Phosphorylation-dependent Binding of Hepatitis B Virus Core Particles to the Nuclear Pore Complex

Purified Escherichia coli–derived core particles, as characterized by Crowther et al. (7), were obtained from Drs. Galina Borisova and Paul Pumpens (Biomedical Research and Study Centre, University of Latvia, Riga, Latvia). The particles were subjected to permeabilization and phosphorylation as described previously (31). Because of the RNA content of the particles, the incubation temperature and time for phosphorylation were increased to 37°C and 30 min, respectively. After reconstitution of particle integrity, entire core particles were separated from free nucleotides, protein kinases, and unassembled core proteins by sedimentation through a 0.5 ml 25% (wt/vol) sucrose cushion in TNE buffer (40 mM Tris/HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA) using a rotor (TL100; Beckman Instruments) at 100,000 rpm at 15°C for 1 h. The pellet was resuspended in 50 μl TNE, aliquoted, and stored at −80°C until further use. The core particle concentration was determined using a core ELISA (>20 ng/ μl; ref. 47) using a standard of E. coli–expressed core particles (Chiron).

20 ng of phosphorylated or unphosphorylated core particles was separated on a 1% agarose gel, using agarose gel loading buffer without SDS. To disintegrate the particle structure, cores in parallel samples were incubated in 0.1% trifluoroacetic acid (TFA), heated to 50°C for 10 min, neutralized by adding 0.5 vol of 0.1 M Tris-HCl, pH 8.8, before loading on the gel. Proteins were blotted on an Immobilon-P membrane overnight (Millipore; 50). The membrane was blocked for 1 h at room temperature in 5% (wt/ vol) fat-free milk in PBS followed by addition of the first antibody (anti-HBc; DAKO) at a dilution of 1:4,000 in 5% milk/PBS for 2 h at room temperature. After washing (3× for 10 min in 0.1% milk/0.1% Tween 20/PBS at room temperature), the second antibody reaction was performed with HRP-labeled donkey anti–rabbit antibody (Dianova Co.) at a dilution of 1:5,000 in 5% milk/PBS for 1 h at room temperature. After washing, the antibodies were visualized by an enhanced chemiluminescence kit (Boehringer Mannheim GmbH). For analyzing the radioactive phosphorylation with [γ-³²P]ATP (Amersham), the membrane was washed three times for 10 min in PBS before analysis by phosphoimaging (BAS-IIIS for Fuji BAS1000 Bioimager; Fuji Photo Film Co.).

For quantification of core phosphorylation, 2 μl of the samples was spotted on GF/C filters (Whatman), air dried, washed twice for 10 min in ice-cold 10% (wt/vol) TCA and twice for 5 min in ice-cold 5% TCA. The filters were dried and protein-bound radioactivity was determined by liquid scintillation counting.

For isopycnic centrifugation a CsCl gradient was performed with a density of 1.51–1.22 g/ml PBS in an SW60 rotor (Beckman). The ³²P-labeled core particle preparation was loaded on top and centrifuged at 35,000 rpm for 16 h at 10°C. The gradient was harvested in 20 fractions of 175 μl. 20 μl of each aliquot was subjected to core ELISA and to liquid scintillation counting. To analyze the dephosphorylation of phosphorylated core particles, 20 ng of particles was incubated with bacterial alkaline phosphatase (200 U/ml; New England Biolabs) or with 4,000 U/ml calf intestinal phosphatase (Boehringer Mannheim GmbH) in 100 mM Tris-HCl, pH 7.5, 150 mM NaCl or 100 mM Tris-HCl, pH 8.5, 150 mM NaCl, respectively. In the control reaction, phosphorylated core particles were disintegrated by TFA, as described above, before dephosphorylation. Samples were incubated for 30 min at 37°C. Samples containing core particles and the unphosphorylated control were separated on an agarose gel, blotted, and exposed as described above. Disintegrated core protein samples were separated on a 16% SDS-PAGE, stained with Coomassie brilliant blue, and dried before autoradiography.

