Sorting of Marburg Virus Surface Protein and Virus Release Take Place at Opposite Surfaces of Infected Polarized Epithelial Cells

Abstract

Marburg virus, a filovirus, causes severe hemorrhagic fever with hitherto poorly understood molecular pathogenesis. We have investigated here the vectorial transport of the surface protein GP of Marburg virus in polarized epithelial cells. To this end, we established an MDCKII cell line that was able to express GP permanently (MDCK-GP). The functional integrity of GP expressed in these cells was analyzed using vesicular stomatitis virus pseudotypes. Further experiments revealed that GP is transported in MDCK-GP cells mainly to the apical membrane and is released exclusively into the culture medium facing the apical membrane. When MDCKII cells were infected with Marburg virus, the majority of GP was also transported to the apical membrane, suggesting that the protein contains an autonomous apical transport signal. Release of infectious progeny virions, however, took place exclusively at the basolateral membrane of the cells. Thus, vectorial budding of Marburg virus is presumably determined by factors other than the surface protein.

Marburg virus (MBGV) and the closely related Ebola virus (EBOV) make up the family of Filoviridae, which, together with the Paramyxoviridae, Rhabdoviridae, and Bornaviridae, constitute the order Mononegavirales. Filoviruses are highly pathogenic for humans and nonhuman primates, causing a severe hemorrhagic fever with fatality rates of up to 90% in the case of EBOV infection (for reviews see references 24 and 29). The emerging potential of filoviruses is underlined by several outbreaks during the last five years (52, 53). Several attempts have been made to detect the natural reservoir of filoviruses with no success (5, 19). The prototype of Filoviridae, MBGV, was isolated in 1967, when several laboratory workers were infected after contact with imported monkeys (39).

The enveloped MBGV particles are composed of seven structural proteins and the RNA genome. Four viral proteins are the components of the nucleocapsid: the nucleoprotein NP (3, 17, 22, 34), the L protein (27), P (formerly called VP35) (26), and the viral protein VP30 (26). VP40 and VP24 are located between the envelope and the nucleocapsid and probably represent matrix proteins (3). The envelope of MBGV is decorated by the only surface protein, GP, which is inserted into the viral membrane as a homotrimer (8). Since GP is the only membrane protein of MBGV, it is assumed to be responsible for virus entry into host cells (4) and to be the major target for the immune response of the infected organism.

The GP gene (2,844 nucleotides) encodes a protein of 681 amino acids (51). In contrast to EBOV, where the virion-associated surface protein is only expressed after mRNA editing (35, 46), MBGV GP is encoded by a single open reading frame. Two hydrophobic regions have been identified in GP, one at the amino terminus, and the other in the carboxy-terminal region. The N-terminal hydrophobic region is not present in the mature protein, indicating that this region serves as signal peptide (51). The carboxy-terminal hydrophobic domain is used as a membrane anchor, which is adjacent to the short cytoplasmic tail composed of the last eight amino acids of the protein. Two fatty acid attachment sites were identified at the boundary between membrane anchor and cytoplasmic domain (11). GP is heavily N- and O-glycosylated, containing 19 potential N-linked glycosylation sites and several clusters of hydroxyamino acids which serve as O-linked glycosylation sites (13).

It was demonstrated earlier that surface transport of GP involves sequential steps of maturation (2, 8, 11). As a late step of maturation, GP is cleaved by the prohormone convertase furin in the trans-Golgi network (48, 50). The precursor GP_1/2 (220 kDa) gives rise to two fragments, GP₁ (170 kDa) and GP₂ (50 kDa), which are connected by disulfide linkage(s).

Vectorial budding is an important biological feature of viruses which has significant impact on the course of disease (43). Most viral membrane proteins are transported to the membrane compartment, where release of the virus takes place. To investigate whether GP undergoes vectorial transport in polarized epithelial cells, we established a Madin-Darby canine kidney cell line constitutively expressing GP (MDCK-GP). The functionality of the recombinant GP in MDCK-GP cells was verified by recombinant vesicular stomatitis virus (VSV) pseudotypes. We examined several transport-related parameters and provide evidence that the great majority of GP molecules are transported to the apical surface of polarized MDCK-GP cells. Furthermore, GP was released exclusively in the apical supernatant of MDCK-GP cells. When MDCKII cells were infected with MBGV, GP was also transported to the apical side. Interestingly, budding of MBGV takes place exclusively at the basolateral surface.

MATERIALS AND METHODS

Viruses and cell lines.

E6 cells, a cloned cell line of Vero cells (ATCC CRL 1586), and Vero cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS) and antibiotics at 37°C in an atmosphere of 5% CO₂. Madin-Darby canine kidney (MDCKII) cells were cultured in minimal essential medium (MEM) supplemented with 10% FCS under the same conditions. MDCK-GP cells, constitutively expressing GP, were cultured like MDCKII cells in the presence of geneticin (1 mg/ml) (Sigma).

