Coordinate Transcriptional Repression of Liver Fatty Acid-binding Protein and Microsomal Triglyceride Transfer Protein Blocks Hepatic Very Low Density Lipoprotein Secretion without Hepatosteatosis*

Unlike the livers of humans and mice, and most hepatoma cells, which accumulate triglycerides when treated with microsomal triglyceride transfer protein (MTP) inhibitors, L35 rat hepatoma cells do not express MTP and cannot secrete very low density lipoprotein (VLDL), yet they do not accumulate triglyceride. In these studies we show that transcriptional co-repression of the two lipid transfer proteins, liver fatty acid-binding protein (L-FABP) and MTP, which cooperatively shunt fatty acids into de novo synthesized glycerolipids and the transfer of lipids into VLDL, respectively, act together to maintain hepatic lipid homeostasis. FAO rat hepatoma cells express L-FABP and MTP and demonstrate the ability to assemble and secrete VLDL. In contrast, L35 cells, derived as a single cell clone from FAO cells, do not express L-FABP or MTP nor do they assemble and secrete VLDL. We used these hepatoma cells to elucidate how a conserved DR1 promoter element present in the promoters of L-FABP and MTP affects transcription, expression, and VLDL production. In FAO cells, the DR1 elements of both L-FABP and MTP promoters are occupied by peroxisome proliferator-activated receptor α-retinoid X receptor α (RXRα), with which PGC-1β activates transcription. In contrast, in L35 cells the DR1 elements of both L-FABP and MTP promoters are occupied by chicken ovalbumin upstream promoter transcription factor II, and transcription is diminished. The combined findings indicate that peroxisome proliferator-activated receptor α-RXRα and PGC-1β coordinately up-regulate L-FABP and MTP expression, by competing with chicken ovalbumin upstream promoter transcription factor II for the DR1 sites in the proximal promoters of each gene. Additional studies show that ablation of L-FABP prevents hepatic steatosis caused by treating mice with an MTP inhibitor. Our findings show that reducing both L-FABP and MTP is an effective means to reduce VLDL secretion without causing hepatic steatosis.

ApoB is a uniquely large (Ͼ500 kDa) amphipathic protein essential for the assembly and secretion of triglyceride-rich VLDL (3,(15)(16)(17). The inability to produce apoB of sufficient size (ϳ35 kDa) is associated with an inability of both the liver and intestine to assemble and secrete VLDL (18). Normally, hepatic expression of apoB is constitutive; changes in hepatic secretion of apoB-containing lipoproteins are the result of variation in the amount of de novo synthesized apoB that is either secreted or degraded within hepatocytes (1,2).
MTP acts as both a lipid transfer protein (19) and as a facilitator of apoB folding and translocation (3,17). MTP facilitates the transfer of four major lipid classes (free cholesterol, phospholipids, triglycerides, and cholesterol esters) to the nascent apoB-containing lipoprotein particle (20) via a two-step process (21,22). Abrogation of one or more of these concerted MTPdependent processes leads to co-translational degradation of nascent apoB by the proteasome (23)(24)(25)(26)(27).
L-FABP, a highly abundant lipid-binding protein in the cytosol of liver parenchymal cells, facilitates fatty acid transport and utilization (28,29). Genetic disruption of L-FABP expres-* This work was supported by Grants HL57974, HL51648 (to R. A. D.), sion impairs the ability of the liver to efficiently import and transfer fatty acids into glycerolipid biosynthesis (9).
Hepatic VLDL assembly and secretion are highly variable among individuals and sensitive to changes in nutritional state (17). It is induced by carbohydrate feeding (30) and repressed by fasting (31). These nutritional changes in VLDL secretion are linked to sterol regulatory element-binding protein-mediated changes in the expression levels of key lipogenic enzymes (32,33). When the rate of hepatic de novo lipogenesis is reduced (i.e. fasting), fatty acids supplied by adipose tissue can provide sufficient substrate for the glycerolipid synthesis and VLDL assembly/secretion (34). Variations in hepatic expression levels of both L-FABP (9) and MTP (35) control the flux of fatty acids into glycerolipid biosynthesis and VLDL assembly/secretion.
In this study we use two distinct lines of hepatoma cells, each displaying distinct abilities to assemble and secrete VLDL (36,37) in order to uncover how the transcriptions of L-FABP and MTP are coordinately regulated. FAO cells, a rat hepatoma cell line used to study hepatic lipoprotein synthesis and secretion (38,39), express both L-FABP and MTP and exhibit the ability to assemble and secrete apoB-containing lipoproteins. In contrast, L35 cells derived as a single cell clone from FAO cells express neither L-FABP nor MTP and lack the ability to assemble and secrete apoB-containing lipoproteins. Our combined data support the proposal that L-FABP and MTP were derived from a common ancestral lipid-binding protein (40). Retention of the DR1 site in the promoter allowed the distinct lipid transfer functions of L-FABP and MTP to evolve while retaining a common mechanism to ensure that their expression varied concomitantly, and their concerted function is readily adaptable to fatty acid substrate supply.

