Wnt Signaling Inhibits Adipogenesis through β-Catenin-dependent and -independent Mechanisms*

Wnt signaling has been reported to block apoptosis and regulate differentiation of mesenchymal progenitors through inhibition of glycogen synthase kinase 3 and stabilization of β-catenin. The effects of Wnt in preadipocytes may be mediated through Frizzled (Fz) 1 and/or Fz2 as these Wnt receptors are expressed in preadipocytes and their expression declines upon induction of differentiation. We ectopically expressed constitutively active chimeras between Wnt8 and Fz1 or Fz2 in preadipocytes and mesenchymal precursor cells. Our results indicated that activated Fz1 increases stability of β-catenin, inhibits apoptosis, induces osteoblastogenesis, and inhibits adipogenesis. Although activated Fz2 does not influence apoptosis or osteoblastogenesis, it inhibits adipogenesis through a mechanism independent of β-catenin. An important mediator of the β-catenin-independent pathway appears to be calcineurin because inhibitors of this serine/threonine phosphatase partially rescue the block to adipogenesis caused by Wnt3a or activated Fz2. These data supported a model in which Wnt signaling inhibits adipogenesis through both β-catenin-dependent and β-catenin-independent mechanisms.

Wnts are a family of secreted proteins that act through paracrine and autocrine mechanisms to regulate many aspects of cell fate and development (1). Binding of Wnts to frizzled (Fz) 1 receptors and low density lipoprotein receptor-related protein (LRP) coreceptors activates several signaling pathways (2,3). In the ␤-catenin-dependent pathway (i.e. canonical), Wnt signaling inhibits glycogen synthase kinase 3, resulting in hypophosphorylation and subsequent stabilization of ␤-catenin in cytoplasm. After translocation to the nucleus, ␤-catenin binds to and coactivates members of the T-cell factor/lymphoid-enhancing factor (TCF/LEF) family as well as other transcriptions factors (4). Wnt signaling through ␤-catenin has been associated with Wnt family members such as Wnt1, Wnt3a, and Wnt10b and frizzled receptors such as Fz1 and Fz5 (1). In the ␤-catenin-independent pathways (i.e. noncanonical), signaling is through increased calcium flux, activation of protein kinase C, inhibition of cGMP phosphodiesterase, or regulation of other effectors (3). Signaling through ␤-catenin-independent pathways has been associated with Wnt5a and Wnt11 and frizzled receptors such as Fz2 and Fz7.
Wnts signaling through the ␤-catenin-dependent pathway has profound effects on differentiation of mesenchymal progenitors. For example, stabilization of ␤-catenin stimulates osteoblastogenesis and inhibits adipogenesis (5)(6)(7) and regulates other aspects of cell fate, including apoptosis (8,9). Inhibition of ␤-catenin signaling with a dominant negative TCF causes spontaneous adipogenesis (6), suggesting that endogenous Wnts act to repress preadipocyte differentiation. Wnt10b is a candidate for the endogenous inhibitory Wnt because this protein stabilizes ␤-catenin and inhibits adipogenesis. Furthermore endogenous Wnt10b is expressed in preadipocytes and declines upon induction of differentiation (10).
The genetic program for adipogenesis has been studied extensively, and a paradigm for the cascade of genetic events has emerged (11)(12)(13)(14). In response to inducers of adipogenesis, there is a transient induction of CCAAT/enhancer-binding protein ␤ (C/EBP␤) and C/EBP␦, which independently activate expression of C/EBP␣ and peroxisome proliferator-activated receptor ␥ (PPAR␥). These master adipogenic transcription factors regulate each other's expression through a positive feedback loop. Activation of ␤-catenin-dependent signaling by Wnt10b, inhibition of glycogen synthase kinase 3, or expression of dominant stable ␤-catenin inhibits adipogenesis by blocking expression of C/EBP␣ and PPAR␥ (6,10).
Inhibitory Wnt signals may be mediated by Fz1, Fz2, or Fz5 as expression of these Wnt receptors is high in preadipocytes and declines upon induction of differentiation (10). Fz family members are seven-transmembrane proteins, and recent reports suggest that they act as G-protein-coupled receptors (15). Fz receptors contain an extracellular cysteine-rich domain, which interacts with Wnt ligands, and an intracellular carboxyl-terminal portion, which is essential for ␤-catenin-dependent signaling. Considerable evidence suggests that Fz1 initiates ␤-catenin-dependent signaling, whereas Fz2 activates the ␤-catenin-independent pathways (16 -20).
