N-cadherin ligation, but not Sonic hedgehog binding, initiates Cdo-dependent p38α/β MAPK signaling in skeletal myoblasts

Abstract

The p38α/β mitogen-activated protein kinase (MAPK) pathway promotes muscle-specific gene expression and myoblast differentiation but how pathway activity is initiated during these processes is poorly understood. During myoblast differentiation, the intracellular region of the promyogenic cell surface protein Cdo (also known as Cdon) binds to Bnip-2 and JLP, scaffold proteins for Cdc42 and p38α/β MAPK, respectively. The Bnip-2/Cdc42 and JLP/p38α/β complexes associate in a Cdo-dependent manner, resulting in Bnip-2/Cdc42-dependent p38α/β activation and stimulation of cell differentiation. Although the Cdo ectodomain binds to several different proteins, it is unclear how Cdo-dependent p38α/β activation is initiated. In myoblasts, Cdo interacts with the cell–cell adhesion molecule N-cadherin. Cdo also binds directly to the secreted morphogen Sonic hedgehog (Shh) to promote Shh pathway signaling. We report here that N-cadherin ligation activates p38α/β in myoblasts in a Cdo-, Bnip-2-, and JLP-dependent manner. Furthermore, these proteins and activated Cdc42 cluster at sites of N-cadherin ligation. In contrast, neither JLP nor Bnip-2 is associated with Cdo bound to Shh, and Shh does not activate p38α/β in myoblasts. Taken together, these results link cadherin-based cell–cell adhesion to a defined signaling pathway (Cdo → p38α/β) that directly regulates a cell-type-specific differentiation program. Furthermore, they are consistent with a model whereby Cdo serves as a multifunctional coreceptor with mechanistically distinct roles in multiple signaling pathways.

During cell differentiation, lineage-restricted precursor cells acquire a specialized transcriptional program that dictates tissue-specific structure and function. The mechanisms whereby extracellular cues and signal transduction pathways promote this process are of fundamental interest, but not well understood; for example, some signaling pathways necessary for differentiation of specific cell types are ubiquitous and also play roles in unrelated processes. An example is the p38α/β MAPK pathway, which is required for differentiation of several cell types, but is also involved in a variety of inflammatory and stress responses (1).

Differentiation of skeletal myoblasts is coordinated by transcription factors of the MyoD family, and p38α/β (hereafter simply p38) promotes myogenesis through phosphorylation of substrates that stimulate MyoD-dependent, muscle-specific gene expression (2–4). However, the mechanisms that initiate p38 activation in differentiating myoblasts are unclear. Multiprotein complexes that contain the promyogenic cell surface receptor Cdo play a significant (although not exclusive) role in p38 activation during myogenesis (5–7). Cdo^−/− mice and myoblasts display delayed skeletal muscle development and defective differentiation in vitro, respectively (7, 8). Similarly, RNAi-mediated depletion of Cdo from C2C12 myoblasts causes them to differentiate inefficiently (5, 7). Differentiation-dependent activation of p38 in cultured myoblasts is largely, but not completely, dependent on Cdo, and differentiation of Cdo^−/− myoblasts and Cdo-depleted C2C12 cells is rescued by expression of an activated form of the immediate upstream p38-activating kinase, MKK6 (7).

Cdo is a multifunctional cell surface protein with Ig and FnIII repeats in its ectodomain and a long intracellular region that does not resemble other proteins (9). During myoblast differentiation, the Cdo intracellular region is bound by Bnip-2, a scaffold protein for the small GTPase, Cdc42, and by JLP, a scaffold protein for the p38 pathway (6, 7). The Cdo–Bnip-2–Cdc42 complex stimulates Cdc42 activity, which in turn is required for differentiation-dependent p38 activity. Bnip-2 and JLP do not interact directly but associate through their mutual ability to bind Cdo, implying that Cdc42 bound to Cdo via Bnip-2 signals to activate p38 bound to Cdo via JLP (6). Cdo/Bnip-2/JLP/p38 signaling also promotes neuronal differentiation (10). In contrast, Cdo and Bnip-2 are dispensable for TNFα- and hyperosmotic stress-induced p38 activity, indicating that modes of p38 activation in cell differentiation and stress responses are distinct (6).

