Pyruvate Dehydrogenase Kinase 4: Regulation by Thiazolidinediones and Implication in Glyceroneogenesis in Adipose Tissue

Dulbecco modified Eagle's medium (DMEM) was from Life Technologies (Cercy-Pontoise, France). Rosiglitazone was from Alexis Biochemicals (Coger, Paris). Leelamine was from Cayman Chemicals (Interchim, Montluçon, France). Small interfering RNA (siRNA) was from Invitrogen (Carlsbad, CA). Fetal bovine serum, essentially fatty acid–free BSA, 5,6-dichloro-1B-d-ribofuranosyl benzimidazole (DRB), dichloroacetate (DCA), and all other products were purchased from Sigma (L'isle d'Abeau Chesnes, France).

Male Zucker fa/fa rats were purchased from Charles River Laboratories (L'arbresle, France), and male Sprague-Dawley rats were purchased from Janvier Laboratories (Bagneux, France) at 6 weeks of age. They were allowed to acclimate 2 weeks before initiation of treatment and were kept on a 12-h light/dark cycle at constant room temperature. Conventional laboratory diet and tap water were provided ad libitum. In vivo experiments with Zucker rats were carried out as previously described (21). Briefly, rats were given a dose of 5 mg · kg⁻¹ · day⁻¹ rosiglitazone (maleate) or vehicle (0.5% methylcellulose) via oral gavage in the morning. Four rats were used per treatment group. After 4 days of treatment, animals were fasted during 4 h before killing. Plasma samples were obtained via the jugular vein, and aliquots were stored at −20°C for further glucose (Accu-Check; Roche, Meylan, France), insulin (Ultrasensitive rat insulin ELISA; EuroBio, Courtaboeuf, France), triglycerides (Vitros chemistry system; Johnson & Johnson), and NEFA (Fatty Acids half micro test; Roche) determinations. Samples of subcutaneous, periepididymal, and retroperitoneal WAT, liver, and soleus muscle were dissected and rapidly frozen in liquid nitrogen and stored at −80°C for further RNA analysis. The protocol for the animal studies was conducted according to the French Guidelines for the Care and Use of Experimental Animals.

Explants of subcutaneous WAT were obtained from eight women undergoing elective surgery. They were aged 45.7 ± 4.4 (mean ± SE) and had BMI of 30.29 ± 1.42 kg/m². None of the subjects suffered from known metabolic or malignant diseases or were taking medications known to alter adipocyte metabolism. The study was performed according to the Declaration of Helsinki. All the patients gave informed written consent.

WAT specimen from the subcutaneous region was obtained within 15 min after the onset of surgery. WAT (300 mg) was cut in small fragments of ∼20 mg and cultured in DMEM containing 12.5 mmol/l glucose, 200 IU/ml penicillin, and 50 mg/l streptomycin at 37°C in 10% CO₂ atmosphere. After 1 h of preincubation, rosiglitazone was added for 5 h. Explants were then frozen in liquid nitrogen before RNA extraction.

Adipocytes from Sprague-Dawley rat periepididymal fat pads were isolated in DMEM containing 5 mmol/l glucose, 0.5 mmol/l pyruvate, and 2% BSA, as described by Weisberg et al. (23). The pellet containing stromavascular fraction was discarded, whereas the floating cells were collected as adipocyte-enriched fraction.

3T3-F442A adipocytes were cultured at 37°C in a humidified atmosphere of 10% CO₂ in DMEM containing 25 mmol/l glucose, 10% newborn calf serum, 200 IU/ml penicillin, 50 mg/l streptomycin, 8 mg/l biotin, and 4 mg/l pantothenate. At confluence, the medium was changed to DMEM supplemented with 10% FCS and 20 nmol/l insulin to promote adipogenesis. This medium was changed every 2–3 days for 6 days. Twenty-four hours before RNA extraction, cells were placed in serum-free and hormone-free medium.

RNA interference with siRNA was performed by electroporation according to the manufacturer (Cell line Nucleofector kit L; Amaxa). Briefly, 3T3-F442A adipocytes on day 8 of differentiation were detached from culture dishes with trypsin-EDTA (Invitrogen). Tubes containing about 2 million cells were centrifuged for 10 min at 127g. Pelleted cells were then resuspended in 100 μl Nucleofector solution. Control (5′-GGAGGUGGUUGACUUUCAUTT; Invitrogen) or murine PDK4-specific (5′-GGAUUUGGUGGAGUUCCAUTT, Invitrogen) siRNAs were delivered into adipocytes (250 nmol/l) by electroporation. Transfection was stopped by adding DMEM containing 10% FCS. Adipocytes were then reseeded into six-well plates. Twenty-four hours after electroporation, medium was changed to FCS-free DMEM containing rosiglitazone or not. Nineteen hours later, cells were used for RNA analysis or metabolic studies.

