Dentoalveolar Defects and Impaired Alveolar Bone Healing in a Neural Crest Directed Conditional Knockout Mouse Model of Hypophosphatasia

Abstract

Hypophosphatasia (HPP) is an inherited error-of-metabolism caused by loss-of-function mutations in ALPL-encoded tissue-nonspecific alkaline phosphatase (TNAP). HPP has wide-ranging severity, including a clinical subtype called odontohypophosphatasia (odonto HPP), which selectively affects craniofacial structures. Dentoalveolar defects in HPP can affect enamel, dentin, and alveolar bone, and deficient acellular cementum contributes to tooth loss. Global Alpl knockout phenocopies effects of severe HPP, but early lethality precludes longer-term studies. Aiming to create a mouse model replicating dentoalveolar effects of HPP, we used Wnt1^Cre2 mice to conditionally delete Alpl in ectomesenchymal cells that make dentin, cementum, periodontal ligament (PDL), and alveolar bone. We compared appendicular and craniofacial skeletal effects of Wnt1^Cre2 to Prx1^Cre conditional Alpl ablation in limb bud mesenchyme. We also tested alveolar bone socket healing in Wnt1^Cre2; Alpl^fl/fl conditional knockout mice and the effect of TNAP-Fc-D₁₀ enzyme replacement therapy (ERT) on socket healing. Prx1^Cre; Alpl^fl/fl mice exhibited 38% reduced circulating alkaline phosphatase (ALP) and long bone defects, but no craniofacial phenotypes. Wnt1^Cre2; Alpl^fl/fl mice featured 60% reduced ALP and profound mineralization defects in dentin, cementum, and alveolar bone, but no appendicular skeleton changes. Defects were noted in neural crest-derived intersphenoid synchondrosis of the cranial base and mandibular condyle of Wnt1^Cre2; Alpl^fl/fl mice. Extraction of maxillary molars in Wnt1^Cre2; Alpl^fl/fl mice revealed profound alveolar bone healing defects that were partially rescued by ERT. Cranial neural crest deletion of Alpl resulted in a mouse model phenocopying odonto HPP that can be used to investigate mechanisms underlying pathologies as well as interventions.

1. Introduction

Hypophosphatasia (HPP) is an inherited error-of-metabolism that causes a wide range of skeletal manifestations, including rickets, growth plate defects, and osteomalacia [1, 2]. HPP is caused by loss-of-function mutations in ALPL, which encodes tissue-nonspecific alkaline phosphatase (TNAP). TNAP is an ectoenzyme expressed in bones, teeth, liver, and kidney [3]. TNAP hydrolyzes inorganic pyrophosphate (PP_i), a circulating inhibitor of mineralization; in HPP, PP_i levels are pathologically increased because of reduced TNAP function. HPP has wide-ranging severity described as a spectrum of clinical subtypes (OMIM#241500, 136300, 146300) [4]; the most severe ones are the early-onset perinatal and infantile forms associated with pyridoxine-dependent seizures. Subtypes that emerge later in childhood and adulthood are less severe but contribute to a substantial burden of disease [5–7]. One subtype called odontohypophosphatasia (odonto HPP) results from typically autosomal dominant ALPL variants and selectively affects the deciduous and permanent dentition, with some reports of craniofacial effects, and with minimal to no appendicular skeletal involvement [1, 2, 8–10]. Dental manifestations across HPP subtypes can include enamel defects, thin and/or hypomineralized dentin, wide pulp chambers, tooth root abnormalities, reduced or absent acellular cementum, periodontal ligament (PDL) detachment, and alveolar bone defects [8, 10–12]. Premature loss of deciduous, fully rooted teeth due to cementum defects is pathognomic for HPP.

There remain many unanswered questions about HPP effects on dentoalveolar tissues, in part due to limitations of animal models used to date. The global Alpl knockout (Alpl^−/−) mouse phenocopies the severe, infantile HPP subtype and has been indispensable for studies of disease mechanisms and therapies [1, 13]. However, Alpl^−/− mice experience seizures and die by weaning age, precluding longer-term studies required to analyze dentoalveolar development, function, and treatments [13]. Additional HPP mouse models have been created. A mouse with a knock-in A116T Alpl mutation associated with odonto HPP exhibited a very mild dental phenotype of slightly thinner cementum and small regions of osteoid accumulation in alveolar bone, but defects were too mild to be useful for continued experiments on pathology or therapeutic interventions [9, 14]. Conditional (i.e. tissue-selective) gene deletion can be a successful strategy to phenocopy aspects of human disease in mice. We previously described a mouse carrying a floxed Alpl allele (Alpl^fl/fl) allowing conditional deletion of Alpl when crossed with mice harboring Cre recombinase [15]. Proof-of-principle studies using Prx1^Cre (targeting limb bud mesenchyme) or Col1a1^Cre (broadly targeting bone and dental tissues) confirmed that later-onset models of HPP could mimic biochemical, skeletal, and dental aspects of HPP. These conditional knockout models carried substantial axial and appendicular skeletal defects, though did not suffer seizures and lived to the end of the study at 6 months. This established the Alpl floxed mouse as a potential tool to employ to better understand craniofacial and dentoalveolar manifestations of HPP, and especially to test therapeutic strategies that may impact oral health.

We aimed to use the conditional knockout approach to create a mouse model replicating the primarily dentoalveolar effects of odonto HPP without the appendicular skeletal burden of disease. To that end, we selected Wnt1^Cre2 mice to target cranial neural crest cells, which give rise to ectomesenchymal cells that contribute to anterior craniofacial tissues and make dentin, cementum, PDL, and alveolar bone [16, 17]. We compared appendicular and craniofacial skeletal effects of Wnt1^Cre2 to Prx1^Cre conditional Alpl ablation, to understand tissue-limited effects of conditional knockout on circulating ALP and local developmental changes in mineralized tissue. We hypothesized that mice conditionally ablated for Alpl in cranial neural crest cells would phenocopy dentoalveolar aspects of HPP in the absence of skeletal effects. Tooth loss is an extremely common consequence of HPP. Currently, there are few case reports and no consensus on effects of tooth loss on alveolar bone healing and impact on placement of dental implants in those with HPP [10, 11, 18, 19]. Therefore, we also aimed to test whether alveolar bone socket healing was affected in this mouse model of HPP and determine if enzyme replacement therapy would improve socket healing.

