Impact of Altered Mineral Metabolism on Pathological Cardiac Remodeling in Elevated Fibroblast Growth Factor 23

Clinical and experimental studies indicate a possible link between high serum levels of fibroblast growth factor 23 (FGF23), phosphate, and parathyroid hormone (PTH), deficiency of active vitamin D (1,25D) and klotho with the development of pathological cardiac remodeling, i.e., left ventricular hypertrophy and myocardial fibrosis, but a causal link has not been established so far. Here, we investigated the cardiac phenotype in klotho hypomorphic (kl/kl) mice and Hyp mice, two mouse models of elevated FGF23 levels and klotho deficiency, but differing in parameters of mineral metabolism, by using histology, quantitative real-time PCR, immunoblot analysis, and serum and urine biochemistry. Additionally, the specific impact of calcium, phosphate, PTH, and 1,25D on hypertrophic growth of isolated neonatal rat cardiac myocytes was investigated in vitro. Kl/kl mice displayed high serum Fgf23 levels, increased relative heart weight, enhanced cross-sectional area of individual cardiac myocytes, activated cardiac Fgf23/Fgf receptor (Fgfr) 4/calcineurin/nuclear factor of activated T cell (NFAT) signaling, and induction of pro-hypertrophic NFAT target genes including Rcan1, bMHC, brain natriuretic peptide (BNP), and atrial natriuretic peptide (ANP) as compared to corresponding wild-type (WT) mice. Investigation of fibrosis-related molecules characteristic for pathological cardiac remodeling processes demonstrated ERK1/2 activation and enhanced expression of Tgf-β1, collagen I, and Mmp2 in kl/kl mice than in WT mice. In contrast, despite significantly elevation of serum and cardiac Fgf23, and reduced renal klotho expression, Hyp mice showed no signs of pathological cardiac remodeling. Kl/kl mice showed enhanced serum calcium and phosphate levels, while Hyp mice showed unchanged serum calcium levels, lower serum phosphate, and elevated serum iPTH concentrations compared to corresponding WT mice. In cultured cardiac myocytes, treatment with both calcium or phosphate significantly upregulated endogenous Fgf23 mRNA expression and stimulated hypertrophic cell growth and expression of pro-hypertrophic genes. The treatment with PTH induced hypertrophic cell growth only, and stimulation with 1,25D had no significant effects. In conclusion, our data indicate that Hyp mice, in contrast to kl/kl mice appear to be protected from pathological cardiac remodeling during conditions of high FGF23 levels and klotho deficiency, which may be due, at least in part, to differences in mineral metabolism alterations, i.e., hypophosphatemia and lack of hypercalcemia.

Clinical and experimental studies indicate a possible link between high serum levels of fibroblast growth factor 23 (FGF23), phosphate, and parathyroid hormone (PTH), deficiency of active vitamin D (1,25D) and klotho with the development of pathological cardiac remodeling, i.e., left ventricular hypertrophy and myocardial fibrosis, but a causal link has not been established so far. Here, we investigated the cardiac phenotype in klotho hypomorphic (kl/kl) mice and Hyp mice, two mouse models of elevated FGF23 levels and klotho deficiency, but differing in parameters of mineral metabolism, by using histology, quantitative real-time PCR, immunoblot analysis, and serum and urine biochemistry. Additionally, the specific impact of calcium, phosphate, PTH, and 1,25D on hypertrophic growth of isolated neonatal rat cardiac myocytes was investigated in vitro. Kl/kl mice displayed high serum Fgf23 levels, increased relative heart weight, enhanced crosssectional area of individual cardiac myocytes, activated cardiac Fgf23/Fgf receptor (Fgfr) 4/calcineurin/nuclear factor of activated T cell (NFAT) signaling, and induction of prohypertrophic NFAT target genes including Rcan1, bMHC, brain natriuretic peptide (BNP), and atrial natriuretic peptide (ANP) as compared to corresponding wild-type (WT) mice. Investigation of fibrosis-related molecules characteristic for pathological cardiac remodeling processes demonstrated ERK1/2 activation and enhanced expression of Tgf-β1, collagen I, and Mmp2 in kl/kl mice than in WT mice. In contrast, despite significantly elevation of serum and cardiac Fgf23, and reduced renal klotho expression, Hyp mice showed no signs of pathological cardiac remodeling. Kl/kl mice showed enhanced serum calcium and phosphate levels, while Hyp mice showed unchanged serum calcium levels, lower serum phosphate, and elevated serum iPTH concentrations compared to corresponding WT mice. In cultured cardiac myocytes, treatment with both calcium or phosphate significantly upregulated endogenous Fgf23 mRNA expression and stimulated hypertrophic cell growth and expression of pro-hypertrophic genes. The treatment with PTH induced hypertrophic cell growth only, and stimulation with 1,25D had no significant effects.
