Ectopic Activation of Fgf8 in Dental Mesenchyme Causes Incisor Agenesis and Molar Microdontia

Putatively, tooth agenesis was attributed to the initiation failure of tooth germs, though little is known about the histological and molecular alterations. To address if constitutively active FGF signaling is associated with tooth agenesis, we activated Fgf8 in dental mesenchyme with Osr-cre knock-in allele in mice (Osr2-creKI; Rosa26R-Fgf8) and found incisor agenesis and molar microdontia. The cell survival assay showed tremendous apoptosis in both the Osr2-creKI; Rosa26R-Fgf8 incisor epithelium and mesenchyme, which initiated incisor regression from cap stage. In situ hybridization displayed vanished Shh transcription, and immunostaining exhibited reduced Runx2 expression and enlarged mesenchymal Lef1 domain in Osr2-creKI; Rosa26R-Fgf8 incisors, both of which were suggested to enhance apoptosis. In contrast, Osr2-creKI; Rosa26R-Fgf8 molar germs displayed mildly suppressed Shh transcription, and the increased expression of Ectodin, Runx2 and Lef1. Although mildly smaller than WT controls prenatally, the Osr2-creKI; Rosa26R-Fgf8 molar germs produced a miniature tooth with impaired mineralization after a 6-week sub-renal culture. Intriguingly, the implanted Osr2-creKI; Rosa26R-Fgf8 molar germs exhibited delayed odontoblast differentiation and accelerated ameloblast maturation. Collectively, the ectopically activated Fgf8 in dental mesenchyme caused incisor agenesis by triggering incisor regression and postnatal molar microdontia. Our findings reported tooth agenesis resulting from the regression from the early bell stage and implicated a correlation between tooth agenesis and microdontia.


