A comparative cell biological analysis reveals only limited functional homology between the NS5A proteins of hepatitis C virus and GB virus B

GB virus B (GBV-B) is the closest phylogenetic relative of hepatitis C virus (HCV) (Muerhoff et al., 1995; Ohba et al., 1996); it shares up to 28 % overall amino acid identity with HCV, rising to 50 % conservation in regions such as the NS3 coding sequence. GBV-B has been proposed as a model for HCV infection and has been used to characterize candidate antivirals as well as for pathogenesis studies (Beames et al., 2001; Bright et al., 2004). It shares the hepatotropism of HCV. Although GBV-B has been reported to lead to persistent infections in some cases in tamarins (Saguinus species) (Martin et al., 2003; Nam et al., 2004), unlike the human virus, GBV-B primarily causes an acute, resolving hepatitis in tamarins and other New World monkeys such as common marmosets (Callithrix jacchus) (Bukh et al., 2001; Jacob et al., 2004; Lanford et al., 2003). In contrast, HCV is a major cause of chronic hepatitis, liver cirrhosis and hepatocellular carcinoma, affecting an estimated 3 % of the world's population.

Both GBV-B and HCV are members of the genus Hepacivirus of the family Flaviviridae and have a positive-sense RNA genome that is translated in a cap-independent fashion to generate a polyprotein, cleaved by host and viral proteases to produce 10 mature viral proteins. There are three structural proteins at the N terminus: core, E1 and E2, followed by a small cation channel, p7 (p13 in GBV-B). The C-terminal two-thirds of the polyprotein comprises six non-structural proteins: NS2, NS3, NS4A, NS4B, NS5A and NS5B.

Much work has focused on the HCV NS5A protein, it is believed to be a component of an RNA replication complex on cytoplasmic membranes together with the other non-structural proteins (Egger et al., 2002; Lohmann et al., 1999; Mottola et al., 2002). In addition NS5A has been shown to bind to and modulate a range of cellular proteins, influencing activation of signalling pathways (Macdonald & Harris, 2004). Of particular interest is the observation that NS5A inhibits the Ras–Erk mitogen-activated protein kinase (MAPK) pathway acting between the epidermal growth factor receptor (EGFR) and the activation of Ras (Georgopoulou et al., 2003; Macdonald et al., 2003, 2005a; Tan et al., 1999). Furthermore, NS5A has also been shown to bind to, and activate, phosphoinositide 3-kinase (PI3K), resulting in activation of the downstream effector serine/threonine kinase Akt/protein kinase B (He et al., 2002; Street et al., 2004) and elevating β-catenin-dependent transcription (Street et al., 2005). Perturbations of these signalling pathways may contribute to the ability of HCV to establish a persistent infection (Macdonald & Harris, 2004).

Given the differences in the pathology of GBV-B and HCV infections it is of interest to determine whether the individual proteins of the viruses have common functions in perturbing host-cell biology. We therefore characterized the subcellular localization of GBV-B NS5A and its ability to modulate host cell Ras–Erk and PI3K pathways. Unlike HCV NS5A, GBV-B NS5A failed to inhibit Ras–Erk signalling but retained the ability to elevate β-catenin-dependent transcription via activation of the PI3K pathway. These characteristics were observed both for GBV-B NS5A expressed alone from a transiently transfected plasmid and also in the context of Huh-7 cells harbouring the GBV-B subgenomic replicon. Lastly, again in comparison with HCV NS5A, GBV-B NS5A protein did not appear to traffic to endosomal compartments.

DNA manipulation and constructs.
The pSG5-NS5A wild-type and PA2.2 mutant plasmids have been described previously (Macdonald et al., 2003). The NS5A coding sequence from the GBV-B genome (GenBank accession no. NC_001655[GenBank] , kindly provided by Jens Bukh, NIH) was amplified by PCR to include a start codon N-terminal to the GBV-B NS5A sequence, a C-terminal FLAG epitope tag (DYKDDDDK) and a stop codon. The resulting PCR product was digested with EcoRI and cloned into the mammalian expression vector, pSG5 (Green et al., 1988). The construct was verified by DNA sequencing. All primer sequences are available on request. The bicistronic GBV-B replicon used in this study (GBneoU-Q1278K+E2346G, referred to as GBneoU) contains the neomycin phosphotransferase gene and GBV-B NS3-NS5B encoding sequences derived from a GBV-B infectious molecular clone (GenBank accession no. AY243572[GenBank] ) in which two cell-culture-adaptive mutations are present at positions 4277 and 7482 of the GBV-B cDNA (L. Warter, L. Cohen & A. Martin, unpublished data). The NS5A protein of this replicon differs from that in the pSG5 expression construct by 1 aa (an Ala to Val mutation at aa 2236 of the polyprotein). A luciferase reporter construct regulated by a promoter sequence responsive to AP-1 (pAP1-luc) was used for assessment of Ras–Erk activation as described previously (Macdonald et al., 2003). Vectors expressing either wild-type β-catenin or HA-tagged dominant-negative Akt (K179M) have been described previously (Street et al., 2005). The β-catenin-responsive reporter M50 contains eight copies of wild-type Tcf-binding sites upstream of luciferase, this and the control construct M51 (containing mutated Tcf sites) were obtained from Randall Moon (University of Washington, Seattle, USA).