HepG2.2.15 cells were grown on 16-cm dishes in 5% CO₂ at 37°C. After reaching confluency DMEM containing 3% FCS was added for 4 d. 500 ml of medium was collected and insoluble components were sedimented by centrifugation at 3,500 g for 30 min at 4°C. HBV of the supernatant was sedimented through 3-ml cushions of 25% (wt/wt) sucrose/TNE in an SW27 rotor at 27,000 rpm for 36 h at 4°C. Each pellet was resuspended in 150 μl TNE adjusted to a CaCl₂ concentration of 5 mM and incubated with 25 U/ml S7 nuclease (Boehringer Mannheim GmbH) for 1 h at 37°C. To remove the surface proteins from the virus, NP-40 was added to 0.1%. After incubation, insoluble components were sedimented through a 0.5-ml cushion of 25% (wt/wt) sucrose/TNE for 10 min at 10°C in a TL100 rotor (Beckman Instruments) at 100,000 rpm. The upper phase, including the interphase, contained the core particles. They were sedimented through a 0.5-ml cushion of 25% (wt/wt) sucrose/TNE for 2 h at 10°C in a TL100 rotor (Beckman) at 100,000 rpm. The pellet was resuspended in 200 μl TNE.

FITC-BSA (Sigma Chemical Co.) was conjugated with NLS according to Görlich et al. (18). The NLSs used were the SV-40TAg NLS (PKKKRKVED; 18) and M9 domain of hnRNP (YNNQSSNFGPMK). Both contain a spacer (amino acid sequence CGGG; 18) at their NH₂ terminus. Matrix-assisted laser desorption/ionization (MALDI) mass spectrometry (Dietmar Linder and Monica Linder, Institute of Biochemistry, Department of Medicine, Giessen, Germany) of the conjugates showed, on average, 19 NLS were linked to 1 BSA molecule.

Cells were grown in DMEM with 7% FCS on collagen (Sigma Chemical Co.) or Cell-Tak (Becton Dickinson) coated 12-mm coverslips in 24-well dishes in 5% CO₂ at 37°C. The cells were washed with 1 ml serum-free DMEM and permeabilized with 80 μg/ml digitonin (Sigma Chemical Co.) in DMEM for 10 min at 37°C (1). They were washed twice for 10 min at 4°C in washing buffer (transport buffer: 2 mM Mg-acetate, 20 mM Hepes, pH 7.3, 110 mM K-acetate, 1 mM EGTA, 5 mM Na-acetate/1 mM DTT) containing 1% BSA and 10% goat serum (Dianova) followed by a 10-min incubation at 37°C in a humidified box using washing buffer. Preincubation of permeabilized cells with 50 μg/ml WGA/ml transport buffer (Boehringer Mannheim GmbH) or 200 μg antibodies (anti-NPC antibody, mAb414 [BAbCO]/ml transport buffer, anti-NS2, or anticalnexin), was performed during this step.

Permeabilized cells were incubated with 5 μg/ml core particles in 20 μl transport buffer containing 30 mg/ml rabbit reticulocyte lysate (RRL; Promega Corp.); 10 μg/ml aprotinin (Sigma Chemical Co.), 10 μg/ml leupeptin (Sigma Chemical Co.), 10 μg/ml pepstatin (Sigma Chemical Co.) for 20 min at 37°C. If required, an ATP-generating system (1 mM ATP, 5 mM creatine phosphate, and 20 U/ml creatine phosphokinase; Sigma Chemical Co.) or ATP-depleting system (7 mM glucose, 1 U/ml hexokinase; Sigma Chemical Co.) was added (1).

