Fibroblast growth factor 23 is upregulated in the kidney in a chronic kidney disease rat model

The hormone fibroblast growth factor 23 (FGF23) is secreted from bone and is involved in phosphorus (P) metabolism. FGF23 mainly binds the FGF receptor, which interacts with αKlotho in the kidney or parathyroid and regulates Na-dependent phosphate co-transporter type IIa (NaPi-IIa) and type IIc (NaPi-IIc) expression, 1,25-dihydroxyvitamin D3 (1,25(OH)2D3) activity, and parathyroid hormone (PTH) secretion. In this study, we utilized hemi-nephrectomized rats fed a high-P diet (HP Nx), rats subjected to a partial nephrectomy (PN) and rats with doxorubicin-induced renal failure (DXR) as chronic kidney disease (CKD) animal models and analyzed the P metabolism and FGF23 expression in the kidneys in each CKD model. We cultured HK2 cells with a high level of P, 1,25(OH)2D3 or transforming growth factor-β1 (TGFβ1) to investigate the FGF23 expression mechanism. In both the HP Nx and PN rats, the blood FGF23 and PTH levels were increased. However, the 1,25(OH)2D3 level was increased in the HP Nx rats and decreased in the PN rats. In all three animal models, the mRNA expression of αKlotho, NaPi-IIa and NaPi-IIc was decreased, and the mRNA expression of TGFβ1, collagen1a1, osteopontin and FGF23 was elevated in the kidney. FGF23 protein and mRNA were expressed at high levels in the extended tubule epithelium, which was an osteopontin-positive region in the HP and PN rats. FGF23 and osteopontin mRNAs were expressed in HK2 cells incubated with TGFβ1; however, these levels were not altered in HK2 cells incubated with 1,25(OH)2D3 and high P levels in vitro. Altogether, FGF23 is expressed in the kidneys in CKD model rats. Following stimulation with TGFβ1, the injured renal tubular epithelial cells are strongly suspected to express both FGF23 and osteopontin. FGF23 produced in the kidney might contribute to P metabolism in subjects with CKD.


Introduction
Fibroblast growth factor 23 (FGF23) is considered a causative gene in tumor-induced osteomalacia (TIO), including low-phosphorus (P) rickets/osteomalacia with P reabsorption failure in the renal tubules and autosomal dominant hypophosphatemic rickets/osteomalacia (ADHR) [1,2]. FGF23-transgenic mice have symptoms similar to those of hypophosphatemic rickets [3]. In addition, FGF23-deficient mice exhibit a failure to grow, bone lesions, a short lifespan, hyperphosphatemia, and high blood vitamin D levels, suggesting that FGF23 is a hormone regulator of P metabolism [4].
The phenotypes of FGF23-deficient mice and αKlotho-deficient mice are similar, and αKlotho is required for FGF23 to bind the fibroblast growth factor receptor (FGFR) [5]. In addition to FGFR, FGF can bind heparin and heparan sulfate (HS). Since HS is abundant in the extracellular matrix, secreted FGF does not diffuse and functions as a paracrine and autocrine factor [6]. In contrast, endocrine FGF isoforms, such as FGF19 (FGF15 in mouse), FGF21 and FGF23, have extremely low affinities to HS and diffuse into the blood. αKlotho has a very high affinity to FGF23 and can substitute for HS when FGF23 binds FGFR. Since αKlotho displays organ-specific expression, e.g., in the kidney and parathyroid, FGF23 functions in a tissue-specific manner [6].
αKlotho expression is decreased in subjects with renal failure, such as renal ischemia [7,8], fibrosis [9], and chronic kidney disease (CKD) [10][11][12]. In contrast, the blood FGF23 level is elevated beginning in the early stage of CKD and further increases as renal function deteriorates [13]. A high concentration of FGF23 is related to a worsening of the prognosis and deterioration of left ventricular hypertrophy in patients with CKD [14][15][16]. These effects are considered FGF23 functions that are independent of αKlotho. The calcineurin-nuclear factor of activated T cells signaling pathway is involved in FGF23-induced left ventricular hypertrophy in an animal model [15].
FGF23 is produced in the bone of patients with CKD [3]. The FGF23 levels increase with P loading, elevated 1,25-dihydroxyvitamin D 3 (1,25(OH) 2 D 3 ) levels, reduced αKlotho expression and reduced parathyroid hormone (PTH) levels [17][18][19][20]. In addition, a phosphate-binder treatment decreases blood FGF23 concentrations in rats with CKD [21]. In this study, we determined the site and mechanism of FGF23 production in kidney tissues using three rat models of CKD and renal proximal tubular epithelial cells incubated with TGFβ1, 1,25 (OH) 2 D 3 or a high level of P.