For vectorial transport studies, cells were grown on permeable membrane filters to confluence. The transepithelial resistance 24 to 48 h after seeding the cells exceeded 300 or 600 Ω/cm², depending on the filter size (see below). Baby hamster kidney (BHK) cells were cultured in DMEM containing 10% FCS. Tetracycline-regulated BHK cells expressing the VSV G protein (37) were cultured in DMEM containing 10% FCS and tetracycline (0.5 μg/ml). The Musoke strain of MBGV isolated 1980 in Kenya (41) was propagated in E6 cells as described previously (11). VSV Indiana was propagated in BHK cells. The VSV recombinant VSVΔG-GFP/G, containing the gene for the green fluorescent protein (GFP) instead of the gene for G (kindly provided by Ralf Wagner, Marburg), was propagated in the BHK-G cell line stably expressing VSV G protein (37).

Infection of cells. (i) MBGV.

MDCKII cells were trypsinized, and cell density was determined. Approximately 10⁵ (diameter of the filter membrane, 6.5 mm) or 10⁶ cells (diameter of the filter membrane, 2.4 cm) were infected at a multiplicity of infection (MOI) of 1 PFU of MBGV per cell in suspension. During incubation, cells were allowed to settle on permeable membrane filters. At 2 h postinfection (p.i.), the inoculum was removed, and cells were washed three times with MEM and further incubated for 48 h with MEM supplemented with 10% FCS. At 24 h p.i., the measured values for transepithelial resistance were above 300 Ω/cm² (diameter of the filter membrane, 2.4 cm) or 600 Ω/cm² (diameter of the filter membrane, 6.5 mm). Infected MDCKII cells were used for indirect immunofluorescence at 48 h p.i. and for surface biotinylation at 36 and 48 h p.i.

(ii) MBGV/215.

MBGV/215 represents an MBGV stock containing in addition to the authentic MBGV an artificial MBGV-specific minigenome with a chloramphenicol acetyltransferase (CAT) gene which was previously shown to be passaged together with the full-length genome (26). MBGV/215 was used to infect MDCKII cells in suspension as described for MBGV. At 12, 24, 42, and 48 h p.i., 500-μl aliquots of the apical and basolateral supernatant were removed and used to infect Vero cells. At 45 min p.i., 2 ml of DMEM containing 2% FCS was added, and the cells were further incubated for 48 h. Then the cells were washed once with phosphate-buffered saline (PBS), lysed, and assayed for CAT activity.

(iii) VSV.

The basolateral membrane of polarized MDCKII cells grown on permeable filter membranes was infected at an MOI of 1 PFU per cell for 1 h at 37°C. After removal of the inoculum, cells were further incubated for 4 h with MEM. Finally, cells were fixed with 4% paraformaldehyde (PFA) and processed for immunofluorescence analysis.

CAT assay.

CAT activity was determined using 50 nCi of [¹⁴C]chloramphenicol (Amersham Buchler) per sample in a standard assay (14). Lysates corresponding to 5 × 10⁵ MBGV/215-infected Vero cells were used. Quantification of radioactivity was done with the Bio-Imaging Analyzer (Fuji BAS-1000) using the Raytest TINA software.

Establishment of a GP-expressing MDCKII cell line (MDCK-GP).

The MBGV GP gene was cut out from plasmid pSP72-GP (kindly provided by Ute Ströher, Marburg) with HindIII. Then, the HindIII site was filled with Klenow enzyme and subsequently cut with EcoRI. The resulting fragment was ligated into EcoRI- and SmaI-digested plasmid pSG5/new (kindly provided by Wolfram Schäfer, Marburg). The resulting plasmid was designated pSG5-GP. MDCKII cells grown on 10-cm petri dishes (40% confluent) were cotransfected with 40 μg of psG5-GP and 4 μg of pIG1 (kindly provided by Wolfram Schäfer, Marburg) using the Lipofectin method of Felgner et al. (9). The vector pIG1 conferred resistance to geneticin. At 24 h after transfection, geneticin (1 mg/ml) was added to the culture medium to select stably transfected cells. At 6 days after transfection, cell clones were isolated and screened for expression of MBGV GP by indirect immunofluorescence microscopy. The finally selected GP-expressing cell line was designated MDCK-GP.

Pulse-chase and immunoprecipitation analysis.

