Molecular Determinants for Subcellular Localization of Hepatitis C Virus Core Protein

Hepatitis C virus (HCV) core protein is a putative nucleocapsidprotein with a number of regulatory functions. In tissue culturecells, HCV core protein is mainly located at the endoplasmicreticulum as well as mitochondria and lipid droplets withinthe cytoplasm. However, it is also detected in the nucleus insome cells. To elucidate the mechanisms by which cellular traffickingof the protein is controlled, we performed subcellular fractionationexperiments and used confocal microscopy to examine the distributionof heterologously expressed fusion proteins involving variousdeletions and point mutations of the HCV core combined withgreen fluorescent proteins. We demonstrated that a region spanningamino acids 112 to 152 can mediate association of the core proteinnot only with the ER but also with the mitochondrial outer membrane.This region contains an 18-amino-acid motif which is predictedto form an amphipathic ${alpha}$ -helix structure. With regard to thenuclear targeting of the core protein, we identified a novelbipartite nuclear localization signal, which requires two outof three basic-residue clusters for efficient nuclear translocation,possibly by occupying binding sites on importin- ${alpha}$ . Differencesin the cellular trafficking of HCV core protein, achieved andmaintained by multiple targeting functions as mentioned above,may in part regulate the diverse range of biological roles ofthe core protein.

INTRODUCTION

Top

Introduction

Hepatitis C virus (HCV), the most important causative agentof posttransfusion and sporadic non-A, non-B hepatitis, is apositive-stranded RNA virus belonging to the family Flaviviridae(7). A precursor polyprotein of about 3,000 amino acids is encodedby a large open reading frame of the genome and undergoes cellularand viral protease-mediated posttranslational modification toproduce a series of structural and nonstructural proteins (8,13, 16).

HCV core protein, which is derived from the N terminus of theviral polyprotein, forms multimers and interacts physicallywith the viral RNA to constitute the nucleocapsid (28, 47, 50).Tissue transglutaminase is responsible for stabilizing the coreprotein by cross-linking it into a dimeric form (26). In addition,the core viral protein has properties which enable it to modulatea number of cellular processes, including transcription, inhibitionor stimulation of apoptosis, and suppression of host immunity,as reviewed previously (21, 29, 51, 52). Several studies suggestthat expression of the core protein affects mitochondrial functionand lipid metabolism. The core protein increases the cellularproduction of reactive oxygen species with subsequent increasesin lipid peroxidation (35, 39). The viral protein also colocalizeswith human apolipoprotein AII, associates with lipid droplets,and has the capacity to influence metabolic events involvinglipid storage (2, 17, 30, 36, 44). In addition, the core proteinreduces microsomal triglyceride transfer, leading to defectsin very low density lipoprotein assembly and secretion (40).Furthermore, the HCV core protein has transforming potentialin some cells under certain conditions (5, 42). Transgenic miceexpressing this protein in the liver develop hepatic steatosisdue to increased oxidative stress in the absence of inflammation,with subsequent development of hepatocellular carcinoma (34,36). These results suggest that the HCV core protein might playa pivotal role in the pathogenesis of hepatitis C in additionto its role as a structural component of the viral capsid.

The amino acid sequence of the core protein is well conservedamong different HCV isolates and genotypes compared to otherHCV proteins. The N-terminal domain of the HCV core proteinis highly basic, while its C terminus is hydrophobic. Althoughseveral core proteins of various sizes exist (17 to 23 kDa)(15, 23, 25, 49, 56), two processing events result in the predominantproduction of a 21-kDa core protein. Both of these events utilizethe endoplasmic reticulum (ER). The first one is to be cleavedfrom downstream envelope protein E1 at position 191, where theC-terminal hydrophobic domain serves as a putative signal peptidesequence. Subsequently, the signal sequence of 13 or 18 residuesis processed by signal peptide peptidase (19, 23, 56).

The HCV core protein is found primarily within the membranesof cytoplasmic organelles, but it is also found in the nucleus(23, 48, 56). Immunofluorescence studies show a punctate pattern,consistent with ER localization, as well as perinuclear localization(15, 24, 32, 46, 56). Some studies suggest direct effects ofthe core protein on mitochondrial function. In fact, the coreprotein localizes to the mitochondria (34, 39). The N-terminaldomain of the core protein contains three stretches of arginine-and lysine-rich sequences. Translocation of the core proteinto the nucleus, mediated by these basic-residue stretches whichfunction as nuclear localization signals (NLSs), is observed(6, 48). In addition, Moriishi et al. demonstrated that theN-terminal region of the core protein is also essential fornuclear retention through its interaction with the proteasomeactivator PA28 ${gamma}$ (33).

In this study, we found a region that is important for localizationof the mature core protein to the ER and to the mitochondrialouter membrane. We also identified a novel bipartite NLS responsiblefor nuclear targeting of the core protein, presumably via animportin-dependent pathway.

