1146 Diabetes Volume 66, May 2017

ATF4/ATG5 Signaling in Hypothalamic Neurons Regulates Fat Mass via Affecting Energy Expenditure

Yuzhong Xiao, Yalan Deng, Feixiang Yuan, Tingting Xia, Hao Liu, Zhigang Li, Zhixue Liu, Hao Ying, Yi Liu, Qiwei Zhai, Shanghai Chen, and Feifan Guo

Diabetes 2017;66:1146–1158 | DOI: 10.2337/db16-1546

Although many biological functions of activating tran- between food intake and energy expenditure (2–4). It has scription factor 4 (ATF4) have been identified, a role of been shown that certain populations of neurons in the arcu- hypothalamic ATF4 in the regulation of energy homeo- ate nucleus (ARC) of the play key roles in the stasis is poorly understood. In this study, we showed regulation of energy homeostasis (5,6). These neurons include that hypothalamic proopiomelanocortin (POMC) neuron– those expressing orexigenic neuropeptides specific ATF4 knockout (PAKO) mice are lean and have and agouti-related , along with neurons expressing higher energy expenditure. Furthermore, PAKO mice were anorexigenic neuropeptides cocaine and amphetamine-related – resistant to high-fat diet induced obesity, glucose intoler- transcript and proopiomelanocortin (POMC) (5,6). ance, and resistance. Moreover, the expression of The activating transcription factor 4 (ATF4) belongs to autophagy protein 5 (ATG5) was increased or decreased the family of basic zipper-containing (bZIP) by ATF4 knockdown or overexpression, respectively, and (7,8), with broad expression in various tissues, including ATF4 inhibited the transcription of ATG5 by binding to the brain (9). Previous studies have shown that ATF4 global

METABOLISM basic zipper-containing protein sites on its promoter. Im- knockout mice are lean (8,10) and resistant to high-fat portantly, mice with double knockout of ATF4 and ATG5 in – – POMC neurons gained more fat mass and reduced energy diet (HFD) or high-carbohydrate diet induced obesity expenditure compared with PAKO mice under a high-fat and hyperglycemia (10,11). Recently, studies using tis- fi fi diet. Finally, the effect of ATF4 deletion in POMC neurons sue-speci c knockout mice have identi ed that ATF4 reg- was possibly mediated via enhanced ATG5-dependent ulates glucose metabolism through its expression in autophagy and a-melanocyte–stimulating pro- and osteoblasts (12,13). In contrast, the role of ATF4 in duction in the hypothalamus. Taken together, these results POMC neurons in metabolic control is largely unknown. identify the beneficial role of hypothalamic ATF4/ATG5 axis Autophagy is a cellular process through which cells in the regulation of energy expenditure, obesity, and engulf and degrade damaged cytoplasmic components obesity-related metabolic disorders, which suggests (14). Recent studies have implicated the involvement of that ATF4/ATG5 axis in the hypothalamus may be a new autophagy in the regulation of obesity and energy expen- potential therapeutic target for treating obesity and diture (15–21). For example, autophagy is reduced in the obesity-related metabolic diseases. hypothalamus of aged mice or mice fed an HFD (17,22). Consistently, mice with POMC neuron knockout of autophagy- related 7 (ATG7), one of the key regulators of autophagy Decreased energy expenditure and/or increased food (23), are obese and have lower energy expenditure (16,17,21), intake contributes to the development of obesity (1,2). The and autophagy protein 5 (ATG5)-transgenic mice have the central nervous system (CNS), especially the hypothalamus, opposite phenotype (19). Previous studies have shown that plays an important role in the regulation of the balance ATF4 can regulate autophagy and the expression of some of

