N-cadherin activation substitutes for the cell contact control in cell cycle arrest and myogenic differentiation: involvement of p120 and beta-catenin.

N-cadherin is expressed throughout skeletal myogenesis and has been proposed to be involved in the differentiation program of myogenic precursors. Here, we further characterize the N-cadherin involvement and its mechanism of action at the onset of differentiation, through controlled N-cadherin activation by plating isolated C2 myoblasts on surfaces coated with a chimeric Ncad-Fc homophilic ligand (N-cadherin ectodomain fused to the immunoglobulin G Fc fragment). We show that N-cadherin activation substitutes for the cell density in myogenic differentiation by promoting myogenin and troponin T expression. In addition, N-cadherin adhesion participates to the associated cell cycle arrest through the nuclear accumulation of cyclin-dependent kinase inhibitors p21 and p27. Mouse primary myoblast cultures exhibited similar responses to N-cadherin as C2 cells. RNA interference knockdowns of the N-cadherin-associated cytoplasmic proteins p120 and beta-catenin produced opposite effects on the differentiation pathway. p120 silencing resulted in a decreased myogenic differentiation, associated with a reduction in cadherin-catenin content, which may explain its action on myogenic differentiation. beta-Catenin silencing led to a stimulatory effect on myogenin expression, without any effect on cell cycle. Our results demonstrate that N-cadherin adhesion may account for cell-cell contact-dependent cell cycle arrest and differentiation of myogenic cells, involving regulation through p120 and beta-catenins.

Skeletal myogenesis is a multi-step process, regulated by spatiotemporal cues in the embryo. The determination of the myogenic lineage, the differentiation of precursors in myoblasts, and their terminal differentiation leading to the formation of innervated muscle fibers are regulated by environmental signals provided by diffusible factors, extracellular matrix, and intercellular contacts (1,2). Some of these developmental events require a high degree of coordination between changes in myoblast adhesion and the progression toward myogenesis, particularly during migration, alignment, and fusion of myoblasts associated with their exit from the cell cycle and the induction of histo-differentiation genes (3). The adhesive properties of myogenic cells are modulated by the orchestrated expression of a combination of cell adhesion molecules, such as cadherins. At least four classical cadherins (R-, N-, M-, and 11) are expressed in embryonic muscles during mouse or chicken development (4,5). Some cell adhesion molecules of the immunoglobulin superfamily have also been proposed to play a role in myogenic differentiation (6 -9). However, the precise function of these cell adhesion molecules in the control of muscle differentiation remains elusive.
Cadherins are major constituents of intercellular junctions mediating cell adhesion through Ca 2ϩ -dependent homophilic interaction of their extracellular domain and anchoring of their cytoplasmic domain to the actin cytoskeleton by catenins (10,11). They contribute to cell aggregation, segregation, and migration associated with dynamic cell-cell contact remodeling (10,12,13). A role for cadherins in myogenesis was first evidenced by functional perturbation of N-cadherin during Xenopus development (14). The inhibition of N-cadherin-mediated contacts by specific antibodies perturbs myoblast fusion in vitro (7) and impairs the induction of muscle-specific genes (5,15). Conversely, forced expression of N-cadherin in myogenic behavior fibroblastic cells promotes the expression of skeletal muscle proteins in three-dimensional cultures (16,17). When beads coated with the extracellular domain of N-cadherin were applied on subconfluent cultures of C2 myoblasts, myogenic differentiation was enhanced (18). However, myoblasts from mice bearing a null mutation of the N-cadherin gene differentiate normally in vitro and in vivo, likely because of a functional redundancy of cadherins expressed in skeletal muscle (19). This hypothesis is strengthened by the fact that cadherins R and M are expressed during myogenesis in vivo and have been shown to participate in skeletal muscle differentiation in vitro (20 -25). These results mainly rely on functional perturbation or overexpression of cadherins. However, no study has investigated whether the activation of cadherins (in the manner of growth factor receptors) is able to trigger myogenic cell differentiation autonomously.
