SUPPLEMENT ARTICLE

Involvement of Vacuolar Sorting Pathway in Ebola Virus Release Independent of TSG101

Interaction Downloaded from https://academic.oup.com/jid/article/196/Supplement_2/S264/859891 by guest on 24 September 2021

Lynn S. Silvestri,1 Gordon Ruthel,1,a George Kallstrom,1,a Kelly L. Warfield,1 Dana L. Swenson,1 Timothy Nelle,1 Patrick L. Iversen,2 Sina Bavari,1 and M. Javad Aman1,b 1United States Army Medical Research Institute of Infectious Diseases, Fort Detrick, Maryland; 2AVI BioPharma, Corvallis, Oregon

Budding of Ebola virus (EBOV) particles from the plasma membrane of infected cells requires viral and host . EBOV virus matrix protein VP40 recruits TSG101, an ESCRT-1 (host cell endosomal sorting complex required for transport–1) complex protein in the vacuolar protein sorting (vps) pathway, to the plasma membrane during budding. Involvement of other vps proteins in EBOV budding has not been established. Therefore, we used VP40 deletion analysis, virus-like particle–release assays, and confocal microscopy to investigate the potential role of ESCRT-1 proteins VPS4, VPS28, and VPS37B in EBOV budding. We found that VP40 could redirect each protein from endosomes to the cell surface independently of TSG101 interaction. A lack of VPS4 adenosine triphosphatase activity reduced budding by up to 80%. Inhibition of VPS4 expression by use of phosphorodiamidite morpholino antisense oligonucleotides protected mice from lethal EBOV infection. These data show that EBOV can use vps proteins independently of TSG101 for budding and reveal VPS4 as a potential target for filovirus therapeutics.

Ebola virus (EBOV) particles exit host cells by budding is increasing evidence that many different viruses, in- from the plasma membrane, a process that involves cluding the filoviruses and retroviruses, use host pro- interaction between virus-encoded and cellular proteins teins normally involved in the vacuolar protein sorting [1, 2]. The viral matrix protein VP40 is a major com- (vps) pathway for the final steps of budding [8, 9]. This ponent of EBOV virions and is necessary for the for- requires movement of the vps proteins from multives- mation of the filamentous particles [3, 4]. Oligomeri- icular bodies to the plasma membrane, an event me- zation of VP40 at the plasma membrane is required for diated or performed by viral proteins. assembly of particles and for virus budding [5–7]. There The cellular protein TSG101 is involved in budding of both HIV and EBOV [5, 10, 11]. TSG101 functions in early steps of the vps pathway, in a complex of pro- Potential conflicts of interest: none reported. Financial support: Defense Threat Reduction Agency, Joint Science and teins termed “ESCRT-1” (host cell endosomal sorting Technology Office for Chemical and Biological Defense research grant (plan complex required for transport–1) [8]. This complex, 9Y0013_05_RD_B. 3 to S.B.). Supplement sponsorship is detailed in the Acknowledgments. which also includes proteins VPS28, VPS37B, and Opinions, interpretations, conclusions, and recommendations are those of the VPS4, sorts ubiquitinated proteins to promote their authors and are not necessarily endorsed by the US Army. Mention of trade names, commercial products, or organization does not imply endorsement by the US incorporation into multivesicular bodies in a budding Government. process. Recruitment of TSG101 to the plasma a G.R. and G.K. contributed equally to this work. b Present affiliation: Integrated BioTherapeutics Inc., Frederick, Maryland. membrane away from its normal site of function in Reprints or correspondence: Dr. M. Javad Aman, Integrated BioTherapeutics endosomes has been shown to occur via binding to Inc., 4539 Metropolitan Ct., Frederick, MD 21704 (javad@integratedbiothera peutics.com). EBOV VP40 [5, 10]. Interaction of VP40 with TSG101 The Journal of Infectious Diseases 2007;196:S264–70 requires a specific amino acid motif, or late domain (L 2007 by the Infectious Diseases Society of America. All rights reserved. domain) [9]. EBOV VP40 contains 2 overlapping 0022-1899/2007/19610S2-0022$15.00 DOI: 10.1086/520610 L domains: PPXY and PT/SAP motifs arranged as