MATERIALS AND METHODS

Top

Materials and Methods

Plasmid construction. The construction of a plasmid expressing the full-length coreprotein of 191 amino acids, pCAGC191, was described previously(49). pGFP, a construct expressing green fluorescent protein(GFP) with a C-terminal Myc epitope tag sequences, was preparedas follows. pCMV/Myc/mito/GFP (Invitrogen Corp., Carlsbad, Calif.)was digested with PmlI, followed by treatment with the Klenowfragment of DNA polymerase I. The resultant linear fragmentwas ligated to a PstI linker (GCTGCAGC) and digested with PstIto remove the mitochondrial targeting signal sequence, followedby self-ligation. A series of HCV core-GFP fusion constructswere made by amplifying the core gene fragments with PCR withprimers containing Flag epitope tag sequences (sense) and aPstI site (both). After digestion with PstI, the segments wereinserted into the PstI site of pGFP. A series of GFP-core-E1fusion constructs were made by amplifying core and E1 gene fragmentswith PCR with primers containing a NotI site. After digestionwith NotI, the segments were inserted into the NotI site ofpGFP.

pGEX-4T-1 (Amersham Bioscience Corp., Piscataway, N.J.) wasused to express core protein fused with glutathione S-transferase(GST) in Escherichia coli. Core cDNA fragments encoding aminoacids 1 to 71 were inserted into the EcoRI site of pGEX-4T-1.Alanine substitutions were introduced into the core proteinby PCR mutagenesis with primers containing base alterations.The PCR products were then cloned into pCR2.1 (Invitrogen Corp.)and verified by DNA sequencing. Individual cDNAs were excisedand inserted separately into pGFP or pGEX-4T-1. The primer sequencesused in this study are available from the authors upon request.

Plasmid pRSET-hSRP1 ${alpha}$ (54), containing importin- ${alpha}$ cDNA under thecontrol of a T7 promoter, was kindly provided by Karsten Weis(University of California, Berkeley). A cDNA clone of importin- ${alpha}$ possessing 14 residues (MYPYDVPDYGGGGS), derived in part fromthe hemagglutinin (HA) tag at the N terminus, was constructedby PCR. The resultant linear fragment was inserted under thecontrol of a CAG promoter of pCAGGS and designated pCAG-HA-imp.

Cell culture and transfection. Human embryonic kidney 293T cells were maintained in Dulbecco'smodified Eagle's medium supplemented with 100 units of penicillinper ml, 100 µg of streptomycin per ml, and 10% fetal bovineserum at 37°C in a 5% CO₂ incubator. Monolayers of 293Tcells were transfected with plasmid DNA in the presence of Lipofectamine(Gibco-BRL, Life Technologies, Gaithersburg, Md.) accordingto the manufacturer's instructions.

Confocal immunofluorescence microscopy. Transfected cells were grown on glass coverslips. Two days aftertransfection, cells were fixed with 4% paraformaldehyde in phosphate-bufferedsaline (PBS) for 20 min at room temperature. Intracellular localizationof HCV core-GFP fusion proteins was visualized in cells transfectedwith a variety of GFP fusion constructs.

In order to detect the HCV core protein by immunofluorescence,fixed cells were permeabilized with 0.2% Triton X-100 in PBSfor 3 min at room temperature, followed by blocking with a nonfatmilk solution (Block Ace; Snow Brand Milk Products Co., Sapporo,Japan). The cells were then incubated with anticore monoclonalantibody B2 (Anogen, Mississauga, Canada) for 60 min at roomtemperature, followed by incubation with fluorescein isothiocyanate-conjugatedrabbit anti-mouse immunoglobulin G (IgG) (ICN Pharmaceuticals,Aurora, Ohio) for 45 min. To visualize mitochondria, MitoTrackerRed CM-H₂XRos (Molecular Probes, Eugene, Oreg.) was added tothe culture medium to a final concentration of 100 nM and incubatedfor 120 min at 37°C prior to fixation. To visualize theER, goat anticalregulin antibody (Santa Cruz Biotechnology,Santa Cruz, Calif.) and rhodamine-conjugated rabbit anti-goatIgG (ICN Pharmaceuticals) were used as the first and secondantibodies, respectively. To visualize HA-importin- ${alpha}$ , mouse anti-HAantibody (Roche Molecular Biochemicals, Indianapolis, Ind.)and rhodamine-conjugated goat anti-mouse IgG (ICN Pharmaceuticals)were used as the first and second antibodies, respectively.All specimens were examined with an LSM510 laser scanning confocalmicroscope (Carl Zeiss, Oberkochen, Germany).

Immunoelectron microscopy. Cells were transfected as described above. After 2 days, cellswere fixed with 3% paraformaldehyde and 0.1% glutaraldehydein 0.1 M PBS (pH 7.4). Free aldehyde groups were quenched with50 mM NH₄Cl in PBS. The cell pellets were embedded at progressivelylower temperatures (down to –35°C) in Lowicryl k4Maccording to an established protocol (43). Ultrathin sectionswere prepared and mounted on carbon-coated nickel grids. Toperform electron microscopy, Lowicryl k4M ultrathin sections,mounted on grids, were floated on a droplet of PBS containing1% bovine serum albumin, 0.1% Triton X-100, and 0.1% Tween 20for 10 min, after which they were exposed to droplets of mouseanticore monoclonal antibody (Anogen) diluted in PBS for 45min. Following this, they were rinsed twice for 5 min each inPBS and incubated with anti-mouse IgG-coated 10-nm immunogoldparticles (British Biocell, Cardiff, United Kingdom) for 45min. After rinsing with PBS and distilled water, the grids andembedded sections were air dried and exposed to uranyl and leadacetate contrast agents.

