Chronic Ethanol Intake Impairs Insulin Signaling in Rats by Disrupting Akt Association with the Cell Membrane

Chronic and excessive alcohol consumption is an important and modifiable risk factor for type 2 diabetes. We previously reported elevations in hepatic Class 1 alcohol dehydrogenase (ADH) expression in ethanol-fed rats correspondent with reduced levels of mature, nuclear sterol-regulatory element-binding protein-1 (SREBP-1), an insulin-induced transcriptional repressor of the ADH gene. In this report, we have studied the effects of insulin and ethanol on ADH gene expression in a highly differentiated rat hepatoma cell line (FGC-4), as well as the in vivo effects of chronic intake of an ethanol-containing diet on hepatic insulin signaling. Insulin inhibited ADH gene expression, and this was abolished by LY294002 (a phosphatidylinositol 3-kinase inhibitor) and small interfering RNA knockdown of SREBP-1. Chronic ethanol intake led to decreased phosphorylation of Akt (protein kinase B) at Thr308, increased phosphorylation of Akt at Ser473, and decreased phosphorylation of glycogen synthase kinase-3β (a downstream effector of Akt). Hepatic membrane-associated Akt content was decreased and cytosolic Akt content was increased in rats fed an ethanol-containing diet. Thus, disruptive effects of ethanol on insulin signaling occurred via impaired phosphorylation of Akt at Thr308. TRB3, a negative regulator of Akt, was induced in liver of ethanol-fed rats. In ethanol-treated FGC-4 cells, small interfering RNA knockdown of TRB3 increased membrane-associated Akt and the phosphorylation of Akt at Thr308. Our results suggest that ethanol induces TRB3, which, through binding to the pleckstrin homology domain of Akt, prevents its plasma membrane association, Akt-Thr308 phosphorylation, and subsequent Akt-mediated signaling. Ethanol inhibition of insulin signaling reduces nuclear SREBP accumulation and results in disinhibition of Class 1 ADH transcription.

Writings from as early as the 17th century, as well as modern epidemiological studies, suggest that chronic and excessive alcohol consumption is positively associated with the onset of type 2 diabetes (1)(2)(3)(4)(5). Ethanol intake has been reported to decrease glucose uptake and utilization consistent with the development of insulin resistance, a central component of diabetes (6). Results from these previous studies suggest that the ethanol impairment of insulin action is likely to be downstream from PI3 2 kinase; however, the molecular mechanisms underlying the effects of alcohol on insulin resistance and type-2 diabetes remain to be determined (7).
We previously reported that chronic intragastric infusion of an ethanol-containing diet to rats results in unique and predictably recurring cyclic fluctuations in plasma ethanol concentrations (8) as a consequence of cyclic expression of the major alcohol metabolizing enzyme, hepatic Class 1 alcohol dehydrogenase (ADH). Further studies from our laboratory demonstrated that alcohol induces hepatic ADH gene transcription via decreased levels of nuclear SREBP-1c protein, a negative regulator of the ADH gene (9). SREBP-1c is encoded by an insulinresponsive gene (10) and is an important early mediator in the pathway of insulin action in the liver (11). These observations led to the current hypothesis under study, namely that ethanol may suppress nuclear SREBP-1c via disruption of insulin signaling, which may be a potential link between alcohol consumption and insulin resistance. In this study, we elucidate the mechanism by which chronic ethanol intake inhibits insulin action and identify TRB3, a previously identified modulator of Akt signaling, as a primary ethanol-responsive modifier of insulin signaling (12). using a computer-driven programmable pump that supplies diet continuously, except for 1 h each day needed to refill the syringes, as described previously (13). The control rats were infused the same diet except that ethanol was isocalorically replaced with carbohydrate. We used urine ethanol concentrations (UECs) to track blood ethanol concentrations since ethanol equilibrates with body water and we have previously demonstrated that UECs accurately reflect blood ethanol concentrations (36). Rats were sacrificed following 21-30 days of continuous diet infusion and when their UECs were greater than 300 mg/dl, as described previously (8,9,14). The actual time of sacrifice occurred while diets were being infused to avoid any changes in insulin levels associated with stopping the diet infusions. Liver was collected and stored at Ϫ70°C. Type 2 diabetic rat livers were from a previously reported study (15). Briefly, Sprague-Dawley rats (100 g) were fed a high fat diet for 14 days and then treated with low dose (45 mg/kg) streptozotocin. Type 2 diabetes was confirmed by hyperglycemia, normoinsulinemia, and oral glucose tolerance testing (15).
