Teeth develop through a series of morphologically distinct stages, and like most organs, this process is controlled by sequential interactions between adjacent epithelial and mesenchymal tissue layers (Pispa and Thesleff, 2003). This progressive development is effected by soluble growth factors, and it is these reiterative signals that pass both within and between the tissue layers that ultimately control the position, shape, size, and structure of the tooth (Tucker et al., 1998; Thesleff and Mikkola, 2002). Finally, teeth develop roots, they erupt into the oral cavity, and an occlusion is established. If the developmental process is disrupted, anomalies in tooth number, morphology, structure, and position can occur.
Development of the mammalian secondary palate starts with palatal shelf primordia budding out from the maxillary processes at embryonic day 12 (E12) in the mouse. These bilateral projections elongate and grow down between the tongue and the floor of the mouth. They then elevate to a position above the dorsum of the tongue where they approximate and fuse in the midline (Ferguson, 1988). Fusion process involves adhesion of the palatal shelves, cell migration, apoptosis, and transdifferentiation of the medial edge epithelial cells (MEE) into mesenchymal cells (Ferguson, 1988; Fitchett and Hay, 1989; Shuler et al., 1992; Martinez-Alvarez et al., 2000; Cuervo et al., 2002). Interactions between the epithelium and mesenchyme control both early palatogenesis and the later fusion process (Tyler and Koch, 1977; Ferguson and Honig, 1984; Rice et al., 2004). Disturbances in any of these events can lead to a failure of the palate to fuse and result in a cleft. Cleft palate is a multifactorial condition not only caused by a genetic mutation but also by the interaction of environmental factors on the genetic background (Saxen, 1973).
Twist proteins are basic helix–loop–helix (bHLH) transcription factors that have been shown to have major roles during mesoderm differentiation, myogenesis, and epithelial mesenchymal transition (Bate et al., 1991; Baylies and Bate, 1996; Spicer et al., 1996; Yang et al., 2004). Mutations in TWIST cause Saethre–Chotzen syndrome (OMIM 101400), characterized by premature fusion of the calvarial sutures (craniosynostosis), a narrow or cleft palate, and limb defects (El Ghouzzi et al., 1997; Howard et al., 1997; Paznekas et al., 1998). Dental anomalies, including abnormal crown and root development have been reported in patients with Saethre–Chotzen syndrome (Goho, 1998). These mutations are largely deletions or nonsense mutations. Homozygous loss of Twist in mice leads to a failure of the cranial neural folds to fuse and to defects in head mesenchyme, somites, and limb buds. Twist−/− mice die at E11.5 before the onset of tooth and calvarial development (Chen and Behringer, 1995). Mice heterozygous for Twist survive until adulthood and exhibit limb and calvarial phenotypes reminiscent of Saethre–Chotzen syndrome (Bourgeois et al., 1998; Carver et al., 2002). Like Twist, Fibroblast growth factor receptors 1, 2, and 3 (Fgfr1–3) are expressed during calvarial suture development (Rice et al., 2000, 2003) and mutations in FGFR1, 2, and 3 have been found to cause human syndromes characterized by craniosynostosis (Wilkie and Morriss-Kay, 2001). We have shown previously that FGF2 up-regulates the expression of Twist and that Fgfr2 distribution is altered in the developing calvaria of Twist+/− mice. These findings suggest Twist to be part of the FGF signalling pathway during osteogenesis (Rice et al., 2000).
