The Nuclear Receptors Peroxisome Proliferator-activated Receptor α and Rev-erbα Mediate the Species-specific Regulation of Apolipoprotein A-I Expression by Fibrates*

Fibrates are widely used hypolipidemic drugs which activate the nuclear peroxisome proliferator-activated receptor (PPAR) α and thereby alter the transcription of genes controlling lipoprotein metabolism. Fibrates influence plasma high density lipoprotein and its major protein, apolipoprotein (apo) A-I, in an opposite manner in man (increase) versus rodents (decrease). In the present study we studied the molecular mechanisms of this species-specific regulation of apoA-I expression by fibrates. In primary rat and human hepatocytes fenofibric acid, respectively, decreased and increased apoA-I mRNA levels. The absence of induction of rat apoA-I gene expression by fibrates is due to 3 nucleotide differences between the rat and the human apoA-I promoter A site, rendering a positive PPAR-response element in the human apoA-I promoter nonfunctional in rats. In contrast, rat, but not human, apoA-I transcription is repressed by the nuclear receptor Rev-erbα, which binds to a negative response element adjacent to the TATA box of the rat apoA-I promoter. In rats fibrates increase liver Rev-erbα mRNA levels >10-fold. In conclusion, the opposite regulation of rat and human apoA-I gene expression by fibrates is linked to differences in cis-elements in their respective promoters leading to repression by Rev-erbα of rat apoA-I and activation by PPARα of human apoA-I. Finally, Rev-erbα is identified as a novel fibrate target gene, suggesting a role for this nuclear receptor in lipid and lipoprotein metabolism.

Fibrates are a class of drugs used in the treatment of atherogenic dyslipidemia (1). Fibrates are ligands for transcription factors belonging to the peroxisome proliferator-activated receptor (PPAR) 1 subfamily of nuclear receptors (2,3). After ligand activation and heterodimerization to the 9-cis retinoic acid receptor RXR, PPARs bind to specific response elements, termed peroxisome proliferator response elements (PPREs), in the regulatory regions of target genes. PPREs consist of a direct repeat of the degenerated hexamer AGGTCA sequence separated by 1 nucleotide (DR-1). Fibrates exert their effects on plasma lipids by altering the transcription of genes involved in lipoprotein metabolism (1). Direct evidence that fibrate action on lipoprotein metabolism is mediated by PPAR␣, the principal PPAR form in liver, came from studies in mice rendered deficient for PPAR␣ by homologous recombination (4). Functional PPREs have been identified in the promoters of a number of genes involved in plasma triglyceride (e.g. human apoC-III (5) and mouse and human lipoprotein lipase (LPL) (6)) and high density lipoprotein (HDL) metabolism (e.g. human apoA-I (7) and apoA-II (8)).
Whereas both in man and rodents fibrates efficiently decrease plasma triglycerides, their effects on HDL occur in an opposite manner in these species. Whereas in man plasma HDL cholesterol consistently increases upon fibrate treatment (9), a pronounced decrease is observed in rats (10). Previously, it was demonstrated that these species differences in HDL response to fibrates are associated with opposite changes of apoA-I expression, the major HDL apolipoprotein. In rats, fibrate treatment lowers hepatic, but not intestinal, apoA-I mRNA levels, due to a decreased transcription rate of the apoA-I gene in liver (10). In contrast, the transcription rate of the human apoA-I gene is induced by PPAR␣, which interacts with a positive PPRE located in the A site of the human apoA-I gene promoter liver-specific enhancer (7). Furthermore, fibrates enhance apoA-I expression and production both in vivo in man and in vitro in human hepatoma cells (11)(12)(13)(14). The observation that fibrates regulate human and mouse apoA-I gene transcription in an opposite manner in livers of transgenic mice overexpressing the human apoA-I gene driven by its homologous promoter including the PPRE sequence (15) suggested that the species-specific differences in fibrate response are due to cis-element sequence differences between the rodent and human apoA-I genes. However, the exact molecular mechanisms implicated have not been determined yet. In addition, although the mechanisms of positive regulation of gene expression by fibrates via PPAR interacting with positive PPRE sequences are well understood (for review, see Schoonjans et al. (16)), little is known on the mechanisms of negative gene regulation by fibrates and PPARs.
