Adenovirus E1B 19,000-Molecular-Weight Protein Activates c-Jun N-Terminal Kinase and c-Jun-Mediated Transcription

Abstract

Adenovirus E1B proteins (19,000-molecular-weight [19K] and 55K proteins) inhibit apoptosis and cooperate with adenovirus E1A to induce full oncogenic transformation of primary cells. The E1B 19K protein has previously been shown to be capable of activating transcription; however, the underlying mechanisms are unclear. Here, we show that adenovirus infection activates the c-Jun N-terminal kinase (JNK) and that the E1B gene products are necessary for adenovirus to activate JNK. In transfection assays, we show that the E1B 19K protein is sufficient to activate JNK and can strongly induce c-Jun-dependent transcription. Mapping studies show that the C-terminal portion of E1B 19K is necessary for induction of c-Jun-mediated transcription. Using dominant-negative mutants of several kinases upstream of JNK, we show that MEKK1 and MKK4, but not Ras, are involved in the induction of JNK activity by adenovirus infection. The same dominant-negative kinase mutants also block the ability of E1B 19K to induce c-Jun-mediated transcription. Taken together, these results suggest that E1B 19K may utilize the MEKK1-MKK4-JNK signaling pathway to activate c-Jun-dependent transcription and demonstrate a novel, kinase-activating activity of E1B 19K that may underlie its ability to regulate transcription.

The adenovirus (Ad) E1B 19,000-molecular-weight (19K) protein functions to protect cells from apoptosis during an Ad lytic infection (8, 47). This antiapoptotic activity of the E1B 19K protein is also essential for its cooperation with Ad E1A proteins to induce full oncogenic transformation of primary rodent cells (8, 47). The E1B 19K oncoprotein is functionally homologous to the mammalian Bcl-2, both of which can act to suppress apoptosis (7, 47, 64, 66). Both Bcl-2 and the E1B 19K protein have been shown to interact with a common set of cellular proteins, including the apoptotic inducer Bax (21), Bak (15), Bik/Nbk (4, 22), and Nip 1, 2, and 3 (5), and these interactions are believed to be important for the biological functions of both proteins. Another important activity of the E1B 19K protein is its ability to regulate transcription. The E1B 19K protein can specifically block tumor suppressor p53-mediated transcriptional repression, and this ability of E1B 19K in part underlies its ability to inhibit p53-induced apoptosis (49, 55). E1B 19K can also activate enhancer-dependent transcription (74). In contrast, E1A represses viral and cellular enhancers, and this inhibitory effect is achieved at least in part by targeting the transcriptional cofactor p300 (14). The transcriptional antagonism between E1A and E1B 19K may be important for the maintenance of a balanced transcriptional program that is critical for a successful adenoviral infection and for the E1A- and E1B-induced oncogenic transformation. Although it has been known for some time that E1B 19K can function as a transcriptional regulator, the mechanisms by which it regulates transcription are largely unknown. In this report, we show that E1B 19K can strongly activate transcription mediated by c-Jun, and we investigate the mechanism by which E1B 19K induces c-Jun-dependent transcription.

c-Jun belongs to a family of related proteins that includes JunB and JunD. These proteins can heterodimerize with the Fos family of transcription factors to form what is known as the AP-1 transcription factor (1). Previous work has shown that c-Jun-dependent transcription is stimulated by c-Jun N-terminal kinase (JNK), which is also known as stress-activated protein kinase (SAPK) (12, 33). JNK was first characterized as a mitogen-activated protein kinase (MAPK) family member that binds c-Jun and phosphorylates serines 63 and 73 located within the transactivation domain of c-Jun (12, 33). Phosphorylation of these serine sites results in increased transactivation potential of c-Jun (12, 24). The substrates of JNK have now been extended to include other transcription factors, such as ATF-2 (19) and Elk-1 (6). The activity of JNK is regulated by dual phosphorylation on specific threonine and tyrosine residues by MKK4 (otherwise known as SEK1) (13, 40). MKK4, in turn, is activated by the upstream protein kinase MEKK1 (13, 35, 70). Growth factors such as epidermal growth factor (EGF) and activated Ras that lie further upstream of this signaling pathway appear to stimulate MEKK1 and subsequently downstream JNK through GTP-binding proteins such as Cdc42 and Rac1 (10, 43) and Rho (61).

In this study, we report that Ad infection of human cells results in the activation of JNK during the later stages of infection. Analysis of Ad early-region mutants suggests that E1B proteins are required for JNK activation. Subsequent transfection assays demonstrate that E1B 19K is sufficient to induce JNK activation. This JNK activation is accompanied by a strong induction of c-Jun-dependent transcription. Interestingly, full induction of c-Jun-dependent transcription by E1B 19K is only partially dependent on c-Jun phosphorylation (Ser63 and Ser73) by the activated JNK. We show that the transcriptional cofactor p300 can synergize with E1B 19K to activate c-Jun-mediated transcription, suggesting that p300 may play a role in the full induction of c-Jun-mediated transcription by E1B. To investigate the biochemical mechanisms by which E1B 19K activates c-Jun-dependent transcription via activated JNK, we analyzed the effect of known upstream JNK-activating kinases. Our results show that activation of c-Jun-dependent transcription by E1B 19K involves MEKK1 and MKK4, two JNK upstream activating kinases. However, Ras is not involved, suggesting that E1B 19K activates JNK in a manner that is different from that of growth factors. The kinase dependence of E1B 19K in activating c-Jun-dependent transcription is in contrast to the Ad E1A proteins which repress c-Jun-mediated transcription by physically interacting with the coactivator p300. Taken together, our results identify a novel biochemical activity of E1B 19K, i.e., its ability to activate JNK, which may represent a potential molecular mechanism that underlies the transcriptional activation activity of E1B 19K.

MATERIALS AND METHODS

Cell culture and viruses.

HeLa or C33A cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% calf serum (HeLa cells) or fetal calf serum (C33A cells). Wild-type Ad type 5 (wt300) (29) or deletion Ad mutants dl312 (29), dl339 (41), dl802 (48), dl327 (62), and dl366 (20), have been previously described, as have the methods to propagate these viruses. The titers of these viruses were determined by plaque assays as previously described (29). For infection with wild-type virus or the various mutants, cells grown on 10-cm plates were passaged 48 h prior to infection to achieve 60% confluence at the time of infection. The cells were washed three times with phosphate-buffered saline and incubated with virus (multiplicity of infection [MOI] of 25) in serum-free DMEM for 60 to 90 min at 37°C with occasional rocking. The virus suspension was then removed, and the infected cells were incubated in normal growth medium at 37°C. The cells were then harvested at the indicated time periods.

