Stabilization of p53 is a novel mechanism for proapoptotic function of NF-κB

Both pro- and antiapoptotic activities of NF-κB transcription factor have been observed; however, less is known about the mechanism by which NF-κB induces apoptosis. To elucidate how NF-κB regulates proapoptotic signaling, we performed functional analyses using wild-type, ikk1-/-, ikk2-/-, rela-/- murine fibroblasts, MDAPanc-28/Puro, MDAPanc-28/IκBαM, and HCT116/p53+/+ and HCT116/p53-/- cells with investigational anticancer agent doxycycline as a superoxide inducer for generating apoptotic stimulus. In this report, we show that doxycycline increased superoxide generation and subsequently activated NF-κB, which in turn up-regulated p53 expression and increased the stability and DNA binding activity of p53. Consequently, NF-κB-dependent p53 activity induced the expression of p53-regulated genes PUMA and p21waf1 as well as apoptosis. Importantly, lack of RelA, IKK, and p53 as well as expression of a dominant negative IκBα (IκBαM) inhibited NF-κB-dependent p53 activation and apoptosis. The doxycycline-induced NF-κB activation was not inhibited in HCT116/p53-/- cells. Our results demonstrate that NF-κB plays an essential role in activation of wild-type p53 tumor suppressor to initiate proapoptotic signaling in response to overgeneration of superoxide. Thus, these findings reveal a mechanism of NF-κB-regulated proapoptotic signaling.

The mammalian Rel/NF-B 1 family, which consists of RelA (p65), Rel (v-rel), RelB, p50 (p105), and p52 (p100), plays a key role in the regulation of immune response, inflammatory reactions, cell proliferation, and apoptosis (1)(2)(3)(4). NF-B is activated through a complex network of kinase signaling cascades in response to various stimuli (5)(6)(7)(8). Recently we have demonstrated the mechanism by which pro-inflammatory cytokines induce biphasic NF-B activation (9). IKK1 and IKK2 are the essential kinases in signal-induced phosphorylation of IB proteins and subsequent activation of NF-B, which in turn induces a large number of genes to affect the subsequent response and phenotype of a cell (5, 10 -13). RelA, the p65 subunit of Rel/NF-B transcription factors has been shown to play a key role in protecting cells from proapoptotic stimuli (14,15). Similarly, radiation-, daunorubicin-, or TNF-␣-induced apoptosis is potentiated in HT1080 human fibrosarcoma and Jurkat cell lines transfected with dominant negative IB␣ (16,17). Many studies have shown that proapoptotic signals can induce Rel/NF-B, which in turn induces expression of the genes involved in suppressing apoptotic signals (4) (18). Inhibitors of apoptosis c-IAP1 and c-IAP2 and Bcl-2 family members Bcl-x L and Bfl1/A1 have been proposed to mediate NF-B-dependent antiapoptotic signaling (19 -23). Conversely, a proapoptotic aspect of RelA activity has also been reported (24 -28). For example, the induction of apoptosis by glucocorticoids is promoted by inhibition of NF-B, whereas apoptosis induced in the same cells by stimulation with phorbol ester and ionomycin for mimicking T-cell activation requires NF-B (28). It has also been shown that NF-B induces cell death following T-cell receptor engagement or DNA-damaging agents (24 -26). Other reports have shown that NF-B activation is required for the onset of apoptosis induced by alphavirus or kainic acid (27,29). These studies further emphasize that the function of NF-B can be proapoptotic or antiapoptotic, depending on cell type, extent of NF-B activation, and nature of the apoptotic signals. However, how NF-B induces apoptosis and which proapoptotic downstream target genes is induced by NF-B still remains unclear.
The tumor suppressor p53 plays an important role in regulating expression of genes that mediate cell cycle arrest and/or apoptosis in response to genotoxic insults (30,31). Reactive oxygen species (ROS) are some of the potent activators of p53 and they appear to be key factors generated in chemotherapeutic agents induced p53 activation (32). The important function of p53 in mediating hydrogen peroxide-induced apoptosis is demonstrated in p53-null cells, suggesting that loss of the function of p53 is a contributing factor to the chemotherapeutic resistance in tumors (33,34). Following environmental insults, p53 is activated by post-translational modifications such as phosphorylation and acetylation that increase its protein stability and enhance its DNA binding activity (35)(36)(37). However, it is unclear how these posttranslational mechanisms that modulate p53 activity are regulated by ROS. Activated p53 up-regulates expression of several of its downstream target genes, including p53 upregulated modulator of apoptosis (PUMA) (38,39) and cell cycle regulator p21 waf1 (40,41), and thus the high levels of p53 activity can either result in cell cycle arrest or directly promote cell death (42,43). A number of factors including the cell type, the specific insults, and the extent of the damage may contribute this decision for apoptosis or cell cycle arrest. In response to many inducing agents, the activity of a well documented antiapoptotic transcriptional factor, NF-B, and proapoptotic factor, p53, are simultaneously activated (44 -46). The functional NF-B and p53 activity may modulate each other, which in turn would affect the subsequent responses (47,48).
Doxycycline and the newly developed chemically modified tetracycline derivatives COL-3 have been evaluated in preclinical cancer models and early clinical trials because these agents inhibit various zinc-dependent enzymes of the matrix metalloproteinase family and induce apoptosis in a number of cancer cell lines (49 -52). However, the mechanism by which doxycycline induces apoptosis remains unclear. Kroon and co-workers (53,54) showed that tetracycline acts as an anticancer agent by preferentially inhibiting mitochondrial protein synthesis, including cytochrome c oxidase, the key components of electron transport chain, and decrease of its synthesis may lead to a disruption of electron transport function and lead to electron leakage from the respiratory chain to O 2 , thus resulting elevated levels of superoxide radicals.
