ARP-1/COUP-TF II determines hepatoma phenotype by acting as both a transcriptional repressor of microsomal triglyceride transfer protein and an inducer of CYP7A1.

L35 and FAO cells were derived as single cell isolates from H35 cells. Whereas L35 cells do not express microsomal triglyceride transfer protein (MTP), which regulates lipoprotein secretion, they express CYP7A1, which regulates bile acid synthesis from cholesterol. FAO cells display the opposite phenotype (i.e. expression of MTP but not CYP7A1). We examined the molecular basis of the transcriptional inactivation of the MTP gene in L35 cells. Nested deletion and mutagenesis studies show that a conserved DR1 element within the 135-bp proximal MTP promoter is responsible for differential expression by L35 and FAO cells. Yeast one-hybrid screening identified apolipoprotein A1 regulatory protein-1/chicken ovalbumin upstream promoter transcription factor II (ARP-1/COUP-TFII) and retinoid X receptor (RXRalpha) as the protein factors that can bind to the conserved DR1 element. Nuclear extracts from L35 cells contained 2-fold more ARP-1/COUP-TFII and 50% less RXRalpha than those from FAO cells. Immunologic studies show that in L35 cells, ARP-1/COUP-TFII is bound to the DR1 element, whereas in FAO cells, a complex containing RXRalpha is bound to the DR1 element. Co-transfection studies show that ARP-1/COUP-TFII repressed MTP promoter activity by approximately 70% in FAO hepatoma cells, whereas RXRalpha and its ligand 9-cis-retinoic acid increased MTP promoter activity by 6-fold in L35 cells. The combined data suggest that in the context of the MTP promoter, ARP-1/COUP-TFII (repressor) and a complex containing RXRalpha (inducer) compete for the DR1 element. Analysis of the CYP7A1 promoter revealed that it is approximately 5-fold more active in L35 cells than in FAO cells. Co-transfection of an ARP-1/COUP-TFII expression vector showed that it enhances CYP7A1 promoter activity by 6-fold in FAO cells. These combined findings indicate that ARP-1/COUP-TFII acts as both a transcriptional repressor (of MTP) and as a transcription activator (of CYP7A1). This dual function of ARP-1/COUP-TFII may play an important role in determining the metabolic phenotype of individual liver cells.

Individual liver cells, distinguished by their anatomical localization within the portal triad and their relative expression of genes, allow the liver to provide diverse anabolic and catabolic functions. Hepatocytes localized near the portal vein express high levels of lipogenic (1) and gluconeogenic (2) enzymes, whereas those localized near the central vein express high levels of catabolic (CYP7A1) (3,4) and glycolytic (2) enzymes. Environmental differences caused by hormonal, nutritional, and xenobiotic variations in portal versus central blood result in zonal distinctions in the gene expression of transcription factors within the hepatic acinus (5). The molecular mechanisms responsible for the distinct subsets of genes expressed by individual liver cells remain unknown. We have derived and characterized stable lines of rat hepatoma cells in order to identify the molecular mechanisms responsible for the phenotypic expression of the anabolic lipoprotein assembly pathway regulated by microsomal triglyceride transfer protein (MTP) 1 and the catabolic pathway through which cholesterol is converted into bile acids by the rate-limiting enzyme CYP7A1.
The liver is the major site for both the production of plasma lipoproteins and their uptake from plasma and catabolism (reviewed in Ref. 6). The production of apolipoprotein B (apoB)containing lipoproteins by the liver is mainly regulated by the relative amount of de novo synthesized apoB that is translocated into the endoplasmic reticulum and assembled into a secretion-competent lipoprotein particle (7)(8)(9)(10). MTP, an intraluminal protein in the endoplasmic reticulum (11), plays an important role in regulating the assembly and secretion of apoB-containing lipoproteins (12)(13)(14)(15)(16). In humans, the loss of MTP function blocks the production of apoB-containing lipoproteins by both the intestine and the liver (11). On the other hand, the catabolic bile acid biosynthetic pathway, regulated by CYP7A1, is the major route through which hepatic (and total body) cholesterol homeostasis is maintained (17)(18)(19)(20). Hepatic cholesterol levels indirectly regulate the expression of low density lipoprotein receptors via changes in serum response element (SRE)-binding proteins (21). Thus, the expression of CYP7A1 indirectly regulates the expression of low density lipoprotein receptors (22,23). Bile acids, which are the end products of the CYP7A1 reaction, stimulate the secretion of phospholipids and cholesterol into bile (24). Thus, the relative expression levels of MTP and CYP7A1 by individual hepatic parenchymal cells are important determinants of whether phospholipids and cholesterol are targeted into plasma (lipoproteins) or bile (biliary lipids).
