In order to develop a comprehensive understanding of how PPi regulates tooth root development, we performed a detailed histological study of developing first mandibular molars and incisors of mice harboring homozygous knock-out (−/−) of Alpl (high PPi), Ank, or Enpp1 (low PPi), compared to age-matched homozygous wild-type (+/+) controls. Days were selected to capture developmental time points of interest during molar root formation, i.e., during acellular cementogenesis (14 days postnatal, dpn), at completion of the root and following cellular cementogenesis (26 dpn), and after more than a month in occlusion (60 dpn). Alpl−/− mice were limited to a maximum age of 21 dpn because of shortened lifespan. Morphological observations on H&E stained sections were paired with in situ hybridization (ISH) and immunohistochemistry (IHC) for selected mineralized tissue-associated factors.
Acellular cementogenesis requires diminution of pyrophosphate
In the infantile form of HPP, the skeleton is properly mineralized at birth, but postnatal skeletogenesis is compromised 
mice phenocopy aspects of infantile HPP, where loss of TNAP was previously reported to have little effect on bone until postnatal day 6 
. At 14 dpn, the majority of alveolar and mandibular bone in Alpl−/−
mice was well developed, though signs of hyperosteoidosis were noted in the bone adjacent to the molar root (). In Alpl+/+
molars, acellular cementum (AEFC) covered the root dentin as a thin and uniform basophilic layer. Alpl−/−
molars were marked by disruption of acellular cementum, visible as reduction of the basophilic layer (cementum aplasia or severe hypoplasia) and direct contact of PDL cells and tissues with dentin. By 21 dpn this cementum defect was sometimes associated with tearing at the PDL-AEFC interface, suggesting poor integration of Sharpey's fibers at the root surface (not seen at the PDL-bone interface) () and consistent with HPP case reports observing premature tooth exfoliation. This is not likely to be a processing artifact, as infiltrating cells were present in the tear zone. These results agree with AEFC disruption described in this Alpl−/−
, as well as a different TNAP loss-of-function mouse 
Acellular cementogenesis requires diminution of pyrophosphate.
To further investigate the mechanism for the cementum defect in Alpl−/− mice, IHC was performed for two cementum markers, extracellular matrix (ECM) proteins bone sialoprotein (BSP) and osteopontin (OPN), which are present at high concentrations in acellular cementum of controls (). Both BSP and OPN immune localization were disrupted on the Alpl−/− root surface (), compared to the strong, even staining on Alpl+/+ controls. Scanning electron microscopy (SEM) provided improved resolution to explore the root surface. While Alpl+/+ molars displayed a cementum layer on the root dentin surface, this layer was absent in the Alpl−/− molar (). The disruption of cementum initiation and concomitant lack of BSP and OPN localization supports the hypothesis that high PPi in Alpl−/− is acting to inhibit cementogenesis and HAP apposition on the root surface.
Lack of acellular cementum on Alpl−/− molar root surfaces.
Attenuation of pyrophosphate increases acellular cementum
Both Ank and Enpp1−/− mice are deficient in extracellular PPi, though by different mechanisms. In molars of both null mice at 14 dpn, the developing cervical cementum was expanded (hypercementosis) compared to Ank and Enpp1+/+ controls (). At the completion of root development at 26 dpn, both Ank and Enpp1−/− molars featured a nearly identical cementum phenotype where cervical cementum width was expanded several fold over controls (). This thick cervical cementum included numerous cell inclusions in the matrix, in a region that is typically acellular type cementum (AEFC). Intriguingly, for both homozygous knock-out models, apical cementum (CIFC) was not morphologically different from controls (), PDL space remained unmineralized, and dentin was not altered compared to Ank and Enpp1+/+ mice.
Attenuation of pyrophosphate increases acellular cementum.
The incisor in the mouse is divided into a (labial) crown analogue featuring enamel, and a (lingual) root analogue featuring strictly AEFC type cementum. Notably, histological changes in Ank and Enpp1−/− incisors paralleled those in molars, featuring expanded cementum (). Sagittal sections of the mandible allowed observation of all three molars. Loss of Ank affected all molars similarly, with thickened cementum evident on all root surfaces compared to controls (). The fact that acellular cementum on all murine teeth was similarly affected by reduced PPi supports this as a central molecular regulator of cementogenesis which is not tooth- or stage-specific in its influence.
