Critical Role of N-terminal End-localized Nuclear Export Signal in Regulation of Activating Transcription Factor 2 (ATF2) Subcellular Localization and Transcriptional Activity*

Background: Alteration of ATF2 subcellular localization is involved in multiple human diseases. Results: A second nuclear export signal was identified in the most N-terminal end of ATF2. Conclusion: The newly identified nuclear export signal negatively regulates ATF2 nuclear localization and transcriptional activity. Significance: Understanding how ATF2 subcellular localization is regulated will provide insight into the pathological role of ATF2 in human diseases. Activating transcription factor 2 (ATF2) belongs to the basic leucine zipper family of transcription factors. ATF2 regulates target gene expression by binding to the cyclic AMP-response element as a homodimer or a heterodimer with c-Jun. Cytoplasmic localization of ATF2 was observed in melanoma, brain tissue from patients with Alzheimer disease, prostate cancer specimens, and ionizing radiation-treated prostate cancer cells, suggesting that alteration of ATF2 subcellular localization may be involved in the pathogenesis of these diseases. We previously demonstrated that ATF2 is a nucleocytoplasmic shuttling protein, and it contains two nuclear localization signals in the basic region and one nuclear export signal (NES) in the leucine zipper domain (named LZ-NES). In the present study, we demonstrate that a hydrophobic stretch in the N terminus, 1MKFKLHV7, also functions as an NES (termed N-NES) in a chromosome region maintenance 1 (CRM1)-dependent manner. Mutation of both N-NES and LZ-NES results in a predominant nuclear localization, whereas mutation of each individual NES only partially increases the nuclear localization. These results suggest that cytoplasmic localization of ATF2 requires function of at least one of the NESs. Further, mutation of N-NES enhances the transcriptional activity of ATF2, suggesting that the novel NES negatively regulates the transcriptional potential of ATF2. Thus, ATF2 subcellular localization is probably modulated by multiple mechanisms, and further understanding of the regulation of ATF2 subcellular localization under various pathological conditions will provide insight into the pathophysiological role of ATF2 in human diseases.

ATF2 (activating transcription factor 2) is a basic leucine zipper (bZIP) 3 protein and is an important member of the activator protein-1 (AP-1) family (1). Like other bZIP transcription factors, ATF2 utilizes its basic region to bind DNA and its leucine zipper region to dimerize with its partners. ATF2 binds the cAMP-response element, T(G/T)ACGTCA, as a homodimer or ATF2⅐c-Jun heterodimer (2)(3)(4), and regulates transcription of a large set of genes, including those involved in anti-apoptosis (5,6), cell growth (7,8), and DNA damage response (9).
The transcriptional activity of ATF2 can be regulated by multiple mechanisms. First, the transcriptional activity of ATF2 is regulated by phosphorylation at Thr-69 and Thr-71 within the N-terminal transactivation domain (10 -12). In response to growth and stress signals, these two residues can be phosphorylated by mitogen-activated protein kinases (MAPKs), including extracellular signal regulated kinase (ERK), c-Jun N-terminal kinase (JNK), and p38 (10 -12). The phosphorylation-induced transcriptional activity may be achieved by increasing the intrinsic histone acetylase activity of ATF2 (13), promoting DNA binding (14), and/or preventing it from ubiquitination (15,16). Second, the transcriptional activity of ATF2 is regulated by an intramolecular interaction between the N-terminal transactivation domain and the bZIP domain (17). This is largely based on the observation that the activity of ATF2 is limited when it is exogenously expressed without coexpression with its dimerization partners or coactivators (17)(18)(19)(20). It was proposed that the activity of ATF2 is repressed by an intramolecular interaction between the N-terminal domain and the bZIP domain (17). However, when coexpressed with coactivators, such as CREB-binding protein or E1A (21), the intramolecular interaction is disrupted. Third, the transcriptional activity is regulated by its subcellular localization. We recently demonstrated that ATF2 is a nucleocytoplasmic shuttling protein, and its subcellular localization is regulated by its dimerization (19). Interestingly, overexpressed ATF2 is predominantly localized in the cytoplasm due to the inadequate level of Jun proteins anchoring ATF2 in the nucleus (19). This finding not only uncovers a novel regulatory mechanism of ATF2 transcriptional activity but also provides an alternative mechanism, that the limited transcriptional activity of overexpressed ATF2 may be due to its insufficient nuclear localization in addition to the intramolecular interaction.
