Qingning Liang,1,2 Ling Zhong,1,2 Jialiang Zhang,1,2 Yu Wang,1,3 Stefan R. Bornstein,4 Chris R. Triggle,5

Hong Ding,5 Karen S.L. Lam,1,2 and Aimin Xu1,2,3

FGF21 Maintains GlucoseHomeostasis by Mediating theCross Talk Between Liver andBrain During Prolonged FastingDiabetes 2014;63:4064–4075 | DOI: 10.2337/db14-0541

Hepaticgluconeogenesis isamain sourceofbloodglucoseduring prolonged fasting and is orchestrated by endocrineand neural pathways. Here we show that the hepatocyte-secreted hormone fibroblast growth factor 21 (FGF21)induces fasting gluconeogenesis via the brain-liver axis.Prolonged fasting induces activation of the transcrip-tion factor peroxisome proliferator–activated receptora (PPARa) in the liver and subsequent hepatic produc-tion of FGF21, which enters into the brain to activate thehypothalamic-pituitary-adrenal (HPA) axis for release ofcorticosterone, thereby stimulating hepatic gluconeogene-sis. Fasted FGF21 knockout (KO) mice exhibit severehypoglycemia and defective hepatic gluconeogenesis dueto impaired activation of the HPA axis and blunted releaseof corticosterone, a phenotype similar to that observedin PPARa KO mice. By contrast, intracerebroventricularinjection of FGF21 reverses fasting hypoglycemia and im-pairment in hepatic gluconeogenesis by restoring cortico-sterone production in both FGF21KOandPPARaKOmice,whereas all these central effects of FGF21 were abrogatedby blockage of hypothalamic FGF receptor-1. FGF21 actsdirectly on the hypothalamic neurons to activate themitogen-activated protein kinase extracellular signal–relatedkinase1/2 (ERK1/2), thereby stimulating theexpres-sionof corticotropin-releasinghormonebyactivationof thetranscription factor cAMP response element bindingprotein. Therefore, FGF21 maintains glucose homeostasisduringprolonged fastingbyfine tuning the interorgancrosstalk between liver and brain.

Hepatic gluconeogenesis is tightly controlled by counter-regulatory hormones such as glucagon, cortisol, and insulin,via regulating the expression of key gluconeogenic enzymes,including glucose 6 phosphatase (G6Pase) and phospho-enolpyruvate carboxykinase (PEPCK). Fibroblast growthfactor 21 (FGF21), a metabolic regulator mainly secretedfrom the liver in response to fasting and starvation underthe control of the nuclear receptor peroxisome proliferator–activated receptor a (PPARa), plays a critical role in main-taining energy homeostasis and insulin sensitivity in bothrodents and nonhuman primates (1–6). A therapeutic doseof FGF21 decreased blood glucose in diabetic animals with-out causing hypoglycemia (4). FGF21 has also been shownto act as a key downstream effector of PPARa, mediatingseveral metabolic adaptation responses to starvation, in-cluding hepatic fatty acid oxidation, ketogenesis, andgrowth hormone resistance (1,2,7). In addition, FGF21is implicated in hepatic gluconeogenesis, although itremains controversial whether hepatocytes are a directaction site of FGF21 (8,9). There is an obvious dichotomybetween the effects of endogenous FGF21 and pharmaco-logical actions of the recombinant peptide with respect tohepatic metabolism (4,6,9).

FGF21 can cross the blood-brain barrier (10) and isdetectable in both human and rodent cerebrospinal fluid(10,11). Continuous intracerebroventricular injection ofFGF21 into obese rats increases energy expenditure and

1State Key Laboratory of Pharmaceutical Biotechnology, The University of HongKong, Hong Kong, China2Department of Medicine, The University of Hong Kong, Hong Kong, China3Department of Pharmacology and Pharmacy, The University of Hong Kong, HongKong, China4Department of Medicine, University of Dresden, Dresden, Germany5Department of Pharmacology, Weill Cornell Medical College in Qatar, Doha, Qatar

Corresponding author: Aimin Xu, amxu@hku.hk, or Karen S.L. Lam, ksllam@hku.hk.

Received 2 April 2014 and accepted 8 July 2014.

This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db14-0541/-/DC1.

© 2014 by the American Diabetes Association. Readers may use this article aslong as the work is properly cited, the use is educational and not for profit, andthe work is not altered.

See accompanying article, p. 4013.

4064 Diabetes Volume 63, December 2014

METABOLISM

insulin sensitivity (12). More recently, FGF21 has beenshown to act on the central nervous system to increasesystemic glucocorticoid levels, suppress physical activity,and alter circadian behavior (13). Furthermore, FGF21acts on the hypothalamus to suppress the vasopressin-kisspeptin signaling cascade, thereby mediating starvation-induced infertility of female mice (14). However, thephysiological roles of FGF21 and its central actions in reg-ulating glucose metabolism during adaptive starvationresponses remain unknown.

In this study, we show that FGF21 is a key metabolicregulator essential for maintaining glucose homeostasis,by sending the starvation signal from the liver to thebrain, where it acts on the hypothalamic neurons to inducephosphorylation of the mitogen-activated protein (MAP)kinase extracellular signal–related kinase 1/2 (ERK1/2),which in turn stimulates the production of corticotropin-releasing hormone (CRH) and subsequent release of adre-nal corticosterone by activation of cAMP response elementbinding protein (CREB), thereby leading to enhanced he-patic gluconeogenesis.

RESEARCH DESIGN AND METHODS

Animal StudiesAll animal experimental protocols were approved by theAnimal Ethics Committee of The University of HongKong. PPARa knockout (KO) mice in C57BL/6N back-ground were originally obtained from The Jackson Labo-ratory (Sacramento, CA). FGF21 KO mice in C57BL/6Jbackground were described previously (15). The micewere housed in a room under controlled temperature(23 6 1°C) with free access to water and standardchow. Male mice at the age of 12–14 weeks were usedin all animal experiments. Mice were studied by hyper-insulinemic-euglycemic clamp to assess endogenous glu-cose production as previously described (16), exceptthat mice were fasted 24 h before the clamp and so-matostatin was not infused. Bilateral or sham adrenal-ectomy was conducted under isoflurane anesthesia 1 weekbefore various fasting experiments. Plasma corticoste-rone levels were measured to confirm the completenessof adrenalectomy.