Subcellular fractionation. All steps were performed at 4°C in the presence of a proteaseinhibitor cocktail called Complete (Roche Molecular Biochemicals).To isolate the ER fraction, transfected cells were washed withPBS, lysed in homogenization buffer A (50 mM Tris-HCl [pH 8.0],1 mM ß-mercaptoethanol, 1 mM EDTA, and 0.32 M sucrose),and then centrifuged at 5,000 x g for 10 min. The supernatantwas then collected and centrifuged at 105,000 x g for 1 h. Thepellet was disrupted in lysis buffer (50 mM Tris-HCl [pH 7.5],150 mM NaCl, 1% NP-40, 1 mM dithiothreitol, 1 mM sodium orthovanadate,and 10 mM sodium fluoride), after which it was centrifuged at15,000 x g for 20 min. The resulting supernatant was used asthe ER fraction.

To isolate the mitochondrial fraction, transfected cells werewashed with PBS and homogenized in ice-cold homogenization bufferB (200 mM mannitol, 50 mM sucrose, 1 mM EDTA, and 10 mM Tris-HCl)at pH 7.4. The supernatant was then centrifuged at 1,000 x gfor 10 min to remove large debris and nuclei. The resultingsupernatant was then centrifuged at 20,000 x g for 20 min toobtain crude mitochondria. The crude mitochondria pellet wassubfractionated in Nycodenz gradients for further purificationof mitochondria. Nycodenz (Axis-Shield PoC AS, Oslo, Norway)solution at 50% (wt/vol) was prepared in buffer containing 5mM Tris-HCl and 1 mM EDTA at pH 7.4. This stock solution wasthen diluted with buffer containing 0.25 M sucrose, 5 mM Tris-HCl,and 1 mM EDTA at pH 7.4 before use. The crude mitochondrialpellets was suspended in 4 ml of 25% Nycodenz solution and overlaidonto the following discontinuous Nycodenz gradients: 1 ml of40%, 1 ml of 34%, and 2 ml of 30%. The samples were topped offwith 2 ml of 23% Nycodenz solution after placement onto thediscontinuous gradients. The tubes were then centrifuged at52,000 x g for 90 min. The dense band seen after centrifugationat the 25 to 30% interface was recovered as the purified mitochondrialfraction.

To determine the submitochondrial localization pattern of thecore protein, mitochondria were resuspended in SH buffer (0.6M sorbitol and 20 mM HEPES-KOH [pH 7.2]) in the absence or presenceof 30 µg of proteinase K per ml after purification byNycodenz density gradient centrifugation. Samples were incubatedfor 30 min at 0°C, after which protease digestion was haltedby the addition of p-aminophenyl methanesulfonyl fluoride hydrochloride(p-APMSF) (5 mM). Proteins lysed in sodium dodecyl sulfate (SDS)sample buffer were analyzed by SDS-polyacrylamide gel electrophoresis(PAGE) and immunoblotted as described below.

Immunoblot analysis. The proteins were transferred to a polyvinylidene difluoridemembrane (Immobilon; Millipore, Tokyo, Japan) after separationby SDS-PAGE. After blocking, the membranes were probed withmonoclonal- or polyclonal-antibody against core protein (Anogen),prohibitin (Neo Markers, Fremont, Calif.), ribophorin I (SantaCruz Biotechnology), translocase of the outer membrane (Tom)20 (Santa Cruz Biotechnology), translocase of the inner membrane(Tim) 17 (Santa Cruz Biotechnology), or GFP (Santa Cruz Biotechnology).Immunoblots were developed as previously described (15).

GST pulldown assay. Escherichia coli BL21 cells were transformed with GST-core fusionplasmids and grown at 37°C. Expression of the fusion proteinwas induced by 1 mM isopropyl-ß-D-thiogalactopyranosideat 37°C for 3 h. Bacteria were harvested, suspended in lysisbuffer (1% Triton X-100 in PBS), and sonicated on ice. GST andGST fusion proteins were purified from bacterial lysates withglutathione-Sepharose beads (Amersham Bioscience Corp.). Thebeads were washed four times with lysis buffer. Approximatelyequal amounts of purified protein, as estimated by Coomassiebrilliant blue staining, were used for the binding assays. Forpulldown assays, in vitro transcription and translation of importin- ${alpha}$ was done with pRSET-hSRP1 ${alpha}$ and the TNT-coupled reticulocyte lysatesystem (Promega Corp., Madison, Wis.) with T7 RNA polymerase.The reaction was carried out at 30°C for 4 h in the presenceof [³⁵S]methionine/cysteine (ICN Pharmaceuticals). The translationproduct was then incubated with glutathione-Sepharose beadsbound to GST fusion proteins in 1 ml of binding buffer (40 mMHEPES [pH 7.5], 100 mM KCl, 0.1% NP-40, and 20 mM 2-mercaptoethanol)at 4°C for 1 h. The beads were washed four times with bindingbuffer, and the pulldown complexes were separated by SDS-PAGEon 15% polyacrylamide gels. The gels were then fixed, dried,and analyzed with autoradiography.

