Murine Frizzled-1 Behaves as an Antagonist of the Canonical Wnt/β-Catenin Signaling*

Activation of the Wnt signaling cascade provides key signals during development and in disease. Wnt signals are transduced by seven-transmembrane Frizzleds (Fzs) and the single transmembrane low density lipoprotein receptor-related proteins 5 or 6. In the course of the analysis of genes regulated by bone morphogenetic protein 2 in mesenchymal cells we found a significant induction of murine Frizzled-1 (mFz1) gene expression. Unexpectedly overexpression of mFz1 dramatically repressed the induction of alkaline phosphatase mediated by either bone morphogenetic protein 2 or Wnt3a in these cells. Moreover mFz1 overexpression significantly repressed both β-catenin translocation into the nucleus and T cell factor signaling mediated by Wnt3a. Importantly microinjection of mFz1 transcript in Xenopus embryo inhibited the ability of Wnt1 to induce the expression of the Wnt/β-catenin target gene Siamois in animal cap assay and secondary axis formation in whole embryo. By using chimeric constructs in which N- and C-terminal segments of mFz1 were replaced by the corresponding parts of Xfz3 we demonstrated that the antagonistic activity resides in the cysteine-rich domain of the N-terminal part. The antagonist activity of mFz1 could be prevented by overexpression of Gαq-(305-359), which specifically uncouples Gq-coupled receptors, suggesting that Gαq signaling contributes to the inhibition of Wnt/β-catenin pathway by mFz1. This is the first time that a Frizzled receptor has been reported to antagonize Wnt/β-catenin.

The transforming growth factor (TGF) 1 -␤ family plays a central role in regulating a broad range of cellular responses including cell growth and differentiation. Among members of this family, bone morphogenetic proteins (BMPs) have been shown to regulate growth and differentiation of chondroblast and osteoblast lineage cells and induce the commitment of mesenchymal cells into osteoblast/chondroblast phenotypes (1)(2)(3). In addition, when ectopically implanted, BMPs possess the capacity to induce bone and cartilage formation (4,5). In a series of murine cell lines including C3H10T1/2, C2C12, and ST2, BMP-2 induces the expression of several osteoblast differentiation markers including alkaline phosphatase (ALP) and osteocalcin (6). In contrast, TGF-␤1 has been shown to repress the expression of osteoblast markers, and it abolishes the increase in osteocalcin induced by calcitriol (7,8). Moreover we have previously showed that TGF-␤1 is able to antagonize BMP-2 activities in several mesenchymal cell lines (9).
Wnts are secreted proteins involved in a wide range of developmental processes such as embryonic axis specification and organogenesis (10). They have been shown to activate distinct pathways both in vertebrates and in invertebrates. The better described pathway is the Wnt/␤-catenin pathway. In this pathway, the interaction between Wnt and Frizzled receptors leads to inactivation of glycogen synthase kinase-3␤. Genetic epistasis experiments suggest that Dishevelled lies upstream and represses the activity of glycogen synthase kinase-3␤. As a consequence, ␤-catenin is stabilized in the cytoplasm and then forms a complex with TCF/lymphocyte enhancer factor to activate transcription of target genes. Frizzled proteins have been shown to function as Wnt receptors (11), and they constitute a large family of seven-transmembrane receptors with at least 10 members in mammals (12). All Frizzled receptors have a conserved extracellular cysteine-rich domain (CRD) followed by seven putative transmembrane segments. Their cytoplasmic regions differ in length and sequence. Functional analyses in Drosophila and Xenopus embryos indicate that Frizzled proteins have distinct functions in Wnt/␤-catenin signaling (13). More recently, members of the low density lipoprotein receptorrelated protein (LRP) family have been shown to be co-receptors for Wnt ligands (14 -18). The two closely related proteins LRP5 and LRP6 are single pass transmembrane receptors and associate with Frizzled receptors in a Wnt-dependent manner. There is an increasing body of evidence showing that Wnt/␤catenin signaling plays an essential role during osteoblast differentiation. Deletion of LRP6 gene in mice results in embryos with spina bifida, absence of tail, and malformed fore and hind limbs. The skeleton of the neonate also shows truncation of the axial skeleton and loss of distal dorsal limb structures (16). More importantly, mice deficient in LRP5 exhibit decreased osteoblast proliferation and osteopenia (19). Furthermore mutations resulting in a loss of function of LRP5 in humans lead to osteopenia (20), and, in contrast, mutation of G171V leads, in two separate families, to a marked increase in bone mass (21,22). These results clearly suggest a genetic link between LRP5, and consequently Wnt, signaling and the regeneration of bone mass.
