Regulation of GLUT4 gene expression by arachidonic acid. Evidence for multiple pathways, one of which requires oxidation to prostaglandin E2.

We have previously described the ability of arachidonic acid (AA) to regulate GLUT4 gene expression (Tebbey, P. W., McGowan, K. M., Stephens, J. M., Buttke, T. M., and Pekala, P. H.(1994) J. Biol. Chem. 269, 639-644). Chronic exposure (48 h) of fully differentiated 3T3-L1 cells to AA resulted in an 90% suppression of GLUT4 mRNA accumulation. This decrease was demonstrated to be due to a 50% decrease in GLUT4 gene transcription as well as a destabilization of the GLUT4 message (t decreased from 8.0 to 4.6 h). In the current study we have identified, at least in part, the mechanism by which AA exerts its effects on GLUT4 expression. Compatible with a cyclooxygenase mediated event, the AA-induced suppression of GLUT4 mRNA was abolished by pretreating the cells with the inhibitor, indomethacin. Consistent with this observation, exposure of the cells to 10 μM PGE mimicked the effect of AA, in contrast to products of the lipoxygenase pathway which were unable to suppress GLUT4 mRNA content. Quantification of the conversion of AA to PGE demonstrated a 50-fold increase in PGE released into the media within 7 h of AA addition. Cyclic AMP levels were also increased 50-fold with AA treatment consistent with PGE activation of adenylate cyclase. Various long chain fatty acids, including the nonmetabolizable analog of AA, eicosatetraenoic acid (ETYA), also decreased GLUT4 mRNA levels. The effect of ETYA, a potent inhibitor of both lipo- and cyclooxygenases and a potent activator of peroxisome proliferator activated receptors (PPARs), suggested the presence of a second pathway where nonmetabolized fatty acid functioned to suppress GLUT4 mRNA levels. Further support for a PPAR-mediated mechanism was obtained by exposure of the cells to the classic PPAR activator, clofibrate, which resulted in a 75% decrease in GLUT4 mRNA content. Nuclear extracts prepared from the adipocytes contained a protein complex that bound to the PPAR responsive element (PPRE) found in the promoter of the fatty acyl-CoA oxidase gene. When the adipocytes were treated with either AA or ETYA, binding to the PPRE was disrupted, consistent with an ability of these fatty acids to control gene expression by altering the occupation of a PPRE. However, a perfect PPRE has yet to be identified in the GLUT4 promoter, but this does not rule the possibility of a PPAR playing an indirect role in the AA-induced GLUT4 mRNA suppression.

Glucose transport across the plasma membrane represents the rate-limiting step in glucose metabolism and is a highly regulated process in the animal cell. Facilitated diffusion of glucose into the cell is carried out by a family of stereospecific transport proteins known as the glucose transporters (GLUT1 through GLUT5 and GLUT7). These integral membrane proteins are members of a gene family in which tissuespecific expression of one or more members will in part determine the net rate of glucose entry into the cell. In addition to tissue-specific expression, hormones, growth factors, and fatty acids can influence the net flux of glucose across the plasma membrane (Kahn, 1992;James et al., 1989;Birnbaum, 1989;Corneilus et al., 1990;Stephens and Pekala, 1992;Tebbey et al., 1994). 1 In addition to roles in the biosynthesis of lipids and energy production, fatty acids are involved in the regulation of glucose metabolism. Fatty acids in general and arachidonic acid (AA) 2 specifically have been demonstrated to be physiological regulators of the adipocyte glucose transport system (Hardy et al., 1991;Murer et al., 1992;Hunnicut et al., 1994;Tebbey et al., 1994). Arachidonic acid (20:4), a major metabolically important polyunsaturated fatty acid present in mammalian cells, is synthesized in the liver from dietary linoleic acid (18:2) and then transported via serum albumin or lipoproteins to various tissues. Serum levels of AA are low relative to other fatty acids except in obesity and diabetes where levels can be significantly elevated over normal matched controls (Distel et al., 1992;Svedberg et al., 1990;Grunfeld et al., 1981). In addition, many cells possess a high affinity arachidonyl-CoA synthetase which facilitates selective accumulation of AA even when other fatty acid species are in excess (Neufeld et al., 1983;Taylor et al., 1985). Such characteristics imply an important role for AA in cellular growth and function and as evidence for such, AA has been implicated in the regulation of gene expression (Tebbey and Buttke, 1992;Glaslow et al., 1992;Ntambi, 1992;Clarke and Abraham, 1992;Tebbey et al., 1994). Specifically, AA was found to suppress the transcription rate of a number of genes, including: the stearoyl-CoA desaturase 2 in T-cells (Tebbey and Buttke, 1992), the hepatic fatty acid synthase (Clarke and Abraham, 1992), and GLUT4 in 3T3-L1 adipocytes (Tebbey et al., 1994). Interestingly, stearoyl-CoA desaturase 2 and GLUT4 are coordinately expressed during the differentiation process in the 3T3-L1 adipocytes (Kaestner et al., 1989(Kaestner et al., , 1990 and results obtained in transient co-transfection assays suggest the possibility of identical mechanisms of activation of the promoters for both the GLUT4 and stearoyl-CoA desaturase 2 genes (Kaestner et al., 1990;Christy et al., 1989). Thus, control of the expression of these genes by AA may be part of a generalized regulation of adipose-specific gene expression.