RESULTS

Top

Results

Subcellular localization of HCV core protein. To assess the subcellular localization of HCV core protein,we first analyzed cells transfected with a full-length core-expressingconstruct by confocal microscopy. In accordance with previousobservations (2, 15, 32, 45, 56), a granular cytoplasmic stainingpattern of the core protein was observed in 293T (Fig. 1) andhuman hepatoblastoma HepG2 (data not shown) cells. Dual stainingof transfected cells with antibody against the ER protein calregulinalong with anticore antibody confirmed the ER localization ofthe core protein (Fig. 1a, b, and c show the core, calregulin,and a merged image, respectively). The pattern of subcellularlocalization of the core protein (Fig. 1d) was compared to thedistribution of mitochondria, as revealed by MitoTracker staining(Fig. 1e). Although distribution of the core protein was notcompletely identical with that of the mitochondrion-selectivedye, overlapping staining was observed, particularly in theperinuclear region (Fig. 1f).

View larger version (31K):
[in this window]
[in a new window]
FIG. 1. Confocal analysis of double staining for HCV core protein and ER or mitochondria. 293T cells transfected with full-length HCV core expression plasmid, pCAGC191 were allowed to express the plasmid for 2 days. Transfected cells were fixed directly (a to c) or fixed after loading with Mitotracker (d to f). After permeabilization with Triton X-100, cells were subjected to immunofluorescence staining with a mouse anticore antibody. A goat anticalregulin antibody was used for ER staining. The green signals corresponding to the core were found with a fluorescein isothiocyanate-conjugated rabbit anti-mouse IgG (a and d). The red signals corresponding to the ER were obtained with a rhodamine-conjugated rabbit anti-goat IgG secondary antibody (b). Mitochondria were stained with the mitochondrion-selective dye Mitotracker (e). Overlay resulted in yellow signals indicative of colocalization (c and f).

Intracellular localization of the core protein was further examinedin 293T cells by subcellular fractionation and Western blotting.The core protein was present in both the ER and mitochondrialfractions (Fig. 2), while it was not detected in the cytosolfraction (data not shown). The purity of the ER and mitochondrialfractions was confirmed with antibodies against ribophorin Ias an ER marker and prohibitin as a mitochondrial marker.

View larger version (29K):
[in this window]
[in a new window]
FIG. 2. Subcellular distribution of HCV core protein analyzed by immunoblotting. ER and mitochondrial (Mt) fractions were isolated from 293T cells expressing the full-length core protein (Core) or nontransfected cells (Mock) 2 days after transfection. Equal amounts of protein from each fraction as well as whole cell lysates (Total) were subjected to immunoblotting with a monoclonal antibody against either HCV core, prohibitin, or ribophorin I.

It is generally difficult to identify the nuclear distributionof proteins of interest due to contamination of the nuclearpreparation with unbroken, intact cells. Thus, to investigatewhether the core protein localizes to the nucleus, we examinedtransfected cells by immunoelectron microscopy. Although goldparticles were primarily observed within cytoplasmic membranes,perhaps highlighting the ER, immunoreactivity to anticore antibodywas also observed in the nucleus (Fig. 3A, arrows). In contrast,no antibody labeling was observed in cells transfected withan empty vector (Fig. 3B).

View larger version (164K):
[in this window]
[in a new window]
FIG. 3. Immunoelectron microscopy of HCV core protein. 293T cells expressing the full-length core protein (A) and nonexpressing cells (B) fixed 2 days after transfection. Immunoelectron microscopic analysis was performed with a mouse anticore antibody and a secondary anti-mouse IgG conjugated with gold particles. The arrows indicate the core protein localized in the nucleus (Nu). Bar, 500 nm.

Thus, HCV core protein predominates in the cytoplasm in a membrane-associatedform(s) with ER and mitochondria, but nuclear localization isalso observed.

Regions responsible for directing core protein to the ER and mitochondria. Given the tendency of the core protein to localize to the ERand to mitochondria, we next investigated whether specific sequencesmight be responsible for transporting the core protein to theseorganelles. Fusion proteins between different regions of thecore protein and GFP were developed, with specific emphasison the region downstream of amino acid 72 because this regioncontains clusters of hydrophobic amino acids and the N-terminal71 residues of the core are known to play a role in nucleartargeting (6, 48).

