E1A Sensitizes Cells to Tumor Necrosis Factor-induced Apoptosis through Inhibition of IκB Kinases and Nuclear Factor κB Activities*

The adenovirus E1A protein has been implicated in increasing cellular susceptibility to apoptosis induced by tumor necrosis factor (TNF); however, its mechanism of action is still unknown. Since activation of nuclear factor κB (NF-κB) has been shown to play an anti-apoptotic role in TNF-induced apoptosis, we examined apoptotic susceptibility and NF-κB activation induced by TNF in the E1A transfectants and their parental cells. Here, we reported that E1A inhibited activation of NF-κB and rendered cells more sensitive to TNF-induced apoptosis. We further showed that this inhibition was through suppression of IκB kinase (IKK) activity and IκB phosphorylation. Moreover, deletion of the p300 and Rb binding domains of E1A abolished its function in blocking IKK activity and IκB phosphorylation, suggesting that these domains are essential for the E1A function in down-regulating IKK activity and NF-κB signaling. However, the role of E1A in inhibiting IKK activity might be indirect. Nevertheless, our results suggest that inhibition of IKK activity by E1A is an important mechanism for the E1A-mediated sensitization of TNF-induced apoptosis.

oncogenes such as ras or E1B (1). However, recent studies indicate that E1A has strong tumor suppression activities, including suppression of transformation, tumorigenicity, and metastasis (2)(3)(4)(5). E1A has also been shown to interact with the tumor suppressor Rb and the transcriptional coactivator p300 in regulating cell differentiation, proliferation, and apoptosis (6,7). When E1A associates with Rb, it leads to the release of the transcription factor E2F, which promotes cells to enter S phase of the cell cycle. However, when E1A associates with p300 and the tumor suppressor p19 ARF , it results in the accumulation and stabilization of p53, which induces both p53-dependent and p53-independent apoptosis (7,8). Moreover, E1A has been shown to sensitize cells to various stimuli causing them to undergo apoptosis. Such stimuli include ionizing irradiation, DNA-damaging agents, serum starvation, and tumor necrosis factor (TNF) 1 (9,10). Although induction of cellular susceptibility to TNF has been reported to depend on the p300 and Rb binding domains of E1A (10), the mechanism by which E1A sensitizes cells to TNF-induced apoptosis is unknown.
We have shown previously that E1A mediates sensitization of radiation-induced apoptosis through inhibition of NF-B activity (9). The NF-B/Rel family of transcription factors plays a crucial role in regulating genes that function in immunologic and inflammatory responses, cell proliferation, and apoptosis (11,12). This family consists of p65 (RelA), p50 (NF-B1), c-Rel, RelB, and p52 (NF-B2) subunits, which can dimerize in various combinations. The primary form of NF-B is a heterodimer of p50 and RelA subunits and is retained as a latent form in the cytoplasm of resting cells by IB, an inhibitor of NF-B (13). NF-B is activated by stimulation of the IB kinase (IKK) complex, which phosphorylates IB and triggers its ubiquitination-dependent degradation (11,14). This results in nuclear translocation of the activated NF-B and activation of the target genes. A cytokine-responsive IKK complex containing IKK-␣, IKK-␤, and IKK-␥ subunits has been identified, and the genes encoding these subunits have been cloned (15).
It is intriguing that TNF on the one hand induces cellular apoptosis and on the other hand activates NF-B, which prevents TNF-induced apoptosis (16,17). The link between these contradictions is unknown. Mice lacking the RelA gene died embryonically from extensive apoptosis within the liver (18). The TNF-treated RelAϪ/Ϫ mouse embryonic fibroblasts, macrophages, and 3T3 cell lines showed a dramatic decrease in viability when compared with the TNF-treated RelAϩ/ϩ cells (19), suggesting that RelA plays an essential role in protecting cells from TNF-induced apoptosis. Moreover, overexpression of a superrepressor IB mutant in the TNF-resistant cell lines results in the blockage of NF-B activity and enhancement of TNF-induced apoptosis (20 -22), implying that inhibition of NF-B activation plays a role in TNF-induced apoptosis.
In light of these findings, we hypothesized that E1A may mediate sensitization of cells to TNF through regulation of the NF-B signaling pathway. Thus, we examined the E1A transfectants of human cancer cell lines for their sensitivity to TNFinduced apoptosis and the role of E1A in regulation of NF-B activation. We found that the cells transfected with E1A, but not E1A mutants, became very sensitive to TNF-induced apoptosis. Furthermore, we found that this E1A-mediated sensiti-zation of TNF-induced apoptosis was due to inhibition of IKK activity, IB phosphorylation, and NF-B activation.

