FAM20A Mutations and Transcriptome Analyses of Dental Pulp Tissues of Enamel Renal Syndrome

Abstract

Aim:

Biallelic loss-of-function FAM20A mutations cause amelogenesis imperfecta (AI) type IG, better known as enamel renal syndrome (ERS), characterized by severe enamel hypoplasia, delayed/failed tooth eruption, intrapulpal calcifications, gingival hyperplasia, and nephrocalcinosis. FAM20A binds to FAM20C, the Golgi casein kinase (GCK), and potentiates its function to phosphorylate secreted proteins critical for biomineralization. While many FAM20A pathogenic mutations have been reported, the pathogeneses of orodental anomalies in ERS remain to be elucidated. This study aimed to identify disease-causing mutations for patients with ERS phenotypes and to discern the molecular mechanism underlying ERS intrapupal calcifications.

Methodology:

Phenotypic characterization and whole exome analyses were conducted for 8 families and 2 sporadic cases with hypoplastic AI. A minigene assay was performed to investigate the molecular consequences of a FAM20A splice-site variant. RNA sequencing followed by transcription profiling and gene ontology (GO) analyses were carried out for dental pulp tissues of ERS and the control.

Results:

Biallelic FAM20A mutations were demonstrated for each affected individual, including 7 novel pathogenic variants: c.590-5T>A, c.625T>A (p.Cys209Ser), c.771del (p.Gln258Argfs*28), c.832_835delinsTGTCCGACGGTGTCCGACGGTGTCCA (p.Val278Cysfs*29), c.1232G>A (p.Arg411Gln), c.1297A>G (p.Arg433Gly), and c.1351del (p.Gln451Serfs*4). The c.590-5T>A splice-site mutation caused Exon 3 skipping, which resulted in an in-frame deletion of a unique region of the FAM20A protein, p.(Asp197_Ile214delinsVal). Analyses of differentially expressed genes in ERS pulp tissues demonstrated that genes involved in biomineralization, particularly dentinogenesis, were significantly upregulated, such as DSPP, MMP9, MMP20, and WNT10A. Enrichment analyses indicated over-representation of gene sets associated with BMP and SMAD signaling pathways. In contrast, GO terms related to inflammation and axon development were under-represented. Among BMP signaling genes, BMP agonists GDF7, GDF15, BMP3, BMP8A, BMP8B, BMP4, and BMP6 were upregulated, while BMP antagonists GREM1, BMPER, and VWC2 showed decreased expression in ERS dental pulp tissues.

Conclusions:

Upregulation of BMP signaling underlies intrapulpal calcifications in ERS. FAM20A plays an essential role in pulp tissue homeostasis and prevention of ectopic mineralization in soft tissues. This critical function probably depends upon MGP (matrix Gla protein), a potent mineralization inhibitor that must be properly phosphorylated by FAM20A-FAM20C kinase complex.

Introduction

Enamel renal syndrome (ERS), also known as amelogenesis imperfecta type IG, (AI1G, OMIM #204690) is an autosomal recessive disorder primarily featured by severe hypoplastic (thin) or aplastic defects of dental enamel and nephrocalcinosis, which usually develops with age and remains asymptomatic (Lubinsky Angle et al. 1985; Hall Phakey et al. 1995; Paula Melo et al. 2005). Concomitant orodental anomalies are also frequently evident, including multiple impacted permanent teeth with dysmorphology, pulp stones, and hyperplastic gingiva with ectopic calcification, although their severities vary significantly among individuals (de la Dure-Molla Quentric et al. 2014; Hassib Shoeib et al. 2020). Pulp stones are intrapulpal calcifications often identified in aged teeth or those under chronic stimulation (Goga Chandler et al. 2008). However, the ERS pulp stones can be found in young and unerupted teeth, suggesting an etiology of pathological soft tissue homeostasis of dental pulp rather than a reactive physiological response (de la Dure-Molla Quentric et al. 2014; Wang Reid et al. 2014). Histological analyses have shown that ERS amorphous intrapulpal calcifications can replace almost the entire dental pulp but never fuse with circumpulpal dentine (Martelli-Junior Bonan et al. 2008). Also, contrary to regular pulp stones, ERS mineralized tissues in pulp consist of fibrodentine, orthodentine, but not osteodentine, further demonstrating their distinctive nature (Berès Lignon et al. 2018). It has been well documented that biallelic loss-of-function mutations of FAM20A (family with sequence similarity member A, OMIM *611062) cause ERS (O'Sullivan Bitu et al. 2011; Jaureguiberry De la Dure-Molla et al. 2012; Wang Aref et al. 2013). However, the molecular pathogenesis underlying ERS intrapulpal calcifications and how FAM20A functions to maintain dental pulp homeostasis are largely unknown.