Cell culture.
Cos-7 (African green monkey kidney cells) were cultured in Dulbecco's modified Eagle's medium supplemented with 10 % fetal calf serum (FCS), 2 mM L-glutamine, 100 IU penicillin ml^–1 and 100 µg streptomycin ml^–1. Huh-7 (human hepatoma cells) and the GBV-B-cured derivative cB76.1/Huh-7 [kindly provided by Cinzia Traboni, Istituto di Ricerche Biologia Moleculare (IRBM), Rome, Italy] were cultured in minimal essential medium supplemented with 10 % FCS, 1 % non-essential amino acids, 2 mM L-glutamine, 100 IU penicillin ml^–1 and 100 µg streptomycin ml^–1. Huh-7 and cB76.1/Huh-7 cells harbouring culture-adapted subgenomic replicons of either HCV (FK5.1) or GBV-B (GBneoU), respectively, were maintained in the presence of 250 µg G418 ml^–1. Cells were incubated at 37 °C in a humidified 5 % CO₂ incubator.

Transfection of plasmid DNA and luciferase assay.
One day prior to transfection, cells were seeded (2x10⁵) into six-well dishes. Cells were transfected using polyethylenimine (PEI; Polysciences) according to the manufacturer's instructions. AP-1 luciferase reporter assays were performed as described previously (Street et al., 2004). Briefly, cells transfected with pAP1-luc and NS5A expression vectors were incubated with the transfection reagent for 24 h, after which the transfection reagent was removed and the cells were overlaid with growth medium. For β-catenin assays, cells were transfected with either M50 or M51 (0.5 µg), together with hemagglutinin-epitope-tagged Akt [HA-Akt(K179M)] (0.5 µg), β-catenin expression vector (1.0 µg) and pSG5 expression vectors. In all experiments a Renilla luciferase reporter construct (pRLTK) acted as an internal control for transfection efficiency and the total amount of DNA was kept constant by inclusion of empty vector DNA. For AP-1 luciferase experiments, transfected cells were maintained under low-serum growth conditions (serum-free or 0.5 % serum) prior to stimulation with 10 % serum-containing growth medium for 6 h. Where appropriate lithium chloride (10 mM) was added for 12 h prior to harvest. Cells were lysed with 200 µl passive lysis buffer (Promega) and luciferase levels were measured using the Stop and Glo reagent (Promega) and a luminometer (EG and Berthold). All assays were performed in triplicate, and each experiment was repeated a minimum of three times.

Western blotting.
To analyse the expression of FLAG-tagged GBV-B NS5A, cells were lysed in Glasgow lysis buffer (10 mM PIPES-NaOH pH 7.2, 120 mM KCl, 30 mM NaCl, 5 mM MgCl₂, 1 % Triton X-100 and 10 % glycerol) plus protease inhibitors (Roche Complete) and phosphatase inhibitors (2 mM Na₃VO₄, 5 mM NaF and 5 mM Na₄P₂O₇). Lysates (50 µg total protein) were resolved by SDS-PAGE and probed with an anti-FLAG monoclonal antibody (1 µg ml^–1; Sigma) or a sheep polyclonal antiserum raised against bacterially expressed HCV NS5A (Macdonald et al., 2003) followed by appropriate horseradish peroxidase-conjugated secondary antibodies (Sigma). Blots were visualized using enhanced chemiluminescence (Amersham Biosciences). To analyse the activity status of Erk1/2 or Akt, lysates were probed with antibodies specific to phospho-Erk1/2, pan-Erk1/2, phospho-Akt or pan-Akt (Cell Signalling Technology). Antibodies to neomycin phosphotransferase and GAPDH were obtained from Upstate Biotechnology and Abcam, respectively.

