DLX3-Dependent Regulation of Ion Transporters and Carbonic Anhydrases Is Crucial for Enamel Mineralization

Abstract

Patients with tricho-dento-osseous (TDO) syndrome, an ectodermal dysplasia caused by mutations in the homeodomain transcription factor DLX3, exhibit enamel hypoplasia and hypomineralization. Here we used a conditional knockout mouse model to investigate the developmental and molecular consequences of Dlx3 deletion in the dental epithelium in vivo. Dlx3 deletion in the dental epithelium resulted in the formation of chalky hypomineralized enamel in all teeth. Interestingly, transcriptomic analysis revealed that major enamel matrix proteins and proteases known to be involved in enamel secretion and maturation were not affected significantly by Dlx3 deletion in the enamel organ. In contrast, expression of several ion transporters and carbonic anhydrases known to play an important role in enamel pH regulation during maturation was significantly affected in enamel organs lacking DLX3. Most of these affected genes showed binding of DLX3 to their proximal promoter as evidenced by chromatin immunoprecipitation sequencing (ChIP-seq) analysis on rat enamel organ. These molecular findings were consistent with altered pH staining evidenced by disruption of characteristic pH oscillations in the enamel. Taken together, these results show that DLX3 is indispensable for the regulation of ion transporters and carbonic anhydrases during the maturation stage of amelogenesis, exerting a crucial regulatory function on pH oscillations during enamel mineralization.

Introduction

Enamel covers the surface of the tooth and is the hardest tissue in the body. It is produced by ameloblasts through a complex sequential mechanism.⁽¹⁾ In rodents, all stages of amelogenesis can be tracked along the proximal-distal axis of the continuously growing incisor. Dental epithelial stem cells are located in the cervical loop at the most proximal end of the incisor. When these cells leave their niche to differentiate into ameloblasts, they migrate towards the distal end of the tooth and go through different phases. The pre-secretion phase is followed by a secretion phase during which enamel matrix proteins are secreted in a highly structured fashion to form the scaffold of enamel where mineralization is simultaneously initiated. Secretory stage ameloblasts then enter a maturation phase characterized by the degradation of the vast majority of the enamel matrix proteins and their replacement by expanding hydroxyapatite crystals.

Defects in enamel formation, classified under amelogenesis imperfecta (AI) in humans, are common and can present with varying degrees of alterations in enamel thickness, enamel hardness, or both.^(2–4) Most of the nonsyndromic cases of AI have been associated with genes encoding enamel matrix proteins (AMELX, AMBN, ENAM, C4ORF26),^(5–8) proteases functioning in the degradation of matrix proteins (MMP20, KLK4),^(9,10) kinases with essential activity in the phosphorylation of enamel matrix proteins (FAM20A, FAM83H),^(11,12) molecules involved in ameloblast attachment to the basement membrane (ITGB6, LAMB3),^(13–15) membrane proteins involved in endocytic vesicle trafficking during enamel maturation (WDR72),⁽¹⁶⁾ and ion transporters (SLC24A4).⁽¹⁷⁾ Syndromic forms of AI are found in association with mutations leading to cases of epidermolysis bullosa (COL17A1, LAMA3, LAMB3, LAMC2, ITGB4),^(18–20) to immunodeficiencies due to calcium entry defects in T cells (ORAI1, STIM1),^(21–23) to renal tubular acidosis with ocular abnormalities (SLC4A4),⁽²⁴⁾ to renal hypomagnesemia (CLDN16),⁽²⁵⁾ and to cone rod dystrophy (CNNM4).^(26,27) The latest report of mutations in patients with AI patients identified a deletion in the gene encoding amelotin (AMTN),⁽²⁸⁾ a protein secreted during enamel maturation and known to promote hydroxyapatite mineralization.⁽²⁹⁾ The connection of most of these genes with amelogenesis has been further characterized in animal models.⁽³⁰⁾

So far, the only form of AI that has been associated with mutations in a transcriptional regulator is tricho-dento-osseous (TDO) syndrome. TDO is an autosomal dominant ectodermal dysplasia characterized by anomalies in hair, teeth, and bone development,^(31,32) and is caused by mutations in the gene encoding the homeodomain transcription factor DLX3.^(33–35) DLX3 is a member of the distal-less family that counts six members in mammals (DLX1–DLX6). The downstream pathways regulated by DLX3 in hair, teeth, and bone have been partially elucidated using conditional deletion in mice.^(36–40) Dental defects are by far the most debilitating trait of TDO syndrome, with hypoplastic and hypomineralized enamel, as well as dentin defects (taurodontism).^(31,32) Using a mouse model of neural crest conditional deletion of Dlx3, we showed previously that DLX3 is essential for normal dentin formation and mineralization through direct regulation of the major dentin matrix components encoded by the Dspp gene.⁽³⁷⁾ Thus far, the study of DLX3 function during amelogenesis has only been performed ex vivo.⁽⁴¹⁾

In the present study, we used a mouse model with conditional deletion of Dlx3 in the dental epithelium to determine the phenotypic consequences of Dlx3 ablation on enamel development in vivo. Furthermore, this animal model has allowed us to characterize the consequences of Dlx3 deletion on gene expression in the enamel organ, using RNA sequencing (RNA-seq) and microarray in two independent transcriptomic analyses. By performing chromatin immunoprecipitation sequencing (ChIP-seq) analysis on rat enamel organ in vivo, we identified genes that are regulated directly by DLX3 during amelogenesis. This in vivo approach has established the regulatory link between the DLX3 transcription factor and downstream targets involved in amelogenesis.