Experimental animal models
Sprague-Dawley rats and Wistar rats were purchased from CLEA Japan, Inc., Tokyo, Japan. The animals were provided free access to standard food and water and received care in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, Guidelines for Proper Conduct of Animal Experiments of the Science Council of Japan and Tokyo Women's Medical University Animal Experiment Regulations. The protocol was approved by the Committee on the Ethics of Animal Experiments of Tokyo Women's Medical University (Permit Number: AE16-66). All animals were anesthetized with isoflurane (induction 4-5%, maintenance 2-3%) for animals (Intervet, Tokyo, Japan) prior to surgery, and all efforts were exerted to minimize suffering. The urine collection, water intake and food intake were managed using a metabolic cage. The rats were subcutaneously injected with 0.01 mg/kg subcutaneous buprenorphine (Otsuka Pharmaceutical Co., Ltd., Tokyo, Japan) for analgesia immediately before the operation. Then, we performed an additional subcutaneous injection of buprenorphine after 6 and 24 hours if the animals did not move, drink or eat. We referred to the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines. The animals were housed in metabolic cages for 96 hours, and the first 24 hours were used for acclimatization. Urine samples collected within the first 24 hours were discarded. Then, we collected and measured the urine samples, body weight, food consumption and amount of water drank by each rat every 24 hours. We used the average values obtained over three days. The animals were sacrificed under deep anesthesia with isoflurane.
Hemi-nephrectomized rats fed a high-P diet. Sprague-Dawley rats were used for this experiment. The food for the rats was purchased from ORIENTAL YEAST CO., LTD. (Tokyo, Japan). In this experiment, the high-phosphate (HP) diet contained 2.0% P and 0.6% calcium (Ca), whereas the non-high-phosphate (NP) diet contained 0.35% P and 0.6% Ca.
We performed a right-side nephrectomy (Nx) or sham operation on 7-week-old rats at week -2. After two weeks (9 weeks of age, week 0), the animals began the NP or HP diet. Seven weeks after the diet change (16 weeks of age, week 7), all animals were sacrificed and their blood, left kidney, calvaria and liver specimens were collected. The rats were divided into the following four groups: NP sham group (n = 8), NP Nx group (n = 7), HP sham group (n = 7) and HP Nx group (n = 9). Partial nephrectomy rat model. Wistar rats were used in this experiment. Forty percent of the left kidney was surgically removed from 6-week-old rats in the mild partial nephrectomy group, and 60% of the left kidney was removed from the rats in the severe partial nephrectomy group. One week after the operation (7 weeks of age), nephrectomy of the right kidney or a sham operation was performed at SANKYO LABO SERVICE CORPORATION, INC. (Tokyo Japan). The rats were fed a standard diet, which contained 0.83% P and 1.07% Ca and was purchased from ORIENTAL YEAST CO., LTD. Nine weeks after the operation (16 weeks of age, week 9), all animals were sacrificed, and their blood, left kidney and liver specimens were collected. The rats were divided into the following three groups: sham group (n = 6), mild partial nephrectomy group (PN mild) (n = 6), and severe partial nephrectomy group (PN severe) (n = 6).
Rat model of doxorubicin-induced renal failure. In this experiment, we used frozen serum, frozen kidneys and kidney sections embedded in paraffin as described in our previous report [22]. The doxorubicin (DXR) used in this study was provided by Kyowa Hakko Kirin (Tokyo, Japan). The DXR-induced renal failure model was produced using a previously reported procedure [23]. Anesthetized rats were initially injected with 3 mg/kg of DXR via the tail vein, and another 2 mg/kg of DXR were injected after two weeks. The rats were divided into the vehicle group (n = 5) and DXR group (n = 5). Sixteen weeks after the initial DXR administration, all animals were sacrificed, and their blood and both right and left kidney specimens were collected.

