Curcumin-primed human BMSC-derived extracellular vesicles reverse IL-1β-induced catabolic responses of OA chondrocytes by upregulating miR-126-3p

Curcumin has anti-inflammatory effects and qualifies as a potential candidate for the treatment of osteoarthritis (OA). However, curcumin has limited bioavailability. Extracellular vesicles (EVs) are released by multiple cell types and act as molecule carrier during intercellular communication. We assume that EVs can maintain bioavailability and stability of curcumin after encapsulation. Here, we evaluated modulatory effects of curcumin-primed human (h)BMSC-derived EVs (Cur-EVs) on IL-1β stimulated human osteoarthritic chondrocytes (OA-CH). CellTiter-Blue Viability- (CTB), Caspase 3/7-, and live/dead assays were used to determine range of cytotoxic curcumin concentrations for hBMSC and OA-CH. Cur-EVs and control EVs were harvested from cell culture supernatants of hBMSC by ultracentrifugation. Western blotting (WB), transmission electron microscopy, and nanoparticle tracking analysis were performed to characterize the EVs. The intracellular incorporation of EVs derived from PHK26 labeled and curcumin-primed or control hBMSC was tested by adding the labeled EVs to OA-CH cultures. OA-CH were pre-stimulated with IL-1β, followed by Cur-EV and control EV treatment for 24 h and subsequent analysis of viability, apoptosis, and migration (scratch assay). Relative expression of selected anabolic and catabolic genes was assessed with qRT-PCR. Furthermore, WB was performed to evaluate phosphorylation of Erk1/2, PI3K/Akt, and p38MAPK in OA-CH. The effect of hsa-miR-126-3p expression on IL-1β-induced OA-CH was determined using CTB-, Caspase 3/7-, live/dead assays, and WB. Cur-EVs promoted viability and reduced apoptosis of IL-1β-stimulated OA-CH and attenuated IL-1β-induced inhibition of migration. Furthermore, Cur-EVs increased gene expression of BCL2, ACAN, SOX9, and COL2A1 and decreased gene expression of IL1B, IL6, MMP13, and COL10A1 in IL-1β-stimulated OA-CH. In addition, phosphorylation of Erk1/2, PI3K/Akt, and p38 MAPK, induced by IL-1β, is prevented by Cur-EVs. Cur-EVs increased IL-1β-reduced expression of hsa-miR-126-3p and hsa-miR-126-3p mimic reversed the effects of IL-1β. Cur-EVs alleviated IL-1β-induced catabolic effects on OA-CH by promoting viability and migration, reducing apoptosis and phosphorylation of Erk1/2, PI3K/Akt, and p38 MAPK thereby modulating pro-inflammatory signaling pathways. Treatment of OA-CH with Cur-EVs is followed by upregulation of expression of hsa-miR-126-3p which is involved in modulation of anabolic response of OA-CH. EVs may be considered as promising drug delivery vehicles of curcumin helping to alleviate OA.


