Dissection of Wnt5a-Ror2 Signaling Leading to Matrix Metalloproteinase (MMP-13) Expression*

Background: Constitutively active Wnt5a-Ror2 signaling in osteosarcoma cells induces MMP-13 expression. Results: MMP-13 is expressed by binding of c-Jun and ATF2 to AP1-binding sites within its promoter during Wnt5a-Ror2 signaling. Conclusion: MMP-13 expression by Wnt5a-Ror2 signaling is mediated by Dvl, Rac1, JNK, and AP1 transcription factors. Significance: This provides novel molecular machineries of Wnt5a-Ror2 signaling, involved in MMP-13 expression. It has been shown that constitutively active Wnt5a-Ror2 signaling in osteosarcoma cell lines plays crucial roles in induced expression of matrix metalloproteinase-13 (MMP-13), required for their invasiveness; however, it remains largely unclear about the molecular basis of MMP-13 gene induction by Wnt5a-Ror2 signaling. Here we show by reporter assay that the activator protein 1 (AP1) (binding site in the promoter region of MMP-13 gene is primarily responsible for its transcriptional activation by Wnt5a-Ror2 signaling in osteosarcoma cell lines SaOS-2 and U2OS. Chromatin immunoprecipitation assays revealed that c-Jun and ATF2 are crucial transcription factors recruited to the AP1-binding site in the MMP-13 gene promoter during Wnt5a-Ror2 signaling in SaOS-2 cells. Using siRNA-mediated suppression or specific inhibitors, we also show that Dishevelled2 (Dvl2) and c-Jun N-terminal kinase are required for MMP-13 gene induction presumably via phosphorylation of c-Jun and ATF2 during Wnt5a-Ror2 signaling in SaOS-2 cells. Interestingly, Dvl2 and Rac1, but not Dvl3, are required for MMP-13 expression in SaOS-2 cells, whereas Dvl3, but not Dvl2 and Rac1, is required for its expression in U2OS cells, indicating the presence of distinct intracellular signaling machineries leading to expression of the same gene, in this case MMP-13 gene in different osteosarcoma cell lines. Moreover, we provide evidence suggesting that Wnt5a-Ror2 signaling might also be required for expression of MMP-13 gene during the development of the cartilaginous tissue.

It has been shown that constitutively active Wnt5a-Ror2 signaling in osteosarcoma cell lines plays crucial roles in induced expression of matrix metalloproteinase-13 (MMP-13), required for their invasiveness; however, it remains largely unclear about the molecular basis of MMP-13 gene induction by Wnt5a-Ror2 signaling. Here we show by reporter assay that the activator protein 1 (AP1) (binding site in the promoter region of MMP-13 gene is primarily responsible for its transcriptional activation by Wnt5a-Ror2 signaling in osteosarcoma cell lines SaOS-2 and U2OS. Chromatin immunoprecipitation assays revealed that c-Jun and ATF2 are crucial transcription factors recruited to the AP1-binding site in the MMP-13 gene promoter during Wnt5a-Ror2 signaling in SaOS-2 cells. Using siRNA-mediated suppression or specific inhibitors, we also show that Dishevelled2 (Dvl2) and c-Jun N-terminal kinase are required for MMP-13 gene induction presumably via phosphorylation of c-Jun and ATF2 during Wnt5a-Ror2 signaling in SaOS-2 cells. Interestingly, Dvl2 and Rac1, but not Dvl3, are required for MMP-13 expression in SaOS-2 cells, whereas Dvl3, but not Dvl2 and Rac1, is required for its expression in U2OS cells, indicating the presence of distinct intracellular signaling machineries leading to expression of the same gene, in this case MMP-13 gene in different osteosarcoma cell lines. Moreover, we provide evidence suggesting that Wnt5a-Ror2 signaling might also be required for expression of MMP-13 gene during the development of the cartilaginous tissue.
The Wnt family of proteins are secreted cysteine-rich glycoproteins with lipid modifications consisting of 19 members in human to date (1). Several lines of evidence demonstrate that Wnt proteins play important roles in developmental, physiological, and/or pathological processes (2). Wnt proteins can elicit a ␤-catenin-dependent (canonical) and ␤-catenin-independent (non-canonical) signaling pathways (3). Wnt5a is a representative of Wnt members that can activate non-canonical Wnt signaling. It has been shown that the sustained or increased expression of Wnt5a is implicated in various types of cancers (4 -7) and that Wnt5a promotes cell migration and invasiveness on osteosarcoma cells or prostate carcinoma cells (8,9).
Ror2, a member of the Ror-family of receptor-tyrosine kinase s, acts as a receptor or co-receptor for Wnt5a to mediate Wnt5a-induced various cellular functions, including cell migration, polarity, or invasion (8,10,11). Both Wnt5a-and Ror2-deficient mouse embryos exhibit similar abnormalities in various tissues and organs, including cardiovascular, inner ear, and skeletal systems that are caused at least partly by disrupted convergent extension movements and planar cell polarity during development (12,13). Ror2 mediates Wnt5a signaling by inhibiting the ␤-catenin-T-cell factor pathway and activating Wnt/c-Jun N-terminal kinase (JNK) pathways (3,12). The Wnt/JNK pathway can be mediated by various signaling molecules, including Dishevelled (Dvl), 2 activator protein 1 (AP1), and members of the Rho-family of small GTPases in addition to JNK (12); however, it remains elusive how this pathway regulates expression of its target genes.