With declining kidney function, the endocrine network of mineral metabolism becomes altered, which leads to elevated serum levels of FGF23, phosphate, and PTH in addition to a defi ciency of 1,25D and klotho (20). All these parameter represent key risk factors for the development of endothelial dysfunction, left ventricular hypertrophy (LVH), myocardial fibrosis, and contribute to the overall cardiovascular mortality in CKD and nonCKD patients (1,6,(21)(22)(23)(24)(25)(26). We previously showed that FGF23 directly targets the heart and promotes LVH by binding to FGFR4 on cardiac myocytes activating phospholipase C gamma (PLCγ)/calcineurin/nuclear factor of activated T cells (NFAT) to induce prohypertrophic gene expression independently of its coreceptor klotho that is not expressed in the heart (16,(27)(28)(29). Moreover, we demonstrated that FGF23 contributes to pathologic cardiac remodeling and promotes the profibrotic crosstalk between cardiac myocytes and fibroblasts resulting in enhanced cardiac hypertrophy and fibrosis in the absence of klotho (30).
Besides high FGF23 levels, klotho deficiency also seems to be associated with cardiac dysfunction in humans and rodents (25,29,31), and rescuing the availability of klotho by genetic overexpression or intravenous delivery of soluble klotho shows beneficial outcomes including amelioration of cardiac hypertro phy in klothodeficient uremic mice and suppression of cardiac fibroblast activation and collagen synthesis (31)(32)(33). In addition, klotho ameliorates FGF23mediated oxidative stress by inducing nitric oxide synthesis and degrading reactive oxygen species in endothelial cells (34). Furthermore, klotho protects against myocardial hypertrophy by reducing indoxyl sulfatemediated oxidative stress in vitro and in vivo (35).
To date, the impact of high phosphate levels for the develop ment of cardiomyopathy in the setting of high FGF23 levels is controversially discussed. Increased serum phosphate is associ ated with cardiovascular mortality in CKD (25,36) and FGF23 mediates renal phosphate excretion only in the presence of klotho (19). Normalization of serum phosphate and FGF23 levels by dietary phosphate restriction in 5/6 nephrectomized klothodeficient mice did not abrogate the development of car diac hypertrophy suggesting that reduced klotho contributes to uremic cardiomyopathy independent of phosphate and FGF23 (32). In contrast, cardiac hypertrophy and fibrosis correlated with high phosphate levels in uremic and nonuremic klotho deficient mice, and genetically induced elevation of soluble klotho ameliorated phosphateinduced hypertrophic growth of cardiac myocytes in vivo (31). The association between FGF23 and cardiac remodeling in klothodeficient animals suggests that FGF23 affects cardiac hypertrophy and fibrosis only in states of high phosphate and klotho deficiency. Thus, the interplay between increased FGF23, phosphate, and reduced klotho par ticipates in the development of CVD. However, the direct impact of each single factor on pathologic cardiac remodeling is still not clear. Here, we test whether elevated FGF23 levels in klotho defi ciency invariably leads to cardiac hypertrophy and fibrosis and whether different alterations in mineral metabolism play a role in these effects. Therefore, we compared two mouse models with genetically elevated serum FGF23 levels and reduced renal klotho availability, Hyp mice, and klotho hypomorphic (kl/kl) mice, with respect to the development of a pathologic cardiac phenotype. The Hyp mouse is a murine homolog to the human disease of Xlinked dominant hypophosphatemia resembling elevated circulating FGF23 concentrations resulting in renal phosphate wasting, hypophosphatemia, decreased renal 1,25D synthesis, and defects in bone mineralization (37,38). Klotho hypomorphic mice present with high plasma concentrations of FGF23, 1,25D, and phosphate, resulting in severe soft tissue calcification and premature aging (16). In addition, we used neonatal rat ventricu lar myocytes (NRVM) to study the role of parameters of altered mineral metabolism including calcium, phosphate, PTH, and 1,25D for the induction of cardiac hypertrophy.

