Identification of 5′ AMP-activated Kinase as a Target of Reactive Aldehydes during Chronic Ingestion of High Concentrations of Ethanol*

Background: Carbonylation of proteins contributes to increased hepatocellular damage during alcoholic liver disease. Results: In a murine model of alcoholic liver disease, AMPK is covalently modified by reactive aldehydes reducing activity. Conclusion: Inhibition of AMPK activity by reactive aldehydes contributes to increased steatosis in alcoholic liver disease. Significance: This is the first report of AMPK carbonylation and inhibition during conditions of increased oxidative stress. The production of reactive aldehydes including 4-hydroxy-2-nonenal (4-HNE) is a key component of the pathogenesis in a spectrum of chronic inflammatory hepatic diseases including alcoholic liver disease (ALD). One consequence of ALD is increased oxidative stress and altered β-oxidation in hepatocytes. A major regulator of β-oxidation is 5′ AMP protein kinase (AMPK). In an in vitro cellular model, we identified AMPK as a direct target of 4-HNE adduction resulting in inhibition of both H2O2 and 5-aminoimidazole-4-carboxyamide ribonucleoside (AICAR)-induced downstream signaling. By employing biotin hydrazide capture, it was confirmed that 4-HNE treatment of cells resulted in carbonylation of AMPKα/β, which was not observed in untreated cells. Using a murine model of alcoholic liver disease, treatment with high concentrations of ethanol resulted in an increase in phosphorylated as well as carbonylated AMPKα. Despite increased AMPK phosphorylation, there was no significant change in phosphorylation of acetyl CoA carboxylase. Mass spectrometry identified Michael addition adducts of 4-HNE on Cys130, Cys174, Cys227, and Cys304 on recombinant AMPKα and Cys225 on recombinant AMPKβ. Molecular modeling analysis of identified 4-HNE adducts on AMPKα suggest that inhibition of AMPK occurs by steric hindrance of the active site pocket and by inhibition of hydrogen peroxide induced oxidation. The observed inhibition of AMPK by 4-HNE provides a novel mechanism for altered β-oxidation in ALD, and these data demonstrate for the first time that AMPK is subject to regulation by reactive aldehydes in vivo.

Previous reports concerning the effects of ethanol on activation of the AMPK pathway in mice vary depending on the amount of ethanol and the duration of feeding. In some studies, AMPK phosphorylation is increased (20 -22), whereas in others AMPK phosphorylation is decreased (23)(24)(25)(26). The use of different types of dietary fats as well as different percentages of dietary fat in these studies may be responsible for the discrepancies. In one report, 40% saturated fat plus ethanol resulted in a 2-fold increase in AMPK␣ phosphorylation (24). Concurrently, 40% PUFA plus ethanol resulted in a slight decrease in AMPK␣ phosphorylation (24). Using an intragastric overfeeding model, ethanol resulted in an increase in Thr(P) 172 AMPK but no corresponding increase in ACC phosphorylation (21). We have previously reported that the addition of ethanol for 6 weeks in conjunction with 30% PUFA suppressed AMPK phosphorylation, whereas ETOH combined with 45% PUFA resulted in increased AMPK phosphorylation but no change in overall phosphorylation of ACC (20). Most recently, in C57BL6/J mice, chronic ETOH decreased AMPK phosphorylation but resulted in an increase in CPT1 mRNA and CPTII protein expression (27). Activity of CPT1, however, did not significantly change, suggesting no change in ␤-oxidation.
Herein, we describe the effects of increased lipid peroxidation/4-HNE on AMPK signaling in cell culture as well as in the liver of mice chronically fed ethanol for 7.5 weeks. We determine that in HepG2 cells, 4-HNE inhibits activation of AMPK by H 2 O 2 and direct modification of recombinant AMPK by 4-HNE inhibits its activity. This research is further translated into the identification of AMPK as a direct target of lipid peroxidation in the livers of chronic ethanol-fed mice. These results provide a novel mechanism for dysregulation of AMPK signaling under conditions of increased oxidative stress that occur during chronic ethanol administration.

EXPERIMENTAL PROCEDURES
Animal Model and Dietary Information-C57BL/6J male mice (The Jackson Laboratory, Bar Harbor, ME) 6 -8 weeks of age in groups of 12 were fed a modified Lieber-DeCarli diet (30% fat-derived calories (Bio-Serv, Frenchtown, NJ) consisting of isocaloric pair-fed control and ethanol-treated animals (28). Ethanol-derived caloric content was ramped from week 1 of 10.8%, with incremental increases weekly to 16.2, 21.5, 26.9, 29.2, 31.8, and 34.7% for the last 1.5 weeks of feeding (3,4). In the control animals, calories derived from ethanol were replaced isocalorically by carbohydrates in the form of maltodextrin. Fresh control and ETOH diet was provided at 7:00 a.m. daily. Food consumption was monitored daily, and body weights were measured once per week. Upon completion of the study, animals were anesthetized via intraperitoneal injection with sodium pentobarbital and euthanized by exsanguination. Blood was collected from the inferior vena cava, and plasma was separated via centrifugation at 4°C and assayed for alanine aminotransferase activity (Sekisui Diagnostics). Blood ethanol concentrations were determined by gas chromatography as previously described from samples obtained at 11:45 p.m. (29). Excised livers were weighed, and subcellular fractions were obtained via differential centrifugation as previously described (6). All procedures involving animals were approved by the Institutional Animal Care and Use Committee of the University of Colorado and were performed in accordance with published National Institutes of Health guidelines.