Western blotting of subcellular fractions with anti-GFP antibodyrevealed the localization of a core (72-173)-GFP fusion proteinto the ER and to mitochondria (Fig. 4). Fusion proteins containingGFP and core proteins with N- or C-terminal deletions (72-152-GFPand 112-152-GFP) were likewise identified within the ER andmitochondrial fractions. In contrast, the ER and mitochondrialfractions did not contain GFP fusion proteins containing coreprotein amino acids 124 to 152, 124 to 173, 152 to 173, or 72to 111. These fusion proteins demonstrated distribution profilessimilar to that of GFP alone. We also tested GFP-core-E1 fusions,which are processed at the C terminus of the core by signalpeptidase and signal peptide peptidase (19, 30). GFP-core fusionsexpressed from pGFP/72-382 and pGFP/112-382 were detected inthe ER and mitochondrial fractions. The fusion expressed frompGFP/152-382 was not identified in these fractions.

View larger version (33K):
[in this window]
[in a new window]
FIG. 4. Identification of segments that mediate association with ER and mitochondria in the core protein. Schematic diagram (left) and nomenclature (middle) of the core-GFP fusions are shown. Gray bars, expressed core and E1 regions. Subcellular distribution of fusion proteins is indicated on the right. ER and mitochondrial (Mt) fractions as well as whole-cell lysates (Total) were subjected to immunoblotting with an anti-GFP antibody.

We further analyzed subcellular localization of the fusion proteinsby confocal immunofluorescence microscopy (Fig. 5). As expected,fusions of (72-173)-GFP and (112-152)-GFP exhibited localizationto the ER and mitochondria. The patterns of subcellular localizationof these fusions are indistinguishable from that of the full-lengthcore protein, as shown in Fig. 1. Expression of (124-152)-GFPor (112-123)-GFP resulted in widespread diffusion of the fusionin the cell. Thus, these results indicate that the region spanningamino acids 112 to 152 can mediate association of the core proteinnot only to the ER but also to the mitochondria.

View larger version (59K):
[in this window]
[in a new window]
FIG. 5. Confocal analysis of double staining for core-GFP fusion protein and ER or mitochondria. 293T cells transfected with core-GFP expression plasmids (72-173, 112-152, 124-152, and 112-123) were allowed to express the plasmid for 2 days. Transfected cells were fixed directly (a to c, g to i, m to o, and s to u) or fixed after loading with Mitotracker (d to f, j to l, p to r, and v to x). After permeabilization with Triton X-100, a goat anticalregulin antibody was used for ER staining. The red signals corresponding to the ER were obtained with a rhodamine-conjugated rabbit anti-goat IgG secondary antibody (b, h, n, and t). Mitochondria were stained with the mitochondrion-selective dye Mitotracker (e, k, q, and w). Overlay resulted in yellow signals indicative of colocalization (c, f, i, l, o, r, u, and x).

We subsequently examined the submitochondrial localization ofthe core protein with a protease protection assay. As shownin Fig. 6A, HCV core protein localized in the mitochondria wascompletely digested upon treatment with proteinase K for 30min at 0°C. Under identical conditions, a marker specificfor the mitochondrial outer membrane, Tom20, was also observedto disappear, whereas digestion of a mitochondrial inner membranemarker, Tim17, was not observed. These findings confirm thatHCV core protein is localized to the mitochondrial outer membrane.

View larger version (23K):
[in this window]
[in a new window]
FIG. 6. (A) Protease protection assay. A mitochondrial fraction isolated from cells expressing the core protein was treated with proteinase K (+) as described in Materials and methods. The sample as well as the nontreated fraction (–) were subjected to immunoblotting with a monoclonal antibody against either HCV core, Tom20, or Tim17. (B) Protein sequence and predicted secondary structure of HCV core, amino acids 72 to 173. The secondary structure prediction was obtained with the self-optimized prediction method, a computer program on the internet (http://npsa-pbil.ibcp.fr/cgi-bin/npsa_automat.pl?page = /NPSA/npsa_sopm.html). H, ${alpha}$ -helix; t, turn; e, extension. (C) ${alpha}$ -Helical plot of amino acids 117 to 134 of the core protein. In the helical wheel plots, the gray shading represents apolar and hydrophobic residues; and the white represents polar residues.

The predicted secondary structure of the region, amino acids72 to 173, is shown in Fig. 6B. The presence of a long helicalsegment, lying between amino acids 116 and 134, and two short ${alpha}$ -helices (amino acids 146 to 152 and amino acids 155 to 159)were predicted. The results of the cell fractionation assayand confocal microscopy with a series of deletion mutants shownin Fig. 4 and 5 suggest that an ${alpha}$ -helix between amino acids 116and 134 may be required for associating the core protein withthe ER and the mitochondrial outer membrane. When amino acids117 to 134 are portrayed as a helical wheel, we found an amphipathicstructure with hydrophobic residues on one side and polar residueson the other side of the ${alpha}$ -helix (Fig. 6C), which is often observedin membrane-associated proteins. This helical conformation mightbe important for directing the core protein to the ER and mitochondrialouter membranes.

