Amino Acid Residue-Specific Neutralization and Nonneutralization of Hepatitis C Virus by Monoclonal Antibodies to the E2 Protein

Abstract

Antibodies to epitopes in the E2 protein of hepatitis C virus (HCV) reduce the viral infectivity in vivo and in vitro. However, the virus can persist in patients in the presence of neutralizing antibodies. In this study, we generated a panel of monoclonal antibodies that bound specifically to the region between residues 427 and 446 of the E2 protein of HCV genotype 1a, and we examined their capacity to neutralize HCV in a cell culture system. Of the four monoclonal antibodies described here, two were able to neutralize the virus in a genotype 1a-specific manner. The other two failed to neutralize the virus. Moreover, one of the nonneutralizing antibodies could interfere with the neutralizing activity of a chimpanzee polyclonal antibody at E2 residues 412 to 426, as it did with an HCV-specific immune globulin preparation, which was derived from the pooled plasma of chronic hepatitis C patients. Mapping the epitope-paratope contact interfaces revealed that these functionally distinct antibodies shared binding specificity for key amino acid residues, including W⁴³⁷, L⁴³⁸, L⁴⁴¹, and F⁴⁴², within the same epitope of the E2 protein. These data suggest that the effectiveness of antibody-mediated neutralization of HCV could be deduced from the interplay between an antibody and a specific set of amino acid residues. Further understanding of the molecular mechanisms of antibody-mediated neutralization and nonneutralization should provide insights for designing a vaccine to control HCV infection in vivo.

INTRODUCTION

Hepatitis C virus (HCV) infects about 180 million people worldwide; HCV infection may lead to chronic liver disease, liver failure, and hepatocellular carcinoma (12, 27). Although acute hepatitis C is typically mild or even subclinical, the infection becomes chronic in more than 75% of those infected (16, 27). The standard antiviral treatment based on pegylated interferon plus ribavirin is successful in ca. 50% of treated patients. While two protease inhibitors have recently been licensed and newer direct-acting antiviral drugs are in the pipeline that will improve the cure rate, antiviral therapy will still carry significant side effects (2). In most parts of the world, such treatments are not economically or logistically feasible; therefore, vaccine development remains an important goal.

At this time, we still do not know with certainty what type of immune response should be induced by a vaccine to effectively prevent HCV infections or at least prevent chronic infections. To the best of our current knowledge, it does not appear that any of the proposed vaccines could induce sterilizing immunity. We have long known that recovered patients can often be reinfected (1, 14) and that experimentally infected chimpanzees that have cleared the infection in the acute stage can be reinfected even with the same strain of virus (4, 6). Experimental vaccines designed either to induce antibodies to the envelope glycoproteins or to induce T lymphocyte responses to nonstructural viral proteins have been developed. T cell vaccines have had very mixed results in animal studies (3, 6, 13, 19, 24, 32), whereas neutralizing antibody-based vaccines appear to be somewhat more promising (21, 28). However, the role played by neutralizing antibodies in the clearance of acute infections is not fully understood (7, 11, 20). Nonetheless, eliciting neutralizing antibodies should be explored as at least an important part of successful vaccine development.

Previously, we demonstrated the existence of neutralizing, nonneutralizing, and interfering antibodies in the plasma samples of chronically infected hepatitis C patients and vaccinated chimpanzees (29, 30, 31). We also mapped two antigenic epitopes in the E2 protein, namely, epitope I (amino acids [aa] 412 to 426) and epitope II (aa 427 to 446) (Fig. 1A). Epitope I-specific neutralization of HCV was detected in vitro with enriched preparations of polyclonal antibodies derived from experimental HCV-specific immune globulins (HCIGIV), whereas neutralization was not observed with antibodies specific for epitope II. In addition, these nonneutralizing antibodies were shown to interfere with the neutralization by epitope I-specific antibodies. Depletion of these interfering antibodies revealed a broader cross-genotype neutralization by the epitope I-specific antibodies (30, 31).

Fig 1.

