Ebola Virus Secretory Glycoprotein (sGP) Diminishes FcγRIIIB-to-CR3 Proximity on Neutrophils¹

Hemorrhagic fever is a highly lethal disease that affects both humans and nonhuman primates. One infectious agent responsible for hemorragic fever is Ebola virus, named after a river in Africa where it was first recognized (1, 2). One viral product is a secreted glycoprotein (sGP)⁴ of 50–70 kDa (3). Recent studies have suggested that sGP binding to human neutrophils is dependent on FcγRIIIB (CD16b) and that it acts in vitro to inhibit L-selectin down-regulation (4), which may affect host defense mechanisms. Because FcγRIIIB is a GPI-linked membrane protein, it is unclear how sGP-FcγRIIIB complexes affect neutrophil activation. One mechanism of FcγRIIIB signaling is via lateral membrane interactions with CR3 (5, 6, 7, 8, 9, 10). Early studies showed that FcγRIIIB cocaps with CR3 and that immune complexes trigger CR3 interactions with microfilaments (5, 11). Moreover, FcγRIIIB has been observed in proximity to CR3 on neutrophil membranes in the absence of a capping stimulus using resonance energy transfer (RET) (7, 8). Further studies have shown that certain saccharides and polysaccharides can affect FcγRIIIB-dependent cell activation and calcium signaling by competing with FcγRIIIB for the lectin-like binding site of CR3 (9). Stockl et al. (10) have mapped the FcγRIIIB binding site to the membrane proximal region of CD11b. Thus, the interactions of FcγRIIIB and CR3 contribute to inflammatory signaling. Therefore, we hypothesized that the inhibitory effect of sGP on neutrophil functions may be due to an sGP-mediated disturbance of the FcγRIIIB interactions with CR3.

N-acetyl-d-glucosamine (NADG) was obtained from Sigma (St. Louis, MO). FITC and tetramethylrhodamine isothiocyanate (TRITC) were obtained from Molecular Probes (Eugene, OR).

sGP was prepared as described previously (12). Briefly, an expression vector for sGP and a plasmid control were transfected into human 293 cells. Culture supernatants containing sGP or controls were used in the experiments described below.

Neutrophils were isolated from peripheral blood obtained from healthy adults using heparinized tubes, by Ficoll-Hypaque (Sigma) step-density gradient centrifugation. The cell preparation was ∼95% neutrophils. The purified cells were typically 95% viable, as assessed by trypan blue exclusion.

Mouse mAbs to CR3 (anti-Mo1, clone 44) were obtained as described previously (5). F(ab′)₂ fragments were prepared as described (5). F(ab′)₂ fragments directed against FcγRIII (clone 3G8) were obtained from Medarex (West Lebanon, NH). Intact mAbs directed against FcγRIII clones DJ130c (IgG1) and 1D3 (IgM) were obtained from Dako (Carpinteria, CA) and Colter/Immunotech (Hialeah, FL), respectively.

mAb or their F(ab′)₂ fragments were dialyzed against 0.15 M carbonate-bicarbonate buffer at pH 9.3 for 4 h at 4°C (5). Samples were incubated with dyes at a F/P ratio of 40 μg TRITC or 30 μg FITC per mg Abs at room temperature for 4 h. The fluorescent conjugates were separated from unreacted fluorochromes by Sephadex G-25 (Sigma) column chromatography. Purified conjugates were dialyzed against PBS at pH 7.4 overnight at 4°C.

Neutrophils in suspension were labeled with fluorochrome-conjugated IgG (DJ130c), IgM (1D3), F(ab′)₂ fragments (3G8) of anti-FcγRIIIB, or F(ab′)₂ fragments of anti-CR3 (clone 44) mAbs for 30 min at 4°C. The cells were washed twice and then labeled with a second mAb to create FITC-TRITC pairs.

Cocapping experiments were performed as described previously (5). CR3 was capped using F(ab′)₂ fragments of clone 44. Cells were capped using goat F(ab′)₂ fragments of a goat anti-murine F(ab′)₂ IgG (5). To avoid interference with sGP binding, capped cells were probed for cocapping using clone DJ130c.