In this study, we determined the molecular mechanisms behind the differential regulation of rat versus human apoA-I gene expression by fibrates. Our results demonstrate that this opposite regulation is due to sequence differences of two distinct enhancer regions in the rat and human apoA-I genes. On the one hand, the human apoA-I gene promoter A site contains a positive PPRE to which PPAR␣ binds. The lack of induction of apoA-I gene transcription by fibrates in rats is due to 3 * This research was supported by the Région Nord-Pas de Calais. 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.
§ Both authors contributed equally to this article.  1 The abbreviations used are: PPAR, peroxisome proliferator-activated receptor; PPRE, peroxisome proliferator response element; RXR, 9-cis retinoic acid receptor; RevRE, Rev-erb␣ response element; TK, thymidine kinase; HDL, high density lipoprotein; TR, thyroid hormone receptor; EMSA, electrophoretic mobility shift assays; TRE, TR response element; CAT, chloramphenicol acetyltransferase. nucleotide differences between the rat and the human apoA-I gene promoter A sites, resulting in the transformation of the positive PPRE in the human apoA-I promoter in a nonfunctional site in rats. On the other hand, rat, but not human, apoA-I gene transcription is repressed via an element overlapping the TATA box to which the nuclear orphan receptor Rev-erb␣ binds. Furthermore, fibrates induce Rev-erb␣ and repress apoA-I gene expression only in rat liver, but not in intestine. Interestingly, the rat apoA-I promoter Rev-erb␣ response element (RevRE) is not conserved in the human apoA-I gene promoter.

EXPERIMENTAL PROCEDURES
Materials-Fenofibric acid and fenofibrate, and ciprofibrate were kind gifts of Dr. Alan Edgar (Laboratoires Fournier, Daix, France) and Dr. Herbert (Sanofi, Toulouse, France) respectively. Bezafibrate and hematoxylin were from Sigma.
Animal Studies-Adult Sprague-Dawley rats (3 months old) were treated for 14 days with the specified fibrates at the indicated concentrations (w/w) mixed in chow. At the end of the treatment period all animals were fasted overnight, weighed, and sacrificed by exsanguination under ether anesthesia between 9 and 11 a.m. Livers were frozen in liquid N 2 or stored in phosphate-buffered saline until RNA or in situ hybridization analysis.
Cell Culture-Human hepatoma HepG2 cells (E.C.A.C.C., Porton Down, Salisbury, UK) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum at 37°C in a humidified atmosphere of 5% CO 2 , 95% air. Medium was changed every other day. Human and rat hepatocytes were isolated by collagenase perfusion and cultured as described previously (17). At the end of the treatment period with fenofibric acid (in Me 2 SO), cells were washed three times with ice-cold phosphate-buffered saline, homogenized in 4 M guanidinium isothiocyanate, and used for RNA analysis.
Transient Transfections-Transfections in HepG2 cells were performed at 50 -60% confluency by the calcium phosphate coprecipitation procedure with a mixture of plasmids that contained in addition to the reporter (1-5 g/60 mm culture dish) and expression vector(s) (100 -1000 ng), 0.35 g of cytomegalovirus-␤-galactosidase expression vector as a control for transfection efficiency. All samples were complemented to an equal total amount of DNA. After 4 h cells were washed with phosphate-buffered saline and incubated for another 16 h with the indicated stimuli in fresh medium containing 5% calf serum delipoproteinized by ultracentrifugation in KBr (1.21 g/ml) and subsequently treated with AG-1-X8 resin (Bio-Rad) plus activated charcoal. Stimuli were dissolved in Me 2 SO and added to the medium at the indicated concentrations and periods of time. Control cells received vehicle only. CAT activity was determined on cell extracts as described by Gorman et al. (25). Autoradiographs were quantified by liquid scintillation counting and results were normalized for transfection efficiency. Transfection experiments were performed at least three times.
Electrophoretic Mobility Shift Assays (EMSA)-hPPAR␣, mRXR␣, and hReverb␣ proteins were synthesized in vitro using the rabbit reticulocyte lysate system (Promega). EMSAs with Rev-erb␣, PPAR and/or RXR were performed as described previously (7,26). For competition experiments, increasing amounts of the indicated cold oligonu-cleotides were included just before adding labeled oligonucleotide. After 15 min of incubation at room temperature, DNA-protein complexes were separated by electrophoresis on a 4% polyacrylamide gel in 0.25ϫ Tris borate-EDTA buffer at 4°C.

