Multiplexed FluoroSpot for the Analysis of Dengue Virus– and Zika Virus–Specific and Cross-Reactive Memory B Cells

Dengue virus (DENV) and Zika virus (ZIKV) are members of the genus Flavivirus and belong to the family Flaviviridae, which also includes other medically important viruses, such as yellow fever virus, West Nile virus, Japanese encephalitis virus, and tick-borne encephalitis virus (1). DENV and ZIKV are transmitted to humans by the bite of mosquitos, predominantly Aedes aegypti (2, 3). Due in part to the global spread of the vector, the four types of DENV circulate in much of the world and cause episodic outbreaks. The recent outbreak of ZIKV in Brazil in 2015 was associated with neurologic symptoms, microcephaly in infants born to pregnant women, and novel patterns of transmission via sexual intercourse (4).

B cells are massively expanded during primary and secondary DENV infections, and several studies have evaluated the specificity of the Abs secreted by plasmablasts (5–9). Long-lived plasma cells and memory B cells (MBCs) reside in the bone marrow and in the peripheral blood, respectively, and contribute to Abs detected in the sera years postinfection (10, 11). One challenge to studying these cell populations is the lack of commercially available reagents to identify Ag-specific B cells. We recently used amine labeling of DENV to identify DENV-specific B cells in the circulation (12, 13). We identified multiple subsets of Ag-specific B cells in the circulation during acute primary and secondary DENV infection, including naive B cells, MBCs, and plasmablasts with a much lower frequency of DENV-specific MBCs 6 mo later (13). Exposure to ZIKV or administration of a ZIKV vaccine will likely occur in children and adults on a background of pre-existing immunity to DENV (14, 15). Therefore, understanding the magnitude and quality of cross-reactive responses at the B cell and Ab levels between ZIKV and DENV is important (16).

The envelope (E), premembrane (prM), and the nonstructural protein 1 (NS1) proteins are the main targets for the humoral immune response after DENV and ZIKV infection (17). However, there are several challenges in measuring Ab responses (14, 18). The four DENV serotypes differ from each other by 30–35%, and DENV differs from ZIKV by 41–46% in amino acid sequence of the E protein (19). Cross-reactive Abs specific for E with poor, moderate, or potent neutralizing activity and against prM and NS1 have been characterized in individuals following natural infection (20–24). A number of human mAbs from DENV immune donors only bind quarternary epitopes presented on mature viruses, and these Abs cannot be detected using recombinant protein (25). Multiple studies have not been able to reliably distinguish DENV from ZIKV infection serologically given the high degree of antigenic similarity between these two viruses (26, 27), but newer assays using recombinant virus particles effectively measure neutralizing Ab responses to ZIKV (28). Cross-reactive neutralizing Abs between DENV and ZIKV are not sustained long term in DENV and ZIKV immune individuals (29, 30). Several in vitro studies have recently reported Ab-dependent enhancement activity of Abs using sera from DENV- or ZIKV-immune donors (23, 31), whereas two macaque studies showed no significant impact of DENV Abs on viral load or severity of subsequent ZIKV-induced disease (32, 33).

The B cell ELISPOT assay to enumerate the number of Ab-secreting B cells (ASCs) was first described in 1983 (34). The ELISPOT assay has been modified to detect Ag-specific plasmablasts and MBCs expanded in vitro into ASCs (6, 35–37). Although the B cell ELISPOT assay is a powerful and highly sensitive method to analyze Abs from ASCs, it is limited to detection of a single Ag in a well. To visualize cross-reactive ASCs, a Quad Color FluoroSpot assay was recently developed in which standardized concentrations of DENV Ags were added to Abs secreted by MBCs captured on plates by Ig-specific Abs (38). Spots representing individual B cells were identified using DENV serotype-specific fluorescently labeled Abs.

In this article, we present a novel approach to detecting cross-reactive ASCs, wherein we directly labeled viruses representing all four serotypes of DENV as well as ZIKV. We hypothesized that fluorescently labeled DENV (FL-DENV) and ZIKV would be useful to measure DENV-specific, ZIKV-specific, DENV serotype cross-reactive, and DENV–ZIKV cross-reactive MBC responses. We used well-characterized PBMC from DENV-immune, ZIKV-immune, and naive donors to simultaneously detect ASCs producing Abs that recognize one, two, three, or four serotypes of DENV and/or ZIKV.