2. Materials and Methods

2.1. Mice

Animal procedures were approved by the Institutional Animal Care and Use Committee at The Ohio State University (Columbus, OH, USA). Mice carrying the Alpl^fl/fl allele were previously described on a C57BL/6 genetic background [15]. Here we report all Alpl conditional knockouts on a 129 genetic background; genetic background was determined to be important for phenotype of Alpl knockout and conditional knockout mice [20]. Prx1^Cre [B6.Cg-Tg(Prrx1-cre)1Cjt/J; stock no. 005584] and Wnt1^Cre2 [B6.Cg-E2f1Tg(Wnt1-cre)2Sor/J; stock no. 022501] mouse strains were acquired from The Jackson Laboratory (Bar Harbor, ME, USA) [16, 21]. Mice were backcrossed to a 129X1/SvJ genetic background (The Jackson Laboratory; stock no: 000691) for at least 5 generations before generating experimental animals for analysis. To generate limb bud mesenchyme selective Alpl deletion, male Prx1^Cre; Alpl^fl/+ mice were crossed with female Alpl^fl/+ mice to produce Prx1^Cre; Alpl^fl/fl conditional knockout mice (also referred to as cKO). Alpl^fl/fl mice, which display no phenotype, were used as healthy controls (CTR) in experiments, and both male and female mice were analyzed and grouped, unless otherwise stated. To generate neural crest (craniofacial and dental ectomesenchyme) selective Alpl deletion, female Wnt1^Cre2; Alpl^fl/+ mice were crossed with male Alpl^fl/+ mice to produce Wnt1^Cre2; Alpl^fl/fl conditional knockout mice (also referred to as cKO). Mice were euthanized at 60 days postnatal (dpn) to study developmental defects at the young adult stage, and tissues were fixed in 10% neutral buffered formalin. Genotyping is shown in Supplemental Figure S1. No unexpected adverse events occurred, no mice or data points were excluded from the study, and confounders were not controlled. Animals were coded to lab personnel throughout analysis.

2.2. Blood Biochemistry

Mice were anesthetized using intraperitoneal administration of ketamine/xylazine (10 mg/1 mg/kg). Blood was collected using cardiac puncture with 25G needle attached to sterile BD 1 ml syringe into BD microtainer blood collection tubes with BD SST, Gold. Blood samples were incubated at room temperature for 30 minutes and centrifuged at 15,000x g for 2 minutes. Serum was collected and aliquoted into fresh tubes and frozen until future use. Serum alkaline phosphatase (ALP) activity was measured using an enzymatic assay at the Goss laboratory at The Ohio State University College of Veterinary Medicine (Columbus, OH, USA).

2.3. Tooth Extraction

Alveolar bone socket healing was compared in 42 dpn Wnt1^Cre2; Alpl^fl/+ vs. control littermate control mice (n=5–7 mice/genotype, including both males and females). Maxillary first molar (M1) extraction was performed as previously described [22, 23]. Mice were anesthetized by nasal isoflurane and bilateral extraction of maxillary M1 was accomplished using a #2 dental explorer under a surgical microscope. Following extraction, mice were subcutaneously administered 5 mg/kg carprofen to alleviate pain every 24 hours for 2 days and rehoused in original cages. Mice were euthanized at 21 days post-procedure (dpp) and tissues were fixed in 10% neutral buffered formalin. For treatment, mice were subcutaneously injected with 8.2 mg/kg TNAP-Fc-D₁₀ or saline (control) starting from 2 days prior to molar extraction and every other day until the end of the study at 21 dpp (n=4/group), at approximately 63 dpn at young adult stage. Assignment of mice to treatment or control group was done based on alternating assignments until n=4/group was achieved.

2.4. Micro-computed tomography

Right hemi-mandibles, hemi-maxillae with healing extraction sockets, and left femurs were scanned using a μCT 50 scanner (Scanco Medical, Bassersdorf, Switzerland) at 70 kV, 85 μA, 0.5 mm Al filter, and 900 ms integration time. A voxel dimension of 6 μm was used for mandibles and maxillae and a voxel dimension of 10 μm was used for femurs. Hemi-mandibles were analyzed as previously described [24, 25]. Reconstructed images were calibrated to five known densities of hydroxyapatite and analyzed using Analyze 14.0 (version 1.0; AnalyzeDirect, Overland Park, KS, USA). For hemi-mandibles, a region of interest (ROI) was defined around the first mandibular molar (M1) and associated alveolar bone flanking between 240 μm mesial to the mesial root for M1 and 240 μm distal to the distal root of M1. Enamel was segmented above 1,600 mg HA/cm³. Alveolar bone and dentin/cementum were segmented at 650–1,600 mg HA/cm³.

We previously used a sample registration approach for analysis of new bone formation in the mouse socket healing model [22, 23]. Briefly, reconstructed maxilla images were oriented to a standard orientation and were then registered to a standard image to automate orientation and ensure very close 3D alignment between samples. To define the ROI for measurement of bone healing, sockets of CTR and cKO mice at 42 dpn (0 days post procedure, i.e. the same day as molar extraction) were traced independently and used to generate an average socket map for each genotype. Genotypes were analyzed separately because of differences in alveolar bone that exist by 42 dpn. This average socket region was then eroded by 120 μm to minimize inclusion of preexisting lamina dura and localize the ROI entirely within the empty socket where only new bone formation occurs after tooth extraction. This ROI was mapped onto oriented maxillae of experimental mice and all tissue within the ROI above 650 mg/cm³ HA was defined as bone. This traced ROI was used to calculate bone volume fraction (BV/TV), bone mineral density (BMD), and tissue mineral density (TMD) for new bone [25, 26].

