Introduction

Dengue fever (DF) is an important mosquito-borne disease caused by infection with dengue virus (DENV). As there is no protective vaccine or specific treatment available currently, dengue fever remains a major public health concern. DENV has four distinct serotypes (DENV-1, -2, -3, and -4; Kuno et al. 1998) and infection with any of the four serotypes can cause a spectrum of clinical features ranging from asymptomatic infections and mild DF to potentially life-threatening dengue hemorrhagic fever (DHF) and dengue shock syndrome (DSS; Guzman and Isturiz 2010). Moreover, infection with one serotype confers only short-lived cross-protection against the other three serotypes, and to make things worse, immunity to one serotype may increase the risk of developing a severe disease upon exposure to subsequent heterologous infections. Antibody-dependent enhancement (ADE), a means of enhancing DENV infection in Fcγ receptor (FcγR)-bearing cells by pre-existing subneutralizing antibodies, has been implicated as one of the main pathogenic mechanisms of disease exacerbation into DHF/DSS (Halstead 2003). An increased risk of ADE is associated with the difficulty of inducing protective immunity against all four DENV serotypes and presents a persistent hindrance to the development of dengue vaccines. Accordingly, current vaccine development has focused on producing a tetravalent vaccine to provide life-long protection simultaneously against all four DENV serotypes to avoid the risk of ADE. However, the lack of an appropriate animal model for dengue disease further limits our understanding of vaccine-induced antibody responses, thus making it necessary to rely on data obtained from in vitro studies. In addition, functional evaluations of enhancing antibodies elicited among DENV-infected populations during natural infections in endemic regions also depend on in vitro approaches.

In a previous study, we established a high-throughput enzyme-linked immunospot-based micro-neutralization test (ELISPOT-MNT) to determine neutralizing antibody activity (Liu et al. 2012). To assess concomitant ADE risks of corresponding dengue antibody response, efficient and high-throughput methods for in vitro ADE assay are also anticipated. Early investigators developed ADE assay methodology in the studies of infection enhancement of DENV, such as the infectious titer determination and the infectious centre assay (Halstead et al. 1983; Halstead and O’Rourke 1977; Halstead et al. 1977; Morens and Halstead 1990). Then flow cytometry was introduced into ADE assay and applied in many studies of dengue infection enhancement (Guy et al. 2004). Currently, the other tools available for measuring ADE activity include an immunofluorescent staining assay to determine the percentage of infected cells (Ito et al. 2010) and reverse transcriptase polymerase chain reaction (RT-PCR) to detect viral genomes (Moi et al. 2011). The labor-intensive and time-consuming procedures inherent in these early and current ADE assays preclude their application in large-scale studies, particularly in large dengue seroprevalence studies, as these studies usually highlight the need for a relatively high sample throughput. Accordingly, we managed to develop an alternative ADE assay with more simplicity and adequate throughput required for ADE assessment, using an appropriate detection target. ELISA, a commonly used immunoassay, can be easily performed in most laboratories and is known to generate considerable throughput. In addition, nonstructural protein 1 (NS1) was demonstrated as a diagnostic marker of DENV infection (Huang et al. 2001; Young et al. 2000) and the amount of secreted NS1 is closely related to infectious dengue virus titers (Alcon-LePoder et al. 2005; Huang et al. 2006). Thus, enhanced infection can be detected through NS1 production in culture supernatants using an NS1 antigen capture ELISA.

In this study, we first developed a novel NS1 capture ELISA-based ADE assay (ELISA-ADE) with FcγR-expressing K562 cells and a known enhancing antibody for ADE risk assessment. Then we evaluated the accuracy of this ELISA-ADE assay using an infectious virus titration method. Enhancing activities quantified by measurement of kinetics of NS1 production were in a good agreement with those of the virus titration assay. We further evaluated and analyzed enhancing and neutralizing antibody activities of two flavivirus cross-reactive murine monoclonal antibodies (mAbs; 4G2 and 2A10G6) and a panel of four DENV-1-infected human serum samples by ELISA-ADE established in this study and a previously developed ELISPOT-MNT. The results demonstrated observable relationship between the dose-dependent profiles of enhancing and neutralizing activities.

