Ligand-Activated Retinoic Acid Receptor Inhibits AP-1 Transactivation by Disrupting C-Jun/C-Fos Dimerization

Ligand-Activated Retinoic Acid Receptor Inhibits AP-1 Transactivation by Disrupting c-Jun/c-Fos Dimerization

Xiao-Feng Zhou, Xi-Qiang Shen, and Lirim Shemshedini Downloaded from https://academic.oup.com/mend/article/13/2/276/2741651 by guest on 27 September 2021 Department of Biology University of Toledo Toledo, Ohio 43606

In the presence of retinoic acid (RA), the retinoid volved in epithelial differentiation (2), has a central role receptors, retinoic acid receptor (RAR) and retinoid as a tissue-specific morphogen during embryogenesis X receptor (RXR), are able to up-regulate transcrip- (3), and represses malignant transformation of epithe- tion directly by binding to RA-responsive elements lial cells both in vitro (4) and in vivo (5). These diverse on the promoters of responsive genes. Liganded and pleiotropic effects of RA are mediated through the RARs and RXRs are also capable of down-regulat- binding of RA to a family of nuclear receptors, which, ing transcription, but, by contrast, this is an indi- together with the receptors for steroid and thyroid rect effect, mediated by the interaction of these hormones and vitamin D, form the nuclear receptor nuclear receptors not with DNA but the transcrip- superfamily (reviewed in Refs. 6–10). Nuclear recep- tion factor activating protein-1 (AP-1). AP-1 is a tors comprise the largest family of transcription fac- dimeric complex of the protooncoproteins c-Jun tors, which, upon binding of their cognate ligands, and c-Fos and directly regulates transcription of modulate transcription initiated from promoters of tar- genes important for cellular growth. Previous in get genes by interacting with specific cis-acting, DNA vitro results have suggested that RARs can block response elements. AP-1 DNA binding. Using a mammalian two-hybrid In contrast to this positive effect of nuclear recep- system, we report here that human RAR␣ (hRAR␣) tors on transcription, which requires receptor-DNA in- can disrupt in a RA-dependent manner the homo- teractions, the retinoid receptors and other nuclear and heterodimerization properties of c-Jun and c- receptors can negatively affect gene expression with- Fos. This inhibition of dimerization is cell specific, out binding to DNA, via their ability to functionally occurring only in those cells that exhibit RA- interact with the transcription factor AP-1 (activating induced repression of AP-1 transcriptional activity. protein-1) (reviewed in Refs. 11–14). AP-1 consists of Furthermore, this mechanism appears to be spe- homodimers and heterodimers of Jun (c-Jun, v-Jun, cific for the RARs, since another potent inhibitor of JunB, and JunD), Fos (c-Fos, v-Fos, FosB, Fra1, and AP-1 activity, the glucocorticoid receptor, does not Fra2), or activating transcription factor (ATF2, ATF3/ affect AP-1 dimerization. Our data argue for a novel LRF1, B-ATF) bZIP (basic region leucine zipper) pro- mechanism by which RARs can repress AP-1 DNA teins (15–17). The transcription of the c-jun and c-fos binding, in which liganded RARs are able to inter- genes, encoding the major components of AP-1, is fere with c-Jun/c-Jun homodimerization and rapidly induced upon stimulation of cellular prolifera- c-Jun/c-Fos heterodimerization and, in this way, tion (37, 38). Like nuclear receptors, AP-1 activates may prevent the formation of AP-1 complexes ca- transcription of target genes by binding to specific pable of DNA binding. (Molecular Endocrinology promoter elements, called TREs (TPA-responsive ele- 13: 276–285, 1999) ments) (17, 18). TPA (12-O-tetra-decanoyl-phorbol- 13-acetate) is a tumor promoter that induces the expression of the c-fos and c-jun genes (19, 20) and INTRODUCTION thereby indirectly stimulates the expression of AP-1 target genes. Retinoic acid (RA), the most biologically active natural Both positive and negative regulatory interactions metabolite of vitamin A (retinol), exerts profound ef- between nuclear receptors and c-Jun/c-Fos have fects on vertebrate development, cellular differentia- been reported (reviewed in Refs. 11–14). The first re- tion, and homeostasis (reviewed in Ref. 1). RA is in- sults showed an inhibition of glucocorticoid receptor (GR)-induced transcription by either c-Fos or c-Jun 0888-8809/99/$3.00/0 (21–24). We (25) and others (26–31) have shown that Molecular Endocrinology Copyright © 1999 by The Endocrine Society this type of interference is not restricted to the GR, but

276 RAR Blocks c-Jun/c-Fos Dimerization 277

seems to be a common characteristic of nuclear re- Using a mammalian two-hybrid system, we provide ceptors, including the receptors for the hormones pro- evidence here that RAR, but not GR, is able to disrupt gesterone (PR), estrogen (ER), androgen (AR), and thy- in vivo c-Jun/c-Fos dimerization in a ligand-dependent roid (TRs), and the RA (RARs, RXRs). Conversely, the manner. This effect is not only receptor specific, but activation of the collagenase and stromelysin genes by also cell specific, paralleling what has been reported AP-1 is repressed in a ligand-dependent manner by previously with RA-induced inhibition of AP-1 tran- several receptors, including GR (21–24), PR (25), AR scriptional activity (27). Our results suggest that c-Jun/ (25), ER (25), TR (28), and RARs/RXRs (26, 27, 29). By c-Fos dimerization may be a third target of nuclear contrast, coexpression of c-Jun, c-Fos, and ER receptor-mediated repression of AP-1 that may be causes synergistic activation of the ovalbumin gene specific for the transrepression activity of RARs and

