Wnt-3A enhances bone morphogenetic protein-2-mediated chondrogenesis of murine C3H10T1/2 mesenchymal cells.

We have recently reported the chondrogenic effect of bone morphogenetic protein-2 (BMP-2) in high density cultures of the mouse multipotent mesenchymal C3H10T1/2 cell line and have shown the functional requirement of the cell-cell adhesion molecule N-cadherin in BMP-2-induced chondrogenesis in vitro (Denker, A. E., Nicoll, S. B., and Tuan, R. S. (1995) Differentiation 59, 25-34; Haas, A. R., and Tuan, R. S. (1999) Differentiation 64, 77-89). Furthermore, BMP-2 treatment also results in an increased protein level of beta-catenin, a known N-cadherin-associated Wnt signal transducer (Fischer, L., Haas, A., and Tuan, R. S. (2001) Signal Transduction 2, 66-78), suggesting functional cross-talk between the BMP-2 and Wnt signaling pathways. We have observed previously that BMP-2 treatment up-regulates expression of Wnt-3A in high density cultures of C3H10T1/2 cells. To assess the contribution of Wnt-3A to BMP-2-mediated chondrogenesis, we have generated C3H10T1/2 cell lines overexpressing Wnt-3A and various forms of glycogen synthase kinase-3beta (GSK-3beta), an immediate cytosolic component of the Wnt signaling pathway, and examined their response to BMP-2. We show that overexpression of either Wnt-3A or kinase-dead GSK-3beta enhances BMP-2-mediated chondrogenesis. Furthermore, Wnt-3A overexpression results in decreases in both N-cadherin and GSK-3beta protein levels, whereas Wnt-3A as well as kinase-dead GSK-3beta overexpression increase total and nuclear levels of both beta-catenin and LEF-1. Direct cross-talk between Wnts and BMP-2 was also indicated by the up-regulated interaction between beta-catenin and SMAD-4 in response to BMP-2. These results suggest that Wnt-3A acts in a manner opposite to that of other Wnts, such as Wnt-7A, which were previously identified as inhibitory to chondrogenesis, and is the first BMP-2-regulated, chondrogenesis-enhancing member of the Wnt family.

Cellular condensation is a requisite step of mesenchymal chondrogenesis, the first differentiation event of endochondral skeletal development, because perturbation of cell-cell contact abrogates chondrogenesis both in vitro and in vivo (1,2). We have shown previously that in the developing limb bud, N-cadherin, a calcium-dependent homotypic adhesion molecule, is expressed in a manner that correlates spatiotemporally with cellular condensation (3,4). Furthermore, N-cadherin is functionally required in developmental chondrogenesis (3,4) as well as the induction of chondrogenesis by bone morphogenetic protein-2 (BMP-2) 1 (5). Finally, we have shown that the chondrogenic modulation of N-cadherin is accompanied by protection of ␤-catenin protein levels (6).
The Wnt family consists of a number of small, cysteine-rich, secreted glycoproteins involved in embryonic development, tissue induction, and axial polarity (17,18). According to the current mechanistic model of canonical Wnt action, in cells lacking Wnt signal, GSK-3␤ phosphorylates ␤-catenin, inducing rapid degradation of ␤-catenin via the ubiquitin/proteasome pathway (19,20). In the presence of Wnt, Disheveled (Dsh), through a protein kinase C mechanism, inhibits GSK-3␤ kinase activity, and ␤-catenin is protected from proteasomemediated degradation. ␤-Catenin instead translocates to the nucleus, where it functions in a transcription complex in association with LEF-1/T-cell factors (for review, see Ref. 21). In recent studies, we have observed increased ␤-catenin and LEF-1 protein levels as well as their interaction during BMP-2-stimulated chondrogenesis of mesenchymal cells in vitro (22), suggesting a possible regulatory role for Wnts during chondrogenesis.
