Daniel Beiroa,1,2 Monica Imbernon,1,2 Rosalía Gallego,3 Ana Senra,1 Daniel Herranz,4

Francesc Villarroya,2,5 Manuel Serrano,4 Johan Fernø,6 Javier Salvador,2,7 Javier Escalada,2,7

Carlos Dieguez,1,2 Miguel Lopez,1,2 Gema Frühbeck,2,7 and Ruben Nogueiras1,2

GLP-1 Agonism StimulatesBrown Adipose TissueThermogenesis and BrowningThrough Hypothalamic AMPKDiabetes 2014;63:3346–3358 | DOI: 10.2337/db14-0302

GLP-1 receptor (GLP-1R) is widely located throughoutthe brain, but the precise molecular mechanisms medi-ating the actions of GLP-1 and its long-acting analogson adipose tissue as well as the brain areas responsiblefor these interactions remain largely unknown. We foundthat central injection of a clinically used GLP-1R agonist,liraglutide, in mice stimulates brown adipose tissue(BAT) thermogenesis and adipocyte browning indepen-dent of nutrient intake. The mechanism controlling theseactions is located in the hypothalamic ventromedialnucleus (VMH), and the activation of AMPK in this areais sufficient to blunt both central liraglutide-inducedthermogenesis and adipocyte browning. The decreasedbody weight caused by the central injection of liraglutidein other hypothalamic sites was sufficiently explained bythe suppression of food intake. In a longitudinal studyinvolving obese type 2 diabetic patients treated for 1year with GLP-1R agonists, both exenatide and liraglu-tide increased energy expenditure. Although the resultsdo not exclude the possibility that extrahypothalamicareas are also modulating the effects of GLP-1Ragonists, the data indicate that long-acting GLP-1Ragonists influence body weight by regulating eitherfood intake or energy expenditure through various

hypothalamic sites and that these mechanisms mightbe clinically relevant.

GLP-1 is an incretin hormone released by L cells locatedin the ileum and colon into the bloodstream postpran-dially (1,2). Among its numerous physiological effects (3),GLP-1 increases insulin and decreases glucagon secretionin a glucose-dependent manner (4,5), slows gastric emptying(6), increases glucose disposal, and decreases appetite (7).Therefore, incretin hormone analogs acting on the GLP-1receptor (GLP-1R) are considered the most promising newtherapies for type 2 diabetes (T2D). Within the centralnervous system (CNS), numerous neuronal populationsexpress GLP-1R, including hypothalamic nuclei crucialfor the regulation of energy balance (8). Furthermore,a large number of extrahypothalamic areas have GLP-1binding sites (9). Indeed, it has been demonstrated thatperipheral administration of GLP-1 leads to neuronal ac-tivation in various parts of the CNS (10), indicating theexistence of brain sites accessible from the bloodstream.Finally, cells with GLP-1 mRNA are widely expressed inhuman brain areas (11).

1Department of Physiology, CIMUS, University of Santiago de Compostela-Institutode Investigación Sanitaria, Santiago de Compostela, Spain2CIBER Fisiopatología de la Obesidad y Nutrición (CIBERobn), Santiago deCompostela, Spain3Department of Morphological Sciences, School of Medicine, University of Santiagode Compostela-Instituto de Investigación Sanitaria, Santiago de Compostela, Spain4Tumor Suppression Group, Spanish National Cancer Research Center (CNIO),Madrid, Spain5Department of Biochemistry and Molecular Biology and Institute of Biomedicine(IBUB), University of Barcelona, Barcelona, Spain6Department of Clinical Science, K.G. Jebsen Center for Diabetes Research, Uni-versity of Bergen, Bergen, Norway

7Department of Endocrinology and Nutrition, Clínica Universidad de Navarra,Pamplona, Spain

Corresponding author: Ruben Nogueiras, ruben.nogueiras@usc.es.

Received 21 February 2014 and accepted 30 April 2014.

This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db14-0302/-/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.

3346 Diabetes Volume 63, October 2014

OBESITY

STUDIES

In the hypothalamus, GLP-1 acts as a physiologicalsatiety factor (7), and dose-dependent central administra-tion of GLP-1 reduces feeding in rats (12). Central GLP-1administration blunted fasting-induced neuropeptide Yand Agouti-related peptide levels and fasting-reducedproopiomelanocortin and cocaine- and amphetamine-regulated transcript expression (13). Stimulation of the cen-tral GLP-1 system not only suppresses food intake but alsoregulates glucose homeostasis (14), behavioral responses tostress (15), and visceral illness (16). Moreover, brain GLP-1modulates lipid metabolism in white adipose tissue (WAT)(17) and brown adipose tissue (BAT) thermogenesis (18)through the activation of the sympathetic nervous system.However, the molecular mechanisms mediating the actionsof brain GLP-1 on adipose tissue as well as the brain areasresponsible for these interactions are unknown. Further-more, whether some of the beneficial effects exerted inthe clinical setting by long-acting GLP-1 agonists could bemediated at the central level is yet unclear. One of theselong-acting GLP-1 analogs is liraglutide, which injected oncedaily improves glycemic control in T2D with the additionalbenefits of weight loss and a low risk of hypoglycemia(19,20). As GLP-1R agonists start to be included in treat-ment guidelines, they are generally being recommended assecond- or third-line therapies after the failure of one ormore oral antidiabetic drugs (21).

In the current study, we show that central stimulationof GLP-1R by the agonist liraglutide leads to body weightloss independent of reduction in food intake. Instead, thisweight loss is caused by the activation of the thermogenicprogram in BAT. Specific injection of liraglutide in thehypothalamic ventromedial nucleus (VMH) is sufficient tocause food intake–independent weight loss and stimula-tion of BAT thermogenic activity. This regulatory mecha-nism depends on AMPK.

