Portal de Periódicos da CAPES

Mutual regulation of c-Jun and ATF2 by transcriptional activation and subcellular localization

2006; Springer Nature; Volume: 25; Issue: 5 Linguagem: Inglês

10.1038/sj.emboj.7601020

1460-2075

Han Liu, Xuehong Deng, Y. John Shyu, Jian Jian Li, Elizabeth J. Taparowsky, Chang‐Deng Hu,

Cancer therapeutics and mechanisms

Article2 March 2006free access Mutual regulation of c-Jun and ATF2 by transcriptional activation and subcellular localization Han Liu Han Liu Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University Search for more papers by this author Xuehong Deng Xuehong Deng Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University Search for more papers by this author Y John Shyu Y John Shyu Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University Search for more papers by this author Jian Jian Li Jian Jian Li School of Health Science, Purdue University, West Lafayette, IN, USA Purdue Cancer Center, Purdue University, West Lafayette, IN, USA Search for more papers by this author Elizabeth J Taparowsky Elizabeth J Taparowsky Department of Biological Sciences, Purdue University, West Lafayette, IN, USA Purdue Cancer Center, Purdue University, West Lafayette, IN, USA Search for more papers by this author Chang-Deng Hu Corresponding Author Chang-Deng Hu Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University Purdue Cancer Center, Purdue University, West Lafayette, IN, USA Walther Cancer Institute, Indianapolis, IN, USA Search for more papers by this author Han Liu Han Liu Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University Search for more papers by this author Xuehong Deng Xuehong Deng Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University Search for more papers by this author Y John Shyu Y John Shyu Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University Search for more papers by this author Jian Jian Li Jian Jian Li School of Health Science, Purdue University, West Lafayette, IN, USA Purdue Cancer Center, Purdue University, West Lafayette, IN, USA Search for more papers by this author Elizabeth J Taparowsky Elizabeth J Taparowsky Department of Biological Sciences, Purdue University, West Lafayette, IN, USA Purdue Cancer Center, Purdue University, West Lafayette, IN, USA Search for more papers by this author Chang-Deng Hu Corresponding Author Chang-Deng Hu Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University Purdue Cancer Center, Purdue University, West Lafayette, IN, USA Walther Cancer Institute, Indianapolis, IN, USA Search for more papers by this author Author Information Han Liu1, Xuehong Deng1, Y John Shyu1, Jian Jian Li2,4, Elizabeth J Taparowsky3,4 and Chang-Deng Hu 1,4,5 1Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University 2School of Health Science, Purdue University, West Lafayette, IN, USA 3Department of Biological Sciences, Purdue University, West Lafayette, IN, USA 4Purdue Cancer Center, Purdue University, West Lafayette, IN, USA 5Walther Cancer Institute, Indianapolis, IN, USA *Corresponding author. Department of Medicinal Chemistry and Molecular Pharmacology, School of Pharmacy, Purdue University, 575 Stadium Mall Drive, RHPH 224D, West Lafayette, IN 47907, USA. Tel.: +1 765 496 1971; Fax: +1 765 494 1414; E-mail: [email protected] The EMBO Journal (2006)25:1058-1069https://doi.org/10.1038/sj.emboj.7601020 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info ATF2 and c-Jun are key components of activating protein-1 and function as homodimers or heterodimers. c-Jun–ATF2 heterodimers activate the expression of many target genes, including c-jun, in response to a variety of cellular and environmental signals. Although it has been believed that c-Jun and ATF2 are constitutively localized in the nucleus, where they are phosphorylated and activated by mitogen-activated protein kinases, the molecular mechanisms underlying the regulation of their transcriptional activities remain to be defined. Here we show that ATF2 possesses a nuclear export signal in its leucine zipper region and two nuclear localization signals in its basic region, resulting in continuous shuttling between the cytoplasm and the nucleus. Dimerization with c-Jun in the nucleus prevents the export of