Galectin 3 Deficiency Alters Chondrocyte Primary Cilium Formation and Exacerbates Cartilage Destruction via Mitochondrial Apoptosis

Mechanical overload and aging are the main risk factors of osteoarthritis (OA). Galectin 3 (GAL3) is important in the formation of primary cilia, organelles that are able to sense mechanical stress. The objectives were to evaluate the role of GAL3 in chondrocyte primary cilium formation and in OA in mice. Chondrocyte primary cilium was detected in vitro by confocal microscopy. OA was induced by aging and partial meniscectomy of wild-type (WT) and Gal3-null 129SvEV mice (Gal3−/−). Primary chondrocytes were isolated from joints of new-born mice. Chondrocyte apoptosis was assessed by Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL), caspase 3 activity and cytochrome c release. Gene expression was assessed by qRT-PCR. GAL3 was localized at the basal body of the chondrocyte primary cilium. Primary cilia of Gal3−/− chondrocytes were frequently abnormal and misshapen. Deletion of Gal3 triggered premature OA during aging and exacerbated joint instability-induced OA. In both aging and surgery-induced OA cartilage, levels of chondrocyte catabolism and hypertrophy markers and apoptosis were more severe in Gal3−/− than WT samples. In vitro, Gal3 knockout favored chondrocyte apoptosis via the mitochondrial pathway. GAL3 is a key regulator of cartilage homeostasis and chondrocyte primary cilium formation in mice. Gal3 deletion promotes OA development.


Introduction
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 [1,2]. 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][3][4]. However, the precise molecular mechanisms responsible for initiation or progression of OA remain to be elucidated.
The expression of galectin 3 (GAL3), a 30-kDa member of the galectin family of lectins, is increased in human OA versus normal cartilages [16,17]. 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 [19,20]. Importantly, recent work revealed the involvement of GAL3 in cilium biogenesis in two tissue types [21,22]. 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.

GAL3 Deficiency Leads to Spontaneous OA in 14-Month-Old Mice and Exacerbates Surgery-Induced OA in 3-Month-Old Mice
We first assessed the consequences of Gal3 deletion in two OA models. The cartilages of old (14 months) and young (3 months) wild-type (WT) and Gal3 −/− mice were compared after safranin-O staining of knee sections ( Figures 1A and 2A). The extent and intensity of safranin-O staining was similar between 3-and 14-month-old WT mice, so the thickness and the proteoglycan content of knee cartilage remained unchanged during aging. Gal3 −/− mice, which did not have an apparent skeletal anomaly during adulthood, had same mean body weight as WT mice. Safranin-O staining of 3-month-old cartilage was indistinguishable from that of 3-month-old WT mice, but severe defects appeared with age. The knee cartilage of 14-month-old Gal3 −/− mice showed a decrease in cartilage thickness with extensive loss of proteoglycan, associated with cartilage fissuration ( Figure 1A). With the OARSI scoring system used to quantify these observations, we found a high mean score for 14-month-old Gal3 −/− mice (4.3 (range 0-8.0), n = 5 mice) and a normal mean score for 14-month-old WT mice (0.7 (0-2.0), n = 7 mice, p = 0.014). We conclude that Gal3 −/− mice spontaneously show development of knee OA lesions with age. Negative controls were nonspecific IgG antibodies. Scale bars: 100 µm. Two-tailed Mann-Whitney U test between WT and Gal3 −/− . Then, we compared the effect of mechanical overload, a major cause of OA, on WT versus Gal3 −/− knee cartilages. Overload was induced by knee joint instability surgery (MNX) of 2 month-old mice. At 4 weeks after surgery, Gal3 −/− cartilages displayed fissuration and proteoglycan loss (mean OARSI score 4.3 (range 1.0-9.0), n = 8 mice), whereas WT cartilages were essentially unaffected (mean OARSI score 1.6 (range 0-4.0), n = 8 mice, p = 0.014) (Figure 2A).
These results establish that mice lacking Gal3 show severe OA during aging and in response to mechanical overload as compared with WT animals.

