Identification of Conserved Residues in Hepatitis C Virus Envelope Glycoprotein E2 That Modulate Virus Dependence on CD81 and SRB1 Entry Factors

ABSTRACT

In spite of the high variability of its sequence, hepatitis C virus (HCV) envelope glycoprotein E2 contains several conserved regions. In this study, we explored the structural and functional features of the highly conserved E2 segment from amino acid (aa) 502 to 520, which had been proposed as a fusion peptide and shown to strongly overlap a potential conserved neutralizing epitope. For this purpose, we used reverse genetics to introduce point mutations within this region, and we characterized the phenotypes of these mutants in the light of the recently published structure of E2. The functional analyses showed that their phenotypes are in agreement with the positions of the corresponding residues in the E2 crystal structure. In contrast, our data ruled out the involvement of this region in membrane fusion, and they indicate that alternative conformations would be necessary to expose the potential neutralizing epitope present in this segment. Of particular interest, we identified three specific mutations (Y507L, V514A, and V515A) located within this neutralizing epitope which only mildly reduced infectivity and showed no assembly defect. These mutations modulated HCV dependence on the viral receptor SRB1, and/or they also modulated virion sensitivity to neutralizing antibodies. Importantly, their characterization also showed that amino acids Y507, V514, and V515 contribute to E2 interaction with HCV receptor CD81. In conclusion, our data show that the highly conserved E2 segment from aa 502 to 520 plays a key role in cell entry by influencing the association of the viral particle with coreceptors and neutralizing antibodies.

IMPORTANCE Hepatitis C virus (HCV) envelope proteins E1 and E2 exhibit sequence variability. However, some segments of the envelope proteins are highly conserved, suggesting that these sequences play a key role at some steps of the HCV life cycle. In this work, we characterized the function and structure of a highly conserved E2 region that is targeted by neutralizing antibodies and had been proposed as a fusion peptide. Our data ruled out the involvement of this region in membrane fusion but allowed for the identification of new residues modulating the interaction of the virus with entry factors and its sensitivity to neutralizing antibodies. Moreover, structural data suggest that alternative conformations could exist for E2, which would explain the presence of a partially masked neutralizing epitope in this segment in the currently available E2 structure. Overall, our findings highlight the importance of conserved regions in the sequences of HCV envelope proteins.

INTRODUCTION

Approximately 160 million people worldwide suffer from chronic hepatitis C and are at risk of liver cirrhosis and cancer. For a long time, standard therapy for chronic hepatitis C virus (HCV) infection consisted of the administration of pegylated alpha interferon and ribavirin, which is effective in only 40 to 60% of cases. Importantly, hepatitis C treatment is rapidly changing, and recent results from clinical trials suggest that many HCV infections may be cured with a relatively brief therapy based on an all-oral combination of direct-acting antivirals (1). However, before HCV-related morbidity and mortality significantly decline worldwide, a certain number of challenges remain to be overcome, such as making anti-HCV drugs more affordable for low-income countries and developing treatments adapted to patients with advance disease and comorbidity, as well as prevention of liver graft infection (2). Furthermore, a vaccine will also be desirable for global control of infection (3).

One important hurdle in HCV therapy and vaccine development is the large genetic heterogeneity of the virus. Indeed, HCV isolates have been grouped into seven genotypes and a number of subtypes, which have different geographical distributions (4). A clear association between HCV genotypes and their susceptibility to antiviral treatments has been widely documented. Moreover, in a single host, HCV exists as a combination of related but genetically different variants that constitute quasispecies. This high variability rate favors the appearance of escape variants resistant to antiviral therapies. Among HCV proteins, the envelope glycoproteins, E1 and E2, present the highest sequence variability. Nevertheless, some regions in these proteins are highly conserved, suggesting that they play a major role in the viral life cycle (5). Interestingly, a certain number of conserved neutralizing epitopes have been identified in the E2 envelope glycoprotein (6). Thus, the characterization of such conserved regions may document mechanisms involved in virus-cell interactions while facilitating the design of an immunogenic vaccine. Among these sequences, the E2 segment composed of amino acids 496 to 515 (aa496-515 segment) contains a novel potential conserved neutralizing epitope (7). Moreover, the overlapping aa502-520 segment, which is rich in glycine and nonpolar amino acids and is composed mainly of uncharged residues, has been suggested to be a potential fusion peptide (5). However, the crystal structure of the E2 ectodomain has been determined very recently, and this protein does not present the expected three-domain organization shared by class II viral fusion proteins but rather shows a globular structure containing many regions with no regular secondary structure (8, 9). Indeed, E2 is composed of a central beta sandwich flanked by front and back layers consisting of loops, short helices, and beta sheets.

Within the E1E2 complex, E2 is the major target of neutralizing antibodies and is also the receptor binding protein which has been shown to interact with two major HCV coreceptors, CD81 tetraspanin (10) and scavenger receptor B1 (SRB1) (11). However, the process of HCV entry remains poorly understood. Indeed, it is a complex, multistep process involving several other entry factors acting at the initial attachment step, cell surface transport, cellular uptake, and membrane fusion (12). Thus, beyond SRB1 and CD81, several other cellular factors are required at different steps of HCV entry, which include the tight-junction proteins claudin-1 and occludin (13, 14), epidermal growth factor receptor (EGFR) (15), Niemann-Pick C1-like1 (NPC1L1), (16) and transferrin receptor 1 (17).

Since the aa502-520 segment has been suggested as a potential fusion peptide and contains a potential conserved neutralizing epitope, we wanted to investigate the potential function of this highly conserved segment in the light of the atomic structure of this protein. For this purpose, we generated viruses carrying point mutations in this region. The functional study of these mutants showed that their phenotypes are in agreement with the positions of the corresponding residues in the E2 structure. However, conformational changes would be necessary to expose the potential neutralizing epitope present in this segment. Our study also led to the identification of three residues that modulate HCV dependence on SRB1 and CD81 receptors. Finally, our data also ruled out the involvement of this region in membrane fusion.

MATERIALS AND METHODS

Peptide synthesis and purification.

The E2-AG peptide, representing amino acids 501 to 525 of E2 from the HCV strain JFH-1 (accession number AB047639; the amino acid sequence is shown in Fig. 1) was synthesized on a Milligen 9050 apparatus, employing N-(9-fluorenyl)methoxycarbonyl (Fmoc) chemistry, and purified by reversed-phase high-performance liquid chromatography (RP-HPLC) on a Nucleosil C₁₈ column (120 Å, 5 μm) using a water-acetonitrile gradient containing 0.1% trifluoroacetic acid. The peptide was eluted as a single peak and identified by mass spectroscopy at its expected molecular mass. To promote intrachain disulfide formation by oxidation with molecular oxygen, purified E2-AG peptide was diluted to 30 μM in 10 mM ammonium acetate (pH 8.5) and incubated at room temperature. The follow-up of cyclization and the final purification of cyclized peptide were done by RP-HPLC.