RESEARCH DESIGN AND METHODS

Animal ModelsSwiss male mice (8–10 weeks old, 20–25 g) and Sprague-Dawley rats (8–10 weeks old, 200–250 g) were housedunder conditions of controlled temperature (23°C) and il-lumination (12-h light/dark cycle). They were allowed adlibitum access to water and standard laboratory chow.Body composition was measured by a nuclear magneticresonance imaging whole-body composition analyzer(EchoMRI, Houston, TX). Animals were killed by decapita-tion, and the tissues were removed rapidly, frozen imme-diately on dry ice, and kept at 280°C until analysis. Allanimal experiments and procedures were reviewed and ap-proved by the Ethics Committee of the University of San-tiago de Compostela in accordance with European Unionnorms for the use of experimental animals.

Implantation of Intracerebroventricular Cannulae andTreatmentsRats and mice were anesthetized by an intraperitonealinjection of ketamine 80 and 8 mg/kg body weight,

respectively, and xylazine 100 and 3 mg/kg body weight,respectively. Intracerebroventricular (ICV) cannulae wereimplanted stereotaxically in mice and rats, as de-scribed previously (22). Animals received vehicle, GLP-1(1 mg/mouse), liraglutide (3 mg/mouse or 10 mg/rat), orAICAR (3 mg/mouse).

Conditioned Taste AversionConditioned taste aversion (CTA) was expressed as thepercent saccharin preference ratio [(100 3 saccharin in-take)/(saccharin intake + water intake)]. A saccharin pref-erence ratio ,50% was considered a signal of CTA (23).

Cold ExposureTwenty-four hours after the liraglutide injection, animalswere placed for 6 h in a room with a stable temperatureof 4°C as previously described (24). Body temperaturewas recorded with a rectal probe connected to a digitalthermometer, and the interscapular temperature wasrecorded with an infrared camera.

Stereotaxic MicroinjectionSprague-Dawley rats were placed in a stereotaxic frameunder ketamine-xylazine anesthetics. Liraglutide (10mg/rat) was injected in the arcuate nucleus (ARC) (anteriorto the bregma [AP],22.85 mm; lateral to the sagittal suture[L], 60.3 mm; and ventral from the surface of the skull [V],10.2 mm), in the lateral hypothalamus area (LHA) (AP,22.85 mm; L, 62 mm; DV, 28.1 mm), in the VMH (AP,22.85 mm; L, 60.6 mm; DV, 210mm), in the dorsomedialnucleus (DMH) (AP, 23.12 mm; L, 60.5 mm; DV, 28.6mm), and in the paraventricular nucleus (PVH) (AP, 21.9mm; L, 60.5 mm; DV, 28 mm) with a 25-gauge needle(Hamilton, Reno, NV) connected to a 1-mL syringe. Acetyl-salicylic acid (Bayer, Leverkusen, Germany) 150 mg/kg wasinjected intraperitoneally after surgery as a painkiller.Liraglutide 5 mg and adenoviral vectors (green fluores-cent protein [GFP] or AMPKa-CA [Viraquest, North Liberty,IA]) (25,26) were injected simultaneously into the VMHwith a 25-gauge needle connected to a 2-mL syringe.

Immunohistochemistry and ImmunofluorescenceRat brains were fixed, and diaminobenzidine immunohis-tochemistry and detection of GFP were performed aspreviously reported (22,25).

RNA Isolation and Real-Time RT-PCRTotal RNA and real-time RT-PCR were performed aspreviously described (22). The primers and probes aredescribed in Supplementary Table 1.

Western Blot AnalysisWestern blot was performed as previously described (22).Briefly, total protein lysates from epididymal WAT (30mg) and BAT (15 mg) were subjected to SDS-PAGE, electro-transferred on a polyvinylidene difluoride membrane, andprobed with the following antibodies: uncoupling pro-tein (UCP) 1, UCP3, b-adrenoreceptor 1 (ADRb1), celldeath–inducing DNA fragmentation factor a–like effectora (CIDEA), fibroblast growth factor 21 (FGF21), and PR

diabetes.diabetesjournals.org Beiroa and Associates 3347

domain containing 16 (PRDM16) (Abcam, Cambridge, U.K.);GAPDH (EMD Millipore, Billerica, MA); and a-tubulin(T-5168; Sigma-Aldrich, St. Louis, MO).

Patient SelectionAll clinical studies were approved from an ethical andscientific standpoint by the hospital’s ethics committeeand were conducted in accordance with the principles ofthe Declaration of Helsinki, with patients giving their in-formed consent for participation. Twenty-five obese T2Dpatients matched for sex, age, and BMI were studied be-fore (baseline) and after (12 6 3 months) antidiabetictreatment instauration with metformin in combinationwith either of the GLP-1 agonists exenatide or liraglutide.The initial dose of metformin was 500 mg b.i.d. orally,which was increased to 1,000 mg b.i.d. orally after 2weeks. Exenatide treatment was initiated with 5 mgb.i.d. administered subcutaneously and increased to 10mg b.i.d. after 2 weeks of good digestive tolerance.Liraglutide treatment was initiated with a single subcuta-neous administration of 0.6 mg a day progressing to 1.2mg a day after 2 weeks of adequate digestive tolerance. Allpatients were of Caucasian origin, attended the Endocri-nology Department at the University Clinic of Navarra,and underwent a clinical assessment. Inclusion criteriawere age between 20 and 80 years, hemoglobin A1c

(HbA1c) levels $7.0%, and no previous treatment withinsulin and/or a sulfonylurea. Exclusion criteria werelactating or pregnant women, uncontrolled treated oruntreated hypertension, fasting C-peptide levels ,0.1ng/mL, recurrent major hypoglycemia or hypoglycemicunawareness, known proliferative retinopathy or mac-ulopathy requiring acute treatment, impaired renal func-tion defined as a serum creatinine level $133 mmol/L formen and $124 mmol/L for women, history of chronicpancreatitis or idiopathic acute pancreatitis, known historyof unstable angina, acute coronary event, heart failure,other significant cardiac event or stroke, thyroid disorders,malignant diseases, hematologic alterations, and concur-rent medication likely to influence energy homeostasis.

