AP-1 Transrepressing Retinoic Acid Does Not Deplete Coactivators Or AP-1 Monomers but May Target Speci®C Jun Or Fos Containing Dimers

Home , Retinoic acid receptor, Transactivation

AP-1 transrepressing retinoic acid does not deplete coactivators or AP-1 monomers but may target speci®c Jun or Fos containing dimers

Kazumi Suzukawa1 and Nancy H Colburn*,1

1Gene Regulation Section, Basic Research Laboratory, National Cancer Institute at Frederick, Frederick, Maryland, MD 21702- 1201, USA

Retinoic acid (RA) inhibits tumor promotion in many Introduction models in vivo and in vitro, among them mouse epidermal JB6 cells. RA treatment suppresses 12-O-tetradecanoyl- Carcinogenesis is a multistep process consisting of phorbol-13-acetate (TPA) induced AP-1 activity, an initiation, promotion and progression (Fearon and activity that is required for transformation of JB6 P+ Vogelstein, 1990; Klein, 1981). Regulators of responsible cells. The molecular mechanism of AP-1 transrepression signaling pathways for each stage are potential targets by retinoids is unclear, especially as related to inhibition for cancer prevention (Boutwell, 1989; Colburn et al., of transformation. Overexpression of AP-1 components 1979, 1982). A causal relationship between tumor did not rescue TPA induced AP-1 activation nor did a promoter-induced neoplastic transformation and activa- GST pull down experiment implicate direct binding, thus tion of transcription factor AP-1 has been demonstrated rendering unlikely both a Jun/Fos-RA-RAR direct in mouse and human cell progression models (Bernstein interaction and a Jun/Fos sequestration mechanism. and Colburn, 1989; Dong et al., 1994; Li et al., 1998) and Overexpression of p300, SRC-1 or pCAF did not in a dominant negative c-jun transgenic mouse (Young et abrogate AP-1 suppression by RA, thus arguing against al., 1999). In the JB6 mouse epidermal cell clonal coactivator competition. Overexpression of the corepres- variants, AP-1 transactivation is induced by 12-O- sor silencing mediator for retinoic acid and thyroid tetradecanoylphorbol-13-acetate (TPA) in transforma- hormone receptors (SMRT) suppressed AP-1 activity. tion sensitive but not in resistant cells (Bernstein and However, SMRT but not RA inhibited cJun transactiva- Colburn, 1989). TPA induced cell transformation is tion, suggesting SMRT does not mediate RA transre- blocked by AP-1 inhibitors such as retinoids, glucocorti- pression. RA treatment also did not block TPA induced coids, and the dominant negative cJun (TAM67) (Dong ERK phosphorylation, Jun/Fos family protein expression et al., 1994; Li et al., 1996, 1997). Retinoic acid (RA) is except for cFos, or DNA binding of the AP-1 complex. eective in inhibiting papilloma formation in mouse skin The transcriptional activities of full-length JunB and and transformation of mouse epidermal JB6 cells full-length Fra-1, but not the transactivation domain (Colburn et al., 1981; Li et al., 1996; Verma and fusions, were increased by TPA treatment and sup- Boutwell, 1977). RA induces dierentiation of keratino- pressed by RA. Since these full-length fusions have bzip cytes and suppresses tumor phenotype in several models domains, the results suggest that JunB and/or Fra-1- (Bollag and Holdener, 1992; Chen et al., 1995; Fuchs and containing dimers may constitute one target of RA for Green, 1980). RA can also cause a reversible phenotypic transrepression of AP-1. suppression of the TPA-transformed JB6 cell line RT101 Oncogene (2002) 21, 2181 ± 2190. DOI: 10.1038/sj/ (Dong et al., 1995). onc/1205281 The biological eects of RA are mediated by its receptors (RARs) in the nucleus. The RARs belong to Keywords: AP-1; JB6 cells; retinoic acid; transrepres- a large superfamily of ligand-inducible transcription sion; GAL4 transactivation assay factors that include vitamin D and thyroid hormone receptors. The retinoid receptors consist of two subunits, RARs (a, b, g) and RXRs (a, b, g) (Chambon, 1996). It has been well documented that liganded nuclear receptors including RAR inhibit AP-1 activation (Fisher and Voorhees, 1996; Li et al., 1996, 1997; Pfahl, 1993; Resche-Rigon and Gronemeyer, 1998; Saatcioglu et al., 1994; Schule and Evans, 1991). Transactivation of target genes through the *Correspondence: NH Colburn, Gene Regulation Section, Basic RARE sequence is functionally dissociated from Research Laboratory Building 560 Room 21 ± 89, National Cancer transrepression of AP-1 (Bollag and Holdener, 1992; Institute at Frederick, PO Box B, Frederick, Maryland, MD 21702- Chen et al., 1995; Fanjul et al., 1994; Fuchs and Green, 1201, USA; E-mail: [email protected] Received 22 August 2001; revised 14 December 2001; accepted 19 1980; Li et al., 1996). Inhibition