Mechanism for peroxisome proliferator-activated receptor-alpha activator-induced up-regulation of UCP2 mRNA in rodent hepatocytes.

Peroxisome proliferator-activated receptor-alpha (PPARalpha)activators, fish oil feeding, or fibrate administration up-regulated mitochondrial uncoupling protein (UCP2) mRNA expression in mouse liver by 5-9-fold, whereas tumor necrosis factor-alpha (TNFalpha) also up-regulated UCP2 in liver. In this study, the mechanisms for PPARalpha activators-induced up-regulation of UCP2 mRNA, related to TNFalpha and reactive oxygen species (ROS), were investigated. PPARalpha activators-induced UCP2 up-regulation in mouse/rat liver tissues was due to their increases in hepatocytes but not in non-parenchymal cells. Addition of PPARalpha activators, WY14,643 or fenofibrate, to cultured hepatocytes up-regulated UCP2 mRNA by 5-10-fold. PPARalpha activators-induced up-regulation of UCP2 mRNA was not due to increased mRNA stability and required cycloheximide-sensitive short term turnover protein(s). However, expression of PPARalpha/retinoid X receptor-alpha and PGC-1 was not rate-limiting for WY14,643-induced UCP2 up-regulation. In primary hepatocytes, an exogenous oxidant, tert-butyl-hydroperoxide (TBHP), which increased ROS production, up-regulated UCP2 mRNA, whereas WY14,643 treatment did not produce detectable ROS under the condition that fibrate markedly up-regulated UCP2. In in vivo studies, PPARalpha activators moderately up-regulated TNFalpha mRNA expression in mouse liver. An anti-oxidant pyrrolidine dithiocarbamate ammonium salt injection completely prevented their TNFalpha mRNA increases but did not prevent most of their UCP2 mRNA increases. These data indicate that PPARalpha activators up-regulate UCP2 expression in hepatocytes through unknown proteins by increased transcription, and neither ROS nor TNFalpha production are the major causes for PPARalpha activators-induced UCP2 up-regulation.

UCPs 1 (UCP1, UCP2, and UCP3) are mitochondrial transporters that are capable of dissipating the proton gradient and increasing thermogenesis while reducing the efficiency of ATP synthesis (1). In addition, there are several other hypotheses concerning the physiological roles of UCPs, regulation of fatty acid and glucose oxidation, and reduction of mitochondrial reactive oxygen species (ROS) generation (2). UCP2 was expressed in liver tissues, but hepatocytes of the adult rat liver did not express UCP2, although non-parenchymal cells especially Kupffer cells did (3). However, UCP2 in hepatocytes will be up-regulated under some metabolic conditions. Bacterial lipopolysaccharide (LPS) stimulation or lipid emulsions (linoleic or oleic acid) treatment induced UCP2 mRNA accumulation in rat hepatocytes (4,5). Furthermore, in in vivo mice studies, fatty liver increased UCP2 expression in hepatocytes (6). We also have shown that compared with safflower oil feeding, fish oil feeding up-regulated UCP2 by 5-fold and fenofibrate administration also induced UCP2 expression by 9-fold in mouse liver tissues (7). Because n-3 fatty acids rich in fish oil and fibrate compounds are peroxisome proliferator-activated receptor-␣ (PPAR␣) activators that stimulate ␤-oxidation of fatty acids in hepatocytes, and because there was no evidence that energy generated by increased oxidation of fatty acids was being channeled into energy-requiring processes, PPAR␣ activators-induced increase of fatty acids oxidation was uncoupled in hepatocytes. In contrast, in rats, PPAR␣ activators did not affect UCP2 mRNA levels when they were measured in whole liver tissues (8). Because hepatocytes are a major site of ␤-oxidation of fatty acids, if we assumed that increased UCP2 expression was coupled to this process, UCP2 expression in hepatocytes should be up-regulated in both mice and rats. To answer this question, fish oil or fibrate was administered in rats and mice; hepatocytes and non-parenchymal cells were isolated from liver tissues by collagenase digestion, and UCP2 mRNA levels in each cell fraction were measured.
PPARs ligands up-regulate UCP2 in adipocytes and skeletal muscles (9 -11) and unsaturated fatty acids are potent inducers of PPARs (12,13). Because in C2C12 cells the overexpression of PGC-1, a coactivator of PPAR␣ and PPAR␥, up-regulated UCP2 mRNA in the absence of ligands (14), it was suggested that PPAR␣ activation was necessary for UCP2 up-regulation in C2C12 cells. However, the PPAR␣ knockout mouse study (15) suggested that cardiac UCP2 expression was regulated via a PPAR␣-independent mechanism. These data suggested that UCP2 was regulated differently in different types of cells.