MaTerials anD MeThODs animal experiments
All experimental procedures were approved by the State Office com mittee on animal welfare Lower Saxony for Hyp mice and Baden Württemberg for kl/kl mice and performed in accordance with national animal protection guidelines from Directive 2010/63/ Mineral Metabolism and FGF23 Cardiac Toxicity Frontiers in Endocrinology | www.frontiersin.org June 2018 | Volume 9 | Article 333 EU of the European Parliament on the protection of animals used for scientific purposes. Hemizygous male B6.CgPhex Hyp /J (Hyp) mice (strain no. 000528; Jackson Laboratory, Bar Harbor, ME, USA) with Xlinked semidominant mutation in the Phex gene causing defects in phos phate metabolism and male wildtype (WT) littermates produced from breeding of heterozygous females (Hyp/+) with C57BL/6J WT males were used in this study. Mice were fed normal rodent chow containing 600 IU/kg cholecalciferol, 0.7% calcium, and 0.5% phosphate (#1324, Altromin, Lage, Germany) ad lib. The origin of homozygous klotho hypomorphic mice (kl/kl), breeding, and genotyping were described previously (16). Five to 13 mice per group were used in this study and sacrificed at 6-8 weeks of age. Blood was collected via cardiac puncture, and hearts were isolated and prepared for histological and biochemical analyses. For serology, blood was centrifuged at 4°C and 13,000 rcf for 20 min. Serum supernatant was collected, stored at −80°C, and subsequently analyzed via ELISA techniques for Cterm FGF23, intact FGF23, and 184 PTH (each from Immutopics, San Clemente, CA, USA), and spectrophotometrically for calcium and phosphate (each from DiaSys Diagnostic Systems GmbH, Holzheim, Germany).

isolation of nrVM
Neonatal rat ventricular myocytes (NRVM) were isolated using a standard isolation system (Worthington Biochemical Corporation) (39). In brief, hearts from 1 to 2daysold Sprague Dawley rats were harvested, minced in calcium and magnesiumfree Hank's Balanced Salt Solution (HBSS) followed by tissue digestion with 50 µg/mL trypsin at 4°C for 20-24 h. Soybean trypsin inhibitor in HBSS was added and the tissue was further digested with colla genase (in Leibovitz L15 medium) under slow rotation (15 rpm) at 37°C for 45 min. Cells were homogenized and resuspended 20 times with a standard 10 mL serological pipette and filtered twice through a 70µm cell strainer (BD Falcon). After incubation at room temperature for 20 min, cells were centrifuged at 100 × g for 5 min and cell pellet was resuspended in plating medium Dulbecco's Modified Eagle Medium (DMEM) with 20% M199 (Invitrogen), 15% fetal bovine serum (FBS; Invitrogen), and 1% penicillin/streptomycin solution (P/S; Invitrogen).
Cells were plated on glass and plastic surfaces precoated with laminin (Invitrogen; 10 µg/mL in PBS) at room temperature for 1 h. For immunofluorescence analysis, 3 × 10 5 cells were seeded per well on precoated glass coverslips in 24well plates and for RNA isolations, 8 × 10 5 cells were seeded in 6 cmculture dishes. Cells were left in plating medium at 37°C over night. After starvation in maintenance medium DMEM with 20% M199, 1% insulintransferrinsodium selenite solution (ITS; Sigma Aldrich), and 1% P/S, isolated NRVM were stimulated in dupli cates in maintenance medium in the presence of 3 mM calcium, 1 mM phosphate, 10 nM PTH, 10 nM calcitriol, or vehicle, respectively, for 48 h. At least six independent cell isolations were used for all experiments.

histological analysis
Formalinfixed paraffinembedded heart tissue samples were deparaffinized in xylene, hydrated through a series of graded alcohols. For the quantification of cardiac myocyte size, fixed car diac midchamber (MC) (33) sections were incubated with wheat germ agglutinin (WGA) Alexa Fluor555 (Invitrogen) at 5 µg/mL in PBS for 1 h to visualize cellular borders of individual cardiac myocytes. 4′,6diamidino2phenylindole (DAPI; 0.2 µg/mL) was used for nuclear staining in the dark for 15 min. Representative immunofluorescence images of cardiac tissue were taken on a Zeiss AxioObserver Z1 microscope (Carl Zeiss) with a PlanApo 63×/N.A. 1.4 oil objective. ZEN Software (Carl Zeiss) was used to measure myocardial crosssectional area in square micrometer of 100 cardiac myocytes in average.
