Targeted Disruption of the Osteoblast/Osteocyte Factor 45 Gene (OF45) Results in Increased Bone Formation and Bone Mass*

We have previously described osteoblast/osteocyte factor 45 (OF45), a novel bone-specific extracellular matrix protein, and demonstrated that its expression is tightly linked to mineralization and bone formation. In this report, we have cloned and characterized the mouse OF45 cDNA and genomic region. Mouse OF45 (also called MEPE) was similar to its rat orthologue in that its expression was increased during mineralization in osteoblast cultures and the protein was highly expressed within the osteocytes that are imbedded within bone. To further determine the role of OF45 in bone metabolism, we generated a targeted mouse line deficient in this protein. Ablation ofOF45 resulted in increased bone mass. In fact, disruption of only a single allele of OF45 caused significantly increased bone mass. In addition, knockout mice were resistant to aging-associated trabecular bone loss. Cancellous bone histomorphometry revealed that the increased bone mass was the result of increased osteoblast number and osteoblast activity with unaltered osteoclast number and osteoclast surface in knockout animals. Consistent with the bone histomorphometric results, we also determined thatOF45 knockout osteoblasts produced significantly more mineralized nodules in ex vivo cell cultures than did wild type osteoblasts. Osteoclastogenesis and bone resorption in ex vivo cultures was unaffected by OF45 mutation. We conclude that OF45 plays an inhibitory role in bone formation in mouse.

We have previously described osteoblast/osteocyte factor 45 (OF45), a novel bone-specific extracellular matrix protein, and demonstrated that its expression is tightly linked to mineralization and bone formation. In this report, we have cloned and characterized the mouse OF45 cDNA and genomic region. Mouse OF45 (also called MEPE) was similar to its rat orthologue in that its expression was increased during mineralization in osteoblast cultures and the protein was highly expressed within the osteocytes that are imbedded within bone. To further determine the role of OF45 in bone metabolism, we generated a targeted mouse line deficient in this protein. Ablation of OF45 resulted in increased bone mass. In fact, disruption of only a single allele of OF45 caused significantly increased bone mass. In addition, knockout mice were resistant to aging-associated trabecular bone loss. Cancellous bone histomorphometry revealed that the increased bone mass was the result of increased osteoblast number and osteoblast activity with unaltered osteoclast number and osteoclast surface in knockout animals. Consistent with the bone histomorphometric results, we also determined that OF45 knockout osteoblasts produced significantly more mineralized nodules in ex vivo cell cultures than did wild type osteoblasts. Osteoclastogenesis and bone resorption in ex vivo cultures was unaffected by OF45 mutation. We conclude that OF45 plays an inhibitory role in bone formation in mouse.
The constant modulation of the balance between skeletal strength and mineral availability in bone is effected by competing cell types in response to physiological needs. Osteoblasts produce, organize, and mineralize bone matrix in forming bone. Osteoclasts break down matrix by forming a lytic pocket in which bone is degraded and calcium is released. The generation and activity of these cell types is tightly regulated to provide equilibrium between formation and resorption and, thereby, an appropriate balance of strength and mineral release. Under certain conditions, such as aging, postmenopausal estrogen deficiency, or some pathophysiological states, there can exist an imbalance between bone resorption and bone formation. As a result, skeletal mass and strength are compromised and osteoporotic fractures can occur in the afflicted individuals.
Bone is produced by the organization and mineralization of the extracellular matrix produced by osteoblasts. The major component of the extracellular matrix of these cells is Type I collagen, which functions as a scaffold for new bone. In addition, non-collagenous matrix proteins have been identified that influence the operations of bone turnover, formation, and repair. These proteins are generally acidic and highly posttranslationally modified by phosphorylation, glycosylation, or sulfation (1).
Targeted deletion of extracellular matrix genes in mice has been a useful method to determine the in vivo functions of several matrix proteins. For example, osteocalcin is an abundant gamma carboxyl glutamic acid-containing bone matrix protein shown to be highly expressed in osteoblasts and is a biochemical marker of the bone remodeling process. Osteocalcin deletion in mice results in increased cortical bone thickness due to increased osteoblast activity, indicating that osteocalcin has negative effects in vivo on osteoblasts and bone formation (2). Another gene of this category, matrix glutamic acid protein (known as "MGP"), 1 was deleted in mice, resulting in extensive cartilage calcification and in the inappropriate calcification of arteries, leading to blood vessel rupture, and lethality in homozygotes (3). Ablation of the extracellular matrix protein osteonectin causes reduced bone formation with a decrease in both osteoblasts and osteoclasts resulting in a net loss of trabecular bone compared with wild type controls (4). Targeted disruption of osteopontin, an RGD-containing protein, results in a defect in the in vitro differentiation of osteoclasts; increased mineral content and maturity in long bones; and resistance to ovariectomy-induced, parathyroid hormone-induced, and mechanical unloading-induced bone resorption in mice (5)(6)(7)(8). The functions of matrix proteins can be further elucidated through the characterization of mice deficient in combinations of genes. For example, mice deficient in both osteopontin and MGP exhibit more vascular calcification than the MGP knockout mice, demonstrating that osteopontin can be an inhibitor of ectopic calcification in vivo (9).