EXPERIMENTAL PROCEDURES
Cell Lines and Cultures-The establishment and culture conditions of ip1-E1A2, ip1-Efs, and SKOV3.ip1 cell lines have been described previously (4). To establish the Edl.0108/ip1 cell line, the SKOV3.ip1 cells were transfected with the pE1A mutant DNA whose p300 and Rb binding domains were deleted. All of these stable cell lines were cell clones which were isolated by G418 selection. The culture conditions for the Edl.0108/ip1 cells, a human prostate cancer cell line (PC3), and its transfectants (PC3-E1A1 and PC3-neo) were the same as those for ip1-E1A2, ip1-Efs, and SKOV3.ip1 cells.
Apoptosis Assays-The luciferase-based in vitro cell viability assay was performed as described previously (23). Specifically, ip1-E1A2, ip1-Efs, and Edl.0108/ip1 cells were transfected with the cytomegalovirus (CMV) promoter-luciferase expressing vector (pCMV-luc), using liposome as a gene delivery vehicle. About 36 h after transfection, the cells were treated with or without TNF (20 ng/ml). After incubation for an additional 12 h, the cells were lysed, and the luciferase activity was determined. The percentage of luciferase activity of the TNF-treated cells was normalized by using the percentage of luciferase activity of the untreated cells (100%) as base line. Standard deviations were calculated from three independent experiments. The apoptotic cells were also analyzed by the deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL) assay as described previously (24). Briefly, the cells were seeded in an eight-chamber slide (1.6 ϫ 10 4 cells/chamber) for 4 h, and TNF was added to the culture. The cells were then cultured for 3 days, and the TUNEL assay was performed. The percentage of apoptotic cells was quantitated and S.D. were calculated from three independent experiments.
Electrophoretic Mobility Shift Assay (EMSA) and Western Blot Analysis-The cells were treated with TNF (20 ng/ml) for 30 min or untreated (as controls). Cell extracts were prepared, and EMSA for NF-B was performed as described previously (9). For Western blot analysis, the cells were treated with TNF (20 ng/ml) for 4 min or left untreated (as controls), cell extracts were prepared, and Western blot analysis was performed as described previously (4). Antibodies against RelA (SC-109, Santa Cruz) and IB-␣ (SC-371, Santa Cruz) were used.
Dephosphorylation Assay-IB-␣ was immunoprecipitated from the cell lysates using an anti-IB-␣ antibody (SC-371, Santa Cruz). The precipitates were then incubated with calf intestinal phosphatase (Promega) at 37°C for 30 min. Subsequently, the samples were dissolved in the loading buffer and subjected to 12% SDS-polyacrylamide gel electrophore- The ip1-Efs, Edl.0108/ip1, and SKOV3.ip1 cells were transfected with a luciferase expression vector (pCMV-luc) for 36 h, treated with human TNF (20 ng/ml) for 12 h, and then harvested. Control cells were harvested 48 h after transfection without TNF treatment. Equal amounts of cell lysates were used for luciferase assays and analyzed by Western blot for the presence of E1A (inset) as described under "Experimental Procedures." The percentage of luciferase in the TNF-treated cells represents cell viability and has been normalized against that of the corresponding untreated control cells (100%). The data presented were the mean of three independent experiments, and S.D. was indicated. B, apoptotic assays of the ip1-E1A2, ip1-Efs, and SKOV3.ip1 cells were performed. The cells were seeded in an eight-chamber slide (1.6 ϫ 10 4 cells/chamber) for 4 h before they were treated with or without TNF. The cells were cultured for 3 days followed by a TUNEL assay. Quantitation of apoptotic cells was indicated as a percentage over the total cell number per field (100%), and error bars depict S.D. of triplicate samples.

FIG. 2. E1A suppresses the activation of NF-B induced by TNF.
