Overexpression of Fibroblast Growth Factor 23 Suppresses Osteoblast Differentiation and Matrix Mineralization In Vitro

Authors

  • Hua Wang,

    1. Oral Growth and Developmental Biology, Hiroshima University Graduate School of Biomedical Sciences, Hiroshima, Japan
    Current affiliation:
    1. Department of Molecular Oral Medicine and Maxillofacial Surgery, Guanghua College of Stomatology, Sun Yat-Sen University, Guangzhou, China
    Search for more papers by this author
    • These authors contributed equally to this work

  • Yuji Yoshiko PhD,

    Corresponding author
    1. Oral Growth and Developmental Biology, Hiroshima University Graduate School of Biomedical Sciences, Hiroshima, Japan
    • Department of Oral Growth and Developmental Biology, Hiroshima University Graduate School of Biomedical Sciences, 1-2-3 Kasumi, Minami-ku, Hiroshima 734-8553, Japan
    Search for more papers by this author
    • These authors contributed equally to this work

  • Ryoko Yamamoto,

    1. Orthodontics and Craniofacial Developmental Biology, Hiroshima University Graduate School of Biomedical Sciences, Hiroshima, Japan
    Search for more papers by this author
  • Tomoko Minamizaki,

    1. Oral Growth and Developmental Biology, Hiroshima University Graduate School of Biomedical Sciences, Hiroshima, Japan
    2. Pediatric Dentistry, Hiroshima University Graduate School of Biomedical Sciences, Hiroshima, Japan
    Search for more papers by this author
  • Katsuyuki Kozai,

    1. Pediatric Dentistry, Hiroshima University Graduate School of Biomedical Sciences, Hiroshima, Japan
    Search for more papers by this author
  • Kazuo Tanne,

    1. Orthodontics and Craniofacial Developmental Biology, Hiroshima University Graduate School of Biomedical Sciences, Hiroshima, Japan
    Search for more papers by this author
  • Jane E Aubin,

    1. Department of Molecular Genetics, Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
  • Norihiko Maeda

    1. Oral Growth and Developmental Biology, Hiroshima University Graduate School of Biomedical Sciences, Hiroshima, Japan
    Search for more papers by this author

  • The authors state that they have no conflicts of interest.

Abstract

Introduction: Fibroblast growth factor (FGF)23 is produced primarily in bone and acts on kidney as a systemic phosphaturic factor; high levels result in rickets and osteomalacia. However, it remains unclear whether FGF23 acts locally and directly on bone formation.

Materials and Methods: We overexpressed human FGF23 in a stage-specific manner during osteoblast development in fetal rat calvaria (RC) cell cultures by using the adenoviral overexpression system and analyzed its effects on osteoprogenitor proliferation, osteoid nodule formation, and mineralization. Bone formation was also measured by calcein labeling in parietal bone organ cultures. Finally, we addressed the role of tyrosine phosphorylation of FGF receptor (FGFR) in mineralized nodule formation.

Results: Nodule formation and mineralization, but not osteoprogenitor proliferation, were independently suppressed by overexpression of FGF23 in RC cells. Increased FGF23 levels also suppressed bone formation in the parietal bone organ culture model. FGF23 overexpression enhanced phosphorylation of FGFR, whereas the impairment of mineralized nodule formation by FGF23 overexpression was abrogated by SU5402, an inhibitor of FGFR1 tyrosine kinase activity.

Conclusions: These studies suggest that FGF23 overexpression suppresses not only osteoblast differentiation but also matrix mineralization independently of its systemic effects on Pi homeostasis.

INTRODUCTION

Fibroblast growth factor (FGF)23 was identified by positional cloning of the gene responsible for autosomal dominant hypophosphatemic rickets (ADHR), where elevated active FGF23 (gain-of-function mutations in the fgf23 gene in ADHR) in serum causes hypophosphatemia with resultant rickets/osteomalacia.(1,2) In addition, FGF23 seems to be involved in other phosphaturic disorders that share pathologies with ADHR, such as tumor-induced osteomalacia (TIO)(3) and X-linked hypophosphatemia (XLH).(4) However, the complexity of understanding the relationships and outcomes of reduced or elevated systemic levels of FGF23 and phosphate is underscored by observations on FGF23-null mice(5,6) versus transgenic mice overexpressing FGF23,(7–9) which have similar abnormalities in bone but distinct cartilage phenotypes.(5–9) Thus, additional data and alternative approaches are needed to elucidate fully the patho/physiological roles of FGF23 in skeletal development.