In some experiments, RRL was preincubated with peptides, representing parts of the COOH terminus of the core protein (amino acids 142–155, TLPETTVVRRRDRG; amino acids 158–168, PRRRTPSPRRR; amino acids 165–175, PRRRRSQSPRR; amino acids 173–183, PRRRRSQSRES), or the NLS of SV-40 T antigen (STPPKKKRKRKV), or lamin B2 (RSSRGKRRRIE) at a concentration of 2 mM for 10 min on ice. The peptides were synthesized by Dr. Ursula Friedrich (Institute of Medical Virology, Giessen, Germany) on an Applied Biosystems 431A Peptide Synthesizer, purified by HPLC, and analyzed by MALDI mass spectrometry (Dietmar Linder and Monika Linder).

For incubation of cores in the absence of cytosolic proteins, BSA was added instead of reticulocyte lysate to the same protein concentration (30 mg/ml). In some experiments, importin (karyopherin) α (Rch 1) and β (gift of D. Görlich, ZMBH, Heidelberg, Germany) were added to BSA-containing buffer at a concentration of 30 ng/μl (16) each. For the transport assay, FITC-labeled BSA conjugate with SV-40TAg NLS was added to a final concentration of 20 μg/ml. After incubating for 20 min at 37°C, cells were washed in washing buffer as described above.

When the phosphorylated E. coli–derived core (P-rHBc) was used for blocking the import of the FITC-BSA–M9 conjugate, P-rHBc was added at a concentration of 100 ng/μl to the assay in the presence of RRL and incubated for 20 min at 37°C as described above. In the control experiment, cells were treated identically but without addition of P-rHBc. After the binding reaction, cells were washed three times in washing buffer for 10 min at 0°C. The coverslips were loaded on a new reaction mixture containing reticulocyte lysate, ATP-generating system and 20 ng/μl of the FITC-BSA–M9 conjugate. After a second incubation period of 20 min at 37°C, the coverslips were washed as described above.

To quantitate the amount of bound core particles, cells were scraped off the coverslip using a rubber policeman, counted in a Neubauer chamber, and lysed by addition of 100 μl 10 mM Tris-HCl, pH 7.5, 0.1% Triton X-100, 10 mM MgCl₂, 150 mM NaCl containing 20 U DNase I, and 2 mg/ ml RNase A. Cells were incubated for 15 min at 37°C and lysis was controlled microscopically before analysis by ELISA.

For immunofluorescence microscopy, cells were fixed onto the coverslips using 3% paraformaldehyde (Merck) in PBS, pH 7.4, for 30 min at room temperature. The coverslips were washed once in PBS and in PBS/0.1% Triton X-100 for 5 min. After two additional washings for 5 min in PBS at room temperature, cells were exposed to a mixture of anti-HBc antibody (1:200) and anti-NPC (1:1,000) in PBS/10% goat serum/1% BSA for 90 min at 37°C. Coverslips were washed in PBS three times for 5 min at room temperature and incubated with secondary antibodies (affinity-purified FITC-conjugated goat anti–rabbit IgG, H+L; Dianova; affinity-purified Texas red–conjugated goat anti–mouse IgG, H+L; Dianova) in PBS/10% goat serum/1% BSA for 40 min at 37°C. Cells were washed as described above and embedded in 50 mg/ml DABCO/moviol (Sigma Chemical Co. and Hoechst, respectively). Microscopy was performed on a Zeiss fluorescence microscope, equipped with a 63× plan apochromat objective (Zeiss). Photographs were taken on Kodak TMAX 400 pro film. Negatives were developed at 1600 ASA.

Confocal immunofluorescence microscopy was performed on a Leica DM IRBE microscope. Analysis of core-NPC colocalization and of the nuclear localization of FITC-labeled conjugates was done by using the TRITC- and FITC-fittings at a pinhole size of 0.