FIG 1.

Structural characterization of the E2 aa502-520 segment. (A) Ribbon diagram of the crystal structure of the HCV E2 core glycoprotein (PDB accession code 4MWF) showing the central β-sandwich scaffold flanked by the front and back protein layers and the CD81 binding loop (8). Disulfide bridges are colored green, and the highly conserved aa502-520 segment is colored red. (B) Amino acid repertoire showing the high sequence conservation of the aa502-520 segment. The sequences of E2 aa 499 to 523 from the HCV H77 consensus clone (GenBank accession number AF009606) and aa 501 to 525 from the HCV JFH-1 clone (accession number AB047639) are indicated on the top and bottom, respectively. Amino acids are numbered with respect to E2 and the HCV polyproteins, with strain H77 used for reference numbering (53). The amino acid repertoire deduced from ClustalW multiple alignments of the 27 representative E2 sequences from confirmed HCV genotypes and subtypes (listed with the accession numbers in Table 1 of reference (4), including the recently described genotype 7a (accession number EF108306; see the European HCV Database for details [http://euhcvdb.ibcp.fr/]), is shown. The degree of amino acid and physicochemical conservation at each position can be inferred from the extent of variability (with amino acids listed in decreasing order of frequency from top to bottom), together with the similarity index according to ClustalW convention (asterisk/red, invariant; colon/green, highly similar; period/blue, similar) (54). Fully conserved cysteine residues are highlighted in green. (C) Far-UV circular dichroism spectra of linear (top) and cyclized (bottom) E2-AG peptide were recorded as previously described (reference 18 and references therein), using a peptide concentration of 30 μM. Analyses were performed in 5 mM sodium phosphate buffer, pH 7.4 (H₂O) (dashed line), complemented with either 100 mM dodecyl phosphocholine (DPC) (large dashed line), 100 mM n-dodecyl-β-d-maltoside (DDM) (dotted line), or 50% 2,2,2-trifluoroethanol (TFE) (solid line). The CD spectra recorded in 100 mM sodium dodecyl sulfate (SDS) and 1% l-α-lysophosphatidyl choline (LPC) were close to those obtained in DPC and DDM (not shown for clarity). (D) Summary of sequential (i, i + 1) and medium-range (i, i + 2 to i, i + 4) nuclear Overhauser enhancements (NOEs) of cyclized E2-AG peptide in 50% TFE. Data were deduced from the NMR spectra recorded at 500 MHz on a Bruker Avance 500 MHz spectrometer using standard homonuclear pulse sequences, as described previously (reference 18 and references therein). Intraresidue backbone resonances and aliphatic side chains were identified from homonuclear ¹Hα clean total correlation spectroscopy (TOCSY) and confirmed with ¹Hα-¹³Cα heteronuclear single quantum correlation (HSQC) in ¹³Cα natural abundance. Sequential assignments were determined by correlating intraresidue assignments with interresidue cross peaks observed in bidimensional ¹Hαβ NOE spectroscopy (NOESY). Intensities of NOEs are indicated by the height of the bars. Sequential NOEs allowing the assignment of proline residues are indicated in red. Asterisks indicate that the presence of an NOE cross peak was not confirmed because of overlapping resonances. (E) NMR ¹Hα and ¹³Cα chemical shift differences were calculated by subtraction of the experimental values from the random-coil conformation values in 50% TFE (55). The dashed lines indicate the standard threshold values for an α-helix (−0.1 ppm for ΔHα and 0.7 ppm for ΔCα). (F) Amino acid sequence and NMR representative structure of cyclized E2-AG peptide. (a) Sequence numbering refers to the H77 strain for reference numbering (53) and JFH-1 sequence numbering for the synthetic peptide. Residues are color coded according to their physicochemical properties: hydrophobic residues (P, V, F, and Y) are dark gray, Gly and Ala residues are light gray, polar residues Ser and Thr are yellow, basic Arg and acidic Asp residues are blue and red, respectively, and cysteine residues are green. (b) Representative structure model of cyclized E2 peptide showing the Cys 503-Cys 508 disulfide bond. The ribbon representation with backbone and side chain residues colored as in panel a. Residue numbering refers to the H77 strain. Panels A and F (b) were generated from structure coordinates using the VMD program (http://www.ks.uiuc.edu/Research/vmd/) and rendered with POV-Ray (http://www.povray.org).

Structural analyses.

Far-UV circular dichroism (CD) spectra of linear and cyclized E2-AG peptides, nuclear magnetic resonance (NMR) spectroscopy, NMR-derived constraint and structure calculation, and molecular modeling and structure representation were performed by standard approaches as detailed previously (18).

Cell culture.

Huh-7 hepatoma cells (19) were grown in Dulbecco's modified essential medium (DMEM) (Invitrogen) supplemented with 10% fetal calf medium.

Antibodies.

Anti-HCV monoclonal antibodies (MAbs) A4 (anti-E1) (20), 3/11 (anti-E2; kindly provided by J. A. McKeating, University of Birmingham, United Kingdom) (21), JS81 (anti-CD81; BD Pharmingen), 9E10 (anti-NS5A; kindly provided by C. M. Rice, Rockefeller University, New York, NY, USA) (22), OM 8A9-A3 (anti-claudin-1) (23), AR3A and AR5A (anti-E2 and anti-E1E2, respectively; kindly provided by M. Law, The Scripps Research Institute, La Jolla, CA, USA) (24), anti-SRB1 clone 25 (BD Biosciences) and anti-beta-tubulin (Sigma) were used in this work. Anti-NS5A polyclonal antibody was kindly provided by M. Harris (University of Leeds, United Kingdom) (25).

Mutagenesis.

The virus used in this study is based on the JFH-1 strain (genotype 2a; GenBank accession number AB237837) (26), kindly provided by T. Wakita (National Institute of Infectious Diseases, Tokyo, Japan). Mutations were introduced in a modified version of the plasmid carrying the full-length JFH-1 genome (pJFH1-CS-A4) engineered to reconstitute the A4 epitope in E1 (27) and titer-enhancing mutations (28). Mutations were constructed by sequential PCR steps as described previously (29), using an Expand High Fidelity Plus PCR system (Roche). The amplicons were digested with BsiWI and NotI and ligated into the BsiWI/NotI-digested pJFH1-CS-A4 plasmid. The mutations in the JFH-1 parental genome were confirmed by sequencing the amplicons. Viral RNAs were produced by in vitro transcription as described previously (30). The nonreplicative control of the HCV genome (GND) contained a GND mutation in the NS5B active site as previously reported (26).

Infectivity assays.