Plasmids and antibodies.

pSRαhemagglutinin (HA) epitope-tagged JNK1, HA-tagged JNK1 (APF), pSRαRas(N17), glutathione S-transferase (GST)–c-Jun (amino acids [aa] 1 to 79) and GST–c-Jun (aa 1 to 79) Ala63 and 73 were kindly provided by T. Deng from the University of Florida, Gainesville, Fla. (12, 43). GAL4–c-Jun (aa 1 to 223), GAL4–c-Jun (aa 1 to 223) Ala63 and 73, GAL4–c-Jun (aa 43 to 223), GAL4–c-Jun (aa 56 to 223), GAL4–Elk-C (aa 307 to 428 of Elk-1), GAL4–v-Jun, and MEKK1 K432M expression vectors (6, 24, 40, 44) were gifts from M. Karin, University of California, San Diego, Calif. pCDNA3MKK4 (Ala) and Rc/RSVMKK3 (Ala) were provided by R. Davis, University of Massachusetts, Worcester, Mass. (46, 67). pGAL4-E1BCAT and expression vector for 13S E1A have been described previously (36, 57). pCMVE1B 19K, pCMVE1B 19K PM 51, pCMVE1B 19K PM87, pCMVE1B 19K PM102, and pCMVE1B 19K PM28 were kind gifts from Eileen White (Rutgers University) and have been described elsewhere (8). Rc/CMV E1B 19K, Rc/CMV E1B 19K dl aa 90 to 96, Rc/CMV E1B 19K 123,124 WR-AS, Rc/CMV E1B 19K 75,76 EK-AS, and Rc/E1B 19K (aa 1 to 146) have been described previously (60) and were kindly provided by G. Chinnadurai (St. Louis University Health Sciences Center). Rc/CMV E1B 19K (aa 1 to 40) and Rc/CMV E1B 19K (aa 1 to 88) were provided by M. L. Schmitz (German Cancer Research Center, Heidelberg, Germany) (39). GAL4-Sp1Q2 was obtained from R. Tjian (University of California, Berkeley, Calif.). GAL4-JunB (aa 1 to 259) and GAL4-JunD have been previously described (37). Cytomegalovirus β (CMVβ HA-p300) was provided by D. Livingston (Dana-Farber Cancer Institute, Boston, Mass.) (14). Monoclonal antibody to E1A (M73) was provided by E. Harlow (Massachusetts General Hospital Cancer Center). Rabbit polyclonal antibody to E1B 19K was provided by Maurice Green (St. Louis University School of Medicine, St. Louis, Mo.). Rabbit anti-human JNK1 was obtained from Santa Cruz Biotechnology, Inc., and mouse anti-HA monoclonal antibody, 12CA5, was a kind gift from Tom Kirchhausen (Center for Blood Research, Harvard Medical School).

Transfections and CAT assays.

Cells were transfected by the calcium phosphate precipitation method as previously described (57). The total amount of DNA was normalized with plasmid PSP72 for each transfection. Cells were harvested 48 h after addition of the precipitate. For virus infections, the cells were infected at an MOI of 25 with Ad 24 h before harvesting. Transfections were carried out with at least two independent DNA preparations and were repeated at least three times. For chloramphenicol acetyltransferase (CAT) assays, whole-cell extracts were prepared as described elsewhere (57) and were quantitated with a Beckman LS6500 scintillation counter. The amount of cell extract used was such that the CAT activity was within the linear range.

Protein kinase assays.

Immunocomplex kinase assays were performed as essentially described elsewhere (34, 45). Briefly, cells were lysed in 800 μl of kinase lysis buffer (20 mM HEPES [pH 7.4], 2 mM EGTA, 50 mM β-glycerophosphate, 1 mM sodium orthovanadate, 1% Triton X-100, 10% glycerol, 1 mM dithiothreitol, 2 mM phenylmethysulfonyl fluoride, and 10 μg each of aprotonin, pepstatin, and leupeptin) for 15 min on ice. Insoluble debris was removed by microcentrifugation at 14,000 × g for 15 min. The supernatant was removed and standardized according to protein concentration as determined by a Bradford assay kit (Bio-Rad). Endogenous JNK1 activity was determined by immunocomplex kinase assays after immunoprecipitation of JNK1 with rabbit anti-JNK1 (Santa Cruz Biotechnology, Inc.) for 3 h at 4°C. HA-JNK1 activity were assayed by immunoprecipitating the kinases with a mouse anti-HA monoclonal antibody (12CA5). One microgram of GST–c-Jun (aa 1 to 79) or GST–c-Jun (aa 1 to 79) Ala63 and 73 was used as a substrate. Quantitation of kinase assays were performed by densitometric analysis with NIH Image 1.59 software.

Western blotting.

Proteins from 500-μg whole-cell lysates were resolved by electrophoresis through a 12% polyacrylamide gel, transferred to nitrocellulose, and probed with the antibodies described above. The proteins were visualized as previously described (28).

RESULTS

Ad activates JNK activity.

To determine whether Ad can induce JNK activity, HeLa cells were infected with wild-type Ad and immunocomplex kinase assays were performed at various time points postinfection. As shown in Fig. 1A, a 17-fold increase in endogenous JNK1 activity was observed at 14 h postinfection, and this activity continued to increase by as much as 60-fold at 48 h postinfection. Mock-infected HeLa cells did not show any elevation of JNK1 activity at any time points (Fig. 1A, lane 1, and data not shown). The level of endogenous JNK1 kinase activity at 48 h postinfection was close to that observed in cells that were irradiated with 40-J/m² UV light (Fig. 1A, lane 7). Immunoblot analysis of the lysates with an anti-JNK1 polyclonal antibody shows that JNK1 protein levels remain roughly constant in Ad-infected cells (Fig. 1B), indicating that the observed increase in JNK1 activity is not due to a change in the protein level. These results demonstrate that Ad activates JNK1 in HeLa cells at late stages of infection.