These in vitro results are supported by dietary experiments demonstrating that high safflower oil content diets (rich in -6 fatty acids) resulted in decreased cellular content and distribution of GLUT4 in rat adipocytes (Ezaki et al., 1992). Insulinstimulated glucose transport activity was decreased to 51% of controls, and GLUT4 protein content of both plasma and microsomal membranes decreased by 35%. Fish oil feeding (increased -3 fatty acids) transiently improved the safflower oil-mediated decrease of insulin-stimulated glucose transport activity by increasing the amount of both GLUT1 and GLUT4 proteins. These data suggest that the regulatory properties of fatty acids may be determined by specific structure-function relationships.
Previous studies from our laboratory (Tebbey et al., 1994) examined the mechanisms by which AA can modulate glucose homeostasis in fully differentiated 3T3-L1 adipocytes. Chronic exposure of the adipocytes to 50 M AA was demonstrated to alter both basal-and insulin-stimulated glucose uptake and render the cells insulin resistant. The AA treatment specifically reduced the GLUT4 content of both the plasma and intracellular membranes, while GLUT1 content increased slightly. Consistent with the down-regulation of the protein, GLUT4 mRNA levels decreased to 10% of the initial content. Mechanistically the decrease was determined to be the result of both transcriptional down-regulation (50% of control) and a destabilization of the message (GLUT4 mRNA t1 ⁄2 decreased by ϳ43%). These data suggested that AA, in a manner similar to insulin (Flores-Riveros et al., 1993a) and TNF Pekala, 1991, 1992), down-regulates expression of the insulinresponsive glucose transporter in adipocytes.
The oxidation of AA to very potent biological molecules is mediated by three different enzyme systems to include: the cyclooxygenase, lipoxygenase, and cytochrome P-450 epoxygenase. Thus, it became important to determine if oxidation of AA was necessary for its effect on GLUT4 or whether the observed effects were mediated by alternative pathways. In the current report we demonstrate that mechanistically fatty acids may act via two independent mechanisms, only one of which requires an oxidized metabolite of AA.

EXPERIMENTAL PROCEDURES
Materials-Dulbecco's modified Eagle's medium was purchased from Life Technologies, Inc. Fetal bovine serum was purchased from Hy-Clone (Logan, UT) and used at a 1:10 dilution in Dulbecco's modified Eagle's medium. Based on specifications provided by HyClone, the mean AA content of culture medium containing 10% fetal bovine serum would be ϳ3 M. Radiolabeled compounds were obtained from DuPont NEN. Hybond-N blotting membrane was purchased from Amersham Corp. Deoxyribonucleotides were obtained from Pharmacia Biotech Inc. Klenow fragment and TRIzol Reagent was obtained from Life Technologies, Inc. The polyclonal antiserum against PPAR␥2 was developed against a peptide containing amino acids 284 -298 as described by the clone; it was obtained from James Stiehr of Affinity Bioreagents Inc. Tumor necrosis factor-␣ was the generous gift of Biogen (Cambridge, MA). The specific activity was 9.6 ϫ 10 6 units/mg protein, based on a cytotoxicity assay using L929 cells. Prostaglandin E 2 -Monoclonal and cAMP Enzyme Immunoassay Kits were purchased from Cayman Chemical. All fatty acids and prostaglandins were purchased from Cayman Chemical. All other chemicals, unless otherwise stated, were of molecular biology grade and purchased from Sigma.
3T3-L1 Cell Culture-The murine 3T3-L1 cells used in this study were originally obtained from Dr. Howard Greene, Harvard University, Boston, MA. Cells were cultured, maintained, and differentiated as described previously (Tebbey et al., 1994). The cells were maintained for 10 days post-differentiation and then treated with the indicated agents for various times prior to RNA isolation.