We have isolated stable cell clones from rat hepatoma cells as a means to understand the mechanisms through which liverspecific genes controlling cholesterol metabolism and lipoprotein production are regulated. Whereas all cell lines express many genes commonly considered as markers of liver parenchymal cells (e.g. albumin and apolipoproteins), they display distinct phenotypes characterized by their differential expression of either CYP7A1 or MTP. One single cell clone (L35 cells) expresses CYP7A1 at levels similar to that of rat liver (25)(26)(27)(28) but expresses no detectable MTP (29). In contrast, another line (FAO cells) do not express detectable levels of CYP7A1 enzyme activity or mRNA but do express high levels of MTP (25)(26)(27)(28)(29).
The lack of MTP expression in L35 cells causes a complete block in their ability to assemble and secrete apoB-containing lipoproteins (29). Detailed analysis of L35 cells showed that their inability to express MTP is caused by a block in the transcription of the MTP gene (29). In the present report, we provide evidence showing that ARP-1/COUP-TFII (30, 31) plays a dual role in producing the L35 cell phenotype. Thus, both the transcriptional repression of the MTP gene and induction of the CYP7A1 gene displayed by L35 cells require the association of ARP-1/ COUP-TFII to specific promoter elements in each gene.

EXPERIMENTAL PROCEDURES
Cell Culture-Cells were cultured and transfected as described previously (29). FAO cells were obtained as a gift from Franz Simon (University of Colorado). L35 cells were obtained as described (25).
Cells were transfected using LipofectAMINE reagent (Invitrogen) according to the protocols from the manufacturer, with minor modifications (29). A day before transfection, L35 and FAO cells (2 ϫ 10 5 ) were seeded on 12-well plates. On the day of transfection, cells were transfected with 0.6 g of promoter/luciferase reporter construct and 6 ng of pRL-CMV plasmid as an internal control to normalize the MTP promoter activity for differences in transfection efficiency. The normalized pRL-CMV activities are reported relative to activity of the empty vector (no insert) from parallel experiments performed at the same time with the same reagents and cells. Varying doses of COUP-TFII or RXR␣ expression vectors were added as indicated in the figure legends. The final DNA concentration for each assay was maintained constant by adding the expression plasmid without an insert. After transfection, cells were incubated for 48 h and harvested using passive lysis buffer (Promega, Madison, WI). Luciferase activities were measured using the Dual-Luciferase Reporter Assay system (Promega).
Reporter Gene Constructs and Expression Vectors-Several chimeric fusion constructs containing various lengths of the 5Ј-flanking MTP sequence cloned upstream of the luciferase reporter gene were prepared. Initially, the Ϫ612/ϩ84 fragment of the human MTP promoter was amplified by PCR using the MTP expression vector (a gift from Dr. N. Hariharan) as a template. The amplified fragment was then cloned into KpnI and BglII sites of promoterless luciferase reporter vector, pGL3Basic (Promega). The subsequent 5Ј deletions for human MTP promoters were generated by PCR, each tailored to have varying restriction enzyme sites at the 5Ј-end. The 3Ј-ends were all terminated at ϩ84 and had the same BglII sites.
To construct the MTP activator/p36 hybrid promoter reporter, the Ϫ36 to ϩ31 fragment of the prolactin promoter was isolated by digesting the original "P36 B(Ϫ) Luciferse" construct (a gift from Dr. Chris Glass, University of California San Diego) with XhoI and HindIII. Subsequently, the digested fragment was subcloned into XhoI and HindIII cloning sites of pGL3Basic (Promega). Two strands of oligonucleotides (sense and antisense) containing the Ϫ115 to Ϫ77 portion of human MTP promoter sequence with flanking 5Ј-ends were synthesized, annealed, and phosphorylated before they were ligated into MluI and XhoI sites of the p36-pGL3basic construct. Subsequently, the digested fragment was subcloned into XhoI and HindIII cloning sites of pGL3Basic (Promega).
A specific mutation in the DR1 sequence of the MTP promoter was generated by using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). In vitro mutagenesis was carried out using rat MTP135-luciferase reporter vector as the template and two oligonucleotide primers (mutated bases are underlined) each complementary to opposite strands of the vector (forward, 5Ј-GGA GTT TGG AAT CTG TGC TTT CCC CTA TAG-3Ј; reverse, 5Ј-CTA TAG GGG AAA GCA CAG ATT CCA AAC TCC-3Ј). All promoter deletion constructs and site-specific mutant constructs were verified by DNA sequencing. Expression plasmid for COUP-TFII was a gift from Dr. Ming-Jer Tsai (Baylor Medical University). Dr. Ronald Evans (Salk Institute, LaJolla, CA) kindly provided us with the RXR␣ expression plasmid. The rat CYP7A (Ϫ416/ϩ32) luciferase reporter plasmid was a gift from Dr. John Chiang (Northeastern Ohio Universities College of Medicine).