(32). GR has been shown to potentiate c-Jun-acti- may partially explain the receptor- and cell-specific Downloaded from https://academic.oup.com/mend/article/13/2/276/2741651 by guest on 27 September 2021 vated transcription from the proliferin- regulatory ele- nature of this repression. ment (33). Similarly, transfected c-Jun enhances AR- induced transactivation, but it does so independently of promoter or cell type specificity (25, 34, 35). RESULTS In contrast to the interaction between AP-1 and AR, PR, or GR, which is nonmutual and can be either RA-Bound human RAR␣ (hRAR␣) Inhibits AP-1 negative or positive (25, 32, 33), the interaction be- Activity in a Cell-Specific Manner tween AP-1 and the retinoid receptors is mutual and solely inhibitory. c-Jun and c-Fos, either individually or The interaction between nuclear receptors and AP-1 together, have been shown to repress the transcrip- has been shown previously to be dependent on cell tional activity of RAR and/or RXR (27). Conversely, type (Refs. 25 and 27; reviewed in Ref. 11). To test for both RAR/RXR heterodimers or homodimers of either cell specificity in RARs’ ability to inhibit AP-1 transcrip- can inhibit AP-1 transactivation of several AP-1- tional activity, cells were transiently transfected with ␣ responsive promoters (27, 29). Indeed, the RAR/RXR expression plasmids for c-Jun and hRAR and the antagonism of AP-1 has been directly implicated in the AP-1-inducible reporter TRE-tk-CAT (34). In keeping ␣ regulation of collagenase (27, 29) and stromelysin (26), with previously published data (27), hRAR inhibited in two genes that play key roles in tumor potential and a ligand- and dose-dependent manner exogenous c- invasiveness. Jun activity in HeLa cells (Fig. 1A), but not in Cos cells While the molecular bases of these diverse regula- (Fig. 1B). The same difference in activity was observed tory interactions between nuclear receptors and AP-1 on endogenous AP-1 activity (data not shown). are not known, recent studies provide several attrac- Mammalian Two-Hybrid System Can Be Used to tive models. Based on the demonstration that CREB- Measure c-Jun/c-Fos Dimerization in Vivo binding protein (CBP) and the related p300 can act as transcriptional coactivators for both nuclear receptors The yeast two-hybrid system has been used previ- (36–38) and AP-1 (39, 40), it has been proposed that ously to measure in vivo protein-protein interactions the nuclear receptor-AP-1 antagonism depends on (45). In the current study, we have used a similar competition for limiting amounts of these two coacti- system to measure c-Jun/c-Fos heterodimerization vator proteins (36). While this model may explain some and c-Jun/c-Jun homodimerization in cultured mam- of the observations made, it is not able to explain all of malian cells. In our system, two fusion proteins are the cell, promoter, and receptor specificity that has expressed, one containing the GAL4 DNA-binding do- been observed in the nuclear receptor-AP-1 interac- main (DBD) fused to either full-length c-Jun or only its tions (25). More recently, it has been suggested that bZIP region and the other containing the VP16 trans- GR, RARs, and TRs can block AP-1 activity by inhib- activation domain fused to either full-length c-Fos or iting the activity of Jun amino-terminal kinase (JNK) only its bZIP region (Fig. 2A). HeLa cells were trans- (41), which enhances c-Jun transcriptional activity by fected with expression plasmids for these different phosphorylating Ser63/73 (42, 43). However, this fusion proteins, and dimerization was monitored with model also is unable to account for the diverse nature the GAL4-inducible reporter 17M-tk-CAT. While GAL- of the interactions between nuclear receptors and cJun had a weak activity and VP16-cFos had no mea- AP-1. Others have argued, based on in vitro results, surable activity, these two fusion proteins together that AP-1 and receptors mutually inhibit each other’s resulted in a 14-fold stimulation in transcription (Fig. DNA-binding ability (21, 24, 33). However, Konig et al. 2B), demonstrating a strong in vivo interaction be- (44) have provided strong evidence that in vivo DNA tween transfected c-Jun and c-Fos. If either c-Fos or binding, at least for GR and AP-1, is not affected, since c-Jun or both were absent from the fusion proteins, no the in vivo footprint of either of these transcriptional interaction was detected (Fig. 2B). Since earlier work activators in the presence of the other did not change. (46) has shown that the respective bZIP regions of Thus, the nuclear receptor-AP-1 antagonism may de- these two protooncoproteins are sufficient for het- pend on multiple mechanisms with the involvement of erodimerization, we tested these same regions in cell- and receptor-specific factors. our system. Indeed, GAL-cFos(137–216) and VP16- MOL ENDO · 1999 Vol 13 No. 2 278 Downloaded from https://academic.oup.com/mend/article/13/2/276/2741651 by guest on 27 September 2021

Fig. 2. c-Jun/c-Fos Heterodimerization Is More Efficient Than Is c-Jun-c-Jun Homodimerization in Vivo A, A schematic representation showing full-length c-Jun and c-Fos and truncations of these two proteins, which were Fig. 1. Ligand-Bound RAR Inhibits AP-1 Activity in HeLa used as fusion proteins with GAL4 or VP16 in the mammalian Cells, but Not Cos Cells two-hybrid screen. Numbers represent amino acid residues. ␮ HeLa (A) and Cos (B) cells were transfected with 1 gofthe B, HeLa cells were transfected with 1 ␮g of the 17 M-tk-CAT ␮ TRE-tk-CAT reporter plasmid together with 1 g of c-Jun and reporter plasmid together with 1 or 3 ␮g each of expression ␮ ␣ 1, 3, or 5 g of hRAR expression plasmids. Cells receiving plasmids for GAL-cJun or VP16-cJun or 1 ␮g each of ex- ␣ Ϫ7 hRAR were treated with 10 M AT-RA as indicated. Note pression plasmids for VP16-cFos, GAL-cFos(137–216), that CAT activity is represented relative to activity of first VP16-cJun(237–331), GAL-DBD, or VP16. Note that CAT ac- condition, which was set to 1. tivity is represented relative to activity of GAL-cJun, which was set to 1.