Introduction
Osteoarthritis (OA) is one of the most common agerelated degenerative diseases of joints and is also one of the most common causes of disability in elderly individuals [1]. OA is characterized by synovial inflammation and hyperplasia, degeneration of the articular cartilage and menisci, sclerosis of the subchondral bone, and formation of osteophytes, plus different degrees of joint pain. Also affected are tendons, ligaments, and fat pads which increase in volume and number plus peri-articular muscles which develop sarcopenia over the time. These observations lead to the concept that OA affects all tissues of the joint and lead to a dysfunction of the whole joint [2]. However, OA pathogenesis is not fully understood, and no regenerative treatment has been approved to prevent or slow down the disease progression. Total joint replacement is still considered to be the most effective treatment for end-stage OA [3], but the risks of periprosthetic joint infections, aseptic loosening, revision surgery, and other complications have become more and more evident, especially in elderly patients [4,5]. Thus, it is desirable to find novel therapeutic approaches to protect articular cartilage and other joint tissues and slow down OA progression at the early stage of the disease.
Curcumin is a kind of a natural polyphenol compound derived from turmeric [6] (Fig. 1). It is reported that curcumin has anti-osteoarthritic and anti-inflammatory effects and qualifies thus a potential candidate for the treatment of OA [7,8]. However, due to its slow intestinal-liver metabolism, low stability, quick systemic elimination, and its hydrophobic property resulting in low solubility, curcumin has limited bioavailability [9][10][11].
Extracellular vesicles (EVs) are nanoscale lipid-bound vesicles secreted into the extracellular space by almost all cell types [12], which have been characterized in preclinical animal and cell culture models for their therapeutic properties [13,14]. Furthermore, since they are easily taken up by target cells through natural membrane fusion or endocytosis, EVs have been investigated as vehicles to deliver miRNA (miR) and drugs [15]. Accumulating evidence suggests that the solubility, stability, and therapeutic potential of curcumin could be enhanced by encapsulation in EVs [16][17][18][19]. The aim of this study was to evaluate the effect of Cur-EVs on OA-CH. OA-CH were stimulated with IL-1β to establish a standard in vitro OA model. Cur-EVs were used as a cell-free therapeutic approach to treat IL-1β-stimulated OA-CH. We hypothesize that Cur-EVs could reverse the catabolic effects induced by IL-1β in OA-CH by inhibiting inflammatory related signaling pathways miRs are noncoding, endogenous, single-stranded, small RNA molecules of 19-22 nucleotides in length, which are involved in multiple biological processes, including cell cycle control, cell metabolism, cellular differentiation, proliferation, and apoptosis via targeting the 3′-untranslated region (3'UTR) of mRNA [20][21][22]. Numerous pre-clinical in vitro and in vivo studies using OA models have indicated that specific miR profiles are associated with OA [23][24][25]. For example, miR-146, one of the earliest discovered miRs in the articular cartilage, appears to be involved in modulating inflammatory pathways in OA pathology [26]. The growing number of miRs is now reported to play critical roles in modulating molecular mechanisms involved in cartilage matrix degradation [27][28][29][30].
IL-1β is well known as one of the most catabolic members of the pro-inflammatory cytokine family, regulating inflammatory and immune responses in OA pathology [31]. In the present study, we used IL-1β stimulation of chondrocytes obtained from OA-patients (OA-CH) as a standard in vitro OA model to induce catabolic effects resembling metabolic changes in OA pathology. Subsequently, we determined modulating effects of curcumin-primed human bone marrowderived mesenchymal stromal cell (hBMSC)-derived EVs (Cur-EVs) on IL-1β-stimulated OA-CH. To investigate the molecular mechanism of putative anti-catabolic functions of Cur-EVs, we analyzed miR microarray profiles from public datasets and identified a role of hsa-miR-126-3p in metabolism of IL-1β-induced OA-CH. We provide evidence that the protective action of Cur-EVs regarding catabolic responses of IL-1β-induced OA-CH is potentially mediated via upregulating the expression of hsa-miR-126-3p, which is capable of inhibiting activation of Erk1/2, PI3K/AKT and p38 MAPK signaling.

Materials and methods
Experimental outline and hypothesis of this study are shown in Fig. 1.

Ethical statement
The use of human material has been approved by the local ethics committee (No. 14-101-0189; Ethikkommission, University of Regensburg, email: ethikkommission@klinik. ukr.de), and the written consent of all patients has been obtained.

