TGFβ regulates epithelial-mesenchymal interactions through WNT signaling activity to control muscle development in the soft palate

The soft palate plays important roles in swallowing, speech, hearing, middle-ear ventilation and respiration (Back et al., 2004; Evans et al., 2010; Marsh, 1991). Despite the important physiological functions of the soft palate, we know very little about the mechanisms that regulate mammalian soft palate development (Bush and Jiang, 2012; Chai and Maxson, 2006). A failure of soft palate development results in a cleft soft palate and a risk of speech problems, middle ear disease and swallowing difficulties (Carvajal Monroy et al., 2012; Precious and Delaire, 1993).

The fleshy soft palate, which makes up approximately the posterior one-third of the secondary palate, is composed of muscles, whereas the anterior two-thirds consists of the bony hard palate (Hilliard et al., 2005; Iwata et al., 2011). Clinically, individuals with a cleft in the soft palate exhibit mis-orientation and delayed development of the soft palate muscles (Cohen et al., 1994). Even after surgical correction of the cleft soft palate and reconstruction of the muscle sling, there are often functional defects in the muscles (Flores et al., 2010; van der Sloot, 2003). Therefore, patients with a defect in the soft palate need long-term therapy and additional surgical corrections after the initial surgical repair for optimal function and normal speech development (Dworkin et al., 2004; Perry, 2011).

There are five muscles in the soft palate: the tensor veli palatini (TVP), the levator veli palatini (LVP), the palatoglossus, the palatopharyngeus and the musculus uvulae (Evans et al., 2010). Their formation, like that of limb and other craniofacial muscles, is controlled by regulatory processes distinct from those governing axial muscle formation (Grenier et al., 2009). The anatomical configuration of the soft palate muscles supports its unique physiological function. Surgical repair of cleft soft palate focuses on the LVP, which is the largest muscle in the soft palate and functions in its elevation (Evans et al., 2010). The LVP may become severely atrophic because of reduced function, and often is only half as thick in cleft soft palate patients as in healthy newborns. The other muscles in the soft palate have not been well studied.

Non-cell-autonomous activation of myogenesis in different regions of the embryo is controlled by a series of complex growth factor networks and niches that ultimately result in the expression of myogenic regulatory factors within nascent and differentiating myoblasts (Snider and Tapscott, 2003). Several epithelial-specific conditional knockout mice, such as Shh^c/n;K14-Cre and Catnb^fl/fl;K14-Cre, exhibit complete cleft palate and a cell proliferation defect in palatal mesenchymal cells, suggesting that oral epithelium-derived factors probably contribute to the cell fate determination of mesenchymal cells through tissue-tissue interactions during soft palate development (He et al., 2011; Lan and Jiang, 2009). In this study, we investigated how loss of transforming growth factor, beta receptor II (Tgfbr2) in the epithelium results in a cleft in the soft palate.

Transforming growth factor, beta (TGFβ; TGFβ1 - Mouse Genome Informatics) signaling plays a crucial role in craniofacial development. TGFβ transmits signals through TGFβ type I and type II receptors that phosphorylate SMAD2 and SMAD3, followed by the formation of transcriptional complexes with SMAD4 and translocation into the nucleus (Massagué and Chen, 2000). In addition, SMAD-independent pathways transduce TGFβ signals in some physiological and pathological conditions (Derynck and Zhang, 2003; Iwata et al., 2012; Xu et al., 2008). TGFβ ligands target a variety of genes in a developmental stage-dependent and cell-type-specific manner (Chai and Maxson, 2006; Iwata et al., 2011). Mice with loss of Tgfbr2 in epithelial cells (Tgfbr2^fl/fl;K14-Cre) exhibit both submucous cleft palate and cleft soft palate. Our previous studies showed that TGFβ signaling regulates gene expression of Irf6 and the fate of the medial edge epithelium (MEE) during palatal fusion in mice (Xu et al., 2006). Interestingly, overexpression of Irf6 fails to rescue cleft soft palate in Tgfbr2^fl/fl;K14-Cre mice although it rescues submucous cleft palate and eliminates MEE persistence, suggesting that TGFβ may regulate the disappearance of the MEE through a different signaling network than the one it uses to control soft palate muscle development (Iwata et al., 2013a). However, it remains unknown how loss of TGFβ signaling in epithelial cells causes cleft soft palate and how mesoderm-derived muscle cells are affected.

