|Home | About | Journals | Submit | Contact Us | Français|
The molecular basis underlying hepatitis C virus (HCV) core protein maturation and morphogenesis remains elusive. We characterized the concerted events associated with core protein multimerization and interaction with membranes. Analyses of core proteins expressed from a subgenomic system showed that the signal sequence located between the core and envelope glycoprotein E1 is critical for core association with endoplasmic reticula (ER)/late endosomes and the core's envelopment by membranes, which was judged by the core's acquisition of resistance to proteinase K digestion. Despite exerting an inhibitory effect on the core's association with membranes, (Z-LL)2-ketone, a specific inhibitor of signal peptide peptidase (SPP), did not affect core multimeric complex formation, suggesting that oligomeric core complex formation proceeds prior to or upon core attachment to membranes. Protease-resistant core complexes that contained both innate and processed proteins were detected in the presence of (Z-LL)2-ketone, implying that core envelopment occurs after intramembrane cleavage. Mutations of the core that prevent signal peptide cleavage or coexpression with an SPP loss-of-function D219A mutant decreased the core's envelopment, demonstrating that SPP-mediated cleavage is required for core envelopment. Analyses of core mutants with a deletion in domain I revealed that this domain contains sequences crucial for core envelopment. The core proteins expressed by infectious JFH1 and Jc1 RNAs in Huh7 cells also assembled into a multimeric complex, associated with ER/late-endosomal membranes, and were enveloped by membranes. Treatment with (Z-LL)2-ketone or coexpression with D219A mutant SPP interfered with both core envelopment and infectious HCV production, indicating a critical role of core envelopment in HCV morphogenesis. The results provide mechanistic insights into the sequential and coordinated processes during the association of the HCV core protein with membranes in the early phase of virus maturation and morphogenesis.
The hepatitis C virus (HCV), a major cause of chronic hepatitis, liver cirrhosis, and hepatocellular carcinoma, is an enveloped, single-stranded, positive-sense RNA virus in the Hepacivirus genus of the Flaviviridae family, which also includes the flaviviruses and pestiviruses (42; for a review, see reference 5). The HCV genome contains a single open reading frame flanked by nontranslational regions at its 5′ and 3′ ends and encodes for an ca. 3,000-amino-acid precursor polyprotein. The polyprotein is cotranslationally and/or posttranslationally processed by cellular and viral proteases into mature structural proteins including core and envelope (E) glycoproteins E1 and E2, p7, and the nonstructural (NS) proteins of NS2, NS3, NS4A, NS4B, NS5A, and NS5B (for a review, see reference 68). The core protein is a multifunctional molecule that constitutes the viral nucleocapsid and is also involved in the pathogenesis and carcinogenesis of HCV (for reviews, see references 40, 53 and 71). The structural envelope glycoproteins, E1 and E2, mediate attachment of the virus to receptor(s) on host cells and acidic pH-dependent membrane fusion between the viral envelope and endosomal membranes. NS proteins, although not thought to be assembled into virus particles, participate in viral replication.
Based on the hydrophobicity and clustering of basic amino acids, the core protein consists of three distinct predicted domains (Fig. (Fig.1A,1A, panel 1): a basic and hydrophilic region covering two-thirds of the N-terminal portion (domain I; residues 1 to 118); a hydrophobic domain (domain II; residues 119 to 173) covering the central one-third; and the last hydrophobic signal sequence of the downstream protein E1 (domain III; residues 174 to 191) (for a review, see reference 40). The mature core is a dimeric, α-helical protein with membrane-associated features (12). Domain I is mainly involved in RNA binding and encapsidation of the viral genome within a nucleocapsid particle and in homotypic interactions necessary for particle formation (for reviews, see references 40 and 45). Domain I also contains three basic residue stretches of Arg- and Lys-rich sequences, which function as nuclear localization signals, to mediate core localization in the nucleus (32, 62, 77). Domain II is not only required for proper folding of domain I but is also critical for the membrane property of the core including core association with endoplasmic reticula (ER) and outer mitochondrial membranes (12, 63). On the other hand, core proteins, via domain II, target lipid storage droplets (3, 23, 41), where they colocalize with apolipoprotein AII and directly associate with triglycerides (3). Two amphipathic α-helices and the intermediate hydrophobic loop, all of which constitute major structural elements within domain II, were shown to be essential for the association with lipid droplets (LDs), which points to an in-plane membrane interaction of the two helices at the lipid bilayer interface (11). This LD targeting event has been implicated in HCV replication and/or pathogenesis (40, 43), and the LD association site in domain II is a critical determinant of the efficiency of HCV assembly (59).
Maturation of the HCV is thought to proceed through budding of nascent particles into the ER lumen, and then viral particles are transported via a constitutive secretory pathway to the extracellular milieu. The internal signal sequence located between the core and E1 proteins targets the nascent polypeptide to ER membranes and induces translocation of E1 into ER. After membrane insertion, the signal peptide is first cleaved from E1 by a signal peptidase to yield the 191-amino-acid core precursor, p23. This initial cleavage is subsequently followed by a second cleavage catalyzed by an intramembrane-cleaving protease, termed the signal peptide peptidase (SPP) (29, 41, 50), to yield the mature core p21 (41). This mature p21 form predominates in both transfected cell cultures and virus particles from infected sera (77). Once cleaved from the polyprotein, the p21 core is thought to assemble into capsids at the cytoplasmic face of the ER and acquires its envelopes by budding into the ER (6-8, 33, 34), during which time the E1 and E2 proteins are embedded into the viral envelope.
The morphogenesis of the core protein coexpressed with or without other structural proteins has been studied in bacteria (28), yeast (1), insect cells (6), and cell-free systems (25). Many mammalian cell culture systems have also been developed to study the capsid assembly pathway. Although HCV capsids can be detected in many mammalian cells using various expression systems by electron microscopy, core assembly into capsids appears to be inefficient, since no virus-like particles (VLPs) are detected at all (7, 8, 18), presumably due to the instability of core complexes in further isolation, quantification, and biochemical studies. It was reported that the core protein expressed in cells or the purified recombinant core protein self-assembles into nucleocapsid-like particles which require the presence of stem-loop RNA structures, such as those in tRNA (28, 48). Moreover, structural proteins including the core are expressed in various replicon systems using the full-length HCV genome; nonetheless, evidence for virion assembly and release from human liver cell lines was not available (9, 10, 24, 35, 55) until recently when cloned HCV genomes, based on the JFH1 clone isolated from a patient with fulminant hepatitis C, which allowed robust replication of infectious HCV in tissue cultures (HCVcc), were developed (31, 70, 78). The advancement in HCV tissue culture systems has tremendously facilitated our understanding of the molecular mechanisms underlying the HCV replication cycle and virus-cell interactions.
In the present study, we first establish a core subgenomic expression system to dissect several coordinated, key events in the early phase of the HCV core maturation process. The results also demonstrate that association with membranes and SPP-catalyzed cleavage at the C terminus of the core are necessary but not sufficient for core envelopment and that domain I also contains sequences important for core envelopment. We also show that the core protein expressed in the robust HCVcc replication system undergoes several key steps in the early phase of core maturation processes similar to those utilized by the core when expressed alone. In addition, the results from treatment with the SPP-specific inhibitor, (Z-LL)2-ketone, or coexpression with the loss-of-function SPP D219A mutant reveals a critical role of core envelopment in HCV propagation. Our study thus reveals the sequential and concerted processes of core multimerization and interactions with and envelopment by membranes during the HCV assembly and/or budding process. To our best knowledge, this is the first demonstration of this kind.
