Regulation of retinoic acid-induced inhibition of AP-1 activity by orphan receptor chicken ovalbumin upstream promoter-transcription factor.

Retinoids are therapeutically effective in the treatment of various cancers, and some of the therapeutic action of retinoids can be ascribed to their potent inhibition of AP-1 activity that regulates transcription of genes associated with cell growth. We recently reported that the expression of orphan receptor chicken ovalbumin upstream promoter-transcription factor (COUP-TF) plays a role in mediating the growth inhibitory effect of trans-retinoic acid (trans-RA) in cancer cells. To gain insight into the molecular mechanism by which COUP-TF regulates trans-RA activity, we evaluated the effect of COUP-TF on antagonism of AP-1 activity by trans-RA. Our results demonstrated a positive correlation between COUP-TF expression and the ability of trans-RA to inhibit AP-1 activity in various cancer cell lines. In transient transfection assay, expression of COUP-TF strongly inhibited tumor promoter 12-O-tetradecanoylphorbol-13-acetate-induced AP-1 transactivation activity and transactivation of c-Jun/c-Fos in both a trans-RA-dependent and -independent manner. In vitro studies demonstrated that the addition of COUP-TF inhibited c-Jun DNA binding through a direct protein-protein interaction that is mediated by the DNA binding domain of COUP-TF and the leucine zipper of c-Jun. Stable expression of COUP-TF in COUP-TF-negative MDA-MB231 breast cancer cells restored the ability of trans-RA to inhibit 12-O-tetradecanoylphorbol-13-acetate-induced c-Jun expression. The effect of COUP-TF in enhancing the trans-RA-induced antagonism of AP-1 activity required expression of retinoic acid receptors (RARs), since stable expression of COUP-TF in COUP-TF-negative HT-1376 bladder cancer cells, which do not express RARalpha and RARbeta, failed to restore trans-RA-induced AP-1 repression. Thus, COUP-TF, through its physical interaction with AP-1, promotes anticancer effects of retinoids by potentiating their anti-AP-1 activity.

nuclear receptor classes, the retinoic acid receptor (RAR) 1 and retinoid X receptor (RXR) (3)(4)(5). Both are encoded by three different genes (␣, ␤, and ␥) and function as ligand-inducible transcription factors in vivo mainly as RXR/RAR heterodimers. trans-Retinoic acid (trans-RA) binds RARs, whereas 9-cis-RA binds both RARs and RXRs. Binding of retinoids to their receptors induces receptor conformational changes that switch on transcription of genes containing RA response elements (RAREs) (3)(4)(5). In addition to their positive regulation of RARE-containing genes, retinoid receptors, in response to their ligands, can inhibit effects induced by the tumor promoter 12-O-tetradecanoylphorbol-13-acetate (TPA) and the transcriptional activity of the proto-oncogenes c-Jun and c-Fos (6), which are components of the AP-1 complex, which often has a role in cancer cell proliferation (7). The activation of AP-1-responsive genes by TPA or c-Jun/c-Fos through TPA response elements (TREs) is repressed by retinoid receptors in a ligand-dependent manner (8 -10). Conversely, AP-1 represses transactivation of retinoid receptors (8,9). This mutual antagonism appears to play a critical role in regulating cell growth and differentiation (6). For example, overexpression of c-Jun conferred retinoid resistance to breast cancer cells (11), while overexpression of retinoid receptors enabled trans-RA to inhibit AP-1 activity in ovarian cancer cells and their growth (12). The functional interaction between AP-1 and retinoid receptors is also observed for other nuclear receptors, including glucocorticoid receptor (13)(14)(15)(16), thyroid hormone receptor (17), vitamin D receptor, estrogen receptor (18,19), and androgen receptor (AR) (18).
