Role of NF-κB and Myc Proteins in Apoptosis Induced by Hepatitis B Virus HBx Protein

Abstract

Chronic infection with hepatitis B virus (HBV) promotes a high level of liver disease and cancer in humans. The HBV HBx gene encodes a small regulatory protein that is essential for viral replication and is suspected to play a role in viral pathogenesis. HBx stimulates cytoplasmic signal transduction pathways, moderately stimulates a number of transcription factors, including several nuclear factors, and in certain settings sensitizes cells to apoptosis by proapoptotic stimuli, including tumor necrosis factor alpha (TNF-α) and etopocide. Paradoxically, HBx activates members of the NF-κB transcription factor family, some of which are antiapoptotic in function. HBx induces expression of Myc protein family members in certain settings, and Myc can sensitize cells to killing by TNF-α. We therefore examined the roles of NF-κB, c-Myc, and TNF-α in apoptotic killing of cells by HBx. RelA/NF-κB is shown to be induced by HBx and to suppress HBx-mediated apoptosis. HBx also induces c-Rel/NF-κB, which can promote apoptotic cell death in some contexts or block it in others. Induction of c-Rel by HBx was found to inhibit its ability to directly mediate apoptotic killing of cells. Thus, HBx induction of NF-κB family members masks its ability to directly mediate apoptosis, whereas ablation of NF-κB reveals it. Investigation of the role of Myc protein demonstrates that overexpression of Myc is essential for acute sensitization of cells to killing by HBx plus TNF-α. This study therefore defines a specific set of parameters which must be met for HBx to possibly contribute to HBV pathogenesis.

The transcription factor NF-κB is involved in a number of cellular processes, including immune cell activation and development, stress responses, expression of inflammatory cytokines, and the control of apoptosis (5). NF-κB transcription factors are hetero- and homodimer complexes of related proteins which contain a Rel homology domain involved in specific DNA binding, protein dimerization, and nuclear importation (6). The Rel proteins predominantly found in mammalian cells consists of two transcriptionally inactive forms, NF-κB1 (p50) and NF-κB2 (p52), and three transcriptionally active subunits known as RelA (p65), c-Rel, and RelB (6). NF-κB is generally sequestered in the cytoplasm in an inactive state associated with inhibitory IκB proteins (30). Upon phosphorylation and degradation of IκB, NF-κB is released and translocated to the nucleus, where it activates dependent genes.

Different NF-κB transcription factors may play diverse and even opposing roles in modulating cell death by apoptosis. In certain settings, c-Rel has been associated with promoting apoptosis. Increased expression of c-Rel protein and its accumulation in the nucleus correlate with induction of apoptosis in various tissues of developing chicken embryos (1, 63). Overexpression of c-Rel in bone marrow cells triggers apoptosis (47). Apoptosis was also observed upon expression of c-Rel in HeLa cells stably expressing the c-rel gene under inducible control (9). In contrast, a variety of studies with knockout mice have demonstrated the importance of RelA and c-rel in prevention of apoptosis. Apoptosis of the liver occurs in relA knockout mice during embryogenesis (10). In addition, overexpression of RelA/NF-κB also protects cells from tumor necrosis factor alpha (TNF-α)- or chemotherapy-mediated apoptosis (10, 37, 67, 106, 107, 116), and RelA-deficient embryonic fibroblasts die upon treatment with TNF-α, while RelA-containing fibroblasts do not (10). Enforced expression of RelA or c-rel blocks apoptosis induced by a variety of proapoptotic agents, including TNF-α (32, 67). In c-Rel-deficient mice, B cells undergo apoptosis in response to antigen receptor ligation due to an inability to induce the Bcl-2 prosurvival protein, known as protein A1 (38). Studies have found that the ability of NF-κB to block TNF-α-mediated apoptosis is related to its induction of prosurvival (antiapoptotic) genes, including Bcl-2 family proteins (38), and inducible NO synthase genes (36). Metabolites of NO have been linked to inhibition of apoptosis (72). Thus, activation of NF-κB transcription factors in different settings can control apoptosis in quite opposite manners.

Many viruses have regulatory proteins that activate NF-κB. These include cytomegalovirus (119), human immunodeficiency virus type 1 (HIV-1) (31), human T-lymphotropic virus type 1 (HTLV-1) (52, 53, 62, 75), Epstein-Barr virus (EBV) (45), influenza virus (81), Sindbis virus (66), dengue virus (73), and hepatitis B virus (HBV) (18, 68, 89, 94, 104). However, for some viruses the activation of NF-κB is associated with induction of apoptosis. For example, Sindbis A virus-induced apoptosis was found to require activation of NF-κB, as suppression of NF-κB blocked virus-mediated apoptosis (66). Replication of dengue A virus in human hepatoma cells was shown to activate NF-κB, which in turn was linked to apoptotic cell death (73). The HIV-1 envelope glycoprotein gp160 was also shown to induce apoptosis via activation of NF-κB (20). Thus, activation of NF-κB for these viruses induces apoptosis. For other viruses, activation of NF-κB prevents host cell apoptosis. For example, the EBV latent membrane protein-1 (LMP-1) induces NF-κB activity, predominantly as p50-RelA and p52-RelA forms (45). EBV appears to protect B cells from apoptosis, at least in part through upregulating expression of Bcl-2 by LMP-1, probably through NF-κB/RelA activation (42). Some viruses are also capable of simultaneously inducing apoptotic death of their host cells while stimulating NF-κB activity, without any correlation between the two activities. For example, HTLV-1 Tax protein has been studied intensively for its ability to activate NF-κB (15, 53, 75, 86, 97, 112). Tax also induces apoptosis in Jurkat cells when overexpressed (22). However, these studies could find no discernable correlation between Tax-induced NF-κB activity and its induction of apoptosis. The HBx protein of HBV is associated with induction of apoptosis or sensitization of cells to apoptosis by subthreshold levels of proapoptotic stimuli such as TNF-α and etoposide (16, 19, 55, 87, 90, 91, 95, 100), and it slightly enhances hepatocyte killing in the liver of transgenic mice (83, 100, 101). HBx also induces NF-κB through multiple molecular mechanisms, resulting in NF-κB complexes of different compositions (18, 94, 113), although the effects on apoptosis of HBx-induced complexes are not known.