Key Laboratory of Nutrition and Metabolism, Institute for Nutritional Sciences, This article contains Supplementary Data online at http://diabetes Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Uni- .diabetesjournals.org/lookup/suppl/doi:10.2337/db16-1546/-/DC1. versity of Chinese Academy of Sciences, Shanghai, China © 2017 by the American Diabetes Association. Readers may use this article as Corresponding author: Feifan Guo, [email protected]. long as the work is properly cited, the use is educational and not for profit, and the Received 15 December 2016 and accepted 13 February 2017. work is not altered. More information is available at http://www.diabetesjournals .org/content/license. diabetes.diabetesjournals.org Xiao and Associates 1147 the autophagy regulators in vitro under different stimula- mice after overnight fasting or 0.75 units/kg to mice tions (24–27); however, the relationship has not been after 4-h fasting, respectively, at 21 weeks old. HOMA of tested in vivo yet. insulin resistance index and areas under the curve were Given the knowledge mentioned above, the aim of our calculated as described previously (33). current study was to explore the role of ATF4 in POMC Serum , Leptin, , neurons in the regulation of obesity and energy expenditure and Corticosterone Measurements and the possible involvement of autophagy in this regulation. Serum norepinephrine (NE), leptin, growth hormone, and RESEARCH DESIGN AND MATERIALS corticosterone levels were measured by an NE ELISA kit (Novus Biologicals, Littleton, CO), leptin ELISA kit (Merck Animals and Treatment Millipore, Frankfurter, Germany), growth hormone ELISA C57BL/6J wild-type (WT) mice were purchased from kit (Merck Millipore), and corticosterone ELISA kit (Novus Shanghai Laboratory Animals Co. Ltd (Shanghai, China). To Biologicals), respectively, according to the manufacturer’s generate hypothalamic POMC neuron–specificATF4knock- instructions. out mice (PAKO) mice, ATF4-floxed mice (13) were inter- crossed with POMC-Cre mice (28), provided by Joel K. Histological Analysis of Tissues Elmquist and Tiemin Liu from Southwestern Medical Paraformaldehyde-fixed, paraffin-embedded sections of Center (Dallas, TX). The mice lacking ATF4 and ATG5 in the white (WAT) and brown adipose tissue POMC neurons (double knockout [DKO]) were generated (BAT) were stained with hematoxylin and eosin for histology. fl fl by intercrossing ATF4- oxed and ATG5- oxed mice (29) fl – Immuno uorescence Staining with POMC-Cre mice. To visualize POMC protein expressing Immunofluorescence (IF) staining with anti-ATF4 anti- neurons under fluorescence microscope, POMC-Cre, PAKO, body (Santa Cruz Biotechnology, Santa Cruz, CA), anti–a- and DKO mice were intercrossed with tdTomato reporter/Ai9 melanocyte–stimulating hormone (a-MSH) antibody mice (30). All animals were under the C57BL/6J background (Merck Millipore), and anti-p62 antibody (Progen, Heidel- and housed in animal cages with a 12-h dark/light cycle at 25°C, berg, Germany) were performed as described previously with free access to water and normal chow diet (NCD). For HFD (31). Phosphorylated (p-)STAT3 staining was performed studies, 8-week-old male mice were fed an HFD with 60% kcal as described previously (32,34). from fat (Research Diets, New Brunswick, NJ) for 3 months. At Construction of Plasmids the age of 22 weeks, mice were sacrificed by CO2 inhalation. All of the experiments were conducted in accordance with the The coding sequence region of ATF4 was amplified from guidelines of the Institutional Animal Care and Use Committee mouse hepatic cDNA and inserted into the eukaryotic of the Institute for Nutritional Sciences, Shanghai Institutes for expression plasmid pCMV-myc. The ATG5 promoter Biological Sciences, Chinese Academy of Sciences. (22085 to +1) was amplified from mouse genomic DNA and inserted into the pGL3-basic report plasmid. The Metabolic Parameter Measurements ATG5 promoter with the bZIP sites (A: 21962/21958 The mice body composition was measured by a nuclear and B: 21321/21317) deleted was generated by site- magnetic resonance system (Bruker, Rheinstetten, Ger- directed mutagenesis (35). many). Indirect calorimetry was measured in a compre- hensive laboratory animal-monitoring system (Columbus Cell Culture and Treatments Instruments, Columbus, OH), as previously described The primary culture of hypothalamic neurons was per- (31). Rectal temperature of mice was measured at 14:00 formed as described previously (31). Recombinant adeno- by a rectal probe attached to a digital thermometer viruses expressing green fluorescent protein (GFP), ATF4, (Physitemp, Clifton, NJ). The measurement of food intake or dominant-negative (DN) ATF4 were purified and admin- 7 was conducted as reported previously (6). istrated at the dose of 1 3 10 plaque-forming units/well in 12-well plates for 48 h (36). The autophagic flux assays Leptin Sensitivity Assay were performed as reported previously (37). 293T cells Leptin sensitivity assay was performed as reported pre- were cultured in DMEM with 10% FBS. Plasmids were viously (32). Mice were individually housed and intraper- transfected by Lipofectamine 2000 (Invitrogen, Carlsbad, itoneally (i.p.) injected with PBS for 5 days prior to i.p. CA) according to the manufacturer’s instructions. injection of leptin (R&D Systems, Minneapolis, MN) twice a day (at 8:00 and 19:00) for 3 days. RNA Isolation and Relative Quantitative RT-PCR The RNA isolation and RT-PCR were performed as Blood Glucose, Serum Insulin, Glucose Tolerance described previously (31). The sequence of primers used Tests, Insulin Tolerance Tests, and HOMA of Insulin Resistance Index is available upon request. Levels of blood glucose and serum insulin were measured by Western Blot Analysis a glucometer and a Mercodia Ultrasensitive Rat Insulin Western blot was performed as described previously (31) ELISA kit (ALPCO Diagnostic, Salem, NH), respectively. with the following primary antibodies: anti-LC3, anti– Glucose tolerance tests (GTTs) and insulin tolerance tests hormone-sensitive lipase (HSL), and anti–phosphorylated (ITTs) were performed by i.p. injection of 2 g/kg glucose to (p-)HSLSer660 (Cell Signaling Technology, Beverly, MA); 1148 POMC ATF4/ATG5 Signaling Regulates Obesity Diabetes Volume 66, May 2017 anti–uncoupling protein 1 (UCP1), anti-ATF4, and anti– by ATF4 knockout. Serum contents of secreted Tribbles homolog 3 (TRB3; Santa Cruz Biotechnology); anti- from the pituitary, including corticosterone and growth p62(Abcam,Cambridge,U.K.);anti-ATG5(Proteintech, hormone (6,17), were comparable between PAKO and Chicago, IL); and anti–b-actin (Sigma-Aldrich, St. Louis, MO). control mice (Supplementary Fig. 1A and B). Expression of pituitary , including T box transcription factor Luciferase Activity Assay (Tpit), Pomc, corticotrophin-releasing hormone receptor 1 293T cells were cotransfected with the internal control (Crhr1), growth hormone (Gh), pituitary-specificpositive plasmid pRL Renilla (Promega, Madison, WI) or plasmids transcription factor 1 (Pit1), and -stimulating hor- as indicated; the firefly and Renilla luciferase activities mone b-chain (Tshb) (6) were also not changed (Supple- were measured by Dual-Glo Luciferase Assay System mentary Fig. 1C). Furthermore, anatomical evaluation of (Promega). POMC neurons throughout the ARC of PAKO and control Chromatin Immunoprecipitation Assays mice revealed no changes in neuronal population size and Chromatin immunoprecipitation (ChIP) assays were con- distribution (Supplementary Fig. 2). ducted according to the manufacturer’sinstructions The body weight, fat mass component, and abdominal (Millipore, Danvers, MA) with anti-ATF4 antibody or normal fat mass were significantly decreased, whereas the lean rabbit IgG (both from Santa Cruz Biotechnology). Immuno- mass component was not changed in PAKO compared precipitated DNA was used as the template to amplify the with control mice maintained on an NCD (Fig. 1C–F). bZIP sites in the promoter of ATG5 with the primers WAT cell volume was decreased, and the protein levels designed as following. Primers for site A are: forward, 59- of p-HSL, a key enzyme regulating lipolysis (31,34), AAGGAGAGGGAAATCTCACCCAAGGGAAAGG-39 and re- were increased in PAKO mice (Supplementary Fig. 3A–C). verse, 59-CTAAATAAACTCCGTTCTATGCTATGCCTGT-39. The energy homeostasis is maintained by a balance Primers for site B are: forward, 59-CTCCACCTACTCAAAG between food intake and energy expenditure (1,4). Food CAGAAATCTCTACTAT-39 and reverse, 59-TCTTGTCAGAC intake was not changed, but the oxygen consumption and TTCTGTTGAGGAGAAGCTGGG-39. The sequences of nega- energy expenditure were higher, and the respiratory ex- tive control primers to amplify a DNA fragment 600 bp change ratio (RER; VCO2/VO2) was lower in PAKO mice upstream and not containing the bZIP site are: forward, 59- (Fig. 1G–J). Although the physical activity was not changed, GTCAAACTAGAAATTCAGGTCGTCAAG-39 and reverse, 59- body temperature was much higher in PAKO mice (Fig. 1K CAAATAGTGCCTGGCCAGCCTCTTCTG-39. and L). The higher body temperature observed in PAKO micewasmostlikelycausedbyincreasedthermogenesis, Transmission Electron Microscope which is regulated by UCP1 in BAT and serum NE that Transmission electron microscope analysis was conducted activates the sympathetic nervous system (SNS) (31,34). as reported previously (38). Consistently, the cell volume and lipid drops in BAT were Statistical Analysis decreased (Supplementary Fig. 3D and E), whereas BAT All values are presented as means 6 SEM. Differences be- UCP1 expression and serum NE levels were increased in tween groups were analyzed by either Student t test or PAKO mice under an NCD (Fig. 1M–O). The expression of one-way ANOVA followed by the Student-Newman-Keuls browning-related genes peroxisome proliferator-activated re- (SNK) test. For GTTs and ITTs, Student t test or one-way ceptor g coactivator 1-a (Pgc1a), Ucp1,andcell death–inducing ANOVA followed by the SNK test was used to compare the DFF-like effector A (Cidea) (6) were also increased in the difference between or among different groups of mice at subcutaneous WAT of PAKO mice (Supplementary Fig. 4). each time point examined. Differences in which P was ,0.05 were considered statistically significant. PAKO Mice Are Resistant to HFD-Induced Obesity, Glucose Intolerance, and Leptin Resistance RESULTS The expression of ATF4 is increased in ARC of mice fed an PAKO Mice Are Lean and Have Higher Energy HFD (34), suggesting that ATF4 in POMC neurons may Expenditure play a role in HFD-induced obesity. Then, we analyzed To investigate whether ATF4 in POMC neurons regulates the metabolic parameters of PAKO and control mice main- obesity and energy expenditure, PAKO mice were gener- tained on an HFD for 3 months. As observed for PAKO ated. Furthermore, to label the POMC-expressing neurons mice under an NCD, the body weight, fat mass component, in vivo, some of the PAKO mice were intercrossed with and abdominal fat mass were also lower, and the lean mass Ai9 (tdTomato reporter) mice (30). IF staining showed component was unchanged in PAKO mice compared with that ATF4 was colocalized with POMC-expressing neu- control mice under an HFD (Fig. 2A–D). WAT cell volume rons in ARC of control mice, but this colocalization was was lower, and the levels of p-HSL were higher in HFD-fed largely reduced in PAKO mice, as evaluated by counting PAKO mice (Supplementary Fig. 5A–C). Moreover, HFD- the number of ATF4-positive POMC neurons (Fig. 1A and fed PAKO mice had higher oxygen consumption, energy B). Because the POMC promoter also drives Cre recombi- expenditure, and body temperature, with lower RER and nase expression in the pituitary (6), we examined whether no change in food intake and physical activity (Fig. 2E–J). the functions of the pituitary–adrenal axis were affected BAT cell volume was lower (Supplementary Fig. 5D and E), diabetes.diabetesjournals.org Xiao and Associates 1149 whereas BAT UCP1 expression and serum NE levels were Consistently, HOMA of insulin resistance index was sig- higher in HFD-fed PAKO mice (Fig. 2K–M). nificantly lower in HFD-fed PAKO mice (Supplementary Because HFD-induced obesity is normally accompanied Fig. 6E). The glucose tolerance and clearance were also by glucose intolerance and leptin resistance, and hypotha- improved in HFD-fed PAKO mice, as demonstrated by lamic signaling has an effect on body glucose homeostasis GTTs and ITTs, respectively (Supplementary Fig. 6F–I). and leptin sensitivity (4,34,39,40), we therefore explored In addition, serum leptin levels were lower in HFD-fed the possible involvement of POMC ATF4 in this regulation PAKO mice (Fig. 3A). To examine the effect of POMC in PAKO and control mice maintained on an HFD. As ATF4 on leptin sensitivity, we i.p. administered leptin expected, the levels of fed or fasted blood glucose and (32) to PAKO and control mice and compared the effects serum insulin were lower in PAKO compared with control of leptin injection on changes in food intake and body mice maintained on an HFD (Supplementary Fig. 6A–D). weight. The effect of leptin on reducing food intake and