For the detection of myocardial fibrosis and visualization of fibrillar collagen fibers, MC sections were stained with picro sirius red (SigmaAldrich) as described previously (28). In brief, deparaffinized MC sections were incubated with picrosirius red for 60 min followed by mounting in nonaqueous mounting medium and analyzed by bright field and polarized microcopy using a Keyence BZ9000 ® microscope with a 20× objective. In addition, for polarized light microscopy, two polarization filters were used in a rectangular orientation, positioned above and below the sample.

immunocytochemistry and Morphometry of nrVM
To analyze hypertrophic growth of isolated NRVM on laminin coated glass coverslips after 48 h of treatment with calcium, phos phate, PTH, and 1,25D, NRVM were fixed in 2% PFA in 5 mg/mL sucrose for 5 min and permeabilized in 1% Triton X100 in PBS for 10 min followed by incubation with mouse monoclonal antibody against sarcomeric αactinin (1:1,000 dilution; EA53; SigmaAldrich). Cy3conjugated goatanti mouse (Jackson Immuno Research) was used as secondary antibody at 1:300 dilution. To visualize nuclei, fixed cells were incubated with DAPI (400 ng/mL in PBS) for 10 min. Immunofluorescence images were taken on a Zeiss AxioObserver Z1 microscope (Carl Zeiss) with a 40× objective. Myocyte crosssectional area was measured based on αactininpositive staining using Carl Zeiss ZEN software. At least 100 cells per stimulation were quantified for the determination of cardiac myocyte cross sectional area.

rna isolation and Quantitative real-Time Pcr (qrT-Pcr) analysis
For RNA isolation of snapfrozen mouse myocardial tissue or NRVM, RNeasy Mini Kit (Qiagen) was used according to the manufacturer protocol. Total RNA (500 ng) was transcribed into cDNA using QuantiTect Reverse Transcription Kit (Qiagen) and qRTPCR was performed in triplicates (20 ng cDNA per reaction) with appropriate primers in 5′-3′orientation ( Table 1) using QuantiFAST SYBR Green PCR Kit including ROX dye (Qiagen). Fortyfive cycles (95°C, 10 s; 60°C, 30 s) were performed on an ABI prism 7900HT Fast system (Applied Biosystems). Relative gene expression values, adjusted for the same CTthreshold and baseline settings, were calculated according to the 2 −ΔΔCT method (40) using Gapdh as housekeeping gene (SDS Software v2.3, Applied Biosystems).  Figures 1A,B). In the heart tissue of Hyp and kl/kl mice, endogenous cardiac Fgf23 mRNA expression and fulllength biological active Fgf23 protein were significantly upregulated compared to respective WT controls (Figures 1C,D). Thus, both mouse models showed elevated circulating and cardiac Fgf23 levels, although serum Fgf23 concentrations appeared to be higher in kl/kl compared to Hyp mice.
homozygous kl/kl, but not Hyp Mice, Develop cardiac hypertrophy Next, we investigated the development of cardiac hypertrophy in both mouse models. Due to growth retardation caused by abnor mal mineral metabolism in Hyp mice, we calculated the relative heart weight on the basis of heart weight to body weight ratio. When compared to WT controls, the relative heart weight of kl/kl mice was 0.84 ± 0.15 mg/g higher (Figure 2A). In addition, kl/kl mice presented with 282 ± 13 μm 2 crosssectional area of individual cardiac myocytes compared to 189 ± 14 μm 2 in WT  littermates demonstrated by WGA staining of the myocyte cell borders in heart tissue sections. In contrast, Hyp mice showed unaltered cardiac myocytes cell size compared to WT littermates (202 ± 19 versus 178 ± 13 μm 2 ; P = 0.287) ( Figure 2B). Thus, it appears that only kl/kl mice display the development of cardiac hypertrophy.
