Glucose Down-regulates the Expression of the Peroxisome Proliferator-activated Receptor-α Gene in the Pancreatic β-Cell*

To better understand the action of glucose on fatty acid metabolism in the β-cell and the link between chronically elevated glucose or fatty acids and β-cell decompensation in adipogenic diabetes, we investigated whether glucose regulates peroxisomal proliferator-activated receptor (PPAR) gene expression in the β-cell. Islets or INS(832/13) β-cells exposed to high glucose show a 60–80% reduction in PPARα mRNA expression. Oleate, either in the absence or presence of glucose, has no effect. The action of glucose is dose-dependent in the 6–20 mm range and maximal after 6 h. Glucose also causes quantitatively similar reductions in PPARα protein and DNA binding activity of this transcription factor. The effect of glucose is blocked by the glucokinase inhibitor mannoheptulose, is partially mimicked by 2-deoxyglucose, and is not blocked by the 3-O-methyl or the 6-deoxy analogues of the sugar that are not phosphorylated. Chronic elevated glucose reduces the expression levels of the PPAR target genes, uncoupling protein 2 and acyl-CoA oxidase, which are involved in fat oxidation and lipid detoxification. A 3-day exposure of INS-1 cells to elevated glucose results in a permanent rise in malonyl-CoA, the inhibition of fat oxidation, and the promotion of fatty acid esterification processes and causes elevated insulin secretion at low glucose. The results suggest that a reduction in PPARα gene expression together with a rise in malonyl-CoA plays a role in the coordinated adaptation of β-cell glucose and lipid metabolism to hyperglycemia and may be implicated in the mechanism of β-cell “glucolipotoxicity.”

glucose-stimulated secretion in this animal model of adipogenic non-insulin-dependent diabetes mellitus (4). The mechanisms causing ␤-cell decompensation and failure in non-insulin-dependent diabetes mellitus are beginning to be understood. What is emerging from the recent literature is that, in addition to genetic predisposition, changes in lipid partitioning may play an important role in the development of ␤-cell failure (1,2). Thus, triglyceride (TG) overload in the ␤-cell correlates with altered glucose-induced secretion and apoptotic cell death (5). In addition, pharmacological (6,7) or molecular biological (8) strategies aiming at reducing islet fat overload improve ␤-cell function. Altered lipid metabolism, rather than glucose metabolism, appears to be implicated in this so-called ␤-cell "lipotoxicity" via a process that apparently does not involve a Randle cycle, in which enhanced fat oxidation reduces glucose metabolism (9). This view is in accordance with the observation that enhanced ␤-cell lipid synthesis and esterification characterize the ␤-cell of ZDF rats (10). Also consistent with the possibility that changes in lipid metabolism are central to the process of ␤-cell "glucolipotoxicity" (1) is that ZDF rats show enhanced lipogenesis in association with increased levels of the lipogenic enzymes acetyl-CoA carboxylase and fatty acid synthase and the esterifying enzyme glycerol-3-phosphate acyltransferase (10). Interestingly, glucose causes a coordinated induction of lipogenic genes and promotes lipid esterification and TG deposition in ␤(INS) cells (11).
Because both elevated glucose and FFA probably contribute to the progressive demise of the ␤-cell in non-insulin-dependent diabetes mellitus (1), there is much interest in defining the genes whose expression are altered by glucose or fatty acids in islet tissue. We know that both glucose and FFA regulate the expression level of ␤-cell genes encoding key regulatory enzymes in the glycolytic (12)(13)(14)(15), lipogenic (11), and fat oxidation (16) pathways and of the insulin gene (17,18). However, little is known about the action of these calorigenic nutrients on the expression of transcription factors in the ␤-cell or other cell types. This may prove particularly relevant to our understanding of obesity-associated diabetes, because transcription factors are expected to translate variations in the nutritional state to late phenotypic changes. Noteworthy is the fact that to date five of the six identified maturity onset diabetes (MODY) genes are transcription factors regulating the insulin gene and vari-ous intermediary metabolism genes (19). Our recent work indicates that both glucose and FFA induce the transcriptional activation of several immediate early response genes in INS ␤-cells and rat islets, in particular the transcription factors c-fos, nur 77, and members of the jun family (20 -22). Glucose also increases the expression of the IDX-1 gene (23), which encodes a key ␤-cell transcription factor implicated in ␤-cell differentiation and the control of the insulin gene (24); by contrast, palmitate, in the presence of elevated glucose, reduces IDX-1 expression in rat islets (25).