FAM20A belongs to an evolutionarily conserved gene family that in humans includes FAM20B and FAM20C (Nalbant Youn et al. 2005). FAM20B is a glycan kinase that phosphorylates xylose at the proteoglycan linkage region (Koike Izumikawa et al. 2009). FAM20C is the long-sought-after Golgi casein kinase (GCK) that phosphorylates numerous secretory pathway proteins at the S-X-E/pS motif (Tagliabracci Engel et al. 2012). Despite its sequence similarity to these kinases, FAM20A lacks a residue critical for catalysis of phosphorylation and is a pseudokinase that potentiates the kinase activity of GCK by dimerizing into a functional complex with FAM20C (Cui Xiao et al. 2015). In addition, distinct from FAM20C, FAM20A has a unique pattern of disulfide bridges mediated by an inserted region of 17 amino acids. Deletion of this region significantly diminishes FAM20A’s ability to facilitate phosphorylation by GCK (Cui Zhu et al. 2017). Mutations in FAM20C cause Raine syndrome (RNS, OMIM #259775), a highly lethal disorder of osteosclerotic bone dysplasia (Simpson Hsu et al. 2007). In mild cases, the patients exhibit orodental phenotypes similar to those of ERS, including severe enamel hypoplasia, failed tooth eruption, gingival overgrowth, and intrapulpal calcification, indicating critical roles for the FAM20A-FAM20C kinase complex in development and homeostasis of both mineralized and soft tissues (Simpson Scheuerle et al. 2009; Fradin Stoetzel et al. 2011). However, while defective phosphorylation of enamel matrix proteins (known protein substrates of GCK), such as AMELX and ENAM, has been considered as the cause of enamel malformations in ERS and RNS (Cui Xiao et al. 2015), how the kinase complex helps maintain soft tissue homeostasis and prevents abnormal mineralization remains to be elucidated.

In this study, we characterized 8 families and 2 individuals with severe enamel hypoplasia and identified biallelic FAM20A mutations in each case. Seven variants have never been reported to be disease-causing. By conducting transcriptome analyses for pulp tissues from a patient, we determined the transcription profile of ERS pulp and identified altered gene expressions, signaling pathways, and biological processes that might be the cause of intrapulpal calcifications in ERS. Accordingly, we hypothesize a molecular mechanism and propose a role for the FAM20A-FAM20C kinase complex in dental pulp homeostasis.

Materials and Methods

The manuscript of this laboratory study has been written according to Preferred Reporting Items for Laboratory studies in Endodontology (PRILE) 2021 guidelines (Figure 1).

Figure 1.

PRILE 2021 flowchart of designs and results for this study.

Subject Recruitment and Mutational Analyses

The study protocols and consent forms for human research were reviewed and approved by the Institutional Review Boards at the University of Michigan and National Taiwan University Hospital. Following comprehensive explanation and thorough discussion of study contents, written consents were obtained from all participants. Oral examinations, collection of photographs/radiographs, and history taking were conducted for phenotypic characterization and construction of family pedigrees. For mutational analyses, 2-mL non-stimulated saliva was procured from each subject to obtain genomic DNA (Wang Lin et al. 2019). All these recruitment procedures were specified in our research protocols and are in compliance with the Declaration of Helsinki.

