DNA Damage–Dependent Translocation of B23 and p19^ARF Is Regulated by the Jun N-Terminal Kinase Pathway

Introduction

A growing body of evidence attributes a new role to the nucleolus, extending beyond its traditional activity of ribosome production, stress response ( 1, 2). Among the proteins that comprise it, some are important regulators of the cell cycle and stress responses. The localization of two of these proteins, p14^ARF/p19^ARF and B23 (nucleophosmin, numartin, or NO38), is regulated by genotoxic stress ( 3– 5).

The human p14^ARF/mouse p19^ARF are important tumor suppressors. Knockout of p19^ARF underscored its importance in cancer prevention, as p19^ARF-deficient animals are more prone to cancer ( 6). The tumor suppressor effects of alternative reading frame (ARF) are mainly relayed by regulation of the p53 pathway ( 7, 8). ARF interacts and inactivates Mdm2, thereby increasing the stability and activity of the tumor suppressor p53 ( 9, 10). Hence, increased expression of ARF results in a p53-dependent growth arrest or apoptosis. However, increased tumor development in triple knockout mice nullizygous for ARF, p53, and Mdm2 showed that ARF acts also in a p53-independent manner ( 11). Some of the p53-independent activity is attributed to its ability to reduce rRNA processing ( 12, 13) and inhibit oncogene-induced transcription ( 14, 15).

ARF augments p53 activity mainly in response to oncogenic stress. Its expression is up-regulated in response to deregulated oncogenic activity due to elevated transcription governed by several transcription factors, such as E2F ( 16), c-myc ( 17), AP-1 ( 18), or by oncogenic Ras ( 19). A second mechanism regulating ARF activity involves the control of its subcellular localization ( 3). The nucleolar localization of ARF raised a model suggesting that ARF sequesters Mdm2 to the nucleolus and thereby releases p53 from its inhibitory effects ( 11). However, later work ( 20) suggested that ARF may also posses growth-suppressing activity when it is located in the nucleoplasm. Moreover, recent studies have actually suggested that retention of ARF in the nucleolus inhibits its activity ( 3, 21), whereas its redistribution to the nucleoplasm enables extended interaction with Hdm2 ( 3). Furthermore, this last study showed that ARF is redistributed in the nucleus in response to DNA damage ( 3).

ARF localization is mostly governed by interactions with the abundant nucleolar protein B23 ( 22, 23). B23 functions as a molecular chaperone that shuttles between the nucleus and the cytoplasm. It affects cellular proliferation, resistance to DNA damage–dependent cell death, and maintenance of genomic stability, in addition to its activities in ribosome biogenesis and rRNA processing ( 24). Its interactions with ARF direct ARF to the nucleolus. In acute myeloid leukemia cells containing cytoplasmic mutants of B23, ARF is cytoplasmic as well ( 25, 26), and in cells deficient in B23, ARF is excluded from the nucleolus ( 27). Independently of ARF, B23 interacts with two key proteins in cell cycle regulation, Hdm2 and p53 ( 5, 28). Both of these interactions seem to lead to p53 stabilization ( 5, 28), in agreement with this hypothesized role of the protein in the maintenance of genomic stability. Moreover, increased expression of B23 has been associated with improved DNA repair ( 29), and B23 1–deficient cells exhibit increased H2AXγ phosphorylation ( 27). Its redistribution from the nucleolus in response to cytotoxic drugs and genotoxic stresses, such as the inhibition of RNA polymerase I and II or exposure to DNA intercalating agents or UV radiation, further supports its role in the response to DNA damage ( 3– 5, 30, 31). It seems that the translocation from the nucleolus is an important regulatory step that controls the effects of B23 on DNA repair.

Several members of the AP-1 transcription factor family are highly responsive to genotoxic stress. c-Jun and JunB are strongly induced by DNA-damaging agents such as UV radiation ( 32). Induction of c-Jun by DNA damage is mostly dependent on its upstream kinase, the c-Jun amino-terminal kinase (JNK; refs. 33, 34). However, JNK-independent inductions of c-Jun and JunB were also observed ( 35). JNK phosphorylates c-Jun at Ser⁶³ and Ser⁷³ and Thr⁹¹ and Thr⁹³, and regulates its transcriptional capacity and ability to interact with other proteins ( 36– 38). Induction of c-Jun is important for UV-driven apoptosis ( 39). Reports on the effects of JNK on JunB are mixed. Although early reports suggested that JunB does not possess JNK phosphorylation sites ( 40), a later report showed phosphorylation of Thr¹⁰² and Thr¹⁰⁴ by JNK ( 41). However, the significance of the induction of JunB by UV is not clear.

Currently, the pathways and mechanisms that regulate the redistribution of B23 and ARF after exposure to DNA-damaging agents are unknown. In light of the importance of these translocations, we explored the molecular mechanisms that regulate them. We show that activation of the JNK pathway is required for dispersal of B23 and p19^ARF from the nucleolus. The stress-induced translocation of ARF is mediated by JunB and c-Jun whose phosphorylation at Thr⁹¹ and Thr⁹³ is required for the translocation. Moreover, we show that B23 directly interacts with c-Jun, and that JNK-dependent translocation of c-Jun translocates associated B23. We also show that when JNK activity is reduced due to senescence, UV radiation does not trigger the translocation of ARF.