Measurement of biochemical parameters
The blood and urine samples were centrifuged at 1,300 ×g for 10 minutes. The serum was separated from the blood sample, and the urine supernatant was used. The creatinine (Cr), Ca and P levels were analyzed using an auto-analyzer at MONOLIS Co., Ltd. (Tokyo, Japan).
(MONOLIS Co., Ltd.) according to the manufacturer's instructions. The serum levels of intact PTH were measured using the Intact PTH ELISA kit (Immutopics, Inc., San Clemente, CA, USA) according to the manufacturer's protocol.

RNA isolation, reverse transcription, and real-time PCR
The total RNA was extracted from the kidneys, calvarias, livers and HK2 cells using TRIzol 1 Reagent (Thermo Fisher Scientific) and an RNeasy Plus Mini Kit (QIAGEN, Tokyo, Japan). The cDNA templates were prepared using 2 μg of total RNA and a High-Capacity RNA-to-cDNA Kit (Thermo Fisher Scientific) and subjected to real-time PCR using a StepOnePlus™ Real-Time PCR System (Thermo Fisher Scientific). The primers used for the real-time PCR are listed in Table 1. The results were normalized to the cyclophilin level as previously described [9].

Western blot analysis
The frozen kidney samples were homogenized in lysis buffer (20 mM Na-HEPES [pH 7.5], 100 mM NaCl, 1% Triton X-100, 15 mM NaF, 1 mM Na 3 VO 4 , 10 mM Na 4 P 2 O 7 , 1 mM EDTA, Protease Inhibitor Cocktail ([Roche, Tokyo, Japan], and Phosphatase Inhibitor Cocktail 2 [Sigma-Aldrich]) on ice and centrifuged at 5,591×g for 10 minutes to remove debris. The protein concentrations in the supernatants were determined using 5 μg/μL. Ten micrograms of protein from each specimen were suspended in loading buffer, separated on a 4-15% Crite-rion™ TGX™ Precast Gel (Bio-Rad, Tokyo, Japan), and electrophoretically transferred to a nitrocellulose membrane. The membrane was blocked with 5% skim milk in TBS and incubated with a 500-fold dilution of the primary antibodies overnight at 4˚C. After three 5-minute washes with washing buffer (0.1% Tween 20 in TBS), the membranes were incubated with the secondary antibodies for 1 hour at room temperature. After three 10-minute washes with washing buffer, the membrane was incubated with Enhanced Chemiluminescence (ECL) Western Blotting Detection Reagents (GE Healthcare, Tokyo, Japan) and exposed using Lumi-noGraph (ATTO, Tokyo, Japan) and ImageSaver6 (ATTO). GAPDH was used as an internal control.

Immunohistochemistry
The kidneys were fixed with 4% paraformaldehyde in PBS for 16 hours at 4˚C and then embedded in paraffin. The paraffin-embedded sections (3-μm-thick) were deparaffinized, hydrated and then incubated with 0.3% H 2 O 2 in methanol for 30 minutes. A VECTASTAIN ABC Standard Kit (Vector Laboratories, Tokyo, Japan) was used for the immunohistochemical staining. The sections were blocked with normal blocking serum for 30 minutes and then incubated with a 500-fold dilution of the primary antibodies for 16 hours. After the incubation with the appropriate biotinylated goat anti-rabbit IgG secondary antibody for 60 minutes, the sections were stained with Vectastain ABC Reagent for 30 minutes, followed by the peroxidase substrate solution using the DAB substrate kit (Vector Laboratories).

