Role for ADP Ribosylation Factor 1 in the Regulation of Hepatitis C Virus Replication

Abstract

We hypothesized that ADP-ribosylation factor 1 (Arf1) plays an important role in the biogenesis and maintenance of infectious hepatitis C virus (HCV). Huh7.5 cells, in which HCV replicates and produces infectious viral particles, were exposed to brefeldin A or golgicide A, pharmacological inhibitors of Arf1 activation. Treatment with these agents caused a reduction in viral RNA levels, the accumulation of infectious particles within the cells, and a reduction in the levels of these particles in the extracellular medium. Fluorescence analyses showed that the viral nonstructural (NS) proteins NS5A and NS3, but not the viral structural protein core, shifted their localization from speckle-like structures in untreated cells to the rims of lipid droplets (LDs) in treated cells. Using pulldown assays, we showed that ectopic overexpression of NS5A in Huh7 cells reduces the levels of GTP-Arf1. Downregulation of Arf1 expression by small interfering RNA (siRNA) decreased both the levels of HCV RNA and the production of infectious viral particles and altered the localization of NS5A to the peripheries of LDs. Together, our data provide novel insights into the role of Arf1 in the regulation of viral RNA replication and the production of infectious HCV.

Hepatitis C virus (HCV) is an important human pathogen that causes chronic hepatitis, which can progress to cirrhosis and liver cancer (78). In many patients, it is difficult to eliminate chronic HCV infection. Because persistent infection contributes to the chronic phase of the disease, it is extremely important to understand the molecular and cellular events underlying the establishment and maintenance of HCV replication.

HCV contains a plus-strand RNA genome that encodes the structural proteins core, E1, E2, and the p7 protein, and the nonstructural (NS) proteins 2, 3, 4A, 4B, 5A, and 5B. The structural proteins are components of the mature viral particle, whereas the NS proteins, which function mainly in RNA replication and viral polyprotein processing, are not known to be packaged in the virion. Recent models propose that HCV infection commences by initial binding of the virus to the low-density lipoprotein (LDL) receptor and scavenger receptor class B type I (24). Subsequently, HCV particles interact with the tetraspanin CD81 and the tight-junction proteins claudin-1 and occludin to facilitate the internalization of the virus into the host cell cytoplasm via clathrin-coated pits (7, 9, 19, 23, 31, 53).

In the past few years, robust cellular model systems that support HCV infection, replication, and viral particle secretion have been developed (27). Using these systems, studies have shown that HCV, like other positive-strand RNA viruses, hijacks intracellular membranes, probably of diverse origins, to generate unique membranous platforms where HCV genome replication and viral particle assembly occur (20). Electron microscopic observations have revealed that parts of the endoplasmic reticulum (ER) in these cells are deformed, forming uniquely shaped membrane structures termed “membranous webs.” These altered membrane structures can be induced by sole expression of the viral protein NS4B (22, 44), and de novo viral RNA synthesis appears to occur in their vicinity (28). Hence, membranous webs have been proposed to contain NS proteins comprising replication complexes (RCs) that promote viral RNA replication.

Lipid droplets (LDs) are dynamic organelles that store neutral lipids. They are thought to originate from the ER and to move through the cytoplasm, likely via interactions with microtubules (63), while interacting with various membranous organelles. These interactions probably serve to facilitate the transport of neutral lipids (45, 71). Several independent observations have recently suggested the involvement of LDs in HCV RNA synthesis and the production of infectious viral particles. For example, association of the HCV structural protein core with LDs has been shown to alter the mobility of LDs, and consequently their intracellular distribution, in a microtubule-dependent manner. The LDs' association with intact microtubules has been proposed to be important for the production of virus progeny (14). Core, localized on LDs, has been shown to interact with NS5A (39). These interactions could facilitate the recruitment of NS proteins and RCs residing in ER-modified membranes to core-associated LDs, an activity proposed to be critical for the production of infectious viruses (42, 61). The potential bridging between LDs and modified ER membranes harboring RCs is further supported by ultrastructural data showing that multilayered and convoluted ER membrane structures surround LDs in cells where HCV is replicating (42, 56) and by data showing close core-dependent apposition of HCV RNA in RCs and LDs (69, 70).

Why does HCV have such a strong affinity with LDs and associated membranes? One possibility may be supported by current hypotheses suggesting that HCV production and release are coordinated with the biosynthesis of very low density lipoprotein (VLDL). HCV virions isolated from patients appear to interact with various lipoproteins, including VLDL (3, 50). Although VLDL assembly is a poorly characterized process, it is postulated that lipid mobilization from cytosolic LDs to the nascent LDs in the ER lumen contributes to VLDL assembly. Targeting of HCV proteins to LDs and associated ER membranes may therefore be needed to facilitate viral entry into the ER lumen and exploiting the VLDL assembly pathway (for a recent review, see reference 41). Although key details of HCV and VLDL coassembly are not known, it is reasonable to suggest that LDs play a central role in multiple stages of HCV assembly and trafficking into ER lumen sites.

ADP-ribosylation factor 1 (Arf1) is a small GTPase that activates phospholipase D1 (34). Arf1 and this phospholipase have been shown to reside in LDs (30, 48). The fungal metabolite brefeldin A (BFA) inhibits Arf1 by targeting specific guanine nucleotide exchange factors (GEFs), including Golgi complex-specific BFA resistance factor 1 (GBF1). BFA forms a stable GDP-Arf1-GEF complex on membranes, thus preventing GDP/GTP exchange of Arf1 on membrane surfaces (2). Arf1, probably in its GDP-bound form, and GBF1 have been localized to LDs (30, 47, 48, 80). Moreover, proteins that interact with Arf1 GEFs and that have been implicated in HCV RNA replication, such as Rab1 (64, 65), have been identified by proteomic analyses as LD-associated proteins in core-expressing cells (59) and other cells (80). Arf1 has been suggested to be involved in the biogenesis of VLDL (4, 5). Given that LDs and VLDL function in the HCV life cycle, these observations prompted our hypothesis that LD-associated Arf1 plays an important role in HCV RNA synthesis and viral particle biogenesis.