To search for pathogenic mutations that cause enamel malformations, whole exome sequencing and analysis were performed as previously described (Wang Zhang et al. 2021). Following identification of sequence variants, target amplification and Sanger sequencing were conducted to validate and discern their distribution in the family. The identified FAM20A variants were described following HGVS recommendations and annotated with NG_029809.1, NM_017565.4, and NP_006185.1 for numbering gDNA, cDNA, and amino acid sequence positions respectively.

Minigene Splicing Assay

A 723-bp DNA fragment, containing human FAM20A Exon 3 and parts of flanking intron sequences, was cloned into the pSPL3 vector (a gift from Dr. Tompson) by using XhoI and BamHI restriction enzyme sites to construct the wild-type minigene (Tompson & Young 2017). A splice-site mutation of c.590-5T>A was subsequently introduced to generate the mutant clone by site-directed mutagenesis. These two minigene constructs were respectively transfected into HEK293T cells for splicing analyses. Cells were harvested 24 hours after transfection, and RNA was extracted, reverse-transcribed, and amplified using V1-F (5’-TCTGAGTCACCTGGACAACC-3’) and V2-R (5’-ATCTCAGTGGTATTTGTGAGC-3’) primers. The amplification products were analyzed by agarose gel electrophoresis and Sanger sequencing.

Sample Collection, RNA sequencing, and analyses

Dental pulp tissues of left maxillary permanent canines (tooth number 11) were harvested respectively from the proband of Family 8 (Figure 2) and an age-comparable healthy female patient who had no systemic diseases and had undergone orthodontic treatment. Both teeth were impacted and had to be surgically removed. The harvested dental pulps from the extracted teeth were immediately immersed and incubated in ~10 volumes of RNAlater^® solution (Thermo Fisher Scientific, Waltham, MA USA) overnight at 4°C, and thereafter stored at −20°C before use. RNA processing and sequencing were conducted as previously described (Wang Lin et al. 2019). Briefly, the integrity of extracted RNA was assessed using the RNA 6000 Nano Kit in an Agilent 2100 Bioanalyzer. A cDNA library was prepared and submitted for 75 bp single-end sequencing on an Illumina NextSeq 500 Sequencer System. Bowtie2 (v2.2.6) and RSEM (RNA-Seq by Expectation Maximization) software were employed to map sequencing reads to the hg19 human reference transcriptome and to quantify transcript abundance respectively (Li & Dewey 2011). Differential gene expression (DGE) between ERS and control pulp tissues was analyzed using EBSeq (v1.16.0), followed by gene ontology (GO) and enrichment analyses performed with clusterProfiler (v3.8.1) (Yu Wang et al. 2012; Leng Dawson et al. 2013).

Figure 2.

Phenotypes of ERS Family 8 and FAM20A mutations. (a) The family pedigree indicates an autosomal recessive pattern of disease inheritance. (b) Clinical photographs of the proband (III:3), a 17-year-old female, show that she had a mixed dentition with over-retained maxillary primary canines. All teeth exhibited yellow-brown discoloration with almost no enamel coverage. (c) The panoramic radiograph reveals that both of her maxillary permanent canines were impacted with a pericoronal radiolucency over the left one. Enamel aplasia was evident on all teeth. Tooth number 11 was later surgically removed, and its pulp tissues used for transcriptome analyses. (d) The DNA chromatograms show two FAM20A pathogenic variants identified from the proband, c.590-5T>A and c.734_735del. Only she and her affected sister (III:2) were compound heterozygous to these 2 mutations.

The results were plotted using the GOBar and GOCluster functions of the GOplot package to present calculated z-scores and hierarchical clustering for fold changes or GO terms (Walter Sanchez-Cabo et al. 2015). The “genelist” item contained all differentially expressed genes with PPEE < 0.05, and the “genes” item was set to include only genes with a FC (fold change) > 2. For gene cluster analyses, 10 GO terms of biological process were selected for the “process” item: GO:0001501 (skeletal system development), GO:0034765 (regulation of ion transmembrane transport), GO:0030198 (extracellular matrix organization), GO:0009306 (protein secretion), GO:0030509 (BMP signaling pathway), GO:0060395 (SMAD protein signal transduction), GO:0061564 (axon development), GO:0050727 (regulation of inflammatory response), GO:0042476 (odontogenesis), and GO:0001894 (tissue homeostasis).