A panel of 19 DENV-immune, ZIKV-immune, and naive subjects was studied (Table I). All DENV and ZIKV PBMC were collected after written informed consent. The studies were approved by the institutional review boards at the University of Rhode Island, National Institute of Allergy and Infectious Diseases, and the Human Subjects Research Review Board for the Commanding General of the U.S. Army Medical Research and Material Command. Blood samples of travelers were collected at the National Institutes of Health Clinical Center from patients displaying symptoms of a suspected ZIKV infection following return to the United States from areas where ZIKV was known to be circulating. PBMC were purified and cryopreserved until used.

DENV-1 strain WP-74, DENV-2 strain S16803, DENV-3 strain CH53489, and DENV-4 strain TVP-360 were propagated in Vero cells, concentrated by tangential flow ultrafiltration, and purified by sucrose gradient ultracentrifugation as described previously (39). The concentrations of purified DENV were determined by A260/280 measurement on a NanoDrop. Protein concentrations of DENV-1, DENV-2, DENV-3, and DENV-4 were 0.44, 0.49, 0.89, and 0.84 mg/ml, respectively. The ZIKV Paraiba_01 strain was propagated in C6/36 cells. The infected cell culture supernatant was harvested on day 5 postinfection. The supernatant was layered on top of a 30% sucrose solution containing 10 mM Tris, 100 mM NaCl, and 1 mM EDTA. The virus was pelleted by ultracentrifugation in a swinging-bucket rotor at 26,000 rpm for 4 h at 4°C to remove low-molecular–weight contaminants such as soluble proteins. The virus pellet was resuspended in PBS. The purity of the viral preparations was verified by SDS-PAGE. The protein concentration of ZIKV was 0.8 mg/ml.

We labeled purified DENV and ZIKV using small scale 3 × 100–200 μg Lightning-Link Rapid Ab protein labeling kits DyLight (DL) 405, DL488, DL594, or DL650 (Innova Biosciences, Cambridge, U.K.). Briefly, 10 μl of Lightning-Link Rapid modifier was added to 100 μl of virus preparation with gentle pipetting, and the mixture was added to the pellet of lyophilized dye. After gentle aspiration, the reaction was allowed to proceed 15–30 min at room temperature protected from light. Lightning-Link Rapid quencher (10 μl) was added to terminate the reaction. After 15 min, RPMI 1640 (no phenol red) supplemented with 5% FBS was added in a volume to normalize initial protein concentrations of the viral preparations. Aliquots of 50 μl labeled virus were stored at −80°C until further use.

Labeling efficiency as well as retention of binding capacity of FL-DENV and ZIKV was verified on all preparations. Goat anti-mouse IgG (Fc) beads (100 μl aliquots of 3.0 μm beads; MPFc-30-5; Spherotech, Lake Forest, IL) were incubated with the optimal concentrations of mouse anti-Dengue 2H2 Ab (Thomas Scientific, Swedesboro, NJ) or 3H5 Ab (MilliporeSigma, Burlington, MA), anti-flavi 4G2 Abs (Novus Biologicals, Littleton, CO), or mouse anti-human CD28 Abs (BD Biosciences, San Jose, CA) and then washed and resuspended in 1 ml of PBS containing 10% FBS. Test aliquots of FL-DENV or ZIKV were incubated with 15 μl aliquots of functionalized beads for 30 min on a rotating platform at room temperature, washed two times, and read on the flow cytometer. A pattern of reactivity was revealed to show both specificity of binding as well as relative fluorescence intensity of the labeled virion preparations. For binding assays, FL-DENV–1, –2, –3, or –4 were added to 1 × 10⁵ U937 DC-SIGN cells. Following incubation at 37°C for ∼45 min, cells were washed and fixed in 100 μl of 1:4 BD Cytofix Fixation Buffer (diluted in PBS) by BD Biosciences. Binding of FL-DENV was assessed on a BD LSR flow cytometer.

