Sympathetic Activation Induces Skeletal Fgf23 Expression in a Circadian Rhythm-dependent Manner

Background: The mechanism whereby the circadian clock regulates phosphate metabolism remains elusive.

Results: Fgf23 expression is regulated by the time of food intake which involves the alteration in circadian profile of sympathetic activity.

Conclusion: The circadian network plays important roles in phosphate metabolism.

Significance: The sympathetic regulation of Fgf23 expression may shed light on new regulatory networks that could be important for phosphate homeostasis.

Abstract

The circadian clock network is well known to link food intake and metabolic outputs. Phosphorus is a pivotal nutritional factor involved in energy and skeletal metabolisms and possesses a circadian profile in the circulation; however, the precise mechanisms whereby phosphate metabolism is regulated by the circadian clock network remain largely unknown. Because sympathetic tone, which displays a circadian profile, is activated by food intake, we tested the hypothesis that phosphate metabolism was regulated by the circadian clock network through the modification of food intake-associated sympathetic activation. Skeletal Fgf23 expression showed higher expression during the dark phase (DP) associated with elevated circulating FGF23 levels and enhanced phosphate excretion in the urine. The peaks in skeletal Fgf23 expression and urine epinephrine levels, a marker for sympathetic tone, shifted from DP to the light phase (LP) when mice were fed during LP. Interestingly, β-adrenergic agonist, isoproterenol (ISO), induced skeletal Fgf23 expression when administered at ZT12, but this was not observed in Bmal1-deficient mice. In vitro reporter assays revealed that ISO trans-activated Fgf23 promoter through a cAMP responsive element in osteoblastic UMR-106 cells. The mechanism of circadian regulation of Fgf23 induction by ISO in vivo was partly explained by the suppressive effect of Cryptochrome1 (Cry1) on ISO signaling. These results indicate that the regulation of skeletal Fgf23 expression by sympathetic activity is dependent on the circadian clock system and may shed light on new regulatory networks of FGF23 that could be important for understanding the physiology of phosphate metabolism.

Introduction

Phosphorus is an indispensable nutritional element involved in numerous biological processes such as cell signaling, energy homeostasis, and bone metabolism (1–4). The regulation of phosphate metabolism is an integrated process involving multiple organs and accumulating evidence has demonstrated the pivotal roles of fibroblast growth factor 23 (FGF23)² in phosphate metabolism (4–9). FGF23 is produced mainly by osteoblastic cells, including osteocytes, and functions as an endocrine factor to regulate genes involved in phosphate and vitamin D metabolism (4). The nodal point of the regulation of phosphate metabolism by FGF23 seems to primarily reside in the suppression of NaP_i-IIa/c expression and 1,25-dihydroxyvitamin D production in the kidney (4). Clinical evidence from genetic disorders in which mutations in the FGF23 gene or mutations causing aberrant FGF23 signaling are associated with dysregulated phosphate metabolism has placed bone-derived FGF23 in the center of regulatory networks of phosphate metabolism (10–12). Hence, it is critical to understand the mode of the regulation of FGF23 expression in the skeleton to fully understand the physiological and pathological functions of FGF23 in phosphate metabolism. Although previous studies have revealed that 1,25-dihydroxyvitamin D can stimulate Fgf23 expression in bone in part by directly activating the Fgf23 gene promoter (13–15), the precise mechanisms by which skeletal Fgf23 expression is regulated remain largely elusive. Because serum phosphate levels have been shown to exhibit circadian profile in humans, it is likely that phosphate metabolism is under the regulation of the circadian clock system (16–18); however, the precise mechanism by which the circadian clock network regulates phosphate homeostasis is still largely unknown.

The circadian clock network is an evolutionarily conserved process by which organisms adapt to environmental cues such as the availability of nutrients (19–21). For example, when food access is restricted in mice in the daytime (light phase) only, the expression profiles of circadian clock genes and circadian-regulated genes related to metabolic outputs have been shown to exhibit a phase shift so that the organisms can utilize ingested nutrients in a timely manner (19, 22–24). The central pacemaker of the circadian clock system is located at the suprachiasmatic nucleus in the hypothalamus and is integrated by multiple steps including transcriptional, translational, and post-translational mechanisms (20). Briefly, Clock (circadian locomotor output cycles protein kaput) heterodimerizes with Bmal1 (brain and muscle ARNT-like 1; also known as ARNTL) and induces the expression of PER (period circadian protein) and CRY (cryptochrome), which in turn suppresses Clock/Bmal1 transcriptional activity, thereby forming a 24-h feedback loop (20).