The anti-AP-1 activities shown by liganded retinoid receptors appear to contribute significantly to the therapeutic efficacy of retinoids against hyperproliferative diseases (6). Transcription of various AP-1-responsive genes, such as collagenase and stromelysin, which have roles in tumor progression and invasiveness (20), is inhibited by retinoids (8 -10) and has been reported to contribute to their reversal of human bronchial epithelial squamous differentiation (21). Retinyl methyl ether, which effectively prevents mammary cancer in animals, strongly suppressed AP-1 activity in breast cancer cell (22). Interestingly, RAR␤, a negative regulator of cancer cell growth (23), potently inhibits AP-1 activity and collagenase expression in both breast and lung cancer cells (24). Recent studies demonstrate that retinoids that specifically inhibit AP-1 activity but antagonize RAR transactivation on RAREs inhibited the growth of many different types of cancer cells (25)(26)(27). Anti-AP-1 retinoids inhibited squamous differentiation of human bronchial epithelial cells (21), TPA-induced transformation and the clonal growth of the promotion-sensitive JB6 mouse epidermal cell line (28), and papilloma formation in animals (29). Thus, anti-AP-1 activity of retinoids contributes to their chemopreventive and chemotherapeutic effects, presumably by blocking the processes of tumor promotion and cell transformation.
The mechanism by which retinoids inhibit AP-1 activity remains largely unclear. Unlike the effect of retinoid receptors on RAREs, inhibition of AP-1 activity by retinoid receptors is independent of retinoid receptor-RARE interaction (8 -10). Previous studies suggested several possible mechanisms for AP-1 inhibition by retinoid receptors. First, retinoid receptors are reported to physically interact with c-Jun and/or c-Fos (8,9). This interaction results in the mutual inhibition of their DNA binding and transactivation functions and could explain the cross-talk occurring between AP-1 and retinoid signaling. However, a large excess of either retinoid receptor protein or c-Jun and c-Fos proteins was required to inhibit in vitro binding to the TRE or RARE, respectively (8,9). Because the in vivo footprint assay for glucocorticoid receptor and AP-1 interaction did not reveal any effect on DNA binding (30), whether RAR and AP-1 directly interact in vivo remains to be established. Subsequently, it was suggested that blocking activation of the Jun N-terminal kinase signaling pathway that activates AP-1 might be responsible for AP-1 inhibition by retinoid and other nuclear receptors (31). Although this mechanism may explain how retinoids inhibit AP-1 activity under some conditions, it does not address how mutual inhibition occurs. Recent results suggest that the molecular basis of receptor-mediated inhibition of AP-1 transcriptional activation might be due to competition for a common coactivator, such as the cAMP-response element-binding protein (CBP), which is required for transcriptional activation by both the AP-1 complex and RARs (32). However, a domain of RAR capable of inhibiting AP-1 activity, such as the DNA-binding domain (33), does not interact with CBP (32). The development of retinoids that specifically inhibit AP-1 activity without transactivating RARs (25,26) also argues against the involvement of RAR coactivators in RARs-AP-1 cross-talk. Thus, the availability of CBP is unlikely to be the sole modulator of RAR and AP-1-signaling pathways, and other adapter proteins may be involved in the antagonism of AP-1 activity by RARs.
Recent studies demonstrate that orphan receptor COUP-TF is involved in regulation retinoid responses (34,35). COUP-TF is encoded by two distinct genes, COUP-TFI (EAR-3) (36,37) and COUP-TFII (ARP-1) (38). Both show exceptional homology and overlapping expression patterns, suggesting their redundant functions (39). COUP-TF can modulate retinoid responses through either its high affinity binding to various RAREs or its heterodimerization with RXR (40 -43). We previously reported that COUP-TF expression was required for cancer cell growth inhibition by trans-RA (35). We further demonstrated that the effect of COUP-TF is partly due to its induction of RAR␤ that mediates growth inhibition by retinoids in various cancer cells (35).
To further understand how COUP-TF is involved in the regulation of trans-RA activity in cancer cells, we investigated the effect of COUP-TF on antagonism of AP-1 activity by trans-RA. Our data demonstrate that COUP-TF expression in various cancer cell lines correlated with the ability of trans-RA to suppress AP-1 transcriptional activity. In addition, we found that COUP-TF effectively suppressed AP-1 transcriptional activity by interacting with c-Jun to cause loss of c-Jun DNA binding. This interaction required COUP-TF DNA-binding domain and the c-Jun leucine-zipper domain. Although AP-1 inhibition by COUP-TF did not require trans-RA, COUP-TF strongly potentiated the AP-1 antagonism by trans-RA when RARs were expressed. Our results demonstrate that interaction between COUP-TF and AP-1 is involved in regulating anti-AP-1 activity of retinoids in cancer cells and suggest that COUP-TF plays a role in the cross-talk between retinoid and AP-1 signalings.

