Transcriptome analysis of gingival tissues of Enamel Renal Syndrome

Abstract

Background and objective:

Biallelic loss-of-function mutations of human FAM20A have been known to cause enamel-renal syndrome (ERS), featured by agenesis of dental enamel, nephrocalcinosis, and other orodental abnormalities, including gingival hyperplasia. However, while the histopathology of this gingival anomaly has been analyzed, its underlying molecular mechanism remains largely unknown. This study aimed to unravel the pathogenesis of gingival hyperplasia in ERS.

Methods:

Whole exome sequencing was conducted for an ERS case. Transcriptome analyses, using RNA sequencing, of the patient’s gingiva were performed to unravel dysregulated molecules and aberrant biological processes underlying the gingival pathology of ERS, which was further confirmed by histology and immunohistochemistry.

Results:

Two novel frameshift FAM20A mutations in Exon 1 (g.5417delG; c.129delG; p.Cys44Alafs*101) and Exon 5 (g.62248_62249delAG; c.734_735delAG; p.Glu245Glyfs*11) were identified. Transcriptional profiling of patient’s gingival tissue revealed a total of 1683 genes whose expression had increased (1129 genes) or decreased (554 genes) at least 2-fold compared to control gingival tissues. There were 951 Gene Ontology (GO) terms of biological process being significantly over-represented or under-represented. While GOs involved in extracellular matrix organization, angiogenesis, biomineralization, and epithelial cell proliferation appeared to be activated in ERS gingiva, genes related to keratinocyte differentiation, epithelial development, and keratinization were of decreased expression. FAM20A immunohistochemistry revealed a strong reactivity at the suprabasal layers of epithelium in control gingiva but showed a significantly diminished and scattered signal in ERS tissues. For genes showing significant over-expression in the transcriptome analyses, namely ALPL, SPARC, and ACTA2, an increased immunoreactivity was observed.

Conclusion:

Our results unraveled a potential role for FAM20A in homeostasis of both gingival epithelium and connective tissues.

INTRODUCTION

Amelogenesis imperfecta type IG (AI1G; OMIM #204690), better known as enamel-renal syndrome (ERS), is an autosomal recessive disorder characterized by severe enamel hypoplasia/aplasia (thin or no dental enamel), delayed or failed tooth eruption, misshapen teeth with intrapulpal calcification, and gingival overgrowth (1–4). Despite typically normal blood chemistry, nephrocalcinosis is usually discovered by a routine renal ultrasound without subjective complaints but may not appear until later in life. Among the oral findings, gingival overgrowth is the one that only involves soft tissues and is consistently found in ERS patients, although its severity varies significantly among reported cases (2, 5, 6). Histologically, it has been demonstrated that gingival tissues from the patients have well-structured epithelium with elongated rete ridges and dense connective tissues containing an increased amount of collagen fiber bundles running in all directions, myofibroblasts, islands of odontogenic epithelium, and calcified psammomatous deposits (ectopic calcification), indicating disturbed epithelial and connective tissue homeostasis (5, 6). However, the molecular mechanism underlying these abnormalities are largely unknown.

Biallelic loss-of-function mutations of FAM20A (family with sequence similarity 20, member A; OMIM *611062) were first identified to be associated with amelogenesis imperfecta and gingival fibromatosis syndrome (AIGFS; OMIM #614253) (7) and subsequently found to cause ERS, which differ from AIGFS only by the presence of nephrocalcinosis (3, 4). On the other hand, mutations in FAM20C, the closest paralog to FAM20A, have been known to cause Raine syndrome (RNS; OMIM #259775) characterized by deadly osteosclerotic bone dysplasia and intracranial calcifications (8). Oral manifestations of RNS have been reported to be similar to those of ERS, including enamel hypoplasia and gingival overgrowth, in some patients with hypomorphic FAM20C mutations and a non-lethal form of RNS (9). Molecular and structural investigations further demonstrated that FAM20C is the long-sought Golgi casein kinase that phosphorylates the vast majority of the secreted phosphoproteome (10, 11), and FAM20A is a pseudokinase that allosterically activates FAM20C to phosphorylate secreted proteins important for biomineralization (12, 13). While these studies might unravel the molecular mechanisms underlying enamel defects of ERS, the pathogenesis of gingival overgrowth and the role of FAM20A in gingival homeostasis remain to be elucidated.