Cryopreserved PBMC were thawed and washed twice with RPMI 1640 media with 10% FBS (Sigma-Aldrich, St. Louis, MO) and 100c U/ml penicillin/streptomycin (HyClone TM). Cells were counted and diluted to 2 × 10⁶/ml in RPMI/10% FBS. PBMC were stimulated with 2.5 μg/ml R848 (InvivoGen, San Diego, CA) and 1000 U/ml recombinant human IL-2 (PeproTech, Rocky Hill, NJ) for 7 d at 37°C and 5% CO₂. Supernatants from MBC cultures were collected and assessed for Ab secretion by ELISA. Cells were collected and tested in ELISPOT and FluoroSpot assays.

PBMC culture supernatants were tested for DENV and ZIKV IgG Abs in ELISA assays. MaxiSorp ELISA plates (Thermo Fisher Scientific) were coated with 10 ng/well DENV or ZIKV virus-like particles (VLPs) (The Native Antigen Company, Oxford, U.K.) and incubated overnight at 4°C. The plates were blocked with 1% BSA for 90 min, and 50 μl supernatant was added per well for 1 h at 37°C. Plates were washed with PBS, containing 0.05% Tween-20. Goat anti-human IgG-Fc fragment coupled to HRP (A80-104P; Bethyl Laboratories, Montgomery, TX) was added as the secondary Ab at a 1:5000 dilution. Then, 100 μl 3, 3, 5, 5′- tetramethylbenzidine peroxidase substrate (Thermo Fisher Scientific, Rockford, IL) was added. After ∼15 min, the enzyme reaction was stopped by the addition of 50 μl sulfuric acid, and the plates were read at 450 nm. The supernatants from each condition were tested in duplicate. The average of duplicates was calculated after background values were subtracted.

ELISPOT assays were performed as previously described (6). Briefly, wells of Millipore ELISPOT plates (MAIPSWU10; MilliporeSigma, Billerica, MA) were treated with 20 μl per well 35% ethanol and then coated with 100 μl DENV or DENV VLP (7.5 μg/ml). To detect total IgG, the wells were coated with 80 μl anti-human Ig capture Ab (Cellular Technology, Shaker Heights, OH). The plates were stored at 4°C overnight. PBMC cultures were collected and counted, and 2 × 10⁵ or 5 × 10⁴ cells in 100 μl was added to each well in duplicate to assess DENV-specific or total IgG ASCs, respectively. The plates were incubated overnight at 37°C and 5% CO₂. After two washes with PBS and 0.05% Tween PBS, 80 μl biotinylated mAb directed against human IgG (Cellular Technology) was added to each well. The plates were incubated for 2 h at room temperature, washed three times with 0.05% Tween PBS, and developed with alkaline phosphatase–conjugated streptavidin for 1 h. The plates were washed two times with PBS-Tween, followed by two times with deionized water and then 80 μl of the substrate solution was added for 15 min before rinsing three times with tap water. The background reactivity (spot counts in negative control wells) observed within the ELISPOT assay was one to two spots per well.

For FluoroSpot assays, low fluorescent PVDF membrane plates (IPFL) (PUV96; Cellular Technology) were treated with 20 μl per well 70% ethanol for 1 min. Ig capture Abs (Human IgG Single-Color FluoroSpot Assay, Cellular Technology) were added to wells, and plates were incubated overnight at 4°C. Following washing and blocking of the plates, MBC cultures (2 × 10⁵) were added to duplicate wells for each condition and then incubated overnight at 37°C and 5% CO₂. The plates were washed and blocked with 1% BSA for 1 h at 37°C. Optimized concentrations of FL-DENV and ZIKV diluted in 100 μl of PBS was then added. The plates were incubated for 1 h at room temperature. Plates were washed three times with PBS, followed by two times with deionized water and rinsed with tap water. The plates were allowed to dry completely. To determine total IgG-secreting cells, 5 × 10⁴ MBC cultures were added to wells coated with human Ig and visualized using a polyclonal PE-labeled IgG detection Ab (Cellular Technology).

The plates were scanned using an S5 (for dual color) or S6 (for quad color) Cellular Technology ImmunoSpot analyzer. Spots representing Abs secreted by Ag-specific MBCs were detected using different filters (440/40 for DL405 DENV-2, 520/40 for DL488 DENV-4 or ZIKV, 630/60 for DL594 DENV-3, and 690/50 for DL650 DENV-1). The settings for the sensitivity of spot counting were established and adjusted manually for each plate using Ag-stimulated and negative controls and to exclude artifacts or background spots. The number of specific or cross-reactive spots were evaluated by the reader and verified manually. The numbers of Ag-specific ASCs were adjusted to ASC/1 10⁶ input cells for statistical analysis. FCS data files of wells were collected on a BD LSR II with BD FACSDiva 8.0.1 software to display and analyze specific and cross-reactive spots. Data analysis and transformation was performed using FlowJo v10 (Tree Star, Ashland, OR).

