Transcriptional Regulation of Apolipoprotein C-III Gene Expression by the Orphan Nuclear Receptor ROR a *

Triglyceride-rich remnant lipoproteins are considered as major risk factors contributing to the pathogen-esis of atherosclerosis. Because apolipoprotein (apo) CIII is a major determinant of plasma triglyceride and remnant lipoprotein metabolism, it is important to un-derstand how the expression of this gene is regulated. In the present study, we identified the orphan nuclear receptor ROR a 1 as a regulator of human and mouse apo C-III gene expression. Plasma triglyceride and apo C-III protein concentrations in staggerer ( sg / sg ) mice, homozygous for a deletion in the ROR a gene, were signif-icantly lower than in wild type littermates. The lowered plasma apo C-III levels were associated with reduced apo C-III mRNA levels in liver and intestine of sg/sg mice. (Promega) following the manufacturer’s instructions. DNA-protein binding assays were conducted as described (68) using the following binding buffer: 10 m M Hepes, 50 m M KCl, 1% glycerol, 2.5 m M MgCl 2 , 1.25 m M dithiothreitol, 0.1 m g/ m l poly(dI-dC), 50 ng/ m l herring sperm DNA, 1 m g/ m l bovine serum albumin containing 10% of programmed or unprogrammed reticulocyte lysate. Double-stranded oligonucleotides were end-labeled using T4 polynucleotide kinase and [ g - 32 P]ATP and used competition experiments, amounts cold oligonucleotide included 15 oligonucleotides.

Several epidemiological studies support the idea that, in addition to elevated low density lipoprotein and reduced high density lipoprotein cholesterol, elevated triglycerides constitute an independent risk factor for coronary heart disease (1)(2)(3)(4). More specifically, triglyceride-rich lipoprotein remnants are positively correlated to the progression of atherosclerosis (5,6). Identifying the factors or genes controlling triglyceride metabolism is therefore of major importance and may provide means for pharmacological intervention in dyslipidemic patients.
Apolipoprotein (apo) 1 C-III is a 79-amino acid glycoprotein synthesized in the liver and, to a lesser extent, in the intestine, that plays a key role in plasma triglyceride metabolism as evidenced by pharmacological (7)(8)(9), clinical (10,11), genetic (12), and experimental data in transgenic animal models (13). Apo C-III concentrations in plasma are positively correlated with plasma triglyceride levels, both in the normal population as well as in hypertriglyceridemic patients (10,11,14) or in transgenic animals (15). Moreover, apo C-III deficiency in humans (16) or apo C-III gene disruption in transgenic mice (17) results in increased catabolism of very low density lipoprotein particles, whereas increased apo C-III synthesis occurs in hypertriglyceridemic patients (18). Results from both in vivo (19 -22) and in vitro (23)(24)(25) studies indicate that apo C-III delays the catabolism of triglyceride-rich particles. Several potential mechanisms may participate in the inhibitory effect of apo C-III on triglyceride catabolism. These include inhibition of lipolysis by lipoprotein (16,26) or hepatic lipase (27), inhibition of triglyceride-rich particle binding to glycosaminoglycans (22), as well as interference with apo E-mediated receptor clearance of remnant particles from plasma (19,20,22,25).
Apo C-III gene expression is tightly regulated, being downregulated by hormones such as insulin (28,29) or thyroid hormones (30), cytokines such as interleukin-1 (31) or tumor necrosis factor ␣ (32), as well as hypolipidemic drugs such as fibrates (7,33) or ␤-blocked fatty acids (8,34). By contrast, its expression is increased by retinoids (9). Regulatory sequences determining the tissue-specific expression pattern of apo C-III have been delineated in its gene (35)(36)(37)(38). The C3P site located at position Ϫ87/Ϫ67 relative to the transcription start site is a major determinant of apo C-III promoter activity (35,36). It contains a direct repeat of two AGGTCA half-sites separated by one nucleotide (DR-1) to which the nuclear hormone receptors HNF-4, PPAR, RXR, Ear2, COUPTF-I, and COUPTF-II are binding (8,9,37,39,40). Whereas Ear2, COUPTF-I, and COUPTF-II repress apo C-III promoter activity via these sites (37,39), HNF-4 activates it (35-37, 39, 40). PPAR/RXR heterodimers also enhance the activity of reporter construct containing the C3P site cloned in front of an heterologous promoter (9). In addition, the Ϫ592/Ϫ792 fragment of the apo C-III * This work was supported by grants from INSERM and Merck-Lipha. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ ‡ To whom correspondence should be addressed. promoter acts as an enhancer that potentiates the strength of the proximal apo C-III promoter (38). Functional positive HNF-4 and Sp1 binding sites as well as negative COUPTF-I and COUPTF-II binding sites have been mapped in this region (38,40). In addition, a CCAAT/enhancer binding protein ␦ binding site located in the proximal apo C-III promoter (Ϫ171/ Ϫ137) seems to be involved in the negative regulation of apo C-III expression by interleukin-1 (31). Finally, T 3 R␤ (41), ATF-2 (42), NFB (43), and Jun (42) regulatory elements have also been identified in the human apo C-III promoter.
