Aberrantly elevated Wnt signaling is responsible for cementum overgrowth and dental ankylosis

Vertebrate teeth are attached to the jawbones using a variety of methods but in mammals, a fibrous connection is the norm. This fibrous periodontal ligament (PDL) allows teeth to move in the jawbones in response to natural eruptive forces, mastication, and orthodontic tooth movement. In some disease states the PDL either calcifies or is replaced by a mineralized tissue and the result is ankylosis, where the tooth is fused to the alveolar bone. To understand how the PDL maintains this fibrous state we examined a strain of mice in which tooth movement is arrested. Da β cat Ot mice express a stabilized form of β -catenin in DMP1-positive alveolar bone osteocytes and cementocytes, which results in elevated Wnt signaling throughout the periodontium. As a consequence, there is an accrual of massive amounts of cellular cementum and alveolar bone, the PDL itself calcifies and teeth become ankylosed. These data suggest that to maintain its fibrous nature, Wnt signaling must normally be repressed in the PDL space.


Introduction
The periodontium is a tripartite structure consisting of a connective tissue, the periodontal ligament (PDL) and two mineralized tissues, the cementum and alveolar bone. The PDL spans the two mineralized tissues and this arrangement permits movement of the teeth within the jawbones. For example, in response to mastication the PDL dissipates potentially damaging strains that would otherwise accumulate in the bone and tooth [1]. Fibrous tissues that serve this type of function are referred to as entheses [2,3]. In health, entheses remain fibrous in nature even though they express osteogenic proteins [4]. In some disease states, entheses mineralize and this results in a condition known as ankylosis, where a joint becomes immobile.
Biological signals are required to maintain the fibrous state of the PDL (reviewed recently in [5]). For example, the transcription factor Scleraxis actively prevents mineralization, in part by opposing the effects of the transcription factor Osterix [6]. Physical signals-both tensile and compressive-are also implicated in maintaining the fibrous nature of the PDL [3]. Whether biological signals, physical stimuli, or a combination of both are required to maintain the fibrous state of the PDL, one thing is certain: if the PDL has inappropriately undergone mineralization then teeth become immobile [7][8][9][10]. Dental ankylosis represents a pathologic condition in mammals but in stem mammals including extinct flying and aquatic reptiles and dinosaurs, dental ankylosis was the natural terminal stage in successful tooth replacement [11].
To gain insights into the mechanisms regulating the development and potentially, evolution of the periodontium, we undertook an analysis of a strain of mice in which tooth movement is defective. Daβcat Ot mice express a stabilized form of β-catenin in osteocytes [12], and exhibit defects in the movement of both incisors and molars. We combined molecular, cellular, histologic, and imaging analyses to understand how a stabilized form of β-catenin led to prolonged Wnt/β-catenin signaling in alveolar bone osteocytes and surprisingly, in cementocytes as well, and how this aberrant Wnt signal led to dental ankylosis through mineralization of the PDL.

Micro-computed tomography analysis
Micro-computed tomography (µCT) scans of murine skulls were performed on daβcat Ot control (N=3) and daβcat Ot mutant (N=3) groups. Mice were sacrificed, tissues were fixed in 70% ethanol and then scanned at 10µm resolution (Micro XCT, Xradia Inc., Pleasanton, CA). Samples were reconstructed, segmented with ScanIP then using the measure function, the length of incisors were calculated. After orienting the µCT slice planes to show longitudinal sections of the incisors, the length of root-to-tip was measurements and the average length of 3 mice were calculated.

