Six-week-old 129c/c mice were housed in wire cages in animal rooms with controlled temperature, humidity, and light cycles. Mice were allowed food and water ad libitum. Recombinant human gal-3 (rh-gal-3) was prepared in our laboratory and sterilized on a 0.2 μm filter. As the amino acid sequence of rh-gal-3 shows 85% identical homology and 91% positive homology with murine gal-3, we injected rh-gal-3 into the knees of wild-type mice. Mice were distributed into 4 groups receiving 100 ng, 1 μg or 10 μg of gal-3 or vehicle (PBS) alone according to previous established protocols 3031. After being anaesthetized with isoflurane, a skin incision was performed on each knee and a single injection of gal-3 or PBS administered under the patellar ligament using a Hamilton syringe with a 26G^3/8 intradermal needle. The day of injection was considered day 0 (D0); the animals were sacrificed 4 days after the injection. The study was performed according to the Canadian Council on Animal Care regulations and was approved by the Animal Care Committee of the University of Montreal Hospital Centre.

Animals were examined daily and knee diameter was measured using a digital calliper (model #2071M, Mitutoyo Corporation, Kawasaki, Japan) as described by Williams and colleagues 32. The swelling corresponded to the difference between joint diameter measured every day and joint diameter prior to the injection.

Histological evaluation was performed on the sagittal sections of the mouse knees removed at D4. Specimens were dissected, fixed in TissuFix #2 (Laboratoires Gilles Chaput, Montreal, QC, Canada), decalcified in RDO Rapid Decalcifier for bone (Apex Engineering, Plainfield, IL, USA), and embedded in paraffin. Serial sections (5 μm) were stained with safranin O and toluidine blue. The modifications in cartilage and subchondral bone were graded on a scale of 0 to 20 by two blinded, independent observers using a histological scale modified from Mankin and colleagues 33. This scale was used to evaluate the severity of modifications based on the loss of staining with toluidine blue (scale 0 to 4), cellular changes (scale 0 to 4), surface/structural changes in cartilage (scale 0 to 5), structure of the deep zone of cartilage (scale 0 to 4), and subchondral bone remodelling (scale 0 to 3). Scoring was based on the most severe histological changes within each cartilage and subchondral bone section.

The sections (5 μm) of each specimen were subjected to safranin O staining, as previously described 34. A Leica DMLS microscope (Leica, Weitzlar, Germany) connected to a personal computer (Pentium III, using Image J software, V.1.27, NIH, USA) was used to perform the subchondral bone morphometry analysis. The subchondral bone surface (μm) was measured on each slide in two 500 μm × 250 μm boxes, using as the upper limit, the calcified cartilage/subchondral bone junction as previously described 34. Two measurements were done and averaged for each section.

Femoral condyles and tibial plateaus were obtained from 15 OA patients (9 female and 6 male; aged 67 ± 9 years) following total knee arthroplasty. All patients were evaluated by a certified rheumatologist and, based on the criteria developed by the American College of Rheumatology Diagnostic Subcommittee for OA 35, were diagnosed as having OA. This procedure was approved by the Ethics Committee of the University of Montreal Hospital Centre.

Chondrocytes were released from the articular cartilage by sequential enzymatic digestion at 37°C, as previously described 3637 and cultured in DMEM (Invitrogen, Burlington, ON, Canada) supplemented with 10% FBS (Invitrogen) and an antibiotic mixture (100 units/ml penicillin base, 100 μg/ml streptomycin base; Invitrogen) at 37°C in a humidified atmosphere of 5% CO₂/95% air. Only first-passage cultured OA chondrocytes were used in the study. OA chondrocytes were seeded at 1 × 10⁵ cells in 12 well plates in DMEM containing 10% FBS for 48 h; the medium was then replaced for 24 h by DMEM containing 0.5% FBS, after which the cells were incubated for 24 h in fresh media containing 0.5% FBS in the absence or presence of rh-gal-3 (0 to 10 μg/ml).

The overlying cartilage was removed from the tibial plateaus, and the trabecular bone tissue was dissected from the subchondral bone plate. Primary subchondral osteoblasts were released as previously described 38. Briefly, subchondral bone samples were cut into small pieces of 2 mm² before sequential digestion in the presence of 1 mg/ml collagenase type I (Sigma-Aldrich, Oakville, ON, Canada) in DMEM without serum at 37°C for 30, 30, and 240 minutes. After being washed with the same medium, the digested subchondral bone pieces were cultured in DMEM containing 10% FBS. This medium was replaced every two days until cells were observed in the petri dishes. At confluence, cells were passaged once in 12- or 24-well plates in DMEM containing 10% FBS. Experiments were performed in DMEM containing 0.5% of charcoaled FBS with or without 50 nM 1,25 [OH]₂ D₃ (1,25-dihydroxycholecalciferol; vitamin D₃) in combination or not with gal-3. To evaluate signalling pathways involved in vitamin D₃-stimulated osteocalcin production that are inhibited by gal-3, cells were pre-incubated for 2 h with specific inhibitors and then incubated for 22 h in the presence of the inhibitors and vitamin D₃ in combination or not with gal-3. The inhibitors used were KT5720 (inhibitor of protein kinase A; final concentration 2 μM), KT5823 (inhibitor of protein kinase G; final concentration 2 μM), Genistein (broad inhibitor of tyrosine kinase; final concentration 20 μM), Taxifolin (an antioxidant flavonoid; final concentration 1 μM), wortmannin (inhibitor of phosphatidylinositol 3-kinase (PI 3-kinase); final concentration 250 nM), PD98059 (inhibitor of mitogen-activated protein kinase kinase-1 (MEK-1) activation; final concentration 10 μM), and SB202190 (inhibitor of p38 mitogen-activated protein kinase; final concentration 2 μM). All inhibitors were purchased from Calbiochem (San Diego, CA, USA).

