Overexpression of Phex in osteoblasts fails to rescue the Hyp mouse phenotype.

Inactivating mutations of Phex, a phosphate-regulating endopeptidase, cause hypophosphatemia and impaired mineralization in X-linked hypophosphatemia (XLH) and its mouse homologue, Hyp. Because Phex is predominantly expressed in bone and cultured osteoblasts from Hyp mice display an apparent intrinsic mineralization defect, it is thought that reduced expression of Phex in mature osteoblasts is the primary cause of XLH. To test this hypothesis, we studied both targeted expression of Phex to osteoblasts in vivo under the control of the mouse osteocalcin (OG2) promoter and retroviral mediated overexpression of Phex in Hyp-derived osteoblasts (TMOb-Hyp) in vitro. Targeted overexpression of Phex to osteoblasts of OG2 Phex transgenic Hyp mice normalized Phex endopeptidase activity in bone but failed to correct the hypophosphatemia, rickets, or osteomalacia. OG2 Phex transgenic Hyp mice did exhibit a small, but significant, increase in bone mineral density and dry ashed weight, suggesting a partial mineralization effect from restoration of Phex function in mature osteoblasts. Similarly, retroviral mediated overexpression of Phex in TMOb-Hyp osteoblasts restored Phex mRNA levels, protein expression, and endopeptidase activity but failed to correct their intrinsic mineralization defect. In addition, we failed to detect the Phex substrate FGF-23 in osteoblasts. Taken together, these in vivo and in vitro data indicate that expression of Phex in osteoblasts is not sufficient to rescue the Hyp phenotype and that other sites of Phex expression and/or additional factors are likely to be important in the pathogenesis of XLH.

Inactivating mutations of Phex, a phosphate-regulating endopeptidase, cause hypophosphatemia and impaired mineralization in X-linked hypophosphatemia (XLH) and its mouse homologue, Hyp. Because Phex is predominantly expressed in bone and cultured osteoblasts from Hyp mice display an apparent intrinsic mineralization defect, it is thought that reduced expression of Phex in mature osteoblasts is the primary cause of XLH. To test this hypothesis, we studied both targeted expression of Phex to osteoblasts in vivo under the control of the mouse osteocalcin (OG2) promoter and retroviral mediated overexpression of Phex in Hyp-derived osteoblasts (TMOb-Hyp) in vitro. Targeted overexpression of Phex to osteoblasts of OG2 Phex transgenic Hyp mice normalized Phex endopeptidase activity in bone but failed to correct the hypophosphatemia, rickets, or osteomalacia. OG2 Phex transgenic Hyp mice did exhibit a small, but significant, increase in bone mineral density and dry ashed weight, suggesting a partial mineralization effect from restoration of Phex function in mature osteoblasts. Similarly, retroviral mediated overexpression of Phex in TMOb-Hyp osteoblasts restored Phex mRNA levels, protein expression, and endopeptidase activity but failed to correct their intrinsic mineralization defect. In addition, we failed to detect the Phex substrate FGF-23 in osteoblasts. Taken together, these in vivo and in vitro data indicate that expression of Phex in osteoblasts is not sufficient to rescue the Hyp phenotype and that other sites of Phex expression and/or additional factors are likely to be important in the pathogenesis of XLH.
Because the discovery that mutations of PHEX, or the Phosphate-regulating gene with homologies to Endopeptidases on the X chromosome, is the genetic defect underlying X-linked hypophosphatemia (XLH) 1 (1)(2)(3)(4), efforts have been underway to determine how this novel endopeptidase regulates phosphorus and mineral homeostasis. Phex is one of six members of the M13 family of zinc-dependent type II cellsurface membrane metalloproteases (5)(6)(7). The presence of renal phosphate wasting secondary to inactivating mutations of the Phex gene suggests that this endopeptidase degrades a novel phosphaturic hormone (referred to as phosphatonin) or inactivates a phosphate-conserving factor (8). Neutral endopeptidase substrates ZAAL-pNA and [Leu]enkephalin, as well as certain parathyroid hormone-related peptides (9 -12), are cleaved by recombinant Phex in vitro, but the biological relevance of these substrates is not certain. Recent studies indicate that FGF-23 may be a physiologically important Phex substrate and a candidate for phosphatonin (14). Not only is FGF-23 cleaved by recombinant Phex in vitro (13), but mutations in the FGF-23 gene product cause the related disorder autosomal dominant hypophosphatemia (14), and FGF-23 induces hypophosphatemia and defective mineralization when administered to mice in vivo (15).
There is significant evidence indicating bone is a physiologically relevant site of Phex expression and is directly involved in the pathogenesis of XLH. Phex is expressed at high levels in osteoblasts and other mineralizing tissues such as teeth and growth plate cartilage (16 -20) where its expression is temporally associated with the formation of mineralized extracellular matrix in cultured osteoblasts (18,20). In addition, available data (21,22) suggest that loss of Phex function in osteoblasts results in a nascent defect that leads to impaired mineralization of extracellular matrix, independent of the hypophosphatemia. Osteoblasts derived from Hyp mice, a murine homologue of XLH, display defective mineralization and other abnormalities in culture (21,22), as well as fail to form mineralized bone after transplantation into normal mice (23). Finally, putative phosphate and mineralization inhibitory activities also have been identified in conditioned media of Hyp osteoblasts (21,24), suggesting that Phex and its substrate are both produced in bone.
Although these data provide compelling evidence that the loss of Phex function in osteoblasts is causally related to the intrinsic abnormality of mineralization, several components of this model have not be substantiated. First, there is no direct evidence that Phex metabolizes endogenous phosphaturic or mineralization inhibitory factors synthesized by osteoblasts. Second, FGF-23, the current best candidate for phosphatonin (14,15), has not been shown to be expressed in bone marrow (14), although its expression in osteoblasts has not been excluded. Third, several studies (25,26) have failed to document nascent defects in Hyp-derived osteoblasts, indicating that the observed abnormalities of cultured osteoblasts may be secondary to the Hyp milieu. Fourth, the observation that parabiosis (27) and cross-kidney transplantation (28) between normal and Hyp mice lead to phosphaturia in the normal animal indicates the presence of extrinsic circulating factors (29) that modulate bone mineralization in the X-linked disorder, either directly or indirectly through the induction of hypophosphatemia. Finally, bone marrow transplantation via the intraperitoneal route, which does not normalize Phex expression in bone, partially rescues the hypophosphatemia in Hyp mice (30). All of these studies indicate that factors extrinsic to the osteoblast might be responsible for the mineralization defect in XLH. No studies to date, however, have examined whether the apparent intrinsic abnormalities of mineralization in Hyp-derived osteoblasts are corrected by restoration of Phex expression and function in osteoblasts.