Recently, it has been reported that Ror2 regulates the expression of genes, encoding various matrix metalloproteinases (MMPs), which belong to a family of zinc-dependent enzymes capable of degrading extracellular matrix, including basement membrane components (8,14,15). Among MMPs, MMP-13 exhibits unique characteristics, i.e. it cleaves not only type IIcollagen but also macromolecules of the extracellular matrix such as a fibrinogen. Spatiotemporally regulated expression of MMP-13 is required for proper matrix remodeling under physiological conditions such as development of cartilaginous tissue (16). In contrast, sustained overexpression of MMP-13 appears to contribute to cartilage destruction in rheumatoid arthritis or to progression of various malignant tumors, including breast cancers, chondrosarcomas, melanomas, esophageal cancers, colorectal cancers, and osteosarcomas (8,(17)(18)(19)(20)(21)(22)(23)(24)(25). In fact it has been shown that constitutive overexpression of MMP-13 is critically required for cell invasiveness of osteosarcoma in a cell-autonomous manner and that expression of MMP-1 or MMP-2 is involved in progression of several other types of cancer cells (8,9,14,15). These findings indicate that cell type-dependent expression of members of MMP-family is implicated in some if not all of cancer cells (26).
In the present study we show that in addition to c-Src, Dvl2, Rac1, and JNK are required for MMP-13 gene expression by Wnt5a-Ror2 signaling via binding of c-Jun and ATF2 to AP1binding site within the MMP-13 gene promoter in SaOS-2 cells. On the other hand, MMP-13 gene expression by Wnt5a-Ror2 signaling in U2OS cells requires Dvl3, but not Dvl2 and Rac1, indicating that distinct intracellular signaling events might play a role in regulating MMP-13 gene expression in different osteosarcoma cell lines.

EXPERIMENTAL PROCEDURES
Animals-The experiments using animals in this study were approved by the Institutional Animal Care and Use Committee (permission number P110307) and carried out according to the Kobe University Animal Experimentation Regulations.
Luciferase Reporter Assay-SaOS-2 or U2OS cells were transfected with the respective siRNAs and cultured for 4 days.
ChIP-ChIP was performed using ChIP assay kit (Millipore) as previously described with minor modifications (36). SaOS-2 cells were harvested after cross-linking with 1% formaldehyde for 5 min. Chromatin was sheared by sonication to an average length of 500 -600 bp followed by immunoprecipitation with control (isotype-matched), anti-c-Jun, anti-ATF2, or anti-c-Fos antibodies. ChIP were then washed and eluted. After reversal of cross-linking, the precipitated DNA pellets were subjected to PCR analyses using specific primers for amplification of the promoter region of MMP-13 gene. Forward and reverse primers were as follows: proximal region containing AP1-binding site in MMP-13 gene promoter, 5Ј-TTTGCCAGATGGG-TTTTGAGACCC-3Ј and 5Ј-TTGAATGGTGATGCCTGGG-GACTG-3Ј; distal region lacking the element in MMP-13 gene promoter, 5Ј-TAGGCTGTCCTCAAACTTCACCCC-3Ј and 5Ј-CTCTCTGATGGTTGCTTAGTTCTACC-3Ј.