How Cdo-dependent p38 activation is initiated is not known. The Cdo ectodomain does not interact with itself in trans and has no obvious adhesive properties, but it binds several proteins as a putative coreceptor (9, 11–14). In myoblasts, Cdo forms cis complexes with the cell–cell adhesion molecule N-cadherin (Ncad) (11), which is itself promyogenic (15–20). Cell–cell contact stimulates myoblast differentiation, and although Ncad is not essential for myogenesis (likely due to redundancy with other cadherins) (21), direct Ncad ligation can substitute for cell–cell contact to enhance muscle-specific gene expression and myoblast differentiation (17, 19). The ubiquitous cadherin-associated protein p120-catenin is important for Ncad’s effects in myoblasts but the signaling mechanisms that drive a muscle-specific response are unknown (17, 22). Cdo also binds directly to the secreted morphogen, Sonic hedgehog (Shh), perhaps as a coreceptor for Patched1, to promote Shh pathway signaling (13, 14). Shh has been reported variously to promote or block differentiation of cultured myoblasts (23–25). In this study, we report that Ncad ligation activates p38 in a Cdo-, JLP-, and Bnip-2-dependent manner. Furthermore, these proteins and active Cdc42 are attracted to sites of Ncad ligation. In contrast, neither JLP nor Bnip-2 is associated with Cdo bound to Shh, and Shh does not activate p38 in myoblasts. These results link cadherin-based cell–cell adhesion to a defined signaling pathway that directly regulates a cell-type-specific differentiation program. Furthermore, they are consistent with a model whereby Cdo serves as a multifunctional coreceptor with mechanistically distinct roles in multiple signaling pathways.

Results and Discussion

Ncad Ligation Activates p38 in a Cdo-, JLP-, and Bnip-2-Dependent Manner.

To assess whether Ncad ligation activates p38 signaling in myoblasts, C2C12 cells were plated at low density (without any cell–cell contact) on culture dishes or coverslips coated with Ncad ectodomain-Fc fusion protein (Ncad-Fc). Under these conditions, Ncad-expressing cells attach and spread as if adhering to neighboring cells and undergo Ncad-dependent signaling, but without other juxtacrine signals that may occur indirectly as a consequence of cell–cell adhesion (26). Dishes and coverslips coated with fibronectin (which like Ncad promotes myogenesis) (27) or poly-L-lysine (PLL) were used as controls. Cells were visualized by staining with phalloidin to reveal F-actin structures and harvested for analysis of the phosphorylated (activated) form of p38 (pp38) by Western blotting. Cells plated on Ncad-Fc for 2 h spread, formed stress fibers and filopodial extensions, and induced pp38 (Fig. 1 A and B). Furthermore, pp38 in these cells was localized in the nucleus and perinuclear region (Fig. 1C). Cells plated on fibronectin formed stress fibers and longer projections and achieved a similar overall surface area to those on Ncad-Fc, but did not produce pp38 (Fig. 1 A, B, and D). Cells plated on PLL attached but did not spread as well and did not produce pp38. Therefore, Ncad ligation specifically activated p38 in myoblasts. Furthermore, cells plated at low density on Ncad-Fc, but not PLL, for 24 h expressed myogenin and troponin T, early and later markers of differentiation, respectively (Fig. S1).

Fig. 1.

Ncad ligation activates p38 in myoblasts. (A) Photomicrographs of C2C12 cells cultured on Ncad-Fc (Ncad), fibronectin (FN), or poly-L-lysine (PLL) for 2 h and stained with rhodamine phalloidin to reveal F-actin structures. (B) Western blot analysis of pp38 and total p38 in cultures shown in A. (C) Photomicrograph of a C2C12 cell on Ncad-Fc substrate for 2 h stained with rhodamine phalloidin (red), anti-pp38 (green), and DAPI to reveal nuclei (blue). (D) Surface area of C2C12 cells cultured on the indicated substrates for 2 h. Values are means ± SD, n > 80 cells. *, P < 0.01 by Student's t test. (Scale bars, 10 μm.)