293T and Huh7 cells were cultured in Dulbecco modified Eagle medium containing 10% heat-inactivated fetal bovine serum. In particular, nonessential amino acids were added to the Huh7 culture medium. To obtain anti-HCV core antiserum, the coding sequence of residues 2 to 173 of the core of the genotype 1 H77 strain was cloned in a pET21(a) vector (Merck Biosciences-Novagen; Darmstadt, Germany). The fusion protein with a His6 tag attached at the N terminus of the core was purified by BD Talon affinity metal resin (BD Biosciences; San Diego, CA), and the purified protein was used to immunize New Zealand White rabbits. A mouse monoclonal antibody (MAb) for the influenza virus hemagglutinin (HA) tag (HA.11, clone 16B12) was purchased from Covance (Richmond, CA). A MAb specific for HCV core (clone C7-50) was obtained from Affinity BioReagents (Golden, CO). MAbs directed against HCV E1 (clone BD1198) and E2 (clone BD1167) were purchased from Biodesign (Saco, ME). A MAb against HCV E2 (clone HCM-0919-5) was purchased from Austral Biologicals (San Ramon, CA). Rabbit anti-HCV E2 peptide antibody (catalog no. GB10416) was from Genesis Biotech (Xindian, Taipei County, Taiwan). A MAb against DsRed (clone 64-1077) was purchased from BD Biosciences. Rabbit anti-green fluorescent protein (GFP) immunoglobulin G (IgG), goat anti-β-Gal IgG, and goat anti-TIP47 IgG (N-15) were all obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Affinity-purified fluorescein isothiocyanate (FITC)-conjugated secondary antibodies were purchased from Kirkegaard & Perry Laboratories (Gaithersburg, MD). 1,3-Di-(N-carboxybenzoyl-l-leucyl-l-leucyl)amino acetone [(Z-LL)2-ketone] was purchased from Merck Biosciences-Calbiochem (Darmstadt, Germany).
To construct the pCAGGS-based core plasmids encoding amino acid residues 1 to 191, 1 to 173, 1 to 131, and 1 to 99 with the HA peptide, YPYDVPDYA, tagging the N terminus (Fig. (Fig.1A,1A, panel 1), the HACore-coding sequences were amplified from pBRTM-HCV-H1-3011, a full-length HCV genotype 1a clone, through a PCR using HA342f-EcoRI as the sense primer and 914r-EcoRI, 860r-EcoRI, 735r-EcoRI, and 639r-EcoRI, respectively, as the antisense primers. The number in the name of a primer indicates the position of the first nucleotide of the primer in the sequence of the pBRTM-HCV-H1-3011 clone. EcoRI-restricted DNA fragments were then inserted into pCAGGS at the same site. To construct pCAGGS-HACore-A180V, -S183L, -C184V, and -ASC180/3/4VLV core mutant plasmids (Fig. (Fig.1A,1A, panel 2), which encoded core variants with a single substitution of Val, Leu, and Val for Ala-180, Ser-183, and Cys-184, respectively, or all three substitutions for the three residues, HA342f-EcoRI was used as the sense primer and A180Vr-EcoRI, S183Lr-EcoRI, C184Vr-EcoRI, and ASC180/3/4VLVr-EcoRI, respectively, were used as the antisense primers in the PCR. The EcoRI-cut DNA fragments were cloned into the pCAGGS vector to generate mutant core plasmids. The pCAGGS-based core plasmids expressing the Δ88-106 and Δ42-68 mutants (Fig. (Fig.1A,1A, panel 3) were, respectively, PCR amplified using del88-106f and del88-106r, and del42-68f and del42-68r, as the paired primers. To construct pcDNA3-HACE1E2, the sense and antisense primers in the PCR were HA342f-HindIII and 2579r-XbaI, respectively, and the HindIII- and XbaI-cut DNA fragments were cloned into the pcDNA3 vector (Invitrogen). pcDNA3-HACE1E2 was then cut with HindIII and XbaI and end blunted with T4 DNA polymerase, and the insert was ligated to the end-blunted EcoRI site in pCAGGS to generate pCAGGS-HACE1E2. Kozak-342f-HindIII and 914r-XbaI were used as primers to generate the DNA fragment, which was then cloned into pcDNA3 at the HindIII and XbaI sites to generate pcDNA3-HACore. Kozak-735f-EcoRI and 2579r-XbaI were used as primers and pBRTM-HCV-H1-3011 as the template to generate the E1E2 insert, which was then cloned into the corresponding sites in pcDNA3 to yield pcDNA3-E1E2. The sequences of oligonucleotides used as PCR primers are available upon request.
Subconfluent 293T cells grown on 6- or 10-cm petri dishes were transfected with 5 or 10 μg of core- or CE1E2-expressing plasmids, respectively, using a standard calcium phosphate coprecipitation method (54). Unless otherwise indicated, the DNA and calcium phosphate complexes were washed away 6 to 8 h after transfection. Huh7 cells were transfected with DNA plasmids by using the Arrest-in transfection method (Openbiosystems, Huntsville, AL) in accordance with the manufacturer's procedures. Unless otherwise indicated, all transfected 293T and Huh7 cells were harvested 48 h after transfection for analyses.
The JFH1 and chimeric F6/JFH1 (also termed Jc1) RNA genomes were synthesized in vitro as previously described (70). In brief, the JFH1 and Jc1 plasmids were first restricted with XbaI and treated with mung bean nuclease (New England Biolabs; Beverly, MA) to remove the four terminal nucleotides, resulting in the correct 3′ end of the HCV cDNAs. Digested plasmid DNAs were then used as a template for in vitro RNA synthesis by using a MEGAscript T7 kit (Ambion, Austin, TX). Synthesized RNAs were treated with DNase I and purified by using an RNeasy minikit (Qiagen, Valencia, CA). Huh7 cells were transfected with 10 μg of RNA by electroporation using a BTX Model ECM 630 Electro cell manipulator (Holliston, MA) with settings at 240 V and 975 μF. Cells were harvested 72 h posttransfection for various biochemical analyses. Alternatively, 16 h after JFH1 RNA transfection, Huh7 cells were transfected with plasmid DNA by the Arrest-in transfection method described above.
Culture supernatants obtained from Huh7 cells electroporated with HCV RNAs were passed through a filter with a 0.45-μm pore size and fivefold serially diluted with fresh medium supplemented with 20 mM HEPES [4-(2-hydroxyethyl)-1-1-piperazine ethanesulfonic acid] and 8 μg of Polybrene/ml. The supernatants were then used to infect Huh7 cells grown in 96-well plates. The inoculum was removed at 8 h postinfection. At 72 h postinoculation, the cells were processed for indirect immunofluorescence staining. First, cells were fixed with methanol at −20°C, followed by blocking. Cells were then successively incubated with an NS5A (clone 9E10) MAb and FITC-conjugated anti-mouse IgG. The immunostained cells were scored for NS5A foci under an inverted immunofluorescence microscope (Eclipse TE2000-U; Nikon, Tokyo, Japan). The virus titer was expressed as focus-forming units per milliliter determined by the average number of NS5A-positive foci.