Treatment of HepG2 Cells-HepG2 cells were purchased from Invitrogen and maintained according to the manufacturer's instructions. Cells were maintained at 50 -80% confluence in RPMI supplemented with 10% fetal bovine serum, 100 mM Hepes, 100 IU/ml penicillin, and 100 g/ml streptomycin. Cells were plated into 6-well plates at a density of 1 ϫ 10 6 cells per well. The following day the cells were washed twice in serumfree RPMI and treated with the indicated concentrations of 4-HNE. For hydrogen peroxide treatments, HepG2 cells were treated with 0 -50 M 4-HNE for 60 min, washed once in serum-free media, and stimulated with 1.0 mM hydrogen peroxide for 5 min (Sigma). For 5-aminoimidazole-4-carboxyamide ribonucleoside (AICAR) experiments, cells were pretreated with 50 M 4-HNE followed by stimulation with 2 mM AICAR for 60 min.
Immunohistochemistry-Sections of freshly excised liver tissue were placed in 10% neutral-buffered formalin (Sigma) for 16 h followed by incubation in 70% ethanol overnight. Samples were then processed, embedded in paraffin, and mounted to slides by the University of Colorado C Denver Histology Core. Immunohistochemical characterization of carbonylation and macrophage infiltration was performed using rabbit polyclonal anti-4-HNE, anti-MDA, anti-acrolein (Abcam, Cambridge, MA), and rat anti-mouse polyclonal antibodies directed against F4/80 antigen (AbSerotec/Bio-Rad), respectively (30).
AMPK Activity Assays-The AMPK activity assay was purchased from MBL International (Woburn, MA) and utilized for AMPK activity. Briefly, 0.5 g of rAMPK was treated in triplicate with increasing molar ratios of 4-HNE (0 -25:1) in NH 4 HCO 3 for 30 min at 37°C. Protein was resuspended in kinase buffer and incubated in a 96-well plate for 30 min at 30°C. To each well 100 l of anti-phospho mouse IRS1 Ser 789 was added and incubated for an additional 30 min. Wells were washed, 100 l of HRP-conjugated anti-mouse IgG added and allowed to incubate for 30 min. Wells were washed, and 100 l of tetramethylbenzidine were added for 30 min followed by the addition of EDTA stop solution. Absorbance was then read at 450 nm.
Measurement of Reduced and Oxidized Glutathione-Glutathione measurements were performed using a modified method from Reed et al. (31). Briefly, fresh liver tissue was homogenized in 0.1 N perchloric acid using a sonicator and centrifuged at 13,000 rpm for 15 min, and the cleared supernatant was further neutralized with sodium bicarbonate and treated with 1:10 iodoacetic acid (14.9 mg/ml) in the dark for 45 min at room temperature. Samples were then derivatized (1:1) with Sanger's reagent (1.5% 1-fluoro-2,4-dinitrobenzen in absolute ethanol) overnight in the dark at room temperature. Derivatized samples were then stored at 4°C in the dark until analyzed by HPLC. Samples were applied (50-l injection) to a Supelco LC-NH 2 5 m, 25 cm ϫ 4.1 mm column (Sigma) separated with a gradient mobile phase system. Solvent A (80% methanol) and Solvent B (80% Solvent B, 20% acetic acid/ammonium acetate solution (756 ml of glacial acetic acid, 244 ml of water, and 308 g of ammonium acetate)) were used for the separation of derivatized reduced (GSH) and oxidized (GSSG) glutathione using a Shimadzu LC-10AD dual pump system (flow rate of 1.0 ml/min) coupled to a Shimadzu SPD-M10AV diode array detector (Kyoto, Japan) set to 350 nm. HPLC running conditions were as follows: isocratic period of 50% solvent B for 10 min followed by a linear gradient from 50% solvent B to 95% solvent B for 15 min for a total run time of 35 min. Typical elution times for GSH were 11.55 and 17.05 min for GSSG. Nanomolar concentrations of GSH and GSSG were calculated against a known standard curve of seven points for each molecule.
Biotin Hydrazide Modification of Alkylated AMPK-To evaluate reactive aldehyde adduction, 1 ϫ 10 7 HepG2 cells were treated with 50 M 4-HNE for 1 h. Cells were then lysed in 50 mM HEPES, 100 mM NaCl, 2 mM EDTA, 0.5% Triton-100, 2.5 mM biotin hydrazide for 2 h. To remove excess biotin, lysates were dialyzed overnight at 4°C. Biotinylated proteins were captured using streptavidin-agarose beads (Pierce) on a rotary mixer at 4°C for 3 h. Beads were washed 5 times in PBS, boiled in SDS loading buffer, and analyzed using SDS-PAGE/Western blotting for AMPK␣, -␤1, and -␤2. For murine studies, whole cell lysates were prepared from liver tissue isolated from mice fed ETOH for 7.5 weeks followed by biotin hydrazide treatment of homogenates containing 0.5 mg of total protein from pairfed and ETOH-fed mice.