Fatty Acids-All fatty acids used were absorbed onto diatomaceous earth and subsequently complexed to essentially fatty acid-free bovine serum albumin (BSA) to yield a FA:BSA ratio of ϳ1.4:1 (Tebbey and Buttke, 1992;Buttke et al.,1989). FA/BSA was added to cells from a 2.5 mM stock solution to yield a final concentration of 100 M. RNA Isolation and Northern Blot Analyses-Total RNA was isolated by extraction with guanidine isothiocyanate and centrifugation through 5.7 M cesium chloride (Chirgwin et al., 1979) or by the TRIzol Method (Life Technologies, Inc.). Northern analyses were performed as described previously (Stephens and Pekala, 1991).
DNA Probes-GLUT4, a 2.8-kilobase pair EcoRI fragment encoding the 3T3-L1 homolog of the adipose/muscle (insulin-responsive) glucose transporter (Kaestner et al., 1989); ␤-actin, a 1.9-kilobase pair HindIII fragment obtained from Dr. D. W. Cleveland (Cleveland et al., 1980). Enzyme Immunoassay for PGE 2 -Cells were treated with 100 M AA/BSA for various times. At the indicated times 100-l aliquots of medium were taken and analyzed for PGE 2 content using an enzyme immunoassay kit for PGE 2 (Cayman Chemical) based on the ELISA method.
Enzyme Immunoassay for cAMP-Cells were treated with 100 M AA/BSA for various times and at the indicated times the cells were washed in phosphate-buffered saline and then scraped in 1 ml of 10% trichloroacetic acid. Cell debris was pelleted and cAMP was then extracted from the supernatant with an equal volume of 3:1 mixture of Freon/tri-n-octylamine. The upper aqueous phase was collected (ϳ500 l) and then assayed for cAMP content using an enzyme immunoassay kit for cAMP (Cayman Chemical) based on the ELISA method.
Sphingomyelin Assay-Cells were labeled with [ 3 H] choline chloride (specific activity of 1 mCi/ml) at 0.5 Ci/ml for 6 days after initiation of differentiation, and the cells were grown in normal media for the remaining 3 days. On day 9 the cells were treated with either 100 M AA or 5 nM TNF for the indicated times, the lipids extracted, and labeled sphingomyelin quantified by the bacterial sphingomyelinase method as described by Jayadev et al. (1994).
Preparation of Nuclear Extract-After the indicated treatments for various times, the cells were pelleted in phosphate-buffered saline for 10 min at 3000 rpm. The pellet was then resuspended in 3 ml of M3 lysis buffer (250 mM sucrose, 25 mM Tris, pH 7.8, 1.1 mM MgCl 2 , 0.2% Triton X-100, 0.5 mM phenylmethylsulfonyl fluoride, 10 mM EDTA, 10 mg/ml leupeptin, and 1 mg/ml aprotinin) and centrifuged at 10,000 rpm for 15 min. The nuclei were extracted using 250 l of cold M4 extraction buffer (20 mM HEPES, pH 7.8, 0.4 M KCl, 2 mM dithiothreitol, 20% glycerol, 0.5 mM phenylmethylsulfonyl fluoride, 10 mM EDTA, 10 mg/ml leupeptin, and 1 mg/ml aprotinin) for 30 min and then the nuclear debris and DNA pelleted at 12,000 rpm for 15 min. Nuclear extracts were assayed for protein concentrations by the method of Bradford (1976).
Electrophoretic Mobility Shift Assay (EMSA)-EMSAs were performed using the direct repeat half-site PPRE (5Ј-AATTTCGAG-AACGTGACCTTTGTCCTGGTCCAGCT-3Ј). Briefly, ϳ100 ng (60,000 cpm) of a 32 P-labeled double-stranded DNA oligomer was incubated with 5 g of nuclear extract for 15 min at room temperature in a total volume of 20 l of reaction buffer (25 mM Tris, pH 7.8, 0.5 mM EDTA, 88 mM KCl, 1 mM dithiothreitol, 150 g of poly(dI⅐dC)-poly(dI⅐dC), 0.05% Triton X-100, and 12.5 g/ml salmon sperm DNA). Reaction mixtures were subjected to electrophoresis on a 5% nondenaturing polyacrylamide gel at 4°C, 40 amps (200 V), for ϳ2 h. The gels were dried and exposed to x-ray film for 12 h at Ϫ80°C.
