|Home | About | Journals | Submit | Contact Us | Français|
Fibroblast growth factor 21 (FGF21) is a fasting-induced hepatokine that has potent pharmacologic effects in mice, which include improving insulin sensitivity and blunting growth. The single-transmembrane protein, βKlotho, functions as a co-receptor for FGF21 in vitro. To determine if βKlotho is required for FGF21 action in vivo, we generated whole-body and adipose tissue-selective βKlotho knockout mice. All of the effects of FGF21 on growth and metabolism were lost in whole-body βKlotho knockout mice. Selective elimination of βKlotho in adipose tissue blocked the acute insulin-sensitizing effects of FGF21. Taken together, these data demonstrate that βKlotho is essential for FGF21 activity and that βKlotho in adipose tissue contributes to the beneficial metabolic actions of FGF21.
Fibroblast growth factor 21 (FGF21) is a member of an atypical subfamily of FGFs, including FGF15/19 and FGF23, that can circulate as hormones (Potthoff et al., 2012). In rodents, FGF21 expression is strongly induced in liver by fasting through a peroxisome proliferator-activated receptor (PPAR) α-dependent mechanism (Badman et al., 2007; Inagaki et al., 2007; Lundasen et al., 2007). Consistent with a role in the response to long-term fasting and starvation, FGF21 stimulates hepatic fatty acid oxidation, ketogenesis and gluconeogenesis while inhibiting somatic growth in lean mice (Badman et al., 2009; Badman et al., 2007; Inagaki et al., 2007; Inagaki et al., 2008; Potthoff et al., 2009). Fgf21 is also induced in white adipose tissue (WAT) by fasting/refeeding regimens and PPARγ agonists (Dutchak et al., 2012; Muise et al., 2008; Oishi et al., 2011; Wang et al., 2008), and in brown adipose tissue (BAT) by cold exposure (Chartoumpekis et al., 2011; Hondares et al., 2011). In WAT, FGF21 acts in an autocrine fashion to regulate PPARγ activity (Dutchak et al., 2012) and to stimulate a thermogenic response (Fisher et al., 2012). In obese rodents and monkeys, FGF21 administration improves insulin sensitivity, normalizes triglyceride and cholesterol levels, and causes weight loss (Coskun et al., 2008; Kharitonenkov et al., 2005; Kharitonenkov et al., 2007; Xu et al., 2009). Thus, FGF21 is a promising drug candidate for treating human metabolic disease.
In vitro studies showed that FGF21 and FGF15/19 act through a cell surface receptor complex composed of conventional FGF receptors and βKlotho, a single-pass transmembrane protein (Kharitonenkov et al., 2008; Ogawa et al., 2007; Suzuki et al., 2008). FGF21 interacts directly with the extracellular domain of βKlotho in the FGFR/βKlotho complex and activates FGF receptor substrate 2α and ERK1/2 phosphorylation. Whereas the FGFRs are expressed in most tissues, βKlotho expression is restricted to just a few, including liver and both WAT and BAT (Fon Tacer et al., 2010). FGF21 modulates the expression of metabolic genes in each of these tissues (Coskun et al., 2008).
Mice lacking βKlotho in all tissues are viable and fertile (Ito et al., 2005). However, these βKlotho-knockout (KO) mice are smaller than their wild-type counterparts and have increased hepatic bile acid synthesis and smaller gallbladders. The latter two phenotypes are consistent with the established roles of FGF15/19 in repressing hepatic bile acid synthesis and stimulating gallbladder filling. Subsequent studies showed that βKlotho-KO mice are refractory to the effects of FGF19 on gene expression in liver and other tissues (Tomiyama et al., 2010). While these results demonstrate that βKlotho is required for FGF15/19 action, to date there is no corresponding in vivo evidence that βKlotho is required for FGF21 action. Indeed, a recent study suggested that βKlotho is not required for FGF21 to induce expression of the transcription factor, early growth response 1 (EGR1), in WAT and BAT (Tomiyama et al., 2010), leaving the relationship between FGF21 and βKlotho in question. In this study, we have examined the requirement of βKlotho for FGF21 action. We show that mice lacking βKlotho are resistant to both acute and chronic effects of FGF21. We further show that βKlotho in adipose tissue is crucial for the acute insulin-sensitizing actions of FGF21.