EXPERIMENTAL PROCEDURES
Cell Culture-HeLa ovarian and MDA-MB231 breast cancer cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS); Calu-6 lung cancer and HT-1376 bladder cancer cells were grown in MEM containing 10% FCS; and T-47D and ZR-75-1 breast cancer and H292 lung cancer cells were grown in RPMI-1460 medium with 10% FCS.
Plasmid Construction-CAT reporter constructs TRE-tk-CAT and Ϫ73Col-CAT have been described (8,17,22,24), as have expression vectors for RAR␣, COUP-TF, c-Jun, and c-Fos (8,17,22,24,44). pcDNA3-COUP-TFII C-terminal deletion mutants pcDNA3-COUP-TFII⌬7, pcDNA3-COUP-TFII⌬30, and pcDNA3-COUP-TFII⌬108 were generated as described (35). pcDNA3-COUP-TFII N-terminal deletion mutants were Ϫ73Col-CAT reporter was transfected into the indicated cancer cell lines. After transfection, the cells were incubated in DMEM medium containing 0.5% FCS for 24 h and treated with either TPA (100 ng/ml) alone or with the indicated concentrations of trans-RA. After 12 h, the cells were harvested, and CAT activity was determined. The activities of cotransfected ␤-galactosidase were used as controls for transfection efficiency. B, effect of COUP-TF expression in COUP-TF-negative cancer cell lines on inhibition of AP-1 activity by trans-RA. The Ϫ73Col-CAT reporter was transfected with or without COUP-TF into COUP-TF-negative cancer cell lines (MDA-MB231, H292, and HT-1376). Cells were incubated in medium (see "Experimental Procedures") containing 0.5% FCS for 24 h and then treated with either TPA (100 ng/ml) alone or with 10 Ϫ6 M trans-RA. After 12 h, cells were harvested, and CAT activity was determined. The activities of cotransfected ␤-galactosidase were used as reference values.
Transient and Stable Transfection Assays-HeLa cells (1 ϫ 10 5 cells/ well) were plated in 24-well plates for 16 -24 h before transfection as described (34,35,45), and other cancer cells (5 ϫ 10 5 cell/well) were seeded in six-well plates. A modified calcium phosphate precipitation procedure was used for transient transfections (34,35,45). Briefly, 200 ng of reporter plasmid, 100 ng of ␤-galactosidase of expression vector (pCH 110; Amersham Biosciences), and various amounts of each expression vector were mixed with carrier DNA (pBluescript) to give 1000 ng of total DNA/well. CAT activity was normalized for transfection efficiency to the responding ␤-Gal activity. For stable transfections, the pRC/CMV-COUP-TF recombinant was transfected into MDA-MB231 and HT-1376 cells by the calcium phosphate precipitation method, and the stable clones were screened with G418 (Invitrogen) as described (35). Integration and expression of transfected cDNA were determined by Southern blotting and Northern blotting, respectively.
Preparation of Receptor Proteins-Receptor proteins for RAR␣, TR3, COUP-TF, and its mutants were synthesized by in vitro transcriptiontranslation using rabbit reticulocyte lysates (Promega) as described previously (34). Amounts of translated proteins were determined by [ 35 S]methionine incorporation and SDS-PAGE with quantitation by incorporated radioactivity after normalization relative to methionine content.
Gel Retardation Assay-A fragment of the collagenase promoter region Ϫ73 to ϩ63 containing one AP-1-binding site (TRE) was excised from a collagenase-CAT construct (8). In addition, the AP-1 binding sequence 5Ј-GATCCGGATGAGTCACCA-3Ј was synthesized. Two fragments were labeled with [ 32 P]dCTP for use as probes for protein-DNA interaction. In vitro translated protein was incubated with the probe in a 20-l reaction mixture containing 10 mM HEPES, pH 7.9, 50 mM KCl, 1 mM dithiothreitol, 2.5 mM MgCl 2 , 10% glycerol, and 1 g of poly(dI-dC) at 25°C for 15 min. DNA-protein complexes were resolved on 5% nondenaturing polyacrylamide gels, and then gels were dried and analyzed by autoradiography.
