|Home | About | Journals | Submit | Contact Us | Français|
Keratocystic odontogenic tumors (KCOTs) are cystic epithelial neoplasias with a high recurrence rate. However, the molecular mechanisms underlying the initiation and progression of KCOTs are still largely unknown. Here, we show that specific ablation of Smad4 in odontoblasts unexpectedly resulted in spontaneous KCOTs in mice. The mutant mice exhibited malformed teeth characterized by fractured incisors and truncated molar roots. These abnormalities were mainly caused by disrupted odontoblast differentiation that led to irregular dentin formation. The cystic tumors arising from the reactivation of epithelial rests of Malassez (ERM), in which Smad4 remained intact, proliferated and formed stratified and differentiated squamous epithelia that exhibited a dramatic upregulation of Hedgehog signaling. Odontoblasts, which are responsive to transforming growth factor beta (TGF-β)/bone morphogenetic protein (BMP) signals, may produce signal molecules to inhibit the activation of ERM. Indeed, we observed a downregulation of BMP signals from Smad4 mutant odontoblasts to the adjacent Hertwig's epithelial root sheath (HERS). Intriguingly, KCOTs frequently emerged from Smad4-deficient ERM in keratinocyte-specific Smad4 knockout mice, suggesting a novel mechanism in which reciprocal TGF-β/BMP signaling between odontoblasts and HERS was required for tooth root development and suppression of KCOT formation. These findings provide insight into the genetic basis underlying KCOTs and have important implications for new directions in KCOT treatment.
Tooth development is a complex multiple-stage process that involves a series of sequential and reciprocal interactions between the dental epithelium and cranial neural-crest-derived mesenchymal cells (49). The ectoderm and the underlying mesenchyme cooperate to thicken, bud, fold, and grow to form the complex shape of the tooth, which can be divided morphologically or developmentally into two parts: the crown and the root. Compared with the crown, the tooth root embedded in the jawbone possesses a unique morphogenetic model. After crown formation, the inner and outer enamel epithelia fuse at the cervix to produce a bilayered tissue named Hertwig's epithelial root sheath (HERS). Thereafter, a mutual induction between HERS and root dentin formation occurs. HERS proliferates apically and induces odontoblast differentiation from dental papilla to form root dentin; then, the odontoblasts signal in turn to regulate the growth and morphogenesis of HERS. As root formation is completed, HERS divides into epithelial nests and cords and remains quiescent in the periodontal ligament known as epithelial rests of Malassez (ERM) (37, 61). Very little is known about the signals controlling the mutual induction between HERS and odontoblasts, as well as maintaining the ERM in quiescence.
Odontogenesis is essentially a fine-tuned communication mediated by growth factors, receptors, and transcription factors between dental epithelial and ectomesenchymal tissues (17, 24, 41, 60, 62), but relatively little is known about tooth root development (36, 55). Emerging evidence highlights an important role of mesenchymal bone morphogenetic proteins (BMPs) and the epithelial transcriptional factor muscle segment homeobox (Msx) in the regulation of the growth and morphogenesis of the root sheath epithelium, which determines the root shape (55). Consistent with this idea, the molar roots of Msx2-deficient mice were markedly truncated and malformed, and enlarged ERM were also observed in the periodontium (2, 7). Nevertheless, the precise mechanisms underlying root development and dental diseases remain poorly understood.
Keratocystic odontogenic tumors (KCOTs) are benign uni- or multicystic, intraosseous tumors of odontogenic origin that account for 11% of all oral cysts. In contrast to other maxillary cysts, such as periodontal lateral cysts, epidermoid cysts, or radicular cysts, the KCOTs stand out for their high recurrence rate, a characteristic lining of parakeratinized stratified squamous epithelium, and the potential for aggressive, infiltrative behavior (29, 42-44). Human KCOTs are known to occur in two modes: sporadic or as part of the nevoid basal cell carcinoma syndrome. In addition to environmental factors involved in the initiation of KCOTs, a portion of human KCOTs may harbor PTCH1 or PTCH2 mutations, which may lead to the derepression of Smoothened (Smo) and constitutive Hedgehog (Hh) signal activities (4, 12). In accordance with the clinical findings, overexpression of the Hh transcriptional effector Gli2 within dental epithelium is sufficient for the induction of highly penetrant KCOTs in K5-Gli2 transgenic mice (13). To date, other signal transduction events that occur during KCOT initiation and progression are still largely unknown. Although tooth development and dental epithelium behavior are believed to be tightly associated, the pathological causes linking tooth root development with KCOT oncogenesis are not well described.