EXPERIMENTAL PROCEDURES
Cell Culture-Cells were cultured and transfected as described (37). FAO cells were obtained as a gift from Franz Simon (University of Colorado). L35 cells were obtained as described (31).
Cells were transfected using Lipofectamine reagent (Invitrogen) according to manufacturer's protocol, with minor modifications (37). One day prior to transfection, L35 and FAO cells (2 ϫ 10 5 ) were seeded on 12-well plates. On the day of transfection, cells were transfected with 0.8 g of promoter/luciferase reporter construct and with 6 ng of pRL-CMV plasmid as an internal control for normalization of L-FABP and MTP promoter activities. The normalized pRL-CMV activities are reported relative to activity of the empty vector from parallel experiments. Varying doses of COUP-TFII expression vector was added as indicated in figure legends. The total DNA concentration for each assay was maintained constant by the addition of empty expression vector pCR 3.1 (Invitrogen). Upon transfection, cells were incubated for 48 h and harvested using passive lysis buffer (Promega). Luciferase activities were measured using the dual-luciferase reporter assay system (Promega).
The PPAR␣ and RXR agonists, WY-14,643 (WY) and 9-cisretinoic acid (RA), respectively (A.G. Scientific, Inc.), were dissolved in dimethyl sulfoxide (Me 2 SO, 0.15% v/v) and used at working concentrations of 10 M (WY) and 1 M (9-cis-RA). Briefly, upon transfection cells were treated for 48 h with ago-nists or Me 2 SO alone as indicated. Cells were harvested and promoter activity assayed as described above.
Reporter Gene Constructs and Expression Vectors-The wild type and mutant rat MTP reporter vectors (Ϫ135/ϩ66) were as described previously (37). To generate the wild type rat L-FABP reporter vector (Ϫ141/ϩ66), genomic DNA was isolated and purified from FAO cells using the DNeasy tissue kit (Qiagen). The promoter fragment was generated by PCR using the primers with indicated restriction enzyme sites as follows: forward, 5Ј(KpnI)-GAA CAA ACT TCT GCC GGT ACC ATT CTG ATT TTT A-3Ј, and reverse, 5Ј(BglII)-TTC ATG GTG GCA ATG AGA TCT CCT TTC CAC AGC TGA-3Ј. The promoter fragment was then cloned into KpnI and BglII sites of the empty luciferase reporter vector PGL3Basic (Promega).
To generate the mutant L-FABP reporter vector, a specific mutation in the proximal DR1 sequence was generated using the QuikChange site-directed mutagenesis kit (Stratagene). For the in vitro mutagenesis, the wild type rat L-FABP (Ϫ141/ϩ66)luciferase reporter vector was used as the template along with two oligonucleotide primers (mutated bases underlined), each complementary to opposite strands of the vector as follows: forward, 5Ј-AAT CGA CAA TCA CTG TGC TAT GGC CTA TAT TT-3Ј; reverse, 5Ј-AAA TAT AGG CCA TAG CAC AGT GAT TGT CGA TT-3Ј. The site-specific mutant construct was verified by DNA sequencing. The expression plasmid for COUP-TFII was a gift from Dr. Ming-Jer Tsai (Baylor College of Medicine).
cDNA Synthesis and Real Time PCR-Total RNA was isolated from either frozen liver using the Versagene RNA tissue kit (Gentra Systems, Inc.) or from cells using the Versagene RNA cell culture kit (Gentra Systems, Inc.) with on-column DNA removal per the manufacturer's instructions. The RNA concentrations were determined by spectrophotometer at 260 nm. First strand cDNA was synthesized from 0.5 g of total RNA using the Bio-Rad iScript for reverse transcription (Bio-Rad). Specific primers for each gene (supplemental Table 1) were designed using gene sequences from GenBank TM . To avoid amplification of genomic DNA, the primers were positioned to span exon junctions. All primers were synthesized by IDT.
Real time PCR analysis was performed with the iCycler using SyBr Green supermix according to manufacturer's instructions (Bio-Rad). The reactions were analyzed in triplicate with specific product monitored using meltcurve analysis. The expression data were normalized to an endogenous control, either 18 S ribosomal RNA or acidic ribosomal phosphoprotein P0 (36B4). The level of both 18 S RNA and 36B4 was invariable among samples of all experiments. The relative expression levels were calculated according to the formula 2 Ϫ⌬Ct , where ⌬Ct is the difference in threshold cycle (Ct) values between the target and either the 18 S or 36B4 endogenous control.
Chromatin Immunoprecipitation Assay and Relative Quantitation-Cells were cultured in complete medium in 150-mm dishes until Ϸ70 -80% confluent. Where indicated the agonists WY (10 M) and 9-cis-RA (1 M) were added to cell culture medium for 48 h prior to harvesting. The cells were then fixed by the addition of 280 l of 37% formaldehyde (Sigma) to 10 ml of culture medium for 10 min at 37°C, harvested, and processed for immunoprecipitation using the ChIP-IT shearing kit (Active Motif) and ChIP-IT chromatin immunoprecipitation kit (Active Motif) for chromatin immunoprecipitation according to the manufacturer's protocol. Immune complexes were eluted, reverse cross-linked using 5 M NaCl at 65°C, treated with proteinase K, and purified using mini-columns provided with ChIP-IT kit.
