| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Papers In Press, published online ahead of print May 1, 2006
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
activators and fasting induce the expression of adipose differentiation-related protein in liver
Department of Nutrition, Institute of Basic Medical Sciences, University of Oslo, N-0316 Oslo, Norway
The online version of this article (available at http://www.jlr.org) contains additional figures. 
Published, JLR Papers in Press, February 17, 2006.
1 S. M. Ulven and B. M. Arntsen contributed equally to this work.
2 To whom correspondence should be addressed. e-mail: h.i.nebb{at}medisin.uio.no
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
(PPAR) in mouse liver and rat and human hepatoma cells through a highly conserved direct repeat-1(DR-1) element. Although the ADFP mRNA is highly increased by a synthetic PPAR agonist, the ADFP protein is only substantially increased in cells containing LDs, such as hepatocytes incubated with fatty acids, and in livers of fasted mice. ADFP is induced by fasting even in the absence of a functional PPAR, in marked contrast to the PPAR target gene acyl-coenzyme A oxidase-1. Activation of LXRs, which stimulates LD formation through the activation of lipogenesis, does not affect ADFP mRNA levels. TIP47, another PAT member known to be expressed in liver, was unaffected by all treatments. This constitutively expressed PAT member seems to be less transcriptionally regulated than ADFP. These observations suggest that ADFP is primarily a fasting-induced protein in liver that coats the newly synthesized triacylglycerol-containing LDs formed during fasting.
Supplementary key words adipose differentiation-related protein • adipophilin • tail-interacting protein of 47 kDa • fatty liver • direct repeat-1 • peroxisome proliferator-activated receptor • peroxisome proliferator-activated receptor response element • promoter • triacylglycerol • lipid droplet
Abbreviations: ACO-1, acyl-coenzyme A oxidase-1; ADFP, adipose differentiation-related protein; CE, cholesterol ester; CMC, carboxymethyl cellulose; DR-1, direct repeat-1; LD, lipid droplet; ME-1, cytosolic malic enzyme-1; OA, oleic acid; PPAR, peroxisome proliferator-activated receptor; PPRE, peroxisome proliferator-activated receptor response element; RXR, retinoid X receptor; SREBP-1, sterol-regulatory element binding protein-1; TAG, triacylglycerol; TIP47, tail-interacting protein of 47 kDa; TTA, tetradecylthioacetic acid
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Vimentin (4), perilipin (5, 6), and mouse adipose differentiation-related protein (ADFP; adipophilin in humans) (7) were the first proteins experimentally demonstrated to associate with the LD surface. This cellular location was more recently also experimentally confirmed for tail-interacting protein of 47 kDa (TIP47) (8, 9) and S3-12 (10). Perilipin, ADFP, and TIP47 exhibit high sequence identity within an N-terminal motif termed PAT-1 (for perilipin, ADFP, and TIP47) and a more distally located PAT-2 domain (9, 11). The N-terminus of S3-12 shares limited identity with the PAT-1 motif, but the protein shares sequence homology with ADFP and TIP47 in the PAT2 domain and the C terminus (12). S3-12 and the PAT proteins have amphipathic 11-mer helical repeats in common. These repeats are also found in other lipid binding proteins, such as synucleins, apolipoproteins, phosphate cytidyltransferases, and dehydrins (13). The PAT proteins and S3-12 are expressed in distinct tissues, suggesting that they have evolved to fine-tune lipid metabolism according to the particular needs of these tissues (12).
At present, the LD-associating properties have only been thoroughly studied for ADFP and perilipin (14, 15). Perilipin expression is confined to adipose tissue and steroidogenic cells, where the protein associates with TAG and CE droplets, respectively (5, 16, 17). Perilipin has been found to have an important role in the hormonal control of basal and stimulated lipolytic activity within adipose tissue (18, 19). Except for the well-established targeting to LDs (1, 7, 9, 15), less is known about the function of ADFP. Initially, when cloned, ADFP was thought to be preferentially expressed in adipocytes, with increasing mRNA expression during adipocyte differentiation (20). Later analyses, however, have demonstrated that ADFP mRNA is more ubiquitously expressed, with tissue enrichment in metabolic organs and tissues (7, 12, 19). The ADFP protein associates with smaller neutral LDs located within most tissues, but rarely in adipose cells that express perilipin (7, 19). Forced expression of ADFP in COS-1 cells stimulates long-chain fatty acid (FA) uptake (21), and stimulation of cells in culture with FAs [i.e., oleic acid (OA)] induces the formation of LDs coated with ADFP (7, 9, 22).
The cellular content of ADFP mRNA is altered by lipids or their metabolites. Incubation of adipogenic cells in vitro by long-chain FAs increases ADFP mRNA content (23). Other lipids, such as oxidized low density lipoprotein (24) and VLDL, increase ADFP mRNA content in macrophages (25). In addition, synthetic ligands for peroxisome proliferator-activated receptor ß/
(PPARß/) have been found to increase ADFP mRNA in macrophages (25–27). The PPAR family consists of PPARß/ and two other isotypes, PPAR and PPAR
. They belong to a subfamily of nuclear receptors that heterodimerize with retinoid X receptors (RXRs) and regulate transcription by binding to specific peroxisome proliferator-activated receptor response elements (PPREs) in the promoter region of target genes (28). PPAR is highly enriched in white adipose tissue and macrophages (29). PPAR is enriched in liver and other metabolic tissues, such as muscle, heart, and kidney (30). PPARß/ is more ubiquitously expressed (31). The PPAR isotypes play distinct roles in fatty acid transport and metabolism. PPAR is irreplaceable for a proper response to fasting in liver (32) and stimulates the transcription of numerous genes encoding for proteins in the FA oxidation pathways (33).