Smads are primary cytoplasmic signal transducers of the transforming growth factor beta (TGF-β)/BMP signaling pathway, which is required for organ development and is commonly involved in diseases (31). Previous studies demonstrated that there were spatial and temporal distributions of Smad1 to -7 during tooth morphogenesis (54). Unfortunately, targeted disruption of Smad4, the unique central cytoplasmic mediator of TGF-β/BMP signaling, resulted in early embryonic lethality in mice (45). The conditional deletion of Smad4 in neural-crest-derived cells resulted in arrested tooth development at the dental lamina stage (22), making it difficult to assess the function of Smad4 in the late stage of tooth development. In order to comprehensively understand the role of Smad4-mediated TGF-β/BMP signaling in tooth development and dental diseases, we specifically deleted the Smad4 gene in odontoblasts using the Cre-LoxP system with the well-established OC-Cre transgenic strain (47). Our data show that Smad4-mediated signals play important roles in mesenchymal-epithelial interactions that control root development, as well as suppress the onset of multiple KCOTs in the jaw.
The generation of Smad4Co/Co OC-Cre and Smad4Co/Co K5-Cre mice has been previously described (46, 47, 56). The offspring were genotyped by PCR analyses using primers described previously (47, 58). The mandibles (with molars) of ROSA26 OC-Cre, Smad4Co/Co ROSA26 OC-Cre, and ROSA mice were stained with X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) using standard procedures (56). Control littermates were used in all experiments. All experimental protocols were designed according to the recommendations of the Beijing Experimental Animal Regulation Board (SYXK/JING/2005/0031).
Five-micrometer sections of mandibles were stained with hematoxylin-eosin or Masson's trichrome stain by standard methods. The primary antibodies used in immunohistochemistry analyses were as follows: Shh, Smo, and Gli-1 (Santa Cruz Biotechnology); keratin 1 (K1), K10, and K14 (Covance); and Smad4 and bromodeoxyuridine (BrdU) (Abcam). Bound antibodies were visualized with diaminobenzidine, and sections were counterstained with hematoxylin. Radiographs of mandibles were obtained in lateral projection with a soft X-ray system (Softex CSM-2). Scanning electron microscopy (SEM) was performed by a protocol described previously (27) using a field emission environmental scanning electron microscope (Hitachi BACPC S4800).
Four-week-old animals were injected with calcein (20 mg/kg of body weight intraperitoneally; Sigma-Aldrich) at day 7 and day 2 before being sacrificed. Five-micrometer cross sections were cut perpendicularly through the erupting incisor, passing through the midsection of the first molar, and were viewed under epifluorescent illumination using a Nikon E600 microscope. We calculated the average dentin mineral apposition rate by eight separate measurements of the distance between the two fluorescent labels in each section using an Osteometrics analytic system and then divided that measurement by the number of days between injections (5 days).
In situ hybridization (ISH) of paraffin sections was performed using standard procedures (57). We designed dentin sialophosphoprotein (DSPP) and Bmp probes to detect their expression (DSPP forward, 5′-GGCTCCGAGTCAATACATGTA-3′, and reverse, 5′-CTCCTTGGTGTCCATTGCTAT-3′; Bmp2 forward, 5′-CGTGGAGGAACTTCCAGAGA-3′, and reverse, 5′-CCACGATCCAGTCATTCCAC-3′; Bmp4 forward, 5′-CCTCAGCAGCATCCCAGAGA-3′, and reverse, 5′-CCACGTCACTGAAGTCCACGT-3′; Bmp7 forward, 5′-CGGAAGTCCATCTCCGTAGTA-3′, and reverse, 5′-GCTGTGGTAGCTGGTAGGATC-3′). Probes were labeled with [35S]UTP using the Maxiscript in vitro transcription kit (Ambion). Slides were dipped in emulsion (Amersham Pharmacia) and exposed for 5 to 10 days before being developed.