Materials and Methods

Animals

Dlx3^K14–cKO mice were generated using K14-cre mice and Dlx3^F/F mice and genotyped as described.⁽³⁶⁾ Cre recombinase activity in K14-cre mice was traced by mating with R26R^YFP (Jax006148) mice (The Jackson Laboratories, Bar Harbor, ME, USA). All animal work was approved by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) Animal Care and Use Committee.

Tissue

All experiments using mouse enamel organ extracts (RNA, protein, activity assays) were performed at postnatal day 10, an early time point in enamel formation when all stages of ameloblast differentiation are present in the continuously growing incisor. Mandibles were dissected and soft tissue was scraped off the mandibular bone using a round tip scalpel blade (#10). For RNA extractions, mandibles were immediately stored in 1 mL of RNAlater solution (Life Technologies, Grand Island, NY, USA) and stored at 4°C for up to 1 week. The basal bone was cut laterally and removed to expose the enamel organ of the mandibular incisor. Using the tip of a sharp edge scalpel blade (#11), the enamel organ was scraped off the surface of the continuously growing incisor and the tissue was placed into the appropriate buffer for further processing.

Immunohistochemical analysis

Samples were fixed overnight at 4°C in 4% paraformaldehyde in 1× PBS, dehydrated, and embedded in paraffin and 10-μm-thick sections were prepared. Yellow fluorescent protein (YFP)-expressing samples were prepared for frozen sections using standard procedure. Immunohistochemical analysis was performed using a blocking solution containing 5% goat serum and 7.5% BlockHen II (Aves Labs, Tigard, OR, USA) in 1× PBS. Primary antibodies used: rabbit anti-DLX3 (Abcam, Cambridge, MA, USA), rabbit anti-AMELX (Abcam), rabbit anti-CA2 (GeneTex Inc., Irvine, CA, USA), rabbit anti-CA6 (Abcam). Alexa-488 and Alexa-555 anti-rabbit (Thermo Fisher Scientific, Waltham MA, USA) were used as secondary antibodies for immunofluorescence. Biotinylated goat anti-rabbit IgG was used for colorimetric signal detection using VECTASTAIN ABC kit (Vector Laboratories, Burlingame, CA, USA) and liquid diaminobenzidine (DAB) substrate (BioGenex, Fremont, CA, USA).

Scanning electron microscopy

Samples were fixed overnight at 4°C in 2% glutaraldehyde, 2% paraformaldehyde (PFA) in 0.1 M phosphate buffer pH 7.4 and dehydrated through a series of 50%, 70%, 95%, and 100% ethanol solutions. They were incubated for 10 min in hexamethyldisilazane, air-dried for 30 min, mounted on aluminum specimen mount stubs covered with conductive carbon adhesive tabs (Electron Microscopy Sciences, Hatfield, PA, USA), sputter-coated with gold, and analyzed under a Field Emission Scanning Electron Microscope S4800 (Hitashi, Toronto, Canada) at 10 kV.

Micro–computed tomography analysis and 3D reconstructions

Mandibles were fixed in 4% PFA in 1 × PBS overnight at 4°C. Micro–computed tomography (μCT) analysis was performed as described⁽³⁷⁾ using the Skyscan 1172 desktop X-ray microfocus CT scanner. CT-an and CT-vol software (Skyscan b.v.b.a., Aartselaar, Belgium) were used for data analysis, measurements, and 3D reconstruction.

Statistical analysis

All quantitative experiments were performed on at least three control and three mutant tissues obtained from animals (mean ± SE). Statistical analyses were performed on Prism 5 statistical software (GraphPad Software Inc., San Diego, CA, USA), using t test with a significance level of 0.05 (*p < 0.05; **p < 0.01; ***p < 0.001).

Other methods

The methods for RNA-seq, microarray, ChIP-seq, qPCR, Western blot, carbonic anhydrase activity measurement, staining of enamel using methyl red and glyoxal bis(2-hydroxyanil) (GBHA), luciferase reporter assay, and systematic motif analysis are detailed in the Supplemental Materials and Methods.

Accession numbers

Complete microarray (Accession number: GSE87079), RNA-seq (Accession number: GSE57984), and ChIP-seq (Accession number: GSE85451) data have been deposited in the Gene Expression Omnibus site (https://www.ncbi.nlm.nih.gov/geo/).

Results

Early deletion of Dlx3 in the dental epithelium does not affect tooth morphogenesis

We performed a thorough analysis of DLX3 protein distribution in the developing tooth at the placode, bud, cap, and bell stages, on tooth sections from K14-cre:R26R^YFP mice, in which cre recombinase activity was tracked by analyzing YFP expression. DLX3 was detected in the dental epithelium at all stages of tooth morphogenesis, with a stronger expression on the labial side (Fig. 1A). The dental mesenchyme did not exhibit any DLX3 staining at the placode and bud stages, whereas strong expression was detected in the dental pulp at the cap and bell stages (Fig. 1A), which is consistent with our previous observations.⁽³⁷⁾ The dental epithelium was positive for YFP at all stages of tooth morphogenesis (Fig. 1A). Therefore, K14-cre mice can be used to delete Dlx3 in the dental epithelium at early stages of tooth development. Specific deletion of Dlx3 from the dental epithelium in Dlx3^K14–cKO mice was confirmed by immunohistochemical analysis using anti-DLX3 antibody on cap stage tooth sections (Fig. 1B). Tooth morphogenesis was not significantly affected in Dlx3^K14–cKO mice (Fig. 1B). These data show that the absence of DLX3 from the dental epithelium at early stages of tooth development does not affect tooth morphogenesis, suggesting a redundant function with other factors.