Peptide specificity and binding affinity of the MAbs. (A) Peptides used for the present study. Peptide A corresponding to amino acid residues 412 to 447 of the E2 protein of HCV H strain (H77) was used to immunize mice for generating the MAbs tested in the present study. The truncated forms of peptide A, i.e., peptide B, B short, and peptide D, are indicated. The locations of epitope I and epitope II within peptide A, as mapped in our previous studies (30, 31), are shown. (B) Peptide A specificity of the MAbs in an ELISA. Biotin-conjugated peptide A was added to streptavidin-coated 96-well plates (200 ng/well). Each MAb (ascites fluid) was diluted 1:1,000 and used as the primary antibody. The y axis indicates the absorbance at 405 nm obtained in the ELISA, representing specific binding of a given antibody to peptide A. The data shown represent three independent experiments. (C) Peptide B specificity of the MAbs in an ELISA. Instead of peptide A, biotin-conjugated peptide B was used for the ELISA as described in panel B. (D) The affinities of B-specific antibodies to peptide A and B were determined by an ELISA using sodium thiocyanate (NaSCN) as a chaotropic agent as described in reference 22. An ELISA was performed with 1:20,000-diluted ascites fluid in the presence or absence of various concentrations of NaSCN as indicated. The specific binding affinity was calculated based on the values obtained with or without NaSCN. The data shown represent three independent experiments. Error bars represent the standard deviation.

In the present study, we generated a panel of monoclonal antibodies (MAbs) in an attempt to represent the critical aspects of antibody-mediated neutralization and nonneutralization (interference) of HCV described previously with polyclonal antibody preparations. In particular, our studies focused on MAbs that bound specifically to epitope II, i.e., the peptide B region of the E2 protein of HCV genotype 1a (H77 strain) (Fig. 1A). We examined these antibodies for their capacity to neutralize HCV or to interfere with neutralization in vitro and investigated further the interactions of these MAbs with their respective binding sites at the amino acid level. Here, we describe the characteristics of four epitope II-specific antibodies differing in their requirement of amino acid residues for binding and their ability to neutralize the virus.

MATERIALS AND METHODS

Peptide synthesis.

All peptides were chemically synthesized by the Core Laboratory of the Center for Biologics Evaluation and Research at the U.S. Food and Drug Administration, with an Applied Biosystems (Foster City, CA) model 433A peptide synthesizer. Biotinylated peptides were synthesized with Fmoc-Lys (Biotin-LC)-Wang resin (AnaSpec, San Jose, CA) as described previously (30).

Generation of MAbs.

MAbs were produced by Harlan Bioproducts for Science (Indianapolis, IN) according to their standard procedures for generating MAbs. Briefly, BALB/c mice were injected intraperitoneally (i.p.) with a chemically synthesized peptide A containing amino acid residues 412 to 447 of the E2 protein from the HCV genotype 1a strain H77 (GenBank accession no. M67463), which was conjugated to keyhole limpet hemocyanin. Mice that produced high titers of antibody to peptide A were selected for cell fusion to generate the hybridomas (Fig. 1A and B). Antibody-positive cells were cloned by the limiting dilution method for several cycles. At each cloning cycle, the tissue culture supernatant of each clone was screened by enzyme-linked immunosorbent assay (ELISA) for the presence of antibodies to peptide B (Fig. 1C), which represented epitope II from our previous studies (30, 31). The selected anti-peptide B-positive clones were injected i.p. into BALB/c mice primed with Pristane (Sigma-Aldrich, St. Louis, MO) to produce ascites fluid.

ELISA.

Biotin-conjugated peptide (200 ng/well) was added to streptavidin-coated 96-well Maxisorp plates (Thermo Fisher Scientific, Rockford, IL), followed by incubation at room temperature for 1 h in Super Block blocking buffer (Thermo Scientific). The wells were blocked further in blocking buffer for another hour at 37°C. After the plate was washed four times with phosphate-buffered saline (PBS; pH 7.4) containing 0.05% Tween 20 to remove unbound peptides, serial dilutions of the test antibodies were added to the plate, followed by incubation at 37°C for 1 h. The plate was then washed four times before the secondary antibody, either a goat anti-mouse peroxidase-conjugated IgG or a goat anti-human peroxidase-conjugated IgG (Sigma-Aldrich) at a 1:5,000 dilution, was added to the wells, followed by incubation at 37°C for 1 h. After four washes, the reaction was developed with ABTS [2,2′azinobis(3-ethylbenzthiazolinesulfonic acid)] peroxidase substrate (KPL, Gaithersburg, MD) and stopped by the addition of 100 μl of a 1% sodium dodecyl sulfate (SDS) solution. The absorbance of each well was measured at 405 nm. Alternatively, the reaction was developed with 1-Step TMB-ELISA substrate solution (KPL) and stopped by adding 100 μl of 4 N sulfuric acid. The absorbance of each well was measured at 450 nm, using a SpectraMax M2e microplate reader (Molecular Devices, Sunnyvale, CA).

The binding affinity of each of these MAbs to a specific antigen was measured by an ELISA with sodium thiocyanate (NaSCN) as a chaotropic agent according to the procedure described previously (22).

Neutralization assay.