Indirect CalorimetryResting energy expenditure (REE) and respiratory quotient(RQ) were determined by indirect calorimetry after a 12-hovernight fast by using an open-air circuit–ventilated can-opy measurement system (Vmax29; SensorMedics Corpo-ration, Yorba Linda, CA) (27). After adjustment for bodycomposition, the measured REE was compared with pre-dicted REE according to age- and sex-specific equations(28). The physical activity level (PAL) was assessed by aquestionnaire validated with doubly labeled water (29).On the basis of the REE determination and obtainedPAL, the total energy expenditure for each individualwas calculated.

Blood DeterminationsBlood determinations were performed as previously de-scribed (22,30).

Statistical AnalysisFor rodents, data are expressed as mean 6 SEM in re-lation (%) to the specific control (vehicle-treated rats).Statistical significance was determined by Student t test(for two groups) or ANOVA and post hoc two-tailed Bon-ferroni test (for more than two groups). P , 0.05 wasconsidered significant. For human data, a mixed ANOVAwith a repeated-measures design was performed to studybetween- and within-subject factors comparing temporalchanges in values between the treatment groups. More-over, a MANOVA was used to assess differences betweengroups during the overall study time course, using Pillaitrace criterion as a test of significance. Comparisons be-tween pre- and postdata within a same treatment groupwere further analyzed by two-tailed paired Student t tests.The calculations were performed by SPSS for Windowsversion 15.0 software (IBM Corporation, Chicago, IL).P , 0.05 was considered statistically significant.

RESULTS

Central GLP-1R Stimulation Suppresses Food Intakeand Decreases Body Weight in MiceA single ICV injection of liraglutide (0.3, 1, and 3 mg/mouse) significantly decreased food intake and bodyweight after 24 h at doses of 1 and 3 mg/mouse, andthese effects remitted at 48 h (Supplementary Fig. 1).We next compared the central effects of liraglutide versusGLP-1 at 1 mg/mouse, a dose adapted from previous stud-ies (31,32). Whereas a single ICV injection of liraglutide (3mg/mouse) suppressed feeding behavior after 24 h, GLP-1failed to do so (Fig. 1A). ICV liraglutide-decreased foodintake was not caused by aversive effects (Fig. 1B). ICVliraglutide also suppressed body weight independent of itsfood intake effect because vehicle pair-fed mice, which atethe same amount of food as liraglutide-treated mice, didnot show a significant decrease in body weight relative tovehicle-treated mice (Fig. 1C). ICV liraglutide-treated miceshowed increased energy expenditure when corrected forlean mass (Fig. 1D), without changes in locomotor activity(Fig. 1E) or RQ (Fig. 1F). Body composition of these micewas determined immediately after keeping them in theindirect calorimetric system. The reduction in bodyweight was accompanied by a decrease in circulating lep-tin levels but without any changes in free fatty acids,triglycerides, cholesterol, insulin, or glucose levels (Sup-plementary Table 1).

To rule out the possibility that centrally infusedliraglutide may leak out of the CNS into the circulationand elicit a response by directly acting at peripheral level,we administered liraglutide peripherally. At the samedoses as those infused centrally, liraglutide did not changecumulative food intake (Fig. 1G) or body weight (Fig. 1H).

Central GLP-1R Stimulation Triggers BATThermogenic Activity in MiceBecause it has been reported that central GLP-1 infusionincreases BAT thermogenesis (18), we investigated the

3348 Liraglutide and Thermogenesis Diabetes Volume 63, October 2014

Figure 1—Effect of a 24-h ICV liraglutide (3 mg/mouse) or GLP-1 (1 mg/mouse) injection on cumulative food intake (A), CTA (B), body weightchange (C), energy expenditure (D), locomotor activity (E), and RQ (F ). Effect of a 24-h intraperitoneal liraglutide (3 mg/mouse) on cumulativefood intake (G) and body weight change (H). Data are mean 6 SEM of 7–8 animals per group. *P < 0.05, **P < 0.01, ***P < 0.001 vs.vehicle. #P < 0.05, ##P < 0.01. EE, energy expenditure; IP, intraperioteneal; LiCl, liraglutide; PF, pair-fed.

diabetes.diabetesjournals.org Beiroa and Associates 3349

effects of liraglutide on this issue. First, we found thatcore body temperature was not affected by ICV liraglutideat any of the studied time points when animals werehoused at room temperature (Fig. 2A). However, BATinterscapular temperature increased significantly at 12 hafter ICV liraglutide injection in the same animals housedat room temperature (Fig. 2B and C). Histomorphologicalanalysis revealed smaller lipid droplets in BAT from mice24 h after a single ICV injection of liraglutide (Fig. 2D).Accordingly, in BAT of centrally liraglutide-treated mice,we found higher gene expression levels of several thermo-genic biomarkers, such as CIDEA, FGF21, bone morpho-genic protein 7 (BMP7), PRDM16, and ADRb1 (Fig. 2E).Protein levels of UCP1, UCP3, ADRb1, FGF21, andPRDM16 were also significantly augmented after theICV liraglutide injection (Fig. 2F).

There were no differences in basal core body temper-atures between vehicle- and liraglutide-treated mice (Fig.2G). However, liraglutide-treated mice showed increasedcore body temperature (Fig. 2G) compared with controlmice when they were kept at 4°C.

Central GLP-1R Stimulation Induces Browning of WATin MiceICV liraglutide induced browning in WAT, evident 24 hafter the liraglutide injection (Fig. 3A). UCP1 protein lev-els (Fig. 3B) and both UCP1 and PRDM16 mRNA expres-sion (Fig. 3C) were significantly upregulated in WAT ofmice treated with ICV liraglutide.

