Relevant use of Klotho in FGF19 subfamily signaling system in vivo

Abstract

α-Klotho (α-Kl) and its homolog, β-Klotho (β-Kl) are key regulators of mineral homeostasis and bile acid/cholesterol metabolism, respectively. FGF15/ humanFGF19, FGF21, and FGF23, members of the FGF19 subfamily, are believed to act as circulating metabolic regulators. Analyses of functional interactions between α- and β-Kl and FGF19 factors in wild-type, α-kl ^−/−, and β-kl ^−/− mice revealed a comprehensive regulatory scheme of mineral homeostasis involving the mutually regulated positive/negative feedback actions of α-Kl, FGF23, and 1,25(OH)₂D and an analogous regulatory network composed of β-Kl, FGF15/humanFGF19, and bile acids that regulate bile acid/cholesterol metabolism. Contrary to in vitro data, β-Kl is not essential for FGF21 signaling in adipose tissues in vivo, because (i) FGF21 signals are transduced in the absence of β-Kl, (ii) FGF21 could not be precipitated by β-Kl, and (iii) essential phenotypes in Fgf21 ^−/− mice (decreased expressions of Hsl and Atgl in WAT) were not replicated in β-kl ^−/− mice. These findings suggest the existence of Klotho-independent FGF21 signaling pathway(s) where undefined cofactors are involved. One-to-one functional interactions such as α-Klotho/FGF23, β-Klotho/FGF15 (humanFGF19), and undefined cofactor/FGF21 would result in tissue-specific signal transduction of the FGF19 subfamily.

The physiological roles of the Klotho family have remained puzzling since the original mutant mouse was developed (1). α-Kl deficiency in mice led to a characteristic phenotype resembling premature aging symptoms in human (1). Thereafter, we found that the overproduction of 1,25(OH)₂D and altered mineral-ion homeostasis are the major cause of these premature aging-like phenotypes observed in α-kl^−/− mice, because the lowering of 1,25(OH)₂D activity by dietary restriction (a regimen in which α-kl^−/− mice are fed a vitamin D–deficient diet) (2) is able to rescue the premature aging–like phenotypes and enable α-kl–deficient mice to survive normally without obvious abnormalities. Recently we have reported that α-Kl interacts with fibroblast growth factor 23 (FGF23) in kidney and plays an essential role in maintaining serum 1,25(OH)₂D levels by regulation of key active vitamin D–metabolizing enzymes, 1α-hydroxylase (Cyp27b1), and 24-hydroxylalse (Cyp24) (3). We also found that α-Kl binds to Na⁺,K⁺-ATPase in choroid plexus, parathyroid glands, and the distal convoluted tubules (DCT) of the kidney where extracellular calcium concentration is coordinately regulated (4). In these tissues, Na⁺,K⁺-ATPase activity is controlled in an α-Kl–dependent manner for transepithelial calcium transport in the choroid plexus and DCT, and for regulated PTH secretion in the parathyroid glands. By associating with both Na⁺,K⁺-ATPase and circulating FGF23, α-Kl plays a multifunctional role in α-Kl expressing tissues to regulate calcium and phosphate concentrations in vivo. This led to the concept that α-Kl is a regulator that integrates mineral homeostasis (5).

We next identified β-kl, which shares structural identity and characteristics with α-kl (6). β-Kl is predominantly expressed in the liver, pancreas, and adipose tissues (6) distinct from α-Kl expressing tissues (1, 2). To understand the biological role(s) of β-Kl, we generated a mouse line lacking β-kl (7). Although there were no gross abnormalities in the appearance of β-kl ^−/− mice, these mice exhibited an altered metabolism of bile acids, a group of structurally diverse molecules that are primarily synthesized in the liver from cholesterol, promote absorption of dietary lipids in the intestine, and stimulate biliary excretion of cholesterol (8). The enterohepatic circulation of bile acids is regulated largely in hepatocytes where bile acid biosynthesis is regulated by rate-limiting enzymes; cholesterol 7α-hydroxylase (Cyp7a1) and sterol 12α-hydroxylase (Cyp8b1) (8). Bile acids and oxysterols act as ligands to nuclear receptors regulating the expression of important genes in cholesterol homeostasis (9). Particularly, bile acids bind to the promoter region of the farnesoid X receptor (FXR), which induces transcription of small heterodimer partner (SHP), a negative regulator of Cyp7a1 and Cyp8b1, resulting in suppression of bile acids synthesis in a negative feedback manner (9).