In the current study, we used chimeras between Xenopus Wnt8 and mouse Fz1 or Fz2 to investigate the effects of these Wnt receptors on mesenchymal cell fate. Previous studies have demonstrated the ability of specific Wnt-Fz chimeras to initiate ␤-catenin-dependent signaling (21,22). We report that activated Fz1 is sufficient to stabilize ␤-catenin, inhibit apoptosis, and stimulate osteoblastogenesis and that activated Fz2 does not influence these cellular events. Although activated Fz1 or Fz2 partially inhibit preadipocyte differentiation, both are re-quired to mimic the complete block to adipogenesis observed with Wnt10b. These data suggested that Wnts inhibit adipocyte conversion through ␤-catenin-dependent and -independent mechanisms. Calcineurin, a calcium-dependent serine/ threonine phosphatase, appears to be an important mediator of the ␤-catenin-independent pathway because inhibitors of this phosphatase partially rescue repression of adipogenesis caused by Wnt3a or activated Fz2. This study suggested that some effects of Wnt on mesenchymal cell fate can be attributed to ␤-catenin-dependent signaling by Fz1, whereas other effects, such as inhibition of adipogenesis, involve activation of more than one Fz and signaling pathway.

EXPERIMENTAL PROCEDURES
Reagents-FK506 and cyclosporin A (Calbiochem) were dissolved in ethanol. Recombinant mouse Wnt3a was dissolved in phosphate-buffered saline containing bovine serum albumin (R&D systems). Troglitazone (Pfizer) was dissolved in dimethyl sulfoxide.
Cell Culture-Mouse 3T3-L1 preadipocytes and human embryonic kidney 293T cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% calf serum (Atlanta Biologicals) as described previously (23). ST2 marrow-derived stromal cells were cultured in minimum essential medium ␣ (Invitrogen) containing 10% fetal calf serum (Cambrex) and ascorbic acid (Sigma). 3T3-L1 cells were induced to differentiate into adipocytes 2 days after confluence as described previously, using methylisobutylxanthine, dexamethasone, and insulin (24). Where indicated, 5 M troglitazone was added for the first 4 days of adipogenesis. To visualize lipid accumulation, 3T3-L1 cells were stained with Oil Red-O (25). Alkaline phosphatase activity in ST2 cells was measured 2 days after confluence using a histochemical kit (Sigma). To induce mineralization, ST2 cells were seeded on 3.5-cm plates coated with 1% gelatin. At 2 days after confluence, ST2 cells were treated with growth medium containing ␤-glycerophosphate (Sigma) for 28 days. Mineralization was visualized by von Kossa staining (26). Briefly, ST2 cells were fixed with 10% buffered formalin, and incubated in the presence of 5% silver nitrate solution under an ultraviolet light for 1 h, and then incubated for 5 min in 5% sodium thiosulfate solution.
Wnt-Fz Expression Vectors-A chimera between a portion of Xenopus Wnt8 and human Fz5 in pCS2ϩ was obtained from Bart Williams (Van Andel Institute, Grand Rapids, MI). Fz5 was replaced with a portion of Fz2 (amino acids 29 -570) lacking its putative signal peptide. Fz2 was amplified by PCR using a mouse Fz2 cDNA provided by Ramesh Shivdasani (Dana-Farber Cancer Institute, Boston, MA). The Wnt-Fz2 chimera was then subcloned into the retroviral vector, pTS13. A chimera between Wnt8 and a portion of Fz1 (amino acids 69 -642) lacking its putative signal peptide was similarly constructed by PCR using mouse Fz1 cDNA (I.M.A.G.E. clone 5697795, American Type Culture Collection). The Wnt-Fz1 chimera was then subcloned into retroviral vectors, pBabe and pTS13. Site-directed mutagenesis was performed using the Stratagene QuikChange kit. Stop codons were inserted prior to the cytosolic tail (Fz1, amino acid 619; Fz2, amino acid 547) or just after the first putative transmembrane domain (Fz1, amino acid 345; Fz2, amino acid 280).
TCF-responsive Reporter Gene Assay-Human embryonic kidney 293T cells (3.5-cm plates) were transiently transfected by calcium phosphate coprecipitation with 2 g of total DNA, including pTOPFLASH luciferase reporter gene (25 ng; Upstate Biotechnology), and expression vectors for human LRP6 (500 ng; Bart Williams, Van Andel Institute, Grand Rapids, MI) and CMV-␤-galactosidase (250 ng). Expression plasmids for human Wnt1 (500 ng), Wnt-Fz1 (500 ng), or Wnt-Fz2 (500 ng) were cotransfected as indicated. A constant amount of cytomegalovirus promoter (pcDNA3.1; Invitrogen) was maintained to control for potential squelching of transcriptional machinery. After transfection, cells were incubated for 48 h and lysed. Luciferase activity was measured, and variations in transfection efficiency were accounted for by normalization to ␤-galactosidase activity as described previously (27).
