Media and reagents

Dulbecco’s minimal essential medium (DMEM) and trypsin–EDTA were purchased from Biowest. Minimal essential medium alpha (MEM-α), Iscove’s modified Dulbecco’s medium (IMDM), HEPES, non-essential amino acids solution (NEAA), sodium pyruvate, β-mercaptoethanol, insulin-transferrin-selenium (ITS + 2), fetal bovine serum (FBS), and TrypLE™ select were from Gibco. L-glutamine, penicillin–streptomycin (P/S), and phosphate-buffered saline (PBS) were from Corning. Human platelet lysate Stemulate® (HPL) was from Sexton Biotechnologies, Cryostor® CS10 from BioLife Solutions, and interferon gamma (IFN-γ) from Biotechne. Collagenase II, Pronase E, cell culture-grade DMSO, methanol-free paraformaldehyde (PFA), bovine serum albumin (BSA), Triton X-100, Tween-20, dexamethasone, ascorbate 2-phosphate, proline, phytohemagglutinin (PHA) were from Sigma. Rapamycin was from Merck. The MesenCult™ ACF Chondrogenic Differentiation Kit, MesenCult™ Osteogenic Differentiation Kit (Human), and MesenCult™ Adipogenic Differentiation Kit (Human) were from STEMCELL™. CellTrace™ Violet, LIVE/DEAD™ Fixable Near-IR, and Hoechst 33,342 were purchased from ThermoFisher.

Antibodies

The anti-SOX9 mouse monoclonal antibody, clone GMPR9 (1:200), anti-collagen II mouse monoclonal antibody, clone 2B1.5 (1:200), goat anti-rabbit IgG (H + L) highly cross-adsorbed secondary antibody, Alexa Fluor™ 647 (1:500) and goat anti-mouse IgG (H + L) cross-adsorbed secondary antibody, Alexa Fluor™ 488 (1:500) were purchased from ThermoFisher. The anti-CDKN2B rabbit polyclonal antibody (orb213719, 1:200) was from Biorbyt and the anti-COL1A1 XP® rabbit monoclonal antibody (E8F4L, 1:500) and anti-PPAR-γ rabbit monoclonal antibody (C26H12, 1:400) were from Cell Signaling. The APC-coupled anti-human CD274 (B7-H1, 1:100) antibody was from BioLegend.

Human adipose-derived stromal cell isolation, expansion and treatment

Human AD-MSCs were isolated from abdominal adipose tissue obtained from four healthy adults after they signed an informed consent, as previously described [19]. The donors’ body mass index ranged from 24.7 to 30.0. AD-MSC production was standardized to comply with the GMP. Cells were expanded twice in Quantum hollow-fiber bioreactors (Terumo BCT) in MEMα containing 5% HPL and were characterized as MSCs according to the ISCT recommendations. AD-MSCs were frozen in Cryostor® CS10 at passage 1 (P1). For experiments, cells were thawed and resuspended in MEMα, 5% HPL, 1% P/S. For amplification, AD-MSCs (4.5 × 103 cells/cm2) were seeded in flasks and maintained at 37 °C in the same medium for 4 days. For IFN-γ stimulation, AD-MSCs (4 donors) were seeded at 1.5 × 104 cells/cm2 in 6-well plates for 24 h before incubation with 20 ng/mL IFN-γ in the presence of 10 nM rapamycin or 0.01% DMSO (control). After 48 h, cells were processed for RNA sequencing or flow cytometry analysis.

Human chondrocytes isolation and culture

Human OA chondrocytes were isolated as previously described by Malaise et coll. [20], from cartilage obtained from patients with OA who underwent total knee arthroplasty at Lapeyronie hospital, France, after obtaining their informed consent. Briefly, cartilage from the tibial plate and condyles was removed and cut into small pieces. The recovered cartilage samples were weighed, washed twice with DMEM, 1% P/S, 1% L-glutamine, and digested with Pronase E (50 µg/mL) in the same medium for 1 h. Then, cartilage samples (5 g per digestion at most) were washed twice and digested with collagenase II (2 mg/mL-10 mL) in the same medium overnight. The next day, samples were neutralized with 1 volume DMEM, 10% FBS, 1% P/S, 1% L-glutamine and passed through a 70 µm sieve. Cell suspensions were centrifuged at 300 g for 5 min, washed once, plated (4,000 cells/cm2) and expanded twice.