The mechanisms by which nutrient availability affects the circadian clock network still need to be determined; however both central and peripheral networks have been implicated as functional in this regulation (20). Centrally, the food-entrainable oscillator, which is anatomically different from the suprachiasmatic nucleus, has been considered to determine food-anticipatory behavior (20). Changes in the circadian profile of sympathetic activity may be one of the central mechanisms connecting food intake and metabolic outputs because food intake has been shown to be associated with enhanced sympathetic activity (25–28). In addition to central regulation, peripheral tissues also possess an oscillator that is synchronized with the central circadian system through retinal, hormonal, nutritional, and neuronal signals (29, 30). Recent advances in our understanding regarding the role of the peripheral oscillator have emphasized its importance in metabolic regulation (21). Furthermore, it has been well established that the circadian clock system in peripheral tissues is entrained by nutritional cues (21, 23). Taken together, these findings led us to hypothesize that phosphate metabolism was regulated by the circadian clock network through the modification of food intake-associated sympathetic activation, which may involve the action of the peripheral clock system.

In the current study, we tested our hypothesis that the circadian profile of circulating phosphate and FGF23 levels is determined by the time of nutrient availability by analyzing the circadian profile of skeletal Fgf23 expression in mice where the timing of food intake was restricted during the light phase and found that light phase-restricted feeding altered the circadian expression profile of skeletal Fgf23, which was in part caused by changes in the circadian profile of sympathetic activity. In addition, we demonstrated that stimulation with a β-adrenergic receptor agonist induced Fgf23 expression, which was suppressed by the overexpression of Cry1. These results underline the important roles of the circadian clock system in the regulation of phosphate metabolism.

EXPERIMENTAL PROCEDURES

Mice

C57BL/6J mice were purchased from CLEA Japan, Inc., and Bmal1 knock-out mice on a C57BL/6J background were generated as reported previously (31). Mice were maintained on a 12-h:12-h light dark cycle (lights on at zeitgeber time (ZT) 0) in a pathogen-free animal facility with free access to water and standard chow (CE-2; CLEA Japan, Inc.), unless otherwise mentioned. The light phase restricted feeding regimen was carried out by allowing mice access to food for 6 h between ZT2 and ZT8. A control diet containing 0.6% phosphate and 1.0% calcium and a high phosphate diet containing 1.65% phosphate and 1.0% calcium were purchased from CLEA Japan, Inc. All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee of Osaka Medical Center and Research Institute for Maternal and Child Health.

Reagents and Cell Lines

Isoproterenol hydrochloride, propranolol hydrochloride, and 3-isobutyl-1-methylxanthine (IBMX) were purchased from Wako Pure Chemical Industries Ltd. (Osaka, Japan). Human parathyroid hormone (PTH)(1–34) was obtained from the Peptide Institute, Inc. (Osaka, Japan). UMR-106 cells were obtained from ATCC (Manassas, VA) and maintained in DMEM supplemented with 10% fetal bovine serum and 1% insulin-transferrin-selenium-G supplement (Invitrogen). Cells were cultured at 37 °C in a 5% CO₂ atmosphere.

Real-time RT-PCR

Total RNA was prepared using TRIzol (Invitrogen) and treated with DNase I (Qiagen). cDNA was generated using a random hexamer and reverse transcriptase (Superscript II; Invitrogen) according to the manufacturer's instructions. The quantification of mRNA expression was carried out using a 7300 Real-time PCR system or a StepOnePlus^TM Real-time PCR system (Applied Biosystems). TaqMan Gene Expression Assays for Fgf23, Cryptochrome1, Rankl, Sost, Slc34a1, Slc34a3, Cyp27b1, Cyp24a1, and Gapdh were purchased from Applied Biosystems. Primer sequences for Rev-erbα, Dbp, and Bmal1 are described in Table 1. Gapdh was used as an internal standard control gene for all quantifications.

TABLE 1.

Primer sequences for real-time RT-PCR

Gene	Forward Primer	Reverse Primer
Rev-erbα	5′-cccaacgacaacaaccttttg-3′	5′-ccctggcgtagaccattcag-3′
Dbp	5′-caccgtggaggtgctaatga-3′	5′-gcttgacagggcgagatca-3′
Bmal1	5′-aggcccacagtcagattgaaa-3′	5′-ccaaagaagccaattcatcaatg-3′

Western Blot Analysis

To prepare whole cell lysates, cells were solubilized in radioimmuneprecipitation assay buffer (1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 150 mm NaCl, 10 mm Tris-Cl (pH 7.4), 5 mm EDTA, 1 mm orthovanadate, and protease inhibitor mixture (Complete^TM; Roche Diagnostics). Equal amounts of protein were separated by SDS-PAGE and transferred electrophoretically to PVDF membranes. Membranes were blocked in BlockAce reagent (Dainippon Pharmaceuticals, Osaka, Japan) or Blocking-one P reagent (Nacalai Tesque, Kyoto, Japan), immunoblotted with anti-CREB (1:1000, 9192; Cell Signaling, Beverly, MA), anti-pCREB (1:1000, 9191; Cell Signaling), anti-V5 (1:5000, 46-0705; Invitrogen), or anti-β-actin (1:2000, sc-47778; Santa Cruz Biotechnology) and developed with horseradish peroxidase-coupled secondary antibodies, followed by enhancement with a chemiluminescence (ECL) detection system (GE Healthcare).

