The HCV E2 glycoprotein encodes clusters of overlapping epitopes that are highly immunogenic with evidence that the more dominant epitopes do not elicit the most broadly protective antibodies. For example, the HVR1 region displays immunodominant epitopes that are mainly targeted by isolate-specific antibodies from which the virus is able to rapidly escape 
. A second cluster of highly immunogenic epitopes, designated as antigenic domain B, contains overlapping conformational epitopes that account for the majority of the identified broadly neutralizing antibodies 
. Although antigenic domain B antibodies exhibit broad neutralization against different HCV genotypes and subtypes, it is probable that HCV is able to escape from immune pressure by the majority of these antibodies 
. A third cluster, designated as antigenic domain A, includes epitopes that induce non-neutralizing antibodies 
. It is also probable that antigenic domain A and other non-neutralizing determinants are highly immunogenic and account for a substantial portion of antibody response to E2 
. Taken together, HCV is able to divert the immune response to these highly immunogenic determinants and thereby gains a selective advantage.
We implemented a screening approach for novel antibodies that avoided these determinants. Heterologous E2 employed in the screening eliminated antibodies to HVR1. Information gained from epitope mapping of previously isolated HCV HMAbs led to the development of mutant E2 antigens that minimized the selection of non-neutralizing antigenic domain A antibodies and neutralizing antibodies to antigenic domain B. Based on broad binding patterns to different genotype and subtype E2 proteins, nine scFvs were selected for conversion to IgG1
and were further studied. Surprisingly, all nine antibodies mediated virus neutralization and neutralized both 1a and 2a HCVcc. Our earlier experiences in isolating HMAbs to HCV, using an initial screen by IFA binding to recombinant E2, yielded nearly 50% of the isolated antibodies that are non-neutralizing 
. A possible implication of this finding is that the majority of the non-neutralizing antibody response to HCV E2 is to antigenic domain A epitopes. If this proves to be the case, a vaccine candidate that avoids an antibody response to antigenic domain A could be an approach to focus the immune response to a repertoire of antibodies that eliminates at least a significant portion of non-neutralizing antibodies. As expected, all of these antibodies bound to a 1a D535A E2 mutant, which is not bound by broadly neutralizing antigenic domain B antibodies 
. The nine HC-84-related antibodies, designated as antigenic domain D, broadly neutralized different HCVcc genotypes and subtypes, and many of these antibodies have greater neutralization potency against 1a and 2a HCVcc than two of the more potent antigenic domain B antibodies, HC-1 and HC-11 
. The IC50
values of antigenic domain D antibodies are substantially lower against 2a HCVcc (JFH1) than 1a HCVcc (H77) (). Subtle differences in the presentation of antigenic domain D epitopes between these two genotypes could account for the different IC50
values. Since these antibodies are derived from B cells of an individual infected with HCV genotype 2b, it is possible that the HC-84 HMAbs are more directed at their respective epitopes as presented in genotype 2. However, it is also possible that the JFH1 isolate is more sensitive to neutralizing antibodies than the H77 isolate. We previously reported on antigenic domain B antibodies isolated from B cells of an individual infected with genotype 1b having lower IC50
values against 2a HCVcc (JFH1) than 1a HCVcc (H77) 
. Moreover, the IC50
values for HC-84.1 and -.26 against the 2a HCVcc (JFH1 and J6) isolates are significantly different. Even though their epitopes have complete sequence conservation between these two isolates, it is possible that global conformation with the variation (approximately 13%) in E2 glycoproteins of JFH1 and J6 could disturb antibody binding to their respective epitope. This, in turn, will be reflected in the observed differences in antibody-mediated neutralization. Additional studies will be required to link the amino acid(s) and precise location(s) that contribute to this possibility. Nonetheless, the patterns of neutralization against HCVcc of different genotypes () suggest that these antibodies are directed at highly conserved epitopes. These antigenic domain D antibodies showed more uniform neutralization against different HCV genotypes and subtypes than antigenic domain B antibodies 
Based on the observation that each antigenic domain D antibody inhibits E2 binding to CD81, epitope mapping studies by alanine scanning focused on E2 segments encompassing aa410–446, aa526–540 and aa611–617. These regions have been reported to contain contact residues that form the E2 binding site to CD81 
. As expected, contact residues were not located within aa526–540 but within aa410–446 and aa611–617. In the proposed model of the tertiary organization of HCV E2, domain I is organized such that β-strands C0
, as well as E0
, are consecutive in sequence, spanning aa418–444 and aa526–542 as two β-hairpins, respectively. While antigenic domain B antibodies are localized on the C0
β-hairpins, antigenic domain D antibodies are localized on C0
and on domain III. The location of antigenic domain D contact residues is in agreement with the model in which domains I and III are close in space. Although antigenic domain D is a distinct cluster of overlapping epitopes, there is some overlap between antigenic domain D and antigenic domain B. Some antigenic domain B antibodies, e.g., HC-11, share contact residues within C0
