Osteocyte-Specific Deletion of Fgfr1 Suppresses FGF23

Increases in fibroblastic growth factor 23 (FGF23 or Fgf23) production by osteocytes result in hypophosphatemia and rickets in the Hyp mouse homologue of X-linked hypophosphatemia (XLH). Fibroblastic growth factor (FGF) signaling has been implicated in the pathogenesis of Hyp. Here, we conditionally deleted FGF receptor 1 (FGFR1 or Fgfr1) in osteocytes of Hyp mice to investigate the role of autocrine/paracrine FGFR signaling in regulating FGF23 production by osteocytes. Crossing dentin matrix protein 1 (Dmp1)-Cre;Fgfr1 null/+ mice with female Hyp;Fgfr1 flox/flox mice created Hyp and Fgfr1 (Fgfr1Dmp1-cKO)-null mice (Hyp;Fgfr1 Dmp1-cKO) with a 70% decrease in bone Fgfr1 transcripts. Fgfr1Dmp1-cKO-null mice exhibited a 50% reduction in FGF23 expression in bone and 3-fold reduction in serum FGF23 concentrations, as well as reductions in sclerostin (Sost), phosphate regulating endopeptidase on X chromosome (PHEX or Phex), matrix extracellular phosphoglycoprotein (Mepe), and Dmp1 transcripts, but had no demonstrable alterations in phosphate or vitamin D homeostasis or skeletal morphology. Hyp mice had hypophosphatemia, reductions in 1,25(OH)2D levels, rickets/osteomalacia and elevated FGF2 expression in bone. Compared to Hyp mice, compound Hyp;Fgfr1 Dmp1-cKO-null mice had significant improvement in rickets and osteomalacia in association with a decrease in serum FGF23 (3607 to 1099 pg/ml), an increase in serum phosphate (6.0 mg/dl to 9.3 mg/dl) and 1,25(OH)2D (121±23 to 192±34 pg/ml) levels, but only a 30% reduction in bone FGF23 mRNA expression. FGF23 promoter activity in osteoblasts was stimulated by FGFR1 activation and inhibited by overexpression of a dominant negative FGFR1(TK−), PLCγ and MAPK inhibitors. FGF2 also stimulated the translation of an FGF23 cDNA transfected into osteoblasts via a FGFR1 and PI3K/Akt-dependent mechanism. Thus, activation of autocrine/paracrine FGF pathways is involved in the pathogenesis of Hyp through FGFR1-dependent regulation of FGF23 by both transcriptional and post-transcriptional mechanisms. This may serve to link local bone metabolism with systemic phosphate and vitamin D homeostasis.


Introduction
The FGF family consists of canonical FGFs, intracellular FGFs, and hormone-like FGF gene products (i.e., FGF19, FGF21 and FGF23). The earliest evolved family members are likely intracellular FGFs, exemplified by high molecular weight FGF2 (HMW-FGF2), which interact with intranuclear FGFR1 to directly activate gene transcription (i.e., integrative nuclear FGFR1 signaling) [1]. Canonical or FGFs evolved later to serve autocrine/paracrine functions. These secreted FGFs have high heparin affinity, limited diffusion capacity and act locally on cell surface FGF receptors [2]. More recently, circulating FGFs emerged with the capability to diffuse from tissues and specifically target FGF receptor/Klotho complexes at distal sites due to their unique C-terminal domain [3]. For example, FGF23 is a ,32 kDa hormone with an N-terminal FGF-homology domain and a novel 71 amino acid C-terminus [4] that allows it to be released into the circulation and to interact with a-Klotho, a type I membrane, bglycosidase to form a trimeric complex with FGFRs in specific tissues [4][5][6][7][8][9].
FGF23 is produced and secreted by osteoblasts/osteocytes in bone. FGF23 activates FGFR/a-Klotho complexes in the kidney to decrease Npt2a co-transporters leading to inhibition of renal tubular phosphate reabsorption and to reduce circulating 1,25(OH) 2 D levels by inhibiting enzymes regulating vitamin D metabolism (i.e., inhibiting Cyp27b1 and stimulating Cyp24) [6,7,[9][10][11][12][13]. The biological functions of FGF23 are essential for maintenance normal mineral metabolism. Ablation of FGF23 in mice is lethal in the early postnatal period due to hyperphosphatemia and excessive 1,25(OH) 2 D production [14,15]. On the other hand, excess FGF23 causes hypophosphatemia, aberrant vitamin D metabolism and rickets/osteomalacia. Increases in FGF23 underlie acquired and hereditary forms of hypophosphatemic rickets and are involved in the pathogenesis of mineral metabolism abnormalities in chronic kidney diseases [16]. Consequently, understanding the factors that regulate FGF23 is of high clinical importance.
Recent studies implicate a role of FGFR1 activation in regulating FGF23 gene transcription [29,[34][35][36]. In this regard, FGF23 is increased in osteoglophonic dysplasia, which is caused by activating mutations in FGFR1 [35,36]. In addition, ligands for FGFR1, including FGF1, FGF2, and FGF7 are significantly increased in the Hyp and Dmp1 knockout mice [37][38][39]. Pharmacological inhibition of FGFR1 also blocks FGF23 transcription in bone both in vitro and ex vivo [29,40,41]. Moreover, recent studies have reported that administration of monoclonal FGFR1 activating antibodies stimulates FGF23 production and induces hypophosphatemia [42]. These findings suggest that autocrine/paracrine FGF/FGFR1 signaling pathways may be involved in regulation of hormonal FGF23. The gain-offunction studies implicating FGFR1 in the regulation of FGF23, however, are confounded by the generalized effects to activate FGF receptors in multiple tissues. Interpretation of the inhibitor studies are confounded by actions to block FGF23 end-organ effects and other potential systemic actions that could lead to feedback regulation of FGF23. These limitations preclude establishing a direct cause and effect relationship between FGFR1 function and FGF23 gene transcription in osteoblasts/osteocytes.
