Jagannath Misra,1 Don-Kyu Kim,1 Yoon Seok Jung,1 Han Byeol Kim,2

Yong-Hoon Kim,3 Eun-Kyung Yoo,4 Byung Gyu Kim,5 Sunghoon Kim,6 In-Kyu Lee,4

Robert A. Harris,7 Jeong-Sun Kim,8 Chul-Ho Lee,3 Jin Won Cho,2 andHueng-Sik Choi1

O-GlcNAcylation of Orphan NuclearReceptor Estrogen-Related Receptor gPromotes Hepatic GluconeogenesisDiabetes 2016;65:2835–2848 | DOI: 10.2337/db15-1523

Estrogen-related receptor g (ERRg) is a major positiveregulator of hepatic gluconeogenesis. Its transcriptionalactivity is suppressed by phosphorylation signaled byinsulin in the fed state, but whether posttranslationalmodification alters its gluconeogenic activity in the fastedstate is not known. Metabolically active hepatocytes directa small amount of glucose into the hexosamine biosyn-thetic pathway, leading to protein O-GlcNAcylation. In thisstudy, we demonstrate that ERRg is O-GlcNAcylated byO-GlcNAc transferase in the fasted state. This stabilizesthe protein by inhibiting proteasome-mediated proteindegradation, increasing ERRg recruitment to gluconeo-genic gene promoters. Mass spectrometry identifies twoserine residues (S317, S319) present in the ERRg ligand-binding domain that are O-GlcNAcylated. Mutation ofthese residues destabilizes ERRg protein and blocksthe ability of ERRg to induce gluconeogenesis in vivo.The impact of this pathway on gluconeogenesis in vivowas confirmed by the observation that decreasing theamount of O-GlcNAcylated ERRg by overexpressing thedeglycosylating enzyme O-GlcNAcase decreases ERRg-dependent glucose production in fasted mice. We con-clude that O-GlcNAcylation of ERRg serves as a majorsignal to promote hepatic gluconeogenesis.

O-GlcNAcylation works as a nutrient sensor in the liverto maintain energy homeostasis in response to varying

nutrient flux (1). O-GlcNAcylation is extensively linkedwith glucose metabolism in liver. L-Glutamine fructose-6-phosphate amidotransferase (GFAT) overexpression leadsto peripheral insulin resistance (2,3). Transgenic mice over-expressing O-GlcNAc transferase (OGT) in skeletal muscleand fat exhibit elevated circulating insulin levels and insulinresistance (4). Insulin receptor substrate 1/2 of insulin sig-naling is O-GlcNAcylated (5), and O-GlcNAcylation hasbeen shown to be a negative regulator of insulin signaling(6). O-GlcNAcylation of forkhead box class O1 (FOXO1),CRTC2, and peroxisome proliferator–activated receptor g co-activator 1a (PGC-1a) modulates expression of gluconeogenicgenes (7–10). Chronic increase in O-GlcNAcylation levels ofPDX1 and NeuroD1 may contribute to hyperinsulinemia intype 2 diabetes (11,12). Thus, by being intimately intertwinedwith metabolism, the hexosamine biosynthetic pathway (HBP)and its end product O-GlcNAc link transcriptional processes tocellular glucose metabolism and insulin resistance.

Estrogen-related receptors (ERRs) are members of theNR3B subfamily of nuclear receptors, which include ERRa,ERRb, and ERRg. ERRg is primarily expressed in heart,brain, kidney, pancreas, and liver tissues and is inducedduring fasting in murine liver (13–15). ERRg plays an im-portant role in the regulation of glucose, lipid, alcohol andiron metabolism in mouse liver (16,17). Hepatic ERRg

1National Creative Research Initiatives Center for Nuclear Receptor Signals andHormone Research Center, School of Biological Sciences and Technology, ChonnamNational University, Gwangju, Republic of Korea2Department of Integrated OMICS for Biomedical Science, Yonsei University,Seoul, Republic of Korea3Korea Research Institute of Bioscience and Biotechnology, Daejeon, Republic of Korea4Department of Internal Medicine, Kyungpook National University School of Med-icine, Deagu, Republic of Korea5Leading-edge Research Center for Drug Discovery and Development and Met-abolic Disease, Kyungpook National University, Daegu, Korea6Medicinal Bioconvergence Research Center Department of Molecular Medicineand Biopharmaceutical Sciences Graduate School of Convergence Science andTechnology College of Pharmacy, Seoul National University, Seoul, Korea

7Department of Biochemistry and Molecular Biology, Indiana University School ofMedicine and the Roudebush VA Medical Center, Indianapolis, IN8Department of Chemistry and Institute of Basic Sciences, Chonnam NationalUniversity, Gwangju, Republic of Korea

Corresponding author: Hueng-Sik Choi, hsc@chonnam.ac.kr.

Received 4 November 2015 and accepted 15 June 2016.

This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db15-1523/-/DC1.

© 2016 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 65, October 2016 2835

METABOLISM

expression is induced in fasting and the diabetic state andcauses insulin resistance and glucose intolerance (18). In-duction of hepatic ERRg impairs insulin signaling throughdiacylglycerol-mediated protein kinase e activation (19),suggesting that ERRg transcriptional activity could be in-volved in insulin action to maintain glucose homeostasis.Recently, our laboratory reported that insulin-dependentphosphorylation of ERRg alters its transcriptional activityto suppress hepatic gluconeogenesis in the fed state (20).

PGC-1a is a transcriptional coactivator involved in he-patic glucose metabolism. Fasting induces hepatic PGC-1aexpression that directly interacts with transcription fac-tors, including hepatocyte nuclear factor 4a, FOXO1, andglucocorticoid receptor, to increase the expression of glu-coneogenic genes (21,22). PGC-1a overexpression leads toincreased expression of G6Pase and PEPCK, key enzymesin the hepatic gluconeogenesis. Conversely, knockdown orknockout of PGC-1a results in lower blood glucose levelsas a result of reduced gluconeogenesis.

As insulin-dependent posttranslational modification reg-ulates the transcriptional activity of ERRg in the fed state,in this study, we investigated whether fasting-dependentactivation of the transcriptional activity of ERRg involvesO-GlcNAcylation. We demonstrate that the fasting condi-tion triggers O-GlcNAcylation of ERRg that results inprotein stabilization. O-GlcNAcylation of ERRg by OGTdecreases its ubiquitination and cooperatively upregulatesgluconeogenesis. In contrast, the fed condition decreasesO-GlcNAcylation of ERRg, resulting in ubiquitin-mediatedprotein degradation. Overall, our study describes howO-GlcNAcylation modulates gluconeogenesis via ERRg.

RESEARCH DESIGN AND METHODS

Animal ExperimentsMale 8-week-old C57BL/6J mice, maintained at the KoreaResearch Institute of Bioscience and Biotechnology (KRIBB)were fed either a high-fat diet (HFD) (D12492; ResearchDiets, New Brunswick, NJ) or a normal chow diet for12 weeks. At the end of 12 weeks, mice were sacrificed, andliver tissue was used for identification of O-GlcNAcylationof ERRg. ob/ob and db/db mice (7–12 weeks old; CharlesRiver Laboratories) were maintained at KRIBB in an animalfacility with ad libitum access to water and a standardlaboratory diet. Liver tissue was used for identificationof O-GlcNAcylation of ERRg. Male 7-week-old C57BL/6Jmice (The Jackson Laboratory, Bar Harbor, ME) wereobtained from Ochang Branch Institute, KRIBB. After2 weeks, adenoviruses (Ad-green fluorescent protein [GFP],Ad–wild-type [wt] ERRg, and Ad-S317A+S319A ERRg;5.9 3 109 plaque-forming units/mouse) were deliveredby tail-vein injection into mice. Glucose tolerance testwas performed at day 5 after a tail-vein injection ofadenoviruses. Briefly, mice fasted 16 h were injected in-traperitoneally with 1 g/kg glucose, and blood glucose wasmeasured in tail-vein blood using a blood glucose meterand test trips (Accu-Chek Aviva meter system; RocheDiagnostics, Indianapolis, IN). All mice were acclimatized

to a 12-h light/dark cycle at 22 6 2°C with free access tofood and water in a specific pathogen-free facility. All ani-mal experiments were approved and performed by the In-stitutional Animal Care and Use Committee of KRIBB.

Glucose Output AssayGlucose production from primary mouse hepatocytes wasmeasured using a colorimetric glucose oxidase assay kitaccording to the manufacturer’s protocol. Briefly, afterthe experimental time period as indicated, the cells werewashed three times with PBS. Then the cells were incu-bated for 3 h at 37°C, 5% CO2, in glucose productionbuffer (glucose-free DMEM [pH 7.4], containing 20mmol/Lsodium lactate, 1 mmol/L sodium pyruvate, and 15 mmol/LHEPES, without phenol red), and the glucose assays wereperformed.

