Background. sGP, which is perceived as nonstructural, secretory glycoprotein, shares 295 amino acids at its N-terminal with GP1,2, which include the specific residue necessary to interact with GP2. In the present study, we tested whether the sGP protein of Zaire ebolavirus (ZEBOV) could substitute for GP1 and form a complex with GP2, thus serving as a structural protein.
Methods. We expressed ZEBOV GP1,2, VP40, and NP proteins, together with sGP protein, from expression plasmids and examined the resultant virus-like particles by using Western blot. Cells expressing GP2 in combination with either GP1 or sGP were analyzed by using flow cytometry with the KZ52 antibody, which recognizes a GP1,2 conformational epitope. A VSV pseudotype, VSVΔG*, which expresses a GFP reporter gene instead of the G protein, was used to produce pseudotyped viruses encoding sGP and variants of GP to test the contribution of sGP to infectivity.
Results. Western blot and flow cytometric analyses suggested the existence of a covalently linked sGP-GP2 molecule. VSVΔG*(sGP + GP2) and VSVΔG*(GP1,2) infected Vero E6 cells and were neutralized by the KZ52 antibody. Overexpression of sGP reduced the titer of VSVΔG*(GP1,2).
Conclusions. ZEBOV sGP can substitute for GP1, forming a sGP-GP2 complex and conferring infectivity. Our studies suggest a novel role for sGP as a structural protein.
Ebola virus (EBOV) is a filamentous, enveloped, nonsegmented, negative-strand RNA virus, which, with Marburg virus, constitutes the family Filoviridae. The fourth gene of the EBOV genome encodes 2 glycoproteins: the secretory glycoprotein (sGP), encoded by the predominantly unedited transcript, and the virion envelope glycoprotein (GP), the product of RNA editing during formation of the mRNA . The molecular ratio of unedited to edited mRNA is ∼3:1 (73% vs 27%) , indicating that sGP is produced more abundantly than is GP in the life cycle of EBOV.
Pre-sGP (60 kDa), the translation product of the unedited mRNA from the fourth gene, is cleaved by furin into secretory glycoprotein (sGP; 50 kDa) and a delta peptide (Δ-peptide; 10–14 kDa) [2, 3]. The sGP consists of a homodimer joined in a parallel manner by 2 inter-subunit disulfide bonds at paired N-terminal and C-terminal cysteines (C53–C53’ and C306–C306’) [4–6]. Although the N-terminal 295 amino acids of sGP are identical to those of the envelope GP, the C-terminal 69 amino acids are unique to sGP [1, 3, 6]. The sGP homodimer is structurally and functionally distinct from the envelope GP . The large amount of sGP produced by EBOV may contribute to some infection processes in its natural hosts but may also contribute to pathogenesis in nonnatural hosts .
For example, sGP may serve as a decoy ( houkutin) for EBOV-specific neutralizing and nonneutralizing antibodies [3, 7]. sGP, but not the Δ-peptide, induces the partial recovery of endothelial cell barrier function after treatment with TNF-α . sGP is also thought to interact with and inactivate human neutrophils via the IgG Fc receptor IIIb (FcγRIIIb, CD16b) [9, 10], a concept that has been challenged by some researchers .
The Δ-peptide is secreted in monomeric form ; its function is unknown.
GP is a transmembrane protein that facilitates viral receptor binding and fusion with target cells [12–14]. It comprises GP1 (140 kDa) and GP2 (25 kDa), which are proteolytically processed from the GP0 polypeptide, disulfide linked, and form the mature GP1,2 complex [2, 15]. Mutational and crystal structure analyses revealed that receptor-binding, protein stability-associated, and GP2-interacting regions exist in the N-terminal 312 amino acids of GP1 [16–20]. After cleavage by furin in the Golgi, the GP1-GP2 interaction becomes unstable and some GP1 is released as a soluble monomer .
The KZ52 antibody, a human monoclonal antibody established from a patient , bridges amino acids 42-43 of GP1 and 505-514 and 549-556 of GP2  but does not recognize sGP. However, in the literature , immunoprecipitants with the antibody contained an additional molecule of 50–70 kDa that is unlikely to be GP1 or GP2.
