge tubes (Beckman Coulter, USA) and centrifuged again sequentially at 10,000 × g for 2.5. MSC-EV Characterization. MSC-EV characterization was performed by electron microscopy (TEM), flow cytometry, and nanoparticle tracking analysis (NTA) [41]. For TEM, 5 μl of MSC-EVs was adsorbed for 30 s onto a carbon-coated, glow-discharged grid. Excess liquid was removed and samples were stained with 1% uranyl acetate (Agar Scientific, UK). Excess uranyl acetate solution was removed, and the MSC-EV-loaded grids were dried then examined using a Hitachi HT7800 transmission electron microscope. Digital images were collected using an Emsis Xarosa camera with Radius software.
EV surface markers CD63, CD9, and CD81 were analyzed by flow cytometry following coating of 4 μm aldehyde/sulfate latex beads (ThermoFisher Scientific) with 10 μl of MSC-EVs suspension. The reaction was stopped by incubation with 1 M glycine (MilliporeSigma), and the MSC-EVbead complex was washed twice with PBS then incubated with mouse anti-human PE CD63 (clone H5C6), PerCPCy5.5 CD9 (clone M-L13), and APC CD81 (clone JS-81) antibodies or corresponding isotype controls (all from BD Biosciences, USA). Following further washes, cells were resuspended in particle-free PBS (ThermoFisher Scientific) and data was acquired using a BD FACS Canto II cytometer (BD Biosciences) and analyzed with FlowJo 10.0 software (Tree Star Inc., USA).
For NTA, MSC-EV pellets were diluted with sterile particle-free PBS and analyzed using Nanosight LM10 (Malvern Panalytical Ltd, UK), as described by the manufacturer's protocol. Three 60 s measurements of the particle size and concentration were measured for each sample. The acquired data was processed using NTA 2.3 software (Malvern Panalytical Ltd).
A second dose of EVs from 6 × 10 4 MSCs was added on day 3, and luminescence measurements were performed on day 5. For this, 100 μl of equilibrated CellTiter-Glo 2.0 reagent was added to the wells and luminescence was measured 10 min afterwards, using the Spark multimode microplate reader (Tecan, Switzerland) and the Sparkcontrol method editor software (Tecan). In these experiments, OA-EVs and H-EVs were from 3 and 6 MSC donors, respectively. 2.7. Three-Dimensional (3D) Pellet Coculture of MSC-EVs and OA Chondrocytes. To characterize whether MSC-EVs (both OA-and H-EVs) have an effect on chondrocytes' expression of genes implicated in the extracellular matrix (ECM) metabolism, EVs from 6 × 10 6 OA or control MSCs were added to 6 × 10 5 chondrocytes resuspended in EVdepleted DMEM:F12/10%FBS media (MSC-EVs to chondrocytes ratio 10 : 1). Afterwards, chondrocytes were separated into 3 different tubes and centrifuged at 450 × g for 10 min to create 3D pellets, for each treatment. In addition, 2 × 10 5 chondrocytes were pelleted in EV-depleted DMEM:F12/10%FBS media without MSC EVs and used as negative controls (No EVs). After 2 and 7 days, EVdepleted media was half-changed and cell pellets were taken for RNA isolation. In these experiments, OA-EVs and H-EVs were from 2 different donors each, and the chondrocyte cultures (n = 3) were not donor-matched to OA-MSCs.
Longer-term effect of OA and healthy MSC-EVs on chondrocyte gene expression in the pellet culture was tested using a single-donor chondrocyte culture and MSC to chondrocyte ratio of 4 : 1, with half-change medium every 3-4 days.

