TBX1 protein interactions and microRNA-96-5p regulation controls cell proliferation during craniofacial and dental development: implications for 22q11.2 deletion syndrome.

T-box transcription factor TBX1 is the major candidate gene for 22q11.2 deletion syndrome (22q11.2DS, DiGeorge syndrome/Velo-cardio-facial syndrome), whose phenotypes include craniofacial malformations such as dental defects and cleft palate. In this study, Tbx1 was conditionally deleted or over-expressed in the oral and dental epithelium to establish its role in odontogenesis and craniofacial developmental. Tbx1 lineage tracing experiments demonstrated a specific region of Tbx1-positive cells in the labial cervical loop (LaCL, stem cell niche). We found that Tbx1 conditional knockout (Tbx1(cKO)) mice featured microdontia, which coincides with decreased stem cell proliferation in the LaCL of Tbx1(cKO) mice. In contrast, Tbx1 over-expression increased dental epithelial progenitor cells in the LaCL. Furthermore, microRNA-96 (miR-96) repressed Tbx1 expression and Tbx1 repressed miR-96 expression, suggesting that miR-96 and Tbx1 work in a regulatory loop to maintain the correct levels of Tbx1. Cleft palate was observed in both conditional knockout and over-expression mice, consistent with the craniofacial/tooth defects associated with TBX1 deletion and the gene duplication that leads to 22q11.2DS. The biochemical analyses of TBX1 human mutations demonstrate functional differences in their transcriptional regulation of miR-96 and co-regulation of PITX2 activity. TBX1 interacts with PITX2 to negatively regulate PITX2 transcriptional activity and the TBX1 N-terminus is required for its repressive activity. Overall, our results indicate that Tbx1 regulates the proliferation of dental progenitor cells and craniofacial development through miR-96-5p and PITX2. Together, these data suggest a new molecular mechanism controlling pathogenesis of dental anomalies in human 22q11.2DS.