Results

AI Families and Novel FAM20A Mutations

Whole exome analyses for a cohort of AI patients identified 8 kindreds and 2 simplex cases carrying biallelic disease-causing FAM20A mutations, 7 of which have never been reported (Table 1). All affected individuals exhibited typical orodental phenotypes of AI type IG or enamel renal syndrome (ERS), including severe enamel hypoplasia (aplasia), impacted teeth, and pulp calcifications, although their expressivity (severity) varied significantly. While, in some cases, failed tooth eruption could not yet be determined due to the patient’s age, in others, tooth impaction only involved a few teeth, except for the proband of Family 5. Gingival overgrowth was evident in some cases, such as the Family 1 proband. Nephrocalcinosis or kidney problems was not noted in any affected individual, although thorough renal examination or sonography was not available for all cases. The clinical phenotypes and identified FAM20A mutations for each family are summarized in Table 1. Figure 2 shows an ERS family (Family 8) in which the proband’s (III:3) dental pulp was used for transcriptome analyses in the study. All other available dental records (photographs and radiographs) and DNA chromatograms for recruited individuals are presented in the Supplementary Figures file.

Table 1.

Phenotypes of affected individuals from 8 families and 2 sporadic cases with biallelic FAM20A mutations. Family pedigrees, clinical photographs, dental radiographs, and DNA sequencing chromatograms upon which phenotyping and genotyping are based are provided in Supplementary Figures. The symbol for each subject refers to that in the corresponding family pedigree. The teeth of delayed/failed eruption (impaction) are specified, when possible, with their tooth numbers. N.I., no available information; cbd, cannot be determined due to the young age of the subject. Novel FAM20A mutations are shown in bold.

Subject	Enamel	Intrapulp calcifications	Delayed/failed tooth eruption (Tooth #)	Gingival overgrowth	Nephro-calcinosis	FAM20A mutations
Family 1
V:2	aplastic	no	cbd	yes	N.I.	c.34_35del
V:3	aplastic	no	cbd	no	N.I.	c.34_35del
Family 2
III:2	hypoplastic	no	cbd	no	N.I.	c.771del c.1297A>G
Family 3
II:1	aplastic	no	cbd	no	N.I.	c.832_835delins(26) § c.1232G>A
Family 4
II:1	Aplastic	yes	cbd	N.I.	N.I.	c.915_918del c.1232G>A
Family 5
II:1	aplastic	yes	2, 3, 4, 5, 6, 11, 12, 15, 18, 21, 22, 27, 29, 31	yes	N.I.	c.406C>T c.406C>T
Family 6
III:1	aplastic	yes	no	yes	N.I.	c.625T>A
III:2	aplastic	yes	18, 31	yes	N.I.	c.1207G>A
Family 7
III:1	aplastic	no	cbd	yes	N.I.	c.129del c.1351del
Family 8
III:2	aplastic	no	6, 31	no	no	c.590-5T>A
III:3	aplastic	yes	6, 11	no	no	c.734_735del
Subject 9	N.I.	N.I.	N.I.	N.I.	N.I.	c.727C>T c.727C>T
Subject 10	N.I.	N.I.	N.I.	N.I.	N.I.	c.625T>A c.1351del