Cloning of Mouse and Rat MTP Promoter-A mouse genomic clone containing an 8.5-kb HindIII fragment in pBlueScript (a gift from Steve Young, University of California San Francisco) was partially sequenced to obtain the proximal mouse promoter sequence. Since the genomic clone contained the complete exon 1 sequence and the 5Ј-flanking region of ϳ5 kb in length, a sequencing primer was designed from the coding region (96 -116 bp downstream of the translation start site, in the inverse direction; 5Ј-GAG TAC GTG AGC TTG TAT AGC-3Ј). Based on the obtained sequencing results, the Ϫ231 to ϩ63 portion of the mouse MTP gene was PCR-amplified and cloned into SacI and BglII sites of pGL3basic luciferase reporter vector.
To identify the rat MTP promoter sequence, the genomic DNA from L35 and FAO cells (which are from rat origin) was isolated and purified using the DNeasy tissue kit (Qiagen, Alameda, CA). Since the proximal promoter region and the coding region are highly conserved among different species, the forward primer designed from the mouse 5Јflanking region (forward, 5Ј-TCT TAA AAG CGA GAG ACT AC-3Ј) and the reverse primer designed from the mouse coding region (reverse, 5Ј-GTA GGA GGA GAA GAA GCA-3Ј) allowed the amplification of L35 and FAO genomic DNA. Subsequently, the amplified PCR products were subcloned into TA cloning vector (Invitrogen), and the DNA sequences were obtained. The amplified region from L35 and FAO genomic DNA displayed no difference in the DNA sequence. For the construction of rat MTP reporter vector, the Ϫ135 to ϩ66 region of the cloned L35 MTP promoter was PCR-amplified and cloned into the SacI and BglII sites of pGL3basic.
Preparation of Nuclear Extracts-Nuclear extracts from L35 and FAO cells were prepared as described (21). The cells were trypsinized and harvested by centrifugation, washed with 1ϫ phosphate-buffered saline, and resuspended in a hypotonic buffer (10 mM HEPES, pH 7.9, at 4°C, 1.5 mM MgCl 2 , 10 mM KCl, 0.2 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol). After a 10-min incubation on ice, the cells were lysed by Dounce homogenizer. The nuclei were pelleted by centrifugation and resuspended in low salt buffer (20 mM HEPES, pH 7.9, at 4°C, 25% glycerol, 1. polynucleotide kinase (New England Biolabs, Beverly, MA), followed by purification on a G50 column. For binding reaction, 15 g of nuclear extracts were incubated with 2 ϫ 10 4 cpm probe on ice for 20 min in a total volume of 15 l of solution containing 20 mM HEPES (pH 7.9), 10% glycerol, 100 mM KCl, 1 mM EDTA, and 2 g of poly(dI-dC). For supershift experiments, 1 l of specific antibodies were added to preincubated DNA-protein complexes for an additional 20 min on ice. Antibodies against the ARP-1/COUP-TFII (sc-6576X) and RXR␣ (sc-553X) were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). DNAprotein complexes were resolved on 4% native polyacrylamide gel electrophoresis containing 0.5ϫ TBE buffer.
Construction of the L35 AD Fusion Library-The SMART cDNA Library Construction Kit (Clontech, Palo Alto, CA) was used to synthesize cDNA from L35 cells. The isolated cDNA was then purified and digested with SfiI and ligated into modified pGADT7 vector, which contains two unique SfiI sites. (The original pGADT7 vector (Clontech) has been modified previously by introducing two distinct SfiI sites into multiple cloning sites to ensure proper ligation and unidirectional expression.) After the overnight ligation, the ligated products were introduced into electrocompetent DH5␣ cells (Invitrogen) via electroporation. Subsequently, the primary library was amplified, and the library plasmids were isolated and purified using the NucleoBond Plasmid Mega Kit (Clontech).
Yeast One-hybrid System-The MATCHMAKER one-hybrid system (Clontech) was used to isolate the cDNA encoding the protein that binds to DR1 element of MTP promoter. The specific procedures were followed according to protocols established by the manufacturer. Three tandem copies of a 30-bp DNA fragment containing the DR-1 sequence from the human MTP promoter (5Ј-TGA CCT TTC CCC AAA GAT AAA CAT GAT TGT-3Ј) were cloned upstream of pHISi and pLacZ genes in yeast reporter vectors. These plasmids were then transformed into yeast strain YM4271 to generate a dual reporter strain. This reporter strain was then transformed with the L35 GAL4-AD cDNA library using LiAc method, and the transformants were selected on SD/ϪHis/ϪLeu medium containing 15 mM 3-amino-1,2,4-triazole. Large colonies from the ϪHis/ϪLeu plates were then transferred to a filter paper (VWR, San Diego, CA) and further screened for ␤-galactosidase activity. The plasmids from positive yeast colonies were isolated and transformed into DH5␣ bacterial competent cells. Subsequently, the plasmid DNA extracted from bacteria was transformed back into the yeast reporter strain to confirm Hisϩ/lacZϩ phenotype. The cDNAs from reproducible positive clones were partially sequenced, and BLAST searched.