The ROR (retinoic acid receptor related orphan receptor; also termed RZR) orphan receptors (44 -46) are a subfamily of orphan nuclear receptors consisting of three different genes ROR␣, ␤, and ␥ (44,46,47). RORs were initially reported to bind as monomers to response elements consisting of a 6-base pair AT-rich sequence preceding the half-core PuGGTCA motif (44,48,49), but more complex response elements have also been described (50,51). Because of alternative splicing and promoter usage, the ROR␣ gene gives rise to four isoforms: ␣1, ␣2, ␣3, and RZR␣ (44 -46), which differ in their N-terminal domains and display distinct DNA recognition and transactivation properties (44). In contrast to ROR␤, the expression of which is restricted to brain, retina, and pineal gland (52), both ROR␣ and ROR␥ are widely expressed in peripheral tissues (44,45,47,49). Based on the presence of putative response elements in their promoter, several target genes for ROR subfamily members were proposed and analyzed in vitro (53)(54)(55)(56). A role for ROR␣1 has been proposed in muscle differentiation (57), whereas ROR␥ expression is induced during adipose tissue differentiation (58). Transgenic mice have been developed that carry a deleted ROR␣ gene (59,60). Their phenotype is similar to the one of staggerer mice, which carry a natural deletion in the ROR␣ gene that prevents the translation of its putative ligand-binding domain, thereby presumably disrupting the normal function of this transcription factor (61). These mice exhibit deficient intestinal apo A-I expression (62), suggesting that the mouse apo A-I gene is an in vivo target of ROR␣. Moreover, when maintained on a high fat atherogenic diet staggerer (sg/sg) mice develop a severe hypo-alphalipoproteinemia and atherosclerosis, suggesting an important role for ROR␣ in cardiovascular and metabolic diseases (63).
In the present study, we investigated the regulation of apo C-III expression and triglyceride metabolism by the orphan nuclear receptor ROR␣ in vivo using the staggerer mouse model. We observed a striking reduction in both triglyceride and apo C-III plasma levels in mutant compared with wild type mice. Next, we studied in vitro the molecular mechanisms regulating apo C-III gene transcription by ROR␣. Our results indicate that ROR␣ enhances the activity of the human Ϫ1415/ ϩ24 apo C-III promoter. Furthermore, a ROR␣ response element was identified at position Ϫ23/Ϫ18 that confers ROR␣ responsiveness to the human apo C-III promoter. This response element is preserved in both the human and the mouse apo C-III gene promoters and confers ROR␣ responsiveness to a heterologous promoter. Taken together, our results identify ROR␣ as a positive regulator of apo C-III gene transcription and support a role of ROR␣ as a regulator of lipid and lipoprotein metabolism.

MATERIALS AND METHODS
Mice-staggerer mutant mice were obtained by crossing heterozygote (ϩ/sg) mice maintained in a C57BL/6 genetic background and identifying homozygous offspring by polymerase chain reaction genotyping and by their clinical ataxia. Mice were maintained on chow diet purchased from UAR (France) as described previously (62). Wild type littermates of the same age as the homozygous mutants were used as control. 10-week-old mice fasted overnight were killed by ether overdose. Blood, liver, and intestine samples were taken and stored for further analysis.
Lipid and Lipoprotein Analysis-Plasma triglyceride concentrations were determined by enzymatic assays using commercially available reagents (Roche Molecular Biochemicals), whereas plasma levels of apo C-III were measured by an immunonephelometric assay using a specific polyclonal antibody as described previously (65).