FIG. 1.

(A) Effect of Ad infection of HeLa cells on endogenous JNK1 activity. HeLa cells were infected with Ad as described in Materials and Methods. Cells were lysed in kinase buffer, JNK1 immunocomplex kinase assays were performed with GST–c-Jun (aa 1 to 79) as the substrate, and densitometric analysis was performed. Fold activation values of endogenous JNK activity as measured by GST–c-Jun phosphorylation induced at various times post-Ad infection and by UV irradiation are indicated at the bottom. The band denoted by an asterisk represents a proteolytic fragment of GST–c-Jun aa 1 to 79. Results are representative of three experiments. (B) Analysis of JNK1 endogenous levels from lysates of Ad-infected cells by immunoblotting.

Since HeLa cells contain the human papillomavirus type 18 early gene products E6 and E7 (54), we asked whether Ad can activate JNK1 in another human cell line, C33A cells, that lacks HPV E6 and E7 (71). As shown in Fig. 2A, Ad infection strongly activates endogenous JNK1 in both HeLa and C33A cells, indicating that HPV E6 and E7 are not involved in the observed JNK activation. The expression levels of endogenous JNK1 remain unchanged in cells infected with Ad (data not shown), suggesting that JNK1 activation is not due to an increase in the protein level. In addition, as shown in Fig. 2B (lanes 1 and 2), Ad infection strongly activates the HA epitope-tagged JNK1 (HA-JNK1) that is transfected into HeLa cells. We further show that a catalytically inactive form of JNK1, JNK1 APF, in which the dual phosphorylation sites Thr183 and Tyr185 have been mutated to alanine and phenylalanine (12), cannot be activated by Ad infection, as shown by its inability to phosphorylate GST–c-Jun (aa 1 to 79) (Fig. 2B, lane 4). This result suggests that Ad is likely to activate an upstream kinase (MKK4, for example) that phosphorylates and activates JNK at these sites. Additionally, the HA-JNK1 immunocomplex isolated from Ad-infected HeLa cells failed to phosphorylate the GST–c-Jun Ala 63 and 73 mutant (Fig. 2B, lane 5), suggesting that Ser63 and Ser73 of c-Jun are the target residues for JNK1 that is activated by Ad. Taken together, the above data indicate that Ad activates JNK indirectly by activating kinases upstream of JNK.

FIG. 2.

(A) Effect of Ad infection on JNK1 activity in HeLa and C33A cells. Immunocomplex kinase assays were performed on lysates of cells infected for 24 h with Ad. The fold increase in endogenous JNK activity above the basal level in mock-infected cells was determined by densitometry. Wt, wild type. (B) Effect of Ad infection of HeLa cells on activation of wild-type (Wt) and mutant JNK1 (APF). HeLa cells were transfected with either HA-tagged wild-type JNK1 or mutated JNK1 (APF) (1 μg each) prior to infection with Ad for 24 h. Cells lysates were then made, and immunocomplex kinase assays were performed with HA epitope-tagged antibodies. Lanes 1 to 4, results with GST–c-Jun aa 1 to 79 as a substrate; lane 5, immunocomplex kinase assay with a JNK phosphorylation-defective GST–c-Jun (GST–c-Jun [aa 1 to 79] Ala 63 and 73) as a substrate.

Activation of JNK by Ad occurs through stimulation of upstream JNK kinases.

A cascade of events involving activation of multiple kinases led to JNK activation. The JNK activating kinases include MKK4/JNKK/SEK1 (13, 40) and MEKK1 (44, 51), as well as further upstream GTPases, such as Ras (Ha-Ras) (12, 43) and the rho subfamily of GTPases (Rac1 and Cdc42) (10, 43). We investigated whether Ad-induced JNK activation was a direct effect on JNK or whether activation of kinases upstream of JNK were involved. We tested various dominant-negative mutants of JNK upstream kinases, including a dominant-negative mutant of Ras, for their abilities to block Ad-induced JNK activation. MKK4 is a protein kinase that phosphorylates and activates both JNK and p38 MAPK (13, 46, 51), while MKK3 is a specific activator of the p38 MAPK (46). MKK4 (Ala) and MKK3 (Ala) are the dominant-negative mutants of MKK4 and MKK3, respectively, in which the activating dual phosphorylation sites have been mutated to alanine) (46, 67). As shown in Fig. 3A, compared with the pcDNA3 vector control, cotransfection of MKK4 (Ala) and MKK3 (Ala) reduced the ability of Ad to activate JNK1 from 34-fold activation to 5- and 13-fold, respectively (compare lanes 4 and 6 with lane 2). The inhibition of Ad-induced JNK activation by MKK4 (Ala) suggests that MKK4 is one of the downstream kinase targets for Ad and lends further support that JNK stimulation by Ad is not direct and occurs by activation of an upstream kinase. The inhibition of Ad-induced JNK activation by MKK3 (Ala) may be due to its ability to sequester common upstream MAPK kinase kinases that activate both JNK and p38 signaling pathways (46). Consistent with this idea, MEKK1, the upstream activator of MKK4 (44, 51), appears to be involved in Ad-induced JNK activation, since the dominant-negative mutant of MEKK1, i.e., MEKK1 K432M (lysine 432 converted to methionine) (40, 44), significantly inhibits JNK activation in Ad-infected cells (Fig. 3B, compare lanes 2 with 4 [32-fold versus 2-fold activation, respectively]). In contrast, the GTP-binding protein Ras does not appear to be involved in Ad-induced activation of JNK since the dominant-negative mutant (Ras N17) virtually had no inhibitory effect on JNK1 activity in Ad-infected cells (Fig. 3B, compare lanes 2 and 3). However, the Ras N17 mutant efficiently inhibited EGF-induced JNK enzymatic activity in HeLa cells (data not shown). These results suggest that Ad activates JNK1 through a MEKK1-MKK4 pathway. In this regard, it is interesting to note that Ras functions as an upstream activator of JNK in response to growth factors (44) but does not appear to be involved in mediating Ad-induced JNK activation. This suggests that Ad and EGF induce the MEKK1-MKK4-JNK signaling cascade by different mechanisms.

FIG. 3.

