Dkk-1 expression in chondrocytes inhibits experimental osteoarthritic cartilage destruction in mice

Authors


Abstract

Objective

Dkk is a family of canonical Wnt antagonists with 4 members (Dkk-1, Dkk-2, Dkk-3, and Dkk-4). We undertook this study to explore the roles of Dkk-1 and Dkk-2 in osteoarthritic (OA) cartilage destruction in mice.

Methods

Expression of Dkk and other catabolic factors was determined at the messenger RNA and protein levels in human and mouse OA cartilage. Experimental OA in mice was induced by destabilization of the medial meniscus (DMM) or by intraarticular injection of Epas1 adenovirus (AdEPAS-1). The role of Dkk in OA pathogenesis was examined by intraarticular injection of AdDkk-1 or by using chondrocyte-specific Dkk1 (Col2a1-Dkk1)–transgenic mice and Dkk2 (Col2a1-Dkk2)–transgenic mice. Primary culture mouse chondrocytes were also treated with recombinant Dkk proteins.

Results

We found opposite patterns of Dkk1 and Dkk2 expression in human and mouse experimental OA cartilage: Dkk1 was up-regulated and Dkk2 was down-regulated. Overexpression of Dkk1 by intraarticular injection of AdDkk-1 significantly inhibited DMM-induced experimental OA. DMM-induced OA was also significantly inhibited in Col2a1-Dkk1–transgenic mice compared with their wild-type littermates. However, Col2a1-Dkk2–transgenic mice showed no significant difference in OA pathogenesis. Wnt-3a, which activates the canonical Wnt pathway, induced Mmp13 and Adamts4 expression in primary culture chondrocytes, an effect that was significantly inhibited by Dkk-1 pretreatment or Dkk1 overexpression.

Conclusion

Our findings indicate that expression of Dkk1, but not Dkk2, in chondrocytes inhibits OA cartilage destruction. The protective effect of Dkk-1 appears to be associated with its capacity to inhibit Wnt-mediated expression of catabolic factors, such as Mmp13, providing evidence that Dkk-1 might serve as a therapeutic target for OA treatment.

Osteoarthritis (OA) is a chronic degenerative joint disease characterized by articular cartilage destruction. Biophysical and biochemical factors, including mechanical stress and proinflammatory cytokines, respectively, are responsible for the activation of catabolic pathways and initiation of OA cartilage destruction (1, 2). Activation of certain biochemical pathways in chondrocytes—unique resident cells that synthesize cartilage-specific extracellular matrix (ECM) components as well as various catabolic and anabolic factors—eventually leads to cartilage destruction. Among the articular chondrocyte biochemical pathways important in this context are those leading to the production of matrix metalloproteinases (MMPs) and ADAMTS, which are involved in the degradation of the ECM (1, 2). However, the molecular pathogenic mechanisms of OA are not understood in sufficient detail to allow the development of effective therapeutic targets for OA treatment.

One possible mechanism involved in the regulation of OA pathogenesis is the action of canonical Wnt signaling (3, 4), as evidenced by the fact that both chondrocyte-specific conditional activation and selective inhibition of β-catenin in mice yield OA-like phenotypes with different mechanisms (5, 6). The canonical Wnt pathway is regulated by various antagonists, including Wnt inhibitory factor 1, sclerostin, secreted Frizzled-related protein (sFRP), and Dkk. Secreted FRP inhibits the canonical Wnt pathway by binding to Wnt proteins (7). Mutations in sFRP-3, encoded by FRZB, are associated with human OA pathogenesis (8, 9), and Frzb-knockout mice show more severe OA cartilage destruction in response to instability, enzymatic injury, or inflammation (10). In contrast, Dkk acts by binding to low-density lipoprotein receptor–related protein 5 (LRP-5) and LRP-6 coreceptors (11). Dkk is a family of soluble canonical Wnt antagonists with 4 members (Dkk-1, Dkk-2, Dkk-3, and Dkk-4). Among members of the Dkk family, Dkk-1 is known to inhibit the canonical Wnt pathway, whereas Dkk-2 inhibition of this pathway depends on cellular context (11). For instance, Dkk-2 blocks the Wnt-1–dependent canonical pathway in OA osteoblasts (12), while it is also known to activate β-catenin signaling in Xenopus embryos (13). These opposing effects of Dkk-2 appear to be modulated by kremen 2, which converts Dkk-2 from an agonist to an antagonist of LRP-6 (14).

Although there is evidence to suggest that Dkk-1 is associated with OA development, reported results are inconclusive and somewhat contradictory. For example, cohort studies indicated that high serum levels of Dkk-1 were associated with reduced progression of radiographic hip OA (15, 16). Additionally, a recent study by Honsawek et al (17) indicated that plasma and synovial fluid levels of Dkk-1 were markedly lower in the knees of OA patients than in those of healthy controls, suggesting that circulating Dkk-1 levels were inversely correlated with radiographic OA severity. However, in contrast to the above reports, it has been shown that up-regulated Dkk1 expression in cartilage is associated with increased OA development, and intraperitoneal administration of Dkk1 antisense oligonucleotides ameliorates chondrocyte apoptosis and cartilage destruction (18, 19). Moreover, the roles of other Dkk members in OA pathogenesis are unknown, and direct evidence of Dkk function in OA pathogenesis in vivo is lacking, especially evidence of direct effects of Dkk-1 and Dkk-2 in articular chondrocytes.