In this study, we investigated the genetic basis of muscle formation and the cellular mechanism that controls muscle development in the soft palate during embryogenesis. We show that epithelial-specific loss of Tgfbr2 results in cleft soft palate, compromised WNT signaling due to upregulated Dickkopf-related proteins 1 and 4 (DKK1 and DKK4), and muscle deformities in the soft palate of Tgfbr2^fl/fl;K14-Cre mice. Treatment with neutralizing antibodies for DKK1 and DKK4 rescued the mesenchymal cell proliferation and muscle development defects in an ex vivo organ culture system. Thus, our data indicate that impaired epithelial-mesenchymal communication contributes to the pathology of cleft soft palate.

Loss of TGFβ signaling in the epithelium of Tgfbr2^fl/fl;K14-Cre mice resulted in cleft soft palate with a phenotype penetrance of 100% (Fig. 1A,B). We performed three-dimensional (3D) microCT imaging analysis to examine muscle formation in the soft palate and found that the TVP and LVP muscles were reduced in volume in newborn Tgfbr2^fl/fl;K14-Cre mice compared with wild-type littermate controls (Fig. 1C-J; supplementary material Fig. S1). In addition, we generated 3D histological images of the soft palate of Tgfbr2^fl/fl;K14-Cre and wild-type control mice (Fig. 1K-P). The TVP and LVP muscles in Tgfbr2^fl/fl;K14-Cre mice were smaller than those in wild-type controls. We concluded from these results that the development of the TVP and LVP are compromised in newborn Tgfbr2^fl/fl;K14-Cre mice as the result of loss of TGFβ signaling in the palatal epithelium.

Next, we investigated the time course of muscle development in the soft palate from embryonic day (E)14.5 to E18.5 (Fig. 2A-H). Although we first detected myofibers in the soft palate of both Tgfbr2^fl/fl;K14-Cre and wild-type control mice at E15.5, the total volume of the muscle in the soft palate was reduced in Tgfbr2^fl/fl;K14-Cre mice compared with wild-type control mice. At E15.5 and later embryonic stages, Tgfbr2^fl/fl;K14-Cre mice exhibited a cleft in the soft palate, consistent with the reduction of muscle volume in the soft palate (Fig. 2I-R; supplementary material Fig. S2). In addition, the muscle fibers were aligned in the anterior-posterior direction in the soft palate of Tgfbr2^fl/fl;K14-Cre mice, in contrast to the lateral-medial alignment in controls. Moreover, the muscles in Tgfbr2^fl/fl;K14-Cre mice were attached to the posterior border of the hard palate, similar to human patients with a cleft in the soft palate. We also found that muscle fibers in Tgfbr2^fl/fl;K14-Cre mice were thinner than in control mice, suggesting that Tgfbr2^fl/fl;K14-Cre mice may have a defect in the maturation of myotubes into myofibers (Fig. 2S-V). Collectively, these findings indicate that Tgfbr2^fl/fl;K14-Cre mice can serve as an excellent model to investigate the mechanism that regulates soft palate development and associated malformations.