As for other mechanism(s) for UCP2 up-regulation by PPARs, it has been hypothesized that generation of reactive oxygen species (ROS) by increased mitochondrial or peroxisomal fatty acid oxidation may lead to up-regulation of UCP2 (16). PPAR␣ activators increase in the expression of H 2 O 2generating peroxisomal fatty acyl-CoA oxidase and microsomal cytochrome P450 4A1 and 4A3 genes. Indeed, it has been reported that PPAR activators increased H 2 O 2 production in rat liver tissues (17,18). In agreement with this hypothesis, TNF␣ stimulation, by which mitochondrial ROS production was increased (19,20), up-regulated UCP2 mRNA in primary cultures of hepatocytes (4). A study of UCP2 promoter analysis revealed that the proximal 3.3 kb of UCP2 promoter contained ROS-sensitive trans-acting factors, including AP-1 and C/EBP, but did not contain a complete NF-B consensus motif (4). Thus, PPAR␣ as well as TNF␣ activation increase H 2 O 2 levels and then may up-regulate UCP2 mRNA through ROS-sensitive transacting factors. In relation to PGC-1, TNF␣, and ROS, the mechanism(s) of increased UCP2 expression in hepatocytes by PPAR␣ activators were examined.

MATERIALS AND METHODS
Animals-Female C57BL/6L mice (8 weeks of age) and female Sprague-Dawley rats (7 weeks of age) were obtained from Tokyo Animals Science Co. (Tokyo, Japan). The animals were fed standard rodent chow (CE2, CLEA, Japan) for 3-7 days to stabilize the metabolic conditions. Animals were exposed to a 12-h light/12-h dark cycle and were maintained at a constant temperature of 22°C.
Diet-In the experiments, C57BL/6J mice were divided into three groups (n ϭ 4 in each group). The first group was a given a high carbohydrate diet, which on a calorie basis contained 63% carbohydrate, 11% fat, and 26% protein. Safflower oil was used as source of fat in the high carbohydrate-fed diet. The second group was given a high safflower oil-rich diet containing 14% carbohydrate, 60% safflower oil, and 26% protein. The third group was given a high fish oil diet containing 14% carbohydrate, 60% fish oil (mainly from tuna), and 26% protein.
Fatty acid compositions of dietary oil were subjected to gas-liquid chromatography. Safflower oil (high oleic type) contained 46% oleic acid (18:1 n-9) and 45% linoleic acid (18:2 n-6) from total fatty acids; fish oil contained 7% eicosapentaenoic acid (20:5 n-3) and 23% docosahexaenoic acid (22:6 n-3). In fenofibrate experiments, fenofibrate (Sigma) was mixed in the high carbohydrate diet (0.5%, w/w). The materials and methods of preparation of diet were the same as those used in our previous study (7). In rat experiments, Sprague-Dawley rats were divided into three groups as follows: high carbohydrate diet-fed rats, high fish oil diet-fed rats, and fenofibrate administered high carbohydratefed rats. Mice and rats were fed each diet for 2 days and were anesthetized at about 10:00 a.m. by intraperitoneal injection of pentobarbital sodium (0.08 mg/g body weight, Nembutal, Abbott).
Isolation of Hepatocytes and Non-parenchymal Cells-Isolated hepatocytes and non-parenchymal cells were obtained from mice or rats fed the abovementioned diet using the two-step perfusion collagenase method (21). Briefly, the liver was first perfused in situ through the portal vein with liver perfusion medium (Invitrogen) at 37°C for a few minutes. Then it was perfused with liver digest medium containing collagenase (Invitrogen). The resulting liver cell suspension was separated into two fractions, hepatocytes and non-parenchymal cells, by differential centrifugation at 50 ϫ g for 1 min. Precipitate cells were used as hepatocytes. The supernatants was recentrifuged at 50 ϫ g for 1 min three or four times to remove hepatocytes and were washed twice by centrifugation at 150 ϫ g for 3 min with addition of 10 ml of PBS to obtain a precipitate, which was used as non-parenchymal cells. RNA was prepared from each cellular fraction by TRIZOL reagent (Invitrogen).
Primary Hepatocytes Culture and Plasmid Transfections-The viability of the isolated rat hepatocytes was over 85% as determined by the trypan blue exclusion test. The hepatocytes were plated at a density of 2 ϫ 10 6 cells/10-cm plastic dish coated with collagen I (Iwaki, Tokyo) in M199 medium (Invitrogen) supplemented with 5% fetal bovine serum, 100 nM triiiodothyronine, 100 nM dexamethasone, 1 nM insulin, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C in a humidified atmosphere of 5% CO 2 , 95% air. After a 4-h attachment period, the medium was replaced with the same medium without serum and was further incubated for 16 h (22).