We initially observed opposite patterns of Dkk1 and Dkk2 expression in OA cartilage: up-regulation of Dkk1 and down-regulation of Dkk2. We therefore hypothesized that Dkk-1 and Dkk-2 have specific functions in OA pathogenesis. Germline Dkk knockout is fatal to embryos (20). We therefore addressed specific in vivo functions of Dkk-1 and Dkk-2 in OA cartilage destruction by using a Dkk1 and Dkk2 misexpression strategy. Specifically, we generated cartilage-specific Dkk1- and Dkk2-transgenic mice by using Col2a1 promoter and enhancer (21) to illustrate in vivo functions of Dkk-1 and Dkk-2 in OA pathogenesis. We report here that cartilage-specific overexpression of Dkk-1 exerts a protective effect in OA cartilage destruction by inhibiting the canonical Wnt pathway. We additionally found no role for Dkk-2 in OA cartilage destruction, although its expression was significantly decreased in OA cartilage.

MATERIALS AND METHODS

Mice.

Male C57BL/6 mice, chondrocyte-specific Dkk1 (Col2a1-Dkk1)–transgenic mice and Dkk2 (Col2a1-Dkk2)–transgenic mice, STR/Ort mice, and CBA/CaCrl (CBA) mice were used for experimental OA studies. For the generation of chondrocyte-specific Dkk1- and Dkk2-transgenic mice, mouse Dkk1 and Dkk2 genes were cloned by polymerase chain reaction (PCR) using primers for Dkk1 (sense 5′-ATGATGGTTGTGTGTGCAGCGGCAGCT-3′ and antisense 5′-TTAGTGTCTCTGGCAGGTGTGGA-3′) and Dkk2 (sense 5′-ATGGCCGCGCTGATGCGGGT-3′ and antisense 5′-TCAGATCTTCTGGCATACATG-3′). The cloned complementary DNA (cDNA) was inserted into the Not I site of the pNass vector (Clontech), which contains the promoter and enhancer of the mouse Col2a1 gene (21). Animals were maintained under pathogen-free conditions. All experiments were approved by the Gwangju Institute of Science and Technology Animal Care and Use Committee.

Arthritic cartilage, experimental OA, and histology of OA cartilage.

International Cartilage Repair Society (ICRS) (22) grade 4 human OA cartilage (23) was sourced from individuals (age 51–72 years) undergoing arthroplasty for OA of the knee joint. The Wonkwang University Hospital Institutional Review Board approved the use of these materials, and all individuals provided written informed consent before the operative procedure. Messenger RNA (mRNA) levels of Dkk-1 and Dkk-2 in the undamaged and damaged cartilage regions from the same patients were quantified by quantitative reverse transcription–PCR (qRT-PCR). Similar locations of the undamaged and damaged regions were selected for biochemical and histologic analysis. Human OA cartilage samples were frozen, sectioned at a thickness of 6 μm, fixed in paraformaldehyde, and stained with Alcian blue (21).

Spontaneous OA cartilage destruction in STR/Ort mice was examined at age 28 weeks (21, 24). Experimental OA was induced by destabilization of the medial meniscus (DMM) surgery or by intraarticular injection of 1 × 109 plaque-forming units (PFU) of mouse Dkk1–expressing adenovirus (AdDkk-1) in 10–12-week-old male mice (21, 25). Sham-operated animals and animals injected with empty adenovirus (mock transduction) were used as controls for DMM and AdEPAS-1 injection models, respectively. Mice were killed 8 weeks after DMM surgery or 2 weeks after virus injection for histologic and biochemical analyses. Cartilage destruction in mice was examined using Safranin O staining and scored using the Mankin scale (26). Briefly, knee joints were fixed in 4% paraformaldehyde, decalcified in 0.5M EDTA (pH 7.4) for 14 days at 4°C, and embedded in paraffin. The paraffin blocks were sectioned at 6 μm thickness. We obtained serial coronal sections (frontal section) from the entire joint at 70-μm intervals, which yielded ∼10 slides (27). The sections were deparaffinized in xylene, hydrated with graded ethanol, and stained with Safranin O (21, 28).

Cartilage destruction was scored by 2 blinded observers (C-HC, J-SC) using the Mankin scale. The results of the Mankin scoring represent the mean of the maximum score in each mouse, and the representative Safranin O–stained image was selected from the most advanced lesion in the serial sections of medial tibial plateaus. The Safranin O–stained slides were further analyzed for osteophyte formation, synovitis, and subchondral bone sclerosis. Osteophyte maturity was scored by examining anteromedial tibia regions in each animal (29). Synovitis was examined by Safranin O and hematoxylin staining of synovium (21). We selected a region of subchondral bone adjacent to the cartilage that does not align with the meniscus in the middle of the frontal sections of the medial tibial plateaus, and we measured bone volume/total tissue volume (%) using the AxioVision microscope image analysis program (Carl Zeiss) (30).