In an effort to further characterize the process of bone me-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  tabolism and identify the proteins involved, we screened RNA of rat bone marrow cell cultures for novel transcripts specific to bone mineralization (10). This screen produced a novel clone that we designated osteoblast/osteocyte factor 45 (OF45). Rat OF45 encodes an RGD protein having 45% homology to the recently cloned human MEPE as well as loose homology to AG-1/DMP1. In rat, OF45 is highly expressed in the tibial shaft and metaphysis as well as in osteoblasts of induced bone marrow cultures, calvaria, and the UMR106 osteoblastic cell line. Immunohistochemistry in rat tibia revealed abundant OF45 protein in the osteocytes. Newly embedded rat osteocytes were also positive for OF45 in a marrow ablation model. Interestingly, the temporal expression pattern of OF45 differs from other well-characterized bone markers such as osteocalcin, osteonectin, and osteopontin in both in vitro and in vivo bone growth models (10). In this report we have cloned and characterized the mouse OF45 cDNA. With the exception of some additional sequence we report at the 5-prime end, the sequence reported here was identical to that recently published as MEPE (11). To demonstrate the functional significance of an elimination or reduction of OF45 expression, we have cloned the mouse OF45 genomic DNA, generated a mouse line with a targeted disruption of the OF45 gene, and subjected these mice to phenotypic analysis.

EXPERIMENTAL PROCEDURES
Cloning of Mouse OF45-A 1255-bp probe encoding most of the rat OF45 cDNA sequence, excluding the 3Ј-untranslated repeat, was used to screen a mouse 129 genomic lambda phage library (12). A 15-kb clone found to include exon 3 was used to generate PCR primers to clone the complete cDNA, the BAC clone, as well as the targeting construct. The 5Ј-end of the mRNA was determined by RNA ligase-mediated RACE as described (13). Briefly, 50 g of total RNA from mouse tibia was dephosphorylated with calf intestinal alkaline phosphatase (Roche Molecular Biochemicals, Indianapolis, IN) and the 5Ј-cap was removed using tobacco acid pyrophosphatase (Epicentre Technologies, Madison, WI). The RNA linker generated from the plasmid pGbx-1 (Gift from Dr. M. A. Frohman) was ligated to the decapped RNA using T4 RNA ligase (Epicentre Technologies, Madison, WI). The RNA was converted to cDNA by using Superscript II reverse transcriptase and p(dT) [12][13][14][15][16][17][18] primer (Invitrogen, Gaithersburg, MD). PCR was used to amplify the 5-end of the cDNA using primer pair NRC-1-288A and 36387.233B and primer pair NRC-1-288B and 36387.233A in the first and second round, respectively. A single PCR product of ϳ550 bp was obtained in the second round PCR. A control sample in which the pyrophosphatase step was omitted did not yield a PCR product, indicating that the ϳ550-bp product likely resulted from full-length mRNA that had been 5Ј-capped. The second round PCR fragment was isolated and ligated using a TA cloning kit (Invitrogen, Carlsbad, CA). Insert-containing plasmids were sequenced.
The 3Ј-end of the mRNA was determined using a 3Ј RACE kit (Invitrogen, Gaithersburg, MD) as suggested by the manufacturer. Briefly, 5 g of mouse tibia total RNA was reverse-transcribed using the AP primer provided. The PCR reaction amplified the sequence between OF45 nucleotide 527 (primer 36387.233E) and the AUAP primer provided. Two fragments close in size were cloned with the TA cloning kit (Invitrogen) and sequenced. The sequences of the two fragments were identical except for a short stretch of additional sequence at the 3Ј-end. Louis, MO) for screening of a mouse ES-129/SVIII BAC library. A single clone was identified and characterized. Chromosomal localization was determined by fluorescence in situ hybridization using the BAC clone by Genome Systems, Inc. according to their protocols. Intron/exon boundaries were determined by alignment of BAC clone sequence with the OF45 cDNA sequence.
Antibody Generation and Western Blot Analysis-High titer polyclonal antiserum was generated by immunization of rabbits with bacterial expressed full-length OF45 (Zymed Laboratories Inc., South San Francisco, CA). OF45 antibody was affinity-purified by chromatography on OF45-coupled agarose beads by standard methods (14). Protein samples were separated by electrophoresis on Novex (San Diego, CA) 10% NuPAGE gels and transferred to nitrocellulose using a semi-dry transfer. Blots were blocked using 1% Western Blocking Reagent from Roche Molecular Biochemicals (Mannheim, Germany) for 1 h and 1 o antibody diluted in 0.5% block for 1 h. After washing, goat anti-rabbit peroxidase 2 o antibody (Roche Molecular Biochemicals) diluted into 0.5% block was applied to the blot for 1 h. Signal was detected using an ECL detection kit (Amersham Biosciences, Buckinghamshire, England).
