These authors contributed equally to this work.
Assembly of the Marburg virus envelope
Version of Record online: 20 DEC 2012
© 2012 Blackwell Publishing Ltd
Special Issue: Scientific Priority Program 1175
Volume 15, Issue 2, pages 270–284, February 2013
How to Cite
Mittler, E., Kolesnikova, L., Herwig, A., Dolnik, O. and Becker, S. (2013), Assembly of the Marburg virus envelope. Cellular Microbiology, 15: 270–284. doi: 10.1111/cmi.12076
- Issue online: 16 JAN 2013
- Version of Record online: 20 DEC 2012
- Accepted manuscript online: 27 NOV 2012 05:30PM EST
- Manuscript Accepted: 19 NOV 2012
- Manuscript Revised: 18 NOV 2012
- Manuscript Received: 17 SEP 2012
- Top of page
- Results and discussion
- Experimental procedures
- Conflict of interest
The key player to assemble the filamentous Marburg virus particles is the matrix protein VP40 which orchestrates recruitment of nucleocapsid complexes and the viral glycoprotein GP to the budding sites at the plasma membrane. Here, VP40 induces the formation of the viral particles, determines their morphology and excludes cellular proteins from the virions. Budding takes place at filopodia in non-polarized cells and at the basolateral cell pole in polarized epithelial cells. Molecular basis of how VP40 exerts its multifunctional role in these different processes is currently under investigation. Here we summarize recent data on structure–function relationships of VP40 and GP in connection with their function in assembly. Questions concerning the complex particle assembly, budding and release remaining enigmatic are addressed. Cytoplasmic domains of viral surface proteins often serve as a connection to the viral matrix protein or as binding sites for further viral or cellular proteins. A cooperation of MARV GP and VP40 building up the viral envelope can be proposed and is discussed in more detail in this review, as the cytoplasmic domain of GP represents an obvious interaction candidate because of its localization adjacent to the VP40 layer. Interestingly, truncation of the short cytoplasmic domain of GP neither inhibited interaction with VP40 nor incorporation of GP into progeny viral particles. Based on reverse genetics we generated recombinant virions expressing a GP mutant without the cytoplasmic tail. Investigations revealed attenuation in virus growth and an obvious defect in entry. Further investigations showed that the truncation of the cytoplasmic domain of GP impaired the structural integrity of the ectodomain, whichconsequently had impact on entry steps downstream of virus binding. Our data indicated that changes in the cytoplasmic domain are relayed over the lipid membrane to alter the function of the ectodomain.
- Top of page
- Results and discussion
- Experimental procedures
- Conflict of interest
Marburg virus (MARV) belongs to the Filoviridae, a family of viruses with extended filamentous morphology that encompasses viruses that are highly pathogenic for humans and non-human primates. Members of the Filoviridae cause severe haemorrhagic fever against which neither an approved vaccine nor antiviral therapy is available (Sanchez et al., 2007). Filoviruses infect a wide range of cells, among which are macrophages, dendritic cells, hepatocytes, adrenocorticocytes, fibroblasts and endothelial cells (Ryabchikova et al., 1999). The filamentous MARV particle is composed of seven proteins; five are associated with the nucleocapsid (NP, VP35, L, VP30 and VP24). The nucleoprotein NP encapsidates the non-segmented negative-strand RNA genome. VP35 and L form the RNA-dependent RNA polymerase. VP30 regulates viral transcription and the function of VP24 is enigmatic. The nucleocapsid is encased by a layer of the matrix protein VP40; the outside of the layer is associated with the viral lipid membrane. The surface of the virion is covered by homotrimers of the surface glycoprotein GP (Bharat et al., 2011). Viral entry is mediated by GP which attaches to the target cell, and also catalyses fusion between viral and endosomal membranes (Feldmann et al., 1991; Volchkov et al., 2000; Carette et al., 2011). After entry, the viral genome is transcribed and replicated and new viral proteins are synthesized in the cytoplasm of infected cells (Mühlberger et al., 1998). Newly formed nucleocapsids are first detected in perinuclearly located viral inclusions (Kolesnikova et al., 2000) from where mature nucleocapsids are transported to the plasma membrane. Budding of MARV takes place at filopodia in non-polarized cells and at the basolateral side of polarized epithelial cells (Sanger et al., 2001; Kolesnikova et al., 2007a,b; Dolnik et al., 2008).
The viral protein responsible for the filamentous morphology of the virions is VP40. VP40 mediates the recruitment of nucleocapsids and GP to the plasma membrane and induces viral budding (Kolesnikova et al., 2004a; 2007b; Mittler et al., 2007). This is remarkable because nucleocapsids, VP40 and GP do not share common transport pathways, suggesting VP40 to be a truly multifunctional protein. Important for the formation of the viral envelope is a close interaction of VP40 and GP. GP is the only MARV surface protein and plays a central role in receptor binding and entry processes. Here we review recent studies on structure–function relations of MARV VP40 and the surface glycoprotein GP, and identify open questions in the field of MARV assembly.
Structure and function of the matrix protein VP40
Membrane association of VP40
MARV VP40 (303 amino acids) is a peripheral membrane protein that becomes associated with cellular membranes immediately after synthesis. At 2 h after synthesis 50% of the molecules are attached to membranes and under equilibrium conditions approximately 80% of VP40 molecules are membrane-associated (Kolesnikova et al., 2002). Time-course analyses revealed that VP40 interacts with different types of cellular membranes (Kolesnikova et al., 2004b). The initial intracellular structures positive for VP40 are small vesicles of 30–50 nm in diameter. Later, VP40 is found on the surface of large membrane compartments morphologically resembling multivesicular bodies (MVBs) and containing markers of late endosomes. Subsequently, VP40 is also associated with peripheral MVBs, the plasma membrane and filamentous extrusions of the plasma membrane. Fission of these filamentous extrusions results in appearance of VP40-enriched filamentous virus-like particles (VLPs) in the supernatant (Kolesnikova et al., 2004a; Swenson et al., 2004). The intracellular membrane-associated transport pathway of VP40 differs from the classical secretory pathway of MARV GP, resulting in a different distribution of the two proteins at their final destination at the plasma membrane. While GP is diffusely distributed throughout the plasma membrane, VP40 accumulates in the clusters located at the cell periphery and in filamentous protrusions of the plasma membrane (Kolesnikova et al., 2004b). The different distribution patterns of GP and VP40 become more obvious in polarized cells, where VP40 and GP are transported to different domains: expression of GP results in its accumulation at the apical domain, while recombinantly expressed VP40 accumulates in clusters at the basolateral domain (Kolesnikova et al., 2007b). Upon coexpression of both proteins, GP is redistributed to the VP40-positive clusters at the basolateral domain where budding of MARV takes place (Kolesnikova et al., 2007b). Coexpression of both viral proteins in non-polarized cells is also accompanied by redistribution of GP to the peripheral VP40-enriched clusters, indicating that VP40-enriched membrane attracts GP and not vice versa (Kolesnikova et al., 2004a).