Figure 1—PAKO mice are lean and have higher energy expenditure under an NCD. A: Representative images showing IF staining of ATF4 in POMC neurons of POMC-Cre/Ai9 mice and PAKO/Ai9 mice, POMC neurons (red), ATF4 staining (green), and merge (yellow). Scale bars = 50 mm. B: The number of ATF4-positive POMC neurons in ARC. C: Body weight curve. D: Body fat mass component. E: Lean mass component. F: Abdominal fat mass. G: Daily food intake. H: Oxygen consumption. I: Energy expenditure (EE). J: RER (VCO2/VO2). K: Locomotor activity. L: Average basal rectal temperature. M: Ucp1 mRNA expression in BAT. N: UCP1 proteins in BAT (Western blot, top; quantitative measurement of UCP1 relative to actin, bottom). O: Serum NE levels. All studies were conducted in male control mice and littermate PAKO mice maintained on an NCD. Values are means 6 SEM (n = 8/group) and analyzed by two-tailed Student t test. *P < 0.05. 3V, third ventricle. 1150 POMC ATF4/ATG5 Signaling Regulates Obesity Diabetes Volume 66, May 2017

Figure 2—PAKO mice are resistant to HFD-induced obesity. A: Body weight curve. B: Body fat mass component. C: Lean mass com- ponent. D: Abdominal fat mass. E: Daily food intake. F: Oxygen consumption. G: Energy expenditure (EE). H: RER (VCO2/VO2). I: Locomotor activity. J: Average basal rectal temperature. K: Ucp1 mRNA expression in BAT. L: UCP1 proteins in BAT (Western blot, top; quantitative measurement of UCP1 relative to actin, bottom). M: Serum NE levels. All studies were conducted in male control mice (ATF4loxp/loxp) and littermate PAKO mice maintained on an HFD for 3 months, starting at 8 weeks old. Values are means 6 SEM (n =8–10/group) and analyzed by two-tailed Student t test. *P < 0.05.

