Retinoic acid is a high affinity selective ligand for the peroxisome proliferator-activated receptor beta/delta.

Retinoic acid (RA) modulates transcription of numerous target genes, thereby regulating a myriad of biological processes. It is well established that RA functions by activating retinoic acid receptors (RARs), which, in turn, control cell differentiation, proliferation, and apoptosis. However, perplexing reports of diverse and sometime opposing actions of RA have been published. Hence, while RA induces apoptosis and inhibits cell growth in some settings, it potentiates proliferation and acts as an anti-apoptotic agent in others. These observations raise the possibility that signaling pathways other than RAR may be involved in mediating RA activities. Here we show that RA is a high affinity ligand for another nuclear receptor, namely the orphan receptor peroxisome proliferator-activated receptor (PPAR) beta/delta. We demonstrate that while RA does not activate PPARalpha and PPARgamma, it binds to PPARbeta/delta with nanomolar affinity, modulates the conformation of the receptor, promotes interaction with the coactivator SRC-1, and efficiently activates PPARbeta/delta-mediated transcription. Transcriptional signaling by RA is thus exerted by a dual pathway, providing a rationale for understanding divergent cellular responses to this hormone.

The vitamin A metabolite all-trans-retinoic acid (RA) 1 is a potent modulator of cell growth, differentiation, and apoptosis in various tissues both in the embryo and in the adult (1,2). Consequently, RA plays key roles in embryonal development (3,4) and is also used as a therapeutic agent in a number of pathologies including several cancers (5,6). It is well established that the pleiotropic activities of RA are mediated by members of the superfamily of nuclear hormone receptors known as retinoic acid receptors (RARs), which upon activation modulate the rates of transcription of numerous target genes.
Like other class 2 nuclear receptors, RARs function as heterodimers with retinoid X receptors (RXRs). These heterodimers bind to specific response elements in the promoter regions of target genes and function as ligand-inducible transcription factors (1,7). Perplexingly, RA displays diverse and sometime opposing activities. For example, while this hormone induces apoptosis and inhibits cell growth in some settings (8 -10), it potentiates proliferation and acts as an anti-apoptotic agent in others (11)(12)(13)(14). These observations raise the possibility that signaling pathways other than RAR may be involved in mediating RA activities. In search for a mechanism that may underlie these inconsistencies, we considered the orphan nuclear receptor PPAR␤/␦, for which no physiological ligand has been reported. The biological functions of PPAR␤/␦ are not as well understood as those of its PPAR␣ and PPAR␥ isotypes. It was, however, recently demonstrated that PPAR␤/␦ mediates differentiation in keratinocytes and oligodendrocytes (15,16). It was also shown that by up-regulating the expression of antiapoptotic genes and down-regulating the expression of proapoptotic genes, PPAR␤/␦ significantly increases cell resistance to apoptosis (15,17). We thus set out to examine whether this receptor may be activated by RA. We show that RA is a high affinity ligand for PPAR␤/␦. We demonstrate that while RA does not activate PPAR␣ and PPAR␥, it binds to PPAR␤/␦ with nanomolar affinity, modulates the conformation of the receptor, promotes interaction with the coactivator SRC-1, and efficiently activates PPAR␤/␦-mediated transcription. These observations suggest that transcriptional signaling by RA is exerted by a dual pathway, which, in turn, may lead to divergent cellular responses.
Fluorescence Titrations-Bacterially expressed mPPAR␣-LBD, mP-PAR␤/␦-LBD, and mPPAR␥-LBD (0.2-1 M) were titrated directly in a cuvette with RA dissolved in ethanol. Ethanol concentration was usually below 1% and never exceeded 2%. To ensure equilibration between protein and ligand, the fluorescence was monitored until a constant value was reached. The progress of titrations was monitored by following the decrease in the intrinsic fluorescence of the protein (excitation, 280 nm; emission, 340 nm), which accompanies RA binding. Inner filtering by the ligand, reflected by linear slopes observed following saturation of the protein, was corrected for as described (20). Corrected data were analyzed to obtain equilibrium dissociation constants (K d ). Analyses were carried out by fitting the data to equation (1) derived from simple binding theory (23), where F is the observed fluorescence, F 0 and F ϱ are the fluorescence in the absence of ligand and at saturation, respectively, P T and R T are the total concentrations of protein and RA, respectively, and K a is the association constant (K a ϭ 1/K d ).