To analyze the mechanism that causes cleft soft palate in Tgfbr2^fl/fl;K14-Cre mice, we investigated cell proliferation and apoptosis activities. We found that cell proliferation was reduced in Tgfbr2^fl/fl;K14-Cre mice at E14.5 and E15.5 (Fig. 3) but apoptosis was unaffected (supplementary material Fig. S3). Next, we investigated whether muscle differentiation and maturation were altered in the soft palate of Tgfbr2^fl/fl;K14-Cre mice. We found that myofibers were thin and disorganized, appearing wavy and lacking striation (Fig. 4A,B). The diameter of myofibers was decreased in Tgfbr2^fl/fl;K14-Cre mice (Fig. 4C). In addition, the percentage of centrally placed nuclei in myofibers was significantly increased (P<0.05) in Tgfbr2^fl/fl;K14-Cre mice compared with wild-type control mice (Fig. 4D-H). Taken together, our data indicate that failure of muscle cell proliferation and differentiation probably leads to muscle defects and cleft soft palate in Tgfbr2^fl/fl;K14-Cre mice.

To explore the molecular mechanism of cleft soft palate formation in Tgfbr2^fl/fl;K14-Cre mice, we performed global gene expression analysis using soft palate tissue from Tgfbr2^fl/fl;K14-Cre and wild-type control mice at E15.5 (n=6 per genotype) and examined the downstream consequences of dysfunctional TGFβ signaling in epithelial cells. In this comparison, we uncovered 291 probe sets representing transcripts that were differentially expressed [≥1.5-fold, <5% false discovery rate (FDR)], 148 of them more abundant in Tgfbr2^fl/fl;K14-Cre mice (supplementary material Table S1) and 143 more abundant in wild-type control mice (supplementary material Table S2; see http://face.usc.edu//category/microarrays/). The genes identified were consistent with muscle defects, with significant reductions in the levels of transcripts related to the control of muscle movement, such as motoneurons, muscular metabolism and channel molecules for ion exchange (supplementary material Table S2). In order to examine how epithelial-specific deletion of Tgfbr2 results in defects in muscle formation, we focused on the expression of genes related to growth differentiation factors as candidates for regulating tissue-tissue interactions. We found that transcript levels of Dkk1 and Dkk4, which are negative regulators of WNT-β-catenin signaling, were significantly increased (Dkk1: 2.4-fold change, FDR=0.0448; Dkk4: 5.6-fold change, FDR=0.0103) in the soft palate of Tgfbr2^fl/fl;K14-Cre mice (supplementary material Table S1). We confirmed the increased gene expression levels of Dkk1 and Dkk4 in E15.5 Tgfbr2^fl/fl;K14-Cre mice by quantitative RT-PCR analysis (Fig. 5A). In order to investigate gene expression patterns of Dkk1 and Dkk4, we performed whole-mount in situ hybridization for Dkk1 and Dkk4. We detected ectopic expression in the posterior palate of E14.0 Tgfbr2^fl/fl;K14-Cre mice (Fig. 5B-E). We also performed section in situ hybridization of Dkk1 and Dkk4 in E15.5 Tgfbr2^fl/fl;K14-Cre and wild-type control mice (Fig. 5F-I). Dkk1 and Dkk4 gene expression was restricted to the epithelial layers and was not detectable in the palatal mesenchyme of Tgfbr2^fl/fl;K14-Cre mice. Next, we performed immunohistological analysis for DKK1 (Fig. 5J-Q), and found that DKK1 expression was increased in the mesenchyme of Tgfbr2^fl/fl;K14-Cre mice compared with wild-type control mice. DKK1 expression was detectable in the MEE in both wild-type control and Tgfbr2^fl/fl;K14-Cre mice at E14.5. At E15.5, the MEE started to dissociate and DKK1 expression decreased in wild-type control mice. By contrast, DKK1 expression persisted in Tgfbr2 mutant MEE cells, and its expression was increased in the palatal mesenchyme (Fig. 5J-Q). To confirm the diffusion of DKK1 protein to the mesenchyme, we isolated palatal mesenchyme from the soft palate of E15.5 control and Tgfbr2^fl/fl;K14-Cre mice and performed immunoblotting analysis for DKK1 (Fig. 5R). We found that DKK1 protein was undetectable in the palatal mesenchyme from wild-type control mice. By contrast, we detected DKK1 expression in the palatal mesenchyme of Tgfbr2^fl/fl;K14-Cre mice. Next, to investigate whether overexpression of DKK1 affected WNT activity in Tgfbr2^fl/fl;K14-Cre mice, we performed immunoblotting analysis using samples from the entire E15.5 soft palate (Fig. 6A,B). In Tgfbr2^fl/fl;K14-Cre mice, DKK1 expression level was significantly upregulated. Moreover, we detected a decrease in the expression of the active form of β-catenin (ABC) and an increase in phosphorylated β-catenin in Tgfbr2^fl/fl;K14-Cre mice compared with wild-type control mice. Taken together, our data indicate that upregulated DKK1 in the palatal epithelium resulted in compromised WNT activity in the palatal mesenchyme during soft palate development.

