Transcriptional regulation of the human hepatic lipase (LIPC) gene promoter Published, JLR Papers in Press, April 7, 2006.

Hepatic lipase (HL) plays a key role in the metabolism of plasma lipoproteins, and its level of activity requires tight regulation, given the association of both low and high levels with atherosclerosis and coronary artery disease. However, little is known about the factors responsible for HL expression. Here, we report that the human hepatic lipase gene (LIPC) promoter is regulated by hepatocyte nuclear factor 4α (HNF4α), peroxisome proliferator-activated receptor γ coactivator-1α (PGC-1α), apolipoprotein A-I regulatory protein-1 (ARP-1), and hepatocyte nuclear factor 1α (HNF1α). Reporter analysis showed that HNF4α directly regulates the LIPC promoter via two newly identified direct repeat elements, DR1 and DR4. PGC-1α is capable of stimulating the HNF4α-dependent transactivation of the LIPC promoter. ARP-1 displaces HNF4α from the DR1 site and blocks its ability to activate the LIPC promoter. Induction by HNF1α requires the HNF1 binding site and upon cotransfection with HNF4α leads to an additive effect. In addition, the in vivo relevance of HNF4α in LIPC expression is shown by the ability of the HNF4α antagonist Medica 16 to repress endogenous LIPC mRNA expression. Furthermore, disruption of Hnf4α in mice prevents the expression of HL mRNA in liver. The overall effect these transcription factors have on HL expression will ultimately depend on the interplay between these various factors and their relative intracellular concentrations.

Hepatic lipase (HL) is a 477 amino acid glycoprotein that plays an important role in lipoprotein metabolism. The majority of HL is synthesized and secreted by the liver, where it has been shown to act as both a lipase and a ligand. As a lipase, it catalyzes the hydrolysis of triglycerides and phospholipids of intermediate density lipoprotein remnants, large buoyant LDLs, and HDLs to form smaller, denser lipoprotein particles (1,2). Studies in humans have shown an association between high HL activity and increased plasma concentrations of small, dense LDL and HDL particles, one of the major risk factors for coronary artery disease (3)(4)(5). As a ligand, HL contributes to the process of reverse cholesterol transport by participating with surface proteoglycans and the low density lipoprotein receptor like-protein in promoting hepatic uptake of lipoproteins, including remnant LDL and HDL particles (6)(7)(8), thus mediating the hepatic uptake of HDL-cholesteryl esters (9). The observed association of low HL activity with coronary artery disease might be attributable to decreased HL-enhanced remnant uptake by the liver (10). Together, these data show that HL is an important enzyme in lipid metabolism that must be highly regulated, because both low and high levels of HL activity appear associated with dyslipidemia.
The expression level of hepatic lipase activity varies widely in normal individuals (5-to 8-fold) and is influenced by genetic variation, obesity, gender, intracellular cholesterol, and lipid-lowering therapy (11). However, the signal transduction pathways and transcription factors that mediate HL regulation by these factors have not been elucidated. To date, only a few transcription factor binding sites have been identified in the hepatic lipase gene (LIPC) proximal promoter (Fig. 1A). The positive regulators include the upstream stimulatory factor (USF; 2527 to 2502) (12) and the hepatocyte nuclear factor 1a (HNF1a; 265 to 253) (13). The negative regulators include the estrogen receptor a (21,557 to 21,175) (14), the activator protein-1 (2564 to 2558) (15), and the farnesoid X receptor (FXR). The exact binding site for FXR has not been elucidated, but it is located between 2698 and 2541 (16). USF, estrogen receptor a, and FXR have been confirmed functionally by transient transfection assays, whereas HNF1a and activator protein-1 are only predicted based on sequence, binding assays, and/or placement within DNase I footprints from hepatic nuclear extracts.
Upon further sequence examination of the proximal LIPC promoter, we identified two additional potential transcription factor binding sites between nucleotides 2302 and 2226 consisting of tandem inverted direct repeats (DRs) that conform to the consensus DNA response element half site 59-RG(G/T)TCA-39 (Fig. 1B). The functional importance of these DRs is suggested by the region's 78% sequence conservation between rat and human (17). The first element, located between 2226 and 2238, is a DR1 (DRs separated by 1 bp; Fig. 1B) and lies within DNase I footprints previously identified with hepatic nuclear extracts (13,15). The second site is a DR4 (DRs separated by 4 bp; Fig. 1B) located between 2287 and 2302. DR1 and DR4 DNA response elements are known to bind various members of the nuclear receptor family as either homodimers or heterodimers. Possible DR1 and DR4 binding factors include retinoid X receptor a (RXRa), apolipoprotein A-I regulatory protein-1 (ARP-1), also know as the chicken ovalbumin upstream promoter transcription factor II (COUP-TFII), hepatocyte nuclear factor 4a (HNF4a), retinoic acid receptor a (RARa), peroxisome proliferatoractivated receptor, liver X receptor, thyroid hormone receptor, and the constitutive androstane receptor (18)(19)(20)(21).
Among the transcription factors mentioned above, RXRa, RARa, HNF1a, HNF4a, and ARP-1 are of particular interest because they have been shown to act together to regulate the expression of a number of genes involved in lipid metabolism, including the apolipoprotein gene promoters APOA1/C3/A4 gene cluster, APOA2, and APOB (22)(23)(24)(25)(26). On a DR1 element, RXRa homodimers act as positive regulators in the presence of ligand (9-cis-retinoic acid), whereas RXRa-RARa heterodimers repress RXRa homodimer activation, in a ligand-dependent manner, by competition for binding to the DR1 element (27). On DR4 elements, RXRa binds as a heterodimer with various partners and acts as a transcriptional regulator (18). HNF4a is a liver-enriched transcription factor essential for hepatocyte differentiation and liver function (28,29). Its disruption leads to embryonic lethality (30). Importantly, HNF4a expression has been shown to correlate with HL expression in several hepatic cells lines (17). HNF4a is a constitutively activated transcription factor whose ligand has been identified as tightly bound endogenous fatty acids (31)(32)(33). Recent analysis has suggested that binding of coactivators rather than ligand binding locks HNF4a into an active conformation (34).
HNF1a is a homeodomain-containing transcription factor important for diverse metabolic functions in pancreatic islets, liver, intestine, and kidney (52,53). It is enriched in the liver, where it transcriptionally modulates numerous liver-specific genes (54). That HNF1a plays a role in HL expression is suggested by the reduction of HL expression in HNF1a-deficient mice (53) and the association of mutations in HNF1a with variations in plasma lipoprotein metabolism (55). However, whether HNF1a plays a direct or indirect role in HL expression has not been elucidated. Lockwood and Frayling (56) used expression data from HNF1a-deficient mice to write software that searched for potential HNF1a binding sites 2 kb upstream of 28 genes downregulated in HNF1a-deficient mice compared with wild-type (WT) mice. This analysis identified LIPC as one of eight genes likely to be directly regulated by HNF1a. Similar to ARP-1, HNF1a can both negatively (57) and positively (58,59) affect transcriptional regulation by HNF4a.
In this study, we investigated the potential regulation of the LIPC proximal promoter by these factors. We demonstrate that the LIPC proximal promoter is regulated by HNF4a, PGC-1a, HNF1a, and ARP-1 via the DR1, DR4, and HNF1 binding sites but is not regulated by RARa and RXRa. In addition, we demonstrate the importance of HNF4a for endogenous hepatic lipase expression in vivo.