Both Ank and Enpp1−/− mice featured a hypercementosis phenotype, indicating both PPi regulators function in controlling cementum formation. Comparative analysis between Ank and Enpp1−/− and their respective controls was accomplished by measuring the growth rate of cervical cementum over time. During early root formation between 14 and 26 dpn, Ank and Enpp1−/− molars featured at least 10-fold greater cementogenesis compared to controls (). Ank and Enpp1−/− cementum continued to increase at a rate of 0.2–0.7 µm/day from 26 to 60 dpn, while over the same period, controls featured tightly controlled apposition, growing at the much slower pace of 0.01–0.05 µm/day.
Increased cementum apposition in Ank and Enpp1−/− teeth.
While cementum was dramatically affected by loss of ANK or NPP1, dramatic changes in other tissues were not observed. Histomorphometry at age 26 dpn was performed to measure cross-sectional widths to determine if PDL and alveolar bone were affected. Cementum was significantly increased in both null models, with Ank−/−
at 14-fold and Enpp1−/−
at more than 13-fold the width of age-matched controls (). A direct comparison of the two homozygous knock-out models revealed that Ank−/−
featured slightly, but significantly, thicker cementum at the age sampled. Histomorphometry confirmed that PDL space was maintained in both null models, even significantly larger in Ank−/−
, despite exuberant cementogenesis. Alveolar bone on the lingual aspect tended towards reduced cross sectional dimension in both Ank
models, though the effect was not statistically significant as measured here. Tartrate resistant acid phosphatase (TRAP) staining confirmed increased numbers of osteoclast-like cells (TRAP positive, multinucleated) on the bone surface adjacent to the tooth root in Ank−/−
. A modeling/remodeling of bone away from the root provides a mechanism for maintenance of the PDL in the face of expanding cementum.
One of the key functional characteristics of the cervical cementum is the extrinsic nature of the collagen fibers, which serve to anchor the tooth to surrounding alveolar bone. Picrosirius red staining in association with polarized light microscopy was used to visualize the birefringent collagen fibers of the periodontia 
. The thick cementum of Ank
molars featured a high concentration of extrinsic collagen fibers, which were continuous with the fibers in the PDL proper (). As this thick cementum in the null molars features dense extrinsic collagen fibers, but also contains numerous cell inclusions, it could properly be labeled cellular extrinsic fiber cementum (CEFC), a form of cementum not typical for cervical molar roots, and furthermore, not previously described in the cementum family. Importantly, the observation of an ongoing, progressive apposition on the root surfaces of Ank
mice confirms this is thickening of the normally present extrinsic fiber cementum, and is not likely to be a different type of ectopic calcification on the root surface. As a comparison, Alpl−/−
molars were examined, and confirmed tearing at the root-PDL interface, osteoid invasion of the PDL space, and poorly organized and sparsely embedded collagen fibers at the cervical root ().
Progressive mineralization of extrinsic collagen fibers in Ank and Enpp1−/− cervical cementum.
Cementum, bone, and dentin are also characterized by their extracellular matrix (ECM) protein composition, and these ECM proteins contribute to crystal growth and regulation, and affect mechanical properties of these tissues. Because of the dramatic changes in cementum apposition, we investigated the ECM profile in PPi deficient mice. In the low PPi environment of the Ank and Enpp1−/− mice, the thick cervical cementum was marked by increased OPN and dentin matrix protein 1 (DMP1), proteins of the SIBLING family (). OPN staining strongly labeled control acellular cementum, and was intensely expressed in the corresponding Ank and Enpp1−/− cervical cementum and associated cementoblast cells. DMP1, a marker for osteocytes, odontoblasts, and cementocytes, was present at low or undetectable levels in acellular cementum in controls, in contrast to intense localization in expanded Ank and Enpp1−/− cementum. OPN and DMP1 levels were not changed in Ank or Enpp1−/− apical cementum, as well as in other dentoalveolar locations. The source of the increased OPN and DMP1 protein was confirmed, as cementoblast gene expression for both Opn and Dmp1 mRNA was increased in Ank and Enpp
−/− mice (). OPN and DMP1 expression changes were not observed in other cell populations in the dentoalveolar complex in these mice. Another characteristic marker for cementum, BSP, was present in control and null cementum (), and where protein concentration was diluted in the larger cementum volume of the Ank and Enpp1−/− mice, mRNA levels in cementoblasts were unaltered ().
