Dual regulation of metalloproteinase expression in chondrocytes by Wnt-1–inducible signaling pathway protein 3/CCN6




Wnt-1–inducible signaling pathway protein 3 (WISP-3)/CCN6 is mutated in progressive pseudorheumatoid dysplasia and may have effects on cartilage homeostasis. The aim of this study was to ascertain additional roles for WISP-3/CCN6 by determining its expression in osteoarthritic (OA) cartilage and by investigating its effects on cartilage-relevant metalloproteinase expression in immortalized (C-28/I2) and primary chondrocytes.


Cartilage steady-state levels of WISP-3/CCN6 messenger RNA and protein production were determined by real-time quantitative reverse transcription–polymerase chain reaction (RT-PCR) and immunohistochemistry, respectively. WISP-3/CCN6 was overexpressed in C-28/I2 cells, and the resultant clones were analyzed by quantitative RT-PCR. The stable clones were analyzed by RT-PCR for metalloproteinase expression, and the signaling pathways involved were investigated using pharmacologic inhibition. The effects of WISP-3/CCN6 on metalloproteinase expression in primary chondrocytes were investigated using a small interfering RNA approach.


WISP-3/CCN6 was highly expressed in OA cartilage compared with undamaged cartilage, at both the RNA and protein levels. WISP-3/CCN6 overexpression in C-28/I2 cells resulted in unexpected dual regulation of metalloproteinases; expression of the potent aggrecanase ADAMTS-5 was down-regulated 9-fold, while expression of MMP-10 was up-regulated 14-fold, and these responses were accentuated in the WISP-3/CCN6 clones grown in suspension. MMP-10 up-regulation was dependent on several MAPKs, but WISP-3/CCN6–mediated ADAMTS-5 repression was independent of these pathways and was partially relieved by activation of β-catenin signaling. WISP-3/CCN6 also suppressed ADAMTS-5 expression in C-28/I2 cells treated with cytokines. In cytokine-treated primary chondrocytes, gene silencing of WISP-3/CCN6 resulted in enhanced ADAMTS-5 expression, while MMP-10 expression was suppressed.


WISP-3/CCN6 was highly expressed in end-stage OA cartilage, suggesting a role for this growth factor in cartilage homeostasis. WISP-3/CCN6–induced repression of ADAMTS-5 expression and regulation of MMP-10 expression suggest complex and context-dependent roles for WISP-3/CCN6 in cartilage biology.

Wnt-1–inducible signaling pathway protein 3 (WISP-3; also known as CCN6) is a member of the CCN family of matricellular proteins (1). Little is known about the biologic activity of WISP-3/CCN6, but different mutations in WISP3 have been identified in several unrelated cases of progressive pseudorheumatoid dysplasia, a rare noninflammatory skeletal disease that manifests in childhood with the loss of articular cartilage in multiple joints and bone abnormalities (2). A single-nucleotide polymorphism in WISP3 has also been correlated with the incidence of polyarticular juvenile idiopathic arthritis (JIA), a disease that is clinically similar to progressive pseudorheumatoid dysplasia (3). WISP-3/CCN6 may therefore have a role in articular cartilage homeostasis. Interestingly, WISP-3/CCN6 is chemotactic for mesenchymal stromal cells (4), which may also suggest that it is involved in tissue repair.

Overexpression of WISP-3/CCN6 in immortalized chondrocytic cell lines leads to increased expression of the 2 major matrix components of articular cartilage, aggrecan and type II collagen (5), suggesting that WISP-3/CCN6 is a potential stimulator of anabolic pathways in cartilage. Furthermore, expression of WISP-3/CCN6 is reduced in the chondrocytes of mice lacking connective tissue growth factor (CCN2), another CCN family member that is known to promote chondrocyte differentiation (6). Given the severe articular cartilage degeneration observed in patients with progressive pseudorheumatoid dysplasia, it is also possible that WISP-3/CCN6 regulates cartilage degradation through effects on proteinase expression. Indeed, chondrocytes from one patient with progressive pseudorheumatoid dysplasia and mutant WISP-3/CCN6 showed dysregulated expression and abnormal secretion of matrix metalloproteinases (MMPs) (7).

Aggrecan degradation is an important event in joint disease. Cartilage aggrecan may protect type II collagen from degradation. and aggrecan degradation precedes type II collagen degradation (8). The proteinases primarily involved in cartilage matrix degradation are ADAMTS proteinases and the MMPs (for review, see ref.9). Once an excess loss of aggrecan over synthesis occurs, type II collagen becomes accessible to collagenases such as MMP-13; the situation may be exacerbated by exposure to chondrocyte type II collagen receptors such as discoidin domain receptor 2, which can up-regulate MMP-13 expression upon ligand binding (10). Following collagen degradation, the process of cartilage destruction becomes irreversible. Thus, the enzymes responsible for pathologic aggrecan degradation play a key role in initiating cartilage degeneration in joint disease.

