Pivotal Role of Tenascin-W (-N) in Postnatal Incisor Growth and Periodontal Ligament Remodeling

The continuously growing mouse incisor provides a fascinating model for studying stem cell regulation and organ renewal. In the incisor, epithelial and mesenchymal stem cells assure lifelong tooth growth. The epithelial stem cells reside in a niche known as the cervical loop. Mesenchymal stem cells are located in the nearby apical neurovascular bundle and in the neural plexus. So far, little is known about extracellular cues that are controlling incisor stem cell renewal and guidance. The extracellular matrix protein tenascin-W, also known as tenascin-N (TNN), is expressed in the mesenchyme of the pulp and of the periodontal ligament of the incisor, and is closely associated with collagen 3 fibers. Here, we report for the first time the phenotype of tenascin-W/TNN deficient mice, which in a C57BL/6N background exhibit a reduced body weight and lifespan. We found major defects in the alveolar bone and periodontal ligament of the growing rodent incisors, whereas molars were not affected. The alveolar bone around the incisor was replaced by a dense scar-like connective tissue, enriched with newly formed nerve fibers likely leading to periodontal pain, less food intake and reduced body weight. Using soft food to reduce mechanical load on the incisor partially rescued the phenotype. In situ hybridization and Gli1 reporter mouse experiments revealed decreased hedgehog signaling in the incisor mesenchymal stem cell compartment, which coordinates the development of mesenchymal stem cell niche. These results indicate that TNN deficiency in mice affects periodontal remodeling and increases nerve fiber branching. Through periodontal pain the food intake is reduced and the incisor renewal and the neurovascular sonic hedgehog secretion rate are reduced. In conclusion, tenascin-W/TNN seems to have a primary function in rapid periodontal tissue remodeling and a secondary function in mechanosensation.


INTRODUCTION
Rodent incisors, unlike human teeth, grow continuously throughout life. In young animals the entire length of the lower incisor renews every month (1). The incisor tip selfsharpens due to the asymmetrical distribution of the enamel, which covers only the labial surface of the tooth enabling abrasion of the softer dentin on the lingual surface resulting in a sharp labial enamel edge. Though the lower incisors are only visible at the most ventral aspect of the mandible, within the jaw itself the incisors occupy almost the entire length of the body of the mandible ( Figure 1A). In the dorsal mandible, epithelial and mesenchymal stem cells reside in the incisor and they provide a continuous supply of hard tissue forming cells. The epithelial stem cell niche is morphologically clearly defined; it is localized in a loop-shaped ending of the epithelial layer called cervical loop ( Figure 1B). Cervical loop is composed of outer and inner epithelia, which enclose loosely arranged stellate reticulum cells, including stem cells. The epithelial stem cells differentiate into enamel secreting ameloblasts. On the other hand, dentin is produced by odontoblasts, which derive from the incisor specific mesenchymal stem cells (MSCs) just recently discovered (2)(3)(4).
Lineage tracing experiments showed that Schwann cell precursors, which express Sox10 and PLP1, generate a major population of dentin producing odontoblasts (3). A second, vasculature associated mesenchymal stem cell population includes slowly dividing population expressing Gli1, as well as aSMA and NG2 expressing cells (2,4,5). Gli1 expressing population is regulated by Hedgehog signaling which is transduced through two Hedgehog ligands; Shh and Dhh (4,6). Proliferation and differentiation of tooth stem cells is directed toward the tip of the tooth and is controlled through epithelialmesenchymal crosstalk (7). In recent years it became clear that extracellular matrix proteins have a key role in controlling the organization of cell compartments: they provide cell attachment sites, form barriers, present growth factors and control the mechanical properties of the tissue (8). Each tissue has an unique composition and topology of its extracellular matrix (9). The structure and function of enamel and dentin matrix proteins such as ameloblastin, amelogenin, dentin matrix protein, and dentin sialophosphoprotein are well studied. However, the microenvironmental factors that control pulp stem cell fate are less well known. In this context, tenascins are highly interesting as they are expressed in embryonic mesenchyme and stem cell niches (10). Tenascins are large multifunctional proteins, which have a common domain structure: an N-terminal assembly domain, epidermal growth factor (EGF)-like repeats, fibronectin type III domains, and a C-terminal fibrinogen-like globular domain. In mammals four tenascins are expressed: Tenascin-C, -R, -W/(-N), and -X. One well known function of tenascins is their ability to modulate cell adhesion and migration in vitro (11). Tenascin-C and -R bind to neural cell adhesion receptors and co-receptors such as contactin (12,13), syndecan (14), and integrins (11). Tenascin-C is expressed in the central nervous system and in embryonic connective tissues. Mice deficient for tenascin-C were originally reported to develop normally (15), but more recently behavioral phenotypes and abnormal central nervous system development were described (16). The expression of tenascin-R is limited to the central nervous system (17) and the knockout mice show cognitive defects, reduced coordination and increased anxiety (18). In contrast, tenascin-X is widely expressed in loose connective tissues (19), and tenascin-X deficient mice show an Ehlers-Danlos syndrome-like phenotype with altered fibrillar collagen density and hyperextensible skin (20).