Specific genomic DNA fragments from immunoprecipitated samples and inputs were quantitated by real time PCR with SyBr Green Supermix (Bio-Rad) as indicated above. As a control for region selectivity of immunoprecipitation-specific enrichment differences, amounts of noncoding distal untranslated regions were determined for each sample. The antibodies used were against COUP-TFII (a gift from Dr. S. Karathanasis), PPAR␣ (sc-9000X; Santa Cruz Biotechnology), and control IgG (Active Motif). Primer sets were designed to amplify the following rat genomic DNA regions: MTP-DR1 (forward, 5Ј-TAG TGA GCC CTT CCA TGA AC-3Ј; reverse, 5Ј-CAG AAT CTG CGA CAA CAG TG-3Ј), L-FABP-DR1 (forward, 5Ј-GAG TTA ATG TTT GAT CCT GGC C-3Ј; reverse, 5Ј-CCA CCC ACT GTT GGC TAT TTT-3Ј); and L-FABP 3Ј-untranslated region (forward, 5Ј-GTC TTC CGC TAC CTA AGA GG-3Ј; reverse, 5Ј-CTG TCA TCT GAC CAG CTC TC-3Ј). All values were normalized to values from both input DNA and immunoprecipitations with IgG using the ⌬⌬Ct method. Briefly, for every promoter studied, a ⌬Ct value was calculated for each sample by subtracting the Ct value for the input DNA from the Ct value obtained for the immunoprecipitated sample. A ⌬⌬Ct value was then calculated by subtracting the ⌬Ct value for the sample immunoprecipitated with the specific antibody (PPAR␣ or COUP-TFII) from the ⌬Ct value for the corresponding sample immunoprecipitated with normal rabbit serum (IgG). Fold differences (factor-specific ChIP relative to control IgG ChIP) were then determined by raising 2 to the ⌬⌬Ct power.
RNA Interference-Knockdown of PPAR␣ in FAO cells was achieved by using the Smartpool siRNA (Dharmacon) specific for rat PPAR␣. The sequence for siRNA directed against rat PGC-1␤ was selected as described previously (41). The rat PGC-1␤ siRNA (sense sequence, 5Ј-GAT ATC CTC TGT GAT GTT A-3Ј) was synthesized by Dharmacon as a 21-nucleotide duplex, using option A4, with 3Ј-dinucleotide (TT) overhangs. The species-specific siCONTROL nontargeting siRNA number 1 (Dharmacon) was utilized as a negative control sequence to monitor nonspecific targeting. Cells were plated in 12-well plates at a concentration of 5 ϫ 10 4 /well 24 h prior to the experiment. Prior to transfection, siRNAs were resuspended in 1ϫ siRNA buffer (Dharmacon) to a concentration of 20 M. All siRNAs were transfected into FAO cells using the Dharma-FECT TM 4 transfection reagent (Dharmacon) according to the manufacturer's instructions. FAO cells were transfected with the indicated siRNAs for 48 -72 h at working concentrations of 100 nM as indicated in the figure legends. For the knockdown of PGC-1␤, cells were transfected with the indicated siRNAs for 72 h; cells were exposed to the agonists WY-14,643 (10 M) and 9-cis-RA (1 M) 24 h post-transfection and treated for 48 h prior to harvesting. RNA isolation, cDNA synthesis, and real time PCR expression analyses were performed as described above.
Animal Studies-Male PPAR␣ Ϫ/Ϫ mice and age-matched WT littermates (on SV/129 background) were fed a standard chow diet, supplemented with either the PPAR␣ agonist GW-7647 (2.5 mg/kg/day) or equivalent amount of solvent vehicle (Me 2 SO), for 7 weeks. All animals had ad libitum access to water. Mice were weighed every 2 weeks, and drug intake was adjusted according to mean weight. Upon end of treatment the animals were sacrificed; the liver was isolated, and total RNA was extracted using Versagene RNA tissue kit (Gentra Systems, Inc.). First-strand cDNA synthesis and real time PCR expression analyses were performed as described above.
For MTP inhibitor (8aR) studies, male L-FABP Ϫ/Ϫ mice and age-matched WT controls (C57BL/6) were fed a standard chow diet and then administered the 8aR compound orally each day for 7 days at a dose of 50 mg/kg body weight. At the end of treatment the animals were sacrificed the plasma and livers were isolated, and the various hepatic and plasma lipid levels were determined by using commercially available kits as described previously (9).