Plasmids and cloning
Plasmid constructs containing the human LIPC proximal promoter segment from 2851 to +29 were prepared by PCR amplification using total human genomic DNA as a template and the following primers: forward, 59tcatgagctcCCAACTCCAGAAG-GTAAGAACC39; reverse, 59tagtaagcttCGGGGTCCAGGCTTTC-TTGG39. The lowercase letters indicate non-LIPC sequences, and the underlined sequences are SstI and HindIII cleavage sites, respectively. The restriction sites were used for directional cloning of the LIPC proximal promoter initially into the pXP1 vector. The pGL4-851 (WT) construct was made by excising the 2851 to 129 LIPC proximal promoter region from pXP1 with XhoI and HindIII and cloning it into the pGL4 luciferase vector from Promega. Dr. Vassilis I. Zannis (Boston University Medical Center) kindly provided the expression vectors pMT2-ARP-1 and pMT2-HNF4a and the control empty vector pMT2, and Dr. Anastasia Kralli (Scripps Institute) provided the pcDNA3/HA-hPGC-1a expression vector (60). The pcDNA3.1-HNF1a expression plasmid was constructed by cloning the HNF1a cDNA, prepared by PCR from cDNA made from HuH7 mRNA, into the pcDNA3.1/V5-His TOPO TA expression vector (Invitrogen). The expressed HNF1a protein does not contain the V5 epitope or the His tag, because the natural stop codon for HNF1a is present in the cDNA clone. The sequences of all constructed plasmids were confirmed by sequencing in both orientations.
Site-directed mutagenesis of the LIPC WT promoter construct All mutants were generated by site-directed mutagenesis using the Quikchange site-directed mutagenesis kit (Stratagene) and the following primers (boldface letters indicate transcription factor binding regions, and lowercase letters show mutated bases): DR1 mutant, 59GCTTTAAGTTGATTAATTTGttcga-CTttCgaTGGCCCCAAAAGG39 and 59CCTTTTGGGGCCA-tcGaaAGtcgaaCAAATTAATCAACTTAAAGC39; DR4 mutant, 59GCAGCCACGTGGAAGCagtCgACCCCtAtggTTGGCAGAAT-TTCC39 and 59CGTCGGTGCACCTTCGtcaGcTGGGGaTacc-AACCGTCTTAAAGG39; HNF1 mutant, 59GTGGATAACATGTT-GAGAGGTTAAgccggAATGGGCAGTCTTCCCTAACAAAGT39 and 59ACTTTGTTAGGGAAGACTGCCCATTccggcTTAAC-CTCTCAACATGTTATCCAC39. Site-directed mutagenesis was performed according to the manufacturer's protocol. The sequences of all mutant promoter constructs were confirmed by sequencing in both orientations.