of tumor promotion December 2001 by RA appears to be mediated by blocking AP-1 AP-1 transrepression by retinoic acid in JB6 cells K Suzukawa and NH Colburn 2182 activation, not by activating retinoic acid response element (RARE)- dependent transcription (Li et al., 1996). Mouse JB6 cells express only RAR g and a, not RAR b, and the tumor promotion inhibiting activity has been attributed to the RAR g (Rudd et al., 1993). RAR g is functionally predominant for transrepression in keratinocytes (Goyette et al., 2000). The precise mechanism of transrepression of AP-1 by RA is not clear. Three mechanisms have been suggested. The ®rst is direct interaction between Jun or Fos proteins and liganded RAR (Pfahl, 1993). The second is competition between liganded RAR and Jun or Fos proteins to bind transcriptional coactivators (Kamei et al., 1996; Lee et al., 1998). The third is inhibition of cJun N- terminal kinase (JNK) (Caelles et al., 1997; Lee et al., 1999). The silencing mediator for retinoic acid and thyroid hormone receptors (SMRT) was identi®ed as a corepressor which interacts with unliganded RAR. The interaction between SMRT and RAR is destabi- lized by ligand (Chen and Evans, 1995). SMRT mediates the repressive eect of unliganded nuclear Figure 1 Transrepression of TPA induced AP-1 activity by RA. RA represses TPA induced AP-1 activity in a dose dependent receptors through the recruitment of histone deacety- manner. JB6 P(+) cells were transfected with 0.4 mg of AP-1- lase complexes (Nagy et al., 1997). SMRT also luciferase reporter construct in a 24 well plate. The next day, the suppresses other transcription factors (e.g. AP-1 SRF, cells were exposed to DMSO or TPA (10 ng/ml)+RA (in various NFkB, and Oct-1) (Kakizawa et al., 2000; Lee et al., concentrations as indicated) in 0.2% FBS+EMEM for 18 h. All 2000). Therefore, there is the possibility that RA experiments were performed in triplicate, and standard error of the mean is indicated. The data are shown as mean+s.d. Three treatment recruits SMRT from RAR to AP-1 com- independent experiments showed similar results plexes. The results presented here render unlikely the previously suggested mechanisms as well as recruitment of corepressor SMRT. Instead, RA appears to target RA (Figure 2). When cJun, Fra-1, Fra-2, or cFos was JunB and/or Fra-1-containing dimers. expressed, the AP-1 activation level with TPA+RA was similar to the solvent control level. Furthermore, a GST pull down assay using GST-retinoic acid receptor g (the functionally predominant RAR in JB6 cells) did Results not detect ligand dependent interaction with AP-1 components of a JB6 nuclear extract, while RXR a was RA transrepresses TPA induced AP-1 activation pulled down ligand independently as expected (data The concentrations of RA that inhibit TPA induced not shown). These results suggest that direct interac- transformation of JB6 cells (Dong et al., 1994), also tion of liganded RAR with Jun or Fos proteins is inhibit TPA induced AP-1 activation (Figure 1). Dose unlikely to contribute to AP-1 transrepression. dependent suppression by RA was observed at concentrations lower than 1076 M, with higher con- Concentration of coactivator is not limiting for centrations showing no greater suppression. transrepression It has been suggested that the amount of available No rescue of AP-1 activity from transrepression by jun or p300/CBP or SRC-1 can be limiting when AP-1 and fos overexpression RAR compete for binding to the coactivator (Kamei et If RAR inhibits AP-1 by direct interaction with Jun or al., 1996; Lee et al., 1998). We asked whether any of Fos proteins to sequester them, overexpression of the these coactivators is limiting for AP-1 activity in RA sequestered AP-1 component should cancel the inhibi- treated cells. Overexpression of coactivator p300 or tion. A luciferase assay of AP-1 activation in cells SRC-1 increased both basal and TPA induced AP-1 cotransfected with Jun or Fos expression constructs activity. However, neither of them reversed transre- was performed. Overexpression of JunD or JunB raised pression by RA (Figure 3), nor did expression of ASC- the basal level and prevented further induction of AP-1 2 (not shown) reverse transrepression. by TPA. Expression of cFos substantially increased Overexpression of pCAF (Figure 3) inhibited AP-1 basal AP-1 activity and TPA treatment produced a activation, suggesting pCAF is not involved in the further increase. Overexpression of cJun, Fra-1, Fra-2 coactivator complex of AP-1 in this context. To or cFos enhanced both basal and TPA induced levels determine whether transrepression between RAR and of AP-1 but failed to abrogate AP-1 transrepression by AP-1 is mutual, endogenous RAR activity was