Materials and Methods

Cell lines and tissue culture. All the cells were grown in DMEM supplemented with 10% fetal bovine serum. The cell lines used in this work have been previously described as follows: c-jun^−/−p53^−/− by Shaulian et al. ( 39), 63/73 Ala by Behrens et al. ( 42), and 91/93 Ala by Laderoute et al. ( 43).

Plasmids. Expression vectors for c-jun, junB, and junD were described previously ( 44). cDNAs for wild-type c-Jun and 63/73Ala were cloned into pcDNA3 to generate cytomegalovirus (CMV)-derived expression vectors. CMV 91/93 Ala was obtained from G. Li and R. Johnson (University of California, San Diego, California). The conversion of Arg 272 and Lys 273 to glutamic acids was performed using the QuikChange site-directed mutagenesis kit II, according to the manufacturer's instructions (Stratagene). Expression vectors for green fluorescent protein (GFP)-fibrillarin and JDP2 were obtained from Y. Shav-Tal and A. Aronheim (Bar Ilan University, Israel and the Technion-Israel Institute of Technology, Israel, respectively). ShRNA for JunB was previously described ( 18).

Antibodies. Polyclonal anti–c-Jun antibody (H-79; Santa Cruz Biotech) was used for c-Jun immunoprecipitation, immunofluorescence staining, and immunoblotting. Monoclonal antibodies KM-1 (Santa Cruz Biotech) and C-J 4C4 (Abcam) were used to detect phospho-Ser⁶³ and phospho-Thr⁹¹-c-Jun, respectively by immunoblotting. Polyclonal anti–p19^ARF ab80 (Abcam) was used for p19^ARF staining and immunoblotting. Monoclonal mouse anti-B23 FC-61991 (Zymed; Invitrogen) was used for immunostaining, immunoprecipitation, and immunoblotting. Monoclonal anti-actin C4 (MP Biomedical) was used for immunoblotting. For detection of JunB, polyclonal anti-JunB N-17 (Santa Cruz Biotech) was used. HA-Tag 6E2 (Cell Signaling Technology) was used for the detection of HA-tagged proteins by immunofluorescence. For detection of JNK activation, anti–phospho-Thr¹⁸³/Tyr¹⁸⁵ JNK (Cell Signaling Technology) was used. Total JNK levels were determined using the monoclonal antibody G151-333 (BD Biosciences PharMingen).

Immunofluorescence staining. Cells were plated on coverslips and usually harvested 3 h after UV irradiation. Cells were fixed by a 30-min incubation in 4% paraformaldehyde at room temperature. After washes with PBS, cells were incubated with the primary antibody for 1 h at room temperature in PBS containing 3% bovine serum albumin and 0.3% triton. The slides were then washed with PBS with 0.1% Tween and incubated with fluorescently labeled secondary antibody for 40 min at room temperature. Slides were mounted using dibutylpthalate xylene mounting solution (Fluka Biochemika). Images were visualized using an Olympus fluorescence microscope. At least 200 randomly selected cells were counted at each point. We defined redistribution of ARF and B23 when cells exhibiting normal staining of 4′,6-diamidino-2-phenylindole (DAPI) did not exhibit visible nuclear foci of ARF and B23. In all cases, the fraction of cells expressing nucleolar ARF after treatment was scored. The Y axis title and the figure legends specify the 100% reference point of each experiment. Each experiment was repeated at least thrice.

Immunoprecipitations and immunoblotting. For immunoprecipitations, cell were lysed with NP40 lysis buffer [50 mmol/L Tris-HCl (pH 8.0), 120 mmol/L NaCl, 0.5% NP40, 1 mmol/L EDTA plus proteinase inhibitors (1:200 Cocktail III; Calbiochem), 0.1 mmol/L sodium orthovanadate, and 0.1 mmol/L sodium fluoride]. Two milligrams of total cell extract were used per reaction, and standard immunoprecipitation protocols were used throughout. Nuclear extracts were used for detection of JunB by immunoblotting analysis. Each experiment was repeated at least thrice.

