FGF21 mediates the lipid metabolism response to amino acid starvation.

Lipogenic gene expression in liver is repressed in mice upon leucine deprivation. The hormone fibroblast growth factor 21 (FGF21), which is critical to the adaptive metabolic response to starvation, is also induced under amino acid deprivation. Upon leucine deprivation, we found that FGF21 is needed to repress expression of lipogenic genes in liver and white adipose tissue, and stimulate phosphorylation of hormone-sensitive lipase in white adipose tissue. The increased expression of Ucp1 in brown adipose tissue under these circumstances is also impaired in FGF21-deficient mice. Our results demonstrate the important role of FGF21 in the regulation of lipid metabolism during amino acid starvation.

Amino acid starvation initiates a signal transduction cascade, starting with the activation of the general control nonderepressible 2 (GCN2) kinase, phosphorylation of eukaryotic initiation factor 2 (eIF2), and increased synthesis of activating transcription factor 4 (ATF4) ( 1 ). We have recently found that fi broblast growth factor 21 ( Fgf21 ) is a target gene for ATF4. Consequently, FGF21 is induced by amino acid deprivation both in mouse liver and HepG2 cells as part of the transcriptional program initiated by increased levels of ATF4 ( 2 ).
FGF21 is a member of the FGF family with endocrine properties. It is predominantly produced by the liver, but is also produced by other tissues such as white adipose incubator, and cultures were replenished with fresh MEM medium and serum for approximately 16 h prior to initiating all treatments to ensure that the cells were in the basal ("fed") state. Amino acid deprivation was induced by transfer of cells to culture medium containing 2 mM histidinol (HisOH) for 8 h, which blocks charging of histidine onto the corresponding tRNA and thus mimics histidine deprivation and triggers activation of the AAR cascade ( 19 ). For mitogen-activated protein/ERK kinase (MEK) inhibition, the direct upstream kinases of ERK1/2, cells were treated with 30 M PD98059 (PD98) for approximately 16 h before harvesting. HisOH (H6647) and PD98059 (P215) were purchased from Sigma-Aldrich (St. Louis, MO). Primary BAT adipocytes were obtained and maintained as previously described ( 20 ). Mouse 3T3-L1 adipocytes were cultured and differentiated as described previously ( 21 ).
Serum measurements. Serum was obtained by centrifugation of clotted blood and stored at Ϫ 80°C. A mouse FGF21 enzymelinked immunosorbent assay (ELISA) kit was obtained from Millipore for the quantifi cation of FGF21 in mouse serum. The assay was conducted according to the manufacturer's protocol. Briefl y, a calibration curve was constructed by plotting the difference in absorbance values at 450 and 590 nm versus the FGF21 concentrations of the calibrators, and concentrations of unknown samples (performed in duplicate) were determined using this calibration curve ( 22 ). The Veterinary Service of Clinical Biochemistry at the Universitat Autònoma de Barcelona determined blood biochemical parameters.
RNA isolation and relative quantitative RT-PCR. Total RNA was extracted from the frozen tissues [liver, epididymal white adipose tissue (eWAT), and BAT] or cells (3T3L1 and primary brown adipocytes) using TRI reagent solution (Ambion) followed by DNase I treatment (Ambion) to eliminate genomic DNA contamination. To measure the relative mRNA levels, quantitative (q)RT-PCR was performed using TaqMan reagents. cDNA was synthesized from 1 g of total RNA by MLV reverse transcriptase (Invitrogen) with random hexamers (Roche Diagnostics) according to the manufacturer's instructions. TaqMan Gene Expression Master Mix and TaqMan Gene Expression assays (Invitrogen/Applied Biosystems) were used for the PCR step. Amplifi cation and detection were performed using the Step-One Plus Real-Time PCR System. Each mRNA from a single sample was measured in duplicate, using 18S rRNA as an internal control. Results were obtained by the comparative threshold cycle (Ct) method and expressed as fold of the experimental control.
