Role of dendritic cells in antibody-dependent enhancement of dengue virus infection

Abstract

Dengue viruses (DV), composed of four distinct serotypes (DV1 to DV4), cause 50 to 100 million infections annually. Durable homotypic immunity follows infection but may predispose to severe subsequent heterotypic infections, a risk conferred in part by the immune response itself. Antibody-dependent enhancement (ADE), a process best described in vitro, is epidemiologically linked to complicated DV infections, especially in Southeast Asia. Here we report for the first time the ADE phenomenon in primary human dendritic cells (DC), early targets of DV infection, and human cell lines bearing Fc receptors. We show that ADE is inversely correlated with surface expression of DC-SIGN (DC-specific intercellular adhesion molecule-3-grabbing nonintegrin) and requires Fc gamma receptor IIa (FcgammaRIIa). Mature DC exhibited ADE, whereas immature DC, expressing higher levels of DC-SIGN and similar FcgammaRIIa levels, did not undergo ADE. ADE results in increased intracellular de novo DV protein synthesis, increased viral RNA production and release, and increased infectivity of the supernatants in mature DC. Interestingly, tumor necrosis factor alpha and interleukin-6 (IL-6), but not IL-10 and gamma interferon, were released in the presence of dengue patient sera but generally only at enhancement titers, suggesting a signaling component of ADE. FcgammaRIIa inhibition with monoclonal antibodies abrogated ADE and associated downstream consequences. DV versatility in entry routes (FcgammaRIIa or DC-SIGN) in mature DC broadens target options and suggests additional ways for DC to contribute to the pathogenesis of severe DV infection. Studying the cellular targets of DV infection and their susceptibility to ADE will aid our understanding of complex disease and contribute to the field of vaccine development.

FIG. 1.

ADE of DV infection. (A) Schematic diagram of flow cytometry-based ADE assay results. The control value is the percentage of DV-infected cells in the absence of DV-immune serum (baseline infection). The PENT is the dilution at which the maximum infection rate occurs for the tested serum. The power is the ratio of the percent infection rate at the PENT divided by the percent infection rate at the control titer. (B) Comparison of ADE effects of DV2 in K562 cells by using anti-DV1 DV-immune serum, two commercially available anti-DV antibodies, 4G2 and 3H5, and healthy human IgG. Three independent experiments were performed in triplicate, and data shown are the means ± standard deviations for all three experiments. (C) Infection of U937 WT and 3T3 WT cells (upper panels) and U937-DS and 3T3-DS cells (lower panels) by DV2 S16803 (MOI = 1). The solid line represents the infectivity with DV-immune serum, and the dashed line represents infectivity without DV-immune serum in one experiment representative of three independent experiments. Surface expression levels of DC-SIGN for each cell line as measured by flow cytometry are shown as insets in the upper panels.

FIG. 2.

ADE in cell lines as a function of DC-SIGN expression. (A) DC-SIGN surface expression on K562 WT (shaded histogram), K562-Lo (dashed line), and K562-Hi (solid line) cells measured by using flow cytometry. (B) ADE patterns obtained from infection of K562 WT (left column), K562-Lo (middle column), and K562-Hi (right column) cells with or without serial dilutions of DV-immune sera. SSC, side scatter. (C) Percent surface expression of DC-SIGN (bars) versus power of enhancement (solid line) for K562 WT, K562-Lo, and K562-Hi cells. The inset graphs the power and the DC-SIGN MFI for each of the three cell types in three independent experiments, and the correlation (r² = 0.96) was determined using a nonlinear curve fitting algorithm. Data shown are the means ± standard deviations from three independent experiments for each cell line.

FIG. 3.

Analysis of ADE pattern in monocytes, immDC, and matDC. (A) DC-SIGN surface expression on immDC, matDC, and monocytes from a single representative donor. (B) MFI of surface DC-SIGN expression from paired immDC and matDC (n = 5). (C) Infection and ADE patterns obtained for monocytes, immDC, and matDC prepared from a single representative donor in the absence or presence of DV-immune serum. All three cell types (monocyte, immDC, and matDC) were tested in three independent experiments with three different donors.

FIG. 4.

Increased viral production and proinflammatory cytokines with ADE in matDC. (A) Detection of viral output in supernatant using quantitative real-time PCR compared with intracellular viral antigen detection using 2H2 MAb in matDC undergoing ADE. (B) Culture supernatants collected from matDC undergoing ADE (as described for panel A) were tested for productive infection by culturing with Raji-DS cells (filled circles). (C) Supernatants from matDC described for panel A were tested in parallel in Vero cell plaque assays to confirm Raji-DS infectivity data. (D) Enhanced proinflammatory cytokines (TNF-α and IL-6) were detected in culture supernatants from matDC undergoing ADE only at enhancement titers. The dashed line indicates the lower limit of detection for the assay (20 pg/ml). Results from an experiment representative of four independent experiments performed in triplicate are shown. All data points shown are means ± standard deviations. Similar cytokine production patterns were obtained from five donors tested across five different sera demonstrating DV ADE (Table 1).

FIG. 5.

Influence of FcγRIIa on ADE in matDC. (A) Scatter plots show the percentages of FcγRIIa-positive cells and FcγRIIb-positive cells in immDC (iDC) and matDC (mDC) from eight paired donors. The black horizontal lines represent means. FcγRIIb expression decreases significantly between immDC and matDC (paired t test; P < 0.001). The histogram shows a single representative donor's changes in FcγRIIa and FcγRIIb expression levels with maturation. (B) Infection and ADE pattern in matDC treated with control IgG1, control IgG2a, control IgG2b, specific anti-FcγRIIa MAb, and anti-FcγRIIb MAb. The dashed line represents the baseline infection rate without DV-immune serum. (C) TNF-α (left panel) and IL-6 (right panel) in matDC (white bar) undergoing ADE, under the influence of control IgG1, IgG2a, IgG2b, specific anti-FcγRIIa MAb, and specific anti-FcγRIIb MAb. The dashed line indicates the lower limit of detection for the assay (20 pg/ml). Results of one experiment representative of three independent experiments using three donors with two different DV-immune sera are shown. The experiments were performed in triplicate and the bars represent means ± standard deviations.

FIG. 6.

ADE requires active virus. mDC, matDC. (A) Percent infection in ADE assay in matDC using active DV2 versus UV-irradiated DV2. (B) Cytokine production (TNF-α [left panel] and IL-6 [right panel]) in ADE assay supernatants with active DV2 versus UV-irradiated DV2. (C) DV infection without DV-immune serum (left panel) in matDC alone and matDC treated with 10 μg/ml anti-DC-SIGN MAb. TNF-α and IL-6 production from matDC and anti-DC-SIGN-treated matDC (right panel). Experiments were performed in triplicate, and results are expressed as means ± standard errors of the means for three different donors. The dashed line indicates the lower limit of detection for the assay (20 pg/ml).

FIG. 7.

Comparison of infection patterns in the presence and absence of DV-immune serum. mDC, matDC. (A) Percent DV infection under the influence of DV-immune serum at the PENT and without immune sera (MOI of 1 or 5). (B) TNF-α (white bar) and IL-6 (gray bar) production from matDC under ADE conditions and without DV-immune serum. (C) Vero cell plaque assay from supernatants of matDC undergoing ADE (MOI of 1 plus 1/640 sera) or inoculated with DV2 at an MOI of 1 or 5. Experiments were performed in quadruplicate, and results are expressed as means ± standard errors of the means for three different donors.