Recently, Ncad-Fc recombinant chimera (extracellular domain of cadherin fused to Fc fragment of immunoglobulin) were developed as tools to control cadherin homophilic binding and to study cellular responses triggered by cadherin activation, which resulted in significant progress in the determination of cadherin-initiated signaling pathways (26 -30). We used this approach to investigate whether isolated N-cadherin activation could control myogenic differentiation. Our work demonstrates that N-cadherin engagement, independently of other cell adhesion molecules, is sufficient to induce cell cycle arrest and myogenic differentiation of C2 cells and primary myoblasts. The p120 catenin is required for myogenic differentiation, whereas ␤-catenin inhibits myogenic differentiation, independently of cell cycle regulation.

EXPERIMENTAL PROCEDURES
Cell Culture and Induction of Myogenic Differentiation-The mouse myogenic C2 cell line (31) was maintained undifferentiated at low density in Dulbecco's modified Eagle's medium (DMEM) 1 supplemented with 10% fetal calf serum (growth medium (GM)) under 7.5% CO 2 at 37°C. To induce differentiation, C2 cells were placed at high density (10 5 cells/cm 2 ) in DMEM plus 2% horse serum (differentiation medium (DM)). For primary myoblast cell cultures, newborn mice (OF1 strain) were sacrificed according to guidelines of the local animal experimental ethics committee. The limbs were rinsed in PBS before dissecting muscle from the bones and cartilage. Tissues were incubated in PBS, 0.125% trypsin without EDTA for 30 min at 37°C and dissociated by several flushes. The cell pellet was resuspended in DMEM supplemented with 10% fetal calf serum and 10% horse serum and preplated for 6 h on uncoated tissue culture dishes. The floating cell population enriched in myoblasts was then allowed to attach on gelatin-coated dishes at 10 4 cells/cm 2 (32). After 2 days (three or four divisions) allowing the formation of small colonies, the culture medium was switched to DMEM plus 10% horse serum for differentiation.
Antibodies and Reagents-The following antibodies were used for immunostaining: rabbit anti-myogenin (1:100, M225 clone; Santa Cruz, Tebu, Le Perray-en-Yvelines, France), mouse anti-Troponin T (1:50, clone JLT-12; Sigma), rabbit anti-p21 (1:200, C-19; Santa Cruz), and rabbit anti-p27 Kip1  DNA and siRNA Transfections-C2 cells grown at 5 ϫ 10 4 cells/cm 2 (mid-confluence) in GM without antibiotics were transfected with the LipofectAMINE 2000 reagent according to the manufacturer's instructions (Invitrogen). 5 l of LipofectAMINE 2000 were mixed with 2 g of DNA expression vector or 100 -500 nM of dsRNA duplexes in 100 l of Opti-MEM medium for 30 min. The cells were incubated with this mixture for 24 h in GM without antibiotics, and differentiation was induced by medium switch when cells reached confluence. The cells were either fixed for immunostaining or submitted to protein extraction for Western blotting, after 2 days in DM corresponding to 3 days post-lipofection. The following DNA vectors were used: (control DNA) pSUPER (33) for mock transfections and pSUPER-p120 (30) for p120 silencing containing the insertion in both sense and antisense orientations of the specific p120 sequence: GATGGTTATCCAGGTGGCA, corresponding to the 655-673 amino acids of the mouse sequence (NM007615). The RNA duplex sequence used for ␤-catenin silencing corresponds to the 492-498 amino acids of the mouse sequence (NM007614) and was noted ␤-cat siRNA: rCrUrGrUrUrGrUrGrGrU-rUrArArArCrUrCrCrUTT (Proligo, Paris, France). As a control RNA, we used inefficient RNA duplex corresponding to the 79 -85 amino acids sequence (control dsRNA): rArGrCrUrGrArUrArUrGrArCrGrGrGrCr-ArGTT.
Preparation of Adhesion Substrates and N-cadherin Activation-Fibronectin (Fn, 1 g/cm 2 ; PAA Laboratories, GmbH Linz, Austria) and N-cadherin homophilic ligand, Ncad-Fc (Ncad-Fc, 5 g/cm 2 ) (26) were coated on silanized glass coverslips or thermosterilized bacterial dishes (Falcon, Becton Dickinson Europe) as reported previously (30). Alternatively, soluble chimera was directly diluted in DM at the final concentration of 50 g/ml in the presence of 3% bovine serum albumin. The C2 cells were mechanically dissociated in PBS, 5 mM EDTA, 2% bovine serum albumin on ice and plated on different coated surfaces in the absence of serum and at low density (ld; 10 3 cells/cm 2 ) (30). After 2 h of adhesion, the medium was changed to GM or DM and renewed every day during the time of the experiments.
