Gene Therapy Using Adeno‐Associated Virus Serotype 8 Encoding TNAP‐D10 Improves the Skeletal and Dentoalveolar Phenotypes in Alpl −/− Mice

ABSTRACT Hypophosphatasia (HPP) is caused by loss‐of‐function mutations in the ALPL gene that encodes tissue‐nonspecific alkaline phosphatase (TNAP), whose deficiency results in the accumulation of extracellular inorganic pyrophosphate (PPi), a potent mineralization inhibitor. Skeletal and dental hypomineralization characterizes HPP, with disease severity varying from life‐threatening perinatal or infantile forms to milder forms that manifest in adulthood or only affect the dentition. Enzyme replacement therapy (ERT) using mineral‐targeted recombinant TNAP (Strensiq/asfotase alfa) markedly improves the life span, skeletal phenotype, motor function, and quality of life of patients with HPP, though limitations of ERT include frequent injections due to a short elimination half‐life of 2.28 days and injection site reactions. We tested the efficacy of a single intramuscular administration of adeno‐associated virus 8 (AAV8) encoding TNAP‐D10 to increase the life span and improve the skeletal and dentoalveolar phenotypes in TNAP knockout (Alpl −/−) mice, a murine model for severe infantile HPP. Alpl −/− mice received 3 × 1011 vector genomes/body of AAV8‐TNAP‐D10 within 5 days postnatal (dpn). AAV8‐TNAP‐D10 elevated serum ALP activity and suppressed plasma PPi. Treatment extended life span of Alpl −/− mice, and no ectopic calcifications were observed in the kidneys, aorta, coronary arteries, or brain in the 70 dpn observational window. Treated Alpl −/− mice did not show signs of rickets, including bowing of long bones, enlargement of epiphyses, or fractures. Bone microstructure of treated Alpl −/− mice was similar to wild type, with a few persistent small cortical and trabecular defects. Histology showed no measurable osteoid accumulation but reduced bone volume fraction in treated Alpl −/− mice versus controls. Treated Alpl −/− mice featured normal molar and incisor dentoalveolar tissues, with the exceptions of slightly reduced molar enamel and alveolar bone density. Histology showed the presence of cementum and normal periodontal ligament attachment. These results support gene therapy as a promising alternative to ERT for the treatment of HPP. © 2021 The Authors. Journal of Bone and Mineral Research published by Wiley Periodicals LLC on behalf of American Society for Bone and Mineral Research (ASBMR).


Introduction
H ypophosphatasia (HPP) is an inborn error-of-metabolism caused by loss-of-function mutations in the ALPL gene, which encodes tissue-nonspecific alkaline phosphatase (TNAP). (1)(2)(3) TNAP is expressed in bones, teeth, liver, and kidney; its deficiency leads to mineralization defects caused by the accumulation of extracellular inorganic pyrophosphate (PP i ), one of the major substrates of TNAP and a potent inhibitor of hydroxyapatite crystal formation and propagation. (4) Murine studies have demonstrated that severe skeletal hypomineralization, respiratory failure due to thoracic cage dysplasia and hypoplastic lungs, and elevated intracranial pressure due to craniosynostosis. (8)(9)(10)(11) Dentoalveolar phenotypes, including premature exfoliation of primary teeth, periodontal disease, and enamel alternations, are commonly observed in patients with all forms of HPP. (12)(13)(14) There are more than 400 mutant alleles identified for the ALPL gene (the ALPL mutation database http://alplmutationdatabase. hypophosphatasie.com), and their genotype/phenotype correlations are not well understood. (15,16) The inheritance pattern of perinatal and infantile HPP is often autosomal recessive, with most patients being compound heterozygotes for pathogenic ALPL mutations that result in almost null alkaline phosphatase (ALP) activity, but some are homozygous for recessive alleles and most adult and odonto-HPP patients harbor a single dominant-negative ALPL allele. (17,18) Asfotase alfa is a recombinant fusion protein comprising the TNAP ectodomain, a human IgG1 Fc domain for one-step purification, and a terminal deca-aspartate (D 10 ) motif for mineral targeting. In a murine model for infantile HPP, TNAP knockout (Akp2 À/À or Alpl À/À ) mice, treatment with daily subcutaneous injections of asfotase alfa preserved life span, improved skeletal phenotypes, and prevented epileptic seizures and dental defects. (19)(20)(21) In humans, subcutaneous injections of asfotase alfa, three to seven times a week, in children or adults with HPP has demonstrated substantial and sustained efficacy with a good safety profile. Asfotase alfa saved lives of severe neonatal and infantile HPP and improved bone mineralization, motor function, and quality of life in adult HPP. (8,(22)(23)(24)(25)(26) At the same time, the patient burden of multiple injections per week to maintain the efficacy of asfotase alfa and the associated medical cost (27,28) have prompted preclinical studies of alternative strategies for treating HPP.