For femurs, trabecular and cortical bones were segmented at 350 and 650 mg HA/cm³, respectively. The trabecular bone was traced using 50 slices (total of 0.5 mm) proximal to the distal femur growth plate to quantify bone volume (BV), total volume (TV), bone volume fraction (Tb.BV/TV), trabecular number (Tb.N), thickness (Tb.Th), spacing (Tb.Sp), connectivity density (1/mm³), and mineral density (Tb.BMD). For the cortical bone, 50 slices of the mid-femur of each bone were used to quantify cortical bone volume fraction (Ct.BV/TV), cortical thickness (Ct.Th), mineral density (Ct.BMD), marrow area (Ma.Ar), cortical porosity (Ct.Po), porosity number (Po.N), porosity volume (Po.V).

For skulls, micro-CT scanning was performed using SkyScan 1172 (Bruker Micro-CT, Kontich, Belgium) scanner with specific parameters set to 55 kV, 181 μA, 0.5 mm Al filter, 280 ms integration time, and 10 μm voxel dimension. The micro-CT images were reconstructed using NRecon software and calibrated to three known densities of hydroxyapatite. The data were then analyzed with Analyze 14.0 (version 1.0; AnalyzeDirect, Overland Park, KS, USA). Skull orientation and linear measurements of skull length and width were performed as previously described in [27]. The intersphenoid synchondrosis (ISS) width was quantified using linear measurements of synchondrosis width at the most midline slice of the cranial base.

2.4. Histology

Left hemi-mandibles were fixed in Bouin’s solution, demineralized in a solution of 10% v/v glacial acetic acid, 0.4% v/v neutral-buffered formalin, and 0.85% w/v sodium chloride (AFS), and processed for paraffin embedding [28]. Histological coronal sections of mandibles at 60 dpn were stained with hematoxylin and eosin (H&E), following a standard protocol. Histological sagittal sections of mandibular condyles at 14 and 60 dpn were stained by H&E and Safranin O. Sections were stained with 0.1% Safranin O (Electron Microscopy Sciences, Hatfield, PA, USA) for 20 minutes and 0.02% Fast Green FCF (Electron Microscopy Sciences) for 5 minutes. Right femurs were fixed in Bouin’s solution at room temperature for 24 hours. Samples were decalcified using AFS solution for 3 weeks and processed for paraffin embedding. Samples were sectioned at 6 μm thickness and then deparaffinized and hydrated in a series of ethanol and distilled water. Femur sections were stained with Safranin O.

The analysis of acellular cementum and predentin was performed on H&E-stained images captured with the same acquisition parameters as previously described in [24]. The values for acellular cementum and predentin represent the average of three linear measurements taken at 90 μm, 100 μm, and 110 μm measured apically from the cemento-enamel junction (CEJ) using the straight-line function in the ImageJ.

For immunohistochemistry (IHC), the following primary antibodies were used: Polyclonal rabbit anti-bone sialoprotein (BSP) IgG (1:200, Dr. Renny Franceschi, University of Michigan, Ann Arbor, MI, USA), polyclonal rabbit anti-osteopontin (OPN) IgG (LF-175; 1:200, Dr. Larry Fisher, NIDCR/NIH, Bethesda, MD, USA), and rabbit anti-dentin matrix protein 1 (DMP1) IgG (1:200; Takara Bio USA, San Jose, CA, USA).

TRAP staining was performed to visualize osteoclasts. TRAP was performed on demineralized, paraffin embedded histological sections of left mandibular molars. Tissue sections were deparaffinized and incubated with the provided staining cocktail from the TRAP staining kit (Sigma-Aldrich), following the manufacturer’s instructions. To quantify the osteoclasts, the alveolar bone surface’s perimeter was outlined using the freehand selection function in ImageJ. The number of osteoclasts per bone perimeter (N.OC/B.Pm) was manually counted and then normalized to the alveolar bone surface’s perimeter facing the M1 tooth.

2.5. Statistical Analysis

Data are displayed as mean ± standard deviation (SD). Statistical analysis was performed with GraphPad Prism 9 software (Version 9.5.1). Two-tailed Student’s t-test was used to compare means between genotypes and one-way ANOVA and post hoc Tukey test were used to compare three groups. A prior study found significant differences between groups of n=3 [15]. Power analysis predicted for α=0.05 and β=0.8 that n=3–4 would detect changes of 10% with SD=4–5%. Statistically significant differences are indicated as *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. ns: not significant.

3. Results

3.1. Appendicular Bone Defects from Limb Bud Conditional Deletion of Alpl

Previous studies confirmed that using Prx1^Cre mice to target Alpl deletion in limb bud mesenchyme resulted in reduced circulating ALP and skeletal defects, including hypomineralized long bones, deformations, and fractures [15]. We employed the Prx1^Cre; Alpl^fl/fl conditional knockout mice as a model of later-onset HPP and a counterpoint to the new Wnt1^Cre2; Alpl^fl/fl mice described below. These studies using the Prx1 driver were conducted for the first time on the 129 genetic background, which expresses a more penetrant HPP phenotype compared to the C57BL/6 genetic background. We also expanded the craniofacial analysis of Prx1^Cre; Alpl^fl/fl beyond the previous report to assess the utility of this line for studying effects on craniofacial tissues.

Compared to the Alpl^fl/fl control group, Prx1^Cre; Alpl^fl/fl mice showed no difference in body weight and a 38% decrease in ALP at 60 dpn (Supplemental Figure S2A, B). Craniofacial measurements showed no differences in skull length or width between the two groups (Supplemental Figure S2C, D). Micro-CT showed abnormal growth plates and reduced cortical thickness in femurs of Prx1^Cre; Alpl^fl/fl vs. control mice (Supplemental Figure S2E–F). Analyses of trabecular and cortical bone parameters revealed several significant differences in Prx1^Cre; Alpl^fl/fl vs. control mice, particularly reduced bone volume fraction for both trabecular (BV/TV) and cortical (Ct.BV/TV) bone, reduced trabecular number (Tb.N), and reduced cortical thickness (Ct.Th) (Supplemental Figure S2G). Histology using Safranin O staining of femur growth plates confirmed abnormal chondrogenesis and expanded and disorganized chondrocytes in Prx1^Cre; Alpl^fl/fl mice compared with controls (Supplemental Figure S3).