Thus, interaction of VP40 with cellular membranes is important for protein transport and budding activity; however, the exact mechanism of how VP40 binds to membranes remains unknown. Hydropathy plot of VP40 shows several hydrophobic parts in the C-terminal domain of the protein, suggesting this domain might be responsible for membrane binding as has been shown for Ebola virus (EBOV) VP40 (Ruigrok et al., 2000). Besides, it is currently unclear (i) which mechanisms trigger interaction of VP40 with different cellular membranes, (ii) whether a direct fusion of the peripheral VP40-containing MVBs or vesicular-dependent transport provides access of VP40 to the plasma membrane and (iii) what is a mechanism of basolateral targeting of VP40.
Oligomerization of VP40
The crystal structure of the closely related EBOV VP40 shows two individually folded domains (Dessen et al., 2000). The C-terminal domain of VP40 mediates binding to membranes, which induces a conformational change in the molecule that might allow polymerization of the protein via protein–protein interaction of the N-termini (Ruigrok et al., 2000). For EBOV VP40 four conformational states are described: a monomeric form (Dessen et al., 2000), a dimeric form (Gomis-Rüth et al., 2003; Timmins et al., 2003a) and higher-ordered oligomeric structures that include hexamers and octamers (Ruigrok et al., 2000; Scianimanico et al., 2000). Hexa- and octamers of filoviral VP40 form ring-like structures (Timmins et al., 2003b; Hoenen et al., 2010). In MARV particles a highly ordered protein layer of VP40 is located beneath the lipid bilayer showing a striated appearance (Kolesnikova et al., 2002; Bharat et al., 2011; 2012). A similar striation was detected in MARV VP40-induced VLPs (Kolesnikova et al., 2002), underlining the ability of VP40 to form a highly ordered protein array. The filamentous MARV VP40-induced VLPs do not contain cellular proteins, indicating that polymerization of VP40 is important for the extrusion of cellular proteins from viral envelope.
Although a recent study showed that amino acid residues from 84 to 89 might be important for overall protein structure and stability of MARV VP40 (Liu et al., 2010), the amino acid residues of the protein involved into protomer–protomer or dimer–dimer interactions remain unknown. The organization of the VP40 oligomers beneath viral membrane remains enigmatic, as well. It is still unclear how identical ring-like structures might be organized to create an outer hydrophobic surface which interacts with lipids, and an inner hydrophilic side surrounding the nucleocapsid.
Late domains in VP40 and NP
Association of VP40 and other viral matrix proteins with late endosomal compartments, like MVBs, has gained significant attention during the last years. MARV VP40 contains a late domain motif PPPY at aa position 16–19, which was shown to mediate interaction with the endosomal sorting complex required for transport (ESCRT)-associated protein Nedd4, an E3-like ubiquitin ligase (Urata and Yasuda, 2010). ESCRT complexes are essential to induce inward budding of vesicles into MVBs, a process necessary for lysosomal targeting and back transport of, e.g. receptor tyrosine kinases to the plasma membrane (Henne et al., 2011). Inward budding of these cellular vesicles is topologically similar to virus budding. Interaction with proteins of the ESCRT complexes has been shown for many virus systems to increase or even enable budding of viral particles (Bieniasz, 2006). In case of VP40, mutation of the Nedd4 binding late domain resulted in a diminished release of VP40-induced VLPs, suggesting a contributing role of the ESCRT machinery in release of viral particles (Urata et al., 2007).
It is assumed that the ability to recruit the ESCRT complexes to the budding site by viral late domain motifs evolved independently in different MARV proteins. While two late domains are present in EBOV VP40, to recruit Tsg101 and Nedd4, the active late domains in MARV are split between VP40 and NP. In line with this assumption NP has been reported to enhance the release of VP40-induced VLPs. NP contains two conserved late domain motifs that allowed interaction with Tsg101. The N-terminal PTAP motif was inactive probably because the NP N-terminus is involved in NP–NP self-interaction. Mutation of the PSAP motif in the C-terminus of NP inhibits the enhancing function of NP on the release of VP40-mediated VLPs (Dolnik et al., 2010). These data indicated that recruitment of Tsg101 via the PSAP motif in NP supports the release of VP40-induced VLPs. A contribution of the ESCRT machinery to MARV budding was also supported by the finding that dominant negative mutants of VPS4, a key ATPase for the recovery of ESCRT components, inhibits release of VLPs and infectious virions, as well (Kolesnikova et al., 2009).
However, mutations in VP40 (or NP) late domains or inhibition of the ESCRT machinery in general do not completely inhibit budding of VLPs or viral particles (Urata et al., 2007; Kolesnikova et al., 2009; Dolnik et al., 2010). These findings tempt us to suggest that the vacuolar protein sorting pathway is only one of several mechanisms that drive MARV budding.
Two types of VP40-positive VLPs
As mentioned above, expression of VP40 results in the release of VP40-positive particles into the supernatant. Close examination of these particles revealed two different classes. On the one hand, VP40 is associated with vesicular particles that contain many cellular proteins. On the other hand, filamentous particles are observed that almost exclusively consist of VP40 and closely resemble MARV particles (Kolesnikova et al., 2009). The existence of these two classes of VP40-containing particles needs to be considered because only the formation and release of filamentous VLPs is dependent on the activity of late domains or the specific interaction with GP.
Phosphorylation of VP40
In SDS-PAGE profiles VP40 migrates as double band at 35–37 kDa and recent investigations showed that the upper band represents a phosphorylated form of VP40 (Kolesnikova et al., 2012). Phosphorylated tyrosine residues are mainly localized to the N-terminus. Interestingly, phosphorylated tyrosine at position 19 is located within the late domain motif PPPY (Kolesnikova et al., 2012). While mutation of the phosphorylation acceptor sites left most functions of VP40 (e.g. membrane association, oligomerization, VLP formation and budding) intact, the non-phosphorylatable VP40 was significantly inhibited in its ability to recruit nucleocapsids to the plasma membrane (Kolesnikova et al., 2012). This is supported by the finding that infectious virus-like particles (iVLPs) induced by the activity of non-phosphorylatable VP40 contain less nucleocapsids and are less infectious than iVLPs induced by wild-type VP40. It is currently unknown which cellular kinase is responsible for phosphorylation of VP40 and where phosphorylation of VP40 takes place. Additionally, it is unclear where phosphorylated VP40 meets the nucleocapsids. Our data suggest that perinuclear VP40 which is detected within the inclusions is not phosphorylated and the highest concentration of phosphorylated VP40 is detected close to the plasma membrane. Interestingly, only a small fraction of virion-associated VP40 is phosphorylated, suggesting a catalytic process that e.g. induces either the entrance of nucleocapsids into the filopodia or envelopment of nucleocapsids.