Results

JNK activity regulates B23 and p19^ARF distribution. The recent studies describing the translocation of the nucleolar proteins B23 and p14^ARF after exposure of cells to genotoxic stresses ( 3– 5) prompted us to examine whether signaling specificity exists in the redistribution of these proteins. To examine the phenomena in relatively normal cells, we irradiated mouse embryo fibroblasts (MEF; passage 3) with 30 J/m² UV and 20 Gy of ionizing radiation and then stained for p19^ARF and B23 at 3 h after irradiation. As previously reported, in untreated cells, the majority of p19^ARF colocalized with B23 in the nucleoli ( Fig. 1A ), but the nucleolar staining of p19^ARF and B23 was significantly reduced after exposure to UV radiation. Interestingly, no change in the distribution of p19^ARF and B23 was observed after exposure of the cells to ionizing radiation. To determine whether the reduction in the staining after exposure to UV radiation reflects a reduction in the levels of p19^ARF or B23 that anchors p19^ARF to the nucleolus, we measured the level of both proteins before and 3 h after UV exposure. Within this time frame, the changes in the levels of p19^ARF and B23 were minor ( Fig. 1B). This suggests that the reduced nucleolar staining reflects changes in p19^ARF distribution and not a change in the total amounts of p19^ARF or B23.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Translocation of p19^ARF from the nucleolus correlates with JNK activation. A, early passage MEFs were irradiated with 30 J/m² UV or 20 Gy of ionizing radiation and stained for p19^ARF, B23, and with DAPI 3 h after irradiation. Pictures were taken through a fluorescent microscope. B, the protein levels of p19^ARF in the irradiated cells were determined 3 h after irradiation with 30 J/m² UV. Actin served as a loading control. C, early passage MEFs were irradiated with 30 J/m² UV or 20 Gy of ionizing radiation or treated with MMC or CDDP and harvested at the indicated times (hours after treatment). The level of phosphorylated JNK was determined using anti–phospho-Thr¹⁸³/Tyr¹⁸⁵-JNK antibody. JNK1 was used as a loading control. D, p19^ARF localization was determined by immunostaining the cells treated with UV, ionizing radiation, MMC, and CDDP; and the portion of cells retaining nucleolar p19^ARF at 4 h after treatment was plotted. Nucleolar ARF, the fraction of cells expressing nucleolar ARF from the total cell population.

One of the most significant differences between UV and ionizing radiations is the pathways that they activate. Specifically, there is a difference in their abilities to activate the JNK pathway ( 39). Therefore, we explored a possible correlation between JNK activation and translocation of p19^ARF in response to the different genotoxic stresses that activate JNK. To this end, we exposed MEFs to 30 J/m² UV, 20 Gy ionizing radiation, 50 μmol/L cisplatin (CDDP), or 50 μmol/L mitomycin C (MMC) and examined their ability to activate JNK and induce p19^ARF translocation. JNK activation was determined by measuring the extent of Thr¹⁸³/Tyr¹⁸⁵ phosphorylation, which is known to activate it (reviewed by ref. 45).

As depicted in Fig. 1C, UV is the most efficient activator of JNK and ionizing radiation is the weakest one. At the concentration examined, MMC activates JNK better than ionizing radiation but less efficiently than CDDP. The stress-dependent translocation by the different agents was closely correlated to the level of JNK activation. Nucleolar staining was retained in 76% of ionizing radiation–treated cells, 47% of MMC-treated cells, 18% of CDDP-treated cells, and in only 15% of the UV-treated cells ( Fig. 1D). To directly examine the involvement of JNK in the distribution of p19^ARF and B23, we used two cellular models. As p19^ARF expression is elevated in p53-deficient cells, it is easily detected by staining. Irradiation of p53^−/− fibroblasts with 30J/m² UV almost completely dispersed both B23 and p19^ARF from the nucleolus ( Fig. 2A–B ). However, a preincubation with 100 μmol/L of JNK inhibitor SP600125 at 1 h before treatment inhibited the translocation of p19^ARF from the nucleolus, and the percentage of cells in which nucleolar p19^ARF was detected increased >15-fold, from 3.98 to 61.8 ( Fig. 2A–B). As depicted in Fig. 2C, the significant increase in p19^ARF staining in the nucleolus is not a result of any increase in its overall quantity. In fact, SP600125 treatment slightly reduced the levels of p19^ARF. Significant repression of c-Jun phosphorylation served as a control for the inhibitor activity. To corroborate the notion that JNK controls p19^ARF and B23 nuclear localization, we also took a genetic approach and examined ARF translocation after UV exposure in JNK^+/+ mouse fibroblasts and JNK 1 and 2 double-knockout fibroblasts (JNK1,2^−/−), in which c-Jun is not phosphorylated by JNK in response to UV, and its induction is relatively limited (Supplementary Fig. S1A). As depicted in Fig. 2D, in the absence of JNK expression, the nucleolar staining of both B23 and p19^ARF is morphologically distinct, the nucleoli are smaller, and more B23 is located in the nucleoplasm. However, UV irradiation does not trigger the export of p19^ARF and B23 from the nucleolus in these cells; thus proving that JNK activity is involved in dispersing them from the nucleolus. Moreover, the dispersal of p19^ARF in JNK1,2^−/− after UV exposure was comparable with that of and JNK^+/+ cells treated with SP600125. Both were significantly higher than in SP600125-untreated, UV-irradiated JNK^+/+ cells (Supplementary Fig. S1B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Export of p19^ARF from the nucleolus is JNK-dependent. A, p53^−/− mouse fibroblasts were irradiated with 30 J/m² UV. Control fibroblasts were left untreated. Some of the cells were preincubated with a JNK inhibitor SP600125 (SP) 1 h before irradiation. The cells were than stained for p19^ARF and B23. B, quantitative analysis of p19^ARF dispersal, which is shown in A. Y-axis, percentage of cells containing nucleolar p19^ARF from the total cell population. Empty bars, SP600125-treated cells; black bars, untreated cells. C, protein extracts were prepared from p53^−/− fibroblasts treated as indicated in A; and the levels of p19^ARF, B23, c-Jun, and phospho-c-Jun were determined by immunoblotting with specific antibodies. Actin was used as a loading control. D, JNK^+/+ and JNK1,2^−/− double-knockouts were exposed to 30 J/m² UV, and p19^ARF and B23 localization was determined by immunostaining 3 h after irradiation. c, control.