The major proteolytic aggrecan fragment identified in osteoarthritis (OA), inflammatory joint disease, and joint injury is that resulting from cleavage of aggrecan at the ADAMTS interglobular site (11, 12). Of all the ADAMTS family members that have been demonstrated to cleave aggrecan, ADAMTS-4 and ADAMTS-5 appear to have the strongest activity at the interglobular domain site (13, 14). In particular, ADAMTS-5 has emerged as an important aggrecanase, because ADAMTS-5−/− mice with induced inflammatory arthritis are protected against cartilage degradation (15). ADAMTS-5−/− mice also had a significant reduction in the severity of cartilage degradation after surgical induction of joint instability compared with wild-type or ADAMTS-4−/− mice (16). Importantly, under physiologic conditions, the aggrecanolytic activity of ADAMTS-5 was 1,000-fold greater than that of ADAMTS-4 (17), and immunolocalization studies showed ADAMTS-5 in pericellular and intercellular locations in human OA cartilage, which are also the sites of the most intense staining for ADAMTS-generated NITEGE aggrecan fragments (18).

Due to the association of WISP-3/CCN6 mutations with cartilage destruction in progressive pseudorheumatoid dysplasia, a possible role for WISP-3/CCN6 in OA was investigated. Our results suggest that WISP-3/CCN6 is highly expressed in cartilage, with elevated expression in damaged cartilage compared with undamaged cartilage. Importantly, overexpression of WISP-3/CCN6 in C-28/I2 cells dramatically reduced the expression of ADAMTS-4 and ADAMTS-5, while the expression of MMP-1 and MMP-10 was elevated. WISP-3/CCN6 knockdown in cytokine-stimulated primary chondrocytes, using a loss-of-function approach, resulted in elevated expression of ADAMTS-5 but repression of MMP-10. Taken together, our results suggest that WISP-3/CCN6 may have multiple effects on cartilage metabolism.


Tissue collection.

Approval was obtained from the North Nottinghamshire Health Authority Local Research Ethics Committee (projects NNHA/420, NNHA/544, and NNHA/673). After informed consent was provided, articular surface specimens were obtained at the time of total knee joint replacement surgery from patients fulfilling the American College of Rheumatology revised criteria for tibiofemoral OA (19). Cartilage specimens were also recovered postmortem from the knee joints of donors who had no history of arthritis or knee pain (based both on information provided by the next of kin and medical records), did not have a diagnosis of OA, and had not undergone surgery or sustained a fracture to that joint. Subjects from whom specimens were obtained portmortem had no Heberden's nodes, rheumatoid nodules, or visible osteophytes. Primary chondrocytes were isolated, as previously described (20), from intact regions of the femoral condyles of patients with OA who were undergoing total knee replacement, with approval by the institutional review board and patient consent. The chondrocytes were cultured to confluence in Dulbecco's modified Eagle's medium (DMEM)/Ham's F-12 containing 10% fetal calf serum (FCS) and used for experimental purposes at passage 1.

Real-time reverse transcription–polymerase chain reaction (RT-PCR).

RNA was extracted from articular cartilage (pooled from femoral condyles and tibial plateaus) obtained from patients with OA and postmortem donors, using phenol–chloroform extraction following SPEXmill powdering of cartilage into TRIzol (Invitrogen). WISP-3/CCN6 messenger RNA (mRNA) transcript levels were analyzed by TaqMan one-step real-time RT-PCR, using a QuantiTect Probe RT-PCR kit (Qiagen). For TaqMan two-step real-time RT-PCR analysis of C-28/I2 gene expression, RNA was harvested from cells using an RNeasy Mini Kit (Qiagen) and reverse transcribed with Moloney murine leukemia virus reverse transcriptase (Promega) according to the manufacturer's instructions. The MMP/tissue inhibitor of metalloproteinases and ADAMTS primer/probe sequences were described previously (21). The WISP-3/CCN6 primer and probe sequences were as follows: forward 5′-CTCCAAAGCTGAAAAATTTGTCTTT-3′, reverse 5′-TTTTTGTGGAATATGCTTGGATAAGA-3′, probe 5′-CTGGATGCTCAAGTACTCAGAGTTACAAACCCA-3′. The expression of each gene was normalized to the expression of 18S from the same complementary DNA preparation, using primers and probe specific for 18S (Applied Biosystems). Statistical analysis of gene expression changes between 2 sets of data was performed using Student's 2-tailed t-test on groups containing at least 3 samples.

Immunohistochemical analysis.

Human articular cartilage was obtained from the femoral condyle of the knees of patients with OA (3 men [ages 60–73 years] and 2 women [ages 75 years and 82 years, respectively]) or age-matched postmortem donors (1 man [age 61 years] and 3 women [ages 59–63 years]). Serial 7-μm cryostat sections were fixed in cold acetone for 10 minutes, washed twice in Tris buffered saline–0.05% Tween (TBST) for 5 minutes, incubated with 0.5 mg/ml (175–350 units/ml) hyaluronidase (Sigma-Aldrich) in phosphate buffered saline for 20 minutes, washed in TBST, incubated with methanol plus 3% hydrogen peroxide for 30 minutes, washed in TBST, blocked for 1 hour in 20% normal sheep serum (Dako) and 1% bovine serum albumin, and incubated overnight at 4°C with primary polyclonal rabbit antibody directed against residues 1–100 of human WISP-3/CCN6 (Abcam; 0.6 μg/ml) or a nonimmune rabbit isotype control (Dako). Sections were washed and incubated with secondary biotinylated sheep anti-rabbit antibody (Serotec) for 30 minutes at room temperature, washed, incubated with streptavidin–horseradish peroxidase (StreptABC/HRP kit; Dako) and then with 3,3′-diaminobenzidine (DAB) (Chromogen System; Dako). The DAB chromogen was allowed to develop for an equal time on all sections, and cell nuclei were counterstained with hematoxylin.