Results are presented as means ± standard deviations (SD). All statistical analyses were done using Excel software. Statistical differences were determined by Student's t test. A P value of <0.05 was considered statistically significant in all experiments.
We previously reported a conditional knockout mouse strain (Smad4Co/Co OC-Cre) in which the Smad4 gene was inactivated in mature osteoblasts (47). In addition to revealing that Smad4 was important for regulating postnatal bone homeostasis, we also found tooth malformation in this strain (Fig. (Fig.1).1). Here, we examined LacZ expression in the teeth of ROSA26 OC-Cre double-transgenic mice. In newborn (postnatal day 1 [P1]) animals, LacZ expression was sparsely detected in odontoblasts at the cusp tip of the first molar (M1) of mandibula (Fig. (Fig.1A,1A, left). At P10, when the root development of M1 had just begun, some positive staining of cells was observed within the primordial root (Fig. (Fig.1A,1A, middle), in sharp contrast to the intense LacZ expression within the developing crown (Fig. (Fig.1A,1A, middle). Thorough and specific LacZ expression was detected in the differentiated odontoblasts lining the coronal and root dentin in P20 transgenic mice (Fig. (Fig.1A,1A, right). Immunohistochemical analysis revealed that Smad4 protein was present in the odontoblastic layer of incisors in Smad4Co/+ OC-Cre mice but was dramatically decreased in that of their Smad4Co/Co OC-Cre littermates (Fig. (Fig.1B).1B). Southern blot analysis confirmed that Cre-mediated recombination occurred exclusively in teeth and bones, which contain odontoblasts or osteoblasts, respectively (Fig. (Fig.1B,1B, bottom). These data indicated that Smad4 was efficiently disrupted in odontoblasts by OC-Cre-mediated recombination.
The general appearance of Smad4Co/Co OC-Cre incisors was characterized by an opaque, chalky-white color and a fragile appearance in contrast to the Smad4Co/+ OC-Cre control incisors. Dentin defects on the lingual side of the mutant incisors were also observed (Fig. (Fig.1C).1C). At P28, M1s were in the process of eruption, since the crown surfaces showed neither mechanical wear nor dental caries and the roots had just completed development. At this stage, compared with the controls, the roots of mutant M1s were significantly shorter (Fig. (Fig.1D),1D), as well as those of the second or third molars (data not shown). At P28, the average root lengths of maxillary and mandibular M1s in control mice were about 1.40 ± 0.16 mm and 1.97 ± 0.22 mm, respectively, while in mutant littermates they decreased to 1.12 ± 0.12 mm and 1.30 ± 0.16 mm (n = 7) (Fig. (Fig.1E).1E). The average crown/root ratios of control maxillary and mandibular M1s were 0.56 ± 0.07 and 0.56 ± 0.08, whereas the ratios were 0.81 ± 0.08 and 1.10 ± 0.11 (n = 7), respectively, in mutant M1s (Fig. (Fig.1F).1F). However, there was no obvious change in the length of the crown (Fig. (Fig.1D1D and data not shown). The number of roots in each molar remained unaffected, reflecting normal initiation of root formation.
Another striking abnormality seen in mutant mice was the prevalence of multiple cysts in the mandibles of mutant mice older than 3 months of age with 100% penetrance (n = 42) (Fig. (Fig.1G).1G). Necropsy of these mice revealed multiple cystic cavities lined by thin walls, which frequently contained white keratinized material (data not shown). Radiographic analysis demonstrated multilocular cystic lesions within the mandible, as indicated by expansile changes and a thinning of the mandibular cortex (Fig. (Fig.1G1G).
