Chronic Ethanol Consumption Inhibits Glucokinase Transcriptional Activity by Atf3 and Triggers Metabolic Syndrome in Vivo

Background: Chronic ethanol consumption induces pancreatic β-cell dysfunction and metabolic syndrome. Results: Ethanol-induced Atf3 inhibits glucokinase transcriptional activity through direct binding or Atf3/Pdx-1/Hdac1 axis on glucokinase promoter. Conclusion: ATf3 fosters β-cell dysfunction via Gck down-regulation and triggers T2D, which is ameliorated by in vivo Atf3 silencing. Significance: The presented data uncover a new role for Atf3 as a potential therapeutic target in treating type 2 diabetes.

Chronic ethanol consumption induces pancreatic ␤-cell dysfunction through glucokinase (Gck) nitration and downregulation, leading to impaired glucose tolerance and insulin resistance, but the underlying mechanism remains largely unknown. Here, we demonstrate that Gck gene expression and promoter activity in pancreatic ␤-cells were suppressed by chronic ethanol exposure in vivo and in vitro, whereas expression of activating transcription factor 3 (Atf3) and its binding to the putative Atf/Creb site (from ؊287 to ؊158 bp) on the Gck promoter were up-regulated. Furthermore, in vitro ethanol-induced Atf3 inhibited the positive effect of Pdx-1 on Gck transcriptional regulation, enhanced recruitment of Hdac1/2 and histone H3 deacetylation, and subsequently augmented the interaction of Hdac1/Pdx-1 on the Gck promoter, which were diminished by Atf3 siRNA. In vivo Atf3-silencing reversed ethanol-mediated Gck down-regulation and ␤-cell dysfunction, followed by the amelioration of impaired glucose tolerance and insulin resistance. Together, we identified that ethanol-induced Atf3 fosters ␤-cell dysfunction via Gck down-regulation and that its loss ameliorates metabolic syndrome and could be a potential therapeutic target in treating type 2 diabetes. The Atf3 gene is associated with the induction of type 2 diabetes and alcohol consumption-induced metabolic impairment and thus may be the major negative regulator for glucose homeostasis.
Chronic ethanol consumption is well established as an independent risk factor for type 2 diabetes (T2D) 2 and has been suggested to cause increased adiposity, impaired glucose homeostasis, and insulin resistance (1,2). Several studies have demonstrated that light-to-moderate ethanol consumption is associated with a lower risk of T2D through increased glucosestimulated insulin secretion and insulin sensitivity in peripheral tissues (3,4). However, it is generally believed that the administration of moderate amounts of ethanol to rodents with metabolic disorders may increase the risk of complications in metabolic diseases or diabetes (5,6). Based on several previous studies, pancreatic ␤-cells may be sensitive to ethanol-induced oxidative stress associated with the production of reactive oxygen species, which is one of the earliest events in glucose intolerance and is associated with mitochondrial dysfunction (7,8). However, the mechanism by which chronic ethanol consumption induces pancreatic ␤-cell dysfunction and how such induced ␤-cell dysfunction contributes to the progression of T2D remain largely unknown.
Glucokinase plays a critical role in blood glucose homeostasis, serving as a glucose sensor for glucose-stimulated insulin secretion in pancreatic ␤-cells and as the major regulator of glucose uptake in hepatocytes (9,10). Recently, we suggested that ethanol consumption may induce pancreatic ␤-cell dysfunction and apoptosis through Gck nitration and down-regulation, correlated with impaired glucose responsiveness and insulin resistance (11). Although we demonstrated that ethanol-induced Atf3 in pancreatic ␤-cells may function as an upstream regulator of Gck down-regulation, little is known about the exact role and regulatory mechanism of Atf3 in ethanol-induced Gck down-regulation.
Atf3, a member of the Atf/Creb family of transcription factors, regulates gene expression by binding to the consensus Atf/ Creb cis-regulatory element via a basic leucine zipper domain (12). Given its frequent induction by various cellular stressors, ectopic expression of Atf3 in heart, liver, and pancreatic ␤-cells causes cardiac enlargement, liver or pancreatic ␤-cell dysfunction and apoptosis, impaired glucose metabolism, and diabetes (13). Although pancreatic duodenal homeobox-1 (Pdx-1) and sterol regulatory element-binding protein-1c directly bind to specific elements of the pancreatic and liver Gck promoter, respectively, and are positive regulators for Gck gene expression (14,15), the relevant upstream activator or repressor regulators involved in Gck transcriptional regulation are little known. We previously suggested that lipotoxicity-induced Atf3 may be associated with the inhibition of Pdx-1-mediated transcriptional activity (16), but the precise action mechanisms of Atf3 are still not clear. Generally, transcription is regulated by various complex processes that require cooperation between transcription factors and co-activators or co-repressors that modulate histone structure (17). Histone modification via acetylation, phosphorylation, and methylation has been implicated in increased or decreased accessibility to transcription machinery, thereby leading to the repression or activation of gene expression (18). The ␤-cell-specific transcription factor Pdx-1 has been shown to interact with the histone acetyltransferase p300/Cbp, and this interaction has been demonstrated to be important for Gck-mediated insulin and Glut2 expression (19). We also demonstrated that endoplasmic reticulum stressinduced Atf3 inhibits Pdx-1-mediated PBE/Luc activity or Gck/ Luc activity by inhibiting the interaction between Pdx-1 and p300, which leads to ␤-cell dysfunction (20). These results prompted us to investigate whether ethanol-induced Atf3 may also act as a transcriptional repressor of Gck gene expression via histone modification, leading to pancreatic ␤-cell dysfunction and apoptosis. This study provides molecular insight into the mechanism by which chronic ethanol-induced Atf3 inhibits the transcriptional activity of Gck through direct binding to the consensus Atf/Creb-binding site and the formation of an Atf3/ Pdx-1/Hdac1/2 axis at the Gck promoter with the deacetylation of histone H3. Clear evidence for the amelioration of these events by Atf3 silencing using in vivo-jetPEI siRNA delivery system may offer a therapeutic strategy to combat the impaired glucose metabolism, insulin insensitivity, and hyperglycemia associated with chronic ethanol consumption.