Introduction 22q11.2 deletion syndrome (22q11.2DS) is the unifying term for patients with a common microdeletion on one of the proximal long arms of chromosome 22. This deletion encompasses the genes responsible for DiGeorge syndrome (DGS, MIM#188400), velo-cardio-facial syndrome (VCFS, MIM# 192430) and conotruncal anomaly face syndrome. Characteristic features include congenital heart defects, hypoplasia or aplasia of the thymus and parathyroid and craniofacial dysmorphisms including tooth defects (1)(2)(3). Three research groups identified Tbx1 as the candidate gene for 22q11.2DS based on the analyses of segmental deletions and single gene knockout mice (4)(5)(6). Although the extensive evidence gathered from these mouse studies and information on human TBX1 mutations strongly support TBX1 as the candidate gene involved in 22q11.2DS, the molecular mechanisms underlying the loss or gain of TBX1 function in the pathogenesis of 22q11.2DS is not fully understood.
Tbx1 is a member of the T-box gene family, a group of evolutionarily conserved transcription factors that share a 180-200 amino acid DNA binding domain called the T-box (7). The expression pattern of Tbx1 is consistent with the critical role Tbx1 plays during pharyngeal apparatus formation, heart development and tooth morphogenesis (8)(9)(10). Moreover, mouse studies have associated a progressive reduction in dosage of the Tbx1 mRNA with a non-linear increase in severity of the phenotype (11), and an increase in Tbx1 mRNA dosage with malformations similar to those observed in 22q11.2DS patients (12,13). Recent studies suggest that Tbx1 plays a role in the regulation of several myogenic genes associated with core mesoderm cell survival and fate required for the formation of the branchiomeric muscles (14). Tbx1 Cre fate mapping experiments from E10.5 to E14.5 reveal Tbx1 positive cells in tooth buds and surface ectoderm (15). These findings highlight the need for precise regulation of Tbx1 expression during embryogenesis.
Because DGS patients have dental anomalies we used Tbx1 conditional knockout mice, over-expression mice and Tbx1 Cre mice to determine the molecular basis for dental defects in DGS. The mouse dentition is unique in that the incisor continually grows through the life of the mouse while the molars do not. Each incisor has two cervical loops (CLs), one on the labial side (LaCL) and the other on the lingual side (LiCL). The epithelial stem cells on the incisor reside in the LaCL, which consists of the stellate reticulum (SR), the inner enamel epithelium (IEE) and the outer enamel epithelium (OEE). The self-renewing stem cells localize to the SR and these stem cells will give rise to transit-amplifying (T-A) cells that differentiate into mature enamel-secreting ameloblasts. This process is necessary for matrix deposition and subsequent enamel mineralization (16,17). Many 22q11.2DS patients suffer from enamel hypoplasia, hypomineralization, hypodontia, delayed tooth eruption and excessive dental caries (3). At E11.5, Tbx1 is expressed in the oral epithelium and during early incisor development, it is expressed in the IEE, OEE, CLs and enamel knot (Ek), and at later stages, it is localized to the IEE in molars and incisors (10).
Recent studies have examined the role of microRNAs (miRs) in tooth development. Discrete sets of miRs are expressed in molars compared with incisors, epithelial compared with mesenchymal compartments of the incisors and differentiated ameloblasts compared with cells of the LaCL (18,19). One study compared miRs expressed in the LaCL, the LiCL and enamel-producing ameloblasts (20). These studies confirm that discrete cohorts of miRs regulate the incisor stem cell niche versus ameloblast maturation.
In this report, we show that Tbx1 and miR-96 interact in a negative regulatory loop to maintain the correct dose of Tbx1 in the dental epithelium and that Tbx1 regulates the proliferation of epithelial progenitor stem cells in the LaCL. We demonstrate new Tbx1 protein interactions and a molecular basis for human TBX1 mutations in the regulation of tooth and craniofacial development. This study reveals Tbx1 to play a central role in the pathway that regulates tooth and craniofacial development in adult mice, with changes in Tbx1 dosage in the dental epithelium affecting tooth size, molar cusping, ameloblast differentiation and enamel production.
Results miR-96-5p regulates Tbx1 expression in the dental epithelium miRs play a critical role in the regulation of tooth stem cell proliferation and differentiation (18,(20)(21)(22)(23). A schematic representation of the mouse lower incisor (LI) and cells that populate the growing incisor during tooth development, including the LaCL (the stem cell niche), is shown (Fig. 1A). The mouse dental epithelium (light blue, which includes the LaCL and ameloblasts, Am) was extracted and the LaCL cells were isolated from the ameloblast cells and used to analyze miR expression. Analyses of miR expression during LI development at P0 showed low miR-96-5p expression in the LaCL, however, miR-96-5p expression increased >2-fold in the differentiating pre-ameloblast and ameloblast cells (Fig. 1B). The Tbx1 3′UTR contains a highly conserved miR-96 binding element, which was cloned into a luciferase reporter to assess miR-96 function (Fig. 1C). miR-96 repressed luciferase expression from the WT Tbx1 3′UTR, but not from a mutated Tbx1 3′UTR, in LS-8 oral epithelial cells (Fig. 1D). Over-expression of miR-96 in LS-8 cells repressed endogenous Tbx1 expression, as shown by real-time PCR (Fig. 1E). Western blots of miR-96 transfected LS-8 cells demonstrated decreased Tbx1 protein, while cells transfected with empty vector or a scrambled miR did not show a change in Tbx1 expression (Fig. 1F). Together these data demonstrate that miR-96 represses Tbx1 and correlates with high levels of Tbx1 expression in the LaCL (low levels of miR-96). Interestingly, a miR screen in Tbx1 over-expression mice (COET K14Cre ) mandibles revealed a decrease in miR-96 expression compared with wild-type (WT). Real-time PCR confirmed decreased miR-96 expression in COET K14Cre mice mandible (Fig. 1G). Thus, we have tentatively identified a Tbx1-miR-96 feedback loop, where miR-96 represses Tbx1 and Tbx1 represses miR-96 expression.
The Tbx1 N-terminus is required for repression of PITX2 transcriptional activity We have previously shown that Tbx1 interacts with the PITX2 C-terminus to repress PITX2 transcriptional activity (24). However, the Tbx1 domain for protein interactions was not known. We generated a series of Tbx1 truncated proteins to test for PITX2 interactions and transcriptional activity ( Fig. 2A). GST-Tbx1 pull-down experiments demonstrate that PITX2 binds to the N-terminus of Tbx1 (Fig. 2B). PITX2 protein (500 ng) was incubated with GST-Tbx1 FL (full-length) and truncated proteins to determine the protein interaction domain of Tbx1. PITX2 bound to Tbx1 FL, Tbx1 ΔC and Tbx1 ΔTC, but not to Tbx1 T-box or Tbx1 ΔNT. PITX2 did not bind to Tbx1 ΔN (data not shown). The PITX2 binding domain was localized to the N-terminus of Tbx1 ( Fig. 2A and B).
PITX2 activates the mouse Pitx2c promoter in LS-8 oral epithelial cells and auto-regulates its expression (Fig. 2C) (24). Tbx1 FL, Tbx1 ΔC and Tbx1 ΔN proteins do not activate the Pitx2c promoter; however, both Tbx1 FL and Tbx1 ΔC repress PITX2 activation of the Pitx2c promoter (Fig. 2C). Deletion of Tbx1 N-terminus (Tbx1 ΔN) does not activate the Pitx2c promoter and does not repress PITX2 transcriptional activity, as would be expected as the Tbx1 N-terminus interacts with PITX2 (Fig. 2C). The Tbx1 truncated proteins were expressed in LS-8 cells (Fig. 2D). Thus, the Tbx1 N-terminus is a site for protein interactions and we show that the Tbx1 N-terminus is required to repress PITX2 transcriptional activity. Furthermore, cells transfected with Tbx1 FL showed decreased endogenous Pitx2 expression (Fig. 2E).
Human TBX1 mutations have variable activity TBX1 mutants associated with 22q.11.2DS bind DNA and we have shown that Tbx1 interactions with PITX2 do not alter PITX2 DNA binding (24)(25)(26)(27). However, one report shows that TBX1 mutant proteins F148Y, H194Q and G310S activate an artificial promoter, with TBX1 H194Q having increased transcriptional activity compared with WT TBX1 (25). Two other reports using artificial reporters (CAT and luciferase reporters with T-Box sites or Brachyury consensus binding site sequences) show reduced or no activation of the reporters with the mutant proteins compared with WT TBX1 (26,27). We also tested TBX1 and mutants with an artificial luciferase reporter (contains T-Box elements from the FGF promoter) and found little activation by WT or mutant proteins (data not shown). However, we asked if the TBX1 mutants altered 96 differentially expressed between the LaCL (stem cell niche) and Am (differentiating cells) region. Five biological samples were assayed for each region, and one sample is shown. (C) The Tbx1 3′UTR miR-96 binding site is highly conserved among species. The sequence of this region that is mutated in the Mut Tbx1 3′UTR (miR-96 seed seq.; underlined) results in an inability to bind miR-96. (D) Normalized luciferase activity of the 3′-UTR Tbx1-luciferase reporter (WT Tbx1 3′UTR) in the presence of empty plasmid (Vector) or CMV-miR-96 (miR-96) shows that luciferase activity is lost when miR-96 is expressed; this is not the case when the miR-96 seed sequence is mutated (Mut Tbx1 3′UTR). Error bars indicate ± SEM, five independent experiments (n = 5); P < 0.001. (E) Expression of miR-96 in LS-8 oral epithelial cells repressed endogenous Tbx1 expression. Real-time PCR experiments measured Tbx1 transcripts with and without miR-96 expression and normalized to control gene (N = 3). (F) Western blot analysis shows that Tbx1 levels decrease when miR-96 is over-expressed in LS-8 oral epithelial-like cells. GAPDH served as a loading control. (G) Tooth germ RNA isolated from WT and Tbx1 K14COET mice mandibles demonstrated significantly decreased miR-96 expression. Real-time PCR was performed on three biological samples and each experiment was performed in triplicate (N = 3). PITX2 transcriptional activity. A schematic representation of the TBX1 gene and mutations is shown (Fig. 3A). TBX1 has three isoforms, A, B and C with C being the most conserved between mice and humans and the most highly expressed in humans (28,29).
To determine if these mutants repressed PITX2 transcriptional activity PITX2 was transfected at 1 μg and TBX1 plasmids at 0.25, 0.5 and 1 μg s, respectively. Interestingly, human TBX1 variant C (VC) showed a slight activation of the Pitx2c promoter in LS-8 cells (Fig. 3B). However, both TBX1 VC and TBX1 G310S (G-S) mutant proteins repressed PITX2 transcriptional activation of the Pitx2c promoter (Fig. 3B). TBX1 G-S and TBX1 H194Q (H-Q) mutant proteins did not activate the Pitx2c promoter. Interestingly, the TBX1 H-Q protein did not repress PITX2 activation (Fig. 3B). In previous reports, the stability of the TBX1 mutant proteins was not analyzed, and we found that both TBX1 G-S and H-Q were expressed and stable in the LS-8 cells. Because PITX2 interacts with the N-terminus of Tbx1 and Tbx1 interacts with the C-terminus of PITX2 the loss of TBX1 repression of PITX2 with the H-Q mutation suggests other protein functions. Further experiments are required to determine the exact mechanisms. These data could explain phenotypic variations among 22q.11.2DS patients.