In the current study, to test the central role of FGFR1 in mediating the increased expression of FGF23 in bone, we use a mouse genetic approach to conditionally deleted Fgfr1 from osteocytes of Hyp mice. We find an important role of FGFR1 signaling in osteocytes in mediating the increase of FGF23 caused by Phex mutations in Hyp mice, thereby linking alterations autocrine/paracrine functions of FGF/FGFR1 pathways in the bone microenvironment with the secretion of circulating FGF23 that activates FGFRs in distal tissues to coordinate bone mineralization with renal regulation of phosphate and vitamin D metabolism.

Animals breeding and genotyping
All animal research was conducted according to guidelines provided by the National Institutes of Health and the Institute of Laboratory Animal Resources, National Research Council. The University of Tennessee Health Science Center's Animal Care and Use Committee approved all animal studies (Protocol number: 12-162.0). CMV-Cre and Hyp mice were originally purchased from the Jackson Laboratory (Bar Harbor, ME, USA) and maintained in C57BL/6J background. The floxed Fgfr1 mice (Fgfr1 flox/+ ) were obtained from Dr. Chuxia Deng at National Institute of Diabetes and Digestive and Kidney Diseases and maintained in C57BL/6J background for at least six generations. 9.6-kb Dmp1-Cre mice were used to delete the floxed Fgfr1 in bone as described previously [43] and maintained in C57BL/6J background for at least five generations. The 9.6-kb Dmp1-Cre has been widely utilized to delete a gene specifically in osteocytes [43,44], where it has been shown to be active in 77% of osteocytes and only ,1% of surface osteoblasts in long bone [45,46]. All mice were maintained on a standard diet (7912, Harlan Teklad, Madison, WI, USA). First, we crossed Fgfr1 flox/+ to CMV-Cre to obtain a germlinespecific deletion of Fgfr1 (Fgfr1 null/+ ). The Fgfr1 null/+ mice were crossed to Dmp1-Cre mice to obtain double heterozygous Dmp1-Cre;Fgfr1 null/+ mice. Second, heterozygous female Hyp (XX Hyp ) mice were crossed to Fgfr1 flox/flox males to obtain Hyp;Fgfr1 flox/+ double heterozygous females. Then Hyp;Fgfr1 flox/+ females were crossed to Fgfr1 flox/flox males to generate Hyp;Fgfr1 flox/flox females. Finally, Hyp;Fgfr1 flox/flox females were crossed to Dmp1-Cre;Fgfr1 null/+ males to obtain the osteocyte-specific deletion of Fgfr1 in Hyp mice. These mice are on a mixed genetic background. For the entire study, samples were collected from 6-week-old Fgfr1 flox/+ (wild-type equivalent) control, conditional Dmp1-Cre;Fgfr1 flox/+ heterozygous, conditional Dmp1-Cre;Fgfr1 null/flox (Fgfr1 Dmp1-cKO )-null, Hyp (X Hyp Y), and compound Hyp;Fgfr1 Dmp1-cKO -null male littermates as well as Fgfr1 flox/null (equivalent to Fgfr1 +/null ) and Hyp;Fgfr1 flox/null (equivalent to Hyp;Fgfr1 +/null ) male mice. Tail clips were collected to genotype the mice. REDExtract-N-Amp Tissue PCR Kit (Sigma-Aldrich, St. Louis, MO, USA) was used for DNA extraction and PCR amplification. Mice were genotyped for Phex mutations and Dmp1-Cre using previously described primers [43,47], for the Fgfr1 flox allele using forward primer 59-CTG GTA TCC TGT GCC TAT C-39 and reverse primer 59-CCA ATC TGA TCC CAA GAC CAC-39 (325 bp product for the Fgfr1 + wild-type allele, 400 bp product for the Fgfr1 flox floxed allele), and for the Fgfr1 null allele using forward primer 59-GTA TTG CTG GCC CAC TGT TC-39 and reverse primer 59-CCA ATC TGA TCC CAA GAC CAC-39 (250 bp product for the Fgfr1 null null allele).

Serum Biochemistry
Serum samples were collected by intracardiac exsanguinations and urine samples were collected overnight in metabolic cages. Calcium was measured using a Calcium CPC Liquicolor Kit (Stanbio Laboratories, Boerne, TX, USA) and phosphorus was measured using the phosphomolybdylate-ascorbic acid method, as previously described [47]. Serum parathyroid hormone (PTH) levels were measured using the Mouse Intact PTH ELISA kit (Immutopics, Carlsbad, CA, USA). Serum full-length FGF23 levels were measured using the FGF23 ELISA kit (Kainos Laboratories, Tokyo, Japan), and serum C-terminal Fgf23 levels were measured using the FGF23 (C-Term) ELISA kit (Immutopics, Carlsbad, CA, USA).