The control of osteoblast differentiation involves a complex combination of signals. However, the relationship and possible cross-talk between different signaling pathways and/or different components of these pathways remain largely unknown. In this study we report that BMP-2 up-regulates the expression of Frizzled-1 in different murine mesenchymal cell lines. Unexpectedly overexpression of Frizzled-1 counteracts the effects of BMP-2 and Wnt3a in inducing the expression of the osteoblast differentiation marker ALP, indicating that mouse Frizzled-1 (mFz1) may exert a feedback mechanism to modulate the effect of BMP-2 in murine mesenchymal cells. More importantly we have demonstrated, using a Xenopus assay, that mFz1 is capable of antagonizing Wnt/␤-catenin signaling. Our data suggest that the CRD of mFz1 may be responsible for its inhibitory function. Whether mFz1 antagonizes Wnt signaling by competing with other members of the Frizzled family or with Wnt binding for LRP5 association or by both mechanisms remains to be determined.
RNA Extraction, Chip Hybridizations, and Quantitative RT-PCR-Total RNA samples were obtained from three C2C12 cell cultures (BMP-2-treated, 1 g/ml; TGF-␤-treated, 2.5 ng/ml; and solvent-treated control, 10 mM HCl) by use of the RNAplus kit provided by Quantum, harvesting from each culture at six time points (4 h, 8 h, 1 day, 2 days, 3 days, and 4 days). Total RNA was also extracted from ST-2 and C3H10T1/2 cells under similar conditions. Total RNA samples from C2C12 cells were hybridized to the complete series of Affymetrix 35,000 mouse chips as described by Theilhaber et al. (23), and data assembly and analysis were carried as described by the authors (23).
Wnt3a-conditioned Medium Preparation-Wnt3a-conditioned medium (Wnt3a-CM) was prepared as described by Shibamoto et al. (24). Briefly, to collect the conditioned medium from cultures of Wnt3aproducing L cells, these cells were seeded at a density of 6 ϫ 10 6 cells in a 125-cm 2 flask containing Dulbecco's modified Eagle's medium with 10% fetal calf serum. 24 h after seeding, medium was changed to Dulbecco's modified Eagle's medium with 2% fetal calf serum, and cells were cultured for 3 days. Then Wnt3a-CM was harvested, centrifuged at 1000 ϫ g for 10 min, and filtered through a nitrocellulose membrane. The activity of Wnt3a-CM was assayed on normal L cells by examining the increase in ␤-catenin as described by Willert et al. (25). Wnt3a-CM was added to cells at 20% final concentration in all subsequent experiments.
Measurement of Alkaline Phosphatase Activity-C2C12 cells were plated at 2 ϫ 10 4 /cm 2 and 24 h later were transfected with the indicated constructs (1 g total) using FuGENE 6 (Roche Applied Science). 16 h after transfection, cells were washed and cultured in media containing 2% fetal bovine serum. When indicated, cells were treated with BMP-2 (100 ng/ml) (6) or with Wnt3a-CM. 48 h later, ALP activity was determined in cell lysates using the alkaline phosphatase Opt kit (Roche Applied Science). Cell lysates were analyzed for protein content using the micro-BCA assay kit (Pierce), and ALP activity was normalized for total protein concentration.
Luciferase Assay-COS-7 cells plated in 24-well plates at 2 ϫ 10 4 /cm 2 were transiently transfected with the indicated constructs (1 g total) using FuGENE 6 (Roche Applied Science). 20 ng of pRL-TK (Promega, Madison, WI), which encodes a Renilla luciferase gene downstream of a minimal herpes simplex virus thymidine kinase promoter, was systematically added to the transfection mixture to assess transfection efficiency. When required, controls were carried out by replacing constructs with corresponding empty vectors. 16 h after transfection, cells were washed and cultured in media containing 2% fetal bovine serum for an additional 24 h. Cell lysates were prepared, and luciferase assays were performed with the Dual Luciferase assay kit (Promega) according to the manufacturer's instructions. 10 l of cell lysate were assayed first for firefly luciferase and then for Renilla luciferase activity. Firefly luciferase activity was normalized to Renilla luciferase activity.