Western Blot Analysis-Nuclear proteins prepared from L35 and FAO cells were separated on a 4 -12% gradient SDS-polyacrylamide gel and transferred onto nitrocellulose membranes. After the blots were blocked with 10% dry milk in water for 1 h at room temperature, they were incubated overnight at 4°C with an antibody specific for ARP-1/ COUP-TFII (a gift from Dr. S. Karathanasis), RXR␣ (sc-553), HNF-4 (sc-6556), or HNF-1 (sc-6547) (Santa Cruz Biotechnology). After rinsing with TBS-T, the blots were incubated with secondary antibodies diluted in TBS-T for 1 h at room temperature. The hybridized bands were visualized by chemiluminescence as in the protocol of the ECL kit (Amersham Biosciences).

Identification of the Cis Element
Responsible for the Repression of the MTP Gene by L35 Cells-We have previously shown that the 612-bp human MTP promoter contains the sequences responsible for the cell type-specific repression of the MTP gene by L35 cells (29). To identify the minimal DNA sequence required for the repression of MTP by L35 cells, nested 5Ј deletions of the MTP promoter were generated and fused to a luciferase reporter gene. The relative activities in L35 and FAO cells of the promoter/luciferase reporters were determined by assaying luciferase activities in extracts from transfected cells. Deletions of the MTP promoter sequences between Ϫ612 and Ϫ135 had no effect on promoter activity in either cell type (Fig.  1). However, further deletions beyond Ϫ135 bp led to a marked decrease in the promoter activity in FAO cells, resulting in a level comparable with that of L35 cells (Fig. 1). These data indicate that sequences within the 135-bp promoter are sufficient to confer the differential expression of MTP in L35 and FAO cells.
Functional and Sequence Conservation of the Proximal Promoter of the MTP Gene-Sequence analysis revealed that the proximal promoter sequences (Ϫ135 to ϩ66 bp) of the MTP gene in FAO and L35 cells are identical (Fig. 2). Studies by others have shown that the proximal promoter sequences of the human and hamster MTP genes are highly conserved (32) (Fig.  2). Indeed, our sequence analysis indicated that the rat MTP promoter sequence (obtained from L35 and FAO cells) is Ͼ80% identical to those found in the mouse, hamster, and human genes (Fig. 2). Furthermore, the proximal MTP promoter sequences from mice, rats, and humans displayed similar differences in their relative activities by L35 cells and FAO cells (Fig.  3). These data indicate that the sequences responsible for the differential expression by L35 and FAO cells are conserved in all three species and are likely to be important in the regulation of MTP expression. The apparent higher activity displayed by the reporter containing the mouse MTP promoter may be caused by the presence of additional 5Ј sequences, which may enhance transcription.
Activator Elements in Ϫ135 to Ϫ77 bp of the MTP Promoter Exhibit Similar Activity in L35 and FAO Cells-The finding that deleting sequences between Ϫ135 and Ϫ89 bp of the MTP promoter markedly reduced the promoter activity in FAO cells (Fig. 1) suggests that they contain key element(s) responsible for the transcriptional activation. It has been proposed that these sequences contain an insulin response element and putative binding sites for HNF-1, AP-1, HNF-4 (32), and SREbinding protein (33). Both insulin response element (Ϫ123 to Ϫ113 bp) and SRE (Ϫ124 to Ϫ116 bp) have previously been identified as negative elements (32,33). However, since deletion of sequences from Ϫ135 to Ϫ111 bp of the promoter did not affect the promoter activity (Fig. 4A), it is unlikely that the insulin response element and SRE are responsible for the differential expression of MTP by L35 and FAO cells.
Deletions beyond Ϫ111 bp reduced the promoter activity in FAO cells to levels comparable with that of L35 cells (Fig. 4A). These findings suggest that a putative HNF-1 binding site (Ϫ111 to Ϫ99 bp) and a putative HNF-4 binding site (Ϫ115 to Ϫ77 bp) may be necessary for transcriptional activation of the MTP gene (32). To determine whether these sequences contribute to the differential expression of MTP by FAO and L35 cells, the Ϫ115 to Ϫ77 bp region of the MTP promoter was inserted 5Ј to the minimal prolactin promoter (P36). Without inserts, the P36 promoter exhibits almost no activity (Fig. 4B). However, the addition of the Ϫ115 to Ϫ77 bp portion of the MTP promoter increased the transcriptional activity of the heterologous P36 promoter by Ͼ7-fold in both cell types (Fig. 4B). Furthermore, EMSAs indicate that nuclear extracts from L35 and FAO cells formed similar DNA-protein complexes with the Ϫ115 to Ϫ77 bp portion of the MTP promoter (Fig. 4C). These data indicate that the sequences containing the putative HNF-1 and HNF-4 binding sites (Ϫ115 to Ϫ77 bp) do not contribute to the differences in the expression of MTP by L35 and FAO cells.
