Role of Conserved E2 Residue W420 in Receptor Binding and Hepatitis C Virus Infection

ABSTRACT

Hepatitis C virus (HCV) enters cells via interactions with several host factors, a key one being that between the viral E2 envelope glycoprotein and the CD81 receptor. We previously identified E2 tryptophan residue 420 (W420) as an essential CD81-binding residue. However, the importance of W420 in the context of the native virion is unknown, as those previous studies predate the infectious HCV cell culture (cell culture-derived HCV [HCVcc]) system. Here, we introduced four separate mutations (F, Y, A, or R) at position 420 within the infectious HCVcc JFH-1 genome and characterized their effects on the viral life cycle. While all mutations reduced E2-CD81 binding, only two (W420A and W420R) reduced HCVcc infectivity. Further analyses of mutants with hydrophobic residues (F or Y) found that interactions with the receptors SR-BI and CD81 were modulated, which in turn determined the viral uptake route. Both mutant viruses were significantly less dependent on SR-BI, and its lipid transfer activity, for virus entry. Furthermore, these viruses were resistant to the drug erlotinib, which targets epidermal growth factor receptor (EGFR) (a host cofactor for HCV entry) and also blocks SR-BI-dependent high-density lipoprotein (HDL)-mediated enhancement of virus entry. Together, our data indicate a model where an alteration at position 420 causes a subtle change in the E2 conformation that prevents interaction with SR-BI and increases accessibility to the CD81-binding site, in turn favoring a particular internalization route. These results further show that a hydrophobic residue with a strong preference for tryptophan at position 420 is important, both functionally and structurally, to provide an additional hydrophobic anchor to stabilize the E2-CD81 interaction.

IMPORTANCE Hepatitis C virus (HCV) is a leading cause of liver disease, causing up to 500,000 deaths annually. The first step in the viral life cycle is the entry process. This study investigates the role of a highly conserved residue, tryptophan residue 420, of the viral glycoprotein E2 in this process. We analyzed the effect of changing this residue in the virus and confirmed that this region is important for binding to the CD81 receptor. Furthermore, alteration of this residue modulated interactions with the SR-BI receptor, and changes to these key interactions were found to affect the virus internalization route involving the host cofactor EGFR. Our results also show that the nature of the amino acid at this position is important functionally and structurally to provide an anchor point to stabilize the E2-CD81 interaction.

INTRODUCTION

Hepatitis C virus (HCV) is a positive-strand RNA virus belonging to the Hepacivirus genus within the Flaviviridae family (1). The viral genome comprises a single open reading frame (ORF), encoding structural and nonstructural (NS) proteins, flanked by two untranslated regions (UTRs) at the 5′ and 3′ ends. The large polyprotein of ∼3,000 amino acids (aa) is cleaved by cellular and viral proteases into 10 different proteins: core, E1, E2, p7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B (2). The structural proteins include core, which forms the viral nucleocapsid, and the envelope glycoproteins E1 and E2, which mediate early cell entry events (3). NS2 and p7 (a viroporin) play crucial roles in virus assembly/egress (4–6), and the remaining nonstructural proteins, NS3, NS4A, NS4B, NS5A, and NS5B, form replication complexes, which synthesize both plus- and minus-strand viral RNAs (7). HCV is classified into seven major genetic groups and further subdivided into numerous subtypes (1, 8). This genetic variability is caused by the error-prone nature of the RNA-dependent RNA polymerase (NS5B), amplified by the high rate of virus production (9) and further accelerated by the selective pressure exerted by the host immune response (10).

The viral particle consists of a nucleocapsid encasing the viral RNA, surrounded by a lipidic cell-derived envelope in which the glycoproteins E1 and E2 are embedded. Numerous reports have shown that both serum-derived HCV and cell culture-derived HCV (HCVcc) are tightly associated with low-density lipoproteins (LDLs) and very-low-density lipoproteins (VLDLs) to form a hybrid particle called a lipoviroparticle (LVP) (11, 12). In vivo, these associations are believed to protect HCV from the humoral immune response by shielding the glycoproteins from circulating neutralizing antibodies (NAbs) (3). HCVcc studies have also shown that the major VLDL component, ApoE, functions in the viral entry process (13). HCV entry into hepatocytes is a multistep process involving a series of interactions between the virus particles and several cellular molecules. Initial attachment of HCV to the cell surface most likely occurs via interactions of virion-associated ApoE with LDL receptor (LDLR) and with glycosaminoglycans (GAGs) present on heparan sulfate proteoglycans (HSPGs) (14–18). To gain entry into the cell, HCV depends on several cellular molecules: scavenger receptor BI (SR-BI) (19), tetraspanin CD81 (20), the tight junction proteins claudin-1 (21) and occludin (22), epidermal growth factor (EGF) receptor (EGFR) (23) and its signal transducer Harvey rat sarcoma viral oncogene homolog (HRas) (24), Niemann-Pick C1-like cholesterol receptor (NPC1L1) (25), and transferrin receptor 1 (26). The interaction of HCV particles with the cell leads to the internalization of particles through clathrin-mediated endocytosis (27, 28) and their subsequent fusion at low pH with the membranes of early endosomes (29). Only two of these host cell molecules, CD81 and SR-BI, have been reported to interact directly with the HCV envelope glycoproteins (19, 20). SR-BI mediates the binding of E2 through an interaction that involves hypervariable region 1 (HVR1), a 27-amino-acid segment located at the N terminus of the HCV E2 glycoprotein. It is believed that HVR1 masks or induces the masking of the E2-CD81-binding site and that the E2–SR-BI interaction facilitates conformational changes within the glycoproteins that cause exposure of the CD81-binding site (30). CD81 has been demonstrated to have a major role in HCV entry and is the best characterized of the cellular entry factors to date. Numerous studies have identified E2 regions and residues that are potentially involved in CD81 interactions based on the characterization of neutralizing antibodies, mutagenesis studies, and structural data (31). However, most of those studies used soluble E2 or HCV pseudoparticle (HCVpp)-derived E2, in which the conformation of E2 is slightly different from those of the envelope glycoproteins in native HCV particles (32). Therefore, it remains unclear which E2 residues or sequences are genuinely involved in the interaction of virus particles with CD81. Prior to the availability of the HCVcc system, we identified several residues required for E2-CD81 binding by alanine mutagenesis (33). One of these residues, W420, was located within highly conserved E2 epitope I, comprising residues 412 to 423. We also showed that mouse monoclonal antibody (MAb) AP33, which inhibits the E2-CD81 interaction, bound to this epitope and specifically recognized residues L413, N415, G418, and W420 (33, 34). Analysis of amino acid variation within the AP33 epitope shows that W420 is 99.9% conserved (35), which suggests that it is functionally or structurally important. Epitope mapping has shown that it is a contact residue for several broadly neutralizing antibodies, including AP33 (reviewed in reference 36). More recently, this has been confirmed by structural analysis, which reveals how extensively this residue is bound by the neutralizing antibodies that recognize this epitope (35, 37–40).

