A perivascular stem cell niche at the tip of mouse incisors regulates tissue regeneration

Abstract

Cells with mesenchymal stem cell (MSC) properties are present in adult tooth pulp as quiescent cells that are mobilised by damage. These MSC-like cells respond to tooth damage by stimulating their proliferation and differentiation into odontoblast-like cells that form dentine to repair (reparative dentine) the damage. In the continuously growing mouse incisor, tissue at the incisor tips is continuously being damaged during feeding by the shearing action between the upper and lower teeth that acts to self-sharpen the tips. We have investigated mouse incisor tips as a model for the role of pulp MSCs in a continuous natural repair/regeneration process. We show that the pulp at the tip of the incisor is composed of a disorganised mass of mineralised tissue that is produced by cells expressing the odontoblast cell marker DSPP. These cells become embedded into the mineralised tissue that is rapidly formed and then lost as the animal feeds. Tetracycline labelling revealed the expected incorporation into newly synthesised dentine mineralised tissue formation along the length of the incisor, but also a zone covering the pulp cavity at the tips of the incisors that is mineralised very rapidly. This tissue was dentine-like, but had a significantly lower mineral content than normal dentine as determined by Raman spectroscopy. The mineral was also more crystalline than normal dentine, suggesting that it was composed of small, defect-free mineral particles.

To identify the origin of the cells responsible for the deposition of mineralised tissue we genetically labelled perivascular cells (pericytes) by crossing NG2ERT2Cre and Nestin Cre mice with reporter mice. A large number of pericyte-derived cells were visible in the pulp at the incisor tip with some having an elongated, odontoblast-like shape. These results show that in the mouse incisor, rapid, continuous mineralisation occurs at the tip to seal-off the pulp tissue from the external environment to prevent infection. The mineral is formed by perivascular-derived cells that differentiate into cells that express DSPP and produce a dentine-like material in a process that appears to function as continuous natural tissue regeneration.

Introduction

Dentine is a specialised mineralised connective tissue that is the main component of teeth and is made by unique polarised mesenchymal-derived cells the odontoblasts [1, 2]. In adult teeth of mammals, dentine is not remodelled but is continuously produced at a very slow rate during life that results in the progressive reduction of the volume of the tooth pulp with age [3]. When dentine is damaged during trauma or from dental caries it can undergo limited repair either from existing odontoblasts or from new odontoblast-like cells that are derived from mesenchymal stem cells present in the soft inner tissue of the tooth, the dental pulp [4–6]. The tooth pulp provides the source of the cells that form reparative odontoblasts and genetic lineage tracing of cells following extensive dentine damage in mice has shown that perivascular cells (pericytes) in the location of the damage are stimulated to proliferate, leave the vessels and differentiate into odontoblast-like cells that produce a reparative form of dentine to form a dentine bridge [7].

Mouse incisors are specially-adapted teeth that grow continuously throughout life [8]. This continuous growth is an adaptation to compensate for the loss of tooth material from the tips of the incisors as they occlude and self-sharpen. The growth of incisors is mediated by stem cell populations located at the proximal end (cervical loop) that include separate epithelial and mesenchymal stem cell niches [9–12]. The mesenchymal stem cells are located in the tooth pulp in the proximal-most region in the neurovascular bundle and consist of a population of slow cycling cells that express the Shh pathway transcription factor Gli1 [12]. The mesenchymal stem cells have at least two separate origins as identified by genetic lineage tracing. Up to 50% can be identified as being derived from neuronal glia and the other 50% are likely to be perivascular derived [13]. When incisors are damaged, the mesenchymal stem cells in the neurovascular bundle niche are able to respond to the damage but the extent to which they are involved in dentine repair is unclear [12, 13].

A direct consequence of the shearing force at the incisor tips that abrades mineral and tissue to produce sharp points is that the pulp is exposed to the external environment. Pulp is a highly vascularised and innervated tissue that provides vitality to the tooth and any exposure makes it susceptible to infection which occurs in dental caries [14]. Since mouse incisors do not become infected following tissue loss we assumed there must be a mechanism that prevents this. We thus set out to investigate the processes that occur at the tips during tissue loss to determine how the tooth pulp is protected from infection. Using a combination of approaches we identify a natural soft tissue mineralisation process that occurs in the exposed pulp tissue that results in the formation of a thin sheet of mineral that protects the pulp. The rapidly produced mineral is made by odontoblast-like cells at the tooth tip and has characteristics of dentine, and we have christened this “restorative dentine”. If animals are fed a soft diet to limit tissue loss, the mineralisation becomes more extensive, suggesting this is a continuous process that appears “cyclic” only as a result of the newly formed mineral being shed as the teeth re-sharpen. We identify a source of odontoblast-like cells that produce restorative dentine as pericytes.