Myc-tagged Xenopus Frizzled-3 (Xfz3) was described previously (29). Xenopus Wnt1 coding sequence was amplified by PCR and subcloned in pCS2 vector (30). mFz1 was obtained by PCR amplification of mFz1 CRD and the downstream sequence with sense primer 5Ј-ATCTCCAT-GGCGGACCACGGCTAC-3Ј and antisense primer 5Ј-AGTTCTAGACG-GTAGTCTCCCCCT-3Ј including a NcoI and XbaI site, respectively. mFz1 amplicon was cloned into pCS2 vector in-frame with the signal peptide and Myc epitopes using convenient restriction sites. Derived truncated or chimera constructs were also cloned in the same vector using convenient restriction sites in-frame with the signal peptide and Myc epitopes. mFz1 was also cloned in the same vector using convenient restriction sites. The Myc-tagged mFz1⌬C that retains seven residues after the seventh transmembrane domain was obtained by digesting the full-length mFz1 with NcoI and EcoRI. mFz1⌬N lacking the CRD was obtained by PCR amplification of the seven transmembrane and the C-terminal cytoplasmic domains using a sense and an antisense primer including a NcoI and a XbaI sites (underlined), respectively: 5Ј-AATC-CATGGGCCAGAACACGTC-3Ј and 5Ј-AGTTCTAGACGGTAGTCTC-CCCCT-3Ј. The PCR product was digested by NcoI and XbaI and then subcloned into the corresponding sites of Myc-tagged Xfz3 in-frame with the signal peptide and Myc epitopes. The chimeric receptors Xfz3N/mFz1C and mFz1N/Xfz3C were generated by PCR amplification of the C-terminal cytoplasmic region of mFz1 or Xfz3 with a sense primer, 5Ј-ACATTCGAATGGAGGAAGTTCTAC-3Ј with a BstBI site or 5Ј-ACTGAATTCCTGGGCCAGCTTTTTC-3Ј with an EcoRI site (underlined), respectively, and the T7 primer. The PCR products were cloned into the same sites in Myc-tagged Xfz3 or mFz1, respectively, and the XbaI site. To obtain Xfz3CRD/mFz1C, the sequence corresponding to the seven transmembrane and the C-terminal cytoplasmic domains of mFz1 was amplified by PCR using sense primer 5Ј-AGAGGAGCTC-CGCTTCTCGCGC-3Ј including a SacI site (underlined) and the T7 primer and subcloned into the SacI and XbaI sites of the Myc-tagged Xfz3.
Immunoblotting-Cells were transfected with expression vectors as described above, and 24 h post-transfection, cells were lysed (20 mM Tris-HCl (28), 150 mM NaCl, 5 mM MgCl 2 , 10% glycerol, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml aprotinin, and 20 g/ml leupeptin), and then the protein content of cell lysates was determined with Bio-Rad DC protein assay. 50 g of each cell lysates were separated by SDS-PAGE. Proteins were then electrophoretically transferred to nitrocellulose membranes and probed with the indicated primary antibody. Immunoreactive bands were visualized by ECL Western blotting technique (Amersham Biosciences) using horseradish peroxidase-labeled secondary antibodies. Anti-Myc antibody was from Santa Cruz Technology, and the anti-␤-catenin antibody was obtained from Upstate Biotechnology.
Xenopus Embryos and Microinjections-Xenopus eggs were obtained from females injected with 500 IU of human chorionic gonadotropin (Sigma) and artificially fertilized with minced testis. Eggs were dejellied with 2% cysteine hydrochloride (pH 7.8) and kept in 0.1ϫ modified Barth solution. Synthetic capped mRNAs were made by in vitro transcription as described previously (29). Embryos were injected at the four-cell stage near the animal pole region (for ectodermal explants) or in the two ventrovegetal blastomeres (for secondary axis induction) in 0.1ϫ modified Barth solution containing 3% Ficoll-400. Ectodermal explants were dissected at blastula stage and cultured to early gastrula stage for RT-PCR analysis.
RT-PCR-For RT-PCR, RNA samples were treated with RNase-free DNase I (Roche Applied Science) and were reverse-transcribed using 200 units of SuperScript (Invitrogen). PCR and primers for the Wnt/␤catenin target gene Siamois and the housekeeping gene ornithine decarboxylase were as described previously (29).
Confocal Microscopy Assay-COS-7 cells are plated at 40,000 cells/ well in a 6-well plate; each well contained a sterile microcoverglass, which can be removed for observation. Cells were transfected with the indicated plasmid (1 g total) using FuGENE 6 (Roche Applied Science). Transfected cells were treated or not with Wnt3a-CM for 24 h, then fixed with 3.7% formaldehyde (Sigma) for 10 min, and washed two times with phosphate-buffered saline. Fixed cells were permeabilized by phosphate-buffered saline, 0.025% Triton X-100 (Sigma) for 5 min and blocked in phosphate-buffered saline, 3% bovine serum albumin for 15 min. After that cells were incubated overnight at 4°C with the primary antibody, a rabbit polyclonal anti-␤-catenin (Upstate Biotechnology), at 5 g/ml in phosphate-buffered saline. After being washed two times cells were incubated for 1 h at room temperature with the secondary antibody, goat anti-rabbit conjugated to FITC (Santa Cruz Biotechnology) at a 1:100 dilution. The cells were washed tree times, mounted on a glass slide, and viewed with a confocal laser scanning microscope (LSM510, Carl Zeiss).