Tbx1 binds to the miR-96 promoter and represses miR-96 expression
To confirm Tbx1 regulation of miR-96, a chromatin immunoprecipitation (ChIP) assay was performed to demonstrate endogenous Tbx1 binding to the miR-96 chromatin. A Tbx1 binding site was identified 3251 base pairs upstream of the miR-96 transcription start site (Fig. 4A). This Tbx1 binding site is similar to a recent report that identified Tbx1 binding sites by SELEX (27). As a control, primers were designed to a site upstream of the Tbx1 binding element in the miR-96 promoter and this DNA was not immunoprecipitated (IP'ed) by IgG serum or Tbx1 antibody (Fig. 4A, lanes 2 and 3, respectively). However, the Tbx1 antibody The PITX2 bound protein was resolved on 10% PAGE gel transferred to PVDF filters, immunoblotted and detected using PITX2ABCDE antibody (Capra Science, Sweden) and ECL reagents. PITX2 bound to Tbx1 full-length (FL), Tbx1ΔC (C-terminus deleted) and Tbx1ΔTC (T-box and C-terminus deleted). PITX2 did not bind to the Tbx1 T-box (Tbx1T-box) or Tbx1 C-terminus (Tbx1ΔNT), also PITX2 did not bind to Tbx1ΔN (data not shown). (C) Tbx1 truncations were tested in transfection assays to determine their activity and ability to repress PITX2 transcriptional activation of the Pitx2c promoter. As expected Tbx1 activated the Pitx2c promoter at low levels and the Tbx1ΔC and Tbx1ΔN proteins did not activate the promoter. However, deletion of the Tbx1 N-terminus (Tbx1ΔN) did relieve the repressive effect of Tbx1 on PITX2 transcriptional activation of the Pitx2c promoter. Luciferase activity is shown as mean-fold activation compared with activity in the context of the empty expression plasmid (Vector). All luciferase activities were normalized to β-galactose expression; five independent experiments were performed in LS- The miR-96 promoter was cloned (∼5 kb) into the luciferase vector to test for Tbx1 functional regulation. Murine Tbx1, human TBX1 VC and TBX1 G-S all repressed the miR-96 promoter (4-fold or greater, P < 0.05), while TBX1 H-Q had no effect (Fig. 4B). Furthermore, because TBX1 H-Q binds DNA (27), the inability of this mutant to repress the miR-96 promoter could suggest defective protein interactions, other than PITX2. Thus, Tbx1 directly represses miR-96 expression and this is the first demonstration of Tbx1 repression of a miR.

Specific Tbx1 expression in the incisor CL controls incisor development
The rodent incisor is a unique model for the differentiation of enamel organ cells from stem cells to enamel-secreting ameloblasts (Fig. 5A). Stem cells located in the LaCL give rise to the pre-secretory, secretory and maturation-stage epithelial or ameloblast cells.
Tbx1 was conditionally knocked out using the K14 Cre mouse crossed to the Tbx1 f/f mouse to generate the Tbx1 K14cKO mice. The K14 promoter is active in surface ectoderm and basal cells from embryonic day E9.5 in developing hair follicles and tooth epithelia (30)(31)(32). At P0, the LI LaCL was smaller and disorganized in Tbx1 K14cKO mice compared with WT, and the inner enamel epithelial (IEE) cells appeared undifferentiated and not well polarized ( Fig. 5B and C). Furthermore, the OEE and stratum intermedium (SI) layer was thin, unorganized and lacked structure (Fig. 5C).
Tbx1 transcripts were specific to the dental epithelium in both the incisors and molars and were detected from E10.5 to E18.5 (10,33,34). Tbx1 protein expression in teeth was confirmed by immunofluorescence of the LI at E16.5 and Tbx1 was expressed in the myogenic core of the tongue (t), dental follicle, dental lamina, oral epithelium and in both the LaCL and LiCL (Fig. 5D-I). Tbx1 expression was seen in the IEE, OEE and SR of the LaCL ( Fig. 5H and I). However, Tbx1 expression decreased in differentiating ameloblasts and was not seen in the odontoblasts.
The Tbx1 Cre mouse was crossed with the ROSA26 LacZ reporter mouse to better understand Tbx1 cell fate during incisor development. Fate mapping of Tbx1-expressing cells revealed Tbx1expressing cells specifically in the LaCL of the LI. High magnification imaging of the LaCL indicated that Tbx1-expressing cells populated the distal region of the SR, and in the OEE and IEE (Fig. 5J). The stem cells located in the SR compartment intercalate into the OEE and move around the CL to the IEE, where they divide and migrate to the distal part of the growing incisor (35) (see arrows, Fig. 5J). Interestingly, cells in the proximal SR were Tbx1 negative (Fig. 5J). Tbx1 daughter cells were observed in the differentiating ameloblasts or secretory cells and in the SI (Fig. 5K). Overall, these results suggest that Tbx1 marks a specific subset of dental epithelial stem cells in the LaCL, different and independent from Sox2 (36).