Generation of Adenoviruses

Adenoviruses carrying GFP or Cry1 were constructed using the ViraPower Adenoviral Expression System (Invitrogen). Briefly, cDNA was inserted into a TOPO pENTR vector and was recombined to the adenovirus expression plasmid pAd/CMV/V5-DEST. The pAd/CMV/V5-DEST plasmid with the cDNA of interest was digested with the PacI endonuclease and transfected with HEK293A cells. The medium supernatant containing the adenovirus was collected and titrated according to the manufacturer's instructions. UMR-106 cells were infected with the adenovirus at a multiplicity of infection of 500 with 4 μg/ml poly-l-lysine (Sigma).

Constructs and Luciferase Reporter Assay

The Cry1 expression construct with a V5 tag was created by subcloning the corresponding PCR products into the pENTR vector using the pENTR Directional TOPO cloning kit (Invitrogen) and transferring to the pcDNA3.2/V5 vectors using the LR recombination reaction system (Invitrogen). Luciferase vectors containing 2000 bp of the mouse Fgf23 gene promoter (−1872 to +128) (2000bp-Luc) was prepared by subcloning the corresponding PCR products into pGL4.20[luc2/Puro] (Promega, Madison, WI) vectors according to the previous report (13). Luciferase constructs containing 766 bp (−638 to +128) of the Fgf23 gene promoter were created by the digestion of 2000bp-Luc with BglII (located at the 5′ region in the multiple cloning site of the pGL4.20 vector relative to the insert and −638 to −633) followed by the ligation of fragments containing luciferase with T4 DNA ligase. Luciferase constructs containing 143 bp (−15 to +128) of the Fgf23 gene promoter were designed by digesting pT7 vectors harboring 2000 bp of the Fgf23 gene promoter with SmaI (located at −18 to −13 and the 3′ region in the pT7 vector relative to the insert) followed by the ligation of fragments containing the promoter region of interest with pGL4.20[luc2/Puro] vectors digested with EcoRV using T4 DNA ligase.

Mutagenesis

Two candidate motifs were detected as possible cAMP responsive elements (CREs) in the promoter region of the Fgf23 gene by in silico analysis, and these motifs were designated as CRE1 and CRE2, respectively. To determine whether these motifs were functional, site-directed mutagenesis was performed using QuikChange II XL (Agilent Technologies, Santa Clara, CA) according to the manufacturer's protocol. The CRE1 located at −620 to −613 and CRE2 located at −46 to −39 were mutated from TGACCTCA to TGAAATCA and TGATGTCA to TGAAATCA, respectively.

Luciferase Assay

UMR-106 cells were seeded in a 24-well plate at a density of 5 × 10⁴ cells/well, and transient transfection was carried out using FuGENE HD (Promega) following the manufacturer's protocol. The total amount of DNA added to each well was equalized using an empty vector. The luciferase assay was performed in triplicate according to the protocol of the dual-luciferase reporter assay system (Promega). Briefly, 24 h after transfection cells were treated with isoproterenol (10 or 100 μm) and/or IBMX (0.5 mm) in DMEM containing 1% FCS overnight, followed by the determination of luciferase activity using specific substrates in a luminometer. Transfection efficiency was normalized by co-transfection with the TK-Renilla luciferase construct (Promega). While Cry1-overexpressing UMR-106 cells were used for the luciferase assay, UMR-106 cells were infected with an adenovirus containing GFP or Cry1-V5. Twenty-four h after the infection cells were trypsinized and plated in a 24-well plate as described above.

Animal Studies

Isoproterenol or PTH(1–34) was dissolved in saline and administered intraperitoneally at a dose of 6 μg/g (32) or 100 μg/kg, respectively. A saline injection was used as a control treatment. Whereas propranolol (PRO) was used for the in vivo study, PRO was dissolved in the drinking water at a concentration of 0.5 g/liter (33), and the drinking water was changed three times a week.

Measurement of Serum and Urine Parameters

The measurement of serum phosphate was carried out using P-test Wako (Wako Pure Chemical Industries Ltd., Osaka, Japan). Total (C-Term) and intact (full-length) FGF23 concentrations were determined by ELISA from Immutopics, Inc., San Clemente, CA and Kainos Laboratory, Tokyo, Japan, respectively, following the manufacturers' instructions. Urine samples were collected in the presence of 5 μl of 5 n HCl, and the volume of urine was measured. Urine epinephrine was determined using ELISA (IBL, Hamburg, Germany). Urine phosphate and creatinine were measured using P-test Wako and Creatinine Test Kit (Wako Pure Chemical Industries Ltd.), respectively.

Statistical Analysis

All data are expressed as the mean ± S.E. Results were examined for significant differences using Student's t test or analysis of variance followed by the Bonferroni multiple comparison post hoc test. Significance was set at p < 0.05.