Structure and function of the surface glycoprotein GP
Glycosylation and oligomerization of GP
The glycoprotein GP (681 aa), a type I transmembrane protein, represents the only MARV surface protein. GP is composed of an ectodomain (ED, aa 1–642), a membrane-spanning domain (TMD, aa 643–673) and a cytoplasmic domain (CD, aa 674–681, Fig. 1). GP is cotranslationally integrated into the membrane of the rough endoplasmic reticulum (ER) from where it exploits the classical secretory pathway and is finally transported to the plasma membrane. During its transport, the glycoprotein GP passes the stacks of the ER, the Golgi apparatus and the trans-Golgi network and undergoes extensive co- and post-translational modifications. GP contains more than 20 potential N-glycosylation sites (N-Xxx-S/T) and numerous potential O-glycosylation sites; most of them are probably used because approximately 50% of the molecular weight of GP (220 kDa) is contributed by sugar side-chains. While N-glycans are distributed over the whole ectodomain, O-glycans are concentrated in the mucin-like domain (aa 289–501). High-mannose N-glycans are cotranslationally attached in the ER and partially converted to complex N-glycans in the Golgi apparatus and trans-Golgi network (Feldmann et al., 1991; Geyer et al., 1992; Mittler et al., 2011). Next to the extensive glycosylation, GP undergoes trimerization within the ER mediated via a hypothetical oligomerization domain (aa 584–599) (Feldmann et al., 1991). Homo-trimerization is essential for fusion activity of GP during viral entry into target cells. In the trans-Golgi network GP is proteolytically cleaved at the sequence RRKR435 by the subtilisin-like endoprotease furin into its two subdomains GP1 (170 kDa) and GP2 (50 kDa). GP2 carries the transmembrane domain, is incorporated into the viral or cellular membrane and stays linked to GP1 via an intra-molecular disulfide bridge. It has been published previously that filovirus infection is dependent on extensive proteolytic priming and complex conformational rearrangements of GP occurring within the late endosome and endo-lysosomal compartment of the target cell (Chandran et al., 2005; Schornberg et al., 2006). Recent investigations elucidated that cleavage of EBOV GP by endosomal cathepsins was essential for binding of GP to the newly described receptor NPC1 located in the endo-lysosomal membrane. Combined with the concurrent activity of an until now unknown fusion trigger, this results into fusion of viral and endo-lysosomal membranes and finally release of the viral nucleocapsid into the cytoplasm (Carette et al., 2011; Cote et al., 2011; Miller et al., 2012).
Acylation of GP
At the boundary between the transmembrane domain and the eight aa long cytoplasmic domain, GP is acylated at two cysteine residues (aa 671 + 673) (Funke et al., 1995). The function of acylation is so far unknown although it has been suggested that acylation mediates association of EBOV GP with cholesterol-rich membrane microcompartments (lipid rafts). Colocalization of GP with the lipid raft marker GM1 at the plasma membrane in transfected cells as well as the incorporation of GM1 into purified MARV and EBOV was reported (Bavari et al., 2002). Other experimental approaches using confocal immunofluorescence microscopy did not confirm colocalization of filoviral GPs with lipid raft markers at the cell surface (Lopez et al., 2012). In addition, costaining of GM1, GP and VP40 in VP40-induced peripheral clusters close to the plasma membrane was suggested to occur because of an extensive accumulation of membranes at the VP40-induced budding sites (Dolnik et al., 2008).
The transmembrane domain of GP determines colocalization with VP40
Previous investigations from our laboratory revealed a significant role of the transmembrane domain of GP for assembly of the viral envelope (Mittler et al., 2007) which is mainly composed of GP and the matrix protein VP40. As mentioned above, coexpression of GP and VP40 results in redistribution of GP to VP40-enriched plasma membrane protrusions. Substitution of the transmembrane domain of GP with the transmembrane domain of the Lassavirus surface protein GPC (a protein that is not transported into MVBs upon coexpression with VP40) prevented the colocalization of GP mutants with VP40 in VP40-enriched plasma membrane protrusions and also their subsequent incorporation into filamentous VLPs. In contrast, when the transmembrane domain of MARV GP is swapped into the background of Lassavirus GPC, the resulting mutant of GPC was able to colocalize with VP40-enriched membranes and was recruited into VLPs. These results demonstrated that the accumulation of GP in VP40-enriched structures is a prerequisite for its integration into the viral envelope and the transmembrane domain is essential and sufficient for recruitment of GP into the viral envelope. So far the mechanism of how the transmembrane domain of GP recognizes the VP40-enriched membrane remains enigmatic as well as how GP is transported to the VP40-enriched membrane protrusions. Acylation, the length, sequence or hydrophobicity of the transmembrane domain can be considered as important factors.
Removal of the cytoplasmic domain impairs function of GP in an iVLP assay
In contrast to the transmembrane domain, the cytoplasmic domain of GP does not play a prominent role in assembly of the viral envelope, although because of its close proximity to the VP40 lattice, a direct interaction of both proteins via the cytoplasmic domain could be assumed. However, previous studies revealed that GP lacking the cytoplasmic domain (GPΔCD) colocalized with VP40 and was incorporated into released (i)VLPs undistinguishable to wild-type GP (GPwt) (Mittler et al., 2007). Interestingly, iVLPs decorated with GPΔCD were less infectious than those decorated with GPwt, suggesting a role of the cytoplasmic domain during viral entry. Further studies showed that the structural integrity of the ectodomain of GP was dependent on the presence of the cytoplasmic domain (Mittler et al., 2011).
Results and discussion
- Top of page
- Results and discussion
- Experimental procedures
- Conflict of interest
Rescue and characterization of a recombinant MARV incorporating GP missing its cytoplasmic domain
Our previous studies indicated a functional link between the ectodomain and the cytoplasmic domain of GP exceeding a simple physical stabilization of the protein in the lipid viral envelope. We obtained these data using several surrogate systems mimicking viral infection (e.g. iVLP assays) (Mittler et al., 2011). It was therefore of major interest to understand the role of the cytoplasmic domain in the context of the MARV infection. To this end, we generated a recombinant MARV which expressed GP lacking the cytoplasmic domain (GPΔCD) instead of GPwt. Successful rescue of the recombinant virus (recMARVGPΔCD) could be achieved with helper plasmids encoding the MARV ribonucleoprotein complex proteins and recMARVGPΔCD was verified by sequencing (data not shown). The protein composition and particle morphology of recMARVGPΔCD were then characterized in comparison with recMARV. Silver staining of the two recombinant MARVs revealed that the amount of GPΔCD in recMARVGPΔCD was comparable to the amount of GPwt in recMARV. In addition, the overall protein composition of recMARVGPΔCD was not altered (Fig. 2A). To confirm these results, we evaluated the quantity of integrated glycoproteins in recMARV and recMARVGPΔCD by immunoelectron microscopy and detected that both GPΔCD and GPwt were present in similar amounts (Fig. 2B; 69 ± 38 vs. 63 ± 27 gold particles/μm). Additionally, the filamentous morphology of recMARVGPΔCD particles appeared to be identical to those of recMARV decorated by GPwt. These results were in line with our previous data showing negligible impact of GP cytoplasmic domain on protein composition and morphology of iVLPs (Mittler et al., 2007; 2011).