Up-regulation of mFz1 Expression by BMP-2 in Mesenchymal
Cell Lines-The myoblast cell line C2C12 displays a certain degree of plasticity in terms of differentiation, and BMP-2 treatment is known to inhibit myotube formation and converts the differentiation pathway of C2C12 cells into that of the osteoblast lineage (31). To identify genes playing a role in the transdifferentiation of these pluripotent cells, we performed a

FIG. 2. Frizzled-1 inhibits BMP-2 and Wnt activities.
A, C2C12 cells were transiently transfected with empty vector (pCS2), Xfz3-, hFz1-, or mFz1-expressing construct and then treated with BMP-2 or with Wnt3a-CM. ALP activity was measured in cell lysates 72 h after transfection, normalized to protein content, and expressed as nmol of p-nitrophenyl phosphate (pnpp)/min/g of protein (prot). B, COS-7 cells were transiently co-transfected with TCF-1 expression construct, TOPflash, and pTK-Renilla. Where indicated either empty vector, Xfz3, hFz1, or mFz1 was added to the co-transfection mixture. 18 h after transfection cells were treated with Wnt3a-CM, and luciferase activity was determined in cell lysates 24 h later and normalized to Renilla signal. All experiments were performed in triplicate and repeated three times. Data Ϯ S.D. from one representative experiment are presented. C, expression of constructs used in C2C12 cells and their effect on cellular ␤-catenin level. C2C12 cells were transfected with mFz1, hFz1, or Xfz3 expression construct or control vector. 24 h later expression of Frizzled constructs was revealed by immunoblotting using anti-Myc antibody (upper panel). ␤-Catenin expression was also assessed in the same cell lysates using a specific ␤-catenin antibody (lower panel).
genome-wide expression analysis by determining changes in expression levels of 27,000 genes with Affymetrix oligonucleotide chips (23). Among these genes Frizzled-1 (mFz1) was found to be significantly increased in the presence of BMP-2 (Fig. 1A). This increase in the expression level of mFz1 was observed 24 h after BMP-2 treatment and was maintained until 4 days of culture. In contrast, TGF-␤1, which inhibits myotube formation of C2C12 cells without inducing the osteoblastic phenotype (31), did not significantly affect mFz1 expression in our experiments. The increase in expression of mFz1 in response to BMP-2, but not to TGF-␤1, was confirmed by quantitative RT-PCR (TaqMan) in C2C12 cells but also in two other murine mesenchymal pluripotent cell lines, ST2 and C3H10T1/2 (Fig. 1B).
Overexpression of mFz1 Represses Activity of BMP-2 and Wnt3a-BMP-2 is known to induce the osteoblast phenotype in distinct pluripotent mesenchymal cell lines including C2C12, C3H10T1/2, and ST2. To investigate the significance of the up-regulation of mFz1 by BMP-2, we examined whether mFz1 overexpression could affect the responsiveness of these cells to BMP-2. mFz1, hFz1, and Xfz3 were cloned and expressed as Myc-tagged proteins and showed comparable expression levels when transfected in C2C12 cells (Fig. 2C) or in the other cells used (data not shown). As shown in Fig. 2A, mFz1 significantly inhibits the BMP-2-induced ALP activity in C2C12. Similar data were obtained using ST-2 and C3H10T1/2 cells (data not shown), suggesting that mFz1 antagonizes the effects of BMP-2. We have recently demonstrated that BMP-2 controls ALP expression by a Wnt autocrine loop (32). As shown in Fig.  2A, mFz1 overexpression also inhibited Wnt3a-mediated ALP activity. Moreover mFz1 was also capable of significantly decreasing Wnt3a-mediated TCF-1 activation in COS cells (Fig.  2B). In contrast, overexpression of human Frizzled-1, Xfz3 (Fig.  2, A and B), or Xfz4 (data not shown) did not decrease but rather enhanced either BMP-2 or Wnt3a activities in our assays. These observations demonstrate that mFz1 antagonizes the activity of BMP-2 and Wnt3a in different cell lines.