Peroxisome proliferator-activated receptors (PPARs) are fatty acid sensors, which transduce nutritional stimuli into changes in gene expression. The PPAR␣ isoform controls many genes of fat metabolism, in particular those implicated in mitochondrial and peroxisomal ␤-oxidation and fatty acid synthesis and transport as well as uncoupling proteins (26). The function of the PPAR␤ isoform is poorly understood. PPAR␥ has a more restricted tissue distribution than the ␣ and ␤ isoforms. It plays an essential role in fat cell differentiation and lipid storage and may be a "thrifty" gene (27). Little is known about the role of PPARs in ␤-cell function. The three isoforms are expressed in purified ␤ and INS cells 2 as well as in rat islets (28). PPAR␣ mRNA expression has been reported to be induced by high concentrations of FFA in rat islets (28). Interestingly, the antidiabetic PPAR␥ ligand troglitazone prevents diabetes in ZDF rats as well as TG deposition and ␤-cell lipoapoptosis both in vivo (29) and ex vivo (6).
Because both excessive glucose and fatty acids may be implicated in the progressive ␤-cell failure in diabetes and because altered lipid partitioning, a process controlled by PPARs, probably contributes to this process, this study was designed to gain insight into glucose and fatty acid regulation of PPAR␣ expression in the ␤-cell.

EXPERIMENTAL PROCEDURES
Materials-Pancreatic islets were isolated using collagenase (type IV) from Worthington (Lakewood, NJ) and Ficoll DL-400 from Sigma. Reverse transcriptase was purchased from Rote Molecular Biochemical (Laval, Quebec, Canada), Taq polymerase was from Life Technologies, Inc., and Pd(N) 6 hexamer was from Amersham Pharmacia Biotech. Protran nitrocellulose membranes for protein analysis were from Schleicher & Schuell (Dassel, Germany), and the bicinchoninic acid protein assay was from Pierce. The enhanced chemiluminescence detection kit for Western blot (ECL) and T4-kinase were from Amersham Pharmacia Biotech. The horseradish peroxidase-conjugated goat antirabbit immunoglobulin antibody and zeta probe membrane were obtained from Bio-Rad.
Pancreatic Islet Isolation and Cell Culture-Islets were isolated from Wistar rats (Charles River, St.-Constant, Quebec, Canada) weighing ϳ200 g by collagenase digestion of the total pancreas and subsequent separation on discontinuous Ficoll DL-400 gradients as described previously by Gotoh et al. (30). Islets and INS-1 (31) and INS(832/13) (32) cells were kept in culture at 37°C in a humidified atmosphere containing 5% CO 2 . At the end of the isolation step, islets were maintained at 15-20 islets/ml in regular RPMI 1640 medium containing 11 mM glucose supplemented with 10% fetal calf serum, 10 mM Hepes (pH 7.4), 1 mM sodium pyruvate, and 50 M ␤-mercaptoethanol. INS(832/13) cells were seeded in 60-mm tissue culture dishes (6 ϫ 10 6 cells/dish) and cultured in complete RPMI medium. On the next day, islets and INS(832/13) cells were preincubated for 24 h in complete culture medium containing 3 mM glucose. Finally, they were incubated in complete culture medium at various glucose concentrations in the absence or the presence of oleate (0.4 mM) bound to 0.5% albumin (9) for the indicated periods of time. The experiments dealing with the regulation of PPAR␣ gene expression were carried out with the new clonal cell line INS(832/ 13) derived from INS-1 cells. The action of chronic elevated oleate and glucose on ␤-cell metabolism and insulin secretion was carried out with the original INS-1 cells at passages below 85. Thus, we performed this work just prior to the availability of the new clone INS(832/13). Albumin-bound oleate was prepared by stirring the fatty acid Na salt at 45°C with defatted bovine serum albumin (BSA)(Sigma Fraction V). After adjustment of the pH to 7.4, the solution was filtered through a 0.22-m filter, and the fatty acid concentration was measured using a NEFAC kit (Wako Chemicals GmbH). For incubations longer than 24 h, the medium was changed daily to maintain a constant concentration of fatty acid. The final concentration of BSA in the culture medium was 0.5%.