FGF23 is expressed primarily in skeletal tissues, most notably in osteoblasts and osteocytes,(10–12) and appears in the systemic circulation,(13) where it targets the renal proximal tubule to reduce the expression of the type II a and c sodium-dependent phosphate co-transporters (NPT2a and c), leading to decreased phosphate reabsorption.(14–16) FGF23 also suppresses the expression of 1α-hydroxylase, an essential enzyme for synthesis of the active vitamin D metabolite, 1,25(OH)2D,(3,5,15) which enhances intestinal phosphate absorption. Together, these two pathways account for the hypophosphatemia caused by an excess of serum FGF23.(3,7–9,14) Parathyroid has been recently established as an additional target of FGF23 to inhibit PTH expression,(17) providing a new aspect for understanding the role of FGF23.

Most FGF family members exert their functions through interactions with FGF receptors (FGFRs). To date, four FGFRs, each with alternatively spliced variants, are known, along with many aspects of their tissue distribution, ligand selectivity, and linkage of signaling pathways to biological effects.(18–21) In particular, FGF23 seems to have the ability to interact with FGFR1c, FGFR2c, FGFR3c, and FGFR4.(22–25) Despite the ubiquitous presence of FGFRs, the induction of FGF23-responsive downstream molecules, such as activation of early growth response-1 and phosphorylation of extracellular signal-regulated kinase (ERK), is detected in a limited number of tissues, including kidney, parathyroid, and pituitary glands, when mice and rats were treated with recombinant FGF23 protein.(17,22–25)

The discovery that Klotho, an aging suppressor protein, is needed to form a complex with FGF23–FGFR for induction of FGF23-specific pathways provides important insights into the unique activities of FGF23 in phosphate metabolism.(24,25) Thus, although the expression of the klotho gene has not been detected in bones even by RT-PCR,(26) its secreted form is present in the systemic circulation.(26,27) Given that FGF23 is most highly expressed in bones(10–12) and that multiple FGFs and FGFRs have been implicated in skeletal development,(20,21) we therefore hypothesize that FGF23 has a direct role in bone formation. To test this, we analyzed the effects of overexpression of human FGF23 (hFGF23) on bone formation in the well-established fetal rat calvaria (RC) cell and organ culture models.

MATERIALS AND METHODS

Construction of adenovirus vector carrying hfgf23

Full-length hFGF23 cDNA was prepared from human thymus total RNA (Clontech Laboratories, Mountain View, CA, USA) (see below), cloned, and introduced into adeno-X viral DNA (BD Biosciences, Palo Alto, CA, USA). The hFGF23 recombinant adenoviruses (Adv-hFGF23) were generated by the transfection of Adv-hFGF23 into 293 cells. The adenovirus particles were purified using the BD-Adeno-X virus purification kit (BD Biosciences). Recombinant adenovirus expressing β-galactosidase (Adv-βgal; BD Biosciences) was used as control.

RC cell culture

Timed-pregnant Wistar rats were maintained in accordance with guidelines for the Committee of Animal Experimentation and Research Facilities for Laboratory Animal Science, Hiroshima University. RC cells were isolated from 21-day-old fetuses as described.(28) Briefly, calvariae were sequentially digested with collagenase (type I; Sigma-Aldrich, St Louis, MO, USA), and osteogenic fractions (the last four of five fractions) were precultured for a day in αMEM supplemented with 10% FCS (Biological Industries, Kibbutz beit Haemek, Israel). Adherent cells were trypsinized and replated at 0.2 × 104/cm2 in the same medium supplemented additionally with 50 μg/ml ascorbic acid (osteogenic medium). To induce matrix mineralization, cultures were treated with 10 mM β-glycerophosphate (βGP) for 48 h before culture termination (Fig. 2A). For immunocytochemical detection of FGFRs and FGF23, relatively mature osteoblastic cells (day 14 of culture) were collected by collagenase digestion and replated onto coverslips, followed by incubation in osteogenic medium for 24 h.

In preliminary experiments, cells were co-infected with Adv-βgal and regulatory adenovirus (Tet-Off; 1:1) at MOIs of 10–100 pfu/cell, according to the manufacturer's instructions. As assessed by 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-gal) staining and cell counts, we determined that cells were optimally infected with the two adenoviruses each at an MOI of 10 pfu/cell (henceforth, the description of Tet-Off is abbreviated). X-gal staining also showed that the transduction efficiency did not differ across different cellular developmental stages (data not shown). FGF23 expression in Adv-hFGF23 cultures was 10-fold higher or more than control (Adv-βgal); Adv-βgal was also shown not to induce cytotoxicity at the MOI (10 pfu/cell) used in experiments reported (see below). FGF23 overexpression was achieved without doxycycline; doxycycline (≤2.5 ng/ml) was used as the negative control in experiments assessing FGF23 turn-off.