Recently published iVLP experiments revealed that, as a result of the deleted cytoplasmic domain, the ectodomain of GP undergoes conformational modifications (Mittler et al., 2011). In order to confirm this finding in the context of recMARVGPΔCD, we determined the dilution of a neutralizing serum necessary to inhibit infection by recMARV and recMARVGPΔCD by 50%. While 50% neutralization of recMARVGPΔCD was accomplished up to serum dilutions of 1:8 to 1:16, the infectivity of recMARV was still significantly inhibited even at a serum dilution of 1:512 (Fig. 2C). These results supported previous data from iVLP experiments and indicated that the removal of the cytoplasmic domain induced conformational changes in the ectodomain of GPΔCD incorporated into a recombinant virus. The conformational changes led to altered presentation and recognition of immunodominant epitopes (Mittler et al., 2011). Modifications in the cytoplasmic domain of other viral surface proteins were also shown to result in structural rearrangements of their ectodomain. For example, when the cytoplasmic domains of the fusion proteins (F) of Nipah virus or simian virus 5 were mutated or truncated, F underwent functional changes (Waning et al., 2004; Aguilar et al., 2007).
Growth kinetics and plaque morphology analyses of recMARVGPΔCD
In order to assess the impact of the cytoplasmic domain of GP on viral infectivity and replication efficiency, VeroE6 cells were infected with recMARVGPΔCD or recMARV and production and infectivity of progeny virus were analysed (Fig. 3A). Western blot analysis of infected cell lysates revealed the expression of viral proteins (NP) starting from day three post infection (p.i.) for both viruses. Release of viral particles into the supernatant was first detected at day three (recMARVGPΔCD) or day four p.i. (recMARV). This result indicated that, although the exchange of GPwt to GPΔCD was tolerated with respect to support virus infection and growth, recMARVGPΔCD seemed to be more rapidly replicated and released (Fig. 3A, upper panel, compare protein content on d4 and d7). However, the increased amount of viral proteins in the tissue culture supernatant did not correlate with an increase in infectious titre (Fig. 3A, lower panel, compare TCID50/ml on d4). On the contrary, recMARVGPΔCD was slightly attenuated. It has been reported for rabies virus and vesicular stomatitis virus that removal of the cytoplasmic domains of their surface glycoproteins modulates viral budding efficiency. However, in contrast to data presented in the present study, a significant reduction in particle release was observed (Mebatsion et al., 1996; Schnell et al., 1998). This reduction in viral budding was discussed to be linked to a missing interplay between surface and matrix proteins. Data presented in the current study underlined our previous findings that the interaction of MARV GP and the matrix protein VP40 was not based on the cytoplasmic tail but on the transmembrane domain of GP (Mittler et al., 2007).
Subsequently, we compared plaque sizes induced by infection with recMARVGPΔCD and recMARV. To this end, infected cell cultures were immunostained and the size of foci of infected cells was measured demonstrating that recMARVGPΔCD generated significantly smaller plaques than recMARV (Fig. 3B). This result substantiated the data shown in Fig. 3A that deletion of the cytoplasmic domain of GP resulted in growth defects of recMARVGPΔCD. Differences in plaque sizes induced by recMARV and recMARVGPΔCD were more pronounced than the differences observed in the TCID50 assay, although both approaches reflected multi-cycle replication. Nevertheless, comparable results have been received with recombinant simian virus type 5 carrying a mutated fusion protein whose cytoplasmic tail was almost completely truncated (Waning et al., 2002). Like with recMARVGPΔCD the plaques originating from this modified simian virus type 5 were substantially decreased in size, but infectious titres were unchanged.
RecMARVGPΔCD showed a reduced capability to establish initial steps of the viral life cycle
As the performed TCID50 and plaque assays both reflected multi-cycle viral replication (Fig. 3), it was of interest to characterize recMARVGPΔCD in a single-cycle replication assay to monitor viral entry. It is conceivable that the observed conformational modifications in the ectodomain of GP, caused by the missing cytoplasmic tail, influenced performance of GP during the complex entry and fusion process (see above). We infected HUH-7 cells with the same particle amounts of recMARV and recMARVGPΔCD and fixed the cells at 16 h p.i., the first time point to clearly detect newly synthesized viral proteins by immunofluorescence. Quantification of the amount of infected cells revealed that the number of infections induced by recMARVGPΔCD was significantly diminished compared to those mediated by recMARV (Fig. 4A; 60 ± 3% vs. 86 ± 8% infected cells). As an impact of GPΔCD on viral transcription and replication was excluded (data not shown), the result obtained here suggested a reduced capability of recMARVGPΔCD to initiate infection. In order to further dissect the mechanism underlying the defects of recMARVGPΔCD, we examined the binding of viral particles to target cells. Purified recMARVGPΔCD and recMARV were incubated with chilled VeroE6 cells and bound particles were quantified by Western blot analysis. Examinations revealed that recMARV and recMARVGPΔCD particles did not show statistically significant binding differences (Fig. 4B; 100% vs. 83.5 ± 15.2% particle binding).
Taken together, our data support the idea that the very short cytoplasmic domain of GP contributes to stabilizing the conformation of the ectodomain. While removal of the cytoplasmic domain did not impair intracellular transport or incorporation of GP into particles, conformation of GPΔCD was changed and recMARVGPΔCD exhibited an attenuated growth phenotype most likely as a result of alterations during viral entry. Currently, it is unclear which step in the complex and highly orchestrated entry process is impaired; as binding to target cells seems not to be inhibited, it is likely that the attenuated phenotype of recMARVGPΔCD results from alterations downstream of this event. Either priming cleavage of GP by endosomal proteases, recognition of NPC1 or GP-mediated fusion of viral and lysosomal membranes could be modified. Investigations are underway to further dissect steps in virus entry that are modified by the conformational changes in the ectodomain of GPΔCD.