To further investigate the role of W420 in CD81 binding and virus infection, we replaced this residue with phenylalanine, tyrosine, alanine, or arginine in the genotype 2a HCVcc JFH-1 background. We then characterized the mutant viruses by testing their viral replication levels, cellular receptor interactions, and sensitivity to neutralizing antibodies. We confirmed that W420 is important for CD81 binding during virus entry and interestingly also modulates virion interactions with the receptor molecules SR-BI and EGFR. Together, our results suggest that the tryptophan, as a large hydrophobic residue, functions in conjunction with other CD81-binding regions to provide an additional anchor point to stabilize the E2-CD81 interaction.

MATERIALS AND METHODS

Cells.

Human epithelial kidney (HEK-293T) (ATCC CRL-1573) and human hepatoma Huh-7 (41) and Huh7-J20 (42) cells were propagated in Dulbecco's modified essential medium supplemented with penicillin-streptomycin, nonessential amino acids, and 10% fetal calf serum (DMEM). CHO-K1 cells were propagated in Ham F-12 medium (Life Technologies) supplemented as described above. The stable cell lines CHO-hSR-BI and CHO-hSR-BI-GFP were generated by cloning sequences encoding the human SR-BI or SR-BI–enhanced green fluorescent protein (EGFP) fusion protein into the retrovirus transfer vector pQCXIP (BD Biosciences). These plasmids were cotransfected with constructs expressing murine leukemia virus (MLV) Gag-Pol and the vesicular stomatitis virus (VSV) G glycoprotein into HEK-293T cells to generate VSV-G pseudoparticles (VSVpp). CHO-K1 cells were transduced with VSVpp carrying the gene encoding human SR-BI or SR-BI–EGFP, and transduced cells were selected in medium containing 4 μg/ml puromycin.

Antibodies.

The anti-E2 rodent MAbs AP33 and 3/11 and the human MAb (HMAb) CBH-5 were described previously (43–46). The anti-E2 MAbs CBH-5 and 3/11 and the anti-NS5A MAb 9E10 (47) were kindly provided by S. Foung, J. McKeating, and C. M. Rice, respectively. The MLV Gag-specific MAb was obtained from rat hybridoma cells (ATCC CRL-1912). The anti-core MAb C7-50 and the anti-CD81 MAb JS-81 were obtained from Bioreagents and BD Biosciences, respectively. The anti-Flag M2 MAb was obtained from Sigma-Aldrich. A derivative of the anti-SR-BI human MAb151 described previously (48) was generated in CHO-K1 cells. Briefly, the variable heavy chain (V_H)- and variable light chain (V_L)-encoding sequences of MAb151 were cloned into mouse IgG1 expression vectors pFUSEss-CHIg-mg1 and pFUSE2ss-CLIg-mk (InvivoGen, CA, USA), respectively. Following cotransfection of these plasmids into CHO-K1 cells, a clone stably secreting human-mouse chimeric IgG (called MAb151-NP1) was selected and expanded. MAb151-NP1 secreted into the medium was purified by using protein G-Sepharose affinity chromatography and confirmed to react specifically with human SR-BI. Cell surface expression of SR-BI was measured by incubating cells with anti-SR-BI MAb151-NP1 or an isotype IgG1 control, followed by an anti-mouse phycoerythrin (PE)-conjugated secondary antibody. The cells were analyzed by flow cytometry on a FACSCalibur instrument with CellQuest Pro software (BD biosciences).

Plasmid constructs and mutagenesis.

Plasmid pUC-JFH-1 carries the full-length cDNA of genotype 2a HCV strain JFH-1. Plasmid pUC-GND JFH-1 is identical except for a GND mutation in the viral NS5B RNA polymerase (49). The plasmids used to generate HCVpp containing strain JFH-1 envelope glycoproteins were described previously (50). Site-directed mutagenesis was carried out by using a QuikChange-XL-II kit (Agilent Technologies), according to the manufacturer's instructions, to introduce amino acid substitutions at the target sites in E2. Briefly, the amino acid substitutions W420F, W420Y, W420A, W420R, and W420V in the E2-coding region were individually introduced into plasmid pUC-JFH-1 by using appropriate primers (the sequences of which are available upon request). The presence of the desired mutations in the resulting clones was confirmed by sequencing of the DNA fragments spanning the mutation site, and these fragments were then subcloned back into pUC-JFH-1 and the HCVpp E1E2 expression vector.

Determination of virus infectivity and RNA replication.

Infectious viruses were generated by electroporation of viral RNA into Huh7 cells as previously described (49). Infectious virus titers in the cell medium were determined by infecting Huh7 cells with a serially diluted inoculum followed by immunostaining for the NS5A viral protein in a focus-forming unit (FFU) assay as described previously (51). The levels of virus infectivity and intracellular RNA replication were determined by infecting the reporter cell line Huh7-J20 and measuring secreted alkaline phosphatase (SEAP) activity in the culture medium at the indicated times postinfection, as described previously (42). To monitor virus infectivity during serial passaging, Huh7 cells electroporated with the viral RNA or inoculated with infectious virus were passaged in T80 flasks containing DMEM. At each passage, the cell culture supernatants were harvested, and the released virus infectivity was determined by an FFU assay. To determine the replication of each mutant virus, 2 × 10⁶ Huh7-J20 cells were electroporated with 10 μg of viral RNA and resuspended in 4 ml of DMEM. Aliquots of 0.5 ml were then seeded into triplicate wells of a 24-well plate. Following incubation at 37°C for 72 h, the virus infectivity/replication levels and infectious virus yields in cell culture supernatants were determined by SEAP and FFU assays, respectively. Cell-associated virus was obtained essentially as described previously (52). Briefly, 3 × 10⁶ Huh7 cells were electroporated with 10 μg of viral RNA, resuspended in 15 ml of DMEM, and seeded into 90-mm culture dishes. Cells were harvested at 72 h postelectroporation, washed in DMEM, resuspended in 0.8 ml DMEM, and freeze-thawed three times. The samples were centrifuged to remove cell debris, and the supernatant was assayed by an FFU assay to determine virus infectivity.

