Longitudinal analysis of DENV-specific neutralizing antibodies following primary DENV3 infection.

Following primary DENV infection, a durable serotype-specific neutralizing response can be detected years after exposure (14). To determine whether the 5J7 epitope is recognized by the DENV3 type-specific antibody response in a setting where dengue is endemic, neutralization assays were performed with primary DENV3 plasma collected 3 and 18 months post-infection and the rDENV4/3-M14 virus displaying the partial 5J7 epitope (D. Widman, E. Young, U. Nivarthi, S. R. Royal, J. Swanstrom, L. Y. Boyd, K. Debbink, M. Begley, A. M. de Silva, W. B. Messer, and R. Baric, unpublished) (Fig. 1A). Sigmoidal dose-response curves representative of one primary DENV3 infection show neutralization of the rDENV4/3-M14 virus at 3 and 18 months post-infection at levels close to parental DENV3 at the later time point (Fig. 1C and ​andD).D). As expected, in all subjects, primary DENV3 plasma strongly neutralized DENV3, with titers significantly higher than those to DENV4 virus (Fig. 1E). The neutralizing titers to the parental strain UNC 3001 and the infecting DENV3 Nicaraguan strain were similar at both time points (see Fig. S1 in the supplemental material), as both are members of DENV3 genotype III (15). Plasma from most individuals neutralized the rDENV4/3-M14 virus, and some neutralized it at levels similar to DENV3 (Fig. 1E). This indicates that DENV3 type-specific neutralizing antibodies in human DENV3-immune plasma recognized the 5J7 amino acid residues transplanted into the DENV4 backbone. The longitudinal analysis showed a significant increase in the neutralizing titers to the rDENV4/3-M14 at 18 months (P < 0.001), while the neutralizing titers to the parental viruses did not show significant variation between the time points analyzed (Fig. 1E).

Open in a separate window

FIG 1 

Neutralizing antibody titers to the chimeric rDENV4/3-M14 virus and parental DENV3 and DENV4 viruses 3 and 18 months post-primary DENV3 infection. (A) Ribbon diagram of the DENV E trimer (domains I, II, and III in red, yellow, and blue, respectively), showing amino acid residues within the 5J7 epitope spanning the A, B, and B′ monomers (purple and teal spheres); the residues represented by the teal spheres were transplanted into a DENV4 backbone, creating the rDENV4/3-M14 virus. (B) As in panel A, but the residues represented by the teal spheres were transplanted into a DENV4 backbone to create the rDENV4/3-M16 virus. (C and D) The raw antibody titration data were fitted with a four-parameter sigmoidal dose-response curve to estimate the 50% neutralizing antibody titer (NT₅₀) to the rDENV4/3-M14 and parental viruses DENV3 and DENV4 in plasma samples 3 and 18 months postinfection, respectively. (E) Geometric mean of the NT₅₀ values to the rDENV4/3-M14 and parental viruses of DENV3 and DENV4 in plasma samples 3 and 18 months postinfection. Data are representative of two independent experiments processed in duplicate for each plasma sample. The NT₅₀ values were compared by two-way ANOVA (n = 30). ***, P < 0.001; ****, P < 0.0001.

Correlation between neutralizing antibody titers to rDENV4/3-M14 and DENV3 at 3 and 18 months post-primary DENV3 infection.

Following primary infection, a type-specific response to the infecting serotype and a cross-reactive response to heterologous DENV serotypes is developed (7). In individuals who experienced a primary DENV3 infection, the correlation between the DENV3 50% neutralizing antibody titer (NT₅₀) and the DENV4 cross-reactive NT₅₀ in plasma was not significant at 3 and 18 months post-infection (Fig. 2A and ​andB).B). To investigate how the rDENV4/3-M14 virus neutralizing antibody titers track with the DENV3 response, correlation analyses between rDENV4/3-M14 and DENV3 NT₅₀ values were performed at 3 and 18 months post-infection. At both time points analyzed, this correlation was significant (Fig. 2C and ​andD).D). Thus, the increased positive correlation between DENV3 and rDENV4/3-M14, compared to the correlation between DENV3 and DENV4, is due to inclusion of the 5J7 epitope transplant in the DENV4 backbone.