EXPERIMENTAL PROCEDURES
Cell Lines and Isolated Islet Cells-Rat INS-1 pancreatic ␤-cells were cultured in RPMI 1640 medium containing 11.1 mM glucose supplemented with 10% fetal bovine serum, 2 mM glutamine, 1 mM sodium pyruvate, 55 M ␤-mercaptoethanol, 10 mM HEPES, 100 IU/ml penicillin, and 100 g/ml streptomycin (Invitrogen). All experiments were performed between passage P4 and P20. Islet cells were isolated from overnight-fasted C57BL/6 mice using a previously described collagenase digestion technique (11). All antibodies were obtained from Cell Signaling Technology (Beverly, MA) or Santa Cruz Biotechnology Inc. (Santa Cruz, CA), and chemicals were purchased from Sigma.
In Vitro Treatment of Ethanol-To determine the concentrations and time points of ethanol showing positive effects on cell physiology (insulin content and ATP production) and the changes of phenotypes (apoptosis), we have treated the INS-1 cells or isolated islet cells with ethanol at the different concentrations (0, 25, 50, 100, and 200 mM) and time points (0, 6, 12, 24, and 48 h). Thereafter, the effective concentration (100 mM) and time (24 h) of ethanol, which show the most effective effects on the decrease of insulin content and ATP production and the increase of apoptosis, were selected. The treatment of ethanol with less than 50 mM or more than 100 mM did not have effective effects. Although there are still conflicting claims in the relative calculation between millimolar and in vivo percentage (%) of blood alcohol levels (22), several previous studies have shown that the selected 100 mM ethanol actually corresponds to about 0.46% (23), which can yield signs of intoxication in in vivo organs. The selected concentration (100 mM) and time (24 h) of ethanol is currently accepted and considered as an acute ethanol consumption in an in vitro model (24,25). When cells were treated with 100 mM ethanol, the final media contained the volume of treated ethanol. However, when cells were treated with ethanol, alcohol exposure of cells may be hampered by evaporation of the alcohol. The fluctuation of alcohol concentration and ethanol effects on the cells was due to evaporation. To avoid this, investigators used settings where ethanol was added into the culture media and the cell culture plates were maintained for the entire duration of stimulation in a microclimate chamber at 37°C with a gas mixture and an alcohol atmosphere (26).
Animals-C57BL/6J male mice (6 weeks old) originally purchased from The Jackson Laboratory (Bar Harbor, ME) were used in all experiments. Individually caged mice were placed on a Lieber-Decarli regular liquid diet (Dyets; control diet number 710027 or ethanol diet number 710260). Mice were pair-fed with the control versus 5% (v/v) ethanol diet for 8 weeks as reported previously (11). All animal experiments were conducted in accordance with guidelines from the Korean National Institute of Health Animal Facility.
Luciferase Activity Analysis-After transfection or treatment, the cells were lysed, and luciferase activity was measured using the Luciferase Assay System (Promega, Madison, WI).
RNA Interference and Transient Transfection-The human (sc-29757) or rat Atf3-(sc-72029) and mouse (sc-38761) or rat Pdx-1 (sc-10840) siRNAs were purchased from Santa Cruz Biotechnology. The cells were plated at 60 -70% confluence and transfected with siRNA complexes or only with transfection reagents using Lipofectamine 2000 (Invitrogen) following the protocol recommended by the manufacturer.
Electrophoretic Mobility Shift Assay (EMSA)-DNA mobility shift assays were performed as described previously (27). In brief, EMSA was performed in 20-l volumes with 20 mM Tris-HCl, pH 7.9, 1.5% glycerol, 50 g/ml BSA, 1 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, 2 g of poly(dI-dC), 1 ng of 32 Plabeled probe (5Ј-TGCAGTGACAGGGTGACAGA-3Ј), and 10 g of nuclear extract. Reactions were incubated at 25°C for 20 min and subsequently analyzed by electrophoresis through nondenaturing stock polyacrylamide gels (6 or 10%) in 0.5ϫ TBE buffer containing 44.5 mM Tris-HCl, pH 8.2, 44.5 mM boric acid, and 1 mM EDTA. After the gel was pre-run at 100 V for 2 h, electrophoresis was performed at 270 V for 2 h at 4°C. The gels were exposed to x-ray films using two intensifying screens at Ϫ70°C.