Northern Blot Analysis-Northern blot analysis was conducted as described previously (8,9). A 400-bp rat Class 1 ADH cDNA probe was used for the detection of ADH mRNA. Rat TRB3 probe was generated using the following primers: 5Ј-TTGCGTCGATTTGTCTTCAG-3Ј (5Ј-primer) and 5Ј-CTCCCTCAACCAGGGATGTA-3Ј (3Ј-primer). 18 S ribosomal RNA antisense oligonucleotide was synthesized (Bio Synthesis, Inc., Lewisville, TX). All filters were probed with the 18 S rRNA oligonucleotide as an internal control. Bands were quantitated by densitometry of the autoradiogram, and the ratio of ADH message to 18 S rRNA in the same sample was determined and expressed as relative RNA units or as percentage of that for the control.
Preparation of Rat Liver Extracts, Crude Plasma Membrane Fractions, Cytosolic Fractions, Nuclear Extracts, and Measurement of Akt Activity-Crude plasma membrane fractions and cytosolic fractions were prepared from liver tissue homogenized with a Dounce homogenizer using a previously described procedure (16). Nuclear extracts were prepared as described previously (9). Akt activity was measured by using an Akt kinase assay kit from Cell Signaling.
Western Immunoblot Analysis-Samples with equal amounts of protein determined by the Bradford protein assay were resolved on polyacrylamide gels. One was transferred to a Hybond-P membrane (Amersham Biosciences), and the other was stained with Coomassie Blue to confirm equal sample loading. Membranes were blocked overnight at 4°C in Tris-buffered saline with Tween plus 5% (w/v) milk powder with gentle shaking and were then incubated with different primary antibodies diluted in Tris-buffered saline with Tween. Proteins were visualized using the enhanced chemiluminescence plus system (ECL Plus; Amersham Biosciences) following incubation with secondary antibodies. Detection by autoradiography was as described previously (9).
Cell Culture, Transfection, and Confocal Microscopy-Rat hepatoma FGC-4 cells (obtained from Mary C. Weiss, Pasteur Institute, Paris, France) were seeded in medium (Dulbecco's modified Eagle's medium containing 100 units/ml penicillin and 100 g/ml streptomycin sulfate and 5% fetal bovine serum). For transfection, FGC-4 cells at 60% confluence were incubated for 6 h with plasmid/Lipofectamine 2000 mixture following the manufacturer's protocol. Subcellular localization of GFP-tagged Akt was studied in cells cultured in Flaskette chamber slides sealed with a gasket (Nunc International, Naperville, IL) after treatment with insulin and/or ethanol and followed by fixation in 4% paraformaldehyde and examination with a Zeiss LSM410 laser scanning confocal microscope. For experiments with ethanol treatment, cells were cultured in glucose-free Dulbecco's modified Eagle's medium supplemented with 1 mM pyruvate and 5% fetal bovine serum; all caps were tightly closed for the remainder of the experiment to prevent the evaporative loss of ethanol.
RNA Interference and Co-immunoprecipitation-Double-stranded stealth RNA duplexes (Invitrogen) corresponding to rat TRB3 (165-189 bp, NM_144755) coding region (5Ј-ATCTCTGGCTGCTTCTGC-CGATGTT-3Ј) were transfected into FGC-4 cells. SREBP-1 siRNA was purchased from Dharmacon Inc. (Lafayette, CO). The effect of RNA interference was measured after 48 h of incubation. The negative control siRNA was used in control transfections (Ambion). For co-immunoprecipitation, cells were transfected with pEGFP-Akt expression vectors or the pEGFP-N1 control vector by Lipofectamine 2000 for 24 h; total cell lysates were obtained and incubated with antiserum to GFP and protein G slurry overnight at 4°C. Immunoprecipitates were collected, washed, and separated by SDS-PAGE (12%) followed by Western blotting using TRB3 antiserum.