Deletion of GAL3 Increases Chondrocyte Catabolic Activity and Hypertrophy
Profound changes in chondrocyte metabolism are a hallmark of OA. To investigate the effect of GAL3 on the anabolic and catabolic activity of chondrocytes, we studied the expression of chondrocyte markers and different proteinases in two experimental settings: directly on cartilage sections and in primary cultures of articular chondrocytes from newborn mice.
First, we stained cartilage sections with antibodies directed against ADAMTS-5, one of the most potent aggrecanases involved in cartilage destruction [23]. The staining was greater in Gal3 −/− than WT cartilages both during aging-related OA and after mechanical-induced OA. In 14-month-old mice ( Figure 1B), this increase in ADAMTS-5 staining was seen in knee cartilages, as confirmed by comparing the percentages of ADAMTS-5+ chondrocytes (Table 1). In mechanical-induced OA ( Figure 2B), the relative increase in ADAMTS-5+ chondrocytes was more pronounced in Gal3 −/− than WT cartilages (Table 1). These data were completed by measuring the quantity of Adamts-5 mRNA in cultured chondrocytes. We found a 60% increase of Adamts-5 mRNA expression in Gal3 −/− than WT cells without treatment (p < 0.05, n = 5 independent experiments), so the difference in level of ADAMTS-5 protein may result, at least in part, from differences at the transcriptional level. In line with the in vivo data showing an overproduction of ADAMST-5 in Gal3 −/− OA cartilages, with IL-1β treatment, Adamts-5 mRNA content was increased more in Gal3 −/− than WT chondrocytes (4.2-fold vs 1.8-fold, p < 0.01, n = 5) ( Figure 3A)

Deletion of GAL3 Increases Chondrocyte Catabolic Activity and Hypertrophy
Profound changes in chondrocyte metabolism are a hallmark of OA. To investigate the effect of GAL3 on the anabolic and catabolic activity of chondrocytes, we studied the expression of chondrocyte markers and different proteinases in two experimental settings: directly on cartilage sections and in primary cultures of articular chondrocytes from newborn mice.
First, we stained cartilage sections with antibodies directed against ADAMTS-5, one of the most potent aggrecanases involved in cartilage destruction [23]. The staining was greater in Gal3 −/− than WT cartilages both during aging-related OA and after mechanical-induced OA. In 14-month-old mice ( Figure 1B), this increase in ADAMTS-5 staining was seen in knee cartilages, as confirmed by comparing the percentages of ADAMTS-5+ chondrocytes (Table 1). In mechanical-induced OA ( Figure 2B), the relative increase in ADAMTS-5+ chondrocytes was more pronounced in Gal3 −/− than WT cartilages (Table 1). These data were completed by measuring the quantity of Adamts-5 mRNA in cultured chondrocytes. We found a 60% increase of Adamts-5 mRNA expression in Gal3 −/− than WT cells without treatment (p < 0.05, n = 5 independent experiments), so the difference in level of ADAMTS-5 protein may result, at least in part, from differences at the transcriptional level. In line with the in vivo data showing an overproduction of ADAMST-5 in Gal3 −/− OA cartilages, with IL-1β treatment, Adamts-5 mRNA content was increased more in Gal3 −/− than WT chondrocytes (4.2-fold vs 1.8-fold, p < 0.01, n = 5) ( Figure 3A)    (Table 1). In addition, the intensity of the VIDPEN staining, located in the pericellular matrix, was greater in Gal3 −/− than WT cartilage ( Figure 2B). In parallel to these in vivo data, IL-1β treatment induced significantly higher mRNA levels of Mmp3 in Gal3 −/− than WT chondrocytes ( Figure 3B).
We next examined the expression pattern of type X collagen, a marker of the terminal (i.e., hypertrophic) stage of chondrocyte differentiation. After staining with anti-type X collagen serum, we observed a relatively weak signal, restricted to the calcified zone, in WT cartilages from old animals or mechanically overloaded animals. In contrast, the intensity of the staining was much stronger and expanded both cartilage zones in corresponding Gal3 −/− cartilages ( Figures 1B and 2B). Consistently, the mRNA content of type X collagen as well as Mmp13 was higher in Gal3 −/− than WT chondrocytes ( Figure 3C).  (Table 1). In addition, the intensity of the VIDPEN staining, located in the pericellular matrix, was greater in Gal3 −/− than WT cartilage ( Figure 2B). In parallel to these in vivo data, IL-1β treatment induced significantly higher mRNA levels of Mmp3 in Gal3 −/− than WT chondrocytes ( Figure 3B).
We next examined the expression pattern of type X collagen, a marker of the terminal (i.e., hypertrophic) stage of chondrocyte differentiation. After staining with anti-type X collagen serum, we observed a relatively weak signal, restricted to the calcified zone, in WT cartilages from old animals or mechanically overloaded animals. In contrast, the intensity of the staining was much stronger and expanded both cartilage zones in corresponding Gal3 −/− cartilages ( Figures 1B and 2B). Consistently, the mRNA content of type X collagen as well as Mmp13 was higher in Gal3 −/− than WT chondrocytes ( Figure 3C).
Finally, we assessed whether Gal3 deletion altered chondrocyte anabolic activity. We observed the same expression pattern for type-2 collagen and aggrecan between WT and Gal3 −/− cartilages after MNX. The staining intensity and proportion of positive-stained chondrocytes were identical between WT and Gal3 −/− cartilages ( Figure 4A). In vitro studies showed the same expression of Col2a, Acan and Sox-9 between WT and Gal3 −/− articular chondrocytes both at the basal level and after IL-1β stimulation ( Figure 4B). Finally, we assessed whether Gal3 deletion altered chondrocyte anabolic activity. We observed the same expression pattern for type-2 collagen and aggrecan between WT and Gal3 −/− cartilages after MNX. The staining intensity and proportion of positive-stained chondrocytes were identical between WT and Gal3 −/− cartilages ( Figure 4A). In vitro studies showed the same expression of Col2a, Acan and Sox-9 between WT and Gal3 −/− articular chondrocytes both at the basal level and after IL-1β stimulation ( Figure 4B). Hence, in the absence of Gal3, although chondrocyte anabolic activity was not altered, the overall catalytic activity of chondrocytes was greatly enhanced and the hypertrophic stage was much more prominent, which indicates an acceleration of the differentiation process.
To ensure that these effects were not secondary to a compensatory overexpression of other galectins, we compared the gene expression of different galectins between WT and Gal3 −/− articular chondrocytes. The expression of Lgal1, 2, 7 and 12 was similar between WT and Gal3 −/− chondrocytes, whereas that of Lgal4, 8 and 9 was slightly changed (1.27-, 0.78-and 0.86-fold, respectively) ( Figure  4C).