Nile Red Staining of Neutral Lipids-HepaRG cells were plated in 6-well plates to Ͼ90% confluence and treated for 1 h with 5% oleic acid conjugated to BSA or with 50 M 4-HNE followed by a 24-h recovery in serum-containing RPMI 1640 media. Nile Red was used to detect cellular neutral lipid using a modification of previously reported techniques, and Hoechst 33342 was used to measure cellular DNA (32). Cells were collected in 1 ml of phenol red-free, serum-free RPMI 1640, and 50 l of cell suspension was incubated in sterile 96 well fluorescence plates for 1 h at 37°C with 200 l of serum-free, phenol red-free RPMI 1640 containing Nile Red at 20 g/ml and Hoechst 33342 at 10 g/ml. Plates were shaken for 10 s before reading, and neutral lipid was assessed by Nile Red fluorescence with excitation 485 nm/emission 565 nm and was normalized against cellular DNA as measured by Hoechst 33342 staining measured with excitation 355 nm/emission 465 nm.
Adduction and Trypsin Digest of Recombinant AMPK␣/␤/␥ by 4-HNE-10 g of rAMPK␣/␤/␥ was treated with 4-HNE at molar ratios of 0.1, 1:1, 2.5:1, 5:1,10:1, and 20:1 for 30 min at 37°C. 0.5 g were removed and loaded onto an SDS-PAGE gel and Western-blotted using anti-4-HNE polyclonal antibodies. To verify AMPK adduction, membranes were then stripped for 15 min using Restore TM stripping buffer (Pierce), washed twice in TBST, and blocked once again for 1 h in TBST, 5% nonfat dry milk. After blocking, membranes were incubated overnight with mouse anti-AMPK (Pierce) and subsequently processed. The rest of the sample was treated with 10 mM sodium borohydride in 100 mM NaOH for 30 min at 37°C. Samples were boiled in 5ϫ SDS loading buffer and subjected to SDS-PAGE. Gels were stained for 15 min with Coomassie Blue R250 and destained overnight in 10% acetic acid, 20% MEOH. From the gel, each band was picked and destained in 50 mM NH 4 HCO 3 for an additional 24 h. Gel slices were dehydrated with 100% acetonitrile for 15 min, reduced in 10 mM DTT, 50 mM NH 4 HCO 3 for 45 min at 60°C, and alkylated in 50 mM iodoacetic acid in 50 mM NH 4 HCO 3 for 45 min in the dark at room temperature. Samples were washed once in 50 mM NH 4 HCO 3 and digested in 0.3 g of trypsin 50 mM NH 4 HCO 3 overnight. Peptide extraction consisted of sonicating for 20 min in 50% acetonitrile, 50 mM NH 4 HCO 3 followed by sonication in 5% formic acid in acetonitrile for an additional 20 min. Samples were dried down and resuspended in 0.1% trifluoroacetic acid.
A nano-flow HPLC system (Agilent 1200, Palo Alto, CA) coupled with LTQ Orbitrap Velos Hybrid FT mass spectrometer and nanospray ion source (Thermo Fisher, San Jose, CA) was used for LC-MS/MS analysis. An 8-l volume of samples was run on a ZORBAX 300SB-C18 trap column (5-m inner diameter ϫ 5 mm, Agilent Technologies, Santa Clara, CA) to remove salts and contaminating analytes, and reverse phase separation of peptides was performed on a C18 column (100-m inner diameter ϫ 17-cm length) packed in-house using a 4-m 80 Å pore size matrix (Synergi, Phenomenex, Torrance, CA). Mobile phase A was composed of 99.9% HPLC grade water, 0.1% formic acid, and mobile phase B was composed of 99.9% acetonitrile, 0.1% formic acid. Peptides were eluted from the column using a linear gradient with mobile phase B increasing from 2-90% over 60 min. The positive ion mode was used for data acquisition of MS of ions m/z 300 -2000, and MS2 scans were performed for the most intense ions. All data were acquired using Thermo Xcalibur software (Version 2.1.0.1140, Thermo Fisher).
MS/MS Data Analysis-An in-house script was used to convert raw spectral data files into mascot generic file (mgf) format. The mgf files were searched against the human Swissprot database Mascot (Version 2.2, Matrix Science Inc., London, UK). The MS/MS tolerance was set at Ϯ0.6 Da for the searches, and the peptide tolerance was set at Ϯ15 ppm. The search allowed for the missed cleavage of one tryptic site as well as one carbon 13. The carbamidomethylation of cysteine was searched as a fixed modification. Variable modifications included acylation of the protein N terminus, pyroglutamic acid formation of the peptide N terminus, oxidation of methionine, 4-HNE modification of cysteine, histidine, and lysine, and reduced 4-HNE modification of cysteine, histidine, and lysine. Spectral images were obtained from the raw data files using Xcalibur, and Mascot was used to assist peak assignment. Only Mascot scores of Ͼ30 were considered significant.
