Distinct functions for Bmp signaling in lip and palate fusion in mice

Orofacial clefting, a common congenital malformation with a prevalence of from 1 in 500 to 1 in 2,500 live births, has a complex etiology that includes strong genetic and environmental components. Clinically, orofacial clefting has been categorized based on the presence or absence of cleft lip and whether or not it is part of a more extensive syndrome. Moreover, cleft lip with or without cleft palate (CL/P) and isolated cleft palate (CP) have been considered to be separate entities based on genetic and developmental grounds(Murray and Schutte, 2004; Schutte and Murray, 1999). Fusion of the lip occurs earlier in development, complete by 11.5 days post-coitum (dpc) in the mouse, and involves the fusion of the medial and lateral nasal processes with the maxillary process (MP)(Moore and Persaud, 1993). Secondary palate formation, completed in the mouse by 14.5 dpc, is a complex process that involves downward growth of the palatal shelves, derived from the MP, followed by elevation and midline fusion(Ferguson, 1988).

Recent findings have shown that in some families CL/P and CP may result from the same genetic lesion. For example, a nonsense mutation (ser104stop) in the Msx1 homeobox gene, a transcriptional regulator that is a downstream component of bone morphogenetic protein (Bmp) signaling pathways,resulted in both cleft palate (CP) and cleft lip/palate (CL/P) in a Dutch family (van den Boogaard et al.,2000). Moreover, common mutations in p63 have also been shown to cause CL/P and CP within a family(Barrow et al., 2002). Recent evidence suggests that p63 is a direct downstream target of Bmp signaling in zebrafish (Bakkers et al.,2002). Taken together, these data suggest that Bmp-signaling has a role in two forms of orofacial clefting that were previously considered to be distinct.

Bmp ligands are expressed in the facial primordia(Ashique et al., 2002; Barlow and Francis-West, 1997; Francis-West et al., 1994),and are known to signal through broadly expressed type I and type II serine/threonine kinase receptors (von Bubnoff and Cho, 2001). Bead-implantation experiments in chick embryos led to the conclusion that both reduction and enhancement of Bmp signaling within facial primordia caused defective lip fusion(Ashique et al., 2002). Although underscoring the importance of tightly regulated Bmp signaling in lip and palate fusion, the transient nature of bead implantation experiments, as evidenced by the requirement for multiple rounds of bead insertion, may fail to uncover the complete requirements for Bmp function in orofacial development.

Direct investigation of the type 1A Bmp receptor gene Bmpr1a (also referred to as Alk3), and Bmp4 in craniofacial development has been hampered by the early embryonic lethality of the germline null mutant mice (Mishina et al., 2002; Mishina et al., 1995; Winnier et al., 1995). To circumvent this early lethality, we used conditional null alleles of Bmpr1a and Bmp4 to directly investigate Bmp signaling in lip and palate fusion in mouse embryos. We report here that the Bmpr1a-dependent pathway is a major regulator of lip and palate fusion and tooth morphogenesis. Inactivation of Bmpr1a in the craniofacial primordia resulted in CL/P with tooth agenesis. However, we provide evidence that the mechanisms underlying lip clefting and cleft secondary palate are distinct. We also report that deficiency of Bmp4resulted in isolated cleft lip. Taken together, our results reveal that Bmp signaling has distinct functions in lip fusion versus secondary palate development.

The Bmpr1a and Bmp4 conditional null alleles have been previously described (Liu et al.,2004; Mishina et al.,2002). Both alleles have been demonstrated to have wild-type activity prior to recombination and to revert to a null allele after recombination. The Nestin cre transgenic line, previously described,directs cre activity in the craniofacial region, as reported here, but is imprinted in somatic tissues when maternally inherited(Trumpp et al., 1999). For our crosses, the Nestin cre transgene was derived from the male.

Palatal shelves were harvested at 13.0 dpc and 14.0 dpc from wild-type and mutant embryos. The dissected palatal shelves were cultured on 6.5-mm transwell (Costar), keeping the paired shelves with their medial edge epithelia (MEE) in close apposition without apparent distortion of the tissue shape. Paired palatal shelves were cultured overnight using DMEM supplemented with 1% penicillin/streptomycin (Chai et al., 1998).