GLP-1R Stimulation in the VMH RegulatesThermogenesis and Browning in RatsWe next aimed to investigate the hypothalamic arearesponsible for the actions of liraglutide on BAT andWAT. Thus, we specifically injected liraglutide in the ARC,LHA, PVH, DMH, or VMH of rats (Fig. 4A–E). C-FOSimmunostaining was assessed to corroborate the effi-ciency of the injections in each area (Fig. 4A–E). We foundthat the specific activation of the GLP-1R in the ARC,LHA, and PVH decreased food intake and body weightof rats (Fig. 4A–C), whereas no effects were detectedwhen liraglutide was injected in the DMH (Fig. 4D). Atthe molecular level, UCP1 was unaltered in BAT and WATof rats injected with liraglutide in those nuclei (Fig. 4A–D). Of note, administration of liraglutide in the VMH ledto a significant weight loss with no significant differencesin food intake, and UCP1 levels were increased in BAT andWAT (Fig. 4E).

AMPK Within the VMH Is Essential for the CentralActions of Liraglutide on BAT and WATBecause previous reports have demonstrated a key role ofAMPK within the VMH in the regulation of BAT thermo-genesis (25,26), we first assessed hypothalamic pAMPKlevels after ICV injection of liraglutide. We found thatpAMPK and its downstream target pACC were significantlydecreased in the whole hypothalamus of ICV liraglutide-treated mice compared with their controls (Fig. 5A). There-fore, to demonstrate the relevance of hypothalamic AMPK

as a mediator of the actions of liraglutide, we used a phar-macological activator of AMPK, AICAR (25). At the doseused, ICV AICAR did not blunt the anorexigenic effect ofICV liraglutide (Fig. 5B) but prevented ICV liraglutide-induced weight loss (Fig. 5C). Second, AMPK activity waselevated by using an adenoviral vector encoding constitu-tively active (CA) AMPKa, with adenoviruses expressingGFP used as controls (25). The adenoviruses were injectedstereotaxically into the VMH of rats together with liraglu-tide, and the specificity of these injections was corrob-orated by GFP immunostaining (Fig. 5D). Liraglutideinjected specifically into the VMH decreased weight, butthe liraglutide-induced weight loss was completely bluntedwhen AMPKa-CA was overexpressed in the VMH (Fig. 5E).Identical to the effect on body weight, AMPK activation inthe VMH also reduced the liraglutide-induced UCP1 ex-pression in BAT of rats (Fig. 5F) and WAT (Fig. 5G).

The SIRT1/p53 system interacts with AMPK at thecentral level, and both molecules are key mediators of theorexigenic action of ghrelin (33). Furthermore, GLP-1inhibits ghrelin-stimulated neuronal activity in the hypo-thalamus as well as its effects on food intake (34). There-fore, we next examined whether SIRT1 or p53 might berelevant for the central actions of liraglutide. To this aim,we used mice with moderate overexpression of SIRT1 un-der the control of its natural promoter (35) and p53 nullmice (33). The findings demonstrate that ICV injection ofliraglutide decreased food intake and body weight simi-larly in both wild-type and SIRT1 transgenic mice (Fig. 5Hand I) or p53 null mice (Fig. 5J and K).

Long-Term GLP-1R Agonism Increases EnergyExpenditure in Obese T2D PatientsIn addition to the functional data on liraglutide from themouse study, we analyzed data from a cohort of obesepatients with T2D who had been treated for 1 year withthe antidiabetic drugs metformin and metformin incombination with the GLP-1R agonists exenatide andliraglutide (Table 1). At baseline, all obese T2D groupsexhibited an identical sex distribution with no significantdifferences in age, BMI, waist circumference, body com-position, REE, and RQ (Table 1) as well as in biochemicaland hormonal variables (Supplementary Table 1). After 1year of antidiabetic treatment instauration, all studygroups showed a decrease in fasting plasma glucose andinsulin concentrations. No significant changes in totalcholesterol, HDL cholesterol, and LDL cholesterol concen-trations at the end of the 1-year study period were ob-served. After 12 months of antidiabetic treatment, thegroups treated with metformin combined with exenatideor liraglutide showed a significant decrease in BMI andtotal body fat percentage and a significant increase in fat-free mass (FFM) (Table 1). None of the study groups sig-nificantly changed in RQ and PAL during the experimentalperiod, which remained within the same sedentary range asbefore treatment started as well as between the diverseantidiabetic administration groups (Table 1). Although no

3350 Liraglutide and Thermogenesis Diabetes Volume 63, October 2014

Figure 2—Effect of ICV liraglutide (10 mg/rat) on core body temperature (A) and BAT interscapular temperature (B) in animals housed at roomtemperature after 6, 12, and 24 h of the injection. C: Representative pictures of BAT interscapular temperature in animals housed at roomtemperature after 12 h of the injection. D: Representative BAT histology pictures (hematoxylin-eosin) 24 h after liraglutide ICV injection (3 mg/mouse).Effect of a 24-h ICV liraglutide (3 mg/mouse) on mRNA expression of PGC1a, ADRb1, ADRb2, ADRb3, CIDEA, FGF21, BMP7, and PRDM16 (E) andprotein levels of UCP1, UCP3, CIDEA, FGF21, and PRDM16 (F). Tubulin was used to normalize protein levels. G: Body core temperature measure-ments over a 6-h period at 4°C. Data are mean 6 SEM of 7–8 animals per group. *P < 0.05, **P < 0.01, ***P < 0.001 vs. vehicle. PF, pair-fed.

diabetes.diabetesjournals.org Beiroa and Associates 3351

statistically significant differences between and withingroups were observed for unadjusted REE, there was a cleartendency to be higher in patients treated with GLP-1 ago-nists. When the REE data were adjusted for FFM, we founda significant increase in REE in patients treated with met-formin in combination with exenatide or liraglutide; how-ever, patients treated only with metformin did not showchanges in any of these parameters (Fig. 6 and Table 1).

DISCUSSION

We report that the CNS GLP-1 system directly activatesBAT thermogenesis and browning of white adipocytes.The primary finding is that these actions are mediated byhypothalamic AMPK, specifically within the VMH. To ourknowledge, these findings are the first to provide in-formation about the brain site and mechanisms by whichfat mass decreases in response to a stimulation of CNS

Figure 3—A: Representative WAT histology pictures (hematoxylin-eosin) 24 h after liraglutide ICV injection (3 mg/mouse). Effect of 24-h ICVliraglutide (3 mg/mouse) on protein levels of UCP1 (B) and mRNA expression of PGC1a, PPARg, PPARa, CEBPa, CEBPb, ADRb1, ADRb2,ADRb3, FGF21, UCP1, and PRDM16 (C). GAPDH was used to normalize protein levels. Data are mean 6 SEM of 7–8 animals per group.*P < 0.05, **P < 0.01 vs. vehicle. PF, pair-fed.