In the present investigation, we directly examined whether Phex deficiency is directly associated with impairment of osteoblast-mediated mineralization by two complementary approaches. We used a retroviral vector to overexpress a functional Phex cDNA in osteoblasts derived from Hyp mice and the osteocalcin (OG2) promoter to achieve targeted overexpression of Phex to Hyp osteoblasts in vivo. Neither of these approaches corrected the Hyp-related mineralization abnormalities, and the targeted expression of Phex to mature osteoblasts in bone did not rescue the hypophosphatemia. Thus, our findings fail to support the simple hypothesis that loss of Phex in osteoblasts is primarily responsible for the Hyp phenotype.

EXPERIMENTAL PROCEDURES
Reagents-␣-Minimum essential medium, Dulbecco's modified Eagle's medium/Ham's F-12, Dulbecco's modified Eagle's medium, penicillin/streptomycin solution, Hanks' balanced salt solution, and Trizol Reagent for single-step isolation of total RNA from cells were obtained from Invitrogen. Fetal bovine serum was obtained from HyClone Laboratories (Logan, UT). Pronase-E, ascorbic acid, ␤-glycerophosphate, bovine serum albumin, and alkaline phosphatase kit were purchased from Sigma. Bio-Rad reagent for protein assay was obtained from Bio-Rad. Oligonucleotide primers were synthesized at the Duke University DNA Core Facility. We used Phex-specific rabbit antisera that recognize the C-terminal end of mouse Phex (17). Other antibodies used in these studies included anti-rabbit IgG antibodies conjugated to horseradish peroxidase (Santa Cruz Biotechnology, Santa Cruz, CA). Recombinant Phex was produced in Sf9 cells as described previously (17). The enhanced chemiluminescence detection kit (PerkinElmer Life Sciences) was used to detect horseradish peroxidase. Nitrocellulose membranes (0.45 m) and other chemicals used for SDS-PAGE and Western blotting were purchased from Bio-Rad.
Cell Culture-E86 cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum as described previously (9). TMOb-N1 and TMOb-Hyp immortalized cells derived from normal and Hyp mice calvaria were grown for periods of up to 14 days in ␣-minimum essential medium containing 10% fetal bovine serum with supplementation with 5 mM ␤-glycerophosphate and 25 g/ml ascorbic acid with media being replaced every 3 days as reported previously (21).
Retroviral Constructs-The modular retroviral vector pLuvM was provided by Dr. Clay Smith at Duke University (31). We modified this construct by adding an internal ribosome entry site (IRES) and enhanced green fluorescent protein coding region (CLONTECH Laboratories, Inc., Palo Alto, CA). The new construct, called pLuv-IRES-GFP, is capable of concurrently overexpressing a gene of interest and enhanced green fluorescent protein from a single bicistronic mRNA. The Phex cDNA was amplified by forward primer 5Ј-GAAGCAGAAACAGG-GAGCAC and reverse primer 5Ј-TGCAGCGGCCGCTACCAGAGTCG-GCAAGAATC using Vent DNA polymerase (New England Biolabs, Beverly, MA). The Phex retrovirus construct, pLuv-Phex-IRES-GFP, was made by cloning the PCR product from second codon of Phex cDNA between blunted NcoI and NotI sites of the pLuvIRES-GFP construct. In addition, we used additional constructs pLuv-Phex-WT and pLuv-Phex-3ЈM as reported previously (9).
Development of Retroviral Producer Cell Lines and Transduction-Separate ecotropic packaging E86 cells were created with the empty retrovirus vector pLuv-IRES-GFP and pLuv-Phex-IRES-GFP as described previously (17). We also used previously created E86 producer cells transfected with either the empty retrovirus vector pLuvM, pLuv-Phex-WT, pLuvPhex-3ЈM, or the pLuvGFP (17). Western blot analysis with the Phex-specific rabbit antisera confirmed Phex protein expression in E86 producer cells. Retroviral supernatant for infecting osteoblasts was collected and stored at Ϫ70°C (31). 100,000 TMOb-Nl or TMOb-Hyp cells were plated in 60-mm dishes 24 h prior to infection. Media were aspirated the following day, and cells were incubated in 1 ml of retroviral supernatant containing 4 g/ml Polybrene. For pLuvM, pLuvPhex-WT, or pLuvPhex-3ЈM transductions, we performed parallel studies with pLuvGFP to evaluate infection efficiency. With studies using pLuv-Phex-IRES-GFP, we used production of GFP from the bicistronic cDNA to monitor infection efficiency and to select cells. Fluorescence-activated cell sorter analysis was performed on a FASCcan apparatus (Becton Dickinson, San Jose, CA), and for all analyses Ͼ10,000 events were collected. Non-transformed cells were included as negative controls for background fluorescence. Cursors were set to define positive cells so that Ͻ1% of the negative control cells were detected as positive. The initial infection efficiency was 70 -80%. We collected and cultured the GFP-positive cells from pLuv-Phex-IRES-GFP transduction to obtain 100% Phex-expressing cells.
Preparation of Membranes Containing Phex-Membranes solubilized with 1% n-dodecyl ␤-D-maltoside were collected by centrifugation as described previously (9). The protein content of each sample was determined by the Nano-Orange TM Protein Quantitation kit (Molecular Probes, Eugene, OR).
