Naturally selected hepatitis C virus polymorphisms confer broad neutralizing antibody resistance

Abstract

For hepatitis C virus (HCV) and other highly variable viruses, broadly neutralizing mAbs are an important guide for vaccine development. The development of resistance to anti-HCV mAbs is poorly understood, in part due to a lack of neutralization testing against diverse, representative panels of HCV variants. Here, we developed a neutralization panel expressing diverse, naturally occurring HCV envelopes (E1E2s) and used this panel to characterize neutralizing breadth and resistance mechanisms of 18 previously described broadly neutralizing anti-HCV human mAbs. The observed mAb resistance could not be attributed to polymorphisms in E1E2 at known mAb-binding residues. Additionally, hierarchical clustering analysis of neutralization resistance patterns revealed relationships between mAbs that were not predicted by prior epitope mapping, identifying 3 distinct neutralization clusters. Using this clustering analysis and envelope sequence data, we identified polymorphisms in E2 that confer resistance to multiple broadly neutralizing mAbs. These polymorphisms, which are not at mAb contact residues, also conferred resistance to neutralization by plasma from HCV-infected subjects. Together, our method of neutralization clustering with sequence analysis reveals that polymorphisms at noncontact residues may be a major immune evasion mechanism for HCV, facilitating viral persistence and presenting a challenge for HCV vaccine development.

Figure 9. I538V/Q546L/T563V mutations in E2 confer resistance to neutralization by plasma from HCV-infected subjects.

Introduction of I538V/Q546L/T563V into sensitive E1E2 clone 1b09 or back mutation of V538 to I, L546 to Q, and V563 to T in resistant clone 1a129 confers increased resistance or sensitivity, respectively, to neutralization by plasma samples from 18 HCV-infected subjects. Each box represents relative infection of the indicated HCVpp in the presence of each of 18 plasma samples. Dots indicate neutralization by plasma from the source donor of resistant E1E2 clone 1a129. Plasma was tested for neutralizing activity at a 1:100 dilution. Black bars indicate medians. Whiskers indicate 10th and 90th percentiles. P values were calculated by Wilcoxon signed-rank test.

Figure 8. I538V/Q546L/T563V mutations in E2 confer resistance to 6 NC1 mAbs by reducing mAb binding to E1E2.

(A) The first bar in each graph indicates relative infection of wild-type pp1b09 in the presence of the indicated mAb, adjusted to 1. Subsequent bars indicate fold change in neutralization resistance of the indicated HCVpp relative to pp1b09. Values are the means of 2 to 6 independent experiments performed in duplicate, and error bars indicate SD. *P < 0.05, **P < 0.005 by t test. (B) Binding of mAbs CBH-2 and AR3C to E1E2 protein in an ELISA. For each mAb, binding to each of the E1E2 variants was compared at a single mAb concentration selected to produce binding within the linear range of the assay. Values are normalized for relative binding of control NC2 mAb HC33.4. Error bars indicate SD between duplicate wells. (C) Crystal structure from Kong et al. of E2 core with AR3C (34), from the Protein Data Bank, accession 4MWF, with E2 colors modified as in Figure 5. Residue 431 is purple, residue 442 is orange, and residues 538 and 563 are blue. The likely position of residue 546 is indicated with a blue asterisk. AR3C Fab is tan.

Figure 7. V538I/L546Q/V563T mutations in E2 confer additive sensitivity to NC1 mAbs.

(A) Alignment of amino acids 526–569 of sensitive E1E2 clone 1b09 and resistant clones 1a129 and 1a142. Homology to the 1b09 amino acid sequence is indicated by a dot. Positions that differ between the sensitive and both resistant clones are highlighted in yellow. Arrows indicate contact residues for AR3C in the E2 core/AR3C crystal structure. (B) The dashed line indicates relative infection of HCVpp with chimeric E1E2 1b09/1a129-β in the presence of the mab AR3A, adjusted to 1. Each subsequent bar indicates the fold change in neutralization resistance after the indicated mutation(s) were introduced. Error bars indicate SD between duplicate wells. (C) Neutralization by serial dilutions of the indicated mAb of HCVpp with chimeric E1E2 1b09/1a129-β (pp1b09/1a129-β) or pp1b09/1a129-β after introduction of the indicated mutation(s). Error bars indicate SD between duplicate wells.

Figure 6. Polymorphisms between amino acids 526 and 569 (β-sheet) of resistant E1E2 clone 1a142 confer resistance to multiple broadly neutralizing mAbs.

Neutralization by the indicated mAb of HCVpp with E1E2 chimeras between resistant clone 1a142 and sensitive clone 1b09 (1b09/1a142-β and 1b09/1a142-F [F442I]), E1E2 with F442I introduced by site-directed mutagenesis (1b09/F442I), and with resistant E1E2 clone 1a142 (1a142 wild type), which carries F442I naturally. Chimeras with the front layer (1a142-F, amino acids 412–444) from resistant clone 1a142 are marked with cyan. Chimeras with β6-β8 (1a142-β, amino acids 526–569) from resistant clone 1a142 are marked in red. Orange indicates clones with F442I polymorphisms. Dashed lines indicate 50% neutralization. Error bars indicate SD between duplicate wells.