3352 Liraglutide and Thermogenesis Diabetes Volume 63, October 2014

Figure 4—Effect of the specific injection of liraglutide (10 mg/rat) in the ARC (A), LHA (B), PVH (C ), DMH (D), and VMH (E) on C-FOS, 24-hfood intake, 24-h body weight change, BAT UCP1 levels, and WAT UCP1 levels. Tubulin was used to normalize protein levels. Data aremean 6 SEM of 7–8 animals per group. *P < 0.05, **P < 0.01 vs. vehicle.

diabetes.diabetesjournals.org Beiroa and Associates 3353

Figure 5—A: Effect of 24-h ICV liraglutide (3 mg/mouse) on hypothalamic protein levels of pAMPK, pACC, and ACC. Food intake (B) and bodyweight change (C) in mice receiving an ICV injection of liraglutide (3 mg/mouse) after previous injection with the AMPK activator AICAR (5 mg/mouse). Effect of the injection of adenoviral particles encoding for GFP or pAMPK-CA in the VMH of rats treated with liraglutide (10 mg/rat)(D) and on body weight change (E), BAT UCP1 protein levels (F), and WAT UCP1 protein levels (G). Effect of 24-h ICV liraglutide (3 mg/mouse)in wild type (WT) and SIRT1 transgenic (TgA) mice on food intake (H) and body weight change (I). Effect of 24-h ICV liraglutide (3 mg/mouse) inWT and p53 null (p53 KO) mice on food intake (J) and body weight change (K). GAPDH was used to normalize protein levels. Data are mean6SEM of 7–8 animals per group. *P < 0.05, **P < 0.01, ***P < 0.001 vs. vehicle. KO, knockout; Lig, liraglutide; PF, pair-fed; Veh, vehicle.

3354 Liraglutide and Thermogenesis Diabetes Volume 63, October 2014

GLP-1 activity independently of anorexigenic actions. Wealso observed that obese T2D patients treated withmetformin in combination with exenatide or liraglutide,which showed a reduction in BMI and fat mass, had anincreased energy expenditure, suggesting that at theclinical level, the effects of GLP-1 agonists on body weightare at least partially mediated by increased energyexpenditure.

The mechanisms by which central GLP-1 stimulationreduces body weight seem to involve food intake (7), de-creased lipid storage in WAT (17), and increased BATthermogenesis (18). The diversity of the biological actionsof CNS GLP-1 is likely caused by the wide expression of itsreceptor throughout the entire CNS (36). Therapies forT2D-based GLP-1R agonists are now commonly used incombination with other drugs and are under investigationfor the treatment of human obesity (20). Liraglutide isa long-acting GLP-1 analog indicated for the treatment ofT2D (37,38). In addition to its antidiabetic properties, thesystemic administration of liraglutide induces weight lossin obese rats (39) and patients (20). Because previousstudies found differences in the central actions of GLP-1and long-acting GLP-1 analogs (31), we aimed to investi-gate the central effects of liraglutide on brown and whiteadipocyte metabolism.

In agreement with previous findings (7,17,18), wefound that the central stimulation of GLP-1R reducesfood intake and body weight. Differently from previousstudies that used a chronic activation of the brain GLP-1R(17,18), the present data show that a single ICV injectionof liraglutide maintains its biological actions for 24 h. Thedecreased body weight in mice receiving a single ICV ad-ministration of liraglutide was independent of feedingbehavior. Instead, the results show that increased energyexpenditure, and more specifically BAT thermogenesisand WAT browning, can explain the liraglutide-induced

weight loss. Although previous reports indicated thatbrain GLP-1 induces BAT thermogenesis (18), the capacityof brain GLP-1 to enhance white-to-brown transdifferen-tiation was previously unknown. This effect is likely me-diated by the increased activity of sympathetic fibersinnervating WAT, as previously reported for BAT (18).

Although GLP-1R is widely distributed throughout therodent hypothalamus (36), the specific sites of action andthe molecular pathways triggered by GLP-1 within thehypothalamus remain largely unknown. In the currentstudy, we demonstrate that injection of liraglutide specif-ically into ARC, LHA, or PVH reduced both food intakeand body weight. The reduced food intake observed afterthe administration of liraglutide in the ARC might seemcontroversial because a previous study showed that ARCGLP-1R regulates glucose homeostasis but not feeding(14). However, important methodological differences ex-ist between these studies, such as the compounds used(GLP-1 vs. liraglutide) and the time at which food wasweighed (2 vs. 24 h). Differently from the hypothalamicsites cited, the administration of liraglutide to the VMHsignificantly decreased body weight without concomitantreduction in food intake, suggesting that the stimulationof GLP-1R in this particular area reduces weight in a foodintake–independent manner, which is corroborated bythe significantly higher levels of UCP1 in the BAT andWAT of VMH in liraglutide-treated rats. Indeed, theVMH has been previously demonstrated to be an essentialmodulator of BAT metabolism, with the AMPK pathwayas an important mediator controlling energy dissipation(25,26). Therefore, we hypothesized that the centraleffects of liraglutide on BAT and WAT could be mediatedby AMPK in the VMH. The data show that ICV liraglutideadministration decreased hypothalamic AMPK activity,measured as decreased levels of pAMPK. To investigatewhether activation of AMPK could counteract the effect

Table 1—Anthropometric characteristics and indirect calorimetry data of T2D patients before and after 1 year of antidiabetictreatment initiation

Metformin monotherapy Metformin in combination with GLP-1 agonists StatisticalsignificanceTreatment variable Premetformin Postmetformin Pre-exenatide Postexenatide Preliraglutide Postliraglutide