Statistical Analyses-The results of these experiments were analyzed using the Sigma-Stat package (Jandel Scientific Software, San Rafael, CA). Comparisons of means among groups were tested for significant differences by a Newman-Keuls procedure following oneway analysis of variance. The level of statistical significance chosen for these experiments was p Ͻ 0.05.

Inhibition of Cyclooxygenase Abolishes
Arachidonic Acid-induced GLUT4 mRNA Suppression-AA is rapidly converted to a number of eicosanoids which express potent physiological properties. In order to identify involvement of either the lipo-or cyclooxygenase in the conversion of AA to a metabolite active in the suppression of GLUT4 gene expression, inhibitors of the two enzyme systems were used. Pretreatment of the 3T3-L1 adipocytes with nordihydroguaiaretic acid (NDGA), a selective inhibitor of the lipoxygenases (IC 50 ϭ 0.2, 30, and 30 M for 5-, 12-, and 15-LO, respectively, as opposed to 100 M for the cyclooxygenase; Tobias and Hamilton, 1978) for 1 h at 50 M did not significantly alter the AA-induced suppression of GLUT4 mRNA content (Fig. 1). In contrast, pretreatment with 50 M indomethacin, which inhibits the cyclooxygenase (IC 50 ϭ 0.1 M) selectively over the lipoxygenases (IC 50 Ͼ 100 M; Tobias and Hamilton, 1978), for 1 h completely abolished the AA-induced decrease of GLUT4 mRNA content, consistent with conversion of AA to a prostanoid which in turn mediated suppression. Similar results were observed using the cyclooxygenase inhibitor ibuprofen (data not shown). While these results were rather compelling, a third oxidative pathway was considered. The cytochrome P-450 epoxygenase pathway has only been considered a major pathway in liver and kidney and a minor pathway in occular and pituitary tissues (Fitzpatrick and Murphy, 1988). Its existence in adipose tissue, to our knowledge, has not been described. However, we attempted to block any potential metabolism through this enzyme system using the inhibitor SKF-541. At concentrations of up to 50 M, this inhibitor had no effect on the AA-mediated suppression of GLUT4 mRNA. In addition, NDGA exhibits an IC 50 value of 15 M for AA-specific epoxygenase (Capdevila et al., 1988) and should have blocked the AA-induced GLUT4 down-regulation had an epoxygenase metabolite been responsible for the regulation; as seen from the data in Fig. 1, this did not occur. These results strongly indicate a cyclooxygenase metabolite rather than AA itself as the suppressive agent in the down-regulation of GLUT4 message.
PGE 2 Suppresses GLUT4 mRNA Levels-Of the cyclooxygenase metabolites, PGE 2 and prostacyclin (PGI 2 ) are the major prostaglandins formed in the fat cell (Axelrod and Levine, 1981;Richelsen, 1987Richelsen, , 1992. Given the inhibition of indomethacin and ibuprofen on AA-induced GLUT4 mRNA suppression, we examined whether PGE 2 could elicit the same effect as AA. The results shown in Fig. 2, indicate that exposure of the cells for 12 h to 10 M PGE 2 resulted in a marked decrease (ϳ70%) in GLUT4 mRNA content. The effect was significantly (p Ͻ 0.05%) more potent than either AA (ϳ55% decrease) or tumor necrosis factor-␣ (TNF) (ϳ60% decrease). Treatment of the cells with the nonmetabolizable analog of prostacyclin, carbaprostacyclin, for 12 h at 5, 10, or 25 M had little or no effect on GLUT4 mRNA levels (data not shown), suggesting the specificity of PGE 2 .
To further establish the role of PGE 2 in the regulation of GLUT4, as well as obtain support for the NDGA inhibitor data, we treated the cells with the hydroperoxy products of the lipoxygenase pathway, the 5(S)-, 12(S)-, and 15(S)-HPETEs for 12 h at 5 and 10 M (Fig. 3). The data clearly demonstrated that, in contrast to PGE 2 , these lipoxygenase metabolites had no effect on GLUT4 mRNA expression.
To determine if AA was being converted to PGE 2 by the adipocytes, we measured PGE 2 levels present in the media at various times after the cells were treated with AA, using an ELISA. As shown in Fig. 4, within 1 h of AA addition to the cells, PGE 2 content increased 10-fold above basal levels. A maximum 50-fold increase above basal was observed 7 h after AA addition (Fig. 4). The duration of the PGE 2 peak was relatively short, lasting only ϳ1 h and then returning slowly to basal levels over the next 15 h. These data suggest that AA is being converted via the cyclooxygenase pathway to PGE 2 , the apparent active metabolite mediating GLUT4 suppression.