Oncogene AP-1 transrepression by retinoic acid in JB6 cells K Suzukawa and NH Colburn 2183

Figure 2 Overexpression of Jun or Fos proteins did not rescue AP-1 activity. Each AP-1 component or pcDNA3.1 as a control DNA was cotransfected with AP-1 reporter construct. JB6 P (+) cells were exposed to 0.1% DMSO or TPA+RA for 18 h. Triplicate samples were run for each condition. All experiments were performed in triplicate, and standard error of the mean is indicated. Three independent experiments showed similar results. Note the dierent scales for the right and left groups

AP-1 and RAR is not mutual in JB6 cells (Figure 4). Taken together the results suggest that coactivators p300, SRC-1 and pCAF are not limiting for AP-1 activation when it is transrepressed by RA in JB6 cells.

SMRT suppresses AP-1 activity differently from RA Liganding of RAR decreases its anity to corepressors (Chen and Evans, 1995). Released corepressors may be recruited to the AP-1 complex resulting in suppression of transcriptional activity. Overexpression of corepressor SMRT suppressed both basal and TPA induced AP-1 activity (Figure 5a). We measured transcriptional activation of full-length cJun using a Gal4 transactivation assay. TPA treatment activated the Gal4 DNA binding domain (dbd)-cJun fusion construct. At a concentration that substantially inhibits AP-1 luciferase activation, RA treatment did not signi®cantly aect the TPA induced activation of GAL4dbd-cJun whereas SMRT expression suppressed cJun activation almost Figure 3 Overexpression of coactivator did not rescue AP-1 completely (Figure 5b). Similar results were seen with activity. An expression construct of p300, SRC-1, or p/CAF or Gal4dbd-JunB or Gal4dbd-JunD. Thus SMRT appears control DNA was transfected into JB6 cells. Then cells were to interact with Jun proteins or with the kinases that treated with DMSO or TPA+RA. Triplicate samples were analysed for each condition. All experiments were performed in activate them. SMRT-mediated suppression of cJun, triplicate, and standard error of the mean is indicated. Three JunB or JunD activation can explain transrepression of independent experiments showed similar results AP-1 by SMRT, but it cannot explain transrepression of AP-1 by RA. Histone deacetylase (HDAC) is a calatytic subunit of a corepressor complex (Ayer, measured as RARE driven luciferase activity. JB6 cells 1999). A speci®c inhibitor of HDAC, trichostatin A showed about 40-fold RARE activation by 1077 M (TSA), increased AP-1 activity in a dose dependent RA. RA treatment transrepressed AP-1 activity while manner. HDAC thus appears to regulate AP-1 activity. TPA treatment enhanced RARE activation, consistent However, TSA treatment did not rescue AP-1 activity with a report published elsewhere (Fukasawa et al., from transrepression (Figure 5C). The RA+TPA value 2000). In contrast to other reports (Kamei et al., 1996; is about 40% of the TPA-only value at all TSA Yang-Yen et al., 1991), the transrepression between concentrations and without TSA. These results suggest

Oncogene AP-1 transrepression by retinoic acid in JB6 cells K Suzukawa and NH Colburn 2184

Figure 4 Transrepression is not mutual between AP-1 and RARE activities. JB6 Cl41 cells were transfected with either AP-1- Luciferase or RARE-Luciferase reporter construct. Cells were treated with DMSO or TPA(10 ng/ml)+RA(1077 M). Three independent experiments showed similar results. Standard error of the mean is indicated as a bar

that recruitment of corepressors SMRT or HDAC is protein expression does not appear to explain RA not likely to be involved in transrepression of AP-1 by transrepression of AP-1, nor does RA inhibit the RA. formation of AP-1 complexes on DNA.

RA does not inhibit TPA induced ERK activation RA targets the transcriptional activation of full-length JunB and Fral Unidirectional transrepression between RAR and AP-1 (Figure 4) suggests that RA treatment blocks the TPA Activation of AP-1 components is the last step to AP-1 signaling pathway leading to AP-1 activation. Activa- regulated transcription and might be a target of tion of MAP kinase ERKs 1 and 2 is required for TPA transrepression. We measured transcriptional activa- induced AP-1 activation and transformation of JB6 tion of the transactivation domain located at the N- cells (Huang et al., 1998; Watts et al., 1998). TPA terminus of cJun, JunB, or JunD using a Gal4 treatment rapidly increased the amount of phospho- transactivation assay (TA). None of the JunTA ERK1/2, an activated form of ERKs. Peak activation proteins was both activated by TPA and suppressed was observed at 30 min followed by a gradual by RA. We then measured the transcriptional activity decrease. Simultaneous RA treatment did not inhibit of full-length Jun genes. All of the Jun family proteins TPA induced ERK activation nor did it accelerate showed increased transcriptional activity with TPA subsequent inactivation (Figure 6). The data suggest treatment. RA speci®cally suppressed full-length JunB that RA does not block the TPA signaling pathway at transactivation (Figure 8a) albeit it a partial transre- the stage of ERK activation or events upstream of pression. To determine the contribution of Fos family ERK required for the activation of ERK. genes to transrepression, we measured the transcriptional activation of all but Fos B, a protein that is absent from JB6 cells and from mouse epidermis RA does not inhibit TPA induced synthesis of most Jun or (Angel et al., 2001; Bernstein and Walker, 1999). At Fos proteins or inhibit DNA binding of the AP-1 complex the serum concentration needed to observe AP-1 TPA treatment induces c-Jun, JunB, c-Fos, and Fra-1 transrepression by RA, neither full-length cFos nor expression within 4 hours (Bernstein et al., 1992; the C-terminal transactivation domain (CTA) was Bernstein and Walker, 1999). However, simultaneous activated by TPA treatment. Both full-length and C- treatment with RA did not inhibit TPA induced Jun terminal Fra-1 were activated by TPA. Full-length but family, Fra-1, or Fra-2 expression (Figure 7a). Some not C-terminal Fra-2 was activated by TPA. Among decrease in cFos expression was seen at 3 and 6 h them, only full-length Fra-1 activation was signi®cantly following TPA+RA treatment. However, the lack of suppressed by RA treatment. Contransfection of either AP-1 rescue (Figure 2) by cFos expression, suggests a full-length JunB or a full-length Fra-1 Gal4 fusion that the RA reduced level of cFos is not limiting for with a cDNA expression construct of a heterodimer AP-1 activation. RA treatment did not block TPA partner (i.e. cFos, Fra-1, or Fra-2, for JunB) showed induced DNA binding of the AP-1 complex (Figure similar results to those shown in Figure 8 (data not 7b). Thus, RA inhibition of TPA induced Jun/Fos shown). That is, expression of cFos, Fra-1, or Fra-2

Oncogene AP-1 transrepression by retinoic acid in JB6 cells K Suzukawa and NH Colburn 2185