RNA Isolation from OA and Healthy MSC-EVs, Parental
MSCs, and MSC-EV-Treated Chondrocytes. For isolation of RNA from EVs (60 μl of the EVs suspension in PBS), total Exosome RNA and Protein Isolation kit (ThermoFisher Scientific) were used following the manufacturer's instructions. The RNA concentration from EVs was assessed using the Bioanalyzer 2100 and the RNA 6000 pico kit (Agilent Technologies, USA).
RNA from parental MSCs (from which EVs were produced from) and from chondrocyte pellets was obtained using the miRNeasy Mini kit (Qiagen) following the manufacturer's instructions. The RNA was quantified using the Nanodrop spectrophotometer (ThermoFisher Scientific).
2.9. miRNA Profiling from OA and Healthy MSC-EVs. miRNA profiling was carried out using the nCounter® Human v3.0 miRNA Expression Assay Kit (NanoString Technologies), based on miRBase v21, from total RNA obtained from OA-EVs and H-EVs as well as parental MSCs (the MSCs from which the respective EVs were obtained from). The code set incorporated 799 mature microRNAs and included 6 positive controls, 8 negative controls, 6 ligation controls, 5 mRNA housekeeping controls (ACTB, B2M, GAPDH, RPL19 and RPLP0) and 5 spike-in controls. miRtag ligation, miRNA CodeSet Hybridization and posthybridization were performed following the manufacturer's instructions. The resulting miRNA expression profiles were 3 Stem Cells International analyzed using the nSolver software V4 (NanoString Technologies). Samples were normalized to the geometric mean of the Top100 miRNAs taking into account background thresholding and positive control normalization (geometric mean). Fold change (FC) expression differences between groups were calculated using nSolver v2.5 (NanoString Technologies) ratio data, based on normalized count data. Further analysis was performed using a pipeline designed by Newcastle University, Haematological Sciences Department. This integrated a number of R (R project) statistical packages in the R programming language; p values between two groups were generated using a two-tailed t-test. Analysis on miRNA targets was performed using mirWalk [44], mir-Path from DIANA Tools [45], Reactome [46] and miRDIP [47], utilizing default parameters for Human or Homo sapiens.
2.10. Quantitative Real Time PCR. Gene expression was assessed in 2, 7, and 21 day-cultured OA and healthy MSC-EV-treated pellets and control chondrocyte pellets. Total RNA was isolated and reverse transcribed with the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems). qRT-PCR was performed using the Taqman assays (ThermoFisher Scientific; Table 1) and Fast Advanced Master Mix (Applied Biosystems, ThermoFisher Scientific) in a QuantStudio 3 Real-Time PCR system (Applied Biosystems). Genes for the analysis were selected based on the literature evidence of their involvement in the chondrogenesis and chondrocyte catabolic and anabolic activity [12,48] and are shown in Table 1.

The Effects of OA and Healthy MSC-EVs on Chondrocyte
Viability. An ATP-based viability assessment was performed following coculture of OA chondrocytes with OA-or H-EVs. When comparing untreated chondrocytes with those treated with control H-EVs, an average 6.27% increase in chondrocyte viability was found (p value = 0.020) (Figure 2(a)). OA-EV treated chondrocytes also showed increased viability compared to untreated chondrocytes (average 5.92%, p value = 0.042) (Figure 2(b)). However, the difference in the percentage increase of viability between OA and H-EVs failed to reach statistical significance (p value = 0.768) (Figure 2(c)).