We have previously demonstrated that transcription of S3-12 and perilipin in adipocytes is stimulated by PPAR (12). Several independent studies suggest that the expression of ADFP is similarly stimulated by lipid metabolites that are believed to activate members of the PPAR family (23, 24, 34, 35). However, to date, we have not been able to demonstrate a direct regulation of ADFP by PPAR in adipose tissue (12). Although highly transcribed in adipose cells, the ADFP protein is almost undetectable in the cytosol or at the surface of LDs when perilipin is expressed (7, 19). Therefore, the presence of ADFP might be more relevant in liver, where ADFP is expressed at high levels and perilipin is not normally expressed (5, 7, 12). By the use of animal and cell culture experiments, we demonstrate that although ADFP mRNA is highly induced by PPAR activators, the increased mRNA does not necessarily result in more cellular ADFP protein. ADFP mRNA and protein levels, however, are highly induced by fasting, a physiological condition that also favors the accumulation of TAG-containing LDs. This accumulation of TAGs seems to be a major regulator of cellular ADFP protein content.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, pSG5-mRXR, and pSG5 (Stratagene, La Jolla, CA) expression vectors were provided by Jan-Åke Gustafsson (Department of Bioscience and Medical Nutrition, Novum, Huddinge, Sweden). LG100268 and GW501516 were kindly provided by J. Lehmann (CareX, Strasbourg, France). Tetradecylthioacetic acid (TTA; C14-S-acetic acid:0) (36) was kindly provided by Jon Bremer (University of Oslo, Norway).
BSA binding of fatty acids
Stock solutions of OA (#O1008), -linolenic acid (1,12,15-octadecatrienoic acid; #L2376), cis-4,7,10,13,16,19-docosahexaenoic acid (#D2534), or arachidonic acid (#A9673) were made by adding ethanol to 25 mg/ml final concentration. To make FA (6 mM)/BSA (2.4 mM) stock solutions, ethanol was removed by evaporation under nitrogen, and the dried FAs were dissolved in 100 µl of 0.1 M NaOH before BSA binding in 1.5 ml of 2.4 mM fat-free BSA (Sigma; #A6003) at 50°C for 5 min. TTA was dissolved in 1 M NaOH before binding to BSA. FA/BSA stock solutions were sterile-filtered through 0.22 µm µStar LB filters (Costar; #8110) and stored at –20°C for no longer than 3 weeks before use.
Culturing, transfection, and lipid staining of cells
COS-1 cells (American Type Culture Collection) were cultured in high-glucose-containing DMEM (Sigma; #6546). Human hepatoma HepG2 cells (American Type Culture Collection) were cultured in minimum essential medium Eagle (Oh Biocoat (BD Biosciences, Bedford, MA) Sigma; #M2279). Rat hepatoma 7800-C1 cells (37) were cultured in low-glucose-containing DMEM (Sigma; #D6046). All culture media were supplemented with 10% fetal calf serum, 2 mM L-glutamine, penicillin (50 U/ml), and streptomycin (50 µg/ml), and cells were incubated at 37°C in 5% CO2. For experiments, 7800-C1 cells were seeded in 12-well dishes (2 x 105 cells/well) and cultured for 2–3 days before the start of experiments. All stimulations were performed in 2 ml medium/well.
In transient transfection experiments, COS-1 cells were seeded in six-well dishes with reporter (5 µg) and cotransfected with pSV-ß-galactosidase (3 µg, as internal control) and pSG5, pSG5-mRXR, and/or pSG5-mPPAR expression vectors (1 µg each) with calcium phosphate precipitation (38). HepG2 cells were seeded at 70% confluency in 12-well dishes in antibiotic-free medium 24 h before transfection with reporter (400 ng), pSV-ß-galactosidase (250 ng), and expression vectors (80 ng of each). DNA and 2 µl of Lipofectamine2000 (Invitrogen Life Technologies, Carlsbad, CA) were dissolved in 2 x 100 µl of OptiMEM-I (Invitrogen Life Technologies), mixed, and incubated for 20 min before transfection of cells in 2 ml of OptiMEM-I for 6 h. After transfection, culture medium was replaced with growth medium containing FAs and synthetic ligands for 24 h. Cells were harvested in 250 µl (COS-1) or 100 µl (HepG2) of reporter lysis buffer (Promega), and luciferase activities were measured with a TD-20/20 Luminometer (Turner Designs, Sunnyvale, CA). Relative luciferase activity was normalized against ß-galactosidase activity (control = 1). All transfections are representative of at least three individual experiments performed in triplicate. Error bars represent standard deviations.