Viral RNAs were delivered into Huh-7 cells by electroporation as described previously (30). Supernatants containing extracellular virus were harvested 48 h, 72 h, and 96 h after electroporation, and cell debris was removed by centrifugation for 5 min at 10,000 × g. Cells were washed with phosphate-buffered saline (PBS), harvested by treatment with trypsin, and pelleted at 100 × g for 5 min. Cell pellets were resuspended in complete medium (medium supplemented with 10% fetal bovine serum) and lysed by three freeze-thaw cycles. Cell lysates were clarified by centrifugation at 10,000 × g for 5 min. Cell supernatants and cell extracts were further processed for immunofluorescence to measure infectivity. The titers of the intracellular virus fractions were expressed relative to the volume of the corresponding supernatant.

HCV core quantification.

The cell supernatants from electroporated Huh-7 cells were harvested, and the cells were lysed in PBS containing 1% Triton X-100 and supplemented with protease inhibitor cocktail (Roche). The lysates were cleared by centrifugation for 15 min at 14,000 × g. Intracellular and extracellular core concentrations were then quantified by a fully automated chemiluminescent microparticle immunoassay according to the manufacturer's instructions (Architect HCVAg, Abbott, Germany) (31).

CD81 interaction and immunoprecipitation assays.

HCV-electroporated Huh-7 cells were lysed in PBS containing 1% Triton X-100 and supplemented with protease inhibitor cocktail (Roche). The lysates were cleared by centrifugation for 15 min at 14,000 × g. Glutathione-Sepharose beads (Amersham Bioscience) were washed twice with PBS. For each cell lysate, 50 μl of glutathione-Sepharose beads was incubated for 2 h at 4°C with 10 μg of the large extracellular loop (LEL) of human CD81 in fusion with glutathione S-transferase (GST) in 1 ml cold PBS. The glutathione-Sepharose beads were then washed twice with cold PBS, and cell lysate samples containing E1E2 proteins were incubated overnight at 4°C with the glutathione-CD81 LEL beads. The beads were then washed with PBS containing 1% Triton X-100, and they were resuspended in 30 μl of Laemmli buffer (200 mM Tris-HCl [pH 6.7], 0.5% SDS, 10% glycerol, and 100 mM beta-mercaptoethanol). Samples were boiled and loaded for 12% SDS-PAGE, followed by Western blotting to determine the presence of HCV envelope glycoproteins.

CD81 inhibition assay and virus neutralization.

Neutralization and inhibition experiments were performed with virus-containing supernatants harvested at 72 h postelectroporation. For inhibition with the CD81 LEL, viruses were preincubated with human CD81 LEL in fusion with GST for 2 h at 37°C before inoculation of Huh-7 cells. For antibody neutralization, the viruses were mixed with antibodies and immediately used for inoculation of Huh-7 cells. At 5 h postinoculation, complete medium was added to the cells, and at 72 h postinfection, cells were fixed with methanol and processed for immunofluorescence to measure infectivity.

RNA interference.

Huh-7 cells were transfected with small interfering RNA (siRNA) pools (Dharmacon) targeting CD81 or SRB1 using RNAiMax (Invitrogen) and according to the manufacturer's instructions. Briefly, 3 μl RNAiMAx Lipofectamine in 500 μl PBS was mixed with 50 pmol siRNA per well in 6-well plates. After 30 min of incubation, 2 ml of complete medium containing 2 × 10⁵ cells was added to the siRNA-Lipofectamine mix for each well. The knockdown effects were determined at 48 h after transfection by Western blotting at the time of virus inoculation. The effects of receptor silencing on virus infection were determined 72 h later by immunolabeling of the infected cells.

Equilibrium density gradient analyses.

At 72 h postelectroporation, viruses were precipitated from the supernatants using polyethylene glycol 6000 at a final concentration of 8%. The mixture was shaken for 1 h on ice and incubated overnight at 4°C. The samples were then centrifuged at 8000 rpm (Beckman JA10 rotor) for 25 min, and the pellets were resuspended in 1 ml PBS. Virus preparations were centrifuged for 5 min at 5,000 × g to remove insoluble materials before loading on continuous 10 to 50% iodixanol gradients. The gradients were spun for 16 h at 36,000 rpm in an SW41 rotor (Beckman). Following centrifugation, 12 fractions of 1 ml were harvested and analyzed for their infectivity.

RESULTS

Structural characterization of the highly conserved aa502-520 segment.

The E2 aa502-520 segment is almost strictly conserved within all HCV genotypes (Fig. 1B) and has recently been suggested to be a potential fusion peptide because it exhibits structural features similar to those of class II fusion proteins (5). It is indeed composed mainly of noncharged residues, glycine and nonpolar amino acids, and includes two cysteine residues, at aa 503 and 508, that were found to form a disulfide bridge (5). However, the recently reported crystal structure of HCV E2 glycoprotein revealed that the aa502-520 segment belongs to the central β-sandwich scaffold and that the two cysteine residues at aa 503 and 508 are involved in two distinct disulfide bridges (503-429 and 508-552) (Fig. 1A) (8). Interestingly, a conserved neutralizing epitope composed of aa 496 to 515 in this region has also been recently described, but examination of the E2 crystal structure indicates that this epitope is only partially accessible at the protein surface (see Discussion). These features suggest that alternative conformations of E2 could exist, at least transiently and/or during its folding process. To gain insights into the structural and lipotropic properties of the highly conserved aa502-520 segment, a peptide encompassing E2 residues 501 to 525 of HCV strain JFH1 and designated E2-AG was chemically synthesized, purified in either its reduced (linear) form or oxidized (cyclized) form, and analyzed by circular dichroism (CD) and nuclear magnetic resonance (NMR). The large negative peaks around 220 nm with shoulder around 230 nm observed in the CD spectra of both peptide forms in aqueous solution indicate a mainly random coil conformation (Fig. 1C). The presence of detergents such as n-dodecyl phosphocholine (DPC) or n-dodecyl-β-d-maltoside (DDM) to mimic the membrane environment did not induce significant peptide folding at near-neutral pH (Fig. 1C) or under acidic conditions (pH 4.4 or 5.5) (data not shown). The broadening of signals observed for the cyclized peptide throughout the spectra is typical of the presence of a disulfide bond. The addition of 2,2,2-trifluoroethanol (TFE) to probe the conformational preferences of the peptides resulted in spectra exhibiting some features of a typical α-helix with negative peaks around 208 and 222 nm. However, the low intensity of the positive signal around 192 nm even at a high TFE concentration (90%) (not shown) indicates a limited propensity of the peptide to fold into an α-helix. Accordingly, the various CD deconvolution methods used indicated a maximum α-helix content limited to 20%.

