Dengue Virus Neutralization in Cells Expressing Fc Gamma Receptors

Activating Fc gamma receptors (FcγRs) in hematopoietic cells serve to remove antibody-opsonized antigens, including dengue virus (DENV), from systemic circulation. While neutralizing antibody concentrations provide humoral immunity, cross-reactive or sub-neutralizing levels of antibody can result in antibody-dependent enhancement of DENV infection that increases overall viral burden. Recently, it has been suggested that the antibody levels needed for DENV neutralization differs when different FcγR is engaged. If this is true, the threshold titer used to infer immunity should be influenced by FcγR usage. Here, using cells that express both activating and inhibitory FcγRs, we show that the type of FcγR engaged during phagocytosis can influence the antibody concentration requirement for DENV neutralization. We demonstrate that phagocytosis through FcγRI requires significantly less antibody for complete DENV neutralization compared to FcγRIIA. Furthermore, when DENV is opsonized with sub-neutralizing levels of antibody, FcγRI-mediated phagocytosis resulted in significantly reduced DENV titers compared to FcγRIIA. However, while FcγRI may remove antibody-opsonized DENV more efficiently, this receptor is only preferentially engaged by clustering when neutralizing, but not sub-neutralizing antibody concentrations, were used. Collectively, our study demonstrates that activating FcγR usage may influence antibody titers needed for DENV neutralization.


Introduction
Dengue is the most common mosquito-borne viral disease globally. It is caused by a positive-strand RNA virus, which exists as four antigenically distinct serotypes. Infection with dengue virus (DENV) results in a spectrum of illness that ranges from undifferentiated fever to severe dengue that comprises hypovolemic shock from plasma leakage, internal hemorrhage or organ dysfunction. While antibody response triggered during the acute infection result in lifelong immunity to the homologous serotype, infection with a heterologous DENV serotype or during a time where maternally acquired antibodies wane in infants have been shown to be epidemiologically associated with increased risk of severe dengue [1][2][3]. Cross-reactive or sub-neutralizing levels of antibodies offer DENV with an alternative pathway of entry into monocytes, macrophages and dendritic cells through the activating Fc gamma receptors (FccRs). This pathway of infection, termed antibody-dependent enhancement of DENV infection (ADE), is hypothesized to be an important mechanism in the pathogenesis of severe dengue [3][4][5][6][7]. FccRs are broadly expressed by cells of hematopoietic origin and is composed of activating (FccRI, FccRIIA, and FccRIIIA) and inhibitory (FccRIIB) receptors [8]. While these receptors could contribute to ADE [9,10], they are important in the removal of DENV opsonized with neutralizing levels of antibody. Delineating the determinants of neutralization or ADE upon FccR-mediated phagocytosis would thus be important for the understanding of immunity and pathogenesis, respectively, which could prove useful in refining vaccine development to overcome the currently observed limited immunity with the leading dengue vaccine candidate [11].
Stoichiometric studies have shown that neutralization of flavivirus is a ''multi-hit phenomenon'', which occurs when the number of antibodies bound to a virus exceeds a required threshold and is dependent on antibody affinity and epitope accessibility [12][13][14]. However, the stoichiometric requirement for DENV neutralization may be different when phagocytosis is mediated by either FccRI or FccRIIA. Rodrigo and colleagues used a panel of monoclonal antibodies to demonstrate that DENV neutralization required significantly lower antibody concentration in CV-1 cells transfected with FccRI compared to FccRIIA [15]. However, the gamma subunit containing the immunoreceptor tyrosine activating motif that signals for phagocytosis upon was covalently linked FccRI in the transfected cells whereas in cells that naturally express this receptor, the gamma subunit is only recruited upon activation of the receptor [16]. Whether the experimental design adopted by Rodrigo and colleagues [15] affected the outcome of the antibody concentration needed for complete DENV neutralization, is unknown. We hence utilized cells that naturally express FccRs to investigate the antibody concentration requirements for DENV neutralization. We show here that more antibodies are required for DENV neutralization with FccRIIA-compared to FccRI-mediated phagocytosis. Furthermore, when both receptors are expressed together, DENV opsonized with neutralizing levels of antibody preferentially engage FccRI by clustering this receptor on the cell membrane.