Identification of the Cis Element Responsible for the Differential Expression of MTP by L35 and FAO Cells-Comparative analysis of the MTP proximal promoter sequences downstream of Ϫ77 bp of rat, mouse, hamster, and human revealed a conserved direct repeat (DR1) hormone response element (Ϫ51 to Ϫ39 bp) (Fig. 2). The 5Ј half-site (Ϫ51 to Ϫ46; TGACCT) conforms to the consensus sequence, whereas the 3Ј half-site (Ϫ44 to Ϫ39; TCCCCT) has a 2-bp mismatch. The functional significance of this element was tested by mutagenesis of the 5Ј half-site (Fig. 5A). In L35 cells, mutation of the 5Ј half-site resulted in a 3-fold increased activity promoter activity, whereas in FAO cells, the 5Ј half-site mutation caused the promoter activity to decrease (by Ͼ65%) (Fig. 5A). These findings strongly suggest that this DR1 element functions as a negative element in L35 cells, whereas it acts as a positive element in FAO cells.
To examine the binding of proteins to the DR1 element, EMSAs were performed with radiolabeled oligonucleotides containing either the wild-type or the mutant DR1 element. Incubation of the wild-type DR1 probe with nuclear extracts from both cell types produced multiple DNA-protein complexes that were observed as four distinct radiolabeled bands (Fig. 5B). L35 cells produced bands A, B, and C (Fig. 5B, lane 1). In contrast, FAO cells produced bands A, C, and D (Fig. 5B, lane 4). Incubation of the radiolabeled mutant DR1 probe with nuclear extracts from both cell types eliminated the cell type-specific differential banding patterns producing only band C (data not shown). The addition of a 500-fold excess of unlabeled oligonucleotides containing the wild-type DR1 sequence blocked the formation of all four DNA-protein complexes formed by nuclear extracts from both L35 and FAO cells (data not shown). In contrast, adding a 500-fold excess of unlabeled oligonucleotides containing the mutant DR1 sequence had almost no effect on the formation of all four DNA-protein complexes produced by nuclear extracts from L35 and FAO cells (data not shown). These findings indicate that cell type-specific DNA-protein complexes are formed by both cell types with the DR1 element. The DNA-protein complexes represented by bands A and C are common to both cell types. The DNA-protein complex represented by band D is unique to FAO cells and may represent the binding of a transcriptional activator. The DNA-protein complex represented by band B, formed by nuclear extracts from L35 cells but not from FAO cells, may represent protein factors responsible for the transcriptional repression of the MTP gene unique to L35 cells.

Differential Regulation of MTP Gene Transcription Is Mediated by Competition between ARP-1/COUP-TFII and RXR␣
Binding to the DR1 Sequence-Putative transcriptional regulatory proteins that bind to the DR1 element were identified FIG. 4. The MTP activator region is functional in L35 cells. A, 5Ј deletion analysis of human MTP promoter. 5Ј deletions of MTP promoters were generated by PCR at the indicated positions and were subcloned into a luciferase reporter vector. For all constructs, the 3Ј-ends of promoters were terminated at ϩ84. Promoter activities of these constructs were determined by transient transfection assay. Filled bars represent the luciferase activity in FAO cells, and empty bars represent the luciferase activity in L35 cells. All luciferase values were normalized as described under "Experimental Procedures." Error bars indicate S.D. of triplicate samples. B, MTP activator/P36 heterologous promoter activity. The MTP activator element (Ϫ115 to Ϫ77) was cloned 5Ј to the heterologous minimal P36 promoter, and the hybrid construct was transfected into both L35 and FAO cells. Filled bars represent the luciferase activity of the MTP activator/P36 hybrid construct, and empty bars represent the activity of P36 promoter alone. All luciferase values were normalized as described under "Experimental Procedures." Error bars indicate S.D. of triplicate samples. C, electromobility shift assay of the activator element. The MTP activator element (Ϫ115 to Ϫ77) was 32 P-end-labeled and incubated with the nuclear extracts (NE) from either FAO or L35 cells. The protein-DNA complexes were resolved by electrophoresis using a 4% polyacrylamide gel.
