Adipocyte Low Density Lipoprotein Receptor-related Protein Gene Expression and Function Is Regulated by Peroxisome Proliferator-activated Receptor γ*

The α2-macroglobulin receptor/low density lipoprotein receptor-related protein (LRP) is a large multifunctional receptor that interacts with a variety of molecules. It is implicated in biologically important processes such as lipoprotein metabolism, neurological function, tissue remodeling, protease complex clearance, and cell signal transduction. However, the regulation of LRP gene expression remains largely unknown. In this study, we have analyzed 2 kb of the 5′-flanking region of the LRP gene and identified a predicted peroxisome proliferator response element (PPRE) from −1185 to −1173. Peroxisome proliferator-activated receptor γ (PPARγ) ligands such as fatty acids and rosiglitazone increased functional cell surface LRP by 1.5–2.0-fold in primary human adipocytes and in the SW872 human liposarcoma cell line as assessed by activated α2-macroglobulin binding and degradation. These agents were found to increase LRP transcription. Gel shift analysis of the putative PPRE demonstrated direct binding of PPARγ/retinoid X receptor α heterodimers to the PPRE in the LRP gene. Furthermore, these heterodimers could no longer interact with a mutated PPRE probe. The isolated promoter was functional in SW872 cells, and its activity was increased by 1.5-fold with the addition of rosiglitazone. Furthermore, the isolated response element was similarly responsive to rosiglitazone when placed upstream of an ideal promoter. Mutagenesis of the predicted PPRE abolished the ability of this construct to respond to rosiglitazone. These data demonstrate that fatty acids and rosiglitazone directly stimulate transcription of the LRP gene through activation of PPARγ and increase functional LRP expression.

The ␣ 2 -macroglobulin receptor/low density lipoprotein receptor-related protein (LRP) is a large multifunctional receptor that interacts with a variety of molecules. It is implicated in biologically important processes such as lipoprotein metabolism, neurological function, tissue remodeling, protease complex clearance, and cell signal transduction. However, the regulation of LRP gene expression remains largely unknown. In this study, we have analyzed 2 kb of the 5-flanking region of the LRP gene and identified a predicted peroxisome proliferator response element (PPRE) from ؊1185 to ؊1173. Peroxisome proliferator-activated receptor ␥ (PPAR␥) ligands such as fatty acids and rosiglitazone increased functional cell surface LRP by 1.5-2.0fold in primary human adipocytes and in the SW872 human liposarcoma cell line as assessed by activated ␣ 2 -macroglobulin binding and degradation. These agents were found to increase LRP transcription. Gel shift analysis of the putative PPRE demonstrated direct binding of PPAR␥/retinoid X receptor ␣ heterodimers to the PPRE in the LRP gene. Furthermore, these heterodimers could no longer interact with a mutated PPRE probe. The isolated promoter was functional in SW872 cells, and its activity was increased by 1.5-fold with the addition of rosiglitazone. Furthermore, the isolated response element was similarly responsive to rosiglitazone when placed upstream of an ideal promoter. Mutagenesis of the predicted PPRE abolished the ability of this construct to respond to rosiglitazone. These data demonstrate that fatty acids and rosiglitazone directly stimulate transcription of the LRP gene through activation of PPAR␥ and increase functional LRP expression.
The ␣ 2 -macroglobulin receptor/low density lipoprotein receptor-related protein (LRP) 1 is a 600-kDa multifunctional endo-cytic receptor that belongs to the low density lipoprotein receptor gene family (1). LRP binds and internalizes a broad range of biologically diverse ligands. These include proteases of the fibrinolytic pathway (2) and serpin-enzyme complexes (3) as well as proteins important in lipoprotein metabolism such as lipoprotein lipase, hepatic lipase, lipoprotein(a), and apoE-rich lipoproteins (4 -9). Targeted deletion of LRP in the mouse results in early embryonic death, demonstrating a critical function for LRP in prenatal development (10). LRP has also been shown to have a dual role in ␤-amyloid metabolism by enhancing ␤-amyloid precursor protein conversion to ␤-amyloid (11) and mediating the clearance of ␤-amyloid (12,13). These data support a potentially complex role for LRP in the pathogenesis of Alzheimer's disease (14). In addition, LRP mediates signal transduction by interacting with cytosolic adaptor and scaffold proteins including DAB-1, JIP-2, and PSD-95 (15). A 39-kDa receptor-associated protein (RAP) is an endoplasmic reticulumresident protein that functions intracellularly as a molecular chaperone for LRP and regulates its ligand binding activity (16 -18). RAP is required for the proper folding and export of the LRP from the endoplasmic reticulum by preventing the premature binding of co-expressed ligands, such as apoE (19 -21). RAP binds LRP directly via adjacent complement-type repeats, both containing a conserved acidic residue (22), and thus stearically interferes with binding of other LRP ligands including ␣2M* and remnant lipoproteins. LRP is expressed in a variety of cells with high expression in hepatocytes, macrophages, neuronal cells, fibroblasts, and adipocytes (23). In human adipocytes, LRP is involved in chylomicron remnant cholesterol clearance (24) and mediates the selective uptake of high density lipoprotein-derived cholesteryl ester (25).
Despite the diverse and biologically important functions of LRP, relatively little is known about the regulation of LRP gene expression. The human LRP gene consists of 89 exons spanning 92 kb, encoding an mRNA of 15 kb (26). Although the coding regions of LRP and the low density lipoprotein receptor share some homology, there is little apparent similarity in their promoter regions. A portion of the 5Ј-flanking region of the LRP gene has been previously described (27,28). In Chinese hamster ovary cells, the minimal promoter driving expression of the LRP gene was shown to be in a 1.6-kb GC-rich fragment that does not contain a classical TATA box. An Sp1 sequence at Ϫ80 and two clusters of Sp1 sequences between Ϫ520 and Ϫ752 were characterized and shown to be critical for expression of the gene. The promoter region also contains a consensus NRF-1 element located at Ϫ152 that may mediate the effects of cAMP and IFN␥ (29,30). There is a consensus sterol response element located at ϩ233 in the 5Ј-untranslated region; however, studies have shown that LRP gene expression is not regulated by cholesterol (27).
