Amelogenin phosphorylation affects key regulatory genes in the enamel organ

Abstract

Amelogenin (AMELX), the predominant enamel matrix protein (EMP), has a single phosphorylation site at Serine 16 (S16), which enhances its capacity to stabilize amorphous calcium phosphate and inhibit crystallization in vitro. To investigate the role of AMELX phosphorylation in vivo, we generated a knock-in (KI) mouse model, Amelx^S16A, in which S16 is substituted with Alanine that prevents AMELX phosphorylation. KI enamel is hypoplastic, lacks the enamel rods, and features multiple ectopic calcifications. Notably, KI ameloblasts lack Tomes’ processes and develop progressive cell pathologies, suggesting that AMELX phosphorylation affects ameloblast biology. To gain a comprehensive view on the effects of AMELX phosphorylation on the biology of the enamel organ (EO) cells, we performed single-cell RNA sequencing on incisal EOs from WT and KI mice. Using this approach, we found that 624 genes were differentially expressed between WT and KI in total EO cell populations. In particular, genes encoding secretory stage EMPs were upregulated in KI mice, while genes encoding maturation-stage EMPs were downregulated. Ingenuity Pathway Analysis revealed that the pathways associated with cell morphology, growth, and proliferation are most affected in KI cells. Ten distinct cell populations of the EO were identified using cluster analysis. Additionally, our analyses showed that AMELX phosphorylation affects the expression of several genes involved in the regulation of odontogenesis. Furthermore, differentially expressed genes were concentrated in the secretory ameloblast (SA) cluster, suggesting that the SAs were most affected by the lack of AMELX phosphorylation. Specifically, there was a substantial 5.1-fold downregulation of Shh, while Wnt5a was upregulated 8.1-fold in KI SAs compared to WT. This dysregulation of regulatory genes may contribute to the observed ameloblast abnormalities in KI. In summary, our transcriptomics analysis reveals that AMELX phosphorylation not only affects numerous extracellular processes during amelogenesis but also influences multiple cellular pathways and processes within the EO.

Introduction

Dental enamel is the hardest tissue of the human body comprising ~96% w/w of mineral and ~1% w/w of enamel matrix (EM) (1). The basic building block of enamel is a parallel array of high-aspect ratio carbonated hydroxyapatite (CHA) crystals called an enamel rod. The enamel rods, surrounded by inter-rod enamel, are further organized into an intricate decussating pattern (2, 3). This hierarchical structural organization of enamel contributes to its unique mechanical properties, combining high hardness and fracture toughness (2, 4, 5).

Enamel formation, amelogenesis, is carried out by ameloblasts - specialized epithelial cells of the enamel organ (EO). Amelogenesis occurs in four stages: presecretory, secretory, transition, and maturation, each associated with distinct ameloblast phenotypes (1, 6). During the presecretory stage, presecretory ameloblasts form a few μm thick rod-less enamel layer, while differentiating into secretory ameloblasts by developing a specialized secretory apparatus at their distal end, called the Tomes’ process. During the secretory stage, each Tomes’ process secretes one enamel rod and a portion of the interrod, as the ameloblasts move away from the dentin-enamel interface (DEJ). Secretory enamel contains only ~30% mineral by weight, in the form of tens of microns long and ~5 nm thick mineral nanoribbons suspended in a self-assembled gel, consisting of EMPs ameloblastin (AMBN) and enamelin (ENAM) and AMELX, the most abundant EMP comprising 90% of the secretory enamel matrix (1, 7). Another important constituent of secretory enamel is matrix metalloproteinase 20 (MMP20), which cleaves EMPs at specific sites. These proteolytic cleavages are essential for proper enamel formation (8–10). During the transition stage, enamel deposition continues, while Tomes’ processes are resorbed, leaving a thin rod-less surface layer. When the full thickness of enamel is achieved at the end of transition, the enamel maturation stage begins. During this next stage, the mineral content increases, via accretion of the existing mineral nanoribbons, while the organic matrix is primarily degraded by kallikrein 4 (KLK4) and removed by endocytosis (10, 11). Mutations in genes encoding EMPs and enamel proteinases cause non-syndromic Amelogenesis Imperfecta (AI) (12), a group of inherited diseases manifested by hypoplastic and/or hypomineralized weak enamel. Multiple mutations in the gene encoding AMELX, located on the X chromosome, cause X-linked AI (12), and although no mutations affecting AMELX phosphorylation site were identified, so far, mutations in Golgi kinase Fam20C which phosphorylates extracellular matrix proteins in mineralized tissues (13) cause AI (14).