Nuclear localization of the HCV core protein is mediated by a bipartite NLS, possibly via an importin-dependent pathway. Although HCV core protein is mainly localized within the cytoplasm,it is also found in the nucleus, as shown in Fig. 3. The resultsof previous studies demonstrate that the N-terminal region ofthe core protein is responsible for nuclear targeting. It containsthree clusters of basic amino acid residues that represent putativeconsensus motifs for NLS sequences PKPQRKTKR (amino acids 5to 13), PRRGPR (amino acids 38 to 43), and PRGRRQPIPKARRP (aminoacids 58 to 71) (6, 48). Nuclear targeting is generally governedby a family of transporters or cytosolic receptor proteins,known as importins or karyopherins, which function in concertwith a guanine nucleotide-binding protein named Ran and otherregulatory proteins such as NTF2/p10. Conventional NLS-dependentnuclear targeting occurs when importin- ${alpha}$ recognizes the NLS sequence,mediating binding to importin-ß1, after which thetrimeric complex translocates to the nucleus (12).

In order to determine whether the putative NLS motifs identifiedwithin the core protein sequence are capable of binding to importin- ${alpha}$ ,we examined the in vitro interaction between bacterially expressedGST-fused core protein and ³⁵S-labeled importin- ${alpha}$ with a GSTpulldown assay. We then substituted lysine and arginine residuesof one or more of the putative NLS motifs of the core protein(all contained within the first 71 amino acids of the N terminus)with alanine and fused the resultant constructs with GST, asshown schematically in Fig. 7A. As shown in Fig. 7B (upper panel),importin- ${alpha}$ was pulled down by a GST fusion protein containingwild-type core (amino acids 1 to 71) protein but not with GSTalone, suggesting that direct binding occurs between the coreprotein and importin- ${alpha}$ . Importin- ${alpha}$ was also pulled down by GST-corefusion proteins containing substitutions in one or two NLS motifs(NLS/m1, NLS/m2, NLS/m3, NLS/m4, NLS/m5, and NLS/m6). However,importin- ${alpha}$ was not pulled down by GST-core fusion proteins containingalanine substitutions in all three NLS motifs (NLS/m7). It shouldbe noted that similar amounts of GST fusion proteins were usedfor each of the in vitro pulldown assays, followed by SDS-PAGEand Coomassie brilliant blue staining (Fig. 7B, lower panel).These results demonstrated that all three putative NLS motifsof the N-terminal region of the core protein can mediate bindingto importin- ${alpha}$ , which suggests that nuclear translocation of thecore protein occurs via an importin-dependent pathway (12).

View larger version (167K):
[in this window]
[in a new window]
FIG. 7. Mutational analyses of NLS motifs in HCV core protein. (A) Schematic structures of fusion proteins and amino acid sequences corresponding to putative NLS motifs, three basic clusters (underlined) in the core protein. Two series of constructs fused with either GFP or GST were created. The mutated basic residues are indicated with outline letters. (B) GST pulldown assay. Equal amounts of GST fusions as described in A or GST alone was immobilized on glutathione-Sepharose 4B beads and incubated with in vitro-translated, [³⁵S]methionine-labeled importin- ${alpha}$ . Bound material was separated by SDS-PAGE, and the amount of importin- ${alpha}$ bound was detected by autoradiography. Direct electrophoretic separation of in vitro translation products served as a control (input). Coomassie brilliant blue staining of GST fusions and GST alone are shown in the bottom panel. (C) Confocal analysis of double staining for core-GFP fusion protein and HA-importin- ${alpha}$ . 293T cells transfected with the wild-type core (1-71)-GFP (a to c) or NLS/m4 (d to f) expression plasmid and pCAG-HA-imp were allowed to express for 2 days. After the cells were fixed and permeabilized, they were incubated with a mouse anti-HA antibody. The red signals corresponding to HA-importin- ${alpha}$ were obtained with a rhodamine-conjugated goat anti-mouse IgG secondary antibody (b and e). Overlay resulted in yellow signals indicative of colocalization (c and f). (D) Subcellular localization of GFP fusion proteins. GFP fusions with and without substitution mutations in the NLS motifs of the core protein as described in A were expressed in 293T cells. GFP images of the fixed cells were recorded.

The interaction between the core and importin- ${alpha}$ was further analyzedby a colocalization assay (Fig. 7C). The GFP fusion containingthe wild-type core (amino acids 1 to 71) was well colocalizedwith HA-importin- ${alpha}$ ; distribution of the two proteins showed similarnuclear staining patterns, confirming the presence of a functionalNLS sequence(s) within the core protein. In contrast, NLS/m4,with substitutions in two NLS motifs, was partly colocalizedwith HA-importin- ${alpha}$ near or around the nuclear membrane, suggestingthat NLS motif double mutants bind to importin- ${alpha}$ but their bindingefficiency is lower than that of wild-type core protein.

Finally, we examined the subcellular localization of core proteinexpressed by the wild-type and NLS mutants (Fig. 7D). As expected,a fusion protein containing wild-type core protein (amino acids1 to 71) and GFP was localized exclusively to the nucleus. Coreproteins from three fusion proteins containing substitutionsin each NLS motif (NLS/m1, NLS/m2, and NLS/m3) were detectedprimarily in the nucleus. Weak fluorescence was also observedin the cytoplasm, suggesting that these mutations caused a slightreduction in the efficiency of nuclear translocation. On theother hand, two or three NLS motif substitution mutations (NLS/m4,NLS/m5, NLS/m6, and NLS/m7) completely abolished nuclear translocation,resulting in a diffuse distribution of core protein, similarto that of GFP alone. Although it is likely that all three putativeNLS motifs play a role, the above results suggest that at leasttwo of the three putative NLS motifs are prerequisite for efficientnuclear translocation of the core protein.