cJun(237–331) exhibited a dimerization capacity that is comparable to that observed with the full-length c-Jun homodimerization (Fig. 3B). Since the bZIP re- proteins (Fig. 2B). As further evidence, we measured gions of c-Jun and c-Fos are sufficient for dimerization c-Jun/c-Jun homodimerization by coexpressing GAL- (see Fig. 2B), we wanted to determine whether these cJun and VP16-cJun(237–331). In agreement with pre- regions are also sufficient for the negative effect of vious work (47), c-Jun/c-Jun homodimerization was hRAR␣ on AP-1 dimerization. Liganded hRAR␣ was significantly weaker than c-Jun/c-Fos heterodimeriza- also able to repress the dimerization between GAL- tion, even with higher (ϳ3-fold) amounts of c-Jun fu- cFos(137–216) and VP16-cJun (Fig. 4A) and that be- sion proteins expressed (Fig. 2B). Identical results tween GAL-cFos(137–216) and VP16-cJun(237–331) have been obtained in Cos cells (data not shown). (Fig. 4B). Note that hRAR␣, with or without AT-RA, did These results together indicate that our mammalian not affect the activity of GAL-VP16 (Fig. 3C), excluding two-hybrid system is faithfully measuring the ability of a possible hRAR␣ interference of either GAL(DBD) or c-Jun and c-Fos to dimerize in vivo. VP16 function. These results clearly show that hRAR␣ is able to block in a ligand-dependent manner both hRAR␣ Disrupts c-Jun/c-Fos Dimerization in Vivo hetero- and homodimerization of AP-1, and thereby in a Ligand-Dependent Manner blocking in vivo DNA binding, by possibly targeting the respective bZIP regions of c-Jun and c-Fos. To determine whether RAR can affect the in vivo dimerization between c-Jun and c-Fos, HeLa cells hRAR␣ Disruption of c-Jun/c-Fos Dimerization Is were transfected with an expression plasmid for Cell Specific Ϫ7 hRAR␣ and treated with 10 M all-trans-retinoic acid (AT-RA). hRAR␣ was able to severely block dimeriza- The cell-specific nature of nuclear receptor-induced tion between full-length c-Jun and c-Fos (Fig. 3A). This inhibition of AP-1 transcriptional activity prompted us negative effect occurred only in the presence of AT- to study hRAR␣-induced inhibition of AP-1 dimeriza- RA, since there was no effect by hRAR␣ in the ab- tion in several different cells. As observed before (see sence of AT-RA (Fig. 3A). A similar ligand-dependent Fig. 4B), liganded hRAR␣ is able to block the dimer- RAR␣-induced inhibition was observed on c-Jun/ ization between the bZIP regions of c-Jun and c-Fos in RAR Blocks c-Jun/c-Fos Dimerization 279 Downloaded from https://academic.oup.com/mend/article/13/2/276/2741651 by guest on 27 September 2021

Fig. 4. RAR Inhibition of c-Jun/c-Fos Dimerization Is Tar- geted to the bZIP Regions of These Protooncoproteins HeLa cells were transfected with 1 ␮g of the 17M-tk-CAT reporter plasmid and 1 ␮g of hRAR␣ expression plasmid together with 1 ␮g each of expression plasmids for GAL- cFos(137–216) and either VP16-cJun (A) or VP16-cJun(237– Ϫ 331) (B). Cells receiving hRAR␣ were treated with 10 7 M AT-RA as indicated. Note that CAT activity is represented relative to activity of GAL-cFos(137–216), which was set to 1.

Fig. 3. Ligand-Bound RAR Disrupts Both AP-1 Homodimer- ization and Heterodimerization HeLa cells were transfected with 1 ␮g of the 17M-tk-CAT tional activation domain and the LexA DBD joined to reporter plasmid and 1 ␮g of hRAR␣ expression plasmid either c-Fos(137–216) for heterodimerization or full- together with 1 ␮g each of expression plasmids for GAL-cJun length c-Jun for homodimerization. B42-cJun(237– and VP-cFos for heterodimerization (A), 1 ␮g each of expres- 331) interacted strongly with both LexA-cFos(137– sion plasmids for GAL-cJun and VP16-cJun for homodimer- 216) and LexA-cJun, but cotransformed hRAR␣, either ␮ ␮ ization (B), or 1 g GAL-cJun and 0.5 g of GAL-VP16 in the absence or presence of AT-RA, had no signifi- expression plasmids (C). Cells receiving hRAR␣ were treated Ϫ cant effect (Fig. 6A). Accordingly, liganded hRAR␣ was with 10 7 M AT-RA as indicated. Note that CAT activity is unable to repress the transcriptional activity of LexA- represented relative to activity in the absence of activator, which was set to 1. cJun (Fig. 6B). These results together suggest that the ability of hRAR␣ to repress AP-1 dimerization is dependent on cell-specific factors that may not be found HeLa cells (Fig. 5A); note that the ligand-dependent in either Cos or yeast cells. activity observed in the absence of transfected hRAR␣ is likely due to endogenous receptor. Importantly, hRAR␣ Disruption of AP-1 in Vitro DNA Binding Is however, when the same experiment was repeated in Also Cell Specific Cos cells, there was no detectable effect of hRAR␣, either in the absence or presence of AT-RA, on c-Jun/ Our transfection results above show that liganded c-Fos dimerization (Fig. 5B). This lack of RAR activity hRAR␣ can block AP-1 dimerization in a cell-specific in Cos cells on AP-1 dimerization correlates with the manner (see Fig. 5). Disruption of AP-1 dimerization lack of RAR activity on AP-1 transcriptional activity in should lead to abolishment of AP-1 sequence-specific these same cells (see Fig. 1B). To further examine the DNA binding. To analyze this, we measured the in vitro cell specificity, we used the yeast two-hybrid system DNA-binding ability of endogenous AP-1 from either to test the activity of hRAR␣ on AP-1 dimerization. In HeLa or Cos cells, which had been transfected with this system, we analyzed the interaction between c- hRAR␣, in the absence or presence of RA. Extracts Jun(237–331) fused to the B42 (acid blob) transcrip- from both cells exhibited significant AP-1 DNA-bind- MOL ENDO · 1999 Vol 13 No. 2 280 Downloaded from https://academic.oup.com/mend/article/13/2/276/2741651 by guest on 27 September 2021

Fig. 5. RAR Disrupts c-Jun/c-Fos Dimerization in a Cell- Specific Manner HeLa (A) and Cos (B) cells were transfected with 1 ␮gof the TRE-tk-CAT reporter plasmid together with 1 ␮g each of expression plasmids for GAL-cFos(137–216), VP16- cJun(237–331), or hRAR␣. Cells receiving hRAR␣ were Ϫ treated with 10 7 M AT-RA as indicated. Note that CAT activity is represented relative to activity of GAL-cFos(137–231), which was set to 1.