Glucagon AssayBlood glucagon levels were measured using the mouse glu-cagon EIA kit (Ray Biotech) following the manufacturer’sprotocol.

Cell Culture and ReagentsPrimary hepatocytes were isolated from C57BL/6J mice(male, 20–30 g) by collagenase perfusion (23) and seededwith Medium 199 (Cellgro). After 3–6 h of attachment,cells were infected with the indicated adenoviruses foroverexpression or treated with various chemicals as in-dicated. HEK 293T and AML12 cells were maintained asdescribed previously (24). Transient transfection was per-formed using Lipofectamine 2000 (Invitrogen) or SuperFect(Qiagen, Hilden, Germany) according to the manufacturers’instructions. b-N-Acetyl-D-glucosamine (GlcN), 6-diazo-5-oxo-L-norleucine (DON), glucagon, insulin, cycloheximide,ANTI-FLAG M2 affinity gel, glucose oxidase assay kit, andstreptozotocin (STZ) were purchased from Sigma-Aldrich.MG-132 protease inhibitors were purchased from Calbiochem.Express Protein Labeling Mix [35S] (NEG072002MC) waspurchased from PerkinElmer. The Mouse Glucagon EIAkit (EIAM-GLU) was purchased from Ray Biotech. Anti-bodies were purchased as follows: O-GlcNAc fromCovance, a-tubulin from Abfrontier, ERRg from PerseusProteomix, OGT from Abcam, FLAG, Anti-FLAG M2, andhemagglutinin (HA) from Cell Signaling Technology, andG6Pase, Mdm2, ubiquitin, Gal4, PGC-1a, and PEPCKfrom Santa Cruz Biotechnology.

Plasmid and Adenovirus Vector ConstructsExpression vectors for HA-ERRg, FLAG-ERRg, HA-PGC-1a, and Sft4-luc containing three copies of the ERRgbinding site were described previously (25). HA-ERRawas described previously (26). FLAG-human OGT wasconstructed by inserting the full PCR fragment of theopen reading frame into the Not1/Sal1 sites of thep3XFLAG-CMV-7.1 vector. FLAG-human O-GlcNAcase(OGA) was constructed by inserting the full PCR frag-ment of the open reading frame into the Bgl2/Sal1sites of the pFLAG-CMV-7.1 vector. FLAG-mutant ERRg’s(S317A, S319A, and S317A+S319A) were constructed using

2836 O-GlcNAcylation of ERRg Diabetes Volume 65, October 2016

wild-type ERRg as a template by the Quick ChangeLightning Site-Directed Mutagenesis kit from AgilentTechnologies. Gal4–DNA binding domain (DBD) andGal4-tk-Luc were described previously (27). Briefly, Gal4–ERRg–ligand-binding domain (LBD) is a fusion proteinconsisting of the Gal4–DBD (amino acids 1–147) andERRg–LBD (amino acids 189–458). This fusion proteinactivates transcription of a reporter construct (Gal4–tk-luc) containing five GAL4 binding sites (upstream activatorsequence) upstream of the firefly luciferase genein the pGL2-promoter. Gal4-DBD-S317A ERRg, Gal4-DBD-S319A ERRg, and Gal4-DBD-S317A+S319A ERRgwere constructed using Gal4-DBD-wild-type ERRg as atemplate by the Quick Change Lightning Site-DirectedMutagenesis kit from Agilent Technologies. wtPEPCK-luc,and ERRg response element (ERRE) mutant PEPCK-lucwere described previously (28). Adenoviruses expressingunspecific short hairpin (sh)RNA, shERRg, control GFP,and ERRg were described previously (28). Ad-OGT encod-ing the human OGT gene, adenovirus OGA (Ad-OGA) encod-ing the human OGA gene, and adenovirus S317A+S319AERRg (Ad-S317A+S319A ERRg) were generated with thepAd-easy system as described previously (29). All viruseswere purified by using CsCl gradient protocol.

Hepatic FLAG-ERRg Complex Purification andMapping of O-GlcNAc Site Using Mass SpectrometryOverexpressed wild‐type FLAG-ERRg proteins from mouseliver were purified using FLAG‐M2 agarose and subjectedto SDS-PAGE. Purified protein was digested with trypsin(Promega, Madison, WI) (25 ng/mL) for 16 h at 37°C.After in-gel digestion, tryptic peptides were separated byonline reversed-phase chromatography using a ThermoScientific Eazy nano LC II autosampler (Thermo Scientific)with a reversed-phase peptide trap EASY-Column (100-mminner diameter, 2-cm length) and a reversed-phase ana-lytical EASY-Column (75-mm inner diameter, 10-cmlength, 3-mm particle size; both from Thermo Scientific).Electrospray ionization was performed using a 30-mm (in-side diameter) nano-bore stainless steel online emitter(Thermo Scientific) and a voltage set at 2.6 V, at a flowrate of 300 nL/min. The chromatography system wascoupled on-line with an LTQ Velos Orbitrap mass spec-trometer (Thermo Scientific). Protein identification wasaccomplished using the Proteome Discoverer v1.3 data-base search engine (Thermo Scientific), and searcheswere performed against IPI.Humanv3.87 FASTA data-base or ERRg FASTA database. A fragment mass toler-ance of 1.2 Da, peptide mass tolerance of 25 ppm, andmaximum missed cleavage of 2 were set.

Real-Time Quantitative RT-PCR AnalysisTotal RNA was isolated using TRIzol reagent (Invitrogen),cDNA was synthesized using a reverse transcriptase kit(Intron Biotechnology), and real-time quantitative RT-PCR(qPCR) was performed with the SYBR green PCR kit(Enzynomics). The amount of mRNA for each gene wasnormalized to that of actin mRNA.

Chromatin Immunoprecipitation AssayNuclear isolation of primary hepatocytes and cross-link-ing of protein to DNA were performed as describedpreviously (28). After sonication, soluble chromatin wassubjected to immunoprecipitation using anti-ERRg anti-body. DNA was recovered by phenol/chloroform extrac-tion and analyzed by PCR using primers against relevantpromoters. Primer sequences were mouse PEPCK1 pro-moter forward, 59-CTAGCCAGCTTTGCCTGACT-39 and re-verse, 59-GGGTCCCCACGACCTTCCAA-39.

Western Blot AnalysisWhole-cell extracts were prepared using radioimmuno-precipitation assay (RIPA) buffer (50 mmol/L Tris [pH7.5], 150 mmol/L NaCl, 1% Nonidet P-40, and 5 mmolEDTA). Proteins from whole-cell lysates were separated by10% SDS-PAGE and then transferred to nitrocellulosemembranes. The membranes were probed with differentantibodies. Immunoreactive proteins were visualized usingan Amersham Biosciences ECL kit (GE Healthcare) accord-ing to the manufacturer’s instructions.

In Vivo ImagingC57BL/6J mice were infected with respective viruses viatail-vein injections. Four days postinjection, mice were fastedfor 16 h and imaged using an IVIS Lumina II imagingsystem (Caliper Life Sciences, Hopkinton, MA) as describedpreviously (28).

Confocal MicroscopyAt 24 h after transfection, the cells were fixed with 2%formaldehyde, immunostained, and subjected to observa-tion by confocal microscopy using a laser-scanning confocalmicroscope (Olympus, Lake Success, NY).

Pulse-Chase ExperimentAML12 cells were transfected with FLAG-wt ERRg, FLAG-S317A ERRg, FLAG-S319A ERRg, and FLAG-S317A+S319AERRg and incubated in methionine and cystine-free me-dium for 2 h. Translabel mixture (PerkinElmer) contain-ing 35S-methionine was added for 30 min, and then cellswere cultured in normal medium up to 3 h. FLAG-ERRg was immunoprecipitated with M2 antibody inRIPA buffer, and radioactive FLAG-ERRg was detected byautoradiography.

Statistical AnalysesAll values are expressed as means6 SEM. The significancebetween mean values was evaluated by two-tailed Studentt test.

RESULTS

ERRg Is Modified by O-GlcNAcERRg is a key positive regulator of hepatic gluconeogen-esis (28). ERRg phosphorylation by protein kinase B/Aktalso contributes to insulin-mediated inhibition of hepaticgluconeogenesis (20). Many factors that regulate hepaticgluconeogenesis are O-GlcNAcylated in various condi-tions (7,30,31). Hence, we sought to determine whether

diabetes.diabetesjournals.org Misra and Associates 2837

ERRg is modified by O-GlcNAc. HBP intermediate GlcNtreatment significantly increased ERRg O-GlcNAcylationlevels as well as ERRg protein content in a dose-dependentmanner (Fig. 1A). Consistent with a rise in ERRg pro-tein levels, GlcN treatment significantly reduced ERRgubiquitination levels and increased ERRg transcrip-tional activity (Supplementary Fig. 1A and B). Glucose(Glc) flux through HBP regulates O-GlcNAcylation (1).Therefore, to determine whether glucose could directlyaffect O-GlcNAcylation of ERRg, mouse primary hepa-tocytes (MPH) were treated with 5 and 25 mmol/L glu-cose. High glucose significantly increased O-GlcNAcylationof ERRg along with ERRg protein levels (Fig. 1B). Thiswas further confirmed when OGT cotransfection mark-edly enhanced O-GlcNAcylation of ERRg (Fig. 1C). Be-cause ERRa and ERRg are both associated with hepaticglucose metabolism, we also examined whether OGToverexpression could lead to ERRa O-GlcNAcylation. Un-like ERRg, O-GlcNAcylation was not detected for ERRa(Fig. 1D).