Overexpression of mFz1 Blocks Nuclear ␤-Catenin Translocation in COS Cells-To investigate how mFz1 inhibits the activity of Wnt, we monitored nuclear ␤-catenin translocation induced by Wnt3a in the presence or absence of mFz1. COS cells were transfected with either a control vector expressing GFP or a construct that expresses both GFP and mFz1, and then cells were treated with Wnt3a-conditioned medium. Endogenous cellular ␤-catenin localization was determined by immunocytochemistry, and exogenous mFz1 expression was determined by GFP fluorescence. As shown in Fig. 3, in cells transfected with GFP expression vector ␤-catenin is located both in the cytoplasm and at the membrane but not in the nucleus (Fig. 3A). Wnt3a treatment resulted in nuclear localization of ␤-catenin (Fig. 3B). Importantly mFz1 transfection precluded the Wnt3a-induced ␤-catenin nuclear localization  mFz1 Inhibits Wnt/␤-Catenin Signaling in Xenopus Embryos-To further characterize the effect of mFz1 in Wnt/␤catenin signaling, we used the Xenopus animal cap assay in which activation of the pathway induces the expression of the target gene Siamois (33,34). Synthetic mRNAs were injected into the animal pole region at the four-cell stage, and animal cap explants were dissected at blastula stage and cultured to early gastrula stage for analysis of Siamois expression by RT-PCR. Injection of mFz1 mRNA (2 ng) alone had no effect on Siamois expression, whereas injection of Wnt1 mRNA (1 pg) strongly induced its expression in animal cap explants (Fig.  4A). Co-injection of mFz1 (2 ng) with Wnt1 (1 pg) mRNAs prevented markedly the induction of Siamois expression by Wnt1 (Fig. 4A). Importantly and in contrast with mFz1, coinjection of the same amounts of mRNAs encoding other Frizzled receptors such as Xfz3, Xfz4, and Xfz7 synergized with Wnt1 to induce Siamois expression (Fig. 4A). Expression of injected Frizzled mRNAs in Xenopus embryos was confirmed by immunoblotting using anti-Myc antibody. Furthermore mFz1 did not affect ␤-catenin expression monitored by immunoblotting (data not shown), thus confirming similar observations made in eukaryotic cells (see Fig. 2C).
We then examined the effect of mFz1 in whole embryos. Ventrovegetal injection of Wnt1 mRNA induces the formation of a complete secondary axis with anterior structures including eyes and cement gland (Fig. 4, B, C, and E). Co-injection of mFz1 with Wnt1 resulted in embryos with essentially partial secondary axis characterized by the absence of anterior tissues such as eyes and cement gland (Fig. 4, D and E). In contrast to other Frizzled receptors such as Xfz3, Xfz4, and Xfz7, dorsovegetal or ventrovegetal injection of mFz1 had no effect on gastrulation movements (data not shown). Thus the inhibitory effect of mFz1 on secondary axis formation induced by Wnt1 did not result from gastrulation defects. Altogether both in vitro and in vivo analyses suggest that mFz1 exerts a negative effect on Wnt/␤-catenin signaling.
The C-terminal Cytoplasmic Domain of mFz1 Is Capable of Transducing Wnt/␤-Catenin Signal-In an effort to determine whether the inhibitory effect of mFz1 on Wnt/␤-catenin signaling could be due to its inability to signal while still being capable of binding Wnts, we tested the ability of the C-terminal cytoplasmic region to activate the Wnt/␤-catenin signaling pathway. To examine this possibility, we generated various chimeric Frizzled receptors. Xfz3N/mFz1C is composed of the CRD and the seven transmembrane domains of Xfz3 fused to the C-terminal cytoplasmic region of mFz1, while mFz1N/ Xfz3C is the converse. Xfz3CRD/mFz1C is composed of the CRD of Xfz3 fused with the seven transmembrane domains and the C-terminal cytoplasmic region of mFz1 (Fig. 5A). The expression of our constructs was confirmed by injection of synthetic mRNAs encoding these chimeric receptors in Xenopus embryos and immunoblotting using anti-Myc antibodies. All the constructs used were found to be expressed at comparable levels (data not shown). Synthetic mRNAs encoding these chi-

FIG. 4. Effect of mFz1 on Wnt/␤-catenin signaling in Xenopus embryo. A, RT-PCR analysis of Wnt/␤-catenin target gene