Cells and Antibodies
BHK-21, THP-1, K562 and Vero cells were purchased from the American Type Culture Collection (ATCC) and cultured according to ATCC recommendation. 3H5 is a monoclonal antibody that binds to domain III of DENV envelope protein. A chimeric human antibody of 3H5 (h3H5) IgG1 was constructed consisting of mouse VH and VL sequences and human c1 and k constant sequences [17]. These antibodies were indistinguishable from the parent 3H5 mAb in their ability to bind to DENV-2 [18]. Antibodies used for flow cytometry staining, western blot and immunofluorescence assay (IFA) were: FccRI antibody clone 10.1 (eBioscience), FccRII clone IV.3 (Stem cell biology), FccRIIB (Abcam), LAMP-1 (BD biosciences, Abcam), Cy3 anti-LAMP-1 (Sigma) and HRP conjugated anti-mouse (Dako). All Alexa Fluor labeled antibodies were purchased from Invitrogen and used at 1:200 dilution.

Virus culture and purification
DENV-2 (ST strain) was first isolated from a clinical sample from Singapore General Hospital. Viruses were propagated in Vero cell line and harvested 5 days post infection (dpi) and purified through 30% sucrose. Virus pellets were resuspended in 5 mM Hepes, 150 mM NaCl, and 0.1 mM EDTA (HNE) buffer, aliquoted and stored at 280uC until use. Infectious titer was determined by plaque assay.

Plaque Assay
Ten fold serial dilutions of virus culture supernatant were added to monolayer of BHK-21 cells in 24-well plates and incubated for 1 h at 37 uC with gentle rocking every 15 mins. The inoculum was aspirated, replaced with 0.8% methyl-cellulose in maintenance medium (RPMI-1640, 2% FCS, 25 mM Hepes, penicillin, and streptomycin) and incubated at 37 uC. After 5 dpi, cells were fixed with 20% formaldehyde at room temperature for 20 mins, washed with water, and stained with 1% crystal violet for additional 20 mins. The plates were washed, dried, and the plaque forming units per milliliter (pfu/mL) were calculated.
Titration of h3H5 antibody for complete neutralization in THP-1 or K562 cells Two fold serial dilutions of h3H5 (200 mg/mL to 0.097 mg/mL) were incubated with DENV at a multiplicity of infection (MOI) 10 for 1 h at 37uC and then added to THP-1 or K562 cells, subjected to synchronization for 20 mins on ice and then incubated at 37uC for 72 h. The virus culture supernatants were harvested and quantified by plaque assay. The antibody dilution required to mediate full virus neutralization was then determined using the following formula:% neutralization = {[virus only (pfu) -(virus + antibody) (pfu)]/virus only (pfu)} 6100
DENV was labeled with Alexa Fluor as described previously [21]. Briefly, ,9610 8 pfu purified DENV was incubated with 100 mM of Alexa Fluor 488 (AF488) succinimidyl ester (Invitrogen) for 1 h at room temperature. The labeling reaction was then stopped by adding 1.5 M hydroxylamine, pH 8.5, and incubated at room temperature for 1 h. The excess dye was then removed by gel filtration on a Sephadex G-25 column. AF488-labeled DENV (AF488-DENV) was stored in 100 mL aliquots at 280uC, retitrated by plaque assay, and tested for fluorescence using IFA on Vero cells before using in experiments.