Statistical analysis was performed using GraphPad Prism 6 (GraphPad Software, La Jolla, CA). FluoroSpot results were analyzed with Mann–Whitney U test for paired comparisons. ELISA binding results were analyzed using one-way ANOVA and t test. A p value ≤0.05 was considered significant.

The sucrose gradient–purified DENV for each of the four serotypes were titrated in a virus capture ELISA, and virus particles bound to capture Ab were detected using the flavivirus cross-reactive Ab 4G2 (Fig. 1A). The end-point titers for DENV-1, -2, -3, and -4 were 3,200, 12,800, 12,800, and 51,200, respectively. We conjugated DENV-1, -2, -3, and -4 to amine-reactive, water-soluble DL dyes with different excitation wavelengths (DL650, DL405, DL594, and DL488, respectively). The specific dyes with absorption spectra ranging from 400–670 nm were selected to match the principal output wavelengths for flow cytometry and FluoroSpot instrumentation. Fluorescently labeled viruses (FLVs) were initially evaluated for binding to beads coated with flavivirus-specific Abs 2H2, 4G2, and 3H5 or control anti-CD28 Abs. In all cases, there was a clear separation in binding of FLVs between control and flavivirus cross-reactive 2H2 and 4G2 Ab-coated beads, although the intensity of fluorescence varied based on the dye used (Fig. 1B). As expected, there was no shift in fluorescence when labeled DENV-1, -3, or -4; preparations were added to DENV-2–specific Ab (3H5)-coated beads. To determine whether the labeling process inhibited the ability of DENV to bind susceptible cells, we evaluated the binding of FLVs to U937 DC-SIGN cells. After incubation of DL650 DENV-1, DL405 DENV-2, DL594 DENV-3, or DL488 DENV-4 with U937 DC-SIGN cells at 4°C for 45 min, we found significant binding of all FLVs to U937 DC-SIGN cells (Fig. 1C). Finally, to determine whether there was competition for binding of FLVs to Ab-coated beads, DL594 DENV-3 and DL488 DENV-4 were added alone or together to 4G2-coated beads and fluorescence evaluated. We found a shift in fluorescence in the PE and FITC channels, respectively, when 4G2 beads bound with DENV-3 or -4 were added alone (Fig. 1D). When DENV-3 and -4 were added together, staining was detected along a diagonal and not skewed toward a particular fluorescence channel, indicating that both FLVs were able to bind 4G2-coated beads equally. Taken together, our data indicate productive labeling of FLVs with no apparent effect on the structure or binding capabilities of the viruses.

A schematic comparing the conventional ELISPOT and FluoroSpot assays using specific Abs or FLVs is shown in Fig. 2. Because circulating MBCs do not actively secrete Abs, PBMC from DENV-immune and naive subjects were stimulated with the TLR7/8 agonist, r848, and rIL-2 in vitro for 7 d to convert MBCs into ASCs (40). Stimulated cells were plated onto low fluorescent PVDF membrane plates (IPFL) coated with human Ig capture Abs and incubated at 37°C overnight. To determine whether FLVs could be used to identify Abs secreted by DENV-specific MBC cultures, we added optimal concentrations of DL594 DENV-3 (red) or DL488 DENV-4 (green) or unlabeled DENV into separate wells. The assay was optimized using ethanol pretreatment, blocking and washing steps, and varying concentrations of individual FLVs to give a distinct signal when ASCs from immune donors were used and minimal staining when ASCs from naive donors were used (Supplemental Fig. 1). Representative images of negative control wells containing MBC cultures from a naive subject with labeled DENV, an immune subject with unlabeled DENV, and unstimulated PBMC from an immune donor with labeled DENV are shown. Red or green spots were detected in wells containing MBC cultures from DENV-immune donors with DL594 DENV-3 or DL488 DENV-4, respectively. No spots were detected when the wavelength-mismatched filter was used, indicating lack of signal spillover between detection channels (Supplemental Fig. 1).