Virus stocks were prepared by transfecting full-length HCV RNA derived from an HCV genotype 2a clone, J6/JFH1 (a gift from Charles Rice, Rockefeller University, New York, NY), into Huh 7.5 cells as previously described (10, 30, 31). An HCV genotype 1a/2a chimera virus was produced by replacing the structural genes of J6/JFH1 with those of the HCV H77 strain, which is known to be genotype 1a. Briefly, Huh 7.5 cells were seeded at a density of 4 × 10³ to 5 × 10³ cells/well in 96-well plates to obtain ca. 60% confluence in 24 h. The virus stock was diluted in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum–1% penicillin-streptomycin–2 mM glutamine to yield ∼50 infected foci per well in the absence of antibodies. Viruses were mixed with a diluted antibody or with cell culture medium, incubated at 37°C for 1 h, and then inoculated into Huh 7.5 cells. After 3 days in culture, virus foci were detected either by immunofluorescence or immunoperoxidase staining and then counted. Neutralization was determined by comparing the infectivity of the viruses incubated with the antibody to the infectivity of the viruses incubated with medium alone or with preimmune plasma. The median 50% inhibitory dilution (ID₅₀) and the median 50% inhibitory concentration (IC₅₀; μg/ml) were determined according to the method of Reed and Muench (23).

Enrichment and removal of peptide-specific antibodies.

A total of 500 ng of biotinylated peptide B, peptide D, or an unrelated peptide control (a pool of overlapping peptides representing the M2 protein from the influenza virus) was mixed with 100 μl of streptavidin-coated Dynabeads (Life Technologies, Grand Island, NY), followed by incubation at room temperature for 1 h. After a washing step with PBS (pH 7.4), the beads were mixed with an appropriate dilution of ascites fluid or plasma, which contained specific antibodies, followed by incubation at room temperature for 1 h. To obtain a solution enriched in peptide-specific antibodies, the beads were collected with a magnet stand. After the beads were washed with PBS, the antibodies were eluted with 50 mM glycine-HCl solution (pH 2.2). The eluates were neutralized by mixing with an equal volume of 1 M Tris-HCl buffer (pH 9.2). In contrast, to remove the peptide-specific antibodies, the beads were pelleted with a magnet stand, and the supernatant was collected for further analysis.

Phage display.

The selection of peptides from random peptide phage display libraries (New England BioLabs, Ipswich, MA) was described previously (30). Briefly, 10¹⁰ phages were incubated with individual MAb-protein G mixtures at room temperature for 20 min. After eight washes with 0.05 M Tris-HCl buffer (pH 7.5) containing 0.15 M NaCl and 0.05% Tween 20, the phages were eluted from the complexes with 0.1 M HCl and neutralized with 1 M Tris-HCl buffer (pH 9.0). The eluted phages were then amplified in the host strain ER2738 for 4 to 5 h. After three additional rounds of selection of amplified phages by the same MAb, the DNA from each single-phage plaque was sequenced, and the corresponding peptide sequence was then deduced from the DNA sequence.

Statistical analysis.

Statistical analysis was performed with GraphPad Prism 4 (GraphPad Software, La Jolla, CA) using an unpaired Student t test with a two-tailed P value (P < 0.05). Error bars represent the standard deviation or the standard error of the mean.

RESULTS

Neutralization and nonneutralization of HCV 1a/2a chimeras by MAbs.

A panel of hybridoma cell lines that produced monoclonal antibodies was obtained after the mice were immunized with peptide A (aa 412 to 447) (Fig. 1A). Using an ELISA, we identified seven MAbs (contained in ascites fluid) from the panel that were capable of binding peptide A and found that their binding capacities to peptide A were comparable at a dilution of 1:1,000 (Fig. 1B). Four of the seven antibodies, namely, antibodies 8, 12, 41, and 50, were specifically reactive with peptide B, which corresponds to residues 427 to 446 of the E2 protein (Fig. 1A and C). The other three antibodies—27, 48, and 49—did not bind peptide B (Fig. 1C).

We determined the relative affinity of peptide B-specific MAbs to peptides A and B (Fig. 1D and E). The data showed in both cases that antibody 41 had the highest affinity among these antibodies since it did not show any reduction of the binding even in the presence of 4 M NaSCN. Antibody 8 had a higher affinity than both antibodies 12 and 50. The order of the affinity toward both peptides was thus: antibody 41 > antibody 8 > antibody 50 > antibody 12.