Retroviral Infection of 3T3-L1 Preadipocytes-293T cells (10-cm plates) were transfected by calcium phosphate coprecipitation with the viral packaging vectors SV-E-MLV-env and SV-E-MLV in addition to retroviral vectors as indicated in the figure legends (7.5 g of each). The S33Y ␤-catenin retrovirus and its control (pNeo) were provided by Eric Fearon (University of Michigan, Ann Arbor, MI). Virus-containing medium was collected 16 h after transfection and passed through a 0.45-m syringe filter. Polybrene (hexadimethrine bromide; Sigma) was added to a final concentration of 8 g/ml. This medium was then applied to subconfluent (30%) 3T3-L1 preadipocytes (10-cm plates). The infection protocol was repeated every 8 -16 h until cells were 80% confluent. Cells were then trypsin-treated and replated in Dulbecco's modified Eagle's medium supplemented with 10% calf serum and the appropriate antibiotic: 150 g/ml hygromycin (Invitrogen) for pTS13-based vectors, 400 g/ml Geneticin (Invitrogen) for pNeo-based vectors, or 2 g/ml puromycin (Sigma) for pBabe-based vectors.
TUNEL Assay-3T3-L1 cells were grown on 18-mm-square glass coverslips. After confluence, cells were incubated for 24 h in serum-free medium (containing 0.5% bovine serum albumin). Cells were fixed with 4% paraformaldehyde and permeabilized in 0.1% sodium citrate and 1% Triton X-100. Nicked DNA was end-labeled with fluorescein-conjugated dNTPs (Roche Applied Science) for 1 h at 37°C. Cells were washed with phosphate-buffered saline and then counterstained with a nuclear dye, Hoechst 33342 (Sigma). Coverslips were mounted on slides, and six random fields for each treatment were analyzed using a Nikon TE200 fluorescent microscope. The total cell number (Hoechst 33342-labeled nuclei) and apoptotic cell number (fluorescein-labeled nuclei) from each field were used to calculate the apoptotic index for each treatment, which is defined as: (number of fluorescein-labeled nuclei)/(number of Hoechst-labeled nuclei) Ϫ 100. Image analysis was performed using the public domain NIH Image program (rsb.info.nih.gov/nih-image/).
Fractionation of Cells-Two days after reaching confluence, 3T3-L1 cells were washed with phosphate-buffered saline and scraped into hypotonic lysis buffer (10 mM Tris-HCl, pH 7.5; 140 mM NaCl; 5 mM EDTA; 2 mM dithiothreitol; and protease inhibitors). Cells were homogenized by 30 strokes of a Dounce homogenizer, and the lysate was centrifuged at 1000 ϫ g for 10 min at 4°C. Membrane and cytosolic fractions were obtained by ultracentrifugation at 100,000 ϫ g for 90 min at 4°C. The supernatant was designated the cytosolic fraction. Pellets resuspended by sonication in hypotonic lysis buffer containing 0.1% SDS were designated the membrane fraction. Equal amounts of total cytosolic protein or equal volumes of the membrane fraction were separated by SDS-PAGE and transferred onto nitrocellulose membranes. Immunoblots were performed with antibodies specific for ␤-catenin and cyclin D1 (BD Transduction Laboratories) and ␤-tubulin (Sigma). Anti-Myc-horseradish peroxidase (Invitrogen) was used to detect Wnt-Fz chimeras in the membrane fractions.
Preparation of Whole Cell Extracts-Cells were washed with phosphate-buffered saline and scraped into lysis buffer (50 mM Tris-HCl. pH 7.5; 150 mM NaCl; 2 mM EGTA; 0.1% Triton X-100; and protease inhibitors) and incubated for 30 min on ice. Lysates were centrifuged at 16,000 ϫ g for 10 min at 4°C. Equal amounts of total protein were separated by SDS-PAGE and transferred onto nitrocellulose membranes. Immunoblots were performed with antibodies specific for C/EBP␤ (M. D. Lane, The Johns Hopkins University, Baltimore, MD), C/EBP␣ and PPAR␥ (Santa Cruz Biotechnology), laminin (Sigma), and FABP4 (David Bernlohr, University of Minnesota, Minneapolis, MN).

Activation of ␤-Catenin-dependent Wnt Signaling by Fz1 but
Not Fz2-Our prior work indicates that Fz1 and Fz2 are expressed in 3T3-L1 preadipocytes and stromal-vascular cells but not in adipocytes (10). To investigate the role each Fz might have in maintaining the preadipocyte state, we constructed chimeras between a portion of Xenopus Wnt8 and mouse Fz1 or -2. Previous studies have shown that Wnt-Fz chimeras increase canonical Wnt signaling, depending on which Fz is fused to Wnt (21,22). As controls, we generated truncated Wnt-Fz chimeras that lack regions carboxyl-terminal to the last transmembrane domain (Wnt-Fz⌬Tail; Fig. 1A) or the first transmembrane domain (Wnt-Fz⌬Cyt). These chimeras were stably expressed in 3T3-L1 preadipocytes by retroviral infection and were present in the isolated membrane fractions at the predicted size (Fig. 1B).