Intracerebroventricular and IntraparaventricularNucleus Injection in MiceMale FGF21 KO and PPARa KO mice and their respectivewild-type (WT) littermates at the age of 12–14 weekswere chronically implanted with 26G stainless cannula(Plastics One, Inc., Roanoke, VA) in the cerebral ventricleor the paraventricular nucleus (PVN) region of the brainas previously described (17). A neutralizing monoclonalantibody against FGF receptor 1 (FGFR1, 0.2 mg/g bodyweight; NBP2–12308, Novus) or CRH antagonist (a-helicalCRH9–41, 0.2 mg/g body weight; C2917, Sigma-Aldrich)was infused by intra-PVN injection before central admin-istration of endotoxin-free recombinant mouse FGF21(rmFGF21, 0.02 mg/g body weight) (18).

Biochemical and Immunological AnalysisSerum levels of insulin and FGF21 were quantified withimmunoassays according to the manufacturer’s instruc-tions (Antibody and Immunoassay Services, The Univer-sity of Hong Kong). Mouse FGF21 assay was based on anaffinity-purified rabbit polyclonal antibody specific tomouse FGF21 and did not cross-react with other membersof the FGF family. The intra- and interassay variationswere 4.2 and 7.6%, respectively. Mouse insulin assaywas based on a monoclonal antibody pair raised usingrecombinant mouse insulin as an antigen, with intra-and interassay variations of 5.3 and 6.2%, respectively.Serum levels of corticosterone (Enzo Life Sciences) andglucagon (Millipore) and plasma levels of adrenocortico-tropic hormone (ACTH; MD Bioproducts) and CRH(Phoenix Pharmaceuticals) were determined using immu-noassays according to the manufacturer’s instructions.For analysis of adrenaline and noradrenaline in the liver,frozen tissue was extracted with 0.1 N HCl and analysesperformed according to the supplier’s instructions (LaborDiagnostika Nord, Nordhorn, Germany).

Ex Vivo StudiesThe livers of male 12-week-old C57BL/6 mice were iso-lated, and primary hepatocytes were prepared from maleWistar rats (200 g) and used to determine glucose pro-duction as previously described (19). The mouse hypotha-lamic explants were prepared as previously described (20).The hypothalamic slices were pretreated without or withPD98059 for 30 min, followed by treatment withrmFGF21 for different periods.

RNA Extraction and Real-Time PCRTotal RNA was extracted from mouse tissues and rathepatocytes, cDNA was synthesized, and the mRNAexpression levels were quantified with real-time PCR aspreviously described (21).

Western BlotTotal cellular proteins in liver and hypothalamic lysateswere separated by SDS-PAGE and probed with a rabbitpolyclonal antibody against peroxisome proliferator–activated receptor-g coactivator 1a (PGC-1a; Abcam),FGF21 (18), phosphorylated ERK (p-ERK), total ERK(t-ERK), p-CREB, t-CREB (Cell Signaling Technology),FGFR1, bKlotho (Santa Cruz), or b-actin (Sigma-Aldrich).

ImmunohistochemistryParaformaldehyde-fixed brains were frozen in liquidnitrogen and stored at280°C until sectioning with a cryo-stat (CM1950, Leica) at 220°C. Slides were washed andthen blocked in 10% goat serum with 3% BSA in TBS for1 h at room temperature, followed by incubation withprimary antibody overnight at 4°C. On the 2nd day, theslides were further incubated with Alexa Fluor 488 or594–labeled mouse-specific secondary antibodies (Invitro-gen) for another hour at room temperature. Images werecaptured with a fluorescence microscope (IX71; Olympus;QImaging).

diabetes.diabetesjournals.org Liang and Associates 4065

Statistical AnalysisAll analyses were performed with Statistical Package forSocial Sciences version 11.5 for Windows (SPSS, Chi-cago, IL). Comparison between groups was performedusing one-way ANOVA followed by least-significantdifference post hoc test. In all statistical comparisons,a P value ,0.05 was used to indicate a statistically sig-nificant difference.

RESULTS

FGF21 KO Mice Exhibit Hypoglycemia DuringProlonged FastingBlood glucose levels were comparable between FGF21KO and WT mice in the fed state and within 6 h of fooddeprivation (Fig. 1A). However, during the prolongedfasting period (24–48 h), FGF21 KO mice exhibiteda much more rapid decline in blood glucose levels andmore severe hypoglycemia as compared with WT mice(Fig. 1A). Moreover, both the mRNA and protein expres-sion levels of hepatic PGC-1a, a transcriptional coactiva-tor that plays a key role in mediating the fasting-inducedexpression of gluconeogenic genes (22), were comparablebetween FGF21 KO mice and WT littermates in the fedstate (Fig. 1A and B). However, the amplitude of fasting-induced elevation in PGC-1a expression was significantlydecreased as compared with WT mice, and this changewas accompanied by a marked impairment in the fasting-induced expression of G6Pase and PEPCK (Fig. 1Aand B). The rate of glucose production in FGF21 KOmice, as estimated by pyruvate tolerance test, wasmuch lower than that in WT littermates in fasted state(Fig. 1C). We further assessed whole-body glucose ho-meostasis in euglycemic glucose clamp studies. Duringconstant hyperinsulinemia in mice after fasting, a higherglucose infusion rate was required to maintain a normalglucose level in KO mice than in WT mice (Fig. 1D).This difference was primarily accounted for by a signifi-cant reduction in glucose production in KO mice, al-though whole-body glucose disposal was similar in WTand KO mice (Fig. 1D). Treatment of mouse liverexplants with rmFGF21 for up to 24 h had no obviouseffect on the expression of gluconeogenic genes (PGC-1a,PEPCK, and G6Pase) and glucose production, althoughthe glucocorticoid receptor agonist dexamethasone ledto a significant induction of the gluconeogenic program(Supplementary Fig. 1A and B). Similarly, dexametha-sone, but not FGF21, increased the expression of thegluconeogenic genes and glucose production in primaryrat hepatocytes (Supplementary Fig. 1C and D). Further-more, there was no difference in circulating levels ofinsulin and glucagon nor the levels of hepatic adrenalineand noradrenaline between FGF21 KO mice and WT lit-termates in either the fed or fasted state (SupplementaryFig. 2A–D), suggesting that insulin, glucagon, and he-patic sympathetic nerves are not the downstream medi-ators of FGF21 in the maintenance of fasting glucosehomeostasis.