We hypothesized that inhibition of WNT signaling activity might cause defects in cell proliferation and differentiation of muscle cells, because previous studies have shown that WNT signaling is important in muscle development (Snider and Tapscot, 2003). To test this hypothesis, we cultured C2C12 cells, a mouse myoblast cell line, in proliferation or differentiation medium with or without an inhibitor for canonical WNT-β-catenin signaling (endo-IWR1). We found that endo-IWR1 suppressed cell proliferation activity in C2C12 cells (Fig. 7A). Similarly, endo-IWR1 inhibited cell proliferation activity in primary cranial neural crest (CNC)-derived MEPM cells from wild-type mice (Fig. 7B). Next, in order to test whether WNT-β-catenin signaling is crucial for muscle differentiation, we cultured C2C12 cells in differentiation medium for 0, 3 or 5 days. We found that endo-IWR1 inhibited muscle differentiation in C2C12 cells (Fig. 7C-H). Moreover, we cultured wild-type palatal explants with endo-IWR1 for 1 or 3 days (Fig. 8A-J). Strikingly, wild-type palates treated with endo-IWR1 in an ex vivo organ culture system (Fig. 8B,E,H) mimicked the cell proliferation and differentiation defects in Tgfbr2^fl/fl;K14-Cre palates (Fig. 8C,F,I), whereas these activities were not affected after treatment with vehicle (Fig. 8A,D,G). Thus, our data indicate that WNT signaling is crucial for soft palate muscle development.

We hypothesized that reduction of DKK1 and DKK4 would rescue the defective cell proliferation and muscle differentiation activities in Tgfbr2^fl/fl;K14-Cre mice. To test this hypothesis, we treated palatal explants from Tgfbr2^fl/fl;K14-Cre mice with neutralizing antibodies (NAb) for DKK1 and DKK4 (Fig. 9A-J). We found that treatment with NAb for DKK1 and DKK4 could rescue the cell proliferation defect in the mesenchyme of Tgfbr2^fl/fl;K14-Cre palates (Fig. 9A-C,J). Similarly, treatment with NAb for DKK1 and DKK4 could rescue both muscle cell proliferation and differentiation defects in Tgfbr2^fl/fl;K14-Cre palates in an ex vivo organ culture system (Fig. 9D-J). Collectively, our findings indicate that inhibition of WNT signaling through elevated DKK1 and DKK4 expression is responsible for the failure of soft palate development and the muscle defects in Tgfbr2^fl/fl;K14-Cre mice (Fig. 9K).