The molars and incisors at different developmental stages were sectioned for histological analyses. We found an evident dentin deficiency within the tooth roots and root analogues of the mutant mice. At P12, right after the initiation of M1 root formation, the arrangement of coronal odontoblasts and the structure of coronal dentin in mutant M1 were relatively normal compared with those in the controls. However, dentin formed rarely and discontinuously in mutant roots, and with irregular morphology (Fig. (Fig.2A,2A, top). At P45, the root dentin of mutant M1s displayed remarkable dysplasia of the osteodentine type: most odontoblasts were entrapped in the matrix (Fig. (Fig.2A,2A, middle and bottom). Similar defects occurred within the lingual side (root analogue) of incisors in P12 and P70 mutant mice (Fig. (Fig.2B).2B). Interestingly, enlarged ERM within the periodontal ligament were always observed (Fig. 2A and B).
We performed resin cast SEM to investigate the ultrastructure of the dentin tubular system. In the dental neck zones of mutant molars, the arrangement and morphology of the dentin tubular system appeared normal at the crown compared with those of the controls; however, the root portion of the system was malformed, as suggested by polygonal cells entrapped in the surrounding matrix (Fig. (Fig.3A).3A). LacZ staining revealed these entrapped cells to be positively stained odontoblasts (Fig. (Fig.3A,3A, right). High magnification of SEM images revealed significant differences between the controls and mutants in the dentin tubular system of the root portions of molars (Fig. (Fig.3B,3B, bottom), but not the coronal portions of molars (Fig. (Fig.3B,3B, top). The polygonal odontoblasts were irregularly arranged, with processes oriented in multiple directions (Fig. (Fig.3B).3B). We then performed a double-fluorescence labeling analysis, a histomorphometric measurement of odontoblast activity in vivo, which revealed a markedly decreased dentin mineral apposition rate in the labial side of incisors and the coronal portions of molars (Fig. (Fig.3C3C and data not shown) associated with Smad4 deficiency in 4-week-old animals. The analysis also revealed discontinuous, wavy, and more diffuse labeling lines on the lingual side of incisors and the root portions of molars from mutant mice (Fig. (Fig.3C3C and data not shown). These data indicated that ablation of Smad4 in odontoblasts caused abnormalities in the morphodifferentiation and functions of odontoblasts in the roots of molars and the lingual side of incisors.
BrdU labeling and terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling assays revealed no significant differences in proliferation and apoptosis between wild-type and mutant odontoblasts (data not shown). We then examined the differentiation of odontoblasts by ISH. As the early marker genes for odontoblasts, Col1a1 and osteocalcin (OC) were intensively expressed along the odontoblastic layer with the declining root-crown expression gradient in wild-type molars, but in the mutants, their expression was elevated within the coronal odontoblasts (Fig. 4A and B) and was dispersed around the root region (Fig. 4A and B). The expression of DSPP, a late marker gene for odontoblasts, was reduced in coronal odontoblasts of mutant molars (Fig. (Fig.4C),4C), and was scarcely detected in root odontoblasts of mutant teeth (Fig. (Fig.4C).4C). These data suggest that in Smad4Co/Co OC-Cre mice, the differentiation of odontoblasts was delayed within the crown and was arrested at the secretory stage within the molar root.
K14 was expressed by many epithelial tissues, such as the oral epithelium and the enamel organ. However, HERS and ERM were the only structures that could be positively stained for K14 along the root surface and in its periphery (28). To characterize in more detail the genesis of jaw cysts in Smad4Co/Co OC-Cre mice, we marked HERS and ERM with anti-K14 antibody at different developmental stages. At P10, a line of individual HERS cells in close proximity to the lingual surfaces of wild-type incisors was clearly observed. At P35, the distance between individual ERM cells increased and the number of ERM cells was markedly reduced; by P72, the ERM were barely detectable (Fig. (Fig.5A,5A, top). Although no obvious change could be found in mutant incisors at P10, HERS cells in mutant mice unconventionally formed larger ERM at P35. Some of the large ERM continued growing and gradually became stratified cysts by P72 (Fig. (Fig.5A,5A, bottom), which were noted as multiple cysts within the periodontal ligaments of the incisors shown in Fig. Fig.2B2B.