Lysates obtained from transfected cells were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), followed by Western blotting with appropriate primary and secondary antibodies and then enhanced chemiluminescent detection. Alternatively, proteins in each fraction after density gradient centrifugation or subcellular fractionation were precipitated with 10% cold trichloroacetic acid prior to SDS-PAGE. An HA MAb was used as the primary antibody to detect core proteins by Western blotting unless otherwise indicated. For quantitation of the core amounts, images within the linear range were scanned by using a Microtek ScanMaker 8700 (Zntennation; Hsinchu, Taiwan) and quantified by using MetaMorph (Molecular Devices; Sunnyvale, CA).
Transfected cells were washed with cold phosphate-buffered saline (PBS) twice, and lysed with PBS containing 1% Nonidet P-40 (NP-40), 1% sodium deoxycholate, and a protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany). After centrifugation at 18,000 × g for 30 min at 4°C, cell lysates were loaded onto 5 to 50% (wt/vol) linear sucrose gradients, and the gradients were centrifuged at 100,000 × g for 16 h at 4°C as previously described (2). After centrifugation, samples were fractionated from the bottom of the gradients, and 1 ml per fraction was collected.
For membrane flotation, previously described procedures (15) were followed. In brief, PBS-washed transfected cells were sonicated twice, each time for 15 s, in 0.5 ml of hypotonic TE buffer (10 mM Tris-HCl [pH 7.5] containing 1 mM EDTA and Complete protease inhibitor cocktail) supplemented with 10% (wt/vol) sucrose on ice. After low-speed centrifugation to remove nuclei, the postnuclear supernatants (0.25 ml) were mixed with 1.25 ml of 85.5% sucrose prepared in TE buffer and placed at the bottom of a Beckman SW41 ultracentrifuge tube, and then 7 ml of 65% sucrose and 3.25 ml of 6% sucrose (both prepared in TE buffer) were successively loaded into the tube. The gradients were centrifuged at 100,000 × g for 18 h at 4°C. Samples were then fractionated from the bottom of the gradients, and 1.2 ml per fraction was collected.
For iodixanol (Axis-Shield, Oslo, Norway)-based subcellular fractionation studies, a previously reported procedure (22, 26) was followed. Briefly, transfected cells were sonicated in 1 ml of 10 mM Tris-HCl (pH 7.5) containing 0.25 M sucrose, 1 mM EDTA, and protease inhibitor cocktail. After removal of the nuclei, the postnuclear supernatants were subjected to subcellular fractionation to separate different membrane compartments. First, 0.33 ml of the postnuclear supernatants was mixed with 0.67 ml of 60% Optiprep and then transferred to the bottom of a Beckman SW60 centrifuge tube, which was successively overlaid with 1 ml each of 30%, 20%, and 10% Optiprep. The gradients were centrifuged at 337,000 × g in a Beckman SW60 rotor (50,000 rpm) for 3 h at 4°C. Fractions (180 μl) were collected from the top of the gradient.
In vitro-coupled transcription/translation was performed using the reticulocyte-based TNT quick coupled transcription/translation kit (Promega; Madison, WI) with 1 μg of pcDNA3 plasmids in a 25-μl reaction mixture. For protein synthesis in the presence of membranes, 1.5 μl of a canine pancreatic microsomal membrane (Promega) was added per 25 μl of reaction mixture. After incubation at 30°C for 90 min, a final concentration of 0.1 mg of cycloheximide/ml was added to stop the reaction.
A membrane protection assay was performed as previously indicated (15). For analysis of proteins expressed in cells, transfected 293T cells were sonicated in 50 mM Tris-HCl (pH 8.0), containing 10 mM CaCl2, 1 mM dithiothreitol, and protease inhibitor cocktail. JFH1 or Jc1 RNA-transfected Huh7 cells were passed through a 27-gauge syringe 30 times. The postnuclear supernatants or samples obtained from in vitro synthesis were divided into three portions. One portion received no treatment. The second portion was treated with 50 μg (for proteins expressed in 293T cells) or 10 μg (for proteins expressed in Huh7 or in vitro) of proteinase K/ml. The third portion was solubilized with a final concentration of 1% Triton X-100 before proteinase K treatment. After incubation on ice for 1 h, samples were supplemented with a final concentration of 5 mM phenylmethylsulfonyl fluoride (dissolved in ethanol), followed by incubation on ice for 10 min. All samples were solubilized with detergents, boiled at 95°C for 10 min, and then analyzed by SDS-PAGE followed by Western blotting.
In vitro-translated core proteins were incubated with or without 3 μl of microsomal membranes in 250 μl of PBS containing 120 mM potassium acetate, 5 mM magnesium acetate, 1 mM dithiothreitol, and 5 mM ATP at 30°C for 3 h. Alternatively, detergent-lysed cell extracts containing core proteins were first subjected to 5 to 50% linear density sucrose gradient centrifugation. The bottom fractions containing high-ordered, multimeric core complexes were pooled, and the concentrated core complexes were incubated with or without microsomal membranes. The reaction mixtures were then mixed with 1.25 ml of 85.5% sucrose prepared in TE buffer and resolved by a membrane flotation analysis as described above.
Transfected 293T cells were sonicated with the proteinase K digestion buffer, and the postnuclear fractions were divided into three parts. The first portion was incubated with 5 μl of rabbit preimmune serum, the second portion was incubated with 5 μl of rabbit anti-core (for 191 and 131 core transfection) or with 1 μg of affinity-purified rabbit anti-E2 (for HACE1E2 transfection), and the third portion was treated with 1% each of NP-40 and sodium deoxycholate prior to incubation with rabbit anti-core or anti-E2. Then, 30 μl of 10% washed protein A-Sepharose beads was added to each mixture, followed by additional incubation. The precipitated antigens were resolved by SDS-PAGE, followed by Western blotting analysis.
Huh7 cells were transfected with JFH1 RNA, and at 72 h posttransfection the cells were washed with PBS and fixed with 4% paraformaldehyde (prepared in PBS) at room temperature for 30 min. The cells were then incubated with PBS alone or with PBS containing 0.5% each of saponin or Triton X-100 at room temperature for 15 min. After three washes with PBS and blocking with PBS containing 1% fetal bovine serum and 5% bovine serum albumin, the cells were incubated with rabbit preimmune serum or an anti-core, followed by incubation with FITC-conjugated goat anti-rabbit IgG. The immunostained cells were quantitated by fluorescence-activated cell sorter (FACS) analysis using a FACSCalibur (Becton Dickinson Immunocytometry Systems, San Jose, CA).
To provide mechanistic insights into the HCV core protein maturation process, we first examined the expression, intracellular localization, and maturation of the full-length HCV H77 strain core protein containing residues 1 to 191, which includes the signal sequence, i.e., residues 174 to 191 at the C terminus, as well as core proteins with a series of truncations from the C terminus (Fig. (Fig.1A,1A, panel 1) in 293T cells. To expedite detection of the core protein, all core proteins encoded by the pCAGGS, a cytomegalovirus immediately-early enhancer/chicken β-actin promoter-driven vector (13), were tagged with an HA epitope prefaced by a start codon at their N termini (Fig. (Fig.1A,1A, panel 1). The shorter forms of core proteins were effectively expressed as the 1~191 core (data not shown). In a chemical cross-linking analysis with glutaraldehyde, all of the truncated core proteins, as well as the 191 core exhibited ladderlike, oligomeric forms (data not shown).