Cloning of Recombinant Plasmids-The plasmid containing the Ϫ1415/ϩ24 sequence of the human apo C-III gene promoter cloned in front of the chloramphenicol acetyltransferase (CAT) reporter gene (Ϫ1415/ϩ24WT-CAT) has been described previously (9). The luciferase gene from the plasmid pGL3 (Promega, Madison, WI) (SacI-BamHI) was subcloned between the corresponding sites of the vector pBKCMV (Stratagene, La Jolla, CA) (pBKCMV-Lucϩ). The CAT reporter gene of the plasmid Ϫ1415/ϩ24WT-CAT was then excised by digestion with KpnI and BamHI and replaced by the luciferase-containing KpnI-BglII fragment of the plasmid pBKCMV-Lucϩ plasmid. The Ϫ1415/ϩ24 fragment of the apo C-III promoter was excised from the resulting construct by HindIII digestion and cloned in the corresponding site of the vectors pGL3 (construct Ϫ1415/ϩ24WTpGL3) and pSL301 (Amersham Pharmacia Biotech) (construct PSL301Ϫ1415/ϩ24hCIII). The pSL301Ϫ1415/ϩ24hCIII construct was partially digested with EcoO109I and religated. The resulting construct was digested with XbaI and HindIII. The insert was cloned in the corresponding sites of pGL3 to create the construct Ϫ108/ϩ24WTpGL3. The plasmid Ϫ1415/ϩ24WTpGL3 was then used as template to polymerase chain reaction amplify fragments of different length of the human apo C-III promoter using forward primers annealing to specific parts of the promoter sequence and containing a NheI restriction site and a reverse primer annealing downstream of the pGL3 polylinker. The polymerase chain reaction products were cut with NheI and HindIII and cloned in the corresponding sites of the pGL3 vector. Site-directed mutagenesis of the construct Ϫ1415/ ϩ24WTpGL3 was performed using the Quick Change site-directed mutagenesis kit (Stratagene) following the manufacturer's instructions. The constructs (Ϫ58/Ϫ27) 8s TkpGL3 and (Ϫ47/Ϫ79) 1s TkpGL3 were obtained following the described previously strategy (66) based on intermediary cloning in the BamHI and BglII sites of the vector pic20H using double-stranded oligonucleotides with sequences corresponding to the indicated fragment of the apo C-III promoter flanked by protruding ends compatible with BamHI and BglII sites. The oligonucleotide multimers were excised from pic20H with SalI and XhoI and cloned in the XhoI site of the described previously vector TkpGL3 (34). Alternatively, these oligonucleotide multimers were directly cloned into the BglII site of the TkpGL3 vector after disruption of the BamHI site by Klenow blunting. The construct pCDNA3-hROR␣1 containing the hROR␣1 cDNA cloned in the KpnI and XbaI sites of the pCDNA3 vector was a gift of Dr. A. Shevelev. The Renilla luciferase gene of the pRLnull construct (Promega) was excised by the enzymes NheI and XbaI and cloned in the XbaI site of the plasmid pBKCMV. The resulting construct was cut by HindIII and XbaI, and the insert was cloned in the corresponding sites of the pGL3 control vector (Promega) to yield the pRenConTϩ construct used to evaluate transfection efficiency.
Cell Culture and Transient Transfection Assays-Human hepatoma HepG2, CaCO2 and RK13 cells were obtained from European Collection of Animal Cell Cultures (Porton Down, Salisbury, UK). Cell lines were maintained in standard culture conditions (Dulbecco's modified Eagle's minimal essential medium supplemented with 10% fetal calf serum at 37°C in a humidified atmosphere of 5% CO 2 /95% air). Medium was changed every 2 days.
Cells were seeded in 24-well plates at a density of 5 ϫ 10 4 , 6 ϫ 10 4 , or 10 5 cells/well for RK13, HepG2, or CaCO2, respectively, and incubated at 37°C for 16 h prior to transfection. Cells were transfected using the cationic lipid RPR 120535B as described previously (34) with reporter plasmids (at 50 ng/well), expression vectors (pCDNA3 or pCDNA3-hROR␣1 at 100 ng/well), and the control plasmids (pRen-Contϩ at 1 ng/well or pSV-␤-gal at 50 ng/well). At the end of the experiment, the cells were washed once with ice-cold 0.15 M NaCl, 0.01 M sodium phosphate buffer, pH 7.2, and the luciferase activity was measured with the Dual-Luciferase TM Reporter Assay System (Promega) according to the manufacturer's instructions. All transfection experiments were performed at least three times. The ␤-galactosidase activity was measured as described previously (66). Protein content of the extract was evaluated by the Bradford assay using the kit from Bio-Rad.
