Conditional inactivation of Tgfbr2 in cranial neural crest causes cleft palate and calvaria defects

The vertebrate neural crest is a pluripotent cell population derived from the lateral ridges of the neural plate during early stages of embryogenesis. During craniofacial development, cranial neural crest (CNC) cells migrate ventrolaterally as they populate the branchial arches. The proliferative activity of these crest cells produces the discrete swellings that demarcate each branchial arch. Following their migration, CNC cells contribute extensively to the formation of mesenchymal structures in the head and neck,of which palate and calvaria development are classic examples. The migration,proliferation and differentiation of CNC cells are regulated by growth factor signaling pathways and their downstream transcription factors before they become committed to an array of different phenotypes(Noden, 1983; Noden, 1991; Lumsden, 1988; Graham and Lumsden, 1993; Le Douarin et al., 1993; Echelard et al., 1994; Imai et al., 1996; Trainor and Krumlauf,2000).

The mammalian palate develops from two primordia: the primary and the secondary palate. The primary palate represents only a small part of the adult hard palate and is the part anterior to the incisive fossa. The secondary palate is the primordium of the hard and soft palate in adults. Palate development is a multi-step 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 CNC-derived ectomesenchyme and pharyngeal ectoderm(Ferguson, 1988; Shuler, 1995; Wilkie and Morriss-Kay, 2001; Zhang et al., 2002).

TGFβ signaling plays a pivotal role in regulating palatogenesis. During mouse palatal development, both TGFβ1 and TGFβ3 are expressed in the medial edge epithelium (MEE) of the palatal shelves, whereas TGFβ2 expression is restricted to the CNC-derived mesenchyme beneath the MEE(Fitzpatrick et al., 1990; Pelton et al., 1990). Upon fusion of the palatal shelves and disappearance of the midline epithelial seam, the expression of TGFβ1 and TGFβ3 is lost, suggesting crucial functions of TGFβ signaling in regulating palatal fusion. Loss-of-function mutation of Tgfb2 or Tgfb3 results in cleft palate. Tgfb2-null mutant mice exhibit anteroposterior cleft of the secondary palate with only 23% phenotype penetrance(Sanford et al., 1997). Significantly, Tgfb3-null mutation results in 100% penetrance of cleft secondary palate (Kaartinen et al.,1995; Proetzel et al.,1995). The etiology of cleft palate in Tgfb3-null mutant mice is apparently due to a failure of fusion of palatal shelves, which has been rescued by addition of exogenous TGFβ3 in an in vitro organ culture system (Brunet et al., 1995; Taya et al., 1999). Subsequent studies have shown that TGFβ3 is specifically required for the fusion of palatal shelves, probably by enhancing the transformation of MEE cells into the palatal mesenchyme and inducing apoptosis in the MEE(Sun et al., 1998; Martinez-Alvarez et al.,2000).

TGFβ IIR is expressed in both the MEE and CNC-derived palatal mesenchyme (Wang et al., 1995; Cui et al., 1998). The physiological function of TGFβ IIR in regulating palatogenesis is not known because Tgfbr2-null mutation results in early embryonic lethality, thus, making it impossible to investigate the functional significance of this signaling molecule in regulating palatogenesis(Oshima et al., 1996). Up until now, most of the palatogenesis studies, such as the ones involving TGFβ signaling, have mainly focused on the molecular regulation of the fate of MEE cells. Although CNC cells are critical for palatogenesis, very little is known about the molecular mechanism that regulates the fate of the CNC-derived palatal mesenchyme during palatogenesis.

The vertebrate skull includes both the neurocranium (such as the calvaria and base of skull) and viscerocranium (such as mandible, zygoma, maxilla,etc.). Calvaria formation is a complex and lengthy developmental process that is initiated during embryogenesis and is completed in adulthood. The size flexibility of the calvaria is crucial for accommodating the rapid growth of the brain. Both the mesoderm and CNC-derived ectomesenchyme contribute to the cranial skeletogenic mesenchyme, which gives rise to bony elements (such as frontal, parietal and occipital bones) collectively known as the calvaria(Wilkie and Morriss-Kay,2001). Studies have shown that the dura mater, a dense fibrous membrane underneath the calvaria, and cranial sutures provides crucial regulatory signals for calvaria development. To date, studies suggest that cranial sutures function as signaling centers for bone growth and remain patent postnatally to accommodate cranium expansion. Premature closure of cranial sutures affects the growth of the calvaria and results in craniosynostosis (Wilkie and Morriss-Kay,2001).