DISCUSSION

Top

Discussion

HCV core protein is released from the viral polyprotein by ahost protease(s) within the ER membrane at a signal peptidesequence lying between the core and envelope (E1) proteins (16,41). Subsequently, the signal peptide is further processed byan intramembranous protease called signal peptide peptidase(38, 53). This mature form of the core protein is then releasedand undergoes subcellular trafficking (30, 53). The core proteinlocalizes mainly to the ER, mitochondria, and lipid droplets.Some reports also describe localization of the core proteinto the nuclei of hepatocytes in HCV-infected patients (10),transgenic mice (34), and cultured cells expressing viral polyproteins(56). Although it has been reported which sequence motifs areresponsible for localization of the HCV core protein to lipiddroplets and nuclei, it is uncertain which sequences targetthe core protein to the ER and to mitochondria. In this study,we identified sequences related to localization of the maturecore protein to the ER and to mitochondria.

Through heterologous expression of core-GFP fusion proteinscontaining a series of deletions, we determined that a sequenceextending from amino acids 112 to 152 of the core protein isrequired for its localization at the mitochondrial outer membrane.Translocation of nucleus-encoded mitochondrial proteins is usuallydependent on N-terminal sequences, referred to as mitochondrialtargeting sequences (37). However, it is also true that a significantproportion of mitochondrial proteins lack these N-terminal mitochondrialtargeting sequences. Specifically, a number of outer membraneproteins do not have cleavable sequences at their N termini;rather, they are targeted to mitochondria by means of internalor C-terminal signals (31).

Since it has been reported that amino acid sequences requiredfor targeting to the outer mitochondrial membrane form a highlyhydrophobic ${alpha}$ -helical wheel, as seen in A-kinase associated protein84/12 (4) and NADH-cytochrome b reductase (14), a predictedstructure of an amphipathic ${alpha}$ -helix present between amino acids116 and 134 (Fig. 6B and C) possibly plays a role in directingthe core protein to the mitochondrial outer membrane. Sequencecomparisons demonstrate conservation of the amino acid sequenceand secondary structure of the region, amino acids 112 to 152,among a variety of HCV isolates, including the infectious H77cclone (55), as well as a full-length adaptive replicon (3).To gain insight into the significance of the secondary structureof the region in targeting to the mitochondria, further structuraland biochemical analyses are needed.

The association of HCV core protein with the mitochondrial membranesuggests that the core protein has the ability to modulate mitochondrialfunction, possibly by altering the permeability of the mitochondrialmembrane. The core protein induces the production of cellularreactive oxygen species in the livers of core-transgenic miceand in core-expressing cell lines (35). Reactive oxygen species,predominantly generated in mitochondria, induce genetic mutationsand act as secondary messengers to regulate a variety of cellularfunctions, including gene expression and proliferation (1).Although the molecular mechanism by which core protein inducesreactive oxygen species production is still unclear, HCV coreprotein is known to impair the mitochondrial electron transfersystem (35). The core protein may also modulate apoptosis, sincemitochondria play a major role in regulating programmed celldeath. Expression of HCV proteins, including the core protein,suppresses the release of cytochrome c from mitochondria tothe cytoplasm in HCV-transgenic mice, thus inhibiting Fas-mediatedapoptosis (27).

Okamoto et al. recently reported that not only the C-terminalsignal sequence but also amino acids 128 to 151 are requiredfor ER retention of the core protein by using a series of N-terminallytruncated core protein constructs (38). Here, in this study,we further showed that amino acids 112 to 152 mediate associationof the core protein with the ER in the absence of the C-terminalsignal sequence. Hope and McLauchlan demonstrated that the centraldomain of the core protein, amino acids 119 to 174, is importantfor association with lipid droplets (17). They also showed thatthis corresponding domain is shared with GB virus B, which ismost closely related to HCV, but not with either pestivirusesor flaviviruses (18). It appears that the 41 residues identifiedas the sequence mediating association with the ER membrane inthe present study are crucial for directing the core proteinto lipid droplets, since the surface of lipid droplets mustderive from the cytoplasmic side of the ER membrane.

The HCV core protein contains NLS sequences which are composedof three stretches of sequences rich in basic residues. Thesesequences were originally identified by experiments with fusedforms of wild-type and mutated core proteins with ß-galactosidase(6, 48). C-terminally truncated versions of the core proteinlocalize exclusively to the nucleus (48). A fraction of thecore protein is detected in the nucleus even when full-lengthHCV core gene is expressed (Fig. 2) and as described (34, 56).However, it is difficult to demonstrate clearly the nuclearlocalization of the core protein by immunofluorescence, presumablybecause of the instability of nuclearly localized core protein(49, 33). We only observed a nuclear staining pattern of thematured core protein after adding proteasome inhibitors to theculture (33).