ATF2 and is essential for the transcriptional activation of the c-jun promoter. Importantly, c-Jun-dependent nuclear localization of ATF2 occurs during retinoic acid-induced differentiation and UV-induced cell death in F9 cells. Together, these findings demonstrate that ATF2 and c-Jun mutually regulate each other by altering the dynamics of subcellular localization and by positively impacting transcriptional activity. Introduction ATF2 belongs to the basic region leucine zipper (bZIP) family of proteins and is an important member of activating protein-1 (AP-1) (Wagner, 2001). ATF2 functions as a homodimer or as a heterodimer with other bZIP proteins to bind specific DNA sequences and activate gene expression. The proto-oncoprotein c-Jun is a major dimerization partner of ATF2, and c-Jun–ATF2 heterodimers are important for many cellular processes. One major role of ATF2 is to regulate the response of cells to stress signals (Gupta et al, 1995; Whitmarsh and Davis, 1996; Karin et al, 1997; Hayakawa et al, 2004; Bhoumik et al, 2005). ATF2 also contributes to cellular transformation induced by several viral proteins, including adenovirus E1A (Liu and Green, 1990, 1994), and, in conjunction with c-Jun, mediates distinct processes of nonviral cellular transformation (van Dam and Castellazzi, 2001; Eferl and Wagner, 2003). ATF2 also plays a role in regulating development of various organs in mice (Reimold et al, 1996; Maekawa et al, 1999) and cellular differentiation in vitro (Monzen et al, 2001). For example, treatment of F9 mouse embryonic tetratocarcinoma with retinoic acid (RA) induces differentiation (Yang-Yen et al, 1990; Alonso et al, 1991), which is associated with the binding of ATF2 and the p300 coactivator to a differentiation response element (DRE) (Kawasaki et al, 1998). Although it has been well documented that c-Jun–ATF2 heterodimers are responsible for the activation of target genes involved in stress response, it remains unknown whether ATF2 alone, or in cooperation with c-Jun, regulates F9 cell differentiation. Accumulated evidence suggests that two events are common to the activation of AP-1 proteins. The first is phosphorylation by a mitogen-activated protein kinase (MAPK) and the second is the selective formation of dimers. In mammals, three major MAPKs can phosphorylate and activate ATF2 and c-Jun. These are the extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinase (JNK) and p38 (Davis, 2000; Kyriakis and Avruch, 2001). In response to growth and stress signals c-Jun is phosphorylated on residues S63 and S73, while ATF2 is phosphorylated on residues T69 and T71. Although these MAPK phosphorylation events are critical for the full transcriptional activity of c-Jun and ATF2 (Pulverer et al, 1991; Smeal et al, 1991; Gupta et al, 1995; Livingstone et al, 1995; Jiang et al, 1996; Stein et al, 1997; Ip and Davis, 1998; Ouwens et al, 2002), the underlying mechanism of this activation is poorly defined. It has been proposed that phosphorylation by JNK may regulate the intrinsic histone acetylase activity of ATF2 (Kawasaki et al, 2000) or the interaction of c-Jun with the coactivator p300 (Arias et al, 1994; Bannister et al, 1995). It also has been proposed that phosphorylation of c-Jun and ATF2 by JNK/p38 may prohibit their ubiquitination (Fuchs et al, 1996, 1997, 1998; Fuchs and Ronai, 1999), leading to increased levels of these bZIP proteins in cells. Given that ATF2 is ubiquitously and abundantly expressed in many tissues while the amount of c-Jun in cells is very limited (Angel et al, 1988; Chiu et al, 1989; Takeda et al, 1991; Stein et al, 1992; van Dam et al, 1993, 1995; Herdegen and Leah, 1998), it is apparent that additional cellular mechanisms, including events leading to increased transcription of c-jun, must be operating to control the levels and activities of these bZIP proteins. Using a bimolecular fluorescence complementation (BiFC) assay and fluorescent protein fusions, we present evidence here that ATF2 monomers and ATF2 homodimers are localized predominantly in the cytoplasm. We have identified a nuclear export signal (NES) in the leucine zipper region and two nuclear localization signals (NLS) in the DNA-binding domain of ATF2. These nuclear transport