A previous study using TOPGAL reporter mice provided evidence for canonical Wnt signaling (Lohi et al., 2010; Suomalainen and Thesleff, 2010). We first evaluated this activity more extensively using WntVis reporter mice at P7.0 (Takemoto et al., 2016). Signals were intense in the odontoblast layer (asterisks in Figure 1A). Weak Signals were detected at the alveolar bone surface and the periodontal ligament (Figure 1A).

10E4 recognizes the sulfated HSPG, and we have previously shown that HSPG is desulfated in mature odontoblasts (Hayano et al., 2012). In developing molars at P0, 10E4 immunoreactivity was absent from the odontoblast layers (Figure 1B), suggesting that cell surface HSPG was desulfated with odontoblast differentiation. Tmem2-CKO.

The cell viability of primary mouse dental papilla mesenchymal cells (mDP cells) after exposure to 0.1–10 ppm MA-T for 48 h was measured using the MTT (3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyltetrazolium bromide) assay. The viability of mDP cells was significantly reduced by MA-T treatment at 10 ppm. MA-T exposure at less than 0.2 ppm did not affect cell viability compared to the control (Figure 1C).

Sodium chlorate is a potent oxidant and a reversible inhibitor of glycosaminoglycan sulfation. Treatment with chlorate significantly reduced the binding of the 10E4 antibody to odontoblasts (Hayano et al., 2012). The main active ingredient of MA-T is chlorine dioxide which has oxidizing properties. However, neither MA-T nor chlorine dioxide has been tested for effectiveness in the desulfating of HSPG in odontoblasts.

10E4 immunoreactivity was evident in odontoblasts, while 72 h MA-T treatment resulted in a marked reduction in the cell-surface 10E4 immunoreactivity in odontoblast mDP cells (Figure 1D), indicating that MA-T desulfated HSPG on odontoblasts, as well as sodium chlorate.

MA-T treatment also upregulated the mRNA expression of Axin2, a canonical Wnt signaling marker, 3 days after MA-T application (Figure 1E).

We then cultured odontoblast cell lines until they formed a mineralized matrix. MA-T treatment increased mineralized matrix deposition as assessed by alizarin red S incorporation 14 days after the start of culture (Figure 1F). MA-T also upregulated the expression of Dspp and Dmp1 (Figure 1G).

To further analyze the influence of MA-T on dentin matrix formation, the growing incisors at E16.0 were cultured with treatment of 1.0 ppm of MA-T for 10 days. Dentin matrix formation was evaluated by micro-CT and histologic evaluation. The enamel was highly calcified in the CT images (Figure 2A), whereas the dentin appeared comparatively less calcified. To determine the thickness of the dentin matrices, we first measured the thickness of calcified tissue on the labial aspect of the incisors. The thickness of the enamel was then measured at the designated measurement site using signal threshold adjustments. Finally, the thickness of dentin was calculated as the difference between the two. The results showed that MA-T increased the thickness of dentin (Figure 2B). Histologic images showed the formation of dentin and an increase in the size of dentin matrices due to MA-T (Figure 2C).

To comprehensively understand the Wnt signaling in odontoblast differentiation, we reanalyzed the public GSE146855 scRNA-seq dataset on the previously analyzed isolated mouse incisors (Chiba et al., 2020).

We obtained 6,260 single-cell transcriptome data, and uniform manifold approximation and projection (UMAP) clustering identified ten cell population clusters of odontoblast, sub-odontoblast and dental pulp, dental follicle 1 and 2, ameloblast, pre-ameloblast, quiescent-IEE, non-ameloblast, leukocyte, erythrocyte, endothelial cell (Figure 3A). The annotation of these cell types for these clusters was based on the expression of the specific markers in different clusters on the UMAP plot (Supplementary Figures S1A, B). In addition, the expression of Dspp and Dmp1 was projected onto the UMAP plot, which showed that the distribution of the Dmp1-expressing cell population was similar to that of Dspp in the odontoblast clusters (Figure 3B).