MDCKII and MDCK-GP cells were starved for 1 h with methionine- and cysteine-deficient DMEM and thereafter labeled for 40 min with 100 μCi of [³⁵S]Promix (Amersham Pharmacia). Labeling medium was removed, and cells were chased in the presence of normal DMEM. At the indicated time, supernatants were saved and cells were washed with ice-cold PBS and lysed with BEP (50 mM Tris-HCl [pH 8], 100 mM NaCl, 20 mM CaCl₂ · 2H₂O, 20 mM MgCl₂, 2% glycerin, 1% NP-40, 0.5% Tween 20, 5% [vol/vol] Trasylol [Bayer], 1 mM phenylmethylsulfonyl fluoride [PMSF]). Cell lysates were subsequently sonicated and cleared by centrifugation. The cell lysates and the chase supernatants were incubated for 1 h at 4°C with protein A-Sepharose, which was then removed by centrifugation. Samples were diluted with 1 volume of TNE (10 mM Tris-HCl [pH 7.4], 0.15 M NaCl, 2 mM EDTA) and further incubated for 2 h at 4°C with a 1:100 dilution of an anti-GP rabbit serum. Complexes of GP and antibody were incubated for 3 h with 40 μl of protein A-Sepharose and sedimented at 14,000 rpm for 1 min. Pellets were washed three times with BEP, resuspended in 20 μl of denaturing buffer (New England Biolabs), and boiled for 10 min. The samples were divided in two fractions and supplied with 2 μl of G5 buffer (New England Biolabs). One of the fractions was treated with 2 μl of endoglycosidase H (EndoH) for 2 h at 37°C. Finally, 2.5 μl of sample buffer (40% glycerin, 12% sodium dodecyl sulfate [SDS], 750 mM Tris-HCl [pH 6.8], 20% mercaptoethanol, 5% saturated bromphenol blue solution) was added. Samples were heated for 5 min to 95°C and subjected to SDS-polyacrylamide gel electrophoresis (PAGE) and autoradiography.

Indirect immunofluorescence analysis of MDCK-GP and MDCKII cells.

Approximately 10⁵ MDCK-GP or MDCKII cells were seeded on glass coverslips (1.2 cm²) and grown for 48 h. For surface immunofluorescence analysis, cells were washed two times with ice-cold PBS and fixed for 5 min with 3% PFA at room temperature (RT). Cells were then rinsed two times with PBS and incubated with 0.1 M glycine for 10 min at RT. Thereafter, samples were washed once with PBS and incubated for 1 h at RT with a rabbit anti-GP antiserum which was diluted 1:100 in PBS–3% bovine serum albumin (BSA). Subsequently, cells were washed twice with PBS and incubated for 1 h with a fluorescein isothiocyanate (FITC)-conjugated donkey anti-rabbit immunoglobulin (Ig) antiserum (Dako; 1:100 dilution in PBS–3% BSA). Finally, the coverslips were washed twice with PBS, dipped once into H₂O, and mounted with Fluoprep (BioMerieux). Microscopic analysis was performed using an Axiomat fluorescence microscope (Zeiss).

For intracellular staining, the same protocol was used except that cells were permeabilized after fixation using PBS containing 0.2% Triton X-100 for 5 min at RT. A monoclonal anti-GP antibody was used as the primary antibody at a dilution of 1:2 in PBS–3% BSA. Bound antibodies were detected using a Texas Red-coupled goat anti-mouse IgG diluted 1:200 in PBS–3% BSA (Calbiochem).

Indirect immunofluorescence analysis of filter-grown MDCKII cells and laser scanning analysis.

Polarized MDCKII and MDCK-GP cells were grown on membrane filters (Transwell, Corning Costar; 6.5 mm diameter; 0.4 μm pore size). Immunofluorescence analysis was performed essentially as described above.

Antibody incubation. (i) MDCK-GP and MBGV-infected MDCKII cells.

The apical and basolateral membranes of the cells were incubated with a rabbit anti-GP antiserum (diluted 1:50 in PBS–3% BSA) for 90 min at RT. As the secondary antibody, the FITC-conjugated donkey anti-rabbit Ig antiserum was employed as described above. Finally, the membranes were cut out from the filters and mounted with Fluoprep. Samples were analyzed with a confocal laser microscope (LS410; Zeiss) using Z-Scan analysis.

(ii) VSV-infected MDCKII cells.

A monoclonal anti-G antibody (kindly provided by Michael Ross, Dallas, Tex.) was used at a dilution of 1:4 in PBS containing 3% BSA. Bound antibodies were detected as described above using a Texas Red-conjugated goat anti-mouse IgG.

Domain-selective surface biotin labeling.

MDCK-GP and MDCKII cells were grown on membrane filters (Falcon; Becton Dickinson; 2.5-cm diameter, 1.0-μm pore size). Polarized cells on filters were washed twice with ice-cold PBS, and each side of the filter membranes was incubated separately twice for 25 min with PBS containing of NHS-biotin (1 mg/ml) (Calbiochem) at RT on a rocker platform. The opposite membranes were incubated with 0.1 M glycine. Thereafter, cells were washed with PBS and further incubated for 5 min with 0.1 M glycine from both sides. After washing the cells three times with PBS, filter membranes were cut out and transferred to BEP. Membranes were incubated for 1 h at 4°C and subsequently sonicated for 2 min. Cell lysates were clarified by centrifugation in a microcentrifuge for 20 min at 4°C. Immunoprecipitation and SDS-PAGE were performed as described above. Proteins were blotted onto polyvinylidene difluoride (PVDF) membranes (Millipore) and subjected to Western blot analysis. Surface biotinylation of MBGV-infected MDCKII cells on filter membranes was performed at 48 h p.i.

Western blot analysis. (i) Detection of biotin-labeled GP.