An axiovert inverted fluorescence microscope with HBO-100 mercury illumination (Carl Zeiss, New York, NY) interfaced to a Dell 410 workstation (Round Rock, TX) via Scion SG-7 video card (Vay Tek, Fairfield, IA) was employed. The fluorescence images were collected by an intensified charge-coupled device camera, model XC-77 (Hamamatsu, Hamamatsu City, Japan) and processed with ScionImage software (Vay Tek). A narrow bandpass-discriminating filter set was used with excitation at 485DF22 nm and emission at 530DF30 nm for FITC and an excitation of 540DF20 nm and emission of 590DF30 nm for rhodamine (Omega Optical, Brattleboro, VT). Long pass dichroic mirrors of 510 nm and 560 nm were used for FITC and TRITC, respectively. For energy transfer imaging, the 485DF22, 5101p, and 590DF30 filter combination was used (13). In some experiments the RET intensity was quantitated using a photomultiplier tube (Hamamatsu) held in a Products for Research (Danvers, MA) housing. The signal was processed using a discriminator/amplifier (model 1762, Photochemical Research Associates, London, Ontario, Canada) and quantitated using a photon counter PRA model 1770. An average of three readings per target cell were used for each datum point. At least 100–150 cells were studied in each trial.

RET was also assessed using a microscope/imaging spectrophotometer system (e.g., see Ref. 14). To minimize light losses, a Zeiss IM-135 axiovert microscope with a bottom port was employed. The bottom port was fiber-optically coupled with an efficiency near 1.0 to the input side of an Acton-150 (Acton, MA) imaging spectrophotometer. The exit side was connected to a liquid nitrogen-cooled intensifier which was, in turn, attached to a Peltier-cooled I-MAX-512 camera (Princeton Instruments, Trenton, NJ). The collection of spectra or images was controlled by a high-speed Princeton ST-133 interface and a Stanford Research Systems (Sunnyvale, CA) DG-535 delay-gate generator. This high-efficiency, high-sensitivity system allowed medium-resolution emission spectra to be collected from individual cells in less than 1 s. These systems were interfaced to a Dell 410 workstation running Winspec software (Princeton Instruments) to manage and analyze data. Cells were illuminated using an optical filter at 485DF22 nm and a 510lp dichroic mirror. By employing a 520lp emission filter we were able to simultaneously collect spectra at both donor and acceptor emission wavelengths. This enabled us to simultaneously monitor the RET-mediated acceptor emission and the donor’s fluorescence-intensity quenching.

To test the hypothesis that Ebola virus sGP interferes with interactions between CR3 and FcγRIIIB, we performed RET experiments with labels attached to CR3 and FcγRIIIB on neutrophils after incubation with sGP or control culture supernatants. For RET imaging, neutrophils were incubated in the presence of cell culture supernatants from sGP or control-transfected cells. Samples were first stained for 30 min at 4°C with rhodamine-conjugated reagents directed against FcγRIIIB (CD16) and then with a fluorescein-conjugated reagent directed against CR3 (CD11b; clone 44). Three reagents directed against FcγRIIIB were employed: F(ab′)₂ fragments of clone 3G8, which compete with sGP for FcγRIIIB binding; an intact IgM of clone 1D3; and intact IgG of clone DJ130c, which do not interfere with the binding of sGP to cells (Ref. 4 and our unpublished observations). Samples were incubated for 20 min. at 4°C with various dilutions of control or sGP supernatants and then they were placed on a 37°C microscope stage for further analyses.