ApoA-I Gene Expression Is Regulated in an Opposite Manner by Fenofibrate in Human Versus
Rat Hepatocytes-To study whether the species differences in response to fibrates are due to a distinct regulation of apoA-I expression in liver, the effects of fenofibric acid, the active form of fenofibrate, on apoA-I gene expression were compared between primary cultures of human versus rat hepatocytes. Addition of fenofibric acid to primary human hepatocytes, at a concentration (100 M) within the range of plasma concentrations attained in fenofibrate-treated patients, resulted in a Ͼ2-fold increase of apoA-I mRNA levels ( Fig. 1A). In contrast, in rat hepatocytes fenofibric acid decreased apoA-I mRNA levels in a dose-dependent manner (Fig.  1B). As a control, mRNA levels of the ribosomal protein 36B4 did not change upon treatment of either human or rat hepatocytes with fenofibric acid (Fig. 1). These data indicate that the species-distinct regulation of the apoA-I gene is due to a direct effect of fenofibrate on the hepatocyte.
The Human, but Not the Rat, ApoA-I Promoter A Site Contains a Positive PPAR-response Element-Since previously a positive PPRE has been identified in the human apoA-I promoter A site (7), we wondered whether the lack of induction of apoA-I gene expression by fibrates in rats could be due to the absence of a functional PPRE in the rat apoA-I promoter. Interestingly, when the human and rat apoA-I A sites, which are composed of 3 degenerated hexameric half-sites separated by 2 and 1 nucleotides, respectively (DR-2 and DR-1), were aligned, 3 nucleotide differences were observed ( Fig. 2A). To determine whether these nucleotide differences conferred a different PPAR binding potential to the rat and human apoA-I A sites, EMSA experiments were performed. Incubation of radiolabeled human A site oligonucleotide with in vitro produced PPAR protein, in the presence of its heterodimeric partner RXR, resulted in the formation of a retarded complex (Fig. 2B,  lane 3). This complex was specific since it could be competed out by adding excess unlabeled human A site oligonucleotide ( Fig. 2, B, lane 4, and C, lanes 7-10). In contrast, unlabeled rat A site oligonucleotide did not compete for binding to the human A site (Fig. 2, B, lane 5, and C, lanes [11][12][13][14]. Next, the human and rat A sites were cloned in front of a heterologous promoter, and transient co-transfection experiments were performed in HepG2 cells. PPAR co-transfection resulted in a significant increase of human A site driven CAT activity (Fig. 3). Due to the high basal activity of PPAR in HepG2 cells, possibly due to the activation by endogenous PPAR ligands, no further activation was observed in the presence of fenofibric acid (Fig. 3). However, activation by fibrates was observed when the same experiments were performed in other cells, such as in HeLa or CV-1 cells (not shown). In contrast, rat apoA-I A site-driven CAT activity was not increased by PPAR, fenofibric acid, or both together (Fig. 3). These data clearly indicate that only the human, but not the rat, apoA-I A site functions as a PPRE.
The Orphan Receptor Rev-erb␣ Represses Transcription of the Rat ApoA-I Promoter-In a previous report, the nuclear receptor ROR␣1 was identified as a positive regulator of apoA-I expression in the intestine (28). Interestingly, nuclear receptors of the ROR and Rev-erb groups regulate transcription through identical response elements, leading to cross-talk between the positive ROR and the negative Rev-erb nuclear receptors (29 -32). Therefore, it was tested whether Rev-erb␣, which is expressed in liver (18,33), could act as a repressor of rat apoA-I transcription. Co-transfection of CAT reporter constructs driven by the rat apoA-I promoter with a Rev-erb␣ expression vector resulted in dose-dependent repression of CAT activity in HepG2 cells (Fig. 4, A and B). Interestingly, human apoA-I promoter activity was not repressed by Rev-erb␣ cotransfection (Fig. 4A). These results indicate that the rat, but not the human, apoA-I gene is a Rev-erb␣ target gene.
To determine whether this repression of apoA-I promoter activity is mediated by the previously identified response element for ROR␣1 (28) located adjacent to the TATA box, the activity of Rev-erb␣ was analyzed on an apoA-I promoter construct carrying a mutation in the ROR␣1 response element (see Fig. 6A). Whereas wild-type apoA-I promoter activity was significantly repressed in a dose-dependent manner, Rev-erb␣ failed to repress the activity of the mutated construct (Fig. 4C), thereby indicating that the Rev-erb␣ and ROR␣ response elements coincide.