Immunostaining of Dkk-1 and Dkk-2.

Control and OA knee joints from wild-type (WT) mice, Col2a1-Dkk1–transgenic mice, and Col2a1-Dkk2–transgenic mice were fixed in paraformaldehyde, decalcified, embedded in paraffin, sectioned (5 μm), and processed for immunofluorescence microscopy (28). Briefly, sections containing articular cartilage, synovium, cruciate ligament, or meniscus were incubated with anti–Dkk-1 (R&D Systems) or anti–Dkk-2 (Abcam) antibodies overnight at 4°C and visualized with a fluorescein isothiocyanate–conjugated or rhodamine-conjugated secondary antibody (Invitrogen).

Construction of AdDkk-1 and AdEPAS-1.

The adenovirus expressing mouse Epas1 was described previously (21, 28). Mouse Dkk1 cDNA was inserted into the Not I site of the pShuttle-CMV vector. AdDkk-1 was produced by Seoulin Bioscience using the pAdEasy System (Qbiogene), as described previously (21, 28). Mouse articular chondrocytes were cultured for 3 days, infected with AdDkk-1 or empty virus (mock) for 90 minutes at a multiplicity of infection of 800, incubated for 12 hours after infection, and treated with recombinant Wnt-3a (R&D Systems) for an additional 24 hours. AdDkk-1 or empty virus was injected into the joints of 8–10-week-old C57BL/6 male mice 15, 25, 35, and 45 days after DMM surgery (21, 28). Briefly, mice were anesthetized by intraperitoneal injection of 2.5% tribromoethanol, and knee joints were injected laterally with 1 × 109 PFU (8 μl) of AdDkk-1 or empty virus (mock). Mice were killed 8 weeks after DMM surgery for histologic analyses.

Primary culture of articular chondrocytes.

Mouse articular cartilage was isolated from femoral heads, femoral condyles, and tibial plateaus from postnatal day 5 mice, as described previously (31). Chondrocytes were maintained as a monolayer in Dulbecco's modified Eagle's medium, and cells on culture day 3 were treated as indicated in each experiment.

RT-PCR and qRT-PCR analyses.

Total RNA was isolated from primary culture mouse articular chondrocytes and OA cartilage tissues using TRI reagent (Molecular Research Center). Knee joint cartilage samples of 3 mice in any group were scraped off with no. 10 surgical blades and homogenized in TRI reagent (21). RNA was reverse-transcribed, and the resulting cDNA was amplified by PCR using Taq polymerase (iNtRON). PCR primers and conditions for Mmp2, Mmp3, Mmp9, Mmp12, Mmp13, Mmp14, Mmp15, Adamts4, Adamts5, Nos2, Ptgs2, and Gapdh have been described previously (21). PCR primers for Dkk1, Dkk1, Dkk2, and Dkk2 were as follows: for mouse Dkk1, sense 5′-TCTGCTAGGAGCCAGTGCC-3′ and antisense 5′-GATGGTGATCTTTCTGTATCC-3′; for human Dkk1, sense 5′-TCCCCTGTGATTGCAGTAAA-3′ and antisense 5′-TCCAAGAGATCCTTGCGTTC-3′; for mouse Dkk2, sense 5′-TGCCACAGTCCCCACCAAGGATC-3′ and antisense 5′-CCTGATGGAGCACTGGTTTGCAG-3′; for human Dkk2, sense 5′-GGTGCTGATGGTGGAGAGCTCACAG-3′ and antisense 5′-CGTTTGGTCACGAGGTAGTCCC-3′. Quantitative RT-PCR was performed using an iCycler (Bio-Rad) and SYBR Premix Ex Taq (Takara Bio). All qRT-PCR were performed in duplicate, and the amplification signal from the target gene was normalized to that of Gapdh in the same reaction.

The relative levels of Dkk1 and Dkk2 were analyzed by the comparative Ct (cycle threshold) method. The average Ct was calculated for Dkk1, Dkk2, and Gapdh. The ΔCt (CtDkks − CtGapdh) and the ΔΔCt (ΔCtDkk − ΔCtDkk1) were analyzed. The data calculated as 2math image indicate fold changes in gene expression relative to Dkk1.

Statistical analysis.

We performed nonparametric statistical analysis for the data quantified based on ordinal grading systems such as the Mankin scale, ICRS grade, and osteophyte formation. For qRT-PCR data evaluated by relative fold change, we first confirmed their normal distribution using the Shapiro-Wilk normality test, and we conducted Student's t-tests and analysis of variance followed by post hoc tests for pairwise comparisons and multiple comparisons, respectively. P values less than 0.05 were considered significant.

RESULTS

Up-regulation of Dkk1 and down-regulation of Dkk2 in OA cartilage.