GST Pull-down Assay-To prepare glutathione S-transferase (GST)c-Jun or GST-c-Jun mutant fusion proteins, each c-Jun DNA or c-Jun mutant fragment was cloned in frame into the expression vector

FIG. 2. Inhibition of AP-1 activity by COUP-TF. A, inhibition of TPA-induced collagenase promoter activity by COUP-TF in HeLa cells. The
Ϫ73Col-CAT reporter was cotransfected without or with COUP-TFI expression vector (10, 20, or 50 ng) or the control pcDNA3 vector into cells. After transfection, cells were incubated in DMEM medium containing 0.5% FCS for 24 h and then treated with or without TPA (100 ng/ml). After 12 h, the cells were harvested, and CAT activity was determined. B, inhibition of c-Jun-induced Ϫ73Col-CAT and TRE-tk-CAT activity by COUP-TF in HeLa cells. The Ϫ73Col-CAT or TRE-tk-CAT reporter was cotransfected with/without c-Jun in the presence or absence of COUP-TFI expression vector (10, 20, and 50 ng) into cells. After transfection, cells were incubated in DMEM medium containing 0.5% FCS for 24 h, and CAT activity was determined. C, inhibition of c-Jun and c-Fos activities by COUP-TF in HeLa cells. The Ϫ73Col-CAT reporter was cotransfected with/without c-Jun (50 ng) and/or c-Fos (50 ng) expression vectors either in the presence or absence of COUP-TFI or COUP-TFII expression vector (10, 20, and 50 ng) into HeLa cells. The cells were then harvested, and CAT activity was determined.
pGEX-3X (Amersham Biosciences). Fusion proteins were expressed in bacteria using the manufacturer's procedure and analyzed by gel retardation assays and Western blotting (data not shown). To determine the interaction between COUP-TF and c-Jun, the fusion proteins were immobilized on glutathione-Sepharose beads. The vector protein (GST), prepared under the same conditions as a control, was also immobilized. Beads were preincubated with bovine serum albumin (1 mg/ml) at room temperature for 5 min. 35 S-labeled in vitro translated COUP-TF proteins (2-5 l, depending on translation efficiency) were then added to the beads. The beads in 200 l in EBC buffer (140 mM NaCl, 0.5% Nonidet P-40, 100 mM NaF, 200 mM sodium orthovanadate, and 50 mM Tris, pH 8.0) were rocked continuously for 1 h at 4°C. After washing 5 times with NETN buffer (100 mM NaCl, 1 mM EDTA, 20 mM Tris, pH 8.0, 0.5% Nonidet P-40), bound proteins were analyzed by SDS-PAGE and autoradiography.
Northern Blotting-For Northern analysis, total RNAs were prepared using an RNeasy Mini Kit (Qiagen, Hilden, Germany). Total RNA (30 g) from different cell lines treated with or without trans-RA (10 Ϫ6 M) in the presence or absence of TPA (100 ng/ml) was analyzed by Northern blotting as described (46).

Correlation between COUP-TF Expression and Inhibition of AP-1 Activity by trans-RA-
We recently reported that expression of COUP-TF is required for growth inhibition and apoptosis induction by trans-RA in various cancer cells (35). Since inhibition of AP-1 activity by retinoids is known to contribute to their anticancer effects, we studied whether COUP-TF expression was involved in the process. The effect of trans-RA on inhibiting AP-1 activity (Fig. 1A) was evaluated in COUP-TF-positive T-47D and ZR-75-1 breast cancer and Calu-6 lung cancer cell lines and in COUP-TF-negative MDA-MB231 breast cancer, H292 lung cancer, and HT-1376 bladder cancer cell lines (35). AP-1 activity induced by TPA was determined by transient transfection using the reporter Ϫ73Col-CAT, which contains a TRE that binds AP-1 (8). TPA-induced reporter activity was strongly inhibited in a trans-RA-dependent manner in the COUP-TF-positive ZR-75-1, T-47D, and Calu-6 cell lines. In contrast, trans-RA showed very little effect on TPAinduced Ϫ73Col-CAT activity in the COUP-TF-negative MDA-MB231, H292, and HT-1376 cell lines. Thus, COUP-TF expression correlates positively with the ability of trans-RA to inhibit AP-1 transcriptional activity.