Histology, immunohistochemistry, and histomorphometric analyses
Animals were sacrificed at the time points indicated by first anesthetizing them with an IP injection of ketamine (80 mg/kg) and Xylazine (16 mg/kg) followed by cervical dislocation. Tissues were fixed in 4% paraformaldehyde overnight then washed in PBS for 30 min and decalcified in 19% EDTA for up to 14 days in 4ºC, with shaking. After decalcification, specimens were dehydrated through an ascending ethanol series and embedded in paraffin. Eight-micron-thick longitudinal, transverse, and coronal sections were cut and collected on superfrost-plus slides. Movat's pentachrome staining was performed on slides [15], further dehydrated in an ethanol series, and then mounted with Permount. Tartrate resistant acid phosphatase (TRAP) staining was performed [16]. Immunohistochemistry was performed [17] with primary antibodies: rabbit anti-PCNA (ab18197, Abcam, Cambrdige, UK), rabbit anti-Periostin (ab14041, Abcam), rabbit anti-Osterix (ab22552, Abcam), mouse anti-Sclerostin (AF1589, R&D systems, Minneapolis, USA), mouse anti-CTNNB1 (β-catenin) antibody (610154, BD Biosciences New Jersey, USA), rabbit anti-green fluorescent protein (GFP; 2956S, Cell Signaling Technology, Danvers, USA) and secondary antibodies including biotinylated goat anti-rabbit IgG antibody (BA-1000, Vector Lab, Burlingame, USA) and biotinylated horse anti-mouse IgG antibody (BA-2000, Vector Lab). The staining was visualized by ABC peroxidase standard staining kit (32020, Thermo Fisher Scientific, Rockford, USA). Tissue sections were photographed using a Leica digital imaging system. Histomorphometry was performed using Image J software (Version 1.49v) to quantify the TRAP activity. At least 4 slides from each sample were used for quantification. The total length of incisor was divided into three part from the tip to the apex. The first 1/3 part was defined as the region of interest (ROI) for the incisive edge and the third 1/3 was used for apex quantification. To quantify the cell number in PDL area, the total DAPI +ve cell in PDL area are counted and normalized to the PDL area.

Statistical analyses
Results were presented as mean ± standard deviation. A two-tailed Student's t-test was used to determine significant differences between data sets. A p value<0.05 was considered statistically significant and all statistical analyses were performed with Microsoft Excel software (Version 15.16, USA).
This study follows ARRIVE guidelines.

Tooth eruption is perturbed in daβcat Ot mice
A functional periodontium is required for active tooth eruption [18,19], and in murine incisors this eruption process continues throughout life. In daβcat Ot mice, however, incisor eruption was arrested. Compared with µCT scans of P42 littermate controls (Fig. 1A,C), the mandibular incisors of aged-matched daβcat Ot mutant mice were not visible in the oral cavity ( Fig. 1B,D). In control littermates, the incisor tip (green asterisk) projected past the most coronal extent of the alveolar ridge (white arrow; Fig. 1E) whereas in daβcat Ot mice the incisor tip (red asterisk) was still encased in alveolar bone (white arrow; Fig. 1F; quantified in G). Upon close examination, the PDL near the incisor tip was thin and discontinuous with no obvious fibrous attachment between the bone and the lingual incisor surface (Fig. 1F).
One explanation for this phenotype could be a lack of adequate tooth structure; in other words, the eruption machinery was intact but there was insufficient tooth material to protrude into the oral cavity. The overall lengths of the maxillary incisors in controls and mutants, however was equivalent (Fig. 1H,I), yet one erupts and the other doesn't fully erupt ( Fig. 1A-D). Therefore, it was unlikely the case that the eruption defect in daβcat Ot mice was solely caused by inadequate tooth structure.
Three-dimensional volumetric reconstruction data revealed that overall, the mandibular daβcat Ot incisors were significantly shorter than controls (Fig. 1J,K). We considered cellular mechanisms of incisor eruption in rodents. For example, cell proliferation at the rodent incisor apex drives eruption [20][21][22]. Using PCNA immunostaining, we identified proliferating cells in controls (Fig. 1L) and the number of PCNA +ve cells was notably reduced in the mutants (Fig. 1M). We also noted a defect in molar eruption (Fig. 1A,B) but PCNA staining failed to reveal any proliferating cells at the apices of these teeth (data were not shown). Therefore, it was unlikely that a lack of cell proliferation was solely responsible for the widespread eruption defect and turned to consider other mechanisms that could contribute to this phenotype.

The daβcat Ot eruption defect is not due to perturbed osteoclast activity
Osteopetrosis is a genetic condition of increased bone mass caused by defects in osteoclast formation and function [23]. Osteopetrosis can arrest tooth movement [24] because the alveolar bone fails to resorb properly [25,26]. To ascertain whether the daβcat Ot eruption defect was attributable to an osteopetrotic phenotype we evaluated osteoblast and osteoclast function around the roots using ALP and TRAP activities ( Fig. 2A,B). In controls, ALP and TRAP activities was readily detectable in the alveolar bone surrounding the molar roots ( Fig. 2A), as well as around the incisor buccal and lingual alveolar bone surfaces (Fig.  2C,D). In daβcat Ot mutants, TRAP activity was evident around the alveolar bone surrounding the molar roots ( Fig. 2B) and along the thickened buccal and lingual alveolar bone surfaces (Fig. 2E,F). The extent of TRAP staining was quantified, and these analyses showed that rather than being depressed, daβcat Ot mutants had significantly higher osteoclast activity compared to controls (Fig. 2G). These data agree with previously published analyses of the daβcat Ot strain [12]. Notably, in some regions, the low density PDL space is missing and the alveolar bone appears to be fused with the tooth (Fig. 2E, yellow boxes).