Abnormal tooth development and amelogenin expression in Tbx1 K14cKO mutant embryos and neonate mice
The differentiation of dental epithelial cells into ameloblasts occurs through several morphological stages, over this period the cells become elongated and polarized, features that are required for the deposition of enamel (37).
In this report we used Tbx1 conditionally deleted mice, K14 Cre X Tbx1 flox/flox (Tbx1 K14cKO ) embryos to study tooth and craniofacial morphogenesis. Hematoxylin and eosin (H&E) staining of sagittal sections of the craniofacial region at E16.5 demonstrated a delay in upper incisor (UI) and LI morphogenesis in Tbx1 K14cKO mutant mice compared with WT counterparts (data not shown). The Tbx1 K14cKO embryos at E18.5 had small LIs (∼25% decrease in size, black bar compared with WT) and small LaCL regions ( Fig. 6A and B). Higher magnification of the differentiating dental epithelium (ameloblasts, Am) and dental mesenchyme (odontoblasts, Od) revealed only minor defects in both Od and Am polarization and differentiation of the LI at E18.5 in the Tbx1 K14cKO embryos ( Fig. 6C and D). However, the small tooth size could have resulted from a lack of stem cell proliferation in the LaCL.
Ameloblasts are responsible for the secretion of the three structural enamel matrix proteins, amelogenin, ameloblastin and enamelin. Amelogenin constitutes ∼90% of the enamel organic matrix and is highly conserved across species (38)(39)(40). Amelogenin and ameloblastin are essential for proper enamel formation (41)(42)(43)(44). At P1, amelogenin was decreased in Tbx1 K14cKO LIs compared with WT ( Fig. 6E-H). At P4, amelogenin levels in the incisors remained low or absent in the mutant mice (data not shown). The low expression of amelogenin can cause enamel defects.
H&E staining of the lower molars (LMs) at E16.5 revealed a delay in early bell stage morphogenesis in Tbx1 K14cKO embryos (data not shown). The Ek, the region where epithelial-mesenchymal signaling regulates tooth size and shape, was normal in WT mice, but underdeveloped or absent in Tbx1 K14cKO mice (data not shown). The Ek is an organizer of cusp formation and sagittal sections of P0 lower molars demonstrated defective molar cusping of the first molar (M1) in the Tbx1 K14cKO mice compared with WT mice (Fig. 6I and J). Higher magnification showed defective odontoblast (Od) and ameloblast (Am) differentiation in the Tbx1 K14cKO lower molars ( Fig. 6K and L). At P4, amelogenin levels in the molars remained low or absent in the mutant mice ( Fig. 6M-P). Interestingly, ameloblastin was not diminished in Tbx1 K14cKO P4 molars and incisors (data not shown). Thus, Tbx1 K14cKO neonate mice have severely reduced amelogenin expression, but not completely absent expression.

Tbx1 regulates epithelial cell proliferation
Ki67 immunohistochemistry (IHC) was carried out on P0 WT and Tbx1 K14cKO embryos to determine if decreased cell proliferation occurred at this later stage of development. The ratio of Ki67-positive cells to the total cell number [2-(4-Amidinophenyl)-6-indolecarbamidine dihydrochloride, 4′,6-Diamidino-2-phenylindole dihydrochloride (DAPI) stain] in the LaCL was calculated to estimate the proliferation ratio. In the LI, the value for the WT LaCL was 75% and that for the mutant was 53% (Fig. 7A). These data demonstrate that the proliferation of dental stem cells in the LaCL is significantly lower in P0 mutant compared with WT neonate mice.
Mouse embryo fibroblasts (MEFs) collected from WT and Tbx1 −/− mice at E14.5 were plated at 100,000 cells per 60-mm  (Fig. 7D). Taken together, these data suggest that Tbx1 regulates cell proliferation, and that Tbx1 in the epithelium alone regulates tooth size by controlling cell proliferation and differentiation.
Tooth size, shape and enamel formation are reduced in adult Tbx1 K14cKO mice Skeletal preparations of P14 WT and Tbx1 K14cKO mice were made, and scanning electron microscopy (SEM) and microCT (μCT) scans were carried out on the samples. The LIs were shorter in the mutant compared with the WT mice and the enamel layer thinner (Fig. 8A-D). At a higher magnification, the incisal edge appeared worn down or chipped in the mutant, indicating a potential decrease in enamel mineralization and/or structural defects. SEM imaging of incisors that were fractured perpendicular to the growth axis in the erupted portion of the tooth showed that the enamel layer was thinner in the mutant than the WT mice ( Fig. 8E and F). At higher magnification, it became evident that the orientation of the enamel crystallite bundles ( prisms) is normal in the mutant mice. However, the prisms are less densely packed and separate easily from the underlying dentin at the enamel dentine junction (Fig. 8G and H).
At P14, the first molar (M1) was smaller in the mutant (Fig. 8I-L). Also, the second molar (M2) lacked a distal cusp (Fig. 8N, O, Q, R, arrow). The cusps of the first two molars in the mutant mice were not as sharp or deep compared with WT mice (Fig. 8Q and R) and development of the third molar (M3) was more advanced in the mutant compared with WT mice (Fig. 8M-T). Overall, molars of the Tbx1 K14cKO mice were smaller ( Fig. 8J and K) and featured decreased enamel formation (Fig. 8I, L, M and P), malformed cusping of the first two molars and premature growth of the third molar. The increased growth of the third molar was surprising, but may correlate to the lack of Pitx2 repression, which initiates dental development.
Tbx1 over-expression regulates incisor size and dental epithelial stem cell proliferation Analyses of E16.5 LIs of the K14 Cre activated Tbx1 over-expression mouse (COET K14Cre ) demonstrated a larger LaCL with less differentiated cells (Am) compared with WT mice (Fig. 9A-D). The LI of the COET K14Cre mice is larger than that of WT at P1, with an increase in the width and length of the LaCL (Fig. 9E-H). Cell proliferation was measured by Ki67 staining of the P1 LI LaCL and analyses of several mice (N = 3) revealed an increase in cell proliferation in the COET K14Cre mice (Fig. 9I-L). These data are consistent with loss of Tbx1 function resulting in small incisors and decreased cell proliferation in the CL regions or stem cell niche. Thus, the expression and dose of Tbx1 control dental stem cell proliferation and differentiation.
Incisor growth, amelogenin expression and enamel formation are increased with Tbx1 over-expression We asked if Tbx1 over-expression increased amelogenin expression and if it correlates with an increase in enamel formation in the COET K14Cre mice. At P1, both UIs and LIs showed increased amelogenin expression in the proximal regions compared with WT mice (Fig. 10A-H). These results suggest that Tbx1 transcriptional mechanisms may regulate amelogenin or that Tbx1 expression expands cell proliferation and increased cell differentiation.
To understand tooth structure and mineralization, we analyzed the incisors and molars of 2-week-old COET K14Cre mice by μCT. Tbx1 over-expression resulted in a larger incisor with increased enamel formation compared with WT ( Fig. 10I and J; see arrows), and increased enamel formation (red) in the molars ( Fig. 10K and L; see arrows). In COET K14Cre mice at 4 weeks of age, the mandibles showed a decrease in alveolar bone development ( Fig. 10M and N; light green arrow and orange bracket). Enamel thickness of the LI was increased in the COET K14Cre mouse ( Fig. 10M-P; blue arrow). Development of the third molar ( Fig. 10O and P; yellow arrow) was delayed and cortical bone formation was decreased ( Fig. 10O and P; white arrow). These results are consistent with a role for Tbx1 in regulating tooth size, shape and dental epithelial cell differentiation, which leads to enamel formation. Interestingly, third molar development was decreased in the Tbx1 over-expression mouse, which correlates to Tbx1 interaction with PITX2 to repress its transcriptional gene network required for normal dental development.