Western blot analysis.

Western blot analysis was performed as described previously (43), with some modifications. To detect intracellular antigens, cultured cells were washed once in phosphate-buffered saline (PBS) and lysed directly in SDS-PAGE sample loading buffer (200 mM Tris-HCl [pH 6.7], 0.5% SDS, β-mercaptoethanol, 10% glycerol). Lysates were homogenized by passage through a 22-gauge needle five times before use. To obtain extracellular virus, 10 ml of culture medium from electroporated cells was harvested, filtered, overlaid onto 1 ml of a 20% (wt/vol) sucrose cushion made with PBS, and centrifuged at 25,000 rpm for 4 h in a Sorvall Discovery 90SE ultracentrifuge. The pellets were then lysed directly in 50 μl of SDS-PAGE sample loading buffer and stored at −20°C until use. The proteins in 20 μl of the sample were resolved by 12.5% SDS-PAGE and transferred onto nitrocellulose membranes (Hybond-ECL; Amersham).

Identification of reversion mutations.

Total RNA was prepared from infected cells by using the RNeasy kit (Qiagen), and the HCV RNA was converted to first-strand DNA by using a Superscript III first-strand synthesis kit (Invitrogen) with the primer 5′-TTGCGAGTGCCCCGGGA-3′. After digestion with 1 U of RNase H (Invitrogen) for 20 min at 37°C, one-quarter of the first-strand DNA mixture was amplified with the appropriate primers to yield four fragments of HCV cDNA (nucleotides [nt] 322 to 930, 538 to 3038, 2544 to 5542, and 5412 to 7890) covering the core region to the NS5A region of the viral genome. The PCR products were gel purified and used directly for nucleotide sequencing.

HCVpp genesis, infection, and analysis by immunoblotting.

HCVpp were generated by transfection of HEK-293T cells with plasmids expressing HCV E1E2, MLV Gag-Pol, and the MLV transfer vector expressing a firefly luciferase reporter. The medium containing HCVpp was collected, filtered through a 0.45-μm-pore-size membrane, and used to infect Huh7 cells as described previously (53). Three days after infection, the cells were lysed, and their luciferase activity was measured by using the Bright-Glo luciferase assay system (Promega). For protein analysis, HCVpp-containing medium was pelleted through a 20% (wt/vol) sucrose cushion in PBS at 100,000 × g for 2 h. The pellets were resuspended in SDS-PAGE sample loading buffer and analyzed by SDS-PAGE followed by immunoblotting for HCV E2 and MLV Gag.

GNA capture and CD81-binding assays.

An enzyme-linked immunosorbent assay (ELISA) to detect MAb binding to the E2 glycoprotein was performed essentially as described previously (54). Briefly, HEK-293T cells were cotransfected with E1E2 expression plasmids, and the expressed glycoproteins present in clarified lysates of these cells were captured onto GNA (Galanthus nivalis agglutinin)-coated Immulon II enzyme immunoassay (EIA) plates (Thermolabsystems). Captured glycoproteins were detected by using anti-E2 MAbs, followed by secondary antispecies antibodies conjugated to horseradish peroxidase (HRP) (Sigma-Aldrich) and the TMB (3,3′,5,5′-tetramethylbenzidine; Sigma-Aldrich) substrate. Absorbance values were determined at 450 nm. The E2-CD81-binding assay was performed essentially as described above. Briefly, E1E2 from cell lysates was captured on an ELISA plate coated with GNA, the wells were washed, and insect cell-expressed FLAG-tagged large extracellular loop (LEL) of human CD81 (soluble CD81-LEL [sCD81-LEL]) was added. Bound CD81 was detected by using an anti-FLAG antibody, followed by anti-mouse horseradish peroxidase, as described above.

HCVcc neutralization assays.

Inhibition assays were performed with Huh7-J20 cells, and virus infectivity levels were determined by an SEAP reporter assay, as described previously (42). Briefly, cells were seeded at a density of 4 × 10³ cells per well into a 96-well plate and incubated at 37°C overnight prior to infection. For anti-E2 antibody neutralization assays, virus was preincubated at 37°C for 1 h with the appropriate antibody prior to infection of cells at a multiplicity of infection (MOI) of 0.1. To test neutralization by sCD81-LEL, virus was preincubated at 37°C for 1 h with purified His-tagged sCD81-LEL expressed in Escherichia coli as described previously (55), prior to infection of cells at an MOI of 0.1. To test neutralization by antireceptor antibodies, cells were preincubated with appropriate antibodies for 1 h at 37°C prior to infection at an MOI of 0.1. At 3 h postinfection, the inoculum was replaced with fresh DMEM and incubated for 48 h.

HCVcc dose-response assays.

Blocker of lipid transport 4 (BLT-4), erlotinib, and sunitinib were obtained from Sigma-Aldrich. The assay was performed similarly to the neutralization assay described above except that cells were pretreated with the inhibitor for 1 h at 37°C prior to infection at an MOI of 0.1 for 3 h, the inoculum and inhibitor were removed, cells were washed, and fresh medium was added. After 72 h of incubation, the SEAP activity in the medium was measured. Cell viability assays were performed with Huh7-J20 cells seeded at 4 × 10³ cells per well in a 96-well plate and incubated at 37°C overnight prior to treatment. Cells were treated with the inhibitor for 4 h at 37°C and washed, and fresh medium was added. Following 72 h of incubation, medium was removed, cells were incubated in 10% cell proliferation reagent WST-1 reagent (Roche), and the absorbance at 440 nm was measured.

RESULTS

Effects of E2 mutations on virus replication.