Immunofluorescence staining
The kidneys were fixed with 4% paraformaldehyde in PBS for 16 hours at 4˚C, sequentially immersed in 10%, 15% and 20% sucrose in PBS for 12 hours at 4˚C, embedded in optimum cutting temperature (OCT) compound and immediately frozen in liquid nitrogen.
The immunofluorescence staining was performed using a primary antibody and a secondary antibody (Alexa Fluor 1 488-conjugated goat anti-rabbit IgG (H+L) secondary antibody or Alexa Fluor 1 568-conjugated goat anti-mouse IgG (H+L) secondary antibody, [Thermo Fisher Scientific]). The sections were blocked for 1 hour and incubated with a 500-fold dilution of the primary antibody for 16 hours at 4˚C. After the incubation with the secondary antibody for 1 hour at room temperature in the dark, the sections were incubated with DAPI (4',6-diamidino-2-phenylindole, dihydrochloride) (Thermo Fisher Scientific) and mounted using VECTASHIELD Mounting Medium (Vector Laboratories). The images of the sections were captured using Axio Observer Z1 and Axio Vision (ZEISS, Tokyo, Japan).

In situ hybridization
In situ hybridization was performed as previously described [26,27]. The kidneys were fixed with 4% paraformaldehyde in PBS for 16 hours at 4˚C and then embedded in paraffin. RNAscope 1 2.5 HD Reagent kit-RED (#322350, Advanced Cell Diagnostics, Hayward, CA, USA) was used for the in situ hybridization. The paraffin-embedded sections (5-μm-thick) were treated according to the manufacturer's instructions.

Statistical analysis
The data are presented as the means ± standard errors of the means (SEM). One-way ANOVA, two-way ANOVA, Kruskal-Wallis tests, unpaired t-tests with Bonferroni correction and Mann-Whitney U-tests with Bonferroni correction were performed using Excel (Microsoft WA, USA) and JMP for Windows (SAS Institute, NC, USA). P-values <0.05 represent statistically significant differences.