E2-AG peptide samples prepared in 100 mM deuterated DPC micelles or 50% deuterated TFE yielded well-resolved homo- and heteronuclear multidimensional spectra (data not shown). Sequential attribution of all spin systems in 50% deuterated TFE was complete except for the ¹Hα of Phe 509 (numbering refers to the H77 strain) (Fig. 1D), which could not be observed for both the linear and cyclized forms of the peptide. Attempts to observe this proton at higher temperatures or upon dissolution of the peptides in deuterated DPC micelles failed, suggesting that its chemical shift might be close to that of water. In addition, the relatively low intensities of backbone atoms of the aa507-510 segment for the cyclized peptide suggest that this region could undergo slow conformational movement. This explains the lack of connectivities reported for this region in the summary of sequential and medium-range nuclear Overhauser enhancement (NOE) connectivities (Fig. 1D). The patterns of NOE connectivities together with the deviation of ¹Hα and ¹Cα chemical shifts from random-coil values (Fig. 1E) along the cyclized peptide sequence indicate a tendency of the C-terminal aa512-522 segment to fold into an α-helix. However, the weak intensities of medium-range NOE connectivities, especially for dαβ(i, i + 3) connectivities, indicate a poorly stable helix. On the basis of NOE-derived interproton distance constraints and of dihedral angles calculated with TALOS+ (32) from ¹Hα and ¹³Cα chemical shifts, a set of 20 structures of cyclized E2-AG peptide was calculated with Xplor-NIH (33). Superimposition of these structures shows that both the N- and C-terminal regions are well organized, including the cyclized aa503-508 segment and a short α-helix for the aa512-522 segment (Fig. 1F and data not shown). However, the poor stability of this helical folding observed only in the presence of the stabilizing organic solvent 50% TFE does not favor the existence of such folding under native conditions. In addition, the conformation of the central region from aa 509 to 511, which remains unfolded under stabilizing conditions and is likely flexible (see above), is consistent with the turn conformation observed for the corresponding segment in the crystal structure of E2 (Fig. 1A). In contrast, the formation of the aa503-508 intradisulfide bond in this segment as observed by Krey et al. (5) would generate a drastically different conformation, incompatible with the existence of the central β scaffold observed in the E2 crystal structure. This suggests that the existence of an aa503-508 intradisulfide bond is unlikely, but one cannot exclude that it could form transiently during the E2 folding step.

Effects of E2 mutations on HCV infectivity.

To determine the potential role of the aa502-520 segment in E2 functions during the HCV life cycle, we generated a series of mutants in the context of a JFH1-derived infectious clone (Fig. 2A). The mutations were chosen to evaluate the importance of the physicochemical properties of residue side chains at the mutated positions but while limiting the structurally deleterious impact of mutations on the basal structural features of the aa502-520 segment: mutation of Tyr 507 to Leu or Phe and mutation of Phe 509 to Leu were chosen to test the importance of the hydrophobicity and aromatic character of these residue side chains, mutation of Thr 510 to Ser or Ala to test the importance of the presence of a polar and/or methyl group, mutation of Pro 511 and 513 to Ala to evaluate the importance of conformational constraints imposed by a proline residue at these positions, and mutation of Val 514 and 515 to Ala to evaluate the importance of the presence of a bulky and rigid beta-branched hydrophobic residue at these positions. To avoid the formation of disulfide bonds, Cys 503 or 508 was mutated to Ser.

FIG 2.

Effects of E2 mutations on HCV infectivity and locations of mutations in the crystal structure of E2. (A) Presentation of the mutations introduced in the E2 aa502-520 segment. (B) Determination of viral protein expression in Huh-7 cells. Huh-7 cells were electroporated with wild-type (wt) or mutant viral RNAs. At 48 h postelectroporation, cells were lysed and processed for Western blotting using primary antibodies specific for E1 (A4), E2 (3/11), NS5A (9E10), and β-tubulin. (C) Quantification of virus infectivity. Extracellular and intracellular infectious titers (in focus-forming units [ffu]/ml) were measured at 48, 72, and 96 h postelectroporation. Errors bars indicate standard deviations from at least three independent experiments. The black line indicates the detection limit of the assay. (D) Crystal structure of E2, showing the positions of mutated residues. Ribbon representations are in the same orientation as in Fig. 1A (front view) and rotated by 90°C to exhibit the E2 protein side bearing the mutated residues (right view). The surface representation (same orientation as for the right view) shows mutated residues (van des Walls spheres) located at the protein surface. Mutated positions corresponding to mutants that exhibited infectious titers very similar to that of the wild-type virus are colored green (510 and 513 [but residue 507 is colored cyan; see the text]), those corresponding to mutants that were noninfectious or almost noninfectious are colored red (503, 508, 509, and 511), and those which showed a mild decrease in infectivity are colored blue (514 and 515 [but residue 507 is colored cyan; see the text]).

To determine the effects of the mutations on HCV infectivity, Huh7 cells were electroporated with in vitro-transcribed wild-type or mutant RNAs, and intra- and extracellular infectivities were quantified at several times postelectroporation. A viral genome containing an inactive mutation in NS5B polymerase (GND) was used as a negative control. As shown in Fig. 2B, the mutations did not affect virus replication, since the levels of expression of E1, E2, and NS5A were very similar for mutant and wild-type viruses. Furthermore, intra- and extracellular viruses showed in general similar infectivity profiles for each mutant, with approximately 10-fold more infectious virus in the supernatant as generally observed (Fig. 2C), indicating that none of the mutations blocked the secretion of infectious viral particles. However, some mutations affected virus infectivity (Fig. 2C), and we observed three different phenotypes. A first group of mutants (Y507F, T510A, T510S, and P513A) exhibited infectious titers very similar to that of the wild-type virus. The corresponding residues side chains are located at the surface of the protein (colored green in Fig. 2D, except residue 507, which is colored cyan [see below]) and are thus not involved in the stability of the beta sandwich scaffold. The second group (C503S, C508S, F509L, and P511A) contained mutants that were noninfectious or almost noninfectious. These residues (colored in red in Fig. 2D) are located within the interior of the protein core (Phe 509 and Pro 511) or are involved in disulfide bridges (Cys 503 and 508), which are obviously essential for the stability of the E2 fold. Indeed, the aromatic side chain of Phe 509 is in the center of a long-range hydrophobic network including Pro 513, Thr 553, Pro 601, and Tyr 611, and its mutation to the aliphatic, flexible side chain of Leu could not ensure the stabilization of the corresponding fold. Similarly, Pro 511 is also involved in a long-range hydrophobic stabilizing network, including Phe 550, Pro 568, and Val 633, and the short alanine methyl side chain could not ensure the stability of this hydrophobic cluster. The third group (Y507L, V514A, and V515A) showed a mild decrease in infectivity, with an approximately 10-fold reduction compared to that of the wild-type virus. These infectivity differences from the wild type were statistically significant at 96 h postelectroporation (unpaired t test). The corresponding residues (colored in blue in Fig. 2F, except residue 507 in cyan) form a hydrophobic patch at the protein surface but are not directly involved in the folding stability of E2. They are thus most likely involved in intra- and/or intermolecular interactions with other viral and/or cell factors. The lack of a significant effect of the Y507F mutation compared to the mild decrease in infectivity observed with the Y507L mutant points out the importance of the aromatic ring of residue 507 in such interaction.