Inhibitors of differentiation (Ids, also called inhibitors of DNA binding) are HLH transcription factors that lack a functional DNA binding domain. They preferentially dimerize with E proteins thus preventing other HLH transcription factors, such as Twist, binding to them and forming functional heterodimers. Thus, they act dominant-negatively (Benezra et al., 1990; Pesce and Benezra, 1993). Ids have multiple roles during development and tumorigenesis, including regulation of cell lineage commitment and cell fate decisions, cell growth, differentiation, proliferation, and death (Ruzinova and Benezra, 2003). There are four Id family members that are expressed at multiple sites during development (Jen et al., 1996). In the craniofacial region, Id1 is expressed in the first and second branchial arches, Rathke's pouch, and during calvarial bone and cartilage development (Rice et al., 2000). Id1−/− and Id3−/− single knockout mice are viable and have no reported craniofacial defects. However, Id1−/−; Id3−/− double mutants die at E12.5, indicating redundancy between these two proteins (Lyden et al., 1999). Ids are downstream targets of bone morphogenetic protein (BMP) signalling (Hollnagel et al., 1999), and in the calvarial mesenchyme, Id1 expression is up-regulated by BMP2 (Rice et al., 2000). Interestingly, Id1+/−Id3+/− mice have widened calvarial sutures and exhibit a suppression of BMP-induced bone formation (Maeda et al., 2004).
Snail family of zinc-finger transcription factors are thought to act as transcriptional repressors. They contain E box binding sites identical to those in bHLH transcription factors, suggesting that they compete for the same target genes (Nieto, 2002). Snail proteins induce during mesoderm specification and emergence of neural crest from the neural tube. Snail may also play a role in EMT during fusion of the palatal shelves (Martinez-Alvarez et al., 2004). Twist has also been associated with EMT (Yang et al., 2004). Snail and Twist have been shown to induce invasion and spread of breast tumors (Nieto et al., 1994; LaBonne and Bronner-Fraser, 2000; Blanco et al., 2002). However, in contrast to Twist, which promotes tumor cell metastasis with no apparent growth effect on the primary tumor, Snail attenuates tumor proliferation and produces a change in cell shape (Vega et al., 2004).
In this study, we investigated the role of Twist, Snail, and Id during tooth and palate development. We performed a detailed analysis of Twist, Snail and Id1 mRNA expression during tooth and palate development. Twist and Snail expression was restricted to mesenchymal domains, whereas Id1 was expressed both in epithelial and mesenchymal domains. As the mesenchymal components of both the palate and teeth are derived from neural crest cells (Tan and Morriss-Kay, 1986; Imai et al., 1996; Francis-West et al., 2003), we decided to investigate whether upstream regulatory mechanisms for these genes are conserved between these two tissues. We show that the mesenchymal Twist and Snail expression is regulated by epithelially derived FGFs during tooth and palate development: organ culture experiments using recombined dental mesenchyme and dental epithelium showed that the mesenchymal expression of Twist and Snail was induced by the epithelium and that this epithelial induction could be mimicked in isolated mesenchymal explants by exogenous FGF4. Similarly, the palatal epithelium induced Twist expression in the palatal mesenchyme in recombined palatal cultures, and in isolated palatal mesenchyme explants, exogenous FGF2 was sufficient to induce expression of both Twist and Snail. We also demonstrate that Id1 expression was regulated by BMP signalling in both tooth and palate. FGF and BMP signalling are known to act antagonistically during tooth development (Neubuser et al., 1997). We found the responses of Twist, Snail, and Id1 to FGFs and BMPs to be the same in the developing tooth and palate. And as Ids are known to inhibit transcription factors such as Twist and Snail, we suggest that FGF-regulated Twist and Snail and BMP-regulated Id1 may mediate antagonistic effects of FGF and BMP signalling during both tooth and palate development through a common mechanism.