Cyclic AMP Increases in the 3T3-L1 Adipocytes with AA Treatment-PGE 2 functions to increase intracellular cAMP levels in various cells by stimulation of adenylyl cyclase in an autocrine fashion (Davies and MacIntyre, 1992). In adipocytes, PGE 2 has been demonstrated to exhibit a concentration dependence with respect to its inhibitory or stimulatory effects on adenylate cyclase (Kather andSimon, 1977, 1979 1982). In intact adipose cells PGE 2 inhibited isoproternol-stimulated lipolysis at nanomolar concentrations, while exposure of cells to concentrations in excess of 1 M resulted in cAMP accumulation and stimulation of lipolysis (Kather, 1981). To determine if cAMP content increases with AA treatment, we measured cAMP levels present in the cytosol at various times after the cells were treated with AA using an ELISA. As shown in Fig. 5, cAMP levels increased 50-fold above the basal level within 3 h of AA treatment. The increase in cAMP occurred consistent with stimulation of adenylate cyclase by PGE 2 (compare Figs. 4 and 5). Unlike the increase in PGE 2 , the rise in cAMP was relatively sharp, and the levels remained high for less than 30 min, potentially due to the desensitization of the enzyme. These data are consistent with AA metabolism to PGE 2 which acts in an autocrine fashion to increase cAMP levels.
Specificity of the AA Regulatory Effect-Specificity of AA in suppressing GLUT4 mRNA levels was investigated using a series of fatty acids as well as a nonmetabolizable analog of AA.
To obtain a maximal response, based on preliminary studies, the cells were exposed to the fatty acid-albumin complex at a final concentration of 100 M for 24 h. The results, displayed in Fig. 6, indicate that all fatty acids examined, stearic acid (18:0), oleic acid (18:1), linoleic acid (18:2), linolenic acid (18:3), ␥-linolenic acid (␥18:3), mead fatty acid (20:3), AA (20:4), eicosapentaenoic acid (20:5), and the acetylenic analog of AA, ETYA, were able to suppress GLUT4 mRNA levels by at least 50% with AA providing the maximum suppression of 80%. While 18:2 and ␥ 18:3 can be converted to AA and potentially lead to increased cAMP, none of the other fatty acids tested exhibit this potential. To confirm this, we examined the potential for 18:0, 18:1, and ETYA to increase cAMP levels; our data indicated that cAMP did not increase over the 5-h time frame of the experiment (mean Ϯ S.D. of the [cAMP] over the 5-h time course, control: 178 Ϯ 32 pmol/ml; 18:0-, 18:1-, and ETYAtreated: 149 Ϯ 28, 139 Ϯ 19, and 159 Ϯ 21 pmol/ml, respectively). Interestingly ETYA, which inhibits AA uptake, AA-specific and nonspecific acyl-CoA synthetases, the cyclooxygenase, and all lipoxygenases, PLA 2 as well as the cytochrome P-450 epoxygenase, and thus cannot be converted to PGE 2 , was as effective as AA in suppressing GLUT4 mRNA accumulation. In addition, the suppression of GLUT4 mRNA accumulation could not be blocked by addition of indomethacin to the cultures. These studies are consistent with the premise that fatty acids may regulate GLUT4 mRNA content without the requirement for further metabolism and are consistent with the presence of a second signal transduction pathway.
AA Does Not Induce Sphingomyelin Turnover- Jayadev et al. (1994) have identified AA as one of several fatty acids that can activate the sphingomyelinase in HL-60 leukemia cells. This results in generation of free ceramide and initiation of a signal transduction cascade that potentially could control GLUT4 gene expression. To determine if this regulatory pathway was viable in the adipocytes, we examined the ability of AA to activate the sphingomyelinase and initiate sphingomyelin turnover. As shown in Fig. 7, after addition of AA to the cells, no significant hydrolysis of sphingomyelin was detected. TNF activation of the sphingomyelinase at a single 40-min time FIG. 4. Arachidonic acid induction of PGE 2 as detected by ELISA. Fully differentiated 3T3-L1 adipocytes were treated with 100 M AA, and aliquots of the media were removed at various times and tested for the presence of PGE 2 . The data are plotted as the mean picomoles of PGE 2 /ml Ϯ S.D. The graph represents two independent determinations. Where not indicated with an error bar, the error of the data lies within the data point itself.  5, 6, 7), and 10 M 5, 12, and 15 (S) HPETEs (lanes 8, 9, 10). Lane 1, No treatment. Total cellular RNA was isolated, and 20 g of each RNA sample was incubated with complementary radiolabeled GLUT4 RNA transcript and then subjected to Ribo-nuclease Protection analysis. The 182-base pair protected fragment is indicated by the arrow.