Micro-CT analyses revealed no deleterious alterations to mandibular first molar (M1) or incisor dentoalveolar structures of Prx1^Cre; Alpl^fl/fl vs. control mice (Supplemental Figure S4). Histology confirmed no apparent differences in periodontal organization between genotypes (Supplemental Figure S5). Identifying growth plate defects in Prx1^Cre; Alpl^fl/fl mouse femurs prompted us to examine cartilaginous structures in craniofacial tissues. However, we did not observe defects in either the cranial base synchondrosis (Supplemental Figure S6A–G) or the mandibular condyles (Supplemental Figure S6H–I).

3.2. Neural Crest Deletion of Alpl Causes Dentoalveolar Mineralization Defects

We next analyzed Wnt1^Cre2; Alpl^fl/fl mice on a 129 genetic background. Unlike Alpl^−/− mice, Wnt1^Cre2; Alpl^fl/fl conditional knockouts did not experience seizures, had comparable body weight to controls, and lived to the study endpoint of 60 dpn (Figure 1A). Wnt1^Cre2; Alpl^fl/fl mice had 60% decreased ALP levels and no differences in serum calcium, phosphorus, or PP_i (Figure 1B–E). Compared with controls, skulls of Wnt1^Cre2; Alpl^fl/fl mice were reduced in width (Figure 1F, G). No defects were observed in cortical or trabecular bone in femurs (Figure 1H–J).

Figure 1. Neural Crest Deletion of Alpl Reduces Circulating ALP But Does Not Cause Appendicular Bone Defects.

(A) Body weights are not different between Alpl^fl/fl (Controls shown by gray bars) and Wnt1^Cre2; Alpl^fl/fl conditional knockout mice (shown by red bars) at 60 dpn (n=6/group). (B-E) Biochemical results shows similar plasma alkaline phosphatase (ALP), calcium, phosphate, and inorganic pyrophosphate (PP_i) levels in Wnt1^Cre2; Alpl^fl/fl vs. control mice (n=5/group). (F) 3D micro-CT images of skulls from control and Wnt1^Cre2; Alpl^fl/fl mice at 60 dpn. Red and blue dotted lines indicate skull length and skull width, respectively. (G) Quantification of skull length and width reveals reduced width in Wnt1^Cre2; Alpl^fl/fl vs. control mice (n=3–4/group). (H, I) 3D and 2D micro-CT images of femurs from Alpl^fl/fl controls and Wnt1^Cre2; Alpl^fl/fl mice. (J) Quantitative micro-CT analysis of trabecular band cortical bone parameters in distal femurs from the two genotypes at 60 dpn (n=4–5/group). *p<0.05; ***p<0.001; ns: not significant. BV: Bone volume; TV: total volume; BV/TV: bone volume fraction; Tb.Th: trabecular thickness; Tb.Sp: trabecular spacing; Tb.N: trabecular number; Conn.D: trabecular connectivity; Tb.BMD: trabecular bone mineral density; Ct.Th: cortical thickness; Ct.BV/TV: cortical bone fraction; Ct. BMD: cortical bone mineral density; Ma.Ar: marrow.

Micro-CT analysis of M1 and surrounding periodontal tissues showed that Wnt1^Cre2; Alpl^fl/fl mice exhibited thin dentin, wide pulp chambers, and increased PDL space, compared with controls (Figure 2A, B). Dentin/cementum volume was decreased 10% and pulp volume was increased 49% in Wnt1^Cre2; Alpl^fl/fl vs. control mice (p<0.01 and p<0.0001, respectively) (Figure 2C). Alveolar bone showed significantly reduced volume and density in Wnt1^Cre2; Alpl^fl/fl vs. control mice (p<0.001 and p<0.05, respectively). Enamel, which is of epithelial origin and not derived from neural crest, was unaffected in molars. Incisors of Wnt1^Cre2; Alpl^fl/fl vs. control mice displayed both reduced enamel and dentin volumes (Figure 2D).

Figure 2. Neural Crest Deletion of Alpl Causes Dentoalveolar Mineralization Defects.

(A, B) 3D and 2D micro-CT renderings of mandibular molars (M1) at 60 dpn reveals dentoalveolar defects in Wnt1^Cre2; Alpl^fl/fl vs. Alpl^fl/fl control mice, including thin dentin (DE), widened pulp chambers (red *), and enlarged periodontal ligament (red arrows). (C) Quantitative micro-CT analysis of the M1 region indicates alterations in dentin, alveolar bone and pulp in Wnt1^Cre2; Alpl^fl/fl vs. controls (n=6/group). (D) Quantitative micro-CT analysis of mandibular incisor (INC) indicates reduced enamel and dentin volumes in Wnt1^Cre2; Alpl^fl/fl vs. controls (n=6/group). EN: enamel; AB: Alveolar bone; CC: cellular cementum. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns: not significant.

Histology revealed a number of alterations indicative of mineralized tissue defects in Wnt1^Cre2; Alpl^fl/fl mice vs. control mice. These included deficient acellular cementum, loss of PDL attachment, wide predentin, and accumulation of osteoid (Figure 3A–D). Compared to normal distribution of key markers in controls, immunostaining of Wnt1^Cre2; Alpl^fl/fl tissues revealed reduced BSP and OPN on root surfaces, in association with reduced cementum thickness (Figure 3E–H). BSP, OPN, and DMP1 all localized to alveolar bone in control tissues, but were reduced or absent in large regions of osteoid in alveolar bone of Wnt1^Cre2; Alpl^fl/fl mice (Figure 3E–H, I, K). TRAP staining indicated no differences in osteoclast numbers (Figure 3J, L, O). Histomorphometry confirmed predentin thickness was 4-fold increased and acellular cementum was nearly absent (p<0.0001 for both) (Figure 3M, N).

Figure 3. Periodontal Defects in Mice with Neural Crest Deletion of Alpl.

