SGK1/FOXO3 Signaling in Hypothalamic POMC NeuronsMediates Glucocorticoid-Increased AdiposityYalan Deng,1 Yuzhong Xiao,1 Feixiang Yuan,1 Yaping Liu,2 Xiaoxue Jiang,1 Jiali Deng,1 Geza Fejes-Toth,3

Aniko Naray-Fejes-Toth,3 Shanghai Chen,1 Yan Chen,1 Hao Ying,1 Qiwei Zhai,1 Yousheng Shu,2 andFeifan Guo1

Diabetes 2018;67:569–580 | https://doi.org/10.2337/db17-1069

Although the central nervous system has been implicatedin glucocorticoid-induced gain of fat mass, the underlyingmechanisms are poorly understood. The aim of this studywas to investigate the possible involvement of hypotha-lamic serum- and glucocorticoid-regulated kinase 1 (SGK1)in glucocorticoid-increased adiposity. It is well known thatSGK1 expression is induced by acute glucocorticoid treat-ment, but it is interesting that we found its expression tobe decreased in the arcuate nucleus of the hypothalamus,including proopiomelanocortin (POMC) neurons, followingchronic dexamethasone (Dex) treatment. To study the roleof SGK1 in POMC neurons, we produced mice that de-veloped or experienced adult-onset SGK1 deletion in POMCneurons (PSKO). As observed in Dex-treated mice, PSKOmice exhibited increased adiposity and decreased en-ergy expenditure. Mice overexpressing constitutively ac-tive SGK1 in POMC neurons consistently had the oppositephenotype and did not experience Dex-increased adiposity.Finally, Dex decreased hypothalamic a-melanocyte-stimu-lating hormone (a-MSH) content and its precursor Pomcexpression via SGK1/FOXO3 signaling, and intracere-broventricular injection of a-MSH or adenovirus-mediatedFOXO3 knockdown in the arcuate nucleus largely reversedthe metabolic alterations in PSKO mice. These resultsdemonstrate that POMC SGK1/FOXO3 signaling mediatesglucocorticoid-increased adiposity, providing new insightsinto the mechanistic link between glucocorticoids and fataccumulation and important hints for possible treatmenttargets for obesity.

Despite the overwhelming beneficial anti-inflammatoryeffects of glucocorticoid, chronic glucocorticoid treatmenthas been shown to cause numerous adverse metabolicoutcomes, including fat mass gain (1). Recent studies haveelucidated several peripheral mechanisms underlying gluco-corticoid-induced increase in fat mass. For example, gluco-corticoid induces adipocyte differentiation (1–3), alters lipidmetabolism in adipose tissue (1–3), and inhibits browningof white adipose tissue (WAT) or thermogenesis of brownadipose tissue (BAT) (4,5). In fact, body fat mass is largelycontrolled by the central nervous system (CNS) (6–8). Spe-cific populations of neurons in the arcuate nucleus (ARC) ofthe hypothalamus also play fundamental roles in the regu-lation of energy balance and lipid metabolism (6–8). Inparticular, neurons coexpressing the orexigenic neuropeptidesagouti-related protein and neuropeptide Y, and neurons coex-pressing the anorexigenic proopiomelanocortin (POMC)precursor and the cocaine- and amphetamine-related tran-script, are extensively involved in the regulation of appetite,body weight, and metabolism (6–8). POMC is a pro-tein expressed and secreted from POMC neurons andcleaved by prohormone convertases to produce a-melanocyte-stimulating hormone (a-MSH) (8). a-MSH binds to themelanocortin 4 receptor and functions as a key hub linkingthe CNS to peripheral organs through the sympathetic ner-vous system (SNS), whereas dysfunction of this signalingaxis leads to obesity in mice and humans (9,10). Activationof the SNS promotes the release of norepinephrine (NE)

1Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences,Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Uni-versity of Chinese Academy of Sciences, Shanghai, China2State Key Laboratory of Cognitive Neuroscience and Learning and IDG/McGovernInstitute for Brain Research, School of Brain and Cognitive Sciences, BeijingNormal University, Beijing, China3Department of Physiology, Geisel School of Medicine, Dartmouth College, Leb-anon, NH

Corresponding authors: Feifan Guo, ffguo@sibs.ac.cn, and Yousheng Shu,yousheng@bnu.edu.cn.

Received 6 September 2017 and accepted 2 January 2018.

This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db17-1069/-/DC1.

© 2018 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, and thework is not altered. More information is available at http://www.diabetesjournals.org/content/license.

Diabetes Volume 67, April 2018 569

METABOLISM

that binds to b-adrenergic receptor 3 and stimulates WATlipolysis and BAT thermogenesis (11–14). Although previousstudies have shown that glucocorticoid regulates food intakeand energy expenditure (15,16), the central signals mediat-ing the effect of glucocorticoid are poorly understood.

Serum- and glucocorticoid-regulated kinase 1 (SGK1)belongs to the family of serine/threonine kinases, and itscoding region was originally isolated from rat mammarytumor cells (17). SGK1 is ubiquitously expressed in varioustissues, including hypothalamus (17), and functions via ac-tivation of the glucocorticoid receptor (GR), retinoid Xreceptor, peroxisome proliferator–activated receptor g,and nuclear factor kB (17). SGK1 is involved in the regula-tion of many processes, including hypertension, epithelialsodium channel activity, and insulin sensitivity (17–19).SGK1 also mediates many important functions of glucocor-ticoids, including insulin secretion and hippocampal neuro-genesis (20,21). Although extensive studies have beencarried out, a role for hypothalamic SGK1 in the regulationof energy homeostasis is unknown. Furthermore, SGK1 iswell known as an early-response gene that can be inducedby acute glucocorticoid treatment in various cells and ani-mal models (20–22); however, the effect of chronic gluco-corticoid treatment on SGK1 expression remains largelyunknown. In fact, the expression of SGK1 in the contextof glucocorticoid-induced metabolic effects could be veryimportant, as studies show that sometimes SGK1 mayhave effects opposing those of glucocorticoid (23).

Despite the aforementioned unknown facts, we canspeculate that SGK1, as a downstream target of glucocor-ticoid (17) expressed in the hypothalamus (17), may con-tribute to the central action of glucocorticoid. Therefore, theaim of this study was to test this hypothesis first by determiningthe expression of SGK1 in the hypothalamus and then byinvestigating its possible contribution to glucocorticoid-increased adiposity.

By creating mice that develop or experience adult-onsetknockout of SGK1, or overexpression of SGK1 in POMCneurons, we demonstrate a crucial role for SGK1 expressedin POMC neurons in glucocorticoid-increased adiposity andprovide a novel mechanistic link between glucocorticoidtreatment and body fat mass gain.

