|Home | About | Journals | Submit | Contact Us | Français|
Reciprocal interactions between epithelium and mesenchyme are crucial for embryonic development. Fibroblast growth factors (FGFs) are a growth factor family that play an important role in epithelial–mesenchymal tissue interaction. We have generated epithelial-specific conditional knockout mice targeting Fibroblast growth factor receptor 2 (Fgfr2) to investigate the function of FGF signaling during craniofacial development. K14-Cre;Fgfr2fl/fl mice have skin defects, retarded tooth formation, and cleft palate. During the formation of the tooth primordium and palatal processes, cell proliferation in the epithelial cells of K14-Cre;Fgfr2fl/fl mice is reduced. Thus, FGF signaling via FGFR2 in the epithelium is crucial for cell proliferation activity during tooth and palate development.
Fibroblast growth factor (FGF) is one of the growth factors that are involved in intercellular signaling during embryogenesis (Thesleff and Sharpe, ’97; Hogan, ’99). To date, 22 FGF ligands and four FGF receptors have been identified (Ornitz and Itoh, 2001; Itoh and Ornitz, 2004). FGFR1, 2, and 3 contain three extracellular, immunoglobulin-like (Ig-like) domains, an acid box (between Ig-like I and II), a single transmembrane domain, and paired intracellular tyrosine kinase domains. Alternative mRNA splicing of the Fgfr gene specifies the sequence of the carboxyl-terminal half of Ig-domain III, called either IIIb or IIIc (Miki et al., ’92; Chellaiah et al., ’94; Naski and Ornitz, ’98). Exon IIIb is expressed in epithelial lineages, whereas exon IIIc tends to be expressed in mesenchymal lineages (Avivi et al., ’93; Scotet and Houssaint, ’95).
FGF signaling is involved in craniofacial development such as that of the craniofacial bone, palate, salivary gland, tooth, and craniofacial muscle (Nie et al., 2006). During tooth development, Fgfr2b (IIIb) is mainly expressed in the dental and palatal epithelia and Fgfr2c (IIIC) is expressed in the dental mesenchyme (Kettunen et al., ’98). FGF10, which is the ligand for FGFR2b, is expressed in the dental and palatal mesenchyme (Harada et al., 2002; Rice et al., 2004). Moreover, the phenotypes of Fgf10 conventional knockout mice are consistent with those of Fgfr2b mice including retardation of limbs, lungs, salivary glands, and thymus (Sekine et al., ’99; De Moerlooze et al., 2000; Ohuchi et al., 2000). These observations imply that reciprocal interactions between mesenchymal and epithelial tissue play a role in this process.
There are exceptions to these patterns of receptor expression that occur in tooth development. For instance, Fgfr2b is expressed in both ameloblasts and odontoblasts at the newborn stage, and Fgfr2c is expressed in the inner dental epithelium at the bell stage (E16-17) (Kettunen et al., ’98). Furthermore, in the absence of Fgfr2b, Fgf10 can bind to other isoforms of the FGF receptor, such as FGFR1b (Beer et al., 2000). To examine these discrepancies, we created mice lacking all isoforms of Fgfr2 specifically in the epithelium (K14-Cre;Fgfr2fl/fl). K14-Cre;Fgfr2fl/fl mice have defects in skin, teeth, plate, and hindlimbs, but not forelimbs. During tooth and palatal development, epithelial-specific loss of Fgfr2 causes defects in cell proliferation in epithelial of tooth and palate. Organization of the rugae of the palate, which is considered a landmark of palatal development (Cui et al., 2005), is also compromised in K14-Cre;Fgfr2fl/fl mouse.
All animal studies were performed according to IACUC guidelines. K14-Cre transgenic mice have been described previously (Indra et al., 2000; Andl et al., 2004). We crossed K14-Cre;Fgfr2fl/+ with Fgfr2fl/fl mice to generate K14-Cre;Fgfr2fl/fl null alleles that were genotyped using PCR primers as previously described (Yu et al., 2003).
The R26R conditional reporter allele has been described previously (Soriano, ’99). We mated K14-Cre and R26R mice to generate K14-Cre;R26R embryos. Detection of β-galactosidase activity in whole embryos (E10.5) was carried out as previously described (Chai et al., 2000).
Whole skeletal preparations of newborn mice were prepared and stained with Alizarin Red and Alcian Blue as previously described (McLeod, ’80).