Reduced pyrophosphate alters acellular cementum matrix composition.
Reduced pyrophosphate alters gene expression in cervical cementoblasts.
Thus, increased cementogenesis in Ank
teeth was linked to increased OPN and DMP1 specifically in cervical cementum. It is notable that OPN was increased in cementum as a result of reduced extracellular PPi
. This change is opposite to the decreased OPN that has been documented in osteoblasts and articular locations in mice lacking ANK or NPP1 
Cementoblasts express pyrophosphate regulators in a time and space restricted manner
Acellular cementum was shown to be exceptionally sensitive to regulation by PPi; with increased PPi (as in Alpl−/− mice) AEFC was severely inhibited, and under reduced PPi conditions (as in Ank and Enpp1−/− mice) cementum thickness increased significantly, a trend not reflected in other dental hard tissues. In order to understand the sensitivity of acellular cementum to PPi metabolism, we mapped the expression of TNAP, ANK, and NPP1 during tooth root formation. We also assayed these factors in all of the null models to determine if there were compensatory or antagonistic expression changes that would contribute to phenotypes under PPi dysregulation.
TNAP was widely expressed during molar root formation, most strongly in mineralizing osteoblasts, odontoblasts, and cementoblasts (). As previously reported, TNAP was also strongly localized to the PDL region 
. TNAP localization was not altered in developing Ank
Cementoblasts express pyrophosphate regulators in a time and space restricted manner.
We previously reported wide expression of ANK gene and protein in the tooth and supporting tissues 
, paralleling previous findings that ANK is expressed in several tissues system-wide 
. Using a refined immunohistochemistry technique, which allowed more sensitive identification of differential ANK protein localization, we discovered that after acellular cementum formed, ANK was labeled most intensely in cementoblasts lining the molar and incisor roots (). Developmental localization of NPP1 protein was similar to that of ANK, with most intense staining found in cementoblasts (). Both ANK and NPP1 stained weakly in other cells, including PDL cells, osteoblasts, and odontoblasts. Immunolocalization revealed compensatory up-regulation, where NPP1 was increased in Ank−/−
and ANK was increased in Enpp1−/−
(). Most interestingly, the observed increase was found only in cementoblasts, and not in other cell populations of the dentoalveolar region. These data suggested that ANK and NPP1 were differentially expressed by cementoblasts and employed to tightly regulate PPi
and developmental cementum apposition. However, it still remained unclear by what mechanism PPi
was controlling cementum apposition and ECM composition.
Pyrophosphate controls mineralization and coupled gene expression in cementoblast cultures
regulators ANK and NPP1 were preferentially expressed by cementoblasts after initiation of cementogenesis, and their expression was modulated under conditions of low extracellular PPi
and increased apposition. Expression levels of cementum ECM proteins OPN and DMP1 were also responsive to PPi
deficiency, reflecting the altered homeostasis of Pi
ratio or increased cementum apposition in Ank
. These data together suggested that cementoblasts associated with AEFC regulate PPi
as a means to tightly control the process of apposition and related gene expression. In vitro
experiments were performed to determine how these genes were regulated during mineral formation, and what potential role PPi
played in their regulation. Because of the technical obstacles in isolating and identifying primary cementoblasts, we opted to use an immortalized cementoblast cell line (OCCM.30) and modulate exogenously added PPi
. OCCM.30 cells were cultured in control media or mineralization media where 5 mM β-glycerophosphate (BGP) was added. BGP served as an organic Pi
source, mimicking similar sources in vivo
and commonly used for in vitro
mineralization experiments 
. Cells receiving control media lacking BGP failed to mineralize during the course of the experiment. While cells cultured with BGP produced mineral nodules by day 6, with increased staining and calcium incorporation at day 8 ().