To generate global βKlotho-KO mice, loxP recombination sites were introduced flanking exon 1 of the βKlotho (Klb) gene, which encodes the start codon and first 275 amino acids of the protein (Figure S1A and S1B). Mice heterozygous for the floxed (fl) Klb allele were bred with mice expressing a Meox-Cre transgene, which deletes in the germ line, to generate Klb+/– heterozygous (HET) mice. The HET mice were then intercrossed to yield Klb–/– (KO) mice. No Klb mRNA or protein was detected in BAT, WAT or liver from KO mice (Figure 1A and 1B). Consistent with a previous report (Tomiyama et al., 2010), Ffg15 expression was significantly increased in the ileum in KO mice (Figure S1C). Fgf21 mRNA in the liver and protein in the plasma also trended higher in the KO mice (Figure S1C).
We examined whether βKlotho is required for FGF21 signaling in BAT, WAT and liver. To generate sufficient numbers of mice for these and subsequent studies, HET-KO crosses were bred to yield HET and KO progeny, which were then compared. As expected (Fisher et al., 2010; Tomiyama et al., 2010), FGF21 injection induced Egr1 expression in BAT, WAT and liver of HET mice (Figure 1C). The effect of FGF21 on Egr1 expression was lost in KO mice. These data show that βKlotho is required for the acute effect of FGF21 on Egr1 expression in multiple tissues.
We next tested whether βKlotho is required for the acute insulin-sensitizing actions of FGF21. Co-injection of insulin and FGF21 into HET mice caused a decrease in blood glucose levels that was not seen when insulin was injected with vehicle alone (Figure 1D). Importantly, this insulin-sensitizing effect of FGF21 was lost in KO mice. Thus, βKlotho is required for the acute insulin-sensitizing actions of FGF21. In complementary experiments, we measured the effect of FGF21 on tissue-specific glucose uptake under these same ITT conditions. As shown in Figure 1E, FGF21 significantly induced glucose uptake in BAT but not WAT or gastrocnemius or soleus skeletal muscle in wild-type mice 30 minutes after injection. These data demonstrate that BAT contributes to the acute glucose-lowering effect of FGF21.
To determine whether βKlotho is required for long-term FGF21 action, KO mice were crossed with Fgf21-transgenic (Tg) mice (Inagaki et al., 2007) to generate four genotypes: HET, HET/Tg, KO and KO/Tg mice. As expected, HET/Tg mice weighed less and had shorter tibiae than HET mice (Figure 2A and 2B), which is consistent with FGF21 inhibiting growth (Inagaki et al., 2008; Kubicky et al., 2012). KO mice also weighed less and had shorter tibiae than HET mice, demonstrating that βKlotho is required for normal growth as previously described (Ito et al., 2005). Notably, the effects of transgenic FGF21 overexpression on body weight and tibia length were lost in KO/Tg mice (Figure 2A and 2B). FGF21 overexpression also increased food intake, lowered body fat content and increased lean body content in HET mice but not in KO mice (Figure 2C and 2D). Thus, βKlotho is required for the effects of FGF21 on growth and body composition in lean mice.
Long-term treatment of obese mice with FGF21 suppresses weight gain and improves insulin sensitivity and other metabolic parameters (Coskun et al., 2008; Kharitonenkov et al., 2005; Xu et al., 2009). To determine if βKlotho is required for these effects, groups of HET, HET/Tg, KO and KO/Tg mice were fed a high fat diet (HFD). HET/Tg mice gained weight more slowly on the HFD than HET, KO and KO/Tg mice despite higher food intake (Figure 2E, S2A and S2B). MRI measurements done after 2 months on the HFD showed that HET/Tg mice had reduced percent fat mass and increased percent lean mass compared to the other three genotypes (Figure 2F). Consistent with these findings, plasma leptin concentrations were lower in HET/Tg mice compared to all the other groups while plasma adiponectin concentrations were higher (Table 1). The liver weight:lean mass ratio was reduced, and hepatic lipids, including triglyceride and cholesterol, were lower in HET/Tg mice compared to the other groups (Figure 2G-J). HET/Tg mice also had reduced plasma cholesterol concentrations (Table 1). These findings show that βKlotho is required for the effects of FGF21 on body composition and lipid homeostasis in HFD-fed mice.