Biochemical Analysis-Blood was collected by cardiac puncture from mice that had fasted overnight. Sera and relevant peripheral tissues were stored at Ϫ20°C and used subsequently to assess the levels of triglycerides. Triglycerides were measured by colorimetric assays. For glucose tolerance test and insulin tolerance test, the mice were fasted at 8:00 a.m. for 6 h and then injected intraperitoneally with 1g/kg glucose or 1.5 units/kg regular human insulin, respectively. Blood samples were collected from the tail vein at time 0 (before injection), 30, 60, 90, and 120 min after injection, and the blood glucose level was measured using a portable glucose meter (Glucocard II Arkray, Kyoto, Japan) (11). The levels of insulin, ATP, and nitrite were also determined as described previously (11). Intracellular ROS generation was measured by using dichlorodihydrofluorescein diacetate (Molecular Probes, Eugene, OR) (11).
Immunoblots and Coimmunoprecipitation-Cells were lysed in RIPA buffer at 4°C, vortexed, and then centrifuged at 16,000 rpm for 10 min at 4°C. The supernatant was mixed in Laemmli loading buffer, boiled for 4 min, and then subjected to SDS-PAGE. For endogenous complexes, mitochondrial lysates (1 mg) fractionated from the cells were immunoprecipitated with 2 g of antibody and immunoblotted.
Histopathology-For immunohistochemistry analysis, after pancreatic tissues were fixed, deparaffinized, and washed, sections were treated with diluted blocking serum for 20 min. Slides were incubated overnight at 4°C in a humidified chamber with specific antibodies diluted in blocking serum. For immunocytochemistry analysis, the INS-1 cells were grown on glass coverslips for 24 h in a 6-well plate. After treatment, the cells were washed with cold PBS and fixed in paraformaldehyde (4% in PBS) for 15 min. All procedures were performed as described previously (11).
Chromatin Immunoprecipitation (ChIP)-The ChIP assays were performed according to the manufacturer's protocol (Upstate Biotechnology, Lake Placid, NY). In brief, after treatment, cells were cross-linked with 1% formaldehyde in medium for 15 min at 37°C and then washed with ice-cold PBS and resuspended in 200 l of SDS sample buffer containing protease inhibitor mixture. The suspension was sonicated 10 times for 30 s with a 1-min cooling period on ice between times and pre-cleared with 20 l of protein A-agarose beads blocked with sonicated salmon sperm DNA for 30 min at 4°C. The beads were removed, and the chromatin solution was immunoprecipitated overnight with anti-Atf3, Pdx-1, Stat5, Ac-K18 histone H3 (1766-1, Epitomics), Ac-K8 histone H4 (1796 -1, Epitomics), and polymerase II (2035-1, Epitomics) antibodies at 4°C, followed by incubation with protein A-agarose beads for an additional hour at 4°C. The immunocomplexes were eluted with 100 l of elution buffer (1% SDS and 0.1 mol/liter NaHCO 3 ), and formaldehyde cross-links were reversed by heating at 65°C for 6 h. Proteinase K was added to the reaction mixtures and incubated at 45°C for 1 h. DNAs of the immunoprecipitates and control input DNAs were purified using a QIAquick PCR purification kit (Qiagen Inc.) and then analyzed by regular PCR using rat Gck promoter-specific primers. Alternatively, quantitative PCR was performed using the same primers in the presence of SYBR Green PCR Supermix (Bio-Rad) to detect the real time quantitative PCR products of chromatin immunoprecipitation samples according to the manufacturer's instructions (20,28). The primer sequences for ChIP and qChIP are as follows: rat Gck promoter (Ϫ283 to Ϫ135), forward 5Ј-CGTAGAGGGCTCTGCTCCTT-3Ј and reverse 5Ј-CAAAG-GCCTCTCTACTCCTGTC-3Ј; rat Gck promoter (Ϫ127 to Ϫ28), forward 5Ј-GTCCCAGTTTTCTGCATGGT-3Ј and reverse 5Ј-GAGTAGATGCCTCCCGTCAG-3Ј; and qChIP analysis of rat Gck promoter (Ϫ125 to ϩ15), forward 5Ј-CCC-AGTTTTCTGCATGGTGG-3Ј and reverse 5Ј-CGAGCTCA-GTCACAGTCGAT-3Ј.