Our mouse model for cleft soft palate, Tgfbr2^fl/fl;K14-Cre, suggests that specific regulators, such as growth factors, from epithelial cells may control the fate and orientation of muscle cells during palatogenesis. Importantly, this mutant mouse model mimics the phenotype of human individuals with cleft soft palate. Using the mouse model, we explored how loss of TGFβ signaling in epithelial cells results in a failure of muscle development and found that TGFβ-controlled WNT-β-catenin signaling activity is crucial for muscle formation. Previous studies indicated that DKK1 and DKK4 can inhibit WNT-β-catenin signaling (Kawano and Kypta, 2003; Nie et al., 2005; Semënov et al., 2001). Dkk1 knockout mice exhibit severe craniofacial deformities and embryonic lethality before palatal development (Mukhopadhyay et al., 2001). Mice lacking β-catenin in CNC-derived cells (Catnb^fl/fl;Wnt1-Cre) exhibit severe craniofacial deformities (Brault et al., 2001), and Wnt9b knockout mice exhibit cleft palate (Jin et al., 2012). Thus, WNT signaling is involved in craniofacial development. Loss of WNT-β-catenin signaling results in downregulation of TGFβ, whereas expression of stabilized β-catenin in the palatal epithelium causes ectopic Tgfb3 expression and fusion of the palatal shelf and mandible (He et al., 2011). Our study clearly shows that TGFβ signaling also affects WNT signaling activity through DKK1 and DKK4. This feedback between TGFβ and WNT signaling may be crucial for maintaining normal growth factor signaling during craniofacial development.

Tgfbr2^fl/fl;K14-Cre mice exhibit both persistence of the MEE in the hard palate (submucous cleft palate) and a cleft in the soft palate (cleft soft palate). Our previous studies showed that TGFβ signaling regulates Irf6 gene expression and the fate of the MEE, and overexpression of Irf6 rescues MEE disappearance although it fails to rescue the cleft in the soft palate in Tgfbr2^fl/fl;K14-Cre mice, indicating that the molecular mechanism of soft palate development is different from that of MEE fate determination (supplementary material Fig. S4) (Iwata et al., 2013a; Xu et al., 2006). Irf6 is only expressed in palatal epithelial cells (Iwata et al., 2013a; Thomason et al., 2010). Hence, TGFβ regulated IRF6 signaling must only control the fate of MEE cells, and other TGFβ downstream targets must control other aspects of palate development. This reasoning is supported by our finding that Dkk1 and Dkk4 expression is specifically upregulated in the posterior palate of Tgfbr2^fl/fl;K14-Cre mice, resulting in compromised WNT-β-catenin signaling activity in the mesenchyme and soft palate muscle defects. These results suggest that TGFβ signaling plays a crucial role in regulating tissue-tissue interactions during palatogenesis. In order to investigate this further, we compared gene expression profiles of the anterior and posterior palate in E15.5 wild-type mice and found that a total of 887 transcripts were differentially expressed in the soft palate. We focused on the TGFβ signaling pathway and found that TGFβ signaling mediators were differentially expressed (supplementary material Fig. S5), suggesting that TGFβ signaling is differently regulated in the anterior versus posterior palate during palate formation. This putative asymmetry is supported by our previous finding that haploinsufficiency of Tgfbr1/Alk5 in Tgfbr2^fl/fl;Wnt1-Cre mice failed to rescue soft palate development, whereas it rescued clefting of the hard palate (Iwata et al., 2012) (our unpublished data). Our previous studies indicate that SMAD and p38 mitogen-activated protein kinase (MAPK) are functionally redundant in regulating the fate of the MEE cells (Xu et al., 2006; Iwata et al., 2013a). Therefore, gene expression of Dkk1 and Dkk4 may be regulated by both canonical and non-canonical TGFβ signaling. We analyzed the promoter sequences of Dkk1 and Dkk4 and found that the promoter regions of both genes contain SMAD-binding sites and p38 MAPK response elements (data not shown). Taken together, our findings suggest that gene expression of Dkk1 and Dkk4 may be regulated by TGFβ signaling in a tissue- and region-specific manner.