The human WISP-3/CCN6 gene was cloned from a cocktail of brain/placenta/testis/chondrocyte RNA by one-step RT-PCR, using a Qiagen OneStep RT-PCR Kit with 50 picomoles of forward primer (5′-CGCGATGCAGGGGCTCCTCTT-3′) and reverse primer (5′-TATATTACAGAATCTTGAGCTCAG-3′). The 1-kb PCR product was cloned into a pCR2.1-TOPO vector using One Shot TOP10 Escherichia coli cells and a 5-minute TOPO Cloning Kit (Invitrogen Life Technologies) according to the manufacturer's protocol. For stable transfection in C-28/I2 cells, which are resistant to Geneticin (22), the WISP-3/CCN6 insert was subcloned into the plasmid pcDNA3.1/Hygro (Invitrogen Life Technologies), which confers hygromycin B resistance.

Cell culture.

Generation of the immortalized chondrocytic cell line C-28/I2 has been described previously (22). C-28/I2 cells were maintained in DMEM (Gibco Invitrogen), 10% FCS (Biosera), 100 units penicillin/streptomycin (Gibco Invitrogen), with 250 μg/ml hygromycin B (Invitrogen) added to transfected cells, at 37°C in an atmosphere of 5% CO2.

Stable transfection of C-28/I2 cells.

Five micrograms of pcDNA3.1/Hygro or pcDNA3.1/Hygro WISP-3/CCN6 plasmid DNA was linearized with Fsp I followed by phenol–chloroform extraction. Twenty-four hours prior to transfection, C-28/I2 cells were seeded at 100,000 cells/well into 6-well plates. Cells were transfected overnight in DMEM plus 10% FCS with 1 μg of pcDNA3.1/Hygro or pcDNA3.1/Hygro WISP-3/CCN6, using FuGENE 6 transfection reagent (Roche) according to the manufacturer's protocol. Forty-eight hours after transfection, cells were changed to selective medium, and hygromycin B–resistant cells were dilution-plated to establish single-cell clonal cell lines.

Suspension cell culture.

For suspension culture, 24-well tissue culture plates were coated with 0.1 ml/cm2 of 12 mg/ml polyHEMA (SigmaUltra; Sigma-Aldrich) in ethanol (23) and allowed to dry. Cells were seeded at 190,000 cells/well in full medium and grown for 48 hours, centrifuged, resuspended in 1 ml serum-free medium, and treated with 5 ng/ml interleukin-1α (IL-1α) and 10 ng/ml oncostatin M (OSM) (both from R&D Systems). In parallel, cells were seeded in monolayer (38,000 cells/well; 24-well plates) in full growth medium and after 48 hours were incubated in serum-free medium for 1.5 hour, then stimulated with IL-1α and OSM as described above. RNA was extracted 24 hours after cytokine treatment. For LiCl treatment, cells in monolayer were incubated with serum-free medium containing 40 mM LiCl (SigmaUltra; Sigma-Aldrich) or 40 mM NaCl (as an osmolarity control) for 24 hours. For signal pathway analysis, 1 control clone and 1 WISP-3/CCN6 clone were seeded in monolayer and treated 24 hours later with 10 μM U0126 (MEK-1/2 inhibitor; Sigma-Aldrich), 10 μM SB202190 (p38 MAPK inhibitor; Calbiochem), or 5 μg/ml BMS-354451 (IKK inhibitor; Calbiochem).

Western blotting.

Clonal C-28/I2 cell lines were seeded in T25 flasks and 24 hours later changed to serum-free medium with 5 μM BB-94 (British Biotech Pharmaceuticals) and 10 μg/ml aprotinin, leupeptin, pepstatin A, and soybean trypsin inhibitor (all from Sigma), then incubated for a further 72 hours before undergoing cell lysis in radioimmunoprecipitation assay buffer. Equal volumes of protein were examined by Western blotting with rabbit anti–WISP-3/CCN6 primary antibody (Abcam) and peroxidase-conjugated donkey anti-rabbit secondary antibody (The Jackson Laboratory) and then were developed with enhanced chemiluminescence Western blot analysis reagents (Amersham Biosciences).

WISP-3/CCN6 gene silencing.

Primary chondrocytes were transfected with control nontargeting small interfering RNA (siRNA; Dharmacon) or siRNA against WISP-3/CCN6 (Dharmacon) at a final concentration of 50 nM, using DharmaFECT transfection reagent. At 48 hours posttransfection, cells were stimulated with IL-1α and OSM for 48 hours. Samples were lysed and reverse transcribed using a Cells-to-cDNA (Ambion) protocol. WISP-3/CCN6 knockdown and any subsequent effects on metalloproteinase expression were assessed by real-time RT-PCR.


Up-regulated expression of WISP-3/CCN6 in OA cartilage compared with cartilage obtained postmortem.

