Interaction of the Hepatitis B Virus X Protein with the Crm1-dependent Nuclear Export Pathway*

The leucine-rich nuclear export signal (NES) is used to shuttle large cellular proteins from the nucleus to the cytoplasm. The nuclear export receptor Crm1 is essential in this process by recognizing the NES motif. Here, we show that the oncogenic hepatitis B virus (HBV) X protein (HBx) contains a functional NES motif. We found that the predominant cytoplasmic localization of HBx is sensitive to the drug leptomycin B (LMB), which specifically inactivates Crm1. Mutations at the two conserved leucine residues to alanine at the NES motif (L98A,L100A) resulted in a nuclear redistribution of HBx. A recombinant HBx protein binds to Crm1 in vitro . In addition, ectopic expression of HBx sequesters Crm1 in the cytoplasm. Furthermore, HBx activates NF k B by inducing its nuclear translocation in a NES-dependent manner. Abnormal cytoplasmic sequestration of Crm1, accompanied by a nuclear localization of NF k B, was also observed in hepatocytes from HBV-positive liver samples with chronic active hepatitis. We suggest that Crm1 may play a role in HBx-mediated liver carcinogenesis. -AATTCTCAGGTCTTGCCCA- -AATTCTCAGGTCTTGCCC-AAGCTCTTACATAAGAGGACTGCTGGAGCCTCAGCAATG-3 -GATCCATTGCTGAGGCTCCAGCAGTCCTCTTATGTA-AGAGCTTGGGCAAGACCTGAG-3 9 GFP-NLS, resulting in a GFP-NLS-NES or GFP-NLS-NESM fusion protein. The NES motif the HBx residues 87–103, NESM HBx-NES motif mutations the conserved leucine residues (L98A L100A). pSVCM GFP-CRM1 pNF B-Luc, John

The leucine-rich nuclear export signal (NES) is used to shuttle large cellular proteins from the nucleus to the cytoplasm. The nuclear export receptor Crm1 is essential in this process by recognizing the NES motif. Here, we show that the oncogenic hepatitis B virus (HBV) X protein (HBx) contains a functional NES motif. We found that the predominant cytoplasmic localization of HBx is sensitive to the drug leptomycin B (LMB), which specifically inactivates Crm1. Mutations at the two conserved leucine residues to alanine at the NES motif (L98A,L100A) resulted in a nuclear redistribution of HBx. A recombinant HBx protein binds to Crm1 in vitro. In addition, ectopic expression of HBx sequesters Crm1 in the cytoplasm. Furthermore, HBx activates NFB by inducing its nuclear translocation in a NES-dependent manner. Abnormal cytoplasmic sequestration of Crm1, accompanied by a nuclear localization of NFB, was also observed in hepatocytes from HBV-positive liver samples with chronic active hepatitis. We suggest that Crm1 may play a role in HBx-mediated liver carcinogenesis.
Hepatocellular carcinoma (HCC) 1 is one of the most prevalent malignant diseases worldwide and hepatitis B virus (HBV) is the major etiologic factor for HCC (1)(2)(3). HBV is a DNA tumor virus, which encodes four open reading frames, S/preS, C/preC, P, and HBx (4). The oncogenicity of HBV is largely the result of HBx, the smallest gene encoding a 17-kDa protein (3,(5)(6)(7)(8). One major cellular function of HBx is its promiscuous transcriptional activation activity, a property that is believed to contribute to its oncogenicity (9). A wide range of cellular genes can be up-regulated or down-regulated by HBx (9). However, HBx is localized in the cytoplasm mostly and does not bind to double-stranded DNA. A "universal" effect of HBx, on otherwise totally different types of promoters with no obvious consensus sequence, has led to the hypothesis that HBx may regulate gene expression by interacting directly with host general transcription factors (10 -12), or indirectly via the activa-tion of protein kinase C and RAS-RAF mitogen-activated protein kinase (MAPK) signaling pathways (13)(14)(15). Although HBx can induce neoplastic transformation, presumably by preventing p53-mediated apoptosis (16 -18), it also can induce apoptosis in a p53-dependent or -independent manner (19,20), 2 or sensitize cells to tumor necrosis factor ␣ (TNF␣)-induced apoptosis (21). Therefore, the precise mechanism related to its effector remains unknown and none of these studies satisfactorily explain the pleiotropic effects associated with HBx.
