Single-Cell Analysis of B Cell/Antibody Cross-Reactivity Using a Novel Multicolor FluoroSpot Assay

This article is featured in In This Issue, p.2921

Half of the global population is at risk for dengue, a mosquito-borne disease caused by four distinct virus serotypes (DENV1–4) (1). Infection with one DENV serotype confers protection from disease upon reinfection with the same serotype, but the immune response can contribute to either protection or increased disease severity upon infection with other serotypes (2). After a primary DENV infection, although cross-reactive Abs predominate, neutralizing Abs are mostly type specific (3). In secondary DENV infections, type-specific Abs derived from primary memory B cells (MBCs) are maintained, but cross-reactive Abs are induced at higher titers and can be strongly neutralizing (4–7). With respect to B cells, secondary DENV infection results in a dramatic plasmablast (PB) expansion in peripheral blood (8, 9). Serotype-specific B cells are more frequent after primary DENV infection, whereas after secondary infection, serotype cross-reactive B cells predominate, with greater reactivity to previously infecting DENV serotypes (10, 11). Dissecting Ab and B cell serotype specificity and cross-reactivity is essential for understanding protective immunity against DENV resulting from natural infections and candidate vaccines.

Human DENV-specific B cells have been studied by generation of B cell clones and mAbs and by ELISPOT. EBV transformation of B cells isolated from dengue patient PBMCs and fusion with myeloma cells or recombinant expression of mAbs generated from DENV-reactive PB/plasma cells (PCs) has enabled the analysis of the DENV serotype cross-reactivity of single MBCs or PB/PCs (3–5, 7, 12). Although this approach provides information about cross-reactivity on a single-cell basis, generation of mAbs can be stochastic due to low EBV transformation and fusion efficiency (4, 5, 13) and potential bias in the selection process of EBV transformation or to limitations in sorting DENV-specific B cells and in Ab sequencing, cloning, and expression. Thus, the resulting clones and mAbs may not be representative of the entire breadth of the B cell response. Conventional ELISPOT has been used to characterize both MBC and PB/PC responses after DENV infection (8–11). However, because the serotype specificity is dictated by the Ag coating each well, this approach can only determine B cell specificity at the population level. Recent advances in ELISPOT readers and fluorescent probe technology permit improved detection of Ag specificity using fluorescently labeled Abs. We have incorporated these innovations into a new dengue Quad-Color FluoroSpot (QCF) assay to enable single cell–based analysis of B cell serotype cross-reactivity in DENV-infected individuals.

Aedes albopictus C6/36 cells were infected with the four DENV serotypes at a multiplicity of infection of 0.01 in RPMI 1640 supplemented with 2% FBS (Life Technologies), 1% penicillin/streptomycin, and 1% HEPES. Cellular supernatant was collected 5–8 d postinfection and concentrated using 100-kDa Amicon filter units (Millipore) by centrifuging at 3250 × g for 20 min at 4°C. The concentrated supernatant containing the virus was layered on a discontinuous 20/55% OptiPrep (Sigma-Aldrich) density gradient and ultracentrifuged at 40,000 rpm for 2 h at 4°C, without the brake, using a SW41Ti rotor. DENV layered between the 20 and 55% gradients was collected and tested by RT-PCR (11) and SDS-PAGE (see below) for serotype specificity and protein purity, respectively.

Three of the mAbs used for detection are of mouse origin (anti-DENV1 E95, anti-DENV2 E96, and anti-DENV4 E88) and were obtained from M. Diamond (Washington University in St. Louis). One human detection mAb (anti-DENV3 5J7) was obtained from J. Crowe (Vanderbilt University). Murine hybridomas secreting mAbs used for detection in the QCF assay were propagated in RPMI 1640 supplemented with 10% FBS, 1% penicillin/streptomycin, and 1% HEPES, adapted to serum-free medium (Hybridoma-SFM; Life Technologies), and transferred to CELLine dual-chamber bioreactors (Integra Biosciences). Supernatant containing the secreted mAbs was collected every 7 d and purified using agarose beads coated with protein G (UltraLink; Pierce) according to the manufacturer’s instructions. mAbs were resuspended in Tris buffer (pH 7.4) and stored at −20°C. mAb purity and DENV serotype specificity were tested by SDS-PAGE and ELISA, respectively.

