Mesenchymal stem cell senescence alleviates their intrinsic and seno-suppressive paracrine properties contributing to osteoarthritis development

Expression of the senescence Cdkn2a-encoding product p16^INK4a is a hallmark of experimental collagenase-induced OA and is partially required for cartilage degradation

First, we wanted to monitor the appearance of senescent cells after OA induction in the collagenase-induced OA (CIOA) model [14], which mimics joint inflammation and synovitis that are observed in 1/3 of patients with OA [5]. To this aim, we performed intra-articular injection (at day 0 and day 2) of collagenase type VII in the left knee and saline solution in the right knee of 2-month-old C57BL/6JRj male mice, as previously described [15], and collected joints at day 14, 28 and 42 post-injection. Analysis of cartilage degradation (OA score) and synovitis, showed progressive cartilage degradation and early synovial activation (Figure 1A and 1B) in the collagenase-injected joint, but not the NaCl control ones.

Figure 1. p16^INK4a is involved in experimental collagen-induced osteoarthritis. Osteoarthritis (OA) was induced by collagenase intra-articular injection in the left knee (NaCl injection in the right knee for control) of 2-month-old C57BL/6JRj male mice. (A) Representative images of OA kinetic development after intra-articular collagenase injection showing synovial inflammation and osteophytosis (top panel) and focus on cartilage degradation (bottom panel). (B) Synovial inflammation quantification (synovitis semi-quantitative score; from 0 to 3) and cartilage degradation score (OA modified score according to van den Berg; from 0 to 30) were analyzed at day 14, 28 and 42 post-injection and compared with NaCl control at day 42. Data are the mean ± SEM (n=8), *=p<0.05, ***=p<0.001, ****=p<0.0001. (C) p16^INK4a, IL-1β, IL-6 and MMP-13 mRNA expression levels in the synovial membrane after NaCl or collagenase injection, measured by RT-qPCR. Results were expressed as fold change compared with NaCl control at day 42. Graphs represent the mean ± SEM (n=8); *=p<0.05, **=p<0.01, ***=p<0.001. (D) Experimental design of p16^INK4A expression analysis in Cdkn2a^+/luc after OA induction. (E) Luminescence analysis in both knees with a CDD camera after intra-peritoneal and intra-articular Cyc-Luc injection. Values for the left knee (collagenase injection) were expressed as fold change relative to the right knee (control). Data are the mean ± SEM (day 14, n=13; day 24, n=6; day 35, n=6; day 42 n=8); *=p<0.05. (F) Representative image of luciferase signal in the left (CIOA) and right (NaCl) knee at day 24. (G) Experimental design of OA induction in Cdkn2a^+/luc and Cdkn2a^luc/luc mice. (H) Cartilage degradation score at day 42 after NaCl (control) or collagenase (CIOA) injection in 2-month-old Cdkn2a^+/luc and Cdkn2a^luc/luc mice. Data the mean ± SEM (n=8 and 11 respectively); *=p<0.05, ***=p<0.001.