RESULTS

Skeletal Fgf23 Exhibited a Circadian Expression Profile

To investigate the mechanisms whereby phosphate metabolism is regulated by the circadian clock network, we first examined the circadian expression profile of Fgf23 in the femur of wild-type mice fed standard chow ad libitum (AL). As reported previously (34), components of the clock network including Rev-erbα (nuclear receptor subfamily 1, group D, member 1; Nr1d1), Dbp (D site of albumin-binding protein), and Cry1 exhibited rhythmic expression patterns in the femur (Fig. 1, A–C, and supplemental Fig. S1). The Fgf23 expression profile showed higher expression levels during the dark phase (DP) compared with the light phase (LP) with the highest at ZT16 (Fig. 1D and supplemental Fig. S1). Because it is well known that food consumption reaches highest at the beginning of DP in mice fed AL, we speculated that skeletal Fgf23 expression was regulated by the food consumption in a manner involving the circadian clock system. To test this speculation, mice were fed during the LP from ZT2 to ZT8 (LP-restricted feeding: LP-RF) for 10 days. Because it is unclear whether skeletal tissue is entrained by nutrient availability despite the fact that food intake is a strong zeitgeber in peripheral tissues such as liver, we examined the circadian expression profiles of genes involved in daily oscillations in the femur. As shown in Fig. 2, A–C, and supplemental Fig. S2, the peak expressions of Rev-erbα, Dbp, and Cry1 shifted by 12 h in mice under LP-RF conditions compared with AL conditions, suggesting that skeletal tissue is also entrained by nutrient availability. Based on this observation, we next analyzed the expression profile of Fgf23 in the femur and found that Fgf23 showed a rhythmic expression pattern with a peak expression at ZT8 (Fig. 2D and supplemental Fig. S2). Taken together, these findings indicate that Fgf23 expression possesses a circadian expression profile that is at least in part determined by the time of nutrient availability.

FIGURE 1.

Fgf23 showed a rhythmic expression pattern in the femur of mice fed ad libitum. WT mice were maintained under a light-dark regimen (12-h:12-h cycle) and fed ad libitum. Samples were collected every 4 h from ZT0. A, expression of Rev-erbα (n = 3–4, *, p < 0.01 versus ZT0 and ZT20; **, p < 0.05 versus ZT20; ***, p < 0.05 versus ZT0, ZT16 and ZT20). B, expression of Dbp (n = 3, *, p < 0.001 versus ZT0, ZT4, and ZT20; **, p < 0.01 versus ZT16; ***, p < 0.05 versus ZT8; †, p < 0.05 versus ZT0). C, expression of Cry1 (n = 5–7, *, p < 0.05 versus ZT8). D, expression of Fgf23 (n = 5–7, *, p < 0.05 versus ZT8). All expression was in the femur and was analyzed using real-time RT-PCR. The white bar and black bar represent the light phase and dark phase, respectively. Values are expressed as the mean ± S.E. (error bars).

FIGURE 2.

Light phase restricted feeding altered the expression profile of Fgf23 in the femur. WT mice were maintained under a light-dark regimen (12-h:12-h cycle) and fed for 6 h from ZT2 to ZT8 for 10 days. Samples were collected every 4 h from ZT0. A, expression of Rev-erbα (n = 3, *, p < 0.05 versus ZT0 and ZT20; **, p < 0.05 versus ZT20). B, expression of Dbp (n = 3, *, p < 0.05 versus ZT0). C, expression of Cry1 (n = 3–4, *, p < 0.05 versus ZT0 and ZT20). D, expression of Fgf23 (n = 3, *, p < 0.05 versus ZT0). All expression was in the femur and was determined by real-time RT-PCR. The white bar and black bar represent the light phase and dark phase, respectively. Values are expressed as the mean ± S.E. (error bars).

Sympathetic Activation Enhanced Fgf23 Expression in the Femur

It is well known that food intake is tightly coupled to an increase in the metabolic rate to adjust for the increase in nutrient influx, which in part involves an elevation in sympathetic activity (25–28). Indeed, urine levels of epinephrine, a marker for sympathetic activity, in mice fed AL were significantly enhanced in DP compared with LP (Fig. 3A). Interestingly, LP-RF caused a phase shift in sympathetic activity with greater levels during LP-RF, but the difference did not reach statistical significance (Fig. 3B). Based on these findings, we speculated that Fgf23 expression may at least in part be regulated by sympathetic activation in a circadian manner. To test this idea, we intraperitoneally administered the β-adrenergic receptor agonist, isoproterenol (ISO), to mice at different time points of the day and analyzed the expression of Fgf23 in the femur 4 h after the injection. The administration of ISO caused an increase in skeletal Fgf23 expression when injected at ZT8 (p = 0.12) and ZT12 (p < 0.05), whereas ISO treatment had no effect on Fgf23 expression when administered at the other time points (Fig. 3C). To determine whether the effect of ISO was specific to Fgf23 induction, we also analyzed the expression of Rankl, one of the target genes of ISO (35), and found that the induction of Rankl showed a pattern similar to that of Fgf23 induction (Fig. 3C). To further understand the involvement of sympathetic activity in the circadian Fgf23 profile, mice were maintained under LP-RF in the presence of the β-blocker, PRO, to analyze the effect of sympathetic activity on the peak expression of Fgf23 noted at ZT8. Interestingly, Fgf23 expression did not exhibit any circadian profile when PRO was concomitantly administered, indicating the involvement of sympathetic activity in the circadian profile of Fgf23 expression in the femur (compare Fig. 3D with Fig. 2D, and see supplemental Fig. S2).

