Cross-talk between Wnt and Bone Morphogenetic Protein 2 (BMP-2) Signaling in Differentiation Pathway of C2C12 Myoblasts*

Loss of function of the Wnt co-receptor, lipoprotein receptor-related protein 5, decreases bone formation, and a point mutation in this gene results in high bone mass, indicating the importance of this signaling pathway in bone formation. However, the exact mechanism is currently unknown. We examined a potential role for Wnt signaling and functional cross-talk of bone morphogenetic protein 2 (BMP-2) in osteoblast differentiation. To assess the contribution of Wnt, we generated C2C12 cells over-expressing Wnt3a or Wnt5a and treated these with BMP-2. We showed that expression of matrix extracellular phosphoglycoprotein was induced by BMP-2 in Wnt3a over-expressing C2C12 cells but not in Wnt5a over-expressing C2C12 cells. Over-expression of Wnt3a blocked BMP-2-induced inhibition of myotube formation in C2C12 cells when switched to low mitogen medium. In these cultures, expression of inhibitor of DNA binding/differentiation (Id) 1, a helix-loop-helix protein induced by BMP-2, decreased in stable Wnt3a- but not in Wnt5a-expressing cells. This suppression is mediated by a GC-rich region of the BMP-2-responsive element of the Id1 gene promoter, and interaction between Smad1/4 and β-catenin is crucial for Wnt-mediated suppression of the BMP-2 response in C2C12 cells. Over-expression of the inhibitor of canonical Wnt signaling, Dickkopf, inhibits this suppression. In contrast, BMP-2 or Smad1/4 up-regulated Wnt3a or activated β-catenin-induced lymphoid-enhancing factor 1/T cell factor-dependent transcriptional activity. These findings identify functional cross-talk of Id1 expression between Wnt and BMP signaling and demonstrate a novel mechanism for Wnt regulation of the BMP-2 response, linking Id1 expression to Wnt/β-catenin signaling.

The Wnt secreted proteins are a family of glycosylated lipid-modified proteins that are powerful regulators of embryonic development, cell differentiation, proliferation, and migration. Signaling is initiated by Wnt ligand binding to two receptor molecules, the seven-transmembrane domain protein receptors of the Frizzled family and lipoprotein receptor-related proteins 5 and 6 (LRP5/6) (1). Two types of Wnt proteins have been identified: one class is ␤-catenin-dependent canonical Wnts such as Wnt1, Wnt2, Wnt3, and Wnt3a; the other class is the so-called "noncanonical" Wnts such as Wnt4, Wnt5a, Wnt5b, Wnt6, and Wnt7a, which is independent of or inhibits ␤-catenin signaling (2,3). According to the current model of canonical Wnt action, in cells lacking Wnt signal, glycogen synthase kinase-3␤ (GSK-3␤) 2 phosphorylates ␤-catenin, inducing rapid degradation of ␤-catenin via the ubiquitin/proteasome pathway. ␤-Catenin is also found cytoplasmically interacting with several molecules including the adenomatous polyposis tumor suppressor protein, lymphoid enhancer factor 1/T cell factor (Lef1/Tcf), and subunits of protein phosphatase 2A and axin. Conventional Wnt signaling causes stabilization of ␤-catenin and its accumulation in a complex with the transcription factor Lef1/Tcf, which regulates expression of target genes such as c-myc and cyclin D1 (2,3). Dickkopf (Dkk) is a secreted protein that acts as a Wnt pathway inhibitor by binding to and antagonizing LRP5/6 (4). Conversely, the socalled "noncanonical Wnt pathway" mediated by the Wnt5a subclass of Wnts triggers intracellular Ca 2ϩ release to activate protein kinase C and Ca 2ϩ /calmodulin-dependent kinase II and then activates TAK1 mitogen-activated protein kinase (MAPK) kinase kinase and nemo-like kinase (NLK) MAPK (5). NLK phosphorylates Lef1/Tcf and inhibits the interaction of the ␤-catenin-Lef1/Tcf complex with DNA (6). This distinct Wnt pathway converges in an antagonistic manner that is conserved between vertebrates and Caenorhabditis elegans.
Recently, Wnt signaling has been suggested to be involved in regulation of bone mass and bone formation. A loss of function mutation in LRP5 was found to associate with osteoporosis-pseudoglioma syndrome, an autosomal recessive disorder (7,8). Moreover, a Gly 171 -to-Val substitution mutation in LRP5 results in a high bone mass phenotype (9,10). These phenotypes associated with the loss of function or substitution mutations of LRP5 indicate that Wnt signaling might be involved in modulating the regulation of bone mass and bone formation (11). During osteogenesis, pluripotent mesenchymal stem cells differentiate into preosteoblasts, which then differentiate into mature osteoblasts that deposit the necessary components to form bone matrix and subsequent mineralization. Upon differentiation into osteoblasts, the cells express differentiation-related phenotypes such as a high level of alkaline phosphatase (ALP), parathyroid hormone receptor, type I collagen, osteocalcin, matrix extracellular phosphoglycoprotein (MEPE), and bone sialoproteins (12). In cultured cells, Bain et al. (13) described that stimulation of canonical Wnt signaling using constitutively active forms of ␤-catenin induces the activity of ALP, which is one of the * This study was supported by Grant-in-aid 16390529 from the Japanese Ministry of differentiation markers in osteoblastic cells. However, the exact mechanisms and signaling regulating osteoblastic differentiation or function and bone formation still remain to be elucidated.