RESULTS
Constitutively Active Wnt5a-Ror2 Signaling Is Critical for Transcriptional Activation of MMP-13 Gene Promoter via AP1binding Site in SaOS-2 Cells-As an attempt to elucidate the molecular basis of MMP-13 gene expression by Wnt5a-Ror2 signaling, we first performed reporter assays to identify a region within the MMP-13 gene promoter required for transcriptional activation by Wnt5a-Ror2 signaling in osteosarcoma cell line SaOS-2. Because Wnt5a-Ror2 signaling has been shown to be constitutively activated by sustained expression of both Wnt5a and Ror2 in SaOS-2 cells (8), we examined the effects of siRNA-mediated suppression of Wnt5a or Ror2 on transcriptional activation of MMP-13 gene in the cells. SaOS-2 cells transfected with Control (Ctrl) #1, Ror2 #1, or Ror2 #2 siRNA with different sequences for Ror2 gene were cultured for 4 days, and WCLs were subjected to immunoblotting with the indicated antibodies. As shown in Fig. 1A, suppressed expression of Ror2 was detected in SaOS-2 cells transfected with Ror2 #1 or Ror2 #2 siRNA compared with Ctrl #1 siRNA. SaOS-2 cells transfected with the respective siRNAs were further transfected with the indicated MMP-13 gene promoter-luciferase (luc) constructs (p3105-luc, p1004-luc, p253-luc, and basic-luc) along with the renilla luciferase construct as described under "Experimental Procedures." Reporter assays with p3105-luc, p1004-luc, or p253-luc revealed that decreased levels of relative luciferase activities were detected in SaOS-2 cells transfected with Ror2 #1 or Ror2 #2 siRNA compared with Ctrl #1 siRNA (Fig. 1B). The basic-luc activity was not affected by suppressed expression of Ror2 (Fig. 1B). Because the identified MMP-13 promoter region (bp Ϫ253 to Ϫ1) contains two well characterized cis-regulatory elements, the OSE2-and AP1-binding sites, we next examined which element was responsible for MMP-13 gene expression by Ror2-mediated signaling. To this end point mutations were introduced into OSE2-and AP1-binding sites in p253-luc (253(OSE2m)-luc and p253(AP1m)-luc, respectively) (Fig. 1C). SaOS-2 cells transfected with the respective siRNAs were further transfected with p253-luc, p253(OSE2m)luc, p253(AP1m)-luc, or basic-luc along with the renilla luciferase construct and cultured for 24 h. Transfection with p253(AP1m)-luc, but not p253(OSE2m)-luc, resulted in a drastic decrease of luciferase activities that were unaffected by Ror2-mediated signaling (Fig. 1C). Because Ror2 is known to act as a receptor or co-receptor for Wnt5a, we also examined the effect of suppressed expression of Wnt5a on MMP-13 gene expression in SaOS-2 cells. SaOS-2 cells transfected with Ctrl #2, Wnt5a #1, or Wnt5a #2 siRNA were cultured for 4 days, and total RNAs from the respective cells were subjected to qRT-PCR to assess suppressed expression of Wnt5a by Wnt5a siRNAs (Fig. 1D). SaOS-2 cells transfected with the respective siRNAs were further transfected with p3105-luc, p1004-luc, p253-luc, and basic-luc along with the renilla luciferase construct. Like Ror2 knockdown experiments (see Fig. 1B), reporter constructs that contain the promoter region spanning bp Ϫ253 to Ϫ1, but not basal-luc, show a drastic reduction of luciferase activities after suppressed expression of Wnt5a (Fig.  1E), indicating that the proximal promoter region (bp Ϫ253 to Ϫ1), containing AP1-binding sites, is required for MMP-13 gene expression by Wnt5a-Ror2 signaling. To further assess the importance of the proximal promoter region in MMP-13 gene expression by Wnt5a-Ror2 signaling, SaOS-2 cells were serumstarved in a prolonged culture to partly reduce the basal activity of Wnt5a-Ror2 signaling (8) and subjected to luciferase reporter assay after Wnt5a stimulation. As shown in Fig. 1F, Wnt5a stimulation resulted in a significant increase of lucifer-ase activities on p1004-luc and p253-luc, but not basal-luc, further emphasizing that the proximal promoter region of MMP-13 gene is responsive to Wnt5a-Ror2 signaling in SaOS-2 cells.
Because we have previously shown that MMP-13 gene is also induced by Wnt5a-Ror2 signaling in another osteosarcoma cell line U2OS (8), we next examined whether the same proximal promoter region is required for MMP-13 gene expression by Wnt5a-Ror2 signaling. To this end U2OS cells transfected with Ctrl #1, Ror2 #1, or Ror2 #2 siRNA were cultured for 4 days, and expression levels of Ror2 protein and MMP-13 transcript were assessed by anti-Ror2 immunoblotting and qRT-PCR, respectively. As shown in Fig. 2A, transfection of U2OS cells with Ror2 siRNAs resulted in a drastic inhibition of expression of Ror2 protein and MMP-13 transcript, respectively. U2OS cells transfected with Ctrl #1, Ror2 #1, or Ror2 #2 siRNA were further Then cells were transfected with the indicated MMP-13 gene promoter-luc vectors, whose promoters contain mutations within the OSE2-and AP1-binding sites, respectively, along with the renilla luciferase reporter construct and were cultured for 24 h. Reporter assays were performed as described under "Experimental Procedures." The data are representative of three independent experiments. The bars represent the mean Ϯ SD of triplicate experiments. *, p Ͻ 0.01, t test. D, SaOS-2 cells were transfected with Ctrl #2, Wnt5a #1, or Wnt5a #2 siRNA and cultured for 4 days. Total RNAs from the respective cells were then isolated and subjected to qRT-PCR to monitor mRNA expression of either Wnt5a or GAPDH. Relative amounts of Wnt5a transcripts were determined after normalization by those of GAPDH transcripts. *, p Ͻ 0.01, t test. E, SaOS-2 cells were transfected with Ctrl #2, Wnt5a #1, or Wnt5a #2 siRNA and cultured for 4 days. Subsequently, SaOS-2 cells were transfected with the indicated MMP-13 gene promoter-luciferase (Luc) vectors along with the renilla luciferase reporter construct and were cultured for 24 h. Reporter assays were performed as described under "Experimental Procedures." The data are representative of three independent experiments. The bars represent the mean Ϯ SD of triplicate experiments. *, p Ͻ 0.01, t test. F, SaOS-2 cells were transfected with the respective MMP-13 gene promoter-luc vectors for 12 h. The cells were further serum-starved for 12 h and then treated with either Wnt5a (400 ng/ml) or vehicle for 12 h. Reporter assays were performed as described under "Experimental Procedures." The data are representative of three independent experiments. The bars represent the mean Ϯ SD of triplicate experiments. *, p Ͻ 0.01, t test.