Gel Retardation Assays-ROR␣ was in vitro transcribed from the pCDNA3-hROR␣ plasmid using T7 polymerase and subsequently translated using the TNT-coupled transcription/translation system (Promega) following the manufacturer's instructions. DNA-protein binding assays were conducted as described (68) using the following binding buffer: 10 mM Hepes, 50 mM KCl, 1% glycerol, 2.5 mM MgCl 2 , 1.25 mM dithiothreitol, 0.1 g/l poly(dI-dC), 50 ng/l herring sperm DNA, 1 g/l bovine serum albumin containing 10% of programmed or unprogrammed reticulocyte lysate. Double-stranded oligonucleotides were end-labeled using T4 polynucleotide kinase and [␥-32 P]ATP and used as probe. For competition experiments, the indicated amounts of cold oligonucleotide were included 15 min before adding labeled oligonucleotides.

Plasma Triglyceride and apo C-III Concentrations as Well as
Hepatic and Intestinal apo C-III mRNA Levels Are Decreased in staggerer Mice-To determine whether staggerer mice display altered triglyceride metabolism, plasma triglycerides were measured in overnight fasted female staggerer mice and compared with age-matched wild type C57BL/6 littermates (Fig. 1). Interestingly, staggerer mice exhibited 50% lower blood triglyceride levels compared with wild type littermates (Fig. 1A). Because apo C-III is a major determinant of plasma triglyceride levels (13), its plasma concentrations were measured next. A 70% decrease of plasma apo C-III concentration was observed in mutant mice (Fig. 1B). To determine whether this reduction was associated with a decreased expression of the apo C-III gene, hepatic and intestinal apo C-III mRNA levels were analyzed by Northern blotting. Both intestinal and liver apo C-III mRNA levels were reduced in mutant mice compared with wild type mice (Fig. 1C). Hepatic and intestinal 36B4 mRNA levels measured as control were similar in both groups.
Overexpression of hROR␣1 Enhances the Activity of the Human apo C-III Gene Promoter-Because staggerer (sg/sg) mice carry a nonfunctional ROR␣ gene (61), the above data suggest that ROR␣ is a positive regulator of apo C-III transcription. Transient transfection assays were performed to determine whether human ROR␣ controls the transcription of the human apo C-III gene. In hepatoma HepG2 cells, which produce apo C-III, cotransfection of a human nuclear receptor ROR␣1 ex-pression plasmid resulted in an increased activity of the luciferase reporter gene driven by the Ϫ1415/ϩ24 fragment of the human apo C-III gene promoter ( Fig. 2A). An activation was also observed in rabbit kidney RK13 cells that do not express apo C-III (Fig. 2B) and in human intestinal CaCO2 cells (Fig.  2C) that produce apo C-III. The effect of hROR␣1 overexpression was promoter-dependent because the promotor-less vector pGL3 was unaffected in all cell lines studied.
Mapping of the Human apo C-III Promoter Sites Conferring Responsiveness to hROR␣1-To identify the response element(s) required for hROR␣1 activation of the apo C-III gene promoter, 5Ј nested deletions of the apo C-III promoter were cotransfected with the hROR␣1 expression vector in HepG2 cells. Deletion of the promoter led to a decrease in its basal activity (Fig. 3), corroborating previous observations that the Ϫ792/Ϫ592 fragment of the apo C-III promoter acts as a strong hepatic enhancer (38). However, hROR␣1 activation was still observed with the shortest construct Ϫ108/ϩ24WT-pGL3, indicating that the first 108 nucleotides of the human apo C-III promoter contain sequence determinants sufficient to confer hROR␣1 responsiveness (compare the 3.1-fold induction of the Ϫ1415/ϩ24wtpGL3 construct with the 2.6-fold induction of the Ϫ108/ϩ24wtpGL3 construct). To verify whether hROR␣1 directly binds to the proximal apo C-III promoter, radiolabeled overlapping oligonucleotides corresponding to portions of the Ϫ108/ϩ24 fragment of the apo C-III promoter were used as probes in gel shift assays. Specific binding of in vitro translated hROR␣1 protein was observed only on the Ϫ33/Ϫ16 and Ϫ90/Ϫ64 fragments (Fig.  4A). Both fragments contain an AGGTCA half-site preceded by a degenerated A/T-rich region that could function as hROR␣1 response element. Binding of hROR␣1 to the Ϫ33/ Ϫ16 fragment of the apo C-III promoter was lost after mutation of the AGGTCA half-site present in position Ϫ23/Ϫ18 (Ϫ33/Ϫ16mt: Ϫ22 G3C , Ϫ21 G3A ) (Fig. 4B). The binding of hROR␣1 to the Ϫ33/Ϫ16 fragment of the apo C-III promoter was displaced by increasing amounts of either the cold Ϫ33/ Ϫ16 double-stranded oligonucleotide or a cold doublestranded oligonucleotide that contains one copy of the hROR␣1 consensus binding site (Fig. 5A). This binding was not displaced by the mutated cold Ϫ33/Ϫ16 double-stranded oligonucleotide (Fig. 5A). Binding of hROR␣1 to the Ϫ90/Ϫ64 fragment corresponding to the C3P site of the human apo C-III gene promoter was specific because it could be displaced by increasing amounts of either cold Ϫ90/Ϫ64 (Fig. 5B) or

FIG. 1. staggerer mice have decreased plasma triglyceride and apo C-III levels associated with decreased hepatic and intestinal apo C-III mRNA levels compared with wild type littermates.