Images of fluorescein, rhodamine and the RET emission channel were collected with an intensified charge-coupled device camera. A significant level of RET emission was observed using all anti-FcγRIIIB reagents (clones 3G8, DJ130c, and 1D3) in the presence of control supernatants (Fig. 1, A–I). In contrast, exposure of neutrophils to sGP supernatants dramatically decreased RET intensity, as assessed by RET imaging, between CR3 and FcγRIIIB (noncompeting clones DJ130c and 1D3) (Fig. 1, L and O). However, the RET level of sGP-treated cells labeled with the sGP-competing anti-FcγRIIIB (clone 3G8) reagent was not affected (the clone 3G8-labeled receptors cannot interact with sGP) (Fig. 1, P–R). Thus, clone 3G8, which blocks sGP binding, served as a negative control. Although the RET level was dramatically decreased between CR3 and noncompeting FcγRIIIB reagents in the presence of sGP (Fig. 1, L and O), the anti-CR3 and anti-FcγRIIIB labels remained colocalized at the cell surface (Fig. 1, J, K, M, and N). Thus, sGP appeared to block CR3 and FcγRIIIB proximity (<7 nm) without affecting colocalization.

The above findings were extended and confirmed quantitatively by microfluorometry. We first verified label stability in the presence of sGP. Cells were first labeled with anti-CR3 or anti-FcγRIIIB (clone DJ103c) as described above followed by exposure to control or sGP supernatants (Table I). After 15 min at 37°C, there was no statistically significant change in the staining intensity of either CR3 or FcγRIIIB when treated with control or sGP supernatants. Therefore, incubation with sGP did not cause the shedding (or displacement) of the labels or quench the fluorescence of either fluorescein or rhodamine. Quantitative dose-response studies were then conducted at the RET emission wavelength. Cells incubated with control supernatants demonstrated significant RET levels (Fig. 2, a–c), as qualitatively seen in the images of Fig. 1, C, F, and I. Incubation of neutrophils with sGP dramatically decreased RET intensity between CR3 and FcγRIIIB for cells labeled with mAb 1D3 and DJ130c (Fig. 2, a and b). Moreover, the RET intensity decreased in an sGP dose-dependent fashion. There was no reduction of the RET intensity from CR3 to FcγRIIIB in cells incubated with sGP and labeled with an anti-FcγRIIIB mAb (clone 3G8) that blocks sGP binding to FcγRIIIB (Fig. 1,C). This finding is consistent with the fact that mAb (clone 3G8) blocks sGP binding to cells. Another GPI-linked protein, urokinase-type plasminogen activator receptor (uPAR; CD87), is also known to interact with CR3 (15, 16). To verify the specificity of sGP-FcγRIIIB-CR3 interactions, we performed quantitative RET measurements between CR3 and uPAR in the presence of sGP. No change in RET intensity was observed (Fig. 2 d) suggesting that sGP is specific for the FcγRIIIB interaction with CR3.

The RET imaging and quantitative microfluorometry experiments described above measure energy transfer by examining one spectral region near the acceptor’s emission. Another physical means of studying RET is the assessment of the donor chromophore’s emission spectrum in the presence and absence of the acceptor. In contrast to the above studies, single-cell spectrophotometry simultaneously measures cell emission properties at many wavelengths across the visible spectrum. We first examined the emission spectra of cells individually labeled with either fluorescein anti-CR3 or rhodamine anti-FcγRIIIB reagents alone (Fig. 3,A). The fluorescence emission spectra and intensities of fluorescein and rhodamine labels were not affected by sGP (Fig. 3,B). For RET studies, emission spectra (>520 nm) were collected from individual cells. For cells treated with control supernatants, robust acceptor emission at ∼590 nm was observed in the presence of both the donor and acceptor but not in the presence of the donor alone (Fig. 3, C, E, G, and I, bold lines vs thin lines). However, emission spectra of sGP-treated neutrophils labeled with CR3 and 1D3 or DJ130c reagents were characterized by minimal emission beyond 570 nm (Fig. 3, D and F). In contrast, RET emission was observed between CR3 and FcγRIIIB (clone 3G8) or uPAR in both the presence and absence of sGP (Fig. 3, G–J). A careful inspection of Fig. 3 reveals a reduction (or quenching) in the FITC emission intensity for all samples demonstrating RET (Fig. 3, C, E, and G–J). This reduction in FITC emission is due to energy transfer to TRITC. Thus, these data provide broad spectral information and confirm RET using FITC quenching, a second physical parameter.