Hepatocytes were transfected with expression vectors encoding mouse PPAR␣ (pNCMV-PPAR␣), mouse RXR␣ (pSG5-RXR␣), and mouse PGC-1 (pSG5-PGC-1) using Lipofectin (Invitrogen). The expression vectors PPAR␣ and RXR␣ were kindly provided by Dr. T. Osumi at Himeji Institute of Technology in Japan and Dr. P. Chambon at CNRS INSERM in France, respectively. Complete cDNA of mouse PGC-1 was obtained by PCR from first strand DNA using mouse skeletal muscle total RNA. First strand cDNA was prepared, using first strand cDNA synthesis kit (CLONTECH) primed with oligo(dT). PCR primers used were as follows: 5Ј primer, 5Ј-ATGGCTTGGGACATGTGC-3Ј, and 3Ј primer, 5Ј-TTACCTGCGCAAGCTTCTCT-3Ј. PCR was performed with FIG. 1. Effects of dietary oil feeding (A and C) or fibrate administration (B and C) on UCP2 mRNA levels in hepatocytes and non-parenchymal cells in mice (A and B) and rats (C). Hepatocytes and non-parenchymal cells were isolated from livers in high carbohydrate, high safflower oil, and high fish oil-fed mice (Carb.Saf. Fish) (A), high carbohydrate diet mice in the absence and presence of fenofibrate (about 500 mg/kg/d) (B), and high carbohydrate, high fish oil, and fenofibrate supplemented high carbohydrate diet-fed rats (Carb.Fish Fib.) (C). Mice and rats were fed these diets for 2 days. Total RNA was isolated from each cell group, and transferred membrane sheets were probed with 32 P-labeled human UCP2 cDNA. A typical autoradiogram (2-5 days of exposure) and its relative levels are shown. These mRNA levels were quantified using an image analyzer. The data for each band are shown as relative value to the UCP2 mRNA level of carbohydrate diet-fed group. A and B, because two independent experiments showed similar results, the results of a representative experiment, using pooled RNAs from 3 mice/group, are shown. C, each value represents mean Ϯ S.E. of three independent experiments, in which one rat is used in one group. Statistical differences are shown as p Ͻ 0.05 (*).
Taq DNA polymerase (Takara, Shiga, Japan). The amplified products were subcloned into pCR2.1-TOPO vector (Invitrogen) and verified by sequencing. The complete cDNA of mouse PGC-1 was subcloned into the BamHI-XhoI site in pSG5 vector.
Formation of liposome-DNA complexes was carried out by mixing 0.4 g of each plasmid DNA with Lipofectin to give a Lipofectin-to-DNA ratio of 6:1 (w/w) in 8 ml of medium (23). Incubation medium of hepatocytes was replaced with this plasmid mixture and incubated for 10 h. Eight ml of medium containing several growth factors, serum and WY14,643 (Calbiochem), was poured into this medium and incubated for further 38 h. Transfection efficiency was estimated by measuring ␤-galactosidase activity by cotransfection of ␤-galactosidase expression vector.

Preparation of Nuclear Extracts and Electrophoretic Mobility Shift
Assay-Nuclear extracts from TNF␣-and fenofibrate-treated cells were prepared as described previously (24). Hepatocytes (3 ϫ 10 6 cells/10-cm plastic dish) were washed once with 3 ml of PBS. Cells were scraped into cold PBS and then centrifuged at 1,500 ϫ g for 5 min. The packed cell volume was measured, and 5 volumes of ice-cold hypotonic buffer (10 mM HEPES, pH 7.6, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM DTT) was supplemented with protease inhibitors (40 g/ml bestatin, 0.5 mM pefabloc, 0.7 g/ml pepstatin A, 2 g/ml aprotinin, and 0.5 g/ml leupeptin) and with phosphatase inhibitors (20 mM ␤-glycerophosphate, 10 mM p-nitrophenyl phosphate, 50 M Na 3 VO 4 , 0.5 mM DTT). These reagents were obtained from Sigma. Cells were swollen on ice for 15 min, and then 0.3% Nonidet P-40 was added. The tube was then mixed vigorously for ϳ10 s. The homogenate was centrifuged at 15,000 ϫ g for 30 s in a microcentrifuge. The supernatant was discarded, and the nuclear pellet was lysed in NUN buffer (17.5 mM HEPES, pH 7.6, 1.1 M urea, 0.33 M NaCl, 1.1% Nonidet P-40, and 1 mM DTT) supplemented with proteinase inhibitors. The tubes were then mixed vigorously for 5 s and incubated on ice for 30 min (25). Nuclear extracts were centrifuged at 15,000 ϫ g for 10 min at 4°C, and supernatants were collected with addition of 10% glycerol, frozen in liquid nitrogen, and stored at Ϫ70°C prior to experiments. Protein concentration was determined using the micro-BCA assay reagent kit (Pierce) by measuring absorbance spectrophotometrically at 562 nm. NF-B consensus (GGGGACTTTCCC) and mutant (GGCGACTTTCCC) oligonucleotides were purchased from Santa Cruz Biotechnology and labeled with [␥-32 P]ATP by T4 polynucleotide kinase. Nuclear extract (10 g of protein) was incubated at room temperature for 20 min in a buffer containing 10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM DTT, 1 mM EDTA, 5% glycerol with 1 g of poly(dI-dC) and 32 P-labeled DNA probe (50,000 dpm). Reaction mixtures were separated by electrophoresis on 4.6% polyacrylamide gels. After electrophoresis, gels were dried and exposed to x-ray film at Ϫ80°C. Specificity of NF-B binding was verified by competition assays, oligonucleotide mutation assay, and ability of specific antibodies to supershift protein-DNA complex. In competition assays, 80-fold excess of unlabeled oligonucleotide was added 10 min before addition of labeled probe. In supershift experiments, 1 g of goat, rabbit antisera against p50 or p65 protein, respectively (Santa Cruz Biotechnology), was added to the reaction mixture 30 min before addition of labeled probe.