To assess the role of E2 residue W420 in virus infection, we generated HCVcc mutants containing the amino acid substitutions W420A, W420R, W420F, and W420Y. Alanine, with its small, nonpolar side chain, is most commonly used for site-directed mutagenesis, but it is more informative to introduce residues that cover a wider range of physicochemical properties. In this case, replacement of tryptophan with the aromatic residue phenylalanine or tyrosine is a relatively conservative change. Substitution by arginine, a positively charged polar residue, is a drastic change, but we decided to include it because this mutation is found in 1 of 2,161 naturally occurring HCV E2 sequences (35). To study the effect of these mutations on virus replication, Huh7-J20 cells were electroporated with RNA of wild-type JFH-1 (JFH-1_WT), the JFH-1 mutant containing the W420F substitution (JFH-1_W420F), JFH-1_W420Y, JFH-1_W420A, or JFH-1_W420R, and after 72 h, the supernatant was harvested for FFU and SEAP assays. All of the mutants showed SEAP activities similar to that of JFH-1_WT (Fig. 1a). Cells electroporated with the replication-deficient JFH-1_GND RNA and replication-competent but assembly-deficient JFH-1_ΔE1E2 RNA served as controls in this experiment. These results indicate that the E2 mutations do not alter intracellular HCV RNA replication. Similar levels of the core, E2, and NS5A viral proteins were detected in all cell lysates, confirming that these mutations do not affect genome or protein synthesis (Fig. 1c). In contrast, major differences in the titers of infectious virus released into the cell medium were observed (Fig. 1b). The JFH-1_W420F and JFH-1_W420Y viruses showed peak titers comparable to those of JFH-1_WT, whereas the infectivities of the JFH-1_W420A and JFH-1_W420R viruses were reduced by ∼1 log₁₀ and ∼4 log₁₀ units, respectively. A similar infectivity profile was obtained by using intracellular virus (recovered from cells lysed by freeze-thawing), indicating that the defect in the infectivity of JFH-1_W420A and JFH-1_W420R is not due to reduced virion secretion (Fig. 1b). To determine whether the W420 mutants were producing noninfectious viral particles, culture medium harvested at 72 h postelectroporation was concentrated and probed for the presence of the core protein by immunoblotting and also tested for infectivity. We found that all the mutants secreted levels of extracellular core protein similar to those secreted by JFH-1_WT, whereas the infectivity of JFH-1_W420A and JFH-1_W420R was very much reduced (Fig. 1c, pelleted), which means that these mutations alter the specific infectivity of particles, with little or no effect on virus assembly and secretion.

FIG 1.

Analysis of W420 mutant viruses. Huh7 cells were transfected with viral RNA transcribed from cDNA encoding JFH-1_WT, JFH-1_GND, JFH-1_ΔE1E2, and a panel of JFH-1_W420 mutants and analyzed after 72 h. (a) Intracellular viral replication was quantified by SEAP activity (in relative light units). (b) Viral titers were quantified by an FFU assay. Black bars show extracellular virus harvested from the medium, and gray bars show intracellular virus harvested from lysed cells. (c) Western blot analysis to detect (i) the viral proteins core, E2, and NS5A and tubulin as a loading control in mock-infected (MI) and infected cell lysates and (ii) core protein in released virus pelleted from the infected cell supernatant. The blots shown are representative. (d) Virus released from cells transfected with viral RNA transcribed from cDNAs for JFH-1_WT, JFH-1_W420A, and JFH-1_W420V was titrated by an FFU assay. Panels a, b, and d show average values from duplicate independent experiments, and error bars show standard errors of the means.

Reversion of E2 mutants during prolonged culture.

To determine whether the infectivity of the JFH-1_W420A and JFH-1_W420R mutants could be rescued by compensatory mutations, cells electroporated with JFH-1_W420R RNA or infected with the JFH-1_W420A virus were serially passaged. The level of infectious virus released into the culture medium was monitored throughout each passaging experiment. A progressive increase in the level of extracellular virus release was observed in both experiments, which eventually achieved peak titers similar to those expected for JFH-1_WT (data not shown). To identify the mutation(s) responsible for this increased infectivity, total RNAs were prepared from cells infected with the virus collected from the final passage, and the core- to NS5A-encoding regions of the HCV genome were sequenced by RT-PCR. Interestingly, sequencing revealed that the JFH-1_W420R virus had reverted back to the wild-type tryptophan residue by a single nucleotide change (CGG to TGG), which emphasizes the functional importance of W420. The passaged JFH-1_W420A virus also contained a single nucleotide change (GCG to GTG), thereby converting the original alanine substitution to a valine residue. To determine if this valine substitution was indeed responsible for the improved infectivity of JFH-1_W420A seen during passaging, we engineered this change into the original JFH-1_WT genome and analyzed infectious virus production at 72 h postelectroporation. As shown in Fig. 1d, the JFH-1_W420V mutant displayed cell-free infectivity comparable to that of JFH-1_WT, suggesting that the JFH-1_A420V change functions as a reversion mutation.

Infectivity profiling with the HCVpp system.

We previously showed that a W420A mutation in the HCV genotype 1a strain H77 E2 protein abolished HCVpp infection (33). However, the results presented here show that the same mutation in the strain JFH-1 HCVcc system reduces infection only 10-fold. To resolve this discrepancy, we assessed the infectivities of the JFH-1 E2 W420 mutants with the HCVpp system. In contrast to the results for HCVcc, we found that W420Y retained 5% of the infectivity of the wild type, and the rest of the W420 mutants were noninfectious in the HCVpp system (Fig. 2a). The HCVpp infectivity data do not correspond to the E2-CD81-binding data (see below and Fig. 5a), as the W420F mutant retained CD81-binding activity but was not infectious. Thus, the reasons for the lower infectivity of mutant HCVpp are not clear. Using an E2 GNA capture ELISA, we confirmed that the wild-type and mutant E2 glycoproteins were expressed intracellularly in comparable quantities (data not shown). However, we consistently found less incorporation of the E2 W420A, W420V, and W420R mutants into secreted HCVpp (Fig. 2b). The level of the E2 W420R mutant was extremely low, while the levels of the E2 W420A and W420V mutants were clearly reduced relative to that of the wild type. In contrast, the level of incorporation of E2 W420F and W420Y mutants into HCVpp was higher than that of the wild-type protein, even though the HCVpp displayed reduced or no infectivity in Huh7 cells (Fig. 2). The lack of infectivity of the W420A, -V, and -R HCVpp mutants is likely explained by reduced E2 incorporation into pseudoparticles.

FIG 2.