Oil Red O (0.5%) was dissolved in isopropanol and filtered to remove unsolved particles. Before use, Oil Red O solution was mixed with MilliQ water, (60:40) to make Oil Red O coloring reagent. For lipid staining, cells were fixed in 4% paraformaldehyde/PBS for 15 min, washed with PBS, incubated with Oil Red O coloring reagent for 15 min, and finally washed in PBS to remove unincorporated stain.
Preparation and analysis of RNA
Total RNA was extracted with TRIZOL® reagent (Invitrogen). RNA (10 or 20 µg) was separated on a 1% agarose formaldehyde/MOPS gel and blotted and hybridized as described (38). 36B4 (acidic ribosomal phosphoprotein P0) mRNA (39) was used as a loading control. Probes were generated by radiolabeling of cDNAs with [-32P]dCTP (Amersham Biosciences, Buckinghamshire, UK) by use of the Megaprime DNA Labeling System (Amersham Biosciences). Northern blots were visualized by Hyperfilm MP, scanned with Personal Densitometer SI, and analyzed using ImageQuantTM software (Amersham Biosciences).
Full-length or partial cDNA sequences used to generate radiolabeled probes have been described elsewhere [m-ADFP, m-S3-12, h-perilipin, m-aFABP/aP2 (12), m-SREBP-1, h-LXR (38)] or were cloned by PCR with the following primers: r-ACO-1-s, 5'-CACGCAATAGTTCTGGCTCA-3'; r-ACO-1-a, 5'-GCCATTCGACATTCTTCGAT-3'; m-ACO-1-s, 5'-AAGCCTGACGGCACGTATGTAAA-3'; m-ACO-1-a, 5'-CAGAAGTCAAGTTCCACGCCACT-3'; m-ME-1-s, 5'-GGATGTCGTCAAGGCTATTGTGG -3'; and m-ME-1-a, 5'-GCTGGTCGGACTACTCAAAGCAA-3'.
Electrophoretic mobility-shift assays
Proteins were synthesized from pSG5-mPPAR, pSG5-mRXR, and pSG5 expression vectors using the TNT T7-coupled in vitro transcription/translation system (Promega). Oligonucleotide probes were labeled using T4 polynucleotide kinase (Promega) and [-32P]ATP (Amersham Biosciences). Binding and separation of protein-DNA complexes from free probe was performed as described (38). Oligos used are as follows (only upper primer is shown): m-ADFP-DR-1-s, 5'-GGCCTTTTGTAGGTGAAAGGGCAAAGAAAGAGG-3'; m-ADFP-DR-1-s-mut, 5'-GGCCTTTTGTAccGAcAcGCAAAGAAAGAGG-3'; h-adipophilin-DR-1-s, 5'-ATTT-TGTAGGTGAAAGGGCGAGAGTCT-3'; and h-adipophilin-DR-1-s-mut, 5'-ATTTTGTAcTGAcAcGCGAGAGTCT-3'.
Identification, cloning, and mutagenesis of promoters
The complete mouse ADFP gene and a short 5' upstream promoter sequence have previously been cloned and partially characterized (40). Full-length cDNA sequences were used to search nonredundant and high-throughput genomic sequence databases by BLAST (41) to obtain longer promoter sequences [human adipophilin mRNA (accession number BC005127) annealed to htgs: #AC024502.2, and mouse ADFP mRNA (accession number L09734.1) annealed to htgs: #AL824707.5]. The human and mouse ADFP promoters were scanned with a consensus PPRE (RGGBSAAAGGTCA) with the use of the GCG program package (Wisconsin Package version 10.0; Genetics Computer Group) and compared by BLAST2 alignment (42).
Full-length promoters were amplified with PfuTurbo (Stratagene) with mouse (m; Clontech; #6650-1) or human (h; Clontech; #6550-1) genomic DNA as template with PCR settings as described (38). Primers used were as follows: 5-h-adipophilin-promoter (KpnI), 5'-TAGGTACCAAAAACGTCTCCTTTGTCCTCTGGA-3'; 3-h-adipophilin-promoter (KpnI), 5'-TAGGTACCTCTAACGCGTTTCCCTTTCGATAAT-3'; 5-m-ADFP-promoter (NheI), 5'-ATGCTAGCATCGCCTTGGGATCTAATCTTGGT-3'; and 3-m-ADFP-promoter (NheI), 5'-ATGCTAGCCTAACAGGAGAGCTGAGGGACGAG-3'. The deletion construct for the human adipophilin reporter was generated by restriction digestion with PvuII followed by religation of the vector. Site-directed mutagenesis of the DR-1 element was performed with PCR as described (38) with the oligonucleotides used in the electrophoretic mobility-shift assays as targeting primers.
Protein measurements and Western blot analysis
Frozen mouse liver tissue (25 mg) was homogenized in 1 ml of lysis buffer [1x PBS, 1% Nonidet P-40, 0.1% SDS, and 1x Compete (Boehringer Mannheim)] using a Mini-BeadbeaterTM (BioSpec Products, Inc., Bartlesville, OK) for 20 s at 4,200 rpm with 0.5 mm glass beads (BioSpec Products). Cell homogenate from 7800-C1 cells (100 µl) was homogenized by sonication for 2 x 2 s. Protein concentrations were measured with the BC assay method (Interchim; #FT-40840).