In this study, we characterized a case of ERS and reported two novel FAM20A disease-causing mutations. By performing RNA sequencing (RNAseq) of gingival tissues from the patient, we characterized the transcriptome of ERS gingiva and identified aberrant gene expressions and biological processes that might underlie the gingival overgrowth and ectopic calcification in ERS. Histological analyses further validated these findings and suggested potential distinct roles of epithelium and connective tissues in altered gingival homeostasis of ERS.

MATERIALS AND METHODS

Subject recruitment and mutational analyses

The human study protocol and subject consents were reviewed and approved by the Institutional Review Boards at the National Taiwan University Hospital. Following informed consent, the study participants were given an oral exam with radiographs and oral photographs. Saliva samples were collected using the Saliva DNA Collection and Preservation kit (Norgen Biotek Corp, Thorold, ON, CA), and DNA was extracted following the manufacturer’s instructions.

Genomic DNA from the proband was characterized by whole-exome sequencing (Genomics, New Taipei City, Taiwan). Briefly, the genomic DNA was captured with SureSelect Human All Exon V5 Kit (Agilent Technologies, Santa Clara, CA USA) and sequenced with Illumina HiSeq X Ten for 100 base paired-end reads. Reads were aligned to human reference genome GRCh37 (hg19) using BWA. Single nucleotide variants and short insertions and deletions (indels) were called using GATK HaplotypeCaller. The called variants were annotated using Ensembl VEP (14). The disease-causing FAM20A sequence variations were confirmed in the proband and his father by Sanger sequencing (4).

RNA sequencing and analyses

Gingival tissues were harvested from the proband during surgical removal of left maxillary molars and an age-comparable healthy male patient undergoing a periodontal procedure of distal wedge over tooth number 15. Tissues were immediately immerged in 5–10 volumes of RNAlater® solution (Thermo Fisher Scientific, Waltham, MA USA) and incubated overnight at 4°C, and then stored at −20°C until needed for RNA extraction. Total RNA was extracted from samples and checked for integrity using an Agilent 2100 Bioanalyzer with RNA 6000 Nano Kit. A cDNA library was created by random priming from fragmented mRNA for synthesis of the first strand cDNA followed by PCR amplification. The library was sequenced on an Illumina NextSeq 500 Sequencer System using 75 bp single-end sequencing. The processed reads were mapped to hg19 human reference transcriptome using bowtie2 (v2.2.6) and selected for following quantification done by RSEM (RNA-Seq by Expectation Maximization) (15). EBSeq (v1.16.0) was used to identify differentially expressed genes between ERS and control gingival tissues (16). Further gene ontology (GO) and enrichment analyses were conducted using clusterProfiler (v3.8.1) (17). GOplot was implemented to plot the results and to perform hierarchical clustering for gene expression patterns (fold changes) or functional categories (GO terms) (18).

