Broad neutralization by a combination of antibodies recognizing the CD4 binding site and a new conformational epitope on the HIV-1 envelope protein

Abstract

Two to three years after infection, a fraction of HIV-1-infected individuals develop serologic activity that neutralizes most viral isolates. Broadly neutralizing antibodies that recognize the HIV-1 envelope protein have been isolated from these patients by single-cell sorting and by neutralization screens. Here, we report a new method for anti-HIV-1 antibody isolation based on capturing single B cells that recognize the HIV-1 envelope protein expressed on the surface of transfected cells. Although far less efficient than soluble protein baits, the cell-based capture method identified antibodies that bind to a new broadly neutralizing epitope in the vicinity of the V3 loop and the CD4-induced site (CD4i). The new epitope is expressed on the cell surface form of the HIV-1 spike, but not on soluble forms of the same envelope protein. Moreover, the new antibodies complement the neutralization spectrum of potent broadly neutralizing anti-CD4 binding site (CD4bs) antibodies obtained from the same individual. Thus, combinations of potent broadly neutralizing antibodies with complementary activity can account for the breadth and potency of naturally arising anti-HIV-1 serologic activity. Therefore, vaccines aimed at eliciting anti-HIV-1 serologic breadth and potency should not be limited to single epitopes.

Figure 1.

Binding and adsorption of HIV-1–reactive antibodies and purified IgGs to GFP-293T^BaL cells. (A) Dot plots show mean fluorescence intensity (MFI) of staining by 51 HIV-1–reactive mAbs (y axis) tested on gp160Δc^BaL-expressing 293T cells versus IC₅₀ measured in TZM-bl assay using the BaL.26 pseudovirus. On the left, antibodies are grouped into those with no or low neutralizing activity (IC₅₀ > 50 µg/ml, black), and those that neutralize (IC₅₀ <50 µg/ml, red). On the right, the IC₅₀ values for neutralizing antibodies (IC₅₀ < 50 µg/ml) are shown. P-values were determined using the Mann-Whitney U test (left); correlation (right) was analyzed by Spearman correlation coefficient (rho). (B) As in A using IgG purified from the serum of 81 HIV-1–infected patients. (C) Graphs show the neutralization activity against BaL.26 of IgG from 4 patients (3B, 7A, 8A, and C69) after adsorption with 293T cells transfected with plasmids encoding gp160Δc^BaL (blue) or no-insert (green). Antibody binding to gp160Δc^BaL-expressing 293T cells (A and B) was measured in at least two independent experiments and one representative dataset is shown. Neutralization activity of adsorbed patients’ IgGs was analyzed in duplicate.

Figure 2.

Antibodies cloned by single-cell sorting using cell surface–expressed gp160Δc^BaL. (A) Cell sorting strategy. Dot plot (left) shows percentage of doublets composed of GFP-293T^BaL cells and CD20⁺ B cells from one HIV-1–infected donor (patient 3B). IgG-expressing B cell/GFP-293T^BaL doublets (middle) were sorted into single wells and subjected to antibody cloning procedure. (right) ImageStream^X visualization of a PE-stained B cell (orange) attached or in close proximity to a larger GFP-positive gp160Δc^BaL-transfected 293T cell. (B) Pie charts depicting expansion of clonally related antibodies (colored) cloned from four HIV-1–infected individuals (3B, 7A, 8A, and C69). The numbers within the inner circles (top row) indicate the total number of IgH sequences analyzed. Percentages of clonally related sequences are shown in the middle row. Total numbers of antibody clones are displayed within the inner circle (bottom row) and expansion of clones are proportionally displayed in pie charts (bottom row). (C) mAbs were tested by FACS (y axis shows MFI) for binding to 293T cells transfected with GFP-harboring plasmids encoding gp160Δc^BaL (black bar), gp160Δc^YU2 (gray bar), or no-insert (white bar). All 15 antibodies that bound to gp160Δc^BaL are shown. PG16, b12 (positive controls), and mGO53 (negative control) were included. (D) The same antibody panel as in C was tested for binding to soluble HIV-1 proteins (gp140^BaL and gp140^YU2) by ELISA. Graphs show OD405nm (y axis) and antibody concentration in µg/ml (x axis). Binding analysis of generated antibodies to soluble protein (D) and cell surface–expressed gp160Δc^BaL/gp160Δc^YU2 (C) was at least performed in duplicates.