RESEARCH DESIGN AND METHODS

Mice and DietsPOMC-Cre (24) and tamoxifen-inducible POMC-cre(POMC-cre:ERT2) (24) mice were kindly provided by JoelK. Elmquist and Tiemin Liu from University of Texas South-western Medical Center (Dallas, TX); floxed SGK1 allele(SGK1loxp/loxp) mice (18) were provided by Dr. Geza Fejes-Toth and Dr. Aniko Naray-Fejes-Toth (Geisel School ofMedicine, Dartmouth College, Hanover, NH). To generatePOMC neuron–specific SGK1 knockout (PSKO) mice,POMC-Cre mice were crossed with SGK1loxp/loxp mice. Togenerate inducible POMC-specific SGK1 knockout mice,POMC-cre:ERT2 mice were crossed with SGK1loxp/loxp mice.

Tamoxifen (0.15 g/kg; Sigma-Aldrich, St. Louis, MO) sus-pended in corn oil (Sigma-Aldrich) was intraperito-neally injected into 8-week-old male SGK1loxp/loxp orSGK1loxp/loxp 3 POMC-cre:ERT2 littermate mice for five con-secutive days to generate mice with adult-onset SGK1 de-letion in POMC neurons (PSKO-ER). Dexamethasone (Dex)treatment was administered to male wild-type (WT), control,and PSKO mice, and to male POMC-Cre mice injected withadeno-associated virus (AAV) expressing constitutively activemutant rat SGK1 (S422D) (AAV-CA SGK1)/AAV-null in ARC(control/POMC neuron–specific SGK1 overexpression [PSOE]mice) by intraperitoneal injection of PBS or 5 mg/kg Dex everyother day for 6 weeks or for 2 h (1,25). All the mice were ona C57BL/6J background. Mice were maintained under a 12-hlight/12-h dark cycle (lights on at 0700 h/lights off at 1900 h)at 25°C, with free access to water and standard chow diet. Invivo studies were conducted in accordance with the guidelinesof the Institutional Animal Care and Use Committee of ShanghaiInstitute for Nutritional Sciences, Chinese Academy of Sciences.

Intracerebroventricular Cannulation and ARCAdministration ExperimentsIntracerebroventricular (ICV) cannulation experiments wereconducted as previously described (26). After surgery, the micewere given 5 days to recover and were then infused with2 mL a-MSH peptide (Abcam, Cambridge, U.K.) at a concen-tration of 1 nmol/mL or 2 mL artificial cerebrospinal fluid(Tocris, Bristol, U.K.), and experiments were conducted 24 hlater. ARC administration experiments were conducted as pre-viously described (6). Mice were anesthetized and receivedbilateral stereotaxic injections of adenovirus expressingFOXO3-specific short hairpin RNA against FOXO3 (Ad-shFOXO3) or scrambled control adenovirus (Ad-scrambled),AAV-CA SGK1, or control AAV-null into the ARC (1.5 mmposterior to the bregma, 60.2 mm lateral to the midline,and 6 mm below the surface of the skull). The AAV-CA SGK1expression plasmid was constructed in pAAV-Ef1a-DIO-mCherry-2A plasmid (Addgene, Cambridge, MA), and SGK1started to express only in the presence of CRE recombinase.

Metabolic Parameter MeasurementsThe body composition of mice was measured with a nuclearmagnetic resonance system (Bruker, Rheinstetten, Ger-many). Indirect calorimetry was performed in a compre-hensive laboratory animal-monitoring system (ColumbusInstruments, Columbus, OH), as previously described(27). Rectal temperature of mice was measured at 1400 and1700 h with a rectal probe attached to a digital thermom-eter (Physitemp Instruments Inc., Clifton, NJ). Food intakewas measured as reported previously (6).

POMC Neuron Identification, Count, and AreaAI9 (tdTomato) reporter mice (The Jackson Laboratory)were mated with, or AAV-CA SGK1 and AAV-null expressedmCherry red fluorescent protein were injected into the ARCof, POMC-Cre mice to reflect POMC neurons, demon-strated by co-localization with POMC antibodies. The dis-tribution and number of POMC neurons were determined

570 POMC SGK1/FOXO3 Signaling Regulates Adiposity Diabetes Volume 67, April 2018

as described previously (6). Average somatic area was ana-lyzed in .500 POMC neurons (n = 4 mice/genotype). Thearea occupied by POMC neurons was manually scored usingImageJ software.

Hypothalamic a-MSH Protein ContentHypothalamus was prepared as previously described (6),and a-MSH was quantified with an ELISA kit (PhoenixPharmaceuticals Inc., Burlingame, CA), according to themanufacturer’s instructions.

Hypothalamic Nuclear and Cytoplasmic FractionsHypothalamic nuclear and cytoplasmic fractions were iso-lated as previously described (28).

Immunofluorescence StainingImmunofluorescence staining was performed, as previouslydescribed (29), with anti-SGK1 and the anti-p-N-mycdownstream-regulated gene 1 (p-NDRG1) (Abcam), anti-POMC(Phoenix Pharmaceuticals Inc.), anti-FOXO3 (Cell Signaling

Technology, Danvers, MA), anti-p-SGK1 and anti-GR (SantaCruz Biotechnology, Santa Cruz, CA), and anti-a-MSH (MerckMillipore, Frankfurter, Germany). Immunofluorescence stainingof p-FOXO3 was performed using the Tyramide Signal Ampli-fication Cyanine 3 system (Perkin-Elmer, Boston, MA), andanti-p-FOXO3 primary antibody (Cell Signaling Technology) wascoincubated with anti-mCherry (Abbkine, Inc., Wuhan, China).

RNA Isolation and Relative Quantitative RT-PCRRNA was isolated and RT-PCR performed as previouslydescribed (27). The sequences of primers used in this studyare available upon request.

Western Blotting AnalysisWestern blotting was analyzed, as previously described (27),with the following primary antibodies: anti-p-FOXO3, anti-FOXO3, anti–lamin B1, and anti-p-GR (Cell Signaling Tech-nology); anti-SGK1 and anti-GR (Abcam); anti–uncouplingprotein 1 (UCP1) and anti-p-SGK1 (Santa Cruz Biotechnol-ogy); and anti-a-tubulin and anti-b-actin (Sigma-Aldrich).