In Silico Molecular Modeling-All simulations were performed using Discovery Studio software (Version 3.1; Accelrys Inc., San Diego, CA). The crystallographic coordinates of the 1.6 Å human-activated AMPK crystal structure (PDB code 2Y94) were obtained from the RCSB Protein Data Bank (33). The 4-HNE modification of the Cys 130 , Cys 174 , and Cys 227 residues on AMPK␣ and Cy 225 on AMPK␤ were typed with the CHARMm force field and subjected to minimization using the conjugate gradient method to a convergence of 0.001 kcal/mol using the Generalized Born implicit solvent model (34,35).
Statistical Analysis-Western blots were quantified using ImageJ (rsb.info.nih.gov) followed by statistical analysis via a paired Student's t test and Prism 4.0 for Windows (GraphPad Software, San Diego, CA). All data are expressed as the means Ϯ S.E., and p values Ͻ0.05 were considered significant.

AMPK Phosphorylation Is Decreased after 4-HNE Treatment in HepG2
Cells-In our previous report we identified 4-HNEdependent modification and inhibition of PTEN as a contributor to lipid accumulation in HepG2 cells (11). We further demonstrated that Akt2, which is the immediate downstream target of PTEN, was also inhibited by 4-HNE in the same cells. This suggests that lipid accumulation induced by 4-HNE exposure may be due to regulation of an alternative pathway. In the liver Akt2 is a primary regulator of fatty acid synthesis, whereas AMPK is a primary regulator of ␤-oxidation (17,36). To evaluate the hypothesis that 4-HNE alters phosphorylation of AMPK␣ and ACC, HepG2 cells were treated with 50 M 4-HNE using a time course of 2 h. Cells were lysed, and lysates were examined for increased phosphorylation of AMPK and its immediate downstream target of ACC via Western blotting. From Fig. 1, incubation of HepG2 cells 4-HNE resulted in a time-dependent decrease in Thr(P) 172 AMPK␣ corresponding to decreased Ser(P) 79 ACC.
Protein phosphatase 2A (PP2A) regulates AMPK activation by dephosphorylation of Thr(P) 172 AMPK (37). To determine if the mechanism of 4-HNE inhibition of AMPK phosphorylation is via increased PP2A activity, HepG2 cells were incubated with the PP2A inhibitor okadaic acid with or without 4-HNE. From Fig. 2, A--C, treatment with okadaic acid (OA) resulted in partial but not complete restoration of AMPK and ACC phosphorylation. This indicates that activation of PP2A contributes to 4-HNE-induced decreases in AMPK phosphorylation.

4-HNE Inhibits Hydrogen
Peroxide-induced Activation of AMPK in HepG2 Cells-We have previously shown that 4-HNE will inhibit activation of both Akt1 and Akt2 by H 2 O 2 in HepG2 cells (38). Reactive species such as H 2 O 2 are also known activators of AMPK (19,37,39). Treatment of rat hepatoma cells with 1 mM H 2 O 2 resulted in a 1.5-fold increase in AMPK phosphorylation (37). To gain a better understanding of the effects of 4-HNE on H 2 O 2 -induced AMPK, HepG2 cells were preincubated with increasing concentrations of 4-HNE followed by H 2 O 2 stimulation. As shown in Fig. 3, A-D, in cells treated with    4-HNE/H 2 O 2 , a concentration-dependent decrease in AMPK␣ phosphorylation was observed. This decrease corresponded to a decrease in phosphorylation of the downstream AMPK targets ACC and Raptor (39). Combined, these data demonstrate 4-HNE inhibits the induction of AMPK signaling by H 2 O 2 .
In addition to H 2 O 2 , AMPK can be activated by the adenosine analog AICAR resulting in increased phosphorylation of ACC (23). To determine if 4-HNE broadly inhibits AMPK activity, HepG2 cells were treated with 4-HNE and 2 mM AICAR, and phosphorylation of ACC was evaluated. From Fig.  4, A and B, 4-HNE significantly inhibits AICAR-induced phosphorylation of ACC in HepG2 cells, indicating that reactive aldehydes inhibit AMPK activation.
4-HNE Treatment Leads to Increased Carbonylation of AMPK in HepG2 Cells-Biotin hydrazide will react with reactive aldehyde groups on proteins forming a covalent biotin hydrazone bond. We have utilized biotin hydrazide modification followed by streptavidin capture to identify carbonylated proteins from cells as well as hepatic tissue (10 -12, 38, 40, 41).
To determine if carbonylation of cellular AMPK␣/␤ occurs after 4-HNE treatment, HepG2 cells were subjected to increasing concentrations of 4-HNE followed by biotin hydrazide modification followed by streptavidin capture and Western blotting. As shown in the Western blots depicted in Fig. 5A, 4-HNE treatment resulted in a concentration-dependent increase in carbonylation of AMPK␣/␤. The quantification of the Western blots demonstrates a 5-10-fold increase in carbonylation after the 25-50 M 4-HNE treatment (Fig. 5, B--D).
These data indicate that intracellular carbonylation of AMPK occurs as a consequence of 4-HNE exposure (8).