On the basis of these reports, we hypothesized that sGP may substitute for GP1 and form an sGP-GP2 molecule on the virions. To test this hypothesis, we expressed Zaire EBOV (ZEBOV) proteins from plasmids and analyzed them by Western blot, flow cytometry, and a VSVΔG* pseudotype system. We also used pre-sGP, which is cleaved into sGP and the Δ-peptide [1, 3], to investigate the influence of the Δ-peptide on the formation of sGP-GP2.
MATERIALS AND METHODS
Cell Lines and Antibodies
Human embryonic kidney 293T cells and African green monkey kidney cells (Vero E6) were cultured in high-glucose Dulbecco’s modified Eagle's medium (SIGMA) supplemented with 10% fetal calf serum and penicillin-streptomycin. The cells were grown in an incubator at 37°C under 5% carbon dioxide.
An affinity-purified anti-FLAG M2 monoclonal antibody, monoclonal anti-HA clone HA-7 mouse ascites fluid, and an anti-c-myc rabbit antibody were obtained from SIGMA. The KZ52 human IgG antibody was kindly provided by Dr. Dennis Burton. An anti-sGP antibody was produced by immunization of rabbits with peptide CLSQLYQTEPKTSVVRVRR, an amino acid sequence unique to sGP. Anti-mouse Alexa Fluor 488 IgG (Invitrogen) and fluorescein-conjugated anti-human IgG F(c) (Rockland) were used for flow cytometry. An anti-GP monoclonal antibody (12/1.1) that recognizes the C-terminus of GP1 , mouse TrueBlotTM ULTRA: horseradish peroxidase anti-mouse IgG (eBioscience), and the universal antibody in the VECTAStain ABC kit (Vector Laboratories) were used for Western blotting.
To express ZEBOV proteins, cDNAs were cloned into the expression plasmid pCAGGS/MCS [24, 25]. Untranslated regions of the ZEBOV genes were deleted to minimize unexpected effects, such as mRNA splicing or the regulation of translation by these regions. Plasmids for full-length GP1,2, sGP, and pre-sGP without tags were designated pCGP, pCsGP, and pCpresGP, respectively. For GP2 expression, the nucleotide sequence of the signal peptide of the murine Ig κ chain (METDTLLLWVLLLWVPGSTGE) was inserted at the 5′ end of the GP2 cDNA (pCGP2). In addition, tagged plasmids were prepared as follows. A FLAG (DYKDDDDK)-tag was inserted between residues Ser (32) and Ile (33) of the GP cDNA of pCGP to generate FLAG-tagged GP1,2 (pCGP-FLAG). To generate a FLAG/myc-tagged GP expressing plasmid (pCGP-FLAG/myc), a myc (EQKLISEEDL) tag was added to the N-terminus of the GP2 sequence in pCGP-FLAG. FLAG-tagged GP1 was prepared by replacing the C-terminal 5 amino acids (RRTRR) of GP1 with a spacer sequence encoding TLE and by adding the FLAG-tag (pCGP1-FLAG). The influenza virus hemagglutinin epitope (HA; YPYDVPDYA)-tagged sGP (pCsGP-HA) and presGP (pCpresGP-HA) were prepared by inserting the HA tag immediately downstream of the signal sequences in pCsGP and pCpresGP, respectively.
VLP Formation and Purification
To generate Ebola virus like particles (VLPs) with or without tags, we transfected 2.4 x 107 293T cells with plasmids for the expression of GP (pCGP-FLAG/myc or pCGP; 15 μg), VP40 (pCVP40; 15 μg), and NP (pCNP: 7.5 μg) by using TransIT-293 (Takara Bio) [26–29]. In addition, to assess the effect of sGP or presGP on the glycoprotein composition of VLPs, an empty plasmid (pCAGGS/MCS; 45 μg) or plasmids for sGP (pCsGP-HA or pCsGP; 45 μg) or presGP (pCpresGP-HA or pCpresGP; 45 μg) were cotransfected with the plasmids for the formation of the VLPs as indicated above. The 1:3 ratio of GP plasmids to empty, sGP, or presGP plasmids was based on the ratio of the GP mRNA transcripts to the sGP mRNA transcripts . At 48 hours after transfection, the culture medium was harvested and filtered through a 0.45 μm PVDF filter (Millipore). The filtrate was then centrifuged through a 20% sucrose cushion in phosphate-buffered saline (PBS) for 2 hours at 27000 rpm (130000 g) to isolate VLPs. VLPs were resuspended in PBS and centrifuged again for 2 hours at 27000 rpm. The VLPs were then resuspended in Tris-Glycine sodium dodecyl sulfate sample buffer (Invitrogen), and protein composition was analyzed using Western blotting.