Figure 1—SGK1 expression in hypothalamic POMC neurons under chronic or acute Dex treatment. A: Sgk1 expression in the hypothalamusunder chronic Dex treatment. B: Western blotting (left) and densitometric quantification (right) of SGK1 and p-SGK1 in the hypothalamus. C andD: Immunofluorescence for SGK1 in ARC sections (C) and integrated density quantification (D). E and F: Immunofluorescence for p-SGK1 inARC sections (E) and integrated density quantification (F). G and H: Immunofluorescence for POMC neurons (red), SGK1 (green), and a merge(yellow) in ARC sections (G) and integrated density quantification in POMC neurons and colocalization (H). I: Sgk1 expression in the hypothal-amus under acute Dex treatment. J: Western blotting (left) and densitometric quantification (right) of SGK1 in the hypothalamus under acute Dextreatment. K and L: Immunofluorescence for POMC neurons (red), SGK1 (green), and a merge (yellow) in ARC sections (K) and integrated densityquantification in POMC neurons and colocalization (L) under acute Dex treatment. Studies were conducted using 14- to 15-week-old male WTmice (A–F), or POMC-tdTomato indicator mice (G and H) treated without Dex (2 Dex) or with Dex (+ Dex) every other day for 6 weeks, or in9-week-old male WT mice (I and J) or POMC-tdTomato indicator mice (for K and L) treated without Dex (2 Dex) or with Dex (+ Dex) for 2 h. Dataare expressed as the mean 6 SEM (n = 6–11 mice/group). *P , 0.05 for the effect of Dex treatment vs. no Dex treatment.

diabetes.diabetesjournals.org Deng and Associates 571

Isolation and Treatment of Primary HypothalamicNeuronsPrimary cultures of hypothalamic neurons were prepared aspreviously described (27). On day 7, primary cultured hy-pothalamic neurons were infected with adenovirus express-ing SGK1-specific short hairpin RNA (108 plaque-formingunits/cells on 60 cm2) or Ad-scrambled, constructed as de-scribed previously (19). Primary hypothalamic neurons weretransfected with small interfering RNA for FOXO3 usingX-tremeGENE siRNA Transfection Reagent (Roche Diagnos-tics, Mannheim, Germany). Constitutively active mutant ratSGK1 (S422D) was subcloned into a PCMV-MYC plasmidand transfected into primary cultured hypothalamic neuronsusing Lipofectamine 2000 (Life Technologies).

Statistical AnalysisAll values are presented as the mean 6 SEM. Differencesbetween groups were analyzed by either the Student t testor one-way ANOVA followed by the Student-Newman-Keulstest. Differences for which P was ,0.05 were consideredstatistically significant.

RESULTS

Chronic Dex Treatment Decreases SGK1 Expressionin Hypothalamic POMC NeuronsTo investigate the metabolic effects of Dex, C57B6J WTmice were intraperitoneally injected with Dex for 6 weeks;this model has been commonly used to study the role ofDex (1,30). Dex treatment did not change body weight,although the total body fat and abdominal fat mass wereincreased compared with those in mice receiving the controltreatment, possibly as a result of decreased lean mass

(Supplementary Fig. 1A–D). A balance between energy in-take and energy expenditure maintains body fat mass (7).Dex treatment did not change food intake but did decreaseenergy expenditure as measured by 24-h indirect calorime-try (Supplementary Fig. 1E and F). No difference was ob-served in locomotor activity, but body temperature, levels ofthe BAT thermogenic marker UCP1 (11), and levels of se-rum NE were significantly lower in Dex-treated mice (Sup-plementary Fig. 1G–J).

To investigate the possible involvement of hypothalamicSGK1 in Dex-increased adiposity, we examined hypotha-lamic SGK1 expression under this condition; it is interestingthat we found that hypothalamic SGK1 and phosphorylated(p-)SGK1 were decreased in Dex-treated mice (Fig. 1A andB). Furthermore, immunofluorescence staining showedthat SGK1 and p-SGK1 were decreased in the ARC ofthe hypothalamus of Dex-treated mice (Fig. 1C–F). Immu-nofluorescence staining of tdTomato and SGK1 showedthat SGK1 was colocalized with POMC neurons in PBS-treated mice but was decreased significantly in POMC neu-rons of Dex-treated mice (Fig. 1G and H). By contrast, acutetreatment increased SGK1 expression in the ARC of thehypothalamus (Fig. 1I–L).

Deletion of SGK1 in POMC Neurons Causes Obesityand Decreases Energy ExpenditureBased on the above results, we speculated that knockout ofSGK1 expression in POMC neurons might mimic Dex-induced metabolic alterations. To test this hypothesis, wegenerated PSKO mice. Immunofluorescence staining oftdTomato and SGK1 showed that SGK1 was colocalizedwith POMC neurons (.90% overlap with tdTomato) in

Figure 1—Continued.

572 POMC SGK1/FOXO3 Signaling Regulates Adiposity Diabetes Volume 67, April 2018

control mice but was almost absent in POMC neurons ofPSKO mice (Fig. 2A and Supplementary Fig. 2A–C), with nodifference in SGK1 staining in the paraventricular nucleus(PVN) and ventromedial hypothalamus (VMH) betweenPSKO and control mice (Supplementary Fig. 2D–G). Sgk1mRNA levels were consistently decreased ;50% in theARC, as the ARC of PSKOmice have other neurons or neuro-gliocytes (31,32), but Sgk1 mRNA levels were unchanged inother brain areas and tissues (Fig. 2B). Anatomical assess-ment of POMC neurons throughout the ARC area revealedno significant alterations in neuronal population size, dis-tribution, or somatic area (Supplementary Fig. 3A and B),indicating that SGK1 deficiency did not alter POMC neurondifferentiation and/or survival. Because the POMC pro-moter also drives CRE recombinase expression in the pituitary(33), we examined serum contents of hormones secreted fromthe pituitary, including corticosterone and growth hormone(34), and found that the levels of these two hormones werenot altered in PSKO mice (Supplementary Fig. 3C and D).

Male PSKO mice exhibited significantly increased bodyweight from the age of 9 weeks compared with control mice(Fig. 2C); this was accompanied by a significant increase intotal body fat and abdominal fat mass (Fig. 2D and E) butunchanged lean mass (Supplementary Fig. 2H). Food intake

was not altered, but the energy expenditure was markedly de-creased and the respiratory exchange ratio (RER; VCO2/VO2)was higher in PSKO mice (Fig. 2F–H). Again, locomotoractivity was not changed, but body temperature, UCP1 inBAT, and serum NE levels were significantly lower in PSKOmice (Fig. 2I–L). Female PSKO mice displayed phenotypessimilar to those observed in male mice (Supplementary Fig.4), so we performed all of the subsequent studies in malemice.