MEPE was one of the bone matrix proteins identified from cDNA as a tumor-derived phosphaturic factor in oncogenic hypophosphatemic osteomalacia (14); this cDNA codes for a matrix protein consisting of a serine/glycine-rich secreted peptide that contains numerous potential phosphorylation sites and one RGD motif (15). This protein is a member of the SIBLING (small integrin-binding ligand, N-linked glycoprotein) family of matrix proteins, which includes MEPE, bone sialoprotein, dentin matrix protein 1, dentin sialophosphoprotein, enamelin, and osteopontin, expressed in bones and/or teeth (16). The tissue distribution of this mRNA is bone-specific and highly expressed in osteoblasts and osteocytes (15). Furthermore, targeted disruption of the MEPE gene in mice results in increased bone formation and bone mass (17). Although MEPE is perhaps involved in the regulation of bone metabolism in vivo, regulation of the expression of this gene during osteoblastic differentiation has not yet been identified.
Bone morphogenetic proteins (BMPs), members of the transforming growth factor-␤ (TGF-␤) superfamily, regulate the proliferation, differentiation, and apoptosis of various types of cells and organs not only in embryonic development but also in postnatal physiological function (18). At least 15 types of BMPs have been identified in humans. Among them, BMP-2 alone is sufficient to induce ectopic bone formation when it is implanted into the tissues of rodents. BMP-2 is reported to trigger osteoblast differentiation and to up-regulate the expression of most genes encoding osteoblastic phenotype-related proteins in vitro (18). Genetic disruptions of BMPs have resulted in various skeletal and extraskeletal developmental abnormalities (19). Signaling by TGF-␤ superfamily members including BMPs is initiated following their binding to two types (type I and II) of their respective serine/threonine kinase receptors. In the intracellular signaling of BMPs, eight Smad proteins have been identified to play critical roles in mammals (20). The receptor Smads (R-Smads), consisting of Smad1, Smad2, Smad3, Smad5, and Smad8, are phosphorylated directly by type I receptors and then form complexes with the Co-Smad, Smad4 and move into the nucleus, where they bind to the regulatory regions of the target genes and regulate their expression.
Inhibitor of DNA binding/differentiation 1 (Id1), one of the basic helix-loop-helix proteins, was identified as a gene for inhibition of myogenesis and as a typical early response gene in BMP treatment in various types of cells in mice and humans (21). We previously reported that the expression of Id1 was stimulated within 1 h after the addition of BMP-2 in C2C12 cells (22). Id family proteins such as Id1, Id2, and Emc, which lack the basic DNA binding domain, are capable of forming heterodimers with other basic helix-loop-helix family transcription factors, such as MyoD family proteins, that have been implicated in transcriptional regulation during differentiation of certain cell types including myoblasts. The helix-loop-helix heterodimers formed with Id proteins inhibit transcriptional, activity as Id proteins lack DNA-binding activity (23). Previously, we identified a 29-bp GC-rich element as a BMP-2responsive element in the 5Ј-flanking region of the human Id1 gene and showed that complexes of Smad1 and Smad4 recognize this element in response to BMP-2 (24).
Here, we have established stimulatory Wnt3a and inhibitory Wnt5a canonical Wnt-expressing pluripotent C2C12 cells to elucidate the responses of canonical Wnt signaling to modulation by BMP-2. In these cultures, expression of MEPE was induced by BMP-2 in Wnt3a overexpressing C2C12 (Wnt3a-C2C12) cells but not in either Wnt5a overexpressing C2C12 (Wnt5a-C2C12) cells or vehicle-transfected C2C12 cells. C2C12 cells form multinucleated myotubes when switched to low mitogen medium, and this can be inhibited by the addition of BMP-2. However over-expression of Wnt3a blocked this BMP-2-induced inhibition of myotube formation. We also show that Wnt3a down-regulates expression of the BMP-2-responsive gene, Id1, and this suppression is mediated by a 29-bp GC-rich region of the BMP-2-responsive element of the Id1 gene promoter. Interaction between ␤-catenin and Smad1/4 is found to be crucial for Wnt-mediated suppression of the BMP-2 response. In contrast, BMP-2 or Smad1/4 up-regulates ␤-catenin-induced Lef1/Tcf-dependent transcriptional activity. These findings reveal that Wnt signaling links BMP target gene expression and illustrate the functional role of Wnt signaling in BMP-2-regulated mesenchymal cell differentiation.

EXPERIMENTAL PROCEDURES
Cell Cultures-Cells of the mouse myoblast cell line C2C12 (Cell Systems, Kirkland, WA) were cultured in ␣-MEM (Sigma) containing 100 g/ml kanamycin (Meiji, Tokyo) supplemented with 10% fetal bovine serum (FBS, Sigma) at 37°C in 100-mm cell culture dishes (Corning, Corning, NY) in a humidified atmosphere of 5% CO 2 in air.
Establishment of Stably Transfected C2C12 Cells-Cells were plated 1 day before transfection in ␣-MEM at a density of 1 ϫ 10 5 cells/well (24-well plate). The cells were transfected with 1.0 g of Wnt3a-pUSEamp, Wnt5a-pUSEamp, or empty vector using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Two days after transfection, the medium was changed to ␣-MEM containing neomycin/Geneticin (G418) (Promega, Madison, WI) at 0.1 g/ml. The cells were passaged, and the clones were selected in ␣-MEM supplemented with G418 and 10% FBS. To rule out the possibility of clonal variation, we characterized at least three independent clones for each stable transfection. To examine the effects of Wnt over-expression on the action of BMP in endogenous differentiation, control C2C12 or stably transfected C2C12 cells were inoculated at 1.5 ϫ 10 6 /cm 2 and cultured in ␣-MEM containing 10% FBS (growth medium). On day 1 the medium was replaced with ␣-MEM containing 2.5% FBS (low mitogen medium), and the cells were cultured with BMP-2 or vehicle for 3-9 days.
Histochemical Analysis-To analyze expression of myosin heavy chain (MHC) histochemically, the cells were fixed for 10 min at room temperature with 3.7% formaldehyde and then stained immunohistochemically using anti-MHC monoclonal antibody (MF-20; purchased from Developmental Studies Hybridoma Bank, Iowa City, IA) and a secondary antibody of biotinylated rabbit anti-mouse immunoglobulin as described (22). The reaction products of biotin-streptavidin were visualized using the AEC substrate kit (Histofine, Nichirei Co., Tokyo).