Statistics-The Student's t test was used to determine whether group means differed at a significance level of p Ͻ 0.05.

RESULTS
Effects of Chronic Ethanol Intake on UECs-UECs did not reach a steady-state level, although the ethanol-containing diet was infused continuously for more than 3 weeks (data not shown). Rather, UECs fluctuated in the form of large regular "pulses" such that concentrations varied between a nadir of near zero to peak levels of greater than 500 mg/dl over a 6-day period, similar to our previous observations (8). The mean blood and urine alcohol concentrations were greater than 300 mg/dl at the time of sacrifice (data not shown); hepatic nSREBP-1 protein was reduced by more than 50%, and hepatic Class I ADH was induced by greater than 2-fold, results nearly identical to those previously reported (9). Serum insulin concentrations did not differ significantly between control and ethanol-fed rats (1.00 Ϯ 0.11 versus 0.85 Ϯ 0.10 ng/ml, respectively).

Effects of Ethanol and Insulin on Class 1 ADH mRNA in Hepatoma FGC4 Cells and Livers of Type 2 Diabetic
Rats-The effects of ethanol and insulin on ADH mRNA expression were studied in highly differentiated rat hepatoma FGC-4 cells. Fig. 1A presents a Northern blot of FGC-4 cells treated with various doses of insulin (0.1-10 nM) in the presence or absence of ethanol (50 mM). The data demonstrate that ethanol induced and insulin reduced ADH mRNA abundance in a dosedependent manner (p Ͻ 0.05), an effect that was completely abolished by 50 mM ethanol, confirming that insulin inhibits in vitro ADH expression. Insulin treatment increased nuclear SREBP-1c protein expression (p Ͻ 0.05), and ethanol treatment not only prevented the insulin induction of nSREBP-1 but reduced (p Ͻ 0.05) it to levels below the control (Fig. 1C). These data are consistent with the model that insulin regulates ADH via changes in nSREBP-1c (9). The PI3K inhibitor LY294002 not only blocked the insulin inhibition of ADH gene expression but also induced ADH mRNA expression more that 2-fold over control ( Fig. 2A).  These data suggest that insulin inhibits ADH gene expression through the PI3K pathway. When SREBP-1 was knocked down by specific siRNA, insulin-inhibition of ADH gene expression was negated, and ADH mRNA was significantly increased (Fig. 2B). This evidence confirmed our previous report that SREBP-1 is a repressor of ADH gene expression (9). The above data indicate that insulin inhibits ADH gene expression through PI3K and SREBP-1.
To substantiate the inhibition by insulin of ADH gene expression, we examined ADH mRNA levels in livers of type 2 diabetic rats and found that ADH mRNA was induced 4-fold as compared with liver from control rats (Fig. 2C). These results are consistent with the elevated ADH activity reported in diabetic male rats (18).
Effects of Ethanol on the Phosphorylation of GSK3, Akt, and Proximal Insulin Signaling in Rat Liver-We examined GSK3 because it inhibits SREBP-1c transcriptional activity (19) and stimulates ubiquitindependent degradation of SREBP-1c (20). Insulin has been reported to inhibit GSK3 as a result of phosphorylation. Western blot analyses were performed on liver whole cell lysates from rats fed an ethanol-containing diet and that were killed when blood ethanol concentrations were greater than 300 mg/dl. Although no ethanol-induced changes were observed in total GSK3 levels, p-GSK3␤ levels were lower (p Ͻ 0.05) in the hepatic whole cell lysates of ethanol-fed rats (Fig. 3, A and B). Ethanol did not affect total Akt protein abundance (Fig. 3A), but Akt phosphorylation of Thr 308 (p-Akt-Thr 308 ) was inhibited by ethanol (p Ͻ 0.05), and Akt phosphorylation of Ser 473 (p-Akt-Ser 473 ) was increased (p Ͻ 0.05) relative to control animals (Fig. 3, A and B). Decreased (p Ͻ 0.05) hepatic Akt activity was observed in ethanol-treated animals as compared with control (0.48 Ϯ 0.11 versus 1 Ϯ 0.12 arbitrary densitometric units (ADU)). Notably, the alcohol-induced decreases in levels of phospho-GSK3␤, p-Akt-Thr 308 , and Akt activity coincided with reduced hepatic nuclear SREBP-1 protein content and increased ADH activity as reported previously (9). Thus, these results are consistent with a role for both Akt and GSK3 in alcohol induction of ADH through SREBP-1.