7 × 10⁶ sheep anti–rabbit conjugated biomagnetic beads (Dynal) were washed twice in PBS according to the vendor's manual, resuspended in 500 μl PBS containing 2 μl anti-core antibody-containing rabbit serum (DAKO), and incubated overnight at 4°C. After washing as described above, 25 ng core was added and incubated overnight at 4°C. The beads were washed three times in 500 μl PBS, resuspended in 500 μl 0.1% BSA/ PBS, and incubated for 1 h at 4°C. Afterwards, the beads were added to 100 μl of RRL for 30 min at 37°C. In a control experiment, 2 mM of lamin B2 NLS (RSSRGKRRRIE) was added to the lysate. Further controls were performed by replacing lysate either with 300 ng importin β and 450 ng importin α in transport buffer, or by 300 ng importin β alone. The beads were washed three times in 500 μl PBS, 1× 500 μl PBS/0.1% Tween 20, and transferred to a new tube. After three cycles of washing in 500 μl PBS, the pellet was resuspended in 30 μl Laemmli buffer (66 mM Tris-HCl, pH 6.8, 10% [vol/vol] glycerin, 5% [vol/vol] β-mercaptoethanol, 2% [wt/vol] SDS, 0.015% [wt/vol] bromphenol blue), and denatured for 5 min at 96°C. The samples were separated on an 12% SDS-PAGE and blotted overnight to an Immobilon-P membrane (Millipore) by wet transfer. The membrane was blocked for 1 h at room temperature in 5% fat-free milk in PBS and the first antibody (anti-importin β, gift from D. Görlich) was added at a dilution of 1:4,000 in 5% milk/PBS for 2 h at room temperature. Washing, incubation with the second antibody, and detection were performed as described above.

The in vitro generated cores (rHBc) used in this study were obtained by expressing the core protein in E. coli in the absence of other viral proteins. In the bacterial cytosol, the core proteins assemble into particles (43) that have the same size and symmetry as authentic cores (33). Instead of viral DNA, they contain unspecific E. coli RNA (3, 13, 39). When analyzed by isopycnic centrifugation in CsCl, the cores produced banded at 1.355 g/ml in CsCl (Fig. 1 A), the same density as authentic cores (1.355–1.36 g/ml; ref. 22). In contrast to cores expressed in eukaryotic cells, E. coli–derived cores are not phosphorylated.

To generate P-rHBc gradient purified cores were permeabilized using low salt, incubated in the presence of PKC and ATP, reconstituted in physiological salt solution, and repurified by gradient centrifugation (31). Using [³²P]ATP for labeling, the extent of phosphorylation was ∼12–24 phosphate groups per particle. In native agarose gel electrophoresis, both P-rHBc and rHBc comigrated as a homogeneous band (Fig. 1 B). Alkaline phosphatase did not remove the phosphorus-32 label of intact core particles (Fig. 1 C, lanes 2 and 3) but after disruption of the particle structure (Fig. 1 D, lanes 2 and 3). Thus, it was apparent that the phosphorylation sites were inaccessible to enzymatic hydrolysis as in authentic cores (15). The phosphorylation sites are located close to the COOH terminus of the core protein overlapping with an arginine-rich region (15, 37) containing several putative NLS (59). Immunoprecipitation of core protein after disruption of particle structure using an antibody specific to a phosphorylated peptide derived from amino acids 166–177 (mAb 2212; ref. 37) demonstrated that serine 172 was phosphorylated by PKC (data not shown). Identical results were obtained for core proteins derived from human liver (37) or HBV DNA-transfected hepatoma cell lines (36).

In addition, infectious HBV particles (48) from the stable HBV DNA-transfected hepatoma cell line HepG2.2.15 were purified from the culture medium. In contrast to in vitro generated cores, these cores are from mature viruses and contain the full complement of viral DNA, polymerase, and host cell components. Treatment of the virus with the detergent NP-40 released the cores from the surface proteins (P-hHBc; 24). These cores migrated in CsCl gradients with a density of 1.36 g/ml and did not show any reactivity to antisurface glycoprotein antibodies in ELISA or immunoblot after separation on a native agarose gel (data not shown). Thus, regardless of the source of the cores, P-rHBc and P-hHBc behaved similarly by several biochemical and immunological criteria.