WISP-3/CCN6 steady-state mRNA expression was assessed by real-time RT-PCR on RNA extracted from different human tissues and cells. WISP-3/CCN6 mRNA was highly expressed in cartilage, and steady-state levels of WISP-3/CCN6 mRNA were 2-fold higher in OA cartilage than in cartilage samples obtained postmortem (Figure 1A). WISP-3/CCN6 expression was considerably lower in the 27 other tissues studied (data not shown). To further explore the elevated expression of WISP-3/CCN6 in OA cartilage, we performed immunohistochemical analysis for WISP-3/CCN6 in cartilage obtained from patients with OA and donor cartilage obtained postmortem. Immunostaining for WISP-3/CCN6 was observed in OA cartilage sections and was often localized to pericellular areas (Figure 1B). Immunostaining for WISP-3/CCN6 was barely detectable in undamaged donor cartilage obtained postmortem (Figure 1B) or in undamaged cartilage from a single knee with OA (Figure 1C), while strong pericellular staining was observed in samples of damaged cartilage from the same OA knee (Figure 1C). Although the subjects in this study who provided cartilage samples postmortem did not have a diagnosis of OA and were not undergoing treatment for OA, some histologic sections from these donors displayed signs of cartilage damage. Interestingly, WISP-3/CCN6 immunostaining was positive in these postmortem specimens with OA-like changes, and WISP-3/CCN6 immunostaining was confined to damaged areas of the cartilage (Figure 1D). These results suggest that WISP-3/CCN6 protein is most highly expressed in damaged cartilage.

Figure 1.

Wnt-1–inducible signaling pathway protein 3 (WISP-3)/CCN6 expression in cartilage obtained from patients with osteoarthritis (OA) and donor cartilage obtained postmortem (PM). A, WISP-3/CCN6 mRNA expression in 20 postmortem cartilage samples and 18 OA cartilage samples was assessed by real-time reverse transcription–polymerase chain reaction. Values are the mean ± SEM. ∗ = P < 0.001. B–D, Immunohistochemical staining for WISP-3/CCN6 protein in cartilage sections was performed. B, Pericellular staining was observed in extensively damaged OA cartilage (top) but not in undamaged postmortem cartilage (bottom). C, No staining was observed in undamaged OA cartilage (top), but staining was observed in damaged OA cartilage (bottom). D, No staining was observed in undamaged postmortem cartilage (top), but diffuse staining was observed in adjacent damaged areas of the same cartilage (bottom). Bars = 100 μm.

Stable transfection of WISP-3/CCN6 in chondrocytic C-28/I2 cells.

In order to examine the biologic effects of elevated WISP-3/CCN6 expression in chondrocytic cells, we stably transfected the WISP-3/CCN6 gene into the immortalized chondrocytic cell line, C-28/I2. Three clonal cell lines stably overexpressing WISP-3/CCN6 or stably transfected with the empty vector alone were generated. Real-time RT-PCR confirmed that the clonal cell lines transfected with WISP-3/CCN6 were overexpressing WISP-3/CCN6 mRNA compared with the empty-vector–transfected clones (Figure 2A). The expression of WISP-3/CCN6 protein was confirmed by Western blotting of lysates from clonal cell lines grown in the presence of protease inhibitors (Figure 2A). As predicted, a protein band was observed at 39 kd in WISP-3/CCN6 clones. A larger band was observed at 100 kd, which may represent dimers or trimers of the protein. In 1 empty-vector clone, a low level of WISP-3/CCN6 protein expression was observed. In the absence of protease inhibitors, WISP-3/CCN6 protein was undetectable (data not shown). Other CCNs are known to be cleaved by MMPs (24).

Figure 2.

Wnt-1–inducible signaling pathway protein 3 (WISP-3)/CCN6 and metalloproteinase expression in clonal cell lines. WISP-3/CCN6 and metalloproteinase expression in clonal cell lines stably overexpressing the empty vector pcDNA3.1/Hygro (vector) or pcDNA3.1/Hygro plus WISP-3/CCN6 (WISP-3) was determined using real-time reverse transcription–polymerase chain reaction (RT-PCR), with expression normalized to the housekeeping gene 18S. A, Steady-state WISP-3/CCN6 mRNA expression as measured by RT-PCR (top), and Western blot obtained using rabbit anti–WISP-3/CCN6 primary antibody (bottom). B, Expression of ADAMTS-5 and ADAMTS-4. C, Expression of matrix metalloproteinase 10 (MMP-10) and MMP-1. Bars show the mean ± SEM fold change in 3 clones compared with the clones stably overexpressing the empty vector. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001.

Metalloproteinase expression in clonal C-28/I2 cell lines stably transfected with WISP-3/CCN6.

To determine the consequences of WISP-3/CCN6 overexpression in C-28/I2 cells, we evaluated the steady-state levels of mRNA for metalloproteinase genes in the 3 clonal WISP-3/CCN6–overexpressing cell lines and the 3 clonal empty-vector–transfected cell lines (Table 1). Of particular interest was the strong down-regulation of the levels of ADAMTS-5 mRNA (9-fold; P < 0.001) (Figure 2B) and ADAMTS-4 mRNA (4-fold; P < 0.05) (Figure 2B) in WISP-3/CCN6–overexpressing clones. Conversely, the steady-state mRNA levels of MMP-1 and MMP-10 were elevated 13-fold (P < 0.05) and 14-fold (P < 0.01), respectively, in the WISP-3/CCN6 clones compared with empty-vector clones (Figure 2C). We confirmed that the induction of MMP-10 was clearly observed at the protein level (additional information is available from the corresponding author), and the repression of ADAMTS-5 may be detectable at the protein level but requires further investigation (data not shown).