Because Ncad and Cdo associate in myoblasts and Cdo-containing complexes are involved in differentiation-dependent p38 activity (6, 7, 11), Cdo and Cdo-binding proteins were assessed for their importance in pp38 production by Ncad ligation. C2C12 cells were depleted of Cdo, JLP, and Bnip-2 by RNAi, and the cells were plated on Ncad-Fc; cells that expressed a control RNAi sequence were also analyzed. Depletion of each protein diminished production of pp38 upon Ncad ligation to an extent that was roughly proportional to the extent of knockdown (Fig. 2 A and B). In contrast, phosphorylation of other MAP kinases, ERK and JNK, induced by Ncad ligation was not significantly affected by RNAi depletion of Cdo, JLP, or Bnip-2, indicating a specific requirement for Cdo and its associated proteins in Ncad-dependent p38 signaling (Fig. 2B and Fig. S2). Cdo also interacts with M-cadherin (11) and, consistent with potential cadherin redundancy in myoblasts, M-caderin ligation activated p38 in a Cdo-dependent manner (Fig. S3). Satellite cell-derived myoblasts from Cdo^+/+ and Cdo^−/− mice were also investigated. Similar to C2C12 cells, Cdo^+/+ myoblasts activated pp38 when plated on Ncad-Fc but not on fibronectin or PLL. In contrast, Cdo^−/− myoblasts displayed only trace production of pp38 upon Ncad ligation (Fig. 2C). Next, we asked whether disruption of Ncad-Cdo association interfered with p38 activation. A Cdo deletion mutant that lacks the first FnIII repeat (designated CdoΔFn1) is deficient in its ability to interact with Ncad but associates normally with other proteins known to interact with the Cdo ectodomain (11–13); expression of this protein in C2C12 cells inhibits differentiation (11). Expression of CdoΔFn1 in these cells also strongly diminished induction of pp38, but not pERK, by Ncad ligation (Fig. 2D).

Fig. 2.

Cdo, JLP, and Bnip-2 are involved in Ncad-induced p38 activation. (A) Western blots of C2C12 cells stably expressing siRNA against Cdo, JLP, or Bnip-2, or a control siRNA (Con), were probed with antibodies against the indicated protein or against β-tubulin as a control. (B) Western blot analysis of pp38, total p38, pERK, and total ERK2 in C2C12 cells expressing the indicated siRNA constructs and cultured for 2 h on Ncad or PLL substrates. (C) Western blot analysis of pp38 and total p38 in Cdo^+/+ and Cdo^−/− myoblasts cultured for 2 h on Ncad, FN, or PLL substrates. (D) Western blot analysis of endogenous Cdo and exogenous Cdo(ΔFn1), pp38, total p38, pERK, and total ERK2, in control or Cdo(ΔFn1)-expressing C2C12 cells cultured on Ncad or PLL substrates for 2 h. Note that although the Con and Cdo(ΔFn1) blots appear split, they are from the same autoradiograms.

Cdo-dependent p38 activation in differentiating myoblasts occurs through Cdo/Bnip-2-dependent Cdc42 activity (6). C2C12 cells plated on Ncad-Fc substrate induced Cdc42 activity significantly above that observed in cells on PLL, as assessed by levels of GTP-bound Cdc42 (Fig. 3A). Furthermore, RNAi-mediated depletion of Cdo strongly decreased levels of GTP-bound Cdc42 in cells plated on either substrate, suggesting that Cdo is not only required for Ncad-initiated Cdc42 activity but also involved in basal Cdc42 activity on PLL (Fig. 3A). We previously showed that stable overexpression of Cdc42GAP in C2C12 cells reduces steady-state levels of GTP-bound Cdc42, differentiation-associated p38 activation, and differentiation itself (6). Overexpression of Cdc42GAP also reduced induction of pp38 by Ncad ligation (Fig. 3B). Taking the results together, Ncad ligation induced pp38 in a short-term signaling assay in a manner very similar to the activation of pp38 seen in differentiating myoblasts (i.e., each is sensitive to disruption of Ncad-Cdo interaction, depletion of Cdo and its intracellular binding proteins JLP and Bnip-2, and reduction of Cdc42 activity) (6, 7, 10).