Sex (male/female) 7/3 7/3 8/3 8/3 10/4 10/4 ns

Age (years) 65 6 11 — 67 6 14 — 66 6 15 — ns

BMI (kg/m2) 33.0 6 6.6 31.4 6 9.5 34.9 6 2.6 32.1 6 1.8* 35.3 6 2.2 32.0 6 2.1* *

Waist (cm) 115 6 13 110 6 19 116 6 10 106 6 15 117 6 11 110 6 16 ns

Fat mass (%) 39.0 6 16.9 37.9 6 9.4 39.5 6 4.3 35.8 6 3.7* 39.2 6 3.6 35.1 6 3.3* *

FFM (%) 63.3 6 17.0 67.1 6 9.3 60.7 6 4.2 64.3 6 3.4* 60.8 6 3.1 64.9 6 3.0* *

REE (kJ/day) 7,670 6 1,030 7,469 6 1,230 7,595 6 832 8,240 6 901 7,623 6 1,014 8,326 6 1,107 ns

REE (kJ/kg/day) 74.9 6 10.0 74.1 6 11.7 75.7 6 8.4 79.1 6 8.7 76.0 6 10.3 80.9 6 11.2 ns

REE (kJ/kg FFM/day) 121.0 6 16.7 114.3 6 18.8 120.6 6 5.1 135.7 6 6.0* 122.4 6 6.9 143.6 6 7.1* *

RQ (vCO2/vO2) 0.85 6 0.02 0.84 6 0.03 0.85 6 0.04 0.85 6 0.05 0.83 6 0.06 0.84 6 0.04 ns

PAL 1.47 6 0.11 1.46 6 0.08 1.45 6 0.15 1.48 6 0.12 1.46 6 0.13 1.47 6 0.16 ns

Data are mean 6 SEM unless otherwise indicated. ns, nonstatistically significant differences by repeated-measures ANOVA, MANOVA,and two-tailed paired Student t test between pre- and postvalues within a same treatment group. *P , 0.05 by repeated-measuresANOVA and compared with baseline (pre) values within the same treatment group by two-tailed paired Student t test.

diabetes.diabetesjournals.org Beiroa and Associates 3355

of liraglutide, we first pharmacologically activated centralAMPK with the compound AICAR and then genetically ac-tivated AMPK specifically in the VMH by using adenoviral-mediated targeting. In both cases, we found that theactions of liraglutide on thermogenesis and browningwere abolished, indicating that VMH AMPK modulatesthe actions of brain GLP-1 on both BAT and WAT. Indeed,it is well established that the GLP-1R is also locatedin extrahypothalamic areas (8,9); therefore, the currentresults do not exclude the possibility that extrahypotha-lamic areas are also involved in the effects of GLP-1Ragonists on BAT thermogenesis and energy expenditure.In particular, GLP-1 binding sites were found in the in-ferior olive and nucleus of the solitary tract (9); both areashave been described as involved in the control of thermo-genesis (40,41). Although at the central level SIRT1/p53and AMPK pathways are interacting to mediate theorexigenic actions of ghrelin (33), neither SIRT1 nor p53modified liraglutide-induced hypophagia or weight loss, in-dicating that the brain SIRT1/p53 system does not interactwith AMPK to mediate the actions of liraglutide on ther-mogenesis and browning.

Finally, because it is known that BAT is active in humans(42–45), we investigated the clinical value of the currentdata and found that obese T2D patients treated with met-formin in combination with the GLP-1R agonists exenatideor liraglutide showed an increase in energy expenditure. TheHbA1c reductions observed were consistent with those fromother studies of exenatide and liraglutide (20,46,47). In thepatient study, both GLP-1 agonists were applied at dosesknown to have good digestive tolerance. Although previousreports showed that liraglutide failed to significantly affectenergy expenditure in patients (46–49), in the currentstudy, a significant increase in the adjusted REE and a re-duction in BMI and the percentage of fat mass were evident.

The apparent discrepancies in the results can be explainedby the length of the treatment period; in previous studies,patients were treated with GLP-1 agonists for 4 (48), 8 (46),

Figure 6—REE in obese T2D patients before and after the treatment for 1 year with metformin (A), metformin and exenatide (B), andmetformin and liraglutide (C ). *P < 0.05.

Figure 7—Schematic overview summarizing the physiological ef-fects of the central injection of liraglutide on energy balance. Activa-tion of GLP-1R in the VMH dephosphorylates AMPK and stimulatesBAT thermogenesis and energy expenditure. Activation of GLP-1R inthe LHA, PVH, and ARC decreases caloric intake. Combined, theseparallel metabolic changes result in a decreased body weight.

3356 Liraglutide and Thermogenesis Diabetes Volume 63, October 2014

or 12 (47) weeks, whereas in the current study, patientswere treated for a mean of 1 year. The 1-year clinical datademonstrate for first time in our knowledge that liraglutide,which is now under evaluation by the Food and Drug Ad-ministration as an antiobesity drug, increases energy expen-diture. Although we cannot rule out the possibility thatliraglutide-induced REE in humans is partially mediated byperipheral mechanisms, the present findings suggest thatliraglutide uses a specific central pathway to increase energyexpenditure and ultimately reduce body weight, and thispathway might be of clinical relevance.

As summarized in Fig. 7, we have provided a combina-tion of pharmacological and genetic evidence to demon-strate that the central stimulation of GLP-1R induces notonly BAT thermogenesis, but also adipocyte browning inWAT. The molecular mechanism controlling these actionsinvolves AMPK in the VMH.