Immunoprecipitates of biotin-labeled GP separated by SDS-PAGE were blotted onto PVDF membranes. Membranes were blocked with 10% milk powder in PBS at 4°C overnight, washed once with PBS, and incubated for 45 min with peroxidase-coupled streptavidin, diluted 1:200 in PBS–0.1% Tween 20. Subsequently, membranes were washed twice with PBS–0.1% Tween 20 and three times with PBS. Bound streptavidin was detected with Super Signal Ultra (Pierce).

(ii) Detection of nonlabeled GP.

Lysates of MDCK-GP cells or HeLa cells expressing GP using the vaccinia virus-T7 system were separated by SDS-PAGE and blotted onto PVDF membranes, and GP was detected essentially as described by Becker et al. (2).

Pseudotyping VSV with MBGV GP.

MDCK-GP or MDCKII cells were inoculated with VSVΔG-GFP/G at an MOI of 0.01 PFU per cell. At 1 h p.i., cells were washed twice with MEM and incubated for 12 h at 37°C. Subsequently, cells were fixed and permeabilized as described above. GFP fluorescence was detected by fluorescence microscopic analysis.

Electron microscopic analysis.

Transmission electron microscopy and immunoelectron microscopy were performed with PFA-fixed MDCKII cells on membranes as described by Kolesnikova et al. (17).

RESULTS

Constitutive expression of GP in MDCKII cells.

To investigate whether GP is vectorially transported to either the apical or basolateral plasma membrane, we established an epithelial cell line stably expressing GP. To this end, we transfected MDCKII cells with plasmid pSG5-GP and plasmid pIG1, conferring geneticin resistance. Transfected cells were further cultivated in medium containing 1 mg of geneticin per ml and screened for GP-expressing clones by indirect immunofluorescence analysis. The finally selected cell clone (MDCK-GP) expressed GP over 20 passages without showing signs of cytopathogenicity which might have been caused by the foreign glycoprotein. Syncytium formation was not detected even after the culture medium was acidified to pH 5. In MDCK-GP cells, GP was found intracellularly in the endoplasmic reticulum (ER) and the Golgi compartment (Fig. 1A). Surface immunofluorescence of MDCK-GP cells revealed GP at the plasma membrane displaying a punctate pattern (Fig. 1B), as was described previously for MBGV-infected cells and recombinant GP expressed using the vaccinia virus-T7 system (2). Coincidentally with the immunofluorescence analysis, metabolically labeled and immunoprecipitated GP is partly sensitive against Endo H, representing the immature molecule located in the ER (Fig. 1E, left panel). The Endo H-resistant GP-specific signal (Fig. 1E, right panel) represents the mature GP₁. The expression rate of GP in MDCK-GP cells was compared to GP expression in the vaccinia virus-T7 system by Western blot analysis. The amount of GP in MDCK-GP cells was 10 times lower than in the vaccinia virus-T7 system (not shown).

FIG. 1.

Expression of GP in MDCK-GP cells. MDCK-GP cells (A and C) and MDCKII cells (B and D) were cultured on coverslips, fixed, and permeabilized with Triton X-100 (A and B) or left untreated (C and D). (A) Intracellular GP was detected using a monoclonal anti-GP antibody followed by a Texas Red-coupled anti-mouse IgG as the secondary antibody. Surface GP was detected using a rabbit anti-GP serum and a FITC-coupled donkey anti-rabbit Ig serum as the secondary antibody. (E) Endo H digestion of intracellular GP. GP was metabolically labeled and immunoprecipitated using a rabbit anti-GP serum. Left panel: untreated GP. Right panel: Endo H-treated GP. The arrows point to the mature GP₁ and to the ER form of GP which is shifted after treatment with Endo H. Sizes are shown in kilodaltons.

The rate-limiting step of the intracellular transport of GP to the plasma membrane is the transport from the ER to the Golgi apparatus. The kinetics of this step were determined in MDCK-GP cells by pulse-chase analysis and following Endo H digestion of immunoprecipitated GP. After approximately 45 min of chase, the first Endo H-resistant GP₁ molecules were detected, representing GP which had reached the med-Golgi (not shown). This value is in accordance with the transport kinetics of GP determined in MBGV-infected cells (2).

In addition, the supernatant of the cells was checked for the presence of GP. As shown in Fig. 2, GP appeared in the supernatant after 60 to 120 min of chase. Both cleavage products of GP, GP₁ (170 kDa) and GP₂ (46 kDa), could be detected. Additionally, a GP-specific protein of 40 kDa was observed which might represent either a degradation product or a differentially glycosylated form of GP₂.

FIG. 2.

Release of GP into the supernatant of MDCK-GP cells. MDCK-GP cells were pulse labeled with [³⁵S]Promix and chased after removal of the label for 240 min. At the indicated times, supernatants were checked for released GP by immunoprecipitation with a rabbit anti-GP antiserum and protein A-Sepharose. Immunocomplexes were separated by SDS-PAGE, and the radiolabeled proteins were detected by autoradiography. Sizes are shown in kilodaltons.

Pseudotyping of recombinant VSV with GP supplied by MDCK-GP cells.