HBx is largely a cytoplasmic protein (25, 27, 49, 92) that moderately activates transcription and is essential for viral replication (17, 121). HBx activates a variety of transcription factors, including NF-κB (65, 68, 71, 76, 89, 104), AP-1 (12, 14, 24, 54, 78, 88, 103), and CREB/ATF2 (8, 70, 99, 115). In addition to RNA polymerase II-dependent transcription, HBx also stimulates transcription by RNA polymerase III (3, 61, 108) and RNA polymerase I (110). Underlying many but not all of the reported activities of HBx is its ability to stimulate cytoplasmic signal transduction pathways (12–14, 18, 23, 24, 27, 54, 58, 69, 79, 94, 95, 108–110). HBx activates the Ras–Raf–mitogen-activated protein kinase pathway (12, 24, 43, 79, 108), the cell stress-induced MEKK1–p38–c-Jun N-terminal kinase (JNK) pathway (14, 43, 94), and the family of Src tyrosine kinases (58), which may be important for viral replication (59). HBx also stimulates deregulation of early cell cycle checkpoint controls (13, 60, 91).

Here we report studies that were conducted to determine whether the composition of HBx-induced NF-κB complexes is important for HBx-mediated apoptosis. There were three reasons for carrying out this study. First, HBx activates different complexes of NF-κB, including RelA/NF-κB and c-rel/NF-κB, which may have opposing effects on cell survival in different settings (32). Second, HBx either directly or indirectly induces apoptosis or sensitizes cells to TNF-α-mediated apoptotic killing in the context of viral replication or when expressed independently of HBV in cultured cells (94). Moreover, TNF-α production occurs during HBV infection (21, 34), and HBx has been reported to stimulate the expression of TNF-α in infected hepatocytes (35, 48, 82). Third, HBx induces myc genes in some cell lines (4, 7, 95), and myc gene expression may be elevated in early neoplastic liver nodules containing HBV or woodchuck hepatitis B virus (WHV) (105, 108). Myc proteins can sensitize cells by about twofold to apoptotic killing in certain settings, such as during exposure to TNF-α or other factors (29, 44, 50, 51, 57) or potent inhibitors in other settings (reviewed in references 32 and 102).

MATERIALS AND METHODS

Cell culture.

Wild type (WT) and RelA/p65 knockout (KO) NIH 3T3 murine cell lines were provided by D. Baltimore (CalTech). 3T3 cells were propagated in Dulbecco's modified Eagle's medium with 10% calf serum and gentamicin sulfate (50 μg/ml). TNF-α (Sigma) was added directly to the medium at 10 ng/ml for the times indicated in the text. Insulin-like growth factor II (IGF-II; Sigma) was added to growth medium at 10 ng/ml as indicated.

Antibodies and plasmids.

Rabbit anti-NF-κB p50, anti-p52, anti-c-Rel, and anti-c-Myc were purchased from Santa Cruz Inc. Anti-NF-κB p65 was a gift from D. Baltimore. Monoclonal anti-goat immunoglobulin G-fluorescein isothiocyanate (FITC) conjugate was purchased from Sigma. Plasmid pCMV-IκBα expresses a superrepressor form of IκBα in which serines 32 and 36 were mutated to alanine (107). The cDNA expression vector for the human c-myc gene controlled by the simian virus 40 (SV40) early promoter was a gift from E. Ziff (New York University). The SV40 early promoter is only slightly stimulated by HBx (68).

Transfection and transduction of cells.

Cells at 50% confluency were transfected with DNA plasmids using Lipofectin (Gibco) with a total of 10 μg of plasmid DNA per 10-cm plate of cells. Where indicated, cells were treated with TNF-α and/or other reagents. Transfection efficiencies were monitored by inclusion of an expression plasmid for green fluorescent protein (pGFP). Ad-HBx and Ad-HBxo constructs express the wild-type HBx gene or a mutant deleted of all AUG codons (subtype ayw), respectively, controlled by the cytomegalovirus promoter in a replication-defective adenovirus type 5 (Ad) vector deleted of Ad early regions E1A and E1B. These vectors were described previously in detail (27). Viral vectors were propagated in complementing 293 cells, and titers were determined. Transduction utilized 100 to 200 virus particles per cell at a 20:1 ratio of particles to PFU (5 to 10 PFU/cell).

