BMP-2-Modulated Chondrogenic Differentiation In Vitro Involves Down-Regulation of Membrane-Bound Beta-Catenin

Abstract

Bone morphogenetic proteins (BMPs) are important regulators of cellular differentiation and embryonic development. Beta catenin mediated nuclear signaling has been implicated in BMP-2-modulated chondrogenic differentiation in the pluripotential stem cell line C3H10T1/2. However, there is little information on the functional role of beta catenin in BMP-2-modulated differentiation of primary nontransformed mesenchymal cells. Here, we present evidence to show that BMP-2-induced chondrogenic differentiation in high-density primary mesenchymal culture is associated with a significant decrease in membrane-bound beta catenin by 72 hours compared to controls. Nuclear localization of beta catenin is not detectable by immunofluorescence and the TCF-responsive reporter vector TOPFLASH shows only background activity during chondrogenic differentiation. BMP-2-treated cultures show reduced cell-cell adhesion by 72 hours, which correlates with the changes in levels of membrane-bound beta catenin. Upregulation of membrane-bound beta catenin blocks the effect of BMP-2 on both chondrogenic differentiation and cell-cell adhesiveness. These findings suggest that BMP-2 can modulate the adhesivity of adherens junctions through regulation of membrane bound beta catenin.

Keywords:

• Cell adhesion
• BMP
• beta catenin
• micromass culture
• adherens junctions

Abbreviations
BCA	=	bicinchoninic acid
BMP	=	bone morphogenetic protein
BSA	=	bovine serum albumin
COX-2	=	cyclo-oxygenase-2
DAPI,	=	4′,6-diamidino-2-phenylindole dihydrochloride
ECL	=	electrochemiluminescent
EDTA	=	sodium ethylenediamine tetraacetate
FCS	=	fetal calf serum
GSK3beta	=	glycogen synthase kinase 3beta
HBSS	=	Hanks buffered saline solution
HCL	=	hydrochloric acid
HMG	=	high mobility group
HRP	=	horseradish peroxidase
LEF	=	lymphoid enhancer binding factor
PBS	=	phosphate buffered saline
PCR	=	polymerase chain reaction
R.T.	=	room temperature
SDS	=	sodium dodecyl sulphate
TCF	=	T-cell transcription factor
TGF-beta	=	transforming growth factor beta

INTRODUCTION

Cartilage formation in vivo from undifferentiated mesenchymal cells involves coordination of a series of complex processes including cell proliferation, cell migration, cell-cell and cell-substrate adhesion, and cell differentiation (1). Isolated mesenchymal cells undergo an initial process of cellular aggregation and proliferation to form precartilaginous condensations that then differentiate into cartilage nodules. The initial condensation process requires high levels of cellular adhesion, and is mediated through cell-cell adhesion molecules such as N-CAM and N-cadherin (2, 3, 4) as well as interactions with extracellular matrix molecules such as fibronectin and tenascin (1). The resulting high cellular density in the condensations is a requirement for chondrogenesis to occur, with the level of chondrogenic differentiation being directly correlated to the extent of initial condensation (1, 5). Following condensation, cell-cell interactions and communication through gap junctions triggers one or more signal transduction pathways that lead to chondrogenic differentiation (6).

Members of the bone morphogenetic protein family (BMPs) are important modulators of chondrogenic differentiation (7, 8, 9) and can initiate ectopic cartilage and bone formation (7). One of the most well studied members, BMP-2, is expressed at sites of bone repair and in developing embryonic skeleton (8, 10, 11) and enhances chondrogenic and osteogenic differentiation in cell lines (12, 13, 14, 15) as well as primary mesenchymal cells (16). Knockout of BMP-2 in transgenic mice is embryonic lethal (17), but other BMP null mutants show skeletal phenotypes, ranging from mild patterning defects (18) to more generalized deficiency in skeletal growth (19). The BMP signaling pathways are complex, but in general BMPs signal through transmembrane receptors that have intrinsic serine/threonine kinase activity, leading to activation of the Smad proteins and altered transcription (17).