Rev-erb␣ Represses Transcription of the Rat ApoA-I Promoter through Both Active and Passive Mechanisms-Since Rev-erb␣ may repress transcription either actively, by recruiting nuclear co-repressors of the N-CoR family (34,35), or passively, by competing for binding with other, positive factors, such as the proteins binding to the TATA box, the activity of a C terminustruncated Rev-erb␣236, which can still bind to DNA, but does not actively repress transcription (36), was tested on rat apoA-I promoter transcription. In contrast to wild-type Rev-erb␣, which repressed transcription already at low concentrations, Rev-erb␣236 started to repress apoA-I transcription only at higher amounts of expression vector (Fig. 5A).
To determine whether Rev-erb␣ could repress transcription from a heterologous promoter, the rat apoA-I TATA site (AI-TATA) was cloned in front of the TK promoter, and transient transfection assays were performed. Compared with the empty expression vector, co-transfection of Rev-erb␣ resulted in a significant decrease of wild-type rat AI-TATA driven CAT activity (Fig. 5B). In contrast, neither mutated AI-TATA nor TK promoter alone driven CAT activity were repressed by Rev-erb␣ co-transfection (Fig. 5B). Altogether, these observations suggest that the repression of rat apoA-I transcription by Rev-erb␣ is likely due to a combination of active and passive transcriptional repression mechanisms.
Rev-erb␣ Binds to the Rat, but Not the Human, ApoA-I Promoter TATA Site-EMSAs were performed to determine whether Rev-erb␣ could bind to the rat AI-TATA site. Incubation of labeled rat wild-type AI-TATA oligonucleotide with Rev-erb␣ resulted in the formation of a retarded complex (Fig. 6, A  and B, lane 2). The binding was specific since it could be competed out by excess amounts of either unlabeled consensus monomeric RevRE (30,37) or rAI-TATA wt oligonucleotides (Fig. 6B, lanes 3-6). In contrast, neither rAI-TATA mt nor human AI-TATA oligonucleotides could compete for binding of Rev-erb␣ to rAI-TATA wt (Fig. 6, lanes 7-10). Similar results were obtained using the consensus RevRE oligonucleotide as a probe (Fig. 6C). Similar as cold RevRE oligonucleotide (Fig. 6C, The human (hAIA wt ) and rat (rAIA wt ) apoA-I A sites were cloned upstream of the TK promoter, and HepG2 cells were transfected with the indicated plasmids (5 g) in the presence of co-transfected mPPAR␣ or pSG5 vector plasmids (1 g). Cells were treated with fenofibric acid (FF) (500 M) or vehicle, and CAT activity was measured and expressed as mean Ϯ S.D.
Fibrates Induce Rev-erb␣ mRNA Levels in Rat Liver-To determine whether the negative regulation of rat apoA-I gene expression by fibrates is mediated via Rev-erb␣, adult rats were treated for 14 days with different doses of fenofibrate, and Rev-erb␣ mRNA levels were measured by Northern blot analysis in livers and intestines, the major sites of apoA-I production. Administration of fenofibrate resulted in a pronounced induction of Rev-erb␣ gene expression (Fig. 7A). This induction of Rev-erb␣ mRNA was dose-dependent, being evident at a dose level of 0.05% (w/w) of fenofibrate and further enhanced at 0.5%. Interestingly, the induction of Rev-erb␣ mRNA levels was inversely proportional to the decrease of apoA-I mRNA levels in liver (Fig. 7A). In contrast to liver, intestinal Rev-erb␣ mRNA levels did not change significantly after fenofibrate treatment (control, 100 Ϯ 19%; fenofibrate 0.005%, 90 Ϯ 3%; fenofibrate 0.05%, 110 Ϯ 10%; fenofibrate 0.5%, 144 Ϯ 5%), which correlates well with the absence of effects of fibrates on intestinal apoA-I mRNA levels (10).