We have studied the regulation of LRP gene expression during human preadipocyte differentiation and in response to free fatty acid availability. LRP mRNA was absent in human preadipocytes, and the appearance of LRP mRNA during differentiation coincided with that of the peroxisome proliferator-activated receptor ␥ (PPAR␥). 2 PPAR␥ is a transcription factor belonging to the nuclear hormone receptor superfamily. The retinoid X receptor ␣ (RXR␣) is the obligate partner of PPAR␥ (31), and together they form a heterodimer that regulates gene transcription following binding to a peroxisome proliferator response element (PPRE) and activation by specific ligands. The PPRE consists of a hexameric nucleotide repeat of the recognition motif (TGACCT) spaced by one nucleotide (DR-1) (32,33). PPAR␥ is activated by a number of ligands including long chain fatty acids (34), prostaglandin J2 derivatives (35), and thiazolidenediones (36,37). The effects of PPAR␥ ligands on gene expression are direct results of increased transcription of the target gene containing a PPRE. In our own analysis of the LRP promoter, we have identified a novel sequence (TGAACTcTGACAT) in the 5Ј-flanking sequence at positions Ϫ1185 to Ϫ1173 with high homology to the PPRE.
We report here that functional cell surface LRP is increased by PPAR␥ ligands via the activation of PPAR␥ transcriptional complexes that bind the newly identified PPRE in the LRP promoter. This is the first report demonstrating regulation of LRP gene expression via a discrete promoter element.

EXPERIMENTAL PROCEDURES
Human Preadipocyte Isolation and Culture-Subcutaneous adipose tissue was collected from healthy normolipemic subjects undergoing reduction mammoplasty procedures. Preadipocytes were isolated and cultured from adipose tissue through collagenase digestion, centrifugation, and filtration as previously described (38 -41). The preadipocytes were cultured in differentiation media for 10 -14 days. The cells were then insulin-starved in the presence or absence of varying concentrations of the PPAR␥ ligand rosiglitazone for 24 h prior to assays. The control cells were treated with vehicle only (Me 2 SO).
Cell Culture-The human liposarcoma cell line SW872 (American Type Culture Collection, Manassas, VA) was previously characterized and has been shown to be a good cell model for adipocyte gene expression (42)(43)(44). The cells were cultured in Dulbecco's modified Eagle's medium/Ham's F-12 medium (3:1) (Invitrogen) supplemented with 5% fetal bovine serum and 1% L-glutamine (Invitrogen), 10 mM NaCO 3 , and 50 g/ml gentamycin (NovoPharm, Toronto, Canada) in the presence of 5% CO 2 at 37°C. Lipoprotein-deficient fetal calf serum was prepared as described previously (45) and dialyzed against PBS for 24 h. The effect of fatty acids and thiazolidenediones on LRP mRNA and protein levels was determined by incubating cells for 24 h in medium containing either lipoprotein-deficient fetal calf serum or CS in the presence or absence of oleic acid (18:1) or arachidonic acid (20:4) (Sigma) or rosiglitazone (Smith Kline Beecham Pharmaceuticals, King of Prussia, PA). All of the conditions were studied in triplicate.

125
I-␣2M Degradation and Binding Assays-SW872 cells were cultured for 24 h with either 160 M arachidonic acid or 500 nM rosiglitazone in media containing CS prior to measuring their ability to degrade 125 I-labeled ␣2M* as previously described (46). Differentiated primary human adipocytes cultured in differentiating medium were starved of insulin in the presence or absence of increasing concentrations of rosiglitazone prior to measuring their ability to degrade or bind 125 I-␣2M* (46). Control cells were treated with vehicle only (Me 2 SO). As a further control, the cells were treated in the presence or absence of RAP because this molecule will impair LRP function.
Transcription Assay-To determine the transcriptional effect of rosiglitazone, 500 nM of this ligand was added to cells cultured as described above in the presence or absence of 10 g/ml ␣-amanitin (Sigma), a potent inhibitor of RNA polymerase II (47).
RNA Extraction, Northern Blot, and RT-PCR-Total cellular RNA was isolated from both differentiated primary human adipocytes and SW872 cells with Tri-Reagent (Bio/Can, Mississauga, Canada) according to the manufacturer's instructions. RNA samples from differentiated primary adipocytes that were to be used in RT-PCR reactions were treated with amplification grade DNase I to deplete the samples of any DNA contamination according to the manufacturer's instructions (Invitrogen). RNA concentration was determined spectrophotometrically using A 260/280 . Total RNA (5 g) was separated by agarose gel electrophoresis using the NorthernMax-Gly kit and transferred to BrightStar-Plus nylon membrane according to the manufacturer's instructions (Ambion, Austin, TX). DNA probes were synthesized by RT-PCR; first strand DNA was synthesized as described below, and PCR was performed using the following primers: LRPf, 5Ј-GAGTACCAGGTCCTGTACATCGCTG-3Ј, and LRPr, 5Ј-CTCGTCAATCATGCCCGAGATGAGC-3Ј; ␤-actin-f, 5Ј-GCCCCTCCATCGTCCACCGC-3Ј, and ␤-actin-r, 5Ј-GGGCACGAAG-GCTCATCATT-3Ј. The PCR products were gel-purified using the Qi-aexII kit (Qiagen), and the purified DNA was subsequently labeled using the Rediprime II random prime labeling kit according the manufacturer's instructions (Amersham Biosciences). The probes were cleaned up with NICK columns (Amersham Biosciences), and the specific activity was determined by use of a scintillation counter. Hybridizations and washes were performed according the NorthernMax-Gly kit instructions (Ambion).