Beta catenin is a multifunctional intracellular protein that can modulate apoptosis, migration, proliferation, and epithelial-mesenchymal transition (20). Beta catenin plays a direct role in promoting cell-cell adhesion through participation in adherens junctions, acting to link the cadherins to the actin cytoskeleton (21). As well, beta catenin plays a role in gene transcription through interaction with members of the TCF/LEF HMG domain family of transcription factors (22). Physiologically, beta catenin is required for normal cellular adhesion, linking the cadherins to the cytoskeleton (23, 24). Pathologically, nuclear translocation of beta catenin has been linked to tumorigenic transformation in many cell types including colon cancer, aggressive fibromatosis (25), pancreatic carcinoma, and other gastrointestinal malignancies (for review see Giles RH et al., 2003) (26). The downstream targets of beta catenin signaling have only been partially characterized but include the proto-oncogene c-myc (27), cyclin D1 (28), fibronectin (29) and COX-2 (30).

Previous immunohistochemical studies in normal human articular cartilage and perichondrium (31) have shown beta catenin expression in the membrane but have not detected any nuclear localization of beta catenin. However, recent overexpression studies using adhesion deficient truncated beta catenin have implicated nuclear beta catenin signaling as a mechanism for BMP accelerated chondrogenic differentiation in C3H10T1/2 cell line (32). Despite this data suggesting a role for beta catenin-mediated signaling in chondrogenic differentiation, little is known about the effect of BMP-induced chondrogenic differentiation on beta catenin in nontransformed primary mesenchymal cells. Here we characterize a new action of BMP-2, providing evidence that BMP-2 modulates the levels of membrane-bound beta catenin and subsequent calcium-dependent cell-cell adhesion during chondrogenic differentiation in primary mesenchymal cells but does not activate beta catenin signaling.

MATERIALS AND METHODS

Materials

Fertilized chicken embryos were staged according to Hamburger and Hamilton (33). Recombinant BMP-2 was obtained from Genetics Institute, San Diego, CA. Primary antibodies were rabbit anti beta catenin, mouse anti beta catenin, and rabbit anti Pan cadherin (Sigma, St Louis, MO).

Isolation and Culture of Cells

Micromass cultures were prepared based on Ahrens (5) and Jiang (34). The distal one third of stage 23/24 limb buds were dissected and the apical etodermal ridge removed. Pooled limb buds were digested with 0.1% trypsin and collagenase (Worthington Corp., Freehold, NJ) at 37°C for 10 minutes, and then the tissues gently triturated. After centrifugation, the cells were resuspended at a concentration of 2 × 10⁷ cells/ml, plated as 10 microliter drops on 35 mm culture dishes (Corning) and cultured in serum-free defined medium (35).

Immunofluorescence

Micromass cultures were fixed with 100% methanol for three minutes and blocked with 0.1 M poly lysine. Specimens were incubated with primary antibodies overnight, followed by secondary fluorescent goat anti-rabbit antibodies (Alexa 488) (Molecular Probes, USA) or fluorescent Cy3conjugated-anti mouse antibodies (Chemicon International INC., Temecula, CA). Confocal laser microscopy was used to visualize the fluorescent signal.

Quantitation of Chondrogenic Differentiation

Micromass cultures were fixed with 2.5% paraformaldehyde in PBS and stained with 1% Alcian blue 8GX in 0.1 N HCl, pH 1 (36, 37) for three hours, which stains cartilage specific sulfated proteoglycans (38). Cultures were then destained with 70% ethanol. For assessment of total sulphated proteoglycans, the bound Alcian blue dye was extracted with 0.5 ml of 4 M guanidine HCL (pH 5.8) and quantified by measuring absorbance at optical density (OD) 600 nm in a Perkin Elmer LambdaBio UV/VIS spectrophotometer. The OD was then normalized to total cell number measured in parallel cultures.