Total RNA was extracted from the cell lines and from rat WAT by the method of Chomczynski and Sacchi (24), whereas RNeasy total RNA kit from Qiagen (Courtaboeuf, France) was used for human WAT explants. Sample quality was controlled by 260:280 nm absorption ratio determination (between 1.8 and 2.0).

For real-time RT-PCR analyses, 1.25 μg total RNA was first reverse transcribed using the High Capacity cDNA Archive kit (Applied Biosystems, Courtaboeuf, France). Reverse transcribed RNA were amplified on a thermal cycler (ABI prism 7900 HT; Applied Biosystems) using the SYBR green fluorescence method and specific oligonucleotides. Results were analyzed with the SDS 2.1 real-time detection system software. Quantification of RNA was carried out by comparison of the number of cycles required to reach reference and target threshold values (δ-δ Ct method).

Mitochondrial fraction was prepared from 3T3-F442A adipocytes, and mitochondrial pellet was resuspended in PBS 1× buffer, containing 1% Nonidet P-40, 0.5% sodium deoxycholate, and 0.1% SDS to extract mitochondrial proteins. Protein concentrations were determined using the Bradford method using BSA as the standard. Fractions (50 μg) were subjected to SDS-PAGE using a 10% resolving gel with a 6% stacking gel. Resolved proteins were transferred electrophoretically to nitrocellulose membranes, blocked for 1 h at 22°C with Tris-buffered saline (TBS) supplemented with 5% (w/v) nonfat powdered milk and incubated overnight at 4°C with polyclonal antisera raised against mouse PDK4, PDK2 from Santa Cruz Biotechnology (Tebu-bio, France), or PDC-E2 (provided by Dr. J.G. Lindsay, University of Glasgow, U.K.). Nitrocellulose membranes were then washed with 1% Tween in TBS (three times for 5 min) and incubated with horseradish peroxidase–linked secondary antibody anti-rabbit IgG (1:5,000 dilution, in 1% [w/v] BSA in TBS with 0.05% Tween) for 1 h at 22°C. Bound antibody was visualized using ECL according to the manufacturer's instructions. The blots were exposed to Hyperfilm, and the signals were quantified by scanning densitometry.

Isolated adipocytes were submitted to incorporation assay immediately after their collection. [1-¹⁴C]pyruvate incorporation was performed with 400 μl adipocyte suspension in 2 ml Krebs buffer containing 2% fatty acid–free BSA, 0.5 mmol/l pyruvate, and [1-¹⁴C]pyruvate at isotopic dilution of 25. When indicated, PDK inhibitors were added in the incorporation medium. Untransfected or siRNA-transfected 3T3-F442A adipocytes were treated or not for 18 h with rosiglitazone. Three hours before [1-¹⁴C]pyruvate incorporation, medium was replaced with glucose-free, serum-deprived DMEM containing 0.3% free fatty acid BSA. Pyruvate incorporation was carried out with 20 μmol/l [1-¹⁴C]pyruvate in Krebs buffer containing 0.3% BSA and leelamine, when indicated. After a 2-h incubation, cells were rinsed in PBS and scraped in 10 mmol/l Tris-Cl, pH 7.4, containing 0.25 mol/l sucrose, 0.1 mmol/l EDTA, 0.1 mmol/l dithiothreitol, and 0.1% Triton and frozen in liquid nitrogen before lipid extraction. Lipids were extracted according to the simplified method of Bligh and Dyer (25), and the subsequent [1-¹⁴C]pyruvate incorporation was estimated by counting the radioactivity associated with the chloroformic fraction. NEFA concentration was monitored in the incubation medium. Under each conditions, cells were submitted to a test of toxicity, using 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT) method, which did not show any drug-induced toxicity.

To test the efficiency of leelamine, a functional PDK assay was performed using a commercially available PDC purified from pig heart and containing intrinsic PDK activity, as extensively described by Jackson et al. (26), with 800 μg/ml purified PDC.

Quantitative results of Western blotting were obtained by densitometry in ImageJ software. The nonparametric Mann-Whitney U test for pairwise comparisons was applied due to the small number of experiments. Analyses were performed using the StatView 4.01 (Abacus Concepts, Berkeley, CA) statistical package. A value of P < 0.05 was considered statistically significant.