H&E staining reveals dentoalveolar mineralization defects in Wnt1^Cre2; Alpl^fl/fl vs. Alpl^fl/fl control mice. (A-D) Compared to controls, Wnt1^Cre2; Alpl^fl/fl mice exhibit widened predentin (PD; orange line in panel D), deficient acellular cementum (AC) (red asterisk in panel D), periodontal ligament (PDL) detachment, and accumulation of osteoid (* in panels G, H, and K) in alveolar bone. (E-H) Immunostaining for BSP and OPN demonstrates reduced AC on root surfaces of Wnt1^Cre2; Alpl^fl/fl vs. control mice. Accumulation of osteoid in Wnt1^Cre2; Alpl^fl/fl mice is indicated by altered localization of BSP, OPN, and (I, K) DMP1 in AB. (J, L) TRAP staining of osteoclasts on alveolar bone (AB) surfaces is similar between genotypes. (M-O) Histomorphometry confirms increased PD, decreased AC thickness, and similar density of osteoclasts (OC) per bone surface (B) perimeter (Pm). ****p<0.0001; ns: not significant.

3.3. Defective Chondrogenesis in the Cranial Base Synchondrosis and Mandibular Condyle in Mice with Neural Crest Deletion of Alpl

HPP can contribute to dysregulated craniofacial mineralization, leading to cranial abnormalities and craniosynostosis [1, 2]. These aspects have been described in the Alpl^−/− mouse, in a different model of neural crest Alpl deletion, and in knock-in sheep models of HPP [20, 29–31]. Cranial neural crest cells contribute significantly to the craniofacial skeleton during embryogenesis, including teeth, jawbones, condyles, and portions of the cranial base [32].

Micro-CT revealed abnormal ossification and chondrogenesis at the cranial base in Wnt1^Cre2; Alpl^fl/fl mice. The basisphenoid bone, shown to have neural crest as well as mesoderm contributions [33], had a porous appearance in Wnt1^Cre2; Alpl^fl/fl vs. control samples (Figure 4A, B). The mid-cranial base is formed by endochondral ossification at intersphenoid synchondrosis (ISS) anteriorly and spheno-occipital synchondrosis (SOS) posteriorly [27, 34]. Compared with controls, micro-CT analysis of Wnt1^Cre2; Alpl^fl/fl mutants showed irregularities in the ISS, including enlarged spaces between presphenoid and basisphenoid bones, along with instances of premature closure, particularly at the sides of the synchondrosis (Figure 4A–D). Conversely, no discernible defects were detected in the SOS region. Notably, one of the Wnt1^Cre2; Alpl^fl/fl skulls exhibited complete premature closure of ISS (Figure 4B, red asterisk). The ISS abnormality was quantified using linear measurements of synchondrosis width at the most midline slice of the cranial base; the prematurely closed ISS could not be measured and was not included in the graph (Figure 4E). Compared to normal chondrocyte arrangements within the resting zone, proliferative zone, and hypertrophic zone in Alpl^fl/fl control synchondroses, the ISS in Wnt1^Cre2; Alpl^fl/fl mice showed widening in the hypertrophic zone due to abnormal chondrocyte accumulation (Figure 4F, G).

Figure 4. Defective Chondrogenesis in the Cranial Base Synchondrosis in Mice with Neural Crest Deletion of Alpl.

(A-D) 3D and 2D micro-CT images of the cranial base from Alpl^fl/fl controls and Wnt1^Cre2; Alpl^fl/fl mice at 60 dpn. Red arrows point to irregularities of space within the intersphenoid synchondrosis (ISS). The red asterisk indicates closure of ISS. No defects are evident in the spheno-occipital synchondrosis (SOS). (E) Measurement of synchondroses reveals increased width of ISS in Wnt1^Cre2; Alpl^fl/fl vs. control mice but no differences between genotypes at the SOS (n=2–3/group). (F, G) H&E staining of cranial base synchondroses showed widening of hypertrophic zone (HZ) depicted by red lines within ISS of Wnt1^Cre2; Alpl^fl/fl vs. control mice. PS: Presphenoid; BS: Basiosphenoid; BO: Basiooccipital; RZ: resting chondrocyte zone; PZ: proliferative chondrocyte zone. *p<0.05; ns: not significant.

Cranial neural crest cells contribute to the condylar cartilage of the temporomandibular joint, an endochondral growth site and articulation of the mandible and cranium [32]. Micro-CT revealed dramatic defects in the subchondral bone in Wnt1^Cre2; Alpl^fl/fl mice compared with Alpl^fl/fl controls (Figure 5A–D). The subchondral bone of the condyle was abnormally enlarged and there was a central, unmineralized region in Wnt1^Cre2; Alpl^fl/fl vs. control mice. Histology of mandibular condyles at 14 dpn revealed no major differences between genotypes; however, by 60 dpn, Wnt1^Cre2; Alpl^fl/fl mice exhibited a dramatic disruption, disorganization, and abnormal accumulation of chondrocytes (Figure 5E, F). Chondrocytes showed growth into the subchondral bone region, explaining the unmineralized gap observed by micro-CT.

Figure 5. Defective Chondrogenesis in the Mandibular Condyle in Mice with Neural Crest Deletion of Alpl.

(A-D) 3D and 2D micro-CT images of mandibular condyles from 60 dpn Wnt1^Cre2; Alpl^fl/fl mice reveal abnormal, enlarged mandibular condyle with a large unmineralized zone in the subchondral bone (red arrowheads), compared with Alpl^fl/fl mice. (E, F) Histology of mandibular condyles at 14 dpn by H&E and Safranin O staining reveals no major differences between genotypes. By 60 dpn, Wnt1^Cre2; Alpl^fl/fl mice exhibit a disorganized overgrowth of the condylar cartilage chondrocytes deep into the subchondral bone (red arrowheads). CC=condylar cartilage; SB-subchondral bone; R=ramus.

3.4. Neural Crest Deletion of Alpl Reduces Alveolar Bone Socket Healing

Deciduous and permanent tooth loss are common consequences of HPP [1, 2, 5, 10, 11]. However, there is no consensus on whether alveolar bone can heal properly and successfully support dental implant placement in affected individuals. Early lethality in Alpl^−/− mice prevents long-term studies of tooth attachment, retention, or effects of HPP on bone healing [13]. As Wnt1^Cre2; Alpl^fl/fl mice replicate alveolar bone hypomineralization defects in a longer lived mouse model, we aimed to study socket healing in these mice using a model of maxillary molar extraction [22, 35, 36]. This represents the first study of bone healing in an animal model of HPP. At 21 days post-procedure, Wnt1^Cre2; Alpl^fl/fl mice exhibited dramatically reduced bone healing compared to controls (Figure 6A–D). This resulted in greater than 50% reduction in bone volume fraction (BV/TV), and significant reductions in both bone and tissue mineral density (BMD, TMD; p<0.001 for all) (Figure 6E).