Western Blotting

Micromass cell culture samples from different time points were solubilized in 60 mM Tris (pH 6.8) and 2% SDS and protein concentration determined by micro BCA kit (Pierce Chemical, USA). Samples were adjusted to equal protein concentrations in a reducing sample buffer and analyzed by electrophoresis in a 10% polyacrylamide SDS gel. The proteins were transblotted to Hybond^TM C membrane (Amersham International Plc, UK) for two hours using Semi-Dry transfer units (Pharmacia Biotech, USA). Nonspecific binding sites were blocked with PBS containing 1% BSA and 0.2% Tween-20. The membrane was then incubated with 1:2000 rabbit anti beta-catenin antibody (Sigma, USA) in PBS with 0.2% Tween at R.T for one hour. After washing overnight, the membrane was incubated with HRP-conjugated goat anti rabbit (Amersham International Plc, UK) at 1:1000. Antibody reactivity was detected by enhanced chemiluminescence (ECL^TM detection reagents, Amersham International Plc, UK). Equal protein loading was confirmed by reprobing the membrane for beta actin.

Cell Adhesion Assay

Control and BMP-2 treated micromass cultures were plated at high density (2× 10⁷ cells/ml) and grown for defined time points. Cells were rinsed with warm PBS, and incubated in 0.01% Trypsin with 1 mM CaCl₂ at 37°C for 15 minutes, at which time all cells were floating in the medium. The reaction was then stopped with FCS, the cells resusupended in PBS and triturated 15× with a 1 ml Pipetman. The cells were then photographed and the number of particles in cell suspension counted with a hemocytometer. Cells were then centrifuged again and incubated in 0.1% trypsin and collagenase at 37°C in calcium free conditions for 10 minutes. After incubation, cells were sheared by passing them through a glass pipette to create a single cell solution. The data was then normalized according to the formula N_o/N_t, where N_o is the total number of cell aggregates after first tituration and N_t is the total number of single cells in the sample.

Reporter Gene Assays

Isolated chick limb cells were suspended in defined medium at a density of 1–2 × 10⁷ cells/ml. A 400 μ l aliquot of the cell suspension was placed into a 400 mm electroporation cuvette (Biorad). 5 μ g of either pTOPFLASH-Luc or pFOPFLASH-Luc (Upstate Biotechnology), and 5 μ g of the beta-galactosidase expression vector (pRCAS-lacZII) were added to the cuvette. The cells were given a single electrical pulse using the Biorad Gene Pulser II electroporation system set at 380 volts (V), 250 micro Farad (μ F), and ∞ ohms (Ω). Cells were then plated at low density overnight and replated at high-density (2× 10⁷cells/ml) the next day. Cell extracts were assayed for luciferase activity using Steady-GLO® Assay System and for beta galactosidase activity using Beta-galactosidase Enzyme Assay System (both from Promega Corporation, USA). Experiments were performed in triplicate.

RNA Extraction and Real Time PCR

Cell monolayers were detached using a rubber policemen. An RNeasy mini kit (Qiagen, Valencia, CA) was used to extract total RNA and reverse transcription carried out using Superscript II reverse transcriptase. Real-time quantitative RT-PCR analyses were performed using an ABI Prism 7700 Sequencing Detection system (Perkin-Elmer Applied Biosystems, Calif. USA). Primers and probes for beta catenin, smad6 and type II collagen were designed using Primer Express software (sequences supplied on request) and amplification performed using TaqMan RT-PCR assay (Perkin-Elmer Applied Biosystems, Calif. USA). Amplification of 28S ribosomal RNA was used as an internal control and all experiments were performed in triplicate. Water only negative controls were run for each experiment.

RESULTS

Beta Catenin and N-Cadherin Protein Levels are Differentially Modulated During BMP-2-Induced Chondrogenic Differentiation in Primary Mesenchymal Cells

To investigate the role of beta catenin in BMP-2-induced chondrogenic differentiation in primary cells, we used high density “micromass” culture of distal limb bud cells derived from stage 23/24 chick embryos (5, 33). This is a well-characterized culture system in which undifferentiated mesenchymal cells mature into defined cartilage nodules separated by fibroblastic cells over a period of four days in serum-free medium (39, 40). The changes in extracellular matrix composition seen in this culture system parallel those seen in vivo, making this a good model of cartilage formation (41). Under the experimental conditions used here, BMP-2 induces widespread cartilage formation by 72 hours after initial treatment (16).

