Indirect modulation of Shh signaling by Dlx5 affects the oral-nasal patterning of palate and rescues cleft palate in Msx1-null mice

Cleft palate represents one of the major groups of congenital birth defects in the human population. Despite recent advancements in medical intervention, babies born with cleft palate often suffer multiple handicaps that significantly compromise the quality of their lives. The mammalian palate develops from two primordia: the primary palate and the secondary palate. The primary palate represents only a small part of the adult hard palate. The secondary palate is the primordium for most of the hard and all of the soft parts of the palate. Palate development is a multistep process that involves palatal shelf growth, elevation, midline fusion of palatal shelves and the disappearance of the midline epithelial seam. The palatal structures are composed of the cranial neural crest (CNC)-derived ectomesenchyme and pharyngeal ectoderm (Ferguson, 1988; Shuler, 1995). Throughout palatal development, there is continuous epithelial-mesenchymal interaction that is essential for the growth and fusion of the palate. The most common type of cleft palate documented in animal studies, which also most closely resembles cleft palate in humans, is the failure of palate shelf expansion following elevation (Chai and Maxson, 2006; Ito et al., 2003; Rice et al., 2004; Satokata and Maas, 1994).

In the developing palate, the epithelia that cover the palatal shelves are divided into oral, nasal and medial edge epithelium (Chai and Maxson, 2006). The nasal and oral epithelia differentiate into pseudostratified and squamous epithelia, respectively, whereas the medial edge epithelium (MEE) is removed from the fusion line by means of programmed cell death and cell migration (Martinez-Alvarez et al., 2000; Vaziri Sani et al., 2005). The palatal mesenchyme is mainly derived from CNC cells (Ito et al., 2003) and has been treated as a homogeneous population in previous studies. The oral-nasal patterning of the palatal mesenchyme and the molecular regulation of the fate of mesenchymal cells must be taken into account when analyzing palatal development.

Dlx5, Shh and Msx1 control the fate of CNC cells. Specifically, Dlx5 plays a critical role in regulating the patterning of craniofacial structures (Depew et al., 1999; Qiu et al., 1997; Yang et al., 1998). Nested Dlx gene expression in the branchial arches patterns proximodistal axes and is crucial in the acquisition and refinement of mammalian jaws through evolution (Depew et al., 2002). Sonic hedgehog (Shh) mediates the ventral inductive signaling during the dorsoventral patterning of the spinal cord (Jessell, 2000). Within the CNC population, Shh is required for cardiac outflow tract and facial primordial development via regulation of CNC cell survival and proliferation (Jeong et al., 2004; Washington Smoak et al., 2005). During palatogenesis, Shh expression is restricted to the oral side of the palatal epithelium, and conditional inactivation of Shh in the ectoderm leads to dramatic shortening of the palatal shelves and cleft palate (Lan and Jiang, 2009; Rice et al., 2004). Exogenous Shh stimulates palatal mesenchyme proliferation in palatal explant culture (Bei et al., 2000). Interestingly, a recent study shows that overexpression of Shh signaling in the palatal ectoderm also leads to cleft palate (Cobourne et al., 2009). Collectively, these studies suggest that Shh signaling needs to be tightly regulated during palatogenesis.

Msx1 is crucial for the development of palate, teeth and other craniofacial structures (Han et al., 2003; Satokata and Maas, 1994). In humans, mutations in the MSX1 gene result in orofacial clefting and tooth agenesis, consistent with the phenotype observed in Msx1 mutant mice (Hu et al., 1998; Jumlongras et al., 2001; van den Boogaard et al., 2000; Vastardis et al., 1996). In mice, Msx1 is required for Bmp4 and Bmp2 expression in the palatal mesenchyme and Shh expression in the palatal epithelium. Shh acts downstream of Bmp4 and upstream of Bmp2 to stimulate mesenchymal cell proliferation to promote the outgrowth of the palatal shelf (Zhang et al., 2002).