Immunostainings and BrdU Incorporation-C2 cells were fixed in PBS with 3% formaldehyde at 37°C for 15 min, washed in PBS with 0.1 M glycine, and permeabilized in PBS with 0.5% Triton for 5 min. Primary antibodies were revealed using goat fluorescein isothiocyanate-or TRITC-conjugated anti-mouse or anti-rabbit immunoglobulins G (Jackson Immunoresearch, West Grove, PA) or goat AlexaFluor 488 or AlexaFluor 546-conjugated anti-mouse or anti-rabbit immunoglobulins G (1:500; Molecular Probes Europe, Leiden, Holland). The nuclei were counterstained with Hoechst-33258 (1:20,000; Molecular Probes). The samples mounted in Mowiol (Calbiochem) were observed by conventional microscopy (Provis; Olympus Optical Co., Tokyo, Japan), and the images were captured with a Micromax CCD camera driven with the Metamorph software (Roper Scientific, Trenton, NJ). To visualize cells in the S phase, BrdU (10 M; Roche Applied Science) was added to the medium 4 h before cell fixation. Fixed cells were treated with PBS-Triton 0.5% plus HCl 2 N at 37°C during 30 min to denature DNA, rinsed extensively with borate 0.1 M pH 8.0 and PBS before immunostaining with mouse fluorescein isothiocyanate-conjugated anti-BrdU antibody (1:200; Roche Applied Science).
Quantitative Analysis-Immunostaining experiments were quantified by manual counts of the positive nuclei (except positive cytoplasm for troponin T labeling) related to the total number of nuclei stained by Hoechst. Microscopy fields were taken under constant time illumination, oriented up to down and left to right on the glass coverslips, always beginning by Hoechst labeling. Five fields were counted corresponding to 500 cells in hd conditions; at least 10 fields were counted corresponding to 100 cells in ld conditions. Each experiment was independently repeated three times. The results are expressed as the percentages of the mean of the positive/total ratios (except ratio of control for siRNA experiment performed after 2 days of DM culture). The 2 test was used for statistical analysis of all the data sets: * for p Ͻ 0.1, ** for p Ͻ 0.01, *** for p Ͻ 0.001, and **** for p Ͻ 0.0001.
Whole Cell Lysates and Western Blotting-To prepare whole cell lysates, the samples were incubated in 200 l of 10 mM Tris/HCl, pH 7.5, 2 mM EDTA, and 1% SDS for 20 min on ice, scraped, and submitted to sonication. Equal amounts of proteins, 10 -50 g/lane estimated by micro-BCA (Pierce) were analyzed by 7 or 13% SDS-PAGE and blotted on 0.2-m nitrocellulose membranes, as previously described (34). The secondary goat horseradish peroxidase-conjugated anti-mouse or antirabbit antibodies were purchased from Dako (1:10000, Glostrup, Denmark). Blots were developed for chemiluminescence using the ECL Western blotting reagents (Amersham Biosciences) and could be rehybridized using the Western stripping kit from Pierce.

RESULTS
Mouse C2 cells (31) are classically induced to differentiate at high density (hd) in low serum-containing medium (DM). Although proliferating C2 cells express the transcription factor MyoD (80% of MyoD-immunopositive cells; data not shown), the related transcription factor myogenin is an early marker of their differentiation (35). The expression of myogenin at hd:DM was detected after 1 day and increased up to day 3 ( Fig. 1A), whereas undetectable in cells cultured for 3 days at low density in DM (ld:DM), or maintained at high density in growth medium (hd:GM). Thus, myogenin expression is dependent both on serum content and cell density, likely through the establishment of cell-cell contacts. We hypothesized that N-cadherin may be a molecular factor involved in the cell-cell contact effect.