Lipid Extraction and Analysis-Livers were homogenized in phosphate-buffered saline, and protein concentration was determined. 300 l of homogenate was extracted with 5 ml of chloroform/methanol (2:1) and 0.5 ml of 0.1% sulfuric acid. An aliquot of the organic phase was collected, dried under nitrogen, and resuspended in 2% Triton X-100. Hepatic free fatty acids, triglyceride, and cholesterol content were determined using commercially available kits as described previously (9). Data were normalized for differences in protein concentration.
Adenovirus Experiment-Adenoviral vectors expressing either PGC-1␤ (Ad-PGC-1␤) or GFP (Ad-GFP) were generous gifts from Dr. B. Spiegelman. L35 cells were infected with either the Ad-PGC-1␤ or Ad-GFP for 2 h in serum-free media and then treated for 48 h with the complete media with the agonists WY-14,643 (10 M) and 9-cis-RA (1 M). Cells were infected ϳ75-85% as determined by GFP expression. RNA isolation, cDNA synthesis, and real time PCR expression analyses were performed as described above.
Immunoprecipitation of Secreted ApoB-Secreted apoB was immunoprecipitated as described previously (23,36). Briefly, L35 cells were cultured in 60-mm dishes in the absence or presence of WY-14,643 (10 M) and 9-cis-RA (1 M) for 72 h. Cells were then switched to methionine-free Dulbecco's modified Eagle's medium for 2 h and then labeled with 3 ml of [ 35 S]methionine (100 Ci/ml) in Dulbecco's modified Eagle's medium for 24 h. Media were collected and incubated with polyclonal anti-apoB antibody overnight at 4°C. Protein A-Sepharose was then added, and the mixture was further incubated for 2 h at 4°C. The immunoprecipitate complex was washed three times with TETN buffer (25 mM Tris, pH 7.5, 5 mM EDTA, 250 mM NaCl, and 1% Triton X-100) and once with phosphate-buffered saline. The pellet was resuspended in SDS-PAGE loading buffer, boiled for 5 min, and resolved on a 4 -12% Tris-glycine gel by electrophoresis. Radioactive proteins were detected by autoradiography. The locations of apoB48 and apoB100 was determined by molecular weight markers and human low density lipoprotein standards.

RESULTS
Transcriptional Activities of L-FABP and MTP Promoter-Reporter Constructs Reflect Cell Type-specific Differences in mRNA Expression-A DR1 element located within the proximal MTP promoter region was shown to be responsible for the lack of expression in L35 cells and the high level expression exhibited by FAO cells (37). Occupation of this DR1 element by COUP-TFII caused transcription repression (37). Because the L-FABP promoter contains a similar DR1 element (Fig. 1A), we examined if its expression would vary in parallel with expression of MTP. Indeed, the cell type-specific differences in L-FABP mRNA displayed by L35 and FAO cells were similar to those of MTP mRNA (Fig. 1B), suggesting that the transcription of both genes may be coordinately regulated.
Luciferase reporter constructs driven by either the L-FABPor MTP-proximal promoter regions displayed similar cell typespecific differences; promoter activities were higher (8-fold) in FAO cells compared with L35 cells (Fig. 1C). Mutation of the DR1 sites decreased L-FABP and MTP promoter activities in FAO cells to levels similar to those of L35 cells (Fig. 1B). These findings show that the proximal DR1 elements of both L-FABP and MTP genes are sufficient to confer the cell type-specific differences in mRNA expression displayed by the L35 and FAO cells.
Binding of COUP-TFII to the Proximal DR1 Site Mediates Transcriptional Repression of the L-FABP Gene-Nuclear extracts from L35 and FAO hepatoma cells formed distinct DNA-protein complexes with the oligonucleotide probe containing the DR1 element of the L-FABP promoter (Fig. 2A). The DNA-protein complex formed using nuclear extracts from L35 cells exhibited a supershift with an antibody specific for COUP-TFII ( Fig. 2A). In contrast, the DNA-protein complex formed using nuclear extracts from FAO cells did not display a supershift with the COUP-TFII antibody ( Fig. 2A). Because mutation of the DR1 site blocked the formation of the cell type-specific DNA-protein complexes (37), they require an intact DR1 site.
Ectopic expression of COUP-TFII in FAO cells resulted in a dose-dependent decrease in L-FABP promoter activity (Fig.  2B). Because increased COUP-TFII expression did not alter the activity of the mutated L-FABP promoter (Fig. 2B), the repression of the L-FABP promoter by COUP-TFII is dependent on a functional DR1 element. The maximal reduction of L-FABP promoter activity exhibited by FAO cells transfected with the COUP-TFII plasmid was only ϳ50 -70% of the low level exhibited by L35 cells, suggesting that additional factors are likely to contribute to the cell type-specific differences in expression. These findings, similar to those obtained using the MTP DR1 element (37), support the conclusion that in L35 cells occupancy of the DR1 element by COUP-TFII is responsible for transcriptional inactivation of both genes ( Fig. 2A).