Daβcat Ot mutants lack a function periodontal ligament
The daβcat Ot eruption defect was not due to an arrest in bone resorption nor was it due to a primary defect in osteoclast activity; we therefore considered other explanations for the tooth eruption defect in daβcat Ot mutants. For example, tooth movement is possible because of the fibrous nature of the PDL (Fig. 3A); we reasoned that changes in the width and/or its fibrous state might adversely affect tooth eruption. We first determined there was a narrowing of the PDL space in the mutant: at post-natal day 24 (P24), histomorphometric comparisons between the strains revealed a slightly reduced PDL space in daβcat Ot mutants (compare Fig. 3A,B; quantified in E). Histomorphometric comparisons were made again at P42. The PDL normally widens until animals reach adulthood, which was seen in the control group (compare Fig. 3A,C). In daβcat Ot mutants, however, the PDL instead had narrowed significantly (Fig. 3D, quantified in E). Closer analyses revealed aberrant mineralization within the PDL proper that appeared like "spot-welds" (arrows) within what is a normally fibrous tissue (compare Fig. 3F,G). These observations together suggested that the mineralized tissues surrounding the PDL e.g., the alveolar bone and/or cementum were either enroaching onto the PDL space, or that the PDL itself was undergoing mineralization. µCT imaging analyses verified the widespread nature of this aberrant mineralization: normally, the PDL appears as a dark "halo" around teeth ( Fig. 3H) but in daβcat Ot mutants, the radiolucent area was replaced in places by a radio-opaque tissue (arrows, Fig. 3I).
To further examine the fibrous versus mineralized state of the daβcat Ot PDL, we examined daβcat Ot control and mutant mice for the expression of Periostin. Periostin is normally expressed in the fibrous PDL ( Fig. 3J) but was minimally detectable in the daβcat Ot mutant (Fig. 3K). On tissue sections from daβcat Ot mutants, we found that in sites that still retained a fibrous PDL, the osteogenic protein Osterix was expressed very strongly (compare Fig.  3L,M). Did this change in Osterix result from a direct increase in the number of osteogenic cells in the daβcat Ot PDL? To address this question, we verified that cell proliferation in the daβcat Ot mutant PDL, just like in an adult wild-type PDL [4], was minimal (data were not shown). We then performed DAPI staining to quantify the density of cells in the PDL area in both daβcat Ot control and mutant mice. The number of DAPI +ve cells in PDL was normalized to the area. These data demonstrated that although the PDL space was narrower in the daβcat Ot group, cell density was comparable to the control group (Fig. 3N,O, quantified in P). Finally, to confirm that the Osterix +ve cells were indeed committed to an osteogenic fate, expression of a late-stage osteogenesis marker [27,28], Osteocalcin, was examined. Osteocalcin is normally absent from PDL cells (Fig. 3Q) but Osteocalcin +ve cells were present in the daβcat Ot PDL (Fig. 3R). Together, these data demonstrated that the increase in osteogenic cells within the daβcat Ot PDL was directly due to a greater percentage of PDL cells committing to an osteogenic fate.

