Protective Capacity of the Human Anamnestic Antibody Response during Acute Dengue Virus Infection

Abstract

Half of the world's population is exposed to the risk of dengue virus infection. Although a vaccine for dengue virus is now available in a few countries, its reported overall efficacy of about 60% is not ideal. Protective immune correlates following natural dengue virus infection remain undefined, which makes it difficult to predict the efficacy of new vaccines. In this study, we address the protective capacity of dengue virus-specific antibodies that are produced by plasmablasts a few days after natural secondary infection. Among a panel of 18 dengue virus-reactive human monoclonal antibodies, four groups of antibodies were identified based on their binding properties. While antibodies targeting the fusion loop of the glycoprotein of dengue virus dominated the antibody response, two smaller groups of antibodies bound to previously undescribed epitopes in domain II of the E protein. The latter, largely serotype-cross-reactive antibodies, demonstrated increased stability of binding at pH 5. These antibodies possessed weak to moderate neutralization capacity in vitro but were the most efficacious in promoting the survival of infected mice. Our data suggest that the cross-reactive anamnestic antibody response has a protective capacity despite moderate neutralization in vitro and a moderate decrease of viremia in vivo IMPORTANCE: Antibodies can protect from symptomatic dengue virus infection. However, it is not easy to assess which classes of antibodies provide protection because in vitro assays are not always predictive of in vivo protection. During a repeat infection, dengue virus-specific immune memory cells are reactivated and large amounts of antibodies are produced. By studying antibodies cloned from patients with heterologous secondary infection, we tested the protective value of the serotype-cross-reactive "recall" or "anamnestic" response. We found that results from in vitro neutralization assays did not always correlate with the ability of the antibodies to reduce viremia in a mouse model. In addition, a decrease of viremia in mice did not necessarily improve survival. The most protective antibodies were stable at pH 5, suggesting that antibody binding in the endosomes, after the antibody-virus complex is internalized, might be important to block virus spread in the organism.

FIG 1

Binding groups of human plasmablast-derived antibodies. (A and B) ELISA of dengue virus-specific antibodies to rE protein (A) and to whole viral particles (B) of all four DENV serotypes. (C) One-way competition ELISA. Nonlabeled antibodies were added in 100-fold excess over biotinylated antibodies to compete for binding sites on the captured rE protein. The values are normalized to antibody binding in the absence of competition. Blocking was defined as reduction in absorbance readings by at least 70%.

FIG 2

Group D antibodies are DENV fusion loop specific. (A) Summary of antibody clones in each of the four binding groups A, B, C, and D and the color code for each group. (B) 4G2 competition ELISA. Nonlabeled antibodies were added in 100-fold excess over biotinylated 4G2 to compete for binding sites on the surface of the captured rE.

FIG 3

Groups of antibodies bind to distinct epitopes in EDI or EDII. (A) Amino acid contact residues engaged by antibodies were identified by shotgun mutagenesis mapping. DENV prM/E mutants were expressed in HEK-293T cells, and binding by test antibodies was detected by a fluorescent secondary antibody, normalizing the results to the mean fluorescence intensity. (B) Sandwich ELISA with rE proteins containing alanine replacement mutations in surface-exposed amino acids. The values were normalized to unmutated rE.

FIG 4

Antibody groups show different stabilities at low pH. (A) The stability of antibody binding at pH 5 or pH 7 was assessed by surface plasmon resonance. A representative sensorgram from an antibody from each linkage group is shown. (B) Stability of antibody binding at pH 5 or pH 7 for the same antibodies was tested in a sandwich ELISA (see Materials and Methods) for all four serotypes. The ratios of EC₅₀ values for pH 7 to the EC₅₀ values for pH 5 are shown for each tested antibody. ND, not done; NA, not applicable, since the curve fit for pH 5 was ambiguous. (C) Epitopes of antibodies 6C-H8L1 and 2C-H3L2 illustrated on the trimeric form of the E protein. The three E proteins are shown in blue, light blue, and purple, and the epitope is indicated in only one of the E proteins. (D) BHK-21 cells 7 min after uptake of the indicated antibodies complexed with DENV-1. Anti-human IgG (green) and anti-EEA-1 (red) were used to detect complexes in the early endosomes (yellow). The insets are magnified areas of the main images.

FIG 5

Group B antibodies promote survival despite low in vitro neutralizing capacity. (A) BHK21 cell-based PRNT of all Abs against all four DENV serotypes ordered according to color-coded groups A to D. The highest concentration tested was 100 μg/ml, indicated by the dashed line. Antibodies that were not neutralizing at 100 μg/ml were arbitrarily assigned a PRNT₅₀ value of 100 μg/ml. Each dot represents one antibody. (B) U937-DC-SIGN cell-based neutralization assay of all the Abs against all four DENV serotypes, ordered according to color-coded groups A to D. Each dot represents one antibody (7). (C) In vivo protection assay in AG129 mice as illustrated in the treatment scheme. Each mouse was treated i.v. with a total of 100 μg of a mixture of Abs from each group (1 for group A, 2 for group B, 3 for group C, 12 for group D, and 18 for group E [A, B, C, and D]). Viremia was measured 3 days after infection with DENV-2, and mouse survival was monitored until day 12. A Kruskal-Wallis test was used to compare the groups, and a P value of <0.05 was considered statistically significant. Survival statistics were calculated using a log-rank Mantel-Cox test and applying a Bonferroni-corrected P value of 0.01 as the threshold for statistical significance, considering five comparisons of isotype control versus treatment.