Close inspection of the HBx sequence revealed a short, hydrophobic, leucine-rich nuclear export signal motif (NES) (Fig.  1A). An NES is located in the center region of HBx (residues 89 -100). The center region of HBx is retained in HCC frequently and is essential for its transactivation (22)(23)(24). This region also is conserved among HBx from different subtypes (Fig. 1A). Several viral proteins including HIV-1 Rev, HTLV-1 Rex, and adenovirus E4 34-kDa proteins contain functional NESs (Fig. 1A). Similar to HBx, Rev and Rex also are potent viral and cellular transactivators with no apparent DNA binding property (25,26). In addition, NESs also have been identified in cellular proteins, many of which are involved in transcription, cell signaling cascade, oncogenic transformation, and cell cycle regulators. Examples include protein kinase inhibitor, MAP kinase kinase (MAPKK), TFIIIA, Mdm2, p53, IB␣, NF-AT, cyclin B1, c-Abl, and 14-3-3 (reviewed in Ref. 27). The activities of these proteins are tightly regulated by their NESs. The nuclear export receptor Crm1 and its cofactor Ran GTPase are essential in this process by recognizing NESs and mediating nuclear protein export (27,28). In addition, previous results indicate that Crm1 may be involved in maintaining chromosomal integrity (29) and Ran may play a key role in regulating mitosis initiation by stimulating spindle formation (30,31). Mutation of the hydrophobic leucine residues to alanines have been shown to disrupt NES function in a number of proteins, including HIV-rev, E4 34-kDa, p53, Mdm2, and cyclin B1 (25,26,32,33).
In this study, we have investigated the hypothesis that the pleiotropic effects associated with HBx may be contributed by the presence of a NES motif, and HBx may activate cellular gene expression and induce oncogenicity through the modulation of Crm1-mediated functions. We have identified a functional NES motif in HBx. This motif is necessary for HBxinduced cytoplasmic sequestration of Crm1, and subsequently, the nuclear translocation and activation of NFB. Cytoplasmic retention of Crm1 also is found in liver samples with chronic active hepatitis infected with HBV, a condition that is predisposing individuals to the development of HCC. We suggest that the inactivation of the Crm1-mediated pathway may be an early step during viral hepatitis-mediated liver carcinogenesis.
Cell Culture and Media-Normal primary human fibroblasts were obtained from Coriell Institute for Medical Research (Camden, NJ). These cells were grown in Ham's F-10 medium supplemented with 15% fetal bovine serum and were used before passage 12. A telomeraseimmortalized human fibroblast line (NHF-hTERT), a gift of Dr. Judy Campisi, was grown in Dulbecco's modified Eagle's media supplemented with 10% fetal bovine serum. Normal primary human hepatocytes were purchased from Clonetics (BioWittaker), grown in LCM (BioWittaker), and used only at passage 1. Hep3B cells were grown in Eagle's minimum essential medium containing 10% fetal bovine serum.
