Cross-talk between Orphan Nuclear Hormone Receptor RZRα and Peroxisome Proliferator-activated Receptor α in Regulation of the Peroxisomal Hydratase-Dehydrogenase Gene*

The genes encoding the peroxisomal β-oxidation enzymes enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase (HD) and fatty acyl-CoA oxidase (AOx) are coordinately regulated by peroxisome proliferator-activated receptor α (PPARα)/9-cis-retinoic acid receptor (RXRα) heterodimers that transactivate these genes in a ligand-dependent manner via upstream peroxisome proliferator response elements (PPRE). Here we demonstrate that the monomeric orphan nuclear hormone receptor, RZRα, modulates PPARα/RXRα-dependent transactivation in a response-element dependent manner. Electrophoretic mobility shift analysis showed that RZRα bound specifically as a monomer to the HD-PPRE but not the AOx-PPRE. Determinants in the HD-PPRE for binding of RZRα were distinct from those required for interaction with PPARα/RXRα heterodimers. In transient transfections, RZRα stimulated ligand-mediated transactivation by PPARα from an HD-PPRE luciferase reporter in the absence of exogenously added RXRα, but did not affect PPARα-dependent transactivation of an AOx-PPRE reporter gene. These data illustrate cross-talk between the RZRα and PPARα signaling pathways at the level of the HD-PPRE in the regulation of the HD gene and characterize additional factors governing the regulation of peroxisomal β-oxidation.

Nuclear hormone receptors are a diverse group of structurally related ligand-activated transcription factors that direct the expression of target genes in response to physiological and environmental stimuli (1,2). Peroxisome proliferator-activated receptors (PPAR) 1 are members of the steroid hormone receptor superfamily that act to regulate a large number of genes involved in differentiation and lipid metabolism (3)(4)(5)(6)(7)(8)(9) in response to a variety of compounds collectively called peroxisome proliferators. Peroxisome proliferators include the fibrate family of hypolipidemic drugs, phthalate ester plasticizers, herbicides, pesticides, antidiabetic thiazolidinediones, as well as natural and synthetic fatty acids (10 -15). Transactivation of target genes by PPARs is mediated through binding to cisacting regulatory sequences called peroxisome proliferator response elements (PPRE) that consist of direct repeats of the hexameric TGACCT/C core motif. PPARs heterodimerize with the 9-cis-retinoic acid receptor, RXR␣, and bind with preference to response elements with spacing of one nucleotide between hexameric repeats (DR1) (16 -19). PPREs have been identified in the regulatory regions of a number of genes, including those encoding the first two enzymes of the peroxisomal ␤-oxidation pathway, fatty acyl-CoA oxidase (AOx) (5,20) and enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase (HD) (6,21).
In addition to PPARs, several other nuclear hormone receptors have been shown to bind to PPREs and differentially modulate PPAR function. These include chicken ovalbumin upstream promoter transcription factor (COUP-TF) (22), hepatocyte nuclear factor-4 (HNF-4) (23), and thyroid hormone receptor (TR) (24). Transcriptional regulation via PPREs is thus a net aggregate response manifested in part by the availability of PPARs and other factors that bind to PPREs, the complexity of response elements, and the interplay of PPARs with other nuclear hormone receptors and cofactors. Overlying this network is a series of corepressors and coactivators that serve to mediate receptor signaling. The overall intricacy of the system enables the integration of information from multiple signaling pathways to ensure appropriate transcriptional responses of target genes to various stimuli.