RT-PCR was performed using a two-step approach. First strand cDNA was synthesized using 2.5 g of total RNA, 10 M random decamer primers (Ambion), and 200 units of Moloney murine leukemia virus reverse transcriptase (Invitrogen) and incubated at 42°C for 1 h. Consecutive PCR reactions were then performed on the first strand cDNA using the primers shown below and the SYBR green "taq-start" polymerase and the LightCycler Apparatus according the manufacturer's instructions (Roche Molecular Biochemicals). The data from the LightCycler was repeated using relative quantitative RT-PCR as described below. For SW 872 samples, relative quantitative RT-PCR was performed using the Quantum RNA 18 S Internal Standards kit from Ambion. This kit has been previously shown to allow the accurate determination of relative changes in gene expression between samples (48). Briefly, first strand cDNA was synthesized using 2.5 g of total RNA, 10 M random decamer primers (Ambion), and 200 units of Moloney murine leukemia virus reverse transcriptase (Invitrogen) and incubated at 42°C for 1 h. LRP forward primer (5Ј-GAGTACCAGGTC-CTGTACATCGCTG-3Ј) and reverse primer (5Ј-CTCGTCAATCATGC-CCGAGATGAGC-3Ј) were designed to amplify a region of the LRP mRNA that is ϳ400 bp in size, whereas the primers provided in the 18 S Internal Standards kit produced a band that is ϳ500 bp. A cycle number of 23 was determined to be within the linear range of PCR and was used for all subsequent PCR reactions. The 18 S primer:competimer ratio of 3:7 was experimentally determined so that the LRP and 18 S PCR products were amplified to give similar yields so that they could be compared between samples. PCR was performed on 1 l of the RT reaction using 20 pmol of each LRP primer and 4 l of 18 S primer/competitor mix with the following PCR conditions; 1 cycle of 95°C for 3 min and 23 cycles of 95°C for 30 s, 66°C for 30 s, and 72°C for 30 s. Cocktails containing all shared components were used to reduce variation between samples. The PCR products were subjected to electrophoresis through a 1.5% agarose gel and visualized with ethidium bromide staining. The band intensities were measured using the ChemiDoc apparatus and Quantity One software (Bio-Rad). Relative intensity was calculated by dividing the 400-bp band corresponding to the LRP message by the 488-bp band corresponding to the 18 S message.
Cell Surface Fluorescent Detection of LRP-The cells were cultured on 35-mm cover glass bottom dishes (MatTek, Ashland, MA) as described above and supplemented with 160 M arachidonic acid, 500 nM rosiglitazone, or control. ␣2M was activated by incubating purified ␣2M with 400 mM methylamine for 16 h at room temperature. ␣2M* was fluorescently labeled using Cy3 monofunctional reactive dye (Amersham Biosciences) to a dye:protein ratio of 1.3 according to the manufacturer's instructions. The cells were placed on ice for 1 h in 3:1 Dulbecco's modified Eagle's medium/Ham's F-12 medium supplement with 2 mg/ml BSA buffered with 10 mM HEPES. Labeled ␣2M* was diluted in the same medium to a concentration of 1 g/ml and was added to the cells at 0°C for 45 min. The cells were washed with ice-cold PBS three times prior to being fixed with 4% paraformaldehyde for 10 min at 0°C. The cells were rinsed with PBS and kept in 2 ml of PBS at room temperature for fluorescence microscopy. Binding was competed with 30-fold excess of unlabelled ␣2M*. The cells were viewed with an Olympus IX50 fluorescent microscope, and the images were taken using a coded CCD camera (MicroMax) and WinView software from Princeton 2 F. Benoist and R. McPherson, unpublished data.

Instruments (Princeton, NJ).
Western Blotting of LRP-Total cellular protein (5 g) from SW872 cells incubated in the presence or absence of various PPAR␥ ligands was subjected to SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose (49). The LRP was detected (49) using a polyclonal rabbit antisera (from Dr. G. Bu) followed by chemiluminescent detection (Pierce) of a secondary antibody conjugated to horseradish peroxidase. The blot was developed, and the bands were quantified using the ChemiDoc apparatus and Quantity One software (Bio-Rad). Triplicate cell samples were processed and are summarized in the graph. Molecular biology techniques were essentially as described by Sambrook et al. (50).
Preparation of Nuclear Protein Extracts-The nuclear proteins were extracted from SW872 cells as previously described (51) or from primary human adipocytes as described (52). Protein concentration was determined using BCA protein reagent (BioLynx, Brockville, Canada) according to the manufacturer's instructions.
Electrophoretic Mobility Shift Assays (EMSA)-Double-stranded oligonucleotides (oligomers) corresponding to the PPRE of LRP (5Ј-CCCCGCTCCTTGAACTCTGACATCGAGACACCTA-3Ј) were radioactively end-labeled with [␥-32 P]dATP (Amersham Biosciences) using T4 polynucleotide kinase (Invitrogen) and purified from unincorporated nucleotides by gel filtration through G-50 spin columns (Amersham Biosciences). The same procedure was used for oligomers corresponding to the PPRE of the human fatty acyl CoA oxidase gene (hACOX) (5Ј-TCCGAACGTGACCTTTGTCCTGGTCCCCTTT-3Ј) and oligomers corresponding to the mutated form of the LRP PPRE (the mutated half-site is underlined) (5Ј-CCCCGCTCCTTGAACTCAACGATCGAGACACC TA-3Ј). The specific activities of the oligomers were ϳ250 cpm/fmol. These were diluted to 60 fmol/l for use in the assay. PPAR␥ (from Dr. Bruce Spiegelman) was subcloned into the MluI and NotI sites of pSPORT1 (Invitrogen) using PCR-based methods. RXR␣ was provided by Dr. Michael Saunders in pSG5. Both constructs are driven by the T7 RNA polymerase promoter for use in the TNT T7-coupled reticulocyte lysate system (Promega) for in vitro transcription/translation. All of the EMSA reactions were carried out on ice in 20 l of binding buffer (12.5 mM HEPES-KOH, pH 7.6, 6 mM MgCl 2 , 5.5 mM EDTA, and 50 mM KCl) supplemented with 5 mM dithiothreitol, 0.25 g of low fat milk, 0.05 g of poly(dI-dC), and 10% glycerol. For EMSA reactions with TNT-purified proteins, 2 l of the TNT reaction was added to the reaction mix along with 1 l (60 fmol) of labeled oligomers. For EMSA reactions with nuclear protein extracts, 6 l of nuclear extracts were added to the reaction mix with 1 l (60 fmol) of labeled oligomers. These reactions were left on ice for 20 min. Following the 20-min incubation, 2 l of 20% Ficoll was added to the samples. DNA-protein complexes were then resolved by electrophoresis through 6% polyacrylamide gels in 0.25ϫ Tris borate running buffer (20 mM Tris borate, pH 7.2, 0.5 mM EDTA).