transfected with p1004-luc, p253-luc, and basic-luc along with the renilla luciferase reporter construct as described under "Experimental Procedures." Similar to SaOS-2 cells, reporter constructs that contain the proximal promoter region (bp Ϫ253 to Ϫ1), but not basal-luc, show a drastic reduction of luciferase activities after suppressed expression of Ror2 (Fig.  2B). We next examined which element was responsible for MMP-13 gene expression by Ror2-mediated signaling. SaOS-2 cells transfected with the respective siRNAs were further transfected with p253-luc, p253(OSE2m)-luc, p253(AP1m)-luc, or basic-luc along with the renilla luciferase construct and cultured for 24 h. Transfection with p253(AP1m)-luc, but not p253(OSE2m)-luc, resulted in a drastic decrease of luciferase activities that were by Ror2-mediated signaling (Fig. 2C). U2OS cells were also serum-starved in a prolonged culture and were subjected to luciferase reporter assay. As expected, Wnt5a stimulation of U2OS cells also resulted in a significant increase of luciferase activities that could be remarkably inhibited by deletion of the proximal promoter region (bp Ϫ253 to Ϫ1) (Fig.  2D). Taken together, the results suggest, like SaOS-2 cells, the proximal promoter region of MMP-13 gene is responsive to Wnt5a-Ror2 signaling in U2OS cells.

AP1 Components c-Jun and ATF2 Are Recruited to MMP-13
Gene Promoter in Wnt5a-dependent Manner-We next examined which members of the AP1-family of transcription factors are involved in MMP-13 gene expression by Wnt5a-Ror2 signaling via their binding to the proximal promoter region containing the AP1-binding site. To this end SaOS-2 cells transfected with Ctrl #2, Wnt5a #1, or Wnt5a #2 siRNA and cultured for 4 days were subjected to a ChIP assay using the indicated antibodies as described under "Experimental Procedures." The positions of the respective primers used in ChIP assays are shown in Fig. 3A. c-Jun, c-Fos, and ATF2 were recruited to the proximal region (bp Ϫ212 to Ϫ27) containing the AP1-binding site but not to distal region (bp Ϫ3104 to Ϫ2919) within the MMP-13 gene in cells transfected with Ctrl #2 siRNA (Fig. 3B). Interestingly, recruitment of c-Fos, but not c-Jun and ATF2, to the proximal promoter region was detected after suppressed expression of Wnt5a (Fig. 3B), suggesting that c-Jun and ATF2 were recruited to the proximal promoter region of MMP-13 gene by Wnt5a-Ror2 signaling. Although c-Fos was recruited to the proximal promoter region of MMP-13 gene, its recruitment was marginally affected by Wnt5a-Ror2 signaling. Furthermore, phosphorylation of both c-Jun and ATF2 detected in  (Fig. 3C). These results suggest that constitutively active Wnt5a-Ror2 signaling in SaOS-2 cells can induce phosphorylation of c-Jun and ATF2, thereby promoting their binding to the proximal promoter region of MMP-13 gene presumably via AP1-binding site, leading to expression of MMP-13 gene.

SaOS-2 cells was inhibited by suppressed expression of Wnt5a
We further tested the effect of Wnt5a stimulation on the recruitment of AP1 components to the MMP-13 gene promoter in SaOS-2 cells after their serum starvation. For this purpose, serum-starved SaOS-2 cells were stimulated with either Wnt5a or vehicle alone for 1 h, and WCLs were prepared and subjected to immunoblotting with anti-Dvl2 antibody. As shown in Fig. 3D, Wnt5a stimulation induced phosphorylation of Dvl2, as assessed by its electrophoretic mobility shift, in serum-starved SaOS-2 cells. Serum-starved SaOS-2 cells treated with either Wnt5a or vehicle alone were also subjected to ChIP analysis with the indicated antibodies as described under "Experimental Procedures." After Wnt5a stimulation, c-Jun and ATF2, but not c-Fos, were recruited to the proximal promoter region (bp Ϫ212 to Ϫ27) of MMP-13 gene (Fig. 3E). On the other hand, after Wnt5a stimulation, ATF2, but not c-Jun and c-Fos, was recruited to the distal promoter region (bp Ϫ3104 to Ϫ2919) of MMP-13 gene (Fig. 3E); however, the role of its recruitment is unclear. We also observed phosphorylation of both c-Jun and ATF2 after Wnt5a stimulation of serumstarved SaOS-2 cells (Fig. 3F), further emphasizing the critical role of c-Jun and ATF2 in MMP-13 gene expression by Wnt5a-Ror2 signaling in SaOS-2 cells.