c-Jun and JunB mediate the JNK-dependent redistribution of p19^ARF by UV. The results presented above show that the JNK pathway regulates the translocation of p19^ARF from the nucleolus. Therefore, it was of interest to identify what JNK substrates mediate p19^ARF redistribution. The plethora of JNK substrates complicates this task; however, AP-1 genes are clear candidates. As depicted in Figs. 2 and 3B , UV induces c-Jun and JunB in a kinetic that correlates with the p19^ARF redistribution. c-Jun is a primary JNK substrate. Therefore, we initially examined whether the status of c-Jun affects the redistribution of p19^ARF after exposure to UV. To that end, we irradiated p53^−/−c-jun^+/+ and p53^−/−c-jun^−/− mouse fibroblasts with 30 J/m² and examined p19^ARF localization 3 h later. As depicted in Fig. 3A, p19^ARF was also dispersed from the nucleoli of p53^−/−c-jun^−/− cells. However, the translocation of p19^ARF was more extensive in c-jun–expressing cells, as only 4% of the p53^−/−c-jun^+/+ cells were stained for nucleolar ARF, whereas nucleolar staining of ARF was observed in 49% of the p53^−/−c-jun^−/− ( Fig. 3A). Preincubation with the JNK inhibitor increased the fraction of cells stained for nucleolar p19^ARF by 64% and 27% in p53^−/−c-jun^+/+ and p53^−/−c-jun^−/− cells, respectively, suggesting that c-Jun is required for p19^ARF translocation, but an additional factor contributes as well. Thus, we examined other AP-1 proteins that may be involved in p19^ARF translocation. We knocked down JunB expression with specific small hairpin RNA (shRNA; ref. 18) in low-passage MEFs and examined the redistribution of p19^ARF after UV irradiation ( Fig. 3C–D). Indeed, repression of JunB expression also inhibited the translocation of p19^ARF from the nucleolus in MEFs ( Fig. 3D). Hence, we conclude that JunB cooperates with c-Jun to translocate p19^ARF. To test if the effect of c-Jun and JunB on p19^ARF translocation is additive, we knocked down junB in p53^−/−c-jun^−/− cells to generate c-jun, junB–deficient cells. junB shRNA significantly reduced the levels of JunB (Supplementary Fig. S2A). Irradiation of p53^−/−c-jun^−/− reduced the portion of cells expressing nucleolar p19^ARF to 14%, and inhibition of the JNK pathway prevented translocation and increased the fraction of cells expressing nucleolar p19^ARF to 84% (Supplementary Fig. S2B). In contrast, the translocation of p19^ARF in p53^−/−c-jun^−/− shjunB cells was significantly reduced, retaining nucleolar p19^ARF in 64% of the cells (Supplementary Fig. S2B). In these cells, the pretreatment with SP600125 only mildly increased the proportion of cells exhibiting nucleolar p19^ARF. This result suggests that the effects of c-Jun and JunB are additive.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

c-Jun and JunB are required for UV-induced p19^ARF translocation. A, p53^−/−c-jun^+/+ (striped bars) and p53^−/−c-jun^−/− (black bars) mouse fibroblasts were exposed to 30 J/m² UV, with or without preincubation with JNK inhibitor (SP). The cells were stained for p19^ARF 3 h after UV exposure, and the percentage of cells containing nucleolar p19^ARF at that time is plotted above. The fraction of cells containing nucleolar p19^ARF in the untreated control was considered as 100%. B, JunB protein levels were determined by immunoblotting in MEFs exposed to UV and then harvested at the indicated times (hours after treatment). Actin was used as a loading control. C, MEFs were infected with a retrovirus-expressing shRNA against junB or the vector virus alone, and selected. Six days after infection, the cells were exposed to 30 J/m² UV, and the level of JunB at 3 h after irradiation was determined by immunoblotting. Lamin A and B served as the loading controls. D, the MEFs whose JunB levels are exhibited in C were also stained for p19^ARF, and the percentage of cells containing nucleolar p19^ARF is plotted. Dark bars, cells expressing the vector; empty bars, cells expressing shRNA for junB.