A, equal amounts of nuclear extracts isolated from the TNFinduced or untreated ip1-E1A2, ip1-Efs, and Edl.0108/ip1 cells were subjected to EMSA with a NF-B oligonucleotide probe. The cold wildtype and mutant NF-B oligonucleotides were included as controls in the TNF-induced ip1-Efs EMSA. As a RelA control, an anti-RelA antibody (SC-109, Santa Cruz) was also included in the assay. The RelA-specific supershifted complex was highlighted (*). B, NF-B (RelA) proteins in the TNF-induced or control ip1-E1A2, ip1-Efs, SKOV3.ip1, and Edl.0108/ ip1 cells were analyzed by Western blot analysis. Equal amounts of total cell proteins were subjected to SDS-PAGE (10% gel) and detected by Western blot with an anti-RelA antibody (top panel). As a control, the same blot was probed with an anti-actin antibody (bottom panel). sis (PAGE) and immunoblot analysis using the same anti-IB-␣ antibody.
Transient Transfections and Immunocomplex Kinase Assays-The various ip1 cell lines were plated the day before transfection at a density of 2 ϫ 10 6 cells per 100-mm dish. Cells were transfected with either an IKK-␣-FLAG or a HA-IKK-␤ expression vector or an empty vector (control), using liposome as described above. Cell extracts were prepared 48 h after transfection, and immunocomplex kinase assays were performed as described previously (25).

RESULTS AND DISCUSSION
To test whether E1A could affect cellular susceptibility to cell death induced by TNF, an in vitro cell viability assay was performed using a luciferase assay (23). We used a human ovarian cancer cell line derivative, SKOV3.ip1, which was stably transfected with the wild-type Ad5 E1A (ip1-E1A2) or an E1A frameshift Efs mutant (ip1-Efs) that has lost the E1A function, or an E1A deletion mutant (Edl.0108/ip1) whose p300 and Rb binding domains were deleted (10). When the cells were transiently transfected with the luciferase reporter gene (pCMV-luc) and treated with TNF, the luciferase activity was strongly reduced in the ip1-E1A2 cells compared with the parental SKOV3.ip1 and the mutant ip1-Efs cells (Fig. 1A). This suggests that E1A sensitizes cells to TNF-induced cell death. However, this E1A-mediated sensitization of TNF-induced cell death was abolished in the Edl.0108/ip1 cells, suggesting that the p300 and Rb binding domains of E1A are required for this sensitization. This is consistent with the previous finding that the E1A function in increasing cellular susceptibility to cell death induced by TNF depends on its binding to either p300 or p105Rb (10). As a control, we examined the expression of E1A in these cell lines (ip1-E1A2, ip1-Efs, SKOV3.ip1, and Edl.0108/ ip1) by Western blot analysis using an anti-E1A monoclonal antibody (mAb) (M58). These results indicated that the ip1-E1A2 and Edl.0108/ip1 cells expressed a significant amount of E1A, but the parental (SKOV3.ip1) and the mutant (ip1-Efs) cells did not (inset, Fig. 1A). As another control, we also examined the expression of TNF receptor (TNFR) in these cell lines (ip1-E1A2, ip1-Efs, and SKOV3.ip1) by Western blot analysis using an anti-TNFR1 antibody (R&D System), and showed that the levels of TNFR were similar in these cell lines (data not shown). To confirm the E1A-mediated sensitization of TNF-induced apoptosis, we examined these TNF-treated ip1-E1A2 cells by a TUNEL assay. As shown in Fig. 1B, TNF indeed induced apoptotic DNA breakage in the ip1-E1A2 cells. As negative controls, the ip1-Efs and SKOV3.ip1 cells did not show apoptotic phenotypes with the TNF treatment. Moreover, many of the ip1-E1A2 cells treated with TNF showed the morphologic changes associated with apoptosis, including cell shrinking and apoptotic body formation (data not shown). Taken together, these results suggest that E1A sensitizes cells to TNF-induced apoptosis.