Histological analyses and immunohistochemistry

Gingival tissues harvested from the proband and the control patient were formalin fixed, paraffin embedded, sectioned, and H&E stained for regular histological analyses. For immunohistochemistry, the staining was performed using a VENTANA BenchMark XT automated IHC/ISH staining instrument (Ventana Medical Systems, Oro Valley, AZ USA). Briefly, 4 μm-thick sections were deparaffinized, rehydrated, and incubated with 3% hydrogen peroxide solution for 5 min. After a washing procedure with the supplied buffer, sections were repaired for 40 min with EDTA, followed by incubation with the primary antibody for 60 minutes at 37°C and then overnight at 4°C. After three rinses in buffer, the slides were incubated with the secondary antibody. The final staining was visualized with a DAB substrate chromogen solution from OptiView DAB IHC Detection Kit (Ventana Medical Systems). Afterwards, slides were counterstained with hematoxylin, dehydrated, and mounted. The primary antibodies used in the study included: anti-FAM20A (1:100 ; A8496 ; ABclonal Technology, Woburn, MA USA), anti-ALPL (1:100 ; 11187–1-AP ; Proteintech, Rosemont, IL USA), anti-SPARC (1:100 ; 15274–1-AP ; Proteintech), anti-ACTA2 (1A4 ; Cell Marque, Rocklin, CA USA). The secondary antibodies were from Histofine® Simple Stain MAX PO (R) (Nichirei Bioscience, Tokyo, Japan).

RESULTS

Novel FAM20A mutations causing Enamel Renal Syndrome

The proband was a 41-year-old male from Taiwan, presumably of East Asian ethnicity. He came to our clinic for treatment of a progressive pain over upper left back teeth. He has been taking gout medication but is otherwise healthy. According to him, his adult teeth came out pretty late and appeared small and widely spaced. His gums have looked large and bumpy. No other members from his family seemed to have teeth that looked like his (Fig. 1A). He was diagnosed with generalized amelogenesis imperfecta by his family dentist and has undergone multiple prosthodontic treatments. Clinically, most teeth were covered with fixed prostheses, except tooth numbers 2, 14, 15, 29, 31 (Fig. 1B). The right maxillary first molar was a dental implant. The uncovered teeth showed generally thin dental enamel with smooth tooth surfaces. Both left maxillary molars exhibited deep periodontal pockets and moderate tooth mobility. Gingival tissues appeared slightly thick and firm over the facial surfaces of the alveolar ridges. Radiographically, multiple impacted teeth were noticed, including tooth numbers 1, 16, 17, 18, 22, 26, 27, 32 (Fig. 1C). These impacted teeth showed a complete lack of dental enamel and an evident sign of pre-eruptive crown resorption without apparent pericoronal radiolucencies. Pulp chambers appeared generally small with intrapulpal calcifications. Tooth numbers 14 and 15 exhibited severe periodontal bone destruction, which might involve impacted tooth number 16. Renal ultrasonographs taken during proband’s routine medical examination revealed medullary nephrocalcinosis of bilateral kidneys (Fig. 1D). Based upon these findings, a clinical diagnosis of enamel-renal syndrome was made.

Figure 1. ERS Family with novel FAM20A mutations.

(A) Pedigree. Dots mark the two persons who donated samples for DNA sequencing. (B) Oral photographs of the proband (III:1). Most teeth were covered with fixed prostheses. The uncovered teeth (tooth numbers 2, 14, 15, 29, 31) show generally thin dental enamel with smooth tooth surface. The attached gingiva is enlarged and bumpy. (C) Panoramic radiograph of the proband. Many impacted teeth (tooth numbers 1, 16, 17, 18, 22, 26, 27, 32) show a complete lack of dental enamel and intrapulpal calcifications. Roots of the teeth are generally short. (D) Kidney ultrasounds of the proband. Hyperechoic signals are evident and suggestive of medullary nephrocalcinosis. (E) DNA sequencing chromatograms of FAM20A mutations. Left: Sequence from the border of Exon 1 and Intron 1 revealing heterozygosity for a one-nucleotide deletion (g.5417delG; c.129delG) that occurs in the father (II:5) and proband. Right: Exon 5 sequence revealing heterozygosity for a two-nucleotide deletion (g.62248_62249delAG; c.734_735delAG) and a synonymous SNP (g.62249G>A; rs2286562) that occur in the proband. The mutation designations are with respect to the FAM20A genomic reference sequence NG_029809.1 and cDNA reference sequence NM_017565.3 (for mRNA transcript variant 1).