Overexpression of Jun proteins translocates p19^ARF from the nucleolus. UV radiation activates multiple pathways and triggers intensive transcriptional responses. To examine the net effects of Jun proteins on the redistribution of p19^ARF, we transfected p53^−/−c-jun^−/− cells with Ha-tagged c-jun, junB, junD, or JDP2 and then localized p19^ARF in the transfected cells. As depicted in Fig. 4A , all Jun proteins dispersed p19^ARF from the nucleolus. However, significant differences in their potency were observed. c-Jun redistributed p19^ARF most efficiently, and only 33% of the transfected cells retained nucleolar p19^ARF. JunB was 1.5-fold less efficient and JunD was relatively inefficient, with p19^ARF retained in the nucleoli of 77% of the transfected cells ( Fig. 4A). Jun specificity in p19^ARF redistribution is supported by the fact that in addition to GFP, JDP2, another JNK target, did not redistribute p19^ARF when overexpressed under the same conditions ( Fig. 4A). The reduced nucleolar staining of p19^ARF is not due to reduction in its level, as transduction of c-Jun by viral expression in these cells did not repress p19^ARF expression (Supplementary Fig. S3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Redistribution of p19^ARF by Jun proteins. A, p53^−/−c-jun^−/− cells were transfected with expression vectors coding for HA-tagged c-Jun, JunB, JunD, or with expression vectors coding for GFP or JDP2. Transfected cells were identified by staining against the HA tag at 36 h after transfection, and the localization of p19^ARF in Jun-expressing cells was determined using coimmunofluorescent staining. The proportion of cells containing nucleolar p19^ARF after GFP transfection was regarded as 100%. Bands above the bars exhibit the level of each of the transfected proteins as determined by immunoblotting. B, c-jun ^+/+ fibroblasts were preincubated with JNK inhibitor or not, exposed to 30 J/m² UV, and harvested at the indicated times (h). The levels of phospho-Thr⁹¹ (p Thr 91) and phospho-Ser⁶³ (p Ser 63)as well as the total level of c-Jun were determined by immunoblotting. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as a loading control. C, p53^−/−c-jun^−/− cells were transfected with plasmids coding for HA-tagged wild-type (wt) and mutated c-Jun, as indicated (left). The cells were stained and subjected to microscopic determination of p19^ARF localization as in A. Levels of transfected c-Jun proteins exhibited above were determined by immunoblotting. Right, p19^ARF localization was determined in mouse fibroblasts expressing the indicated c-Jun before or 3 h after exposure to 30 J/m² UV. The fraction of cells expressing nucleolar ARF from the total cell population is presented. D, expression vectors for wild-type c-jun, c-jun mutant 272/273E, and for GFP were transfected to p53^−/−c-jun^−/− cells. Costaining for c-Jun and p19^ARF was used to determine the localization of the last-in Jun-transfected cells. The proportion of cells containing nucleolar p19^ARF after GFP transfection was regarded as 100%.

Two clusters in c-Jun are phosphorylated by JNK after exposure of cells to UV, Ser⁶³ and Ser⁷³, and Thr⁹¹ and Thr⁹³ ( Fig. 4B). We determined which JNK phosphorylation sites are required for c-Jun–dependent redistribution of p19^ARF by transfecting wild-type c-jun or mutants, in which JNK phosphorylation clusters were substituted to alanines (Ser^63/73Ala and Thr^91/93Ala) into p53^−/−c-jun^−/−, and subsequent examination of p19^ARF localization ( Fig. 4C, left). Interestingly, the canonical JNK phosphorylation sites, Ser⁶³ and Ser⁷³, were dispensable for p19^ARF redistribution as Ser^63/73Ala dispersed p19^ARF comparably to wild-type c-Jun. On the other hand, Thr⁹¹ and Thr⁹³ are required as the Thr^91/93Ala mutant is impaired in its ability to disperse p19^ARF. To avoid any possible transfection artifacts, we also examined the localization of p19^ARF in mouse fibroblasts in which Ser^63/73Ala or Thr^91/93Ala were substituted for the endogenous wild-type c-jun ( 42, 43). In agreement with the lack of ability of Thr^91/93Ala mutant to redistribute p19^ARF when overexpressed, p19^ARF was not significantly dispersed in the Thr^91/93Ala cells 3 h after exposure to UV, whereas in cells expressing wild-type c-Jun, a significant redistribution was observed ( Fig. 4C, right). The p19^ARF was redistributed less in Ser^63/73Ala cells probably due to the lower level of c-Jun induction in these cells shortly after UV exposure (Supplementary Fig. S4A; ref. 42). These results suggest that Thr⁹¹ and Thr⁹³ of c-Jun are important for redistribution of p19^ARF. c-Jun is a transcription factor that regulates the expression of numerous genes ( 32). To examine whether the transcriptional activity of c-Jun is required for p19^ARF redistribution, we generated a transcriptionally inactive c-Jun mutant by replacing two basic residues, Arg272 and Lys273, in its basic DNA-binding domain with two glutamic acids to generate a mutant incapable of binding DNA. Its ability to induce the transcription of AP-1–controlled reporter gene (3XTRE-Luc) was compared with that of wild-type c-Jun (Supplementary Fig. S4B). Unlike wild-type c-Jun, which enhances the transcription of the reporter gene in a dose-dependent manner and whose transcriptional capacity is augmented by coexpression of its upstream MAP3K, MEKK1, the mutant 272/273E did not activate transcription (Supplementary Fig. S4B). Nevertheless, its transfection into p53^−/−c-jun^−/− culminated in the dispersal of p19^ARF from the nucleolus, comparable with that of wild-type c-Jun ( Fig. 4D). This result suggests that the transcriptional activity of c-Jun is dispensable for the redistribution of p19^ARF. To further corroborate this notion, we examined the induction of the c-jun gene, which is subject to autoregulation by the endogenous Jun protein and can therefore serve as an indicator for its activity, in wild-type and Thr^91/93Ala-expressing cells after exposure to UV (Supplementary Fig. S4C). Again, the ability to regulate gene expression does not correlate with the ability to redistribute p19^ARF, as c-jun mRNA expression increased to the same extent in both cell lines 3 h after exposure to 30 J/m² UV (Supplementary Fig. S4C), whereas p19^ARF was dispersed only in wild-type c-jun–expressing cells ( Fig. 4C).