Simultaneously, Inagaki et al. reported that FGF15 dramatically suppresses expression of Cyp7a1 through a gut–liver signaling pathway that is different from the FXR/SHP-mediated negative feedback system (10). Moreover, the association of bile acids with FXR leads to the increase of Fgf15 expression in intestine, resulting in repression of Cyp7a1 in the liver. Importantly, this negative feedback effect was not observed in Fgf15 ^−/− and Fgfr4 ^−/− mice, and highlighted a concept that the binding of FGF15 with FGFR4 is involved in a second negative feedback system in bile acid metabolism. Taken together, analogous to the role of α-Kl in FGF23/FGFR1-mediated signal transduction, it was hypothesized that β-Kl plays a critical role in FGF15/FGFR4 mediated negative feedback regulation of Cyp7a1 and Cyp8b1 expression in the liver (11).

The mammalian FGF family currently consists of 22 members subdivided into seven subfamilies based on their structural similarity and modes of action (12). Most FGFs play an important role as paracrine factors regulating cell growth, regeneration, differentiation, and morphogenesis (13). However, it has been established that members of the FGF-19 subfamily, which also includes FGF21 and FGF23, differ in two important aspects from other FGF proteins. First, they have no or very small mitotic effects; and second, they exert their action via systemic, hormone-like effects as metabolic regulators. In fact, human FGF19 (hFGF19) and its murine ortholog FGF15, as well as FGF23, are secreted from ileal enterocytes and bone, respectively, and then circulate in the bloodstream to target tissues (12 –14). The third member, FGF21 is predominantly synthesized in the liver (15) and has beneficial effects on several metabolic parameters in different animal models of obesity; recently, FGF21 has been postulated to be a newly found regulator of glucose metabolism through induction of glucose transporter 1 (GLUT 1) (16).

As first shown for FGF23 and subsequently for FGF19, FGF21 has been predicted to require a specific cofactor for its binding to a certain type of FGFR and subsequent activation of FGF21/FGFR signaling pathway. β-Kl has been reported as a candidate cofactor essential for bioactivity of FGF21 in in vitro studies (16 –22). However, these have not been confirmed in in vivo studies. It is particularly important to examine (i) whether FGF21 signal transduction is abolished in β-kl ^−/− mice and (ii) whether the phenotypes of β-kl ^−/− mice significantly overlap with those of Fgf21-deficient mice (Fgf21 ^−/−) (23, 24).

Recent advances in understanding the signaling of FGF19 subfamilies have mainly been based on conventional in vitro experiments (13, 17, 19, 22), whereas in vivo verification of the association of FGF ligands and FGF receptor or of FGF ligands and Klotho family proteins, as well as the signal transduction (phosphorylation) cascades triggered by FGF 19 subfamilies have yet to be confirmed.

In the present study, we demonstrate the first manifest evidence revealing that whereas α-Kl and β-Kl are required for FGF23 and FGF15/hFGF19-mediated signaling pathways in vivo, respectively, β-Kl appears not to be essential for FGF21-mediated signal transduction in vivo.

Results

α-Kl–Dependent Vitamin D Regulation by FGF23.

FGF23 is derived from bone and is essential for maintaining phosphate homeostasis and regulation of vitamin D metabolism. In WT mice, administration of hFGF23 results in remarkable suppression of serum 1,25-dihydroxyvitamin D [1,25(OH)₂D] through the repression of Cyp27b1 and induction of Cyp24 in the kidney. As we previously reported, serum concentrations of 1,25(OH)₂D in both α-kl ^−/− and Fgf23 ^−/− mice were remarkably higher than that of WT mice (2, 25). Intriguingly, serum FGF23 in α-kl ^−/− mice was >8,000-fold that of WT mice (Fig. 1A). To analyze how α-Kl and FGF23 coordinately regulate vitamin D metabolism in the kidney, we analyzed the interactive actions of FGF23 and α-Kl in vivo. The FGFs used in these experiments (hFGF23, hFGF19, hFGF21) were prepared from CHO cell culture media, and their activities were estimated by measuring Egr-1-promotor directed Luciferase activities using Peak rapid cells with or without exogenous expression of α-kl or β-kl (Fig. S1). Furthermore the activity of hFGF21 was confirmed by up-regulation of Glut1 mRNA in 3T3-L1 adipocyte. To minimize the effects of hypervitaminosis D, a major cause of the abnormalities observed in α-kl ^−/− mice, we used α-kl ^−/− mice fed with a vitamin D–deficient diet, in which serum 1,25(OH)₂D levels were normal and consequently most of the premature aging–like phenotypes were alleviated (2).