Localization of Twist, Id1, and Snail mRNA During Murine Tooth Development
We examined the expression patterns of Twist, Id1, and Snail during molar tooth development by in situ hybridization. The mouse molar tooth develops through a series of morphologically distinct stages. This process starts with a thickening of the dental lamina at E11 and progresses through bud (E12–E13.5) and cap (E14.5–E15.5) stages to bell stage at E16.5. We focused our studies at these stages using frontal (E11.5–E14.5) and parasagittal head sections (E16.5). At E11.5 Twist was expressed at a minimal level in the mesenchyme immediately adjacent to the thickened dental epithelium (Fig. 1A). At E13.5, Twist transcripts were detected throughout the condensing dental papilla, adjacent to the dental epithelium. Intense expression in the dental papilla and future follicle continued through the bud stage (Fig. 1B). At the cap stage, E14.5, Twist expression formed a gradient from the dental papilla to the dental follicle so that the highest level of expression was seen adjacent to the inner enamel epithelium and the least amount of expression was seen in the dental follicle (Fig. 1C). At E16.5 (bell stage), expression was restricted to the dental papilla mesenchyme in the region of the cusps.
Unlike Twist, Id1 was not only expressed in the dental mesenchyme but also in the epithelial compartment. Epithelial expression started at E11.5 when the tooth morphogenesis commences (Fig. 1E). At E13.5, Id1 continued to be expressed in the dental epithelium and was now also expressed in the dental mesenchyme. Expression of Id1 in both the dental epithelium and mesenchyme continued throughout dental development. The epithelial expression of Id1 became more restricted to the primary, secondary, and tertiary enamel knots (Fig. 1G,H). The enamel knots are signalling centers that control the change in the shape from a flat structure into characteristic tooth shape with distinctive cusps and fossas (Jernvall et al., 1998).
Snail transcripts were localized in the dental mesenchyme throughout development (Fig. 1I–L). During the bud, cap, and bell stages, this finding was similar to that of Twist but was seen in a more diffuse pattern.
Localization of Twist, Id1, and Snail mRNA During Murine Palate Development
We used tissue sections at the different stages of palatogenesis: before the budding of the palatal shelves from the maxillary processes (E11.5), before shelf elevation (E13.5), after elevation but before fusion (E14.5 anterior), and during the fusion process (E14.5 central region). As fusion of the palatal shelves starts in the central region of the palate and progresses in both posterior and anterior directions, sections through the anterior region are slightly delayed in development compared with those in the central region. Twist was intensely expressed in the palatal mesenchyme adjacent to the epithelium at all ages studied (Fig. 2A–D). At E11.5, Twist transcripts were mainly located on the medial side on the palatal bud (Fig. 2A). At E13.5, Twist expression was adjacent to the future nasal epithelium, MEE, and oral epithelium (Fig. 2B). Expression was particularly intense just before and during the fusion process (Fig. 2C,D).
Id1 transcripts were detected in both the palatal mesenchyme and epithelium. At E11.5, Id1 was detected in the mesenchyme on the future nasal side of the palatal process. Subsequently, mesenchymal expression intensified at the tip of the palatal shelves adjacent to the MEE (Fig. 2F–H). Id1 was also expressed in the oral epithelium during palate development (Fig. 2E–H). Like Twist and Id1, Snail was also expressed in the palatal mesenchyme just before and during palatal fusion (Fig. 2J–L). Thus, expression patterns of all three genes were spatially and temporally overlapping, indicating possible molecular interactions.
Dental Epithelium Induces Twist and Snail and This Induction Can Be Mimicked by Epithelially Derived Fgfs
As Twist and Snail were expressed in the dental mesenchyme in close proximity to the epithelium, we decided to test whether Twist and Snail are regulated by the epithelium. In tissue culture, we recombined isolated dental mesenchyme and epithelium. To test whether Twist and Snail are autonomously regulated within the mesenchyme, we also maintained isolated dental mesenchyme in culture without epithelium. Whole-mount and sectioned tissues were then probed for Twist and Snail. The dental epithelium was found to induce both Twist and Snail in the adjacent mesenchyme (Twist 10/10 explants; Snail 7/9 explants; Fig. 3A,E). In isolated dental mesenchyme Twist and Snail expression were not maintained (Fig. 3B and data not shown).