We have investigated the establishment of O-N patterning in the palate by assaying the expression of various asymmetric gene markers and investigating the palatal phenotype associated with the loss of Dlx5 in mice. We find that oronasal (O-N) patterning is associated with the expansion and fusion of the palatal shelves and that Dlx5 is required in the O-N patterning of palatal mesenchyme. Dlx5 is specifically required for Fgf7 expression in the nasal side of palatal mesenchyme. Furthermore, Fgf7 strongly inhibits Shh expression in the nasal side of palatal shelf epithelium. Loss of Dlx5 results in downregulation of Fgf7 and an expansion of Shh expression into the nasal side of the palatal epithelium. This expanded Shh signaling is sufficient to rescue palatal fusion, as Msx1/Dlx5 double-null mutant mice show restored CNC cell proliferation and palate fusion. Furthermore, Msx1 and Dlx5 antagonistically regulate the expression of Shh, which in turn controls the fate of CNC cells through tissue-tissue interaction during palatogenesis. Finally, we report that Dlx5 is crucial for the patterning of soft palate.

Mice carrying Msx1^+/− and Dlx5^+/− alleles have been described previously (Depew et al., 1999; Satokata and Maas, 1994). We crossed Msx1^+/−; Dlx5^+/− mice to generate Msx1/Dlx5 double-null mutants. All samples were fixed in 10% buffered formalin and processed through serial ethanol, and paraffin embedded and sectioned using routine procedures. For general morphology, deparaffinized sections were stained with Hematoxylin and Eosin (H and E) using standard procedures. Skeletal structures were stained using Alcian Blue for non-mineralized cartilage and Alizarin Red for bone, as described previously (Ito et al., 2003). For SEM, samples were fixed with 10% buffered formalin at 4°C overnight. After dehydration through a graded ethanol series, samples were trimmed and dried in a Balzer Union (FL-9496) apparatus, and coated with colloidal silver liquid (Ted Pella) by a Technics Hummer V Sputter Coater. Samples were examined with a Cambridge 360 scanning electron microscope.

Timed-pregnant mice were killed on post-coital day 13.5 (E13.5). Genotyping was carried out as previously described (Depew et al., 1999; Satokata and Maas, 1994). Paired secondary palatal shelves were microdissected and cultured in serumless, chemically defined medium as previously described (Ito et al., 2003). For bead implantation, Affi-Gel blue agarose beads (BioRad) were soaked in proteins as previously described (Zhang et al., 2002). Tissues were harvested after 24 hours of culture and fixed in 4% paraformaldehyde for processing. Shh N-terminal peptide (R&D Systems) was used at 1 mg/ml, anti-Shh antibody (Developmental Studies Hybridoma Bank) was used at 0.30 mg/ml and BSA was used at 10 ng/ml. Neutralizing antibodies to Fgf7 (MAB251) and Mouse IgG1 (MAB002) (R&D Systems) were added to the culture medium at concentrations of 50 μg/ml.

Following treatment with 20 mg/ml proteinase K for 15 minutes at room temperature, apoptotic cells were assayed by the TUNEL procedure using the In Situ Cell Death Detection Kit, Fluorescein (Roche Molecular Biochemicals) by following the manufacturer's protocol. Cell proliferation was scored by injection of 5-bromo-2′-deoxyuridine (BrdU, Sigma, 100 μg/g body weight) into pregnant females 1 hour before recovery of embryos. For palatal organ culture, BrdU was supplemented into culture medium at concentration of 100 μM for 2 hours before harvest. Detection of BrdU-labeled cells was carried out using a BrdU Labeling and Detection Kit following the manufacturer's protocol (Zymed).

Samples for whole-mount and section in situ hybridization were fixed in freshly made 4% paraformaldehyde/PBS. In situ hybridizations were performed as previously described (Xu et al., 2005). The following cDNAs were used to generate antisense riboprobes: a 1.2 kb fragment of mouse Msx1, a 900 bp mouse Dlx5, a 1.6 kb mouse Gli1 and 640 bp mouse Shh. Non-radioactive RNA probes were generated by in vitro transcription labeling with digoxigenin-UTP according to the manufacturer's protocol (Roche Molecular Biochemicals).

E13.5 palatal shelves (six wild type and six mutants, in two independent experiments) were precisely dissected. Total RNA was extracted using the RNeasy Kit (Qiagen, USA). Quality of RNA, primer efficiency and correct size were tested by RT-PCR. Real-time PCR was performed with Cycler (Bio-Rad) using iQ SYBR Green (Bio-Rad). 18S RNA was used to normalization.

Primer sequences are: 18S, 5′-CGGCTACCACATCCAAGGAA-3′; 18S AS, 5′-GCTGGAATTACCGCGGCT-3′; Fgf7 F, 5′-CTCTACAGGTCATGCTTCCACC-3′; Fgf7 R, 5′-ACAGAACAGTCTTCTCACCCT-3′.