Supershift assays were performed using the PPAR␥ NuShift kit following the manufacturer's protocol (Active Motif). The nonspecific antibody used was mouse monoclonal anti-␤-actin (Santa-Cruz).
LRP Reporter Gene Constructs-Complementary primers with flanking NheI and XhoI restriction sites (5Ј-CTAGCCTCCTTGAACTCTGA-CATGCAGACC-3Ј) were annealed and subcloned into the luciferase reporter vector, pGL3-Promoter (Promega). This new vector contains a single copy of the putative PPRE upstream of an ideal promoter and is designated pGL3-PPRE.
We also prepared 1.9 kb of the 5Ј-flanking region of LRP by PCR amplification from the LRP-BAC construct prepared by Dr. Jan Boren (53), which contains the entire 92-kb gene of human LRP, using the primers 5Ј-GCAACGAGCTCCGTAAAAGGGGGAAG-3Ј and 5Ј-GCAG-CAGATCTTTCCCCGGACTGAAG-3Ј. This fragment was subcloned into the SacI and BglII sites of the luciferase reporter vector, pGL3-Basic (Promega), and designated pGL3-LRP. Mutagenesis of the PPRE was performed by PCR using PFUTurbo (Stratagene, La Jolla, CA) according to their Quikchange site-directed mutagenesis protocol. The complementary primers (5Ј-CCCGCTCCTTGAACTCAACGATGCAGA-CACC-3Ј) were designed to mutate a single half-site of the PPRE so that PPAR␥ would no longer bind the response element (mutated nucleotides are underlined). This construct was designated pGL3-LRP mutant PPRE . Both pGL3-LRP and pGL3-LRP mutant PPRE were sequenced to confirm that the promoter sequence was correct (compared with GenBank TM accession number Y18524) and to verify the mutagenesis.
Transient Transfection Assays-Confluent SW872 cells were trypsinized and seeded at a density of 1.25 ϫ 10 5 cells/well in 12-well plates 48 h prior to transfection. The cells were ϳ70 -80% confluent at the time of transfection. Fresh medium containing CS was added 12 h preceding transfections. The cells were co-transfected with 4 g of the firefly luciferase reporter vector (either pGL3-basic, pGL3-LRP, PGL3-PPRE, or pGL3-LRP mutant PPRE ) and 0.25 g of the Renilla luciferasebearing reporter vector, pRL-CMV (Promega) using the calcium phosphate-DNA precipitate method (54). The cells were shocked with 15% glycerol for 2 min, 4 h after the transfection, and washed three times with PBS before the addition of medium. The cells were treated 12 h later with Me 2 SO alone (vehicle control) or varying concentrations of rosiglitazone. After 24 h (36 h total post-transfection) the cells were scraped in 250 l of reporter lysis buffer (Promega) and kept on ice until assayed.
Luciferase activities derived from both firefly (LRP constructs) and Renilla (pRL-CMV) proteins were measured using the dual luciferase reporter assay system (Promega) and recorded using a Monolight 2010c luminometer (Analytical Luminescence Laboratory, Ann Arbor, MI). Renilla luciferase activity was then used to standardize for transfection efficiency.
Statistical Analysis-The results are expressed as the means Ϯ S.E. Where indicated, the statistical significance of the differences between groups was determined using Student's t test or analysis of variance.

RESULTS
Functional Cell Surface LRP Is Increased upon Exposure to PPAR␥ Ligands-The effects of PPAR␥ ligands on differentiated primary human adipocytes were examined by an 125 I-␣2M* binding assay (Fig. 1A). The B max of cells incubated with 1 M rosiglitazone (27.0) is ϳ1.5 times greater than that of the vehicle-treated cells (17.3), indicating that there is an increase in the levels of functional cell surface LRP. The difference in binding was found to be highly significant with a two-tailed p value of less than 0.0001. When the cells were treated with RAP (30 g/ml), the amount of binding was reduced to background levels, demonstrating that this process is LRP-specific. There was no statistically significant difference in the K d between the treated and control cells as illustrated in the Scatchard plot (Fig. 1B), indicating that the binding affinities have not changed. 125 I-␣2M* degradation assays were also performed in the presence or absence of PPAR␥ ligands for the differentiated primary human adipocytes and SW872 cells. In primary adipocytes, there was a direct relationship between the amount of rosiglitazone added and the amount of ␣2M* degradation over 8 h ( Fig. 2A) with very significant increases ranging from 1.2-to 1.7-fold over control (p Ͻ 0.009). The RAP is an antagonist of all identified LRP ligands including ␣2M*; therefore we used purified RAP to block LRP function in our assays. When cells were treated with RAP, the amount of 125 I-␣2M* degraded was diminished to background levels, demonstrating that this process is LRP-specific. Degradation assays were also performed in the SW872 cells treated with rosiglitazone ( Fig. 2B) or arachidonic acid (Fig. 2C). There was a 1.5-fold increase in the amount of 125 I-␣2M* degraded over 8 h in the treated cells versus the control cells for both of the PPAR␥ ligands.