Immunofluorescence.
Huh-7 cells grown on glass coverslips were transfected with plasmids expressing FLAG-tagged GBV-B NS5A or HCV NS5A and the indicated marker proteins. At 48 h post-transfection cells were fixed with 4 % paraformaldehyde or ice-cold methanol for 10 min followed by permeabilization in ice-cold methanol/acetone for 10 min. Cells were washed with PBS and blocked in PBS/1 % BSA for 30 min, and then incubated with a mouse monoclonal anti-FLAG antibody (1 µg ml^–1) at room temperature for 1 h in PBS/1 % BSA. After washing with PBS, Alexa Fluor 488 nm- or 594 nm-conjugated anti-mouse secondary antibodies (Molecular Probes) in PBS/1 % BSA were added for 1 h at room temperature. After washing, the cells were mounted onto microscope slides using Citifluor. Labelled cells were then viewed on a Zeiss 510-META laser scanning confocal microscope under an oil-immersion x63 objective lens (numerical aperture=1.40). Alexa Fluor 488 nm was excited using an argon laser fitted with a 488 nm filter and Alexa Fluor 594 nm with a helium/neon laser fitted with a 543 nm filter. Detection of HCV NS5A was accomplished using a polyclonal sheep NS5A serum (Macdonald et al., 2003) followed by Alexa Fluor 488/594 nm-conjugated anti-sheep secondary antibodies.

Expression of a C-terminally FLAG-tagged GBV-B NS5A
The NS5A coding sequence from GBV-B, corresponding to polyprotein residues 1864–2274, was amplified by PCR and cloned into the mammalian expression vector pSG5 with a C-terminal FLAG tag. A C-terminal tag was chosen so as not to interfere with the function of the N-terminal amphipathic helix present in GBV-B NS5A (Brass et al., 2007). Expression of the FLAG-tagged GBV-B NS5A was verified by Western blot analysis (Fig. 1a), demonstrating that the expressed protein was smaller than its HCV counterpart with an apparent molecular mass of approximately 52 kDa. In common with HCV NS5A, both proteins exhibited a significantly larger apparent molecular mass in comparison with their actual molecular mass, for FLAG-tagged GBV-B NS5A this value is 45 kDa. This discrepancy may be due to the high proline content of the protein (10 %) or to post-translational modifications.

(69K): Fig. 1. Expression and distribution of GBV-B NS5A. (a) Western blot analysis of GBV-B NS5A expression. Lysates from Huh-7 cells transfected with either pSG5.NS5A-FLAG (GBV-B) (lanes 1 and 3) or pSG5.NS5A (wild-type) (lanes 2 and 4) were probed with an anti-FLAG antibody (left panel) and then reprobed with a sheep polyclonal anti-HCV NS5A (Macdonald et al., 2003) (right panel). Note that the residual signal from the anti-FLAG blot can be seen upon reprobing, illustrating the difference in apparent molecular masses of the two NS5A proteins. (b) Cells were co-transfected with pSG5.NS5A (wild-type) and pSG5.NS5A-FLAG (GBV-B), fixed 48 h post-transfection with 4 % paraformaldehyde and permeabilized with methanol–acetone. GBV-B NS5A was visualized with a mouse anti-FLAG antibody followed by Alexa Fluor 594 nm-conjugated anti-mouse secondary antibody, HCV NS5A with a polyclonal sheep NS5A serum (Macdonald et al., 2003) followed by Alexa Fluor 488 nm-conjugated anti-sheep secondary. Representative confocal images are shown. Bar, 10 µm. The lower images are higher magnifications of the boxed areas.

The subcellular distribution of the GBV-B NS5A protein differs from that of the HCV NS5A protein
When expressed either alone, or in conjunction with other HCV proteins in the context of the subgenomic replicon, HCV NS5A displays a punctate cytoplasmic staining that is at least partially associated with the endoplasmic reticulum (Mottola et al., 2002; McCormick et al., 2006). Both HCV and GBV-B NS5A proteins contain an N-terminal amphipathic helix reported to mediate membrane association (Brass et al., 2002, 2007). To investigate if GBV-B NS5A exhibited the same subcellular distribution as HCV NS5A, cells were co-transfected with pSG5 vectors expressing the FLAG-tagged GBV-B NS5A and HCV NS5A and analysed by immunofluorescence and confocal microscopy. As shown in Fig. 1(b), although both proteins exhibited a punctate cytoplasmic staining pattern, there was only limited co-localization. HCV NS5A exhibited a broader distribution, as judged by the more widespread green fluorescence.