Cell culture and transient transfection assays
The human hepatoma HuH7 and African green monkey kidney COS7 cell lines were maintained in DMEM (Gibco) supplemented with 10% fetal bovine serum (Hyclone Laboratories, Inc.).
For transient transfection assays, HuH7 cells were seeded at 1 3 10 5 cells/well in 24-well plates 24 h before transfection. Transfections in HuH7 cells were carried out using the Lipofectamine 2000 (Invitrogen) reagent in DMEM without fetal bovine serum at a 1:2.5 mg DNA/ml Lipofectamine ratio. A total of 600-850 ng of plasmid DNA was used per well, consisting of the appropriate promoter construct plasmid, 10 ng of phRG-TK renilla reporter vector (Promega), used to correct for differences in transfection efficiencies, and the appropriate combination of expression plasmids (see figure legends for the exact amounts used). The total amount of DNA was kept constant by supplementing with the pMT2 empty vector. The DMEM-transfection mixture was replaced 4-6 h after transfection with complete medium (listed above).
Luciferase activity was determined 24 h after transfection using the Dual-Luciferase kit (Promega). Briefly, the cells were lysed directly in the 24-well plate using 120 ml of 13 passive lysis buffer provided with the Dual-Luciferase kit. One hundred microliters of LARII reagent (firefly luciferase substrate) was added to 50 ml of lysis supernatant, and firefly luciferase activity was measured with a Lumat LB 9507 luminometer for 10 s. Then, 100 ml of Stopn-Glo reagent (renilla luciferase substrate) was added to the reaction, and renilla luciferase activity was measured for an additional 10 s. Firefly luciferase activities were normalized by renilla activities to correct for differences in transfection efficiencies. All transfection experiments were repeated at least two times in triplicate. Statistical significance was analyzed by Student's t-test. Firefly luciferase activities normalized by renilla activities are presented as fold induction relative to the normalized firefly luciferase activity in cells transfected with the pMT2 empty vector only, which was taken as 1.0.

Preparation of whole cell and nuclear extracts
Whole cell extracts (WCEs) were prepared using transiently transfected COS7 cells in 10 cm plates using 10 mg of expression vector in the presence of 30 ml of Lipofectamine 2000 (Invitrogen). After 24 h, cells were washed and collected at 200 g in 40 mM Tris-HCl, pH 7.4, 1 mM EDTA, and 0.15 M NaCl (43). Cells were lysed in 20 mM Tris-HCl, pH 7.4, 0.4 mM KCl, 2 mM DTT, and 20% (w/v) glycerol by three rounds of freezing and thawing in liquid nitrogen, essentially as described previously (61). Cell debris was removed by centrifugation at 48C (10,000 g) for 15 min. Phenylmethylsulfonyl fluoride was added to the supernatant (WCE) to a final concentration of 1 mM, and aliquots were frozen at 2708C. Nuclear extracts from HuH7 cells were prepared using the CelLytic NuCLEAR extraction kit (Sigma-Aldrich, Inc.) according to the manufacturer's protocol. Protein concentrations of whole cell and nuclear extracts were determined using the 96-well format of the Coomassie Protein Assay kit (Pierce) and quantified using the SpectraCount plate reader (Packard).

Electrophoretic mobility shift assays
Electrophoretic mobility shift assays (EMSAs) were performed using the Gel Shift Assay system (Promega) at room temperature according to the manufacturer's protocol. Double-stranded synthetic probes (see Fig. 2 for sequences of individual probes) were prepared by annealing equal amounts of complementary synthetic oligonucleotides with single-stranded 59 overhangs (cgcg), as described previously (20). The resulting double-stranded fragments were labeled with ( 32 P)dCTP using the Klenow fragment of DNA polymerase I. MicroSpin G-25 spin columns (Amersham Pharmacia Biotech) were used to remove the unincorporated label. Unlabeled double-stranded oligonucleotides were used as cold competitors (see Fig. 2 for sequences of individual competitors). Specific antisera were added to the DNA binding reaction for "supershifting" of DNA transcription factor complexes. Antibodies against HNF4a (sc-6557X), RARa (sc-551X), and RXRa (sc-774X) were obtained from Santa Cruz Biotechnology, Inc., and anti-Coup-TF, which recognizes both ARP-1/Coup-TFII and Coup-TFI, was kindly provided by Dr. Ming-Jer Tsai (Baylor College of Medicine). One microliter of anti-HNF4a, anti-RARa, and anti-RXRa antibodies and 1 ml of diluted anti-CoupTF (1:10 dilution in 1 mg/ml BSA) was used in supershift assays. Competition and supershift assays were carried out by preincubating cold competitors and antisera with protein extracts for 20 min before adding the binding probe. After addition of the binding probe, incubation was continued for an additional 20 min. Binding reactions were electrophoresed on 4% polyacrylamide gels, dried, and autoradiographed using the Molecular Dynamics Storm 820 PhosphorImager.
The dissociation constant (K d ) values of HNF4a and ARP-1 were calculated from binding reactions performed with a constant amount of WCE from COS7 cells overexpressing either HNF4a or ARP-1 (0.5 or 1 mg, respectively) and increasing concentrations of radiolabeled LIPC DR1 probe (see Fig. 2 for probe sequence), as described previously (43). Binding reactions were incubated at room temperature for 30 min. After gel electrophoresis and PhosphorImager scanning, the bands corresponding to the bound and free oligonucleotides were quantified using ImageQuant 5.2 software (Molecular Dynamics).