A human chimeric recombinant alkaline phosphatase, ChimAP, and several forms of virus vectors expressing TNAP-D 10 have shown to prolong life, prevent seizures, and improve the skeletal phenotype of Alpl À/À mice. (29)(30)(31)(32)(33) A single intravenous injection of a lentiviral or adeno-associated virus type 8 (AAV8) vector encoding TNAP-D 10 led to sustained correction of the skeletal phenotype of Alpl À/À mice, but the consequential wide distribution of vector genome to the whole body raised concern about possible transduction into germ cells. (30) In this study, we treated Alpl À/À mice with a single intramuscular injection of AAV8 vector encoding TNAP-D 10 and performed a detailed analysis of bone microstructure, dentoalveolar phenotype, inorganic PP i metabolism, and ectopic calcification to assess the efficacy and safety of viral vector-mediated gene therapy.

Materials and Methods
Mouse model of infantile HPP TNAP knockout (Alpl À/À ) mice were created by inserting a Neo cassette into exon 6 of the mouse Alpl gene via homologous recombination. (34) Alpl À/À mice phenocopy human infantile HPP, showing normal appearance and being indistinguishable from other siblings at birth. (35) They display almost zero circulatory ALP activity, develop epileptic seizures, become cachectic, and die by 10 to 12 days postnatal (dpn) without additional supportive treatment. (35) Alpl À/À mice were maintained in a 12.5% C57Bl/6 and 87.5% 129 J background and genotyped by PCR using genomic DNA extracted from toe samples within 5 days after birth. All animals (breeders, nursing mothers, pups, and weanlings) in this study were given free access to regular diet (2018 Teklad global 18% protein extruded rodent diets or 2019 Teklad global 19% protein rodent diets, Envigo, Indianapolis, IN, USA) with a standard level of vitamin B6, increased level of which was reported to improve the life span of Alpl À/À mice. (19,34,35) The institutional Animal Care and Use Committee (IACUC) approved all the animal studies.
Virus vector encoding the human TNAP-D 10 cDNA TNAP-D 10 contains recombinant human soluble TNAP (sALP) and a deca-aspartate (D 10 ) sequence at the C terminus, which enables TNAP to target mineralized tissues, such as bone and teeth. (19) The human IgG1 Fc domain present in asfotase alfa to enable a one-step purification is absent from the product of this vector as purification is not required for viral-vector mediated in vivo expression. Recombinant AAV type 8 vector encoding TNAP-D 10 (AAV8-TNAP-D 10 , aka ARU-2801) was generated using the HEK293 cell line by the triple transfection method, purified, and then titrated as previously reported. (36)(37)(38) Recombinant AAV type 8 vector encoding GFP (AAV8-GFP) was used as a control to evaluate viral vector-related adverse effects. (39) Eleven Alpl À/À mice (male n = 6, female n = 5) and 14 wild-type (WT) controls (male n = 7, female n = 7) were included in this study. After genotyping, all the Alpl À/À mice received a single injection of AAV8-TNAP-D 10 at a dose of 3 Â 10 11 vector genomes (vg)/ body into the quadriceps femoris within 5 dpn. Three control WT mice received the same dose of AAV8-TNAP-D 10 to assess its effect on PP i metabolism and soft-tissue calcification when endogeneous TNAP activity is present. Seven WT mice received the same amount of control AAV8-GFP vector and 4 WT mice were untreated; their data were combined and analyzed together as WT because of the lack of substantial differences between them. Mice were euthanized at 70 dpn by exsanguination, after intraperitoneal administration of Avertin. In mice, 70 dpn is the end of adolescence, equivalent to a human age of 20 years, (40) and root formation and cellular cementum formation have been completed. (20,29,41) For the analysis of untreated Alpl À/À , 17 WT, 7 heterozygote (Alpl +/À ), and 10 Alpl À/À pups were collected at 10 dpn and euthanized by exsanguination after intraperitoneal administration of Avertin.

Sample collection and biochemical analyses
Body weight of the mice was measured at 35 and 70 dpn. Blood was collected from the orbital sinus of isoflurane-anesthetized mice using Pasteur pipets every 4 weeks after injection. Spot urine samples were collected simultaneously.
Blood was collected into two types of BD Microtainers coated with either clot activator or lithium heparin (Becton, Dickinson and Company, Franklin Lakes, NJ, USA). Blood for plasma collection and urine samples were placed on ice. Microtainers were then spun at 7000g for 10 minutes. Twenty microliters of heparin-plasma were deproteinized using a Microcon-10 kDa Centrifugal Filter Unit with Ultracel-10 membrane (MilliporeSigma, Merck KGaA, Darmstadt, Germany), centrifuged at 14,000g for 20 minutes. Urine was diluted 1:3 with 10 mM HEPES. Samples were stored at À80 C for further analyses.
Plasma PP i concentration was measured according to the protocol described previously. (42) Five microliters of deproteinized plasma sample and the PP i standard ranging from 0.125 μM to 20 μM (sodium pyrophosphate decahydrate, Sigma-Aldrich, St. Louis, MO, USA) were added to 45 μL of assay mixture containing 90 μM adenosine 5 0 phosphosulfate sodium salt (APS) (Sigma-Aldrich), 22.5 μM MgCl 2 , 11.25 mM HEPES with 0.9 U/mL recombinant yeast ATP-sulfurylase/MET3 (R&D Systems, Inc., Minneapolis, MN, USA). The mixture was incubated at 37 C for 30 minutes and heat-inactivated at 90 C for 10 minutes. Ten microliters of each sample were then transferred into a 96-well white-bottom plate and mixed with 50 μL of BacTiter-Glo Microbial Cell Viability Assay (Promega Corporation, Madison, WI, USA). Luminescence was measured by FilterMax F5 Multimode Microplate Readers (Molecular Devices, LLC, San Jose, CA, USA).