In this study, we performed functional analyses using ikk1 Ϫ/Ϫ , ikk2 Ϫ/Ϫ , and rela Ϫ/Ϫ fibroblasts and MDAPanc-28/ IB␣M and HCT116/p53 Ϫ/Ϫ cells to decode the role of NF-B in regulating p53-dependent proapoptotic signaling in response to doxycycline-induced ROS. Our study reveals a mechanism by which NF-B functions as a proapoptotic factor by activating the p53 signaling pathway for initiating cell apoptosis.

EXPERIMENTAL PROCEDURES
Reagents-Doxycycline was purchased from Sigma. Oligodeoxyribonucleotides were purchased from Sigma-Genosys. A stock solution of 8 mg/ml doxycycline was prepared immediately before its use in phosphate-buffered saline (PBS) with a pH of 7.4. Stock solutions were diluted in PBS to allow a 1% volume addition of diluted solutions to the experimental wells. Control wells were added with 1% volume with PBS. Primary antibodies for immunoblotting (IB␣, IB␤, and ␤-actin) were purchased from Santa Cruz Biotechnology, Inc. Antibodies for p53 immunoblotting and EMSA were purchased from Calbiochem. The proteasome inhibitor PS-341 was reconstituted in PBS and used in 100 nM final concentrations. Radioisotopes were purchased from Amersham Biosciences.
Cell Culture-The human pancreatic tumor cell line MDAPanc-28, which was originally established by Frazier et al. (55), and HCT116 cell lines were maintained in Dulbecco's modified Eagle's medium supplemented with 10% heat-incubated fetal bovine serum. The wild-type, ikk1 Ϫ/Ϫ , ikk2 Ϫ/Ϫ , and rela Ϫ/Ϫ cells were grown on gelatin-coated tissue culture dishes in Dulbecco's modified Eagle's medium containing 4.5 g/liter of glucose, glutamine, sodium pyruvate, and nonessential amino acids and supplemented with 12% heat-inactivated fetal bovine serum. Cells were incubated at 37°C in a humidified 5% CO 2 atmosphere.
Cell Proliferation and Cell Cycle Analysis-Cells were grown at a concentration of 10 6 cells/ml in 6-well Costar plates in Dulbecco's modified Eagle's medium at 37°C for 6 days in a 95% O 2 , 5% CO 2 incubator. Aliquots of cells and medium were removed at 2-day intervals. Cultures containing 40 g/ml of doxycycline were examined, and cell numbers were counted at each time interval with a Coulter Z1 particle counter (Beckman) after the cells were harvested by trypsinization. Subconfluent cells remained untreated or were treated with 40 g/ml of doxycycline for 12 and 24 h. Cells were then harvested and fixed with ice-cold 70% (v/v) ethanol for 24 h. After centrifugation at 200 ϫ g for 5 min, the cell pellet was washed with PBS (pH 7.4) and resuspended in PBS containing propidium iodide (50 g/ml), Triton X-100 (0.1%, v/v), 0.1% sodium citrate, and DNase-free RNase (1 g/ml). Cells were then incu-bated at room temperature for 1 h, and DNA content was determined by flow cytometry using a FACScan flow cytometer (BD Biosciences).
Measurement of Cellular Superoxide-Superoxide was measured using hydroethidine (HEt, Molecular Probes). Hydroethidine emits light blue fluorescence and, on interaction with O 2 , is converted to ethidium, which intercalates into the DNA and emits a red fluorescence. After doxycycline treatment, cells were labeled with HEt (100 ng/ml, 60 min) and then analyzed by flow cytometry as described previously (56). Data were analyzed using the BD Biosciences CellQuest Pro software package.
DNA Fragmentation-DNA fragmentation in apoptotic cells was determined by gel electrophoresis. The cells (50 ϫ 10 6 ) treated with doxycycline for the specified times were collected and washed with PBS, followed by incubation with extraction buffer (10 mM Tris, pH 8.0; 0.1 mM EDTA, 0.5% SDS; and 20 g/ml of RNase) at 37°C for 1 h. Then, 100 g/ml of proteinase K was added, and the sample was incubated at 50°C for 3 h. DNA was extracted with phenol/chloroform and chloroform. The aqueous phase was precipitated with two volumes of 100% ethanol and 1/10 volume of 3 M sodium acetate for 30 min on ice. The DNA pellet was then washed with 70% ethanol and resuspended in 50 l of Tris-EDTA buffer. The absorbance of the DNA solution at 260 and 280 nm was determined by spectrophotometry. The extracted DNA (40 g/lane) was subjected to electrophoresis on 2% agarose gels. The gels were stained with ethidium bromide and then photographed.
Transfection and Luciferase Assays-IB␣M expression plasmid and control plasmid containing puromycin-resistant gene were transfected into HCT116 tumor cells by the LipofectAMINE method (Invitrogen) as described (57) and according to the manufacturer's recommendation.
Retroviral Infections of Cell Lines-The CMV-FLAG-IB␣M/puromycin, RelA, and puromycin-alone control retroviruses were generated, and infections were performed as described previously (58). Supernatants containing the viruses were harvested 48 h later, and filtered supernatants were used to infect cells. Forty-eight hours after infection, cells were seeded in a 100-mm dish at a density of 5 ϫ 10 5 cells in medium containing 500 g of puromycin (Clontech). Puromycin-resistant cells were pooled for subsequent analysis.
Northern Blot Analysis and RT-PCR-For Northern blot analysis, total RNA was extracted using the TRIzol reagent (Invitrogen). Fifteen micrograms of RNA was electrophoresed on a 1% denaturing formaldehyde agarose gel, transferred to a nylon membrane in the presence of 20ϫ SSC, and UV cross-linked. The blots were hybridized with human p53, PUMA, or p21 waf1 cDNA probes labeled with [␣-32 P]deoxycytidine triphosphate using a random labeling kit (Roche Applied Science). Equal loading of mRNA samples was monitored by hybridizing the same membrane filter with the cDNA probe for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as described previously (61). Semiquantitative RT-PCR analysis was performed using a pair of PUMA primers (sense primer CCGCCACTGCAGTTAGAGC and antisense primer CATGGTGCAGAGAAAGTCCC) and separated on an 8% polyacrylamide gel.