using the yeast one-hybrid system (Clontech). When a tandem repeat containing the DR1 sequence was used as the bait for screening the GAL4-AD cDNA library prepared from L35 cells, ARP-1/COUP-TFII was identified as a positive clone. When the same tandem repeat containing the DR1 sequence was used as the bait for screening a GAL4-AD cDNA library prepared from rat liver, RXR␣ was identified as a positive clone. To verify the binding activity of ARP-1/COUP-TFII and RXR␣ to the DR1 sequence, we performed EMSA/supershift analysis using an antibodies specific for ARP-1/COUP-TFII and RXR␣ (Fig. 5B). Adding an anti-ARP-1/COUP-TFII antibody to L35 nuclear extracts incubated with radiolabeled oligonucleotides containing the wild-type DR1 sequence resulted in the formation of a distinct supershifted band and the loss of band B (Fig. 5B, lane  2). In contrast, an anti-RXR␣ antibody caused no discernible change in the EMSA bands (Fig. 5B, lane 3). Opposite results were obtained using nuclear extracts from FAO cells. Thus, adding an anti-ARP-1/COUP-TFII antibody to FAO nuclear extracts caused no discernible change in the EMSA bands (Fig.  5B, lane 5); adding the anti-RXR␣ antibody to FAO nuclear extracts incubated with radiolabeled oligonucleotides containing the wild-type DR1 sequence decreased EMSA band D while forming a distinct supershift band (Fig. 5B, lane 6). The combined data suggest that ARP-1/COUP-TFII is bound to the DR1 site in L35 cells but not in FAO cells, whereas RXR␣ is bound to the DR1 in FAO cells but not in L35 cells. The simplest interpretation of these findings is that ARP-1/COUP-TFII and a complex containing RXR␣ compete with each other for the DR1 site. In L35 cells, ARP-1/COUP-TFII is bound to this site, causing repression of the MTP gene. In FAO cells, a complex containing RXR␣ activates MTP gene transcription by binding to this site.
Increased Levels of ARP-1/COUP-TFII (L35 cells) and RXR␣ (FAO Cells) Correlate with Repression and Activation of MTP Gene Transcription-To determine whether the differential re-cruitment of either ARP-1/COUP-TFII or RXR␣ to the conserved DR1 site in L35/FAO cells is due to their relative levels in these cells, we estimated their relative protein levels by Western blotting. Whereas the expression levels of HNF-1 and HNF-4 were similar in these cells, L35 cells contained 2-fold more ARP-1/COUP-TFII protein and 50% less RXR␣ as compared with FAO cells (Fig. 6). The combined data suggest that the relative expression levels of ARP-1/COUP-TFII and RXR␣ may contribute to the cell type-specific expression of MTP by L35 and FAO cells.
Expression of ARP-1/COUP-TFII and RXR␣ Differentially Regulates the Activity of the MTP Promoter-If the differential expression and subsequent binding of ARP-1/COUP-TFII and RXR␣ to the DR1 element is responsible for the differences in MTP gene transcription exhibited by L35 and FAO cells, cotransfection of FAO cells with an ARP-1/COUP-TFII expression vector should repress MTP promoter activity, whereas co-transfection of L35 cells with an RXR␣ expression vector should activate MTP promoter activity. Indeed, as predicted, co-transfection of an ARP-1/COUP-TFII expression vector (50 ng) decreased MTP promoter activity in FAO cells by 48% (Fig.  7A). At the highest concentration examined (200 ng), MTP promoter activity was reduced by ϳ70% (Fig. 7A). In contrast, co-transfection of an RXR␣ expression vector (200 ng) without ligand (i.e. 9-cis-retinoic acid) increased MTP promoter activity in L35 cells by 2.5-fold (Fig. 7B). Adding the 9-cis-retinoic acid (1 M) with the RXR␣ expression vector (200 ng) to L35 cells resulted in a 6.5-fold increase in MTP promoter activity (Fig.  7B). These findings show that ARP-1/COUP-TFII can repress the activity of the MTP promoter in FAO cells toward levels displayed by L35 cells, whereas RXR␣ together with its ligand 9-cis-retinoic acid can induce the activity of the MTP promoter in L35 cells toward levels displayed by FAO cells.