JNK Is Critical for Expression of MMP-13 Gene by Wnt5a-Ror2
Signaling in SaOS-2 Cells-To further confirm the role of c-Jun and ATF2, in particular their phosphorylation, in MMP-13 gene expression by Wnt5a-Ror2 signaling in SaOS-2 cells, we examined the effect of suppressed expression of Wnt5a and Ror2 on expression of MMP-13 and phosphorylation of c-Jun and ATF2 by qRT-PCR and immunoblotting with phospho-specific antibodies, respectively. Consistent with the results of luciferase reporter assay (see Fig. 1, D and E), suppressed expression of Wnt5a in SaOS-2 cells resulted in a drastic decrease of MMP-13 transcripts as assessed by qRT-PCR (Fig. 4A). Immunoblot analysis revealed that suppressed expression of Wnt5a also resulted in a drastic inhibition of phosphorylation of c-Jun and ATF2 without affecting the total amounts of c-Jun and ATF2, respectively (Fig. 4A). Similarly, suppressed expression of Ror2 resulted in a remarkable decrease of MMP-13 transcripts and in a remarkable inhibition of phosphorylation of both c-Jun and ATF2 (Fig. 4B). Correlation between the levels of MMP-13 expression and of phosphorylation of c-Jun and ATF2 suggests that phosphorylation of c-Jun and ATF2 might be required for MMP-13 gene expression by Wnt5a-Ror2 signaling in SaOS-2 cells.
It has been shown that JNK is activated by Wnt5a-Ror2 signaling and that JNK is involved in phosphorylation of c-Jun and/or ATF2 (37). Thus, we examined a possible involvement of JNK in phosphorylation of c-Jun and/or ATF2 induced by Wnt5a-Ror2 signaling in SaOS-2 cells. To this end, SaOS-2 cells were treated with either SP600125 (an inhibitor of JNK, 20 M) or vehicle alone for 10 h, and expression levels of MMP-13 and phosphorylation status of c-Jun and ATF2 were monitored by qRT-PCR and immunoblotting with phospho-specific antibodies, respectively. As shown in Fig. 4C, MMP-13 expression and phosphorylation of c-Jun and ATF2 in SaOS-2 cells were remarkably inhibited by SP600125, indicating that JNK is responsible for phosphorylation of c-Jun and ATF2 and for possible subsequent induction of MMP-13 gene by Wnt5a-Ror2 signaling. Consistent with the results, ChIP analysis revealed that treatment of SaOS-2 cells with SP600125 resulted in failure to recruit c-Jun and ATF2, but not c-Fos, to the proximal promoter region, but not to the distal region, within the MMP-13 gene promoter (supplemental Fig. S1). To further clarify the role of JNK in MMP-13 gene expression, SaOS-2 cells transfected with the respective MMP-13 gene promoter-luc vectors (p1004-luc, p253-luc, and basic-luc) were cultured for 12 h. Subsequently, the cells were serum-starved for 12 h and further treated with either SP600125 (20 M) or vehicle for 12 h. As shown in supplemental Fig. S2, inhibition of JNK resulted in a significant decrease in relative luciferase activities in SaOS-2 cells. Taken together, these results indicate that JNK plays a critical role in the expression of MMP-13 gene by Wnt5a-Ror2 signaling via phosphorylation and subsequent recruitment of c-Jun and ATF2 to its proximal promoter region in SaOS-2 cells.
Dvl2, but Not Dvl3, Is Required for Expression of MMP-13 Gene via Phosphorylation of Both c-Jun and ATF2 in SaOS-2 Cells-Because we recently showed that Dvl2 and Rac1 are involved in Wnt5a-induced AP1-luc activation in L cells expressing Ror2 (35), we next addressed a question of whether Dvl proteins (Dvl2 and Dvl3) and Rac1 are also involved in MMP-13 gene expression by Wnt5a-Ror2 signaling in SaOS-2 cells. To this end, SaOS-2 cells transfected with Ctrl #1, Dvl2 #1, Dvl2 #2, Dvl3 #1, or Dvl3 #2 siRNA were cultured for 4 days, and WCLs and total RNAs prepared from the respective transfectants were subjected to immunoblotting with anti-Dvl2 and Dvl3 antibodies and qRT-PCR, respectively. Interestingly, suppressed expression of Dvl2, but not Dvl3, resulted in a remarkable inhibition of MMP-13 gene expression in SaOS-2 cells (Fig.  5A). We also performed a similar experiment to examine whether Dvl2 is indeed required for phosphorylation of c-Jun and ATF2 presumably via JNK by Wnt5a-Ror2 signaling in SaOS-2 cells. As expected, suppressed expression of Dvl2, but not Dvl3, resulted in a significant inhibition of phosphorylation of both c-Jun and ATF2 (Fig. 5, B and C), indicating that Dvl2, but not Dvl3, plays a critical role in JNK-mediated phosphorylation of c-Jun and ATF2, leading to expression of MMP-13 gene in SaOS-2 cells. It has been shown that the Dvl2 mutant, Dvl2(KM), bearing an amino acid substitution (K446M) within the DEP domain (corresponding to K417M mutant of Drosophila Dsh that is defective in activating the Wnt/ planar cell polarity pathway but functional for the ␤-catenin-dependent signaling pathway) fails to mediate the Wnt5a-induced AP1 activation in the Dvl2-knockdown L cells expressing Ror2 (35). Thus, we examined whether Dvl2(KM) can restore the failure of Dvl2-knockdown SaOS-2 cells to induce MMP-13 gene expression. As shown in Fig. 5D, siRNA-resistant Dvl2(WT), but not Dvl2(KM), could restore MMP-13 gene expression by Wnt5a-Ror2 signaling in the Dvl2-knockdown SaOS-2 cells, suggesting that Dvl2 function, mediating the Wnt/ planar cell polarity pathway, is required for MMP-13 gene expression by Wnt5a-Ror2 signaling.