SDS-PAGE and Western Blot Analyses-Immunoblot analysis was carried out by modifications of methods described previously (32). The specified amounts of membrane proteins were dissolved in SDS gel loading buffer (500 mM Tris-HCl, pH 6.8; 8.5% SDS; 27.5% sucrose; 0.03% bromphenol blue). Separated proteins were transferred to a nitrocellulose membrane (0.45 m, Bio-Rad) over a 30-min period at 2.5 mA/cm 2 at room temperature using a semi-dry blotting system (Millipore Corp., Chicago, IL). Immunoblotting was performed using affinitypurified rabbit anti-Phex polyclonal antisera (1:2000 dilution). In addition, after washing the blots three times with TBST at room temperature for 40 min each, immunoreactivity was detected by a chemiluminescence system.
Assessing Phex Enzyme Activity-We assessed Phex activity by modifications of methods described previously (11). We used membrane fractions from the Sf-9 cells (50 g) or TMOb-Hyp osteoblasts (100 g) expressing vector or various Phex constructs as well as homogenized calvaria-derived protein (80 g) from 6-week-old non-transgenic normal (ϩ/ϩ), Hyp (Ϫ/Ϫ), OG2-Phex transgenic, and OG2-Phex Hyp mice (see below). These protein samples were incubated with 50 M Z-Ala-Ala-Leu-p-nitroanilide (ZAAL-pNA) (Bachem Biosciences, Inc., King of Prussia, PA) in 100 l of 100 mM MES, pH 6.5, for 1 h at 37°C. After completion of the initial incubation, the reaction mixture was further incubated with 0.4 milliunits of leucine aminopeptidase (Sigma) for 20 min at 37°C. The reaction was stopped by the addition of 100 mM EDTA, and the absorbance was measured at 405 nm after centrifugation. In some studies membrane fractions were preincubated with 100 mM EDTA for 30 min before the addition of ZAAL-pNA. Phex endopeptidase activity also was assessed using [Leu]enkephalin (Bachem Biosciences, Inc., King of Prussia, PA) using modifications of methods described previously (9).
Alkaline Phosphatase Activity-We analyzed alkaline phosphatase in cell layers by calorimetric assay of enzyme activity using alkaline phosphatase kit (Sigma) as described previously (21). The alkaline activity was expressed as nanomoles of substrate transferred per min per g of DNA.
Mineralization Assays-The formation of in vitro mineralization nodules was determined by alizarin red-S histochemical staining, and mineralization was quantified by modification of methods described previously (21). Briefly, the stained matrix was washed with water and phosphate-buffered saline, and the dye was destained with 10% (w/v) cetylpyridinium chloride in 10 mM sodium phosphate, pH 7.0, for 15 min at room temperature. The alizarin red-S was quantified at 562 nm.
Generation of Transgenic Mice-The transgene was constructed in the pW1 vector, which contains a multiple cloning site and an SV40 intron plus polyadenylation site flanked by the rare restriction enzyme sites Tth111I and SfiI. A1.3-kb fragment of the mouse OG2 promoter (34) was released from p1316-luc by restriction endonuclease digestion and subcloned into the blunted SacI site and SalI site of pW1, thereby generating pW1 5ЈOG2. pW1 5ЈOG2-Phex was constructed by releasing a 2.2-kb fragment from pBS-Phex (18) by digestion with XbaI and SpeI and subcloned downstream of the osteocalcin promoter into the same sites of pW1. Plasmid DNA was first isolated using the Qiafilter Maxi Kit (Qiagen, Valencia, CA), followed by CsCl banding. OG2-Phex DNA for microinjection was released from the vector backbone using Tth111I and SfiI, separated by agarose gel, and then gel-purified using Qiaex II gel extraction (Qiagen, Valencia, CA) according to the manufacturer's instructions. OG2-Phex DNA was further purified using the EndoFree Kit (Qiagen, Valencia, CA) to remove endotoxins and quantified using the PicoGreen Assay Kit (Molecular Probes, Eugene, OR). Transgenic mice were made by the Duke Transgenic Facility by microinjection of C57Bl6F1/J fertilized mouse eggs with DNA at a concentration of 2-3 ng/l according to standard techniques (35). Mice that carried the transgene were identified by PCR using tail DNA, followed by Southern blot analysis. For Southern blots we used 10 g of DNA from each mouse digested with KpnI and SpeI separated on 1% agarose gel and hybridized by 32 P-radiolabeled Phex cDNA probe as described previously (21).
We obtained heterozygotic female mice (Hyp ϩ/Ϫ) with 3Ј deletions of Phex gene from The Jackson Laboratory (Bar Harbor, ME). Mice were maintained and used in accordance with recommendations in the Guide for the Care and Use of Laboratory Animals, prepared by the Institute on Laboratory Animal Resources, National Research Council (Department of Health and Human Services Publication NIH 86-23), and by guidelines established by the Institutional Animal Care and Use Committee of Duke University.
Dry Ash Weights of Femurs-Femurs collected from 12 week-old mice were cleaned of muscle and dried to constant weight at 110°C. The dried femurs were ashed overnight at 600°C and weighed. The results are expressed as a ratio of ashed to dry weight. This value represents the total bone mineral content (37).
Measurement of Bone Mineral Density-We assess the bone mineral density (BMD; g/cm 2 ) of the whole skeleton and femur in 3-, 6-, and 12-week-old mice with a LUNAR PIXIMUS bone densitometer (Lunar Corp., Madison, WI). The instrument was calibrated before each scanning session using a Phantom with known BMD according to the manufacturer's guidelines. The animals were anesthetized by an intraperitoneal injection of ketamine (90 g/g body weight) and xylazine (10 g/g body weight) and then were placed in prone position on the specimen tray to scan the entire skeleton.
Analysis of Nondecalcified Bone-Skeletons of mice were prelabeled with calcein (Sigma C-0875, 30 g/g body weight, subcutaneous injection) 1 and 3 days prior to collection of tibias. Femurs and tibias were removed from 13-week-old mice, fixed in 70% ethanol, prestained in Villanueva stain (38), and processed for methyl methacrylate embedding. Five-m sections were stained with Goldner's stain and analyzed under transmitted light, and 10-m Villanueva pre-stained sections were evaluated under fluorescent light as reported previously (38) by our laboratory.