Next, the odontoblast cluster was characterized. The FindAllMarkers function in Seurat identified the 524 differentially expressed genes in the odontoblast cluster (0.05 > p-value adjusted). KEGG pathway analysis of the top 300 up- and downregulated genes in the odontoblast cluster revealed that the “Wnt signaling pathway” was highly enriched (-log10 p values = 3.33, Figure 3C). In this pathway, enriched DEGs were Bambi, Dkk1, Lgr6, Nkd1, Notum, Serpinf1, Tcf7, Wnt10a, Wnt5a, Wnt6. Among these genes, feature analysis revealed that the distribution of Dspp and Dmp1 was uniquely overlapped with that of Wnt6, Wnt10, Dkk1, and Notum (Figure 3D). The distribution of Tcf7 expression also overlapped with Dspp and Dmp. Tcf7 was more specifically distributed in the odontoblast cluster than Lef1, Tcf3, or Tcf4 (Figure 3D and Supplementary Figure S2). Bambi and Serpinf1 were highly expressed in odontoblasts, but their expression was rather non-specific. Wnt5a expression was also non-specific among odontoblasts, pre-odontoblasts, and dental mesenchymal cells (Supplementary Figure S2).

DotPlot showed that the Dmp1 expression is more specific than Dspp. Dotblot also confirmed that Wnt6, Wnt10a, and Wnt5a were more expressed in the odontoblast cluster. Tcf7 is more specific in the odontoblast cluster than Tcf3 and Tcf4 (Figure 3E).

Previous studies have shown that both Wnt10a and Wnt6 have anabolic effects on odontoblast differentiation. The scRNA-seq reanalysis showed the specific overlap of Wnt6 and Wnt10a expressing cells among odontoblast clusters. Therefore, we hypothesized that they have some functional redundancy.

The combination of Wnt10a siRNA with Wnt6 siRNA showed a more remarkable reduction in the expression of Dspp and Dmp1, respectively, compared to the single treatment (Figure 4A), suggesting that there might be functional redundancy between Wnt10a and Wnt6 in the induction of odontoblast differentiation.

Wnt signaling is activated in odontoblasts as shown by WntBis reporter mice (Figure 1A). In our scRNA-seq, both Wnt ligands and Wnt inhibitory genes are specifically activated in odontoblasts. Thus, we speculate that Wnt activity is maintained in the odontoblast by both direct activation and inhibition of Wnt signaling in an autocrine manner.

Our in vitro studies showed that MA-T treatment at 0.2 ppm and 1.0 ppm and LiCl treatment upregulated the mRNA expression of Wnt10a and Wnt6 but not Wnt5a in odontoblasts (Figure 4B). On the other hand, Wnt inhibitors of both Dkk1 and Notum were also activated by MA-T and LiCl treatment (Figure 4B). Thus, these results reveal the existence of a positive and negative feedback loop in which Wnt signaling in odontoblasts is maintained between Wnt ligands and Wnt inhibitors.

To evaluate the possible direct control of Wnt6, Wnt10a, Notum, and Dkk1 by Wnt signaling, we investigated whether TCF/LEF binding sites exist around the promoter region of Wnt10a, Wnt6, Dkk1, and Notum by analyzing the public ChIP-seq data (SRX9641211, SRX1411163, SRX360319, SRX5825990) using Lef1, Tcf7, and RNA polymerase II. In silico ChIP analysis was performed using the ChIP-Atlas 23.

ChIP-Atlas analysis showed that Tcf7 occupies the proximal promoter region of Wnt6, Wnt10a, Notum, and Dkk1 with RNA pol II (Figure 4C). Lef1 also bind to binds to the promoters of Wnt6, Notum, and Dkk1 (Figure 4C).

MA-T contains sodium chlorite, but cationic detergents acting as Lewis acids allow sodium chlorite to be converted to chlorine dioxide only in the presence of reactive targets (Shibata and Konishi, 2020).