RNA Extraction and Reverse Transcriptase-PCR Analysis-Total RNA was extracted from 50 islets and INS(832/13) cells by the guanidinium thiocyanate/phenol/chloroform extraction method (33) using 20 g of yeast tRNA as a carrier for islet RNA. First strand cDNA was generated from either the total amount of islet RNA or from 5 g of INS(832/13) cell RNA in 50 l (final volume) of a buffer containing the oligonucleotides Pd(N) 6 (34). 3 l (PPAR␣) and 1 l (␤-actin) of the reverse transcription mixtures were amplified in a final volume of 50 l with specific primers for the following rat cDNAs. Amplification of rat PPAR␣ (28), uncoupling protein 2 (UCP2) (35), acyl-CoA oxidase (ACO) (6), and carnitine palmitoyltransferase I (CPT I) (6) was done using the primers and experimental conditions described in the cited references. Amplification of rat ␤-actin was done using the following parameters and primers: 92°C for 30 s, 63°C for 1 min, and 72°C for 1 min during 25 cycles; 5Ј-GTGCCCATCTATGAGGGTTACGCG-3Ј (sense); 5Ј-GGAACCGCTCATTGCCGATAGTG-3Ј (antisense). The linearity of each PCR was tested by amplification of 500 ng of total RNA/reaction. It was linear from 20 to 45 cycles. Ten l of the PCR products were subjected to electrophoresis on 1.2% agarose gels followed by Southern blotting on a zeta probe membrane. The following specific radiolabeled probe was used for ␤-actin: 5Ј-TCATGCCATCCTGCGTCTGGACCT-3Ј. The probes for PPAR␣ (28), UCP2 (35), ACO (6), and CPTI (6) are indicated in the cited references. The hybridization signals were quantitated by scanning the autoradiograms.
Nuclear Extract Preparation and Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared from INS(832/13) cells as described (36). Briefly, cells were collected, washed twice with cold phosphate-buffered saline, and lysed in 500 l of ice-cold buffer A (10 mM Hepes (pH 7.8), 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, and 2 mM phenylmethylsulfonyl fluoride). After a 15-min incubation on ice, Nonidet P-40 (0.5% final concentration) was added, samples were gently vortexed, and nuclei from lysed cells were centrifuged at 1000 rpm for 2 min. Nuclei were washed twice with buffer A without Nonidet P-40. Nuclei were then lysed in buffer A containing 26% glycerol, 0.2 mM EDTA, and 450 mM NaCl. After a 15-min incubation on ice, nuclear extracts were centrifuged for 2 min at 1000 rpm. Aliquots (50 l) of the supernatants were frozen at Ϫ70°C. The protein concentration of the nuclear extracts was then measured. The oligonucleotides used to assess the DNA binding activities of PPAR␣ and adenovirus-5 major late transcription factor (MLTF) by electrophoretic mobility shift assays were purchased from Alpha-DNA (Montreal, Quebec, Canada). A 20mer sense oligonucleotide (5Ј-CGTCTGCCCTTTCCCCCTCT-3Ј) and a 20-mer antisense oligonucleotide (5Ј-GAGAAGAGGGGGAAAGGGCA-3Ј) containing the PPAR responsive element (PPRE) of the lipoprotein lipase promoter were annealed and used to assess PPAR␣ binding activity. A 28-mer sense oligonucleotide (5Ј-TAGGTGTAGGCCACGT-GACCGGGTGTTC-3Ј) (and its corresponding antisense) described to bind MLTF (37) was used to assess MLTF DNA binding activity. The conditions used for the electrophoretic mobility shift assays were as described (36). Briefly, 5 g of nuclear extracts were incubated with a radiolabeled probe (20,000 cpm per sample) for 15 min in 15 l of binding buffer (25 mM Hepes, 10% glycerol, 50 mM NaCl, 0.05% Nonidet P-40, and 1 mM dithiothreitol, pH 7.5) in the presence or absence of 2 l of anti-PPAR␣, anti-PPAR␤ (kindly provided by Dr. W. Wahli, University of Lausanne, Lausanne, Switzerland), and anti-PPAR␥ antisera (Calbiochem). A 100-fold molar excess of cold probe was used to assess the specificity of the binding of nuclear proteins. Samples were analyzed on 4% nondenaturating polyacrylamide Tris-glycine gels (37).