signals contribute to the shuttling of the protein between the cytoplasm and the nucleus. We also demonstrate, for the first time, that heterodimerization with c-Jun prevents nuclear export of ATF2 and is the key event leading to nuclear localization of c-Jun-ATF2 dimers and the transcriptional activity of this complex towards targets such as c-jun. Results Distinct subcellular localization of AP-1 dimers and AP-1 proteins We previously developed a BiFC assay using yellow fluorescent protein (YFP) to visualize protein–protein interactions in living cells (Hu et al, 2002). Since chromophore maturation and protein folding of YFP is sensitive to higher temperatures (Tsien, 1998), a preincubation at 30°C for a few hours is necessary before visualization of BiFC signals (Hu et al, 2005). To circumvent this problem, we have recently identified several new combinations of fluorescent protein fragments that significantly increase the BiFC signal at 37°C culture conditions and display a two-fold increase in specificity (Shyu et al, 2006). The combination using N-terminal residues 1–172 (VN173) and C-terminal residues 155–238 (VC155) of Venus fluorescent protein showed higher BiFC signals when bJun–bFos interactions were examined at 37°C (Supplementary Figure 1). To determine the subcellular localization of AP-1 dimers with the newly identified BiFC fragments, we expressed c-Fos, c-Jun and ATF2 as fusion proteins with either VN173 or VC155 in COS-1 cells. Although fluorescent signals derived from JunVN173–FosVC155 and JunVN173–JunVC155 were localized, as predicted, in the nucleus (Figure 1A and B), 90% of fluorescent signals derived from ATF2VN173–ATF2VC155 were located in the cytoplasm. Interestingly, we observed that the BiFC signals derived from JunVN173–ATF2VC155 were located equally in the cytoplasm and the nucleus, whereas the majority of JunYN155–ATF2YC155 was localized in the cytoplasm when YFP fragments were used (Hu et al, 2002). This difference may be accounted for by two major differences in experimental approaches: the lack of the quantification of fluorescence intensity and the exposure of cells to lower temperatures in our previous work. Figure 1.Subcellular localization of AP-1 dimers and proteins. (A) Plasmids encoding c-Fos, c-Jun and ATF2 fused to N-terminal residues 1–172 (VN), C-terminal residues 155–238 (VC) of Venus, or full-length Venus (Venus) were cotransfected into COS-1 cells. Representatives of fluorescent images of different AP-1 dimers and proteins captured at 12 h post-transfection are shown. Venus alone was included as a control. (B) Quantification of subcellular localization of different AP dimers and proteins from (A). The error bar indicates standard deviation. Download figure Download PowerPoint Since subcellular localization of transcription factors is determined by all interacting partners (Hu et al, 2002; Hu and Kerppola, 2003; Grinberg et al, 2004), we examined the subcellular localization of ATF2 by itself. Using ATF2 fused to full-length Venus, we again observed predominant cytoplasmic localization, whereas Fos-Venus and Jun-Venus were localized in the nucleus (Figure 1A and B). A similar profile of ATF2-Venus localization was also observed in human HEK293 and MCF-7 cells (data not shown). The distinct cytoplasmic localization of ATF2-Venus was in sharp contrast to the nuclear localization of Fos-Venus and Jun-Venus, suggesting that it was unlikely that the cytoplasmic localization of ATF2 was due to saturation of nuclear import machinery by overexpressed proteins. To rule out the possibility that the fusion tag of Venus to the C-terminus of ATF2 may uniquely impede nuclear import of ATF2, we expressed ATF2 as a FLAG fusion protein in COS-1 cells. Similar cytoplasmic localization of the FLAG–ATF2 fusion proteins was detected by immunostaining with anti-FLAG antibody (Supplementary Figure 2). Thus, we conclude that cytoplasmic localization of exogenously expressed ATF2 is not an artifact caused by protein overexpression or by the presence of a fusion tag. ATF2 shuttles between the cytoplasm and the nucleus Since subcellular localization of transcription factors can be affected by the relative rates of nuclear import versus export, we examined if the