The orphan nuclear hormone receptor LXR alpha interacts with the peroxisome proliferator-activated receptor and inhibits peroxisome proliferator signaling.

The yeast two-hybrid system was used to isolate novel cellular factors that interact with the mouse peroxisome proliferator-activated receptor alpha (PPARalpha). One of the interacting clones isolated encoded LXRalpha, a recently described human orphan nuclear hormone receptor. LXRalpha bound directly to PPARalpha, as well as to the common heterodimerization partner 9-cis-retinoic acid receptor (RXRalpha). LXRalpha did not form a DNA binding complex with PPARalpha on synthetic hormone response elements composed of direct repeats of the TGACCT consensus half-site or on naturally occurring peroxisome proliferator response elements (PPREs) or LXRalpha response elements. However, LXRalpha inhibited binding of PPARalpha/RXRalpha heterodimers to PPREs, and coexpression of LXRalpha in mammalian cells antagonized peroxisome proliferator signaling mediated by PPARalpha/RXRalpha in vivo. These findings identify a novel partner for PPARalpha and suggest that LXRalpha plays a role in modulating PPAR-signaling pathways in the cell.

cently described members of the ligand-activated nuclear hormone receptor superfamily, which includes receptors for steroids, vitamin D, and thyroid and retinoid hormones (1). PPARs have been shown to regulate a broad spectrum of genes involved in lipid metabolism, cellular growth, and differentiation (2). Consequently, there is a great deal of interest in understanding their specificity and mechanisms of action. PPARs were originally identified as factors that mediate transcriptional responses to peroxisome proliferators, a broad class of xenobiotic chemicals that include fibrate hypolipidemic drugs and other nongenotoxic rodent hepatocarcinogens (3,4). Subsequently, PPARs were shown to be differentially activated by a variety of long chain fatty acids and lipid-like compounds (5), suggesting that fatty acids or fatty acid derivatives serve as physiological activators. PPARs exist in a variety of pharmacologically distinct subtypes and isoforms that are differentially expressed and which mediate distinct patterns of tissuespecific gene expression (4, 6 -8). For example, mouse (m) PPAR␥ triggers adipogenesis in cultured cells (8) and is selectively activated by 15-deoxy-⌬ 12,14 -prostaglandin J2, a recently identified high affinity ligand of this PPAR subtype (9,10).
PPARs activate expression of target genes by recognizing peroxisome proliferator response elements (PPREs) composed of TGACCT-related direct repeats that are spaced by one nucleotide (DR1) (11,12). Specific DNA binding is manifested through heterodimerization with the 9-cis-retinoic acid receptor, RXR␣ (13,14), another member of the nuclear hormone receptor superfamily that also serves as a heterodimerization partner for thyroid hormone, retinoic acid, and vitamin D receptors (1). The involvement of PPARs in multiple and diverse cellular functions suggests that these receptors may be integrated with other cellular signaling pathways, in addition to the well characterized RXR␣ pathway. Indeed, the reciprocal modulation of thyroid hormone and peroxisome proliferatorresponsive genes through cross-talk between thyroid hormone receptors and PPARs has recently been demonstrated (15)(16)(17). Moreover, it has been reported that rat PPAR␣ heterodimerizes with the thyroid hormone receptor (18), although this conclusion remains controversial (15). Unraveling the pleiotropic functions of PPARs requires identification of the full spectrum of factors that interact with PPARs. In this report, we used the yeast two-hybrid system (19) to isolate novel factors that interact with mPPAR␣. One mPPAR␣-interacting factor isolated was identified as LXR␣, a recently described human orphan nuclear hormone receptor that appears to be involved in a novel retinoid signaling pathway (20). LXR␣ inhibited the binding of mPPAR␣/RXR␣ to PPREs in vitro and antagonized transcriptional activation by mPPAR␣ in vivo. Our findings demonstrate that nuclear receptors other than RXR␣ bind directly to PPAR␣ and may play a role in modulating the cellular functions of this receptor.