ing activity (Fig. 7, lanes 1 and 3). This AP-1 acitivity Fig. 6. RAR Is Unable to Disrupt c-Jun/c-Fos Dimerization was confirmed by both addition of antibody, which or LexA-cJun Activity in Yeast disrupts c-Jun DNA binding, (compare lanes 5 and 6) Yeast cells EGY48 were transformed with 1 ␮gofthe and competition with unlabeled DNA (compare lanes 7 pSH18–34 reporter plasmid and 1 ␮g each of expression and 8). Importantly, RA-bound hRAR␣ was able to plasmids for hRAR␣, LexA-cJun, LexA-cFos(137–216), and VP16-cJun (A) or hRAR␣, LexA-cJun, and LexA (B). Cells repress AP-1 DNA binding in HeLa cells (compare Ϫ receiving hRAR␣ were treated with 10 6 M AT-RA as indi- lanes 1 and 2), but not Cos cells (compare lanes 3 and cated. Bars in panel A represent the following: open, B42; 4), paralleling what was observed with the mammalian black, B42-cJun(237–331); gray, B42-cJun(237–331) ϩ two-hybrid system (see Fig. 5). hRAR␣; stippled, B42-cJun(237–331) ϩ hRAR␣ ϩ AT-RA. hRAR␣ Disruption of c-Jun/c-Fos Dimerization Is Receptor Specific shown). Importantly, GR, with or without Dex, had no significant influence on dimerization between GAL- Several nuclear receptors, including that for glucocor- cFos(137–216) and VP16-cJun(237–331) in HeLa cells ticoids (GR), have been shown to block AP-1 tran- (Fig. 8) Thus, inhibition of AP-1 dimerization in vivo is scriptional activity in a ligand-dependent manner. In- not only cell specific, but also receptor specific. terestingly, however, GR has been shown to have no effect on in vivo AP-1 DNA binding, as measured by in vivo footprint analysis (44). Therefore, we tested the activity of GR in our mammalian dimerization assay in DISCUSSION HeLa cells, which have previously been shown to exhibit dexamethasone (Dex)-induced repression of Previous in vitro DNA binding studies suggested that AP-1 transcriptional activity (Refs. 20–24 and data not RARs are able to antagonize AP-1 activity by blocking RAR Blocks c-Jun/c-Fos Dimerization 281

Fig. 8. GR Is Unable to Disrupt c-Fos/c-Jun Dimerization Downloaded from https://academic.oup.com/mend/article/13/2/276/2741651 by guest on 27 September 2021 HeLa cells were transfected with 1 ␮g of the 17M-tk-CAT reporter plasmid and 1 ␮g each of expression plasmids GAL- cFos(137–216) and/or VP16-cJun(237–331), and 1, 3, or 5 ␮g of hGR expression plasmid. Cells receiving hGR were treated Ϫ with 10 7 M Dex as indicated. Note that CAT activity is represented relative to activity of GAL-cFos(137–216), which was set to 1.

via the conserved bZIP regions found within c-Jun, c-Fos, and their protein families (reviewed in Ref. 16). Our transfection studies show that the bZIP regions of c-Jun and c-Fos are sufficient for the ligand-dependent RAR inhibition of dimerization. Interestingly, the bZIP regions of c-Jun and c-Fos have been previously Fig. 7. RAR Disrupts AP-1 in Vitro DNA Binding in a Cell- shown to be essential for the transcriptional interac- Specific Manner tions between these protooncoproteins and nuclear HeLa and Cos cells were transfected with 10 ␮g of hRAR␣. receptors (23, 25). Thus, the bZIP regions of c-Jun and Ϫ These cells were treated with 10 7 M AT-RA as indicated. c-Fos may provide a common surface through which Nuclear extracts were tested for AP-1 DNA binding using a nuclear receptors can engage in a protein-protein in- gel mobility shift assay. Antibody disruption analysis was teraction with c-Jun and c-Fos. done with either an anti-c-Jun or antiandrogen receptor (AR) Several studies (21, 24, 27, 33, 48), using chemical antibody. Note that the anti-c-Jun antibody is directed cross-linking and coprecipitation approaches, have against the bZIP region and thus disrupts c-Jun DNA binding; suggested that nuclear receptors can physically asso- therefore, no supershift is detectable. DNA competition was ciate with both c-Jun and c-Fos, and thus block their carried out with a 50-fold excess of unlabeled DNA elements (TRE, ARE). The arrow points to the AP-1-TRE complex, and ability as AP-1 to bind to DNA. However, these pro- FP represents the free probe. The protein amount (in micro- tein-protein interactions appear to be weak or indirect, grams) used in each reaction is the following: lane 1, 11.5; since we and others (22, 23, 25, 26, 34) have been lane 2, 11.8; lane 3, 12.8; lane 4, 14.1; lane 5, 11.4; lane 6, unable to detect a stable interaction between RAR and 11.5; lane 7, 11.5; lane 8, 11.5. the AP-1 components. In fact, our current data support the model proposed by Pfahl (11), that additional factors, which appear to be expressed in a cell-spe- its DNA-binding ability (21, 23, 24, 27–29). We show in cific manner, are essential for the antagonism between this paper that this interference is cell specific, occur- receptors and AP-1. Our previous results showed that ring in HeLa but not Cos cells, and, significantly, pro- the direction and magnitude of the nuclear receptor- vide the first in vivo evidence supporting this mecha- AP-1 interaction is dependent on the type of cell, nism of transcriptional interference. Moreover, these promoter, receptor, and AP-1 component (25). We data provide a potential basis for some of the cell- and now provide evidence that liganded hRAR␣ is able to receptor-specific effects that have been reported (25, inhibit AP-1 dimerization in cell-specific manner, oc- 27). Our results indicate that hRAR␣ is able to disrupt curring in HeLa cells, in which this receptor can inhibit in a RA-dependent manner the in vivo dimerization AP-1 transcriptional activity, but not in Cos cells, capacity of c-Jun with either itself or c-Fos, which where it is has no effect on AP-1 transcriptional activ- would preclude the formation of DNA-binding-compe- ity. Further, liganded hRAR␣ does not disrupt c-Jun/ tent AP-1 complexes. In support of previous work (27), c-Fos homo- and heterodimerization reconstituted in this RA-dependent effect on dimerization is cell spe- yeast, and, accordingly, as would be predicted, it is cific, paralleling the previously observed activity of unable to repress LexA-cJun transcriptional activity in RARs on AP-1 transcriptional activity. yeast. Thus, it is possible that RARs, and other nuclear How might RAR prevent the formation of AP-1 ho- receptors, can directly associate with c-Jun and/or mo- and heterodimers? AP-1 dimerization is mediated c-Fos, but the affinity of this direct interaction is not MOL ENDO · 1999 Vol 13 No. 2 282

sufficient in vivo to modulate transcription. Additional including TR (36, 52), GR (53), ER (54), PR (53, 55), and factors, expressed in a tissue-specific manner, may be AR (51), then these same receptors should be in a needed to stabilize the RAR-AP-1 interaction and thus competitive and mutually inhibitory interaction with the prevent AP-1 dimerization (see Fig. 9 for a scheme). An AP-1 components c-Jun and c-Fos. Although this is example of a nonreceptor, cell-specific factor mediat- generally the case, there are several exceptions. First, ing the interaction of a nuclear receptor with another the AP-1 interaction with AR can be either negative or transcription factor comes from a study on the trans- positive, depending on the AP-1 component. Indeed, activation of the RAR␤2 promoter. Berkenstam et al. our laboratory has shown that c-Jun strongly en- (49) have found that this promoter is synergistically hances AR-induced transcription and c-Fos can inhibit activated by RAR and the TATA box-binding protein this activity, independent of cell or promoter specificity