body weight was much more significant in PAKO mice changes in the expression of TRB3, a well-known down- (Fig. 3B and C). Consistent with the stronger effect of stream effector of ATF4 (42) (Fig. 4A and B). Autophagy leptin in PAKO mice, leptin injection produced more sig- was increased (as demonstrated by the increased LC3-II nals of p-STAT3, the marker of leptin signaling activation and decreased p62 protein levels) in the hypothalamic neu- (32), in POMC neurons of PAKO mice (Fig. 3D and E). rons infected with adenoviruses expressing DN ATF4 Similar effects of POMC ATF4 on glucose metabolism and (Ad-DN ATF4) (Fig. 4A). The opposite effect was observed leptin sensitivity were also observed in mice maintained when ATF4 was overexpressed by adenoviruses expressing on an NCD (Supplementary Figs. 7 and 8). ATF4 (Ad-ATF4) (Fig. 4B). Similar results were obtained in the ARC of PAKO mice under either diet (Fig. 4C and D). ATG5 Expression Is Directly Regulated by ATF4 In Moreover, p62 staining in POMC neurons of PAKO mice Vivo and In Vitro and transmission electron microscope analysis of the num- It is shown that ATF4 can regulate the activity of autophagy ber of autophagosomes in the ARC of PAKO mice under an and the expression of autophagy regulators in different cell HFD also indicated that the autophagy was increased in the lines under different treatments (24–27), and several auto- hypothalamus of PAKO mice (Supplementary Fig. 9). phagy regulators have been indicated in metabolic control Furthermore, the autophagic flux was increased by (16–20,41), suggesting that POMC ATF4 may regulate en- ATF4 inhibition and decreased by ATF4 overexpression, ergy homeostasis by affecting the expression of some of the respectively, in primary hypothalamic neurons transfected autophagy regulators. To test this possibility, we examined with GFP-LC3, simultaneously infected with Ad-DN ATF4 the protein levels of autophagy markers LC3-II (a positive or Ad-ATF4. Because autophagy is a dynamic process, we autophagy marker) and p62 (a negative autophagy marker) further treated the primary hypothalamic neurons with (37,41) in primary cultured hypothalamic neurons with chloroquine (CQ), a lysosomotropic weak base, which ATF4 overexpression or inhibition. The effects of ATF4 over- blocks the fusion of autophagosome with lysosome (37). expression or inhibition were reflected by the corresponding In the presence of CQ, the amount of autophagosome was diabetes.diabetesjournals.org Xiao and Associates 1151

Figure 3—PAKO mice are resistant to HFD-induced leptin resistance. A: Serum leptin levels. B and C: Food intake change and body weight change in mice i.p. injected with 1.5 mg/kg leptin for 3 days. Arrows indicate leptin administration. D: Representative images showing IF staining of p-STAT3 (green) in the POMC neurons (red) of mice i.p. injected with 3 mg/kg leptin or PBS for 45 min. Scale bars = 50 mm. E: Statistical analysis of p-STAT3–positive POMC neurons for D. All studies were conducted in male control mice and littermate PAKO mice maintained on an HFD for 3 months, starting at 8 weeks old. Values are means 6 SEM (n =8–10/group in A–C; n = 4/group in D and E) and analyzed by two-tailed Student t test. *P < 0.05. 3V, third ventricle.

increased in control cells, but the increase was much more Knockout of ATG5 in POMC Neurons of PAKO Mice significant in Ad-DN ATF4 cells or smaller in Ad-ATF4 cells Reversed the Lean Phenotype in These Mice (Fig. 4E and F). Under an HFD fi In order to identify through which autophagy-related To con rm a role of ATG5 in mediating the effects of ATF4 regulators ATF4 regulates autophagy, we measured the knockout in POMC neurons, we generated POMC neuron mRNA expression of autophagy-related proteins Atg4a, ATF4 and ATG5 DKO mice and maintained them and Atg5, Atg6 (Becn1), and Atg7, all important for the produc- corresponding control mice on an HFD. Knockout of ATG5 fi tion of autophagosome (23), in primary hypothalamic neu- (as con rmed by RT-PCR) largely reversed the effects of rons with ATF4 overexpression or inhibition. The mRNA ATF4 knockout in POMC neurons on body weight, fat levels of Atg4a and Atg5 were increased by ATF4 inhibition, mass component, and abdominal fat mass under an HFD – whereas only Atg5 mRNA levels were inhibited by ATF4 (Fig. 6A C and E). Moreover, the higher oxygen consump- overexpression (Fig. 5A and B). Consistently, Atg5 mRNA tion, energy expenditure, body temperature, serum NE lev- levels were increased, with inconsistent changes in other els, BAT UCP1 expression, and WAT p-HSL protein levels Atgs expression, were observed in the ARC of PAKO mice and the lower RER, WAT, and BAT cell volume were also undereitherdiet(Fig.5C and D). The ATG5 protein was also largely reversed by ATG5 knockout in POMC neurons of similarly regulated by ATF4 in vitro and in vivo (Fig. 5E–H). PAKO mice under an HFD, with no changes in lean mass Because two potential ATF4 binding bZIP sites (A: 21962/ component, food intake, and physical activity (Fig. 6D and – 21958 and B: 21321/21317) were identified on the pro- F N and Supplementary Fig. 10). Furthermore, ATG5 moter of ATG5 by the online tool MatInspector (43), we knockout in POMC neurons also attenuated the improved conducted a luciferase activity assay to investigate whether glucose tolerance and leptin sensitivity in PAKO mice un- ATF4 can regulate ATG5 transcription by direct binding to der an HFD (Fig. 7 and Supplementary Fig. 11). these sites. As predicted, overexpression of ATF4 inhibits The Effect of ATF4 Knockout in POMC Neurons Is the promoter activity of ATG5, and the inhibitory effect Possibly Mediated via Stimulation of ATG5-Dependent was abolished by the deletion of the two bZIP sites on the Autophagy and a-MSH Production promoter of ATG5 (Fig. 5I). Moreover, ChIP assay showed ATG5 is an important regulator for autophagy (23), and that the two bZIP sites in the promoter of ATG5 were autophagy has been indicated in the regulation of energy immunoprecipitated by ATF4 antibodies in the ARC of WT homeostasis (15–17,21,41), suggesting that the effect of mice (Fig. 5J). ATF4 knockout in POMC neurons is possibly mediated via 1152 POMC ATF4/ATG5 Signaling Regulates Obesity Diabetes Volume 66, May 2017