Introduction
Tooth development relies on the intensive interactions between the dental epithelium and the underlying mesenchyme, in which a lot of growth factors, such as Wnts, Transformation Growth Factor-β (TGF-β)/Bone Morphogenic Protein (BMP), Sonic Hedgehog (SHH) and Fibroblast Growth Factor (FGF) families, are involved [1].The FGF family consists of 22 members, which are divided into seven groups (Fgf1, Fgf4, Fgf7, Fgf8, Fgf9, Fgf11 and Fgf15 groups) according to their affinities to FGF receptor 1-4 (Fgfr1-4) [2].In the past decades, a lot of studies revealed the roles of FGF signaling in the patterning, initiation, morphogenesis and differentiation of tooth development [3].During early tooth development, Fgf8-expressing epithelial progenitors in mandible migrated toward the Shh-expressing domain to decide the position of tooth primordium.Suppression of FGF and SHH signaling resulted in tooth agenesis by disrupting the oriented cell migration of cells [4].In contrast, the deletion of Sprouty2 (Spry2) and Sprouty4 (Spry4), the intracellular suppressor of FGF signaling, led to supernumerary teeth by persistently activating FGF signaling [5,6].During tooth morphogenesis, the expressions of Fgf4, Fgf9 and Fgf20 in dental epithelium acted as the major downstream effectors of ectodysplasin (Eda) that could rescue the reduced size, number and shape of teeth resulting from the loss-of-function of Eda [7].In contrast to the epithelium-expressed Fgf8, Fgf4, Fgf9 and Fgf20, Fgf3 and Fgf10 were activated in dental mesenchyme from E12.5 and maintained by Runt Related Transcription Factor 2 (Runx2) to form interaction loop with the epithelial Fgf4 and Fgf10.Runx2 null mice showed tooth germs arrested at early cap stage with the reduced Fgf3, Fgf10, Fgf4 and Fgf9 expression [8].Moreover, the Fgf10 expressed in muse incisor mesenchyme adjacent to the cervical loop maintained the stem cell niche for the self-renewal of ameloblasts and odontoblasts [9].The incisor germs in Fgf10 null mice proceeded to cap stage without the formation of a cervical loop because of tremendous cell death [10].Therefore, disruptions in the expression of Fgfs, as well as the FGF signaling in dental epithelial and/or mesenchymal cells, could lead to the developmental failure of tooth germs, which causes tooth agenesis in clinics.
Tooth agenesis is categorized into hypodontia, referring to the absence of six or less teeth, and oligodontia, referring to the lack of more than six teeth [11].Genes involved in tooth agenesis have been identified in humans, such as Paired Homeobox Gene 9 (PAX9), Muscle Segment Homeobox Homolog 1 (MSX1), Axis Inhibition Protein 2 (AXIN2), WNT10A, WNT10B, EDA and EDAR, and verified in genetic mouse models [12].A series of studies indicated that the mesenchymal Wnt signaling was modulated by the antagonistic interaction between Axin2 and Runx2, and an increased Wnt signaling in dental mesenchyme impaired tooth development [13].Mutations of AXIN2, an inhibitor to Wnt signaling, are associated with tooth agenesis [14], while the haplo-insufficiency of RUNX2 results in supernumerary teeth [15]; thus, the exact mechanism of Runx2 in tooth development remains to be elucidated.On the other hand, although the roles of these genes been intensively studied during tooth development, few investigations have concerned the pathogenesis of tooth agenesis.Since few studies have reported on the degeneration of tooth germs in the late cap or bell stage, tooth agenesis is putatively attributed to the failure of tooth initiation, or the arrest at the bud or early cap stage [12,16].
Compared to tooth agenesis, microdontia, which is characterized by the miniature teeth, is relatively rare.However, the definition of microdontia is still in debate, because microdontia is usually companied by micrognathia and regarded as a complication of micrognathia.Actually, regardless of micrognathia, the etiology of microdontia also remains elusive.Most microdontia was reported as a complication in syndromes such as LAMM syndrome (labyrinthine aplasia, microtia and microdontia), Axenfeld-Rieger syndrome, Lacrimo-auriculo-dento-digital syndrome, Kenny-Caffey Syndrome, etc. [17][18][19][20].Because of the mutations detected in above syndromes, FGF3, FGF10, FGFR2 and FGFR3, especially FGF3, are candidates for the causative gene of microdontia [17,19,21].Since tooth eruption takes place after birth, tooth agenesis or microdontia can only be found in childhood, and that impeded our understanding of their pathological processes.
As mentioned above, FGF signaling is involved in both tooth agenesis and microdontia.However, the exact roles of FGF signaling in tooth agenesis and microdontia are still being debated, which requires both the loss-and gain-of-function models for further exploration.Since most of the mutations reduce gene expression or protein activity, previous reports on the mutations of FGF genes or FGF signaling could be regarded as the model of loss-offunction [22].By contrast, there are a few reports on the gain-of-function models to clarify the roles of FGF genes or signaling in tooth agenesis and microdontia.In this study, by activating Fgf8 in dental mesenchyme, we constructed a gain-of-function model of FGF signaling.Since the attenuated FGF signaling in dental mesenchyme is correlated with the reduced tooth number, we hypothesized that Osr2-cre KI ; Rosa26R-Fgf8 mice would give rise to supernumerary tooth as RUNX2 haplo-insufficient cases did [15].However, we found the agenesis of incisors and the microdontia in molars.Our study not only revealed a precise pathogenesis of tooth agenesis but also suggested a potential correlation between tooth agenesis and microdontia.