Concomitantly, in the synovial tissue, we observed transient gene expression at day 14 of the senescence marker p16^INK4a, but also with expression of the SASP factors (IL1-β, IL-6 and MMP13) (Figure 1C). To confirm p16^INK4a expression during experimental OA, we used 2-month-old Cdkn2a^+/luc mice in which the promoter of gene encoding p16^INK4a drives the expression of the reporter gene luciferase [16]. At day 14, 24, 35 and 42 after the first injection of collagenase type VII in the left knee and saline in the right knee (as before), we injected Cyc-Luc^R intra-peritoneally and intra-articularly, and then determined the luminescence signal intensity using a CDD camera (Figure 1D). Comparison of the two knees in each mouse showed a significant and transient peak of luciferase activity at day 24 in the OA joint following its mRNA induction at day 14 and reflecting the presence of p16^INK4a-positive senescent cells (Figure 1E and 1F). We next asked whether p16^INK4a was required for cartilage and joint alteration following OA onset. To this aim, we induced CIOA in heterozygous Cdkn2a^+/luc mice that retains p16^INK4a protein expression and in homozygous Cdkn2a^luc/luc mice lacking p16^INK4a expression [16]. At day 42 post-injection, histological analysis of paraffin-embedded knee tissue sections stained with Safranin-O/Fast Green (Figure 1G, 1H and Supplementary Figure 1) showed that the mean (± SEM) OA score in collagenase-injected knees was 15.78 ± 5.8 in Cdkn2a^+/luc mice and 10.24 ± 5.8 in Cdkn2a^luc/luc mice (p<0.05). This significant difference in OA score demonstrates that p16^INK4a ablation partially protects against OA-driven cartilage degradation. Micro-CT analysis of histo-morphometric parameters did not reveal any difference between Cdkn2a^+/luc and Cdkn2a^luc/luc mice in term of collagenase-induced subchondral bone remodeling (bone volume, bone surface, bone surface/bone volume ratio and subchondral bone height) (data not shown). This finding and the recent demonstration that specific somatic Cdkn2a gene inactivation in joint chondrocytes has no impact on OA onset [17] suggest that other joint cell types acquire a deleterious p16^INK4a-driven senescence phenotype during disease development.

Senescent p16^INK4a-positive MSCs show impaired self-renewal and in vitro cartilage formation capacities altogether with specific secretory profile

Cartilage homeostasis relies primarily on the cartilage self-repair mechanisms and on MSCs found mainly in the bone marrow of sub-chondral bones and in synovial tissue [18, 19]. MSCs contribute to cartilage homeostasis through their self-renewal capacities and chondrogenic differentiation into neocartilage [20, 21]. However, MSCs might also contribute to OA onset because increased TGF-β signaling in MSCs is sufficient to induce cartilage breakdown in a transgenic mouse model [18]. Therefore, we first asked in vitro whether MSCs senescence could be implicated in OA pathogeny.

To this aim, we incubated human bone marrow-derived MSCs isolated from healthy donors with etoposide, a senescence-promoting DNA-damage inducer, for 7 hours and then cultured them for 2 weeks. Following this acute treatment, MSCs showed an increase in the percentage of SA β-galactosidase-positive cells compared to untreated MSCs (64.4% ± 12.2 and 9.6% ± 3.2, respectively, p<0.0001) (Figure 2A and 2B). Proliferative rate (Figure 2C), BrdU incorporation (Figure 2D), and colony formation unit capacity (Figure 2E) were reduced in treated MSCs compared with control. Moreover, the mRNA and protein expression levels of the two main senescence markers (p16^INK4a and p21^cdkn1a) were increased (Figure 2F and 2G) confirming senescence onset. We next used TGF β3-driven chondrogenic protocol to in vitro differentiate MSCs into cartilage. We showed that senescence induction alleviates MSC capacities to form cartilage in micro-mass setting as revealed by phase contrast imaging (Figure 2H). Finally, we performed a differential protein array analysis of senescent or proliferating MSCs allowing us to determine SASP protein expression. Among the 105 inflammation-related proteins tested, the comparative analysis identified 13 candidates that were common to three MSC isolated from healthy donors. We thus found DKK1 (a WNT antagonist factor known to trigger OA when overexpressed in a mouse model [22]), CHI3L1/YKL-40 and IGFBP-3 (two OA synovial fluid biomarkers that correlate with disease progression [23, 24]), and IL-8 and CCL2/MCP1 (two inflammatory cytokines that trigger in vitro senescence onset and that are usual component of the SASP [25, 26]) (Figure 2I). This profile specific of senescent MSCs by its composition could affect thus joint homeostasis by production of promoting OA and senescence spreading factors. Altogether, our findings demonstrate that following 7 hours of etoposide treatment, MSCs undergo senescence, lose their stemness properties and produce specific factors.