The effect of the epithelium could be mimicked by placing FGF4-impregnated beads onto isolated dental mesenchyme (Twist 10/11 explants; Snail 8/8 explants; Fig. 3C,G). Fgf4 is normally expressed by the dental epithelium. Mesenchymally derived FGF10 could induce neither Twist nor Snail (data not shown). Bovine serum albumin (BSA) beads did not induce Twist or Snail (Fig. 3D,H).
BMP4 Induces Id1, Msx1, and Msx2 in Isolated Dental and Palatal Mesenchyme
Several BMPs are expressed during tooth development (Aberg et al., 1997). Notably, Bmp2, 4, and 7 are expressed in the enamel knot and inner enamel epithelium at cap stage. We have shown previously that exogenous BMP4 will induce Id1 in the developing calvarial mesenchyme (Rice et al., 2000), and here we demonstrate that BMP4 also induces Id1 in both the developing dental and palatal mesenchyme (tooth 8/8 explants; palate 4/5 explants; Figs. 3I,J, 4J). Neither FGF4-impregnated nor BSA-impregnated beads induced Id1 (Fig. 3J,K). We also have shown that BMP4 up-regulates Msx1 and Msx2 in dental mesenchyme (Vainio et al., 1993). In this study, we show that BMP4 similarly up-regulates Msx1 and Msx2 in isolated palatal mesenchyme (Msx1 4/5 explants; Msx2 5/5 explants; Fig. 4I,K,L). Also, BMP4 did not up-regulate Twist or Snail (Fig. 4F and data not shown).
Palatal Epithelium Induces Twist, and Exogenous FGF2 Induces Twist and Snail in Isolated Palatal Mesenchyme
Tissue recombination and bead assay experiments gave similar results in the palate as we found in the tooth. The palatal epithelium induced Twist in the palatal mesenchyme (6/8 explants; Fig. 4C). Exogenous FGF2 induced Twist and Snail in the palatal mesenchyme (Twist 7/9 explants; Snail 7/7 explants; Fig. 4A,B,D,E).
Using two developmental models, we have shown that Twist, Snail, and Id1, three interacting transcription factors, are regulated by FGFs and BMPs in a common regulatory network. Twist and Snail are expressed in the palatal and dental mesenchyme immediately adjacent to the epithelium. By placing isolated dental or palatal epithelia onto isolated mesenchymal explants, we show that mesenchymal expression of Twist and Snail is regulated by the overlying epithelium. We also show that epithelial FGFs can mimic this effect. Id1, which is regulated by BMP signalling, is expressed in a pattern that overlaps the expression of Twist and Snail. As Id1 acts in a dominant-negative manner to Twist and Snail, it may limit the action of Twist and Snail during tooth and palate morphogenesis.
FGF and BMP Regulate Twist, Snail, and Id1 During Tooth Development
During tooth morphogenesis, FGFs regulate reciprocal interactions between the adjacent epithelium and mesenchyme. Fgf4, 8, and 9 are expressed in the dental epithelium. Fgf8 has been implicated as an early epithelial signal that regulates the expression of mesenchymal transcription factors (Neubuser et al., 1997), and conditional deletion of its function in branchial arch epithelium arrests tooth development at the initiation stage (Trumpp et al., 1999). The epithelial Fgfs regulate proliferation both in the dental epithelium and mesenchyme, whereas Fgf10 produced in the mesenchyme stimulates proliferation only in the epithelium through its receptor Fgfr2b (which is located in the epithelium; Kettunen et al., 2000). FGFs and BMPs antagonistically regulate the function of many genes during tooth development, including Pax9, Barx1, Pitx2, and L-Fng: FGFs stimulate these transcription factors, and BMPs inhibit them (Neubuser et al., 1997; Tucker et al., 1998; St. Amand et al., 2000; Mustonen et al., 2002). As we found that FGF2 and FGF4 stimulated Twist and Snail, and BMP induced Id1 but not Twist or Snail, and Id1 is known to inhibit Twist and Snail, we suggest that the antagonistic effects of FGFs and BMPs may be mediated by Twist, Snail, and Id1.