Inducible Loss of SGK1 in POMC Neurons in Adult MiceRecapitulates Aberrant Energy HomeostasisWe next asked whether adult-onset loss of SGK1 inPOMC neurons had effects similar to those of ablationduring development. We used the POMC-cre:ERT2 mousemodel (24) that allows temporal control of CRE recombi-nase activity and can be combined with SGK1flox/flox mice toproduce mice with adult-onset deletion of SGK1 (PSKO-ER).We observed phenotypes similar to those found when usingconstitutive POMC-Cre mice (Supplementary Fig. 5).

Mice With SGK1 Overexpression in POMC Neurons AreLean and Resistant to Dex-Induced Fat AccumulationWe then asked whether overexpression of SGK1 in POMCneurons in mice would have the opposite phenotype of

Figure 2—PSKO mice exhibit an obese phenotype and decreased energy expenditure the same as Dex-treated WT mice. A: Immunofluores-cence for POMC neurons (red), SGK1 (green), and a merge (yellow) in ARC sections from male POMC-tdTomato indicator mice. B: Sgk1expression in different tissues. COR, cortex; LV, liver. C: Body weight curve. Graphs show total body fat mass (D), abdominal fat mass (E), dailyfood intake (F), daily energy expenditure (EE) (G), daily RER (VCO2/VO2) (H), daily locomotor activity (I), and basal rectal temperature (J). K: Westernblotting (top) and densitometric quantification (bottom) of UCP1 in BAT. L: Serum NE. All studies were conducted in 12- to 14-week-old malecontrol (2 PSKO) and PSKO (+ PSKO) mice. Data are expressed as the mean 6 SEM (n = 6–16 mice/group). *P , 0.05 for the effect of thePSKO group vs. the control group.

diabetes.diabetesjournals.org Deng and Associates 573

that observed in PSKO mice and whether Dex-increasedadiposity would be avoided. For this purpose, we generatedPSOE mice by bilateral stereotaxic injection into the ARCof AAV-CA SGK1 or control AAV-null male POMC-Cremice. The effect of SGK1 overexpression was validated byimmunofluorescence staining of the phosphorylated NDRG1,which reflects the activation status of SGK1 (35) (Fig. 3A),and increased SGK1 signals in POMC neurons (.90% overlapwith mCherry), but not in the PVN and VMH, of PSOE mice(Supplementary Fig. 6A–H). As predicted, body weight de-creased (starting from 6 weeks after AAV injection), accom-panied by a decrease in total body fat and abdominal fat mass,in PSOE mice (Fig. 3B–D). Food intake was not affected, butthe energy expenditure was increased and RER was de-creased in PSOE mice (Fig. 3E–G). No difference was ob-served in locomotor activity, but body temperature, UCP1in BAT, and serum NE levels were increased in PSOE mice(Fig. 3H–K). Furthermore, PSOE mice were resistant toDex-induced fat accumulation and other metabolic alterations(Fig. 4), and Dex injected 5 weeks after AAV injection createdno difference in lean mass and fat mass between control andPSOE mice (Supplementary Fig. 6I and J). By contrast, Dexhad a very mild effect on PSKO mice, as demonstrated bythe slightly decreased body weight and lean mass as well as

increased fat mass, and no significant effect on food intakeor energy expenditure (Supplementary Fig. 7).

Dex Decreases Hypothalamic a-MSH Content via SGK1,and Administration of a-MSH Reverses ObesePhenotype in PSKO MiceBecause previous studies have shown that a-MSH playsa critical role in the regulation of energy homeostasis(33), we asked whether it might be involved in Dex-induced metabolic alterations. As predicted, a dramaticreduction of a-MSH staining was observed in PVN of Dex-treated mice (Fig. 5A and B). Similar results were obtainedin PSKO mice (Fig. 5C and D). The amount of a-MSH, asanalyzed by ELISA, was consistently significantly decreasedin the hypothalamus of PSKO mice (Fig. 5E). Notably, Dex-reduced a-MSH staining was reversed in PSOE mice (Fig.5F and G).

To investigate whether a-MSH could mediate SGK1 reg-ulation of energy homeostasis, we administered a-MSHpeptide ICV to PSKO or control mice. ICV injection ofa-MSH to PSKO mice markedly reduced body weight andabdominal fat mass and increased rectal temperature com-pared with those values in mice injected with the vehicle(control; Fig. 5H–J). ICV injection of a-MSH in PSKO mice

Figure 3—PSOE mice show a lean phenotype and increased energy expenditure. A: Immunofluorescence for POMC neurons (red), p-NDRG1(green), and a merge (yellow) in ARC sections. Graphs show a body weight curve (B), total body fat mass (C), abdominal fat mass (D), daily foodintake (E), daily energy expenditure (EE) (F), daily RER (VCO2/VO2) (G), daily locomotor activity (H), and basal rectal temperature (I). J: Westernblotting (top) and densitometric quantification (bottom) of UCP1 in BAT. K: Serum NE. All studies were conducted in 19- to 20-week-old malecontrol (2 PSOE) and PSOE (+ PSOE) mice. Data are expressed as the mean6 SEM (n = 6–9 mice/group). *P, 0.05 for the effect of the PSOEgroup vs. the control group.

574 POMC SGK1/FOXO3 Signaling Regulates Adiposity Diabetes Volume 67, April 2018

also blocked a UCP1 protein decrease (Fig. 5K). Similareffects were observed in control mice after ICV injectionof a-MSH (Fig. 5H–K).

Dex Reduces a-MSH Precursor POMC Expression Viathe SGK1/FOXO3-Dependent Pathway, andDownregulation of FOXO3 Largely Reverses the ObesityPhenotype in PSKO Micea-MSH levels are determined by the levels of its precursorPOMC and the expression of prohormone convertases thatare responsible for cleaving POMC to a-MSH (8). The re-duced a-MSH concentration in Dex-treated mice did notseem to be a consequence of decreased expression of process-ing enzymes, including prohormone convertase 1 (Pc1/3),prohormone convertase 2 (Pc2), carboxypeptidase E (Cpe),a-amidating monooxygenase (Pam), and prolylcarboxypep-tidase (Prcp) (8), as gene expression of these enzymes wasunchanged (Fig. 6A). On the other hand, POMC expressionwas decreased in Dex-treated mice (Fig. 6A–C). Similarresults were obtained in PSKO mice (Supplementary Fig.8A–C). The effect of Dex on reducing POMC expression,however, was reversed by overexpression of SGK1 (Fig. 6Dand E). Similarly, SGK1 knockdown decreased Pomc expres-sion and SGK1 overexpression increased Pomc expressionin primary cultured hypothalamic neurons (SupplementaryFig. 8D and E).