Western Blot Analysis-Cytoplasmic extracts were prepared as described previously (59,60) and were separated on 10% SDS-PAGE by electrophoresis and transferred onto a polyvinylidene difluoride membrane (Osmonics) electrophoretically. The membrane was blocked with 5% nonfat milk in PBS containing 0.2% Tween-20 and incubated with affinity-purified monoclonal antibodies against p53 (Calbiochem); Ser 20 -phosphorylated p53, which recognizes only Ser 20 phosphorylated p53 (New England Biolabs); ␤-actin and FLAG M2 (Sigma); and rabbit polyclonal antibodies against IB␣ and IB␤ (Santa Cruz Biotechnology). The membranes were washed in PBS containing 0.2% Tween-20 and probed with horseradish peroxidase-coupled secondary goat antirabbit or anti-mouse IgG antibodies (Amersham Biosciences). The Lumi-Light Western blot substrate (Roche Applied Science) was used for detection. For determining half-life for p53, MDAPanc-28/Puro, or MDAPanc-28/IB␣M cells were treated with 50 g/ml of doxycycline for 24 h, followed by cycloheximide (10 g/ml) addition. The cytoplasmic and nuclear protein extracts were isolated 2, 4, and 8 h after addition of cycloheximide. p53 levels were examined by Western blot analysis with the monoclonal antibody against p53 (Calbiochem).
Metabolic Labeling and Immunoprecipitation-Pulse chase experiments with [ 35 S]methionine labeling were carried out by culturing the cells in methionine-free medium including 10% dialyzed fetal calf serum for 1 day, followed by adding 300 Ci of L-[ 35 S]methionine (Amersham Biosciences) in 8 ml of methionine-free medium for 24 h with simultaneous stimulation either with PBS or 50 g/ml of doxycycline, washing with PBS twice, and further incubating in a medium containing 10% fetal calf serum and 10 mM non-labeled methionine for the indicated time. To equilibrate detection of p53, 200 and 25 g of crude cell extracts from PBS-and doxycycline-treated cells were used for p53 immunoprecipitation, respectively, and separated by SDS-PAGE, and visualized by phosphorimaging and autoradiography.

NF-B Activation Induced p53
Activity in Response to Doxycycline Stimulation-To provide a better understanding of the molecular basis of doxycycline-induced cytotoxicity in cancer cells, we first determined the effect of doxycycline on NF-B activation using MDAPanc-28/Puro and MDAPanc-28/IB␣M cells. MDAPanc-28/IB␣M human pancreatic tumor cells were generated by pooling puromycin-resistant cells after infection with a retrovirus with or without expression of a FLAG-tagged, phosphorylation-defective mutant of IB␣ (IB␣M) (23). These results showed that the expression of IB␣M efficiently inhibits not only constitutive or high basal NF-B activity but also cytokine-dependent NF-B activation in MDAPanc-28 cells (Fig. 1, A-C). The B-DNA binding specificity and subunit composition were confirmed by competition, supershifting, and B reporter gene assays, (data not shown). The dose-dependent activation of NF-B and p53 by doxycycline are shown in Fig.  1D. These results suggested that 20 g/ml of doxycycline induced a detectable NF-B and p53 activities and 40 and 60 g/ml of doxycycline stimulated the peak level of NF-B and p53 activities. As shown in Fig. 1E, DNA binding activity of the RelA/p50 heterodimer was induced at 12 h and lasted up to 24 h in MDAPanc-28/Puro cells (Fig. 1E, lanes 1-3); the expression of IB␣M inhibited doxycycline-induced NF-B activation in MDAPanc-28/IB␣M cells (Fig. 1E, lanes 4 -6). Additional experiments for time-dependent NF-B activation by doxycycline showed that NF-B activity was barely detectable at 8 h (data not shown). Consistent with NF-B activation, the level of IB␣ protein, but not the level of IB␣M protein, was substantially decreased at 12 h and the effect was persistent until 24 h after doxycycline stimulation (Fig. 1F). Collectively, these results suggested that doxycycline activated NF-B.
Since both p53 and NF-B are activated in response to various anticancer agents such as DNA-damaging agents and UVand ␥-irradiation (62-66), we next investigated the p53-DNA binding activity with a specific p53 probe containing the p53 binding site of the p21 waf1 promoter using MDAPanc-28/Puro and MDAPanc-28/IB␣M cells in the presence and absence of doxycycline stimulation. While the basal p53 DNA binding activity in unstimulated MDAPanc-28/Puro was very low, p53 DNA binding activity was induced after 12 and 24 h of doxycycline stimulation (Fig. 1G, lanes 1-3) but was undetectable in MDAPanc-28/IB␣M cells (Fig. 1G, lanes 4 -6). Competition with a wild-type or a control probe (Sp-1) and supershift with anti-p53 antibody showed that the detected DNA binding complex was p53-specific (Fig. 1H). Thus, these results suggested that NF-B activity is required for doxycycline-mediated p53 activation.
Doxycycline-induced Overgeneration of Superoxide Resulted in Activation of NF-B-Doxycycline, which preferentially inhibits mitochondrial protein synthesis, may cause overgeneration of reactive oxygen species by specifically blocking the synthesis of the enzyme complexes involved in the electron transport chain (53,54). To test this notion, we determined the levels of superoxide at 0, 12, and 24 h of doxycycline stimulation. As shown in Fig. 2, A-C, doxycycline stimulation increased superoxide in both MDAPanc-28/Puro and MDAPanc-28/IB␣M cells. Although it is well documented that superoxide activates NF-B (67-69), we verified our findings. As shown in Fig. 2D, NAC inhibited constitutive and doxycycline-induced NF-B activation, while protein synthesis inhibitor cycloheximide has little effect on doxycycline-induced NF-B activation. These results suggested that the increase of superoxide may be involved in doxycycline-induced activation of NF-B in MDAPanc-28/Puro cells, but doxycycline-mediated NF-B activation is inhibited in MDAPanc-28/IB␣M cells (Fig. 1D).