Differential Gene Expression between Osr2-cre KI ; Rosa26R-Fgf8 Incisor and Molar Germs
To address the apoptosis in the enamel organs of Osr2-cre KI ; Rosa26R-Fgf8 incisors, the transcription of Shh, a growth factor expressed in dental epithelial cells key to cell proliferation, survival and differentiation, was examined by in situ hybridization.When robust in the E14.5 EKs of WT incisors, the Shh transcription was excluded in the enamel organs of Osr2-cre KI ; Rosa26R-Fgf8 incisors (Figure 3A,B).Similarly, the Shh transcription confined to the differentiating ameloblasts of E16.5 WT incisors was also absent from the epithelial compartments of Osr2-cre KI ; Rosa26R-Fgf8 incisors (Figure 3C,D).In contrast to the absence in the incisors, the Shh-expressing EKs were obvious in the E14.5 Osr2-cre KI ; Rosa26R-Fgf8 molars, though smaller and more concentrated to the buccal sides compared to WT molars (Figure 3E,F).At E16.5, Shh transcription extended into the entire inner ameloblast layer in the Osr2-cre KI ; Rosa26R-Fgf8 molars, as in WT molars, but the cusps of the Osr2-cre KI ; Rosa26R-Fgf8 molars were still inconspicuous (Figure 3G,H).Similarly, Runx2, the transcription factor mediating Fgf3 and Fgf10 activation in dental mesenchyme, was almost diminished in E14.5 Osr2-cre KI ; Rosa26R-Fgf8 incisor mesenchyme compared to the robust expression in WT controls (Figure 3I,J), while the Runx2 expression in E14.5 Osr2cre KI ; Rosa26R-Fgf8 molar mesenchyme was comparable to, or even a little robuster than that in WT molars (Figure 3K,L).The transcription of Ectodin, a dual inhibitor to Wnt and BMP signaling, was remarkably enhanced in both the E14.5 Osr2-cre KI ; Rosa26R-Fgf8 incisor and molar epithelium compared to the WT counterparts (Figure 3M-P).Taken together, the suppressed Shh and Runx2 expression specific to Osr2-cre KI ; Rosa26R-Fgf8 incisor germs indicated an attenuated FGF signaling that resulted in the incisor agenesis.Meanwhile, the enhanced Ectodin transcription, implicating an inhibition of Wnt and BMP signaling in both Osr2-cre KI ; Rosa26R-Fgf8 incisor and molar germs, was suggested to indiscriminately repress incisor and molar development.

Altered Gene Expression in Osr2-cre KI ; Rosa26R-Fgf8 Incisor and Molar Germs
Then, we examined several key signaling pathways in the E13.5 incisor and molar germs.Compared with the WT controls (Figure 4A,B), Fgfr1, the main receptor for Fgf8, was also ectopically activated in incisor enamel organs (Figure 4A′), and drastically upregulated in the presumptive molar papilla of Osr2-cre KI ; Rosa26R-Fgf8 mice (Figure 4B′).The p-Erk1/2 distribution was sporadic in the E13.5 WT incisor mesenchyme beside the epithelial stalk (Figure 4C), but it was more intensive in the counterpart of Osr2-cre KI ; Rosa26R-Fgf8 incisor germs (Figure 4C′).In the molar mesenchyme, the intensive p-Erk1/2 signal in the WT molar papilla (Figure 4D) was almost diminished in the Osr2-cre KI ; Rosa26R-Fgf8 molar papilla, but it was ectopically activated in the mesenchyme lingual to epithelial stalks (Figure 4D′).Lef1, a nuclear mediator of canonical Wnt signaling, which was distributed in the epithelium and mesenchyme of WT incisor and molar germs (Fig- ure 4E,F), became enlarged in the incisor mesenchyme (Figure 4E′), but fainter in the molar mesenchyme of Osr2-cre KI ; Rosa26R-Fgf8 mice (Figure 4F′), suggesting an up-and downregulated Wnt signaling in incisor and molar mesenchyme, respectively.Compared to WT tooth germs (Figure 4G,H), the distribution of p-Smad1/5/8 seemed little difference in both 2.4.Altered Gene Expression in Osr2-cre KI ; Rosa26R-Fgf8 Incisor and Molar Germs Then, we examined several key signaling pathways in the E13.5 incisor and molar germs.Compared with the WT controls (Figure 4A,B), Fgfr1, the main receptor for Fgf8, was also ectopically activated in incisor enamel organs (Figure 4A ′ ), and drastically up-regulated in the presumptive molar papilla of Osr2-cre KI ; Rosa26R-Fgf8 mice (Figure 4B ′ ).The p-Erk1/2 distribution was sporadic in the E13.5 WT incisor mesenchyme beside the epithelial stalk (Figure 4C), but it was more intensive in the counterpart of Osr2-cre KI ; Rosa26R-Fgf8 incisor germs (Figure 4C ′ ).In the molar mesenchyme, the intensive p-Erk1/2 signal in the WT molar papilla (Figure 4D) was almost diminished in the Osr2-cre KI ; Rosa26R-Fgf8 molar papilla, but it was ectopically activated in the mesenchyme lingual to epithelial stalks (Figure 4D ′ ).Lef1, a nuclear mediator of canonical Wnt signaling, which was distributed in the epithelium and mesenchyme of WT incisor and molar germs (Figure 4E,F), became enlarged in the incisor mesenchyme (Figure 4E ′ ), but fainter in the molar mesenchyme of Osr2-cre KI ; Rosa26R-Fgf8 mice (Figure 4F ′ ), suggesting an up-and down-regulated Wnt signaling in incisor and molar mesenchyme, respectively.Compared to WT tooth germs (Figure 4G,H), the distribution of p-Smad1/5/8 seemed little difference in both the incisor mesenchyme (Figure 4G ′ ) and molar mesenchyme of Osr2-cre KI ; Rosa26R-Fgf8 mice (Figure 4H ′ ).