To ask what effect BMP-2 has on beta catenin and N-cadherin mRNA and protein expression, we performed a time-course experiment exposing cultures either to BMP-2 200 ng/ml, or vehicle only (control cultures). This dose of BMP-2led to increased levels of smad6 mRNA (a BMP-early response gene) within two hours of exposure to BMP-2 and widespread chondrogenic differentiation by 72 hours as shown by increased Alcian blue staining and significantly increased levels of type II collagen mRNA (Figure 1A-D). Cultures were then harvested for either real-time PCR or for Western blotting at the time points shown.

Figure 1 BMP-2 treatment induces smad6 mRNA (a BMP early response gene) by two hours and induces chondrogenic differentiation by 72 hours, shown by increased type II collagen mRNA expression and increased cartilaginous extracellular matrix. Cells were plated as micromass cultures and cultured in serum free conditions with or without BMP-2. Cultures were either fixed with paraformaldehyde at 96 hours and stained with 1% Alcian blue 8GX in 0.1 N HCl, pH 1 for three hours to demonstrate cartilaginous extracellular matrix (A: Control; B: BMP-2 treated), or RNA was extracted at defined time points for measurement of relative levels of smad6 mRNA (C) or type II collagen mRNA (D). (See Color Plate III at the end of this issue).

COLOR PLATE III See W. V. Zhang and N. S. Stott, Figure 1.

Beta catenin mRNA levels showed little variation over the time period of the culture (Figure 2A). However, the levels of beta catenin mRNA were mildly reduced in the BMP-2-treated cultures compared to the control cultures. N-cadherin mRNA levels peaked at 18 hours after high density plating in both the control and the BMP-2-treated cultures and then fell progressively at 48 and 72 hours (Figure 2B). BMP-2 treatment led to a further reduction in N-cadherin mRNA levels compared to control 72 hours after high-density plating (p < 0.009).

Figure 2 Cells undergoing BMP-2-induced chondrogenic differentiation show little change in beta catenin mRNA levels but reduced N-cadherin mRNA levels at 72 hours. Cells were plated as micromass cultures and cultured in serum free conditions, either with BMP-2, 200 ng/ml or vehicle only (PBS). (A) Real-time PCR shows a small reduction in beta catenin mRNA levels in the BMP-2-treated cultures but little variation over the different time points. (B) N-cadherin mRNA levels peak at 18 hours in both the control and BMP-2-treated cultures, with a progressive fall in mRNA levels after that time. There is a significant reduction in N-cadherin mRNA levels by 72 hours after high density plating in the BMP-2-treated cultures (p < 0.009).

Beta catenin protein levels rose from 2 to 18 hours after high density plating in the control cultures and then stabilized (Figure 3A). In contrast, N-cadherin protein levels peaked later at 48 hours in both the control cultures and the BMP-2-treated cultures. BMP-2 treatment led to a significant fall in beta catenin and N-cadherin protein levels at 72 hours. To further quantify the changes in beta catenin protein levels, three separate immunoblots derived from cultures prepared during independent experiments were scanned, and the density of the bands estimated using Image Quant (Molecular Dynamics) software. In the control cultures, beta catenin protein levels increase shortly after high-density plating (Figure 3B) but then remain constant from 18 to 72 hours. The BMP-2-treated cultures show a biphasic response with an initial 1.5-fold increase in beta catenin and then a progressive fall in beta catenin levels after 48 hours with a two-fold reduction in beta catenin protein levels compared to controls by 72 hours.