FGF21 Deficiency Impairs Fasting-Induced Activationof the Hypothalamic-Pituitary-Adrenal Axis andRelease of Corticosterone in MiceGlucocorticoids are fasting-inducible hormones that playa central role in activating gluconeogenic genes andsustaining glucose homeostasis during starvation. Ourrecent clinical study showed that FGF21 and cortisoldisplay a similar circadian rhythm in humans (23). There-fore, we next investigated the functional relationship be-tween FGF21, corticosterone, and its upstream regulatorsin mice. In WT mice, serum levels of corticosterone wereprogressively increased in response to fasting (Fig. 1E).However, the magnitude of fasting-induced elevation ofserum corticosterone in FGF21 KO mice was muchsmaller than WT mice (Fig. 1E). Likewise, fasting for24 h caused a marked induction in hypothalamic expres-sion of CRH (Fig. 1F) and pituitary release of ACTH (Fig.1H), whereas such fasting-induced expression of thesetwo corticotropic hormones was severely impaired inFGF21 KO mice (Fig. 1F and H), suggesting that FGF21deficiency causes defective fasting-induced release of cor-ticosterone possibly by impairing the activation of thehypothalamic-pituitary-adrenal (HPA) axis. On the con-trary, the expression of arginine vasopressin (AVP) wascomparable between WT and FGF21 KO mice under bothfed and fasted states (Fig. 1G), suggesting that these hor-mones do not contribute to the hypoglycemia phenotypesin FGF21 KO mice. Moreover, the difference in fastingblood glucose between WT and KO mice totally disap-peared after the production of corticosterone was blockedby adrenalectomy in mice (Fig. 1I), suggesting that corti-costerone mediates the effects of FGF21 in regulatingfasting glucose homeostasis.

We next investigated whether the fasting hypoglyce-mia phenotype in FGF21 KO mice can be reversed bysupplementation with corticosterone. Injection of thesynthetic glucocorticoid dexamethasone into FGF21 KOmice reversed fasting hypoglycemia to a level comparableto WT mice (Supplementary Fig. 3A). The dexamethasone-mediated normalization of fasting glucose homeostasiswas associated with significantly elevated expression ofPGC-1a and its target genes PEPCK and G6Pase (Supple-mentary Fig. 3B–D).

FGF21 Acts in the Brain to Induce CorticosteroneProduction and Hepatic GluconeogenesisConsistent with a previous report (10), we found thatFGF21 was present in the hypothalamus and its hypotha-lamic level was markedly elevated in response to fasting(Fig. 2A), despite that there was not any detectable FGF21mRNA expression in this tissue even in the fasted state(Fig. 2A). Furthermore, intraperitoneal injection of exog-enous rmFGF21 led to its marked accumulation in thehypothalamus region (Fig. 2B), confirming that FGF21can cross the blood-brain barrier into the hypothalamus.

To investigate whether FGF21 induces fasting gluco-neogenesis through its central actions, rmFGF21 wasinfused directly into the cerebroventricle of FGF21 KO

4066 FGF21 Mediates Cross Talk of Liver and Brain Diabetes Volume 63, December 2014

Figure 1—Fasting-induced gluconeogenesis and activation of the HPA axis are impaired in FGF21 KO mice. A: Blood glucose levels andthe mRNA abundance of hepatic gluconeogenic genes PGC-1a, PEPCK, and G6Pase in FGF21 KO mice and WT littermates in either fedstate or after fasting. B: The protein abundance of PGC-1a was determined by Western blot analysis and quantified using scanningdensitometry. The results were expressed as fold changes relative to internal control b-actin. C: Pyruvate tolerance test was performedin 24 h–fasted FGF21 KOmice and WT littermate mice by using 2 g pyruvate per kg body weight. D: Glucose infusion rate, glucose disposalrate, and glucose production during a hyperinsulinemic-euglycemic clamp experiment. E–H: The amount of hormones of HPA axis inFGF21 KO mice and WT littermates either in fed or fasted state; the serum levels of corticosterone (E) were determined by ELISA, themRNA abundance of CRH (F ) and AVP (G) in the hypothalamus was determined by qRT-PCR, and plasma ACTH levels (H) were determinedby ELISA. I: Blood glucose in FGF21 KO mice and WT littermates receiving adrenalectomy (ADX) or sham surgery in either fed state orvarious time points after fasting. n = 8 for WT mice and n = 10 for KO mice. *P < 0.05; **P < 0.01. Data were expressed as mean 6 SEM.

diabetes.diabetesjournals.org Liang and Associates 4067

Figure 2—FGF21 acts in the brain to activate HPA axis and induce hepatic gluconeogenesis. A: Protein levels of FGF21 in the hypothal-amus and mRNA abundance of FGF21 in the livers and hypothalamus of C57BL/6J mice in fed state or fasted for 24 h. B: FGF21 proteinlevels in the hypothalamus, which was collected at 30 min after intraperitoneal injection of rmFGF21 (1 mg/kg) or vehicle in FGF21 KO mice.The mice were perfused with 13 PBS by transcardiac perfusion before the collection of hypothalamic tissues. Representative Westernblotting analysis and qRT-PCR assays from two independent cohorts run in duplicate are shown (n = 4). C–G: FGF21 KO mice were treatedwith rmFGF21 (0.02 mg/g, i.c.v.) or vehicle. C: Blood glucose levels and the mRNA abundance of hepatic gluconeogenic genes PGC-1a,PEPCK, and G6Pase in the livers. CRH (D) and AVP (E) mRNA in the hypothalamus, plasma levels of ACTH (F ), and serum levels ofcorticosterone (G) were determined at 3 h after intracerebroventricular injection. *P < 0.05 and **P < 0.01 vs. vehicle control (n = 6). Datawere expressed as mean 6 SEM. H: The hypothalamic slices from FGF21 KO mice were incubated with different concentrations ofrmFGF21 for 2 h. The culture medium was collected for determination of CRH by ELISA. *P < 0.05 and **P < 0.01 vs. vehicle control(n = 6). Data were expressed as mean 6 SEM. I–J: FGF21 KO mice were pretreated with CRH antagonist a-helical CRH9–41 (anti-CRH, 0.2mg/g, intra-PVN) for 30 min, followed by injection with rmFGF21 (0.02 mg/g). Serum levels of blood glucose (I) and corticosterone (J) weredetected at 3 h after injection. *P< 0.05; n = 6. Data were expressed as mean6 SEM. K–L: Twelve-week-old FGF21 KO mice were pretreatedwith RU486 by intraperitoneal injection (20 mg/kg) for 1 h, followed by intracerebroventricular infusion of rmFGF21 (0.02 mg/g) or vehicle for 3 h.K: Blood glucose levels were determined at various time points after treatment. *P < 0.05 and **P < 0.01 vs. vehicle. #P < 0.05 vs. rmFGF21group. n = 6. Data were expressed as mean 6 SEM. L: The mRNA abundance of hepatic PGC-1a, PEPCK, and G6Pase was determined byqRT-PCR. *P< 0.05 and **P< 0.01 (n = 6). Data were expressed as mean6 SEM.M: FGF21 KOmice receiving adrenalectomy (ADX) or shamsurgery were intracerebroventricularly infused either with rmFGF21 or vehicle. Blood glucose levels were detected at various time points afterinjection. **P < 0.01 vs. vehicle + sham surgery; #P < 0.05 vs. rmFGF21 + sham surgery. n = 6. Data were expressed as mean 6 SEM.