We showed previously that treating male Zucker fa/fa rats with rosiglitazone for 4 days significantly lowered both plasma insulin and serum lipid levels (21). This normalization in lipidemia was linked to a large induction in WAT glyceroneogenesis, thus decreasing NEFA release. Under these conditions, the expression of one of the key genes involved in glyceroneogenesis, PEPCK-C, was stimulated two- to threefold in subcutaneous, periepididymal, and retroperitoneal WAT. A similar rosiglitazone-induced normalization in lipidemia was obtained in the present study (Table 1). We used real-time RT-PCR to analyze the effect that rosiglitazone would have on the expression of Pdk2, Pdk4, Pdp1, and Pdp2 in the same three WAT depots. PDP1 and PDP2 mRNAs remained unaffected by rosiglitazone treatment (data not shown). In contrast, rosiglitazone induced a large 3- to 4.5-fold increase in PDK4 mRNA in all depots but neither in liver nor in soleus muscle, demonstrating a tissue-specific response of this gene under this acute treatment (Fig. 1A). In contrast, Pdk2 expression remained insensitive to rosiglitazone whatever tissue, demonstrating a selective action of the drug on PDK isoforms (Fig. 1B).

To determine whether rosiglitazone acts on PDK4 gene expression through a direct effect on WAT, we treated ex vivo rats and humans WAT explants for 5 h with rosiglitazone and monitored PDK4 mRNA. Both species responded to rosiglitazone by a large three- to fourfold increase in Pdk4 expression, whereas mRNA concentrations of PDK2 and pyruvate carboxylase (PC) were unaffected (Fig. 2A and B). WAT is a heterogeneous tissue containing not only adipocytes but also many other cell types, including preadipocytes, macrophages, fibroblasts, and endothelial cells. To study the mechanism of rosiglitazone action on Pdk4 gene expression in a homogeneous source of adipocytes, we used the 3T3-F442A adipocyte cell line. These cells were previously used to study Pdk1 regulation and glyceroneogenesis by PPARγ activators (27). A 2-h treatment of differentiated 3T3-F442A adipocytes with rosiglitazone induced PDK4 mRNA 2.5-fold whereas the concentrations of PDK2 and PC mRNAs were unchanged (Fig. 2C). We used Western blot experiments with specific antibodies for either PDK2 or PDK4 to monitor the relative expression of both protein isoforms in 3T3-F442A adipocytes at day 6 after confluence. Taking PDC-E2 as a control, results strongly suggest that PDK4 protein amount is about threefold that of PDK2 in adipocytes (Fig. 2D).

Rosiglitazone induction in Pdk4 expression was acute, with a maximum attained at 2 h in 3T3-F442A adipocytes (Fig. 3A). We chose this 2-h treatment to determine the rosiglitazone concentration that produced the half-maximal effect (EC₅₀). Incubation of 3T3-F442A adipocytes with rosiglitazone caused a dose-dependent increase in the abundance of PDK4 mRNA (Fig. 3B). EC₅₀ was at about 1 nmol/l, in agreement with the affinity of rosiglitazone for PPARγ, strongly suggesting that this receptor was involved. The rosiglitazone-induced increase in PDK4 mRNA could be the result of stimulation of Pdk4 transcription and/or of mRNA stability. To address these issues, we used DRB, a transcription inhibitor, and puromycin, a protein synthesis inhibitor. 3T3-F442A adipocytes were first treated with 1 μmol/l rosiglitazone with or without 80 μmol/l DRB for 0.5–2 h before analyzing PDK4 mRNA abundance by real-time RT-PCR. Whatever the time of treatment, DRB abolished the induction of PDK4 mRNA by rosiglitazone (Fig. 3A). Under the same conditions of rosiglitazone treatment, puromycin had no effect, demonstrating that de novo protein synthesis was not required (data not shown). These results are strongly in favor of rosiglitazone directly increasing transcription rate of Pdk4. Second, we examined the possibility of a rosiglitazone-induced PDK4 mRNA stabilization by incubating 3T3-F442A adipocytes for 2 h with rosiglitazone and then for various times with DRB before monitoring PDK4 mRNA abundance by real-time RT-PCR. In the absence of rosiglitazone, PDK4 mRNA half-life was estimated at ∼1.5 h as previously obtained in other cell types (Fig. 3C). After a 2-h treatment with 1 μmol/l rosiglitazone, PDK4 mRNA half-life rose to ∼2 h, showing a slight rosiglitazone-induced stabilization of PDK4 messenger in adipocytes. mRNA induction was followed by a significant twofold increase in PDK4 protein at 18 h of rosiglitazone treatment (Fig. 3D).