S264 • JID 2007:196 (Suppl 2) • Silvestri et al. 7-PTAPPEY-13. Similar to the HIV-1 Gag protein, the VP40 L domain motif PTAP binds to TSG101 [10]. The PPEY motif binds to Nedd4 ligase [12]. L domain–containing viruses may recruit the entire ESCRT machinery to sites of virus budding through interactions with such molecules as TSG101 or Nedd4. In the final step of the vps pathway, the ATPase activity of VPS4 generates the energy needed for dis- sociation of the protein complex, allowing for subsequent rounds of sorting. A dominant-negative VPS4 mutant, lacking ATPase activity, can inhibit release of both HIV and EBOV [12, Downloaded from https://academic.oup.com/jid/article/196/Supplement_2/S264/859891 by guest on 24 September 2021 13]. EBOV VP40 may redirect other ESCRT-1 proteins from endosomes to the plasma membrane, but these interactions have not been fully characterized. We investigated the interactions of EBOV VP40 with the ESCRT proteins VPS4, VPS28, and VPS37B by examining the effect of VP40 L domain mutations on ESCRT protein local- ization and virus-like particle (VLP) release. Our results show that recruitment of VPS4, VPS28, and VPS37B to the plasma membrane was not dependent on VP40-TSG101 interaction. A lack of VPS4 ATPase activity did not affect its recruitment to the plasma membrane, but it did inhibit VLP release. Ad- ministration of phosphorodiamidite morpholino oligonucleo- tides (PMOs) to knock down VPS4 protected mice from lethal EBOV infection. The results presented here demonstrate that EBOV can use the vps pathway independently of TSG101 interaction and suggest a potential for VPS4 as a therapeutic target for the development of countermeasures against EBOV and, possibly, against other viruses.

MATERIALS AND METHODS

Cells, plasmids, mutagenesis, and transfections. 293T cells were maintained in Dulbecco’s MEM (DMEM) containing 10% fetal bovine serum at 37 C with 5% CO2. Cells were seeded 24 h before transfection. DNA was mixed with OptiMem (Invi- trogen) and transfected using 1 mL of Lipofectamine2000 (In- vitrogen), in accordance with the manufacturer’s instructions. Tetracysteine-tagged VP40 has been described elsewhere [5]. Mutations were made by use of the QuickChange Mutagenesis Kit (Stratagene). Mutagenic primers are listed in figure 1A. Plasmids encoding VPS28, VPS37B, and a green fluorescence protein (GFP) fusion of VPS4A were kindly provided by W. Sundquist (University of Utah, Salt Lake City). Figure 1. Analysis of multimerization and virus-like particle (VLP) re- lease of VP40 deletion mutants. A, Mutagenic primers used to create VLP-release assay. The assay was performed in 2–3 rep- VP40 deletion mutants. B, VP40 constructs transfected into 293T cells licates, as described elsewhere [14]. Briefly, supernatants were and tested by Western blot for multimerization. Samples were either centrifuged and adjusted to 1.5% Triton X-100. ELISA plates heated (H) or unheated (U) before loading. VP40 was detected using a were coated with mouse anti-VP40 (AE11; 1:100) in PBS, mixture of mouse anti-VP40 (AE11) antibody and rabbit anti-VP40 antibody. blocked with 5% milk/PBST (PBS Tween; 0.02%) for 2 h, then Oligomeric and monomeric forms of VP40 are indicated. C, VLP release measured in supernatants of transfected cells by ELISA specific to VP40. washed with PBST. One hundred microliters of supernatant The amount of VP40 present in supernatants from cells transfected with was added to each well and incubated at room temperature for wild-type VP40 was set at 100%. 4 h. Rabbit anti-VP40 and HRP-conjugated goat anti-rabbit

VPS Pathway in Ebola Virus Release • JID 2007:196 (Suppl 2) • S265 Research was conducted in compliance with the Animal Wel- fare Act and other federal statutes and regulations relating to animals and experiments involving animals and adhered to principles stated in the Guide for the Care and Use of Laboratory Animals [15]. The facility where this research was conducted is fully accredited by the Association for Assessment and Ac- creditation of Laboratory Animal Care International.