Because Dvl2 is co-localized with Rac1, a member of the Rho-family of small GTPases, after stimulation of serum- #1, or Wnt5a #2 siRNA. After 4 days in culture, total RNAs from the respective transfected cells were isolated and subjected to qRT-PCR to monitor mRNA expression of either MMP-13 or GAPDH. WCLs were also prepared and subjected to immunoblotting with the indicated antibodies. *, p Ͻ 0.01, t test. B, SaOS-2 cells were transfected with Ctrl #1, Ror2 #1, or Ror2 #2 siRNA. After 4 days in culture, total RNAs from the respective transfected cells were isolated and subjected to qRT-PCR to monitor mRNA expression of either MMP-13 or GAPDH. WCLs were also prepared and subjected to immunoblotting with the indicated antibodies. *, p Ͻ 0.01, t test. C, SaOS-2 cells were treated with vehicle or SP600125 (an inhibitor of JNK, 20 M) for 10 h. Total RNAs from the respective cells were isolated and subjected to qRT-PCR to monitor mRNA expression of either MMP-13 or GAPDH. WCLs were also prepared and subjected to immunoblotting with the indicated antibodies. *, p Ͻ 0.01, t test.
starved SaOS-2 cells with Wnt5a (data not shown) (35), we next addressed the question of whether Rac1 is also involved in MMP-13 gene expression by Wnt5a-Ror2 signaling in SaOS-2 cells. Suppressed expression of Rac1 resulted in significant inhibition of MMP-13 gene expression in SaOS-2 cells as assessed by qRT-PCR analysis (Fig. 5E). It should be noted that Dvl2 phosphorylation by Wnt5a-Ror2 signaling in SaOS-2 cells was unaffected by suppressed expression of Rac1 (Fig. 5E), suggesting that Rac1 might act downstream of Dvl2 in Wnt5a-Ror2 signaling in SaOS-2 cells. It has been shown that an Src-family protein-tyrosine kinase (SFK) is involved in MMP-13 gene expression by Wnt5a-Ror2 signaling in SaOS-2 cells (8). Thus, we also examined whether a SFK acts upstream or downstream of Dvl2 in Wnt5a-Ror2 signaling, leading to MMP-13 gene expression, in SaOS-2 cells. For this purpose SaOS-2 cells were serum-starved in the presence of either PP2 (an inhibitor of SFKs, 2 M) or PP3 (a relevant control compound, 2 M) for 12 h and then treated with either Wnt5a or vehicle alone for 1 h. WCLs and total RNAs from the respective cells were subjected to immunoblotting with anti-Dvl2 antibody and qRT-PCR, respectively. As expected, treatment with PP2, but not PP3, significantly inhibited Wnt5a-induced expression of MMP-13 gene (supplemental Fig. S3). Importantly, Dvl2 phosphorylation induced by Wnt5a stimulation was abrogated by treatment with PP2, but not PP3, suggesting that a SFK might act upstream of Dvl2. Collectively, these results indicate that Dvl2, but not Dvl3, is required for MMP-13 gene expression in SaOS-2 cells and that a SFK-Dvl2-Rac1-JNK pathway might play an important role in MMP- The histograms indicate the relative extents of phosphorylated forms of c-Jun or ATF2 (phosphorylated c-Jun or ATF2 levels normalized by total c-Jun or ATF2 levels, respectively). Quantifications were performed as described under "Experimental Procedures." The data are representative of three independent experiments. The bars represent the mean Ϯ SD of triplicate experiments. D, SaOS-2 cells were transfected with either Ctrl #2 or Dvl2 #3 siRNA for 4 days in culture, and then Dvl2 #3 siRNA-treated cells were further transfected with either siRNA-resistant Dvl2 WT or KM (K446M) mutant expression vector and cultured for 24 h. Total RNAs from the respective transfected cells were isolated and subjected to qRT-PCR to monitor mRNA expression of either MMP-13 or GAPDH. WCLs were also prepared and subjected to immunoblotting with the indicated antibodies. E, SaOS-2 cells were transfected with Ctrl #2, Rac1 #1, or Rac1 #2 siRNA. After 4 days in culture, total RNAs from the respective transfected cells were isolated and subjected to qRT-PCR to monitor mRNA expression of either MMP-13 or GAPDH. WCLs were also prepared and subjected to immunoblotting with the indicated antibodies. Note that the asterisk (*) and double asterisks (**) indicate hypophosphorylated and hyperphosphorylated forms of Dvl2, respectively. 13 gene expression by Wnt5a-Ror2 signaling in SaOS-2 cells.