We (J. Mankouri, S. Griffin & M. Harris, unpublished data), and others (Stone et al., 2007; Tang et al., 2007), have recently observed partial co-localization of HCV NS5A with early endosome markers. To determine whether this was also the case for GBV-B NS5A, Huh-7 cells transfected with plasmids expressing either HCV or GBV-B NS5A were co-stained with antibodies to either HCV NS5A or FLAG, together with one to EEA1, a marker of early endosomes (Fig. 2a). In comparison with HCV NS5A, which exhibited a partial co-localization with EEA1, GBV-B NS5A did not co-localize at all with EEA1. To confirm this result we also co-transfected cells with plasmids expressing either HCV or GBV-B NS5A and a Rab5–GFP fusion protein, again targeted to the early endosomal compartment. As shown in Fig. 2(b), HCV NS5A partially co-localized with Rab5–GFP, whereas GBV-B NS5A failed to co-localize with the early endosomal marker. These data suggest that HCV NS5A displays a broader subcellular distribution than GBV-B NS5A, which may aid in its ability to bind host-cell proteins that circumvent host-cell signal transduction. We postulated this might result in differential effects on host-cell signalling pathways between the two proteins.

(98K): Fig. 2. Subcellular localization of GBV-B NS5A. Huh-7 cells were transfected with either pSG5.NS5A (wild-type) or pSG5.NS5A-FLAG (GBV-B) alone (a), or together with a Rab5–GFP expression plasmid (b). At 48 h post-transfection cells were fixed and permeabilized as in Fig. 1. NS5A was visualized with a polyclonal sheep anti-NS5A serum followed by Alexa Fluor 488 nm- (a) or 594 nm (b)-conjugated anti-sheep secondary antibodies. EEA1 was labelled with a mouse monoclonal antibody followed by an Alexa Fluor 594 nm-conjugated anti-mouse secondary antibody. GBV-B NS5A-FLAG was labelled as in Fig. 1. Bar, 10 µm. The images on the right are higher magnifications of the boxed areas. Representative confocal images are shown.

GBV-B NS5A does not inhibit activation of the AP-1 transcription factor via the Ras–Erk MAPK signalling pathway
Previous work by our group (Macdonald et al., 2003, 2005a) and others (Georgopoulou et al., 2003) has shown that NS5A, when expressed either alone or in the context of the complete polyprotein, is able to block Ras–Erk-pathway-mediated activation of the AP-1 transcription factor in a manner that was dependent on the presence of a conserved C-terminal polyproline motif (consensus sequence PxxPxR) within NS5A. As GBV-B NS5A lacks this PxxPxR motif we predicted that GBV-B NS5A would be unable to modulate this mitogenic signalling cascade. To test this we utilized a previously described luciferase reporter assay in Cos-7 and Huh-7 cells (Macdonald et al., 2003). Cos-7 cells were utilized as GBV-B infects small New World monkeys, such as tamarins (Bukh et al., 2001) and we thought it was important to investigate this effect additionally in a cell line of non-human origin. GBV-B- or HCV NS5A-expressing plasmids were co-transfected with a reporter construct in which luciferase expression was regulated by three tandem sequences corresponding to the AP-1-binding site (pAP1-luc). As shown in Fig. 3(a), incubation of control transfected cells in medium containing 10 % serum for 6 h resulted in a threefold (Huh-7) or ninefold (Cos-7) stimulation of the AP-1 luciferase reporter construct, compared with cells maintained in low (0.5 %) serum. As previously demonstrated (Macdonald et al., 2003), in cells expressing wild-type HCV NS5A, serum-stimulated levels of luciferase were reduced by up to 75 %, whereas the NS5A PA2.2 mutant (containing alanine substitutions for the proline residues within the PxxPxR motif) had no effect. Consistent with the absence of a PxxPxR motif in GBV-B NS5A, cells expressing this protein displayed either no reduction (Cos-7 cells), or only a modest reduction (Huh-7) in AP-1 induction levels in comparison with control experiments, the reduction in Huh-7 cells was not statistically significant (as determined by one-way ANOVA and the Bonferroni test). We conclude from these data that the GBV-B NS5A protein is unable to block activation of the Ras–Erk pathway.