Tissues were fixed in 4% paraformaldehyde in phosphate buffered saline, decalcified in Decalcifying solution (Richard-Allan Scientific, Kalamazoo, MI), paraffin embedded, sectioned at 5 mm, and stained with hematoxylin/eosin.
A 10 mg/mL stock of bromodeoxyuridine (BrdU; Sigma, St Louis, MO) was injected intraperitoneally into mice at E13.5 at a dose of 100 μg BrdU per gram of body weight. Mice were sacrificed 1 hr after injection and bones were harvested. BrdU staining was carried out on paraffin sections by using a BrdU staining kit according to the manufacturer's directions (Zymed, South San Francisco, CA).
Whole-mount and sectioned in situ hybridizations were performed according to standard procedure (Wilkinson, ’98). Several negative controls (sense probe and no probe) were run in parallel with the experimental reaction. RNA probes were generated as reported previously: Fgfr2b (Sasaki et al., 2006) and Shh (Hui and Joyner, ’93).
We generated an epithelial-specific gene deletion of Fgfr2 in mice using Cre recombination driven by the epithelial-specific K14 promoter (Andl et al., 2004). K14-Cre;Fgfr2fl/1 mice were indistinguishable from their wild-type littermates (Fig. 1A and C). We crossed K14-Cre;Fgfr2fl/+ mice with Fgfr2fl/fl mice and recovered approximately 25% K14-Cre;Fgfr2fl/fl mice, implying that no significant lethality occurred before birth. K14-Cre;Fgfr2fl/fl mice died within 24 hr of birth without feeding, likely owing to their cleft palate (Fig. 2). The eyes of K14-Cre;Fgfr2fl/fl mice were open owing to the absence of eyelids, and they had shiny skin (Fig. 1B). We detected limb dysgenesis in the hindlimb but not in the forelimb (Fig. 1A–D). Bone staining of K14-Cre;Fgfr2fl/fl mice revealed no bones in the lower limb except the pelvic girdle (Fig. 1E and F).
To confirm the tissue-specific deletion of Fgfr2 gene expression, we performed in situ hybridization. The epithelium and primordium of the hair follicle expressed Fgfr2b at E12.5 (Fig. 1G, arrow). In contrast, Fgfr2b expression was not detectable in K14-Cre;Fgfr2fl/fl mice (Fig. 1H), consistent with successful epithelial-specific loss of Fgfr2b. Interestingly, K14-Cre;Fgfr2fl/fl mice had limb defects only in the hindlimb, although Fgfr2b and Fgf10 conventional knockout mice form neither fore- nor hindlimbs. To confirm the timing of Cre recombination under the control of the K14 promoter, we crossed K14-Cre mice with Rosa26R mice (Soriano, ’99) and performed X-gal staining of the embryo. E10.5 samples clearly indicated that Cre recombination occurred earlier in the hindlimb than in the forelimb (Fig. 1I). At later stages, Cre recombination was detectable in all epithelial cells (Fig. 1J). This time gap might be a critical reason that the forelimb escaped having defects in formation in K14-Cre;Fgfr2fl/fl mice. We also confirmed the expression pattern in the oral epithelium of the K14 promoter of our construct. K14-Cre;R26R mice exhibited mosaic X-gal staining of the first branchial arch at E10.5 (Fig. 1I, arrowhead) (Chai and Maxson, 2006). At E11.5, we detected strong X-gal staining in the oral epithelium (Fig. 1J). At E12.5, the dental and palatal epithelium are strongly positive for X-gal staining (Fig. 1K) (Xu et al., 2006, 2008).
Newborn K14-Cre;Fgfr2fl/fl mice lacked teeth (data not shown). At E12.5, the lamina stage, we detected thickened epithelium in the presumptive tooth development area of K14-Cre;Fgfr2fl/fl mice, which was indistinguishable from that of control mice (Fig. 2A and B). At E13.5, the teeth of control mice had reached the bud stage, but the dental epithelium of K14-Cre;Fgfr2fl/fl mice still appeared to be in the lamina stage (Fig. 2C and D). At E14.5, control mice had reached the cap stage (Fig. 2E), but tooth development remained retarded in K14-Cre;Fgfr2fl/fl mice (Fig. 2F).