Analysis of W420 mutants in the pseudoparticle system. Pseudoparticles were harvested from HEK-293T cells transfected with the wild type (WT), position 420 mutant E1E2, or no envelope (NE). (a) HCVpp infectivity in Huh7 cells expressed as a percentage relative to wild-type infectivity. The results shown are the averages of data from 3 independent experiments, and error bars show standard errors of the means. (b) Representative Western blot of pelleted wild-type and mutant HCVpp probed with anti-HCV E2 and anti-MLV Gag.

FIG 5.

Virus-CD81 receptor interactions. (a) Reactivity of sCD81-LEL in a modified GNA ELISA with the same panel of wild-type and mutant E1E2-containing lysates as those used for Fig. 3. (b and c) Neutralization of JFH-1_WT, JFH-1_W420F, and JFH-1_W420Y viruses by anti-CD81 (b) and sCD81-LEL (c). Each data set shows the average of results from 2 (b and c) or 3 (a) independent experiments, and error bars show standard errors of the means.

E2 mutations alter sensitivity to neutralizing antibodies.

We assessed the reactivity of three broadly neutralizing MAbs, AP33, 3/11, and CBH-5, to each E2 mutant by a GNA capture ELISA. Both AP33 and 3/11 bind to distinct but overlapping epitopes within the highly conserved region of E2 spanning residues 412 to 423 (QLINTNGSWHIN), with W420 being a critical contact residue for both antibodies (34). Recent structural data, however, revealed that this region is flexible. AP33 binds to a β-hairpin structure; in contrast, 3/11 recognizes an open conformation of this region (37–39). HMAb CBH-5 binds to an epitope within the CD81-binding region: replacement of E2 residues G523, P525, G530, D535, and N540 with alanine was reported to ablate CBH-5 binding, whereas a mutation at W420 did not reduce binding by >50% (56). Further studies have shown that this antibody maps to immunodomain B and directly competes with CD81 for binding to E2 (44, 57, 58). A panel of HEK-293T cell lysates containing wild-type or mutant HCV E1E2 was first tested for reactivity to an anti-E2 MAb, DAO5, that recognizes a linear epitope spanning aa 532 to 540 (I. Vasiliauskaite, A. Owsianka, P. England, W. Witkowskac, S. K. H. Foung, R. E. Swann, F. A. Rey, A. H. Patel, and T. Krey, unpublished data). As expected, all lysates had similar reactivities to this MAb, indicating that the proteins were expressed in equivalent quantities (Fig. 3a). All W420 mutants showed undetectable levels of binding to MAb AP33 in the ELISA, confirming that this is a critical residue for the MAb AP33-E2 interaction (Fig. 3b). In contrast, while the W420A, -V, and -R mutants had no detectable binding to MAb 3/11, both aromatic substitution mutants W420F and W420Y retained 50% and 75% of the binding activities, respectively (Fig. 3c). The same panel of lysates was assessed for reactivity with HMAb CBH-5, which binds a conformational epitope within antigenic domain B of E2 (47). Although all W420 mutants retained some level of binding activity, this correlated with the type of residue substituted. Both aromatic mutants (W420F and W420Y) retained 55% of the binding activity of the wild type, and both aliphatic residues (W420A and W420V) had similar binding, with 36% and 41% of the binding activities of the wild type, respectively, while E2 containing the positively charged W420R substitution bound HMAb CBH-5 with only 20% of the wild-type activity. Subsequently, we investigated if these antibodies could neutralize W420 mutant viruses, using the two aromatic mutants, which had levels of infectivity similar to those of wild-type JFH-1. We found that, as predicted by the ELISA E2-binding assay results, JFH-1_W420Y was completely resistant to neutralization by MAb AP33. Surprisingly, JFH-1_W420F showed a low but consistent level of inhibition at the highest concentrations of antibody AP33 tested (Fig. 4a), suggesting that this mutant can still bind MAb AP33 albeit at a level undetectable by an ELISA. In contrast, we found that both mutants were neutralized by the 3/11 antibody that binds to E2_412–423 in the open conformation. Although JFH-1_W420F binds MAb 3/11 at only ∼50% of wild-type levels, the inhibition profile did not show a corresponding change. Instead, the mutant virus was inhibited similarly to wild-type JFH-1 despite the reduction in binding affinity. Indeed, the JFH-1_W420Y mutant that retained 75% binding activity was found to be more sensitive to neutralization by MAb 3/11 than the wild-type virus (Fig. 4b). We also tested neutralization using the CBH-5 antibody, which binds to an epitope within the CD81-binding region of E2 (Fig. 4c). Notably, both mutants were significantly more sensitive to inhibition with this antibody, with ∼2,000-fold less antibody being required to inhibit the mutant viruses by 50% than that required to inhibit the wild-type virus. The differences observed in the sensitivities of these mutant viruses to neutralization by these human and rodent antibodies versus their glycoprotein reactivities by ELISA indicate that the mutations may enhance the exposure of E2 on the virion.

FIG 3.

Binding of E2 antibodies. Shown are data for the reactivity of anti-E2 linear (a to c) and conformational (d) antibodies with lysates from cells expressing wild-type and mutant E1E2 proteins in a GNA ELISA. Bound antibodies were detected by using secondary antispecies antibodies conjugated to HRP. Background levels were removed by subtracting the binding of a lysate lacking E1E2. Each data set shows the average of results from 3 independent experiments, and error bars show standard errors of the means.

FIG 4.

Neutralization by E2 antibodies. The JFH-1_WT, JFH-1_W420F, and JFH-1_W420Y viruses were neutralized by MAb AP33 and MAb 3/11, which bind to E2_412–423 (a and b), and by HMAb CBH-5, which binds immunodomain B (c). Each data set shows the average of results from 2 independent experiments, and error bars show standard errors of the means.

E2 mutations alter virus-receptor interactions.