O-GlcNAcylation is linked with protein stability (30–33).From our results (Fig. 1A and B), we speculated thatO-GlcNAcylation could stabilize ERRg by decreasing protein

degradation. To test this, HEK 293T cells were cotrans-fected with ERRg and OGA or OGT expression vectorsfollowed by MG-132 treatment to inhibit proteasome-mediated protein degradation. ERRg ubiquitination lev-els were significantly raised and O-GlcNAcylation levelswere decreased in presence of OGA, whereas OGT had anentirely opposite effect, suggesting that OGT triggersERRg O-GlcNAcylation that inhibits ERRg ubiquitinationand stabilizes it (Fig. 1E). Next, to examine the functionalimplications of O-GlcNAcylation, reporter gene assay withtransient transfection was carried out in the 293T cellline. ERRg significantly enhanced the Sft4-luc reporteractivity, which was further augmented in presence ofOGT. OGA had an inverse effect to that of OGT andsignificantly reduced ERRg transcriptional activity. How-ever, ERRa could not markedly activate the reporter geneeven in presence of OGT (Fig. 1F). Activation of the Gal4-tk-luc reporter gene by Gal4–ERRg–LBD was significantlyaugmented by OGT, whereas OGA significantly repressedit, indicating that O-GlcNAcylation of ERRg might occurin its LBD (Fig. 1G). Overall, these results suggest thatERRg is subject to O-GlcNAcylation, which in turn in-creases protein stability by decreasing ubiquitin mediated

Figure 1—ERRg is modified by O-GlcNAcylation. A: MPH were treated with GlcN in a dose-dependent manner for 6 h and subsequentlyimmunoprecipitated (IP) with ERRg antibody. Total proteins were analyzed for the O-GlcNAcylation by immunoblot with O-GlcNAc, OGT,ERRg, and a-tubulin antibodies. B: MPH were incubated in 5 mmol/L or 25 mmol/L glucose media overnight followed by immunoprecip-itation with ERRg antibody and immunoblot with O-GlcNAc, ERRg, and a-tubulin antibodies. HEK 293T cells were transfected withexpression vector for HA-ERRg (C) and HA-ERRa (D) alone or with expression vector for FLAG-OGT, followed by immunoprecipitationwith HA antibody and immunoblot with O-GlcNAc, HA, and a-tubulin antibodies. E: HEK 293T cells transfected with plasmids encodingHA-ERRg and FLAG-OGA or FLAG-OGT. At 24 h after transfection, cells were treated with MG-132 (10 mmol) for 6 h followed byimmunoprecipitation with HA antibody and immunoblot with O-GlcNAc, ubiquitin (Ub), HA, and a-tubulin antibodies. F: HEK 293T cellswere transfected with Sft4-luc along with the indicated plasmids. G: HEK 293T cells were transfected with Gal4-tk-luc along with theindicated plasmids. Error bars show 6 SEM. *P < 0.05, **P < 0.005 using Student t test.

2838 O-GlcNAcylation of ERRg Diabetes Volume 65, October 2016

protein degradation and also enhances ERRg transcrip-tional activity.

Glucagon Increases O-GlcNAcylation of ERRgHepatic ERRg expression is increased by fasting-dependentactivation of the CREB-CRTC2 pathway (18,28). Hence, wesought to determine whether fasting-dependent increasein ERRg expression is associated with O-GlcNAcylation.Fasting significantly enhanced O-GlcNAcylation of ERRgcompared with the fed state. This stands in contrast toERRg ubiquitination levels that were higher in the fedcondition than the fasting condition, suggesting thatO-GlcNAcylation mediates ERRg stability in the fastingcondition (Fig. 2A). To confirm that O-GlcNAcylationmediates ERRg stability in the fasting condition, weoverexpressed OGA in this condition. Overexpressionof OGA significantly reduced O-GlcNAcylation of ERRgand simultaneously increased ubiquitination of ERRg,resulting in lowering of ERRg abundance, unambigu-ously establishing that ERRg stability is governed byO-GlcNAcylation in the fasting condition (Fig. 2B). In-creased protein stability is associated with decreasedinteraction between protein and E3 ubiquitin-ligasessuch as Mdm2 (32). We observed that fasting increasedOGT–ERRg interactions in a time-dependent manner,resulting in enhanced O-GlcNAcylation and reducedMdm2–ERRg interactions in mice (Fig. 2C). The bindingof OGT and ERRg was reduced, but the levels of ERRgO-GlcNAcylation were still intensified after 3 h of fast-ing (Fig. 2C). To clarify this, we tested whether theinteractions between ERRg and OGT or OGA might bedynamically altered during the fasting period. The in-teraction between ERRg and OGT increased in a fastingtime manner, reaching a peak between 3 and 6 h andthen declining, whereas the interaction between ERRgand OGA decreased in a fasting time manner (SupplementaryFig. 1C), suggesting that in the fed state, OGA interacts withERRg and destabilizes it, whereas in the fasting conditionOGT interacts with ERRg and stabilizes it. Interestingly,protein levels of both OGT and OGA were increased in thefasting (Supplementary Fig. 1C), which was supportedby elevated OGT and OGA mRNA levels in fasting (Sup-plementary Fig. 1D). In spite of low circulating glucose,fasting promoted O-GlcNAcylation of ERRg. Hence, wespeculate that glucagon could be more important thanblood glucose for O-GlcNAcylation of ERRg during fasting.To test this, we measured blood glucose and glucagon levelsin a fasting time-course experiment (Supplementary Fig. 1Eand F). Blood glucose levels in fasting mice decreased from0 to 12 h, whereas serum glucagon levels steadily increasedfrom 0 h, reaching a peak at 6 h, followed by a decrease at12 h, though it remained significantly higher at 12 h than at0 h. ERRg O-GlcNAcylation levels gradually increased from0 h, reaching a peak at 6 h and then remaining almost thesame until 12 h, suggesting that circulating glucagon levelsare more important than circulating glucose levels forO-GlcNAcylation of ERRg during fasting.

Glucagon increases hepatic ERRg transcriptional ac-tivity during fasting (28). Thus, we sought to ascertainwhether glucagon increases ERRg transcriptional activityby affecting O-GlcNAcylation. Glucagon treatment sig-nificantly increased ERRg O-GlcNAcylation levels, result-ing in higher protein stability through reduced ERRgubiquitination (Fig. 2D). As both high glucose (Fig. 1B)and glucagon (Fig. 2D) induced O-GlcNAcylation, wetested which is more important for O-GlcNAcylation ofERRg. Both high levels of glucose and glucagon increasedERRg O-GlcNAcylation levels and total protein levels.They have a cumulative effect when applied together(Supplementary Fig. 1G), but unlike glucagon, glucosedid not increase ERRg mRNA levels (SupplementaryFig. 1H). The notion that O-GlcNAcylation mediatesthe effect of glucagon on ERRg protein stability wasfurther corroborated when overexpression of OGA sig-nificantly reduced both O-GlcNAcylation and proteinlevels of ERRg (Fig. 2E). Insulin induced by feeding sup-presses ERRg transcriptional activity as well as gene ex-pression (20,28). Thus, we speculate that insulin mayinhibit O-GlcNAcylation of ERRg, leading to decreasedprotein stability. As we expected, glucagon-induced ERRgO-GlcNAcylation was significantly suppressed by insulintreatment in AML12 cells. ERRg ubiquitination was mark-edly enhanced in presence of insulin compared with glu-cagon treatment, resulting in lower protein stability (Fig.2F). Taken together, these results indicate that fastingincreases O-GlcNAcylation of ERRg, leading to proteinstability, whereas feeding acts reciprocally to degrade ERRgby suppressing its O-GlcNAcylation.