Skeletal Radiography-Six-week-old mice were radiographed using a Hewlett-Packard Faxitron 43807 and X-Omat film.
RT-PCR Analysis-We isolated total cellular RNA by a single-step method using Trizol reagent as described previously (39). RNA samples were pretreated with DNase to remove any contaminating DNA and were quantified by absorbance at 260 nm. RT-PCR was performed using a two-step RNA PCR kit (PerkinElmer Life Sciences). 2.5 g of DNasetreated total RNA was reverse-transcribed into cDNA in a total volume of 50 l with random primers. The reverse transcription reaction was incubated at 42°C for 15 min. The resulting cDNA was PCR-amplified using various primer sets. To identify Phex expression, we performed RT-PCR using the following primers: Mϩ786 forward (5Ј-TAAT-AGCTCTCGAGCTGAACATGA-3Ј) and Mϩ1983 reverse (5Ј-TATC-CATTTCCTGTAAGCCC-3Ј) to amplify the 3Ј end of Phex, Mϩ786 forward, 5Ј-TAATAGCTCTCGAGCTGAACATGA, and V5 reverse, 5Ј-GAAGATCTCACGTAGAATCGAGACCGAG, to amplify Phex-3ЈM. For evaluating OG2-Phex transgene expression, we used the primers Mϩ1156 forward, 5Ј-GGCTAGAATTCTCAAGGGTA, and pW1-Spe reverse, 5Ј-GATCTGCATCACTAGTAACG located in the 3Ј-untranslated region of the transcript. In addition, we used previously reported (21) primer sets and RT-PCR to characterize osteoblast gene expression TMObs-Hyp infected by pLuv-IRES-GFP and pLuv-Phex-IRES-GFP. The conditions of PCR were 2 min at 94°C, followed by 38 cycles of 94°C for 1 min, 60°C for 1 min, 72°C for 1-2 min, and 72°C for 10 min for final extension. Samples without reverse transcriptase treatment were analyzed as controls. All products were separated by agarose gel electrophoresis and stained with ethidium bromide. We also used primers mFGF23 forward, 5Ј-ATTCTTGCGGCCGCTGTGCAATGCTAGG-GACCT-3Ј, and mFGF23 reverse, 5Ј-ATTCTTGGGCCCGACGAACCT-GGGAAA-3Ј, to amplify FGF-23. In these studies the pGEM-FGF23 containing the full-length FGF-23 cDNA was used as a positive control (40). The products from these experiments were separated by agarose gel electrophoresis, stained with ethidium bromide, and confirmed by Southern blot using a 32 P-labeled full-length FGF-23 cDNA.
In Situ Hybridization-Tibias were fixed in 4% paraformaldehyde/ phosphate-buffered saline for 4 h at 4°C. Specimens were embedded in paraffin and sectioned at 5 m. In situ hybridization was performed using biotin-labeled Phex antisense and sense riboprobes (41). The Phex probe is a 432-nucleotide fragment of the 1822-2253 region. Hybridization was performed for 16 h at 42°C, and washes were performed at 55°C. Hybridization was detected by GenPoint CSA kit (Dako Corp., Carpenteria, CA) (42).
Statistics-We evaluated differences between groups by one-way analysis of variance (43). All values are expressed as mean Ϯ S.E. All computations were performed using the Statgraphic statistical graphics system (Rockville, MD).

Restoration of Full-length Phex Expression in Hyp Mouse
Osteoblasts-First, we examined Phex transcript expression by RT-PCR analysis. Consistent with previous observations (21), TMOb-Hyp cells have a 3Ј deletion of the Phex gene resulting in the absence of an RT-PCR product using reverse primers designed to amplify both the 5Ј and 3Ј ends of Phex (Fig. 1a, 2nd  lane). Retroviral mediated overexpression of the full-length Phex cDNA restored Phex mRNA (Fig. 1a, 4th lane) in TMOb-Hyp cells, whereas the full-length Phex transcripts were not detected in vector-infected TMOb-Hyp cells (Fig. 1a, 3rd lane). Next, we evaluated protein expression by Western blot analysis using a polyclonal anti-Phex antibody (Fig. 1b). We were unable to detect Phex expression in wild-type TM-Ob osteoblasts (data not shown), consistent with our prior studies that required immunoprecipitation to detect Phex in normal osteoblasts using this antibody (9). The failure to detect Phex in wild-type osteoblasts is likely related to low abundance, because we detected Phex using this antibody in positive control Sf9 cells expressing high levels of recombinant wild-type Phex (Fig. 1b,  3rd lane). Moreover, consistent with attaining high levels of Phex expression in TMOb-Hyp cells transduced with pLuv-Phex-IRES-GFP, we observed in these cells the presence of the expected 100-kDa Phex protein representing the glycosylated monomer (Fig. 1b, 2nd lane). No endogenous Phex protein was detected in the TMOb-Hyp cells transfected with pLuv-IRES-GFP vector alone (Fig. 1b, 1st lane). In addition, we demonstrated that the level of the Phex transcripts remained elevated in retrovirally infected TMOb-Hyp cells throughout a 14-day culture period (Fig. 1c), a period when these cells undergo a temporal sequence of osteoblastic maturation (see below).