Several lines of additional evidence provide indirect information on the role of Arf1 in the HCV life cycle. For example, HCV replication has been shown to depend on the Arf1 effector COPI coatomer complex and phosphatidylinositol-4-kinase (68). HCV replication also depends on the Arf1 GEF GBF1 (29). Treatment of cells with golgicide A (GCA), a drug that specifically inhibits GBF1 (57), or with BFA, has been shown to decrease HCV RNA replication (29, 68). However, these studies examined the drugs' effects only on early stages of viral replication. In this study, we tested the sensitivity of later stages in the viral life cycle to Arf1. Our results emphasize the importance of Arf1 in the regulation of HCV RNA replication and in the biogenesis of infectious HCV.

MATERIALS AND METHODS

Antibodies.

Monoclonal antibodies against NS5A and NS3 were obtained from Virostat (Portland, ME). Anti-core monoclonal antibodies (clone 6G7) were a generous gift from Harry Greenberg (Stanford University School of Medicine, Stanford, CA). Anti-β-actin was obtained from Sigma Immunochemicals (Rehovot, Israel). Arf1-specific monoclonal antibodies (clone 3F1) were from Novus Biologicals (Littleton, CO). The anti-α-COP polyclonal antibody was kindly provided by Michal Goldberg (Hebrew University of Jerusalem, Jerusalem, Israel). Fluorescently (Alexa Fluor 488 and Alexa Fluor 594) labeled secondary antibodies were from Molecular Probes (Eugene, OR). Cy5-conjugated anti-mouse and horseradish peroxidase (HRP)-conjugated secondary antibodies were from Jackson ImmunoResearch Labs (West Grove, PA).

Plasmids.

Standard recombinant DNA technology was used to genetically manipulate and purify all constructs. The sequences of PCR-amplified DNA segments were confirmed by DNA sequencing using BigDye Terminator cycle sequencing chemistry from Applied Biosystems (Foster City, CA). NS5A and core, subcloned into the pEF6 mammalian expression vector, were constructed by amplifying the full-length coding regions of these proteins from the Bart79I HCV subgenomic replicon and ligating them into the pEF6 myc-His vector (Invitrogen, Carlsbad, CA). A stop codon was inserted at the 3′ end of NS5A, upstream of the sequences encoding the Myc and His tags. Plasmids expressing wild-type (WT) Arf1 tagged with green fluorescent protein (GFP) (Arf1 WT-GFP), constitutively active Arf1 Q71L-GFP, and dominant negative Arf1 T31N-GFP were prepared as follows. An Arf1 clone, obtained from the Human ORFeome Collection (Open Biosystems, Huntsville, AL), was inserted into pDEST-47 with a GFP C-terminal tag, using the Gateway system (Invitrogen). The T31N and Q71L mutations were constructed using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). Plasmid pFL-J6/JFH1, encoding the full-length chimeric HCV J6/JFH of genotype 2a, was used to generate infectious HCV in cell culture (HCVcc), as described previously (36) and below. The plasmid was kindly provided by Charles M. Rice (The Rockefeller University, NY).

Cell culture.

The human hepatoma cell line Huh7 was cultured routinely in RPMI 1640-Dulbecco's modified Eagle's medium (DMEM; Invitrogen) (1:1, vol/vol) supplemented with 10% (vol/vol) fetal bovine serum (FBS), penicillin (100 U/ml), streptomycin (100 μg/ml), and 2 mM l-glutamine (Biological Industries, Kibutz Beit Haemek, Israel). Huh7.5 cells were maintained in DMEM containing 10% (vol/vol) heat-inactivated fetal calf serum (FCS; Invitrogen) and 0.1 mM nonessential amino acids (Invitrogen). FLRP1 Huh7 cells expressing a full-length replicon of genotype 1b (12) were cultured in RPMI-DMEM-10% FBS supplemented with 750 μg/ml G418 (Gibco-BRL, Carlsbad, CA). Cells were grown routinely at 37°C under a 5% CO₂ atmosphere and were passaged every 3 days at a 1:3 split ratio. Cells were kept in culture up to about 30 passages.

In vitro transcription of recombinant JFH1 genomic RNA.

The plasmid encoding the full-length chimeric HCV J6/JFH1 was linearized by XbaI digestion, followed by treatment with mung bean nuclease to remove 5′-end overhangs. The linearized DNA template, purified by phenol-chloroform extraction and ethanol precipitation, was resuspended at a final concentration of 1 μg/μl. The linearized DNA template was transcribed with T7 RNA polymerase by using a MEGAscript T7 kit (Ambion, Austin, TX) according to the manufacturer's instructions. After transcription, synthesized RNA was treated with DNase I. The integrity of the RNA was analyzed by nondenaturing agarose gel electrophoresis, and the yield was determined by spectrophotometry. The RNA concentration was adjusted to 2 μg/μl, and the RNA was stored at −70°C until electroporation.

Electroporation of cells with HCV RNA.