Materials and Methods

Tetracycline incorporation

Four injections of 41.6 nmol/g body weight of tetracycline hydrochloride was administered to adult CD1 wild type mice at 5 day intervals before sacrificing 5 days after the last injection. In addition, a single tetracycline injection was administered in other wild type mice and collected 24 hours later. The mandibles were dissected out and fixed overnight in 4% paraformaldehyde at 4°C. The tissues were dehydrated in sucrose without prior decalcification, embedded in optimum cutting temperature (OCT) medium and cryosectioned (approximately 100μm thickness). Samples were imaged using a Leica SP5 laser-scanning confocal microscope with an ultraviolet laser (LD405 nm) and 405- to 488-nm excitation filter.

41.6nmol/g body weight of tetracycline was injected into adult mice four times at 5 day intervals and the incisors were collected 5 days after the last injection followed by cryosectioning without decalcification. Images of the incisor tips clearly show four bands of tetracycline incorporation into newly synthesised dentine of the growing incisor as expected. In addition a separate patch of incorporation was visible in the pulp cavity at the very tip, indicative of mineralisation of the pulp

Micro-Raman spectroscopy

Raman spectra were recorded on incisor tips using a Renishaw inVia mapping microRaman spectrometer outfitted with a Leica microscope and 785 nm in-line diode laser with a line shaped profile. A 1200 line/mm grating was used to maximize spectral resolution (~1 cm^-1). Wavelength and intensity calibrations were completed using the 520 cm^-1 Si-band from an internal silicon standard. For line mapping studies, a pinhole was inserted into the beam path to produce a circular excitation spot on the tooth surface. Ten second integration times were used at 10% laser power to avoid damaging the tooth. Data were collected at three-micron steps along the length of the maps. Two dimensional maps were assembled by removing the pinhole and rastering across the x-axis using the Renishaw Streamline collection feature. This produced steps of 1 µm in the x-direction and 1.4 µm in the y-direction while autofocusing the z-direction. Spectra were processed using Wire 4.1 software. Prior to peak fitting, the spectral areas were normalized to an area of 1 and an offset baseline was subtracted from all spectra. Peaks were fit using a full Voigt curve. Using this fit, peak areas as well as full width at half maximum (FWHM) and peak position were determined. A gage repeatability and reproducibility study of the Raman spectrometer and peak fitting algorithms confirmed the ability to determine peak width within 0.3 cm^-1.

Tip response to soft/hard feeding regime

To investigate the response of the incisal tips towards an external stimulus, mice were placed onto soft “mash” diets for a period of 1 or 4 days in contrast to their normal diet which consists of hard rodent chow pellets. Maxillas were dissected out and fixed overnight in 4% paraformaldehyde (PFA) at 4°C followed by decalcification in 10% EDTA pH7.4 for 4-6 weeks. After demineralisation, the samples were dehydrated through a series of ascending ethanol concentrations followed by paraffin embedding in a sagittal orientation. The samples were sectioned at a thickness of 8µm and mounted onto Superfrost-plus slides prior to histological processing.

Histology

Haematoxylin and eosin (H&E) staining was performed to visualise cell morphology. Haematoxylin stains the nuclei blue while cytoplasm, connective tissue and other extracellular structures are stained pink or red by eosin. After staining, sections were viewed in brightfield using a Zeiss microscope (Axioskope 2 plus) and captured with an AxioCam HRC using Axiovision software.