In addition to epithelial signals, the defects in the muscle fibers in the soft palate could also be the result of improper signaling from the cranial neural crest (CNC)-derived palatal mesenchyme. In the craniofacial region, CNC-derived cells give rise to tendons and connective tissues in the skeletal muscles, whereas these tissues are derived from the lateral mesoderm in the rest of body. CNC-specific Tgfbr2 mutant (Tgfbr2^fl/fl;Wnt1-Cre) mice exhibit muscle abnormalities in the tongue (Iwata et al., 2013b) and soft palate (supplementary material Fig. S6), following defects in cell proliferation and maturation or organization. Moreover, in our microarray analyses of palates from E14.5 Tgfbr2^fl/fl;Wnt1-Cre mice (Iwata et al., 2012), we found that expression of genes in the WNT signaling pathway, including Dkk2 and Limd1, was altered (supplementary material Figs S6, S7). Taken together, our results suggest an important positive role for CNC cells in craniofacial muscle development.

We found that elevated WNT inhibitors DKK1 and DKK4 diminished WNT-β-catenin signaling in the soft palate and adversely affected the proliferation and differentiation of myoblasts as well as the maturation of myotubes into myofibers in the palatal mesenchyme. These findings have important implications for clinical studies that aim to identify patients with cleft soft palate and families with a high risk of cleft soft palate. Moreover, this study lays the groundwork for efforts to reorient soft palate muscles and to accelerate muscle regeneration in patients with cleft soft palate. Our study of soft palate development could also be useful for understanding tissue-tissue interaction mechanisms in regulating skeletal muscle development and regeneration.

To generate Tgfbr2^fl/fl;K14-Cre mice, we mated Tgfbr2^fl/+;K14-Cre with Tgfbr2^fl/fl mice. To generate Tgfbr2^fl/fl;K14-Cre;Irf6^Tg mice, we mated Tgfbr2^fl/+;K14-Cre;Irf6^Tg with Tgfbr2^fl/fl mice. To generate Tgfbr2^fl/fl;Wnt1-Cre mice, we mated Tgfbr2^fl/+;Wnt1-Cre with Tgfbr2^fl/fl mice. Genotyping was performed using PCR primers as described previously (Iwata et al., 2012; Iwata et al., 2013a; Xu et al., 2006).

MicroCT analysis was performed using a SCANCO μCT50 device at the University of Southern California Molecular Imaging Center. The microCT images were acquired with the X-ray source at 70 kVp and 114 μA. The data were collected at a resolution of 10 μm. The three-dimensional (3D) reconstruction was done with AVIZO 7.1 software (Visualization Sciences Group).

Total RNA samples (1 μg per sample) were isolated from E15.5 soft palate of control and Tgfbr2^fl/fl;K14-Cre mice, and converted into biotin-labeled cRNA using the GeneChip IVT Labeling Kit with Biotin-ddUTP and standard protocols recommended by Affymetrix (Santa Clara, CA, USA). Fragmented cDNA was applied to GeneChip Mouse Genome 430 2.0 Arrays (Affymetrix) containing probe sets designed to detect over 39,000 transcripts. Microarrays were hybridized, processed and scanned as previously described (Karaman et al., 2003), using the manufacturer’s recommended conditions. WebArray software was used to generate scaled log₂-transformed gene expression values using the RMA algorithm (Wang et al., 2009; Xia et al., 2005). Probes sets showing ≥1.5-fold differential expression with a <5% false discovery rate (FDR) were identified through liner models for microarray data (LIMMA)-based liner model statistical analysis (Smyth, 2004) and FDR calculations made using the spacings LOESS histogram (SPLOSH) method (Pounds and Cheng, 2004). All scaled gene expression scores and .cel files are available at the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) repository under the series accession number GSE46211.

Hematoxylin and Eosin (H&E), BrdU and immunohistochemical staining were performed as described previously (Iwata et al., 2010; Iwata et al., 2012; Iwata et al., 2013a). Antibodies used for immunohistochemistry were mouse monoclonal antibody against myosin heavy chain (MyHC; Sigma-Aldrich) and DKK1 (Santa Cruz Biotechnology). Fluorescence images were obtained using a fluorescence microscope (Model IX71, Olympus).