Materials and methods

Viruses and cells

Four dengue serotypes (DENV-1, Hawaii; DENV-2, New Guinea-C; DENV-3, Guanxi-80-2; and DENV-4, H241) were used in this study. LLC-MK2 cells (ATCC Number: CCL-7) were grown at 37 °C with 5 % CO2 in MEM containing 10 % FBS (Gibco). Viral propagation was performed in C6/36 cells (ATCC Number: CRL-1660) in MEM containing 10 % FBS at 33 °C for 3–5 days with 5 % CO2. The infectivity of each virus pool was titrated and viral stocks were kept frozen at −80 °C until used. Human erythroleukaemic FcγRIIa-expressing K562 cells (ATCC Number: CCL-243) were grown in RPMI 1640 medium supplemented with 10 % FBS.

Monoclonal antibodies and human serum samples

Flavivirus cross-reactive fusion-loop murine mAbs 4G2 (IgG2a; Henchal et al. 1985) and 2A10G6 (IgG1; Deng et al. 2011) were used in this study. The collection and use of human immune sera was approved by the Ethics Committee of the Zhujiang Hospital of Southern Medical University under the reference number ZJYY-2010-YXJYZX-003. The written informed consent was obtained from all human participants. The human convalescent serum samples were collected from four patients exposed to primary DENV-1 infection, which were confirmed by RT-PCR and a dengue IgG and IgM capture ELISA (Panbio, Australia). Serum samples were heat inactivated at 56 °C for 30 min before the determination of ADE and neutralizing activities. Irrelevant IgG1 and IgG2a mAbs and healthy human serum were included as negative controls.

NS1 capture ELISA-based ADE assay (ELISA-ADE)

The mAbs were serially diluted two-fold from 200 μg/mL and human serum samples from 1:10 dilution in serum-free RPMI 1640 medium. The antibody–virus complexes were prepared in 96-well plates by mixing 50 μL of the mAb or serum dilution with an equal volume of DENV at a multiplicity of infection (MOI) of 0.5 or 0.125 and incubated at 37 °C for 1 h. For viral control in the absence of antibodies, 50 μL of serum-free RPMI 1640 medium was substituted for the antibody dilution. Next, K562 cells were washed once, adjusted to a concentration of 5 × 104/100 μL with serum-free RPMI 1640 medium, and added to the virus–antibody mixture in the 96-well plates. Following incubation at 37 °C for 2 h, K562 cells were washed twice and incubated at 37 °C for 96 h in RPMI 1640 medium supplemented with 5 % FBS. The 96-well plates were placed at −80 °C overnight. To quantify viral infection levels with or without antibodies, NS1 level was determined using an NS1 antigen-capture ELISA developed previously (Ding et al. 2011). Specifically, the −80 °C freeze–thawed 96-well plates were subjected to inactivation at 56 °C for 30 min. Then, after clarification by centrifugation, inactivated supernatants were harvested and transferred (100 μL/well) to 96-well plates (Costar Corning, Inc., Corning, NY) coated with four anti-NS1 mAbs. Another HRP-conjugated anti-NS1 mAb was used as a detection antibody. The following procedures were essentially the same as described previously. To differentiate enhancing and non-enhancing antibody activities, a cut-off value was set as a mean absorbance at 450 nm (A 450) plus three times of the standard deviation value from eight wells of the viral control in the absence of antibodies.

Virus titration of ADE by TCID50

A TCID50 assay established previously (Li et al. 2011) to titrate infectious virus titers was used to validate our current ELISA-ADE. Infection enhancement of representative DENV-1 by mAb 4G2 was performed using the infection procedures described above. After freeze-thaw cycles at −80 °C, enhanced supernatants were clarified, inactivated, and divided into two aliquots. One aliquot was used for the ELISA-ADE assay as mentioned above and the other for infectious titer analysis by TCID50.