Effect of E2 mutations on HCV particle assembly.

The decrease in infectivity observed for some mutants could be due either to a defect in virus assembly or to the release of less-infectious viral particles. To distinguish between these two possibilities, we quantified the intra- and extracellular levels of HCV core protein at 72 h after electroporation. Since mutants in the first group did not affect infectivity, we selected only one of them for core protein quantification. For all the tested mutants, the intracellular level of core protein expression was similar to that of the wild type, confirming that the mutations did not affect HCV genomic replication (Fig. 3A). In contrast, more variability was observed for the secreted core protein (Fig. 3B). Indeed, three out of the four mutants exhibiting no infectivity or almost no infectivity (C508S, F509L, and P511A) showed a dramatic reduction in core release, suggesting that these mutations impair virus assembly. In contrast, core protein secretion was only mildly affected for C503S virus, suggesting that this mutant secretes noninfectious particles. Within group three, the Y507L and V514A mutants showed a slight decrease in core protein secretion that correlates with the mild reduction in virus infectivity, whereas the V515A mutant was not affected in core protein secretion, suggesting that the secreted virions are slightly less infectious for this mutant. However, when specific infectivities were calculated at 72 h postelectroporation by normalizing the infectivity (Fig. 2C) to the level of core protein secretion, the differences between the wild type and the mutants were similar to those observed for their infectivity before normalization (data not shown). This suggests that the lower infectivity observed for the Y507L, V514A, and V515A mutants is due to the production of less-infectious viral particles.

FIG 3.

Effects of E2 mutations on HCV assembly. Huh-7 cells were electroporated with wild-type or mutant viral RNAs. Intracellular (A) and extracellular (B) core protein concentrations were measured at 72 h postelectroporation. The error bars indicate standard deviations. The unpaired t test was used to compare the core protein concentrations of the wild-type and mutant viruses. Differences were considered statistically significant if the P value was <0.05 (*).

Y507L, V514A, and V515A mutations affect E2 recognition by CD81.

To further characterize our mutants, we analyzed the effects of the mutations on the folding of HCV envelope glycoproteins. For this purpose, we used the large extracellular loop (LEL) of human CD81, an HCV coreceptor (10), as a probe to determine the formation of properly folded E1E2 complexes in a GST pulldown assay, since this interaction depends on the proper conformation of E2 (34). As expected, E1E2 complexes from all the mutants of the first group were recognized by the CD81 LEL at the same level as the wild-type proteins, indicating that the envelope proteins of the Y507F, T510A, T510S, and P513A viruses demonstrate at least some degree of proper folding (Fig. 4A). None of the mutants of the second group was recognized by the CD81 LEL, suggesting a defect in E1E2 folding for the C503S, C508S, F509L, and P511A viruses, which could explain their defect in assembly and/or infectivity. Interestingly, two mutants of the third group (Y507L and V515A) were not recognized by the CD81 LEL, and for the third mutant (V514A), the levels of E1 and E2 recognized by this receptor were strongly reduced (Fig. 4A). Since the infectivities of the Y507L, V514A, and V515A viruses were only mildly reduced, the dramatic effect on CD81 interaction prompted us to further investigate these mutants. We therefore used well-characterized conformation-sensitive MAbs AR3A and AR5A to further analyze these mutants. MAb AR3A has been shown to recognize a discontinuous epitope on E2 and to block E2-CD81 interaction, whereas MAb AR5A recognizes a discontinuous epitope on E1E2, which does not interfere with CD81 binding (24). As shown in Fig. 4B, MAb AR3A immunoprecipitated E1E2 from the V514A and V515A mutants as efficiently as the wild-type proteins, whereas recognition of Y507L was slightly affected. Furthermore, the V515A mutant was well recognized by MAb AR5A, whereas mutants Y507L and V514A were less efficiently detected. In the case of the V514A mutant, a smaller amount of E1 than of E2 was immunoprecipitated by AR5A, although this mutation did not affect the level of expression of E1 in electroporated Huh7 cells (Fig. 4A). This result might not be due to a deficiency in E1E2 heterodimerization, since equivalent amounts of E1 and E2 were immunoprecipitated by AR3A. This difference could be explained by an impact of the mutation on the presentation of the E1E2 discontinuous epitopes recognized by the AR5A antibody. In contrast to the V514A mutant, Y507L slightly reduced the recognition by both the AR3A and AR5A antibodies, suggesting a more global change in E2 for this mutant. The C503S mutant was not detected or was only slightly detected by these MAbs (Fig. 4B), confirming the folding defect for this virus. Together, these data indicate that HCV envelope proteins of mutants Y507L, V514A, and V515A are properly folded and that Y507 and V515 and to a lesser extent V514 are involved in CD81 binding. This finding is in keeping with the fact that these residues are close to the residues identified to contribute to the CD81 binding site (see Fig. 9 [colored in yellow]).

FIG 4.

Effects of E2 mutations on E1E2 conformation. (A) Interaction of viral envelope glycoproteins with HCV entry factor CD81. At 48 h postelectroporation, E1 and E2 proteins from cell lysates were analyzed by GST pulldown with the help of a CD81 LEL-GST fusion protein. Pulled-down E1 and E2 proteins were revealed by Western blotting with MAbs A4 and 3/11, respectively. (B) Recognition of HCV envelope glycoproteins by envelope-specific conformational antibodies. At 48 h postelectroporation, E1 and E2 proteins from cell lysates were analyzed by immunoprecipitation with MAbs AR5A (E1E2 specific) or AR3A (E2 specific). Immunoprecipitated proteins were revealed by Western blotting with MAbs A4 and 3/11. The amount of lysate loaded in the Western blot (4A) corresponds to 1/3 of the input amount used for the immunoprecipitation and the pulldown assays. These experiments were repeated several times, and the results were very reproducible.

FIG 9.

Positions of mutated residues in the E2 crystal structure compared to the CD81 binding region and the aa496-515 neutralizing epitope. (A) Ribbon and surface representations (right and left panels, in the same orientation as in panels b and c of Fig. 2D) showing the proximity between mutated residues belonging to the highly conserved aa502-520 segment (colored red) and side chains of residues involved in the binding of CD81 (colored yellow, residues 420, 421, 422, 424, 523, 526, 527, 530, 535, 538, and 540; reviewed in Fig. S3 of reference 8). (B) Ribbon and surface representations (right and left panels) of front (a), right (b), and top (c) views of the E2 crystal structure, showing the relative positions of the CD81 binding region (colored yellow), the aa496-515 neutralizing epitope (colored violet), and mutation G451R (colored orange). Note that the aa496-515 epitope includes all residues mutated in this study, which are thus colored violet, but their labels are colored according to panel A. This epitope overlaps the highly conserved aa502-520 segment, for which only the backbone of residues 516 to 520 are colored red (left panels).