Chromatin immunoprecipitation assay
Chromatin immunoprecipitation (ChIP) assays were performed using the EZ ChIP kit from Upstate. The HNF4a antibody (H171X, sc-8987X) and HNF1a antibodies (C19X, sc-6547X or H140, sc10791) were obtained from Santa Cruz Biotechnology, Inc. Briefly, HuH7 cells were seeded in 100 mm plates with 10 ml of medium and grown to 7 3 10 6 cells/plate. The HuH7 cells on each plate were cross-linked with 1% formaldehyde for 10 min at room temperature. Glycine was added to a final concentration of 0.125 M for 5 min at room temperature to quench the reaction. Each 100 mM plate was washed two times with 5 ml of 13 PBS plus protease inhibitors. Cells were then scraped from each plate into a microfuge tube and spun at 700 g at 48C for 4 min. The cell pellet was resuspended in 350 ml of SDS lysis buffer containing protease inhibitors per 7 3 10 6 starting cells. Cell lysates were divided into two tubes and sonicated at 30% amplitude for 10 pulses (10 s pulse followed by a 30 s pause) using a Branson model 102C sonicator. The sonicated cell lysate was spun at 12,000 g at 48C for 10 min, and the supernatant was separated into 100 ml aliquots of z2 3 10 6 cell equivalents each.
One 100 ml aliquot (2 3 10 6 cell equivalents) was used per ChIP assay. Each 100 ml aliquot was diluted with 900 ml of ChIP dilution buffer plus protease inhibitors. To preclear the chromatin, 60 ml of protein A or G agarose (plus sonicated salmon sperm DNA) was added to each sample and incubated for 1 h at Binding of nuclear extract proteins from HuH7 cells to the LIPC DR1 or DR4 WT and mutant sequences, and supershifting electrophoretic mobility shift assays (EMSAs) to identify factors present in the resulting DNA/protein complex. A, C: Autoradiographs of EMSA gels using no nuclear extract (mock) or 1 mg of nuclear extract proteins from HuH7 cells bound to a double-stranded WT or mutant HL DR1 (A) or HL DR4 (C) probe. Nuclear extracts were competed with and without a 250-fold molar excess of cold WT competitor (A or C, lanes 3 and 6) or mutant (mut) competitor (A or C, lanes 4 and 7), as specified above the lanes. B, D: Supershifting EMSA assays to identify factors present in the DNA/protein complex from the nuclear extracts of HuH7 cells. Nuclear proteins were bound to WT or mutant HL probes with and without antibodies against HNF4a, chicken ovalbumin upstream promoter transcription factor (COUP-TF) [apolipoprotein A-I regulatory protein-1 (ARP-1)], retinoic acid receptor a (RARa), and retinoid X receptor a (RXRa), as indicated above the lanes. E: Supershifting EMSA assay to show that antibodies against RXRa and RARa can detect RXRa/RARa binding to the known DR5responsive element of the rat cholesterol 7a-hydroxylase promoter (rCyp7a). Arrows indicate the DNA/protein bands shifted by the protein extracts, brackets denote bands supershifted by the indicated antibodies, and FP designates the free probe. Under the autoradiographs are the sequences of the probes and cold competitors. Boldface letters and arrows indicate the two inverted half-sites. Uppercase letters represent WT sequences, and lowercase letters represent mutated bases.
48C with rotation. The beads were pelleted by centrifugation at 4,000 g for 1 min. The precleared chromatin was immunoprecipitated with no antibody, 10 mg of the appropriate antibody, or corresponding IgG and incubated with rotation at 48C overnight. The protein/antigen/DNA complexes were collected by adsorption to 60 ml of protein A or G agarose (plus sonicated salmon sperm DNA) and incubated for 1 h at 48C with rotation. The agarose beads were pelleted by brief centrifugation at 4,000 g for 1 min, and the supernatant fraction was removed. The beads were washed once with 1 ml of low-salt immune complex wash buffer, once with 1 ml of high-salt immune complex wash buffer, once with 1 ml of LiCl immune complex wash buffer, and twice with 1 ml of Tris-EDTA (TE) buffer. After every wash, the beads were pelleted by brief centrifugation at 4,000 g for 1 min and the supernatant fraction was removed. The protein/DNA complexes were eluted in 200 ml of elution buffer (1% SDS and 0.1 M NaHCO 3 in water) and reverse cross-linked at 658C overnight by the addition of 8 ml of 5 M NaCl, 4 ml of 0.5 M EDTA, 8 ml of 1 M Tris-HCl, and 1 ml of proteinase K. RNA was eliminated by the addition of 1 ml of RNase A and incubation at 378C for 30 min. The DNA was purified using the PCR purification kit from Qiagen and eluted in 50 ml of elution buffer (EB; Qiagen).

Medica 16 treatment, RNA isolation, reverse transcription, and real-time PCR
HuH7 cells were seeded in six-well plates and, after reaching 60-70% confluence, were treated with 100, 250, or 400 mM MEDICA 16 (M16; Cayman Chemical) or the vehicle DMSO for 48 h. Total RNA was extracted 48 h after treatment using the TRIzol reagent (Invitrogen). Five hundred nanograms of each RNA preparation was reverse-transcribed to cDNA using the iScript cDNA synthesis kit from Bio-Rad. The cDNA from each preparation was diluted 1:50, and 2 ml of each dilution was used for real time PCR. Real-time PCR was performed on the 7500 Real-Time PCR System from Applied Biosystems using the Universal PCR Master Mix (Applied Biosystems) and primer/probe mixtures from Applied Biosystems for hGAPDH, hLIPC (hepatic lipase), or hAPOC3. The housekeeping gene GAPDH was used as an internal control for sample normalization. The mRNA values were calculated using a standard curve method and are presented as mRNA levels relative to the mock level, which was set at 1.0. Detection of HL mRNA by RT-PCR in livers of mice with liver-specific targeted disruption of HNF4a RNA from livers of the recently generated (29) mouse embryos whose livers lacked HNF4a (Hnf4 loxP/loxP Alfp.cre) and control mouse embryos (Hnf4 loxP/1 Alfp.cre) were tested for the expression of HL. RT-PCR was carried out as described previously (28) using mouse HL primers (forward, 59GCTGTCGTCTCAGAC-CTCAGC39; reverse, 59GAGCAGGATCAACTCGCCGATC39) and primers for mouse Hprt, Hnf4a, aldolase B, and albumin (described in 28).