We first examined Dkk expression levels in OA-affected human cartilage obtained from individuals undergoing arthroplasty. Human OA cartilage damage was confirmed by Alcian blue staining and scored according to the ICRS grading system. Dkk1 expression levels in damaged regions of human OA cartilage, as determined by qRT-PCR, were significantly increased compared to levels in undamaged regions (Figure 1A). We also examined mRNA levels of Dkk-1 and Dkk-2 in STR/Ort mice. The majority (>85%) of male STR/Ort mice have been shown to display signs of cartilage destruction at age 6 months (24). Dkk-1 mRNA levels were significantly increased in OA cartilage from STR/Ort mice compared with those in cartilage from CBA control mice (Figure 1B). Finally, Dkk expression levels were determined in 2 experimental mouse OA models: intraarticular injection of AdEPAS-1 (21) and DMM surgery (25). In both models, Dkk1 transcript levels were significantly elevated in OA cartilage compared with control cartilage (Figures 1C and D). In contrast to the Dkk1 expression pattern, Dkk2 expression in all examined OA cartilage was significantly decreased compared with that in control cartilage (Figure 1). Thus, Dkk1 and Dkk2 expression were oppositely regulated in OA cartilage, suggesting possible specific functions of the corresponding proteins in OA development and progression.

Figure 1.

Up-regulation of Dkk1 and down-regulation of Dkk2 in osteoarthritic (OA) cartilage. A, Undamaged (UD) and damaged (D) regions of human OA cartilage from the same patients, stained with Alcian blue and scored according to the International Cartilage Repair Society (ICRS) grading system (top). A result typical of those from 10 independent patients is presented. Dkk-1 and Dkk-2 mRNA levels in undamaged and damaged cartilage regions from the same patient were quantified by quantitative reverse transcription–polymerase chain reaction (qRT-PCR). Results presented are typical of those from 10 independent patients (bottom). B, Safranin O staining of cartilage from CBA/CaCrl (CBA) control and STR/Ort (STR) mice (top). Cartilage destruction was scored using the Mankin scale (n = 7 independent experiments), and Dkk-1 and Dkk-2 mRNA levels were quantified by qRT-PCR (n = 7 independent experiments) (bottom). C, Safranin O staining in sham-operated (Sham) control cartilage and cartilage with OA induced by destabilization of the medial meniscus (DMM) (top). Cartilage destruction was scored using the Mankin scale (n = 13 independent experiments), and Dkk-1 and Dkk-2 mRNA levels were quantified by qRT-PCR (n = 13 independent experiments) (bottom). D, Safranin O staining in cartilage sections from mice injected with AdEPAS-1 or empty virus (Mock) (top). Cartilage destruction was scored using the Mankin scale (n = 9 independent experiments), and Dkk-1 and Dkk-2 mRNA levels were quantified by qRT-PCR (n = 6 independent experiments) (bottom). Bars = 50 μm in A; 150 μm in BD. Values are the mean ± SEM. ∗ = P < 0.05; ∗∗ = P < 0.005; ∗∗∗ = P < 0.0001, by Mann-Whitney U test or Student's t-test.

We next examined Dkk-1 and Dkk-2 mRNA levels by qRT-PCR analysis in OA cartilage at different time points after DMM surgery. Dkk2 expression levels were a mean ± SEM 2.74 ± 0.59–fold higher than Dkk1 expression levels in WT mice (n = 7) (P = 0.008). Cartilage destruction, as determined by the Mankin score, was significant 2 weeks after DMM surgery. However, up-regulation of Dkk1 expression and down-regulation of Dkk2 expression in chondrocytes was significant at 6 and 3 weeks, respectively, after DMM surgery (Figure 2A), indicating that the reciprocal changes in the expression of Dkk1 and Dkk2 occur after the onset of cartilage destruction. We additionally determined Dkk expression levels in OA cartilage with various degrees of damage (Mankin scores 1–10) caused by DMM surgery. Increases in Dkk-1 expression and decreases in Dkk-2 expression after DMM surgery were observed at Mankin scores of 2–3 (Figure 2B). Thus, up-regulation of Dkk-1 and down-regulation of Dkk-2 were observed in cartilage tissue that was damaged to a relatively small degree. Dkk-1 is expressed in all layers of OA cartilage and subchondral bone, and its expression is weak in areas of cartilage that have lost proteoglycans (i.e., with loss of Safranin O staining). In contrast, Dkk-2 in normal cartilage is predominantly expressed in chondrocytes of the noncalcified cartilage.

Figure 2.

Expression patterns of Dkk1 and Dkk2 in degenerating knee joint articular cartilage after destabilization of the medial meniscus (DMM) surgery. A, Sham operation in the left knee joint and DMM surgery in the right knee joint were performed in C57BL/6 mice. At the indicated times after surgery, Mankin scores and mRNA for Dkk-1 and Dkk-2 were determined. Values are the mean ± SEM (n = 7 independent experiments). ∗ = P < 0.05; ∗∗ = P < 0.005; ∗∗∗ = P < 0.0005 versus sham operation at each indicated time point, by Mann-Whitney U test or analysis of variance followed by post hoc test. B, Articular cartilage obtained 8 weeks after DMM surgery was stained with Safranin O. Various degrees of cartilage destruction, as indicated by Mankin scores, are shown in sections from 5 independent experimental animals. Dkk-1 and Dkk-2 expression was determined by immunofluorescence microscopy. Bar = 200 μm.

Chondrocyte-specific Dkk-1 overexpression inhibits OA cartilage destruction.