Figure 3 Cells undergoing BMP-2-induced chondrogenic differentiation show modulation of beta catenin and N-cadherin protein levels. Cells were plated as micromass cultures and cultured in serum free conditions, either with BMP-2, 200 ng/ml or vehicle only (PBS). (A). Western blot analysis shows reduced levels of beta catenin protein at 48 and 72 hours in the BMP-2-treated cultures. N-cadherin levels are also reduced at 72 hours in the BMP-2-treated cultures. Equal protein loading is confirmed by reprobing for beta actin. (B). Quantitative analysis of the levels of beta catenin protein at each time point show a biphasic response to BMP-2 treatment with an initial rise in levels followed by a two-fold fall in total beta catenin protein levels by 72 hours. * p < 0.05. Experiment performed in triplicate. (C). Equal concentrations of Triton X-100-soluble (noncadherin-bound) and insoluble (cadherin-bound) cell fractions were analyzed by gel electrophoresis and western blot. Levels of Triton-X soluble beta catenin do not change over the 72 hours, levels of Triton-X insoluble beta catenin are increased at 2 and 18 hours after high density plating in BMP-2-treated cells but reduced by 72 hours.

BMP-2-Induced Chondrogenic Differentiation in Primary Mesenchymal Cells Is Not Associated with Changes in Cytoplasmic Levels of Beta Catenin or Catenin-Mediated Nuclear Signaling

Beta catenin is a multi-functional protein whose activities depend on its subcellular localization. Cytoplasmic accumulation often correlates with the entry of beta catenin into the nucleus and transcriptional activity, while membrane-bound beta catenin acts as a component of the adherens junctions (20). Dynamic changes in the distribution of beta catenin have been correlated with keratinocyte differentiation (42). We, therefore, examined the subcellular localization of beta catenin during BMP-2-induced chondrogenic differentiation.

Figure 3C shows that beta catenin is located in both the Triton-X insoluble and the Triton-X soluble fractions during chondrogenic differentiation. No change is seen in the levels of beta catenin protein present in the Triton-X soluble fraction during differentiation. Biphasic changes are seen in the levels of Triton-X insoluble beta catenin, which correlate well with the changes in total beta catenin protein levels seen during BMP-2-enhanced chondrogenic differentiation.

Confocal microscopy was further used to confirm the subcellular localization of beta catenin. Immunofluorescent staining and analysis by confocal microscopy at the previously used time points confirmed expression of beta catenin in the membrane, both in control and BMP-2-treated cultures. Figure 4 shows one representative immunofluorescent staining following high-density plating. Beta catenin is localized to the membrane at points of cell-cell contact. Beta catenin immunoreactivity could not be detected in the nucleus in the limb mesenchymal cells, however low levels of beta catenin immunoreactivity were detected in the cytoplasm in a perinuclear staining pattern. Similar patterns of localization of beta catenin were seen at all time points in both control and BMP-2-treated cultures.

Figure 4 Beta catenin protein is immunolocalized to the membrane and cytoplasm during chondrogenic differentiation. Micromass cultures were plated at a density of 1× 10⁷cells/ml, fixed at defined time points and labeled for beta catenin using an affinity-purified polyclonal antibody against beta catenin (Sigma) and secondary goat anti rabbit Cy3. Image overlay of immunolabeled cells (A) and nuclei labeled with DAPI (B) shows that beta catenin is localized to the cell membrane in the regions of cell-cell contact. Nuclear staining with beta catenin antibody is not detected (C). Similar findings were seen at all time points in both control cells and in cells treated with BMP-2 (not shown). (See Color Plate IV at the end of this issue).

COLOR PLATE IV See W. V. Zhang and N. S. Stott, Figure 4.

To detect activation of beta catenin signalling, we employed a beta catenin responsive reporter vector that has been shown to be functional in chicken cells (43). The reporter consists of a beta catenin responsive promoter (TOPFLASH) or a mutated nonfunctional version (FOPFLASH) linked to a luciferase reporter. Levels of beta catenin signaling can be quantified by measuring the activity of the luciferase reporter gene product. Primary mesenchymal cells were electroporated with TOPFLASH or control FOPFLASH vectors plus a beta galactosidase reporter and then cultured at low density overnight prior to being plated at high density and treated with BMP-2. Cell extracts were prepared at defined time points and assayed for luciferase activity normalized to beta galactosidase activity. There was no increase in luciferase activity in the TOPFLASH transfected cultures, suggesting that nuclear signaling was not occurring at these time points (Figure 5).