We then investigated the downstream signaling of SGK1in mediating Dex-decreased POMC expression. Previousstudies showed that SGK1 phosphorylates FOXO3 (36) andthat another member from the same FOXO family, FOXO1,inhibits Pomc expression (37), suggesting that FOXO3might have a function similar to that of FOXO1 down-stream of SGK1 in Dex-induced metabolic alterations.Consistent with this possibility, hypothalamic FOXO3 phos-phorylation was decreased in Dex-treated mice (Fig. 6F).Similar reduction was observed in the hypothalamic ARCof PSKO mice (Supplementary Fig. 9A). Furthermore, Dex-decreased hypothalamic FOXO3 phosphorylation was re-versed in PSOE mice (Fig. 6G and H). Similar regulatoryeffects of SGK1 on p-FOXO3 were observed in primarycultures of hypothalamic neurons (Supplementary Fig. 9Band C).

The inhibitory effect of SGK1 knockdown on Pomc ex-pression was reversed by small interfering RNA–mediatedFOXO3 inhibition (Supplementary Fig. 9D), therebyprompting us to investigate the in vivo function of FOXO3downstream of SGK1. For this purpose, we knocked downFOXO3 expression in the ARC of PSKO and control mice byARC administration (6) of adenovirus expressing short hair-pin RNA directed against the coding region of FOXO3 (Ad-shFOXO3) (38) or Ad-scrambled. The effect of Ad-shFOXO3was demonstrated through decreased expression of Foxo3

Figure 4—PSOEmice are resistant to Dex-induced fat accumulation and decreased energy expenditure. A: Body weight curve. B: Total body fatmass. C: abdominal fat mass. D: Daily food intake. E: Daily energy expenditure (EE). F: Daily RER (VCO2/VO2).G: Daily locomotor activity. H: Basalrectal temperature. I: Western blotting (left) and densitometric quantification (right) of UCP1 in BAT. J: Serum NE. All studies were conducted in19- to 20-week-old male control (2 PSOE) and PSOE (+ PSOE) mice treated with Dex (+ Dex). Data are expressed as the mean6 SEM (n = 6–12mice/group). *P , 0.05 for the effect of the PSOE group vs. the control group.

diabetes.diabetesjournals.org Deng and Associates 575

and the corresponding change in Pomc expression in theARC of PSKO mice (Fig. 6I). Immunofluorescence consis-tently showed that FOXO3 was decreased in the ARC, butnot the PVN or VMH, of these mice (Supplementary Fig.10A and B). Ad-shFOXO3 decreased the body weight, totalbody fat, and abdominal fat mass in PSKO mice (Fig. 6J–L).Although food intake was not affected (Supplementary Fig.10C), the decreased energy expenditure and increased RERin PSKO mice were largely reversed by Ad-shFOXO3 (Fig.6M and N). No significant difference in locomotor activitywas detected (Supplementary Fig. 10D); however, the de-creased body temperature, UCP1 in BAT, and serum NE inPSKO mice were upregulated by Ad-shFOXO3 (Fig. 6O–Q).Moreover, the reduced a-MSH staining in PSKO mice wasalso blocked by Ad-shFOXO3 (Supplementary Fig. 10E andF). Except for the unaltered body weight, similar effectswere observed in control mice following administration ofAd-shFOXO3 (Fig. 6I–Q and Supplementary Fig. 10).

Because glucocorticoid functions via the GR (28), we in-vestigated the spatial regulation of GR and SGK1/FOXO3using previously validated GR antibodies (39). Althoughhypothalamic Gr mRNA was unchanged, total GR and

phosphorylated GR expression were significantly decreasedin Dex-treated mice (Supplementary Fig. 11A and B). Fur-thermore, these three proteins were all expressed in POMCneurons and hypothalamic nuclear p-GR was decreased andFOXO3 was increased, but the cytoplasmic total and phos-phorylated proteins examined were all decreased, in Dex-treated mice (Supplementary Fig. 11C and D).

DISCUSSION

Fat mass accumulation is a serious side effect of glucocor-ticoid therapy (1). Recent studies have elucidated severalperipheral mechanisms underlying glucocorticoid-inducedgains in fat mass (1–5). In this study we demonstrateda novel central mechanism, mediated by SGK1, underlyingglucocorticoid-increased adiposity. SGK1 is a well-knowndownstream target of Dex (20–22). It has been widely dem-onstrated that acute Dex treatment induces SGK1 (20–22).It is interesting to note that we found decreased SGK1expression in POMC neurons in the ARC of Dex-treatedmice. The importance of POMC SGK1 in mediating Dex-induced adiposity was demonstrated by the observation thatknockout of SGK1 expression in POMC neurons increased

Figure 5—Dex decreases hypothalamic a-MSH content via SGK1, and ICV administration of a-MSH reverses the obese phenotype in PSKOmice. A and B: Immunofluorescence for a-MSH in PVN sections (A) of, and integrated density quantification (B) in, 14- to 15-week-old male WTmice treated with no Dex (2 Dex) or with Dex (+ Dex). C–E: Immunofluorescence for a-MSH in PVN sections (C), integrated density quanti-fication (D), and relative hypothalamic a-MSH content based on ELISA (E) in 12- to 14-week-old male control and PSKO mice. F and G:Immunofluorescence for a-MSH in PVN sections (F) and integrated density quantification (G) in 19- to 20-week-old male control and PSOE micetreated with no Dex (2 Dex) or with Dex (+ Dex). H–K: Body weight (H), abdominal fat mass (I), basal rectal temperature (J), and Western blotting(top) and densitometric quantification (bottom) of UCP1 in BAT (K) in 10- to 12-week-old male control (2 PSKO) and PSKO (+ PSKO) micetreated with no a-MSH (2 a-MSH) or with a-MSH (+ a-MSH). Data are expressed as the mean 6 SEM (n = 6–8 mice/group). *P , 0.05 for anytreatment compared with the control group (A–E). *P , 0.05 for the effect of any group vs. control mice treated with no Dex; #P , 0.05 for theeffect of PSOE mice vs. control mice after Dex treatment (G). *P, 0.05 for the effect of any group vs. control mice treated with no a-MSH; #P,0.05 for the effect of treatment with vs. without a-MSH in PSKO mice (H–K).

576 POMC SGK1/FOXO3 Signaling Regulates Adiposity Diabetes Volume 67, April 2018

adiposity, whereas overexpression of SGK1 in POMC neuronsresulted in a lean phenotype and prevented Dex-induced fatmass gain in mice. Furthermore, the Dex-induced gain in fatmass was much lower in PSKO than in control mice. Fat masscould, however, still be increased by Dex treatment in PSKOmice, suggesting the existence of other central or peripheralsignals involved in Dex-increased adiposity (40). Our studyprovides a novel mechanistic link between glucocorticoidtreatment and fat mass gain. This is important for under-standing the mechanisms of glucocorticoid-induced meta-bolic phenotypes and provides an important hint for apossible treatment target for glucocorticoid-induced sideeffects. In addition, our study reports an unrecognized novelfunction of SGK1 in POMC neurons of the hypothalamus inthe regulation of energy homeostasis. These results are im-portant for understanding the signals in specific neuronsthat are critical for metabolic control.