Because, like SaOS-2 cells, the same proximal promoter region of MMP-13 gene is required for MMP-13 gene expression by Wnt5a-Ror2 signaling in another osteosarcoma cell line U2OS (see Fig. 2, A-D), we next examined possible roles of Dvl2, Dvl3, and Rac1 in MMP-13 gene expression by Wnt5a-Ror2 signaling in U2OS cells. To this end, U2OS transfected with Ctrl #1, Dvl2 #1, Dvl2 #2, Dvl3 #1, or Dvl3 #2 siRNA were cultured for 4 days, and WCLs and total RNAs from the respective transfected cells were subjected to immunoblotting with anti-Dvl2 and Dvl3 antibodies and qRT-PCR, respectively. Surprisingly, suppressed expression of Dvl3, but not Dvl2, resulted in a drastic inhibition of MMP-13 gene expression in U2OS cells (Fig. 6A), indicating that Dvl3, but not Dvl2, is required for MMP-13 gene expression by Wnt5a-Ror2 signaling in U2OS cells. In addition, unlike SaOS-2 cells, suppressed expression of Rac1 failed to affect MMP-13 gene expression by Wnt5a-Ror2 signaling in U2OS cells (Fig. 6B). Taken together, these results indicate that different signaling molecules (Dvl2 and Rac1 in SaOS-2 and Dvl3 in U2OS cells) are utilized in different osteosarcoma cell lines, i.e. SaOS-2 and U2OS cells, to mediate Wnt5a-Ror2 signaling, leading to expression of the same gene, MMP-13 gene.
Wnt5a-Ror2 Signaling Is Involved in Endochondral Ossification of Cartilaginous Tissues-We next addressed a question of whether Wnt5a-Ror2 signaling is also involved in physiological conditions. For this purpose the paraffin specimens, prepared from the cartilaginous tissues from both Ror2 ϩ/ϩ and Ror2 Ϫ/Ϫ male mouse embryos at E15.5, were fixed and stained with the antibodies against Ror2 and MMP-13 as described under "Experimental Procedures." IHC analysis revealed that endochondral ossification was detected in cartilaginous tissues from male Ror2 ϩ/ϩ , but not Ror2 Ϫ/Ϫ mouse embryos at E15.5 (Fig.  7). Expression of MMP-13 was clearly detected at the ossifica-tion centers of the cartilaginous tissues from Ror2 ϩ/ϩ , but not Ror2 Ϫ/Ϫ mouse embryos (Fig. 7). Furthermore, expression of Ror2 was detectable at the ossification centers from Ror2 ϩ/ϩ , but not Ror2 Ϫ/Ϫ mouse embryos (Fig. 7). These results suggest that Wnt5a-Ror2 signaling might be required for expression of MMP-13 during the development of the cartilaginous tissues.

DISCUSSION
We have previously shown that both Wnt5a and Ror2 are expressed in osteosarcoma SaOS-2 cells at least partly by endogenous expression of Snail in the cells (15) and that as a result Wnt5a-Ror2 signaling is constitutively activated in SaOS-2 cells, leading to the induction of MMP-13, which plays an important role in their invasiveness via activation of a SFK (8). However, it remains largely unknown about the molecular mechanism underlying MMP-13 gene expression by Wnt5a-  Ror2 signaling. Here we show by using reporter assays that the proximal promoter region of MMP-13 gene containing both the OSE2 and AP1-binding site is primarily required for expression of MMP-13 gene in SaOS-2 cells (Fig. 1). In addition, mutagenesis within either the OSE2-or AP1-binding site revealed that the AP1-binding site, but not the OSE2-binding site, is critically required for expression of MMP-13 gene by Wnt5a-Ror2 signaling in SaOS-2 cells (Fig. 1). As reported previously, MMP-13 gene is also expressed by constitutively active Wnt5a-Ror2 signaling in another osteosarcoma cell line U2OS cells (8). Importantly, we also identified the AP1-binding site within the proximal promoter region of MMP-13 gene as the critical responsive element to Wnt5a-Ror2 signaling in U2OS cells (Fig. 2), indicating that AP1-binding site-mediated activation of MMP-13 gene by Wnt5a-Ror2 signaling may be a common feature observed in osteosarcoma cells. Consistent with the results obtained by our reporter assays, ChIP assays also revealed that both c-Jun and ATF2 are recruited to the AP1binding site within the proximal promoter region of MMP-13 gene by Wnt5a-Ror2 signaling in SaOS-2 cells (Fig. 3). With respect to this, it should be noted that expression of MMP-1 or MMP-2, other members of the MMP-family of proteinases, is induced by Ror2 and/or Wnt5a-Ror2 signaling in prostate carcinoma cells or renal cell carcinoma and snail-expressing epidermoid carcinoma cells, respectively (9, 14 -15). Because the promoter regions of MMP-1 and MMP-2 gene also contain the AP1-binding site (9,14,15), it is of interest to identify whether the AP1 family of transcription factors is recruited to the AP1binding sites in these genes by Ror2 and/or Wnt5a-Ror2 signaling.