Colony-forming units (CFUs)

AD-MSCs (3 donors) were seeded (100 cells per well) in 6-well plates in triplicate (9.6 cm2 ~ 10 cells/cm2) in MEMα, 5% HPL, 1% P/S. The next day, 10,000X stock solution of rapamycin (in DMSO) was diluted 1/10 in medium and added (1/1000) to the wells. Medium was not changed during the 10 days of culture. Then, cell colonies were fixed with ice-cold methanol for 5 min, washed once with water, and stained with 2% Giemsa solution for 10 min with agitation. Cells were washed three times with water and dried. CFUs were counted manually.

Reverse-transcription and quantitative polymerase chain reaction (RT-qPCR)

RT was performed using M-MLV Reverse Transcriptase, a Random Hexamer (both from Invitrogen™) and 250 ng of RNA at the final concentration of 12.5 ng/μL. The resulting cDNA was diluted five times (2.5 ng/μL) for qPCR analysis that was performed using 5 ng (2 μL) cDNA, the LightCycler® 480 SYBR Green I Master (Roche) and a Roche LightCycler® 480 Instrument II (384-well plates). The primer sequences are in Table S1.

AD-MSC differentiation into chondrocytes, adipocytes and osteoblasts

For adipogenesis and osteogenesis, P1 AD-MSCs (4 donors) were thawed, seeded (4.5 × 103 cells/cm2) in MEMα, 5% HPL, 1% P/S. Cells were plated in 24-well, 6-well and black 96-well half surface plates for Oil Red O staining (adipogenesis), Alizarin Red staining (osteogenesis), RT-qPCR and immunofluorescence analysis, respectively. After 4 days (cells at 95% of confluence), medium was removed and replaced with the reagents from the Mesencult™ Osteogenic or Adipogenic Differentiation Kit (STEMCELL) in the presence of 10 nM rapamycin or 0.01% DMSO (control). Osteogenic differentiation was evaluated after 10 days (adipo) and 14 days (osteo). Analysis by RT-qPCR, Oil Red O and Alizarin Red staining were then performed (see Supplementary data for details). For chondrogenesis, AD-MSCs (5 × 105 cells from 3 donors) were differentiated in 3D conditions in the presence of the Mesencult™ chondrogenic medium. with/without 10 nM rapamycin. After 21 days, cartilage pellets were lysed for chondrogenic marker quantification by RT-qPCR or stained with Alcian blue.

Immunosuppression assay

Peripheral blood mononuclear cells (PBMCs) from three healthy donors (buffy coats from the Etablissement Français du Sang) were isolated using the Ficoll procedure, pooled and cryopreserved in DMSO supplemented with 10% FBS. Immunosuppression assays were performed by co-culturing AD-MSCs (4 donors) with activated PBMCs at different ratios, as previously described [19] (see Supplementary data for details). To evaluate AD-MSC inhibitory effect on PBMC proliferation, the percentage of PBMC proliferation in co-culture was compared to the proliferation of PHA-activated PBMC alone (set to 100%). The difference between conditions represents the percentage of inhibition.

ELISA analysis of GAS6 secretion by AD-MSCs

GAS6 secretion by AD-MSCs was assessed using the Human GAS6 ELISA Kit (Invitrogen – BMS2291) following the manufacturer’s instruction, but by initially diluting samples five times in diluent buffer (see Supplementary data for details).

AD-MSC and chondrocyte transwell co-cultures

To mimic the physiological conditions of intra-articular injection of AD-MSCs, 5 × 105 OA chondrocytes (3 donors) were seeded into wells of 6-well plates with/without 7.2 × 104 AD-MSCs (3 donors) on inserts as previously described [21]. Before co-culture, chondrocytes and AD-MSCs were grown alone for 24 h. For co-culture, medium was replaced with minimal chondrogenic medium (DMEM, 1% P/S, 1% L-glutamine, 1 mM sodium pyruvate, 50 mg/mL ascorbate 2-phosphate, 40 mg/mL proline, 1% ITS + 2, 100 nM dexamethasone) in the presence of 10 nM rapamycin or 0.01% DMSO for 3 days. Total RNA was extracted for RNA sequencing analysis.