We used the panel of W420 E1E2 lysates to examine the E2-CD81-binding reactivity of the intracellular viral glycoproteins to sCD81-LEL by an ELISA. Remarkably, the majority of the residue 420 mutants had no detectable binding to CD81, confirming the importance of this residue in the E2-CD81 interaction (Fig. 5a). The JFH-1_W420F mutant, however, could still bind to sCD81-LEL, although the level of binding was ∼40% of the wild-type levels. To determine if the mutations also reduced the affinity of E2 on the virion for the virus receptor CD81, we monitored neutralization of the JFH-1_WT, JFH-1_W420F, and JFH-1_W420Y viruses. First, we used an anti-CD81 neutralizing antibody that binds to the CD81 receptor on the cell surface and found that all viruses were similarly inhibited irrespective of their ability to bind sCD81-LEL in the ELISA (Fig. 5b). Second, a competition assay using sCD81-LEL was performed. In this experiment, we found that both W420 mutants were more sensitive to inhibition than JFH-1_WT, with JFH-1_W420F being the most sensitive (Fig. 5c). Even though no E2-CD81 interaction was detected in the ELISA for JFH-1_W420Y, the inhibition profile with sCD81-LEL indicates that this mutant retains some affinity for CD81. This parallels the above-described observations with MAb AP33 (Fig. 3b and 4a). Notably, the sensitivity of the mutant viruses to inhibition by sCD81-LEL (Fig. 5c) would not be predicted by their CD81-binding activity (Fig. 5a), although the W420F mutant, which retained 40% binding to sCD81-LEL, was the most sensitive to inhibition. This is in line with their greatly increased sensitivity to neutralization by HMAb CBH-5 (Fig. 4c) despite weaker binding to CBH-5 in the ELISA (Fig. 3d). Together, these data suggest increased exposure of CD81-binding sites on the residue 420 mutant virions.

Having established that these mutations influence the HCV-CD81 interaction, we investigated their effects on SR-BI-dependent entry. SR-BI has been reported to have three distinct functions in HCV entry: modulation of primary attachment via interactions with apolipoproteins such as ApoE on the HCV virion; an access function that depends on the lipid transfer activity of SR-BI; and, finally, an infectivity enhancement function (30). Only the latter function is thought to require the E2–SR-BI interaction. To investigate whether a mutation at position 420 alters the interaction of the virion with the SR-BI receptor, we monitored SR-BI neutralization of the JFH-1_WT, JFH-1_W420F, and JFH-1_W420Y viruses. The human-mouse anti-SR-BI MAb151-NP1 was expressed and purified, and its specificity for human SR-BI was then confirmed. The ability of anti-SR-BI MAb151-NP1 to bind to CHO-K1 cells expressing human SR-BI or human SR-BI–EGFP was assessed by fluorescence-activated cell sorter (FACS) analysis. First, the detection of GFP in CHO-hSR-BI-GFP cells confirmed the expression of the SR-BI–EGFP fusion protein (Fig. 6a). As expected, anti-SR-BI MAb151-NP1 bound to CHO-K1 cells expressing human SR-BI or human SR-BI–EGFP but not to the parental CHO-K1 cell line, confirming that the antibody specifically recognizes human SR-BI (Fig. 6b to d). Naive Huh7-J20 cells were preincubated with various concentrations of neutralizing human-mouse anti-SR-BI MAb151-NP1 prior to infection with each virus. Interestingly, both mutants were considerably less sensitive to neutralization than the wild type, suggesting that these mutants are significantly less dependent on SR-BI for virus entry (Fig. 6e). However, the residue 420 mutants were inhibited by the highest concentrations of anti-SR-BI tested. Inhibitory SR-BI antibodies have been shown to inhibit both primary attachment and lipid transfer activity functions (30). Therefore, to investigate which SR-BI function targeted by antibody treatment was responsible for the reduced sensitivity, we used the chemical inhibitor BLT-4, which blocks SR-BI lipid transfer activity (59). Huh7-J20 cells were preincubated with increasing concentrations of BLT-4 before infection with virus. In parallel, cell viability in the presence of BLT-4 was assessed and found not to be affected (Fig. 6g). The JFH-1_WT virus was sensitive to BLT-4 treatment; in contrast, the JFH-1_W420F and JFH-1_W420Y viruses were not inhibited, even at the highest dose (Fig. 6f). These data suggest that the W420 mutant viruses access cells independently of SR-BI lipid transfer activity but can still use SR-BI for primary attachment.

FIG 6.

Virus–SR-BI receptor interactions. (a) Cell surface expression of the fusion protein SR-BI–GFP was measured by detection of EGFP by flow cytometry. The filled gray peak represents CHO-SR-BI cells, and the black line represents CHO-SR-BI-GFP cells. (b to d) Expression of human SR-BI was measured by comparing the binding of anti-SR-BI MAb151-NP1 (black line) and an IgG1 isotype control (filled gray peak) to CHO-K1 (b), CHO-hSR-BI (c), and CHO-hSR-BI-GFP (d) cells. (e) Neutralization of JFH-1_WT, JFH-1_W420F, and JFH-1_W420Y viruses by anti-SR-BI MAb151-NP1. (f and g) Dose-response curves of BLT-4 for infectivity of JFH-1_WT, JFH-1_W420F, and JFH-1_W420Y viruses (f) and cell viability (g). Each data set shows the average of results from 2 (e) or 3 (f and g) independent experiments, and error bars show standard errors of the means.