Multiple growth and transcription factors play pivotal roles in regulating the osteogenic ability of cranial sutures. In particular, TGFβ signaling stimulates osteogenic progenitor cell proliferation and can induce premature suture obliteration in cultured fetal rat calvaria, suggesting that TGFβsignaling plays an important regulatory role in postnatal calvaria development(Opperman et al., 2000). In addition, TGFβ signaling within the immature dura mater (in newborn and immature animals) possesses the ability to induce calvaria bone repair, while diminished TGFβ signaling within the mature dura mater fails to repair calvarial defect, suggesting that TGFβ signaling is a crucial regulator for calvarial ossification (Greenwald et al., 2000). TGFβ IIR is expressed in the dura mater and cranial sutures, presumably playing an important role during skull development(Pelton et al., 1990; Lawler et al., 1994; Wang et al., 1995). Collectively, these studies have demonstrated that TGFβ signaling has an important regulatory function for postnatal cranial suture patency and skull repair. However, it remains unclear what the physiological function of TGFβ signaling is in regulating the initiation and development of the calvaria during embryogenesis.

To investigate the role of TGFβ signaling in regulating the fate of CNC cells during palate and calvaria development, we performed tissue-specific Tgfbr2 gene ablation using Cre/loxP recombination exclusively in the cranial neural crest lineage. Our study shows that loss of Tgfbr2 in the CNC cells results in cleft secondary palate and calvaria defects with 100% phenotype penetrance. Specifically, conditional Tgfbr2 mutation inhibits cyclin D1 expression and affects CNC cell proliferation in the palatal mesenchyme. The midline epithelium of the mutant palatal shelf remains functionally competent to mediate palatal fusion once the palatal shelves are placed in close contact in vitro. Disruption of TGFβ signaling in the CNC severely impairs cell proliferation in the dura mater, consequently resulting in calvaria agenesis. We provide the first in vivo evidence that TGFβ signaling within the CNC-derived dura mater provides essential inductive instruction for both the CNC- and mesoderm-derived calvarial bone development.

Both Wnt1-Cre transgenic line and R26R conditional reporter allele have been described previously(Danielian et al., 1998; Soriano, 1999). Mating Wnt1-Cre and R26R mice generated transgenic mice with progenies of neural crest cells labeled with β-gal because once Wnt1-Cre expression commences in premigrating neural crest cells, theβ-galactosidase is indelible. Detection of β-galactosidase(lacZ) activity in both whole embryos and tissue sections was carried out as previously described (Chai et al.,2000).

All mouse embryos used in this study were maintained on C57BL6/J background. Mating Tgfbr2^fl/+;Wnt1-Cre with Tgfbr2^fl/fl mice generated Tgfbr2^fl/fl;Wnt1-Cre null alleles that were genotyped using PCR primers as previously described(Chytil et al., 2002). All samples were fixed in 10% buffered formalin and processed into serial paraffin wax-embedded sections using routine procedures. For general morphology,deparaffinized sections were stained with Hematoxylin and Eosin using standard procedures.

DNA synthesis activity within the palate or skull was monitored by intraperitoneal BrdU (5-bromo-2′-deoxy-uridine, Sigma) injection (100μg/g body weight) at E12.5, E13.4 and E14.5. One hour after the injection,mice were sacrificed and embryos were fixed in Carnoy's fixative solution and processed. Serial sections of the specimen were cut at 5 μm intervals. Detection of BrdU labeled cells was carried out by using a BrdU Labeling and Detection kit and following manufacturer's protocol (Boehringer Mannheim). BrdU-positive and total number of cells within the palatal mesenchyme or MEE of palatal shelf were counted from five randomly selected sections per sample. Five palate samples were evaluated from each experimental group. TUNEL assay was performed using the In Situ Cell Death Detection (fluorescein) kit (Roche Molecular Biochemicals) by following the manufacturer's protocol. Cell density analysis was performed by counting the number of cells per unit area from 20 randomly selected sections per experimental group. Student's t-test was applied for statistical analysis. A P value of less than 0.05 was considered statistically significant.

Timed-pregnant mice were sacrificed on postcoital day 13.5 (E13.5). Genotyping was carried out as described above. The palatal shelves were microdissected and cultured in serumless chemically defined medium as previously described (Shuler et al.,1991). After 3 days in culture, palates were harvested, fixed in 10% buffered formalin and processed.