The retinoid Z receptor (RZR) family, also known as the retinoid orphan receptor (ROR) family, is a recently described group of nuclear hormone receptors shown to be involved in regulating genes responsible for cellular differentiation, the inflammatory response, and lipid metabolism (25)(26)(27)(28). RZR/ ROR target response elements have been identified in the promoters of genes for chicken ␥F-crystallin, human and rat bone sialoprotein, human 5-lipoxygenase, N-myc proto-oncogene, and apolipoprotein A-I (28 -32). Three RZR/ROR isoforms have been characterized (␣, ␤, ␥) that show differential tissue localization. The ␣ isoform includes four splice variants (ROR␣1, ROR␣2, ROR␣3, and RZR␣) and is widely expressed, RZR␤ is found primarily in brain tissues, and ROR␥ is localized to skeletal muscle (26,27,33,34). Although nuclear hormone receptors typically bind response elements as hetero-or homodimers, RZR/ROR family members are distinct in that they are able to bind to, and transactivate from, target response elements as monomers (25,26). RZR/ROR receptors recognize a TGACCT/C consensus half-site and have a demonstrated preference for sites flanked by a 6-base A/T-rich region (25,26,30,35). This consensus half-site is embedded in the HD-PPRE, and recently RZR␣ was shown to bind to this element, suggesting that RZR␣ might be a candidate regulator of peroxisomal ␤-oxidation (30). To explore this possibility, we undertook to determine how RZR␣ might interact with PPAR␣ and RXR␣ in regulating the transcription of genes encoding the first two enzymes of the peroxisomal ␤-oxidation pathway, AOx and HD. The results presented here indicate that RZR␣ binds to the HD-PPRE but not the AOx-PPRE, and selectively potentiates transactivation from the HD-PPRE in a manner dependent on the relative availability of RXR␣.

EXPERIMENTAL PROCEDURES
Plasmids-The in vitro and in vivo expression plasmids for rat PPAR␣ and human RXR␣, and the luciferase reporter constructs pCPSluc containing the minimal promoter for the gene encoding carbamoyl-phosphate synthetase, pHD(X3)luc containing two copies of the minimal HD-PPRE, and pAOx(X2)luc containing two copies of the minimal AOx-PPRE, have been described previously (6,17,36). The plasmid pM2(X3)luc, constructed by cloning the mutant HD-PPRE synthetic oligonucleotide M2 (5Ј-GATCCTCTCCTTTAAAATATTGAACTA-TTACCTACATTTGA) and its complement into the BamHI site of pCPSluc, contains three direct tandem copies of the M2 element. The expression plasmid for human RZR␣ (RZR␣/SG5) (30) was a kind gift of Carsten Carlberg (Hôpital Cantonal Universitaire de Genève, Geneva, Switzerland).
In Vitro Transcription/Translation-Transcription/translation of cDNAs encoding PPAR␣, RXR␣, and RZR␣ was performed using the TNT T7 coupled rabbit reticulocyte lysate system according to the manufacturer's protocol (Promega). Translation products labeled with L-[ 35 S]methionine were analyzed on 15% SDS-polyacrylamide gels. Synthesis of proteins for use in electrophoretic mobility shift analysis (EMSA) was carried out in parallel with unlabeled methionine.
Electrophoretic Mobility Shift Analysis-EMSA was carried out as described previously (22,36) with oligonucleotide probes corresponding to the HD-PPRE (5Ј-CCTCTCCTTTGACCTATTGAACTATTACCTAC-ATTTGA and its complement) and the AOx-PPRE (5Ј-CCTTTCCCGA-ACGTGACCTTTGTCCTGGTCCCCTTTTGCT and its complement). Underlined sequences indicate TGACCT-like direct repeats. Complementary oligonucleotides were annealed and end-labeled with the Klenow fragment of DNA polymerase I and [␣-32 P]dATP. Programmed lysate (1 to 2 l) was incubated with 0.2 pmol of labeled probe at 25°C in a final volume of 15 l containing 6 mM Hepes (pH 7.9), 120 mM NaCl,  Transient Transfections and Measurement of Luciferase Activity-Transfections were performed in African green monkey kidney BSC40 monolayer cells by the calcium phosphate method (6). Cells at 60 -80% confluence were transfected in medium without phenol red and supplemented with 5% (v/v) charcoal-stripped fetal bovine serum. Transfections were carried out with 5 g of luciferase reporter construct (pCPSluc, pHD(X3)luc, pM2(X3)luc, or pAOx(X2)luc), 2 g each of RXR␣ and PPAR␣ expression plasmids, and 0.5-4 g of RZR␣ expres-sion plasmid per 10-cm plate, as indicated. 2 g of the ␤-galactosidase expression plasmid pCH110 (Amersham Pharmacia Biotech) was included in transfections as an internal reference. Effector plasmid dosage was kept constant by addition of corresponding amounts of empty expression plasmids, and total DNA was maintained at 20 g/10-cm plate by addition of sonicated salmon sperm DNA. The peroxisome proliferator Wy-14,643 or the putative RZR␣ ligand melatonin (both in dimethyl sulfoxide) was added to fresh medium to a final concentration of 0.1 mM and 1 M, respectively, where indicated. An equivalent volume of dimethyl sulfoxide was added to control medium. Cells were harvested and lysates prepared 48 h after transfection, and luciferase activity was quantitated as described previously (6).