Preparation of curcumin solutions
Curcumin (#C7727-500MG; Sigma-Aldrich, Germany) stock solution (100 mM) was prepared using DMSO (#A994.2; ROTH, Karlsruhe, Germany) as a solvent. Curcumin stock solution was diluted in cell culture medium (hBMSC were cultured in StemMACS Expansion Medium supplemented with 0.2% MycoZap and OA-CH were cultured in DMEM F12 medium, supplemented with 10% FCS and 1% penicillin-streptomycin) to achieve different final concentrations. The pH of the curcumin solution was adjusted to 7.35-7.45 before application. DMSO alone was used as the vehicle control.
Cur-EV isolation 2 × 10 6 hBMSC (passage 3) were seeded in triple T175 flasks (#132867; ThermoFisher, Germany) with Stem-MACS Expansion Medium (Miltenyi Biotec, Bergisch Gladbach, Germany) supplemented with 0.2% MycoZap (#923c1069, Lonza, Switzerland). Expansion medium was exchanged with α-MEM (M8042; Lot: RNBH2983. Sigma-Aldrich GmbH, Steinheim; Germany) medium, supplemented with 10% FCS depl-uc , 10 μM curcumin, and 1% penicillin-streptomycin when cells were 80% confluent. After 48 h incubation, differential centrifugation and ultracentrifugation steps were applied to pellet Cur-EVs from curcumin-primed hBMSC conditioned culture supernatant according to protocols previously published [34,35]. Briefly, we applied sequential centrifugation steps for 10 min at 300×g to remove cells, followed by 10 min at 2000×g to remove dead cells and finally for 30 min at 10,000×g to remove cell debris. This procedure was followed by ultra-centrifugation of the supernatant at 120,000×g for 70 min at 4°C and subsequently dispersing the pellet in PBS, after which the solution was filtered through a 0.22-μm filter. Finally, the Cur-EVs were pelleted at 120,000×g for 70 min at 4°C and the pellet was resuspended in the presence of 25 mM Trehalose (#8897.1, ROTH, Germany) in PBS and stored at − 80°C for use in the experiments. The concentration of Cur-EVs was determined by a BCA Protein assay according to manufacturer's protocol (#23227, Thermo Scientific, Rockford; USA). α-MEM medium, supplemented with 10% FCS depl-uc and 10 μM curcumin, underwent the same isolation procedure by ultracentrifugation as Cur-EVs and is defined as Cur-medium ultra . It functions as a control for the Cur-EV effects.

RNA extraction and real-time RT-PCR analysis
Total RNA of cells was isolated using the Absolutely RNA Miniprep Kit (Agilent Technologies; USA) according to the manufacturer's instructions and reversetranscribed into cDNA using AffinityScript QPCR cDNA Synthesis Kit (#600559, Agilent Technologies; USA). Subsequently, qRT-PCR for assessing relative mRNA levels was performed with Brilliant III Ultra-Fast SYBR® Green QPCR Master Mix (#600882, Agilent Technologies; USA) using a MX3005P QPCR System (Agilent Technologies, Santa Clara; USA). All genes were analyzed relatively, calibrated to the expression of control groups, and normalized to GAPDH and TBP. All qPCR experiments were performed in duplicates using 10-50 ng cDNA.
For hsa-miR-126-3p expression analysis, total cellular RNA was isolated using the miRNeasy Mini Kit (#217004, QIAGEN; USA) according to the manufacturer's instructions, and 30 ng of total RNA was reversely transcribed with the TaqMan MicroRNA Reverse Transcription Kit (#4366596, ThermoFisher; USA) using RT primers of TaqMan MicroRNA Assays for hsa-miR-126-3p and U6 (reference gene) (Applied Biosystems, USA). qPCR was performed in duplicates using 2 ng cDNA, TaqMan primers (hsa-miR-126-3p and U6) of TaqMan MicroRNA Assays and TaqMan Universal PCR Master Mix II, No AmpErase UNG (#4428173, Applied Biosystems). Sequences of all primers used for qRT-PCR in this study are listed in Table 1.