cytoplasmic localization of ATF2-Venus was a result of rapid nuclear export. To this end, we treated cells with leptomycin B (LMB), a specific inhibitor of chromosome region maintenance 1 (CRM1) involved in nuclear export (Nishi et al, 1994; Kudo et al, 1997, 1998; Ossareh-Nazari et al, 1997; Wolff et al, 1997; Yashiroda and Yoshida, 2003). After treatment for 12 h, 85% of ATF2-Venus was sequestered in the nucleus (Figure 2A). This demonstrates that ATF2 can be exported in a CRM1-dependent manner and that the ATF2-Venus fusion protein is competent to shuttle between the cytoplasm and the nucleus. Treatment of cells expressing Jun-Venus and Fos-Venus with LMB did not alter their persistent nuclear localization (data not shown). Figure 2.Identification of NLS and NES motifs in ATF2. (A) Plasmids encoding wild-type ATF2, an NES mutant of ATF2 [ATF2(NES4A)], or ATF2(400–415) fused to Venus were transfected into COS-1 cells. At 12 h post-transfection, cells were treated with or without 20 ng/ml of LMB for an additional 12 h and fluorescent images were captured. The number indicates the percentage of quantified nuclear distribution ±standard deviation. (B) Alignment of ATF2 NES with several characterized NES motifs. The conserved hydrophobic residues are highlighted. L can be replaced with V, F and I, and X can be any amino acid. (C) Alignment of ATF2 bipartite NLS with several characterized NLS motifs. The conserved basic residues are highlighted. (D) Functional analysis of NLSs. Cells expressing the indicated ATF2 mutants fused to Venus were treated with 20 ng/ml of LMB, or left untreated, for 12 h and the fluorescence intensity localized in the nucleus was quantified. The number indicates the percentage of quantified nuclear distribution ±standard deviation. Download figure Download PowerPoint CRM1-dependent nucleocytoplasmic shuttling proteins contain an NES, which is composed of one or more leucine-rich motifs. The predicted consensus motif of an NES is L–X(2–3)–L–X(2–3)–L–X–L (Mattaj and Englmeier, 1998). Sequence analysis of ATF2 revealed that residues 405–413 of ATF2 matched perfectly with several well-characterized NES motifs (Figure 2B), suggesting that this region may act as an NES. To test this, the conserved valine and three leucine residues within the region were replaced with alanines and the subcellular localization of the mutant ATF2-Venus (ATF2(NES4A)-Venus) was examined. Consistent with our prediction, 67% of ATF2(NES4A)-Venus was localized in the nucleus (Figure 2A). Also, the fusion of this region only to Venus localized 87% of Venus to the cytoplasm (Figure 2A). The basic region of many bZIP transcription factors has a dual role as a sequence-specific DNA-binding domain and an NLS. To provide evidence that the basic region of ATF2 functions as an NLS, we first identified two potential bipartite NLS motifs in the ATF2 basic region (Figure 2C). Next, we used deletion mutagenesis to test each motif for function. Deletion of either NLS alone did not affect the nuclear sequestration of ATF2-Venus in the presence of LMB, whereas deletion of both NLS motifs completely abolished the nuclear sequestration of ATF2-Venus by LMB (Figure 2D). These results indicate that both NLS motifs are functional and either of them is sufficient to translocate ATF2 to the nucleus. We conclude that ATF2 represents the first AP-1 protein discovered to possess both NES and NLS motifs, and that the coordinated function of these sequences mediates the observed nucleocytoplasmic shuttling. Nuclear localization of ATF2 depends on physical interactions with c-Jun The identification of an NES in ATF2 coupled with the cytoplasmic localization of exogenously expressed ATF2 prompted us to examine how the nuclear localization of ATF2 is regulated. As c-Jun can form heterodimers with ATF2 in both the cytoplasm and the nucleus (Figure 1A and B), we reasoned that c-Jun may facilitate the nuclear localization of ATF2. To test this possibility, we examined the effect of c-Jun expression on the subcellular localization of ATF2-Venus. Indeed, increased expression of c-Jun significantly enhanced the nuclear accumulation of ATF2 (Figure 3A). With a 3:1 ratio of plasmids encoding c-Jun to those encoding ATF2-Venus, over 