(TBP) in embryonal carcinoma (EC) cells but not Cos, (25, 34, 35). On GR-inducible promoters, c-Jun gen- Downloaded from https://academic.oup.com/mend/article/13/2/276/2741651 by guest on 27 September 2021 and that this synergy can be restored in Cos cells by erally blocks GR activity except in several T cell lines ectopically expressing E1A (50). It is also possible that (56). On the AP-1-inducible proliferin promoter, which RAR has a secondary effect, by inducing the expres- contains a composite response element, GR and min- sion of a protein that is directly involved in inhibiting eralocorticoid receptor (MR) can act either coopera- AP-1 dimerization. tively or antagonistically with AP-1, depending on the In the case of nuclear receptor- and AP-1-mediated identity of receptor (GR, MR) and AP-1 component transcription, there is compelling evidence that CBP (c-Jun/c-Jun or c-Jun/c-Fos) (33). In view of this com- and p300 can act as coactivators for both pathways plexity, it is likely that additional modes of interaction (36–40). It was recently reported that CBP, and per- occur between nuclear receptors and AP-1. Recently, haps p300, can also facilitate the homodimerization of it was reported that several nuclear receptors, includ- AR (51). Since CBP and p300 can associate with both ing RARs, are able to repress JNK activation of c-Jun, c-Jun and c-Fos, it is possible that they are serving a providing a second mechanism of nuclear receptor- similar role in AP-1 dimerization. However, our mam- induced inhibition (41). Our data suggest that an ad- malian two-hybrid assay detected no p300 influence ditional mode of action exists, in which RARs can on either c-Jun/c-Fos dimerization or RA-induced dis- disrupt the ability of c-Jun to homodimerize with itself ruption of dimerization (data not shown). It has been and heterodimerize with c-Fos. In contrast to the other suggested that RARs and other nuclear receptors can mechanisms, the last one closely parallels the cell- compete with AP-1 for limiting amounts of CBP and specific nature of RAR-induced repression of AP-1 p300 (36), thereby resulting in transcriptional efficacy transcriptional activity, appears to be specific for going to one activator at the expense of the other. RARs, and may depend on involvement of cell-specific Since CBP and p300 are known to interact with and factors. Thus, it appears that RARs can use multiple mediate the activities of several nuclear receptors, mechanisms by which to blunt the transcriptional activity of AP-1. It is possible that one of these proposed mechanisms is the preferred choice in vivo or that these different modes of actions may cooperate to ensure the appropriate expression of AP-1-responsive genes. RARs’ effect on AP-1 activity has been proposed to be responsible for the clinical effects of retinoids as antineoplastic, antiinflammatory, and immunosup- pressive agents (26, 27, 29). In this regard, it is note- worthy that selective retinoids have been reported that allow a separation of the transactivation and transrepression activities of RARs (57, 58). Future work will be needed to determine whether some of these anti-AP- 1-selective retinoids act by inducing RAR-mediated disruption of c-Jun/c-Fos dimerization.

MATERIALS AND METHODS

Fig. 9. A Model of How Ligand-Bound RAR Can Inhibit AP-1 Plasmids Transcriptional Activity In the absence of RA-activated RAR, c-Jun and c-Fos are ␣ able to dimerize to form AP-1, and this dimeric complex can For mammalian expression, hRAR (59), hGR (59), GAL-VP16 (59), and c-Jun (25) in pSG5 have been described. GAL-c-Jun bind to promoter elements (AP-1 elements) of AP-1-respon- and GAL-c-Fos were also expressed from the mammalian sive genes. When RA is bound to RAR and the necessary expression plasmid pGAL0 (60). GAL-cFos(137–216) was cell-specific factor(s) (factor x) is present, the c-Jun/c-Fos constructed by PCR amplification of c-Fos amino acids 137– heterodimer does not form; thus, the AP-1 element is vacant 216 using the upstream oligo 5Ј-GATCGAATTCATGGAA- and the gene remains transcriptionally silent. GAGAAACGGAGA-3Ј and the downstream oligo 5Ј-GATCG- RAR Blocks c-Jun/c-Fos Dimerization 283