^§c.832_835delins(26), c.832_835delinsTGTCCGACGGTGTCCGACGGTGTCCA

The 7 novel FAM20A mutations include 4 single-nucleotide changes and 3 indels. Only individuals homozygous or compound heterozygous for these mutations exhibited AI, while family members carrying only wild-type or a single defective allele were unaffected. Four of these mutations are rare sequence variants listed in Genome Aggregation Database (gnomAD) (Karczewski Francioli et al. 2020), including c.590-5T>A (global minor allele frequency, MAF=0.000046), c.625T>A (0.000011), c.1232G>A (0.000131), and c.1351del (0.000007) (Supplementary Table 1). The c.590-5T>A mutation substitutes an A for a highly conserved T at the acceptor splice site of Intron 2 and is likely to alter normal splicing of FAM20A transcripts. A splicing assay using pSPL3 minigene clones containing FAM20A Exon 3 and partial flanking intron sequences was conducted. While the wild-type clone generated the expected 315-bp amplification product, the mutant minigene yielded a smaller amplicon that precisely lacked the 51 bps of Exon 3, indicating that the c.590-5T>A mutation causes recognition of the acceptor splice site to fail and leads to exon skipping (Figure 3a). Exclusion of Exon 3 results in an in-frame deletion and replaces 18 amino acids with a single valine at the middle of FAM20A protein (NP_006185.1:p.Asp197_Ile214delinsVal) (Figure 3a,b). The other 3 single nucleotide variants, c.625T>A, c.1232G>A, and c.1297A>G, all cause missense mutations that alter evolutionarily conserved amino acid residues, p.Cys209Ser, p.Arg411Gln, and p.Arg433Gly, and are predicted to be possibly damaging with PolyPhen-2 scores of 0.996, 1.000, and 1.000 respectively (Figure 3c). Among the three indel mutations, two were single-nucleotide deletions (c.771del and c.1351del) that would cause a −1 frameshift and premature translation termination (p.Q258Rfs*28 and p.Q451Sfs*4). While the mutant transcript resulting from the former variant would presumably undergo nonsense mediated decay (NMD), that of the later would probably not undergo NMD and would translate a truncated FAM20A protein lacking the C-terminal ~100 amino acids. The indel, c.832_835delinsTGTCCGACGGTGTCCGACGGTGTCCA, identified from Family 3, is located within Exon 6. This mutation, causing a net insertion of 22 exotic nucleotides, will probably cause a +1 frameshift or introduce a novel splice site, either of which would lead to a loss of FAM20A function.

Figure 3.

Analyses of FAM20A pathogenic mutations. (a) The schematic gene structure shows that the minigene contains human FAM20A Exon 3 and its flanking sequences cloned into the pSPL3 vector. The agarose gel image reveals an RT-PCR amplification product of 315 bp from wildtype (WT) minigene and a 264-bp amplicon from the mutant (Mut), which results from whole Exon 3 skipping. The mutant transcript would presumably generate FAM20A protein with an in-frame deletion, p.Asp197_Ile214delinsVal. (b) Alignment of the amino acid sequence (Phe191-Val220) of human FAM20A with its orthologs from mouse (Mus musculus), rat (Rattus norvegicus), chick (Gallus gallus), frog (Xenopus tropicalis), and fish (Danio rerio) along with the paralog, FAM20C. The 18 amino acids shown in bold are encoded by Exon 3 of FAM20A, which is lacking in FAM20C, and replaced by an exotic valine in the mutant protein from c.590-5T>A mutation. (c) Alignment of the amino acid sequences surrounding the substituted amino acids (in bold) of the three novel missense FAM20A mutations: p.Cys209Ser (top), p.Arg411Gln (middle), and p.Arg433Gly (bottom).

Transcriptome Analyses of ERS Dental Pulp Tissues

To investigate the molecular mechanism underlying ERS intrapulpal calcifications, we conducted transcriptome analysis for dental pulp tissues from the affected younger sister (III:3) of Family 8 using RNAseq. By comparing with control non-ERS dental pulp, expression of 1142 genes were identified to be significantly increased or decreased with PPEE <0.05 in ERS pulp tissues (Supplementary Table 2). Of these differentially expressed genes (DEGs), 662 showed upregulated expression and 480 were downregulated in the ERS pulp. Among the over-expressed genes, many were involved in biomineralization, such as MEPE (FC = 358.2) and BGLAP (FC = 7.6). Particularly, genes known to be expressed by odontoblasts during dentine formation appeared to be significantly upregulated, including WNT10A (FC = 4.9), MMP20 (FC = 4.4), MMP9 (FC = 4.0) and DSPP (FC = 3.6). In contrast, many genes involved in inflammation and axon development, were among the most down-regulated genes. Noticeably, FAM20A was reduced by an FC of 0.4 relative to the control, suggesting that ~60% of its transcripts were likely degraded by NMD due to the truncation mutation, p.Glu245Glyfs*11, shown in Family 8.