The palatal epithelia is heterogeneous in newborn mice, with pseudostratified, ciliated, columnar epithelia covering the nasal side of the palatal shelf and stratified keratinizing, squamous epithelia covering the oral side (Fig. 1A,B). Mesenchymal heterogeneity of the nasal and oral regions of the palate is subtle, with the palatine bone in the nasal region and soft tissue in the oral region (Fig. 1B). We hypothesized that these morphological differences are likely to be the result of molecular heterogeneity established earlier along the oronasal (O-N) axis of developing palate. In fact, Dlx5, Fgf7 and phospho-Smad1/5/8 expression is restricted to the nasal region at E13.5 (Fig. 1C-F), whereas Gli1 and Fgf10 are expressed mainly in the oral region of the palatal mesenchyme (Fig. 1H-K). Shh is expressed only in the oral side of palatal epithelium at E13.5 (Fig. 1G).

Previous studies examining Dlx5^−/− mice were based on skeletal staining, which revealed incomplete palatal bone fusion (Depew et al., 1999; Levi et al., 2006). In order to characterize the palatal phenotype thoroughly, we performed scanning electron microscopy (SEM) on Dlx5^−/− mice. In the anterior region of the palate, the soft tissue was fused in all Dlx5^−/− mice (n=45) (Fig. 2A-D). Although some Dlx5^−/− mice (n=5, 11%) were born with a groove (a folding of the palatal shelf) in the palate (Fig. 2C), histological analysis clearly revealed that there was normal soft tissue connection between the two palatal shelves (Fig. 2D). The landmark of the oral palatal epithelium, the rugae, are detectable in SEM images of the oral side of newborn palatal shelves (Fig. 2A,B, white arrows). Compared with wild-type littermates, the rugae in Dlx5^−/− mice appear more prominent, both in the SEM image (Fig. 2A,B) and the H and E staining of sagittal sections through the rugae (Fig. 2E,F). We detected approximately a 40% increase in the height of the ruga and 40% increase in the thickness of the squamous epithelium in Dlx5^−/− mutants relative to wild type (Fig. 2F). We did not detect any difference in stratification of the oral side of the palatal epithelium in Dlx5^−/− mutants. By contrast, the thickness of the nasal side palatal epithelium in Dlx5^−/− mice is comparable to that of the wild-type control (see Fig. S1 in the supplementary material). The expansion of the oral side of the palate shelf is most obvious in Dlx5^−/− mutants with excencephaly (representing 11% of Dlx5^−/− mice), in which the oral palatal epithelium protrudes into the palate forming a groove (Fig. 2C). The expansion of the oral side of the palate shelf in Dlx5^−/− mice can be detected at E13.5 based on the expansion of Shh expression into the MEE and nasal side of the palatal epithelium (Fig. 2I-L). This is an expansion, not a shift, as the oral side Shh expression persists (Fig. 2J). In wild-type mice, Shh is not expressed in the MEE and nasal side of the palatal epithelium. We confirmed previous studies that reported diminished Shh expression in the anterior part of the oral palatal epithelium in Msx1^−/− mutant mice (Fig. 2O) (Zhang et al., 2002). We also found that the expression of Gli1, a key hedgehog (Hh) pathway target (Hooper and Scott, 2005; Lum and Beachy, 2004), expanded from the oral region of the palate into the nasal region in Dlx5^−/− mice (Fig. 3A,B). Phospho-Smad1/5/8, and Fgf7 were detectable in the nasal region of the palatal mesenchyme in wild-type mice, but their expression was reduced in Dlx5^−/− mutants (Fig. 3D,E,G,H,J). Our quantitative PCR analysis confirmed greater than 40% reduction in Fgf7 expression in the palate of Dlx5^−/− mice (Fig. 3J). Although the expression of Shh is diminished in the palatal epithelium of Msx1^−/−mice, the expression of Gli1, phospho-Smad1/5/8 and Fgf7 were indistinguishable from the wild-type control (Fig. 3C,F,I).