The Y507L, V514A, and V515A mutations modulate HCV dependence on CD81 and/or SRB1 receptors.

To further investigate the roles of Y507, V514, and V515 in HCV infection, we sought to determine the effect of their mutation on HCV interaction with its receptors in the context of a cell culture infection. To determine the effects of the Y507L, V514A, and V515A mutations on CD81 recognition by virus-associated glycoproteins, we analyzed the sensitivities of the mutants to inhibition by the CD81 LEL. Interestingly, the infectivities of the Y507L and V515A viruses were not affected by preincubation with the CD81 LEL (Fig. 5A), which is in agreement with the lack of interaction observed in our pulldown assay (Fig. 4A). Furthermore, these two mutants were more sensitive to neutralization by the anti-CD81 antibody than the wild-type virus (Fig. 5B). Indeed, the 50% inhibitory concentrations (IC_50s) were about 0.09 μg/ml and 0.07 μg/ml for Y507L and V515A, respectively (P = 0.015 and P = 0.0026 in the unpaired t test), whereas an IC₅₀ of 0.13 μg/ml was observed for the wild-type virus. The higher sensitivity of these mutants to inhibition by the anti-CD81 antibody could be explained by the lower affinity of E2 for CD81, which can be better outcompeted by the anti-CD81 antibody. In contrast, the V514A mutant was more sensitive to CD81 LEL inhibition and less susceptible to anti-CD81 neutralization (Fig. 5). Indeed, the CD81 LEL inhibited the V514A and wild-type viruses with IC_50s of 0.84 μg/ml and 2.60 μg/ml, respectively (P = 0.026 in the unpaired t test), and MAb JS81 neutralized the V514A and wild-type viruses with IC_50s of 0.25 μg/ml and 0.13 μg/ml, respectively (P = 0.0015 in the unpaired t test). These results are in contradiction with the lower level of interaction observed in our pulldown experiment for the V514A mutant (Fig. 4A). This could be due to the reorganization of the envelope glycoproteins at the surface of the virion as observed for the wild-type virus (35). Indeed, one cannot exclude that this reorganization increases the accessibility of the CD81 binding site on the V514A virion, which could lead to an increased affinity of virion-associated E2 for CD81. Together, these data indicate that the Y507L, V514A, and V515A mutations modulate HCV interaction with CD81 in the context of HCV infection in cell culture. This is also in line with the fact that these residues are close to the residues identified to contribute to the CD81 binding site.

FIG 5.

The Y507L, V514A, and V515A mutations modulate HCV dependence on CD81. Interactions between HCV particles and CD81 were measured by inhibition of virus infectivity in the presence of the CD81 LEL (A) or CD81-specific antibody JS81 (B). Wild-type, Y507L, V514A, and V515A viruses were incubated with various concentrations of CD81 LEL for 2 h at 37°C. The mixtures were then used to infect Huh-7 cells. For anti-CD81 antibody mediated neutralization, viral infections were performed in the presence of different concentrations of MAb JS81. At 72 h postinfection, virus infectivity was measured by immunofluorescence with the help of MAb A4. Infectivity is expressed as the percentage of infection performed in the absence of antibody or CD81 LEL. Mean values and standard deviations from three independent experiments are shown.

To further determine whether the Y507L, V514A, and V515A mutations affect the way HCV interacts with cellular entry factors, additional experiments with anti-SRB1 and anti-CLDN1 antibodies were performed. As shown in Fig. 6A, no change in sensitivity to anti-CLDN1 antibody neutralization was observed. In contrast, mutants Y507L and V514A were less sensitive to neutralization by anti-SRB1 antibody, whereas the V515A mutant was neutralized as efficiently as the wild-type virus (Fig. 6B). To further investigate the lack of SRB1 dependence of the Y507L and V514A mutants, we also analyzed the effect of silencing SRB1 expression on their infectivity. As shown in Fig. 6C, virus infectivity was not affected or was much less affected by SRB1 knockdown for mutants Y507L and V514A, respectively, whereas infectivity of the V515A virus was reduced at the same level as for the wild type. Finally, CD81 silencing was used as a control of the knockdown experiment. However, in contrast to what was observed for SRB1, infectivity was strongly reduced for all the mutants (Fig. 6C), indicating that CD81 remains an essential entry factor for these mutants. Together, these data indicate that the Y507L and V514A mutations modulate the dependence of HCV on SRB1 during infection.

FIG 6.

Dependence of wild-type and mutant HCV on SRB1 and CLDN1. (A and B) Huh-7 cells were infected with wild-type, Y507L, V514A, or V515A viruses in the presence of increasing concentrations of antibody directed against CLDN1 (A) or SRB1 (B). Infectivity was determined as described for Fig. 5. (C) SRB1 or CD81 expression was downregulated by siRNA specific to SRB1 or CD81 mRNA. A nontargeting siRNA was transfected in parallel and served as a control. Knockdown efficiency was determined by Western blotting with the help of anti-SRB1, anti-CD81, and anti-β-tubulin. Virus infectivity was measured by immunofluorescence with the help of MAb A4. Infectivity is expressed as the percentage of infection performed in the presence of the control siRNA. Mean values and standard deviations from three independent experiments are shown. The unpaired t test was used to compare the infectivities of the wild-type and mutant viruses. Differences were considered statistically significant if the P value was <0.05 (*).

The Y507L, V514A, and V515A mutations modulate HCV sensitivity to antibody neutralization.

Since mutations in the E2 envelope glycoprotein can also affect HCV sensitivity to antibody neutralization (36), we also tested the Y507L, V514A, and V515A mutations in an antibody neutralization assay with two well-characterized MAbs (AR3A and 3/11). AR3A is a conformation-sensitive human antibody recognizing a discontinuous epitope located within the CD81 binding region (37), and 3/11 is a rat antibody recognizing a linear epitope also located within the CD81 binding region (38). As shown in Fig. 7, the Y507L mutation did not alter the sensitivity of the virus to neutralization by MAbs AR3A and 3/11, whereas the V514A virus was approximately twice as sensitive to neutralization by these two antibodies. In the case of MAb AR3A, the IC_50s were about 0.05 μg/ml and 0.09 μg/ml for the V514A and wild-type viruses, respectively (P = 0.0009 in the unpaired t test). With MAb 3/11, the IC_50s were about 5.90 μg/ml and 10.89 μg/ml for the V514A and wild-type viruses, respectively (P = 0.0379 in the unpaired t test). In the case of the V515A mutant, a change in neutralization sensitivity was only observed with MAb AR3A (Fig. 7A). Indeed, the IC_50s were about 1.17 μg/ml and 0.09 μg/ml for the V515A and wild-type viruses, respectively (P = 0.0004 in the unpaired t test). Together, these data suggest that the V514A virus has a global increase in sensitivity to antibody neutralization, whereas the reduced sensitivity of V515A to neutralization by MAb AR3A could be due to a local alteration by the mutation of the epitope recognized by this antibody.