Restoration of Phex Endopeptidase Activity to Membranes of Hyp Osteoblasts-Recent studies (9,11) indicate that Phex hydrolyzes several neutral endopeptidase substrates, including ZAAL-pNA and Leu[enkephalin], which can be used to assess Phex endopeptidase activity. We initially confirmed that recombinant Phex produced in Sf9 cells cleaves ZAAL-pNA (Fig.  2a), and we then used this substrate to demonstrate that the overexpression of Phex in Hyp mouse osteoblasts imparted endopeptidase activity to levels equal to that of normal osteoblasts (Fig. 2b). Sf9 membranes expressing recombinant Phex cleaved ZAAL-pNA in an EDTA-dependent fashion, whereas the 3Ј-truncated Phex mutant lacking the catalytic domain resulted in cleavage not different from vector-transfected control cells (Fig. 2a). We found that the EDTA-dependent cleavage of ZAAL-pNA was restored in Hyp mouse osteoblasts membranes overexpressing Phex to levels comparable with that of normal 14-day-old osteoblasts, whereas Hyp osteoblasts transduced with the empty retroviral vector displayed low activity, consistent with the presence of other endopeptidases known to be present in these cells (Fig. 2b). Similar results were obtained with [Leu]enkephalin as a substrate (Fig. 2c). [Leu]enkephalin alone migrated as a single peak (Fig. 2c, upper panel) and as a faster migrating peak after incubating [Leu]enkephalin with Hyp membranes overexpressing Phex (Fig. 2c, middle panel). Incubation with membrane preparations from vector-transfected Hyp membranes did not result in cleavage of [Leu]enkephalin (Fig. 2c, lower panel), similar to our prior reports with recombinant Phex (9). Thus, in addition to restoring high levels of Phex protein expression, we normalized Phex endopeptidase activity, a requirement for correcting this haplo-insufficient state.
Phenotype Characteristics of Hyp Osteoblasts Overexpressing Phex-Next, we examined if retroviral mediated transduction of Hyp-derived osteoblasts retained their capacity to undergo a temporal sequence of maturation characterized by an initial period of replication and subsequent expression of osteoblastic characteristics. Similar to our previous report, Hyp-derived osteoblasts infected with the empty vector as well as nontransfected osteoblast derived from normal and Hyp mice (21) underwent a progressive period of cell proliferation that was characterized by increments in DNA content as an index of cell number (Fig. 3a). The overexpression of Phex did not affect the growth characteristics of Hyp-derived osteoblasts. During the period of rapid cell growth, both vector alone and Phex-expressing TMOb-Hyp cells expressed low levels of alkaline phosphatase (Fig. 3b), consistent with their immaturity. The slowing of cell replication was associated with a significant increase in the expression of alkaline phosphatase activity to similar degrees in both TMOb-Hyp vector alone and Phex-expressing cells (Fig.  3b). Similarly, the process of osteoblast maturation in both vector and Phex-expressing Hyp-derived cells was marked by similar levels of osteocalcin, osteopontin, and type 1 collagen transcripts in 14-day-old cultures (Fig. 3c).
Failure to Correct the Impaired Mineralization in TMOb-Hyp Osteoblast Cultures by Overexpression of Phex-In ensuing experiments, we assessed mineralization in normal and Hyp mouse osteoblasts using alizarin red-S histochemical staining. We confirmed our previous report (21) that TMOb-Nl cells exhibited marked increments in alizarin red-S-stained mineralization nodules by day 14 of culture (Fig. 3, d and e). Similar to our previous report (21) of impaired mineralization in mature Hyp mouse derived osteoblasts, the Hyp osteoblasts infected with the vector also exhibited minimal alizarin red-S staining characterized by only ill-defined patches with limited dye uptake and the absence of discrete mineralization nodules (Fig. 3, d and e). Overexpression of Phex and restoration of Phex endopeptidase activity (Figs. 1 and 2) failed to correct the impaired mineralization in Hyp osteoblasts (Fig. 3d, middle  panel). Indeed, the levels of alizarin red-S staining were not statistically different between Hyp osteoblasts infected with the vector alone and Hyp osteoblasts overexpressing Phex (Fig.  3e). In addition, we showed that the impaired mineralization in Hyp mouse derived osteoblasts was not attributable to differences in cellular composition of the cultures, as clonal cell lines obtained from the parent TMOb-Hyp cultures failed to mineralize after retroviral mediated transfection of Phex (data not shown).
As additional controls, we demonstrated that neither the overexpression of wild-type Phex nor its inactive mutant affect mineralization of normal osteoblasts. We achieved high levels of retroviral mediated Phex expression in 4-day-old normal TMOb cells (Fig. 4a, left panel, 2nd lane) as well as attained high levels of expression of the 3Ј-truncated Phex mutant in normal TMOb-Nl cells (Fig. 4a, right panel, 2nd  lane). As documented previously (18), normal osteoblasts display a maturation-dependent up-regulation of Phex, with little expression at 4 days and significant expression at 14 days of culture (Fig. 4a, middle panel, 1st and 2nd lanes). Regardless of whether they were transduced with vector alone, full-length Phex, or the 3Ј-truncated Phex constructs, normal TMOb osteoblast cultures exhibited indistinguishable maturation-dependent mineralization (Fig. 4, b and c). In addition, in separate studies we demonstrate that retroviral mediated overexpression of GFP did not affect osteoblast growth, differentiation, or mineralization (data not shown).
Osteoblasts Do Not Express the Phex Substrate FGF-23-Current data indicate that FGF-23 is a substrate for Phex (13,15). Therefore, we examined whether FGF-23 might be present in osteoblasts and differentially expressed in osteoblasts derived from Hyp mice. By using RT-PCR analysis with FGF-23specific primers, however, we failed to identify FGF-23 transcripts in either normal or Hyp-derived osteoblasts (Fig. 5).
Generation of Hyp Transgenic Mice Expressing Phex in Osteoblasts-Because we failed to correct the Hyp osteoblast phenotype in culture and failed to identify FGF-23 in osteoblasts, we reasoned that factors required for Phex function may be lacking in the in vitro environment. Because Phex is predominantly expressed in mature osteoblasts and is temporally coexpressed with osteocalcin (20), we next determined whether restoring Phex expression to mature Hyp osteoblasts in vivo using the osteocalcin gene 2 (OG2) promoter corrected the mineralization defect as well as increased serum phosphorus levels in Hyp mice. We constructed transgenic mice expressing full-length Phex under the control of a 1.3-kb fragment of the mouse OG2 promoter (Fig. 6a). A total of five transgenic mouse lines were obtained (Fig. 6b). We selected two lines with the highest level of expression for further study (founder lines 1 and 2). Male founder mice expressing Phex, designated OG2-Phex, were bred with heterozygous female Hyp mice (X hyp X) to create offspring that co-express the full-length Phex transgene and mutant Hyp allele, designated OG2-Phex-Hyp. The OG2-Phex-Hyp mice were identified and compared with nontransgenic and transgenic normal and non-transgenic Hyp littermates (Fig. 6c). To date we have analyzed results from 68 mice from line 2, including 19 nontransgenic normals, 21 OG2-Phex mice normals, 11 non-transgenic Hyp, and 17 OG2-Phex-Hyp and a more limited number from line 1. Expression of the Phex transgene was predominantly restricted to bone and bone marrow, with minimal expression in brain and no expression in other tissues tested by RT-PCR from 6-week-old animals (Fig. 6d).