Branchial arches were dissected and placed in a Petri dish containing HBSS(Sigma) with 10% FCS. After dissection, HBSS was replaced with dispase II (0.8 U/ml from Roche). Explants were incubated for 1 hour at room temperature. Forceps were used to gently separate the epithelium from mesenchyme.

Mouse tissues were fixed in buffered formalin overnight, dehydrated through graded alcohols, and embedded in paraffin wax. Paraffin blocks were sectioned at 7-10 μm and stained with Hematoxylin and Eosin.

After scalding, mice were eviscerated, fixed in 95% ethanol and stained with Alcian Blue (0.015% Alcian Blue dissolved in 20% acetic acid and 76%ethanol). Following two washes with 95% ethanol, the sample was cleared by 2%KOH at room temperature. Bone was stained with Alizarin Red (50 mg Alizarin Red in one liter of 2% KOH).

Pregnant mice were intraperitoneally injected with BrdU (100 mg/kg body weight) 1 hour prior to sacrifice. Embryos were fixed overnight in 10%formalin at 4°C, dehydrated through an ethanol series, cleared in xylene,embedded in paraffin wax and sectioned at 5 μm. BrdU was detected immunohistochemically with a cell proliferation kit (Invitrogen), according to the manufacturer's instructions.

After dissection, the embryos were fixed in fresh fix buffer [(0.2 glutaraldehyde, 2% formalin, 5 mM EGTA, 2 mM MgCl₂, in 0.1 M phosphate buffer (pH 7.3)] for 30 minutes. Following three washes with rinse buffer [0.1% sodium dexycholate, 0.2% NP40, 2 mM MgCl₂, in 0.1 M phosphate buffer pH 7.3)], the samples were stained with staining buffer (1 mg/ml X-gal, 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, in rinse buffer) overnight at room temperature.

The Bmpr1a^flox allele, a conditional null allele of Bmpr1a, has LoxP sites flanking exon 2, which is essential for Bmpr1a function (Mishina et al.,2002). We crossed Bmpr1a^flox homozygous mice,which are normal and fertile, to Nestin cre;Bmpr1a^nullheterozygous mice (Trumpp et al.,1999), to generate embryos that were deficient for Bmpr1ain the craniofacial primordia, the Nestin cre;Bmpr1a^null/flox(n/f) mutant embryos. Crosses to Rosa26 Reporter (R26R) mice revealed that the Nestin cre transgene directed cre activity in the oral and dental epithelium, in epithelium of the medial and lateral nasal processes, and in the MP mesenchyme(Fig. 1A-D). At 13.5 dpc, Nestin cre;R26R compound heterozygous embryos had cre activity in the mesenchyme and epithelium of the palatal shelves(Fig. 1E,F). In situ experiments using the Bmpr1a exon2 probe, which is deleted in the Nestin cre;Bmpr1a n/f mutant embryos, failed to provide reproducible data. Therefore, to understand the timing and completeness of the Bmpr1a recombination, we performed a microdissection experiment followed by PCR.

Microdissection of craniofacial tissues, followed by PCR, indicated that Bmpr1a had been deleted from the MP mesenchyme and epithelium in Nestin cre;Bmpr1a n/f mutant embryos by 10.5 dpc(Fig. 1G). The PCR data also consistently showed that recombination at the Bmpr1a locus was incomplete in the epithelium of the mandibular and lateral and medial nasal processes, as a low intensity PCR product was obtained from these tissues. Taken together, the PCR data and the R26R data reveal that the Nestin cre;Bmpr1a n/f embryos have complete removal of Bmpr1a from the epithelium and mesenchyme of the MP by 10.5 dpc, and mosaic deletion in the epithelium of the mandibular and nasal processes.