Previous studies have shown that high-density lipoprotein (HDL) enhances HCVcc entry via SR-BI (60, 61). More recently, Diao et al. demonstrated that HDL enhancement can be inhibited by treatment with erlotinib, which targets EGFR, a known host factor for HCV entry (62). This finding indicates that SR-BI and EGFR may use the same internalization pathway. Therefore, we investigated whether mutation of W420 affected EGFR-dependent entry. Naive cells were pretreated with the EGFR kinase inhibitor erlotinib prior to infection with virus. As expected, the JFH-1_WT virus was sensitive to erlotinib in a dose-dependent manner with a 50% inhibitory concentration (IC₅₀) of 0.244 μM, which is in agreement with the range observed previously (Fig. 7a) (23, 62). In comparison, the JFH-1_W420F andJFH-1_W420Y viruses were markedly less sensitive to erlotinib treatment. In parallel, cell viability was assessed, and erlotinib treatment was found to have no effect (Fig. 7b). Erlotinib targets the tyrosine kinase domain of EGFR; however, previous studies also indicate a role for the ligand-binding domain of EGFR in HCV entry. Although previous studies agree that EGFR ligands such as EGF promote HCV infection, the data are conflicting with regard to the effect of an EGFR neutralizing antibody that prevents ligand interactions (23, 62). We determined the response of JFH-1_WT, JFH-1_W420F, or JFH-1_W420Y to incubation in the presence of an EGFR antibody (LA-1) (Fig. 7c). In agreement with the results reported by Diao and coworkers, we found that antibody treatment did not block HCVcc JFH-1_WT infection. Moreover, there was no difference between the wild-type virus and the W420 mutants. The main target of erlotinib is EGFR; however, Neveu et al. recently reported that erlotinib also inhibits cyclin G-associated kinase (GAK) during HCV entry (63). GAK is a regulator of clathrin-mediated endocytosis that recruits clathrin and AP-2 to the plasma membrane. GAK has been shown to regulate EGFR internalization and promote EGF uptake (64). Therefore, erlotinib treatment targets two steps of the EGFR pathway. The resistance to erlotinib inhibition indicates that JFH-1_W420F and JFH-1_W420Y do not require the EGFR/GAK route for entry. The host cell kinase AP-2-associated protein kinase 1 (AAK1) is a second regulator of AP-2 clathrin-mediated endocytosis. AAK1 was also shown to regulate EGFR-mediated HCV entry (63). Therefore, we used sunitinib, a kinase inhibitor that targets AAK1, to investigate if the W420 mutants require AAK1. Huh7-J20 cells were pretreated with increasing concentrations of sunitinib prior to infection with the virus. We found that the JFH-1_WT, JFH-1_W420F, and JFH-1_W420Y viruses were inhibited by sunitinib in a dose-dependent manner (Fig. 7d). Interestingly, at the highest concentrations tested, the mutants were more sensitive to sunitinib inhibition than the JFH-1_WT virus, indicating that the mutants were more dependent on this entry route (Fig. 7e). Sunitinib treatment was found to affect cell viability only at the highest concentration tested (Fig. 7f).

FIG 7.

Virus-EGFR interactions. (a and c to e) Dose-response curves for infectivity of JFH-1_WT, JFH-1_W420F, and JFH-1_W420Y viruses in the presence of erlotinib (a), anti-EGFR (c), and sunitinib (d and e). For panel e, wild-type and residue 420 mutant viruses were analyzed by the Student t test; asterisks show statistically significant differences (*, P < 0.05; **, P < 0.05). (b and f) Dose-response analysis of cell viability for erlotinib (b) and sunitinib (f). Each data set shows the average of results from 2 (c) or 3 (a, b, and d to f) independent experiments, and error bars show standard errors of the means.

DISCUSSION

We investigated the role of the highly conserved tryptophan residue at aa 420 within the AP33 epitope of the E2 glycoprotein by replacing this residue in the full-length viral genome with phenylalanine, tyrosine, alanine, arginine, and valine. The phenotypes of the E2 mutants in this study highlight the importance of tryptophan at position 420 during virus infection. None of the substitutions at position 420 had an effect on viral replication levels, but overall, the viral titers of the JFH-1_W420A and JFH-1_W420R viruses were significantly decreased. Analysis of the amount of HCV core protein present in the medium indicates that there is no effect on virus assembly and secretion. The decrease in viral titers is apparent in the FFU assay, which requires infection of naive cells and therefore reveals a defect in the entry step of the viral life cycle.

It is remarkable that the only mutation at this position detected in 1 out of 2,161 naturally occurring HCV sequences was W420R, particularly as our results indicate that this virus is very disabled. In addition, the fact that the JFH-1_W420R virus reverted to the wild-type sequence upon serial passaging is strong evidence that W420 is important for function. However, these mutations were assessed only in a JFH-1 background; therefore, it is possible that other compensatory mutations were present in the natural variant that improved viral fitness. In terms of amino acid properties, phenylalanine and tyrosine are the most conservative mutations to replace tryptophan. Indeed, the unaltered infectivity of the JFH-1_W420F and JFH-1_W420Y mutants suggests that other aromatic residues can replace the tryptophan residue at this position. Furthermore, our results with JFH-1_W420A and JFH-1_W420V also demonstrate that smaller residues can substitute for the tryptophan albeit less efficiently in the case of alanine. Indeed, considering our mutagenesis data from the HCVcc system, it is somewhat surprising that W420 is so strictly conserved in nature. In contrast, the data from the HCVpp system clearly show that tryptophan is essential at this position, as all substitutions severely reduced infectivity. Our data also suggest a requirement for W420 for efficient assembly of HCVpp, as W420A, -V, and -R HCVpp contained low levels of the E2 glycoprotein. Further analysis would be required to assess if a similar defect is observed in the HCVcc system, although this is unlikely to be the case, at least with the JFH-1_W420V mutant, which exhibited infectivity levels that were similar to those of wild-type HCVcc (Fig. 1d).

At first glance, a comparison of the data for binding and for neutralization by the E2 conformational HMAb CBH-5 or by sCD81-LEL shows a lack of a direct correlation between the two properties. This is in concordance with observations for other virus systems, as neutralization activity is dependent on many additional factors (65–67). Our assays showed that the mutant E2 glycoproteins bound more weakly than the wild type to HMAb CBH-5 or sCD81-LEL, whereas the corresponding mutant virions were more sensitive than the wild-type virus to neutralization by these same molecules, suggesting a clear difference between the wild-type and mutant virions. There are numerous possible mechanisms that may influence sensitivity to neutralization. These mechanisms include differences in binding affinity between the wild-type and mutant E2 proteins, the number of functional E1E2 complexes present on the virion surface, and the number of antibody or receptor molecules required for neutralization. For example, a change in the angle of the antibody relative to the virion surface will affect the amount of steric hindrance caused by a single molecule. The HCV virion is protected by a glycan shield (68); therefore, it is possible that mutation of W420 affects the position of the glycans and increases the exposure of the antibody epitope. However, in the core E2 structure, the majority of glycans were positioned on a different face of E2, and the CD81/CBH-5-binding site was relatively exposed, suggesting that this is not a likely explanation (35). There is evidence that this region is masked by HVR1 (69–71) in native virions. During virus entry, an interaction with a host factor induces a conformational change that exposes the CD81-binding site and enables interactions with CD81. Both wild-type and mutant viruses were similarly neutralized by anti-CD81, which prevents the virion-CD81 interaction by blocking CD81 receptors on the cell, indicating that a substitution at position 420 did not alter the CD81 dependence of infectivity. This demonstrates that there was no significant net decrease in virion-CD81 interactions despite the decreased affinity for CD81, suggesting that this was counteracted by the improved accessibility of the CD81-binding region. Indeed, the increased sensitivity to competition with sCD81-LEL indicates that the CD81-binding region on the mutant virion is more exposed than that on the wild-type virion. This observation is analogous to data from a previous study of viruses with point mutations within this region. E1E2 containing the mutations N415D, T416A, N417S, and I422L bound sCD81-LEL similarly to or better than the wild type (53). This is in contrast to E1E2 containing mutations at position 420, which had reduced or undetectable levels of binding compared to wild-type E1E2, supporting the hypothesis that tryptophan at position 420 is important for the E2-CD81 interaction. The fact that the reduced affinity for E1E2 does not appear to have a detrimental effect on the virion-CD81 interaction demonstrates either that increased exposure of CD81-binding sites on the virion can compensate completely or that the affinity of E1E2 on the virion is not as strongly reduced by mutations at this position.