Reporter Constructs-Luciferase reporter plasmids of Id985WT-luc, Id985mutB-luc, IdWT4F-luc, and IdmutB4F-luc were generated as described previously (24). In brief, the 1079-bp Id1 promoter fragment (Ϫ985 to ϩ94) was subcloned to pGL3 Basic (Promega) (Id985WT-luc). Id985mutB-luc was generated by PCR using the specific primer of 5Ј-ttctcgagCATGGCGACCGCCCGCGCTTTGCCA-3Ј (lowercase and underlining indicate flanking sequences and mutations, respectively). To generate IdWT4F-luc and IdmutB4F-luc, the specific oligo DNAs of the 29-bp BMP-responsive element with the wild-type or the mutB sequence flanked by six nucleotides were annealed, digested with XhoI and SalI, and subcloned into the pGL3 promoter (Promega). Topflash is a Tcf reporter plasmid containing two sets (with the second set in the reverse orientation) of three copies of the Lef1/Tcf binding site (wild type) upstream of the thymidine kinase minimum promoter and luciferase open reading frame (Upstate Biotechnology, Lake Placid, NY). Fopflash is a Tcf reporter plasmid containing two full copies and one incomplete copy of the Lef1/Tcf binding site (mutated) followed by three copies, in the reverse orientation, upstream of the thymidine kinase minimum promoter and luciferase open reading frame (Upstate Biotechnology).
Transient Cell Transfection and Assay for Luciferase Activity-For the reporter assay, transfection of plasmid DNA into cells was performed using the cationic lipid reagent Lipofectamine 2000 (Invitrogen). In brief, cells were plated 24 h before transfection at a density of 1 ϫ 10 5 cells/well (24-well plate) and cultured in ␣-MEM supplemented with 10% FBS. The cells were transfected with reporter plasmid and expression plasmid as indicated in the legends for Figs. 3 and 4. BMP-2 or vehicle was then added at the concentration indicated in the legends for Figs. 3 and 4. The cells were subsequently cultured in an atmosphere of 5% CO 2 and 95% air at 37°C for 24 h. The cells were then harvested, and luciferase activities in the cell extracts were determined using a PicaGene ® (Toyo Ink, Tokyo) and read using a Mini Lumat LB 9506 (PerkinElmer Life Sciences). For each transfection, additional cells were transfected with 100 ng of the pTK␤ plasmid (Invitrogen) for ␤-galactosidase assays, to account for the variation in transfection efficiency, and luciferase activity was normalized against ␤-galactosidase activity. All experiments were performed on triplicate samples and were independently repeated four to six times.
Electrophoresis Mobility Shift Assay (EMSA)-EMSA was performed essentially as previously described (24). C2C12 cells were incubated for 1 h, with or without 300 ng/ml BMP-2, and then nuclear extracts were prepared from the cells by the method of Muller et al. (36) with a protease inhibitor mixture and phosphatase inhibitor (Roche Diagnostics, Mannheim, Germany). Oligonucleotide probes were annealed by heating to 95°C for 5 min followed by cooling slowly to room temperature in the presence of 50 mM Tris-HCl (pH 8.0) and 10 mM MgCl 2 . Some sequences also contained cohesive SalI sites at the ends. The annealed probes were labeled by in-filling 5Ј protruding ends with [␣-32 P]dCTP (PerkinElmer Life Sciences) using the Klenow fragment (Amersham Biosciences) (37). The oligonucleotide sequences used as probes or competitors in EMSA were as follows: wild-type BMP-responsive element top strand, 5Ј-tcgacCATGGCGACCGCCCGCGCGGCGC-CAGCCT-3Ј; wild-type BMP-responsive element bottom strand, 5Ј-tcgacAGGCTGGCGCCGCGCGGGCGGTCGCCATG-3Ј. Six micrograms of the nuclear extracts were incubated with a labeled BMPresponsive element probe with the wild-type sequence. The reaction mixture was loaded onto a 5% polyacrylamide gel in 0.5ϫ TBE (44.5 mM Tris base, 44.5 mM boric acid, and 1 mM EDTA) and resolved by electrophoresis.

Establishment and Characterization of Wnt3a or Wnt5a Stably
Expressing C2C12 Cells-To examine a potential role for Wnt canonical signaling in mesenchymal cell differentiation and the functional contribution of BMP-2, we generated a pluripotent C2C12 cell line stably expressing Wnt3a or Wnt5a. C2C12 cells are a myoblastic cell line and a well characterized model system; C2C12 cells have been reported to differentiate into not only myotubes but also osteoblasts, depending upon the specific culture conditions, when incubated in the presence of BMPs for 48 -72 h (22). These culture conditions were previously shown to give rise to cells that express numerous mRNAs characteristic of osteoblasts including bone matrix proteins, such as osteocalcin and osteopontin, and transcription factors involved in osteoblastic differentiation, such as Runx2 (22,38). First, we examined several Wnt gene expression in C2C12 cells by RT-PCR analysis. Although Wnt4, Wnt5b, and Wnt10b were expressed during day 1-5, Wnt1, Wnt2, Wnt2b, Wnt3a, Wnt5a, Wnt7a, Wnt8, Wnt8b, Wnt10a, Wnt11, and Wnt15 were not detected during all days of cultures (results not shown). Therefore, C2C12 cells were transfected with the Wnt3a expression plasmid for activating canonical Wnt signaling or the Wnt5a expression plasmid for inhibiting canonical Wnt signaling and then were selected using G418 to establish cell lines. We designated these cell lines Wnt3a-C2C12 cells and Wnt5a-C2C12 cells, respectively. Although vehicletransfected C2C12 cells did not produce any detectable Wnt3a or Wnt5a mRNA expression, Wnt3a-C2C12 cells or Wnt5a-C2C12 cells expressed high levels of Wnt3a or Wnt5a mRNA as detected by RT-PCR analysis at day 2 in cultures ( Fig. 1A).