N-cadherin-based Adhesion Controls the Induction of Myogenic Differentiation-In the high density conditions, cells integrate information from both cell-cell and cell-matrix contacts. Selective engagement of N-cadherin was achieved by spreading cells at low density on Ncad-Fc coated surfaces (30), allowing N-cadherin "activation" or "engagement" independently of other stimulations emerging from cell-cell or cell-matrix contacts (Fig. 1B). As a control for the contribution of cell-matrix contact, C2 cells were spread on the extracellular matrix protein Fn or laminin (data not shown) at low density. The contribution of N-cadherin-based adhesion to the C2 cell differen- 1 The abbreviations used are: DMEM, Dulbecco's modified Eagle's medium; siRNA, small interfering RNA; dsRNA, double strand RNA; DM, differentiation medium; GM, growth medium; ld, low density of cells; hd, high density of cells; md, medium density of cell; Fn, fibronectin; PBS, phosphate-buffered saline; TRITC, tetramethylrhodamine isothiocyanate; BrdU, bromodeoxyuridine; ␤-cat, ␤-catenin.
tiation was assessed through the expression of myogenin and the histodifferentiation marker troponin T. Myogenin and troponin T immunolabeling were each present in around 30% of cells after 3 days of culture on Ncad-Fc in DM, similar to hd:DM conditions (Fig. 1, C and D). In contrast, the relative numbers of myogenin-positive nuclei and troponin T-positive cells were significantly lower on fibronectin (Fig. 1, C and D). The induction of troponin T expression either by hd or by Ncad-Fc was confirmed by Western blot analysis (Fig. 1E). The time course of appearance of myogenin-positive nuclei in C2 cells cultured at low density on Ncad-Fc was indistinguishable from that observed in hd:DM conditions (Fig. 1F), demonstrating that N-cadherin activation promotes myogenin expression from day 1, as efficiently as hd. Because myogenin-positive nuclei were undetectable in GM on Ncad-Fc (data not shown), myogenin expression requires serum withdrawal in addition to N-cadherin activation. Thus, these results indicate that Ncadherin activation is sufficient to mimic the effect of cell-cell contact on the myogenic differentiation of C2 cells.
To characterize the involvement of N-cadherin in this process, we performed a dose-response analysis of N-cadherin-triggered myogenin expression. The myogenin-positive nuclei at day 3 was low for Ncad-Fc concentration inferior to 2 g/cm 2 and increased with Ncad-Fc concentration to reach a plateau at 5 g/cm 2 (Fig. 1G). Soluble Ncad-Fc was unable to induce myogenin expression at ld:DM with concentrations as high as 50 g/ml (data not shown). Our data suggest that the N-cadherin adhesion receptor induces myogenin expression through a dose-dependent, solid phase homophilic binding to the Ncadherin extracellular domain (i.e. N-cadherin activation). Therefore, N-cadherin is sufficient to substitute for the cell-cell contact in the myogenic differentiation.
N-cadherin Activation Triggers Cell Cycle Arrest Prior to Differentiation-Previous reports show that N-cadherin overexpression leads to the up-regulation of p27 in Chinese hamster ovary fibroblastic cells (36), controlling cell cycle arrest. We wondered whether N-cadherin activation could be involved autonomously in the cell cycle arrest during differentiation program of myoblasts. The nuclear accumulation of the cyclindependent kinase (cdk) inhibitors p21 and p27 was analyzed by immunostaining after 3 days in DM. Isolated C2 cells on Ncad-Fc exhibited 57 and 76% of p21-and p27-positive nuclei, respectively (Fig. 2, A and B), similar proportions of cdk inhibitor-positive nuclei (75 and 82% for p21 and p27, respectively) were observed for hd:DM conditions (Fig. 2B). In contrast, C2 cells cultured on fibronectin exhibited only 30% of cdk inhibitor-positive nuclei (Fig. 2, A and B). In addition to this nuclear accumulation of cdk inhibitors, Western blot analysis reveals higher levels of p27 in hd:DM or in ld:DM on Ncad-Fc, compared with fibronectin (Fig. 2C). Thus, the activation of Ncadherin induces p21 and p27 nuclear accumulation during myogenic differentiation, supporting the notion that N-cadherin activation mimics the cell-cell contact action on cell cycle arrest.