Antioxidant (PDTC) Effects on TNF␣ and UCP2 Expression-To examine the effects of pyrrolidine dithiocarbamate ammonium salt (PDTC) on LPS-induced increase of serum TNF␣ concentration and UCP2 mRNA, mice were injected intraperitoneally with PDTC (250 mg/kg body weight, Sigma), followed 30 min later by LPS (0.5 mg/kg body weight, Sigma) injection. Serum concentration of TNF␣ drawn at 90 min after LPS injection was measured using an enzyme-linked immunosorbent assay kit (BioSource International, Inc.). Mice were sacrificed 17 h thereafter. RNA from liver was used for measurement of UCP2 mRNA.
To examine the effects of PDTC on dietary oil feeding or fibrate administration-induced increases on TNF␣ and UCP2 mRNAs, PDTC (250 mg/kg body weight) was injected intraperitoneally just prior to initiation of feeding of high safflower oil diet, high fish oil diet, high carbohydrate diet, and fenofibrate-administered high carbohydrate diet to mice. RNA was isolated from livers 17 h after PDTC injection and used for Northern blots to measure TNF␣ and UCP2 mRNAs.
Northern Blotting-Northern blot analysis was performed as described previously (7). A portion of RNA (15 g per lane) was denatured with glyoxal and dimethyl sulfoxide and analyzed by electrophoresis in 1% agarose gels. After transfer to nylon membranes (PerkinElmer Life Sciences) and UV cross-linking, RNA blots were stained with methylene blue to locate 28 S and 18 S rRNAs and to ascertain the amount of loaded RNAs. Human UCP2, mouse PPAR␣, mouse RXR␣, mouse PGC-1, rat ACS, and human TNF␣ cDNAs were labeled with [␣-32 P]dCTP (PerkinElmer Life Sciences) using a random prime labeling kit (Multiprime DNA Labeling Systems, Amersham Biosciences). The amounts of UCP2, PPAR␣, RXR␣, PGC-1, ACS, and TNF␣ mRNAs were quantitated with an image analyzer (BAS 2000, Fuji Film, Tokyo, Japan) and expressed as the intensity of phosphostimulated luminescence.
Hydroperoxide Measurement-Hepatocytes were washed with PBS and resuspended medium containing 100 nM triiiodothyronine, 100 nM dexamethasone, 1 nM insulin (Lilly), and agents such as TBHP (30, 100 M, Sigma) and WY14,643 (50 M), incubated for 14 h, and then harvested. A fluorescent probe, 2,7-dichlorofluorescin diacetate (DCF-DA, 10 M, Sigma), was added at 30 min prior to cell harvesting. DCF-DA was used for assessing intracellular hydroperoxides (hydrogen peroxides and lipid hydroperoxides) (26,27). DCF-DA diffuses through the cell membrane and is converted to DCF by intracellular esterases. Intracellular hydroperoxides convert non-fluorescent DCF to fluorescent DCF. Because DCF-DA located outside of the cells does not react with hydroperoxides, only intracellular hydroperoxides that react with DCF are detected. Fluorescence intensity was measured in a Hitachi F2000 fluorescence spectrophotometer with excitation wavelength at . ␣-Amanitin, polymerase II inhibitor, was added to the culture medium 2 h prior to addition of WY14,643. Total RNA was isolated, and transferred membrane sheets were probed with 32 P-labeled human UCP2 cDNA. A typical autoradiogram (2-day exposure) and its relative levels are shown. In the autoradiogram, each line represents a sample from one dish. The radioactivity in each band was quantified using an image analyzer. The data for each band are shown as relative value to UCP2 mRNA level of control, untreated cells. Each value represents mean Ϯ S.E. of 3 dishes. Statistical differences are shown as p Ͻ 0.01 (**) and p Ͻ 0.001 (***).

Fish Oil Feeding or Fibrate Administration Up-regulates UCP2 mRNA in Hepatocytes from Both Mice and Rats-We
reported previously (7) that fish oil feeding for 5 months or fibrate administration for 2 weeks up-regulated UCP2 mRNA in whole liver tissue by 5-and 9-fold, respectively. To examine whether hepatocytes or non-parenchymal cells are responsible for UCP2 up-regulation observed in whole liver tissues, the level of UCP2 mRNA in two types of liver cells was measured separately by Northern blotting (Fig. 1). In carbohydrate-fed mice, UCP2 mRNA level in non-parenchymal cells (mostly Kupffer cells) was 4 -7-fold higher than hepatocytes (Fig. 1A). Compared with carbohydrate-fed mice, safflower oil-fed mice for 2 days showed 1.6 -1.8-fold increases of UCP2 expression in both cell types. However, compared with safflower oil-fed mice, fish oil-fed mice showed a 5-fold increase of UCP2 mRNA in hepatocytes, whereas its increase in non-parenchymal cells was very small (1.3-fold). Similar different expression profiles were observed in fenofibrate administration (Fig. 1B). Fibrate administration for 2 days increased UCP2 mRNA by more than 18-fold in hepatocytes and by 2.5-fold in non-parenchymal cells.