Serum ALP activity was measured using an enzymatic assay. Five microliters of serum and recombinant human ALP standards (0.099 to 216.0 ng/mL) were mixed with 95 μL of 10 mM pNPP in diethanolamine (DEA) buffer (pH 9.8) containing 1.0 mM MgCl 2 and 20 μM ZnCl 2 . The increase in A405 nm was measured using OptiMax Microplate Absorbance Reader (Molecular Devices) for 15 minutes. QuantiChrom Calcium Assay Kit, QuantiChrom Creatinine Assay Kit, and QuantiChrom Urea Assay Kit (BioAssay Systems, Hayward, CA, USA) were used to measure serum and urine calcium, urine creatinine, and serum urea concentrations, respectively. Stanbio Phosphorus Liqui-UV (EKF Diagnostics-Stanbio Laboratory, Boerne, TX, USA) was used to measure serum and urine phosphorus concentrations.

Quantitative polymerase chain reaction (qPCR)
Total RNA was extracted from the kidney using RNAeasy Plus Kit (Qiagen LLC, Germantown, MD, USA), and reverse transcription was carried out using PrimeScript RT Master Mix (Takara Bio USA, Inc., Mountain View, CA, USA). Real-time qPCR was performed in a 384-well plate in an Applied Biosystems 7900HT Fast Real-Time PCR system (Thermo Fisher Scientific, Waltham, MA, USA) using cDNA equivalent to 25 ng total RNA and DyNAmo Flash SYBR Green qPCR Kit (Thermo Fisher Scientific). The reaction was run for 40 cycles at an initial temperature of 95 C for 7 minutes and then at 95 C for 10 seconds followed by 60 C for 15 seconds. Ct values were determined by the software, and the amplification of the target gene was normalized to that of 18S ribosomal RNA (Rn18s). Sequences of the primer pairs used for PCR are as follows: Alpl (NM_007431.3) F-CTGCCACTGCCTACTTGTGT and R-GATGGATGTGACCTCATTGC; Ank (progressive ankylosis protein) (NM_020332.4) F-CTGCTGCTACAGAGGCAGTG and R-GACAAAA CAGAGCGTCAGCGA; Enpp1 (ectonucleotide pyrophosphatase phosphodiesterase 1) (NM_001308327.1) F-TCACGCCACCGAGA CTAAATA and R-TGCAGTAGGGTGTCATGAAGG; Abcc6 (ATP binding cassette subfamily C member 6) (NM_018795.2) F-CA TCTTGCCAGGAATCAACACT and R-ACCAGGGACAAGCACAGGTA; Il6 Radiography and micro-computed tomography (μCT) Radiographic images of entire skeletons and forelimbs, hindlimbs, and skulls were obtained with a Faxitron MX-20 DC4 (Chicago, IL, USA), using energy of 20 kV. Lengths of the femur, tibia, humerus, and radius were measured using ImageJ (Rasband WS, ImageJ, National Institutes of Health, Bethesda, MD, USA, https://imagej.nih.gov/ij/, 1997-2018). Head measurements were performed using the following landmarks (43,44) : nose length, the length from the rostal point of intersection of nasal bones to the caudal point of intersection of nasal bones; cranial length, the length from the rostal point of intersection of nasal bones to the median (midline) point of the posterior margin of the foramen magnum; cranial width, the length from the right joining of squamosal body to zygomatic process of squamous portion of temporal bone to the left counterpart.
The first mandibular molar and associated alveolar bone were quantitively analyzed as previously described. (45) The alveolar bone region of interest (ROI) included the area between 240 μm mesial to the most mesial point of the first molar mesial root and 240 μm distal to the most distal point of the distal root. Enamel was segmented above 1600 mg HA/cm 3 , while dentin/cementum and alveolar bone were segmented at 550 to 1600 mg HA/cm 3 .

Tissue collection and histological studies
Skeletal and soft tissues were fixed in 4% paraformaldehyde/PBS solution and processed for histological analyses. Undecalcified fixed bone samples were placed in 30% sucrose/PBS solution and then cryo-embedded in Optimal Cutting Temperature (OCT) compound (Tissue-Tek, Torrance, CA, USA) in hexane dry ice bath and were sectioned by the Kawamoto method. (46) The tibial and femur bones were placed in 0.125 M EDTA/10% formalin (pH 7.3) solution for 7 days for decalcification and were then paraffin embedded. Soft organs were either paraffin embedded or embedded in an OCT in an ethanol dry ice bath. Hematoxylin and eosin (H&E), von Kossa, von Kossa/van Gieson, and Safranin O staining were performed according to standard methods. Tissue ALP activity was assayed by incubating the OCT-embedded sections in freshly mixed substrate solution made of one volume of 0.2 mg of Naphthol AS-MX phosphate disodium salt per mL of water and one volume of 1.2 mg FAST Violet B salt per mL of 0.2 M Tris-HCl (pH 8.9) at room temperature for 60 minutes, and counterstaining in methyl green solution. (47) Slides were observed under IX81 Olympus Microscope (Olympus Corporation, Center Valley, PA, USA) or scanned by Aperio AT2 system (Leica Biosystems of Leica Microsystems Inc., Buffalo Grove, IL, USA). Bone volume fraction (BV/TV) was measured using ImageJ.
