Equine ASCs were isolated from subcutaneous AT from the horse hip and obtained during surgery by veterinary doctors on tranquilized horses. The protocol was approved by the Ethical Committee of VetAgro Sup (permit number: 69 127 800, certified number: B 69 127 0501). AT was cut into small pieces and digested with 250 U/mL type 1 collagenase (Worthington, Serlabo Technologies, Entraigues) at 37°C for 1.5 h. The stroma vascular fraction was collected by centrifugation (300 g, 10 min), and cells were filtered successively through a large sterile filter and, then, through 100 and 70 μm porous membranes (Cell Strainer, BD Biosciences, Le Pont de Claix). Erythrocytes were lysed using red lysis buffer (NH₄Cl 155 mmol/L, KHCO₃ 10 mmol/L, EDTA 0.11 mmol/L, pH 7.3). Cell counting and viability were evaluated with calibrated Vicell Beckman Coulter. Cells were plated as passage 0 at the initial density of 4000 cells/cm² in αMEM GlutaMAX™ (ThermoFisher Scientific, Illkirsch) supplemented with 100 U/mL penicillin, 100 mg/mL streptomycin, 10 μg/mL gentamicine/0.25 μg/mL amphotericin B (ThermoFisher scientific), 10% fetal calf serum (FCS, ThermoFisher scientific), and 1 ng/mL basic fibroblast growth factor (bFGF; Sigma-Aldrich, Saint-Quentin-Fallavier). After 1 week, cells were trypsinized and expanded at 2000 cells/cm² till day 14, where ASCs at passage 1 were used. For IFNγ priming, 100 ng/mL of equine IFNγ (R&D Systems, Lille) was added in the culture medium for 24 h before use of the cells. mMSC were isolated from BM of C57BL/6 mice and characterized, as previously described (21).

Equine ASCs (200,000 cells) in PBS containing 0.2% bovine serum albumin (BSA) were incubated with different antibodies: non-conjugated ELA-class I (clone CVS22, AbD Serotec, Bio-Rad, Marnes-la-Coquette) or -class II (clone CVS20, Bio-Rad) primary antibodies coupled with Alexa Fluor 488 secondary antibody, PE-conjugated CD29 (clone 4B4LDC9LDH8, Beckman Coulter), CD44 (clone CVS18, Bio-Rad) or CD90 (clone 5E10, Bio-Rad), or the respective isotype control (clone MOPC-21, Bio-Rad) at room temperature for 60 min in the dark. The labeled cells were then analyzed by multiparameter flow cytometry using a LSR II cytometer and the BD FACSDiva™ software V.6.1.3 (BD Biosciences).

Adipogenic differentiation was induced after plating eASCs at the density of 1500 cells/cm² in proliferative medium for 7 days. Medium was changed by adipogenic medium (DMEM-F12, 100 U/mL penicillin, 100 μg/mL streptomycin, 15% Rabbit Serum, 1 μM dexamethasone, 60 μM indomethacin, 10 μg/mL insulin, 0.5 mM IBMX), for 21 days. Adipogenesis was assessed by quantification of adipocyte markers by RT-qPCR, and lipid droplets were visualized by HCS LipidTOX™ Green neutral lipid stain (ThermoFisher Scientific). Osteogenesis was induced by culture at low density (1500 cells/cm²) in osteogenic medium (DMEM, 10% FCS, 100 U/mL penicillin, 100 μg/mL streptomycin, 0.1 mM ascorbic acid, 0.1 μM dexamethasone) for 21 days. Osteogenic differentiation was assessed by quantification of osteoblast markers by RT-qPCR and extracellular matrix mineralization detected by alizarin red S staining. Chondrogenic differentiation was induced by pelleting and culturing eASCs (250,000 cells/tube) in chondrogenic medium {high glucose DMEM, 0.1 μM dexamethasone, 1 mM sodium pyruvate, 170 μM ascorbic-2-phosphate acid, 0.35 mM proline, ITS [insulin/transferrin/selenic acid (Lonza, Koln, Germany)], 10 ng/mL human TGF-β3}. Chondrogenesis was assessed by quantification of chondrocyte markers by RT-qPCR and Safranin O–Fast green staining.