Figure 5 Recruitment of corepressors SMRT or HDAC is not involved in RA induced transrepression. (a) SMRT mimics RA eect on AP-1. 0.2 mg of AP-1 reporter construct was transfected with 0.2 mg of pcDNA control vector or SMRT expression construct. Transfected cells were treated with DMSO or TPA+RA for 18 h. (b) SMRT does not mimic RA but suppresses transactivation of cjun. Twenty-®ve ng of pFR-luc, 50 ng of pMcjun, 0.2 mg of pcDNA control vector, and 0.2 mg of pcDNA or SMRT was transfected. Cells were treated with DMSO or TPA+RA for 6 h. (c) Treatment with HDAC inhibitor TSA does not rescue AP-1 activity. 0.4 mg of AP-1 reporter construct was transfected. Indicated concentrations of TSA were added in addition to TPA+RA. Cells were cultured for 18 h. Shown are the means for three independent experiments+standard error with cotransfected Gal4-JunB produced no statistically a target of RA treatment. A role for JunB/cFos dimers signi®cant enhancement of retinoid transrepression. cannot be excluded. Also consistent with these results The Gal4-JunB activity with RA+TPA was about is the possibility of a JunB or Fra-1 binding partner 74% of the TPA induced level with or without that is not an AP-1 protein. expression of cFos, Fra-1, or Fra-2. This suggests that the concentration of endogenous Fos family proteins is not limiting for forming the RA-sensitive JunB- Discussion containing complex. Since full-length JunB and Fra-1 have dimerization domains, these results are consistent Using the mouse JB6 model of AP-1 transactivation by with the possibility that the JunB/Fra 1 heterodimer is TPA and transrepression by RA, the above observa-

Oncogene AP-1 transrepression by retinoic acid in JB6 cells K Suzukawa and NH Colburn 2186 tions render unlikely three suggested mechanisms for However, the details of the mechanism of AP-1 AP-1 transrepression as follows, (i) direct interaction transrepression by RA have been uncertain. A cross- between a Jun or a Fos protein and RAR. (ii) linking and coprecipitation approach suggested that Competition between AP-1 and RAR to bind RAR can physically associate with both c-Jun and c- coactivators p300, SRC-1, or pCAF (iii) recruitment Fos (Yang-Yen et al., 1991). However, these protein- of corepressor SMRT or HDAC. An AP-1 complex protein interactions appear to be weak or indirect, consisting of JunB and Fra-1 or another JunB- or Fra- since our data and other reports suggest that direct 1-containing complex appears to be one target of interaction between AP-1 and RAR is unlikely transrepression by RA. (Nicholson et al., 1990). A recent study using RAR null keratinocytes p300 and SRC-1 can act as coactivators for RAR or revealed that RAR is essential for the transrepression AP-1 mediated transcription (Kamei et al., 1996; Lee et of AP-1 activity by RA (Goyette et al., 2000). RARg al., 1998). pCAF has been shown to be a coactivator of plays a major role in AP-1 transrepression by RA in RAR (Blanco et al., 1998). In this study, we showed JB6 cells (Rudd et al., 1993). Mutational analysis of that pCAF is not involved as a coactivator in AP-1 RAR de®ned the DNA binding domain and the C- mediated transcription. Competition between AP-1 and terminal transactivation domain (AF2) as crucial for RAR to bind p300 or SRC-1 appears not to explain transrepression of AP-1 (DiSepio et al., 1999). transrepression in JB6 cells. Unidirectional transrepression also argues against the simple competition mechanism. RA might repress AP-1 activity through inhibition of Jun N-terminal kinase (JNK) (Caelles et al., 1997; Lee et al., 1998, 1999). Transformation by TPA does not, however, require JNK in JB6 cells. Expression of dominant negative Jun kinase did not inhibit transformation by TPA while it did inhibit transformation by tumor necrosis factor alpha (Huang et al., 1999). Since TPA induced transformation requires AP-1 but not JNK, it appears likely that JNK is not required for Figure 6 RA did not block ERK activation by TPA. One6105 AP-1 activation. TPA treatment produced no detect- JB6 P(+) cells were seeded in six well plates. The next day, cells able activation of JNK whereas TNFa did activate were treated with DMSO or TPA (10 ng/ml)+RA (1077 M) for indicated time (hour). The same blot was used to detect total JNK (data not shown). The lack of signi®cant amounts of ERK or phospho-ERK, sequentially activation of the transactivation domain of Jun family

Figure 7 RA inhibited TPA induced cFos expression but not expression of other Jun or Fos proteins or DNA binding activity of AP-1. (a) Western blot analysis of AP-1 components with nuclear protein at dierent times. Cells were treated with TPA (10 ng/ ml)+RA (1077 M) for indicated time (hour). Equal amount of nuclear protein was loaded. (b) EMSA with nuclear extracts (2 mg) from JB6 P(+) cells. The cells were exposed to DMSO or TPA (10 ng/ml)+RA (1077 M) in 0.2% FBS+EMEM for 18 h. 3, 6, 9 and 12 h treatment of TPA+RA did not decrease DNA binding relative to TPA only treatment (data not shown)

Oncogene AP-1 transrepression by retinoic acid in JB6 cells K Suzukawa and NH Colburn 2187