The increase in functional cell surface LRP was confirmed in SW872 cells by cell surface labeling experiments using fluorescently labeled ␣2M* in the presence of 160 M arachidonic acid or 500 nM rosiglitazone (Fig. 3A). In addition, total cellular LRP was increased in those cells as determined by Western blot analysis (Fig. 3B). The increases seen in LRP using these methods were ϳ1.5-2-fold, supporting the binding and degradation data above.
Endogenous LRP mRNA Levels Are Increased upon Exposure to PPAR␥ Ligands-The effect of PPAR␥ ligands on levels of LRP mRNA was determined in the adipocytic cell line, SW872, by relative quantitative RT-PCR. The relative intensity was determined by the ratio of the LRP band intensity compared with the 18 S band intensity, and these values were normalized to the control samples to give values of fold increase. The fold increases of LRP mRNA in cells upon treatment with oleic acid, arachidonic acid, and rosiglitazone are summarized in Fig. 5  below (A, B, and C, respectively). PPAR␥ ligands, rosiglitazone (500 nM), arachidonic acid (160 M), and oleic acid (0.8 mM) increased LRP mRNA levels by 1.5-, 1.6-, and 1.3-fold, respectively. The maximum effect of the ligands was observed after 24 h of treatment. Increases in LRP mRNA levels were not observed for concentrations of rosiglitazone above 1 M (data not shown). When cells were cultured in medium containing lipoprotein-deficient fetal calf serum instead of CS, the increases in LRP mRNA levels were similar to those shown in Fig. 4 (B and C).
The effect of rosiglitazone on LRP mRNA levels was also determined in differentiated primary human adipocytes, as measured by Northern blot and real time PCR, and these results are summarized in Fig. 5. According the Northern blot analysis, LRP mRNA is increased by rosiglitazone treatment by ϳ1.6-fold. Using real time RT-PCR, there was a dose-dependent increase in LRP mRNA levels (ranging from 1.4-to 1.8-fold) after 24 h of ligand treatment. The levels of LRP mRNA were also verified using relative quantitative RT-PCR, and the fold increases in LRP expression were similar to those shown in Fig. 5.
PPAR␥ Ligands Act at the Transcriptional Level to Increase LRP mRNA-The increase in LRP mRNA could be due to an mRNA stabilization effect or to increased transcription of the mRNA. To distinguish between these possibilities, a potent inhibitor of RNA polymerase II activity, ␣-amanitin, was used to inhibit new transcriptional activity. If the level of mRNA were increased by a stabilization effect of rosiglitazone or other PPAR␥ ligands, then the increase in mRNA levels would still be observed when both ␣-amanitin and rosiglitazone were added to cells concomitantly. We did not observe an increase in LRP mRNA in cells cultured with both rosiglitazone and ␣-amanitin (Fig. 4C). Although there was a small decrease in LRP mRNA levels with the addition of ␣-amanitin, this decrease was identical in both the vehicle-treated and rosiglita- This analysis is presented as a Scatchard plot with bound/free plotted against the bound 125 I-␣2M*. The K d is a measure of the binding affinity between the ligand and the receptor and is determined from the slope. The B max is a measure of the number of receptors on the cell surface and is represented by the x intercept. The data points represent means from triplicate experiments, and the error bars represent the standard error of the mean. *, the two-tailed p value is Ͻ0.0001.

FIG. 2. Increased cellular degradation of 125 I-␣2M* by cells treated with PPAR␥ ligands.
A, differentiated primary human adipocytes were treated with various concentrations of rosiglitazone in the absence of insulin for 24 h prior to the experiment. The cells were then incubated at 37°C with HBSS, 25 mM HEPES, pH 7.45, 10 mg/ml BSA containing 1 mg/ml 125 I-␣2M* with or without RAP (30 g/ml). At the beginning of the incubation period and at each time point, the medium was precipitated with ice-cold tricarboxylic acid, and the tricarboxylic acid-soluble material was used as a measure of the degraded protein.
The data points represent the means from triplicate experiments, and the error bars represent the standard error of the mean. *, p ϭ 0.009; **, p ϭ 0.007; ***, p ϭ 0.002. B and C, SW872 cells (treated with vehicle, 500 nM rosiglitazone, and 160 M arachidonic acid) were washed at 37°C to remove any residual fetal calf serum. The cells were incubated at 37°C for the indicated times in HBSS, 25 mM HEPES, pH 7.45, 10 mg/ml BSA containing 1 mg/ml 125 I-␣2M*. At the beginning of the incubation period and at each time point, the medium was precipitated with ice-cold tricarboxylic acid, and the tricarboxylic acid-soluble material was used as a measure of the degraded protein. zone-treated cells, suggesting this is due to normal turnover of the mRNA. These results support a role for these ligands as transcriptional up-regulators of LRP gene expression.