Protein extract preparation. Whole-cell lysates from HepG2 cells were isolated using NP40 lysis buffer [50 mM Tris-HCl (pH 8), 150 mM NaCl, 1% Nonidet P-40]. To obtain liver nuclear extracts, frozen liver was triturated within a mortar in liquid nitrogen and immediately homogenized with a Dounce homogenizer in 1 ml of HB buffer [15 mM Tris-HCl (pH 8), 15 mM NaCl, 60 mM KCl, 0.5 mM EDTA], and centrifuged at 800 g for 5 min. The resulting pellet was resuspended in 100 l of HB buffer supplemented with 0.05% Triton X-100 (Sigma), and centrifuged for 10 min at 1,000 g . Nuclear pellets were washed with 1 ml of HB buffer supplemented with 0.05% Triton X-100 and 1 ml of HB buffer. Nuclei were incubated at 4°C for 30 min in 50 l of HB buffer containing 360 mM KCl and centrifuged for 5 min at 10,000 g . The supernatant corresponding to the nuclear extract was collected, frozen, and stored at Ϫ 80°C. To obtain liver, WAT, and BAT total extracts, tissues were homogenized in RIPA buffer and centrifuged at 12,000 g for 15 min at 4°C. The supernatant was collected and frozen at Ϫ 80°C until analysis. Protein concentration was assayed using Bio-Rad reagent. All of the buffers were The coincidence between the metabolic response to essential amino acid deprivation and to FGF21, the induction of Fgf21 under amino acid deprivation ( 2 ), together with the repression of the transcription and maturation of sterol regulatory element binding protein 1c (SREBP1c) induced by FGF21 in HepG2 cells ( 18 ), led us to consider that FGF21 could be an important mediator between amino acid deprivation and lipid metabolism in liver, WAT, and BAT.
To investigate this hypothesis, we examined the response of FGF21-defi cient mice to deprivation of the essential amino acid leucine. As expected, we found a huge increase in Fgf21 expression in the liver of wild-type (WT) animals, along with a repression of lipogenic genes after 7 days of leucine deprivation. In this condition, FGF21-deficient mice developed liver steatosis caused by unrepressed expression of lipogenic genes. In WAT, the expression of lipogenic genes was also repressed and the phosphorylation of hormone-sensitive lipase (HSL) was increased under leucine deprivation. The absence of leucine also induced an increase in the expression of Ucp1 and type 2 deiodinase ( Dio2 ) in BAT. We found that all these effects in WAT and BAT were also impaired in FGF21-defi cient mice. Here, we show the involvement of FGF21 in the regulation of lipid metabolism during amino acid starvation, thus reinforcing its important role as an endocrine factor in coordinating energy homeostasis under a variety of nutritional conditions.

Animals and diets
Male WT and Fgf21 -null mice (B6N; 129 S5-Fgf21tm1Lex/ Mmucd), obtained from the Mutant Mouse Regional Resource Center, were housed in a temperature-controlled room (22 ± 1°C) on a 12/12 h light/dark cycle and were provided free access to commercial rodent chow and tap water prior to the experiments. Control (nutritionally complete amino acid) and leucinedefi cient [( Ϫ )leu] diets were obtained from Research Diets, Inc. (New Brunswick, NJ). All diets were isocaloric and compositionally the same in terms of carbohydrate and lipid components. At the beginning of the feeding experiment, 12-15 week old male mice were fi rst acclimated to the control diet for 7 days, and then randomly assigned to either the control diet group, with continued free access to the nutritionally complete diet, or the ( Ϫ )leu diet group, with free access to the diet devoid of the essential amino acid leucine for 7 days. Food intake and body weight were recorded at least every 2 days. Animals were anesthetized by isofl urane inhalation, and blood was collected from the heart for the assay described below. After euthanization, tissues were isolated and immediately snap-frozen and stored at Ϫ 80°C for future analysis. The Animal Ethics Committee of the University of Barcelona approved these experiments.

Cell culture and treatment conditions
HepG2 cells were cultured in Eagle's Minimal Essential Medium (MEM) (GIBCO, Invitrogen) supplemented to contain 1× nonessential amino acids, 4 mM glutamine, 100 g/ml streptomycin sulfate, 100 units/ml penicillin G, and 10% (v/v) fetal bovine serum. Cells were maintained at 37°C in a 5% CO 2 /95% air Lipolysis assay. Lipolysis assay was performed after 24 h treatment with FGF21 using the Free Glycerol Determination Kit (Sigma).
Data analysis/statistics. All data are expressed as means ± SEM. Signifi cant differences were assessed by a two-tailed Student's t -test. P < 0.05 was considered statistically signifi cant. a , P < 0.05 versus control (Ctl) WT mice (or DMSO-treated HepG2 cells); b , P < 0.05 versus Ctl Fgf21 -knockout (KO) mice; c , P < 0.05 versus ( Ϫ )leu WT (or HisOH-treated HepG2 cells) (n = 6 per group of mice, or at least three independent experiments with HepG2 cells).