Table 1. Changes in steady-state mRNA levels of metalloproteinases in cells overexpressing WISP-3/CCN6*
MetalloproteinaseRegulation in cells overexpressing WISP-3/CCN6P
  • *

    The expression of matrix metalloproteinase (MMP) and ADAMTS mRNA in 3 Wnt-1–inducible signaling pathway protein 3 (WISP-3)/CCN6–overexpressing C-28/I2 clones and 3 empty-vector C-28/I2 clones was measured by real-time polymerase chain reaction and normalized to 18S expression.

  • Versus empty-vector clones.

MMP-113-fold increase<0.05
MMP-22-fold decrease<0.05
MMP-3No effect 
MMP-7No effect 
MMP-8No expression 
MMP-9No effect 
MMP-1014-fold increase<0.01
MMP-111.5-fold decrease<0.05
MMP-12No expression 
MMP-13No effect 
MMP-142-fold decrease<0.05
MMP-152.5-fold decrease<0.01
ADAMTS-1No effect 
ADAMTS-44-fold decrease<0.05
ADAMTS-59-fold decrease<0.001
ADAMTS-8No expression 
ADAMTS-9No effect 
ADAMTS-15No effect 

It was particularly interesting that WISP-3/CCN6 overexpression modulated ADAMTS-5 and MMP-10 expression in C-28/I2 cells, because ADAMTS-5 is a major aggrecanase, and MMP-10 may be involved in the formation of bony osteophytes in OA (25). We determined whether the modulation of ADAMTS-5 and MMP-10 expression by WISP-3/CCN6 was still observed when cells were grown in 3-dimensional (3-D) culture rather than in monolayer, because C-28/I2 cells grown in 3-D culture have a more chondrocytic phenotype (26). Three WISP-3/CCN6–overexpressing clones and 3 empty-vector clones were seeded on tissue culture plastic coated with polyHEMA in order to prevent cell adhesion to the substratum. Cells grew in large groups, forming nodule-like structures (results not shown). Suppression of ADAMTS-5 expression was maintained in WISP-3/CCN6–expressing clones compared with empty-vector–transfected clones (Figure 3A). In fact, growth in suspension significantly increased ADAMTS-5 expression in the empty-vector clones but not in the WISP-3/CCN6 clones (Figure 3A). Meanwhile, MMP-10 expression was further enhanced in the WISP-3/CCN6 clones grown in suspension (Figure 3A).

Figure 3.

ADAMTS-5 and MMP-10 expression in monolayer culture versus 3-dimensional culture in the presence and absence of cytokines, as determined by real-time RT-PCR, with expression normalized to 18S. A, ADAMTS-5 and MMP-10 expression in clonal cell lines stably overexpressing the empty vector (vector) or stably overexpressing WISP-3/CCN6 grown in monolayer or in suspension over polyHEMA. Bars show the mean ± SEM fold change in 3 clones compared with the clones stably overexpressing the empty vector in monolayer culture. B, ADAMTS-5 and MMP-10 expression in 1 empty-vector clone and 1 WISP-3/CCN6 clone grown in monolayer in serum-free medium (SF) or serum-free medium supplemented with 5 ng/ml interleukin-1α (IL-1α) and 10 ng/ml oncostatin M (OSM). C, ADAMTS-5 and MMP-10 expression in 1 empty-vector clone and 1 WISP-3/CCN6 clone grown in suspension over polyHEMA in serum-free medium or serum-free medium supplemented with 5 ng/ml IL-1α and 10 ng/ml OSM. Bars in B and C show the mean ± SEM fold change in 3 wells compared with cells stably overexpressing the empty vector under control conditions (SF, monolayer culture). ∗ = P ≤ 0.05; ∗∗ = P ≤ 0.01; ∗∗∗ = P ≤ 0.001. See Figure 2 for other definitions.

Inflammation can contribute to the progression of OA (27). To determine the effects of WISP-3/CCN6 overexpression in an inflammatory environment, cells were treated with IL-1α and OSM, which are known to stimulate metalloproteinase expression and matrix catabolism in cartilage explant cultures (28). In both monolayer and suspension cultures, these cytokines stimulated the expression of ADAMTS-5 mRNA in empty-vector cells and in WISP-3/CCN6–expressing cells, partially reversing WISP-3/CCN6 suppression of ADAMTS-5 in both monolayer (Figure 3B) and suspension (Figure 3C) cultures. However, the overall expression of ADAMTS-5 still remained significantly lower in the cytokine-treated WISP-3/CCN6–expressing cells compared with the cytokine-treated empty-vector cells (Figures 3B and C). Meanwhile, cytokines completely de-repressed the suppression of ADAMTS-4 expression in WISP-3/CCN6–expressing cells in monolayer culture. Cytokine treatment had no effect on MMP-10 mRNA levels in the empty-vector cell line but completely abrogated induction of MMP-10 to control levels in WISP-3/CCN6–expressing cells (Figures 3B and C). MMP-1 expression in WISP-3/CCN6–expressing cells grown in suspension and treated with cytokines was slightly but not significantly repressed (additional information is available from the corresponding author).