Western blot analysis was used to monitor the effects of ethanol on the subcellular localization of PDK1, p110␣ (the catalytic subunit of PI3K), p85, and Rictor (a critical component of the PDK2 complex) (Fig.  4A). In agreement with a previous report demonstrating that ethanol treatment increased PI3K activity in rat liver (6), we found significant recruitment of PDK1, p110␣, p85, and Rictor to the hepatic plasma membrane of rats fed an ethanol-containing diet (Fig. 4A). Cellular distribution of Rictor abundance was assessed and found to be a membrane-enriched protein (Fig. 4B). There were no statistically significant differences in hepatic phospho-PTEN (1.00 Ϯ 0.017 versus 0.99 Ϯ 0.012 ADU) or PTEN (1.00 Ϯ 0.041 versus 1.095 Ϯ 0.041 ADU) between control and ethanol-fed rats, respectively. Subcellular Localization of Akt in Rat Liver-Data in Fig. 3 demonstrate that ethanol impairs phosphorylation of Akt at Thr 308 . Translocation of Akt from the cytosol to the plasma membrane and binding to phosphatidylinositol-3, 4, 5-triphosphate (PIP 3 ) are crucial for the phosphorylation on Thr 308 by membrane-located PDK1 (21,22). We, therefore, examined the subcellular location of Akt in response to ethanol consumption. Ethanol reduced (p Ͻ 0.05) membrane-localized Akt (Fig. 5A), in agreement with the observed ethanol suppression of p-Akt-Thr 308 (Fig. 3A), whereas ethanol increased (p Ͻ 0.05) the amount of Akt in the cytosolic fractions (Fig. 5A). These data suggest that ethanol prevents association of Akt with the plasma membrane-bound PIP 3 , a prerequisite for PDK1 phosphorylation of Akt-Thr 308 .
To visualize the subcellular localization of Akt in response to insulin and ethanol, we transfected a GFP-Akt (full-length) expression plasmid into FGC-4 cells and monitored intracellular localization of GFP-Akt by confocal microscopy. Upon insulin stimulation, a significant portion of GFP-Akt was distributed to the plasma membrane, whereas ethanol abolished the insulin-stimulated GFP-Akt membrane localization (Fig.   FIGURE 3. Ethanol effects on the levels of GSK3 and Akt proteins in whole cell lysates of rat liver. A, ethanol-fed rats were sacrificed when the UECs were high on the descending limb of the UEC pulse. Control rats were fed non-ethanol-containing diets and were sacrificed at the same time (n ϭ 6 rats/group). Each lane represents an individual rat. Western blot analyses of phospho-GSK3␤ (p-GSK3␤), total GSK3(GSK3) levels; Akt phosphorylated at Thr 308 (p-Akt-Thr 308 ); and Akt phosphorylated at Ser 473 (p-Akt-Ser 473 ) and Akt are shown. B, the corresponding densitometric analysis (means Ϯ S.E.) of data in panel A. Asterisks indicate significant differences from the respective control at p Ͻ 0.05. 6A). To confirm these results, plasma membrane fractions were examined for GFP-Akt protein in Western blot analysis. As can be seen in Fig.  6, B and C, insulin-stimulated cells had five times greater levels of GFP-Akt (p Ͻ 0.05), and this was blocked by ethanol treatment. It is noteworthy that there are 3 amino acid residues that differ between our cloned Akt2 cDNA from Sprague-Dawley rats (GenBank number DQ198085) and Akt2 previously cloned from Rattus norvegicus, but these 3 amino acid residues are identical to mouse Akt2 at the corresponding positions.