Similar anomalies developed on mutant molars as early as P10, immediately after the beginning of root formation (Fig. (Fig.5B).5B). No obvious dentin formation could be observed within the mutant root by P10; the continuous HERS cells still outlined the shape of the future root. At P15, some irregular dentin formed, but the fenestration of HERS of the coronal portion often failed, resulting in the connection of continuous HERS with the cervical margin of the ameloblast cell layer. The continuous HERS proliferated into stratified epithelium by P35 (Fig. (Fig.5B,5B, bottom). The odontogenic epithelium also continually proliferated and formed cysts within the periodontal ligaments of the molars shown in Fig. Fig.2A2A.
These data indicated that deletion of Smad4 in odontoblasts, which changed the fate of the odontoblasts, could also alter the fate of HERS and ERM, leading to the formation of multiple odontogenic cysts in jaws.
The distinct features, including the ERM origin, a stratified squamous epithelial lining, and corrugated corneal layers with an aggressive nature, strongly suggested the cysts in mutant mice were odontogenic keratocysts. Therefore, we analyzed the expression of keratins in the jaw cysts. K1 and K10 are markers for cornified epithelium and are generally expressed in human KCOTs (42). K1 and K10 were expressed in early cysts within the periodontal ligaments of mutant mice but were not detected in the ERM of control mice. Moreover, the expression pattern of the terminal differentiation marker K10 within the innermost lining of the cyst epithelium was similar to that in human KCOTs, whereas K14, the wide-spectrum marker for epithelium, was found in the cyst epithelium of mutant mic, as well as in ERM of control mice. BrdU incorporation demonstrated that robust proliferation in early cysts was limited within the basal layer but was undetectable within the ERM of control mice (Fig. (Fig.6A6A).
Human KCOTs have been reported to be related to PTCH1 or PTCH2 gene mutations, which lead to the pathological activation of Hh signaling (4, 12). By immunohistochemical analyses, we observed remarkable upregulation of the expression of a series of Hh signaling components, Shh, Smo, and Gli1, in the cysts of Smad4Co/Co OC-Cre mice (Fig. (Fig.6B).6B). Based on these evidences of positive keratin expression, as well as the dramatically reactivated Hh signaling within the hyperplastic epithelium, the cysts in Smad4Co/Co OC-Cre mice were interpreted as typical KCOTs.
Observations of the hyperplastic, aggressive behavior, in addition to the upregulated Hh signaling, of KCOTs in Smad4 mutant mice suggested a phenotype resembling that of Smad4 deficiency. However, as seen in Fig. Fig.6C,6C, regardless of the expression within odontoblasts, the Smad4 protein was consistently present within HERS of either Smad4Co/Co OC-Cre or Smad4Co/+ OC-Cre mice. In addition, positive immunohistochemical staining of Smad4 in the cystic epithelium corroborated the availability of intact Smad4-mediated signaling within these cells (Fig. (Fig.6D).6D). Consistent with this, LacZ staining of P70 Smad4Co/Co ROSA26 OC-Cre mice verified that, although the odontoblasts of incisors were markedly stained, the cysts did not carry evidence of the gene knockout (data not shown). Taken together, the data suggest that oncogenesis of KCOTs would not be a consequence of genetic ablation of Smad4 within the dental epithelia.
Since KCOTs should not have arisen from an intrinsic genomic mutation per se, altered paracrine signals from the microenvironment might account for the aberrant behavior of root sheath epithelia. Bmp genes, including Bmp2, Bmp3, Bmp4, and Bmp7, have been reported to be coexpressed in early odontoblasts in the apical region and are likely to act as mesenchymal factors that regulate the growth and morphogenesis of HERS during tooth root development (55). Therefore, we performed ISH to examine the expression of these Bmp genes on sections of P15 molars. In wild-type mice, Bmp2 and Bmp7 transcripts were mainly present in early odontoblasts in the apical region; Bmp4 was expressed similarly to Bmp2 and Bmp7 but with the greatest abundance in the preodontoblastic cells proximate to the epithelial root sheath (Fig. (Fig.7,7, top). In the roots of mutant molars, however, we witnessed a striking reduction of Bmp expression within the poorly differentiated odontoblasts (Fig. (Fig.7,7, bottom). Consistent with this, the expression of Msx2, encoding an essential transcriptional factor regulated by BMPs during tooth development (41, 55), was remarkably reduced within the root sheath epithelium of mutant mice (Fig. (Fig.7,7, far right). Our in situ data led us to the conclusion that, by deletion of Smad4, the dysplastic odontoblasts failed to produce the mesenchymal BMP signals that might direct the development or recession of adjacent HERS.