Effect of dominant-negative mutants on JNK1 activation by Ad infection. (A) HeLa cells were transfected with HA-JNK1 (1 μg) with or without dominant-negative expression plasmids for MKK4 and MKK3 (10 μg each). PCNDA3 represents empty vector control. Twenty-four hours after transfection, the cells were mock infected or infected with wild-type Ad for 24 h. Immunocomplex kinase assays were then performed. The fold increase in HA-JNK1 activity above the basal level in mock-infected cells was determined by densitometry. (B) HeLa cells were transfected with dominant-negative Ras (10 μg) or dominant-negative MEKK1 (MEKK1 K432M [2 μg]) along with HA-tagged JNK1 (1 μg) prior to infection with wild-type Ad. Immunocomplex kinase assays were performed as described for panel A. Results represent three experiments.

E1B proteins are necessary for JNK1 activation by Ad.

To identify the viral proteins that are responsible for JNK activation during Ad infection, we analyzed Ad mutants that carry deletions of individual early regulatory genes. As shown in Fig. 4A, mutations of E2, E3, and E4 regions did not affect the ability of these Ad mutants to activate JNK1 (lanes 5, 6, and 7). However, deletion of E1A (dl312) or E1B (dl339) severely impaired the ability of the mutant viruses to activate JNK1 (Fig. 4A, compare lanes 3 and 4 with lane 2 [64-fold activation for the wild type, but 1.3- and 3-fold for dl312 and dl339, respectively]). The lack of JNK activation in these cells was not due to changes in the JNK protein level, as shown by Western blot analysis (Fig. 4B). This suggests that the Ad E1 region is necessary for activating JNK1.

FIG. 4.

(A) Effect of deletion of Ad early regulatory proteins on the ability to activate JNK1. HeLa cells were infected with wild-type or various Ad deletion mutants for 24 h before lysis of cells and determination of endogenous JNK1 activity by immunocomplex kinase assays. The fold increase in endogenous JNK activity above the basal level in mock-infected cells was determined by densitometry. Results represent three experiments. (B) Analysis of endogenous JNK1 levels by Western blot analysis.

This E1 region encodes both E1A (12S and 13S proteins) and E1B (55K and 19K) proteins (for reviews, see references 16, 56, and 68). E1A is necessary for the activation of other viral genes, including that of E1B during infection, while one of the main functions of E1B proteins is inhibition of apoptosis (8, 47). To determine which of these two viral gene products might be responsible for JNK activation, we first analyzed the expression of E1A and E1B proteins during Ad infection. Figure 5A shows that Ad infection of HeLa cells results in activation of endogenous JNK during later time points in infection and that this activation correlates with the expression of E1B 19K. As shown in Fig. 5B, under our assay conditions, expression of E1A is evident as early as 8 h postinfection and peaks at 24 h postinfection. However, by 48 h postinfection, the level of E1A proteins dramatically declined (Fig. 5B, lane 6), while endogenous JNK1 activity was maximally induced (Fig. 5A, lane 6). In contrast, expression of E1B 19K became evident at 12 h postinfection (Fig. 5C, lane 4) and continued to increase for as much as 48 h (Fig. 5C, lane 6), which is concomitant with the maximal endogenous JNK1 activity seen at that time point (Fig. 5A, lane 6). At time points when JNK1 enzymatic activity was high, JNK1 protein levels remained relatively constant (Fig. 5D), indicating that the increase in JNK1 activity was not due to an increase in levels of the protein. These results suggest that the E1B proteins are the likely candidates that induce JNK activation during Ad infection.

FIG. 5.

Time course of expression of viral early region 1 (E1) proteins. (A) HeLa cells were infected with wild-type Ad for various time points, and JNK1 immunocomplex kinase assays were performed. The fold increase in endogenous JNK activity above the basal level in mock-infected cells was determined by densitometry. Western blot analyses of the same lysates were performed with the indicated antibodies, including anti-E1A monoclonal antibody M73 (B), rabbit anti-E1B 19K peptide antibody (C), and rabbit anti-JNK1 antibody (D).

To further define the role of E1A and E1B in JNK1 activation, we analyzed the ability of these proteins to activate JNK in transient transfection assays. As shown in Fig. 6, transfection of E1B 19K together with HA-JNK1 resulted in an increase in JNK1 activity comparable to that of Ad-infected HeLa cells (Fig. 6, compare lane 4 with 2). In contrast, cotransfection of 13S E1A had no effect on JNK1 activity (Fig. 6, lane 5). Taken together, these results suggest that E1B 19K, but not E1A, is directly responsible for the activation of JNK1 during viral infection. Since E1A is essential for the transcriptional activation of E1B, the fact that the E1A-negative virus fails to activate JNK1 is likely due to the inability of this mutant virus to synthesize sufficient amounts of E1B. The ability of E1B 19K to activate JNK is not restricted to the JNK1 isoform. Transient transfection experiments show that E1B 19K also activates HA-JNK2 (data not shown). Therefore, our results show that both JNK isoforms can be activated by E1B 19K. Lastly, we found that the E1B 55K protein can also activate JNK in transfection experiments but that maximal activation of JNK by E1B 55K may require additional adenoviral gene products (data not shown).

FIG. 6.

Activation of HA-JNK1 by transfected E1B 19K. HeLa cells were transfected with either CMV vector control or E1B 19K or 13S E1A expression plasmid (10 μg each) together with HA-tagged JNK1 (1 μg). Forty-eight hours after transfection, the cells were harvested for immunocomplex kinase assays. For infection experiments, HeLa cells were transfected with HA-JNK1 prior to infection with Ad for 24 h and were then lysed for immunocomplex kinase assays. The fold increase in HA-JNK1 activity above the basal level in either mock-infected cells or in cells transfected with CMV empty vector was determined by densitometry. Results represent three experiments. Wt, wild type.

Activation of c-Jun-dependent transcription by E1B 19K.