Tenascin-W, also known as tenascin-N, is widely expressed during embryonal development (21), but in adult mammals its expression is mainly restricted to the periodontal ligament (22), periosteum (21), and stem cell niches (10). In cancer, TNN expression is strongly induced and for breast cancer a role in migration and metastasis has been proposed (23,24). In vitro experiments show that tenascin-W/TNN promotes migration of several cell types, including breast cancer cells, osteoblasts, and endothelial cells (23,25,26). Hence, TNN may play an important A B FIGURE 1 | Graphical overview of the mandibular incisor anatomy. (A) A large fraction of the mouse incisor is embedded in the mandible. Enamel covers (magenta) only the labial surface of the tooth. Mesenchymal stem cells (MSC) and epithelial stem cells are found in the dorsal mandible. (B) Mesenchymal stem cells are regulated by sonic hedgehog (Shh) released from the neurovascular bundle and express Gli1, which is a hedgehog signaling effector. Those stem cells eventually proliferate in the transient amplifying cell (TAC) zone and are regulated by epithelial-derived Wnt. Axin2, a signaling molecule from the Wnt pathway, is a marker for those highly proliferative cells. The mesenchymal TAC cells then differentiate into odontoblasts and form dentin or are differentiating into pulp stroma cells. The epithelial stem cells are found in the labial cervical loop and they differentiate into ameloblasts and form enamel. role in cell migration and differentiation in the continuously growing incisor tooth.
In this study, we have analyzed the role of tenascin-W/TNN using a global tenascin-W/TNN deficient mouse. This first analysis of tenascin-W/TNN deficient animals showed that tenascin-W/TNN has a function in incisor periodontal ligament remodeling, in incisor eruption and in tooth renewal. Tenascin-W/TNN deficient mice show under hard pellet diet weight loss and reduced food uptake.

Micro-Computed Tomography
Micro-computed tomography (mCT) was performed using a mCT 35 Scanner (Scanco Medical). Complete mandibular bones (mice, n = 6) were scanned with an isotropic voxel size of 12 mm using 70 kVp tube voltage, 114 mA tube current, 400 ms integration time, segmentation support of 1, and a sigma correction of 0.8. For the enamel reconstruction of 6 and 12 month old mice a threshold of 27% for dentin and 60% for enamel was chosen. In 1 month old animals the threshold for dentin was selected at 23% for dentin and 60% for enamel, respectively.

Statistics
The data are shown as mean ± SD. Statistical analysis was performed using GraphPad Prism 5 software. Two-tailed, paired t-test was used for experiments involving two groups; p values lower than 0.05 were considered significant.

RESULTS
First, we analyzed the expression of tenascin-W/TNN in the incisor pulp mesenchyme during tooth development, using immunofluorescence. The expression started at the bell stage of tooth development, marked by ameloblast and odontoblast differentiation (34) (Supplementary Figures 1A, B). Tenascin-W/TNN is mainly found in the distal part of the dental papilla and colocalizes with thick spiral fibers (Figure 2A), while it is not detected in the apical part. Co-immunostainings showed that the tenascin-W/TNN labeled fibers contain collagen 3 ( Figures 2B, B´, B´´). In newborn mice, tenascin-W/TNN specific immunostaining is concentrated in the area of preodontoblasts and mature odontoblasts (Supplementary Figure  1C). In adult mice, only a weak signal for tenascin-W/TNN is persistent in the distal pulp ( Figure 2C); in contrast however tenascin-W/TNN is strongly expressed in the periodontal ligament of incisors and molars ( Figures 2C, D).