Microinjection, Transfection, Luciferase, and Indirect Immunofluorescence Assays-Microinjection of NHF cells was done essentially as described previously (35). Transfection of NHF-hTERT and Hep3B cells was carried out by the LipofectAMINE Plus reagent based on the recommended protocol described by the manufacturer (Life Technologies, Inc., Gaithersburg, MD). For the luciferase reporter assay, cells were seeded into 6-well dishes (Costar) at 50% confluence, co-transfected with 0.2 g of pNFkB-Luc or pNFAT-Luc, along with 2 g of each of various HBx constructs, in the presence or absence of TNF␣ (10 ng/ml). A total of 0.5 g of pRL-null or 0.003 g of pRL-CMV was added into each transfection as an internal control. pEGFPC1 was used to keep the total amount of plasmid DNA constant. Cells were then incubated for 24 h prior to harvesting. The luciferase activity was carried out by the Dual Luciferase Reporter Assay System (Promega) according to the manufacturer's instructions and were measured in a Monolight 2010 luminometer (Analytical Luminescence Laboratory). The luciferase activity was expressed as fold activation against the untreated reporter by itself, which was normalized by Renilla luciferase activity for the transfection efficiency. The reported results represent at least three independent transfections. For immunocytochemistry analysis, cells were fixed in 4% paraformaldehyde in phosphate-buffered saline for 10 min, followed by methanol for 20 min at 24 h postmicroinjection, and washed in phosphate-buffered saline-plus (0.15% glycine, 0.5% bovine serum albumin in phosphate-buffered saline). The GFP fusion proteins can be visualized directly in a fluorescence microscope equipped with an fluorescein isothiocyanate filter with or without fixation. The HBx expression also was verified by an anti-HBx monoclonal antibody (data not shown). NFB was detected using an anti-NFB polyclonal antibody (1:100) (Santa Cruz) and Crm1 was detected using an anti-Crm1 polyclonal antibody (1:100). Both antibodies were incubated for 1 h at 37°C. Anti-rabbit IgG conjugated to fluorescein or Texas Red (Vector Laboratories) was used (1:300) for 1 h at room temperature. Nuclei were stained with 4,6-diamidino-2-phenylindole. The following criteria were used to define the subcellular localization, preferentially cytoplasmic localization: protein signals mostly intense in the cytoplasm; preferentially nuclear localization: protein signals mostly intense in the nucleus. Some of the slides were analyzed with a blind fashion to avoid any bias. For GFP-NLS reporter expression, cells on the same coverslip were microinjected with GFP-NLS, GFP-NLS-XNES, or GFP-NLS-XNESM construct and incubated for 6 h. Cells were monitored under a Zeiss Axioskop fluorescence microscope and representative images were taken using a high-performance CCD imaging system (IP Lab Spectrum) with the same exposure time. For co-localization of Crm1 and HBx analysis, cells were co-stained with anti-HBx and anti-Crm1 antibodies with the corresponding secondary antibodies conjugated with Texas Red or fluorescein, and analyzed by a Bio-Rad MRC 1024 confocal system. Sequential excitation at 488 and 568 nm was provided by a krypton-argon gas laser. Emission filters of 598/40 and 522/32 were used for sequentially collecting red and green fluorescence in channels 1 and 2, respectively. Z-sections were collected at 0.5-m intervals for each cell using LaserSharp software. Images were analyzed by Confocal Assistant software.
In Vitro Protein-binding Assay-In vitro binding between HBx and hCrm1 was determined with the in vitro translated HBx and hCrm1 (16). The HA-tagged HBx and the hCrm1 proteins were made by the one-step TnT in vitro transcription and translation system (Promega), as described previously (16), from pcDNA-HA-HBx and hCrm1Blue744 vectors (gift of Dr. Gerard Grosveld), respectively. Aliquots of HBx (25 l) and hCrm1 (45 l) were mixed together in CBBL buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, and 0.1% Nonidet P-40) in the presence of 0.2 mM GTP␥S and 80 l of anti-HA affinity matrix (Roche Molecular Biochemicals, Indianapolis, IN). The mixture was incubated at 37°C for 60 min. After centrifugation, beads were washed four times with CBBL buffer. The bound proteins were separated on SDS-polyacrylamide gel electrophoresis and Crm1 was detected by Western blotting using anti-Crm1 antibody and followed by the Amersham ECL system. A 10% of in vitro translated hCrm1 was loaded in parallel as the input.