Proteins (10 µg from tissue/50 µg from cells) were denatured by boiling for 5 min in loading buffer (final concentration, 40 mM Tris-HCl, pH 6.8, 5% glycerol, 2% SDS, 0.002% bromphenol blue, and 0.1 M mercaptoethanol), resolved on a 10% Tris-HCl SDS gel (CriterionTM Precast Gel; Bio-Rad), and transferred to a polyvinylidene difluoride membrane (Hybond-P; Amersham Biosciences; #RPN 303F) by electrotransfer (2.5 mM Tris-HCl, 19 mM glycine, and 20% methanol). Five microliters of Precision Plus ProteinTM Standard (Bio-Rad; #161-0374) was used as molecular weight marker. The membranes were blocked (PBS, 5% nonfat dry milk, and 0.1% Tween) overnight at 4°C before incubation with primary antibody [guinea pig anti-adipophilin diluted 1:2,000 (Research Diagnostic; #RDI-PROGP40) against mouse samples, and mouse anti-adipophilin diluted 1:200 (#RDI-PRO610102) against rat samples] for 1 h at room temperature. The membrane was washed three times (PBS containing 0.1% Tween) followed by incubation with secondary antibody [horseradish peroxidase-labeled goat anti-guinea pig antibody (Southern Biotech; #6090-05) or anti-mouse antibody (#1031-05)] diluted 1:8,000 for 1 h. Finally, the membranes were washed again and developed using enhanced chemiluminescence (ECLplus; Amersham Biosciences; RPN 2132) and visualized with Hyperfilm MP (Amersham Biosciences). To ensure equal loading of protein, the membranes were stripped and incubated with primary antibody against ß-actin (diluted 1:10,000; Sigma; #A5441).
Measurement of lipid composition in liver and 7800-C1 cells
Liver tissue samples (20–50 mg) were homogenized in PBS (1.5 ml) as described for Western blot analysis. Lipids were extracted from the liver homogenate (200 µg of protein) and separated by high-performance thin-layer chromatography essentially as described (43, 44). In brief, lipids were extracted from the homogenate by organic phase extraction using methanol-chloroform and dried under nitrogen. Cholesterol acetate (2 µg) was added during homogenization as an internal standard. Samples were resolved in 90 µl of chloroform and spotted to an activated high-performance thin-layer chromatography plate [Nano-sil-20 (10 cm x 20 cm); Alltech Associates, Inc.], and lipid fractions were separated in 87% hexane, 10% diethyl ether, and 3% isopropanol in a CAMAG Horizontal Developing chamber (Camag, Switzerland). The lipid fractions were visualized with acidic Mg solution and quantified by scanning the plates using the Gel Logic 100 Imaging system (Eastman Kodak Co., Rochester, NY) and integrating the density areas using Kodak 1D version 3.6.1 software.
7800-C1 cells were harvested in dH2O, frozen to disrupt the cell membrane, and homogenized by ultrasound treatment (2 x 5 s at a working frequency of 47 kHz ± 6%) before TAG measurement using the Triglycerides Enzymatique PAP 150 kit (BioMérieux; #61236) essentially as described. TAG concentration was related to protein concentration.
Animal experiments
All animal use was approved and registered by the Norwegian Animal Research authority. Mice were maintained in a temperature-controlled (22°C) facility with a strict 12 h light/dark cycle and given free access to normal chow diet (Special Diets Services No. 3, containing 64% carbohydrates, 31.5% proteins, and 2.0% fat; Scanbur BK AS) and water. WY-14643 and T0901317 were dissolved in vehicle [1% (w/v) carboxymethyl cellulose (CMC) dissolved in water]. Each mouse was gavaged with 10 µl vehicle/g body weight. Mice were euthanized by cervical dislocation, and tissues were rapidly dissected, quickly frozen in liquid nitrogen, and stored at –80°C.
C57Bl/6J mice (B&K Universal, Ltd., Sollentuna, Sweden; 10 weeks, 25 g) were fed twice (24 and 8 h before they were euthanized) with vehicle (CMC) or WY-14643 (2 x 50 mg/kg). The PPAR-deficient (PPAR–/–) mice (Sv/129, Jae substrain), obtained from the Jackson Laboratory (Bar Harbor, ME), have been described (33). 129/SvEvTac@Bom (Bomholtgård, Denmark) was used as a wild-type control (PPAR+/+). In the 7 day feeding experiment, PPAR+/+ and PPAR–/– mice (12–18 weeks, 25–35 g) were fed daily with vehicle (CMC) or WY-14643 (10 mg/kg), with the last dose given 4 h before mice were euthanized.
In the 24 h feeding experiment, PPAR+/+ and PPAR–/– mice (12–13 weeks, 25–30 g) were fed twice (24 and 8 h before they were euthanized) with vehicle (CMC), T0901317 (2 x 10 mg/kg), or WY-14643 (2 x 50 mg/kg).