NF-B Is Required for Doxycycline-induced Apoptosis-To determine whether expression of the p53 downstream target genes is inhibited by blocking NF-B activation, we performed Northern blot analysis to compare the expression of p21 waf1 and PUMA in doxycycline-stimulated MDAPanc-28/Puro and MDAPanc-28/IB␣M cells. The results showed that the expression of both p21 waf1 and PUMA was induced after doxycycline stimulation (Fig. 3, A and B). The basal and doxycycline-induced p21 waf1 mRNA levels and doxycycline-induced PUMA expression were almost completely inhibited in MDAPanc-28/ IB␣M cells (Fig. 3, A and B, lanes 4 -6). These results demonstrate that doxycycline induces p21 waf1 and PUMA expression and suggest that doxycycline-induced expression of these p53 target genes is NF-B-dependent.
Since doxycycline-mediated induction of p53 activity and p53 downstream target gene expression depends on NF-B activity, we sought to determine whether inhibition of NF-B activity augmented or suppressed doxycycline-mediated cell cycle arrest or apoptosis. First, cell cycle profiles of these cells were assayed by flow cytometry analysis. Results show that no significant difference was observed between unstimulated MDAPanc-28/Puro and MADPanc-28/IB␣M cells and doxycycline-induced NF-B activity plays a role in inducing G 1 /S arrest in doxycycline-treated MDAPanc-28/Puro cells (Fig. 3C). We also found that doxycycline stimulation significantly reduced the number of viable MDAPanc-28/Puro cells, but not of viable MDAPanc-28/IB␣M cells, suggesting that NF-B activity may be required for doxycycline-mediated cytotoxicity (Fig. 3D).
To determine whether doxycycline-mediated cytotoxicity is caused by induction of apoptosis in these cells, a DNA fragmentation assay was performed by agarose gel electrophoresis. As shown in Fig. 3E, DNA fragmentation occurred from 48 to 96 h after doxycycline stimulation in MDAPanc-28/Puro cells (Fig.  3E, lanes 3-5). Interestingly, DNA fragmentation was detected only at 96 h after doxycycline stimulation in MDAPanc-28/ IB␣M cells (Fig. 3E, lane 11). These results suggest that doxycycline-induced DNA fragmentation was delayed by overexpression of IB␣M, and that NF-B activity is involved to initiate proapoptotic signaling in doxycycline stimulation.
p53 Activation by NF-B Is Involved in Doxycycline-mediated Apoptosis-To determine whether p53 activity is required for doxycycline-induced apoptosis, we carried out our analyses using HCT116p53 ϩ/ϩ and HCT116p53 Ϫ/Ϫ cells (57). Fig. 4, A and  B show that the expression of p53 mRNA and protein was only detected in HCT116p53 ϩ/ϩ cells. Doxycycline stimulation sub-stantially reduced the number of viable HCT116p53 ϩ/ϩ cells to a greater extent than that in HCT116p53 Ϫ/Ϫ cells, suggesting that p53 activity is important for doxycycline-mediated cytotoxicity (Fig. 4C). Doxycycline induced NF-B activation in both HCT116p53 ϩ/ϩ and HCT116p53 Ϫ/Ϫ cells, as shown in Fig.  4D. Doxycycline-induced p53 activity and expression of p53 and PUMA are detected only in HCT116p53 ϩ/ϩ , not in HCT116p53 Ϫ/Ϫ cells (Fig. 4, D and E). Together, these results indicated that doxycycline-induced NF-B activation did not require p53 activity, whereas doxycycline-induced p53 activation required NF-B activity.
To determine whether doxycycline-induced NF-B activity was required to initiate proapoptotic signaling in HCT116p53 ϩ/ϩ cells, we expressed IB␣M in HCT116p53 ϩ/ϩ cells by liposome-mediated transient transfection with greater than 85% transfection efficiency (data not shown). The expression of IB␣M almost completely suppressed doxycycline-induced NF-B and p53 activation (Fig. 4F), and inhibited the doxycycline-induced expression of p53 and its downstream target gene, PUMA (Fig. 4G). DNA fragmentation assay demonstrated that doxycycline-induced DNA fragmentation only occurred in HCT116p53 ϩ/ϩ cells (Fig. 4H, lanes 3-5), whereas doxycycline-induced DNA fragmentation was significantly delayed in HCT116p53 Ϫ/Ϫ cells (Fig. 4H, lane 10) and in HCT116p53 ϩ/ϩ cells expressing IB␣M (Fig. 4H, lane 16). These results suggested that NF-B plays a key role in initiating apoptosis by activating p53 in doxycycline-stimulated cells.
Doxycycline-induced p53 Activation Is Dependent on NF-B Activation-To confirm our finding that doxycycline-inducible p53 activity was dependent on NF-B activity, we examined the DNA binding activities of NF-B and p53 in wild-type, ikk1 Ϫ/Ϫ , ikk2 Ϫ/Ϫ , and rela Ϫ/Ϫ murine fibroblasts (11,12,14). As shown in Fig. 5A, both NF-B-and p53-DNA binding activities were induced by doxycycline in wild-type fibroblasts, but not in ikk1 Ϫ/Ϫ , ikk2 Ϫ/Ϫ , and rela Ϫ/Ϫ fibroblasts, suggesting that IKK1, IKK2, or RelA is required for doxycycline-mediated activation of NF-B and that doxycycline-induced p53 activation depends on NF-B function. To further demonstrate that NF-B induces p53 activation, we infected rela Ϫ/Ϫ murine fibroblasts with retrovirus encoding a RelA or puromycin-resistant gene as the control. Our results showed that re-expression of RelA transiently increased the level of p53 protein and activated p53 DNA binding activity, which induced p21 waf1 expression (Fig. 5B). Collectively, our data revealed that doxycycline-stimulated NF-B activation induced p53 activity in MDAPanc-28 pancreatic cancer cells, HCT116 colon cancer cells, and murine embryonic fibroblasts.