PPAR␣-RXR␣ Heterodimers Compete with COUP-TFII for Binding to the DR1 Promoter Elements of Both the L-FABP and MTP Genes-EMSA supershift analysis of the MTP-DR1 site revealed an FAO cell-specific complex containing RXR␣ (37). Because PPAR␣ and RXR␣ agonists have been shown to activate the transcription of L-FABP via the DR1 site in the proximal promoter region (43), we assessed whether the MTP gene was regulated in a similar manner. EMSA supershift analyses using nuclear extracts from FAO cells demonstrated that DNA probes containing either the L-FABP-or MTP-DR1 sites formed similar FAO-specific complexes (Fig. 3A, 1st lane), which did not supershift with a COUP-TFII-specific antiserum (Fig. 3A, 2nd lane), but did supershift with an antiserum recog-nizing either RXR␣ (Fig. 3, 3rd lane) or PPAR␣ (Fig. 3, 4th lane).
In contrast, PPAR␣-RXR␣ supershifts were not detected with the nuclear extracts obtained from L35 cells (see Ref. 37 and data not shown). These findings indicate that in FAO cells, the DR1 sites of both the L-FABP and MTP promoters are occupied by PPAR␣-RXR␣, whereas in L35 cells these sites are occupied by COUP-TFII.
ChIP analyses of both cell types showed that chromatin from L35 cells immunoprecipitated by a COUP-TFII-specific antiserum was enriched with the DR1 elements of both L-FABP and MTP genes by 4-fold compared with chromatin from FAO cells (Fig. 3B). In contrast, using a PPAR␣-specific antiserum, DR1 element-specific chromatin immunoprecipitated from FAO cells was enriched 4-fold compared with chromatin from L35 cells (Fig. 3B). Because immunoprecipitated chromatin obtained using both antisera contained similar enrichment of DNA sequences corresponding to regions distal to the DR1 regions from both cell lines (Fig. 3B), the ϳ4-fold enrichment of DR1 elements from both the L-FABP and MTP genes reflect cell type-specific differences in binding of PPAR␣-RXR␣ (FAO

cells) or COUP-TFII (L35 cells). The concordant findings obtained from both EMSA supershift analyses (Figs. 2A and 3A)
and ChIP analyses (Fig. 3B) suggest that the cellular content of activator (PPAR␣ and RXR␣) relative to repressor (COUP-TFII) in each cell type would determine which complex occupies the DR1 site and controls transcriptional activity. This interpretation is supported by the findings showing that both cell lines contained relative levels of COUP-TFII mRNA that were inversely related to the amounts of both PPAR␣ and RXR␣ (Fig. 3C). (The relative differences in protein levels for each of the nuclear receptors correlates with those observed for their respective mRNAs (see Ref. 37 and data not shown).

PPAR␣ and RXR␣ Agonist Treatment of L35 Cells Results in the Transcriptional Induction of L-FABP and MTP
Expression and a Restored Ability to Secrete ApoB-Treatment of L35 cells with either a PPAR␣ or RXR␣ agonist increased the expression of both L-FABP (ϳ60 -75-fold) and MTP (ϳ55-65-fold) mRNAs (Fig. 4A). Treatment with both agonists synergistically increased L-FABP and MTP mRNA expression by ϳ300-fold (Fig. 4A). L-FABP and MTP promoter luciferase reporters exhibited similar responses to the PPAR␣ and RXR␣ agonists (Fig. 4B). Mutation of the DR1 sites blocked the activation of both the L-FABP and MTP by PPAR␣ and RXR␣ agonists (Fig. 4B). Thus, PPAR␣ and RXR␣ agonists coordinately activate L-FABP and MTP gene transcription via their respective DR1 elements.
It should be noted that the vehicle Me 2 SO alone caused a modest but significant increase in both the expression of L-FABP and MTP mRNA and their respective promoter activities (Fig. 4, A and B,  respectively). Clearly, PPAR␣ and RXR␣ agonists added to cells using Me 2 SO as a vehicle exhibited a much greater induction of L-FABP and MTP compared with Me 2 SO alone (Fig. 4, A and B). The induction of PPAR␣-activated genes by Me 2 SO has been described (44).
The inability of L35 cells to assemble and secrete apoB-containing lipoproteins is strictly caused by a lack of MTP; transfection with a constitutively expression MTP transgene allows L35 cells to secrete apoB-containing lipoproteins (36,37). We now show that treating L35 cells with PPAR␣-RXR␣ agonists markedly enhanced the secretion of de novo synthesized 35 S-labeled apoB (Fig. 4C). Thus, PPAR␣-RXR␣ agonists restored expression of L-FABP and MTP allowed L35 cells to assemble and secrete apoB-containing lipoproteins.