Immunofluorescence (IF) staining

For IF analyses, cells were plated in black 96-well plates with half-area well and µclear® bottom (Greiner Bio-One). Cells were fixed with 3.6% PFA/PBS for 30 min, washed twice with PBS and stored at 4 °C for less than 1 week. Before staining, cells were washed once with PBS, permeabilized with PBS/3% BSA/0.3% Triton X-100 for 3 min, washed twice and incubated with a solution containing PBS, 3% BSA, 0.05% Tween-20 for 30 min. Then, cells were incubated with two primary antibody pairs (SOX9/p15INK4B and COL1A1/COL2 for chondrocytes) diluted in PBS/3% BSA/0.05% Tween-20 at room temperature in humidified chambers for 4 h. After two washes with PBS/0.05% Tween-20, cells were incubated with secondary antibodies diluted in PBS/3% BSA/0.05% Tween-20 at 37 °C for 1 h, washed twice with PBS/0.05% Tween-20, and incubated with Hoechst 33,342 (2 µg/mL) in PBS for 5 min. After washing once with PBS, a 90% glycerol-PBS solution was used as mounting medium. A Leica TCS SP8 confocal microscope with LAS X navigator mode and motorized stage was used for image acquisition.

Flow cytometry analysis

AD-MSCs were trypsinized using TrypLE™ Select, incubated with 250 ng of APC-coupled anti-CD274 (PD-L1) antibody or APC-coupled isotype on ice, in the dark for 30 min. Then, cells were washed twice with PBS/1% BSA and resuspend in 500 µL of this buffer before analysis with a FACS Canto II flow cytometer (BD Biosciences). PD-L1 expression in AD-MSCs was quantified with Kaluza Analysis 2.1. The inhibitory effect of AD-MSCs on PBMC proliferation was quantified with the FlowJo software (FlowJo, Ashland, OR, USA).

RNA preparation and sequencing

Total RNA was isolated from AD-MSCs or OA chondrocytes using the RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. The RNA Integrity Number was evaluated on Agilent chips and for all samples it was > 9. Libraries were prepared by BGI Genomics (Hong Kong, China) using 500 ng of total RNA and the MGIEasy RNA Library Prep Kit (MGI, China). Library quality was assessed on a Bioanalyzer 2100 (Agilent) and quantity by qPCR. The final libraries were formatted as single-stranded circular DNA. Libraries were sequenced on a DNBSEQ G400 sequencer (MGI, China) (paired-end 150) and at least 30 M clean reads were obtained for each library.

RNA sequencing data analysis

RNA sequences were aligned using HiSAT2 to the reference human genome. Differentially expressed genes (DEGs) were identified with the Deseq2 R package. Results were considered significant when Q-value < 0.05. To obtain a reference genome-wide signature from RNA sequencing data of healthy chondrocytes, published data from Fisch et coll. [22] on chondrocytes isolated from cartilage samples of 18 healthy donors and 20 patients with OA were used. The raw count table from GSE114007 was retrieved for the Gene Expression Omnibus databank and analyzed using the EdgeR package. A pre-selection threshold of at least 380 total counts per row (at least 10 counts per sample) was applied. All genes with FDR ≤ 0.05 without log2(FC) thresholding were selected. Heatmaps and volcano plots were generated with the Complex Heatmap and Enhanced Volcano packages, respectively. The list of all OA dysregulated genes was submitted to the KEGG Pathways and REACTOME databases to identify pathways that were significantly affected in OA. Then, the lists of upregulated and downregulated gene underwent the same process to identify upregulated/downregulated genes in the affected pathways. Pathways significantly affected by OA and the number of "hit" genes were extracted and transferred to GraphPad to be displayed as a double-sided histogram. Then, to identify markers associated with OA physiopathology and senescence, normalized counts of these genes were extracted, converted to log10(Norm.count + 1) and plotted using GraphPad. Similarly, cytokines and growth factors deregulated in OA were identified.

Statistical analysis

All data are presented as median or mean ± SEM. Results from multiple conditions were compared by two-way ANOVA. The Mann–Whitney Student’s t-test was used for comparisons between two experimental groups. Data were analyzed using the Prism software v10 (GraphPad Software Inc.). P values < 0.05 were considered significant (*p < 0.05; **p < 0.01; ***p < 0.001).