Total RNA was extracted from cells using the RNeasy kit (Qiagen, Courtaboeuf). RNA (500 ng) was reverse transcribed using the M-MLV enzyme (ThermoFisher Scientific). qPCR, for evaluating eASC differentiation, was performed using MSC PCR array (Qiagen). Analysis of cytokine expression was done using horse cytokines and chemokines RT² profiler PCR array (Qiagen). Other qPCR analyzes were done with Taqman gene expression assay (Life Technologies, Courtaboeuf). PCR reactions were carried out on 20 ng of cDNA samples according to supplier’s recommendations on a 7500 FAST real-time PCR system (Applied Biosystems) and analyzed with the dedicated software. All values were normalized to GAPDH housekeeping gene and expressed as relative expression or fold change using the respective formulae 2^−ΔCT or 2^−ΔΔCt.

Equine peripheral blood mononuclear cells (PBMC) were collected from adult allogeneic horses into EDTA-containing tubes via jugular venipuncture. Mononuclear cells were isolated from peripheral blood by Ficoll density gradient centrifugation (density 1.077 g/L; Sigma) and suspended in IMDM GlutaMAX™ medium (ThermoFisher Scientific) supplemented with 10% inactivated FCS, 2 mM glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, 0.1 mM non-essential amino acids, 1 mM sodium pyruvate, 20 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid), and 5 × 10⁻⁵M 2-mercaptoethanol. eASCs were plated in 96-well flat-bottom plates at different densities (10⁵, 10⁴, 10³, or 10² cells/well) ± 100 ng/mL equine recombinant INFγ (R&D Systems) at 37°C for 24 h with 5% CO₂. PBMCs were added at 10⁵ cells/100 μL/well, and proliferation was induced by adding the mitogen phytohemaglutinin (PHA) at 5 μg/mL (Sigma-Aldrich). Unstimulated PBMCs were used as negative control. After 4 days incubation, cultures were pulsed with 5 μCi/mL ³H thymidine (Amersham, Buckinghamshire, UK) for 18 h. PBMCs were harvested and thymidine incorporation was expressed as proliferation in counts per minute. The inhibitory effect of eASCs on PBMC proliferation was quantified by subtracting the signal for PHA-stimulated PBMCs from unstimulated PBMCs. Proliferation rate was calculated referring to 100% the value of PHA-stimulated PBMCs.

Fetlock joints were recovered from the local slaughter house. The joints were aseptically opened with a scalpel in a dissection room to allow access to articular cartilage. Joints were graduated given the OARSI grade (32), and grade 1 joints were selected. A biopsy punch was used to extract 3 mm³ explants (average weight of 10 mg) from cartilage. Five explants per well in 24 multi-well plates were cultured in 1 mL of explant culture medium (DMEM, 20 mM Hepes, 5 μg/mL ascorbic acid, ITS, 1% Gentamycin) and incubated at 37°C, in 5% CO₂ humidified atmosphere. After 2 days, recombinant equine IL-1β (1 ng/mL; R&D Systems, Lille) and recombinant human oncostatin M (10 ng/mL; Invitrogen, Courtaboeuf) were added in the culture medium, and explants were cultured with or without 25,000 or 100,000 eASCs on a transwell membrane (pore 0.4 μm) during 7, 9, or 12 days. eASCs were primed or not with equine IFNγ (100 ng/mL) for 24 h before addition to the explant cultures. Media were changed every 2–3 days and aliquots of supernatants were stored at −20°C until GAG quantification.

Glycosaminoglycan quantification was performed using Blyscan™ GAG Kit on triplicate wells following supplier’s recommendations (Tebu-Bio, Le Perray en Yvelines).

Collagenase-induced OA was induced as previously described (18). Animal experimentation was conducted in agreement with the Languedoc-Roussillon Regional Ethics Committee on Animal Experimentation (approval CEEA-LR-10041). All surgery was performed under isoflurane gas anesthesia, and all efforts were made to minimize suffering. Briefly, right knee joints of mice were injected with 1 U type VII collagenase from Clostridium histolyticum (Sigma-Aldrich) in 5 μL of saline at day 0 and day 2, causing disruption of the ligaments and local instability of the joint. Groups of 10 mice received at day 7, either saline, mMSC (2 × 10⁵ cells/8 μL), or eASCs [2 × 10⁴ or 2 × 10⁵ cells/8 μL primed or not with IFNγ (100 ng/mL for 24 h)]. Mice were sacrificed at day 42, and the joints were collected, fixed in 4% formalin for 1 week, decalcified in 5% formic acid for 2 weeks, and embedded in paraffin. Knee frontal sections (three sections separated from 140 μm) were cut and stained with Safranin O–Fast Green staining. Histological OA score was determined using the modified Pritzker OARSI score (32). At euthanasia, blood was collected by intracardiac puncture, centrifuged at 1000g during 10 min. Sera were recovered and stored at −20°C.