It is well-documented that JNK activation leads to phosphorylation of c-Jun at Ser63 and Ser73 within its transactivation domain that results in a significantly enhanced transcriptional activity of c-Jun (12, 33). Therefore, we examined whether E1B 19K also activates c-Jun-dependent transcription as a result of its ability to activate JNK. As shown in Fig. 7, E1B 19K significantly enhanced GAL4–c-Jun-mediated transcription compared with CMV vector control (compare lanes 1 and 2 [37-fold increase]). This effect of E1B 19K appeared to be most pronounced for GAL4–c-Jun, while transcription mediated by GAL4-JunD and GAL4–Elk-C was unaffected (Fig. 7, lanes 15 to 18). The inability of E1B 19K to induce transcription mediated by these transcription factors is not due to an inherent defect of these proteins to activate transcription, since GAL4-JunD can activate transcription on its own while constitutively activated MEKK1 strongly induces GAL4–Elk-C activity (data not shown). E1B 19K also moderately stimulated transcription mediated by GAL4–v-Jun and GAL4-JunB (lanes 11 to 14), but the level of activation was not as robust as that for GAL4–c-Jun.

FIG. 7.

Ability of E1B 19K to activate GAL4–c-Jun, GAL4–c-Jun mutants, GAL4–v-Jun, or other transcription factors fused to the GAL4 DNA binding domain. CMV vector control or CMV E1B 19K (10 μg each) was cotransfected into HeLa cells, along with GAL4-E1BCAT reporter plasmid (10 μg) and GAL4 chimeric proteins (1 μg each). The fold activation in CAT activity in the absence of E1B 19K was normalized to 1. The results are the means plus or minus standard deviations from four independent transfections and CAT assays.

To determine the mechanism by which E1B 19K activates GAL4–c-Jun-dependent transcription, we analyzed a series of GAL4–c-Jun mutants. As shown in Fig. 7, the ability of E1B 19K to activate transcription mediated by c-Jun is significantly impaired by mutations of the JNK phosphorylation sites in c-Jun (GAL4–c-Jun Ala63 and 73; compare lanes 2 and 4). This suggests that JNK activation is important for E1B 19K to activate c-Jun-mediated transcription. Consistent with this idea, other deletion mutants of c-Jun (Fig. 7, lanes 5 to 10) as well as v-Jun (Fig. 7, lanes 11 to 12), all of which lack a binding site for JNK, also show a significantly reduced response to E1B 19K compared to wild-type GAL4–c-Jun (compare lanes 6, 8, 10, and 12 with 2). Taken together, these results suggest that E1B 19K specifically activates c-Jun-dependent transcription by activating JNK.

As described earlier, MEKK1 is critical for Ad to activate JNK (Fig. 3B, lane 4); therefore, we tested whether MEKK1 was also important for E1B 19K to activate c-Jun-dependent transcription. As shown in Fig. 8, the dominant-negative mutant MEKK1 K432M significantly reduced GAL4–c-Jun-mediated transcriptional activation by E1B 19K (compare lanes 2 and 3 [24-fold activation in its absence versus 3-fold activation in its presence]). In contrast, MEKK1 K432M failed to interfere with GAL4-Sp1Q2-dependent transcription (Fig. 8, lanes 5 and 6), suggesting that the inhibition of E1B 19K-induced, c-Jun-dependent transcription is not due to a general inhibition of transcription by the MEKK1 dominant-negative mutant. The inhibition of GAL4–c-Jun-mediated transcription by MEKK1 K432M was also not due to inhibition of E1B 19K protein expression, since Western blot analysis indicated that E1B 19K protein levels did not change in the presence or absence of MEKK1 K432M or of other inhibitors [e.g., MKK4 (Ala) and Ras N17] used in this study (data not shown). As expected, the Ras dominant-negative mutant (Ras N17), which did not affect the ability of Ad to activate JNK, also failed to inhibit E1B 19K-induced, GAL4–c-Jun-dependent transcription (data not shown). In summary, these results suggest that E1B 19K activates JNK via the MEKK1-MKK4 pathway, and one of the downstream consequences of this JNK activation is the stimulation of c-Jun-dependent transcription.

FIG. 8.

Effect of dominant-negative MEKK1 on induction of GAL4–c-Jun-mediated transcription by E1B 19K. CMV empty vector or CMV E1B 19K was cotransfected into HeLa cells, along with GAL4-E1BCAT (10 μg) and GAL4–c-Jun (1 μg) in the presence or absence of MEKK1 K432M (2 μg). Negative controls consisted of cotransfecting GAL4 DNA-binding domain alone (aa 1 to 147) or GAL4-Sp1Q2 (10 μg), along with GAL4-E1BCAT (10 μg) in the presence or absence of MEKK1 K432M. CAT activity in the presence of CMV vector or GAL4 (aa 1 to 147) only was normalized to 1. The results are the means plus or minus standard deviations from four independent transfections and CAT assays.

Analysis of E1B 19K mutants for activation of GAL4–c-Jun-mediated transcription.

As mentioned earlier, E1B 19K cooperates with E1A in transformation assays (8, 60, 66), probably due to its ability to inhibit apoptosis (8, 49, 60, 66). To determine whether the induction of GAL4–c-Jun-mediated transcription by E1B 19K contributes to its ability to inhibit apoptosis and to induce transformation, we tested several previously described E1B 19K transformation-defective mutants for their abilities to induce c-Jun-mediated transcription (8, 39, 60, 66). As shown in Fig. 9A, most of these mutants (PM51, PM87, PM102, and dl aa 90 to 96]) can induce c-Jun-dependent transcription like the wild-type E1B 19K protein. In addition, they can also potently activate JNK (Fig. 9C and data not shown). These results suggest that activation of JNK and c-Jun-mediated transcription by E1B 19K may not be critical for its transformation and antiapoptotic activity.

FIG. 9.

(A) Mapping the region of E1B 19K required for GAL4–c-Jun-mediated transcription. Wild-type E1B 19K or mutants were cotransfected into HeLa cells, along with GAL4-E1BCAT (10 μg) and GAL4–c-Jun (1 μg). Negative controls consisted of empty vector alone. The fold activation of GAL4–c-Jun-mediated transcription by wild-type (WT) E1B 19K was normalized to 100%. The results represent three independent experiments. The mutants used in this study are indicated on the left. The shaded boxes indicate regions homologous to Bcl-2. Asterisks indicate regions where wild-type amino acids were replaced by alanine and serine. Plus and minus signs indicate whether the mutants were defective for transformation or for inhibition of apoptosis based on previously published results (8, 60). ND, not determined. (B) Expression levels of E1B 19K mutants as determined by Western blot analysis with rabbit anti-E1B 19K antibodies. (C) Activation of HA-JNK1 by wild-type E1B 19K or E1B 19K mutant s. HeLa cells were cotransfected with vector alone, wild-type E1B 19K, E1B 19K 123,124 WR-AS, or E1B 19K PM87, along with HA-JNK1 expression vector. Immunocomplex kinase assays were performed with GST–c-Jun aa 1 to 79 as a substrate. Fold increase in HA-JNK1 activity above vector control levels was determined by densitometry. Western blot analysis of the protein levels for wild-type E1B 19K or the two mutants is shown in the bottom panel.