ERRg Is Modified by O-GlcNAc at S317 and S319In order to identify the O-GlcNAc site of ERRg, we in-fected mice with Ad-FLAG-ERRg and extracted the ERRgprotein from liver tissue through immunoprecipitation(Fig. 3A and B). The mass spectrometry results showthat S317 and S319 of ERRg are O-GlcNAcylated (Supple-mentary Figs. 2 and 3). The primary structure of ERRgrevealed that S317 and S319 were present in the LBD ofERRg (Fig. 3C), which was consistent with the previousresult that suggested O-GlcNAcylation site is in the LBD(Fig. 1G). However, only S317 was conserved in ERRa,ERRb, and ERRg (Fig. 3C). The three-dimensional struc-ture of ERRg–LBD provides a comprehensive view of theO-GlcNAcylation site. The two O-GlcNAcylated serine res-idues are located at the end of helix5 of the reportedERRg–LBD structures, which is locally stabilized withhelix 6, helix 7, and a couple of strands. Although S319is completely exposed to the solvent-accessible surface,S317 is partially hidden (Fig. 3D). To verify the specificO‐GlcNAcylation sites indicated by mass analysis, weconstructed three site‐specific point mutant cDNAsof ERRg, S317A ERRg, S319A ERRg, and S317A+S319AERRg. The single mutants (S317A ERRg and S319A ERRg)showed less O-GlcNAcylation compared with wild-typeERRg, whereas the double mutant (S317A+S319A ERRg)

diabetes.diabetesjournals.org Misra and Associates 2839

showed complete absence of an O-GlcNAc signal in thepresence of GlcN (Fig. 3E). Interestingly, even thoughthe single mutants showed higher O-GlcNAcylation thanthe double mutant, all three mutants were highly ubiq-uitinated compared with wild-type in presence of GlcN,unambiguously suggesting that O-GlcNAcylation of bothS317 and S319 is required for ERRg protein stability (Fig.3E). Consistent with these findings, we noticed that allthree mutants showed significantly lower stability com-pared with wild-type in a pulse-chase experiment, demon-strating the importance of O-GlcNAcylation in ERRgprotein stability (Fig. 3F). This was further confirmed

when all three mutants showed significantly reducedprotein levels in the presence of cycloheximide, a pro-tein synthesis inhibitor (Supplementary Fig. 4A). Next,reporter gene assay reveled that unlike wild-type, themutant ERRg had no significant transcriptional activ-ity, suggesting that O-GlcNAcylation is required forERRg transcriptional activity as well (Fig. 3G). All threemutants were unable to bind to the PEPCK promoterrevealing why they had no transcriptional activity(Fig. 3H). Collectively, these results illustrate thatO-GlcNAcylation stabilizes ERRg and increases its tran-scriptional activity.

Figure 2—Glucagon increases O-GlcNAcylation of ERRg. A: C57BL/6 mice (n = 5) were fasted for 6 h and refed for 2 h, followed by livertissue isolation. Immunoprecipitation (IP) with ERRg antibody and immunoblot with O-GlcNAc, ubiquitin (Ub), ERRg, and a-tubulin anti-bodies were performed. All of the samples were combined for Western blot analyses. B: Ad-green fluorescent protein (GFP) or Ad-OGA wasadministered via tail-vein injection into C57BL/6 mice (n = 6). At day 6 after injection, mice were fasted for 6 h and sacrificed. Liver tissuefrom mice was homogenized for immunoprecipitation with ERRg antibody and immunoblot with O-GlcNAc, Ub, ERRg, and a-tubulinantibodies. All of the samples were combined for Western blot analyses. C: Liver tissue from C57BL/6 mice (n = 5), fasted for differenttime points as indicated, was homogenized for immunoprecipitation with ERRg antibody and immunoblot with O-GlcNAc, OGT, Mdm2,ERRg, and a-tubulin antibodies. All of the samples were combined for Western blot analyses. D: AML12 cells were transfected with FLAG-ERRg. At 24 h posttransfection, cells were incubated with vehicle or glucagon (100 nmol/L) for different time points as indicated. Immu-noprecipitation with FLAG antibody and immunoblot with O-GlcNAc, Ub, FLAG, and a-tubulin antibodies were performed. E: AML12 cellswere transfected with FLAG-ERRg. At 24 h posttransfection, cells were infected with Ad-GFP or Ad-OGA as indicated. At 24 h postinfection,cells were incubated with glucagon (100 nmol/L) for 6 h followed by immunoprecipitation with FLAG antibody and immunoblot with O-GlcNAc,Ub, FLAG, and a-tubulin antibodies were performed. F: AML12 cells were transfected with FLAG-ERRg. At 24 h posttransfection, cells wereincubated with glucagon (100 nmol/L) for 6 h or glucagon (100 nmol/L) for 6 h followed by insulin (100 nmol/L) for 6 h. Immunoprecipitation withFLAG antibody and immunoblot with O-GlcNAc, OGT, Ub, FLAG, and a-tubulin antibodies were performed. hr, hour.

2840 O-GlcNAcylation of ERRg Diabetes Volume 65, October 2016

Figure 3—ERRg is modified by O-GlcNAcylation at S317 and S319. A: C57BL/6 mice (n = 5) were injected via tail veins with adenoviralvector expressing FLAG-ERRg. Six days later, mice were fasted for 6 h, and liver extracts were prepared. The FLAG-ERRg was purifiedusing anti-FLAG M2 agarose. B: Proteins were separated using SDS-PAGE and stained by colloidal Coomassie Blue. C: Schematicdiagram of ERRg structure (amino acids 1–458). ERRg consists of N-terminal domain, DBD, hinge domain, LBD, and COOH-terminaldomain. ERRg is O-GlcNAcylated at S317 and S319. The amino acid sequences of ERR family members ERRa, ERRb, and ERRg are givenbelow. D: Two serine residues of ERRg-LBD. Twelve LBD structures (Protein Data Bank identification numbers 2ewp, 2gpp, 2gp7, and2p7z) were superimposed, which showed root mean SD values of <0.4 Å for all the aligned residues. ERRg-AF-2 domain is represented asan orange color surface and ERRg–LBD as a gray-colored surface with two serine residues in space-filling models in the left and bottomright panels. In the top right panel, the superimposed ERRg–LBD structures are displayed with ribbons and differentiated by colors;two serine residues are drawn as stick models. E: AML12 cells were transfected with plasmids encoding FLAG-wt ERRg, FLAG-S317AERRg, FLAG-S319A ERRg, and FLAG-S317A+S319A ERRg. At 24 h after transfection, cells were treated with GlcN (10 mmol/L) for 6 hand MG-132 (10 mmol/L) for 6 h followed by immunoprecipitation (IP) with FLAG antibody and immunoblot with O-GlcNAc, ubiquitin(Ub), FLAG, and a-tubulin antibodies. F: AML12 cells transfected with FLAG-wt ERRg, FLAG-S317A ERRg, FLAG-S319A ERRg, andFLAG-S317A+S319A ERRg were metabolically labeled for 30 min followed by chase for the indicated times. FLAG-ERRg wasimmunoprecipitated with M2 antibody and detected by autoradiography. G: ERRg-dependent activation of the Sft4-luc. Transienttransfection was performed in HEK 293T cells with the indicated plasmid DNAs. H: ChIP assay showing the occupancy of wild-typeERRg, S317A ERRg, S319A ERRg, and S317A+S319A ERRg on ERRE1 of PEPCK1 promoter from DMSO and GlcN-treated AML12cells. Error bars show 6 SEM. ***P < 0.0005 using Student t test.

diabetes.diabetesjournals.org Misra and Associates 2841

O-GlcNAcylation Affects ERRg–PGC-1a Interactionbut Not ERRg Cellular LocalizationIncrease in transcriptional activity of transcription factorsin response to O-GlcNAcylation was previously linked totheir nuclear transport (7,34). Because the O-GlcNAcylationmutant ERRg had practically no transcriptional activitycompared with wild-type (Fig. 3G and H), we speculatethat the O-GlcNAcylation mutant may not translocateinto the nucleus. Unexpectedly, the S317A+S319A ERRgwas located in the nucleus, indicating that subcellular lo-calization of ERRg was not governed by O-GlcNAcylation(Fig. 4A). Transcriptional coactivator PGC-1a interactswith ERRg and is critical for ERRg transcriptional activity(28). Hence, we examined whether PGC-1a could interactwith S317A+S319A ERRg. In AML12 cells, GlcN treat-ment significantly augmented O-GlcNAcylation of ERRgas well as ERRg–PGC-1a interaction, whereas inhibit-ing HBP by treating with DON, an inhibitor of GFAT, therate-limiting enzyme of HBP, significantly reduced O-GlcNA-cylation of ERRg as well as ERRg–PGC-1a interaction. We

could not detect any interaction between S317A+S319AERRg and PGC-1a, supporting the idea that the HBP me-diates ERRg–PGC-1a interaction (Fig. 4B). We also observedthat GlcN treatment increased PGC-1a protein levels (Fig.4B), but not mRNA levels (Supplementary Fig. 4B). The risein PGC-1a protein levels could be because O-GlcNAcylationalso stabilizes PGC-1a protein (30). Next, STZ treatment, aninhibitor of OGA and promoter of O-GlcNAcylation (7,35),significantly augmented PGC-1a–ERRg interaction, demon-strating that OGA inhibition enhances O-GlcNAcylation ofERRg that results in substantial increase in PGC-1a–ERRginteraction (Fig. 4C). Along with ERRg (FLAG), PGC-1a (HA)protein levels were also elevated in response to STZ treat-ment (Fig. 4C), which is consistent with Fig. 4B, in which GlcNtreatment elevated PGC-1a protein levels. Next, we per-formed in vitro interaction study between PGC-1a and Gal4construct containing either wild-type or S317A+S319AERRg-LBD as the LBD contains the O-GlcNAcylation site(Fig. 3C). Similar to Gal4–wild-type ERRg–LBD, Gal4-S317A+S319A ERRg-LBD protein was as stable as the