Increased Levels of ARP-1/COUP-TFII in L35 Cells Correlate with Activation of CYP7A1 Gene Transcription-Unlike most other stable lines of hepatoma cells including FAO cells, L35 cells express the liver-specific enzyme CYP7A1 at levels similar to those expressed in rat liver (25). The expression of CYP7A1 by L35 cells is the result of transcriptional activation of the endogenous CYP7A1 gene via a mechanism not yet defined (25). In human (HepG2) hepatoma cells, ARP-1/COUP-TFII has been shown to bind to and activate both rat and human CYP7A1 promoters (34,35). We examined the possibility that the increased levels of ARP-1/COUP-TFII observed in L35 cells (Fig. 6) may bind to and activate the CYP7A1 promoter. EMSA/ supershift analysis using labeled oligonucleotides containing a defined segment of the rat CYP7A1 promoter (34,35) showed that nuclear extracts from L35 cells produced a distinct supershifted band with an antibody against ARP-1/COUP-TFII (Fig.  8A, lane 3). In contrast, nuclear extracts from FAO cells produced almost no detectable supershifted band (Fig. 8A, lane 4). These results suggest that ARP-1/COUP-TFII is associated with the CYP7A1 promoter in L35 cells but not in FAO cells.
Additional studies show that the activity of the rat CYP7A1 promoter was ϳ7-fold higher in L35 cells compared with FAO cells (Fig. 8B). Furthermore, co-transfecting FAO cells with increasing amounts of the ARP-1/COUP-TFII expression plasmid resulted in a dose-dependent increase in the activity of a rat CYP7A1 promoter-reporter (Fig. 8C). At the highest concentration of ARP-1/COUP-TFII expression plasmid used (200 ng), the activity of the CYP7A1 promoter was increased by 6-fold. These data suggest that in addition to repressing the MTP gene, the increased content of ARP-1/COUP-TFII in L35 cells may also induce the CYP7A1 gene. DISCUSSION In this report, we show that the nuclear orphan receptor ARP-1/COUP-TFII acts on two distinct metabolic pathways that define the phenotype of cultured hepatoma cells. The binding of ARP-1/COUP-TFII to a conserved DR1 site located in the Ϫ51 to Ϫ39 region of the MTP promoter repressed MTP gene transcription. ARP-1/COUP-TFII-mediated repression of the MTP gene results in an inability of L35 cells to assemble and secrete apoB-containing lipoproteins (29). Within the same cellular context, ARP-1/COUP-TFII bound to the rat CYP7A1 promoter, resulting in enhanced transcription. This dual repressor/activator characteristic of ARP-1/COUP-TFII results in the unique phenotype characterized by L35 cells: blocked MTP expression and enhanced CYP7A1 expression. Our findings provide further support for the proposal that ARP-1/COUP-TFII plays an important role in regulating the expression of genes controlling hepatic lipid metabolism (31,36,37) To our knowledge, our findings are the first to show that ARP-1/COUP-TFII represses MTP gene transcription. ARP-1/ COUP-TFII was first identified as a member of the nuclear hormone receptor superfamily shown to bind to a DR1 promoter element in the chicken ovalbumin gene and activate its transcription (30). ARP-1/COUP-TFII is highly expressed in the liver (38) and has been shown to repress the transcription of several apolipoprotein genes (e.g. human apoA1 (31), human apoCIII (36), and apoB (37,39)). Together with our findings showing that ARP1/COUP-TFII represses the MTP promoter, these data suggest that ARP1/COUP-TFII acts to repress the transcription of the network of genes involved in the very low density lipoprotein assembly/secretion pathway. A similar regulatory mechanism exists in invertebrates. In the mosquito Aedes aegypti, AaSv, a homologue of COUP-TF, negatively regulates ecdysone receptor complex-mediated transactivation of vitellogenin, an invertebrate lipoprotein (40).
Our data show that the DR1 sequence located in the Ϫ51 to Ϫ39 region of the MTP promoter is required for the cell typespecific repression of the MTP gene in L35 cells. Mutation of the DR1 site abolished both the repression of MTP transcription exhibited by L35 cells, as well as the activation of MTP transcription exhibited by FAO cells. As a result, the MTP promoter construct containing the DR1 mutation was expressed at similar levels by both cell types (Fig. 5A). The additional findings showing that an antibody specific to ARP-1/COUP-TFII supershifted the 32 P-labeled DR1 probe incubated with nuclear extracts from L35 cells but not FAO cells, whereas an antibody specific to RXR␣ supershifted the 32 Plabeled DR1 probe incubated with nuclear extracts from FAO cells but not L35 cells (Fig. 5C) indicate that ARP-1/COUP-TFII and RXR␣ compete for the DR1 site. Furthermore, co-transfection studies provide functional evidence demonstrating that in FAO cells, ARP-1/COUP-TFII acts as a repressor of MTP promoter activity (Fig. 7A), whereas in L35 cells, RXR␣ acts as an inducer (Fig. 7B), suggesting that the relative cellular content of ARP-1/COUP-TFII to RXR␣ (Fig. 6) is an important determinant of MTP promoter activity. Since, under identical conditions, increasing the cellular content of ARP-1/ COUP-TFII by co-transfection of the ARP-1/COUP-TFII expression plasmid increased the activity of the CYP7A1 promoter by 6-fold (Fig. 8C), the repressive effect of ARP-1/COUP-TFII on the activity of the MTP promoter is specific and not caused by a general impairment of transcription. In regard to the activation of the CYP7A1 promoter by ARP-1/COUP-TFII, our findings are consistent with those obtained from human hepatoma cells showing that ARP-1/COUP-TFII activates the transcription of the rat and human CYP7A1 promoters (34,35).