The neutralization experiment with anti-SR-BI found that a substitution at W420 reduced the requirement for SR-BI for the infectivity of these viruses. Again, this finding is consistent with data from several studies of viruses with mutations within this region. Viruses containing the point mutations I414T, N415D, T416A, N417S, and I422L in E2 were all shown to have reduced sensitivity to inhibition by anti-SR-BI (53, 72). The HVR1 region of E2 lies immediately upstream, and a virus with a complete deletion of HVR1 (delHVR1) was shown to be completely resistant to neutralization by anti-SR-BI (69). The delHVR1 result was expected, as several studies have shown that the E2–SR-BI interaction maps to HVR1 and, more recently, that inhibition by anti-SR-BI or SR-BI inhibitors maps to this region (19, 69, 70). However, HVR1 is still present in the W420 mutants, suggesting that the SR-BI-binding site is less accessible in the mutant virions. The complete resistance of the W420 mutant viruses to the inhibitor BLT-4 indicates that the reduced dependence on SR-BI is linked to the lipid transfer function, which is required for both the access and enhancement functions of SR-BI. This is consistent with previously reported observations of N415D, T416A, N417S, and I422L mutant viruses that were resistant to HDL-mediated enhancement, which requires the SR-BI lipid transfer activity (53, 73, 74). The observation that substitution of tryptophan at position 420 also rendered the viruses resistant to erlotinib, which targets EGFR and GAK, two components required for EGFR internalization, suggests that the position 420 mutant viruses do not use this route for entry. The reduced dependence on SR-BI together with the observation by Diao et al. that erlotinib can also block SR-BI-dependent HDL-mediated enhancement strongly suggest that SR-BI, EGFR, and GAK are involved in the same entry route. The small but significant increase in sensitivity to sunitinib indicates that a substitution at position 420 makes the virus more reliant on AAK1-dependent internalization. Therefore, together with the CD81 data, this indicates a model where an alteration at position 420 causes a subtle change in the conformation or flexibility of HVR1 that prevents the interaction with SR-BI and increases accessibility to the CD81-binding site. This in turn alters the entry requirements for clathrin-mediated uptake blocking the SR-BI–EGFR-GAK pathway and favoring the CD81-EGFR-AAK1 route.

Two of the antibodies tested (AP33 and 3/11) bind to E2_412–423, and W420 has been identified as a critical binding residue for both antibodies by alanine scanning and structural analyses of antibody bound to peptide (34, 37–39). E2_412–423 is structurally flexible and is recognized by each antibody in a different conformation. It adopts a β-hairpin conformation when bound to MAb AP33 and a completely different linear, open conformation when bound to MAb 3/11. Our data suggest that W420 is less critical for the interaction with MAb 3/11, as both hydrophobic substitutions retained binding activity for MAb 3/11. Scrutiny of the antigen/antibody interfaces reveals that the interaction of the W420 side chain with both antibodies is dominated by (i) hydrophobic interactions and (ii) hydrogen bonds of the NE1 atom with the carbonyl oxygens of T96 and T97 (MAb 3/11) or N91 (MAb AP33) of the light chain (Fig. 8). The spatial organization of the two antigen/antibody complexes is such that the distance between a hydrophobic phenyl ring as part of a phenylalanine or tyrosine side chain in position 420 and the respective carbonyl oxygens would be shorter for the AP33 complex (∼2.7 Å) than for the 3/11 complex (∼3.2 Å). This disadvantageous interaction could explain the observed lower tolerance of MAb AP33 for a phenylalanine or tyrosine residue at this position. Nevertheless, a hydrophobic residue at this position is essential, as all other mutations ablated binding activity.

FIG 8.

Antibody-peptide interfaces. Shown is a detailed view of the interface between the common part of a synthetic E2 peptide (aa 413 to 423) and Fabs AP33 (left) and 3/11 (right). The Fab is colored according to a normalized hydrophobicity scale from white (hydrophobic) to orange (hydrophilic). The hydrogen bonds to carbonyl groups of residues NL91 (AP33) and TL96/TL97 (3/11) are shown as dashed black lines, and the carbonyl groups are shown as sticks and colored red.

Inspection of the neutralization profile of sCD81-LEL shows that the JFH-1_W420F virus is more sensitive to neutralization than JFH-1_W420Y, suggesting that the hydroxyl group of the tyrosine side chain does not favor interactions with CD81. This is in line with the fact that only the JFH-1_W420F mutant and not the JFH-1_W420Y mutant partially retained CD81-binding activity and suggests that a polar group within the side chain of residue 420 is not beneficial for receptor binding. One possible interpretation of these results could be that this residue provides an additional hydrophobic anchor point to bind the mostly hydrophobic binding site within CD81 (75). Of note, a second antigenic region contributing to the CD81-binding site displays a hydrophobic protrusion constituted by F442 and Y443 that is essential for virus propagation (76). In conjunction with the conformational flexibility around W420 and the essential role of a hydrophobic side chain at position 420, this suggests a stabilizing role of W420 in E2-CD81 binding.

The aim of the present study was to determine if E2 residue W420 was indeed a contact residue for the CD81 receptor during HCVcc entry. Our results clearly show that W420 is required for virus entry and is required for E2-CD81 binding in the virion. In addition, our data highlight the relationship between E2-CD81 and E2–SR-BI interactions, as a mutation at this position also modulates the interaction of the virion with the SR-BI receptor and the subsequent internalization route. The strong requirement for a hydrophobic residue at position 420 also provides new insights into the mode of binding to the cellular receptor.