The identification of this group of MAbs that were capable of binding the peptide B region of the E2 protein prompted us to test their capacity to neutralize HCV in vitro or to interfere with neutralization by antibodies directed at epitope I. We used a genotype 1a/2a chimeric virus for the neutralization assay given that the peptide sequence of the genotype 1a virus corresponded to that of peptide A, which was used to generate these MAbs. Antibodies 8 and 41 were able to neutralize the genotype 1a/2a virus with IC₅₀s of 65.14 and 0.74 μg/ml, respectively. In contrast, antibodies 12 and 50, which also recognized peptide B in an ELISA, were unable to neutralize the virus (Fig. 2A). The neutralization ability appeared to be correlated with overall binding affinity of these MAbs to peptide B.

Fig 2.

Neutralization of chimeric viruses by MAbs in Huh 7.5 cells. (A) Neutralization of genotype 1a/2a virus. Each antibody at the concentrations of 0.1, 1, 10, and 100 μg/ml was incubated with the appropriately diluted genotype 1a/2a virus before the mixture was added to Huh 7.5 cells. Cell culture medium (DMEM) was used as a negative control for the antibody. The x axis indicates the particular antibody tested. The y axis indicates the relative infectivity of the virus (%), i.e., the percentage of the negative control. Error bars represent the standard error of the mean. (B) Peptide B-specific neutralization of genotype 1a/2a virus by antibody 41. Antibody 41 was adsorbed with (+) or without (−) peptide B prior to performing an ELISA to test its binding to peptide B (left panel) and a neutralization assay to assess its neutralizing activity in Huh 7.5 cells (right panel). Each of these samples shown on the x axis was tested at the dilution of 1:10⁵ in an ELISA. The y axis indicates the absorbance at 405 nm, representing the specific binding of the antibody to peptide B. The data shown represent at least three independent experiments. Error bars represent the standard deviation. For the neutralization assay (right panel), the supernatant was diluted at 1:400, followed by incubation with the genotype 1a/2a virus before the mixture was added to Huh 7.5 cells. The cell culture medium (Med) was used as the negative control in place of the tested antibodies. The x axis indicates the samples tested. The y axis indicates the relative infectivity of the virus (%), i.e., the percentage of the negative control. The data shown represent the results from three independent experiments. The error bars represent the standard error of the mean. The statistical significance of the difference in infectivity is indicated. (C) Inability of the antibodies to cross-neutralize the J6/JFH1 virus. The antibodies were tested for their abilities to neutralize J6/JFH1, a genotype 2a virus, in Huh 7.5 cells with the procedure described in panel A. The data shown represent three independent experiments. The error bars represent the standard error of the mean.

To ascertain whether the observed neutralization of the 1a/2a chimeric virus was peptide B specific, the neutralization assay was repeated after depleting the peptide B-specific binding activity from the ascites fluid that contained neutralizing antibody 41. As demonstrated by the ELISA results shown in Fig. 2B (left panel), the peptide B-specific binding activity was substantially depleted by the adsorption. Concurrently, its neutralizing activity was also significantly diminished (Fig. 2B, right panel).

We then examined the cross-neutralizing activity of antibodies 8 and 41, and found that these MAbs, along with the nonneutralizing antibodies 12 and 50, were not able to neutralize the genotype 2a virus, J6/JFH1 (Fig. 2C), and other chimeric viruses 1b/2a and 3a/2a (data not shown). These results demonstrated that antibodies 12 and 50 had no detectable neutralizing activity against any of the genotypes tested and that antibodies 8 and 41, through the direct binding of peptide B, could only neutralize the genotype 1a virus.

Interference of peptide D-specific neutralization of HCV by nonneutralizing MAbs.

In our previous studies (30, 31), we identified two epitopes, epitope I and epitope II, within peptide A of the E2 protein (Fig. 1A), which includes the sequences of peptide D (aa 412 to 426) and peptide B (aa 427 to 446), respectively. We also observed that the peptide B-specific antibodies enriched from experimental HCIGIV preparations were not able to neutralize HCV. Moreover, these nonneutralizing antibodies could reduce the neutralizing activity of the polyclonal antibodies that bound peptide D, i.e., epitope I, one of highly conserved epitopes in HCV. Given that both nonneutralizing antibodies 12 and 50 were found to be peptide B specific, we tested whether these nonneutralizing MAbs had interfering activity. For these studies, the plasma collected from an immunized chimpanzee, Ch1587, was selected as the source of peptide D-specific neutralizing antibodies because this plasma had been shown to neutralize HCV in vitro and that its antibodies bound specifically to peptide D (21, 30, 31). The ID₅₀ of Ch1587 plasma was found to be 1:400, i.e., at which dilution the plasma could reduce the virus infectivity to ca. 50% of the negative control (cell culture medium alone) (Fig. 3A). To test for interference, 1 volume of Ch1587 plasma at the ID₅₀ was initially mixed with 1 volume (1:1 [vol/vol]) or 4 volumes of antibody 12 (1:4 [vol/vol]) and then added to genotype 1a/2a virus in a total assay volume of 100 μl before the whole mixture was added to Huh 7.5 cells. Antibody 12 at the ratio of 1:4 completely erased the previously observed neutralizing effects of Ch1587 plasma

Fig 3.