To determine whether the viral cores can associate with nuclei, we used the human hepatoma cell line, HuH-7, grown on coverslips. The plasma membrane was permeabilized using digitonin (1), the cytosol was washed out, and the E. coli–generated HBV cores (rHBc or P-rHBc) were added in the presence or absence of RRL, which is a source of cytosolic factors. In some samples, the lysates were complemented with either an ATP generating (+ATP) or depleting system (−ATP). After fixation, cores and NPCs were localized by double indirect immunofluorescence microscopy using core specific antibodies and antibodies to NPC. Immunoblotting of cores electrophoresed in agarose gels (Fig. 1 B) and quantitative ELISA assays (not shown) confirmed that both rHBc and P-rHBc reacted equally well with the anti-core antibody. The antibodies were selective for intact cores, since they bound to heat-denatured particles with ∼100-fold lower efficiency.

The P-rHBc gave a strong rimlike fluorescence pattern characteristic of ligand association with the nuclear envelope (Fig. 2, P-rHBc, +RRL, top row). The labeling colocalized extensively with the one for NPCs (Fig. 2, lower row). In contrast, core particles that had not been phosphorylated (not shown) or had been mock phosphorylated in the absence of PKC (rHBc) did not bind to the permeabilized cells (Fig. 2, rHBc, top row), to the nucleus, nor to other cellular structures. Thus, the lack of association of the rHBc with the nucleus cannot be explained by competition with nonnuclear structures. Additionally, a differential entry of the different types of cores into the permeabilized cells seemed unlikely since they looked identical in electron microscopy (31) and behaved identically in biochemical analysis (Fig. 1). The binding of the P-rHBc to the nucleus indicated that phosphorylated core proteins were essential for nuclear association.

P-rHBc bound to nuclei only if a cell lysate was present (Fig. 2, P-rHBc, +RRL). Inclusion of ATP-generating or -depleting systems had no detectable effect (Fig. 2, +RRL +ATP vs. +RRL −ATP). When the reticulocyte lysate was replaced by buffer containing BSA, no binding was observed (Fig. 2, −RRL). Taken together, these findings showed that HBV cores interacted with the nucleus of permeabilized cells, and that binding depended on core phosphorylation and cytosolic factors but not on metabolic energy. Identical results were obtained with the human hepatoma cell line HepG2 and with the mouse fibroblast cell line LTK (data not shown).

Confocal immunofluorescence microscopy showed that the bound cores were localized at the nuclear envelope and not imported deeper into the nucleoplasm (Fig. 3), at least not in a form detected by the antibodies used. The nuclei were import-competent as shown by the efficient uptake of FITC-labeled SV-40TAg NLS-linked BSA (Fig. 4, positive control). Preincubation of the cells with WGA, a lectin that binds to N-acetylglucosamine containing proteins of the NPC (11), prevented detectable nuclear import of the substrate (Fig. 4, negative control).

Phosphorylation-dependent binding of cores to nuclei was quantitated by an ELISA (Table I). As already mentioned, the ELISA detected both types of cores equally well (Fig. 1 A). 17% of the P-rHBc were bound to permeabilized cells. This corresponds to ∼3,400 core particles per nucleus. Of the rHBc, only 4% associated with the permeabilized cells. If the cores were added to cells without prior digitonin permeabilization, a background association of <4% was observed. Analysis of the supernatants after binding verified the presence of intact core particles (data not shown). Thus, unbound core particles were neither disassembled nor degraded during the assay.

To characterize the site of association, binding assays were performed in the presence of agents known to block binding of karyophilic proteins to NPCs. Preincubation with WGA prevented detectable binding of cores and also reduced the signal of the anti-NPC labeling (Fig. 5, WGA). Binding was also blocked by pretreating the permeabilized cells with an excess of anti-NPC antibodies (Fig. 5, anti-NPC). Antibodies against irrelevant proteins, e.g., anti-NS2 (Fig. 2, anti-NS2) or against the cytoplasmic tail of calnexin (20), an integral membrane protein of the ER (data not shown), had no effect. These observations strongly suggested that cores bound to components of the NPC. Note, in Fig. 2, anti-NS2, no labeling for NPC was performed since the competing NS2-antibody was generated in the same species.