Figure 4—ATF4 regulates autophagy in hypothalamic neurons in vivo and in vitro. A and B: LC3-II, p62, TRB3, and ATF4 protein levels in primary hypothalamic neurons with ATF4 inhibition or overexpression (Western blot, left; quantitative measurement of LC3-II, p62, TRB3, and ATF4 relative to actin, right). C and D: LC3-II and p62 protein levels in ARC of PAKO mice and control mice (ATF4loxp/loxp) under an NCD or HFD (Western blot, left; quantitative measurement of LC3-II and p62 relative to actin, right). E and F: GFP-LC3 dots in primary hypothalamic neurons infected with Ad-DN ATF4 or Ad-ATF4 and treated with CQ (10 mmol/L) or not for 6 h before fixed by 4% paraformaldehyde. Scale bars = 10 mm. Means 6 SEM shown are representative of at least three independent in vitro experiments or two independent in vivo experiments, with the number of mice included in each group in each experiment indicated (n =8inC and D) and analyzed by two-tailed Student t test. *P < 0.05.

ATG5-dependent autophagy. Autophagy was activated (as reversed by ATG5 knockout (Fig. 8A–C). Furthermore, demonstrated by the decreased p62 staining) in POMC autophagy affects POMC protein splicing to produce neurons and non-POMC neurons of PAKO mice under a-MSH (17), a key regulator released to the paraventricular an HFD; however, this increased autophagy was largely nucleus (PVN) of the hypothalamus controlling energy diabetes.diabetesjournals.org Xiao and Associates 1153

Figure 5—ATG5 expression is directly regulated by ATF4 in vivo and in vitro. A and B: Atg4a, Atg5, Becn1,andAtg7 mRNA levels in primary hypothalamic neurons with ATF4 inhibition or overexpression. C and D: Atg4a, Atg5, Becn1,andAtg7 mRNA levels in ARC of PAKO mice and control mice (ATF4loxp/loxp) under an NCD or HFD. E and F: ATG5 protein levels in primary hypothalamic neurons with ATF4 inhibition or overexpression (Western blot, top; quantitative measurement of ATG5 relative to actin. bottom). G and H: ATG5 protein levels in ARC of PAKO mice and control mice (ATF4loxp/loxp) under an NCD or HFD (Western blot, top; quantitative measurement of ATG5 relative to actin, bottom). I: The luciferase (LUC) activity of ATG5 promoter and ATG5 promoter with bZIP sites deleted in 293T cells with ATF4 overexpression or not. J: ChIP assay in ARC of WT mice. Means 6 SEM shown are representative of at least three independent in vitro experiments or two independent in vivo experiments, with the number of mice included in each group in each experiment indicated (n =8inC, D, G,andH; n =3inJ)and analyzed by two-tailed Student t test in A–H:*P < 0.05 or one-way ANOVA followed by the SNK test; in I:*P < 0.05 for the effects of any group vs. corresponding control group; #P < 0.05 for the effect of ATG5 promoter with bZIP sites deleted group vs. ATG5 promoter group.

homeostasis (6,16,17). a-MSH staining, using validated The expression of Pc2 and Cpe was increased in the a-MSH antibody (Supplementary Fig. 12), was increased hypothalamus of PAKO mice, and the increase was re- in the PVN of HFD-fed PAKO mice, and the increased versed by ATG5 knockout, whereas the expression of the a-MSH staining was also reversed by ATG5 knockout other processing enzymes showed no significant changes (Fig. 8D and E). The processing enzymes prohormone con- (Fig. 8F). vertase 1 (Pc1/3), prohormone convertase 2 (Pc2), carboxy- peptidase E (Cpe), a-amidating monooxygenase (Pam), and DISCUSSION prolylcarboxypeptidase (Prcp)areinvolvedinthepro- Many populations of neurons in the hypothalamus play cessing of POMC protein to produce a-MSH, consistently. key roles in metabolic control (1,4). In this study, we 1154 POMC ATF4/ATG5 Signaling Regulates Obesity Diabetes Volume 66, May 2017

Figure 6—Knockout ATG5 in POMC neurons of PAKO mice reverses the lean phenotype in these mice under an HFD. A: Atf4 and Atg5 mRNA levels. B: Body weight curve. C: Body fat mass component. D: Lean mass component. E: Abdominal fat mass. F: Daily food intake. G: Oxygen consumption. H: Energy expenditure (EE). I: RER (VCO2/VO2). J: Locomotor activity. K: Average basal rectal temperature. L: Ucp1 mRNA expression in BAT. M: UCP1 proteins in BAT (Western blot, left; quantitative measurement of UCP1 relative to actin, right). N: Serum NE levels. All studies were conducted in male mice with DKO of ATG5 and ATF4 and corresponding control mice (Dloxp/loxp) and PAKO and corresponding control mice (ATF4loxp/loxp), as indicated, maintained on an HFD for 3 months, starting at 8 weeks old. Values are means 6 SEM (n =8–10/group) and analyzed by one-way ANOVA followed by the SNK test: *P < 0.05 for the effects of PAKO group vs. ATF4loxp/loxp group; #P < 0.05 for the effects of DKO group vs. PAKO group.

demonstrated a novel function for ATF4 expressed in ATF4 deletion in POMC neurons, as there were no sig- POMC neurons in the regulation of fat mass: deletion of nificant defects in the pituitary–adrenal axis observed in ATF4 in POMC neurons results in a lean phenotype in PAKO mice. mice maintained on an NCD and helps to prevent HFD- Because body fat mass is maintained by a balance induced obesity. It suggests that ATF4 in POMC neurons between energy intake and energy expenditure (4,34), the may be a new potential therapeutic target for treating lean phenotype in PAKO mice is likely to be caused by obesity. POMC promoter also drives Cre recombinase increased energy expenditure, as food intake is not expression in the pituitary (6,17), a region also involved changed in PAKO mice. In contrast to the unaltered phys- in the regulation of metabolism (44), suggesting that ical activities in PAKO mice, thermogenesis, another as- signals from pituitary might contribute to the phenotype pect responsible for energy expenditure occurring in BAT in PAKO mice. We speculated that, however, the pheno- via increased UCP expression by activation of SNS (45), typesobservedinPAKOmicewerelargelycausedby was increased in PAKO mice. Consistently, the expression diabetes.diabetesjournals.org Xiao and Associates 1155