Huh7.5 cells were grown to 60 to 80% confluence, trypsinized, and washed twice in cold RNase-free phosphate-buffered saline (PBS) (BioWhittaker Inc., Walkersville, MD). Cells were resuspended in cold PBS at a concentration of 1.5 × 10⁷/ml; 0.4 ml of the cell suspension was mixed with 10 μg in vitro-transcribed J6/JFH1 RNA. The mixture was dispensed into a 2-mm-diameter gap cuvette (BTX, San Diego, CA), and electroporation was performed using a BTX model 830 electroporator (820 V; 5 99-μs pulses given at 220-ms intervals). Cells were left to recover for 15 min at 22°C and were then mixed with 10 ml of prewarmed (37°C) growth medium. Cells were then seeded in 10-cm-diameter dishes; 48 h postelectroporation, cells were trypsinized, seeded in 12-well plates (2 × 10⁵ cells/well), and incubated for an additional 24 h before the various treatments.

BFA and GCA treatments.

Cells were plated at 2 × 10⁵/well in a 12-well plate. The growth medium was replaced with the same medium supplemented with either 5 μg/ml of BFA (Epicentre Technologies, Madison, WI) (stored as a 10 mg/ml stock in dimethyl sulfoxide [DMSO] at −20°C); 10 μM GCA, a GBF1-specific inhibitor (stored as a 10 mM stock in DMSO at −20°C) (57); or the equivalent volume of DMSO alone as a control. Cells were incubated with BFA for 12 h or with GCA for 20 h. The cell health indicator Alamar Blue (49) was used to determine cell viability. The viability of treated cells was similar to that of controls (see Fig. S1 at http://www.bio.huji.ac.il/eng/staff_in.asp?staff_id=77&chapter_id=198), suggesting that the treatments were not toxic to the cells. Moreover, fluorescence data showed that the Golgi marker α-COP was dispersed from perinuclear regions into the cytoplasm upon BFA or GCA treatment and was redistributed back to the original perinuclear sites following drug washout (see Fig. S2 at http://www.bio.huji.ac.il/eng/staff_in.asp?staff_id=77&chapter_id=198). These data further suggest that the cells remained alive following the prolonged BFA or GCA treatments. Bartz et al. (8) reported that LDs are lost upon a 12-h exposure of different kinds of nonhepatic cells to BFA. Here, however, with similar BFA or GCA treatments, LDs were detected in all cells, and in many cells they were even visualized as inflamed globular bodies (see Fig. S3 at http://www.bio.huji.ac.il/eng/staff_in.asp?staff_id=77&chapter_id=198). We have no explanation for this discrepancy, but we assume that it can be attributed to the different cell types used.

Gene silencing by siRNA.

A pool of four small interfering RNA (siRNA) duplexes targeting different sites within the coding region of human Arf1 (ON-TARGETplus SMART pool [62, 66, 75]) and nontargeting siRNAs (siGlo or siControl) for use as controls were purchased from Dharmacon (Lafayette, CO).

Arf1 siRNA sequences were as follows: 5′-UGACAGAGAGCGUGUGAAC-3′, 5′-CGGCCGAGAUCACAGACAA-3′, 5′-ACGAUCCUCUACAAGCUUA-3′, and 5′-GAACCAGAAGUGAACGCGA-3′. The nontargeting siRNA control sequence was 5′-UGGUUUACAUGUCGACUAA-3′. Cells (2 × 10⁵ per well) cultured in 12-well plates were transfected with 50 to 100 nM siRNA by using the Lipofectamine 2000 reagent (Invitrogen). Typically, Arf1 siRNA was transfected in triplicate for 72 h. The number and viability of the cells were nearly identical in all siRNA treatments, and thus, no corrections for these factors were required (see Fig. S1 at http://www.bio.huji.ac.il/eng/staff_in.asp?staff_id=77&chapter_id=198). To confirm the functional knockdown of the targeted gene, mRNA levels were determined by real-time PCR (RT-PCR) (see below), and protein levels were determined by Western blotting.

Determination of RNA levels by real-time PCR.

RNA extracted from 2 × 10⁵ cells (TRI reagent [Sigma]) was subjected to reverse transcription using the High Capacity RNA to cDNA kit (Applied Biosystems). RT-PCR (7900HT RT-PCR system; Applied Biosystems) was performed on 1 μg of the resultant cDNA. RNA levels were determined quantitatively using the ΔΔC_T method, as described previously (37). Levels of HCV genotype 1b (FLRP1) RNA were determined using primers 5′-AGAGCCATAGTGGTCT-3′ (forward) and 5′-CCAAATCTCCAGGCATTGAGC-3′ (reverse) and a 6-carboxyfluorescein (FAM)-CACCGGAATTGCCAGGACGACCGG-6-carboxytetramethylrhodamine probe. Levels of HCV genotype 2a (J6/JFH1) RNA were determined using primers 5′-CTTCACGCAGAAAGCGTCTA-3′ (forward) and 5′-CAAGCACCCTATCAGGCAGT-3′ (reverse) and a FAM-TATGAGTGTCGTGCAGCCTC probe. Human TAT binding protein (hTBP) and Arf1 RNA levels were determined by TaqMan Gene Expression assays (assay ID numbers, Hs99999910_m1 for hTBP and Hs00796826_s1 for Arf1; Applied Biosystems) according to the manufacturer's instructions. All experiments were repeated at least three times, with triplicates of each sample. Importantly, the hTBP RNA levels were not altered in response to the different cell treatments.

Determination of viral RNA levels by a luciferase activity assay.