Cell killing.

The amount of cell killing was determined as described previously (95) by scoring the fraction of dead or dying cells stained with trypan blue (Sigma) and by Hoechst 33258-stained condensed nuclei. A minimum of 200 cells were scored for each sample, and each experiment was performed in duplicate at least three times. Standard errors were calculated from all three experiments, and results represent the mean of three studies.

Preparation of cytoplasmic and nuclear extracts.

Cytoplasmic and nuclear extracts were prepared as described previously (2) with modifications (94). Cell pellets were resuspended in 400 μl of cold buffer A (10 mM HEPES [pH 7.9], 1.5 mM MgCl₂, 10 mM KCl, 0.5 mM dithiothreitol [DTT], 1 mM phenylmethylsulfonyl fluoride [PMSF], leupeptin [10 mg/ml], and aprotinin [10 mg/ml]), swollen on ice for 10 min, vortexed for 10 s, and centrifuged for 10 s at 12,000 × g. Nuclear pellets were resuspended in 30 to 70 μl of cold buffer B (20 mM HEPES [pH 7.9], 25% glycerol, 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 1 mM DTT, 1 mM PMSF, leupeptin [10 mg/ml], and aprotinin [10 mg/ml]) and incubated on ice for 20 min, and nuclear extracts were obtained by centrifugation at 12,000 × g for 2 min at 4°C.

EMSAs.

Electrophoretic mobility shift assays (EMSAs) were carried out essentially as described elsewhere (93, 94). Double-stranded DNA oligonucleotides for probes or competitors consisted of the following double-stranded sequence corresponding to an NF-κB binding site sequence: 5′-GATCCAGAGGGGCCACTTTCCGAGAGGA-3′. Nuclear extracts for band shift analysis were prepared as described above. All extracts were standardized for protein concentration before use. Binding reactions were carried out in 12- to 20-μl reaction volumes containing 5 μg of nuclear protein extract, 10 fmol of ³²P-5′-end-labeled double-stranded oligonucleotide, approximately 10⁶ cpm/reaction, and 1 μg of poly(dI-dC) in buffer C (5 mM MgCl, 1 mM DTT, 1 μg of single-stranded DNA, 0.3 mM PMSF, 0.5 mM EDTA, 10% glycerol, 10 mM HEPES [pH 7.9], 50 mM NaCl). Binding reactions were incubated for 30 min at 23°C, and DNA-protein complexes were resolved from free-labeled DNA by electrophoresis in 4.5% polyacrylamide gels containing 50 mM Tris-HCl (pH 8.5), 200 mM glycine, and 1 mM EDTA. Gels were dried and autoradiographed, and radioactivity was quantitated by digital imaging software. For antibody competitor assays, specific antibodies against p50, p52, and p65 and c-Rel subunits of NF-κB or preimmune sera were added to the binding reaction prior to addition of labeled oligonucleotide for 15 min at 4°C. Unlabeled competitor assays were carried out by adding a 100-fold molar excess of unlabeled double-stranded DNA NF-κB oligonucleotide simultaneously with labeled probe.

Immunoblot analysis.

For each sample, 50 to 100 μg of nuclear protein was resolved on sodium dodecyl sulfate (SDS)–10% polyacrylamide gels and transferred to nitrocellulose filters (Schleicher and Schuell) using an electroblotting transfer system. Filters were incubated for 2 h in Tris-buffered saline (TBS) containing 5% nonfat dry milk at room temperature and then for 3 h with primary antibody to specific proteins in TBS plus 0.5% sodium azide. Filters were washed three times for 10 min each in TBS–0.2% Tween 20. Filters were then incubated with secondary antibodies in TBS for 1 h. Filters were washed five times for 10 min each in TBS–0.2% Tween 20. Immune complexes were visualized with the enhanced chemiluminescence system (ECL; Amersham).

Immunofluorescence analysis.

Cells were grown on coverslips in dishes and transfected or transduced as above, washed in phosphate-buffered saline after transfection, then fixed or permeabilized with fresh 70% acetone–30% methanol at −20°C for 7 min as described previously (27). Cells were incubated with primary antibodies to Flag (IBI, Inc.) for HBxFlag protein, washed, and then incubated with FITC-conjugated secondary antibody as described (27). Staining with Hoechst 33258 was carried out as described previously (95). Stained cells were examined and photographed using a Zeiss Axiophot fluorescence microscope.

RESULTS

Characterization of HBx-activated NF-κB.

To determine whether NF-κB plays a role in controlling induction of apoptosis by HBx, experiments were conducted in a RelA-KO mouse 3T3 fibroblast cell line (RelA KO), which was derived from embryonic fibroblasts of RelA-KO mice (11). A matched wild-type 3T3 cell line (WT) was derived from the control mice (11). HBx was delivered by transduction from a replication-defective Ad vector that fails to express detectable levels of Ad early and late genes in noncomplementing cells, such as murine 3T3 cells. The Ad-HBx vector, which was described previously, transduces and expresses the HBx gene in ∼95% of the cells (12, 13, 27, 94). The Ad-HBx vectors replicate only in complementing 293 cells that provide Ad E1A and E1B gene products in trans.