FIGURE 3.

Sympathetic activation induced skeletal Fgf23 expression in vivo. A, urine was collected from WT mice under ad libitum (AL) conditions either during the light phase (LP: ZT0–12) or the dark phase (ZT12–24). The volume of urine and urine epinephrine levels were measured, and the amount of urine epinephrine/h was determined (n = 9). B, urine was collected from WT mice under LP-RF conditions either during ZT2–8 or ZT8–26. The volume of urine and urine epinephrine levels were measured, and the amount of urine epinephrine/h was determined (n = 7). C, ISO was administered intraperitoneally to WT mice under AL conditions at different time points of the day as indicated, and 4 h after the injection the expression of Fgf23 and Rankl in the femur was determined by real-time RT-PCR (n = 3–4). D, WT mice were maintained under LP-RF conditions for 10 days, and 0.5 g/liter propranolol was added to the drinking water from day 7 to day 10. Femurs were collected at indicated time points, and Fgf23 expression was measured by real-time RT-PCR (n = 3). The white bar and black bar represent the light phase and dark phase, respectively. ns, not significantly different. Values are expressed as the mean ± S.E. (error bars). *, p < 0.001; **, p < 0.05.

ISO Trans-activated the Fgf23 Gene Promoter and Induced Fgf23 Expression in UMR-106 Cells

Because in vivo administration of ISO enhanced the expression of Fgf23 in the femur, we investigated whether ISO signaling trans-activated Fgf23 gene promoter using osteoblastic UMR-106 cells in which endogenous Fgf23 was expressed. For this purpose, we generated a luciferase construct containing a 2000-bp promoter region of the mouse Fgf23 gene. The treatment with ISO showed a significant increase in luciferase activity in a dose-dependent manner (Fig. 4A). In line with this, the ISO treatment enhanced the expression of Fgf23 in these cells (Fig. 4B). These results suggest that ISO induces Fgf23 expression at least in part by activating the transcription of the Fgf23 gene. Because in silico analysis pointed out the existence of two motifs whose sequences were very similar to CRE, we next tested whether these motifs, designated as CRE1 and CRE2, respectively, were involved in ISO-induced Fgf23 trans-activation. For this purpose, we generated luciferase vectors containing the truncated forms of the Fgf23 gene promoter and found that the luciferase vectors containing CRE1 and CRE2 were responsive to the ISO/IBMX treatment (Fig. 4C). To determine the responsible motif(s) for this trans-activation of the Fgf23 gene promoter, we introduced mutations in CRE1 and/or CRE2 and found that CRE2 was responsible for ISO/IBMX-induced activation of the Fgf23 gene promoter (Fig. 4D).

FIGURE 4.

ISO trans-activated Fgf23 transcription in UMR-106 cells. A, UMR-106 cells were seeded in 24-well plates and transfected with 2000bp-Luc (200 ng) and phRL-TK (10 ng). Twenty-four h after transfection, cells were treated with ISO at a dose of 10 or 100 μm overnight, and luciferase activity was measured. 0.5 mm IBMX was used as a positive control for the activation of cAMP signaling (n = 3). B, UMR-106 cells were treated with 100 μm ISO overnight, and expression of Fgf23 was determined by real-time RT-PCR (n = 4). C, UMR-106 cells were seeded in 24-well plates and transfected with 2000bp-Luc (200 ng), 766bp-Luc (200 ng), or 143bp-Luc (200 ng) and phRL-TK (10 ng). Twenty-four h after transfection, cells were treated with ISO (100 μm) and IBMX (0.5 mm) overnight, and luciferase activity was measured (n = 5). D, UMR-106 cells were seeded in 24-well plates and transfected with 766bp-Luc (200 ng) or 766bp-Luc containing mutations in CRE1 and/or CRE2 (200 ng), and phRL-TK (10 ng). Twenty-four h after transfection, cells were treated with ISO (100 μm) and IBMX (0.5 mm) overnight, and luciferase activity was measured (n = 4). Values are expressed as the mean ± S.E. (error bars). ns, not significantly different. *, p < 0.001; **, p < 0.01; †, p < 0.05.