Reduction of virus neutralizing activity of Ch1587 plasma and D-eluate by nonneutralizing antibody 12. Ch1587 plasma was prepared at the ID₅₀ (1:400 dilution) and used as a positive control antibody. The plasma was mixed with an equal (1:1) or 4-fold volume (1:4) of antibody 12 (A) or antibody 50 (B) and incubated with genotype 1a/2a virus in a final assay volume of 100 μl at 37°C for 1 h, after which, the mixture was added to the Huh 7.5 cells. The cell culture medium (Med) was used as the negative control. (C) Ch1587 plasma was affinity purified by using biotinylated peptide D. The resulting D-eluate was confirmed in the ELISA for its specific binding to peptide D. (D) The IC₅₀ of the D-eluate was predetermined to be 0.348 μg/ml. At the IC₅₀, the D-eluate was mixed with various concentrations of nonneutralizing MAb 12 (D) or with MAb 50 (E). The x axis indicates the samples used in this assay. The relative infectivity, i.e., the percentage of the negative control indicated by the y axis, was calculated as described in panel A. The P values were determined for the differences in the infectivity (*, P < 0.05; **, P < 0.01). The data shown represent at least three independent experiments. The error bars represent the standard errors of the mean.

(P < 0.05) (Fig. 3A). In contrast, the nonneutralizing antibody 50 did not reduce the neutralizing activity of Ch1587 plasma under the same experimental conditions (Fig. 3B).

To confirm that the observed interference was peptide D specific, we further purified the peptide D-specific antibodies from Ch1587 plasma, namely, the D-eluate. We confirmed in an ELISA that the D-eluate was peptide D specific (Fig. 3C). The IC₅₀ of the D-eluate for 1a/2a virus was determined to be 0.348 μg/ml. To test the interference, we mixed the D-eluate at the IC₅₀ (0.348 μg/ml) with nonneutralizing antibody 12 or 50 at various amounts ranging from 0 to 120 μg/ml (Fig. 3D and E). Peptide D-specific neutralization was significantly diminished in the presence of antibody 12 at 30, 60, and 120 μg/ml (P < 0.05, P < 0.01 and P < 0.01, respectively). In contrast, nonneutralizing MAb 50 did not show any interference under the same experimental conditions. These data confirmed that MAb 12, through the binding to the peptide B region, could interfere with the antibody-mediated neutralization occurring at the peptide D region in a manner consistent with our previous observations in patients (30, 31).

To establish the clinical relevance of the observed interference, we further examined whether these MAbs could affect the virus neutralization by the antibodies in the HCIGIV, which was made solely from pooled plasma of chronically HCV-infected patients (29). Under the experimental conditions similar to that for the analysis of D-eluate of chimpanzee plasma, the ability of HCIGIV to neutralize 1a/2a virus was tested in the presence or absence of nonneutralizing MAb 12 or 50. As shown in Fig. 4A, the neutralizing activity of HCIGIV was inhibited by MAb 12 in a dose-dependent manner. At 60 μg/ml, the interfering effect was clearly detected (P < 0.05). In contrast, MAb 50 did not reduce the neutralizing ability of HCIGIV (Fig. 4B). In addition, in a separate set of experiments, we found that neither antibody 12 nor antibody 50 showed any interfering effect on the neutralizing MAb 41 (Fig. 4C) or MAb 8 (data not shown).

Fig 4.

Effects of nonneutralizing antibodies on virus neutralization by HCIGIV and antibody 41. HCIGIV at the ID₅₀ (1:3,000 dilution) was used to neutralize genotype 1a/2a virus in the presence or absence of different concentrations of antibody 12 (A) or antibody 50 (B). The cell culture medium (Med) was used as the negative control. (C) The ability of antibody 41 to neutralize genotype 1a/2a virus was measured in the presence or absence of nonneutralizing antibody 12 or antibody 50 according to the procedure described in Fig. 3. Antibody 41 was diluted at 1:400 in DMEM, and then mixed with antibody 12 or antibody 50 with the ratios of 1:1 and 1:4 (vol/vol). The antibody mixture was subsequently incubated with genotype 1a/2a chimeric virus at 37°C for 1 h before being added to Huh 7.5 cells. The data shown represent three independent experiments, and the standard errors of the mean are indicated by error bars. The x axis indicates the samples used in this assay. The y axis indicates the relative infectivity of the virus (%).