At 18.5 dpc, we found that all Nestin cre;Bmpr1a n/f mutant embryos showed a strong phenotype with bilateral cleft lip and palate(Fig. 2A-D). Histological sections of the forming secondary palate at multiple time points revealed that at 13.5 dpc, both the wild-type and the Nestin cre;Bmpr1a n/f palatal shelves grew vertically down the side of the tongue, although the mutant shelves were slightly smaller than the control(Fig. 2E,F). Moreover, at 14.5 dpc, the palatal shelves elevated correctly in both wild-type and mutant embryos (Fig. 2G,H). However,after palatal shelf elevation, the mutant palatal shelves failed to grow together and fuse (Fig. 2I,J). These results suggest that the cleft secondary palate in Nestin cre;Bmpr1a n/f embryos was secondary to failure of palatal shelf growth.

To investigate this, we studied cell proliferation in the MP mesenchyme of wild-type and Nestin cre;Bmpr1a n/f mutant embryos using a BrdU-incorporation assay. At 10.5 dpc, prior to lip fusion, we found that the cell proliferation rates in the maxillary mesenchyme were reduced in Nestin cre;Bmpr1a n/f mutant embryos. In 11.5 dpc embryos, the proliferative index in MP mesenchyme was 43.1% in Nestin cre;Bmpr1a n/f mutants, whereas in wild-type embryos it was 49.3%(P<0.05, χ²). In the epithelium and mesenchyme of the medial and lateral nasal processes, the proliferative index in the mutant was similar to that in the wild-type embryo (data not shown). From these data,we conclude that Bmpr1a function is required for normal cell proliferation in MP mesenchyme.

If the cleft secondary palate of Nestin cre;Bmpr1a n/f mutants was caused by a proliferation defect in the maxillary mesenchyme, then mutant palatal shelves may still retain the ability to fuse in vitro. To investigate this notion, organ culture experiments were performed. Palatal shelves were harvested at 13.0 and 14.0 dpc and cultured overnight with the medial edge epithelium (MEE) in contact. The Nestin cre;Bmpr1a n/f palatal shelves harvested at both 13.0 and 14.0 dpc fused in organ culture(Fig. 2K-N). Moreover, in the palatal shelves harvested at 13.0 dpc, MEE degeneration was clearly visible(Fig. 2M,N). These data indicate that the Nestin cre;Bmpr1a n/f palatal shelves retain the ability to fuse and support the hypothesis that the mechanism underlying cleft secondary palate in the Nestin cre;Bmpr1a n/f embryos was a failure of palatal shelf outgrowth. However, further experiments are required to prove that the MP proliferation defect caused cleft secondary palate in Nestin cre;Bmpr1a n/f embryos.

Recent work has suggested that Bmp signaling plays a role in anterior posterior (AP) patterning of the palatal shelves(Zhang et al., 2002). To investigate whether AP patterning was defective in the Nestin cre;Bmpr1a mutant palatal shelves, we examined the expression of Msx1 in the wild-type and Nestin cre;Bmpr1a mutant secondary palates. Msx1 continued to be expressed in the anterior secondary palate of Nestin cre;Bmpr1a mutant embryos(Fig. 3A-B). The expression of Msx1 in the MP of mutant embryos suggests that Bmp signals are transduced to the MP mesenchyme in the absence of Bmpr1a. This is likely to reflect the contribution of other type I receptors, such as Bmpr1b (also known as Alk6), to Bmp signaling in the MP.

In contrast to Msx1, we found that expression of the paired box gene Pax9, required for normal palate development(Peters et al., 1998), was expanded in the Nestin cre;Bmpr1a mutant embryos(Fig. 3C,D). Similarly, Barx1 expression was also expanded in the secondary palates of Nestin cre;Bmpr1a mutant embryos at 12.5 dpc(Fig. 3G,H). Other genes important for palate development, such as Tbx22 and Osr2(Lan et al., 2004; Marcano et al., 2004), were expressed normally in Nestin cre;Bmpr1a n/f embryos(Fig. 3G-J). From these data,we conclude that signaling through Bmpr1a is necessary for correct spatiotemporal expression of crucial transcriptional regulators of palate development, such as Barx1 and Pax9.