PPAR␥-RXR␣ Heterodimers Selectively Bind the PPRE Identified in the LRP Promoter Region-We identified a PPRE located at Ϫ1185 to Ϫ1173 of the LRP promoter through sequence scanning of the 5Ј-flanking region. The sequence homology to the consensus PPRE is 83%; there is a single mismatch per half-site of the DR-1 (Fig. 6A). The TNT in vitro transcription/translation of pSPORT1-PPAR␥ and pSG5-RXR␣ was first shown to express PPAR␥ and RXR␣; when [ 35 S]methionine was added to the reaction mix, proteins of the correct molecular mass were synthesized as verified by SDS-PAGE and autoradiography. For the EMSA, these constructs were subjected to the TNT reaction (without [ 35 S]methionine) and incubated with 32 P-end-labeled oligomers corresponding to the PPRE of LRP. A clear shift was evident in the presence of these transcription factors (Fig. 6B, lane 2) compared with when the TNT reaction was carried out with the empty pSPORT1 vector (unprogrammed lysate) (lane 1). The interaction between the protein complex and the DNA could be competed by the addition of increasing amounts of excess of unlabeled hACOX oligomer (lanes 3-6). Identical experiments using the hACOX probe yielded very similar results (Fig. 6B, lanes 7-12). The doublet bands that appear have been observed in other studies where rabbit reticulocyte lysate was used to produce PPAR␥ and RXR␣ (55)(56)(57). These doublets cannot be homodimers because PPAR␥ and RXR␣ TNT reactions, added individually to the EMSA reaction, did not give shifts (data not shown). Incubation of the probes with nuclear protein extracts from primary adipocytes gave a shift of the same size as the TNT proteins (Fig.  6C, lane 2). This shift could also be inhibited by increasing amounts of excess unlabeled hACOX oligomer (lanes 2-5). Nuclear extracts from the SW872 cell line yielded similar results. These results are comparable with those obtained using oligomers corresponding to the hACOX PPRE (lanes 6 -10). A probe containing a mutated PPRE half-site, LRP mut , could no longer interact and bind TNT proteins (Fig. 6B, lane 14) or nuclear extracts (Fig. 6C, lane 11).
The shift caused by the nuclear extracts was further analyzed by gel supershift analysis to confirm that PPAR␥ was a component of this complex (Fig. 6D). Free probe was run in lane 1 as a control reaction, whereas a reaction containing nuclear protein extracts from primary human adipocytes is shown in FIG. 3. LRP protein levels are increased in SW872 cells upon exposure to PPAR␥ ligands. A, cells were incubated with 1 g/ml fluorescently labeled activated ␣2M (␣2M*) (Cy3 monofunctional reactive dye) on ice for 45 min in 3:1 Dulbecco's modified Eagle's medium/ Ham's F-12 medium supplement with 2 mg/ml BSA buffered with 10 mM HEPES after an equilibration period of 1 h. The cells were washed with ice-cold PBS three times prior to being fixed with 4% paraformaldehyde for 10 min at 0°C. The cells were rinsed with PBS and kept in 2 ml of PBS at room temperature for fluorescence microscopy. The photographs are representative of cells treated with the ligands indicated. A 30-fold excess of unlabeled ␣2M* was used to compete for binding with the fluorescently labeled ␣2M*. The photographs are normalized so that an increase in intensity on the cell surface and the cell circumference between photographs represents an increase in the fluorescence (i.e. total binding). B, Western blot of cells treated with Me 2 SO (vehicle control), 160 M arachidonic acid, or 500 nM rosiglitazone for 24 h. 5 g of total protein was loaded in each lane. The blot was developed using chemiluminescent techniques, and the bands were visualized using the ChemiDoc apparatus. The band at 515 kDa corresponds to the ␣-subunit of LRP. C, quantification of Western blot. The 515-kDa band corresponding to LRP was quantified using Quantity One software. All of the samples were normalized to the vehicle-treated cells (control). The Results are shown as the means of triplicate experiments, and the error bars represent the standard error of the mean. *, the two-tailed p value is 0.002; **, the two-tailed p value is 0.003.

FIG. 4. Endogenous LRP mRNA levels in SW872 cells are modulated by PPAR␥ ligands at the transcriptional level.
The effect of PPAR␥ ligands on levels of LRP mRNA was determined by incubating SW872 cells for 24 h in Dulbecco's modified Eagle's medium/Ham's F-12 medium (3:1) supplemented with complete serum in the presence or absence of the PPAR␥ ligands. A, oleic acid. B, arachidonic acid. C, rosiglitazone. Also a potent inhibitor of RNA polymerase II, ␣-amanitin, was added to SW872 cells in the presence or absence of 500 nM rosiglitazone to inhibit transcription (C). Reverse transcription was performed on total RNA isolated from these cells, and multiplex PCR was performed using LRP and 18 S gene-specific primers (relative quantitative RT-PCR). PCR products were visualized by EtBr staining on the Chemi-Doc, and the band intensities were determined using Quantity One software. The intensity of the LRP product was divided by the intensity of the 18 S product to obtain a value termed relative intensity. These results were normalized to the control and shown as fold increases. The bars in each graph represent the means of triplicate experiments, and the error bars represent the standard error of the mean. One-tailed p values from paired t test are shown above the bars to indicate significant differences.
the second lane. The nuclear protein extracts were preincubated with anti-PPAR␥ antibody prior to being added to the reaction containing labeled oligomers, and this was run in the third lane. A band of the same size as that in the second lane was present as well as a larger band that was absent in all other lanes (supershift). The same results were obtained for the oligomer corresponding to the hACOX PPRE. A supershift was not observed when a nonspecific antibody was used in place of the anti-PPAR␥ antibody. Similar results were also obtained using nuclear extracts from SW872 cells supporting the involvement of PPAR␥ in the protein-DNA complex.
LRP PPRE Luciferase Constructs Are Responsive to Rosiglitazone in Dual Luciferase Assay-Although the basal transcrip-tional activity of the pGL3-PPRE construct was ϳ2-fold higher than the pGL3-LRP construct because of the context of the SV40 ideal promoter (data not shown), the promoter context of the endogenous LRP promoter, which is present in pGL3-LRP, is the more physiologically relevant of the two constructs. This portion of the promoter drove the basal activity of the reporter gene as determined by the dual luciferase reporter assay system (Fig. 7A). The values were normalized to the cells treated with Me 2 SO only (vehicle control). The empty pGL3-basic vector had no basal activity above the background measurements of the instrument (data not shown). For pGL3-LRP (Fig. 7A), as well as pGL3-PPRE (data not shown), there was a dose-dependent response of the luciferase activity upon treatment with rosiglitazone that corresponded to the increases in LRP mRNA shown in Fig. 4. The ratio of firefly and Renilla luciferase activities are shown in the figures; it is important to note that the luciferase activity increases with rosiglitazone treatment and that the Renilla does not decrease.