ELISPOT-MNT

The ELISPOT-MNT was performed as described previously with some modifications (Liu et al. 2012). In brief, an antibody–virus mixture was prepared by the addition of 75 μL DENV (diluted to give about 200 PFU per well) to an equal volume of mAb or serum dilution and incubated at 37 °C for 1 h. Next, 100 μL of each mixture was added to the LLC-MK2 cell monolayers and incubated at 37 °C for 1.5 h. LLC-MK2 cells were overlaid with MEM medium containing 1 % methylcellulose (Sigma) and 5 % FBS and maintained at 37 °C for 2 or 3 days. After fixation, immunostaining of DENV NS1 spots was performed by incubating with the anti-DENV NS1 mAb 5F10A7, followed by HRP-conjugated anti-mouse IgG. Immunostained spots were finally developed using the AEC substrate kit (SK-4200, Vector Laboratories) and then counted using an automated ELISPOT instrument (Cellular Technology Ltd.). A 50 % reduction in immunostained spot number compared with the control in the absence of a test antibody was calculated as described previously.

Data analysis

All statistical analyses were performed using GraphPad Prism version 5.04. Correlations between ELISA-ADE and TCID50 were evaluated by linear regression analysis. The 50 % inhibitory concentration (IC50) value of each mAb and 50 % neutralizing titers (NT50) of the serum samples, which corresponds to a 50 % plaque reduction, were calculated by nonlinear regression (curve fit) analysis.

Results

Development and evaluation of the ELISA-ADE assay

To develop an effective ELISA-ADE assay, we first determined and optimized several factors, including the cell lines, MOI, and cultivation periods to obtain maximum NS1 production. The flavivirus cross-reactive mAb 4G2 (IgG2a), which is known to enhance infection of all four DENV serotypes (Henchal et al. 1985) was employed to establish the ELISA-ADE assay. K562 cells were chosen with emphasis on method establishment in this current study for their slightly higher NS1 production, demonstrated in a preliminary experiment, in a dose-dependent manner in comparison with U937 cells at the same MOI (data not shown). The optimal MOI was 0.5 for strains DENV-1, -3, and -4, and 0.125 for strain DENV-2 considering both the preferentially lower background of NS1 production in the absence of antibodies (approximate A 450 value of 0.15) and higher antibody-enhanced NS1 production. The infection level of K562 cells by DENV alone in the absence of antibodies was set as the infection baseline from which fold enhancement was determined.

Since the amount of secreted NS1 is closely related to infectious dengue virus titers (Alcon-LePoder et al. 2005), we next evaluated whether the ADE profile reflected by elevating, peaking and subsequently decreasing NS1 production corresponded to an identical or similar pattern measured via viral titer levels. 4G2-mediated enhanced profiles of DENV-1 infection were determined simultaneously by this ELISA-ADE assay and a TCID50 assay. The ELISA-ADE revealed a typical ADE phenomenon as measured by NS1 production (Fig. 1a), which was consistent with the graphically enhanced ADE profile in a concentration-dependent pattern (Morens et al. 1987). Infectious titers determined by the TCID50 assay also demonstrated a similar characteristic enhancement profile (Fig. 1a). Enhanced infectious virus titers, after subtraction of unenhanced control titers without 4G2 antibody, ranged from 3.6 × 101 to 1.93 × 104 PFU/mL, corresponding to the lowest and peak levels of viral replication. This simultaneous determination depicted the same trend of 4G2-mediated infection enhancement of DENV-1 in a dose-dependent manner (Fig. 1a). The NS1 A 450 values and infectious virus titers of the same culture supernatants were significantly correlated, with a correlation coefficient of 0.938 (P = 0.000; Fig. 1b). ADE was not observed for the IgG2a control mAb (with the same isotype as 4G2; data not shown).