The total protein concentration in the palates was determined by comparison with BSA standards. Seventy-five micrograms total protein from each sample was loaded in each well on a 12% polyacrylamide gel. Western analysis was carried out as previously described (Chai et al.,1999). Antibodies used: anti-cyclin D1 and anti-CDK4 (BD Biosciences), anti-Msx1 (kindly provided by P. Denny, USC) and anti-β-actin (Santa Cruz Biotechnology).

The three-dimensional architecture of the skeleton was examined using a modified whole-mount Alcian blue-Alizarin Red S staining protocol (details available upon request).

Sectioned immunohistochemistry was performed with an Immunostaining kit(Zymed) according to manufacturer's directions. The following antibodies were used for this experiment: anti-BrdU (Sigma), anti-cyclin D1 (BD Biosciences)and anti-p21 (Santa Cruz Biotechnology). Positive staining was shown in orange-red for immunohistochemistry. The slides were counterstained with Hematoxylin.

To date, little is known about the fate of the CNC-derived palatal mesenchyme or the molecular mechanism that regulates the specification of these progenitor cells during palate development. We provide in vivo analysis of the dynamic distribution of CNC cells during palatogenesis by using the Wnt1-Cre;R26R animal model for indelibly marking the progenies of CNC cells (Chai et al., 2000). During extension of the palatal shelf, CNC-derived cells (blue) are mixed with non-CNC-derived cells (pink, mesenchymally derived) at both anterior and posterior regions of the developing palate at E13.5(Fig. 1A-D). Between E13.5 and E14.5, rapid growth of the palatal shelves brings the two processes into horizontal apposition above the tongue. Subsequently, the opposing palatal shelves fuse following the disappearance of midline epithelial cells at around E14.5. At this point, the anterior region of the secondary palate shows disruption of the midline epithelium at the fusion site(Fig. 1E,F). The palatal mesenchyme is mainly populated with CNC-derived cells, especially in the region adjacent to the midline epithelium, indicating the important biological function of the CNC cell during palatal fusion(Fig. 1F). In the posterior portion of the secondary palate, the opposing palatal shelves have fused,leaving a remnant of a continuous midline epithelium at the fusion site(Fig. 1G,H). At E15.5, palatal fusion is complete and the palatal mesenchyme is mainly populated with CNC-derived cells (Fig. 1I,J). After fusion, CNC-derived cells have begun to form an aggregated cell mass to initiate palatal bone formation in the palatal mesenchyme(Fig. 1K,L).

Although TGFβ IIR is strongly expressed in the CNC-derived palatal mesenchyme, mice deficient for the Tgfbr2 gene die on embryonic day 10.5 (E10.5) as the result of defects of yolk sac hematopoiesis and vasculogenesis (Wang et al.,1995; Oshima et al.,1996). To circumvent this early lethality and to investigate the specific function of TGFβ IIR in regulating CNC cells during palatogenesis, we crossed a Tgfbr2 conditional allele with the Wnt1-Cre transgenic mouse line and generated Tgfbr2^fl/fl;Wnt1-Cre embryos, in which the CNC-derived palatal mesenchyme was homozygous for the Tgfbr2^fl/fl null allele. Genetically, in the presence of Cre recombinase, the second exon of the Tgfbr2 gene is removed, resulting in a null allele as previously described (Chytil et al.,2002). The Wnt1 transgene drives Cre expression specifically in the neural crest lineage(Chai et al., 2000). In control(normal) embryos, one or both active Tgfbr2 allele(s) was retained.

The complete failure of mouse secondary palate fusion was first detected in Tgfbr2^fl/fl;Wnt1-Cre mutant embryos at E14.5 when normal palatal fusion had just occurred (Fig. 2A,B). We compared cross-sections of E14.5 Tgfbr2^fl/fl;Wnt1-Cre mutant embryonic heads with the ones of Tgfbr2^+/fl;Wnt1-Cre or Tgfbr2^fl/fllittermate embryos. There was decreased cellular density (P<0.05)in the elevated palatal shelf mesenchyme of Tgfbr2^fl/fl;Wnt1-Cre mutant embryos (4298±275 cells/mm²) when compared with the normal developing palate(5174±168 cells/mm²), in which fusion occurred with the partial disappearance of the midline epithelium(Fig. 2C-F). At E16.5, both of the palatal shelves had elevated into horizontal position but failed to fuse at the midline in Tgfbr2^fl/fl;Wnt1-Cre mutant embryos,while completed palatal fusion was observed in the control samples(Fig. 2G,H). The CNC-derived palatal mesenchyme began to form an aggregated cell mass as a prelude to palatal bone development in the control sample(Fig. 2I), while CNC cell condensation was not observed in the Tgfbr2^fl/fl;Wnt1-Cremutant embryo (Fig. 2J). At birth, complete cleft secondary palate was observed in Tgfbr2^fl/fl;Wnt1-Cre mutant mice with 100% (36/36 newborn pups) phenotype penetrance (Fig. 2L).