Purified DENV or purified mAbs were denatured in 5× SDS loading buffer (0.25 mol Tris-HCl [pH 6.8], 0.25% bromophenol blue, 0.5 mol DTT, 10% SDS, and 50% glycerol) at 96°C for 15 min. For mAbs, 12 μl denatured samples was loaded on a 12% precasted SDS-PAGE gel (Bio-Rad) and run at 100 V for 30–60 min. For DENV, 5 μg DENV/well was loaded on a precast 4–20% gel (Bio-Rad) and run for 2 h at 60 V. Gels were then stained with Coomassie brilliant blue staining solution (Coomassie brilliant blue dissolved in 50% methanol, 10% glacial acetic acid, and 40% H₂O) for 10–15 min. Subsequently, gels were destained for 20–50 min using a destaining solution (50% methanol, 10% glacial acetic acid, and 40% H₂O) until the protein bands were visible.

A qualitative ELISA was used to assess the serotype specificity of the detection mAbs (anti-DENV1 E95, anti-DENV2 E96, anti-DENV3 5J7, and anti-DENV4 E88) prior to quantum dot (Qdot) conjugation (Qdot 565, 705, 625, and 525, respectively). Briefly, plates were coated with a 1:100 dilution of the OptiPrep-purified DENV in coating buffer at 37°C for 2 h. After three washes with PBS-0.05% Tween (PBS-T), plates were blocked with blocking buffer (5% milk in PBS-T) for 2 h at 37°C. Detection mAbs were incubated with the primary Ab at a 1:1000 dilution in blocking buffer at 37°C for 2 h. After three washes with PBS-T, the plates were incubated with either anti-human-HRP or anti-mouse-HRP secondary Ab depending on the origin of the primary Ab (human versus mouse) at 37°C for 2 h. Plates were then washed three times with PBS-T and developed with 3,3′,5,5′-tetramethylbenzidine (Sigma-Aldrich).

Purified mAbs (100 μg) were conjugated enzymatically to Qdots (Live Technologies) via the N-acetylglucosamine of the glycosylation chain on the Fc region of the mAb after hydrolysis of a galactose residue. SiteClick technology (Invitrogen) was used according to the manufacturer’s instructions. Briefly, after the Tris buffer was exchanged with the Ab preparation buffer, galactosidase was used to cleave the galactose residue on the glycosylation chain, leaving the N-acetylglucosamine exposed. Subsequently, β-1,4-galactosyltransferase was used to attach an azide group (N-azidoacetylgalactosamine) to the cleaved galactose moiety. Microcentrifugal filter units (m.w. cutoff 50 kDa) were used to purify the azide-modified mAbs from the enzymes, and the mAbs were incubated with dibenzocyclooctynol-modified Qdots overnight at room temperature. The conjugated mAbs were then ready for use. For Qdots 625 and 700, the manufacturer offers a microcentrifugal filter of 300 kDa to remove unconjugated Qdots and mAbs.

Flow cytometry was used to test the conjugation efficiency of the detection mAbs. For this purpose, Raji cells stably transfected to express dendritic cell–specific ICAM3-grabbing nonintegrin were infected with each of the DENV serotypes separately in infection media (RPMI 1640 supplemented with 2% FBS, 1% penicillin/streptomycin, and 1% HEPES) for 48 h. Cells were then centrifuged and washed with FACS buffer (1× PBS, 0.5% BSA, and 0.02% sodium azide). Washed cells were then fixed with 2% paraformaldehyde for 30 min and permeabilized for 10 min with 0.05% saponin. Qdot-conjugated detection mAbs were incubated with the cells for 30 min at 1:100 in permeabilization buffer containing 1% normal mouse or human sera. After the incubation, three washes with permeabilization buffer and two washes with FACS buffer were performed. Samples were analyzed on a BD LSR Fortessa flow cytometer (BD Biosciences). Analysis was performed using FlowJo 8.8.7 software (Tree Star).