Acylation and the cytoplasmic domain of MARV GP synergistically stabilize the conformation of the ectodomain of GP
The missing cytoplasmic domain presumably impaired anchorage of GP in the viral membrane which relayed to the conformational changes in the ectodomain (see above). We were interested whether further destabilization of the GPΔCD–lipid membrane interaction would lead to a more severe phenotype during viral entry and whether incorporation of such severely destabilized GP molecules into the viral membrane would be inhibited.
MARV GP is acylated at the boundary between its transmembrane anchor and the cytoplasmic domain (Funke et al., 1995). It is conceivable that attached fatty acids (mainly palmitic acid) dive into the lipid membrane and thus mediate stabilization of GP in the lipid bilayer. We therefore mutated the two experimentally determined acylation sites at aa 671 and 673 in GPΔCD and in GPwt (Fig. 5A; GPΔCD_C/A and GP_C/A respectively). Transient expression of GPΔCD_C/A and GP_C/A revealed a prominent perinuclear accumulation of the recombinant proteins in line with an accumulation in the ER/Golgi region which was indistinguishable from GPwt (data not shown). Coexpression of GPwt, GPΔCD_C/A or GP_C/A with VP40 resulted for all glycoproteins in colocalization with VP40 in peripheral clusters, which were described to serve as sites for MARV envelope assembly (data not shown) (Kolesnikova et al., 2004b; Mittler et al., 2007). GPΔCD, GP_C/A and GPΔCD_C/A were then employed in a MARV-specific iVLP assay, which served as a model system for filoviral infection and investigated the incorporation of GPwt and the mutant GPs into released iVLPs (Wenigenrath et al., 2010). As shown in Fig. 5B, intracellular expression of GP_C/A and GPΔCD_C/A as well as incorporation into released iVLPs were similar to GPwt. Additionally, the overall protein composition of released iVLPs was not altered (exemplified by analysing NP and VP40). Purified iVLPs were used to infect HUH-7 cells, pretransfected with plasmids encoding the MARV nucleocapsid proteins (Wenigenrath et al., 2010). Infected cells were lysed at 60 h p.i. and reporter protein activity was determined. Interestingly, iVLPs containing GP_C/A were similarly infectious as those containing GPwt, suggesting that removal of only the acylation sites in GP had no substantial effect in GP-mediated infectivity (Fig. 5C). In contrast, iVLPs incorporating the mutant GPΔCD_C/A were significantly less infectious than iVLPs containing either GPwt or GPΔCD (Fig. 5C; 6.7 ± 4.8% infectivity). Based on these results we suggested a synergistic effect of the acylation and the cytoplasmic domain to maintain the structural integrity of GP.
In order to confirm these results in the context of infectious MARV, we generated recombinant viruses expressing either GP_C/A or GPΔCD_C/A. Both recombinant viruses (recMARVGP_C/A and recMARVGPΔCD_C/A) were rescued and verified by sequencing (data not shown). Subsequently, growth kinetics of recMARVGP_C/A and recMARVGPΔCD_C/A were analysed and compared with recMARVGPΔCD and recMARV (Fig. 5D). Lysates of infected cells were examined via Western blot for the presence of NP as a representative of viral protein expression. Traces of NP were detected with all viruses at 3 days p.i. and signal strength of NP increased during the following days peaking at days four to seven p.i. with recMARVGP_C/A lagging slightly behind. The reduced amount of intracellular NP correlated with less viral proteins in the supernatant emanating from recMARVGP_C/A-infected cells, suggesting an attenuated viral growth. TCID50 assays revealed the largest difference in growth kinetics between recMARV and recMARVGPΔCD_C/A; the end-point titres differed by more than 1 log (Fig. 5D; 3.8 × 106 vs. 2.8 × 105 TCID50/ml). RecMARVGP_C/A and recMARVGPΔCD showed an intermediate growth curve.
Function of the acylation of filoviral GP has been investigated previously with pseudotyped viruses or VLP systems (Ito et al., 2001; Bavari et al., 2002). While loss of the acylation sites in EBOV GP did not affect infectivity of pseudotyped vesicular stomatitis virus particles, other studies showed that acylation was crucial for the release of recombinantly expressed EBOV GP in form of spherical particles. In contrast to the previously published data, our approaches were more closely related to the native filovirus infection. We made use of an iVLP system expressing all the MARV structural proteins which did not reveal an effect of acylation on release or infectivity of iVLPs. Only in combination with the removal of the cytoplasmic domain did mutation of the acylation acceptor sites in GP effectively reduce iVLP infectivity. Investigation of the effects of GP acylation using recombinant MARV confirmed the results gained with the iVLP system.
In summary, these data supported the assumption that the cytoplasmic domain and acylation of the transmembrane domain of GP serve to stabilize the molecule in the viral envelope preventing conformational changes in the ectodomain, which interfered with functions of GP during viral entry.
Current model of MARV assembly and release
During recent years understanding of structure–function relationship of VP40 and GP was considerably improved. According to our current hypothesis of MARV assembly, nucleocapsids, GP and VP40 use different transport routes (Fig. 6). Recombinant expression of GP results in the transport of GP along the secretory pathway to the plasma membrane and its release in spherical vesicles (yellow panel). Recombinant VP40 is transported to filopodia along the retrograde late endosomal transport pathway and induces the release of filamentous VLPs. These VLPs exclusively contain VP40; the majority of cellular proteins is excluded because of oligomerization of VP40 (green panel). Upon coexpression of viral proteins or in the course of viral infection, GP is transported from the secretory pathway to peripheral VP40-positive compartments that colocalize with late endosomal markers (beige panel). The accumulation of GP in the VP40-positive membrane compartments is mediated by the transmembrane domain of GP. Only GP that crossed the VP40 compartment is further recruited into viral particles. It is suggested that the VP40-positive membrane compartments serve as the assembly site for the viral envelope. Data obtained so far indicate that VP40 plays a leading role in recruitment of GP, as well as in extrusion of cellular proteins from viral envelope. Nucleocapsids are formed in the perinuclear region and accumulate in inclusions. Upon a trigger by VP40, transport-competent nucleocapsids are transported in direction to the plasma membrane along the cytoskeleton. It is not clear whether microtubules, actin or both are involved. Nucleocapsids are recruited into filopodia, the sites of viral budding, by phosphorylated VP40, while the budding function of VP40 is independent of phosphorylation (see green panel). Transport of VP40 inside filopodia and budding is dependent on actin polymerization. Finally, budding of progeny virions takes place at the sides or the tips of filopodia.
- Top of page
- Results and discussion
- Experimental procedures
- Conflict of interest
Cells and viruses
HUH-7 (human hepatoma), HEK293 (human embryonic kidney) and Vero/VeroE6 (African green monkey kidney) cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum, L-glutamine and penicillin-streptomycin (Gibco, Karlsruhe, Germany) at 37°C under 5% CO2.