Figure 7—Knockout ATG5 in POMC neurons of PAKO mice reverses the improved leptin sensitivity in these mice under an HFD. A: Serum leptin levels. B and C: Food intake change and body weight change in mice i.p. injected with 1.5 mg/kg leptin for 3 days. Arrows indicate leptin administration. All studies were conducted in male mice with DKO of ATG5 and ATF4 and corresponding control mice (Dloxp/loxp) and PAKO and corresponding control mice (ATF4loxp/loxp), as indicated, maintained on an HFD for 3 months, starting at 8 weeks old. Values are means 6 SEM (n =8–10/group) and analyzed by one-way ANOVA followed by the SNK test: *P < 0.05 for the effects of PAKO group vs. ATF4loxp/loxp group; #P < 0.05 for the effects of DKO group vs. PAKO group.

of browning-related genes (6) were also increased in sub- factor that regulates the size of fat mass that is also con- cutaneous WAT of PAKO mice. These results suggest that trolled by the activation of the SNS (31,34). Our results the lean phenotype in PAKO mice is caused by the in- showed that the activity of HSL, the key enzyme regulating creased thermogenesis. In addition, lipolysis is another lipolysis (17,31,34), is increased in WAT of PAKO mice,

Figure 8—The effect of ATF4 knockout in POMC neurons is possibly mediated via stimulated ATG5-dependent autophagy and a-MSH production under an HFD. A: Representative images showing IF staining of p62 in POMC neurons of POMC-Cre/Ai9 mice, PAKO/Ai9 mice, and DKO/Ai9 mice: POMC neurons (red), p62 staining (green), and merge (yellow). Scale bars = 50 mm. B and C: Statistical analysis of p62- positive POMC neurons and non-POMC neurons in ARC. D: Representative images showing IF staining of a-MSH in the PVN of mice as indicated. Scale bars = 100 mm. E: Quantification analysis of a-MSH in the PVN. F: Pomc, Pc1/3, Pc2, Cpe, Pam, and Prcp mRNA levels. All studies were conducted in male mice as indicated, maintained on an HFD for 3 months, starting at 8 weeks old. Values are means 6 SEM (n = 4/group in A–E; n =8–10/group in F) and analyzed by one-way ANOVA followed by the SNK test: *P < 0.05 for the effects of PAKO group vs. ATF4loxp/loxp group; #P < 0.05 for the effects of DKO group vs. PAKO group. 3V, third ventricle; Dloxp/loxp, control mice. 1156 POMC ATF4/ATG5 Signaling Regulates Obesity Diabetes Volume 66, May 2017 suggesting increased lipolysis might also contribute to the POMC neurons has a significant effect on energy homeo- lean phenotype in PAKO mice. The increased energy ex- stasis (15–17,21). In addition, POMC is a prohormone penditure observed in PAKO mice is consistent to that that needs posttranslational cleavage by several conver- obtained in ATF4 global knockout mice (8,10), which fur- tases to generate a-MSH (6,48), the process of which is ther demonstrates an important role for ATF4 expressed in shown to be attenuated by the inhibition of autophagy in POMC neurons in the regulation of energy homeostasis. mice (17). Inhibition of autophagy also decreases the A change in fat mass is always associated with a change POMC-derived a-MSH projections to PVN of the hypo- in glucose metabolism and leptin sensitivity, which thalamus by attenuating POMC neuron axon projections controls blood glucose and energy homeostasis, respec- in development (15,16). Our results showed that PAKO tively (34,46). In this study, the glucose tolerance and mice exhibited stimulated autophagy and production of clearance and leptin sensitivity were also improved in a-MSH in PVN, both of which were largely reversed by PAKO mice under either diet at 21 weeks old, which ATG5 knockout in POMC neurons of these mice under an might be a consequence of the reduced fat mass in these HFD, suggesting that POMC ATF4 may regulate obesity mice, as no differences in glucose tolerance and clearance and energy expenditure via stimulated ATG5-dependent were observed between PAKO and control mice at 8 weeks autophagy and a-MSH levels in the hypothalamus. Though old when there was no difference in body weight (Supple- it was unclear, we speculated that the increased a-MSH mentary Fig. 13). Furthermore, leptin-induced STAT3 levels in PVN of PAKO mice was possibly caused by the phosphorylation was increased not only in ARC but also increased cleavage of POMC precursors and the change in in the ventromedial nucleus, dorsomedial nucleus, and POMC neuron axon projections in development. These pos- lateral hypothalamic area of PAKO mice (Supplementary sibilities, however, require nvestigation in the future. Fig. 14). Another possibility is that POMC ATF4 may have Although the hypothalamus of PAKO mice exhibits a direct effect on glucose metabolism and leptin sensitiv- significant alterations in leptin sensitivity, autophagy ity, as adenovirus-mediated ATF4 knockdown in hypo- activity, and a-MSH levels, factors that are important for thalamus improves insulin sensitivity (36), and many the regulation of food intake (6,17), we did not observe any signaling pathways in POMC neurons regulate leptin sen- changes in daily or fasting-induced food intake in PAKO sitivity (2,4,6,47). These possibilities, however, require mice (Supplementary Fig. 15). Our results are consistent further investigation. with some other studies showing that pharmacological or Several autophagy regulators have been demonstrated genetic blockade of melanocortin receptors in the CNS or to play important roles in energy homeostasis (16–19,21,41). global deletion of ATF4 has no effect on food intake Forexample,knockoutofATG7orATG12inPOMC (49,50), and mice with XBP1 overexpressed in POMC neu- neurons causes obesity and decreases energy expendi- rons remain unaltered with regard to food intake in the ture (16,17,21,41). It has been previously shown that presence of enhanced leptin sensitivity (47). We hypothe- knockdown of ATF4 expression activates autophagy in sized that the additional signals in POMC neurons or sig- human HT1080 fibrosarcoma cells (24), and ATF4 is also nals in other neurons or brain areas may play dominant required for stress-induced autophagy regulators ATG5, roles in the regulation of food intake in PAKO mice. ATG7, and BECN1 expression in human cancer cells Taken together, our results identify a new function for or mouse embryonic fibroblasts (26,27). The regulatory central ATF4/ATG5 axis in the regulation of energy mechanisms underlying ATF4 regulation of ATG5 expres- expenditure, obesity, and obesity-related glucose metab- sion and whether this relationship exists in vivo, however, olism and leptin sensitivity. It also suggests that ATF4 in have not been reported. We showed that the transcription the CNS may be a new potential therapeutic target for of ATG5 was regulated by ATF4 via direct binding to the treating obesity and obesity-related metabolic diseases. bZIP sites in its promoter and also provided the first evidence showing that ATG5 expression is regulated by ATF4 in vivo. Acknowledgments. The authors thank Joel K. Elmquist and Tiemin Liu The important role of ATG5 in ATF4-regulated energy from Southwestern Medical Center for providing POMC-Cre mice. The authors homeostasis on an HFD was demonstrated by the larger also thank Noboru Mizushima from the University of Tokyo for providing ATG5 amount of fat mass and lower energy expenditure observed floxed mice. in DKO mice compared with PAKO mice. In contrast, Funding. This work was supported by grants from National Natural Science Malhotra et al. (41) reported that mice with ATG5 knock- Foundation of China (81325005, 81471076, 81570777, 81390350, 81130076, out in POMC neurons have no obvious phenotypes in body 31271269, 81400792, 81500622, and 81600623), Basic Research Project weight and fat mass compared with control mice under an of Shanghai Municipal Science and Technology Commission (16JC1404900 and 17XD1404200), and International S&T Cooperation Program of China (Singapore HFD. The induced ATG5 expression in POMC neurons of 2014DFG32470), and research was supported by the Chinese Academy of HFD-fed PAKO mice in our study may be responsible for Sciences/State Administration of Foreign Experts Affairs International Partner- the different observations. However, this possibility re- ship Program for creative research teams. F.G. was also supported by the 100 quires further study in the future. Talents Program of the Chinese Academy of Sciences. It is well established that ATG5 is required for the Duality of Interest. No potential conflicts of interest relevant to this article production of autophagosome (23), and autophagy in were reported. diabetes.diabetesjournals.org Xiao and Associates 1157