The membrane of a sterile IPFL clear 96-well plate (Millipore) was activated with 50 μl/well 70% ethanol for 2 min. After three washes with PBS, the plates were incubated with 2 μg/well Fcγ fragment-specific anti-human IgG (The Jackson Laboratory) overnight at 4°C. Wells were blocked with 200 μl RPMI 1640 supplemented with 10% FBS for 2 h at 37°C after the wells were washed three times with PBS. Human hybridomas, EBV-transformed B cells, or activated PBMCs from DENV-positive or -negative individuals were counted and plated in a total of eight serial dilutions of 1:2. PBMCs were activated with IL-2 (1000 U/ml)/R848 (2.5 μg/μl) for 5 d. Cells were incubated at 37°C in 5% CO₂ for 48 h to allow sufficient Ab deposition to occur. After the incubation, cells were washed away with 200 μl PBS-T 0.05% three times with a final PBS rinse. All four OptiPrep-purified DENV serotypes were plated at a final concentration of 1 μg/serotype/well for 1 h at 37°C. After the DENV incubation, wells were washed three times with 200 μl PBS-T 0.05%, with a final PBS rinse. The four Qdot-conjugated, serotype-specific mAbs (50 μl/well at a 1 μg/ml final concentration for each detection mAb) were added to each well and incubated for 1 h at 37°C. Finally, the plate was washed two to three times with PBS and once with double-distilled H₂O and dried using a 96-well MultiScreen Vacuum Manifold (Millipore). Visualization of the fluorescent spots was performed using a CTL ImmunoSpot S6 Micro-Analyzer (Cellular Technology Limited). Specifically, four different emission filters are used, each designated to one serotype. Each emission filter, and therefore serotype, is read by the CTL reader individually, and a merged picture is generated that is the combination of all four images. The software that is installed in the reader counts the spots for each filter/serotype individually. Subsequently, the software identifies the spots that are found in one filter (type-specific) or more than one filter (bi-, tri-, or tetra-serotype specific) through pixel colocalization. To control for background, two negative controls were used. First, stimulated PBMCs from each DENV-positive Nicaraguan sample were processed as above; however, after PBMC incubation and Ab deposition through IgG binding, Qdot-conjugated serotype-specific detection mAbs were added directly without the DENV Ag incubation step. In all cases, no reactivity (background) was observed. In addition, stimulated PBMCs from DENV-negative Oakland Blood Bank donors were used as a negative (background) control. In this instance, the complete QCF procedure was followed, including incubation with DENV Ag. However, because the donors were DENV negative, this control also showed no reactivity.

To determine total IgG-secreting cells, PBMCs were plated in 1:2 serial dilutions of the untreated (ex vivo) or stimulated (in vitro) cells starting from 60,000 PBMCs/well. Wells were coated with anti-human IgG (Fc-specific) Ab. Following cell incubation and Ab deposition, IgG-positive spots were visualized using a polyclonal PE-labeled anti-human IgG detection mAb (H+L). Spots were then counted using the CTL reader.

The QCF is a modified version of an ELISPOT that uses purified virus from all four DENV serotypes and fluorophore-linked serotype-specific mAbs to identify to which DENV serotype(s) Abs from Ab-secreting cells (ASCs) bind. Flipping the configuration of the assay from Ag coated on the plate followed by detection of captured Abs (conventional ELISPOT), to the capture of Abs secreted by B cells on the plate followed by detection of Ag specificity (QCF) enables determination of serotype-specificity on a single-cell basis, rather than only at the cell-population level (Fig. 1A). Specifically, eight serial 1:2 dilutions of ASCs consisting of either EBV-transformed B cells/hybridomas (1 × 10³ in the first well) or activated PBMCs from dengue patients or DENV-immune blood-bank donors (1 × 10⁶ stimulated PBMCs) are incubated for 48 h in a sterile IPFL clear 96-well plate that has been activated, bound with anti-human Fc-specific IgG, and blocked. This captures ASC-secreted Abs, after which the cells are removed via washes. The wells are then incubated with purified virions from all four serotypes in equal concentration (1 μg/serotype/well). Spot specificity is determined by the addition of fluorophore-linked serotype-specific detection mAbs. Excess Ab is removed, and spots are visualized on a fluorescent ELISPOT reader.