A number of Wnt genes are expressed in the developing limb (18,23), including Wnt-5A in distal mesenchyme, Wnt-7A in dorsal ectoderm, and Wnt-3A in mouse apical ectodermal ridge (24,25). Wnt-7A has been shown to inhibit chondrogenesis in vitro (25,26), and Wnt-4, localized to developing joint regions (28), has been implicated in chondrocyte differentiation (29). Wnt-3A has been shown to enhance bmp-2 expression (25), and wnt-3A-and lef1-deficient mice both display similar phenotypes, with the latter developing limb deformities (30). In addition, Wnt-3A is able to induce cytoskeletal reorganization (31), an event shown previously to be critical for the requisite cell shape change that occurs during mesenchymal chondrogenesis (32,33). It is noteworthy that in the developing limb, Wnt-7A and Wnt-5A have been found to act through a pathway other than ␤-catenin/LEF-1, whereas Wnt-3A and Wnt-4 appear to utilize ␤-catenin and LEF-1 specifically in their signal (25,29). Interestingly, recent evidence has shown that a combination of Frizzled receptor specificity and Dsh acting as a molecular switch can transduce a Wnt signal into both ␤-catenin pathways and Jun N-terminal kinase regulation separately or concomitantly (34 -36). Wnts have been classified into the Wnt-1 class, which functions through ␤-catenin/LEF, and the Wnt-5A class, which acts via the modulation of intracellular calcium levels (14). It has been postulated that the two classes of Wnts may act antagonistically when present together (37).
Our laboratory has established high density micromass cultures of the murine multipotent cell line C3H10T1/2 as an experimental model of mesenchymal chondrogenesis under the stimulation of the chondro-inductive factor, BMP-2 (5,38). We have observed that BMP-2-induced chondrogenesis in C3H10T1/2 cells is accompanied by down-regulation of Wnt-7A and up-regulation of Wnt-3A, whereas the level of other Wnts, including Wnt-3 and Wnt-5A, remains unchanged (22). This observation raises the intriguing possibility that BMP-2-mediated mesenchymal chondrogenesis results from the regulated, antagonistic relationship between the chondro-inhibitory Wnts, such as Wnt-7A (26,27), and chondro-enhancing Wnts. Such a scenario in the developing vertebrate limb would thus bear some analogy to the interactive relationship between the actions of decapentaplegic (a BMP homolog) and Wingless (a Wnt homolog), which synergistically regulate a number of developmental and patterning events in Drosophila (39,40,44). Consistent with this mechanistic scheme is the recent evidence suggesting direct cross-talk between transforming growth factor-␤s and Wnts through interaction of SMAD-4 and ␤-catenin (44), as well as SMAD-3 and LEF-1 (46).
The central hypothesis of this study is that Wnt-3A is a candidate as a BMP-2 up-regulated chondro-enhancing Wnt. To test this hypothesis, we have investigated whether BMP-2induced chondrogenesis in high density cultures of C3H10T1/2 cells can be altered by changes in Wnt-3A expression and signaling. Specifically, we tested the effects of stable overexpression of Wnt-3A, wild-type GSK-3␤, or kinase-dead GSK-3␤ on the level of chondrogenesis, N-cadherin expression, and the subcellular distribution of ␤-catenin and LEF-1. Our results suggest that Wnt-3A plays a positive role in mesenchymal chondrogenic differentiation, specifically by enhancing BMP-2induced chondrogenesis through regulation of events important for both cellular adhesion and nuclear transcriptional modification.