Our present study using siRNA-mediated knockdown or specific inhibitors for particular signaling molecules identified Dvl2, Rac1, JNK, c-Jun, and ATF2 in addition to a SFK as candidate signaling molecules involved in MMP-13 gene expression by Wnt5a-Ror2 signaling in SaOS-2 cells. Treatment of SaOS-2 cells with a SFK inhibitor (PP2) inhibits phosphorylation of Dvl2 (supplemental Fig. S3), a surrogate marker for Dvlmediated Wnt signaling (35,38), suggesting that an SFK may act upstream of Dvl2 in Wnt5a-Ror2 signaling. Because it has been reported that Dvl proteins can be phosphorylated on serine and threonine residues by protein serine/threonine kinases including casein kinase I (39), it can be assumed that regulation of Dvl2 phosphorylation by a SFK may be indirect. Future study will be required to clarify the molecular mechanism of Dvl2 phosphorylation by a SFK. Furthermore, consistent with our previous observation using L cells expressing Ror2 (35), it was found that Dvl2 is co-localized with Rac1 (data not shown) and Dvl2-Rac1 is required for MMP-13 gene expression presumably via phosphorylation of ATF2 and c-Jun (Fig. 5, B and C), leading to their recruitment to the AP1-binding site within the proximal promoter region of MMP-13 gene in SaOS-2 cells (see Fig. 3). Because suppressed expression of Rac1 failed to inhibit phosphorylation of Dvl2 by Wnt5a-Ror2 signaling (Fig. 5E), it can be envisaged that Dvl2 may act upstream of Rac1 in Wnt5a-Ror2 signaling. In fact, Rac1 has been shown to be involved in JNK activation (40). It has been reported that JNK is responsible for phosphorylation of c-Jun or ATF-2 (37). In fact, treatment of SaOS-2 cells with a JNK inhibitor (SP600125) resulted in the inhibition of phosphorylation of both c-Jun and ATF2 and of their recruitment to the AP1-binding site within the proximal promoter region of MMP-13 gene (Fig. 4C, supplemental Fig.  S1). Taken together, the results indicate that the c-Src-Dvl2/ Rac1-JNK-c-Jun/ATF2 pathway plays a crucial role in expression of MMP-13 gene by Wnt5a-Ror2 signaling in SaOS-2 cells.
We also performed partial dissection of Wnt5a-Ror2 signaling leading to MMP-13 expression in another osteosarcoma U2OS cells. Unlike SaOS-2 cells, suppressed expression of either Dvl2 or Rac1 in U2OS cells failed to inhibit MMP-13 gene expression by Wnt5a-Ror2 signaling (Fig. 6). Interestingly, suppressed expression of Dvl3, a different Dvl isoform, in U2OS cells resulted in drastic inhibition of MMP-13 gene expression (Fig. 6), indicating the presence of distinct intracellular signaling machinery leading to expression of MMP-13 gene in different osteosarcoma cell lines. It is worth noting that Wnt5b, a close relative of Wnt5a, has also been found to be expressed in osteosarcoma tissues (41), although its expression levels and Wnt5b-Ror2 signaling in SaOS-2 and U2OS cells remain to be determined.
In this study we have performed partial dissection of Wnt5a-Ror2 signaling using the osteosarcoma cell lines SaOS-2 and U2OS cells. Considering that both SaOS-2 and U2OS cells are osteoblast-like cells, one can assume that Wnt5a-Ror2 signaling is also involved in the development of the skeletal system. In fact, we show that endochondral ossification was clearly detected at the ossification centers of the cartilaginous tissues from Ror2 ϩ/ϩ , but not Ror2 Ϫ/Ϫ mouse embryos at E15.5 (Fig.  7). Expression of Ror2 was detectable at the ossification centers at E15.5 (Fig. 7). The results suggest that Wnt5a-Ror2 signaling might play a role in endochondral ossification. It has been shown that MMP-13 is also involved in endochondral ossification via extracellular matrix degradation at the ossification centers (16,42). In this study we show that expression of MMP-13 was clearly detected at the ossification centers of the cartilaginous tissues from Ror2 ϩ/ϩ , but not Ror2 Ϫ/Ϫ mouse embryos at E15.5 (Fig. 7). Distribution of MMP-13 expression was overlapped if not identical with that of Ror2 expression in the cartilaginous tissues (Fig. 7). Based on these findings, it is conceivable that MMP-13 might be one of target genes of Wnt5a-Ror2 signaling under a physiological condition, i.e. this case the development of the cartilaginous tissues. It should be noted that Runx2, which is known to bind to the OSE2 within the proximal promoter region of MMP-13 gene (43), also plays a crucial role in the development of the cartilaginous tissue (44). Therefore, it is of interest to examine how Wnt5a-Ror2 signaling and Runx2 function might be integrated to achieve the proper development of the cartilaginous tissues.