Overnight fasted 11-week-old homozygous female staggerer mice carrying a nonfunctional ROR␣1 gene and their wild type littermates were killed by ether overdose. Plasma triglyceride (A) and apo C-III levels (B) were measured as described under "Materials and Methods" (four animals/group). Total RNA was extracted from liver and intestinal tissues and analyzed by Northern blotting as described under "Materials and Methods." C shows the relative apo CIII mRNA levels (three animals/ group) as evaluated by quantitative scanning densitometry (Bio-Rad hROR␣1 consensus binding site double-stranded oligonucleotides (data not shown). By contrast, binding of hROR␣1 to the Ϫ90/Ϫ64 fragment of the human apo C-III gene promoter was abrogated when the AGGTCA half-site located at position Ϫ82/Ϫ77 was mutated (Ϫ90/Ϫ64mt: Ϫ78 G3C , Ϫ79 G3A ; Fig.  4B). No significant binding was observed to other fragments of the proximal human apo C-III promoter (Fig. 4A). Taken together, our results suggest the presence of two binding site for hROR␣1 on the proximal human apo C-III promoter, the first downstream of the TATA box (Ϫ23/Ϫ18) and the second within the C3P site (Ϫ82/Ϫ77).
Functional Characterization of hROR␣1 Response Elements Present in the Proximal Human and Mouse apo C-III Promoters-To evaluate whether these two putative response elements were functional in the context of the proximal human apo C-III promoter, the half-sites present downstream of the TATA box in position Ϫ23/Ϫ18 (Ϫ22 G3C , Ϫ21 G3A ) or in the C3P site in position Ϫ82/Ϫ77 (Ϫ78 G3C , Ϫ79 G3A ) of the human apo C-III promoter were mutated by site-directed mutagenesis in the Ϫ1415/ϩ24WTpGL3 construct either alone (Ϫ82/Ϫ77mt, Ϫ23/Ϫ18mt, respectively) or in combination (Ϫ82/Ϫ77mt ϩ Ϫ23/Ϫ18mt). Mutation of the Ϫ82/Ϫ77 half-site reduced the basal activity of the apo C-III promoter but did not prevent its activation by hROR␣1 (Fig. 6). Mutation of the Ϫ23/Ϫ18 halfsite enhanced the basal activity of the apo C-III promoter in HepG2 cells and abrogated hROR␣1 responsiveness. The activity of the promoter and its hROR␣1 responsiveness were lost when both half-sites were mutated simultaneously. These data suggest that the Ϫ23/Ϫ18 half-site plays a major role in the hROR␣1 responsiveness of the apo C-III promoter in HepG2 cells.