Residue-specific binding of neutralizing and nonneutralizing antibodies.

The key residues involved in the binding of the neutralizing and nonneutralizing antibodies were mapped by screening random peptide phage display libraries (Fig. 5). All of the antibodies tested in the present study, regardless of their neutralizing activity, selected preferentially the peptides that contained W and L residues that were arranged in a linear consensus sequence of WLxxL, albeit a few (primarily hydrophobic) substitutions were tolerated. From these data, we predicted that W⁴³⁷, L⁴³⁸, or L⁴⁴¹ within peptide B were the direct contact points for the binding of these antibodies. Interestingly, the L⁴³⁸ was sometimes replaced by a P residue without affecting the binding capacity of either neutralizing antibody 8 or nonneutralizing antibody 50, which was presumably associated with the presentation of these peptides on the phage. We also observed that the L⁴⁴¹ could be replaced by various amino acids, as seen in the peptides selected by neutralizing antibody 41. This variability, however, was not apparent in the peptides selected by the nonneutralizing antibodies, raising the possibility that the presence of L⁴⁴¹ may be more related to the binding of nonneutralizing antibodies rather than that of neutralizing antibodies. In addition, one of the nonneutralizing antibodies, MAb 50, tended to select preferentially the peptides that contained an additional W residue at the position corresponding to F⁴⁴², whereas the other antibodies did not show such a preference, indicating that the residue specificity was different between nonneutralizing antibody 50 and the rest of the antibodies.

Fig 5.

Epitope mapping by screening random peptide phage display libraries. The amino acid sequences of phage clusters identified after at least three rounds of screening phage-display libraries (12-mer and 7-mer) with neutralizing antibodies 8 and 41 and nonneutralizing antibodies 12 and 50 are listed. The numbers of peptides identified over the total numbers of peptides sequenced are shown in parenthesis. The candidate core residues at the epitope-paratope contact interfaces are indicated in bold font. The letter “X” denotes any amino acid residue other than L at that position.

To further determine the residue specificity of these antibodies, each of the key binding residues predicted by the phage display analysis was replaced, one at a time, by an alanine residue (Fig. 6A) in a truncated version of peptide B termed “B short” (aa 434 to 446) and then tested by ELISA to determine the effect of the specific substitution on the binding of these antibodies (Fig. 6B). We found that all of the antibodies tested in this study, irrespective of their function, lost their binding to B short when W⁴³⁷ or L⁴³⁸ was replaced by an A residue, thus confirming that W⁴³⁷ and L⁴³⁸ were the “core” residues that were commonly recognized by these antibodies. As predicted by the phage display analysis, neutralizing antibodies 8 and 41 were indeed less affected by the substitution at L⁴⁴¹ in contrast to its effect on the binding capacities of the nonneutralizing antibodies 12 and 50. The alanine substitution of F⁴⁴² reduced the binding of nonneutralizing antibody 50 (P < 0.05), while F⁴⁴² appeared to be an insignificant residue for neutralizing antibodies 8 and 41 and nonneutralizing antibody 12 (P > 0.05). These data were consistent with the observations made in the phage display analysis and allowed us to differentiate which set of residues within peptide B were critical for the neutralizing antibodies or for the nonneutralizing antibodies, including antibody 12, which confer an interfering function (Fig. 7A).

Fig 6.

Use of mutational analysis to identify the residues that are critical for antibody recognition. (A) Biotin-conjugated peptides were chemically synthesized to represent the truncated peptide B, i.e., B short, from the E2 protein of HCV genotype 1a (H77 strain) and its mutations. The B short mutant peptides contained a single alanine substitution at positions 437, 438, 440, 441, and 442, respectively. A hyphen indicates an amino acid residue identical to that of the H77 sequence. (B) Biotin-conjugated B short peptide and its mutants were added to streptavidin-coated 96-well plates at 200 ng/well in an ELISA. Each MAb (ascites fluid) was diluted 1:10⁵ dilution, and applied as the primary antibody. The data shown represent at least three independent experiments. The x axis indicates the antibodies used in the assay. The y axis indicates the absorbance at 405 nm, representing specific binding of a given antibody to each individual peptide.

Fig 7.

Summary of the amino acid residues critical for antibody functions. (A) Key residues identified in peptide B that are critical for the binding property of each antibody. (B) Alignment of amino acid sequences of the E2 region from aa 412 to 446 of various HCV genotypes. Key residues identified in the present study are indicated in boldface underlined letters. A hyphen indicates an amino acid residue identical to that of the H77 sequence.