In Vivo Atf3 Gene Silencing-Chemically synthetic siRNA against Atf3 was synthesized by Dharmacon RNA Technology. A commercially available cationic polymer transfection reagent (in vivo-jetPEI TM , PolyPlus-Transfection, Illkirch, France) was used to deliver siRNA via intravenous injection (29 -31). Briefly, 150 g of siRNA diluted in 200 l of 5% glucose solution was mixed with in vivo-jetPEI solution including transfection reagent jetPEI. The mixture was incubated for 15 min at room temperature to allow the complexes to form. This mixture was then injected into the ethanol-fed mice for 5 weeks once to six times with a 3-day interval between injections and continuously exposed to ethanol by 8 weeks. The scrambled siRNA solution mixed with in vivo-jetPEI solution served as a control siRNA. Three-day interval injection of siRNA was found to be optimum to sustain the effects of siRNA via preliminary experiments. The in vivo delivery efficiency of siRNA was assessed by performing RT-PCR and Western blotting in the removed tissues, lung, liver or pancreas, and isolated islet cells. The four siRNA candidates for mouse-specific Atf3 are as follows: Atf3 mouse-736, forward 5Ј-GCGGCGAGAAAGAAAUAATT-3Ј and reverse 5Ј-UAAAGA-GGUUCCUCUCGUCTT-3Ј; Atf3 mouse-587, forward 5Ј-GCA-GAAAGAGUCAGAGAAATT-3Ј and reverse 5Ј-UUUCUCUG-ACUCUUUCUGCTT-3Ј; Atf3 mouse-516, forward 5Ј-GCGGC-GAGAAAGA AAUAAATT-3Ј and reverse 5Ј-UUUAUUUCUU-UCUCGCCGCTT-3Ј; and Atf3 mouse-356, forward 5Ј-CACCC-UUUGUCAAGGAAGATT-3Ј and reverse 5Ј-UCUUCCUUG-ACAAAGGGUGTT-3Ј.
Statistical Analysis-For comparing values obtained in three or more groups, one-factor analysis of variance was used, followed by Turkey's post hoc test, and p Ͻ 0.05 was taken to imply statistical significance.

Chronic Ethanol Consumption Represses Gck Transcriptional
Activity in Pancreatic ␤-Cells-In agreement with previous reports (11), Gck expression was reduced in the liver and pancreas of ethanol-fed mice and in ethanol-exposed pancreatic ␤-cells, and this reduction was correlated with increased hepatic steatosis and reduced islet cell mass (Fig. 1, A-C). Similarly, mRNA levels ( Fig. 1, D-F) of Gck and Glut2, a glucose transporter, and the Luc activity (Fig. 1, G and H) of rat pancreatic Gck promoter pRGP-1003/Luc were significantly reduced in islet cells or INS-1 cells exposed to 100 mM ethanol, which was ameliorated by 4-methylpyrazole, an inhibitor of cytochrome P4502E1 (32), or peroxynitrite scavengers such as L-NMMA, uric acid, and deferoxamine (11). These results suggest that ethanol metabolism and peroxynitrites play a critical role in ethanol-mediated reduction of Gck transcriptional activity.
To determine the region of Gck promoter crucial for ethanolmediated Gck down-regulation, we overexpressed rat Gck promoter constructs with 5Ј-serial deletions and analyzed their responsiveness to ethanol treatment in INS-1 cells ( Fig. 2A). Ethanol treatment significantly decreased the Luc activity of the Gck promoter between Ϫ1003 and Ϫ287 bp, although we did not detect significant reductions between Ϫ158 and Ϫ84 bp, suggesting that the consensus binding site for a repressive transcription factor induced by ethanol was located between Ϫ287 and Ϫ158 bp. These results were confirmed using the pRGP-287/Luc and pRGP-158/Luc constructs (Fig. 2B). Also, we know that de novo protein synthesis as a repressive transcriptional factor is required for ethanol-repressed Gck transcriptional activity using cycloheximide, a de novo protein synthesis inhibitor (Fig. 2, C and D).
Ethanol-enhanced Atf3 Plays a Critical Role in the Downregulation of Gck Transcriptional Activity-To understand the molecular mechanisms by which ethanol inhibits Gck promoter activity, we investigated how Gck promoter activity is regulated by searching a 129-bp consensus sequence between Ϫ287 and Ϫ158 bp of the 5Ј-flanking region of the Gck promoter using the TRANSFAC transcription factor binding database. Illustrated in Fig. 3A, five potential binding sites for Atf/Creb that contained the TAAC, TGAA, TGAG, or TGAC element, in addition to two binding sites for Ap-1 (GTCA), were found in the Gck promoter region (Fig. 3A). Because stress-inducible Atf3 is still considered as a dual-face transcription factor that activates or represses a gene expression (13), we first determined the role of Atf3 in repressing Gck transcriptional activity. Gck expression and the pRGP-1003/Luc or pRGP-287/Luc promoter activities were decreased by Atf3 overexpression but not by Atf3 siRNA or C-terminal domain-deleted ATF3(⌬C) (Fig. 3,  B-F). So, we wondered whether ethanol exposure affects Atf3 gene expression in vivo and in vitro (Fig. 3, G and H). Atf3 was strongly increased in pancreatic tissue and islet cells of ethanol-fed mice or ethanol-treated INS-1 cells, indicating that de novo protein synthesis of Atf3 protein by eth-  SEPTEMBER 26, 2014 • VOLUME 289 • NUMBER 39 anol may act as a negative transcription regulator to reduce Gck gene expression.