Effects of 4-HNE on Lipid Accumulation in HepaRG Cells-We previously demonstrated that the addition of 4-HNE results in increased lipid accumulation in HepG2 cells (11). HepG2 cells are a hepatocarcinoma cell line and may not best represent primary liver cells (42). HepaRG cells were treated with 50 M 4-HNE or 5% oleic acid (positive control) for 60 min followed by incubation for 24 h, and lipid accumulation was measured using Nile Red. From Fig. 6A, incubation of HepaRG cells with 4-HNE results in a significant 34% increase in lipid accumulation. This is in agreement with data we obtained in HepG2 cells (11).
The effects of 4-HNE on expression of ␤-oxidative enzymes in cell culture has not been reported. From Fig. 6B, expression of carnitine palmitoyl transferase (CPT1␣) and acyl-CoA synthetase 1 (ACSL1) was not significantly changed after overnight 4-HNE exposure, suggesting that 4-HNE does not have a direct effect on the expression of ␤-oxidative enzymes.
We sought to determine if AMPK kinases are also carbonylated after HepaRG cell exposure to 4-HNE by using biotin hydrazide modification and streptavidin capture (12). From Fig. 6C, an increase in carbonylation is clearly evident in the 4-HNE treated cells, indicating that AMPK is indeed a target of

4-HNE in HepaRG cells.
Combined with these data, we demonstrate that increased 4-HNE concentrations contribute to lipid accumulation as well as increased carbonylation of AMPK.
Effects of Chronic Ethanol on Hepatic Inflammation and Oxidative Stress-We have identified AMPK␣ as a novel target of reactive aldehydes in hepatocellular cell culture models. Our previous data indicate that ACC phosphorylation is not altered in response to increased AMPK phosphorylation in the liver after chronic ethanol consumption. We hypothesized that AMPK may also be carbonylated in our chronic ETOH con-sumption model. To determine if AMPK carbonylation was increased after chronic ETOH consumption, mice were fed increasing concentrations of ETOH for 7.5 weeks. As shown in Table 1, chronic ethanol consumption resulted in a significant increase in liver/body weight, serum alanine aminotransferase, hepatic triglycerides, and in overall carbonylation. After ETOH consumption, a significant decrease in GSH and in the ratio of GSH:GSSG indicate an increase in hepatocellular oxidative stress.
In other murine models, chronic ethanol consumption results in an increase in hepatic inflammation (43). The F4/80 antigen is an accepted marker of increased hepatic monocyte/ macrophage recruitment during chronic inflammation (6). To verify that this model of ETOH consumption results in an increase in inflammation, immunohistochemical staining of F4/80 ϩ cells was performed using formaldehyde-fixed, paraffin-embedded liver tissue isolated from pair-fed and ETOH-fed mice. As shown in Fig. 7A, chronic ethanol consumption resulted in a marked increase in F4/80 ϩ -stained cells in both the periportal and central venous regions of the liver. To determine the percent increase in monocytes/macrophages after ETOH, F4/80 ϩ cells were counted, and the ratio F4/80 ϩ cells: total cells (based on numerical analysis of nuclei) was determined. As shown in Fig. 7B, chronic ethanol consumption significantly increased monocytes/macrophage recruitment by Ͼ2-fold in both the periportal and centrilobular hepatic regions but not in the midzonal region.
We previously determined a blood ethanol concentration of 245 mg/dl after 6 weeks of chronic ethanol consumption (30). As shown in Fig. 7C, after 3 weeks serum ETOH concentrations increase linearly in our ETOH-fed animals, reaching an average concentration of 375 mg/dl at 7.5 weeks. Although high, these concentrations are representative of concentrations found in human alcoholics after hospital admission as well as in forensic analysis of cadaveric alcoholics upon death (44,45). Although this is higher than levels obtained after 6 weeks of chronic ETOH consumption (30), these concentrations are similar to concentrations found in rodents in other models after chronic ETOH consumption (46,47).
Although we have previously shown using immunohistochemistry that the reactive aldehydes 4-HNE, malondialdehyde . Cells were washed in PBS followed by incubation for an additional 24 h in serum containing media. B, exposure to 4-HNE does not affect expression of proteins involved in ␤-oxidation. HepaRG cells were treated with 4-HNE (50 M for 60 min) or oleic acid (5% 60 min-positive control). Cells were washed in PBS followed by incubation for an additional 24 h in serum containing media. Cells were lysed and examined for expression of ACSL and CPT1␣ via Western blotting. C, AMPK is carbonylated after 4-HNE exposure in HepaRG cells. HepaRG cells were treated with 50 M 4-HNE followed by cell lysis and purification of carbonylated proteins using biotin hydrazide capture (BH/SA). Whole cell extracts (WCE) and purified biotinylated proteins were analyzed using SDS-PAGE/Western blotting probing for AMPK␣. Experiments was performed in at least triplicate and subjected to one-way ANOVA with Tukey's multiple comparison test. *, p Ͻ 0.05; ***, p Ͻ 0.001.

TABLE 1 Liver parameters obtained from pair-fed and ETOH-fed mice
Basic liver parameters were obtained from chronic ETOH consumption for 7.5 weeks. Serum alanine aminotransferase (ALT; units/ml), liver/body weight ratio, hepatic triglycerides (mmol/mg of tissue), percent control carbonylation, GSH (mol/g of tissue), GSSG (mol/g of tissue), and GSH:GSSG were all determined as described under "Experimental Procedures." n ϭ 6 mice/group. (MDA), and acrolein are all increased by chronic ETOH consumption after 6 weeks, these effects have not been characterized after 7.5 weeks of ETOH consumption (48). From Fig. 8, in the control animals low levels of staining of 4-HNE, MDA, and acrolein occur. Compared with the control animals, 7.5 weeks of ETOH consumption resulted in a marked increase in immunostaining surrounding the portal triad of proteins modified by acrolein, MDA, and 4-HNE.