To investigate possible mechanisms of this tooth retardation, we analyzed gene expression of candidates expressed in the dental epithelium. Fgf8 is one of the molecules indicating the onset of tooth development (Zhang et al., 2003). Fgf8 was expressed in the presumptive tooth region in control mice at E11.5, whereas it was dramatically reduced in K14-Cre;Fgfr2fl/fl mice, even though Fgf8 expression in the nasal pit area was indistinguishable in K14-Cre;Fgfr2fl/fl and control mice (Fig. 2G and H, open arrowheads). Pitx2, another marker for tooth initiation (Chai and Maxson, 2006), was unchanged in K14-Cre;Fgfr2fl/fl mice relative to control (data not shown).
Next, we analyzed cell proliferation activity in the lamina and bud stages. At E12.5 (lamina stage), cell proliferation activity of K14-Cre;Fgfr2fl/fl mice was indistinguishable from that of control mice in both epithelial and mesenchymal cells (data not shown). At E13.5 (bud stage), we detected a reduction of cell proliferation activity in dental epithelial cells of K14-Cre;Fgfr2fl/fl mice compared with those of control mice, although there was no change in proliferation in dental mesenchymal cells (Fig. 2I–K).
We observed complete cleft palate in 75% of K14-Cre;Fgfr2fl/fl mice (Fig. 3B), and the remaining 25% displayed submucous cleft palate at the newborn stage. We were unable to detect rugae formation in K14-Cre;Fgfr2fl/fl mice, although rugae were clearly visible on the hard palate of newborn mice (Fig. 3A, black arrows, and and3B).3B). To confirm the presence or absence of rugae, we examined Shh expression, which is a marker for rugae formation (Sasaki et al., 2007), using whole mount in situ hybridization. We detected seven rugae on the secondary palate of control mice at E15.5, based on Shh expression (Fig. 3C). In contrast, Shh expression appeared dotted on the secondary palate of K14-Cre;Fgfr2fl/fl mice, implying that the rugae were not well organized (Fig. 3D). Interestingly, hair follicle formation in the nose in K14-Cre;Fgfr2fl/fl mice was well organized when compared with control mice (Fig. 3C and D). Histological analysis demonstrated that the palatal shelves of K14-Cre;Fgfr2fl/fl mice transformed from vertical to horizontal positions, but the left and right sides of the palatal shelves did not contact each other (Fig. 3F). We hypothesized that cleft palate was owing to insufficient palatal shelf growth, which might be the result of a reduction of mesenchymal cell proliferation. We did detect a reduction in cell proliferation in K14-Cre;Fgfr2fl/fl palatal epithelium at E13.5. We conclude that the reduction of epithelial cell caused the cleft palate in K14-Cre;Fgfr2fl/fl mice.
Furthermore, we investigated whether morphological changes of the palatal shelf occurred before shelf elevation owing to the reduction of epithelial cell proliferation. At E13.5, the size of the palatal shelves of K14-Cre;Fgfr2fl/fl mice was reduced from anterior to posterior when compared with that of control mice (Fig. 4A–D). The shape of the palatal shelves in K14-Cre;Fgfr2fl/fl mice was also compromised. The palatal shelves of K14-Cre;Fgfr2fl/+ (control) mice formed an asymmetric, fingerlike shape (Fig. 4A and C), whereas this asymmetry was largely lost in K14-Cre;Fgfr2fl/fl mice (Fig. 4B and D).
In this study, we examined the biological significance of FGF signaling through Fgfr2 in epithelial cells using a tissue-specific deletion of the Fgfr2 gene. Through the analysis of conditional knockout (K14-Cre;Fgfr2fl/fl) mice, we conclude that the FGF signal pathway is crucial for cell proliferation activity in epithelial cells and is involved in the organization of the rugae, the thickened lines on the secondary palate.
Previous studies have reported the expression pattern of FGF ligands and their receptors during embryonic development. There are multiple FGF receptor isoforms owing to RNA splicing, and these isoforms have tissue-specific expression patterns. For instance, Fgfr2 IIIb is expressed in the epithelium and Fgfr2 IIIc is expressed in mesenchymal cells, although exceptions exist at different developmental stages (Kettunen et al., ’98; Rice et al., 2004). These observations suggest that Fgfr2b conventional knockout mice may be highly similar to epithelial tissue-specific conditional knockout mice. Surprisingly, K14-Cre;Fgfr2fl/fl and Fgfr2b-/- mice exhibit some phenotypes that differ as well as common defects in the eye and skin. First of all, the severity of the limb formation defect is significantly different. Fgfr2b-/- mice exhibited failure of limb initiation (De Moerlooze et al., 2000), whereas K14-Cre;Fgfr2fl/fl mice showed normal forelimb development and compromised hindlimb development. X-gal staining of K14-Cre;R26R mice demonstrates that Cre recombinase derived from the K14 promoter is active around E10.5 in the hindlimb but not in the forelimb, implying that epithelial cells in the forelimb escape the deletion of Fgfr2 at early developmental stages. This observation in K14-Cre;Fgfr2fl/fl mice is consistent with the previous finding that Fgfr2b plays a critical role in the apical endodermal ridge (AER) during limb formation (Revest et al., 2001).