Left hemi-mandibles were fixed in Bouin's solution overnight, decalcified in an acetic acid/formalin/sodium chloride solution, processed for paraffin embedding, and sectioned at 5 μm thickness in the coronal plane. Paraffin sections were stained with H&E to assess tooth and associated periodontium. Immunohistochemistry (IHC) procedures were performed as described previously. (48) Primary antibodies included monoclonal rat anti-human alkaline phosphatase IgG (TNAP) (R&D Systems) (49) ; polyclonal rabbit anti-mouse bone sialoprotein (BSP) IgG (Dr Renny Franceschi, University of Michigan, Ann Arbor, MI, USA) (50) ; and polyclonal LF-175 rabbit antimouse osteopontin (OPN) IgG (Dr Larry Fisher, NIDCR/NIH, Bethesda, MD, USA). (48) In situ hybridization was performed for mouse Alpl with RNAscope 2.5 HD Detection reagent kit-RED assay (Advanced Cell Diagnostics, Newark, CA, USA) following the manufacturer's instructions as previously described. (51) For acellular AAV8-TNAP-D 10 improves survival and altered PP i metabolism in AAV8-TNAP-D 10 -treated Alpl À/À mice. (A) Average body weight of AAV8-TNAP-D 10 -treated male Alpl À/À mice is comparable to WT littermates, although female Alpl À/À mice have lower body weight than WT. Blue and red circles in boxplots represent male and female data, respectively. (B) AAV8-TNAP-D 10 -treated Alpl À/À and AAV8-TNAP-D 10 -treated Alpl À/À WT mice show a significant increase in serum ALP activity compared with untreated WT mice. (C) Plasma PP i levels of AAV8-TNAP-D 10 -treated Alpl À/À mice are significantly lower than those of WT controls. (D) Urine PP i concentrations of AAV8-TNAP-D 10 -treated Alpl À/À mice remain significantly higher than both treated and untreated WT mice (70 dpn). (E) Urine PP i concentrations of untreated Alpl À/À and heterozygote (Alpl +/À ) mice are higher than those of WT mice (10 dpn). (F) Histochemical staining of the kidney reveals no ALP activity in the proximal tubules of AAV8-TNAP-D 10 -treated Alpl À/À mice (upper panel). Light microscopy does not reveal any apparent structural changes in the H&E-stained glomeruli or renal tubules of AAV8-TNAP-D 10 -treated Alpl À/À mice (lower panel). Statistical analysis was performed by unpaired t test with Welch's correction in male mice in (A) and one-way ANOVA followed by Tukey's multiple comparison test in female mice in (A) as well as (B-E).
Journal of Bone and Mineral Research n 1838 KINOSHITA ET AL. cementum/periodontal ligament (PDL) analysis, H&E-stained images captured with the same acquisition parameters were segmented using the color map function (5 Ramps) in ImageJ. This method is to pseudo-color the images to make differences between pixel values more apparent for improved tissue visualization. The values for acellular cementum and mantle dentin thickness represent the average of three linear measurements taken at 90 μm, 100 μm, and 110 μm from the cemento-enamel junction (CEJ) using the ImageJ straight-line function. Mantle dentin is the outer layer, less mineralized dentin adjacent to acellular cementum. For assessing cellularity, a region of 5.5 mm 2 area was defined 100 μm apical to the CEJ for counting cells in the PDL space.

Statistics
All the statistical analyses were performed using GraphPad Prism version 9.0.0 (GraphPad Software, San Diego, CA, USA). Data are expressed as mean AE standard deviation (SD) in charts. Statistical analysis was performed by one-way analysis of variance (ANOVA) followed by Tukey's multiple comparison test to explore the differences among control WT, treated Alpl À/À , and treated WT mice. Unpaired t test with Welch's correction was performed to test the difference between WT and treated Alpl À/À mice. Significance was determined by p < 0.05.

Results
Improved survival and correction of plasma PP i in AAV8-TNAP-D 10 -treated Alpl À/À mice While untreated Alpl À/À pups die within 10 to12 days after birth, all Alpl À/À mice treated with AAV8-TNAP-D 10 in this study did not develop epileptic seizures and were viable until the endpoint of the study at 70 dpn. The untreated Alpl À/À pups did not feed well and were significantly smaller than WT littermates at 10 dpn (Supplemental Fig. S1A). In contrast, the Alpl À/À mice injected with AAV8-TNAP-D 10 within 5 dpn showed catch-up growth, and their body weight was not significantly different from WT littermates at 15 dpn (Supplemental Fig. S1B). The body weight of AAV8-TNAP-D 10 -treated male Alpl À/À mice were comparable to their WT littermates, but treated female Alpl À/À mice had lower body weight than their WT littermates both at 35 and 70 dpn (Fig. 1A).