Hemi-nephrectomized rats fed a high-P diet
Renal function and mineral metabolism. Regarding renal function, the serum creatinine (Cr) levels did not significantly differ between the HP sham group and NP sham group. The 24-hour Cr clearance (24-hr CCr) at week 6 was significantly lower in the HP sham group than that in the NP sham group. The serum Cr level was significantly higher in the HP Nx group than that in the NP Nx group. The 24-hr CCr was also significantly lower in the HP Nx group than that in the NP Nx group (Fig 1a and 1b).
The serum P levels detected in the HP Nx group were higher than those detected in the NP Nx group at all time points throughout the experiment. The serum P level increased at weeks 2 and 4 after the animals in the HP sham group started the high-P diet and then gradual decreased. At week 6, no significant differences were observed compared with the animals fed the NP diet. In contrast, the 24-hour urine P excretion (24-hr U-P) was higher in the HP groups than that in the NP groups. However, a significant difference was not observed between the HP sham group and HP Nx group. The serum Ca level was lower in the HP groups than that in the NP groups at all time points throughout the experiment. The 24-hour urine Ca  (Fig 1c-1f). The serum 1,25(OH) 2 D 3 levels detected in the HP groups were higher than those detected in the NP groups. The serum level of intact PTH was higher in the HP groups than that in the NP groups (Fig 1g and 1h). The NaPi-IIa and NaPi-IIc mRNAs were measured because FGF23-αKlotho signaling suppresses NaPi-IIa and NaPi-IIc expression [6]. Compared with the NP groups, the αKlotho, NaPi-IIa and NaPi-IIc mRNAs were expressed at lower levels in the kidneys in the HP groups. In addition, the mRNA expression of cytochrome P450 family 27 subfamily B member 1 (CYP27B1), which encodes 25-hydroxyvitamin D3 1-alpha-hydroxylase (1α-hydroxylase), showed changes that were similar to the changes in the 1,25(OH) 2 D 3 levels (Fig 1i-1l). Transforming growth factor-β1 (TGFβ1), collagen1a1 (Col1a1), osteopontin and osteocalcin are markers of fibrosis and calcification. The TGFβ1, Col1a1, osteopontin and osteocalcin mRNAs expressed in the kidneys in the HP groups were higher than those expressed in the kidneys in the NP groups (Fig 1m-1p).
FGF23 expression. The serum FGF23 levels detected in the HP groups were higher than those detected in the NP groups. The mRNA expression of FGF23 was measured in the calvaria. FGF23 mRNA in the bone specimens from the HP groups was expressed at higher levels than that in the NP groups (Fig 2a and 2b). The mRNA expression of FGF23 was also measured in the kidney. Although FGF23 mRNA was barely detectable in the kidneys from the NP groups, it was detectable in the kidneys from the HP groups. The difference between the two groups was statistically significant. We did not observe any significant differences between the HP sham group and the HP Nx group (Fig 2c). The expression of FGF23 mRNA was not detected in the liver specimens from each group by real-time PCR (S3a Fig). Therefore, the FGF23 protein levels in the kidney were measured by performing Western blotting. The levels of FGF23 protein were elevated in the kidneys in the HP sham group and HP Nx group (Fig 2d).
Renal histology. In the HP groups, we observed interstitial fibrosis in the outer medulla, calcification of the tubule lumen, expansion of the renal tubule lumen and a decreased number of tubule cells (Fig 3, HE, MT and VK). The immunohistochemical staining of FGF23 was detected in the expanded tubular cells in the HP groups. In the NP sham group, we also detected weak positive signals in several glomerular and tubular cells. The area of FGF23 staining was greater in the HP groups (Fig 4, FGF23). Osteopontin is a biomarker of bone matrix. Osteopontin was detected in the expanded tubular cells expressing FGF23 in serial sections of the kidney (Fig 4, osteopontin). αSMA is a biomarker of interstitial fibrosis, tubule cells undergoing the epithelial-mesenchymal transition (EMT) and vascular smooth muscle cells. We performed double immunofluorescence staining of αSMA and FGF23. In the HP groups, the FGF23 and αSMA signals were not merged in the expanded tubules (Fig 5, FGF23, αSMA and merge). According to the FGF23 in situ hybridization, FGF23 mRNA was detected in the expanded tubular cells in the HP groups. We did not detect positive signals in the NP sham group (Fig 4, FGF23 in situ hybridization).