In situ hybridization

In situ hybridization was performed on paraffin sections following standard procedures under RNase-free conditions. mRNA digoxigenin-labelled antisense probes were prepared from complementary DNA (cDNA) templates for Dspp. Briefly, sections were dewaxed and washed in histoclear and rehydrated in descending ethanol washes. Tissues were permeabilised using Proteinase K and re-fixed in 4% PFA. Sections were treated with acetic anhydride (0.25%) in 0.1M triethanolamine to minimise background signals, washed and dehydrated followed by incubation with in situ hybridisation buffer at 65°C overnight with antisense RNA Dspp probe. The next day, to remove unhybridised probe, the sections were placed in high stringency washes with sodium citrate. Followed by washing in maleic acid buffer and blocked in buffer containing 10% sheep serum with 2% blocking reagent (Roche) for 1 hour at room temperature then incubated in blocking buffer supplemented with a 1:5000 dilution of anti-digoxygenin-AP Fab fragments (Roche) conjugated to alkaline phophatase overnight at 4°C for probe detection. Sections were treated with 0.5mg/mL levamisole to reduce background alkaline phosphatase activity and incubated in NTMT buffer with 50% polyvinylalcohol, 100mM Tris-HCL pH9.5, 100mM NaCl, 5mM MgCl₂ and 0.1% Tween-20, 4-nitro blue tetrazolium (Roche) and 5-brom-4-chloro-3-indoyl- phosphate (Roche) to develop the colour reaction.

Tamoxifen administration

In NestincreERT;R26R transgenic mice, Nestin-Cre expression is induced by tamoxifen administration and leaves indelible labelling only in cells expressing Nestin during the period of tamoxifen administration, including pericytes in the dental pulp. We therefore used this system to linage-trace the contribution of pericytes in the tooth pulp to identify the origin of cells producing mineralized tissue at the incisor tips. These double transgenic mice were given 3 intraperitonial injections of 4 mg tamoxifen (200µl of 20mg/ml tamoxifen in corn oil solution) per 30 g body weight over 3 weeks to activate the cre-expression in Nestin expressing cells.

Immunohistochemistry

For the NG2cre; Z/EG transgenic animals, the fate of NG2 cells can be lineage traced by permanent expression of EGFP in cells that have undergone Cre-mediated recombination. Therefore, green fluorescent protein immunohistochemistry following standard immunohistological staining procedures was used. Briefly, after tissue sections were deparaffinized and rehydrated, heat-based antigen retrieval using a microwave oven at full power for 10 minutes in 0.1M Tris-HCl pH9.5 solution was performed. After PBS washes, endogenous peroxidise activity was blocked with 3% H₂O₂ in methanol for 5 minutes and washed again in PBS before incubation in blocking buffer (10% fetal bovine serum, 1% bovine serum albumin in PBS) for 1 hour at room temperature. Rabbit polyclonal Anti-GFP antibody (Abcam, ab6556) was diluted 1:500 in blocking buffer and added to the sections for overnight incubation at 4ºC in a humidified chamber. To perform peroxidase visualization for the antibodies, the sections were incubated in ABC solution (Vectastain kit) for 1 hour at room temperature and then washed in PBS. The colour reaction (brown) was developed by DAB Peroxidase Substrate Kit, following manufacturer instuctuctions, and was stopped by rinsing with H₂O for 5-10 minutes. The sections were then counterstained in Hematoxylin, dehydrated and coverslipped with Neomount (Merck) mounting medium. Sections were viewed in light-field using a Zeiss microscope (Axioskope 2 plus) and captured with an AxioCam HRC (Zeiss) using Axiovision software.

β-galactosidase staining

In adult Nestincre; R26R mice, cells positive for Nestin express the lacZ gene product β-galactosidase which is detected by X-gal staining. Incisor samples were harvested and fixed in 0.2% glutaraldehyde solution with 2mM MgCl₂ at 4°C overnight. Followed by decalcification in 10% EDTA pH7.4 for 4-6 weeks. Sucrose dehydration was performed by submerging the samples in PBS containing 15% sucrose and 2mM MgCl₂ overnight at 4°C and then in PBS containing 30% sucrose, 2mM MgCl₂ in 50% volume of OCT (optimum cutting temperature) medium for 1.5h at room temperature. The samples were then embedded in a sagittal orientation in OCT medium using dry ice in ethanol and sectioned at a thickness of 12µm sections. Sections were then refixed in 0.2% glutaraldehyde solution with 2mM MgCl₂ for 10 minutes on ice. After rinsing with PBS, the slides were stained with X-gal (50mg/ml) overnight at 37°C, rinsed in PBS, refixed in 4% PFA and counterstained in nuclear fast red, dehydrated and coverslipped with Neomount (Merck) mounting medium.