Cells were infected with adenoviruses at day 4 or day 14 (immature or mature stages of osteoblast development; Fig. 2A).(28) Cells infected early (day 4) were used to assess cell proliferation and differentiation, whereas more mature cells (day 14) were used to assess matrix mineralization. To evaluate FGFR phosphorylation immunocytochemically, some infected cells (day 14) were collected at day 15 as above, replated onto coverslips, and maintained in osteogenic medium for a day, followed by incubation under low serum conditions (0.5% FCS) for 24 h. Some of these subcultures were maintained in the presence of SU5402, an inhibitor of FGFR1 tyrosine kinase (≤10 μM; Signaling Technology, Beverly, MA, USA)(29) or vehicle (DMSO) alone. In other cases, the cells infected at day 14 were treated with SU5402 from day 15 until culture termination.

Conditioned media were collected at day 16 from cell cultures infected with Adv-βgal or Adv-hFGF23 at day 14. Media in noninfected cells (day 16) were replaced with the conditioned media together with βGP, and cultures were maintained for an additional 2 days.

All cultures were maintained at 37°C in a humidified atmosphere with 5% CO2, and medium was changed every second or third day, except as designated in specific experiments.

Fetal rat parietal bone organ culture

Fetal rat (embryonic day 21) parietal bones were cut into halves along the sagittal suture and placed onto cell culture inserts (Becton Dickinson, Franklin Lakes, NJ, USA) in BGJb medium supplemented with 20% FCS. The bones were co-infected with Adeno-X Tet-Off and either Adv-hFGF23 or Adv-βgal at an MOI of 1 × 106 pfu/bone and incubated for 4 days as above; an appropriate MOI was estimated by counting the number of cells obtained by digestion from a piece of the parietal bones. Medium was changed by half every day. To quantify bone formation, calcein (1 μg/ml) was added into cultures 2 days before culture termination. After fixation with 4% paraformaldehyde (PFA) in PBS, images of frozen sections (7 μm in thickness) were recorded on a fluorescence microscope and measured using an image densitometer. Thirty-six fluorescence images randomly selected from 12 parietal bone pieces (three images in each) in each group were recorded, and data are expressed as an average fluorescence intensity. Immunohistochemical staining (see below) for FGF23 was also performed with counterstaining (methyl green).

X-gal staining

The percentage of lacZ-positive cells was determined with X-gal staining. Cells were fixed in 4% PFA in PBS, and X-gal–positive (X-gal+) cells, as well as total cells, were counted under a phase-contrast microscope. The mean transduction efficiency of randomly selected fields (five independent areas in each field) was calculated.

Alkaline phosphatase and von Kossa staining

To evaluate osteoblast development and matrix mineralization, cells were fixed in 10% neutral buffered formalin and treated with naphthol AS MX/red violet LB in 0.1 M Tris-HCl (pH 8.3) to detect alkaline phosphatase (ALP) activity. Matrix mineralization was confirmed by further incubation with 2.5% silver nitrate solution.(28) The number of ALP+ nodules and mineralized foci in each well or dish was counted under a stereoscopic microscope.

MTT assay

To evaluate the effect of FGF23 overexpression on cell proliferation and viability, an MTT assay was carried out. Briefly, cells infected at day 4 with Adv-hFGF23 or Adv-βgal were evaluated before (day 6) and after (day 18) differentiation. To determine osteoprogenitor proliferation, parallel cell cultures were trypsinized and replated (0.3 × 104/cm2) at day 5 and subjected to the MTT assay daily until cells reached confluence.

Western blotting

Cells were lysed with 100 mM KCl, 1 mM EDTA, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonylfluoride, and complete protease inhibitor (Roche Diagnostics, Penzberg, Germany) in 50 mM Tris HCl (pH 7.5). After centrifugation, supernatants were collected and stored at −80°C until use. Aliquots of samples were subjected to SDS-PAGE on 15% gels under reducing conditions and electroblotted onto nitrocellulose membranes (Millipore, Bedford, MA, USA). The membranes were incubated with TTBS (0.1% Tween 20 and 0.1 M NaCl in 0.1 M Tris-HCl, pH 7.5) containing 0.3% casein and with primary antibodies (polyclonal; 1:1000, FGF23, FGFR1, FGFR2, and FGFR3; Santa Cruz Biotechnology, Santa Cruz, CA, USA; and 1:1000, phospho-FGFR; Cell Signaling Technology, Danvers, MA, USA) at 4°C overnight. The membranes were incubated with horseradish peroxidase–conjugated secondary antibodies (1:2000; Santa Cruz Biotechnology) for 1 h, followed by chemiluminescence detection (Lumi-LightPLUS Western Blotting Detection System; Amersham Biosciences, Buckinghamshire, UK).