Concordant Changes in Cellular Expression of PPAR␣-RXR␣ and COUP-TFII Orchestrate Agonist-mediated Changes in
Transcription by Determining Which Complex Occupies the DR1 Element-PPAR␣-RXR␣ agonists altered the levels of nuclear receptors in L35 cells so that they resembled the levels in FAO cells (Fig. 5A). Upon treatment of L35 cells with PPAR␣-RXR␣ agonists mRNA expression of both PPAR␣ (5-fold) and RXR␣ (2.3-fold) increased, whereas COUP-TFII expression decreased (Ϫ60%) (Fig. 5A). Me 2 SO alone also reduced COUP-TFII mRNA expression by L35 cells, without affecting expression of either PPAR␣ or RXR␣ (Fig. 5A). The Me 2 SO-mediated reduction in COUP-TFII may explain why its addition to L35 cells increased transcription and expression of L-FABP and MTP (Fig. 4). The agonist-mediated changes in nuclear receptor protein levels parallel those seen for the mRNAs (data not shown).
The effect of agonist-mediated changes in the relative expression of the DR1-associated factors, on the occupancy of the DR1 sites, was determined by ChIP analyses. PPAR␣-RXR␣ agonist treatment of L35 cells decreased (ϳ3-fold) the amount of DR1 sequences in chromatin immunoprecipitated with the COUP-TFII-specific antiserum (Fig. 5B), although it led to an enrichment of (ϳ3-fold) the DR1 sequences immunoprecipitated with a PPAR␣-specific antiserum (Fig. 5B). Thus, the ChIP analyses demonstrated that changes in the relative expression of the DR1-associated factors (Fig. 5A) resulted in concordant changes in the occupancy of each DR1 site (Fig. 5B).

PPAR␣ Is Necessary for High Expression Levels of L-FABP and MTP in Hepatoma Cells and in Vivo-
The combined data suggest that cellular levels of transcriptional activators (PPAR␣-RXR␣) relative to the transcriptional repressor (COUP-TFII) dictate which complex occupies the DR1 site and controls the coordinate transcriptional activity of L-FABP and MTP genes. PPAR␣-RXR␣ agonist treatment induces L-FABP and MTP transcription by both increasing the cellular content of PPAR␣-RXR␣/COUP-TFII and by ligand binding. To examine if changes in cellular expression of PPAR␣-RXR␣/COUP-TFII might change transcription independent of PPAR␣-RXR␣ agonists, we utilized RNA interference to knock down the expression levels of PPAR␣ in FAO cells. FAO cells transfected with PPAR␣-specific siRNAs demonstrated a 75% reduction PPAR␣ mRNA compared with the cells transfected with control siRNA (Fig. 6A). This decrease in PPAR␣ mRNA was associated with a reduced cellular content of both L-FABP and MTP mRNAs to nearly 50% of the control (Fig. 6A). The PPAR␣-specific RNA interference did not alter the mRNA levels of PPAR␣-independent apoB, suggesting that the decreases in L-FABP and MTP were PPAR␣-reduction specific (Fig. 6A).
We also examined whether a PPAR␣ agonist would coordinately induce the hepatic expressions of L-FABP and MTP mRNAs in vivo and, if so, whether the agonist-mediated increases were PPAR␣dependent. Control SV/129 and PPAR␣ Ϫ/Ϫ mice were treated with the PPAR␣ agonist GW-7647 (45). Although control mice displayed increased levels of both L-FABP and MTP mRNAs following treatment with the PPAR␣ agonist (Fig. 6B), mice lacking functional PPAR␣ displayed no significant changes in either L-FABP or MTP expression levels (Fig. 6B). These findings obtained in vivo corroborate those obtained using FAO and L35 hepatoma cells indicating that ligand activation of PPAR␣ coordinately induces both L-FABP and MTP genes.
PGC-1␤ Acts in Concert with PPAR␣ to Coordinately Induce the L-FABP and MTP Genes-The observations that L35 cells express relatively high levels of PGC-1␣ and nearly undetectable levels of PGC-1␤, whereas FAO cells express high levels of PGC-1␤ and nearly undetectable levels of PGC-1␣ (Fig. 7A), are consistent with the proposal that PGC-1␤ activates MTP gene expression (41). This proposal is further supported by our findings showing both L35 cells and mice treated with PPAR␣ agonists caused an ϳ3-fold induction of PGC-1␤ mRNA (Fig. 7, B and C, respectively). To examine if PGC-1␤ is essential for PPAR␣-mediated induction of L-FABP and MTP, FAO cells were treated with PPAR␣-RXR␣ agonists, and the effect of siRNA knockdown of PGC-1␤ was determined. PPAR␣-RXR␣ agonists-treated FAO cells given the siRNA specific for PGC-1␤ demonstrated a 65% reduction in PGC-1␤ mRNA levels, which was associated with coordinate decreases in both L-FABP (Ϫ38%) and MTP (Ϫ48%) mRNA levels, whereas PGC-1␣ mRNA levels remained unchanged (Fig. 7D). PPAR␣-RXR␣ agonists-treated FAO cells given a negative control siRNA exhibited no change in any of these mRNA levels (Fig.  7D). Thus, the siRNA demonstrated target specificity and the associated reductions in L-FABP and MTP mRNA expressions were because of the reduction in PGC-1␤ content.