To confirm the clinical diagnosis, exome analysis of proband’s DNA was performed. Two deletion mutations of FAM20A were identified, g.5417delG (c.129delG; p.Cys44Alafs*101) and g.62248_62249delAG (c.734_735delAG; p.Glu245Glyfs*11), and subsequently validated with Sanger sequencing (Fig. 1E). Proband’s father was a heterozygous carrier of p.Cys44Alafs*101 mutation. However, we were unable to recruit proband’s mother, who should carry the p.Glu245Glyfs*11 mutation. These two mutations, located at Exon 1 and 5 respectively, would cause a frameshift followed by a premature stop codon. The mutant transcripts would presumably undergo nonsense-mediated decay and produce no FAM20A protein. Neither of the two mutations were previously reported in cases of ERS (Supplementary Table 1).

Transcriptome analyses of ERS gingival tissues

To investigate the underlying mechanism of gingival hyperplasia in ERS, we harvested gingival tissues from the proband during surgical removal of left maxillary molars. Transcriptional profiling of the tissues was performed using RNAseq and compared with that of control non-ERS gingiva. A total of 1683 genes were identified with PPEE ≦ 0.05 and a minimum fold change (FC) ≧ 2 as being significantly increased or decreased in ERS relative to control gingivae (Supplementary Table 2). Of these, 1129 showed higher expression and 554 lower expression in the ERS gingiva. Among highly over-expressed genes, there were many involved in extracellular matrix biosynthesis, such as COL1A1 (FC = 13.8) and COL1A2 (FC = 11.0). Genes related to biomineralization also appeared to be highly upregulated, such as LTF (FC = 1084, the highest in the rank), SPARC (FC = 6.2), and ALPL (FC = 4.7). On the other hand, many genes involved in keratinocyte differentiation and keratinization, such as those of LCE (late cornified envelope) gene family, were among the most down-regulated genes (Supplementary Table 2).

Further gene ontology (GO) and enrichment analysis revealed that 951 GO terms of biological process (bp) were significantly over-represented or under-represented with p-values ≦ 0.05 (Supplementary Table 3). To get an impression if a biological process is more likely to be decreased or increased, a z-score was further calculated for each bp GO terms, and the first 100 most significant terms were plotted (Fig. 2) (Supplementary Table 4). Z-score is a simplified value to assess if the biological process is more likely to be decreased (negative value) or increased (positive value) (18). While biological processes involved in extracellular matrix organization (GO:0030198, GO:0043062), angiogenesis (GO:0048514, GO:0001525), and biomineralization (GO:0001501) had the highest z-scores, those related to epithelial development and keratinization (GO:0030216, GO:0031424, GO:0009913, GO:0008544, GO:0043588) had the lowest. Noticeably, while keratinocyte differentiation (GO:0030216 ; z-score = −5.06) is under-represented, epithelial cell proliferation (GO:0050673 ; z-score = 3.78) is over-represented, suggesting a role of FAM20A in regulating epithelial proliferation and differentiation (Supplementary Table 4).

Figure 2. Barplot of the GO terms of biological process with a significant change.

Only the first 100 most significant over-represented or under-represented GO terms of biological process are plotted. The y-axis shows the significance of the terms on a logarithmic scale, and the bars are ordered according to their z-score (Supplementary Table 4). The red color indicates a positive z-score, and the blue color a negative one.

To further understand the relationships between genes with large fold changes and biological processes of significant enrichment, 10 biological processes of particular interest were selected and plotted with the highly up-regulated and down-regulated genes (Fig. 3). Many of these genes were related and functioning in similar biological processes. Noticeably, genes involved in MAPK, ERK1, and ERK2 cascades, such as IL6, CCL4, XCL2, CCL8, and NTRK1, were mainly over-expressed in ERS gingivae. In contrast, expressions of genes that function in proteolysis and keratinocyte differentiation were significantly down-regulated. These relationships were further illustrated by performing hierarchical clustering for gene expression patterns (fold changes) or functional categories (GO terms) to unbiasedly obtain clusters that were likely to contain sets of co-regulated or functionally related genes (Fig. 4). The results were consistent with other analyses.