Serum ALP activity was significantly higher in Alpl À/À and WT mice injected with AAV8-TNAP-D 10 , approximately 500 to 800 times higher than that of control WT mice ( Fig. 1B; Table 1). There were no significant differences in serum calcium, serum phosphorus, serum urea, urine calcium, or urine phosphorus concentrations among experimental groups (Table 1). While plasma PP i concentrations were almost undetectable in the Alpl À/À and WT mice treated with AAV8-TNAP-D 10 ( Fig. 1C; Table 1), urine PP i concentrations of AAV8-TNAP-D 10 -treated Alpl À/À mice remained significantly higher than WT mice at 70 dpn ( Fig. 1D; Table 1). Elevated urine PP i concentrations were also observed in untreated 10 dpn Alpl À/À mice and Alpl +/À mice (Fig. 1E). qPCR performed on RNA isolated from the kidney showed no expression of the Alpl gene in the Alpl À/À mice, and no significant differences were observed in other genes related to PP i metabolism (Ank, Enpp1, and Abcc6), nor in genes related to inflammation (Il6 and Tnf ) ( Table 2). Histochemical staining of the kidney and liver revealed no ALP activity in the renal proximal tubules and the branches of hepatic artery of AAV8-TNAP-D 10 -treated Alpl À/À mice ( Fig. 1F; Supplemental  Fig. S2). Light microscopy did not reveal any apparent structural changes in the H&E-stained glomeruli or renal tubules (Fig. 1F). No ectopic calcifications were observed in the aorta, coronary arteries, brain, or kidney by 70 dpn (Supplemental Fig. S3).  Improved bone microstructure in AAV8-TNAP-D 10 -treated Alpl À/À mice In previous studies, untreated 20 to 23 dpn Alpl À/À mice demonstrated profound skeletal abnormalities including reduced tissue mineral density and bone fractures. (29,52) Radiography of 70 dpn Alpl À/À mice treated with AAV8-TNAP-D 10 revealed grossly normal skeletal development comparable to WT littermates ( Fig. 2A). Treated Alpl À/À mice did not show features characteristic of rickets or osteomalacia, such as bowing of the long bones, enlargement of the ends of epiphyses, or fractures (Fig. 2B). Long bone lengths of treated male Alpl À/À mice were not significantly different from those of WT mice, while treated female Alpl À/À mice were significantly smaller than their WT littermates (Fig. 2C). Treated Alpl À/À mice did not show craniosynostosis (Fig. 2B). Nose length, cranial length, and cranial width were not significantly different among each treatment group in female mice, while the cranial lengths of AAV8-TNAP-D 10 -treated Alpl À/À male mice were longer than those of WT male mice (Table 3).
Three-dimensional μCT renderings of AAV8-TNAP-D 10 -treated Alpl À/À femurs showed bone morphology similar to WT femurs; however, the bone was $7% to 9% shorter in treated Alpl À/À females compared with WT controls (Figs. 2C and 3A). Notably, treated Alpl À/À mice had abnormal articular surfaces the femur joints compared with that of WT controls. Quantitative μCT analysis showed similar bone trabecular and cortical parameters in treated male and female Alpl À/À femurs compared with WT controls. Metaphyses of distal femurs showed no significant differences in TV, BV, or Tb.BV/TV among groups (Fig. 3B). However, treated Alpl À/À females showed increased Tb.BV/TV associated with improved trabecular connectivity compared with treated WT mice (Fig. 3B). No significant differences were detected in Tb.N, Tb.Th, Tb.Sp, or Tb. BMD among the groups. However, AAV8-TNAP-D 10À treated Alpl À/À females showed reduced cortical BV/TV associated with decreased cortical thickness. Treated Alpl À/À males showed increased cortical porosity, but this did not significantly affect Ct.BV/TV. Also, AAV8-TNAP-D 10 partially restored the Ct.BMD, both in males and females (Fig. 3C), indicating an incomplete rescue of cortical bone defects.
Von Kossa/van Gieson staining of femurs and lumbar spines showed no measurable osteoid surface (Fig. 4A), but the BV/TV values of lumbar spines of AAV8-TNAP-D 10 -treated Alpl À/À mice were significantly lower than that of WT mice ( Fig. 4B; WT n = 5; male n = 2, female n = 3, knockout [KO] n = 3; male n = 1, female n = 2). H&E and Safranin O staining of the decalcified tibias did not show the expansion of the tibial epiphyseal growth plate but showed an abnormal distribution of chondrocytes in secondary ossification centers in AAV8-TNAP-D 10 -treated Alpl À/À mice ( Fig. 4C; Supplemental Fig. S4). Histochemical staining of the femurs revealed strong ALP activity in the hypertrophic zone of the epiphyseal growth plate, metaphysis, and diaphysis of WT mice, while only subtle ALP activity was observed in the growth plate and diaphysis of AAV8-TNAP-D 10 -treated Alpl À/À mice (Fig. 4D).