When the postnuclear extract containing the 191 core was subjected to 5 to 50% linear density sucrose gradient centrifugation, the 191 core was mainly detected in the bottom one to four fractions (data not shown), which had a buoyant density ranging 1.148 to 1.192 g/ml, similar to those found for VLPs produced by heterologous expression systems (18). The core complexes located in top fractions 6 to 12 might represent membrane-associated multimeric structures, since treatment with 1% each of NP-40 and sodium deoxycholate prior to gradient centrifugation significantly reduced their distribution in these fractions (data not shown). The cell-free human immunodeficiency virus type 1 immature Gag VLPs were located in the bottom fractions 1 to 4, with a peak buoyant density of 1.164 g/ml (data not shown) as previously reported (27, 75). Therefore, the species detected in bottom fractions 1 to 4 represent high-ordered, multimeric core complexes (data not shown). In addition, the shorter core proteins were all distributed in bottom fractions 1 to 4 (data not shown), indicating that the C-terminal segment of core is not critical for core multimerization into high-ordered structural complexes.
In these analyses, we found that the 191 and 173 core proteins comigrated on SDS-PAGE (data not shown). To provide evidence for SPP-mediated conversion of the 1-191 innate p23 to the processed p21, a pulse-chase was performed. With a 15-min pulse time, a fraction of p23 was already processed to p21; nevertheless, the p23 form was still predominant (data not shown). After a 5-min chase, the level of innate p23 had decreased, which was accompanied by an increase in the level of the processed p21, which was at a slightly higher level than that of p23 (data not shown). The instability of the HCV core observed after chasing for 5 min (data not shown) was in line with previous results that degradation of the HCV core protein (64) was mediated by E6AP ubiquitin ligase (60) or the proteasome activator, PA28γ (46). Cotransfection of the 191 core plasmid with or without pcDNA3-SPP(D219A)-HA (50), which encodes an Asp-to-Ala substitution for Asp-219, one of two protease-active sites in SPP, revealed unprocessed p23 and processed p21 forms, respectively (data not shown). These observations show the precursor-product relationship of p23 and p21.
To determine the intracellular transport of the core complex, a cell fractionation method using iodixanol-based, isopycnic linear density gradient centrifugation (22, 26) was used. As a control, the human immunodeficiency virus type 1 Env glycoproteins gp160, gp120, and gp41 were mainly distributed in the ER/late-endosomal compartments, while a significant fraction of gp41 was also detected in plasma membranes (data not shown), which is consistent with the intracellular transport and maturation process of Env. The Pr55 Gag precursor was recovered from all four fractions (data not shown), in accordance with the maturation process of Gag after its biosynthesis in the cytoplasm and targeting to the plasma membrane for budding. Cellubrevin and TI-VAMP (vesicle-associated membrane protein) are markers for recycled endosomes and late endosomes/secretory vesicles, respectively (17, 37). As previously shown (22), GFP-Cellubrevin was mainly associated with plasma membranes, ER/late endosomes, and small vesicles, whereas GFP-TI-VAMP was predominantly associated with small vesicles and ER/late endosomes (data not shown). Lysosome-associated membrane protein (LAMP) 1, a late endosome marker, was mainly associated with endosomal and plasma membranes (data not shown). Rab11, which was detected in the trans-Golgi network, secretory vesicles, and the pericentriolar recycling endosomes, was recovered from endosomal membranes, small vesicles, and plasma membranes (data not shown). The distribution profiles of LAMP1 and Rab11 were similar to those previously reported (26).
When the postnuclear fraction of the 191 core was analyzed, the processed 191 core was primarily recovered from ER/late endosomes and small vesicles, and a small amount of this protein was associated with plasma membranes (data not shown). Only a small amount of the 191 core was recovered from soluble proteins (data not shown). A similar distribution pattern of the 191 core associated with intracellular membranes was obtained when the core was expressed in COS-1 cells (data not shown). In contrast, the 173, 131, and 99 core proteins were primarily associated with soluble fractions, and a fraction of these proteins was associated with small vesicle membranes (data not shown). Also, a small amount of 173 core protein was transported to plasma membranes (data not shown). Nevertheless, these truncated core proteins were not efficiently associated with ER/late endosomes, indicating that the signal sequence is critical for core association with ER/late endosomes.
We then examined the intracellular localization of recombinant core complexes using confocal immunofluorescence microscopy with known intracellular organelle markers. The 191 core protein exhibited a finely reticular perinuclear staining pattern and also formed granules disseminated throughout the cytoplasm (data not shown). Both the perinuclear and the cytoplasmic staining showed “ringlike” and “dot” structures. The 173, 131, and 99 core proteins were primarily localized to the nucleus, and only a small percentage of these proteins formed patches or speckles in the cytoplasm (data not shown), which is consistent with previous findings that removal of the signal sequence between the C and E1 proteins causes the core to detour from the cytoplasm to the nucleus (14, 46, 50, 57, 62). We also observed a considerable degree of colocalization of the 191 core protein with calnexin, an ER marker, GM130, a Golgi marker, and with LAMP3, a late endosome marker (data not shown). Nevertheless, the 191 core protein did not colocalize with EEA1, an early endosome marker (data not shown).
The core protein expressed from subgenomic-expressing systems and in the context of JFH1 RNA replication were shown to target LDs, which exist in virtually all eukaryotic cells and even in some prokaryotes (for reviews, see references 47 and 52). A confocal microscopic analysis showed that a significant fraction of the cytoplasmic 191 core which formed a “ring” structure was localized at the surface of LDs, which were stained with the LD marker, BODIPY 493/503, whereas those core complexes present as “punctates” or “dots” did not seem to be associated with LDs (data not shown). The cytoplasmic 173 core was predominantly localized at the surface of LDs, whereas the cytoplasmic 131 and 99 core proteins were barely detected there (data not shown). The LD association property of 191 and 173 core proteins was also observed when these core proteins were expressed in Huh7 cells (data not shown), a hepatoma cell line which supports HCV replication.
To understand the relationship between membrane-free and -bound 191 core proteins, DNA-calcium phosphate coprecipitates were washed away 4 h posttransfection, and the cultures were left to recover in fresh medium for an additional 1, 2, 16, and 20 h. This type of time course study was previously used to monitor intracellular trafficking of the Marburg virus matrix protein VP40 between cellular compartments (26). The percentage of the 191 core distributed in the soluble fraction of the total core population decreased as the cultures were harvested and analyzed at later posttransfection times (Fig. (Fig.1B),1B), revealing a progressive shift in the 191 core from soluble to membrane fractions as the posttransfection time proceeded.
To understand whether the structural organization of membrane-free and -associated 191 core proteins in a steady state differs, which thus affects their membrane association properties, a cell-based protein-membrane interaction assay, which separates membrane proteins from cytosolic proteins (16, 61), was used. As previously reported (39), the 191 core could float to membrane fractions 7 and 8 (data not shown), which corresponded to the 6 and 65% sucrose interfaces. When the membrane-associated fractions 7 and 8 and cytosolic fractions 1 to 3 of the 191 core after membrane flotation were separately collected, concentrated, supplemented with a final concentration of 1% Triton X-100, incubated at 37°C for 10 min, and then subjected to a 5 to 50% linear sucrose gradient centrifugation, they all predominantly formed high-ordered, multimeric structures (Fig. (Fig.1C,1C, panels 1 and 3, respectively). In contrast, these membrane-free and -associated core proteins were both located in the top fractions of the gradient when treated with 1% SDS prior to sucrose gradient centrifugation, (Fig. (Fig.1C,1C, panels 2 and 4, respectively), indicating that SDS not only interferes with core-membrane interactions but also strongly disrupts intersubunit interactions of high-ordered, multimeric core complexes, resulting in monomeric core molecules.