HBx was introduced into serum-starved WT or RelA KO 3T3 cells by transduction with 100 to 200 virus particles (5 to 10 PFU) of Ad-HBx vector per cell. Control cells were transduced with a virus vector that expresses a mutated HBx mRNA lacking AUG codons and therefore cannot synthesize HBx protein (Ad-HBxo) (12, 13, 27, 94). Eight hours after vector transduction, when HBx activation of Src-Ras signaling was elevated (58), cells were harvested and nuclear extracts were prepared. Formation of NF-κB DNA-binding complexes was examined by EMSA, using equal amounts of nuclear protein extracts and a ³²P-labeled, double-stranded oligonucleotide DNA probe containing the consensus NF-κB binding site. Induction of NF-κB DNA-binding activity was observed in WT cells expressing HBx but not those expressing HBxo (Fig. 1A). HBx stimulated NF-κB DNA binding to approximately fivefold-lower levels than in cells treated with murine TNF-α (Fig. 1A; 10-fold less TNF-α sample was loaded in this particular gel). In RelA KO 3T3 cells, which still express other forms of NF-κB, both TNF-α and HBx stimulated formation of NF-κB DNA-binding complexes. Formation of NF-κB complexes was ablated by an excess of unlabeled competitor DNA (Fig. 1B), demonstrating specificity.

FIG. 1.

Induction of NF-κB DNA-binding activity by HBx and TNF-α in NIH 3T3 cells. Serum-starved (A) WT 3T3 cells and (B) RelA KO 3T3 cells were transduced by Ad-HBx or Ad-HBxo vector for 8 h or treated with mouse TNF-α (10 ng/ml) for 30 min. Nuclear extracts were prepared, and 10 μg was used to measure NF-κB DNA-binding activity by EMSA with a double-stranded ³²P-labeled oligonucleotide probe containing one NF-κB binding site. For competition studies, a 100-fold molar excess of unlabeled competitor (c. comp.) probe was added. Protein-DNA complexes were resolved by electrophoresis in 4.5% polyacrylamide gels and visualized by autoradiography, and radioactivity was quantitated by digital densitometry. Control cells were neither transduced nor treated. Note that 10-fold less TNF-α sample was loaded in panel A.

The NF-κB DNA-binding complexes activated by HBx in WT and RelA KO cells were characterized by EMSA using NF-κB member-specific antibodies. Polyclonal antibodies directed against p50, p52, RelA, and c-Rel proteins were added to binding reactions prior to addition of ³²P-labeled oligonucleotide DNA probe. Partial or full ablation of specific Rel protein band shifts is diagnostic of DNA binding by that particular protein. In WT 3T3 cells, antibodies to p50, p52, RelA, and c-Rel each partially prevented formation of NF-κB DNA-binding complexes, which was similar in TNF-α-treated HBx-expressing cells (Fig. 2A). In RelA KO cells treated with TNF-α or transduced by Ad-HBx, antibody addition studies demonstrated similar formation of the NF-κB DNA-binding complexes (but induced more strongly by TNFα), containing p50, p52, and c-Rel proteins but not RelA, since addition of antibody to RelA had no effect on complex formation (Fig. 2B). The ablation of p50 protein complexes by p52 antiserum and p52 protein complexes by p50 antiserum is probably due to cross-reactivity (data not shown) and does not alter these conclusions. HBx may also induce low levels of RelB and p105 NF-κB in both cell lines (94) (data not shown). The WT and RelA KO cells therefore provide an excellent model system for investigating the contributions of Rel protein activation by HBx in sensitization or resistance of cells to apoptotic killing.

FIG. 2.

Composition of NF-κB DNA-binding complexes induced by HBx protein and TNF-α in WT and RelA KO 3T3 cells. Serum-starved (A) WT and (B) RelA KO cells were transduced with the Ad-HBx vector for 8 h or treated with mouse TNF-α (10 ng/ml) for 30 min. Nuclear extracts were prepared, and 10 μg of protein extract was used to detect NF-κB DNA-binding activity by EMSA as described in the legend to Fig. 1. Specific polyclonal antibodies (Ab) were added to binding reactions corresponding to p50 (NF-κB1), p52 (NF-κB2), RelA, and c-Rel proteins by addition 15 min prior to that of ³²P-labeled oligonucleotide probe. Control refers to extracts from untreated quiescent cells. n. serum, normal serum.

Inactivation of RelA and c-Rel/NF-κB enhances HBx-induced apoptosis but not sensitization to TNF-α.