GATCCTCACATCTCCTCTGGGAA-3Ј and inserting into KCl, and 150,000 cpm of 32P-labeled probe (5Ј-TCGAGTT- pGAL0. VP16-cJun was constructed by inserting full-length GCATGAGTCAGACATCGATTGCA-3Ј). After the addition of x c-Jun into the BamHI/BglII sites of VP16/pTL1, pTL1 (34) ␤-gal units of nuclear extract, the reactions were gently vor- containing the activation domain of VP16. VP16-cJun(237– texed and incubated for 15 min at 25 C. The samples were 331) was constructed by PCR amplification of c-Jun amino run on a 6% polyacrylamide gel for 1.5 h at room tempera- acids 237–331 using the upstream oligo 5Ј-GATCGAATTC- ture, after which the gel was dried and exposed to autora- CTCGAGATGGGCGAGACACCGCCC-3Ј and the downstream diography. In some reactions, 1 ␮l of either anti-c-Jun (sc-44, oligo 5Ј-GATCGGATCCTCAAAATGTTTGCAA-3Ј and inserting Santa Cruz Biotechnology, Santa Cruz, CA) or anti-hAR (sc- into VP16/pTL1. 815, Santa Cruz Biotechnology) were added before addition For yeast expression, the expression plasmids pEG202 of probe. Other reactions received a 50-fold excess of either (45), pJG4–5 (45), and pYE10 (61) were used. LexA- unlabeled AP-1 element (given above) or androgen-response cFos(137–231) was constructed by digestion of cFos(137– element (ARE) (5Ј-GATCCAAAGTCAGAACACAGTGTTCT- 231) from pGAL0 and insertion into pEG202. LexA-cJun was GATCAAAGA-3Ј). constructed by digestion of c-Jun from pTL1 with BglII, ex- Downloaded from https://academic.oup.com/mend/article/13/2/276/2741651 by guest on 27 September 2021 tension with klenow, and cutting with BamHI. This fragment was inserted into pEG202. B42-cJun(237–331) was con- Yeast Two-Hybrid System structed by inserting the PCR fragment encoding c-Jun amino acids 237–331 into pJG4–5. hRAR␣ was expressed Yeast two-hybrid analysis and LexA-cJun activity were mea- from the plasmid pYE10 (61). sured by quantifying ␤-gal activity using o-nitrophenyl-␤-D- For mammalian cells, the reporter plasmids have the gene galactoside as a substrate as described (62). AT-RA was Ϫ for chloramphenicol acetyl transferase (CAT) driven by the added to a final concentration of 10 6 M. RA-inducible RARE-tk, AP-1-inducible TRE-tk, or GAL4-inducible 17M-tk promoters (34). Transfection efficiency was Acknowledgments standardized by measuring the ␤-galactosidase (␤-gal) activity, originating from the cotransfected plasmid pCH110 or CMV-LacZ (25). For yeast cells, the reporter was pSH18–34 We would like to thank Roger Brent, Barak Cohen, and Lau- (45), which has the LacZ gene under the control of a LexA- ren Ha for providing the materials for the yeast two-hybrid inducible promoter. system, Gordon Tomaselli for GAL-cFos and GAL-cJun, Pierre Chambon and Hinrich Gronemeyer for the hRAR␣ plasmid, Athanasios Bubulya for helpful discussions, Yun Cell Transfections and CAT Assays Zhou for technical support, and Scott Leisner for critical reading of the manuscript. HeLa and Cos cells were grown and transfected as described previously (34). hRAR␣ and GR were activated by the addi- Ϫ tion of 10 7 M of either AT-RA and Dex, respectively. CAT assays were performed and standardized according to the Received July 9, 1998. Re-revision received November 2, measured ␤-gal activity as previously described (25). For all 1998. Accepted November 4, 1998. transfections, we used different amounts of expression plas- Address requests for reprints to: Dr. L. Shemshedini, ␮ mid, 1 g of reporter plasmid (RARE-tk-CAT, 17 M-tk-CAT, or Department of Biology, University of Toledo, Toledo, ␮ ␮ TRE-tk-CAT), 2 g of pCH110 for Cos cells, and 0.5 g Ohio 43606. Email: [email protected]. CMV-LacZ for HeLa cells, and enough carrier DNA (Blue- This work was supported in part by American Heart Asso- script) to bring the final plasmid amount to 9 ␮g per dish. CAT ciation Grant NW-95–16-YI and NIH Grant DK-51274 to L.S. assay results were quantified by densitometric scanning of autoradiograms of at least three repeats for each transfection, and each value represents the average of three to four repetitions plus standard deviation. REFERENCES Gel Mobility Shift Assay 1. Mendelsohn C, Ruberte E, Chambon P 1992 Retinoid HeLa and Cos cells were transfected with 10 ␮g of hRAR␣ receptors in vertebrate limb development. Dev Biol 152: and 2 ␮g of pCH110. Cells receiving ligand were treated with 50–61 100 nM RA 24 h before harvesting. Cells were harvested in 2. Lotan R 1980 Effects of vitamins A and its analogs (reti- ice-cold PBS and spun at 5000 rpm for 5 min. Ten percent of noids) on normal and neoplastic cells. Biochim Biophys the cells were used to perform a ␤-gal assay for quantification Acta 605:33–91 of transfection efficiency. The remainder of the cells were 3. Thaller C, Eichele G 1987 Identification and spatial dis- resuspended in buffer I (10 mM Tris-HCl, pH 7.5; 10 mM NaCl; tribution of retinoids in the developing chick limb bud.

5mM MgCl2) and incubated at 4 C for 5 min. Sucrose (0.3 M) Nature 345:815–819 was then added and cells were lysed with a dounce homog- 4. Merriman R, Bertam J 1979 Reversible inhibition by reti- enizer. Nuclei were pelleted by centrifuging lysed cells at noids of 3 methylcholanthrene-induced neoplastic trans- 2500 rpm (600 ϫ g) for 10 min. The nuclear pellet was washed formation in C3H/10T1/2 clone 8 cells. Cancer Res once with buffer II (buffer I containing 0.3 M sucrose). Then, 39:1661–1666 the nuclear pellet was resuspended in buffer III (50 mM Tris- 5. Hong WK, Lippman SM, Itri LM, Karp DD, Lee JS, Byers HCl, pH 8; 150 mM NaCl; 5 mM EDTA; 0.1% Nonidet P-40) RM, Schantz SP, Kramer AM, Lotan R, Peters LJ, Dimery with protease inhibitors and incubated with shaking at4Cfor IW, Brown BW, Goepert H 1990 Prevention of second 30 min. The lysed nuclei were centrifuged at 15,000 rpm for primary tumors with isoretinoin in squamous-cell carci- 15 min, and the supernatant, constituting the nuclear extract, noma of the head and neck. N Engl J Med 323:795–801 was saved. The amount of extract used was standardized 6. Mangelsdorf DJ, Thummel C, Beato M, Herrlich P, according to ␤-gal activity. Schutz G, Umesono K, Blumberg B, Kastner P, Mark M, Gel mobility shift assays were performed with nuclear ex- Chambon P, Evans RM 1995 The nuclear receptor su- tracts containing the same amount of ␤-gal activity. These perfamily: the second decade. Cell 83:835–839 reactions were performed in a final volume of 20 ␮l in DNA- 7. Mangelsdorf DJ, Evans RM 1995 The RXR heterodimers binding buffer (10 mM Tris, pH 8; 0.1 mM EDTA; 4 mM dithio- and orphan receptors. Cell 83:841–850 threitol) which also contained 1 ␮g of poly(dI-dC), 100 mM 8. Beato M, Herrlich P, Schutz G 1995 Steroid hormone MOL ENDO · 1999 Vol 13 No. 2 284