Sustained Wnt/β-catenin signaling in the periodontium causes ankylosis
Thus far our data demonstrated that the replacement of a fibrous PDL by mineralized tissue was the primary cause of ankylosis in daβcat Ot mutants. Our next analyses set out to determine how amplified Wnt signaling in the daβcat Ot mice specifically caused dental ankylosis.
We confirmed that Wnt signaling was elevated in the daβcat Ot PDL using an antibody to phosphorylated β-catenin. Compared to controls (Fig. 4E), phosphorylated β-catenin expression was higher in daβcat Ot PDL (Fig. 4F). Sclerostin expression domains were also expanded: normally, Sclerostin is expressed in osteocytes and cementocytes (Fig. 4G) but in daβcat Ot mutants, its expression was also detected in the PDL (Fig. 4H). We returned to the DMP1 reporter strain of mice, this time focusing on how the daβcat Ot mutation affected DMP1 expression itself. In comparison to controls, DMP1 expression itself was dramatically increased in daβcat Ot mutants (Fig. 4I,J, compared with Fig. 4 B,D). Collectively, these data demonstrated that Wnt signaling was aberrantly elevated in the periodontium of daβcat Ot mice.
We examined the longer-term sequelae of this elevated Wnt signaling in the periodontium. Normally, acellular cementum covers most of the root surface, and cellular cementum is restricted to the root apices (Fig. 4K,L). In daβcat Ot mutants, however, the elevated state of Wnt signaling caused an dramatic expansion in cementum (Fig. 4M,N). At the apex of the root, cellular cementum is normally separated from the adjacent alveolar bone by an intervening fibrous PDL (Fig. 4O). In daβcat Ot mutants, however, the fibrous PDL was replaced by a solid mass of mineralized tissue (Fig. 4P). This series of events obliterated the distinction between alveolar bone and cellular cementum, and effectively fused the tooth to its socket. Thus, elevated Wnt signaling in multiple components of the periodontium results in the formation of mineralized tissue at the expense of fibrous tissue, thereby resulting in dental ankylosis.

Regulating PDL mineralization
Entheses are tissues that must remain fibrous in order to function. In some disease states, as we have shown here, the fibrous PDL enthesis mineralizes and the result is ankylosis. A wide variety of factors-both biological and physical-can influence the mineralization state of entheses. For example, the mineralization state of the cementum can be influenced by a phosphate/pyrophosphate ratio [29] and it is tempting to speculate that a similar general mechanism might underlie mineralization of the PDL. There is no direct data to support this hypothesis, however, and it must be emphasized that, rather than a generalized phenomenon, the aberrant mineralization events in the daβcat Ot PDL are not uniform. Rather, the mineralization nodes appear more like "spot-welds" within a normally fibrous tissue (Figs.  2,3). This complicates a biochemical analysis of the mutant PDL as a tissue, since the mineralization spots are surrounded by fibrous PDL.
Matrix Gla protein (MGP) has also been recognized as a potent calcification inhibitor, which also suggests a potential role in preventing PDL mineralization. In vitro data support this conjecture [30], but again, whether a similar mechanism is at work to maintain a fibrous PDL is simply not known. At least in vitro, the aberrant expression of an extracellular matrix protein such as bone sialoprotein is initially sufficient to promote mineralization, but the effect is transient [31], and so is unlikely to be the causal link in the daβcat Ot ankylosis phenotype.
Another potential clue may come from analysis of ANK, a transmembrane protein that control movement of intracellular inorganic pyrophosphate across the cell membrane [32]. Genetic modulation of ANK in humans and other vertebrates have identified a role in regulating tissue calcification [32] and potentially, mineral deposition in tendons and ligaments. There is, however, no mention of dental ankylosis in ank/ank mice [33]. Clearly, there is still much to be learned about what factors, both biological and physical, are key regulators of mineralization within the functioning PDL.

Wnt signaling is essential for the homeostasis of the periodontium
An abundance of data demonstrates that Wnt signaling is required for proper development of the periodontium. For example, if Wnt signaling is disrupted early during embryogenesis then eventually the development of the periodontium is arrested [34][35][36]. If Wnt signaling is disrupted after the periodontium has formed, then homeostasis of the periodontium is disrupted [37,38]. Wnt signaling is also required to maintain alveolar bone [39] and cementum [37]. Given the interdependence of the cementum, bone, and the PDL on one another during development and homeostasis, it is challenging to unravel just when and where endogenous Wnt signals are having their primary action. Daβcat Ot mutant mice offered rare insights into this conundrum because Wnt signaling was amplified only after the embryonic periodontium had formed [40], and then only in very specific cell niches.