Bone and kidney RNA isolation and real-time reverse transcriptase (RT)-PCR
For quantitative real-time RT-PCR, 1.0 mg total RNA isolated from tibias without bone marrow and kidney of 6-week-old four genotypes mice was reverse transcribed as previously described [48]. PCR reactions contained 20 gg template (cDNA or RNA), 375 gM each forward and reverse primers, and 16 SsoFast EvaGreen supermix (Bio-Rad, Hercules, CA, USA) in a total of 10 ml reaction volume. The threshold cycle (Ct) of tested gene product from the indicated genotype was normalized to the Ct for cyclophilin A. Expression of total Fgfr1 transcripts was performed using the following Fgfr1-allele-specific primers in exons 9 and 10: forward primer of normal Fgfr1 + transcripts (Fgfr1 + plus Fgfr1 flox ): 59 -ACC AAG AAG AGC GAC TTC CA -39 and reverse primer: 59 -AAC CAG GAG AAC CCC AGA GT -39. The normal Fgfr1 + vs. cyclophilin A was normalized to the mean ratio of five control mice, which was set to 1. The percentage of Fgfr1 null (Fgfr1 null ) and/or conditional deleted (Fgfr1 Dflox ) transcripts was calculated from the relative levels of the normal Fgfr1 + transcripts in different Fgfr1-deficient mice [49]. All primer information of other genes used in real-time RT-PCR can be found in our previous report [44].

High resolution 3D microtomography
The femurs were collected, fixed and dehydrated in 70% ethanol. High-resolution micro-Computed Tomography (mCT40, Scanco Medical, Basserdorf, Switzerland) was used to scan and evaluate the metaphyseal trabecular bone microarchitecture and the midshaft cortical bone parameters. The entire femurs were scanned in a 12.3 mm diameter sample holder at 6 mm resolution: energy level of 55 KeV and intensity of 145 mA. Evaluation of the bone growth was obtained by measuring the length of the scanned femur. The trabecular bone volume (BV/TV%) was measured within the secondary spongiosa on a set of 50 sections (0.6 mm) underneath the growth plate at a threshold of 200 as previously described [50]. The cortical bone thickness (CtTh, mm) was analyzed from 100 sections chosen at the midshaft of each femur at a threshold of 350.

Bone histology and histomorphometry
Evaluation of the bone growth was obtained by measuring the length of the femur of 6-week-old mice with a slide caliper. Femurs were fixed and dehydrated in 70% ethanol, and embedded in methylmetacrylate at low temperature. Nonserial longitudinal frontal slices (5 mm) were cut from the embedded bones with a microtome (Polycut-S; Reichert-Jung, Wetzlar, Germany) and were either left unstained or used for modified Goldner staining. Alizarin complexone dehydrate (Sigma, St. Louise, MO) double labeling of bone and histomorphometric analyses of periosteal mineral apposition rate (MAR) in femurs were performed using the osteomeasure analysis system (OsteoMetrics, Decatur, GA, USA).

Western blot analysis
Bone marrow-free femurs from four groups of mice were homogenized into a fine powder in liquid nitrogen using a porcelain mortar and pestle. The powder was transferred into T-PER Tissue Protein Extraction Reagent with 16 Halt protease inhibitor (Thermo Scientific, Rockford, IL) and 1 mM phenylmethylsulfonyl fluoride (PMSF). After three 30-second sonications, samples were centrifuged at 13,0006 g for 10 minutes and supernatants were stored at 280uC until use.
To examine if Fgfr1 signaling has a role in translational control of FGF23 protein expression in vitro, a number of 1.2610 6 of MC3T3-E1 cells were transfected with pcDNA3.1-FGFR1 cDNA expression plasmids (3.0 mg) and human pcDNA3.1-FGF23-V5-His cDNA expression plasmids (3.0 mg) conducted by electroporation using Cell Line Nucleofector Kit R according to the manufacturer's protocol (Amaxa Inc, Gaithersburg, MD). 2.0610 5 cells were seeded in 6-well plates in a-MEM media (Life technologies, Grand Island, NY) with 10% FBS at 37uC in 5% CO 2 and humidified incubator for 8 hours. For time-course experiments, the cells were changed into fresh growth medium containing 5% FBS and FGF2 (50 ng/ml, Sigma-Aldrich, St. Louis, MO) and heparin (10 mg/ml, Sigma-Aldrich, St. Louis, MO) were added to the cell culture media for 8, 24. And 48 hours before cells were harvested. For signaling mechanism study, the cells were changed into fresh growth medium containing 5% FBS and FGF2 (50 ng/ml) and heparin (10 mg/ml) in the presence and absence of Wortmannin (PI3K inhibitor, 1.0 mM) or Cycloheximide (0.5 mg/ml, Sigma-Aldrich, St. Louis, MO) were added to the cell culture media for 24 hours before cells were harvested. The cells were lysed with 150 ml of T-PER with 16 Halt protease inhibitor and 1 mM per well, After three 30-second sonications, total cell lysates were centrifuged at 13,0006g for 10 minutes and supernatants were stored at 280uC until use. Protein concentrations of the supernatant were determined with a Bio-Rad protein assay kit (Bio-Rad, Hercules, CA). Equal quantities of protein were subjected to NuPAGE 4-12% Bis-Tris Gel (Invitrogen, Carlsbad, CA) and were analyzed with standard Western blot protocols (HRP-conjugated secondary antibodies from Santa Cruz Biotechnology and ECL from GE Healthcare Bio-Sciences (Pittsburgh, PA). Antibody against Fgf23 (MAB2629) was obtained from R&D Systems, Inc. (Minneapolis, MN). Antibody against Fgf2 (610073) was purchased from BD Biosciences (San Jose, CA). Anti-V5-HRP antibody (R961-25) was purchased from Life technologies (Carlsbad, CA). Anti-b-actin (sc-47778) antibodies were from Santa Cruz Biotechnology (Paso Robles, CA). The intensity of bands was quantified using Image J software (http://rsb.info.nih. gov/ij/).