Western Blot Analysis-The proteins obtained from nuclear extracts were analyzed on 10% SDS-containing polyacrylamide gels. Proteins were then transferred to nitrocellulose filters, and PPAR␣ was detected using a rabbit antibody raised against this protein (dilution 1:500) and a horseradish peroxidase-conjugated goat anti-rabbit immunoglobulin antibody (dilution 1:8000). Detection was carried out with the enhanced chemiluminescence detection technique.
Malonyl-CoA and Insulin Secretion Measurements-Following cell culture for 3 days in 21-cm 2 Petri dishes at 5 or 25 mM glucose with or without 0.4 mM oleate bound to bovine serum albumin (0.5%), INS-1 cells were washed with phosphate-buffered saline and preincubated at 37°C for 30 min in Krebs-Ringer bicarbonate buffer (KRBB) with 25 mM Hepes (pH 7.4) (9), at 5 mM glucose, containing 0.07% BSA (Fraction 2 N. Voilley, F. Schuit, and M. Prentki, unpublished data. V, Sigma). Cells were then incubated for 30 min in fresh KRBB-Hepes at 5 or 25 mM glucose. Incubation media were discarded, and malonyl-CoA was extracted from cells with 10% trichloroacetic acid. After centrifugation of precipitated proteins, cell extracts were brought to pH 5-6 by successive ether extractions. Samples were lyophilized and stored at Ϫ70°C. Malonyl-CoA was assayed with a radioactive method using fatty acid synthase (38). For insulin secretion determinations, INS-1 cells were plated (10 5 cells/well) into 24-well plates. They were cultured and incubated as described above for malonyl-CoA measurements. The insulin concentration in the medium was determined by radioimmunoassay using rat insulin as a standard (9). Total cellular insulin content was measured after acid-ethanol (1.5% HCl, 75% ethanol) extraction.
Fatty Acid Metabolism and Triglyceride Measurements-Fatty acid oxidation was measured in INS-1 cells cultured in 21-cm 2 Petri dishes. Following a preexposure for 3 days at 5 or 25 mM glucose in the absence or presence of 0.4 mM oleate, cells were washed with phosphate-buffered saline and preincubated at 37°C for 30 min in KRBB-Hepes (pH 7.4) medium, at 5 mM glucose, containing 0.07% BSA. Cells were then incubated for 1 h at 37°C in 5 ml of fresh KRBB-Hepes at 5 or 25 mM glucose in the presence of 0.1 mM palmitate, 0.5% defatted BSA, 1 mM carnitine, and 0.11 Ci of [1-14 C]palmitate (55 mCi/mmol) (Amersham Pharmacia Biotech). At the end of the incubations media were collected and transferred to 25-ml Erlenmeyer flasks covered with septa caps. Media were acidified by injecting perchloric acid (6% final concentration) with a syringe. The liberated CO 2 was trapped in a plastic well suspended from the septa caps containing 0.4 ml of methanolic benzethonium hydroxide. After 1 h of incubation at 37°C, the wells were removed, and the trapped 14 CO 2 was measured by liquid scintillation counting. Cells were scraped in cold phosphate-buffered saline, pelleted by centrifugation, and resuspended in 4 ml of Folch reagent (9). Total lipids were extracted and separated by thin layer chromatography to measure the incorporation of labeled palmitate into phospholipids (9). The cellular triglyceride content was measured as described previously (9). Triglyceride recovery assessed with triolein was 94 Ϯ 5% (n ϭ 4).
Statistical Analysis-All results are expressed per mg of protein or g of DNA as the mean Ϯ S.E. of the indicated number of experiments. Statistical significance was calculated with the Student's t test.