Fig. 1.

Expression pattern of DLX3 in the developing tooth relative to cre recombinase activity in K14-cre mice. (A) Immunohistochemical analysis of DLX3 expression at the placode, bud, cap, and bell stages of tooth morphogenesis using anti-DLX3 antibody (red staining) on coronal sections from K14-cre:ROSA^YFP mice at E12.5, E13.5, E14.5, and E16.5, respectively. Green staining (YFP) reflects cre recombinase activity in K14-cre mice. White arrows indicate stronger staining on the lingual side of the dental epithelium. Scale bars = 100 μm. (B) Immunohistochemical detection of DLX3 in cap stage teeth from Dlx3^WT and Dlx3^K14–cKO mice showing specific absence of DLX3 in the dental epithelium in Dlx3^K14–cKO mice. DAPI (blue) marks the nuclei. Scale bar = 100 μm.

Dlx3 expression during the secretion and maturation stages of amelogenesis

Immunohistochemical analysis of DLX3 expression during amelogenesis was performed on longitudinal sections of mandibular incisors (Fig. 2A) from K14-cre:RosaYFP mice at postnatal day 5. Absent from the cervical loop, DLX3 expression became evident during the pre-secretion stage and its expression persisted during differentiation (Fig. 2A). DLX3 expression was detected throughout the dental mesenchyme (Fig. 2A), and the whole dental epithelium was YFP positive (Fig. 2A), which confirms that Dlx3 can be deleted from all stages of amelogenesis using K14-cre mice. Conditional deletion was validated by qPCR analysis of Dlx3 mRNA expression in enamel organs dissected from Dlx3^WT and Dlx3^K14–cKO mice at postnatal day 10, showing a 50-fold decrease in Dlx3 expression (Fig. 2B). Western blot and immunohistochemical analyses showed the absence of DLX3 protein in enamel organ from Dlx3^K14–cKO mice (Fig. 2C, D). DLX3 was deleted from both secretion stage and maturation stage ameloblasts, whereas its expression remained unaffected in the dental mesenchyme (Fig. 2D, E).

Fig. 2.

DLX3 expression at the secretion and maturation stages of amelogenesis and deletion in Dlx3^K14–cKO mice. (A) Immunohistochemical detection of DLX3 (red) at different positions of the developing mandibular incisor at P5: cervical loop (1, CL), pre-secretory (2), and secretory (3) stages, as indicated on H&E section above. Immunostaining was performed on longitudinal sections of mandibles from K14-cre:R26R^YFP mice. YFP signal in green marks the lineage of cre-expressing cells in K14-cre mice. The white arrowhead indicates the increase in DLX3 expression during the pre-secretion stage. The white arrow indicates the start of the secretion stage. Scale bar = 100 μm. (B) qPCR analysis of Dlx3 mRNA levels in the enamel organ of Dlx3^WT and Dlx3^K14–cKO mice at P10. (C) Western blot analysis of DLX3 and actin expression in the enamel organ of Dlx3^WT and Dlx3^K14–cKO mice at P10. (D) Immunohistochemical detection of DLX3 (red) in the enamel organ from Dlx3^WT and Dlx3^K14–cKO mice at P5 showing specific deletion of DLX3 in the dental epithelium in Dlx3^K14–cKO mice at the secretory stage. DAPI (blue) was used to stain nuclei. Scale bar = 100 μm. (E) Immunohistochemical detection of DLX3 (brown staining) in the enamel organ from Dlx3^WT and Dlx3^K14–cKO mice at P15 showing that DLX3 is present in both secretory and maturation stage ameloblasts and absent from both stages in Dlx3^K14–cKO mice. Scale bars = 50 μm. DE = dental epithelium; DM = dental mesenchyme; Od = odontoblasts; De = dentin.

Dlx3 deletion in the enamel organ results in hypomineralized enamel

The enamel surface of incisors from Dlx3^K14–cKO mice appeared white and chalky (Supplemental Fig. 1), which is a characteristic sign of severe AI.⁽²⁾ To further analyze enamel defects in Dlx3^K14–cKO mice, we performed scanning electron microscopy on adult teeth at 8 weeks. As opposed to teeth from Dlx3^WT mice that exhibited a smooth and homogeneous enamel surface, all teeth from Dlx3^K14–cKO mice had a rough and irregular surface (Fig. 3A). Analysis of broken mandibular incisors from Dlx3^K14–cKO mice revealed that the enamel of erupted teeth appeared thin and did not exhibit the characteristic lattice pattern seen in control enamel (Fig. 3B).

Fig. 3.