Figure 2. Senescence modulates MSCs intrinsic properties. (A) Beta-galactosidase staining in human MSCs at day 14 after DNA damaged-induced senescence (Senescent) or not (Proliferative). Data are the mean ± SEM (n=5); ****=p<0.0001. (B) Representative images of beta-galactosidase staining in proliferative and senescent human MSCs. (C) Proliferation rate (mean ± SEM) in proliferative and senescent human MSCs (n=6); ****=p<0.0001. (D) BrdU incorporation in proliferative and senescent human MSCs. BrdU-positive cells relative to all DAPI-positive cells were counted using an optical microscope (mean ± SEM; n=5). ****=p<0.0001. (E) Colony forming units in proliferative and senescent human MSCs (mean ± SEM; n=3); *=p<0.05. (F) p16^INK4a and p21^cdkn1a mRNA expression in human MSCs at day 14 after DNA damaged-induced senescence (Senescent) or not (Proliferative) by RT-qPCR. Data are the fold change relative to proliferative cells (mean ± SEM; n=4 for each condition); *=p<0.05. (G) p16^INK4a, p21^cdkn1a and p27^KIP1 protein expression in human MSCs at day 14 after DNA damaged-induced senescence (Senescent) or not (Proliferative) by western blotting. Representative images of MSCs from n=3 independent donors. (H) Representative images of one cartilage pellet after chondrogenesis induction in proliferating and senescent human MSCs (from n=3). (I) Protein expression profiles of total cell extracts from senescent (bottom) and proliferating (top) human MSCs from three different healthy donors. The table showed the 13 proteins that were overexpressed in all three senescent MSC samples.

Senescence alleviates in vitro MSCs seno-suppressive paracrine functions.

To maintain articular homeostasis following cartilage injury, MSC rely also on their paracrine chondroprotective properties toward functionally-altered chondrocytes. For instance, using trans-well co-culture systems, previous studies have shown that MSCs can reduce the expression of some hypertrophy and fibrotic markers in OA chondrocytes [8]. To determine whether senescence could change this central MSC tissue support function, we co-cultured for 7 days, using the same trans-well setting, OA chondrocytes with proliferative (control) or senescent p16^INK4a-positive MSCs (Figure 3A). As previously published, control MSCs significantly reduced, in OA chondrocytes, the expression of TGF-β1, ADAMTS3 and ADAMTS5 three hypertrophic markers associated with the SASP as well as the fibrotic marker collagen 3 (Col3) (Figure 3B). Remarkably, although the secretory profile of these MSCs neither promote the re-expression of aggrecan (Figure 3B) nor collagen type 2 expression in OA chondrocytes (data not shown), it can nevertheless reduce the expression for senescence-associated cell cycle inhibitors namely p16^INK4a, p15^INK4b and p27^KIP1 (Figure 3B). Our results emphases then a chondro-protective, anti-fibrotic but also an anti-senescent paracrine effect for these stem cells towards OA chondrocytes. However, when senescent, MSCs retain only their anti-fibrotic properties by repressing Col3 but loss their suppressive capacities on p16^INK4a, p15^INK4b p27^KIP1 and the SASP factors TGF-β1 and ADAMTS3 in OA chondrocytes (Figure 3B). Thus, MSC seno-suppressive paracrine effect is strongly affected upon senescence onset arguing here that senescent MSC maintains and/or participates to chondrocytes loss of function during OA progression or cartilage aging.

Figure 3. Senescence modulates MSCs extrinsic properties in vitro. (A) Experimental design of the without-contact co-culture system to assess the effect of senescent human MSCs on chondrocytes from patients with OA. (B) Expression analysis by RT-qPCR in OA chondrocytes without co-culture (black columns; control), or co-cultured with proliferating MSCs (grey columns), or with senescent MSCs (white columns) for 7 days. Data are expressed as fold change relative to control (mean ± SEM of n=5); *=p<0.05, **=p<0.01.