During palatogenesis, Shh signaling from the oral region of the palatal epithelium is required for palatal mesenchymal proliferation (Rice et al., 2004; Zhang et al., 2002). At E13.5, the expression of Shh in the palatal epithelium is restricted to the epithelial thickening, the developing ruga (Fig. 4A). In sagittal sections, epithelial cells that express Shh are not actively proliferating, whereas the mesenchymal cells underlying the Shh-expressing cells are actively proliferating and show a higher proliferative activity than neighboring regions of mesenchymal cells underlying non-Shh-expressing epithelia (Fig. 4B-D). We hypothesized that expanded Shh signaling in the absence of Dlx5 might rescue the palatal mesenchymal proliferation defect in Msx1^−/− mutant mice, a cleft palate model in which the two palatal shelves fail to meet at the midline following elevation. By comparing cell-proliferation activities in wild type, Msx1^−/− and Msx1^−/−; Dlx5^−/− palatal mesenchyme (Fig. 4E-H), we found that inactivation of Dlx5 significantly stimulated palatal mesenchymal proliferation in the background of Msx1-null mutation. At E14.0, Msx1^−/− mice showed significantly reduced cell proliferation (13.2±2.3%) in the palatal mesenchyme compared with the control (29.4±5.1%) (Fig. 4E,F,H). Significantly, CNC cell proliferation activity was restored (23.6±3.8%) in the palatal mesenchyme of Msx1^−/−; Dlx5^−/− mutants (Fig. 4G,H). We did not detect altered cell proliferation activity in Dlx5-null mutants compared with wild-type littermates (data not shown). There was no difference in apoptotic activity in the palatal mesenchyme of wild type, Msx1^−/− or Msx1^−/−;Dlx5^−/− mutant samples (data not shown). The two palatal shelves were able to reach the midline in Msx1^−/−; Dlx5^−/− mutants at E14.0 (Fig. 4L), probably as a consequence of the rescued palatal mesenchymal proliferation. As expected, the anterior part of the secondary palatal shelf showed insufficient expansion towards the midline in Msx1^−/− mutants (Fig. 4J). The expansion of the palatal shelves in Dlx5^−/− mutants was comparable to wild type (Fig. 4K,I).

To investigate whether restored cell proliferation in the palatal mesenchyme and palatal shelf extension was sufficient to restore proper palatal fusion in the Msx1 mutant, we examined palatogenesis in Msx1^−/−; Dlx5^−/− mice. We found that there was complete rescue of the Msx1^−/− cleft palate defect in Msx1^−/−; Dlx5^−/− mutant mice (n=15) (Fig. 5A-D). This rescued palate development required the complete absence of Dlx5, as Msx1^−/−; Dlx5^+/− palates failed to fuse (data not shown). At E17.5, Msx1^−/− mice showed a complete cleft of the secondary palate, the palatal shelves failed to meet at the midline (Fig. 5B,F) and the palatine processes of the maxilla and of the palatine bones were missing, leaving the vomer visible in an oral view (Fig. 5J). In the Dlx5^−/− sample, fusion of the anterior palate was indistinguishable from that of the wild-type control; the soft tissue covering the anterior part of the palate was fused completely (Fig. 5C,G), and the bony parts of the hard palate were present (Fig. 5K). In Msx1^−/−; Dlx5^−/− mutant mice, palatal fusion was rescued, with confluence of the mesenchyme and reappearance of the palatal bones that were missing in the Msx1^−/− mutant (Fig. 5D,H,L).

Next, we examined whether Msx1 and Dlx5 regulate each other's expression in an upstream or downstream manner by determining if Msx1 expression was altered in Dlx5^−/− mice or vice versa. Msx1 is expressed in the anterior part of the developing palate at E13.5. In the Dlx5^−/− sample, Msx1 expression was comparable to that of wild type (Fig. 5M,N,Q,R). A recent study failed to show that Msx1 and Dlx5 expression overlaps in the anterior palatal mesenchyme (Levi et al., 2006). Our data clearly demonstrate that Dlx5 is expressed in the anterior palatal mesenchyme (Fig. 1C) and was unaffected in the Msx1^−/− sample (Fig. 5O,P,S,T). Thus, Msx1 and Dlx5 do not appear to regulate each other's expression in the palate.