HK2 cells treated with TGFβ1, 1,25(OH) 2 D 3 or a high level of P
In the HP diet-fed rats, the renal FGF23 expression was increased under special circumstances, such as high P loading, high serum 1,25(OH) 2  the renal tubules. P loading and elevated 1,25(OH) 2 D 3 levels have been reported to increase the FGF23 levels. Subsequently, we cultured HK2 cells with high levels of P, 1,25(OH) 2 D 3 or TGFβ1 in vitro to investigate the mechanism underlying FGF23 expression in the renal tubules.
A TGFβ1 stimulation for 6 and 24 hours (short-term stimulation) tended to reduce the FGF23 mRNA expression, but a TGFβ1 stimulation for 72 and 144 hours (long-term stimulation) significantly induced FGF23 mRNA expression in the HK2 cells (Fig 6a). The osteopontin mRNA expression was also decreased in the cells subjected to the short-term TGFβ1 treatment but increased in the cells subjected to the long-term TGFβ1 treatment (Fig 6b). The E-cadherin mRNA expression was decreased by TGFβ1 (Fig 6c). The PAI1 mRNA expression was increased in response to the short-term TGFβ1 stimulation but decreased in response to the long-term TGFβ1 stimulation (Fig 6d).
For the 1,25(OH) 2 D 3 stimulation, the HK2 cells were challenged with 0.4 μM 1,25(OH) 2 D 3 for 6, 24, 72 and 144 hours. For the high P stimulation, the HK2 cells were challenged with 3 mM Na 2 HPO 4 /NaH 2 PO 4 (P) and 2.5 mM β-glycerophosphate medium for 1, 7 and 21 days. The 1,25(OH) 2 D 3 and P stimulations did not change the FGF23 mRNA expression levels (Fig  6e and 6f). Therefore, we used a partial nephrectomy rat model, which usually displays high TGFβ1 levels, normal serum P and 1,25(OH) 2 D 3 levels, and a lack of calcification during the early and middle stages of CKD, and examined the FGF23 expression in the kidney.
Regarding renal function, the serum Cr level was significantly higher in the PN mild and severe groups than that in the sham group. At week 9, the 24-hr CCr was also significantly lower in the PN mild and severe groups than that in the sham group (Fig 7a and 4b).
Although the serum P levels in the PN mild and severe groups did not differ from those in the sham group, the 24-hr U-P was higher in the PN mild and severe groups than that in the sham group at week 9. The serum Ca levels in the PN mild and severe groups did not differ from those in the sham group. At week 9, the 24-hr U-Ca in the PN mild and severe groups did not significantly differ from that in the sham group (Fig 7c-7f). FGF23 production in the kidney is upregulated in a rat CKD model Although the serum 1,25(OH) 2 D 3 level was lower in the PN mild group than that in the sham group at week 9, the difference was not significant. However, the serum 1,25(OH) 2 D 3 level was significantly lower in the PN severe group than that in the sham group. The serum level of intact PTH was significantly higher in the PN mild and severe groups than that in the sham group (Fig 7g and 7h).
The αKlotho, NaPi-IIa and NaPi-IIc mRNAs were expressed at significantly lower levels in the kidneys of the PN severe group than those in the kidneys of the sham group. In addition, the expression of CYP27B1 mRNA was significantly higher in the PN mild and severe groups than that in the sham group. The CYP27B1 mRNA expression level did not correlate with the serum 1,25(OH) 2 D 3 level (Fig 7i-7l). The levels of TGFβ1, Col1a1, osteopontin and osteocalcin mRNAs found in the kidneys of the PN severe group were higher than those found in the kidneys of the sham group (Fig 7m-7p).
FGF23 expression in the kidney. The serum FGF23 levels detected in the PN mild and severe groups were higher than those detected in the sham group (Fig 8a). Although the expression of FGF23 mRNA was barely detectable in the kidneys of the sham group, it was detected in both the PN mild and severe groups, and the levels were significantly higher than those in the sham group (Fig 8b). The expression of FGF23 mRNA was not detectable in the livers of each group by real-time PCR (S3b Fig). The levels of FGF23 protein in the kidney were measured by Western blotting. The levels of FGF23 protein were elevated in the kidneys of the PN mild and severe groups (Fig 8c).
Renal histology. In the PN mild and severe groups, we observed interstitial fibrosis, expansion of the renal tubule lumen and a decreased number of tubule cells (Fig 8d, HE and MT). Calcification was not observed by VK staining (Fig 8d, VK). Through immunohistochemical staining, FGF23 and osteopontin were detected in the expanded tubular cells (Fig 9, FGF23 and osteopontin). According to the FGF23 in situ hybridization, FGF23 mRNA was detected in the expanded tubular cells in the PN severe and mild groups (Fig 9, FGF23 in situ hybridization).