HCV genotype 1a has a W⁴³⁷ in its E2 protein, whereas the other HCV genotypes often contain an F residue at the same position (Fig. 7B) (25). We hypothesized that the genotype 1a-specific neutralizing antibodies are limited to one genotype because of their inability to recognize F⁴³⁷. To test this hypothesis, we switched W⁴³⁷ to an F residue in B short (Fig. 8A). The W437F switch resulted in a significant loss of binding in an ELISA by neutralizing antibodies 41 and 8 to B short (Fig. 8B). This result confirmed that the presence of W⁴³⁷ was required for the shared recognition by these antibodies and further suggested that residue W⁴³⁷ is an important attributing factor to the observed genotype 1a-specific neutralization of HCV. Of interest was the observation that the W437F switch also resulted in a loss of binding by antibodies 12 and 50 to B short, a finding consistent with their inability to neutralize any of the genotypes tested.

Fig 8.

Effect of the W437F switch on antibody binding. (A) Schematic representation of the peptide mutations used in the ELISA. Biotin-conjugated peptides were chemically synthesized to represent peptide B, the truncated peptide B (B short), and B short with specific single mutations at position 437. A hyphen indicates an amino acid residue identical to that of the H77 sequence. (B) Detection of the effect of the W437F switch on antibody binding in an ELISA. Biotin-conjugated peptide B, B short peptide, and its mutants were added to streptavidin-coated 96-well plates at 200 ng/well. Each antibody (in ascites fluid) was used at a 1:10⁵ dilution as the primary antibody in the ELISA. Cell culture medium was used as the negative control of the antibody. The x axis indicates the antibodies used in this assay. The y axis indicates the absorbance obtained at 450 nm, which represents the measurement of specific binding of a given antibody to each individual peptide. The ELISA results of the W437A mutant here were comparable to the ELISA results obtained separately in Fig. 6B.

DISCUSSION

Although HCV was identified more than 20 years ago, there are still no licensed vaccines available to prevent this virus from infecting humans. Because of the high variability of the virus and the limited number of animal models to study HCV infection, vaccine development remains a challenge. Theoretically, the exposed epitopes on the surface of the HCV E2 protein should be the best candidates for vaccine design, since these antigens are most likely accessible to neutralizing antibodies. Unfortunately, the role of such antibodies in clearance and protection is still unknown. Patients who clear acute HCV infections have little or no neutralizing antibodies present in their serum at the time of viral clearance. On the other hand, patients with chronic infections have high levels of neutralizing antibodies (18).

Our previous studies showed the existence of neutralizing, nonneutralizing, and interfering antibodies in patients with chronic HCV infection and in vaccinated chimpanzees (30, 31). However, the complexity of polyclonal antibodies makes it difficult to directly characterize the phenomenon of antibody-mediated neutralization and nonneutralization. Instead, we addressed the key aspects of this phenomenon by studying a panel of MAbs that bind specifically to the peptide B region (epitope II). We demonstrated that both neutralizing and nonneutralizing antibodies that are specific for the peptide B region could be generated by immunizing mice with an HCV E2 peptide consisting of amino acid residues 412 to 447. Amino acid residues W⁴³⁷ and L⁴³⁸ were mapped to be the core sequence critical for the binding of both neutralizing and nonneutralizing antibodies. However, the neutralization of HCV obtained with our approach turned out to be genotype 1a virus specific. The genotype 1a specificity was found to be linked, if not limited, to the neutralizing antibodies' recognition of the W⁴³⁷ residue. This conclusion is supported by our observations that non-genotype 1a viruses contain an F residue at position 437 and that simply replacing W⁴³⁷ by F was sufficient to severely reduce the binding of neutralizing antibodies 8 and 41 to peptide B. Intriguingly, the ⁴³⁶GWLAGLFY motif of the E2 protein, where the W⁴³⁷ and L⁴³⁸ reside, has been shown to be involved in the CD81-mediated entry of virus (8, 9). In addition, a broadly neutralizing antibody was recently reported to recognize a discontinuous conformational epitope composed of amino acid residues 396 to 424, 436 to 447, and 523 to 540 of the E2 protein (17). Noticeably, the second segment of this conformational epitope mapped to the same region as peptide B in our study. Furthermore, after observing that this segment had a high degree of amino acid sequence variability, Lapierre et al. pointed out that the broadly neutralizing activity of the antibody could have resulted from its recognition of structural determinants rather than the specific residues in this conformational epitope (15). These studies suggest the importance of the spatial arrangement of amino acids within an epitope in defining the ability of an antibody to bind and to block CD81-mediated virus entry. Further structural determination of whether the genotype-specific neutralization described here is due to the disruption of the association between the core residues, such as W⁴³⁷ and L⁴³⁸, and CD81 by MAbs 8 and 41 should shed light on the generation of antibodies that can broadly neutralize across HCV genotypes.