ELISA

Conditioned media from cells were collected and stored at −80°C until use. Levels of FGF23 were determined with an FGF23 ELISA kit (Kainos Laboratory, Tokyo, Japan), according to the manufacturer's directions.

Immunofluorescence staining

Newborn rat calvariae were fixed in 4% PFA in PBS, decalcified with EDTA, and embedded in paraffin. Cultured cells on coverslips were fixed with 4% PFA in PBS and permeabilized with 0.1% Triton-X 100 in PBS or by freeze-thawing. Multiple serial sections (5 μm in thickness) and cells were routinely processed by blocking (Protein Block; DakoCytomation, Carpinteria, CA, USA), followed by incubation with primary antibodies (1:100, see above) at 4°C overnight. The specimens were treated with fluorescein isothiocyanate (1:2000; Santa Cruz Biotechnology) – or Cy 3 (1:400; Jackson ImmunoResearch Laboratories, West Grove, PA, USA) –conjugated secondary antibodies for 1 h and viewed with a fluorescence microscope. To evaluate FGFR phosphorylation in RC cells, 12 fluorescence images randomly selected from three coverslips (4 images in each coverslip) were analyzed as above. For histological examination, paraffin sections were stained with H&E.

RNA extraction and real-time RT-PCR analysis

Total RNA was routinely isolated from tissues or cells with TRIzol reagent (Invitrogen, Carlsbad, CA, USA). cDNA was synthesized from 2 μg of total RNA using ReverTraAce (Toyobo, Osaka, Japan) at 50°C for 40 min. Real-time PCR analysis was performed with LightCycler (Roche Diagnostics, Mannheim, Germany) by using SYBR Green (SYBR Green Realtime PCR Master Mix; Toyobo), according to the manufacturer's directions. Each primer set for rat Runx 2, type I(α) collagen [COLI(α)], osteopontin (OPN), ALP, bone sialoprotein (BSP), osteocalcin (OCN), FGF23, FGFR1, FGFR2b, FGFR2c, FGFR3, FGFR4, and ribosomal protein L32 (internal control) was designed using Primer Picking (primer 3) or as described elsewhere(12,23,28) (Table 1). The expression of FGFR isoforms and FGF23 was assessed by regular RT-PCR. PCR amplification conditions were as follows: denaturing, 95°C; annealing, 62°C for COLI(α), OPN, ALP, BSP, OCN, FGFR1, FGFR2b and c, and L32, and 59°C for Runx2, FGFR3 and FGFR4; extension, 72°C. PCR products were verified by melting curve or sequencing.

Table Table 1.. Primer Sequences for FGF23, FGFRs, and Osteoblast Markers
original image

Statistical analysis

Data are expressed as the means ± SD of replicate samples (n = 3–12, depending on the experiments), and experiments were repeated, at a minimum, two times. Differences between groups were analyzed by one-way ANOVA using Fisher's test for multiple comparisons and t-test for two groups comparisons.

RESULTS

Overexpression of hFGF23 in RC cell and organ cultures inhibits not only osteoblast development but also matrix mineralization

In preliminary experiments, we used X-gal staining to determine the optimal conditions for overexpression in cells infected with Adv-βgal and found that 10 pfu/cell worked well and without toxicity, with βgal expression detected at least from day 2 to day 5, but not day 8 after infection at either earlier (day 4; Fig. 1A) or later (day 14, data not shown) time in culture. Consistent with this, Western blotting and/or ELISA confirmed a marked increase in FGF23 expression and/or secretion at day 2 and day 5 after infection of day 4 cells with Adv-hFGF23 (Figs. 1B and 1C). Furthermore, overexpression of FGF23 was downregulated by the addition of doxycycline in a dose-dependent manner, indicating the fidelity of the Tet-Off regulation of FGF23 overexpression (Fig. 1D).

Figure Figure 1.