Statistics
We evaluated differences between two groups by unpaired t-test and multiple groups by one-way analysis of variance. All values are expressed as means 6 SD. All computations were performed using GraphPad Prism5 (GraphPad Software Inc. La Jolla, CA, USA).

Osteocyte specific deletion of Fgfr1 in mice
We selectively deleted Fgfr1 in osteocytes by crossing Dmp1-Cre;Fgfr1 +/null mice with homozygous Hyp;Fgfr1 flox/flox mice. Conditional Fgfr1 Dmp1-cKO (or Dmp1-Cre;Fgfr1 flox/null )-null mice with intact Phex but reduced Fgfr1 in osteocytes; Hyp (X Hyp Y) mice with intact Fgfr1 but mutant Phex; compound Hyp (X Hyp Y);Dmp1-Cre;Fgfr1 flox/null mice with reduced Fgfr1 in osteocytes and mutant Phex, hereafter referred to as Hyp;Fgfr1 Dmp1-cKO -null mice; and Fgfr1 flox/+ control mice with both intact Fgfr1 and Phex, were all born with the expected Mendelian frequency. There was no difference in survival between genotypes produced by this breeding strategy. All mice survived to adulthood.
To confirm Dmp1-Cre-mediated conditional deletion of Fgfr1 in bone, we performed PCR analysis of genomic DNA using a combination of primers that specifically detect floxed Fgfr1 alleles (Fgfr1 flox ) and the excised floxed Fgfr1 alleles (Fgfr1 Dflox ), as well as wild type alleles (Fgfr1 + ) in Dmp1-Cre;Fgfr1 flox/+ heterozygous mice ( Figure 1A). We found that Dmp1-Cre-mediated excision of the Fgfr1 Dflox alleles occurred in bone tissues, including calvaria and femur. Low levels of recombinase activity were detected in brain, muscle, and intestine, indicating that Dmp1-Cre promoter is not restricted to bone ( Figure 1B). We found no evidence for the Fgfr1 Dflox alleles in a wide range of other tissues tested, including the kidney. The floxed Fgfr1 flox alleles were detected in all tissues tested ( Figure 1B). Real-time RT-PCR analysis revealed that Fgfr1 mRNA levels in bone tissues were reduced by about 30% from conditional Dmp1-Cre;Fgfr1 flox/+ mice and about 70% from conditional Fgfr1 Dmp1-cKO -null mice compared with control mice ( Figure 1C), respectively. Consistent with the absence of Dmp1-Cre expression in kidney, there were no reduction in Fgfr1 transcripts in the kidney of Dmp1-Cre;Fgfr1 flox/+ , but Fgfr1 transcripts were reduced by 50% in Fgfr1 Dmp1-cKO -null mice, consistent with the presence of a null Fgfr1 allele ( Figure 1D).
Effects of osteocyte specific deletion of Fgfr1 on mouse bone phenotype Dmp1-Cre;Fgfr1 flox/+ mice were indistinguishable from Fgfr1 flox/+ littermates and wild-type mice. We focused our studies on Fgfr1 Dmp1-cKO -null and compound Hyp;Fgfr1 Dmp1-cKO -null mice. The overall appearance, body weight and femur length of Fgfr1 Dmp1-cKO -null mice were indistinguishable from Fgfr1 flox/+ littermates, which we used as controls (Figure 2A and 3A). In spite of the known role of Fgfr1 in bone development [54], conditional deletion of Fgfr1 in osteocytes, which represent the terminal stage of osteoblasts differentiation, had minimal effects on the skeletal phenotype. Micro-CT 3D image analyses of the femur revealed that deletion of Fgfr1 in osteocytes resulted in no measurable changes in cortical or trabecular bone structural parameters. Indeed, Fgfr1 Dmp1-cKO -null mice had normal femur length, trabecular bone volume, and cortical thickness ( Figure 3). The histological appearance of bone was indistinguishable in Fgfr1 Dmp1-cKO -null mice (i.e., normal growth plate, trabecular bone and osteoid volume and mineral apposition rate) compared with age-matched control mice ( Figure 4).
Assessment of mRNA expression in bone, however, revealed that the conditional deletion of Fgfr1 in osteocytes did result in significant but selective reductions in osteocyte-specific gene expression, including Fgf23, Sost, Phex, Dmp1, and Mepe, but not Osteopontin, Osteocalcin, Fgf1, Fgf2, Fgfr2, Fgfr3, Fgfr4, and Mmp13 (Table 1). Consistent with the 50% decrease of Fgf23 transcripts, western blot analysis showed a significant reduction of FGF23 protein in bone from Fgfr1 Dmp1-cKO -null mice ( Figure 2B and 2C). Conditional deletion of Fgfr1 in osteocytes also resulted in a 2-fold reduction in serum FGF23 protein levels in the Fgfr1 Dmp1-cKO -null mice ( Table 2).