Restoration of Phex Endopeptidase Activity in Calvaria of OG2-Phex-Hyp Mice-Similar to our studies of Phex activity in isolated osteoblasts (Fig. 2), we used ZAAL-pNA (Fig. 6e) to evaluate Phex activity in calvaria. In non-transgenic normal (ϩ/ϩ) mice, calvarial homogenates cleaved ZAAL-pNA, whereas the non-transgenic Hyp calvaria displayed low levels of activity (Fig. 6e). This nonspecific cleavage likely represents contaminating endopeptidases. In contrast, we found that calvaria from transgenic OG2-Phex had a marked increase in activity, consistent with the possible additive effect of endogenous Phex and the Phex transgene. Finally, calvaria from OG2-Phex-Hyp mice degrades ZAAL-pNA similar to non-transgenic normals (Fig. 6e), indicating that Phex expression by the osteocalcin promoter normalized endopeptidase activity in bone derived from Hyp mice.
Failure of Phex Overexpression to Affect Serum Phosphorus or Bone Mineral Density-Phex expressing transgenic OG2-Phex mice lines 1 and 2 were indistinguishable from their wild-type littermates (ϩ/ϩ), regardless of sex. Detailed analysis was limited to the higher expressing OG2-Phex line 2. Transgenic OG2-Phex mice displayed normal serum phosphorus levels that decreased with age (Fig. 7a). Bone mass as assessed by DEXA (Fig. 7b), and a dry ashed weight (Fig. 7c) in OG2-Phex mice were identical to non-transgenic normal (ϩ/ϩ) littermates. Non-transgenic Hyp littermates exhibited the expected reduction in serum phosphorus during the 12-week period of observation (Fig. 7a), as well as lower total bone density (data not shown) and decreased bone density and dry ashed weight of the femur compared with non-transgenic (ϩ/ϩ) and OG2-Phex transgenic littermates (Fig. 7, b and c). Despite restoring Phex expression (see Fig. 7a, below) and endopeptidase activity (Fig. 6e) to osteoblasts in bone, OG2-Phex-Hyp transgenic mice exhibited hypophosphatemia identical to nontransgenic Hyp mice littermates (Fig. 7a). OG2-Phex transgenic Hyp mice derived from founder line 1 also were hypophos-phatemic (average phosphorus 4.5 mg/dl) and had low bone mass. Although total bone mineral density was not significantly different from non-transgenic littermates (data not shown), we did observe a slight, but statistically significant, increase in bone mineral density and dry ashed weight of the femur in OG2-Phex-Hyp compared with non-transgenic Hyp mice, but this value remained significantly less that non-transgenic normal and OG2-Phex transgenic littermates (Fig. 7, b  and c). There were no differences between males and females in the respective groups.
Radiographic in Situ and Histologic Characterization of OG2 Phex Transgenic Mice-Overexpression of Phex did not affect the radiographic appearance of skeleton in OG2-Phex mice compared with non-transgenic normals (Fig. 8a, compare 1st  and 2nd columns). In contrast, the skeletal radiographs of 6-week-old non-transgenic Hyp mice showed evidence of rickets (Fig. 8a, 3rd column). The caudal vertebrae demonstrated smaller vertebrae and an increase in the apparent distance between the vertebral bodies. The knees also demonstrated widening and irregularity of the growth plate. OG2-Phex-Hyp mice were indistinguishable from their non-transgenic Hyp littermates (Fig. 8a, compare the 3rd and 4th columns).
To document that we successfully targeted the expression of Phex to osteoblasts, we performed in situ hybridization studies in 4-day-old normal (ϩ/ϩ), Hyp, and OG2-Phex-Hyp male mice to confirm the localization of the Phex transgene in bone and to compare the distribution of the Phex transgene to endogenous Phex (Fig. 8b). We found, similar to previous reports (20), that endogenous Phex is expressed at high levels in osteoblasts and osteocytes of trabecular bone (Fig. 8b, upper left panel) and in lower abundance in hypertrophic chondrocytes in the growth plate (data not shown). In contrast, Phex was absent in nontransgenic male Hyp mice at all sites (Fig. 8b, upper middle  panel). The OG2-driven expression of Phex in transgenic Hyp mice was similar to that of endogenous Phex in trabecular bone, where it was expressed in osteoblasts and osteocytes (Fig. 8b,  upper right panel). Transgenic Hyp mice, however, did not express Phex in chondrocytes (data not shown). Sections probed with the sense riboprobe gave only low level background staining in all groups (Fig. 8b, lower panels).
