Flexibility in surface-exposed loops in a virus capsid mediates escape from antibody neutralization

Abstract

New human norovirus strains emerge every 2 to 3 years, partly due to mutations in the viral capsid that allow escape from antibody neutralization and herd immunity. To understand how noroviruses evolve antibody resistance, we investigated the structural basis for the escape of murine norovirus (MNV) from antibody neutralization. To identify specific residues in the MNV-1 protruding (P) domain of the capsid that play a role in escape from the neutralizing monoclonal antibody (MAb) A6.2, 22 recombinant MNVs were generated with amino acid substitutions in the A'B' and E'F' loops. Six mutations in the E'F' loop (V378F, A382K, A382P, A382R, D385G, and L386F) mediated escape from MAb A6.2 neutralization. To elucidate underlying structural mechanisms for these results, the atomic structure of the A6.2 Fab was determined and fitted into the previously generated pseudoatomic model of the A6.2 Fab/MNV-1 virion complex. Previously, two distinct conformations, A and B, of the atomic structures of the MNV-1 P domain were identified due to flexibility in the two P domain loops. A superior stereochemical fit of the A6.2 Fab to the A conformation of the MNV P domain was observed. Structural analysis of our observed escape mutants indicates changes toward the less-preferred B conformation of the P domain. The shift in the structural equilibrium of the P domain toward the conformation with poor structural complementarity to the antibody strongly supports a unique mechanism for antibody escape that occurs via antigen flexibility instead of direct antibody-antigen binding.

Importance: Human noroviruses cause the majority of all nonbacterial gastroenteritis worldwide. New epidemic strains arise in part by mutations in the viral capsid leading to escape from antibody neutralization. Herein, we identify a series of point mutations in a norovirus capsid that mediate escape from antibody neutralization and determine the structure of a neutralizing antibody. Fitting of the antibody structure into the virion/antibody complex identifies two conformations of the antibody binding domain of the viral capsid: one with a superior fit and the other with an inferior fit to the antibody. These data suggest a unique mode of antibody neutralization. In contrast to other viruses that largely escape antibody neutralization through direct disruption of the antibody-virus interface, we identify mutations that acted indirectly by limiting the conformation of the antibody binding loop in the viral capsid and drive the antibody binding domain into the conformation unable to be bound by the antibody.

FIG 1

Generation of monoclonal antibody (MAb) A6.2 escape viruses by serial passaging in cell culture. (A) MNV-1 was passaged 5 times through RAW 264.7 cells in the presence of increasing concentrations of the neutralizing MAb A6.2 or a nonneutralizing isotype control antibody. Passage 0 (P0) virus, P3 virus grown with MAb A6.2, and P5 viruses from both conditions were then analyzed in a neutralization assay with 0, 20, 60, and 200 ng MAb A6.2. Data are from two independent experiments. The error bars indicate standard errors. Statistical analysis was performed using the t test. *, P < 0.05; **, P < 0.01. (B) Viruses from RAW 264.7 cell lysates obtained after passage in MAb A6.2 or the isotype control were analyzed by 454 DNA sequencing. Mutations were graphed onto the MNV-1 P domain crystal structure. Ribbon thickness and color (blue to red) are approximately equal to the log frequency of each mutation. The dominant MNV-1 P domain mutants in passages 3 (P3) and 5 (P5) are shown in red (D385G) and orange (V378F), respectively. (C) Clonal sweep trajectory of MNV-1 mutants under MAb A6.2 pressure. Wild-type MNV-1 was the dominant phenotype at P0. MNV-1 containing the mutation 385G became the dominant genotype by P3, with a shift to 378F-containing MNV-1 mutants by P5. The frequency of indicated single point mutations is shown in dark boxes, while gray boxes show the same mutation combined with other mutations.

FIG 2

Single amino acid substitutions did not affect MNV-1 growth. RAW 264.7 cells were infected with recombinant viruses having mutations in the A′B′ loop (A and B) or E′F′ loop (C and D) at an MOI of 2 on ice for 1 h. The inoculum was removed, and cells were infected at the indicated time points. Virus titers were determined by plaque assay. The titer for each mutant virus was compared to that for wild-type (WT) MNV. Data are presented as means ± standard errors (SE) from duplicate samples in three independent experiments.