FGF21 gene expression is induced by leucine deprivation specifi cally in liver but not in BAT or WAT
According to our previously reported results ( 2 ), mice maintained on a ( Ϫ )leu diet show a dramatic increase in FGF21 circulating levels ( Fig. 1A ). To check the origin of this circulating FGF21 we analyzed Fgf21 gene expression in several tissues. Consistent with the liver as the main site of FGF21 production and release into the blood, Fgf21 mRNA levels in liver paralleled those in serum, whereas mRNA levels were unchanged in BAT and, unexpectedly, signifi cantly decreased in eWAT in WT mice maintained on a ( Ϫ )leu diet ( Fig. 1B ). As expected, Fgf21 mRNA levels were undetectable in any analyzed tissue in the Fgf21 -KO mice. In accordance with previous reports ( 23 ), the circulating levels of FGF21 in KO mice were below the threshold for correct quantifi cation .

FGF21 defi ciency signifi cantly attenuates weight loss under leucine deprivation
When fed a leucine-deprived diet, mice undergo rapid weight loss ( 16 ). The goal of the present study is to investigate whether this phenomenon is FGF21 dependent. For this purpose, WT and Fgf21 -KO mice were fed a control or ( Ϫ )leu diet for 7 days. Together with total body weight supplemented with a mixture of protease inhibitors (Sigma-Aldrich), 0.1 mM phenylmethylsulfonyl fl uoride (PMSF), and a phosphatase inhibitor cocktail (IPC3, Sigma-Aldrich).
Immunoblotting. Total and nuclear proteins were resolved by SDS-polyacrylamide gel electrophoresis and transferred onto a Hybond-P PVDF membrane (Millipore). Membranes were blocked for 1 h at room temperature. The blots were then incubated with primary antibody in blocking solution overnight at 4°C. Antibodies were diluted according to the manufacturer's instructions. The blots were washed three times and incubated with horseradish peroxidase-conjugated secondary antibody in blocking buffer for 2 h at room temperature. After three washes, the blots were developed using the EZ-ECL Chemiluminescence Detection Kit for HRP (Biological Industries). The quantifi cation of phosphorylation was performed by densitometry of phosphorylated protein normalized to total protein using Image J software.
Histological examinations. For the histological analysis, tissues (liver and eWAT) were fi xed in 10% formalin (Sigma-Aldrich) and embedded in paraffi n. Then, 4 m thick sections were cut and stained with hematoxylin and eosin (H and E). Images were acquired using a Leica CTR 4000 microscope. Quantitative data were obtained using the IMAT program developed in the Science and Technology Center of the University of Barcelona. The selection of the test objects was performed according to color and choosing the same limits for binari zation for all images. Adipocyte size and lipid accumulation were measured using at least three different randomly chosen fi elds of eWAT and liver sections, respectively, from each mouse. suggested that FGF21 stimulates lipolysis in WAT during normal feeding, but inhibits it during fasting ( 25 ). Therefore, to examine the role of FGF21 in lipolysis, we evaluated the mRNA levels of adipose triglyceride lipase ( Atgl ), Hsl , and perilipin 1 ( Plin1 ), and the levels of phosphorylated HSL. We did not fi nd any statistically signifi cant changes in Atgl , Hsl , or Plin1 mRNA levels upon leucine deprivation ( Fig. 3B ). Consistent with changes in body weight, lack of FGF21 signifi cantly decreased levels of phosphorylated HSL in WAT under leucine deprivation ( Fig. 3C ), which suggests that lipolysis was impaired in these mice. Despite the evidence of increased lipolysis under leucine deprivation, levels of free fatty acids in serum were not signifi cantly altered in the conditions analyzed ( Table 1 ), suggesting increased fatty acid utilization by other tissues.

FGF21 defi ciency prevents changes in liver and WAT in leucine-deprived mice
A link between FGF21 and SREBP1c during lipogenesis in cultured hepatocytes has recently been proposed ( 18 ). As lipogenic genes are downregulated in the liver of mice deprived of leucine ( 15 ), we speculated that FGF21 might regulate their expression. To investigate this possibility, we examined the expression of genes involved in the regulation of lipid metabolism in the liver loss ( Figs. 2A, B ), mice fed a leucine-deprived diet undergo a fat mass loss, both epididymal and subcutaneous ( Fig. 2D ). We found that these effects were partially blunted in Fgf21 -KO mice ( Figs. 2A, B, D ), while the reduction in food intake caused by leucine deprivation ( ‫ف‬ 30%) was unchanged between genotypes ( Fig. 2C ).