To evaluate whether these two sites could confer ROR␣ responsiveness to an heterologous promoter and to exclude that the Ϫ108/ϩ24 fragment of the apo C-III promoter contains other hROR␣1-responsive elements, overlapping fragments of the apo C-III promoter (covering the Ϫ100/Ϫ16 region of the apo C-III promoter) were cloned in front of a thymidine kinase (Tk) promoter-driven luciferase reporter vector. These constructs were cotransfected with a hROR␣1 expression vector in HepG2 cells. The (Ϫ33/Ϫ16) 3S TkpGL3 construct was strongly HepG2 cells were cotransfected with pCDNA3-hROR␣1 expression vector (hROR␣1; 100 ng) or the empty pCDNA3 vector as control (Cont.) and reporter constructs (50 ng) containing the wild type or site-directed mutated Ϫ1415/ϩ24 fragments of the human apo C-III promoter cloned in front of the luciferase reporter gene. The empty pGL3 vector was used as negative control. The AG-GTCA half-sites present in position Ϫ23/Ϫ18 (Ϫ22 G3C , Ϫ21 G3A ) and Ϫ82/Ϫ77 (Ϫ78 G3C , Ϫ79 G3A ) were mutated alone (Ϫ23/Ϫ18, Ϫ82/ Ϫ77mt, respectively) or in combination (Ϫ23/Ϫ18 ϩ Ϫ82/Ϫ77mt) as indicated. Cells were transfected and luciferase activity measured and expressed as described under "Materials and Methods." The fold induction above control level is indicated for each constructs .   FIG. 3. Identification of the human apo C-III promoter elements conferring its responsiveness to hROR␣1. HepG2 cells were cotransfected with pCDNA3-hROR␣1 expression vector (100 ng; hROR␣1) or the empty pCDNA3 vector as control (Cont.) and reporter constructs (50 ng) containing the indicated nested fragments of the human apo C-III promoter cloned in front of the luciferase reporter gene. The empty pGL3 vector (50 ng) was used as control. Cells were transfected and luciferase activity measured and expressed as described under "Materials and Methods." The fold induction above control level is indicated for each construct.
FIG. 4. hROR␣1 binding to labeled probes covering the ؊16/ ؊125 region of the human apo C-III promoter. Double-stranded oligonucleotides corresponding to overlapping fragments of the human apo C-III promoter comprised between positions Ϫ16 and Ϫ125 were prepared and labeled as described under "Materials and Methods." These probes were incubated as indicated with in vitro translated hROR␣1 protein or unprogrammed lysate as control. DNA/protein complexes were resolved by nondenaturating PAGE (A) as described under "Materials and Methods." In B, double-stranded oligonucleotides corresponding to the wild type or mutated Ϫ33/Ϫ16 (Ϫ22 G3C , Ϫ21 G3A ) and Ϫ90/Ϫ64 (Ϫ78 G3C , Ϫ79 G3A ) fragments of the human apo C-III promoter were labeled and incubated with in vitro translated hROR␣1 protein or unprogrammed lysate as control. DNA/protein complexes were resolved by nondenaturating PAGE as described under "Materials and Methods." Specific complexes not observed with unprogrammed lysate are indicated by arrows.
FIG. 5. Specificity of hROR␣1 binding to ؊33/؊16 or the ؊90/ ؊64 fragments of the human apo C-III promoter. Double-stranded oligonucleotides corresponding to the Ϫ33/Ϫ16 (A) and Ϫ90/Ϫ64 (B) fragments of the human apo C-III promoter were labeled as described under "Materials and Methods." In vitro translated hROR␣1 protein or unprogrammed lysate were incubated with 10-, 50-, and 100-fold excess of the indicated unlabeled double-stranded oligonucleotides for 15 min at 4°C before labeled probes were added for 5 min at room temperature. DNA/protein complexes were resolved by nondenaturating PAGE as described under "Materials and Methods." Specific complexes not observed with unprogrammed lysate are indicated by arrows. activated by hROR␣1, whereas the (Ϫ83/Ϫ67) 4S TkpGL3 construct was weakly stimulated (Fig. 7A). The other constructs were not activated by hROR␣1 (Fig. 7A). Both the Ϫ33/Ϫ16 and Ϫ83/Ϫ67 fragments contain an AGGTCA half-site preceded by a degenerated A/T-rich region. To evaluate the specificity of ROR␣ action, these half-sites were next mutated to create the constructs (Ϫ33/Ϫ16mt) 3S TkpGL3 (Ϫ22 G3C , Ϫ21 G3A ) and (Ϫ87/Ϫ67mt) 4S TkpGL3 (Ϫ78 G3C , Ϫ79 G3A ) that were cotransfected with a hROR␣1 expression vector in HepG2 cells. In contrast to the wild type constructs, these mutated constructs were not activated by hROR␣1 (Fig. 7B). These data suggest that the two AGGTCA half-sites to which hROR␣1 bind in the proximal human apo C-III promoter are also functional in the context of a heterologous promoter.