In the fasting/refeeding experiment, male PPAR+/+ (7–9 weeks, 20–25 g) and PPAR–/– mice (6–8 weeks, 20–25 g) were divided into three groups: nonfasted (control), fasted, and refed. The control group was fed ad libitum, the fasted group was fasted for 24 h, and the refed group was fasted for 24 h and then refed for 12 h. The experiment was staggered to ensure that all mice were euthanized at the same time of the day at the end of the dark period.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
agonist in rat hepatoma 7800-C1 cells is the most active PPAR isotype in liver (31), and to test if activation of PPAR affects ADFP expression in liver, we stimulated rat 7800-C1 hepatoma cells (37) with increasing concentrations of a potent synthetic PPAR agonist (WY-14643) for 24 h. A robust induction of the ADFP transcript was observed at 10 µM concentration (Fig. 1A
). The same was observed for the two characterized PPAR target genes acyl-coenzyme A oxidase-1 (ACO-1) (45) and cytosolic malic enzyme-1 (ME-1) (46, 47). The induction of ADFP mRNA by WY-14643 (10 µM) was initiated already at 4–6 h, similar to ACO-1 and ME-1, suggesting a direct effect of PPAR activation on ADFP transcription (Fig. 1B). Stimulation with the RXR agonist LG100268 (1 µM), which activates the heterodimeric partner of PPAR, increased the transcription of ACO-1 and ME-1 similar to what was observed for WY-14643, but LG100268 had less effect on ADFP mRNA transcription (Fig. 1C). TIP47, another PAT member expressed at significant levels in liver (12), was not regulated by any of these treatments.
|
regulates ADFP expression in vivo, we treated C57Bl/6J mice with vehicle (CMC) or the synthetic PPAR agonist (WY-14643; 2 x 50 mg/kg) for 24 h. Liver samples were dissected and subjected to RNA analysis. A strong induction of the ADFP, ACO-1, and ME-1 transcripts was observed after administration of the PPAR agonist in liver (Fig. 2A
, B). As found also in 7800-C1 cells, the induction of ADFP mRNA by WY-14643 treatment was much higher compared with the induction observed for the PPAR target genes ACO-1 and ME-1.
|
. Therefore, we analyzed the human adipophilin and the mouse ADFP promoter for the presence of a typical DR-1-like PPRE. Only one conserved direct repeat-1 (DR-1) element, located 
2,000 bp upstream of the transcription start site, was identified (see supplementary Fig. IA, B). This element was found to bind all three PPAR isotypes in electrophoretic mobility-shift assays (see supplementary Fig. IC; results not shown). We next cloned the mouse ADFP (–2,781 to +50) and the human adipophilin (–3,592 to +415) promoters into the pGL3-basic luciferase reporter vector and transfected these reporters transiently into COS-1 cells to test the functionality of the identified DR-1 element. These analyses established the DR-1 element as a functional PPRE (see supplementary Fig. II).
Transcriptional stimulation of the ADFP gene by long-chain FAs depends on the DR-1 element
Long-chain FAs increase ADFP mRNA levels in adipocyte precursor cells (23), but the molecular mechanism involved has not been investigated. Long-chain FAs are believed to function as physiological PPAR ligands (35); hence, we tested whether selected long-chain FAs, such as OA (C18:1; 100 µM), -linolenic acid (C18:3; 100 µM), arachidonic acid (C20:4; 50 µM), docosahexaenoic acid (C22:6; 50 µM), and the FA analog TTA (C14-S-acetic acid:0; 50 µM) (36), were able to stimulate transcription of the ADFP gene. Wild-type and DR-1-mutated mouse ADFP reporter constructs were transfected into HepG2 cells in the absence or presence of RXR and PPAR expression vectors and stimulated with WY-14643, TTA, and the FAs listed above (Fig. 3
). Only the wild-type reporter responded to the added FAs and the synthetic PPAR ligands, suggesting that the FA regulation of ADFP transcription is mediated through the identified DR-1 element.
|
–/– mice at lower doses (35), but in vitro analyses suggest that it can activate PPAR at higher doses (
100 µM) (48). To determine if the effect seen in WY-14643 treatment is mediated exclusively through PPAR, we fed PPAR+/+ and PPAR–/– mice with vehicle (control) or WY-14643 (10 mg/kg) daily for 1 week. In agreement with previous results (33), WY-14643 treatment resulted in a marked increase in liver size in PPAR+/+ mice but not in PPAR–/– mice (Table 1
). WY-14643 strongly induced the expression of ADFP, ACO-1, and ME-1 mRNA in the liver of PPAR+/+ mice but had no effect in PPAR–/– mice (Fig. 4A
and results not shown). This demonstrates that the effect obtained by WY-14643 on ADFP mRNA levels is dependent on the presence of an intact PPAR protein. The mRNAs for TIP47 (results not shown) and RXR remained unchanged by the treatment or disruption of the PPAR gene.
|
|
+/+ mice, whereas no change in ADFP protein content was observed in PPAR–/– mice (Fig. 4B). Analysis of the hepatic lipid content revealed no evident differences in the lipid classes among the mice (Fig. 4C).