To determine the role of NF-κB proteins in HBx-mediated apoptosis and in HBx sensitization to apoptosis by TNF-α, both WT and RelA KO cells were transduced with Ad-HBx or Ad-HBxo vectors for 8 h, followed by treatment with TNF-α (10 ng/ml) for 16 h. Cell death was then quantified by trypan blue exclusion assay, as described previously (95). This assay measures the percentage of dead and dying cells that become permeable to dye, which is a sensitive and quantitative measure of cell death (94). The viability of WT cells expressing HBxo was the same as that of control cells that were not transduced by Ad-HBx or treated with TNF-α (Fig. 3). WT 3T3 cells expressing HBx demonstrated a slight but reproducible increase in killing (∼8%) compared to control HBxo-expressing cells (4 to 5%) (Fig. 3A, Fig. 4; P < 0.05). Surprisingly, WT 3T3 cells expressing HBx were not significantly sensitized to apoptotic killing by treatment with TNF-α (Fig. 3A), in contrast to previous studies in HepG2 and Chang cells (16, 95). Treatment of cells with much higher levels of TNF-α (up to 1 μg/ml) was also without effect (data not shown), indicating that HBx was incapable of sensitizing WT 3T3 cells to TNF-α-mediated killing. In RelA KO 3T3 cells expressing HBx, a twofold increase in cell killing was observed compared to WT 3T3 cells expressing HBx. Treatment with TNF-α only slightly, but reproducibly, increased killing of RelA KO cells expressing HBx (Fig. 3B). HBxo expression had no effect on RelA KO cell viability, as expected. Treatment of control or HBxo-expressing RelA KO cells with TNF-α showed a twofold increase in killing compared to untreated cells. Thus, RelA KO cells, which still express c-Rel and other forms of NF-κB, are only modestly sensitized to killing by HBx plus TNF-α. As a control, cotreatment of WT or RelA KO cells with cycloheximide rendered them acutely sensitive to killing by TNF-α, demonstrating that these cells can be made susceptible to TNF-α and can be induced to apoptose.

FIG. 3.

Cell killing by HBx and TNFα in WT and RelA KO 3T3 cells. Serum-starved (A) WT and (B) RelA KO 3T3 cells were transduced by Ad-HBx or Ad-HBxo vector for 8 h, followed by 16 h of mock treatment or treatment with murine TNF-α (10 ng/ml) or TNF-α (10 ng/ml) and cycloheximide (cycl.) (10 μg/ml). Cell death was quantified spectrophotometrically by trypan blue dye exclusion. Results represent the means of three independent experiments, with derived standard errors shown. Control (Ctrl) refers to cells not transduced by Ad vectors.

FIG. 4.

Combined HBx expression and TNF-α treatment induce death of 3T3 cells by apoptosis. WT and RelA KO 3T3 cells were grown on coverslips, transduced by Ad-HBx, and either mock treated or treated with TNF-α as described in the legend to Fig. 3. Cells were processed for staining with Hoechst DNA dye 33258 and then photographed under UV light using a Zeiss Axiophot photomicroscope at ×600 magnification. Apoptotic cells contain condensed nuclei that stain brightly, compared to diffuse and pale staining of normal cells.

Studies confirmed that the death of WT and RelA KO cells occurred by apoptosis. WT and RelA KO cells grown on coverslips were transduced with Ad-HBx. Although virtually all cells were transduced and express HBx (shown in Fig. 6C), only a small fraction of cells contained condensed and fragmented nuclei, a hallmark of apoptosis, as determined from bright punctate staining with Hoechst 33258 (Fig. 4). Diffuse and pale staining is typical of nuclei in nonapoptotic cells. These data confirm that very few WT 3T3 cells expressing HBx undergo apoptosis, and they indicate that elimination of RelA in RelA KO cells enhances the ability of HBx to mediate apoptotic cell killing only about twofold. Treatment of cells with TNF-α (10 ng/ml) resulted in only a small increase in apoptotic killing of WT and RelA KO cells expressing HBx (Fig. 4), consistent with the results in Fig. 3 and in contrast to other cell lines examined previously.

FIG. 6.

Effect of c-Myc overexpression on HBx-induced cell killing. (A) WT and RelA KO 3T3 cells were transfected with a plasmid that expresses the human c-myc gene under the control of the SV40 promoter, which is only slightly induced by HBx (68). Equal amounts of whole-cell extracts were resolved by SDS-polyacrylamide gel electrophoresis, and c-Myc protein was identified by immunoblotting with a specific antibody. Cell treatment with TNF-α was carried out as described below. (B) WT 3T3 cells were transfected by 1 μg or 10 μg (10×) of the Myc expression plasmid, followed 8 h later by transduction with Ad-HBxo (vec) or Ad-HBx (HBx). At 10 h following transduction, cells were treated with TNF-α (10 ng/ml) for 16 h or mock treated. Cell killing was quantified by trypan blue dye exclusion assay. (C) Cells grown on coverslips were transfected as above, followed by transduction with an Ad-HBx-Flag vector (vec) (27), which expresses HBx protein containing the Flag foreign epitope. HBx-Flag behaves identically to HBx protein (27). Cells were processed for staining with Hoechst 33258 DNA dye, and indirect immunofluorescence was performed using FITC-conjugated M2 anti-Flag antibodies to visualize HBx. Cells were photographed as described in the legend to Fig. 4.

Studies were therefore performed to determine whether expression of other forms of NF-κB affords significant protection against HBx-induced apoptosis and sensitization to killing by TNF-α. Cells were transfected with a plasmid expressing a superrepressor mutant form of IκBα, which cannot be phosphorylated at serine residues 32 and 36 and therefore blocks activation of NF-κB (107). Cells were transduced with the Ad-HBx vector 6 to 8 h after transfection. Transduction by virus drives the uptake of plasmid into a much larger number of cells (13, 120). Cotransfection with a green fluorescent protein expression plasmid indicated that ∼70% of cells were transfected by this approach (data not shown). Transfection of the IκBα superrepressor expression plasmid followed by transduction with Ad-HBx suppressed HBx activation of all forms of NF-κB by about 75% in WT cells and slightly more so in RelA KO cells, as measured by NF-κB EMSA (Fig. 5A). The residual NF-κB DNA-binding complexes likely arise from the 30% of cells that were not transfected but were transduced by the Ad-HBx vector. These data demonstrate downregulation of NF-κB activation in the majority of cells that express HBx. As expected, transfection of the IκBα superrepressor and transduction of cells by the Ad vector alone did not demonstrate activation of NF-κB DNA-binding complexes. We therefore examined whether downregulation of NF-κB is associated with a concomitant increase in cell killing by HBx and an increased sensitivity to TNF-α.