Fig. 1.

FGF23 is dependent on α-Kl for regulation of vitamin D synthesis in kidney (A–E). (A) Serum concentrations of FGF23 in WT and α-kl ^−/− were measured by ELISA. WT (open bars) and α-kl ^−/− (filled bars) mice (n = 4/group) were injected with recombinant hFGF23 (0.2 mg/kg) or PBS control. Mice were killed 4 h after injection, and serum concentrations of 1,25(OH)₂D (B) were measured. Cyp27b1 (C), Cyp24 (D), and α-kl (E) mRNA levels in kidney were analyzed by RT–quantitative PCR. In this and all other figures, error bars represent mean ± SD and are plotted as fold change. Data were derived from 8- to 10-week-old male mice on vitamin D–deficient diets. *P < 0.05; **P < 0.01. FGFR1 binds to α-Kl and is phosphorylated by FGF23 in the kidney (F–I). (F) Kidney lysates were precipitated with anti-FGFR1 antibody or with control IgG. Input is 0.01% of the kidney whole extract used for the immunoprecipitation. (G) Kidney lysates were precipitated with anti-FGFR4 antibody or with control IgG. Input is 0.2% of the kidney whole extract used for the immunoprecipitation. (H) The kidney lysates of WT and α-kl ^−/− mice were immunoprecipitated with the anti-phospho-FGFRs or the anti-FGFR1 antibody. The immunoprecipitates were separated by SDS/PAGE and blotted with anti-FGFR1 antibody. (I) Kidney lysates of WT mice were immunoprecipitated with the anti–phospho-FGFRs or the anti-FGFR4 antibody and then blotted with anti-FGFR4 antibody.

hFGF23 administration induced a significant decrease in serum 1,25(OH)₂D levels in WT mice, whereas no effect was observed in α-kl ^−/− mice (Fig. 1B). Consistently, in WT kidneys injected with hFGF23, Cyp27b1 expression was reduced >13-fold, whereas Cyp24 expression was induced >5-fold (Fig. 1 C and D). However, no significant effect of hFGF23 was found on the expression of Cyp27b1 and Cyp24 in α-kl ^−/− mice. These results offer direct evidence that α-Kl is essential for FGF23-derived repression of Cyp27b1 and induction of Cyp24 in vivo. In addition, we found that the administration of hFGF23 resulted in down-regulation of α-Kl expression (Fig. 1E), probably because α-Kl is a target of FGF23 signal transduction and/or because of a secondary effect of decreased 1,25(OH)₂D, an inducer of α-Kl gene expression (2). These data implicate an elaborate mutual negative feedback system composed of α-Kl, FGF23, and 1,25(OH)₂D in mineral-ion maintenance (Fig. 5A).

Fig. 5.

Schematic representation of α-Kl/FGF23 and β-Kl/FGF15 systems. (A) Regulatory network of mineral homeostasis illustrated by the mutual positive/negative feedback actions of α-Kl, FGF23, and 1,25(OH)₂D. (B) Regulatory network of bile acid/cholesterol metabolism represented by the mutual positive/negative feedback actions of β-Kl, FGF15, and bile acids.

α-Kl–Dependent FGFR1 Phosphorylation by FGF23 in Vivo.

Generally FGFs can bind to and activate cell surface tyrosine kinase FGF receptors and transduce signals to downstream molecules including MAP kinase (26). The FGF receptor family consists of four members, FGFR1–4. With the exception of FGFR4, splicing variants in the third Ig-like domain (IIIb and IIIc types) have been identified for each member (12, 26). Recently, it has been reported that α-Kl binds to FGFRs in cultured cells (19) and converts the canonical FGFR1(III)c to a receptor specific for FGF23 (3). We therefore tested whether the above observations were valid in vivo. We first examined the interactions between α-Kl and FGFR1, and α-Kl and FGFR4 in the kidney. As observed in in vitro experiments, α-Kl was coprecipitated not only with FGFR1 but also with FGFR4 in the kidney (Fig. 1F and G). We then investigated whether these two receptors are activated by hFGF23 in the kidney (procedures are as in SI Text and Fig. S2). In WT mice, FGFR1 was activated in the kidney 10 min after injection of hFGF23 (Fig. 1H). In contrast, phosphorylation of FGFR4 was not detectable even after the injection of hFGF23 (Fig. 1I), suggesting that FGFR4 is not a major receptor responsible for FGF23 signaling in the kidney. As expected, we could not detect phosphorylation of FGFR1 in the kidney of α-kl^−/− mice even after hFGF23 injection (Fig. 1H). In summary, we concluded that FGFR1 is preferentially activated by FGF23 in a α-Kl–dependent manner in the kidney.