FIG 7.

The V514A and V515A mutations modulate the sensitivity of HCV to antibody neutralization. Wild-type, V514A, and V515A viruses were used to infect Huh-7 cells in the presence of increasing concentrations of MAb AR3A or 3/11. At 72 h postinfection, virus infectivity was measured by immunofluorescence with the help of MAb A4. Infectivity is expressed as the percentage of infection performed in the absence of antibody. Mean values and standard deviations from three independent experiments are shown.

Effect of mutations Y507L and V514A on the density of infectious viral particles.

HCV associates with lipoproteins to form lipo-viro particles (LVP), and this interaction is important for its infectivity (reviewed in reference 39). Moreover, HCV-associated lipoproteins have been shown to play a role in virus interaction with SRB1 (40–42). Finally, E2 mutations increasing virus sensitivity to antibody neutralization have also been associated with a shift in virion density (36, 43, 44). In this context, we sought to determine if mutations Y507L and V514A, which modulate the dependence of the virus on SRB1 for entry, as well as the recognition of E2 by neutralizing antibodies in the case of V514A, would affect the association of the virus with lipoproteins. For this purpose, we analyzed the density of infectious viral particles by density gradient ultracentrifugation. We compared the distributions of wild-type and mutant particles across gradients by quantifying the infectivity in each fraction. As shown in Fig. 8, no striking difference could be observed between the density of wild-type infectious particles and those of the Y507L and V514A mutants. Thus, these mutations do not seem to affect the association of the virus with lipoprotein.

FIG 8.

Comparison of the buoyant densities of wild-type and Y507L and V514A mutant viruses. Concentrated cell-cultured HCV from the wild type and mutants Y507L and V514A were separated by sedimentation through a 10 to 50% iodixanol gradient. Fractions were collected from the top and analyzed for their densities and infectivities.

DISCUSSION

Due to their exposure to selection by the humoral immune system, envelope glycoproteins contain the most variable regions in the HCV genome. On the other hand, these proteins are assigned to a wide range of functions, since they participate in virus morphogenesis, they interact with cell surface receptors, and they induce fusion of the viral envelope with a cellular membrane. These various functions may constrain the degree of structural flexibility of envelope protein sequences. As a consequence, some regions of these glycoproteins are highly conserved and might play an important role in E1E2 structure and functions. In this study, we explored the structural and functional features of the highly conserved E2 aa502-520 region, which had been proposed as a fusion peptide and has been shown to strongly overlap a conserved neutralizing epitope including residues 496 to 515 (7). We used reverse genetics to introduce point mutations within this region, and we characterized the phenotypes of these mutants, which are summarized in Table 1. Of particular interest, we identified three specific mutations (Y507L, V514A, and V515A), located within the above-mentioned neutralizing epitope, which only mildly reduced infectivity and showed no assembly defect. Importantly, we demonstrated that these three residues modulate the dependence on the SRB1 and CD81 receptors.

TABLE 1.

Summary of the phenotypes of the mutants

Mutant	Infectivitya	Core releaseb	Precipitation exptsc			Infection inhibition assaysd
			Immunoprecipitation		CD81 pulldown
			AR3A	AR5A		hCD81 LEL	Anti-CD81	Anti-SR-BI	Anti-E2 3/11	Anti-E2 AR3A
Wild type	++	+++	++	+++	++	++	++	++	++	++
C503S	−	++	−	+	−	ND	ND	ND	ND	ND
Y507L	+	++	+	++	−	+/−	+++	+/−	++	++
Y507F	++	ND	ND	ND	++	ND	ND	ND	ND	ND
C508S	−	+	ND	ND	−	ND	ND	ND	ND	ND
F509L	−	+	ND	ND	−	ND	ND	ND	ND	ND
T510A	++	ND	ND	ND	++	ND	ND	ND	ND	ND
T510S	++	ND	ND	ND	++	ND	ND	ND	ND	ND
P511A	−	+	ND	ND	−	ND	ND	ND	ND	ND
P513A	++	+++	ND	ND	++	ND	ND	ND	ND	ND
V514A	+	+++	++	++	+	+++	+	+/−	++	+++
V515A	+	+++	++	+++	−	+/−	+++	++	++	+

^aThe infectivity of cell culture-grown HCV harboring the different E1E2 glycoproteins in the supernatant of electroporated Huh-7 cells was quantified at 96 h post electroporation (Fig. 2). +++, infectious titer greater than or equal to 10⁴ focus-forming units (FFU)/ml; ++, titer of 10³ FFU/ml; −, titer between 0 and 50 FFU/ml.

^bQuantification of core released in the supernatant at 72 h postelectroporation of Huh-7 cells (Fig. 3). +++, concentration greater than or equal to 10³ fmol/liter; ++, concentration between 5 × 10² and 10³ fmol/liter; +, concentration less than or equal to 10² fmol/liter; ND, not determined.

^cThe recognition of E1E2 protein by AR3A and AR5A conformational antibodies and their interaction with the hCD81 LEL were determined by precipitation experiments (Fig. 4). For AR3A: ++, similar amount of E1E2 immunoprecipitated to that of the wild type; +, 2-fold less E1E2 protein immunoprecipitated; −, no envelope protein precipitated. For AR5A: +++, similar amount of E1E2 immunoprecipitated to that of the wild type; ++, 2- to 5-fold less E1E2 protein immunoprecipitated; +, 8-fold less E1E2 protein immunoprecipitated. For CD81 pulldown: ++, similar amount of E1E2 precipitated to that of the wild type; +, 2.5-fold less of E1E2 precipitated; −, no E1E2 precipitated. ND, not determined.

^dThe sensitivities of the mutants to inhibition of infectivity by different antibodies or the hCD81 LEL were assessed (Fig. 5 to 7). +++, 1.5- to 3-fold-higher sensitivity to inhibition than the wild type; ++, wild-type sensitivity to neutralization; +, 2- to 13-fold less sensitivity to inhibition than the wild type; +/−, less than 25% of infectivity inhibition. ND, not determined.