Fig. 3.

Cdc42, but not RhoA, is involved in Ncad-induced p38 activation. (A) Quantification by G-LISA of GTP-bound (activated) Cdc42 in C2C12 cells stably expressing siRNA against Cdo or a control siRNA plated on Ncad or PLL substrates. *, P < 0.01; **, P < 0.001 by Student's t test. See Methods for details. (B) Western blot analysis of pp38, total p38, and flag epitope in C2C12 cells stably transfected with control (Con) or Cdc42GAP(flag) expression vectors. (C) Photomicrographs of C2C12 cells cultured with or without cell-permeable C3 RhoA inhibitor and plated on Ncad-Fc substrate for 2 h and then stained with rhodamine phalloidin (red), anti-pp38 (green), and DAPI to reveal nuclei (blue). (Scale bar, 10 μm.) (D) Western blot analysis of pp38 and total p38 in C2C12 cells cultured with or without cell-permeable C3 RhoA inhibitor and plated on Ncad-Fc or PLL substrates for 2 h.

The small GTPase RhoA positively regulates myogenic differentiation (15), and stable expression of constitutively active or dominant-negative forms of RhoA enhances or reduces pp38 levels, respectively, in C2C12 cells cultured for 24–72 h in differentiation medium (20). Furthermore, RhoA is activated by cadherin ligation in several cell systems, including myoblasts (15, 26). To assess whether RhoA is directly involved in Ncad ligation-induced pp38, C2C12 cells were treated with the specific Rho inhibitor, C3 transferase. C3 decreased RhoA-GTP levels in C2C12 cells by >70% (Fig. S4), and C3-treated cells on Ncad-Fc substrate displayed a virtually complete loss of stress fibers, production of which is known to require RhoA activity (28) (Fig. 3C). In contrast, Ncad-mediated induction of pp38 was unaffected by C3, nor was the nuclear/perinuclear localization of pp38 (Fig. 3 C and D). Therefore, RhoA is not directly involved in Ncad-dependent p38 activation. The ability of RhoA to positively regulate p38 activity over a longer time course in differentiating myoblasts (19) may be related to its effects on MyoD expression and a consequent effect on MyoD’s linkage to p38 activity through feed-forward mechanisms (4, 29).

JLP, Bnip-2, and Active Cdc42 Cluster at Sites of Ncad-Ligation and Ncad-Cdo Interaction but Not at Sites of Shh-Cdo Interaction.

The results described above strongly suggest that the Cdo-dependent signaling complexes that activate p38 in differentiating myoblasts lie in a pathway initiated by Ncad ligation. High cell density promotes both Ncad ligation and p38 activity in C2C12 cells (15, 20). Ncad and Cdo associate in both high- and low-density cultures (i.e., Ncad ligation is not necessary for Ncad-Cdo interaction) (11), but Cdo and p38 coimmunoprecipitate only in high-density cultures (Fig. 4A), suggesting that Ncad ligation triggers this association. It would therefore be predicted that components of the Cdo complex would cluster at sites of active Ncad ligation. To test this notion, Ncad ectodomain-coated microspheres were allowed to settle onto adherent NIH 3T3 cells that transiently expressed fluorescently tagged forms of Ncad, Cdo, JLP, Bnip-2, or the Cdc42-binding domain of N-Wasp (wGBD), which interacts specifically with active, GTP-bound Cdc42 (30). Cadherin-coated beads attach to cells via cognate cellular cadherins, and additional cellular proteins that cluster at these sites of adhesion can be visualized as a fluorescent signal surrounding the bead (31–33). When Ncad-coated beads attached to cells that coexpressed DsRed-tagged Ncad and Cdo-GFP, each fluorescent protein was concentrated at such beads. In contrast, GFP itself clustered at Ncad beads much less efficiently (Fig. 4 B and C). The percentage of beads that clustered a given fluorescent protein in these assays was quantified. Eighty-five percent of Ncad beads clustered Ncad-DsRed, and 77% clustered Cdo, whether or not exogenous Ncad was expressed (exogenous Ncad is unnecessary as NIH 3T3 cells express abundant endogenous Ncad) (34), whereas only 20% of Ncad beads clustered GFP (Fig. 4C). Furthermore, PLL-coated beads also attached to NIH 3T3 cells but only 11% of these beads clustered any fluorescent protein, regardless of its identity (Fig. 4C).