Previous studies have suggested that activation of rat Fz1 stimulates ␤-catenin-dependent signaling, whereas rat Fz2 initiates ␤-catenin-independent signaling (28). To determine whether activated forms of Fz1 or Fz2 initiate ␤-catenin signaling, we isolated cytosolic fractions from confluent 3T3-L1 preadipocytes stably expressing the Wnt-Fz chimeras or Wnt10b. As expected, levels of free cytosolic ␤-catenin increased with expression of Wnt10b (Fig. 1C). ␤-Catenin levels were also increased in cells expressing Wnt-Fz1. However, levels of ␤-catenin were not altered by expression of Wnt-Fz2. To verify increased activity of ␤-catenin, we examined expression of cyclin D1, a known target of ␤-catenin signaling (29,30). As expected, increased levels of cyclin D1 were observed in 3T3-L1 preadipocytes expressing Wnt10b or Wnt-Fz1 but not in cells expressing Wnt-Fz2 (Fig. 1C). These results suggested that activated Fz1 stabilizes ␤-catenin and increases expression of a known target of ␤-catenin, whereas activated Fz2 does not.
␤-Catenin is a coactivator of TCF/LEF transcription factors (4). To determine whether Wnt-Fz chimeras increase TCFresponsive promoter activity, we performed luciferase reporter assays with pTOPFLASH, a TCF-responsive reporter gene. Expression constructs for Wnt1, Wnt-Fz1, and Wnt-Fz2 were transiently transfected into 293T cells along with pTOPFLASH and the Wnt coreceptor LRP6 (Fig. 1D). As expected, ectopic expression of Wnt1 or Wnt-Fz1 increased TCF-dependent reporter activity, whereas expression of Wnt-Fz2 did not. Taken together, these results suggested that ␤-catenin-dependent signaling is activated by Fz1 but not by Fz2.
Apoptosis of Preadipocytes Is Inhibited by Wnt-Fz1 but Not by Wnt-Fz2-Wnt-Fz1 stabilizes ␤-catenin and increases expression of a known target of ␤-catenin signaling (Fig. 1, C  and D). We have previously shown that ␤-catenin-dependent Wnt signaling prevents apoptosis of preadipocytes (9). To determine whether activated Fz1 or Fz2 inhibits apoptosis, 3T3-L1 preadipocytes stably expressing Wnt10b, Wnt-Fz1, or Wnt-Fz2 were evaluated for apoptosis induced by serum deprivation (Fig. 2). As expected, expression of Wnt10b protected preadipocytes from apoptosis after incubation for 24 h in serum-free medium. Expression of Wnt-Fz1 also protected preadipocytes from apoptosis, whereas Wnt-Fz2 had no effect. Protection from apoptosis in response to serum deprivation correlates with the level of stabilized ␤-catenin within the cells (Fig. 1C). These results suggested that activated Fz1, but not Fz2, initiates ␤-catenin-dependent signaling to block preadipocyte apoptosis.
Induction of Osteoblastogenesis by Wnt-Fz1 but Not by Wnt-Fz2-We have observed in ST2 marrow-derived stromal cells that Wnt signaling increases alkaline phosphatase activity and mineralization through a ␤-catenin-dependent mechanism (31). Thus, to determine whether activated Fz1 or -2 is sufficient to stimulate osteoblastogenesis, we treated 3T3-L1 preadipocytes expressing Wnt10b, Wnt-Fz1, Wnt-Fz2, or a dominant stable form of ␤-catenin with mineralization media for 2 weeks followed by staining for alkaline phosphatase activity (Fig. 3A). Although 3T3-L1 preadipocytes are committed to the adipocyte lineage (32), expression of Wnt10b or a dominant stable form of ␤-catenin was sufficient to induce alkaline phosphatase activity, an early marker of osteoblast differentiation (Fig. 3A). Although expression of Wnt-Fz1 also increased alkaline phosphatase staining, this was not observed with Wnt-Fz2. Once again, these results suggested that activated Fz1, but not activated Fz2, increases ␤-catenin-dependent signaling in 3T3-L1 preadipocytes.