c-Jun interacts with B23 in a JNK-independent manner and changes its subnuclear localization in a JNK-dependent manner. As the transcriptional activity of c-Jun is dispensable for redistribution of p19^ARF, we tested whether its interactions with other proteins contribute to the ability of c-Jun to disperse p19^ARF from the nucleolus. We explored two possible interactions: interactions between c-Jun and p19^ARF and interactions of c-Jun with B23. Coimmunoprecipitation experiments were performed to determine the validity of the hypothetical interactions. We directly examined interactions of endogenous, not overexpressed, proteins in p53^−/− fibroblasts that were exposed to UV (or left untreated) and harvested 3 h later. In addition, we preincubated the same cells with SP600125 to determine the dependency of the interactions on JNK activity. In these experiments, immunoprecipitates of c-Jun did not contain p19^ARF (data not shown). However, precipitated c-Jun coprecipitated B23 reproducibly ( Fig. 5A ). The weak interactions observed between c-Jun and B23 in untreated cells were strengthened in UV-irradiated cells. However, these interactions do not correlate with c-Jun phosphorylation by JNK, as preincubation with SP600125, which significantly reduced c-Jun phosphorylation after UV exposure, actually increased the association of c-Jun with B23 (compare UV − SP600125 to UV + SP600125). These interactions are independent of p19^ARF expression as they were also observed in extracts from p19^ARF-deficient cells (Supplementary Fig. S5). These results suggest that c-Jun interacts with B23 and that this interaction is not dependent on JNK phosphorylation. Indirect interactions between c-Jun and p19^ARF were observed in NIH3T3 cells, containing metallothionein promoter–regulated p19^ARF 24 h after induction with ZN²⁺. Immunoprecipitation of p19^ARF revealed the presence of B23 as well as low levels of c-Jun in the precipitate, thus, suggesting the presence of a complex containing p19^ARF/B23 and c-Jun ( Fig. 5B, left). The hypothesis that c-Jun directly displaces p19^ARF from the complex with B23 was directly tested by transfecting HEK293 cells with expression vectors for B23, p19^ARF, and c-Jun and examining the amounts of p19^ARF that coprecipitated with B23 in the presence of increasing amounts of c-Jun. As depicted in Fig. 5B (right), B23 was precipitated even in nontransfected cells, and comparable high levels were precipitated by the antibody after transfection. Increasing the amount of transfected c-Jun resulted in an increase in the association between B23 and c-Jun ( Fig. 5B, right). Surprisingly, p19^ARF interactions with B23 increased rather than decreased, in parallel to the increase in B23 to c-Jun interactions ( Fig. 5B, right). These results show that c-Jun does not prevent p19^ARF from interacting with B23. Therefore, we examined the alternative possibility that JNK might control the subnuclear localization of c-Jun and, thus, the localization of interacting B23 as well. It should be mentioned that the localization of several AP-1 proteins was previously found to be regulated by UV ( 46). We, therefore, stained untreated and UV-irradiated JNK^+/+ and JNK1,2^−/− mouse fibroblasts for c-Jun to detect JNK-dependent changes in its subnuclear localization after UV exposure ( Fig. 5C). In JNK^+/+ mouse fibroblasts, c-Jun is distributed throughout the nucleus, including in the nucleolus where B23 is concentrated. After exposure to UV, however, c-Jun is excluded from certain areas suspected to be the nucleoli ( Fig. 5C), in correlation with B23 dispersal. To determine whether the unstained areas are indeed the nucleoli, we transfected the JNK^+/+ mouse fibroblasts with an expression vector for GFP-tagged fibrillarin, which is often used as a nucleolar marker, and detected its expression relative to c-Jun in UV-irradiated cells ( Fig. 5C, bottom). This analysis confirmed that the areas from which c-Jun is excluded are indeed the nucleoli. c-Jun staining is also observed in the entire nuclear area of untreated JNK1,2^−/− cells. Moreover, regions of intense c-Jun staining that coreside with the nucleolar B23 were also observed ( Fig. 5C, arrows). However, unlike in JNK-expressing cells, c-Jun was not excluded from the nucleolus of irradiated JNK-deficient cells. Quantitative analysis revealed a good correlation between c-Jun exclusion and the export of B23 from the nucleoli ( Fig. 5D). These results suggest that JNK-dependent translocation of c-Jun is correlated with the exclusion of B23 from the nucleolus. Additional experiments showed that B23 is redistributed after UV exposure also in cells expressing Ser^63/73Ala c-Jun. However, it is not dispersed in the Thr^91/93Ala cells (Supplementary Fig. S6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