Role of TRB3 in Activation of Akt-TRB3 has been reported to inhibit insulin signaling by binding directly to Akt and preventing phosphorylation at Thr 308 (12). Thus, induction of TRB3 is a potential mechanism by which ethanol could block Akt activation. We found that both protein (Fig. 5A) and mRNA (Fig. 7A) abundance of TRB3 were greater (p Ͻ 0.05) in hepatic tissues from rats fed ethanol-containing diet. In whole cell lysates of FGC-4 cells, ethanol increased (p Ͻ 0.05) the abundance of TRB3 protein and decreased (p Ͻ 0.05) p-Akt-Thr 308 and nuclear SREBP-1 (Fig. 8A), suggesting the possibility that ethanol blocks Akt activation by promoting the association between TRB3 and Akt. To confirm the inhibitory action of TRB3 on the activation process, RNA interference was used to suppress TRB3 levels in ethanol-treated FGC-4 cells. In these cells, siRNA decreased TRB3 (p Ͻ 0.05) and increased p-Akt-Thr 308 and nuclear SREBP-1 (Fig. 8B), supporting elevated translocation of Akt to the plasma membrane upon reduced TRB3. We confirmed this by demonstrating that siRNA knockdown of TRB3 in FGC-4 cells also prevented the ethanol-induced inhibition of insulin-stimulated membrane translocation of GFP-Akt (Fig. 8, C and D).
The above data raised the possibility that TRB3 binds to the PH domain of Akt, prevents this domain from binding PIP 3 , and thereby localizes Akt to the cytoplasm. To test this possibility, we constructed a truncated Akt (1-147 amino acids)-GFP fusion protein containing the PH domain of Akt2, which mediates interaction between myosin II and Akt (23). Full-length and truncated GFP-Akt constructs and control FIGURE 4. Ethanol effects on the levels of PDK1, p110␣, p85, and Rictor proteins in whole cell lysates and membrane fractions of rat hepatic tissues. A, Western blot analyses of PDK1, p110␣, and p85 in whole cell lysates (Cell Lysate) or plasma membrane fractions (Membrane) of rat liver; also shown is a blot of Rictor protein in the plasma membrane fractions of rat liver. Ethanol-fed rats were sacrificed when the UECs were high on the descending limb of the UEC pulse. Control rats were fed non-ethanol-containing diets and were sacrificed at the same time (n ϭ 6 rats/group). Each lane represents an individual rat. Bar graphs represent corresponding densitometric analysis (means Ϯ S.E.). Asterisks indicate significant differences from the respective control at p Ͻ 0.05. N.D., not detected. B, whole cell lysates, cytosolic fractions, and plasma membrane fractions were separated by SDS-PAGE (90 g of protein) and analyzed for Rictor by immunoblot analysis. Each lane represents an individual rat. W. Cell, whole cell lysates.
GFP vector were transfected into FGC-4 cells, and cell extracts were prepared. Antibody against GFP was used to immunoprecipitate the GFP-Akt fusion proteins, and immunoprecipitates were probed with TRB3 antiserum in Western blots (Fig. 8E). Truncated GFP-Akt as well as full-length GFP-Akt were co-immunoprecipitated with TRB3, whereas the GFP control protein did not complex with TRB3. These data indicate that TRB3 associates with the PH domain of Akt2.

DISCUSSION
Chronic alcohol consumption can result in alcoholic liver disease, insulin resistance, and type 2 diabetes. Although there have been extensive studies on alcoholic liver disease, less research has been conducted on the mechanisms underlying ethanol impairment of insulin signaling with some inconsistencies in reported results apparent. For example, although both acute and chronic ethanol intake have been reported to cause insulin resistance, as evidenced by decreased glucose utilization and impaired insulin suppression of hepatic glucose production, some reports suggest that this is accompanied by enhanced phosphorylation of insulin receptor, IRS-1 and IRS-2, and by increased PI3K activity (6), whereas other reports using chronic ethanol exposure models have found reduced phospho-IRS-1 levels and lower activities of PI3K and Akt in liver (24). Thus, mechanisms underlying ethanol-induced insulin resistance remain to be clarified.
