Complement-dependent virion lysis mediated by dengue-Zika virus cross-reactive antibodies correlates with protection from severe dengue disease

Abstract

Primary infection with one of four dengue virus serotypes (DENV1-4) may generate antibodies that protect or enhance subsequent secondary heterotypic infections. However, the characteristics of heterotypic cross-reactive antibodies associated with protection from symptomatic infection and severe disease are not well-defined. We selected plasma samples collected before a secondary DENV heterotypic infection that was classified either as dengue fever (DF, n = 31) or dengue hemorrhagic fever/dengue shock syndrome (DHF/DSS, n = 33) from our longstanding pediatric cohort in Nicaragua. We screened various antibody properties to determine the features correlated with protection from DHF/DSS. Protection was associated with high levels of binding of various antibody isotypes, IgG subclasses and effector functions, including antibody-dependent complement deposition, ADCD. Although the samples were derived from DENV-exposed, Zika virus (ZIKV)-naïve individuals, the protective ADCD association was stronger when assays were conducted with recombinant ZIKV antigens. Further, we showed that a complement-mediated virion lysis (virolysis) assay conducted with ZIKV virions was strongly associated with protection, a finding reproduced in an independent sample set collected prior to secondary heterotypic inapparent versus symptomatic DENV infection. Virolysis was the main antibody feature correlated with protection from DHF/DSS and severe symptoms, such as thrombocytopenia, hemorrhagic manifestations, and plasma leakage. Hence, anti-DENV antibodies that cross-react with ZIKV, target virion-associated epitopes, and mediate complement-dependent virolysis are correlated with protection from secondary symptomatic DENV infection and DHF/DSS. These findings may support the rational design and evaluation of dengue vaccines and development of therapeutics.

Fig. 1.. Binding and neutralization titers of pre-infection antibodies are not significantly associated with protection from subsequent severe dengue disease.

(A) Plasma samples collected prior to a secondary heterotypic infection resulting in non-severe dengue (DF; green) or severe dengue disease (DHF/DSS; gold) were selected for this systems serology study. (C) Inhibition enzyme-linked immunosorbent assay (iELISA) binding antibody and (C-G) neutralizing antibody titer (NT₅₀) of pre-DF and pre-DHF/DSS groups measured against DENV1-4 and ZIKV by FRNT on Vero cells with mature virions. Shown are median NT₅₀ (middle line), 25^th to 75^th percentile (box), and 5^th to 95^th percentile (whiskers) as well as the raw data (points). Asterisks indicate Benjamini-Hochberg-adjusted p-values for Mann-Whitney U tests (ns, non-significant).

Fig. 2.. Complement-mediated virolysis with ZIKV-cross-reactive antibodies is associated with protection from subsequent DHF/DSS and symptomatic DENV infection.

(A-D) Pre-infection plasma samples prior to DF (green; n=31) vs. DHF/DSS (gold; n=33) were incubated with mature DENV2, mature DENV3, ZIKV, and YFV virions. (A,B) Left panel, 1:10 dlution of plasma; right panel, 1:30 dilution of plasma. (E) Pre-inapparent (blue circles; n=30) and pre-symptomatic (orange circles; n=29) secondary DENV3 infection plasma samples were incubated with ZIKV virions. After incubation with plasma samples in the presence of guinea pig complement, mixtures were freeze-thawed once, followed by RNAse A digestion and RNA extraction from intact virions. Virion lysis was quantified by measuring viral RNA via RT-qPCR (to measure the number of genome copies) and calculated as percentage (%) of virion lysis as described in Methods. Shown are median NT₅₀ (middle line), 25^th to 75^th percentile (box), and 5^th to 95^th percentile (whiskers) as well as the raw data (points). Asterisks indicate Benjamini-Hochberg-adjusted p-values for Mann-Whitney U tests (****p < 0.0001, and ns, non-significant).

Fig. 3.. Variables discriminating DHF/DSS and DF vary depending on incoming serotype but overall, ZIKV virolysis is the main discriminating antibody feature.

Multivariate regression models with regularization, incorporating variables that showed significant differences in the bivariate analysis, were used to identify discriminant features between subsequent DF vs. DHF/DSS. (A-B) Multivariate regression model with regularization without stratifying by incoming serotype. (C-D) Multivariate regression model with regularization stratified for incoming DENV2. (E-F) Multivariate regression model with regularization stratified for incoming DENV3. (A,C,E) Receiver operator curve (ROC) analysis of the regression models, with the area under the curve (AUC) values displayed. (B,D,F) Variable of importance plot (VIP) derived from the regression models. The red dashed line indicates a VIP value of 1. The regression model without stratification (regardless of the incoming serotype) identified ZIKV virolysis as the most relevant variable enriched in DF (A-B). The analysis stratified by incoming serotype of pre-DENV2 (C-D) and pre-DENV3 (E-F) infection improved the performance of the model and revealed differences in variables discriminating subsequent DHF/DSS vs. DF, including ZIKV virolysis.

Fig 4.. Complement-dependent virolysis mediated by ZIKV-cross-reactive antibodies is correlated with protection from symptoms and signs of DHF/DSS.

Forest plot showing the adjusted odds ratios (OR), confidence intervals (CI), and p-value of the correlation between DHF/DSS or symptoms of DHF/DSS (thrombocytopenia, hemorrhagic manifestations, and plasma leak) and ZIKV virolysis. Odds ratios (OR) were adjusted by incoming serotype (interacting variable), infection history, and time since last infection. OR are shown with 95% confidence intervals (95% CI).