Senescence-accelerated SAMP8 mice develop spontaneous early osteoarthritis

To demonstrate the causal role for senescent MSCs in OA pathogeny in vivo, we rely on the senescence accelerated mouse-prone (SAMP8) mouse. This strain harbors mutations in the mitochondrial genome causing an progressive increase in systemic oxidative stress, high level of cell senescence in epididymal adipose tissue [27], retinal cells [28], aorta [29], astrocytes [30], or cochlea [31], and premature aging phenotypes [27–30]. We first verify that SAMP8 prematurely develop OA. To this aim, we analyzed by micro-CT, bone and cartilage structures in 8-month-old mice revealing a significant higher bone surface/bone volume ratio and a significant lower subchondral bone volume on the lateral femoral subchondral bone area of SAMP8 compared with SAMR1 control mice (Figure 4A and 4B). This finding indicates early subchondral bone structure remodeling and degradation as found in OA patients. Ectopic ligamental and meniscal knee calcifications were also present in 50% of the analyzed SAMP8 animals (Figure 4A and 4C). Furthermore, the knee OA degradation score was significant higher in 11-month-old SAMP8 than SAMR1 control animals (Figure 4D and 4G), and was associated with higher synovitis (Figure 4E) and osteophytosis scores (Figure 4F). Altogether, this indicates that senescence-prone SAMP8 mice develop spontaneous joint OA signs that resemble the human disease.

Figure 4. SAMP8 mice display a spontaneous OA phenotype. (A) Histo-morphometric analysis by micro-CT of the left knee in SAMR1 and SAMP8 mice. Graphs represent the mean ± SEM (n=8 for SAMR1, n=5 for SAMP8); **=p<0.01, ***=p<0.001. (B) Representative micro-CT images showing higher sub-chondral bone modification and (C) ligament calcifications in SAMP8 mice compared with SAMR1 mice. Knees from SAMR1 and SAMP8 mice were stained with Safranin-O/Fast Green to quantify: (D) spontaneous cartilage degradation (OA modified score according to van den Berg, from 0 to 30), (E) spontaneous synovial membrane inflammation (synovitis semi-quantitative score, from 0 to 3), and (F) osteophytes (osteophyte semi-quantitative score, from 0 to 3). Data are the mean ± SEM (n=5 for each condition). **=p<0.01. (G) Representative images of the spontaneous OA phenotype in SAMP8 mice with cartilage degradation compared with SAMR1 mice.

Intra-articular injection of senescent MSCs isolated from SAMP8 mice is sufficient to induce cartilage degeneration in young wild-type mice

We next wanted to link premature OA development in these SAMP8 mice and the presence of senescent MSCs. To this aim, we isolated enriched mesenchymal stromal/stem population from 6-months SAMP8 and SAMR1 mice after bone marrow flushing, and 2 weeks of selection for cell culture adhesion as previously published [32]. Several cell cycle inhibitors associated with senescence state such as p16^INK4a, p19^ARF and p21^Cdkn1a, and some SASP factors namely TGF-β1 and MMP-13, were upregulated in SAMP8 MSCs compared with SAMR1 MSCs (Figure 5A) [18, 33, 34]. As expected for senescent cells, these SAMP8 MSCs showed also an increase in SA-β-galactosidase activity (Figure 5B) but a surprisingly non-classical inflammatory SASP as demonstrated by absence in IL-6 and IL-1β expression levels (Figure 5A). Then, to challenge our hypothesis whether senescent MSCs can contribute to OA development, we intra-articulary delivered 2.5x10⁵ MSCs from SAMP8 or SAMR1 mice in 2-month-old C57BL/6JRj male mice at day 0. At day 42 post-injection, each injected mouse was sacrificed and dissected joints analyzed as in Figure 4 (Figure 5C). Histological analysis after Safranin-O/Fast green staining showed that injection of SAMP8 MSCs strongly promoted cartilage degradation compared with SAMR1 MSCs (mean OA score = 12.2 ± 0.7 and 6.1 ± 1.6, respectively; p<0.05) (Figure 5D and 5E). Furthermore, subchondral bone analysis by micro-CT revealed, even if not significant, a trend toward a higher bone surface/bone volume ratio and a lower subchondral bone volume in the medial compartment in SAMP8 MSCs injected mice (Figure 5F). These results demonstrate that senescent MSCs are sufficient to trigger alone cartilage and joint dysfunction in young wild-type mice, similar to the phenotypes observed in senescence-accelerated SAMP8 mice and in human patients.