Infection for localization studies in THP-1 or K562 cells
Concentrations of h3H5 required for complete neutralization in THP-1 (3.125 mg/mL) or K562 (25 mg/mL) were incubated with DiD-DENV (MOI 10) for 1 h at 37uC. The immune complexes were added to cells, synchronized on ice for 20 mins and incubated for 30 mins at 37uC. Cells were then fixed with 3% paraformaldehyde (PFA) in 16PBS for 30 mins at 4uC. Fixed cells were processed for IFA.
Neutralizing (3.125 mg/mL) or sub-neutralizing (0.390 mg/mL) concentrations of h3H5 were incubated with AF488-DENV (MOI 10) for 1 h at 37uC. Infection was then carried out in THP-1 for 15, 30, 60 and 120 mins as mentioned above. The infected cells were then fixed and sorted using Fluorescence-activated cell sorting (FACS) before processing for IFA.
To quantify co-localization of AF488-DENV with FccRs, 10 cells were selected to calculate percentage co-localization of DENV with FccRI or FccRII at 120 mins post infection by overlap coefficient using Zen 2009 software. The mean intensity of FccRI or FccRII when co-localized with DENV was evaluated using a tool Histo in Zen 2009 software for 15, 30, 60 and 120 mins post infection. An area of 70.560.28 mm 2 was analyzed on each cell selected from 10 different fields for all time points.

Fluorescence-activated cell sorting (FACS)
THP-1 infected with AF488-DENV reacted with neutralizing, sub-neutralizing antibody or in absence of antibody for different time points (15,30, 60 and 120 mins) of infection was fixed and sorted using BD FACS Aria II cell sorter at Duke-NUS FACS facility. The AF488 positive sorted cells were then subjected to IFA.

siRNA transfection in THP1 or K562
siRNA knockdown studies in THP-1 have been previously described [18]. 50 nM of human FccRI or FccRIIA siRNA (Qiagen) or All-Stars scrambled control siRNA (Qiagen) were used for the knockdown studies. For K562, studies were performed with slight modifications. Human FccRIIB siRNA (Qiagen) or All-Stars scrambled control siRNA (Qiagen) (50 nM) were incubated with DharmaFect2 (Dharmacon) in serum-free media for 20 mins and then added to cells at a density of 2610 5 cells/mL. After 6 h incubation, cells were replaced with RPMI supplemented with 10% fetal calf serum (FCS) for 24 h to allow recovery. This was followed by a second round of siRNA transfection. Knockdown efficiency was determined by western blot or flow cytometry.

Flow Cytometry to determine surface expression of FccRs
THP-1 or K562 cells were stained with FccRI or FccRII antibody for 30 mins on ice, washed three times using 1XPBS with 1% FCS followed by 30 mins of staining with secondary antibody, AF488 anti-mouse IgG, on ice. After final washes using 1XPBS supplemented with 1% FCS, FACS data acquisition was performed on a BD LSR Fortessa.

Western Blot
Cells were washed once in 1XPBS and lysed in 1% NP-40 with protease inhibitor (Sigma). The cell lysates were centrifuged to remove insoluble aggregates, mixed with loading buffer and separated by SDS-PAGE before transferring to PVDF (Millipore). FccRIIB and LAMP-1 were detected with specific antibodies followed by addition of anti-mouse IgG-horseradish peroxidase (HRP). Bands were visualized using ECL (Amersham) for chemiluminescence development.

Statistical Analysis
Two-tailed unpaired Student's t-test or one-way ANOVA were done using GraphPad Prism v5.0. Results with P,0.05 were considered significant.