ADFP mRNA and protein content are increased during fasting and refeeding independent of PPAR
PPAR has been ascribed an important role in the adaptive response to fasting (32, 33). Fasting is a powerful stimulator of adipose tissue lipolysis, which gives rise to a profound increase in plasma FFA content. These FFAs are readily taken up by the liver and oxidized, converted to ketone bodies, or incorporated into LDs. Such LDs have been found to accumulate in the liver of both fasted wild-type and PPAR–/– mice (32). Because our previous analyses established ADFP as a direct PPAR target gene, we analyzed the effect of PPAR and fasting/refeeding on ADFP mRNA and protein levels and hepatic lipid content. PPAR+/+ and PPAR–/– mice were fed ad libitum (control), fasted for 24 h (fasted), or fasted for 24 h and refed for 12 h (refed). Surprisingly, the ADFP mRNA was highly induced by fasting in both PPAR+/+ and PPAR–/– mice, with a prolonged higher expression after refeeding compared with control fed mice (Fig. 5A
). This result differed from those seen in ACO-1 and PPAR mRNAs, which were only induced by fasting in the PPAR+/+ mice. Yet another expression pattern was observed for the ME-1 mRNA, which was relatively unaffected by fasting or removal of PPAR but increased slightly by refeeding. The expression of ME-1, however, was very similar to the expression of RXR. The TIP47 mRNA remained relatively unchanged by the fasting and refeeding treatments.
|
+/+ mice compared with PPAR–/– mice, which suggests that the presence of PPAR favors accumulation of the ADFP protein. Analysis of the liver lipid content showed a considerable increase in accumulated lipids, mainly as TAG, in fasted and refed mice compared with control fed mice. The amount of TAG accumulated in PPAR+/+ and PPAR–/– mice was similar, but the net increase was larger in PPAR+/+ mice, because of the higher TAG content in control fed PPAR–/– mice. Nevertheless, the similar amount of TAG found in fasted and refed PPAR+/+ and PPAR–/– mice suggests that a functional PPAR gene is not required for TAG accumulation during fasting (Fig. 5C).
Stimulation of 7800-C1 cells with long-chain FAs induces ADFP mRNA and protein content and LD formation
The two in vivo experiments described above reveal an interesting difference in ADFP protein levels. Even though both WY-14643 treatment and fasting stimulate ADFP mRNA content to approximately similar levels, fasting seems to favor a much higher increase in ADFP protein content. This suggests that physiological changes that occur during fasting favor accumulation of the ADFP protein. The ADFP protein furthermore seems to be stabilized during the refeeding period, when ADFP protein level remains high, even though the ADFP mRNA declines to basal levels.
Fasting and refed states are almost contradictory when it comes to hormonal signaling. This argues against a hormonal regulation of ADFP protein accumulation. However, it is well established that ADFP associates on the surface of LDs (1, 7, 9, 15). TAG-containing LDs are present in the liver during both fasting and refeeding, which suggests that LDs might favor accumulation of the ADFP protein. To test this hypothesis, we stimulated rat 7800-C1 cells in vitro with selected long-chain FAs, such as OA (100 µM), linolenic acid (100 µM), arachidonic acid (50 µM), docosahexaenoic acid (50 µM), and the FA analog TTA (50 µM) (36), and measured ADFP mRNA and protein levels and LD content.
The ADFP mRNA was moderately increased after 8 h of stimulation by most FAs, whereas the same FAs had more effect after 24 h (Fig. 6A
). The polyunsaturated long-chain FAs (-linolenic acid, arachidonic acid, and docosahexaenoic acid) and TTA were more potent compared with the monounsaturated FA (OA). This fatty acid-selective effect on induction of the ADFP mRNA is in agreement with the results obtained from the transfection studies (Fig. 3). No alteration of TIP47 mRNA was found with any of the fatty acids. All of the selected FAs increased the ADFP protein content (Fig. 6B). We next analyzed the intracellular lipid pool by coloring the cells with Oil Red O. The cells stimulated with FAs contained larger and more LDs compared with control cells (receiving BSA) or cells stimulated with TTA (Fig. 6C). These LDs are likely formed by an increased uptake of FAs from the medium and incorporation of these into LDs and are not attributable to increased de novo lipogenesis, because the mRNAs for lipogenic genes such as sterol-regulatory element binding protein-1 (SREBP-1) and FAS were not affected by the addition of the FAs (Fig. 6A and results not shown). Interestingly, no evident correlation among ADFP mRNA and ADFP protein levels were found after stimulation with the selected FAs. As examples, OA treatment resulted in the largest increase in ADFP protein levels but barely stimulated ADFP mRNA levels. The opposite effect was found for TTA. A much better correlation was observed among ADFP protein levels and the amount of accumulated Oil Red O-colored lipids.
|
antagonist (MK-886), a PPAR agonist (GW501516), and a RXR agonist (LG100268).