Trans-RA-dependent and -independent Antagonism of AP-1 Activity by COUP-TF-The positive association between COUP-TF expression and trans-RA-induced anti-AP-1 activity suggested that COUP-TF was required for trans-RA to inhibit AP-1 activity and that the absence of COUP-TF expression in MDA-MB231, H292, and HT-1376 cells might be responsible for the lack of trans-RA activity. Therefore, we transiently transfected the COUP-TF expression vector and the Ϫ73Col-CAT reporter into the COUP-TF-negative cell lines, in which trans-RA did not inhibit AP-1 transcriptional activity (Fig. 1B). Upon COUP-TF transfection, trans-RA inhibited TPA-induced reporter activity in both MDA-MB231 and H292 cells in a COUP-TF concentration-dependent manner. Interestingly, COUP-TF transfection alone produced some inhibition of AP-1 activity. This observation suggested that COUP-TF inhibited AP-1 activity in both trans-RA-dependent and -independent manners in MDA-MB231 and H292 cells. When evaluated in HT-1376 cells, COUP-TF expression was able to inhibit AP-1 activity independently of trans-RA (Fig. 1B). However, it failed to confer the ability of trans-RA to inhibit TPA-induced reporter activity, even at high transfection levels, suggesting that trans-RA-dependent inhibition of AP-1 activity by COUP-TF is impaired in this cell line.
The above results suggested that COUP-TF alone inhibited AP-1 activity. We then examined the anti-AP-1 activity of COUP-TF in HeLa cells, in which TPA strongly induced the transcription of the transfected Ϫ73Col-CAT reporter. Similar to that observed in other cancer cell lines (Fig. 1B), COUP-TFI cotransfection inhibited TPA-induced reporter activity in a COUP-TFI concentration-dependent manner ( Fig. 2A). Cotransfection of COUP-TFII produced almost identical results (data not shown). Activation of collagenase promoter by TPA occurs mainly through induction of AP-1 activity that activates the TRE in the promoter (7). We therefore examined whether COUP-TF expression also interfered with transactivation activity of c-Jun homodimer and c-Jun/c-Fos heterodimer. Cotransfection of Ϫ73Col-CAT with the c-Jun expression vector into HeLa cells led to about 5-fold induction of reporter expression (Fig. 2B), presumably due to activation of the collagenase promoter by the c-Jun homodimer. The c-Jun-induced Ϫ73Col-CAT reporter activity was repressed when COUP-TFI expression vector was cotransfected. Similarly, Ϫ73Col-CAT reporter activity induced by c-Jun/c-Fos heterodimers was inhibited by COUP-TFI and COUP-TFII (Fig. 2C). To determine that inhibition of collagenase promoter activity by COUP-TF was mediated by the TRE, we evaluated how COUP-TFI affected the TRE-tk-CAT reporter, which has a TRE sequence fused to the thymidine kinase promoter (Fig. 2B). Our result showed that cotransfection of COUP-TFI significantly inhibited c-Jun-induced TRE-tk-CAT activity (Fig. 2B). These results clearly demonstrate that COUP-TF inhibits transcriptional activity of c-Jun homodimer and c-Jun/c-Fos heterodimer and that the inhibition of AP-1 activity by COUP-TF on the TRE is respon- sible for its antagonism of TPA activity.
Inhibition of c-Jun Binding to DNA by COUP-TF-Inhibition of AP-1 activity by several nuclear receptors has been shown to be due to their inhibition of AP-1 binding to DNA (8, 9, 13-15, 17, 22, 47). To study whether inhibition of AP-1 DNA binding by COUP-TF was involved in its inhibition of AP-1 activity, the purified collagenase promoter fragment was used as a probe in gel shift assays for binding of in vitro synthesized c-Jun protein in the absence or presence of COUP-TF protein (Fig. 3A). c-Jun alone formed a strong complex with the promoter. However, upon preincubation with COUP-TF protein, c-Jun binding was significantly inhibited. The addition of COUP-TF protein did not show any new complex formed with the collagenase promoter, indicating that the inhibitory effect of COUP-TF is not due to its direct binding to the collagenase promoter. We also studied the effect of COUP-TF on binding of c-Jun to the TRE. Preincubation with COUP-TF protein similarly inhibited c-Jun binding to the TRE derived from the collagenase promoter (Fig.  3B). As a comparison, preincubation of c-Jun with TR3 orphan receptor (34) that could not antagonize c-Jun transactivation (data not shown) did not inhibit c-Jun binding to the TRE. This result demonstrated that the inhibitory effect of COUP-TF on c-Jun binding to the collagenase promoter is not due to regions other than the TRE in the promoter. Next, we compared the inhibitory effect of COUP-TF with that of RAR␣ that is known to repress AP-1 activity and inhibit c-Jun binding to TRE (8).