To determine potential pathway(s) involved in WISP-3/CCN6–induced regulation of ADAMTS-5 and MMP-10 expression, cells were treated with inhibitors of MEK-1/2, p38 MAPK, and IKK. These inhibitors did not reverse the suppression of ADAMTS-5 expression in WISP-3/CCN6–expressing cells, but each reduced WISP-3/CCN6–induced MMP-10 expression by 50% (Figure 4A), suggesting that WISP-3/CCN6–induced regulation of MMP-10 and ADAMTS-5 expression may operate via different pathways. Meanwhile, the IKK inhibitor reduced ADAMTS-5 expression in both the empty-vector and WISP-3/CCN6–expressing cells, suggesting that NF-κB signaling regulates ADAMTS-5 expression in C-28/I2 cells.

Figure 4.

ADAMTS-5 and MMP-10 expression in clonal cell lines treated with pathway inhibitors or LiCl. A, ADAMTS-5 and MMP-10 expression in 1 empty-vector control clone and 1 WISP-3/CCN6 clone treated with 10 μM U0126 (an MEK-1/2 inhibitor), 10 μM SB202190 (SB, a p38 MAPK inhibitor), or 5 μg/ml BMS-354451 (BMS, an IKK inhibitor), as measured by real-time RT-PCR. Bars show the mean ± SEM fold change in 3 wells compared with control cells stably overexpressing the empty vector under control conditions. B, ADAMTS-5 and MMP-10 expression in 3 empty-vector clones (vector) and 3 WISP-3/CCN6 clones (WISP3) incubated in 40 mM NaCl (as an osmolarity control) or 40 mM LiCl for 24 hours. Bars show the mean ± SEM fold change in 3 clones compared with the clones stably overexpressing the empty vector in NaCl. Expression of ADAMTS-5 and MMP-10 is normalized to 18S. ∗ = P ≤ 0.05; ∗∗ = P ≤ 0.01; ∗∗∗ = P ≤ 0.001. See Figure 2 for other definitions.

WISP-3/CCN6 has previously been shown to inhibit Wnt/β-catenin signaling in zebrafish and in human embryonic kidney and mouse embryo cell lines (29). We treated our clonal cell lines with LiCl, which can stimulate β-catenin signaling by inhibiting glycogen synthase kinase 3β (GSK-3β) (30). LiCl had no effect on empty-vector clones but partially reversed WISP-3/CCN6–induced suppression of ADAMTS-5 and enhanced MMP-10 expression in WISP-3/CCN6 clones (Figure 4B). This suggests that WISP-3/CCN6 and β-catenin signaling may interact in C-28/I2 cells, and this may affect MMP-10 and ADAMTS-5 expression. In agreement with these findings, treatment of the clonal cell lines with a GSK-3β inhibitor, SB216763 (30), produced similar results. SB216763 more potently enhanced MMP-10 expression in WISP-3/CCN6–overexpressing cells and also increased ADAMTS-5 expression in WISP-3/CCN6–overexpressing cells, although statistical significance was not reached (P = 0.056) (additional information is available from the corresponding author).

We used a gene-silencing approach to determine whether modulation of WISP-3/CCN6 expression also affected metalloproteinase expression in primary chondrocytes. WISP-3/CCN6 siRNA transfection resulted in a 90% reduction of WISP-3/CCN6 mRNA (Figure 5A) and elevation of ADAMTS-5 expression in 4 independent cell isolates treated with cytokines (Figure 5B). ADAMTS-4 showed a trend toward increased expression (Figure 5B). Conversely, MMP-10 expression was suppressed (Figure 5C), but MMP-1 expression was slightly increased.

Figure 5.

A–C, Effect of WISP-3/CCN6 knockdown in primary chondrocytes and potential model for WISP-3/CCN6 function in C-28/I2 cells and chondrocytes. Silencing of WISP-3/CCN6 by small interfering RNA (siRNA) transduction (A) significantly increased ADAMTS-5 expression (B) and decreased matrix metalloproteinase 10 (MMP-10) expression (C) in primary chondrocytes treated with interleukin-1α (IL-1α)/OSM. Bars show the mean ± SEM fold change in expression (3 replicate determinations), normalized to 18S. ∗∗ = P ≤ 0.01; ∗∗∗ = P ≤ 0.001. D, In C-28/I2 cells, MEK-1/2, p38 MAPK, and IKK are involved in induction of MMP-10 expression by WISP-3/CCN6 (1), but in the presence of IL-1α/OSM, WISP-3/CCN6 does not induce MMP-10 expression (2). WISP-3/CCN6–induced suppression of ADAMTS-5 expression may involve suppression of β-catenin signaling, although another pathway, perhaps opposing cytokine signaling, may be involved (3). β-catenin enhances WISP-3/CCN6 induction of MMP-10 by an unknown mechanism. This also demonstrates the complex- and context-dependent regulation of gene expression by WISP-3/CCN6, which may interact with canonical Wnt/β-catenin signaling in both positive and negative ways. In primary chondrocytes, WISP-3/CCN6 induces MMP-10 expression (4) and suppresses ADAMTS-5 expression (5) in the presence of cytokines, by unknown mechanisms. See Figure 3 for other definitions.


Here, we showed that WISP-3/CCN6 expression is significantly elevated in end-stage OA cartilage compared with clinically normal cartilage. A key finding in this study is that stable WISP-3/CCN6 overexpression in chondrocytic C-28/I2 cells led to down-regulation of the steady-state levels of ADAMTS-4 and ADAMTS-5 mRNA. If WISP-3/CCN6 has the same activity in vivo, this may suggest that WISP-3/CCN6 is up-regulated in response to cartilage damage as a feedback mechanism following early induction of ADAMTS-5 in order to suppress aggrecanase mRNA levels, which are known to decrease in end-stage OA (31, 32).