cardiac-specific Fgfr4/calcineurin/nfat signaling is Only activated in kl/kl Mice Since Fgfr1 and Fgfr4 are the main Fgfrs expressed in the myo cardial tissue of humans and rodents (27,29), and each induces pathological cardiac remodeling through a different pathway, we next wanted to know whether both receptors were altered in the heart of Hyp and kl/kl mice. Cardiac Fgfr1 mRNA levels were unchanged in kl/kl mice, but tended to be lower in Hyp mice; however, the difference did not reach statistical significance (P = 0.184) (Figure 3A). In contrast, cardiac Fgfr4 mRNA expres sion was 3.8 ± 0.3fold upregulated in kl/kl mice compared to controls. In Hyp mice, a trend of Fgfr4 elevation was recognized when compared to their respective WT controls (2.4 ± 0.7fold; P = 0.097) (Figure 3B). Calcineurin protein levels, activated by Fgfr4 via PLCγ, were 1.6 ± 0.1fold enhanced in heart tis sue lysates from kl/kl mice (P = 0.017), but not from Hyp mice  Figure 3C). Moreover, the transcription factor Nfat was clearly dephosphorylated (0.41 ± 0.16fold; P = 0.027) and thereby activated in heart tissue only from kl/kl mice compared to WT littermates ( Figure 3D). Taken together, the prohypertrophic Fgfr4/calcineurin/NFAT signaling pathway is induced in kl/kl, but not in Hyp mice.

Pro-hypertrophic nFaT Target Markers are Upregulated in kl/kl but not in Hyp Mice
In order to evaluate prohypertrophic genes targeted by activated calcineurin/NFAT pathway, we analyzed the expression of regu lator of calcineurin 1 (Rcan1), brain natriuretic peptide (BNP), atrial natriuretic peptide (ANP), and both alpha and betamyosin heavy chain (aMHC, bMHC) in myocardial tissue of these two mouse models. In kl/kl mice, cardiac Rcan1 expression was 2.1 fold upregulated on both mRNA and protein level when compared to WT controls but not in Hyp mice (Figures 4A,B). The switch to a fetal cardiac gene expression pattern was only detected in kl/kl mice demonstrated by an enhanced bMHC to aMHC ratio when compared to respective WT littermates ( Figure 4C). Moreover, the mRNA expression of the prohypertrophic factors ANP and BNP was clearly induced in kl/kl mice (5.3 ± 1.1fold, P = 0.0035; 2.0 ± 0.3fold, P = 0.0094) (Figures 4D,E). Interestingly, BNP levels were even 0.6fold lower in Hyp mice in comparison to their WT controls (P = 0.010).

Only klotho hypomorphic Mice Display enhanced cardiac Fibrosis
One major characteristic of pathologic cardiac hypertrophy is the concomitant development of myocardial fibrosis (3,41). Therefore, we next investigated collagen synthesis and remodeling as well as the expression levels of fibrosisrelated molecules involved in the transforming growth factorbeta (Tgfβ) signaling cascade. Demonstrated by picrosirius red staining of myocardial tissue sec tions, the accumulation of fibrillar collagens was clearly increased in kl/kl mice but not in Hyp mice ( Figure 5A). This was confirmed by a 2.0fold enhanced mRNA expression of both collagen 1 (Col1) and matrix metallopeptidase 2 (Mmp2) only in kl/kl mice com pared to WT littermates (Figures 5B,C). One major profibrotic pathway in cardiomyopathy among others is the induction of con nective tissue growth factor (Ctgf) via Tgfβ1mediated activation of extracellular signalregulated kinases (ERK)1/2 (41-43). Only in kl/kl mice, Tgfb1 mRNA levels were 3.0fold upregulated compared to respective WT controls ( Figure 5D). The enhanced profibrotic gene expressions in kl/kl mice were confirmed by higher Col1, Mmp2, and Tgfβ1 protein expressions in myocardial tissue of kl/kl mice but not in Hyp mice (Figures 5E-G). Moreover, ERK1/2 was activated in kl/kl mice demonstrated by a 2.6fold enhanced phosphorylation in total heart tissue lysates (Figures 5H,I). Finally, Ctgf protein expression was 1.5fold induced in kl/kl mice but not in Hyp mice (Figures 5H,J). In summary, interstitial cardiac fibrosis was only present in kl/kl mice.