The specific in vivo functions of Dkk-1 and Dkk-2 were evaluated by DMM surgery using chondrocyte-specific Col2a1-Dkk1–transgenic and Col2a1-Dkk2–transgenic mice and their corresponding WT littermates. Col2a1-Dkk1–transgenic mice were generated using a Col2a1 promoter and enhancer (Figure 3A), and overexpression of Dkk-1 in articular cartilage was confirmed by immunostaining (Figure 3B). Dkk-1 was also overexpressed in meniscus but not in other tissues, such as synovium and ligaments of Col2a1-Dkk1–transgenic mice (Figure 3C). Additionally, the transgenic mice did not show apparent synovitis (Figure 3C). Consistent with the increase in Dkk-1 mRNA in OA cartilage, DMM surgery enhanced Dkk-1 protein expression in the cartilage of WT mice. The DMM-induced increase in Dkk-1 expression was further elevated in the chondrocytes of Col2a1-Dkk1–transgenic mice (Figure 3B). DMM in WT mice caused severe cartilage destruction. However, DMM-induced cartilage destruction was significantly inhibited in Col2a1-Dkk1–transgenic mice, as determined by Safranin O staining (Figure 3D) and Mankin score (Figure 3E). Additionally, DMM-induced osteophyte formation and subchondral bone sclerosis observed in WT mice were significantly reduced in Col2a1-Dkk1–transgenic mice (Figures 3D and E), indicating that Dkk-1 overexpression suppresses DMM-induced OA pathogenesis.

Figure 3.

Chondrocyte-specific overexpression of Dkk-1 inhibits DMM-induced OA cartilage destruction. A, Shown are vector constructs used for the generation of Col2a1-Dkk1–transgenic mice. Poly A = poly adenine tail. B, Left, Expression of Dkk-1 in the articular cartilage from 2-week-old Col2a1-Dkk1–transgenic mice and their wild-type (WT) littermates was determined by immunofluorescence microscopy. Right, Dkk-1 immunostaining in knee joint cartilage obtained 8 weeks after sham operation and DMM surgery in Col2a1-Dkk1–transgenic mice and their WT littermates is shown. C, Shown is Dkk-1 detection in the meniscus, ligament, and synovium of Col2a1-Dkk1–transgenic mice and their WT littermates. Synovitis was examined by Safranin O and hematoxylin staining. D, Articular cartilage destruction, osteophyte formation, and subchondral bone sclerosis were determined by Safranin O staining in Col2a1-Dkk1–transgenic mice and their WT littermates after sham operation and DMM surgery (n = 8 independent experiments). Arrow indicates osteophyte. E, Shown are Mankin scores, osteophyte maturity, and bone volume/total tissue volume (BV/TV) in the knee joints of Col2a1-Dkk1–transgenic mice and their WT littermates after sham operation and DMM surgery. Bars in B–D = 200 μm. Values are the mean ± SEM (n = 8 independent experiments). ∗ = P = 0.05; ∗∗ = P = 0.005; ∗∗∗ = P = 0.0001, by Mann-Whitney U test or Student's t-test. NS = not significant (see Figure 1 for other definitions).

The inhibitory effects of Dkk-1 in OA cartilage destruction were further examined by ectopically expressing Dkk-1 in mice by intraarticular injection of AdDkk-1. AdDkk-1 injection caused overexpression of Dkk-1 in chondrocytes of articular cartilage, meniscus, synovium, and cruciate ligament of joints (Figure 4A). AdDkk-1 injection did not cause apparent synovitis (Figure 4A). Similar to the effects of Dkk-1 overexpression in chondrocytes of transgenic mice, AdDkk-1–mediated overexpression of Dkk-1 also significantly inhibited DMM-induced OA cartilage destruction, determined by Safranin O staining (Figure 4B) and Mankin score (Figure 4C). Dkk-1 overexpression by intraarticular injection of AdDkk-1 also significantly reduced DMM-induced osteophyte formation and subchondral bone sclerosis (Figures 4B and C). Thus, our overexpression strategies clearly revealed the protective effects of Dkk-1 in OA cartilage destruction in mice.

Figure 4.

Ectopic expression of Dkk-1 inhibits DMM-induced OA cartilage destruction. A, Expression of Dkk-1 in chondrocytes from articular cartilage, meniscus, cruciate ligament, and synovium 2 weeks after AdDkk-1 or mock injection was examined by immunofluorescence microscopy. Synovitis was determined by Safranin O and hematoxylin staining and quantified. B, Articular cartilage destruction, osteophyte formation, and subchondral bone sclerosis were visualized by Safranin O staining in 16-week-old C57BL/6 male mice after sham operation or DMM surgery. AdDkk-1 or empty virus (1 × 109 plaque-forming units [PFU]) was injected into knee joints 15, 25, 35, and 45 days after DMM surgery. Arrows indicate osteophytes. C, Mankin score, osteophyte maturity, and bone volume/total tissue volume (BV/TV) were determined in samples from 16-week-old C57BL/6 male mice after sham operation or DMM surgery. AdDkk-1 or empty virus (1 × 109 PFU) was injected into knee joints 15, 25, 35, and 45 days after DMM surgery, or no injections were given after surgery (none). Bars in A and B = 200 μm. Values are the mean ± SEM (n = 9 independent experimental animals). Statistical significance was evaluated by Mann-Whitney U test or Student's t-test. See Figure 1 for other definitions.