We evaluated the cell viability of the mDP cells in response to sodium chlorite treatment by MTT assay and the influence on odontoblast differentiation. The viability of mDP cells was significantly reduced by sodium chlorite treatment at 10 ppm. However, sodium chlorite exposure at less than 1.0 ppm did not affect the cell viability in contrast to the control (Supplementary Figure S3A). On the other hand, sodium chlorite upregulated the expression of both Dmp1 and Dspp expression at the dose of 100 ppm. Such upregulation was not evident at the dose of less than 2.0 ppm (Supplementary Figures 3B, C), suggesting that sodium chlorate could promote the odontoblast differentiation only at a toxic dose to mDP cells.

3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyltetrazolium bromide (MTT) is the reagent for an assay of the metabolic activity of viable cells. The MTT assay is performed according to the manufacturer’s instructions (Sigma-Aldrich, United States). Briefly, 10 μL of MTT solution is added to each well of a 96-well plate. Incubate for 2 h in an incubator. Add 100 μL of formazan crystal solubilizer to the medium. Optical density was then measured using a microplate reader at a wavelength of 570 nm.

Primary mouse dental papilla mesenchymal cells (mDP cells) were used in this study. The first molars of the mandibles of 3-day-old C57BL/6 mice were isolated with forceps under a stereomicroscope and digested for 1 h at 37 C in a solution containing 3 mg/mL collagenase type I (Sigma-Aldrich, United States) and 4 mg/mL dispase (Worthington Biochem, United States) as previously described (Wang F. et al., 2013). The enzymatically digested tissues were then filtered through a cell strainer (40 μm) to remove clumps and debris. The remaining cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen, United States) containing 10% fetal calf serum plus penicillin (100 units/mL) and streptomycin (100 μg/mL) at 37 C in a humidified atmosphere of air containing 5% CO₂. Then, mDP cells were isolated by the side population (SP) discrimination assay based on efflux of the fluorescent dye Hoechst 33342 detected by FACS. Cells were maintained as a standard procedure at density ∼0.8–1.0×10⁵ cells/cm² with DMEM containing 10% fetal calf serum plus penicillin (100 units/mL) and streptomycin (100 μg/mL) at 37 C in a humidified atmosphere of air containing 5% CO₂. The medium was refreshed every 2 days, and the cells were plated after reaching confluence. The mDP cells expressed mesenchymal stem cell marker and exhibited differentiate capacity into odontoblasts, chondrocytes, and adipocytes (Supplementary Figure S4). The antibody used for immunophenotyping were purchased from BioLegend (United States, San Diego) as follows: APC-conjugated anti-CD105 (cat.120413), anti-CD73 (cat.127209), anti-CD29 (Cat.102215), anti-CD90 (cat. 140311) or APC-conjugated Isotype control IgG were used for immunolabeling of positive markers. PerCP-conjugated anti-CD45 (cat.103129), anti-CD34 (cat.119327) and PerCP-conjugated Isotype control IgG were used for immunolabeling of negative markers.

MA-T is a stable and mild chlorine dioxide-generating reagent. MA-T contains sodium chlorite, in combination with one of two kinds of cationic detergents that serve as the Lewis acid catalyzing the generation of chlorine dioxide (the Lewis acidity of both were (α: LUMO (Lowest Unoccupied Molecular Orbital) = −4.12 eV) and (γ: LUMO = −4.02)), in a buffer stabilizing the solution at a neutral pH. MA-T was provided by Acenet Inc. (Tokyo, Japan).

Immunohistochemistry was performed on frozen sections of developing tooth germs and mDP cells. The sulfated HSPG distribution was visualized with 10E4 antibody (1:100; Seikagaku, Tokyo, Japan), as described previously (Hayano et al., 2012).

The immunofluorescence image was visualized using an all-in-one fluorescence microscope (BZ-X700, Keyence, Osaka, Japan) that is equipped with filters for GFP (excitation: 475 nm, emission: 525 nm) and DAPI (excitation: 360 nm, emission: 460 nm) channels. The instrument was controlled by the BZ Viewer version 1.0 software (Keyence, Osaka, Japan).