FIG. 5.

Suppression of NF-κB activation promotes HBx-mediated apoptosis. WT and RelA KO 3T3 cells were transfected by a plasmid expressing a superrepressor (S.R.) mutant form of IκBα, which blocks activation of NF-κB. Eight hours later, cells were transduced for 10 h with Ad-HBx or vector alone (Ad dl312, vec), a control virus vector that is functionally identical to control virus AdHBxo. At 10 h following transduction, cells were either mock treated or treated with murine TNF-α (10 ng/ml) for 16 h. Approximately 70% of cells were transfected by this approach, as determined by inclusion of a pGFP expression vector (data not shown). (A) Nuclear extracts were prepared and 10 μg of protein was analyzed for NF-κB DNA-binding complexes using a ³²P-labeled DNA probe and EMSA. Killing of (B) WT 3T3 cells and (C) RelA KO cells was quantified spectrophotometrically by the trypan blue dye exclusion assay. Results represent the means of three independent experiments, with calculated standard errors shown.

WT and RelA KO 3T3 cells were transfected by the IκBα superrepressor expression plasmid or a control plasmid and transduced by Ad-HBx or control Ad vector (Ad dl312). Ad-HBx-transduced cells expressing the IκBα superrepressor were threefold more sensitive to killing than those without the superrepressor or cells transduced by vector alone (Fig. 5B and C). IκBα superrepressor expression resulted in killing of about 20% of WT 3T3 cells expressing HBx, compared to ∼7% without HBx expression. In RelA KO cells expressing HBx, inhibition of NF-κB resulted in ∼30% cell killing compared to cells without HBx or without the repressor (Fig. 5B and C). Thus, NF-κB partially protects cells against HBx-mediated killing. Nevertheless, there was still only a small increase in sensitization to TNF-α-mediated killing in HBx-expressing WT and RelA KO cells.

c-myc gene expression sensitizes HBx-expressing cells to TNF-α-mediated apoptosis.

Sensitization of cells to TNF-α-mediated killing is strongly associated with upregulation of myc gene expression in a variety of cell types (reviewed in reference 32). Moreover, HBx protein can enhance myc gene expression in certain hepatic cell lines (4, 7, 95). In addition, 3T3 cells were shown previously to synthesize low levels of Myc proteins but to become about twofold more sensitive to TNF-α killing if c-Myc protein is overexpressed by transfection (56, 57). We next investigated whether the absence of myc gene expression in 3T3 cells is the basis for the failure of HBx to sensitize these cells to killing by TNF-α.

WT and RelA KO 3T3 cells contain low levels of c-Myc protein, shown by Western immunoblot analysis of equal amounts of whole-cell protein lysate (Fig. 6A, left). Immunoblot analysis of WT 3T3 cells transfected with a c-myc expression plasmid resulted in overexpression of c-Myc protein relative to endogenous levels, as expected. Treatment of c-myc-transfected cells with TNF-α and/or expression of HBx did not alter the level of ectopically expressed c-Myc protein (Fig. 6A, right). About 70% of the cells were transfected in this manner, based on inclusion of a pGFP reporter (data not shown). The expression of c-Myc in WT 3T3 cells transduced by Ad-HBxo (vec samples) resulted in a slight but reproducible increase in killing from 5 to 9% (Fig. 6B; P < 0.05). Treatment of these cells with TNF-α demonstrated a twofold increase in sensitivity to killing compared to a lack of sensitivity to TNF-α in the absence of c-Myc overexpression. Cells transfected by c-Myc and expressing HBx displayed an increase in killing to 15%, compared to 7% in the absence of c-myc transfection. Cells transfected by c-myc and transduced by HBx and then treated for 8 h with TNF-α showed a dramatic increase in killing, representing 65 to 70% of the cells, almost equivalent to the fraction that were transfected (Fig. 6B). Killing was not observed in the vector-alone plus Myc controls that were treated with TNF-α. Thus, the combination of HBx and c-Myc increases the sensitivity of cells to TNF-α killing by about 10-fold. Transfection of cells with 10-fold-higher levels of c-myc expression plasmid, in the absence of HBx expression, did not significantly increase the sensitivity of cells to TNF-α killing compared to the lower level of plasmid (Fig. 6B). These data are consistent with those previously reported by Ueda and Ganem (105), in which it was shown that HBx and N-myc2 coexpression alone was not sufficient to promote apoptosis in a rodent hepatocyte cell line. They are also consistent with results from NIH 3T3 cells demonstrating a moderate (twofold) increase in killing by c-Myc at 16 h of TNF-α treatment (57). Coimaging analysis of HBx, by indirect immunofluorescence staining with anti-Flag antibody and by staining for apoptotic condensed chromatin with Hoechst stain (Fig. 6C), showed that WT 3T3 cells which expressed HBx-Flag also underwent apoptotic killing when treated with TNF-α. The poor resolution of HBx-Flag results from the death of cells and because the conditions for Hoechst dye staining cause poor imaging of HBx (95). In RelA KO cells, the combination of HBx and overexpression of c-Myc in the absence of TNF-α treatment was sufficient to induce killing in 70% of the cells (data not shown). Previous studies demonstrated that coexpression of HBx and treatment of cells with TNF-α do not impair HBx or TNF-α activation of NF-κB (94, 95). These results demonstrate that increased expression of Myc protein strongly sensitizes HBx-expressing cells to enhanced killing by TNF-α. They also demonstrate that cell killing by HBx plus TNF-α is suppressed by activation of NF-κB but can be overridden by a high level of c-Myc protein.