Chondrocyte-specific Dkk-2 overexpression does not affect OA cartilage destruction.

The in vivo functions of Dkk-2, which is down-regulated in OA cartilage (Figure 1), were evaluated by using Col2a1-Dkk2–transgenic mice and their corresponding WT littermates (Figure 5A). Immunostaining indicated overexpression of Dkk-2 in articular chondrocytes (Figure 5B). DMM surgery caused down-regulation of Dkk-2 expression in WT mice, as determined by immunostaining (Figure 5B) and qRT-PCR (Figure 1C). Dkk-2 expression in Col2a1-Dkk2–transgenic mice was also decreased after DMM surgery, but Dkk-2 expression still remained at a relatively high level compared with that in WT mice (Figures 5B and C). We have previously shown that type II collagen expression is decreased in OA cartilage destruction caused by DMM surgery (32), suggesting that the promoter activity of Col2a1 in transgenic mice may be reduced in DMM-induced OA cartilage destruction. Therefore, the decrease of Dkk-2 expression in transgenic mice may be due to the nature of the Col2a1 promoter and enhancer that we used to generate transgenic mice. Although Dkk-2 expression was down-regulated in OA cartilage, overexpression of Dkk-2 in cartilage did not significantly affect DMM-induced OA cartilage destruction, determined by Safranin O staining (Figure 5D) and Mankin score (Figure 5E). Thus, our results clearly indicate that Dkk-2, unlike Dkk-1, does not directly modulate OA cartilage destruction.

Figure 5.

Ectopic expression of Dkk-2 does not inhibit DMM-induced OA cartilage destruction. A, Shown are vector constructs used for the generation of Col2a1-Dkk2–transgenic mice. Poly A = poly adenine tail. B, Left, Expression of Dkk-2 in the articular cartilage of 2-week-old Col2a1-Dkk2–transgenic mice and their wild-type (WT) littermates was determined by immunofluorescence microscopy. Right, Dkk-2 immunostaining in knee joint cartilage obtained 8 weeks after sham operation and DMM surgery in Col2a1-Dkk2–transgenic mice and their WT littermates is shown. C, Relative Dkk-2 mRNA levels in articular cartilage from Col2a1-Dkk2–transgenic mice 8 weeks after sham operation and DMM surgery were determined by qRT-PCR. Values are the mean ± SEM (n = 7 independent experiments). D and E, Shown are Safranin O staining (D) and Mankin scores (E) in knee joint cartilage from Col2a1-Dkk2–transgenic mice and their WT littermates after sham operation and DMM surgery. Bars in B and D = 200 μm. Values in E are the mean ± SEM (n = 9 independent experiments). Statistical significance was evaluated by Mann-Whitney U test. NS = not significant (see Figure 1 for other definitions).

Dkk-1 inhibits Wnt-mediated expression of the catabolic factors Mmp13 and Adamts4 in articular chondrocytes.

In an attempt to elucidate the inhibitory mechanisms of OA cartilage destruction by Dkk-1, we examined expression of Dkk1 under pathogenic conditions, treating primary culture mouse articular chondrocytes with interleukin-1β (IL-1β), a major proinflammatory cytokine involved in OA pathogenesis (1, 2). Consistent with the results observed in OA cartilage, IL-1β significantly increased Dkk1 expression and decreased Dkk2 expression in chondrocytes (Figure 6A). The role of Dkk-1 in chondrocytes was more directly determined by treatment with recombinant Dkk-1 protein or overexpression of Dkk-1 using an adenoviral system. The adenoviral system effectively caused Dkk-1 overexpression in primary culture chondrocytes: a mean ± SEM 29.5 ± 1.28% of cells were infected by AdDkk-1.

Figure 6.

Dkk-1 inhibits Wnt-3a–induced Mmp13 and Adamts4 expression in articular chondrocytes. A, Dkk-1 and Dkk-2 mRNA was detected and quantified by qRT-PCR in chondrocytes treated with interleukin-1β (IL-1β) for the indicated times. Values are the mean ± SEM (n = 5 independent experiments). ∗ = P = 0.05; ∗∗ = P = 0.005; ∗∗∗ = P = 0.0001 versus no IL-1β, by post hoc test. B and C, Expression of catabolic factors was determined by RT-PCR (C) and quantified by qRT-PCR (B) in control (vehicle-treated) chondrocytes or in chondrocytes treated with Wnt-3a (100 ng/ml for 24 hours) with or without 1 μg/ml of recombinant (Re) Dkk-1 (Wnt-3a/Dkk-1 in B) or infected or not infected with 800 multiplicities of infection of AdDkk-1 (Wnt-3a/AdDkk-1 in B). Values in B are the mean ± SEM (n = 5 independent experiments). D, Expression of catabolic factors was determined by RT-PCR (left) and quantified by qRT-PCR (right) in untreated (none) chondrocytes in chondrocytes treated with Wnt-3a (100 ng/ml for 24 hours) with or without 1 μg/ml of recombinant Dkk-2. Values are the mean ± SEM (n = 9 independent experiments). In B and D, ∗ = P = 0.05; ∗∗ = P = 0.005; ∗∗∗ = P = 0.0001, by analysis of variance followed by post hoc test. BSA = bovine serum albumin; NS = not significant (see Figure 1 for other definitions).