RET is sensitive to the relative densities of donor and acceptor chromophores in cell membranes (17). Therefore, we further developed these analyses by acquiring emission spectra at various donor:acceptor ratios. Cells were labeled using reagents directed against CR3 (clone 44; donor) and FcγRIIIB (clone 1D3; acceptor) at subsaturating doses with ratios of 5:1, 1:1, and 1:5. Fig. 4 shows representative emission spectra illustrating the relative amounts of donor-quenching and RET acceptor emission at three labeling ratios (Fig. 4, A, C, and E) for control supernatant-treated neutrophils. We next added a saturating dose of sGP to the same cells of Fig. 4, A, C, and E and collected emission spectra of the donor (>520 nm) and acceptor (>570 nm) chromophores (Fig. 4, B, D, and F). Although RET is not observed in the presence of sGP (e.g., Fig. 4,B, thin line), the presence of sGP allows the direct confirmation of amount of donor and acceptor chromophores associated with the cells without the RET effects of donor quenching or acceptor emission. The experiments of Fig. 4, B, D, and F confirm the approximate levels of labeling anticipated by the mAb additions to the samples. When the density of the donor exceeded that of the acceptor (Fig. 4,B), donor-quenching and the longer-wavelength RET emission were observed in the absence of sGP (Fig. 4,A). Equal amounts of the donor and acceptor chromophores were used for the spectra of Fig. 4, C and D. A bimodal spectrum showing fluorescence emission from the donor and RET was noted in the absence of sGP. When the number of the acceptor molecules exceeded the donors (Fig. 4,F), the relative RET emission intensity increased dramatically (Fig. 4, F vs C). These findings further confirm the results described above by demonstrating the dependence on donor:acceptor ratio and the ability of sGP to block RET at these ratios. Moreover, to our knowledge, this is the first demonstration of the label density-dependence of RET in cells.

These experiments demonstrated that energy transfer between CR3 and FcγRIIIB was reduced by sGP. These data reflect the molecular proximity (7 nm) of receptors, not their global distribution. The colocalization experiments mentioned above (Fig. 1, J, K, M, and N) suggest that CR3 and FcγRIIIB remain associated, although the labeled receptors no longer display RET. To directly test this possibility, we performed cocapping experiments. As previously reported (5), significant levels of CR3 capping and CR3-FcγRIIIB cocapping were observed (Fig. 5, A–C; Table II); however, sGP had no influence on cocapping (Fig. 5, D and E; Table II). Fig. 5 shows micrographs of the same cell recorded before (A–C) and after (D–F) exposure to sGP. In this experiment, sGP was added directly to cells on a tissue culture plate at 37°C on a heating stage. Thus, sGP appears to intercalate into preformed caps to reduce RET. To ensure the specificity of this effect, we also treated cells with the saccharide NADG, which has been previously shown to reduce CR3-FcγRIIIB cocapping (5). NADG treatment substantially decreased CR3-FcγRIIIB cocapping without affecting the ability of CR3 to cap on cells or the amount of tagged mAb on cells (data not shown). Thus, sGP cannot dislodge FcγRIIIB from CR3 caps, although it reduces receptor proximity. However, NADG can dislodge FcγRIIIB from CR3 caps in both the presence and absence of sGP (Table II).

Early studies have demonstrated that FcγRIIIB, a GPI-linked membrane protein, could mediate transmembrane signaling and physiological functions (18, 19). This paradoxical ability of FcγRIIIB can be explained, at least in part, by its ability to physically interact with CR3 (5, 6, 7, 8), which apparently is linked to the functional activity of FcγRIIIB (9). Indeed, several additional neutrophil membrane proteins including the urokinase receptor (CD87), CD14, and FcγRII (CD32) apparently interact with CR3 (15, 20, 21, 22, 23). These interactions have been observed by a variety of techniques including cocapping, RET, lateral diffusion, immunoprecipitation, gene transfection, and gene knock-out experiments (5, 6, 7, 8, 9, 15, 16, 20, 21, 22, 23). Clinical aberrations in these interactions have also been observed in neutrophils from certain pyoderma gangrenosum patients (24, 25). These studies suggest a potential mechanism that accounts for transmembrane signaling by certain GPI-anchored receptors. To our knowledge, the present studies provide the first report of a product of an infectious agent that affects receptor interactions on plasma membranes, presumably to blunt host responses.