We also observed significant reductions in Sost and dickkopf WNT signaling pathway inhibitor 1 (Dkk1) but not frizzled class receptor 2 (Fzd2), which are antagonist of the Wnt signaling; however we failed to see an increase in either Wnt10b or Axin-2 in the Fgfr1 Dmp1-cKO -null mice. Loss of Fgfr1 in osteocytes had no effect on osteoclast, chondrocyte, or adipocyte transcripts in bone (Table 1). There were no differences in the Fgf2 protein expression between control and Fgfr1 Dmp1-cKO -null mice ( Figure 2B and 2C).
In spite of this reduction in FGF23, there were no significant changes in serum PTH, 1,25(OH) 2 D, phosphorus, and calcium levels in Fgfr1 Dmp1-cKO -null mice compared with age-matched control mice ( Table 2). As expected due to the inclusion of an Fgfr1 null allele in the conditional knockout strategy, Fgfr1 transcripts were reduced by ,50% reduction in kidney of Fgfr1 Dmp1-cKO -null mice. However, there were no significant differences in the FGF23-regulated genes, including Npt2a, Npt2c, Cyp24a1 and Cyp27b1, or a-Klotho transcripts in the kidney of Fgfr1 Dmp1-cKO -null mice ( Table 3).

Phex-mutant Hyp mice exhibit elevated FGF23 expression, hypophosphatemia and rickets/osteomalacia
Hyp mice with inactivating Phex mutations have very high production of FGF23 in osteocytes and markedly elevated circulating FGF23 levels leading to hypophosphatemia and abnormalities in vitamin D metabolism, as well as impaired mineralization of cartilage and bone leading to rickets and osteomalacia [31,39,47]. These features were present in Hyp mice in this study. Hyp mice derived from the current breeding strategy were considerably smaller, had significantly lower body weight, and demonstrated reductions in tail and femur lengths due to the presence of rickets/osteomalacia (Figure 2A and 3A). Micro-CT 3D image analyses of the femur of Hyp mice revealed that loss of Phex function resulted in shortened bone length, widened growth plate and reduced cortical or trabecular bone structural parameters, consistent with the presence of rickets and osteomalacia. Consistent with the abnormal mineralization in Hyp mice, histological evaluation revealed a reduction in trabecular bone volume, cortical thickness, and mineral apposition rate as well as an apparent increase of un-mineralized osteoid in both trabecular and cortical bone areas (Figure 3 and 4).
Bone samples from Hyp mice showed a 23-fold increase of Fgf23 transcripts (Table 1) and western blot analysis showed a 7fold increment in FGF23 protein levels in bone ( Figure 2B and 2C). Hyp bone also showed a significant increase in other osteocyte-specific markers, including Dmp1 and Mepe. Consistent with the observations in Hyp osteocytic cells [55], Fgf1, Fgf2, Fgfr2, Fgfr3, and Fgfr4 as well as Osteopontin, Osteocalcin, Galnt3, Mmp13, OPG, and RankL mRNA levels were also increased in bone derived from Hyp mice (Table 1). In addition, Hyp mouse bone exhibited an increase in Fzd2, Wnt10b, and Axin-2 expression, as well as suppression of Sost and Dkk1, and Collagen II and Mmp9 were also increased in Hyp bone. We also observed more than 2-fold increase of Fgf2 protein expression in Hyp mice compared to control Fgfr1 flox/+ mice ( Figure 2B and 2C).
Similar to previous reports [39,47], Hyp mice displayed the anticipated elevations in serum Fgf23 concentrations (,18-fold) that were associated with significant reductions in serum phosphate. There was also evidence for the previously reported diminished degradation of FGF23 in Hyp mice, as evidenced by alterations in the ratio of intact FGF23 to total FGF23, which was 0.31 in control mice and 0.65 in Hyp mice (Table 2). In addition, serum PTH levels were significantly elevated in Hyp mice, whereas 1,25(OH) 2 D levels were significantly decreased in Hyp mice compared with age-matched control mice. Consistent with known effects of FGF23 on the kidney, Hyp mice exhibited reduction of Npt2a and Klotho transcripts (Table 3), increments in Cyp24a1 and decrements in Cyp27b1 (Table 3).
Osteocyte specific deletion of Fgfr1 in Phex-mutant Hyp mice partially rescues the hypophosphatemic rickets phenotype The conditional deletion of Fgfr1 in osteocytes of compound Hyp;Fgfr1 Dmp1-cKO -null mice resulted in partial rescue of the Hyp gross phenotype. In this regard, the tail length, a marker of rickets, was 20 shorter in Hyp mice compared to controls, but was only 0.50 shorter in compound Hyp;Fgfr1 Dmp1-cKO -null mice compared to Fgfr1 flox/+ control mice. Similarly, compound Hyp;Fgfr1 Dmp1-cKO -null mice displayed body weight and femur lengths intermediate to those of Hyp and control mice (Figure 2A and 3A).