Rescue of recombinant MARV (recMARV) was achieved as described by Kraehling et al. (2010). Work with recMARV was performed in the biosafety level 4 (BSL-4) facility of the Philipps University, Marburg, Germany.
Plasmids encoding MARV NP, VP35, VP30, L, VP40, VP24 and GP under the control of a chicken β-actin promoter (vector pCAGGS) were constructed as described elsewhere (Bamberg et al., 2005; Mittler et al., 2007; Wenigenrath et al., 2010). The construction of a MARV-specific artificial minigenome encoding Renilla luciferase as a reporter protein under the control of a T7 promoter was described by Wenigenrath et al. (2010). The plasmid pCAGGS-T7, encoding the T7 DNA-dependent RNA polymerase, was kindly provided by Y. Kawaoka (University of Tokyo, Japan, and University of Wisconsin, Madison, WI, USA). Generation of pCAGGS-MARV GPΔCD, in which the cytoplasmic domain coding sequence was completely deleted, was described previously by Mittler et al. (2007). For MARV GP_C/A, the cysteines at aa position 671 and 673 were substituted to alanines by site-directed mutagenesis using pCAGGS-MARV GPwt as template. Cloning of MARV GPΔCD_C/A was performed by amplifying GPΔCD_C/A by PCR with pCAGGS-MARV GPΔCD as template using a reverse primer including base pairs encoding for alanines instead of cysteines (aa 671 + 673). The amplified product was cloned into pCAGGS via the restriction enzymes SmaI and SacI.
Cloning of full-length MARV cDNA clones was performed by amplifying non-coding 3′ and 5′ ends of the MARV genome from the MARV-specific minigenome (Mühlberger et al., 1998). As RT-PCR template for coding and intragenic regions, genomic viral RNA was used. The anti-genomic sequence of MARV Musoke (accession number: NC 001608) was cloned in three parts [Fragment 1 (FR1): T7leader-NP-VP35-VP40-GP-; FR2: GP-VP30-VP24-L-; FR3: L-trailer-ribozyme] flanked by unique restriction sites into three individual pBlueScript plasmids. Assembly of the full-length plasmid containing the whole anti-genome of MARV Musoke was performed by standard ligation of the three DNA fragments into a minimal pBlueScript vector under the control of the T7 polymerase promoter (Mühlberger et al., 1998). To distinguish between recombinant and wild-type virus, silent mutations at position 6498 (C > T) and 7524 (A > G) were introduced resulting in the deletion of a KpnI restriction site and insertion of a SacII restriction site respectively. Site-directed mutagenesis was used to introduce point mutations or deletions into the GP gene resulting in deletion of the cytoplasmic domain (GPΔCD), mutation of the acylation sites (GP_C/A) or mutants devoid of both the cytoplasmic domain and the acylation sites (GPΔCD_C/A). All constructs were verified by DNA sequencing. Detailed cloning strategies as well as primer sequences are available upon request.
Mouse monoclonal antibodies were used for detection of MARV VP40 [dilution for Western blotting (WB), 1:200], MARV NP (dilution for WB, 1:200; dilution for immunofluorescence analysis, 1:50) and MARV GP (dilution for WB, 1:100; dilution for immunoelectron microscopy, 1:50). Cellular β-tubulin was stained with a monoclonal antibody in a dilution of 1:3000 for WB (Sigma-Aldrich, Munich, Germany). For immunoplaque assays (dilution 1:150) and neutralization assays, a goat anti-MARV serum raised against γ-irradiated MARV was used. Secondary IgGs conjugated with Rhodamine or fluorescein isothiocyanate (FITC) were used for immunofluorescence analysis (dilution, 1:200; Dianova, Hamburg, Germany). Secondary antibodies coupled to Alexa Fluor 680 (Molecular Probes, Karlsruhe, Germany) and to horseradish peroxidase (HRP; Dako, Glostrup, Denmark) were used for WB analysis (dilutions 1:5000 and 1:30 000 respectively). Secondary antibodies conjugated to 5 nm colloidal gold particles were used in immunoelectron microscopy (dilution 1:30; BB International, Cardiff, UK).
Infectivity of recMARV particles released into the supernatant of cells was determined by a 50% tissue culture infective dose (TCID50) assay: VeroE6 cells were grown in 96-well plates to 50% confluence and inoculated in quadruplicate with 10-fold serial dilutions of supernatants of VeroE6 cells infected with multiplicity of infection (MOI) 0.01 TCID50/cell of recMARV for 1 to 7 days. Assays were evaluated 10 days p.i. and TCID50 values were calculated using the Spearman–Kärber method (Hierholzer and Killington, 1996).
VeroE6 cells were inoculated with a twofold serial dilution of recMARV to determine the optimal dilution of viruses producing distinct measurable plaques. After removal of the inoculum, cells were overlaid with a viscous polymer limiting virus diffusion in the cell culture supernatant (2% carboxymethyl cellulose in MEM/2% FCS). At 72 h p.i., the overlay was discarded and cells were fixed and inactivated with 4% paraformaldehyde in DMEM for 48 h. After removal of samples from the BSL-4 facility, foci of infected cells were detected by staining with a goat α-MARV serum and a secondary donkey α-goat IgG coupled with Rhodamine. Determination of the plaque size was carried out with the help of the fluorescence microscope Leica DMI6000B and the software Leica MMAF (Leica, Wetzlar, Germany).
Virus binding assay
RecMARV particles were purified from the supernatant of infected HUH-7 cells at 72 h p.i. by centrifugation over a sucrose cushion for 90 min at 40 000 r.p.m., 4°C in a Beckman ultracentrifuge using a SW60.1 rotor (Beckman Coulter, Palo Alto, CA, USA). Pellets were resuspended in DMEM; an aliquot was inactivated by incubation with SDS sample buffer [40% glycerol, 8% SDS, 200 mM Tris-HCl (pH 6.8), 0.4% bromphenol blue, 20% β-mercaptoethanol], boiled for 10 min and removed from the BSL-4 laboratory and the amount of viral protein was assessed by quantitative Western blot analysis detecting MARV NP. VeroE6 cells were incubated with recMARV or recMARVGPΔCD, normalized to the same protein amount, for 1 h on ice, the inoculum was removed, and cells were washed with DMEM and lysed with SDS sample buffer. After inactivation of the samples by boiling at 100°C for 10 min and removal from BSL-4 conditions, they were subjected to quantitative Western blot analysis evaluating the concentration of bound viral proteins by staining MARV NP. For quantification, the signal intensities of NP were normalized to signal intensities of cellular β-tubulin, reflecting the amount of VeroE6 cells used. Quantifications were conducted with help of the Bio-Rad Laboratories Chemidoc System (Bio-Rad, Munich, Germany).