To stimulate the canonical Wnt pathway, we treated chondrocytes with recombinant Wnt-3a protein. Wnt-3a caused up-regulation of Mmp13 and Adamts4 (Figures 6B–D), which encode proteins known to play a role in OA cartilage destruction (29, 33). The expression of other catabolic factors examined was not modulated by Wnt-3a treatment (Figures 6B–D). Pre-exposure of chondrocytes to recombinant Dkk-1 protein or overexpression of Dkk-1 significantly inhibited Wnt-3a–mediated up-regulation of Mmp13 and Adamts4 (Figures 6B and C). In contrast, pre-exposure of chondrocytes to recombinant Dkk-2 protein did not modulate Wnt-3a–induced expression of Mmp13 and Adamts4 (Figure 6D), indicating that Dkk-2 protein does not effectively inhibit the Wnt-3a–mediated canonical Wnt pathway. Therefore, these results suggest that Dkk-1 inhibits OA cartilage destruction by inhibiting Wnt-induced expression of catabolic factors, such as Mmp13, which play a crucial role in OA cartilage destruction (29).

DISCUSSION

We observed up-regulation of Dkk1 and down-regulation of Dkk2 in human and experimental mouse OA cartilage. An evaluation of the specific functions of Dkk-1 and Dkk-2 in OA pathogenesis showed that overexpression of Dkk-1 in chondrocytes of articular cartilage, either through intraarticular injection of AdDkk-1 or in the context of chondrocyte-specific Dkk1-transgenic mice, exerted a protective effect against OA pathogenesis including articular cartilage destruction, osteophyte formation, and subchondral bone sclerosis. However, although Dkk-2 expression was significantly decreased in OA cartilage, Col2a1-Dkk2–transgenic mice showed no significant difference in OA cartilage destruction compared to their WT littermates. Our results additionally suggest that the inhibitory effects of Dkk-1 are associated with its capacity to inhibit Wnt-mediated expression of catabolic factors, such as Mmp13, in articular chondrocytes.

Dkk-1 has been suggested to play a role in OA pathogenesis. However, its function in OA cartilage destruction is a subject of controversy. For instance, a cohort study suggested a positive correlation between serum levels of Dkk-1 and reduced progression of radiographic hip OA (15, 16). In addition, investigators in another study reported that circulating Dkk-1 levels were inversely correlated with radiographic OA severity (17). Meanwhile, it has also been reported that increased Dkk-1 expression in OA cartilage leads to cartilage destruction (18, 19). Given the avascular nature of articular cartilage, however, circulating levels of Dkk-1 may not directly reflect biochemical events that occur in chondrocytes of articular cartilage during degeneration. We therefore examined Dkk1 and Dkk2 expression in chondrocytes, in human OA cartilage, and cartilage in various experimental mouse models of OA. We observed differential regulation of Dkk1 and Dkk2 expression in degenerating cartilage tissue, and we found that up-regulation of Dkk1 and down-regulation of Dkk2 occurred after the onset of cartilage destruction. The increase in Dkk1 expression in OA cartilage is consistent with previous reports (18, 19), whereas the down-regulation of Dkk2 has not been previously reported.

Although it is known that Dkk1 expression is increased in OA cartilage (18, 19), its function in OA development and progression has remained incompletely understood. Recently, Weng et al (19) reported that down-regulation of Dkk1 by intraperitoneal administration of Dkk1 antisense oligonucleotide ameliorated cartilage destruction and bone erosion in OA knee joints of rats. They induced OA by anterior cruciate ligament transection or collagenase injection, which cause more severe OA than DMM surgery (25). Additionally, intraperitoneal injection of anti–Dkk-1 antibody reverses the overall bone-destructive pattern of a mouse model of rheumatoid arthritis to the bone-forming pattern of OA (34). Therefore, it is likely that systemic inhibition of Dkk-1 by intraperitoneal administration of either Dkk1 antisense oligonucleotide or anti–Dkk-1 antibody has inhibitory effects on cartilage destruction and bone erosion in arthritic joints. However, those studies did not directly target the roles of Dkk-1 in chondrocytes of articular cartilage. Indeed, the function of Dkk-1 in OA pathogenesis has never been specifically evaluated using cartilage-specific Dkk1-transgenic mice.

To evaluate Dkk-1 function in OA pathogenesis, we overexpressed Dkk-1 directly in chondrocytes by generating chondrocyte-specific Dkk1-transgenic mice. Notably, we found that the transgenic mice were resistant to experimental OA, whether caused by DMM surgery or intraarticular injection of AdEPAS-1. To the best of our knowledge, this study is the first to illustrate that chondrocyte-specific overexpression of Dkk-1 protects against experimental OA cartilage destruction. Chondrocyte-specific Dkk1- and Dkk2-transgenic mice show normal cartilage and bone development (35). Compared with their WT littermates, Col2a1-Dkk1–transgenic mice and Col2a1-Dkk2–transgenic mice showed no differences in the overall patterns of cartilage and skeletal development. Body weight and size were also similar in chondrocyte-specific postnatal transgenic mice and their WT littermates. Thus, it is likely that the protective effects on experimental OA cartilage of Dkk-1 overexpression in Col2a1-Dkk1–transgenic mice are not due to any defects in cartilage and bone development in this mouse model.