BMPs and FGFs are known to control Msx1 expression in many developing organs, including the tooth (Vainio et al., 1993; Alappat et al., 2003). Mutations in Msx1 have been found, in both mice and humans, to cause a cessation of tooth development and cleft palate (Satokata and Maas, 1994; Vastardis et al., 1996). Both these phenotypes can be rescued in mice by overexpressing Bmp4, which is a downstream target of Msx1 (Zhao et al., 2000; Zhang et al., 2002). Thus, not only is Msx1 activated by BMP4, but Msx1 regulates Bmp4 expression in a regulatory loop.
We have shown previously that FGF2 will up-regulate Twist in embryonic calvarial explants (Rice et al., 2000), and Montero et al. (2001) have shown that FGF will up-regulate Snail and decrease apoptosis in the limb bud. In the limb, it seems that the response of Snail to exogenous FGF2 or 8 is rapid, within 1 hr, when compared with the up-regulation of Twist (Isaac et al., 2000). In this study, we show that both isolated epithelia and exogenous FGF2 or 4 can up-regulate Twist and Snail in isolated mesenchymal explants. Both Twist−/− and Snail−/− mice die before the start of tooth or palate morphogenesis. However, abnormal molar tooth morphology has been reported in Twist+/− mice (Bourgeois et al., 1998), and dental abnormalities are seen in patients with Saethre–Chotzen syndrome, which is caused by haploinsufficiency of TWIST. These abnormalities include multiple peg-shaped teeth, broad crowns, and thin roots (Marchesi and Leoni, 1993; Goho, 1998). Narrow, high-arched or cleft palate are also features of Saethre-Chotzen syndrome (Gorlin et al., 2001).
Twist, Snail, and Id1 May Have Multiple Roles During Palate Development
Until recently, there has been little evidence that epithelial–mesenchymal interactions play a role during early palate development before shelf elevation. It is now known that Fgf10 from the palatal mesenchyme signals to its receptor Fgfr2b in the adjacent epithelium where one of the target genes of Fgf10 signalling is Sonic hedgehog (Shh;Rice et al., 2004). Shh from the epithelium acts on the mesenchyme to induce Bmp2 and cell proliferation (Zhang et al., 2002; Rice et al., 2004). This Fgf10/Shh/Bmp2 signalling effects morphological change in the palatal shelves as they bud and grow from the maxillary processes. Disruption of any of these interactions causes cleft palate. In this study, using epithelial–mesenchymal dissociation and recombination experiments, we demonstrate that the epithelium regulates the expression of Twist in the mesenchyme during palate development and that epithelial FGFs can mediate this induction.
Later in palate development, the palatal shelves join, forming a barrier between the oral and nasal cavities. Snail and Twist have been implicated in this fusion process and our data, that Snail and Twist are intensely expressed in the palatal mesenchyme adjacent to the MEE immediately before and during the fusion process, would support this implication (Bloch-Zupan et al., 2001; Martinez-Alvarez et al., 2004). Snail, the closely related Slug and Twist are all known to induce EMT at many sites during development and invasive tumorigenesis (Nieto, 2002; Yang et al., 2004). Misexpression of Snail in the palatal epithelial compartment is responsible, in part, for the cleft palate seen in Tgfβ3−/− mice (Martinez-Alvarez et al., 2004).
It has been demonstrated in many biological systems, including this study, that BMPs can up-regulate Id1 gene expression. Of interest, TGFβs have also been shown to regulate Ids during development (reviewed in Ruzinova and Benezra, 2003). The expression domains of Tgfβ1, 2, and 3 overlap with that of Id1 during tooth and palate development (Fitzpatrick et al., 1990; Pelton et al., 1990; Vaahtokari et al., 1991). Thus, Id1 may perform a pivotal role (1) to control the function of BMP and TGFβs and (2) to control FGF-regulated Twist and Snail function through its dominant-negative activity. BMP-activated Ids can cause cells to switch fate, as well as induce cell migration and cell proliferation (Ruzinova and Benezra, 2003). All these processes have been suggested to play a role during palatal fusion.