Figure 4—O-GlcNAcylation affects ERRg–PGC-1a interaction but not ERRg cellular localization. A: Immunocytochemistry showing sub-cellular localization of wild-type and mutant ERRg in AML12. The cells were transfected with plasmids expressing FLAG-wt ERRg andFLAG-S317A+S319A ERRg. At 24 h after transfection, cells were fixed, immunostained, and observed by confocal microscopy. B: AML12cells were transfected with plasmids encoding FLAG-wt ERRg and FLAG-S317A+S319A ERRg. At 24 h after transfection, cells were treatedwith MG-132 (10 mmol/L) and DMSO or GlcN (10 mmol/L) or DON (40 mmol/L) for 6 h followed by immunoprecipitation (IP) with FLAGantibody and immunoblot with O-GlcNAc, PGC-1a, FLAG, and a-tubulin antibodies. C: AML12 cells were transfected with plasmidsencoding FLAG-wt ERRg and HA-PGC-1a. At 24 h after transfection, cells were treated with DMSO or OGA inhibitor STZ (5 mmol/L)for 6 h followed by immunoprecipitation with HA antibody and immunoblot with FLAG, HA, and a-tubulin antibodies. D: HEK 293T cellswere transfected with plasmids encoding Gal4-wt ERRg–LBD, Gal4-S317A+S319A ERRg-LBD, and HA-PGC-1a. At 24 h posttransfection,immunoprecipitation with HA antibody and immunoblot with Gal4 and HA antibodies were performed. E: HEK 293T cells were transfectedwith Gal4-tk-luc along with the indicated plasmids. Error bars show 6 SEM. *P < 0.05 using Student t test.

2842 O-GlcNAcylation of ERRg Diabetes Volume 65, October 2016

wild-type one, but we could not detect any interactionbetween PGC-1a and Gal4-S317A+S319A ERRg-LBD, sug-gesting that O-GlcNAcylation in the LBD regulates ERRg–PGC-1a interaction (Fig. 4D). Similar to the double mutant,the two single mutants also could not interact with PGC-1a(Supplementary Fig. 4C). Gal4–tk-Luc reporter gene assayrevealed that Gal4-S317A+S319A ERRg-LBD was incapableof activating the reporter gene in the presence of PGC-1a(Fig. 4E). Moreover, Gal4-S317A+S319A ERRg-LBD wasalso unable to activate the reporter gene even in presenceof GlcN or OGT (Supplementary Fig. 4D). Taken together,these results demonstrate that O-GlcNAcylation regulatesERRg–PGC-1a interaction critical to ERRg transcriptionalactivity.

O-GlcNAcylation Regulates ERRg-MediatedGluconeogenic Gene ExpressionHBP induces gluconeogenic enzymes gene expressionthrough O-GlcNAcylation (7), (30). ERRg is a key positiveregulator of gluconeogenic enzymes gene expression (28),(18). Our previous results described that ERRg wasO-GlcNAcylated through HBP (Figs. 1A, 4B and C). There-fore, to determine the contribution of O-GlcNAcylatedERRg in HBP-induced gluconeogenesis in MPH, weknocked down endogenous ERRg. ERRg knockdown

markedly reduced GlcN-induced PEPCK and G6Pase pro-tein levels (Fig. 5A). OGT overexpression leads to induc-tion of gluconeogenesis and OGT knockdown improvesglucose homeostasis in diabetic mice (7,30). OGT over-expression significantly increased PEPCK and G6PasemRNA levels, and this increase in mRNA levels wasgreatly suppressed by ERRg knockdown in MPH (Fig.5B, from left, first and second panel). In line with PEPCKand G6Pase mRNA levels results, glucose production wasalso significantly reduced in response to ERRg knockdown(Fig. 5B, rightmost panel). Effectiveness of OGT overex-pression was confirmed by Western blot analyses (Supple-mentary Fig. 4E). Next, to examine the effect of glucose orGlcN on gluconeogenic gene promoter activity, we usedwild-type and ERRE mutant PEPCK promoter, which isdevoid of ERRg binding site. Exposure to glucose or GlcNincreased wild-type promoter activity, but this increase wasgreatly reduced with the ERRE mutant PEPCK pro-moter (Fig. 5C). In a parallel approach, 293T cells weretransfected with wild-type PEPCK promoter along withwild-type and S317A+S319A ERRg. Wild-type ERRg con-siderably increased the promoter activity that wasfurther augmented in presence of OGT, whereas OGAcoexpression greatly impaired ERRg effect. At the same

Figure 5—O-GlcNAcylation regulates ERRg transcriptional activity. A: Effect of ERRg knockdown on GlcN-induced gluconeogenic genesin MPH. Hepatocytes were infected with Ad-unspecific (US) or Ad-shERRg. At 48 h postinfection, cells were treated with DMSO or GlcN(10 mmol/L) for 6 h followed by immunoblot with PEPCK, G6Pase, ERRg, and a-tubulin antibodies. B: Effect of ERRg knockdown on basaland OGT-induced gluconeogenic genes in MPH. Glucose output assay was performed in MPH. C: Effect of Glc and GlcN on wild-type andERRg binding site (ERRE) mutant (mut) PEPCK-luc. HEK 293T cells were transfected with indicated plasmids followed by Glc or GlcNtreatment. D: O-GlcNAcylation–dependent activation of the PEPCK1-luc by ERRg. HEK 293T cells were transfected with indicatedplasmids. E: ChIP assay showing the occupancy of ERRg on ERRE1 of PEPCK1 promoter from GlcN- and DON-treated MPH. Error barsshow 6 SEM. *P < 0.05, **P < 0.005 using Student t test.

diabetes.diabetesjournals.org Misra and Associates 2843

time, S317A+S319A ERRg could not greatly activate thepromoter (Fig. 5D). Next, chromatin immunoprecipitation(ChIP) assay was performed in MPH to monitor the effectof HBP inhibition on ERRg recruitment to the endogenousPEPCK gene promoter. Under basal conditions, ERRg occu-pied the PEPCK promoter. However, GlcN treatment sig-nificantly augmented ERRg occupancy on the PEPCKpromoter, whereas HBP inhibition by DON treatmentmarkedly diminished the occupancy. We also observed asimilar binding pattern of PGC-1a (Fig. 5E). Together,these results demonstrate that O-GlcNAcylation by HBPgoverns gluconeogenic activity of ERRg.

ERRg O-GlcNAcylation Is Required for HepaticGluconeogenesisDiabetic conditions induce ERRg gene expression and pro-mote gluconeogenesis (18). Hence, we speculated thatERRg could be O-GlcNAcylated under diabetic conditions.As we expected, O-GlcNAcylation of ERRg was greatly in-creased in HFD-fed, ob/ob, and db/db mice (Fig. 6A and B).Previously, it was reported that diabetic conditions pro-moted OGT gene expression (7). We also noticed a sig-nificant increase in OGT mRNA levels in HFD-induceddiabetic mouse, although OGA mRNA levels were alsoelevated (Supplementary Fig. 4F). Overexpression of ERRgpromotes hepatic gluconeogenesis and elevates blood glu-cose levels (18,28). Therefore, we compared the effect ofwild-type and S317A+S319A ERRg overexpression inmouse liver. In accordance with previous results, glucoseexcursion during intraperitoneal glucose tolerance testwas significantly higher in Ad–wild-type ERRg-injectedmice compared with control mice, but Ad-S317A+S319AERRg–injected mice showed normal blood glucose levels(Fig. 6C). Fasting blood glucose levels and PEPCK andG6Pase mRNA levels were significantly higher for Ad–wild-type ERRg infection compared with Ad-S317A+S319AERRg (Fig. 6D). Effectiveness of ERRg overexpression wasconfirmed by Western blot analyses (Supplementary Fig.4G). Next, to negate the effect of O-GlcNAcylation, we over-expressed OGA in mouse liver. Disrupting O-GlcNAcylationof ERRg in mice infected with Ad–wild-type ERRg throughoverexpression of hepatic OGA by Ad-OGA greatly loweredthe gluconeogenic profile (Fig. 6E). Effectiveness of ERRgoverexpression was confirmed by Western blot analyses(Supplementary Fig. 4H). Finally, based on the previousresult (Fig. 6C and D), we performed in vivo imaging anal-ysis to verify the effect of O-GlcNAcylation of ERRg onhepatic gluconeogenesis at the transcriptional levels. Ad–wild-type ERRg-dependent induction of PEPCK promoteractivity was significantly reduced in mice injected withAd-S317A+S319A ERRg (Fig. 6F). Overall, these resultssuggest that O-GlcNAcylation is prerequisite for ERRg totrigger hepatic gluconeogenesis.