We further examined the role of PGC-1␤ in L-FABP and MTP gene transcription by enhancing the expression of PGC-1␤ in untreated and PPAR␣-RXR␣ agonist-treated L35 cells via transduction with an adenovirus expressing PGC-1␤ (41,46). Whereas PGC-1␤ adenovirus infection did not alter L-FABP or MTP mRNA levels in untreated L35 cells (Fig. 8), L35 cells treated with the PPAR␣-RXR␣ agonists demonstrated PGC-1␤ adenovirus-mediated increases (3-fold) in both L-FABP and MTP mRNAs relative to agonist-treated uninfected and GFP-infected controls (Fig. 8). These data show that PGC-1␤ plays an essential role in the PPAR␣-mediated induction of L-FABP and MTP gene transcription.
Coordinate Inactivation of L-FABP and MTP Prevents Hepatic Steatosis in Vivo-Chemical inhibition of MTP, a strategy developed to ameliorate hyperlipidemia (47)(48)(49)(50), is of limited use because of the development of hepatic steatosis (35,51). Although L35 cells do not express MTP or secrete apoBcontaining lipoproteins, they do not accumulate triglycerides even when challenged with fatty acids (e.g. oleic acid) (36,37). In other experiments we have shown that blocking hepatic VLDL secretion by inhibiting MTP was associated with the accumulation of hepatic triglycerides (35). Because L-FABP facilitates hepatic fatty acid uptake and formation of triglycerides (9), we hypothesize that loss of MTP by L35 cells did not cause the accumulation of triglycerides because they also lack L-FABP. To examine whether coordinate repression of L-FABP and MTP would block hepatic VLDL assembly/secretion without causing hepatic steatosis, control L-FABP null mice (L-FABP Ϫ/Ϫ ) were treated with an MTP inhibitor (8aR) (35).
After 7 days of 8aR treatment, plasma levels of triglyceride and cholesterol were markedly reduced to similar levels in both groups of mice (Fig. 9A). Although treating control C57BL/6 mice with the MTP inhibitor resulted in a 4-fold increase in hepatic triglyceride levels, hepatic triglyceride accumulation was completely prevented by ablation of L-FABP (Fig. 9B). These data show that coordinate transcription regulation of L-FABP and MTP genes allows variation in VLDL assembly/secretion in the absence of hepatic steatosis.

DISCUSSION
Our findings show that the transcription of L-FABP and MTP is regulated by competitive binding to similar DR1 elements by either the fatty acid ligand-activated transcription factors (PPAR␣-RXR␣) or COUP-TFII. Thus, expression of two lipid transfer proteins (L-FABP and MTP), which function in concert with each other, can be coordinately regulated in response to the availability of fatty acid substrate. There are many functionally distinct pathways competing for the utilization of fatty acids by the liver. These include cellular uptake, esterification in the production of other lipids (e.g. glycerolipids and cholesterol esters), ␤-oxidation, storage, and export mainly in the form of VLDL lipids. The delivery of fatty acids into one or more of these pathways is a dynamic process. Changes in fatty acid flux through these pathways must occur rapidly and selectively. Substrate-driven "feed-forward" transcriptional regulation is a common mechanism allowing changes in gene expression to occur concomitantly with variations in metabolic needs (52). Because fatty acids also can activate PPAR␣-dependent gene transcription of L-FABP (53) and MTP (54), fatty acid flux to the liver both induces the enzymes controlling VLDL assembly/secretion as well as providing lipogenic substrate.
Several lines of evidence indicate that similar DR1 elements present in the L-FABP and MTP promoters provide coordinate transcriptional regulation necessary for the interdependent role of these two lipid transfer proteins in delivering fatty acids to the VLDL assembly/secretion pathway. Mutational deletion of the DR1 element, in both the L-FABP and MTP promoter reporter constructs, caused the following: 1) a reduction of the relatively high activity levels exhibited by FAO cells to the lower levels exhibited by L35 cells (Fig. 1C), and 2) abrogation of both the ability of COUP-TFII (Fig. 2B) and PPAR␣-RXR␣ agonist (Fig. 4A) to repress or activate transcription, respectively. Fur- ther evidence showing that these DR1 elements are the functionally relevant cognate binding sites responsible for competitive occupation by COUP-TFII (repressor) or PPAR␣-RXR␣ (activator) are provided by EMSA supershift (Fig. 2, L-FABP), MTP (37), and ChIP (Fig. 3) analyses. The combined analyses obtained from the complementary EMSA supershift and ChIP experiments of the DR1 elements indicate that occupation by PPAR␣-RXR␣ is associated with transcriptional activation of both genes (FAO cells), whereas occupation by COUP-TFII is associated with repression (L35 cells).