Exposure of cells to buffer alone or to supernatants from control cultures had no effect on RET between FcγRIIIB and CR3. The presence of energy transfer under control conditions was confirmed by optical imaging, quantitative microfluorometry, and single-cell imaging spectrophotometry. Both the acquisition of acceptor fluorescence emission and quenching of donor fluorescence emission were noted (thus demonstrating that the donor lost energy while the acceptor gained energy). Because RET is strongly distance-dependent, the labels attached to FcγRIIIB and CR3 must be within ∼7 nm of one another (approximately molecular proximity). The intensity of RET is also dependent upon the relative numbers of donor and acceptor chromophores. Using an imaging spectrophotometry system, we have confirmed the dependence of RET on the relative number of donor and acceptor labels in this system. Therefore, as we have previously reported (5, 6, 7), FcγRIIIB and CR3 are in close physical proximity on human neutrophils.

We have found that the sGP of Ebola virus dramatically reduces FcγRIIIB-to-CR3 RET on living human neutrophils. This reduction was inferred by studies employing RET imaging, quantitative measurements, and spectrophotometry of single cells. The decrease in RET was found to be dependent upon the concentration of sGP, as illustrated by Fig. 2. One potential mechanism for the reduction in FcγRIIIB-to-CR3 proximity is the dissociation of FcγRIIIB-CR3 complexes, as we have previously noted in studies of saccharides and polysaccharides (e.g., Ref. 9). However, this is apparently not the case for sGP. Surprisingly, the fluorescence staining patterns demonstrated colocalization and cocapping of FcγRIIIB and CR3 in the presence of sGP (Figs. 1 and 5; Table II). A simple explanation consistent with these observations is that sGP may intercalate between FcγRIIIB and CR3, thereby separating the labels by a greater distance and reducing RET, without causing dissociation of FcγRIIIB from CR3. Alternatively, sGP could induce an extraordinary conformational change in the receptors causing the orientation and/or distance between the chromophores to be altered, or sGP could bind to FcγRIIIB in such a fashion that the line of sight between the labels is interrupted, thereby blocking RET. Thus, several lines of evidence support the hypothesis that sGP induces a physical change in the nature of FcγRIIIB-to-CR3 interactions in cell membranes.

Although a simple sGP intercalation model may be a considerable advance in our understanding of sGP’s effect on neutrophils, it is only a partial explanation. Although previous studies have shown that anti-FcγRIIIB clone 3G8 blocks sGP binding to cells (4), FcγRIIIB apparently is a necessary but insufficient condition because transfectants expressing CR3 and FcγRIIIB were unable to bind sGP (our unpublished observations). We speculate that another unidentified component of these CR3-containing membrane complexes participates in sGP-to-cell binding. Nonetheless, sGP binding diminishes cell function. Immune complexes, which activate neutrophil metabolic flux (26), are unable to do so in the presence of sGP, but can in the presence of control supernatants (our unpublished observations). Similarly, the percentage of neutrophils binding or internalizing IgG-opsonized SRBCs is reduced from 51 ± 15% in the presence of Fab fragments of anti-FcRII (clone IV.3) to 8 ± 8% in the presence of both anti-FcRII and sGP (data not shown). Thus, whatever the sufficient conditions are for sGP binding to cells might be, FcγRIIIB-mediated functions and its normal physical interactions with CR3 are disrupted.

We suggest that a step in the pathogenic effects of Ebola virus infections involves the disruption of normal FcγRIIIB-to-CR3 interactions by sGP. However, this disruption may not be limited to FcγRIIIB-mediated cell activation. Our findings raise the interesting possibility that rationally designed compounds directed at exodomain receptor-integrin interactions may be useful as anti-inflammatory drugs.