c-Jun interacts with B23 in a JNK-independent manner and is excluded from the nucleolus in a JNK-dependent manner. A, p53^−/− fibroblasts were either exposed to 30 J/m² UV or not, in the presence or absence of JNK inhibitor SP600125. Endogenous c-Jun was immunoprecipitated from cells harvested 3 h after irradiation, and the levels of immunoprecipitated c-Jun and coprecipitated B23 were detected by immunoblotting using specific antibodies. The right part of the panel shows the level, Ser⁶³ phosphorylated c-Jun, and B23 in the extracts subjected to immunoprecipitation. NC, negative control immunoprecipitate from UV-exposed cells that were incubated with irrelevant antibody. B, p19^ARF was precipitated from extracts of control or UV-irradiated NIH3T3 containing metallothionein promoter–regulated p19^ARF, 24 h after induction with ZN²⁺, and levels of coprecipitated endogenous, B23 and c-Jun were determined using specific antibodies (left). Right, HEK293 cells were cotransfected with constant levels of B23 and p19^ARF (2 and 4 μg, respectively) and the indicated amounts of c-Jun (in μg). Twenty-four hours after transfection, the cells were harvested, and B23 was immunoprecipitated with a specific antibody. The levels of precipitated B23, c-Jun, and p19^ARF were determined by immunoblotting with a specific antibody. C, JNK^+/+ and JNK1,2^−/− mouse fibroblasts were exposed to UV and stained for B23 and c-Jun 3 h after irradiation. Bottom, JNK^+/+ cells were transfected with an expression vector for GFP-fibrillarin. The cells were irradiated at 24 h after transfection and then stained for c-Jun at 3 h after irradiation, using a specific antibody. D, quantitative analysis of nucleolar c-Jun (left graph) and B23 (right graph) before and 3 h after UV exposure. Black bars, staining in JNK^+/+ cells; white bars, staining in JNK1,2^−/− cells.

p19^ARF is not dispersed from the nucleoli of senescent cells. ARF levels increase during progressive passages of explanted MEFs in culture. JNK activity, on the other hand, is reduced in senesced fibroblasts, resulting in defective stress responses in these cells ( 47). To show the importance of the JNK pathway for p19^ARF redistribution in a physiologic system in which JNK activity is reduced without the application of experimental manipulation, we simultaneously examined JNK activity and the location of p19^ARF in early and late passage MEFs before and after exposure to UV radiation. As depicted in Fig. 6A , both basal and UV-induced JNK-dependent c-Jun and phospho–c-Jun levels are higher in early passage MEFs. Concordantly, in early passage MEFs, p19^ARF was efficiently translocated from the nucleoli after UV exposure. In contrast, and in agreement with the hypothesized effects of low JNK activity, little dispersal of p19^ARF from the nucleoli of late passage MEFs was observed after UV exposure ( Fig. 6B). This result shows that low JNK activity can result in the prevention of p19^ARF redistribution after UV exposure.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

UV does not affect p19^ARF localization in senesced cells. A, levels of c-Jun and Ser⁶³-phosphorylated c-Jun were examined in passage 2 and passage 6 MEFs by immunoblotting with specific antibodies. Actin was used as a loading control. B, early passage (P2) and late passage (P6) MEFs were irradiated with UV and the localization of p19^ARF was detected by fluorescent immunostaining. C and D, a model describing the activity of the JNK pathway in stress-induced translocation of B23 and ARF. C, cells not exposed to UV. D, UV-irradiated cells.

Discussion

The nucleolus functions in the stress response ( 1, 2, 4) and redistribution of B23 and p19^ARF may relay this response ( 3, 4). In this study, we define for the first time a pathway and a mechanism that underlies B23 and p19^ARF redistribution. Both are redistributed after the exposure of cells to DNA damage, but this redistribution is selective and occurs efficiently when the JNK pathway is activated. UV radiation, the most potent JNK activator, triggers the redistribution of p19^ARF and B23 most effectively, whereas exposure to ionizing radiation, which is a poor activator of JNK, does not result in efficient p19^ARF redistribution. Furthermore, inhibition of JNK activity by the synthetic inhibitor SP600125, or in cells deficient in JNK, efficiently prevented the redistribution of p19^ARF and B23, thus, proving that the DNA damage-dependent redistribution is JNK-dependent. However, our data suggest that although the JNK pathway is a major regulator of DNA damage–dependent localization of p19^ARF and B23, it is not the only one. We show that impairment of JNK activity prevents only 60% to 80% of the UV-dependent redistribution p19^ARF ( Figs. 2– 3). In addition, at later times after UV-irradiation, p19^ARF may exit the nucleolus in a JNK-independent manner (data not shown). Hence, we suggest that the JNK-dependent redistribution is the early response to the stress.