We transfected 293T cells with plasmids to investigate the expression of glycoproteins on the cell surface. The plasmids transfected were pCAGGS/MCS, pCGP-FLAG, pCsGP-HA, pCpresGP-HA, pCGP2, pCsGP-HA + pCGP2, pCpresGP-HA + pCGP2, pCGP1-FLAG,, and pCGP1-FLAG + pCGP2. Cells were pelleted at 4°C for 2 minutes at 4000 rpm, then resuspended in wash buffer (PBS with 1% fetal calf serum and 0.05% sodium azide) and aliquoted for staining. Glycoproteins were detected by staining cells with the KZ52 human antibody  and exposing them to a fluorescence-conjugated anti-human IgG F(c) antibody. Cells were also stained with an anti-HA antibody and proteins detected with an anti-mouse Alexa Fluor 488 antibody. All staining was performed on ice for 30 minutes, followed by washing. For each sample, 10000 events in the live cell gates were analyzed. Data were collected on a Becton Dickinson FACSCalibur and analyzed using CellQuest software.
Vesicular Stomatitis Virus (VSVΔG*) Pseudotyped With Ebola Virus Glycoprotein
Vesicular stomatitis virus, VSVΔG*, which expresses a GFP reporter gene instead of the G protein, was used to produce VSVΔG* pseudotyped with a different glycoprotein, as described elsewhere . 293T cells were transfected with pCAGGS/MCS, pCGP-FLAG, pCGP1-FLAG and pCGP2, pCsGP-HA and pCGP2, or pCpresGP-HA and pCGP2 and then used for the preparation of VSVΔG* pseudotypes. Twenty-four hours after transfection, cells were infected with VSVΔG*G and after 24 hours incubation, culture fluid was collected, centrifuged, and filtered through a 0.45-μm filter to remove cells and cell debris. The VSVΔG* pseudotyped viruses were stored at -80°C until use. VSVΔG* pseudotyped with EBOV glycoproteins were incubated with a neutralizing anti-VSVG antibody I1 for 30 minutes at room temperature to eliminate the effect of VSVG (used to prepare the initial pseudotype viruses) on the infectivity of the VSVΔG* pseudotypes. The titers of the VSVΔG* pseudotypes were determined in Vero E6 in triplicate.
Ebola Virus GP Complexes on Purified VLPs
VLPs, purified by ultracentrifugation, were analyzed using Western blot, in which GP1 and GP2 were tagged with FLAG- and myc-epitopes, respectively. When VLPs were prepared by expressing pCGP-FLAG/myc and pCAGGS/MCS, pCGP-FLAG/myc and pCsGP-HA, or pCGP-FLAG/myc and pCpresGP-HA, under non-reducing conditions, the anti-FLAG and anti-myc antibodies detected 150 kDa bands, corresponding to the disulphide-bonded GP1 and GP2 complex (Figure 1A, lanes 1–6), and under reducing conditions, the anti-myc antibody detected the 25 kDa GP2 band (Figure 1A, lanes 7–9). Under nonreducing conditions, the anti-HA and anti-myc antibodies detected molecules of ∼75 kDa (Figure 1A), which did not correspond to monomers of GP1 (140 kDa), GP2 (25 kDa), sGP (50 kDa), or pre-sGP (60 kDa) (although the intensity of the band is weak for pre-sGP) (Figure 1A). The anti-HA antibody also detected a molecule of ∼100 kDa, although the band intensity was weaker for pre-sGP than for sGP (Figure 1A). Because sGP is produced as a soluble homodimer [4, 6, 30], this result suggests that the sGP dimer does not covalently associate with VLPs. Under reducing conditions, the molecules at 75 kDa and 100 kDa were not observed (data not shown).
Similar experiments were performed with untagged EBOV proteins and antibodies against GP1 and sGP. GP1,2 was again observed under nonreducing conditions with the anti-GP1 antibody (Figure 1B). The anti-sGP antibody detected 75- and 100-kDa proteins (Figure 1B), although the intensity of the band is weak for pre-sGP (Figure 1B). These results indicate that the use of epitope tags did not cause artifacts, and sGP was present on the purified VLPs as 75- and 100-kDa proteins. The molecules at 75 kDa and 100 kDa were not detected under reducing conditions (data not shown).