Delayed Odontoblast Differentiation and Enhanced Ameloblast Degeneration in Osr2-cre KI ; Rosa26R-Fgf8 Molars
Then, we examined the ameloblastic and odontoblastic differentiation in Osr2-cre KI ; Rosa26R-Fgf8 molars by in situ hybridization.Amelogenin, the matrix essential for enamel formation and dominantly expressed in ameloblasts, showed no difference between WT (Figure 5A) and Osr2-cre KI ; Rosa26R-Fgf8 molars (Figure 5B) after 3 days of sub-renal culture.In contrast, Dentin sialophosphoprotein (DSPP), encoding the dentin-specific protein and activated in both the WT molar ameloblasts and odontoblasts, was detected only in the E17.5 Osr2-cre KI ; Rosa26R-Fgf8 ameloblasts but absent in the odontoblasts after 3 days of sub-renal culture (Figure 5C).However, after 5 days of sub-renal culture, although still expressed in both the E17.5 WT molar ameloblasts and odontoblasts (Figure 5E), the DSPP

Delayed Odontoblast Differentiation and Enhanced Ameloblast Degeneration in Osr2-cre KI ; Rosa26R-Fgf8 Molars
Then, we examined the ameloblastic and odontoblastic differentiation in Osr2-cre KI ; Rosa26R-Fgf8 molars by in situ hybridization.Amelogenin, the matrix essential for enamel formation and dominantly expressed in ameloblasts, showed no difference between WT (Figure 5A) and Osr2-cre KI ; Rosa26R-Fgf8 molars (Figure 5B) after 3 days of sub-renal culture.In contrast, Dentin sialophosphoprotein (DSPP), encoding the dentin-specific protein and activated in both the WT molar ameloblasts and odontoblasts, was detected only in the E17.5 Osr2-cre KI ; Rosa26R-Fgf8 ameloblasts but absent in the odontoblasts after 3 days of sub-renal culture (Figure 5C).However, after 5 days of sub-renal culture, although still expressed in both the E17.5 WT molar ameloblasts and odontoblasts (Figure 5E), the DSPP expression was only in the E17.5 Osr2-cre KI ; Rosa26R-Fgf8 molar odontoblasts and vanished from the ameloblasts (Figure 5F).Thus, it is suggested that the persistent activated Fgf8 in molar mesenchyme delayed odontoblast differentiation but enhanced the degeneration of ameloblasts.
expression was only in the E17.5 Osr2-cre KI ; Rosa26R-Fgf8 molar odontoblasts and vanished from the ameloblasts (Figure 5F).Thus, it is suggested that the persistent activated Fgf8 in molar mesenchyme delayed odontoblast differentiation but enhanced the degeneration of ameloblasts.