We also performed different approaches to directly evaluate the action of isolated N-cadherin activation on cell cycle progression. The proliferation of C2 cells on Ncad-Fc or fibronectin was estimated by counting viable cells cultivated in ld:GM or ld:DM conditions (Fig. 2D). The number of cells was increased by 10-fold on fibronectin and only by 4-fold on Ncad-Fc on the third day in GM. C2 cells also proliferated on fibronectin in low serum medium (6-fold increase), whereas the combination of Ncad-Fc and low serum led to a complete abrogation of cell multiplication. The involvement of N-cadherin in the adhesiondependent cell cycle arrest was further evaluated through BrdU incorporation and p21 coimmunostaining after 24 h. Although 42% (DM) and 60% (GM) of cells on fibronectin displayed BrdU incorporation together with negative staining for p21 (BrdU ϩ /p21 Ϫ ), they were only 11% (DM) and 23% (GM) on Ncad-Fc (Fig. 2E). Similar proportions of BrdU ϩ /p21 Ϫ nuclei were observed at hd:DM (Fig. 2E). Conversely, only 26% of cells cultivated on fibronectin were negative for BrdU and positive for p21 (BrdU Ϫ /p21 ϩ ), whereas 55% of cells on Ncad-Fc were BrdU Ϫ /p21 ϩ (compared with 63% for hd; Fig. 2E). Thus, our experiments demonstrate that N-cadherin activation is sufficient to diminish the proportion of cells that passes through the S phase, again suggesting that N-cadherin negatively regulates cell cycle progression during myogenic differentiation.
The link between cell cycle arrest and myogenic differentiation was then investigated through BrdU incorporation in combination with myogenin or p21 immunostaining at ld:DM on Ncad-Fc (Fig. 2F). After 1 day, 18% of cells were engaged in differentiation (BrdU Ϫ /myogenin ϩ ) whereas 50% were stopped in the cell cycle (BrdU Ϫ /p21 ϩ ). By 3 days, more than 76% of the cells belonged to the BrdU Ϫ /p21 ϩ population, but they were only 31% with BrdU Ϫ /myogenin ϩ staining. Moreover, less than 5% of myogenin-positive cells had incorporated BrdU (BrdU ϩ / myogenin ϩ ), suggesting that the induction of myogenin in cycling cells is a rare event. The uncycled and differentiated population (BrdU Ϫ /myogenin ϩ ) belongs at least partially to the population of cells stopped in cell cycle (BrdU Ϫ /p21 ϩ ). Altogether, our results show that N-cadherin receptor activation regulates cell cycle arrest associated with myogenic differentiation.
N-cadherin Activation Induces Differentiation of Mouse Primary Myoblasts-To expand the results obtained in the C2 cell line to a more physiologically relevant system, the involvement of N-cadherin activation was examined in primary myoblasts from newborn mice. Proliferating myoblasts were allowed to adhere on gelatin-coated dishes and to divide three or four FIG. 2. N-cadherin activation regulates cell cycle arrest of myogenic cells. A, C2 cells were plated at low density on Ncad-Fc or Fn and grown for 3 days in DM before fixation and immunostaining for p21 and p27. White arrowheads point toward p21-or p27-positive nuclei, and the open arrowheads point toward p21-or p27-negative nuclei. Bars, 10 m. B, quantitative analysis of p21 and p27 nuclear accumulation in three independent experiments. C, the expression level of p27 was analyzed by Western blot in cells grown for 3 days at hd:DM or at ld:DM on Ncad-Fc or Fn (representative of two independent experiments). D, proliferation of C2 cells plated at ld on Ncad-Fc or Fn in DM or GM was estimated by the number of viable cells 1, 2, and 3 days after plating. Time 0 corresponds to the time at which DM or GM is added after the 2 h for initial attachment in the absence of serum (two independent experiments). E, C2 cells were plated at ld on Ncad-Fc or Fn and grown for 1 day in DM or in GM. The cells were incubated with BrdU 4 h prior to fixation and coimmunostaining for BrdU and p21. The histograms present the quantitative analysis of BrdU ϩ /p21 Ϫ or BrdU Ϫ /p21 ϩ nuclei in three independent experiments. F, C2 cells grown at low density on Ncad-Fc for 1, 2, or 3 days were incubated with BrdU 4 h prior to fixation and immunostaining for BrdU and myogenin. BrdU Ϫ /myog ϩ , BrdU ϩ /myog Ϫ , BrdU ϩ /myog ϩ , and BrdU Ϫ /myog Ϫ nuclei were scored in two independent experiments and reported in the histogram with the independent count of BrdU Ϫ /p21 ϩ nuclei. times as small colonies (2 days), as previously described (32). Myoblasts were induced to differentiate by serum switch, and their proliferation/differentiation state was evaluated both by Western blotting and by immunostaining for myogenin or p21. We first observed by Western blot the expected induction of myogenin in differentiating myoblasts. Two-day-old mouse primary myoblasts did not express detectable levels of myogenin in GM, but myogenin expression was induced as early as 0.5 day in DM (Fig. 3A). Alternatively, 2-day-old proliferating myoblasts were platted as isolated cells on Ncad-Fc or fibronectin in DM. N-cadherin activation induced the nuclear accumulation of myogenin (69%) or p21 (67%) (Fig. 3, B and C), similar to typical differentiating conditions (colonies in DM on gelatin). Significantly less primary myoblasts were stained for myogenin (33%) or p21 (47%) on fibronectin (Fig. 3, B and C). Thus, as in the C2 cell line, N-cadherin activation regulates cell cycle arrest and differentiation of mouse primary myoblasts.