We additionally demonstrated that overexpression of Dkk-1 in chondrocytes by intraarticular injection of AdDkk-1 inhibits experimental OA. We routinely observed effective infection of articular cartilage by adenoviral genes through intraarticular injection of adenovirus (21, 28, 36). Although intraarticular injection of AdDkk-1 causes overexpression of Dkk-1 not only in articular cartilage, but also in other joint tissues, such as synovium, ligament, and meniscus, Dkk-1 overexpression in articular cartilage appears to make the main contribution to the protective effects of Dkk-1 against OA pathogenesis. This conclusion is based on the observations that AdDkk-1 does not cause apparent synovitis and that cartilage-specific overexpression of Dkk-1 in the context of Col2a1-Dkk1–transgenic mice similarly shows protective effects on experimental OA. Thus, our results clearly indicate that overexpression of Dkk-1 in chondrocytes is sufficient to exert the protective effects of Dkk-1 in experimental OA.

Because Dkk-1 inhibits the canonical Wnt pathway, our results are consistent with the observation that targeted deletion of the Frzb gene in mice (Frzb−/−), which results in activation of the canonical Wnt pathway, leads to up-regulation of Mmp3 and causes more severe OA cartilage destruction in response to instability, enzymatic injury, or inflammation (10). Moreover, we have observed that Dkk1 expression is high in undifferentiated mesenchymal cells and is markedly decreased during in vitro chondrocyte differentiation of micromass cultures of mesenchymal cells (35). Thus, we speculate that Dkk1 expression levels are low in differentiated chondrocytes in cartilage tissue and increase under pathogenic conditions, such as OA, to protect against cartilage degeneration.

Various Wnt molecules, including Wnt-16 and Wnt-2B, are up-regulated in OA cartilage (37, 38). In the current study, we used Wnt-3a to evaluate the function of Dkk-1 in chondrocytes, although we did not identify which Wnt molecules were specifically up-regulated in OA cartilage in our experimental system. Wnt-3a activates the canonical Wnt pathway and stimulates the expression of Mmp3, Mmp13, Adamts4, and Adamts5 in rabbit chondrocytes (39, 40). We found that stimulation of the canonical Wnt pathway in mouse articular chondrocytes by treatment with recombinant Wnt-3a caused up-regulation of catabolic factors Mmp13 and Adamts4. Additionally, we demonstrated that recombinant Dkk-1 or infection with AdDkk-1 inhibited Wnt-3a–induced expression of Mmp13 and Adamts4. MMP-13 plays an essential role in OA pathogenesis, as Mmp13-deficient mice are resistant to OA cartilage erosion (29). Although ADAMTS-4 appears to be associated with OA pathogenesis (33), Adamts4-knockout mice do not appear to have protection against OA cartilage destruction (41). Therefore, the capacity of Dkk-1 to inhibit Wnt-induced Mmp13 expression appears to be associated with its protective effects against OA cartilage destruction.

Although an association of Dkk-1 with OA pathogenesis has been suggested, the roles of Dkk-2, Dkk-3, and Dkk-4 in OA pathogenesis are unknown. In this study, we found that Dkk-2 expression was significantly decreased during OA cartilage destruction. However, experimental OA was unchanged in Col2a1-Dkk2–transgenic mice, indicating that overexpression of Dkk-2 alone in chondrocytes does not modulate the pathogenesis of OA. Whereas Dkk-1 acts as a pure inhibitor of the canonical Wnt pathway, the capacity of Dkk-2 to inhibit the canonical Wnt pathway depends on cellular context (11, 13, 14). We observed in the present study that recombinant Dkk-2 protein does not modulate Wnt-3a–induced catabolic factor expression in articular chondrocytes, suggesting that Dkk-2 is not involved in the regulation of the canonical Wnt pathway. This is consistent with our finding that Dkk-2 overexpression in Col2a1-Dkk2–transgenic mice has no significant effects on experimental OA.

In summary, our results suggest that overexpression of Dkk-1, but not Dkk-2, in chondrocytes of articular cartilage inhibits OA pathogenesis, such as articular cartilage destruction, osteophyte formation, and subchondral bone sclerosis. Although systemic inhibition of Dkk-1 functions by intraperitoneal administration of Dkk1 antisense oligonucleotide or anti–Dkk-1 antibody could result in the inhibition of cartilage destruction and bone erosion in arthritic joints (19, 34), our results clearly indicate that local overexpression of Dkk-1 in cartilage tissue protects against OA pathogenesis.

AUTHOR CONTRIBUTIONS

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. Jang-Soo Chun 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. Oh, J.-S. Chun.

Acquisition of data. Oh, C.-H. Chun, J.-S. Chun.

Analysis and interpretation of data. Oh, J.-S. Chun.

Ancillary