Immunohistochemistry Analysis-A total of 12 samples were included in this part of the study. Histopathologically, they included 5 normal liver samples, 7 viral hepatitis cases showing grades 1 to 2 activity according to the criteria of Scheuer (36). In addition to H & E light microscopy, tissue sections were cut from frozen samples with a thickness of 4 m and mounted on electrically charged glass slides. The sections were fixed in absolute alcohol and then washed with 2 changes of phosphate-buffered saline for 5 min each. They were then quenched with 3% hydrogen peroxide to block the endogenous peroxidase activity for 30 min. Following incubation with 10% horse serum to block the nonspecific binding, the sections were incubated overnight at 4°C with anti-CRM1 antibodies. The sections were prepared in duplicate and two anti-CRM1 antibodies were used, including a goat anti-Crm1 antibody from Santa Cruz at a 1:10 dilution (Santa Cruz) and a rabbit anti-Crm1 polyclonal antibody (Gift from Gerard Grosveld) at a 1:100 dilution. A rabbit anti-NFB antibody (anti-p65) (Santa Cruz) was used at a 1:100 dilution for staining NFB. Biotinylated secondary antibodies and streptavidin peroxidase complex (LSAB) were used. Chromogenic development was obtained by the immersion of sections in 3,3Ј-diaminobenzidine solution (0.25 mg/ml with 3% hydrogen peroxide). The slides were counterstained with Harris' hematoxylin and re-hydrated with alcohol and xylene. Slides were evaluated in a blind fashion.

HBx Has a Functional Leucine-rich Nuclear Export Signal-To test if
HBx-NES functions in nuclear protein export, we first investigated whether the subcellular localization of HBx is sensitive to LMB, an antitumor agent that selectively binds to and blocks Crm1-mediated nuclear export (37,38). Consistent with previous findings (18), HBx was localized in the cytoplasm predominately with a punctate staining pattern in appearance when expressed transiently in normal human fibroblasts (NHF) (Fig. 1) and in a HCC-derived cell line (Hep3B) (data not shown). However, LMB treatment causes an increase in nuclear accumulation of HBx (Fig. 1, B and C).
Next, we used a GFP protein export reporter assay described previously (34) to further test if the HBx-NES oligopeptide functions in nuclear protein export (34). This reporter uses a nuclear localization signal (NLS) fused to a GFP, thereby in- ducing nuclear localization of the fusion protein. Consistently, the GFP-NLS reporter was localized in the nucleus of NHF exclusively. However, when an HBx-NES was fused to this reporter, the resulting GFP-NLS-XNES fusion protein can be found in the cytoplasm with a diffused staining pattern in appearance, although no nuclear exclusion of this reporter protein could be observed (Fig. 1D). In contrast, GFP-NLS-XNESM that contains a mutated HBx-NES oligopeptide (L98A,L100A) does not localize in the cytoplasm (Fig. 1D). To further examine if cytoplasmic localization of a full-length HBx is dependent on the presence of an NES, a GFP-HBx fusion gene (HBx) or an NES-mutated GFP-HBx fusion gene (HBx-NESM) was constructed. Again, GFP-HBx was localized in the cytoplasm predominantly with a punctate pattern in appearance, while HBxNESM was found mainly in the nucleus with a diffused pattern (Fig. 1E). Taken together, these data demonstrate that HBx-NES is functional in Crm1-mediated nuclear protein export.
Interaction between Crm1 and HBx-Nuclear export receptor Crm1 is known to interact with proteins containing NESs and to mediate nuclear protein export. The majority of Crm1 protein is found in the nucleus of normal cells (39). Consistently, every NHF cell displays an endogenous nuclear-diffused Crm1 distribution (Fig. 2). When HBx was expressed transiently in NHF, Crm1 was often found in the cytoplasm with a punctate staining pattern. Approximately 39% Ϯ 2 of the HBxexpressing NHF cells displayed co-localization of Crm1, and over 99% of cytoplasmic punctate Crm1 signals were co-localized with HBx ( Fig. 2A). The punctate staining patterns containing both Crm1 and HBx signals were not always the same in appearance ( Fig. 2A, panels c-l). These results suggest that the cytoplasmic HBx and Crm1 are not necessarily associated with a single subcellular component. In contrast, no cytoplasmic co-localization was found in NHF cells expressing HBx-NESM. These data indicate that HBx was able to partially retain Crm1 in the cytoplasm and suggest that HBx binds to Crm1 in vivo. Consistently, in vitro translated HBx can physically interact with in vitro translated Crm1 (Fig. 2B).