The Effects of OA and Healthy MSC-EVs on Chondrocyte
Gene Expression. In these experiments, pellet culture of chondrocytes and EVs was used to facilitate EV uptake by chondrocytes, as reported previously [49]. The media used in these experiments did not contain any chondrogenic inducers because the latter could 'mask' potentially smaller effects by EVs. To confirm that pellet culture environment was sufficient to induce chondrogenic gene up-regulation in chondrocytes, and to investigate how it affected the expression of the selected catabolic and anabolic genes, a time-course experiment in untreated pelleted chondrocytes from 4 donors was performed (Figure 3(a)).
As expected, gradual increases in chondrogenesis markers SOX9, COL2 and ACAN, as well as COL1, were seen from day 2 onwards confirming that pellet culture in the  and ADAMTS5) did not similarly continue beyond day 7 (Figure 3(a)).
An average 18% up-regulation of SOX9 transcript compared to untreated chondrocytes was observed in coculture with H-EVs and the effect was smaller (average 4% increase) in coculture with OA-EVs however the differences failed to reach statistical significance (Figure 3(b)). A trend for lower-level upregulation of SOX9 by OA-EVs compared to H-EVs, as well as of transcripts for mature cartilage ECM proteins COL2 and ACAN was evident following longerterm culture (Figure 3(c)). Higher COL2/COL1 ratio indicating chondrogenic lineage commitment displayed the same trend, unlike the expression of cartilage catabolic molecules where no specific trends were found (both short-and longterm) (Figure 3(c)).

miRNA Profiling.
Albeit not being statistically significant, average of 1.5-fold differences in chondrocytes' chondrogenesis gene expression (SOX9, COL2 and ACAN expression at 21 days), indicated potential differences in microRNA cargo between H-EVs and OA-EVs. microRNA expression profiling was next performed on H-EVs and OA-EVs (n = 4 cultures each) using NanoString technology (n = 799 microRNA). Expression of 590 mature microRNAs was detected across all samples after correcting for background (Supplementary Data 1). Healthy and OA-EVs showed distinct microRNA expression profiles, as demonstrated using unsupervised hierarchical clustering analysis (Figure 4). There were 75 miRNAs that were significantly    (Figure 4).
In addition to microRNAs that were significantly differentially expressed between OA-EVs and H-EVs, we also assessed the top 20 most highly expressed microRNAs in each population, as well as those that were commonly highly expressed in both OA and H-EVs. 17 microRNA were expressed at a high level in both populations, while miR-6721-5p, miR-579-3p and miR-199a-5p were uniquely highly expressed in OA-MSC-EVs, and miR-145-5p, miR-126-3p and miR-15b-5p were uniquely highly expressed in control healthy MSC-EVs (Table 3).
Assessing the top 20 most highly expressed microRNAs, 14 microRNAs were expressed at a high level in both parental OA-MSCs and their EVs, while 6 miRNAs were uniquely highly expressed in OA-MSC-EVs and other 6 were uniquely highly expressed in parental MSCs (Table 4).

Discussion
Previous studies on EVs in OA have almost exclusively focused on the synovial fluid (SF) EVs, which mediate the cross-talk between the damaged cartilage and the inflamed synovium [51,52] and have shown their considerable influence on chondrocyte catabolism, senescence, and death [53][54][55][56][57]. In contrast, the cross-talk between subchondral bone (SB) cells and chondrocytes in OA is comparatively less explored, despite the fact that SB pathology is implicated in all stages of OA progression [13,58,59]. Furthermore, in previous studies of SF-resident EVs, their tissues of origin, as well as the cells of origin, remained unknown. In contrast, our study is uniquely focused on the cross-talk between cartilage and SB in OA and investigated EV production from a specific type of SB cells, SB MSCs, which are directly implicated in SB OA pathology and cartilage destruction "from below" [60][61][62]. Although SB MSCs from healthy amputees would be a better control than the MSCs from bone marrow used in this work, these samples are difficult to access while MSCs from amputated knees from diabetics could carry underlying negative effects such as described previously [63][64][65].
In this work, when comparing EV size and yields (particle/cell) released from OA and control healthy MSCs, no differences were found, similar to results previously described for healthy and OA-EVs from SF [55,57]. Previous works on SF-EVs have presented slightly lower EV sizes [53][54][55], which could be explained by different methods used for EV isolation in these studies: using ultracentrifugation combined with precipitation [54,55] or using immunoaffinity [53]. EVs obtained from ultracentrifugation alone, as was in our study, are known to be less purified due to the overlap in sizes between EVs and microvesicles [25], however ultracentrifugation is less harmful for EVs [66].
One of the principal aims of this study was to identify specific sets of miRNAs in the OA-EVs derived from SB-MSCs. Depending on the cell of origin, EVs (including exosomes) can contain many constituents of a cell including DNA, RNA, lipids, metabolites, and cytosolic and cell-surface proteins [67]. We have identified several miRNAs that in our experimental conditions were packaged into their EVs and may be characteristic of OA-EVs from these cells.