Figure 5. Intra-articular injection of senescent MSCs induces OA-like cartilage degradation. (A) p16^INK4A, p19^ARF, p21^cdkn1a, TGF-β, MMP-13, IL-1β and IL-6 mRNA expression in MSCs derived from bone marrow (BM) of 6-month-old SAMP8 or SAMR1 mice by RT-qPCR. Data are the mean ± SEM (n=3 for each conditions); *=p<0.05. (B) Beta-galactosidase staining in MSCs derived from BM of SAMP8 or SAMR1 mice. Data are the mean ± SEM (n=5); ****=p<0.0001. (C) Experimental design of BM-MSC injection in the knee of 2-month-old C57BL/6JRj mice. (D) Cartilage degradation (OA modified score according to van den Berg, from 0 to 30, after Safranin-O/Fast Green staining) at day 42 after SAMP8 or SAMR1 BM-MSC intra-articular injection. Data are the mean ± SEM (n = 5 for both conditions); *=p<0.05. (E) Representative images of cartilage degradation in a C57BL/6JRj mouse after SAMP8 or SAMR-1 BM-MSC injection. (F) Histo-morphometric analyses by micro-CT of the left knee (medial and lateral compartment) at day 42 after injection of SAMR1 or SAMP8 BM-MSCs. Data are the mean ± SEM (n = 5 for each condition).

Animal experiments

6-weeks-old C57BL/6JCj male mice were obtained from Janvier Laboratory. SAMP8 and SAMR1 mice were bought from Envigo laboratory. Transgenic Cdkn2a^+/luc and Cdkn2a^luc/luc mice were obtained from K.B. Sharpless’s laboratory. Animal experiments were performed in accordance with the guidelines by the local ethics committee on animal research and care (approval CEEA-LR-10042). After sacrifice, soft tissues and muscles around the knee were dissected and removed. Patella was removed and the supra-patellar synovium tissue was isolated and stored in 200μL Trizol at -80°C. The knee joint was separated, fixed in 4% paraformaldehyde at 4°C for 2 days, and decalcified by incubation at +4°C in a large volume of 14% EDTA for 21 days. The EDTA solution was changed every 3 days. Samples were then stored in PBS at +4°C.

Experimental osteoarthritis and intra-articular injections

Experimental OA was induced in 6-weeks-old male mice according to the CIOA procedure. Under general anesthesia (isoflurane inhalation), the knee area was disinfected with ethanol and a small cut (a few millimeters) was performed in the cutaneous and subcutaneous tissue to visualize and to access the knee. Then, each mouse received an intra-articular injection of 5μL (1IU) collagenase VII in the left knee to induce OA and 0.9% NaCl in the right knee as control, followed by a second injection two days later. Intra-articular injections of BM-MSCs were performed in 6-weeks-old male mice following the same procedure: 8μL of PBS containing 2.5x10⁵ BM-MSCs isolated from SAMP8 or from SAMR1 mice in the left knee at day 0.

Bioluminescence imaging

For luciferase signal detection, mice were anaesthetized before intraperitoneal (100μL) and intraarticular (10μL) injection of the luciferase substrate Cyc-Luc (300μg/mL) (Merck). A few minutes later, mice were placed in the dark chamber and a grey-scale image was first recorded with dimmed light followed by acquisition of luminescence images using a cooled charged-coupled device (CCD) camera (PIXIS 1024B; Princeton Instruments).