In other studies, ARP-1/COUP-TFII represses transcription indirectly by competing with an activator protein for a binding site. For example, ARP-1/COUP-TFII reduces the transcriptional activity of the human apoA1 gene by competing with HNF-4 (36). Furthermore, ARP-1/COUP-TFII represses transcription by competing with RXR for binding to a DR1 element FIG. 8. Differential regulation of CYP7A1 by ARP-1/COUP-TFII. A, EMSA/supershift analysis of the CYP7A promoter element with ARP-1/COUP-TFII-specific antibodies. The CYP7A promoter element (Ϫ74/Ϫ53) was 32 P-end-labeled and incubated with the nuclear extracts (NE) from both L35 and FAO cells. Anti-ARP-1/COUP-TFII-specific antibodies were added to lanes 3 and 4. B, comparison of CYP7A promoter activities in L35 and FAO cells. The CYP7A (Ϫ416/ϩ32) luciferase reporter plasmid was transiently transfected into FAO (filled bar) and L35 (empty bar) cells. All luciferase values were normalized as described under "Experimental Procedures." Error bars indicate S.D. from the mean of triplicate samples. C, effects of cotransfection of ARP-1/COUP-TFII on CYP7A promoter activity in FAO cells. The CYP7A (Ϫ416/ϩ32) luciferase reporter plasmid was cotransfected with the indicated amounts of an ARP-1/COUP-TFII expression plasmid into FAO cells. All luciferase values were normalized as described under "Experimental Procedures" and then reported as percentages relative to transfections with the reporter alone (% Control). (41) as well as recruiting the co-repressors N-CoR and SMRT (42). Our studies show that the binding of RXR␣ to the conserved DR1 element in the MTP promoter activates transcription. The activation (RXR␣) and repression (ARP-1/COUP-TFII) of the MTP promoter by the differential binding of RXR␣ and ARP-1/COUP-TFII, respectively, to the DR1 site is consistent with the findings showing that mutation of the DR1 site decreases the activity of the MTP promoter in FAO cells, whereas it increases the activity in L35 cells (Fig. 5A).
Two lines of evidence indicate that the relative cellular content of ARP-1/COUP-TFII versus RXR␣ is a major determinant of MTP promoter activity. First, the relative cellular content of ARP-1/COUP-TFII compared with RXR␣ (Fig. 6) varies inversely to the relative activity of the MTP promoter in L35 and FAO cells. Second, co-transfection of the ARP-1/COUP-TFII expression vector reduced MTP promoter activity by 70% in FAO cells (Fig. 7A), whereas co-transfection of the RXR␣ expression vector with its ligand 9-cis-retinoic acid increased the activity by 6-fold in L35 cells (Fig. 7B), showing that changing the relative cellular content of the repressor/activator results in a correlating change in promoter activity.
Our results indicate that the increased cellular content of ARP-1/COUP-TFII both repressed MTP gene transcription and activated CYP7A1 gene transcription. However, it is important to emphasize that ARP-1/COUP-TFII is one of many transcription factors that act in concert to ultimately determine the levels of MTP and CYP7A1 gene expression. As shown in our nested deletion analysis of the MTP promoter (Fig. 4A), there are many individual nuclear receptor binding sites that are required for transcription of the MTP gene. Expression of CYP7A1 is determined by an aggregate of factors that regulate gene transcription (reviewed in Refs. 19, 20, and 43) and turnover of CYP7A1 mRNA (44). Additional data are required in order to determine the hierarchical role of ARP-1/COUP-TFII in the regulation of expression of MTP and CYP7A1. Since deletion of ARP-1/COUP-TFII in mice is embryonic lethal (45), liver-specific knockouts of ARP-1/COUP-TFII should uncover its importance in regulating the expression of hepatic genes.
Because of its role in nutrition and as the precursor for many of the plasma lipoproteins, the rate of very low density lipoprotein secretion by the liver has an important impact on diverse physiological and disease processes (6). The rate of very low density lipoprotein secretion by the liver is closely linked to the expression and activity of MTP (15, 46 -48). Multiple casecontrolled studies have revealed associations between a specific polymorphism in the promoter of the human MTP gene and changes in plasma and liver lipids as well as susceptibility to vascular and hepatic diseases (49 -51). Our findings showing that the cellular content of ARP-1/COUP-TFII regulates MTP promoter activity provide a new insight into the complexity determining the expression of MTP, a physiologically important parameter of hepatic function.