FccRIIA-mediated phagocytosis requires increased antibody concentration for DENV neutralization
We recently reported the use of humanized 3H5 monoclonal antibody (h3H5) to investigate FccR-mediated phagocytosis in THP-1, a human monocytic cell line that expresses both FccRI and FccRIIA [18]. However, titration of h3H5 in K562, a human myelogenous erythroleukemic cell line that expresses FccRIIA but not FccRI ( Figure 1A) required eight-fold more antibody for complete DENV neutralization compared to THP-1 ( Figure 1B). Using DiD labeled DENV that emits fluorescence only upon phagocytosis [19,20], we observed that neutralization of DENV in K562 occurred at an antibody concentration where FccRmediated phagocytosis was inhibited as indicated by reduced DiD positive cells ( Figure 1C and 1D). These findings suggest the h3H5 concentration required for complete DENV neutralization in K562 coincides with that which aggregates DENV to co-ligate FccRIIB that inhibits phagocytosis, a mechanism that we demonstrated recently [18].
That complete DENV neutralization in K562 coincided with FccRIIB-mediated inhibition of phagocytosis raises the possibility that an even greater amount of antibody is needed to neutralize DENV if phagocytized by FccRIIA. To test this possibility, we knocked down the expression of FccRIIB in K562 using siRNA (Figure 2A). This resulted in increased uptake DiD-DENV opsonized with h3H5 ( Figure 2B and 2C). However, plaque assay on the culture supernatant indicated that reduced FccRIIB expression did not result in a further increase in h3H5 antibody concentration needed for complete DENV neutralization ( Figure 2D).
As THP-1 and K562 are two different cell lines, we examined if significantly different antibody concentration is needed for DENV neutralization if FccRI or FccRIIA expression were respectively altered in THP-1. Reduced expression of FccRI ( Figure 3A and B) resulted in a four-fold increase in the h3H5 concentration needed for DENV neutralization compared to cells with reduced FccRIIA expression ( Figure 3C). Interestingly, reduced expression of FccRIIA but not FccRI also resulted in lowered DENV titers even with enhancing levels of h3H5 ( Figure 3D). These findings collectively indicate that removal of antibody-opsonized DENV is more efficient with FccRI than FccRIIA.

DENV opsonized with neutralizing but not subneutralizing levels of antibody preferentially engage FccRI
Besides reduced antibody concentration requirement, immunofluorescence examination of THP-1 suggests that FccRI is preferentially engaged by DENV opsonized with neutralizing levels of antibody. As only a subset of THP-1 actively phagocytize antibody-opsonized DENV [18], we enriched for DENV containing cells by sorting for AF488-DENV [21] before affixing the cells on a glass slide for microscopic examination ( Figure 4A). At 120 mins post-synchronization, co-localization of DENV, FccRI and LAMP-1 was observed ( Figure 4B). Quantification of the colocalization signals between DENV and FccRI or FccRIIA in 10 cells obtained from 10 fields at 636magnification, using Zen 2009 software indicated a significantly higher co-localization signal with FccRI than FccRIIA, but only when neutralizing levels of h3H5 was used ( Figure 4C). At sub-neutralizing concentrations of h3H5, however, no difference was observed between the co-localization of AF488-DENV with either FccRI or FccRIIA ( Figure 4B and 4C). As expected, no co-localization could be observed between either FccRI or FccRIIA with DENV only infection.
The increased co-localization between DENV and FccRI suggests that a more efficient pathway is preferentially activated for removal of virus opsonized with neutralizing antibodies. However, the lack of difference in FccR engagement when subneutralizing h3H5 was used is intriguing. Although FccRI is known to have a greater affinity for IgG1 than FccRIIA [22], it cannot explain this difference in FccRI engagement between neutralizing and sub-neutralizing h3H5 since both experiments made use of the same IgG isotype. Instead, the observation may be explained by an antibody-concentration dependent clustering of FccRI, which has previously been shown as a mechanism to activate this receptor [23,24]. A time course examination coupled with sorting for cells containing AF488-DENV ( Figure 5A) showed increased clustering of FccRI but not FccRIIA with increasing time post-synchronization ( Figure 5B). To quantify the clustering of FccRs when co-localized with AF488-DENV, we selected an area of 70.560.28 mm 2 on 10 cells from 10 separate fields under 636 magnification and measured signal intensity of FccRI and FccRIIA using Histo tool in Zen 2009 software. Increased FccRI signal intensity could be observed with increasing time postsynchronization and this was significantly higher than that for FccRIIA at 120 mins ( Figure 5C). This indicates that neutralizing levels of h3H5 was able to cluster and preferentially engage FccRI for phagocytosis.