MATERIALS AND METHODS
Two-hybrid Library Screening-Yeast two-hybrid vectors expressing full-length human RXR␣ and mPPAR␣ as fusions to the GAL4 DNAbinding domain (GBD-RXR␣ and GBD-mPPAR, respectively) and RXR␣ fused to the GAL4 activation domain (GAD-RXR␣) have been described (21). GBD-mPPAR was not suitable for two-hybrid library screening, since it induced a low level of constitutive activity of the ␤-galactosidase reporter gene when expressed alone in yeast (21). We therefore constructed a modified vector, GBD-N⌬mPPAR (missing the amino-terminal 83 amino acid codons of mPPAR␣) that was devoid of this intrinsic activation function but remained capable of interacting with RXR␣ (see Table I). This derivative was used as bait in the two-hybrid system. Saccharomyces cerevisiae strain HF7c (MATa, ura3-52, his3-200, lys2-801, ade2-101, trp1-901, leu2-3, 112, gal4 -542, gal80-538, LYS::GAL1 UAS -GAL1 TATA -HIS3, URA3::GAL4 (17-mers)x3 -CyC1 TATA -lacZ) harboring GBD-N⌬mPPAR was subjected to electroporation (22) with a HeLa cell cDNA library constructed in the GAD vector pGADGH (Clontech). Transformants were plated onto synthetic complete media plates lacking histidine, leucine, and tryptophan and containing 25 mM 3-amino-1,2,4-triazole. His ϩ , leu ϩ , trp ϩ colonies were assayed for expression of the ␤-galactosidase reporter gene by agarose overlay assay using 0.2% (v/v) Triton X-100 as a permeabilization agent. Library plasmids were rescued by electroporation into Escherichia coli ElectroMAX DH10B (Life Technologies, Inc.), and isolated plasmids were used to retransform S. cerevisiae HF7c. Candidate clones were tested for interaction against GBD-N⌬mPPAR and to GBD-RXR␣ and also to irrelevant fusion proteins and the empty fusion vector so as to eliminate false positives (23) and to ensure that the his ϩ and ␤-galactosidase phenotypes were dependent on the presence of both the respective GAD-cDNA library vectors and GBD-N⌬mPPAR. Positive clones were recovered and subjected to partial 5Ј-and 3Ј-sequence analysis. Two clones contained the same partial cDNA that encoded a novel member of the nuclear hormone receptor superfamily, subsequently shown to be nearly identical to the recently described human LXR␣ (20). One of these clones (GAD-SM1; encoding amino acid residues 61-447 of LXR␣) was selected for further analysis. The remaining positive clones were not related to LXR␣ or to other nuclear receptors and will be the focus of future studies.
Cloning a Full-length LXR␣ cDNA-The 1.2-kilobase pair (kbp) insert of GAD-SM1 was labeled with [␣-32 P]dATP by random priming and used to probe, under high stringency conditions, a 5Ј-stretch gt11 human liver cDNA library (Stratagene). The largest hybridizing clone contained a 1.7-kbp insert. The cDNA was sequenced in both directions and shown to contain the entire 447-amino acid long open reading frame corresponding to the published sequence of LXR␣ (20).
In Vitro Transcription/Translation-In vitro expression vectors for mPPAR␣ and human RXR␣ have been described (14). The entire open reading frame of LXR␣ was amplified from the human liver cDNA library plasmid by the polymerase chain reaction (forward primer, 5Ј-GCGCGGATCCGGTACCATGTCCTTGTGGCTGGGGGC; reverse primer, 5Ј-GCGCGGATCCGATATCTCATTCGTGCACATCCCAGATC; initiator codon is underlined) and cloned into the BglII site of the SP6 transcription vector pSPUTK. Transcription of the different cDNAs, followed by translation in rabbit reticulocytes, was performed using a coupled system (Promega), as described previously (14,24).
Protein Binding Assays-Maltose-binding protein (MBP)-mPPAR␣ and MBP-RXR␣ fusion protein expression vectors were constructed in pMAL-2c (New England Biolabs), and the fusion proteins were purified by affinity chromatography according to the manufacturer's instructions. Purified MBP-RXR␣, MBP-mPPAR␣, and MBP proteins (2 mg/ml in column buffer (20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 20% (v/v) glycerol)) were adsorbed to amylose resin (4.5 mg/ml settled resin) and resuspended as a 50% slurry in column buffer. Twenty l of slurry was adjusted to 500 l with column buffer containing 0.2% (v/v) Nonidet P-40 and 2% bovine serum albumin (BSA) and incubated with 1-5 l of programmed reticulocyte lysate for 30 min at room temperature with continuous mixing. Beads were washed sequentially with 20 volumes of column buffer containing 0.2% (v/v) Nonidet P-40 and 2% BSA; column buffer containing 0.1% (v/v) Nonidet P-40 and 2% BSA; and column buffer alone. Bound material was eluted from beads by boiling in SDS sample buffer and analyzed by SDSpolyacrylamide gel electrophoresis.