RZR␣ Binds as a Monomer to the HD-PPRE but Not the
AOx-PPRE-Several potential RZR␣ consensus binding sites exist within the HD-and AOx-PPREs (Fig. 1), and interaction of RZR␣ with the HD-PPRE has recently been observed (30). To explore how RZR␣ might cooperate with PPAR␣ and RXR␣ to regulate transactivation from PPREs, we first examined the DNA-binding characteristics of these nuclear hormone receptors to the HD-and AOx-PPREs. EMSA was performed with radiolabeled HD-and AOx-PPREs and in vitro translated receptors. PPAR␣ and RXR␣ formed a characteristic heterodimer that bound strongly to the HD-PPRE ( Fig. 2A). RZR␣ also bound to the HD-PPRE, apparently as a monomer based on a comparison of the mobility of the RZR␣ complex with that of the heterodimeric PPAR␣/RXR␣ complex. Coincubation of RZR␣ with either PPAR␣ or RXR␣ alone did not yield heterodimeric complexes. When all three receptors were present in the binding reaction, only heterodimeric PPAR␣/RXR␣ and monomeric RZR␣ complexes were observed, suggesting that these receptors do not co-occupy the HD-PPRE under these conditions. Therefore, RZR␣ binds as a monomer to the HD-PPRE and does not form complexes with PPAR␣ or RXR␣ on the HD-PPRE in vitro. PPAR␣/RXR␣ heterodimers readily formed a complex on the AOx-PPRE, as expected; however, RZR␣ was unable to bind to this element either alone or in combination with PPAR␣ or RXR␣ (Fig. 2B).
Competition analysis demonstrated that the interaction of RZR␣ with the HD-PPRE was specific. As shown in Fig. 3, both the RZR␣ and the PPAR␣/RXR␣ complexes were refractory to competition by nonspecific unlabeled oligonucleotide (compare lanes d-f to lanes a-c and g), whereas increasing amounts of unlabeled HD-PPRE oligonucleotide effectively competed for binding of both RZR␣ and PPAR␣/RXR␣ (compare lanes h-j to  lanes a-c and g). These results show that binding of RZR␣ is specific for the HD-PPRE.
We next examined whether PPAR␣ might modulate binding of RZR␣ to the HD-PPRE. EMSA was performed with radiolabeled HD-PPRE probe, constant amounts of RXR␣ and RZR␣, and decreasing amounts of PPAR␣. Dilution of PPAR␣ resulted in progressively decreased amounts of the PPAR␣/RXR␣ complex, as expected; however, there was no effect on formation of the RZR␣ monomer complex (Fig. 4A), suggesting that RZR␣ binding to the HD-PPRE in vitro is not influenced by PPAR␣. Similarly, increasing amounts of RZR␣ resulted in correspondingly greater amounts of the RZR␣ monomeric complex, but no change in the PPAR␣/RXR␣ complex (Fig. 4B). Additional analyses carried out using different amounts of radiolabeled HD-PPRE probe and titration of either RZR␣ (Fig. 5A) or PPAR␣/ RXR␣ (Fig. 5B) showed that PPAR␣/RXR␣ and RZR␣ did not affect each other's binding to the HD-PPRE. Therefore, RZR␣ and PPAR␣/RXR␣ bind independently to the HD-PPRE.