Mineral Metabolism Differs between
Hyp and kl/kl Mice As we presented in this study, circulating Fgf23 levels and cardiac Fgf23 synthesis were significantly enhanced in both Hyp and kl/kl mice. However, only kl/kl mice developed cardiac hypertro phy and fibrosis. To identify additional causes mediating patho logic cardiac remodeling in kl/kl mice and to evaluate why Hyp mice might be protected from cardiovascular disease, we next investigated parameters of mineral metabolism. Serum calcium levels were normal to slightly reduced in Hyp mice (8.7 ± 0.8 versus 9.6 ± 0.5 mg/dL; P = 0.327) but significantly enhanced in kl/kl mice compared to WT littermates (11.3 ± 0.5 versus 9.5 ± 0.3 mg/dL; P = 0.007) (Figure 6A). In addition, Hyp mice displayed hypophosphatemia (5.6 ± 0.5 versus 8.7 ± 0.3 mg/dL; P < 0.0001) while kl/kl mice were hyperphosphatemic (9.9 ± 0.6 versus 7.4 ± 0.2 mg/dL; P = 0.0012) ( Figure 6B). Furthermore, Hyp mice showed 6.5fold enhanced serum PTH concentra tions compared to WT animals (Figure 6C), whereas previous reports describe suppressed PTH levels in kl/kl mice (44,45). Others and we showed previously that Hyp mice are deficient for 1,25D, whereas kl/kl mice have significantly higher serum 1,25D levels (37,(44)(45)(46). Finally, Hyp mice showed markedly reduced renal Klotho mRNA expression levels (0.6 ± 0.07fold; P = 0.0012), which were barely detectable in renal tissue of kl/kl mice ( Figure 6D). Taken together, despite enhanced circulating Fgf23 levels in both mouse models, serum calcium, phosphate, and 1,25D levels were increased in kl/kl mice. In Hyp mice, serum calcium was unchanged and serum phosphate and 1,25D levels were decreased compared to their WT littermates. In contrast, serum PTH levels were increased in Hyp mice, but suppressed in kl/kl mice.
calcium and Phosphate stimulate hypertrophic cell growth and induce Pro-hypertrophic genes in isolated cardiac Myocytes Since both animal models significantly differ with respect to serum calcium, phosphate, PTH, and 1,25D levels but only kl/kl mice developed cardiac remodeling processes ( Table 2), we next investigated whether each single parameter was able to affect the endogenous Fgf23 expression and to promote cardiac hyper trophy in vitro. Therefore, we stimulated NRVM, which do not express klotho, with calcium, phosphate, PTH, and 1,25D and evaluated hypertrophic growth of individual cardiac myocytes, changes of endogenous Fgf23 expression, and finally induc tion of prohypertrophic markers. Treatment of NRVM with calcium, phosphate, and PTH but not with 1,25D significantly enhanced cardiac myocytes crosssectional area demonstrated by immunocytochemical staining with antiαactinin antibody followed by quantification of the cell size (Figures 7A,B). Interestingly, Fgf23 mRNA levels of NRVM were significantly induced by calcium and phosphate treatment but neither by PTH nor by 1,25D (Figure 7C), which is in contrast to their Fgf23stimulating properties reported for bone (9,47). Rcan1 mRNA expression was significantly induced by treatment of NRVM with phosphate and tended to be increased by calcium treatment, a difference, however, not reaching statistical sig nificance ( Figure 7D). Moreover, the mRNA levels of bMHC, BNP, and ANP were significantly induced by both calcium and phosphate but not significantly modified after stimulation with PTH or 1,25D (Figures 7E-G). Taken together, among the altered parameters of mineral metabolism differing between Hyp and kl/kl mice, only calcium and phosphate clearly induced cardiac Fgf23 expression and promoted myocytes hypertrophy in vitro, suggesting that elevated phosphate and calcium levels may promote Fgf23mediated cardiac hypertrophy in states of klotho deficiency.  (48,49) --Blood pressure (44,50) ↑ ↓ Calcification (44,51) - (44,51) ↑ ↓ Serum 1,25D (37,44,51) ↓ ↑↑

DiscUssiOn
High FGF23 levels and klotho deficiency are postulated as key risk factors for the development of uremic cardiomyopathy (2,23,25,31,52) and increase of FGF23 is further associated with poor outcome in patients with heart failure, cardiogenic shock, or arrhythmia at normal kidney function (4)(5)(6)(7)(8). Even though FGF23 directly targets the heart and contributes to hypertrophy and fibrosis independent of klotho (27,28), it is not known whether the increase of FGF23 in parallel with reduction of klotho alone is sufficient to lead to CVD. In the present study, both Hyp and kl/kl mice presented with enhanced circulating levels of Cterm and intact FGF23 as well as upregulated cardiac FGF23 synthesis in addition to reduced renal Klotho expression, although both alterations were more pronounced in kl/kl mice. However, kl/kl mice, but not Hyp mice, developed