RESULTS
Binding of nuclear receptors to the DR4 and DR1 elements of the proximal LIPC gene promoter Sequence evaluation of the proximal LIPC gene promoter revealed the presence of two inverted DR elements (Fig. 1A) that conform to the consensus DNA response element half-site 59-RG(G/T)TCA-39 (Fig. 1B). The first element was a DR1 located between 2226 and 2238. The second site was a DR4 located between 2287 and 2302. DR elements are known to bind a number of nuclear receptors, including HNF4a, ARP-1, RARa, and RXRa. Binding of these nuclear receptors to the HL DR1 and DR4 elements was evaluated by EMSA using nuclear extracts from the human hepatocarcinoma cell line, HuH7, which expresses all of these nuclear receptors as well as HL (Fig. 2). For DR1, significant amounts of a single DNA/ protein complex were observed with nuclear extracts from HuH7 cells with the WT probe ( Fig. 2A, lane 2) but not with the mutant DR1 probe (Fig. 2A, lane 5). The nuclear extract binding was specific to DR1, being effectively outcompeted by an excess of nonlabeled DR1 WT probe ( Fig. 2A, lane 3) but not by an excess of nonlabeled DR1 mutant probe (Fig. 2A, lane 4). The DR1 DNA/protein complex from HuH7 nuclear extracts on the WT DR1 probe, but not the mutant DR1 probe, was shown by supershifting with specific antiserum to contain the nuclear receptors HNF4a and Coup-TF (ARP-1) (Fig. 2B, compare lanes 2, 3 with lanes 7, 8) but neither RARa nor RXRa (Fig. 2B, lanes 4, 5). Given that the anti-COUP-TF antibody used in this study does not discriminate between ARP-1/ Coup-TFII and Coup-TFI, one or both COUP-TFs could be components of the DNA/protein complex. Similarly, nuclear extracts from HuH7 cells formed a single major complex with the WT probe (Fig. 2C, lane 2) but not the mutant DR4 probe (Fig. 2C, lane 5). Binding specificity for the DR4 element was shown by competition for binding by an excess of nonlabeled DR4 WT probe (Fig. 2C, lane 3) but not by an excess of the DR4 mutant probe (Fig. 2C,  lane 4). The DR4 DNA/protein complex on the WT DR4, but not the mutant DR4 probe, was supershifted with antiserum specific for HNF4a and Coup-TF (ARP-1) (Fig. 2D, compare lanes 2, 3 with lanes 7, 8) but not by anti-RARa or anti-RXRa (Fig. 2D, lanes 4, 5). Figure 2E shows the binding and supershifting of RXRa and RARa to the rCyp7a1 DR5 site, a known RXRa/RARa responsive element (19).
Transcriptional regulation of the LIPC promoter by HNF4a, HNF1a, and ARP-1 Based on the EMSA data described above and the previously identified binding of HNF1a to the LIPC promoter (13), we investigated, by transient transfection assays, the ability of HNF4a, ARP-1, and HNF1a to modulate the expression of the LIPC promoter in HuH7 cells. HuH7 cells were cotransfected with the pGL4-851 (WT) luciferase reporter construct and vectors expressing HNF4a, HNF1a, or ARP-1 (Fig. 4A). Transfection of either HNF4a or HNF1a resulted in a 2.5-fold increase of LIPC promoter activity, whereas transfection with ARP-1 showed a slight, but statistically significant, repression of z15%. Because both HNF1a and ARP-1 have been shown to affect transcriptional regulation by HNF4a, we also tested the effect of cotransfecting HNF4a, HNF1a, and ARP-1 in various combinations (Fig. 4A). Coexpression of HNF4a and HNF1a had an additive effect, whereas ARP-1 repressed induction of the LIPC promoter by HNF4a but not induction by HNF1a. These results provide functional support for the importance of these transcription factors in modulating LIPC expression.
Because both HNF4a and ARP-1 are known to bind to the DR1 site, the loss of LIPC induction by HNF4a upon cotransfection with ARP-1 could be the result of competition for DR1 binding. To test this hypothesis, transient transfection assays were performed with the pGL4-851 promoter construct and various ratio combinations of HNF4a and ARP-1 expression vectors (Fig. 4B). The results showed that increasing amounts of ARP-1 led to greater repression of pGL4-851 promoter construct expression. Conversely, ARP-1 repression of LIPC promoter induction by HNF4a was completely overcome by increasing amounts of transfected HNF4a expression vector. Interestingly, cotransfection of equal amounts of ARP-1 and HNF4a expression vectors did not increase the ARP-1 repression. Only when HNF4a was in excess compared with ARP-1 was the repression overcome.
Because HNF4a and ARP-1 appear to compete for transcriptional control of the LIPC promoter via the DR1 element, the relative binding affinities of each protein to the DR1 site were examined. Saturation binding assays were performed by incubating increasing concentrations of radiolabeled DR1 probe with a constant amount of WCEs from COS7 cells that were transiently transfected with HNF4a or ARP-1 expression vectors (Fig. 4C). The K d values were determined from the Scatchard plots depicted in Fig. 4C in the panels at right. The results indicate that ARP-1 binds to the DR1 element with a much higher affinity (K d 5 1.5 nM) than HNF4a (K d 5 13.8 nM). This could explain way HNF4a needs to be in excess to outcompete ARP-1 for binding to DR1. These results support the notion that ARP-1 antagonizes HNF4a transactivation of the LIPC promoter via competition for the DR1 element.