Figure 8 RA targets transactivation of full-length JunB and Fra1. (a) Gal4 transactivation assay of jun family genes. Cells were treated with DMSO or TPA+RA for 6 h after transfection. cJun, JunB, and JunD; Gal4dbd fused to full-length cJun, JunB and JunD, respectively. cJunTA, JunBTA, JunDTA; Gal4dbd fused to transactivation domain of cJun (amino acids (aa) 1 to 223), JunB (aa 1 to 203) and JunD (aa 1 to 151), respectively. The asterisk in 8A indicates statistical signi®cance at P50.01, for RA transrepression of FL JunB. (b) Gal4 transactivation assay of fos family genes. Cells were treated with DMSO or TPA+RA for 12 h after transfection. cFos, Fra1, and Fra2; Gal4dbd fused with full-length cFos, Fra1, and Fra2 gene, respectively. cFosCTA, Fra1C, and Fra2C; Gal4dbd fused with C-terminal transactivation domain of cFos (aa 208 ± 313), C-terminal part of Fra1 (aa 128 to 271), and C-terminal part of Fra2 (146 to 327), respectively. The asterisk in 8B indicates statistical signi®cance at P50.02 for transrepression of FL Fra-1. Each experiment had triplicate samples. Representative data of three independent experiments is shown. Bar indicates standard error for fold induction. The value for untreated sample was set at 1.0, thus showing no standard error genes by TPA (Figure 8a) suggests that TPA does not ®rst is that RA inhibits a coactivator (or activates a activate JNK in JB6 cells. Another MAP kinase, ERK, corepressor) that regulates transactivation of a speci®c is an essential molecule in TPA induced AP-1 AP-1 dimer. The second is that RA inhibits the activation (Huang et al., 1998; Watts et al., 1998). dimerization of Jun and Fos, a possibility that is Fra-1 activation is one of the subsequent steps suggested by Zhou et al., 1999). The third possibility is following ERK activation leading to activation of that the full-length JunB or full-length Fra-1 binds to a AP-1 dependent transcription (Young et al., 2001). Our non-AP-1 transcription factor protein through the bzip data (Figure 8b) suggest these steps are not aected by domain. Precedent for this is found in the binding of RA treatment. Therefore, RA might target the the NFkB p65 rel homology domain to the bzip interaction between the AP-1 complex and the basal domain of some but not all Jun or Fos proteins (Stein transcriptional machinery. et al., 1993). The inhibition of dimerization is unlikely DNA binding studies using in vitro translated cJun in the present case because RA treatment did not block suggested that RARs are able to antagonize AP-1 the DNA binding of AP-1, a process that requires activity by interfering with DNA binding ability dimerization. In conclusion, RA appears to target (Schule et al., 1991; Yang-Yen et al., 1991). Our data JunB- and/or Fra-1-containing dimers in which the using a nuclear extract (Figure 3b) suggest that RA binding partner binds to the bzip domain. The dimer treatment does not block the DNA binding activity of may be an AP-1 heterodimer or it may be JunB or Fra- AP-1, a ®nding that is consistent with a report using 1 bound to a non-AP-1 transcription factor, and this glomerular mesangial cells (Simonson, 1994). Except may contribute at least in part to the transrepression of for cFos, TPA induced expression of Jun and Fos AP-1 by RA in JB6 cells. proteins was not inhibited by RA treatment. The fact that cFos expression failed to reverse RA transrepression suggests that the RA-reduced level of cFos is not Materials and methods limiting for AP-1 transactivation. When fused to a heterologous DNA binding domain Reagents and cell culture like Gal4, Jun or Fos proteins can activate transcription with their intrinsic activation domain (Sutherland All trans retinoic acid was purchased from Sigma (St. Louis, et al., 1992). The leucine zipper motif has no MO, USA). TPA was from LKB Laboratories, Inc. (St. Paul, MN, USA). DMSO was from Pierce (Rockford, IL, USA). transactivation activity by itself but a fusion construct Antibodies against c-Jun (H-79), JunD (329), JunB (N-17), of Gal4dbd with a leucine zipper might recruit a dimer cFos (H-125), Fra-1 (N-17), Fra-2 (L-15) and FosB (102) were partner molecule resulting in activation of a reporter purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, construct (Smith and Bohmann, 1992). Speci®c inhibi- CA, USA). Trichostatin A (TSA) was from Wako chemicals tion of full-length JunB and full-length Fra1 transacti- USA (Richmond, VA, USA). Transformation sensitive JB6 vation by RA suggests three possible mechanisms. The mouse epidermal cell line (Cl 41) were grown at 368C in Eagles