To investigate the stimulation of palatal mesenchymal cell proliferation by Shh signaling during palatogenesis, we treated E13.5 wild-type palatal explants with either BSA or Shh beads. Palatal mesenchyme cell proliferation was enhanced following Shh treatment (Fig. 6A,B insets, Fig. 6C). Moreover, ectopic Shh did not affect apoptosis in the MEE or palatal fusion (Fig. 6A-E). Explants treated with BSA or Shh beads were both able to fuse following 2 days of culture. Thus, increased Shh signaling has a stimulatory role on cell proliferation in the palatal mesenchyme. Furthermore, overexpressing Shh does not appear to have a major impact on the MEE cells.

To further support a role for expanded Shh signaling in the absence of Dlx5 is responsible for the increased palatal mesenchymal cell proliferation and the rescued palatal fusion in Msx1^−/−; Dlx5^−/− mice, we blocked Shh signaling with anti-Shh antibodies. In wild type, Dlx5^−/− and Msx1^−/−; Dlx5^−/− palatal explants, anti-Shh antibody treatment resulted in reduced mesenchymal cell proliferation activity compared with the BSA-treated explants (Fig. 6F-L). Treatment with anti-Shh antibody blocked the rescued cell proliferation in the palatal mesenchyme of Msx1^−/−; Dlx5^−/− mice (Fig. 6H,K,L). Our data are consistent with a role for increased Shh signaling being responsible for the rescue of cell proliferation in the Msx1^−/−; Dlx5^−/− mutant palate.

To confirm the restored mesenchymal proliferation is a result of loss of Fgf7 expression in the nasal side of the palatal shelf, we blocked Fgf7 signaling with neutralizing antibody. In both wild type and Msx1^−/− palatal explants, anti-Fgf7 antibody treatment resulted in increased mesenchymal proliferation activity compared with mouse IgG1-treated explants (Fig. 6Q). Furthermore, in Msx1^−/− palatal explants treated with anti-Fgf7, the mesenchymal proliferation rate was restored to a level comparable to that of wild-type controls without anti-Fgf7 treatment (Fig. 6Q).

The requirement for Shh signaling in the genetic rescue of cleft palate in Msx1^−/−; Dlx5^−/− mice suggests that Shh may be regulated by both Msx1 and Dlx5. This hypothesis predicts that Shh expression should be altered as a result of null mutation of either Msx1 or Dlx5. In wild-type palatal shelves at E13.5, Shh expression was restricted to the oral side of the palatal epithelium and limited to defined stripes that correspond to the developing rugae (Fig. 7A, Fig. 2I, Fig. 4B). In the epithelial cells covering the palatal shelf of Msx1^−/− mice, Shh expression was not detectable (Fig. 7B, Fig. 2O). Loss of Dlx5 resulted in expanded Shh expression in the palatal epithelium (Fig. 7C, Fig. 2J). This medial expansion of Shh signaling suggests that Dlx5 was required to suppress Shh expression in the palate. We also detected a similar expansion of Shh expression in the palatal epithelium of Msx1^−/−; Dlx5^−/− mutant mice (Fig. 7D). Thus, expanded Shh expression is independent of Msx1 signaling in the Dlx5 mutant. Gli1 is a mediator and target for the Shh pathway that enables us to detect Shh-responsive cells. We found that Gli1 was expressed in the oral half of wild-type palatal mesenchyme adjacent to the epithelium where Shh was expressed (Fig. 1G-I, Fig. 3A). In Dlx5^−/− palate, Gli1 expression expanded into the nasal half of the palatal shelf (see Fig. 3B) corresponding to the expansion in Shh expression (Fig. 7C). In Msx1^−/− palate, however, Gli1 expression remained restricted to the oral half of the palate (Fig. 3C). Therefore, we conclude that Msx1 and Dlx5 antagonistically regulate Shh expression in the palatal epithelium; however, the capacity of palatal mesenchyme to respond to Shh signaling is independent of Msx1 status.