Figure 6. Alveolar Bone Socket Healing Defects in Mice Lacking Neural Crest Expression of Alpl.

(A-D) 3D (occlusal and buccal views) and 2D (transverse and sagittal views) images of alveolar bone (AB) healing at 21 days post-extraction (dpe). Red color in 3D images highlights new AB within sockets. Yellow dotted lines in 2D images indicate estimated location of original socket. Micro-CT analysis shows reduced socket healing in Wnt1^Cre2; Alpl^fl/fl mice compared with Alpl^fl/fl controls (n=5–7/group). (E) Quantification shows reduced bone volume fraction (BV/TV), bone mineral density (BMD), and tissue mineral density (TMD) in Wnt1^Cre2; Alpl^fl/fl vs. control mice. ***p<0.001.

After establishing the alveolar bone socket healing defect in Wnt1^Cre2; Alpl^fl/fl mice, we aimed to provide ERT to determine whether intervention had the potential to improve bone healing. Subcutaneous delivery of 8.2 mg/kg TNAP-Fc-D₁₀ ERT twice weekly starting 2 days before extraction significantly improved socket healing in Wnt1^Cre2; Alpl^fl/fl mice by 21 dpp, increasing BV/TV more than 50% (p<0.05), but not changing BMD or TMD measurements (Figure 7A–G). Histology was employed to provide insights into healing bone architecture and composition. Masson’s trichrome stains showed normal bone architecture, with greater bone in treated HPP mice, and IHC for OPN showed expected bone marker localization in new bone within healing sockets (Figure 7H–J).

Figure 7. Enzyme Replacement Therapy Ameliorates Defective Alveolar Bone Socket Healing in Mice Lacking Neural Crest Expression of Alpl.

(A-F) 3D (occlusal and buccal views) and 2D (transverse and sagittal views) images of alveolar bone healing at 21 dpe. Red color in 3D images highlights new AB within sockets. Yellow dotted line in 2D images indicates the location of original socket. Micro-CT analysis shows reduced socket healing in Wnt1^Cre2; Alpl^fl/fl mice is ameliorated with administration of enzyme replacement therapy (ERT) (panels E and F) (n=4/group). (G) ERT improves bone volume fraction (BV/TV) in socket healing in Wnt1^Cre2; Alpl^fl/fl vs. untreated mice. (H-J) Masson’s trichrome (MT) stain and OPN immunostaining show normal bone architecture and marker localization in healing sockets (n=4/group). *p<0.05; **p<0.01; ***p<0.001.

4. Discussion

We analyzed Alpl conditional deletion by Wnt1^Cre2 in cranial neural crest cells in a new mouse model of HPP that phenocopies dentoalveolar effects of odonto HPP. We compared Wnt1^Cre2-mediated phenotype to effects of Alpl deletion by Prx1^Cre. Whereas Prx1^Cre; Alpl^fl/fl mice exhibited reduced circulating ALP levels and only appendicular bone defects, Wnt1^Cre2; Alpl^fl/fl mice had reduced ALP and phenocopied defects in dentin, cementum, and alveolar bone, with loss of PDL attachment. Wnt1^Cre2; Alpl^fl/fl mice also displayed defects in endochondral craniofacial tissues, with abnormalities in the ISS of the cranial base and in the mandibular condyle. Cranial neural crest deletion of Alpl represents a new approach to create a mouse model of HPP that can be used to investigate pathological mechanisms and therapies in craniofacial and dentoalveolar tissues. As proof-of-principle, we studied bone healing for the first time in an animal model of HPP and demonstrated defective alveolar bone healing after maxillary molar extraction in Wnt1^Cre2; Alpl^fl/fl mice. Delivery of ERT ameliorated substantial bone healing defects in HPP mice, providing a model for studying mechanisms of bone healing in HPP conditions with and without ERT. Further analyses of the craniofacial phenotype of this mouse, particularly cranial base and condyles, are needed. However, this model represents an advance in the field of HPP by establishing a mouse where tissue-limited effects of Alpl loss-of-function provide opportunities for both basic and translational studies.

4.1. Lack of Craniofacial Effects from Prx1-directed Alpl Deletion

Paired-related homeobox 1 (Prx1) is a transcriptional factor expressed during early development in limb bud mesenchyme of the appendicular skeleton and a subset of craniofacial/dental mesenchyme [21]. Using the Prx1-Cre driver, we observed cartilage defects in long bones, characterized by growth plate widening, regional expansion of the hypertrophic chondrocyte zone, and disorganized chondrocyte arrangement. Femur alterations included both cortical and trabecular bone defects. This was the first analysis of appendicular skeletal defects of Alpl conditional deletion on a 129 genetic background as the previous report used a C57BL/6 background [15]. We and others have found that genetic background contributes to severity of phenotype in skeletal and dental mineralization phenotypes [31, 37–41]. Based on these results, Prx1-directed Alpl deletion provides significant reduction in ALP levels and causes obvious cortical and trabecular bone defects as a model for appendicular bone defects from HPP.