The Stability of p53 Is NF-B-dependent in Doxycycline-stimulated Cells-To determine whether NF-B-dependent p53 activation is induced by transcriptional regulation or posttranscriptional modifications, Northern blot analysis was performed. The results showed that doxycycline-stimulated ulated with 50 g/ml of doxycycline for the indicated times were subjected to NF-B EMSA. An Oct-1 probe was used as the control. NS, nonspecific bands; FP, free probe. F, doxycycline induces IB␣ protein degradation. The cytoplasmic extracts prepared from E were subjected to Western blot analysis using anti-IB␣ polyclonal antibodies as indicated. The membrane was reprobed for expression of ␤-actin as a loading control. G, doxycycline-induced p53 DNA binding activity requires NF-B activation. Nuclear extracts used in E also were analyzed by EMSA with a p53 probe corresponding to the p53-binding site of the human p21 waf1 promoter. An Oct-1 probe was used as a loading control for the nuclear extracts, and doxycycline-induced p53 DNA binding activity is indicated.  (Fig. 6A). In contrast, the basal level of p53 mRNA expression was reduced and doxycycline-induced p53 expression was attenuated in MDAPanc-28/IB␣M cells (Fig. 6A, lanes 4 -6). These results suggest that the doxycycline-induced expression of p53 mRNA may be in part NF-Bdependent. Western blot analysis performed with nuclear extracts using an anti-p53 monoclonal antibody demonstrated that a large increase in the level of p53 protein was observed following doxycycline stimulation of MDAPanc-28/Puro cells (Fig. 6B, lanes 1-3). In contrast, the level of p53 protein was undetectable in MDAPanc-28/IB␣M cells (Fig. 6B, lanes 4 -6). Since the increase in the level of p53 mRNA alone, as shown in Fig. 5A, cannot explain the doxycycline-induced high levels of p53 protein in MDAPanc-28/Puro cells, suggesting that additional regulatory steps such as post-translational stabilization of p53 protein are possibly involved in doxycycline-induced p53 activation.
To examine the stability of p53 protein in doxycycline-stimulated MDAPanc-28/Puro cells, we directly measured its halflife in MDAPanc-28/Puro and MDAPanc-28/IB␣M cells. Fig.  6, C and D show that the half-life of p53 protein in MDAPanc-28/Puro cells was about 20 min, whereas in doxycycline-stimulated MDAPanc-28/Puro cells, it was much longer (Fig. 6C,   lane 9). The level of p53 protein was barely detectable in MDAPanc-28/IB␣M cells (Fig. 6C, lanes 4 -6), even though 8 times as much protein extracts (200 g) from unstimulated cells were loaded (Fig. 6C, lanes 1-6). To further analyze the stability of p53 protein in doxycycline-stimulated MDAPanc-28 cells, we directly measured its half-life in MDAPanc-28/Puro and MDAPanc-28/IB␣M cells in the presence of the protein synthesis inhibitor cycloheximide. The stability of p53 protein was greatly increased and its half-life in doxycycline-stimulated MDAPanc-28/Puro cells is more than 8 h (Fig. 6E, lanes  9 -12 and Fig. 6F,). No change in the level of p53 protein was detected in the extracts from MDAPanc-28/IB␣M cells with or without doxycycline stimulation (To aid the detection of the extremely low level of p53 protein, 4 times as much protein extracts were loaded to gels) (Fig. 6E, lanes 5-8 and lanes  13-16). Thus, our results showed that doxycycline-induced p53 expression and increased p53 stability are NF-B-dependent.
NF-B-dependent Reduction in the Level of Hdm2 Proteins May Be Involved in Stabilization of p53-To identify a possible mechanism for p53 stabilization, we examined the level of Hdm2 protein from MDAPanc-28/Puro and MDAPanc-28/ IB␣M cells stimulated with doxycycline at the different time points. Fig. 7A shows the locations of the epitopes for the anti-HDM2 monoclonal antibodies used in the immunoblotting.  (Fig. 7, B and C). Interestingly, the level of a 55 kDa Hdm2 protein, known to be identified by anti-HDM2 antibody (M7815), increased by doxycycline stimulation as the 90 kDa Hdm2 protein decreased in doxycycline-stimulated MDAPanc-28/Puro cells (Fig. 7B). The 55 kDa Hdm2 protein was not detected by anti-HDM2 antibody (M7815) in doxycycline-stimulated MDAPanc-28/IB␣M cells (Fig. 7B). Another anti-HDM2 antibody (M4308), which recognizes the N terminus of Hdm2 protein, did not detect the 55 kDa Hdm2 protein in both doxycycline-stimulated MDAPanc-28/Puro and MDAPanc-28/IB␣M cells (Fig. 7C). These results suggested that NF-B-dependent p53 stabilization involves the reduction of the level of 90 kDa Hdm2 protein, a key p53 regulator, which functions as an ubiquitin E3 ligase to promote p53 degradation.
Because phosphorylation of p53 is a major post-translational modification that increases p53 stability and activity, we determined whether phosphorylation of p53 is induced by doxycycline. As shown in Fig. 8A, p53 phosphorylation at Ser 20 was induced by doxycycline stimulation in MDAPanc-28/Puro cells. IB␣ degradation, a key step in NF-B activation, was induced by doxycycline in both HCT116p53 Ϫ/Ϫ and HCT116p53 ϩ/ϩ cells (Fig. 8B), and doxycycline-induced p53 phosphorylation on Ser 20 was not detected in HCT116p53 ϩ/ϩ cells expressing a transfected IB␣M (Fig. 8C), possibly because of the lack of detectable p53 protein. Together, these results appeared to suggest that doxycycline-induced phosphorylation of Ser 20 on p53 significantly increase the level of p53, thus extending its activity to initiate proapoptotic signaling cascades. However, it is unclear whether doxycycline-induced activation of NF-B and Ser 20 phosphorylation on p53 are regulated independently or Ser 20 phosphorylation on p53 is regulated by a NF-Binduced kinase.
