A Combination of Human Broadly Neutralizing Antibodies against Hepatitis B Virus HBsAg with Distinct Epitopes Suppresses Escape Mutations

Abstract

Although there is no effective cure for chronic hepatitis B virus (HBV) infection, antibodies are protective and correlate with recovery from infection. To examine the human antibody response to HBV, we screened 124 vaccinated and 20 infected, spontaneously recovered individuals. The selected individuals produced shared clones of broadly neutralizing antibodies (bNAbs) that targeted 3 non-overlapping epitopes on the HBV S antigen (HBsAg). Single bNAbs protected humanized mice against infection but selected for resistance mutations in mice with prior established infection. In contrast, infection was controlled by a combination of bNAbs targeting non-overlapping epitopes with complementary sensitivity to mutations that commonly emerge during human infection. The co-crystal structure of one of the bNAbs with an HBsAg peptide epitope revealed a stabilized hairpin loop. This structure, which contains residues frequently mutated in clinical immune escape variants, provides a molecular explanation for why immunotherapy for HBV infection may require combinations of complementary bNAbs.

Keywords: Hepatitis B infection; broadly neutralizing antibodies; crystal structure; elite neutralizing activity; escape mutations; humanized mice; passive immunotherapy; prophylaxis.

Conflict of interest statement

Declaration of Interests Q.W. and M.C.N. have a provisional patent application with the U.S. Patent and Trademark Office (62898735). Other authors have no conflicts of interest to declare.

Copyright © 2020 Elsevier Inc. All rights reserved.

Figures

Figure 1.. Antibody responses in HBV vaccinated and recovered individuals.

(A) Donor screen. Sera from 159 volunteers were evaluated for anti-HBs binding by ELISA (x-axis) and HBV serum neutralization capacity using HepG2-NTCP cells (y-axis). Serum neutralization capacity on the y-axis was calculated as the reciprocal of the relative percentage of infected HepG2-NTCP cells. The values for unexposed naïve donors are ≅1. Neutralization tests were performed at 1:5 serum dilution in the final assay volume. Each dot represents an individual donor. Green indicates unvaccinated and unexposed, black indicates vaccinated, and red indicates spontaneously recovered. The dashed line indicates the no serum control. Top neutralizers (serum neutralization capacity higher than 4) are indicated (top right). Boxed are representative samples shown in Figure 2A. Spearman’s rank correlation coefficient (rs) and significance value (p). (B and C) Dose-dependent HBV neutralization by serum (B) or by purified IgG (C). Two assays were used to measure percent infection: ELISA to measure HBsAg protein in the medium (upper panels) and immunofluorescent staining for HBcAg in HepG2-NTCP cells (lower panels). Dashed line indicates virus-only control. (D) Schematic representation illustrating the three forms of the HBV surface protein: L-, M- and S-protein. These three forms of envelope protein all share the same S-region, with PreS1/PreS2 and PreS2 alone as the N-terminal extensions for L- and M-protein, respectively. (E) S-protein produced in Chinese hamster ovary (CHO) cells blocks serum neutralizing activity. Graphs show infection efficiency as a function of the amount of S-protein added. The concentration of polyclonal IgG antibodies (pAb) is indicated. Upper and lower panels are as in (B) and (C). A representative of at least two experiments is shown. See also Figure S1 and Table S1.

Figure 2.. S-protein-specific antibodies.

(A) Frequency of S-protein-specific memory B cells. Representative flow cytometry plots displaying the percentage of all IgG+ memory B cells that bind to both allophycocyanin- and phycoerythrin-tagged S-protein (S-protein-APC and S-protein-PE). Flow cytometry plots from other individuals are shown in Figure S2A. Experiments were repeated two times. (B) Dot plot showing the correlation between the frequency of S-protein-binding IgG+ memory B cells and the serum neutralizing activity. Spearman’s rank correlation coefficient (rs) and significance value (p). (C) Each pie chart represents the antibodies from an individual donor, and the total number of sequenced antibodies with paired heavy and light chains is indicated in the center. Antibodies with the same combination of IGH and IGL variable gene sequences and closely related CDR3s in each individual are shown. The slices with the same color indicate shared antibodies with the same or similar combination of IGH and IGL variable genes between individuals (Figure S2B). Grey slices indicate antibodies with closely related sequences that are unique to a single donor. In white are singlets. (D) V(D)J alignments for representative IGHV3–30/IGLV3–21, IGHV3–33/IGLV3–21 and IGHV3–23/IGLV3–21 antibodies from donors #60/#146 (H006 and H008), #146/#13 (H014 and H012), and #13/#60/#146 (H021, H003 and H004) respectively. Boxed grey residues are shared between antibodies. See also Figure S2 and Table S2.

Figure 3.. Broad cross-reactivity.

(A) Binding to S-protein (adr serotype). 50% effective concentration (EC50 in ng/ml) required for binding of the indicated human monoclonal antibodies to the S-protein. Libivirumab (Eren et al., 2000; Eren et al., 1998) and anti-HIV antibody 10–1074 (Mouquet et al., 2012) were used as positive and negative controls, respectively. All antibodies were tested. (B) Comparative binding of the mature and unmutated common ancestor (UCA) of antibodies H006, H019, and H020 to S-protein by ELISA. (C) Anti-HBs antibody binding to 5 different serotypes of HBsAg. Similar to panel (A), EC50 values are color-coded: red, ≤50 ng/ml; orange, 50 to 100 ng/ml; yellow, 100 to 200 ng/ml; and white, > 200 ng/ml. The abbreviation b.d. indicates below detection. All antibodies were tested. All experiments were performed at least two times. See also Figure S3.