We also examined the role of βKlotho in carbohydrate homeostasis. HET/Tg mice had markedly lower fed-state plasma insulin and glucose concentrations compared with HET, KO and KO/Tg mice (Table 1). In glucose tolerance tests, HET/Tg mice had a reduced glucose excursion compared to the other groups (Figure 2K). In insulin tolerance tests, HET/Tg mice showed a much more pronounced decrease in glucose levels than HET, KO and KO/Tg mice (Figure 2L). We conclude that βKlotho is required for the effects of FGF21 on carbohydrate homeostasis, including its insulin sensitizing actions.
We next compared the expression of metabolic genes in BAT, WAT and liver of HET, HET/Tg, KO and KO/Tg mice. In BAT, uncoupling protein-1 (Ucp1), deiodinase 2 (Dio2), and elongation of very long chain fatty acids like-3 (Elovl3) were elevated in HET/Tg compared to HET mice (Figure S2C). In epididymal WAT (eWAT), peroxisome proliferator-activated receptor γ coactivator 1α (Pgc1a), phosphoenolpyruvate carboxykinase (Pck1), acetyl-CoA carboxylase α (Acaca), stearoyl-Coenzyme A desaturase 1 (Scd1), adiponectin (Adipoq) and resistin (Retn) were increased in HET/Tg compared to HET mice (Figure S2D). In subcutaneous WAT (scWAT), Ucp1, Pgc1a, Pck1, Acaca, Scd1 and resistin Retn were increased in HET/Tg compared to HET mice (Figure S2E). The finding that FGF21 induces Ucp1 in scWAT is in agreement with a recent report in which it was shown that FGF21 efficiently stimulates the browning of inguinal and perirenal WAT (Fisher et al., 2012). In liver, Pgc1a was increased and fatty acid synthase (Fasn), Scd1, Acaca, Cd36 and Elovl6 were decreased in HET/Tg compared to HET mice (Figure S2F). Notably, all of these FGF21-dependent changes in gene expression were lost in the KO background (Figure S2C-F). Thus, βKlotho is essential for FGF21 to exert its effect on metabolic gene expression in multiple tissues.
WAT and BAT are well documented targets of FGF21 expression and activity (Chartoumpekis et al., 2011; Dutchak et al., 2012; Hondares et al., 2011; Kharitonenkov et al., 2005; Muise et al., 2008; Oishi et al., 2011; Wang et al., 2008). To begin to assess the contributions of adipose tissue to the pharmacologic actions of FGF21, we crossed Klbfl/fl mice with aP2-Cre mice, which express Cre in adipose tissue, to yield Klbfl/fl;aP2-Cre (KOaP2) mice. Klb mRNA was almost entirely absent in BAT, reduced by ~70% in WAT and unchanged in liver of KOaP2 mice (Figure 3A). Western blot analysis confirmed that these changes in Klb mRNA caused corresponding changes in βKlotho protein (Figure S3A). Injection of KOaP2 mice with FGF21 induced Egr1 in liver but not BAT (Figure S3B). FGF21-mediated induction of Egr1 was also reduced in WAT of KOaP2 mice (Figure S3B) coincident with the reduction of βKlotho (Figure 3A). KOaP2 mice had normal body weight and composition (Figure S3C). In insulin tolerance tests, the effect of FGF21 was significantly reduced in KOaP2 mice compared to control Klbfl/fl mice (Figure 3B). These data show that βKlotho in adipose tissue is required for at least part of the insulin-sensitizing effect of FGF21, but not the overall growth effects of βKlotho that were observed in the whole body KO (compare Figure 2A to Figure S3C).