Previous studies have shown that TGFβ signaling plays a pivotal role in regulating the fate of the medial edge epithelium (MEE) during palatal fusion (Pelton et al., 1990; Pelton et al., 1991; Fitzpatrick et al., 1990; Kaartinen et al., 1995; Sun et al., 1998; Martinez-Alvarez et al.,2000). To determine whether the Tgfbr2^fl/flmutant MEE had any altered cellular function and was competent to mediate palatal fusion, we first evaluated cell proliferation and apoptosis activity in the MEE and found no difference between the Tgfbr2^fl/fl;Wnt1-Cre mutant and wild-type control samples at E12.5, E13.5 and E14.5 (data not shown), thereby suggesting that altered TGFβ signaling in the CNC-derived palatal mesenchyme did not adversely affect the fate of MEE cells during palatal fusion.

Next, we hypothesized that the failure of palatal fusion in the Tgfbr2^fl/fl;Wnt1-Cre mutant mice was due to insufficient extension of the palatal shelves towards the midline. To test our hypothesis,we performed palatal fusion analysis by using a palatal shelf organ culture model. At E13.5, the developing palatal shelves were pointing downwards on both sides of the tongue. Each isolated pair of palatal shelves was placed in culture with the two segments just touching at the medial edge and kept in the original anteroposterior orientation, thus preventing any variability in growth rates from adversely affecting palatal development. During the 3 day culture period, both wild-type and Tgfbr2^fl/fl mutant palatal specimens fused. All cultured wild-type palatal shelves (n=32 pairs) showed complete fusion with normal disappearance of the MEE and development of a confluent palatal mesenchyme(Fig. 3A,C). Furthermore,osteoid-like structure was present in the cultured palatal shelf, suggesting that palatal bone formation was initiated in vitro(Fig. 3C, insert). Although all cultured Tgfbr2^fl/fl;Wnt1-Cre mutant palatal shelves also showed fusion (n=9 pairs), some fused palates (4/9, 44%) had residual epithelium (arrow) at the midline, indicating a possible delay in the fusion process (Fig. 3B,D). Nevertheless, the MEE cells were competent to facilitate palatal fusion in Tgfbr2^fl/fl mutant samples once the palatal shelves were placed in close contact. In addition, osteoid-like structure was present(Fig. 3D, insert), suggesting that there was normal palatal bone formation in the cultured Tgfbr2^fl/fl;Wnt1-Cre mutant palatal shelf.

In order to test whether a CNC migration defect might have contributed to the deficiency of the CNC-derived palatal mesenchyme, and was responsible for the failure of palatal fusion, we crossed the Tgfbr2 conditional allele with R26R transgenic mice and generated embryos with Tgfbr2^fl/fl;R26R;Wnt1-Cre mutation. All of these embryos had identical malformations (such as complete cleft secondary palate) to the ones seen in the Tgfbr2^fl/fl;Wnt1-Cre mutant mice. Whole-mount and sectioned β-gal staining showed no difference in migration or distribution of CNC cells within the first branchial arch and the frontonasal prominence between Tgfbr2^fl/fl;R26R;Wnt1-Cremutant and the wild-type control embryos from E8.5 to E11.5(Fig. 4A-F and data not shown). At E14.5, palatal fusion was well under way in the wild-type embryos with CNC-derived cells populating the majority of the palatal mesenchyme(Fig. 4G). In Tgfbr2^fl/fl;R26R;Wnt1-Cre mutant embryos, the palatal shelves were populated with the CNC-derived mesenchyme, without any indication of a deficiency in CNC migration (Fig. 4H). Taken together, our data suggest that there is no CNC migration defect that might have resulted in inadequate palatal shelf extension and failure of palate fusion in Tgfbr2^fl/fl;Wnt1-Cre mutant mice. We infer that TGFβsignaling is specifically required in the CNC-derived mesenchyme prior to palatal fusion.