The reported observation that white adipocytes from FGF21 transgenic mice are substantially smaller than those from WT mice ( 8,24 ) and that leucine deprivation decreased adipocyte volume ( 16 ), led us to examine whether this effect was FGF21 dependent. A histological analysis of eWAT ( Fig. 3A ) showed that leucine deprivation resulted in a reduction in adipocyte volume compared with mice fed a control diet. By contrast, the adipocyte volume was only slightly reduced in ( Ϫ )leu-fed Fgf21 -KO mice and remained unchanged in Fgf21 -KO mice on the control diet ( Fig. 3A ). We note that other groups reported increased ( 25 ) or decreased ( 26 ) adipocyte size in Fgf21 -KO mice on regular diets. We have no explanation for these discrepancies beyond potential minor differences in the composition of the diet.

Increased phosphorylation of HSL under leucine deprivation is FGF21 dependent
It has been previously described that leucine deprivation increases lipolysis in WAT ( 16 ). It has also been [cluster of differentiation 36 ( Cd36 ) and fatty acid binding protein 4 ( Fabp4 )] or oxidation [carnitine palmitoyltransferase 1a ( Cpt1a )] was decreased in the absence of Fgf21 . The analysis of FASN, SREBP1c, and ACC1 protein abundance showed a good correlation with the gene expression data ( Fig. 4B ). By contrast ACC1 phosphorylation was decreased under leucine deprivation in both WT and Fgf21 -KO mice.
of WT and Fgf21 -KO mice maintained either on the control or ( Ϫ )leu diet. As expected, levels of Fasn , Srebp1c , and Acc1 mRNA were signifi cantly decreased on the ( Ϫ ) leu diet. However, in mice lacking FGF21, the reduced expression of Fasn was statistically signifi cantly increased, and, although not statistically signifi cant, showed a tendency to increase in both Srebp1c and Acc1 ( Fig. 4A ). The expression of other genes involved in fatty acid uptake  a decrease in the expression of Fasn , Srebp1c , and Acc1 and also of Atgl and Plin ( Fig. 5D ).
The FGF21-dependent phenotype during leucine deprivation is not related with the mitogen-activated protein kinase ERK1/2 signaling pathway As described, mitogen-activated protein kinase (MAPK) ERK1/2 signaling is required for the amino acid starvation response ( 27,28 ). Accordingly, histidinol treatment of HepG2 cells, which blocks charging of histidine onto the corresponding tRNA and thus mimics histidine deprivation, induced the phosphorylation of ERK1/2 ( Fig. 6A ). In addition, increased FGF21 and asparagine synthetase ( ASNS ), mRNA levels ( Fig. 6C ), as well as increased ATF4 protein levels in HisOH-treated cells ( Fig. 6B ) were reduced in the presence of the MEK inhibitor PD98. In these cells, FASN expression is opposed to that of FGF21 ( Fig. 6D ), mimicking what we observed in leucine-deprived WT (high FGF21) or Fgf21 -null mice ( Fig. 4A ).
As it has been described that exogenous FGF21 is able to induce ERK1/2 phosphorylation in the liver and WAT Although liver triglyceride levels, measured by extraction and posterior quantifi cation, did not refl ect the expression pattern of the lipid synthesis genes, H and E staining revealed it. This suggested a decreased lipid accumulation under leucine deprivation in WT animals that does not seem to occur in the Fgf21 -KO mice ( Fig. 4C ).
As expected, gene expression analysis in eWAT revealed that the mRNA levels of the lipogenic genes Fasn , Srebp1c , and Acc1 were also lower in this tissue in mice maintained on the ( Ϫ )leu diet. These changes were blunted in the Fgf21 -KO mice, particularly for Fasn ( Fig. 5A ). The analysis of FASN protein abundance showed a good correlation with the gene expression data ( Fig. 5B ).
We have analyzed the expression of the FGF21-receptor complex ␤ -Klotho and fi broblast growth factor receptor 1 (FGFR1) in WAT. There was a signifi cant increase in both ␤ -Klotho and Fgfr1 mRNA levels induced by both Fgf21 knockout and leucine deprivation ( Figs. 5C ).