4068 FGF21 Mediates Cross Talk of Liver and Brain Diabetes Volume 63, December 2014

mice by intracerebroventricular injection via a cannulapreimplanted 1 week before the treatment. The resultsshowed that the central infusion of rmFGF21 at 0.02 mg/gbody weight, a dose at which there was still no detectableFGF21 in the peripheral blood of FGF21 KO mice, led toa significant elevation of blood glucose levels as comparedwith the vehicle-treated mice (Fig. 2C). The central admin-istration of rmFGF21 also induced the expression of he-patic PGC-1a, PEPCK, and G6Pase (Fig. 2C) and increasedthe hypothalamic level of CRH (Fig. 2D) and pituitaryrelease of ACTH (Fig. 2F) and serum corticosterone (Fig.2G) but had no effect on hypothalamic expression of AVP

(Fig. 2E). Furthermore, direct incubation of mouse hypo-thalamic slices with rmFGF21 was sufficient to induce theCRH expression in a dose-dependent manner (Fig. 2H),whereas the central effects of FGF21 on stimulation ofcorticosterone release and elevation of blood glucose lev-els were blunted by intra-PVN administration of the CRHantagonist a-helical CRH9–41 (Fig. 2I and J). On the con-trary, intravenous injection of a low dose of rmFGF21(0.02 mg/g), which was insufficient to cause any notableelevation of FGF21 in the brain of FGF21 KO mice, hadno obvious effect on either blood glucose level or theexpression of the gluconeogenic genes in the liver or the

Figure 3—FGFR1 and bKlotho are expressed in the hypothalamus of mice. The abundance of FGFR1 mRNA and protein (A) and bKlothomRNA and protein (C) in liver, white adipose tissue (WAT), and hypothalamus (hypo) in 12-week-old male C57BL/6J mice, as determined byreal-time PCR and Western blot analysis, respectively. Representative images of mouse hypothalamic slices immunostained for FGFR1 (B),bKlotho (D), and CRH (B). Note that the merged image shows the colocalization between FGFR1 and CRH (B), and the specificity of anti-CRH antibody (catalog no. ab11133, Abcam) was validated by using the CRH blocking peptide (Sigma-Aldrich, catalog no. C3042) toneutralize the antibody. No positive fluorescent signal was detected when the hypothalamic slices were coincubated with anti-CRHantibody and blocking peptides (data not shown). Representative Western blotting analysis and immunostained figures from threeindependent cohorts are shown; n = 6.

diabetes.diabetesjournals.org Liang and Associates 4069

hormones of the HPA axis (CRH, ACTH, and corticoste-rone) (Supplementary Fig. 4A–E).

Central injection of rmFGF21 did not increase thelevels of noradrenaline or adrenaline in the liver (Supple-mentary Fig. 5A and B), suggesting that sympatheticnerves are not involved in FGF21-mediated hepatic glu-cose production. Plasma ketone bodies and hepatic expres-sion of key ketogenic genes and several genes governinglipid metabolism in the liver were not altered by cen-tral treatment of rmFGF21 (Supplementary Fig. 5C andD), indicating that the effects of FGF21 on ketogenesisand lipid metabolism are not attributed to its centralactions.

To further confirm the role of the HPA axis in FGF21-mediated gluconeogenesis, FGF21 KOmice were pretreatedwith the glucocorticoid receptor antagonist RU486 orsubjected to adrenalectomy prior to intracerebroventricularinjection of rmFGF21. The results showed that the centraleffects of rmFGF21 on elevation of blood glucose andinduction of the hepatic gluconeogenic genes were largelyabrogated by either pharmacological blockage of theglucocorticoid receptor or depletion of endogenous corti-costerone by adrenalectomy (Fig. 2K–M).

Central Effects of FGF21 on Activation of the HPA Axisand Hepatic Gluconeogenesis Are Mediated byHypothalamic FGFR1The metabolic actions of FGF21 are mediated by FGFR1in complex with bKlotho (24). FGFR1 is highly expressedin hypothalamic areas such as PVN and supraoptic nuclei(25). Both real-time PCR and Western blot analysisshowed that the expression level of FGFR1 in the hypo-thalamus was comparable to that in white adipose tissue,a major peripheral target of FGF21 (Fig. 3A). FGFR1 isabundantly expressed in the PVN of the hypothalamus(Fig. 3B), and a large portion of FGFR1 colocalized withCRH neurons (Fig. 3B). In addition, both mRNA and pro-tein expressions of bKlotho were detectable in the hypo-thalamus, although its expression level was much lowerthan that in adipose tissues and livers (Fig. 3C). Specifi-cally, bKlotho was detectable in the PVN of the hypothal-amus (Fig. 3D).

To interrogate the roles of hypothalamic FGFR1 inFGF21-induced hepatic gluconeogenisis via the HPA axis,we performed bilateral intra-PVN injection to directlydeliver the neutralizing antibody into PVN. The intra-PVN injection of anti-FGFR1 antibody largely blockedrmFGF21-induced activation of ERK1/2 in the hypothal-amus (Fig. 4). The central actions of rmFGF21 on ele-vation of blood glucose and induction of hepaticgluconeogenic gene expression were abrogated by a singleintra-PVN administration of a neutralizing monoclonalantibody against FGFR1 (Fig. 5A and B). Furthermore,blockage of hypothalamic FGFR1 by its neutralizing anti-body also negated the stimulatory effects of centraladministration with rmFGF21 on hypothalamic CRHexpression and serum levels of ACTH and corticosterone

(Fig. 5C–F). On the other hand, intra-PVN administrationof the anti-FGFR1 neutralizing antibody had no obviouseffect on glucose levels, hepatic gluconeogenic gene ex-pression, and the HPA activity in FGF21 KO mice withoutFGF21 treatment (Fig. 5A–F). Physiologically, a single intra-PVN administration of the anti-FGFR1 antibody loweredfasting blood glucose and reduced the expression of hepaticgluconeogenic genes in WT mice to a level comparable tothose in KO mice (Fig. 5G and H). Furthermore, fasting-induced elevations in hypothalamic CRH expression andserum levels of corticosterone were significantly dimin-ished by the neutralizing antibody-mediated blockage ofhypothalamic FGFR1 (Fig. 5I and J).