BrdU incorporation

To detect rapid cell division, per 10-gram body weight, 0.5 mg BrdU (50 µl of 10mg/ml BrdU stock in 0.9% saline) was administrated intraperitoneally to CD1 wild type adult mice and sacrificed 24 hours after injection and processed through histology and immunohistochemistry analysis. The same immunohistochemical staining procedure was performed as previously described using Rat monoclonal Anti-BrdU primary antibody (Abcam ab6326) diluted to 1:500.

Results

Formation of mineralised tissue in the pulp of incisor tips

Light field microscope images of the tips of mouse incisors clearly show a white opaque material in the centre of the pulp that has the appearance of mineral (Fig. 1A,B). Since newly forming mineral has the ability to incorporate tetracycline [15, 16] we made use of this to determine if mineral was being produced in the pulp at the tips of incisors. 41.6nmol/g body weight of tetracycline was injected into adult mice four times at 5 day intervals and the incisors were collected 5 days after the last injection followed by cryosectioning without decalcification. Images of the incisor tips clearly show four bands of tetracycline incorporation into newly synthesised dentine of the growing incisor as expected. In addition a separate patch of incorporation was visible in the pulp cavity at the very tip, indicative of mineralisation of the pulp (Fig. 1C). In order to reveal the kinetics of this mineralisation a single injection of tetracycline was administered and incisors analysed 24h later. Using this regime, incorporation into normal dentine was barely detectable. However a very clear area of incorporation in the pulp cavity at the tip was observed (Fig. 1D). This shows that mineral is being deposited in the pulp at the tip of the incisor and that this is a process that is more rapid than the deposition of dentine that occurs during incisor growth.

Figure 1.

Adult CD1 mouse incisor tip morphology. The tips of the mouse maxillary (A) and mandibular (B) incisors indicate a region central to the occlusal surface of the tooth containing morphologically irregular mineral in comparison to the surrounding dentine that is smooth in appearance (arrows in A and B). Non-decalcified mandibular incisor tips from mice injected with 4 cycles of tetracycline at 5 day intervals indicate incorporation of tetracycline in newly synthesised mineral shown by the UV fluorescent bands (C). A single dose of tetracycline followed by a 24 hour chase period revealed an intense fluorescent patch demonstrating freshly deposited tooth mineral (arrow in D), corresponding to the irregular mineralised region on the tip surface (arrows in A and B). Abbreviations d: dentine. Scale bars represent 500um (A, C) and 250um (B, D).

The rapid mineralisation of the pulp at the incisor tips suggested that this mineral is being produced to restore pulp integrity and prevent exposure. The location of the mineral was indicative of it being dentine-like, rather than enamel-like. To further understand the mineral composition of the incisor tips and compare differences between the region of normal and irregular dentine at the occlusal surface of the tips based on the morphological differences observed, micro-Raman spectroscopy was carried out. This laser based technique enables biochemical analysis of cells and tissues using the inelastic scattering of light by chemical bonds, allowing the biomolecular composition of cells or tissues to be determined by the relative intensities of characteristic molecular vibrations [17]. Raman spectra of the newly synthesised mineral more closely resembled that of dentine than enamel, as peaks for both mineral and matrix components were dentine-like (Fig. 2), confirming our initial assumption. Analysis of mineral to matrix ratio and full width at half maximum (FWHM) of the ν1 PO₄^3- peak by line mapping across the pulp region at the incisor tip, however, showed a marked drop in both (Fig. 3A,B). Indeed, when we conducted univariate analysis of the spectra, we found that the material in the pulp region was significantly less mineralised than normal dentine and the ν1 PO₄^3- had a significantly lower FWHM when compared to normal dentine (Table 1, p < 0.01). Taken together, these data suggested that the mineral that was forming in the incisor tip was composed of small, highly crystalline/defect free particles that had newly nucleated in the organic matrix and not yet accumulated defects characteristic of more mature dentine. Detailed spectral maps (Fig. 3C,D) of the pulp region revealed variability in both mineral to matrix ratio and FWHM of the PO₄^3- peak. This suggested that mineral formed in discrete patches, creating a two phase mineral-matrix structure. The presence of such mineral particles in a matrix phase may account for the large variability in the FWHM of the PO₄^3- peak in this region (Table 1).