Whereas an equal amount of COUP-TF significantly inhibited c-Jun binding to the collagenase promoter, an equal amount of RAR␣ did not show any detectable inhibition (Fig. 3A). Inhibition of c-Jun binding required a 5-fold excess of RAR␣, similar to that observed before (8). The addition of trans-RA did not enhance the inhibitory effect of RAR␣ (Ref. 8 and data not shown). Together, these results demonstrate that the inhibition of c-Jun binding to the TRE contributes to its suppression of AP-1 transcriptional activity by COUP-TF and that COUP-TF is a more effective inhibitor than RAR␣.
Interaction of c-Jun and COUP-TF-Our observation that COUP-TF inhibits transactivation and DNA binding by c-Jun prompted us to examine whether COUP-TF and c-Jun interact directly using GST pull-down assay (Fig. 4). In vitro synthesized radiolabeled COUP-TFII was specifically pulled down by bacterially expressed GST-c-Jun hybrid protein but not by GST protein (Fig. 4B), demonstrating that COUP-TF and c-Jun interact in solution. To identify the domain of COUP-TFII responsible for interacting with c-Jun, COUP-TFII deletion mutants (Fig. 4A) were constructed and analyzed on the Ϫ73Col-CAT reporter for antagonism of AP-1 activity in HeLa cells (Fig. 4A) and for their interaction with c-Jun using the Ϫ73Col-CAT reporter was cotransfected without or with c-Jun (100 ng) alone or together with the indicated COUP-TFII mutants (50 ng). After transfection, cells were incubated in DMEM containing 0.5% FCS for 24 h. After 12 h, the cells were harvested, and CAT activity was determined. Reporter activity is shown as percentage of inhibition. B, interaction between c-Jun and COUP-TFs. c-Jun was synthesized in bacteria using pGEX-3X expression vector. GST-c-Jun fusion protein was immobilized on the glutathione-Sepharose beads. As a control, the same amount of glutathione S-transferase was also immobilized on the beads. In vitro translated 35 S-labeled COUP-TF and its mutant proteins were then mixed with the beads. After extensive washing, the bound proteins were analyzed by SDS-PAGE. The input proteins are shown for comparison.
We also identified the region of c-Jun required for COUP-TF interaction (Fig. 5). Deletion of amino acids 73-232 from c-Jun (c-Jun⌬AvaI) did not affect its ability to pull down COUP-TFII. In contrast, deletion of the C-terminal domain of c-Jun from amino acid 191 to 331 (c-Jun⌬bZIP), which encompasses the leucine-zipper region and basic region, completely abolished its ability to pull down COUP-TFII. Thus, the C-terminal domain, but not the N-terminal domain, of c-Jun is responsible for COUP-TF interaction.

Stable Expression of COUP-TF in COUP-TF-negative MDA-MB231 Cells Restores the Ability of trans-RA to Inhibit AP-1
Activity-To further examine the role of COUP-TF in trans-RAinduced inhibition of AP-1 activity, we analyzed the effect of trans-RA on TPA-induced c-Jun expression in MDA-MB231 breast cancer cells and MDA-MB231 cells stably transfected with COUP-TF (MB231/COUP#16) (35). Treatment of both cell lines with TPA strongly induced c-Jun expression as determined by Northern blotting (Fig. 6). Induction of c-Jun expression by TPA was probably due to an AP-1-binding site in the c-Jun promoter (7). Treatment with trans-RA did not affect basal c-Jun expression in both lines and did not inhibit TPAinduced c-Jun expression in MDA-MB231 cells. However, pretreatment of MB231/COUP#16 cells with trans-RA completely abolished TPA-induced c-Jun expression. These results demonstrate that COUP-TF expression is required for inhibition of AP-1 activity in MDA-MB231 cells by trans-RA.