In ALD, inflammation is a precursor for fibrosis. To assess if our model exhibited an increase in fibrosis, picrosirius red staining was performed on tissue sections isolated from both pair-and ETOH-fed mice (49). Picrosirius staining of chronic ETOH liver sections did not indicate a substantial increase in hepatic fibrosis in this model (data not shown). This is further demonstrated by using polarized light and the same sections (Data not shown). Nevertheless, these data verify that after ETOH consumption hepatocellular damage, inflammation and carbonylation all are significantly increased in response to ETOH consumption.
In Fig. 5, we identified both AMPK␣ as a target of carbonylation in 4-HNE-treated HepG2 cells. In our previous report, we determined that although AMPK phosphorylation increased after ETOH consumption, the increase did not cor-relate with an increase in phosphorylation of ACC (20). In that study, mice were fed ETOH for 6 weeks, and hepatocellular damage as shown by alanine aminotransferase was only increased by 2.5-fold (20). Using whole cell lysates from the current 7.5-week feeding model, phosphorylation of AMPK␣ and ACC was assessed. As shown in the Western blot, presented in Figs. 9, A--C), AMPK phosphorylation was increased 1.5-fold in ETOH-fed mice, whereas phosphorylation of ACC was not significantly changed. By its ability to convert longchain fatty acids into long-chain acylcarnitines, CPT1␣ is the rate-limiting enzyme in ␤-oxidation. ACSLs activate longchain fatty acids to form acyl-CoAs, facilitating their transport across membranes for subsequent ␤-oxidation (50). In a recent report, CPT1 expression was increased after 7 weeks of ETOH consumption in mice (27). The effects of chronic ETOH on expression of CPT1␣ and ACSL in this study were examined by Western blotting. From Fig. 9A, expression of CPT1␣ and ACSL was not significantly changed after chronic ETOH consumption. Combined, these data support our hypothesis that expression of proteins associated with ␤-oxidative processes is not significantly changed in the presence of increased AMPK phosphorylation in our ETOH model. Instead, these data indicate that it may be the lack of increased ␤-oxidation in the presence of increased AMPK phosphorylation that contributes to lipid accumulation. Furthermore, they also provide additional evidence of regulation of AMPK signaling by carbonylation of AMPK.
In Figs. 1-5, 4-HNE treatment resulted in an increase in carbonylation and inhibition of AMPK activity in our HepG2 cell culture model. We, therefore, sought to determine if chronic ETOH consumption results in an increase in carbonylation of hepatic AMPK. As shown in Fig. 10, carbonylation of AMPK␣ is significantly increased in the liver of ETOH-fed mice. Figs. 1-5, 4-HNE treatment of HepG2 cells inhibits AMPK activity and modifies AMPK␣/␤. To further elucidate the effects of 4-HNE on AMPK, recombinant AMPK (rAMPK␣/␤/␥) was incubated with increasing molar ratios of 4-HNE (51). As shown in Fig. 11A, exposure of rAMPK␣/␤/␥ to increasing molar concentrations of 4-HNE results in increased modification of AMPK␣/␤ but not AMPK␥. We have previously shown that 4-HNE directly inhibits the activity of Akt1 as well as Akt2 kinases. To determine if 4-HNE treatment of rAMPK␣/␤/␥ inhibited activity, rAMPK␣/␤/␥ was treated with increasing molar concentrations of 4-HNE followed by activity assays. As shown in Fig. 11B, rAMPK activity was significantly decreased by 30% at a molar ratio of 5:1. Combined, these data indicate that AMPK is a direct target of reactive aldehydes, and modification of AMPK inhibits activity.

4-HNE Modifies and Inhibits AMPK Activity in Vitro-Based on data in
Identification of Amino Acid Residues on rAMPK␣␤ as Targets for 4-HNE Modification-Although 4-HNE inhibited AMPK activity, the mechanism of inhibition is unknown. To identify specific residues in AMPK␣ modified by 4-HNE, rAMPK␣ was treated with increasing molar ratios of 4-HNE followed by SDS-PAGE, band excision, tryptic digestion, and LC/MS/MS analysis. The resulting peptides obtained and their respective m/z ratios (mass to charge ratio) are presented in supplemental Table 1. Only peptides with a mascot ion score of Ͼ30 were considered as statistically significant and included. From supplemental Table 1, on AMPK␣, Cys 130 , Cys 174 , Cys 227 , and Cys 304 (all in bold) were identified as targets of 4-HNE modification. On AMPK␤, Cys 225 was also identified as a target of 4-HNE. The data in supplemental Fig. 1, A-E show MS/MS ion chromatograms obtained for each 4-HNE-modified peptide on AMPK␣/␤.