Bone parameter analyses and Safranin-O/Fast Green staining for cartilage analysis

Hind leg paws were dissected to remove smooth tissues and scanned in a micro-CT scanner SkyScan 1176 (Bruker, Belgium, 0.5 mm aluminum filter, 45 kV, 500 μA, 18 μm resolution, 0.5° rotation angle). Scans were reconstructed using CTAn v1.9, Nrecon v1.6 (Bruker, Belgium) and a three-dimensional (3D) model visualization software program (CTVol v2.0). Misalignment compensation, ring artifacts and beam-hardening were adjusted to obtain the correct reconstruction of each paw. Bone degradation was quantified in subchondral bone and the epiphysis region of the medial and/or lateral plateau for each tibia (CTAn software, Bruker, Belgium). Meniscal and ligament calcifications were quantified on the entire knee joint. Reconstructed 3D images of joints were obtained using the Avizo software (Avizo Lite 9.3.0, FEI, France). For cartilage analysis, mouse knees were fixed in 4% paraformaldehyde at 4°C for 48h, washed in PBS, and then processed for routine histology. Knees were decalcified in 14% EDTA/PBS for three weeks and then paraffin-embedded. Tissue sections (5 μm) were rehydrated through a gradient of ethanol and xylene. Sections were then stained with Safranin-O/Fast Green solution to evaluate cartilage degradation, synovial inflammation and osteophytosis. Cartilage damage was analyzed using an arbitrary score of 0–30, based on the OARSI cartilage OA grading system histopathology [59], and modified by van den Berg for the assessment of murine knee joints (grading scale of 0–6 for the severity of cartilage destruction and of 0–5 for the extent of damaged cartilage surface) [14, 60]. A semi-quantitative scoring system was used to analyze synovitis, with scores ranged from 0, indicating no thickening of the lining layer (1 cell layer), to 3, indicating the maximal observed thickening of the synovial lining layer [14, 60]. The same semi-quantitative scoring scale, from 0 to 3, was used for osteophytosis, with 0 = no osteophytosis and 3 = the largest osteophytosis observed [61, 62].

Cell types and culture conditions

Human MSC cultures were established from bone marrow aspirates of healthy donors after signature of the informed written consent and approval by the local and national ethics committees (committee of Hannover Medical school with votum No. 2562). MSCs were isolated and amplified in complete alpha-minimum essential medium (Ozyme; BE12-169F-12) supplemented with 10% fetal bovine serum (FBS), 1% penicillin/streptomycin, 2 mM L-glutamine, and 1 ng/mL of basic fibroblast growth factor-2 (Miltenyi Biotec; GER 130-104-924). For DNA damage-induced senescence, MSCs were incubated with 12.5 μM etoposide (E1383 Sigma, USA) for 7h. Senescence levels were evaluated after 14 days (medium changed every 3 days). Human OA chondrocytes were isolated from cartilage of patients with OA undergoing knee arthroplasty after signature of the informed written consent and approval by the national ethics committee (‘Cellule de bioéthique de la direction générale pour la recherche et innovation, Ministère de l’Enseignement Supérieur et de la Recherche’; registration number DC-2009-1052). OA chondrocytes were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal calf serum (Sigma, USA), 1% penicillin/streptomycin (Lonza, CH), and 2 mM L-glutamine (Lonza, CH).

Bone marrow mesenchymal stem cell isolation

Bone-marrow MSCs (BM-MSC) were isolated from the right leg of SAMP8 and SAMR1 mice. Femur and tibia were dissected, and diaphyses were flushed with PBS. Cell suspensions were filtered and plated at a concentration of 1x10⁶ cells/cm² in DMEM medium supplemented with 10% FBS (Hyclone, Thermo Fisher Scientific), 2mM glutamine, 100U/mL penicillin, 100mg/mL streptomycin (Lonza, Levallois-Perret, France) and 2ng/mL human basic fibroblast growth factor (R&D Systems, Lille, France). Medium was changed daily to remove red blood cells and non-adherent cells. BM-MSCs were stored at -80°C after passage 1.