Gene ontology (GO) enrichment analysis based upon DEGs involved in annotated biological processes (BPs) was conducted. A total of 295 GO:BP terms showed a significant over or under representation with a p-value < 0.05 (Supplementary Table 3). For each term, a z-score was calculated to assess if the BP was likely to be increased (positive value) or decreased (negative value). The first 100 terms with the highest significance were presented in Figure 4 (Supplementary Table 4). While BPs involved in biomineralization (GO:0031214, GO:0001501, GO:0070167, GO:0030282), sodium ion transport (GO:0006814), and extracellular matrix organization (GO:0043062, GO:0030198) had the highest z-scores, the lowest ones included terms related to inflammation (GO:0002526, GO:0050727) and axon development (GO:0061564, GO:0008366, GO:0007409). Four co-occurring terms associated with tooth formation (GO:0042476, GO:0042481, GO:0042487, GO:0042475) all had a z-score higher than 1.00. Furthermore, whereas osteoblastogenesis-related BPs (GO:0033687, GO:0001649) showed positive z-scores, those involved in osteoclastogenesis (GO:0030316, GO:0045670) had negative ones, reaffirming a transcriptional profile towards biomineralization in ERS pulp tissues (Figure 3). Noticeably, 5 terms related to BMP (GO:0071772, GO:0071773, GO:0030509) and SMAD signaling pathways (GO:0060395, GO:0060393) appeared to be over-represented, demonstrating a plausible molecular mechanism underlying the increased biomineralization.

Figure 4.

Barplot for the GO terms of biological process with a calculated z-score. A hundred GO terms of biological process with the greatest significance of over-representation or under-representation in ERS pulp tissues compared to the control are plotted. The y-axis shows the adjusted p-value (significance) of the terms on a negative logarithmic scale. The bars are arranged in order of their calculated z-scores (Supplementary Table 4). A red and blue color scale for the bars is used to represent a positive or negative value of a z-score.

To investigate the hypothesis that intrapulpal calcification in ERS is caused by upregulated BMP signaling, DEGs assigned to GO:0030509 (BMP signaling pathway) were analyzed. Seven genes of TGF-β/BMP family, including GDF7, GDF15, BMP3, BMP8A, BMP8B, BMP4, and BMP6, were upregulated, while BMP5 appeared to be downregulated (Figure 5). On the other hand, 3 genes that encode known secreted BMP antagonists, GREM1, BMPER, and VWC2, showed significantly reduced expression with FCs of 0.0003, 0.37, and 0.46 respectively (Walsh Godson et al. 2010). This expression profile of BMP-related genes confirmed upregulated BMP signaling in dental pulp tissues when FAM20A is mutated. The relationships between genes with significant FCs and ten BPs of particular interest were evaluated with hierarchical clustering. Specific clusters were unbiasedly obtained for gene expression patterns (FCs) and functional categories (BP terms) and plotted (Figure 6). The results agreed with the above analyses.

Figure 5.

Comparative bar graph for differentially expressed genes associated with BMP signaling pathway. The 18 genes, under the term of GO:0030509, of which the expression is significantly increased or decreased in ERS pulp tissues compared to the control are presented. The y-axis shows the TPM (transcripts per million) of each gene transcript on a logarithmic scale with base 2. The table on the right indicates the fold change of each differentially expressed gene. While eleven BMP signaling-related genes are overexpressed in ERS pulp, seven are less expressed.

Figure 6.

Dendrograms for hierarchical cluster analyses of gene expression. Left: The fold changes (FC) of genes are used for grouping based on the pattern of gene expression. The inner ring next to the dendrogram, on a red-blue color scale, shows the logFC of the genes. The outer ring represents the 10 selected GO terms of biological process that are assigned to the genes and color-coded. Right: The assigned functional categories (GO terms) are used for gene clustering.