Fig. 4.

Ncad ligation clusters Cdo, JLP, Bnip-2, and wGBD. (A) C2C12 cells were cultured at high or low cell density as indicated. Lysates were immunoprecipitated with antibodies to Cdo and immunoprecipitates and straight lysates blotted with antibodies to Cdo, p38, and Ncad. (B) (Upper) Photomicrograph of Ncad-coated beads attached to a cell that coexpresses Ncad-DsRed and Cdo-GFP. Arrow, a bead scored positive for clustering Ncad and Cdo; arrowhead, a bead scored negative for clustering Ncad and Cdo. (Lower) Photomicrograph of an Ncad-coated bead attached to a cell that expresses GFP. (C) Percentage of Ncad-coated beads that clustered the indicated fluorescent protein. Values are means ± SD, n > 100 beads. *, P < 0.001 by Student's t test. (D) Photomicrographs of Ncad-coated beads attached to cells that coexpress nonfluorescent Cdo plus JLP-GFP, Bnip-2-GFP, or wGBD-GFP. Arrow, a bead scored positive for clustering Bnip-2-GFP; arrowhead, a bead scored as negative. (E) Percentage of Ncad-coated beads that clustered the indicated fluorescent protein. Values are means ± SD, n > 100 beads. *, P < 0.01; **, P < 0.001 with differences referring to both the respective −Cdo control and the +Cdo/+GFP control. (Scale bars, 10 μm.)

NIH 3T3 cells express low levels of Cdo (35), and clustering of JLP-GFP, Bnip-2-GFP, and wGBD-GFP at Ncad beads was largely dependent on coexpression of nonfluorescent Cdo [Fig. 4 D and E; note that the exogenous expression levels of Cdo-GFP and nonfluorescent Cdo were similar (Fig. S5)]. Seventy-six percent, 59%, and 43% of these beads were positive for JLP-GFP, Bnip-2-GFP, and wGBD-GFP clustering, respectively, all significantly above the 20% seen with GFP. The percentage of beads positive for clustering Ncad-DsRed, Cdo-GFP, JLP-GFP, Bnip-2-GFP, and wGBD-GFP progressively diminished in a manner consistent with the model in which cellular Ncad binds directly to Cdo, Cdo binds directly to JLP and Bnip2, and wGBD binds to Bnip-2 indirectly via Cdc42 (6, 7, 11). The diminishing percentages would be expected if the efficiency of each interaction were <100% and binding was specific (i.e., only beads that clustered Ncad were able to cluster Cdo, and only beads that clustered Cdo were able to cluster JLP and Bnip-2). When the percentage of “available” beads for a given interaction is considered, efficiency at each step ranged from 77 to 99%.