RESULTS

Analysis of VP40 deletion mutants in VLP release and Downloaded from https://academic.oup.com/jid/article/196/Supplement_2/S264/859891 by guest on 24 September 2021 oligomerization. EBOV VP40 forms oligomers at the plasma membrane, an important step for virus budding [5–7]. To search VP40 for regions important in oligomerization and bud- ding, VP40 mutants containing progressive 15-aa deletions were constructed. Mutants were tested for the ability to produce Figure 2. Late (L)–domain mutants in virus-like particle (VLP) release. VP40 oligomers and release of VLPs from transfected cells. All Plasmids encoding VP40 L domain mutants and glycoprotein (GP) were of the mutants, except those with deletions within aa 76–105, transfected into 293T cells. Culture supernatants (sup) were analyzed by were well expressed (figure 1B). Consistent with our previous ELISA specific for VP40. The level of VP40 indicates the amount of VLPs released from the cells. The sample containing tetracysteine-tagged wild- observations [5], deletion of the C-terminal 15 aa resulted in type VP40 was set at 100%. PPEA, 7-PTAPPEA-13 mutant; PTAL, 7-PTAL- a completely oligomerized protein (figure 1B). In contrast, de- PEY-13 mutant. letion of the N-terminal 30 aa did not significantly influence the degree of oligomerization, whereas deletion of aa 30–60 antibody were used to detect VP40. 3,3,5,5-Tetramethylben- significantly impaired oligomerization (figure 1B). As predicted zidine substrate was added, and absorption was read at 650 nm by structural data [16–18], no deletions in the core folded by use of a plate reader. region of the N-terminal domain (aa 61–195) were tolerated Confocal microscopy. 293T cells were transfected with the with respect to oligomerization. indicated plasmids in 2–3 replicates, incubated at 37C for 24 VLP release by each mutant was analyzed by ELISA specific h, then fixed with 4% buffered formalin. VP40 was detected for VP40 [14] (figure 1C). A reduced level of VLP release (40% with rabbit anti-VP40 antibody, and TSG101 and VPS37B were detected using anti-FLAG monoclonal antibody M1 (Sigma). VPS28 was detected using a rabbit polyclonal antibody. The VPS4 plasmid produces a GFP-VPS4 fusion protein for direct visualization. Secondary antibodies were Alexa 488 and Alexa 568 conjugated antibodies (Molecular Probes). Nuclei were la- beled with Hoechst stain (Molecular Probes). Images were ob- tained with a Bio-Rad 2000MP confocal/multiphoton system attached to a Nikon TE300 inverted microscope. Design of PMOs and EBOV challenge. PMOs were syn- thesized by AVI Biopharma. VPS4A PMO complements the region spanning the ATG start codon, a region 100% conserved between human and mouse. The control VPS4B PMO com- plements the region immediately following the ATG start codon of human VPS4B (positions +5 to +25), a region that is only 66% identical to the mouse sequence. The PMO sequences are as follows: VPS4A, 5-GGAGGGTTGACGTTGTCATTTC-3; and VPS4B, 5-GGAGGTTGGGCGAAGTGGATG-3. Groups of 10 C57BL/6 mice were injected intraperitoneally (ip) with 0.5 Figure 3. Effect of the VP40 aa 2–13 mutant on distribution of TSG101. mg of PMO VPS4A or VPS4B or with PBS alone, 24 h before 293T cells were transfected with plasmids encoding TSG101, wild-type and 24 and 48 h after challenge with an ip dose of 1000 pfu VP40, and the VP40 aa 2–13 mutant, as indicated. VP40 and TSG101 of mouse-adapted Zaire EBOV. Mice were monitored for signs were detected with rabbit anti-VP40 antibody and anti-FLAG M1 antibody, of disease for 28 days after infection. respectively.