Adenovirus-mediated overexpression of βgal and hFGF23 in RC cell cultures. (A) Transduction efficiency of Adv-βgal. Cells at day 4 were infected with Adv-βgal at an MOI of 10 pfu/cell. The percentage of X-gal+ cells is shown at each day up to day 8 after infection. (B–D) Adenoviral overexpression of hFGF23. Cell lysates and conditioned media were collected at day 2 and/or day 5 after infection with Adv-hFGF23 or Adv-βgal and analyzed by Western blot analysis of cell lysates for FGF23 (day 2) (B) and ELISA for secreted FGF23 (days 2 and 5) (C). (D) Effect of doxycycline on FGF23. Cells were incubated in the presence or absence of doxycycline for 2 days and analyzed by Western blot.

Based on these results, in experiments reported here, we overexpressed hFGF23 during the three typical developmental time windows in the bone formation process in RC cultures: progenitor cell proliferation, osteoblast differentiation with matrix deposition (nodule formation), and mineralization(28) (Fig. 2A). We first infected proliferating progenitor cells (day 4) with Adv-hFGF23 or Adv-βgal, trypsinized and replated the cells the next day, and continued culturing until cells reached confluence (6 days). The MTT assay confirmed that Adv-hFGF23 did not significantly affect cell growth compared with control (Adv-βgal; Fig. 2B). We therefore cultured the same cells and maintained them up to day 18 to determine the effect of FGF23 overexpression on osteoblast differentiation (Fig. 2A). Although we found no cytotoxic effect in either culture (Fig. 3A), significantly fewer ALP+ nodules formed in Adv-hFGF23– versus Adv-βgal–infected cultures and Adv-hFGF23 plus doxycycline cultures (183.5 ± 13.2 in Adv-βgal and 170.3 ± 20.9 in Adv-hFGF23 plus doxycycline versus 28.6 ± 0.9 in Adv-hFGF23; Figs. 3B and 3C). In all further experiments, we used a combination of Adv-βgal and Tet-Off as control. Real-time RT-PCR analysis confirmed a marked decrease in ALP (3-fold); concomitantly, OCN expression was markedly reduced (40-fold), whereas BSP levels were changed slightly (increased, 1.8-fold) and Runx2, OPN, and COLI(α) were not significantly changed in Adv-hFGF23 versus Adv-βgal cells (Fig. 3D).

Figure Figure 2.

Outline of differentiation and treatment protocols during well-established time windows in RC cell cultures (A) and cell growth assay (B). FGF23 was overexpressed during three typical osteoblast-bone developmental time windows (i.e., progenitor cell proliferation, osteoblast differentiation with matrix deposition, and mineralization). The MTT assay was carried out daily in subcultures of Adv-hFGF23– vs. Adv-βgal–infected cells at day 4. Results are means ± SD of triplicate determinations. All data are representative of three independent experiments.

Figure Figure 3.

FGF23 overexpression during nodule formation stages suppresses osteoblast differentiation in RC cell cultures. Cells were infected with Adv-hFGF23 or Adv-βgal at day 4 and maintained up to day 18. Some of the Adv-hFGF23 cells were cultured in the presence of doxycycline to downregulate FGF23 overexpression (Tet-Off). (A) Cell viability before and after adenoviral infections. The MTT assay was carried out as shown in Fig. 2B. (B) The number of ALP+ nodules in the presence or absence of 2.5 ng/ml doxycycline. bp < 0.01 compared with other groups. (C) Macroscopic images of ALP staining. (D) Real-time RT-PCR analysis of osteoblast markers. L32 was used as internal control. ap < 0.05; bp < 0.01; cp < 0.001. Results in A, B, and D are means ± SD of triplicate determinations. All data are representative of three independent experiments.

To evaluate the effect of Adv-hFGF23 on matrix mineralization separately from the differentiation effect described above, cells were infected with the recombinant adenoviruses when ALP+ nodules were already well developed but not yet mineralized (day 14; Fig. 2A). Two days later, when adenoviral overexpression is active (Fig. 1), cultures were treated with βGP to induce matrix mineralization and were further incubated for 2 days. Adv-hFGF23 significantly decreased the number of mineralized foci (301.3 ± 37.3 in Adv-βgal versus 21.1 ± 3.3 in Adv-hFGF23) without affecting the number of ALP+ nodules present (217.4 ± 13.2 in Adv-βgal versus 231.4 ± 18.3 in Adv-hFGF23; Figs. 4A–4C). We also confirmed mineralization defects when we treated day 16 cells with conditioned media from cells infected with Adv-hFGF23 but not with Adv-βgal (data not shown). Therefore, an excess of secreted FGF23 seems to be involved in hypomineralization in our model. Among osteoblast markers, mRNA levels of COLI(α) (1.3-fold decrease), OCN (2.1-fold decrease), and ALP (1.7-fold increase) were significantly changed but not those for Runx2, OPN, and BSP in Adv-hFGF23–infected late-stage cultures (Fig. 4D).