80% of ATF2-Venus was concentrated in the nucleus (Figure 3B). In contrast, c-Jun(ΔL3), a dimerization-deficient c-Jun mutant (Hu et al, 2002), failed to localize ATF2-Venus to the nucleus (Figure 3A and B), demonstrating that nuclear localization of ATF2-Venus depends on dimerization with c-Jun. To further confirm this result, we employed subcellular fractionation to prepare cytosolic and nuclear extracts and detected the amount of FLAG-ATF2-Venus localized in both cytosolic and nuclear fractions. Consistent with fluorescence microscopic analysis, we observed that coexpression with increasing amount of c-Jun increased nuclear localization of FLAG-ATF2-Venus from 5 to 70% (Supplementary Figure 3). Figure 3.Jun-dependent nuclear localization of ATF2. (A) Nuclear localization of ATF2 is dependent on its heterodimerization. Plasmids encoding wild-type or mutant ATF2 fused to Venus were cotransfected with 10-fold excess of plasmids encoding wild-type c-Jun or the indicated mutants into COS-1 cells. Fluorescent images shown were captured at 12 h post-transfection. The number indicates the percentage of quantified nuclear distribution ±standard deviation. (B) Dose-dependent nuclear localization of ATF2 induced by c-Jun. Different amounts of plasmid encoding c-Jun (•) or its dimerization-deficient mutant c-JunΔL3 (▪) were cotransfected with a fixed amount of plasmid encoding ATF2-Venus into COS-1 cells. The total amount of plasmids was adjusted to 0.5 μg. At 12 h post-transfection, fluorescent images were captured and the percentage of nuclear-localized ATF2-Venus was determined as described in 'Materials and methods.' The inset shows the levels of wild-type and mutant c-Jun detected with anti-c-Jun antibody in cells transfected with 10-fold excess of plasmids encoding c-Jun or c-JunΔL3 (upper panel). The same blot was used for the detection of β-actin (lower panel). (C) Immunostaining of ATF2 in F9, COS-1 and MCF-7 cells. Endogenous ATF2 was detected with an anti-ATF2 antibody and DNA was stained with DAPI. (D) Immunostaining of ATF2 in control siRNA-expressing MCF-7 cells (upper panel) or c-Jun siRNA-expressing MCF-7 cells (lower panel). The arrow indicates the cell showing even distribution of ATF2 in both the cytoplasm and the nucleus. Download figure Download PowerPoint To determine whether c-Jun is also required for the nuclear localization of endogenous ATF2, we examined the subcellular localization of ATF2 in F9 cells that lack detectable levels of c-Jun (Yang-Yen et al, 1990; van Dam et al, 1995). In agreement with c-Jun-dependent nuclear localization of exogenously expressed ATF2, the majority of ATF2 was localized in the cytoplasm of F9 cells (Figure 3C). Consistent with this notion, the majority of ATF2 was localized in the nucleus of c-Jun-expressing cells, such as COS-1 and MCF-7 (Figure 3C). More importantly, knockdown of c-jun by siRNA in MCF-7 cells increased the cytoplasmic localization of ATF2 from 20 to 50% (Figure 3D and Supplementary Figure 4). These findings lead us to conclude that c-Jun is required for the nuclear localization of ATF2 at both exogenous and endogenous levels. Since phosphorylation by the JNK/p38 MAPKs on c-Jun and ATF2 is required for the activation of c-Jun and ATF2 (Whitmarsh and Davis, 1996; Karin et al, 1997), we examined the impact of phosphorylation on the dimerization and nuclear localization of ATF2. Several experiments were performed and the results indicated that the phosphorylation status of neither c-Jun nor ATF2 regulates their interactions or subcellular localization. First, cells expressing ATF2-Venus coexpressed with JNK or p38, or treated with the JNK-specific inhibitor, SP600125, or the p38 inhibitor, SB203580, did not alter ATF2 cytoplasmic localization (Supplementary Figure 5A and B). Second, the interactions between c-Jun(63A, 73A) and ATF2(69A, 71A) or ATF2(69D, 71D) as measured by BiFC were essentially the same as their wild-type counterparts (Supplementary Figure 5C). Third, nuclear localization of ATF2-Venus was facilitated by c-Jun(63A, 73A) (Figure 3A) as efficiently as wild-type c-Jun. Fourth, ATF2(69A, 71A)-Venus and ATF2(69D, 71D)-Venus also were localized predominantly in the cytoplasm (Figure 3A) and their