Fig. 1
figure 1

Correlation between ELISA-ADE and TCID50 assay of 4G2-mediated infection enhancement of DENV-1. a A 450 values determined by ELISA-ADE assay were plotted on the left Y axis and infectious virus titers by TCID50 on the right Y axis. b Scatter plots of A 450 values and infectious virus titers from the two assays are shown. The central solid line is the best fit line to the data by linear correlation with 95 % confidence intervals indicated by the two dashed lines. The two assays displayed an excellent agreement (R = 0.938, P = 0.000)

Full size image

Relationship between enhancing and neutralizing activities of the flaviviral reactive mAbs 4G2 and 2A10G6

We next determined the enhancing and neutralizing activities of the flaviviral cross-reactive mAbs 4G2 (IgG2a) and 2A10G6 (IgG1) using ELISA-ADE and ELISPOT-MNT assays. The two mAbs exhibited similar concentration-dependent ADE curves, demonstrating enhancing activities over an approximate 103-fold dilution range (Figs. 2 and 3). The fold enhancement of 4G2 was approximately 5-fold for DENV-1, 3.5-fold for DENV-2 and -3, and 6-fold for DENV-4, respectively (Fig. 2). Similarly, fold enhancement of 2A10G6 was 5-, 3-, 4-, and 8-fold for DENV−1, -2, -3, and -4, respectively (Fig. 3).

Fig. 2
figure 2

Neutralizing and enhancing activities of 4G2 against DENV-1, -2, -3, and -4. Inhibition and enhancement of viral replication of DENV-1 (a), DENV-2 (b), DENV-3 (c), and DENV-4 (d) by mAb 4G2 in LLC-MK2 and K562 cells, respectively, are shown. The solid lines indicate the ADE trend of 4G2 and dashed lines represent neutralization curve fits generated by nonlinear regression analysis. The representative results of three independent experiments performed in duplicate are shown

Full size image
Fig. 3
figure 3

Neutralizing and enhancement activities of 2A10G6 against DENV-1, -2, -3, and -4. Inhibition and enhancement of viral replication of DENV-1 (a), DENV-2 (b), DENV-3 (c), and DENV-4 (d) by mAb 2A10G6 in LLC-MK2 and K562 cells, respectively, are shown. The solid lines depict the ADE curve of 2A10G6 and dashed lines represent neutralization curve fits generated by nonlinear regression analysis. Representative results of three independent experiments performed in duplicate are shown

Full size image

Moreover, for both mAbs, typical infection enhancement occurred at sub-neutralizing concentrations. The 4G2 concentration at maximum enhancement was approximately 2.0 μg/mL for DENV-1, 5.0 μg/mL for DENV-2, 3.0 μg/mL for DENV-3, and 1.5 μg/mL for DENV-4, whereas approximate IC50 values for each serotype were 2.7, 1.4, 1.4, and 1.4 μg/mL for DENV-1, -2, -3, and -4, respectively (Fig. 2). In contrast, 2A10G6 demonstrated higher neutralizing activities against DENV-1, -2, -3, and -4, with IC50 values of 0.4, 0.6, 0.4, and 0.2 μg/mL, respectively (Fig. 3). Accordingly, ADE of mAb 2A10G6 was readily detected in a concentration-dependent pattern, with peak enhancements at concentrations of 0.6, 0.2, 0.3, and 0.2 μg/mL, and shifted to lower antibody concentrations compared with mAb 4G2.