The core protein translated in vitro provides a ready system to study SPP-mediated signal anchor processing (44, 57, 73, 77). To understand whether the 191 core synthesized in vitro can be enveloped, the 1-191 core translated in vitro was subjected to a standard membrane protection assay. If the core is bound to the inner leaflet of vesicles which are right side out, it will be protected from proteolysis, whereas the cytosolic core and core proteins present on inside-out vesicles are protease sensitive. In an in vitro transcription/translation reaction that synthesized E1 and E2, the E1 translated in the presence of a microsomal membrane was resistant to proteinase K digestion; however, treatment with detergent prior to proteolysis rendered E1 sensitive to proteinase K digestion (Fig. (Fig.2A,2A, lanes 1 to 3), indicating that the synthesized E1 had been translocated across the membrane. Although the in vitro-synthesized core was processed by SPP in the presence of membranes (Fig. (Fig.2A,2A, compare lanes 4 and 7), it was not enveloped, as judged by its sensitivity to protease digestion (Fig. (Fig.2A,2A, compare lanes 7 and 8) (25).
We next determined whether the intracellular 191 core complex was enveloped by membranes. Previous studies suggested that HCV capsids acquire an envelope by interacting with E proteins and budding through the ER membrane to form VLPs (7). Thus, the core and E1 proteins encoded by pCAGGS-HACE1E2 were included as a control. The majority of the coexpressed core and E1 proteins and the core protein expressed alone were both resistant to proteinase K treatment in the absence of Triton X-100; however, they all became protease sensitive when treated with detergent before protease digestion (Fig. (Fig.2B).2B). To strengthen the viewpoint that the core expressed in cells is enveloped by membranes, the ability of core- and E2-specific antibodies to bind to these two proteins was assessed. The E2 antibody was able to bind E2 and coprecipitate the core if the postnuclear fraction was treated with NP-40 and sodium deoxycholate prior to immunoprecipitation (Fig. (Fig.2C,2C, lane 9). However, this antibody bound to E2 at a much smaller degree in the postnuclear fraction without detergent treatment (Fig. (Fig.2C,2C, compare lanes 8 and 9). These observations indicate that the majority of E2 and core proteins are enveloped inside membrane vesicles, that E2 is accessible to anti-E2 upon disruption of the membrane by detergents, and that precipitation of the core by anti-E2 occurs via its association with E1 in the E1-E2 heterodimer. Moreover, the anti-core antibody was able to precipitate a considerable amount of the 131 core, but not the 191 core, even when the membrane remained intact (Fig. (Fig.2C,2C, compare lanes 2 and 5). Nevertheless, the anti-core was able to bind the 191 core upon disruption of the membranes by detergents (Fig. (Fig.2C,2C, lane 3).
When the sensitivities of the 191 and other truncated core proteins to proteinase K were assessed and quantified, 82% of the 191 core molecules were resistant to protease treatment in the absence of Triton X-100 (Fig. (Fig.2D,2D, compare lanes 1 and 2), whereas detergent treatment abolished the protease resistance of the processed 191 core (Fig. (Fig.2D,2D, lane 3). Although the other three truncated core proteins possibly remained associated with small vesicles (data not shown), they were predominantly associated with cytoplasmic leaflets of right-side-out vesicles, as evidenced by their sensitivity to proteolysis even in the absence of detergent (Fig. (Fig.2D,2D, lanes 4 to 12). These results indicate that most of the 191 core molecules coexpressed with E1 and E2 proteins or expressed alone in cells are enveloped by membranes, and also suggest that the signal sequence located at the C terminus of the core is required for core envelopment in cells.
To determine whether the high-ordered, multimeric core complexes can bind to membranes, the membrane-free multimeric 191 core and cytosolic 173 core complex after treatment with detergents were first isolated by sucrose gradient centrifugation and then incubated with or without microsomal membranes prior to membrane flotation. The cytosolic 191 and 173 complexes could float to the 6 and 65% sucrose interfaces only after incubation with membranes (Fig. (Fig.3A,3A, compare panels 2 and 4 to panels 1 and 3, respectively). The 191 core translated in vitro in the presence of membranes was also associated with membranes (Fig. (Fig.3A,3A, compare panels 5 and 6). Nevertheless, all of the membrane-free and -bound processed 191 and 173 core complexes were sensitive to protease digestion (Fig. (Fig.3B).3B). These results indicate that in spite of their ability to bind to membranes, the preassembled 191 and 173 core complexes were not enveloped by membranes in this reconstituted in vitro membrane binding assay.
Next, we determined whether the structural organization of the core expressed in cells and in vitro might differ. In contrast to the core complex expressed in cells, which were located in the bottom fractions 1 to 4 of the gradient, the core proteins synthesized in vitro in the presence or absence of microsomal membrane all peaked at fraction 5 (Fig. (Fig.3C),3C), indicating that the in vitro-translated core assembles into a complex with a buoyant density lighter than that of the core complex synthesized in cells.
To understand the role of cleavage of the membrane-anchoring domain in membrane association of the core complex, cells expressing the 191 core were treated with or without (Z-LL)2-ketone, a cysteine protease inhibitor which specifically targets SPP intramembrane cleavage activity (29, 41, 73). Unlike the core expressed without (Z-LL)2-ketone, the innate core expressed in the presence of the SPP inhibitor predominated over the processed form and was primarily recovered from the soluble protein fraction, while only a small fraction of the innate core was coassociated in ER/late endosomes with the processed core (Fig. (Fig.4A),4A), the cleavage of which by SPP was apparently incompletely blocked by the inhibitor under these experimental conditions. It was also noted that this SPP inhibitor was not effective at inhibiting SPP activity (76).
In the membrane protection assay, a small fraction of the innate and processed core proteins synthesized in the presence of the SPP inhibitor was still resistant to proteolysis despite the level of proteinase K resistance of the innate and processed forms being lower than that without inhibitor treatment (Fig. (Fig.4B,4B, compare lanes 4 and 5 to lanes 1 and 2). This result shows that the inhibition of SPP activity decreased the membrane association of the core and the core's subsequent envelopment. Nevertheless, the innate and processed core proteins synthesized in the presence of the inhibitor were also localized in the bottom fractions of the sucrose linear density gradient (Fig. (Fig.4C),4C), indicating that treatment with the SPP inhibitor did not affect high-ordered, multimeric core complex formation.