nuclear localization can be facilitated by wild-type and mutant c-Jun. Finally, we observed that TAM67, a deletion mutant of c-Jun lacking the N-terminal 123 residues, which is transcriptionally inert (Brown et al, 1994), remained capable of facilitating the nuclear localization of ATF2-Venus (Figure 3A). Taken together, these results clearly demonstrate that physical interaction with c-Jun is the major driving force of ATF2 nuclear localization. Heterodimerization with c-Jun prevents nuclear export of ATF2 Since the increased nuclear localization of ATF2-Venus by heterodimerization with c-Jun could be attributed to either increased nuclear import or decreased export, we compared the import rate of the dimeric versus monomeric form of ATF2. ATF2(L408P) contains a substitution of proline for the fourth leucine (L408) in the leucine zipper region. This mutation abolishes both ATF2 homodimer and c-Jun–ATF2 heterodimer formation (Abdel-Hafiz et al, 1993; Fuchs and Ronai, 1999). Although cytoplasmic localization of ATF2-Venus and ATF2(L408P)-Venus was similarly observed 12 h after transfection (Figure 4A), treatment of cells with LMB resulted in a rapid nuclear accumulation of ATF2(L408P)-Venus, but not ATF2-Venus, in less than 2 h (Figure 4B). This result suggests that ATF2 monomers are translocated into the nucleus more efficiently than ATF2 dimers. To provide direct evidence that heterodimerization with c-Jun does not facilitate nuclear import of ATF2, we coexpressed ATF2(ΔNLS1+2)-Venus with an excess amount of c-Jun. Again, overexpression of c-Jun failed to localize ATF2(ΔNLS1+2)-Venus to the nucleus, although they interacted with each other in the cytoplasm (Figure 4A). Figure 4.Heterodimerization of ATF2 with Jun family of proteins facilitates nuclear localization. (A) Jun-independent nucleocytoplasmic shuttling of ATF2. Subcellular localization of wild-type or ATF2(L408P) fusions with Venus with or without treatment of 20 ng/ml of LMB for 2 h (top two panels), or ATF2(ΔNLS1+2) coexpressed without or with excess amount of c-Jun (the third panel). The bottom panel shows the interaction of ATF2(ΔNLS1+2) with c-Jun in BiFC assay. (B) Time course of nuclear sequestration of ATF2-Venus and ATF2(L408P)-Venus by LMB. (C) Alignment of leucine zipper regions of AP-1 proteins. The NES in ATF2 is highlighted. (D) Jun-dependent nuclear localization of ATF2. Plasmids encoding ATF2-Venus were cotransfected with 10-fold excess of plasmids encoding the indicated proteins into COS-1 cells. Fluorescent images were captured at 12 h post-transfection and nuclear localization of ATF2-Venus was quantified. The error bar indicates standard deviation. (E) Inhibition of CRM1–ATF2 interaction by c-Jun. Plasmids encoding CRM1(566–720) fused to VN173 (CRM1VN) and plasmids encoding ATF2(342–505) or ATF2(342–505, NES4A) fused to VC155 (ATF2VC or ATF2NES4AVC) were cotransfected into cells in the absence or presence of full-length c-Jun or ATF2(342–505) coexpression. As an internal control, plasmids encoding ECFP were cotransfected. At 12 h after transfection, images were acquired using both CFP and YFP filters. The median of YFP/CFP ratios derived from more than 100 transfected cells for each group was determined as described in 'Materials and methods', and is presented on the right of fluorescent images. Download figure Download PowerPoint Next, we examined if the nuclear accumulation of ATF2 in the presence of c-Jun was due to impaired nuclear export. The NES of ATF2 is located in the fourth heptad of the leucine zipper region (Vinson et al, 2002), suggesting that heterodimerization with c-Jun may mask the NES and prevent the nuclear export. Interestingly, c-Jun, JunB and JunD have identical sequences in their fourth heptads with variable substitutions across the first three heptads (Figure 4C). This implies that JunB and JunD should affect the subcellular localization like c-Jun. Consistent with this prediction, JunB and JunD also sequestered ATF2-Venus in the nucleus (Figure 4D). To further confirm this, we examined if c-Fos sequesters ATF2 in the nucleus. c-Fos has been shown to interact with ATF2 in vitro (Kerppola and Curran, 1993) and in vivo (unpublished observations). A comparison of the fourth heptad