Defective enamel mineralization in DLX3^K14–cKO mice. (A) Scanning electron microscopy of first mandibular molars and maxillary incisors from Dlx3^WT and Dlx3^K14–cKO mice at 8 weeks. Insets show a closer view of the enamel surface on maxillary incisor. (B) Scanning electron microscopy of broken mandibular incisors at 8 weeks. Inset shows magnified view of enamel in Dlx3^K14–cKO mice. (C) μCT 3D reconstruction of mandibles from Dlx3^WT and Dlx3^K14–cKO mice at 8 weeks. Insets 1, 2, and 3 show cross-sections at different levels along the proximal-distal axis of the mandible: (1) protected early maturation enamel; (2) protected late maturation enamel; (3) exposed mature enamel. (D) Quantification of enamel and dentin mineral density in the areas corresponding to early (position 1 in C) and late (position 2 in C) stages of enamel maturation.

To investigate the structure and mineral density of enamel in Dlx3^K14–cKO mice in conditions in which mechanical alteration is not involved, μCT analysis was performed on whole mandibles at 8 weeks. μCT sections at different levels of the mandible showed that the enamel of both molars and incisors was hypomineralized (Fig. 3C). At the level of the most distal root of the first molar, mandibular incisors are at an early stage of maturation. In that region, enamel density in Dlx3^WT mice was significantly higher than dentin density, whereas enamel density in Dlx3^K14–cKO mice was significantly lower than dentin density (Fig. 3C, position 1; Fig. 3D). At a late stage of enamel maturation, while the mineral density of enamel in Dlx3^WT mice had continued to increase, there was no significant difference between the mineral density of enamel and dentin in Dlx3^K14–cKO mice (Fig. 3C, position 2; Fig. 3D). These data demonstrate that deletion of Dlx3 in the enamel organ results in hypomineralized enamel.

Major enamel matrix components and proteases involved in amelogenesis are not affected by Dlx3 deletion

To identify molecular markers affected by the absence of DLX3 in the enamel organ in vivo, RNA-seq analysis was performed on enamel organs dissected from mandibular incisors of Dlx3^WT and Dlx3^K14–cKO mice (Fig. 4A). Transcriptomic analysis was also performed on an independent set of tissues using microarray analysis. In order to determine direct targets of DLX3 in vivo, ChIP-seq analysis was performed on dissected rat enamel organs.

Fig. 4.

Unaltered expression of enamel matrix proteins and proteases in Dlx3^K14–cKO mice. (A) Schematics illustrating the dissection and isolation of RNA from the enamel organs of continuously growing incisors from mandibles of Dlx3^WT and Dlx3^K14–cKO mice at P10. Arrowheads at the tip of the mandibles point at the just erupted mandibular incisor and mark a stage where all phases of ameloblast differentiation are represented. Lower panel shows visualization of RNA-seq coverage data confirming complete deletion of Dlx3 exons 2 and 3 in the enamel organ in Dlx3^K14–cKO mice. (B) Heat map showing log(RPKM) levels of Dlx3 and major enamel differentiation markers that have been associated with amelogenesis imperfecta, in enamel organs from Dlx3^WT and Dlx3^K14–cKO mice. Except for Dlx3, none of the fold-changes indicated are significant. (C) Western blot analysis of AMELX and MMP20 protein levels in the enamel organ of Dlx3^WT and Dlx3^K14–cKO mice at P10. (D) Immunohistochemical detection of AMELX (green) in mandibular incisor sections from Dlx3^WT and Dlx3^K14–cKO mice. White arrowheads point at ameloblasts, asterisks mark the enamel space. Scale bar: 100 μm. (E) ChIP-seq analysis of DLX3 and H3K4me3 DNA binding domains in enamel organ dissected from rat mandibular incisors (continuously growing), showing DLX3 binding near the proximal promoter (arrowhead marks transcription start site) of Ambn and Mmp20. DM = dental mesenchyme.

Visualization of RNA-seq coverage data shows the absence of reads aligned to the second and third exons of Dlx3 in the enamel organ from Dlx3^K14–cKO mice, confirming complete inactivation of the gene in the tissue (Fig. 4A). Interestingly, none of the major enamel matrix components (Amelx, Ambn, Enam) or proteases (Mmp20, Klk4) involved in enamel formation and mutated in cases of AI were affected significantly in the absence of DLX3 (Fig. 4B). The same observation was made for most of the genes in which mutations have been associated with AI (Fig. 4B; Supplemental Fig. 2). These results were consistent with microarray data acquired from independent samples (Supplemental Fig. 3) and confirmed at the protein level for AMELX and MMP20 (Fig. 4C, D). Interestingly, ChIP-seq analysis revealed DLX3 binding at the proximal promoter of several of these ameloblast markers, including Ambn and Mmp20 (Fig. 4E). These results demonstrate that, even though DLX3 potentially regulates the expression of enamel matrix proteins and proteases during amelogenesis, these markers are not affected significantly by in vivo deletion of Dlx3.