Figure 3.

Heat map showing neutralization activity of mAbs and total IgG isolated from patient 3B. The IC₅₀ values for 3BC176, 3BC315, and total IgG on a panel of 39 viral strains comprising multiple clades compared with published data on the neutralization of the same viruses by 3BNC117 and 3BNC55 (Scheid et al., 2011). Top group displays viruses neutralized by both sets of antibodies (clone 1 and clone 2). Second and third group lists viruses that are neutralized only by 3BNC117 and 3BNC55 or only by 3BC176 and 3BC315, respectively. Three virus strains (bottom) were not neutralized by any of these antibodies. IC₅₀ values are color-coded: IC₅₀ < 0.1, red; IC₅₀ between 0.1 and 1 µg/ml, dark orange; IC₅₀ between 1 and 10 µg/ml, light orange; IC₅₀ between 10 and 100 µg/ml, dark yellow; and IC₅₀ > 100 µg/ml, yellow. IC₅₀ values above the measured concentration, as well as viruses that were not determined (nd), are not highlighted. All neutralization assays were performed in duplicate. Additionally, 3BC176 and 3BC315 were reproduced and retested against a panel of 10 different virus strains, confirming the neutralizing activity.

Figure 4.

Characterization of the epitope recognized by 3BC176 and 3BC315. (A) MFIs of the antibodies 3BC176, 3BC315, and controls measured at 20 µg/ml on GFP-293T^BaL wt and the indicated mutants (N160K, D368R, and I420R). (B) ELISA for binding to SF162 wt (left) and to SF162^K160N mutant (right) by PG9, PG16, 3BC176, and 3BC315. Graphs show OD405 nm (y axis) and antibody concentration in µg/ml (x axis). (C) Pseudovirus neutralization measured in TZM-bl assay. Graphs show percent neutralization (y axis) by increasing concentrations of 3BC176 or 3BC315 (x axis) of wt SF162 and mutants lacking the V1 or the V2 loop (SF162 DV1 and SF162 DV2, respectively). (D) IC₅₀ of 3BC176 and 3BC315 neutralization of SF162 wt and 11 SF162 pseudoviruses carrying single mutations at different glycosylation sites or the K160N mutatio.n Position of mutated glycosylation site (x axis) according to HXBc2. For both antibodies, the fold increase or decrease of the IC₅₀ values are visualized. (E) Microarray analyses of 3BC176, 3BC315, and 2G12 binding using a set of 50 oligosaccharide probes as neoglycolipids. Table S4 gives designations of the oligosaccharide probes. The binding signals (fluorescence intensities) shown are the mean values of duplicate spots, printed at 2 and 5 fmol per spot (the error bars represent half of the difference between the two values). (F) Graph shows enhancement of Alexa Fluor 647–labeled antibody binding to GFP-293T^BaL in the presence of sCD4. Staining intensity is measured by MFI (y axis), and the starting value normalized to MFI 1,000. 3–67 is a CD4-induced site (CD4i) antibody and was used as control (Scheid et al., 2011). (G) Inhibition of 3BC176 and 3BC315 binding to gp160 Δc^BaL-transfected 293T cells in the presence of sCD4 (10 µg/ml). Graphs show inhibition (in percentage) of Alexa Fluor 647–labeled 3BC176 or 3BC315 (y axis) in the presence of increasing concentrations of the indicated antibodies (x axis). All experiments were at least performed in duplicate, and results of representative experiments (D and G) are shown. Standard errors are shown for A–C and F.