(24K): Fig. 3. GBV-B NS5A fails to inhibit AP-1 activation. (a) Huh-7 (left chart) or Cos-7 (right chart) cells were co-transfected with pAP1-luc, and either pSG5 vector, pSG5.NS5A-FLAG (GBV-B), pSG5.NS5A (WT, wild-type) or pSG5.NS5A (PA2.2) (Macdonald et al., 2003), and placed in low-serum (0.5 %) growth medium overnight. Cells were stimulated by the addition of 10 % serum and harvested 6 h later. The level of expression of the luciferase reporter was assayed using a luminometer and normalized for transfection efficiency by using a co-transfected Renilla luciferase (RLU) control plasmid (n=9). (b) Naïve Huh-7 cells, or Huh-7 cells harbouring the culture-adapted subgenomic replicons of HCV (FK5.1) or GBV-B (GBneo-U) were transfected and assessed for AP-1 luciferase activation as in (a). **, Values significantly different (P<0.05) from control as determined by one-way ANOVA and the Bonaferroni test.

Given that NS5A is expressed as part of a polyprotein, we considered it important to determine whether Ras–Erk signalling was modulated by GBV-B NS5A in the context of the other viral non-structural proteins. To address this possibility, we assessed AP-1-driven luciferase activity in Huh-7 cells stably harbouring culture-adapted NS3-5B subgenomic replicons of either HCV (Krieger et al., 2001) or GBV-B (L. Warter, L. Cohen & A. Martin, unpublished data). Fig. 3(b) illustrates that following serum stimulation of either naïve Huh-7 cells or those harbouring the GBV-B subgenomic replicon, levels of AP-1-driven luciferase increased by three- to fourfold, whereas in cells harbouring the HCV subgenomic replicon the activation of AP-1 by serum treatment was abrogated. As in Huh-7 cells expressing GBV-B NS5A alone (Fig. 3a), there was a modest reduction in AP-1 activation in GBV-B subgenomic replicon cells, however, again this difference from control cells was not significant. These data confirm that, unlike HCV NS5A, when expressed as part of an RNA replication competent subgenomic replicon (i.e. in a physiologically relevant context) GBV-B NS5A is unable to block Ras–Erk signalling in Huh-7 cells.

Phosphorylation of Erk1/2 is not inhibited in cells harbouring the GBV-B subgenomic replicon
The NS5A-mediated inhibition of AP-1 has been shown previously to involve only the Ras–Erk pathway (Macdonald et al., 2003). To provide further evidence that GBV-B is unable to inhibit this pathway we sought to confirm that phosphorylation of Erk1/2 is maintained in cells harbouring the GBV-B subgenomic replicon. Naïve Huh-7 cells were compared with Huh-7 cells harbouring either GBV-B or HCV subgenomic replicons following serum starvation and EGF stimulation (Fig. 4a). After stimulation, levels of phospho-Erk1/2 increased in naïve cells, peaking at 0.5 h post-stimulation and declining thereafter (lanes 1–4). As previously observed (Macdonald et al., 2003) cells harbouring the HCV subgenomic replicon showed a block to the appearance of phospho-Erk1/2 (lanes 5–8). Consistent with the AP-1–luciferase data there was no block to Erk1/2 phosphorylation in cells harbouring the GBV-B subgenomic replicon (lanes 9–12), blotting with a phosphorylation state-independent antibody showed similar levels of Erk1/2 in all cells, confirming that the reduction in Erk1/2 phosphorylation was not due to a reduction in the abundance of Erk1/2. Representative Western blots are shown in Fig. 4(a), in addition densitometry analysis was performed on multiple repeat experiments; the levels of phospho-Erk1/2 were quantified at the 0 and 0.5 h time points. These data are plotted in Fig. 4(b) and confirm that there is no block to EGF-stimulated Erk1/2 phosphorylation in cells harbouring the GBV-B subgenomic replicon. We conclude that the inability of GBV-B NS5A to inhibit the induction of AP-1 activity is due to its failure in blocking EGF-stimulated Erk1/2 phosphorylation and subsequent signalling.

(50K): Fig. 4. Ras–Erk signalling is not blocked in cells harbouring the GBV-B subgenomic replicon. (a) Naïve Huh-7, FK5.1 or GBV-B cells harbouring the subgenomic replicon were maintained in serum-free growth medium overnight. Cells were harvested prior to treatment (lanes 1, 5 and 9) or at 0.5, 2 and 4 h after a 2 min treatment with 20 ng EGF ml–1. Lysates were resolved by SDS-PAGE (12 %) and subjected to immunoblot analysis with antibodies specific to phospho-Erk1/2 (top panel), total Erk1/2 (middle panel) and neomycin phosphotransferase (NPT, lower panel). (b) Densitometry was performed on Western blots (n=3). The phospho-Erk1/2 signals at 0 and 0.5 h post-EGF treatment were normalized to the corresponding total Erk1/2 levels and are presented graphically relative to the signal in control Huh-7 cells at 0.5 h post-EGF treatment.

