Fig 1.
Workflow to identify broadly neutralizing antibodies (bnAbs) from donor samples.
(A) Serum neutralization profile of 4 cohort participants chosen for downstream analysis based on potent neutralizing activity against DENV1-4 and ZIKV. The mean reciprocal serum dilution that neutralized 50% of virus infection (NT50) in 3 independent experiments is depicted as a heatmap with a darker color indicating greater potency according to the key. (B) B cells isolated from the peripheral blood mononuclear cells (PBMCs) of donors selected in (A) were processed for (C) single-cell RNA sequencing of both global gene expression (GEX) and B cell receptor (BCR)-specific libraries. (D) We analyzed BCR libraries using the software package partis [44], which groups antibodies into clonal families and infers their shared ancestry. (E) Antibody sequences most likely to encode broadly neutralizing antibodies (bnAbs) were bioinformatically downselected for functional characterization. (F) We recombinantly expressed selected antibodies as IgG1 and screened them for the ability to neutralize DENV1-4 and ZIKV. This figure was created with Biorender.com.
Fig 2.
Distribution of B cell subsets and antibody isotypes within clonal families.
Graphs depict the number of antibodies encoded (A) by distinct B cell subsets and (B) as various isotypes in clonal families of different sizes in each of the four donor samples analyzed. B cell subset and antibody isotype were determined by analysis of the cell’s transcriptome as captured by the gene expression library (see Methods and S4 Table for details). Only B cells for which a corresponding antibody sequence was observed in the B cell receptor library were included. “Undetermined” B cell subset indicates that the cell had too few reads or unique molecular identifiers to yield accurate gene expression information as analyzed by 10X Genomics Cell Ranger. “Undetermined” isotype indicates insufficient sequence coverage to determine the constant gene segment within the antibody.
Fig 3.
Neutralization profile of top bnAbs expressed as IgG1.
(A) Representative dose-response neutralization curves of each antibody against DENV1 WP-74, DENV2 16681, DENV3 CH53489, DENV4 TVP376, and ZIKV H/PF/2013 reporter virus particles performed in at least 3 biological replicates in duplicate wells. The data points and error bars represent the mean and range of the duplicates, respectively. (B) Mean IC50 values for antibody-virus pairs shown in (A) and compiled from S1 Table. *The final column displays the geometric mean IC50 values against neutralized viruses. (C) IC50 values against additional DENV variants selected due to known antigenic divergence from the panel in (B). Values shown are means from at least two biological replicates. (D) Mean IC50 values against fully infectious DENV clinical isolates from 2004–2007. Values were obtained from at least two biological replicates. *The final column displays the geometric mean IC50 of each antibody against the four viruses. In (B-D), IC50 values are displayed as heatmaps according to the key. Gray indicates that 50% neutralization was not observed at the highest antibody concentration tested (10,000 ng/ml).
Fig 4.
Determinants of E protein binding by bnAbs.
Relative binding efficiency of the indicated antibodies to (A) E protein monomers (B) or reporter virus particles of DENV2 16681 measured by ELISA. Results are from two independent experiments, each performed in duplicate wells. The absorbance of each duplicate, reported in arbitrary units (AU), was normalized to the wells that received positive control antibody B10 [39]. The HIV-specific antibody PGT121 [80] was used as a negative control. Data points represent the normalized means of each experiment and the bars represent the means of the two experiments. (C-F) DENV2 16681 E protein sites important for binding by antibody (C) F25.S02 or (E) F05.S03 are shown on the ribbon structure of the DENV2 E dimer (PDB: 1OAN) and labeled on one monomer. Sites in E domains I, II, and III are depicted in red, yellow, and blue, respectively. Bar graphs show the mean binding reactivity to individual alanine mutants that selectively impact (D) F25.S02 or (F) F05.S03 as a percentage of wildtype (WT) DENV2 E protein reactivity. Binding of control antibodies EDE1-C10 and J9 to these mutants was tested in parallel. Error bars show the range of binding reactivity from two independent experiments. The dotted line in (D-E) indicates 70% reduction in antibody binding activity to mutant compared to WT.
Fig 5.
E protein residues critical for neutralization by bnAbs.
(Left panel) Bar graphs show the mean IC50 fold change against DENV2 16681 reporter virus particles encoding E protein variants relative to wild type (WT) DENV2 for antibodies (A) F09.S05, (B) F05.S03, (C) J9, (D) F25.S02, and (E) EDE1-C10. Sites in E domains I, II, and III are shown in red, yellow, and blue, respectively. Values of 1, >1, and <1 indicate no change, decreased sensitivity, and increased sensitivity of mutant relative to WT DENV2, respectively. Mean values were obtained from at least 2 independent experiments shown as individual data points in which WT and mutant DENV2 were tested in parallel. WT ZIKV H/PF/2013 (gray) was included as a control. Error bars indicate the standard deviation (n>2) or range (n = 2). In each graph, the dotted horizontal line represents a 4-fold IC50 change. (Right panel) For each bnAb, sites of mutations that reduced neutralization potency when tested either individually or in combination by > 4-fold are depicted as spheres on both monomers of the DENV2 E dimer subunit (PDB 1OAN).
Fig 6.
Neutralization profile of antibodies expressed as IgA1.
(A) IgG1 (open circles), monomeric IgA1 (blue circles), and polymeric IgA1 (orange circles) versions of F25.S02 (top row), EDE1-C10 (middle row), and SigN-3C (bottom row) were tested for their ability to neutralize DENV1 WP-74, DENV2 16681, DENV3 CH53489, DENV4 TVP376, and ZIKV H/PF/2013 reporter virus particles. Dose-response curves are representative of 3 independent experiments, each tested in duplicate wells. Data points and error bars represent the mean and range of the duplicates, respectively. (B) Comparison of IC50 values of F25.S02 (left), EDE1-C10 (middle), SigN-3C (right) expressed as IgG1, monomeric IgA1, and polymeric IgA1 against the viruses indicated on the x-axes. Color scheme is similar to (A). Each data point represents an independent experiment in which antibody isotypes were tested in parallel. Horizontal bars indicate the mean. Error bars represent the standard error of the mean.
Fig 7.
Effect of antibody isotype on antibody dependent enhancement (ADE) of infection of U937 monocytes.
(A) DENV2 16681 reporter virus particles were pre-incubated with serial dilutions of IgG1 (filled circles), monomeric IgA1 (open circles), or polymeric IgA1 (open squares) forms of F25.S02 (blue), EDE1-C10 (orange), or SIgN-3C (green) prior to infection of U937 cells, which express Fc receptors for both IgG and IgA. IgG1 and IgA1 antibodies were tested individually in the assay. (B) Competitive ADE assays in U937 monocytes. F25.S02 (top row, blue), EDE1-C10 (middle row, orange) or SIgN-3C (bottom row, green) IgG1 was mixed with either autologous IgA1 (left panel, solid lines) or an IgA1 isotype control (right panel, dashed lines) at the indicated ratios by mass before serial dilution and pre-incubation with concentrated DENV2 16681 reporter virus particles. The assay was performed in two independent experiments, each in duplicate wells. The data points and the error bars represent the means and the range of the duplicates, respectively, from one representative experiment. (C) Area under the curve analysis for experiments represented in (B). For both biological replicates the area of the curve for each infection condition was calculated and normalized to infection in the 100% IgG1 condition. The data points and error bars represent the mean and the range of two independent experiments, respectively.