Figure 2.

(A) Typical Raman spectra of mouse enamel, dentine and restorative dentine over the spectral range 400 to 3200 cm^-1. Raman spectra for mammalian mineralised tissues are dominated by a large peak at ~960cm^-1, which is attributable to phosphate in biological apatite. Note that the full intensities of the 960 cm^-1 peak in the spectra for dentine and enamel have been attenuated to better visualise peaks indicative of other components. (B) Typical Raman spectra of mouse enamel, dentine and restorative dentine in the 2700 to 3200 cm^-1 spectral range showing that restorative dentine is more alike to dentine than it is to enamel. In (A), Note that restorative dentine, like dentine, displays a peak at 1660 cm^-1, which is attributable to Amide I, but is absent in enamel. Moreover, in (B) the CH, CH₂, and CH₃ stretching modes centred at 2935 cm^-1 - which are produced by the proteinaceous matrix, are alike in restorative dentine and dentine but differ in enamel.

Figure 3.

(A) Photograph of mouse incisor showing the positions at which Raman line scans (i and ii) and 2D (iii) maps were collected. (B) Mineral to matrix ratio and full width at half maximum (FWHM) of the 960 cm^-1 ν1 PO₄^3- peak in a typical line scan across the pulp region at the tip of the mouse incisor. Notice that the pulp region has a lower mineral content, as determined by the mineral to matrix ratio, but the mineral is more crystalline, as indicated by the lower FWHM of the ν1 PO₄^3- peak. (C) 2D map of FWHM of the ν1 PO₄^3- peak and (D) mineral to matrix ratio. Notice that both the FWHM and mineral to matrix ratio in the pulp region are quite variable, suggesting that mineral forms in discrete patches creating a 2 phase mineral-matrix structure. Scale bar represents 100µm in A.

Table 1.

Univariate analysis of Raman spectra of normal dentine and restorative dentine. Values represent means +/- standard deviations. Enamel has a negligible matrix component, so mineral to matrix ratio was not calculated. * indicates a significant difference when compared to normal dentine. (n=151, normal dentine; n=40, restorative dentine; n=15, enamel)

	Normal Dentine	Restorative Dentine	Enamel
Mineral to Matrix Ratio	2.685 +/- 0.367	0.639 +/- 0.137*	N/A
FWHM ν1 PO43- (cm-1)	17.889 +/- 0.697	16.58 +/- 3.20*	12.101 +/-0.274*

The Raman results indicated that the composition of the restorative dentine included a significant proportion of organic material consistent with cells. We thus carried out histological sectioning through the incisor tips that revealed the mineralised area was very disordered and included cells embedded in the mineral and remains of blood cells that had undergone a form of coagulation (Fig. 5B). The mineral structure together with its composition suggests that restorative dentine is a form of dentine that is produced very rapidly in response to pulp exposure. This suggested that pulp exposure may provide a stimulus for pulp cells to differentiate into odontoblast-like cells and that restorative dentine formation is a controlled, natural repair-like response. To explore this further, mice were fed a soft diet in order to reduce the wear of the incisor tips. In this situation once the pulp is protected by its layer of restorative dentine we assumed that if this was a repair process, further mineralisation would then cease if the mineral was not lost by abrasion. Surprisingly however we observed that mineralisation continued, to the extent that the pulp chamber began to mineralise more proximally and mineral also protruded distally from the tip in comparison to the control mice fed on normal hard chow (Fig. 4). This suggests that the formation of restorative dentine is a continuous process rather than a repair process that is specifically stimulated by damage. Interestingly, mineral deposition after 4 days of soft diet exposure contained lacunae-like structures (Fig. 4C,C’) that was morphologically distinct from those fed on a soft diet for a shorter period of 1 day, which was less intricately developed and more granulated in appearance (Fig. 4B,B’).

Figure 5.

Dspp expression at incisor tips. H&E sections of the tips indicate very disordered mineral matrix in the pulp cavity which contained embedded cells and remnants of coagulated blood cells (arrows in A, B). In situ hybridisation showed Dspp expression in odontoblast-like cells at the tips, Dspp positive cells were noticeably embedded within the mineralised tissue at the tips (C, D). Scale bars indicate 100µm (A, C) and 50µm (B, D).