Assays for Chondrogenesis
Passage 10 C3H10T1/2 cells obtained from ATTC were plated 1 day before transfection in complete Dulbecco's modified Eagle's medium at a density of 1.5 ϫ 10 5 cells/30-mm dish. Transfection was carried out per manufacturer's specifications using Superfect Transfection reagent (Qiagen). The Wnt-3A, wild-type GSK-3␤ or ⌬GSK-3␤, or for control, empty pCMV expression constructs were added with the carrier plasmid, pSV2NEO, at a ratio of 4:1. Two days after transfection, cells were split into 100-mm dishes at a density of 10 5 cells/dish and fed with 10 ml of complete Dulbecco's modified Eagle's medium containing neomycin/ Geneticin (G418) at 50 g/ml. After ϳ14 days in selective medium, cloning cylinders dipped in sterile vacuum grease were placed around individual colonies, and 0.25 ml of 0.1% trypsin and EDTA was added to dissociate the colonies from the plates. Cells were plated in 75-cm 2 flasks for expansion and fed with Dulbecco's modified Eagle's medium containing G418 at 50 g/ml.

RNA Isolation and Reverse Transcription (RT)-PCR
The following RT-PCR primers (forward and backward, 5Ј 3 3Ј) were used: 1) mouse Wnt-3A, AACC ACG GGA GCA GGG TTC ATT C 3 and AAG GGG GTC TCC AAA AGT TCC ACC, to amplify a 534-bp region within the 3Ј-untranslated region of mouse Wnt-3A; and for the purpose of normalizing mRNA load 2) mouse glyceraldehyde-3-phosphate dehydrogenase, CCA CCC ATG GCA AAT TCC ATG GCA and TCT AGA CGG CAG GTC AGG TCC ACC, to amplify a 600-bp region of mouse glyceraldehyde-3-phosphate dehydrogenase.
Total cellular RNA was isolated from micromass cultures or day 10.5 mouse limb buds using Tri-Reagent RNA/DNA/Protein Isolation Reagent (Molecular Research Center, Inc.) and reverse transcribed using oligo(dT) primers with the Superscript Amplification System (Invitrogen) per the manufacturer's instructions. PCR was carried out in a volume of 50 l using 25 cycles of 1 min at 94°C, 2 min at 60.5°C, and 3 min at 72°C. Products of the RT-PCR were analyzed by electrophoresis on a 1.5% Tris-acetate EDTA-agarose gel stained with ethidium bromide, and band intensities were quantified using a Kodak Digital Science Image Station (model 440 CF).
For Western analysis, 20 g of each protein sample was separated by SDS-PAGE (6 -10%) and electrotransferred onto nitrocellulose membranes (0.2 m, Schleicher & Schuell). The blots were blocked with buffer containing 0.05% Tween 20 and 3% bovine serum albumin and reacted sequentially with primary and secondary antibodies (see below). Primary antibodies were 13A9 at 1:200, ␤-catenin and SMAD-4 at 1:500, LEF-1 at 1:3,000, GSK-3␤ at 1:3,800, for normalization of protein loading, and ␤-actin at 1:10,000. Alkaline phosphatase-conjugated secondary antibodies (Sigma) were at a 1:3,800 dilution. Blots were developed with 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (Zymed Laboratories Inc.) and densitometrically scanned as described above. Negative controls for all immunoblots consisted of a lane of total cell extracts immunoprobed with secondary antibodies in the absence of primary antibody. Either Ponceau S staining of immunoblots or parallel Coomassie Brilliant Blue-stained gels confirmed equal protein loading.
Immunoprecipitation-Protein extracts were incubated with the indicated antibodies in 1 ml of immunoprecipitation buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1.5 mM CaCl 2 , 1% Triton X-100, 0.5% Nonidet P-40, containing protease inhibitors) for 1 h at 4°C with agitation. Immunoprecipitation was done by adding 30 l of protein G-agarose beads (50% suspension; Invitrogen) followed by a 45-min incubation at 4°C. After washing, the resin was resuspended into SDS-polyacrylamide electrophoresis buffer, boiled, centrifuged, and the supernatant fractionated by SDS-PAGE. Electroblotting and Western analysis were performed as described above. Positive controls for all immunoprecipitation experiments consisted of a lane of total cell extract immunoprobed for the protein(s) of interest.