A balance between energy intake and energy expendituremaintains body fat mass (7). BAT oxidizes fat to produceheat via increased expression of uncoupled proteins, whichis stimulated by the activation of the SNS. Deletion of UCP1induces obesity and upregulation of UCP1 increases thermo-genesis and energy expenditure in mice (11). Other studiesalso showed that disruption of SNS activity has a significantnegative impact on energy expenditure (6,41,42). Our studyshowed that Dex increased adiposity mainly by decreasingenergy expenditure, as food intake was not changed in Dex-treated mice. Furthermore, the decrease in energy expendi-ture resulting from Dex treatment was most likely due todecreased thermogenesis in BAT, as demonstrated by thedecreased body temperature, UCP1 expression in BAT, andserum NE in these mice. Lipolysis in WAT is also regulatedby SNS activity (41,42), which might also affect Dex-induced adiposity and should be studied in the future.

Figure 6—Dex reduces a-MSH precursor POMC expression via the SGK1/FOXO3-dependent pathway, and downregulation of FOXO3 in theARC largely reverses the obesity phenotype in PSKO mice. A–C: Neuropeptide expression in the hypothalamus (A), immunofluorescence forPOMC in ARC sections (B), and integrated density quantification (C) in 14- to 15-week-old male WT mice treated with no Dex (2 Dex) or withDex (+ Dex). D and E: Immunofluorescence for POMC in ARC sections (D) and integrated density quantification (E) in 19- to 20-week-old malecontrol and PSOE mice treated with no Dex (2 Dex) or with Dex (+ Dex). F: Western blotting (top) and densitometric quantification (bottom) ofp-FOXO3 and FOXO3 in the hypothalamus in 14- to 15-week-old male WT mice treated with no Dex (– Dex) or with Dex (+ Dex). G and H:Immunofluorescence for POMC neurons (red), p-FOXO3 (green), and a merge (yellow) in ARC sections (G) and integrated density quantificationin POMC neurons and colocalization (H) in 19- to 20-week-old male control and PSOEmice treated with no Dex (2 Dex) or with Dex (+ Dex). I–Q:The expression of Sgk1, Foxo3, and Pomc in the ARC (I), body weight (J), total body fat mass (K), abdominal fat mass (L), daily energyexpenditure (EE) (M), daily RER (VCO2/VO2) (N), basal rectal temperature (O), Western blotting (top) and densitometric quantification (bottom)of UCP1 in BAT (P), and serum NE (Q) in 16- to 18-week-old male control (2 PSKO) and PSKO (+ PSKO) mice injected with Ad-scrambled(2 Ad-shFOXO3) or Ad-shFOXO3 (+ Ad-shFOXO3). Data are expressed as the mean 6 SEM (n = 6–11 mice/group). *P , 0.05 for the effect ofDex treatment vs. no Dex treatment (A, C, and F). *P , 0.05 for the effect of any group vs. control mice treated with no Dex; #P , 0.05 for theeffect of PSOE mice vs. control mice after Dex treatment (E and H). *P , 0.05 for the effect of any group vs. control mice not injected withAd-shFOXO3; #P , 0.05 for the effect of Ad-shFOXO3 injection vs. no Ad-shFOXO3 injection in PSKO mice (I–Q).

diabetes.diabetesjournals.org Deng and Associates 577

Extensive evidence indicates that melanocortin signalingin the hypothalamus plays an important role in regulatingenergy homeostasis and lipid metabolism through its effectson SNS activity in BAT (11–14). In this study we dem-onstrated a possible role for a-MSH in mediating Dex reg-ulation of adiposity, as a-MSH levels were decreased inDex-treated mice via SGK1, and restoration of hypotha-lamic a-MSH levels by ICV administration of this peptidenormalized inadequate energy homeostasis in PSKO mice.Although the beneficial effects of this pharmacologicaltreatment are most likely mediated through direct actionson POMC neurons, we cannot exclude its potential effectson other hypothalamic areas as a result of the delivery routeused.

Our results suggest that the reduced a-MSH content inDex-treated mice was not caused by an altered proteolysisprocess, but the decreased Pomc expression is possibly dueto glucocorticoid resistance, as Dex was shown to inducePomc expression (43). Furthermore, we found that Dex de-creased Pomc expression via the SGK1/FOXO3-dependentpathway, as the inhibitory effect of Dex on Pomc expressionwas blocked in mice with SGK1 overexpression or FOXO3inhibition. Many studies, including those conducted onFOXO3 knockout mice, have demonstrated that FOXO3 isvital for many functions of the CNS and has roles in neu-ral stem cell homeostasis, stress, and Huntington disease(44,45). We showed that it functions as a downstream sig-nal of SGK1 in the regulation of energy homeostasis. We

also demonstrated the spatial relationships among GR,SGK1, and FOXO3, providing a basis for the interactionamong and regulation of these proteins.

In this study we also demonstrated that adult-onset lossof SGK1 in POMC neurons results in a phenotype similar tothat of ablation during development. This is a key issue,because some works report that multiple hypothalamicneurons express POMC in adult mice (24) and prenataland postnatal ablation of certain neurons result in dispa-rate feeding behavior, suggesting that phenotypes causedby prenatal ablation may be influenced by developmentalcompensation (24). The POMC promoter also drives CRErecombinase expression in corticotrophs and melanotrophs(46). The contribution of the pituitary might not be thatsignificant in this study, as no changes were observed inserum corticosterone and growth hormone, which reflectthe function of the pituitary (34), between PSKO and con-trol mice.

Previous studies have shown that POMC neurons areinvolved in the regulation of food intake (29,47). For rea-sons unknown, however, we found that food intake was notsignificantly affected by Dex treatment, or in PSKO or PSOEmice. Consistent with our study, however, previous worksindicate that genetic blockade of the CNS–melanocortin3 receptor promotes fat accumulation in the absence ofhyperphagia (48).

In contrast to the stimulatory effect of glucocorticoidon SGK1 expression (20–22), we observed decreased

Figure 6—Continued.