To further confi rm the results obtained in mice, we have analyzed the expression of lipogenic and lipolytic genes in 3T3L1 treated with FGF21. This treatment induced BAT of WT mice . These changes were blocked in Fgf21 -KO mice ( Fig. 7A ). Because UCP1 expression is related to energy expenditure, the absence of induction of Ucp1 in Fgf21 -KO mice under leucine deprivation may contribute to the decrease in weight loss observed under these circumstances. mRNA levels of Pgc1 ␣ , which regulates the expression of UCP1 ( 32 ), were also increased. However, they did not differ between WT and Fgf21 -KO mice under either control or ( Ϫ )leu diet conditions ( Fig. 7B ). The same pattern as for Ppar ␥ and Pgc1 ␣ was observed for Adrb3 mRNA levels ( Fig. 7C ). We tested the possibility that the activation of p38 MAPK by the ␤ 3-adrenergic receptor ( ␤ 3AR) participated in the induction of the Ucp1 gene in BAT under this situation. However, although p38 phosphorylation, but not PKA activity, is induced in primary brown adipocytes treated with recombinant FGF21 (data not shown), phosphorylated p38 levels remained unchanged between diets and genotypes in BAT ( Fig. 7C ).
We have also analyzed the expression of lipogenic and lipolytic genes and glycerol release in primary brown adipocytes treated with FGF21. This treatment decreased the expression of Fasn and increased glycerol release without signifi cant changes in the expression of Atgl and Plin of acutely treated mice ( 29 ); we checked ERK1/2 phosphorylation in leucine-deprived WT and FGF21-null mice. Effectively, ERK1/2 phosphorylation was induced in the liver of leucine-deprived mice, although there were no differences between genotypes ( Fig. 6E ). Accordingly, the amino acid starvation response program was correctly initiated in Fgf21 -KO mice, as shown by the increased levels of ATF4 protein and Asns , a prototypical ATF4 target gene, mRNA levels ( Fig. 6F ). Moreover, despite the fact that FGF21 serum levels are highly increased in leucine-deprived mice ( Fig. 1A ), ERK1/2 phosphorylation is not modifi ed in WAT ( Fig. 6G ).

FGF21 defi ciency prevents increases in BAT activation in leucine-deprived mice
Thermogenesis in BAT is mediated by the upregulation of UCP1 ( 30 ). It has been proposed that the induction of FGF21 production by the liver mediates direct activation of brown fat thermogenesis during the fetal-to-neonatal transition ( 20 ). FGF21 also regulates PPAR ␥ coactivator 1 (PGC1) ␣ and browning of WAT in adaptive thermogenesis ( 31 ). Consistent with previous results ( 16 ), leucine deprivation increased levels of Ucp1 and Dio2 mRNAs in the in liver but not in BAT or WAT where they were not affected or even downregulated.
We have seen that FGF21 defi ciency signifi cantly attenuates weight loss under leucine deprivation; although the established reduction in food intake induced by a leucinedeprived diet was not changed by the absence of FGF21. This means that FGF21 is in part responsible for the loss of weight under amino acid deprivation, independently of food intake . Consistent with our observation, previous studies have shown that FGF21 transgenic mice are resistant to diet-induced obesity (DIO) and that FGF21 treatment induced weight loss in genetically obese (ob/ob) mice ( 24,33,34 ).
It has been shown that white adipocytes from FGF21 transgenic mice are substantially smaller than those from WT mice ( 8,24 ). Here, we demonstrate that the reduction ( Fig. 7E, F ). Accordingly, mRNA levels of Hsl and Atgl as well as HSL phosphorylation did not change in BAT of Fgf21 -defi cient mice ( Fig. 7B, D ).