The expansion of Shh expression into the nasal side of the palatal epithelium in Dlx5^−/− mice suggests that Dlx5 is required for restricting Shh expression to the oral side of the palatal epithelium. However, Shh is expressed in the oral side of the palatal epithelium (Fig. 1G), whereas Dlx5 is expressed in the nasal side of the palatal mesenchyme (Fig. 1C). This spatial relationship and the fact that Dlx5 functions as a transcription factor suggest that a Dlx5 downstream target gene in the palatal mesenchyme mediates the restriction of Shh expression. Endogenous Fgf7 is expressed in the nasal half of the palatal mesenchyme (Dlx5-expressing domain), whereas Shh is absent from the corresponding palatal epithelium (Fig. 1C,D,G). In the Dlx5 mutant, Fgf7 expression is specifically diminished in the nasal half of the palatal mesenchyme but Fgf7 expression in the craniobase persists (Fig. 3H), suggesting that Dlx5 is required for Fgf7 expression in the nasal half of the palatal mesenchyme. Fgf7 has previously been shown to inhibit Shh signaling during lung and limb development (Bellusci et al., 1997; Yonei-Tamura et al., 1999). To test if Fgf7 can inhibit Shh expression in the palatal epithelium, we placed Fgf7 beads into E13.5 palatal shelves in vitro and analyzed Shh expression. After 1 day of culture in serumless, chemically defined media, Fgf7-bead-treated samples showed a dramatic reduction of Shh expression in the palatal epithelium compared with BSA-treated samples (Fig. 7E,F). Furthermore, Fgf7 beads were able to inhibit Shh expression in the palatal epithelium of Dlx5^−/− samples (Fig. 7G), suggesting that Dlx5-dependent Fgf7 expression is sufficient to inhibit the expression of Shh in the nasal half of palatal epithelium during palatogenesis. To confirm the inhibitory effect of Fgf7 on Shh expression, we then treated wild type and Msx1^−/− palatal explants with anti-Fgf7 neutralizing antibody. Inhibition of Fgf7 signaling enhanced Shh expression in both wild type and Msx1 mutant mice (Fig. 7I,K) and restored Shh expression in the anterior region of the palatal shelf (Fig. 7K). Interestingly, exogenous Shh repressed Fgf7 expression (Fig. 7L-N), suggesting a feedback loop in Fgf7/Shh signaling interaction in regulating palatogenesis.

In the posterior part of the palate, we discovered additional defects with complete phenotype penetrance in Dlx5^−/− mice. These included a shortened, detached soft palate and the presence of a uvula-like structure (Fig. 8B). Wild-type mice have a soft palate with a posterior border attached to the pharyngeal wall (Fig. 8A,C). The shortened soft palate in Dlx5^−/− mice resembles velopharyngeal insufficiency and fails to provide an adequate seal between the nasal and oral pharynx (Fig. 8B,D). This anatomical defect also appears to cause air to enter the gastrointestinal tract, as the stomachs of all Dlx5^−/− newborns were inflated with air (Fig. 8E,F).

In this study, we report the existence of O-N patterning and its molecular regulation in the developing palatal shelf. Specifically, phospho-Smad1/5/8 antibody staining marks the activation of BMP signaling and appears to be restricted to the nasal side of the palatal mesenchyme. Interestingly, the expression of Bmp4 and Bmp2 is uniform throughout the palatal mesenchyme (Zhang et al., 2002). The expression of other members of the BMP and GDF families has not been thoroughly investigated. The functional significance of Bmpr1 has been demonstrated in a recent study, in which conditional inactivation of Bmpr1a in the CNC-derived mesenchyme results in cleft lip and palate (Liu et al., 2005). However, the distribution of Bmpr1a in the palatal mesenchyme still needs to be investigated in order to provide an explanation for the asymmetrical activation of BMP signaling in the palatal mesenchyme. It remains possible that the regional differential expression of Bmp receptor controls the establishment of the BMP-responsive domain in the palatal mesenchyme. Alternatively, it is plausible that there is an inhibition of BMP signaling activation on the oral side of the palatal mesenchyme. Further analysis will elucidate the molecular mechanism that controls the asymmetrical activation of BMP signaling in the palate that ultimately results in the formation of palatal bone towards the nasal side of the palate. Significantly, our study demonstrates that, although the palatal mesenchyme is populated with CNC-derived cells, they are not a homogenous population and need to be evaluated and distinguished by molecular marker analysis. This insight will guide our approach in the evaluation of the CNC-derived palatal mesenchyme during palatogenesis.

The restricted expression of Dlx5 on the nasal side and the expansion of the oral side growth of Dlx5^−/− palate clearly suggest that Dlx5 actively participates in the O-N patterning of the developing palate. By contrast, loss of Msx1 does not alter the O-N patterning of the palate. Thus, we conclude that Msx1 is not involved in the O-N patterning of the palate.