Quantification of murine IL1β, IL6, IL10, osteoprotegerin (OPG) and prostaglandin E2 (PGE2) (R&D Systems), CTX2 (ImmunoDiagnostic Systems, Pouilly en Auxois), and cartilage oligomeric matrix protein (COMP) (MD Bioproducts, Zürich) was performed in sera by specific enzyme-linked immunosorbent assays (ELISA) as recommended by the suppliers.

Data were expressed as the mean ± SEM. Statistical analysis was performed with GraphPad Prism software. The comparison between the different groups was analyzed by one-way analysis of variance (ANOVA) followed by Dunnett post hoc test, when data were parametric, or by a Kruskal–Wallis test, when the data distribution was not Gaussian. Comparison between two groups was analyzed with a Mann–Whiney test for non-parametric data. The test used was indicated in the figure legends. A p value <0.05 was considered significant.

Equine ASCs isolated from subcutaneous AT exhibited a classic fibroblastic morphology at passage 1 (Figure 1A). The majority of cells were positive for CD29, CD44, CD90, and ELA class I markers and negative for ELA class II as detected by flow cytometry (Figure 1B). In absence of available antibodies, expression of ASC markers CD34, CD73, and CD105 was confirmed at the mRNA level, while CD45 hematopoietic and von Willebrand factor (vWF) endothelial markers were not detected (Figure 1C). Under adipogenic conditions, eASCs stored triglycerides in lipid droplets as shown by HCS LipidTOX™ Green neutral lipid staining and the expression of the transcription factor regulating adipogenesis PPARγ, which was significantly increased after 21 days of differentiation compared to day 0 (Figure 1D). eASCs differentiated into osteoblasts, as shown by Alizarin Red S staining and increased Runx2 expression at day 21 (Figure 1D). They also differentiated toward chondrocytes, as suggested by the presence of sulfated GAG evidenced by Safranin O staining and demonstrated by an enhanced expression of Sox9, Aggrecan, and Collagen type II markers (Figure 1D). These results indicated that eASCs possess all the characteristics of mesenchymal stem cells (33).

With the rationale in mind that priming eASCs will improve their immunosuppressive properties, we evaluated their capacities to inhibit the proliferation of ePBMCs. eASCs were pretreated or not with IFNγ for 24 h and then cultured with PHA-activated ePBMCs at different ratios for 3 days. When compared with ePBMCs alone, both eASCs- and IFNγ-primed eASCs were able to inhibit T cell proliferation in a dose-dependent manner (Figure 2A). The antiproliferative effect of IFNγ-primed eASCs was, however, significantly higher, whatever the eASC/ePBMC ratio. In order to determine whether this higher suppressive effect of IFNγ-primed eASCs was related to higher paracrine functions, expression of several known immunosuppressive mediators was quantified by RT-qPCR. We noticed that TGFβ1 expression levels were decreased in IFNγ-primed eASCs, while IL6 expression was increased (Figure 2B). No change in expression of COX-2 transcripts or PGE2 secretion was observed (Figures 2B,C).

In parallel, we investigated whether eASCs could exert chondroprotective properties and inhibit cartilage degradation in an in vitro model of OA cartilage. We used equine cartilage biopsies cultured with IL1β and Oncostatin M for 12 days to induce cartilage degradation. Cartilage destruction was evaluated by quantifying GAG release in the culture medium. In order to determine the effect of eASCs on cartilage degradation, a coculture assay was designed using a transwell system to avoid eASCs–cartilage contact. Cartilage explants (in the lower part) were cultured with 2.5 × 10⁴ or 10⁵ naïve or IFNγ-primed eASCs (upper part). In these conditions, naïve eASCs did not exert a protective effect on cartilage degradation, even though a significant decrease of GAG release was observed with the dose of 10⁵ eASCs after 9 days of coculture (Figure 2D). By contrast, the two doses of IFNγ-primed eASCs significantly reduced GAG release at the three time points. Interestingly, IFNγ-primed eASCs tended to be more efficient at the lowest dose as compared with the highest dose, but the difference was not significant. Altogether, our data demonstrated that IFNγ-priming of eASCs enhanced their immunosuppressive and chondroprotective function.