Gene Targeting, ES Cell Culture, and Microinjection-Genomic DNA fragments from the 15-kb genomic clone were subcloned into the JNS2 targeting plasmid (15) E14Tg2a embryonic stem (ES) cell line derived from the 129sj mouse strain was used for gene targeting of the OF45 locus (16). Electroporation, selection, expansion, and microinjection of ES cells into C57BL/6 embryos were as described previously (17). Out of 70 neomycin and gancyclovir-resistant clones, four were positive for the desired recombination event, resulting in a targeting frequency of 1 in 17. Chimeric animals were mated with C57BL/6 males and females. An 850-bp SpeI/MscI probe, homologous to a region directly 3Ј of the targeted mutation, was used to genotype agouti offspring by Southern blot analysis. Digestion of tail DNA with BamHI/BglII double digest resulted in a 10-kb hybridizing band in the wild type allele and a 9-kb hybridizing band in the targeted allele.
Animals-F2 or F3 129/Bl6 mix heterozygote mice were bred to generate successive populations of littermates used for in vivo experimentation at either 4 months or 1 year of age. The same parent animals were used to generate both 4-month and 1-year age groups. 10-monthold males backcrossed onto a C57/Bl6 background for nine generations were used for dynamic histomorphometry studies. Animals were maintained on a 12-h light/12-h dark cycle and were provided food and water ad libitum. 12 and 2 days prior to sacrifice, mice were given subcutaneous injection of the fluorochrome calcein at 10 mg/kg (Sigma Chemical Co., St. Louis, MO). Animals were euthanized by cervical dislocation, and the femurs were placed in 70% ethanol for later pQCT and histomorphometric analysis. The experiment was conducted according to Pfizer Animal Care and Use-approved protocols, and the animals were maintained in accordance with the ILAR (Institute of Laboratory Animal Research) Guide for the Care and Use of Laboratory Animals.
Serum Mineral Content-Calcium and inorganic phosphate levels were measured from blood serum at DNX Transgenic Sciences (Cranbury, NJ) using the ACE Clinical Chemistry System (Alfa Wassermann Inc., West Caldwell, NJ) by the manufacturer's protocols.
One-year-old animals were used for this study. Sera from 12 male animals of each genotype were used. Seven wild type and seven knockout females were also studied.
High Resolution X-ray-Femurs were examined at 2ϫ and 3ϫ magnification on a Faxitron model MX-20 Specimen Radiography System (Buffalo Grove, IL) with Kodak Min-R 2000 Mammography film in Min-R 2000 cassettes with an intensifying screen. Magnification was calculated by SID/SOD ϭ IS/OS where SID is source to image distance, SOD is source to object distance, IS is image size, and OS is object size. The femurs of 12 to 27 animals of each sex, age, and genotype were examined.
Peripheral Quantitative Computerized Tomography Analysis-Excised femurs were scanned by a pQCT x-ray machine (Stratec XCT Research M, Norland Medical Systems, Fort Atkinson, WI) with software version 5.40. A 1-mm-thick cross-section of the femur metaphysis was taken at 2.5 mm proximal from the distal end with a voxel size of 0.07 mm. Cortical bone was defined and analyzed using contour mode 2 and cortical mode 4. An outer threshold setting of 340 mg/cm 3 was used to distinguish the cortical shell from soft tissue and an inner threshold of 529 mg/cm 3 to distinguish cortical bone along the endocortical surface. Trabecular bone was determined using peel mode 4 with a threshold setting of 655 mg/cm 3 to distinguish (sub)cortical from cancellous bone. An additional concentric peel of 1% of the defined cancellous bone was used to ensure (sub)cortical bone was eliminated from the analysis. Volumetric content, density, and area were determined for both trabecular and cortical bone. Using the above setting, we have determined that the ex vivo precision of volumetric content, density, and area of total bone, trabecular, and cortical regions ranged from 0.99% to 3.49% with repositioning (18 Bone Histomorphometry-Following pQCT analysis, the distal half of the femur from each animal of 4-month-old and 1-year-old animals was dehydrated and embedded in methyl methacrylate. 4-m longitudinal sections were prepared with a Reichert-Jung Polycut S microtome (Leica, Deerfield, IL) and stained with modified Masson's Trichrome stain. Trabecular bone volume, trabecular number, and trabecular thickness were determined on the distal femoral metaphysis for these 4-month-old and 1-year-old animals (19,20). To understand the changes in osteoblasts and osteoclasts on the cancellous bone, a group of wild type (n ϭ 8) and knockout (n ϭ 10) male mice were necropsied at 10 months of age. The left distal half of the femur from these 10-month-old animals was decalcified, and 4-m longitudinal sections were prepared and stained with Toluidine Blue (Sigma, St. Louis, MO). These decalcified sections were used to determine the trabecular bone volume (BV/TV), percent osteoblast surface (Ob.S/BS), number of osteoblast per millimeter of bone surface (N.Ob/BS), percent osteoclast surface (Oc.S/BS), and number of osteoclast per millimeter of bone surface (N.Oc/BS) on the same area of the distal femoral metaphysis as described above (19,20). The right distal half of the femur from 10-monthold animals was dehydrated and embedded in methyl methacrylate. 10-m longitudinal sections were prepared and left unstained for the determination of mineral apposition rate, bone formation rate/bone surface referent (BFR/BS), and bone formation rate/tissue volume referent (BFR/TV) on the distal femoral metaphysis (19,20). An Image Analysis System (Osteomeasure, Inc., Atlanta, GA) was used for all histomorphometric analysis. Histomorphometric measurements were performed on cancellous bone tissue of the distal femoral metaphyses between 0.5 and 2 mm proximal to the growth plate-epiphyseal junction and extended to the endocortical surface in the lateral dimension (21).