More importantly, both the elevated serum FGF23 and hypophosphatemia in Hyp mice were significantly improved in compound Hyp;Fgfr1 Dmp1-cKO -null mice. Conditional deletion of Fgfr1 in Hyp osteocytes resulted in about 70% reduction in serum FGF23 (from 3607 to 1099 pg/ml) and about 50% increase in serum phosphorus (from 6.0 to 9.3 mg/dl, values not significantly different from age-matched control mice). Ablation of Fgfr1 in Hyp osteocytes also resulted in reductions of PTH to the normal range, changes associated with significant increments of serum 1,25(OH) 2 D and phosphate concentrations. We observed no significant differences in serum calcium concentrations across the four genotypes (Table 2). Serum FGF23 and phosphorus concentrations (207684 pg/ml and 9.860.63 mg/dl, respectively) were not significantly different in Fgfr1 flox/null (equivalent to Fgfr1 +/null ) mice compared to Fgfr1 flox/+ controls (155670 pg/ml and 9.661.34 mg/dl, respectively) ( In contrast to the nearly 70% reduction in both bone FGF23 proteins and circulating FGF23 concentrations in Hyp;Fgfr1 Dmp1-cKOnull mice, conditional deletion of Fgfr1 in Hyp osteocytes resulted in only a 28% reduction of Fgf23 transcripts in bone of Hyp;Fgfr1 Dmp1-cKO -null compared with Hyp mice ( Figure 3C &3D   and Table 1 & 2). In Hyp mice the ratio of intact and C-terminal FGF23 in serum was consistent with impaired FGF23 degradation in Hyp mice, as previously described [56,57], however conditional deletion of Fgfr1 in Hyp osteocytes did not affect the ratio of intact and C-terminal Fgf23 in the compound Hyp;Fgfr1 Dmp1-cKO -null mice ( Table 2).
Conditional deletion of Fgfr1 in Hyp osteocytes normalized Npt2a and Klotho expression in the kidney. The alterations in Cyp27b1 and Cyp24 transcripts that regulate vitamin D metabolism were also corrected in the kidney from compound Hyp;Fgfr1 Dmp1-cKO -null mice. These results are consistent with an effect of reduced circulating FGF23 to increase serum phosphate and 1,25(OH) 2 D levels in the compound Hyp;Fgfr1 Dmp1-cKO -null mice.
In addition to the improvement in serum biochemistries in Hyp mice by conditional deletion of Fgfr1 in osteocytes, there was a  partial rescue of the hypophosphatemic rickets phenotype in the compound Hyp;Fgfr1 Dmp1-cKO -null mice. Although trabecular bone volume was not corrected in compound mutant mice skeletal abnormalities remained, consistent with an intrinsic mineralization defect caused by Phex mutations, as previously described [58]; there was a partial correction of the length of femur and cortical thickness in the compound Hyp;Fgfr1 Dmp1-cKO -null mice. Growth plate abnormalities were also qualitative improved, as evidenced by a more organized structure. In addition, the area of unmineralized osteoid in cortical bone was reduced and an increase in mineral apposition rate was observed in Hyp;Fgfr1 Dmp1-cKOnull mice compared to age-matched Hyp mice (Figure 3 and 4). Fgf1, Fgf2, Fgfr2, Fgfr3, Fgfr4 as well as Dmp1, Mepe, Osteopontin, Osteocalcin, Galnt3, Mmp13, Mmp9, and Collagen II message levels were elevated in Hyp mice; and all of these transcripts were reduced in compound Hyp;Fgfr1 Dmp1-cKO -null mice, consistent with partial healing of bone and cartilage (Table 1). Conditional deletion of Fgfr1 in Hyp osteocytes also partially corrected the observed abnormalities of the Wnt signaling pathway, including reductions in Wnt10b, Axin-2, and Fzd2. However, Sost and Dkk1 remained suppressed in Hyp;Fgfr1 Dmp1-cKO -null mice.

Fgfr1 signaling regulates FGF23 protein expression in vitro via a PI3K-AKT dependent pathway
Since we found disproportionate reductions in serum circulating FGF23 protein and bone Fgf23 mRNA levels in compound Hyp;Fgfr1 Dmp1-cKO mice, and FGFR1 is known to enhance recruitment of RNA to polysomes to increase in protein expression in other systems [62], we examined if Fgfr1 regulate Fgf23 protein expression through an effect to stimulate mRNA translation. For these studies we transfected human FGF23-V5-His cDNA expression plasmid into mouse MC3T3-E1 osteoblasts and assessed epitope tagged FGF23 protein production before and after stimulation with FGF2. We found that FGF2 resulted in a time-dependent induction of FGF23-V5-His protein expression, achieving a maximal 3-fold stimulation at 24 hours after Fgf2 stimulation ( Figure 6A and 6B). Both the PI3K inhibitor Wortmannin and protein synthesis inhibitor Cycloheximide completely blocked FGF2-induced FGF23 protein increase ( Figure 6C and 6D), consistent with an effect of FGF2 to stimulate FGF23 protein translation through activation of PI3K/Akt pathway.

Discussion
Osteocytes, the most abundant cells in bone, play a central role in regulating bone remodeling and mineralization through the release of paracrine factors and systemic phosphate and vitamin D metabolism through the release of the hormone FGF23 [63]. FGFR1 is known to be involved in the regulation of bone development and remodeling [64,65]. Inactivating mutations of FGFR1 in autosomal dominant Kallmann syndrome [66] and the Pfeiffer craniosynostosis [67] and conditional deletion of Fgfr1 gene in osteo-chondro-progentitors using Col1a1(2.3)-Cre enhances osteoblast differentiation and endochondral bone formaton [65], suggesting that FGFR1 has osteoblast-stage specific effects. The specific role of FGFR1 in osteocytes, however, has not been previously defined. The current study is the first to selectively ablate Fgfr1 gene in osteocytes in bone.