Because the sequence of the Ϫ33/Ϫ16 fragment of the human apo C-III promoter is almost fully conserved in the mouse promoter (67), hROR␣1 binding to the mouse sequence corresponding to this region was analyzed. As shown in Fig. 8A, hROR␣1 bound with similar affinity to the Ϫ33/Ϫ14 fragment of the mouse apo C-III gene promoter as to the Ϫ33/Ϫ16 fragment of the human apo C-III gene promoter. To compare its activity with the corresponding human promoter sequence, three copies of the wild type Ϫ33/Ϫ14 fragment of the mouse apo C-III promoter were cloned in front of the thymidine kinase promoter and tested in cotransfection assay. As shown in Fig.  8B, the human (Ϫ33/Ϫ16) 3as TkpGL3 and the mouse (Ϫ33/ Ϫ14) 3as TkpGL3 were similarly activated by hROR␣ overexpression in HepG2 cells. This indicates that the AGGTCA halfsites located downstream of the TATA box both in the mouse and human apo C-III promoters are equally functional in the context of an heterologous promoter. DISCUSSION In the present study, we report that staggerer mice lacking functional orphan nuclear receptor ROR␣ (61) have significantly reduced plasma triglyceride levels compared with wild type controls. These data indicate a physiological role of this receptor in the regulation of plasma triglyceride metabolism in mice.
Because apo C-III plays an important role in intravascular triglyceride metabolism (13), we subsequently evaluated the role of ROR␣ in the control of apo C-III expression. Our observation that the decrease in plasma triglyceride levels observed in staggerer mice is associated with a strong decrease in apo C-III plasma concentrations and hepatic as well as intestinal gene expression provides a possible mechanistic explanation of the phenotype and suggests that apo C-III is a ROR␣ target in mice. Despite the severe phenotype of staggerer mice, few ROR␣ target genes have been identified to date. To the best of our knowledge, beside rat apo A-I, apo C-III is the second mouse ROR␣ target gene identified in peripheral tissues using both molecular and physiological approaches. To evaluate whether our results obtained in mice could be extended to humans, we measured the effects of hROR␣1 overexpression on human apo C-III promoter activity by cotransfection assays in HepG2 cells. The observed activation of human apo C-III promoter activity suggests that human apo C-III could be a ROR␣ target gene in humans as well.
The demonstration that the Ϫ108/ϩ24 apo C-III promoter fragment remained sensitive to the action of hROR␣1, although slightly less than the Ϫ1415/ϩ24 fragment indicated the presence of at least one response element to hROR␣1 in this region. Two potential response elements located respectively in position Ϫ23/Ϫ18 and Ϫ82/Ϫ77 have been identified. The Ϫ23/Ϫ18 site consists of a perfect AGGTCA half-site preceded by an A/T rich region that deviates from the optimal consensus only by a C in position Ϫ1 (44). However, this sequence was shown to bind hRZR␣ in vitro, although more weakly than the optimal consensus site (46). Our results confirm these binding data on FIG. 7. hROR␣1 response elements present in the proximal ؊198/؉24 fragment of the human apo C-III promoter confer hROR␣1 responsiveness to heterologous promoters. A, HepG2 cells were cotransfected with the pCDNA3-hROR␣1 expression vector (hROR␣1; 100 ng) or the empty pCDNA3 vector as control (Cont.) and reporter constructs (50 ng) containing multiple copies of overlapping fragments (Ϫ33/Ϫ16, Ϫ58/Ϫ24, Ϫ78/Ϫ46, Ϫ83/Ϫ67, and Ϫ100/Ϫ80) of the human apo C-III promoter inserted in front of the Herpes simplex thymidine kinase promoter, cloned upstream of the luciferase reporter gene as described under "Materials and Methods." The empty TkpGL3 reporter plasmid was used as negative control. B, reporter constructs (50 ng) containing multiple copies of mutated oligonucleotides corresponding to the Ϫ33/Ϫ16 or Ϫ83/Ϫ67 fragments of the the human apo C-III promoter cloned in front of the Herpes simplex thymidine kinase promoter of the TkpGL3 reporter plasmid ((Ϫ33/Ϫ16mt) 3S TkpGL3 (Ϫ22 G3C , Ϫ21 G3A ) and (Ϫ87/Ϫ67mt) 4S TkpGL3 (Ϫ78 G3C , Ϫ79 G3A )) and their wild type homologues were cotransfected in HepG2 cells with pCDNA3-hROR␣1 expression vector (100 ng; hROR␣1) or empty pCDNA3 vector as control (Cont.) (B). Cells were transfected and luciferase activity measured and expressed as described under "Materials and Methods. "   FIG. 8. Functional