578 POMC SGK1/FOXO3 Signaling Regulates Adiposity Diabetes Volume 67, April 2018

hypothalamic SGK1 expression following chronic Dex treat-ment, which is, to our knowledge, a novel observation. Wespeculate that this inhibition is not a direct effect of Dexon SGK1 expression but rather a consequence of attenuatedDex-mediated signaling, as it has been previously shownthat prolonged Dex treatment causes glucocorticoid resis-tance (49). Because glucocorticoid normally functions viathe GR (28), the difference in SGK1 expression under acuteor chronic Dex treatment may be caused by differences in GRactivity under different conditions, as shown by our work andthat of others (28,39). In addition, because chronic Dextreatment affects the activity of several regulatory mole-cules that influence SGK1 transcription and/or mRNA de-cay (28,50), and because hypothalamic signals might also beaffected by peripheral events (6,16,26), the possible contri-bution of these factors to hypothalamic SGK1 expressioncannot be excluded in Dex-treated mice. These possibilitieswill be explored in future studies.

In summary, our results demonstrate that SGK1/FOXO3signaling in POMC neurons is crucial for Dex-inducedadiposity. These results provide novel insights into thecentral mechanisms underlying Dex-induced obesity. In thisstudy we also established that SGK1 in POMC neurons is anessential regulator of systemic energy balance. This previ-ously unrecognized role for hypothalamic SGK1 also indicatesa potential novel drug target in treating obesity and itsrelated metabolic disorders.

Funding. This work was supported by grants from the National Natural ScienceFoundation of China (81325005, 81390350, 81471076, 81570777, 81130076,31271269, 81400792, 81500622, and 81600623), a Basic Research Project of theShanghai Science and Technology Commission (16JC1404900 and 17XD1404200),and the Chinese Academy of Sciences/State Administration of Foreign Experts AffairsInternational Partnership Program for Creative Research Teams. F.G. was supportedby the One Hundred Talents Program of the Chinese Academy of Sciences.Duality of Interest. No potential conflicts of interest relevant to this articlewere reported.Author Contributions. Y.D. researched data and wrote, reviewed, andedited the manuscript. Y.X., F.Y., Y.L., X.J., and J.D. researched data. G.F.-T.and A.N.-F.-T. generated and provided the floxed SGK1 mice. S.C. providedresearch material. Y.C., H.Y., and Q.Z. directed the project and contributed tothe discussion. Y.S. and F.G. directed the project; contributed to the discussion;and wrote, reviewed, and edited the manuscript. Y.S. and F.G. are the guarantorsof this work and, as such, had full access to all the data in the study and takeresponsibility for the integrity of the data and the accuracy of the data analysis.

References1. Ma X, Xu L, Mueller E. Forkhead box A3 mediates glucocorticoid receptorfunction in adipose tissue. Proc Natl Acad Sci U S A 2016;113:3377–33822. Cha JY, Kim HJ, Yu JH, et al. Dexras1 mediates glucocorticoid-associatedadipogenesis and diet-induced obesity. Proc Natl Acad Sci U S A 2013;110:20575–205803. Chapman AB, Knight DM, Ringold GM. Glucocorticoid regulation of adipocytedifferentiation: hormonal triggering of the developmental program and induction ofa differentiation-dependent gene. J Cell Biol 1985;101:1227–12354. Kong X, Yu J, Bi J, et al. Glucocorticoids transcriptionally regulate miR-27bexpression promoting body fat accumulation via suppressing the browning of whiteadipose tissue. Diabetes 2015;64:393–404

5. Soumano K, Desbiens S, Rabelo R, Bakopanos E, Camirand A, Silva JE. Glu-cocorticoids inhibit the transcriptional response of the uncoupling protein-1 gene toadrenergic stimulation in a brown adipose cell line. Mol Cell Endocrinol 2000;165:7–156. Schneeberger M, Dietrich MO, Sebastián D, et al. Mitofusin 2 in POMC neuronsconnects ER stress with leptin resistance and energy imbalance. Cell 2013;155:172–1877. Schwartz MW, Porte D Jr. Diabetes, obesity, and the brain. Science 2005;307:375–3798. Wardlaw SL. Hypothalamic proopiomelanocortin processing and the reg-ulation of energy balance. Eur J Pharmacol 2011;660:213–2199. Huszar D, Lynch CA, Fairchild-Huntress V, et al. Targeted disruption of themelanocortin-4 receptor results in obesity in mice. Cell 1997;88:131–14110. Farooqi IS, Keogh JM, Yeo GS, Lank EJ, Cheetham T, O’Rahilly S. Clinicalspectrum of obesity and mutations in the melanocortin 4 receptor gene. N EnglJ Med 2003;348:1085–109511. Lowell BB, Spiegelman BM. Towards a molecular understanding of adaptivethermogenesis. Nature 2000;404:652–66012. Holm C. Molecular mechanisms regulating hormone-sensitive lipase andlipolysis. Biochem Soc Trans 2003;31:1120–112413. Hücking K, Hamilton-Wessler M, Ellmerer M, Bergman RN. Burst-like controlof lipolysis by the sympathetic nervous system in vivo. J Clin Invest 2003;111:257–26414. Morrison SF, Madden CJ, Tupone D. Central neural regulation of brown adiposetissue thermogenesis and energy expenditure. Cell Metab 2014;19:741–75615. Veyrat-Durebex C, Deblon N, Caillon A, et al. Central glucocorticoid adminis-tration promotes weight gain and increased 11b-hydroxysteroid dehydrogenasetype 1 expression in white adipose tissue. PLoS One 2012;7:e3400216. Zakrzewska KE, Cusin I, Stricker-Krongrad A, et al. Induction of obesity andhyperleptinemia by central glucocorticoid infusion in the rat. Diabetes 1999;48:365–37017. Lang F, Artunc F, Vallon V. The physiological impact of the serum andglucocorticoid-inducible kinase SGK1. Curr Opin Nephrol Hypertens 2009;18:439–44818. Fejes-Tóth G, Frindt G, Náray-Fejes-Tóth A, Palmer LG. Epithelial Na+ channelactivation and processing in mice lacking SGK1. Am J Physiol Renal Physiol 2008;294:F1298–F130519. Liu H, Yu J, Xia T, et al. Hepatic serum- and glucocorticoid-regulated proteinkinase 1 (SGK1) regulates insulin sensitivity in mice via extracellular-signal-regulatedkinase 1/2 (ERK1/2). Biochem J 2014;464:281–28920. Anacker C, Cattaneo A, Musaelyan K, et al. Role for the kinase SGK1 in stress,depression, and glucocorticoid effects on hippocampal neurogenesis. Proc Natl AcadSci U S A 2013;110:8708–871321. Ullrich S, Berchtold S, Ranta F, et al. Serum- and glucocorticoid-induciblekinase 1 (SGK1) mediates glucocorticoid-induced inhibition of insulin secretion.Diabetes 2005;54:1090–109922. van Gemert NG, Meijer OC, Morsink MC, Joëls M. Effect of brief corticosteroneadministration on SGK1 and RGS4 mRNA expression in rat hippocampus. Stress2006;9:165–17023. Reiter MH, Vila G, Knosp E, et al. Opposite effects of serum- and glucocorticoid-regulated kinase-1 and glucocorticoids on POMC transcription and ACTH release. AmJ Physiol Endocrinol Metab 2011;301:E336–E34124. Berglund ED, Liu C, Sohn JW, et al. Serotonin 2C receptors in pro-opiomela-nocortin neurons regulate energy and glucose homeostasis [published correctionappears in J Clin Invest 2014;124:1868]. J Clin Invest 2013;123:5061–507025. Hoekstra M, Meurs I, Koenders M, et al. Absence of HDL cholesteryl esteruptake in mice via SR-BI impairs an adequate adrenal glucocorticoid-mediatedstress response to fasting. J Lipid Res 2008;49:738–74526. Kleinridders A, Schenten D, Könner AC, et al. MyD88 signaling in the CNS isrequired for development of fatty acid-induced leptin resistance and diet-induced obesity. Cell Metab 2009;10:249–25927. Xia T, Cheng Y, Zhang Q, et al. S6K1 in the central nervous system regulatesenergy expenditure via MC4R/CRH pathways in response to deprivation of anessential amino acid. Diabetes 2012;61:2461–2471