To gain insight into the molecular mechanism underlying cleft lip in Nestin cre;Bmpr1a mutants, we examined markers of the edge epithelium of the MP and medial nasal process (MNP). Fgf8 is expressed in the epithelium of the nasal processes and the MP in chick embryos(Helms et al., 1997). Moreover, Fgf signaling is known to closely interact with Bmp signaling pathways to regulate craniofacial and tooth development(Neubuser et al., 1997; Tucker et al., 1998). In addition to Fgf8, we examined expression of p63 and Pitx1, which are expressed in the edge ectoderm of the fusing lip. P63 has also been implicated in cleft lip and palate in human patients (Celli et al., 1999),and Pitx1 is a known target gene for Fgf8 in mandibular development(St Amand et al., 2000).

To precisely define the onset of the lip fusion defect, we examined Fgf8 expression in embryos that had been carefully staged by counting somite number (Fig. 4A-D). In 29-somite embryos (10.0 dpc), Fgf8 expression was similar in the wild-type and Nestin cre;Bmpr1a mutant embryos(Fig. 4A,B). Fgf8 was expressed in the proximal part of the nasal process ectoderm but was not yet expressed in the fusing region. By contrast, 31-somite embryos (10.25 dpc)expressed Fgf8 in the ectoderm at the lip fusion point, whereas Nestin cre;Bmpr1a mutants failed to upregulate Fgf8 in this region (Fig. 4C,D). This difference in Fgf8 expression was more pronounced in 10.5 dpc (34-35 somites) embryos (Fig. 4E,F). In addition to diminished Fgf8 expression, we found that Nestin cre;Bmpr1a mutants also had downregulated expression of both p63and Pitx1 in the fusing ectoderm of the nasal processes in 10.5 dpc embryos (Fig. 4G-L).

We next studied apoptosis in the fusing lip region of wild-type and Bmpr1a mutant embryos at 10.5 dpc (34-35 somites). In wild-type embryos, little apoptosis was detected in fusing ectoderm or underlying mesenchyme, but in mutant embryos upregulation of apoptosis was detected in the ectoderm and mesenchyme of the MNP(Fig. 4M-O). Taken together,these data indicate that reduced Bmp signaling results in elevated apoptosis in the ectoderm and mesenchyme of the MNP. Moreover, Bmp signaling in the edge epithelium of the frontonasal process and the MP is required for the expression of markers of the edge epithelium of the MP and MNP, such as Fgf8 and p63.

Consistent with the role of Msx1 in tooth agenesis in human patients, we detected tooth defects in Nestin cre;Bmpr1a mutants. At 16.5 dpc, mandibular and maxillary first and second molars are normally in the late and early cap stage of development, respectively(Fig. 5A,B). The inner and outer enamel epithelium in the developing molars are separated by the stellate reticulum. The maxillary molars in Nestin cre;Bmpr1a n/f 16.5 dpc embryos were arrested at the bud stage, with an invaginated dental lamina encircled by condensed mesenchyme (Fig. 5A,C). In contrast to the defective maxillary molars, Nestin cre;Bmpr1a n/f 16.5 dpc embryos had well-developed cap-staged mandibular molars (Fig. 5B,D). In 16.5 dpc wild-type embryos, there are two well-developed incisor tooth germs in each arch (Fig. 5E,F). In the Nestin cre;Bmpr1a mutant embryos, development of the mandibular incisors was normal but mutant embryos lacked maxillary incisor teeth(Fig. 5G,H).

At 18.5 dpc, the wild-type mandibular and maxillary molars have progressed to the bell stage and contain newly differentiated odontoblasts that secrete pre-dentine and contain polarized, presecretory ameloblasts(Fig. 5I,J). The bud-staged mutant maxillary molars had failed to progress at 18.5 dpc (compare Fig. 5C with 5K). The epithelial invagination lacked the typical cap morphology and was surrounded by dental papilla-like mesenchyme (Fig. 5K). Mandibular molars in 18.5 dpc mutant embryos were arrested at the cap stage of development. First and second mandibular molars were present but were in close proximity to each other, giving the appearance of fused tooth buds (Fig. 5J,L). Taken together, these data reveal that signaling through Bmpr1a is required for maxillary tooth development to progress beyond the bud stage.