In contrast to the majority of extracellular matrix proteins in mineralized tissues which are heavily phosphorylated (15–17), AMELX contains a sole phosphorylation site at Serine 16 (S16) (18), which is the only post-translational modification of this protein. AMELX phosphorylation is likely carried out by FAM20C, the Golgi kinase responsible for phosphorylating the extracellular matrix proteins of mineralized tissues (19–21). Our in vitro mineralization experiments revealed that phosphorylated AMELX stabilizes amorphous calcium phosphate (ACP) and inhibits apatitic crystal formation (22–24). ACP is the first mineral phase formed in enamel that later undergoes transformation to hydroxyapatite-like enamel crystals during the secretory stage of amelogenesis (8, 25). To understand the role of AMELX phosphorylation in vivo, we generated a knock-in mouse model Amelx^S16A (KI) in which S16 is substituted with Alanine (A) that prevents AMELX phosphorylation as determined by immunohistochemistry using a custom antibody raised against the phosphorylated AMELX peptide (26). This single amino acid substitution causes a severe enamel phenotype manifested by weak, hypoplastic enamel, the loss of enamel rods, and multiple ectopic calcifications (26, 27). Notably, in KI mice lacking AMELX phosphorylation mineralization and ACP to CHA transformation are accelerated during the secretory stage of amelogenesis, leading to increased release of protons and acidification of the enamel space (26, 28). These findings indicate that in vivo phosphorylated AMELX stabilizes amorphous calcium phosphate (ACP) and regulates its transformation to CHA, in agreement with our in vitro experimental results (22–24). Importantly, KI ameloblasts lack Tomes’ processes and develop progressive cell pathology (26). This observation strongly suggests that in addition to the extracellular effects of AMELX phosphorylation on the biological control of enamel biomineralization, it has a profound effect on the ameloblast cell biology, either directly, by affecting interactions of AMELX with other molecules, or indirectly, by preventing the acidification of the enamel space, as observed in KI enamel (28).

The goal of the current study was to test our hypothesis that AMELX phosphorylation not only affects extracellular events but also influences multiple cellular processes and pathways during amelogenesis. To test this hypothesis, we conducted single-cell RNA sequencing (scRNA-seq) analyses of the incisal EOs from wild type (WT) and KI mice.

Identification of triads of cluster specific markers.

Cluster 5 was identified as secretory stage ameloblasts (SA) based on the high expression levels of AMELX, Enam and Mmp20, and cluster 7 based on its high expression of Odam, Amtn and Klk4 was identified as maturation stage ameloblasts (MA) (1). Cluster 1 contains two subpopulations. One is stratum intermedium (SI) based on the strong expression of Cldn10, Enpp2 and Alpl (31, 32). Another is stellate reticulum (SR) expressing Krt17, Krt14 and Notch2 (33). Krt15 and Gja1 are associated with the papillary layer in cluster 2 (30, 34, 35). Cluster 3 was identified as the periodontium (PD) based on strong expression of Bmp3 (36) Col6a1 (37), and Ogn (38). Cluster 4 contains cells of the macrophage lineage (ML) based on the expression of Ccl7 (C-C motif chemokine ligand 7), Ccl2 (C-C motif chemokine ligand 2), and C1qa (Complement C1q A chain) (39, 40). Cluster 6 contains progenitor cells of cervical loop (PCCL), based on the high expression levels of Pthlh, Pitx2 and Epcam (32, 41). Cluster 8 contains endothelial progenitor cells (EPC), based on the expression of Cd34, Kdr (VEGFR2) and Ecscr (42–45). Cluster 9 contains cells of odontoblastic lineage (OBL) based on the strong expression of Smdp3, Dlk1, and Dmp1 (29, 46–48). Cluster 10 consisted of the lymphocytes (LC) based on the high expression of Ltb, Ptprcap, Cd53 (49–51).