Identification of the transcription factor binding sites responsible for HNF4a and HNF1a induction of the LIPC promoter
To determine whether the transactivation by HNF4a and HNF1a occurs through the identified DR1, DR4, and HNF1 binding elements (Fig. 1), transient cotransfections were carried out in HuH7 cells using promoter constructs with mutations in the DR4 (pGL4-851DR4mut), DR1 (pGL4-851DR1mut), and HNF1 (pGL4-851HNF1BSmut) binding sites as well as combinations of these mutants (pGL4-851-DR4/DR1mut, pGL4-851-DR4/DR1/HNF1BSmut) (Fig. 5). Analysis of the basal level of activity of these promoter constructs without the addition of any exogenously expressed vectors showed a significant reduction upon mutation of the DR4 and HNF1 binding site but not upon mutation of the DR1 site (Fig. 5A). The combination of all three mutations nearly eliminated all basal activity of the promoter. These data suggest a role for the DR4 and HNF1 binding sites in the endogenous expression of the LIPC proximal promoter in HuH7 cells. The lack of effect on basal activity of the LIPC promoter by the DR1  mutant suggests that DR1 is not required for the basal level of expression.
Cotransfection of these mutant constructs with HNF4a revealed that both DR4 (Fig. 5C) and DR1 (Fig. 5D) are essential for the observed 2.5-fold induction of the LIPC proximal promoter by HNF4a (Fig. 5B), because mutation of either site resulted in a complete loss of induction. Neither mutation had an effect on LIPC induction by HNF1a (Fig. 5C, D, F). Interestingly, the DR4/DR1 double mutant showed a slight repression upon addition of HNF4a (Fig. 5F, G). Mutation of the HL HNF1 binding site completely abolished the induction produced by overexpression of HNF1a but had no effect on induction by HNF4a (Fig. 5E). These data corroborate the EMSA findings and show that binding of these proteins to the DR1, DR4, and HNF1 binding sites is functionally relevant.

Stimulation of HNF4a-mediated transactivation by PGC-1a
It has been shown that PGC-1a stimulates the induction of numerous promoters through coactivation of HNF4a. To elucidate whether PGC-1a can stimulate the HNF4amediated transactivation of the LIPC promoter, transient cotransfections were carried out in HuH7 cells using PGC-1a and HNF4a. PGC-1a alone was unable to stimulate expression of the pGL4-851 WT LIPC promoter construct. However, cotransfection of PGC-1a with HNF4a resulted in a significantly greater activation than transfection of HNF4a alone (Fig. 6). The induction by PGC-1a and HNF4a was promoter-dependent, because no activation was observed with the promoterless pGL4 vector. To identify the regulatory elements mediating the effect of PGC-1a, reporter constructs containing mutations in the HL DR1, DR4, and HNF1 elements were tested (Fig. 6). Although some remaining response was observed, mutations in both the DR1 and DR4 elements, but not in the HNF1 site, markedly diminished the response to PGC-1a and HNF4a. These results show that PGC-1a moderately stimulates the HNF4a-mediated induction of the LIPC proximal promoter preferentially via the DR1 and DR4 elements.

Determination of whether endogenously expressed HNF4a and HNF1a interact with the LIPC proximal promoter in vivo
In vivo interaction of endogenously expressed HNF4a and HNF1a with the LIPC proximal promoter was evalu-ated by ChIP analysis of anti-HNF4a or anti-HNF1a antiserum in HuH7 cells. Endogenous HNF4a was not observed to be associated with the LIPC DR1 region or any other region of the LIPC proximal promoter, as determined by the lack of enrichment by the HNF4a antibody over rabbit IgG (data not shown). ChIP analysis for binding of endogenously expressed HNF1a to the LIPC proximal promoter was inconclusive, because both HNF1a antibodies tested failed with the LIPC promoter region as well as with the positive control (the HNF4a proximal promoter) for HNF1a antibody binding (data not shown). Because ChIP assays have a narrow window of detection and may not work on all DNA elements, these results do not necessarily exclude endogenous HNF4a and HNF1a binding to the HL proximal promoter.

Suppression of LIPC expression by HNF4a antagonists
Fatty acyl-CoAs longer than C12 specifically bind to the ligand binding domain of HNF4a and result in activation or suppression of its transcriptional activity based on the degree of saturation of the fatty acyl-CoAs (64). Fibrates and Medica analogs (fatty acyl-CoAs C18-C22) are HNF4a antagonists that suppress the expression of HNF4a-responsive genes. We treated HuH7 cells with M16 to test whether endogenously expressed HNF4a is important for hepatic lipase expression in vivo (Fig. 7). Endogenous hepatic lipase mRNA expression in HuH7 cells was inhibited in a dose-dependent manner upon treatment with 250-400 mM M16. Expression of the APOC3 gene, which has been shown to be an HNF4a-responsive gene (64,65), was also repressed by M16. Surprisingly, treatment with 100 mM M16 resulted in a slight, but statistically significant, increase in hepatic lipase mRNA expression. This was not seen with APOC3, and further analysis will be required to determine the cause of this increase. The repression observed by M16 supports the notion that HNF4a is involved in LIPC regulation in vivo.

HNF4a is essential for in vivo hepatic LIPC expression in mice
We next examined the role of HNF4a in the regulation of LIPC expression in an animal model with a liver-specific targeted disruption of the mouse Hnf4a gene. Disruption of Hnf4a in the mouse was observed to cause lethality during early embryogenesis (30). Recently, however, this early embryonic arrest was circumvented (66) by the generation of a liver-specific targeted disruption of Hnf4a using the Cre-lox system (29,67). This allowed for the generation of mouse embryos with livers lacking HNF4a that are viable at 18.5 days after coitus. To determine whether HNF4a is essential for hepatic Lipc expression in mice, we compared the steady-state level of HL mRNA by RT-PCR in livers from mouse embryos both containing and lacking HNF4a. HL mRNA was undetectable in total cellular RNA from livers of Hnf4a-deficient mice. Expression of Hprt was included as a control for the amount of mRNA used in the RT-PCR, and expression of the albumin and aldolase B genes was used as examples of HNF4aindependent and -dependent genes, respectively (28) (Fig. 8). Like the M16 data, these results support the importance of HNF4a for hepatic lipase expression in vivo but do not indicate whether the effect is direct or indirect.