Overexpression of Cry1 Blunted the Effects of ISO on Fgf23 Induction in UMR-106 Cells

These findings may support the concept that Fgf23 expression is regulated by sympathetic activity, but it is still unclear as to why Fgf23 induction by ISO is regulated in a time-dependent manner in vivo. To solve this issue, we assessed whether Cry1 was involved in the ISO-induced activation of Fgf23 expression because Cry1 has been implicated in the suppression of ISO-induced cAMP accumulation in HEK293 cells (36). Indeed, ISO-induced Fgf23 expression was evident when Cry1 expression was low in the femur (Figs. 1C and 3C). Based on these results, we tested our hypothesis that Cry1 suppressed ISO-induced Fgf23 induction by blocking CREB signaling in osteoblastic cells. To test this hypothesis, we overexpressed Cry1 in UMR-106 cells and investigated the effect of ISO on the phosphorylation of CREB. As shown in Fig. 5A, the ISO-induced phosphorylation of CREB was impaired in Cry1-overexpressing cells compared with GFP-expressing control cells. In line with this, ISO-induced trans-activation of the Fgf23 promoter was decreased in cells overexpressing Cry1 (Fig. 5B). To further determine the role of the circadian clock network in ISO-induced Fgf23 induction, we utilized Bmal1-deficient mice in which Cry1 expression was higher than that in WT littermate controls (Fig. 6A). The administration of ISO at ZT12 showed a significant increase in Fgf23 expression in the femur of WT mice, whereas the induction of Fgf23 was weaker in the femur of Bmal1-deficient mice (Fig. 6B).

FIGURE 5.

Cry1 suppressed ISO-induced phosphorylation of CREB in UMR-106 cells. A, UMR-106 cells were infected with an adenovirus containing GFP or Cry1-V5 and treated with 100 μm ISO for 5 min. The expression of pCREB, CREB, V5, and β-actin was determined by Western blotting, and expression of pCREB was quantified by normalizing to the levels of CREB by densitometric analysis (n = 5). B, UMR-106 cells infected with an adenovirus containing either GFP or Cry1 were seeded in 24-well plates and transfected with 2000bp-Luc (200 ng) and phRL-TK (10 ng). Twenty-four h after transfection cells were treated with 100 μm ISO overnight, and luciferase activity was measured (n = 3). The figures shown are the representative from at least three independent experiments. Values are expressed as the mean ± S.E. (error bars). *, p < 0.05.

FIGURE 6.

Fgf23 induction by the ISO treatment in the femur was impaired in Bmal1-deficient mice. A, the expression of Bmal1 and Cry1 was determined by real-time RT-PCR in the femur collected from WT mice and Bmal1-deficient mice at ZT16 (n = 3–5). B, WT mice and Bmal1-deficient mice were fed a control diet for 2 weeks from 8 weeks of age, and ISO was injected intraperitoneally at ZT12. Four h after the injection, the expression of Fgf23 in the femur was analyzed by real-time RT-PCR (n = 4–5). Values are expressed as the mean ± S.E. (error bars). *, p < 0.01; **, p < 0.05.

Parathyroid Hormone Induced Fgf23 Expression When Administered at ZT12

Because PTH has been shown to activate the CREB pathway and induce Fgf23 expression (4, 37), we next tested whether the PTH-induced activation of Fgf23 was also regulated in a circadian rhythm-dependent manner. To test this hypothesis, we administered PTH(1–34) intraperitoneally to WT mice at ZT0 or ZT12. Four h after the injection, the expression of Fgf23 was analyzed in the femur. The skeletal expression of Fgf23 exhibited a significant response to PTH when injected at ZT12, which was associated with a trend toward an increased expression of Rankl and a decreased expression of Sost, known to be regulated by PTH activation in osteoblastic cells (38), whereas PTH had no effect on Fgf23 expression when injected at ZT0 (Fig. 7). These results suggest the possibility of the circadian regulation of PTH action with respect to Fgf23 induction in the skeleton.

FIGURE 7.

PTH induced skeletal Fgf23 expression in a circadian fashion. PTH was intraperitoneally administered in WT mice at ZT0 or ZT12, and samples were collected 4 h after the injection. The expression of Fgf23, Rankl, and Sost in the femur was determined using real-time RT-PCR (n = 3–6). Values are expressed as the mean ± S.E. (error bars). *, p < 0.05.

Fgf23 Induction by Dietary Phosphate Load Was Not Likely Caused by Sympathetic Activation

Because it is still unclear as to whether circadian Fgf23 expression is regulated by food intake itself or the influx of phosphate from the diet, we finally tested whether sympathetic activity was involved in Fgf23 induction by dietary phosphate load. For this purpose, WT mice were fed either a control diet or high phosphate diet for 2 weeks in the presence or absence of PRO. Ingesting the high phosphate diet caused a significant elevation in Fgf23 expression in the femur (Fig. 8), but the concomitant administration of PRO did not affect the levels of Fgf23 expression in the femur (Fig. 8), which indicates that the timing of food intake, and not the amount of ingested phosphate, may determine the circadian profiles of skeletal Fgf23 expression.

FIGURE 8.

Propranolol did not affect skeletal Fgf23 expression induced by dietary phosphate load. WT mice were fed either control diet (CD) or high phosphate diet (HPD) for 2 weeks from 8 weeks of age in the presence or absence of PRO in the drinking water. The expression of Fgf23 in the femur was determined using real-time RT-PCR (n = 3–5). Values are expressed as the mean ± S.E. (error bars). *, p < 0.01.