To confirm the activation or suppression of canonical Wnt signaling in these cells, we transfected with Topflash the reporter plasmid that carries six tandem repeats of the Lef1/Tcf binding site. The promoter activity of Topflash is enhanced in Wnt3a-C2C12 cells. In contrast, Topflash activity is suppressed in Wnt5a-C2C12 cells (Fig. 1B). To determine whether Wnt canonical signaling might reflect osteoblastic phenotypic expression, we examined ALP activity, a defined marker of early osteoblast differentiation. Wnt3a-C2C12 cells had detectable ALP activity, whereas neither Wnt5a-C2C12 cells nor control C2C12 cells stained for ALP activity (Fig. 1C). However any staining was weak compared with that of the osteoblastic cell line MC3T3-E1 (Fig. 1C). To determine the formation of a mineralized matrix in these Wnt-expressing cells, the cells were continuously cultured until days 7, 14, and 28 and then assayed by von Kossa staining. Wnt3a mRNA was maintained at similar levels throughout the culture period; thus the ALP activity was induced in these cultures though calcium deposition was not detected (results not shown). Cells expressing Wnt3a exhibited a distinct morphological change compared with Wnt5a-C2C12 cells or vehicle-transfected C2C12 cells (Fig. 1D). The cells were more elongated and, upon visual inspection, appeared to be denser.
Matrix Extracellular Phosphoglycoprotein and Osteocalcin mRNA Expressions and Their Regulation by BMP-2 and Wnt-We next examined whether Wnt signaling combined with BMP-2 specifically modulates expression of bone matrix proteins in osteoblastic differentiation. Osteocalcin and MEPE transcripts were not detectable in Wnt3a-C2C12, Wnt5a-C2C12, and vehicle-C2C12 cells, indicating that Wnt3a expression alone was not sufficient to induce all of the phenotypic markers for osteoblasts. Similar levels of osteocalcin mRNA expression were induced by BMP-2 treatment not only in vehicle-C2C12 cells but also in Wnt3a-C2C12 cells and Wnt5a-C2C12 cells, and Wnt3a did not change the level induced by BMP-2 (Fig. 1E). Interestingly, culturing the Wnt3a-C2C12 cells with BMP-2 dramatically induced MEPE expression as well (Fig. 1E). Time-dependent changes in MEPE expression in response to the addition of BMP-2 revealed that expression was induced after 24 h and then decreased slightly after 48 h (Fig. 1E). This induction was similar to that of the osteocalcin transcript, except that osteocalcin expression was induced after 24 h and then reached maximal levels by 48 h after BMP-2 treatment in Wnt3a-C2C12 cells as well as in Wnt5a-C2C12 or vehicle-C2C12 cells. The levels of MEPE mRNA were increased by BMP-2 in a dose-dependent manner in Wnt3a-C2C12 cells (Fig. 1F). However, no induction of MEPE expression was seen in Wnt5a-C2C12 cells or vehicle-C2C12 cells, indicating that Wnt3a and FIGURE 1. Establishment and characterization of Wnt3a or Wnt5a stable expressing C2C12 cells. C2C12 cells were transfected with 1.0 g of Wnt3a-pUSEamp, Wnt5a-pUSEamp, or empty vector, and then transfected cell clones (Wnt3a-C2C12, Wnt5a-C2C12, or C2C12 cells) were selected as described under "Experimental Procedures." A, expression of Wnt3a or Wnt5a mRNA in transfected C2C12 cells. Wnt3a-C2C12, Wnt5a-C2C12, or C2C12 cells were plated at 1 ϫ 10 5 cells/cm 2 in 100-mm cell culture dishes. After 2 days, total cellular RNA was extracted, and then RT-PCR was conducted to estimate the level of Wnt3a or Wnt5a mRNA expression. Equal loading of RNA samples was checked by amplification of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA. B, transcriptional activity of Topflash reporter. Cells were transiently transfected in 24-well plates with 0.1 g of Topflash, after which cells were cultured for a further 24 h. Luciferase activity was determined as described under "Experimental Procedures." Normalized luciferase activity is represented as -fold induction over C2C12 cells. C, ALP enzyme activity. The cells were cultured until confluent and then stained for ALP activity as described under "Experimental Procedures." D, cell morphology viewed by phase contrast microscopy. Magnification: ϫ100. E and F, time-and dose-dependent induction of MEPE and osteocalcin mRNA expression with the addition of BMP-2 in Wnt3a or Wnt5a stably expressed C2C12 cells. Wnt3a-C2C12, Wnt5a-C2C12, or C2C12 cells were plated at 1 ϫ 10 5 cells/cm 2 in 100-mm cell culture dishes. After 24 h, the medium was changed, and 300 ng/ml (E) or the indicated concentrations (F) of BMP-2 were added, after which cells were cultured for further indicated times (E) or for 24 h (F). Total cellular RNA was extracted, and then RT-PCR was performed to estimate the level of MEPE or osteocalcin mRNA expression as described under "Experimental Procedures." The data represent one of three experiments with similar results.
BMP-2 are required for MEPE mRNA induction in C2C12 cells. The results shown indicate that the combination with canonical Wnt signaling induced by Wnt3a and BMP-2 positively regulates MEPE expression in osteoblastic differentiation.