Figure Figure 4.

FGF23 overexpression during nodule mineralization stages blocks βGP-induced matrix mineralization in RC cell cultures. Cells infected with Adv-hFGF23 or Adv-βgal at day 14 were maintained up to day 18. To induce mineralization, βGP was added into cultures at day 16. (A and B) The number of ALP+ nodules and mineralized foci, respectively. bp < 0.01. (C) Macroscopic images of ALP/von Kossa staining. (D) Real-time RT-PCR analysis of osteoblast markers. Relative mRNA levels are indicated as shown in Fig. 3D. ap < 0.05; cp < 0.001. Results in A, B, and D are means ± SD of triplicate determinations. All data are representative of three independent experiments.

To assess whether cellular responses to FGF23 overexpression are reflected in other measures of bone formation, we next infected rat parietal bones (day 21 fetuses) with Adv-hFGF23 or Adv-βgal and labeled with calcein in an organ culture model. Consistent with the increased expression of FGF23 as shown by immunohistochemistry (Fig. 5A) and the cell culture results (Figs. 3 and 4), calcein labeling of bone surfaces was markedly decreased (2-fold) in Adv-hFGF23– versus Adv-βgal–infected organ cultures (Figs. 5A and 5B). Taken together, these results suggest that an excess of FGF23 in osteoblastic cultures causes severely impaired bone formation in vitro.

Figure Figure 5.

FGF23 overexpression decreases bone formation in parietal bone organ cultures. Parietal bones from 21-day-old rat fetuses were infected with Adv-hFGF23 or Adv-βgal and maintained for 4 days. Calcein was added into cultures at day 2. (A) Immunohistochemical detection of FGF23 and fluorescence images of accumulated calcein in newly formed bone. (B) Histomorphometric analysis of calcein accumulation. Results are means ± SD. ap < 0.05. n = 12 in each group. All data are representative of two independent experiments. Scale bars, 10 μm.

FGF23 acts directly on osteoblasts through FGFR in RC cell cultures

Because FGF23 is known to bind to the c splice isoforms of FGFR1, FGFR2, FGFR3, and FGFR4,(22–25) we next asked which FGF23 receptors are expressed in RC cells. RT-PCR analysis of developing osteoblastic cells (d14) showed the expression of the c splice variants but not other splice forms of FGFR1, FGFR2, FGFR3, and FGFR4 (Fig. 6A). During osteoblast development, real-time RT-PCR indicated that FGFR3c was relatively more abundant at mature stages, whereas FGFR1c and FGFR2c were stably expressed throughout the culture and differentiation time course (Fig. 6B). Immunofluorescence staining confirmed the presence of FGFR1, FGFR2, FGFR3, and FGF23 in RC cells (Fig. 6C). These results were further supported by immunofluorescence staining of parietal bones, which provided evidence that FGF23 and FGFR1, FGFR2, and FGFR3 were co-localized in osteoblasts lining the bone surface (Fig. 6D).

Figure Figure 6.

Expression profiling of FGF23 and FGFRs in RC cells. (A) RT-PCR analysis of FGFRs and FGF23. Total RNA was extracted from mature RC cells at day 14. These cells (day 14) express FGF23, FGFR1c (R1c), R2c, and R3c but not FGFR1b (R1b), R2b, R3b, and FGFR4 (R4). (B) Real-time RT-PCR analysis of the c splice isoforms of FGFR1–FGFR3 during osteoblast development. Osteoblast markers (OPN, ALP, BSP, and OCN) are also shown. Results are means ± SD of triplicate determinations. (C) Immunofluorescence staining of FGF23 and FGFRs (R1–R3) in osteoblastic cells. Cells at day 14 were subcultured for 24 h, fixed, and stained. NC, negative control. (D) Immunofluorescence staining for FGF23 and R1, R2, and R3 in fetal rat parietal bones. Parietal bones from 21-day-old rat fetuses were fixed and embedded in paraffin. FGFRs are co-localized with FGF23 in osteoblasts on bone surfaces. NC, negative control. Scale bars, 10 μm. All data are representative of three independent experiments.