In developing molars, Prx1 is expressed as early as E11.5, with Prx1⁺ cells reportedly giving rise to odontoblasts, PDL cells, and dental pulp cells [42]. Alpl is expressed in hypertrophic chondrocytes of cranial base synchondroses and global Alpl deletion leads to the expansion of the hypertrophic zone, delayed endochondral ossification, and reduced anterior-posterior growth contributing to dome-shaped skulls in mice [43]. However, in Prx1^Cre; Alpl^fl/fl mice, we did not observe dentoalveolar defects or evident defects in cartilaginous tissues, i.e. cranial base synchondrosis (ISS and SOS) or in mandibular condyles. While femurs showed clear differences, dentoalveolar and craniofacial structures from Prx1^Cre; Alpl^fl/fl mice were indistinguishable from those of controls. This observation of lack of phenotype in the current study raises the possibility that either this Prx1-Cre driver is insufficient to delete Alpl from the necessary craniofacial mesenchyme or that Prx1-Cre expression does not occur in sufficient numbers of cells that express Alpl in teeth, cranial synchondroses, or mandibular joints [44, 45]. We previously found that Prx1-Cre did [15] or did not [46] produce dentoalveolar phenotypes; the cause of this discrepancy is unclear. In this study, the deletion of Alpl in Prx1⁺ cells led to 38% reduction in serum ALP levels, whereas previous studies reported higher reductions of 75% and 48% ALP, corresponding to presence or absence of the dentoalveolar phenotype, respectively [15, 46]. Mice were on different genetic backgrounds in these studies. Caution should be used in choice of and interpretation of craniofacial phenotypes using the Prx1-Cre line, and additional studies are required to identify mechanisms underpinning these findings.

4.2. Neural Crest Deletion of Alpl Results in an Odontohypophosphatasia-Like Phenotype

Wnt1 is selectively expressed by cranial neural crest cells that migrate during early embryonic development to the developing craniofacial region and first branchial arch, becoming ectomesenchymal cells that participate in craniofacial development and odontogenesis [17]. The ectomesenchyme contributes to dental papilla and dental follicle [17]. Therefore, Wnt1-driven Cre recombinase is an ideal choice to target dentin, cementum, PDL, and alveolar bone. Importantly, Wnt1^Cre2; Alpl^fl/fl mice were fed normal chow, did not experience seizures (as did Alpl^−/− mice) and lived to the study endpoint of 60 dpn. Surprisingly, Wnt1-directed deletion of Alpl led to a greater than expected 60% reduction in serum ALP levels at 60 dpn, similar to levels previously reported in wider deletion of Alpl in skeletal and dental tissues using the Col1a1 2.3kb promoter [15]. This large disruption in ALP levels was unanticipated and suggests a larger than expected contribution of craniofacial tissues to circulating ALP. The contributions of various cells and tissues to circulating ALP have not been well characterized and may provide new insights into HPP. While Wnt1-Cre should direct conditional Alpl ablation to primarily neural crest-derived craniofacial mesenchymal cells, osteoblasts and osteocytes of bone likely represent the largest numbers of neural crest-derived cells in the craniofacial area. While osteocytes can be affected by Wnt1-targeted deletion [47] and represent the majority of bone cells [48], they do not produce appreciable amounts of TNAP so would not be expected to contribute significantly to reduced ALP levels as much as major TNAP-producing cells like osteoblasts, odontoblasts, cementoblasts, and PDL fibroblasts [49].

Nonetheless, developmental defects in Wnt1^Cre2; Alpl^fl/fl mice appeared to be limited to craniofacial and dentoalveolar tissues, phenocopying the most prominent aspects of HPP-associated dental effects, including thin dentin, enlarged pulp space, reduced cementum, PDL detachment, and hypomineralized alveolar bone [8, 11]. These dentoalveolar defects in molars resulting from Wnt1-Cre-directed Alpl knockout closely match those reported in Alpl^−/− mice [50–52] and those described in dental case reports [8, 10, 11, 53]. It is of particular interest that dentin and bone defects in HPP mouse models are detected by micro-CT as larger reductions in tissue volumes but minimal changes to bulk density values, in this conditional knockout model as well as the global Alpl knockout, we previously reported [53]. This is likely the result of multiple phenomena. First, dentin mineralization is delayed, with wider predentin and thinner dentin proper; a similar trend exits in bone with greater amounts of osteoid [50–53]. The predentin and osteoid are unmeasurable by micro-CT so that registers as reduced volume of mineralized tissues. Second, TNAP seems to play a critical role in the initiation and growth of mineralization foci, but once those “breakthrough” areas show hydroxyapatite deposition and growth and merging of foci, they appear to proceed somewhat normally, albeit delayed in HPP. This is an area that requires further investigation through techniques such as FIB-SEM, which have provided new insights into defective mineralization in other disorders such as X-linked hypophosphatemia (XLH) [54]. Third, despite it being a powerful 3D approach for analyzing mineralized tissues, micro-CT carries the limitation of diluting local effects when reporting on bulk tissue properties. In human and sheep teeth affected by HPP, we found that the outermost mantle dentin layer was specifically rendered hypomineralized by HPP [30, 53]. However, the remaining majority of the circumpulpal dentin appeared to hold the normal density, and when dentin was only analyzed in bulk, the mantle dentin effects were minimized. This mantle dentin effect is very difficult to detect in the small mouse dentition.

At first glance, the enamel defects and more severe dentin defects in Wnt1^Cre2; Alpl^fl/fl mouse incisors may be surprising. Disrupted incisor enamel may result in part from documented hypomineralization of mantle dentin in HPP, shown in human and sheep teeth affected by HPP [30, 53]. This dynamic, continually erupting tooth has a different enamel structure and may be more sensitive to disruptions in developmental processes than molars [55]. Enamel defects secondary to developmental dentin defects or related to disrupted periodontal attachment and eruption speed have been documented [56–58].