FIG 3

Mutations in the E′F′ loop of the P domain mediate MNV-1 escape from MAb A6.2 neutralization in culture. WT MNV and nine A′B′ loop single point mutants (A), nine E′F′ loop single point mutants (B), and three naturally occurring mutants as well as one double mutant (C) were subjected to in vitro MAb A6.2 neutralization. Virus was incubated with indicated concentrations of MAb A6.2 for 30 min at 37°C before infection of RAW 264.7 cells. Viral titers were measured by plaque assay. Percent infectivity of virus titers at different concentrations of MAb A6.2 was calculated relative to the control without MAb A6.2 set to 100%. Data are presented as means ± SE from at least three independent experiments.

FIG 4

Mutations in the E′F′ loop of the MNV-1 P domain mediate escape from MAb A6.2 neutralization in mice. STAT1^−/− mice were simultaneously infected with 10⁶ PFU of the indicated recombinant MNV-1 orally and injected with MAb A6.2 (open box) or isotype control MAb (black box) intraperitoneally. Virus titers were measured 48 h postinfection in indicated tissues from animals infected with the wild type (A) or one of the Q298E (B), S299A (C), A382K (D), A382R (E), or L386F (F) mutants with 500 μg MAb, WT-infected animals injected with 15 (G) and 100 (H) μg MAb, and L386F mutant-infected animals injected with 15 (I) and 100 (J) μg MAb, respectively. Data are presented as means ± standard deviations (SD) from at least five mice per condition from at least two independent experiments. Statistical analysis was performed using the t test to compare the virus titer for each tissue between MAb A6.2-injected mice and the isotype control. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

FIG 5

Analysis of recombinant MNV-1 P domains. Shown here are the SDS-PAGE and Western analyses of the various purified P domain mutants. The upper panel shows the Western blots. The lower panel shows the Coomassie-stained SDS-PAGE gels to demonstrate the purity of each and that approximately the same amount of P domain was loaded in each well.

FIG 6

Binding of recombinant MNV-1 P-domains to MAb A6.2. ELISA plates were coated overnight at 4°C with bacterially expressed, recombinant MNV-1 P domain A′B′ loop (A) and E′F′ loop (B) mutants at 50 μg/ml in each well. Diluted purified MAb A6.2 followed by secondary antibodies was incubated for 60 min at 37°C. MNV.CR3 P domain was used as a negative control. Data are presented as means ± SE from three independent experiments. The three bars in each set of bars represent three dilutions of purified MAb A6.2: 1, 0.1, and 0.01 μg/ml.

FIG 7

Crystal structure of anti-MNV-1 Fab A6.2. (A) Shown here is a stereo diagram of a typical example of the final electron density of the refined structure of Fab A6.2 to an ∼2.5-Å resolution. The carbon, nitrogen, oxygen, and sulfur atoms are colored white, blue, red, and yellow, respectively. (B) In this stereo figure, the molecular surface of the Fab MAb A6.2 paratope is colored according to the hydrophobic index, ranging from blue for nonpolar to red for charged. The main chain atoms for the whole Fab were assigned a relatively neutral hydrophobicity index corresponding to the color green. The white line denotes the border between the heavy and light chains. The approximate location of the very hydrophobic CDR3 loop is also indicated.

FIG 8

Refitting MAb A6.2 Fab structure into the cryo-TEM for Fab-virus interaction. With the crystal structures of both the Fab and the P domain determined, the fitting of the structures into the molecular envelopes from the cryo-TEM image reconstructions was revisited. (Top) As detailed in Materials and Methods, the program package SITUS was used to refit the structures in an unbiased manner. In the previous crystal structure (27), there are two conformations for the P domain that were designated subunits A (green) and B (brown). The heavy and light chains for the Fab are shown in red and blue, respectively. (Bottom) Details of the fitted ensemble. The two loops that comprise the epitope are splayed apart in the A conformation compared to the B conformation. This causes severe clashes in the case of the B conformation (left side) compared to the A conformation (right side). Furthermore, the splayed conformation of A exposes a hydrophobic patch that complements the similarly nonpolar CDR3 heavy-chain loop.

FIG 9

MNV-1 P domain mutant viruses may escape MAb A6.2 neutralization by stabilization of the B conformation. (A) The location of the L386F mutation in the MNV-1 P domain. In the B conformation (blue), L386 lies between the two antigenic loops and does not contact the antibody. (B) As shown in this surface rendering of the top of the P domain, L386 (mauve) is fully exposed in the A conformation (tan). This suggests that L366F mutation will stabilize B over A and block antibody binding since the A conformation appears to be preferred for antibody binding. (C) Similarly, the A382R mutation may destabilize the A conformation since it would place an Arg residue in the middle of a highly basic patch of the P domain.