DISCUSSION

Several transcription factors promote gluconeogenesisin the fasted state and type 2 diabetes. We hypothesize

that in the fasting and diabetic states, gluconeogenesisgenerates fructose-6-phosphate, which is used by HBP toO-GlcNAcylate transcription factors and coactivators tofurther promote gluconeogenesis. In the current study,we show that ERRg is stabilized by O-GlcNAcylation in thefasted and diabetic states to promote gluconeogenic geneinduction. Hence, we suggest a positive feed-forward loopin which glucose entry into the HBP promotes gluconeo-genesis in the fasting and diabetic conditions in the liver.

Glucagon–insulin crosstalk regulates ERRg proteinstability and transcriptional activity (20,28). In this study,we investigated whether ERRg protein stability and transcrip-tional activity is influenced by O-GlcNAcylation. In fact, glu-cagon stabilized ERRg by promoting its O-GlcNAcylation (Fig.2D–F). O-GlcNAcylation can increase protein stabilityand transcriptional activity by inhibiting ubiquitinationor promoting deubiquitination (30–33). It can also re-duce protein stability and transcriptional activity by in-creasing ubiquitination (36,37). However, we observedthat O-GlcNAcylation stabilized ERRg protein by inhibit-ing its ubiquitination (Fig. 2D–F). Our results clearly dem-onstrated that incremental O-GlcNAcylation–mediatedreduction in ubiquitination was a result of greater in-hibition of the interaction between ERRg and E3 ubiq-uitin ligase Mdm2 that resulted in increased proteinstability (Fig. 2C). Phosphorylation of GFAT1 by cAMP-dependent protein kinase blocks its enzyme activity (38),whereas phosphorylation of GFAT2 by cAMP-dependentprotein kinase increases its enzyme activity (39). Our ob-servation that glucagon robustly enhances O-GlcNAcylationof ERRg and insulin inhibits it (Fig. 2F) could be due toGFAT2 activation. O-GlcNAcylation can affect transcrip-tion factors by modifying key residues involved in theirinteraction with coactivators (40). It can also induce im-portant conformational changes within transcription fac-tors, which might have a direct impact on their activity, asdemonstrated for the estrogen receptor (41). Three-dimensional structural features of ERRg also suggest thatO-GlcNAcylation may trigger a conformational change ofERRg–LBD that contains the AF-2 domain (Fig. 3D). Thisprobable conformational change may be crucial for inhibi-tion of ubiquitination and ERRg–PGC-1a interaction. Per-haps, S317A ERRg and S319A ERRg, in spite of beingpartially O-GlcNAcylated, were heavily ubiquitinated (Fig.3E) and unable to interact with PGC-1a (SupplementaryFig. 4C). The two single mutants may not attain the de-sired conformational change required to inhibit ubiqui-tination and interact with coactivator PGC-1a due toincomplete O-GlcNAcylation. Further investigation isrequired to confirm whether O-GlcNAcylation influencesERRg–LBD structure. Previously, Yang et al. (42) report-ed that the transcripts of ERRg followed a cyclic patternin its diurnal rhythmicity in the liver. ERRg transcriptsreach maximum levels in the liver during the daytime.The observed result of O-GlcNAcylation–mediated stabilityof ERRg may have been influenced by the diurnal rhyth-micity in the liver.

2844 O-GlcNAcylation of ERRg Diabetes Volume 65, October 2016

O-GlcNAcylation takes place in response to high glucoseor insulin (31,43,44). PGC-1a and CRTC2 undergoO-GlcNAcylation in hyperglycemic and hyperinsulinemicdiabetic mouse (7,30). Interestingly, low-glucose conditionsalso promote O-GlcNAcylation (7,9,30). Glucose dep-rivation induces protein O-GlcNAcylation and OGTexpression as well (45). All of these reports suggest

that both high- and low-glucose conditions promoteO-GlcNAcylation in vitro and in vivo. Glucose in theform of fructose-6-phosphate is used by hexosamine bio-synthetic pathway to O-GlcNAcylate proteins under dif-ferent conditions. In the fed conditions, high circulatingglucose could be converted to fructose-6-phosphate andused in O-GlcNAcylation. Although the circulating glucose

Figure 6—ERRg O-GlcNAcylation required for hepatic gluconeogenesis. O-GlcNAcylation of ERRg in diabetic conditions. ERRg wasimmunoprecipitated (IP) with ERRg antibody followed by immunoblot with O-GlcNAc, ERRg, PEPCK, and a-tubulin antibodies from livertissue of C57BL/6 mice (n = 3) fed with normal chow diet (NCD) and HFD (A) from wild-type, ob/ob, and db/db mice (n = 3) (B). C: Ad-GFP,Ad-wt ERRg, or Ad-S317A+S319A ERRg were administered via tail-vein injection into C57BL/6 mice (n = 5). Glucose tolerance test at day 5 afterinjection. Glucose was measured at the indicated times after 1 g/kg intraperitoneal glucose injection. D, left panel: 4-h fasting blood glucoselevels in C57BL/6 mice (n = 5) at day 6 after Ad-GFP, Ad-wt ERRg, or Ad-S317A+S319A ERRg injection. D, middle and right panels: qPCRanalyses of total RNA isolated from liver at day 7 after injection. E, left panel: 4-h fasting blood glucose levels in C57BL/6 mice (n = 5) at day6 after Ad-GFP, Ad-ERRg, or Ad-ERRg + Ad-OGA injection. E, middle and right panels: qPCR analyses of total RNA isolated from mice liver atday 7 after injection. F, left panel: in vivo imaging of hepatic PEPCK1-luciferase (Ad-PEPCK1-luc) activity in presence of Ad-GFP or Ad-wt ERRgor Ad-S317A+S319A ERRg in fasted (16 h) C57BL/6 mice (n = 3). F, right panel: quantitation of luciferase activity is also shown. G: Schematicrepresentation of the proposed model that the fasting and diabetic conditions promote O-GlcNAcylation of ERRg. O-GlcNAcylation is imperativefor ERRg protein stability and enhances gluconeogenic activity of ERRg. Fed state, however, reduces O-GlcNAcylation of ERRg, resulting inubiquitin-mediated degradation of ERRg. Error bars show 6 SEM. sec, second. *P < 0.05, **P < 0.005, ***P < 0.0005 using Student t test.

diabetes.diabetesjournals.org Misra and Associates 2845

concentration is high in the diabetic condition, the dia-betic condition differs from the fed condition in that theactivity of glucokinase, which converts glucose to glucose-6-phosphate, is low in the liver (46), (47,48). Therefore,circulating glucose is less readily converted to fructose-6-phosphate in the diabetic condition. In the diabetic condi-tions hepatic gluconeogenesis is significantly upregulated(49), producing fructose-6-phosphate, which could be usedin O-GlcNAcylation. Furthermore, in low circulating glucose(fasting) conditions, glycogenolysis and gluconeogenesis areboth called into play to maintain blood glucose levels. Duringearly fasting, glycogenolysis, stimulated by glucagon, pro-duces glucose-6-phosphate, which is in equilibriumwith fructose-6-phosphate (50). Therefore, glycogenolysis,especially during early fasting, is surely a major source offructose-6-phosphate. During short-term fasting, glucagonalso triggers the initial induction of hepatic gluconeogene-sis through activation of CREB-CRTC2 (7). In prolongedfasting, PGC-1a, FOXO1, and ERRg promote hepatic glu-coneogenesis (28,30,51). During that initial inductionperiod (short-term fasting), hepatic glycogenolysis and gluco-neogenesis produce fructose-6-phosphate, which might beused to O-GlcNAcylate ERRg to further promote gluconeo-genesis during prolonged fasting. We observed that bloodglucose levels were decreased during fasting in a time-dependent manner, whereas serum glucagon levels andERRg O-GlcNAcylation levels were steadily increased (Sup-plementary Fig. 1E and F), indicating that circulating gluca-gon levels are more important than circulating glucose levelsfor O-GlcNAcylation of ERRg in fasting. Moreover, OGTmRNA levels were significantly enhanced in response tofasting, even though OGA mRNA levels were also enhanced,but it was less significant compared with OGT (Supplemen-tary Fig. 1D). Enhanced OGT mRNA levels could also beresponsible for fasting-dependent O-GlcNAcylation of ERRg.