* This work was supported by Grants CA68150 and DK060684 from the National Institutes of Health. 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.
Transactivation Assays-COS-7 cells were maintained with Dulbecco's modified Eagle's medium supplemented with 10% charcoal-treated newborn calf serum. Cells were transfected using FuGENE (Roche Diagnostics) with a (PPRE) 3 -luciferase reporter (0.3 g/well), expression vectors for receptors (0.3 g/well), and pCH110 (0.2 g/well). Me-dium was replaced with serum-free Dulbecco's modified Eagle's medium for 24 h. Cells were treated with denoted ligands for 24 h and lysed. Lysates were analyzed for reporter activity and normalized for ␤-galactosidase.

RESULTS AND DISCUSSION
Binding of RA to PPAR␣, PPAR␤/␦, and PPAR␥ was studied by fluorescence titrations. This method relies on the considerable overlap of the absorption spectrum of RA with the fluorescence emission spectra of tryptophanes and tyrosines, a consequence of which is that RA-binding results in quenching of the intrinsic fluorescence of a binding protein (20 -22). PPARs were titrated with RA and the ligand-induced decrease in protein fluorescence was monitored (Fig. 1). Titration curves were corrected for inner filtering and fitted to a simple binding equation (23) to obtain equilibrium dissociation constants (K d , Table I). RA bound to PPAR␣ and PPAR␥ with a low affinity demonstrated by K d values of 100 -200 nM. In contrast, RA associated with PPAR␤/␦ with a K d of 17 nM, revealing both high affinity and isotype selectivity. The measured affinity of PPAR␤/␦ for RA is thus comparable with that of various PPAR ligands for  these receptors (24) and with that of the synthetic PPAR␤/␦selective ligand L165041 (25).
Ligand binding by nuclear receptors induces a conformational transition that results in a more compact protein fold (26) and, consequently, decreases their sensitivity to proteolysis (27). Protease sensitivity assays were carried out to compare the effects of RA to those of the synthetic PPAR␤/␦-selective agonist L165041 (Fig. 2). As shown previously (28), addition of L165041 decreased the sensitivity of the receptor to chymotrypsin digestion and resulted in preservation of protease-resistant fragments (denoted by lines on the right in Fig.  2). RA also increased the resistance of the receptor to proteolysis but gave rise to a different pattern of protected fragments.
It was reported previously that different PPAR agonists may give rise to differential protection profiles (29). Nevertheless, the observations in Fig. 2, showing differences between the effects of RA and L165041, raise the question whether RA is a bona fide PPAR␤/␦ agonist. The hallmark of active nuclear receptor ligands is that they enable the recruitment of transcriptional coactivators (30). The ability of RA to promote association of PPAR␤/␦ with the coactivator SRC-1 was thus investigated by GST-pull-down assays (Fig. 3). Similarly to L165041, the interactions of SRC-1 with PPAR␤/␦ were significantly stabilized in the presence of RA but not upon addition of the RAR-selective ligand TTNPB (Fig. 3a). Moreover, RA-induced enhancement of SRC-1 binding by the receptor occurred in dose-depended fashion (Fig. 3, b and c).