Elevated Wnt/β-catenin signaling causes mineralization of the PDL
Daβcat Ot mutant mice demonstrate a progressive expansion of Wnt responsive cells that has an unusually detrimental effect on the periodontium. Wnt/β-catenin signaling is normally restricted in alveolar bone osteocytes and cementocytes, but in daβcat Ot mutant mice, Wnt signaling spreads to envelop the entire PDL (Fig. 4). From our analyses of the DMP1 reporter strain, we knew that DMP1 becomes aberrantly expressed in the daβcat Ot mutant PDL (Fig. 4B,D), a conclusion that was validated using DMP1 immunostaining (Fig. 4I,J). Precisely how this expansion in DMP1 expression occurs was not immediately obvious. Our current hypothesis is that the gradual accretion of bone and cementum constricts the PDL space in daβcat Ot mice (Fig. 3A-D). Both osteoblasts and cementoblasts normally produce and respond to Wnt signals (Fig. 4) therefore it is reasonable to speciulate that as the PDL space narrows, Wnt signals emanating from the mineralizing tissues begin to impact cells in the middle of the PDL space. The subsequent expression of DMP1 in PDL cells may therefore be a cause-rather than an instigating factor-in PDL mineralization.
While some parts of this hypothesis are still speculative, one thing is clear: as the Wnt responsive cell populations expand so, too, does the mineralized tissues. Within short order, both the volume and distribution of cementum and alveolar bone have enlarged in daβcat Ot mutants (Fig. 4), which together obliterates the fibrous PDL (Fig. 4). Other investigators have shown that stabilizing beta-catenin via an Osteocalcin-Cre reporter strain leads to an accumulation of cellular cementum deposited over the entire molar root surfaces of molar in the mutant mice, which causes a narrowing of the PDL space [41,42]. Likewise, stabilizing beta-catenin via an Col1a1-Cre reporter strain to leads to abnormal tooth eruption and excessive cementum hypoplasia [36]. Neither group, however, reported an eruption defect due to ankylosis, as we have shown here for daβcat Ot mice. Finally, there is the possibility that the ankylosis phenotype in daβcat Ot mice is that because of sustained Wnt signaling, the PDL itself undergoes calcification (Fig. 4). Whether the cementum and bone are simultaneously encroaching on the PDL space, or the PDL itself undergoes mineralization, the end result is a fusion of the cementum to the alveolar bone, which arrests tooth eruption. Collectively, these data also raise a possibility that in order to maintain its fibrous nature, Wnt signaling must normally be repressed in the PDL space. Mechanosensory studies support this relationship between repressed Wnt signaling and a fibrous fate [43,44].

A pattern of ankylosis
The pattern of ankylosis we observe in the daβcat Ot mutant may reflect the pattern of tooth eruption. For example, the daβcat Ot phenotype is based on conditional expression of a stabilized form of beta-catenin, driven by the DMP1 promoter. Since mandibular molars erupt prior to maxillary molars, we speculate that expression of stabilized beta catenin has not yet reached its maximum in the mandible by the time molars initiate their eruption. Conversely, by the time the maxillary molars erupt the expression of stabilized beta-catenin may have reached its zenith. Analyses of other strains of mice in which Wnt signaling is aberrrantly elevated in the periodontium may shed light on this hypothesis.

Ankylosis may represent a developmental atavism
In mammals, ankylosis represents a pathological condition that prevents tooth movement, and our study suggests that its etiology is related to pathologically sustained Wnt signaling. In some animals, however, ankylosis is not a pathologic condition. Rather, it represents a natural stage in odontogenesis [45]. For example, in stem mammals including extinct flying and aquatic reptiles and dinosaurs, dental ankylosis was the norm [46]. In the marine lizard Mosasauridae, cellular cementum and alveolar bone accumulated throughout the animal's life, leading to ankylosis [11,47].
This same interpretation aptly describes the daβcat Ot mutant phenotype, where accelerated accumulation of both mineralized tissues eventually causes ankylosis of the molars and incisors. In the case of daβcat Ot mutant mice, the molecular basis for the phenotype is clear; it is tempting to speculate that a similar molecular mechanism of sustained Wnt signaling in the periodontium might have been shared by the Mosasauridae tooth attachment apparatus [11].  In the PDL of P42 daβcat Ot control mice, Osteocalcin is absent. Relative to this control, (R) Osteocalcin expression is observed in daβcat Ot mutant mice in the cleft of tissue remaining between the tooth surface and alveolar bone (N=4). Abbreviations: as indicated previously and M1, first molar; M2, second molar; Scale bars: H,I=1000 μm, others =100 μm. cellular cementum at the root apex is juxtaposed to a fibrous PDL. (P) In daβcat Ot mutants, the mineralized tissue of the alveolar bone and cementum are contiguous. Abbreviations: as indicated previously. Scale bars: K,M=1000 μm, others =100 μm.