Oncogene AP-1 transrepression by retinoic acid in JB6 cells K Suzukawa and NH Colburn 2188 Minimum Essential Media (EMEM) (BioWhitaker, Frederick, Nuclear protein extraction MD, USA) supplemented with 4% Fetal Bovine Serum (FBS), 2mML-Glutamine and 25 mg/ml Gentamicin (Life Technol- JP6 P(+) cells were grown in 15 cm diameter dishes. Cells ogy/Gibco, Gaithersburg, MD, USA). were treated with DMSO or TPA (10 ng/ml)+RA (1077) for 18 h. Cells were washed with PBS(7) once, then harvested with scraper and centrifuged. The collected cells were lysed Plasmids with Lysis Buer containing 25 mM HEPES at pH 7.7, AP-1 Luciferase reporter plasmids consisting of Luciferase 50 mM KCl, 2 mM PMSF, 2 mg/ml Leupeptin, 4 mg/ml reporter gene driven by the promoter harboring the Aprotinin, 100 mM DTT and 0.5% NP-40. The resulting appropriate element were constructed as previously described nuclei were washed with the above buer minus NP-40 (Dong et al., 1995, 1996). The AP-1 reporter plasmid (Washing Buer) and subsequently lysed with Extraction consists of Fire¯y Luciferase genes driven by an AP-1 Buer containing 25 mM HEPES at pH 7.7, 500 mM KCl, responsive promoter containing four copies of ¯anked AP-1 1mM PMSF, 1 mg/ml Leupeptin, 2 mg/ml Aprotinin, 100 mM consensus sequence (TCGACTATGATGAGTCATGGGGC) DTT and 10% Glycerol (or 8% Ficoll). All of the above from GCN4 and a minimal Albumin promoter region with procedures were performed at 48C and aliquots of nuclear TATA box (AAGCTTAGAATCTAGTATATTAGAGC- extracts were stored at 7708C. Protein concentration was GAGTCTTTCTGCACACAGATCACCTTTCCTATCAAC- determined by Coomassie Blue Protein Assay Reagent from CCCACTACCATACCCTTCCTCCATCTATACCACCCT- Pierce (Rockford, IL, USA). The supernatants were stored at ACTCTGC AGGTCGAC). Expression constructs were 7708C as nuclear extracts. generated by inserting full-length cDNA of c-jun (human), junB, junD, c-fos, fra-1 or fra-2, and fosB (mouse) into the EMSA (electrophoretic mobility shift assay) cloning site of eukaryotic expression vector pcDNA3 (Invitrogen, Carlsbad, CA, USA). Full length p300 expres- Oligonucleotides containing AP-1 consensus sequence were sion vector controlled by the CMV promoter was kindly purchased from Santa Cruz or synthesized by Gibco followed provided by Dr Lisa Felzien (Felzien et al., 1998). SalI±NotI by annealing. Fifty nanograms of double stranded oligos fragment of full-length p300 cDNA was cloned into XhoI± were end-labeled using 32P-g-ATP and T4 polynucleotide NotI site of pcDNA3.1(7) because original construct did not kinase (Roche Molecular Biochemicals, Indianapolis, IN, give p300 expression in JB6 cells. SRC-1 and p/CAF USA). The reaction mixture for EMSA consists of expression constructs were kindly provided by Dr Mitchell 50 000 c.p.m. of labeled and puri®ed oligonucleotides, 1 mg/ Lazar and Dr Yoshihiro Nakatani, respectively. RARg ml of poly (dI-dC) (Roche), and 1 ± 2 mg of nuclear extracts expression vector was provided by Dr Pierre Chambon. A in binding buer containing 20 mM of Tris-HCl at pH 7.5, full- length cDNA of the RARg gene was ampli®ed by PCR, 50 mM KCl, 1 mM DTT, 2 mM EDTA and 4% Glycerol. The digested with EcoRI/BamHI, and cloned into the pGEX-3X reaction was carried out at room temperature for 30 min. vector (Amersham Pharmacia Biotech, Inc., Piscataway, NJ, The protein-DNA complexes were resolved on a 6% TBE/ USA) to make a GST-RARg fusion expression construct. Gel Retardation gel (Invitrogen). The gel was dried and pFR-luc, pFA-cfos for Gal4 transactivation assay were exposed to Kodak ®lm at 7708C overnight. purchased from Stratagene (La Jolla, CA, USA). Expression constructs of Gal4 DNA binding domain (Gal4dbd) fused to Preparation of whole cell lysate and Western blots full-length jun family gene (pMcjun, pMjunD, and pMjunB) were provided by Dr Sunita Agarwal (Agarwal et al., 1999). One6105 cells/well of cells were seeded in six well plates. The Expression constructs of Gal4dbd fused with transactivation next day, cells were treated with or without TPA+RA in domain (amino acids (aa) 1 to 223) of c-jun (pFA-cJun) or C- EMEM with 0.2% FBS for appropriate time. Cells were terminal transactivation domain (aa 208 ± 313, pFA-cFos) washed with ice cold PBS once then lysed with lysis buer was purchased from Stratagene. C-terminal part of fral (aa (2% SDS, 50 mM Tris-HCl pH 7.5). Western immunoblotting 128 to 271) or fra2 (146 to 327) with partial leucine zipper was carried out according to the ECL protocol from fused to Gal4dbd fusion constructs were described previously Amersham Pharmacia Biotech Inc. In brief, 10 mg of nuclear (Young et al., 2001). An expression construct of Gal4dbd extracts or 20 ± 40 mg of whole cell lysates were boiled and fused with the transactivation domain (aa 1 to 151) of junD denatured in Sample Buer containing SDS and DTT (pBindjunD) was provided by Dr Tim Bowden (Rosenberger (Invitrogen) followed by gel electrophoresis using NuPage et al., 1999). An expression construct of Gal4dbd fused with 10% Bis-Tris pre-packed gel (Invitrogen) in MOPS buer. the transactivation domain of junB (aa 1 to 203) was made Equal protein was loaded for each sample. The proteins were by excising SacI±HindIII fragment of pMjunB. The RARE- electro-transferred to nitrocellulose membrane (Schleicher Luciferase construct was described previously (Li et al., and Schuell, Keene, NH, USA) using a semi-dry transfer 1996). Mouse full-length SMRT expression construct was blotting system from Enprotech Co. (Hyde Park, MA, USA). kindly provided by Dr R Evans (Chen et al., 1996). The resulting protein-bound membrane was blotted with Full-length mouse cfos, fra-1, and fra-2 cDNA were ampli®ed selected antibodies and visualized using ECL reagents by PCR with following sets of primers. cfos: 5'-ATGGATC- (Amersham Pharmacia). The cells for all samples were CATGATGTTCTCGGGTTTCAAC-3' and 5'-TAGAATTC- cultured and induced at the same time and the protein TCACAGGGCCAGCAGCGTGG-3' fra-1: 5'-ATGGATCC- extracts were also prepared at the same time. CAGGGCATGTACCGAGAC-3' and 5'-TAGAATTCTCA- CAAAGCCAGGAGTGTAG-3' fra-2: 5'-ATGGATCCATG- Luciferase assay for AP-1 dependent transactivation TACCAGGATTATCCCGG-3' and 5'-TAGAATTCTTAC- AGGGCTAGAACTGTGG-3'. Ampli®ed cDNAs were di- Two6104/well of JB6 P (+) (Clone 41) cells were seeded in gested with EcoRI and BamHI then cloned into EcoRI ± 24 well plates. The next day cells were transfected with BamHI site of pFA expression vector (Stratagene) to make Fugene 6 (Roche Molecular Biochemicals), 0.4 mg AP-1 fusion constructs with the DNA binding domain of the Gal4 46reporter construct, and 0.4 mg jun or fos cDNA expression gene. vector. Total DNA amount was adjusted by adding pcDNA3