Open in a separate window

FIG 2 

Correlation analysis between DENV3 and DENV4 or rDENV4/3-M14 neutralizing antibody titers over time. (A and B) Correlation between DENV3 NT₅₀ and DENV4 NT₅₀ at 3 and 18 months following primary DENV3 infection, respectively. (C and D) Correlation between DENV3 NT₅₀ and rDENV4/3-M14 NT₅₀ at 3 and 18 months following primary DENV3 infection, respectively. Data are representative of two independent experiments performed in duplicate for each plasma sample (n = 30). Pearson correlation coefficients (r) were calculated between the pairs of NT₅₀ values. Linear regression lines are shown.

Proportion and kinetics of the DENV3 type-specific neutralizing response attributable to the 5J7 epitope.

The proportion of the DENV3 type-specific response attributable to the 5J7 epitope recognition was measured as (rDENV4/3-M14 NT₅₀ − DENV4 NT₅₀)/(DENV3 NT₅₀ − DENV4 NT₅₀), the ratio between rDENV4/3-M14 titers and DENV3 NT₅₀ from which the cross-reactive component (DENV4 NT₅₀) had been subtracted. The mean proportion of DENV3 type-specific response targeted to the 5J7 epitope was 22.6% at 3 months post-infection, and at 18 months post-infection, the proportion significantly increased to 41.4% (P = 0.0184) (Fig. 3A). The proportions were not significantly different between individuals who experienced DF versus DHF (see Fig. S2 in the supplemental material).

Open in a separate window

FIG 3 

Proportion and kinetics of the DENV3 type-specific neutralizing antibody response attributable to the 5J7 epitope. (A) The proportion of the DENV3 type-specific response directed to the 5J7 epitope obtained with 3- and 18-month post-infection samples. Results for 3 and 18 months (n = 30) were compared by paired t test, where * represents P < 0.05. (B and C) Proportion of the rDENV4/3-M14 and DENV4 neutralizing antibody response relative to DENV3 NT₅₀, where the y axis represents 100% of the DENV3 NT₅₀ values at 3 and 18 months post-infection, respectively. Individual codes are ordered by the proportion of DENV3 attributable to rDENV4/3-M14 observed at 18 months. (D) Individual trajectories of the proportion of the DENV3 type-specific response directed to the 5J7 epitope at 3 and 18 months (n = 30). (E) Pie chart representing the percentage of individuals who displayed a constant, decreased, or increased proportion of the rDENV4/3-M14 DENV3 response between 3 and 18 months post-primary DENV3 infection (n = 30). (F) Antigenic cartography map generated from NT₅₀. The antigenic cartography map positions viruses (closed circles [rDENV4/3-M14, DENV3, and DENV4]) and plasma (open squares) as points, with the distance between each virus and plasma derived from its respective neutralization titer (see Materials and Methods). Each grid square corresponds to a 2-fold dilution in the neutralization titer. Each open square represents one of the 30 plasma samples analyzed.

To visualize the proportions of the DENV3 neutralizing antibody response attributable to cross-reactive neutralization (DENV4 NT₅₀) or the 5J7 epitope (rDENV4/3-M14 NT₅₀ not explained by DENV4 NT₅₀), we plotted the ratios of DENV4 NT₅₀ to DENV3 NT₅₀ and rDENV4/3-M14 NT₅₀ to DENV3 NT₅₀, where the percentage of the DENV3 neutralizing antibody response is represented on the y axis (Fig. 3B and ​andC).C). The 30 individuals are aligned in an ascending order of proportion of the DENV3 response explained by rDENV4/3-M14 at 18 months. It is evident that, at both 3 and 18 months, there was extensive variability in the amount to which the DENV3 response is explained by 5J7 or cross-reactive neutralization. Further, there was variation by individual in changes in neutralizing antibody responses over time, with most individuals having an increasing proportion of the DENV3 response attributable to the 5J7 epitope, while several lost 5J7 recognition (Fig. 3C).