Atf3/Pdx-1/Hdac1 Axis Prompts Metabolic Syndrome
Atf3 Directly Binds to the Putative Atf/Creb Site on the Gck Promoter- Fig. 4A shows that ATF3 overexpression repressed Gck transcriptional activity through a direct interaction with the consensus binding sites of Atf/Creb located between Ϫ287 and Ϫ158 bp of the Gck promoter. The specific role of Atf3 within the proximal Ϫ287-bp region of the Gck promoter was confirmed by cotransfection with ATF3 or inactive ATF3(⌬C) and pRGP-287/Luc or pRGP-158/Luc (Fig. 4B). The pRGP-287/Luc activity was significantly decreased by Atf3, but there was a small reduction by Atf3 in pRGP-158/Luc-transfected cells. A TRANSFAC search revealed that the pRGP-158/Luc promoter also has two potential binding sites for Atf/Creb (TGAC) and one for Ap-1 (GTCA), but mutations of their consensus elements were not responsive to Atf3 (data not shown). Therefore, to confirm the functional activity of putative binding sites for Atf3 in pRGP-287/Luc, plasmids containing the putative binding sites were reconstructed by point mutagenesis (Fig.  4C). As shown in Fig. 4D, only pRGP-287(mCreb6)/Luc, which contains point mutations at the neighboring sites for mCreb4 and mCreb5, did not respond to ATF3 overexpression or ethanol treatment, whereas other constructs containing potential Atf3-binding site mutations still responded to them, suggesting that Creb4 and Creb5, located between Ϫ185 and Ϫ174 bp of the Gck promoter, are essential for the inhibitory responsiveness of ethanol-induced Atf3. These are also confirmed by electrophoretic mobility shift assays (EMSAs) using a labeled oligonucleotide covering the Ϫ190 to Ϫ171-bp region (Creb4/5(W)) in cells overexpressing ATF3, ATF3(⌬C), or ATF3(⌬N) (Fig.  4E). Binding complexes were observed in nuclear extracts of ATF3-or ATF3(⌬N)-transfected cells but not in those of ATF3(⌬C)-transfected cells. Addition of an excess amount of the unlabeled competitor Creb4/5(W) (50-or 100-fold) completely blocked the binding of Atf3 to the Gck promoter. To increase the specificity of the Ϫ190 to Ϫ171-bp region, three 20-bp wild-type oligonucleotides (Ap-1(W), Creb1(W), and Creb2/3(W)) that cover the putative binding site containing the sequence of response elements for Ap-1 or Atf/Crebs, respectively, and the mutated oligonucleotide Creb4/5(M), originating from wild-type Creb4/5(W), were designed (Fig. 4G). A 100fold molar excess of unlabeled oligonucleotide for wild-type Creb4/5(W) (2nd lane) completely abolished the formation of the binding complex in nuclear extracts of ethanol-treated cells, whereas any oligonucleotides for mutated Creb4/5(M), including wild-type Ap-1(W), Creb1(W), and Creb2/3(W), did not compete (Fig. 4F). The specificity of Atf3 in the binding complexes was also confirmed in the results of a supershift assay using an anti-Atf3 antibody. The Atf3-DNA binding was much stronger in ethanol-treated INS-1 cells (no competitor lane) than those in nontreated cells (control). The recruitment of Atf3 to the Gck promoter was confirmed in vivo by chromatin immunoprecipitation. A 149-bp PCR product covering the putative Atf/Creb (from Ϫ283 to Ϫ135) of Gck promoter was detected in the nuclear extracts of ethanol-treated cells, but not in untreated cells, immunoprecipitated with the anti-Atf3 antibody (Fig. 4, H and I). There was no binding in chromatin immunoprecipitated with the anti-Stat5 antibody. As well, chromatin binding in anti-Atf3 immunoprecipitates was significantly increased in Atf3-transfected or ethanol-treated cells, which was strongly decreased by Atf3 siRNA (Fig. 4J), indicating that Atf3 is specifically recruited to Atf/Creb binding region of the Gck promoter to result in a decrease of its transcriptional activity in ethanol-treated cells.