Twofold serial dilutions of a goat α-MARV serum were incubated with recMARV particles with a MOI of 1 TCID50/cell for 1 h under frequent swirling at room temperature. Samples were then transferred to HUH-7 cells, inoculum removed after 1 h and replaced by fresh DMEM/3% FCS. At 16 h p.i., cells were fixed in 4% paraformaldehyde for 48 h, removed from the BSL-4 laboratory and subjected to quantitative immunofluorescence analysis. Immunostaining of infected cells was conducted as described in the section ‘indirect immunofluorescence analysis’ with a monoclonal α-MARV NP IgG. The amount of infected cells was quantified with the help of the fluorescence microscope Leica DMI6000B and the software Leica MMAF (Leica, Wetzlar, Germany). Number of infected cells induced by recMARV or recMARVGPΔCD in the absence of the serum was set to 0% infection inhibition.
Infectious VLP (iVLP) assay
The iVLP assay was performed as described by Wenigenrath et al. (2010).
SDS gel electrophoresis, Silver staining and immunoblot analysis
SDS gel electrophoresis and Western blot analysis were carried out as described elsewhere (Kolesnikova et al., 2004a). Samples from experiments conducted in the BSL-4 facility were inactivated by incubating with 1% SDS or SDS sample buffer, boiling for 10 min and transferring into fresh tubes before removal from the BSL-4 laboratory. After separation on 10% SDS polyacrylamide gels, gels were subjected to either Silver staining (PageSilver™ Silver Staining Kit; Fermentas, St. Leon-Rot, Germany) or Western blot analysis. Silver staining assay of SDS gels was performed following the manufacturer's instructions. In Western blot analysis, proteins were detected with the following antibodies: monoclonal α-MARV NP IgG, α-MARV VP40 IgG, α-MARV GP IgGs and monoclonal β-tubulin antibody. Quantification of Western blots was performed by staining proteins with secondary Alexa Fluor 680- or HRP-coupled antibodies using the software of the Odyssey Infrared Imaging System (LI-COR, Lincoln, NE, USA) or the Bio-Rad Laboratories Chemidoc System (Bio-Rad, Munich, Germany) respectively.
Indirect immunofluorescence analysis
Immunofluorescence analyses were carried out as previously described elsewhere (Mittler et al., 2007). For the detection of NP-induced inclusion bodies in recMARV-infected HUH-7 cells (after fixation of the cells with 4% paraformaldehyde in DMEM for 48 h), we used a monoclonal α-MARV NP antibody and a secondary FITC-coupled goat α-mouse antibody. Quantification of infected cells was carried out with the fluorescence microscope Leica DMI6000B and its respective software Leica MMAF (Leica, Wetzlar, Germany).
Immunoelectron microscopy analysis
Viral particles were purified at 72 h p.i. from the supernatant of infected HUH-7 cells by centrifugation over a 20% sucrose cushion (90 min at 40 000 r.p.m., 4°C in a Beckman ultracentrifuge using a SW60.1 rotor). The particles were resuspended in PBS, inactivated and fixed by incubation with 4% paraformaldehyde in DMEM for 48 h, removed from the BSL-4 laboratory and subsequently subjected to immunoelectron microscopic analysis. For quantification of the number of gold particles bound to GP incorporated into the envelope of recMARV, we used the method of microscopic analysis described by Mittler et al. (2007).
- Top of page
- Results and discussion
- Experimental procedures
- Conflict of interest
We thank Dirk Becker, Sonja Heck and Katharina Kowalski for expert technical assistance, as well as Gotthard Ludwig and Michael Schmidt for technical support with BSL-4 procedures. We gratefully acknowledge Verena Krähling for helpful discussions. This research was supported by grants of the DFG priority program 1175.
Conflict of interest
- Top of page
- Results and discussion
- Experimental procedures
- Conflict of interest
The authors state that there are no conflicts of interest regarding the publication of this article.
- Top of page
- Results and discussion
- Experimental procedures
- Conflict of interest
- 2007) Polybasic KKR motif in the cytoplasmic tail of Nipah virus fusion protein modulates membrane fusion by inside-out signaling. J Virol 81: 4520–4532. , , , , , and (
- 2005) VP24 of Marburg virus influences formation of infectious particles. J Virol 79: 13421–13433. , , , , and (
- 2002) Lipid raft microdomains: a gateway for compartmentalized trafficking of Ebola and Marburg viruses. J Exp Med 195: 593–602. , , , , , , et al. (
- 2012) Structural dissection of Ebola virus and its assembly determinants using cryo-electron tomography. Proc Natl Acad Sci USA 109: 4275–4280. , , , , , , et al. (
- 2011) Cryo-electron tomography of Marburg virus particles and their morphogenesis within infected cells. PLoS Biol 9: e1001196. , , , , , , et al. (
- 2006) Late budding domains and host proteins in enveloped virus release. Virology 344: 55–63. (
- 2011) Ebola virus entry requires the cholesterol transporter Niemann-Pick C1. Nature 477: 340–343. , , , , , , et al. (
- 2005) Endosomal proteolysis of the Ebola virus glycoprotein is necessary for infection. Science 308: 1643–1645. , , , , and (
- 2011) Small molecule inhibitors reveal Niemann-Pick C1 is essential for Ebola virus infection. Nature 477: 344–348. , , , , , , et al. (
- 2000) Crystal structure of the matrix protein VP40 from Ebola virus. EMBO J 19: 4228–4236. , , , , and (
- 2008) Filoviruses: interactions with the host cell. Cell Mol Life Sci 65: 756–776. , , and (
- 2010) Tsg101 is recruited by a late domain of the nucleocapsid protein to support budding of Marburg virus-like particles. J Virol 84: 7847–7856. , , , and (
- 1991) Glycosylation and oligomerization of the spike protein of Marburg virus. Virology 182: 353–356. , , , , and (
- 1995) Acylation of the Marburg virus glycoprotein. Virology 208: 289–297. , , , , and (
- 1992) Carbohydrate structure of Marburg virus glycoprotein. Glycobiology 2: 299–312. , , , , and (
- 2003) The matrix protein VP40 from Ebola virus octamerizes into pore-like structures with specific RNA binding properties. Structure 11: 423–433. , , , , , , et al. (
- 2012) Marburg virus glycoprotein GP2: pH-dependent stability of the ectodomain alpha-helical bundle. Biochemistry 51: 2515–2525. , , , and (
- 2011) The ESCRT pathway. Dev Cell 21: 77–91. , , and (
- 1996) Virus isolation and quantitation. In Virology Methods Manual. Kangro, B.W.M.a.H.O. (ed.). London: Academic Press Limited, pp. 36–38. , and (