Akt phosphorylation is stimulated in Huh-7 cells expressing both GBV-B and HCV NS5A
HCV NS5A has been reported previously to bind to the regulatory (p85) subunit of PI3K, activating the enzyme and leading to increased phosphorylation and activity of the downstream serine/threonine kinase, Akt (Street et al., 2005). To determine if GBV-B NS5A retained this property we analysed the levels of Akt phosphorylation in naïve Huh-7 and cells harbouring either the GBV-B or HCV subgenomic replicons. Cells were serum-starved overnight and stimulated with either insulin or EGF (known activators of PI3K). Cell lysates were immunoblotted with a phospho-Akt-specific antibody. Phospho-Akt was undetectable in Huh-7 cells grown under reduced-serum conditions; however, low levels of phospho-Akt could be detected in both cells harbouring the GBV-B or HCV subgenomic replicon (Fig. 5a, compare lanes 1, 4 and 7). After EGF stimulation, phospho-Akt levels increased in control cells but were unaffected in the two cell lines harbouring the subgenomic replicons (lanes 2, 5 and 8), consistent with the block to EGFR signalling illustrated in Figs 3 and 4. However, after stimulation with insulin, phospho-Akt levels were elevated in all three cell lines, this increase in phospho-Akt was significantly higher in both cells harbouring the GBV-B and HCV subgenomic replicons compared with naïve Huh-7, consistent with the increased activity of PI3K in these cells. Immunoblotting with a phosphorylation state-independent Akt antibody showed that overall levels of Akt were similar in all three cell lines, either in reduced serum or following insulin or EGF stimulation. Representative Western blots are shown in Fig. 5(a), in addition densitometry analysis was performed on multiple-repeat experiments. These data are shown in Fig. 5(b) and confirm that both basal and insulin-stimulated phospho-Akt levels were higher in the cells harbouring the subgenomic replicon compared with naïve Huh-7 cells. These data demonstrate that, in common with HCV, the GBV-B subgenomic replicon retains the ability to stimulate the PI3K pathway, presumably as a result of the activity of NS5A.

(25K): Fig. 5. (a) PI3K and Akt are activated in cells expressing GBV-B NS5A. Naïve Huh-7 FK5.1 or GBV-B cells harbouring the subgenomic replicon were maintained in serum-free growth medium overnight. Cells were harvested prior to treatment (lanes 1, 4 and 7), or immediately after treatment with 20 ng EGF ml–1 (lanes 2, 5 and 8) or 10 µg insulin ml–1 (lanes 3, 6 and 9) for 5 min. Lysates were resolved by SDS-PAGE (12 %) and subjected to immunoblot analysis with an anti-phospho-Ser473 Akt antibody (top panel) or a phosphorylation state-independent antibody to detect total levels of Akt (middle panel). Equal protein loading was confirmed by immunoblotting for GAPDH (bottom panel). (b) Densitometry was performed on Western blots (n=3). The phospho-Akt signals in unstimulated or insulin-stimulated samples were normalized to the corresponding total Akt levels and are presented graphically. (c) NS5A increases transcriptional activation of β-catenin. All cells were transfected with a β-catenin expression vector together with the appropriate luciferase reporter M50 or M51. Plasmids expressing HA-AKT(K179M), GBV-B NS5A-FLAG or HCV NS5A were transfected as indicated. Luciferase reporter activity was assayed as in Fig. 3. **, Values significantly (P<0.05) different from controls as determined by one-way ANOVA and the Bonaferroni test.