To test the potential involvement of PDC and PDK in the control of pyruvate flux toward glyceroneogenesis, pyruvate incorporation into neutral lipids was monitored in isolated adipocytes in the presence or not of PDK inhibitors. We tested both DCA, one of the few known highly specific synthetic PDK inhibitor, and leelamine, a more recent PDK inhibitor (28,29). Results of MTT test demonstrated that neither of these compounds was cytotoxic at the concentrations used (data not shown). Figure 4A shows that both inhibitors significantly reduced glyceroneogenic flux in isolated adipocytes from rat WAT. However, the concentrations used to reach a statistical significant reduction differ for the two inhibitors. At 50 μmol/l, leelamine inhibited pyruvate incorporation by 40% (P < 0.001), whereas a 10-fold higher DCA concentration was required to produce a significant 60% inhibition (P < 0.01) (Fig. 4A). Hence, leelamine is more efficient to reduce pyruvate incorporation into triglycerides than DCA. To ensure that leelamine effect was due to PDK inhibition, an in vitro assay was performed, monitoring residual PDC activity after kinase action. At the concentration of 50 μmol/l, leelamine induced an inhibition of 41.00 ± 2.14% in PDK activity (n = 8, P < 0.001).

To further study the functional involvement of PDK in both basal and rosiglitazone-activated conditions, we treated or not 3T3-F442A adipocytes with 1 μmol/l rosiglitazone for 18 h and then monitored the incorporation of [1-¹⁴C]pyruvate into lipids in the presence or not of leelamine for 2 h. The addition of 50 and 100 μmol/l leelamine reduced basal glyceroneogenic flux 25 and 56%, respectively, with a concomitant increase in NEFA release (Fig. 4B and C). As expected, rosiglitazone induced glyceroneogenesis and reduced NEFA release. The addition of 50 and 100 μmol/l leelamine on rosiglitazone-pretreated cells reduced glyceroneogenic flux 59 and 115%, respectively, with a concomitant increase in NEFA release (Fig. 4B and C). Under these conditions, glycerol release was not modified (data not shown). Hence, PDK inhibitors did not affect lipolysis. To ascertain the specific implication of PDK4 in glyceroneogenesis, a selective PDK4 siRNA was transfected in 3T3-F442A adipocytes. Forty-eight hours after transfection, a specific 60% (P < 0.001) decrease in PDK4 mRNA was observed whereas concentrations in PDK2, PDP1, PDP2, PC, and PEPCK-C mRNAs remained unchanged (Fig. 5A). Furthermore, PDK4 siRNA did not induce interferon (IFN)-β mRNA, showing the lack of IFN response (Fig. 5A). Incorporation of [1-¹⁴C]pyruvate into lipids was reduced 40% after transfection of adipocytes with PDK4 siRNA (P < 0.05) (Fig. 5B). Under these experimental conditions, rosiglitazone induced [1-¹⁴C]pyruvate incorporation 3.1-fold in control cells (P < 0.001) (Fig. 5B). A 35% decrease in rosiglitazone-induced [1-¹⁴C]pyruvate incorporation was observed in the presence of PDK4 siRNA (P < 0.01).

For a better understanding of the early mechanisms involved in PPARγ effects in WAT, we investigated the regulation of potent primary targets for thiazolidinediones. One important observation of our present report is that the expression of Pdk4 is acutely induced by the specific PPARγ agonist rosiglitazone in WAT explants from rats and humans. Induction of PDK4 mRNA occurs in the same time frame as that of PEPCK-C transcript. Furthermore, as we observed previously for PEPCK-C, a 4-day administration of rosiglitazone to Zucker fa/fa rats, a model of insulin resistance and dyslipidemia, induced an increase in Pdk4 expression specifically in WAT from several depots. A longer term in vivo effect of COOH, a PPARγ agonist, on PDK4 mRNA was observed by Laplante et al. (30). Our results are in accordance with these previous results in which, however, the issue of tissue-specific action of the drug was not addressed. In the course of our experiments, we showed that rosiglitazone induction of Pdk4 expression was tissue-specific because liver and muscle did not respond to such a treatment. Furthermore, in our study, Pdk2 expression remained unaffected by rosiglitazone treatment, whereas Laplante et al. (30) observed a simultaneous induction of both Pdk2 and Pdk4 expression in visceral WAT, probably because of the longer duration of treatment used (3 weeks). In contrast to our results, Way et al. (31) observed that short-term (1-day) treatment with GW1929, a different PPARγ agonist, reduced the concentration of PDK4 transcript in muscles from ZDF Zucker diabetic rats (31). One explanation for this discrepancy could be that ZDF rats are insulin-resistant and diabetic at earlier times than the Zucker fa/fa rats we used here. Hence when ZDF rats are 9 weeks old, their insulin-resistant state is correlated with low PDK4 activity in muscle, which could be raised by rosiglitazone. Such an effect is probably indirect as the result of thiazolidinedione-induced rapid decrease in blood NEFA. In 8-week-old fa/fa rats, blood NEFA concentrations are near to normal and therefore unable to stimulate PDK4 in muscle (21). Whatever happens in muscle, our data clearly demonstrate that PDK4 is an early-responsive positive target for PPARγ in WAT.