Impaired Mineralization and Reduced Size of Osr2-cre KI ; Rosa26R-Fgf8 Molars
Since Osr2-cre KI ; Rosa26R-Fgf8 mice died immediately after birth, we transplanted the E17.5 Osr2-cre KI ; Rosa26R-Fgf8 molar germs into a sub-renal capsule for a 3-or 6-week culture to evaluate the in vivo mineralization.After 3 weeks of sub-renal culture, the X-ray images of the grafted E17.5 Osr2-cre KI ; Rosa26R-Fgf8 molar germs (Figure 6B) were similar to those of the grafted WT molar germs (Figure 6A).However, histological sections revealed that the predentin layer in the grafted Osr2-cre KI ; Rosa26R-Fgf8 molar germs (Figure 6D,D ′ ) was much thicker than that in WT molar grafts (Figure 6C,C ′ ).After 6 weeks of sub-renal culture, the grafted E17.5 WT molar germs gave rise to dental root, crown and alveolar bone (Figure 6E).In contrast, the grafted E17.5 Osr2-cre KI ; Rosa26R-Fgf8 molar germs only formed atypical tooth morphology and hypomineralized tooth miniatures (Figure 6F), implicating the impaired mineralization, especially in dentin, and the microdontia resulting from the dental mesenchyme-derived Fgf8.Combined with the prenatally regressed incisor germs, the persistent Fgf8 expression in dental mesenchyme was suggested to impair not only tooth development by repressing Shh transcription and FGF signaling but also postnatal tooth size, root formation and dentin mineralization.6F), implicating the impaired mineralization, especially in dentin, and the microdontia resulting from the dental mesenchyme-derived Fgf8.Combined with the prenatally regressed incisor germs, the persistent Fgf8 expression in dental mesenchyme was suggested to impair not only tooth development by repressing Shh transcription and FGF signaling but also postnatal tooth size, root formation and dentin mineralization.Structure and histology of the grafted Osr2-cre KI ; Rosa26R-Fgf8 molar germs.After 3 weeks of sub-renal culture, plain X-ray images showed the tooth structure of the grafted E17.5 WT (A) and Osr2-cre KI ; Rosa26R-Fgf8 molar germs (B).HE staining shows the histological features of the E17.5 WT (C) and Osr2-cre KI ; Rosa26R-Fgf8 (D) molar grafts after 3 weeks of sub-renal culture.(C′,D′) are the magnified views in the boxes of (C,D), respectively.After 6 weeks of sub-renal culture, plain Xray images showed the tooth morphology of the grafted E17.5 WT (E) and Osr2-cre KI ; Rosa26R-Fgf8 molar germs (F).The black arrows point to the predentin, and the arrowhead points to the dentin in molar grafts.Scale bar, 100 µm.

Discussion
Figure 6.Structure and histology of the grafted Osr2-cre KI ; Rosa26R-Fgf8 molar germs.After 3 weeks of sub-renal culture, plain X-ray images showed the tooth structure of the grafted E17.5 WT (A) and Osr2-cre KI ; Rosa26R-Fgf8 molar germs (B).HE staining shows the histological features of the E17.5 WT (C) and Osr2-cre KI ; Rosa26R-Fgf8 (D) molar grafts after 3 weeks of sub-renal culture.(C ′ ,D ′ ) are the magnified views in the boxes of (C,D), respectively.After 6 weeks of sub-renal culture, plain X-ray images showed the tooth morphology of the grafted E17.5 WT (E) and Osr2-cre KI ; Rosa26R-Fgf8 molar germs (F).The black arrows point to the predentin, and the arrowhead points to the dentin in molar grafts.Scale bar, 100 µm.