Tbx1 loss of function and gain of function embryos have cleft palate
The E16.5 Tbx1 K14cKO embryos exhibited a cleft palate. In our Tbx1 conditional deletion mice the palatal shelves have elevated but do not fuse ( Fig. 11A and B). Interestingly, Tbx1 over-expression also caused cleft palate in COET K14Cre mice ( Fig. 11C and D). This is consistent with a previous report showing cleft palate in embryos over expressing Tbx1 (COET Ap2Cre ) in the surface ectoderm (45). The palatal shelves appeared to elevate but did not fuse at the midline.

Discussion
The cloning and characterization of Tbx1 in mice established this gene as essential for embryonic development. To understand the role of Tbx1 in odontogenesis, gene and miR expression was analyzed in WT, Tbx1 K14cKO and COET K14Cre embryonic stage-specific mouse mandibles, maxilla and dental epithelial tissue. Bioinformatics analyses of the expression data revealed genes and miRs regulated by Tbx1. Comparison of increased gene expression with decreased miR expression or the inverse in these tissues revealed tentative correlations of miR-regulated gene expression controlled by Tbx1. A regulatory loop was identified between Tbx1 and miR-96, which further correlated with their expression patterns in the developing incisor epithelium. To further establish a potential link to 22q11.2DS, several TBX1 mutants were assayed for their transcriptional activity and their ability to regulate the genes and miRs identified in the bioinformatics screens. The staining of WT and COET K14Cre P1 LIs. At this later stage, the COET K14Cre LI is larger (width and length) than its WT counterpart and the LaCL remains increased in size and shape (the LaCL is elongated in the COET K14Cre incisor). Scale bar = 500μm. There are more cells in the SR region of the CL in the COET K14Cre incisor (outlined with white dotted line, higher magnification G and H). Scale bar = 100 μm. (I-L) P1 LIs from WT or COET K14Cre mice were processed, sectioned and stained for Ki67 to assess cell proliferation. Ki67 expression was higher in the COET K14Cre LaCLs compared with WT, as established by quantifying the Ki67/DAPI ratio; 10 l of pups and 20 mutants were sectioned, all showed the same phenotype. Abbreviations: Am, ameloblast; t, tongue; LI, lower incisor; LaCL, labial cervical loop.
TBX1 mutations were analyzed in cell-based assays to understand their function compared with WT TBX1. The in vivo bioinformatics approach identified direct targets of Tbx1 in mouse models for 22q11.2DS and these targets could account for the molecular underpinnings of dental and craniofacial anomalies observed in DiGeorge patients. Clearly some Tbx1 molecular mechanisms between humans and mouse models are different however using the approach in this report revealed new genetic pathways potentially associated with 22q11.2DS. TBX1, 22q11.2DS and associated dental anomalies TBX1 is a candidate gene for 22q11.2DS and is responsible for the majority of the phenotypes seen in 22q11.2DS patients. Various clinical studies have shown that 22q11.2DS patients have tooth defects, ranging from hypodontia to enamel defects (3). Independent of the role of TBX1 in the pharyngeal apparatus, epithelial Tbx1 expression in a maturing tooth has been shown to be specific to the IEE, and cells in this region become mature ameloblasts that secrete enamel (10).
Dental anomalies such as enamel hypoplasia and hypomineralization, hypodontia and aberrant tooth shape are documented in 22q11.2DS patients (3). Tooth defects in 22q11.2DS patients have been linked to hypocalcemia from hypoplasia of the parathyroid, and by micrognathia. Traditionally, enamel disturbances in 22q11.2DS patients were thought to be secondary effects of hypocalcemia caused by hypoparathyroidism. A recent study concluded that a diagnosis of hypoparathyroidism did not affect the prevalence of enamel anomalies (46). Thus, our research demonstrates that the dental and craniofacial defects in 22q11.2DS patients involve a gene regulatory network modulated by Tbx1 regulating cell proliferation and differentiation.