Siamois expression in ectodermal explants. Embryos at the two-cell stage were injected with the indicated mRNAs near the animal pole region. Ectodermal explants were dissected at blastula stage and cultured to early gastrula stage for RT-PCR. Controls were performed with uninjected embryos either using the whole embryo without or with the RT reaction (Embryo/RTϪ and Embryo, respectively) or only ectodermal explant (Uninjected). Ornithine decarboxylase (ODC) was used as a loading control. B, C, D, and E, mFz1 blocks secondary axis formation induced by Wnt1. B, control embryo at larval stage. C, an embryo with a complete secondary axis induced by ventral injection of Wnt1 mRNA. D, co-injection of mFz1 with Wnt1 blocks the effect of Wnt1 and results in an embryo with a partial secondary axis. E, the results are expressed as total number of injected embryos, and the scoring of normal, complete secondary axis or partial secondary axis phenotypes for each conditions is shown. The experiment was repeated three times; data from one representative experiment are presented. mFz1 Inhibits Wnt Signaling meric receptors were then either injected in Xenopus embryos alone or co-injected with Wnt1 as above. We have previously shown that Xfz3 alone is able to activate Wnt/␤-catenin signaling in the absence of exogenous ligands (29). Like Xfz3, injection of Xfz3N/mFz1C mRNA (1 ng) induced Siamois expression (Fig. 5B). These results demonstrate that the C-terminal cytoplasmic tail of mFz1 is functional in transducing Wnt/␤-catenin signal. This is consistent with the fact that this region contains a conserved Lys-Thr-X-X-X-Trp motif necessary for Wnt/␤-cate-nin signaling (29). In contrast, injection of mFz1N/Xfz3C mRNA (1 ng) failed to induce Siamois expression (Fig. 5B). When co-injected with Wnt1, this construct indeed inhibited Siamois expression similar to inhibition by the wild-type mFz1 (Fig. 5B). The chimeric receptor Xfz3CRD/mFz1C had no effect on Siamois expression either alone or co-injected with Wnt1 (Fig. 5B). Similar results were obtained when those chimeric receptors were tested for their capacity to interfere with Wnt3a signaling using the TOPflash reporter assay in COS-7 cells (data not shown). Thus, mFz1 negatively regulates Wnt/␤catenin signaling independently of the C-terminal cytoplasmic region.
Inhibition of Wnt/␤-Catenin Signaling by Different Truncated Forms of mFz1-To further examine the mechanism by which mFz1 negatively regulates the Wnt/␤-catenin pathway, we generated different truncated forms of mFz1 (see Fig. 6A). The expression of the truncated forms of mFz1 was confirmed by immunoblotting in both Xenopus embryos and COS cells (data not shown). mFz1⌬C, which retains only seven residues Siamois expression similar to wild-type Xfz3. MFz1N/Xfz3 that is the converse of Xfz3N/mFz1C inhibited Siamois expression induced by Wnt1. Xfz3CRD/mFz1C that is composed of the Xfz3 CRD and the mFz1 seven-transmembrane domain (7TM) and the C-terminal cytoplasmic tail had no effect on Wnt1. ODC, ornithine decarboxylase; RTϪ, no reverse transcription reaction.
FIG. 6. Effect of truncated mFz1 on Wnt/␤-catenin signaling in Xenopus explants. A, schematic representation of wild-type and truncated mFz1 (see text for details). B, RT-PCR analysis of Wnt/␤-catenin target gene Siamois expression in ectodermal explants. Notice that mFz1⌬C inhibits more efficiently Wnt1-induced Siamois expression than the wild-type receptor, whereas mFz1⌬N has no effect. ODC, ornithine decarboxylase; RTϪ, no reverse transcription reaction; 7TM, seven transmembrane domain.

FIG. 7. Effect of C-and N-terminal truncated murine
Frizzled-1 on Wnt3a activity. A, COS-7 cells were transiently co-transfected with TCF expression construct, TOPflash, and pTK-Renilla. Where indicated either empty vector, wild-type mFz1, C-terminal truncated murine Frizzled-1 (mFz1⌬C), or N-terminal truncated murine Frizzled-1 (mFz1⌬N) expression constructs were added to the co-transfection mixture. Wnt3a-CM was added to cells 18 h after transfection, and then luciferase activity was determined in cell lysates 24 h later and normalized to Renilla signal. B, C2C12 cells were transiently transfected with either empty vector, wild-type mFz1, C-terminal truncated murine Frizzled-1 (mFz1⌬C), or N-terminal truncated murine Frizzled-1 (mFz1⌬N) expression constructs. Wnt3a-CM was added to cells 18 h after transfection, and then ALP activity was measured in cell lysates 72 h later, normalized to protein content, and expressed as nmol of p-nitrophenyl phosphate (pnpp)/min/g of protein (prot). All experiments were performed in triplicate and repeated three times. Data Ϯ S.D. from one representative experiment are presented.