Monoclonal Antibodies-The N-cadherin monoclonal antibody, 13A9, which recognizes the intracellular C-terminal domain of human Ncadherin, was a generous gift from Dr. Karen Knudsen (Lankenau Medical Research Center, Wynnewood, PA; 50). Monoclonal antibody to ␤-catenin and GSK-3␤ were purchased from Transduction Laboratories. Rabbit antibodies to mouse LEF-1 were a generous gift from Dr. R. Grosschedl, University of Munich, Germany (16). Monoclonal antibodies to SMAD-4 and anti-Myc clone 9E10 were purchased from Santa Cruz Biochemicals. Monoclonal antibody to mouse ␤-actin was purchased from Sigma.

In Vitro GSK-3␤ Kinase Assay
Micromass cultures were harvested in 20 mM Tris-HCl, pH 7.5, 270 mM sucrose, 1 mM EDTA, 1 mM EGTA, 0.5% Triton X-100 with protease inhibitors as well as 0.5 mM NaVO 3 and 1 nM okadaic acid and extracted at 4°C for 20 min. Extracts were cleared of membranes by centrifugation at top speed in a microfuge, and then protein concentration was determined. 15-g aliquots of the total lysates were subjected to Western blot analysis and Ponceau S staining to verify loading. Another 100-g aliquot of the total lysates was immunoprecipitated with antibody to GSK-3␤. The immune complex was then washed in 50 mM Tris-HCl, pH 7.5, with 1 mM dithiothreitol and resuspended in 45 l of kinase reaction buffer containing 200 M ATP, 4.5 l 10 ϫ GSK-3␤ kinase buffer (NEB), 100 M [␥-32 P]ATP (0.25 Ci/ml) (PerkinElmer Life Sciences) with or without (as negative control) 60 M phospho-CREB peptide, or 0.5 unit rabbit GSK-3␤ (as positive control). Kinase reactions were carried out for 15 min at 30°C and stopped with 1.5% phosphoric acid. 32 P-Labeled reactions were spotted on P81 Whatman paper and washed 3 time for 5 min each with 0.5% phosphoric acid. Filters were air dried, placed right side up in scintillation vials with 400 l of 0.2 N NaOH overnight at room temperature, and radioactivity of the CREB peptide determined by liquid scintillation counting.

Overexpression of Wnt Signaling Components on BMP-2-
induced Chondrogenesis-Stable, Wnt-3A-transfected clones were analyzed for Wnt-3A overexpression. Both Western immunoblot (Fig. 1A) and RT-PCR (Fig. 1B) analyses revealed the presence of Wnt-3A at protein and mRNA levels, respectively. Myc-tagged Wnt-3A protein was detected in monolayer cultures of several clones examined (Fig. 1A), and increased mRNA was also detected in overexpressing Wnt-3A cell lines (clones 2 and 6 shown; Fig. 1B). We next examined overexpression of GSK-3␤ in C3H10T1/2 micromass at the protein level using the Myc tag as an identifier of exogenously introduced GSK-3␤. A significant level of expression of both exogenously introduced wild-type and kinase-dead GSK-3␤ was detected, with Myc-reactive proteins migrating at the expected 45 kDa, similar to the endogenous product, detected using anti-GSK-3␤ antibody, in untransfected parental cells (Fig. 1C).