Figure 5 BMP-2 does not increase beta catenin nuclear signaling. Primary limb cells were electroporated with either the TCF responsive plasmid TOPFLASH or the FOPFLASH control vector, together with a beta galactosidase reporter vector, and then plated as micromass cultures. Luciferase activity at defined time points after plating was normalized to beta galactosidase activity and graphed as relative luciferase units. The TOPFLASH-transfected cells show the same level of luciferase activity as do FOPFLASH-transfected cells.

Cell-Cell Adhesion Is Decreased in Cultures Treated with BMP-2

We modified an assay described by Hinck et al. 1994 (44) to explore changes in the strength of calcium dependent cell-cell adhesion in micromass culture as chondrogenic differentiation proceeded. This assay tests the strength of cell-cell adhesion following vigorous trituration in either calcium-free or calcium-present conditions. Micromass cultures were initially treated with trypsin in HBSS with 0.1 mM CaCl₂ and then scraped off the dish into PBS. Trypsin with calcium has been reported to leave cadherins intact, unlike trypsin EDTA. Quantification was carried out by counting the number of single cells released by mechanical disruption with a 1 ml Pipetman after 15 minutes digestion with trypsin-calcium normalized to the total number of single cells subsequently recovered after further treatment of the same cells with calcium-free trypsin-collagenase and trituration. This assay confirmed that there is a progressive increase in calcium dependent cell-cell adhesion in control cultures over the first 72 hours after plating in control cultures (Figure 6). BMP-2-treated cultures have the same level of calcium dependent cell-cell adhesion as control cultures in the first 18 hours after plating but then show decreased calcium dependent cell-cell adhesion by 48 hours, detaching easily as single cells by 72 hours (Figure 6).

Figure 6 BMP-2 treatment reduces the adhesiveness of primary limb bud cells: Cells were plated as micromass cultures and cultured in serum free conditions, either with BMP-2, 200 ng/ml or vehicle only (PBS). Cells were harvested at defined time points and assayed for cell-cell adhesiveness. The data shows that BMP-2 treatment does not change cell-cell adhesiveness in the first 48 hours after plating, but that at 48 and 72 hours there is a significant reduction in cell-cell adhesiveness.

Lithium Chloride Treatment Modulates the Response to BMP-2

Lithium chloride is a Wnt mimic and has been shown to elevate levels of both membrane-bound and cytoplasmic beta catenin through inhibition of beta catenin degradation (44, 45), leading to changes in cell-cell adhesiveness. We asked if treatment with lithium chloride could modulate the effect of BMP-2 on beta catenin levels and cell adhesiveness. Treatment of control micromass cultures with 5 mM lithium chloride did not further increase the high level of cell-cell adhesiveness at 72 hours (data not shown). However, addition of 5 mM lithium chloride to the culture medium at the time of high-density plating did inhibit the BMP-2-modulated change in cell-cell adhesiveness seen at 72 hours (Figure 7A, B). No changes were seen in cell proliferation or cell number during this time (data not shown). Immunoblot confirmed that the levels of both membrane-bound and cytoplasmic beta catenin were increased at 72 hours in the BMP-2-treated cultures after addition of lithium chloride (Figure 7C). The level of BMP-2-induced chondrogenic differentiation at 96 hours was also reduced by the addition of 5 mM lithium chloride to the medium at the time of high density plating (Figure 8).

Figure 7 Lithium chloride treatment reverses the change in cellular adhesiveness seen at 72 hours in BMP-2-treated micromass cultures. (A): Micrographs showing the morphology of control and BMP-2-treated 72-hour micromass culture after digestion with trypsin in the presence of 1 mM CaCl₂ and mechanical disruption. Control cells are resistant to disruption in the presence of calcium while BMP-2-treated cells dissociate easily into single cells. Lithium chloride treatment in addition to BMP-2 reverses the effect of BMP-2 on cell adhesion. All cultures dissociate into single cells after trituration in calcium-free conditions. (B): Histogram representing the percentage of single cells following digestion in the presence of calcium normalized to the total number of cells in the culture prior to trituration under the three different conditions. (C): Western Blot analysis shows the levels of both Triton-X soluble (cytoplasmic) and insoluble (membrane bound) beta catenin are increased at 72 hours in the BMP-2-treated cultures after the addition of 5 mM lithium chloride at the time of high density plating. (See Color Plate V at the end of this issue)

COLOR PLATE V See W. V. Zhang and N. S. Stott, Figure 7.