Co-culture experiments

Co-culture experiments were performed as previously described [8]. Briefly, 5x10⁵ OA chondrocytes/well were plated on the bottom of 6-well plates and cultured with MSCs (7x10⁴ cells/insert) in cell culture inserts (PET membranes, 0.4 μm pores, BD Biosciences, UK). Cultures were maintained in 3 mL of minimal medium (DMEM supplemented with 100U/mL penicillin and 100mg/mL streptomycin, 0.35 mmol/L proline, 0.17 mmol/L ascorbic acid, and 1 mmol/L sodium pyruvate) for 7 days when chondrocytes were collected for RT-qPCR analysis.

Gene expression analysis by real-time quantitative PCR

For qPCR experiments, total RNA was extracted from cells using the RNeasy® Mini kit (Qiagen; 74106) according to the manufacturer's instructions. RNA quality was checked by spectral analysis (A260/ 280 nm), and then samples were stored at −80°C. Reverse transcription was performed using the M-MLV reverse transcriptase (Invitrogen; 28025013; 5U/μL final concentration), 500ng total RNA, a random hexamer primer (Thermo Scientific, GER; S0142; 10ng/μL final concentration), and DNTPs (Roche, CH; 1 277 049; 5mM final concentration) in M-MLV reverse transcriptase buffer (Invitrogen; 18057-018), for a total volume of 20μL. SYBR Green-based quantitative PCR was performed using the LightCycle® 480 SYBR Green I master reaction mix (Roche, 04707516001), 10 ng of cDNA, and the LightCycler 480 real-time PCR system (Roche) (40 cycles of amplification). Raw data (Ct values) were analyzed using the comparative Ct method. Gene expression data were calculated as relative to the expression of housekeeping genes. The comparative threshold cycle method (ΔΔCT) was used to quantify relative gene expression, and the obtained quantification was transformed to exponential 2−ΔΔCT values. A p-value below 0.05 (Student’s t-test) was considered as statistically significant. Primers sequences used for RT-qPCR are listed in Supplementary Table 1.

Proliferation, Colony Forming Unit (CFU) assays and BrdU incorporation

For CFU assays, 3x10³ MSCs were homogeneously plated in 60mm dishes. After 10 days, when cell colonies emerged, cultures were fixed with methanol and stained with Giemsa dye solution diluted 20X for counting. For BrdU incorporation analysis, 12x10³ MSCs were plated on coverslips 14 days after etoposide treatment or not. Cells were incubated with 10μM of BrdU in fresh culture medium for 21 hours, and then fixed with 3.7% paraformaldehyde solution at room temperature (RT) for 15min. Cells were permeabilized by adding 0.3% Triton X-100 in PBS at RT for 30min. After Triton X-100 removal, 500μL of 2N HCl solution was added at RT for 10min, followed by incubation with 10% donkey serum in PBS at RT for 30 min. Then, cells were incubated with rat anti-BrdU IgG (1:250 in 3% BSA/PBS, Abcam; ab6326) at 37°C for 2 hours, followed by 488 Alexa Fluor® conjugated goat anti-rat IgG (1:500 in PBS, Thermo Fisher Scientific; A-11006) at RT for 1 hour. Coverslips were mounted with Prolong Gold Antifade Reagent (Thermo Fisher Scientific; 11569306) containing 4′-6-diamidino-2-phenylindole (DAPI). Cells with nuclear BrdU signal were counted as positive among at least 100 DAPI-positive cells from 10 images of randomly chosen fields for each sample, using the ImageJ software.