Among all of the transformation or apoptosis mutants tested, only one E1B 19K point mutant (123,124 WR-AS) showed a compromised ability to induce JNK activation and GAL4–c-Jun-mediated transcription (Fig. 9A and C). This defect is not due to the lack of stable protein expression, since Western blot analysis indicated that this mutant (123,124 WR-AS) is expressed at levels comparable to that of the wild-type protein (Fig. 9B). The electrophoretic mobility of E1B 19K 123,124 WR-AS was slower than those of other point mutants, consistent with the observations of others (60). The alteration in mobility may suggest additional posttranslational modifications of this particular mutant (60). Taken together, our results suggest that there is no simple correlation between the ability of E1B 19K to activate JNK and c-Jun-mediated transcription and the ability to inhibit apoptosis and to induce transformation.

Analysis of a number of E1B 19K deletion mutants showed that E1B 19K from aa 1 to 146 but not from aa 1 to 88 can induce c-Jun-dependent transcription (Fig. 9A), suggesting that the region between aa 88 and 146 is necessary for E1B 19K to induce c-Jun-dependent transcription. However, it appears that the BH (Bcl-2 homology) domain 3 of E1B 19K is not involved in the activation of GAL4–c-Jun-mediated transcription, since the E1B 19K mutant with a deletion from aa 90 to 96 lacks this domain and is capable of inducing GAL4–c-Jun-mediated transcription at levels comparable to that of the wild-type protein (Fig. 9A).

Synergistic activation of c-Jun-mediated transcription by p300 and E1B 19K.

As shown in Fig. 7, although mutations of the JNK phosphorylation sites in c-Jun significantly reduced the ability of c-Jun to respond to E1B 19K-induced transcriptional activation, these JNK phosphorylation-defective c-Jun mutants still responded to E1B 19K, albeit at a lower level (8-fold versus 40-fold activation of the wild-type c-Jun). This suggests that additional factors are involved in E1B 19K-induced, c-Jun-dependent transcription. Previously, we and others have shown that p300 can act as a transcriptional cofactor for c-Jun (3, 37, 59). Therefore, we tested whether p300 plays a role in the response of GAL4–c-Jun-mediated transcription to E1B 19K. The amounts of E1B 19K and p300 were titrated to give low levels of activation of GAL4–c-Jun-mediated transcription in order to observe potential additive or synergistic effects of p300 and E1B 19K. As shown in Fig. 10A, cotransfection of E1B 19K and p300 together resulted in a further increase of approximately 10-fold in reporter activity compared with either alone (compare lanes 4 with 2 and 3), suggesting that E1B 19K and p300 can synergistically activate c-Jun-mediated transcription. This synergistic effect of E1B 19K and p300 appears largely independent of JNK phosphorylation of c-Jun at Ser63 and 73, since E1B 19K and p300 also synergistically activated transcription mediated by the c-Jun mutant (c-Jun [aa 1 to 223] Ala63 and 73) which no longer could be phosphorylated by JNK at these sites (Fig. 10A, lanes 6 to 8). In contrast, JunD-mediated transcription was mildly activated by E1B 19K or p300 alone, and the combined effect of E1B 19K and p300 appears to be additive rather than synergistic (Fig. 10A, lanes 10 to 12). Finally, E1B 19K 123,124 WR-AS, which was defective for JNK activation and induction of GAL4–c-Jun-mediated transcription (Fig. 9A and 9C), did not synergize with p300 to activate GAL4–c-Jun-mediated transcription (Fig. 10B, compare lanes 8 and 4). Taken together, our results suggest that as a cofactor for c-Jun, p300 may play a role in E1B 19K-induced GAL4–c-Jun-dependent transcription.

FIG. 10.

(A) Overexpression of p300 enhances E1B 19K-induced, GAL4–c-Jun-dependent transcription. CMV E1B 19K (2 μg), GAL4–c-Jun (1 μg), and GAL4-E1BCAT (10 μg) were cotransfected into HeLa cells, along with either CMVβ vector control or CMVβ p300 expression vector (2 μg each). The results are the means plus or minus the standard deviations from three independent transfections and CAT assays. CAT activity in the presence of CMV vector control was normalized to 1. V, CMV empty vector; E, E1B 19K; P, p300. (B) Wild-type E1B 19K but not E1B 19K 123,124 WR-AS synergizes with p300 to activate GAL4–c-Jun-dependent transcription. Wild-type E1B 19K or E1B 19K 123,124 WR-AS was cotransfected with GAL4–c-Jun, CMVβ HA-p300, and GAL4E1BCAT into HeLa cells as described above. Results are the means plus or minus standard deviations from three independent transfections and CAT assays. CAT activity in the presence of CMV vector control was normalized to 1. V, empty vector; E, wild-type E1B 19K; P, p300; EM, E1B 19K 123,124 WR-AS.

DISCUSSION

The Ad E1B proteins play an important role in the inhibition of E1A-mediated apoptosis during a productive infection of human cells and in the transformation of primary rodent cells (64–66). It is believed that the ability of E1B proteins to regulate transcription is important for this biological activity (49, 72, 73). Therefore, understanding the mechanisms by which E1B proteins regulate transcription is of particular importance. In this report, we have described a potential molecular mechanism that underlies the ability of the E1B 19K protein to activate c-Jun-dependent transcription, i.e., the involvement of a MEKK1-MKK4-JNK signaling pathway. The dependence of a signaling cascade to activate transcription suggests that E1B 19K regulates transcription in an indirect manner, as opposed to a direct protein-protein interaction in the case of regulation of p53 by E1B 55K (53, 72). Analysis of dominant-negative mutants of upstream activating kinases of JNK reveals the involvement of MKK4 and MEKK1 but not of the further upstream GTP-binding protein, Ras, in E1B 19K-induced JNK activation and subsequent c-Jun-mediated transcription. Investigation of c-Jun-dependent transcription in response to E1B 19K shows that JNK phosphorylation of c-Jun at Ser63 and 73 is critical but cannot account for the full activation induced by E1B 19K. We have provided evidence that p300 can synergize with E1B 19K to activate c-Jun-mediated transcription, suggesting that p300 may play a role in this activation process. Taken together, our results demonstrate a novel activity of E1B 19K, i.e., its ability to activate a signaling pathway that leads to an alteration of transcription.