In addition to the residue specificity of these MAbs, we observed differences in affinity among these MAbs to peptide B. The antibody affinity appeared to be consistent with the ability to neutralize the virus. Apparently, further characterization of these MAbs is needed to determine the precise role of their binding affinities at peptide B region in neutralizing the virus.

The identification of peptide B-specific neutralizing MAbs described here suggests that this subset of neutralizing antibodies might also be induced during HCV infection. In line with our findings, Tarr et al. showed that affinity-enriched human immune globulin fractions that target epitopes overlapping the aa 434 to 446 region of E2 can exhibit neutralization of pseudoparticles and cell-cultured HCV (26).

In the production of neutralizing antibodies 8 and 41, peptide B-specific nonneutralizing antibodies 12 and 50 could also be elicited simultaneously. Furthermore, these nonneutralizing antibodies share the W⁴³⁷ and L⁴³⁸ binding residues with the neutralizing antibodies. These observations connote that the binding of an antibody to its epitope in the E2 protein is necessary but may not be sufficient for the neutralization of the virus. The finding of nonneutralizing antibodies to HCV is not unprecedented. Burioni et al. previously reported that several human monoclonal antibodies with nonneutralizing activity could be isolated from patients with chronic HCV infection. These antibodies were capable of recognizing, at least partially, the epitope II region (5). At present, the roles of these nonneutralizing antibodies in HCV infection have yet to be fully characterized.

In the present study, we identified L⁴⁴¹ in the E2 protein as the residue specific for the binding of nonneutralizing antibodies 12 and 50. In addition, the F⁴⁴² residue was found to be crucial, but only for the binding of nonneutralizing antibody 50. These findings agreed with our previous results, which showed that the skein ⁴⁴¹LFY⁴⁴³ was linked to the nonneutralization of the virus (30). However, caution must be taken in differentiating neutralizing antibodies from nonneutralizing antibodies solely on the basis of their residue specificity because a subtle change of residues in the antibody-antigen interface or a change in binding affinity may alter the recognition capabilities of these antibodies and thereby may modulate their functions accordingly.

Our further characterization of nonneutralizing MAb 12 revealed its ability to interfere with polyclonal neutralizing antibodies derived from immunized chimpanzees and chronically HCV-infected patients that specifically target the peptide D region (epitope I, aa 412 to 426). This result was consistent with the previous observation that interfering antibody activity could be detected in HCIGIV preparations and in the plasma of an immunized chimpanzee (30, 31). However, we have also established that not all nonneutralizing antibodies have the capacity to interfere with neutralizing antibodies.

It should be mentioned that such an interfering activity was not detected in the study by Tarr et al. In their experiments, two samples from patients with chronic HCV infection of genotypes 1a and 3a were analyzed after affinity enrichment by using a peptide that overlapped the aa 434 to 446 region of the E2 protein (26). We do not know the precise cause for these different observations. One possibility is that the relatively low binding affinity of interfering antibodies to epitope II in the samples studied, as demonstrated here with MAb 12, makes it difficult to enrich for these interfering antibodies. Another possibility is that the affinity-enriched antibodies are of multiple specificities for distinct sets of amino acids within the epitope II. The composition of these residue-specific antibodies in the samples at the time of collection (which could represent a particular stage in the course of chronic HCV infection) may also be another factor that potentially affects the chances of detecting the interfering antibodies. Given the complexity of the dynamic interplay between polyclonal antibodies and multifaceted antigenic interface during natural HCV infection and the challenge to study further the phenomenon of neutralization and interference in plasma from multiple patients or immune globulins prepared from pooled plasma samples, our identification of a peptide B-specific MAb with interfering properties provides a unique platform to address this issue.

Based on the evidence presented here, we propose two possibilities as an attempt to provide a yet-to-be-defined mechanism of antibody-mediated neutralization and nonneutralization of HCV. First, the antibodies that are induced during natural HCV infection have different biochemical and biophysical properties, even though they may exhibit similar binding specificities and/or affinities in vitro. Second, each antibody may recognize a unique local and global conformational state of the E2 protein displayed on the infectious virion as a transient intermediate during the entry process. The demonstration of a group of MAbs with neutralizing, nonneutralizing, and interfering properties, each recognizing the same linear stretch of amino acid residues in HCV E2, provides us with an opportunity to further study the molecular and structural basis for the antibody-mediated neutralization and to explore the potential of these antibodies in the prophylaxis and treatment of HCV infections.