Stem Cells International
Many previous studies compared miRNA expression in MSC-EVs and their parental MSCs [68][69][70][71][72], but these studies were all focused on healthy MSCs. Our study has identified 14 miRNAs that were highly expressed in both OA-MSC-EVs and their parental SB MSCs. These included miR-125b, which has been previously shown to regulate the expression of matrix-degrading enzymes in human chondrocytes [73][74][75] as well as regulate MSC osteogenesis [76], and miR-199a involved in repression of chondrogenesis [77], regulation of chondrocyte ageing and cartilage metabolism [78], and highly expressed in OA SF EVs [57].
Previous studies including ours [18,61,79] have described gene expression profiles of OA SB MSCs as predisposed towards bone formation and cartilage extracellular matrix regulation. The current study complements these findings by showing that highly expressed miRNA in these MSCs, including those packaged in their EVs, may be regulating cartilage homeostasis in OA. Although further functional analysis would be necessary to confirm that these miRNAs contained in the EVs may produce this chondrogenic effect. Furthermore, the present study has identified four miR-NAs that are higher-expressed in OA-MSC-EVs compared to parental MSCs and may be involved in regulating the viability and the metabolic status of OA chondrocytes. Of a particular interest are miR-142-3p previously shown to be selectively packaged into EVs [80] and miR-223-3p that have been both shown to inhibit cell apoptosis and inflammation in OA chondrocytes [81,82], as well as miR-630 that has been earlier documented to regulate chondrogenic lineage commitment [83]. In our previous work we have used qPCR and validated the enrichment of miR-142-3p and miR-223-3p in H-EVs coming from MSCs [41]. The role of miR-630 and other miRNAs has not yet been explored in OA and is awaiting further validation.
Pathway analysis has indicated that OA-EVs and/or OA-MSCs expresses miRNAs implicated in chondrogenesis, stem cells, or other pathways related to cartilage and OA. Interestingly, "fatty acid metabolism" points out to the already known role of fatty acids in OA [84]. Also, FOXO transcription factors have been described as chondroprotectors and regulating autophagy and inflammation [85]. Altogether, these data suggest that despite their altered gene expression and predisposition to osteogenesis in OA, SB MSCs produce miRNAs and release EVs that contain miR-NAs involved in positive regulation of chondrocyte viability and differentiation.
MSC-EVs from various human tissues including bone marrow (BM) and adipose tissue are beginning to be used as a therapy for OA [52,86]. In this context, it was interesting to compare the miRNA cargo of OA-EVs with H-EVs obtained from BM aspirates from healthy individuals. Amongst top highly expressed miRNAs, 17 were shared in both types of EVs (Table 3), 13 of which were also abundant in OA-MSCs (Table 4). Common abundant miRNAs included miR-16-5p previously used as an endogenous reference gene for the normalization of urinary exosomal miRNA expression [87], and some let family members previously described amongst the top 10 most abundant miRNAs in humans' samples [88,89]. Common miRNAs between OA-EVs and H-EVs that were not present in OA-MSCs included previously noted miR-142-3p and miR-630 [86] that have been described as "cartilage-protective." Also miR-29a-3p was described as protective for cartilage [86]. Healthy EVs higher-expressed miR-145-5p (uniquely highly expressed in H-EVs compared to OA-EVs) described as having a dual role in OA [90], while OA-EVs higherexpressed hsa-miR-199a-5p involved in chondrogenesis regulation and bone formation [91], as well as highly expressed in OA SF EVs [57] and in plasma of early OA patients [92]. Our in vitro data have revealed that both H-EVs and OA-EVs marginally but significantly improved the viability of OA chondrocytes, with no difference to each other, which could be a result of consorted activity of multiple miRNAs.