Alteration of ATF2 subcellular localization appears to be implicated in the pathogenesis of several human diseases. In melanoma specimens, cytoplasmic localized ATF2 is associated with a better outcome, whereas strong nuclear staining correlates with a poor prognosis (22). In skin cancer, reduced nuclear ATF2 was observed in most basal and squamous cell carcinoma cells (23). In the brain, cytoplasmic ATF2 was observed in cortical neurons of patients with Alzheimer disease (24). In prostate cancer, increased cytoplasmic phosphorylated ATF2 was linked to disease progression (25). Moreover, we recently reported that ionizing radiation-induced cytoplasmic sequestration of ATF2 is involved in neuroendocrine differentiation in prostate cancer cells (26,27). Although the impact of ATF2 subcellular localization appears to be dependent on disease context, these reports highlight the importance of understanding the molecular mechanisms regulating ATF2 subcellular localization.
Currently, the mechanism by which ATF2 subcellular localization is regulated is not well understood. We have found that ATF2 is a nucleocytoplasmic shuttling protein with a nuclear export signal (NES) and two nuclear localization signals (NLSs) in its leucine zipper and basic region, respectively (19). We have also demonstrated that heterodimerization with the Jun family proteins anchors ATF2 in the nucleus by preventing its nuclear export. On the contrary, homodimerization of ATF2 in the cytoplasm hampers the nuclear import (19). Here, we further show that ATF2 possesses another NES located in the most N-terminal end, which functions in cooperation with the previously identified NES to export ATF2 out of the nucleus. Importantly, this NES negatively regulates ATF2 transcriptional activity.
Cell Culture and Fluorescence Imaging Analysis-COS-1 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum and antibiotics at 37°C in 5% CO 2 . Forty thousand cells per well were seeded in 12-well plates to grow overnight and then transfected with 2 g of expression vectors indicated in each experiment using Fugene 6 (Roche Applied Science). At 16 -18 h post-transfection, cells were fixed with 3.7% formaldehyde in phosphate-buffered saline (PBS) for 20 min. To inhibit nuclear export of ATF2 fusion proteins, transfected cells, at 14 h after transfection, were treated with 20 ng/ml leptomycin B (LMB) for 3.5 h before fixation. Fixed cells were permeabilized with 0.2% Triton X-100 in PBS for 5 min followed by counterstaining with 2.5 g/ml 4Ј,6-diamidino-2-phenylindole (DAPI) to visualize nuclei. Localization of various ATF2-Venus(A206K) fusion proteins were examined by a Nikon TE2000-U inverted fluorescence microscope with YFP filters as reported previously (19). All images for purposes of quantification were taken at ϫ300 magnification. To determine ATF2 localization, five regions of interest, with a diameter of ϳ4.5 m, were randomly selected from the nucleus and perinuclear regions, respectively, followed by measuring the intensity of Venus fluorescence within regions of interest using MetaMorph II software. The acquired emission intensities were background-corrected by subtracting mean fluorescence intensity of fields without cells. The localization of wild-type or mutant ATF2-Venus(A206K) within a cell was determined by calculating the ratio of mean nuclear fluorescence intensity to mean cytoplasmic fluorescence intensity. More than 50 individual cells were quantified in each set of experiments, and the ratio of nuclear to cytoplasmic intensity of each fusion protein was presented as mean Ϯ S.E. from at least three independent experiments. The integrity of all Venus fusion proteins was confirmed by Western blot (supplemental Fig. 1).
To determine the effect of the LZ-NES mutation on ATF2⅐c-Jun heterodimerization, HEK 293T cells were transfected with 5 g of FLAG-ATF2-Venus(A206K) or FLAG-ATF2(LZ-NES3A)-Venus(A206K) for 1 day. Cell lysate was prepared as described above and then subjected to co-IP with anti-FLAG antibody (10 g) at 4°C overnight, followed by incubation with protein G beads at 4°C for 2 h. The beads were washed with co-IP buffer six times with a 3-min rotation at 4°C each time. The associated c-Jun was then detected by Western blot.