GAL3 Deletion Increases Chondrocyte Apoptosis via the Mitochondrial Pathway
During OA, chondrocyte hypertrophy leads to chondrocyte death [24]. TUNEL assay showed a greater number of apoptotic cells in Gal3 −/− than WT cartilages, both with aging and after MNX ( Figure 5A, Table 1). Hence, in the absence of Gal3, although chondrocyte anabolic activity was not altered, the overall catalytic activity of chondrocytes was greatly enhanced and the hypertrophic stage was much more prominent, which indicates an acceleration of the differentiation process.
To ensure that these effects were not secondary to a compensatory overexpression of other galectins, we compared the gene expression of different galectins between WT and Gal3 −/− articular chondrocytes. The expression of Lgal1, 2, 7 and 12 was similar between WT and Gal3 −/− chondrocytes, whereas that of Lgal4, 8 and 9 was slightly changed (1.27-, 0.78-and 0.86-fold, respectively) ( Figure 4C).

GAL3 Deletion Increases Chondrocyte Apoptosis via the Mitochondrial Pathway
During OA, chondrocyte hypertrophy leads to chondrocyte death [24]. TUNEL assay showed a greater number of apoptotic cells in Gal3 −/− than WT cartilages, both with aging and after MNX ( Figure 5A, Table 1). To determine which apoptotic pathway(s) was/were affected in the absence of GAL3, we treated primary cultures of articular chondrocytes with TNF-α (20 ng/mL), which triggered the extrinsic pathway, or with actinomycin D (ActD; 50 nmol/L), which triggered the intrinsic pathway. With ActD treatment, both the number of TUNEL-positive cells and level of Cas3 activity increased to a higher extent in Gal3 −/− than in WT chondrocytes (p < 0.01) ( Figure 5B,C). In contrast, the increase in Cas3 activity with TNF treatment was similar in Gal3 −/− and WT chondrocytes. Of note, TNF had no detectable effect on the number of TUNEL-positive cells under these experimental conditions.
Taken together, these results suggest that in cultured chondrocytes, the extrinsic pathway of apoptosis is GAL3 independent while the intrinsic pathway is overstimulated in the absence of GAL3. This latter conclusion could be confirmed by directly monitoring the state of mitochondrial homeostasis. Indeed, we found that the cytochrome C release induced by ActD treatment was significantly more extensive in Gal3 −/− than WT chondrocytes ( Figure 5D).

GAL3 Deletion Induced Alteration of Chondrocyte Primary Cilia
As GAL3 is associated with the basal body of primary cilia in renal epithelial cells [22] and primary cilia was involved in apoptosis, we studied the primary cilia of WT and Gal3 −/− mouse To determine which apoptotic pathway(s) was/were affected in the absence of GAL3, we treated primary cultures of articular chondrocytes with TNF-α (20 ng/mL), which triggered the extrinsic pathway, or with actinomycin D (ActD; 50 nmol/L), which triggered the intrinsic pathway. With ActD treatment, both the number of TUNEL-positive cells and level of Cas3 activity increased to a higher extent in Gal3 −/− than in WT chondrocytes (p < 0.01) ( Figure 5B,C). In contrast, the increase in Cas3 activity with TNF treatment was similar in Gal3 −/− and WT chondrocytes. Of note, TNF had no detectable effect on the number of TUNEL-positive cells under these experimental conditions. Taken together, these results suggest that in cultured chondrocytes, the extrinsic pathway of apoptosis is GAL3 independent while the intrinsic pathway is overstimulated in the absence of GAL3. This latter conclusion could be confirmed by directly monitoring the state of mitochondrial homeostasis. Indeed, we found that the cytochrome C release induced by ActD treatment was significantly more extensive in Gal3 −/− than WT chondrocytes ( Figure 5D).