Recent in vitro studies on MSC-EVs in OA have used inflammatory stimulation of chondrocytes before coculturing with H-EVs [93][94][95][96], and they obtained a reduction in the expression of chondrocyte inflammatory markers. It could be possible that this stimulation made the EVeffect on chondrocytes stronger than observed in our study. In our study, we aimed to reflect more closely the in vivo conditions of chondrocytes based on our previous findings that OA chondrocytes grown in the same hsa-miR-199b-5p hsa-miR-15a-5p hsa-let-7c-5p hsa-miR-99a-5p hsa-miR-15b-5p hsa-let7g-5p 11 Stem Cells International conditions already present the signs of inflammation, evident by high-level expression of matrix-degrading enzymes and pro-inflammatory cytokines [18].
Even though our data on MSC-EV (both OA-and H-EVs) effects on chondrocyte gene expression are limited, they point towards a worse induction of chondrogenic gene expression by OA-EVs compared to H-EVs. The lack of statistical significance could be explained by donor-to-donor variation in both chondrocytes and EVs preparations that we have used. To minimise donor effects and generate larger EV batches for use in multiple experiments, future studies could utilise EVs [54] or chondrocytes [49] which are pooled from several donors.
Smaller than expected effects of EVs on chondrocyte gene expression could be also explained by the 3D assay conditions we used. Mortati and colleagues [49] added MSC-EVs to 4-week cultured-chondrocyte pellets and showed that EVs could diffuse into the matrix. Differently from their study, we mixed the EVs with the chondrocytes before the 3D pellet formation as investigated their effect at different time points of chondrogenesis. Also, no chondrogenic inducers such as TGFβ1 [49] were added to the media in our study because these strong inducers could "mask" or conflict with potentially smaller effects from MSC-EVs. In support of our assay design, previous studies have documented the presence of TGFβ1 in MSC-EVs [97], which could explain long-term, enhanced chondrogenic gene expression with the addition of H-EVs in our experiments. While some studies report on MSC-EV enhancement of chondrocyte collagen II expression in 2D formats [94], 3D pellets [48], cells encapsulated in cartilage-mimicking hydrogels [49] are likely to provide conditions more representative to EV traffic in vivo. Specific to OA osteochondral pathology, such constructs should take into the account intercellular distances between SB MSCs and chondrocytes, which can be established from histological studies [6,18], as well as a known heterogeneity of chondrocyte subpopulations in OA [98,99], which is difficult to re-capitulate in in vitro systems.
In summary, OA-EVs were capable to enhance the viability of OA chondrocytes; also, the expression of chondrogenic genes was enhanced in the chondrocytes by the EVs from OA-MSCs. However, both effects were less evident than the ones shown for H-EVs. This variation could be partially explained by the differential expression of miRNAs found in the cargo of H-EVs and OA-EVs.

Conclusions
The present study reports on the production, characterization and miRNA profile of EVs from SB MSCs in OA. For the first time, it shows that SB MSCs in OA are capable of producing EVs that support chondrocyte viability and chondrogenic gene expression, although future studies would be needed to confirm it using functional assays for chondrogenesis. This supports the notion that modulation of SB MSCs represents a valid strategy for endogenous cartilage regeneration in OA. This may be achieved via enhancement of their trophic activity and EV production in addition to triggering their chondrogenic differentiation, thus representing a new potential tool for cartilage regeneration.

Data Availability
miRNA profiling data are attached as supplementary files.

Disclosure
The views expressed in this article are those of the authors and not necessarily those of the BRC, NIHR, or the Department of Health and Social Care.