Figure 8 Lithium chloride treatment reduces chondrogenic differentiation in BMP-2-treated cultures. Micromass cultures were plated at a density of 2 × 10⁷ cells/ml and cultured in serum-free conditions, with addition of 5 mM lithium chloride or BMP-2, 200 ng/ml; or BMP-2, 200 ng/ml, plus 5 mM lithium chloride. Cultures were fixed at 96 hours with paraformaldehyde and stained with 1% Alcian blue dye 8GX in 0.1N HCl, pH 1 for three hours to measure levels of cartilaginous extracellular matrix. Alcian blue dye was then extracted and the levels quantitated by spectrophotometer. The cultures treated with BMP-2 and lithium chloride show lower levels of cartilaginous extracellular matrix at 96 hours.

DISCUSSION

We have used the primary limb bud micromass culture system to study BMP-2-induced chondrogenic differentiation and the role of beta catenin signaling. Here we present evidence that the majority of beta catenin protein is membrane-bound in primary limb bud cells plated at high density. Furthermore, we show that BMP-2 treatment leads to a biphasic modulation of the levels of Triton-X insoluble membrane-bound beta catenin protein over the 72 hours of culture while the levels of Triton-X soluble nonmembrane-bound beta catenin are unchanged. The fall in levels of membrane-bound beta catenin 48 hours after BMP-2 treatment correlates with decreased calcium dependent cell-cell adhesion by 48 hours and a fall in N-cadherin protein levels by 72 hours. Localization and signaling activity of beta catenin in the nucleus is not detectable during BMP-2-induced chondrogenic differentiation. Elevation of membrane-bound beta catenin levels by treatment with lithium chloride blocks the effect of BMP-2 on cell-cell adhesion and leads to a reduction in subsequent chondrogenic differentiation. These data show that BMP-2-induced chondrogenic differentiation is associated with rapid modulation of membrane-bound beta catenin and N-cadherin protein levels with subsequent changes in calcium dependent cell-cell adhesion and thus, by implication, regulation of the adhesivity of adherens junctions.

Our data is at odds with the report by Fischer et al. (2001) who found that BMP-2-modulated chondrogenic differentiation of the pluripotential mouse cell line C310T1/2 is associated with nuclear accumulation of beta catenin, possibly leading to activation of cartilage specific genes (32). Hartman and Tabin (2000) also reported acceleration of chondrogenic differentiation following over-expression of beta catenin in the developing chick limb bud (46). However, in contradistinction to these findings, are the recent data reported by Ryu et al. (2002), which suggest that accumulation of nuclear beta catenin in articular chondrocytes is associated with dedifferentiation, not chondrogenic maturation (47). Ryu et al. (2002) further postulated that failure to down-regulate membrane-bound beta catenin may block chondrogenic differentiation (47). Our data confirms and extends their work by showing that a member of the TGF beta super-family, BMP-2, induces rapid chondrogenic differentiation through down-regulation of membrane-bound beta catenin and N-cadherin in nontransformed primary mesenchymal cells.

We did not find any fluctuation in beta catenin mRNA expression throughout the time-course of the study, either in control or BMP-2-treated cultures. However, N-cadherin mRNA levels peaked at 18 hours in our control high-density micromass culture, leading to a transient rise in N-cadherin protein levels at 48 hours, the time of maximal cellular aggregation. Temporospatial regulation of cellular adhesion is crucial for the progression from undifferentiated single limb mesenchymal cells to fully differentiated cartilage nodules (1, 48, 49). Reduced cellular adhesiveness through functional inhibition of N-cadherin or N-CAM has been shown to lead to impairment of condensation and a consequent reduction in the extent of chondrogenic differentiation (2, 3, 4, 49). Thus the rise in N-cadherin protein levels in control cultures at the time of maximum cellular aggregation is consistent with its known role in modulation of cell-cell adhesion via participation in the formation of adherens junctions (1).