RZR␣ and PPAR␣/RXR␣ Bind to Different Regions of the HD-PPRE in Vitro-The above results suggest that RZR␣ recognizes distinct determinants on the HD-PPRE compared with the requirements for PPAR␣/RXR␣ interaction. The HD-PPRE is a complex response element that consists of 4 consensus TGACCT hexameric half-sites (sites I-IV; Fig. 1). To determine which elements are responsible for RZR␣ binding specificity, oligonucleotide probes containing mutations in each of the four hexameric half-sites were used in binding studies (Fig. 6). EMSA analysis showed that mutations in sites I, III, or IV did not adversely affect binding of RZR␣, whereas disruption of site II eliminated binding (Fig. 6). This result is consistent with the fact that site II most closely matches the RZR␣ consensus sequence (TGACCT/C/(A/T) 6 ). By comparison, the integrity of sites III and IV were essential for PPAR␣/RXR␣ interaction, whereas sites I and II were dispensable (Fig. 6). This latter result is consistent with previous observations that the downstream DR1 element is necessary and sufficient for PPAR␣/ RXR␣ signaling in the cell (36). Thus, RZR␣ and PPAR␣/RXR␣ binding specificity is mediated by distinct and non-overlapping determinants in the HD-PPRE.

RZR␣ Potentiates Transactivation from the HD-PPRE by PPAR␣-To define a cellular role for the observed in vitro
interaction of RZR␣ with the HD-PPRE, a series of transient transfections was performed with the HD-PPRE luciferase reporter plasmid, pHD(X3)luc, in BSC40 African monkey kidney cells. This cell line was chosen because it contains low levels of endogenous PPAR␣ and RXR␣ (6,22). Transfection of a PPAR␣ expression plasmid resulted in a 2-fold increase in luciferase activity over control transfections containing the reporter plasmid alone (Fig. 7A). Treatment of PPAR␣-expressing cells with the peroxisome proliferator and PPAR␣ ligand, Wy-14,643, increased luciferase activity an additional 4-fold. Transfection of increasing amounts of RZR␣ expression plasmid in the presence of fixed amounts of PPAR␣ expression plasmid led to a dose-dependent increase in the level of luciferase activity in cells treated with Wy-14,643, but did not affect the ligandindependent response. At the maximal level of RZR␣ plasmid used (4 g), ligand-dependent PPAR␣ activity was 23 times the basal level of activity of control cells transfected with reporter plasmid alone (and compared with the 8-fold PPAR␣, liganddependent induction observed in the absence of RZR␣ expression plasmid). RZR␣ expression plasmid alone showed at most a 1.5-fold increase in luciferase activity over basal levels of the HD-PPRE luciferase reporter (data not shown), in agreement with previous studies that demonstrated that RZR␣ only weakly activates transcription via this element (30). The modest increase observed likely reflects cooperation of RZR␣ with endogenous PPAR␣, as RZR␣ had no effect on basal expression levels of the parental pCPSluc reporter gene, which lacks a PPRE (see below). Importantly, functional cooperativity between PPAR␣ and RZR␣ in transactivation from the HD-PPRE is abrogated in the reporter construct pM2(X3)luc (Fig. 7B), in which site II of the HD-PPRE required for binding of RZR␣ has been mutated. These results show that binding of RZR␣ to the HD-PPRE is apparently required to achieve transcriptional stimulation by PPAR␣. FIG. 7. RZR␣ potentiates transactivation by PPAR␣ from the HD-PPRE but not the AOx-PPRE. BSC40 monolayer cells were transfected with 5 g of the luciferase reporter pHD(X3)luc (A), pM2(X3)luc (B), or pAOx(X2)luc (C), and expression plasmids for PPAR␣ (2 g) and RZR␣ (0.5-4 g), in the absence or presence of 0.1 mM Wy-14,643. Plasmid dosage was normalized by addition of the appropriate empty vectors, where required. Cells were harvested 48 h after transfection, and luciferase activity was quantitated, as described previously (6). All values are normalized to ␤-galactosidase activity from the cotransfected reporter plasmid pCH110 and represent the average of three independent experiments carried out in duplicate. Values from individual transfections did not vary by more than 15%. The values are presented relative to the value obtained for cells transfected with PPAR␣ in the presence of Wy-14,643 (taken as 100%). The above experiments demonstrate that RZR␣ cooperates with PPAR␣ to potentiate transactivation from the HD-PPRE in response to treatment with a peroxisome proliferator. In contrast, RZR␣ did not significantly affect PPAR␣-dependent transactivation from the AOx-PPRE, in either the absence or presence of ligand (Fig. 7C). These data are in keeping with the in vitro binding data presented above, and together the results demonstrate that RZR␣ binds specifically to the HD-PPRE and modulates transactivation by PPAR␣ from this response element.