Figure 4.

The incisal tip niche in CD1 adult mice. CD1 mice were exposed to different feeding regimes including control ordinary hard pellet diet, 1 day and 4 days on soft mash diet. Incisor tip mineral of the mice on normal hard chow are shown in panels (A and A’). Both the incisor tips of the mice on the 4 days and 1 day soft diet contained irregular mineral indicated by the arrows in B and C in comparison to A. Upon higher magnification, the mineral produced after 4 days on the soft diet appears to contain lacunae structures and cells (C’) compared to 1 day (B’) where the mineral appears more granulated. Scale bars indicate 200µm (A,B,C) and 50µm (A’, B’, C’). Abbreviation d: dentine

Odontoblast differentiation at the incisor tip

Sections of adult mouse incisors through the distal most tips showed that rather than the homeogenous soft tissue of the pulp, the pulp cavity was filled with disorganised cellular and mineralised material (Fig. 5A). The mineral superficially resembled dentine but was less mineralised and did not contain any tubules. Moreover cells were embedded within the mineral and thus this tissue resembled osteodentine that has previously been described [18]. Obvious coagulation of red blood cells was also clearly visible at the tips (arrows in Fig. 5A,B).

Odontoblasts, the only cell type known to make dentine, are neural crest-derived cells that express the dentin sialoprotein gene (DSPP) during their maturation [19–21]. Expression of DSPP using in situ hybridisation was therefore used to detect expression in cells at the incisor tip. DSPP expression was observed in incisor odontoblasts but not in pulp cells of the tooth body. At the incisor tip extensive expression of DSPP was observed, especially in the cells embedded in the newly synthesized mineralised tissue (Fig. 5C,D).

When mouse incisors are experimentally damaged, NG2-expressing perivascular cells (pericytes) differentiate into odontoblast-like cells to facilitate the production of reparative dentine [22]. A similar process is believed to occur during human tooth damage where cells expressing perivascular markers have been suggested to act as mesenchymal stem cells [7]. Using the same NG2-cre-mediated recombination as previously we crossed these mice with the Z/EG reporter mice where recombination results in expression of enhanced green fluorescent protein upon Cre-mediated excision [23]. During normal incisor growth, labeled cells are sparse in the tooth pulp, being predominantly associated with blood vessels (Fig. 6A). However at the tips of the incisors, there was a large accumulation of labeled cells in the pulp (Fig. 6B). Significantly, some of these labeled cells, particularly those located more peripherally, had an obvious columnar shape, consistent with an odontoblast phenotype (Fig. 6C). To confirm the NG2 results we used another marker often used to identify pericytes, namely Nestin. In the incisor, Nestin is expressed in odontoblasts and and pericytes [24–26]. However, NestinCre; R26R reporter mice did not show any expression in odontoblasts (my figure 6E shows some LacZ+ remnants of odontoblast processes), we therefore used this to specifically identify pericytes close to the incisor tips. Following x-gal staining of cryosections, it is clear that blue lacZ+ve pericyte-derived cells were in abundance at the tip of the incisors and also appear visibly embedded within the irregular tip mineral at the apex of the pulp chamber (Fig. 6D,E). We observed blood vessels with Nestin positive cells and a large accumulation of cells in the pulp at the tip (Fig. 6F). This accumulation of perivascular–derived cells in the mineralising tissue area at the tips of incisors, is consistent with previous evidence of the perivascular origin of mesenchymal cells (MSCs) in the pulp that can differentiate into odontoblast-like cells during dentine repair [22]. The accumulation of pericyte–derived cells at the tip most likely occurred as a result of increased proliferation as indicated by BrdU incorporation after a 24 hour chase period (Fig. S1).

Figure 6.