Ethanol and Atf3 Counteract the Positive Effect of Pdx-1 on Gck Transcriptional Regulation-Although the effects of Pdx-1 on Gck expression are controversial (33), our data show the functional significance of the inter-regulation between Atf3 and Pdx-1 in Gck expression (Fig. 5A). The reduction of the Gck promoter activity by Atf3 or ethanol was potently increased by

Atf3/Pdx-1/Hdac1 Axis Prompts Metabolic Syndrome
Pdx-1 depletion (Fig. 5B), whereas the recovering effects of Pdx-1 were potentiated by Atf3 silencing or ATF3(⌬C) (Fig. 5, C  and D), suggesting that Pdx-1 may counteract the repressive effects of Atf3 or ethanol on Gck transcriptional activity. Two putative Pdx-1-binding sites, TAAT and TGAT from Ϫ105 to Ϫ94 bp, were 72 bp from the core Atf/Creb binding sequence in the pRGP-287/Luc reporter (Fig. 5E). To determine whether these sites include the binding region for Pdx-1, we generated constructs via the deletion (Ϫ105/Ϫ94 bp, ⌬Pdx-1) or point mutation (ATG3 GAT, mPdx-1) of putative binding sites for Pdx-1 in the pRGP-287/Luc-promoter. Pdx-1-enhanced pRGP-287/Luc activity was significantly decreased in pRGP-287(⌬Pdx-1)/Luc or pRGP-287(mPdx-1)/Luc-transfected cells, indicating that the conserved binding site of Pdx-1 is located between Ϫ105 and Ϫ94 bp of the Gck promoter. The counteracting roles of Atf3 and Pdx-1 on Gck transcriptional regulation were further confirmed by using the construct in which Atf3binding sites in Gck promoter are deleted (Ϫ190/Ϫ171, ⌬Atf; Fig. 5F). The reduction of its activity by Atf3 was potentiated by the deletion of the Pdx-1-binding site (⌬Pdx-1), which was rescued by the deletion of the Atf3-binding site, indicating that Atf3 and Pdx-1 may have opposite effects on Gck transcriptional activity through a direct interaction. Next, we hypothesized that the negative effect of ethanol on Pdx-1-mediated Gck  SEPTEMBER 26, 2014 • VOLUME 289 • NUMBER 39 transcriptional activation may be caused via the up-regulation of Atf3, which prevents Pdx-1 recruitment to its putative binding site in the Gck promoter. To this, cells exposed to ethanol in the presence of Atf3 siRNA were assayed by ChIP analysis using primer sequences covering the Pdx-1-binding site (from Ϫ127 to Ϫ28). The binding of Pdx-1 to the putative binding region of the Gck promoter was reduced in ethanol-treated cells but not in Atf3 siRNA (Fig. 5G) or ATF3(⌬C)-transfected cells (data not shown), suggesting that induced Atf3 may indirectly suppress Gck transcriptional activity by preventing Pdx-1 recruitment to its putative binding site.

Ethanol-induced Atf3 Interacts with Pdx-1 and Subsequently
Promotes the Recruitment of Hdac1/2 and Histone H3 Deacetylation on Gck Promoter-Next, to examine whether Atf3 may inhibit Pdx-1-mediated Gck transcription by decreasing histone acetylation on the Gck promoter, we have quantified the amount of Gck gene promoter associated with acetylated histone H3 or H4. To this, ChIP analysis was performed by using the primers covering Pdx-1-binding sites from Ϫ127 to Ϫ28 bp. As expected, histone H3 acetylation on the Gck promoter was lower in ethanol-treated (Fig. 6A) or ATF3-overexpressed cells (Fig. 6B), correlated with reduction in p300 binding, whereas

Atf3/Pdx-1/Hdac1 Axis Prompts Metabolic Syndrome
the recruitment of histone deacetylase, Hdac1 and -2, to the Pdx-1 binding region of Gck promoter was increased in ethanol-treated cells, which were markedly attenuated and reversely rescued by Atf3 depletion or trichostatin A (TSA), an inhibitor of histone deacetylase (27). Concomitantly, Gck mRNA was significantly decreased by ethanol or ATF3 overexpression, which was restored by TSA or Atf3 siRNA (Fig. 6C). Also, quantitative PCR using ChIP samples revealed that ethanol-induced Atf3 directly binds to the Gck promoter despite no Atf3-binding sites, whereas recruitment of Pdx-1, Ac-H3, and  -1 siRNA (B), Atf3 siRNA (C), or ATF3(⌬C) (D), the luciferase activity was measured (*, p Ͻ 0.01; **, p Ͻ 0.05). E, identification of Pdx-1-binding sites on pRGP-287/Luc promoter. Deletion-or point-mutated constructs of putative Pdx-1-binding sites within Ϫ105 and Ϫ94 region of pRGP-287/Luc vector were generated and transfected into the cells (*, p Ͻ 0.01; **, p Ͻ 0.05). F, ATF3 cDNA was cotransfected into the cells together with pRGP-287(⌬Pdx-1)/Luc or pRGP-287(⌬Atf)/Luc vector and measured the luciferase activity (*, p Ͻ 0.01; **, p Ͻ 0.05). G, ChIP assay for Pdx-1 binding to the Gck promoter. The immunoprecipitated (IP) DNAs were used as templates for PCR with primers covering Pdx-1-binding sites from Ϫ127 to Ϫ28 bp. The results represent the average Ϯ S.E. from three independent experiments. Ab, antibody; CTL, control. SEPTEMBER 26, 2014 • VOLUME 289 • NUMBER 39 RNA polymerase II to the promoter was significantly reduced in ethanol-treated cells, which were reversely attenuated by Atf3 depletion (Fig. 6D), demonstrating the specificity of their recruitment by Atf3. Additionally, the inhibition of histone deacetylase activity by TSA may partially restore the recruitment of Pdx-1, Ac-H3, and polymerase II to the Gck promoter after ethanol treatment, although there was no impact on Atf3 recruitment (Fig. 6E), indicating that Atf3 may be an upstream regulator of Hdac1/2 activation.