To validate whether dysregulated functions of dental epithelia were due to disturbed TGF-β/BMP paracrine signaling in Smad4Co/Co OC-Cre mice, we utilized Smad4Co/Co K5-Cre mice (48, 58, 59), in which Smad4-mediated TGF-β/BMP signaling was ablated in the dental epithelia via K5-Cre-mediated recombination. We hypothesized that Smad4-deficient HERS, which cannot respond to the paracrine TGF-β/BMP signals released by odontoblasts, would also develop KCOTs. As expected, the Smad4Co/Co K5-Cre mice also had a high incidence of jaw cysts. Histological analyses located the source of the cysts within the periodontal ligaments (Fig. (Fig.8A).8A). As revealed by immunohistochemical staining of K1, K10, and K14, these cysts were confirmed to be KCOTs largely resembling those seen in Smad4Co/Co OC-Cre mice (Fig. (Fig.8B).8B). However, the cystic epithelium of Smad4Co/Co K5-Cre mice was Smad4 deficient, which was intrinsically different from that of the Smad4Co/Co OC-Cre mice (Fig. (Fig.8B).8B). The in vivo evidence strongly indicated that impairing the TGF-β/BMP pathway by either disturbing BMP paracrine signals or inhibiting intracellular signal transduction was sufficient to interrupt the normal receding process of HERS during root development and led to the formation of KCOTs.
In the present study, we demonstrated that odontoblast-specific knockout of Smad4 resulted in abnormal root development due to the disturbed differentiation of root odontoblasts. This disruption impaired the mesenchymal BMP signaling pathways and consequently altered the fate of HERS and ERM, which led to the onset of multiple KCOTs. Our data provided convincing evidence that the pivotal role of intact TGF-β/BMP communication between dental epithelial and ectomesenchymal tissues was required for the normal development of tooth roots and HERS.
Here, we have provided direct genetic evidence to suggest an indispensable function of Smad4 in the differentiation of root odontoblasts. Within the TGF-β family, TGF-β1, -2, and -3 and BMP2, -4, and -7 were believed to regulate the differentiation of odontoblasts through an autocrine mode of action based on examining their expression patterns in the inner dental epithelium and in the polarizing and functional odontoblasts (5, 6, 50). Some in vitro studies have shown that administration of TGF-β1 and -3 and BMP2 and -4 can induce the polarization or differentiation of odontoblasts (51). Targeted disruption of TGF-β type II receptor in cranial neural crest-derived cells demonstrated a cell-autonomous requirement for TGF-β signaling during odontoblast differentiation and dentin matrix formation (35). BMP2 has been shown to regulate DSPP gene expression and odontoblast differentiation via NF-Y signaling during tooth development (9). Recent studies have revealed that ectodermal Smad4 and p38 are functionally redundant in mediating TGF-β signaling in the proper patterning of dental cusps during early tooth development (53). Here, our in vivo data implied an important role for Smad4-dependent TGF-β/BMP signaling in the terminal differentiation of root odontoblasts. We showed that ablation of Smad4 in odontoblasts resulted in altered polarity of the odontoblasts, reduced expression of DSPP, and a decreased dentin mineral apposition rate. These were largely correlated with downregulation of BMPs in odontoblasts. Therefore, intracellular Smad4 is required for a positive-feedback loop of TGF-β/BMP signaling in regulating the terminal differentiation of root odontoblasts.