Funding. This work has been supported by grants from Ministerio deEconomia y Competitividad (BFU2011-29102 to C.D. and RYC-2008-02219 andBFU2012-35255 to R.N.), Xunta de Galicia (10PXIB208164PR and 2012-CP070to M.L. and EM 2012/039 and 2012-CP069 to R.N.), Fondo de InvestigacionesSanitarias (PI12/01814 to M.L. and FISPI12/00515 to G.F.), and CIBERobn.CIBERobn is an initiative of the Instituto de Salud Carlos III of Spain, which issupported by FEDER (European Fund for Regional Development) funds. This researchhas also received funding from the European Community’s Seventh FrameworkProgramme under the following grants: to C.D., M.L., and R.N.: FP7/2007-2013:no. 245009: NeuroFAST; and R.N.: ERC StG-2011-OBESITY53-281408.Duality of Interest. No potential conflicts of interest relevant to this articlewere reported.Author Contributions. D.B., M.I., R.G., and A.S. contributed to the experi-ments and data analysis. D.H. contributed to the experiments. F.V., M.S., J.F., J.S.,J.E., C.D., M.L., and G.F. contributed to the development of the analytical tools anddiscussion. R.N. contributed to the experimental design and writing of the manu-script. R.N. is the guarantor of this work and, as such, had full access to all thedata in the study and takes responsibility for the integrity of the data and theaccuracy of the data analysis.

References1. Drucker DJ. The role of gut hormones in glucose homeostasis. J Clin Invest2007;117:24–322. Göke R, Fehmann HC, Göke B. Glucagon-like peptide-1(7-36) amide isa new incretin/enterogastrone candidate. Eur J Clin Invest 1991;21:135–1443. Campbell JE, Drucker DJ. Pharmacology, physiology, and mechanisms ofincretin hormone action. Cell Metab 2013;17:819–8374. Kreymann B, Williams G, Ghatei MA, Bloom SR. Glucagon-like peptide-17-36: a physiological incretin in man. Lancet 1987;2:1300–13045. Mojsov S, Weir GC, Habener JF. Insulinotropin: glucagon-like peptide I(7-37) co-encoded in the glucagon gene is a potent stimulator of insulin releasein the perfused rat pancreas. J Clin Invest 1987;79:616–6196. Willms B, Werner J, Holst JJ, Orskov C, Creutzfeldt W, Nauck MA. Gastricemptying, glucose responses, and insulin secretion after a liquid test meal: ef-fects of exogenous glucagon-like peptide-1 (GLP-1)-(7-36) amide in type 2(noninsulin-dependent) diabetic patients. J Clin Endocrinol Metab 1996;81:327–3327. Turton MD, O’Shea D, Gunn I, et al. A role for glucagon-like peptide-1 in thecentral regulation of feeding. Nature 1996;379:69–728. Shimizu I, Hirota M, Ohboshi C, Shima K. Identification and localization ofglucagon-like peptide-1 and its receptor in rat brain. Endocrinology 1987;121:1076–1082

9. Göke R, Larsen PJ, Mikkelsen JD, Sheikh SP. Distribution of GLP-1 bindingsites in the rat brain: evidence that exendin-4 is a ligand of brain GLP-1 bindingsites. Eur J Neurosci 1995;7:2294–230010. Skibicka KP, Alhadeff AL, Grill HJ. Hindbrain cocaine- and amphetamine-regulated transcript induces hypothermia mediated by GLP-1 receptors. J Neu-rosci 2009;29:6973–698111. Holst JJ. Incretin hormones and the satiation signal. Int J Obes (Lond) 2013;37:1161–116812. Tang-Christensen M, Larsen PJ, Göke R, et al. Central administration ofGLP-1-(7-36) amide inhibits food and water intake in rats. Am J Physiol 1996;271:R848–R85613. Seo S, Ju S, Chung H, Lee D, Park S. Acute effects of glucagon-like peptide-1on hypothalamic neuropeptide and AMP activated kinase expression in fasted rats.Endocr J 2008;55:867–87414. Sandoval DA, Bagnol D, Woods SC, D’Alessio DA, Seeley RJ. Arcuateglucagon-like peptide 1 receptors regulate glucose homeostasis but not foodintake. Diabetes 2008;57:2046–205415. Kinzig KP, D’Alessio DA, Herman JP, et al. CNS glucagon-like peptide-1receptors mediate endocrine and anxiety responses to interoceptive and psy-chogenic stressors [published correction appears in J Neurosci 2003;23:8158].J Neurosci 2003;23:6163–617016. Kinzig KP, D’Alessio DA, Seeley RJ. The diverse roles of specific GLP-1receptors in the control of food intake and the response to visceral illness.J Neurosci 2002;22:10470–1047617. Nogueiras R, Pérez-Tilve D, Veyrat-Durebex C, et al. Direct control of pe-ripheral lipid deposition by CNS GLP-1 receptor signaling is mediated by thesympathetic nervous system and blunted in diet-induced obesity. J Neurosci2009;29:5916–592518. Lockie SH, Heppner KM, Chaudhary N, et al. Direct control of brown adiposetissue thermogenesis by central nervous system glucagon-like peptide-1 re-ceptor signaling. Diabetes 2012;61:2753–276219. Marre M, Shaw J, Brändle M, et al.; LEAD-1 SU study group. Liraglutide,a once-daily human GLP-1 analogue, added to a sulphonylurea over 26 weeksproduces greater improvements in glycaemic and weight control compared withadding rosiglitazone or placebo in subjects with type 2 diabetes (LEAD-1 SU).Diabet Med 2009;26:268–27820. Astrup A, Rössner S, Van Gaal L, et al.; NN8022-1807 Study Group. Effectsof liraglutide in the treatment of obesity: a randomised, double-blind, placebo-controlled study. Lancet 2009;374:1606–161621. Nathan DM, Buse JB, Davidson MB, et al.; American Diabetes Association;European Association for the Study of Diabetes. Medical management of hy-perglycaemia in type 2 diabetes mellitus: a consensus algorithm for the initiationand adjustment of therapy: a consensus statement from the American DiabetesAssociation and the European Association for the Study of Diabetes. Diabetologia2009;52:17–3022. Imbernon M, Beiroa D, Vazquez MJ, et al. Central melanin-concentratinghormone influences liver and adipose metabolism via specific hypothalamicnuclei and efferent autonomic/JNK1 pathways. Gastroenterology 2013;144:636–64923. López M, Lelliott CJ, Tovar S, et al. Tamoxifen-induced anorexia is asso-ciated with fatty acid synthase inhibition in the ventromedial nucleus of thehypothalamus and accumulation of malonyl-CoA. Diabetes 2006;55:1327–133624. Czyzyk TA, Romero-Picó A, Pintar J, et al. Mice lacking d-opioid receptorsresist the development of diet-induced obesity. FASEB J 2012;26:3483–349225. López M, Varela L, Vázquez MJ, et al. Hypothalamic AMPK and fatty acidmetabolism mediate thyroid regulation of energy balance. Nat Med 2010;16:1001–100826. Martínez de Morentin PB, Whittle AJ, Fernø J, et al. Nicotine inducesnegative energy balance through hypothalamic AMP-activated protein kinase.Diabetes 2012;61:807–81727. Moreno-Navarrete JM, Catalán V, Ortega F, et al. Circulating omentinconcentration increases after weight loss. Nutr Metab (Lond) 2010;7:27