Atf3/Pdx-1/Hdac1 Axis Prompts Metabolic Syndrome
Atf3 Attenuates Pdx-1-dependent Gck Transcriptional Activity by Enhancing the Physical Interaction with Hdac1-We next investigated whether ethanol-induced Atf3 could inhibit Pdx-1-dependent Gck transcriptional activity by enhancing the activity of histone deacetylase (Fig. 7A). The increase of pRGP-158/Luc activity by Pdx-1 was significantly decreased by ethanol treatment or ATF3 overexpression, which was restored by TSA or Atf3 siRNA, indicating that the activated histone deacetylase may play an essential role in the inhibitory effects of Atf3 on Pdx-1-mediated Gck transcriptional activity. Also, ethanol-induced Atf3 inhibits Pdx-1-mediated Gck transcriptional activity by triggering a physical in vitro and in vivo interaction of Pdx-1 with Hdac1 rather than p300 (Fig. 7, B and C). Immunoprecipitated Pdx-1 directly interacted with endogenous Atf3 induced by ethanol or overexpressed GFP-ATF3 but not with GFP-ATF3(⌬C). Conversely, the interaction of p300 with Flag-Pdx-1 was significantly decreased by ethanol treatment or ATF3 overexpression. Additionally, Gck down-regulation in ethanol-treated ␤-cells was correlated with the induction of Atf3 and the reduction of nuclear localization of Pdx-1, which were reversely rescued by Atf3 siRNA (Fig. 7D).
In Vivo Atf3 Silencing Ameliorates Metabolic Syndrome and ␤-Cell Damage in Ethanol-fed Mice-To confirm the specific upstream role of Atf3 on ethanol consumption-induced Gck down-regulation and metabolic syndrome, a loss-of-function study was performed through in vivo Atf3 siRNA injection. Among four siRNA candidates, we selected 356-Atf3 siRNA, which had the highest efficiency for Atf3 silencing in various tissues (Fig. 8, A and B). As expected, the impaired glucose tolerance (GTT and OGTT; Fig. 8, C-E) and insulin tolerance (ITT; Fig. 8F) induced by ethanol consumption were markedly inhibited by the administration of Atf3 siRNA. Triglyceride accumulation and reduced plasma insulin levels in ethanol-fed mice were also attenuated by Atf3 knockdown (Fig. 8, G and H), correlated with restoring the reduced ATP (Fig. 8I). Similarly, the reduction in glucose-stimulated insulin secretion (GSIS; Fig. 8J) and the production of nitrite or ROS (Fig. 8K) in islet cells of ethanol-fed mice also depended on the presence of Atf3. Together with the amelioration of pancreatic ␤-cell dysfunction, in vivo Atf3 silencing inhibited ethanol-induced ␤-cell apoptosis, which was associated with the restoration of Gck, insulin, and Pdx-1 expression (Fig. 8, L and M). These results suggest that the Atf3 gene is associated with the induction of T2D and alcohol consumption-induced metabolic impairment and thus may play a major negative regulator for glucose homeostasis.

DISCUSSION
In this study, we provide evidence that ethanol-induced Atf3 suppresses Gck gene expression through direct interaction with the Atf/Creb-binding site (Ϫ185/Ϫ174) of the Gck promoter. Additionally, Atf3 directly interacts with Pdx-1 and specifically recruits Hdac1/2 to the Gck promoter in ethanol-treated ␤-cells, thereby repressing functional Pdx-1 on Gck transcriptional regulation. In vivo Atf3 silencing ameliorates ethanolmediated metabolic syndrome and pancreatic ␤-cell dysfunction. All of these findings are summarized in Fig. 9.
This study demonstrates a functional role of Atf3 as a major upstream regulator that negatively regulates Gck transcriptional activity in ethanol-fed mice, based on the following observations. (i) The reduction of Gck transcriptional activity by ethanol was potentiated by Atf3, which was inhibited by Atf3 depletion. (ii) Two putative Atf3-binding sites that are essential for the inhibitory responsiveness by Atf3 were identified between Ϫ185 and Ϫ174 bp of the Gck promoter. (iii) Atf3 also binds to the Pdx-1-binding sites between Ϫ102 and Ϫ92 of the Gck promoter, resulting in chromatin remodeling via histone H3 deacetylation.
