Osteoarthritis (OA) is a multifaceted joint disease characterized by cartilage degradation, bone modifications and mild synovial inflammation. OA cartilage displays extracellular matrix and cell modifications including increased production of metalloproteases (MMPs) and aggrecanases (ADAMTS-4 and 5, A Disintegrin And Metalloprotease with ThromboSpondin-like repeat) and increased chondrocyte catabolic and hypertrophic differentiation [1
]. Factors associated with OA development include genetics, sex, metabolic syndrome, obesity and diabetes, but aging and mechanical overload are the two most prominent risks [2
]. However, the precise molecular mechanisms responsible for initiation or progression of OA remain to be elucidated.
Several reports have recently unveiled the importance of the primary cilium in cartilage physiology and pathology [5
]. Primary cilium located at the chondrocyte surface has the capacity to sense multiple stimuli including mechanical stress and inflammatory cytokine stimulation [5
]. Thus, stimulation with inflammatory interleukin 1β (IL-1β) increases the chondrocyte primary cilium length within minutes [10
]. Furthermore, mechanical stimulation promotes primary cilium-driven intracellular Ca2+
mobilization and Indian Hedgehog (Ihh) signaling [5
]. For instance, chondrocytes isolated from mice with mutant Polaris/intraflagellar transport (ITF) 88, a protein necessary for cilium formation, fail to increase intracellular Ca2+
under mechanical stimulation [11
]. Moreover, loss of primary cilia in chondrocytes reduces cartilage mechanical properties in Col2Cre/Ift88ft/ft
transgenic mice and promotes OA development characterized by increased expression of MMP13, ADAMTS5, COLX and RUNX2 [12
]. Similarly, mice mutant for Bardet-Biedl syndrome 1 (Bbs1−/−)
, a family of proteins involved in primary cilium formation/functions, develop OA-like cartilage abnormalities including proteoglycan loss, small surface fibrillations, reduced cartilage thickness and increased MMP13 expression [14
The expression of galectin 3 (GAL3), a 30-kDa member of the galectin family of lectins, is increased in human OA versus normal cartilages [16
]. GAL3 is multifunctional protein implicated in cellular interactions, cell differentiation, survival and death [18
]. The role of GAL3 in OA has not been completely elucidated: intracellular GAL3 has an anti-apoptotic effect whereas extracellular GAL3 stimulates ADAMTS-5 production [19
]. Importantly, recent work revealed the involvement of GAL3 in cilium biogenesis in two tissue types [21
]. In epithelial renal cells, GAL3 is normally localized at the base of primary cilia, and its absence causing primary cilium abnormalities is associated with major defects in epithelial cell polarity [22
]. In tracheal cells, GAL3 is a component of the basal-foot cap of motile cilia, and its absence causes perturbed ciliary organization, which provokes defects in mucus clearance [21
]. Whether GAL3 is involved in chondrocyte primary cilium formation is unknown.
In this study, we used the Gal3−/− mouse to (1) directly assess the role of GAL3 in OA and (2) further explore the importance of GAL3 in chondrocyte primary cilium formation.
We observed that GAL3 was a key regulator of cartilage homeostasis in mice, particularly in response to mechanical stress induced by joint instability. GAL3 deletion promoted OA and altered chondrocyte functions including formation of primary cilium, catabolic activities and cell death.
In this study, we reported that Gal3−/−
mutant mice showed OA development in two different models and that GAL3 had major role in chondrocyte primary cilium formation. Classical changes in chondrocyte metabolism detected included enhanced catabolic activity, accelerated rate of differentiation and high apoptosis, with no change in anabolic activity. Furthermore, using the MNX OA model, at one month after surgery, Gal3−/−
knees already displayed histological OA damages similar to that observed in older animals, whereas WT knees still appeared unchanged. Although a sham operation was performed in the contralateral legs, the contralateral knee cartilages of WT and Gal3−/−
mice were normal 4 weeks after sham surgery. Taken together, our results further establish GAL3 as a major factor involved in maintaining cartilage homeostasis. Compared to our previous study where we only observed pre-OA signs in 4-month-old Gal3−/−
mice, we clearly showed in this work cartilage destruction induced by aging and joint destabilization OA models [19
]. While, in the former study, 4-month-old Gal3−/−
mice only displayed cartilage structure parameter modifications without cartilage damage, we clearly showed, in this study, that 14-month-old Gal3−/−
mice had severe OA cartilage destructions [19
]. Moreover, while cartilage damages induced by mono-iodoacetate (MIA) injection were similar between Gal3−/−
and WT mice, they were more severe in Gal3−/−
mice than in WT mice when induced by joint instability [19
]. Taken together, these results implicated GAL3 in the cartilage response to mechanical stress. This situation might depend in part on primary cilium functions.