To determine whether the induction of liver Rev-erb␣ gene expression is a general property of fibrates, rats were treated for 14 days with different fibrates. All fibrates tested significantly increased Rev-erb␣ mRNA levels, which was most pro- FIG. 4. Rev-erb␣ represses transcription of the rat, but not the human, apoA-I promoter. A, HepG2 cells were transfected with the rat (r(Ϫ256/ϩ91)AI-CAT) or human (h(Ϫ256/ϩ26)AI-CAT) apoA-I promoter reporter plasmids (2 g) in the presence of co-transfected hRev-erb␣ plasmid (1 g). B, HepG2 cells were transfected with the indicated rat apoA-I promoter reporter plasmids (2 g) in the presence of increasing amounts (0, 100, 500, and 1000 ng) of co-transfected hRev-erb␣ plasmid. C, HepG2 cells were transfected with the wild-type (r(Ϫ252/ϩ26) wt AI-CAT) or mutated (r(Ϫ252/ϩ26) wt AI-CAT) rat apoA-I promoter constructs (2 g) in the presence of increasing amounts (0, 100, 250, 500, and 1000 ng) of co-transfected hRev-erb␣ plasmid. Empty pSG5 vector was added to an equal amount of plasmid DNA. CAT activities were measured after 24 h and expressed as mean Ϯ S.D.

FIG. 5. Repression of the rat apoA-I gene promoter by Rev-erb␣ is due to both active and passive repression mechanisms.
A, HepG2 cells were transfected with the r(Ϫ252/ϩ26)AI-CAT reporter plasmid (2 g) in the presence of increasing amounts (0, 500, and 1000 ng) of either the wild-type (Rev-erb␣) or truncated (Rev-erb␣236) Rev-erb␣ expression plasmids. B, HepG2 cells were transfected with the rat wild type (rAI-TATA wt -TK-CAT) and mutated (rAI-TATA mt -TK-CAT) apoA-I TATA site-containing TK promoter plasmids (2 g) in the presence of increasing amounts (0, 250, 500, and 1000 ng) of Rev-erb␣ expression plasmid. Empty pSG5 vector was added to an equal amount of plasmid DNA.
Finally, the induction of Rev-erb␣ gene expression by fibrates was also observed by in situ hybridization experiments using liver slices from rats treated with different fibrates (data not shown). Compared with controls, hybridization with an antisense Rev-erb␣ riboprobe resulted in an enhanced signal intensity in livers from fibrate-treated rats, which was evenly distributed throughout the liver, indicating that Rev-erb␣ mRNA levels increase after fibrate treatment in the apoA-Iproducing parenchymal cells of the liver. DISCUSSION Fibrates are hypolipidemic drugs, which have been discovered by virtue of their cholesterol-lowering activity in rats (for review, see Staels and Auwerx (1)). Ironically, this cholesterol lowering is entirely due to a decrease of plasma HDL cholesterol concentrations and its major apolipoprotein, apoA-I (10), an effect nowadays considered to be negative, due to the protective activities of HDL and apoA-I against atherosclerosis (38,39). In contrast, in man fibrates consistently elevate plasma HDL and apoA-I and increase the production rate of its major apolipoproteins, apoA-I and apoA-II (8,13,40). In the present study, we addressed the mechanisms of the differential response of apoA-I expression to fibrates in man and rat. Using primary human and rat hepatocytes we demonstrate that the opposite action of fenofibrate is due to a direct effect on the hepatocyte. These data are in line with a previous report showing opposite regulation of human (increase) versus mouse (decrease) apoA-I gene expression by fibrates in livers of transgenic mice overexpressing the human apoA-I gene under control of its homologous promoter (15). The experiments in these mice also indicated that this species-specific opposite regulation between human versus mouse is due to differences in cis-, but not in trans-acting elements (15). The results presented in this study indicate that the differences in fibrate response between the human and rat apoA-I genes are due to a combination of two distinct mechanisms implicating two different cis-elements binding two distinct nuclear receptors, PPAR␣ and Rev-erb␣ (Fig. 8).
On the one hand, the absence of induction of apoA-I gene expression by fibrates in rats is linked to 3-nucleotide differ-ences between the rat and human apoA-I promoter A sites, thereby rendering the positive PPRE in the human apoA-I promoter nonfunctional in rats. Since in vitro promoter mapping studies identified a positive PPRE in the human apoA-I promoter A site (7), and since human apoA-I gene expression is induced at the transcriptional level by fibrate treatment in livers of transgenic mice (15) overexpressing the human apoA-I gene under control of its homologous promoter including this PPRE, it is highly likely that this element indeed confers the positive response of the human apoA-I gene to fibrates. The demonstration that this element is not functional in the rat apoA-I A site would therefore explain the lack of induction of rat apoA-I gene expression by fibrates.