Figure 3. Circos plot of the relationship between genes with a large fold change (FC) and the biological process (bp) GO terms with significant enrichment.

While several highly-upregulated genes are related to biomineral tissue development (GO:0031214; firebrick), many down-regulated ones are involved in keratinocyte differentiation (GO:0030216; bisque). Genes within ERK1 and ERK2 cascade (GO:0070371; green-yellow) appear to be over-expressed in ERS gingival tissues.

Figure 4. Dendrograms of hierarchical clustering for gene expression analyses.

Left: Genes are grouped together based on their expression patterns (FC; fold changes). The first ring next to the dendrogram represents the logFC of the genes, whose values are color- coded. The next ring represents the terms assigned to the genes from the 10 selected GO terms of biological process, which are also color-coded. Right: Genes are grouped based on the functional categories (GO terms) from a functional analysis of the differentially expressed genes performed with clusterProfiler (v3.8.1).

Histological analyses of ERS gingival tissues

We further performed histological analyses on proband’s gingiva to study the histopathology of ERS. The tissue exhibited a normal architecture of oral mucosa with a well-structured parakeratinized epithelium and a cell-dense lamina propria (Fig. 5A). The epithelium was mildly acanthotic and had thick broad-based rete ridges and a thin parakeratinization layer, compared to that of normal gingiva. The connective tissue showed dense collagen fibers running in various directions with fewer papillae inserted into epithelium. Numerous basophilic and laminated calcified structures, resembling psammoma bodies, were observed at both superficial and deep areas of the connective tissue (Fig. 5A, 5B). Immunohistochemical analyses against FAM20A on normal gingiva revealed a strong reactivity at the suprabasal layers of epithelium except the parakeratinized surface layer (Fig. 5D). The positive staining was of a dot-like pattern and localized within the keratinocytes. The nuclei, cell membranes, and extracellular space were negative for the signal. Noticeably, in the connective tissue, the endothelial cells of blood vessels and some fibroblasts were also moderately stained. In contrast, the FAM20A immunoreactivity on proband’s gingiva was significantly diminished with only scattered weak staining in the epithelium, which confirmed the depletion of the protein due to FAM20A null mutations (Fig. 5C).

Figure 5. ERS gingival histology and FAM20A immunohistochemistry.

(A, B) H&E staining of proband’s gingival tissue (100X). A: The epithelium shows mild acanthosis and thick broad-based rete ridges. Dense collagen fibers running in different direction are seen in the lamina propria with some basophilic calcifications (arrowhead). B: Clusters of psammomatous calcifications (arrowheads) are observed at different areas of gingival connective tissue with no apparent epithelial nests around. (C, D) FAM20A localization in gingival tissues from the proband and a control (100X). C: In proband’s gingiva, FAM20A immunoreactivity is minimal except some scattered staining in the epithelium. D: In control gingiva, strong FAM20A signals are observed throughout the whole layer of epithelium, except the basal cell layer and the parakeratinized surface layer. Moderate signal is also detected in endothelium of blood vessels and some connective tissue fibroblasts.

To confirm the enriched biological processes of biomineralization and angiogenesis in ERS gingiva revealed by the transcriptome data, immunostainings of ALPL (Alkaline Phosphatase), SPARC (Secreted Protein, Acidic and Cysteine Rich), and α-SMA (Alpha Smooth Muscle Actin ; ACTA2) were conducted. While ALPL weakly stained in the connective tissue of control gingiva, proband’s tissues showed a moderate to strong ALPL immunoreactivity in fibroblasts and endothelial cells of blood vessels (Fig. 6A, 6B). Particularly, areas surrounding the psammomatous calcifications were strongly stained (Supplementary Fig. 1A). For SPARC, a matrix-associated protein, the positive staining was detected all over the connective tissue of proband’s gingiva with the strongest signal around the psammomatous calcifications (Fig. 6C) (Supplementary Fig. 1B). On the contrary, the control gingiva showed a relatively weak reaction in the connective tissue except some moderate staining at blood vessels in the papillae (Fig. 6D), which was not observed in proband’s tissue. These results of increased ALPL and SPARC expression in the connective tissue were consistent with the findings from transcriptome analyses and indicative of an activation of biomineralization-associated genes in fibroblasts when FAM20A was depleted. On the other hand, α-SMA immunohistochemistry revealed an increased number of blood vessels (α-SMA-positive endothelial cells) and myofibroblasts (α-SMA-positive spindle cells) in the connective tissue of proband’s gingiva compared to the control, which showed barely detectable signals outside of blood vessels (Fig. 6E, 6F).