(Ser-Gly-Lys-Thr-Leu-Asn-Ser) after the seventh transmembrane domain, thus lacking the conserved Lys-Thr-X-X-X-Trp motif (29), inhibited Wnt1-induced Siamois expression more efficiently than did the wild-type mFz1 (Fig. 6B). In contrast, mFz1⌬N in which the N-terminal CRD is absent failed to inhibit Siamois expression induced by Wnt1 (Fig. 6B). The activity of mFz1⌬C and mFz1⌬N constructs was further investigated in COS-7 cells using the TOPflash reporter system. Wild-type mFz1 as well as mFz1⌬C were able to inhibit Wnt3ainduced luciferase activity, whereas mFz1⌬N did not affect TOPflash activation in response to Wnt3a (Fig. 7A). Similar results were obtained when we measured ALP activity induced by Wnt3a in C3H10T1/2 cells transiently transfected with the different constructs (Fig. 7B). These results further indicate that the inhibitory effect of mFz1 on Wnt/␤-catenin signaling is mediated by the extracellular CRD.
G␣ q Signaling Mediates mFz1 Inhibition of Wnt Pathway-Frizzled proteins are G protein-coupled receptors that appear to act through heterotrimeric guanine nucleotide-binding proteins (G proteins) (35). Recently G q signaling has been showed to inhibit Wnt/␤-catenin pathway (36). We have therefore investigated whether the effect of G␣ q -(305-359), the C-terminal peptide of the ␣-subunit G␣ q that specifically uncouples G qcoupled receptors (28), in the inhibitory activity is mediated by mFz1. As shown in Fig. 8A, overexpression of G␣ q -(305-359) almost totally restores Wnt3a activity in the presence of mFz1 in COS-7 cells. G␣ q stimulates phospholipase-C␤ that subsequently causes calcium release from internal stores. We have assessed the effect of phospholipase-C␤ inhibitor U-73122 and the cell-permeable calcium chelator 1,2-bis(2-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetate acetoxymethyl ester (BAPTA/ AM). Both U-73122 and BAPTA/AM were able to relieve the mFz1 inhibition of Wnt3a activity (Fig. 8B). To determine that the effects of G␣ q -(305-359), U-73122, or BAPTA/AM were not due to differences in the expression of ␤-catenin or mFz1, we looked at the expression of ␤-catenin and mFz1 both in cells overexpressing G␣ q -(305-359) and in cells treated with U-73122 or BAPTA/AM, and no differences compared with control cells were found (Fig. 8C). These data strongly suggest that G␣ q signaling contributes to the inhibition of Wnt/␤-catenin pathway mediated by mFz1.

DISCUSSION
The Wnt family of secreted signaling molecules binds to two co-receptors: the Frizzled-type receptor and LRP5/6. The Frizzled gene family encodes proteins that have a seven-transmembrane-spanning motif characteristic of receptors that couple to G proteins (G protein-coupled receptors). To date there are at least nine members identified in mammals (10). In a similar way to Wnt proteins, vertebrate Frizzled homologues have been shown to activate distinct signaling pathways. The downstream targets of Frizzled actions include the canonical Wnt pathway, which acts through ␤-catenin, and the Ca 2ϩ pathway in which Frizzled appears to act through G proteins (for a review, see Ref. 37). In this study we demonstrated that in contrast with other Frizzled family members, mFz1 overexpression leads to the repression of Wnt/␤-catenin canonical signaling. Actually overexpression of mFz1 significantly represses TCF signaling mediated by either Wnt3a or a constitutive active mutant form of ␤-catenin and precludes Wnt3ainduced ␤-catenin nuclear localization in COS-7 cells. Moreover microinjection of mFz1 transcript in Xenopus embryo inhibits the ability of Wnt1 to induce the expression of the Wnt/␤catenin target gene Siamois in animal cap assay and secondary axis formation in whole embryo.
Previous reports have suggested that some Wnts display inhibitory effects on other Wnts. Mainly Wnt5 was shown to block Wnt8 promoting dorsal cell fate in Xenopus (38) and to activate nemo-like kinase, which inhibits TCF transcriptional activity (39). Very recently, genetic evidence has demonstrated the antagonistic activity of Wnt5a toward ␤-catenin signaling (40,41), thus confirming that Wnt signaling pathways can act in an antagonistic manner. Interestingly the Wnt5a antagonistic activity is Ca 2ϩ -dependent, and therefore one can speculate that mFz1 could be acting as a receptor for Wnt5a or other potential "antagonistic Wnts." Further investigations are required to elucidate this possibility.