DISCUSSION
Hepatic lipase plays a key role in the modulation of plasma lipoprotein particle size and density. High levels of HL are associated with an atherogenic lipoprotein profile composed of high levels of small, dense LDL and HDL particles. Both genetic and physiologic factors influence the level of HL. However, very few cis-regulatory elements that regulate LIPC expression have been identified. Sequence analysis of the human LIPC proximal promoter revealed the presence of two phylogenetically conserved inverted DRs (DR1 and DR4) with the motif 59-RG(G/ T)TCA-39 as well as a putative HNF1 binding site. These sites bind proteins such as RXRa, RARa, HNF4a, HNF1a, and ARP-1, which are known to regulate genes important in lipid metabolism. In this report, we present evidence that HNF4a, PGC-1a, ARP-1, and HNF1a may play roles in regulating LIPC gene expression.
Transient cotransfection experiments showed that HNF4a induced the LIPC proximal promoter by 2.5-fold (Fig. 4A). Site-directed mutagenesis of either the DR1 or DR4 element abrogated this induction (Fig. 5). Because HNF4a has been shown to exclusively form homodimers that bind DRs separated by zero, one, or two nucleotides (DR0, DR1, or DR2) (31,(68)(69)(70), the requirement of both the DR4 and DR1 elements for activation by HNF4a suggested potential interaction between HNF4a bound to the DR1 element and another transcription factor bound to DR4 (Fig. 5C, D). There is precedent for this type of interaction in regulating gene expression by HNF4a. Numerous studies have shown that transactivation by HNF4a depends on synergistic interactions between HNF4a and additional factor(s) bound to other regulatory elements. For example, transactivation of the APOC3 gene is mediated by direct physical interaction between HNF4a bound to the I4 element of the APOC3 enhancer with the specificity protein 1 bound to multiple sites (71,72). In addition, cooperative binding of USF and HNF4a has been shown to drive the transcription of APOA2 (73) and APOC3 (74). Interestingly, USF has been shown to activate the LIPC proximal promoter (12), but its ability to transcriptionally synergize with HNF4a has not been tested. Moreover, activation of the APOB promoter by HNF4a requires the interaction of HNF4a bound to element III with C/EBPa bound to the adjacent element IV (43). Although specificity protein 1, USF, and C/EBPa are good  4) and Hnf4adeficient (HNF4 loxP/loxP ;AlfpCre) livers (lanes 5-7) were determined by RT-PCR. All samples contained comparable amounts of mRNA, as indicated by Hprt amplification in the presence of reverse transcriptase (RT). No amplification was observed in the absence of mRNA (lane 1). In mouse embryos whose livers lack HNF4a (lanes 5-7), the mRNA levels of HL and the positive control, aldolase B, in liver were undetectable. The same mRNA level of the negative control, albumin, was present in both normal and Hnf4adeficient livers.  examples of factors that show transcriptional synergism with HNF4a, none of them are known to bind DR4 elements.
Because DR4 elements are known to bind homodimers and heterodimers, the factor binding to DR4 could represent more than one protein. We tested HNF4a, ARP-1, RXRa, and RARa for their ability to bind and induce the LIPC promoter via the DR4 site. EMSA results showed binding of HNF4a and ARP-1 to the DR4 site (Figs. 2, 3). However, despite these results, we do not believe that HNF4a or ARP-1 represents the factor hypothesized to bind to DR4. To begin with, HNF4a has not been shown to bind DR4 elements (68,70). However, it is possible that this is a novel finding and that the HNF4a-mediated activation of the HL promoter requires HNF4a binding to both the DR1 and DR4 elements. Numerous promoters contain two HNF4a binding sites [e.g., APOC3 (70), ornithine transcarbamylase (44), and microsomal triglyceride transfer protein (58)]. However, the HNF4a binding sites are generally both DR1 elements, and loss of activation by HNF4a requires mutation of both sites, not just one, as seen with the HL promoter. Additionally, despite the fact that homodimers of ARP-1 have been shown to bind DR4 elements (19,20) and actively induce or repress transcription (reviewed in 47), overexpression of ARP-1 alone showed little effect on LIPC promoter expression (Fig. 4A). Because the factor binding to HL DR4 appears to play a positive role in HL expression, the lack of activation upon overexpression of ARP-1 suggests that it is not the factor binding to the HL DR4 element. DR4 elements are also known to bind RXRa homodimers and heterodimers with various partners. However, supershifting studies using anti-RXRa antiserum showed no RXRa in the DNA/protein complex bound to the DR4 element (Fig. 2D, lane 5). It is possible that RXRa may not be expressed at high enough levels in HuH7 cells to see binding from nuclear extracts. However, the lack of HL promoter activation by RXRa in transient transfection assays (data not shown) suggests that it is not involved in the regulation of the HL promoter. Therefore, our analysis has ruled out the proteins most commonly known to bind DR4 sequences as being the DR4-bound factor. Sequence analysis of the DR4 region failed to reveal any additional transcription factor consensus sites. We are in the process of trying to determine the identity of the unknown factor(s) binding to DR4 (protein X in Fig. 9).