Experimental Modulation of Cadherin-Catenin Complexes during Myogenic Differentiation-Because p120 and ␤-catenin are directly associated to cadherins and regulate their functions (37), we postulated that they may be involved downstream of N-cadherin activation during myogenic differentiation. To investigate the contribution of these proteins to the control of myogenic differentiation, RNA interference strategy was chosen to reduce their expression. For p120, we used a previously described vector coding for a p120-specific dsRNA (30). Western blot analysis revealed that transfected cells exhibited a significant decrease in p120 protein after 3 days of culture either in proliferation (md:GM) or differentiation (hd: DM) conditions (Fig. 4A). Interestingly, p120 silencing led to a global decrease in the expression of cadherin and ␤-catenin, as previously reported (30,38). For ␤-catenin, we designed a synthetic siRNA duplex in its coding region. Western blot analysis revealed that efficient ␤-catenin silencing was achieved 3 days post-transfection with concentrations of ␤-catenin siRNA duplex ranging from 100 to 500 nM in GM (Fig. 4B) and also in DM at the median concentration (200 nM) used for further experiments (Fig. 4C). In contrast to p120 silencing, ␤-catenin knockdown did not alter significantly total cadherin expression. Interestingly, ␤-catenin siRNA led to an up-regulation of ␥-catenin/plakoglobin (Fig. 4, B and C), which may functionally replace ␤-catenin at the plasma membrane to link N-cadherin to ␣-catenin and to the actin cytoskeleton (10). p120 or ␤-Catenin Silencing Alters Myogenic Differentiation-The effect of p120 and ␤-catenin silencing on myogenic differentiation was evaluated by following myogenin expression in C2 cells in hd:DM. Immunostaining counts revealed a 40% decrease of myogenin-positive nuclei in p120 siRNA-expressing cells (Fig. 4D). Western blot analysis confirmed this decrease of myogenin expression (Fig. 4E). Thus, p120 siRNA associated with ␤-catenin and cadherin down-regulation counteracts myogenic differentiation. In contrast, ␤-catenin siRNA led to a 40% increase of myogenin-positive cells (Fig. 4D) and a global increase of myogenin expression (Fig. 4E). These results suggest that ␤-catenin silencing potentiates myogenic differentiation. This effect of ␤-catenin could be independent of its association with N-cadherin at the membrane, given the fact that ␥-catenin level was augmented. At the opposite of ␤-catenin silencing, C2 cells were submitted to lithium treatment known to stabilize and increase cytosolic pools of ␤-catenin (39). As previously reported (40), this treatment actually led to a global increase of ␤-catenin (Fig. 4F) and counteracted myogenin expression, but only if added after 1 day in DM (Fig. 4F). Thus, our results with silencing and stabilization of ␤-catenin converge to the hypothesis that ␤-catenin exerts a negative action on myogenic differentiation. To determine whether the negative effect of ␤-catenin on myogenic differentiation affects myogenin induction per se or through an effect on cell cycle, BrdU incorporation was followed in combination with myogenin immunostaining in ␤-catenin siRNA-treated cells in hd:DM conditions. ␤-Catenin silencing led to a 3-fold increase in the number of cells accumulating myogenin in their nuclei (Fig. 4,  G and H). A small but significant percentage of nuclei accumulated myogenin in GM (6%), which has never been observed in any other condition with GM. However, ␤-catenin silencing did not significantly interfere with BrdU incorporation (Fig. 4H) or p27 level of expression (data not shown), indicating no major effect on cell cycle regulation. Our results suggest together that ␤-catenin regulates negatively the induction of myogenic differentiation during the step of myogenin expression, by a process independent of cell cycle exit.