Effect of Overexpression of Wnt Signaling Components on BMP-2-induced Chondrogenesis-We next examined the effect of stable overexpression of Wnt-3A and GSK-3␤ on BMP-2 induction of chondrogenesis in micromass cultures of C3H10T1/2 cells. On the basis of Alcian blue staining ( Fig. 2A) and metabolic sulfate incorporation into sulfated proteoglycans (Fig. 2B), Wnt-3A-overexpressing cultures were found to be significantly more responsive to BMP-2 in terms of chondrogenesis at all time points examined (days 5, 9, and 13). It is interesting to note that Wnt-3A overexpression alone was insufficient to induce chondrogenesis in the micromass cultures, i.e. BMP-2 treatment was still required for the induction of chondrogenesis. In the case of GSK-3␤, it appeared that increased levels of wild-type GSK-3␤ negatively affected BMP-2induced chondrogenesis, whereas expression of ⌬GSK-3␤ contributed positively to the effect of BMP-2 on C3H10T1/2 micromass cultures (Fig. 3, A and B), during the early culture period (before day 9). As controls, C3H10T1/2 cells were transfected with empty pCMV vector and showed no difference from nontransfected parental micromass cultures with respect to BMP-2 treatment.
In view of our recent observation that BMP-2 treatment of C3H10T1/2 micromass cultures up-regulates LEF-1 (21), we examined whether the effect of Wnt-3A overexpression functioned through regulation of LEF-1. As shown in Fig. 6, BMP-2 treatment of clonally derived Wnt-3A-overexpressing micromass cultures increased the nuclear levels of LEF-1 protein over that of nontransfected parental cultures on days 5-13 (Fig. 6A). By means of coimmunoprecipitation analysis, the level of BMP-2-induced association of ␤-catenin with LEF-1 was shown to be decreased slightly in Wnt-3A-overexpressing cultures during days 9 -13 compared with parental controls (Fig. 6B); however, the effects of Wnt-3A were not remarkable.
Overexpression of Wild-type and ⌬GSK-3␤ Regulates N-Cadherin and ␤-Catenin Levels-Because both ␤-catenin and N-cadherin were responsive to Wnt-3A overexpression, suggesting Wnt regulation, we analyzed changes in the protein levels of these molecules in response to wild-type and ⌬GSK-3␤ overexpression. As shown in Fig. 7, BMP-2 increases N-cadherin and ␤-catenin protein levels on days 5-13 in parental nontransfected cultures. In comparison, overexpression of

Wnt-3A Signaling in BMP-2-mediated Chondrogenesis
⌬GSK-3␤ in the presence of BMP-2 dramatically increased the levels of both N-cadherin and ␤-catenin early (up to day 5) with little effect thereafter (Fig. 7). It is noteworthy that overexpression of ⌬GSK-3␤ also resulted in increased protein levels of N-cadherin and ␤-catenin in the absence of BMP-2, whereas parental nontransfected cultures required BMP-2 treatment for such increases. Furthermore, overexpression of wild-type GSK-3␤, although not remarkably affecting levels of N-cadherin or ␤-catenin during early (days 1-5) culture, decreased the levels of both molecules during later time points of chondrogenesis (days 9 -13).

Wnt-3A Signaling in BMP-2-mediated Chondrogenesis
response to BMP-2. Nuclear extracts showed an elevated SMAD-4 signal after BMP-2 treatment throughout the entire culture period, whereas control cultures showed only a slight temporal increase in SMAD-4 (Fig. 9A). To analyze the interaction between ␤-catenin and SMAD-4, ␤-catenin immunoprecipitates from C3H10T1/2 micromass cultures were probed for SMAD-4. SMAD-4 interaction with ␤-catenin was indeed detected in these cultures and was enhanced strongly by BMP-2 treatment from days 5 to 13 (Fig. 9B). For comparison, immunoprecipitates of ␤-catenin from micromass cultures were immunoprobed for LEF-1 (Fig. 9C). Interestingly, increased association between ␤-catenin and LEF-1 did not appear in response to BMP-2 until day 9 of culture, suggesting that ␤-catenin interaction with SMAD-4 was an early response that preceded the association of LEF-1 with the complex.