In vitro chondrogenic differentiation

Chondrogenic differentiation of MSCs was induced by 21-day culture in micropellet, as explained previously [63]. Briefly, MSCs (2.5 × 10⁵ cells) were pelleted by centrifugation in 15 ml conical tubes and cultured in chondrogenic medium (DMEM supplemented with 0.1 μM dexamethasone, 0.17 mM ascorbic acid, 1% insulin-transferrin-selenic acid (ITS), 350 nM proline, 1 mM sodium pyruvate) supplemented with 20 ng/mL TGFβ-3.

Senescence-associated β-galactosidase activity

Induction of MSC senescence was assessed by staining MSCs with a senescence-associated β-galactosidase (SA-β-Gal) Staining Kit (Sigma-Aldrich; CS0030) according to the manufacturer's instructions. In detail, 12x103 MSCs were plated on coverslips and after 24h, they were fixed with the Kit fixation solution at RT for 8min, and then they were incubated with the SA-β-Gal staining solution overnight to reveal SA-β-Gal activity. Cells displaying blue signal in the cytosol were counted as positive, using the ImageJ software. Ten images of random fields for each sample were taken by microscopy, and at least 100 cells were counted for each sample.

Inflammation protein profile analysis and western blotting

The secretome of senescent and proliferating MSCs was analyzed with the Protome Profiler Human XL Cytokine Array Kit (R&D Systems; ARY022B) according to the manufacturer’s instructions for the parallel determination of the relative level of 105 human cytokines, chemokines, and acute phase protein. Capture antibodies were spotted in duplicate on nitrocellulose membranes to bind to specific target proteins present in the cell supernatant. Captured proteins were detected with biotinylated antibodies and visualized using chemiluminescence detection reagents. For western blot analysis, proteins were isolated from cultured MSCs using Lysis buffer 17 (R&D systems; 895943), supplemented with protease inhibitors (Roche; 11 873 580 001), according to the manufacturer's instructions. Protein concentration was estimated using the BCA Protein assay (Fisher Scientific; 10678484) and 50μg of total proteins was loaded per well after denaturation in Laemmli blue with beta-mercaptoethanol (at 95°C, for 7min). Samples were resolved through Mini-PROTEAN® TGX™ 4-15% gradient gels (Bio-Rad: 456-1084) in Tris-glycerin buffer, and then were transferred to polyvinylidene fluoride (PVDF) membranes (Bio-Rad Transblot® Turbo™: 170-4157) using the Bio-Rad Transblot® Turbo™ system (1.3A – 25V – 7 min). After transfer, membranes were washed in Tris-buffered saline (TBS), blocked in 5% non-fat dry milk in TBS-0.1% Tween 20 (TBST) at RT for 1 hour, followed by incubation with primary antibodies at 4°C overnight, except for the anti-β-actin antibody (RT for 3h). The following primary antibodies were used: rabbit IgG anti-human p16 (1:750 in 5% NFDM-TBST, Proteintech; 10883-1-AP), rabbit IgG anti-human p21 (1:1000 in 3% BSA-TBST, Abcam; 109199), mouse IgG anti-human p27 (1:200 in 5% NFDM-TBST, Abcam; 193379), rabbit anti human β-actin (1:1000 in 5% BSA-TBST, Abcam; ab8226). After incubation, membranes were washed in TBST and incubated with goat anti-rabbit IgG (1:2000 in 5% BSA-TBST; Cell Signaling, 7074) or goat anti-mouse IgG (1:2000 in 5% BSA-TBST, Cell Signaling; 7076) horseradish peroxidase (HRP)-conjugated at RT for 2h. Reactions were visualized using the Immobilon ECL HRP Substrate (Millipore; WBLS0500).

Statistical analysis

All data are presented as the median or mean ±SEM or SD. The Student’s t-test was used for comparisons between experimental groups, and ANOVA for multiple comparisons followed by a Sidak post-test. Data were analyzed using the Prism software v6 (GraphPad Software Inc.); p-values <0.05 were considered significant (*p <0.05; **p <0.01; ***p <0.001; ****p <0.0001).