FIG. 5. Endogenous LRP mRNA levels in differentiated primary human adipocytes are modulated by rosiglitazone. A,
Northern blot analysis of total RNA from differentiated primary human adipocytes. Differentiated primary human adipocytes were cultured for 24 h in differentiating medium without insulin in the presence or absence of 1 M rosiglitazone Total RNA isolated and separated by agarose gel electrophoresis and transferred to BrightStar-Plus nylon membrane. The membrane was probed with either an LRP or ␤-actin random labeled probe. A representative blot is shown here. B, quantification of Northern blots. Northern blots were scanned using the ChemiDoc, and densitometry was performed using Quantity One software. The data shown are the averages of three independent Northern blots from the RNA of three individual tissue samples (each done in triplicate), and the error bars represent the standard error of the mean. C, quantification of mRNA changes using real time RT-PCR. Differentiated primary human adipocytes were cultured for 24 h in differentiating medium without insulin in the presence or absence of various concentrations of rosiglitazone. Reverse transcription was performed on total RNA isolated from these cells, and independent real time PCR reactions were performed using LRP and 18 S gene-specific primers on the LightCycler. These results were normalized to the control and shown as fold increases. The bars in each graph represent the means of triplicate experiments, and the error bars represent the standard error of the mean. The same RT reaction was subjected to relative quantitative RT-PCR, and the results were similar to those shown here. Twotailed p values from Student's t test are shown above the bars to indicate significant differences. A p value of 0.02 for all of the values is obtained using analysis of variance.  1, 7, and 13). TNT in vitro transcribed/translated PPAR␥ and RXR␣ were incubated with radiolabeled PPRE oligomers (lanes 2,  8, and 14). To control for specificity of binding competition experiments were performed by adding increasing amounts of unlabeled oligomer to the binding reactions. LRP binding reactions were competed using cold hACOX oligomers at 5ϫ, 10ϫ, 25ϫ, and 50ϫ excess (lanes 3-6), and the hACOX binding reactions were competed using cold LRP oligomers at 5ϫ, 10ϫ, 25ϫ, and 50ϫ excess (lanes 9 -12). C, nuclear extracts from primary human adipocytes were incubated with radiolabeled oligomer (lanes 1, 6, and 11) on ice for 20 min. Specificity of binding was again tested using unlabeled oligomers in competition experiments as described above (lanes 2-5 and 7-10). D, gel supershift analysis was used to determine whether PPAR␥ was present in shift seen with nuclear extracts. Free probe (first lane) was used as a control reaction. The nuclear extracts (18 g) were incubated with labeled oligomer and run in the second lane. NuShift anti-PPAR␥ antibody (4 l) was incubated with 18 g of nuclear extracts from SW872 cells prior to being added to the EMSA reaction mix containing labeled oligomers (third lane). A nonspecific antibody (mouse monoclonal anti-␤-actin) was used as a negative control for the supershift (fourth lane). The arrows indicate the shift and the supershift.

Mutagenesis of the PPRE Results in the Loss of Enhancer
Activity-A single half-site of the PPRE, to which PPAR␥ binds, was mutated using the Quikchange mutagenesis protocol (Fig. 7B) and designated pGL3-LRP mutant PPRE . This construct was sequenced, and the point mutations within the PPRE were verified. The basal activity of this reporter construct (vehicle control) was similar to that of the pGL3-LRP construct (Fig. 7C). In the presence of 500 nM rosiglitazone there was approximately a 1.5-fold increase in promoter activity of the pGL3-LRP construct. This increase, however, was lost in the pGL3-LRP mutant PPRE construct, indicating a role for this PPRE as a transcriptional enhancer. DISCUSSION Despite the importance of LRP in lipoprotein, serum protease, and ␤-amyloid metabolism, this is the first study to demonstrate that LRP expression is regulated at both mRNA and protein levels via a discrete promoter element. We have demonstrated that functional LRP expression is regulated by PPAR␥ ligands by a mechanism that involves the ligand-induced up-regulation of transcription via the activation of PPAR␥-RXR␣ heterodimers that bind a newly identified PPRE in the promoter of the LRP gene. Levels of functional cell surface LRP were measured by binding and degradation of a well characterized LRP ligand, ␣2M* (58,59). Furthermore, we have demonstrated that LRP mRNA levels are modulated at the transcriptional level by ligands that activate PPAR␥ and that this response is dose-dependent. The increase in LRP mRNA levels with rosiglitazone was shown to result from direct binding of PPAR␥-RXR␣ heterodimers to the PPRE identified in the promoter.
As anticipated, there was an inverse correlation between the amount of ligand required and the affinity of the ligand for PPAR␥ (36,60). Rosiglitazone, the most potent ligand used, was found to have a maximal effect at ϳ500 nM in SW872 cells, although there was significant up-regulation of LRP at 50 nM, a concentration much closer to the reported K d of 40 nM (36). Concentrations of 750 nM or higher did not alter LRP mRNA abundance (data not shown) or promoter activity of pGL3-LRP in SW872 cells. In differentiated human adipocytes, LRP mRNA levels were up-regulated by rosiglitazone in a dose-dependent manner at concentrations up to 1 M, and the decreased efficacy was observed only at a concentration of 2 M. It has been suggested that rosiglitazone might act as a partial antagonist at these high concentrations (61). In addition, activation of PPAR␥ at the AF2 domain enhances its degradation (62), which would explain the reduced efficacy of rosiglitazone at higher concentrations (Figs. 5 and 7A). In both cell types, LRP expression and function were increased by 1.5-2-fold, similar to that reported for other genes containing a PPRE in the promoter region (57,63,64). For example, the PPAR␣ agonist fenofibrate increases apoA-I expression by 1.5-2.0-fold; yet this translates into a clinically important high density lipoprotein raising effect (65).