DISCUSSION
Our previous studies (5, 6) have suggested a role for ␤-catenin, LEF-1, and GSK-3␤ in BMP-2-induced chondrogenesis in high density micromass cultures of the mesenchymal cell line C3H10T1/2. Interestingly, we have also observed recently upregulation of Wnt-3A in response to BMP-2 (22). In this study we have determined that overexpression of Wnt-3A and kinasedead GSK-3␤ both significantly enhance, whereas that of wildtype GSK-3␤ inhibits, BMP-2-mediated chondrogenesis of micromass cultures. Interestingly, these molecules could not regulate chondrogenesis in the absence of BMP-2, suggesting the necessity of cross-talk between BMPs and Wnts. It is also noteworthy that although Wnt-3A enhances BMP-2-induced chondrogenesis as early as day 5 and continuing to day 13, overexpression of either form of GSK-3␤ did not exert a significant effect post day 9, i.e. during late stages of mesenchymal differentiation into chondrocytes. Because Wnt-3A affects chondrogenesis throughout the culture period, these results imply that both early and late availability of Wnt signal can enhance BMP-2-mediated chondrogenesis, whereas introduction of kinase-dead GSK-3␤ may mimic only the aspect of Wnt signal related to the acceleration of chondrocyte differentiation. Thus, it is possible that Wnt-3A induces changes in C3H10T1/2 micromass cultures via pathways in addition to ␤-catenin protection.
In view of the fact that some Wnts do not act via the inhibition of the serine/threonine kinase activity of GSK-3␤ (14), we have analyzed GSK-3␤ enzymatic activity in BMP-2-stimulated micromass stably transfected with Wnt-3A. Indeed, Wnt-3A overexpression inhibits GSK-3␤ enzyme activity during the entire culture period, with inhibition most dramatic at earlier time points. This finding also suggests that GSK-3␤ inhibition is more important during early chondrogenesis, supporting our results that introduction of both wild-type and ⌬GSK-3␤ affects younger rather than older cultures during BMP-2-mediated chondrogenesis. Wnt-3A overexpression in C3H10T1/2 micromass cultures also regulates the level of Ncadherin protein, supporting the postulate that Wnts act, at least in part, via regulating cadherin levels (27,37,54). Previously we have observed that overexpression of both wild-type and mutant N-cadherin disrupts BMP-2-induced chondrogenesis (5), indicating that tight regulation of Ncadherin is required for the correct series of event to move mesenchymal cells through condensation to differentiation. Our data suggest that one function of Wnt-3A may be optimization of N-cadherin levels during BMP-2 stimulation of chondrogenesis.
The BMP-2-mediated increase of both total and nuclear levels of ␤-catenin, as described previously (6), is enhanced further by Wnt-3A overexpression. Increases in ␤-catenin levels are indicative of Wnt canonical inhibition of GSK-3␤ and are presumed to result in increased ␤-catenin nuclearization and interaction with LEF-1 in a transcription complex (55)(56)(57). It is likely that Wnt-3A protection of ␤-catenin in BMP-2-stimu- Wnt-3A Signaling in BMP-2-mediated Chondrogenesis lated C3H10T1/2 micromass cultures results in translocation of ␤-catenin to the nucleus and subsequent regulation of chondrogenesis-enhancing gene expression, such as those involved in adhesion and cytoskeletal rearrangement (31). There is also evidence that cadherins can sequester catenins away from the Wnt signal (58,59). Therefore, down-regulation of Ncadherin by Wnt-3A may serve a dual function in our system, i.e. regulation of adhesion and increased ␤-catenin signal transduction. In further support of this hypothesis, we have consistently observed a Wnt-3A-induced decrease in ␤-catenin association with N-cadherin (data not shown), another possible mechanism to modulate adhesion and strongly suggestive of ␤-catenin being utilized in a nonmembraneassociated function.