Activation of JNK by Ad E1B 19K.

Previous studies showed that Ras and members of the Rho subfamily of GTPases, Rac1 and Cdc42, are involved in JNK activation in response to growth factors or Ha-Ras (10, 43). The results from the present study suggest that the GTP-binding protein Ras is not involved in Ad-induced JNK activation (Fig. 3B). Therefore, the immediate target(s) for Ad-induced activation of the MEKK1-MKK4-JNK signaling pathway appears to be different from that of growth factors. Recently, additional kinases that can act upstream of the MEKK1-JNK pathway have been described, including MUK (25), Tpl-2 (50), HPK-1 (26, 32), and MLK-3 (63). It is possible that these kinases are stimulated by Ad, resulting in the subsequent downstream activation of JNK. Alternatively, Ad may inhibit a phosphatase(s) such as MKP-1 which has been shown to regulate JNK (18).

Mutations of individual Ad early regulatory genes identified the E1 gene products as being important for activation of JNK by Ad (Fig. 4). Subsequent transfection experiments demonstrated that the E1B gene products (E1B 19K and E1B 55K) but not E1A were important for JNK activation. The ability of E1B 19K to activate JNK is correlated with an increase in c-Jun-dependent transcription, which is blocked by the MEKK1 dominant-negative inhibitor that abrogates Ad-induced JNK activation. These results strongly suggest that E1B 19K activates JNK via the MEKK1-MKK4 signaling pathway, which results in activation of c-Jun-dependent transcription.

What are the immediate effectors of E1B 19K that lead to JNK activation? E1B 19K has been shown to interact with a number of apoptosis-inducing proteins including Bax (21), Bak (15), and Bik/Nbk (4, 22), as well as cellular proteins that are termed Nip 1, Nip 2, and Nip 3 (5). Many of the E1B 19K mutants that have been shown to interact poorly with Bax (21) or Bik/Nbk (4, 22) were strong activators of GAL4–c-Jun-mediated transcription in the present study, suggesting that interaction with these proteins is not involved in JNK activation by wild-type E1B 19K. Nip 1 has been demonstrated to share some homology with the catalytic domain of three mammalian calcium-calmodulin-dependent cyclic nucleotide phosphodiesterases; Nip 2 has strong homology to RhoGap (5), whereas Nip 3 has homology to a rat calbindin D protein, which may have some role in mitochondrial function (5). However, the activation of JNK does not appear to correlate with Nip binding, since the E1B 19K mutant with a deletion from aa 90 to 96 strongly activated GAL4–c-Jun-dependent transcription in our study but was found to bind poorly with Nip proteins in the other study (5). As a result, the immediate targets leading to activation of the MEKK1-MKK4-JNK pathway by E1B 19K remain to be determined.

As described above, the E1B 19K protein is not the only adenoviral protein capable of activating JNK. We have also observed that in transient transfection experiments, the E1B 55K protein alone can activate JNK, but at a level lower than that of E1B 19K. Interestingly, E1B 55K can fully complement the Ad E1B mutant dl339 to activate JNK activity to a level that is comparable to that of the wild-type virus (54a). This result suggests that unlike E1B 19K, E1B 55K may induce maximal JNK activation in cooperation with another viral gene product, possibly the E4 34K protein which can complex with E1B 55K in Ad-infected cells (52). In parallel with the activation of JNK, we have also observed induction of c-Jun-mediated transcription by E1B 55K (54a). Whether the mechanism of activation of c-Jun-dependent transcription by E1B 55K is the same as that of E1B 19K is currently under investigation.

Mechanisms that underlie the ability of E1B 19K to activate c-Jun-mediated transcription.

The ability of E1B 19K to activate c-Jun-dependent transcription is specific, since it does not augment transcription mediated by other activators such as GAL4-JunD, GAL4–Elk-C (Fig. 7), or GAL4-Sp1Q2 (data not shown). This argues against the possibility that E1B 19K might e increasing c-Jun-dependent transcription by stabilizing transfected plasmid DNA (23). The data presented in the present paper strongly suggest that E1B 19K induces c-Jun-mediated transcription by activating the JNK pathway. This is supported by the observation that all E1B 19K mutants that can activate JNK also activate c-Jun-mediated transcription, while the E1B 19K mutant (123,124 WR-AS) which is defective in JNK activation is also defective in the induction of c-Jun-dependent transcription. It is not clear why the E1B 19K 123,124 WR-AS mutant is defective in JNK activation and induction of GAL4–c-Jun-dependent transcription. One possibility is that this mutant protein is mislocalized in the cells (60, 65). Alternatively, the inability to induce GAL4–c-Jun-mediated transcription may be due to posttranslational modifications specific to this mutant (Fig. 9B) (60).

The activation of c-Jun-mediated transcription by E1B 19K involves at least two components (depicted in Fig. 11). The first is the activation of JNK by E1B 19K and the subsequent phosphorylation of c-Jun at Ser63 and Ser73. The second component is independent of JNK phosphorylation of c-Jun. The existence of the second component is suggested by the observation that a c-Jun point mutant (Ser63 and 73 to Ala63 and 73) as well as c-Jun gross deletion mutants (aa 43 to 223 and aa 56 to 223) is still able to activate transcription in response to E1B 19K. These gross deletion mutants lack the JNK binding site and therefore cannot be phosphorylated by JNK at either the major site Ser63 and 73 or at minor sites such as Thr91, Thr93, or Thr95 (24). In addition, E1B 19K also activates JunB- and v-Jun-mediated transcription, despite the fact that neither of these two Jun proteins is a JNK substrate (12, 24, 30, 31). Therefore, additional factors must be involved in the full induction of c-Jun-mediated transcription by E1B 19K.