Author Contributions. Y.X. researched data and wrote, reviewed, and 22. Morselli E, Criollo A, Rodriguez-Navas C, Clegg DJ. Chronic high fat diet edited the manuscript. Y.D., F.Y., T.X., and H.L. researched data. Z. Li, Z. Liu, H.Y., consumption impairs metabolic health of male mice. Inflamm Cell Signal 2014;1: and S.C. provided research materials. Y.L. and Q.Z. contributed to discussion. F.G. e561 directed the project, contributed to discussion, and wrote, reviewed, and edited the 23. Mizushima N, Yoshimori T, Ohsumi Y. The role of Atg proteins in auto- manuscript. F.G. is the guarantor of this work and, as such, had full access to all phagosome formation. Annu Rev Cell Dev Biol 2011;27:107–132 the data in the study and takes responsibility for the integrity of the data and the 24. Ye J, Kumanova M, Hart LS, et al. The GCN2-ATF4 pathway is critical for accuracy of the data analysis. tumour cell survival and proliferation in response to nutrient deprivation. EMBO J 2010;29:2082–2096 References 25. Chen R, Zou Y, Mao D, et al. The general amino acid control pathway 1. Schwartz MW, Porte D Jr. Diabetes, obesity, and the brain. Science 2005; regulates mTOR and autophagy during serum/glutamine starvation. J Cell Biol 307:375–379 2014;206:173–182 2. Leshan RL, Greenwald-Yarnell M, Patterson CM, Gonzalez IE, Myers MG Jr. 26. B’chir W, Maurin AC, Carraro V, et al. The eIF2a/ATF4 pathway is essential Leptin action through hypothalamic nitric oxide synthase-1-expressing neurons for stress-induced autophagy gene expression. Nucleic Acids Res 2013;41: controls energy balance. Nat Med 2012;18:820–823 7683–7699 3. Ryan KK, Woods SC, Seeley RJ. Central nervous system mechanisms 27. Rouschop KM, van den Beucken T, Dubois L, et al. The unfolded protein linking the consumption of palatable high-fat diets to the defense of greater response protects human tumor cells during hypoxia through regulation of the – adiposity. Cell Metab 2012;15:137 149 autophagy genes MAP1LC3B and ATG5. J Clin Invest 2010;120:127–141 4. Myers MG Jr, Olson DP. Central nervous system control of metabolism. 28. Balthasar N, Coppari R, McMinn J, et al. signaling in POMC – Nature 2012;491:357 363 neurons is required for normal body weight homeostasis. Neuron 2004;42:983– 5. Ruan HB, Dietrich MO, Liu ZW, et al. O-GlcNAc transferase enables AgRP 991 – neurons to suppress browning of white fat. Cell 2014;159:306 317 29. Kuma A, Hatano M, Matsui M, et al. The role of autophagy during the early 6. Schneeberger M, Dietrich MO, Sebastián D, et al. Mitofusin 2 in POMC neonatal starvation period. Nature 2004;432:1032–1036 neurons connects ER stress with leptin resistance and energy imbalance. Cell 30. Chiappini F, Catalano KJ, Lee J, et al. Ventromedial hypothalamus-specific – 2013;155:172 187 Ptpn1 deletion exacerbates diet-induced obesity in female mice. J Clin Invest 7. Mielnicki LM, Pruitt SC. Isolation and nucleotide sequence of a murine cDNA 2014;124:3781–3792 homologous to human activating transcription factor 4. Nucleic Acids Res 1991; 31. Xia T, Cheng Y, Zhang Q, et al. S6K1 in the central nervous system reg- 19:6332 ulates energy expenditure via MC4R/CRH pathways in response to deprivation of 8. Wang C, Huang Z, Du Y, Cheng Y, Chen S, Guo F. ATF4 regulates lipid an essential amino acid. Diabetes 2012;61:2461–2471 metabolism and thermogenesis. Cell Res 2010;20:174–184 32. Long L, Toda C, Jeong JK, Horvath TL, Diano S. PPARg ablation sensitizes 9. Mamady H, Storey KB. Coping with the stress: expression of ATF4, ATF6, proopiomelanocortin neurons to leptin during high-fat feeding. J Clin Invest 2014; and downstream targets in organs of hibernating ground squirrels. Arch Biochem 124:4017–4027 Biophys 2008;477:77–85 33. Ueda H, Ikegami H, Kawaguchi Y, et al. Age-dependent changes in phe- 10. Seo J, Fortuno ES 3rd, Suh JM, et al. Atf4 regulates obesity, glucose ho- notypes and candidate gene analysis in a polygenic animal model of Type II meostasis, and energy expenditure. Diabetes 2009;58:2565–2573 diabetes mellitus; NSY mouse. Diabetologia 2000;43:932–938 11. Li H, Meng Q, Xiao F, et al. ATF4 deficiency protects mice from high- 34. Xiao Y, Xia T, Yu J, et al. Knockout of inositol-requiring enzyme 1a in pro- carbohydrate-diet-induced liver steatosis. Biochem J 2011;438:283–289 opiomelanocortin neurons decreases fat mass via increasing energy expenditure. 12. Yoshizawa T, Hinoi E, Jung DY, et al. The transcription factor ATF4 regulates Open Biol 2016;6:160131 glucose metabolism in mice through its expression in osteoblasts. J Clin Invest 35. Wu Y, Xiao Y, Ding X, et al. A miR-200b/200c/429-binding site poly- 2009;119:2807–2817 morphism in the 39 untranslated region of the AP-2 gene is associated with 13. Li K, Zhang J, Yu J, et al. MicroRNA-214 suppresses gluconeogenesis by a targeting activating transcriptional factor 4. J Biol Chem 2015;290:8185–8195 cisplatin resistance. PLoS One 2011;6:e29043 14. Mizushima N. Physiological functions of autophagy. Curr Top Microbiol 36. Zhang Q, Yu J, Liu B, et al. Central activating transcription factor 4 (ATF4) Immunol 2009;335:71–84 regulates hepatic insulin resistance in mice via S6K1 signaling and the vagus – 15. Coupé B, Bouret SG. Weighing on autophagy: a novel mechanism for the nerve. Diabetes 2013;62:2230 2239 CNS regulation of energy balance. Cell Cycle 2012;11:1477–1478 37. Xiao Y, Liu H, Yu J, et al. Activation of ERK1/2 ameliorates liver steatosis in fi 16. Coupé B, Ishii Y, Dietrich MO, Komatsu M, Horvath TL, Bouret SG. Loss of leptin receptor-de cient (db/db) mice via stimulating ATG7-dependent autophagy. autophagy in pro-opiomelanocortin neurons perturbs axon growth and causes Diabetes 2016;65:393–405 metabolic dysregulation. Cell Metab 2012;15:247–255 38. Xu DQ, Wang Z, Wang CY, et al. PAQR3 controls autophagy by integrating 17. Kaushik S, Arias E, Kwon H, et al. Loss of autophagy in hypothalamic POMC AMPK signaling to enhance ATG14L-associated PI3K activity. EMBO J 2016;35: neurons impairs lipolysis. EMBO Rep 2012;13:258–265 496–514 18. Kaushik S, Rodriguez-Navarro JA, Arias E, et al. Autophagy in hypothalamic 39. Seeley RJ, Sandoval DA. Targeting the brain as a cure for type 2 diabetes. AgRP neurons regulates food intake and energy balance. Cell Metab 2011;14: Nat Med 2016;22:709–711 173–183 40. Flak JN, Patterson CM, Garfield AS, et al. Leptin-inhibited PBN neurons 19. Pyo JO, Yoo SM, Ahn HH, et al. Overexpression of Atg5 in mice activates enhance responses to hypoglycemia in negative energy balance. Nat Neurosci autophagy and extends lifespan. Nat Commun 2013;4:2300 2014;17:1744–1750 20. Martinez-Lopez N, Garcia-Macia M, Sahu S, et al. Autophagy in the CNS and 41. Malhotra R, Warne JP, Salas E, Xu AW, Debnath J. Loss of Atg12, but not periphery coordinate lipophagy and lipolysis in the brown adipose tissue and Atg5, in pro-opiomelanocortin neurons exacerbates diet-induced obesity. Auto- liver. Cell Metab 2016;23:113–127 phagy 2015;11:145–154 21. Quan W, Kim HK, Moon EY, et al. Role of hypothalamic proopiomelanocortin 42. Ohoka N, Yoshii S, Hattori T, Onozaki K, Hayashi H. TRB3, a novel ER stress- neuron autophagy in the control of appetite and leptin response. Endocrinology inducible gene, is induced via ATF4-CHOP pathway and is involved in cell death. 2012;153:1817–1826 EMBO J 2005;24:1243–1255 1158 POMC ATF4/ATG5 Signaling Regulates Obesity Diabetes Volume 66, May 2017