After we had verified that recombinant GP in MDCK-GP cells was transported correctly to the plasma membrane with kinetics similar to those of authentic GP, the functionality of the recombinant protein was tested. To this end, we took advantage of a recombinant VSV whose gene encoding the surface protein G was replaced by the GFP gene (VSVΔG-GFP; kindly provided by Ralf Wagner). It has been shown previously that recombinant VSV is able to incorporate foreign surface glycoproteins into progeny virions and, more important, to use the incorporated glycoproteins for infection of target cells (37, 42). The recombinant VSVΔG-GFP was first pseudotyped with its own surface protein using a G-expressing cell line (54). The resulting virus (VSVΔG-GFP/G) was used to infect either MDCKII cells or the recombinant MDCK-GP cell line. It was expected that the recombinant virus would replicate in both cell lines but only the MDCK-GP cells give rise to infectious progeny virions containing the MBGV GP in their envelope. Only these should be able to infect fresh cells. In Fig. 3 it is shown that indeed in the course of infection of MDCK-GP cells with VSVΔG-GFP/G, the number of cells showing a cytopathic effect (Fig. 3C) and GFP expression (Fig. 3D) was increased. Infection of MDCKII cells resulted only in single cells with cytopathic effects and GFP fluorescence (Fig. 3A and B). This indicated that recombinant GP is incorporated into progeny VSVΔG-GFP particles, enabling the infection of target cells.

FIG. 3.

Stably expressed GP is functional in recognizing target cells and mediation of infection. A confluent monolayer of MDCKII cells (A and B) and MDCK-GP cells (C and D) was infected with recombinant VSVΔG-GFP/G at an MOI of 0.01 PFU per cell. At 12 h p.i., cells were fixed and analyzed by light microscopy (A and C) and by immunofluorescence microscopy for GFP-expressing cells (B and D).

Vectorial transport of GP.

Intracellular transport of surface proteins in polarized cells often results in the selective targeting of either the apical or basolateral plasma membrane. To investigate whether GP undergoes vectorial transport, MDCK-GP cells were grown on permeable filters until the transepithelial electric resistance measured between the basolateral and apical chamber of the culture vessel showed values above 600 Ω, indicating a completely polarized monolayer. Then the cells were fixed for a short time with PFA and subjected to surface immunofluorescence analysis. Cellular distribution of GP was checked by laser scan analysis. A vertical scan of MDCK-GP revealed that GP was mainly transported to the apical membrane compartment of the cells (Z-scan, Fig. 4A). Weak staining was also detected basolaterally. To investigate whether other viral proteins might influence the apical transport of GP, the distribution of surface GP was then examined in MBGV-infected MDCKII cells. Since MBGV infection of filter-grown MDCKII cells was highly inefficient, cells were infected in suspension and allowed to settle and reach confluence. At 48 h p.i., the polarized monolayer was subjected to surface immunofluorescence analysis. Laser scans revealed that GP was also located almost exclusively at the apical membrane (Fig. 4C).

FIG. 4.

Vectorial transport of GP. (A and B) MDCK-GP cells (A) and MDCKII cells (B) were grown for 2 days on membrane filters (6.5-mm diameter). Polarized cells were fixed, and both sides of the membranes were subjected to immunofluorescence analysis. GP was detected using a rabbit anti-GP followed by a FITC-coupled donkey anti-rabbit Ig serum as the secondary antibody. Pictures show a laser Z-scan through the cell monolayer. (C and D) MDCKII cells in suspension were either infected with MBGV at an MOI of 1 PFU per cell (C) or mock-infected (D) and subsequently cultivated on filter membranes. At 48 h p.i., cells were fixed and subjected to immunofluorescence analysis as described above. (E and F) MDCKII cells were grown for 2 days on membrane filters (6.5-mm diameter). Polarized cells were either infected with VSV at an MOI of 1 PFU per cell (E) or mock-infected (F) and incubated for 5 h at 37°C. Subsequently, cells were fixed and membranes were stained from both sides with a monoclonal anti-VSV G antibody and Texas Red-coupled donkey anti-mouse IgG as the secondary antibody. ap., apical; bas., basolateral.

It was now necessary to confirm that these results were not influenced by the filter membrane, which might have limited the access of antibodies to the basolateral membrane. For this purpose, the distribution of VSV G, a protein known to be transported to the basolateral side (30), was investigated. Polarized MDCKII cells were infected with VSV and analyzed by immunofluorescence using an anti-G antibody. The Z-scan of the monolayer showed that, indeed, G is located mainly at the basolateral membrane, as expected (Fig. 4E). Thus, the absence of GP-specific signals at the basolateral membrane was not due to restricted diffusion of antibodies through the filter.