Cdo binds directly to Shh, and the ability of beads coated with an amino-terminal Shh signaling fragment (ShhN)-Fc fusion protein to cluster Cdo and Cdo-binding proteins was also assessed. Fifty-six percent of ShhN beads clustered Cdo, less than seen with Ncad beads but significantly above the GFP background of 20% (Fig. 5 A and B). In contrast, the percentage of ShhN beads that clustered JLP-GFP or Bnip-2-GFP (even in the presence of nonfluorescent Cdo) was no different from that seen with GFP alone. Therefore, Ncad beads clustered Cdo and its associated proteins that are involved in activation of p38, but Shh beads clustered only Cdo. Shh is reported to activate p38 signaling in primary astrocyte cultures (36). However, consistent with its inability to cluster Cdo-binding proteins, pp38 was not induced in C2C12 myoblasts by recombinant ShhN, whereas it was induced in response to a known pathway activator, hyperosmotic stress (0.7 M NaCl) (Fig. 5C). Despite the failure to induce pp38, the cells were responsive to ShhN, as demonstrated by activation of Gli1 expression (Fig. 5D).

Fig. 5.

Shh-coated beads cluster Cdo but not JLP or Bnip-2. (A) Photomicrographs of Shh-coated beads attached to cells that express Cdo or coexpress nonfluorescent Cdo plus JLP-GFP or Bnip-2-GFP. (Scale bar, 10 μm.) (B) Percentage of Shh-coated beads that clustered the indicated fluorescent protein. Values are means ± SD, n > 100 beads. *, P < 0.001. (C) Western blot analysis of pp38 and total p38 in C2C12 cultures treated with recombinant ShhN (0.5 μg/mL) for the indicated amounts of time. As a positive control cells were treated with NaCl (0.7 M) for 20 min. (D) qRT-PCR analysis of Gli1 expression in C2C12 cells treated ± ShhN (0.5 μg/mL) for 24 h. (E) Model for Ncad-stimulated p38 activation in myoblasts and for Cdo as a multifunctional coreceptor. Cdo bound in cis to ligated Ncad exists in a state that permits stable interaction with JLP/p38 and Bnip-2/Cdc42. In contrast, Cdo bound to Shh does not interact stably with these factors, although it permits the Shh signal to be transmitted to Ptch1, activating canonical Hedgehog pathway signaling. Note that activated MKK6 rescues the defective differentiation program caused by loss of Cdo or Bnip-2, but the role of MKK3/6 or other p38 activators has not been established, so a question mark accompanies their position. Note also that, Shh binds both Cdo and Ptch1, but the ternary complex shown is hypothetical.

Collectively, these results link cadherin-based cell–cell adhesion to a defined signaling pathway (i.e., Cdo → p38) that directly regulates the activity of a cell-type-specific differentiation program and suggest the following model: When Cdo is associated with ligated cadherins, its intracellular region undergoes a change in conformation and/or posttranslational modification that permits its stable association with Bnip-2 and JLP and, consequently, activation of p38. In contrast, such changes do not occur in Cdo when it binds Shh (Fig. 5E). ShhN-coated beads were somewhat less effective at clustering Cdo than Ncad-coated beads, even though Cdo was presumably bound directly to the ShhN bead and only indirectly to the Ncad bead (via cis interaction with endogenous Ncad, which bound the bead directly). The ability of cadherins to cluster into large adhesive complexes (37) may increase the avidity and/or stability of the association of Cdo and its binding partners at these sites of adhesion.

The results also allow a distinction to be drawn between Cdo’s actions as a putative coreceptor for cadherin and for Hedgehog signaling pathways. In the Hedgehog pathway, Cdo appears to sensitize cells to a given level of ligand, with signaling occurring via the canonical, Smoothened-dependent pathway, and no requirement for the Cdo intracellular region (13, 35). However, in Ncad-initiated signaling, Cdo confers to the cadherin a signaling capability (activation of the p38 pathway) that it does not possess intrinsically and that depends on the Cdo intracellular region for activity. Therefore, Cdo plays mechanistically distinct roles in modulating the signaling output of two different pathways: cytoplasmic signaling in the Ncad pathway and ligand binding in the Shh pathway. Furthermore, the notion of a shared coreceptor raises the possibility for cross-regulation between these pathways, as may occur in development of the cerebral cortex (38). This concept may also extend to other signaling pathways that share membrane-associated components.