FIG. 11.

Model for E1B 19K-induced, c-Jun-mediated transcription. E1B 19K activates the MEKK1-MKK4-JNK signaling cascade, resulting in phosphorylation of c-Jun at serines 63 and 73. E1B 19K also enhances the ability of c-Jun and p300 to function together to activate transcription by still unidentified mechanisms (discussed in the text). X, protein partner for c-Jun at the AP-1 site.

Since p300 significantly augments GAL4–c-Jun-mediated transcription in response to E1B 19K, it is possible that p300 is part of this second component that leads to full, E1B 19K-induced c-Jun-mediated transcription. How E1B 19K synergizes with p300 to activate c-Jun-mediated transcription is not clear at this time. We have been unable to detect any physical interactions between E1B 19K and either p300 or c-Jun (unpublished results), suggesting that E1B 19K may be exerting its effects through either intermediary proteins and/or through signaling molecules. It is possible that E1B 19K functions to facilitate recruitment of p300 by c-Jun and/or to induce posttranslational modifications of p300 that may increase its transcriptional activity. Consistent with the involvement of a signaling pathway in the synergism between E1B 19K and p300, we observed that the MEKK1 dominant-negative mutant can completely abrogate the ability of E1B 19K to induce c-Jun-dependent transcription (Fig. 8). Furthermore, E1B 19K 123,124 WR-AS, which was defective for JNK activation, was unable to synergize with p300 to activate GAL4–c-Jun-dependent transcription (Fig. 10B). We are currently investigating the possibility that the observed synergism involves activation of the MEKK1-JNK pathway by E1B 19K, which then leads to the posttranslational modification not only of c-Jun but also of p300. We speculate that the posttranslational modifications of both p300 and c-Jun enhance the interaction between both proteins as well as the overall transcriptional potency of the c-Jun–p300 complex. In this regard, it is interesting to note that the p300-related protein CBP has been shown to be phosphorylated by MAPK in vitro and that this phosphorylation event may be responsible for an increased transcriptional activity of CBP in vivo (27).

The ability of the E1B 19K protein to regulate transcription has been reported previously, e.g., E1B 19K can activate virus (e.g., simian virus 40 [SV40] and the polyomavirus enhancer) and cellular enhancer-driven transcription (49, 58, 74). Interestingly, E1A has been shown to repress the SV40 enhancer, and this repressive effect seems to be achieved by inactivating the transcriptional cofactor p300 (14). In light of our finding that p300 and E1B 19K can synergize to activate c-Jun-mediated transcription, it is possible that activation of the SV40 enhancer by E1B 19K and repression of the enhancer by E1A may both involve p300. Recently, E1B 19K has been shown to augment p53-mediated transactivation and to alleviate p53-mediated transcriptional repression (49, 55). Like c-Jun, p53 also utilizes p300 as a cofactor to activate transcription (2, 17, 38). Taken together, it is possible that the ability of E1B 19K to regulate transcription and to antagonize the transcriptional effects of E1A may involve p300 and specific signaling molecules.

Lastly, it is interesting to note that not all transcription factors that are substrates for JNK are affected by E1B 19K. For instance, we show that E1B 19K does not affect GAL4–Elk-1-dependent transcription despite the fact that Elk-1 is a JNK substrate (6). Although the underlying mechanisms are currently unclear, this observation is not unprecedented. Previously, it has been reported that v-src activates JNK and GAL4–c-Jun-mediated transcription but has little effect on transcription mediated by ATF-2 (69). It is possible that the specificity of E1B 19K for c-Jun is related to its ability to activate specific isoforms of JNK. At least 10 isoforms of JNK have been identified, and comparisons of their binding activities demonstrate that the JNK proteins differ in their interactions with ATF-2, Elk-1, and Jun transcription factors and that each JNK group may target specific transcription factors selectively (18).

At present, the biological significance of E1B 19K-induced JNK activation and increase in c-Jun-mediated transcription is not known. We have tested several mutants of E1B 19K that are defective either for cooperativity with E1A in transformation assays or for inhibition of either tumor necrosis factor alpha (TNF-α) or cisplatin-induced apoptosis (8, 60, 66) and found that most retained the ability to activate JNK and GAL4–c-Jun-dependent transcription (Fig. 9A and C and data not shown). Only one E1B 19K point mutant, 123,124 WR-AS, shows a parallel defect in activating JNK–c-Jun-dependent transcription, in inducing transformation, and in inhibiting cisplatin-induced apoptosis (60). Taken together, it appears that there is no simple correlation between the ability of E1B 19K to induce JNK activation and c-Jun-dependent transcription and its ability to inhibit apoptosis or to induce transformation in cooperation with E1A. In further support of the possibility of a dissociation between these two activities of E1B 19K, we found that the dominant-negative MEKK1 inhibitor does not block the ability of E1B 19K to inhibit TNF-α-induced apoptosis (data not shown). Similarly, the activation of GAL4–c-Jun-mediated transcription by E1B 19K does not appear to be related to its ability to inhibit NF-κB activity, since the mutant E1B 19K 123,124 WR-AS was defective in inducing both JNK activation and GAL4–c-Jun-dependent transcription (Fig. 9A and C) but is fully active in NF-κB inhibition (39). Currently, we are investigating whether activation of JNK and c-Jun-mediated transcription by E1B 19K may play a role in adenoviral lytic infection.

In summary, we have identified a molecular mechanism that underlies the ability of E1B 19K to activate c-Jun-dependent transcription. The ability of E1B 19K to activate c-Jun transcription is indirect and involves the MEKK1-MKK4-JNK pathway, p300, or possibly other c-Jun coactivators such as the recently identified Jun-activating-domain-binding protein 1 (JAB1) (9). It will be interesting to determine whether this is a general mechanism underlying E1B 19K-mediated transcriptional activation. Previous work has provided ample evidence that Ad E1A and E1B antagonize one another by converging on cellular targets such as p53 (11, 42, 72). Our findings suggest that c-Jun may be another such cellular factor upon which the opposing activity of E1A and E1B converge and that the balance of these two regulatory pathways may be critical for viral lytic infection and/or viral oncogenesis.