Rat model of DXR-induced renal failure
We conducted experiments in a rat model of DXR-induced renal failure as a third model of CKD to investigate whether FGF23 expression is a common phenomenon in renal tubulointerstitial disorder.
Regarding renal function, the serum Cr level was significantly higher in the DXR group than that in the vehicle group. The serum P level was significantly higher in the DXR group than that in the vehicle group. The serum Ca levels did not differ between the DXR group and vehicle group (Fig 10a-10c; these figures were reproduced from our previous report [22]). The αKlotho, NaPi-IIa and NaPi-IIc mRNAs were expressed as significantly lower levels in the kidneys of the DXR group than in the kidneys of the vehicle group (Fig 10d-10f). The expression of CYP27B1 mRNA was significantly decreased in the DXR group (Fig 10g). In the DXR group, we observed interstitial fibrosis in the tubule lumen, an expansion of the renal tubule lumen and a decreased number of tubule cells (Fig 10i, MT).
The FGF23 mRNA expression was measured in the kidney and was significantly higher in the DXR group than that in the vehicle group (Fig 10h). The TGFβ1, Col1a1, osteopontin and osteocalcin mRNAs were expressed at higher levels in the kidneys of the DXR group than in the kidneys of the vehicle group (Fig 10j-10l).

Discussion
This study showed that FGF23 is produced in the kidneys of CKD model animals. The main site of FGF23 production has been reported to be the bone [3], and osteoblasts secrete FGF23  [6]. In addition to bone, FGF23 is also produced in the brain [28] in healthy subjects. Under pathological conditions, the FGF23 production sites include the kidney in rodent models of polycystic kidney disease (PKD) [29] and a Zucker diabetic fatty (ZDF) rat model of human type 2 diabetic nephropathy [30] and 5/6 nephrectomy [26] and the liver in diethyl-nitrosamine (DEN)-treated mice [31], patients on the Liver-Transplant Waiting List [31] at the time of liver failure, and vascular smooth muscle cells cultured in calcifying medium [24]. FGF23 protein has been detected in cells lining renal cysts in rodent models of PKD [29]. FGF23 expression has been speculated to be unrelated to serum and urinary P. Moreover, FGF23 produced in the kidneys in a PKD model and subjects with liver failure originates from the increased FGF23 concentration in the blood [31].
Vascular smooth muscle cells have been reported to transform and differentiate into osteoblast-like cells in response to high-P medium or CKD [10,24,32]. In this study, we assumed that vascular smooth muscle cells in the kidney secrete and express FGF23. However, according to the double staining of FGF23 and αSMA, αSMA-positive vascular smooth muscle cells FGF23 production in the kidney is upregulated in a rat CKD model do not express FGF23. The positive FGF23 and αSMA staining did not merge in these tissues. αSMA is an EMT marker. The osteopontin, osteocalcin, Col1a1 (bone matrix markers) and TGFβ1 (a fibrosis marker) mRNAs were expressed at higher levels in the kidneys from the CKD rats than those in the control kidneys. Thus, the renal tubular epithelial cells transformed into stromal-like cells that express FGF23. In particular, strong FGF23 staining was detected in the extended tubules in the kidneys of HP diet-fed rats, which showed positive Von Kossa staining. Since the kidney is a target organ of FGF23, positive signals were detected in the FGF23 production in the kidney is upregulated in a rat CKD model kidney through immunostaining. However, the FGF23 mRNA expression was increased, and a positive signal of the FGF23 mRNA was detected in the renal tubular epithelial cells that had attenuated their original function, suggesting that FGF23 is produced in these cells. In addition, focal FGF23 staining was previously observed in both proximal and distal tubular sections at a more advanced phase of the disease in a Zucker diabetic fatty (ZDF) rat diabetic nephropathy model [30]. The expression of FGF23 mRNA was found of be increased in the following three rat models of CKD used in this study: HP diet-induced renal failure, partial nephrectomy, and DXR-induced renal failure. Based on these findings, renal expression of FGF23 mRNA is a common phenomenon in CKD.