Indeed, the primary cilium modulates several signaling pathways involved in chondrocyte maturation, survival, or death [25
]. We previously observed that in renal and tracheal epithelia, a major site of GAL3 accumulation is the basal body of primary and motile cilia, respectively [22
]. We report here that GAL3 is indeed highly enriched at the basal body of chondrocyte primary cilia and in mutant chondrocytes the primary cilia were misshapen or straight but significantly longer in mutant than WT cells. These results agree with the ciliogenesis defects observed in tracheal epithelial cells [21
]. Because the primary cilium displayed mechanosensing properties in several cell types [6
], its defects, consecutive to the absence of GAL3, might participate in OA development via an abnormal mechanical response. Structural abnormalities of primary cilium morphology altered cilium functions as described in Bbs2−/−
cilia which displayed a reduced cilium beat frequency compared to normal cilia [33
]. Moreover, cilium length regulated anterograde and retrograde ITF transport, protein cargo traffic, protein signaling and mechanical transduction [34
]. Primary cilium length was regulated by intracellular calcium, AMPc and protein kinase A activation, mechanical compression and osmotic challenge [7
]. Increased cilium length accelerated anterograde ITF transport leading to increase calcium flux delivery [34
]. The cilium length response created a negative feedback loop leading to cilium shortening and reduction of mechanical transduction [34
]. Similarly, low-intensity ultrasound mechanical stimulation promoted elongation and bending of chondrocyte primary cilium and these length and shape changes were reversible [35
]. Thus, the increased primary cilium length induced by GAL3 deletion might dysregulate chondrocyte responses to mechanical stimulation. Alternatively, misshapen primary cilia might increase chondrocyte apoptosis as described in tubular cells [36
]. However, the direct link between these two results (misshapen and elongated primary cilium and OA phenotype) needs further study. We hypothesized that GAL3 might regulate Ihh signaling and chondrocyte hypertrophy differentiation and chondrocyte apoptosis [25
In our original study, we observed accelerated and disorganized chondrocyte differentiation in the growth plate of Gal3−/−
embryonic bones along with increased expression of Ihh in mature and hypertrophic zones, which raises the question of how GAL3 might impinge on cartilage physiology [38
]. The present results reveal GAL3 involved in articular chondrocyte differentiation and death. The role of galectins in apoptosis has been extensively documented. Depending on the cell type and/or subcellular localization, GAL3 could be proapoptotic or antiapoptotic [30
]. Here, we observed that cytosolic GAL3 prevented chondrocytes from mitochondrial-dependent apoptosis, which agrees with a previous study showing that GAL3 inhibited cytochrome C release from mitochondria via synexin interaction [41
]. In this study, we could not exclude that extracellular GAL3 contributed to the OA phenotype induced by GAL3 deletion.
4. Materials and Methods
4.1. Animals and Protocol for OA-Induced Joint Surgery
Non-littermate wild-type (WT) and Gal3−/−
129 SvEV mice were maintained in a specific pathogen-free animal facility and handled according to the French regulation for animal care. Gal3−/−
129 SvEV mice have been generated by F. Poirier since 1998 [31
]. Experiments followed the local Guidelines for Animal Experimentation (Ethics Committee Lariboisière-Villemin no.CEEALV/2012-02-01, Paris, France; approval date: 01 February 2012). OA was observed during aging (at 3 and 14 months) and on induction by partial resection of the medial meniscus (meniscectomy (MNX)) in the right knee of 2-month-old male WT (n
= 8) and Gal3−/−
= 8) mice as described [42
].A sham operation was performed on the left knee. Surgery was carried out under sedation with ketamine/xylazine (160 and 6 mg/kg, respectively) and buprenorphin (50 µg/kg). Four weeks after MNX, mice were sacrificed by cervical dislocation and knee articulations were collected. OA lesions were analyzed according to OARSI recommendations [44
]. Tissue samples were fixed in 4% paraformaldehyde (PFA) for 24 h, then soaked in 0.5 M EDTA at 4 °C for 10 days for complete decalcification before paraffin embedding. For the aging model, analysis involved 8 each of 3-month-old WT and Gal3−/−
mice and 7 and 5 of 14-month-old WT and Gal3−/−
4.2. Safranin O Staining and OARSI Scoring
Decalcified knee samples were embedded in paraffin. Sagittal sections 5-µm-thick were stained with safranin-O. Because OA lesions only occurred in the medial compartment in the MNX model, OA scoring was performed in sagittal sections of medial femorotibial joints. Double-blind (to surgery type and mouse genotype) OA grading was performed at three levels (2–3 slides/level) separated by a 50- to 60-µm interval by two researchers (N. Hafsia, HK Ea). Cartilage lesions in tibias and femurs were scored for OA on a scale from 0 to 12. The average of the worst total score observed by each researcher was used as the OA score for each mouse [44
]. The kappa score of inter-observer agreement was 0.62 which corresponded to a strong correlation.