Several key technical considerations of this assay are as follows. First, the Ab on the plate membrane used to capture the Abs secreted by ASCs is an Fc-specific anti-human IgG that binds the Fc region of the secreted Ab, leaving its V region accessible for binding virus. Second, equal amounts of each Ag (purified DENV) are used to allow binding to the captured Ab. Third, a majority of the Ab response is directed against the surface of the DENV virion—envelope (E) and premembrane, and many human neutralizing Abs bind to quaternary epitopes on the assembled virion that are not found on recombinant monomeric E (7, 13, 14). Furthermore, using virions instead of recombinant E increases the brightness of the spots due to amplification (up to 180 E protein epitopes per virion, with occupancy determined by each mAb) (15). DENV was propagated in A. albopictus cells (C6/36) and purified via a discontinuous OptiPrep density gradient (16). SDS-PAGE and Coomassie blue staining confirmed the purity of the DENV serotype preparations (Supplemental Fig. 1), and total protein concentration for each batch of Ag was measured by bicinchoninic acid (BCA) (Fig. 1B). By using equal in-excess protein concentrations of the four DENV virions, the Ag specificity of the ASCs is determined by the affinity and avidity of the Ab secreted without bias toward any DENV serotype.

We initially tried to visualize DENV-specific spots by direct or indirect fluorescent labeling of the captured DENV. Directly labeling the purified DENV with protein-amine binding dyes (e.g., Alexa Fluor) or membrane dyes (e.g., DiO, DiI) was not bright enough for visualization in the QCF (data not shown). Therefore, we screened by ELISA a panel of mAbs to identify those with serotype-specific reactivity. Three murine mAbs, all binding E domain III (anti-DENV1 E95, anti-DENV2 E96, and anti-DENV4 E88) (17–19) and one human E domain I/II–directed mAb (anti-DENV3 5J7) (14, 20), were selected. The epitopes targeted are maturation insensitive; thus, the maturation state of the virion should not affect binding of the detection mAbs (20, 21). The serotype specificity of each Ab was assessed by ELISA and flow cytometry, confirming no cross-reactivity between the different DENV serotypes (Fig. 1C). Qdots (22) are tunable nanocrystals (10–30 nm) for which fluorescent properties can be altered by size modification. Their bright fluorescence and minimal spectral overlap made Qdots an attractive candidate for the QCF, so the four serotype-specific mAbs were conjugated to four distinct Qdots (Supplemental Fig. 2).

To assess the ability of the QCF to distinguish the ASC serotype specificity, we first used a human serotype-specific anti-DENV2 hybridoma (mAb 2D22) and human cross-reactive pan-DENV EBV-transformed B cells or hybridomas (mAbs D11C and 1G17). When incubated with the four DENV virions and all four serotype-specific detection mAbs, the QCF distinguished between serotype-specific and cross-reactive ASCs (Fig. 2). To rule out steric hindrance or interference among the four detecting mAbs when used simultaneously, captured Abs secreted by ASCs (pan-DENV D11C) were incubated with all four DENV serotypes and detected using each individual serotype-specific detection mAb separately (Fig. 3A). Conversely, to determine whether addition of four viruses simultaneously hindered determination of the specificity of the ASCs, captured Abs were incubated with a single DENV serotype per well and detected with all four serotype-specific detection mAbs (Fig. 3B). Testing parallel wells plated with equivalent numbers of cells, the number of spots detected with either the four DENV serotypes separately or four individual detection mAbs was comparable to the total number of IgG-secreting cells and to the number of spots detected when all four serotypes and four mAbs were used simultaneously (Fig. 3C, 3D). Thus, inclusion of all four DENV serotypes and all four serotype-specific mAbs in a single well did not interfere with the sensitivity or serotype-specificity of the assay.

We first sought to determine optimal stimulation conditions to activate MBCs into ASCs. PBMCs from six different donors from the Nicaraguan National Blood Center were stimulated under three different conditions: 1) CD40L/IL-21/BAFF; b) CD40L/IL-21; and 3) IL-2/R848 for either 3 or 5 d in vitro. Following stimulation, the total number of IgG-secreting cells was evaluated via FluoroSpot. Five days of in vitro stimulation with IL-2/R848 consistently yielded the highest number of IgG-positive ASCs in every donor (Fig. 4A). We then evaluated the ability of the QCF to determine DENV-specific responses from DENV-exposed or DENV-naive blood donors from Nicaragua or California, respectively. PBMCs from blood bank donors were stimulated for 5 d with IL-2 and R848 and analyzed by QCF. Both serotype-specific and cross-reactive MBCs were observed in the MBC population from DENV-exposed blood donors (Fig. 4B, 4C), whereas no DENV-reactive spots were seen in the DENV-naive control (Fig. 4D).