To further understand the effects of signal peptide cleavage on core envelopment, core variants with mutations in the signal peptide were examined. Ala-180, Ser-183, and Cys-184 in the signal sequence were individually replaced by Val, Leu, and Val, respectively, or replaced with a combination of these residues (Fig. (Fig.1A,1A, panel 2). The triple substitutions were designed to disrupt a critical helix break required for SPP processing (29, 41). The two mutant proteins, A180V and C184V, were fully processed, as judged by their comigration with the processed wild-type (WT) p21 (data not shown). The triple mutant, ASC180/3/4VLV, and the S183L mutant migrated slightly faster than the innate WT p23 but slightly slower than the processed WT p21 on SDS-PAGE (data not shown) (29, 69). Optiprep gradient analyses showed that the A180V and C184V mutants exhibited membrane distribution profiles similar to that of the processed WT core, whereas the S183L and triple mutants exhibited increased distribution to the soluble protein fraction (Fig. (Fig.5A).5A). In the membrane protection assay, the A180V and C184V mutants, just like the processed WT core, displayed a higher degree of resistance to protease treatment, while the S183L and triple mutants were more sensitive to proteolysis compared to the WT core (Fig. (Fig.5B5B).
To confirm that processing of the signal sequence by SPP is required for core envelopment, the 191 core was coexpressed with the WT or D219A mutant SPP, both of which contained the influenza virus HA tag and the ER retrieval signal, KEKK, at the C terminus (50). The WT and mutant SPP proteins were expressed as three forms migrating around a molecular mass of 45 kDa and a dimer migrating at ~90 kDa was also observed (Fig. (Fig.6A,6A, lanes 5 and 6). The top, middle, and bottom bands with an apparent molecular mass of 45 kDa corresponded to SPP containing two, one, and no glycans, respectively (50). The core was predominantly processed when expressed alone or coexpressed with the WT SPP (Fig. (Fig.6A,6A, lanes 1, 2, 4, and 5). The core was mainly present as the innate p23 form, and only a small amount of processed p21 was detected when coexpressed with the SPP mutant (Fig. (Fig.6A,6A, lanes 3 and 6), indicating that D219A mutant SPP confers dominant-negative interference with endogenous SPP-catalyzed SPP cleavage (50). In addition, a smaller amount of the processed p21 core was detected when the core was coexpressed with the WT SPP compared to that of the processed p21 core expressed alone (Fig. (Fig.6A,6A, compare lanes 2 and 5 to lanes 1 and 4, respectively). This result was consistent with a previous observation that SPP-catalyzed cleavage of the core protein may destabilize the viral capsid (69). In the membrane protection assay, the WT and mutant SPP proteins were sensitive to protease digestion (Fig. (Fig.6B,6B, top panel, compare lanes 5 and 8 to lanes 4 and 7, respectively); this is because the C terminus of SPP, to which the HA tag was attached, is located in the cytoplasm (72). Coexpression with the WT SPP did not greatly affect the protease resistance of the processed p21 core, whereas coexpression with the mutant SPP rendered the innate p23 core sensitive to proteolysis (Fig. (Fig.6B,6B, top and middle panels). Two bands under the core on SDS-PAGE were visible with WT or mutant SPP coexpression (Fig. (Fig.6B,6B, top panel, lanes 4, 5, 7, and 8). These two bands were not further processed products of the core but were nonspecific bands due to SPP coexpression. This notion was supported by their appearance in mutant SPP expression alone when a similar proteinase K digestion assay was performed (Fig. (Fig.6B,6B, bottom panel, lanes 4 and 5).
To further understand the molecular basis of core envelopment, the effects of deletions of residues 88 to 106 and residues 42 to 68 in the N-terminal half of the core (Fig. (Fig.1A,1A, panel 3) on core envelopment were assessed. These two core mutants still possessed an oligomerization ability, as shown by chemical cross-linking with glutaraldehyde (data not shown), and were able to assemble into high-ordered multimeric complexes, as shown by sucrose linear density gradient centrifugation (Fig. (Fig.7A).7A). Since multiple regions in domain I of the core may contribute to core oligomerization (for a review, see reference 65), it is not surprising that these two core N-terminal deletion mutants still formed high-ordered, multimeric complexes. These two mutants also exhibited patterns of association with intracellular membranes similar to those of the 191 core in the Optiprep subcellular fractionation study (Fig. (Fig.7B).7B). Coexpression with or without the catalytically inactive D219A mutant SPP showed that these two core mutants were efficiently processed by SPP in cells (Fig. (Fig.7C).7C). Nevertheless, a membrane protection assay showed that deletions of these two sequences greatly increased the core's sensitivity to proteinase K digestion (Fig. (Fig.7D),7D), indicating that these two mutants were not efficiently enveloped by membranes.
To understand whether core proteins expressed from the HCVcc system also undergo assembly and maturation processes similar to those observed for the 191 recombinant core expressed alone, core proteins expressed from JFH1 and Jc1 RNAs were analyzed in Huh7 cells. The core proteins expressed by the JFH1 and Jc1 clones were mainly located in fractions 4 to 7 (Fig. (Fig.8A,8A, panels 2 and 3, respectively), which corresponded to buoyant densities ranging from 1.12 to 1.16 g/ml. These observed densities were close to those of intracellular JFH1 infectious particles reported by other investigators (20) but lighter than that of the recombinant core complex (data not shown). As a control, both Aequorea coerulescens GFP and Discosoma DsRed fluorescence protein, respectively, which are known not to form aggregated proteins, peaked in fraction 11 (Fig. (Fig.8A,8A, panels 4 and 5). In the recombinant core expression system, the core may nonselectively enclose cellular RNAs, resulting in a complex with a buoyant density higher than that of authentic HCV particles, in which a single copy of the viral genome is encapsidated. This notion was further supported by the lower buoyant density of 1.13 g/ml of the core complex synthesized in vitro (Fig. (Fig.3C),3C), which might incorporate no or a only low level of RNA in the core complex.
In the Optiprep subcellular fractionation analysis, the core proteins expressed by the subgenomic system, as well as JFH1 and Jc1 infectious clones, were all predominantly fractionated into the ER/late-endosomal fractions (Fig. (Fig.8B).8B). The ability of the HCVcc core complex to float to the lighter ER/late-endosomal membranes indicates that these high-ordered, multimeric complexes are not core aggregates. Unlike the recombinant core which was detected at the plasma membrane, the JFH1 and Jc1 core proteins were only barely detected at the plasma membrane (Fig. (Fig.8B).8B). In the membrane protection assay, a significant fraction, ranging between 66 and 72%, of the core synthesized from JFH1 and Jc1 RNAs was still resistant to proteolysis in the absence of detergent, but they were highly sensitive to proteinase K digestion upon detergent treatment (Fig. (Fig.8C8C).
We then examined the intracellular localization of the core expressed by the HCVcc system using markers of intracellular organelles and the endocytic pathway. We did not observe colocalization of the core and calnexin, an ER marker; GM130, a marker of the Golgi apparatus; EEA1, a marker of early endosomes; or LAMP3, a marker of late endosomes (data not shown). Nevertheless, a fraction of the core located in the perinuclear region was associated with the surface of LDs (data not shown).
To further address the membrane envelopment property of the HCVcc core, the accessibility of the core's epitopes to a core-specific antibody was assessed in JFH1 RNA-transfected Huh7 cells permeabilized by saponin or Triton X-100, followed by immunostaining and flow cytometry. Permeabilization with saponin moderately increased the accessibility of the core's epitopes to a rabbit anti-core antibody; however, treatment with Triton X-100 greatly exposed the core's epitopes to the anti-core compared to saponin permeabilization (Fig. (Fig.9A9A).