Discussion

In this study, 7 novel FAM20A pathogenic variants were reported, which broadened the genotypic spectrum of AI1G or enamel renal syndrome (ERS). These mutations presumably cause a loss of function and significantly gain in vivo insights into the structural functionality of FAM20A protein. The crystal structure of human FAM20A has previously been determined and shown to have a distinct insertion of 17 amino acid residues, encoded by Exon 3, that is not found in FAM20B or FAM20C, two paralogs of FAM20A in vertebrates (Cui Zhu et al. 2017). Deletion of this unique inserted region significantly abolished FAM20A’s ability to allosterically increase kinase activity of FAM20C, which is critical for phosphorylating secreted proteins involved in biomineralization. Consistent with the molecular consequence of a previously reported splice site mutation, c.590-2A>G (Cho Seymen et al. 2012), our minigene assay showed that the c.590-5T>A mutation, identified in Family 8, caused Exon 3 skipping and deletion of the 17 residues encoded within FAM20A. These 2 AI-causing FAM20A variants provide in vivo evidence demonstrating the functional significance of this unique region. It has also been shown that FAM20A exhibits a distinct pattern of disulfide bridges mediated by Cys209 and Cys211 within the inserted region, which are critical for its function (Cui Zhu et al. 2017). The p.Cys209Ser mutation found in Family 6 and Subject 10 further supports the importance of this unique disulfide bond pattern of FAM20A. Moreover, two other missense mutations demonstrated in this study, p.Arg411Gln and p.Arg433Gly, both replace evolutionarily-conserved arginines. The corresponding residues in FAM20C, the Golgi casein kinase (GCK), have been proposed to play potentially significant roles in structural stability. Based upon the crystal structure of the FAM20C ortholog of C. elegans (ceFAM20), Arg367^ceFAM20 (corresponding to Arg411^FAM20A) and Arg390^ceFAM20 (Arg433^FAM20A) interact with Asp359^ceFAM20 and Asp366^ceFAM20 respectively, which surround and stabilize the catalytic segment of ceFAM20 (Xiao Tagliabracci et al. 2013). The p.Arg411Gln and p.Arg433Gly mutations found in our families provide in vivo evidence to support the structural significance of these two residues and the Arg-Asp interactions.

Intrapulpal calcification is a consistent feature found in ERS patients, although the severity varies among individuals (de la Dure-Molla Quentric et al. 2014). It involves not only erupted teeth but those that fail to erupt, suggesting that it results from a pathological alteration of tissue homeostasis rather than a physiological pulpal response to chronic stimuli, such as dental attrition (Wang Aref et al. 2013; Lignon Beres et al. 2017). Our transcriptome analyses indicated that the intrapulpal calcification is likely caused by an increased process of dentinogenesis, rather than osteogenesis, as several odontoblast-expressed genes are significantly upregulated, including DSPP and MMP20 (Caterina Shi et al. 2000; Yamashiro Zheng et al. 2007). Previous studies of histological analysis have demonstrated that the ERS pulp is obliterated by amorphous calcified tissues composed of globular dentine and irregular dentinal tubules, which corresponds to the current findings (Martelli-Junior Bonan et al. 2008; Berès Lignon et al. 2018). However, the dentine of ERS teeth appeared to be normal and never fused with the pulp calcifications, suggesting that the increased expression of dentinogenesis genes we observed occurs de novo in pulp cells rather than in the existing odontoblasts lining the circumpulpal dentine. Furthermore, this ectopic dentine formation might result from an upregulation of canonical BMP signaling pathway, as our GO analyses found BMP/SMAD-related terms were significantly over-represented in ERS pulp tissues. BMP signaling has been known to play critical roles in dentinogenesis (Liu Goldman et al. 2022). Depletion or mutation of BMP signaling-associated genes in mouse models caused dentine defects (Omi Kulkarni et al. 2020). These transcriptome analyses demonstrated significant overexpression of several genes that are known to be involved in dentine development, including BMP3, BMP4, and BMP6. Bmp3 and Bmp4 have been shown to be expressed by pre-odontoblastic cells and early, but not late odontoblasts in developing mouse teeth, suggesting that their roles are in odontoblastic differentiation (Yamashiro Tummers et al. 2003). Whereas Bmp6 expression was barely detected during murine dentinogenesis (Aberg Wozney et al. 1997), differentiating and early odontoblasts of human tooth germs were shown to be BMP6-positive (Heikinheimo 1994). Therefore, increased expression of these 3 genes supports the hypothesis of de novo dentine formation underlying intrapulpal calcification in ERS pulp tissues. Moreover, these analyses also unravel many other BMP-associated signaling molecules that presumably contributed to this pathological process of mineralization, including ligands of TGF-β/BMP family, BMP8A, BMP8B, GDF7, and GDF15, as well as BMP antagonists, GREM1, BMPER, and VWC2 (Walsh Godson et al. 2010). The altered expression of these molecules in ERS pulp tissues suggested that they serve potentially unappreciated functions during normal dentinogenesis, which warrants further investigation.