We performed similar experiments to assess insulin action in diet-induced obese (DIO) Klbfl/fl and KOaP2 mice. DIO Klbfl/fl and KOaP2 mice had similar body weights and composition (Figure S3D). Importantly, while Klb mRNA was reduced by ~70% in WAT from lean KOaP2 mice, it was virtually absent from WAT as well as BAT of DIO KOaP2 mice (Figure 3C). Thus, the HFD improves the efficiency by which aP2-Cre disrupts βKlotho expression in WAT. Consistent with our previous findings in Fgf21-KO mice (Dutchak et al., 2012), DIO KOaP2 mice had reduced expression of Pgc1a, Acaca, Scd1 and Pck1 in WAT (Figure S3E). Notably, the insulin-sensitizing effect of FGF21 was completely lost in DIO KOaP2 mice (Figure 3D).
Complementary euglycemic-hyperinsulinemic clamp experiments were performed to further evaluate the role of βKlotho in adipose for mediating the insulin-sensitizing effect of FGF21. DIO Klbfl/fl and KOaP2 mice were intravenously injected with vehicle or FGF21 and then continuously infused with insulin in the presence or absence of FGF21. Basal and clamped glucose and insulin levels are shown in Table S1. In Klbfl/fl mice, FGF21 treatment increased the glucose infusion rate by ~5 fold (Figure 3E). Under clamp conditions, FGF21 administration significantly increased whole-body glucose uptake in Klbfl/fl but not KOaP2 mice (Figure 3F). Analysis of different tissues showed that glucose uptake was significantly increased only in BAT (Figure 3G). There were also trends towards decreased endogenous glucose production and enhanced suppression of endogenous glucose production in Klbfl/fl but not KOaP2 mice under clamp conditions (Figure 3H). Taken together, these data highlight the importance of adipose tissue in mediating the acute insulin-sensitizing actions of FGF21. While our results showing that FGF21 enhances insulin sensitivity might appear to contradict the previous finding that FGF21 enhances hepatic gluconeogenesis (Potthoff et al., 2009), we note that these previous experiments were done after chronic exposure to FGF21 and in the absence of exogenous insulin. Since insulin suppresses gluconeogenesis, the insulin sensitizing effect that we observed under the current conditions is likely due to the combination of FGF21 stimulating glucose uptake in BAT and suppressing glucose production in liver.
aP2-Cre mice also express Cre in the nervous system (Martens et al., 2010). Since we found that βKlotho is expressed in the suprachiasmatic nucleus in the hypothalamus and dorsal vagal complex in the brainstem (Figure S3G), and intracerebroventricular injection of FGF21 improves insulin sensitivity in rats (Sarruf et al., 2010), we measured Klb mRNA levels in these brain regions of KOaP2 mice. Klb expression was reduced in both the suprachiasmatic nucleus and dorsal vagal complex in KOaP2 mice (Figure S3F). To determine whether βKlotho in the nervous system contributes to the acute insulin-sensitizing actions of FGF21, we crossed Klbfl/fl and CamK2a-Cre mice, which express Cre in the nervous system (Casanova et al., 2001), to generate KOCamK2a mice. As shown in Figure S3G, CamK2a-Cre efficiently reduced Klb mRNA expression in both the suprachiasmatic nucleus and dorsal vagal complex but not in BAT, WAT or liver. Klbfl/fl and KOCamK2a mice have similar body weights and body composition either on normal chow or after HFD for three months (data not shown). In insulin tolerance tests, DIO Klbfl/fl and KOCamK2a mice showed identical response to FGF21 treatment (Figure S3H). Thus, eliminating βKlotho in the nervous system does not affect the acute insulin-sensitizing activity of FGF21. aP2-Cre mice also express Cre in macrophages (Urs et al., 2006). However, QPCR analyses showed that Klb was not expressed in the stromal vascular fraction of adipose tissue or in peritoneal, bone marrow-derived and spleen-derived macrophages (Figure S3I). These findings further support the conclusion that the insulin-sensitizing effects of FGF21 require βKlotho in adipose tissue.