To explore the mechanism responsible for causing the failure of palatal shelf extension in Tgfbr2^fl/fl;Wnt1-Cre mutant embryos, we investigated whether there was a decrease in cell proliferation, an increase in apoptosis, or a combination of both in the CNC-derived palatal mesenchyme. Cell proliferation activity within the CNC-derived palatal mesenchyme, as measured by BrdU incorporation, appeared to be identical between control and Tgfbr2^fl/fl;Wnt1-Cre mutant embryos at E12.5 and E13.5(Fig. 5A-D). However, at E14.5,there was a significant reduction (P<0.01) in the cell proliferation rate within the CNC-derived palatal mesenchyme of the Tgfbr2^fl/fl mutant embryos (18±2.9%) when compared with the ones of the wild-type control (29.6±5.9%)(Fig. 5E,F; Fig. 6). To rule out the possibility that palatal fusion itself was responsible for maintaining proliferation in the CNC-derived palatal mesenchyme, we had analyzed BrdU labeling indices in the palatal shelves both prior to and right after fusion(Fig. 5E and insert in E). The cell proliferation rate remained identical (30-32%) in the CNC-derived palatal mesenchyme of the wild-type embryos before and after the fusion process. Furthermore, by processing single slide for β-gal and BrdU double staining, we found that the cell proliferation defect was exclusively associated with the CNC-derived palatal mesenchyme (not with non-CNC-derived mesenchyme) in the Tgfbr2^fl/fl;Wnt1-Cre mutant embryos(Fig. 2G,H). We concluded that TGFβ signaling specifically controls cell cycle progression in the CNC-derived palatal mesenchyme prior to palatal fusion. As Wnt1-Credid not cause Tgfbr2 deletion in the non-CNC-derived palatal mesenchyme, it was not possible to determine whether TGFβ IIR played a significant role in regulating the non-CNC-derived mesenchymal cell proliferation in the developing palate.

In order to understand the mechanism of TGFβ signaling in regulating the progression of the CNC-derived palatal mesenchymal cell cycle, we investigated possible alteration of cell cycle regulator expression in the Tgfbr2^fl/fl;Wnt1-Cre mutant embryos. Cyclin D1, a member of the cyclin D family, functions to regulate phosphorylation of the retinoblastoma gene products, thereby activating E2F transcription to facilitate cell cycle progression. We show that the expression of cyclin D1 was comparable in the palatal mesenchyme between the Tgfbr2^fl/fl mutant and the control samples at E12.5 and E13.5 (Fig. 5I-L). Significantly, cyclin D1 expression was greatly reduced in the palatal mesenchyme of the Tgfbr2^fl/fl mutant embryos at E14.5 when compared with the ones of wild-type control(Fig. 5M,N). The reduction of cyclin D1 expression was further confirmed by western and microarray analyses(Fig. 5 and data not shown). To rule out the possibility that palatal fusion itself was responsible for maintaining cyclin D1 expression in the CNC-derived palatal mesenchyme, we analyzed cyclin D1 expression in the palatal shelves both prior to and immediately after fusion (Fig. 5M and insert in M). Cyclin D1 expression remained in a similar pattern pre- and post-palatal fusion. We concluded that palatal fusion at E14.5 did not play a role in maintaining cyclin D1 expression in the CNC-derived palatal mesenchyme. We have also examined the expression of other cell cycle regulators (such as CDK4, CDK6, CDK inhibitors p21 and p18^INK4c) and found no significant difference between the wild-type and the Tgfbr2^fl/fl;Wnt1-Cre mutant samples(Fig. 5O,P, and data not shown). In addition, we have analyzed whether increased cell death might have contributed to compromised palatal shelf development in the mutant samples. TUNEL assay showed no difference in cellular apoptotic activity in the CNC-derived palatal mesenchyme between the Tgfbr2^fl/fl;Wnt1-Cre mutant and wild-type embryos (data not shown).

TGFβ signaling is known to regulate the expression of transcription factors which in turn may regulate the fate of CNC cells by controlling the progression of cell cycle (Moses and Serra, 1996; Han et al.,2003). Exogenous TGFβ can repress the transcriptional activity of the Msx1 gene in the palatal mesenchyme in vitro(Nugent and Greene, 1998). We examined the expression level of Msx1 in the developing palate by western analysis. Msx1 expression level was identical between the wild type and the Tgfbr2^fl/fl;Wnt1-Cre mutant samples at E13.5(Fig. 5Q). Significantly, Msx1 expression level was significantly elevated (2.5 times) in the palate of the Tgfbr2^fl/fl;Wnt1-Cre mutants when compared with the Msx1 expression level in the controls (Fig. 5).