Luciferase Reporter Gene Assay-COS-1 cells were cultured in 12-well plates and serum-starved for 18 h. Cells were transfected with 2 g of each indicated plasmid along with 0.5 g of the jun2-luc reporter plasmid (luciferase reporter construct driven by five tandem repeats of Jun2) (32) and 100 ng of pRL-TK (Promega) for 24 h. Firefly and Renilla luciferase activities were measured by the Dual Luciferase Reporter Assay kit (Promega) according to the manufacturer's protocol with the following minor modifications. Firefly luciferase activity was measured by mixing 10 l of cell lysate with 50 l of LARII reagent. Renilla luciferase activity was then determined by mixing the sample with 50 l of Stop & Glo reagent. The luciferase activity was measured by a TopCount NXT microplate luminescence counter (Packard). Luciferase data were presented as mean Ϯ S.E. from at least three independent experiments performed in duplicate.
Immunostaining-COS-1 cells were fixed at 16 h post-transfection or at the end of LMB treatment with 3.7% formaldehyde for 20 min. Cells were then permeabilized with 0.2% Triton X-100 for 5 min, followed by incubation with mouse monoclonal anti-Myc or anti-HA antibody (1:200 dilution) for 1 h. After three washes with PBS, cells were stained with Texas Red-conjugated anti-mouse IgG secondary antibody and DAPI for another 1 h. Microscopy was similarly performed as reported previously (26). To quantify the nuclear distribution of ATF2 mutants in Fig. 5, COS-1 cells were seeded on coverslips and co-transfected with indicated plasmids and pHA-Venus for 16 h. After performing Myc and DAPI staining, the coverslips were mounted on glass slides with Vectashield Hard Set mounting medium (Vector Laboratories, Burlingame, CA). Venus fluorescent signal was used to define the region of the whole cell body. The percentage of nucleus-localized ATF2 in each individual cell was calculated as (integrated intensity of Myc signal in the nucleus/integrated intensity of Myc signal in the whole cell body) ϫ 100%. Results were presented as means Ϯ S.E. quantified from more than 60 cells.
UV Irradiation-COS-1 cells expressing ATF2(LZ-NES3A)-Venus were irradiated with UVC (120 J/m 2 ) in a Spectrolinker XL-1000 UV cross-linker (Spectronic Corp.). Cells were harvested at the indicated time points and subjected to Western blot with anti-phospho-ATF2 antibody. The phosphorylation status of the exogenous ATF2 was quantified by ImageJ software.
Statistical Analysis-The data were expressed as means Ϯ S.E. and analyzed by GraphPad Prism 5 (La Jolla, CA). Student's t test was used to determine the difference between two groups. For multiple-group comparisons, data were analyzed by oneway analysis of variance, followed by Tukey post-test if significant differences were found.

N-terminal Domain Promotes ATF2 Nuclear Export in CRM1-dependent Manner-
We previously demonstrated that ATF2 contains an NES in the leucine zipper region (termed LZ-NES), which, in conjunction with two NLSs in the basic region, regulates ATF2 nucleocytoplasmic shuttling (Fig. 1A). We also observed that deletion of the NTD (residues 1-341) of ATF2 increased nuclear localization of Jun-ATF2 heterodimers (33), whereas overexpressed full-length ATF2 was predominantly localized in the cytoplasm despite possession of the two NLS motifs (19). These results suggest that the NTD of ATF2 may negatively regulate ATF2 nuclear localization. To test this possibility, we examined the effect of an NTD deletion on ATF2 subcellular localization. Wild-type and NTD-deleted ATF2 were expressed as fusion proteins with a monomeric form (A206K mutant) of the fluorescent protein Venus in COS-1 cells. As shown in Fig. 1B, deletion of the NTD alone resulted in a predominant nuclear localization, demonstrating that the NTD does play a negative role in the nuclear localization of ATF2. To determine the subcellular localization of the NTD, we expressed ATF2(1-341) as a fusion with Venus(A206K) and observed that ATF2(1-341)-Venus(A206K) was predominantly localized in the cytoplasm (Fig. 1B). This result suggests that the NTD of ATF2 may possess a cytoplasmic localization activity. Interestingly, treatment of cells with LMB, a CRM1specific nuclear export inhibitor (34), significantly increased the nuclear localization of ATF2(1-341)-Venus(A206K) (Fig.  1C). Thus, predominant cytoplasmic localization of the ATF2 NTD is probably mediated by a CRM1-dependent nuclear export mechanism. To determine whether the nuclear export activity conferred by the NTD is sufficient to drive ATF2 export, we tested the effect of LMB on the localization of ATF2(LZ-NES3A), a previously identified LZ-NES mutant ( 405 VAQLKQLLL 413 to 405 VAQAKQALA 413 ) (19). As we observed previously (19), mutation of LZ-NES significantly increased nucleus-localized ATF2 (Fig. 1D). Interestingly, treatment