GBV-B NS5A increases transcriptional activation of the proto-oncogene β-catenin via a PI3K-dependent pathway
We wished to confirm that the effects of the GBV-B subgenomic replicon on Akt were indeed the result of NS5A activity. To test this we reasoned that one approach would be to investigate the downstream consequences of Akt activation. In this regard one of the key substrates of Akt is glycogen synthase-3β (GSK-3β). Akt-mediated phosphorylation of GSK-3β results in its inactivation thus blocking phosphorylation and concomitant proteosomal targeting of the GSK-3β substrate β-catenin (Doble & Woodgett, 2003). In cells expressing HCV NS5A there is thus an increased level of β-catenin-responsive transcription, as measured using a reporter in which luciferase expression is driven by tandem copies of the β-catenin-responsive Tcf-binding site (Street et al., 2005). Unfortunately, it was not possible to conduct these experiments in Huh-7 cells as these cells have a high constitutive level of basal GSK-3β phosphorylation and concomitantly high β-catenin levels (Desbois-Mouthon et al., 2002). We therefore transfected Cos-7 cells with GBV-B- or HCV-NS5A expressing plasmids, a β-catenin expression vector and a reporter construct carrying eight copies of the wild-type Tcf-binding sites (M50). To verify that luciferase expression was indeed β-catenin dependent, a control plasmid (M51), which contained mutated Tcf-binding sites, was also used. As shown in Fig. 5(c), consistent with our previous results (Street et al., 2005) expression of HCV NS5A resulted in a 2.5-fold increase in β-catenin-responsive transcription. GBV-B NS5A was also able to effect a similar level of upregulation of β-catenin-responsive transcription. To confirm that this induction was mediated via the PI3K–Akt pathway we co-transfected a construct expressing a dominant-negative form of Akt (K179M). As expected this resulted in a modest reduction in the stimulatory effect of GBV-B NS5A. As a positive control, transfected cells were treated with the GSK-3β inhibitor, LiCl, resulting in a sixfold increase in β-catenin-responsive transcription. We conclude that, like HCV NS5A, GBV-B NS5A is able to activate the PI3K–Akt pathway, leading to concomitant stabilization of β-catenin. Whereas in at least 80 % of individuals infected with HCV the virus establishes a chronic infection that can last for decades, GBV-B infection of New World monkeys (believed to represent the natural host) generally results in an acute, self-limiting infection. This suggests that HCV has developed mechanisms to circumvent host-cell antiviral responses, which are not conserved in GBV-B and are likely to contribute to HCV persistence in vivo. A recent study demonstrated that both GBV-B and HCV NS3/4A shared the ability to cleave MAVS and disrupt RIG-I signalling (Chen et al., 2007), so this led us to evaluate whether the differences in pathogenesis of the two viruses might be explained by differences in the functions of their respective NS5A proteins. In this regard NS5A is implicated in the establishment of chronic HCV infection, due to its effects on IFN signalling, modulation of cell growth and the inhibition of apoptosis (for review see Macdonald & Harris, 2004). Since it has been shown previously that HCV NS5A inhibits the Ras–Erk pathway (Tan et al., 1999; Macdonald et al., 2003; Georgopoulou et al., 2003) and simultaneously activates PI3K (Street et al., 2005), the current study investigated if this ability was shared by the GBV-B NS5A protein. We demonstrated that the GBV-B NS5A protein differs both in its ability to inhibit Ras–Erk signalling (Figs 3 and 4) and its cellular distribution when compared with HCV NS5A (Figs 1 and 2). GBV-B NS5A does share with HCV NS5A the ability to stimulate phosphorylation of Akt, presumably through activation of PI3K, leading to a subsequent elevation of β-catenin transcriptional activation (Fig. 5).

Importantly, we demonstrated these effects in the context of both transient expression of GBV-B NS5A alone and in Huh-7 cells harbouring the GBV-B subgenomic replicon, thus expressing GBV-B NS5A as part of a membrane-bound replication complex. This is consistent with our previous observations that the effects of HCV NS5A on Ras–Erk or PI3K signalling are identical, whether the protein is expressed alone or in the context of an HCV subgenomic replicon or full-length polyprotein (Macdonald et al., 2003; Street et al., 2004, 2005). We are confident therefore that the data presented here can be extrapolated to GBV-B-infected cells. However, such experiments offer significant technical challenges, whereas the JFH-1 isolate of HCV can replicate in Huh-7 cells, GBV-B can only be grown in primary tamarin or marmoset hepatocytes (Buckwold et al., 2005).

As the PI3K–Akt pathway is the major anti-apoptotic effector in the cell this suggests that both viruses are able to prevent the induction of apoptosis, thus ensuring the survival of infected cells. Viral inhibition of apoptosis is a common theme, both in viruses that establish chronic infections and those that have more short-term replicative cycles, and in many cases this involves activation of the PI3K–Akt pathway. Indeed both dengue virus and Japanese encephalitis virus (members of the genus Flavivirus) have been shown to activate PI3K and block caspase-dependent apoptosis early in infection (Lee et al., 2005), thus activation of this pathway appears to be conserved throughout the family Flaviviridae. Clearly therefore the ability to activate PI3K and render infected cells resistant to apoptosis does not explain the differences between HCV and GBV-B.