Ion transporters involved in pH regulation during enamel maturation are affected by Dlx3 deletion

During enamel maturation, the nucleation of hydroxyapatite minerals generates high levels of protons that make the pH of enamel very acidic. The buffering of these protons is essential for proper enamel mineralization. The importance of ion transport in enamel pH regulation during maturation has been supported by both clinical reports and animal studies.^(1,42) The expression of cystic fibrosis transmembrane conductance regulator (Cftr), encoding the chloride channel mutated in cystic fibrosis, was significantly downregulated in the enamel organ from Dlx3^K14–cKO mice (Fig. 5A; Supplemental Fig. 4A). Cystic fibrosis is a chronic disorder of exocrine function that affects primarily the lung, pancreas, liver, kidney, and intestine, but has also been associated with enamel defects.⁽⁴³⁾ Moreover, Cftr^−/− mice exhibit soft and chalky enamel⁽⁴⁴⁾ due to altered pH regulation during enamel maturation.⁽⁴⁵⁾ Two bicarbonate transporters, SLC4A2 (bicarbonate/chloride transporter) and SLC4A4 (sodium bicarbonate transporter), are known to play an essential role in enamel pH regulation. Mutations in SLC4A4 in humans lead to renal tubular acidosis, with patients featuring defects in the enamel of the permanent teeth (OMIM# 604278).⁽²⁴⁾ Inactivation of Slc4a2 or Slc4a4 in mice result in hypomineralized enamel.^(46,47) The mRNA levels of Slc4a2 and Slc4a4 were not affected significantly by the absence of DLX3 (Fig. 5A; Supplemental Fig. 4A). However, Slc24a4 (potassium-dependent sodium/calcium exchanger) was downregulated significantly in the enamel organ from Dlx3^K14–cKO mice (Fig. 5A; Supplemental Fig. 4A). Mutations in SLC24A4 lead to hypomaturation-type AI both in humans and in mice.⁽¹⁷⁾

Fig. 5.

Altered expression of ion transporters and defects in enamel pH regulation in Dlx3^K14–cKO mice. (A) Heat map showing log(RPKM) levels of ion transporters in enamel organs from Dlx3^WT and Dlx3^K14–cKO mice. Fold-change and significance are indicated. (B) qPCR analysis showing increased expression of Slc26a1 and Slc26a7 during the transition between the secretion (first mandibular molar at P4) and the maturation (first mandibular molar at P11) stages. Amelx and Klk4 were used as specific secretion and maturation markers, respectively. (C) Staining of mandibular incisors from Dlx3^WT and Dlx3^K14–cKO animals using methyl red as a pH indicator and GBHA as a calcium chelator dye that results in calcium staining only in areas of alkaline pH. (D) ChIP-seq analysis of DLX3 and H3K4me3 DNA binding domains in enamel organ dissected from rat mandibular incisors (continuously growing), showing DLX3 binding near the proximal promoter (arrowhead indicate the transcription start site) of Cftr, Slc24a4, Slc26a7, Slc26a7, Slc34a2. Inserts showing magnification of the transcription start site show that, for several of these loci, the DLX3 peak aligns with a groove on the H3K4me3 track. (E) Same as D for Slc4a2 and Slc4a4.

Other ion transporters of the Slc superfamily were significantly affected by the absence of DLX3 in the enamel organ, including Slc26a1 (sulfate/anion transporter), Slc26a7 (sulfate/anion transporter), Slc34a2 (pH-sensitive sodium-dependent phosphate transporter), and Slc39a2 (zinc transporter) (Fig. 5A; Supplemental Fig. 4A). The expression of these Slcs was significantly upregulated during the transition between the secretion and the maturation stages of ameloblast differentiation (Fig. 5B), consistent with previous reports.^(48,49)

This gene signature suggests that enamel hypomineralization in Dlx3^K14–cKO mice may be due to defects in pH regulation during enamel maturation. This was confirmed by methyl red staining of continuously growing incisors, which showed disruption of the characteristic oscillations between acidic and physiological pH in the enamel matrix (Fig. 5C). These oscillations in pH can also be assessed by staining the enamel with GBHA, a calcium chelator dye that will result in calcium staining only in areas of alkaline pH.⁽⁵⁰⁾ The characteristic banding pattern observed with GBHA staining was lost in the enamel from Dlx3^K14–cKO mice (Fig. 5C), probably due to excess acidity. It may also be hypothesized that the loss of GBHA staining reflects defects in calcium transport. However, because the only calcium transporter we found affected in this model is SLC24A4 with less than a twofold decrease, the loss of GBHA staining is likely a consequence of defective pH.

ChIP-seq analysis confirmed the putative direct regulation of Cftr, Slc26a1, Slc26a7, Slc34a2, and Slc39a2 by DLX3 (Fig. 5D). Interestingly, binding of DLX3 was found at the proximal promoter of Slc4a2 and Slc4a4 (Fig. 5E), even though their expression was not affected by Dlx3 deletion.