Effects of 4-HNE Modification on AMPK Structure-We examined known crystal structures of AMPK and determined of the identified modified cysteines, only Cys 130 , Cys 174 , Cys 227(␣) , and Cys 225(␤) have known locations in the overall AMPK structure (33). The available crystal structures of AMPK do not indicate the location of Cys 304 , but based on structural predictions, it is located in an exposed loop between the AMPK␣ and AMPK␥ subunits (data not shown) (33). Computational-based minimization simulations were performed using the crystal structure of AMPK (PDB code 2Y94) combined with each modification. A global view of all the modifications is shown in Fig. 12. Based on the information obtained from the modeling simulation, 4-HNE treatment did not result in any substantial predicted changes to the overall structure of AMPK (data not shown). Examining each adducted residue, 4-HNE Cys 225(␤) and Cys 227(␣) were located on the surface and did not appear to significantly alter either the active site or the activation loop (data not shown). As shown in Fig. 13, A and B, the minimized structure of Cys 174 -4HNE resulted in 4-HNE freely rotating out of the activation loop immediately adjacent to Thr(P) 172 . This suggests that Cys 174 -4HNE may decrease the efficiency of phosphorylation and/or dephosphorylation of Thr 172 in the activation loop. An examination of the minimized structure of Cys 130 -4HNE revealed a surprising result. In the crystal structure, the ATP competitive inhibitor staurosporine is shown bound to the active site. As shown in Fig. 13, C and D, in the crystal structure of AMPK␣, Cys 130 -4HNE is buried below the active site under staurosporine. This suggests that the adduction of Cys 130 by 4-HNE also has the potential to inhibit ATP binding and or substrate access. Although the specific location of Cys 304 is unknown, H 2 O 2 will oxidize Cys 304 , resulting in an increase in AMPK activity in HEK 293 cells (19). Relative to the location of Lys 299 and Arg 331 , Cys 304 is positioned on the opposite face from Cys 174 (data not shown). Mutation of Cys 304 will inhibit H 2 O 2 -induced increases in AMPK activity (19). This suggests that alkylation of Cys 304 may also play a role in regulating AMPK activity.
An attempt was made to immunoprecipitate AMPK followed by detection of direct modification by LC/MS/MS in our animal model as well as our cell culture model. Unfortunately these attempts were not successful (data not shown). Likewise, we were previously unsuccessful in identifying carbonylation of both PTEN as well as Akt2 in our murine models (10,12). We hypothesize that this is due to the low abundance of AMPK expression in hepatocytes.

DISCUSSION
By its ability to be activated by AMP and subsequently activate ␤-oxidation, AMPK is a critical cellular energy sensor in hepatocytes. In ethanol models, consumption has been implicated in both activation of AMPK as well as suppression of AMPK (20,21,23,26). These reports, however, utilize different lengths of ETOH consumption, different diets, and different ethanol concentrations. Recent publications have indicated that AMPK is regulated by oxidative stress (52). In this report we characterize the effects of preincubation with 4-HNE on AMPK activation in HepG2 cells. We further translate our cellular results into both in vitro and in vivo models.
The downstream effects of lipid aldehydes such as 4-HNE are multifactorial as well as cell-specific. As a major indicator of oxidative stress and inflammation, 4-HNE is proposed to simultaneously initiate cellular damage as well as promote cellular survival (53). Recent proteomic studies have identified numerous protein targets of 4-HNE in cell culture as well as in a murine model of early ALD (48,54). In cells, concentrations of 4-HNE are highest within membranes with calculated hepatocellular concentrations in CCl 4 -and BrCCl 3 -treated rats of 3.8 and 11.3 mM, respectively (8,55,56). These pathophysiological concentrations are significantly higher than concentrations used in this study (0 -50 M) and provide in vivo relevance to these experimental results.
AMPK has been shown to localize to cellular membranes. Localization of AMPK occurs during cellular metabolic stress via myristoylation of the ␤ subunit (57). In MCF-7 cells, another member of the protein A, G, and C kinase family (LKB1) is a known target of 4-HNE (58). Mutational studies identified a critical cysteine residue (Cys 210 ) within the activation loop of LKB1 to be alkylated by 4-HNE. In our study we also identified a critical cysteine residue near the active site. As shown in Figs. 5, 6, and 10, alkylation/carbonylation of AMPK␣ is significantly increased after 4-HNE treatment of hepatocytes (HepG2 cells as well as HepaRG cells) and in a murine model of ETOHinduced inflammation.  In our cellular model 4-HNE also inhibited AICAR-induced phosphorylation of ACC. This clearly indicates that reactive aldehydes can inhibit AMPK not only under conditions of increased oxidative stress but also during conditions of increased AMP. These conditions occur during chronic ETOH consumption, and based on our cellular data we hypothesize that the inhibition of AMPK by reactive aldehydes occurs not only in conditions of increased oxidative stress but under conditions of increased cellular ATP/AMP ratios (59 -61).