The mDP cells were plated in 6-well plates in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) until they reached confluence. The mDP cells were then cultured without or with MA-T solution at 1.0 ppm or 0.2 ppm for 48 h, or 2 weeks before processing for qPCR analysis.

qPCR was performed in triplicate on three independent sets of samples as previously described (Hayano et al., 2012; Ishimoto et al., 2015). The relative amount of transcripts was determined using a standard curve and normalized compared to the expression of glyceraldehyde-3-phosphate dehydrogenase gene (Gapdh) mRNA. The sets of synthetic primers used for the amplification were as follows: Gapdh, 5′-GTC​CCG​TAG​ACA​AAA​TGG​TG-3' (sense) and 5′-CAA​TGA​AGG​GGT​CGT​TGA​TG-3' (antisense); Mouse Dspp, 5′-AAC​TCT​GTG​GCT​GTG​CCT​CT-3' (sense) and 5′-TAT​TGA​CTC​GGA​GCC​ATT​CC-3' (antisense); Mouse Dmp1, 5′-CAG​TGA​GGA​TGA​GGC​AGA​CA-3' (sense) and 5′-TCG​ATC​GCT​CCT​GGT​ACT​CT-3' (antisense); Mouse Wnt5a, 5′-CGC​TAG​AGA​AAG​GGA​ACG​AAT​C-3' (sense) and 5′-TTA​CAG​GCT​ACA​TCT​GCC​AGG​TT-3' (antisense); Mouse Wnt6, 5′-TTC​CAG​TTC​CGT​TTC​CGA​CG-3' (sense) and 5′-CTG​TCT​CTC​GGA​TGT​CCT​GC-3' (antisense); Mouse Wnt10a, 5′-CAT​CTT​CAG​CCG​AGG​TTT​TCG-3' (sense) and 5′-AGC​CTT​CAG​TTT​ACC​CAG​AGC-3' (antisense); Mouse Notum, 5′-CAA​CCG​GGA​GAA​CTG​TGA​TT-3' (sense) and 5′-ACC​ACC​TCC​TGG​ATG​ATG​AG-3' (antisense); Mouse Dkk1, 5′-TCA​ATT​CCA​ACG​CGA​TCA​AGA-3' (sense) and 5′-GGC​TGG​TAG​TTG​TCA​AGA​GTC​TGG-3' (antisense).

Each amplification reaction was performed and checked for the absence of nonspecific PCR products by melting curve analysis using the LightCycler™ system (Roche, Germany). Relative cDNA copy numbers were calculated using data obtained from serial dilutions of representative samples for each target gene. Comparison of quantitative variables in two groups was performed using the Mann-Whitney U test. p values less than 0.05 were considered significant. Significance values were calculated using statistical analysis software with GraphPad Prism 8 (Graph Pad Software, United States).

The day before transfection, the mDP cells were plated and seeded at 150,000–300,000 cells in 35-mm cell culture dishes. Medium was replaced immediately before transfection. For transfection (all volumes per dish), 4 μL Lipofectamine RNAiMAX (Life Technologies, United States) was combined with 250 μL Opti-MEM (Life Technologies, United States) and separately 30 pmol of RNAi construct (Ambion Silencer Select Pre-designed siRNAs, Wnt10a ID:s76061; Wnt6 ID:s76095 and Universal Negative Control Number 1, Life Technologies) was used as negative control siRNA) were combined with 250 μL Opti-MEM. Both solutions were combined after 5 min and incubated for a further 20–30 min at room temperature before the transfection complexes were added dropwise to the cells. All transfections were performed in triplicate. The transfected cells were incubated for 4–6 h at 37°C in 5% CO₂, and the medium was replaced with fresh DMEM, Ham’s F12, 10% FBS, 2 mM glutamine. Western blotting was performed after 48 h unless otherwise noted. For growth assays, transfected cells were detached by trypsinization and seeded at 4,000–5,000 cells/well.

Mouse incisor organ culture has been previously described previously (Sarper et al., 2018). Briefly, the lower incisor was dissected from the E16.0 embryo and cultured on a track-etched polycarbonate membrane filter (Nuclepore) in Trowell-type organ culture with serum-free, chemically-defined medium (BGJB, Gibco) with or without MA-T (0.2 ppm). Tissues were harvested after 10 days of culture and processed for micro-CT and histologic evaluation.