In order to gain insight into the ligands that would be involved in cleft lip and palate, we examined expression of Bmp4 and Bmp2. At 10.5 dpc, we found that Bmp2 and Bmp4 are co-expressed in the edge epithelium at the point of fusion between the MNP and the MP(Fig. 6A,B). We next used a conditional null allele of Bmp4, the Bmp4^floxneoallele, to address the functional role of Bmp4 in CL/P. The Bmp4^floxneo allele has LoxP sites flanking exon 4 that encodes the mature Bmp4 ligand. Removal of the LoxP flanked region has been shown to result in a null allele (Liu et al., 2004).

To determine when Bmp4 had been removed from the branchial arch ectoderm in Nestin cre;Bmp4^null/flox(n/f)embryos, we performed in situ analysis with the Bmp4 exon 4 probe that hybridizes to the deleted region of the Bmp4^floxneoallele. At 10.5 dpc, we found that the Bmp4 exon 4 probe hybridized to the edge ectoderm of the fusing lip in wild-type embryos, but, in the Nestin cre;Bmp4 n/f mutant embryos, no hybridization signal was detected (Fig. 6C,D). We conclude that in Nestin cre;Bmp4 n/f embryos, Bmp4 has been deleted by 10.5 dpc in the edge epithelium of the MNP and MP.

We examined Nestin cre;Bmp4 n/f mutant embryos at two developmental timepoints. At 12.0 dpc, we found that all Nestin cre;Bmp4 n/f mutant embryos had a bilateral delay in fusion of the MNP and MP(Fig. 6E,F). At 14.5 dpc, two out of nine Nestin cre;Bmp4 n/f mutants had unilateral isolated cleft lip (Fig. 6G,H). Taken together, these data reveal that Bmp4 functions in the ectoderm of the nasal processes to regulate lip fusion. Moreover, these findings suggest that most Nestin cre;Bmp4 n/f mutants spontaneously repair the cleft lip that we observed at 12.0 dpc.

Our data uncover a crucial role for Bmp signaling in lip and secondary palate development. In the Nestin cre;Bmpr1a n/f embryos, defective proliferation was the likely cause for secondary palate clefting. In the lip,we provide evidence that premature apoptosis in the MNP edge epithelium resulted in clefting. Moreover, the phenotype of the Nestin cre;Bmp4 n/f embryos suggest that a Bmp4-BmpR1A planar signaling pathway functions in lip fusion. These findings provide insight into the mechanisms underlying the mixed clefting phenotypes observed in human patients.

Development of the secondary palate is a complex process that requires outgrowth, elevation and fusion of the palatal shelves(Ferguson, 1988). Our analysis revealed that Bmpr1a is required for proliferation of MP mesenchyme. The in vitro explants showed that the Bmpr1a mutant palatal shelves are competent to fuse when placed together in culture. Previous work in secondary palate development uncovered a genetic pathway in which Msx1 and Bmp4 function in an autoregulatory loop to regulate proliferation in anterior palate mesenchyme (Zhang et al.,2002). Our data reveal that Bmp signaling is also required for cell proliferation at earlier stages, in the MP mesenchyme.

Two previous studies looking at Bmp signaling in chick embryos reached conflicting conclusions regarding the regulation of cell proliferation in the facial prominences. One study concluded that implantation of Bmp-soaked beads in MP mesenchyme resulted in elevated proliferation(Barlow and Francis-West,1997). The second study, while concluding that Nogginreduced cell proliferation, found that implanted Bmp2 beads failed to influence cell proliferation (Ashique et al., 2002). Our data support the notion that Bmp signaling promotes cell proliferation of MP mesenchyme.

We found that Bmp signaling is required to restrict expression of Barx1 and Pax9 in the forming secondary palate. It is interesting to note that in the developing mandible, Bmp signaling has been shown to repress expression of Pax9 and Barx1, suggesting a conservation of pathways in the palate and mandible(Neubuser et al., 1997; Tucker et al., 1998; Zhang et al., 2002). Moreover,the changes in gene expression in the Nestin cre;Bmpr1a mutant secondary palates support the idea that cleft secondary palate is not a consequence of failed lip fusion.