FGF21 Mediates PPARa-Induced CorticosteroneProduction and Hepatic Gluconeogenesis in Mice viaCentral Nervous SystemDuring fasting, the transcription factor PPARa is acti-vated in the liver, which in turn promotes hepatic ex-pression of FGF21 and its release into the circulation(1–3). Therefore, we investigated whether PPARa activa-tion induces corticosterone production by induction ofFGF21. Chronic treatment of WT mice with the PPARaagonist fenofibrate led to a progressive elevation in serumlevels of FGF21 as well as corticosterone (Fig. 6A and B).However, the magnitude of the fenofibrate-induced in-crease of serum corticosterone was significantly attenuatedin FGF21 KO mice (Fig. 6B). Likewise, fenofibrate-inducedhypothalamic expression of CRH was also impaired inFGF21 KO mice as compared with WT controls (Fig. 6C).Furthermore, fasting-induced elevation of serum cortico-sterone levels and hypothalamic expression of CRH inPPARa KO mice were markedly reduced as comparedwith WT littermates (Fig. 6D and E). On the other hand,the impairments in the fasting-induced production of cor-ticosterone and CRH were partially reversed by intracere-broventricular injection of rmFGF21 (Fig. 6D and E).

Figure 4—FGF21 signaling in hypothalamus is blocked by FGFR1neutralizing antibody. Twelve-week-old FGF21 KOmice were treatedwith an anti-FGFR1 monoclonal antibody (AF, 0.2 mg/g body weight)or mouse nonimmune IgG (NI) by a single hypothalamic paraventric-ular (intra-PVN) injection, followed by another intra-PVN injection withrmFGF21 (0.02 mg/g body weight) or vehicle control. Hypothalamicprotein levels of p-ERK and t-ERK were determined by Westernblotting analysis after 30 min of injection and were quantified usingscanning densitometry. Representative Western blotting analysisfrom three independent cohorts run in duplicate are shown; n = 6.*P < 0.05. Data were expressed as mean 6 SEM.

4070 FGF21 Mediates Cross Talk of Liver and Brain Diabetes Volume 63, December 2014

Consistent with previous reports (26), we found thatPPARa KO mice exhibited similar blood glucose levelswith WT controls in the fed state but developed severehypoglycemia after 16 h of fasting (Fig. 6F). The hypogly-cemic phenotype of PPARa KO mice further deterioratedafter prolonged fasting (24 and 48 h). On the other hand,intracerebroventricular administration of rmFGF21 into

PPARa KO mice not only led to a significant alleviation offasting hypoglycemia but also reversed the impairment offasting-induced expression of several gluconeogenic genes(PGC-1a, PEPCK, and G6Pase) (Fig. 6F and G). Likewise,replenishment of PPARa KO mice with dexamethasonealso resulted in a partial reversal of hypoglycemia anddecreased gluconeogenic gene expression in fasted PPARa

Figure 5—Hypothalamic FGFR1 is obligatory for FGF21-induced activation of the HPA axis and hepatic gluconeogenesis in mice. Twelve-week-old FGF21 KO mice were treated with an anti-FGFR1 monoclonal antibody (AF; 0.2 mg/g body weight) or mouse nonimmune IgG (NI)by a single hypothalamic paraventricular (intra-PVN) injection, followed by another intra-PVN injection with rmFGF21 (0.02 mg/g bodyweight) or vehicle control. A: Blood glucose levels. B: mRNA abundance of PGC-1a, PEPCK, and G6Pase in the livers. C–F: The amount ofhormones of HPA axis. C: mRNA abundance of CRH in the hypothalamus of mice collected 3 h after treatment. D: Representative imagesof immunofluorescent staining with anti-CRH antibody in the PVN region of the hypothalamus (scale bar, 200 mm). Plasma ACTH levels (E)and serum corticosterone levels (F ) were determined at different time intervals. *P < 0.05 and **P < 0.01 vs. vehicle + NI group; #P < 0.05and ##P < 0.01 vs. rmFGF21 + NI group (n = 5). Data were expressed as mean 6 SEM. G–J: Twelve-week-old WT mice were treated withAF (0.2 mg/g body weight) or NI by intra-PVN injections, followed by fasting for 24 h. Fed WT mice and FGF21 KO mice treated with NI wereused as control. Blood glucose levels (G) and the mRNA abundance (H) of PGC-1a, PEPCK, and G6Pase in the livers. mRNA abundance ofCRH in the hypothalamus (I) and serum corticosterone levels (J) were determined. *P< 0.05 and **P< 0.01 (n = 5). Data were expressed asmean 6 SEM.

diabetes.diabetesjournals.org Liang and Associates 4071

KO mice, suggesting that PPARa maintains fasting glu-cose homeostasis via activation of the FGF21-glucocorticoidaxis (Fig. 6F and G).

FGF21 Stimulates CRH Production by Activation of theERK1/2-CREB Signaling AxisCREB is a key transcription factor responsible for trans-activation of the CRH gene in PVN neurons (27). TheMAP kinase ERK1/2 activates CREB by phosphorylatingthe transcription factor at serine 133 (28). Notably,FGF21 induces phosphorylation and activation of ERK1/2both in adipocytes (4) and in whole hypothalamusof mice (29). Therefore, we next investigated the rolesof ERK1/2 and CREB in mediating the central actions ofFGF21 in mice. Consistent with previous reports (30),we found that phosphorylation of both ERK1/2 andCREB was significantly enhanced by fasting for 24 h in

WT mice (Fig. 7A). However, fasting-induced phosphor-ylation of ERK1/2 and CREB was impaired in FGF21 KOmice (Fig. 7A).