GAL3 Deletion Induced Alteration of Chondrocyte Primary Cilia
As GAL3 is associated with the basal body of primary cilia in renal epithelial cells [22] and primary cilia was involved in apoptosis, we studied the primary cilia of WT and Gal3 −/− mouse chondrocytes.
With confocal microscopy, we could visualize the primary cilia of cultured WT chondrocytes stained with antibodies directed against the acetylated form of α-tubulin ( Figure 6A). Double labelling experiments revealed that Gal3 was highly enriched at the base of the cilium, where it colocalized with γ-tubulin, a marker of the basal body. We concluded that Gal3 is present at the level of the basal body of primary cilia of chondrocytes ( Figure 6A). chondrocytes. With confocal microscopy, we could visualize the primary cilia of cultured WT chondrocytes stained with antibodies directed against the acetylated form of α-tubulin ( Figure 6A). Double labelling experiments revealed that Gal3 was highly enriched at the base of the cilium, where it colocalized with γ-tubulin, a marker of the basal body. We concluded that Gal3 is present at the level of the basal body of primary cilia of chondrocytes ( Figure 6A). The total number of ciliated cells was lower in Gal3 −/− than WT chondrocytes (59% vs. 68%); this difference was even greater in serum-free cultures (66% vs. 81%). We observed normal looking (i.e., "straight shaped") primary cilia at the surface of 60% of WT chondrocytes (n = 642) but only 10% of Gal3 −/− chondrocytes (n = 769); moreover, cilia were significantly longer in Gal3 −/− than WT cultures ( Table 2). We also found cells with stunted or curved primary cilia and occasional double-ciliated cells. Although these "abnormalities" were present in cultures of WT chondrocytes, they were far more abundant in mutant chondrocytes. Upon serum starvation, a condition that favors ciliogenesis, the ratios of aberrant primary cilia differed between WT and mutant chondrocytes ( Table 2). The most frequent type of defect was the stunted shape, which was observed in 44.4% of Gal3 −/− chondrocytes and only 16.4% of WT chondrocytes (p < 0.001) ( Figure 6B).
Taken together, these data show that the primary cilia located at the surface of Gal3 −/− chondrocytes are straight but are abnormally long or are misshapen. The total number of ciliated cells was lower in Gal3 −/− than WT chondrocytes (59% vs. 68%); this difference was even greater in serum-free cultures (66% vs. 81%). We observed normal looking (i.e., "straight shaped") primary cilia at the surface of 60% of WT chondrocytes (n = 642) but only 10% of Gal3 −/− chondrocytes (n = 769); moreover, cilia were significantly longer in Gal3 −/− than WT cultures ( Table 2). We also found cells with stunted or curved primary cilia and occasional double-ciliated cells. Although these "abnormalities" were present in cultures of WT chondrocytes, they were far more abundant in mutant chondrocytes. Upon serum starvation, a condition that favors ciliogenesis, the ratios of aberrant primary cilia differed between WT and mutant chondrocytes ( Table 2). The most frequent type of defect was the stunted shape, which was observed in 44.4% of Gal3 −/− chondrocytes and only 16.4% of WT chondrocytes (p < 0.001) ( Figure 6B).
Taken together, these data show that the primary cilia located at the surface of Gal3 −/− chondrocytes are straight but are abnormally long or are misshapen.

Discussion
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][26][27][28][29][30]. 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 [22,31]. 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 [6,22,31,32], 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 −/− and Bbs4 −/− 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 [34,35]. Primary cilium length was regulated by intracellular calcium, AMPc and protein kinase A activation, mechanical compression and osmotic challenge [7,9,34,35]. 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,[27][28][29][30]37].
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 [30,39,40]. 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.

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 −/− (n = 8) mice as described [42,43].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 −/− mice, respectively.

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 50to 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][45][46]. The kappa score of inter-observer agreement was 0.62 which corresponded to a strong correlation.
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. Table 3. Primer sequences of different genes.

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% H 2 O 2 , 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.

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.

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.

Conclusions
We provide genetic evidence that GAL3 is a key regulator of cartilage homeostasis in mice, particularly in response to mechanical stress induced by joint instability. GAL3 deletion favors OA development and severity and disturbs chondrocyte functions including formation of primary cilium, catabolic activities and cell death. Some reports have implicated GAL3 in human OA, so further characterization of this OA animal model could lead to novel therapeutic strategies to tackle this debilitating disease.

Conflicts of Interest:
The authors declare no conflict of interest.