RESULTS
Glucose Down-regulates the Expression Level of PPAR␣ mRNA in Insulin-secreting Cells-Total RNA was isolated from batches of 50 islets, and the expression level of the PPAR␣ transcript was measured by a semiquantitative reverse transcriptase-PCR technique. The quantitation of PPAR␣ isoform is expressed as the ratio of PPAR␣ to actin mRNA. Fig. 1 shows that the expression of the PPAR␣ mRNA species is downregulated by elevated glucose in rat islets. Glucose also decreases the expression of the PPAR␣ transcript in the highly glucose-responsive clonal ␤-cell line INS(832/13) (32) (Fig. 1), although the effect is slightly more prominent (80% reduction) than in rat islets. Oleate, in contrast to glucose, does not alter the mRNA expression level of PPAR␣ in either rat islets or INS(832/13) cells (Fig. 1). In addition, the action of glucose is similar in the absence or presence of oleate. A previous report documented that a FFA mixture of palmitate/oleate (1:2) (2 mM; concentration of albumin unknown) induced the PPAR␣ transcript in rat islets (28). We measured PPAR␣ expression in INS(832/13) cells following a 72-h incubation period in the presence of a 0.5 or 2 mM palmitate/oleate (1:2) mixture bound to 0.5 and 2% albumin, respectively. The palmitate/oleate mixture did not change the abundance of the PPAR␣ transcript (data not shown). The reason for the dichotomy between the present work and this previous study remains to be explained. Perhaps the albumin concentration is important in this respect. The results indicate that INS(832/13) cells can serve as a model system to further characterize and understand the mechanism whereby glucose modulates the expression of the PPAR␣ gene, which coordinates the expression of several metabolic genes in the fat oxidation pathway (26).
Increasing glucose from 3 to 20 mM causes a dose-dependent down-regulation of PPAR␣ mRNA with a threshold concentration of 6 mM and 70% reduction at 20 mM glucose ( Fig. 2A). The inhibitory action of glucose is rapid because it occurs with a lag time of about 2 h and is maximal at 6 h (Fig. 2B). The same reduction observed in PPAR␣ mRNA expression at 6 h was observed after a 24-h (Fig. 2B) or 48-h (data not shown) incubation period.
Glucose Reduces the DNA Binding Activity of PPAR␣ and the Expression of the PPAR␣ Protein-The DNA binding activity of PPAR␣ for its cis-regulatory element in the lipoprotein lipase promoter was tested by an electrophoretic mobility shift assay using nuclear extracts of low (3 mM) and high (20 mM) glucosetreated cells. Fig. 3A shows the results of a representative experiment, and Fig. 3B shows the quantitation of six separate experiments. Glucose decreases the DNA binding activity of PPAR␣ for its PPRE cis-element by 80% (Fig. 3A, lane 1 versus  lane 4). The higher intensity of the low molecular weight band in lane 2 compared with that in lane 1 was not observed reproducibly. To determine that the shifted band is due to PPAR␣ binding to the DNA probe, supershift assays were also performed with anti-PPAR␣, anti-PPAR␤, and anti-PPAR␥ antibodies. Lane 2 shows the complete disappearance of the shifted band in the presence of PPAR␣ antiserum. In blots other than the one shown in Fig. 3, the labeled probe was not supershifted with specific antibodies against PPAR␤ or PPAR␥ (data not shown). This demonstrates that the retardation of the migration of the PPRE probe on the gel is due to its binding to PPAR␣. To assess the integrity of the samples and further test the specificity of the glucose effect, the MLTF cis-element was used as a control in the electrophoretic mobility shift assay. MLTF is a ubiquitously distributed DNA-binding protein that is insensitive to changes in glucose concentration in rat islets (37). As shown in Fig. 3A, the DNA binding activity of MLTF is not changed at high glucose (lane 6 versus lane 7), which attests to the specificity of the action of the sugar on PPAR␣. The reduced DNA binding activity of PPAR␣ at high glucose is probably due to a decreased expression of the PPAR␣ protein. Indeed Fig. 4 shows that chronic exposure of INS(832/13) cells to elevated glucose reduces the amount of the PPAR␣ protein by about 75% as assessed by the Western blot technique.
Glucose Down-regulates the mRNA Expression Level of Some Genes Containing a PPRE Sequence-The mRNA expression of three genes implicated in the control of fat oxidation was measured after a 48-h period of exposure to high glucose (Fig. 5). The genes encoding UCP2 and ACO have been documented to contain PPRE sequences in their promoters (26). By contrast, only the muscle type CPT I gene, but not the liver-type CPT I gene, which is expressed in the ␤-cell (40), was shown to contain a PPRE regulatory sequence (54). UCP2 and ACO mRNA levels are reduced by 70 and 50%, respectively in cells chronically exposed to high glucose (Fig. 5). By contrast, and in agreement with previous work (16), the sugar has no effect on liver-type CPT I mRNA expression. The PPAR␣ agonist clofibrate does not increase CPT I mRNA in INS cells (16), and fatty acids and elevated glucose can still induce the islet CPT I gene in PPAR␣ Ϫ/Ϫ mice. 3 Therefore, the observation of a lack of an associated decreased expression of the CPT I transcript when PPAR␣ is down-regulated by glucose is consistent with the view that PPAR␣ is not implicated in the control of the livertype CPT I gene.