Fgf7 is expressed in the nasal region of the palatal mesenchyme and exogenous Fgf7 beads inhibit the expression of Shh in the nasal region of the palatal epithelium, strongly suggesting that Fgf7 is a member of the hierarchy that determines O-N patterning of the palatal mesenchyme. The null mutation of Fgf7 in mice did not generate an obvious palatal phenotype (Guo et al., 1996), as is also the case for the majority of Dlx5^−/− mice (Depew et al., 1999).

The restricted expression of Shh in the oral side of the palatal epithelium and of Gli1 and Fgf10 in the oral side of the palatal mesenchyme suggests that Shh signaling is crucial for the development of the oral side of the palate. In the developing palatal shelf, the cell-surface receptors for Shh (Ptch1 and Smo), hedgehog signaling inhibitors (Hhip1 and Gas1), and Hh-signaling mediator (Gli-family zinc-finger transcription factors) are expressed in the palatal epithelium and mesenchyme (Rice et al., 2006). Conditional inactivation of Shh in the epithelium results in dramatic shortening of the palatal shelves and a wide cleft palate, whereas conditional inactivation of Smo in the epithelium does not disrupt palatogenesis (Hilliard et al., 2005), consistent with the model that the underlying palatal mesenchyme is the recipient of Shh signals. Furthermore, mutations in Gli2 or Gli3 cause facial abnormalities including cleft palate (Mo et al., 1997). We conclude that it is likely that the restricted expression of Shh in the palate is the result of the patterning, in contrast to the setting in the spinal cord, where Shh functions as an inductive signal of patterning. Gli1, which marks the Shh-responding cells (Hooper and Scott, 2005; Lum and Beachy, 2004), is a stronger candidate for the molecule responsible for the patterning of the oral side palatal mesenchyme, because its expression is restricted to the oral half of the palate (see Fig. 3). Our results suggest that, although Shh is not involved in the O-N patterning of the palatal shelf, Shh responsiveness determines the oral half of the palate.

The persistence of Gli1 expression in Msx1^−/− palates indicates that the Gli1 expression in the oral side of the palatal mesenchyme does not require the presence of Shh and is not under the regulation of Msx1. Interestingly, a recent study shows that Gli1 and Ptch1 expression are downregulated in the palatal mesenchyme following the ectoderm-specific inactivation of Shh gene (Lan and Jiang, 2009). To date, studies have shown that hedgehog and TGF-β signaling induces Gli1 expression, whereas Snail/Slug, and Notch signaling, inhibits Gli1 expression (Katoh and Katoh, 2009). Further analysis is necessary to reveal the comprehensive regulation of Gli1 expression during palatogenesis.

Previous studies have shown that Shh functions as a crucial mitogenic factor for the palate mesenchyme through its coordination of Fgf10-Fgfr2b and Msx1-Bmp4 signaling networks (Rice et al., 2004; Zhang et al., 2002). We have shown here that the actively proliferating palatal mesenchymal cells are those adjacent to the Shh-expressing epithelial cells. This spatial correlation is consistent with the restricted range of Shh activity in the limb bud (Li et al., 2006). Shh beads are able to stimulate mesenchymal proliferation in palate explants in vitro, and mesenchymal cell proliferation is reduced in palate explants treated with anti-Shh antibody. In Dlx5^−/− palates, expanded Shh signaling and increased cell proliferation in the palatal mesenchyme might contribute to the overgrowth of rugae and eventual folding of the palatal shelf. However, compromised Shh expression is accompanied by a reduction in cell proliferation in both palatal mesenchyme and epithelium in Fgf10^−/− mice (Rice et al., 2004).

Previous studies have shown that the Shh signaling cascade is regulated at multiple levels. Cholesterol modification can restrict the spread of Shh and control the range and shape of the Shh morphogen gradient (Li et al., 2006). At the intracellular level, the combined activities of hedgehog signaling inhibitors (Hhip and Ptch1) are crucial for tightly controlled Shh activity during pancreas development (Kawahira et al., 2003) and during the initiation of tooth development (Cobourne et al., 2004). Another important mechanism affecting Shh signaling is tissue-tissue interaction. A previous study shows that Msx1 controls the expression of Bmp4, which in turn positively regulates Shh expression during palatogenesis (Zhang et al., 2002). Fgf10 positively regulates Shh expression through Fgfr2b in the palatal epithelium (Rice et al., 2004). We have shown in this study that transcriptional antagonism between Msx1 and Dlx5 in regulating Shh expression ensures the precise spatiotemporal control of Shh signaling during palatal shelf development. More specifically, Msx1-mediated BMP4 signaling is responsible for inducing Shh signaling in the palatal epithelium, whereas Dlx5 is responsible for the indirect inhibition of Shh signaling in the nasal side of the palatal epithelium. As the result of this antagonistic control, there is an asymmetrical distribution of Shh signaling in the palatal epithelium. This specific Shh expression pattern is crucial for the growth and elevation of palatal shelf before fusion.