The ability of RA to enhance the transcriptional activity of PPAR␣, PPAR␤/␦, and PPAR␥ was then directly examined by transactivation assays. In these, a luciferase reporter driven by a consensus PPAR response element (PPRE) was utilized. Cells were cotransfected with the reporter construct together with expression vectors for each of the PPAR isotypes, and the effect of RA treatment of reporter expression was studied. RA did not activate PPAR␥ and poorly activated PPAR␣. In contrast, RA efficiently enhanced the transcriptional activity of PPAR␤/␦ (Fig. 4), displaying an EC 50 for this subtype that was an order of magnitude lower than that observed for PPAR␣ (Fig. 4a). Control experiments showed that all three PPAR isotypes were similarly activated in these cells by synthetic isotype-selective Retinoic Acid Is a High Affinity PPAR␤/␦ Ligand 41591 ligands, verifying that the inability of RA to activate PPAR␣ and PPAR␥ did not stem from poor expression of these isotypes (Fig. 4b). The data thus demonstrated that RA is a potent and selective activator of PPAR␤/␦. The transcriptional activities of PPARs are mediated by their heterodimers with RXR (31). Both PPAR-RXR heterodimers and RXR homodimers bind to similar response elements comprised of two repeats separated by a single base pair (DR-1). In either complex, RXR can be activated by 9-cis-RA. It can thus be argued that RA may have been converted to an RXR ligand within the cells and that its transcriptional activity reflected activation of RXR rather than PPAR␤/␦. While the observations that RA selectively activated PPAR␤/␦ renders this possibility unlikely, we examined it directly. An RXR construct lacking its AF-2 domain (RXR⌬H12) was used. RXR⌬H12 can heterodimerize with PPAR (32) and properly associate with cognate DNA (33), but it cannot activate transcription on its own. Consequently, PPAR-RXR⌬H12 dimers do not respond to RXR ligands but can be activated by PPAR agonists (albeit at a somewhat lower efficiency than in the presence of WT-RXR). Cells were cotransfected with expression vectors for RXR⌬H12 and PPAR␤/␦, and the ability of RA to activate the PPRE-reporter was studied (Fig. 4c). As expected, the RXR-selective agonist LG268 was ineffective. In contrast, both the PPAR␤/␦ ligand L165041 and RA activated the reporter, demonstrating that RA acts in these assays through PPAR␤/␦ and not via RXR. Note as well that the observed reporter activity was not mediated by RAR, which does not function through the PPRE used in these assays (Fig. 4d).
These observations lead to the surprising conclusion that in addition to activating RAR, RA is a potent and selective ligand for the orphan receptor PPAR␤/␦. The biological functions of PPAR␤/␦ have only began to be elucidated as it was recently reported that this receptor mediates differentiation in keratinocytes and oligodendrocytes (15,16). It was also demonstrated that expression and activation of PPAR␤/␦ increases cell resistance to apoptosis (15). This activity was traced to PPAR␤/␦mediated transcriptional up-regulation of the expression of ILK and PDK1, leading to activation of the anti-apoptotic Akt1 signaling pathway (17).
The current use of RA as a chemotherapeutic agent in several cancers is based on the pronounced growth inhibitory activities of this hormone, which are believed to be mediated by RARs. It was reported that in various carcinomas, RA induces apoptosis (34,35), differentiation (8), cell cycle arrest (36), and that in some cells, and it inhibits proliferation by a combination of effects (37,38). However, it has also been reported that in some settings, RA treatment may result in enhanced proliferation (11)(12)(13)(14). For example, it was shown that RA potentates epidermal growth factor-induced proliferation in cultured explants of mouse mammary gland (14), inhibits lobuloalveolar differentiation of cultured mouse mammary glands (13), and promotes at late stages of malignant progression in some cell variants in breast cancer tumors (39). The basis for the opposing activities of RA on cell growth is unknown at present. However, an intriguing possibility suggested by our observations is that while RA exerts growth inhibition through activating RAR, it may evoke anti-apoptotic activities, and thus promote growth via activation of PPAR␤/␦. An important ques-tion in regard to this possibility is how the partitioning of RA between its dual transcriptional signaling may be regulated.