Effects of Wnt3a or Wnt5a Over-expression on Inhibition of Myogenic Terminal Differentiation by BMP-2 in C2C12 Cells-Next we examined the effects of Wnt over-expression on myogenic differentiation regulated by BMP-2 in the low mitogen culture medium. C2C12 cells generated numerous multinucleated myotubes at day 6, and the observed morphological transition from myoblasts to myotubes was also confirmed by immunohistochemical staining of MHC as a marker for mature muscle cells. By day 6, in the low mitogen medium, even Wnt3aor Wnt5a-over-expressing C2C12 cells were also fused to form multinucleated myotubes ( Fig. 2A). As shown previously, BMP-2 at 300 ng/ml inhibited the myotube formation in C2C12 cells, and the cells that remained as unfused mononuclear polygonal cells did not express MHC. As in the control C2C12 cells, neither myotubes nor MHC-positive cells were detected in Wnt5a-C2C12 cells treated with BMP-2. However, the MHC-positive multinucleated myotubes were observed in Wnt3a-C2C12 cells even in the presence of BMP-2 ( Fig. 2A). The mRNA expressions for myogenic differentiation markers were examined by RT-PCR analysis. Each type of cells expressed similar levels of myogenin and MCK mRNAs at day 3; these expressions increased in a time-dependent manner, indicating that Wnt signaling did not modulate the endogenous myogenic differentiation in these cells (Fig. 2B). Treatment with BMP-2 completely suppressed the induction of these expressions in Wnt5a-C2C12 cells as shown previously in C2C12 cells ( Fig. 2B and Ref. 22). However, even in the presence of BMP-2 (300 ng/ml), both myogenin and MCK mRNA expressions were detected at day 3 in Wnt3a-C2C12 cells and these expressions increased in a timedependent manner until day 9 (Fig. 2B). These results indicate that Wnt canonical signaling suppresses the inhibitory effect of BMP-2 on the myogenic differentiation in C2C12 cells. They also suggest that Wnt canonical signaling modulates BMP-2 signaling during the myogenic cell differentiation.
Wnt3a Modulates BMP-2-induced Id1 mRNA Expression in C2C12 Cells-It has already been shown that BMP-2 transiently induces Id1 mRNA expression in C2C12 cells. To investigate a potential role for canonical Wnt and the functional relationship of BMP-2, we first examined whether Wnt signaling specifically modulates BMP-2-induced Id1 mRNA expression. In Wnt-transfected cells, we added 300 ng/ml BMP-2 for 24 h and then analyzed gene expression by RT-PCR. The Id1 mRNA level induced by BMP-2 decreased in Wnt3a-C2C12 cells compared with vehicle-transfected C2C12 cells (Fig. 3A). In contrast, no significant change of Id1 mRNA level was observed in Wnt5a-C2C12 cells with the addition of BMP-2 (Fig. 3A). However, the mRNA level of osterix, a transcription factor essential for osteoblast differentiation, induced by BMP-2 was not changed in Wnt3a-C2C12 cells or Wnt5a-C2C12 cells (Fig. 3A).
To investigate the molecular mechanism underlying modulation of the BMP-2 response of the Id1 gene by canonical Wnt signaling, we analyzed promoter activity in the 5Ј-flanking region of the Id1 gene. First, we used a luciferase reporter plasmid, Id985WT-luc, which carries a 985-bp fragment of the 5Ј-flanking region of the Id1 gene (24). As shown in Fig. 3B, BMP-2 stimulates the promoter activity of Id985WTluc in C2C12 cells, in agreement with our previous findings (24). By transfection of Wnt3a into C2C12 cells, promoter activity of Id985WTluc induced by BMP-2 was reduced compared with that of vehicletransfected cells (Fig. 3B). In contrast, Wnt5a increased Id985WT-luc induced by BMP-2, indicating that Wnt canonical signaling regulates Id1 promoter activity in response to BMP-2 (Fig. 3B). As the 29-bp GC-rich region between Ϫ985 and Ϫ957 was identified as the BMP-2responsive element of the human Id1 gene, we used Id985mutB-luc with a mutation in the BMP-2-responsive element sequence to investigate whether canonical Wnt signaling regulates the response element. Neither Wnt3a nor Wnt5a regulates the transcriptional activity of Id985mutB-luc (Fig. 3B), suggesting that this element involves not only the BMP-2 response but also a canonical Wnt response in regulation of the Id1 gene.
The BMP-responsive Element of the Id1 Gene Involves Canonical Wnt Signaling-dependent Suppression-As the 29-bp GC-rich region between Ϫ985 and Ϫ957 was identified as the BMP-2-responsive element of the human Id1 gene, we used four copies of this 29-bp fragment with the wild-type or mutated sequence in tandem with the forward sequence (IdWT4F-luc) in front of the SV40 promoter (Fig. 3C). Wnt3a suppressed promoter activity of IdWT4F-luc induced by the addition of BMP-2 compared with that of vehicle-transfected cells, whereas Wnt5a enhanced the luciferase level induced by BMP-2 (Fig. 3C).
In the absence of conventional Wnt signaling, ␤-catenin is phosphorylated by GSK-3␤ and then degraded by proteasomes (4). When conventional Wnt signaling does not cause phosphorylation of ␤-catenin by GSK-3␤ and its accumulation and transport to the nucleus, it forms a complex with Lef1/Tcf that regulates target gene expression. An acti-

FIGURE 2. The effects of Wnt3a or Wnt5a over-expression on inhibition of myogenic terminal differentiation by BMP-2 in C2C12 cells.
A, immunohistochemical analysis for MHC expression. The cells were cultured in low mitogen medium without (a, c, and e) or with (b, d, and f) 300 ng/ml BMP-2. After 6 days, the cells were fixed and stained immunohistochemically for MHC with a specific antibody. a and b, C2C12 cells; c and d, Wnt3a-C2C12 cells; e and f, Wnt5a-C2C12 cells. Magnification: ϫ100. B, mRNA expression of proteins associated with terminal myogenic differentiation. C2C12, Wnt3a-C2C12, or Wnt5a-C2C12 cells were inoculated at 1.5 ϫ 10 6 /cm 2 and cultured in growth medium; then the medium was replaced on day 1 with low mitogen medium containing 300 ng/ml BMP-2 (ϩ) or vehicle (Ϫ). After the indicated days, total cellular RNA was extracted, and RT-PCR was conducted to estimate myogenin and MCK mRNA expression in these cells. RT-PCR for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was performed with the same samples as a control for the amount of reverse-transcribed cDNA present in the samples.
vating form of ␤-catenin that lacks sites of phosphorylation by GSK-3␤ (␤-catenin ⌬GSK) also suppressed IdWT4F-luc activity induced by BMP-2 (Fig. 4A). In contrast, the promoter activity of IdmutB4F-luc, which contains a 3-bp mutation introduced into the 29-bp GC-rich region, was not affected by BMP-2 and/or Wnt3a or by Wnt5a or the constitutively activated form of ␤-catenin, indicating that this BMP-2responsive element of the Id1 gene also contains a canonical Wnt modulating region (Fig. 4A).