Tbx1-protein interactions regulate development
Pitx2, a bicoid/paired-related homeobox gene, was initially identified as the mutated gene in the autosomal-dominant, haploinsufficient Axenfeld-Rieger syndrome (47). Patients with this disorder display many tooth abnormalities, including dental hypoplasia, abnormally shaped teeth and anodontia vera. Within the craniofacial region, Pitx2 is the earliest detected transcription factor in the oral epithelium, and the expression patterns of Tbx1 and Pitx2 overlap during tooth morphogenesis (48,49). Tbx1 represses PITX2-mediated activation of the cyclin-dependent kinase inhibitor p21 in teeth by physically interacting with the PITX2 C-terminus, providing a molecular mechanism for the proliferation of dental epithelial cells (24).
A recent study identified a hierarchical network of transcription factors expressed in the pharyngeal mesoderm that coordinates both heart and craniofacial development. This network includes genetic interactions between Tbx1, Pitx2, Lhx2, Tcf21 and bHLH genes (14,50). This study suggests that Tbx1 levels can be fine-tuned by interactions with other transcription factors, and that these factors may be modifiers for 22q11.2DS (50).
We have dissected the role of Tbx1 protein domains and their interaction with PITX2 to demonstrate that the Tbx1 N-terminus is required for PITX2 binding and repression of PITX2 transcriptional activity. Tbx1 repression of PITX2 during tooth development may regulate tooth initiation and the size and shape of both incisors and molars. We have shown previously that Pitx2 −/+ /Tbx1 −/+ double het mice form an extra premolar, demonstrating that these two factors interact genetically to regulate tooth initiation and formation (24). In this report, we show that Tbx1 K14cKO mice have increased third molar development while the COET K14Cre mice have decreased third molar development, suggesting that the dose of Tbx1 regulates tooth initiation and the timing of tooth development. Because TBX1 is a potent regulator of PITX2 transcriptional activity, which initiates tooth development, the TBX1-PITX2 interaction appears to control tooth initiation and patterning. Tbx1 is expressed early during tooth development and is co-expressed with Pitx2 in the developing incisor and molar. Tbx1 and Pitx2 are early regulators of a gene expression network that define cell proliferation and differentiation of several cell types.
Human TBX1 and DGS associated mutations also regulate PITX2 transcriptional activity. Both TBX1 H194Q and TBX1 G310S proteins are stable and not degraded in cells. TBX1 G310S represses PITX2 activation but at reduced levels compared with WT TBX1; however, TBX1 H194Q has no effect on PITX2 activity. The TBX1 N-terminal tail is highly conserved in all three isoforms while the C-terminal tail varies (51). Because the interaction between PITX2 and TBX1 is crucial for embryonic craniofacial development, it makes sense that PITX2 binds to a highly conserved portion of TBX1. Both TBX1 H194Q and G310S mutations were predicted to affect TBX1 DNA binding and protein stabilization (25), however neither is affected in the mutant proteins, (27) and our study. However, both mutations may alter dimer formation and/or protein function (25). However, these mutations do not appear to change the N-terminal structure of the TBX1 protein, which binds PITX2. Thus other mechanisms are likely responsible for their differential transcriptional activity.

Tbx1 in palatogenesis
Craniofacial malformations occur in more than half of 22q11.2DS patients, and cleft palate (complete, submucosal and soft) is one of the most frequent features (52). Tbx1-null mice exhibited abnormal epithelial adhesion between the palate and the mandible, leading to clefts similar to those observed in 22q11.2DS (53,54). However, during palate development it was suggested that Tbx1-null epithelium was hyperproliferative and did not differentiate and that Tbx1 over-expression inhibited cell growth (53). Funato et al. suggested that Tbx1 regulated the balance between proliferation and differentiation in the epithelium of the palatal primordial (53). Our studies demonstrate that both loss and gain of Tbx1 function causes clefting supporting the dosedependent regulation of palatogenesis and development by Tbx1. However, during odontogenesis Tbx1 acts to increase epithelial cell proliferation consistent with a role for Tbx1 in regulating dental epithelial progenitor cells. This is also consistent with the role of Tbx1 in cardiac progenitor cells where it increases proliferation and inhibits differentiation (55).

Tbx1 in odontogenesis
The CL regions (stem niche) of both UIs and LIs displayed decreased proliferation in Tbx1 K14cKO mice and increased proliferation in the Tbx1 K14COET mice. These data demonstrated that Tbx1 is essential for maintenance of dental progenitor cells. When Tbx1 (Tbx1 K14cKO ) was conditionally deleted from the oral and dental epithelia, we observed microdontia, underdeveloped CLs and defective ameloblast differentiation and these effects can be explained by a decrease in the proliferation of progenitor cells.
The size of the developed incisors and molars appeared smaller in the Tbx1 K14cKO mice at P14 and P20. In molars a distal cusp is missing in the second molar. These abnormalities may be explained by decreased progenitor cells in the LaCL and delayed formation of the Ek. Both incisors and molars show wear of the enamel layer, and on close examination, the enamel prisms formed are less dense than in WT mice, especially in the region closest to the enamel-dentin junction (EDJ). Tbx1 regulates the timing of ameloblast differentiation and the subsequent production of enamel proteins, and the effect seen is a delay and not a complete absence of enamel production.
Tbx1 expression has been shown to be restricted to the IEE at E12.5 and is maintained by mesenchyme-derived Fgf signaling (56). Tbx1 and Fgf8 interact genetically during development (57). However, using Tbx1 Cre fate mapping we demonstrate that Tbx1 expressing cells are located in a unique and distinct region of the LaCL during incisor development. It appears that Tbx1 expressing cells are located in a defined region of the SR that may define both lineages of the pre-ameloblast cells and cells of the OEE and SI. Sox2 expressing cells mark a region of the LaCL that appear not to express Tbx1 (36). We define this region as Tbx1 negative, which suggests that Tbx1 marks a cell lineage separate from Sox2. However, more experiments are required to determine the exact differences.
Tbx1 and microRNA regulation miRs have been shown to be essential regulators of embryogenesis. Bmp signaling promotes outflow tract (OFT) myocardial differentiation by regulating miRs (58). Bmp signals through a conserved Smad-binding element to regulate miR-17-92, which results in decreased Tbx1 expression. Smad1 is a critical negative regulator of SHF proliferation in vivo, and ablation of Smad1 in the SHF enhances cell proliferation (59). Tbx1 binds to Smad1 and Figure 12. Model for the role of Tbx1 in tooth and craniofacial development. PITX2 and TBX1 are two of the first transcription markers for dental development and both a co-expressed in the early dental epithelium, dental lamina and oral epithelium. PITX2 is a transcriptional activator, which activates a gene regulatory network required for dental development (62). TBX1 can repress PITX2 transcriptional activity, but also activate other genes required for cell proliferation. TBX1 is part of a negative feedback loop with miR-96. TBX1 represses miR-96 expression and miR-96 represses TBX1 expression. This feedback loop allows dental epithelial cells to differentiate and produce ameloblasts, which express amelogenin.
negatively modulates the Bmp-Smad signaling pathway by interfering with the Smad1-Smad4 interaction (45). A bidirectional negative feedback loop connecting Bmp, Smad, miR-17-92 and Tbx1 may regulate heart development. A recent paper showed that Tbx1 acts as an on-off switch for Bmp signaling in the hair follicles, acting as a regulator of stem cells transitioning between the quiescent and proliferative states (60). Furthermore, a nice study recently demonstrated that miR-17-92 null mice have craniofacial defects including cleft palate and that Tbx1 was a target of miR-17-92 (61). Tbx1 also inhibits Bmp signaling in the incisor LaCL to prevent immature ameloblast differentiation (unpublished data).
We hypothesized that miRs play a role in ameloblast proliferation and differentiation by regulating the expression of Tbx1. miR-96 is expressed in the mouse incisor LaCL, but at low levels. This might reflect regulation of its expression by factors in the LaCL (20). We demonstrate that Tbx1 repressed miR-96 expression, which maintains the expression of miR-96 at low levels in the LaCL. Such control is important for Tbx1 regulation of the dental stem cell niche (Fig. 12). However, increased miR-96 expression may be required to repress Tbx1 to produce differentiated ameloblasts, but we cannot rule out that other factors may either activate miR-96 or repress Tbx1 expression in ameloblasts. miR-96 indirectly regulates Pitx2 expression and transcriptional activity by inhibiting Tbx1 expression and the ability of Tbx1 to repress Pitx2 transcriptional activation. This would modulate Pitx2 transcriptional activity and fine-tune dental and craniofacial development and the gene regulatory network activated by Pitx2. miRs act as modulators of gene networks to define the timing of gene expression and the patterns of expression. We demonstrate that miR-96 defines the expression pattern of Tbx1 in the LaCL, where Tbx1 expression is high and miR-96 is low. However, in the ameloblasts where miR-96 expression increased it facilitates decreased Tbx1 expression and cell differentiation. This is a great example of the dose response of miRs and transcription factors in regulating cell proliferation versus differentiation during development.
TBX1 G310S mutant protein represses the miR-96 promoter, however, TBX1 H194Q does not repress or activate the miR-96 promoter. The change in miR-96 expression could modulate craniofacial and dental development. Human mutations in the miR-96 seed region, disrupting its function, are associated with progressive hearing loss through the modulation of gene expression in hair cells affecting their normal function (63). Therefore, the altered TBX1-miR-96 regulatory loop is a candidate to explain the dental defects such as enamel hypoplasia, hypomineralization, hypodontia and aberrant tooth shape documented in 22q11.2DS patients.
Tbx1 regulates dental stem cell proliferation and differentiation controlled by miR-96. Our data suggest that Tbx1 regulates a unique set of progenitor cells during dental development. We have uncovered potential molecular underpinnings for the tooth defects in patients with 22q11.2DS. The human TBX1 mutations result in altered proteins that cannot effectively repress PITX2 activity and modulate miR-96 expression.