Discussion
Whether antibodies neutralize or enhance DENV infection is determined by both antibody affinity and epitope occupancy [12][13][14]. However most of these studies for DENV neutralization have made use of cells derived from kidney of various animals such as LLC-MK2, Vero, and BHK-1 cells [25]. These cells neither express FccRs nor are primary targets of DENV in human infections. Recently, it is becoming evident that neutralizing antibody measurements on epithelial cells result in different titers compared to assays that use FccR-expressing cells [10,15,26,27]. Furthermore, we have also shown that besides blocking specific epitope receptor interaction, antibodies can also aggregate DENV in a concentration-dependent manner to co-ligate the lowly expressed FccRIIB that inhibits phagocytosis and hence ADE  [18]. This also appears to be the mechanism in which neutralization of heterologous DENV serotype occurs [28]. Understanding DENV neutralization in cells that express FccR thus represents an area for urgent investigation given the recently observed lack of efficacy in vaccines that have relied on traditional virus neutralization test as surrogate of protection [11].
Our findings using cells that naturally express FccRs corroborate earlier observations that used epithelial cells transfected with FccRs [15]. We observed that the antibody requirement for DENV neutralization was increased when either K562 or THP-1 with reduced FccRI expression, was used. In contrast, THP-1 with reduced FccRIIA expression resulted in reduced antibody requirement for DENV neutralization. We have chosen FccRI and FccRIIA for our investigation as they have been previously shown to mediate specific DENV immune complex infectivity in monocytes [6,7,10,29,30]. FccRIIIA, on the other hand, is expressed at low levels in a small subset of monocytes [31] and does not affect susceptibility to DENV infection [7]. We demonstrate that depending on whether FccRI or FccRIIA mediates phagocytosis, the required threshold of epitopes that must be bound by antibody is different.
FccRI is an activating receptor that recruits the gamma subunit with immunoreceptor tyrosine-based activating motif to phosphorylate kinases that signal for phagocytosis [16], pro-inflammatory responses [32], protection from bacteria [33] and viruses [34]. Our study highlights the involvement of FccRI in phagocytosis and neutralization of DENV. Even when DENV was opsonized with enhancing levels of h3H5, phagocytosis through FccRI produced significantly lower DENV titers. This is consistent with previous report showing that DENV titers were enhanced to a greater effect with FccRIIA instead of FccRIA/c-expressing COS cells [10]. The advantage offered by FccRI can perhaps be explained by differences in the signaling pathway. A recent study has shown differences in intracellular signaling pathways, receptor trafficking and antigen processing at the early stages between FccRI and FccRIIA activation [35]. While both FccRI and FccRIIA phagocytize and traffic antibody-opsonized antigens to early endosome compartment (EEA-1), only antigens taken up by FccRI were trafficked to late endosomal/lysosomal compartments (LAMP-1) [35]. Hence, FccRI signaling pathways may traffic DENV opsonized with neutralizing levels of antibody into compartments that leads to virus degradation. Conversely, FccRIIA trafficking may direct DENV into an intracellular environment favorable for replication. Further studies will be needed to substantiate this notion.
That FccRI possibly offers a more efficient pathway for the clearance of antibody-opsonized DENV also led us to ask how this receptor could be preferentially engaged. We have observed that increased co-localization of DENV with FccRI relative to FccRIIA when neutralizing but not sub-neutralizing level of antibody was used. This observation suggests that neutralizing levels of antibody bound on viral surface not only serves to meet the threshold of epitope occupancy, it also clusters FccRI for preferential activation of this receptor for phagocytosis. It may also be possible that a positive feedback loop exists to augment FccRImediated phagocytosis as activation of FccRI can induce potent inflammatory response [35] that could increase the clustering of FccRI and thus binding of immune complexes with it for phagocytosis [23,24].
Given the role of FccRI in the clearance of antibody-opsonized DENV suggests that, it is possible that the variable expression of this receptor between different ethnic groups [36] and age [37] could influence the outcome of antibody-enhanced DENV infection. Furthermore, FccRI expression has been found to be correlated with interferon-gamma (IFNc) levels [38], which may partially explain the observed reduced ADE in IFNc-treated human peripheral blood monocytes [39]. It may also be possible that reduced IFNc expression in the early febrile stage of illness resulted in reduced FccRI expression and hence viral clearance in patients that go on to develop DHF [40]. Studies that address these questions may further clarify the role different types of FccR play in dengue immunity and pathogenesis.
In conclusion, FccRI-mediated phagocytosis plays an important role in the removal of antibody-opsonized DENV.