Proteasome Inhibitor Abrogates Both NF-B and p53 Activation-PS-341, a proteasome inhibitor, blocks NF-B activation by preventing degradation of IB␣ proteins (70,71), and it may also increase the stability of p53 by inhibiting proteasome. Therefore, we used it at a nontoxic concentration to test whether the inhibition of NF-B activity resulted in suppression of p53 expression and activity. MDAPanc-28/Puro cells were treated with 100 nM of PS-341 for specified periods. The results showed that DNA binding activity of both RelA/p50 heterodimer and p53 was reduced after 12 h of PS-341 treatment and that the inhibitory effect of PS-341 on the activation of these transcription factors was further enhanced in a timedependent manner (Fig. 9A). The expression of p53 and p21 waf1 was down-regulated in a time-dependent manner after PS-341 treatment (Fig. 9, B and C), and these results are consistent with the inhibition of NF-B-and p53-DNA binding activity. Fig. 9D shows a PS-341 dose-dependent inhibition of doxycycline-mediated apoptosis in MDAPanc-28/Puro cells. Thus, our results suggested that pharmacological proteasome inhibitors,

FIG. 3. Doxycycline-inducible expression of p53 downstream target genes p21 waf1 and PUMA is NF-B-dependent.
A, induction of p21 waf1 by doxycycline is dependent on NF-B activity. RNA was isolated from MDAPanc-28/ Puro (lanes 1-3) and MDAPanc-28/ IB␣M (lanes 4 -6) cells stimulated with 50 g/ml of doxycycline for the indicated times and analyzed by Northern blotting with a human p21 waf1 cDNA probe; the same blot was washed and rehybridized to a gapdh cDNA probe. B, induction of PUMA by doxycycline is also dependent on NF-B activity. Northern blot analyses were carried out to determine PUMA expression in the doxycycline-stimulated MDAPanc-28/Puro (lanes 1-3) and MDAPanc-28/IB␣M (lanes 4 -6) cells using the same RNA from 3A. C, doxycycline-induced cell cycle arrest is NF-B-dependent. MDAPanc-28/Puro and MDAPanc-28/IB␣M cells were incubated with 50 g/ml of doxycycline for the indicated times and were collected for fluorescence-activated cell sorting analysis. D, IB␣M inhibits doxycycline-mediated cytotoxicity. MDAPanc-28/Puro (Puro) and MDAPanc-28/IB␣M (IB␣M) cells were incubated with 50 g/ml of doxycycline. At each indicated interval, cells were collected and their survival was determined as described under "Experimental Procedures." E, doxycycline-induced apoptosis is dependent on NF-B activity. MDAPanc-28/Puro (lanes 1-5) and MDAPanc-28/ IB␣M (lanes 7-11) cells were treated with 50 g/ml of doxycycline for the indicated times, and DNA fragmentation assays were performed. Lane 12 is a DNA marker. when used as adjuvant therapy for suppressing NF-B mediated antiapoptotic responses, may lead to inhibition of wildtype p53 activation, expression of its downstream target genes, and p53-mediated apoptosis. DISCUSSION Apoptosis is regulated by the integration of both proapoptotic and antiapoptotic signals. Why environmental stress or therapeutic agents induce apoptosis in some cancer cells but not in others is unclear. Therefore, delineating proapoptotic and antiapoptotic signaling cascades should provide insight for the control mechanisms of apoptosis. NF-B and a number of its downstream target genes have been demonstrated to play a critical role in modulating resistance to apoptosis in response to many stimulations and cancer therapeutic agents (6,72). Thus, inhibitors of NF-B activation have a potential use in overcoming the resistance to apoptosis induced by various anticancer agents (73). However, accumulating evidence show apoptosis-promoting functions of NF-B, although how NF-B functions as a proapoptotic factor and which downstream target genes are induced by NF-B activation to initiate proapo- FIG. 4. p53 function is essential in doxycycline-mediated apoptosis. The p53 status of the HCT116 cells used was verified by Northern (A) and Western blot (B) analyses. C, doxycycline-mediated cytotoxicity is p53-dependent. HCT116p53 ϩ/ϩ and HCT116p53 Ϫ/Ϫ cells were incubated with 50 g/ml of doxycycline. At each indicated time interval, cells were collected, and the number of viable cells was determined as described under "Experimental Procedures." D, doxycycline induces NF-B activation in HCT116p53 ϩ/ϩ and HCT-116p53 Ϫ/Ϫ cells. NF-B, p53, and Oct-1 EM-SAs were performed using the nuclear extracts isolated from HCT116p53 ϩ/ϩ and HCT116p53 Ϫ/Ϫ cells stimulated with doxycycline at the various time intervals indicated. E, doxycycline-induced expression of PUMA is p53-dependent. RNA isolated from HCT116p53 ϩ/ϩ and HCT116p53 Ϫ/Ϫ cells stimulated with doxycycline for various intervals as indicated were analyzed by Northern blotting using p53, PUMA, and gapdh probes. F, IB␣M completely suppresses doxycycline-induced NF-B and p53 activation. NF-B, p53, and Oct-1 EMSAs were performed using the nuclear extracts isolated from HCT116p53 ϩ/ϩ cells transfected with a control vector (CTL) and with IB␣M expression plasmid (IB␣M) were stimulated with doxycycline for the times indicated. G, IB␣M inhibits doxycycline-induced expression of p53 and PUMA. RNA was isolated from HCT116p53 ϩ/ϩ cells (lanes 1-3) and HCT116p53 ϩ/ϩ cells transfected with IB␣M expression plasmid (lanes 4-6), which were stimulated with 50 g/ml of doxycycline for the indicated times and analyzed by Northern blotting with p53 and PUMA cDNA probes. The same blot was washed and rehybridized to a gapdh cDNA probe. H, IB␣M and lack of wild-type p53 inhibit doxycycline-induced apoptosis in HCT116p53 ϩ/ϩ cells. HCT116p53 ϩ/ϩ cells (lanes 1-5), HCT116p53 Ϫ/Ϫ cells (lanes 6-10), and HCT116p53 ϩ/ϩ cells transfected with IB␣M (lanes 12-16) were stimulated with 50 g/ml of doxycycline for the indicated times, and DNA fragmentation assays were performed. Lanes 11 and 17 are DNA size markers.