Fgfr2b null mice exhibited tooth defects and cleft palate (De Moerlooze et al., 2000), but the mechanism of the tooth defect has remained unknown. In this study, we conclude that the tooth defect in the absence of Fgfr2 is likely the result of a cell proliferation defect in the dental epithelium. Although Fgf8 is not the ligand for Fgfr2 (Ornitz et al., ’96), Fgf8 expression is reduced in K14-Cre;Fgfr2fl/fl mice in the presumptive tooth region, implying that Fgf8 is a downstream target gene of FGF signaling through FGFR2. We have recently shown that the antagonistic interaction between BMP4 and Fgf8 and the setup of proximal vs. distal domains of the first branchial arch epithelium is dependent upon Smad4-mediated BMP signaling in the neural crest-derived mesenchyme (Ko et al., 2007). Interestingly, however, the induction of Fgf8 expression in the first branchial arch epithelium is not dependent on the neural crest cells. For example, Fgf8 expression is unaffected in Msx1 null or Msx1/Msx2 double null mice (Bei and Maas, ’98; Ishii et al., 2005). This study provides crucial evidence that FGF signaling through Fgfr2 is required for the induction of Fgf8 in the first branchial arch epithelium during craniofacial development. In spite of the reduction of Fgf8 expression, tooth development is normal until E12.5, implying there is likely compensation by other molecules. Pitx2 expression in K14-Cre;Fgfr2fl/fl mice is indistinguishable from control mice. Recently, it has been reported that Pitx2 cooperates with Lef-1 for activation of β-catenin during tooth initiation (Amen et al., 2007). These results suggest that members of the Wnt canonical pathway with Pitx2 might compensate for Fgf8 function during tooth initiation. A previous study of Fgfr2b null mice has reported that Fgf8 is not downstream of Fgfr2 IIIb isoform during limb development (Revest et al., 2001). Recently Lu and coworkers report that AER-specific knockout of all isoforms of Fgfr2 causes the reduction of Fgf8 during limb development, which is consistent with our observations (Lu et al., 2008). Furthermore, mice with epithelial-specific Fgfr2b gene deletion do not exhibit tooth retardation and cleft palate (Grose et al., 2007). Taken together, these observations imply that another isoform of Fgfr2 such as IIIc not only is involved in the induction of Fgf8 but also compensates for the function of Fgfr2 IIIb.
In regards to cleft palate, Fgf2b conventional knockout mice exhibit reduced cell proliferation defect in both epithelial and mesenchymal cells (Rice et al., 2004). Furthermore, the palatal shelves of Fgfr2b conventional knockout mice were shorter and did not show palatal projection lateral with the tongue. In contrast, we detect a cell proliferation defect only in the palatal epithelium in K14-Cre;Fgfr2fl/fl mice. Although the size of the palatal shelves in K14-Cre;Fgfr2fl/fl mice are smaller than control mice, palatal projection lateral to the tongue does occur. Taken together, these observations imply that there is an intrinsic requirement for FGF signaling in the epithelium and Fgfr2b expressed in the palatal mesenchyme is likely to be responsible for mediating palatal mesenchymal cell proliferation.
In summary, our epithelial-specific gene deletion of Fgfr2b shows that FGF signaling functions through FGFR2. Future exploration of the difference in the phenotypes of conventional and conditional knockout mice may provide insights into the reciprocal interactions of epithelial and mesenchymal cells during tooth and palate development.
We thank Dr. Julie Mayo for critical reading of the manuscript, Drs. David Ornitz and Sarah Millar for Fgfr2 floxed and K14-Cre mice, respectively. This study was supported by grants from the National Institute of Dental and Craniofacial Research, NIH (DE012711 and DE014078) to Yang Chai.
Grant sponsors: The National Institute of Dental and Craniofacial Research; NIH; Grant numbers: DE012711; DE014078.