β-Kl–Dependent Bile Acid Regulation by FGF15.

Because the unusually elevated expression of Cyp7a1 was observed not only in β-kl ^−/− mice (7) but also in Fgf15 ^−/− and Fgfr4 ^−/− mice (10, 27), we predicted that β-Kl was involved in FGF15/FGFR4-signaling system. Based on studies in cultured cells, it was recently proposed that β-Kl is necessary for FGF15/hFGF19-mediated signal transduction in the liver (18, 22). To confirm this hypothesis in vivo, we first measured the mRNA levels of Fgf15 in the terminal ileum of WT and β-kl ^−/− mice. Interestingly, Fgf15 expression levels were ∼12-fold increased in β-kl ^−/− mice compared with those of WT (Fig. 2A), analogous to the elevation of FGF23 expression in α-kl ^−/− mice (Fig. 1A). To evaluate the effect of FGF15 in vivo, we administered hFGF19 and analyzed Cyp7a1 and Cyp8b1 expression. In WT mice, the expression levels of Cyp7a1 and Cyp8b1 6 h after hFGF19 injection resulted in >100-fold and >10-fold reductions, respectively (Fig. 2 B and C). These were comparable to findings in a previous study examining FGF15 (10), and thus we concluded hFGF19 could be used to evaluate bile acid regulation in mice. In contrast, the expression levels of Cyp7a1 and Cyp8b1 remained elevated in β-kl ^−/− livers even after the administration of hFGF19 (Fig. 2 B and C), demonstrating that β-Kl is essential for the negative regulation of Cyp7a1 and Cyp8b1 by FGF15/hFGF19 in vivo. β-Kl-regulated bile acid synthesis by FGF15/hFGF19 is further described in SI (Fig. S3).

Fig. 2.

FGF19 is dependent on β-Kl for regulation of bile acid synthesis in liver (A–D). (A) mRNA levels of Fgf15 in terminal ileum in WT and β-kl ^−/− were measured by RTQ-PCR. WT mice (open bars) and β-kl ^−/− mice (filled bars) (n = 5/group) were injected with recombinant hFGF19 (1 mg/kg) or control medium. Mice were killed 6 h after injection and Cyp7a1 (B) and Cyp8b1 (C) mRNA levels in liver were measured by RT–quantitative PCR. Data were derived from 14- to 16-week-old male mice on standard diet. Egr-1 induction mediated by FGF19 in liver (D). hFGF19 (1 mg/kg) or control medium were injected into WT and β-kl^−/− male mice (12–14 weeks old) on standard diet. Thirty minutes after injection, tissues in WT (Upper) and β-kl ^−/− (Lower) mice (n = 4/group) were excised. Egr-1 mRNA levels were measured by RT-quantitative PCR. The expression levels of FGF19-injected mice (filled bars) and vehicle injected mice (open bars) are plotted as fold change. *P < 0.05; **P < 0.01.

β-Kl/FGFR4 Coexpression Is Required for FGF15 Signaling in Vivo.

To monitor whether FGF15/hFGF19 signals are transduced in tissues other than the liver, we verified Egr-1 (a zinc-finger transcription factor identified as an immediate-early gene induced by cellular stimulation) mRNA levels in β-Kl–expressing tissues (liver, adipose, pancreas, and salivary gland) as well as several β-Kl–nonexpressing tissues, since hFGF23 administration remarkably increased Egr-1 expression and induced phosphorylation of 44/42 MAP kinase (ERK1/2) in the kidney where α-Kl is expressed (3). In WT liver, Egr-1 expression level increased by >120-fold 30 min after hFGF19 administration compared with vehicle (Fig. 2D). With respect to other tissues, we observed a faint, but statistically significant Egr-1 increase in pancreas (>5-fold) and in white adipose tissue (WAT) (∼3-fold). Nonetheless, no remarkable changes were observed in other tissues including brown adipose tissue (BAT) and salivary gland despite β-Kl expression. As expected, in β-kl ^−/− mice injected with hFGF19, no significant induction of Egr-1 was observed in any of the tissues evaluated (Fig. 2D), demonstrating that β-Kl is necessary but not sufficient for FGF15/hFGF19-mediated signal transduction. To address the question of why FGF15/hFGF19 signal is transduced in the liver, pancreas, and WAT, but not in BAT and salivary glands, we profiled the expression of various FGF receptors (FGFRs) in β-Kl–expressing tissues. As reported previously (10), FGFR4 is postulated to be the major receptor responsible for FGF15-mediated signal transduction in the liver. As for the pancreas and WAT, we did observe >2-fold and >6-fold lower Fgfr4 expression compared with that in the liver, respectively. On the contrary, Fgfr4 mRNA was not detected in the salivary glands and BAT (Fig. S4). Hence Egr-1 up-regulation by hFGF19 could be observed in tissues where β-Kl and FGFR4 are coexpressed.