Mechanism of Glucose Regulation of PPAR␣ Gene Expression-Two sets of experiments were performed to gain insight into the mode of action of glucose on the PPAR␣ gene. First, we tested the action of protein and RNA synthesis inhibitors. Cycloheximide does not alter the action of glucose (Fig. 6A). This indicates that the action of glucose occurs in the absence of de novo protein synthesis. The stability of PPAR␣ mRNA is the same at low (3 mM) and high (20 mM) glucose in the presence of actinomycin D, which inhibits RNA synthesis (Fig.  6B). This indirectly suggests that the action of glucose occurs at the transcriptional level. Secondly, we tested several glucose analogues, nutrient stimuli, and pharmacological agents (Fig.  7). Mannose, an epimer of glucose that is well metabolized in the ␤-cell (41) glucose is suppressed by the glucokinase (GK) inhibitor mannoheptulose. 6-deoxyglucose and 3-O-methylglucose, two glucose analogues that rapidly enter the ␤-cell via Glut2 but are not phosphorylated by GK (39), have no effect. 2-deoxyglucose, which is phosphorylated by GK, reduces PPAR␣ expression, although less effectively than glucose. Pyruvate or glutamine plus leucine, which promote insulin secretion in INS cells (41), does not alter the mRNA expression levels of PPAR␣. Elevated K ϩ , phorbol 12-myristate 13-acetate, and forskolin, which are good secretagogues and activate the Ca 2ϩ -, protein kinase C-, and cAMP-signaling systems (42), respectively, were tested. A depolarization of the plasma membrane with high K ϩ and an increase of intracellular cAMP levels reduces PPAR␣ by about 40%. Activation of the protein kinase C pathway with phorbol 12-myristate 13-acetate has no effect. To evaluate whether the effect of glucose occurs indirectly via an autocrine action of insulin, two inhibitors of insulin secretion were used. Epinephrine, which suppresses insulin secretion at a late step in the exocytic process (43), and nifedipine, which curtails secretion by impairing Ca 2ϩ influx via L-type Ca 2ϩ channels (42), have no effect. Likewise, exogenous insulin does not alter PPAR␣ mRNA expression. This series of experiments rules out an autocrine action of insulin on PPAR␣ expression.
Chronic As expected from previous work (44), high glucose acutely increases the malonyl-CoA content of INS-1 cells (Fig. 8A) and reduces fat oxidation (Fig. 8B). Chronic oleate slightly decreases the cellular malonyl-CoA content and increases fat oxidation. A long term exposure to high glucose causes a 6-fold rise in malonyl-CoA and a more than 60% reduction of fat oxidation, even in cells that are subsequently incubated at low (5 mM) glucose (Fig. 8). Qualitatively similar observations were made in cells chronically exposed to both elevated glucose and fatty acids, although the effects are quantitatively less prominent, with a 2-fold rise in malonyl-CoA. This is consistent with the fact that FFA partly antagonize the inductive action of glucose on acetyl-CoA carboxylase, which synthesizes malonyl-CoA (14). Finally, the results in Fig. 8 show that malonyl-CoA is elevated by about 2.5-fold and that fat oxidation is reduced by more than 80% in cells preexposed to both elevated glucose and oleate for 3 days and in which high glucose is present during the incubation period.
Chronic elevated glucose and oleate alone increase the esterification of palmitate into total lipid and phospholipids and promote TG deposition in INS-1 cells (Fig. 9). A similar observation was made in the combined presence of high glucose and oleate. In addition, glucose and oleate synergize to cause TG accumulation in INS cells (Fig. 9).
In accordance with previous work done on rat islets and INS cells (9, 13), a long term exposure of ␤-cells to high glucose or oleate alone results in high basal secretion at 5 mM glucose and a lack of insulin release in response to a further rise in glucose (Fig. 10). Interestingly, the actions of chronic elevated glucose and oleate are additive, with a very high basal insulin secretion rate and also a lack of secretory response to the sugar at concentrations between 5 and 25 mM (Fig. 10).