Oncogene AP-1 transrepression by retinoic acid in JB6 cells K Suzukawa and NH Colburn 2189 vector. Six hours after the start of transfection, cells were speci®c promoter was normalized to its respective Renilla cultured in complete medium for 24 h. Then medium was Luciferase activity driven by thymidine kinase promoter as a changed to EMEM with 0.2% FBS with or without TPA control for transfection eciency. (10 ng/ml)+RA (1077) for 18 h. Cells were washed with PBS and lysed in the passive lysis buer (Promega Corp., Madison, WI, USA). Luciferase activity was measured with Note added in proof Luciferase assay kit (Promega) and a microtiter plate This article is a `United States Government Work' paper as luminometer (MLX, Dynex technologies Inc.). de®ned by the US Copyright Act.

Gal4 transactivation assay One6104 cells were seeded in 24 well plates. The next day, Acknowledgments 25 ng of pFR-luc (reporter construct with Gal4 binding site) We are grateful to P Chambon, S Agarwal, Y Nakatani, L and 50 ng of fusion construct of Gal4dbd and Jun or Fos Feltzien,TBowden,REvansandMLazarforkindly gene were transfected using Fugene 6 (Roche). One day later, providing constructs, and to Matthew Young for valuable cells were treated with or without TPA+RA in 0.2% FBS discussions. Kazumi Suzukawa was supported by a Japan with EMEM for 6 h. Luciferase activity was measured as Society for Promotion of Science Fellowship for Japanese described above. Each Fire¯y Luciferase activity driven by a Biomedical and Behavioral Researchers at NIH.