Discussion
Tooth agenesis is putatively regarded as a consequence of failed tooth initiation, because most mouse models of tooth agenesis showed tooth development arrested at the laminar or bud stage, except for Runx2 null mice at the early cap stage [8,10,11].In this study, we reported the incisor agenesis in Osr2-cre KI ; Rosa26R-Fgf8 mice in which the cap-stage incisor germs failed to enter the bell stage but triggered tremendous apoptosis in both epithelial and mesenchymal compartments.Compared to our previous reports using Wnt1-cre; Rosa26R-Fgf8 mice in which incisor and molar germs arrested at the bud stage [23], our present study showed a special tooth agenesis due to both the maxillary and mandibular incisor regression at the later cap stage or early bell stage (Supplementary Figure S1), expanding our putative notion that tooth agenesis could result from the failure not only in initiation but also in morphogenesis.The diminished Shh expression in the incisor epithelium was suggested to enhance apoptosis directly because Shh plays important roles in morphogenesis, cell proliferation, survival and differentiation during tooth development [22].The diminished Shh transcription was implicated to result from the inactivation of Runx2 in dental mesenchyme because Runx2 is believed to activate Fgf3 and Fgf10 expression in dental mesenchyme, which, in turn, activated epithelial Fgf4 and Fgf9 expression to maintain Shh transcription [8,22].On the other hand, since the suppressed Runx2 expression also implicated the reduced FGF signaling in the incisor mesenchyme, the canonical Wnt signaling in the incisor mesenchyme was supposed to be elevated [8,13], which was supported by the enlarged Lef1-expressing domain.Series studies have demonstrated that the increased mesenchymal canonical Wnt activity, especially in the incisor mesenchyme, led to tooth agenesis by inhibiting tooth formation [13,24].Therefore, the ectopic activation of Fgf8 in the dental mesenchyme was suggested to result in incisor agenesis by suppressing epithelial Shh transcription and Runx2 expression but increasing mesenchymal canonical Wnt signaling.Actually, in the E13.5 sub-mandibular salivary glands of Osr2-cre KI ; Rosa26R-Fgf8 mice, the transcription of Shh and Fgf10 in epithelial buds and ducts was also suppressed.Thus, how Fgf8 inhibits the expression of Shh and Runx2 in incisor germs still requires further investigation.
Compared to the incisor agenesis, the microdontia of Osr2-cre KI ; Rosa26R-Fgf8 molars was not so evident in the prenatal stage.The slightly inhibited epithelial Shh expression did not result in excessive apoptosis in the Osr2-cre KI ; Rosa26R-Fgf8 molar germs at the cap or bell stage.On the other hand, the mildly elevated mesenchymal Runx2, which implicated the increased FGF signaling, was suggested to promote cell proliferation [25].Consistently, the obviously reduced Lef1 staining suggested the decreased canonical Wnt activity, and the activation of Fgfr1 implied an increased FGF signaling in E13.5 Osr2-cre KI ; Rosa26R-Fgf8 molar mesenchyme, implicating little impact on the prenatal odontogenic capacity of Osr2-cre KI ; Rosa26R-Fgf8 molars [25,26].However, the persistent activation of Fgf8 resulted in the delayed odontoblastic and premature ameloblastic differentiation of the sub-renal capsule-implanted Osr2-cre KI ; Rosa26R-Fgf8 molar germs, which impaired the following mineralization of enamel and dentin.The extended sub-renal culture indicated that the Osr2-cre KI ; Rosa26R-Fgf8 molar germs lost the typical tooth shape and structure but develop into a miniature tooth with the blurred cusps, which seemed to be mimicking microdontia [21].It is strongly suggested that the persistent Fgf8 expression impaired tooth root development by inactivating Shh transcription [22] and dentin mineralization by suppressing odontoblast differentiation and maturation [23,27].
The differential tooth phenotypes, agenesis and microdontia in Osr2-cre KI ; Rosa26R-Fgf8 incisors and molar germs, respectively, raised the question why dental mesenchymalderived Fgf8 produced the different impacts on incisor and molar development.Our previous study also displayed the regressed maxillary and mandibular incisor germs, as well as the mildly impaired molar germs in Osr2-cre KI ; Ctnnb1 ex3f mice [28], which enforced the speculation that there was a potential difference between the incisor and molar.Actually, we indeed found that the expression of Shh and Runx2 was differential in the incisor and molar germs of Osr2-cre KI ; Rosa26R-Fgf8 mice.However, we still tend to interpret the differential tooth phenotypes by the differential pattern of Osr2-cre KI allele in incisor and molar mesenchyme.At E13.5, Osr2-cre KI was activated throughout the incisor mesenchyme (Supplementary Figure S2), while it was activated only in the lingual mesenchyme of molar germs, which extended to entire molar mesenchyme at E15.5 [29].The early and wider activation of Fgf8 in incisor mesenchyme disabled the early events of tooth development [23], such as fate determination, cell proliferation and survival, leading to the more severe phenotype, tooth agenesis.In contrast, the late and partial activation of Fgf8 in molar mesenchyme allowed tooth development into cytodifferentiation, which resulted in microdontia by disrupting cusp formation and mineralization.Such an interpretation integrated tooth agenesis and microdontia into a single pathogenesis which provides a novel perspective for the molecular mechanism of congenital tooth deformities.
Our study revealed that tooth agenesis could result from not only initiation failure but also the regression at late stage by excessive cell death.Furthermore, we indicated that the persistently active FGF signaling also led to microdontia with impaired root formation and mineralization, which disclosed the comprehensive effects of Fgf8 on tooth development.This finding provides a novel etiological interpretation for tooth agenesis and microdontia by uniting two clinical manifestations into one pathogenesis, which would benefit the therapeutic strategy for tooth regeneration.Due to the recent and latest advances in the application potential of dental stem cells [30][31][32], our study also implicated that FGF signaling had to be finely tuned in the stem cells for tooth regeneration, and only a moderate activity would allow the normal differentiation and maturation of the stem cells.However, how the overactive FGF signaling impacts tooth root development, odontoblast differentiation and dentin mineralization still requires further investigation.