As the prevalence of this type-specific epitope among individuals living in regions where dengue is endemic is an important guide for vaccine design, the individual variation of the proportion was further analyzed. The proportion of the DENV3 neutralizing antibody response directed to the 5J7 epitope ranged from 0 to 97% at 3 months and from 6 to 100% at 18 months post-infection across the 30 individuals studied (Fig. 3A to ​toC).C). The proportions of the DENV3 neutralizing response attributable to the 5J7 epitope were defined as low (0 to 10%), moderate (10 to 30%), and high (above 30%). At 3 months, 57% (17/30) showed inexistent to low recognition of the 5J7 epitope, 13% (4/30) displayed moderate recognition, and 30% (9/30) displayed high recognition of the 5J7 epitope. At 18 months, 13% (4/30), 47% (14/30), and 40% (12/30) of the individuals displayed low, moderate, and high recognition of the 5J7 epitope, respectively. Matched-pair analysis enabled assessment of the change of the proportion in each individual over time (Fig. 3D). Using 1 standard deviation of the average of the proportions at 3 and 18 months as the criterion to define change, 50% (15/30) of the individuals displayed an increase in the proportion over time, while 37% (11/30) showed no alteration and 13% (4/30) had a decreased proportion (Fig. 3E).

Next, we analyzed the data using antigenic cartography, where the NT₅₀ was treated as a “distance” between antisera and rDENV4/3-M14 as well as its parental viruses. A convergent pattern over time was observed, with rDENV4/3-M14 and DENV3 moving toward each other at 18 months post-primary DENV3 infection and DENV4 virus staying about the same distance from the primary DENV3 anti-sera (Fig. 3F). This indicates greater antigenic similarity between DENV3 and rDENV4/3-M14 as recognized by the primary infection antisera at the later time point, with the antibody response becoming increasingly specific to the 5J7 epitope over time.

Trajectories of the rDENV4/3-M14, DENV3, and cross-reactive neutralizing responses.

While the cross-reactive neutralizing antibody response has been described to wane over time (7), recent studies in settings where dengue is endemic have described the maintenance of cross-reactive titers, potentially due to “boosting” by homotypic or heterotypic reinfection that fell short of the antibody threshold for a new infection (16, 17). To investigate whether the constant, increased, or decreased proportion of the neutralizing antibody titers targeting the 5J7 epitope between 3 and 18 months is associated with the change in magnitude of homologous and heterologous titers, samples collected at 6 and 12 months from a subset of individuals (n = 12) were used for neutralization assays to rDENV4/3-M14 and all DENV serotypes (DENV1 to 4) (Fig. 4; see Fig. S3 in the supplemental material). In this analysis, in addition to DENV4, the cross-reactive titers as averages of anti-DENV1, 2, and 4 titers were plotted. As an example of a constant proportion between 3 and 18 months, one subject (Fig. 4A) displayed rDENV4/3-M14 titers that followed similar kinetics to the DENV3 NT₅₀ at 3, 6, 12, and 18 months, while the cross-reactive titers decayed over time. This pattern was observed in 42% (5/12) of the individuals analyzed. As an illustration of an increased proportion, subject 1228 (Fig. 4B) showed an increase in DENV3 and rDENV4/3-M14 titers between 3 and 6 months in parallel with a slight decrease of cross-reactive titers; one case (8%) with this pattern was seen. Another example of increased titers from 3 to 18 months is shown with subject 1231 (Fig. 4C), where the increase of DENV3 and rDENV4/3-M14 titers is observed at the later time points—between 12 and 18 months. This pattern, observed in 33% (4/12) of the subjects, suggests a possible homotypic boost, whereby reinfection with DENV3 may have occurred (18), as the effect is only seen in the homologous titers (Fig. 4C). Finally, the pattern of decreased proportion is observed in subject 1513 (Fig. 4D), where an increase in neutralizing titers to the heterologous serotypes is seen at the later time points—between 12 and 18 months—which could indicate heterotypic boosting. This pattern was observed in 17% (2/12) of the individuals analyzed.

Open in a separate window

FIG 4 

Trajectories of neutralizing antibody responses 3, 6, 12, and 18 months post-primary DENV3 infection. The neutralizing antibody titers to rDENV4/3-M14, DENV3, DENV4, and the overall cross-reactive titers (geometric mean of DENV1, DENV2, and DENV4 NT₅₀ values) from 12 individuals were analyzed. (A) rDENV4/3-M14 titers follow similar trajectory to DENV3 titers in an individual with a constant proportion of the 5J7 response. (B) Increase of rDENV4/3-M14 titers between 3 and 6 months, with parallel decay of cross-reactive titers, indicates affinity maturation in an individual who displayed an increased proportion of the 5J7 response. (C) Increase in rDENV4/3-M14 titers between 12 and 18 months indicates possible homotypic reexposure in an individual with an increased proportion of the 5J7 response. (D) Increase in cross-reactive titers between 12 and 18 months indicates possible heterotypic reexposure in an individual who displayed a decreased proportion of the 5J7 response over time.