Several studies have suggested that Pdx-1 is a ␤-cell master gene that regulates the expression of ␤-cell-specific genes such as insulin, Glut2, and Gck (34 -36). Although we previously suggested that activated Atf3 directly interacted with Pdx-1 (20), the correlation of Pdx-1 function with Gck expression in the transcriptional hierarchy is unclear, and the upstream regulator that serves as the activator or repressor for Pdx-1-mediated gene expression is also unknown. This study suggested the opposing effects of Atf3 on Pdx-1-mediated Gck transcriptional regulation, which were caused by their direct interaction in vitro and in vivo (Figs. 5-7). Pdx-1 overexpression reversely attenuated or rescued the inhibitory effects of Atf3 on Gck expression, but how Pdx-1 counteracts Atf3-repressed Gck transcriptional activity in response to ethanol is still unclear. Our data show that two putative Pdx-1-binding sites may be essential for counteracting the inhibitory effects of Atf3 on Gck transcriptional activity by using a pRGP-287/Luc with a deleted or point-mutated Pdx-1-binding site (Fig. 5D). Despite the lack of an Atf3-binding site, Atf3 still induced a small reduction in pRGP-287(⌬Atf)/Luc activity, which was similar to the small  SEPTEMBER 26, 2014 • VOLUME 289 • NUMBER 39 reduction in the pRGP-158/Luc promoter (Fig. 4), suggesting that Atf3 may also inhibit Pdx-1 functional activity within the Ϫ158-bp region. The association of Pdx-1 with insulin pro-moter is enhanced by its interaction with p300, and the initial acetylation of histone H4 may stabilize Pdx-1 binding (37). However, when glucose levels are low, Pdx-1 interacts with

Atf3/Pdx-1/Hdac1 Axis Prompts Metabolic Syndrome
Hdac1/2 rather than p300, and the complexes are recruited to the insulin promoter, thus leading to histone H4 deacetylation and insulin gene down-regulation (38). Similarly, ethanol-induced Atf3 may first interact with Pdx-1 and subsequently prevent the formation of Pdx-1/p300, followed by chromatin remodeling via alternative recruitment of histone deacetylase to the consensus Pdx-1-binding sites of the Gck promoter. Furthermore, because the primers used in ChIP assay cover the putative Pdx-1-binding sites (Ϫ28/Ϫ127) on the pRGP-158/ Luc promoter, the product does not contain Atf3-binding sites and may present only low responsiveness for Atf3. Therefore, it is important to determine how Atf3 acts as a negative upstream regulator for Pdx-1 functional activity at the Ϫ158-bp proximal region of the Gck promoter. Quantitative ChIP assay revealed that ethanol-induced Atf3 binds to the proximal Ϫ158-bp region, whereas recruitments of Pdx-1, acetylated H3, and RNA polymerase II were significantly decreased in ethanol-treated cells, which were ameliorated by Atf3 depletion (Fig. 6D), demonstrating that Atf3 association with Gck promoter is a critical step for histone H3 deacetylation. Despite the decreased recruitment to the Gck promoter and expression of Pdx-1 in ethanol-treated cells, how the formation of Atf3/Pdx-1/ Hdac1/2 complexes is increased in response to ethanol is unknown. Previously, it was demonstrated that Pdx-1 phosphorylation may determine the interaction of Pdx-1 with p300/ Cbp or Hdac1/2 in the regulation of the insulin gene, which may depend on the concentration of glucose (35)(36)(37). Similarly, it is possible that ethanol-induced Atf3 directly interacts with Pdx-1, subsequently decreasing Pdx-1 phosphorylation and recruiting Hdac1/2 rather than p300 (Fig. 7); this possibility should be explored further.
We also confirmed the specific upstream role of Atf3 in chronic ethanol-induced Gck down-regulation and pancreatic ␤-cell dysfunction by performing a loss-of-function study using the in vivo Atf3 siRNA delivery system (Fig. 8). In vivo Atf3 silencing ameliorated the impaired glucose metabolism and pancreatic ␤-cell dysfunction in ethanol-treated cells, and therefore prevented the development of T2D by reducing RNS production. However, in our model, the injected Atf3 siRNA may act on several organs such as liver and lung and is not specific to pancreatic ␤-cells (Fig. 8, A and B). Therefore, to define the precise role of Atf3 in ethanol-mediated Gck downregulation and ␤-cell dysfunction, additional studies using pancreas-specific knock-out models are required. Nevertheless, we propose that in vivo Atf3 silencing may be sufficient to improve metabolic syndrome and pancreatic ␤-cell dysfunction by ameliorating ethanol-induced Gck down-regulation. In contrast with the ameliorating effects of Atf3-silencing mice in vivo, it will be possible that Atf3 knock-out mice may conversely potentiate ethanol-mediated Gck down-regulation and ␤-cell dysfunction. Because stress-inducible Atf3 is a dual-face transcription factor that activates or represses gene expression (13), it may initially develop a compensatory or adaptive response for the acute oxidative stress or endoplasmic reticulum stress. Although the data are not shown here, we have determined genetic susceptibility of Atf3 to T2D risk by performing conditional analyses using our previous GWAS data (KARE study) based on the Korean population (39,40). Furthermore, we have identified a noncoding variant in the Atf3 gene as a novel predisposing factor to T2D after controlling for alcohol consumption in T2D groups (data not shown). From these results, we propose that Atf3 may play an important role in ethanol-mediated metabolic alteration, and it may be strongly associated with T2D.
In this study, the direct interaction of Atf3 with the Atf/Crebbinding site of the Gck promoter may have been a major molecular mechanism by which ethanol reduced Gck expression and pancreatic ␤-cell dysfunction and apoptosis. Also, Atf3-dependent interaction of Pdx-1 with Hdac1/2 and subsequent inactivation of Pdx-1 and chromatin remodeling may represent an additional molecular mechanism by which ethanol-induced Atf3 decreases Gck down-regulation (Fig. 9). Taken together, these results show for the first time that Atf3 may act as an upstream negative regulator for Gck transcriptional activation via direct binding to the putative binding site and by indirect binding with Pdx-1/Hdac1/2 on the Gck promoter. Additionally, our findings suggest that strategies based on the inhibition of Atf3 may be of benefit in the treatment of ethanol-mediated glucose impairment and metabolic syndrome.