Nanoparticle tracking analysis
Nanoparticle tracking analysis (NTA) was performed to measure distribution of particle size and concentration in EV preparations (control-EVs and Cur-EVs) using a NanoSight NS300 (Malvern Instruments; Malvern; UK). Accuracy of NTA was confirmed with 100 nm polystyrene beads (Sigma-Aldrich, Germany) immediately before measurements. Control-EV samples were diluted 1:100 and Cur-EV samples were diluted 1:200 in PBS and measurements performed at 25°C, five measurements of 30 s were recorded for each EV sample.

Transmission electron microscopy
For the transmission electron microscopy (TEM), freshly isolated EVs were resuspended in cold PBS. For negative staining, 20 μl of the solution was added onto parafilm and a formvar (polyvinyl formal)-carbon coated 400 copper mesh grid (#G400-CU; Science Services, Munich, Germany) was placed on top of the fluid for 10 min at RT. The grid was incubated with 2% phosphotungstic acid (#19500; Science Services, Munich, Germany) for 1 min and dried at RT for 10 min. EVs were investigated with an acceleration voltage of 100 kV and × 100,000 magnification using a Leo 912 AB (Carl Zeiss, Oberkochen, Germany).

Protein extraction and western blot analysis
OA-CH were washed 2 times with PBS and lysed with RIPA buffer (Thermo Scientific, Waltham, MA) containing phosphatase (#04906845001, Roche, Germany) and  proteinase inhibitors (#04693116001, Roche, Germany). The concentration of cellular protein and EVs was quantified using a BCA protein kit assay (see the "Cur-EV isolation" section). Either 10 μg cell lysates or EVs was mixed with SDS-sample loading buffer, followed by boiling for 5 min at 95°C, and subjected to a 12% SDS-PAGE. The proteins were transferred onto 0.22 μm PVDF membranes (Roche, Penzberg, Germany) after electrophoretic separation. Blot membranes were first stained with Ponceau Red solution (loading control), washed and then blocked with 5% BSA for 1 h at room temperature, and incubated with primary antibodies on a shaker overnight at 4°C. After washing, the membranes were incubated with the appropriate horseradish peroxidase-coupled secondary antibodies (Santa Cruz Biotechnology, and Jackson Immuno Research, West Grove, PA). Protein bands were visualized using ECL detection reagents (Thermo Scientific, Germany). Image J (1.8.0, National Institutes of Health, Bethesda, USA) software was used to quantify the density of bands of phosphorylated proteins and normalize them to the total amount of protein bands. Antibodies used for this study are listed in Table 2.

Determination of curcumin concentration in EVs
The concentration of curcumin in Cur-EVs samples was determined using a Nanodrop 1000 spectrophotometer (Thermo Scientific, Waltham, MA) at 420 nm. Before determination, a standard curve of curcumin was established using different concentrations (0 μM, 5 μM, 10 μM, 20 μM, 50 μM, 100 μM, 200 μM) of curcumin solution. The Cur-EVs samples were lysed with RIPA buffer (Thermo Scientific,) for 5 min, sonicated twice for 10 s, and the curcumin quantity was calculated based on OD 420 nm according to the included standard curve of curcumin.

Statistical analysis
All data are expressed as the mean ± SD. Statistical analysis was performed using Prism 8.02 software (Graph-Pad Software, USA). Comparisons between two groups were analyzed by independent two-tailed Student's t tests, and comparisons between more than two groups were analyzed by one-way ANOVA with Newman-Keuls multiple comparison test. Each assay was performed in replicates and repeated at least in three independent experiments. P < 0.05 was regarded as statistically significant.