We have found evidence for genetic redundancy among Bmp receptors in the MP. The Nestin cre;Bmpr1a mutant embryos still express Msx1in the palatal mesenchyme, suggesting that Bmp signals are transduced in the absence of Bmpr1a. Neural crest specific ablation of another type 1 Bmp receptor (Alk2) also resulted in cleft palate, although this may be secondary to mandibular defects (Dudas et al., 2004). Another possible candidate for redundancy is Bmpr1b, which is expressed in craniofacial processes (W.L. and J.F.M., unpublished). It will be important to explore the question of genetic redundancy between the type 1 Bmp receptors in future experiments.

Our data suggest that a Bmp4-Bmpr1a autoregulatory loop in the edge epithelium is necessary for lip fusion. These findings implicate the Bmp4-BmpR1A pathway in human clefting syndromes, and are consistent with previous observations showing the association between the Bmp-target Msx1 and CL/P (van den Boogaard et al., 2000). It was previously shown that an Arg239Pro Msx1 mutation resulted in tooth agenesis but no clefting in human patients (Vastardis et al.,1996). The second family with a Ser104stop mutation in Msx1 had both CL/P and selective tooth agenesis(van den Boogaard et al.,2000). The defects in both families are likely to result from haploinsufficiency and suggest that the Arg239Pro Msx1 mutation retained more residual function than the Ser104 stop mutation(Hu et al., 1998).

Our observation that most Nestin cre;Bmp4 n/f mutant embryos repair their cleft lip also implicates Bmp4 in clinical cases of microform cleft lip that have been observed in human patients. It is notable that congenital heart disease has been associated with microform clefting(Castilla and Martinez-Frias,1995; Grech et al.,2000). We, and others, have recently reported that Bmp4has a direct role in cardiac morphogenesis, suggesting an important role for Bmp4 in human congenital heart disease(Jiao et al., 2003; Liu et al., 2004).

Our findings reveal that Bmpr1a is required for survival of the edge epithelium and mesenchyme of the MNP. Work in chick embryos has shown that during lip fusion, the outermost periderm epithelium of the frontonasal mass and MP undergoes apoptosis. This exposes a fusion-competent basal epithelium, allowing basal cells to fuse through a process that may involve desmosomes. Cells within the fusing region, or the seam, undergo an epithelial to mesenchymal transition during remodeling of the lip(Sun et al., 2000). Our data suggest that, in Bmpr1a mutants, premature apoptosis in the edge epithelium of the MNP is responsible for clefting of the lip.

The role of programmed cell death in facial process fusion is controversial. The epithelial-mesenchymal transformation of the MEE to palate mesenchyme has been considered to be the critical determinant for secondary palate fusion (Fitchett and Hay,1989). However, recent studies have challenged this model by suggesting that apoptosis is required for normal fusion to occur(Cuervo and Covarrubias, 2004; Cuervo et al., 2002). With regard to the data presented here, it is known that premature induction of apoptosis in palatal shelves inhibited fusion, perhaps as a result of reduced cell adhesion (Cuervo et al.,2002). Further experiments will be required to determine whether failed fusion of the edge epithelium in Nestincre;Bmpr1a n/f mutants is secondary to reduced cellular adhesion.

It is also notable that, in addition to elevated apoptosis in Nestin cre;Bmpr1a n/f mutant nasal ectoderm, there was elevated apoptosis in the underlying mesenchyme (Fig. 4M-O). This suggests that the defect in lip fusion may result from loss of the MNP mesenchyme in addition to the mechanisms discussed above. Further experiments will be required to investigate this.

The in situ analysis presented here revealed defects in the expression of ectodermal markers of the fusing lip, Fgf8, Pitx1 and p63,in Nestin cre;Bmpr1a n/f mutants. Fgf8 expression in the fusing lip region of the Nestin cre;Bmpr1a n/f mutants revealed a failure to upregulate Fgf8 in 31-somite embryos prior to induction of apoptosis. In the mandibular process, Fgf8 has an important role in promoting the survival of underlying mesenchyme(Trumpp et al., 1999). Therefore, in Nestin cre;Bmpr1a n/f mutants, the failure of Fgf8 upregulation may directly result in elevated apoptosis and lip clefting. This notion is also supported by recent observations in chick embryos treated with retinoic acid inhibitors. In these retinoid deficient embryos, defective closure of the nasal pit was associated with elevated apoptosis and downregulation of Fgf8 in the lateral nasal process(Song et al., 2004). However,further experiments will be required to definitively show that apoptosis is the cause of lip clefting in the Nestin cre;Bmpr1a n/f mutants.