Mouse Lines
The Osr2-cre KI , Rosa26R-Fgf8 and Rosa26R-mT/mG mice were genotyped as described previously [33].All mice were housed and bred in the Specific Pathogenic Free System of the institute of Genome Engineered Animal Models for Human Diseases at Dalian Medical University.Morning of the day of vaginal plug detected in breeding females was designated as E0.5.Animals and procedures used in this study were approved by the Animal Care and Use Committee at Dalian Medical University (Protocol No. AEE18011).

Histological and Cryostat Section
The timed pregnant mice were treated by carbon dioxide and cervical dislocation for euthanasia.The dissected mouse embryonic heads were fixed in 4% paraformaldehyde (PFA) at 4 • C overnight, dehydrated through graded ethanol, embedded in paraffin and then sectioned in 10 µm for Masson staining [33].The implanted molar germs from the sub-renal capsule culture were also subjected to the identical protocol for histological sections but stained by hematoxylin and eosin.For cryostat section, the heads were fixed in 4% PFA, dehydrated with 15% and 30% sugar solution overnight, embedded in O.C.T. and eventually sectioned into slices with a thickness of 10 µm.To ensure the penetrance of tooth phenotype and Cre pattern, the embryonic heads from at least three litters were examined for each stage.

BrdU Labeling Assay
For BrdU labeling assays, the pregnant mice were intraperitoneally injected with 10 mmol/L BrdU at a dosage of 5 mL/kg body weight at 30 min before sacrifice.The harvested embryonic heads were fixed in Carnoy's fixative for 4 h, prepared for paraffin sections in 10 µm according to the above-mentioned protocols.The BrdU-labeled cells were detected using the Detection Kit II (Roche, Basel, Swiss) and counterstained with nuclear fast red solution.The density of BrdU-positive cells in incisor and molar germs was present by the numbers of BrdU labeling nuclei in the defined areas of dental epithelium and mesenchyme, respectively.The BrdU labeling assay was repeated in the samples from three different litters for each stage.