The aa502-520 segment is not a fusion peptide. When we initiated this study, this sequence had been proposed as a putative fusion peptide (5). However, the absence of effects of detergents and pH on the conformation of the corresponding synthetic peptide (this study) together with the absence of membranotropic properties for this region (45) does not support this hypothesis. Moreover, none of our mutants could confirm a direct role for the aa502-520 segment in membrane fusion. Indeed, our mutations affected either virus assembly or interactions between the virion and HCV entry factors CD81 and SRB1. Furthermore, the crystal structure of the E2 ectodomain was very recently determined (8, 9), and the data indicate that the aa502-520 segment includes two beta strands connected by a turn that are essential for the β-sandwich scaffold located within the E2 hydrophobic core and therefore unlikely to serve as a fusion peptide. Finally, structural analyses at low pH indicate that E2 does not undergo structural rearrangement at low pH, indicating that it is unlikely to play a direct role in membrane fusion (8, 9). It is therefore likely that the fusion peptide is located somewhere else in E2, as previously suggested (46), or, as proposed for the pestiviruses, that E1 is the fusion protein (47–49).

Residues Y507 and V515 contribute to E2 interaction with CD81. Indeed, the Y507L and V515A mutants were no longer neutralized by the CD81 LEL, and E2 was not recognized by the CD81 LEL in a GST pulldown assay (Fig. 4A). However, these mutants were only attenuated in their infectivity, suggesting that CD81-E2 interaction might not be essential for these mutants, but data on inhibition with anti-CD81 MAb JS81 and CD81 silencing indicate that the Y507L and V515A viruses remain dependent on this receptor for HCV entry (Fig. 5 and 6). The major difference between these assays is the use of a truncated LEL form of CD81 in some of our experiments, and one cannot exclude a contribution of other regions of CD81 for interaction with E2. Finally, the contribution of Y507 and V515 to E2 interaction with CD81 is in agreement with the localization of these residues close to the identified CD81 binding site (Fig. 9A), and it suggests a role for these amino acids in CD81 binding.

Residue V514 can also contribute to E2 interaction with CD81. Indeed, the V514A mutant showed a reduced sensitivity to inhibition by CD81-specific MAb JS81 and an increased sensitivity to inhibition by the CD81 LEL. These observations could be explained by an increased affinity of the V514A virus for CD81. Thus, this mutant would be able to more effectively compete with anti-CD81 MAb JS81 for binding to cell surface-expressed CD81. Interestingly, similar observations have been reported for the G451R mutant (36) as well as the murine CD81-adapted HCV mutant Jc1/mCD81 (50). However, concerning G451R, it has to be noted that this residue is not in the vicinity of V514 but is rather located on the opposite side of E2 (Fig. 9). Similarly to V514A, the G451R and Jc1/mCD81 variants also exhibited a reduced dependence on SRB1, but the G451R and Jc1/mCD81 viruses were not affected in their infectivities, whereas V514A exhibited a 10-fold decrease in infectivity. This difference could be explained by the fact that the Jc1/mCD81 and G451R mutants were selected in cell culture, whereas the V514A mutation was introduced by site-directed mutagenesis in a reverse genetics approach. Finally, the contribution of V514 to E2 interaction with CD81 is also in agreement with the localization of this residue close to the identified CD81 binding site.

The V514A mutant shows an increased sensitivity to antibody neutralization. This phenotype is again similar to what was observed for the G451R and Jc1/mCD81 mutants (36, 50). However, in contrast to the G451R mutant, this phenotype is not associated with an alteration in the V514A particle density. Again, in the case of G451R, the difference could be linked to the positions of residues G451 and V514 on the opposite side of the protein. In the case of G451R, higher-density particles are more infectious, whereas low-density particles are less infectious (36, 50), and the change in the lipid content of HCV particles has been suggested to explain the lower dependence of this mutant on SRB1 for entry. The observation that the V514A mutation reduces SRB1 dependence without affecting viral particle density suggests that other parameters might influence the interaction of the particle with this receptor, and the same is true for the Y507L mutant virus.

The high degree of conservation of the aa502-520 segment is in agreement with its role in E2 structure and function. Indeed, our mutations targeting amino acid residues buried in the E2 structure (C503S, C508S, F509L, and P511A) (Fig. 9A) were deleterious for HCV infectivity. None of these mutants was recognized by the CD81 LEL, indicating a defect in E1E2 folding, which could explain their defect in assembly and/or infectivity. Furthermore, the effects of the C503S and C508S mutations on HCV infectivity are in agreement with previous data from McCaffrey and collaborators showing a loss of infectivity for HCV pseudoparticles harboring these mutations (51). Together with the recently reported E2 structure (8, 9), these data indicate that disulfide bonds 429-503 and 508-552 are essential for HCV glycoprotein folding, virus assembly, and infectivity. It has to be noted that the formation of a disulfide bond between residues 503 and 508, as observed by Krey et al. (5), would lead to a different folding of the E2 protein, and one cannot exclude that such a disulfide bond could form transiently.

The lack of alteration in the phenotypes of the T510A, T510S, and P513A mutants is in agreement with their localization at the surface of E2 protein (Fig. 9A). Furthermore, the absence of effects of these mutations on virus assembly also suggests that these residues are not involved in interactions with HCV glycoprotein E1, and the lack of effect on virus entry and interaction with CD81 is in agreement with their position away from the CD81 binding site.

Although the aa502-520 segment is located in the beta sandwich structure of E2, its sequence overlaps with a novel conserved neutralizing epitope located within the aa 496-515 region (7). This epitope was identified after immunization of mice and chimpanzees with HCV glycoproteins. However, antibodies targeting this epitope are poorly induced during natural infection in patients and chimpanzees (7), suggesting that it is poorly presented to the immune system in the context of the viral particle. This is in line with the fact that this epitope is partially inaccessible on the structure of E2 (Fig. 9B). Thus, it seems that virus- and cell-associated envelope proteins expose the aa502-520 region differently. The fact that residues present in the aa496-515 segment are close to CD81 binding region and are also involved in CD81 binding provides a further element to understand the mechanism of neutralization by antibodies targeting this region. It is indeed very likely that such antibodies would inhibit E2-CD81 interaction. However, the surface exposure of this neutralizing epitope would need conformational changes around the beta sandwich in E2, which could potentially occur at some point of the entry process.

In conclusion, we show that, besides its essential structural role in the E2 core, the highly conserved E2 aa502-520 segment plays a key role in cell entry by modulating the interplay between particle association with coreceptors and neutralizing antibodies. Furthermore, the analyses of the aa502-520 sequence also suggest that alternative conformations could exist in E2, which would explain, for instance, the presence of a neutralizing epitope in this segment. The interaction of E2 with E1 might indeed potentially affect the conformation of E2 (52). Further structural studies on the E1E2 complex will therefore be required to better understand how these different elements interact to modulate the cellular entry process of HCV.