of cells with LMB further sequestered ATF2(LZ-NES3A) in the nucleus. Taken together, these results suggest that the ATF2 NTD may possess an additional CRM1-dependent nuclear export activity. ATF2 NTD Does Not Contain Canonical NES-To determine whether the NTD has any NES motifs, we searched the N-terminal 341 residues of ATF2 using Net NES1.1 (35). We also manually compared the amino acid sequence within the NTD against the canonical NES sequence ⌽ 1 X 2-3 ⌽ 2 X 2-3 ⌽ 3 X⌽ 4 (where ⌽ represents Leu, Ile, Val, Phe, or Met, and X is any amino acid) (35,36). These searches identified four putative NES motifs: PNES1, PNES2, PNES3, and PNES4 ( Fig. 2A). To determine if these motifs are functional NESs, we mutated the conserved hydrophobic residues (shown in boldface type in Fig.  2A) within each putative NES to alanines in the context of ATF2(LZ-NES3A). As shown in Fig. 2, B and C, all of these mutations did not increase the nuclear localization of ATF2(LZ-NES3A) in the absence of LMB, whereas LMB treatment still sequestered the mutant fusions in the nucleus. Moreover, ATF2(PNES1-4m, LZ-NES3A)-Venus(A206K), a mutant with all four putative NESs and LZ-NES mutated, remained responsive to LMB. These results suggest that these four putative NES motifs in the NTD do not confer the CRM1dependent nuclear export activity of ATF2 and that they are not functional NESs.
Residues 1-73 of ATF2 Contain Sequence Required for CRM1-dependent Nuclear Export-Because mutation analysis suggested that the ATF2 NTD does not possess a canonical NES, we then made an N-terminal truncation and a series of internal deletions in the NTD to map the region responsible for mediating the CRM1-dependent nuclear export. To exclude the potential impact of LZ-NES on ATF2 subcellular localization, all of these mutants were generated in the context of ATF2(LZ-NES3A) (Fig. 3A). As shown in Fig. 3, B and C, ATF2(⌬1-73, LZ-NES3A)-Venus(A206K) was almost completely localized in the nucleus, whereas other ATF2 mutants did not show any significant increase in the nuclear localization. Importantly, LMB treatment further increased the nuclear localization of all other ATF2 mutants but not ATF2(⌬1-73, LZ-NES3A)-Venus(A206K). These results suggest that the first 73 residues of ATF2 contribute to the CRM1-dependent nuclear export of ATF2 NTD. Consistent with this notion, the ATF2(1-341) fragment only was predominantly localized in the cytoplasm, whereas the ATF2(74 -341) fragment was distributed in both the nucleus and the cytoplasm (supplemental Fig. 2).
Identification of Non-canonical NES in Most N-terminal End of ATF2-To confirm that the first 73 residues of ATF2 may contain an autonomous NES, we fused ATF2  to the N terminus of 2xVenus(A206K). The fusion to two Venus proteins increases the molecular mass (to ϳ65 kDa) of the fusion protein and thus attenuates its passive diffusion rate across the nuclear pore complex. Similar strategies have been utilized by others to evaluate NES activity (37)(38)(39). As shown in Fig. 4A, ATF2(1-73)-2xVenus(A206K) was predominantly localized in the cytoplasm. Importantly, treatment of cells with LMB altered its predominant cytoplasmic localization and increased its localization in the nucleus, reinforcing our hypothesis that a CRM1-dependent NES resides in this region. Because canonical NES is a hydrophobic residue-enriched motif, we then searched for hydrophobic residue-enriched regions within the first 73 residues and identified 1 MKFKLHV 7 (termed N-NES) as the only potential hydrophobic stretch. Although this sequence does not match the consensus NES, a similar atypical NES has been reported in Eps15 (40) (Fig. 4B). To determine if the hydrophobic stretch functions as an NES, we mutated the four hydrophobic residues to alanines (N-NES4A: 1 MKFKLHV 7 to 1 AKAKAHA 7 ) in ATF2(1-73). As shown in Fig. 4B, transiently expressed ATF2(1-73, N-NES4A)-2xVenus(A206K) was distributed throughout the cell, suggesting that the hydrophobic residues in the most N-terminal end of ATF2 may function as an NES. We also confirmed that the N-NES is functional in other cell lines, including HEK 293T, LNCaP, and PC3 cells (supplemental Fig. 3). To determine that this atypical NES can be recognized by CRM1, we performed a co-IP assay to verify the interaction between CRM1 and ATF2(1-341). We coexpressed Myc-CRM1 and FLAG-Venus(A206K)-ATF2(1-341) or FLAG-Venus(A206K)-ATF2(1-341, N-NES4A) in COS-1 cells and then performed co-IP with anti-FLAG antibody. The insertion of Venus between FLAG and ATF2 is to avoid masking of the FLAG epitope when CRM1 binds to the NTD of ATF2. We found that Myc-CRM1 was efficiently co-immunoprecipitated with FLAG-Venus(A206K)-ATF2(1-341) but not FLAG-Venus(A206K)-ATF2(1-341, N-NES4A) (Fig. 4C), suggesting that the hydrophobic stretch 1 MKFKLHV 7 is recognized and bound by CRM1.