RESULTS AND DISCUSSION
Interaction Cloning of PPAR-interacting Factors-Yeast harboring GBD-N⌬mPPAR␣ were used to screen a HeLa cell cDNA library fused to the GAL4 transactivation domain. Of 5 ϫ 10 5 independent transformants screened, 9 colonies remained blue in color and his ϩ upon clonal purification. Retransformation of rescued plasmids confirmed that all 9 candidate clones conferred the his ϩ and ␤-gal ϩ phenotypes only in the presence of GBD-N⌬mPPAR. Two positive clones, designated GAD-SM1 and GAD-KM1, contained overlapping sequences of the same cDNA that partially encoded a previously undescribed member of the nuclear hormone receptor superfamily. Interestingly, the same clone was isolated in an independent screen for RXR␣-interacting factors (data not presented), demonstrating that this factor also interacts with RXR␣ (Table I). A 1.7-kbp cDNA was isolated from a gt11 human liver cDNA library using GAD-SM1 as a probe and shown to contain an open reading frame encoding a 447-amino acid polypeptide. After this work was completed, Willy and co-workers (20) reported the cloning of a cDNA encoding a novel orphan nuclear hormone receptor using low stringency screening of a human liver cDNA library. The cDNA for this receptor, designated LXR␣, is essentially identical at both the nucleotide and deduced amino acid sequence level to the cDNA clone isolated here by mPPAR␣ interaction cloning. The only differences noted in the deduced polypeptide are a phenylalanine in place of the leucine at position 192 and a histidine in place of the leucine at position 414. LXR␣ was shown to interact with RXR␣ (20), consistent with our two-hybrid analysis results. By convention, we will refer to our mPPAR␣-interacting receptor as LXR␣.
LXR␣ Interacts with mPPAR␣ and RXR␣ in Vitro-To confirm that LXR␣ physically interacts with mPPAR␣ and RXR␣, protein binding assays were carried out using immobilized MBP fusion proteins and [ 35 S]methionine-labeled LXR␣ syn- SM1 Blue a ␤-Galactosidase activity was determined by colony filter assay using S. cerevisiae HF7c harboring the indicated GBD and GAD plasmids, respectively. SM1 refers to GAD-SM1, which encodes amino acid residues 61-447 of LXR␣.
LXR␣ Binds to Direct Repeat Hormone Response Elements Cooperatively with RXR␣ but Not with mPPAR␣-To identify potential mPPAR␣/LXR␣ DNA-binding sites, we synthesized a series of oligonucleotides that contained TGACCT direct repeats separated by 0 -5 nucleotides (DR0 -DR5) and tested them in electrophoretic mobility shift assays with in vitro translated LXR␣, RXR␣, and mPPAR␣ alone or in combination. Binding studies were also carried out with natural PPREs from the HD and AOx genes and with a natural LXRE previously identified in the mouse mammary tumor virus promoter (20). As shown in Fig. 2a, mPPAR␣, LXR␣, and RXR␣ on their own did not bind to any of the elements tested. However, LXR␣ bound cooperatively with RXR␣ to a DR4 element, the LXRE, and to a lesser extent, to a DR5 element, as reported previously (20). As expected, mPPAR␣/RXR␣ heterodimers bound preferentially to a DR1 element as well as to the AOx-and HD-PPREs. In contrast, cooperative binding of LXR␣ and mPPAR␣ was not observed on any of these synthetic direct repeat or natural response elements. Whether there exist natural and specific DNA target sites recognized by mPPAR␣/LXR␣ heterodimers remains to be determined.
LXR␣ Inhibits Binding of mPPAR␣/RXR␣ Heterodimers to PPREs-Coincubation of LXR␣ with RXR␣ and mPPAR␣ re-sulted in a decrease in the binding of mPPAR␣/RXR␣ to the HD-and AOx-PPREs (Fig. 2a, compare lanes 5 and 8). Similarly, mPPAR␣ reduced binding of LXR␣/RXR␣ to DR4 and DR5 elements and to the LXRE (compare lanes 7 and 8). Coincubation with increasing amounts of LXR␣ resulted in a progressive decrease in the binding of mPPAR␣/RXR␣ heterodimers to the HD-and AOx-PPREs (Fig. 2b). Since LXR␣ binds directly to mPPAR␣ and to RXR␣, this inhibition is likely the result of LXR␣ sequestering one, or both, of these receptors into non-binding complexes. Reciprocally, mPPAR␣ can competitively inhibit the binding of LXR␣/RXR␣ to the LXRE and to synthetic DR target sites through the formation of both LXR␣/mPPAR␣ and mPPAR␣/RXR␣ heterodimers.