To study the in vivo function of tenascin-W/TNN, we generated a global tenascin-W/TNN knockout mouse line. These mice are vital and fertile, but show an increased death rate and stress intolerance. Otherwise, tenascin-W/TNN mice develop and grow normally, with the exception of visible enamel defects in the upper incisors at 12 months ( Figures 3A, B). To determine whether the lower incisors are affected as well, we analyzed lower jaws by micro-computed tomography (µCT). There were no obvious macroscopic differences in 1 month old animals, but 6 and 12 month old tenascin-W/TNN deficient mice showed a flattening of the incisor tip, suggesting a defect in abrasion and self-sharpening and/or altered occlusion. In addition, the mandibular bone is fenestrated in 12  The absence of tenascin-W/TNN expression at the level of the first molar in month old TNN-/-incisors (Supplementary Figure 1E) indicates that the changes in dentin thickness observed at this stage must be due to a secondary effect. Next, we analyzed the morphology of the stem cell niches and no morphological changes were observed in 1 month old mice ( Figure 5A). In 3 month old TNN(-/-) mice both the dentin and enamel formation have increased and the enamel epithelium contains cyst-like structures ( Figure 5B). Immunohistochemical staining for ameloblastin revealed that those epithelial inclusions are composed of enamel matrix ( Figure 5B). Next, we analyzed the position of pre-ameloblasts and mature ameloblasts, and of odontoblasts in lower incisors. In 1 month old TNN(-/-) mice dentin sialophosphoprotein (Dspp), sonic hedgehog (Shh), and ameloblastin (Ambn) were expressed normally as determined by in situ hybridization. In 3 month old mice the differentiation of both ameloblasts (Ambn probe) and odontoblasts (Dspp probe) was shifted to the apex (Supplementary Figure 2). Stainings for endogenous alkaline phosphatase further confirmed this shift in differentiation (Supplementary Figure 5), and revealed that alkaline phosphatase activity is high in 1 year old TNN(-/-) mice (Supplementary Figure 3).
These findings indicate either a defect in growth factor signaling or a reduced number of stem cells in TNN-deficient incisors. We therefore performed in situ hybridizations with Axin2, Fgf10, and Gli1 probes (Supplementary Figure 4). Expression of Axin2, a target and negative regulator of Wnt signaling, was chosen to analyze Wnt pathway activity, which controls dentin formation. FGF10 is produced in the pulp mesenchyme and maintains epithelial stem cells. Gli1 is a transcription factor that is expressed at sites of highly active hedgehog signaling (27). We found that tenascin-W/TNN deficiency leads to reduced expression of these three factors, indicating compromised Wnt, FGF, and hedgehog signaling in the region of the mesenchymal stem cell niche of the incisors, or, alternatively, loss of specific cell stem cell populations.
We next performed lineage tracing experiments in Gli1CreER T2 x R26mTmG x TNN(-/-) and Gli1CreER T2 x R26mTmG control mice. 72 hour chase experiments showed that the pulp mesenchyme of 3 month old tenascin-W/TNN knockout mice contained only few Gli1 expressing GFP positive cells ( Figures 6A, B) compared to controls. Ki67 immunostainings showed that the proliferation rate of putative mesenchymal stem cells was reduced in the incisors of 3 month old TNN(-/-) mice ( Figures 6C, D).
These findings led to the question whether the mesenchymal stem cell themselves were affected or not. To resolve this issue, we performed a flow cytometry analysis of isolated apical pulp cells with CD45.  Balic and Mina (37). This experiment showed no differences in the number of CD90.2 or Sca1 positive progenitor cells in 1 month old control or knockout animals (Supplementary Figure 5).