S266 • JID 2007:196 (Suppl 2) • Silvestri et al. Downloaded from https://academic.oup.com/jid/article/196/Supplement_2/S264/859891 by guest on 24 September 2021

Figure 4. Image of 293T cells expressing VPS28 or VPS37B plus VP40 late domain mutants. Cells were transfected with plasmids expressing the indicated VP40 constructs and either VPS28 (A) or VPS37B (B). VPS28 and VPS37B are stained green, and VP40 is red. Nuclei are stained with Hoechst’s (blue). PTAL, 7-PTALPEY-13 mutant; wt, wild type. of wild type) was seen in the aa 2–15 mutant, in which the 7- of VLPs in the media to cell-associated VP40 levels were de- PTAPPEY-13 L domain motifs were deleted (figure 1B). In termined and expressed as percentages of VLP release relative contrast, deletions within aa 15–45 did not alter the ability of to wild-type VP40 plus GP. Figure 2 shows that deletion of 7- VP40 to bud from the cells (figure 1C). This is consistent with PTAPPEY-13 caused an ∼40% reduction in the amount of a proposed role for the vps pathway in VP40 vesicular release. VLPs released (aa 2–13 mutant). The mutants 7-PTALPEY-13 However, the data also suggest that VP40 can use alternative (PTAL), in which both L domains are mutated, and 7-PTAP- pathways for budding in the absence of L domain. PEA-13 (PPEA) slightly reduced the amount of released VLPs, Deletions in the rest of the N-terminal domain and almost by ∼10%–20%, compared with the control (figure 2). Thus, the entire C-terminal domain greatly inhibited VLP release (fig- although some reduction in VLP release was achieved by single- ure 1C). Previous results with the aa 91–105 mutant suggested amino-acid mutations within the L domain motifs, none of the that these amino acids constitute at least a portion of the AE11 mutations completely inhibited the release of VLPs. These data epitope (data not shown), making interpretation of the results suggest that the presence of GP moderately reduces the de- for this mutant problematic. Overall data from deletions in the pendence of VLP release on the L domain. N-terminal domain show a strong correlation between oligo- Necessity of VP40 L domains for localization of TSG101 to merization and the ability of VP40 to bud from the cells (com- the cellular membrane. EBOV VP40 recruits TSG101 to the pare figure 1B and 1C). It is apparent that C-terminal deletions plasma membrane and lipid rafts [5]. Because the VP40 aa 2– reduce budding without a noticeable effect on oligomerization 13 mutant had only a 40% reduction in VLP release, compared (figure 1B and 1C), most likely as a result of disruption of with wild-type VP40 (figure 2), we examined whether this mu- membrane binding, as has been suggested elsewhere [5, 6]. tant was able to recruit TSG101 to the cell surface. Cells were Reduction of VP40–glycoprotein (GP) VLP release by mu- transfected with each VP40 plasmid and a plasmid encoding tations in the L domains. Release of VLPs is enhanced by TSG101, stained, and analyzed by confocal microscopy. When coexpression of GP. To further analyze the role of the L do- TSG101 was expressed alone, it was found compartmentalized mains, constructs bearing L domain deletions and point mu- within the cytoplasm of the cell (figure 3). TSG101 colocalized tations were tested for the release of VP40-GP VLPs. 293T cells with full-length VP40 at the cell surface, whereas, in the pres- were transfected with a VP40 construct and a GP expression ence of the VP40 aa 2–13 mutant, TSG101 remained concen- plasmid and were analyzed for VLP release by ELISA. The ratios trated within endosomes (figure 3). These data show that the