We previously reported that ethanol induces ADH in part by disinhibition of ADH gene expression. This is accomplished by reducing the abundance of mature nSREBP-1c, which is a negative regulator of the ADH gene and which functions in concert with the CCAAT/enhancerbinding proteins to mediate ethanol regulation of ADH (9,14). Because SREBP-1c has been documented as an important early mediator in the pathway of insulin action in liver (11) and alcoholism is a known risk factor for diabetes and insulin resistance, we became interested in the potential linkage of alcohol, insulin, and regulation of ADH gene expression. In this report, we assessed insulin effects on ADH gene expression as well as ethanol effects on insulin signaling pathways. We found that insulin inhibits ADH gene expression. This inhibition occurs through the PI3K pathway and involves SREBP-1, as both LY294002 (a PI3K inhibitor) and knockdown of SREBP-1 abolished insulin effects and disinhibited ADH gene expression. These data establish a relationship between insulin-PI3K-SREBP1-ADH gene expression and confirm that SREBP-1 is a negative regulator of the ADH gene, as suggested in our previous report (9). SREBP-1c must translocate from the cytosol to the nucleus, where it  . TRB3 mRNA levels in rat liver. A shows TRB3 mRNA levels in rat liver. 20 g of total RNA from rat hepatic tissues were subjected to Northern blot hybridization with 32 P-labeled rat TRB3 probe and 18 S rRNA oligonucleotide. Ethanol-fed rats were sacrificed when the UECs were on the high UEC limb of the UEC pulse. Control rats were fed non-ethanol-containing diets and were sacrificed at the same time (n ϭ 6 rats/group). Each lane represents an individual rat. B depicts the densitometric analysis of the above Northern blots (means Ϯ S.E.). Asterisk indicates a significant difference from the control at p Ͻ 0.05. CON, control. binds SREs on the promoter of ADH gene and prevents gene transcription. Ethanol consumption favors binding of enhancer-binding proteins ␤ and ␦ (C/EBP ␤ and ␦) to the promoter and decreases SREBP-1 abundance and nuclear binding to disinhibit ADH gene transcription (9). GSK3␤ negatively regulates SREBP-1c nuclear accumulation (19,20). GSK3␤ phosphorylated at Ser 9 (p-GSK3␤) is inactive. We found that p-GSK3␤ was decreased in hepatic lysates of rats fed the ethanol-containing diet. This suggests that ethanol could be reducing nuclear accumulation of nSREBP-1 by inhibiting phosphorylation of GSK3␤, and thus, activating this kinase to increase SREBP phosphorylation and decrease its stability. Insulin stimulates SREBP-1c signaling by inhibiting GSK3␤ kinase activity, and this inhibition involves Akt-dependent phosphorylation of GSK3␤ (25). Full activation of Akt requires phosphorylation at both Thr 308 and Ser 473 amino acid residues (26). We found that ethanoltreated rats had lower Akt activity, and this was associated with a significantly altered phosphorylation pattern at Akt-Thr 308 and Akt-Ser 473 . Ethanol reduced the abundance of hepatic p-Akt-Thr 308 , an effect probably responsible for the observed decrease in Akt activity in our rats Akt is phosphorylated at Thr 308 by the membrane-bound protein, PDK1 (27). We examined the Akt protein level in whole cell lysates and membrane fractions in livers of ethanol-fed rats. Ethanol decreased membrane-associated Akt and blocked the Akt-membrane translocation and effectively prevented phosphorylation at the Thr 308 . To corroborate these findings, a fusion protein (GFP-Akt2) was constructed for visualizing its subcellular localization. Upon insulin stimulation, the GFP-Akt2 protein increased in the membrane fraction, and ethanol abolished this insulin action. It is worth noting that ethanol decreased membrane-localized GFP-Akt2 as compared with the control group because these data may explain why ethanol also induced ADH in FGC-4 cells in the absence of insulin.
TRB3 is a cytosolic protein that is reported to directly associate with Akt and to block phosphorylation at Akt-Thr 308 (12). Increased TRB3-Akt binding has been associated with reduced Akt activity and with inhibition of insulin signal transduction. We found that ethanol increased cytosolic TRB3 protein levels. We surmise that TRB3 binds and prevents Akt from membrane translocalization. This is based on the following data from ethanol-fed rats showing: 1) decreased levels of membrane-associated Akt; 2) increased cytosolic Akt levels; and 3) elevated TRB3 in the cytosol. Indeed, when TRB3 expression was suppressed by RNA interference, the phosphorylation of Akt at Thr 308 and the membrane-localized GFP-Akt2 were increased, and ϳ60% of insulin stimulation was restored in the presence of ethanol. This is consistent with previous data showing that overexpression of TRB3 in liver increases glucose production (12) and that knockdown of TRB3 by RNA interference improves glucose tolerance (28).