- 2010) Oligomerization of Ebola virus VP40 is essential for particle morphogenesis and regulation of viral transcription. J Virol 84: 7053–7063. , , , , , , and (
- 2010) Biochemical and structural characterization of cathepsin L-processed Ebola virus glycoprotein: implications for viral entry and immunogenicity. J Virol 84: 2972–2982. , , , , , and (
- 2001) Ebola virus glycoprotein: proteolytic processing, acylation, cell tropism, and detection of neutralizing antibodies. J Virol 75: 1576–1580. , , , and (
- 2012) Crystal structure of the Marburg virus GP2 core domain in its post-fusion conformation. Biochemistry 51: 7665–7675. , , , , , , et al. (
- 2000) Ultrastructural organization of recombinant Marburg virus nucleoprotein: comparison with Marburg virus inclusions. J Virol 74: 3899–3904. , , , and (
- 2002) VP40, the matrix protein of Marburg virus, is associated with membranes of the late endosomal compartment. J Virol 76: 1825–1838. , , , and (
- 2004a) Multivesicular bodies as a platform for formation of the Marburg virus envelope. J Virol 78: 12277–12287. , , , and (
- 2004b) The matrix protein of Marburg virus is transported to the plasma membrane along cellular membranes: exploiting the retrograde late endosomal pathway. J Virol 78: 2382–2393. , , , and (
- 2007a) Budding of Marburgvirus is associated with filopodia. Cell Microbiol 9: 939–951. , , , and (
- 2007b) Basolateral budding of Marburg virus: VP40 retargets viral glycoprotein GP to the basolateral surface. J Infect Dis 196: S232. , , , and (
- 2009) Vacuolar protein sorting pathway contributes to the release of Marburg virus. J Virol 83: 2327–2337. , , , , , , and (
- 2012) Phosphorylation of Marburg virus matrix protein VP40 triggers assembly of nucleocapsids with the viral envelope at the plasma membrane. Cell Microbiol 14: 182–197. , , , , and (
- 2010) Establishment of fruit bat cells (Rousettus aegyptiacus) as a model system for the investigation of filoviral infection. PLoS Negl Trop Dis 4: e802. , , , , , , et al. (
- 2006) Conserved receptor-binding domains of Lake Victoria marburgvirus and Zaire ebolavirus bind a common receptor. J Biol Chem 281: 15951–15958. , , , , , , et al. (
- 2008) Structure of the Ebola virus glycoprotein bound to an antibody from a human survivor. Nature 454: 177–182. , , , , , and (
- 2010) Conserved motifs within Ebola and Marburg virus VP40 proteins are important for stability, localization, and subsequent budding of virus-like particles. J Virol 84: 2294–2303. , , , , , and (
- 2012) Anti-tetherin activities of HIV-1 Vpu and Ebola virus glycoprotein do not involve removal of tetherin from lipid rafts. J Virol 86: 5467–5480. , , , , , and (
- 1996) Budding of rabies virus particles in the absence of the spike glycoprotein. Cell 84: 941–951. , , and (
- 2012) Ebola virus entry requires the host-programmed recognition of an intracellular receptor. EMBO J 31: 1947–1960. , , , , , , et al. (
- 2007) Role of the transmembrane domain of marburg virus surface protein GP in assembly of the viral envelope. J Virol 81: 3942–3948. , , , , and (
- 2011) The cytoplasmic domain of Marburg virus GP modulates early steps of viral infection. J Virol 85: 8188–8196. , , , , and (
- 1998) Three of the four nucleocapsid proteins of Marburg virus, NP, VP35, and L, are sufficient to mediate replication and transcription of Marburg virus-specific monocistronic minigenomes. J Virol 72: 8756–8764. , , , and (
- 2000) Structural characterization and membrane binding properties of the matrix protein VP40 of ebola virus. J Mol Biol 300: 103–112. , , , , , , et al. (
- 1999) An analysis of features of pathogenesis in two animal models of Ebola virus infection. J Infect Dis 179 (Suppl. 1): S199–S202. , , and (
- 2007) Filoviridae: Marburg and Ebola viruses. In Fields Virology. Knipe, D. , and Howley, P. (eds). Philadelphia, PA: Lippincott Williams and Wilkins, pp. 1409–1439. , , and (
- 2001) Sorting of Marburg virus surface protein and virus release take place at opposite surfaces of infected polarized epithelial cells. J Virol 75: 1274–1283. , , , , , and (
- 1998) Requirement for a non-specific glycoprotein cytoplasmic domain sequence to drive efficient budding of vesicular stomatitis virus. EMBO J 17: 1289–1296. , , , , , and (
- 2006) Role of endosomal cathepsins in entry mediated by the Ebola virus glycoprotein. J Virol 80: 4174–4178. , , , , , and (
- 2000) Membrane association induces a conformational change in the Ebola virus matrix protein. EMBO J 19: 6732–6741. , , , , , and (
- 2004) Generation of Marburg virus-like particles by co-expression of glycoprotein and matrix protein1. FEMS Immunol Med Microbiol 40: 27–31. , , , , , , et al. (
- 2003a) Ebola virus matrix protein VP40 interaction with human cellular factors Tsg101 and Nedd4. J Mol Biol 326: 493–502. , , , , , , and (
- 2003b) Oligomerization and polymerization of the filovirus matrix protein VP40. Virology 312: 359–368. , , , , , and (
- 2010) Regulation of Marburg virus (MARV) budding by Nedd4. 1: a different WW domain of Nedd4. 1 is critical for binding to MARV and Ebola virus VP40. J Gen Virol 91: 228–234. , and (
- 2007) Interaction of Tsg101 with Marburg virus VP40 depends on the PPPY motif, but not the PT/SAP motif as in the case of Ebola virus, and Tsg101 plays a critical role in the budding of Marburg virus-like particles induced by VP40, NP, and GP. J Virol 81: 4895–4899. , , , , , and (
- 2000) Proteolytic processing of Marburg virus glycoprotein. Virology 268: 1–6. , , , , , , et al. (
- 2002) Roles for the cytoplasmic tails of the fusion and hemagglutinin-neuraminidase proteins in budding of the paramyxovirus simian virus 5. J Virol 76: 9284–9297. , , , and (
- 2004) Activation of a paramyxovirus fusion protein is modulated by inside-out signaling from the cytoplasmic tail. Proc Natl Acad Sci USA 101: 9217–9222. , , , and (
- 1998) The central structural feature of the membrane fusion protein subunit from the Ebola virus glycoprotein is a long triple-stranded coiled coil. Proc Natl Acad Sci USA 95: 6032–6036. , , , , and (
- 2010) Establishment and application of an infectious virus-like particle system for Marburg virus. J Gen Virol 91: 1325–1334. , , , , and (
- 1993) Marburg virus gene 4 encodes the virion membrane protein, a type I transmembrane glycoprotein. J Virol 67: 1203–1210. , , , , , and (