Our data also enabled the construction of a model that revealed a new mechanism underlying the formation of KCOTs. Numerous investigations have recently suggested that odontogenic keratocysts are neoplastic tumors rather than simple cystic lesions (42-44). The totally unexpected phenotype of Smad4Co/Co OC-Cre mice was the noticeable prevalence of highly penetrant odontogenic keratocysts, which exhibited pathological characteristics resembling those in human KCOTs. Mutations of the human PTCH genes have frequently been found to be involved in nevoid basal cell carcinoma syndrome and sporadic KCOTs (1, 4, 14, 18, 23). Similar KCOTs have been confirmed in a few mouse models. For example, 25.4% of heterozygous ptc knockout mice or 95% to 100% of K5-Gli2 transgenic mice developed KCOTs, which arose from the mutant ERM due to activation of Hh signaling (13, 21). The constitutively activated Hh signaling was also detected in the cyst epithelia of Smad4Co/Co OC-Cre mice. However, the altered fate of Smad4-intact HERS in Smad4Co/Co OC-Cre mice was tightly correlated with the abnormality of root development, which was not evident in either ptc heterozygous or K5-Gli2 transgenic mice. While it is widely accepted that the development of KCOTs is largely due to the accumulation of somatic mutations in epithelial cells, we have demonstrated here for the first time that normal HERS epithelial cells that fail to receive the right paracrine signals from root odontoblasts continued to proliferate and form KCOTs.
We have provided the first in vivo genetic evidence to show that odontoblastic TGF-β/BMP pathways provide key mesenchymal signals that regulate the development of HERS. Previous studies suggested that the BMP-Msx pathway, particularly BMP2, -4, and -7 and Msx2, might be involved in the interaction between the odontoblasts and HERS during root formation (39, 55). In this study, several lines of evidence implied that compromised BMP signaling was responsible for the dysregulated behavior of HERS. First, by deletion of Smad4, we showed that the dysplastic odontoblasts produced fewer mesenchymal BMPs toward the adjacent HERS. Second, decreased Msx2 expression within HERS could also elucidate the phenotype of hyperplastic ERM and differentiated squamous epithelium of KCOTs in mutant mice (2, 16). Consistent with this, a recent bead implantation experiment underscored the importance of BMP4 in directing HERS development by preventing elongation and maintaining cell proliferation (15). In addition, identical KCOTs occurring in Smad4Co/Co K5-Cre mice with Smad4-deficient HERS epithelium conclusively demonstrated that unimpeded TGF-β/BMP signaling was required to control the activity of Hh signaling in HERS and ERM, direct their fates, and, finally, restrain the formation of KCOTs. BMPs and their antagonists have been shown to mediate Hh signaling during various developmental processes in mice (3, 8, 10, 11, 30, 52, 64). However, dual roles for BMPs in the regulation of Hh signaling have been reported. Previous studies have shown that BMP signaling activates Ihh in chondrocytes (33, 34, 40), while several in vivo and in vitro studies have also suggested that BMP signaling has a long-range and antagonistic effect on Hh signaling (19, 25, 26, 32, 38). BMPs can act upstream to suppress Shh expression in the developing mouse tooth germ, and during mouse palatogenesis, beads soaked with BMP4 protein and the BMP4 transgene are able to restore the expression of both Shh and Bmp2 in the Msx1 mutant epithelium (10, 62, 63). Ablation of Smad4 or BMP receptor type IA can bring on Hh signaling in tissues formed through ectoderm-mesenchyme interactions, such as in the hair follicles and limbs (58). In this regard, the upregulation of Hh signaling in KCOTs of Smad4Co/Co OC-Cre mice could be interpreted as a result of reduced TGF-β/BMP signaling. Direct evidence linking compromised BMP signaling to KCOT initiation is required, for example, by generating constitutively active BMP receptor conditional transgenic mice that could be bred with Smad4 mutant mice to rescue the KCOT phenotype. A previous study revealed that Smad4-dependent signaling in T cells is required for suppressing intestinal tumorigenesis (20). All these data suggest that TGF-βs or BMPs serve as important signals controlling normal communication between the epithelium and the adjacent cells, which is required for maintaining epithelium homeostasis.
We thank Chuxia Deng for Smad4Co/Co mice.
This work was supported by the Chinese National Key Program on Basic Research (2005CB522506, 2006CB943501, 2006BAI23B01-3, and 2007CB947304), the Chinese Key Program for Drug Invention (2009ZX09501-027), the National Natural Science Foundation of China (30430350), the National High-Tech Research and Development Program (2006AA02Z168), the Beijing Major Scientific Program (D0906007000091), and grant Z0006303041231.
Published ahead of print on 24 August 2009.