Conserved ERRE on the PEPCK promoter is required forthe transcription of that gene in response to fasting anddiabetes-mediated gluconeogenesis (18,28); hence, we ex-plored the role of ERRg in HBP mediated gluconeogenesis.Loss of endogenous ERRg in primary hepatocytes led to asignificant decrease in HBP-induced gluconeogenic profile(Fig. 5), implying the importance of O-GlcNAcylation ofERRg in the context of gluconeogenesis. Diabetic condi-tions promote O-GlcNAcylation mediated gluconeogene-sis in the diabetic mice (7,30). As a matter of fact, ERRgwas highly O-GlcNAcylated in diabetic mice (Fig. 6A and B).Hepatic overexpression of wild-type ERRg caused glucoseintolerance with hyperglycemia, whereas O-GlcNAcylationmutant ERRg overexpression showed glucose tolerance witheuglycemia in mice (Fig. 6C and D). O-GlcNAcylation wasblocked by using either OGT or GFAT inhibitor or by enzy-matically modulating OGA or OGT expression to investigatethe effect of their inhibition on glycemia in diabeticmice. The OGT inhibitor alloxan was used in many stud-ies (52–54), but it has wide off-target effects and generalcellular toxicity (55). Another OGT inhibitor, Ac4-5S-GlcNAc,was used in wide range of studies (56–58), but it affects other

glycosyltransferase and impairs N-glycosylation and extracel-lular glycan synthesis in cultured cell lines (59). More-over, the enzymatic approach modulating OGA or OGTexpression was successfully used to restore glucose homeo-stasis. Overexpression of hepatic OGA or knockdown ofhepatic OGT inhibited aberrant gluconeogenesis and signif-icantly improved glycemic conditions in diabetic mice(7,30,31). We observed that disruption of O-GlcNAcylationby hepatic OGA overexpression reduced gluconeogenic pro-file in normal mice expressing Ad-ERRg (Fig. 6E), unambig-uously illustrating that O-GlcNAcylation is prerequisite forgluconeogenic function of ERRg.

We conclude that the fasting and diabetic conditionspromote O-GlcNAcylation of ERRg. O-GlcNAcylation isimperative for ERRg protein stability and enhances glu-coneogenic activity of ERRg. The fed state, however, reducesO-GlcNAcylation of ERRg, resulting in ubiquitin-mediateddegradation of ERRg (Fig. 6G). Our results indicate a vitalrole for O-GlcNAcylation of ERRg in maintaining normalglucose levels during fasting and also in mediating theelevated blood glucose levels in type 2 diabetes. Hence,pharmacological inhibition of O-GlcNAcylation–mediatedhyperactivation of ERRg might provide a pathway forpreventing hyperglycemia and treating type 2 diabetes.

Acknowledgments. The authors thank David Moore (Baylor College ofMedicine, Houston, TX) and Seok-Yong Choi (Department of Biomedical Sciences,Chonnam National University Medical School, Gwangju, Republic of Korea) forhelpful discussion and Ji Min Lee, Yong Soo Lee, Ki Sun Kim, Soon-Young Na,and Yaochen Zhang (Chonnam National University, Gwangju, Republic of Korea)for technical assistance.Funding. This work was supported by National Creative Research InitiativesGrant 20110018305 through the National Research Foundation of Korea (NRF)funded by the Korean government (Ministry of Science, ICT & Future Planning) (toH.-S.C.). This research was supported by Grant HI16C1501 of the Korea HealthTechnology R&D Project through the Korea Health Industry Development Institute(KHIDI) and funded by the Ministry of Health & Welfare, Republic of Korea (to I.-K.L.).This research was also supported by the NRF funded by the Ministry of Educa-tion, Science, and Technology (Grants NRF-2013R1A2A1A01008067 and NRF-2015M3A9B6073840 to J.W.C.).Duality of Interest. No potential conflicts of interest relevant to this articlewere reported.Author Contributions. J.M. conceived and designed research, performedexperiments, analyzed the data, and wrote the paper. D.-K.K., Y.S.J., E.-K.Y. providedreagents/materials/analysis tools. H.B.K. and B.G.K. performed liquid chromatography-tandem mass spectrometry collision-induced dissociation site mapping analysis foridentification of O-GlcNAcylation sites. Y.-H.K. performed mouse in vivo imaging. S.K.performed liquid chromatography-tandem mass spectrometry collision-induced dis-sociation site mapping analysis for identification of O-GlcNAcylation sites and analyzedthe data. I.-K.L., R.A.H., J.-S.K., and C.-H.L. analyzed the data. J.W.C. analyzed thedata and provided reagents/materials/analysis tools. H.-S.C. conceived and designedresearch, analyzed the data, and wrote the paper. H.-S.C. is the guarantor of thiswork and, as such, had full access to all the data in the study and takes responsibilityfor the integrity of the data and the accuracy of the data analysis.

References1. Hart GW, Slawson C, Ramirez-Correa G, Lagerlof O. Cross talk betweenO-GlcNAcylation and phosphorylation: roles in signaling, transcription, andchronic disease. Annu Rev Biochem 2011;80:825–858

2846 O-GlcNAcylation of ERRg Diabetes Volume 65, October 2016

2. Hebert LF Jr, Daniels MC, Zhou J, et al. Overexpression of glutamine:fructose-6-phosphate amidotransferase in transgenic mice leads to insulinresistance. J Clin Invest 1996;98:930–9363. Veerababu G, Tang J, Hoffman RT, et al. Overexpression of glutamine:fructose-6-phosphate amidotransferase in the liver of transgenic mice results inenhanced glycogen storage, hyperlipidemia, obesity, and impaired glucose tol-erance. Diabetes 2000;49:2070–20784. McClain DA, Lubas WA, Cooksey RC, et al. Altered glycan-dependent sig-naling induces insulin resistance and hyperleptinemia. Proc Natl Acad Sci U S A2002;99:10695–106995. Whelan SA, Dias WB, Thiruneelakantapillai L, Lane MD, Hart GW. Regulationof insulin receptor substrate 1 (IRS-1)/AKT kinase-mediated insulin signaling byO-Linked beta-N-acetylglucosamine in 3T3-L1 adipocytes. J Biol Chem 2010;285:5204–52116. Yang X, Ongusaha PP, Miles PD, et al. Phosphoinositide signalling linksO-GlcNAc transferase to insulin resistance. Nature 2008;451:964–9697. Dentin R, Hedrick S, Xie J, Yates J 3rd, Montminy M. Hepatic glucosesensing via the CREB coactivator CRTC2. Science 2008;319:1402–14058. Housley MP, Rodgers JT, Udeshi ND, et al. O-GlcNAc regulates FoxO ac-tivation in response to glucose. J Biol Chem 2008;283:16283–162929. Housley MP, Udeshi ND, Rodgers JT, et al. A PGC-1alpha-O-GlcNActransferase complex regulates FoxO transcription factor activity in response toglucose. J Biol Chem 2009;284:5148–515710. Kuo M, Zilberfarb V, Gangneux N, Christeff N, Issad T. O-GlcNAc modifi-cation of FoxO1 increases its transcriptional activity: a role in the glucotoxicityphenomenon? Biochimie 2008;90:679–68511. Andrali SS, Qian Q, Ozcan S. Glucose mediates the translocation of NeuroD1by O-linked glycosylation. J Biol Chem 2007;282:15589–1559612. Gao Y, Miyazaki J, Hart GW. The transcription factor PDX-1 is post-translationally modified by O-linked N-acetylglucosamine and this modifi-cation is correlated with its DNA binding activity and insulin secretion inmin6 beta-cells. Arch Biochem Biophys 2003;415:155–16313. Hong H, Yang L, Stallcup MR. Hormone-independent transcriptional acti-vation and coactivator binding by novel orphan nuclear receptor ERR3. J BiolChem 1999;274:22618–2262614. Lui K, Huang Y, Choi HL, et alMolecular cloning and functional study of ratestrogen receptor-related receptor gamma in rat prostatic cells. Prostate 2006;66:1600–161915. Zhang Z, Teng CT. Interplay between estrogen-related receptor alpha(ERRalpha) and gamma (ERRgamma) on the regulation of ERRalpha gene ex-pression. Mol Cell Endocrinol 2007;264:128–14116. Kim DK, Kim YH, Jang HH, et al. Estrogen-related receptor g controls he-patic CB1 receptor-mediated CYP2E1 expression and oxidative liver injury byalcohol. Gut 2013;62:1044–105417. Kim DK, Jeong JH, Lee JM, et al. Inverse agonist of estrogen-related re-ceptor g controls Salmonella typhimurium infection by modulating host ironhomeostasis. Nat Med 2014;20:419–42418. Kim DK, Gang GT, Ryu D, et al. Inverse agonist of nuclear receptor ERRgmediates antidiabetic effect through inhibition of hepatic gluconeogenesis.Diabetes 2013;62:3093–310219. Kim DK, Kim JR, Koh M, et al. Estrogen-related receptor g (ERRg) is anovel transcriptional regulator of phosphatidic acid phosphatase, LIPIN1,and inhibits hepatic insulin signaling. J Biol Chem 2011;286:38035–3804220. Kim DK, Kim YH, Hynx D, et al. PKB/Akt phosphorylation of ERRg con-tributes to insulin-mediated inhibition of hepatic gluconeogenesis. Diabetologia2014;57:2576–258521. Yoon JC, Puigserver P, Chen G, et al. Control of hepatic gluconeogenesisthrough the transcriptional coactivator PGC-1. Nature 2001;413:131–13822. Puigserver P, Rhee J, Donovan J, et al. Insulin-regulated hepatic gluco-neogenesis through FOXO1-PGC-1alpha interaction. Nature 2003;423:550–55523. Koo SH, Satoh H, Herzig S, et al. PGC-1 promotes insulin resistance in liverthrough PPAR-a-dependent induction of TRB-3. Nat Med 2004;10:530–534