Cell Surface Expression of EBOV Glycoproteins
The reactivities of EBOV glycoproteins expressed from plasmids on 293T cells were analyzed using flow cytometry with the human KZ52 antibody that recognizes an epitope that spans the GP1-GP2 subunits of GP . The reactivity of full-length GP1,2 (pCGP-FLAG)-expressing cells is shown in Figure 2A. Neither GP1 (pCGP1-FLAG) nor GP2 (pCGP2) reacted with this antibody (Figure 2B and C). However, cells that coexpressed GP1 and GP2 from separate plasmids reacted, indicating the formation of the epitope recognized by KZ52 (Figure 2D). Although cells expressing either sGP (pCsGP-HA) or pre-sGP (pCpresGP-HA) alone did not show any significant reactivity (Figures 2E and 2F), cells coexpressing sGP and GP2 (pCsGP-HA and pCGP2) or pre-sGP and GP2 (pCpresGP-HA and pCGP2) reacted with the KZ52 antibody (Figures 2G and 2H). These results indicate that coexpression of the N-terminal 295 amino acids of GP (the region common to both full-length GP and sGP) and GP2 formed the KZ52 epitope.
The reactivities of Ebola virus glycoproteins to an anti-HA antibody were also analyzed. Cells expressing HA-tagged sGP (pCsGP-HA) or pre-sGP (pCpresGP-HA) alone reacted with the anti-HA antibody (Figures 2I and 2J), indicating that, although the soluble sGP homodimer was produced in the culture supernatant [4, 6, 30], a part of sGP was retained on the surface of the sGP-producing cells by an unknown mechanism. Coexpression of GP2 without the HA tag increased the reactivity with the anti-HA antibody, suggesting that GP2 could increase sGP retention on the cell surface (Figures 2K and 2L).
sGP Contribution to EBOV Infectivity and Neutralization by the KZ52 Antibody
To examine whether sGP contributes to infectivity, 293T cells were transfected with EBOV glycoprotein expression plasmids and VSVΔG* pseudotypes were prepared. Plasmids used for controls and for expression of Ebola virus glycoproteins were as follows: pCAGGS/MCS, pCGP-FLAG, pCGP1-FLAG, pCsGP-HA, pCpresGP-HA, pCGP2, pCGP1-FLAG + pCGP2, pCsGP-HA + pCGP2, and pCpresGP-HA + pCGP2. As shown in Figure 3, VSVΔG* pseudotyped with full-length GP1,2 (pCGP-FLAG) exhibited 4.14 x 106 IU/mL, whereas the titer for the mock-pseudotype with pCAGGS/MCS was below the detectable level (detection limit 6.67 IU/mL). Expression of GP1 (pCGP1-FLAG), sGP (pCsGP-HA), pre-sGP (pCpresGP-HA), or GP2 (pCGP2) alone resulted in no detectable infectivity. In contrast, VSVΔG* prepared with coexpression of GP2 and GP1 exhibited infectivity of 8.05 x 105 IU/ml. Furthermore, coexpression of GP2 and either sGP or pre-sGP also resulted in infectious titers of 1.40 x 104 and 9.33 x 102 IU/mL, respectively (Figure 3). Because the KZ52 antibody recognizes a GP1,2 conformation-dependent epitope and neutralizes EBOV infectivity [19, 31], we then asked whether this antibody neutralizes the infectivities of VSVΔG* bearing GP-FLAG VSVΔG*(GP-FLAG), sGP-HA plus GP2 VSVΔG*(sGP-HA + GP2), or presGP-HA plus GP2 VSVΔG*(presGP-HA + GP2). These VSVΔG* pseudotypes were preincubated with the KZ52 antibody for 30 minutes at the indicated concentrations and inoculated on Vero E6 cells. As shown in Figure 4, a similar pattern of dose-dependent neutralization by the KZ52 antibody was observed for all of the VSVΔG* pseudotyped viruses.