On the other hand, a negative site for the nuclear receptor Rev-erb␣ is present in the rat, but not in the human, apoA-I promoter, and fibrate treatment results in a pronounced increase of liver expression of the Rev-erb␣ gene. Together with Rev-erb␤ (also termed RVR, BD73) and the Drosophila receptor E75, Rev-erb␣ (also termed ear-1) belongs to a subfamily of orphan receptors, which repress transcription of target genes (41). Rev-erb␣ appears to be ubiquitously expressed (18,33), but its functions are ill defined. A number of indirect observations suggested a role for Rev-erb␣ in metabolic control and energy homeostasis. First, Rev-erb␣ expression is induced during differentiation of preadipocytes into adipocytes (42). Second, Rev-erb␣ is transcribed from the opposite strand of the thyroid hormone receptor (TR) ␣ gene, with its 3Ј exons overlapping with TR␣2, a dominant negative regulator of TR action, suggesting a role for Rev-erb␣ as a modulator of thyroid hormone signaling (18,33,43,44). Indeed, the level of Rev-erb␣ expression appears to be a determinant of the ratio of TR␣1: TR␣2 mRNA levels (42,45,46). In addition, Rev-erb␣ has been suggested to bind to a subset of TR response elements (TREs), composed of DR-4 sequences (47), as well as to compete with TR for binding to TREs (48). Finally, overexpression of Rev-erb␣ in mouse myoblasts leads to inhibition of muscle differentiation, possibly due to interference with thyroid hormone-induced muscle differentiation pathways (48). Interestingly, a significant level of cross-talk exists also between the peroxisome proliferator and thyroid hormone signaling pathways. Indeed, in rodents peroxisome proliferators and thyroid hormones often regulate the same target genes (49). In addition, certain of these genes have been shown to carry both TREs and PPREs (50,51). Cross-talk between PPAR and TR may furthermore occur by competition for binding to DNA (52,53) or for the common heterodimeric partner RXR (53)(54)(55)(56), or by formation of PPAR:TR heterodimers, as has been suggested by one report (57). Our data identifying Rev-erb␣ as a fibrate target gene suggests another layer of cross-talk between the TR and PPAR signaling pathways. Indeed it is conceivable that the induction of Rev-erb␣ expression after fibrates results in decreased TR␣2 mRNA levels, resulting both in a higher level of expression of functionally active TR␣1 as well as lower levels of the dominant negative TR␣2 protein, altogether leading to enhanced thyroid hormone action. Studies to test this hypothesis are currently underway in our laboratory.
The identification of the rat apoA-I gene, a major determinant of HDL metabolism, as a target for Rev-erb␣, is the first demonstration of a role for this nuclear receptor in lipoprotein metabolism, albeit in rats. Although this site is not conserved in the human apoA-I gene, the fact that, both in rats (this study) and in man, 2 Rev-erb␣ expression is controlled by fibrates, which are major regulators of lipoprotein metabolism, points to a role for this nuclear receptor in lipoprotein metabolism and possibly atherogenesis. It will therefore be of interest to determine which human genes are regulated by Rev-erb␣.
Rev-erb family members bind in vitro either to a monomeric half-site, to a subset of DR-2 (termed RevDR2) or to DR-4 TRE sequences (29 -31, 36, 37, 47, 58, 59). In addition to the human Rev-erb␣ gene, which is negatively autoregulated by Rev-erb␣ binding to a RevDR2 sequence (36), the rat apoA-I gene is only the second natural Rev-erb␣ target gene identified to our knowledge. The Rev-erb␣ response element in the rat apoA-I promoter is composed of a AGGTCA half-site preceded by an AT-rich region consisting of the TATA box. Previously, it was suggested that Rev-erb␣ represses transcription only from RevDR2, but not from monomeric sites (31). Our results demonstrating that Rev-erb␣ represses transcription via a monomeric site, both in the context of the homologous as well as a heterologous promoter, suggest that Rev-erb␣ is also active on monomeric sites. Similarly, Rev-erb␤ actively represses transcription of the N-myc gene via a monomeric sequence (32). Active transcription repression by Rev-erb␣ may occur by recruting nuclear receptor corepressors of the N-CoR family, which interact with the C terminus of Rev-erb␣ (34,35). However, the truncated Rev-erb␣236 form (36), which lacks the the N-CoR binding domains, also represses transcription, albeit to a smaller extent than wild-type Rev-erb␣, suggesting that the repression of apoA-I transcription by Rev-erb␣ is due to a combination of both active and passive mechanisms. Passive transcriptional repression may occur by competition for binding of either TATA-box binding protein (TBP) to the TATA box or for positive transcription factors to the monomeric site.