LXR␣ Antagonizes Peroxisome Proliferator-mediated Signaling in Vivo --To investigate the effect of LXR␣ on peroxisome proliferator-mediated signaling in vivo, a luciferase reporter plasmid containing the AOx-PPRE (pAOx(X2)luc) was cotransfected along with LXR␣, RXR␣, and mPPAR␣ expression vectors into BSC40 cells, and luciferase activity was monitored. LXR␣ and mPPAR␣ individually or in combination had little effect on reporter gene expression either in the presence or absence of the peroxisome proliferator Wy-14,643 (Fig. 3a) or the RXR␣ ligand 9-cis-retinoic acid (not shown). However, LXR␣ potently inhibited induction mediated by mPPAR␣/ RXR␣. As shown in Fig. 3b, cotransfection with increasing amounts of the LXR␣ expression plasmid led to a progressive reduction in drug-independent and -dependent induction of the AOx-PPRE reporter gene construct by mPPAR␣/RXR␣. The inclusion of 9-cis-retinoic acid in the transfections along with Wy-14,643 had little effect on LXR␣-mediated repression. Sim- ilar findings were obtained using an HD-PPRE luciferase reporter gene construct (data not presented). Thus, LXR␣ antagonizes transactivation by mPPAR␣/RXR␣ in vivo.
Control transfections demonstrated that LXR␣ was functionally expressed under these conditions. Thus, co-expression of LXR␣ and RXR␣ activated expression of a reporter gene that contained a synthetic DR4 (pDR4(X2)luc) response element but not a DR1 (pDR1(X2)luc) response element (Fig. 3c). Activation of the DR4 reporter gene by LXR␣/RXR␣ was increased in the presence of 9-cis-retinoic acid but was not dependent upon its presence. Willy and co-workers (20) have also demonstrated that LXR␣ interacts with endogenous RXR␣ to mediate transcriptional activation through DR4 elements; however, under these conditions, activation was observed only in the presence of RXR-specific ligands. Our findings indicate that LXR␣ also activates transcription in the absence of RXR-specific ligands when RXR␣ is co-expressed. This is consistent with previous findings that LXR␣ homologs isolated from rat (RLD-1, rUR) and human (NER, hUR) (25)(26)(27) also activate transcription via DR4 target sites in the absence of added RXR-specific ligands.
In summary, we have shown that PPAR␣ can interact directly with other members of the nuclear hormone receptor superfamily in addition to RXR␣, suggesting that combinatorial receptor interactions involving PPARs are more extensive than previously anticipated. Moreover, our findings indicate that both LXR␣ and PPARs play a broader physiological role in the convergence of distinct receptor signaling pathways. Since LXR␣ binds to the two identified components that are necessary for peroxisome proliferator responsiveness, there is potential for complex and diverse effects on both retinoid and peroxisome proliferator signaling pathways. The physiological importance of LXR␣ and related receptors in PPAR signaling is not known at present. Thus far, we have demonstrated that LXR␣ is a negative regulator of PPAR-mediated activation of peroxisome proliferator-responsive genes. Whether PPAR/ LXR␣ heterodimers may also positively regulate gene expression awaits the definition of natural high affinity binding sites and the identification of potential LXR␣ ligands and target genes.
FIG. 3. LXR␣ antagonizes mPPAR␣/RXR␣-mediated transcriptional activation by peroxisome proliferators. Panel a, the pAOx(X2)luc reporter plasmid was cotransfected into BSC40 cells with LXR␣, mPPAR␣, and RXR␣ expression plasmids (0.5 g each) in the absence or presence of the peroxisome proliferator Wy-14,643, as indicated, and luciferase activity was measured. Panel b, effect of increasing amounts of LXR␣ expression vector in the presence of constant amounts of mPPAR␣ and RXR␣ expression vectors, with and without 9-cis-retinoic acid and Wy-14,643. Panel c, activation of a DR4 reporter gene construct by LXR␣/RXR␣. Transfections were carried out with LXR␣ and RXR␣ expression plasmids and with luciferase reporter gene constructs that contained a synthetic DR4 element (pDR4(X2)luc) or a DR1 element (pDR1(X2)luc), as indicated. Luciferase activity (ϮS.D.) in panels a and b is the average (corrected against the ␤-galactosidase internal reference) from three independent transfections, each carried out in duplicate. The values shown were normalized to the value obtained with cotransfected RXR␣ and mPPAR␣ expression plasmids in the presence of Wy-14,643, which was taken as 100%. In panel c, the values are from duplicate transfections (values did not vary by more than 15%) and were normalized to the value obtained with the respective reporter gene alone, which was taken as 1.