Recently, NF-B has been shown to have an important role in the antiapoptotic pathway (9, 18 -22). Although NF-B plays a role in blocking apoptosis induced by TNF (19 -22), the involvement of NF-B in the E1A-mediated cellular susceptibility to TNF has not been examined. Therefore, the ip1-E1A2, ip1-Efs, and Edl.0108/ip1 cells for their NF-B DNA binding activities before and after TNF treatment were analyzed by EMSA. Our results showed that TNF induced NF-B DNA binding activity in the ip1-Efs and Edl.0108/ip1 cells, but not in the ip1-E1A2 cells (Fig. 2A). The activated NF-B complex FIG. 3. E1A inhibits IB phosphorylation induced by TNF. A, phosphorylation of IB-␣ induced by TNF was analyzed by Western blot analysis. The ip1-E1A2, ip1-Efs, SKOV3.ip1, and Edl.0108/ip1 cells were treated with or without TNF (20 ng/ml) for 4 min before harvest. Equal amounts of total cell lysates isolated from these cells were subjected to SDS-PAGE (12% gel), and the phosphorylated IB-␣ (p-IB-␣) and IB-␣ were detected by Western blot with an anti-IB-␣ polyclonal antibody (top two panels). As a control, the same blot was probed with an anti-actin antibody (bottom panel). B, phosphorylation of IB-␣ induced by TNF could be dephosphorylated by a phosphatase. Equal amount of total cell lysates from the TNF-induced ip1-Efs and SKOV3.ip1 cells were immunoprecipitated with an anti-lB-␣ antibody. The precipitates were incubated with or without calf intestine phosphatase (CIP) for 30 min at 37°C, before they were subjected to SDS-PAGE (12% gel). IB-␣ was detected by Western blot with an anti-IB-␣ antibody. C, phosphorylation of IB-␣ induced by TNF in PC3 cells was detected by Western blot analysis. The PC3-E1A1, PC3-neo, and PC3 cells were treated with or without TNF, and p-IB-␣ and IB-␣ were detected by Western blot analysis as described above.
FIG. 4. E1A down-regulates the activities of IKK-␣ and -␤ induced by TNF. The endogenous IKK-␣ (A) and IKK-␤ (B) activities were examined in the various ip1 cell lines. The SKOV3.ip1, ip1-E1A2, ip1-Efs, and Edl.0108/ip1 cells were treated with human TNF (20 ng/ml) for 10 min before harvest, and equal amounts of cell lysates were used for kinase assays. After immunoprecipitation with either anti-IKK-␣ or anti-IKK-␤ antibody, IKK-␣ (A) or IKK-␤ (B) kinase activity was measured by an immunocomplex kinase assay, using GST-IB-␣ (1-54) as a substrate. The ectopically expressed IKK-␣-FLAG (C) and HA-IKK-␤ (D) activities were examined in the various ip1 cell lines. The SKOV3.ip1, ip1-E1A2, ip1-Efs, and Edl.0108/ip1 cells were transfected with either IKK-␣-FLAG (C) or HA-IKK-␤ (D), and the cells were harvested 48 h after transfection with the stimulation of TNF as described above. As a control, the SKOV3.ip1 cells were transfected with an empty vector (lane 5 in C and D) and treated with TNF. After immunoprecipitation with either an anti-FLAG or anti-HA mAb, IKK-␣ (A) or IKK-␤ (B) kinase activity was measured as described above. As controls, the bottom panels show Western blots indicating the expression levels of both endogenous IKK-␣ or IKK-␤ and the transfected IKK-␣-FLAG or HA-IKK-␤ in the various ip1 cell lines. induced by TNF in the ip1-Efs cells was eliminated in the presence of excess cold wild-type, but not mutant, NF-B oligonucleotides, suggesting that the activated NF-B DNA binding activity is NF-B DNA specific ( Fig. 2A). To confirm the presence of the RelA subunit of NF-B, we performed the same EMSA in the presence of an anti-RelA polyclonal antibody. A RelA-specific super-shifted complex was detected, indicating that the binding complex is indeed an activated NF-B ( Fig.  2A), presumably the RelA/p50 heterodimer as reported previously (9). Thus, these results indicate that E1A is capable of inhibiting NF-B activation induced by TNF. To test whether inhibition of NF-B activity correlates with inhibition of NF-B protein expression, whole-cell extracts from the ip1-E1A2, SKOV3.ip1, ip1-Efs, and Edl.0108/ip1 cells with or without TNF treatment were examined by Western blot analysis using an anti-RelA antibody. While NF-B (RelA) protein expression was not inhibited by E1A, its expression was slightly enhanced with TNF stimulation (Fig. 2B). Although this increase of NF-B protein by TNF may partially contribute to the TNFinduced NF-B activity, it does not account for the inhibition of TNF-induced NF-B activity by E1A. Our data suggest that E1A inhibits NF-B activation but not its protein expression.