To further demonstrate the contribution of β-Kl in the hepatic FGF15/hFGF19-mediated signaling cascade, we evaluated the phosphorylation of FGFR4 and downstream signaling molecules in vivo using the methods shown in Fig. 1 (Fig. S2 and SI Text). β-Kl could be efficiently precipitated by an anti-FGFR4 antibody (Fig. 3A) and the phosphorylation of FGFR4 was confirmed after hFGF19 treatment in WT liver (Fig. 3B). Unexpectedly, the amount of FGFR4 protein was significantly reduced in livers of β-kl ^−/− mice (Fig. 3C). To obtain an amount of FGFR4 equivalent to that obtained from WT mice, we concentrated the liver lysates from β-kl ^−/− mice and performed immunoprecipitation (Fig. S2 and SI Text). However, we could not detect enhanced activation of FGFR4 in β-kl ^−/− livers even after injection of hFGF19 (Fig. 3D). Consistent with a previous report (18), clear phosphorylation of ERK1/2 was observed in WT livers 10 min after hFGF19 injection, whereas it was undetectable in β-kl ^−/− livers (Fig. 3E), demonstrating that β-Kl is essential for the FGF15/hFGF19 directed activation of FGFR4 and downstream signaling cascade in the liver.

Fig. 3.

FGFR4 binds to β-Kl and is phosphorylated by FGF19 in liver. (A) Liver lysates were precipitated with anti-FGFR4 antibody or with control IgG. Input is 1% of the liver whole extract used for the immunoprecipitation. The immunoprecipitates were separated by SDS/PAGE and blotted with anti-FGFR4 antibody. (B, D, and E) Ten minutes after injection of hFGF19 or control medium, livers were excised. Liver lysates from WT mice (B) and β-kl ^−/− mice (D) were immunoprecipitated with the anti–phospho-FGFRs or the polyclonal anti-FGFR4 antibody. Immunoprecipitates were blotted with anti-FGFR4 antibody. The arrowhead indicates a nonspecific band (Fig. S2). (C) Whole liver extracts of WT and β-kl ^−/− mice were blotted with anti-FGFR4 antibody or anti–β-actin antibody for a loading control. (E) Liver lysates from WT and β-kl ^−/− were immunoblotted with anti-phospho-ERK1/2 or anti-ERK1/2 antibodies (n = 2 in each case).

β-Kl Is Not Essential for FGF21-Mediated Signaling in Adipose Tissues.