During skull development, TGFβ ligand and its type II receptor are colocalized within the craniofacial mesenchyme and may regulate its differentiation (Fitzpatrick et al.,1990; Pelton et al.,1990; Lawler et al.,1994). A high level of TGFβ IIR mRNA expression is apparent in the meninges surrounding and covering the developing brain, suggesting an important functional role of this receptor in regulating the dura mater development (Wang et al.,1995). Recently, it was shown that CNC cells contribute to the formation of the meninges, which underlies the entire calvaria(Jiang et al., 2002). Remaining unclear is the functional significance of TGFβ signaling in regulating the development of the dura mater as well as the consequence of an impaired dura formation in regulating the patterning of intramembranous bone development.

By analyzing the Wnt1-Cre;R26R embryos, we found that the CNC-derived dura mater covered the entire surface of the developing brain in the wild-type sample at E14.5 (Fig. 7A, blue). In Tgfbr2^fl/fl;R26R;Wnt1-Cre mutant embryos, dura development was severely impaired on the surface of the developing brain (Fig. 7B). Specifically, instead of having a well-defined dura that contained blood vessels as seen in the wild-type samples, the Tgfbr2^fl/flmutant embryos showed a single cell layer, poorly developed dura mater(Fig. 7C,E). As shown in Fig. 4, there was no CNC migration defect in the Tgfbr2^fl/fl mutant embryos. This dura development defect resulted from severely impaired CNC cell proliferation activity in the Tgfbr2^fl/fl mutant embryos, while active CNC cell proliferation was observed in the dura of wild-type controls at E14.5(Fig. 7D,F). Although there was only a poorly defined dura in the Tgfbr2^fl/fl mutants at E14.5, it suggested that CNC cells were able to contribute to early dura development. However, there was a specific requirement for TGFβ signaling during the continued dura development. As craniofacial development continued,the impaired TGFβ signaling in the CNC-derived dura mater failed to induce parietal bone formation (rostral region), while there was proper parietal bone development in the wild type samples at E16.5(Fig. 7G,H). Eventually, the failure of inducing bone formation by the dura led to severely impaired calvaria development.

At birth, the Tgfbr2^fl/fl;Wnt1-Cre mutant mice showed severe skull defects, including a missing frontal and severely retarded parietal bone (with only the development of posterior border portion), as well as a smaller mandible and maxilla (Fig. 7I,J). The overall size of the skull of the Tgfbr2^fl/fl mutants was about 25% smaller than those of the wild-type littermates (Fig. 7K). As a result of compromised calvaria development, skeletal elements of the cranial base of the Tgfbr2^fl/fl mutant became visible when viewed from above (Fig. 7K). Tgfbr2^fl/fl mutation also affected the proper development of the mandible, with a dramatically reduced coronoid process and condyle, and a missing mandibular angle(Fig. 7L).

To date, most of the palate development studies have focused on the molecular regulation of the fate of midline epithelial cells during palatal fusion, while little is known about the molecular mechanism that controls the fate of CNC cells during palatogenesis(Kaartinen et al., 1997; Martinez-Alvarez et al.,2000). CNC fate determination is an important developmental event because successful migration, proliferation and differentiation of these pluripotent cells are crucial for normal craniofacial development. Here, we have investigated the molecular mechanism by which the Tgfbr2 gene regulates CNC-cell migration, proliferation and, ultimately, the formation of an aggregated cell mass prior to palatal bone formation during palatogenesis. By systematically following the lineage of CNC cells as they contribute to palate formation, our study shows that CNC cells contribute to the vast majority of the palatal mesenchyme and may possess crucial roles to regulate the epithelial-mesenchymal interaction during the extension and fusion of the palatal shelves. Evidently, the mesoderm-derived cells also contribute to the formation of the palatal mesenchyme. The dynamic distribution and close association between the CNC- and non-CNC-derived palatal mesenchyme suggest that these two cell populations may interact constantly throughout various stages of palatal development.