Bone Marrow Harvest and Culture-Long bones were isolated from 3-to 6-month-old animals. Femur and tibia bone marrow cells harvested by centrifugation were plated at a density of 15 ϫ 10 6 cells per 100-mm plate in ␣-minimal essential medium (Invitrogen), with 10% fetal bovine serum and 50 g/ml Gentamicin (Invitrogen). 50 g/ml L-ascorbic acid, and 10 mM ␤-glycerophosphate were added at day 10 of culture. Cultures were fed three times per week, maintained for 4 weeks, and then stained by the Von Kossa method (22).
Calvaria Harvest and Culture-Calvaria were dissected from postnatal day 3 mice. 0.2 mg/ml collagenase P/0.25% Trypsin digestions were performed as described previously (23). The cells liberated in the second digest were plated at 2 ϫ 10 4 cells/cm 2 . At confluence, cultures were supplemented with 50 g/ml L-ascorbic acid and 10 mM ␤-glycerophosphate. Cells were later harvested for RNA or Alizarin Red staining (24). Stain solubilization and quantitation were as described (25).
Osteoclast Cultures-Bone marrow cultures were established as described above, plated at 1 ϫ 10 6 cells per well in a 24-well plate. Cells were stimulated with 10 nM 1,25-dihydroxyvitamin D 3 and TRAP stained as described (26). Bone resorption was assessed by culturing bone marrow cells on bone slices for 21 days and counting the resorption pits as has been described by Grasser et al. (27).
Statistical Methods-Statistical significance of in vivo and in vitro parameters was determined by two-tailed Student's t test. A p value of less than 0.05 was considered statistically significant.

OF45 cDNA Cloning and Characterization-Mouse OF45
was cloned using low stringency hybridization of a rat OF45 sequence to a murine 129 strain lambda phage genomic library. The full-length cDNA was then cloned by RT-PCR and 5Ј-and 3Ј-rapid amplification of cDNA ends (RACE) from mouse tibia RNA. The complete cDNA was 1741 bp in length. An additional 1679-bp cDNA was cloned, which was identical to the longer, more abundant message with the exception of a shorter 3Јextension. Three early methionine codons allow alternative start sites for the OF45 protein, the first of which would result in a 441-amino acid protein. A Kyte-Doolittle hydrophilicity plot revealed that the peptide sequence contained a hydrophobic leader sequence followed by a hydrophilic protein (Fig. 1A). Analysis of the amino acid sequence with PSORTII software, which predicts the subcellular localization sites of proteins from their amino acid sequences, indicated that this cDNA likely encodes an extracellular protein (28 -30). The predicted site of cleavage would be between Ala-24 and Ala-25, yielding a 417-amino acid secreted peptide with a calculated molecular mass of 44247 Da. Experimental confirmation of this analysis was obtained by Western blot analysis of media from transiently transfected CHO cells. An OF45-specific antibody detected a secreted protein of ϳ44 kDa (Fig. 1B). An OF45 peptide product of lower molecular weight was also detected in variable amounts, suggesting that proteolytic processing or degradation occurred in the media. The predicted amino acid composition of the basic OF45 peptide (predicted pI 9.17) was rich in serine, glycine, and charged residues. Several consensus protein kinase C, casein kinase II, tyrosine kinase, and cAMPdependant kinase phosphorylation sites, one N-glycosylation site, and a SDGD glycosaminoglycan binding site offer the potential for post-translational modifications to increase the acidic character of the protein.