The Increase in Skeletal Fgf23 Expression during DP Was Associated with Elevated FGF23 Levels in Serum and Enhanced Phosphate Excretion in the Urine

We finally investigated the association between the increase in skeletal Fgf23 expression during DP and systemic phosphate metabolism. Consistent with the rhythmic expression pattern of skeletal Fgf23, total FGF23 levels in the serum exhibited a circadian expression profile with peak levels at ZT 16 (Fig. 9A). Circulating biologically active (intact) FGF23 levels also showed greater levels during DP with the highest at ZT20 compared with ZT0 (Fig. 9B). In line with the increased serum FGF23 levels during DP, expression of Slc34a1 and Slc34a3 coding for NaP_i-IIa and NaP_i-IIc, respectively, showed decreased expressions during DP (Fig. 9, C and D, and supplemental Fig. S1), suggesting the enhanced phosphate excretion in the urine during this period. Indeed, serum phosphate concentrations showed a decline during DP associated with enhanced phosphate excretion in the urine (Fig. 9, E and F). We also analyzed the expression profile of Cyp27b1 and Cyp24a1, other target genes of FGF23 signaling in the kidney, and found that these genes exhibited circadian profiles (Fig. 10). The expression of Cyp27b1, which was down-regulated by FGF23 activation, was lower during the DP when FGF23 levels were greater, but the circadian profile of Cyp24a1, which was up-regulated by FGF23 activation, did not show any association with that of FGF23, suggesting that circadian profile of genes involved in vitamin D metabolism is mainly regulated by the circadian network independent of circadian FGF23 profiles.

FIGURE 9.

Increase in skeletal Fgf23 expression during the dark phase was associated with elevated circulating FGF23 levels and enhanced phosphate excretion in the urine. WT mice were maintained under a light-dark regimen (12-h:12-h cycle) and fed ad libitum. Samples were collected every 4 h from ZT0. A and B, serum concentrations of total (C terminus) FGF23 (n = 6–7, *, p < 0.05 versus ZT0 and ZT8) (A) and intact (full-length) FGF23 (n = 6–7, *, p < 0.05 versus ZT0) (B) were measured. C and D, the expression of slc34a1 (n = 6–7, *, p < 0.05 versus ZT4) (C) and slc34a3 (n = 6–7, *, p < 0.001 versus ZT8; **, p < 0.01 versus ZT4; ***, p < 0.05 versus ZT0; †, p < 0.05 versus ZT4 and ZT8) (D) in the kidney was determined by real-time RT-PCR. E, serum concentration of phosphate was measured (n = 7, *, p < 0.01 versus ZT0; **, p < 0.05 versus ZT0). F, urine was collected either during the light phase (ZT0–12) or dark phase (ZT12–24), and phosphate and creatinine (Cr) levels in the urine were measured (n = 9, *, p < 0.001). The white bar and black bar represent the light phase and dark phase, respectively. Values are expressed as the mean ± S.E. (error bars).

FIGURE 10.

Circadian expression profiles of genes involved in vitamin D metabolism in the kidney. WT mice were maintained under a light-dark regimen (12-h:12-h cycle) and fed ad libitum. Samples were collected every 4 h from ZT0. The expression of Cyp27b1 (n = 8–10, *, p < 0.01 versus ZT0 and ZT20; †, p < 0.01 versus ZT20; ††, p < 0.05 versus ZT0) and Cyp24a1 (n = 9–10, *, p < 0.001 versus ZT0; **, p < 0.01 versus ZT16 and ZT20; †, p < 0.01 versus ZT0; ††, p < 0.05 versus ZT0) in the kidney was determined by real-time RT-PCR. The white bar and black bar represent the light phase and dark phase, respectively. Values are expressed as the mean ± S.E. (error bars).

DISCUSSION

In the present study, we demonstrated that skeletal Fgf23 expression possessed a circadian expression profile. Importantly, the peak in skeletal Fgf23 expression shifted when mice were maintained under LP-RF regimen, suggesting that timing of food intake is an important determinant for rhythmic expression profile of skeletal Fgf23. To investigate the mechanism responsible for generating the rhythmicity in skeletal Fgf23 expression, we had a hypothesis that sympathetic activity may be the responsible factor linking food intake and skeletal Fgf23 expression based on the previous findings that sympathetic activation has been shown to be associated with food intake (28) and display a circadian profile with greater levels during the DP when food intake is active in mice (34). Indeed, we found that the administration of ISO caused an elevation in skeletal Fgf23 expression in a circadian rhythm-dependent manner. Furthermore, the blockade of sympathetic activity by PRO in mice under LP-RF conditions altered the circadian Fgf23 expression profile such that the peak expression of Fgf23 was not observed. Importantly, LP-RF caused a phase shift in sympathetic activity with greater levels during the LF. These results imply that sympathetic activation driven by food intake is a positive regulator for skeletal Fgf23 expression.