Non-decalcified bone sections from tibias of non-transgenic normals and OG2-Phex mice were indistinguishable by qualitative analysis, including normal appearing bone volume, osteoid thickness, and mineralization (Fig. 8c, compare 1st and  2nd column). In contrast, non-transgenic Hyp mice exhibited profound defects in mineralization characterized by hyperosteoidosis and defective mineralization as evidence by the complete absence of double fluorescent bone labels (Fig. 8c, 3rd  column). Only diffuse, non-quantifiable label was present beneath widened osteoid seams (Fig. 8c, 3rd column, 3rd panel). In addition, the growth plate and underlying metaphyseal bone of non-transgenic Hyp mice was disorganized and exhibited impaired mineralization. The bone histology of OG2-Phex-Hyp was indistinguishable from their non-transgenic Hyp mice littermates and exhibited histologic evidence of severe osteomalacia and rickets (Fig. 8c, 4th column). Thus, despite the successful overexpression of the Phex transgene in a pattern overlapping that of endogenous Phex, we failed to observe any effect of the Phex transgene on bone histology in either OG2-Phex transgenic or OG2-Phex Hyp mice. DISCUSSION Because Phex is predominantly expressed in bone and osteoblasts derived from Hyp mice have an apparent intrinsic mineralization defect, we (8) and others (16,17,20) favored the hypothesis that diminished Phex expression in osteoblasts was primarily responsible for the pathogenesis of the XLH and the Hyp phenotype. Direct evidence that co-expression of Phex and  (14) were amplified from total RNA derived from normal osteoblasts (TMOb-Nl) 4 days following transduction with the pLuvM empty vector, pLuv-Phex-WT, or pLuv-Phex-3ЈM. Amplified products were separated by agarose gel electrophoresis and stained with ethidium bromide. ␤-Actin was amplified as a control. Results were compared with non-transduced TMOb-Nl osteoblasts grown for 4 and 14 days (left panel). No transcript could be amplified from RNAderived 4-day-old normal osteoblasts, but the predicted size product was obtained from 14-day-old normal osteoblasts, consistent with the maturational increase in endogenous Phex as reported previously (21). its putative substrates in the bone milieu participated in bone mineralization and regulation of systemic phosphate homeostasis, however, was lacking (21,24). The results from the current study indicate that Phex expression in osteoblasts is not sufficient to explain the pathogenesis of XLH. Rather, we found that the targeted overexpression of Phex to osteoblasts in Hyp mice using the osteocalcin promoter failed to correct either the mineralization defect of bone or the systemic hypophosphatemia (Figs. 6 -8). Despite attaining expression of Phex in mature osteoblasts to levels (Fig. 8b) and activity (Fig. 6e) comparable with endogenous Phex, OG2-Phex-Hyp mice exhibited hypophosphatemia (Fig. 7a), radiographic evidence of rickets (Fig. 8a), and histologic evidence of osteomalacia (Fig. 8c) identical to that of non-transgenic Hyp littermates. The overexpression of Phex also had no demonstrable effect in normal OG2-Phex mice, which displayed serum phosphate levels (Fig.  7a) and skeletal morphology (Fig. 8) indistinguishable from non-transgenic littermates, despite higher endopeptidase ac- Hyp male mice. PCR was performed with genomic DNA using primers specific for the transgene or intronic primers flanking exon 19 to detect the 3Ј Phex deletion in male Hyp mice. Normal mice generate a product with the exon 19 primer set, whereas Hyp mice fail to amplify a product with these primers. d, bone-specific expression of the OG2-Phex transgene. Glyceraldehyde-3-phosphate dehydrogenase (G3PDH) amplification was used as an internal control. e, calvarial homogenates (80 g) from 6-week-old mice were analyzed for peptidase activity using ZAAL-pNA as a substrate. Data represent the average of three independent determinations. Data are representative of three independent experiments and are shown with mean Ϯ S.E. Values sharing the same superscript are not significantly statistic different at p Ͻ 0.05. tivity against a synthetic Phex substrate (Fig. 6e). Thus, our findings fail to support the hypothesis that abnormal Phex function in mature osteoblasts plays a primary role in the pathogenesis of hypophosphatemia and raises questions regarding whether Phex has a direct or indirect role in regulating osteoblast-mediated mineralization of bone.
Based on our findings, it is likely that expression of Phex at sites other than those controlled by the osteocalcin promoter is responsible for the persistent accumulation of Phex substrates, resulting in hypophosphatemia and the failure to rescue the Hyp phenotype. Although Phex and osteocalcin expression are concordant in mature osteoblasts and osteocytes (20), the osteocalcin promoter does not express Phex in teeth, cartilage, or other sites where Phex is normally present (20) and initiates expression later during embryogenesis at embryonic day 15.5 compared to day 11 for endogenous Phex. 2 Consequently, persistent abnormalities in the spatial and temporal expression of Phex, and/or the failure to restore the full complement of Phex activities in the whole mouse (which may be the sum of Phex expression in many tissues), may account for the failure to rescue the Hyp phenotype in the current studies. In addition, other genes and environmental factors have been shown to affect the severity of hypophosphatemic rickets (44) and also could impact upon the inability of Phex to rescue the Hyp phenotype.
Other potential explanations for our in vivo findings seem unlikely. Even though recent reports indicate that mutant Phex proteins accumulate in the endoplasmic reticulum (45), we have shown that the 3Ј deletion Phex mutant does not interfere with the function of the wild-type transfected Phex. For example, overexpression of the mutant Phex construct neither interferes with the mineralization of normal osteoblasts in vitro (Fig. 4) nor disrupts the enzymatic activity of wild-type Phex in vitro (9). In addition, our experimental design, using the 1.3-kb OG2 promoter fragment to drive expression of transgenes restricted to mature osteoblasts in vivo (34), achieved the successful restoration of Phex endopeptidase activity in bone (Fig. 6e). Demonstration of Phex activity in osteoblasts and calvaria also lessens the concern that endogenous Osteocalcin may inhibit the activity of co-expressed Phex, as suggested by a 2 S. Liu, R. Guo, Q. Tu, and L. D. Quarles, unpublished observations. recent in vitro studies (12). Finally, because Phex is not expressed in osteoblastic precursors (20), it seems unlikely that targeting Phex to earlier states of the osteoblast lineage would have altered the results.