To further study the mechanisms by which rosiglitazone stimulates Pdk4 expression, we used the 3T3-F442A cell line. Rosiglitazone rapidly stimulated Pdk4 expression in fully differentiated 3T3-F442A adipocytes as previously found in 7,800 C1 hepatoma cells treated with a PPARα ligand (32). Using transcription and translation inhibitors, we showed that PDK4 was a transcriptional and direct target for rosiglitazone. Our results are in slight contrast with those from Sears et al. (33) who showed that in 3T3-L1 adipocytes, rosiglitazone induction of PDK4 mRNA is delayed, suggesting that in these cells, PDK4 is an indirect PPARγ target. Furthermore, we found that half-life of the PDK4 transcript was short, i.e., ∼2 h, as previously observed for hepatoma cells (32). Interestingly, PC gene expression was not modulated. PC was previously shown as a PPARγ-induced gene during adipocyte differentiation of the 3T3-L1 preadipocytes (34). Our observation that rosiglitazone does not modulate pyruvate carboxylase gene expression in differentiated 3T3-F442A adipocytes is in agreement with a delayed action of PPARγ agonist on this gene during adipocyte differentiation.

We show unambiguously here that PDK4 participates in adipocyte glyceroneogenesis because leelamine, a PDK inhibitor, and a specific PDK4 siRNA strongly reduces basal and rosiglitazone-induced pyruvate incorporation into neutral lipids, with a coordinated stimulation of NEFA release. These results are illustrated in Fig. 6 in which pyruvate crossroad is shown as the central switch for the modulation of G3P production. We postulate that the circulating substrate for the whole process is lactate, which is abundant in the blood under the physiological situation during which lipolysis is activated, i.e., fasting. Rosiglitazone induction of both PDK4 and PEPCK-C results, respectively, in reduction of pyruvate flux toward acetyl-CoA because PDC activity is strongly inhibited by PDK4 and in increased G3P synthesis via PEPCK-C. Such a combined action of thiazolidinediones allows increased reesterification during lipolysis leading to decreased NEFA release from adipocytes. The reduced PDC activity under rosiglitazone treatment raises the intriguing possibility that thiazolidinediones may reduce glucose oxidation to CO₂ in adipocytes, in contrast with what occurs in muscle and at the whole-body level. Although this issue requires further development, we can emphasize here that glucose oxidation is not a predominant pathway during fasting and in the absence of insulin.

Elevated concentration of plasma NEFA is now recognized as a key factor in the onset of insulin resistance and type 2 diabetes (35). Our previous results indicate that WAT glyceroneogenesis is important to lipid homeostasis and that a disregulation in this pathway has profound pathophysiological effects (19–21,36). This would result in dramatic changes in net NEFA output, hence in the concentration of plasma NEFA, which in turn would mediate metabolic alterations, for instance by affecting insulin sensitivity of peripheral tissues. Here, we provide clear evidence that PDK4 is a new player in the glyceroneogenic process and is a thiazolidinedione target selectively in WAT. Therefore, by means of PEPCK-C and of PDK4, glyceroneogenesis is a target for thiazolidinediones, the antidiabetic action of which seems clearly to be linked to their hypolipidemic effect (22,37).

T.C. has received a fellowship from the Ministère de l'Education Nationale, de l'Enseignement Supérieur et de la Recherche. E.D. has received a fellowship from the Ministère de l'Education Nationale, de l'Enseignement Supérieur et de la Recherche. J.-M.B. has received a fellowship from the Ligue Nationale contre le Cancer. C.F. has received support from the Institut National de la Recherche Médicale and the University Paris Descartes and a grant from the Association de Langue Française pour l'Etude du Diabète et des Maladies métaboliques ALFEDIAM-GlaxoSmithKline.

We thank P. Juraver (Institut National de la Santé et de la Recherche Médicale Unité Mixte de Recherche-S 747) for rat handling, Dr. V. Nivet (University Paris Descartes) for triglyceride determination, and Dr. L. Benelli (Plastic Surgery Unit, Hartman Clinic, Paris) for his helpful involvement with human studies.