Methods

Cell Culture and Production of Chimeric Proteins.

C2C12, NIH 3T3, and 293T cells were cultured as previously described (12, 39). Myoblasts derived from hindlimbs of Cdo^+/+ and Cdo^−/− mice were obtained by the method of Rando and Blau (40) and cultured as described (7, 8). 293T cells were transfected with Ncad-Fc expression vector (a gift of R. M. Mege) and FuGene6 (Roche) and 36 h later, cultures were passaged into medium containing 0.5 mg/mL hygromycin (Invitrogen). Drug-resistant cells were cultured in serum-free AIMV medium (Invitrogen), with conditioned medium collected every 3 days and assayed for Ncad-Fc by Western blot analysis. Shh-Fc (a gift of P. T. Chuang) was prepared similarly, but the cells were not selected for drug resistance. For RNAi experiments with C2C12 cells, pSuper vectors containing previously validated siRNA sequences to Cdo, JLP, or Bnip-2 (6, 7, 35), or an irrelevant sequence as a control, were transfected with FuGene6 and cultures selected with 5 μg/mL puromycin. Drug-resistant cells were pooled and analyzed. For osmotic shock, C2C12 cells were treated with 0.7 M NaCl for 20 min. Recombinant Shh (R&D Systems) was used at 0.5 μg/mL

Preparation of Adhesion Substrates and Microspheres.

Adhesion substrates and microspheres were prepared largely as described in refs. 32 and 41. Full details are described in SI Text.

Analysis of Cdc42 and RhoA and Western Blot Analysis.

Cdc42 and RhoA G-LISA kits (Cytoskeleton) were used according to the manufacturer’s instructions and as described in detail in SI Text. Cell permeable C3 Transferase (Cytoskeleton) was added to C2C12 cells cultured in DMEM + 15% FBS at a final concentration of 5 μg/mL for 2 h, and then cells were dissociated and inoculated on adhesion substrates with the same medium containing C3 for another 2 h. As a control, the cells were treated with the same volume of PBS carrier. Western blot analyses were performed as described previously by Kang et al. (39). To remove protein A and Ncad-Fc from lysates of cells spread on Ncad-coated plates, magnetic goat anti-mouse IgG beads (Pierce) were added to the lysates and incubated for 1 h with rotation, and beads were removed magnetically before sample buffer was added. Antibodies used were anti-Cdo (Zymed); anti-p38α/β and anti-troponin T (Sigma); anti-pp38α/β, anti-ERK2, anti-pERK, anti-JNK, and anti-pJNK (Cell Signaling Technology); anti-Bnip-2 (gift of B. C. Low); anti-JLP (Abcam); and anti-myogenin and anti-β-tubulin (Santa Cruz).

Immunofluorescence and Microscopy.

C2C12 cells were fixed in PBS with 3% formaldehyde at 37 °C for 15 min, washed with PBS, quenched with NH₄Cl for 10 min at room temperature, washed again with PBS, and permeabilized with 0.1% Triton X-100 for 10 min at room temperature. Coverslips were then blocked with 5% goat serum (Jackson ImmunoResearch) in 3% BSA/PBS for 1 h and were incubated with anti-rabbit pp38 antibody (Cell Signaling Technology, 1:100 dilution) at 4 °C overnight. Primary antibody was revealed with either Alexa Fluor 546- or Alexa Fluor 488-conjugated goat anti-rabbit secondary antibody (Invitrogen). Cells were stained for F-actin using Rhodamine–Phalloidin (Invitrogen) and nuclei were stained with DAPI (Molecular Probes). Cells were mounted in anti-fade reagent (Invitrogen) and observed by confocal microscopy on a Zeiss LSM-510 Meta and analyzed with ImageJ software in the Mount Sinai Microscopy Shared Research Facility.

RNA Preparation and Real-Time RT-PCR.

Analysis of Gli1 mRNA levels by RT-PCR is as described in ref. 35; full details are described in SI Text.