To further study how E1A might inhibit the TNF-induced NF-B activity, we investigated whether E1A could down-regulate NF-B activity through inhibition of IB phosphorylation. The changes of phosphorylation and expression of IB-␣ in the ip1-E1A2, SKOV3.ip1, ip1-Efs, and Edl.0108/ip1 cells treated with TNF for 4 min were examined by Western blot analysis. As shown in Fig. 3A, there was only one IB-␣ band observed in the ip1-E1A2 cells with or without treatment with TNF. The same band was also detected in the non-TNF-treated SKOV3.ip1, ip1-Efs, and Edl.0108/ip1 cells. However, two IB-␣ bands were detected in the TNF-treated SKOV3.ip1, ip1-Efs, and Edl.0108/ip1 control cells (Fig. 3A); the upper band should be the phosphorylated form of IB-␣. To confirm this, we immunoprecipitated IkB-␣ from the cell extracts with an anti-IB-␣ antibody. Then the precipitated pellets were treated with or without calf intestinal phosphatase and subjected to immunoblotting with the same anti-IB-␣ antibody. The results showed that the upper band disappeared after the calf intestinal phosphatase treatment (Fig. 3B), indicating that the upper band is indeed the phosphorylated form of IB-␣, and E1A inhibits TNFinduced IB-␣ phosphorylation. The finding that the level of IB-␣ protein was elevated in the ip1-E1A2 cells might be due to slower degradation of the nonphosphorylated IB-␣ protein (Fig.  3A). Furthermore, the E1A-mediated suppression of IB phosphorylation induced by TNF was confirmed by using a human prostate cancer cell line PC3 and its E1A transfectants (Fig. 3C) and a human ovarian cancer cell line 2774 (data not shown).
It has been well documented that TNF induces the activation of IKK, which in turn phosphorylate IB-␣ with the subsequent activation of NF-B (11,14). To examine whether E1A could regulate IKK activity, the ip1-E1A2, SKOV3.ip1, ip1-Efs, and Edl.0108/ip1 cells were treated with TNF, and the endogenous IKK-␣ and IKK-␤ activities were determined by immunocomplex kinase assays. The endogenous IKK-␣ (Fig. 4A) and IKK-␤ (Fig. 4B) activities were readily detected in the SKOV3.ip1, ip1-Efs, and Edl.0108/ip1 cells, whereas the activities of both IKKs in the ip1-E1A2 cells were inhibited. To confirm that E1A could inhibit the activities of IKK-␣ and IKK-␤ expressed ectopically, we transiently transfected an IKK-␣-FLAG or a HA-IKK-␤ expression vector or an empty vector (control) into the ip1-E1A2, SKOV3.ip1, ip1-Efs, and Edl.0108/ip1 cells. The activities of IKK-␣ and IKK-␤ in these transfected cells were measured after treatment with TNF. Our results showed that the activities of IKK-␣ (Fig. 4C) and IKK-␤ (Fig. 4D) were signif-icantly suppressed in the ip1-E1A2 cells compared with the same vector transfected in the SKOV3.ip1, ip1-Efs, and Edl.0108/ip1 cells, suggesting that E1A inhibits the TNF-induced IKK-␣ and IKK-␤ activities. Moreover, this inhibition was abrogated in the Edl.0108/ip1 cells, implying that the p300 and Rb binding domains are essential for the E1A function in blocking IKK activity in the presence of TNF. Thus, taken together, our results strongly suggest that E1A down-regulates the TNF-induced NF-B signaling pathway through inhibition of IKK activity, and this mechanism contributes significantly to the E1A-mediated sensitization of cells to TNF-induced apoptosis.
In order to discern the role of E1A in inhibition of IKK activity, direct binding between E1A and IKK was examined by immunoprecipitation and Western blot analysis. An E1A expression vector (pE1A) was cotransfected with either IKK-␣-FLAG or a HA-IKK-␤ into 293T cells. Whole cell lysates were immunoprecipitated with an anti-E1A mAb (M58), and the immunoprecipitates were subjected to Western blot analysis with either an anti-FLAG or anti-HA mAb. While M58 mAb immunoprecipitated E1A protein (ϳ43 kDa) from the cell lysates, neither anti-FLAG nor anti-HA mAb detected IKK-␣-FLAG or HA-IKK-␤ coimmunoprecipitated with E1A in Western blot analysis (data not shown). Similarly, cell lysates were immunoprecipitated with either an anti-FLAG or anti-HA mAb, and the immunoprecipates were then subjected to Western blot analysis with M58 mAb. Again, no detectable E1A protein was coimmunoprecipitated with either IKK-␣-FLAG or HA-IKK-␤ (data not shown). These results indicated no direct binding between E1A and IKK, suggesting that E1A may act on IKK indirectly. Further investigation of the direct cellular target(s) of E1A in this signaling pathway is necessary to elucidate the mechanism underlying the E1A-regulated IKK activity and apoptosis.