It has been demonstrated that Fam20a is expressed by odontoblasts and sporadic cells within the pulp of developing mouse molars (Wang Reid et al. 2014). However, while secretory odontoblasts show strong Fam20a expression, no apparent dentine defects are shown in patients with FAM20A mutations (Wang Aref et al. 2013; de la Dure-Molla Quentric et al. 2014). Correspondingly, Fam20a null mice do not exhibit overt dentine malformations, further demonstrating its dispensable role during dentinogenesis (Li Saiyin et al. 2019). In contrast, the intrapulpal calcifications found in ERS individuals as well as in older Fam20a null mice suggest a critical role for FAM20A in homeostasis of dental pulp (Vogel Hansen et al. 2012). As FAM20A facilitates FAM20C to phosphorylate secretory pathway proteins within the Golgi apparatus, it is reasonable to hypothesize that certain substrate proteins of FAM20A-FAM20C kinase complex are essential for preventing ectopic mineralization of dental pulp. Matrix Gla protein (MGP) is a 14-kDa secreted protein that has been shown to be a potent mineralization inhibitor (Cancela Laizé et al. 2014). Loss-of-function mutations of human MGP cause Keutel syndrome (OMIM #245150) characterized by abnormal calcification of cartilage (Munroe Olgunturk et al. 1999). Mgp deficient mice exhibit pathological vascular and chondral calcifications (Luo Ducy et al. 1997). However, this function of mineralization inhibition depends on proper post-translational modifications of MGP, including γ-carboxylation of glutamates by a vitamin K-dependent carboxylase and serine phosphorylation by FAM20A-FAM20C kinase complex (Schurgers Spronk et al. 2007). Furthermore, a single-cell transcriptome study of human dental pulp has recently demonstrated that dental pulp stem cells (DPSCs) specifically express not only stem cell markers but also MGP (Pagella de Vargas Roditi et al. 2021). Therefore, it is possible that MGP secreted by DPSCs is essential to maintain their stem cell niche and cell stemness. In ERS pulp, defective FAM20A-FAM20C kinase fails to produce functional MGP and therefore induces odontoblastic differentiation of DPSCs and ectopic dentine formation via upregulation of BMP signaling. Further investigations are warranted to test this hypothesis.

Conclusions

This study significantly expanded the genotypic spectrum of ERS and provided in vivo evidence that the unique inserted region and disulfide bond pattern of FAM20A are critical for its function. Intrapulpal calcification of ERS is likely to result from a process of de novo dentinogenesis and an increased activity of canonical BMP signaling. FAM20A plays an essential role in pulp tissue homeostasis and preventing pathological mineralization of soft tissues, which probably depends upon functional MGP protein that requires proper phosphorylation by FAM20A-FAM20C kinase complex.

Data Availability Statement

Whole exome sequencing data and analysis are available at dbGaP under Genetics of Disorders Affecting Tooth Structure, Number, Morphology and Eruption. dbGaP Study Accession: phs001491.v1.p1. Additional data used to support the findings of this study are available from the corresponding author upon reasonable request.