When examined in whole liver tissues, fish oil feeding or fibrate administration did not increase UCP2 mRNA in rats (data not shown). In hepatocytes isolated from rat liver, fish oil feeding or fenofibrate administration for 2 days up-regulated UCP2 mRNA by 1.5-and 5.2-fold, respectively, whereas in non-parenchymal cells, either fish oil feeding or fenofibrate administration rather down-regulated UCP2 expression by 27 and 30%, respectively (Fig. 1C). However, only up-regulation of hepatocytes UCP2 by fenofibrate administration reached statistical significance (p Ͻ 0.05). PPAR␣ activators selectively up-regulated UCP2 mRNA in rat hepatocytes, but their increases are smaller than those in mice. These data well ex-  h to measure RNA preparation. Generation of ROS in hepatocytes was evaluated with oxidation-dependent fluorogen, DCF-DA as described under "Materials and Methods." Total RNA was isolated, and transferred membrane sheets were probed with 32 P-labeled human UCP2 cDNA. The radioactivity in each band was quantified using an image analyzer. The data for fluorescence intensity and UCP2 mRNA levels are shown as relative value to those of Me 2 SO-treated cells. Each value represents mean Ϯ SE of 3-6 dishes. plained why PPAR␣ activators did not up-regulate UCP2 mRNA in whole liver tissues of rat (8). In rats, small increases of UCP2 mRNA in hepatocytes and down-regulation of UCP2 mRNA in non-parenchymal cells did not result in up-regulation of UCP2 mRNA in whole liver tissues. At present, the reasons for the differential responses of PPAR␣ activators to UCP2 expression in hepatocytes and non-parenchymal cells between mice and rats are not clear. However, these differences might not be due to cross-contamination of hepatocytes and nonparenchymal cell fractions, because there were no significant differences of PPAR␣ mRNA (a marker of hepatocytes) and TNF␣ mRNA (a marker of non-parenchymal cells) distributions in hepatocytes and non-parenchymal cell fractions between mice and rats (data not shown). The down-regulation of rat UCP2 in non-parenchymal cells was also reported in LPStreated rats; LPS treatment up-regulated UCP2 mRNA in hepatocytes and down-regulated it in macrophages (4). It is conceivable that a decrease of UCP2 in rats may contribute to enhanced macrophage immunity against bacterial infection as observed in UCP2 knockout mice (28). Direct Addition of Fenofibrate or WY14,643 in Hepatocyte Cultures Up-regulates UCP2 mRNA-Next, to examine whether UCP2 up-regulation observed in hepatocytes from fish oil feeding or fibrate administration was due to the direct effect of PPAR␣ activation in hepatocytes, PPAR␣ activators were directly added to primary mouse/rat hepatocytes culture, and their expression levels of UCP2 were examined (Table I). Fish oil was not examined because of its insolubility. Up-regulation of UCP2 by PPAR␣ activators was observed in both mouse and rat hepatocytes. Addition of 30 M fenofibrate or 50 M WY14,643, a specific activator of PPAR␣, to mouse hepatocytes up-regulated UCP2 by 5.4-and 5.5-fold, respectively; addition of 50 M WY14,643 to rat hepatocytes also up-regulated UCP2 by 4.5-fold. Thus, these data suggest that PPAR activators act directly on both mouse and rat hepatocytes and up-regulate UCP2 expression.
PPAR␣ Activators Up-regulate UCP2 mRNA via Increased Transcription-To examine whether PPAR-induced UCP2 mRNA increase was due to the increase of RNA transcription or the increase of mRNA stability, the effects of ␣-amanitin, polymerase II inhibitor, on the decay of UCP2 mRNAs in the presence and absence of WY14,643 were examined (Fig. 2). The decay of UCP2 mRNA by amanitin treatment was not altered in the presence of WY14,643. The data suggested that PPAR␣ activators up-regulated UCP2 mRNA via an increased transcription but not via an increased UCP2 mRNA stability.