Perictye contribution to the incisor tip niche

In the NG2cre;Z/EG adult incisor, GFP labelled cells of the pulp body are exclusively associated with vasculature representing typical pericyte morphology (A). In contrast, an abundance of NG2 positive pericyte-derived cells is present at the incisor tip pulp chamber (B). More specifically, at higher magnification, genetically labelled NG2+ve pericytes appear consistent in columnar odontoblast morphology (arrows in C). Genetically labelled Nestin+ve cells in Nestincre;R26R lacZ stained incisors indicate Nestin+ve cells visibly embedded beneath the mineralised tip (arrows in D). A pool of LacZ+ve pericyte cells is located in the apical end of the pulp mesenchyme, notice the lacZ+ve dentinal tubules suggestive of pericyte contribution to odontoblast differentiation (E). Furthermore, accumulation of Nestin+ve cells is evident even at early postnatal stage (P2). Scale bars indicate 100μm (A), 50μm (B, C), 150μm (D, E) and 250μm (F). Abbreviations d: dentine, pm: pulp mesenchyme

Discussion

The formation of a dentin bridge composed of reparative dentine following dentine/pulp damage acts to seal the exposed pulp from the external environment, preventing both blood loss and infection [27, 28]. This fundamental natural tooth repair process remains poorly understood although the discovery of cells with mesenchymal stem cell-like properties in tooth pulp has prompted further investigations of the mechanisms of reparative dentine formation [29]. Reparative dentine does not have the structure or same mineral composition as tubular dentine and is often described as “osteodentine” since it is a hybrid material that can contain embedded cells reminiscent of osteocytes in bone [30–32]. The likely reason for the differences between dentine and reparative dentine is the speed with which is the latter is made and the fact that the majority of odontoblasts involved do not possess the usual columnar shape. This material needs to be synthesised rapidly following pulp exposure to ensure a seal is formed quickly. Reparative dentine can thus be considered to be the tooth equivalent of a dermal scar in wound healing.

The cells in tooth pulp that have MSC-like properties that can respond to damage by differentiating into odontoblast-like cells almost certainly include cells of perivascular origin. Perivascular cells (pericytes) share many surface proteins in vivo with MSCs in vitro [33–35]. Cre-mediated genetic lineage tracing of pericytes during tooth damage shows that these cells can give rise to odontoblasts during tooth repair [22]. These results prompted us to investigate an example of continuous dentine formation during normal homeostasis in mouse incisors where the shearing action at the tips continuously abrades the dentine and exposes the pulp. We observed very rapid deposition of a dentine-like material in the distal-most region of the pulp cavity in adult mouse incisors. This dentine contained embedded cells and resembled what has been called osteodentine previously described in the rat incisor [18]. Raman spectroscopy analysis showed this tissue to be dentine-like, but less mineralised and likely composed of small, highly crystalline, defect-free particles.

The cells embedded in this pulp dentine and surrounding it expressed high levels of DSPP and although most had a round shape, some adopted the classic columnar shape of odontoblasts. This suggests that the cells making this reparative dentine are immature odontoblasts, perhaps indicative of their rapid formation. A large accumulation of pericyte-derived cells was also observed in the pulp at the incisor tips. This suggests that the source of progenitor cells likely to give rise to the “secondary odontoblast-like” cells, belong to the vascular-derived pericytes which is perhaps unsurprising given that compelling evidence suggests that pericytes act not only as generic sources of MSCs [36] but have also been demonstrated more specifically to contribute to both tooth growth and repair [22]. Mouse incisors thus exhibit a form of tissue homeostasis where wear is perfectly counterbalanced balanced by growth. Part of this process involves the mineralisation of pulp cells exposed at the tip during wear to protect the tooth from infection.

The nature of the signalling processes that mediate MSCs within the incisal tip niche warrants further investigation. However, the combination of morphological, histological, Raman microspectroscopy as well as lineage tracing data thus far suggests that the incisal tip represents a specialised niche devoted to constant restoration of a “mineral plug” to defend the pulp from damage and infection. To distinguish this from other types of dentine made during tooth repair, we have termed this restorative dentine.

Importantly our proposed incisal tip niche contained a pericyte contribution demonstrated by a resident Nestin and NG2 positive population in the pulp. This novel finding supports the notion that pericytes mainly function in injury repair as we argue that the tip of the incisors undergo a form of natural, consistent “injury” through abrasion. This work thus provides an innovative perspective to investigating different injury repair processes and highlights the diverse applications of the mouse incisor model in studying different MSC populations.

Supplementary Material

Supp Fig. 1

Proliferation at the incisor tip. Wild-type adult mice were injected with BrdU and collected 24 hours post injection to label rapidly dividing cells. BrdU labelled odontoblast-like cells were present at the incisor tips (A), while limited proliferating cells were observed in the pulp body (B). Scale bar represents 50µm (A) and 100µm (B).