Injection of rmFGF21 into the PVN region of FGF21KO mice induced phosphorylation of both ERK1/2 andCREB, whereas such a stimulatory effect of rmFGF21 wascompletely blocked by pretreatment with the ERK in-hibitor PD98059 in the hypothalamic PVN region (Fig. 7Band Supplementary Fig. 6). Likewise, the central effects ofrmFGF21 on induction of hypothalamic CRH expression,elevation of serum corticosterone, and blood glucose lev-els, and stimulation of hepatic gluconeogenic gene expres-sion, were all abrogated by intra-PVN injection of PD98059(Fig. 7C–F). Furthermore, direct incubation of mouse hypo-thalamic slices with rmFGF21 was sufficient to induce thephosphorylation of ERK and CREB (Fig. 7G) and to increase

Figure 6—FGF21 mediates PPARa-induced activation of HPA axis and hepatic gluconeogenesis in mice. A–C: FGF21 KO mice and WTlittermates were treated with fenofibrate (Feno, 200 mg/kg) or vehicle by daily intraperitoneal injection for 7 days. Serum levels of FGF21 (A)and serum corticosterone (B) and the mRNA abundance of CRH (D) in the hypothalamus were determined at days 3 and 7 after treatment.**P < 0.01 vs. WT mice treated with vehicle. #P < 0.05 vs. WT + Feno group (n = 6–8). D and E: PPARa KO mice and WT littermates weretreated with rmFGF21 (0.02 mg/g) or vehicle by intracerebroventricular injection every 6 h during 48 h of fasting. D: Serum levels ofcorticosterone were measured at different time points after treatment. E: mRNA abundance of CRH in the hypothalamus was determinedby real-time PCR analysis after 24 h of fasting. *P < 0.05 and **P < 0.01 vs. WT + vehicle group. #P < 0.05 vs. KO + vehicle group (n = 8).F and G: Twelve-week-old PPARa KO mice and WT littermates were treated with rmFGF21 (0.02 mg/kg) or vehicle by intracerebroven-tricular infusion every 6 h during fasting or dexamethasone (DEX, 2 mg/kg) via intraperitoneal injection, followed by fasting for 48 h. F: Bloodglucose levels were measured at different time intervals. *P< 0.05 and **P< 0.01 vs. KO + vehicle (n = 5). Data were expressed as mean6SEM. G: The mRNA abundance of PGC-1a, PEPCK, and G6Pase in the livers of mice collected at fed state or 24 h after fasting. *P < 0.05and **P < 0.01 (n = 5). Data were expressed as mean 6 SEM.

4072 FGF21 Mediates Cross Talk of Liver and Brain Diabetes Volume 63, December 2014

the mRNA expression and protein release of CRH (Fig. 7Hand I). However, all these effects of rmFGF21 on hypotha-lamic slices were blocked by PD98059 (Fig. 7G–I). Takentogether, these findings suggest that FGF21 stimulates hy-pothalamic CRH production via the ERK1/2-CREB signalingcascade, which in turn triggers the release of corticosteronefor induction of gluconeogenesis (Fig. 8).

DISCUSSION

Appropriate counter-regulatory hormone responses tohypoglycemia are critical for maintaining blood glucoselevels within a narrow range. However, the molecularevents that coordinate this process remain poorly defined.In both rodents and humans, fasting-induced hepaticproduction of FGF21 is mediated by PPARa, which isa master regulator coordinating metabolic adaptions tofasting and starvation (1–3). In addition to its central

role in controlling hepatic fatty acid metabolism, PPARais also a critical mediator of fasting-induced hepatic glu-coneogenesis in mice (26,31). PPARa KO mice exhibitsevere fasting hypoglycemia despite normoglycemia inthe fed state (26). However, PPARa is not a direct tran-scriptional activator of the key gluconeogenic genes, sug-gesting that it induces hepatic gluconeogenesis via anindirect mechanism(s) (32). In the current study, wefound that both PPARa and FGF21 KO mice exhibiteda similar degree of hypoglycemia due to defective hepaticgluconeogenesis during prolonged fasting, whereas hypo-glycemia in fasted PPARa KOmice and FGF21 KOmice canbe rectified by central administration of FGF21, suggestingan obligatory role of the PPARa-FGF21 axis in inducinghepatic gluconeogenesis during metabolic adaptation toprolonged fasting. In support of this notion, transgenicexpression of FGF21 was sufficient to induce hepatic

Figure 7—FGF21 increases CRH production via ERK-CREB-CRH signaling pathways. A: The hypothalamic tissue lysates of 12-week-oldFGF21 KO mice and WT littermates in either fed or fasted (24 h) state were subjected to immunoblotting analysis with antibodies againstp-ERK, t-ERK, p-CREB, or t-CREB. B–F: FGF21 KO mice were treated with the ERK inhibitor PD98059 (2 mg/mouse) or vehicle control(DMSO) by a single intra-PVN injection, followed by another intra-PVN injection with rmFGF21 (0.02 mg/g body weight) or vehicle control. B:Hypothalamic protein levels of p-ERK, t-ERK, p-CREB, and t-CREB were determined by Western blotting analysis after 30 min of injection.Blood glucose (C ) and the mRNA abundance of PGC-1a, PEPCK, and G6Pase in the livers (D), the mRNA abundance of CRH in thehypothalamus (E ), and serum corticosterone levels (F ) were measured after 3 h of injection. G–I: The hypothalamic slices from FGF21 KOmice were preincubated with PD98059 (20 mmol/L) for 30 min, followed by treatment with rmFGF21 (1 mg/mL). G: Hypothalamic proteinlevels of p-ERK, t-ERK, p-CREB, and t-CREB were determined by Western blotting analysis after 15 min of incubation with rmFGF21. ThemRNA abundance of CRH in the hypothalamus (H) and the CRH levels in culture medium (I) were determined after 3 h of incubation withrmFGF21. Representative Western blotting analysis and qRT-PCR assays from three independent cohorts run in duplicate are shown. *P<0.05 (n = 6). Data were expressed as mean 6 SEM.

diabetes.diabetesjournals.org Liang and Associates 4073

gluconeogenesis in the fed state to a level normally attainedduring prolonged fasting, whereas FGF21 KO mice showedimpaired hepatic gluconeogenesis after 24-h fasting (9). Bycontrast, Badman et al. (33) found that fasting glucoselevels were comparable between WT and FGF21 KO micegenerated in their laboratory. This discrepancy may be dueto different ages of mice or different fasting schedules.Nevertheless, they also showed impaired gluconeogenicgene induction (PGC-1a and PEPCK) in their FGF21 KOmice, which is consistent with our present study.