Interaction of Smad and ␤-Catenin Modulates the BMP-2 Response by Canonical Wnt Signaling through the BMP-responsive Element of the Id1
Gene-Previously we showed that transcription factors, including Smad1 and Smad4, bound to the 29-bp BMP-responsive element using EMSA and supershift assay. Co-transfection of both Smad1 and Smad4 cDNA also up-regulated basal promoter activity of IdWT4F-luc in C2C12 cells as well as that induced by the addition of BMP-2 (24). In this induction, the constitutively active mutated form of ␤-catenin, ␤-catenin ⌬GSK, also inhibited the promoter activity of IdWT4F-luc induced by Smad1 and Smad4 (Fig. 5A), indicating that interaction between Smads and canonical Wnt signaling molecules such as ␤-catenin could associate in the suppression of the BMP-2-responsive element in the Id1 gene. To investigate DNA binding activity in the nucleus, we prepared nuclear extracts from C2C12 cells, Wnt3a-C2C12 cells, and Wnt5a-C2C12 cells within 1 h after the addition of BMP-2 and performed EMSA using the 29-bp BMP-responsive element as a probe. As shown as Fig. 5B, binding activity of the 29-bp probe in nuclear extracts prepared from control C2C12 cells was very low, whereas the binding activ-ity in nuclear extracts prepared from C2C12 cells treated with BMP-2 was significantly increased. In contrast, binding activity was decreased in BMP-2-treated C2C12-Wnt3a cells (Fig. 5B). The mobility of the DNA-protein complexes from nuclear extracts of BMP-2-treated C2C12-Wnt5a cells was increased to that observed in C2C12 nuclear  A, RT-PCR analysis of Id1 mRNA expression. C2C12 cells, Wnt3a-C2C12 cells, and Wnt5a-C2C12 cells were stimulated with 300 ng/ml BMP-2 (ϩ) or vehicle (Ϫ) for 24 h, and then RT-PCR analysis was performed using Id1 primers or osterix primers as described under "Experimental Procedures." Equal loading of RNA samples was checked by amplification of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA. B, C2C12 cells were transiently co-transfected in 24-well plates with 0.1 g of Id985WT-luc or Id985mutB-luc as a reporter plasmid and 0.1 g of Wnt3a, Wnt5a, or vehicle expression plasmid, as indicated, and pTK␤. Then, 300 ng/ml BMP-2 (ϩ) or vehicle (Ϫ) was added, after which cells were cultured for a further 24 h. Luciferase activity was determined as described under "Experimental Procedures." Normalized luciferase activity is represented as -fold induction over Id985WT-luc with vehicle. Data are means Ϯ S.D. (n ϭ 5). C, C2C12 cells were transiently co-transfected in 24-well plates with 0.1 g of IdWT4F-luc or IdmutB4F-luc and 0.2 g of Wnt3a, Wnt5a, or vehicle expression plasmid as indicated. Then 300 ng/ml of BMP-2 (ϩ) or vehicle (Ϫ) was added, after which cells were cultured for a further 24 h. Normalized luciferase activity is represented as -fold induction over IdWT4F-luc with vehicle. Data are means Ϯ S.D. (n ϭ 5).
extracts (data not shown). These findings suggest that the binding activity of the BMP-responsive element in nuclear extracts is also regulated by canonical Wnt signaling in C2C12 cells and that its binding activity is closely correlated to the promoter activity and mRNA expression of the Id1 gene.
BMP-2 Up-regulates Transcriptional Activity of Canonical Wnt Signaling-In turn, to examine further the relationship between Wnt signaling and BMP-2, we performed a reporter assay for promoter activity using Topflash, which carries six tandem repeats of the Wnt response element, to elucidate whether BMP-2 regulates the canonical Wnt response. When BMP-2 was added to the medium for 24 h, the promoter activity of Topflash was not altered in C2C12 cells (Fig. 6A). As shown previously, transfection of a canonical Wnt signaling molecule such as Wnt3a or ␤-catenin ⌬GSK induced promoter activity of Topflash. To our surprise, the addition of BMP-2 to transfected cells with activated canonical Wnt signaling involving Wnt3a or ␤-catenin ⌬GSK caused a further dramatic increase in Topflash activity, and these inductions were dose-dependent (Fig. 6A). No inducible activity was detected with BMP-2 in cells transfected with Wnt5a (Fig. 6A). Inducible activity was also observed with the transfection of Smad1/4 instead of the addition of BMP-2 in cells transfected with either Wnt3a or ␤-catenin ⌬GSK (Fig. 6B). Therefore, BMP-2 signaling could increase the canonical Wnt response in this reporter system in these cells.