To examine the impact of N-NES on ATF2 subcellular localization, we introduced the mutations into ATF2 in the context of full-length ATF2 and determined the localization of ATF2. Mutation of N-NES significantly increased nuclear distribution of ATF2, indicating that N-NES does regulate the subcellular localization of ATF2 (Fig. 4D). Importantly, ATF2 was predominantly localized in the nucleus when both N-NES and LZ-NES were mutated. Moreover, LMB treatment did not further increase the nuclear localization of this mutant (supplemental Fig. 4), suggesting that no additional CRM1-dependent NES exists in ATF2. Function of N-NES in full-length ATF2 was also confirmed in HEK 293T and LNCaP cells, albeit N-NES seems to be a dominant export signal in both cell types (supplemental Figs. 5 and 6). Taken together, these results demonstrate that both NESs contribute to the nuclear export of ATF2 and suggest that a complete nuclear localization of ATF2 may require functional inactivation of both NESs.
Transcriptional Activity of ATF2 Is Negatively Regulated by N-NES-We previously demonstrated that the lack of transcriptional activity of exogenously expressed ATF2 is probably due to the predominant cytoplasmic localization of overexpressed ATF2 (19). The increase in nuclear localization of N-NES-disrupted ATF2 prompted us to examine if its transcriptional activity is enhanced. We similarly used the jun2 luciferase reporter gene to examine the ATF2 transcriptional activity (9,19,41). Consistent with our previous observation (19), wild-type ATF2 barely activated the jun2-luc reporter (Fig.  5A). Interestingly, ϳ3-fold activation of jun2-luc by ATF2(N-NES4A) was observed. On the contrary, ATF2 with mutations in LZ-NES or both NESs did not activate the jun2-luc reporter. Because LZ-NES of ATF2 spans between the fourth and fifth heptad of the leucine zipper region and replacement of the leucine in the fourth heptad (corresponding to the Leu-408 of ATF2) with alanine was shown to reduce the stability of leucine-zipper dimer (42, 43), we tested if LZ-NES3A mutation affects heterodimerization of ATF2 and c-Jun, the complex binding to the jun2 response element (44). A co-IP was used to determine if this mutation affects the binding of exogenous ATF2 to endogenous c-Jun. HEK 293T cells were used due to the high transfection efficiency, which allowed us to efficiently co-immunoprecipitate endogenous c-Jun. As shown in Fig. 5B, LZ-NES3A mutation reduced the interaction between ATF2 and c-Jun, suggesting that the stability of ATF2⅐c-Jun complex is essential for its transcriptional activity.
We next sought to confirm that the subcellular localization of these ATF2 proteins is similar to that of Venus-fused ones.  MARCH 9, 2012 • VOLUME 287 • NUMBER 11

JOURNAL OF BIOLOGICAL CHEMISTRY 8627
To avoid the potential masking of FLAG epitope by the CRM1/ N-NES interaction, we fused a Myc tag to the C-terminal end of each protein and determined the percentage of nucleus-localized ATF2 mutants by immunostaining with anti-Myc antibody. As shown in Fig. 5C, mutation of N-NES increased the nuclear distribution of ATF2 from 14 to 37%, whereas mutation of both NESs further brought the percentage of ATF2 in the nucleus up to 79%. We then performed the same reporter assay and found that mutation of N-NES also significantly enhanced transcriptional activity of the ATF2-Myc fusion protein, albeit to a lesser extent when compared with the counterpart without Myc tag (supplemental Fig. 7). In summary, these results suggest that N-NES negatively regulates ATF2 nuclear localization and the transcriptional activity.