DISCUSSION
The differentiation of skeletal myogenic cells, classically induced by serum deprivation in vitro, has been reported to depend on cell density. We demonstrate here that N-cadherindependent homophilic adhesion recapitulates the cell contactdependent effect on the myogenic differentiation process: (i) N-cadherin engagement controls both cell cycle arrest and myogenin induction in C2 cells or in primary myoblasts, (ii) the catenin p120 positively affects myogenic induction likely through its stabilizing properties on the cadherin-catenin com- FIG. 3. N-cadherin activation promotes differentiation of mouse primary myoblasts. A, myoblasts were isolated from limb muscles of newborn mice and grown in proliferating conditions for 2 days on gelatin. The cells were switched to DM for 0.5 or 1 day, and the expression of myogenin was evaluated by Western blot. B, proliferating 2-day-old myoblasts were plated at low density on Ncad-Fc or Fn and cultured for 1 additional day in the presence of DM. They were fixed and immunostained for myogenin or p21 (red), and the nuclei were counterstained with Hoechst (blue). Bars, 10 m. C, quantitative analysis performed on two independent experiments of myogenin and p21-positive nuclei in ld on Ncad-Fc or Fn, in comparison with small colonies of 3-day-old myoblasts on gelatin (Gel). plex, and (iii) ␤-catenin contributes to a negative pathway during myogenic differentiation.
Previous reports based either on functional perturbation or overexpression of N-cadherin suggested a modulator role for N-cadherin in myogenic differentiation (5,7,(15)(16)(17). The involvement of N-cadherin in myogenic differentiation was also investigated through application of N-cadherin-coated beads upon C2 cells in culture, which in this case established actual cell-cell and cell-matrix contacts (18). Our results show that cell density-dependent myogenic differentiation of C2 cells is mimicked by N-cadherin engagement, without contribution of any other adhesion molecules: N-cadherin activation is indeed sufficient to promote myogenin and troponin T expression in isolated myoblasts. A similar positive effect on myogenin expression was observed in response to N-cadherin engagement in mouse primary myoblasts, extending our data on C2 cell line. However, because N-cadherin Ϫ/Ϫ myoblasts differentiate normally in vitro (19), M-or R-cadherins, which are also expressed by myogenic precursors (4, 5), could equally regulate the cell contact-dependent differentiation. In conclusion, cadherins may act in vivo as adhesion and signaling sensors between neighboring cells for the induction of myogenic differentiation. In addition, these cadherins of different specificity may differentially control the site of differentiation, specifying myogenic cell subpopulations, as suggested for motoneurons (41).
Cadherins E, N, or VE exert a negative regulation on cell proliferation in various cell types (36,(42)(43)(44) and during tumor cell growth (45,46). In particular, the overexpression of Ncadherin leads to the up-regulation of p27 in Chinese hamster ovary cells, modulating cell cycle arrest (36). In myogenic cells, N-cadherin activation triggers also the accumulation of p21 and p27 and diminution of DNA synthesis and cell proliferation, suggesting that N-cadherin activation regulates cell cycle exit during the course of myogenic differentiation. A main function for N-cadherin during myogenesis in vivo may be to regulate the proliferation/differentiation switch of myogenic precursors, through mechanisms corresponding in vitro to the contact inhibition of cell growth.