ADAMTS-5 appears to be a key player in aggrecan degradation (for review, see ref.33), and few agents are known to suppress ADAMTS-5 expression. Glucosamine and chondroitin sulfate supplements can decrease IL-1–induced ADAMTS-5 expression in bovine cartilage explants (34), and incorporation of the n-3 polyunsaturated fatty acid eicosapentaenoic acid into bovine articular chondrocyte membranes significantly decreased ADAMTS-5 mRNA expression induced by IL-1α (35). Importantly, several recent studies showed a key role for fibroblast growth factor 2 (FGF-2) in ADAMTS-5 suppression. FGF-2 suppresses IL-1α–induced ADAMTS-5 expression in human chondrocytes and inhibits glycosaminoglycan release in cartilage explants induced by cytokines or retinoic acid (36) and reduces basal ADAMTS-5 mRNA expression in unstimulated human chondrocytes (37). Furthermore, FGF-2–knockout mice had elevated levels of ADAMTS-5 mRNA and accelerated progression of OA following surgical induction of joint instability, and subcutaneous administration of recombinant FGF-2 reduced the expression of ADAMTS-5 mRNA to the levels observed in wild-type mice (38).

WISP-3/CCN6–induced suppression of ADAMTS-4 and ADAMTS-5 may explain the development of postnatal cartilage damage in patients with progressive pseudorheumatoid dysplasia and lacking wild-type WISP-3/CCN6. To date, metalloproteinase expression has been examined in chondrocytes from only 1 patient with progressive pseudorheumatoid dysplasia, and ADAMTS-5 expression was not increased in these cells lacking wild-type WISP-3/CCN6 compared with a control (7). It would be interesting to investigate the induction of aggrecanases in chondrocytes from patients with progressive pseudorheumatoid dysplasia upon mechanical injury or loading of chondrocytes, because this may be the key event that induces WISP-3/CCN6 activity and later suppresses ADAMTS-4 and ADAMTS-5 expression. Interestingly, mice lacking full-length WISP-3/CCN6 protein do not show any phenotypic similarities with patients with progressive pseudorheumatoid dysplasia (39). Similarly, unchallenged mice deficient in ADAMTS-5 activity do not have an obvious phenotype (15, 16). One possible explanation for the lack of a phenotype in WISP-3/CCN6–deficient mice is that WISP-3/CCN6 is induced as part of a repair response. A potential mechanism for induction of WISP-3/CCN6 expression in response to injury is via FGF-2, which is stored in the pericellular matrix surrounding cartilage chondrocytes (40) and is released in response to injury in porcine cartilage (41) and mechanical loading (42).

In our initial screening, MMP-1 and MMP-10 expression was strongly up-regulated in WISP-3/CCN6–overexpressing C-28/I2 cells. MMP-1 (collagenase 1) expression is elevated in OA knee cartilage compared with normal cartilage (43). MMP-10 mRNA is also up-regulated by cytokines in primary human articular chondrocytes, synovial fibroblasts, and chondrosarcoma cells (44, 45). Moreover, recombinant MMP-10 enhances IL-1α/OSM–induced collagenolysis of bovine cartilage explants, probably relating to MMP-10–induced activation of proMMP-1 (44).

The up-regulation of MMP-10 in WISP-3/CCN6–overexpressing cells therefore complements the up-regulation of MMP-1. MMP-10 protein is present in the synovial fluid of patients with OA, patients with RA, and patients with JIA and in the cartilage and synovial tissue of patients with OA and patients with RA (44). It may therefore seem surprising that WISP-3/CCN6 overexpression is associated with up-regulation of MMP-1 and MMP-10 when WISP-3/CCN6 has been hypothesized to be a chondroprotective gene. However, chondroprotective FGF-2 suppresses ADAMTS-5 expression but also induces MMP-1 expression in porcine chondrocytes (41), human articular cartilage explants (36), and human chondrocytes (37).

It thus could be contended that MMP-1 and MMP-10 are involved in matrix remodeling for attempted reparative/anabolic responses in OA. In support of this hypothesis, MMP-10 expression is associated with osteophyte formation (25). Interestingly, WISP-3/CCN6 promotes the migration of mesenchymal stromal cells (4). Although chondrocyte migration in vivo remains little studied, one could envisage that WISP-3/CCN6–mediated induction of MMP-1 and MMP-10 may promote migration of chondrocytes in an attempted wound-healing response through release of matrix-bound factors such as FGF-2.

ADAMTS-5 expression has previously been shown to increase in chondrocytes cultured in alginate beads (46). In the current study, culture in a 3-D environment elevated the expression of ADAMTS-5 in empty-vector clones but not in WISP-3/CCN6 clones, suggesting that the suppressive effect of WISP-3/CCN6 on ADAMTS-5 expression can inhibit or out-compete stimulatory signaling induced by cell-shape changes. Conversely, 3-D culture further enhanced the inductive effect of WISP-3/CCN6 on MMP-10 expression in C-28/I2 cells and increased MMP-10 expression in empty-vector clones. Treatment of cells with IL-1α/OSM increased ADAMTS-5 expression in both empty-vector and WISP-3/CCN6–expressing cells cultured in either a 2-D environment or a 3-D environment. However, ADAMTS-5 expression remained significantly lower in WISP-3/CCN6 clones treated with IL-1α/OSM compared with IL-1α/OSM–treated empty-vector clones, suggesting that WISP-3/CCN6 can still have a suppressive effect on ADAMTS-5 expression in an inflammatory environment.