We have observed recently that the consequences of BMP-2 treatment of C3H10T1/2 micromass cultures are remarkably similar to those known for Wnt signaling, i.e. both total and nuclear LEF-1 and ␤-catenin levels are increased in response to BMP-2, most likely as a result of BMP-2 up-regulation of Wnt-3A (22). In this report, we have shown that there is a remarkable decrease in levels of GSK-3␤ protein in response to Wnt signal, as has been reported in other systems (60), as well as an increase in total and nuclear LEF-1 protein in response to Wnt-3A, again suggesting inhibition of GSK-3␤. Of remarkable interest is our observation that Wnt-3A-induced increases in ␤-catenin and LEF-1 nuclearization are not accompanied by an increased association between LEF-1 and ␤-catenin. This suggests that ␤-catenin in BMP-2-stimulated C3H10T1/2 micromass, in response to Wnt-3A, may interact with another highmobility group transcription partner, such as T-cell factor, or may increase interaction with factors not directly involved in Wnt signal (see below).
Because GSK-3␤ is a ubiquitous and multifunctional kinase (61)(62)(63)(64), its contribution to chondrogenesis could be through various pathways. Our results of GSK-3␤ overexpression, along with the Wnt-3A data, suggest that early nuclearization of ␤-catenin and LEF-1 can accelerate BMP-2 induction of chondrogenesis. However, why would both ⌬GSK-3␤ and Wnt-3A enhance chondrogenesis while exerting opposite effects on Ncadherin levels during early condensation? Because ⌬GSK-3␤ overexpression increases N-cadherin levels in cultures with or FIG. 8. Effect of GSK-3␤ overexpression on nuclear levels of ␤-catenin and LEF-1 in C3H10T1/2 micromass cultures. Immunoblot analysis of nuclear extracts probed for ␤-catenin (A) or LEF-1 (B). All densitometric intensities of immunoreactive protein bands are expressed as a percentage of day 1 control signal. A, parental nontransfected cultures display increased nuclearization of the largest form of ␤-catenin in response to BMP-2 on culture days 1-9. Overexpression of wild-type GSK-3␤ does not markedly affect nuclear ␤-catenin levels. However, overexpression of ⌬GSK-3␤ increased nuclear ␤-catenin in response to BMP-2 throughout the culture period. For comparison, a lane of whole cell extracts (Whole Cell) probed for ␤-catenin shows the three isoforms of ␤-catenin found in total cell lysates of C3H10T1/2 cultures. B, parental nontransfected cultures display increased nuclearization of LEF-1 in response to BMP-2 in older cultures, whereas overexpression of wild-type GSK-3␤ does not affect nuclear LEF-1 levels remarkably. However, overexpression of ⌬GSK-3␤ results in an increase in nuclear LEF-1 throughout the culture period compared with parental cultures. There was no remarkable difference on day 9 between GSK-3␤-overexpressing cell lines and parental cultures in terms of either ␤-catenin or LEF-1 nuclear levels. As a positive control, a lane of mouse thymus extract (Mouse Thymus) probed for LEF-1 accompanies the nuclear immunoblots for LEF-1. Equal protein loading was confirmed by Ponceau S staining of the immunoblot.