Like many aspects of HPP, craniofacial size and shape alterations are variable. More severe clinical HPP subtypes of HPP like infantile or severe childhood forms, are associated with craniosynostosis [1, 2]. Changes in facial shape or size are also reported in more severe forms of HPP, often resulting from growth constrictions of craniosynostosis with potential contributions from cranial base alterations or other craniofacial bone changes that have not been well characterized in this disorder [2, 59]. Craniosynostosis and craniofacial alterations are mimicked to some degree in Alpl^−/− mice, which phenocopy severe infantile HPP [29, 31]. We observed mineralization defects in the cranial base in Wnt1^Cre2; Alpl^fl/fl mice. Endochondral ossification in the synchondroses of the cranial base contributes to craniofacial growth [60] and TNAP is expressed in hypertrophic chondrocytes [13]. The ISS, an anteriorly positioned growth plate derived from neural crest ectomesenchyme, showed several perturbations in Wnt1^Cre2; Alpl^fl/fl mice, including enlarged spaces between basal bones, regions of premature closure, and enlarged chondrocyte hypertrophic zones. The more posterior SOS is largely derived from mesoderm so would be expected to be unaffected, in agreement with our observations. These cranial bases changes are similar, but less severe, than those recently reported for P0^Cre; Alpl^fl/fl mice [20]. P0^Cre; Alpl^fl/fl mice exhibited ISS changes including increased proliferation and decreased apoptosis of chondrocytes and altered expression of ColX and Sox9 indicative of dysregulated chondrocyte differentiation. These disturbances contributed to altered cranial base bone lengths. P0 is also a cranial neural crest marker used to target gene deletion in craniofacial ectomesenchyme. However, substantial differences have been noted in midbrain and hindbrain neural crest localization of Wnt1 vs. P0 expression that likely reflects different distributions of neural crest subpopulations, their contributions to craniofacial connective tissues, and interpretation of phenotypes resulting from Cre-directed gene deletion [61].

While the ISS showed examples of premature closure, we did not observe craniosynostosis, a condition of premature cranial suture fusion that causes increased intracranial pressure and prevents proper skull and brain growth. No differences were noted in Wnt1^Cre2; Alpl^fl/fl mice directly by examination of cranial sutures at 60 dpn, but changes were suggested indirectly by measurements of skull length and width at 60 dpn, when skull width was reduced in Wnt1^Cre2; Alpl^fl/fl mice vs. controls. Craniosynostosis occurs in some children with early-onset, severe HPP forms, including infantile and childhood HPP [62]. Alpl^−/− mice exhibit premature fusion of coronal sutures by 15–20 dpn, resulting in altered cranial size and morphology [29, 31]. Lack of craniosynostosis in Wnt1^Cre2; Alpl^fl/fl mice may be attributed to findings that cranial vault suture origins are complex, with contributions from both neural crest ectomesenchyme and mesoderm [63]. Notably, craniosynostosis was not reported in P0^Cre; Alpl^fl/fl mice [20]. These are preliminary craniofacial studies accompanying a more focused dentoalveolar analysis and are limited by low sample numbers and relatively simple analysis approach. This new mouse model and the recently reported P0^Cre; Alpl^fl/fl mice, represent tools for more detailed explorations of effects of HPP on craniofacial growth, as well as treatment effects.

Effects of HPP on the mandibular condyle and association with temporomandibular joint dysfunction have been reported anecdotally, but are an aspect of HPP almost wholly unstudied [64]. Severe dentoalveolar defects may contribute secondarily to temporomandibular dysfunction because of malocclusion, but our results here support potential direct effects on condylar endochondral ossification. The mandibular condyle showed dramatic defects in Wnt1^Cre2; Alpl^fl/fl mice, including increased size and a large, hypomineralized region of subchondral bone that was found to result from disorganized accumulation of chondrocytes into the underlying subchondral bone. Condylar enlargement and cartilage disorganization may reflect similar mechanisms as growth plate abnormalities in other joints that contribute to rickets [3], and may additionally share mechanisms with the disorganized and expanded cranial growth plates at the ISS. Presence and functions of TNAP in various endochondral growth plates have been only minimally studied [20, 29, 65] and require additional experiments with new models like the conditional knockouts described here.

4.3. Alveolar Bone Healing in a Mouse Model of Odontohypophosphatasia

Tooth mobility and loss are amongst the most common dental consequences of HPP, there are very few case reports of dental implants in individuals with HPP [10, 18, 19]. All preclinical animal studies focusing on dentoalveolar tissues have administered ERT from very early postnatal ages during odontogenesis, effectively preventing dental mineralization defects [50–52, 66]. However, the more common clinical scenario involves individuals affected by HPP being diagnosed at variable ages and going untreated for years or decades. It is currently unclear how much effect ERT will have on ameliorating defects that have accumulated over these years. To date, no studies have been done on ERT and alveolar bone healing. We used maxillary first molar extraction as a first proof-of-principle experiment to test the ability of alveolar bone to heal in this new mouse model. Wnt1^Cre2; Alpl^fl/fl mice exhibited both deficient quantity and quality of new bone. To take a first translational step in this direction, we treated Wnt1^Cre2; Alpl^fl/fl mice with TNAP-Fc-D₁₀, a mineralized tissue-targeted ERT that is FDA-approved and currently used to treat HPP in several countries [1, 52, 67]. Even short-term ERT improved socket healing in the odonto HPP mice, mainly increasing bone volume compared to untreated Wnt1^Cre2; Alpl^fl/fl mice. These findings suggest ERT may improve outcomes for dental procedures in individuals affected by HPP, though additional studies are necessary.

5. Conclusions

These results establish a new mouse model of HPP that may be particularly useful for focused studies of craniofacial and dentoalveolar defects, including their underlying mechanisms and response to therapeutic interventions. The combination of craniofacial and dental mineralization defects, lack of substantial appendicular bone involvement, lack of seizures, and long life, is unique among HPP mouse models reported to date. As a first proof-of-principle translational study using this model, we provided ERT and challenged bone healing using a model of molar tooth extraction. Even short term administration of ERT provided clear benefits for alveolar bone socket healing in these mice. While additional preclinical studies must be completed, including dose-response and timed interventions in different challenge models, these results suggest that ERT should be considered for individuals affected by HPP who are candidates for dental treatments such as orthodontics, periodontal treatment, and dental implant placement. Preclinical studies combined with clinical case reports may provide guidance for evidence-based approaches that will significantly improve oral health outcomes for individuals affected by HPP.

Supplementary Material

Supplementary Materials: Appendix with 6 supplemental figures

HIGHLIGHTS.

• Mice with ablation of the Alpl gene in cranial neural crest cells phenocopy dentoalveolar defects of hypophosphatasia (HPP)

• Cranial base and mandibular condyle growth plates are affected by conditional deletion of the Alpl gene

• Alveolar bone socket healing is reduced by conditional deletion of the Alpl gene

• Enzyme replacement therapy improves alveolar bone healing after tooth extraction in Alpl conditional knockout mice