ptotic signaling was unclear. Our findings from this study suggest that a mechanism by which NF-B regulates proapoptotic signaling cascades. In response to superoxide stimulation, NF-B is induced which in turn activate p53-dependent apoptotic pathways.
We investigated the role of NF-B and p53 in doxycyclinemediated apoptosis and have demonstrated the following: (i) doxycycline increases superoxide generation and subsequently activates NF-B; (ii) doxycycline-induced p53 activation is inhibited by IB␣M overexpression and the lack of RelA, IKK1, or IKK2; (iii) re-expression of RelA/p65 induces p53 activation in relA Ϫ/Ϫ fibroblasts and superoxide-inducible NF-B activation is not inhibited in HCT116p53 Ϫ/Ϫ cells; (iv) superoxide-induced p53 activation leads to the expression of its downstream target genes p21 waf1 and PUMA; (v) overexpression of IB␣M or loss of wild-type p53 function postponed doxycycline-mediated apo-p53, 25 g of protein samples from MDAPanc-28/Puro cells and 100 g from MDAPanc-28/B␣M were used for Western blots with monoclonal antibodies for p53. F, graphic representation of p53 half-lives as determined by an image analyzer. Levels of p53 were calculated as a percentage of the 0-h time interval.  1-3) and MDAPanc-28/IB␣M (lanes 4 -6) cells were stimulated with doxycycline for the indicated times. A, RNA isolated from MDAPanc-28/Puro (Puro; lanes 1-3) and MDAPanc-28/IB␣M (IB␣M; lanes 4 -6) cells stimulated with 50 g/ml of doxycycline for the times indicated was analyzed by Northern blotting using p53 and gapdh probes. B, nuclear extracts were isolated from MDAPanc-28/Puro and MDAPanc-28/IB␣M cells, stimulated with 50 g/ml of doxycycline for the indicated times, and subjected to Western blot analysis using an anti-p53 monoclonal antibody. The membrane was then reprobed for expression of ␤-actin for a loading control. C, doxycycline-mediated increase of p53 protein stability is NF-B-dependent. Panc-28/Puro and Panc-28/IB␣M cells were unstimulated or were stimulated with 50 g/ml of doxycycline for 24 h, followed by the pulse-and-chase experiments for indicated times. To equilibrate the detection of p53, 25-g and 200-g aliquots of the crude cell extracts from the cells with or without doxycycline stimulation, respectively, were used for p53 immunoprecipitation. D, graphic representation of p53 half-lives as determined by phosphorimaging. E, MDAPanc-28/ Puro and Panc-28/IB␣M cells were treated with either PBS or 50 g/ml of doxycycline for 24 h. After treatment, protein synthesis was blocked by cycloheximide, then nuclear proteins were isolated 2, 4, and 8 h after the addition of cycloheximide and analyzed by Western blotting with monoclonal antibodies to p53. To equilibrate the detection of Our results, summarized in Fig. 9E, reveal a possible mechanism by which NF-B regulates proapoptotic signaling cascades by activating p53-dependent apoptotic pathways in doxycycline-induced cell death. Our study also suggests that NF-B inhibitor, when used as adjuvant therapy, can block anticancer agent-induced wild-type p53 activity and apoptosis, and that the therapeutic response may be effectively enhanced by using chemotherapeutic regimens and adjuvants to target the appropriate genetic defects in cancers.
Previous studies suggested that oxidative stress is involved in induction of NF-B activation . However, several report showed that H 2 O 2 -induced NF-B activation is highly cell typedependent, which may reflect difference in ROS metabolism (74). Tetracycline is known for preferential inhibition of mitochondrial protein synthesis and decreases the level of cytochrome c oxidase, the key components of electron transport chain (53,54). It was thought that the reduction of the synthesis of cytochrome c oxidase may lead to a disruption of electron transport function and lead to electron leakage from the respiratory chain to O 2 , thus resulting elevated levels of superoxide radicals. Our results supported this notion and showed that doxycycline induced superoxide formation, which may in turn induce NF-B activation. Recently, Hayakawa et al. (75) showed that a commonly used antioxidant, NAC, inhibited TNF-induced NF-B activation independently of antioxidative FIG. 8. Phosphorylation of p53 protein is NF-B-dependent. A, nuclear extracts from MDAPanc-28/Puro cells stimulated with or without 50 g/ml of doxycycline for the indicated times were subjected to immunoprecipitation with an anti-p53 monoclonal antibody to obtain similar level of p53 proteins and followed by Western blot analysis using an anti-Ser 20 phosphorylated p53-specific monoclonal antibody. B and C, HCT116p53 Ϫ/Ϫ , HCT116p53 ϩ/ϩ cells, and HCT116p53 ϩ/ϩ cells transfected either with a control plasmid or IB␣M expression plasmid, were stimulated with 50 g/ml of doxycycline for the indicated times. The extracts from these cells as indicated were subjected to Western blot analysis using an anti-Ser 20 phosphorylated p53-specific monoclonal antibody and an anti-IB␣ antibody. Bottom, the membrane was reprobed with an anti-␤-actin antibody as a loading control.