FGF21, a member of the FGF19 subfamily that is synthesized in the liver, has been reported to be a newly found regulator of glucose metabolism (16) and β-Kl has been postulated to be essential for its activity in in vitro studies (17, 20, 21). To examine the possible contribution of β-Kl in the FGF21 signaling system in vivo, we first administered recombinant hFGF21 to WT mice and analyzed Egr-1 mRNA levels in multiple tissues (Fig. 4A). Before its use, we confirmed the biological activity of the synthesized hFGF21. As shown in Fig. S5, our hFGF21 could enhance Egr1-derived luciferase reporter expression in a β-Kl–dependent manner at doses that were equivalent to those previously reported (17, 21). Moreover the hFGF21 up-regulated Glut1 mRNA in 3T3-L1 adipocyte (Fig. S5) (16). Consistent with the in vitro results, in WT mice, Egr-1 expression levels were significantly up-regulated by ∼10-fold in WAT and >6-fold in BAT 30 min after injection of hFGF21 (Fig. 4A and Fig. S5). However, administration of hFGF21 also resulted in significant up-regulation of Egr-1 mRNA levels in WAT and BAT from β-kl ^−/− mice (Fig. 4B). We next analyzed the serum levels of FGF21 and hepatic mRNA levels of Fgf21 in WT and β-kl ^−/− mice. Unexpectedly, there was no significant genotype-dependent difference in mean serum protein concentrations of FGF21 nor hepatic Fgf21 mRNA levels (Fig. 4 C and D). We also confirmed that β-kl expression was not affected by FGF21 administration (Fig. 4E). These results suggest that β-Kl is not essential for FGF21-mediated signaling in WAT and BAT. In addition, our prediction was further supported by the following experiments. First, to address the binding properties between β-Kl and FGF21, we performed pull-down assays using recombinant proteins. Although FGF19 was significantly bound by β-Kl in the presence of FGFR4, FGF21 could not be precipitated by β-Kl even with 10 fold amounts of FGF21 (Fig. 4 F and G and SI Text). Second, we compared the phenotypes of Fgf21 ^−/− and β-kl ^−/− mice. Recently, Hotta et al. developed Fgf21 ^−/− mice and reported that expression levels of hormone-sensitive lipase (Hsl) and adipose triglyceride lipase (Atgl) in WAT were decreased in Fgf21 ^−/− mice to almost 50% compared with those of WT mice (23). The adipose phenotypes in Fgf21 ^−/− mice may be an outcome of a deficiency in FGF21 signaling. Thus we analyzed the expression levels of these genes in the adipose tissues of β-kl ^−/− mice. Consequently, in both WAT and BAT, mRNA levels of Hsl and Atgl were not significantly altered between WT and β-kl ^−/− mice (Fig. 4 H and I). These data suggest that β-Kl may not necessarily be involved in the phenotypes observed in FGF21-deficient mice. Taken together, these results provide strong evidence that β-Kl is not essential for FGF21-mediated signaling in WAT and BAT. We therefore asked whether FGF21 signaling might require other unidentified components than β-Kl. We also analyzed Glut1 mRNA levels 4 h after hFGF21 injection at concentrations that could induce Egr-1 expression in WAT and BAT, but no apparent induction was observed, suggesting that Glut1 is not a direct target of FGF21 signaling (28, 29).

Fig. 4.

β-Kl is not essential for FGF21-mediated signaling (A–I). Thirty minutes after injection, tissues in WT (n = 5/group) (A) and β-kl ^−/− mice (n = 4/group) (B) were excised. Egr-1 mRNA levels were measured by RT-quantitative PCR. The expression levels of hFGF21 injected mice (filled bars) and vehicle injected mice (open bars) mice are plotted as fold change. Data were derived from 7- to 10-week-old male mice on standard diet. (C) Serum FGF21 concentrations of WT and β-kl ^−/− mice (n = 5–6/group) were measured by RIA. (D) Fgf21 mRNA levels in livers of WT and β-kl ^−/− mice (n = 5–6/group) were measured by RT-quantitative PCR and are plotted as fold change. Data were derived from 15- to 20-week-old male mice on standard diet. hFGF21 (0.4 mg/kg) or control medium were injected into WT and β-kl^−/− mice. (E) WT mice were injected with recombinant hFGF21 (0.4 mg/kg) or control medium (n = 5/group). Mice were killed 4 h after injection and β-kl mRNA levels in WAT were analyzed by RT-quantitative PCR. Data were derived from 10-week-old male mice on standard diet. (F) A 15-ng quantity of each FGF was pulled down (PD) by α-/β-Kl in the presence (R1, R3, or R4) or absence (–) of FGFRs. Input was 8% of samples used for the pull-down assay. Samples of pulled down by α-/β-Kl were analyzed by SDS/PAGE and blotted with antibodies (anti-His for FGFs, anti-Human Fc for FGFRs, and anti-GFP for α-/β-Kl). Arrowhead indicates a nonspecific band. (G) A 50-ng quantity of FGF19, 150 ng of FGF23, or 500 ng of FGF21 was precipitated by β-Kl in the presence or absence of FGFR4. Input was 1% of samples used for the assay. (H and I) hsl and atgl mRNA levels in WAT and BAT were analyzed by RT-quantitative PCR. Data were derived from 9- to 14-week-old female β-kl ^+/+ and β-kl ^−/− mice (n = 5/group). *P < 0.05; **P < 0.01.