receptors: many actors in search of a plot. Cell 83: pression of c-Jun or c-Fos but not Jun D inhibits oestro- 851–857 gen receptor activity in human breast cancer derived 9. Kastner P, Mark M, Chambon P 1995 Nonsteroid nuclear cells. EMBO J 10:2237–2245 receptors: what are genetic studies telling us about their 31. Tzuckerman M, Zhang X-K, Pfahl M 1991 Inhibition of role in real life? Cell 83: 859–869 estrogen receptor activity by the tumor promoter 12–0- 10. Thummel C 1995 From embryogenesis to metamorpho- tetradecanoylphorbol-13-acetate: a molecular analysis. sis: the regulation and function of Drosophila nuclear Mol Endocrinol 91:1983–1992 receptor superfamily members. Cell 83:571–877 32. Gaub MP, Bellard M, Scheuer I, Chambon P, Sassone- 11. Pfahl M 1993 Nuclear receptor/AP-1 interaction. Endocr Corsi P 1991 Activation of the ovalbumin gene by the Rev 14:651–658 estrogen receptor involves the Fos-Jun complex. Cell 12. Ponta H, Cato ACB, Herrlich P 1992 Interference of path- 63:1267–1276 way specific transcription factors. Biochim Biophys Acta 33. Diamond MI, Miner JN, Yoshinaga SK, Yamamoto KR 1129:255–261 1990 Transcription factor interactions: selectors of pos- Downloaded from https://academic.oup.com/mend/article/13/2/276/2741651 by guest on 27 September 2021 13. Schuele R, Evans R 1991 Cross-coupling of signal trans- itive or negative regulation from a single DNA element. duction pathways: zinc finger meets leucine zipper. Science 249:1266–1272 Trends Genet 7:377–381 34. Bubulya A, Wise SC, Shen X-Q, Burmeister LA, Shem- 14. Miner JN, Diamond MI, Yamamoto K 1991 Joints in the shedini L 1996 c-Jun can mediate androgen receptor- regulatory lattice: composite regulation by steroid recep- induced transactivation. J Biol Chem 271:24583–24589 tor-AP-1 complexes. Cell Growth Differ 2:525–530 35. Wise SC, Burmeister LA, Zhou X-F, Bubulya A, Oberfield 15. Karin M, Zheng-gang L, Ebrahim Z 1997 AP-1 function JL, Birrer MJ, Shemshedini L 1998 Identification of do- and regulation. Curr Opin Cell Biol 9:240–248 16. Angel P, Karin M 1991 The role of Jun, Fos and the AP-1 mains of c-Jun mediating androgen receptor transacti- complex in cell- proliferation and transformation. Bio- vation. Oncogene 16:2001–2009 chim Biophys Acta 1072:129–157 36. Kamei Y, Xu L, Heinzel T, Torchia J, Kurokawa R, Gloss 17. Vogt PK, Bos TJ 1990 jun: oncogene and transcription B, Lin SC, Heyman RA, Rose DW, Glass CK, Rosenfeld factor. Adv Cancer Res 55:1–35 MG 1996 A CBP integrator complex mediates transcrip- 18. Verma IM, Sassone-Corsi P 1987 Proto-oncogene fos: tional activation and AP-1 inhibition by nuclear receptors. complex but versatile regulation. Cell 51:513–514 Cell 85:403–414 19. Lamph WW, Wamsley P, Sassone-Corsi P, Verma IM 37. Chakravarti D, LaMorte VJ, Nelson MC, Nakajima T, 1988 Induction of proto- oncogene Jun/AP-1 by serum Schulman IG, Juguilon H, Montminy M, Evans RM 1996 and TPA. Nature 334:629–631 Role of CBP/P300 in nuclear receptor signalling Nature 20. Ryder K, Lan LF, Nathans D 1988 A gene activated by 383:99–103 growth factors is related to the oncogene v-jun. Proc Natl 38. Kawasaki H, Eckner R, Yao TP, Taira K, Chiu R, Living- Acad Sci USA 85:1487–1491 ston DM, Yokoyama KK 1998 Distinct roles of the co- 21. Jonat C, Rahmsdorf HJ, Park KK, Cato ACB, Ponta H, activators p300 and CBP in retinoic-acid-induced F9-cell Herrlich P 1990 Antitumor promotion and antiinflamma- differentiation. Nature 393:284–289 tion: down-modulation of AP-1 (Fos/Jun) activity by glu- 39. Bannister AJ, Oehler T, Wilhelm D, Angel P, Kouzarides cocorticoid hormone. Cell 62:1189–1204 T 1995 Stimulation of c-Jun activity by CBP: c-Jun res- 22. Lucibello F, Slater EP, Jooss K, Beato M, Muller R 1990 idues Ser63/73 are required for CBP induced stimulation Mutual transrepression of Fos and the glucocorticoid in vivo and CBP binding in vitro. Oncogene 11: receptor: involvement of a functional domain in Fos 2509–2514 which is absent in FosB. EMBO J 9:2827–2834 40. Brockmann D, Bury C, Kroner G, Kirch HC, Esche H 1995 23. Schuele, R, Rangarajan P, Kliewer S, Ransone LJ, Yang Repression of the c-Jun transactivation function by the N, Verma IM, Evans RM 1990 Functional antagonism adenovirus type 12 E1A 52R protein correlates with the between oncoprotein c-Jun and the glucocorticoid re- inhibition of phosphorylation of the c-Jun activation do- ceptor. Cell 62:1217–1226 main. J Biol Chem 270:10754–10763 24. Yang-Yen H-F, Chambard J-C, Sun Y-L, Smeal T, 41. Caelles C, Gonzalez-Sancho JM, Munoz A 1997 Nuclear Schmidt TJ, Drouin J, Karin M 1990 Transcriptional in- hormone receptor antagonism with AP-1 by inhibition of terference between c-Jun and the glucocorticoid recep- the JNK pathway. Genes Dev 11:3351–3364 tor: mutual inhibition of DNA binding due to direct pro- 42. Hibi M, Lin A, Smeal T, Minden A, Karin M 1993 Identi- tein-protein interaction. Cell 62:1205–1215 fication of an oncoprotein- and UV-responsive protein 25. Shemshedini L, Knauthe R, Sassone-Corsi P, Pornon A, kinase that binds and potentiates the c-Jun activation Gronemeyer H 1992 Cell-specific inhibitory and stimula- domain. Genes Dev 7:2135–2148 tory effects of Fos and Jun on transcription activation by 43. Derijard Y, Hibi M, Wu IH, Barret T, Su B, Deng T, Karin nuclear receptors. EMBO J 10:3839–3849 M, Davis RJ 1994 JNK1: A protein kinase stimulated by 26. Nicholson RC, Mader S, Nagpal S, Leid M, Rochette- UV light and Ha-Ras that binds and phosphorylates the Egly C, Chambon P 1990 Negative regulation of the rat stromelysin gene promoter by retinoic acid is mediated c-Jun activation domain. Cell 76:1025–1037 by an AP-1 binding site. EMBO J 9:4443–4454 44. Konig H, Ponta H, Rahmsdorf HJ, Herrlich P 1992 Inter- 27. Yang-Yen H-F, Zhang X-K, Graupner G, Tzuckerman M, ference between pathway-specific transcription factors: Sakamoto B, Karin M, Pfahl M 1991 Antagonism glucocorticoids antagonize phorbol ester-induced AP-1 between retinoic acid receptors and AP-1: implications activity without altering AP-1 site occupation in vivo. for tumor promotion and inflammation. New Biol 3: EMBO J 11:2241–2246 1206–1219 45. Gyuris J, Golemis EA, Chertkov H, Brent R 1993 Cdi1, a 28. Zhang X-K, Wills KN, Husmann M, Hermann T, Pfahl M human G1 and S phase protein phosphatase that asso- 1991 Novel pathway for thyroid hormone receptor action ciates with Cdk2. Cell 75:791–803 through interaction with jun and fos oncogene activities. 46. Gentz R, Rauscher FJ 3d, Abate C, Curran T 1989 Par- Mol Cell Biol 11:6016–6025 allel association of Fos and Jun leucine zippers juxta- 29. Schuele R, Rangarajan P, Yang N, Kliewer S, Ransone poses DNA binding domains. Science 243:1695–1699 LJ, Bolado J, Verma IM, Evans RM 1991 Retinoic acid is 47. Halazonetis TD, Georgopoulos K, Greenberg ME, Leder a negative regulator of AP-1-responsive genes. Proc Natl P 1988 c-Jun dimerizes with itself and with c-Fos, form- Acad Sci USA 88:6092–6096 ing complexes of different DNA binding affinities. Cell 30. Doucas V, Spyrou LG, Yaniv M 1991 Unregulated ex- 55:917–924 RAR Blocks c-Jun/c-Fos Dimerization 285