cardiac hypertrophy, demonstrated by an increase in relative heart weight and cardiac myocyte crosssectional area. In addition, we demonstrated for the first time that FGFR4 is upregulated in heart tissue of kl/kl mice and this resulted in activation of calcineurin/NFAT signaling and finally induction of prohypertrophic NFAT target genes Rcan1, bMHC, ANP, and BNP. Furthermore, kl/kl mice further displayed enhanced myocardial fibrosis with concomitant induction of collagen 1, Mmp2 and Tgfβ1 expression, activated ERK1/2 and Ctgf protein. Thereby, we support the findings of Hu et al. showing that pERK1/2 is elevated and myocardial fibrosis is present in heart tissue of homozygous kl/kl mice as well (31). Moreover, this group investigated the Tgfβ1mediated activation of ERK1/2 in neonatal rat cardiac myocytes and fibroblasts in vitro, which was only present in the absence of klotho suggesting that there is a strong connection between klotho deficiency and the induc tion of profibrotic pathways. In addition, treatment of neonatal mouse cardiac fibroblasts with the 65 kDa soluble klotho isoform was shown to suppress myofibroblast proliferation and col lagen synthesis (33). Taken together, our data suggest that the wellestablished cardiac FGF23/FGFR4 signaling involved in the development of pathologic cardiac hypertrophy with the parallel appearance of cardiac fibrosis is induced in kl/kl mice with high FGF23 levels and klotho deficiency.
Our results in homozygous kl/kl mice are well in line with Yang and colleagues who showed that heterozygous kl/+ mice had elevated relative heart weight and enhanced left ventricular posterior wall thickness with reduced left ventricular internal diastolic diameter (35). In another study by Hu et al., heterozy gous kl/+ mice showed reduced ejection fraction, stroke volume, and cardiac output in addition to enlarged septum and posterior wall thickness within the diastole (31). In contrast, neither kl/+ nor kl/kl mice with comparable age had baseline cardiac abnormalities published in other studies by Xie and colleagues (32,53). Interestingly, according to this study, serum FGF23 and phosphate levels were similar in WT and kl/+ mice. The different cardiac outcome in klothodeficient mice within these studies might be, at least partially, due to differences in dietary phosphate content ranging from 0.35 to 0.6% (31,32,35,53).
The development of pathologic cardiac remodeling includ ing hypertrophy and fibrosis is less clear in Hyp mice. Recently, it was described that 8weekold Hyp mice showed increased relative heart weight, and administration of an FGFR1 activat ing antibody resulted in normalization of relative heart weight, left ventricular wall thickness, and blood pressure. Treatment of Hyp mice with soluble klotho ameliorated systolic, diastolic, and mean arterial blood pressure (MAP) (54). Andrukhova and colleagues showed enhanced FGF23 serum levels in 3month old Hyp mice with enhanced MAP in addition to significantly higher relative heart weight compared to WT littermates (50). Thus, both studies suggest that Hyp mice on a regular phos phate diet present with enhanced blood pressure and cardiac hypertrophy. However, a detailed evaluation of the cardiac phenotype in Hyp mice, i.e., investigation of cardiac myocyte crosssectional area, prohypertrophic signaling pathways, and concomitant fibrosis, in order to evalute pathological cardiac hypertrophy indepth, was not yet been performed in Hyp mice. In the present study, we investigated histological and molecular biological analyses of heart tissue in 6 to 8weekold Hyp mice on a regular phosphate diet and compared our findings to their WT littermates. In contrast to the abovementioned studies (50,54), relative heart weight was not significantly altered in Hyp mice in the present study, and additionally, cardiac myocyte crosssectional area was similar to controls. Although, Hyp mice showed comparably increased circulating FGF23 levels and the presence of enhanced cardiac FGF23 synthesis in the present study, FGFR4 and its respective downstream signaling pathway, i.e., calcineurin/NFAT, Rcan1, bMHC, ANP, and BNP, were not induced at all. Neither accumulation of fibrillar collagens nor fibrosisrelated pathways, including collagen 1 expression, Tgfβ1mediated activation of ERK1/2, or Ctgf were induced in Hyp mice compared to controls. Thus, our data point out that high FGF23 levels in addition to klotho deficiency does not necessarily result in pathological cardiac remodeling in vivo. The fact that blood pressure is elevated in Hyp mice (50) while low in kl/kl mice (44) further supported the hypothesis that FGF23mediated cardiac hypertrophy might be blood pressure independent.