In vitro-translated RNA was prepared from the J6/JFH (5′C19Rluc2AUbi) clone as previously described (32). For RNA transfection, cells were washed with PBS, trypsinized, and resuspended in complete growth medium. Cells were pelleted by centrifugation (1,200 × g for 5 min at 4°C), washed twice with ice-cold PBS (Lonza, Switzerland), and resuspended in ice-cold PBS at 1.5 × 10⁷ cells/ml. Cells (0.4 ml) were mixed with 5 μg of the RNA transcript, placed in 2-mm-diameter gap electroporation cuvettes (BTX Genetronics, San Diego, CA), and electroporated with five pulses of 99 μs at 820 V over 1.1 s in an ECM 830 electroporator (BTX Genetronics). Following a 10-min recovery period, cells were mixed with complete growth medium and were then plated. Luciferase activity was measured at the indicated time points using a Mithras LB 96V luminometer (Berthold, Germany).

Determination of the levels of infectious HCV in cells and media following BFA, GCA, or siRNA treatment.

Huh7.5 cells were first electroporated with the J6/JFH1 RNA; then, 72 h later, they were treated with either BFA (12 h), GCA (20 h), or the equivalent amount of DMSO as a control. In gene-silencing experiments, cells were electroporated with the J6/JFH1 RNA, and 72 h later, they were transfected with Arf1 siRNA (siArf1) or with control siRNA (siControl) for an additional 72 h to allow gene silencing. Following these treatments, the cells were processed for RT-PCR (see above), for immunofluorescence experiments (see below), or for determination of the levels of infectious virus, as follows. Cell media and cell lysates, prepared as described by Gastaminza et al. (26), were subjected to 10-fold serial dilutions (from 10⁻¹ to 10⁻⁴) with growth medium. The titer of infectious virus particles was determined in Huh7.5 cells by endpoint dilution combined with immunofluorescence, as described previously (81). Briefly, 50 μl taken from each of the diluted samples was used to inoculate naïve Huh7.5 cells cultured on a 96-well plate. Infection was examined 72 h postinoculation by immunofluorescence using a monoclonal antibody directed against core and the appropriate Alexa 594-conjugated secondary antibodies (Invitrogen). Titers of infectious HCV particles were determined by the 50% tissue culture infective dose (TCID₅₀) method according to the work of Reed and Muench (55).

Immunofluorescence analysis and Oil Red O staining.

Cells grown on coverslips were fixed for 20 min at 22°C with 4% (vol/vol) paraformaldehyde (PFA) in PBS. Cells were then permeabilized by incubation with 0.2% (wt/vol) saponin and 0.1% (vol/vol) Triton X-100 in permeabilization buffer (3% FBS in PBS [pH 8.0]). Nonreacted PFA was quenched for 15 min at 22°C with a quenching solution (3% FBS, 50 mM NH₄Cl, 50 mM glycine in PBS [pH 8.0]). Samples were then incubated for 30 min at 37°C in a blocking solution (PBS [pH 8.0] containing 3% FBS and 0.1% saponin). Subsequently, cells were subjected to immunofluorescent staining using primary and secondary antibodies (60 min at 37°C) suspended in the blocking solution. Following extensive washes, coverslips were mounted on glass slides using a polyvinyl alcohol mounting medium (Mowiol; Thermo Scientific, Waltham, MA). Confocal images acquired with an Olympus (Melville, NY) FV1000 microscope were processed as previously described (58). In several experiments, the cells were exposed to Oil Red O staining of LDs prior to the mounting step, as previously described (40).

Confocal microscopy.

Confocal microscopy was used to acquire all fluorescent images, which were subsequently processed by Adobe Photoshop, as described previously (58).

Pulldown assays.

The amount of active (GTP-bound) Arf1 in cells was estimated by pulldown assays using glutathione S-transferase (GST) fused to the Golgi apparatus-localized gamma adaptin ear-containing ARF-binding protein 3 (GGA3) and Tom1 (GAT) domain coupled to glutathione-Sepharose beads, as described previously (79). Western blots were probed with anti-Arf1 monoclonal (clone 3F1) antibodies and with anti-NS5A and anti β-actin antibodies. The pulled-down Arf1 signal was quantified and normalized to the total amounts of Arf1 and actin detected in the cell lysates.

SDS-PAGE and quantitative Western blotting.

Samples were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (40 mA, constant voltage; Hoefer) using 10 or 13% gels. Following electrophoresis, proteins were transferred to nitrocellulose membranes using a semidry transfer apparatus (Bio-Rad, Hercules, CA) at 15 V for 1 h. Nitrocellulose membranes were probed with primary antibodies followed by appropriate HRP-conjugated secondary antibodies. Protein bands were detected by an enhanced chemiluminescence (ECL)-based detection system (Beit Haemek Ltd.) and were exposed to X-ray film. For quantitative analyses, protein bands were scanned, and ImageJ software was used to determine band intensities within the linear range.

RESULTS

BFA and GCA treatments decrease HCV RNA levels in FLRP1 cells.

Observations suggesting that GDP-Arf1 is associated with LDs (30, 38, 68) led to the hypothesis that increasing GDP-Arf1 levels would somehow alter specific steps in the HCV life cycle following the establishment of RCs. To address this hypothesis, we examined the effects of BFA and GCA, a compound recently identified as a GBF1-specific inhibitor (57), on HCV infectious viral particle production and RNA replication. Note that BFA inhibits the activation of Arf small GTPases by targeting the large GEFs BIG1, BIG2, and GBF1. Specifically, BFA locks GDP-Arf in a complex with the GEF, consequently blocking GEF activity at an early stage, prior to guanine nucleotide release. Unlike the BIG GEFs, which act in the trans-Golgi complex-to-endosome pathway, the Arf1-GBF1 complex functions in the cis-Golgi region to mediate COPI-coat-dependent Golgi complex-to-ER vesicular transport (16). Since HCV RCs occupy primarily ER-like membranes attached to LDs, we assumed that the effects of BFA on HCV replication were elicited by targeting GBF1 rather than the BIG proteins. Consequently, we hypothesized that selective inhibition of GBF1 would yield effects similar to those observed in response to BFA treatment.