DISCUSSION
C2C12 myoblasts differentiate into multinucleated myotubes when the concentration of FBS in culture medium is reduced from 10 to 2.5%. Our previous studies have shown that BMP-2 inhibits the myogenic differentiation of C2C12 cells into mature muscle cells but also converts their differentiation pathway into that of osteoblasts (22). We show that signaling induced by both canonical Wnt and BMP-2 is needed to induce the expression of late osteoblastic differentiation marker genes such as MEPE. Recent studies have shown that antibodies to MEPE specifically immunostain highly differentiated osteoblasts in adult bone tissues and that MEPE mRNA expression in cultured cells is greatly increased when cells are maintained at confluence in the presence of dexamethasone to stimulate differentiation along the osteoblastic lineage (15). Therefore, this body of evidence indicates that MEPE is a specific marker of the osteoblast. Previously we reported that MEPE expression is down-regulated by FGF2 in osteoblastic cells derived from high-density cultures of primary rat bone marrow stromal cells incubated with dexamethasone, ␤-glycerophosphate, and ascorbic acid (39). In our present study, not only BMP-2 alone but also BMP-2 in combination with Wnt3a was found to have inductive effects on MEPE mRNA expression, which opens up the possibility that such canonical Wnt signaling may be a potential stimulator of osteoblastic differentiation and function in bone tissue. Here we also found that Wnt canonical signaling induced by Wnt3a over-expression suppressed the inhibitory effects of BMP-2 on the myogenic differentiation in C2C12 cells, indicating that the Wnt canonical signaling is capable of regulating BMP-2 signaling. Therefore, a functional cross-talk between Wnt and BMP-2 signaling was further examined in this study.
Id1 is one of a group of negative regulatory proteins for muscle differentiation, and its mRNA and protein levels are down-regulated upon initiation of myogenic differentiation. Id1 is the most significantly upregulated gene upon BMP-2, BMP-6, or BMP-9 stimulation in various types of human and mouse cells (20). Previously, we have shown that the GC-rich element between Ϫ985 bp and Ϫ957 bp of the human Id1 gene was the response element for BMP-2 (24). This 29-bp element showed 100% homology between the human and mouse Id1 gene, indicating that this element has a critical role in the expression of Id1 in mammals. Korchynskyi and ten Dijke (40) also reported that BMPs, but not TGF-␤, rapidly activate the mouse Id1 promoter via two regions, one of which contains CGCC sequence elements flanked by CAGC motifs and the other of which harbors two Smad-binding elements. Recently, BMPdependent activation of Msx2 was shown to be mediated via binding of Smad1 and Smad4 at Smad-binding elements of the Msx2 gene in murine embryonic stem cells (41). In this study, we show that downregulation of Id1 gene promoter activity by canonical Wnt signaling is mediated via the Smad-binding element of the Id1 gene promoter. Thus ␤-catenin could bind Smad1/4, leading to reduced DNA binding activity and transcriptional activation by complexes of transcription factors including Smad1/4. We hypothesized that ␤-catenin might interact with Smad and repress transcription by masking the DNA binding domain of Smad, leading to inhibition of the BMP-2 target gene, Id1. Several lines of evidence suggest that the transcriptional activator p300/ CBP is required for Smad-dependent transactivation of target genes and that Wnt and BMP signaling could occur via enhanced accumulation of p300/CBP on the response element of the promoter. Ghosh et al. (42) demonstrated that p53 also represses TGF-␤ stimulation of promoter activity driven by minimal Smad-binding elements and that over-expression of the transcriptional coactivator p300 rescued TGF-␤ stimulation of collagen ␣2(I) promoter activity in fibroblasts over-expressing p53, suggesting that ligand-dependent interaction of Smad3 with p300 may be one of the targets of p53-mediated inhibition of TGF-␤ responses. Thus, it is possible that the transcriptional complexes decrease the affinity or the levels of coactivator molecules by binding of ␤-catenin to Smad1/4 in the Smad-binding element of the Id1 gene promoter.
BMP and Wnt signaling pathways have been studied extensively in regard to the regulation of early embryonic development and the control of cell differentiation and proliferation in adult tissues (2). The Wnt signaling pathway acts as a dorsal modifier and the BMP signaling pathway as a mesoendoderm inducer (43). Genetic interaction between these highly conserved and ubiquitous signaling pathways has been observed in multiple settings in fruit flies, amphibians, zebrafish, and mammals. Several studies have reported that these two signaling pathways link to biological responses via the formation of a nuclear tran-  3). B, electrophoretic mobility shift assay of the GC-rich 29-bp binding activity in C2C12 cells, C2C12-Wnt3a cells, and C2C12-Wnt5a cells. Nuclear extracts were prepared from cells treated with and without BMP-2. EMSA was performed using the BMP-responsive element as a probe as described under "Experimental Procedures." scription factor complex that acts on target genes, resulting in Wnt signaling repression of BMP-induced activation. For example, over-expression of Dkk1 enhances BMP4-triggered apoptosis in vertebrate limb development, indicating that Wnt signaling inhibits apoptosis induced by BMP signaling (44). In embryonic stem cells, Wnt signaling also inhibits neural differentiation; this differentiation can be partially restored by the addition of the BMP antagonist Noggin, suggesting that Wnt signaling inhibits BMP signaling (45). Our observations provide evidence that Wnt may negatively regulate BMP signaling not only apoptosis in limb development or neural differentiation but also during differentiation of mesenchymal cells. Another report have indicated that the Wnt signaling pathway indeed directly represses transcription of some genes including the osteocalcin gene mediated by Runx2 (46). Our results confirm the cooperative interaction between Smad1/4 and ␤-catenin in gene expression during mesenchymal cell differentiation.
In our data, BMP-2 or Smad1/4 enhance the response of the Lef1/Tcf reporter. Several reports demonstrated that TGF-␤ signaling results in Smad3 or Smad4 directly interacting with ␤-catenin/Lef-1 and the transcriptional activation of Lef1-responsive promoters (47,48). Takizawa et al. (49) reported that BMP-2 enhances Notch-induced transcriptional activation of the Hes-5 gene via the recombination signal-binding protein-J binding sequence in the Hes-5 gene promoter by recruitment of co-activator p300/CBP interacting with ␤-catenin and Smad1. Our present results suggest the possibility that the promoter of the MEPE gene interacts with ␤-catenin/Tcf/Lef1 in cooperation with Smads, and then these complexes induce expression of this gene. It is therefore a possible mechanism that Smad1/4 could form a complex with ␤-catenin/Tcf1 in response to either the Wnt/␤-catenin or BMP signaling pathway via the recruitment of a co-activator such as p300/CBP.