Comparison of OF45 Sequence to Known Genes-Based on sequence and expression analysis, we believe that the mouse OF45 cDNA reported here represents the mouse orthologue of the rat OF45 gene (10) (GenBank TM accession number 260922). Alignment with the rat OF45 amino acid sequence demonstrated 67% identity. Importantly, key structural features were conserved. Both proteins were rich in serine, glycine, and charged amino acids and included an amino-terminal hydrophobic signal sequence that targeted the peptides for secretion. The RGD sequence at amino acids 183-185 is an element traditionally involved in cell-matrix interactions through integrin binding and signaling and was also conserved between species. Comparison of the OF45 cDNA sequence to Gen-Bank TM using the BLAST 1.4 algorithm (31) showed 45% amino acid identity to the recently identified human matrix extracellular phosphoglycoprotein (MEPE) gene (32). The macaque MEPE gene showed similar identity to OF45 (Gen-Bank TM accession number AB05025). The homology between mouse OF45 and human MEPE was distributed throughout the sequences and the RGD motif was conserved. MEPE was cloned as a candidate gene for a tumor-secreted phosphaturic factor responsible for tumor-induced osteomalacia. Like MEPE, OF45 shares structural features common to a class of extracellular matrix phosphoglycoproteins that includes osteopontin, dentin sialophosphoprotein, dentin matrix protein 1, and bone sialoprotein II. These genes are highly expressed in bone or dentin and play significant roles in mineralization (33)(34)(35). The mouse OF45 sequence was also 100% identical to the mouse MEPE cDNA recently published by Argiro et al. (11), although the cloned OF45 cDNA we report contained additional sequence at both the 5Ј-and 3Ј-ends. Comparison of mouse OF45, rat OF45, macaque MEPE, and human MEPE show several tightly conserved regions across all four species. These include the RGD motif, the signal sequence cleavage site, the putative N-glycosylation site, the SGDG glycosaminoglycan binding site, and several putative phosphorylation sites. Based on the sequence comparison and the restricted tissue expression, we concur with Argiro et al. that the sequence we report as mouse OF45 is the orthologue of human MEPE.
OF45 Gene Expression-The expression of mouse OF45 was analogous to that reported for rat OF45 (10). Mouse OF45 mRNA was induced during mineralization in primary mouse calvarial cultures, appearing at 2 weeks and maintained throughout 3 weeks of culture (Fig. 1C). Expression correlated DNAs digested with BamHI and BglII and hybridized with probe 2. A 10-kb band is detected at a wild type allele, and a 9-kb band is detected at a targeted allele. C, Northern blot analysis of 20 g of total RNA prepared from tibia of wild type, heterozygous, and knockout animals. Note the absence of OF45 expression in the RNA from knockout mice and the reduction of expression in the RNA from the heterozygous mice. FIG. 1. OF45 encoded a secreted protein and was expressed primarily in bone. A, a Kyte-Doolittle hydrophilicity plot of the predicted amino acid sequence of OF45 indicated a hydrophobic leader sequence followed by a hydrophilic peptide. The complete cDNA nucleotide sequence was deposited as GenBank TM accession number AF298661. B, Western blot analysis of OF45 expression in the media from CHO cells transfected with pcDNA3-OF45 expression vector. Media from CHO cells transfected with the pcDNA3 vector control are also shown. The major band corresponded to ϳ44 kDa as predicted from the cDNA sequence. A lower molecular weight species, likely the product of proteolysis, were also detected in media from OF45-transfected cells. C, Northern blot analysis of 20 g of total RNA from mouse calvarial bone cultures stimulated to mineralize with ascorbate and ␤-glycerophosphate. RNA samples were taken over a time course of 1 day to 3 weeks as indicated and analyzed for with increasing differentiation of the osteoblast lineage. Mouse OF45 mRNA expression was highly bone-specific. A high level of OF45 message was detected in total RNA prepared from mouse tibia, but all other tested tissues were negative, even by Northern blot analysis of poly(A) ϩ mRNA (Fig. 1D). Using highly sensitive RT-PCR, low levels of OF45 mRNA were detected in brown fat, white fat, testes, brain, and aorta (Fig. 1E).
Localization of OF45 protein was determined by immunohistochemical analysis of longitudinal tibia sections (Fig. 2). Mouse OF45 protein was expressed in all osteocytes, which are cells of osteoblast lineage embedded within bone matrix. In contrast to the OF45 mRNA expression observed in cell culture, osteoblasts and lining cells observed in the bone sections were negative for OF45 immunostaining. There was no specific staining detected in osteoclasts, chondrocytes, or periosteal cells. Also, OF45 was not detected in hematopoietic cells of the marrow compartment.
OF45 Genomic Organization and Targeting-The genomic DNA encoding mouse OF45 was subcloned from a hybridizing mouse 129 BAC clone and mapped. The mouse OF45 gene was comprised of three exons, two small 76-and 78-bp exons separated from the third 1597-bp exon by 10.5 kb (Fig. 3A). Fluorescence in situ hybridization analysis was used to determine that the OF45 gene is located on mouse chromosome 5, region 5E3-E5.
A targeting vector was constructed to delete the third and largest exon of OF45 and replace it with the neomycin-selectable marker gene (Fig. 3A). This deletion removed the DNAencoding Met-27 through the end of the translated sequence, including the final Asp-441. This targeting construct was electroporated into E14Tg2a embryonic stem cells and then screened for positive recombinants. Blastocyst microinjection of a targeted cell line resulted in chimeric animals that were subsequently bred to C57/Bl6 mice to yield mixed-background offspring heterozygous for the OF45 mutation. Heterozygote intercrosses produced litters with Mendelian ratios of wild type, heterozygous, and homozygous pups (Fig. 3B). Northern analysis of mRNA prepared from tibia detected no OF45 message in knockout animals and reduced expression in heterozygous animals (Fig. 3C). Immunohistochemistry of tibia sections confirmed the absence of OF45 protein expression in osteocytes of homozygous mutant mice (Fig. 2).