To test this hypothesis, we examined whether BFA or GCA treatments alter HCV and Arf1 transcript levels in cells expressing FLRP1 genotype 1b replicons. Using an RT-PCR-based assay, we showed that BFA and GCA treatments decrease Arf1 and HCV RNA levels relative to those for DMSO-treated controls (Fig. 1 A). The effect of GCA treatment was not observed after a short period (1 h) of treatment and was partially reversible upon drug washout (Fig. 1B). Interestingly, the protein levels of Arf1, NS5A, and NS3 did not change markedly in response to these treatments (Fig. 1C), suggesting that GCA and BFA reduce transcription but not the levels of proteins encoded by these transcripts.

FIG. 1.

Effects of BFA and GCA treatments on levels of HCV RNA (A and B) and protein (C) in FLRP1 replicons. (A and B) Total cellular RNA was harvested, and levels of HCV and Arf1 RNA were determined by quantitative RT-PCR, as described in Materials and Methods. Data are means ± standard errors for three independent experiments. (A) Cells were treated with either BFA, GCA, or an equivalent volume of DMSO, as described in Materials and Methods. (B) In some experiments, cells were treated with GCA for only 1 h or for 20 h, and immediately thereafter, cells were incubated in drug-free growth medium for an additional 16 h (washout). (C) Equal protein levels were loaded, and after BFA or GCA treatment, levels were analyzed by Western blotting using specific antibodies directed against the indicated proteins.

NS5A redistributes to the rims of LDs in BFA- and GCA-treated cells.

The observation that inhibition of Arf1 by BFA or GCA treatment does not impact either Arf1 or viral protein levels raised the hypothesis that the effects of these agents on viral RNA levels are obtained by altering the intracellular distribution of viral proteins to locations that may not support RNA replication. For instance, it has been shown that the majority of NS5A is localized in speckle-like structures distributed throughout the cell cytoplasm in FLRP1 cells (40). As noted above, these speckles may represent ER-associated organelles that reside in the vicinity of LDs and promote the activity of RCs. Thus, redistribution of viral proteins involved in RNA replication from these speckle sites may be linked to the reduced activity of RCs. We performed a series of confocal imaging analyses to address this hypothesis. Representative images (Fig. 2 A and B) and quantitative image analyses (Fig. 2C) are presented. Indeed, the distribution of NS5A (Fig. 2A) and NS3 (Fig. 2B) in FLRP1 cells shifted considerably from speckles in control cells to the rims of the LDs in BFA- or GCA-treated cells. Notably, the relocalization of NS3 was somewhat less prominent than that of NS5A (see Fig. 2C), and in both cases weak staining was associated with reticular structures distributed throughout the cytoplasm.

FIG. 2.

NS5A and NS3, but not core, are redistributed from speckle-like structures to the peripheries of LDs following BFA and GCA treatments. (A and B) FLRP1 cells were treated with either BFA, GCA, or DMSO (see Materials and Methods for details). In another experiment, cells were first treated with BFA or GCA and were then incubated for 16 h in drug-free medium (washout). Cells were immunostained with anti-NS5A (A), anti-NS3, or anti-core (B) primary antibodies and Alexa 488-labeled secondary antibodies (green). Oil Red O and 4′,6-diamidino-2-phenylindole (DAPI) were used to label LDs (red) and nuclei (blue), respectively. Cells were visualized by confocal microscopy. Images are representative of three independent experiments. (C) LDs whose peripheries were labeled with the viral protein (A, arrow) and LDs that did not show associated staining of viral proteins (A, arrowhead) were scored. A total of 50 to 100 LDs were counted in each experiment. The fraction of LDs (percentage of total) exhibiting peripheral viral protein labeling was determined per microscopic field. Results are means ± standard errors for six fields of view.

The effect was largely reversible; following drug removal (washout), both proteins spread back to their speckled and ER-like reticular membrane distribution. Core, unlike NS5A, showed significant basal levels of LD association prior to the treatment of cells with BFA (6, 40) and no apparent shift from speckles to an LD-associated signal (Fig. 2B). Taken together, these results suggest that some of the viral NS proteins migrate from speckle-like organelles to the peripheries of LDs in response to Arf1 inhibition by BFA or GCA treatment.

BFA and GCA treatments decrease HCV RNA levels and cause redistribution of NS5A and accumulation of infectious viral particles in J6/JFH1 cells.

In agreement with the findings for FLRP1 cells (Fig. 1A), two independent assays showed that BFA and GCA treatments diminish Arf1 and HCV RNA levels in the JFH1-HCVcc cell system (Fig. 3 A). A massive, but reversible, redistribution of NS5A from speckle-like to reticular structures surrounding the LDs was also observed (Fig. 3B).

FIG. 3.