Although ␤-catenin-dependent signaling induces early stages of osteoblastogenesis, the 3T3-L1 preadipocyte model is not ideal because Wnt signaling is not sufficient to stimulate later stages of osteoblast differentiation. Therefore, we decided to characterize the effects of Wnt-Fz chimeras in ST2 cells, which are mesenchymal precursors with the potential to undergo osteoblastogenesis (33). To determine whether activated Fz1 increases osteoblastogenesis in ST2 cells, we assayed con- 3T3-L1 preadipocytes were infected with control retrovirus (pTS13) or a retrovirus expressing the indicated Wnt-Fz chimera. Two days after confluence, cytosolic and membrane fractions were isolated, and Wnt-Fz chimeras in the membrane fraction (B) were detected by immunoblot for the Myc epitope. Expression of ␤-catenin, cyclin D1, and ␤-tubulin (loading control) in the cytosolic fraction (C) was analyzed by immunoblot. D, 293T cells were transfected with pTOPFLASH, which is a TCF-responsive reporter gene, along with an expression vector for LRP6. Expression vectors for Wnt-1, Wnt-Fz1, or Wnt-Fz2 were cotransfected as indicated. Samples were normalized to ␤-galactosidase activity to correct for variations in transfection efficiency. Luciferase activity is reported as activation (mean ϩ standard deviation) relative to pTOPFLASH with LRP6 alone. fluent ST2 cells expressing Wnt10b, Wnt-Fz1, Wnt-Fz2, or dominant stable ␤-catenin for alkaline phosphatase activity. Similar to the results observed with 3T3-L1 preadipocytes, expression of Wnt10b, dominant stable ␤-catenin, or Wnt-Fz1 increased alkaline phosphatase activity, whereas Wnt-Fz2 had no effect (Fig. 3B). In addition, Wnt10b, dominant stable ␤-catenin, or Wnt-Fz1 increased the mineralization of ST2 cells treated with mineralization media, whereas Wnt-Fz2 did not (Fig. 3C). These results further suggested that activated Fz1, but not Fz2, mimics ␤-catenin-dependent Wnt signaling in both 3T3-L1 and ST2 cells.
Both Wnt-Fz1 and Wnt-Fz2 Inhibit Adipogenesis of ST2 and 3T3-L1 Cells-Our results indicated that activated Fz1, as expected, stabilizes ␤-catenin (Fig. 1C) and initiates ␤-catenindependent signaling to inhibit apoptosis and increase osteoblastogenesis (Figs. 2 and 3 (A-C)). However, Fz2 did not influence these cellular events, suggesting either that Wnt-Fz2 does not initiate ␤-catenin-dependent signaling or that the chimera is inactive. To investigate further the biological effects of Fz1 and Fz2, we examined their ability to regulate adipogenesis. Prior work demonstrated that expression of a dominant stable form of ␤-catenin partially inhibited preadipocyte differentiation, suggesting that the repressive effects of Wnt signaling are mediated, at least in part, through this pathway (6,34). We induced adipocyte differentiation of ST2 cells expressing Wnt-Fz chimeras and observed that both Wnt-Fz1 and Wnt-Fz2 were able to partially inhibit adipogenesis as assessed by decreased lipid accumulation (Fig. 3D). The conversion of ST2 cells to adipocytes is less robust than 3T3-L1 and other preadipocyte models. Thus, visualization of an entire plate stained with Oil Red-O is not an effective method for assessing extent of adipogenesis. Inhibition of adipogenesis by Wnt10b, Wnt-Fz1, Wnt-Fz2, or ␤-catenin is reflected in reduced expression of PPAR␥ (Fig. 3D). Taken together, these data suggested that Wnt-Fz2 is active and inhibits adipogenesis of pluripotent progenitor cells.
To determine whether these chimeric proteins also inhibit differentiation of committed preadipocytes, we induced adipogenesis of 3T3-L1 cells expressing Wnt10b, Wnt-Fz1, Wnt-Fz2, or control chimeras. As expected, both Wnt10b and Wnt-Fz1 blocked adipogenesis as assessed by lipid accumulation and expression of PPAR␥ (Fig. 4A). Interestingly, Wnt-Fz2 also inhibited adipogenesis, although to a lesser extent. These results suggested that the Wnt-Fz2 chimera is indeed active but does not appear to require an increase in ␤-catenin signaling to have its effects on adipogenesis. Inhibition of adipogenesis was not observed with Fz chimera lacking portions of the carboxyl terminus (Wnt-Fz⌬Tail, Wnt-Fz⌬Cyt), suggesting that signaling occurs through the cytosolic tail (Fig. 4A).