FIG. 9. BMP-2 modifies SMAD-4 protein levels and its association with ␤-catenin in C3H10T1/2 micromass cultures. A, immunoblot analysis of SMAD-4 in nuclear extracts. BMP-2 treatment increases SMAD-4 nuclearization over that of untreated, control cultures on all culture days. A parallel gel was stained with Coomassie Brilliant Blue to verify equal loading. Densitometric intensities of immunoreactive protein bands are expressed as a percentage of day 1 untreated control signal. B, immunoblot analysis of ␤-catenin immunoprecipitates probed for SMAD-4. SMAD-4 interacts with ␤-catenin in both treated and untreated cultures, and this interaction is enhanced remarkably upon BMP-2 addition on culture days 5-13. Densitometric intensities of immunoreactive bands are expressed as a percentage of day 1 untreated cultures. A SMAD-4 immunoreactive band derived from total cell extract is shown as a control. C, immunoblot analysis of ␤-catenin immunoprecipitates probed for LEF-1. LEF-1 interacts with ␤-catenin in both treated and untreated cultures, and this interaction is enhanced upon the addition of BMP-2 on culture days 9 -13. The positive control consists of a lane containing mouse thymus cell extracts immunoprobed for LEF-1 (Mouse Thymus). Densitometric intensities of immunoreactive protein bands are expressed as a percentage of day 1 untreated control signal. Equal protein loading was confirmed by Coomassie Blue staining of parallel gels. without BMP-2 treatment but does not enhance chondrogenesis in untreated cultures, we suggest that it is not the induced N-cadherin up-regulation that is chondrogenesis-enhancing, but rather nuclearization of LEF-1 and ␤-catenin. Thus, on day 13, when ⌬GSK-3␤ overexpression no longer dramatically affects nuclear ␤-catenin or LEF-1, compared with parental nontransfected micromass, its chondrogenesis-enhancing activity is also abrogated. We suggest that Wnt-3A indeed inhibits GSK-3␤ activity, as does expression of ⌬GSK-3␤, by virtue of competition and results in nuclearization of LEF-1 and ␤-catenin, but that in addition, Wnt-3A acts through a separate pathway to control N-cadherin-mediated adhesion, possibly via Dsh activation of Jun N-terminal kinase (35). Because GSK-3␤ acts downstream of Dsh in Wnt signal, overexpression of ⌬GSK-3␤ only partially mimics Wnt-3A introduction.
Of particular interest here is the recent discovery that transforming growth factor-␤ signaling results in SMAD-4 and SMAD-3 directly interacting with ␤-catenin/LEF-1 and the transcriptional activation of LEF-1-responsive promoters (45,46). We observe here that, although total levels of SMAD-4 do not increase in response to BMP-2 (data not shown), both nuclear levels of SMAD-4 as well as the interaction between SMAD-4 and ␤-catenin are enhanced upon BMP-2 treatment. The temporal profiles show that by day 5, there is a remarkable increase between SMAD-4 and ␤-catenin in response to BMP-2, likely because of the increasing levels of nuclear ␤-catenin. Then as LEF-1 begins to accumulate in the nucleus in response to BMP-2 during mid to late chondrogenesis, increased interaction between ␤-catenin and LEF-1, presumably in complex with SMAD-4, is observed. Although previous reports have shown that SMADs can interact with LEF and/or ␤-catenin in response to transforming growth factor-␤ (45,46), this is the first evidence that BMP-2 induction of chondrogenesis facilitates SMAD/catenin interaction, reminiscent of the well documented coordination of Wingless and decapentaplegic in Drosophila (40 -45, 65).
In conclusion, our results strongly suggest that Wnt-3A has the capacity to enhance BMP-2-mediated chondrogenesis of mesenchymal micromass cultures through the regulation of N-cadherin-mediated adhesion, the inhibition of GSK-3␤ kinase activity, and the nuclear signaling of ␤-catenin and LEF-1. Given the fact that BMP-2 up-regulates Wnt-3A during the early period of culture, and expression of exogenous Wnt-3A regulates N-cadherin during this same time period, Wnt-3A is likely to act primarily during mesenchymal condensation rather than during chondrocyte maturation. Furthermore, introduction of an inactive GSK-3␤ can mimic some but not all of the activities of Wnt-3A, suggesting that Wnt-3A may function in a second yet-to-be-defined pathway, not involving GSK-3␤, to modify cadherin-mediated adhesion. To our knowledge, Wnt-3A is the first described Wnt member that positively affects chondrogenesis in coordination with BMP-2 during early mesenchymal induction and may do so via SMAD/catenin-mediated nuclear signal. Our continuing studies focus on transcriptional regulation of cartilage-specific genes by ␤-catenin and LEF-1, coordination of SMAD/␤-catenin, and the functional contribution of GSK-3␤ to BMP-2-mediated chondrogenesis.