We evaluated the specificity of the assay by analyzing PBMC from six DENV-immune and seven DENV-naive subjects (Fig. 3). Representative images of wells containing MBC cultures from a naive and immune donor with FL-DENV–1, –2, –3, –4, and total IgG are shown in Fig. 3A. The frequency of DENV-1, -2, -3, or -4 binding ASCs was higher in DENV-immune donors compared with naive donors, which was statistically significant (Fig. 3B, 3D, 3F, 3H). We found 3–12 spots per well when using MBC cultures from naive donors in the FluoroSpot. Cell culture supernatants collected 7 d after stimulation from the same immune PBMC also had detectable DENV-specific Abs (Fig. 3C, 3E, 3G, 3I). We found no Abs that recognized DENV-1, -2, -3, or -4 in supernatants from unstimulated PBMC or MBC cultures from naive donors.

Optimized concentrations of DL594 DENV-3 and DL488 DENV-4 were added together in the same well to distinguish DENV-specific and cross-reactive ASCs. Images were initially analyzed using a single filter and then a computerized overlay image was generated to identify overlapping spots. In the overlay image, the red spots represent ASCs that bound DL594 DENV-3, green spots represent ASCs that bound DL488 DENV-4, and yellow spots represent cross-reactive ASCs that bound DENV-3 and -4, all of which were seen in the ASC cultures from the immune but not naive donor (Fig. 4A). We evaluated ASCs using MBC cultures from an immune donor who received a monovalent DENV-4 vaccine (Fig. 4B). Because the donor was exposed only to DENV-4, we expected to see DENV-4–specific spots (green) and DENV-3/4 cross-reactive spots (yellow) and no DENV-3–specific spots (red). In the overlay image, both green and yellow spots were detected with minimal red spots. Furthermore, we found a wide range in the size and color intensity of the green and yellow spots (Fig. 4B, expanded image), which we speculate indicate differences in relative avidities for the different viruses. We compared the results from the FluoroSpot to a conventional ELISPOT assay and found similar frequencies of DENV-3– and DENV-4–specific ASCs (Fig. 4C). However, because the conventional ELISPOT assay cannot differentiate between specific and cross-reactive Abs secreted by individual ASCs, we could not compare the frequency of cross-reactive cells between the two assay formats. Supernatants collected from immune PBMC cultures also contained Abs that bound DENV-3 and DENV-4 (Fig. 4D).

To demonstrate that the FluoroSpot assay could detect cross-reactive ASCs, we used well-defined PBMC from subjects who had natural DENV infections or received candidate DENV vaccines (Table I). ASCs that bound a single (red or green spots) or two DENV serotypes (yellow spots) were detected in PBMC from all subjects (Fig. 5A); however, the frequencies of specific and cross-reactive B cells varied among the donors (Fig. 5B). The ratios of DENV-specific and cross-reactive ASCs to total IgG-secreting cells in immune subjects ranged from 0.2 to 5% (Table II). FCS files were generated using the ImmunoSpot software, which enabled bivariate analysis of spots using FlowJo software and showed clear separation of red, green, and yellow spots (Fig. 5C). In depth analysis in FlowJo, identified cells that recognized DENV-3 and -4 equally, cells that recognize DENV-3 better than DENV-4, and cells that recognized DENV-4 better than DENV-3 (Fig. 5D). The FluoroSpot results analyzed using FlowJo reflected the patterns we detected in the FluoroSpot overlay images, which identified red, green, and yellow spots of different intensities and sizes, including red spots with varying sizes of yellow centers and green spots with varying sizes of yellow centers. Taken together, the dual-color FluoroSpot retained the sensitivity of the single-color assay and, in addition, was able to identify ASCs that bound two different DENV serotypes.

We next used four different FL-DENV together to assess the cross-reactivity of B cells in MBC cultures. Shown in Fig. 6 are images of one well that contained MBC cultures from a DENV-immune donor who reported receiving a tetravalent DENV vaccine. Using a single filter (Fig. 6A), dual filters (Fig. 6B), and four filters (Fig. 6C), we found serotype-specific MBCs (gold, blue, red, or green spots representing DENV-1, -2, -3, or -4–specific MBCs) and MBCs that recognized two DENV serotypes (purple and pink), three serotypes, as well as all four serotypes (white spots). The size and intensity of the DENV-specific and cross-reactive spots varied in MBC cultures from this donor.