Carbonic anhydrases involved in pH regulation are affected by Dlx3 deletion

Carbonic anhydrases catalyze the interconversion of carbon dioxide and water to bicarbonate and protons (CO₂ + H₂O ↔ HCO₃⁻ + H⁺) and are therefore involved in pH regulation. Expression of several carbonic anhydrases by ameloblasts have been reported⁽⁵¹⁾, and their implication in enamel pH regulation has been proposed.^(1,42) We found that the expression of Car6 and Car12 was downregulated significantly in the enamel organ from Dlx3^K14–cKO mice, whereas Car2, Car3, and Car13 were not significantly affected (Fig. 6A; Supplemental Fig. 4B). Consistent with mRNA levels, CA6 protein levels in maturation stage ameloblasts were significantly reduced in Dlx3^K14–cKO mice, while unaltered in the salivary gland where DLX3 was not detected (Fig. 6B, C). CA2 protein levels were unaffected in both the enamel organ and the salivary gland from Dlx3^K14–cKO mice (Fig. 6B, C). Overall, total carbonic anhydrase activity was reduced in the enamel organ from Dlx3^K14–cKO mice while unaffected in the salivary gland (Fig. 6D). ChIP-seq analysis performed on rat enamel organ indicates that DLX3 potentially regulates most of the carbonic anhydrases expressed in the enamel organ, even those unaffected by Dlx3 deletion (Fig. 6E).

Fig. 6.

Altered expression of carbonic anhydrases in the enamel organ from Dlx3^K14–cKO mice. (A) Heat map showing log(RPKM) levels of carbonic anhydrases in enamel organs from Dlx3^WT and Dlx3^K14–cKO mice. Fold-change and significance are indicated. (B) Western blot analysis of DLX3, CA2, CA6, and actin protein levels in the enamel organ (E.O.) and salivary gland (S.G.) of Dlx3^WT and Dlx3^K14–cKO mice at P10. (C) Immunohistochemical detection of CA2 and CA6 (brown precipitate) in mandibular incisor longitudinal sections from Dlx3^WT and Dlx3^K14–cKO mice at P15. Secretory (s), transition (t), and maturation (m) stages are indicated. Scale bar = 100 μm. (D) Measurement of carbonic anhydrase activity in extracts from E.O. and S.G. from Dlx3^WT and Dlx3^K14–cKO mice at P10. (E) ChIP-seq analysis of DLX3 and H3K4me3 DNA binding domains in enamel organ dissected from rat mandibular incisors (continuously growing), showing DLX3 binding near the proximal promoter (arrowhead) of Car2, Car3, Car6, Car12, and Car13, but not Car1. E.O. = enamel organ; S.G. = salivary gland.

Mutant DLX3^TDO affects the transcriptional response of human CA6 and SLC26A1 proximal promoters to DLX3

In order to assess the relevance of our findings to human enamel development and the pathogenicity of TDO syndrome, we compared the DNA sequences of rat Car6 and Sc26a1 proximal promoter regions where DLX3 binds in rat enamel organ to the homologous human regions. The alignment of the rat and human sequences for these regions shows high conservation and reveals conserved putative binding sites for factors that may interact with DLX3 on these promoters (Fig. 7A, B). The SLC26A1 proximal promoter contains a conserved homeodomain binding site (TAATT), which suggests that DLX3 can potentially bind directly to this DNA region (Fig. 7). To further test the response of these human proximal promoter regions to DLX3 and the potential effect of the DLX3^TDO mutant isoform on this response, we performed a luciferase reporter assay using Saos2-TetOff cells (tetracycline-inducible human osteosarcoma cell line) (Supplemental Fig. 5A). As previously reported,⁽⁵²⁾ when using a luciferase reporter construct that contains three copies of the consensus DLX3 homeodomain binding site (TAATT), the overexpression of DLX3 in these cells resulted in a significant increase in luciferase activity that is attenuated when the mutant DLX3^TDO isoform is present (Supplemental Fig. 5B). DLX3 overexpression also resulted in a significant increase in luciferase activity when using a reporter construct containing the CA6 proximal promoter, and the presence of the DLX3^TDO isoform in this context resulted in enhanced increase in activity (Supplemental Fig. 5C). When using a reporter construct containing the SLC26A1 proximal promoter, DLX3 overexpression resulted in a significant decrease in luciferase activity that was more substantial in the presence of the DLX3^TDO mutant isoform (Supplemental Fig. 5D). These results demonstrate that the human proximal promoter regions of both CA6 and SLC26A1 are responsive to DLX3 expression and that the presence of the mutant DLX3^TDO isoform alters the extent of this response. Therefore, the expression of CA6 and SLC26A1 is likely to be altered during enamel development in TDO patients. The diverse effects observed using this ex vivo reporter approach highlight the fact that DLX3 function in vivo involves complex transcriptional regulatory machineries which include tissue-specific interacting partners and remote regulatory regions (enhancers) that remain to be characterized.

Fig. 7.

Alignment of rat and human CA6 and SLC26A1 proximal promoter regions showing DLX3 binding. The DNA sequences corresponding to the DLX3 binding peak in the proximal promoter region of Car6 (A) and Slc26a1 (B) in rat enamel organ (rCar6 and rSlc26a1, respectively) were aligned to their homologous regions in human (proximal promoter of hCA6 and hSLC26A1, respectively). Performing systematic motif analysis (see Supplemental Materials and Methods) on the promoters of Car6 and Slc26a1, we identified a number of conserved regulatory sequences shared between human and rat. These regions harbor potential binding sites for a diverse range of transcription factors including factors from the STAT, SOX, PAX, HNF, FOX, and AP1 families, as well as a complex between EWSR1-FLP1, and sites known to bind leucine-zipper (LZ) and zinc-finger (ZF) domains. In the promoters of rat and human SLC26A1, the conserved region “TAAATAATTT/GAAC” at −25 bp/−30 bp to the transcription start site (TSS) showed a strong affinity for several homeobox binding models (eg, PAX and HNF) based on our computational motif analysis. We hypothesize that DLX3 may bind directly this region. CDS = coding sequence; SC = start codon; LZ = leucine-zipper; ZF = zinc-finger; TSS = transcription start site.