Proportion of the DENV3 neutralizing response targeted to the 5J7 epitope years after infection.

To test whether the durability of the DENV3 neutralizing antibody response targeted to the 5J7 epitope is maintained beyond 18 months post-infection, primary DENV3 plasma samples collected up to 4 years after infection in a long-term cohort study were analyzed. While the rDENV4/3-M14 neutralizing titers did not change across the years (Fig. 5A), the proportion trended toward an increase between year 1 and year 2 after infection (Fig. 5B). From year 1 through year 4, the mean proportions were 8, 25, 29, and 34%, respectively, across the 10 individuals analyzed (Fig. 5C). The individual trajectories of the proportion indicate variations of 0 to 0.26, 0 to 0.80, 0 to 0.87, and 0 to 0.89 at years 1 to 4 (Fig. 5C). The long-term analysis of the titers to the rDENV4/3-M14 virus indicates that the response to the 5J7 epitope is stable and durable.

Open in a separate window

FIG 5 

Proportion of the DENV3 type-specific neutralizing response targeted to the 5J7 epitope up to 4 years after infection in a cohort study design. (A) Neutralizing titers to the rDENV4/3-M14 virus from year 1 through year 4 post-primary DENV3 infection in a cohort study setting. (B) The ratio of the 5J7 epitope neutralizing antibody response (rDENV4/3-M14) to that of the parental DENV3 was analyzed from 1 to 4 years after primary DENV3 infection. (C) Individual trajectories of the DENV4/3-M14/DENV3 proportions from year 1 through year 4 post-primary DENV3 infection. Data are representative of two independent experiments processed in duplicate for each plasma sample. The NT₅₀ and the proportion of the 5J7 response were compared by one-way ANOVA (n = 10); no significant differences were observed.

Study population. (i) Dengue hospital-based study.

Study enrollment took place in the Nicaraguan national pediatric reference hospital, Hospital Infantil Manual de Jesús Rivera (HIMJR). Children between 6 months and 14 years old suspected of dengue (<7 days of illness) were eligible to participate in the hospital study, as described previously (35). Laboratory-confirmed dengue cases were classified by severity using a computerized algorithm that compiled all clinical data meeting each criterion for dengue fever (DF), dengue hemorrhagic fever (DHF), or dengue shock syndrome (DSS), as detailed in the 1997 WHO guidelines (1, 35). Based on these guidelines, 20 individuals were classified as DF, while 10 experienced DHF. Plasma samples were collected in the acute (days 1 to 6) and convalescent (days 14 to 28) phases, as well as 3, 6, 12, and 18 months post-onset of illness.

(ii) Pediatric Dengue Cohort Study.

The Nicaraguan Pediatric Dengue Cohort Study (PDCS; 2004 to present) is a community-based prospective study of approximately 3,500 children 2 to 14 years of age in Managua, Nicaragua (36). Healthy annual blood samples collected from 10 participants 1, 2, 3, and 4 years after experiencing a primary DENV3 infection were used.

Laboratory tests.

DENV infection was confirmed by RT-PCR for detection of viral RNA (37, 38), isolation of DENV on C6/36 cells (37), and/or seroconversion by IgM enzyme-linked immunosorbent assay (ELISA) or a ≥4-fold increase in total antibody titer as measured by inhibition ELISA in paired acute- and convalescent-phase samples. In the hospital study, primary dengue cases were determined by inhibition ELISA (39, 40), where antibody titers of <2,560 in the convalescent-phase sample (day 14 to 28 post-onset of symptoms) define primary infection status (13). In the cohort study, a primary DENV infection was classified by seroconversion (a titer of <1:10 to >1:10 as determined by inhibition ELISA) in paired consecutive annual samples (36).

Cells and viruses.