Phenotypic Analysis of OF45 Knockout Animals-To examine the phenotypic effects of OF45 mutation, heterozygote parents were bred to generate groups of at least 12 animals of each sex and genotype to be examined at 4 or 12 months of age. Mice heterozygous and homozygous for the OF45 mutations were normal, fertile, and healthy, with no obvious pathology. Serum chemical analysis revealed no statistically significant alterations in blood phosphate or calcium content (Table I). Wholebody x-rays revealed normal skeletal patterning and structure (not shown). However, high resolution x-rays of femurs revealed a pronounced increase in the amount of trabecular bone in both heterozygote and knockout animals at 1 year (Fig. 4). This difference was observable with high resolution x-rays at 4 months but was less dramatic (not shown). Quantitative anal-ysis of the distal femoral metaphysis by peripheral quantitative computerized tomography (pQCT) demonstrated significantly increased volumetric trabecular bone content and density in both males and females (Table II). There was no consistent change in the volumetric cortical bone content at this site. The periosteal and endosteal circumference was increased in the knockout animals with no significant change in cortical thickness (Table II). Caliper measurement revealed no difference in femur length between genotypes (Table III).
To further investigate the trabecular bone structure of these mice, we performed cancellous bone histomorphometric analysis at the distal femoral metaphysis (19). We observed a 50% increase in trabecular bone volume (BV/TV(%)) in 4-month-old female knockout mice compared with wild type mice (Fig. 5). In males, there was a 32% greater trabecular bone volume in knockout mice at 4 months. The effect of OF45 knockout mutation was more pronounced at 1 year with Ͼ2-fold and Ͼ3-fold more trabecular bone volume in male and female knockout mice, respectively. The increased trabecular bone volume was reflective of both increased trabecular number and increased trabecular thickness (Fig. 5).
The loss of even one OF45 allele and the resulting reduced gene expression (Fig. 3C) caused significant phenotypic effects in bone. As measured by both pQCT and histomorphometry, heterozygous animals exhibited the phenotypic effect of increased trabecular bone volume. Trabecular bone volume in 4-month males and 1-year females was equivalent in knockout and heterozygous animals with both being significantly in -FIG. 4. High resolution x-ray indicated that OF45 mutant mice had increased trabecular bone. Femurs from male and female mice at 1 year of age were subjected to high resolution x-ray imaging. In the magnification of the distal femur increased trabecular bone can be observed in both heterozygous and knockout mice when compared with the wild type mice. The increased trabecular bone was evident in both males and females at 1 year of age. creased over wild type mice. Heterozygote phenotype appears to be influenced by sex and age, appearing in only older females, but in both young and old males. An additional important observation was that both heterozygous and knockout mice exhibited less aging-associated bone loss than wild type animals. The male and female 1-year-old wild type groups possessed ϳ40 and 70%, respectively, less trabecular bone volume than did the matched 4-month-old groups (Fig. 5). In knockout animals, however, males exhibited no trabecular bone loss at 1 year. Female trabecular bone volume was reduced by only 15% in the 1-year-old mice. Thus, OF45 mutation appeared to have protected against age-related bone loss. This protective effect of OF45 mutation was also observed in heterozygote animals.
The net increase in trabecular bone volume observed in the heterozygous and knockout animals could theoretically arise from either 1) decreased osteoclastic bone resorption, 2) increased osteoblast-mediated bone formation, or 3) a combination of osteoblast and osteoclast effects. To investigate the effect of OF45 mutation on bone resorption and bone formation, we performed static and dynamic bone histomorphometric analysis of distal femoral metaphyseal cancellous bone using wild type and knockout male mice at 10 months of age. Percentage of bone surface occupied by osteoblasts (Ob.S/BS) and the osteoblast number per millimeter of bone surface (N.Ob/ BS) were significantly increased by ϳ2-fold in knockout animals compared with wild type controls (Table IV). A trend toward slightly but non-significantly decreased osteoclast surface (Oc.S/BS) and osteoclast number (N.Oc/BS) was observed (Table IV). Dynamic histomorphometry measurements utilizing calcein double-labeled bones to quantitatively measure the bone formation rate within a 10-day labeling interval demonstrated significantly increased mineral apposition rate (an index of osteoblast activity), bone formation rate/bone surface reference (BFR/BS), and bone formation rate/tissue volume referent (BFR/TV) in knockout animals (Table IV). These data demonstrate that the increased cancellous bone mass in OF45 knockout mice is a result of increased osteoblast-mediated bone formation with unchanged osteoclast-mediated bone resorption.