Cytotoxicity of curcumin regarding hBMSC and OA-CH
In order to test the effect of curcumin on OA-CH and find a biocompatible concentration of curcumin for incubation of hBMSC to isolate EVs containing curcumin as part of their cargo (Cur-EVs), the cytotoxicity of different curcumin concentrations was analyzed after 24 h treatment. Curcumin induced changes in the size and morphology of hBMSC and OA-CH that was particularly evident at higher concentrations (Fig. 2a, b). Concomitantly, the live cell count of hBMSC and OA-CH significant decreased and number of dead hBMSC cells increased significantly when the concentration of curcumin was 20 μM or higher (Fig. 2c-e). Consistent with the results of the live/dead assay, treatment with curcumin at concentrations higher than 20 μM significantly reduced the viability of both hBMSC and OA-CH (Fig. 2f, h). Caspase 3/7 activity of hBMSC was strongly induced by curcumin at concentrations of 20 μM or above, correlating with increased apoptosis (Fig. 2g).
According to these data, 10 μM was considered as the most suitable concentration of curcumin for priming of hBMSC to isolate the Cur-EVs.
Furthermore, the effect of EVs on IL-1β-induced OA-CH was analyzed using a Caspase 3/7 activity assay. IL-1β-induced OA-CH were treated with different concentrations of EVs and the data revealed that EVs with a concentration of ≥ 5 μg/ml reversed increased caspase 3/ 7 activity of OA-CH induced by IL-1β (Fig. 2i).

EV and Cur-EV isolation and characterization
EVs were isolated from control hBMSC and Curcumin hBMSC conditioned culture supernatants, and classical surface makers (CD9, CD63, CD81) of EVs were verified by western blotting (Fig. 3a). Morphology of control EVs/Cur-EVs was monitored by TEM, which revealed particles with an average diameter of 90-120 nm (Fig.  3b, c). Particle size distribution of control EVs/Cur-EVs was determined by NTA, which indicated the average size of both EVs samples to be approximately 120 nm, corresponding to the standard size of EVs Fig. 3d, e).

Quantitative comparison of counts and size between
EVs and Cur-EVs by NTA showed that there was no significant difference between these EV groups, even though number of Cur-EVs was increased by trend (Fig.  3f, g). Spectrophotometry was used to detect the presence of curcumin in Cur-EVs. Cur-EVs emitted a higher absorbance at 420 nm than control EVs, indicating the presence of curcumin in their cargo. The concentration (12 ± 2 μM, n = 3) of curcumin in Cur-EVs was determined according to the standard curve (Fig. 3h). In order to evaluate internalization of EVs/Cur-EVs into OA-CH, the cells were incubated with PKH26-labeled EVs/Cur-EVs for 12 h. Strong intracellular fluorescence signal (red) was detected in OA-CH cytoplasm indicating that both control EVs and Cur-EVs were taken up by OA-CH (Fig. 3i).

Cur-EVs promoted viability and inhibited apoptosis of IL-1β-treated OA-CH
IL-1β significantly inhibited viability and induced apoptosis of OA-CH. The viability of IL-1β-treated OA-CH was significantly higher in the presence of the control EV and Cur-EV treatment groups compared with the groups without EV treatment (Fig. 4a), whereas apoptosis of IL-1β-treated OA-CH was significantly decreased in the control EV and Cur-EV treated groups (Fig. 4b). Notably, treatment with Cur-EVs resulted in stronger reversion of the effects induced by IL-1β compared with control EVs.
Treatment with EVs and Cur-EVs increased the number of vital IL-1β-treated OA-CH significantly. Notably, the cell count of living OA-CH in the Cur-EV treated group was significantly higher than in the control EV group. Concomitantly, the number of dead OA-CH cells was significantly decreased in the IL-1β-treated group in the presence of control EVs and Cur-EVs (Fig. 4c-e). These results suggest that both control EVs and Cur-EVs could promote viability of IL-1β-treated OA-CH by reducing IL-1β-induced apoptosis, but that Cur-EVs were more effective.
Treatment of OA-CH with 10 μM free curcumin (not encapsulated in EVs) did not induce similar anabolic effects when compared to both EV groups but reduced number of dead cells (Fig. 4d). Concentration of 10 μM was chosen according to the concentration of curcumin in EVs (Fig. 3h) and to concentration titration data shown under Fig. 2b, e, h, i.