Until now, the function of TGFβ signaling in regulating the CNC-derived palatal mesenchyme is not well understood. TGFβ subtype expression is conspicuous in the cranial neural crest-derived mesenchyme during early mouse craniofacial development(Heine et al., 1987; Massague, 1990). The presence of TGFβ and its cognate receptors is obvious in the mesenchyme during crucial epithelial-mesenchymal interactions related to the formation of the palate, tooth, and Meckel's cartilage(Nugent and Greene, 1998; Hall, 1992; Chai et al., 1994; Wang et al., 1995; Lumsden and Krumlauf, 1996; Ito et al., 2002). Although the TGFβ type II receptor is strongly expressed in the CNC-derived palatal mesenchyme, mice deficient in Tgfbr2 die before the formation of the palate, making it impossible to investigate the functional significance of TGFβ signaling in regulating the fate of CNC cells during palatogenesis (Wang et al.,1995; Oshima et al.,1996). Our animal model of Tgfbr2 conditional gene ablation in the neural crest cells offers a unique opportunity to investigate the functional mechanism of TGFβ signaling in regulating the fate of the CNC-derived palatal mesenchyme. Owing to the lack of a CNC migration defect in Tgfbr2^fl/fl;Wnt1-Cre mutant mice, we conclude that TGFβ IIR is not crucial for the proper migration of CNC cells into the first branchial arch. The cell proliferation defect in the CNC-derived palatal mesenchyme of Tgfbr2^fl/fl;Wnt1-Cre mutant mice clearly indicates that TGFβ signaling is specifically required in the palatal mesenchyme prior to palatal fusion. We propose that TGFβ directly or indirectly regulates the expression of cell cycle regulators (such as cyclin D1) to control the progression of the cell cycle in CNC-derived palatal mesenchyme, and this regulation is crucial for proper palatal mesenchymal cell proliferation. Decreased palatal mesenchyme cell proliferation has resulted in compromised palatal shelf extension and failure of palatal fusion in Tgfbr2^fl/fl mutant mice. It is important to note that our animal model does not address whether TGFβ IIR regulates the non-CNC-derived palatal mesenchymal cell proliferation, because Wnt1-Cre does not cause Tgfbr2 deletion in this particular cell population. In addition, although cyclin D1 expression is significantly downregulated in the palatal mesenchyme of Tgfbr2^fl/flmutant mice, it is unlikely that a compromised cyclin D1 expression is directly responsible for causing the cleft palate defect in Tgfbr2^fl/fl mutant mice because cyclin D1-null mutant mice do not have cleft palate (Fantl et al.,1995). Other cycle regulators (such as CDK inhibitors p21 or p18^INK4c) appear to be unaffected when we compared their expression patterns within the palatal mesenchyme between the wild-type and the Tgfbr2^fl/fl mutant mice. Clearly, the method by which TGFβ signaling controls the progression of the CNC-derived palatal mesenchyme cell cycle during palatogenesis is complex; it will be the focus of our future studies.

Contrary to the successful fusion of our cultured Tgfbr2^fl/fl;Wnt1-Cre mutant palatal shelves, cultured Tgfb3-null mutant palatal shelves fail to fuse, even when they are placed in close contact in vitro(Kaartinen et al., 1997). Despite clear adherence, the cultured Tgfb3-null mutant palatal shelves show persistent MEE cells and intact basement membrane. Significantly,supplementation of exogenous TGFβ3 facilitates the successful fusion of Tgfb3-null mutant palatal shelves in vitro with transformation of the MEE and degradation of the underlying basement membrane. Clearly, TGFβ3 is specifically required in regulating the fate of MEE cells during palatal fusion. The successful signaling of TGFβ3 requires an integral TGFβreceptor complex. Indeed, TGFβ IIR is also expressed in the MEE prior to palatal fusion (Cui et al.,1998). Our palatal organ culture experiment suggests that the basic TGFβ signaling cascade in MEE cells is intact despite the null mutation of Tgfbr2 in the CNC-derived palatal mesenchyme. It also demonstrates that there is a cell-autonomous requirement for TGFβsignaling in the CNC-derived palatal mesenchyme during palatogenesis. In human clefting birth defects, failure of palatal fusion after proper palatal adhesion (such as the one in Tgfb3-null mutant mice) only represents a small percentage of the cleft palate cases, while failure of palatal shelf extension (such as the one in Tgfbr2^fl/fl;Wnt1-Cre mutant mice) is associated with the majority of the cleft palate cases. Hence, the Tgfbr2^fl/fl;Wnt1-Cre mutant mice will serve as an important animal model for the investigation of the molecular etiology of human cleft palate.