diabetes.diabetesjournals.org Deng and Associates 579

28. Arango-Lievano M, Lambert WM, Bath KG, Garabedian MJ, Chao MV, Jean-neteau F. Neurotrophic-priming of glucocorticoid receptor signaling is essential forneuronal plasticity to stress and antidepressant treatment. Proc Natl Acad Sci U S A2015;112:15737–1574229. Koch M, Varela L, Kim JG, et al. Hypothalamic POMC neurons promote can-nabinoid-induced feeding. Nature 2015;519:45–5030. Poggioli R, Ueta CB, Drigo RA, Castillo M, Fonseca TL, Bianco AC. Dexa-methasone reduces energy expenditure and increases susceptibility to diet-inducedobesity in mice. Obesity (Silver Spring) 2013;21:E415–E42031. Morton GJ, Cummings DE, Baskin DG, Barsh GS, Schwartz MW.Central nervous system control of food intake and body weight. Nature 2006;443:289–29532. Zhang G, Li J, Purkayastha S, et al. Hypothalamic programming of systemicageing involving IKK-b, NF-kB and GnRH. Nature 2013;497:211–21633. Plum L, Lin HV, Dutia R, et al. The obesity susceptibility gene Cpe links FoxO1signaling in hypothalamic pro-opiomelanocortin neurons with regulation of food in-take. Nat Med 2009;15:1195–120134. Wilson CA, Buckingham JC, Morris ID. Influence of growth hormone,corticosterone, corticotrophin and changes in the environmental temperature onpituitary-ovarian function in the immature rat. J Endocrinol 1985;104:179–18335. Murray JT, Campbell DG, Morrice N, et al. Exploitation of KESTREL to identifyNDRG family members as physiological substrates for SGK1 and GSK3. Biochem J2004;384:477–48836. Brunet A, Park J, Tran H, Hu LS, Hemmings BA, Greenberg ME. Protein kinaseSGK mediates survival signals by phosphorylating the forkhead transcription factorFKHRL1 (FOXO3a). Mol Cell Biol 2001;21:952–96537. Ma W, Fuentes G, Shi X, Verma C, Radda GK, Han W. FoxO1 negatively reg-ulates leptin-induced POMC transcription through its direct interaction with STAT3.Biochem J 2015;466:291–29838. Mammucari C, Milan G, Romanello V, et al. FoxO3 controls autophagy inskeletal muscle in vivo. Cell Metab 2007;6:458–471

39. Sarabdjitsingh RA, Meijer OC, de Kloet ER. Specificity of glucocorticoid receptorprimary antibodies for analysis of receptor localization patterns in cultured cells andrat hippocampus. Brain Res 2010;1331:1–1140. Smart JL, Tolle V, Low MJ. Glucocorticoids exacerbate obesity and insulinresistance in neuron-specific proopiomelanocortin-deficient mice. J Clin Invest 2006;116:495–50541. Kaushik S, Arias E, Kwon H, et al. Loss of autophagy in hypothalamic POMCneurons impairs lipolysis. EMBO Rep 2012;13:258–26542. Nogueiras R, Wiedmer P, Perez-Tilve D, et al. The central melanocortin systemdirectly controls peripheral lipid metabolism. J Clin Invest 2007;117:3475–348843. Wardlaw SL, McCarthy KC, Conwell IM. Glucocorticoid regulation of hypotha-lamic proopiomelanocortin. Neuroendocrinology 1998;67:51–5744. Scarpa JR, Jiang P, Losic B, et al. Systems genetic analyses highlight a TGFb-FOXO3 dependent striatal astrocyte network conserved across species and associ-ated with stress, sleep, and Huntington’s disease. PLoS Genet 2016;12:e100613745. Renault VM, Rafalski VA, Morgan AA, et al. FoxO3 regulates neural stem cellhomeostasis. Cell Stem Cell 2009;5:527–53946. Stefaneanu L, Kovacs K, Horvath E, Lloyd RV. In situ hybridization study of pro-opiomelanocortin (POMC) gene expression in human pituitary corticotrophs and theiradenomas. Virchows Arch A Pathol Anat Histopathol 1991;419:107–11347. Yang Y, Atasoy D, Su HH, Sternson SM. Hunger states switch a flip-flop memorycircuit via a synaptic AMPK-dependent positive feedback loop. Cell 2011;146:992–100348. Chen AS, Marsh DJ, Trumbauer ME, et al. Inactivation of the mouse mela-nocortin-3 receptor results in increased fat mass and reduced lean body mass. NatGenet 2000;26:97–10249. Sousa e Silva T, Longui CA, Rocha MN, et al. Prolonged physical trainingdecreases mRNA levels of glucocorticoid receptor and inflammatory genes. HormRes Paediatr 2010;74:6–1450. Cho H, Park OH, Park J, et al. Glucocorticoid receptor interacts with PNRC2 ina ligand-dependent manner to recruit UPF1 for rapid mRNA degradation. ProcNatl Acad Sci U S A 2015;112:E1540–E1549

580 POMC SGK1/FOXO3 Signaling Regulates Adiposity Diabetes Volume 67, April 2018