Pyrophosphate regulates cementoblast mineralization and nucleotide pyrophosphohydrolase (NTPPPH) activity, in vitro.
Cells were introduced to exogenous PPi
to create culture conditions of low (10 µM) and high (100 µM) PPi
. The lower dose of 10 µM PPi
did not affect mineralization, while the higher dose of 100 µM was confirmed as an inhibitor of mineral nodule formation under these conditions. While PPi
is an inhibitor of HAP crystal precipitation, it has also been reported to have cell signaling effects in osteoblasts 
. Neither dose of PPi
affected OCCM.30 cell proliferation, viability, or collagen synthesis compared to controls (), therefore these processes were not indirectly affecting mineralization. Cementoblast ALP enzyme activity was uniform across treatments and times, and added 100 µM PPi
did not appreciably affect ALP (), indicating the effect of PPi
on mineralization was not by inhibition of TNAP. An enzymatic assay for 5′-nucleotide phosphodiesterase I and nucleotide pyrophosphohydrolase (NTPPPH) activity demonstrated significantly increased NPP1 function with mineralization at days 4, 6, and 8, while 100 µM PPi
brought activity back to basal levels of non-mineralizing cultures ().
Pyrophosphate does not affect cementoblast proliferation or collagen synthesis, in vitro.
PPi associated and cementoblast marker genes were assayed by quantitative PCR. Under non-mineralizing conditions, Ank, Enpp1, Opn, and Dmp1 did not change over the course of the experiment (). However, all four genes increased significantly under mineralizing conditions at days 3 and 5, when mineral nodules were forming. At day 3, when increases were most dramatic, Ank increased almost 10-fold, Enpp1 increased 30-fold, Opn increased more than 30-fold, and Dmp1 increased 140-fold in mineralizing cultures compared to controls. These four genes also responded in parallel fashion to PPi. While inclusion of 10 µM PPi had a mild effect on gene expression compared to ascorbic acid (AA) + BGP cultures (paralleling effects on mineralization), the higher dose of 100 µM PPi significantly depressed Ank, Enpp1, Opn, and Dmp1 expression on day 3 compared to mineralizing cells. Cells receiving the 100 µM dose also maintained significantly lower expression of Ank, Enpp1, and Opn on day 5. By day 7, expression levels of the four genes were low, and there were no differences between any of the treatment conditions. Other cementoblast marker genes assayed, including Alpl, Bsp, and Col1, did not show a coherent pattern in response to addition of PPi. Notably, the increase in Enpp1 gene expression associated with mineralization corresponds to the increase in NTPPPHase activity recorded, and inclusion of PPi decreased mineralization and correspondingly decreased Enpp1 gene expression and NPP1 enzyme activity. In contrast, PPi did not perturb cementoblast mineralization by affecting Alpl expression or ALP activity.
Pyrophosphate regulates cementoblast mineralization-coupled gene expression.
These results showed that PPi regulated cementoblast mineralization and associated gene expression, in vitro. In an additional experiment of similar design, the addition of 100 µM PPi was discontinued in some wells midway through the experiment. Cells with 100 µM PPi for the duration did not mineralize, while cultures relieved of PPi inhibition at day 4 showed mineralization by day 6, increased Ank, Enpp1, Opn, and Dmp1 by day 5, coincident with mineralization (). This experiment demonstrated that even if PPi inhibited initiation of mineralization for the first 4 days, its removal facilitated both mineralization and concomitant gene expression. These results support expression of Ank, Enpp1, Opn, and Dmp1 as being functionally coupled to matrix mineralization, i.e. linked to changes in the mineralizing matrix. Importantly, these results parallel in vivo observations, where ANK, NPP1, OPN, and DMP1 were all increased by cementoblasts under conditions of reduced PPi, i.e. Ank or Enpp1 ablation.
Timing of pyrophosphate removal determines cementoblast mineralization and coordinated gene expression in vitro.