The hypothalamus has emerged as a key player in theregulation of glucose homeostasis, by sensing hormonesand nutrients to initiate metabolic responses (34). Hor-mones such as insulin, glucagon, and leptin have beenshown to regulate hepatic glucose metabolism throughboth central and peripheral actions (34,35). In this study,we provide several lines of evidence demonstrating thatthe stimulatory effect of FGF21 on hepatic gluconeogen-esis is mainly mediated by its central actions on hypotha-lamic neurons. First, fasting hypoglycemia and impairedexpression of hepatic gluconeogenic genes in both FGF21KO and PPARa KO mice can be reversed by intracere-broventricular injection of FGF21 at a low dose that isnot detectable in the peripheral blood. Second, FGF21-induced corticosterone release and hepatic gluconeogene-sis can be completely blocked by central administration ofa neutralizing antibody against its receptor FGFR1 ora pharmacological inhibitor of ERK1/2. Furthermore,both our animal and ex vivo studies showed that FGF21

stimulates CRH expression in the hypothalamus via ERK1/2-mediated activation of CREB, suggesting that CRH neu-rons are the direct target of FGF21.

Consistent with our observations in PPARa KO mice,administration of the PPARa agonist fenofibrate has beenshown to elevate fasting serum levels of corticosterone,enhance hepatic gluconeogenesis, and increase blood glu-cose levels (36). These findings support the notion thatPPARa coordinates metabolic adaptation responses tostarvation through a bipartite mechanism: direct trans-activation of the genes involved in hepatic fatty acid ox-idation and ketogenesis (26,31) and indirect activation ofthe HPA axis via induction of FGF21. The synergic effectof this dual action of PPARa is to provide glycerol, fattyacids, ketones, and alanine as sources for gluconeogenesisto maintain glucose homeostasis.

In peripheral tissues, adipocytes have been identifiedas a main target that confers the pleiotropic metabolicactions of FGF21 (4,19). In white adipocytes, adiponectinexpression and secretion is induced by FGF21 and medi-ates several metabolic benefits of FGF21 treatment(18,37). These findings together with our present reportfurther support the notion that FGF21 exerts its diversemetabolic effects indirectly, by modulating the productionof other hormones such as adiponectin and corticoste-rone. The central and peripheral actions of FGF21 onactivation of the HPA axis and its peripheral actions onpromoting adipose secretion of adiponectin appear tohave opposite effects on hepatic glucose production. How-ever, these two actions of FGF21 may occur under differ-ent physiological conditions. FGF21 secretion in adiposetissue is elevated in the fed state, which in turn serves asan autocrine signal to induce PPARg activity and adipo-nectin secretion, thereby leading to enhanced hepatic in-sulin sensitivity (18). Conversely, the central effect ofFGF21 on hepatic glucose production consequent to in-creased release of corticosterone occurs only during pro-longed fasting. Such a dual effect of FGF21 may enablethe control of blood glucose within a narrow range indifferent nutritional states.

In conclusion, our results demonstrate FGF21 asa critical hormonal regulator of glucose homeostasisduring prolonged fasting, by coupling hepatic PPARa ac-tivation to corticosterone release via stimulation of theHPA axis, thereby leading to enhanced hepatic gluconeo-genesis (Fig. 8). This novel liver-PPARa-FGF21-brain-corticosterone circuit can fine tune the cross talk betweenbrain and liver to avoid hypoglycemia during nutrientdeprivation or other adverse clinical events. Interestingly,a novel “hepato-adrenal syndrome” has recently been sug-gested, demonstrating a defect in HPA activation inpatients with severe liver disease (38). Hypoglycemia isa common clinical problem in patients with a defect inHPA regulation such as hypopituitarism and Addison dis-ease. It is also frequently seen in diabetic patients treatedwith insulin or other medications improperly. Unravelingthe liver-adrenal circuit with FGF21 as key player may

Figure 8—A working model by which FGF21 maintains glucosehomeostasis during fasting via mediating the cross talk betweenbrain and liver. FFA, free fatty acids; HSL, hormone-sensitivelipase; TG, triglycerides.

4074 FGF21 Mediates Cross Talk of Liver and Brain Diabetes Volume 63, December 2014

open new avenues for the treatment of these metabolicdiseases with hypoglycemia unawareness.

Acknowledgments. The authors thank N. Itoh and M. Konishi (Universityof Kyoto) for FGF21 KO mice.Funding. This work is supported by the National Basic Research Program ofChina (2011CB504004), the General Research Fund (HKU 783413 and 784111 toA.X.), the Collaborative Research Fund (HKU2/CRF/12R) from the Research GrantsCouncil of Hong Kong, the Matching Fund for the State Key Laboratory of Phar-maceutical Biotechnology from The University of Hong Kong, and the QatarNational Research Fund (NPRP 6-428-3-113).Duality of Interest. No potential conflicts of interest relevant to this articlewere reported.Author Contributions. Q.L. designed and performed the experiments,analyzed the data, and wrote the manuscript. L.Z. and J.Z. performed the experi-ments. Y.W., S.R.B., C.R.T., and H.D. provided technical assistance, reagents,and valuable advice for this study. K.S.L.L. supervised the study and edited themanuscript. A.X. conceived the concept, supervised the project, and wrote themanuscript. Q.L. is the guarantor of this work and, as such, had full access to allthe data in the study and takes responsibility for the integrity of the data and theaccuracy of the data analysis.