Effects of BFA or GCA treatment on viral RNA levels (A), NS5A distribution (B), and the level of infectious viral particles (C). (A) (Top) GCA and BFA reduce HCV and Arf1 RNA levels. J6/JFH1-expressing cells were treated with BFA or GCA. Arf1 and HCV RNA levels were determined by RT-PCR. The results are means ± standard errors for three independent experiments. (Bottom) GCA inhibits HCV replication. Huh7.5 cells were electroporated with J6/JFH (5′C19Rluc2AUbi), a monocistronic reporter virus expressing Renilla luciferase, and were cultured in the presence or absence of 10 μM GCA for 4, 24, and 48 h. Control cells were treated identically with DMSO. Cells electroporated with a replicon containing the GND inactivating mutation in the viral polymerase (Pol⁻) served as a negative control. Duplicate samples were harvested for luciferase assays at the indicated time points posttreatment. Values were normalized to the input RNA levels measured at 4 h. Results are means ± standard errors. (B) NS5A is redistributed from speckle-like structures to the peripheries of LDs following GCA treatment. (Top) J6/JFH1-expressing cells were treated with GCA and stained with anti-NS5A, Oil Red O, and DAPI, as for Fig. 2A. (Bottom) NS5A associated with the peripheries of LDs was quantified as for Fig. 2C. (C) Effects of BFA and GCA on infectious viral particle levels. Huh7.5 cells were electroporated with J6/JFH1 RNA, and 72 h later, they were treated with either BFA (12 h), GCA (20 h), or an equivalent volume of DMSO as a control. Titers of infective HCV particles were determined in the cell medium (Extracellular) and lysates (Intracellular) by the TCID₅₀ method (see Materials and Methods). Results are means ± standard errors for three experiments performed independently, each in triplicate. P, <0.001 by a two-tailed t test of treated cells versus untreated controls.

Next, we asked how Arf1 inhibition affects the biogenesis and secretion of HCV infectious particles in this cell culture. To address this question, Huh7.5 cells were electroporated with pJ6/JFH1 RNA and were allowed to establish replication and to produce infectious HCV particles for 3 days. Immunofluorescent data indicated that under these conditions, all of the cells express NS5A and core, suggesting that active RCs were generated in essentially all of them (not shown). Cells were then treated with BFA or GCA, and levels of infectious particles both in cell lysates and in the extracellular medium were determined. The results indeed showed that both treatments resulted in a decrease in the levels of infectious viral particles associated with the extracellular medium and an increase in their levels within the cells (Fig. 3C). Since it has been suggested that prolonged BFA treatment does not impair viral infectivity (25) but rather disrupts key steps in the secretory pathways, the observed effects are likely due to inhibition of HCV secretion. Combining these data with those obtained in the FLRP1 system, we conclude that Arf1 inhibition by BFA and GCA treatments yields similar effects on viral RNA levels and NS5A localization; hence, these data implicate GBF1 and its target Arf1 in mediating these processes.

NS5A expression suppresses GTP-Arf1 levels.

We hypothesized the existence of cross talk between HCV viral proteins and Arf1. To address this hypothesis, we examined whether NS5A is capable of modulating the GTP/GDP-bound state of Arf1. Huh7 cells were transfected with an NS5A-encoding plasmid, and cells were allowed to express the protein for 3 days. Cellular GTP-Arf levels were assayed quantitatively in untreated versus NS5A- or mock-transfected cells by a pulldown assay. In this assay, the GGA domain, fused to GST and coupled to glutathione-Sepharose beads, was used to pull down active (GTP-bound) Arf1 (79), as described in Materials and Methods. The results showed a significant reduction in GTP-Arf1 levels associated with NS5A-expressing cells but not with mock-transfected cells (Fig. 4). This result raises the possibility that the expressed viral protein reduces the GTP-bound state of Arf1, an activity that might be elicited either by facilitating GTPase-activating protein (GAP) activity (e.g., through interactions with Arf1-GAP) or by inhibiting Arf1-GEFs. The former scenario could be supported by previous findings suggesting that NS5A is complexed to the ArfGAP ARAP1 (1). Of course, other factors, such as NS5A-induced covalent modification of the Arf protein, may also contribute to the decreased Arf1 pulldown. In any case, our results provide first evidence for the possible existence of functional interactions between NS5A and Arf1, or Arf1 modulators.

FIG. 4.

Ectopic expression of NS5A reduces GTP-Arf1 levels. Huh7 cells were either transfected with pEF6-NS5A, mock transfected with the pEF6 vector, or left untreated (cont.). Seventy-two hours after transfection, cells were lysed and subjected to a pulldown assay using GGA-GST bound to glutathione-Sepharose beads, as described in Materials and Methods. (Top) Representative Western blot. Equal amounts of cell lysates of the pulled-down material were probed with anti-Arf1, anti-NS5A, and anti-actin antibodies. (Bottom) Quantitative analyses of three experiments. Results are means ± standard errors for three independent experiments (P, <0.001 by a two-tailed t test of NS5A- versus mock-transfected cells).

NS5A colocalizes with dominant negative Arf1 in the vicinity of LDs.

The ability of NS5A to diminish GTP-Arf1 levels suggests that the two proteins can interact physically or maintain functional interactions by modifying signaling pathways in organelles that harbor them. Attempts to disclose these interactions by coimmunoprecipitation have thus far been unsuccessful. To gain further insight into the possible cross talk between these proteins, we analyzed the colocalization of viral proteins and ectopically expressed wild-type (WT) and mutant GFP-tagged Arf1 forms.