We found that Dmp1-Cre-mediated loss of Fgfr1 in osteocytes resulted in selective alterations of multiple genes involved in bone remodeling and mineral metabolism. Fgfr1 Dmp1-cKO -null mice exhibited reductions in Fgf23 transcripts in bone and circulating levels of FGF23. In addition, targeted disruption of Fgfr1 in osteocytes resulted in reductions of Dmp1 and Mepe, two members Table 3. Gene-expression profiles in kidney in 6-week-old mice. of the SIBLING family of proteins that are expressed in osteocytes as well as Phex, an endopeptidase, that are involved in the regulation of bone mineralization. We also found in the Fgfr1 Dmp1-cKO -null mice evidence for reductions in Sost and Dkk1, which are produced by osteocytes and function as antagonists of the Wnt-signaling pathway, which is an important anabolic signal for bone. In contrast, Fgfr1 Dmp1-cKO -null mice exhibited no changes in more general markers of osteoblast differentiation, such as Osteopontin and Osteocalcin, and bone expression of Wnt10b and Axin-2 were decreased, not increased, in conditional the Fgfr1 Dmp1-cKO -null mice, as would be expected from reductions in Sost and Dkk1. In spite of these changes in osteocyte gene expression, we observed no abnormalities of bone structure or mineralization (i.e., no changes in bone volume or mineral apposition rates) in the Fgfr1 Dmp1-cKO -null mice under normal conditions. This may be due to the complexity of FGFR1 signaling in bone and cross-talk between downstream molecular targets affecting bone remodeling [68] that may have offset the effects of the loss-of-FGFR1 in osteocytes. Indeed, additional pathological perturbations caused by Phex mutations in Hyp mice uncovered a more evident role of Fgfr1 in osteocytes on bone structure and gene expression [29]. We found that conditional deletion of Fgfr1 in osteocytes of Hyp mice decreased Fgf23 transcripts in bone, as well as a lead to a striking reduction in circulating FGF23 levels in compound Hyp;Fgfr1 Dmp1-cKO -null mice. The reductions in FGF23 in Hyp mice partially rescued the hypophosphatemic rickets phenotype, as evidenced by increased serum phosphate levels, improved bone parameters and bone-related gene expression as well as improved kidney gene expression and 1,25(OH) 2 D metabolism. In addition, alterations in Fgf1 and Fgf2, as well as Fgfr2, Fgfr3, and Fgfr4 expression that are known to be increased in osteocytes derived from Hyp bone [55] were normalized in Hyp;Fgfr1 Dmp1-cKO -null mice ( Table 1), indicating that Fgfr1 regulates other members of the FGF/FGFR signaling family in Hyp osteocytes. Since FGF23 has dose-dependent effects on the kidney [69,70], the 3-fold reduction in circulating FGF23 in Hyp;Fgfr1 Dmp1-cKO -null mice likely accounts for the improvement in the hypophosphatemic rickets phenotype, whereas the persistent elevations of FGF23 would explain the residual hypophosphatemia and bone abnormalities in Hyp;Fgfr1 Dmp1-cKO -null mice. Alteration in matrix mineralization caused by loss of Phex function are persistent [41,58,71] and could contribute to persistent skeletal abnormalities are present in the compound Hyp;Fgfr1 Dmp1-cKO null mice.
Studies of FGF23 gene transcription and translation in osteoblast cultures confirmed an important role of FGFR1 in regulating FGF23 production. In this regard, we demonstrated that a dominant negative FGFR1(TK2) construct inhibits FGF2mediated activation of FGF23 gene transcription in cultured osteoblasts. Interestingly, inhibition of MAPK and PLCc pathways, but not PI3K/Akt pathways, blocked FGFR1 stimulation of FGF23-promoter activity in cultured osteoblasts. MAPK regulation of FGF23 transcription is consistent with the recent findings that activating somatic mutations of RAS causes FGF23-mediated hypophosphatemia in humans [72] and that ERK1/2 activation is involved in FGFR-mediated FGF23 transcription in UMR-106 osteoblasts in vitro [40]. Transfection of a HMW-FGF2 cDNA, an intracrine activator of FGFR1, also stimulated FGF23 promoter activity in osteoblast cultures. Thus, both autocrine/paracrine and intracrine activation of FGFR1 regulates FGF23 expression in osteoblasts.
Our studies also identified a possible role of FGFR1 in the posttranscriptional regulation of FGF23. In this regard, reductions in circulating Fgf23 levels were greater than the decrease in Fgf23 mRNA expression in bone of Hyp;Fgfr1 Dmp1-cKO -null mice, indicating a possible role of FGFR1 in regulation of Fgf23 mRNA translation. Indeed, we found that FGF2-FGFR1 signaling regulates FGF23 protein translation in vitro via a PI3K-AKT dependent pathway. These results are consistent with prior reports showing translational control role FGFR1 signaling in cancer and smooth muscle cells [62,73,74]. The dual transcriptional and posttranscriptional control of FGF23 by FGFR1 may explain discrepancies between FGF23 message and circulating levels of this hormone [75], as well as provide another therapeutic target, along with regulation of transcription and degradation [76], to modify circulating FGF23 levels in disease states.