To validate the reliability of the scRNA-seq results, we have assessed mRNA expression of four genes, Amelx, Odam, Shh and Wnt5a, in total EOs of 12-wo maxillary incisors by RT-qPCR, using primers listed in Table 1. The entire EO layer was collected from the labial side of the maxillary incisors. At least three biological replicates containing four EOs from two mice were collected and mRNA was extracted as described in our published protocol (6). Briefly, the tissue was isolated with chloroform and isoamyl alcohol (Sigma C0549) solution (24:1), mixed with 70% ethanol followed by purification with a PureLink RNA Mini Kit (Invitrogen). Approximately, 200 ng of total RNA, quantified by a NanoDrop One C (Thermo Scientific) spectrophotometer, was obtained from pooling 4 maxillary incisors (= 2 mice). The obtained samples were subjected to reverse transcriptase reaction using SuperScript IV enzyme (Thermo Fisher) for 20 min at 50° C. RT-qPCR analyses were performed with QuantStudio 6 Flex Real Time PCR system (Applied Biosystems) and PowerTrack^™ SYBR Green Master Mix (Thermo Fisher). The gene expression levels of the genes of interest were normalized to β-actin mRNA levels and expressed as the mean ± SD (n = 3 to 6 replicates). Statistical analyses were carried out using two-tailed t-test in GraphPad Prism software. OriginLab software (2017) and CorelDraw (2020) were used to generate graphs.

Table 1.

List of primers for RT-qPCR experiments.

Gene	Amplicon size	Primer	Sequence
Amelx (Amelogenin)	254 bp (WT)	F	AAG CTA CCA CCT CAT CCT GGA AGC
	289 bp (KI)	R	TTG AGT CAG AGT GAC CAG GTA GGA AG
Odam (Odontogenic ameloblast-associated protein)	197 bp	F	TTGACAGCTTTGTAGGCACA
		R	GACCTTCTGTTCTGGAGCAA
Wnt5a (Wnt Family Member 5A)	247 bp	F	TGCCATGTCTTCCAAGTTCTTCCTAATG
		R	TGATGCCTGTCTTCGCACCTTCTCCA
Shh (Sonic Hedgehog Signaling Molecule)	127 bp	F	GTTCATTGCAGCTTTCTAGAGG
		R	TACGGAATGGCTCCATTGG
β-actin (Beta actin)	76 bp	F	GGGAAATCGTGCGTGACATC
		R	GCGGCAGTGGCCATCTC

Results.

Transcriptome analysis of the total EO cell population.

Our scRNA-seq analysis revealed that 624 out of more than 23,000 genes were differentially expressed between WT and KI in the total cell population of incisal EOs. The most differentially expressed gene was Krt71, a hair keratin expressed in the EO (52, 53) (Fig. 1A; Suppl. Table). Its expression was 10-fold higher in KI vs. the WT. Furthermore, the major regulator associated with tooth development, Shh (54), and in amelogenesis in particular (55, 56), was downregulated by 2.9-fold in KI, relative to the WT (Fig. 1A; Suppl. Table). We further assessed the differential expression of genes associated with secretory stage amelogenesis, including Amelx, Enam, Ambn and Mmp20, and Amtn, Odam and Klk4, which are primarily expressed during the maturation stage (1). The secretory stage genes were found to be upregulated in KI, while the maturation stage genes were downregulated in KI relative to the WT (Fig. 1B). We have further applied IPA to our dataset. This analysis revealed that pathways associated with cell morphology, cell growth, and proliferation are the most affected by AMELX phosphorylation (Fig. 1C).

Figure 1.

Whole cell population analysis. A. Top differential expressed genes in the total cell population of the incisal enamel organs of KI vs WT mice. B. Differential expression of enamel matrix protein genes in the total cell population of the incisal enamel organs of KI vs WT mice. C. The results of Ingenuity Pathway Analysis.

Cluster analysis and identification of cell populations of the EO using triads of highly expressed genes.

Cluster analysis of the EO transcriptome has identified 11 cell populations (clusters), which were visualized using Uniform Manifold Approximation and Projections (UMAPs) to construct a 3-D representation of the WT and KI EO cell cluster data (Figs. 2A,B). Cluster 11 contains genes involved in the regulation of mitochondrial activity, neurogenesis and angiogenesis; however, it was excluded from further analysis since its cell population identity could not be firmly established, and because of its small size – 4.4% of WT and 1.7% of KI total cell population. Notably, there was a substantial difference in the relative abundance of cells in several clusters between WT and KI (Figs. 2A–D). The results of our cluster analysis are in good agreement with those from prior sc-RNAseq studies (29, 30, 32).