AAV8-TNAP-D 10 prevents HPP-associated dentoalveolar defects in Alpl À/À mice Data from male and female dentoalveolar tissues were combined because of lack of substantial sex-related differences in previous reports of HPP mouse models (20,49,53) and because we found no sex-related trends in these data. Compared with untreated and treated WT controls, first mandibular molars and surrounding alveolar bone of AAV8-TNAP-D 10 -treated Alpl À/À mice appeared grossly normal (Fig. 5A). Enamel volumes were 12% to 14% reduced in molars of AAV8-TNAP-D 10 -treated Alpl À/À compared with WT groups, although no significant differences were observed in enamel mineral density (Fig. 5B). No differences in molar dentin volume or density were observed between AAV8-TNAP-D 10 -treated Alpl À/À and WT groups (Fig. 5B). Continually erupting incisors of Alpl À/À mice, which harbor mineralization defects previously shown to be relatively resistant to treatment, (20,29) showed no differences in the volumes or densities of enamel or dentin compared with WT groups (Fig. 5C). Alveolar bone showed no differences in the volume between AAV8-TNAP-D 10 -treated Alpl À/À mice and WT controls, although exhibited $4% lower mineral density than WT groups (Fig. 5D).

Discussion
We demonstrated in this study that a single, intramuscular injection of AAV8 vector encoding TNAP-D 10 during early neonatal and immunostaining showed near-normal development of the limbs, skull, teeth, and associated periodontal tissues in treated Alpl À/À mice, while detailed analyses by μCT and histology   AAV8-TNAP-D 10 prevents HPP-associated dentoalveolar defects in Alpl À/À mice. (A) 3D and 2D μCT renderings of first molars (M1) and incisors (INC) exhibit normal tooth structures in AAV8-TNAP-D 10 -treated Alpl À/À mice similar to those in WT controls (70 dpn). (B) First molars show no significant differences in enamel density, dentin volume, or dentin density among the groups. Tooth enamel shows decreased volume in treated Alpl À/À molars compared with WT controls. (C) Continually erupting incisor teeth show no significant defects in the volume or density of either enamel or dentin among the groups. (D) AAV8-TNAP-D 10 significantly improves alveolar bone volume in treated Alpl À/À versus WT mice, but alveolar bone shows 4% less mineral density in treated Alpl À/À versus WT mice. Statistical analysis was performed by one-way ANOVA followed by Tukey's multiple comparison test. EN = enamel; DE = dentin; AB = alveolar bone. revealed a remarkable improvement, yet slightly lower ratio in vertebral BV/TV and an incomplete rescue of cortical bone defects in the long bones of the AAV8-TNAP-D 10 -treated Alpl À/À mice compared with WT mice. These corrections resulting from a single injection of vector-mediated TNAP-D 10 are comparable to the skeletal and dental improvements observed in Alpl À/À mice under daily subcutaneous injection of asfotase alfa up to 52 days. (19) Efficacy of intramuscular delivery of AAV8-TNAP-D 10 Previous approaches to correct HPP-associated mineralization defects have met with mixed success or have had inherent limitations or drawbacks. ERT using asfotase alfa was shown to be very effective at correcting skeletal and dentoalveolar defects in Alpl À/À mice, with these translational studies leading to its approval in 2015 for treatment of patients with perinatal/infantile-and juvenile-onset HPP, except in Japan, where asfotase alfa is approved for all ages. (25) However, asfotase alfa ERT requires multiple injections per week, is associated with injection site reactions, and is expensive, (27,28) prompting additional preclinical studies of alternative strategies for HPP treatment. Daily subcutaneous injection of a soluble, non-mineral-targeting, recombinant chimeric alkaline phosphatase (ChimAP) prevented seizures, increased survival, but only partially improved the skeletal and dentoalveolar phenotypes in Alpl À/À mice. (29) Studies using several different types of viral vectors with different modes of administration have been reported. Single intravenous injection of either lentiviral vector containing TNAP-D 10 (HIV-TNAP-D 10 ) or AAV8-TNAP-D 10 resulted in sustained elevation of circulatory TNAP and phenotypic correction in Alpl À/À mice, but the viral sequence was detected in soft tissues, including the liver, lung, and heart, raising a safety concern for oncogenicity of the integrated lentiviral vector in these tissues. (30,31) When compared with these previous methods, our approach described here using a single intramuscular injection of AAV8-TNAP-D 10 combines a more practical gene therapy, potentially improved safety profile due to limited tissue distribution, and yet remains highly effective at correcting mineralization defects. All forms of HPP include dental involvement, with premature tooth loss being one of the most common manifestations. Characteristic dentoalveolar abnormalities observed in untreated Alpl À/À mice include inhibition of tooth root acellular cementum formation, PDL detachment, and enamel and dentin mineralization defects. (5,20,21) In our study, qualitative and quantitative analyses suggest that teeth and associated periodontal tissues significantly improved from AAV8-TNAP-D 10 -mediated gene therapy. AAV8-TNAP-D 10 -treated Alpl À/À mice showed normal