References1. Inagaki T, Dutchak P, Zhao G, et al. Endocrine regulation of the fastingresponse by PPARalpha-mediated induction of fibroblast growth factor 21. CellMetab 2007;5:415–4252. Badman MK, Pissios P, Kennedy AR, Koukos G, Flier JS, Maratos-FlierE. Hepatic fibroblast growth factor 21 is regulated by PPARalpha and is a keymediator of hepatic lipid metabolism in ketotic states. Cell Metab 2007;5:426–4373. Gälman C, Lundåsen T, Kharitonenkov A, et al. The circulating metabolicregulator FGF21 is induced by prolonged fasting and PPARalpha activation inman. Cell Metab 2008;8:169–1744. Kharitonenkov A, Shiyanova TL, Koester A, et al. FGF-21 as a novel met-abolic regulator. J Clin Invest 2005;115:1627–16355. Kharitonenkov A, Wroblewski VJ, Koester A, et al. The metabolic state ofdiabetic monkeys is regulated by fibroblast growth factor-21. Endocrinology2007;148:774–7816. Xu J, Lloyd DJ, Hale C, et al. Fibroblast growth factor 21 reverses hepaticsteatosis, increases energy expenditure, and improves insulin sensitivity in diet-induced obese mice. Diabetes 2009;58:250–2597. Inagaki T, Lin VY, Goetz R, Mohammadi M, Mangelsdorf DJ, Kliewer SA.Inhibition of growth hormone signaling by the fasting-induced hormone FGF21.Cell Metab 2008;8:77–838. Adams AC, Astapova I, Fisher FM, et al. Thyroid hormone regulates hepaticexpression of fibroblast growth factor 21 in a PPARalpha-dependent manner. JBiol Chem 2010;285:14078–140829. Potthoff MJ, Inagaki T, Satapati S, et al. FGF21 induces PGC-1a andregulates carbohydrate and fatty acid metabolism during the adaptive starvationresponse. Proc Natl Acad Sci U S A 2009;106:10853–1085810. Hsuchou H, Pan W, Kastin AJ. The fasting polypeptide FGF21 can enterbrain from blood. Peptides 2007;28:2382–238611. Tan BK, Hallschmid M, Adya R, Kern W, Lehnert H, Randeva HS. Fibroblastgrowth factor 21 (FGF21) in human cerebrospinal fluid: relationship with plasmaFGF21 and body adiposity. Diabetes 2011;60:2758–276212. Sarruf DA, Thaler JP, Morton GJ, et al. Fibroblast growth factor 21 action inthe brain increases energy expenditure and insulin sensitivity in obese rats.Diabetes 2010;59:1817–182413. Bookout AL, de Groot MH, Owen BM, et al. FGF21 regulates metabolism andcircadian behavior by acting on the nervous system. Nat Med 2013;19:1147–115214. Owen BM, Bookout AL, Ding X, et al. FGF21 contributes to neuroendocrinecontrol of female reproduction. Nat Med 2013;19:1153–1156

15. Hotta Y, Nakamura H, Konishi M, et al. Fibroblast growth factor 21 regulateslipolysis in white adipose tissue but is not required for ketogenesis and tri-glyceride clearance in liver. Endocrinology 2009;150:4625–463316. Combs TP, Berg AH, Obici S, Scherer PE, Rossetti L. Endogenous glucoseproduction is inhibited by the adipose-derived protein Acrp30. J Clin Invest 2001;108:1875–188117. Franklin K, Paxinos G. The Mouse Brain in Stereotaxic Coordinates. SanDiego, Academic, 199718. Lin Z, Tian H, Lam KS, et al. Adiponectin mediates the metabolic effects of FGF21on glucose homeostasis and insulin sensitivity in mice. Cell Metab 2013;17:779–78919. Chen W, Hoo RL, Konishi M, et al. Growth hormone induces hepatic pro-duction of fibroblast growth factor 21 through a mechanism dependent on li-polysis in adipocytes. J Biol Chem 2011;286:34559–3456620. Tadross JA, Patterson M, Suzuki K, et al. Augurin stimulates thehypothalamo-pituitary-adrenal axis via the release of corticotrophin-releasingfactor in rats. Br J Pharmacol 2010;159:1663–167121. Cheng KK, Iglesias MA, Lam KS, et al. APPL1 potentiates insulin-mediatedinhibition of hepatic glucose production and alleviates diabetes via Akt activationin mice. Cell Metab 2009;9:417–42722. Yoon JC, Puigserver P, Chen G, et al. Control of hepatic gluconeogenesisthrough the transcriptional coactivator PGC-1. Nature 2001;413:131–13823. Yu H, Xia F, Lam KSL, et al. Circadian rhythm of circulating fibroblast growthfactor 21 is related to diurnal changes in fatty acids in humans. Clin Chem 2011;57:691–70024. Kharitonenkov A, Dunbar JD, Bina HA, et al. FGF-21/FGF-21 receptor in-teraction and activation is determined by betaKlotho. J Cell Physiol 2008;215:1–725. Gonzalez AM, Berry M, Maher PA, Logan A, Baird A. A comprehensiveanalysis of the distribution of FGF-2 and FGFR1 in the rat brain. Brain Res 1995;701:201–22626. Kersten S, Seydoux J, Peters JM, Gonzalez FJ, Desvergne B, Wahli W.Peroxisome proliferator-activated receptor a mediates the adaptive response tofasting. J Clin Invest 1999;103:1489–149827. Aguilera G, Liu Y. The molecular physiology of CRH neurons. Front Neu-roendocrinol 2012;33:67–8428. Song XS, Cao JL, Xu YB, He JH, Zhang LC, Zeng YM. Activation of ERK/CREBpathway in spinal cord contributes to chronic constrictive injury-induced neu-ropathic pain in rats. Acta Pharmacol Sin 2005;26:789–79829. Yang C, Jin C, Li X, Wang F, McKeehan WL, Luo Y. Differential specificity ofendocrine FGF19 and FGF21 to FGFR1 and FGFR4 in complex with KLB. PLoSONE 2012;7:e3387030. Morikawa Y, Ueyama E, Senba E. Fasting-induced activation of mitogen-activated protein kinases (ERK/p38) in the mouse hypothalamus. J Neuro-endocrinol 2004;16:105–11231. Peeters A, Baes M. Role of PPARa in hepatic carbohydrate metabolism.PPAR Res 2010;2010:57240532. Im SS, Kim MY, Kwon SK, et al. Peroxisome proliferator-activated receptora is responsible for the up-regulation of hepatic glucose-6-phosphatase geneexpression in fasting and db/db mice. J Biol Chem 2011;286:1157–116433. Badman MK, Koester A, Flier JS, Kharitonenkov A, Maratos-Flier E. Fibro-blast growth factor 21-deficient mice demonstrate impaired adaptation to ke-tosis. Endocrinology 2009;150:4931–494034. Lam TK. Neuronal regulation of homeostasis by nutrient sensing. Nat Med2010;16:392–39535. Mighiu PI, Yue JT, Filippi BM, et al. Hypothalamic glucagon signaling inhibitshepatic glucose production. Nat Med 2013;19:766–77236. Chen X, Li M, Sun W, et al. Peroxisome proliferator-activated receptor aagonist-induced down-regulation of hepatic glucocorticoid receptor expression inSD rats. Biochem Biophys Res Commun 2008;368:865–87037. Holland WL, Adams AC, Brozinick JT, et al. An FGF21-adiponectin-ceramide axiscontrols energy expenditure and insulin action in mice. Cell Metab 2013;17:790–79738. Fede G, Spadaro L, Tomaselli T, et al. Adrenocortical dysfunction in liverdisease: a systematic review. Hepatology 2012;55:1282–1291

diabetes.diabetesjournals.org Liang and Associates 4075