Mitogenic growth factors suppress HBx and c-Myc sensitization of cells to apoptotic killing by TNF-α.

Previous studies showed that many hepatocellular carcinomas derived from WHV and human HBV chronic infections overexpress any of several mitogenic growth factors, particularly IGF-II for WHV and transforming growth factor-α for HBV. It was also shown that a variety of mitogenic growth factors can suppress apoptotic killing by c-Myc protein (33, 41), by N-Myc in WHV-infected cells (105, 118), and by HBx in TNF-α-sensitive cells (95). We therefore determined whether a mitogenic growth factor suppresses the pathway that sensitizes 3T3 cells to TNF-α killing, which is mediated by combined HBx and Myc expression. WT 3T3 cells grown on coverslips were transfected with a c-Myc expression plasmid, infected with Ad-HBx for 16 h, then treated with TNF-α (10 ng/ml) for 8 h, or simultaneously with 10 ng each of TNF-α and IGF-II per ml for 8 h. Cells were examined for induction of apoptosis by Hoechst staining (Fig. 7A). Data were quantified by the dye exclusion assay (Fig. 7B). HBx and c-Myc expression induced only a slight apoptotic killing of cells (intense, bright punctate staining), consistent with results presented above. The slight increase in cell killing by HBx and c-Myc was fully suppressed by cotreatment of cells with IGF-II. Cells expressing HBx and c-Myc and treated with TNF-α underwent a strong increase in apoptosis, accounting for about 80% of the cells, consistent with the results in Fig. 6. Cotreatment of these cells with IGF-II largely but not completely blocked induction of apoptosis. The inability of IGF-II to fully suppress apoptosis by TNF-α in cells expressing HBx and c-Myc might be due to the inability to fully block death signaling by TNF-α or possibly because not all 3T3 cells are strongly responsive to human IGF-II. Nevertheless, these data demonstrate that a mitogenic growth factor can block the sensitization of HBx- and c-Myc-expressing cells to killing mediated by TNF-α.

FIG. 7.

Apoptotic killing of 3T3 cells expressing c-Myc and HBx by TNF-α is blocked by IGF-II. WT 3T3 cells were grown on coverslips, transfected by a human c-myc expression plasmid, and transduced by Ad-HBx, as described in the legend to Fig. 6. Cells were treated with either murine TNF-α (10 ng/ml) for 16 h, or TNF-α (10 ng/ml) and IGF-II simultaneously. (A) Cells were photographed under UV light as described in the legend to Fig. 4. (B) Quantification of cell killing was performed by the dye exclusion assay.

DISCUSSION

HBx protein has been variously reported to be either a promoter or an inhibitor of apoptotic cell death. HBx was shown to induce apoptosis or to sensitize normally resistant cells in culture to apoptotic killing by proapoptotic stimuli, such as etoposide and TNF-α (16, 19, 55, 83, 90, 91, 95). Importantly, HBx also stimulates apoptotic turnover of hepatocytes when expressed in the livers of transgenic mice or in hepatocytes in vivo in a p53-independent manner (100, 101). It is important to point out that while replication of HBV in the livers of transgenic mice is not associated with pathological killing of hepatocytes, these animals express little if any HBx mRNA (39). Other studies support the p53-independent apoptotic killing of cells by HBx (90, 91). Furthermore, studies have found that HBx suppresses transformation of rat embryo fibroblast cells (which contain wild-type p53) by Ras plus c-Myc by promoting apoptotic killing of these cells (87). However, one study reported that HBx induces apoptosis in a p53-dependent manner (19). Thus, a number of independent studies report proapoptotic effects of HBx protein in a variety of experimental settings, although the role of p53 independence may be less clear. In contrast, two studies reported that HBx can inhibit p53-dependent apoptosis of human fibroblasts and hepatocytes (28, 111). Nevertheless, more recent studies cannot substantiate that HBx has any detectable effect on p53 activity or location (84, 96). As a further complication, HBx possesses several activities that are often proapoptotic, including stimulation of Ras, activation of Src kinases, transcriptional activation of myc genes, constitutive activation of JNKs, and loss of cell cycle checkpoint controls, possibly resulting in a G₁ cell cycle block (4, 7, 12–14, 19, 58, 59, 79, 91, 95, 105). Two reports implicate possible interaction of HBx with mitochondria (85, 98), which, if substantiated and shown to alter mitochondrial function or calcium utilization, might also contribute to proapoptotic effects. The major presumptive antiapoptotic activity of HBx protein is activation of NF-κB. We therefore carried out the present study to understand how key factors that are acted on by HBx protein might promote or prevent apoptotic killing of cells.