Primary antibodies were rat monoclonal antibody directed against GAL3 kindly provided by Dr. HakonLeffler (Lund University, Sweden), anti-VDIPEN IgG by Irwin Singer and Ellen Bayne (Merck Research Laboratories, Rahway, NJ, USA), mouse monoclonal anti-acetylated α-tubulin and rabbit polyclonal anti-γ-tubulin from Sigma-Aldrich (Saint-Louis, MO, USA), mouse monoclonal anti-cytochrome C (Bd Pharmingen, Le pont de Claix, France), mouse monoclonal anti-type X collagen (Diagomics, Blagnac, France), rabbit polyclonal anti-ADAMTS-5 (Abcam, Cambridge, UK), mouse monoclonal anti-type 2 collagen (Abcam), and rabbit polyclonal anti-aggrecan (Millipore, Guyancourt, France). Secondary antibodies were goat anti-mouse Alexa-488 and goat anti-rabbit Alexa-568 (Life Technologies, Paisley, UK).
4.4. Immunostaining of Knee Sections
Antigen retrieval in serial 5-µm-thick sagittal paraffin sections of decalcified knee samples was obtained by incubation with citrate buffer 10 mM, pH 6, for 4 h at 70 °C, then hyaluronidase (1 mg/mL, 37 °C, 15 min) (collagen 2, aggrecan, ADAMTS-5) or pepsin (0.1% in phosphate buffered saline [PBS], 37 °C, 2 h) (collagen X) was added. Sections were incubated with rabbit primary polyclonal antibodies against type 2 collagen (1/200, 2 µg/mL), aggrecan (1/200, 2.5 µg/mL) and mouse primary monoclonal anti-type X collagen antibody (1/100, 2 µg/mL) at 4 °C overnight. ADAMTS-5 and VDIPEN immunostaining was performed as described [42
]. Nonspecific binding sites were blocked with MOM Blocking Reagent (Vector Labs, Peterborough, UK). Negative controls were non-specific IgG antibodies. For these antibodies, the number of positive staining cells was manually counted (4 fields per section, histolab software (Excilone, Elancourt, France) by two researchers (N. Hafsia, HK Ea) who were blinded to each other’s results. In case of discrepancy, a third analysis was performed by both researchers to reach consensus.
4.5. Primary Cultures of Chondrocytes
Articular chondrocytes were isolated from 6-day-old newborn mice as described [48
]. WT and Gal3−/−
chondrocytes were collected and pooled from a litter of WT and Gal3−/−
pups, respectively. Chondrocytes were then plated (at least triplicate/condition) at 105
cells/well in 24-well plates in DMEM 4 g/L glucose (Gibco, Les Ulis, France) containing 10% heat-inactivated fetal calf serum (FCS) (GE Healthcare, Illkirch, France), 4 mM glutamine (Gibco), penicillin (100 U/mL) (Gibco), and streptomycin (100 µg/mL) (Gibco). Chondrocyte cultures were kept until confluence. FCS starvation was performed 24 h before stimulation or fixation. For gene expression studies, confluent WT and Gal3−/−
chondrocytes in serum-free medium were stimulated for 24 h with mouse recombinant IL-1β (10 ng/mL; 201-LB/CF; R&D Systems, Lille, France).
4.6. Immunostaining of Primary Chondrocytes
Before immunostaining with anti-GAL3 (1/100), anti-γ-tubulin (1/200), or anti-acetylated α-tubulin (1/200) antibodies, confluent chondrocytes were fixed in −20 °C precooled methanol for 5 min, permeabilized for 30 min in PBS containing 0.025% saponin and 1% BSA, then incubated with primary antibodies overnight at 4 °C in PBS containing 0.025% saponin and 1% BSA. For cytochrome c immunostaining, chondrocytes were fixed for 5 min in acetone at −20 °C, then for 15 min in 4% PFA, permeabilized for 30 min in PBS containing 0.02% Triton X-100 and 3% BSA, before incubation with anti-cytochrome c antibody (1:200, 2.5 µg/mL) for 1 h at room temperature. Incubation with secondary antibodies was performed for 1.5 h at room temperature. Nuclei were detected by Hoechst 33342 staining (Life Science, Villeneuve-La-Garenne, France).