Stable Expression of COUP-TF Does Not Lead to trans-RAdependent Inhibition of AP-1 Activity in HT-1376 Cells Lacking RAR␣/␤ Expression-We also evaluated whether stable expression of COUP-TF affected TPA-induced c-Jun expression in HT-1376 bladder cancer cells (Fig. 7). The stable clone, HT-1376/COUP#7, which expresses high levels of COUP-TF, was analyzed for the effect of trans-RA on TPA activity (Fig. 7). RAR Expression Is Required for COUP-TF to Facilitate Antagonism of AP-1 Activity by trans-RA-Our observations that transient (Fig. 1B) or stable (Fig. 7) expression of COUP-TF in HT-1376 cells failed to modulate antagonism of AP-1 activity by trans-RA led us to investigate RAR expression. Unlike MDA-MB231 cells that express RAR␣ (48) and RAR␤ when COUP-TF is expressed (35), HT-1376 cells did not express detectable RAR␣ or RAR␤ (Fig. 7), although RAR␥ was expressed (data not shown). This result suggested that modulation of trans-RA-induced anti-AP-1 activity by COUP-TF might require RAR␣ or RAR␤. The effect of RAR␣ expression on the ability of COUP-TF to regulate trans-RA activity was then examined using transient transfection in HT-1376 cells. The Ϫ73Col-CAT reporter was transfected into HT-1376 cells with or without c-Jun and COUP-TF and/or RAR␣ vector. Transfected cells were treated with or without trans-RA. As shown in Fig. 8A, transfected COUP-TF repressed AP-1 activity in the absence of trans-RA. In contrast, transfected RAR␣ (10 ng) led to inhibition of AP-1 activity in a trans-RA dependent manner. The addition of COUP-TF significantly enhanced the trans-RAinduced inhibition of AP-1 activity by RAR␣ (Fig. 8A). Similar results were observed using HeLa cells (Fig. 8B). Thus, efficient inhibition of AP-1 activity by trans-RA activity by COUP-TF requires both COUP-TF and RAR␣. DISCUSSION Retinoids are effective growth inhibitors of cancer cells. Inhibition of AP-1 activity has been proposed as one mechanism by which retinoids exert their anticancer effects (6). Despite extensive studies in the last few years, how retinoids specifically antagonize AP-1 activity remains largely unknown. Here, we provide evidence that COUP-TF is involved in regulating the antagonism of AP-1 transactivational activity by trans-RA. COUP-TF, by physically interacting with c-Jun, inhibits AP-1 DNA binding and transactivation and is required for efficient inhibition of AP-1 activity by liganded RARs. Our results suggest that COUP-TF plays a role in the cross-talk between retinoid and AP-1 signaling pathways.
We recently reported that the expression of COUP-TF positively correlates with the inhibition of the growth of various cancer cell lines by trans-RA and that COUP-TF is underexpressed in many trans-RA-resistant cancer cell lines (35). Stable expression of COUP-TF in COUP-TF-negative cancer cells restores their sensitivity to trans-RA, demonstrating that COUP-TF can mediate anticancer effects of trans-RA (35). By studying anti-AP-1 activity of trans-RA in various cancer cell lines, we found a close correlation between COUP-TF expression and the anti-AP-1 activity of trans-RA (Fig. 1). In COUP-TF-positive ZR-75-1, T-47D, and Calu-6 cancer cell lines (35), trans-RA strongly inhibited the ability of TPA to activate transcription of the collagenase promoter, whereas in COUP-TFnegative MDA-MB231, H292, and HT-1376 cell lines, trans-RA failed to suppress TPA activity (Fig. 1). This finding is consistent with a previous study showing that trans-RA effectively inhibited AP-1 activity in ZR-75-1 and T-47D cells but not in MDA-MB231 cells (49). The requirement of COUP-TF in trans-RA-mediated AP-1 inhibition was further demonstrated by our findings that transient expression of COUP-TF in COUP-TFnegative cells (Fig. 1B) restored the ability of trans-RA to inhibit TPA-induced AP-1 transactivation and that stable expression of COUP-TF in MBA-MB231 cells enabled trans-RA to inhibit TPA-induced c-Jun expression (Fig. 6).
We have reported that COUP-TF expression contributes to growth inhibition by trans-RA in cancer cells because COUP-TF on binding to the RAR␤ promoter induces RAR␤ expression (35). RAR␤ is reported to be a potent AP-1 inhibitor (24), suggesting that COUP-TF-induced RAR␤ probably contributes to the inhibition of AP-1 activity (Fig. 6). Our observations that COUP-TF can effectively interact with c-Jun in vitro (Figs. 3 and 4) and inhibit AP-1 transcriptional activity on transient transfection (Fig. 2) also suggest that COUP-TF is directly involved in antagonizing AP-1 activity.