HBx-mediated Activation of NFB Is Dependent on the Presence of NES-HBx can activate NFB and other cellular transcription factors including NF-AT and MAP kinase cascade, although the mechanism of activation is unknown (14,40,41). Because NFB is inactivated normally through its binding to IB␣ which is sequestrated in the cytoplasm by Crm1-mediated nuclear export (42), we examined whether HBx-NES is responsible for the activation of NFB. Over 97% of normal dividing NHF cells display a predominant localization of NFB in the cytoplasm (Fig. 3A). In contrast, ϳ55 Ϯ 7% of HBxexpressing NHF cells show a preferentially nuclear distribution of NFB, while an NES mutated HBx (HBxNESM) has lost this activity (Fig. 3, A and B). Next, we used a luciferase reporter that contains NFB responsive elements (pNFkB-Luc) to investigate if the HBx-induced nuclear localization of NFB correlates with its activity. pNFB-Luc was co-transfected with HBx or HBxNESM into an hTERT-immortalized NHF (NHF-hTERT) or a hepatocellular carcinoma cell line (Hep3B) in the absence or presence of TNF␣, an agent known to sensitize HBx-mediated transactivation (9). Consistently, HBx alone induced the luciferase activity weakly which could be further enhanced by the treatment of TNF␣ in both NHF-hTERT and Hep3B cells. In contrast, HBxNESM is devoid of its ability to increase luciferase activity even in the presence of TNF␣ (Fig. 3C).
Cytoplasmic Localization of Crm1 Associated with HBV-infected Liver Samples-To investigate if Crm1 is sequestrated cytoplasmically in liver samples from chronic active hepatitis patients with HBV infection, we performed immunohistochemistry analysis of Crm1 in frozen tissue sections or in paraffinembedded sections by anti-Crm1 polyclonal antibody. A total of 12 liver samples with samples from 5 normal individuals were included in this study. The presence of HBV DNA including Core, preS, and HBx genes in these samples was verified by PCR (data not shown). Among 7 HBV positive cases, liver samples show chronic active hepatitis with grades 1 to 2 activity (Table I). While 5/5 normal samples revealed nuclear Crm1 staining exclusively, 6/7 HBV samples showed cytoplasmic Crm1 staining (Fig. 4 and Table I) in hepatocytes. An anti-NFB antibody was also used to determine the status of NFB in these liver samples. Again, normal liver samples revealed mainly a cytoplasmic localization of NFB while HBV positive liver samples showed a nuclear distribution of NFB. The above findings provide an in vivo physiological significance for the presence of a cytoplasmic sequestration of Crm1 accompanied by an activation of NFB associated with HBV infection. DISCUSSION We have demonstrated that the oncogenic HBx contains a functional NES, and that this NES is required for the HBxmediated activation of the NFB signaling cascade. We also have shown that HBx is able to bind to and sequester Crm1 in the cytoplasm, which correlates with the nuclear localization and activation of NFB. Furthermore, liver samples from HBVinfected patients with chronic active hepatitis, a condition predisposing individuals for the development of liver cancer, also display cytoplasmic sequestration of Crm1 and nuclear localization of NFB. It is possible that HBx may use a Crm1-dependent mechanism to modulate cellular gene transcription. Our data provide a plausible model whereby HBx may induce oncogenicity through the modulation of the Crm1-mediated pathway.