FIG. 5. Arachidonic Acid induction of cAMP as detected by ELISA. Fully differentiated 3T3-L1 adipocytes were treated with 100 M AA, and at the indicated times the cells were harvested and cAMP was extracted as described under "Experimental Procedures." The content of cAMP was then determined by an ELISA. The data are plotted as the mean picomoles of cAMP/ml Ϯ S.D. The graph represents two independent determinations. Where not indicated with an error bar, the error of the data lies within the data point itself. point resulted in a 50% decrease in sphingomyelin and was utilized as a positive control. The data suggest that in the 3T3-L1 adipocytes, AA is not mediating activation of the sphingomyelinase and generation of free ceramide.
Involvement of Peroxisome Proliferator-activated Receptors (PPARs)-The newest members of the nuclear hormone receptor superfamily are the PPARs, which are known to be activated, in addition to the classic peroxisome activators such as clofibrate, by long chain saturated (Gottlicher et al., 1992;Gulick et al., 1994) and polyunsaturated fatty acids (Tontonoz et al., 1994a;Keller et al., 1993;Gottlicher et al., 1992). Thus, a predicted activity profile for an effect mediated by fatty acid activation of a PPAR would be similar to that displayed in Fig.  6 and discussed in the previous section. To establish a role for PPARs in the regulation of GLUT4 mRNA expression, fully differentiated 3T3-L1 cells were exposed to 0.5 and 1.0 mM clofibric acid for 12 h and GLUT4 mRNA levels were analyzed (Fig. 8). In a manner similar to AA and ETYA, 0.5 and 1.0 mM clofibric acid decreased GLUT4 mRNA content ϳ35 and 75%, respectively (the average area by quantification of phosphorimage scans Ϯ experimental variation for: control, 4450 Ϯ 273; 0.5 mM clofibrate, 2850 Ϯ 140; 1.0 mM clofibrate, 1187 Ϯ 112). These data suggest that PPAR activation may result in GLUT4 mRNA suppression. PPAR␥2, the predominant isoform expressed in adipose tissue (Tontonoz et al., 1994a;1994b) has been shown to regulate the expression of adipocyte-specific genes and is activated by a diverse group of compounds including ETYA and fatty acids. In order to determine the potential for PPAR␥2 involvement in the fatty acid-induced GLUT4 mRNA suppression, we performed two experiments using a DNA gel mobility shift assay (Fig. 9). Oligonucleotides containing the classic PPRE (TGACCTTTGTCCT) from the promoter of the fatty acyl-CoA oxidase gene were end-labeled with [␣-32 P]dATP and then incubated with 10 g of nuclear extracts. In the first experiment (Fig. 9A), nuclear extracts were prepared from the 3T3-L1 adipocytes and then treated with 50 M AA, ETYA, and 10 M retinoic acid. Separation of the complexes on a 5% nondenaturing polyacrylamide gel demonstrated that addition of AA or ETYA to the nuclear extracts results in a loss of binding of the PPAR to its responsive element. In the second experiment, nuclear extracts were prepared from cells that had been treated with 100 M AA for 2 and 6 h or 100 M ETYA for 12 h. Binding reactions were performed as described above. The results (Fig. 9B) demonstrate that treatment of the intact cells with these fatty acids markedly diminished protein binding to the PPRE. A partial supershift, performed with an antibody to PPAR␥2, demonstrated that the protein-DNA complex was a heterodimer with one partner being PPAR␥2 (data not shown). The data from the two experiments suggest that both AA and its nonmetabolizable analog are capable of disrupting the occupancy of a PPAR and that neither protein synthesis nor oxidation to PGE 2 is necessary for the alteration of binding.