BMP-2 treatment did not affect the peak in N-cadherin mRNA expression or the transient rise in N-cadherin protein but did augment the fall in N-cadherin protein and mRNA at 72 hours. This is in contrast to C3H10T1/2 mesenchymal stem cell line where BMP-2 treatment enhances chondrogenic differentiation through an up-regulation of N-cadherin mRNA and protein expression (50). One explanation for this difference may be that the action of BMP-2 in the C3H10T1/2 cell line is to enhance early N-cadherin expression, leading to increased cellular aggregation and formation of precartilaginous condensations which then undergo chondrogenic differentiation. In contrast, the action of BMP-2 on the primary limb mesenchymal cells may be to modulate levels of N-cadherin and beta catenin after the period of condensation, leading to effects on differentiation rather than on condensation. Interestingly, prolonged expression of N-cadherin in micromass culture, either through retroviral overexpression or retinoic acid treatment, has been shown to lead to a reduction in chondrogenic differentiation (48, 49), possibly due to a failure to down-regulate cellular adhesiveness. Likewise, Haas and Tuan (1999) noted that stably transfected cells with a high (four-fold) increase in N-cadherin showed an inhibition of chondrogenic differentiation rather than enhancement of differentiation (50). Our data extends this information, showing that an appropriately timed fall in the expression levels of N-cadherin and beta catenin is associated with enhanced chondrogenic differentiation. Overall, our data suggests that BMP-2 does not affect cellular aggregation and the formation of the precartilaginous condensation in this culture system but rather modulates chondrogenic differentiation by inducing a decrease in cellular adhesiveness following the formation of the precartilaginous condensation.

Our data does not delineate whether there is a causal relationship between the changes in levels of beta catenin protein and N-cadherin protein. However, the increased cell-cell adhesion and inhibition of BMP-2-induced chondrogenic differentiation seen after lithium chloride induced rise in levels of membrane-bound beta catenin suggests that modulation of beta catenin expression levels is another area of potential control of cellular adhesiveness. The involvement of beta catenin in chondrogenic differentiation has also been suggested by the chondro-inhibitory effect of various Wnt gene family members, such as Wnt-7a and Wnt-1, which have been shown to increase cellular adhesiveness (44, 45, 51, 52). This data would also be consistent with the transgenic mouse knockout studies that have shown that cell-cell adhesion is markedly reduced in the absence of beta catenin (24).

Recent work supports the concept of a dynamic exchange of beta catenin between the membrane and the cytoplasm with continuous synthesis and degradation of beta catenin protein (23, 53). Newly synthesized beta catenin protein can displace existing beta catenin from adherens junctions in the membrane within 30 minutes of synthesis, with the released beta catenin either participating in further exchange cycles with the cadherins or becoming part of the cytoplasmic pool of beta catenin. Intriguingly, once in the cytoplasmic pool, the displaced beta catenin cannot compete with the newly synthesized beta catenin protein for interaction with the cadherins. This rapid turnover of beta catenin between the membrane and cytoplasm represents a potential site of action for BMP-2.

In summary, the work presented here supports a role for BMP-2 in the induction of chondrogenic differentiation in primary mesenchymal cells through modulation of membrane-bound beta catenin and N-cadherin and thus cellular adhesiveness. Our findings also contribute to the emerging understanding of the role of beta catenin in differentiation pathways in nontransformed primary mesenchymal cells.

Acknowledgements

pRCAS-lacZII was a kind gift from Dr. C. M. Chuong, University of Southern California. We thank Genetics Institute, San Diego, USA for the BMP-2. This study was funded by a Marsden Fund grant to Susan Stott together with grants from the Auckland Medical Research Foundation, the Wishbone Trust, and the University of Auckland Research Committee.