Furthermore, we performed immunostainings with a CD90.2 and Sox10 antibodies. Sox10 is a marker for pluripotent neural crest cells and glia cells (38) and was used to identify nerve-derived mesenchymal stem cells in the incisor (3). The stainings showed an accumulation of Sox10 positive cells in the apical pulp mesenchyme of 3 mt old tenascin-W/TNN deficient animals; however, CD90.2 immunolocalization was not altered. This indicates that the nerve-derived mesenchymal stem cells of TNN  Figure 6). Interestingly, aged TNN(-/-) mice never develop obesity. The body weight of 1 year old knockout animals was in both sexes significantly reduced ( Figure 7A). We hypothesized that TNN (-/-) induced damage and tooth pain prevent proper food intake. To test this hypothesis we fed our mice with a soft food diet, which showed that the physical form of the diet was sufficient to rescue the body weight. Hence, the malformed teeth are the cause for the observed weight differences (Figures 7A, B). The weight of the wild type mice did not change with a soft diet. Soft food also ameliorated the incisor phenotype of knockout mice: mCT reconstructions of the enamel layer showed that the enamel layer remained continuous ( Figure 7C), and histological sections of the incisor apical region showed improved epithelial and mesenchymal differentiation ( Figure 7D). Next, we analyzed mCT cross sections of soft food fed 6 month old mice ( Figures  8A, B). The pulp was now obliterated both in wild type and TNN deficient mice, indicating a slower tooth growth rate under soft food diet. Interestingly, the alveolar bone loss of TNN deficient mice persisted as well as the increased nerve branching under soft food diet (Figures 8C, D). This finding demonstrates that the periodontal ligament defect is the primary cause of the incisor eruption phenotype.

DISCUSSION
Our data reveal that tenascin-W/TNN is distinctly localized in the extracellular matrix and partially co-localizes with collagen 3 FIGURE 4 | Analysis of lower incisor cross-sections (A-F) mCT scans at the heights of the first molar showed that compared to wildtype (A-C) the incisor dentin layer is much thicker in 1 month old tenascin-W/TNN deficient mice (D, asterisk), in 6 and 12 month old mice the incisor pulp is completely obliterated (E, F, asterisk). A second phenotypic change in the mutant is the wider incisor periodontal space, indicated by an arrow (D-F). However, the periodontal space of the molars seems normal. (G, H) Sirius red stainings showed a significant alveolar bone loss and a replacement by a dense connective tissue in the mutant compared to wildtype. (I, J) An analysis of blood vessels (CD31 in red, smooth muscle actin (SMA) in white, Tuj1 in green) and nerve fibers revealed that in TNN-deficient mice, the remaining pulp tissue does not containing any projections (I´, J´); in contrast the bone replacement tissue contains a well developed network of nerve fibers (I´´,J´´) (a ameloblasts; pdl periodontal ligament; mice, n = 6, scale bar mCT: 0.5 mm, histology: 100 mm).

Imhof et al.
Analysis of the Tenascin-W/N Knockout fibers. In teeth, this distribution pattern is different from tenascin-C, which is more diffusely distributed in the connective tissues. Earlier studies on tenascin-W/TNN have mostly focused on the in vitro function or tissue expression in cancer (23,24). Here, we report the first tenascin-W/TNN deficient mouse for functional studies. The most striking phenotype of tenascin-W/TNN deficient mice is found in the incisors, which show irregular dentin and enamel formation, while no phenotype changes are seen in molars. Furthermore, we have observed alveolar bone loss and increased nerve branching in the significantly wider periodontal ligament of TNN deficient mice. Changes in the periodontal ligament observed in the TNN-/-mice suggest that connective tissue remodeling and continuous eruption are disturbed. Initial tooth eruption in rodents relies heavily on monocyte/ macrophage lineage cells as demonstrated by toothless rats (39) and in Csf1 deficient mice neither molar nor incisors erupt (40). However, little is known about the role of monocyte/macrophage lineage cells in continuous incisor eruption. The periodontal ligament is very different from the dental sac which is responsible for initial tooth eruption. Tenascins are known to support cell migration in vitro (23) and to modulate macrophage activity by TLR4 binding (41). These known functions are suggesting that in TNN-deficient mice remodeling of the periodontal ligament might be affected through altered macrophage migration or activity. Another hypothesis is that tenascin-W/TNN affects the mechanical properties of the periodontal ligament and periodontal fibroblasts are differently regulated. Changes in the connective tissue stiffness are known to modulate extracellular matrix remodeling (42). Tenascin-X knockout mice have altered collagen fibers mimicking Ehlers-Danlos syndrome (20). However, in tenascin-W/TNN deficient mice ultrastructural analysis of the periodontal ligament showed normal collagen fibers (data not shown). Furthermore, we have observed alveolar bone loss and increased nerve branching in the periodontal ligament of TNN deficient mice. Since the rodent incisor periodontal ligament is rich in Ruffini-and free nerve endings (43), it is likely that this defect leads to pain or discomfort. In agreement with this assumption we found in aged tenascin-W/TNN deficient mice a significant reduction of the body weight.