Figure 6. ALPL, SPARC, and α-SMA immunohistochemistry.

(A, B) ALPL immunostaining in gingival tissues from the proband and a control (100X). A: In proband’s gingiva, immunoreactivity is strongly detected in the connective tissue fibroblasts. B: In control gingiva, the reactivity is scattered and weak. (C, D) SPARC immunostaining in gingival tissues from the proband and a control (100X). C: In proband’s gingiva, signals are strong in the connective tissue fibroblasts but weak in blood vessel endothelium. No signals show in the epithelium. D: In control gingiva, signals are generally weak, except some strong staining in endothelial cells of blood vessels at connective tissue papillae (arrowhead). (E, F) α-SMA immunostaining in gingival tissues from the proband and a control (200X). E: In proband’s gingiva, reaction is strong in both the connective tissue (myofibroblasts) and blood vessel endothelium. F: In control gingiva, only smooth muscle cells on vessel walls and endothelial cells show strong signals.

DISCUSSION

In this study, we presented an ERS individual and identified two novel FAM20A frameshift mutations, p.Cys44Alafs*101 and p.Glu245Glyfs*11. Although these two mutations have never been reported in ERS kindreds, they are documented in databases of human variants (19). In gnomAD (Genome Aggregation Database), the p.Glu245Glyfs*11 (c.734_735delAG) mutation was identified in 7 out of 18,870 chromosomes, giving a MAF (minor allele frequency) of 0.000371, in East Asian population (EAS), while p.Cys44AlafsTer101 (c.129delG) was found in one chromosome from EAS in the whole database. Given that ERS is a recessive disorder, it is expected to identify heterozygous carriers in general population. Of the 41 disease-causing FAM20A mutations reported to date, 14 are documented in either gnomAD or ExAC (Exome Aggregation Consortium) database (Supplementary Table 1). These mutations (variants) are considerably rare and mostly identified only in a specific population. For example, the p.Leu117Cysfs*22 (c.349_367del) mutation is found in 3 out of 13,260 chromosomes in EAS with a MAF of 0.0002262 and has been identified in two Thai families of ERS, suggesting inheritance from a common ancestor (5, 20). Noticeably, a two-nucleotide deletion in Exon 1, c.34_35del (p.Leu2Alafs*67), has been reported in 5 families of ERS, but not documented in gnomAD or ExAC database, which seems to be a mutation hotspot (3, 21–24). However, those cases were all homozygotes for the mutation and mostly from Mediterranean region with probably similar genetic background, indicating a potential founder effect. Therefore, identity by descent (IBD) seems to underlie the genetic etiology of many ERS cases reported all over the world.