To investigate a potential molecular mechanism for inhibition of adipogenesis, we examined the effects of Wnt-Fz1 or Wnt-Fz2 on expression of adipogenic transcription factors. PPAR␥ and C/EBP␣ are required for adipogenesis (35), and the expression of these transcription factors is largely dependent upon expression of C/EBP␤. Consistent with prior reports (6), Wnt10b inhibited expression of PPAR␥ and C/EBP␣ without affecting C/EBP␤ (Fig. 4B). Although Wnt-Fz1 or Wnt-Fz2 did not alter expression of C/EBP␤, their repression of PPAR␥ and C/EBP␣ correlated closely with adipogenesis, as assessed by lipid accumulation and expression of the adipocyte lipid-binding protein, FABP4.
Wnt-Fz1 and Wnt-Fz2 Inhibit Adipogenesis through Independent Mechanisms-Our results suggested that activated Fz2 inhibits adipogenesis through a mechanism independent of ␤-catenin (Figs. 1, 3, and 4). To investigate further the mechanisms of Fz1 and Fz2 action, we expressed Wnt-Fz1 and/or Wnt-Fz2 in 3T3-L1 cells and induced adipocyte differentiation (Fig. 5A). In this experiment, Wnt-Fz1 largely inhibited adipogenesis, whereas Wnt-Fz2 was not as potent. However, when coexpressed, Wnt-Fz1 and Wnt-Fz2 fully inhibited adipogenesis with only a few adipocytes observed upon microscopic examination of the plate. Importantly, Wnt-Fz2 did not increase free cytosolic ␤-catenin in control preadipocytes or cause a further increase in those expressing Wnt-Fz1 (Fig. 5B).
We have observed that a dominant stable form of ␤-catenin only partially inhibits adipogenesis, raising the possibility that Wnt10b blocks adipogenesis through an additional pathway (6,34). Thus, we examined the differentiation potential of 3T3-L1 preadipocytes that express dominant stable ␤-catenin and/or activated Fz2. Although expression of ␤-catenin or Wnt-Fz2 caused a partial block to preadipocyte differentiation, coexpression of these proteins resulted in an almost complete inhibition of adipogenesis (Fig. 5C), without a further increase in free cytosolic ␤-catenin (Fig. 5D). Taken together, these data provided further evidence that activated Fz2 inhibits adipogenesis through a pathway independent of ␤-catenin.
Calcineurin Inhibitors Partially Rescue the Block of Adipogenesis by Activated Fz2 or Wnt3a-Our results indicated that activated Fz1 regulates apoptosis (Fig. 2), osteoblastogenesis (Fig. 3, A-C), and adipogenesis (Figs. 3D and 4) through a mechanism dependent on ␤-catenin. Although activated Fz2 does not influence apoptosis or osteoblastogenesis, it inhibits adipogenesis through a mechanism independent of ␤-catenin. Previous studies have shown that Wnts can initiate ␤-catenin-independent signaling through calcium fluxes, decreased cGMP, rhoA, and other signaling events (3).

FIG. 3. Wnt-Fz1 increases osteoblastogenesis, but both Wnt-Fz1 and Wnt-Fz2 inhibit adipogenesis.
A, 3T3-L1 preadipocytes were infected with control retrovirus (pTS13), retroviruses expressing Wnt10b, Wnt-Fz1, or Wnt-Fz2, a dominant stable form of ␤-catenin (␤-cat), or its control retrovirus (pNeo). Two days after confluence, cells were incubated in mineralization medium for 2 weeks and stained for alkaline phosphatase activity. B, ST2 marrow-derived stromal cells were infected with control retrovirus (pTS13), retroviruses expressing Wnt10b, Wnt-Fz1, or Wnt-Fz2, a dominant stable form of ␤-catenin, or its control retrovirus (pNeo). Two days after confluence, cells were stained for alkaline phosphatase activity. C, control ST2 cells (pTS13 or pNeo) or those expressing Wnt10b, Wnt-Fz1, or Wnt-Fz2, or dominant stable ␤-catenin were incubated for 4 weeks in the presence of mineralization media. Mineralization was visualized by von Kossa staining. D, control ST2 cells or those expressing Wnt10b, Wnt-Fz1, Wnt-Fz2, or dominant stable ␤-catenin were grown to confluence. Two days after confluence, cells were treated with inducers of adipogenesis in the presence of troglitazone. Eight days later, cells were stained with Oil Red-O to visualize the accumulation of lipid and whole cell extracts were analyzed by immunoblot for PPAR␥.
One effector potentially activated by ␤-catenin-independent signaling is the serine/threonine phosphatase, calcineurin (36). To determine whether calcineurin activity is necessary for inhibition of adipogenesis, 3T3-L1 cells expressing dominant stable ␤-catenin, Wnt-Fz1, or Wnt-Fz2 were treated with the pharmacological inhibitors of calcineurin, FK506, or cyclosporin A during differentiation (Fig. 6A). Inhibition of calcineurin had no effect on lipid accumulation or PPAR␥ expression in cells expressing a dominant stable form of ␤-catenin and had only a modest effect on preadipocytes expressing Wnt-Fz1. However, both FK506 and cyclosporin A partially rescued differentiation in cells expressing Wnt-Fz2, as demonstrated by increased accumulation of lipid and expression of PPAR␥. These data suggested that activated Fz2, and to some extent activated Fz1, partially require the phosphatase activity of calcineurin to inhibit adipogenesis.