References

Agarwal SK, Guru SC, Heppner C, Erdos MR, Collins RM, Fearon ER and Vogelstein B. (1990). Cell, 61, 759 ± 767. Park SY, Saggar S, Chandrasekharappa SC, Collins FS, Felzien LK, Woendin C, Hottiger MO, Subbramanian RA, Spiegel AM, Marx SJ and Burns AL. (1999). Cell, 96, Cohen EA and Nabel GJ. (1998). Proc. Natl. Acad. Sci. 143 ± 152. USA, 95, 5281 ± 5286. Angel P, Szabowski A and Schorpp-Kistner M. (2001). Fisher GJ and Voorhees JJ. (1996). FASEB J., 10, 1002 ± Oncogene, 20, 2413 ± 2423. 1013. Ayer DE. (1999). Trends Cell. Biol., 9, 193 ± 198. Fuchs E and Green H. (1980). Cell, 19, 1033 ± 1042. Bernstein LR, Bravo R and Colburn NH. (1992). Mol. FukasawaH,YamaguchiM,HashimotoY,EndoYand Carcinog., 6, 221 ± 229. Shudo K. (2000). Biol. Pharm. Bull., 23, 1414 ± 1417. Bernstein LR and Colburn NH. (1989). Science, 244, 566 ± Goyette P, Feng Chen C, Wang W, Seguin F and Lohnes D. 569. (2000). J. Biol. Chem., 275, 16497 ± 16505. Bernstein LR and Walker SE. (1999). Biochim. Biophys. Huang C, Li J, Ma WY and Dong Z. (1999). J. Biol. Chem., Acta., 1489, 263 ± 280. 274, 29672 ± 29676. Blanco JC, Minucci S, Lu J, Yang XJ, Walker KK, Chen H, HuangC,MaWY,YoungMR,ColburnNandDongZ. Evans RM, Nakatani Y and Ozato K. (1998). Genes Dev., (1998). Proc. Natl. Acad. Sci. USA, 95, 156 ± 161. 12, 1638 ± 1651. Kakizawa T, Miyamoto T, Ichikawa K, Takeda T, Suzuki S, Bollag W and Holdener EE. (1992). Ann. Oncol., 3, 513 ± 526. Mori J, Kumagai M, Yamaguchi K and Hashizume K. Boutwell RK. (1989). Prog. Clin. Biol. Res., 298, 3±15. (2000). J. Biol. Chem., 27, 27. Caelles C, Gonzalez-Sancho JM and Munoz A. (1997). Genes Kamei Y, Xu L, Heinzel T, Torchia J, Kurokawa R, Gloss B, Dev., 11, 3351 ± 3364. Lin SC, Heyman RA, Rose DW, Glass CK and Rosenfeld Chambon P. (1996). FASEB J., 10, 940 ± 954. MG. (1996). Cell, 85, 403 ± 414. Chen JD and Evans RM. (1995). Nature, 377, 454 ± 457. Klein G. (1981). Nature, 294, 313 ± 318. Chen JD, Umesono K and Evans RM. (1996). Proc. Natl. Lee HY, Sueoka N, Hong WK, Mangelsdorf DJ, Claret FX Acad. Sci. USA, 93, 7567 ± 7571. and Kurie JM. (1999). Mol. Cell. Biol., 19, 1973 ± 1980. Chen JY, Penco S, Ostrowski J, Balaguer P, Pons M, Starrett LeeSK,KimHJ,NaSY,KimTS,ChoiHS,ImSYandLee JE, Reczek P, Chambon P and Gronemeyer H. (1995). JW. (1998). J. Biol. Chem., 273, 16651 ± 16654. EMBO J., 14, 1187 ± 1197. Lee SK, Kim JH, Lee YC, Cheong J and Lee JW. (2000). J. Colburn NH, Former BF, Nelson KA and Yuspa SH. (1979). Biol. Chem., 275, 12470 ± 12474. Nature, 281, 589 ± 591. Li JJ, Dong Z, Dawson MI and Colburn NH. (1996). Cancer Colburn NH, Ozanne S, Lichti U, Ben T, Yuspa SH, Wendel Res., 56, 483 ± 489. E, Jardini E and Abruzzo G. (1981). Ann. NY Acad. Sci., Li JJ, Rhim JS, Schlegel R, Vousden KH and Colburn NH. 359, 251 ± 259. (1998). Oncogene, 16, 2711 ± 2721. Colburn NH, Wendel E and Srinivas L. (1982). J. Cell. Li JJ, Westergaard C, Ghosh P and Colburn NH. (1997). Biochem., 18, 261 ± 270. Cancer Res., 57, 3569 ± 3576. DiSepio D, Sutter M, Johnson AT, Chandraratna RA and Nagy L, Kao HY, Chakravarti D, Lin RJ, Hassig CA, Ayer Nagpal S. (1999). Mol. Cell. Biol. Res. Commun., 1, 7 ± 13. DE,SchreiberSLandEvansRM.(1997).Cell,89,373 ± 380. Dong Z, Birrer MJ, Watts RG, Matrisian LM and Colburn NicholsonRC,MaderS,NagpalS,LeidM,Rochette-EglyC NH. (1994). Proc. Natl. Acad. Sci. USA, 91, 609 ± 613. and Chambon P. (1990). EMBO J., 9, 4443 ± 4454. Dong Z, Lavrovsky V and Colburn NH. (1995). Carcinogen- Pfahl M. (1993). Endocr. Rev., 14, 651 ± 658. esis, 16, 749 ± 756. Resche-Rigon M and Gronemeyer H. (1998). Curr. Opin. Dong Z, Xu R-H, Kim J, Zhan S-N, Ma W-Y, Colburn NH Chem. Biol., 2, 501 ± 507. and Kung H-F. (1996). J. Biol. Chem., 271, 9942 ± 9946. Rosenberger SF, Finch JS, Gupta A and Bowden GT. Fanjul A, Dawson MI, Hobbs PD, Jong L, Cameron JF, (1999). J. Biol. Chem., 274, 1124 ± 1130. Harlev E, Graupner G, Lu XP and Pfahl M. (1994). Rudd CJ, Mansbridge JN, Suing KD and Dawson MI. Nature, 372, 107 ± 111. (1993). Cancer Lett., 73, 41 ± 49.

Oncogene AP-1 transrepression by retinoic acid in JB6 cells K Suzukawa and NH Colburn 2190 Saatcioglu F, Claret FX and Karin M. (1994). Semin. Cancer Watts RG, Huang C, Young MR, Li JJ, Dong Z, Pennie WD Biol., 5, 347 ± 359. and Colburn NH. (1998). Oncogene, 17, 3493 ± 3498. Schule R and Evans RM. (1991). Trends Genet., 7, 377 ± 381. Yang-Yen HF, Zhang XK, Graupner G, Tzukerman M, Schule R, Rangarajan P, Yang N, Kliewer S, Ransone LJ, Sakamoto B, Karin M and Pfahl M. (1991). New Biol., 3, Bolado J, Verma IM and Evans RM. (1991). Proc. Natl. 1206 ± 1219. Acad. Sci. USA, 88, 6092 ± 6096. Young MR, Li JJ, Rincon M, Flavell RA, Sathyanarayana Simonson MS. (1994). Am. J. Physiol., 267, F805 ± F815. BK, Hunziker R and Colburn N. (1999). Proc. Natl. Acad. Smith SE and Bohmann D. (1992). Cell Growth Dier., 3, Sci. USA, 96, 9827 ± 9832. 523 ± 529. Young MR, Nair R, Bucheimer N, Tulsian P, Brown N, SteinB,BaldwinJrAS,BallardDW,GreeneWC,AngelP Chapp C, Hsu T-C and Colburn NH. (2002). Mol. Cell. and Herrlich P. (1993). EMBO J., 12, 3879 ± 3891. Biol., 22, 587 ± 598. Sutherland JA, Cook A, Bannister AJ and Kouzarides T. Zhou XF, Shen XQ and Shemshedini L. (1999). Mol. (1992). Genes Dev., 6, 1810 ± 1819. Endocrinol., 13, 276 ± 285. Verma AK and Boutwell RK. (1977). Cancer Res., 37, 2196 ± 2201.

Oncogene