We, then, investigated whether eASCs could exert a therapeutic effect in vivo using a xenogeneic relevant model of OA. We used the CIOA mouse model which is characterized by moderate inflammation of the synovial membrane, osteophyte formation, and cartilage degradation (18). First, we evaluated the effect of intra-articular injection of eASCs (2 × 10⁵ cells) on knee joint OA score by comparison with mMSCs (2 × 10⁵ cells). Histological analysis of knee joint sections revealed protection against cartilage degradation in animals treated either with mMSCs or eASCs as compared with the CIOA control group (Figure 3A). Cartilage protective effect of mMSCs and eASCs was confirmed by scoring both femoral condyles and tibial plateaux in the lateral and medial compartments (Figure 3B). Significantly, lower mean OA scores were obtained after eASCs or mMSCs treatment, and OA score of the eASC group was highly significant compared to control group. These results confirmed that the CIOA mouse model was effective for evaluation of the effect of xenogeneic eASC injection on OA symptoms.

We, therefore, compared the effect of naïve and IFNγ-primed eASCs on OA progression. Injection of a high dose of eASCs (2 × 10⁵ cells) significantly improved the OA score, while the low dose (2 × 10⁴ eASCs) had no effect (Figure 4A). By contrast, both doses of IFNγ-primed eASCs reduced the OA score, but only the lowest dose of IFNγ-primed eASCs induced a significant decrease. Of importance, 2 × 10⁴ IFNγ-primed eASCs were as efficient as 2 × 10⁵ naïve eASCs to protect cartilage from degradation. The therapeutic effect of eASCs was confirmed by measuring the level of OPG, a bone metabolism marker, in the sera of mice at euthanasia. It was significantly decreased in mice treated with 2 × 10⁴ IFNγ-primed eASCs (Figure 4B). Levels of COMP, a marker of cartilage turnover, did not change in the sera of treated mice compared with control group (Figure 4C), and IL6, IL10, IL1β, and C-terminal cross-linked telopeptide of type II collagen (CTX2) were not detected in the sera of mice (data not shown).

Finally, we evaluated whether naïve or IFNγ-primed eASCs could inhibit murine T cell proliferation since the CIOA mouse model is known to be associated with local moderate inflammation. We, therefore, tested naïve and IFNγ-primed eASCs on murine splenocytes in a proliferative assay. Both eASCs significantly decreased splenocyte proliferation in a dose-dependent manner, and this inhibitory effect was higher with IFNγ-primed eASCs at the lowest ratios of eASCs/splenocytes (Figure 4D). These data suggested that the effect of eASCs in vivo could be mediated at least in part by their inhibitory role on inflammatory cells. Overall, we present evidence that eASCs were efficient to reduce the clinical score of OA in the xenogeneic CIOA model, and IFNγ-pretreatment increased their immunosuppressive and chondroprotective functions.

In order to understand the effect of IFNγ priming on the immunosuppressive properties of eASCs, a gene expression analysis was performed on 89 genes related to cytokine or chemokine families. Hierarchical clustering analysis revealed that the gene expression pattern of IFNγ-primed eASCs greatly differed from that of naïve eASCs (Figure 5A). IFNγ priming modified the transcriptomic program of eASCs by significantly dysregulating 13 genes. Gene expression of CXCL9, CXCL11-like, IL32-like, and CXCL10 was significantly induced in IFNγ-primed eASCs compared with naïve eASCs (Figure 5B). In parallel, gene expression of CSF1, IL6, IL7-like, IL-15, CCL2, CCL13, and TNFS13B was significantly upregulated in IFNγ-primed eASCs, while expression of TNFSF11-like and TGFβ2-like was downregulated as compared with naïve eASCs (Figure 5C). Indeed, the cytokine/chemokine profile of eASCs was dramatically altered by IFNγ priming.