Thus, ethanol may be blocking the insulin-mediated SREBP-1 pathway by regulating TRB3. We next used procedures reported by Brazil et al. (29) to study lipid and protein binding to the PH domain of Akt. We showed that endogenous TRB3 co-immunoprecipitated with the PH domain of Akt, suggesting that there is a TRB3 binding site(s) at the PH domain. Since TRB3 has been shown previously to bind Akt (lacking PH domain) near Thr 308 to prevent its phosphorylation (12), our data suggest a second TRB3 binding site on Akt. We propose that TRB3 binds to the PH domain and blocks the lipid binding site, thereby preventing membrane association and providing a second site on Akt at which TRB3 can inhibit insulin signaling. Ethanol appears to be impairing this insulin signaling pathway by increasing TRB3 expression, which leads to subsequent inhibition of Akt translocation and decreased Akt phosphorylation. The increased PI3K activity may be due to the decreased "retrophosphorylation" of insulin receptor and insulin receptor substances by downstream kinase cascades, and also, acetaldehyde (metabolite of ethanol) reduces protein tyrosine phosphatase activity (30).
Interestingly, p-Akt-Ser 473 was elevated in liver of rats fed an ethanolcontaining diet. PDK2 was recently identified as an Akt-Ser 473 kinase (31,32). We measured Rictor (a critical component of this PDK2 enzymatic complex) protein levels in whole cell extracts, cytosol, and cytoplasmic membrane extracts and found that Rictor is a membrane-associated protein. Ethanol increased the membrane association of Rictor, PDK1, p110␣, or p85. However, ethanol-induced TRB3 binds to Akt and prevents its membrane association. Therefore, one might assume that phosphorylation at both Thr 308 and Ser 473 should be similarly affected by ethanol. Since this was not the case, we surmise that Akt phospho-status is determined by a balance between kinase and phosphatase activities. Phosphatases that specifically dephosphorylate Ser 473 , such as PH domain leucine-rich repeat protein phosphatase (PHLPP) (33), might be inhibited by downstream effectors or by ethanol (or its metabolite, acetaldehyde), resulting in increased phosphorylation at Ser 473 . Moreover, since phosphorylation of Ser 473 boosts the phosphorylation of Thr 308 (31), the increased phosphorylation of Ser 473 might be compensatory as TRB3 disrupts insulin signaling.
Based on the data presented herein and previously published data from our laboratory and well as others, we propose the following insulindependent scheme by which chronic ethanol intake can induce expression of Class I ADH in the rat (Fig. 9). Insulin binding to its receptor results in tyrosine phosphorylation of IRS-1 and IRS-2, which provides docking sites for PI3 kinase and also activates this kinase to phosphorylate phosphatidylinositol-4, 5-biphosphate, generating PIP 3 (34,35). Akt binds PIP 3 via its PH domain, triggering the phosphorylation of Akt-Thr 308 by PDK1. To fully activate Akt, PDK2 (mammalian target of rapamycin (mTOR)-Rictor-G␤L) phosphorylates Akt-Ser 473 (21,22,31). Activated Akt in turn phosphorylates and inactivates GSK3␤, a kinase that inhibits SREBP-1c transcriptional activity (19) and stimulates ubiquitin-dependent degradation of SREBP-1c (20). SREBP-1c is then able to translocate to the nucleus, where it binds the SRE sites on the ADH gene promoter and inhibits transcription of the gene. TRB3, a mammalian homolog of the Drosophila Tribbles, previously identified as neuronal cell death-inducible putative protein kinase (NIPK), functions as a negative modulator of Akt by binding directly to Akt and disrupting insulin signaling (12). Ethanol increases TRB3 by an as yet unknown mechanism, which binds the PH domain of cytosolic Akt, thereby blocking Akt phosphorylation, resulting in the maintenance of GSK3 activation. GSK3-induced phosphorylation of SREBP-1c reduces the abundance of nSREBP-1c, thereby disinhibiting ADH gene transcription. Future studies in our laboratory will examine the mechanisms underlying ethanol effects on TRB3 mRNA and protein expression.