sequences of c-Fos and ATF2 reveals one repulsive force in positions 'g' and 'e', as well as the lack of one hydrophobic interaction in positions 'a' and 'd'. These unfavorable interactions could decrease the stability of a c-Fos–ATF2 dimer across this critical region (Vinson et al, 2002). In support of our hypothesis, overexpression of c-Fos failed to sequester ATF2-Venus in the nucleus (Figure 4D). Additionally, expression of other ATF2 coactivators, such as p65 (Kim and Maniatis, 1997), p300 (Kawasaki et al, 1998; Sano et al, 1998) and E1A (Liu and Green, 1990, 1994), failed to sequester ATF2-Venus in the nucleus (Figure 4D). Finally, a specific interaction between CRM1 and ATF2 was observed using the BiFC assay (Figure 4E, left two panels). Furthermore, the specific interaction between CRM1 and ATF2 was almost completely abolished by the coexpression with c-Jun, and to a lesser extent by the coexpression with ATF2. These results provide evidence that formation of a coiled-coil structure between the fourth heptads of ATF2 and the Jun proteins masks the NES of ATF2 and prevents the nuclear export of the protein. Activation of c-jun promoter by ATF2 requires dimerization with c-Jun, but is independent of c-Jun phosphorylation A major ATF2 target gene is c-jun, which has been reported to be activated by c-Jun–ATF2 heterodimers and ATF2 homodimers (Devary et al, 1991; Stein et al, 1992; van Dam et al, 1993, 1995; Herr et al, 1994). Interestingly, transient expression of ATF2 in cells barely activates reporter genes (Liu and Green, 1990; Ivashkiv et al, 1992; Li and Green, 1996; Sano et al, 1998). Based on our findings here, we reasoned that this may be the result of the cytoplasmic localization of ATF2. Since heterodimerization with c-Jun is essential for ATF2 nuclear localization, it is therefore likely that the nuclear anchoring of ATF2 by c-Jun is required for its transcriptional activity. To examine this possibility, we utilized COS-1 cells to monitor expression of a luciferase reporter gene controlled by five tandem copies of the second AP-1-binding site (jun2) in the c-jun promoter. c-Jun–ATF2 heterodimers and ATF2 homodimers are known to bind the jun2 site (Devary et al, 1991; Stein et al, 1992; van Dam et al, 1993, 1995; Herr et al, 1994). Consistent with previous reports, the jun2-luc reporter was not activated by exogenous ATF2, but was activated by exogenous c-Jun in a dose-dependent manner (Figure 5A). Interestingly, coexpression of c-Jun and ATF2 synergistically activated the reporter gene, showing at least two to three times higher activation than c-Jun alone. To test if the transcriptional activity of c-Jun also is required for the activation of the jun2-luc reporter, we expressed ATF2 with c-Jun(63A, 73A), a dominant-negative c-Jun known to be a weak transactivator of the collagen promoter and other AP-1 reporter plasmids (Hu et al, 2002). Results showed that the jun2-luc reporter was activated by the c-Jun(63A, 73A) alone, or in combination with ATF2 (Figure 5B). In contrast, the transcriptionally impaired, dominant-negative ATF2(69A, 71A), produced a 50% reduction in luciferase expression compared to wild-type ATF2 when coexpressed with either wild-type c-Jun or c-Jun(63A, 73A). Equivalent expression of all activator proteins was confirmed by immunoblotting analysis. These results demonstrate that activation of c-jun transcription by ATF2 requires c-Jun as a nuclear anchor and dimerization partner and that phosphorylation of ATF2, but not c-Jun, has an impact on the transcriptional activity of ATF2. Figure 5.Synergistic activation of c-jun transcription by c-Jun and ATF2. (A) The indicated amount of plasmids encoding c-Jun and ATF2 were transfected separately, or cotransfected into serum-starved COS-1 cells along with 0.5 μg of the reporter plasmid jun2-luc and 50 ng of pRL-TK using Fugene 6. Fold increase of F/R ratio was calculated when compared with vector control only. (B) Similar experiments were performed as described in (A), except that 1 μg of each plasmid encoding the indicated wild-type or mutant proteins was used for transfection. *P<0.05 when compared with c-Jun+ATF2 (Student's t-test). Download figure Download PowerPoint c-Jun induction is associated with