Therefore, the question arises if alterations in the mineral metabolism may modulate the cardiac phenotype in these two mouse models. In line with previous published studies by others (37,55) and us (46), Hyp mice showed normal to reduced serum calcium levels, presented with hypophosphatemia, secondary hyperparathyroidism, and vitamin D deficiency. In contrast, kl/kl mice were hypercalcemic and hyperphosphatemic, had elevated 1,25D levels, and suppressed serum PTH levels (44,45). Thus, the observed differences in calcium and phosphate serum concentrations may at least partly explain the different cardiac phenotypes in the two mouse models. The importance of phosphate as a risk factor for cardiomyopathy was previously demonstrated in a variety of experimental studies showing that highphosphate diet induced cardiac hypertrophy and fibrosis in rodents (25,28,56). Moreover, stimulation with phosphate induced pERK1/2, Ctgf, and collagen 1 in isolated neonatal rat cardiac fibroblasts and in addition pERK1/2, Ctgf, and pSmad2/3 in cardiac myocytes in vitro (31). Interestingly, cotreatment with soluble klotho only ameliorated Ctgf and collagen 1 levels in cardiac fibroblasts and Smad2/3 phosphorylation in myocytes suggesting that klotho primarily targets phosphateinduced profibrotic signaling pathways. The impact of phosphate on the development of cardiac hypertrophy on cellular level was not investigated so far. Here, we present that both calcium and phosphate stimulated hypertrophic growth of cultured cardiac myocytes with upregulation of Rcan1, bMHC, ANP, and BNP, respectively. Interestingly, among the known parameters of mineral metabolism that induce FGF23 synthesis in bone (9), only calcium and phosphate upregulated endogenous Fgf23 expression in NRVM.
Circulating FGF23 levels were considerably higher and the degree of klotho deficiency was lower in kl/kl mice compared to Hyp mice in the present study. Therefore, we cannot exclude that further stimulation of circulating FGF23, e.g., by increased phosphate load may result in pathological cardiac remodeling in Hyp mice despite the presence of hypophosphatemia. In addition, kl/kl mice presented with hypercalcemia, which was shown to stimulate cardiac myocyte hypertrophy in vitro, although the effects of calcium were lower compared to phos phate. However, concomitant hypercalcemia may have further promoted the cardiac phenotype in kl/kl mice. Furthermore, in kl/kl mice, the regulation of cardiac hypertrophy and fibrosis by other pathological factors cannot be ruled out (57)(58)(59). Clearly, further studies are needed to confirm an association of altered mineral metabolism at high FGF23 levels and pathological cardiac remodeling.
In extension to previously published studies, our results support the hypothesis that both altered parameters of mineral metabolism and elevated FGF23 levels contribute to cardiac hypertrophy and fibrosis in the setting of klotho deficiency, and consequently, Hyp mice might be protected from pathologic cardiac remodeling by the presence of concomitant hypophos phatemia and lack of hypercalcemia. In addition, high FGF23 levels are only cardiotoxic in the presence of high calcium and/ or high phosphate and independent of klotho deficiency. The latter supports the concept of early initiation of phosphate lowering treatment in states of elevated FGF23 and klotho deficiency, e.g., uremia, in order to prevent pathological cardiac remodeling.

eThics sTaTeMenT
All experimental procedures were approved by the State Office committee on animal welfare Lower Saxony for Hyp mice and BadenWürttemberg for kl/kl mice, and performed in accord ance with national animal protection guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.