We further verified that saponin permeabilized only the plasma membrane, which exposes epitopes on the core that are membrane-free or located at the LD surface, while Triton X-100 permeabilizes plasma membrane and intracellular membrane vesicles, which also exposes epitopes on the membrane-enveloped core as well. To do this, the accessibility of epitopes on HCV E2 and TIP47 to their respective antibodies was assessed. As expected, E2 was largely translocated into the ER lumen, as judged by its resistance to proteinase K digestion (Fig. (Fig.9B).9B). Permeabilization with Triton X-100 greatly increased the levels of anti-E2-positive cells compared to permeabilization with saponin (Fig. (Fig.9B).9B). TIP47 is known to move from the cytosol to coat nascent LDs during rapid fat storage and thus plays a critical role in packaging fat into LDs (for reviews, see references 19 and 74). TIP47 was completely digested by proteinase K even in the absence of detergent (Fig. (Fig.9C).9C). Treatment with saponin and Triton X-100 showed a similar level of anti-TIP47-positive cells (Fig. (Fig.9C).9C). These observations are consistent with the nonenveloped nature of TIP47.
Based on these results, we defined the degree of core envelopment using this differential permeabilization approach as the percentage of the differential level of the core that was specifically exposed by Triton X-100, but not by saponin, relative to the total core level exposed by Triton X-100 (also refer to the legend to Fig. Fig.9A9A for details). Thus, the percentage of core envelopment shown in Fig. Fig.9A9A was 62%. As controls, results from three independent studies showed that the extents of E2 envelopment as determined by proteinase K digestion and differential membrane permeabilization methods were 82.4% ± 5.1% and 80.3% ± 9.5%, respectively, while those of TIP47 were negligible by both assays.
To provide evidence for the involvement of membrane envelopment in HCV replication, (Z-LL)2-ketone was added to Huh7 cells transfected with JFH1 RNA. This SPP inhibitor reduced the core's resistance to proteolysis to 32.7% compared to 79.8% of the core without inhibitor treatment (Fig. 10A, top and middle panels). When the culture supernatants were titrated to determine viral infectivity, the SPP inhibitor decreased the titer of virus released into the culture medium to 49.7% of that without inhibitor treatment (Fig. 10A, bottom panel).
To confirm that treatment with (Z-LL)2-ketone decreases core's envelopment, the accessibility of epitopes to the anti-core for the HCVcc core expressed in the presence or absence of the SPP inhibitor was examined by a differential membrane permeabilization assay. In cells treated with (Z-LL)2-ketone, permeabilization with saponin showed a higher level of anti-core-positive cells compared to cells without the inhibitor; however, permeabilization with Triton X-100 showed similar levels of anti-core-positive cells regardless of the presence or absence of the SPP inhibitor (Fig. 10B). These observations indicated that treatment with (Z-LL)2-ketone increases the percentage of unenveloped core in total core population. The results from three independent studies showed that treatment with the SPP inhibitor reduced core envelopment to 38% as opposed to 63.6% of the core envelopment when core was synthesized without (Z-LL)2-ketone (the diagram shown in Fig. 10B).
Next, we confirmed the role of core envelopment in infectious virus production using the D219A mutant SPP. Coexpression of the D219A mutant SPP not only decreased the core's resistance to proteolysis to 27% compared to 65.7% of the core without mutant SPP coexpression (Fig. 10C, top and middle panels), it also suppressed the production of infectious virus to 33.8% of that of the virus produced in the absence of mutant SPP coexpression (Fig. 10C, bottom panel).
In these studies, the core proteins expressed in the presence or absence of (Z-LL)2-ketone or coexpressed with or without the D219A mutant SPP migrated indistinguishably on SDS-PAGE, suggesting that the SPP inhibitor and D219A mutant SPP may inhibit core processing at a level which is below the detection sensitivity of SDS-PAGE in this infectious virus system. Since HCVcc core proteins with various mutations designed to disrupt the SPP cleavage site in the signal sequence were still efficiently processed and even (Z-LL)2-ketone only partially inhibited the processing of these mutants, Targett-Adams et al. (67) therefore speculated that the HCV core-E1 signal peptide is an extremely efficient substrate for SPP. Okamoto et al. (49) recently showed that treatment with L685,458, another SPP inhibitor, resulted in production of the unprocessed core and the processed form in a Huh7-derived Huh7OK1 cell clone. It is not clear whether different Huh7 clones can cause differences in the efficiency of blocking SPP activity. Because presenilins, which belong to the same aspartic protease family as SPP (72), are able to cleave the β-amyloid precursor protein at multiple sites (51, 58), it is also likely that SPP can cleave the core at multiple sites in the HCVcc system, which thus complicates discrimination of the unprocessed core from the processed one. Alternatively, the possibility that the SDS-PAGE system used may not truly differentiate the unprocessed and processed core proteins expressed by the HCVcc system cannot be completely ruled out despite its ability to differentiate the two forms of the recombinant core.
Thus far, mechanistic insights into the consecutive steps of core multimerization and its interaction with membranes during the early phase of HCV morphogenesis are still poorly understood. Several groups have reported a complex intracellular localization of core protein; it predominantly resides in the cytoplasm, either in association with the ER, LDs, or mitochondria or in the nucleus (for reviews, see references 4 and 66). When signal sequence cleavage is inhibited by mutations in the signal sequence or at the signal peptide cleavage site between the core and E1, the innate p23 is retained in ER membranes (29, 41). Nevertheless, we showed that the presence and appropriate processing of the signal anchor by SPP are critical for retention of the recombinant core on ER/late-endosomal membranes (Fig. (Fig.5A5A and data not shown). In line with this notion, it was reported that substitutions of Leu-139, Val-140, and Leu-144 in the core with Ala residues inhibit SPP processing and prevent the core from attaching to ER membranes (50).
In the present study, we provide several lines of evidence to support the notion that the core is assembled into oligomeric complexes before or upon binding to ER/late endosome membranes. This includes (i) cofractionation of the innate 191 core with the processed core to the ER/late endosomes and small vesicles in the presence of (Z-LL)2-ketone (Fig. (Fig.4A);4A); (ii) inhibition of the core's association with membranes, but not of multimeric complex formation, by deletion of the signal sequence or inhibition of core processing by an SPP-specific inhibitor (Fig. (Fig.44 and data not shown); (iii) the ability of the isolated multimeric 191 and 173 core complexes to bind to membranes in vitro (Fig. (Fig.3A);3A); and (iv) formation of a high-ordered, multimeric complex of core translated in vitro (Fig. (Fig.3C).3C). On the other hand, proteinase K digestion of the recombinant core synthesized in the presence of (Z-LL)2-ketone exhibited both protease-sensitive and -resistant complexes, both of which contained innate and processed core proteins (Fig. (Fig.4B),4B), implying that envelopment of the core occurs after intramembrane cleavage. If intramembrane processing occurred after membrane envelopment, it would be expected that all of the innate p23 core molecules would be sensitive, whereas all of the processed p21 core molecules would be resistant to protease treatment.