DISCUSSION
To address the mechanism by which AA can suppress GLUT4 gene expression, we have investigated the metabolism and the signal transduction properties of AA in the 3T3-L1 adipocytes. Our studies have demonstrated that the major metabolic fate of exogenously added AA is the oxidative metabolism through the cyclooxygenase pathway to a prostanoid. These conclusions are based on the data demonstrating that the effect of AA on GLUT4 can be blocked by indomethacin, a cyclooxygenase inhibitor, while NDGA, a lipoxygenase inhibitor, cannot inhibit the effect of AA. Of the prostanoids derived from AA, isolated adipocytes have been demonstrated to produce only the prostaglandins, PGE 2 and PGI 2 (prostacyclin) in considerable amounts (Axelrod and Levine, 1981;Richelsen, 1987). To determine if either of these compounds contributed to the suppression of GLUT4, the fully differentiated 3T3-L1 adipocytes were exposed to PGE 2 and carbaprostacyclin, the chemically stable analog of prostacyclin. Prostaglandin E 2 treatment resulted in an 70% decrease in GLUT4 mRNA levels, while carbaprostacyclin had no effect on GLUT4 mRNA content. Quantification of the conversion of AA to PGE 2 demonstrated a 50-fold increase in PGE 2 released into the media within 3 h of AA addition (Fig. 3). These data, demonstrating the time course and magnitude of PGE 2 formation, are consistent with PGE 2 synthesis as being the major metabolic fate of exogenously added AA. Moreover, they suggest that PGE 2 may be an intermediate in regulation of GLUT4 gene expression.
In a number of systems, the oxidative metabolism of AA appears to be a requirement to derive the metabolite capable of regulating specific gene expression. In Swiss 3T3 fibroblasts, AA was shown to induce c-fos and Egr-1 mRNA through formation of PGE 2 and subsequent activation of protein kinase C (Danesch et al., 1994). Lipoxygenase metabolites have also been shown to stimulate the accumulation of c-fos and c-jun mRNA in human monocytes while they increase c-fos accumulation in quiescent TA1 cells (Haliday et al., 1991). In both cases, message accumulation occurs through increased gene transcription (Stankova and Rola-Pleszczynski, 1992;Haliday et al., 1991). While underlying mechanisms have not as of yet been identified, these reports demonstrate a role for AA and its metabolites in the regulation of specific genes.
Prostanoids are local hormones that once released from the cell bind to cell surface receptors to act in an autocrine or paracrine fashion. Depending on the cell type, there are three distinct receptors for PGE 2 . Ligand occupation of the appropriate receptor can lead to activation or inhibition of adenylate cyclase, each mediated through a specific G-protein, while occupation of the third class of receptors can stimulate Ca 2ϩ mobilization and subsequent activation of protein kinase C (Davies and MacIntyre, 1992). Interestingly, the 3T3-L1 adipocytes have recently been shown to lack the Ca 2ϩ -activated protein kinase C isoforms (McGowan et al., 1996) and thus a mechanism dependent on PGE 2 activation of protein kinase C is ruled out. Our data are consistent with a cAMP-mediated mechanism with cAMP levels rising ϳ50-fold above basal (10 M) within 3 h of AA addition. The decrease in cAMP content, with levels returning to basal within ϳ30 min while PGE 2 levels remained markedly elevated, suggest a rapid desensitization of the PGE 2 receptor to adenylate cyclase coupling mechanism.
We (Stephens and Pekala, 1992) and others (Ezaki et al., 1993;Flores-Riveros, et al., 1993b) have demonstrated the ability of cAMP to suppress transcription of the GLUT4 gene, leading to a decrease in the content of GLUT4 mRNA. Stephens and Pekala (1992) demonstrated that the effect of 8-bromo-cAMP on GLUT4 transcription was rapid, with a 40% decrease in rate detected within 1 h and a 52% decrease within 4 h. With respect to both magnitude and temporal considerations, these results were very similar to those we have reported for AA, where a 50% decrease in GLUT4 gene transcription occurred in response to AA treatment (Tebbey et al., 1994). Studies by Flores-Riveros et al. (1993b) indicated that the regulatory element mediating transcriptional repression by cAMP resides in the proximal promoter of the GLUT4 gene between positions Ϫ469 and Ϫ78. Thus, based on these observations it is likely that a cAMP-dependent protein kinase is involved in the transcriptional repression of the GLUT4 gene where an interaction between the cis-regulatory element and the activity of a transacting factor is regulated by cAMP-mediated phosphorylation.
Examination of a diverse family of fatty acids for the specificity of the AA effect demonstrated that suppression of GLUT4 was not limited to AA (Fig. 6). Perhaps the strongest evidence for an alternative pathway is that ETYA, a nonmetabolizable AA analog reported to be a potent inhibitor of AA transport and metabolism (Tobias and Hamilton, 1978), is nearly as potent as AA in its ability to suppress GLUT4 mRNA accumulation. These data would suggest that the fatty acids can function as a regulator of GLUT4 mRNA expression by a pathway that does not involve oxidative metabolism.