Wnts have been shown to play important roles in the regulation of many aspects of development, which may include chondrogenesis and osteogenesis (11). Because Wnt3a plays a key role in the development of the axial and appendicular skeleton and appears to regulate stem cell proliferation and differentiation, we sought to explore its effect on cells cultured under osteogenic conditions. Intriguingly, the synergism of expression of Wnt3a or Wnt5a with BMP-2 on gene transcription occurred without altering expression of Runx2 (data not shown), suggesting that canonical Wnt actions are independent or downstream of this osteoblast-specific transcription factor (50). This is consistent with recent observations in early events during skeletal development such as limb patterning. The phenotypic abnormalities observed in Wnt-deficient mice occur in the context of normal expression of a gene controlling osteogenesis, Runx2 (2), indicating a role for a Runx2-independent pathway in control of osteoblast proliferation and function. From our observations, BMP-2 did not regulate Topflash activity in C2C12 cells, indicating that it is unlikely that BMP-2 regulates GSK-3␤ enzymatic activity or increases ␤-catenin nuclearization or interaction with Lef1/ Tcf in a transcription complex, at least in these cells. It has been reported that BMP-2 activates not only Smad proteins but also MAPK and phosphatidylinositol 3-kinase to induce the respective intracellular signals (51). We cannot rule out the possibility that MAPK and/or phosphatidylinositol 3-kinase are involved in the effects of BMP-2, but the contribution of these signaling pathways would be minor, at least in the expression of Id1 in C2C12 cells.
In osteogenic differentiation, activating ␤-catenin induces ALP expression, one of the early markers of osteoblastic differentiation, in the pluripotent mesenchymal progenitor cells C3H10T1/2, and this differentiation induces mineralization of the osteoblast-like cells MC3T3-E1 and participates in BMP-2-mediated signal transduction (13). We also observed ALP activity in Wnt3a-C2C12 cells but not in Wnt5a-C2C12 cells. Recently, Mbalaviele et al. (52) showed that BMP-2 and ␤-catenin synergize to promote osteoblast differentiation such as stimulation of ALP activity during osteogenic differentiation of multipotent embryonic cell lines and new bone formation. Rawadi et al. (53) proposed that BMP-2 induction of ALP in these cells partially relies on the Wnt expression cascade and that ␤-catenin might regulate osteoblastic differentiation via multiple mechanisms. However, in our experiments using C2C12 cells, BMP-2 did not induce canonical Wnt expression and Tcf/Lef1-dependent transcriptional activation. It could be ruled out that BMP-2 regulates canonical Wnt expression leading to direct regulation of ␤-catenin nuclearization or interaction with Tcf/ Lef1 in a transcription complex, at least in C2C12 cells. Our results indicate that BMP-2-induced osteoblastic differentiation may be regulated by canonical Wnt signaling and that ␤-catenin can also participate in non-BMP-2-dependent osteoblastic differentiation processes. Constitutive over-expression of the Id1 gene inhibited osteoblast differentiation initiated by BMP-9, and this BMP-9-regulated Id1 expression is Smad4-dependent (54,55). Taken together, our findings suggest that canonical Wnt signaling may enhance osteoblastic differentiation via down-regulation of the Id1 helix-loop-helix protein. Moreover, Id1 expression must be down-regulated during the terminal differentiation of committed osteoblasts, suggesting that a balanced regulation of Id1 expression may be critical to BMP-induced osteoblast lineage-specific differentiation of mesenchymal progenitor cells.
During bone repair process in skeletal fractures, it is reported that expression certain Wnt gene members and their target genes are expressed (56). The healing of skeletal fractures is essentially a replay of bone development, involving the closely regulated processes of osteo-FIGURE 6. BMP-2 enhances Wnt3a-or ␤-cateninmediated transcriptional activation of Lef1/Tcf reporter in C2C12 cells. A, C2C12 cells were transiently co-transfected in 24-well plates with 0.1 g of Topflash and expression plasmid of Wnt3a or ␤-catenin ⌬GSK as indicated. Then 300 ng/ml BMP-2 was added, after which cells were cultured for a further 24 h. B, C2C12 cells were transiently co-transfected in 24-well plates with 0.1 g of Topflash, expression plasmid of Wnt3a, ␤-catenin ⌬GSK or vehicle, and Smad1 and Smad4 expression plasmid as indicated; then cells were cultured for a further 24 h. Normalized luciferase activity is represented as -fold induction over Topflash with vehicle. Data are means Ϯ S.D. (n ϭ 5).
genesis. We attempted to deliver plasmid DNA containing the Wnt3a gene to the repair process of fractured bone by electrogene transfer in rat. Micro-computed tomography analysis at 14 days after fracture exhibited enlarged callus volume with Wnt3a expression, suggesting that Wnt3a may promote cell differentiation in the callus and the repair of fractured bone. 3 Several Wnt have recently been shown to be targets of sonic hedgehog (4). It was shown previously that hedgehog is expressed in the developing skeletal system and thus holds an important roles as a regulator of skeletogenesis (12). The hedgehog pathway was also shown to be linked to the Wnt and BMP signaling pathways during skeletal development (12). Taken together, our results showing the interaction of multiple signaling pathways may play a pivotal role in the successful study of mammalian bone regeneration, skeletal development, and wound repair.
In conclusion, our results reveal the molecular mechanisms of how BMP-2 and Wnt signaling pathways cooperate to regulate graded expression of Id1. Moreover, we propose that the results of this study suggest that canonical Wnt signaling has putative physiological significance for BMP-2-regulated cell differentiation from mesenchymal progenitor cells. The physiological and functional interactions between ␤-catenin and Smad1/4 and the cooperation of the two pathways, permitting tight regulation of this regulatory process, may link the two pathways and contribute to the specification of cell fates and bone formation.