4.7. Primary Cilium Analysis
For primary cilium detection, confluent articular chondrocytes were fixed, incubated with primary antibodies (anti-γ-tubulin and anti-acetylated α-tubulin), then secondary antibodies (Alexa 568 anti-rabbit and Alexa 488 anti-mouse, respectively, (Life Technologies, Paisley, UK). Confocal images were acquired on a Leica TCS SP5 microscope with x63 lens magnification (Leica Microsystems, Wetzlar, Germany). Six random fields were analyzed for each sample. Confocal images were assessed by two researchers (M. Forien, D. Delacour). Primary cilium analyses involved use of 3D software, IMARIS 7.0 (Bitplane, Zurich, Switzerland). Six independent experiments were performed. In total, n = 769 Gal3−/− and n = 642 WT chondrocytes were analyzed to describe cilium abnormalities.
4.8. RNA Isolation and RT-qPCR
RNA extracts were prepared from chondrocytes in culture by using TRIzol reagent according to the manufacturer’s instructions (Life Technologies, Paisley, UK), followed by a purification step with RNeasy Mini Kit (Qiagen, Courtaboeuf, France). cDNA synthesis involved the High Capacity kit (Fisher Scientific, Illkirch, France). Relative mRNA levels were evaluated by quantitative PCR analysis with LightCycler (Roche Applied Science, Meylan, France) and ABsolute Blue qPCR SYBR Green Mix (Fisher Scientific). HPRT6 mRNA level was a normalization control. Primer sequences of different genes are listed in Table 3
. Primer sequences were designed using the PrimerBlast website then checked for self-annealing sites, 3’ complementarity, and potential hairpin formation using the Oligo Calc: Oligonucleotide Properties Calculator. Finally, the specificity of the primers was checked by doing an in silico PCR on the UCSC website. The efficiency of the primers was checked by doing a PCR with different quantities of cDNA (12.5, 25, 50, 125 and 250 ng) and the corresponding melting curves. Results of qRT-PCR were the mean of at least 3 to 5 independent triplicate experiments.
4.9. TUNEL Assay
Paraffin sections underwent TUNEL assay according to the manufacturer’s instructions (Millipore, Guyancourt, France). Sections were deparaffinised, rehydrated, treated for 15 min with 1 mg/mL proteinase K (Sigma-Aldrich) in PBS, then 8 min in 3% H2O2, then incubated for 4 min in equilibration buffer before adding TdT enzyme for 1 h at 37 °C. The reaction was stopped with Stop/Wash Buffer, and diaminobenzidine-conjugated anti-digoxigenin antibody was used to detect apoptotic cells. Sections were counterstained with methyl green.
Primary chondrocytes seeded on glass coverslips after fixation with acetone/methanol (vol/vol) for 5 min at −20 °C underwent TUNEL staining. Then chondrocytes were incubated with equilibration buffer for 10 s before adding TdT enzyme for 1 h at 37 °C. The reaction was stopped with Stop/Wash Buffer, and rhodamine-conjugated anti-digoxigenin antibody was used to detect apoptotic cells. Chondrocytes were counterstained with DAPI (Sigma-Aldrich) for cell counting. Ten random immunofluorescent images/slide were obtained and Axiocam software (Zeiss Microscopy, Marly Le Roi, France) was used for counting.
4.10. Induction of Apoptosis
After 5 days in culture, subconfluent chondrocytes were washed, starved overnight in serum-free medium, then cultured for 48 h in complete medium containing 20 ng/mL tumor necrosis factor α (TNF-α) (R&D Systems, Lille, France) or 50 nM actinomycin D (ActD) (Sigma-Aldrich) for inducing the extrinsic or intrinsic apoptotic pathway, respectively. Three independent quadruplicate experiments were performed.
4.11. Caspase3 Activity
After apoptosis induction, chondrocytes were incubated on ice for 30 min in lysis buffer (10 mM Tris pH 7.4, 200 mM NaCl, 5 mM EDTA pH 7.4, 10% glycerol and 1% NP40). Lysates were centrifuged at 10,000× g
for 5 min, and supernatants were collected and stored at −20 °C. Caspase3 (Cas3) activity was determined as described [49
], then normalized for protein contain.
4.12. Statistical Analysis
Data are reported as mean ± SEM or median and ranges. Medians (and ranges) are presented in a box plot where the box edges are the first and third quartiles, the line within the box represents the median and the lines outside the box represent the spread of the values. The non-parametric two-tailed Mann–Whitney U test was used for comparing groups. Statistical analysis involved use of StatView (SAS Institute). p < 0.05 was considered statistically significant.