diabetes.diabetesjournals.org Beiroa and Associates 3357

28. Frühbeck G, Alonso R, Marzo F, Santidrián S. A modified method for theindirect quantitative analysis of phytate in foodstuffs. Anal Biochem 1995;225:206–21229. Johansson G, Westerterp KR. Assessment of the physical activity level withtwo questions: validation with doubly labeled water. Int J Obes (Lond) 2008;32:1031–103330. Catalán V, Gómez-Ambrosi J, Ramirez B, et al. Proinflammatory cytokines inobesity: impact of type 2 diabetes mellitus and gastric bypass. Obes Surg 2007;17:1464–147431. Barrera JG, D’Alessio DA, Drucker DJ, Woods SC, Seeley RJ. Differences inthe central anorectic effects of glucagon-like peptide-1 and exendin-4 in rats.Diabetes 2009;58:2820–282732. Hwa JJ, Ghibaudi L, Williams P, Witten MB, Tedesco R, Strader CD. Dif-ferential effects of intracerebroventricular glucagon-like peptide-1 on feeding andenergy expenditure regulation. Peptides 1998;19:869–87533. Velásquez DA, Martínez G, Romero A, et al. The central Sirtuin 1/p53 pathwayis essential for the orexigenic action of ghrelin. Diabetes 2011;60:1177–118534. Riediger T, Eisele N, Scheel C, Lutz TA. Effects of glucagon-like peptide 1and oxyntomodulin on neuronal activity of ghrelin-sensitive neurons in the hy-pothalamic arcuate nucleus. Am J Physiol Regul Integr Comp Physiol 2010;298:R1061–R106735. Pfluger PT, Herranz D, Velasco-Miguel S, Serrano M, Tschöp MH. Sirt1protects against high-fat diet-induced metabolic damage. Proc Natl Acad SciU S A 2008;105:9793–979836. Shughrue PJ, Lane MV, Merchenthaler I. Glucagon-like peptide-1 receptor(GLP1-R) mRNA in the rat hypothalamus. Endocrinology 1996;137:5159–516237. Buse JB, Rosenstock J, Sesti G, et al.; LEAD-6 Study Group. Liraglutideonce a day versus exenatide twice a day for type 2 diabetes: a 26-week ran-domised, parallel-group, multinational, open-label trial (LEAD-6). Lancet 2009;374:39–4738. Garber A, Henry R, Ratner R, et al.; LEAD-3 (Mono) Study Group. Liraglutideversus glimepiride monotherapy for type 2 diabetes (LEAD-3 Mono): a randomised,52-week, phase III, double-blind, parallel-treatment trial. Lancet 2009;373:473–481

39. Larsen PJ, Fledelius C, Knudsen LB, Tang-Christensen M. Systemic ad-

ministration of the long-acting GLP-1 derivative NN2211 induces lasting and

reversible weight loss in both normal and obese rats. Diabetes 2001;50:2530–

253940. Kotz CM, Grace MK, Briggs J, Levine AS, Billington CJ. Effects of opioid

antagonists naloxone and naltrexone on neuropeptide Y-induced feeding and

brown fat thermogenesis in the rat. Neural site of action. J Clin Invest 1995;96:

163–17041. Uno T, Shibata M. Role of inferior olive and thoracic IML neurons in non-

shivering thermogenesis in rats. Am J Physiol Regul Integr Comp Physiol 2001;

280:R536–R54642. Wu J, Boström P, Sparks LM, et al. Beige adipocytes are a distinct type of

thermogenic fat cell in mouse and human. Cell 2012;150:366–37643. Virtanen KA, Lidell ME, Orava J, et al. Functional brown adipose tissue in

healthy adults. N Engl J Med 2009;360:1518–152544. Cypess AM, Lehman S, Williams G, et al. Identification and importance of

brown adipose tissue in adult humans. N Engl J Med 2009;360:1509–151745. van Marken Lichtenbelt WD, Vanhommerig JW, Smulders NM, et al. Cold-

activated brown adipose tissue in healthy men. N Engl J Med 2009;360:1500–

150846. Harder H, Nielsen L, Tu DT, Astrup A. The effect of liraglutide, a long-acting

glucagon-like peptide 1 derivative, on glycemic control, body composition, and

24-h energy expenditure in patients with type 2 diabetes. Diabetes Care 2004;

27:1915–192147. Bradley DP, Kulstad R, Racine N, Shenker Y, Meredith M, Schoeller DA.

Alterations in energy balance following exenatide administration. Appl Physiol

Nutr Metab 2012;37:893–89948. Horowitz M, Flint A, Jones KL, et al. Effect of the once-daily human GLP-1

analogue liraglutide on appetite, energy intake, energy expenditure and gastric

emptying in type 2 diabetes. Diabetes Res Clin Pract 2012;97:258–26649. Tan TM, Field BC, McCullough KA, et al. Coadministration of glucagon-like

peptide-1 during glucagon infusion in humans results in increased energy ex-

penditure and amelioration of hyperglycemia. Diabetes 2013;62:1131–1138

3358 Liraglutide and Thermogenesis Diabetes Volume 63, October 2014