RNA extraction and real time RT-PCR were performed as previously described 29. Primers for the genes encoding a disintegrin and metalloproteinase with thrombospondin type 1 motif (ADAMTS)-5 (aggrecanase-2), matrix metalloproteinase (MMP)-3 (stromelysin), osteocalcin, alkaline phosphatase and type I collagen α1 chain were synthesized by Invitrogen (Table 1). Data analysis was carried out using the Gene Amp 5700 Sequence Detector System software (Applied Biosystem, Foster City, CA, USA) and values normalized to the ribosomal subunit 18S. Specific primers for type I collagen α1 chain were designed using Primer3 software 39.

The assay measured only intact human osteocalcin and was performed on human osteoblast-conditioned media using a specific enzyme immunoassay (EIA) kit with a sensitivity of 0.5 ng/ml (Biomedical Technologies Inc., Stoughton, MA, USA).

Cells were lysed in 0.5% sodium dodecylsulfate and proteins quantified with the bicinchoninic acid assay 40.

Data are expressed as mean ± SEM or median (range). Statistical analyses were the Mann-Whitney U and the two-tailed Student's t-tests for animal experiments and cell culture, respectively. Results of p < 0.05 were considered significant.

As Ohshima and colleagues 15 showed that gal-3 was markedly present in OA synovial tissues during the inflammatory phase and could be recovered in the synovial fluid, we explored the potential extracellular role of gal-3. We injected gal-3 (0.1, 1, and 10 μg) into the knee joints of mice. To evaluate the potential role of gal-3 in the inflammation process we first determined if this molecule induces joint swelling. Data show that the vehicle alone (control) induced a joint swelling at D1 (p ≤ 0.0002 versus D0) (Figure 1). Although joint swelling at D2 was significantly lower compared to D1 (p < 0.005), a significant difference was still seen when D2 was compared to D0 (p < 0.004). Values gradually returned to the basal conditions. Gal-3 exacerbated and extended the swelling; thus, at D2, gal-3 injections of 0.1, 1, and 10 μg significantly induced higher swelling than the vehicle alone (p < 0.05, p < 0.004 and p < 0.002, respectively). This effect was sustained the third day post-injection (p < 0.006 for 0.1 μg, p < 0.002 for 1 μg, p < 0.0001 for 10 μg). Finally, at D4, values tended to return to those of the control group, although gal-3-induced joint swelling was still statistically significant (p < 0.006) with the highest gal-3 concentration used.

Furthermore, we investigated the effect of gal-3 on cartilage and subchondral bone using histological means. The global histological score (median and (range)) in the control group was 5.0 (3.5 to 6.0) whereas it reached 9.5 (7.0 to 12.5) (p < 0.04 versus control), 10.5 (8.5 to 12.5) (p < 0.02 versus control) and 13 (p < 0.04 versus control) in the gal-3-injected group with 0.1, 1, and 10 μg gal-3, respectively (Figure 2a). The cartilage score in the control group was 3.0 (2.0 to 4.0) whereas it reached 4.0 (3.5 to 5.5), 6.5 (7.5 to 5.5) (p < 0.02 versus control) and 8 (p < 0.04 versus control) in the gal-3-injected group with 0.1, 1, and 10 μg gal-3, respectively (Figure 2b). The subchondral score in the control group was 2.0 (1.0 to 2.5) whereas it reached 3.5 (3.0 to 4.5) (p < 0.04 versus control), 4.0 (3.0 to 5.0) (p < 0.04 versus control) and 5 (p < 0.04 versus control) in the gal-3-injected group with 0.1, 1, and 10 μg gal-3, respectively (Figure 2c). Therefore, both the cartilage parameters (structure/surface, cellularity, and toluidine blue staining) and the subchondral bone surface were modified by the gal-3 injection (Figure 3). These modifications are illustrated in Figure 3, which shows changes in the surface, in cellularity and remodelling of the deep layers in the presence of gal-3 (left panel (b-d)) compared to the control group. Destaining and modification of cell columns were also noticed in the presence of gal-3 (left panel (f-h)) compared to the control group.