Since the eruption rate of incisors is under neural and mechanical control (44) we wondered whether soft food diet accompanied with wooden gnaw sticks would rescue the body weight defect. We found that the TNN deficient mice gained weight comparably to control mice under these conditions, and the irregular dentin and enamel formation of the incisor disappeared. However, the periodontal defect including alveolar bone loss remained in the soft food treated TNN deficient animals. This important finding indicates that the odontoblast and ameloblast anomalies observed in TNN deficient incisors under standard hard food diet are secondary to a primary defect in the periodontium, and might be caused by mechanical damage to the pulp and/or pain-related processes.
Next, we examined the impact of tenascin-W/TNN deficiency on the incisor stem cell niche under standard food diet. The decreased Gli1 expression and cell proliferation suggests reduced hedgehog activity (4), and again indicate that the irregular tooth formation is of secondary nature, as tenascin-W/TNN is not expressed in the stem cell niche and the percentage of progenitor cells is not altered. Hendaoui et al. (45) proposed that tenascin-C and -W/N modulate Wnt signaling by sequestering the growth factor. However, tenascin-W/TNN is not expressed in the incisor stem cell niche itself and therefore cannot modulate growth factor signaling directly. Hence, axin2 and Gli1 expression and cell proliferation are clearly due to secondary effects. Since it is very difficult obtaining comparable sections, we decided to include only unquantified data.
Similar phenotypes in the incisors have been observed in periostin or integrin a11 deficient mouse lines; both proteins play central roles in periodontal ligament remodeling and mechanosignaling. One of them, periostin, is a characteristic extracellular matrix component of the periodontal ligament, and periostin deficiency in mice leads to periodontal defects in incisors and molars. Periostin has a known function in collagen 1 fibrillogenesis (46) by supporting proteolytic activation of lysyl oxidase (47). Furthermore, periostin plays a role in vascular smooth muscle cell migration (48) and in the migration of bone lining cells (49). Interestingly, soft chow also ameliorates the phenotype observed in periostin deficient animals (50). One major difference to the phenotype observed in TNN knockout mice is that periostin deficiency affects also the molar periodontal ligament (50), not just that of the continuously erupting incisor. The only other extracellular matrix related knockout where the incisors are affected is that of integrin a11 (51). Integrin a11b1 functions as a collagen receptor and it has been suggested that it plays a role in mechanotransduction. Interestingly, soft food diet ameliorates also the integrin a11 phenotype (51). Both periostin and integrin a11 play a role in the remodeling of collagen 1 fibers. The irregular tooth formation observed in periostin deficient mice might be due to a periodontal ligament-neurosecretory feedback mechanism. Based on published data (41), we hypothesize that tenascin-W/TNN acts on macrophages or fibroblasts directly, and that the underlying mechanism of the remodeling defect is different. In conclusion, our data show that tenascin-W/TNN colocalizes with collagen 3 fibers in the pulp and the periodontal ligament. Here, we present the first characterization of tenascin-W/TNN deficient mice and we conclude that tenascin-W/TNN plays a pivotal role in incisor periodontal remodeling and diet uptake. The periodontal remodeling phenotype is furthermore connected with nerve branching. Since soft food diet results in equal body weights, but the periodontal defect persists in these mice, we assume that nerve branching might lead to pain. Since tenascin-W/TNN itself is not expressed in the stem cell compartment nor in the proliferation zone, we hypothesize that the effect in the pulp is of secondary nature, most probably due to reduced neural input. The diminished Gli1 expression in the pulp supports this assumption. Neurovascular secreted sonic hedgehog is a known regulator in this stem cell niche and Gli1 is the main effector of activated hedgehog signaling. In summary, we report for the first time about the in vivo function of tenascin-W/TNN and we show that this extracellular matrix protein plays a crucial role in periodontal remodeling.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ACKNOWLEDGMENTS
We thank Dr. Sarah Appel, Cologne for providing mouse embryo tissues and Semra Özcelik for technical assistance. We would like to thank to Dr. Mats Paulsson for correcting the manuscript and for helpful advice.