To map NLS on the cores, various synthetic peptides were added to the RRL before incubation with cores and cells. Four of the peptides corresponded to positively charged sequences in the COOH-terminal domain of the core protein. They were chosen because they resemble classical nuclear localization sequences found in karyophilic proteins and because they are close to the core phosphorylation sites. They had the following positions within the core protein and amino acid sequences: 142–155 (TLPETTVVRRRDRG), 158–168 (PRRRTPSPRRR), 165–175 (PRRRRSQSPRR), and 173–183 (PRRRRSQSRES). As controls, two peptides corresponding to the known NLS sequences of SV-40 T antigen (STPPKKKRKV) and lamin B2 were used (27).

As shown by immunofluorescence microscopy, four of the peptides blocked binding of P-rHBc to the NPCs. These were peptides 158–168 and 165–175, and both control peptides (Fig. 5, 158–168, 165–175, SV-40TAg-NLS, lamin B2-NLS). The other two peptides 142–155 and 173–183 did not block binding (Fig. 5, 142–155 and 173–183), although peptide 173–183 differed only by two amino acids from peptide 165–175. The similarity between these two peptides with regard to their basic amino acids strongly argues against an unspecific blocking caused by their positive charges. Peptides 158–168 and 165–175 also inhibited nuclear import of SV-40TAg-NLS linked FITC-BSA (Fig. 4). These results suggested that the binding of HBV cores to the NPCs was mediated by factors involved in the nuclear import of proteins containing classical NLS. Furthermore, the data implied that the COOH-terminal domain of the core protein contained sequences that served as a nuclear pore binding signal (NBS).

To determine whether binding of cores was mediated by importins, the experiment was performed in the presence of a buffer complemented with recombinant importins α and β. No other cytosolic components were present. The added importins conferred full binding activity to the P-rHBc (Fig. 6, importin α + β), indicating that core-NPC association was mediated by importins. To confirm that the binding of the cores was mediated by the classical nuclear import pathway that involves both importin α and β, the assay was performed using either importin α or β separately. Neither component alone (Fig. 6, importin α, importin β) could promote binding of P-rHBc to the NPCs. The requirement of both importins suggested that an interaction of importin α to the NBS signal on the core particle surface was required for binding of importin β.

To confirm the binding of importin β, authentic core particles were purified from the supernatant of HepG2.2.15 cells. As P-rHBc and cores from human liver (37), these cores are phosphorylated at serine residues between amino acids 166 and 177, which is a prerequisite for secretion of enveloped virus (58). For our immunofluorescence experiments ∼10¹⁰ particles were used, an amount which cannot be purified from HBV of culture medium (∼10⁶/ml). Therefore, the binding of P-hHBc to importins was studied by coimmunoprecipitation performed with smaller amounts of cores. P-hHBc immobilized to biomagnetic beads by the particle-specific anti-core antibody reaction that precipitated importin β from reticulocyte lysate (Fig. 7 B, lane 1). Comparison with a standard dilution series of importin β (Fig. 7 A) showed that about four importin molecules were associated with one core particle. In vitro phosphorylated P-rHBc also coimmunoprecipitated importin β (Fig. 7 B, lane 3) but not the unphosphorylated rHBc (Fig. 7 B, lane 5).

To confirm that P-hHBc and HBV-derived cores do not interact with importin β directly without importin α, as shown for P-rHBc (Fig. 6), a lamin B2 NLS that binds importin α was added to RRL. The loss of importin β precipitation by P-rHBc and HBV-derived cores (Fig. 7, C and D, fourth lanes) showed the involvement of importin α in the importin β binding to these cores. To further confirm this finding, RRL was replaced by importin β containing buffer. The results documented in Fig. 7, C and D (third lanes), showed that regardless of the origin or phosphorylation status of the core particles no coimmunoprecipitation of importin β with core occurred. In the control reaction, where importin α and β were added, importin β was precipitated by the P-rHBc and HBV-derived cores (Fig. 7, C and D, first lanes).