Up-regulated UCP2 mRNA by PPAR␣ Activator Required de Novo Protein Synthesis-To examine whether PPAR␣-induced UCP2 mRNA required a newly synthesized protein, the effects of cycloheximide that inhibit protein synthesis were examined (Fig. 3). Addition of cycloheximide prevented WY14,643-induced UCP2 up-regulation (Fig. 3A). The data suggested that up-regulation of UCP2 by PPAR␣ activator requires a newly synthesized protein. As a positive control, the effects of cycloheximide on acyl-CoA synthetase (ACS) expression were also examined (Fig. 3B). ACS is one of the target genes of PPAR␣ activators (29). In contrast to UCP2, cycloheximide rather increased WY14,643-induced ACS up-regulation. In agreement with our study, a recent mouse UCP2 promoter study revealed that the double E box motif is required for PPAR␥-dependent FIG. 5. Effects of TNF␣ and fenofibrate on nuclear NF-Bbinding activity in mouse hepatocytes. Isolated hepatocytes from standard animal chow-fed mice was cultured as described under "Materials and Methods." A, nuclear proteins were isolated from hepatocytes that had been cultured for 1, 3, 17, and 24 h (fibrate (Fib.) only) in the presence of TNF␣ (10 ng/ml) or fenofibrate (30 M). As control of TNF␣ and fenofibrate treatment, PBS and Me 2 SO treatment, respectively, were included. Gel shift assay was performed using these 10-g nuclear protein and 32 P-labeled oligonucleotide fragments containing the NF-B-binding site. Nuclear extracts were separated by electrophoresis on 4.6% polyacrylamide gels. A typical autoradiogram (7-day exposure) is shown. B, to show that protein binding to labeled oligonucleotide probe is specific for the active form NF-B, nuclear proteins were isolated from hepatocytes that had been cultured for 3 h in the presence of TNF␣ (10 ng/ml) and then incubated with 32 P-labeled oligonucleotide fragments containing the NF-B-binding site (lane 2). Lane 1 shows labeled probe without adding nuclear extract. In competition assay, 80-fold excess of unlabeled oligonucleotide was used (lane 3). For mutational analysis, 32 P-labeled NF-B mutant oligonucleotide was used (lane 4). In supershift experiments, 1 g of goat, rabbit antisera against p50 or p65 protein was added, respectively (lanes 5 and 6).

FIG. 6. Effects of an antioxidant, PDTC on LPS-induced increases of UCP2 gene expression in mouse liver tissues.
Mice were injected intraperitoneally with sterile saline (as control), LPS (0.5 mg/kg body weight), or PDTC (250 mg/kg body weight) 30 min prior to intraperitoneal LPS injection. Mice were sacrificed 17 h after treatment to harvest the whole liver. Total RNA from liver was isolated, and transferred membrane sheets were probed with 32 P-labeled human UCP2 cDNA. A typical autoradiogram (1-day exposure) and its relative levels are shown. These mRNA levels were quantified using an image analyzer. The data for each band are shown as relative value to the mRNA level of the control saline group. Each data point represents mean Ϯ S.E. of 3-4 mice. Statistical differences are shown as p Ͻ 0.05 (*).
activation, but PPAR␥ does not bind to this region (30). Thus, PPAR␥ as well as PPAR␣ activates the UCP2 gene indirectly by altering the activity or expression of other transcription factors that bind to UCP2 promoters.
Expressions of PPAR␣, RXR␣, and PGC-1 mRNAs Are Not the Rate-limiting Step for PPAR␣ Activator-induced UCP2 Upregulation-To examine the involvement of PPAR␣, RXR␣, and PGC-1 for PPAR␣ activator-induced UCP2 up-regulation, these proteins were overexpressed transiently in rat hepatocytes, and UCP2 expression levels were measured in the absence and presence of WY14,643 (Fig. 4). Increased expression of PPAR␣/ RXR␣ mRNAs and/or PGC-1 did not affect UCP2 expression in both basal (without PPAR␣ activators) and WY14,643-stimulated conditions. Thus, expression levels of PPAR␣, RXR␣, and PGC-1 mRNAs are not the rate-limiting step for WY14,643induced UCP2 up-regulation in primary hepatocytes. Rather, other transcription factors and cofactors that are rapid turnover proteins might be involved in PPAR␣ activators-induced UCP2 mRNA increases.
ROS Generation Is Not a Major Cause for PPAR␣ Activatorinduced UCP2 Up-regulation in Hepatocytes-ROS generation by PPAR␣ activators through increased mitochondrial or peroxisomal oxidation may lead to up-regulation of UCP2 mRNA. To examine whether ROS produced in hepatocytes up-regulate UCP2 mRNA under our cultured conditions, the effects of an oxidant, tert-butyl-hydroperoxide (TBHP), on ROS production and UCP2 expression in hepatocytes were examined (Table II). TBHP induces mitochondrial permeability transition in hepatocytes by ROS production in hepatocytes (31). Generation of ROS in hepatocytes was evaluated with oxidation-dependent fluorogen 2Ј,7Јdichlorofluorescein diacetate (DCF-DA). Exposure of 100 M TBHP increased ROS production by 7.3-fold, but 30 M TBHP did not increase detectable ROS production. TBHP increased UCP2 mRNAs, but they were not dose-dependent. Exposure of 30 and 100 M TBHP increased UCP2 expression by 1.6-and 1.4-fold, respectively. The latter did not reach statistical significance. This might be due to some cell detachment (cell death) that was observed in 100 M TBHP concentration. WY14,643 treatment did not produce significant ROS but markedly up-regulated UCP2. These data indicated that ROS production might not be involved in PPAR␣ activators-induced UCP2 up-regulation in hepatocytes.