We next investigated whether melatonin, which has been suggested to be a putative ligand for RZR␣ (31,37), could influence the functional cooperation between PPAR␣ and RZR␣ in stimulating transcription from the HD-PPRE. Under our experimental conditions, we were unable to observe any effect of melatonin on the functional cooperativity of these two receptors in stimulating transcription from the HD-PPRE (Fig. 8).
Exogenous RXR␣ Abrogates the Potentiation Effect of RZR␣ on Transactivation by PPAR␣ from the HD-PPRE-RXR␣ is an obligate heterodimerization partner for PPAR␣, and the PPAR␣dependent activation presented above presumably arises from cooperation with low levels of endogenous RXR␣ present in BSC40 cells. We therefore examined the effects of RZR␣ under conditions where RXR␣ expression plasmid was included in the transfections. Coexpression of exogenous RXR␣ and PPAR␣ increased luciferase activity 14-and 10-fold over basal levels for the HD-PPRE (Fig. 9A) and AOx-PPRE (Fig. 9B) luciferase reporter plasmids, respectively. In the presence of Wy-14,643, activity was increased 60-fold for the HD-PPRE reporter (Fig.  9A) and 20-fold (Fig. 9B) for the AOx-PPRE reporter. The robust activity obtained with coexpressed RXR␣ is consistent with the fact that the endogenous level of this receptor is limiting for PPAR␣-mediated transactivation. Interestingly, under these conditions, transactivation from the HD-and AOx-PPRE reporter plasmids was not significantly affected by addition of RZR␣ (Fig. 9, A and B, respectively). Control transfections with the parental construct pCPSluc, which lacks a PPRE, showed that the presence of PPAR␣, RXR␣, and RZR␣ did not influence the levels of relative luciferase activity observed (Fig. 9C), indicating that these receptors did not alter the basal activity of the parental reporter construct. Together, these data suggest that the stimulatory effect observed with RZR␣ on transactivation by PPAR␣ is attenuated by increasing the level of RXR␣.

DISCUSSION
Several observations have suggested a degree of interplay between the RZR/ROR and PPAR nuclear hormone receptor families in the regulation of genes, particularly those encoding proteins involved in lipid metabolism. For instance, both PPAR␣ and RZR␣ regulate transcription of the apolipoprotein A-I gene (28,38). PPAR␥ has been shown to be a critical regulator of the adipogenic program, and ROR␣ and ROR␥ mRNAs are induced early in adipogenesis (4,39). Finally, the antidiabetic thiazolidinediones, which are potent activators of PPAR␥, have recently been shown to also be specific ligands for RZR␣ (15,34).
Both the HD and AOx genes are regulated by a number of nuclear hormone receptors from several different signaling networks that converge on the respective PPREs or that directly modulate PPAR␣ activity (22)(23)(24)40). Consistent with these observations, Schrä der et al. (30) reported that RZR␣ weakly interacted with the HD-PPRE and was able to minimally activate transcription via this element (30). Our results extend this finding to show that RZR␣ can strongly potentiate transactivation by PPAR␣ from the HD-PPRE when RXR␣ levels are limiting. We found that the monomeric binding of RZR␣ to the HD-PPRE was specific and required for potentiation of transactivation by PPAR␣ from the HD-PPRE, and that RZR␣ did not bind to, or stimulate PPAR␣-dependent transactivation from, the AOx-PPRE.