Out of the plethora of JNK substrates, the Jun proteins regulate the JNK-dependent B23 and p19^ARF translocations. p19^ARF redistribution in the absence of c-Jun or JunB is reduced, demonstrating that both are required for the redistribution. However, reduced expression of each of these individual proteins partially prevented the redistribution ( Fig. 3A and D), and combined inhibition of both prevents the redistribution after exposure to UV almost as efficiently as inhibition of JNK activity, suggesting that most of the effect is mediated by these proteins (Supplementary Fig. S2). Inhibition of the JNK-dependent activity of JunD may also be required for a complete inhibition of p19^ARF redistribution. This notion is supported by two experimental findings: transient expression of JunD into mouse fibroblasts resulted in moderate redistribution of the endogenous p19^ARF ( Fig. 4A). In addition, a previous study reported an intense nucleolar staining of p19^ARF in junD^−/− MEFs ( 48). This was intuitively attributed to the ability of JunD to down-regulate p19^ARF expression ( 48). Alternatively, the strong inhibition of translocation when JNK activity is impaired may also stem from direct JNK phosphorylation of either p19^ARF or B23 as a precedence for the involvement of JNK in the trafficking of other nucleolar proteins was described in other systems ( 49). This notion, however, is not supported by the fact that in cells expressing c-Jun mutated at the JNK phosphorylation sites Thr^91/93 no translocation is observed after UV exposure. We also showed the requirement of the JNK pathway for the distribution of p19^ARF in a physiologically relevant system of senesced cells, in which JNK activity and c-Jun levels are low, and p19^ARF was not translocated after UV treatment ( Fig. 6A and B).

To study the mechanism of the redistribution, we focused our attention on the c-Jun protein, due to the presence of extensive data and reagents related to its modulation by JNK. Interestingly, our results show for the first time a specific role for the phosphorylation at Thr^91/93. These sites are phosphorylated by UV, with a kinetic similar to that of the phosphorylation of Ser^63/73 ( Fig. 4B). However, as of today, no biological activity has been assigned to this phosphorylation. Using both cell lines expressing knock-in c-Jun mutants and transient overexpression experiments, we show that the presence of Thr^91/93 is important for the redistribution, whereas the activity of Ser^63/73 is dispensable. We also show that c-Jun–dependent transcriptional regulation of target genes is not involved in the redistribution of p19^ARF. DNA-binding and transcriptionally inactive mutant c-Jun (272/273E) can efficiently redistribute p19^ARF ( Fig. 4D). Instead, we identified a more direct mechanism: c-Jun interacts with B23. Surprisingly, nonphosphorylated c-Jun interacts with B23 better than phosphorylated c-Jun ( Fig. 5A). Therefore, our results suggest that the interactions between B23 and c-Jun are JNK-independent. Hence, the activity of JNK may be important for augmenting the reservoir of c-Jun ( 32, 34, 50). The interaction between B23 and c-Jun is p19^ARF-independent (Supplementary Fig. S5) and does not enhance the dissociation between B23 and p19^ARF ( Fig. 5B). On the other hand, we show that the translocation of Jun and B23 out of the nucleolus is JNK dependent. Therefore, we suggest that, just as for the effects on transcription initiation factor-IA ( 49), JNK is involved in the shuttling of the c-Jun–associated B23. In contrast to the situation in JNK-expressing cells, c-Jun is not excluded from the nucleoli of JNK-deficient cells. This is in perfect correlation with B23 translocation ( Fig. 5C–D). The possibility that the nucleolus could be totally disrupted after irradiation is not valid as fibrillarin, often used as nucleolar marker, is concentrated at nucleolar-shaped structures from which c-Jun is excluded after UV exposure. In addition, when p19^ARF was ectopically overexpressed to high levels, UV irradiation caused only partial redistribution of p19^ARF (data not shown), thus, suggesting that p19^ARF redistribution is a stoichiometric process. In this respect, the involvement of at least two UV-inducible Jun proteins can provide a reasonable stoichiometric relationship to explain the B23 translocation. A recent study by Liu et al. ( 46) offered a possible mechanism by which c-Jun localization might be changed by UV radiation; Jun homodimers are concentrated in nucleolar-like structures; and upon exposure to UV, c-Jun interacts with activating transcription factor 2 and changes its subnuclear position.

A recent model suggested that upon UV exposure, p14^ARF dissociates from B23 and interacts with Hdm2 ( 3). B23 itself is also translocated to the nucleoplasm after UV exposure by a mechanism obscure before this study ( 4, 5). Whether B23-associated p19^ARF is translocated along with c-Jun and the dissociation occurs at later stage is currently studied. Our model suggests that the induction of the AP-1 pathway is the initial and essential step. We suggest that the JNK activity elevates c-Jun levels, thus, increasing the association between c-Jun and B23, and that the phosphorylation of Thr^91/93 by JNK enhances translocation of c-Jun and the associated B23 out of the nucleolus. As B23 anchors p19^ARF to the nucleolus, its translocation will inevitably result in the exit of p19^ARF from the nucleolus as illustrated in our model ( Fig. 6C and D).