comparison of putative hROR␣1 response elements present in the proximal fragment of the human and mouse apo C-III promoters. A, double-stranded oligonucleotides corresponding to the Ϫ33/Ϫ16 fragment of the human apo C-III promoter or the Ϫ33/Ϫ14 fragment of the mouse apo C-III promoter were prepared and labeled as described under "Materials and Methods." These probes were incubated as indicated with in vitro translated hROR␣1 protein or unprogrammed lysate as control. In vitro translated hROR␣1 protein or unprogrammed lysate were incubated with 10-, 50-, and 100-fold excess of the indicated unlabeled double-stranded oligonucleotides for 15 min at 4°C before labeled probes were added for 5 min at room temperature as indicated. DNA/protein complexes were resolved by nondenaturating PAGE as described under "Materials and Methods." Specific complexes not observed with unprogrammed lysate are indicated by arrows. B, HepG2 cells were cotransfected with the pCDNA3-hROR␣1 expression vector (hROR␣1; 100 ng) or the empty pCDNA3 vector as control (Cont.) and reporter constructs (50 ng) containing multiple copies of the wild type or mutated Ϫ33/Ϫ16 fragment of the human apo C-III promoter or the wild type Ϫ33/Ϫ14 fragment of the mouse apo C-III promoter ((Ϫ33/Ϫ16wt) 3as TkpGL3, (Ϫ33/ Ϫ16mt) 3as TkpGL3 (Ϫ22 G3C , Ϫ21 G3A ) and (Ϫ33/Ϫ14wt) 3as TkpGL3), respectively) inserted in front of the Herpes simplex thymidine kinase promoter and cloned upstream of the luciferase reporter gene as described under "Materials and Methods." The empty TkpGL3 reporter plasmid was used as negative control. Cells were transfected and luciferase activity measured and expressed as described under "Materials and Methods." a natural response element with hROR␣ and demonstrate that this sequence is transcriptionally active in the context of a natural promoter. The Ϫ82/Ϫ77 site also consists of a perfect AGGTCA half-site preceded by an A in position Ϫ1 and two Gs in positions Ϫ3 and Ϫ4. This larger divergence from the consensus sequence likely explains the weaker binding of hROR␣ to this sequence and the weaker transactivation of a construct containing three copies of this sequence cloned in front of a heterologous promoter. Site-directed mutagenesis of both sites confirmed that the Ϫ23/Ϫ18 half-site plays a major role in hROR␣1 responsiveness of the apo C-III promoter in HepG2 cells. The better sensitivity of the Ϫ1415/ϩ24 fragment of the apo C-III promoter, however, suggests either the presence of additional hROR␣ response elements in the upstream region of the promoter or the cooperation with other nuclear factors binding to upstream sites of the promoter that remain to be identified.
The sequence of the Ϫ33/Ϫ16 fragment of the human apo C-III promoter is almost fully conserved in the mouse promoter (67). We observed hROR␣1 binding to the Ϫ33/Ϫ14 fragment of the mouse apo C-III promoter and activation of a reporter construct containing three copies of this site cloned in front of an heterologous promoter. This suggests that ROR␣ could also act via this site on the mouse promoter.
Although elevated triglycerides likely affect atherosclerosis in humans, the effect of hypertriglyceridemia in mice is modest (69). This might explain why, despite their decreased hepatic apo C-III gene expression (63), staggerer mice fed an atherogenic diet develop more severe atherosclerosis than wild type mice. In humans, a different picture may be expected because elevated serum triglycerides are considered as an independent risk factor for coronary heart disease and because the human apo A-I promoter is unaffected by hROR␣1 (data not shown). The enhancement of human apo C-III gene promoter activity by overexpressing hROR␣1 in HepG2 cells suggests that it acts at the transcriptional level. Hence, hROR␣1 could be a valuable target for the development of hypotriglyceridemic agents.
In conclusion, the strong decrease in plasma triglyceride and apo C-III levels associated with lowered hepatic apo C-III expression observed in staggerer mice lacking functional ROR␣ identifies ROR␣ as a modulator of triglyceride levels in mice. Furthermore, the observation that human apo C-III promoter activity was also enhanced by hROR␣1 extends the mice data to humans and suggests that hROR␣1 is another modulator of human apo C-III promoter activity and, hence, a valuable target for the development of hypotriglyceridemic agents.