The principal determinant responsible for the phenotypic difference in the expression of L-FABP and MTP exhibited by FAO and L35 cells is the relative cellular content of COUP-TFII (repressor) to PPAR␣-RXR␣ (activator) (Fig. 3C). Three independent experiments demonstrate that the plasticity in the cellular phenotype of FAO and L35 cells is dependent upon the cellular content of COUP-TFII (repressor) relative to PPAR␣-RXR␣ (activator) as follows. 1) Treating L35 cells with PPAR␣-RXR␣ agonists increases the expression of PPAR␣-RXR␣ (activator), while decreasing COUP-TFII (repressor) expression (Fig. 5A). 2) These changes in cellular content of PPAR␣-RXR␣/COUP-TFII are reflected by similar changes in the occupancy of the DR1 elements present in both the L-FABP and MTP promoters (Fig. 5B). 3) These changes resulted in DR1 site-dependent increases in the transcriptional activities of both L-FABP and MTP promoter-luciferase reporter constructs (Fig. 4B). 4) They enhanced expression of L-FABP and MTP mRNAs (Fig. 4A). Furthermore, the agonist-mediated changes in L35 cells were associated with a restored ability to assemble and secrete apoB-containing lipoproteins (Fig. 4C). PPAR␣ and RXR␣ agonists have been shown to activate the transcription of L-FABP in hepatoma cells, an effect dependent on the DR1 site in the proximal promoter region (43). Treating wild type but not PPAR␣ knock-out mice with a PPAR␣ agonist increased hepatic expression of MTP (54). These and additional findings obtained from studies examining transcriptional regulation of genes involved in fatty acid metabolism (55-58) support the proposal that PPAR␣-RXR␣ and COUP-TFII compete with each other for binding to DR1 promoter elements. In the context of these genes, occupation of these elements by PPAR␣-RXR␣ is associated with activation of transcription, whereas occupation by COUP-TFII is associated with transcriptional repression. Thus, this is a common regulatory paradigm for substrate-driven modulation of various fatty acid utilization pathways at the level of transcription.  Our findings show PGC-1␤ is required for PPAR␣ agonistmediated induction of L-FABP, and MTP expression suggests that PGC-1␤ is required for PPAR␣ agonist induction of L-FABP and MTP (Fig. 7D). We also found that although PPAR␣ agonist treatment of PPAR␣ Ϫ/Ϫ mice failed to induce L-FABP and MTP mRNA expression (Fig. 4), hepatic expression of PGC-1␤ mRNA was increased 2-fold (data not shown). These data suggest that in the absence of PPAR␣, PGC-1␤ is not sufficient to increase the expression of L-FABP and MTP. Additional experiments using adenovirus-mediated expression of PGC-1␤ indicate that PPAR␣-RXR␣ is necessary in order to enhance the transcription of L-FABP and MTP (Fig. 8). Our combined findings support the proposal that PGC-1␤ participates in the transcriptional activation of MTP and subsequent induction of apoB-dependent VLDL secretion (41,59). In ob/ob diabetic mice, adenovirus-mediated expression of PGC-1␤ and Foxa2 induced MTP expression, suggesting a mechanism through which insulin blocks VLDL assembly/secretion (59). It has been demonstrated that Foxa2 is completely excluded from the nucleus of ob/ob mice (60), whereas both MTP expression and VLDL assembly/secretion are increased (61). Furthermore, PGC-1␤-mediated increase in MTP expression was retained in the ob/ob mice indicating that induction of MTP transcription can occur via Foxa2-independent mechanisms (59). In fat-fed hyperlipidemic mice, PGC-1␤ activation of sterol regulatory element-binding protein and liver X receptor ␣ is associated with enhanced MTP expression and VLDL assembly/secretion (41). PGC-1␤ has been demonstrated to interact with PPAR␣ (62) and activate the transcription of PPAR␣ target genes (63). Our findings show that in the context of rat hepatoma cells, PGC-1␤ activates PPAR␣-RXR␣ (Fig. 8).
Our combined data suggest that retention of similar DR1 elements in the promoters of L-FABP and MTP ensures that their expression will be sufficiently induced to provide an efficient delivery of fatty acids to the VLDL assembly/secretion pathway. The same coordinate transcriptional regulation can attenuate the expression of both genes. The importance of coordinate decreased expression of L-FABP and MTP was clearly shown by additional studies showing that ablation of L-FABP blocked the accumulation of triglycerides in the livers of mice treated with the MTP inhibitor (Fig. 9).
Our findings have important implications regarding the efficacy of MTP inhibitors to ameliorate hyperlipidemia (47)(48)(49)(50). In a recent study, treatment of hyperlipidemic apoE knock-out mice with an MTP inhibitor caused a marked reduction in both plasma lipids and the progression of atherosclerosis (64). In other studies, liver-specific MTP gene ablation prevented the development of hyperlipidemia and atherosclerosis in mice made susceptible due in part to hepatic overproduction of apoB (65). The utility of MTP inhibitors has been greatly diminished because of the associated development of fatty liver (35,51). Our studies show that ablation of L-FABP completely blocks the accumulation of triglycerides in the liver of mice treated with the MTP inhibitor 8aR (Fig. 9B). Thus, our findings indicate that blocking the function of both L-FABP and MTP would reduce hyperlipidemia without causing the development of fatty liver.