In the present study, Western blotting was performed and detected the levels of FGF23 protein, and an immunohistochemical study showed that FGF23 was located in the glomeruli and tubular cells in the NP sham group. In contrast, the FGF23 mRNA expression was barely detectable by real-time PCR and in situ hybridization in the same group. Therefore, FGF23 proteins located in the glomerular capillary and tubular cells might originate from bone in subjects with normal kidney function. Since the kidney is a target organ of FGF23, FGF23 from bone might exist and bind the FGFR with αKlotho in the normal kidney.
We cannot easily explain the induction of FGF23 expression in the kidneys of the rats with CKD by the high serum P and 1,25(OH) 2 D 3 levels in this study because the serum P and 1,25 (OH) 2 D 3 levels in the hemi-nephrectomized rats fed the HP diet were high, but normal and low levels were observed in the partial nephrectomy rat model, and HK2 cells stimulated with high-P or 1,25(OH) 2 D 3 did not express FGF23 mRNA. In all three animal CKD models, the expression of TGFβ1 mRNA and FGF23 was elevated in the kidney. The FGF23 mRNA was expressed in HK2 cells incubated with TGFβ1 in vitro. These results indicate that renal tubular epithelial cells are highly likely to express FGF23 following stimulation with TGFβ1.
In CKD, the blood FGF23 levels increase beginning at the early stage. FGF23 expression is increased by P loading, increased 1,25(OH) 2 D 3 levels, decreased αKlotho expression, and reduced PTH levels [17][18][19][20]. In addition, αKlotho expression decreases during the early stage of CKD. The mechanism includes not only direct injury to the renal tubule cells that express αKlotho but also circulatory disorders, diabetes, uremic toxins, and oxidative stress [6,8,36]. A reduction in αKlotho expression induces FGF23 resistance, which increases the blood FGF23 level. The administration of an FGF23 neutralizing antibody in a rat model of early stage CKD increases the blood P concentration, reduces the fractional P excretion, and increases the 1,25 (OH) 2 D 3 levels [37]. Based on these findings, the elevation in the FGF23 levels during the early stage of CKD prevents the elevation of blood P levels. One possible mechanism is that the elevated FGF23 levels bind to reduced levels of αKlotho-FGFR, which reduces NaPi-IIa and NaPi-IIc expression. Consequently, urinary P excretion is promoted by suppressing P resorption in the tubules [3,33], thus attenuating the elevated blood P concentration in subjects with CKD.
In the present study, the renal expression of αKlotho, NaPi-IIa and NaPi-IIc mRNAs was reduced, the serum FGF23 and PTH levels were elevated, and FGF23 expression was increased in the calvaria and kidneys of the rat models of mild and severe CKD. The 24-hr U-P was higher in the HP groups than that in the NP groups and tended to be higher in the PN mild and severe groups than that in the sham group. FGF23 expressed in the kidney might play a role in the control of blood and urine P levels and acts on αKlotho and FGFR as a paracrine and autocrine factor to directly affect P resorption in the kidneys of CKD rats.
In the partial nephrectomy rat model, the mRNA expression of CYP27B1, which encodes 1α-hydroxylase, was increased, whereas the serum 1,25(OH) 2 D 3 concentration was decreased. Thus, we detected a dissociation between CYP27B1 expression and the 1,25(OH) 2 D 3 levels. This dissociation might have been caused by the reduced kidney volume in the PN groups. Unexpectedly, during the early stage of HP diet-induced CKD, the 1,25(OH) 2 D 3 concentration was increased, and the expression of the CYP27B1 mRNA was elevated in the partial nephrectomy rats fed the HP diet. Since this finding cannot be explained by FGF23 alone, a complex interaction likely exists among Ca, P and PTH in the blood and other molecules involved in mineral metabolism.

Conclusions
In CKD rat models, FGF23 mRNA is expressed in the kidney, and the FGF23 protein is expressed at high levels in osteopontin-positive renal tubule epithelium cells likely via TGFβ1 stimulation. FGF23 produced in the kidney might contribute to P metabolism in subjects with CKD.