Cells and culture media

The study was approved for specimen recovery by the French Ministry of Research and Innovation and the Personal Data Protection Ethics Committee (CPP) of Languedoc-Roussillon (approval DC-2010-1185). ASCs were obtained from 7 healthy donors (aged 48 ± 5 years), from surgical residues obtained after aesthetic liposuction. Isolation and characterization of ASCs have been performed as reported previously and expanded in α-MEM medium containing 10% fetal calf serum (FCS), 100 µg/mL penicillin/streptomycin (PS), 2 mM glutamine (Glu), and 1 ng/mL basic fibroblast growth factor (CellGenix, Freiburg, Germany) [16]. ASCs were used in passage 2.

Chondrocytes were isolated from the femoral condyles of patients with OA undergoing total knee replacement surgery (mean age: 74 ± 2 years). Briefly, cartilage slices were incubated in 2.5 mg/mL pronase (Sigma-Aldrich, Saint-Quentin-Fallavier) at 37 °C for 1 h followed by 2 mg/mL collagenase type II (Sigma-Aldrich), at 37 °C overnight. Cell suspensions were then filtrated through cell strainer filters with 70 μm pores and cultured in DMEM/PS/Glu/10% FCS (proliferative medium) at the density of 25,000 cells/cm² till the end of passage 0.

Synoviocytes were isolated from synovial tissue from patients with OA undergoing total knee replacement surgery (mean age: 68 ± 6 years). Synovial tissue was fragmented and incubated with collagenase II (2 mg/mL) at 37oC for 2 h. Digested tissue was passed through cell strainers and the flowthrough was analyzed by cellular viability and number of cells. Cells were resuspended in FCS supplemented with 10% of DMSO and frozen before any expansion. For analysis of fibroblast-like synoviocytes, total synoviocytes were thawed and expanded in DMEM supplemented with 10% FCS, 100 µg/mL PS and 2 mM Glu until passage 4, before being used in experiments.

For coculture assays, chondrocytes (500,000 cells/well) were plated on the bottom of 6-well plates and treated with IL1β (10 ng/mL) for 48 h. Different doses of EVs (100, 500 and 2500 ng) were then added on top of chondrocytes and cultured in 3 mL of minimal medium (DMEM supplemented with PS, 0.35 mmol/L proline, 0.17 mmol/L ascorbic acid and 1 mmol/L sodium pyruvate) for 7 days or in proliferative medium containing 10 ng/mL IL1β for 2 days. Synoviocytes were thawed and plated on 6-well plates (500,000 cells/well) and treated with LPS (100 ng/mL) or a combination of IFNγ (20 ng/mL) and TNFα (10 ng/mL) for 24 h. Thereafter, 500 ng of EVs was added in 2 ml of minimal medium for 24 h before cells were analyzed.

Model of etoposide-induced senescence

ASCs were seeded in proliferative medium in 6-wells or 24-well plates at the density of 2,000 cells/cm² for control cells and 6,000 cells/cm² for etoposide (ETO) treatment. ETO (25 µM) was added for 24 h and ASCs were then rinsed three times with PBS before fresh medium addition for 7 or 12 days.

Cell proliferation assay

ASCs were trypsinized and the number of live cells was counted using Trypan Blue exclusion dye. Proliferation was assessed using the Cell Proliferation Elisa BrdU assay as described by the manufacturer (Roche SAS, Boulogne-Billancourt).

Senescence-associated β-galactosidase

ASCs were fixed with 2.5% glutaraldehyde at room temperature (RT) for 5 min. After three washes with PBS, cells were incubated with a solution of 5 mM Potassium ferrocyanide, 5 mM Potassium ferricyanide, 200 mM citric acid/sodium phosphate buffer pH 6, 150 mM sodium Chloride, 2 mM magnesium chloride, 1 mg/mL of X-gal (Promega, Charbonnières-les-Bains, France) at 37 °C for 6 h. Cells were then examined and photographed under a microscope (EVOS M5000, Invitrogen, Illkirch). SA-β-Gal+ ASCs were quantified with the ImageJ software using the cell counter complement. Additionally, SA-β-Gal activity was assessed using the 96-Well Cellular Senescence Activity Assay following the manufacturer’s instructions (Cell Biolabs, Clinisciences, Nanterre). Fluorescence was measured using a Varioskan Flash microplate reader (ThermoFisher Scientific, Illkirch-Graffenstaden).

Immunofluorescence assays

ASCs were fixed with 4% formaldehyde at RT for 15 min and permeabilized with 0.3% Triton X-100 in PBS for 1 h. For γH2AX staining, 5% normal goat serum was added during the permeabilization step. After three washes with PBS, ASCs were incubated with Phospho-histone γH2AX (Ser139) antibody (Cell Signalling Technology, Ozyme, St-Cyr-l’Ecole) at 4 °C overnight. Cells were then washed three times and incubated with Alexa Fluor 594 goat anti-rabbit IgG (H + L) (ThermoFisher Scientific) at RT for 2 h and DAPI for 10 min. Finally, cells were photographed under a microscope (EVOS M5000, Invitrogen). The γH2AX foci and nucleus area were quantified using the Image J software and the cell counter complement.