Our data suggest that an excess of secreted FGF23 may act on osteoblasts. We therefore next asked whether FGFR signaling is needed for the inhibition of mineralized nodule formation by overexpressing FGF23 in the RC cell model with and without the inhibitor of FGFR1 kinase activity, SU5402. Under low serum conditions (0.5% FCS) for 24 h, immunofluorescence staining showed that the intensity of phospho-FGFR was 2.5-fold higher in Adv-hFGF23–infected compared with Adv-βgal–infected cells (Figs. 7A and 7B). Treatment with SU5402 decreased the levels of FGFR1 phosphorylation seen in Adv-hFGF23 cells (Figs. 7A and 7B). We determined whether the upregulation of FGFR phosphorylation by FGF23 overexpression mediates the inhibition of mineralized nodule formation. SU5402 slightly increased the number of mineralized nodules in Adv-βgal– (not significant, 212.8 ± 31.2 versus 246.6 ± 18.2) but markedly increased the number in Adv-hFGF23–infected cultures (41.6 ± 8.9 versus 169.3 ± 10.7; Figs. 7C and 7D). These results suggest that the FGFR signaling response to an excess of FGF23 in osteoblasts negatively regulates mineralized nodule formation in RC cell cultures.

Figure Figure 7.

FGFR signaling mediates impaired bone nodule formation induced by FGF23 overexpression in RC cell cultures. (A and B) Relatively mature osteoblastic cells at day 14 were infected with Adv-hFGF23 or Adv-βgal. The next day, cells were collected, replated, and maintained in osteogenic medium for a day, followed by incubation in low serum conditions (0.5% FCS) for another day. In some cases as indicated, SU5402 (10 μM) was added into subcultures. (A) Immunofluorescence images of phospho-FGFR. (B) Densitometric analysis of fluorescence intensity of phospho-FGFR. Results are means ± SD of 12 randomly selected images in each group. ap < 0.05 compared with other groups. (C and D) Effects of SU5402 on mineralization defects induced by FGF23 overexpression. Cells at day 14 were infected with Adv-hFGF23 or Adv-βgal and maintained in the presence or absence of SU5402. Cultures at day 18 were stained with the ALP/von Kossa method. (C) Macroscopic images of ALP/von Kossa staining. (D) The number of mineralized foci. Results are means ± SD of triplicate determinations. bp < 0.01 compared with other groups. Data are representative of three independent experiments.

DISCUSSION

Adenoviral overexpression of FGF23 has no effect on osteoprogenitor cell proliferation but markedly suppresses not only osteoblast differentiation but also matrix mineralization in RC cell cultures. Increased FGF23 levels also reduce calcein accumulation in rat parietal bone organ cultures. RC cells express FGFR1c, FGFR2c, FGFR3c, and FGF23, and FGF23 overexpression in RC cells upregulates FGFR phosphorylation. SU5402, an inhibitor of the tyrosine kinase activity of FGFR1, rescues RC cell cultures from the inhibition of matrix mineralization induced by an excess of FGF23. These results suggest that overexpression of FGF23 negatively regulates bone formation independently of its systemic effects on Pi homeostasis.

Establishing unambiguously the direct effects of FGF23 on mineralization has been difficult and somewhat controversial and confusing. For example, genetically engineered mice lacking FGF23(5,6) or overexpressing wildtype or mutant FGF23(7–9) exhibit similar bone anomalies, although serum phosphate levels in the two groups exhibit an opposite pattern; the former show hyperphosphatemia, whereas the latter show hypophosphatemia. Transgenic mice expressing FGF23 under the control of the COLI(α) promoter exhibit high serum PTH and normal 1,25(OH)2D and calcium levels,(7) whereas their counterparts expressing FGF23 under the control of the chicken β-action promoter show low levels in all three parameters.(8) Thus, our in vitro data uniquely and clearly show the direct effect of FGF23 overexpression on bone formation without confounding variables of changes in systemic parameters.