HBx has been widely established as an activator of NF-κB, which it may do through several distinct mechanisms of action. Previous studies (18, 92, 94, 113) and Fig. 1 and 2 demonstrate that HBx activates all major forms of NF-κB, including those containing RelA (p65), c-Rel, p52, p50, and p105. Since activation of c-Rel in some settings is associated with proapoptotic effects (1, 9, 38, 63), we examined the effect of HBx activation of c-Rel in the absence of RelA by using NIH 3T3 cells deficient in the relA gene. HBx activation of c-rel in RelA KO cells was only associated with a small (ca. twofold) increase in cell killing and no significantly greater sensitivity to killing by TNF-α (Fig. 3 and 4). These data indicate that induction of c-Rel/NF-κB is not proapoptotic in HBx-expressing cells. In fact, inhibition of all forms of NF-κB by overexpression of an IκBα superrepressor revealed apoptotic killing of cells by HBx protein (Fig. 5). Activation of multiple forms of NF-κB therefore masks apoptosis by HBx. Importantly, even in the absence of NF-κB activation, there was no evidence for increased sensitivity of NIH 3T3 cells expressing HBx to killing by TNF-α (Fig. 5).

NIH 3T3 cells typically express only low levels of Myc proteins (56, 57) (Fig. 6). In addition, Myc proteins can sensitize cells to killing by TNF-α by severalfold (56, 57, 102). We therefore examined the combined effect of HBx and c-Myc expression on TNF-α-mediated cell killing. Coexpression of c-Myc and HBx protein very significantly sensitized WT 3T3 cells to killing by TNF-α (Fig. 6). These data indicate that the combined expression of HBx and c-Myc can override the antiapoptotic effect of NF-κB activation. In the absence of elevated myc gene expression and at physiologically relevant levels of HBx protein, these data suggest that HBx should have little or no proapoptotic effect, even during exposure of cells to TNF-α. Results obtained from woodchuck livers that are chronically infected by WHV support this view (25, 118). Further evidence can be found in studies demonstrating a slight enhancement of hepatocyte apoptosis only in transgenic mice that coexpress elevated levels of HBx and c-Myc proteins in their livers, compared to independent expression of either gene alone (100). Thus, if HBx is a cofactor in promoting greater pathogenesis during virus infection, our data indicate that it would require a special set of conditions, either ablation of NF-κB or strong elevation of myc gene expression plus exposure of hepatocytes to TNF-α.

Studies of other hepatotropic viruses can be instructive and support a limited role for viral factors in enhanced sensitivity of hepatocytes to TNF-α in liver disease. For example, cytomegalovirus infection of the liver in mice induces hepatitis with necrotic foci due to enhanced sensitivity of infected hepatocytes to killing by TNF-α, which occurs in a T-cell- and NK cell-independent manner (80). In this system, TNF-α is an important component for persistent liver pathogenesis by the virus, which accords with its established role in promoting hepatic inflammation and nonviral liver disease (46, 64, 77, 82; reviewed in reference 74). In the case of WHV, there is evidence for enhanced hepatocyte turnover during acute and chronic infection in some settings (40). Increased hepatocyte apoptosis during WHV and HBV infection involves specific cytokine- and T-cell-mediated killing (40). In this regard, studies still need to critically evaluate whether HBx protein plays a role in promoting greater sensitivity to cytokine (TNF-α)-mediated killing of hepatocytes. In particular settings, such as WHV infection and elevated c-myc expression, it is possible that HBx may be a cofactor in hepatocyte killing.

While HBx has been shown to display proapoptotic effects during replication of HBV and WHV in tissue culture (95), a possible proapoptotic effect of HBx during natural infection has not yet been documented. It has been proposed for other viruses that induction of apoptosis can actually facilitate propagation of viral infection by permitting efficient particle release from cells while minimizing the antiviral inflammatory response (26, 114, 117). Such a strategy could conceivably aid in establishment of HBV infection as well. This notion can find partial support in a study showing that HBx protein mutants which have lost proapoptotic function are enriched in human hepatocellular carcinomas, an end-stage event in chronic HBV infection (91). However, because HBx can also suppress cellular transformation in experimental systems by induction of apoptosis (87, 91), it was suggested that HBx might function in a proapoptotic manner early in HBV infection. Apoptosis would then be selected against during chronic HBV infection, possibly promoting carcinogenesis (91). It is also possible that stimulation of apoptosis by HBx might promote release of hepatocyte growth factors, enhancing regenerative responses and increasing the level of uninfected hepatocytes for new infection. Alternatively, it is equally possible that proapoptotic functions represent an unavoidable consequence of HBx activity and that this is in fact somewhat deleterious to viral propagation. For instance, HBx activation of Src signaling, should it prove to be important in natural infection as in cell culture (59), can be proapoptotic. Similarly, accumulation of HBx-expressing cells at the G₁/S junction of the cell cycle (91, 92), should it occur during natural infection as in culture, generally involves stimulation of Src and Myc proteins and may therefore be proapoptotic as well in some settings. Studies now need to address these important issues in model systems relevant to HBV replication and pathogenesis.