TUNEL Assay
Cell apoptosis within the epithelium and mesenchyme of incisor and molar germs was monitored by TUNEL assay, using the In Situ Cell Death Detection Kit, POD (Roche, Basel, Swiss) on 10 µm thick paraffin sections.Nuclei were counter-stained with DAPI.Cell density in incisor and molar germs was calculated by the numbers of TUNEL-positive cells in the defined areas of dental epithelium and mesenchyme, respectively.The TUNEL assay was performed in at least three repeats in the samples from different litters for each stage.

Statistical Assay
The number of BrdU-or TUNEL-labeled nuclei was counted with ImageJ version 1.51j8.The areas of dental epithelium and mesenchyme in incisor and molar germs were also quantified by ImageJ version 1.51j8.All quantitative data were expressed as means ± Standard Deviation from three independent repeats.Student's t-test was used to determine the statistical significance of the differences, which were considered statistically significant at p-value < 0.05.

In Situ Hybridization
The molar germs of sub-renal culture were harvested in the ice-cold phosphate buffer solution treated with diethyl pyrocarbonate (DEPC).The DEPC-treated 4% PFA was used for fixation.After dehydration in DEPC-treated gradient ethanol, the molar germs were embedded in paraffin and sectioned in 10 µm for in situ hybridization, as previously described [33].Antisense RNA probes for Shh, Ectodin, Amelogenin and Dspp were generated as reported previously [28,29].Eosin was used for counter-staining.For Shh and Ectodin probes, the samples from three different litters were examined.For the probes of Amelogenin and Dspp, three samples were examined for each stage.

X-ray Plain Film for Sub-Renal Capsule Implanted Molar Germs
The E17.5 molar germs were isolated and transplanted under the kidney capsule of adult host mice for 3-day, 5-day, 3-week or 6-week culture, respectively, as previously described [24].The grafted molar germs were collected and fixed in 4% PFA.The X-ray plain films of the implanted molars were obtained by using X-ray radiography (Faxitron MX-20DC12, Faxitron Bioptics, Boston, MA, USA).For each period of sub-renal culture, at least three molar germs were implanted and examined by X-ray radiography.

Conclusions
In summary, our study reported that ectopic activation of Fgf8 in dental mesenchyme by Osr2-cre KI resulted in incisor agenesis and molar microdontia.The incisor agenesis in Osr2-cre KI ; Rosa26R-Fgf8 mice was attributed to the tremendous apoptosis at the early bell stage, which resulted from the suppressed expression of Shh and Runx2 and the increased mesenchymal Wnt canonical signaling.In contrast, although affected a little prenatally, Osr2-cre KI ; Rosa26R-Fgf8 molars only formed a miniature tooth with impaired odontoblast and ameloblast differentiation, and mineralization after sub-renal culture, which resembled microdontia.The severe phenotype in the Osr2-cre KI ; Rosa26R-Fgf8 incisor was most likely attributed to the early and wider activation of Osr2-cre KI in the developing incisor mesenchyme.Thus, it implicated that tooth agenesis and microdontia could result from a single pathogenesis, which displayed the differentially spatial-temporal pattern in tooth development.

Figure 6 .
Figure 6.Structure and histology of the grafted Osr2-cre KI ; Rosa26R-Fgf8 molar germs.After 3 weeks of sub-renal culture, plain X-ray images showed the tooth structure of the grafted E17.5 WT (A) and Osr2-cre KI ; Rosa26R-Fgf8 molar germs (B).HE staining shows the histological features of the E17.5 WT (C) and Osr2-cre KI ; Rosa26R-Fgf8 (D) molar grafts after 3 weeks of sub-renal culture.(C′,D′) are the magnified views in the boxes of (C,D), respectively.After 6 weeks of sub-renal culture, plain Xray images showed the tooth morphology of the grafted E17.5 WT (E) and Osr2-cre KI ; Rosa26R-Fgf8 molar germs (F).The black arrows point to the predentin, and the arrowhead points to the dentin in molar grafts.Scale bar, 100 µm.