DISCUSSION
The results indicate that elevated glucose markedly downregulates the expression of the PPAR␣ gene in the ␤-cell. The action of glucose on PPAR␣ mRNA expression occurs in less than 2 h and does not require de novo protein synthesis. The rapidity of the effect and the absence of a requirement for protein synthesis indicate that the PPAR␣ gene behaves as an early response gene in INS(832/13) cells. Therefore, this transcription factor could play an important role in the coordinated control of glucose and lipid metabolism in the ␤-cell in response to hyperglycemia.
To date there are very few examples of genes whose expression level is down-regulated by glucose. Glucose reduces the expression of the phosphoenolpyruvate carboxykinase gene in hepatocytes (45). By contrast, elevated glucose induces several glycolytic and lipogenic genes, in particular liver-type pyruvate kinase (13) and acetyl-CoA carboxylase (14), by a mechanism that requires phosphorylation of the sugar. The mediator of the action of glucose is not known, but it might be glucose-6phosphate or ribulose 5-phosphate (46), an intermediate of the pentose phosphate pathway. The results in this study indicate that many of the characteristics of the action of glucose on the PPAR␣ gene (e.g. the specificity for sugars that are phosphorylated by GK, the absence of an effect of other calorigenic nutrient secretagogues) are similar to those previously observed for the glucose induction of the liver-type pyruvate kinase (47), acetyl-CoA carboxylase (14), and fatty acid synthase (48) genes. This suggests that a similar metabolic intermediate, possibly glucose-6-phosphate, mediates the action of glucose by up-or down-regulating the expression of some genes. We do not favor the view that the pentose phosphate intermediate ribulose 5-phosphate is implicated in the action of glucose on gene expression, because glucose rapidly reduces the flux of glucose carbon through this pathway in the ␤-cell (49).
The DNA response elements mediating the action of glucose on the PPAR␣ gene remain to be defined. Interestingly, two sequences similar to the consensus glucose responsive element (CACGTG) found in many glucose responsive genes (50,51) are present in the PPAR␣ promoter. These sequences correspond to positions Ϫ867/Ϫ862 (CACCTG) and Ϫ130/Ϫ125 (CACGCG). It will be of interest to determine whether these putative glucose response elements are implicated in the down-regulation of the PPAR␣ gene by glucose. This will reveal whether the same consensus response element mediates both the induction and the repression of some genes by carbohydrates.
The only tested agents besides glucose and mannose that reduce PPAR␣ mRNA expression are forskolin and high K ϩ , although both are clearly less effective than glucose. Whether cAMP directly regulates PPAR␣ expression should be considered. However, elevated cAMP might indirectly regulate PPAR␣ expression via intracellular glucose production through glycogen breakdown, because INS cells in culture contain appreciable amounts of glycogen (13). The action of glucose is unlikely to be modulated via cAMP signaling because glucose does not modify cAMP levels in INS cells (21). Elevated KCl slightly reduces PPAR␣ mRNA expression. Nonetheless, the Ca 2ϩ signaling system is unlikely to be implicated in the action of glucose because the Ca 2ϩ channel antagonist nifedipine, which suppresses the glucose-induced Ca 2ϩ rise (42), does not alter the action of glucose. In addition, pyruvate and Gln plus Leu, which promote Ca 2ϩ influx in INS cells, are ineffective.
Islets from ZDF fa/fa rats show reduced PPAR␣ mRNA expression (28) in association with reduced fat oxidation and increased lipid esterification processes (10). As far as lipid metabolism gene expression is concerned, ZDF islets exhibit increased expression of the esterifying enzyme glycerol-3-phosphate acyltransferase (10) and reduced expression of the UCP2 transcript (35), whereas the liver-type CPT I mRNA level is unaltered (28). It is interesting to note that similar changes are observed in INS cells chronically exposed to elevated glucose. Thus, high glucose causes a sustained increased in malonyl-CoA, reduced fat oxidation, and increased phospholipids and TG synthesis. In addition, the sugar reduces the expression of the PPAR␣, ACO, and UCP2 genes, with no change in CPT I mRNA level. Because the ACO (52) and UCP2 (53) genes contain PPRE in their promoters and are regulated by PPAR␣ (26), it is tempting to suggest that the glucose-induced down-regulation of PPAR␣ in the ␤-cell might be implicated in the redirection of lipid metabolism from fat oxidation to esterification during hyperglycemia (1). We will directly evaluate this hypothesis by changing the expression of PPAR␣ in the ␤-cell using the tools of molecular biology.