, at residues 442 and 443. The HC-84 epitope mapping data makes possible several adjustments of the tertiary model to accommodate the distance between contact residues and the spatial orientation of a connecting loop between the C0
β-strands (). One adjustment involves the sequence aa441–443. This sequence was proposed to be located on β-strand D0
, implying that at least one of the three residues would be buried in the β-sheet and not exposed to participate in binding to either CD81 or antibodies to this region, as discussed above. Of the three amino acids, L441 and Y443 are absolutely conserved but F442, although 100% conserved in genotypes 1, 2, 3 and 4, has low frequency changes to either M442 or L442 in genotypes 5 and 6. If F442 is buried and required for the structural integrity of the β-sheet, substitution with a large bulky residue like tryptophan is likely to distort the β-sheet less than substitution with a small side-chain like alanine. Interestingly, CD81 binding experiments showed that when phenylalanine was replaced by tryptophan, HCVpp entry was reduced by 70%, while for a F442A mutant no CD81 binding could be observed 
. A second adjustment is that W420 and I422 may be buried in the β-sheet, such that a substitution at these residues provides only a local distortion. If that is the case, a question is the accessibility of these residues to participate in binding to CD81, as well as binding to antibodies directed at this region 
. Moreover, these residues are absolutely conserved in the entire HCV database indicating functional or structural constraints preventing mutations at this site. Additional studies are required to confirm these modifications to the model. Overall, our findings provide support for this model, in which these antibodies bind to the same tertiary structure that interacts with CD81.
Since all of the antigenic domain D antibodies are affected by substitution to alanine at L441, F442 or Y443, we tested the binding of each antibody to a synthetic peptide encompassing aa434–446. This peptide has been proposed to encode non-neutralizing epitopes 
. However, several neutralizing monoclonal antibodies to linear epitopes in this region have been described 
. In our study, not only do the neutralizing antigenic domain D antibodies bind to contact residues within this sequence, but four of the nine antibodies, HC-84.1, -.25, -.26 and -.27, bound to this peptide indicating that the region aa434–446 forms an integral part of their epitopes. Although HC-84.25, -.26 and -.27 HMAbs are antibodies to conformational epitopes, their ability to bind to the aa434–446 synthetic peptide indicates that their epitopes contain a significant linear component. At the same time, binding by these antibodies to denatured E1E2 was significantly reduced (). One explanation could be that the synthetic peptide is sufficiently flexible to be shaped by the interaction with HC-84.1, -.25, -.26 and -.27 HMAbs leading to binding, but this cannot occur when the aa434–446 region is expressed in the context of denatured E1E2. Taken together, this region includes residues that are involved in conformational epitopes of potent and broadly neutralizing antibodies, although it remains possible that the E2 region aa434–446 encodes for non-neutralizing antibodies.
The initial expectation was for a minor subset of the antigenic domain D epitopes to be invariant because of functional or structural constraints. Surprisingly, when 2a HCVcc was grown in the presence of HC-84 HMAbs, under a viral escape selection protocol to maximize the likelihood of escape variants, five of five selected antibodies led to no escape variants, under the conditions tested. This can be explained in part by the findings in analysis of the effect of each contact residue within the HC-84 epitopes on H77C HCVpp entry (). Seven of ten contact residues, aa420, 429, 437, 441, 442, 613 and 616, when substituted with alanine led to >90% reduction in HCVpp entry compared to wt. Two other residues, aa428 and 443, when substituted reduced entry by approximately 80%. The only contact residue with moderate reduction was aa446, which is restricted to HC-84.27. This antibody was not selected for escape study. Reduction in entry correlated with reduction in binding to CD81 with each of these HC-84-related contact residues. The only residue when substituted without significant decrease in binding to CD81 was K446A. Previously studies have identified residues W420, W437, L441, F442, Y443, Y613 and W616 on E2 as participants in the interaction with CD81 
. Substitution at these sites would be expected to negatively modulate entry. Substitution of the cysteine at aa429 also would be expected to alter the structure required for this interaction. However, it remains possible that escape from HC-84 HMAbs can occur. The leading candidate would be HC-84.27, which includes a contact residue at aa446. Mutations at this site would not significantly decrease virus entry. Although rare, mutations at W437 and F442 have been documented in the HCV database, but the mutations could be associated with a reduction in virus fitness. Moreover, since the JFH1 2a HCVcc is highly sensitive to antibody-mediated neutralization, escape studies with a less sensitive isolate will need to be performed to confirm our findings. Collectively, the HC-84 cluster of epitopes, designated as antigenic domain D, is highly conserved among HCV genotypes and subtypes, and mediates broad and potent virus neutralization that is not likely to lead to virus escape. Thus, these epitopes are relevant in vaccine design for this highly diverse virus.