Discussion

In the present study, we show that the deletion of Dlx3 in the enamel organ in vivo leads to the formation of hypomineralized enamel. At the molecular level, the absence of DLX3 does not affect significantly the expression of major enamel matrix proteins and proteases involved in amelogenesis, despite evidence of DLX3 potential role in the regulation of these genes (Fig. 8, left and central panels). In contrast, the absence of DLX3 in the enamel organ results in a significant decrease in the expression of several ion transporters and carbonic anhydrases (Fig. 8, right panel). These findings demonstrate that DLX3 is a master regulator of genes implicated in enamel pH regulation, a mechanism which is essential for proper maturation of enamel hydroxyapatite crystals.^(1,42) These findings have tremendous implication in the understanding of the pathogenicity of TDO, most importantly for the hypomineralization of enamel in this disease.^(32,33)

Fig. 8.

DLX3 involvement in secretory and maturation stage ameloblasts and effects of in vivo deletion of Dlx3 on gene expression in the enamel organ. DLX3 is expressed in both secretion and maturation stage ameloblasts. ChIP-seq analysis revealed that DLX3 has the potential to regulate the expression of (1) enamel matrix proteins during the secretion phase, (2) proteases expressed at the secretory and maturation stages, and (3) ion transporters and carbonic anhydrases expressed in maturation stage ameloblasts. In the enamel organ of Dlx3^K14–cKO mice, the expression of enamel matrix proteins and proteases was not significantly affected, suggesting that other factors (TF-X) are compensating for the absence of DLX3 in regulating these genes. However, the expression of ion transporters and carbonic anhydrases was significantly affected by Dlx3 deletion, which shows that DLX3 is a master regulator of enamel mineralization though control of enamel pH.

During tooth morphogenesis, we detected DLX3 protein in the labial dental epithelium as early as the placode and bud stages, which is consistent with previous reports on Dlx3 mRNA expression pattern.⁽⁵³⁾ Expression in the dental epithelium persisted at the cap and bell stages, as we and others reported previously.^(37,54) Despite its early expression in the tissue, Dlx3 deletion in the dental epithelium does not affect tooth morphogenesis. Because Dlx2 and Dlx3 exhibit overlapping expression patterns in the dental epithelium at the placode and bud stages of tooth morphogenesis,⁽⁵³⁾ they may have partially redundant functions during early tooth development.

During amelogenesis, we show that DLX3 protein levels increase during the pre-secretion stage and remain high during both the secretion and maturation stages, which is consistent with previous reports of Dlx3 mRNA expression in this tissue.^(54,55) Expression of DLX3 in ameloblasts during the pre-secretion and secretion stages, together with the results from our ChIP-seq analysis of DLX3 binding sites in the enamel organ, support the involvement of DLX3 in the regulation of enamel matrix proteins. This is also supported by a recent study carried out in an ameloblast cell line.⁽⁴¹⁾ However, the expression of major enamel matrix proteins was not affected significantly by the absence of DLX3. This suggests that the absence of DLX3 is compensated for by other DLXs or other homeodomain transcription factors. Similar observations were made for the involvement of DLX3 in the regulation of major enamel matrix proteases known to be essential for enamel maturation. However, the fact that the absence of DLX3 in the enamel organ in vivo does not significantly impact the expression of enamel matrix protein and proteases, does not exclude that these same markers could be affected in TDO syndrome. Indeed, mutant DLX3 is likely to disrupt the regulation of these genes.^(41,52,56)

The enamel hypomineralization phenotype observed in TDO patients,^(31,32) the persistent expression of DLX3 in maturation stage ameloblasts, and the severe enamel hypomineralization phenotype observed in Dlx3^K14–cKO mice, strongly support a crucial involvement of DLX3 in enamel maturation as well. Moreover, the gene signature observed in the enamel organ from Dlx3^K14–cKO mice, reveals the indispensable function of DLX3 in the regulation of major pH regulators, namely ion transporters and carbonic anhydrases.