Discussion

Various roles of Klotho family members have been reported (18, 22, 30); however, a consensus on the molecular functions of α-Kl and β-Kl has not been reached. Based on these findings, we propose a comprehensive regulatory scheme of mineral homeostasis that is illustrated by the mutually regulated positive/negative feedback actions of α-Kl, FGF23, and 1,25(OH)₂D (Fig. 5A). In the present study, we found that FGF23 represses the expression of α-Kl and identified an essential role of α-Kl in FGF23-mediated phosphorylation of FGFR1 in the kidney. This leads to Cyp27b1 down-regulation and Cyp24 up-regulation, and results in inhibition of the synthesis of 1,25(OH)₂D, an active form of vitamin D (3). 1,25(OH)₂D has prominent effects on the kidney, intestine, and bone. In the kidney, 1,25(OH)₂D activates vitamin D receptor (VDR) by binding to its ligand binding domain and negatively regulates the expression of Cyp27b1 while positively regulating Cyp24 and α-Kl expression (2). In the bone, 1,25(OH)₂D binds to VDR and induces FGF23 synthesis in osteocytes and osteoblasts (31) in hours/days. In turn, secreted FGF23 suppresses 1,25(OH)₂D synthesis and inorganic phosphate reabsorption in the kidney to adjust extracellular mineral concentrations. Collectively, α-Kl, in combination with FGF23, is involved in a signaling cascade that maintains extracellular calcium/phosphate levels within a narrow range.

The roles of β-Kl, FGF15, and FGFR4 in bile acid/cholesterol metabolism are schematically summarized in Fig. 5B. Consistent with a previous study (10, 22), i.v. injection of hFGF19 dramatically represses the expression of Cyp7a1 and Cyp8b1 and results in the inhibition of bile acid synthesis from cholesterol in WT livers. This suppression of Cyp7a1 and Cyp8b1 was not observed in β-kl ^−/− mice. Indeed, phosphorylation of FGFR4 and ERK1/2 was not detected in β-kl ^−/− livers even after hFGF19 administration. Our findings provide conclusive evidence proving the essential role of β-Kl in FGF15/hFGF19-mediated activation of FGFR4 and subsequent signal transduction that regulates bile acid synthesis. Particularly, by binding to FXR, bile acid induces SHP expression in the liver and FGF15 transcription in the terminal ileum. In turn, increased SHP and secreted FGF15 differentially suppress Cyp7a1/Cyp8b1 expression to down-regulate bile acid synthesis (8, 9, 11). In addition, we found mutual negative feedback regulations between β-Kl and FGF15, namely, a decrease in β-kl after hFGF19 administration and an increase in Fgf15 in β-kl deficiency. In other words, β-kl ablation leads to impaired negative feedback regulation of bile acid metabolism, resulting in the overflow of bile acid pools. Consequently, in the β-kl ^−/− terminal ileum, chronic stimulation by elevated bile acid would lead to an unusual increase in Fgf15 mRNA. We propose a scheme illustrating the bile acid/cholesterol homeostasis regulated by mutual negative/positive feedback actions of β-Kl, FGF15, and bile acids (Fig. 5B).

As shown in Fig. 5, the scheme for bile acid regulation by β-Kl/FGF15 is conceptually analogous to that of vitamin D metabolism, which involves α-Kl and FGF23. Both systems are regulated by the coordination of two types of feedback mechanisms mediated by end-metabolites, 1,25(OH)₂D or bile acids, that are in situ negative feedback regulation and target tissue mediated negative feedback loop. In the former pathway, the end-metabolite functions as a nuclear receptor ligand and negatively feeds back by repressing the expression of key regulatory enzymes (Cyp27b1 in the kidney or Cyp7a1/Cyp8b1 in the liver) in the relevant metabolic pathway responsible for the generation of end-metabolite itself. In the latter system, the end-metabolite is transported to the target tissue (bone or intestine) from a distal site and enhances the expression of FGF (FGF23 or FGF15) by binding to the nuclear receptor; VDR or FXR. Subsequently, secreted FGF acts as the regulator of a target tissue–mediated negative feedback loop in collaboration with α-Kl or β-Kl. The next question to be addressed is how these two pathways are coordinately involved in the rapid adjustment and long term maintenance of mineral homeostasis and bile acid metabolism.