To underline the results gained with immunofluorescence analyses, a domain-specific surface biotinylation analysis was carried out (Fig. 5). Either the apical or the basolateral membrane of filter-grown polarized MDCK-GP cells was labeled with NHS-biotin under conditions which allow the specific detection of surface proteins. After labeling, cells were lysed, and the biotinylated proteins were precipitated using anti-GP antibodies. The pellets were separated by SDS-PAGE, and proteins were blotted onto PVDF membranes. Biotinylated GP was then detected using peroxidase (POD)-coupled streptavidin and subsequent chemiluminescence analysis. Under these conditions, the great majority of surface GP was shown to be located at the apical membrane (Fig. 5A). Only traces of GP were detected at the basolateral side. Surface biotinylation was subsequently performed with MBGV-infected MDCKII cells as well. For this purpose, MDCKII cells were infected in suspension, and the apical or the basolateral plasma membrane was biotinylated at 36 and 60 h p.i. and further analyzed as described above. Again, GP was found to be transported mainly to the apical plasma membrane of the infected cells (Fig. 5B). The amount of GP at the basolateral membrane increased with time after infection. At 36 h (Fig. 5B), approximately 8% of GP was found basolaterally, and at 60 h p.i., approximately 26% of GP was present at the basolateral membrane (not shown). This result might reflect, on one hand, that GP is released into the culture medium from the apical membrane and, on the other hand, the decrease in cell polarity in the course of infection. Taken together, the surface immunofluorescence and surface biotinylation analyses consistently revealed that recombinant GP in the established cell line and GP in MBGV-infected cells is transported mainly to the apical plasma membrane. These data indicated that GP contains an autonomous transport signal which is not influenced by the presence of other viral proteins.

FIG. 5.

Surface biotinylation of MDCK-GP and MBGV-infected MDCKII cells. (A) MDCK-GP and MDCKII (Mock) cells were grown for 48 h on membrane filters (24-mm diameter). Then, the cells were incubated with NHS-biotin from either the apical (lanes a) or the basolateral (lanes b) side. After lysis of the cells, GP was immunoprecipitated using a rabbit anti-GP antibody, and immunocomplexes were subjected to SDS-PAGE and Western blot. Biotinylated proteins were detected with peroxidase-coupled streptavidin followed by chemiluminescence analysis. (B) MDCKII cells in suspension were infected with MBGV at an MOI of 1 PFU per cell or left untreated (Mock) and cultivated on membrane filters for 48 h. Polarized cells were subjected to surface biotinylation as described above.

Release of GP into the culture medium facing the apical membrane compartment of MDCK-GP cells.

It was now of interest whether the detected release of GP into the culture medium of MDCK-GP (Fig. 2) also took place in a vectorial manner. Filter-grown polarized MDCK-GP cells were metabolically labeled with [³⁵S]Promix for 30 min and, after removal of the labeling medium, chased for 60 and 240 min. GP was then immunoprecipitated from the chase medium and separated by SDS-PAGE. After 60 min of chase, GP could be detected exclusively in the culture medium facing the apical compartment (Fig. 6, apical). Since both cleavage products of GP were found, it is likely that GP is released into the medium in virosomes, as has been shown for the EBOV GP (49). In addition, the detected GP₂ fragments with a higher migration velocity point to proteolytically degraded molecules.

FIG. 6.

Release of GP into the apical supernatant of MDCK-GP cells. MDCK-GP cells and MDCKII cells were cultivated for 2 days on membrane filters (24-mm diameter). Polarized cells were pulse labeled for 30 min with [³⁵S]Promix and chased for 240 min with unlabeled medium. At the indicated times, supernatants were collected from both compartments, and GP was precipitated with a rabbit anti-GP antibody. Immunocomplexes were separated by SDS-PAGE and subjected to autoradiography. Mock, supernatants of MDCKII cells after 240 min of chase. Left panel: basolateral culture medium used for immunoprecipitation. Right panel: apical culture medium used for immunoprecipitation.

Vectorial budding of MBGV.

It has been shown for several viruses that the plasma membrane compartment, to which the surface glycoproteins are transported, is identical to the site where virus budding occurs (10). To investigate whether the apical transport of GP triggers the vectorial budding of MBGV from the apical plasma membrane, MDCKII cells were infected with MBGV/215 in suspension and grown on permeable filters as described before. The medium facing the apical and the basolateral plasma membrane of the infected cells was transferred to Vero cells at 12, 24, 41, and 48 h p.i. Infected Vero cells were then examined at 48 h p.i. for CAT activity. MBGV/215 represents an MBGV stock containing an artificial minigenome with a CAT reporter gene. It was shown previously that the minigenome was passaged together with the full-length genome (26). Thus, CAT activity detected in Vero cells which were infected with the culture supernatants of polarized MBGV/215-infected MDCKII cells indicates released virions. This method represented a valuable tool for detecting and quantifying even small amounts of released virus by taking advantage of the high sensitivity of the CAT assay. CAT activity was detected in Vero cells infected with apical and basolateral culture supernatants of MDCKII cells at 12 h p.i. (Fig. 7, upper and lower panel). CAT activity decreased in Vero cells which were infected with MDCKII supernatants harvested at 24 h p.i. but was found exclusively in cells inoculated with the basolateral supernatant of MDCKII cells. Thus, CAT activity at 12 h presumably represented nonremoved inoculum virus present in the apical and basolateral supernatant of the MDCKII cells. Regarding virus release from MDCKII cells at later time points after infection, to our surprise, CAT activity was exclusively detected in Vero cells which were inoculated with the basolateral supernatants harvested at 24, 41, and 48 h p.i. (Fig. 7). The CAT activity increased with time, pointing to augmented virus release from the basolateral membrane of the MBGV-infected MDCKII cells. Since this result was unexpected, we tried to reconfirm the result using transmission electron microscopy. To this end, MDCKII cells were infected with MBGV in suspension and grown on permeable filters as described before. At 48 h p.i., cells were fixed and processed for transmission electron microscopy and immunoelectron microscopy. We detected electron-dense particles representing progeny virions only at the basolateral membrane of the cells (Fig. 8A [arrows] and 8B [arrowheads]). Virions in the state of budding were also found (Fig. 8B, insect, arrowhead). MBGV particles differed from microvilli and cellular protrusions in their high electron density and the presence of nucleocapsids (Fig. 8B, inset). Polarization of the cells was ensured by the presence of the electron-dense tight junction (Fig. 8A, upper left corner). The released virions displayed the characteristic shape of filoviruses (Fig. 8C) and contained GP (Fig. 8E, arrowheads). The presence of GP in the apical membrane was also detected by immunoelectron microscopy (Fig. 8D). GP is located at the plasma membrane (arrows) and in vesicular structures (arrowheads).