formation and mineralization of molar and incisor enamel and dentin, and the restoration of acellular cementum, PDL cellularity, and PDLcementum attachment. Treated Alpl À/À mice also showed welldeveloped alveolar bone with mildly reduced mineral density. An interesting finding is that AAV8-TNAP-D 10 augmented acellular cementum in AAV8-TNAP-D 10 -treated WT teeth, suggesting a potential effect on cementogenesis even in healthy mice. Circulating ALP levels were previously significantly correlated to mouse incisor acellular cementum thickness, supporting this mechanism. (54) These results were comparable to those shown in the previous studies using daily subcutaneous injections of asfotase alfa. (19)(20)(21)52) Insights into the distribution of TNAP-D 10 and the role of chondrocytes in HPP In our study, histochemical staining of the femur showed significantly lower ALP activity in AAV8-TNAP-D 10 -treated Alpl À/À mice in the epiphyseal growth plate and endosteum (Fig. 4D). Previous studies suggested that ALP activity observed in the bone of viral vector-treated Alpl À/À mice decreases with age. Neonatal intravascular injection of AAV8-TNAP-D 10 resulted in positive staining on the growth plate and the surface of the endosteal bone in 10 dpn mice. (30) The tibias of mice that underwent transuterine intraperitoneal injections of AAV8-TNAP-D 10 showed strong ALP activity in hypertrophic chondrocytes and on the surface of endosteal bones on day 14 but only scattered ALP signals in the cartilage zone on day 56. (32) TNAP might not be actively incorporated into those sites at 70 dpn when skeletal maturity is achieved and avascularity of cartilage might limit accessibility of the therapeutic protein. (55) Additionally, abnormal distribution of hypertrophic chondrocytes was observed in the tibial epiphysis in our AAV8-TNAP-D 10 -treated Alpl À/À mice ( Fig. 4C; Supplemental Fig. S4). The abnormal distribution of chondrocytes and irregularly arranged growth plates were previously reported in untreated 18-to 25-dpn Alpl À/À mice and the Alpl À/À mice injected with a muscle-directed vector, scAAV8-TNAP-D 10 . (35,56) In vitro experiments have shown that the inhibition of endogenous TNAP in hypertrophic chondrocytes impacts chondrocyte differentiation. (57) Therefore, we hypothesize that the distribution of TNAP-D 10 may result in the normalization of growth plate irregularity and the remaining abnormal chondrocyte distribution in our treated Alpl À/À mice.
Lastly, three-dimensional μCT showed abnormal articular surfaces of medial and distal femurs in the treated Alpl À/À mice (Fig. 3A). Articular surface disruption is often observed in the clinical syndrome of posttraumatic osteoarthritis, which occurs after the frequent injuries that cause chondral and subchondral bone damage. (58) Joint complications such as arthropathies and periarticular calcification were reported in a mouse model of lateonset HPP (59) and adult patients with HPP. (23,60) Since the initial repair by chondrocytes is considered important in preventing the further progression of joint degeneration, (58) the chondrocytes that lack TNAP activity might have caused an osteoarthritis-like phenotype in our treated Alpl À/À mice. Further Improved cementum and PDL attachment in AAV8-treated Alpl À/À mice. (A) H&E staining shows no evident dental defects in treated Alpl À/À mice compared with WT controls. Boxed regions are shown at higher magnification (left). (B) In situ hybridization with Alpl probe (left) confirms the absence of Alpl expression in treated Alpl À/À mice. TNAP IHC (right) shows weak staining (brown) around treated Alpl À/À alveolar bone (AB) and more robust staining in WT mice. (C) BSP IHC (left) shows evident staining (brown) in acellular cementum (AC, white arrow) and alveolar bone (AB) in treated Alpl À/À and WT mice. OPN IHC (right) shows comparable staining (brown) in AC and AB in treated Alpl À/À versus WT mice. (D) ImageJ color map (left) shows improved cellularity (yellow symbols) within the PDL space of treated Alpl À/À versus WT mice, and (right) shows the width of AC (yellow lines) in treated Alpl À/À versus WT mice. (E-G) Quantification of the thickness of acellular cementum, mantle dentin (MD), and cellularity in the PDL space. Statistical analysis was performed by one-way ANOVA followed by Tukey's multiple comparison test.
investigation is needed to clarify the etiology of joint degeneration in adult Alpl À/À mice.
Considerations for AAV8-TNAP-D 10 gene therapy Therapies for HPP other than asfotase alfa have been tested in a clinical setting without rigorous preclinical and translational studies. In a cross-sectional study that enrolled 51 patients with childhood and adult-onset HPP, 2 of 4 patients who were treated with teriparatide (parathyroid hormone 1-34) showed clinical and radiological improvement. (61) In a phase IIA open-label study targeting 8 adult patients with HPP, monoclonal antisclerostin antibody (BPS804) treatment resulted in increases in bone formation markers and bone mineral density. (62) However, these anabolic agents have not been approved for HPP and an effective medical therapy for adult-onset HPP patients is needed. At the same time, a novel therapy with fewer injections may further benefit patients with childhood-onset HPP as well.