24. Kim YD, Park K-G, Lee Y-S, et al. Metformin inhibits hepatic gluconeo-genesis through AMP-activated protein kinase-dependent regulation of theorphan nuclear receptor SHP. Diabetes 2008;57:306–31425. Xie YB, Park JH, Kim DK, et al. Transcriptional corepressor SMILE recruitsSIRT1 to inhibit nuclear receptor estrogen receptor-related receptor gammatransactivation. J Biol Chem 2009;284:28762–2877426. Sanyal S, Kim JY, Kim HJ, et al. Differential regulation of the orphan nuclearreceptor small heterodimer partner (SHP) gene promoter by orphan nuclearreceptor ERR isoforms. J Biol Chem 2002;277:1739–174827. Heard DJ, Norby PL, Holloway J, Vissing H. Human ERRgamma, a thirdmember of the estrogen receptor-related receptor (ERR) subfamily of orphannuclear receptors: tissue-specific isoforms are expressed during developmentand in the adult. Mol Endocrinol 2000;14:382–39228. Kim DK, Ryu D, Koh M, et al. Orphan nuclear receptor estrogen-relatedreceptor g (ERRg) is key regulator of hepatic gluconeogenesis. J Biol Chem2012;287:21628–2163929. Park YY, Ahn SW, Kim HJ, et al. An autoregulatory loop controlling orphannuclear receptor DAX-1 gene expression by orphan nuclear receptor ERRgamma.Nucleic Acids Res 2005;33:6756–676830. Ruan HB, Han X, Li MD, et al. O-GlcNAc transferase/host cell factor C1complex regulates gluconeogenesis by modulating PGC-1a stability. Cell Metab2012;16:226–23731. Guinez C, Filhoulaud G, Rayah-Benhamed F, et al. O-GlcNAcylation in-creases ChREBP protein content and transcriptional activity in the liver. Diabetes2011;60:1399–141332. Yang WH, Kim JE, Nam HW, et al. Modification of p53 with O-linkedN-acetylglucosamine regulates p53 activity and stability. Nat Cell Biol 2006;8:1074–108333. Park SY, Kim HS, Kim NH, et al. Snail1 is stabilized by O-GlcNAc modifi-cation in hyperglycaemic condition. EMBO J 2010;29:3787–379634. Guinez C, Morelle W, Michalski JC, Lefebvre T. O-GlcNAc glycosylation: asignal for the nuclear transport of cytosolic proteins? Int J Biochem Cell Biol2005;37:765–77435. Konrad RJ, Janowski KM, Kudlow JE. Glucose and streptozotocin stimulatep135 O-glycosylation in pancreatic islets. Biochem Biophys Res Commun 2000;267:26–3236. Ji S, Park SY, Roth J, Kim HS, Cho JW. O-GlcNAc modification of PPARg reducesits transcriptional activity. Biochem Biophys Res Commun 2012;417:1158–116337. Guinez C, Mir AM, Dehennaut V, et al. Protein ubiquitination is modulated byO-GlcNAc glycosylation. FASEB J 2008;22:2901–291138. Chang Q, Su K, Baker JR, Yang X, Paterson AJ, Kudlow JE. Phosphorylationof human glutamine:fructose-6-phosphate amidotransferase by cAMP-dependentprotein kinase at serine 205 blocks the enzyme activity. J Biol Chem 2000;275:21981–2198739. Hu Y, Riesland L, Paterson AJ, Kudlow JE. Phosphorylation of mouseglutamine-fructose-6-phosphate amidotransferase 2 (GFAT2) by cAMP-dependentprotein kinase increases the enzyme activity. J Biol Chem 2004;279:29988–2999340. Gewinner C, Hart G, Zachara N, Cole R, Beisenherz-Huss C, Groner B. Thecoactivator of transcription CREB-binding protein interacts preferentially with theglycosylated form of Stat5. J Biol Chem 2004;279:3563–357241. Hart GW, Sakabe K. Fine-tuning ER-beta structure with PTMs. Chem Biol2006;13:923–92442. Yang X, Downes M, Yu RT, et al. Nuclear receptor expression links thecircadian clock to metabolism. Cell 2006;126:801–81043. Walgren JL, Vincent TS, Schey KL, Buse MG. High glucose and insulinpromote O-GlcNAc modification of proteins, including alpha-tubulin. Am J PhysiolEndocrinol Metab 2003;284:E424–E43444. Hanover JA, Krause MW, Love DC. The hexosamine signaling pathway:O-GlcNAc cycling in feast or famine. Biochim Biophys Acta 2010;1800:80–9545. Cheung WD, Hart GW. AMP-activated protein kinase and p38 MAPK activateO-GlcNAcylation of neuronal proteins during glucose deprivation. J Biol Chem2008;283:13009–13020

diabetes.diabetesjournals.org Misra and Associates 2847

46. Caro JF, Triester S, Patel VK, Tapscott EB, Frazier NL, Dohm GL. Liverglucokinase: decreased activity in patients with type II diabetes. Horm Metab Res1995;27:19–2247. Torres TP, Catlin RL, Chan R, et al. Restoration of hepatic glucokinaseexpression corrects hepatic glucose flux and normalizes plasma glucose inzucker diabetic fatty rats. Diabetes 2009;58:78–8648. Osbak KK, Colclough K, Saint-Martin C, et al. Update on mutations inglucokinase (GCK), which cause maturity-onset diabetes of the young, permanentneonatal diabetes, and hyperinsulinemic hypoglycemia. Hum Mutat 2009;30:1512–152649. Consoli A. Role of liver in pathophysiology of NIDDM. Diabetes Care 1992;15:430–44150. Agius L. Targeting hepatic glucokinase in type 2 diabetes: weighing thebenefits and risks. Diabetes 2009;58:18–2051. Zhang K, Yin R, Yang X. O-GlcNAc: A Bittersweet Switch in Liver. FrontEndocrinol (Lausanne) 2014;5:22152. Konrad RJ, Zhang F, Hale JE, Knierman MD, Becker GW, Kudlow JE. Alloxanis an inhibitor of the enzyme O-linked N-acetylglucosamine transferase. BiochemBiophys Res Commun 2002;293:207–212

53. Liu J, Yu Y, Fan YZ, et al. Cardiovascular effects of endomorphins in alloxan-induced diabetic rats. Peptides 2005;26:607–61454. Xu C, Liu G, Liu X, Wang F. O-GlcNAcylation under hypoxic conditions andits effects on the blood-retinal barrier in diabetic retinopathy. Int J Mol Med 2014;33:624–63255. Zhang H, Gao G, Brunk UT. Extracellular reduction of alloxan results inoxygen radical-mediated attack on plasma and lysosomal membranes. APMIS1992;100:317–32556. Donovan K, Alekseev O, Qi X, Cho W, Azizkhan-Clifford J. O-GlcNAc modifi-cation of transcription factor Sp1 mediates hyperglycemia-induced VEGF-Aupregulation in retinal cells. Invest Ophthalmol Vis Sci 2014;55:7862–787357. Perez-Cervera Y, Dehennaut V, Aquino Gil M, et al. Insulin signaling controlsthe expression of O-GlcNAc transferase and its interaction with lipid micro-domains. FASEB J 2013;27:3478–348658. Ma Z, Vocadlo DJ, Vosseller K. Hyper-O-GlcNAcylation is anti-apoptotic andmaintains constitutive NF-kB activity in pancreatic cancer cells. J Biol Chem2013;288:15121–1513059. Ortiz-Meoz RF, Jiang J, Lazarus MB, et al. A small molecule that inhibitsOGT activity in cells. ACS Chem Biol 2015;10:1392–1397

2848 O-GlcNAcylation of ERRg Diabetes Volume 65, October 2016