FIG. 7.

Budding of MBGV/215 from the basolateral membrane of infected MDCKII cells. MDCKII cells were infected in suspension with MBGV/215 at an MOI of 1 PFU per cell and grown on permeable membrane filters. Supernatants of MDCKII cells were harvested and transferred to Vero cells at 12, 24, 41, and 48 h p.i. At 48 h p.i., Vero cells were washed once with PBS and lysed, and CAT activity was assayed. Lanes b, lysates of Vero cells infected with the basolateral culture supernatant of MBGV/215-infected MDCKII cells; lanes a, lysates of Vero cells infected with the apical culture supernatant of MBGV/215-infected MDCKII cells. Lanes Mock, lysates of Vero cells infected with the culture supernatant of mock-infected MDCKII cells. Lower panel: quantification of CAT signals of Vero cells infected with the basolateral culture supernatant of MBGV/215-infected MDCKII cells at different times after infection. CAM, chloramphenicol.

FIG. 8.

Budding of MBGV from the basolateral membrane of infected MDCKII cells. MDCKII cells were infected in suspension with MBGV at an MOI of 1 PFU per cell and grown on permeable membrane filters. At 48 h p.i., the MBGV-infected MDCKII cells were washed once with PBS, fixed with 4% PFA in PBS overnight, and subsequently subjected to transmission electron microscopy and immunoelectron microscopy. (A) Infected MDCKII cell, grown on filter membranes. Arrows point to released virions in the intercellular space. At the upper left corner, a part of the apical membrane is shown. The basal plasma membrane is attached to the membrane filter (MF). Bar, 500 nm. (B) Infected MDCKII cell grown on filter membranes. At the basal membrane, budding particles can be recognized (outlined in black). The inset shows a higher magnification of the outlined area. Arrowheads show released and budding virions. The asterisk marks a cellular protrusion. Note the difference in electron density. Bar, 300 nm. (C) Single virion in the intercellular space between infected MDCKII cells. Surface spikes can be seen. Bar, 100 nm. (D) Immunoelectron micrograph of the apical membrane of an infected MDCKII cell. GP is labeled with a rabbit anti-GP serum and colloidal gold (10 nm). Arrows show GP at the plasma membrane. Arrowheads show GP in vesicular structures. Bar, 90 nm. (E) Immunoelectron micrograph of the basal membrane of an infected MDCKII cell. GP is labeled with a rabbit anti-GP serum and colloidal gold (10 nm). Arrows show GP in the basolateral membrane. Arrowheads show GP incorporated into a released virion. Bar, 120 nm. Vi, viral inclusion.

DISCUSSION

Viral infection of polarized cells often results in budding of progeny virions from either the apical or the basolateral membrane. The polarity of budding might have great impact on the course of disease, as shown for Sendai virus (44). While budding of Sendai virus from the apical membrane of bronchial epithelium resulted in a relatively mild respiratory disease, additional basolateral budding is followed by a systemic infection. The prerequisite for polarized budding is the vectorial transport of viral components like nucleocapsids, matrix, and surface protein(s) to either the basolateral or apical membrane. A number of viral surface proteins are transported in a vectorial manner to either the apical or basolateral plasma membrane. Mostly, the destined membrane compartment is identical to the site of viral release. Thus, it has been presumed that viral surface proteins determine the site of viral budding (10, 20, 21, 28).

To investigate whether MBGV is released from polarized cells in a vectorial manner, which might influence the course of the disease, and whether this process is triggered by the single transmembrane protein GP, we analyzed MBGV-infected MDCKII cells and an MDCK cell line constitutively expressing GP. The data presented show that stable expression of GP in MDCKII cells could be achieved without the need for inducible promoters preventing the cell from the putatively toxic effects of the overexpressed viral glycoprotein. It is presumed that the moderate expression level of GP in the MDCK-GP cell line is mandatory for the low cytotoxic effects of GP on the cells.