Several lines of evidence point to a functional role for Rev-erb␣ in the transcriptional repression of rat apoA-I gene expression by fibrates. First, the repression of rat apoA-I expression by fibrates occurs at the transcriptional level (10). Second and in contrast to the induction by fibrates of direct PPARtarget genes, such as acyl-CoA oxidase, the repression of apoA-I gene expression is a relatively slow process requiring between 12 and 24 h, pointing to an indirect action of fibrates (60). Third, the Rev-erb␣ response element is only present in the rat, but not in the human, apoA-I promoter, which correlates well with the difference in response of human versus rat apoA-I gene expression to fibrates. Finally, Rev-erb␣ gene expression is only induced in the liver, but not in the intestine, and fibrates repress apoA-I gene expression only in rat liver, but not in intestine (10).
In addition to the rat apoA-I gene, fibrates have been shown to repress the transcription of a wide array of genes involved in lipid and energy metabolism, such as rat S14 (56), rat apoA-IV (61), rat lecithin:cholesterol acyltransferase (62), rat hepatic lipase (63), and the human, rat, and mouse apoC-III (4, 5, 64, 65) genes in liver. Furthermore, the antidiabetic thiazolidinediones, which are ligands for PPAR␥, have been shown to repress the transcription of the human and rat leptin genes in adipose tissue (66 -69). Although the mechanism of positive regulation by fibrates via PPAR interacting with PPRE sequences in target gene promoters has been fairly well studied, relatively little is known about the mechanisms of negative transcriptional regulation by these agents. Since both apoC-III and apoA-I gene expression are increased in livers of PPAR␣ knock-out mice and since their expression is not repressed by fibrate treatment in the same mice (4), PPAR␣ expression appears to be required for the negative regulation by fibrates, at least with respect to the mouse apoA-I and apoC-III genes. A previous study indicated that fibrates may repress transcription of the human and rat apoC-III and rat transferrin genes in liver by down-regulating the expression of the strong positive transcription factor HNF-4 as well as by displacement of HNF-4 binding by non-productive PPAR/RXR heterodimers (5,70). It is unlikely, however, that the repression of rat apoA-I gene transcription occurs through this mechanism. Indeed, we could not confirm any significant regulation of HNF-4 gene expression by fibrates, whereas both Rev-erb␣ and apoA-I mRNA levels changed much more pronouncedly in the livers of animals treated with all fibrates tested. Moreover, whereas HNF-4 is an important regulator of hepatic apoC-III gene transcription, this transcription factor appears of lesser importance for human apoA-I transcriptional regulation in HepG2 cells (71,72). In addition, the role of HNF-4 in controlling rat apoA-I gene transcription is unclear, since HNF-4 may inhibit and activate its transcription via the A and C sites, respectively (73,74). Finally, down-regulation of HNF-4 expression and competition for its binding would be expected to affect a wide array of genes in liver whose expression is controlled by HNF-4, such as human apoA-II and apoB (75). However, whereas apoB expression is unaffected by fibrates (10), human apoA-II gene expression is induced by fibrates via PPAR:RXR binding to the HNF-4 site (8). It is tempting to speculate that the induction of Rev-erb␣ expression by fibrates may be a general mechanism leading to the transcriptional repression of a certain number of genes. For instance, since Rev-erb␣ increases upon adipocyte differentiation (42), it will be of interest to determine whether this nuclear receptor also regulates the transcription of a number of genes in adipose tissue, such as possibly the leptin gene.
In conclusion, the species-distinct regulation of apoA-I gene expression by fibrates is linked to sequence differences in cisacting elements (see Fig. 8 for overview). In man, apoA-I transcription is induced via PPAR␣ binding to a positive PPRE. This site is not conserved in rats resulting in the lack of binding of PPAR to the rat apoA-I promoter. In contrast, rat apoA-I transcription is repressed by Rev-erb␣, whose expression is induced by fibrates and which binds to a RevRE site adjacent to the TATA box in the rat, but not in the human apoA-I promoter. These observations indicate that the expression of the transcription factor Rev-erb␣ may be under control not only by pharmacological, but also by dietary factors (such as fatty acids) as well as inflammatory cytokines (such as PG-J2 and LT-B4), which all bind to and activate PPARs (2, 3, 76 -78). Altogether these observations suggest a role for the nuclear orphan receptor Rev-erb␣ in metabolic control.