The current study substantiates the potential importance of Ras–Erk signalling in HCV pathogenesis. Although the role of HCV NS5A modulation of Ras–Erk signalling in viral replication remains to be determined, HCV NS5A may modulate this host-cell mitogenic signalling pathway to enhance viral replication (Huang et al., 2006), regulating cell growth and activation and enabling the establishment of a chronic infection. One physiological consequence of this dysregulation might be perturbation of host cell-cycle control. Indeed, the activation of the Ras–Erk pathway by EGF is required for primary hepatocytes to progress through a restriction point in the late G₁ phase (Talarmin et al., 1999). Furthermore, it has been demonstrated recently that liver biopsies from HCV-positive individuals are enriched for cells in the G₁ phase of the cell cycle in comparison with other chronic liver conditions (Marshall et al., 2005), consistent with the effect of HCV NS5A on the Ras–Erk pathway. The observation that GBV-B NS5A is unable to block Ras–Erk signalling suggests that perturbation of this pathway may be important for the establishment of chronic infection.

We (Macdonald et al., 2004, 2005b), and others (Tan et al., 1999), have previously shown that HCV NS5A interacts with the SH3 domains of the adaptor protein Grb2 and members of the Src family of tyrosine kinases via a C-terminal polyproline motif (PxxPxR) located between domains 2 and 3. This motif was required for inhibition of Ras–Erk signalling and the observation that GBV-B NS5A lacks this motif at this position within the protein provides a mechanistic explanation for the inability to modulate Ras–Erk signalling. In contrast, the amino acid sequences within HCV NS5A required for the interaction with the SH3 domain of the p85 regulatory subunit of PI3K have not yet been defined (Street et al., 2004) and comparison of the amino acid sequences of the two proteins might help to shed light on this question. However, the two proteins share only 25 % identity and there are no obvious conserved motifs; it is thus possible that they interact with PI3K by distinct mechanisms.

Both HCV and GBV-B NS5A possess an N-terminal amphipathic helix that mediates membrane association (Brass et al., 2007), and conserved cysteines that co-ordinate a zinc ion (Tellinghuisen et al., 2004). In the case of HCV both these elements are essential for HCV RNA replication; thus it is likely that, unlike the situation regarding perturbation of host-cell signalling, the RNA replication functions of the two NS5A proteins are conserved. Consistent with this, immunofluorescence analysis showed a partial co-localization of both GBV-B and HCV NS5A. However, HCV NS5A appeared to show a more extensive distribution than the GBV-B protein, our data (Fig. 2 and J. Mankouri, S. Griffin & M. Harris, unpublished) and that of others (Tang et al., 2007) suggest that HCV NS5A can co-localize with markers of early endosomes such as Rab5 and EEA1. It is tempting to speculate that the ability of HCV NS5A to localize to such compartments may be related to its ability to inhibit Ras–Erk signalling, particularly in the light of studies showing that EGFR signalling events require receptor endocytosis into endosomal compartments (Stasyk et al., 2007; Vieira et al., 1996).

In conclusion, the current study demonstrates that GBV-B NS5A has evolved a strategy similar to that employed by HCV, resulting in the activation of PI3K signalling with potential anti-apoptotic consequences. There are, however, important differences in the characteristics of HCV and GBV-B NS5A proteins, both in their subcellular distribution and modulation of Ras–Erk signalling, which may contribute to the inability of GBV-B to establish chronic infections. Further studies to ascertain the differences in HCV and GBV-B NS5A function may in the future provide new avenues for therapeutic interventions in HCV infection.

We are grateful to Jens Bukh (NIH), Nigel Bunnett (UCSF), Bo Van Deurs (University of Copenhagen, Copenhagen, Denmark), Cinzia Traboni (IRBM, Rome, Italy), Randall Moon (University of Washington, Seattle, USA) and Ralf Bartenschlager (University of Heidelberg, Heidelberg, Germany) for the kind gifts of reagents. We thank Andrew Macdonald for critical reading of the manuscript. This work was supported by grants to M. H. from the Medical Research Council (grant no. G0401577) and Yorkshire Cancer Research (grant no. L321), and to A. M. from Agence Nationale de Recherches sur le SIDA et les Hépatites Virales.