Ex Vivo Cell Culture-To further examine the cellular mechanisms behind the increased trabecular bone volume in heterozygote and knockout animals, we examined the in vitro mineralization and osteoclastogenesis potential of cultures derived from wild type and mutant animals. Primary cultures of bone marrow cells from all three genotypes were grown in the presence of ascorbate and ␤-glycerophosphate to induce osteoblast differentiation and promote bone nodule formation. After 3 weeks, mineralization was detected by Von Kossa stain and quantitated by image analysis (Fig. 6A). In several separate experiments, heterozygote-and knockout-derived cultures gave rise to significantly more bone nodules than those from wild type bone marrow, suggesting that there is more osteoblastic precursor cells present in the marrow (36). In addition, the mineralized area per nodule was increased in knockout cultures. A second culture system derived from mouse calvarial bone of 3-day-old mice produced comparable results over multiple experiments. In this more highly osteoblast-enriched system, calvarial cultures from OF45 knockout mice produced 60% more mineralized matrix as assayed by quantitation of solubilized Alizarin Red stain (Fig. 6B). Importantly, this experiment indicated that the osteoblastic cells derived from knockout mice at only 3 days after birth already exhibited increased mineralization potential.
To determine if a deficiency in the formation or activity of osteoclasts contributed to the increased bone phenotype, bone marrow cultures from wild type and knockout animals were established under osteoclast promoting conditions. In multiple cultures, the ability of marrow cultures to form multinucleated tartrate-resistant acid phosphatase (TRAP)-positive cells was unimpaired (Fig. 7, A and C). In fact, knockout bone marrow cultures exhibited a trend toward higher numbers of TRAPpositive multinucleated cells than wild type-derived cultures. The osteoclasts generated in the marrow culture system had the ability to resorb bone when cultured on bone slices (Fig.  7B). Resorption pits appeared to be identical between cells from wild type and knockout mice. No difference in the number of resorption pits was observed (Fig 7C).

DISCUSSION
We successfully generated a strain of mice with a deletion in the OF45 locus and ablated OF45 protein expression. This mutation did not affect the gross skeletal morphology or overall health of these animals. Refined examination of long bone revealed significantly increased cancellous bone mass in both male and female knockout animals, with differences in magnitude based on sex and age. These differences were visible by high resolution x-ray and quantifiable by both pQCT and histomorphometry. In addition, analysis of sex-matched populations at 4 months and 1 year revealed that deletion of OF45 prevented the loss of trabecular bone characteristic of aged mice. The bone loss normally observed in aged mammals is thought to be due to the reduced abundance and reduced survival of osteoblasts, with no change in the abundance or activity of osteoclasts (37)(38)(39). The increased bone mass in OF45ablated mice was due to increased abundance and activity of osteoblasts as indicated by the increased N.Ob/BS and BFR/BS levels observed relative to wild type mice. Ex vivo cell culture from bone marrow indicated increased number osteoblast precursors in OF45 knockout animals. Although not specifically addressed in this study, it is possible that OF45 deletion also affected osteoblast survival or lifetime.
Heterozygous mice also exhibited significantly increased trabecular bone. Therefore, even a reduction of OF45 levels markedly impacted bone structure suggesting that OF45 is a limiting control point in the regulation of bone. The haploinsufficiency nature of OF45 suggests that genetic variation in OF45 expression levels could result in variation of peak bone mass or aging-related bone loss in individuals. It will be of interest to determine if polymorphisms in the OF45 gene contribute to differences in bone mass within the human population.
The spectrum of tissue expression of mouse OF45 closely matched that of rat, with abundant production restricted to bone. In differentiating calvarial osteoblast cultures, OF45 expression increased with progressive differentiation and maintained a steady state of expression in mature cultures. Immunohistochemistry of tibial bone sections revealed abundant staining of the embedded osteocytes. We did not observe OF45 immunostaining of osteoblasts in either rats (10) or mice despite the abundant mRNA expression in osteoblast-rich ex vivo cultures. This discrepancy could be explained as a technical artifact in the immunostaining methods. For example, perhaps only the embedded cells accumulate sufficient amounts of this secreted protein to allow detection. However, as we previously reported, Northern blot analyses of mRNA isolated from the rat marrow ablation model support the conclusion that OF45 mRNA is expressed only by cells embedded within bone and not by the osteoblastic cells that are not imbedded (10). Taking  these data together, we believe that OF45 protein is expressed in vivo in mature osteoblasts during the process of embedding in new bone and maintained throughout osteocyte development.
In addition to bone expression, we report low mRNA levels in other organs, including testes, fat, and the aorta. The significance of these low levels of mRNA is unclear. Adipose tissue expression may be meaningful given the common mesenchymal stem cell origin of both adipocytes and osteoblasts. Expression of OF45 in the aorta may also have significance, especially under pathophysiological conditions of vascular calcification (40). A recent report has demonstrated OF45 mRNA expression in odontoblasts, the tooth cells analogous to osteoblasts (41). Furthermore, the human tissue expression of OF45 (called MEPE) was reported in brain as well as bone (32). We have not been able to detect OF45 mRNA in mouse brain, although we have been able to confirm relatively abundant OF45 expression in human brain mRNA samples (T.A.B., not shown). To determine if the differences in expression pattern between humans and mice translate to additional functions for OF45 in humans will require further investigation.