One of the important issues to be addressed is the effect of circadian regulation of skeletal Fgf23 expression on systemic phosphate metabolism. To address this issue we performed series of analyses and found that circulating both total and biologically active (intact) FGF23 levels were greater during DP. In line with the increase in FGF23 levels during DP, the expression of slc34a1 and slc34a3, known to be down-regulated by FGF23 activation in the kidney, was decreased during this period associated with enhanced phosphate excretion in the urine and decreased phosphate levels in the serum. Because food intake is increased at the beginning of DP in mice, these findings may suggest that the increase in skeletal Fgf23 expression during DP has an important role in handling the phosphate influx from the diet. It is important to note that the amplitude of circadian profile of intact FGF23 levels is not as obvious as that of total FGF23 levels, suggesting the possibility of the existence of additional mechanism regulating circadian profile of intact FGF23 in the circulation. Because FGF23 protein is known to be cleaved between Arg-179 and Ser-180 (39), it is possible that the post-translational modification or processing of FGF23 protein may create the difference in amplitude between total and intact FGF23 circadian rhythms, although the mechanisms of how FGF23 is cleaved in the circulation are not well defined and still need to be determined. Thus, these lines of evidence may imply that the circadian clock network may function in a coordinated manner involving multiple organs to maintain systemic phosphate homeostasis.

The present findings demonstrate that the timing of food intake regulates the circadian profile of skeletal Fgf23 expression, but it is still unclear as to whether circadian Fgf23 expression is regulated by the influx of phosphate from the diet. Because inorganic phosphate has been shown to induce Fgf23 expression in osteocyte-like IDG-SW3 cells (40), it is possible that the influx of phosphate from the diet may regulate circadian Fgf23 expression; however, previous in vivo studies failed to demonstrate an acute effect of dietary phosphate on FGF23 induction (41, 42), making this concept unlikely to be operative. Despite the lack of an increase in FGF23 levels in response to an acute phosphate load, it has been well established that a chronic phosphate load causes elevations in FGF23 levels (43). Consistent with previous reports, we detected a significant increase in skeletal Fgf23 expression in mice fed a high phosphate diet, but this increase was not associated with sympathetic activation because the concomitant administration of PRO did not affect Fgf23 expression. These findings may indicate that the timing of food intake, and not the amount of ingested phosphate, is a predominant determinant for the circadian profiles of skeletal Fgf23 expression.

The involvement of Cry1 in the regulation of G protein-coupled receptor signaling pathways has been demonstrated previously in a mouse model in which CRE activated luciferase activity in the liver (36). In these mice luciferase activity in the liver was markedly higher at ZT13 than at ZT1, which was associated with increased CREB phosphorylation, and was inhibited when Cry1 was overexpressed (36). The underlying mechanisms described in this study included the suppression of cAMP accumulation by binding of Cry1 to G_sα proteins (36). Based on these previous findings, we investigated whether a similar mechanism was operative in osteoblastic cells and found that overexpression of Cry1 suppressed ISO-induced CREB phosphorylation and trans-activation of the Fgf23 promoter in osteoblastic UMR-106 cells, which indicated that Cry1 also suppressed ISO-induced Fgf23 expression in osteoblastic cells. CREB has been shown to bind to the palindromic sequence (TGACGTCA) called CRE with strong affinity and regulate the transcription of target genes. In addition to this canonical CRE, CREB has also been shown to bind to CRE variants albeit with low affinity (44). The sequence of CRE detected in this study is one such variant that is known to mediate CREB signaling (44), and this may be one of the reasons why the induction of luciferase activity by ISO was not intense. Because in vivo analysis revealed a significant increase in Fgf23 expression by the ISO treatment, other CRE motif(s) may be present in other regions and affect the in vivo expression of Fgf23.

The regulation of Fgf23 expression in bone has also been implicated in the action of PTH. For example, circulating FGF23 levels were shown to be elevated under conditions in which PTH signaling was continuously active such as chronic kidney disease and Jansen metaphyseal chondrodysplasia caused by a mutation in the PTH1R gene, which suggests that PTH is a positive regulator for Fgf23 expression (37, 45). However, the negative action of PTH on Fgf23 expression has also been demonstrated in a mouse model in which PTH was intermittently administered (46). Thus, the effect of PTH on Fgf23 expression is context-specific and controversial (37, 45–47). Hence, the result showing that the induction of skeletal Fgf23 by a single injection of PTH was observed in a circadian manner may provide a clue to solve this controversial issue of PTH action on Fgf23 expression although further studies are needed to precisely determine PTH action on Fgf23 regulation.

In conclusion, in the present study we have provided evidence that the time of food intake determined the circadian profile of skeletal Fgf23 expression which involved a systemic activation of sympathetic tone and that sympathetic activation was peripherally regulated by Cry1 expression in the skeleton. Given the paucity of data as to the mechanisms regulating skeletal Fgf23 expression, these lines of evidence may shed light on new regulatory networks of FGF23 which could be important for understanding the physiology of phosphate metabolism.