We propose that Shh signaling is a potential target for the repair of the group of cleft palate cases that result from the failure of palatal shelves to meet at the midline due to compromised palatal mesenchymal proliferation. Restored Shh expression in the palatal epithelium of Msx1^−/−; Dlx5^−/− mice is sufficient to trigger the Shh signaling cascade leading to palatal mesenchyme proliferation and fusion, supported by the evidence that inhibition of Shh signaling in Msx1^−/−; Dlx5^−/− mice reverses the rescue of palatal mesenchyme proliferation. Furthermore, Shh is the converging point for Bmp signaling and Fgf signaling during the expansion stage of palatogenesis (Rice et al., 2004; Zhang et al., 2002). Taking these findings together, we conclude that modulating Shh signaling may provide an opportunity to direct palatal shelf growth and rescue palatal fusion in mutant models with insufficient palatal shelf growth and cleft palate defect.

The discovery of a soft palate defect in Dlx5^−/− mice may have significant implications from an evolutionary perspective. Unlike humans, mice have a soft palate that is attached to the posterior pharyngeal wall. The epiglottis is above the level of soft palate; mice can therefore suckle and breathe at the same time. In humans, the posterior border of the soft palate is free and the epiglottis is below the posterior border of soft palate. This anatomical feature is an important advancement in human evolution because proper function of the soft palate is crucial for speech development. Furthermore, soft palate, along with epiglottis and soft tissue structures within the larynx, acts as a valve to prevent food and liquid from entering lower parts of the respiratory tract. We have found that loss of Dlx5 results in a shortened soft palate in mice. Because Dlx genes are known for their function in regulating the identity of craniofacial structures and morphological novelty in the vertebrate lineage (Beverdam et al., 2002; Depew et al., 2002; Neidert et al., 2001), our data suggest that Dlx5 plays an important role in patterning the proximal region of the pharyngeal arch derivatives.

Our data clearly demonstrate that Msx1 and Dlx5 operate in parallel in regulating downstream target gene expression during palatogenesis. Recent studies show that the function of Msx genes is to control CNC cell cycle progression, as loss of both Msx1 and Msx2 genes results in defects in CNC cell proliferation and survival but does not affect the expression of Dlx5 and the patterning of the branchial arch (Han et al., 2003; Ishii et al., 2005). However, members of the Dlx gene family mainly control the patterning of the craniofacial skeleton, as loss of both Dlx5 and Dlx6 genes results in homeotic transformation of the lower jaw to upper jaw (Depew et al., 2002). Interestingly, the expression of Msx1 and Msx2 is reduced in the first branchial arch of the Dlx5; Dlx6 compound mutant samples, suggesting that Msx genes may function downstream of Dlx5 and Dlx6 in regulating the patterning of the branchial arch derivatives (Depew et al., 2002). In tooth development, however, Msx1 is required for Dlx5 expression in the dental mesenchyme. Msx1 and Dlx5 appear to work synergistically to regulate tooth and alveolar bone development (Zhang et al., 2003). Clearly, Msx1 and Dlx5 can work either antagonistically or synergistically to regulate downstream target gene expression, depending on the context. The outcome of this interaction and whether Msx1 and Dlx5 work in parallel or in a sequential manner depends on the cell and tissue type where the interaction takes place. This operating logic allows for diverse outcomes associated with Msx1/Dlx5 interaction in regulating organogenesis. Equally important, perturbation of the Msx1/Dlx5 interaction may affect an array of downstream target genes and sets the stage for dysmorphogenesis.

We thank Henry Sucov for critical reading and comments. This study was support by grants from the NIDCD, NIH (R01 DC005667) to John Rubenstein and the National Institute of Dental and Craniofacial Research, NIH (DE012711 and DE014078) to Yang Chai. Deposited in PMC for release after 12 months.