It is also known that ROS activate transcription factor NF-B and then alter various gene expressions (32). To examine whether NF-B activation was also involved in PPAR␣ activatorinduced up-regulation of UCP2 in hepatocytes, gel shift analysis was made using nuclear extracts from fenofibrate-treated hepatocytes (Fig. 5A). Whereas addition of TNF␣ to primary hepatocytes increased DNA binding activity of NF-B at 1, 3, and 17 h after TNF␣ exposure, fenofibrate addition did not increase DNA binding activity of NF-B up to 24 h under the condition that UCP2 was up-regulated. Specificity of DNA binding activity of NF-B in this labeled oligonucleotide was verified in the following experiment (Fig. 5B). In the absence of nuclear proteins, no protein-DNA complex was detected (lane 1). DNA binding with 32 P-labeled probe (lane 2) was markedly reduced in the presence of an excess of unlabeled oligonucleotide containing the NF-Bbinding site (lane 3). DNA binding with mutated 32 P-labeled probe was also markedly reduced (lane 4). Furthermore, addition of antiserum to p50 or p65 subunit NF-B reduced the intensity FIG. 7. Effects of an antioxidant, PDTC, on dietary oil feeding or fibrate administration-induced increases on TNF␣ (A) and UCP2 (B) mRNA in mouse liver tissues. PDTC (250 mg/kg body weight) was injected intraperitoneally just prior to initiation of feeding high safflower (Saf.) oil diet, high fish oil diet, high carbohydrate (Carb.) diet, and fenofibrate administered high carbohydrate (Carb.ϩFib.) diet to mice. Total RNA was isolated from livers at 17 h after PDTC injection. Total RNA were subjected to electrophoresis and transferred to nylon membrane. The membrane was hybridized with 32 P-labeled human TNF␣ (A) and human UCP2 (B) cDNAs. A typical autoradiogram (6day exposure for TNF␣ and 1-day exposure for UCP2) and its relative levels are shown. In the autoradiogram, each line represents a sample from an individual mouse. The radioactivity in each band was quantified using an image analyzer. The data for each group are shown as relative value to the mRNA levels of high carbohydrate-fed, saline-treated mice. Each data point represents mean Ϯ S.E. of 4 mice. Statistical differences are shown as p Ͻ 0.05 (*), p Ͻ 0.01 (**), and p Ͻ 0.001 (***). of the complex and produced supershifted complexes with a higher molecular mass (lanes 5 and 6, respectively). Thus, all these data indicate that PPAR␣ activators-induced up-regulation of UCP2 may not be due to NF-B activation.
PPAR␣ Activators Up-regulate Liver UCP2 mRNA in Mice via TNF␣-independent Pathway-Because it has been reported (4) that TNF␣ up-regulates UCP2 mRNA in hepatocytes, we examined whether PPAR␣ activators up-regulate UCP2 in vivo via TNF␣ production. It is conceivable that TNF␣ was produced in Kupffer cells from PPAR␣ activator-administered mice and then increased UCP2 expression of hepatocytes.
Peritoneal bacterial LPS injection in mice increases TNF␣ mRNA markedly in liver and blood TNF␣ concentration (33)(34)(35). It is also known that PDTC prevents TNF␣ production and also inhibits TNF␣ action by inhibition of transcription factor, NF-B (34,35). Indeed, in control non-treated mice, blood TNF␣ concentration was not detected, but LPS injection (0.5 mg/kg) markedly increased blood TNF␣ concentration up to 1160 pg/ml, and peritoneal PDTC injection (PDTC, 250 mg/kg) prior to TNF prevented LPS-induced increased blood TNF␣ production; TNF␣ concentration became 140 pg/ml (data not shown). Although increase of UCP2 mRNA by TNF␣ stimulation was relatively low, the levels of UCP2 mRNA changed in parallel with TNF␣ concentration; LPS treatment up-regulated UCP2 mRNA by 1.8-fold in whole liver, but PDTC treatment completely prevented its increase (Fig. 6).
Next, because fish oil feeding or fibrate administration did not increase detectable blood TNF␣ concentration, we meas-ured TNF␣ mRNA levels in mouse liver tissues (Fig. 7). Fish oil feeding for 17 h significantly up-regulated TNF␣ mRNA by 2-fold compared with safflower oil feeding (Fig. 7A). Fenofibrate administration increased TNF␣ mRNA by 1.4-fold compared with carbohydrate diet, but the increase did not reach statistical significance. PDTC treatment prior to either fish oil feeding or fenofibrate administration completely inhibited TNF␣ mRNA increase. However, PDTC treatment did not prevent fish oil feeding-induced UCP2 up-regulation (Fig. 7B). Although a significant decrease of UCP2 expression by PDTC treatment was observed in fenofibrate-fed mice, the decrease was very small. Thus, most of increased expression of UCP2 by fish oil feeding or fibrate administration was not due to PPAR␣ activator-induced TNF␣ mRNA increase. DISCUSSION In this study, we have shown that PPAR␣ activators, either fish oil feeding or fibrate administration, up-regulated UCP2 mRNA in hepatocytes both in mice and rats and that these effects were also observed in vitro. Up-regulation of UCP2 mRNA by PPAR␣ activators in hepatocytes was due to increased transcription and required a newly synthesized protein. In addition, PGC-1, ROS, and TNF␣ were not involved in this process.