The mechanism by which RZR␣ stimulates PPAR␣ activity on the HD-PPRE is not known. EMSA analysis did not show any obvious antagonizing or stabilizing effects between RZR␣ monomers and PPAR␣/RXR␣ heterodimers on the HD-PPRE in vitro. This result agrees with our findings that the RZR␣ monomer and PPAR␣/RXR␣ heterodimer target distinct and nonoverlapping hexameric determinants for binding to the HD-PPRE. Moreover, we did not observe a higher order ternary complex containing PPAR␣/RXR␣/RZR␣ on the HD-PPRE. However, this does not preclude the possibility of such a com-  Fig. 7). Values from individual transfections did not vary more than 15%. The values presented were relative to the value obtained for cells transfected with PPAR␣ and RXR␣ in the presence of Wy-14,643 (taken as 100%). C, transfection of BSC40 cells with the reporter plasmid pCPSluc in the presence or absence of PPAR␣ (2 g), RXR␣ (2 g), and RZR␣ (0.5-4 g), as indicated. Values represent the average of duplicate transfections. plex forming in vivo, which may be dependent on cooperativity or interaction with auxiliary cofactors. The involvement of auxiliary factors such as SRC-1, p300, and N-COR in transcriptional regulation by PPARs and other nuclear hormone receptors is well established (41)(42)(43). Additionally, PPAR␣, RXR␣, and/or RZR␣ might require phosphorylation or other modification not provided in the in vitro transcription/translation system. We are currently investigating whether auxiliary factors and receptor post-translational modification affect cooperative transcriptional regulation by PPAR␣, RXR␣ and RZR␣.
The HD-PPRE is a complex response element and among a select few that contain four hexameric direct repeats. In the HD-PPRE, these hexameric half-sites are organized in two tandem DR1 arrays that are separated by 2 nucleotides, an arrangement that is thought to facilitate diverse receptor interactions and thereby permit multiple levels of control. A model has been proposed in which either one or two PPAR␣/ RXR␣ heterodimers bind to the HD-PPRE to determine the state of transcriptional activation (16). As we have shown previously, the 3Ј DR1 (sites III/IV) array is essential and sufficient for PPAR␣/RXR␣ binding and activity (36); however, the arrangement that has been suggested to yield the highest level of transactivation has PPAR␣/RXR␣ heterodimers bound to both DR1 sites. Under certain conditions, a single PPAR␣/ RXR␣ heterodimer may also bind to the DR2 element (sites II/III), but this complex is in a transcriptionally inactive form. Since RZR␣ occupies site II within the HD-PPRE, this may preclude binding of PPAR␣/RXR␣ to the DR2 element and thereby favor binding of the heterodimer to the transcriptionally competent 3Ј DR1 array. RZR␣ may also contribute directly to transcriptional responses by stabilizing the PPAR␣/ RXR␣ complex in vivo or facilitating interactions with an auxiliary factor(s). This pathway of stimulation may be operative when RXR␣ is present in limiting amounts, for example in transfections carried out with PPAR␣ alone, as RZR␣ had no effect when transfections were carried out in the presence of excess exogenous RXR␣. Cotransfection of RXR␣ significantly increased the overall level of transactivation by PPAR␣, as one would expect if endogenous RXR␣ is limiting. It is possible that this level of activity is beyond a threshold level at which RZR␣ might be expected to have a stimulatory effect. Moreover, when RXR␣ is present in excess in vivo, PPAR␣/RXR␣ heterodimers may occupy both DR1s, thereby resulting in a maximal transcriptional response and preventing RZR␣ from accessing site II. This model suggests ligand-mediated regulation of the HD-PPRE is dependent on input and dynamic interplay among the PPAR␣, RXR␣, and RZR␣ signaling pathways.
In summary, our findings demonstrate that the orphan receptor RZR␣ can work in cooperation with PPAR␣ to regulate expression of the gene encoding HD, the second enzyme of the peroxisomal ␤-oxidation pathway. Transcriptional regulation of peroxisomal ␤-oxidation has proven to be a dynamic process, integrating cues from a host of signaling pathways and rapidly responding to variations in levels of key components. An understanding of the principal regulators of peroxisomal ␤-oxidation, including transcription factors controlling and modulating the expression of the genes encoding the enzymes of this pathway, may provide for the development of pharmacologic agents that specifically target the peroxisomal ␤-oxidation system as a means to influence overall lipid metabolism and homeostasis.