Inhibition of AP-1 activity appears to be a common characteristic of nuclear receptors, as has been demonstrated for the progesterone, estrogen, androgen, thyroid hormone, glucocorticoid, and retinoid receptors, which can functionally interact with the AP-1 complex. This study reveals that the orphan receptor COUP-TF behaves similarly. The molecular basis of the interaction between nuclear receptors and AP-1 pathways remains largely unknown. Models proposed include direct protein-protein interaction (8,9,(13)(14)(15), inhibition of Jun N-terminal kinase activity (31), and competition for CBP (32). Our results of GST pull-down assays (Figs. 4B and 5) and mutation analysis (Figs. 4 and 5) support the first model. The requirement for the COUP-TF DBD for c-Jun binding is similar to observations that the DBDs of glucocorticoid receptor (14,15) and RAR (47) are essential for their interaction with AP-1. Interestingly, the in vitro DNA-binding study showed that COUP-TF is a more effective inhibitor of AP-1 binding than RAR␣ (Fig. 3) on the basis of the levels of each receptor to exert this effect.
In contrast to inhibition of AP-1 activity by other receptors that require their respective ligands, COUP-TF effectively inhibited AP-1 transactivation in the absence of any ligand (Fig.  2). Such a ligand-independent inhibition of AP-1 activity may restrict expression levels of AP-1-responsive genes in cancer cells (Figs. 6 and 7). Interestingly, we observed that COUP-TF expression potentiates antagonism of AP-1 activity by trans-RA. However, regulation of trans-RA activity by COUP-TF appears to require RAR␣ or RAR␤, because transiently transfected COUP-TF was unable to confer anti-AP-1 activity to trans-RA in HT-1376 cells lacking RAR␣ or -␤ (Fig. 8A) or HeLa cells (Fig. 8B) unless RAR␣ was cotransfected. In addition, stable expression of COUP-TF restored trans-RA-induced anti-AP-1 activity in MDA-MB231 cells (Fig. 6) having RARs (48) but not in HT-1376 cells lacking RAR␣ and -␤ (Fig. 7). Thus, expression of both RAR␣/␤ and COUP-TF is required for optimal inhibition of AP-1 activity by trans-RA.
Although we do not fully understand how COUP-TF and RAR␣ coexpression maximizes inhibition of AP-1 activity by trans-RA, our finding of physical interaction between COUP-TF and RAR␣ (35,50) suggests its involvement in this process. Based on our observation that COUP-TF more effectively inhibits c-Jun binding to DNA than RAR␣ (Fig. 3), it is tempting to speculate that COUP-TF, with its ability to interact with both RAR (35,50) and c-Jun (Fig. 3), may function as a bridging factor to mediate RAR/AP-1 interaction. In this model, the COUP-TF/RAR heterodimer would strongly interact with AP-1 in the presence of trans-RA to inhibit the DNA binding of AP-1 and its transactivation. Similarly, COUP-TF, by interacting with c-Jun, might also potentiate the inhibition of RAR␣/␤ transactivation by AP-1. Unfortunately, perhaps due to rapid formation of the COUP-TF homodimer after COUP-TF expression in vitro (40,51), we were unsuccessful in demonstrating the interaction between RAR/COUP-TF heterodimer and c-Jun in gel shift or GST pull-down assays (data not shown). Another possible mechanism for inhibiting AP-1 transactivation by COUP-TF may involve its recruiting of a transcriptional corepressor. COUP-TF has been widely considered as a potent negative transcriptional regulator (40 -43) due to its effective interaction with corepressors, such as N-CoR and SMRT (52). The interaction between COUP-TF and both RAR and AP-1 may recruit transcriptional corepressors to the complex, thereby enhancing the mutual repression of AP-1 and RAR transcriptional activity. Interestingly, COUP-TF interacts with a variety of nuclear receptors, including RXR (41), TR (40,50), estrogen receptor (53,54), and peroxisome proliferator-activated receptor (55). It remains to be determined whether and how COUP-TF is involved in modulating the anti-AP-1 activity by their ligands.