Effect of sGP on the Infectivity of VSVΔG* Bearing Full-Length GP1,2
To examine the effect of overexpression of sGP or pre-sGP on the infectivity of VSVΔG* pseudotyped with full-length GP, pCAGGS/MCS, pCsGP, or pCpresGP was cotransfected with the plasmids needed to produce VSVΔG* pseudotyped with full-length GP-FLAG, and titers were determined. Coexpression of sGP-HA or presGP-HA resulted in a reduction in the titer of VSVΔG* pseudotyped with full-length GP by 0.5 or 1.1 log units (Figure 5).
Western blot analyses suggested that a molecule of ∼75 kDa that associated with VLPs was likely to be disulphide bond-linked sGP and GP2, because the molecule reacted with anti-tag antibodies for which the epitopes, HA and myc, were fused to sGP and GP2, respectively, and the anti-sGP antibody detected the 75-kDa protein from untagged VLP (Figure 1). Moreover, the use of a reducing reagent resulted in the dissociation of the 75-kDa molecule into 50-kDa and 25-kDa bands, corresponding to sGP and GP2, respectively (Figure 1). The KZ52 antibody showed reactivity to cells expressing both sGP and GP2 in flow cytometry and neutralizing activity against VSVΔG*(sGP-HA + GP2) and VSVΔG*(presGP-HA + GP2) (Figures 2–4), strongly suggesting that sGP and GP2 form a complex that confers infectivity and is structurally similar to full-length GP1,2. The N-terminal 295 amino acids of sGP and GP1,2 are identical and include the Cys53 residue that is necessary to form the disulphide bond with Cys609 of GP2 [1, 15]. The N-terminal 295 amino acids region contains the Cys53 residue as a part of a base subdomain, being in contact with GP2 , a receptor-binding site [16–18, 20] and a putative protein-stability region (amino acids 214 to 270) [16, 18]. The rest of GP1 includes the last β18 strand of the glycan cap and a mucin like domain, both of which do not interact with GP2 and can be deleted without affecting infectivity to VSVΔG* or lenti/retroviral vectors [15, 16, 19, 32, 33]. Therefore, sGP likely substitutes for the functional domains in GP1 that are necessary for the ability to pseudotype VSVΔG*.
We speculate that the sGP-GP2 complex may be synthesized as follows: sGP takes the place of GP1 and forms a disulphide bond between Cys53 of sGP and Cys609 of GP2 in the endoplasmic reticulum before GP1 and GP2 can form a disulphide bond. Formation of the disulphide bond is probably more efficient between GP1 and GP2 than between sGP and GP2, because the former pair exists on one polypeptide ; however, the abundance of sGP [1, 34] may make the sGP-GP2 complex possible. GP1 is then removed from sGP-GP0 by furin in the Golgi apparatus, resulting in the release of the GP1 monomer from EBOV-infected cells [21, 22]. Thus, monomeric sGP may be linked to GP2, and the heterodimer may form a heterotrimer like GP1,2 peplomer.
The infectivity of VSVΔG*(presGP-GP2) was lower than that of VSVΔG*sGP-GP2 (Figure 3), suggesting that the Δ-peptide may inhibit the formation of sGP-GP2. Further investigation is necessary to determine the function of the Δ-peptide.
The reduced infectivity of VSVΔG* pseudotyped with GP by sGP overexpression (Figure 5) suggests that sGP may interfere with the infectivity of Ebola virus. The sGP-GP2 molecule likely associates with GP1,2 molecules in the peplomer. Consequently, this heterotrimeric peplomer may result in lower infectivity. However, the net effect of sGP on entry may be relatively small. In addition, by replacing GP1, which has a cytotoxic mucin like domain, with sGP, the cytotoxicity (CPE) of sGP-GP2 would be reduced because sGP lacks this cytotoxic region. Previously, it was reported that an EBOV mutant whose genome encodes GP1,2, but not sGP, showed significantly increased cytotoxicity, compared with authentic EBOV, which encodes GP1,2 and sGP . Our results may suggest a new role for sGP in the life cycle of EBOV.
This work was supported by Grants-in-Aid for Specially Promoted Research and for Scientific Research; a Contract Research Fund for the Program of Founding Research Centers for Emerging and Reemerging Infectious Diseases; Exploratory Research for Advanced Technology; the Special Coordination Funds for Promoting Science and Technology from the Ministry of Education, Culture, Sports, Science, and Technology of Japan; the Region V “Great Lakes” Regional Center of Excellence; and National Institute of Allergy and Infectious Diseases Public Health Service research grants.