After 72 h of stimulation, accumulated lipids were stained with Oil Red O. Cells cultured in BSA and stimulated with WY-14643 (PPAR agonist), LG100268 (RXR agonist), or T0901317 (LXR agonist) increased the lipid content, whereas the lipid content in cells stimulated with GW501516 (PPARß/ agonist) or MK-886 (PPAR antagonist) remained unchanged (Fig. 7A
). The lack of any effect by the PPARß/ agonist compared with the PPAR agonist is likely attributable to the low expression of PPARß/ compared with PPAR in these cells (results not shown). In agreement with the findings described above (Fig. 6C), supplementation of the culture medium with OA increased the lipid content. Interestingly, stimulation with both OA and WY-14643 resulted in an additional increase in accumulated lipids. Stimulation with LG100268 and T0901317 also increased the lipid content compared with OA alone, but not as much as observed for WY-14643. GW501516 and MK-886 had no clear effect. We expect that the variable degree of Oil Red O coloring is the result of the unequal ability of the cells to take up OA and incorporate it into TAG-containing LDs. Therefore, we measured the TAG content in cells stimulated as described above. The intracellular TAG content corresponded well with the degree of Oil Red O coloring (Fig. 7B).
|
At the protein level, treatment with WY-14643 increased ADFP protein content, which was further increased with a combination of WY-14643 and OA treatment (Fig. 7D). Essentially the same effects were observed with LG100268 treatment, but the amount of ADFP protein in the samples was lower. T0901317 treatment had no effect on ADFP protein content in the absence of OA, but a weak increase was observed in combination with OA. MK-888 and GW501516 had no particular effect on ADFP protein levels.
A comparison of the ADFP mRNA, ADFP protein, and TAG accumulation showed a much better correlation among ADFP protein and cellular TAG content than between ADFP protein and ADFP mRNA levels. However, despite the clear accumulation of LDs in T0901317-stimulated cells, the level of ADFP protein was much lower than expected compared with the amount of ADFP protein found in WY-14643- and LG100268-treated cells. Therefore, we examined the level of ADFP protein in 7800-C1 cells incubated with BSA or OA and stimulated with WY-14643 or T0901317 for 24 or 72 h. The amount of ADFP protein remained stable in cells cultured in the presence of either WY-14643 or T0901317, but a reduction in ADFP protein was found in cells treated with OA alone (Fig. 7E). This suggests that activation of the lipogenic pathway through LXRs does not affect the transcription of ADFP, but the increase in FA synthesis and LD formation with prolonged stimulation seems to have a stabilizing effect on the ADFP protein content.
Liver ADFP mRNA is not regulated upon stimulation of de novo lipogenesis by activation of LXRs
It is impossible to study the complex regulations involved in fasting and de novo lipogenesis in a cell culture system. To clarify whether the ADFP gene is transcriptionally regulated by a LXR agonist in vivo, we fed PPAR+/+ and PPAR–/– mice with vehicle (control), the LXR agonist (T0901317; 2 x 50 mg/kg), or the PPAR agonist (WY-14643; 2 x 50 mg/kg) for 24 h. As expected, ADFP and ACO-1 were only regulated in wild-type mice by WY-14643 treatment (Fig. 8A
). Similar to the results obtained in vitro in 7800-C1 cells, treatment with the LXR agonist had no effect on ADFP mRNA in the liver. Treatment with T0901317, however, did increase the transcription of the known LXR target genes SREBP-1 and FAS, even in the absence of a functional PPAR gene. The induction of genes in the lipogenic pathway without any effect on ADFP mRNA argues against ADFP playing a role in LD coating during lipogenesis in the liver. Rather, the increased ADFP expression by externally provided fatty acids, PPAR agonists, and fasting suggests that ADFP is likely more important for LD coating during fasting.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
in adipose tissue (12). Here, we provide substantial evidence to establish ADFP as a novel direct target gene for the PPAR/RXR heterodimer in liver. Increased transcription of ADFP after activation of PPAR with a synthetic PPAR agonist occurs rapidly in rat hepatoma 7800-C1 cells, with a transactivating potential in agreement with the potency of PPAR activation (35). Transcription of ADFP is furthermore increased by WY-14643 treatment in C57Bl/6J and Sv/129 mice. ADFP mRNA level is not altered in PPAR–/– mice upon PPAR agonist treatment, which demonstrates that the induction of ADFP mRNA is dependent on the expression of a functional PPAR protein. PPAR-mediated transcriptional regulation of ADFP in liver was recently also demonstrated by others (50), which supports our findings.
Activation of PPAR stimulates transcription of the ADFP gene through a DR-1 element that is conserved among the human and mouse ADFP promoters. This DR-1 element has previously been found to bind PPARß/ (25) and more recently also PPAR (51) by other investigators. Long-chain FAs, which are believed to be functional ligands for PPAR (35), have been found to increase ADFP mRNA levels in adipocyte precursor cells cultured in vitro (23). Our analyses suggest that FAs increase ADFP mRNA levels by activating the transcription of ADFP through the conserved DR-1 element, possibly by functioning as physiological ligands for the PPAR/RXR heterodimer. Similarly, it has been proposed that VLDL/LDL activates the transcription of ADFP in macrophages through PPARß//RXR hererodimerization (25). Thus, although there are likely other important transcription sites in the ADFP promoter, the DR-1 element seems to be fundamentally important in the lipid-mediated regulation of ADFP transcription.