48. Touray M, Ryan F, Jaggi R, Martin F 1991 Characteriza- tional role of Jun oncogene family members on glucocor- tion of functional inhibition of the glucocorticoid receptor ticoid receptor-dependent transcription. Mol Endocrinol by Fos/Jun. Oncogene 6:1227–1234 7:570–584 49. Berkenstam A, del Mar Vivanco Ruiz M, Barettino D, 57. Fanjul A, Dawson MI, Hobbs PD, Jong L, Cameron JF, Horikoshi M, Stunnenberg HG 1992 Cooperativity in Harlev E, Graupner G, Lu XP, Pfahl M 1994 A new class transactivation between retinoic acid receptor and TFIID of retinoids with selective inhibition of AP-1 inhibits pro- requires an activity analogous to E1A. Cell 69:401–412 liferation. Nature 372:107–110 50. Scholer HR, Ciesolka T, Gruss P 1991 A nexus between 58. Chen JY, Penco S, Ostrowski J, Balaguer P, Pons M, oct-4 and E1A: implications for gene regulation in em- Starrett JE, Reczek P, Chambon P, Gronemeyer H bryonic stem cells. Cell 66:291–304 1994 RAR-specific angonist/antagonists which disso- 51. Aarnisalo P, Palvimo JJ, Janne OA 1998 CREB-binding ciate transactivation and AP-1 transrepression inhibit protein in androgen receptor-mediated signaling. Proc anchorage-independent cell proliferation. EMBO J Natl Acad Sci USA 95:2122–2127 14:1187–1197 Downloaded from https://academic.oup.com/mend/article/13/2/276/2741651 by guest on 27 September 2021 52. Monden T, Wondisford FE, Hollenberg AN 1997 Isolation 59. Leid M, Kastner P, Lyons R, Nakshatri H, Saunders M, and characterization of a novel ligand-dependent thyroid Zacharewski T, Chen JY, Staub A, Garnier JM, Mader S, hormone receptor-coactivating protein. J Biol Chem Chambon P 1992 Purification, cloning, and RXR identity 272:29834–29841 53. Smith CL, Onate SA, Tsai MJ, O’Malley BW 1996 CREB of the HeLa cell factor with which RAR or TR het- binding protein acts synergistically with steroid receptor erodimerizes to bind target sequences efficiently. Cell coactivator-1 to enhance steroid receptor-dependent 68:377–395 transcription. Proc Natl Acad Sci USA 93:8884–8888 60. Finkel T, Duc J, Fearon ER, Dang CV, Tomaselli GF 1993 54. Hanstein B, Eckner R, DiRenzo J, Halachmi S, Liu H, Detection and modulation in vivo of helix-loop-helix pro- Searcy B, Kurokawa R, Brown M 1996 p300 is a com- tein-protein interactions. J Biol Chem 268:5–8 ponent of an estrogen receptor coactivator complex. 61. Heery DM, Zacharewski T, Pierrat B, Gronemeyer H, Proc Natl Acad Sci USA 93:11540–11545 Chambon P, Losson R 1993 Efficient transactivation by 55. Jenster G, Spencer TE, Burcin MM, Tsai SY, Tsai MJ, retinoic acid receptors in yeast requires retinoid X recep- O’Malley BW 1997 Steroid receptor induction of gene tors. Proc Natl Acad Sci USA 90:4281–4285 transcription: a two-step model. Proc Natl Acad Sci USA 62. Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman 94:7879–7884 JG, Smith JA, Struhl K 1995 Current Protocols in Molec- 56. Maroder M, Farina AR, Vacca A, Felli MP, Meco D, ular Biology. John Wiley & Sons, Inc., vol 2:20.1.1– Screpanti I, Frati L, A Gulino 1993 Cell-specific bifunc- 20.1.28

8th Meeting of the Latin American Thyroid Society Iguaçu Falls PR, Brazil, May 27–30, 1999 For information, please contact Dr. Antonio C. Bianco, fax: 55-11-8187285 or visit our web site at http://www.lats.org.