Defects in the development of the dura mater and calvaria bone have significant implications. A recent study has shown that the mammalian frontal bones are neural crest derived (still controversial for avian) and that the parietal bones are of mesodermal origin. Furthermore, the dura mater that underlies the parietal bones is neural crest-derived and is sensitive to retinoic acid exposure during parietal bone ossification, suggesting that intramembranous ossification of this mesodermal bone requires interaction with the CNC-derived meninges (Jiang et al.,2002). Here, the defects of both frontal and parietal bones suggest that the CNC-derived dura mater is crucial for the induction of CNC-derived frontal bone and mesoderm-derived parietal bone formation. We hypothesize that the dura mater produces inductive signaling which interacts with the overlaying mesenchyme, whether neural crest or mesodermally derived,to control the initiation and patterning of frontal and parietal bones during calvaria development. Furthermore, our study indicates that TGFβsignaling plays a pivotal role in regulating the proliferation of the CNC-derived dura mater. Aberrant TGFβ signaling results in compromised dura mater development and consequently, in calvaria development defects.

TGFβ is known to regulate the fate of multipotential progenitor cells instructively by regulating the expression or function of tissue-specific transcription factors (Moses and Serra,1996). For example, TGFβ downregulates the expression of homeobox gene Msx1 and affects cell fate determination in limb development (Ganan et al.,1996). The expression patterns of TGFβ and Msx1 have significant overlaps during palatal development and suggest an epistatic relationship between these genes when CNC-derived cells become committed to form the palatal mesenchyme (Pelton et al., 1990; Ferguson,1994). Overexpression of TGFβ suppresses transcriptional activity of the Msx1 gene in the palatal mesenchyme in vitro(Nugent and Greene, 1998). Similarly, TGFβ signaling may regulate the expression of the Msx2 gene during calvaria development. TGFβ IIR and Msx2 are co-expressed in the CNC-derived meninges prior to calvaria formation. We have shown here that Msx1 expression is significantly elevated while cyclin D1 expression is greatly reduced in the palatal mesenchyme of the Tgfbr2^fl/fl;Wnt1-Cre mutant embryos, suggesting that TGFβ may regulate the expression of the Msx1 gene, which in turn controls the progression of the CNC cell cycle during palatogenesis. A recent in vitro study has shown that Msx1 gene expression maintains cyclin D1 gene expression and controls cell cycle progression, thereby regulating terminal differentiation of progenitor cells during embryonic development(Hu et al., 2001). Our in vivo data suggests that the outcome of Msx1-regulated cyclin D1 expression might be tissue type-dependent. As suggested in the previous study, cyclin D1 is likely to be an indirect target of Msx1 during embryonic development. Furthermore, our study supports the previously proposed model that reconciles the observed phenotype similarities between the Msx1 loss- and gain-of-function mutations in the context of cell cycle regulation(Hu et al., 2001). In addition, mutations of the TGFβ IIR may also impinge on BMP signaling within the developing CNC and the CNC-derived mesenchyme, because there is significant overlap between the expression patterns of BMP and TGFβduring craniofacial development. TGFβ IIR can bind to BMPs, and the dominant-negative mutation of TGFβ IIR attenuates both BMP and TGFβsignaling (Massague, 1990; ten Dijke et al., 1994; Dumont and Arteaga, 2003). Potentially useful regionally restricted branchial arch and/or palatal mesenchyme markers (such as members of the homeobox-containing genes) need to be analyzed to dissect the TGFβ signaling cascade in regulating the fate of CNC cells during craniofacial morphogenesis.

The broad spectrum of phenotypic abnormalities suggests that TGFβsignaling is crucial for the transcriptional regulation of multiple regulatory signaling cascades during embryogenesis. We provide an animal model for investigating the molecular mechanism of cleft palate, calvaria agenesis and other CNC-related congenital malformations and demonstrate that TGFβ IIR signaling is specifically required in regulating the fate of CNC cells during craniofacial development. Future studies using this animal model will provide useful information on the mechanism of TGFβ IIR signaling in both normal and abnormal human development. In addition, genetic screening of the Tgfbr2 mutation among individuals with secondary palate cleft and skull malformations may provide crucial information in linkage analysis to investigate the etiology of congenital malformations.

We thank Andy McMahon and Phil Soriano for Wnt1-Cre and R26R mice, respectively. We also thank Hal Slavkin, Henry Sucov, Rob Maxson and Xun Xu for critical reading of the manuscript. This study was supported by grants from the National Institute of Dental and Craniofacial Research, NIH (DE12711 and DE14078) and the March of Dimes (6-FY02-137) to Yang Chai.