N-NES Is Not Regulated by MAPK Phosphorylation-Transcriptional activity of ATF2 is regulated by MAPKs, which phosphorylate Thr-69/71 in the NTD. We previously demonstrated that MAPK phosphorylation did not affect the subcellular localization of full-length ATF2 with intact N-NES and LZ-NES (19). To determine whether MAPK-mediated phosphorylation has any impact on N-NES, we substituted Thr-69 and -71 with non-phosphorylatable alanine, phosphomimic aspartic acid, or glutamic acid in the background of ATF2(LZ-NES3A). As shown in Fig. 6A, none of the substitutions altered the subcellular localization of ATF2. We also found that the exogenous ATF2(LZ-NES3A)-Venus(A206K) but not the alanine mutant was readily phosphorylated at Thr-71 in cells (Fig.  6B). To further evaluate the effect of ATF2 Thr-69/Thr-71 phosphorylation on N-NES activity, COS-1 cells expressing ATF2(LZ-NES3A)-Venus(A206K) were subjected to UV irradiation, which induces ATF2 phosphorylation by activating JNK and p38 kinases (9,12,45). No significant change in subcellular localization of the exogenous ATF2 was observed (Fig.  6C), although UV irradiation indeed enhanced ATF2 phosphorylation (Fig. 6D). Taken together, these results indicate that the N-NES is unlikely to be regulated by MAPK-mediated phosphorylation.

DISCUSSION
We previously demonstrated that ATF2 possesses two NLS and one NES (LZ-NES) and constitutively shuttles between the cytoplasm and nucleus in a monomeric form (19). In the present study, we identified another NES (N-NES) in the most N-terminal end of ATF2, suggesting that the subcellular localization of ATF2 is modulated by multiple layers of regulations. Nuclear localization of ATF2 requires formation of a heterodimer with c-Jun, which masks the LZ-NES of ATF2 and anchors ATF2 in the nucleus. With limited c-Jun protein, ATF2 is predominantly located in the cytoplasm due to two mechanisms. First, ATF2 homodimerization in the cytoplasm can inhibit nuclear import (19). Second, localization of the ATF2 monomer is determined by the balance between N-NES, LZ-NES, and two NLSs. The nuclear export rate contributed by both N-NES and LZ-NES seems to outweigh the nuclear import rate conferred by the two NLSs because the homodimerizationdeficient mutant ATF2(L4P) is predominantly located in the cytoplasm (19).
Since identification of the consensus NES motifs (36), many nucleocytoplasmic shuttling proteins have been demonstrated to possess the canonical NES motif. The NES motif is a hydrophobic-rich region, defined as ⌽ 1 X 2-3 ⌽ 2 X 2-3 ⌽ 3 X⌽ 4 (where ⌽ represents Leu, Ile, Val, Phe, or Met, and X is any amino acid) (35,36). Interestingly, only 72% of reported NES motifs match this well defined consensus NES (35), and the remaining motifs have only partially conserved NESs. As presented in this study, the N-NES of ATF2, MKFKLHV, has only one amino acid residue between ⌽ 1 /⌽ 2 and ⌽ 2 /⌽ 3 , respectively. Because similar sequence presented in Eps15 acts as a functional NES (40), we have provided several lines of evidence to demonstrate that the N-NES is also functional. First, the N-terminal 73 residues are sufficient to render a tandem Venus fusion to be localized in the cytoplasm (Fig. 4A). Second, this tandem Venus fusion is responsive to LMB treatment. Third, mutation of N-NES in the context of full-length ATF2 increased ATF2 nuclear localization (Fig. 4D). Finally, CRM1 can be coimmunoprecipitated with the wild-type N-terminal domain of ATF2 but not the mutant form (Fig. 4C). Thus, ATF2 has a canonical NES in its leucine zipper domain and a non-canonical NES in the most N-terminal end.
The N-terminal end localization of N-NES may allow it to be more accessible to CRM1 (46). Interestingly, due to its unique location, a careful experimental design is needed in order to determine ATF2 subcellular localization. We observed that an N-terminal FLAG tag immediately adjacent to the N-NES prevented us from staining the nucleus-localized FLAG-ATF2 (data not shown), which is probably due to the binding of CRM1 to the N-NES of FLAG-ATF2 in the nucleus. Similarly, we also observed a slight but significant increase in nuclear localization of ATF2 when Venus is fused to the N terminus of the protein (supplemental Fig. 8), which may be due to a steric hindrance for the CRM1/N-NES interaction. Therefore, C-terminal fluorescent protein or tag fusions are preferred when exogenous ATF2 fusion proteins are used for the analysis of ATF2 subcellular localization.