In other cell lines, protein phosphatase 2A has been demonstrated to be the primary mechanism involved in the dephosphorylation of Thr 172 on AMPK. In HEK293 cells, inhibition of PP2A resulted in increased cell survival after exposure to H 2 O 2 for 12 h (39). In Jurkat cells, 4-HNE has been demonstrated to activate PP2A (62). In our system, preincubation of cells with okadaic acid only partially rescued AMPK phosphorylation in the presence of 4-HNE, indicating a combinatorial effect of 4-HNE in the inhibition of AMPK in HepG2 cells. Concerning our ETOH model, these data suggest that increasing the length of ETOH consumption from 6 weeks to 7.5 weeks along with the final concentration of ETOH (31.5-34.7%) further increases serum ETOH concentrations as well as hepatocellular damage (12,20,30). In our ETOH model, AMPK phosphorylation increased in the ETOH-fed animals, but this increase did not correspond to an increase in Ser(P) 79 ACC. This is in agreement with our previous sixweek study as well as data obtained in an overfeeding ETOH model (21). Concurrently there was not a significant change in expression of CPT1␣ and ACSL (quantification not shown). This suggests that in the ETOH-fed liver, LKB1 is active as shown by increased AMPK phosphorylation, but decreases in AMPK downstream signaling are in part due to alternative mechanisms such as direct modification by reactive aldehydes. It also indicates that in this model, hepatic fatty acid accumulation may be due to the inability of AMPK to increase ␤-oxidation.
Examining 4-HNE inhibition of AMPK, the discrepancy between our cell culture model and animal model of ALD is intriguing. In our cell culture model, 4-HNE is administered in the medium and under conditions of short term incubation. This results in a time-dependent inhibition of AMPK phosphorylation. In ALD, there is chronic induction of oxidative stress and activation of Cyp2E1 that contributes to the production of reactive aldehydes in the mitochondria as well as in the endoplasmic reticulum (30,48,63). The discrepancy between the intracellular production of reactive aldehydes in ALD and the use of reactive aldehydes as an external cellular stimulus provides a plausible explanation for our cell culture data and our in vivo data.
Using recombinant AMPK and LC/MS/MS, we identify Cys 130 (located within the active site of AMPK␣), Cys 174 (located in the activation loop adjacent to Thr 172 ), Cys 227 (located on the external surface of AMPK␣), Cys 225 (located on AMPK␤), and Cys 304 (location not yet determined) as selective targets of 4-HNE Michael addition adducts at molar ratios of 5:1 (Fig. 11, supplemental Table 1 and Fig. 1). The identification of the AMPK␣ Cys 130 and Cys 174 as modified amino acids suggests some intriguing possibilities concerning the mechanism Purified rAMPK␣/␤/␥ was incubated with increasing molar ratios of 4-HNE for 30 min at room temperature. Samples were boiled in 5ϫ SDS loading buffer, run on an 8% SDS-PAGE gel, blotted, and probed for 4-HNE using anti-4-HNE polyclonal antibodies. B, inhibition of rAMPK␣ by 4-HNE. Purified rAMPK␣ was incubated with increasing ratios of 4-HNE, and activity assays were performed as under "Experimental Procedures." All samples were performed in at least triplicate. Statistical analysis was via one-way analysis of variance with Tukey's multiple comparison test. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. FIGURE 12. Global in silico molecular modeling of 4-HNE modified human AMPK␣/␤/␥. Shown is a ribbon diagram demonstrating locations of adducts found on AMPK␣/␤ (AMPK␣ backbone is in green, AMPK␤ backbone is in yellow, and 4-HNE adducted residues are in red). Human recombinant protein was used. of 4-HNE-mediated AMPK inhibition. The presence of a bulky hydrophobic lipid aldehyde group could prevent substrate access and, therefore, phosphorylation.
In vitro, although the specific location of Cys 304 is not known, the structure indicates that it is in a loop that is accessible to modification by 4-HNE. In HEK 293 cells, treatment with H 2 O 2 results in glutathionylation of Cys 304 and a concomitant increase in AMPK activity (19). If Cys 304 is mutated to Ala 304 , H 2 O 2 activation of AMPK is suppressed. The adduction by 4-HNE of AMPK Cys 304 may prevent this mechanism of AMPK activation by H 2 O 2 , but it is unknown if adduction would prevent AMPK activation by AMP/AICAR. In summary, due to its ability to be activated by AMP and subsequently activate ␤-oxidation, AMPK is a critical cellular energy sensor in hepatocytes. In ethanol models, consumption has been implicated in both activation of AMPK as well as suppression of AMPK/␤-oxidation (20,21,23,26). Recent publications have also clearly indicated that AMPK is regulated by oxidative stress (52). As summarized in Fig. 14, we identify AMPK as a direct target of reactive aldehydes in cell culture as well as in a murine chronic ETOH model. Carbonylation of AMPK prevents AMPK activation by H 2 O 2 as well as by AMP by directly inhibiting enzyme activity and thereby preventing phosphorylation of ACC. This marks the first time that AMPK␣ has been demonstrated to be a target of reactive aldehydes. Using in vitro phosphorylation assays as well as examining downstream targets of AMPK, we further determine that carbonylation of AMPK directly inhibits enzymatic activity both in vitro as well as in the liver of mice chronically fed ETOH. These data provide a plausible mechanism for the occurrence of increased AMPK phosphorylation in the absence of significant changes in phosphorylation of ACC in our ETOH model (20). Furthermore, this also provides a mechanistic explanation for our data indicating no significant change in ␤-oxidation after chronic ETOH consumption (29).