Figure 4.. HBsAg epitopes.

(A) Competition ELISA defines 3 groups of antibodies. Results of competition ELISA shown as percent of binding by biotinylated antibodies and illustrated by colors: black, 0–25%; dark grey, 26–50%; light grey, 51–75%; white, >76%. Weak binders (H002, H012, H013, H014, H018) were excluded. Representative of two experiments. (B) Results of ELISA on alanine scanning mutants of S-protein. Only the amino acids vital for antibody binding are shown. Binding to mutants relative to wild-type S-protein: black, 0–25%; dark grey, 26–50%; light grey, 51–75%; white, >75%. Additional details are provided in Figure S4. (C) Results of ELISA on human escape mutations of S-protein. Wild-type S-protein and empty vector serve as a positive and negative controls, respectively. Binding to mutants relative to wild-type S-protein: black, 0–25%; dark grey, 26–50%; light grey, 51–75%; white, >75%. Amino acid mutations in bold represent frequently observed mutations in humans (Ma and Wang, 2012). The antibodies tested in (B and C) were selected from Group-I, -II, -III based on their neutralizing activity (Figure 5A–5C). All experiments were performed at least two times. See also Figure S4.

Figure 5.. In vitro neutralization by the monoclonal antibodies.

(A and B) In vitro neutralization assays using HepG2-NTCP cells. Percent infection in the presence of the indicated concentrations of bNAbs measured by ELISA of HBsAg in medium (A) and anti-HBcAg immunofluorescence (B). Anti-HIV antibody 10–1074 (Mouquet et al., 2012) and libivirumab (Eren et al., 2000; Eren et al., 1998) were used as negative and positive controls respectively. The corresponding IC50s are shown in the left and middle column of panel (C). All experiments were repeated a minimum of two times. (C) bNAb 50% maximal inhibitory concentration (IC50) calculated based on HBsAg ELISA (left column) and HBcAg immunofluorescence (middle column) for the in vitro neutralization assays using HepG2-NTCP cells, or HBeAg ELISA (right column) for in vitro neutralization using primary human hepatocytes. The abbreviation b.d. and n.d. indicate below detection and not done respectively. (D) In vitro neutralization using primary human hepatocytes. The levels of HBeAg in medium were measured by ELISA. The calculated IC50 values are shown in the right column of panel (C). Experiments were repeated three times. (E) In vitro neutralization assay using HepG2-NTCP cells. IgG antibodies were compared to their corresponding Fab fragments. Concentrations of Fab fragments were adjusted to correspond to IgG. Experiment was performed two times. See also Figure S5.

Figure 6.. Crystal structure of H015 bound to its recognition motif.

A single crystal was used to obtain a high resolution (1.78 Å) structure. (A) Synthetic peptides spanning the antigenic loop region were subjected to ELISA for antibody binding. Among the tested antibodies, only H015 binds peptides-11 and −12. Experiments were performed three times and details are in Figure S6A. (B and C) The peptide binds to CDR1 (R31), CDR2 (W52 and F53) and CDR3 (E99, P101, L103, and L104) of H015 heavy chain (green) and CDR3 (P95) of the light chain (cyan) (B). The interacting residues (C) on the heavy chain (green) are R31 (main chain), W52, F53 (main chain), E99, P101 (main chain), L103 (main chain), L104 (hydrophobic). One contact with the light chain (cyan) is with P95. (D) Electron density map of the bound peptide as seen in the 2Fo-Fc map contoured at 1 RMSD indicating high occupancy (92%). (E) The recognition motif, KPSDGN, adopts a sharp hairpin conformation due to the salt-bridge between K141 and D144 and is facilitated by kinks at P142 and G145. Glycine 145 (G145, circled) is the residue that escapes the immune system when mutated to an arginine. See also Figure S6.

Figure 7.. Anti-HBs bNAbs are protective and therapeutic in vivo.

(A and E) Diagram of the prophylaxis and treatment protocols, respectively. (B) Prophylaxis with isotype control antibody 10–1074 (Mouquet et al., 2012). (C and D) Prophylaxis with H020 and H007 respectively. The dashed line in (B-D) indicates the detection limit. (F) Treatment of viremic huFNRG mice with control antibody 10–1074. (G and H) Treatment of viremic huFNRG mice with H020 alone or H007 alone, respectively. HBV DNA levels in serum were monitored on a weekly basis. Two independent experiments comprising a total of 5 to 8 mice were combined and displayed. (I) Mutations in the S-protein sequence from the indicated mice (red arrows) in (G), (H) and (J). S-protein sequence chromotograms are shown in Figure S7. (J-L) Treatment of viremic huFNRG mice with combination of anti-HBs bNAb H006 + H007 (J), or H017 + H019 (K), or H016 + H017 + H019 (L), respectively. Sequencing showed that none of the mice in (K) and (J) carried viruses with escape mutations in the S-protein. See also Figure S7.