Total RNA was isolated from mouse soft palate dissected at E15.5 using QIAshredder and RNeasy Micro extraction kits (Qiagen), as described previously (Iwata et al., 2010). The following PCR primers were used: Dkk1, 5′-GAGGGGAAATTGAGGAAAGC-3′ and 5′-GGTGCACACCTGACCTTCTT-3′; Dkk4, 5′-TAGAGTTCGCAGGAGGTGTC-3′ and 5′-TGAGGTCTGTTTTCCTCTGC-3′; Dkk2, 5′-ACCCTTGCAGCAGTGATAAG-3′ and 5′-TGGCTTTGGAAGAGTAGGTG-3′; Limd1, 5′-CGGACTACTTCGGTTCCTGT-3′ and 5′-TTACAGATGACGCAGCGGAA-3′; and Gapdh, 5′-AACTTTGGCATTGTGGAAGG-3′ and 5′-ACACATTGGGGGTAGGAACA-3′. A two-tailed Student’s t-test was applied for statistical analysis of quantitative PCR data. A P-value ≤0.05 was considered statistically significant. For all graphs, data are means ± standard deviation (s.d.).

Immunoblotting was performed as described previously (Iwata et al., 2012; Iwata et al., 2013b). Antibodies used for immunoblotting were as follows: mouse monoclonal antibodies against DKK1 (Santa Cruz Biotechnology), active form of β-catenin (ABC) and GAPDH (Millipore), and rabbit polyclonal antibody against phosphorylated β-catenin (Cell Signaling Technology).

Timed-pregnant mice were sacrificed at E15.5. Genotyping was carried out as described above. The palatal shelves were micro-dissected and cultured in BGJb medium supplemented with penicillin, streptomycin, vitamin C and 10% fetal bovine serum. After 24 or 72 hours in culture with neutralizing antibodies (1 μg/ml each) for DKK1 and DKK4, or WNT inhibitor endo-IWR1 (50 μM; Tocris Bioscience), soft palate explants were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) and processed for histological analysis as previously described (Iwata et al., 2012).

The probes for Dkk1 and Dkk4 were obtained from Dr Irma Thesleff (University of Helsinki, Finland). In situ hybridization was performed as described previously (Iwata et al., 2010). Several negative controls (e.g. sense probe and no probe) were run in parallel with the experimental reaction.

Primary mouse embryonic palatal mesenchymal (MEPM) cells were obtained from E13.5 embryos as described previously (Iwata et al., 2012). C2C12 cells (ATCC) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum supplemented with penicillin, streptomycin, L-glutamate, sodium pyruvate and nonessential amino acids. Cell proliferation assays were performed using a cell counting kit (Dojindo Molecular Technologies). MEPM and C2C12 cells were treated with endo-IWR1 (Tocris Bioscience) at 10 μM for 24 hours or left untreated. Myogenic differentiation was induced by culture in a monolayer in differentiation medium (DMEM supplemented with 2% donor equine serum, 1 μM insulin, penicillin, streptomycin and L-glutamate) for 3 or 5 days. Myofibers were stained with MyHC. All fluorescence images were obtained using a fluorescence microscope (Model IX71, Olympus). Pictures were taken using MicroSuite Analytical Suite software (Olympus).

Two-tailed Student’s t-tests were applied for statistical analysis. For all graphs, data are mean ± s.d. A P-value of less than 0.05 was considered statistically significant.

We thank Drs J. Mayo and B. Samuels for critical reading of the manuscript; Dr H. Moses for Tgfbr2^fl/fl mice; Dr H. Slavkin for discussion; P. Bringas, Jr, O. Chaudhry, the Molecular Imaging Center and the Microarray core of CHLA for technical assistance.