Vertebrate Frizzled homologues have been shown to activate distinct signaling pathways. Some Frizzled receptors induce the expression of target genes of ␤-catenin when overexpressed in Xenopus ectodermal cells, whereas others stimulate calcium release and protein kinase C activation (13,42,43). Nevertheless the lack of activity of a Frizzled receptor in Wnt/␤-catenin signaling in a particular context may be due to the absence of an appropriate ligand. For example, Xfz3 activates Wnt/␤-catenin in the absence of exogenous ligands, while Xfz4 and Xfz7 interact with Xwnt5A to activate this pathway (29). Two Frizzleds have been previously found to reduce the activity of Wnt: Drosophila Dfz3 (44) and Caenorhabditis elegans MOM-5 (45,46). These two Frizzleds do not display the Lys-Thr-X-X-X-Trp motif conserved among all Frizzled receptors reported to be required for activation of the Wnt/␤-catenin pathway and for membrane relocalization and phosphorylation of Dishevelled (29). Dfz3 and mom-5 may therefore represent two naturally occurring "defective" Wnt receptors that reduce Wnt/␤-catenin signaling. In contrast to Dfz3 and mom-5, mFz1 contains the conserved Lys-Thr-X-X-X-Trp motif in its C-terminal cytoplasmic tail. In fact, the rat and human Frizzled-1 counterparts have been reported to positively affect Wnt/␤-catenin signaling in different models (13,43,47). Although the structure of mFz1 is not distinguishable from that displayed by the closest homologues, human or rat Frizzled-1, 2 subtle differences, especially in the extracellular region, exist. Interestingly the data obtained with chimeric receptors show that the antagonistic activity of mFz1 resides in the cysteine-rich domain of the Nterminal part of mFz1, while the intracellular portion of the receptor is capable of activating Wnt/␤-catenin. Recently it was demonstrated that the CRD of certain Frizzled receptors (Frizzled-3 and Frizzled-8) exhibit the dimerization potential, which may be a feature of Wnt/␤-catenin signaling (48). Therefore, it would be interesting to evaluate the capacity for dimerization of mFz1.
It is important to note that mFz1 not only blocks ␤-catenin nuclear translocation induced by Wnt, but it also dramatically reduces the transcriptional activity of a constitutive active mutant form of ␤-catenin. Previous studies have implicated calcium signaling in the regulation of Wnt/␤-catenin pathway, and Li et al. (36) have recently shown that the G q pathway inhibits the Wnt/␤-catenin signaling at the level of ␤-catenin. Importantly the G q pathway promotes ␤-catenin nuclear export. Interestingly we have shown that the antagonist activity of mFz1 could be prevented by overexpression of G␣ q -(305-359), which specifically uncouples G q -coupled receptors (28). Moreover the phospholipase-C␤ inhibitor U-73122 and the cellpermeable calcium chelator BAPTA/AM were able to relieve the mFz1 inhibition of Wnt3a. These data strongly suggest that G q signaling contributes to the Wnt/␤-catenin inhibitory activity displayed by mFz1.
BMP-2 is a potent activator of osteoblast differentiation, and it induces the expression of a number of osteoblast differentiation markers in distinct pluripotent mesenchymal cell lines. We have shown here that mFz1 expression is up-regulated by BMP-2 and that mFz1 overexpression represses BMP-2induced osteoblast differentiation. The regulation of mFz1 by BMP-2 is specific because another member of the TGF-␤ family, TGF-␤1, had little effect on its expression. Although the present data did not allow us to conclude whether mFz1 represents a direct target gene of BMP-2, it is likely that the increase in mFz1 expression level may be a result of BMP-2induced osteoblast differentiation. Indeed the time course analysis showed that the up-regulation of mFz1 was observed 24 h after BMP-2 treatment and was maintained until 4 days of culture. Our functional analyses suggest that this up-regulation may serve as a fine tuning of BMP-2 activity. Transfection experiments in different mesenchymal cell lines clearly show that mFz1 inhibits BMP-2-induced ALP expression. These data strongly suggest that the activity of mFz1 on BMP-2 is not direct but mediated by inhibition of Wnt/␤-catenin signaling. Consistent with this hypothesis is the fact that mFz1 also inhibits the induction of ALP by Wnt3a. Moreover we have recently reported that dominant negative forms of the Wnt co-receptor LRP5 are capable of inhibiting the BMP-2-mediated ALP induction in mesenchymal cells (20). There are several lines of evidence showing that Wnt/␤-catenin signaling plays a crucial role in bone biology. In particular, loss of function of the Wnt co-receptor LRP5 affects bone development both in mouse and human (20 -22). One possibility is that other Frizzled receptors cooperate with LRP5 to activate Wnt/␤-catenin signaling during bone development, whereas mFz1 represents a feedback mechanism to modulate this signal. Preliminary experiments show that overexpression of LRP5 does not overcome mFz1 inhibition, indicating that mFz1 is not interfering with LRP5 signaling to inhibit the ␤-catenin pathway.
In summary our data show that mFz1 is capable of antagonizing Wnt/␤-catenin. The induction of this receptor by BMP-2 in mesenchymal cells may represent a feedback mechanism to modulate the activity of BMP-2.