Previous data have implicated FGFR1 signaling in the regulation of FGF23 expression, but these observations lack the cell-and organ-specificity of our studies [77,78]. For example, pharmacological inhibition of FGFR1 in vivo in Hyp mice [78] results in inhibition of both FGF23 end-organ effects as well as variable effects on FGF23 production by bone. Similarly, Y372C missense gain-of-function mutation in FGFR1 in osteoglophonic dysplasia (OGD) and systemic administration of activating antibodies to FGFR1 in mice elevate circulating FGF23 levels [79], but have the confounding effects of activation of FGFRs in both bone and kidney. Thus, use of non-specific inhibitors of FGFR1 or FGFR1 activating antibodies in vivo are limited by generalized effects that prevent distinguishing between actions on FGFR1 in osteocytes or indirect effects through other actions on FGFRs in the kidney or other tissues. In contrast, the selective deletion of Fgfr1 in osteocytes provides in vivo confirmation of prior in vitro studies showing that pan-FGFR inhibitors directly suppress FGF23 expression in Hyp-derived osteoblasts and FGF1 and FGF2-mediated FGFR1 activation directly stimulates FGF23 promoter activity in osteoblasts cultures [39,77].
Perhaps more importantly, we have discovered that the evolutionarily linked canonical (or autocrine/parcrine) and intracellular (or intracrine) FGFs are also physiologically coupled to the hormone-like FGF23 in bone through FGFR1-dependent mechanisms. FGF1 and FGF2 (both low and HMW forms), produced by the osteoblast lineage and stored in bone matrix, are increased in Hyp bone [37][38][39]. Osteocytes also participate in a peri-cellular demineralization process called osteolytic osteolysis that involves matrix degradation and growth factor activation by enzymes, such as Matrix metalloproteinase-9 and 13 (MMP-9 and -13), which are also increased in Hyp mice [80]. It is tempting to speculate that either the local release of stored FGFs from the extracellular matrix, or other factors regulating, FGF1, FGF2 and HMW-FGF2, are regulating osteocyte functions through FGFR signaling. Further studies are needed to define how alterations in bone metabolism and mineralization caused by Phex mutations regulate FGF ligands and FGFR signaling pathways in osteocytes. Since hormonal-like FGFs emerged early in vertebrate evolution, our findings raise the possibility that the mineralized skeleton developed a means to coordinate local and systemic functions by the linkage of autocrine/paracrine and intracrine FGF signaling with hormonal FGF23.
Our study has several limitations. First, conditional deletion of Fgfr1 failed to completely restore FGF23 expression to normal in the compound Hyp;Fgfr1 Dmp1-cKO -null mice. The persistent elevations of FGF23 might be due to incomplete ablation of Fgfr1 in osteocytes or continued production of FGF23 from osteoblasts not affected by Dmp1-Cre (e.g., 30% of Fgfr1 transcripts persisted in bone of Fgfr1 Dmp1-cko -null mice and Dmp1-Cre targets ,1% of osteoblasts in long bone). Alternatively, FGFR1-independent pathways known to regulate FGF23 expression may account for the elevations of FGF23. The possibility that other FGFRs in osteocytes might mediate the residual stimulation of Fgf23 in Hyp mice appears unlikely, since we observed a decrease in Fgfr2, Fgfr3, and Fgfr4 in Hyp;Fgfr1 Dmp1-cKO -null mice. In addition, prior studies indicate that conditional deletion of Fgfr2 in mice leads to no abnormalities of serum FGF23 or phosphate [81], and global loss of Fgfr3 and Fgfr4 enhances, rather than inhibits, FGF23 expression in bone, due to impaired end organ sensing of FGF23 by the kidney [82,83]. Second, although Dmp1-Cre had no kidney expression, to achieve the 70% reduction of Fgfr1 expression in osteocytes using Dmp1-Cre required using a strategy that paired a flox with a null Fgfr1 allele, which led to a 50% reduction in Fgfr1 expression in the kidney, and other tissues. Meaningful conclusions can be made about FGFR1 function in bone, however, because heterozygous Fgfr1 flox/null (equivalent to Fgfr1 +/null ) mice had no demonstrable renal phenotype and the reduction in FGF23 expression was dosedependently related to reductions in Fgfr1 expression in bone. We lack evidence that end-organ resistance to FGF23 actions on the kidney in Hyp;Fgfr1 Dmp1-cKO -null mice, since this would lead to hyperphosphatemia and elevated FGF23 levels, a phenotype different from the one observed in these mice.
In conclusion, conditional deletion of Fgfr1 in osteocytes from Hyp mice reduced circulating FGF23 concentrations through regulation of both the transcription and translation of FGF23. These studies establish a function of FGFR1 in osteocytes and show a physiological coupling between canonical FGFRs and hormone-like FGFs that may provide a mechanism to link local regulation of bone metabolism with systemic phosphate and vitamin D homeostasis. This evolutionary relation may permit local changes in bone metabolism regulated by FGFs-FGFRs to communicate with the kidney to permit FGFR1 regulation of bone formation and mineralization to be coordinated with phosphate absorption and/or vitamin D metabolism to match the bone needs for mineral with the renal handling of these minerals.