A Common Regulatory Loop Between Twist, Snail, and Id1 During Palate and Tooth Formation
We have investigated how FGF and BMP signalling may regulate cell and tissue interactions during tooth and palate morphogenesis. We show that Snail and Twist expression in the palatal and dental mesenchyme is regulated by the oral epithelium and that exogenous FGF can replace this epithelial action. Our data show that BMP4 induces Id1 expression, and we suggest that as Id1 may dominant-negatively compete with Twist and Snail, this may be one route by which TGFβ superfamily members may regulate the functions of Twist and Snail. In addition, Twist is known to repress bHLH transcription factors such as Snail, and Snail is known to up-regulate Twist in Xenopus neural crest cells, thus completing the feedback loop (Spicer et al., 1996; Aybar et al., 2003).
Humans and mice with mutations in MSX1 exhibit hypodontia and cleft palate. We have shown previously in the tooth model that FGF2 and BMP4 will up-regulate Msx1 (Vainio et al., 1993; Kettunen et al., 1998), and in this study, we show that a similar signalling exists in the developing palate. Without functioning teeth or palate, most mammals fail to survive. It is not surprising, therefore, that Twist and Snail may be regulated by more than one genetic pathway to ensure that tight developmental control is maintained.
Preparation of the Tissues
Whole heads of mice (NMRI or CD1) between embryonic day 11 and embryonic day 16 (E11.5–E16.5) were dissected in Dulbecco's phosphate buffered saline (pH 7.3) under a stereomicroscope. Noon on the day the vaginal plug was found was calculated as E0.5. After overnight fixation in 4% paraformaldehyde in PBS at 4°C, the tissues were dehydrated in an ethanol series, stained with eosin to aid tissue orientation in paraffin, treated with xylene, and embedded in paraffin. Sections of 7 μm were cut and mounted on triethoxysilane treated slides, dried overnight at 37°C, and stored at 4°C. After overnight fixation, tissues were dehydrated in a methanol series and stored in absolute methanol at −20°C.
Tissue Separation and Recombination Cultures
Tissue separation and recombination culture experiments in the tooth and palate have been described previously (Vainio et al., 1989; Rice et al., 2004). Briefly, E13.5 lower molar teeth or palatal shelves were dissected free from the mandible and maxilla, respectively. After pancreatic-trypsin digestion, the epithelia were separated from the mesenchyme. Pancreatic-trypsin enzymatic digestion took 5–10 min (teeth) 35 min (palates) on ice then the tissues were allowed to “recover” for 25 min in medium with fetal bovine serum on ice, before the tissues were used. Isolated mesenchymal tissues were placed on filters in Trowell-type organ culture and then epithelia recombined.
Heparin-coated acrylic beads were incubated in recombinant human FGF2, FGF4, FGF10 (100 ng/μl, R&D Systems), or in BSA at 37°C for 40 min and stored at 4°C before being placed on the explant. Bead assays were cultured for 24 hr. Similarly, Affi-gel agarose beads were impregnated with human BMP4 (100 ng/μl, R&D Systems).
In Situ Hybridization
Preparation of Twist and Id135S- and digoxigenin-labeled riboprobes, in situ hybridization, and image analysis have all been described previously (Vaahtokari et al., 1996; Kettunen et al., 1998, 2000; Rice et al., 2000; Gritli-Linde et al., 2001). Antisense Snail probe (a kind gift from Dr. MA Nieto) was generated from pGEM-T by linearizing with SalI and transcribing with RNA polymerase T7.
We thank Kaija Kettunen, Merja Mäkinen, Heide Olsen, and Riikka Santalahti for their excellent technical assistance. D.R. was funded by the MRC, I.T. was funded by the Academy of Finland, and R.R. was funded by the Finnish Cultural Foundation.