To detect DENV/ZIKV cross-reactive MBCs, we analyzed PBMC from DENV- and ZIKV-immune subjects using a pan-DENV/ZIKV dual-color FluoroSpot. We labeled a pan-DENV preparation using equal amounts (measured by protein concentration) of the four DENV serotypes with DL594 (red) and a preparation of purified ZIKV with DL488 (green). PBMC from ZIKV-immune donors were obtained approximately 2 mo after natural ZIKV infection, and PBMC from DENV-immune donors were obtained months to years postinfection or after vaccination (Table I). Using DL594 pan-DENV and DL488 ZIKV, we found many more green spots compared with red or yellow spots in the wells from four of four ZIKV-immune donors, indicating low cross-reactivity between DENV and ZIKV ASCs in MBC cultures (Fig. 7A). Although the frequency of cross-reactive MBCs was low, they were higher in PBMC from donors 9 and 10 compared with PBMC from donors 11 and 12 (Fig. 7B, Table III). Interestingly, donors 9 and 10 reported receiving a yellow fever vaccination. We next evaluated MBC cultures from DENV-immune donors. MBC cultures from one of four DENV-immune donors secreted Abs that bound DENV and ZIKV strongly, which resulted in many more yellow spots in the well (Fig. 7C left-most image). This donor received a monovalent DENV-4 vaccine and had traveled to an area where yellow fever was endemic (41). Supernatants from MBC cultures contained Abs that bound DENV and ZIKV VLPs (Fig. 7E, 7F). PBMC from DENV-immune donors had a higher frequency of DENV-specific ASCs compared with DENV/ZIKV cross-reactive ASCs (Fig. 7D; Table II).

We established a novel FluoroSpot assay using DENV and ZIKV labeled with different DL dyes, which allowed us to distinguish DENV-specific, DENV serotype cross-reactive, ZIKV-specific, and DENV/ZIKV cross-reactive ASCs at the single-cell level. Using PBMC from DENV-immune, ZIKV-immune, and naive donors, our data indicate that FLVs can be used to study the specificity of Abs secreted by MBCs generated after natural DENV or ZIKV infection or after vaccination. In each experiment performed, we used FLVs to evaluate PBMC from flavivirus-naive donors as specificity controls. Although no DENV-binding Abs were detected in these individuals by ELISA, a very low number of spots were seen in the FluoroSpot assay. Given that members of flavivirus family circulate globally, it is difficult to exclude the possibility that these background spots represent true flavivirus cross-reactive immunity to other flaviviruses. Our findings clearly demonstrate that the FluoroSpot is better adapted than the conventional ELISPOT to distinguish cross-reactive B cell responses.

In this study, we were able to efficiently label DENV-1–4 and ZIKV with DL dyes using commercially available protein-labeling kits, similar to our previous study using Alexa Fluor dyes (13). We chose to label DENV and ZIKV with Lightning-Link DL dyes, because they were easy to use and did not involve spin or separation steps. The excitation and emission spectra of the DL dyes used were also compatible with common laser and filter sets. We have not extensively compared Alexa Fluor–labeled DENV and DL-labeled DENV in multiparametric flow cytometry assays used to identify MBCs in the circulation. We have not performed structural or detailed infectivity studies on DL-labeled DENV and ZIKV. It is possible that some aberrant epitopes are presented because of the modifications made to the viruses during the labeling process. Furthermore, we cannot rule out that the labeling process blocked some epitopes on the surface of the virion. All labeled FLVs, however, bound beads that were coated with multiple flavivirus-specific (2H2, 4G2, and 3H5) but not control (CD28) Abs and the susceptible cell line U937 DC-SIGN. Furthermore, Abs in supernatants from stimulated DENV-immune PBMC bound unlabeled and FL-DENV equally in ELISA assays (data not shown).