Generally, NLS sequences fall into two categories; (i) monopartiteNLSs, which contain a single cluster of basic residues, and(ii) bipartite NLSs, which contain two clusters of basic residuesseparated by an unconserved linker sequence of variable length(reviewed in reference 12). Nuclear translocation of an NLS-containingcargo protein is initiated when the soluble import receptor(importin) recognizes the NLS-containing protein within thecytoplasm. Importin- ${alpha}$ contains an NLS-binding site(s), and importin-ßdocks importin-cargo complexes to the cytoplasmic filamentsof a nuclear pore complex, after which translocation occursthrough the nuclear pore. Thus, importin- ${alpha}$ functions as an adaptorbetween the bona fide import receptor and the NLS-containingprotein.

We further characterized the NLS of the core protein and foundthat each of the NLS motifs of the core protein is able to bindto importin- ${alpha}$ and that at least two NLS motifs are required forefficient nuclear distribution of the core protein in cells.It appears that double mutations among three NLS motifs decreasethe ability of the core protein to bind importin- ${alpha}$ . These observationssuggest that the binding of the double mutants with importin- ${alpha}$ leads to no or little active translocation of the core proteininto the nucleus. The double mutations may also block subsequentinteractions with importin-ß1, GTPase Ran, and/orNTF2/p10, which are required for translocation through the nuclearpore complexes.

The findings obtained in this study suggest that HCV core proteinNLS motifs have a bipartite function. Crystallographic studiesof monopartite (e.g., simian virus 40 large T antigen) and bipartite(e.g., nucleoplasmin) NLSs show that the basic residue clustersof bipartite NLSs occupy separate binding sites on importin- ${alpha}$ .In contrast, while monopartite NLSs can bind to the same sitesas bipartate NLSs on importin- ${alpha}$ , they mainly bind to the N-terminalbinding site, which is referred to as the major binding siteon importin- ${alpha}$ (9, 11). A recent report describes an importin- ${alpha}$ variant with a mutation in the major site which results in decreasedability to bind both monopartite and bipartite NLSs. Anothervariant with a mutation in the minor site exhibits decreasedbinding only to bipartite NLS-containing proteins, making importin- ${alpha}$ nonfunctional in vivo (22). Thus, we favor a model in whichthe core protein bipartite NLS, composed of any two of the threebasic clusters, occupies both major and minor binding siteson importin- ${alpha}$ , resulting in efficient nuclear translocation.Importin- ${alpha}$ may be equally accessible to all clusters, given theirclose proximity to one another, as well as the distinct conformationalflexibility of the ${approx}$ 70-residue N-terminal region of the coreprotein.

With regard to the molecular mechanisms participating in nuclearlocalization of the core protein, Moriishi et al. found thatPA28 ${gamma}$ is involved in nuclear localization of the core protein.Interaction of the core protein with PA28 ${gamma}$ plays an importantrole in retention of the core protein in the nucleus (33). Furthermore,in yeast cells, nuclear transport of the core protein requiresthe activity of the small GTPase Ran/Gsp1p and is mediated byKap123p, but neither importin- ${alpha}$ nor importin-ß is involved(20). Differences in nucleocytoplasmic transport between yeastand mammalian cells might explain the inconsistencies observedin the present study. Further experiments are required to characterizethe exact nature of the interaction between the core proteinand components of the nuclear import machinery, particularlyin cells where HCV is replicating.

In conclusion, the mature HCV core protein has an internal 41-amino-acidsequence mediating association of the viral protein with theER and mitochondria. We also provide evidence for a novel classof bipartite NLS contained within the core protein, which comprisestwo of three basic motifs, thus enabling efficient nuclear targeting.Multiple functional domains influence the subcellular localizationof the core protein, which ultimately depends on the balanceof the respective signals.

ACKNOWLEDGMENTS

We thank colleagues in the laboratories of the Department ofVirology II at the National Institute of Infectious Diseasesof Japan for providing advice and help. We especially thankMami Matsuda and Makiko Yahata for assistance in sequencingand the preparation of experimental reagents and Tomoko Mizoguchifor secretarial work. We are grateful to Karsten Weis for providingus with the plasmid containing importin- ${alpha}$ cDNA.

This work was supported in part by Second Term Comprehensive10-Year Strategy for Cancer Control and Research on Emergingand Reemerging Infectious Diseases, Health Sciences ResearchGrants of the Ministry of Health, Labor and Welfare, and bythe Program for Promotion of Fundamental Studies in Health Sciencesof the Organization for Drug ADR Relief, R&D Promotion andProduct Review of Japan (ID:01-3). This work was also supportedin part by a Grant-in-Aid for Young Scientists from the Ministryof Education, Culture, Sports, Science and Technology to R.S.(15790244).

FOOTNOTES

* Corresponding author. Mailing address: Department of Virology II, National Institute of Infectious Diseases, 1-23-1 Toyama, Shinjuku-ku, Tokyo, Japan 162-8640. Phone: (81) 3-5285-1111. Fax: (81) 3-5285-1161. E-mail: tesuzuki{at}nih.go.jp.

REFERENCES

Top