To verify that recombinant GP expressed by the MDCK-GP cell line was functional, we took advantage of the ability of VSV to incorporate foreign transmembrane proteins (37) by pseudotyping a recombinant VSV (VSVΔG-GFP) with the recombinant GP. The presence of GP enabled the recombinant VSV to infect target cells, indicating that stably expressed GP is able to bind to the receptor and mediate infection. The possibility of creating recombinant VSV pseudotyped with GP or GP mutants provides a valuable tool for investigating functions of GP like receptor recognition and fusion activity (4).

Investigation of MBGV-infected MDCKII cells and the MDCK-GP cell line revealed that GP is vectorially transported, mainly to the apical membrane compartment. This result suggests that GP contains an autonomous apical transport signal which is not influenced by the presence of the other viral proteins. While the signals leading to basolateral transport of transmembrane proteins are well characterized (16, 25), signals leading to apical sorting of proteins are only poorly understood. Recently, the concept emerged that transport of membrane proteins to the apical compartment is achieved by incorporating the proteins directly into glycosphingolipid-cholesterol-enriched membrane subcompartments (rafts) which are transported by default to the apical membrane (40). Structural elements which are presumed to influence apical sorting of surface proteins are (i) the transmembrane domain, as shown for influenza virus hemagglutinin (36), (ii) N-linked glycans (15), and (iii) O-linked glycans (1). It is hypothesized that the sugar side chains are able to connect the proteins to components of the rafts via a lectin-like bond. Whether MBGV GP is directly embedded into rafts, attached to rafts via its N- or O-linked sugar side chains, or directed to the apical membrane by an unknown mechanism is currently under investigation.

Our experiments revealed that apical transport of GP did not result in apical budding of progeny virions. On the contrary, MBGV particles were found exclusively at the basolateral membrane of infected MDCKII cells. As mentioned above, most investigated viruses are released from the infected cells at the same membrane compartment which is targeted by the respective viral surface protein (10, 18, 20, 21, 28, 31, 45). However, Maisner et al. (23) found that measles virus particles are released from the apical membrane, whereas one of the two surface proteins, F, is preferentially transported to the basolateral membrane. This result suggests that vectorial budding might in some cases also be determined by factors other than the viral surface proteins. This is underlined by a study of Rindler et al. (30), who have shown that although VSV G protein is mainly transported to the basolateral surface and budding takes places exclusively basolaterally, a significant part of G is also found at the apical membrane. These molecules, however, did not lead to viral budding, suggesting that other viral or cellular factors in addition to G might trigger virus release at the basolateral membrane. Prime candidates for viral factors determining the vectorial budding of MBGV are the matrix proteins VP40 and VP24 (3). VP40 was found inside in the cytoplasmatic inclusion bodies, which are places of nucleocapsid storage (12; unpublished data). It is hypothesized that VP40 interacts with components of the nucleocapsid and induces the vectorial transport of nucleocapsids by specific interaction with proteins of the cytoskeleton. Once the nucleocapsids have reached the basolateral membrane, the small amount of GP which is clustered to specific areas (2) might be sufficient to mediate budding of infectious particles. This hypothesis is currently under investigation.

Infection of humans or monkeys with MBGV resulted in a systemic infection affecting, among others, several organs with polarized cells like the liver, kidney, and lung (33, 39). The role of vectorial budding from these polarized cells in the course of MBGV disease is poorly understood. Experimental infection of endothelial cells revealed that the release of MBGV takes place in a vectorial manner predominantly from the apical membrane of the infected cells (38). Hepatocytes of experimentally infected animals also displayed vectorial budding of MBGV; however, virus release took place preferentially at the basolateral plasma membrane (E. Ryabchikova, personal communication). Thus, for hitherto unknown reasons, different types of polarized cells obviously support MBGV budding at different membrane compartments. A similar situation is found with Coronaviridae and Togaviridae. Infection of different epithelial cells with the mouse hepatitis virus resulted in budding from different membrane compartments (32). Also, infection of two different epithelial cell types with Semliki Forest virus or Sindbis virus gave rise to virus release from either the apical or basolateral membrane, depending on the cell type (55).

Release of progeny virions from MBGV-infected epithelial cells takes place at the basolateral membrane and is presumed to have an impact on the spread of the virions in the infected organism, as has been shown for Sendai virus (44). GP, however, is shed from the apical membrane into the medium. Shedding of GP is also detected with EBOV, where the protein is found in large quantities in the supernatant of infected cells (49). The role of shed GP molecules, either soluble or incorporated into virosomes, in the course of MBGV or EBOV hemorrhagic fever is not yet understood. For EBOV it was presumed that nonvirion GP may impair the cellular immune response against EBOV. This might be exerted when antigen-presenting cells are lysed by CD4⁺-bearing T cells after presenting GP in context with major histocompatibility complex class II antigen (49). Additionally, an immunosuppressive function of GP is discussed for both filoviruses (6, 7, 47). Future experiments will elucidate the role of basolateral budding of MBGV and the apical release of GP in the pathogenesis of MBGV hemorrhagic fever.