We chose to initially focus on labeling DENV-3 and -4 to develop the dual-color FluoroSpot for multiple reasons. Amino acid sequences in the E protein indicate significant homology between DENV-1 and -3 and DENV-2 and -4 (42); thus, one might expect to find more cross-reactive MBCs using these serotypes together, as the E protein is a dominant target of Ab recognition. Fluorescence was detected on a diagonal when FL-DENV–3 and –4 were added together, indicating equivalent binding by both DENV-3 and -4 to 4G2-coated beads. The amount of prM and E proteins present in each of the four DENV preparations varied but DENV-3 and -4 had similar amounts of prM and E proteins by Western blot analysis compared with DENV-1 and -2 (data not shown). Because FLVs were diluted in a volume calculated to normalize initial protein concentrations of the viral preparations, it is possible that in the multiplexed FluoroSpot assay in which four FL-DENV were used together, we underestimated the extent of cross-reactivity between serotypes that differed in prM and E protein content. Although most of the spots should represent prM or E-specific MBCs, a small number could be NS1-specific as our purified viruses contain trace amounts of secreted NS1.

A quad-color FluoroSpot assay was recently developed using four FL-DENV serotype-specific mAbs to detect DENV serotype-specific and cross-reactive B cells (38). Our approach involves direct labeling of the viruses themselves and has several advantages. Our FluoroSpot protocol is simple to use. PBMC stimulated 6–7 d prior were added to Ig-coated wells and ASCs detected using FLVs in our assay while purified DENV was added to ASCs followed by detection with Abs that recognized each of the four serotypes in the recently published study. Using FLVs is not dependent on having a specific mAb for detection, so our assay system is more flexible in terms of which viruses/Ags can be used and multiplexed together. For example, we compared responses to DENV and ZIKV using all four DENV serotypes labeled with the same dye. Alternatively, it would be feasible to compare responses to different strains within a serotype, such as representative of different genotype or wild-type versus vaccine strains. Using an Ag-specific mAb in the detection step may compete with Abs from ASCs that potentially recognize the same epitope, which could affect sensitivity of the signal. Labeling of rNS1 or recombinant proteins in the future would be easier compared with validating five different NS1 or E-specific Abs for detection of NS1 and recombinant-specific B cells.

We detected a wide range of spot size and color intensity. We speculate that the size and intensity in each channel relate to the rate of Ab secretion and Ab avidity for each virus. The large, bright yellow spots would be expected to reflect DENV-3/4 or DENV/ZIKV cross-reactive ASCs with high avidity, whereas small yellow spots represent low-avidity ASCs. Analysis of the data using FCS files and FlowJo software supported the presence of Abs with varying avidities, although information regarding which spot on the FluoroSpot well corresponds to which dot on the flow plot is not preserved when exporting the FCS files.

Pre-existing immunity against DENV has been proposed to influence the response to subsequent ZIKV infection, potentially leading to protection or to enhanced infection and possibly disease (14). It is likely that a ZIKV vaccine will be available in the next few years, and therefore it is crucial that we understand whether prior experience with DENV is expected to enhance ZIKV vaccine-induced responses or potentially hinder it and whether pre-existing ZIKV immunity influences subsequent DENV infection. Our data using PBMC from DENV- and ZIKV-immune donors indicate minimal cross-reactivity between DENV and ZIKV MBCs using PBMC from seven of eight donors tested. The timing when samples are obtained after vaccination or natural infection is likely to influence the extent of MBC cross-reactivity between these two more distantly related flaviviruses.

The data from this study showed that FLVs are sensitive and specific tools for the analysis of flavivirus-specific and cross-reactive ASCs. With further development of the FluoroSpot and the incorporation of new filters and fluorophores, it is likely that inclusion of more Ags or analysis of other functions of B cells, such as cytokine production, and determination of the involvement of the Ig subclasses in DENV-specific and cross-reactive responses is achievable. The FluoroSpot assay should enable assessment of the breadth as well as the durability of DENV/ZIKV B cell responses generated in the context of vaccination or natural infection. Moreover, this method may be very useful in vaccine efficacy trials in which the number of cells available are limited and to investigate the contribution of cross-reactive immunity to vaccine-induced protection. In addition, the assay can be adapted to assess ASCs specific for other diseases, such as influenza, in which prior immunity may influence subsequent responses to infection with different virus strains.

We thank the donors who generously provided PBMC for use in our studies. We thank Villian Naeem from Cellular Technology Ltd. for his technical expertise and helpful discussion.

The authors have no financial conflicts of interest.