Based on both clinical data and animal studies, ion transporters are known to play a crucial role in the regulation of enamel pH during enamel maturation.^(1,42) This has been established for CFTR,^(43–45) SLC4A2,⁽⁴⁶⁾ and SLC4A4.^(24,47) These ion transporters have been proposed to contribute directly or indirectly to the buffering of the protons generated by the formation of hydroxyapatite crystals, mostly through the secretion of bicarbonate in the enamel matrix.^(1,42) The import of chloride through CFTR at the apical membrane of the ameloblasts is thought to be coupled with the export of bicarbonate through SLC4A2 (in exchange for chloride), whereas SLC4A4 would be involved in the import of bicarbonate through the basolateral membrane.⁽⁴²⁾ The involvement of SLC24A4 in enamel mineralization has been clearly established, with a major function in calcium transport.⁽¹⁷⁾ However, this potassium-dependent sodium/calcium exchanger may also have an indirect effect on enamel pH. We found other genes encoding ion transporters from the Slc superfamily to be regulated by DLX3, namely Slc26a1, Slc26a7, Slc34a2, and Slc39a2. Several of these are potentially involved in enamel pH regulation and mineralization, even though no AI-associated mutations have been identified in these genes so far. Mutations of the gene encoding SLC26A1, which is implicated in sulfate-oxalate, sulfate-bicarbonate, and oxalate-bicarbonate exchange, lead to the formation of calcium oxalate kidney stones, both in mouse⁽⁵⁷⁾ and in humans suffering from nephrolithiasis.⁽⁵⁸⁾ Slc26a7^−/− mice exhibit distal renal tubular acidosis and gastric acid secretion defects.⁽⁵⁹⁾ Deletion of these genes in mice does not lead to enamel defects but rather in the upregulation of other ion transporters that are crucial for pH homeostasis, which suggests compensatory mechanisms used by ameloblasts in their absence.⁽⁴⁹⁾ Although the expression of Slc34a2 and Slc39a2 is increased in maturation stage ameloblasts,⁽⁴⁸⁾ their function during amelogenesis has not been investigated. In humans, mutations in SLC34A2, which encodes a sodium-phosphate co-transporter, lead to pulmonary alveolar microlithiasis (OMIM# 265100), a rare disease characterized by the deposition of calcium phosphate microliths in the lungs.⁽⁶⁰⁾ The Slc39a2 gene encodes a zinc transporter. The importance of zinc transport in mineralization processes has been shown in association with bone formation⁽⁶¹⁾ and in the context of kidney stone disease.⁽⁶²⁾ Their function is most likely through the regulation of metalloenzymes that require zinc for their activity, such as carbonic anhydrases.⁽⁶³⁾ It is therefore highly probable that Slc34a2 and Slc39a2 contribute to enamel mineralization.

In addition to ion transporters, carbonic anhydrases have been proposed to be major players in enamel pH regulation and mineralization, with CA2 and CA6 being the most abundant carbonic anhydrases (cytosolic and extracellular, respectively) in the enamel organ.⁽⁵¹⁾ Carbonic anhydrases are ancient enzymes that are essential for the formation of calcium carbonate exoskeleton in corals, which demonstrates their ancestral function in biomineralization processes.⁽⁶⁴⁾ In vertebrates, the importance of pH regulation by carbonic anhydrases in the process of matrix vesicle mineralization is well established.⁽⁶⁵⁾ CA2 is highly expressed in osteoclasts and known to play an essential role in bone resorption through modulation of intracellular pH and Ca²⁺,⁽⁶⁶⁾ and mutations in CA2 in humans lead to osteopetrosis with renal tubular acidosis (OMIM# 259730).⁽⁶⁷⁾ Even though dental defects are not fully penetrant in patients with CA2 deficiency, some patients exhibit enamel hypoplasia and severe caries.⁽⁶⁸⁾ However, enamel defects have not been reported in Car2^−/− mice.⁽⁶⁹⁾ An ex vivo study of CA2 function in ameloblast differentiation, using inhibitor and antisense approaches, suggests that the reduction of intracellular pH caused by CA2 results in downregulation of enamel matrix protein expression and upregulation of proteases involved in enamel maturation through a pH-dependent activation of JNK signaling pathway.⁽⁷⁰⁾ The role of CA6 in enamel maturation is more perplexing, partly because it is also abundant in the saliva and the enamel pellicle (layer between the enamel and bacterial plaque). A higher prevalence of caries has been associated with lower concentrations of CA6 in the saliva of human subjects, which suggests that CA6 serves to protect the enamel surface from caries by neutralizing excess acidity.⁽⁷¹⁾ More recently, a genetic association study identified several polymorphisms in CA6 in humans that were correlated with tooth decay.⁽⁷²⁾ However, Car6^−/− mice do not exhibit a major dental phenotype and further are resistant to oral colonization by Streptococcus mutans and caries development.⁽⁷³⁾ These observations were made with a focus on CA6 function in the saliva without addressing CA6 function in enamel formation per se. Even though the expression of Car3 and Car12 increases during enamel maturation,⁽⁵¹⁾ their function in this process has not been determined. CA12 mutations in humans have been associated with hyperchlorhidrosis (OMIM# 143860), a disorder characterized by excess chloride secretion in sweat.⁽⁷⁴⁾ This membrane-bound carbonic anhydrase may work in concert with CFTR and bicarbonate transporters in regulating enamel pH. Overall, even though the direct involvement in enamel mineralization remains to be established for most of the carbonic anhydrases expressed in the enamel organ, their role in pH regulation and their influence on biomineralization is undeniable. Based on the function of carbonic anhydrases in corals, an alternative or additional mechanism of action for these enzymes in enamel could be the production of amorphous calcium carbonate that would serve as a precursor to the formation of hydroxyapatite minerals. Indeed, this sequential mineralization process was recently described for the formation of hydroxyapatite crystals in osteoblast cultures.⁽⁷⁵⁾ However, the presence of amorphous calcium carbonate precursors in enamel remains to be tested.

Taken together, our findings establish that DLX3 is involved in both the secretion and maturation stages of amelogenesis, with a particularly indispensable role in regulating pH during the maturation stage. Although ion transporters and carbonic anhydrases had been previously identified in enamel and linked to pH regulation, this is the first study that characterizes a common upstream transcriptional mechanism that acts as a master regulator of these families of genes.