Gingival hyperplasia has been considered as one of the characteristic features of ERS, although the severity of individual cases can vary significantly (2). Based upon our transcriptome and histological analyses of ERS gingiva, this pathology seems attributable to aberrations of both epithelium and underlying connective tissues. In ERS epithelium, while cell proliferation is activated (positive z-score), keratinocyte differentiation and keratinization are significantly suppressed (negative z-score), suggesting a regulatory role of FAM20A in balance between epithelial proliferation and differentiation. A significant decrease in parakeratinization and mild acanthosis of proband’s gingiva on histology are consistent with the transcriptome result. Recently, Li et al. reported that loss of epithelial FAM20A in a conditional knockout mouse model (K14-Cre;Fam20a^flox/flox mice) caused a significant thickening of the gingival epithelium, which further supports our findings and the role of FAM20A in epithelial homeostasis (25). On the other hand, aberrations in the connective tissue also contribute to gingival hyperplasia of ERS. Many biological processes are activated, including extracellular matrix (ECM) organization, angiogenesis, and biomineralization (high positive z-scores), which are respectively corresponding to the dense collagen fibers, abundant blood vessels, and psammomatous calcifications on histopathology of ERS gingiva. While many genes involved in collagen metabolic processes and fibril organization are over-expressed in the connective tissue, expression of genes regulating proteolysis are significantly affected, demonstrating a modulating role for FAM20A in ECM homeostasis. Showing FAM20A expression in endothelial cells of blood vessels in normal gingiva and increased blood vessel formation in ERS gingiva, we revealed potential unappreciated functions of FAM20A in angiogenesis. However, although α-SMA, an endothelial marker, is over-expressed in the gingiva when FAM20A is depleted, the increased expression seems to result from an increased number of myofibroblasts which also express α-SMA in addition to endothelial cells of blood vessels. The functions of FAM20A in angiogenesis and myofibroblasts differentiation require further investigations. Ectopic gingival calcifications have been consistently observed in ERS patients (2, 5, 6). It has also been suggested that odontogenic epithelial cells were responsible for formation of these calcifications due to their close proximity to the nests of epithelial cells (6). However, there are few, if any, epithelial nests observed around the psammoma bodies in our case. Also, genes involved in biomineralization, such as ALPL and SPARC, are over-expressed in the connective tissue surrounding the calcified structures, suggesting that a pathological calcification process is activated in connective tissue fibroblasts when FAM20A is depleted. The lack of gingival calcifications in Fam20a epithelial conditional knockout mice further supports this hypothesis (25). Moreover, this aberrant mineralization is probably not due to a disturbance in systemic calcium homeostasis, since the blood chemistry is usually normal in ERS patients (2), including our proband. Therefore, localized activation of biomineralization-related genes in connective tissue fibroblasts, when FAM20A is depleted, seems to underlie gingival ectopic calcification in ERS. This mechanism might also apply to parenchymal calcification in kidneys (nephrocalcinosis) of ERS patients.

Recently, FAM20A was shown to be a pseudokinase that forms a functional complex with Golgi casein kinase (G-CK) Fam20C and enhances extracellular protein phosphorylation within the secretory pathway (12, 13). However, how loss of this function leads to all the pathologies in ERS remains puzzling. An alternative explanation is that the ERS phenotypes might result from loss of unidentified G-CK-independent functions of FAM20A. However, this is unlikely, since FAM20C loss-of-function mutations cause similar and overlapping phenotypes of ERS, including gingival hyperplasia (9, 26, 27). As discussed above, depletion of FAM20A might promote proliferation, while inhibiting differentiation, in gingival epithelium. Disturbance in this proliferation/differentiation balance probably results from impaired phosphorylation of secreted signaling molecules, such as EGF, since many of these proteins contain G-CK phosphorylation motif (11). Although diminished phosphorylation of enamel matrix proteins is considered as the main cause of enamel defects in ERS (12, 28), aberrant epithelial proliferation/differentiation balance might also partly contributes to the pathogenesis. It has been shown that ameloblast differentiation at pre-secretory stage, prior to secretion of enamel matrix proteins, was disturbed in Fam20a knockout mice, which supports this hypothesis (25, 29). Like epithelial abnormalities, loss of FAM20A’s function in facilitating phosphorylation of secreted signaling molecules by G-CK might also underlie the pathology in the connective tissue, including excessive accumulation of extracellular matrix and ectopic calcifications. Comprehensive characterization and analysis of “phospho-secretome” from diseased tissues will be required to test the hypothesis and unravel the key affected phosphoproteins driving the various pathologies of ERS.