The above data showed an importin-mediated binding of the core particles to the NPC. To test whether the binding of cores to the NPC occurred so close to the nuclear pore that nuclear import of other substrates was inhibited, nuclei of digitonin-permeabilized cells were pretreated with an excess of P-rHBc. Next, the cells were used in a transport assay with FITC-BSA linked to the M9-domain transport signal of hnRNPs. This substrate was chosen because its import does not require importin α and β involved in P-rHBc binding. A control without pretreatment of the cells with P-rHBc showed nuclear import of this substrate (Fig. 8, control). In contrast, preincubation with P-rHBc abolished the nuclear import of the M9-FITC-BSA (Fig. 8, preincubation with P-rHBc). The bound cores thus prevented uptake of an independently targeted karyophilic substrate.

The target and transport of viral capsids to the nucleus constitutes a key step in the replication cycle of most DNA viruses and some RNA viruses. With the exception of influenza and retroviruses, little information exists about these processes (57). We studied the targeting of HBV cores by applying concepts and in vitro assays originally developed to analyze the cell biology of nuclear import of karyophilic cell proteins. These included the use of digitonin permeabilized cells and binding and uptake experiments with isolated ligands (16, 40).

We found that only phosphorylated HBV cores were specifically targeted to the nucleus by mechanisms and host factors involved in the NLS-mediated uptake of cellular proteins. Attachment occurred to NPCs and blocked the import of other karyophilic substrates, thereby giving evidence for association with the nuclear pore. The binding required the presence of importins α and β. The involvement of importins was confirmed by the coimmunoprecipitation of importin β from reticulocyte lysate using HBV-derived core particles. Thus, the contents of the authentic core particles, e.g., polymerase, hsp90, and viral DNA, had apparently no influence on nuclear binding. However, phosphorylation of the core subunits was necessary for exposure of NBS. The inhibition by specific peptides strongly implied that nuclear binding of HBV cores involved sequences present in the COOH-terminal portion of the core protein.

During core assembly the pregenomic RNA, the viral polymerase (2, 23), hsp90 (25), and PKC (32) are first encapsidated into a particle containing multiple core protein subunits. The COOH-terminal domain of the core protein is highly positively charged. In unphosphorylated cores, cryoelectronmicroscopy has shown that it is located inside the cores close to small holes in the capsid wall (60). It binds to nucleic acid (21, 41) and also provides the phosphorylation site(s) for the protein kinase trapped in the central cavity (21, 31, 32, 36, 37). Since the phosphorylation sites on core proteins compete with the encapsidated RNA (31), phosphorylation of the core protein is most likely linked with DNA synthesis.

Peptide inhibition studies suggested that the region exposed as an NBS may include the sequence between amino acids 158 and 175. As in authentic cores (15, 37), the phosphate groups were not removed by alkaline phosphatase unless the particles were destroyed. Thus, the phosphates most likely remained hidden. This suggests that only the nonphosphorylated part of the COOH terminus was exposed on the particle surface.

Many cases have been described in which phosphorylation either up- or downregulates nuclear transport of proteins. This is known for lamins (35), SV-40 T antigen (46), PKC (4), v-Jun (52), and NF-AT (49). The export of influenza viral RNPs also requires phosphorylation (56). Moreover, it has been suggested that phosphorylation of a small number of M proteins of HIV-1 by kinases trapped in the virion triggers M protein release from the viral membrane and exposure of NLSs close to the phosphorylation sites (12). Core proteins contain an NLS in their COOH terminus (10) overlapping with a nucleic acid binding domain (21) and serine phosphorylation sites (36). Unassembled core proteins, where this sequence is exposed, are imported into the nucleus of HBV-infected cells (36). Inside the nucleus, these imported core proteins spontaneously assemble into core particles (47) devoid of nucleic acids (19).