The dominant negative mutant form of Arf1 fused to GFP (Arf1 T31N-GFP), which is thought to be analogous to GDP-Arf1 and whose overexpression triggers BFA-like phenotypes (15), was expressed in FLRP1 replicon cells (Fig. 5 A) and was coexpressed with NS5A in plain Huh7 cells (Fig. 5D). The cells were then fluorescently immunostained with anti-NS5A, and LDs were labeled with Oil Red O. The mutated Arf protein was associated mainly with the rims of LDs (indicated by arrows). A quantitative analysis is presented in Fig. 5G. In contrast, expression of constitutively active Arf1 (Arf1 Q71L-GFP) or Arf1 WT-GFP in FLRP1 (Fig. 5B and C) or plain Huh7 cells (Fig. 5E and F) localized mainly to the juxtanuclear region, which most likely represents the Golgi apparatus and recycling endosomes. The level of colocalization of these Arfs with the NS protein at the peripheries of the LDs appeared to be marginal (Fig. 5G). Arf1 T31N-GFP had also preferentially labeled the peripheries of LDs in plain Huh7 cells (see Fig. S4 at http://www.bio.huji.ac.il/eng/staff_in.asp?staff_id=77&chapter_id=198), suggesting that the mere presence of viral proteins did not determine the LD-associated appearance of the mutant Arf. In parallel experiments, Arf1 WT-GFP, Arf1Q71L-GFP, or Arf1 T31N was coexpressed with core, which, unlike NS5A, shows predominant localization to LDs. Imaging of these cells showed clear segregation between Arf WT or Arf Q71L, which localized in the juxtanuclear region, and core (see Fig. S5 at http://www.bio.huji.ac.il/eng/staff_in.asp?staff_id=77&chapter_id=198). In contrast, Arf1T31N colocalized significantly with core at the peripheries of the LDs (see Fig. S5, arrows). These results suggest that GDP-Arf1 and viral proteins coreside in the peripheries of LDs. At these sites, Arf1 may modulate the localization and function of the viral proteins.

FIG. 5.

NS5A colocalizes with Arf1 T31N-GFP at the peripheries of LDs. (A to C) FLRP1 cells were transfected with the indicated Arf1-GFP constructs. (D to F) Huh7 cells were cotransfected with Arf1-GFP constructs and pEF6-NS5A. Cells were immunostained with anti-NS5A antibodies (blue) and with Oil Red O (red). Cells were then visualized by confocal microscopy. Arrows indicate sites of NS5A and Arf colocalization at the peripheries of LDs. (G) Quantitative analysis of GFP-Arf and NS5A associated with the peripheries of LDs was performed as for Fig. 2C.

Downregulation of Arf1 expression by siRNA diminishes levels of HCV RNA and infectious particles.

Our data up to this point suggested that GDP-Arf1 regulates HCV RNA replication and the release of infectious viral particles to the external milieu. Next, we asked how the reduction in Arf1 expression levels by siRNA affects HCV replication and viral particle production. siRNA treatment of FLRP1 cells suppressed Arf1 protein expression levels by ∼80% (Fig. 6 A, top) and Arf1 RNA levels by about 75% (Fig. 6A, bottom) relative to those for controls. Arf1 siRNA did not appear to affect the expression levels of NS5A, NS3, and actin in these cells (Fig. 6A, top), but it certainly resulted in considerable inhibition (∼50%) of HCV RNA replication (Fig. 6A, bottom). The half-life of the NS proteins (11 to 16 h [52]) is comparable to that of the RNA (about 11 h [51]). However, each positive-strand RNA template is translated numerous times, giving rise to anywhere from 1,000 to 10,000 polyprotein copies (54). The latter characteristic may have contributed to the rather stable protein levels, despite the partial decrease in viral RNA levels in siRNA-treated cells.

FIG. 6.

Arf1 knockdown by siRNA treatment decreases HCV RNA replication and the levels of infectious viral particles. (A and B) FLRP1 cells. Cells were transfected with Arf1 siRNA (siArf1) or with a control nontargeting siRNA (siControl), as described in Materials and Methods. (A) (Top) Cells were then lysed and subjected to Western blot analysis using antibodies directed against NS5A, NS3, Arf1, or actin. (Bottom) In parallel experiments, total cellular RNA was extracted and used to determine HCV and Arf1 RNA levels by quantitative RT-PCR. Results are means ± standard errors for three experiments, each performed in triplicate. Asterisks indicate significant differences (*, P < 0.05; **, P < 0.0001) by a two-tailed t test. (B) (Top) Cells treated with siArf1 or siControl were immunostained with anti-NS5A (green) and Oil Red O (red). (Bottom) NS5A associated with LDs was scored as for Fig. 2C. (C and D) J6/JFH1-expressing cells. Huh7.5 cells were electroporated with J6/JFH1 and were transfected with siArf1 or siControl. (C) Total RNA was extracted, and quantitative RT-PCR was used to determine the levels of HCV and Arf1 RNA. The results are means ± standard errors for three independent experiments (P < 0.0001). (D) Levels of infectious HCV particles were determined in the cell medium (Extracellular) and cell lysates (Intracellular). Results are means ± standard errors for three experiments performed independently, each in triplicate. P, <0.0001 by a two-tailed t test for siArf1-treated versus siControl-treated cells.

As with the BFA and GCA treatments, NS5A shifted its distribution from speckles in the siControl cells to regions decorating the peripheries of LDs in siRNA-treated cells (Fig. 6B).

Finally, Arf1 siRNA treatment also caused a significant reduction in Arf1 (∼75%) and HCV (∼40%) RNA levels in the JFH1-HCVcc system (Fig. 6C). The treatment dramatically reduced the levels of infectious viral particles associated with these cells and with their extracellular milieu (Fig. 6D). In conclusion, these results suggest that Arf1 expression is essential for HCV replication and NS5A localization in speckles, which may represent RCs.