Figure 2.

Transcriptome cluster analysis and cell population identification of the in WT and KI incisal enamel organs. UMAPs of WT (A) and KI (B) transcriptomes with 10 clusters. The clusters were identified as the enamel organ distinct cell populations based on the expression profiles of triads of gene markers listed next to the corresponding clusters and color-coded accordingly. Bar graphs presenting cluster sizes, as a proportion of the total cell population, in WT (C) and KI (D) enamel organs. (E) and (F) Bubble-maps of the expression of selected marker genes in clusters 1 through 10. Cluster identification: 1 – stratum intermedium/stellate reticulum (SI/SR), 2-papillary layer (PL), 3 – periodontium (PD), 4 – monocytes and macrophages (MC), 5 – secretory ameloblasts (SA), 6 – progenitor cells of the cervical loop (CL), 7 – maturation stage ameloblasts (MA), 8 – endothelial progenitor cell (EPC), 9--odontoblast lineage cells (OLC), 10 – lymphocytes (LC).

We selected several EMP genes associated with amelogenesis and applied heat mapping to identify cell populations expressing these genes, based on the z-score (Figs. 2E,F). Expression of the genes of interest mainly localized to two clusters: cluster 5 showed a higher expression of secretory EMPs genes Amelx and Enam, and Mmp20, while cluster 7 showed a higher expression of maturation stage-related EMPs genes Odam and Amtn and Klk4 (Figs. 2A,B,E,F). Based on this expression profile we have assigned cluster 5 as containing secretory stage ameloblasts and cluster 7 as containing maturation stage ameloblasts. Notably, the proportion of secretory ameloblasts in KI EOs increased from 4% to 26% of the total cell population, while the proportion of maturation ameloblasts decreased from 8% to 4% of the total cell population, This finding might suggest a prolonged secretory stage in KI mice (Figs. 2C,D).

To identify other cell populations comprising the clusters, we have analyzed the lists of the twenty top expressed genes in each cluster (Supplementary Materials) and conducted a comprehensive literature search to select triads of gene markers characteristic of different cell populations of the EO (Fig. 2 and 3). The details of the marker triads identification procedures are presented in the Materials and Methods section. Cluster 1 was identified as stratum intermedium and stellate reticulum (SI/SR) (Fig. 2 and 3). Cluster 2 was identified as the papillary layer (PL) (Fig. 2 and 3). Cluster 3 was identified as the periodontium (PD) and cluster 4 was identified as monocytes and macrophages (MC) (Figs. 2 and 3). As mentioned above, clusters 5 and 7 were identified as secretory (SA) and maturation (MA) stage ameloblasts, respectively (Fig. 2 and 3). Cluster 6 contains progenitor cells of the cervical loop (CL) and cluster 8 contained endothelial progenitor cells (EPCs) (Fig. 2 and 3). Cluster 9 contained odontoblast lineage cells (OLC) and cluster 10 contained lymphocytes (LC) (Fig. 2 and 3).

Figure 3.

Identification of individual clusters using triads of marker genes. Pairs of UMAPs of WT and KI transcriptomes showing distributions of cells expressing marker genes for each cluster, with trinary plots of the marker gene expression in corresponding clusters. Only cells with the expression levels above 50% of the maximum were included.

Differential gene expression analysis shows differences in KI secretory and maturation stage ameloblasts relative to those found in the WT.

We have performed a differential expression analysis of genes of interest associated with the enamel matrix in secretory (cluster 5) and maturation stage (cluster 7) ameloblasts between KI and WT. In the secretory stage ameloblasts in KI, Ambn expression was more than 2.3-fold higher in KI vs WT, while Amtn expression was 2.3-fold lower (Fig. 4A). Odam was slightly downregulated in both clusters, but this difference was below two-fold threshold (Fig. 4). In KI maturation stage ameloblasts, secretory EMPs Amelx, Enam and Ambn were overexpressed compared to the WT by factors of 2.9, 2.2. and 1.8, respectively, while Mmp20 expression did not differ between the two genotypes (Fig. 4B).