An alternative hypothesis is suggested by the observation that p63has previously been shown to play an important role in differentiation of embryonic epidermis (Mills et al.,1999; Yang et al.,1999). Furthermore, recent work uncovered multiple, functional Smad regulatory elements in the zebrafish p63 gene(Bakkers et al., 2002). Taken together with our observation that p63 is reduced in the edge epithelium of the MNP and MP of Nestin cre;Bmpr1a n/f mutants, these data suggest the existence of a linear genetic pathway important for maturation and subsequent fusion of the edge epithelium. The connection of p63 to clefting in humans makes this idea worthy of further investigation (Celli et al.,1999).

Bead implantation experiments performed in chick embryos concluded that, in the maxillary epithelium, Bmp signaling promoted cell death and loss of epithelium. Beads soaked in Noggin resulted in reduced cell death in the epithelium of the frontonasal mass, and thus epithelial integrity was maintained in the pre-fusion phases. Moreover, Fgf8 expression in the frontonasal process was increased (Ashique et al., 2002). By contrast, our data suggest that Bmp signaling in the epithelium is required for cell survival and upregulation of Fgf8expression. The different results may stem from the large doses of Noggin that were necessary to obtain phenotypes in the chick experiments. It is also notable that we inactivated Bmp signaling in the nasal process epithelium,whereas Noggin was placed in the mesenchyme and so may have indirectly disrupted an unidentified signal from mesenchyme to epithelium.

It has been proposed that Bmp signaling has multiple roles in tooth development, including tooth type morphogenesis, tooth organ placement, and signaling within and from the enamel knot(Jernvall et al., 1998; Neubuser et al., 1997; Tucker et al., 1998). The Nestin cre;Bmpr1a n/f mutants display different tooth phenotypes in the mandible and maxilla. In the maxilla, molars arrest at the bud stage,while incisors arrest at an earlier stage. In the mandible, molars arrest at the cap stage, while incisor development is normal. Recently published observations showing that inactivation of Bmpr1a in dental epithelium resulted in arrest of tooth development at the bud stage(Andl et al., 2004) suggest that the distinct tooth phenotypes we observed are likely to result from variable cre activity. Nonetheless, our data also raise the possibility that different teeth may have different requirements for levels of Bmp signaling to complete organogenesis.

Our RT-PCR analysis showed that the Nestin cre;Bmpr1a n/f mutants mandibular ectoderm had mosaic cre activity, whereas maxilla had a more complete deletion of Bmpr1a at 11.5 dpc. The cap stage arrest is consistent with previous data showing that Bmp4 signaling to the forming enamel knot is crucial for the progression of tooth development(Bei et al., 2000; Chen et al., 1996; Jernvall et al., 1998).

Bmp4 in mesenchyme also has been proposed to induce expression of p21, and cell cycle arrest in cells of the prospective enamel knot. Once induced in enamel knot, Bmp4 functions to regulate tooth organ shape by controlling cell cycle progression and apoptosis. It is plausible that mosaic disruption of Bmpr1a in the dental epithelium would allow tooth germ development to progress beyond the bud stage and arrest at stages when enamel knot function would become important. Future experiments will be needed to investigate these ideas in more detail.

We thank R. Jiang for comments on the manuscript and J. Murray for discussions; A. Bradley, P. Soriano and R. Behringer for reagents; A. McMahon,G. Martin, R. Balling, T. Mitsiadis, B. Hogan and E. Olson for in situ probes;and B. Hogan and Y. Furuta for the Bmp4^lacZ allele. Supported by NIH grants (2R01DE/HD12324-06 and R01DE013509) to J.F.M., DE08682 to M.M., and AR42919 to R.R.B.