Cur-EVs induced migration of IL-1β-treated OA-CH
A scratch (wound healing) assay was used to analyze the effect of control EVs and Cur-EVs on migration of IL-1β-treated OA-CH. Treatment with IL-1β reduced migration capacity of OA-CH significantly almost to zero (Fig. 4f). We observed a higher wound-healing (migration) rate in the OA-CH control group, EV control group, and Cur-EV group compared with the IL-1βtreated group. Additionally, the migration rate in the Cur-EV group was significantly higher than in the control EV group, whereas free curcumin only slightly induced migration rate (Fig. 4f, g).
These results indicate that EVs have a positive effect on migration capacity of OA-CH. Of note, Cur-EVs have a stronger capability to restore migration capacity of IL-1β-treated OA-CH than control EVs whereas free curcumin did not achieve comparable effects.

Effect of Cur-EVs on gene expression of IL-1β-treated OA-CH
qRT-PCR was performed to determine relative expression of anabolic and catabolic genes in IL-1β-treated OA-CH after treatment with control EVs and Cur-EVs. As shown in Fig. 5, expression levels of anabolic genes (BCL2, ACAN, SOX9, COL2A1) were increased in the control EVs and Cur-EVs treated groups compared with IL-1β-treated groups (Fig. 5a-d). Furthermore, Cur-EVs induced gene expression of BCL2 and ACAN significantly stronger, compared to control EVs. In contrast, Fig. 2 Effect of different concentrations of curcumin on cell death, viability and apoptosis of hBMSC and OA-CH. Live/dead staining, CellTiter-Blue assay, and Caspase-3/7 assay were used to determine the viability and apoptosis of hBMSC and OA-CH after treatment with different concentrations of free curcumin for 24 h. a, b Vital and dead hBMSC and OA-CH were visualized by fluorescence microscopy after labeling cells with calcein and ethidium homodimers. Living cells were labeled with calcein (green fluorescence) and dead cells were stained with ethidium homodimer (red fluorescence). Scale bar = 200 μm. c-e Quantification of live/dead staining conducted with hBMSC and OA-CH after stimulation with different concentrations of free curcumin (0-100 μM). f-h Quantification of viability and apoptosis assays conducted with hBMSC and OA-CH after treatment with different concentrations of free curcumin (0-100 μM). The 0 μM group without curcumin treatment is defined as control (DMSO only). i Quantification of caspase 3/7 assay conducted with IL-1β-induced OA-CH after treatment with different concentrations of EVs. The OA-CH group (no treatment) is defined as control. Difference to control: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; # difference between groups: # p < 0,05; ## p < 0,01; 1-way ANOVA with Newman-Keuls multiple comparison test; n = 4 expression levels of catabolic genes (IL-1beta, IL6, MMP13, COL10A1) were significantly decreased in the Cur-EVs treated groups compared to the IL-1β groups (Fig. 5e-h). Moreover, gene expression levels of IL6 and MMP13 were significantly decreased in the Cur-EV groups compared with the control EV groups. Free curcumin did not affect gene expression of anabolic genes but also reduced expression of catabolic genes except COL10A; however, it is less effective than both EV groups. These results indicated that although both control EVs and Cur-EVs induced anabolic gene expression and inhibited catabolic gene expression in OA-CH in the presence of IL-1β, the Cur-EVs had a significantly greater effect.