43. Cartharius K, Frech K, Grote K, et al. MatInspector and beyond: promoter 47. Williams KW, Liu T, Kong X, et al. Xbp1s in Pomc neurons connects ER stress analysis based on transcription factor binding sites. Bioinformatics 2005;21: with energy balance and glucose homeostasis. Cell Metab 2014;20:471–482 2933–2942 48. Wardlaw SL. Hypothalamic proopiomelanocortin processing and the regu- 44. Ristic N, Stevanovic D, Nesic D, et al. Diet-induced obesity and lation of energy balance. Eur J Pharmacol 2011;660:213–219 effects on pituitary gonadotrophs: immunohistomorphometric study in male 49. Nogueiras R, Wiedmer P, Perez-Tilve D, et al. The central melanocortin rats. Cell J 2016;17:711–719 system directly controls peripheral lipid metabolism. J Clin Invest 2007;117: 45. Lowell BB, Spiegelman BM. Towards a molecular understanding of adaptive 3475–3488 thermogenesis. Nature 2000;404:652–660 50. Perino A, Beretta M, Kilic A, et al. Combined inhibition of PI3Kb and PI3Kg 46. Flak JN, Myers MG Jr. Minireview: CNS mechanisms of leptin action. Mol reduces fat mass by enhancing a-MSH-dependent sympathetic drive. Sci Signal Endocrinol 2016;30:3–12 2014;7:ra110