PGC-1 is a transcriptional coactivator identified recently based on its ability to interact with PPAR␥ (36). In addition, PGC-1 coactivates NRF-1, a key transcription factor of nuclear and mitochondrial genes coding respiratory proteins (36,37) and also coactivates PPAR␣, a key transcription factor of mitochondrial and peroxisomal fatty acid oxidation enzyme gene expression (38). Forced overexpression of PGC-1 in C2C12 and in 3T3-L1 cells up-regulated UCP2 mRNA and PPRE-containing reporter gene expression, respectively, and in rat neonatal cardiac myocytes down-regulated UCP2 mRNA (14,38,39). In an in vitro study (38), glutathione S-transferase "pull-down" studies revealed that in contrast to the ligand-independent interaction with PPAR␥, PGC-1 binds PPAR␣ in a ligand-dependent manner. However, in the case of primary hepatocytes, major PPAR␣ target cells, overexpression of PGC-1 did not up-regulate UCP2 mRNA (Fig. 4). The data suggest that PGC-1 as well as PPAR␣/RXR␣ is not a rate-limiting step of UCP2-mRNA expression. In hepatocytes, PGC-1 may be preferentially recruited to nuclear receptors, such as glucocorticoid receptor, hepatic nuclear factor-4␣ (HNF-4) rather than PPAR␣, to promote gluconeogenesis (40).
Increased ROS production might be a clue to find a critical molecule responsible for UCP2 up-regulation. ROS can be produced through increased fatty acid peroxisomal and mitochondrial ␤-oxidation (18). In addition, because macrophages from UCP2 knockout mice generate more ROS, the role for UCP2 in the limitation of ROS and macrophage-mediated immunity has been proposed (28). These data suggested that ROS production might be a cause of up-regulation of UCP2. However, our study suggested that ROS generation by PPAR␣ activators might not be a cause of UCP2 up-regulation. Direct addition of PPAR␣ activators to isolated hepatocytes did not increase detectable ROS but up-regulated UCP2 mRNA by 7-fold, while as a positive control, an oxidant, TBHP treatment up-regulated UCP2 expression by 1.6-fold with increased ROS production (Table  II). In addition, the evidence that PPAR␣ activator did not activate NF-B, which is activated by ROS, further supports this hypothesis (Fig. 5). Although it is unlikely, however, the possibility is not ruled out that ROS, which was not detected by the currently employed ROS detection method, may up-regulate UCP2 mRNA.
Our data indicated that PPAR␣ activators up-regulated UCP2 via hepatocytes but not via increased TNF␣ production An oxidant, TBHP produced ROS in hepatocytes and then up-regulated UCP2. However, PPAR␣ activators up-regulated UCP2 via a different pathway. PPAR␣ activators target hepatocytes directly, activate the PPAR␣/RXR␣ heterodimer complex, and then via unknown proteins up-regulate UCP2 mRNA. Oral administration of PPAR␣ activators increased TNF␣ production slightly in Kupffer cells, but TNF␣ did not contribute to most of PPAR␣ activators-mediated UCP2 up-regulation. from Kupffer cells for the following reasons. First, direct addition of fibrate to hepatocyte cultures up-regulated UCP2 markedly (Table I and Figs. 2-4). Second, because fish oil feeding and fenofibrate administration did not increase blood TNF␣ concentrations but slightly increased their message levels (Fig.  7A), the contribution of TNF␣ in Kupffer cells to hepatocytes UCP2 up-regulation might be very small. Third, an anti-oxidant, PDTC pretreatment that prevents TNF␣ production, did not affect most of the PPAR␣-induced UCP2 up-regulation (Fig.  7B). As activated Kupffer cells are known to release several cytokines (e.g. epidermal growth factor, TNF␣, and hepatocyte growth factor), superoxide, nitric oxide, and eicosanoids could theoretically be involved in regulation growth and metabolic modulation of nearby hepatocytes (41). In agreement with our study, in which PPAR␣ activators increased liver TNF␣ production, oral administration of WY14,643 in rats moderately elevated the level of TNF␣ mRNA and cell replication in the liver, and antibodies to TNF␣ prevented increases in cell replication in liver due to WY14,643 (42). In addition, inactivation of Kupffer cells with methyl palmitate completely prevented the mitogenic effects of WY14,643 but did not affect WY14,643induced peroxisome proliferation (43). These data support that PPAR␣ activators increased TNF␣ production in Kupffer cells but did not affect hepatocyte, which might lead to up-regulation of UCP2 mRNA, possibly due to its small amount.
In Fig. 8, a proposed model of PPAR␣ activator-induced UCP2 gene expression in liver is presented. In hepatocytes, PPAR␣ activators, ROS, and TNF␣ can increase UCP2 expression. Oral administration of PPAR␣ activators increased TNF␣ production slightly in Kupffer cells, but TNF␣ did not contribute to most of the PPAR␣ activator-mediated UCP2 up-regulation. PPAR␣ activators directly affect hepatocytes and then up-regulate UCP2 mRNA. This step requires additional proteins other than PPAR␣/RXR␣ and PGC-1. These proteins might be other cofactors and transcription factors, in which their expressions might be regulated by PPAR␣ activators.