Since calcineurin activity appears to be partially required for ␤-catenin-independent inhibition of adipogenesis, we predict that Wnt3a, a Wnt reported to activate both ␤-catenin-dependent and ␤-catenin-independent signaling (37,38), may also require calcineurin activity for complete inhibition of preadipocyte differentiation. To test this hypothesis, we treated 3T3-L1 preadipocytes with Wnt3a for the first 4 days of adipocyte conversion in the presence or absence of calcineurin inhibitors. Consistent with our prior results using a brown preadipocyte model (39), Wnt3a inhibited adipogenesis of 3T3-L1 cells in a concentration-dependent manner and increased stability of ␤-catenin (Fig. 6, B and C). Importantly, repression of adipogenesis by Wnt3a was partially rescued by inhibition of calcineurin with FK506 or cyclosporin A, as assessed by lipid accumulation and expression of C/EBP␣ and PPAR␥ (Fig. 6B). Taken together, these data suggested that calcineurin in part mediates the ␤-catenin-independent inhibition of adipogenesis by Wnt signaling. FIG. 5. Wnt-Fz1 and Wnt-Fz2 inhibit adipogenesis through independent mechanisms. A, 3T3-L1 preadipocytes were infected with a control retrovirus (pBabe) or a retrovirus expressing Wnt-Fz1. After selection, cells were reinfected with a control retrovirus (pTS13) or a retrovirus expressing Wnt-Fz2. Two days after confluence, cells were treated with insulin (Ins), methylisobutylxanthine, dexamethasone, and insulin (MDI), or MDI and a PPAR␥ agonist, troglitazone (MDIT). Two weeks later, cells were stained with Oil Red-O to visualize the accumulation of lipid. B, cells from A were grown to confluence, and 2 days later, cytosolic fractions of cell lysates were isolated, and ␤-catenin (␤-cat) and ␤-tubulin (loading control) were analyzed by immunoblot. C, 3T3-L1 preadipocytes were infected with a control retrovirus (pNeo) or a retrovirus expressing a dominant stable form of ␤-catenin. After selection, cells were reinfected with a control retrovirus (pTS13) or a retrovirus expressing Wnt-Fz2. Two days after confluence, cells were treated with insulin, methylisobutylxanthine, dexamethasone, and insulin, or MDI and a PPAR␥ agonist, troglitazone. Two weeks later, cells were stained with Oil Red-O to visualize the accumulation of lipid. D, cells from C were grown to confluence, and 2 days later, cytosolic fractions of cell lysates were isolated, and ␤-catenin and ␤-tubulin (loading control) were analyzed by immunoblot.

DISCUSSION
In this study, we used constitutively active chimeras between Wnt8 and Fz1 or Fz2 to investigate functions of these Wnt receptors in mesenchymal precursors. Activated Fz1 stabilizes ␤-catenin and increases levels of Cyclin D1 (Fig. 1C), a known target of ␤-catenin signaling, whereas activated Fz2 does not. In addition, activated Fz1 mimics the effects of stabilized ␤-catenin by inhibiting apoptosis and adipogenesis in 3T3-L1 preadipocytes (Fig. 2) and by increasing osteoblastogenesis in ST2 cells (Fig. 3, B and C). Despite the apparent inability of activated Fz2 to initiate ␤-catenin-dependent signaling, activated Fz2 partially blocks adipogenesis in 3T3-L1 preadipocytes and pluripotent ST2 cells (Figs. 3D, 4, and 5). Repression of adipogenesis by activated Fz2 requires calcineurin activity as FK506 and cyclosporin A partially restore lipid accumulation and expression of PPAR␥ (Fig. 6A). Although inhibition of adipogenesis by recombinant Wnt3a is also partially rescued by calcineurin inhibitors (Fig. 6B), FK506 and cyclosporin A do not increase preadipocyte differentiation in cells expressing a dominant stable form of ␤-catenin. Taken together, these data suggested that Wnt signaling inhibits adipogenesis via both ␤-catenin-dependent and ␤-catenin-independent mechanisms.