For F-actin staining, ASCs were incubated with 50 µg/mL fluorescent phalloidin conjugated with Tetramethyl-rhodamine B isothiocyanate (TRITC) in PBS at RT for 40 min. Cells were then stained with DAPI for 10 min and washed three times with PBS before being photographed. Using the Image J software, the cell surface and the corrected total cell fluorescence (CTCF) were quantified by the formula: CTCF = Integrated Density – (Area of selected cell x Mean fluorescence of background readings).

RNA extraction and RT-qPCR

Total RNA was extracted from cells using 350 µL RLT buffer from the RNeasy Mini Kit according to the supplier’s recommendations (Qiagen, Les Ulis). Reverse transcription of 500 ng RNA was obtained by M-MLV reverse transcriptase (ThermoFisher Scientific). Real-time PCR was done on 10 ng cDNA using SYBR Green I Master mix (Roche Diagnostics, Meylan) and specific primers (Table 1).

Table 1 List of primers for PCR

Total RNA was extracted from joint tissues after mechanical dissociation of joints previously stored in Trizol using Ultra-Turrax. Chloroform (200 µL) was then added to the suspension of crushed tissue and incubated at RT for 3 min before centrifugation at 12,000 ×g, at 4 °C for 15 min. The aqueous phase was recovered and mixed with 600 µL of 70% ethanol. The suspension was then loaded onto columns of the RNeasy Mini Kit according to the supplier’s recommendations (Qiagen, Les Ulis).

Values were normalized to the Ribosomal Protein S9 (RPS9) housekeeping gene and expressed as a relative expression or fold change using the respective formulae 2−ΔCT or 2−ΔΔCT.

Protein detection and ELISA assays

Cells and EV pellets were resuspended with RIPA buffer (Sigma) and incubated for 20 min on ice for protein extraction. After protein quantification by Micro BCA Protein Assay Kit (ThermoFisher Scientific), protein preparations were analysed by Western Blot and ELISA.

For Western blot analysis, 20 µg of proteins were mixed with lithium dodecyl sulfate (LDS) and reducing buffer to a final volume of 40 µL, and incubated at 70oC for 10 min. The protein preparation was loaded and separated in Novex 4 to 20%, Tris-Glycine polyacrylamide gels (Thermo Fischer Scientific), and then transferred to nitrocellulose membrane using the iBlot2 system (Life Technologies). The transferred membrane was then blocked with Tris-buffered saline, 0.1% Tween 20 (TBS-T) with 5% milk, incubated with primary antibodies for LOXL4 detection (ab88186, 1:2000, Abcam) and β-actin (AC-15, 1:5000, Sigma), and then the secondary anti-rabbit (#7074, 1:2000, Cell Signaling Technology) and anti-mouse (A9044, 1:100000, Sigma) antibodies. The membrane was treated with Sirius HPR substrate (Advansta, San Jose, USA) and analysed with the ChemiDoc (BioRad, Hercules, USA) imaging system.

Supernatants from cells were recovered and stored at -20 °C until use. Quantification of MMP-3, MMP-13, IL-6, IL8, HGF and VEGF was performed using specific Enzyme-Linked Immunosorbent Assays (Bio-Techne, RnD Systems, Rennes). ICAM-1 (Abcam, Cambridge, UK), GPC1 (RnD Systems), APOE (Abcam), COL15A (Novus Biologicals, Littleton, USA) were detected in cell supernatant, cell lysate (1 µg) and/or EV lysate (5 µg) using ELISA kits as per manufacture recommendation.

Flow cytometry for marker detection on synoviocytes

Synoviocytes were gently detached from cell culture plates by incubation with Versene (Gibco) for 45 min at 37oC before being stained with REAfinity antibodies anti-human CD45, CD73, CD11b, CD86, CD80 and CD163 (Miltenyi) at RT for 15 min and washed in PBS containing 0.5% BSA and 2mM EDTA, as per manufacturer’s recommendation. Cells were analyzed using the FACS Canto II Cytometer (BD).