The induction of myogenin expression by N-cadherin acti- FIG. 4. Opposite effects of p120 and ␤-catenin silencing on myogenic differentiation. A, C2 cells were transfected with a p120 dsRNA producing vector (p120 siRNA) or with the empty vector (control DNA) and grown in DM or GM. Transfected cells were analyzed 3 days later by Western blot for their content in cadherin, p120, and ␤-catenin. B, ␤-catenin silencing in GM was evaluated by Western blot analysis of cells transfected with specific synthetic siRNA duplexes (␤-cat siRNA), from 100 to 500 nM) or control duplexes (con dsRNA, at 200 nM) (see also "Experimental Procedures"). The expression of cadherin and ␥-catenin was also analyzed. C, the cadherin and ␤and ␥-catenin cellular content was evaluated at day 3 in cells treated by ␤-cat siRNA or control dsRNA in DM or GM both at 200 nM. D, mid-confluent C2 cells (md:GM) were transfected with control dsRNA, ␤-cat siRNA, control DNA, or p120 siRNA, switched to DM at day 1 when cells reached confluence, and analyzed for myogenin expression at day 3. Quantitative analysis of myogenin immunostaining is presented as the ratio between ␤-cat siRNA/control dsRNA or p120 siRNA/control DNA conditions (three independent experiments). E, gene silencing efficiency and its effect on myogenin induction were evaluated by Western blot at day 3. F, C2 cells at hd:DM were treated with NaCl (30 mM) or LiCl (30 mM) from day 0 to day 3 (D0-D3) or from day 1 to day 3 (D1-D3). The effect of LiCl on ␤-catenin stability and myogenin expression was evaluated by Western blot at day 3. G, C2 cells in md:GM were transfected at day 0 either with control dsRNA or ␤-cat siRNA, grown in GM to reach confluence, and maintained in GM or switched to DM. The effect of ␤-catenin silencing on myogenic differentiation and cell cycle regulation was evaluated at day 2 by costaining for BrdU (green) and myogenin (red). H, the histogram presents the counts of BrdU ϩ /myogenin Ϫ (BrdU) and myogenin ϩ /BrdU Ϫ (myogenin)-positive nuclei. Each experiment was repeated three times.
vation is a dose-dependent process. The maximal cell response requires a significant level of molecules engaged in adhesion, likely mimicking extensive cell-cell contacts encountered in high density conditions. Moreover, myogenin expression required immobilized N-cadherin ligand, suggesting that the signaling activity of N-cadherin receptor is dependent on its adhesive engagement. Doherty and co-workers (47) reported that N-cadherin signaling associated to neurite outgrowth is triggered by soluble extracellular domain of N-cadherin involving fibroblast growth factor receptor clustering by direct binding of N-cadherin ectodomain. Although tyrosine kinase receptors (47,48) could be involved in the present response to N-cadherin activation, we favor a signaling pathway through the cadherin-associated catenins in the myogenic context.
Recent studies propose a positive function for p120 in cadherindependent adhesion through the regulation of stability and trafficking of the cadherin-catenin complexes (38, 49 -51). p120 silencing led to an important decrease in the number of differentiated cells expressing myogenin and to a global decrease in cadherin and ␤-catenin expression. Hence, p120 silencing may counteract myogenin induction through its negative action on cadherin adhesion. In addition p120 has been shown to modulate Rho family GTPase activity in response to balanced signals provided by integrins and cadherins (52)(53)(54) and may be involved in the regulation of myogenesis by controlling the activity of small GTPases (15,55). Further studies will be required to establish whether p120 physiologically regulates myogenic cell differentiation by modulating cadherin activity either through the cadherin-catenin availability and/or small GT-Pases activities.
Signaling initiated by N-cadherin homophilic binding may be also mediated by ␤-catenin, as proposed by Geiger and coworkers (40). Indeed, the accumulation of ␤-catenin triggered by LiCl or by the overexpression of stabilized mutant forms had a negative effect on myogenin (40). We did not observe changes in ␤-catenin subcellular localization during differentiation of C2 cells (data not shown). Moreover, the ␤-catenin knockdown was compensated by an augmentation in ␥-catenin/plakoglobin, substituting for ␤-catenin at the membrane (56,57). These observations suggest that cytosolic ␤-catenin might act negatively on myogenic differentiation through a mechanism independent of N-cadherin adhesion as previously suggested for ␤-catenin function during tumor cell proliferation (45,46). Although we might have suspected that the canonical Wnt/␤catenin/T cell factor pathway regulates the cell cycle progression of myogenic cells, the positive effect of ␤-catenin silencing on myogenin expression was not related to a modulation of cell cycle. Thus, ␤-catenin appears as a negative effector of cell differentiation in committed myogenic cells, acting independently of cell cycle arrest.
In conclusion, our results show that cadherin-based adhesion is required for myogenic differentiation. This process regulated by cadherin-associated proteins may ensure that only myogenic precursors located at the right place and having the appropriate neighbors withdraw from the cell cycle and activate the differentiation program. The function of environmental sensor described here for N-cadherin in myogenic cells may be a general property of cadherin-catenin adhesion complexes contributing to coordinated cell differentiation in other embryonic cell lineages.