HNF4a also appears essential for in vivo hepatic LIPC expression. HL mRNA expression was shown to be completely abolished in livers of Hnf4a-deficient mice (Fig. 8). Moreover, treatment of HuH7 cells with the HNF4a antagonist M16 led to the reduction of endogenous HL mRNA expression as a result of the inhibition of transactivation by HNF4a (Fig. 7). These data show the importance of HNF4a to the in vivo expression of HL but do not indicate whether the role of HNF4a is direct, indirect, or both. To further elucidate the in vivo role of HNF4a, we tested the ability of endogenously expressed HNF4a to bind the DR1 element by ChIP. ChIP analysis failed to show binding of endogenous HNF4a to the DR1 region in HuH7 cells. This lack of endogenous HNF4a binding to the LIPC DR1 element is supported by the lack of effect on basal LIPC promoter activity upon mutation of the DR1 site (Fig. 5A). However, it is possible that, as a result of the DR1 binding competition between HNF4a and ARP-1, the amount of endogenous HNF4a binding to the HL DR1 is below the detection limit of the assays performed. On the other hand, if endogenously expressed HNF4a does not regulate the LIPC proximal promoter in vivo, it still could play a direct role in LIPC expression by binding to sequences outside the 2851 to 129 LIPC proximal promoter. However, we failed to detect any further increase in LIPC promoter activation by HNF4a over that found with the proximal promoter in transient transfection assays using LIPC promoter constructs containing sequence up to 24.7 kb upstream of the transcription start site (data not shown). HNF4a could also have an indirect effect whereby HNF4a could activate the transcription of another protein that is important for direct regulation of the LIPC proximal promoter. This other protein could be the factor binding to DR4 or HNF1a, which is positively regulated by HNF4a (75,76).
ARP-1 appears to block binding and activation by HNF4a through competition for DR1 binding. Blocked binding through direct competition with HNF4a for the DR1 site is a commonly observed mechanism of repression for ARP-1 (47). Data from various experiments support the competition model. EMSA studies showed that both ARP-1 and HNF4a bind the DR1 site (Figs. 2, 3), and determination of the relative DR1 binding affinities indicated that ARP-1 binds the DR1 element with much higher affinity than HNF4a (Fig. 4C). The ability of ARP-1 to out-compete HNF4a for binding to DR1 explains how cotransfection of equal amounts of ARP1 and HNF4a results in the loss of LIPC promoter induction by HNF4a (Fig. 4A). However, when HNF4a is expressed in excess compared with ARP-1, induction by HNF4a is restored (Fig. 4B). The higher binding affinity of ARP-1 could explain why higher concentrations of HNF4a are needed to out-compete ARP-1 for binding to DR1.
We have also determined that HNF1a is important for LIPC expression. Transfection studies showed a significant loss of basal activity of the LIPC promoter construct upon mutation of the HNF1 binding site (Fig. 5A) and a 2.5-fold induction upon overexpression of HNF1a (Fig. 4A). Sitedirected mutagenesis of the HNF1 binding site abolished this induction (Fig. 5E). Coexpression of both HNF1a and HNF4a normally leads to a synergistic effect (58,59) or repression of HNF4a-induced transcription (57) attributable to physical interaction between these two proteins. However, we have demonstrated that cotransfection of both HNF1a and HNF4a produces an additive effect on the LIPC proximal promoter (Fig. 5B). This could indicate that on the LIPC promoter these two proteins do not interact physically but instead interact with the basal transcription machinery individually. Despite the fact that we have been unable to confirm the binding of endogenous HNF1a to the LIPC proximal promoter by ChIP analysis, we feel that the loss of basal activity and induction by HNF1a upon mutation of the LIPC HNF1 binding site and the fact that only members of the HNF1 family (HNF1a and HNF1b) have been shown to bind the HNF1 consensus sequence (54) strongly support the argument that HNF1a is important to the in vivo expression of HL. In addition, microarray analysis of HNF1a-deficient versus WT mice showed a dramatic reduction (80%) in mouse LIPC mRNA expression, further supporting the importance of HNF1a for the regulation of LIPC expression in vivo (53).
In conclusion, we have provided in vitro and in vivo evidence for the importance of HNF4a in the activation of the LIPC proximal promoter via the DR1 and DR4 elements. The coactivator PGC-1a appears to moderately stimulate this HNF4a-mediated transactivation. We have also shown that HNF1a independently activates the LIPC proximal promoter using the HNF1 binding site and that ARP-1 acts as an antagonist to induction by HNF4a through competition for DR1 binding. Based on these results, we propose the model shown in Fig. 9.
The level of hepatic lipase in each individual will ultimately be determined by the interplay between these various transcription factors and modulated by signals that affect the levels of each factor. HNF4a, HNF1a, and ARP-1 are part of a complex regulatory network responsible for defining the hepatic phenotype. Cross-regulation and autoregulation between these and other factors are important mechanisms for the establishment of strictly controlled interdependent regulatory pathways that lead to balanced high-level expression of the main factors needed in hepatocytes (51,57,62,(75)(76)(77). Therefore, signals that affect the expression of one or all of these transcription factors will in turn influence the expression of the others and ultimately determine the level of expression of the many proteins involved in lipid metabolism that they regulate. The better we understand how HNF4a, HNF1a, and ARP-1 are regulated and interact to modulate hepatic lipase expression, the better we will be able to help in the management of lipoprotein profiles that predispose to coronary artery disease and atherosclerosis.