The OF45 sequence described here is the mouse orthologue of the human MEPE gene. The human MEPE gene was cloned FIG. 6. Increased mineralization in OF45 mutant ex vivo cell cultures. A, primary bone marrow cultures from wild type, heterozygote, and knockout animals were cultured under mineralizing conditions with ascorbate and ␤-glycerophosphate. Mineralization was detected by Von Kossa staining. The black-stained mineralized nodules were quantified using Optimas image analysis software. Graphs depict quantitation of nodule number, total nodule area, and average area per nodule. B, primary calvarial cultures derived from wild type and knockout animals at 3 days postnatal were cultured under mineralizing conditions with ascorbate and ␤-glycerophosphate. Alizarin Red stain was used to detect mineralization. The graph depicts concentration of alizarin red stain quantitated by stain solubilization and optical density reading. Error bars denote standard error. *, statistical significance at p Ͻ 0.05. as a candidate for the long sought after phosphaturic factor secreted by tumors causing oncogenic osteomalacia (32). A parallel field of research on a phenotypically similar condition, X-linked hypophosphatemic rickets, has suggested that the phosphaturic factor is processed or degraded by the PHEX endoprotease. However, subsequent investigation has implicated FGF23 as a phosphaturic factor in both tumor-induced osteomalacia and autosomal-dominant hypophosphatemic rickets (42,43). Furthermore, it has been demonstrated that FGF23 is a substrate for the PHEX endoprotease (42). It has been reported that MEPE, in contrast to FGF23, does not exhibit phosphaturic activity on renal cells (42) and is not a direct substrate of the PHEX endoprotease (44). Our observation that OF45 knockout animals exhibited no statistically significant differences in serum calcium or phosphate also argues against MEPE being a critical phosphaturic factor. Nonetheless, it is still significant that OF45/MEPE was identified due to its overexpression by osteomalacia-inducing tumors, and there remains a link between PHEX and OF45/MEPE. Although Bowe et al. (44) have claimed that OF45/MEPE is not a direct substrate of PHEX, Argiro et al. (11) have demonstrated that the osteoblasts of PHEX-deficient Hyp mice do express markedly higher levels of MEPE mRNA than do those from wild type mice. We have demonstrated that co-transfection of OF45 and PHEX resulted in decreased OF45 protein detected in the media than observed with OF45 alone. 2 Conversely, in a cell-free system, Guo et al. (45) demonstrated that PHEX in-hibits the cleavage of OF45/MEPE. It seems increasingly clear that FGF23 and PHEX are members of a phosphate regulatory pathway. MEPE may be a downstream target of this pathway and a member of a bone regulatory branch of this system.
In summary, we have determined that reduction or total ablation of the OF45 gene in mouse resulted in increased bone mass. Although the molecular mechanism of OF45 action is unknown, the etiology of this increased bone mass cannot be attributed to a failure in the differentiation or abundance of osteoclasts. Bone histomorphometry and primary cell culture experiments did not detect any impairment in the osteoclast cells that mediate bone resorption. The increased osteoblast number and activity (mineral apposition rate) observed by bone histomorphometry and the mineralization potential observed in both marrow and calvarial cultures indicate that the phenotypic effect in the knockout animals resulted from an increased number and activity of osteoblasts and osteoblastic precursors. One interpretation of these results is that OF45 functions in vivo as a negative regulator of osteoblast number and activity. The concept that regulation of bone mass can be controlled by negative regulators has recently been demonstrated with the characterization of osteocalcin-deficient (2) and leptin-deficient mice (46). Further experimentation will be needed to determine if OF45 acts directly on osteoblasts or through indirect mechanisms. The expression of OF45 within osteocytes, the terminally differentiated osteoblasts that become encased in the mineralized matrix during bone formation (47), opens the possibility that secreted OF45 may act locally on osteoblasts and osteoblast precursors. Osteocytes can communicate through 2 T. A. Brown, unpublished results. FIG. 7. Bone marrow cells from OF45 mutant mice were fully capable of osteoclast formation and osteoclast-mediated bone resorption. A, osteoclastogenesis in bone marrow cells derived from wild type and knockout mice. Tartrate-resistant acid phosphatase (TRAP) cytochemical stained cultures from wild type and knockout animals. TRAP-positive cells with three or more discrete nuclei were considered osteoclasts. B, bone-resorptive activity of OF45 wild type and knockout osteoclasts. Representative resorption pits formed on bone slices by osteoclasts from wild type and knockout animals are shown. C, osteoclastogenesis was assessed by counting TRAP-positive multinucleated cells. Bone resorption was assessed by counting the number of resorption pits per bone slice. Error bars denote standard error. No statistically significant differences were observed. extensive dendritic processes and have been theorized to be the primary sensors of mechanical strain within bone (48 -51). OF45 may function as an integral component of the intimate connection between the sensing and signaling of osteocytes and the osteoblast response.