FIG. 9. Proteasome inhibitor PS-341 inhibits apoptosis. A, PS-341 inhibits NF-B and p53-DNA binding activity. MDAPanc-28/Puro cells were treated with 100 nM PS-341 for the indicated times. After treatment, nuclear protein extracts were isolated and subjected to EMSA analysis using NF-B and p53 probes. B and C, PS-341 inhibits p53 expression. MDAPanc-28/Puro cells were treated with 100 nM PS-341 for the times indicated. Whole cell extracts were analyzed by Western blotting using anti-p53 and anti-␤-actin antibodies, and total RNA was isolated and analyzed by Northern blotting with human p53 and p21 waf1 cDNA probes; the same blot was washed and rehybridized to a gapdh cDNA probe. D, PS-341 inhibits doxycycline-induced apoptosis. MDAPanc-28/Puro cells were stimulated with 40 g/ml of doxycycline in the absence (lane 2) or presence of PS-341 at the indicated concentrations (lanes 3, 4, and 5) for 24 h, and DNA fragmentation assays were performed. E, proposed model for NF-B-regulated proapoptotic signaling cascade. In response to chemotherapeutic agent doxycycline, the transcription factor NF-B is required to activate wildtype tumor suppressor p53 activity through stabilization of p53, possibly by down regulating the expression of Hdm2, and inducing a kinase to phosphorylate p53. p53 in turn induces the expression of its downstream target genes, such as PUMA, to initiate proapoptotic signaling. Doxycycline-induced apoptosis is inhibited/or delayed in the cells expressing IB␣M or lacking IKK, RelA, or p53, suggesting that NF-B may function as a proapoptotic factor by activating p53 signaling cascades. function by lowing the affinity of TNF receptor to its ligand. The recent finding raised question about the specificity of NAC as antioxidant and the role of ROS as a mediator for TNFinduced NF-B activation. Therefore, whether or not NAC functioned as an antioxidant in blocking doxycycline-induced NF-B activation is unclear. The role of doxycycline-induced superoxide in activation of NF-B remains an ongoing study.
Transcription factor NF-B can regulate both pro-and antiapoptotic signaling pathways; however, less is known about the mechanism by which NF-B induces apoptosis. Kasibhatla et al. (24,25) showed that activation of the two transcription factors NF-B and AP-1 is crucially involved in FasL expression induced by etoposide, teniposide, and UV irradiation. There are a number of reports described that show the NF-Bmediated proapoptosis involves up-regulation of p53 (76 -78). For example, NF-B may promote an apoptotic response in striatal medium-sized neurons to excitotoxic insult through up-regulation of c-Myc and p53 (76). However, it is unclear how p53 is up-regulated by NF-B activation. Ryan et al. (79) reported that expression of p53 induced NF-B activation in Saos-2 cell line transfected with a tetracycline-inducible p53 expression vector. In our analysis, doxycycline-induced NF-B activation is not inhibited in HCT116p53 Ϫ/Ϫ colon cancer cells. Activation of p53 completely inhibited in MDAPanc28 pancreatic cancer cells and HCT116 colon cancer cells expressing IB␣M and in IKK-or RelA-deficient mouse fibroblasts. Reexpression of RelA in relA Ϫ/Ϫ mouse fibroblasts induces p53 activity and expression of its downstream target gene p21 waf1 , further indicating that NF-B activation induces p53 in response to ROS. It is possible that this difference may be due to the different cell lines and the amount of inducing agents used in the experiments. Similarly, the difference in growth inhibition between MDAPanc-28/IB␣M and HCT116p53 Ϫ/Ϫ cells may be cell-specific. It is also possible that p63 and p73, the members of the tumor suppressor p53 family, may partially compensate the p53 function in HCT116p53 Ϫ/Ϫ cells. Our results are consistent with a recent report that the NF-B signaling cascade is a potential modulator of p53 activity in response to chemotherapeutic agents (48), even though the opposite effects of NF-B activation on p53 stability have been observed. In comparison to wild-type fibroblasts, ikk1/2 Ϫ/Ϫ fibroblasts, in which activation of NF-B is defective, showed increased p53 stability and cell death in response to doxorubicin (48). In our study, doxycycline induces NF-B activation, which in turn activates p53 activity. The difference in the stability of p53 protein mediated by NF-B activation may reflect the different cellular targets that the two well-known chemotherapeutic agents, doxycycline and doxorubicin, act on. Doxycycline inhibits mitochondrial protein synthesis, which may cause overgeneration of reactive oxygen species by specifically blocking the synthesis of the components in the enzyme complexes involved in the electron transport chain such as cytochrome c oxidase, whereas doxorubicin damages DNA by multiple mechanisms, including DNA intercalating and topoisomerase inhibition, and also generates reactive oxygen species (80). The multiple cellular targets of doxorubicin suggest a possible explanation for the simultaneous activation of both NF-B and p53. Our study showed that doxycycline induced NF-B activation, which in turn activates p53 activity, mainly by increasing its half-life possibly through down-regulation of Hdm2, expressing shorter form of Hdm2 lacking the N terminus, and inducing phosphorylation of p53 at Ser 20 . However, the mechanisms for NF-B-dependent regulation of Hdm2 expression remain unclear, and an NF-B-regulated kinase for phosphorylation of p53 has not been identified.
In summary, doxycycline-induced apoptosis is inhibited in cells expressing IB␣M or lacking IKK, RelA, and p53, and doxycycline-induced p53 activation is blocked in cells lacking functional NF-B activity. In responding to doxycycline, transcription factor NF-B is required to activate the tumor-suppressor activity of wild-type p53, primarily by reducing Hdm2 full-length protein to stabilize p53, and by inducing a kinase to phosphorylate p53 protein at Ser 20 , which in turn induces the expression of its downstream target genes, such as PUMA, for initiating proapoptotic signaling. Thus, our results suggest that a mechanism by which NF-B may function as a proapoptotic factor is to activate the p53 signaling pathway.