Nucleocytoplasmic shuttling of transcription factors has emerged as an important research area. Many transcription fac- tors critical to cell fate determination are tightly regulated by multiple NLS and NES motifs. This may allow regulation of these transcription factors by multiple mechanisms. For example, p53 contains two NESs and three NLSs (47)(48)(49)(50). The N-terminal NES is regulated by phosphorylation events in response to DNA damage (50), whereas the NES in the tetramerization domain is masked upon p53 oligomerization (49). In the case of ATF2, we have shown that the function of LZ-NES is inhibited when ATF2 heterodimerizes with the Jun family proteins. However, it is currently unknown whether and how the N-NES of ATF2 is regulated. Because phosphorylation at Thr-69 and Thr-71 by MAPKs is known to stimulate transactivation activity of ATF2, we initially thought that MAPK phosphorylation may regulate ATF2 subcellular localization. However, it does not appear to be the case because substitutions of these two residues to non-phosphorylatable alanines or phosphorylationmimicking aspartic acid or glutamic acid residues did not affect the localization of ATF2(LZ-NES3A) (Fig. 6A). In addition, further induction of phosphorylation at these two residues by UV did not affect the localization of this mutant (Fig. 6C). Because ATF2 is also phosphorylated at other sites in the NTD (51,52), it remains to be investigated whether phosphorylation of these additional sites regulates the activity of N-NES. Interestingly, mutation of LZ-NES increased nuclear localization of ATF2 in COS-1 cells but not in HEK 293T or LNCaP cells, suggesting that N-NES is a dominant export signal in the latter two cell types. One possible explanation is that N-NES activity is enhanced by signaling that is activated in these cell types. In this case, comparing the difference of cellular signaling between cell types with differential N-NES activity may help unravel mechanisms that regulate N-NES. Alternatively, activity of N-NES may be modulated by ATF2-interacting proteins, such as E1A and CREB-binding protein (21). Also, undifferentiated embryonic cell transcription factor 1 (UTF1) (53) and activating signal cointegrator-2 (ASC-2) (54) are reported to bind the NTD of ATF2 and function as coactivators. It will be interesting to test if these coactivators enhance ATF2 transcriptional activity by masking N-NES and anchoring ATF2 in the nucleus.
Our results showed that mutation of N-NES enhanced transcriptional activity of FLAG-ATF2 and FLAG-ATF2-Myc by ϳ90 and 30%, respectively ( Fig. 5 and supplemental Fig. 7). Because C-terminal Myc tag does not appear to affect ATF2 stability (data not shown), it may reduce the transcriptional potential of ATF2 by causing a conformational change. Indeed, it has been reported that function of transcription factors can be affected by N-or C-terminal tags (55,56). Nevertheless, the  reporter gene activation by ATF2 with mutations in the N-NES correlates with its increased nuclear localization, suggesting that regulation of N-NES by itself could increase the transcriptional activity of ATF2. In this sense, it is worth noting that there are multiple ATF2 alternative spliced variants that have been deposited to the NCBI database, although their functions remain largely uninvestigated. For example, ATF2-Var13 (GenBank TM number AAY17215.1) lacks the N-terminal 18 residues due to the loss of exon 3, where N-NES is located. Based on our finding here, it is very likely that the transcriptional activity of ATF2-Var13 may be less subject to the regulation by its subcellular localization. Therefore, a switch between spliced variants may be an alternative mechanism to regulate ATF2 subcellular localization and its activity. Consistent with this notion, alignment of full-length ATF2 with those from several species reveals that N-NES is only present in Homo sapiens, Canis familiaris, and Bos taurus ATF2 and not in Gallus gallus, Mus musculus, Rattus norvegicus, and Xenopus laevis (supplemental Fig. 9). It is therefore possible that this additional NES is evolutionarily acquired to deal with more complex regulatory mechanisms in higher organisms. Given that alteration of ATF2 subcellular localization has been observed in several human diseases (22)(23)(24)(25)(26), further analysis of the regulation of N-NES under these pathological conditions will probably provide exciting insight into the role of ATF2 in the pathogenesis of human diseases.