Relationship between enhancing and neutralizing activities of serum samples from DENV-1-infected patients

To further examine the utility of the ELISA-ADE assay in analyzing infection enhancement of human immune sera, we selected serum samples from four convalescent patients with primary DENV-1 infections and analysed them first via ELISPOT-MNT to determine their neutralizing antibody titers. As expected, each serum sample exhibited higher NT50 values against DENV-1 in comparison with other serotypes (Figs. 4a, c, e, g and 5a, b), demonstrating strong homotypic neutralizing antibody activity against DENV-1 and weak heterotypic neutralizing activities against DENV-2 to -4. The average NT50 values of the four serum samples against DENV-1, -2, -3, and -4 were 5546, 382, 1084, and 251, respectively (Fig. 5b). Subsequently, we examined the serum samples for ADE activity via ELISA-ADE and found that all demonstrated dose (serum dilution)-dependent response patterns for enhancing activities (Fig. 4b, d, f, h). Peak enhancement against DENV-1 was observed at serum dilutions of 1:1280, 1:640, 1:80, and 1:320 for serum samples #1, #2, #3, and #4, respectively (Fig. 5a). For each serum sample, maximum enhancement occurred at higher serum dilution with DENV-1 in contrast to the other three serotypes (Fig. 5a, c). The average serum dilution of the four serum samples at peak enhancement against DENV-1, -2, -3, and -4 was 1:580, 1:20, 1:70, and 1:130, respectively (Fig. 5c). The control serum did not enhance infection (data not shown).

Fig. 4
figure 4

Neutralizing and enhancing activities of human serum samples against DENV-1, -2, -3, and -4. Neutralizing activities of serum samples from the four primary DENV-infected patients against DENV-1, -2, -3, and -4 are shown in a, c, e, and g, respectively. Correspondingly, serum ADE activities against DENV-1 to DENV-4 are presented in b, d, f, and h. For each serum sample, neutralization curve fits generated by nonlinear regression analysis are presented on the left and ADE curves on the right

Full size image
Fig. 5
figure 5

Neutralizing and enhancing parameters of serum samples from primary DENV-1-infected patients. a Neutralizing and enhancing parameters of convalescent-phase sera from four primary DENV-1 infected patients against DENV-1, -2, -3, and -4 are shown. b, c Scatter plots of serum NT50 values against the four serotypes and serum dilutions (reciprocal) at corresponding peak enhancements are shown, respectively

Full size image

Discussion

In this present study, we report a high-throughput quantitative alternative assay for enhanced DENV infection, which was performed in 96-well format appropriate for automatic readout. This ELISA-ADE enhancement assay demonstrated a wide application to mAbs with different isotypes directed against envelope protein of DENV, including IgG1 (2A10G6), IgG2a (4G2), and IgG2b (unpublished data). mAbs of different isotypes can bind FcγRII, but with different affinities (Rodrigo et al. 2009b), which may explain the wide isotype applicability of this ELISA-ADE assay. Of note, human K562 cells that we used in this study express only one type of FcγR (FcγRIIa; Littaua et al. 1990), while human monocytes/macrophages and related U937 cells express at least FcγRI and FcγRII (Boonnak et al. 2011; Looney et al. 1986). Considering that the neutralizing and enhancing activities determined with different IgG subclass can be modulated by interacting with different FcγR subtype (Rodrigo et al. 2009b; Rodrigo et al. 2006), difference of neutralizing/enhancing antibody activities might exist between cell lines expressing different FcγRs. However, K562 cells have been widely accepted and frequently used human indicator cells and in vitro infection model in ADE determination and assessment (Balsitis et al. 2010; Beltramello et al. 2010; Boonnak et al. 2008; Goncalvez et al. 2007; Guy et al. 2004; Williams et al. 2012; Yamanaka et al. 2012). Furthermore, FcγRIIa was shown strikingly more efficient than FcγRI in enhancing DENV immune complex infectivity (Rodrigo et al. 2006) and K562 cell line was preferred sometimes for its simple and well-characterized system with only FcγRII (Schieffelin et al. 2010). Also combined with the observation that relatively higher sensitivity of NS1 production in K562 cells was observed in comparison with that in U937 cells, we employed K562 cells in developing the current ELISA-ADE assay.