DISCUSSION

Previous publications have indicated the importance of BFA- and GBF1-sensitive factors in HCV replication (25, 29). In addition, coat proteins that require Arf1 for their recruitment onto membranes, such as COPI, and phosphatidylinositol-4-kinase III alpha (PI4KIII alpha), recently described to be essential for the recruitment of GBF1 to Golgi membranes (21), are also required for HCV replication (10, 68). However, data implicating Arf1 directly in viral RNA replication and infectious particle production have not been provided. Here, by using Arf1 inhibitors and siRNA-mediated knockdown experiments, we show that HCV RNA levels, the quantity of infectious virions, and the localization of viral proteins depend on the activity and expression of Arf1.

Inhibition of Arf1 caused a significant shift in the localization of viral proteins (e.g., NS3 and NS5A) from speckles to the peripheries of LDs (Fig. 2). Arf1 knockdown also relocalized NS5A to regions lining the peripheries of LDs (Fig. 6B), suggesting that this phenomenon is due to interference with Arf1 functions. Interestingly, a recent genome-wide RNA interference (RNAi) screen in Drosophila revealed that knockdown of members of the COPI machinery, including Arf1 and its GEF, results in similar changes in the morphology and lipid utilization of LDs (30). Moreover, inhibition of Arf1 by BFA treatment or expression of a dominant negative version of Arf caused a similar effect (30). On the basis of these observations, we hypothesize that GTP-Arf1 could be required for retaining the distribution of NS proteins in the ER and associated RCs (speckles), where they mediate HCV RNA replication. Extensive inhibition of Arf1 by BFA and GCA, or knockdown of Arf1 expression, caused modifications in the morphology and functions of the LDs, as well as a shift in the localization of NS proteins from areas where they are required for efficient viral replication (e.g., speckles) to areas that do not support replication (e.g., the peripheries of LDs). These alterations may have contributed to the observed inhibition both of viral RNA synthesis and of the production of infectious viral particles.

However, the effects of BFA/GCA treatments differed from that of siRNA with respect to the distribution of infectious progeny viruses within the cells and the extracellular environment. Whereas the former caused accumulation of infectious particles within the cells and decreased their levels in the external milieu (Fig. 3C), siRNA treatment caused decreases in infectious particle levels in both environments (Fig. 6D). These differential effects might be explained by studies showing that unlike BFA treatment, knockdown of Arf1 by RNA interference has no effect on membrane traffic or Golgi morphology (75). This finding leads to the conclusion that BFA or GCA, but not siRNA, caused the accumulation of infectious viral particles in the cells due to inhibition of HCV secretion. Thus, our studies suggest that HCV may act like other positive-strand RNA viruses that manipulate the early secretory pathway by targeting GBF1 and Arf1 (74, 76, 77).

Recent studies have implicated Rab1 and its GAP in HCV replication (64, 65). Rab1 has been suggested to interact with the Arf1 GEF GBF1 (2, 43) and to be associated with LDs (59, 80). Moreover, a recent study has intriguingly suggested that Rab1 is required for the timely and specific recruitment of GBF1 to membranes by activating PI4KIII alpha (21). Thus, an intriguing hypothesis is that HCV somehow modulates the activities of all these proteins to promote its own replication. We hypothesize, for instance, that downregulation of the signaling pathways emanating from these proteins, e.g., by the viral proteins themselves, produces a localized BFA/GCA-like effect that shuttles essential membrane components, including some viral proteins (35, 60), in the Golgi complex-ER-LD interfaces to regulate HCV replication and biogenesis. Our observation suggesting that NS5A expression inhibits Arf1 activity (e.g., through Arf1 GAP activation or GEF inhibition [Fig. 4]) may provide initial supportive evidence for this prediction.

Interesting in this context are observations suggesting that HCV negatively regulates the synthesis and secretion of VLDLs, components shown to be required for the morphogenesis and secretion of HCV (for a recent review, see reference 67). The dominant negative Arf1 T31N mutant has been suggested to decrease the assembly of Apo-B-VLDLs (5). Hence, we hypothesize that HCV hijacks components of the VLDL secretory pathway to downregulate its own secretion by NS5A-mediated inhibition of Arf1.

The LD-associated adipose differentiation-related protein (ADRP) is a major member of the PAT family of LD-associated proteins. ADRP and other PAT family proteins that are positioned on the surfaces of LDs have been suggested to regulate the access of proteins, such as lipases, to these droplets (11). Data have shown that progressive coating of LDs with HCV core correlates with ADRP dissociation from LDs and alterations in intracellular LD localization, events that appear to be important for virus production (14). This notion might also be supported by proteomic profiling analyses (59). Thus, the mere presence of ADRP on LDs may hinder the association of viral proteins with these organelles, and ADRP may serve as a regulator of HCV RNA replication and virus assembly. Interestingly, expression of a dominant negative mutant of Arf1, which mimics GDP-Arf1 and the action of BFA (15), has been reported to cause the dissociation of ADRP from LDs (47). Moreover, inhibition of GBF1/Arf1/COPI also inhibited the delivery of ADRP from ER exit sites to LDs (66). These data suggest that inhibition of Arf1 increases the access of viral proteins to LDs through a decrease in the ADRP-LD association. This process could have facilitated the enhanced localization of viral proteins at the peripheries of LDs observed here in BFA- and GCA-treated cells (Fig. 2A and B and 3B).

Arf1 and its modulators have been implicated in diverse cellular functions, including the remodeling of phosphoinositide signaling and of the actin cytoskeleton (17, 18, 46). Since the actin cytoskeleton (13) and signaling elements associated with Arf1 have been implicated in HCV pathogenesis (1, 10, 33, 68, 72, 73), it is entirely conceivable that a reduction in Arf1 expression (by siRNA) would result in the collapse of Arf1-dependent signaling networks that play a role in HCV RNA replication and infectious virus biogenesis.