We did observe a slight, but significant, increase in dry ashed weight and bone mineral density of femurs derived from OG2 Phex-Hyp mice (Fig. 7, b and c). These later findings, representing changes in bone mineralization not detected by conventional radiographic or histologic methods, raise the possibility that Phex may have a role in regulating bone mineralization that is masked by the persistent hypophosphatemia in the OG2 Phex-Hyp mice. It is possible that Phex directly or indirectly regulates the production of matrix proteins and/or local bone substrates, which in turn regulate the mineralization process and account for the small increase in bone mineral density in persistently hypophos-phatemic OG2-Phex-Hyp mice. Proof of an effect independent of hypophosphatemia mediated by Phex direct regulation of osteoblast-mediated mineralization requires identification of additional physiologically important Phex substrates in bone. Nevertheless, the observed increase in bone mass in OG2 Phex-Hyp mice is in keeping with the increased production of factors that regulate mineralization and phosphate transport by cultured osteoblasts derived from Hyp mice (21,23). Recent studies have observed abnormalities of bone extracellular matrix proteins (46) and the accumulation of MEPE (47) in Hyp mice. The failure to detect FGF-23, the only known physiologic Phex substrate and the leading candidate for phosphatonin (13,14,35), in bone (Fig. 5) and bone marrow (15), however, indicates that this phosphaturic factor is not the putative Phex substrate in the local bone environment. Related members of the M13 family of metalloproteases have There is evidence of growth plate widening consistent with rickets. b, in situ hybridization analysis of endogenous Phex expression and OG2-Phex transgene expression in the tibia of 4-day-old wild-type male mouse (1st column), non-transgenic Hyp (2nd column), and OG2-Phex-Hyp (3rd column) incubated with the antisense Phex riboprobe (upper panels) and sense Phex riboprobes (lower panels) as described under "Experimental Procedures." Endogenous Phex is expressed in osteoblasts and osteocytes, whereas Phex is not detected in male Hyp mice. OG2-Phex-Hyp male mice express Phex transcripts in osteoblasts and osteocytes. Hybridization with the sense probe produced only weak nonspecific background staining in all groups (lower panels). c, representative non-decalcified histologic sections of the tibia in 13-week-old non-transgenic normals, OG2-Phex transgenic normals, non-transgenic Hyp, and OG2-Phex transgenic Hyp littermates. Goldnerstained sections (ϫ200 magnification) show the growth plate and metaphyseal bone (upper panel) and endosteal/cortical bone (middle panel). All sections represent tibial bone cut at the mid-sagittal level. In Goldner-stained sections the mineralized bone is blue, and unmineralized osteoid is reddish-brown in color, whereas the growth plate is weakly stained. View under fluorescent light of Villanueva-stained sections (ϫ200) of trabecular bone (lower panel) shows the abundant and distinct dual fluorescent labels deposited beneath narrow osteoid seams of bone surfaces of non-transgenic normals and OG2-transgenic mice compared with the diffuse ill-defined single fluorescent labels or unlabeled wide osteoid surfaces in both Hyp and OG2-Phex-Hyp mice, indicating a disorderly deposition of mineral characteristic of osteomalacia. A minimum of 5 animals was evaluated per group. multiple substrates whose tissue-specific actions are derived from their co-localization (8). Therefore, Phex may separately metabolize distinct substrates that regulate phosphaturia and mineralization.
We confirmed previous reports (21) that osteoblasts derived from Hyp mice exhibit the inability to form mineralization nodules in culture (Figs. 3 and 4). However, restoration of Phex expression (Fig. 1) and enzymatic activity (Fig. 2) to Hyp osteoblasts did not restore their capacity to mineralize extracellular matrix in vitro (Fig. 3) under culture conditions supporting mineralization in normal osteoblasts (Fig. 4). The inability to rescue the Hyp osteoblastic phenotype by Phex overexpression in vitro is consistent with prior studies where co-culture of cells expressing Phex also failed to correct the mineralization defect in Hyp osteoblasts (21). The current negative findings are not due to the effects of the retroviral transduction or inadequate restoration of endopeptidase activity. Moreover, the presence of a truncated Phex in Hyp osteoblast does not interfere with the restoration of Phex function, because overexpression of a 3Ј-truncated mutant Phex failed to disrupt mineralization of normal osteoblasts (Fig. 4). It also is unlikely that aberrant temporal and/or the excessive amounts of retroviral mediated Phex expression or GFP could have influenced our results in vitro, because transduction of normal osteoblasts with the retroviral Phex construct with and without GFP did not affect their ability to mineralize (Fig. 4). We cannot exclude, however, potential variability resulting from differences in the cell culture models used to assess mineralization by alizarin red staining. Indeed, a preliminary report from another laboratory (48), unlike our in vitro studies, shows that overexpression of Phex in Hyp-derived osteoblasts, although not sufficient to fully normalize mineralization, results in partial rescue of their mineralization capacity.
Regardless, the in vitro studies of Hyp-derived osteoblasts indicate a more complex pathogenesis of the defective mineralization in cultured osteoblasts. There are differences in the gene expression profiles in osteoblastic cultures compared with bone in Hyp mice, indicating that these culture models do not fully mimic the in vivo state (21,46). In addition, the exposure to hypophosphatemia or the Hyp milieu may somehow limit the expression of an accessory factor necessary for Phex function and/or lead to deficiencies in the full complement of genes necessary for mineralization. There is a precedent for hypophosphatemia to induce a similar intrinsic mineralization defect in osteoblasts derived from mice in which the renal sodium-dependent phosphate transporter has been ablated (49). Extracellular phosphate also may alter osteoblast gene expression through its actions to modulate nuclear export of the osteoblast transcriptional regulator Cbfa1 in bone cells (33). Identification of the potential Phex substrates in bone and these modulating factors will be necessary to unravel the relative contribution of local and systemic regulation of in osteoblast-mediated mineralization.
In conclusion, our current findings fail to support the simple hypothesis that the lack of Phex in Hyp osteoblasts is directly responsible for the impaired mineralization and abnormalities in systemic phosphate homeostasis. Rather, other sites and/or temporal aspects of Phex expression appear to be physiologically important in the metabolism of the phosphaturic factor phosphatonin. It is likely that the successful rescue of the Hyp phenotype will require restoration of normal Phex activity in one or more of these additional sites. In addition, we failed to establish a cause and effect relationship between Phex expression in osteoblasts and their ability to form a mineralized extracellular matrix in culture, indicating that hypophosphatemia has a predominant role in the defective mineralization. Additional studies that restore endogenous Phex expression in transgenic animals (possibly by using Phex promoter) and/or selectively disrupt Phex by tissue-specific targeted deletion strategies will be necessary to establish a cause and effect relationship between the site of Phex expression and the Hyp phenotype. In addition, confirming the identify of phosphatonin, as well as the identification of possible bone-derived substrates for Phex and/or phosphate-dependent accessory factors that regulate the mineralization process, will be important in unraveling the complex pathogenesis of XLH.