DISCUSSION
We have previously shown that leucine deprivation signifi cantly increased FGF21 hepatic expression and serum protein levels ( 2 ), suggesting an important role of this hormone in the amino acid starvation phenotype. In the current study, we demonstrate that in response to leucine deprivation, weight loss, downregulation of liver and WAT key lipogenic genes, as well as BAT activation are partly FGF21 dependent. Our results show that the FGF21 serum levels positively correlate with the mRNA levels measured Fig. 6. Activation of the MEK/ERK pathway in the liver by leucine deprivation is independent of FGF21. HepG2 cells were incubated for 8 h with HisOH (2 mM) and the MEK inihibitor PD98 (30 M), when indicated. Phosphorylation of ERK1/2 (A) and ATF4 protein levels (B) were analyzed by Western blot in total and nuclear HepG2 extracts, respectively. mRNA levels for FGF21 , ASNS (C), and FASN (D) were analyzed by qRT-PCR. E: Phosphorylated ERK1/2 (P-ERK1/2) and total ERK1/2 (t-ERK1/2) levels were measured in liver extracts of WT and FGF21-KO mice by Western blot analysis. The right panel shows quantifi cation by densitometry of phosphorylated ERK1/2 normalized to total ERK1/2 using Image J software. F: Asns and Atf4 mRNA levels and nuclear ATF4 protein levels (insert), were measured in FGF21-KO mouse liver by qRT-PCR and Western blot, respectively. Actin was used as a loading control. A representative blot is shown. G: Phosphorylated and total ERK1/2 levels were measured in WT mouse eWAT homogenates by Western blot analysis. The right panel shows quantifi cation by densitometry of phosphorylated ERK1/2 normalized to total ERK1/2 using Image J software. Error bars represent the mean ± SEM. a , P < 0.05 versus Ctl WT mice; b , P < 0.05 versus Ctl FGF21-KO mice; c , P < 0.05 versus ( Ϫ )leu WT mice (n = 6/group). Fig. 7. FGF21 is required for inducing BAT activation during amino acid deprivation. A: Ucp1 and Dio2 gene expression was measured by qRT-PCR in mouse BAT. B: The expression of genes related with lipid metabolism was measured by qRT-PCR in mouse BAT. C: Phosphorylated p38 MAPK (P-p38 MAPK ) and total p38 MAPK (t-p38 MAPK) levels were measured in WT and FGF21-KO mouse liver extracts by Western blot analysis. The right panel shows quantifi cation by densitometry of phosphorylated p38 MAPK normalized to total p38 MAPK using Image J software. D: Phosphorylated and total HSL protein levels were measured in WT and FGF21-KO eWAT homogenates by Western blot analysis. The right panel shows quantifi cation by densitometry of phosphorylated HSL normalized to total HSL, using Image J software. A representative blot is shown. E: Fasn, Srebp1c, Atgl and Plin1 gene expression was measured by qRT-PCR in differentiated primary BAT treated with FGF21 (50nM) for 24h. F: Glycerol release in differentiated primary BAT treated with FGF21 (5nM and 50nM) for 24h. Error bars represent the mean ± standard error of the mean (SEM). Error bars represent the mean ± SEM. A representative blot is shown. a , P < 0.05 versus Ctl WT mice; b , P < 0.05 versus Ctl FGF21-KO mice; c , P < 0.05 versus ( Ϫ )leu WT mice (n = 6/group). its origin and other factors, which may refl ect the metabolic state and the energy requirements of the organism.
An inhibitory action in the lipid synthesis had been already described for FGF21 ( 36 ). It was demonstrated that the reduction of hepatic triglyceride levels was associated with FGF21 inhibition of nuclear sterol regulatory element binding protein 1c ( Srebp1c ) and the expression of an extended array of genes involved in fatty acid and triglyceride synthesis. Accordingly, in liver, we have seen that the expected reduction in the mRNA levels of Srebp1c , Fasn , and Acc1 by ( Ϫ )leu diet was not observed in Fgf21 -KO mice. We have also seen a direct effect in lipogenic gene expression in both 3T3L1 cells and primary brown adipocytes treated with recombinant FGF21.
The induction, in liver from Fgf21 -KO mice, of Asns expression, for which the gene product catalyzes the glutamine and ATP-dependent conversion of aspartic acid to asparagine, suggests that FGF21 is not involved in the control of amino acid metabolism under amino acid starvation. We have also observed that contrary to the exogenous FGF21 induction of ERK1/2 phosphorylation in the liver and WAT of acutely treated mice ( 29 ), the FGF21-dependent phenotype in leucine deprivation is unexpectedly not related with the MAPK ERK1/2 signaling pathway both in liver and WAT.