Although the downstream effectors of FGF23 activity is still undefined, we also clearly show that FGFR signaling is involved in FGF23-dependent bone formation defects in the RC cell model. Thus, our findings provide new insight into the role of FGF23 in skeletal development (i.e., FGF23 acts on bone formation directly and independently of its systemic effect on Pi homeostasis). Although we did not compare the biological properties of FGF23 with other members of the FGF family in our models, several aspects of our results suggest that the skeletal activity of FGF23 is unique. First, we found no evidence that FGF23 stimulates osteoprogenitor cell proliferation that all of FGF2, 4, 8, 9, and 18 do in multiple bone formation studies in vitro and in vivo.(20,21,30–39) On the other hand, subcutaneous insertion of FGF2 beads in fetal mouse heads has been reported to reduce proliferation and increase suture fusion.(32,36) Thus, we cannot attribute the effects of FGF23 overexpression on osteoblast marker expression, decreased nodule formation, and mineralization in Adv-hFGF23–infected cells to an effect on proliferation. Second, we previously showed that matrix mineralization can be analyzed separately from and independently of other development stages in RC cell cultures.(40) Therefore, by overexpressing FGF23 after nodules had formed but before their mineralization, we uncovered a direct role for FGF23 in matrix mineralization independent of its effects on differentiation. Although other FGF members have been reported to play roles in matrix mineralization,(30,34,38,39) their effects have been proposed to reflect primarily effects on cellular proliferation.(30–32,34,36,37,39) However, treatment with recombinant human FGF2 (≥10 ng/ml; R&D Systems, Mountain View, CA, USA) inhibited matrix mineralization under the same conditions as we used for our FGF23 studies (data not shown). In fact, FGF2 is known to have a higher affinity for FGFR1c than to the other FGFR isoforms(19) and is not bound with Klotho.(24,25) Thus, a systematic and comprehensive analysis of FGF actions in osteoblasts, at physiological and pathological levels and including cross-talk among FGF members, remains of interest.

Previously, in situ hybridization showed an increase and decrease, respectively, in OPN and OCN mRNA levels in tibial osteoblasts of fgf23−/− versus wildtype mice,(6) although exactly how this relates to the bone anomalies seen in fgf23−/− mice remains unclear. The effects of FGF23 overexpression on osteoblast marker expression reported here are also complex but presumably depend on developmental and maturational stages of osteoblasts and their precursors. It is interesting that Runx2, COLI(α), BSP, and OPN were relatively unaffected by FGF23, whereas OCN and ALP were significantly affected in both differentiation and mineralization time windows. Whereas functional roles of OCN continue to be dissected, it is known to influence mineralization through binding to hydroxyapatite and functions in cell signaling and recruitment of osteoblasts,(41) which should be considered as components downstream of FGF23 signaling. The roles of ALP in mineralization and cellular phosphate homeostasis are also notable and deserve further attention.(42) In any case, a search of the downstream effectors of FGF23 activity in osteoblasts, as suggested above, should be useful to further address these issues.

Our expression profiling of FGFRs in RC cell cultures is consistent with previous results from in situ hybridization of developing fetal mouse calvariae.(33) FGFR1c, FGFR2c, and FGFR3c are all expressed in the sutural osteogenic front with relatively weak staining for FGFR1c and intense staining for FGFR2c; FGFR1b and FGFR4 mRNA were not detected in the developing calvarial bones.(33) Our results showed that treatment with Adv-hFGF23 increased tyrosine phosphorylation of FGFR and that the selective FGFR1 tyrosine kinase inhibitor, SU5402, rescued mineralization in FGF23-treated cultures, suggesting that biologically relevant FGF23 signaling in osteoblasts may be mediated, at least in part, by FGFR1.

Although Klotho may be needed for at least some FGF23-specific signaling events,(24,25) its role in the osteoblast effects we present here is not yet known. However, it is interesting to note that Klotho and FGF23 knockout mice display overlapping bone phenotypes, which include reduced bone volume and matrix mineralization.(5,6,43,44) ALP activity and mineralized matrices in osteoblast cultures from the Klotho-deficient mice are significantly lower than those from wildtype mice,(45,46) even though Klotho is not detectably expressed in skeletal tissues.(26) Taken together with the results of FGF2 in our model, further studies will be necessary to identify the direct roles of Klotho in osteoblast development; our data do not negate the possibility that Klotho, especially its secreted form,(26,27) may participate in FGF23-specifc signaling in osteoblasts.

In conclusion, this study showed that an excess of FGF23 negatively regulates both osteoblast differentiation and matrix mineralization in an autocrine/paracrine manner, at least in part through FGFR1. Further studies to understand the local effects of FGF23 on bone formation will be of great importance to understanding the pathology of rickets/osteomalacia and phosphate homeostasis.

Acknowledgements

This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan (16591828 and 18592001 to YY), Research for Promoting Technological Seeds from Japanese Science and Technology Agency (10-068 to YY), and the Canadian Institutes of Health Research (FRN 483033 to JEA).

Ancillary