hsa-miR-126-3p was downregulated in OA-CH and IL-1βtreated OA-CH compared with non-OA-CH After analysis of miRNA (miR) microarray datasets from a database, we focused on investigation of hsa-miR-126-3p as the significantly downregulated miR in OA compared to non-OA chondrocytes (Fig. 6a). We observed that expression of hsa-miR-126-3p was significant lower in chondrocytes derived from cartilage tissue of OA patients compared with chondrocytes from non-OA cartilage biopsies (Fig. 6b). Target genes of hsa-miR-126-3p were indicated by evaluation of the public database, KEGG pathway enrichment, which showed that hsa-miR-126-3p is involved in regulation of proinflammatory signaling pathways, including PI3K/AKT signaling pathway, chemokine signaling pathway, MAPK signaling pathway, NF-κB signaling pathway, and the Toll-like receptor signaling pathway (Fig. 6c). These data suggested that reducing expression of hsa-miR-126-3p probably plays a critical role in the pathophysiology of OA.
Cur-EVs reduced phosphorylation levels of Erk1/2, PI3K/ Akt, and p38 MAPK in IL-1β-treated OA-CH via upregulating hsa-miR-126-3p expression To investigate whether the hsa-miRNA-126-3p expression in IL-1β-treated OA-CH is influenced by the control EVs or Cur-EVs, qPCR was performed after 24 h of treatment with IL-1β, Cur-medium ultra , free Curcumin, and both groups of EVs. As shown in Fig. 8a, free curcumin, EVs, and Cur-EVs reversed the IL-1β-induced inhibition of hsa-miR-126-3p expression in OA-CH with Cur-EVs revealing the strongest effect.  b FPKM analysis of hsa-miR-126-3p in Non-OA and OA cartilage samples; independent two-tailed Student's t tests, ****p < 0,001. c KEGG pathway enrichment of targeted genes of hsa-miR-126-3p. d Relative expression of transfected hsa-miR-126-3p mimic, inhibitor, and respective controls measured by qRT-PCR in OA-CH after IL-1β stimulation, n = 4. e Viability and apoptosis of OA-CH after treatment with IL-1β and transfection of the different miR groups; n = 4; g-i 24 h after treatment with IL-1β and transfection with different miR groups, live/dead OA-CH were visualized by fluorescence microscopy after labeling cells with calcein and ethidium homodimers. Living cells were labeled with calcein (green fluorescence) and dead cells were stained with ethidium homodimer (red fluorescence); n = 4. Difference to control (OA-CH): *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; # difference between groups: # p < 0.05; ## p < 0.01; ### p < 0.001; #### p < 0.0001; 1-way ANOVA with Newman-Keuls multiple comparison test. NC, negative control To analyze the effect of Cur-EVs on components of these pro-inflammatory signaling pathways in IL-1βtreated OA-CH, western blotting was performed to evaluate the protein and phosphorylation levels of Erk1/ 2, PI3K/Akt, and p38 MAPK. We detected an increased phosphorylation of Erk1/2, PI3K/AKT, and p38 MAPK in IL-1β-treated OA-CH compared to controls (absence of IL-1β). In the presence of free curcumin, control EVs, or Cur-EVs, phosphorylation levels of all three kinases were reduced. Of note, in the presence of Cur-EVs, the decrease of phosphorylation of Erk1/2, PI3K/Akt, and p38 MAPK in IL-1β-treated OA-CH was stronger than after treatment with control EVs and free curcumin which shows the least effect ( Fig. 8b-d). These results suggested that Cur-EVs are more effective than control EVs and free curcumin in inhibiting IL-1β-induced activation of Erk1/2, p-AKT, and p38MAPK signaling pathways in OA-CH, probably via upregulating hsa-miR-126-3p expression.