A number of pathways are reported to be downstream of Fz2 signaling, including calcium signaling, induction of protein kinase C activity, and induction of cGMP phosphodiesterase activity (3). We used pharmacological inhibitors to address whether one or more of these pathways were required for inhibition of adipogenesis by activated Fz1 or Fz2. Of the signaling pathways tested (i.e. cGMP phosphodiesterase, protein kinase C, c-Jun amino-terminal kinase, calmodulin-dependent kinase II, RhoA, calcineurin), only inhibitors of the serine/threonine phosphatase, calcineurin, rescued adipogenesis in cells expressing activated Fz1 and Fz2 (Fig. 6A and data not shown). A role for calcineurin in regulating adipogenesis is supported by work from the Clipstone laboratory (40) in which expression of a constitutively active form of calcineurin inhibits adipogenesis. Well characterized substrates for calcineurin are nuclear factor of activated T cell (NFAT) family members (41), and expression of a constitutively active NFATc1 has been reported to transform 3T3-L1 preadipocytes and inhibit adipo-cyte differentiation (42). In our experiments in 3T3-L1 preadipocytes, we have no evidence that NFATc1 activity or subcellular localization is affected by Wnt-Fz1, Wnt-Fz2, or Wnt3a treatment (data not shown). Although calcineurin has established targets in other cell types, including Elk-1 (43) and MEF-2A (44), the substrate of calcineurin involved in inhibition of adipogenesis remains unknown.
Inhibition of calcineurin was unable to rescue differentiation in preadipocytes ectopically expressing Wnt10b, possibly due to overexpression (data not shown). Instead we chose to investigate the ability of calcineurin inhibitors to rescue the effects of recombinant Wnt3a on adipogenesis. Although this Wnt family member is not expressed in preadipocytes, Wnt3a is thought to signal through ␤-catenin-dependent pathways (20,45). Intriguingly, calcineurin inhibitors partially relieved the repression of adipogenesis caused by recombinant Wnt3a. These data suggested that Wnts normally associated with ␤-catenin-dependent signaling, such as Wnt3a, may inhibit adipogenesis via both ␤-catenin-dependent and ␤-catenin-independent pathways, potentially through interactions with both Fz1 and Fz2. These findings are supported by recent studies in which Wnt1 and Wnt3a are reported to activate the small GTPase, rhoA, to influence cell motility and neurite outgrowth (37,38,46).
The Wnt-Fz chimeras appear to require the carboxyl terminus because truncation of Wnt-Fz1 or Wnt-Fz2 inhibits their activity (Fig. 4A). This is consistent with previous studies demonstrating that the cytosolic tail of Fz is important for recruiting Dishevelled and mediating ␤-catenin-dependent signaling (21,47,48). Recruitment of Dishevelled also appears to be required for ␤-catenin-independent signaling, including activation of c-Jun amino-terminal kinase signaling (49 -51). These findings suggested that the conserved Dishevelled binding motif located in the cytosolic tail is essential for both ␤-catenindependent and ␤-catenin-independent Wnt signaling.
Despite sharing considerable sequence identity, Fz1 and Fz2 appear to activate distinct signaling pathways. Previous studies in Drosophila suggest that the specificity of signaling initiated by Fz may be intrinsic to the cytosolic and transmembrane portions and independent of the amino-terminal Wnt binding domain (52,53). In addition, chimeras between Wnts and hFz5 activate ␤-catenin signaling even when fused to Wnts normally FIG. 6. Calcineurin inhibitors partially rescue the inhibition of adipogenesis by activated Fz2 or Wnt3a, but do not rescue adipogenesis in cells expressing ␤-catenin. A, 3T3-L1 preadipocytes expressing a dominant stable form of ␤-catenin (␤-cat), Wnt-Fz1, Wnt-Fz2, or control cells (pTS13) were induced to differentiate in the presence of vehicle or the calcineurin inhibitors FK506 (10 ng/ml) or cyclosporin A (CsA; 1 g/ml) as indicated. Two weeks later, cells were stained with Oil Red-O to visualize lipid accumulation or lysed for whole cell extracts. Expression of PPAR␥ and laminin (loading control) was determined by immunoblot analysis. B, 3T3-L1 preadipocytes were induced to differentiate in the presence of the indicated concentrations of recombinant Wnt3a and either vehicle or the calcineurin inhibitors, FK506 or CsA. Two weeks later, cells were stained with Oil Red-O to visualize lipid accumulation, and expression of PPAR␥ and C/EBP␣ was analyzed by immunoblot. C, two days after confluence, 3T3-L1 preadipocytes were incubated with the indicated amounts of recombinant Wnt3a (rmWnt3a) for 2 h. ␤-Catenin and ␤-tubulin (loading control) in cytosolic fractions were analyzed by immunoblot. associated with ␤-catenin-independent signaling (22). These studies suggested that the specificity of Fz signaling may be intrinsic to the Fz rather than dependent on the binding of a particular Wnt.