We next sought to assess whether the QCF could be used to detect MBC and PB/PC responses in primary human PBMCs from DENV-infected individuals during either acute- or convalescent-phase time points in participants in our Hospital-based Dengue Study in Nicaragua (11, 23). PBMCs from a DENV-positive patient (secondary DENV3) were plated directly ex vivo and assessed via QCF to determine the PB/PC response. Cells were serially diluted as above, and DENV-specific ASCs were detected with the four DENV serotypes and four DENV serotype-specific detection mAbs. The Nicaraguan PB/PCs ex vivo were serotype cross-reactive (Fig. 5A). Next, convalescent PBMCs from a patient with primary DENV3 infection, a patient with secondary DENV3 infection, and a DENV-negative blood bank donor from Oakland as a negative control were stimulated in vitro with IL-2/R848 for 5 d and analyzed via QCF. MBCs from the primary convalescent sample revealed a predominantly serotype-specific response against DENV3 (Fig. 5B), consistent with previous reports (10), whereas the secondary convalescent sample displayed more serotype cross-reactive MBCs (Fig. 5C). The number of B cells specific for one, two, three, or four DENV serotypes was enumerated (Fig. 5). The DENV-negative control displayed no B cell reactivity to DENV (data not shown).

We have developed a novel modification of the conventional ELISPOT, the QCF, to detect DENV serotype-specific and cross-reactive B cell responses on a single-cell basis. Using DENV-specific human EBV-transformed B cells, we showed that visualization of ASCs with all four DENV viruses and all four DENV-specific mAbs was as sensitive as detection with a single virus or a single detection mAb. Furthermore, the QCF was able to quantify serotype-specific and cross-reactive single B cells from the PBMCs of DENV-exposed blood donors and DENV-infected patients, in both PB/PCs and MBCs.

The correlates of protection against DENV infection and/or severe disease are not well understood, but are thought to be associated with a strongly neutralizing Ab response. However, infection with one DENV serotype imparts long-term protection against reinfection with the same serotype but not against infection from heterologous serotypes, and secondary infection with a different serotype increases the risk for developing severe disease, thought to be due at least in part to Ab-dependent enhancement (24). Recent studies have provided valuable information about MBC specificity at a B cell population level by conventional ELISPOT (10, 11) and by using human mAbs generated from DENV-exposed individuals (4, 5, 7, 14). Although informative, these studies have not addressed how Ab and MBC serotype-specific or cross-reactive responses develop over time and mature after primary or secondary natural DENV infection on a single-cell basis. It is critical to understand how the MBC response evolves and what constitutes immune correlates of protection in natural DENV infections. Thus, it is imperative to better understand B cell and Ab responses to natural DENV infections, which will in turn facilitate evaluation of vaccine candidates in clinical trials. The QCF technology will enable an in-depth analysis of the generation and evolution of the specificity of B cell responses in primary and secondary DENV infections as well as vaccinees and will permit B cell serotype specificity or cross-reactivity to be evaluated as a potential immune correlate of protection in natural DENV infections and in trials of candidate dengue vaccines.

Finally, the QCF can be adapted to assess other diseases in which identifying the serotype- or subtype-specific B cell response is paramount. In addition to DENV, other infections such as influenza (25), HIV (26, 27), and severe acute respiratory syndrome-coronavirus (28) have been reported to generate Ag-specific Ab responses that can enhance infection and disease progression. In the case of HIV and influenza, it has been hypothesized that vaccination with the wrong virus could lead to enhanced infection within the vaccinated population (26). Thus, understanding the specificity and cross-reactivity of B cells generated by a vaccination protocol against different viral subpopulations could be vital to predicting vaccination safety and efficacy; the QCF is a tool that could be used to facilitate this analysis.

We thank Michael S. Diamond (Washington University, St. Louis, MO) for mAbs E88, E95, and E96 and the CD40L-producing cell line; John S. Schieffelin (Tulane University) for D11C EBV-transformed B cells; and James Crowe (Vanderbilt University) for the 2D22 and 1G17 hybridomas and 5J7 mAb. We also thank Zoltán Megyesi for image processing support and Tameem Ansari (Cellular Technology Limited) for technical support.

The authors have no financial conflicts of interest.