Figure 5. Polymorphisms between amino acids 526 and 569 (β-sheet) of resistant E1E2 clone 1a129 confer resistance to multiple broadly neutralizing mAbs.

(A) Crystal structure of H77 E2 core, published by Kong and colleagues (34), from the Protein Data Bank, accession 4MWF, with colors modified. E2 core front layer (amino acids 412–444) is cyan, β2-β5/partial CD81-binding loop (amino acids 445–525) is yellow, and partial CD81-binding loop/β6-β8 (amino acids 526–569) is red. Remaining structure is green. Residue 442 is marked in orange. (B) Neutralization by the indicated mAb of HCVpp with E1E2 chimeras between sensitive clone 1b09 and neutralization-resistant E1E2 clone 1a129. Chimeras with β6-β8 (1a129-β, amino acids 526–569) from resistant clone 1a129 are marked in red. Chimeras with the front layer, CD81-binding loop, and the central β-sheet (1a129-FCβ, amino acids 412–569) from resistant clone 1a129 are marked with cyan, yellow, and red. Dashed lines indicate 50% neutralization. Error bars indicate SD between duplicate wells.

Figure 4. D431E, F442I, and F560Y mutations each confer resistance to multiple broadly neutralizing mAbs.

The dashed lines indicate relative infection of wild-type 1b09 HCVpp in the presence of the indicated mAb, adjusted to 1. Each bar indicates the fold change in HCVpp neutralization resistance relative to wild-type 1b09 after the indicated mutation(s) were introduced. Values are the means of 2 to 6 independent experiments performed in duplicate, and error bars indicate SD. *P < 0.05, **P < 0.005 by t test. Black triangles indicate HCVpp with D431E mutations. White triangles indicate HCVpp with F442I mutations. Gray triangles indicated HCVpp with F560Y mutations. pp1b09, HCVpp with wild-type E1E2 1b09.

Figure 3. Sequence analysis reveals polymorphisms associated with neutralization resistance.

(A) Representative E1E2 sequence analysis to identify mAb HC84.22 resistance-associated polymorphisms. Similar analyses for the remaining NC1 mAbs are shown in Supplemental Figure 4. E1E2 clones are ranked by increasing resistance to HC84.22. Clones are grouped into the 5 most sensitive, 7 with intermediate resistance, and 7 with greatest resistance, separated by horizontal black lines. Gray vertical bars indicate positions with a substitution in any resistant E1E2 clone but in none of the 5 most sensitive E1E2 clones. Black vertical bars indicate CD81-binding sites in E2 determined by alanine scanning. Blue vertical bars indicate critical binding residues for HC84.22 determined by alanine scanning. Sites with substitutions in at least 2 resistant clones but in no sensitive clones are included in the summary panel in the bottom row. (A–C) Sites marked with red are predominantly polymorphic in the 7 most resistant clones. Orange indicates sites that are polymorphic in an equal number of highly resistant and intermediate resistant clones. Green indicates sites that are predominantly polymorphic in the 7 clones with intermediate resistance. (B) Compiled results for all NC1 mAbs. (C) Resistance-associated polymorphisms, with the more commonly observed polymorphism listed first. The amino acid found at each position in the 5 most neutralization-sensitive E1E2 clones is indicated in the top row. Black boxes indicate positions at which results from this analysis are concordant with alanine scanning mutagenesis results with the same mAb.

Figure 2. All neutralizing mAbs segregate into 3 neutralization clusters.

A heat map was generated using pairwise correlations among all mAb resistance profiles. Circles at each intersection are scaled by the magnitude of the Spearman correlation (r) between neutralization profiles of the indicated mAbs, with darker blue indicating higher r values. Open circles indicate identity. *P < 0.05. Hierarchical clustering analysis using pairwise correlations of neutralization resistance patterns is depicted as a tree. Numbers at tree nodes are approximately unbiased (AU) test values(42), which indicate strength of support for a particular cluster.

Figure 1. Ranking of library HCVpp resistance reveals relationships among mAbs.

(A) Each bar represents neutralization of a unique HCVpp by the indicated mAb. Representative mAbs are shown here, and neutralization results for all 18 mAbs are shown in Supplemental Figure 1. Relative infection is infection in the presence of 10 μg/ml neutralizing mAb relative to infection in the presence of nonspecific IgG. Error bars indicate SDs between duplicate wells. Pseudoparticles with murine leukemia virus (MLV) envelope serve as a negative control for neutralization. (B–D) Each blue diamond indicates neutralization of an individual pseudoparticle by one antibody on the x axis and a second antibody on the y axis. Spearman correlations (r) are indicated for each mAb pair. *P < 0.0001. (B) Comparison of HCVpp resistance to HC33.4 and HC33.8, 2 closely related mAbs targeting the same linear epitope. (C) Comparison of HCVpp resistance to 2 unrelated mAbs, HC33.4 and HC84.22. (D) Comparison of HCVpp resistance to AR3C and HC84.22, 2 mAbs with distinct epitopes based on alanine scanning but overlapping epitopes based on crystal structures.