Fatty acid esterification processes may be implicated in the regulation of insulin secretion through the generation of lipid mediators such as diacylglycerol (41). Thus, provided it can be verified in the in vivo situation, a down-regulation of PPAR␣ expression by episodes of postprandial glucose elevations might play a role in the ␤-cell compensation phase in the development of adipogenic diabetes (see Fig. 11). Consistent with this view, chronic (3 days) high glucose and oleate have an additive effect in causing high basal insulin secretion in INS cells. However, at later stages in the progression toward diabetes, with increasing insulin resistance, long episodes of postprandial hyperglycemia, and elevated FFA, the glucose-induced reduction of PPAR␣ expression might be implicated in the ␤-cell demise by virtue of its key role in lipid partitioning. Indeed, the three known pathways of lipid detoxification implicating fat oxidation are mitochondrial ␤-oxidation, which is limited by CPT I activity (40), peroxisomal fatty acid oxidation, where ACO is the limiting enzyme (26), and uncoupled fat oxidation, which may involve UCP2 (55,56). Thus, in response to chronic elevations in glucose the three fat oxidation pathways would be inhibited by 1) the chronically elevated malonyl-CoA, which suppresses CPT I activity, and 2) the reduction of the two remaining pathways mediated by glucose-induced decreases in ACO and UCP2 expression caused by the decrease in PPAR␣. Consistent with this view, malonyl-CoA is elevated, fat oxidation is almost totally suppressed, and lipid esterification processes are markely increased following long term exposure of INS cells to both elevated glucose and oleate. With respect to this new working hypothesis implicating both reduced PPAR␣ expression and a sustained rise in malonyl-CoA in the mechanism of ␤-cell glucolipotoxicity (see Fig. 11), it is interesting to note that the PPAR␥ ligand troglitazone (at very high concentrations that may cause PPAR␣ activation) reverses the secretory and lipid metabolism alterations of the ␤-cell in ZDF rats (6). In addition, treatments of yellow KK mice with BM17.0744, a new PPAR␣ agonist, results in an amelioration of the diabetic status and dyslipidemia (57).
In conclusion, glucose at concentrations in the range of 3-20 mM rapidly down-regulates the expression of the PPAR␣ gene by a mechanism that requires phosphorylation of the sugar by FIG. 11. Model implicating PPAR␣ and malonyl-CoA in the pancreatic ␤-cell adaptation to chronic hyperglycemia and the mechanism of ␤-cell glucolipotoxicity. In the absence of hyperglycemia, fatty acids, even when elevated as occurs during the early prediabetic phase of adipogenic diabetes, are oxidized (detoxification) and do not accumulate to a large extent in the ␤-cell. The three fatty acid oxidation pathways are mitochondrial ␤-oxidation, in which CPT I is limiting, peroxisomal ␤-oxidation, in which ACO is limiting, and uncoupled mitochondrial oxidation, which involves UCP2. With hyperglycemia (postprandial or chronic) the three pathways of fat oxidation are reduced because of 1) a sustained elevation in malonyl-CoA that inhibits CPTI, and 2) a reduction in PPAR␣ expression that downregulates ACO and UCP2. The glucose-induced suppression of fat oxidation causes fatty acyl-CoA (FA-CoA) accumulation in the cytoplasm and alters ␤-cell lipid partitioning by deriving fatty acyl-CoA from oxidation towards esterification. This results in increased phospholipid and diacylglycerol synthesis as well as triglyceride deposition. This process may initially contribute to ␤-cell compensation of insulin resistance via the accelerated production of lipid mediators such as the protein kinase C activator diacylglycerol that would increase insulin secretion. However, with time, the chronic suppression of fat oxidation will result in massive fat deposition in the ␤-cell, thus causing the ␤-cell demise (decompensation phase) and overt diabetes. G3P, glycerol 3-phosphate; DAG, diacylglycerol; PL, phospholipid; ␤-ox., ␤-oxidation; detox., detoxification; IRI, immunoreactive insulin.
GK. The action of glucose does not require de novo protein synthesis and probably occurs at the transcriptional level. It is suggested that a reduction of PPAR␣ gene expression together with chronic elevated malonyl-CoA plays a role in the coordinated adaptation of the ␤-cell glucose and lipid metabolism to hyperglycemia and may be implicated in the mechanism of ␤-cell glucolipotoxicity. The synergistic accumulation of TG in the ␤-cell in the combined presence of elevated glucose and FFA is consistent with this view, as illustrated in Fig. 11.