In a previous report, we showed that i.v. injection of hFGF23 induces phosphorylation of ERK1/2 and specifically up-regulates the expression of Egr-1 in the murine kidney (3). Here we demonstrate that α-Kl is required for FGF23 signal transduction in vivo. Likewise, i.v. injection of hFGF19 results in ERK1/2 phosphorylation and up-regulation of Egr-1 in the liver in a β-Kl–dependent manner. Among β-Kl–expressing organs, significant up-regulation of Egr-1 was observed in tissues where β-Kl and FGFR4 are coexpressed. Although induction of Egr-1 in pancreas and WAT are slight, it occurs in a β-Kl–dependent manner. FGF15-mediated signal in pancreas and WAT could therefore be involved in bile acid homeostasis, but its functional importance has yet to be elucidated. Furthermore, the very high Egr-1 induction in the liver strongly suggests that other elements, in addition to the coexpression of β-Kl and FGFR4, may endow this prominent hepatic signal activation. Recently, several groups have reported that α-Kl and β-Kl can bind to certain types of FGFRs. Those studies report preferences and differences for this binding that might be dependent on assay conditions. α-Kl solely binds to FGFR1(IIIc) in vitro (3), however α-Kl binds to not only FGFR1(IIIc) but also FGFR4 and weakly to FGFR3(IIIc) in cultured cells (30). Even though FGFR4 could precipitate α-Kl in the kidney, activation of FGFR4 by hFGF23 could not be detected. Further studies are required to understand how FGFR(s) is definitively and preferentially used for a particular FGF signal in vivo.

Serum levels of FGF23 and ileac Fgf15 mRNA expression were intensively increased in α-kl^−/− and β-kl^−/− mice, respectively. Furthermore, administrations of hFGF23 and hFGF19 apparently suppressed the expression of α-kl and β-kl, respectively. In contrast, the serum levels of FGF21 and hepatic Fgf21 mRNA expression were not increased in β-kl^−/− mice, and β-kl expression was not significantly suppressed by hFGF21 in adipose. Consistent with a previous report that FGF21 induces ERK1/2 phosphorylation specifically in WAT (17), administration of FGF21 to WT mice significantly induced Egr-1 mRNA expression in WAT and BAT, suggesting that WAT and BAT were the possible target tissues of FGF21. However, surprisingly, remarkable Egr-1 inductions in WAT and BAT were also observed in β-kl ^−/− mice, indicating that β-Kl is not essential for FGF21 signal transduction in vivo. These in vivo results contrasted with those obtained from in vitro assays, as β-Kl is essential for FGF21-mediated signal transduction in vitro. We reproduced the direct binding of α-Kl and hFGF23 and also confirmed tricomplex formation of β-Kl, hFGF19, and FGFR4 but were unable to detect binding of α-/β-Kl and hFGF21 in our pull-down assay (Fig. 4G). Furthermore, we confirmed that the adipose phenotypes in Fgf21 ^−/− mice did not overlap with those of β-kl ^−/− mice. This inconsistency leads to a postulation that β-Kl is not necessary for FGF21 signaling.

Currently, β-Kl is believed to be a common player essential for FGF15- and FGF21-mediated signal transduction. However, our present results, together with the data from Hotta et al. (23), do not support this hypothesis. Possible explanations are that the response found in cultured cells might be caused by: (i) an artificial abundance of β-Kl and/or FGF21, (ii) peculiar characteristics of the cultured cells used in these experiments, and/or (iii) a combination of these two factors (17, 20, 21).

Recent studies have reported that FGF21 stimulates lipolysis in WAT and ketogenesis in the liver (32, 33). However, those results represent the pharmacological effects of sustained FGF21 treatment and thus include consequences that are secondary and indirectly induced by FGF21. We propose a β-Kl–independent response directly triggered by hFGF21 administration. Significant Egr-1 up-regulation in WAT and BAT are indicative that FGF21 mediates lipid metabolism in adipose tissues. The physiological target(s) of FGF21 signaling need to be clarified to understand how FGF21 functions as a regulator of lipid metabolism. Observations using genetic manipulation will lead us to a precise understanding of the roles of the FGF19 subfamily in metabolic homeostasis in vivo.

Materials and Methods

Measurement of Serum Parameters.

Blood samples were collected from orbital cavities or hearts under anesthesia and were centrifuged to obtain sera. Serum FGF23 levels were measured by sandwich ELISA (Kainos Laboratory), which can quantify the intact form or FGF23 using human recombinant FGF23 as a standard. Serum 1,25(OH)₂D levels were analyzed by SRL, Inc. Serum FGF21 levels were measured by specific RIA (Phoenix Pharmaceuticals, Inc.).

Statistical Analysis.

Unless otherwise noted, all values are expressed as mean ± SD. All data were analyzed by the Mann–Whitney U test. P values less than 0.05 were considered to be statistically significant.

More details are described in SI Materials and Methods.