Animals
All animals were housed at the University of Iowa, in the Program of Animal Resources and were handled in accordance with the principles and procedure of the Guide for the Care and Use of Laboratory Animals. All experimental procedures were approved by the University of Iowa IACUC guidelines. Tbx1 +/− (64), Tbx1 flox/+ (65), Tbx1 Cre (15,66), COET (Conditional Over-expression of Tbx1) (12), K14 Cre (18) and ROSA26 LacZ (Jackson Labs) were previously described. Observation of a vaginal plug was counted as embryonic (E) day 0.5, and embryos were collected at E14.5, E16.5, E18.5, P0 and P4. Mice and embryos from WT, K14 Cre ; Tbx1 flox/flox (Tbx1 K14cKO ) were genotyped from DNA extraction of tail biopsies. Mice and embryos were genotyped by PCR using DNA extracted from tail biopsies and previously published PCR primers.

Cell proliferation and MTT assays
Mouse embryonic fibroblasts (MEFs) were obtained from E14.5 WT and mutant mice. MEFs were passaged twice and plated at 100 000 cells per 60-mm cultured plate. Cell numbers were counted at 24, 48 and 72 h time points. Tbx1, empty vector or non-transduced LS-8 cells were plated in triplicate at 5000, 2500, 1250 and 625 cells/well in 96 well plates. After 4 h, when 98% of the cells had adhered, the media was removed from one plate (0 h) and 100 µl of media + MTT (Thiazolyl blue tetrazolium bromide, Sigma) were added to 0.12 m and allowed to incubate at 37°C for 4 h. After 4 h, 100 µl of 1% SDS and 0.1 N HCl were added to the wells and the plate was further incubated at 37°C for 4 h. Plates were then analyzed on a microtiter plate reader at 570 nm. Plates were compared for 0, 24 and 48 h. Values at 0 h were set to 1 and the fold increase was calculated at 24 and 48 h. Each point cell dilution was done in triplicate and the experiment was done twice (n = 2).

Histology and IHC
Embryonic heads were fixed in 4% paraformaldehyde (PFA) for 0.5-4 h at room temperature (RT). Samples were dehydrated with increasing concentrations of ethanol, followed by xylene, embedded in paraffin and sectioned at 7-μm thickness. H&E staining was used to examine craniofacial and tooth morphology. For IHC, sections were de-paraffinized and boiled with 0.1  sodium citrate buffer for 15 min and cooled in solution to RT. Slides were blocked with 10% serum for 1 h followed by overnight incubation at 4°C with anti-amelogenin (Santa Cruz, 1:200), antiameloblastin (Santa Cruz, 1:200), anti-Tbx1 (Invitrogen, 1:200) and anti-Ki67 (Abcam, 1:200). For fluorescein immunochemistry (IF), secondary antibodies conjugated to FITC were used at a dilution of 1:200 (Invitrogen). DAPI was used for counter staining. For IHC, slides were treated with a biotinylated goat anti-rabbit IgG conjugate (Vector Labs, 1:200) using the avidin-biotin complex (Vector Labs) and an AEC staining kit (Sigma).