Discussion
There is accumulating evidence that pro-inflammatory cytokines and pro-inflammatory signaling pathways are involved in the pathophysiology of OA [37]. Specifically, IL-1β, IL-6, and TNF-α have been reported to play a critical role in the progress of OA in vivo and vitro [38,39]. Additionally, the effects of pro-inflammatory signaling by MAPK, NF-κB, and PI3K/Akt controlled pathways have been implicated in the progression of OA [40][41][42].
hBMSC-derived EVs were reported to exert antiinflammatory roles in OA pathophysiology [43,44]. Curcumin was shown to be an anti-inflammatory agent modulating MAPK, NF-κB, and PI3K/Akt signaling pathways in a rat model of OA and in cell culture experiments with OA chondrocytes [8,[45][46][47]. However, the bioavailability of curcumin is poor because of its low solubility, instability, and rapid systemic elimination [48]. For these reasons, curcumin was encapsulated in EVs in this study by stimulating hBMSC with curcumin and subsequently isolating the EV fraction. These curcumin-primed EVs (Cur-EVs) were used as a nanoscale delivery system of curcumin and to analyze the effects of curcumin on IL-1β-stimulated OA-CH. Our hypothesis was that we would see stronger antiinflammatory and anti-catabolic effects of Cur-EVs compared with control EVs or free curcumin on IL-1βstimulated OA-CH.
Curcumin effects, after being encapsulated in EVs, have been studied already in different in vivo and in vitro experimental settings [19,49,50]. Our characterization methods using western blotting, TEM, NTA, and internalization assays did not reveal significant differences between control EVs and Cur-EVs in terms of standard surface markers, size distribution, counts, shape, and internalization rate, which is consistent with previous reports [17,51].
Previous studies demonstrated that pro-inflammatory signaling pathways play critical roles in chondrocyte catabolism and articular cartilage matrix degradation [52][53][54]. When we treated OA-CH with inhibitors of MAPK/ Erk, PI3K/Akt13, and p38 MAPK before cells were stimulated with IL-1β, we observed that the decreased viability and induced caspase 3/7 activity of IL-1β-induced OA-CH were reversed when Erk1/2, p38, and PI3K/ AKT signaling were inhibited, confirming their role in IL-1β-induced catabolic effects.
In order to analyze the putative anti-inflammatory effects of Cur-EVs, their effect on critical pro-inflammatory signaling pathway components was investigated in IL-1βstimulated OA-CH by determining the phosphorylation status of Erk1/2, PI3K/Akt, and p-38MAPK. Our data indicated that Cur-EVs could reduce IL-1β-induced phosphorylation of Erk1/2, PI3K/Akt, and p-38MAPK and thus reduce activity of pathways involving these kinases. These findings suggest that anti-inflammatory effects of Cur-EVs exist which might be transduced via modulation of IL-1βinduced pro-inflammatory signaling pathway activity.
According to the data of this study, the question arises do we suggest to rather apply miR-loaded EVs?
Several studies demonstrated that miRs could be efficiently loaded into the EVs [65][66][67]. Yujie et al. reported that EVs which are encapsulated with miR-140 alleviate OA progression via inhibiting ADAMTS-5 and MMP-13 [68]. Compared with Cur-EVs, the production process of miR loaded EVs (miR-EVs) requires additional steps to transfer nucleic acids to donor cells. Furthermore, in order to increase the loading efficiencies of miR-EVs, extensive chemical modifications will be needed to be performed, which requires expensive chemically modified oligonucleotides and reagents [69]. Accordingly, we would prefer to load EVs with curcumin and not with miRs as preparing Cur-EVs only requires to stimulate cells with curcumin.
In conclusion, our study demonstrated that Cur-EVs attenuated IL-1β-induced catabolic effects in OA-CH by promoting viability and migration and inhibiting apoptosis. In addition, expression of catabolic genes was downregulated whereas expression of anabolic genes was upregulated in IL-1β-induced OA-CH in the presence of Cur-EVs. The EVs prepared from curcumin-primed BMSC were clearly more effective than EVs from control BMSC and free curcumin, indicating that curcumin encapsulated in EVs potentiates the anabolic effects of BMSC-derived EVs. These effects may in part be attributed to upregulated expression of hsa-miR-126-3p in the target cells induced by Cur-EVs, with the consequence of reducing phosphorylation of components of proinflammatory signaling pathways. Our data indicated that EVs may be considered as a promising drug delivery vehicle of curcumin for treatment of OA, providing an efficient anti-inflammatory intervention strategy.