In summary, we provide definitive evidence that βKlotho is essential for the diverse actions of FGF21. We note that another group, using an independently-derived βKlotho-KO mouse line, previously concluded that βKlotho is not required for at least some of the actions of FGF21 (Tomiyama et al., 2010). However, this group used unpurified, concentrated culture medium from FGF21-overexpressing CHO cells as their source of recombinant FGF21 and only examined the effect of this protein on a single parameter, namely the induction of Egr1 expression in WAT and BAT (Tomiyama et al., 2010). In contrast, using highly purified FGF21, we show that FGF21-mediated induction of Egr1 in WAT, BAT and liver requires βKlotho. Importantly, we further demonstrate that βKlotho is required for the metabolic actions of FGF21 ranging from its insulin-sensitizing and lipid-lowering actions to its inhibitory effects on growth. Finally, we demonstrate that βKlotho in adipose tissue plays an important role in mediating the acute effects of FGF21 on insulin sensitivity and glucose uptake. While we were unable to determine definitively whether these effects are mediated by FGF21 acting directly on WAT or BAT, our data suggest that both tissues are likely to be important. Two recent studies showed that the metabolic effects of FGF21 are lost in lipodystrophic mice (Veniant et al., 2012; Wu et al., 2011), providing additional and complementary evidence that FGF21 acts on adipose tissue. Future studies will be aimed at dissecting the contributions of the different types of adipose tissue to the actions of FGF21.
Fgf21-Tg, CamK2a-Cre and Meox-Cre mice were previously described (Casanova et al., 2001; Inagaki et al., 2007; Tallquist and Soriano, 2000). aP2-Cre (#005069) and FLP mice (#003800) were purchased from Jackson Labs. Mice were fed standard chow (Harlan Teklad, TD.2916) or a HFD containing 60% fat (Research Diets, D12492i) ad libitum. All experiments were performed with male mice and were approved by the Institutional Animal Care and Research Advisory Committee of the University of Texas Southwestern Medical Center.
The βKlotho-KO targeting vector was constructed as diagrammed in Figure S1A. The βKlotho gene was deleted in the germline by crossing Klbfl/+ mice with Meox-Cre mice. aP2-Cre and CamK2a-Cre mice were used to delete βKlotho in adipose tissue and the nervous system, respectively. All mice were maintained on a mixed C57Bl/6;129/Sv background.
Human FGF21 (residues 29–209) with a hexahistidine-tag on the amino terminus was expressed in E. coli, purified by sequential Ni-chelating and size-exclusion chromatography, and stored at 2-3 mg/ml in 25 mM HEPES (pH 7.5), 150 mM NaCl, 50% glycerol. For acute studies, stock FGF21 (or vehicle) was diluted to 0.1 mg/ml in saline and injected intraperitoneally at 1 mg/kg.
Body composition was measured using an EchoMRI-100 Body Composition Analyzer (Houston).
Glucose and insulin tolerance tests were performed on mice fasted for 16 h and 4 h, respectively.
Statistical analyses were performed using Student's t test (Microsoft Excel 2007). Data are presented as the mean±SEM; p < 0.05 was considered significant.
We thank Yuan Zhang, Kevin Vale, Heather Lawrence, Li Peng, and Dr. Vicky Lin for assistance with animal experiments; Yihong Wan for bone measurements and bone marrow-derived and spleen-derived macrophage cDNA samples; Michihisa Umetani for peritoneal macrophage cDNA samples; Joseph Takahashi for providing CamK2a-Cre mice; Mindy Kim for help with the clamp study; and Moosa Mohammadi and Regina Goetz for providing recombinant FGF21 protein. This work was supported by National Institutes of Health Grants DK067158, P20RR20691, and 1RL1GM084436-01 (to S.A.K. and D.J.M.), U19DK62434 (to D.J.M.), GM007062 (A.L.B.), the Robert A. Welch Foundation (Grant I-1558 to S.A.K.; Grant I-1275 to D.J.M.), and the Howard Hughes Medical Institute (to D.J.M.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Supplemental information includes Extended Experimental Procedures, three figures and one table and can be found with this article online.