The studies described in this report may offer a common mechanism to explain the pleiotropic effects of HBx. First, Crm1 transports and controls many cellular proteins including NFB, NF-AT, MAPKK, and p53, whose activities are known to be regulated by HBx (14,17,40,41,43). Second, Crm1 also regulates cytoplasmic mRNA transport. Inactivation of Crm1 by HBx, in principle, would alter the stability of mRNA thereby influencing gene expression. Consistent with this hypothesis is our findings that HBx is able to sequester Crm1 in the cytoplasm. The HBx-NES is essential for sequestering Crm1 in the cytoplasm and for the activation of NFB. Inactivation of Crm1-mediated nuclear protein export is known to activate this target (44). Therefore, the promiscuous transactivationmediated by HBx may be a secondary effect resulting from alteration of Crm1 function.
An alternative mechanism for HBx to activate NFB is through the degradation of IB␣ in the cytoplasm by inducing its phosphorylation and ubiquitination. However, we did not observe any detectable degradation of IB␣ in cells expressing HBx by Western blot analysis with anti-IB␣ antibody (data not shown). This also correlates with our results that HBx alone only displays a weak activation of NFB that can be enhanced significantly by TNF␣ (Fig. 3C). It is possible that a small degree of dissociation of the NFB-IB␣ complex, followed by the degradation of IB␣ induced by HBx in the cytoplasm, may contribute to the nuclear localization of NFB. A recent study by Weil and colleagues (45) indicate that HBx may induce the nuclear import of NFB/IB␣ by binding to IB␣. Interestingly, the region of IB␣ that is critical for HBx-induced nuclear import of IB␣ contains a functional NES (45). These findings are consistent with our data that HBx-induced nuclear localization of NFB/IB␣ is dependent on the IB␣-NES.
FIG. 3. HBx-mediated activation of NFB is dependent on a functional nuclear export motif. A, normal primary human fibroblasts were expressed transiently without (a-c), or with HBx (d-f) or HBxNESM (g-i) for 24 h. Cells were fixed and stained with anti-NFB. Texas Red-conjugated secondary antibody was used to visualize NFB (a, d, and g). The HBx signal was detected directly with an fluorescein isothiocyanate filter (e and h). Percentages of cells with nuclear staining of NFB that are positive for HBx are shown B. C, a luciferase reporter containing 4 NFB responsive elements was transfected with HBx or an HBx mutant (HBxNESM, L98A, G99A, L100A) into a telomerase-immortalized human fibroblast line (NHF-hTERT) or a hepatocellular carcinoma cell line (Hep3B). Cells were then treated with or without 3 ng/ml TNF␣ for 24 h. The luciferase activity was measured and data were normalized by an internal Renilla luciferase vector.
HBx is thought to be small enough to diffuse passively through the nuclear pore complex. The fact that this oncogenic viral protein has acquired NES activity, is preferentially localized in the cytoplasm, and modified nuclear export implies that nuclear export may be an important target for viral-mediated oncogenesis. Interestingly, cellular oncoproteins such as c-Abl and Mdm2 are known to be involved in a Crm1-dependent nuclear export. Analogous to viral hepatitis oncoprotein, the cellular oncoproteins, in principle, may induce neoplastic transformation also through the disruption of the Crm1-mediated mechanism. It is known that TPR and CAN are nuclear pore complex-associated proteins that act as the docking sites and are essential for Crm1-mediated nuclear export (27). Ear-lier findings demonstrated that the TPR and CAN genes are mutated through translocation in thyroid carcinoma, gastric carcinoma, and acute myeloid leukemia (46 -52). We also found that cytoplasmic sequestration of Crm1 is associated frequently with HCC (data not shown). In addition, Crm1 and Ran may play a role in mitosis initiation (29 -31). It is possible that the inactivation of Crm1 and Ran may induce genomic instability, a predisposing factor for cancer development. Thus, we postulate that the inactivation of the Crm1-mediated nuclear export is a common event during viral and cellular oncogenesis.