It is thought that the HCV particles assemble and bud into the ER lumen and are transported via exocytosis through the Golgi apparatus to the plasma membrane, where the membrane of vesicles is fused with the plasma membrane to release virions into the extracellular milieu. Nevertheless, this proposition has not been verified in the HCVcc system. The recombinant core may ultimately be routed and attach to the plasma membrane without further fission or secretion, resulting in a greater tendency of a recombinant core to be located at plasma membranes than would an HCVcc core (Fig. (Fig.8B).8B). Similar to the HCVcc core which targets the surface of LDs (data not shown), where the core recruits viral NS proteins and replication complexes (40, 43), a significant fraction of the recombinant 191 core and the majority of cytoplasmic 173 core were also located on LDs (data not shown). These results indicate a critical role of LDs in the maturation process of the recombinant core as well. Rouillé et al. previously noted that the core expressed in the context of the infectious cycle is not localized in well-defined intracellular compartments such as the ER or the Golgi complex or the intermediate compartment, nor is it associated with markers of the endocytic pathway such as EEA1 or LAMP1 (56). We did not observe colocalization of the HCVcc core with markers of ER, the Golgi complex, or early or late endosomes by confocal microscopy (data not shown) either. These findings are in contrast with the general belief that the HCV core interacts with the ER and that the core protein of the genotype 1b BK strain expressed by stably transfected cells is localized in the ER, intermediate compartment, and cis-Golgi complex region (38). Rouillé et al. proposed that this disparity can be attributed to the rapid transfer of the core from the ER to LDs in infected cells or to the differential levels of LDs in various cell types used, since the latter factor may also affect different extents to which the core is associated with the ER or LDs (56).
Importantly, we provide compelling evidence to show that the recombinant and HCVcc core proteins are indeed enveloped by a membrane. Besides the conventional membrane protection assay that is generally used to assess enclosure of a protein by membranes, we performed coimmunoprecipitation with specific antibodies to detect envelopment of the recombinant core (Fig. (Fig.2C).2C). We also established a differential membrane permeabilization assay to differentiate the HCVcc core population that is membrane-enveloped from those that remain membrane-free or are localized at the surface of LDs (Fig. (Fig.9A).9A). The fidelity of this method was further validated by examining the HCV E2 protein (Fig. (Fig.9B)9B) and an LD-associated protein, TIP47 (Fig. (Fig.9C).9C). Moreover, we showed that treatment with the SPP inhibitor decreased membrane envelopment of the HCVcc core using both proteinase K digestion and differential membrane permeabilization assays (Fig. 10A and B). Remarkably, the extent of core envelopment as determined by the differential membrane permeabilization assay was similar to that assessed by the membrane protection assay regardless of the addition of SPP inhibitor or not (compare Fig. Fig.8C8C and 10A to Fig. Fig.9A9A and 10B, respectively). All of these results confirmed that a significant portion of the membrane-enveloped core, which is not localized at the LD surface or in the vicinity of LDs (data not shown), is present inside the cytoplasmic membranous compartments or vesicles.
Just like the recombinant core, the core expressed by Huh7 cells from infectious JFH1 and Jc1 RNAs formed a high-ordered, multimeric complex, was mainly associated with ER/late-endosomal membranes, and was enveloped by membranes (Fig. (Fig.8).8). Both the SPP inhibitor and the D219A mutant SPP interfered with the production of infectious HCV production as well as core envelopment (Fig. 10A and C), linking a function of core envelopment to HCV maturation and morphogenesis. This finding is in line with our subgenomic core analyses which showed that processing of the core by SPP is required for core envelopment and with recent results from other researchers that SPP-mediated core processing is required for HCV propagation (49, 67). Our results also underline the more-profound effect of the SPP inhibitor and mutant SPP on inhibition of core envelopment and infectious virus production than its seemingly unnoticed effect on retarding the mobility of the core on SDS-PAGE. These observations collectively imply that during the early phase of core assembly and budding, the recombinant core expressed alone undergoes steps similar to those utilized by the HCVcc core.
The envelopment of HCV nucleocapsids is thought to be achieved by interactions at the ER membrane between the first hydrophobic domain of E1 and the C-terminal hydrophobic domain of the core protein (33, 36). Nevertheless, our subgenomic core and infectious RNA clone studies showed that the HCV core protein, which constitutes the budding apparatus of the virus, can direct the core's assembly into a high-ordered multimeric complex and envelopment by membranes without the help of other viral proteins. It is likely that once it is folded and assembled, the core multimeric complex may possess the intrinsic capacity to assemble laterally into an isometric lattice on ER membranes. The core complex may be sufficient to introduce the membrane curvature needed to form a bud and may trigger the fission reaction required for pinching off membrane vesicles. Deletion of the E1 and E2 open reading frame was found to result in no production of infectious viral particles into the culture medium (70), indicating that E1 and E2 are required for the release of infectious viral particles into the medium. By taking our results and this observation into consideration, we hypothesize that the core itself is sufficient for assembly into VLPs and budding, likely into the lumen of ER. Nevertheless, these VLPs are incapable of being released into the culture fluid without acquiring E1 and E2 proteins on their viral envelope.
On the other hand, we showed that the association with membranes and processing of core signal sequences are required but not sufficient for envelopment of the recombinant core and that sequences in domain I also play crucial roles in core envelopment (Fig. (Fig.7).7). The core protein binds to a variety of cellular proteins and modulate signaling pathways, cell proliferation, cellular and viral gene expressions, cell transformation, apoptosis, and lipid metabolism (for reviews, see references 21, 30, and 68). In particular, the binding sites in the core of many cellular proteins have been mapped to domain I (for a review, see reference 40). It is likely that cellular proteins are involved in the nucleocapsid envelopment process. The differential protease-resistance of core complexes expressed in cells and in vitro (Fig. (Fig.2)2) supports this notion.
The results presented here using a subgenomic core and HCVcc expression systems provide a model to delineate the sequential and coordinated events of core complex formation and interactions with and envelopment by membranes during the core assembly process (Fig. (Fig.11).11). The core is first synthesized in the cytoplasm, and forms oligomeric structures prior to or upon attaching to the cytoplasmic side of ER membranes. On the ER membrane, the core is processed by SPP to yield the mature core. Once the signal sequence is removed, the processed core complex is subsequently enveloped into the lumen of the ER. In particular, core envelopment, a critical step in the infectious virus life cycle, has not previously been identified. An understanding of the mode of core assembly into multimeric complexes and interactions with and envelopment by membranes will not only provide better insights into nucleocapsid maturation and virus assembly and budding but may also reveal potential viral and/or cellular targets with antiviral implications.
We thank Yi-Miao Chen for technical assistance. We are also indebted to Charles M. Rice (Rockefeller University, New York, NY) for kindly providing pBRTM-HCV-H1-3011, the Jc1 construct, and the NS5A MAb (clone 9E10) and to Takaji Wakita (National Institute of Infectious Diseases, Tokyo, Japan) for generously providing pUC-JFH1. pEGFP-C3-derived GFP-Cellubrevin and GFP-TI-VAMP plasmids were generous gifts from Thierry Galli (Institut du Fer-à-Moulin, Paris, France). pcDNA3-SPP-HA and pcDNA3-SPP(D219A)-HA were kindly provided by Kohji Moriishi (Osaka University, Osaka, Japan).
This study was supported by a grant from the National Science Council (NSC95-2320-B-001-032-MY3), a Genomic and Proteomic Research grant (94M001-2), a Foresight Research grant (AS97FP-L15-1 and AS98FP-L15-1), and a Genomic Research grant (Summit Project III) from Academia Sinica, Taipei, Taiwan, R.O.C.
Published ahead of print on 15 July 2009.