The application of this simple high-throughput assay to the two flavivirus cross-reactive mAbs (4G2 and 2A10G6) further validated its accuracies and feasibility. 2A10G6, with epitope residues localized to a highly conserved flavivirus fusion loop peptide (98DRXW101 motif; Deng et al. 2011) exhibited higher neutralizing activities against DENV-1 to -4 compared with 4G2. As expected, the peak enhancement of 2A10G6 occurred at a lower concentration than 4G2 and the ADE dose-dependent curve for 2A10G6 shifted to a lower concentration accordingly. Thus, our assay not only showed dose-dependent ADE profiles for the two flavivirus cross-reactive mAbs and the occurrence of infection enhancement at sub-neutralizing concentrations, but also demonstrated that the higher the neutralizing activity, the lower the mAb concentrations at peak enhancement.

Further application of the simple high-throughput assay to serum samples from primary infection patients extended its utility to the measurement of enhancing activities of DENV-specific antibodies in polyclonal sera. In vitro neutralization against the homotypic infecting DENV serotype occurred at much higher serum dilutions than that against heterotypic serotypes, with results similar to previous studies (Midgley et al. 2011). Consistently, they exhibited characteristic ADE dose–response patterns at higher serum dilutions against primary infecting DENV-1 strains than against heterotypic serotypes (DENV-2, -3, and -4). In the presence of higher titers of homologous neutralizing antibodies, neutralization eclipses infection enhancement and defers the occurrence of ADE to lower concentrations of human antibodies, while lower neutralizing capacity of cross-reactive antibodies against heterotypic infection might not antagonize enhancing antibodies at initial serum dilutions. Thus, the patient sera in our study, when undiluted as in vivo, likely exhibit neutralizing antibody capacity against subsequent homotypic (DENV-1) infections, while against heterotypic (DENV-2, -3, and -4) infections, the ultimate combined activities might depend on the balance or compromising effect of enhancing/neutralizing antibodies, especially for patient serum #1 and #2. By increasing the dilution factor, which is parallel to the waning of serum antibody levels in vivo, neutralization reduction will be accompanied by an increase in infection enhancement against the homotypic serotype. As such, our novel ADE assay can provide considerable informative data conveniently and rapidly using vaccinated and DENV-infected human serum samples to assist evaluation of the risk of developing severe dengue diseases.

Quantitative measurement of DENV replication levels in infected K562 cells using ELISA format allows for simple, rapid, and high-throughput detection of enhancing activities, avoiding cumbersome and time-consuming procedures. The high-throughput assay can measure functional activities of DENV-specific mAbs of interest and as well as human sera, exhibiting promising applicability and thus helping determine enhancing mAb properties mapped to different structural epitopes and DENV-specific antibody responses in human immune sera. Moreover, ADE, as an injurious phenomenon, constitutes one of the most critical aspects regarding evaluations of DENV vaccine candidates. In the pursuit of safe dengue vaccine candidates, it would be informative to measure neutralizing activities of vaccinated sera using ELISA-ADE in a large-scale study to facilitate functional evaluation of dengue vaccines.

It should be pointed out that some caution might be exercised when interpreting the relationship between functional activities and protective versus exacerbating status of DENV-infected or vaccinated human serum samples using data from ELISA-ADE in combination with ELISPOT-MNT. This is because the enhancing and neutralizing assays using ELISA-ADE and ELISPOT-MNT are performed in K562 and FcγR-negative LLC-MK2 cell line, respectively. Considering the limited predictive values of plaque reduction neutralization test (PRNT) (Endy et al. 2004; Wu et al. 2012), and that neutralizing activity determined by FcγR-expressing cells might better reflect protection to DENV infection in vivo (Moi et al. 2012; Rodrigo et al. 2009a), conducting both enhancing and neutralizing assays in the same FcγR-expressing cell line to reflect the combined enhancing/neutralizing activities deserves exploration in further studies.