Surprisingly, IL-1α/OSM strongly suppressed WISP-3/CCN6–induced MMP-10 expression but had no effect on this proteinase in empty-vector cells. It is possible that in C-28/I2 cells, the WISP-3/CCN6 pathway that induces MMP-10 expression is suppressed in the presence of inflammatory cytokines as a protective mechanism. Taken together, these results suggest that different signaling pathways are involved in WISP-3/CCN6–induced regulation of these 2 metalloproteinases in C-28/I2 cells.

We used a pharmacologic approach to begin to uncover potential signaling pathways involved in WISP-3/CCN6–induced modulation of metalloproteinase expression. The IKK inhibitor BMS-354451 suppressed basal ADAMTS-5 expression in empty-vector cells and further reduced ADAMTS-5 expression in WISP-3/CCN6–overexpressing cells, indicating that ADAMTS-5 expression is normally stimulated by the NF-κB pathway in C-28/I2 cells, mediated possibly by endogenous stress responses. ADAMTS-5 expression is also induced by NF-κB in nucleus pulposus cells (47). Partial suppression of WISP-3/CCN6–induced MMP-10 expression by all signaling pathway inhibitors tested suggests that this growth factor can impact on multiple regulatory networks.

We hypothesized that WISP-3/CCN6 may suppress ADAMTS-5 expression by inhibition of canonical Wnt/β-catenin signaling, because WISP-3/CCN6 harboring the mutations associated with cartilage degeneration lacks the ability to inhibit canonical Wnt signaling (29). Other investigators have shown that canonical Wnt/β-catenin signaling induces ADAMTS-5 expression in chick and rabbit chondrocytes (48, 49). An interpretation of our data is that WISP-3/CCN6 suppresses ADAMTS-5 expression by repressing basal β-catenin signaling, which is relieved by LiCl. Stimulation of Wnt/β-catenin signaling did not affect MMP-10 expression in empty-vector clones but strongly enhanced MMP-10 expression in WISP-3/CCN6–overexpressing cells. Thus, in contrast to its role in ADAMTS-5 expression, β-catenin may be positively involved in WISP-3/CCN6 induction of MMP-10 or can synergize with the WISP-3/CCN6 signaling pathway that induces MMP-10 expression. Interestingly, Wnt/β-catenin signaling induces osteophyte formation in animal models (50), and MMP-10 is expressed in osteophytes (25). WISP-3/CCN6 could interact with Wnt/β-catenin–activated processes in both positive and negative ways to modulate the outcome of Wnt/β-catenin signaling, reflecting the context-dependence of this pathway. Other CCN proteins can affect Wnt/β-catenin–activated pathways both positively and negatively in Xenopus Laevis (51).

In primary chondrocytes, WISP-3/CCN6 gene silencing revealed effects that partially mirrored those seen in WISP-3/CCN6–overexpressing C-28/I2 cells. Figure 5D summarizes WISP-3/CCN6–induced regulation of ADAMTS-5 and MMP-10 in C-28/I2 cells and primary chondrocytes. The mechanisms underlying the changes in ADAMTS-5 and MMP-10 following depletion of WISP-3/CCN6 in primary chondrocytes remain to be determined. The contrasting effects of WISP-3/CCN6 on MMP-10 and MMP-1 expression in primary chondrocytes and C-28/I2 cells in the presence of cytokines may reflect differences in signaling pathways and/or epigenetic effects in these 2 cellular contexts. Importantly, the effect of WISP-3/CCN6 on ADAMTS-5 expression was confirmed in primary chondrocytes, and the overall anticatabolic effects suggest important future areas of investigation of the role of this CCN family member in OA.

In conclusion, WISP-3/CCN6 is overexpressed in end-stage OA cartilage, and our in vitro studies suggest that this late expression may be an attempt by cartilage to reduce further damage through inhibition of aggrecan breakdown. The mechanisms by which WISP-3/CCN6 regulates metalloproteinase expression are unknown, but the results of the current study suggest that in certain circumstances WISP-3/CCN6 can regulate ADAMTS-5 and MMP-10 expression by distinct pathways and could modulate existing signaling cascades, thereby resulting in complex context-dependent control of metalloproteinase expression.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Gavrilović had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Baker, Sharpe, Newham, Barker, Langham, Goldring, Gavrilović.

Acquisition of data. Baker, Culley, Otero, Bevan, Barker, Clements, Langham.

Analysis and interpretation of data. Baker, Sharpe, Bevan, Barker, Clements, Langham, Goldring, Gavrilović.


Authors Sharpe, Newham, Barker, Clements, and Langham are employees of AstraZeneca.


We thank our colleagues at the Gavrilović laboratory and in the Cellular Protease Group at the University of East Anglia and in the AstraZeneca Respiratory Inflammation Research Area, especially Dr. W. M. Abbot and M.J. Snow, for helpful discussions throughout this project.