Figure 4.

Differential expression of the enamel matrix protein genes in secretory (A) and maturation (B) clusters of the incisal enamel organs of KI vs WT mice.

Differential expression analysis of genes involved in the regulation of tooth development.

The fact that KI ameloblasts undergo major phenotypic changes including the absence of the Tomes’ processes, the progressive loss of cell polarity and ameloblast layer integrity (26), as well as the dramatic increase in the relative size of SA cluster prompted us to assess the differential expression of regulatory genes involved in odontogenesis (54, 57, 58). We found that several members of Wnt family – Wnt5a, Wnt6, Wnt10a and Wnt7b and Wnt receptor Fzd2, as well as Pitx2 and Shh are differentially expressed in at least one cluster (Fig. 5A). All but one gene, Wnt5a, were downregulated in KI vs WT. Out of the six genes tested, five were differentially expressed in the SA cluster, while in other clusters no more than 2 genes crossed two-fold threshold (Fig. 5A). Wnt5a was affected greatly, with an 8.1-fold upregulation in the SA cluster (Figs. 5A,B), while Wnt6 and Wnt10a were downregulated, 3.7 and 3-fold, respectively. Wnt receptor Fzd2 was downregulated 4.6-fold, while Shh was downregulated 5.1-fold in SA (Figs. 5A,C). Together these results indicate that AMELX phosphorylation critically affects regulatory gene expression patterns, especially in the secretory stage ameloblasts.

Figure 5.

Expression patterns of selected regulatory genes. (A) Differential expression of regulatory genes in 10 identified cell populations of the incisal enamel organs of KI vs. WT mice. UMAPs showing the distribution of Wnt5a (B) and Shh (C) among 10 identified cell populations of the incisal enamel organs of WT and KI mice.

Validation Experiments.

To validate the scRNA-seq results, we conducted mRNA expression analysis of four highly differentially expressed genes. We selected Amelx which has expression levels are more than 5 folds higher in the total EO cell population (Fig. 1A,B) and are substantially higher in SA and MA clusters (Fig. 3). We also selected Odam, which is downregulated in KI (Fig. 1A,B, Fig. 3) and two key regulatory genes – Shh and Wnt5a. Shh expression in KI is more than 3 fold lower in the total cell population of EO (Fig. 1A) and more than 5 fold lower in SA (Fig. 5A), while Wnt5a expression is 8 times higher in SA cluster in KI (Fig 3A). The results of the PCR experiments demonstrate significant increases in Amelx and Shh expression, and significant decreases in Odam and Wnt5a expression in KI (Fig. 6), in excellent agreement with the scRNA-seq results.

Figure 6.

mRNA expression of Amelx (A), Odam (B), Wnt5a (C) and Shh (D) in the incisal EOs of 12-wo WT and KI mice.

Discussion

The present study was designed to test the hypothesis that the state of AMELX phosphorylation influences ameloblast cell biology. Accordingly, we conducted scRNA-seq analysis of the incisal EOs from WT mice and from KI mice that lack AMELX phosphorylation. Indeed, our results demonstrate that 624 genes are differentially expressed confirming our hypothesis. Importantly, among the affected genes, 511 were downregulated and 113 were upregulated in KI ameloblasts relative to the WT. Among the differentially expressed genes, there were several genes of interest, such as Shh, which is the major regulator of odontogenesis in general and in amelogenesis in particular (55, 56), several keratins, and genes associated with cell-cell junctions and cell adhesion, among others. Our differential expression analysis of seven EMPs and proteinases in the total cell population revealed that the secretory stage genes Amelx, Enam and Ambn are upregulated in KI mice, relative to the WT, while maturation stage genes Amtn and Odam are downregulated, potentially suggesting that the transition from secretory to maturation stage is delayed during KI enamel formation.