The expression of ADFP can be stimulated by PPARß/ (25, 26) or by PPAR ligands (51; this report), but there are conflicting reports on whether ADFP is regulated by PPAR ligands. Increased ADFP mRNA levels have been reported after stimulation with PPAR ligands in colorectal cancer cell lines (52), human blood monocytes (34), and human trophoblasts (53) but not in human macrophages (27) or adipose cells (12). A recently characterized hepatic cell line overexpressing PPAR2 accumulates LDs with a corresponding increase in ADFP protein (54). Although aFABP/aP2 mRNA, a well-characterized PPAR target gene (29), is induced by PPAR2 overexpression, the ADFP mRNA remains unchanged (54). The human adipophilin promoter is not activated by PPAR2 cotransfection in COS-1 cells (results not shown), but others have reported that it might respond to PPAR1 cotransfection and activation in CV-1 and JEG-3 cells (53). Therefore, transcriptional regulation of ADFP by PPAR might depend on the PPAR isoform-expressed or -specific coactivators/corepressors present in the cell. Another, and perhaps more likely, possibility is that transcriptional regulation of ADFP by PPAR agonists is secondarily initiated in some cell types but not in others (e.g., as a result of altered FA uptake after PPAR activation).
The ADFP mRNA and ADFP protein levels were highly induced in mice after 24 h of fasting. PPAR has been ascribed an important role in the adaptive response to fasting and is known to initiate the transcription of genes that encode key proteins involved in liver ß-oxidation and peroxisomal proliferation (32, 33). Although expression of ADFP is induced by fasting, this effect is apparently not dependent on PPAR, because ADFP mRNA and protein levels were induced similarly during fasting in PPAR+/+ and PPAR–/– mice. In contrast, induction of ACO-1 mRNA by fasting was dependent on the presence of PPAR. ME-1 mRNA remained unchanged after fasting but was induced after refeeding in a pattern very similar to RXR mRNA expression. Hence, there are marked differences in the requirements for PPAR and the response to fasting/refeeding among characterized PPAR target genes. It should be emphasized that ME-1 was recently demonstrated to be alternatively regulated by RXR homodimers through the DR-1 element previously characterized as a PPRE (55). Even though ME-1 mRNA is enhanced after treatment with a potent PPAR agonist (46, 47; this report), the gene might not be activated by PPAR in the presence of more physiological PPAR ligands, which questions the classification of ME-1 as a PPAR target gene. Our results from the fasting/refeeding experiment clearly demonstrate that ADFP is not solely regulated by PPAR in liver. Another compensatory stimulating transcription mechanism exists that ensures the induction of ADFP mRNA during fasting in the absence of a functional PPAR protein. Further studies will be required to identify these transcription factors.
An interesting finding from our experiments performed in vivo and in vitro is the close correlation between accumulation of the ADFP protein and the formation of LDs. Increased expression of ADFP mRNA by the synthetic PPAR agonist resulted in only a moderate increase in ADFP protein content. In contrast, increased ADFP mRNA under situations in which LDs were formed (i.e., FA supplementation or fasting) resulted in a much greater cellular content of ADFP. Such a difference in ADFP mRNA and protein levels has also been found in mature adipocytes, in which perilipin is thought to displace ADFP from the LD surface (7, 19). It is known that forced expression of ADFP might increase FA uptake (21) and enhance the formation of LDs in cell culture experiments (7, 9, 22). In agreement with this, increased mRNA expression and intracellular ADFP protein content have been observed in different cell types in various diseases (24, 56–58) and in a rodent model of obesity (12). These observations were all obtained in cells or tissues that had been exposed to high levels of FAs with enhanced formation of intracellular LDs. All of these observations jointly suggest that the cellular availability of FAs and the subsequent formation of LDs are required for cellular accumulation of the ADFP protein. A more recent publication suggests that ADFP is actively degraded by proteasomes when not bound to LDs (59). This might explain the low content of ADFP protein in cells lacking intracellular LDs compared with the high ADFP content observed in cells containing LDs.
In marked contrast to the induction of ADFP by fasting or the PPAR agonist, we could not observe any transcriptional regulation of ADFP with an LXR agonist. Similarly, the LXR agonist barely changed the TAG or ADFP protein content in 7800-C1 cells. Activation of LXRs stimulates lipogenesis and the synthesis of TAG in liver and increases the secretion of VLDL particles (49). The transcriptional regulation of ADFP in liver cells is apparently stimulated by increased cellular FA uptake but not by increased cellular FA synthesis. This suggests that ADFP likely does not have a role in the binding of lipid bodies synthesized through activated lipogenesis. The expression and presence of ADFP in the liver are likely much more important during fasting.
TIP47 is another PAT protein known to be expressed at significant levels in liver (12). We failed to detect any relevant changes in TIP47 mRNA levels in any of our experiments. TIP47 also seems not to be transcriptionally regulated by PPAR in adipose tissue (12). Although we cannot exclude the possibility that the TIP47 protein is targeting to LDs when they are formed, it seems reasonable that ADFP is the major PAT protein that coats the LDs in these experiments. Therefore, we propose that TIP47 is a less specialized and more ubiquitously expressed PAT protein whose function is the packaging of small LDs under normal physiological conditions. In contrast, ADFP seems to be a more specialized and physiologically regulated PAT protein in liver. The main function of ADFP in liver is to ensure the efficient uptake and trapping of FAs that are released from peripheral tissues by coating the TAG-containing LDs formed in that organ upon fasting.
|
ACKNOWLEDGMENTS |
|---|
Manuscript received October 19, 2005 and in revised form January 27, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