Concerns for AAV8-TNAP-D 10 administration in human patients include continuously elevated circulatory ALP activity and its potential effects on PP i metabolism and development or aggravation of ectopic calcification. In our studies, a single intramuscular injection of AAV8-TNAP-D 10 resulted in extremely high circulatory ALP activity, more than 500 times higher than that of WT mice. Extracellular PP i is known as a central regulator of biomineralization and is critical for controlling inappropriate soft-tissue calcification in the body. Hydrolysis of PP i by TNAP-D 10 may result in oversuppression of extracellular PP i in soft tissues. However, AAV8-TNAP-D 10 -treated Alpl À/À mice, with suppressed plasma PP i concentrations, did not develop any soft-tissue calcifications during our observational window of 70 dpn. Although it is unclear if the significant drop of PP i concentrations in the treated animals to levels below the control group is due to in vivo metabolism of PP i or ex vivo catabolism during the processing of the samples, this contrasts with previous results from genetically modified mouse models with targeted overexpression of TNAP in the vasculature. TNAP overexpression in the smooth muscle cells using Tagln-Cre resulted in massive arterial calcifications in the ascending and descending aorta, carotid, and subclavian arteries. (63) TNAP overexpression in endothelial cells using Tie2-Cre resulted in the partial calcification in the arteries of the heart, kidney, mesentery, pancreas, and spleen. (64) These mouse models showed 20 to 30 times higher circulatory ALP activity, while plasma PP i concentrations were similar to those of WT. From these results, we conclude that neither circulatory ALP activities nor circulatory PP i concentrations correlate with local PP i concentrations, which cannot be measured in vivo but presumably are low in target tissues with high TNAP expression. Additionally, the elevated urine PP i concentrations in TNAP-D 10 -treated Alpl À/À mice indicate that urine PP i metabolism is mostly dependent on TNAP expressed in the luminal surface of kidney proximal tubules and independent of serum ALP activity and circulatory PP i concentrations. The qPCR analysis of the kidneys of treated mice indicated that genes for primary regulators of systemic PP i metabolism, including Ank, Enpp1, and Abcc6, were not affected by treatment.
In patients with HPP, with or without treatment, several types of ectopic calcifications are commonly observed: ocular calcification, nephrocalcinosis, and painful periarthritis. (22,23,(65)(66)(67)(68)(69)(70)(71) Our AAV8-TNAP-D10-treated Alpl À/À mice did not develop nephrocalcinosis, kidney stones, or structural changes in the glomeruli or renal tubules as observed under light microscopy. qPCR did not show a significant increase in the inflammatory markers in the kidney. These mice did not develop hypercalcemia or hypercalciuria (Table 1). Therefore, we hypothesized that abnormal mineral metabolism, such as hypercalcemia and hypercalciuria observed in untreated HPP, might facilitate the development of nephrocalcinosis rather than an inflammatory response associated with TNAP deficiency. (72) This study was terminated at 70 dpn; therefore, long-term effects of AAV8-TNAP-D 10 with the presence of age-related ectopic calcifications remain unknown but are an area for further study. We have previously shown that asfotase alfa binds to the sites of ectopic calcification in the TNAP overexpression-induced vascular calcification mouse models (Tagln-Cre and Tie2-Cre). (73) Therefore, we cannot rule out the possibility that the continuous administration of TNAP-D 10 , either subcutaneously or as a gene therapy, may lead to the development or aggravation of ectopic calcification in some HPP patients.
There are several limitations to our study. First, the skeletal growth of treated female Alpl À/À mice was not as well corrected as in male Alpl À/À mice, which suggests the possibility of sex differences in response to AAV8-TNAP-D 10 ; however, there are no established sex differences among patients with HPP, (3) and most previous in vivo studies did not indicate the sex differences in Alpl À/À mice. Therefore, further study is necessary to explore if sex affects phenotype and treatment response in HPP. Second, we did not assess the reproductive capacity of the mice treated with the virus vector and the health of their offspring in our study design. Finally, vector-host interactions such as immunemediated responses and the emergence of neutralizing antibodies (NAb) against the virus capsid and/or transgene product are known as the major challenges that virus-vector-mediated gene therapy has to overcome before its clinical application. (74)(75)(76) We did not perform a dose-ranging study to determine the optimal dose of virus vector that can minimize the risk of immune responses and can correct the skeletal phenotype of the Alpl À/À mice. We did not analyze the presence of circulating NAb against AAV8 or TNAP-D 10 in our study. These limitations will be the basis for continuing studies.
In conclusion, a single intramuscular injection of AAV8-TNAP-D 10 successfully corrected the skeletal and dentoalveolar phenotypes of otherwise lethal HPP mice without apparent complications in other tissues. Our results support gene therapy as a safe and effective approach and a potential alternative to ERT for the treatment of HPP.

Disclosures
All authors state that they have no conflicts of interest. scanning and Dr Noriko Miyake (Nippon Medical School) for the preparation of AAV8-TNAP-D 10 .
Authors' roles: YK, BLF, FFM, FA, and SN contributed to the conception and design of the study, data acquisition and analysis, and manuscript drafting and revision. KM contributed to the production of AAV8-TNAP-D 10 and the conception and design of the study. JLM contributed to the conception and design of the study and manuscript drafting and revision. All authors gave final approval of the submitted version and agreed to be accountable for all aspects of the work.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.