Furthermore, the IPA of the total cell population revealed that the pathways affected by AMELX phosphorylation are associated with the regulation of cell morphology and polarity, cell differentiation, growth and proliferation, which are clearly reflected in the observed KI ameloblast phenotype manifested by the lack of Tomes’ processes, loss of cell polarity and of the integrity of the ameloblast cell layer (26). In addition, genes associated with AI, such as Acp4 and Relt (59–62), were identified using Ingenuity Pathway Analysis as molecules affected by AMELX phosphorylation status (Fig. 1C). Of note, mutations in Acp4 cause nonsyndromic AI with hypoplastic pitted enamel (60, 61) and Acp4^R110C KI mice feature enamel phenotype closely resembling Amelx^S16A KI (26) characterized by hypoplastic rod-less enamel with multiple ectopic calcifications (63).

The cluster analysis of the total cell population of the incisal EOs revealed 10 clusters, representing the major cell types including SA, MA, SI and PL, among others. In line with the total cell population analysis, the differential expression analysis in SA and MA clusters revealed that secretory EMP genes were upregulated in KI MA cluster. Taken together the results of our differential expression analysis reflect the delay of histodifferentiation processes in the cells of ameloblast lineage and specifically the transition from secretory to maturation stage ameloblasts. Recent results obtained in our laboratory are in line with this suggestion of an extended secretory stage in KI enamel (unpublished findings). Dysregulation of regulatory genes in KI enamel development greatly affected their expression profiles.

We identified six regulatory genes which were differentially expressed between WT and KI, and all of them except Wnt5a are downregulated. Notably, the differentially expressed regulatory genes are concentrated in the SA cluster, suggesting that the secretory ameloblasts are most affected by the lack of AMELX phosphorylation. Shh was the most downregulated gene in SA cluster in the KI. Shh is the major regulator of odontogenesis, in particular it induces differentiation of the cells of the ameloblast lineage (55, 56). Notably, HhAntag (a hedgehog signaling antagonist) treatment in mice leads to cessation of enamel formation (55). It was also reported that in the absence of the major Shh receptor Smo in the enamel epithelium, ameloblasts failed to properly differentiate (54). Specifically, the ameloblasts failed to polarize and form the Tomes process and to form a cohesive cell layer. In Amelx^S16A KI the secretory ameloblasts similarly do not form Tomes’ processes, gradually dedifferentiate, lose their polarity and ameloblast layer integrity (26). These phenotypic similarities between ameloblasts in Smo co-KO and AMELX^S16A KI mice and the fact that Shh is downregulated in KI suggests that the Shh deficit contributes to the observed ameloblast pathology. Even though Shh is expressed in several in cell populations of the EO, such as SI and preameloblasts, our scRNA-seq results revealed its downregulation primarily in SA cell population. Together, these results indicate that the lack of AMELX phosphorylation selectively and directly affects SA population.

At this point we can only surmise how the sole phosphorylation of the major EMP AMELX affects biological processes in ameloblasts in such a profound way. However, there are several plausible scenarios. We have recently demonstrated that the absence of AMELX phosphorylation accelerates enamel mineralization during the secretory stage of amelogenesis, which is accompanied by the acidification of the enamel space due to excessive proton release (28). The most plausible explanation for the observed changes in gene expression is that this acidification directly impacts ameloblast biology, altering expression patterns, as has been shown in other systems (64). A similar phenomenon is observed in mice lacking Cftr, a H⁺/Cl⁻ exchanger involved in pH regulation during the maturation stage. In these mice, enamel acidification leads to pronounced phenotypic changes, most notably the cessation of ameloblast modulation (65, 66). This would also explain why the expression changes are primarily concentrated in ameloblast cell population, since they are facing the enamel space. A second possibility is that AMELX phosphorylation affects its interactions with pericellular and transmembrane proteins on the apical ends of ameloblasts, such as ACP4, which can also affect the ameloblast cell biology. Further investigations are necessary to identify the exact mechanisms by which AMELX phosphorylation affects the ameloblast cell biology and other EO cell populations.

Overall, our scRNA-seq analysis demonstrates that AMELX phosphorylation affects numerous pathways and processes in the EO, and in ameloblasts, in particular. Our new findings highlight the critical role phosphorylated AMELX plays in regulating enamel formation and in the maintenance of secretory ameloblast integrity and function. Presented results provide new insights into the process of amelogenesis and should contribute to new hypotheses for further mechanistic studies of cell-matrix and mineral-matrix interactions which will improve our understanding of enamel formation and ultimately will aid in the development of new treatments and biomaterials for enamel repair.