U937 cells expressing DC-SIGN (dendritic cell-specific intercellular adhesion molecule-3-grabbing nonintegrin), a known DENV attachment factor, were maintained as suspension cell cultures at 37°C with 5% CO₂ in RPMI 1640 (Gibco) supplemented with 1% non-essential amino acids, 1% penicillin and streptomycin, and 5% fetal bovine serum (FBS; HyClone). Aedes albopictus C6/36 cells were grown at 32°C in 5% CO₂ and were used for propagation of the Nicaraguan strains of DENV1 (N1265), DENV2 (N172), and DENV3 (N7236), parental strains of DENV3 (UNC 3001) and DENV4 (Haiti 74), and recombinant viruses rDENV4/3-M14 and rDENV4/3-M16. The rDENV4/3-M14 virus contains 29 amino acid residues within the 5J7 epitope, spanning two E protein monomers, whereas the rDENV4/3-M16 virus contains 36 amino acid residues from the 5J7 epitope spanning three E protein monomers (Widman et al., unpublished).

DENV neutralization assay.

A flow cytometry-based neutralization assay was used to measure DENV-specific neutralizing antibodies, as previously described (41). Briefly, DENV-immune plasma at an initial dilution of 1:5 were serially diluted 3-fold 8 times in RPMI supplemented with 2% FBS. An amount of virus that infects 15% of the cells (previously determined by virus titration) was added to the plasma dilutions and incubated for 45 min at 37°C. After 24 h, the cells were centrifuged at 252 × g for 5 min and resuspended in 100 µl of RPMI medium. Next, cells were fixed in 4% paraformaldehyde, incubated for 10 min at room temperature, and centrifuged at 252 × g for 5 min. Then, cells were blocked in permeabilization buffer (0.1% saponin, 5% bovine serum albumin in 1× phosphate-buffered saline [PBS]) for 30 min at room temperature. Cells were subsequently incubated with anti-E MAb 4G2 conjugated to Alexa 488, diluted 1:1,000 in blocking buffer (0.5% bovine serum albumin and 0.02% sodium azide in 1× PBS) for 25 min at room temperature. Cells were washed and resuspended in PBS. The percentage of infected cells was determined using a Guava flow cytometer (EMD Millipore) by gating Alexa 488-positive cells. The plasma dilution that reduced viral infection by 50% (50% neutralizing antibody titer [NT₅₀]) was calculated by a nonlinear, 4-parameter dose-response regression analysis with Prism software (GraphPad), which is expressed as the reciprocal serum dilution. To eliminate the effect of nonspecific cross-reactivity in human plasma, DENV-naïve titers (average of 10 DENV-naïve individuals enrolled in the PDCS) were subtracted from DENV-immune titers. Quality control criteria for the sigmoidal dose-response regression fit included an absolute sum of squares of <0.2 and a coefficient of determination (R²) of >0.9.

Antigenic cartography map.

Antigenic maps were generated from NT₅₀ against DENV3 UNC 3001, DENV4 Haiti 74, and rDENV4/3-M14 obtained from antisera collected at 3 and 18 months post-infection, as previously described (42). The measured antigenic distance, D_ij, between virus i and antiserum j was estimated as D_ij = log₂(b_j) − log₂(N_ij), where b_j is the NT₅₀ titer for the virus best neutralized by each antiserum j, and N_ij is the NT₅₀ for virus i and antiserum j. Error function, E = ∑ije(D_ij, d_ij), was minimized (5,000 independent optimizations) to estimate coordinates for viruses and antisera that produced antigenic maps with the least disagreement between Euclidean distances, d_ij, and measured antigenic distances, D_ij. Error was defined as e(D_ij, d_ij) = (D_ij − d_ij)² when the NT₅₀ was within the limit of detection of the assay and e(D_ij, d_ij) = (D_ij − 1 − d_ij)²{1/[1 + e^{−10(dij − 1 − dij)}]} for NT₅₀ values below the assay limit of detection. The minimum error antigenic map is shown on a grid indicating antigenic distances as 2-fold dilutions, in any direction, on the antigenic map. The shift in virus position from the 3- to 18-month maps is shown with arrows.

We thank members of the study team based at the Hospital Infantil Manuel de Jesús Rivera, the Centro de Salud Sócrates Flores Vivas, the National Virology Laboratory in the Centro Nacional de Diagnóstico y Referencia, and the Sustainable Sciences Institute in Nicaragua for their dedication and high-quality work, as well as the children who participated in the studies and their families.

This research was supported by NIAID/NIH grant P01 AI106695 (to E.H.) and a Coordenacao de Aperfeicoamento de Pessoal de Nivel Superior (CAPES) fellowship (to D.V.A.).