BAT is a major site of adaptive thermogenesis, and it is used to preserve both thermal and caloric homeostasis in response to environmental temperature or diet ( 37 ). Fgf21 -KO mice under leucine deprivation exhibited decreased induction of genes defi ning BAT identity (i.e., Ucp1 and Dio2 ), while its transcriptional regulators Pgc1 ␣ and Ppar ␥ were identically induced in both WT and Fgf21 -KO mice. Increased Ucp1 expression may be regulated by the sympathetic nervous system through the activation of ␤ -adrenergic receptors. We found that the mRNA expression of Adrb3 was induced by leucine deprivation in BAT, although there were no signifi cant changes between genotypes. The ␤ 3AR stimulates p38 MAPK, which is required for the ␤ AR-dependent increase in Ucp1 expression in brown adipocytes ( 38 ). Nevertheless, p38 phosphorylation levels were not affected by leucine deprivation compared with control, both in the WT in the adipocyte volume that occurs under leucine deprivation depends on the increased FGF21 expression and secretion that takes place under this diet.
We have found that the levels of phosphorylated (P)-HSL were increased upon ( Ϫ )leu feeding in eWAT, as it was described in ( 16 ), so activation of the HSL protein probably allowed increased lipolysis. Interestingly, our results show that lack of FGF21 signifi cantly decreases phosphorylated HSL levels in the eWAT of ( Ϫ )leu-fed mice, suggesting an important role of FGF21 action on leucine deprivation-induced lipolysis.
The analysis of blood biochemical parameters, however, did not show signifi cant changes in NEFA levels, likely due to increased fatty acid utilization by other tissues. We assume that there is an increased glucagon signaling under leucine deprivation, indicated by the increased PKA-dependent HSL Ser660 phosphorylation in WAT. Additionally, a reduction in insulin levels under leucine deprivation is observed in WT but not in KO mice; these changes are well correlated with the differences observed in body weight and fat mass.
We have also examined whether the impaired reduction in body weight observed in Fgf21 -KO mice under leucine deprivation was not only related to lipolysis in WAT, but also to other factors that infl uence adipose tissue mass as lipogenesis. We have observed a signifi cant reduction in Fasn mRNA levels in eWAT upon ( Ϫ )leu feeding that was totally blocked in Fgf21 -KO mice. Srebp1c and Acc1 mRNA levels presented the same pattern, although with distinct statistical signifi cances. The endocrine effect of FGF21 on WAT under leucine deprivation that we observe here is different from the recently described autocrine effect on this tissue in a fed state, in which FGF21 induces lipogenesis through the regulation of PPAR ␥ activity ( 26 ). The decreased Fgf21 expression in WAT under leucine deprivation also contrasts with its observed induction after one day of fasting ( 35 ). FGF21 administration to DIO mice leads to a dramatic decrease in the WAT Fgf21 transcript. However, controversially, it induces an increase in the expression of adipogenic genes ( 33 ). It seems, therefore, that the response to increased levels of FGF21 depends on and Fgf21 -KO mice. One of the best-known inducers of BAT and its function is norepinephrine ( 39 ), which has also been shown to be induced in leucine-deprived mice ( 16 ). These fi ndings raise the possibility that FGF21 might induce Ucp1 (and also Dio2 ) through an indirect mechanism involving the central nervous system. Of interest, it has been recently proposed that increased expression of FGF21 secreted from liver enters the brain and stimulates the hypothalamic-pituitary-adrenal axis ( 40 ). However, a more extensive analysis of other candidate factors should be performed in the future, and further investigation will be required to determine the exact mechanism by which FGF21 induces BAT activation.
In summary, we found that FGF21 is an important factor, although not the only one, in mediating the changes in lipid metabolism observed upon leucine deprivation ( Fig. 8 ). We have shown that Fgf21 -defi cient mice under these circumstances showed unrepressed lipogenesis in liver and WAT, decreased phosphorylation of HSL in WAT indicating impaired lipolysis, and impaired induction of Ucp1 expression in BAT. Thus, our results suggest that FGF21 plays an important role in the regulation of lipid metabolism during amino acid starvation.
While this work was under revision, a study was published ( 41 ) that showed the Atf4-dependent induction of FGF21 in mice with autophagy defi ciency in skeletal muscle or liver. As a result of this induction, these mice are protected from diet-induced obesity and insulin resistance. In the supplementary results of this paper, the authors confi rm that the ATF4-FGF21 axis also has a physiologically relevant role under conditions of leucine deprivation. In agreement with our results, they also show how the effects of leucine deprivation on body weight, fat weight, and blood glucose are partially diminished in Fgf21 -KO mice.