Our previous studies with AA demonstrated that in addition to a suppression of transcription, exposure of the adipocytes to AA resulted in a destabilization of the GLUT4 mRNA (t1 ⁄2 decreased from 9.3 to 4.5 h). However, in the studies described above (Stephens and Pekala, 1992;Ezaki et al., 1993) no effect was observed on the stability of the GLUT4 mRNA when the cells were exposed to 8-bromo-cAMP, thereby localizing the effect on transcription to cAMP. These data taken collectively suggest that AA must initiate a second signal transduction cascade that is responsible for the alteration of GLUT4 mRNA stability. In an attempt to further define this issue, we examined whether ETYA or 18:1 could alter GLUT4 mRNA stability. After exposure of the cells to ETYA for 24 h, the half-life of the GLUT4 mRNA was determined to be 8. 1 Ϯ 1.1 h (n ϭ 3). This did not test significantly different than control. However, incubation of the cells in the presence of 18:1 resulted in a stability Nuclear extracts were then prepared as described under "Experimental Procedures," and 10 g of protein from each sample were incubated with radiolabeled PPRE and then subjected to EMSA analysis. decrease similar to that observed for AA (5.5 Ϯ 0.5 h and 4.2 Ϯ 0.3 h, respectively). Thus, while ETYA appears to exert its effect solely by means of transcriptional mechanisms, the metabolizable fatty acids appear to also influence mRNA stability.
Fatty acid levels fluctuate in normal metabolism as well as in certain pathological diseases and are appropriately considered to serve as biological effectors as well as metabolic substrates. Arachidonic acid has become recognized as a novel second messenger regulating cell growth through activation of various enzyme systems such as neutral sphingomyelinase (Jayadev et al., 1994), protein kinase A (Doolan and Keenan, 1994), protein kinase C (Bell and Burns, 1991), and the platelet-derived growth factor receptor (Tomaska and Resnick, 1993). In addition, AA as well as its metabolites have been shown to regulate gene expression either directly (Danesch et al., 1994) or as mediators of various stimuli including tumor necrosis factor-␣ and growth factors (Jayadev et al., 1994). Long chain fatty acids have also been shown to activate the recently cloned PPARs, the newest members of the nuclear hormone receptor superfamily. Interestingly, ETYA, which suppressed GLUT4 mRNA as strongly as AA, is one of the most potent activators of this class of nuclear receptors. Signaling through these receptors may provide the second mechanism by which AA can function to suppress GLUT4 gene expression. The fact that the classic PPAR activator, clofibrate, can also suppress GLUT4 mRNA levels further supports a potential role for activation of PPARs in the regulation of GLUT4 mRNA accumulation. The observation that AA and ETYA can disrupt PPAR binding to its response element indicates that these fatty acids are capable of regulating occupation of the PPARs in the 3T3-L1 adipocytes. Whether or not this AA-induced disruption of PPAR binding to its response element is involved in GLUT4 mRNA suppression has yet to be elucidated. To date, a perfect PPRE has not been found in the GLUT4 promoter, and as such, activation of a PPAR by a fatty acid or clofibrate may play an indirect role in GLUT4 mRNA suppression.
In addition to regulation of GLUT4 gene expression by a direct mechanism, the PPAR may potentially regulate indirectly by altering expression or activation of another transcription factor. The CCAAT/enhancer-binding protein (C/EBP␣) Herrera et al., 1989), a transcription factor demonstrated to bind to and activate the GLUT4 gene promoter upon differentiation of the 3T3-L1 adipocytes , has been demonstrated to act synergistically with PPAR␥2 in the development of adipose cells from uncommitted mesodermal precursers (Tontonoz et al., 1994b). Thus, perhaps an essential interaction between the PPAR␥2 and C/EBP␣ necessary for GLUT4 expression is altered on exposure to the fatty acids. This remains to be resolved.
In summary, these studies demonstrate that in 3T3-L1 adipocytes AA/fatty acids regulate GLUT4 gene expression by potentially two independent mechanisms: 1) via oxidative metabolism to the cyclooxygenase metabolite PGE 2 and the subsequent elevation of cAMP levels, based on our previous studies it is likely that the increase in cAMP represents the potential mechanism for suppression of GLUT4 transcription. We note, however, that as of yet we have not ruled out a mechanism by which PGE 2 functions to suppress GLUT4 independent of the increase in cAMP and 2) via nonoxidized fatty acid with the potential involvement of a PPAR. In addition, the demonstration that AA did not activate sphingomyelinase confirms the absence of another potential fatty acid activated signaling pathway in the adipocytes.