Previous studies have examined the effect of sterols on LRP transcription and reported that LRP, unlike the low density lipoprotein receptor and other members of this receptor family, was not down-regulated by sterols (27). Further study identified a sequence corresponding to a sterol response element in the 5Ј-untranslated region of the LRP transcript (28), which appears inactive because LRP does not show any response to sterols. This is in agreement with our data demonstrating that LRP mRNA levels are similar when cells are cultured in CS as compared with lipoprotein-deficient fetal calf serum.
PPAR␥ is not expressed in preadipocytes and is turned on during differentiation prior to the expression LRP 3 and other adipocyte genes (66). In addition to LRP, many adipocyte proteins important in triglyceride accumulation, such as lipoprotein lipase, fatty acid transport protein-1, acyl-CoA synthase, CD36, and aP2 (67)(68)(69)(70), all contain at least one PPRE in their 5Ј-flanking sequences and are all regulated by PPAR␥. Adipocyte LRP functions in chylomicron remnant cholesterol clearance both in vitro and in vivo (24). We have recently demonstrated that LRP also plays a role in the selective uptake of high density lipoprotein-CE by human adipocytes (71). Thus, co-ordinate regulation of LRP and fatty acid transporters may be a mechanism by which adipocytes can regulate cholesterol 3 F. Benoist 7. LRP promoter confers regulation of luciferase activity by rosiglitazone. A, the promoter region of human LRP (0 to Ϫ1200) was cloned into pGL3-Basic. This construct was then transiently cotransfected into SW872 cells along with the Renilla luciferase reporter vector pRL-CMV using the calcium-phosphate precipitation method. Luciferase activities for each of firefly and Renilla luciferase were determined using the dual luciferase assay system. The intensity of firefly luciferase is shown as a function of Renilla luciferase (relative activity). The relative intensity is normalized to the control (vehicle treatment) values. The means of triplicate experiments are graphed, and the error bars represent the standard error of the mean. The one-tailed p values from paired t tests are shown for significant differences. B, pGL3-LRP was subjected to site-directed mutagenesis, in which the half-site of the DR-1 to which PPAR binds, was mutated so that 5 of 6 nucleotides no longer matched the consensus. PPREs of pGL3-LRP and pGL3-LRP mutant PPRE are shown with the consensus sequence for comparison. Point mutations are underlined, and the mismatches present in endogenous PPRE are in italics. C, pGL3-LRP mutant PPRE , was transiently co-transfected with pRL-CMV into SW872 cells, and their respective luciferase activities were determined using the dual luciferase assay system. The relative intensity was calculated, and the means of triplicate experiments are shown. The values of relative intensity are normalized to the control (vehicle-treated) cells. The error bars represent the standard error of the mean. Rosiglitazone has a significant effect on pGL3-LRP; however, it is not significant for pGL3-LRP mutant PPRE . The one-tailed p values from t tests were 0.02 and 0.2, respectively. uptake to match TG synthesis during differentiation and maturation of the preadipocytes into fat cells.
LRP is expressed in various tissues (23), and its function in each cell type differs widely. Thus, the regulation of LRP gene expression and function by insulin-sensitizing agents of the glitazone class or by fibrates could have considerable clinical importance. Three PPAR subtypes (␣, ␥, and ␦) have been identified. Within a given species, the DNA-binding domains of the three PPARs are 80% identical (slightly higher between PPAR␥2 and PPAR␣ (72)). However, their ligand-binding domains only share ϳ65% homology (73). It has been demonstrated that PPAR␣ and PPAR␥ bind the same core DR-1 PPRE (74). The distinct tissue-specific expression of the different PPARs as well as their specific activation by ligands suggests a mechanism for highly tissue-specific regulation of genes with a PPRE, including LRP. PPAR␣ is predominantly expressed in liver, heart, kidney, intestinal mucosa, and brown adipose tissue (33). These are all sites with high fatty acid catabolism and peroxisomal metabolism. PPAR␦ is ubiquitously expressed, whereas PPAR␥ is expressed mainly in adipose tissue, skeletal muscle, heart, brain, vascular smooth muscle cells, and monocyte/macrophages (31,33,75). PPAR␥2 is relatively adiposespecific, although in animal models of obesity, hepatic expression of PPAR␥2 has been documented (76). The transcriptional activity of the PPAR subtypes is enhanced by a multitude of compounds. Prostaglandin J2 is a natural ligand for PPAR␥, whereas thiazolidenediones (e.g. BRL49653 or rosiglitazone) are synthetic ligands for PPAR␥ (36) and do not activate PPAR␣. PPAR␣ ligands include 8(S)hydroxyeicosapentanoic acid, leukotriene B4, and the synthetic fibrates. Long chain fatty acids are less specific ligands recognizing all PPAR subtypes (33,34). By administration of selective PPAR ligands, it may be possible to regulate the expression of LRP in a tissuespecific manner.
We have demonstrated that adipocyte LRP expression and function is up-regulated by rosiglitazone, a widely used insulinsensitizing agent. Rosiglitazone has been shown to enhance plasma triglyceride clearance by an unknown mechanism (75), which we hypothesize may involve adipocyte LRP. Fibrates substantially decrease plasma triglyceride levels, and this effect has been primarily attributed to an increase in lipoprotein lipase activity and decreased expression of apoCIII (33,34,75). We propose that a PPAR␣-mediated increase in hepatic LRP expression may explain in part the triglyceride-lowering effects of fibric acid derivatives. In ongoing studies, we are investigating PPAR-mediated regulation of LRP expression and function in other cell types, including hepatocytes and neuronal cells. The involvement of LRP in a variety of important metabolic processes including amyloid precursor protein processing, ␤-amyloid clearance, lipoprotein metabolism, cellular remodeling, and protease complex clearance suggest a possible therapeutic role for LRP up-regulation by PPAR ligands in a number of disease states.