Fluorescence immunocytochemistry
For cell-based IF, cells were seeded on cover slides for 24 h, followed by fixation with cold acetone for 5 min. Cells were washed twice with PBS-Tween (PBS-T) and incubated with 10% goat serum for 30 min at RT and then incubated with 1/500 Myc-tagged Ab (Cell Signaling) or 1/250 Beta-catenin Ab (Santa Cruz) for 2 h at RT. Cells were washed three times in PBS-T and then incubated with Alexa-488 or Alexa-555 at 1/250 for 30 min at RT. Finally, the cells were washed three times with PBS-T for 10 min each, and counter stained using mounting solution containing DAPI.
Cell culture, transient transfection, luciferase and β-galactosidase assays Tbx1 expression plasmid was previously described (65,67). miR-96 expression plasmid was constructed as previously described for other miRs (68). The pre-miR-96 was cloned into the expression vector. The Tbx1 3′UTR was cloned after the luciferase gene in pGL3 vector (Promega) (68). Truncations of the mouse Tbx1 gene were made using sequence-specific primers to Tbx1 by PCR with EcoRI and KpnI restriction enzyme sites. Tbx1 FL (full length), Tbx1 ΔC (deletion of C-Terminal tail) and Tbx1 ΔN (deletion of N-terminus) were cloned into pcDNA3.1. The Pitx2c luciferase promoter construct has been previously described (24), and the 5 kb miR-96 promoter was PCR amplified from mouse genomic DNA and cloned into the luciferase plasmid. All plasmid constructs were confirmed by DNA sequencing. LS-8 (oral epithelial) cells (69) were cultured in 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin and transfected via electroporation as previously described (70). Cells were fed fresh media 24 h before transfection, and electroporated with 2.5 μg of expression plasmid, 5 μg of reporter plasmid and 0.5 μg of SV-40 β-galactosidase plasmid. Transfected cells were incubated for 48 h in 60-mm culture dishes, lysed and assayed for β-galactosidase activity (Tropix Inc.), luciferase activity (Promega) and protein content (Bio-Rad). All luciferase activities were normalized to β-galactosidase activity and protein concentration. Experiments were repeated three to five times and the results are shown ±SEM. Transfection protocol for miR-96 to knock down endogenous Tbx1 used 1 μg DNA, 3 μl of X-tremeGene HP DNA transfection reagent, 200 μl of serum free media and LS-8 cells in a 6-well plate. Cells were plated at 10-20% confluence and harvested after 48 h and assayed for endogenous Tbx1 expression by western blot.

Expression and purification of GST-Tbx1 mutants and GST-PITX2A fusion proteins
Cloning of Tbx1 and PITX2A into pGEX6P-2 GST vector was previously described (24,71). Tbx1 deletion constructs were PCR amplified from a cDNA clone and ligated into pGEX6P-2 GST vector (Amersham Pharmacia Biotech, Piscataway, NJ) using EcoRI and XhoI restriction enzyme sites. The plasmids were confirmed by DNA sequencing and transformed in BL21 cells. Proteins were extracted as previously described (71,72). PITX2A was cleaved from GST moiety using 80 units of PreScission protease (Pharmacia Biotech) per milliliter of glutathione sepharose. Protein concentrations were quantified with Bradford Reagent (Bio-Rad Laboratories, Hercules, CA) and stored in 10% glycerol. Commassie Blue staining of denatured SDS-polyacrylamide gels was used to verify production of protein.

Western blotting
Expression of transiently expressed Tbx1 was demonstrated using a 1:500 dilution of anti-myc antibody (Cell Signaling).
Approximately 15-40 μg of cell lysates were used for sodium dodecyl sulfate gel electrophoresis. The protein was transferred to PVDF filters (Millipore), immunoblotted and detected using specific secondary antibodies and enhanced-chemiluminescene ECL reagents (GE HealthCare).
Quantitative real-time PCR of mRNA and microRNA Total RNA and miR from MEFs, LS-8, CHO and HEK 293 FT cells were prepared using the miRNeasy Mini Kit (Qiagen). The quantity and integrity of the RNA samples were assessed by measurements at 260 and 280 nm and verified using gel analysis. LIs and molars from E18.5 control and Tbx1 K14COET were also dissected and all of the RNA was prepared using the miRNeasy Mini Kit (Qiagen). MicroRNA were reversed transcribed using TaqMan microRNA assay probes (Applied Biosystems) and the TaqMan microRNA reverse transcription kit (Applied Biosystems) according to the manufacturer's instruction. Quantitative real-time PCR (qPCR) analysis of miRs was performed using TaqMan microRNA assay probes and normalized using the U6B probe (Taqman Universal PCR mastermix, Applied Biosystems).
Total RNA was reversed transcribed into cDNA using oligo (dT) primers according to the manufacturer's instructions (iScript Select cDNA Synthesis Kit, Bio-Rad). The MyiQ single colored Real-Time Detection System (Bio-Rad) was used for the reactions, and quantities were analyzed using the MyiQ Optical System Software 2.0 (Bio-Rad). cDNA levels were normalized to β-actin (F: 5′-GCCTTCCTTCTTGGGTATG-3′ and R: 5′-ACCACCAGACAGC ACTGTG-3′). Primers used for qPCR are as follows: Tbx1 (F: 5′-CGA CAAGCTGAAACTGACCA-3′ and R: 5′-GTGACTGCAGTGAAGC GTGT-3′) and Sox2 (F: 5′-ATGAGAGCAAGTACTGGCAAG-3′ and R: 5′-TCGGCAGCCTGATTCCAATAA-3′). The thermal cycling profile consisted of 95°C for 4 min, 40 cycles of denaturation at 95°C for 30 s, annealing at 60°C for 30 s and elongation at 72°C for 30 s. Samples were run in triplicate. Melting curves were generated to confirm the amplification specificity of the PCR products. In addition, all of the PCR products were sequenced to verify that the correct band was amplified.

Statistical analysis
For each condition, three experiments were performed and the results are presented as the mean ± SEM. The differences between two groups of conditions were analyzed using an independent, two-tailed t-test.

SEM imaging and microCT
Hemi-mandibles of littermate WT and Tbx1 K14cKO , COET or Tbx1 Pitx2cKO were dissected, fixed in 4% PFA overnight in 4°C and stored in 70% ethanol. EM images of uncoated specimens were taken with a Zeiss Evo LS 10 scanning electron microscope (Carl Zeiss, Peabody, USA) in secondary electron mode at 7 kV, 3 pA under high vacuum to assess gross morphology. Subsequently the incisors and molars were fractured, mounted on stubs using adhesive copper tape, and gold coated (Denton V Sputter coater, Denton, Moorestown, NJ). Images were then collected in secondary electron mode at 10 kV and 8 pA in high vacuum mode and at a working distance of 5-9 mm. MicroCT samples were analyzed in ethanol using a MicroCT 40 (Scanco Medical, Brüttisellen, Switzerland) at 70 kV, 114 µA, 8 W and 10 µm resolution, with an integration time of 300 ms. All samples to be compared were scanned in one batch to ensure identical conditions and allow for comparison of mineral densities. The images in DICOM format were processed using Fiji imaging software (http://fiji.sc/Fiji) to standardize the orientation of the samples. All samples were processed identically, setting an arbitrary, but identical threshold to remove any background and to allow for a comparison of mineral densities between samples in slices and maximum intensity projections.