Cultured mouse incisors were fixed in 4% paraformaldehyde overnight and scanned by micro-CT (R_mCT2, Rigaku) at 90KV, 200 μA, microfocus 5 µm/voxel size. Volume Graphics (VGstudio) MAX 2.2 software was used to reconstruct the three-dimensional images.

Sagittal and frontal sections of mandibular micro-CT images were utilized to measure dentin thickness. Dentin thickness was measured on a micro-CT image of the frontal section at the central position in the sagittal section. From the frontal cross-sectional image of the incisor, a line was drawn through the center of the enamel shape. Dentin thickness was then measured at the lingual and labial sides along this line using ImageJ. Thresholds for identifying enamel and dentin on the labial side were determined based on segmented objects with and without dentin. The same intensity thresholds were applied to all specimens.

After micro-CT scanning, the fixed mouse incisors were soaked in a mild decalcifier, Osteosoft (Sigma-Aldrich), for tooth decalcification. Sagittal sections of paraffin-embedded mandibles were prepared and used for hematoxylin-eosin (HE).

Cell viability was determined using the 3-(4, 5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) dye. Odontoblast cells were incubated with a series of different concentrations of MTT or sodium chlorite. After incubation, 10 μL MTT (Sigma-Aldrich, Japan) was added to each well of a 96-well microplate, and the microplates were placed in an incubator at 37°C for 4 h. One hundred and 50 μL of dimethyl sulfoxide (DMSO) was added to all wells and thoroughly mixed to lyse the cells and dissolve the dark blue crystals. After 10 min, the absorbance was measured at 570 nm using a microplate reader (Bio-Rad, Japan).

The public scRNA-seq dataset for GSE146855 of mouse incisors was downloaded from the GEO database to reanalyze the expression profile in odontoblast differentiation. The data were analyzed using a package from Seurat (version: 4.0.5) with R studio (Butler et al., 2018). In total, 6,260 cells were reanalyzed. After normalization, scaling, and principal component analysis (PCA) of the data, cells were clustered using FindNeighbors (dims = 1:6) followed by FindClusters (resolution = 0.35). The RunUMAP function was used to visualize the cell clusters.

Differential expression and cell identification were performed using FindAllMarkers (min.pct = 0.25, logfc. threshold = 0.25) with the Wilcoxon rank sum test. Visualization of gene expression by feature plot and dot plot was performed using the Seurat function FeaturePlot and DotPlot, respectively. KEGG enrichment analysis was performed on differentially expressed genes using DAVID online tools (https://string-db.org/).

The bioinformatics in silico ChIP analysis was performed using ChIP-Atlas (https://chip-atlas.org), an integrative and comprehensive data mining suite of public ChIP-Seq data. BigWig data of Lef1, Tcf3, Tcf4, Tcf7 and RNA poly ChIP-Seq performed in different cell types were visualized around the Wnt6, Wnt10a, Notum and Dkk1 TSS. In addition, all data were mapped to the reference human genome (mouse 9) using the Integrative Genomics Viewer (IGV). Our analyses included the following publicly available ChIP atlas data sets SRX9641211, SRX1411163, SRX360319, SRX5825990.

To identify binding sites for key regulators, we searched collections in the ChIP Atlas database (https://chip-atlas.org/). This database is comprehensive and integrative, allowing visualization of public ChIP-seq data from over 131,000 experiments submitted to the Sequence Read Archives at NCBI, DDBJ, or ENA (Oki et al., 2018). The antigen class was set to ‘TFs and others’, the cell type class to ‘gonad’ and ‘embryo’, and the threshold for statistical significance values calculated by MACS2 (−10*Log10 [MACS2 Q-value]) was 20, 200, 500.

Statistical methods were not used to predetermine sample size. Statistical analyses were performed with GraphPad Prism 8. Student’s two-sided t-test and two-way ANOVA were used under the assumption of normal distribution and observance of similar variance. A p-value of <0.05 was considered significant. Bonferroni post hoc analysis was performed where applicable. Values are expressed as mean ± SD. Data shown are representative images; each analysis was performed on at least three mice per genotype. Immunostaining was performed at least in triplicate.