Our results suggest that phosphorylation of core subunits induces a conformational change that exposes the COOH-terminal NLSs in such a way that the phosphoserine residues remain inaccessible to added phosphatases. The positively charged COOH-terminal domain may dissociate from the nucleic acids inside the core. They may protrude through the neighboring holes in the capsid wall and become partially exposed on the surface to serve as a nuclear targeting signal. This structural change may involve only a fraction of proteins in a core. In the P-rHBc used in this study, on average 12–24 of the 240 core proteins per core were phosphorylated. This was evidently sufficient to target the cores to the nucleus.

Both importin α and β were necessary and sufficient for core binding to NPCs. Thus, HBV cores followed the classical nuclear targeting pathway and not alternative pathways used by protein A1 of U snRNPs (38), M9-like domains of hnRNPs (44), influenza virus nucleoprotein (55), and glycoconjugates (8). Importin α binds to positively charged, T antigen–like NLSs, whereas importin β is thought to bind to this complex and mediate attachment of the complex to the fibrils that extend from the NPCs into the cytosol. This binding reaction is energy independent (for review see 6, 17, 42). With a diameter of >25 nm, the core exceeds the size limit for NLS-coated colloidal gold particles that can enter through nuclear pores (9). Thus, import of DNA should involve disassembly or deformation of the core particle.

We postulate that phosphorylation of core protein is an important control element in the viral life cycle. During encapsidation, the interaction with the COOH-terminal arginine-rich sequences of the core proteins helps to condense the pregenomic RNA into the small space offered by the central cavity (Fig. 9, step 1). The binding of RNA prevents the encapsidated PKC from phosphorylating this core domain (Fig. 9, step 2). During reverse transcription, the RNA is degraded. This allows the PKC to phosphorylate the COOH-terminal domain of some core protein subunits (Fig. 9, step 3). Phosphorylation induces a conformational alteration in the core proteins. As a consequence, the COOH terminus is exposed on the particle surface (Fig. 9, step 4) that serves as a nuclear pore targeting signal.

By association with importins α and β (Fig. 9, step 5), the cores are targeted to the nuclear membrane and bind to the nuclear pores (Fig. 9, step 6). There, they release their DNA/polymerase complex by unknown mechanisms (Fig. 9, step 7), which is transported into the nucleoplasm (Fig. 9, step 8). Alternatively, the viral cores can undergo budding at the intermediate compartment between the ER and the Golgi apparatus leading to virus formation (Fig. 9, step 9). When the viral cores enter new host cells (Fig. 9, step 10), they can again make use of the importins and the classical pathway of nuclear import. One of the potential advantages of this strategy is that as long as the viral DNA is packaged it can be moved through the cytosol to reach its defined targets. Therefore, the release of the bulky DNA molecule occurs only upon reaching the nuclear envelope. The molecular mechanisms of this pathway can now be elucidated experimentally.

We thank Drs. Maria Seifer (Schering-Plough., Kenilworth, NJ) for performing core ELISAs, Ursula Friedrich for synthesis of the peptides, Monica and Dietmar Linder for MALDI analysis of the peptides, Paul Pumpens and Galina Borisova for providing the purified E. coli–derived core particles, Stefan Schmehl (Institute for Medical Virology, Giessen, Germany) for purification of the HBV from HepG2.2.15 cells, Dirk Görlich for importin α and β and the anti-importin β antibody, and Päivi Ojala (University of Helsinki, Helsinki, Finland) and Gary Whittaker (Cornell University, Ithaca, NY) for many stimulating discussions.

This work was supported by a grant of the Deutsche Forschungsgemeinschaft to M. Kann (Ka 1193/1-1 and SFB 535, B5) and W.H. Gerlich (SFB 272, C4); B. Sodeik and A. Helenius were supported by a grant from the National Institutes of Health to A. Helenius (AI 18599) and a European Molecular Biology Organization postdoctoral fellowship to B. Sodeik (ALTP 2S4-1993).