EV Production, isolation and characterization

Production of conditioned medium (CM) and isolation of EVs were performed as previously described [17]. Briefly, ASCs were cultured in α-MEM medium containing 3% EV-depleted FCS for three days. Total EVs were recovered from the conditioned medium by differential ultracentrifugation (last step at 100,000×g, 4 °C for 2 h). EVs were characterized as recommended by the International Society of Extracellular Vesicles (ISEV) [18]. EV size and concentration were determined by Nanoparticle Tracking Analysis, as described [17]. The structure was observed using cryo-TEM while the total RNA and protein contents were determined using the RNeasy Micro Kit (Qiagen, Les Ulis) and the Micro BCA Protein Assay Kit (ThermoFisher Scientific), respectively. Surface marker expression was analyzed using fluorophore-conjugated antibodies. EVs (1 µg of equivalent proteins) were coated onto 4 μm aldehyde/sulfate latex beads by incubation at 4 °C overnight. Beads coated with EVs were then washed 3 times in PBS and incubated with specific antibodies for CD11b, CD29, CD44, CD45, CD63, CD81, CD90 and HLA-ABC (BD Biosciences) by flow cytometry. EVs were used freshly prepared.

Mass spectrometry and analysis

EVs (2 µg) were lysed and total protein content was reduced, alkylated and digested with trypsin. The peptides were then desalted and injected on a nanoLC-Q-TOF Impact II (Brüker). Each sample was injected in triplicate.

Protein identification was performed with Maxquant software (version 1.6.17.0). The parameters used were the following: the digestion enzyme is trypsin, the number of missed cleavages is 1, a mass tolerance of 10 ppm for parent ions and 0.05 Da for MS/MS spectra was used, the minimum peptide size is 5 amino acids, the maximum peptide mass is 4600 Da and a protein identification FDR was set at 2.5%. The Uniprot database of the human proteome was used as reference (version 01/02/2021). Some modifications, induced by the sample preparation protocol, were studied: asparagine deamidation and methionine oxidation as variable modifications and cysteine carbamidomethylation as fixed modification. The maximum number of modifications for a peptide was 5.

LFQ intensities for each protein were normalized by the initial protein amount before data processing, performed with the LFQ-Analyst platform. Proteins considered as contaminants and redundant were removed. For 3 biological replicates, proteins without at least 2 valid values for at least one group were eliminated. The LFQ data for each protein were transformed by applying the log2(x) formula and grouped according to donors and conditions (4 donors: D1, D2, D3, D4 and 4 conditions: Control, 2DG, TI, Oligo). The data was then normalized to meet a normal distribution and missing values were imputed. Protein-wise linear models combined with empirical Bayes statistics were used for the differential expression analyses. The limma package from R Bioconductor was used to generate a list of differentially expressed proteins for each pair-wise comparison. A cutoff of the adjusted p-value of 0.05 (t-statistic correction) along with a |log2 fold change| of 0.5 has been applied to determine significantly regulated proteins in each pairwise comparison.

Collagenase-induced osteoarthritis model and joint imaging

The collagenase-induced OA (CIOA) murine model was performed as previously described [19] and following guidelines and regulations of the Ethical Committee for animal experimentation of the Languedoc-Roussillon (Approval APAFIS#5349-2016050918198875). Briefly, 5 µL of 1Unit type VII collagenase or saline (control group; CT) were administered intra-articularly (IA) in the knee joint of 10 weeks-old C57BL/6 mice at days 0 and 2. At day 7, collagenase-injected groups of 22 mice received IA injections of EVs (250 ng/5 µL) or 5 µL saline (CIOA). Part of the mice (n = 7/group) were euthanatized on day 14 and knee joints were recovered. Soft tissues around the joints were carefully removed, and the joints were stored in 1 mL Trizol at -80 °C.

On day 42, paws were recovered from the other part of the mice (n = 15) and fixed in 4% formaldehyde. For bone analysis, hind paws were scanned in a Micro-Computed Tomography (µCT) scanner (SkyScan 1176, Bruker, Kontich, Belgium) and 3D image stacks were reconstructed using the NRecon software (Bruker). The quantification of the subchondral bone of the tibia and calcification of the meniscus and ligaments was performed using the CTAn software (Bruker). Reconstructed 3D images of joints were obtained using Avizo software (Avizo Lite 9.3.0, FEI, France). For cartilage analysis, tibia plateaus were scanned with a confocal laser scanning microscope (CLSM; TCS SP8, Leica Microsystems, Nanterre, France). Stacks of images were analyzed to quantify several cartilage morphometric parameters using the Avizo software (FEI Visualization Sciences Group, Lyon).

Histological analysis

After µCT and CLSM analyses, hind paws were decalcified using 5% formic acid at RT for 2 weeks and then embedded in paraffin. Frontal sections of tibias were cut (3 slices of 7 μm each 100 μm; first section at 50 μm below the cartilage surface) and stained with safranin O and fast green. Cartilage degradation was quantified using the modified Pritzker OARSI score.

Statistical analysis

Statistical analyses were performed using the GraphPad 9 Prism Software. Data distribution was assessed using the Shapiro–Wilk normality test. The statistical analyses are indicated in the figure legends. Data are presented as mean ± SEM. * p < 0.05; ** p < 0.01; *** p < 0.001; ****< 0.0001.