VPS Pathway in Ebola Virus Release • JID 2007:196 (Suppl 2) • S267 uneven distribution pattern at the plasma membrane (figure 4B). When expressed with wild-type VP40, the majority of the VPS37B was found at the membrane and was more evenly distributed. However, VPS37B and VP40 were not clearly co- localized. The VP40 aa 2–13 mutant and PTAL also caused VPS37B to localize to the plasma membrane, with no significant colocalization of the proteins, similar to wild-type VP40. These findings suggest that VPS28 and VPS37B can be redirected to the plasma membrane by VP40 in the absence of TSG101 recruitment. Downloaded from https://academic.oup.com/jid/article/196/Supplement_2/S264/859891 by guest on 24 September 2021 Redistribution of VPS4A to the plasma membrane caused by VP40. Localization of VPS4A in presence and absence of VP40 was analyzed by confocal microscopy. VPS4A, when ex- pressed alone, had an intracellular globular pattern that most likely represented VPS4A associated with endosomes (figure 5). In the presence of wild-type VP40, however, a portion of VPS4A was redistributed to the plasma membrane, where VPS4A and VP40 colocalized. VPS4A that remained in the cytoplasm still appeared to be associated with endosomes. This distribution pattern of VPS4A was also observed in the presence of the VP40 aa 2–13 mutant. When coexpressed with mutants PTAL and PPEA, VPS4A was still distributed between the surface and endosomes, with a subset diffusely distributed throughout the Figure 5. VPS4 redistribution to the plasma membrane caused by VP40. cytoplasm. These data suggest that, similar to VPS28 and 293T cells were transfected with the indicated VP40 plasmids and either VPS37B, VPS4 can be recruited to plasma membrane inde- wild-type VPS4 or mutant VPS4 K173Q. VPS4 (green fluorescence protein pendently of TSG101. [GFP]) was detected on the cell surface in the presence of wild-type or mutant VP40 (red). Nuclei are stained blue. PPEA, 7-PTAPPEA-13 mutant; Because VPS4A is an ATPase, we sought to determine PTAL, 7-PTALPEY-13 mutant; wt, wild type. whether this enzymatic activity is required for VP40-induced redistribution of VPS4A using K173Q, a dominant-negative ATPase mutant of VPS4A. When expressed without VP40, movement of TSG101 to the plasma membrane requires the L VPS4A K173Q appeared as large globular structures within the domains and suggests that the 60% release of VLPs achieved with the aa 2–13 mutant (figure 2) was via a TSG101-inde- pendent pathway. Effect of VP40 on cellular localization of ESCRT proteins VPS28 and VPS37B. VPS28 and VPS37B bind to TSG101 within the ESCRT-I complex and are involved in the sorting of proteins on endosomal compartments [19]. To determine whether VP40 L domains mediate redistribution of VPS28 and VPS37B from endosomes to the cell surface, localization of VPS28 or VPS37B was analyzed in the presence and absence of VP40. In the absence of VP40, VPS28 was distributed throughout the cell, but it also appeared to be concentrated in some regions on the plasma membrane (figure 4A). In the presence of wild-type VP40 and the aa 2–13 mutant, VPS28 was slightly more concentrated at the surface than was observed when VPS28 was expressed alone. Membrane-bound VPS28 appeared to colocalize with VP40. The VP40 mutant PTAL was Figure 6. Effect of VPS4 ATPase-negative mutants on virus-like particle less effective in recruiting VPS28 to the plasma membrane than (VLP) release. VLP release was measured from 293T cells transfected wild-type VP40 or the aa 2–13 mutant. with plasmids encoding wild-type VP40, GP, and wild-type or ATPase- VPS37B, when expressed alone, had a punctuate appearance negative VPS4 mutants. After 48 h, the amount of VP40 was determined within the cytoplasm and, to a lesser degree, was seen in an by ELISA from cell lysates and supernatants (sup).

S268 • JID 2007:196 (Suppl 2) • Silvestri et al. tively, indicating that the ATPase activity of VPS4A is needed for efficient VLP release with an impact greater than that caused by deletion of the L domains. Protection of mice from EBOV infection by knockdown of VPS4A gene expression. We have recently demonstrated that antisense PMOs targeted against several EBOV can inhibit viral replication and protect mice and monkeys from lethal infection [20]. To examine the role of VPS4A in viral replication in vivo, we constructed a PMO against a 100% conserved amino acid region that spans the start codon of human and mouse Downloaded from https://academic.oup.com/jid/article/196/Supplement_2/S264/859891 by guest on 24 September 2021 VPS4. We first examined the ability of this PMO to inhibit the expression of VPS4A in 293T, Vero E6, and dendritic cells. Treatment of human myeloid dendritic cells with this PMO for 24 h completely blocked the expression of VPS4A (figure 7A), and this blockage was sustained for at least 72 h (data not shown). Interestingly, blockage of the ATG start codon forced the expression of a truncated form of VPS4A, presumably re- sulting from initiation of translation at an alternative down- stream ATG codon, most likely at position 595 (“Alt. ORF” in figure 7A). Similar results were also obtained in 293T cells and Vero E6 cells (data not shown). These data demonstrated that PMOs could potently inhibit the expression of VPS4 in primary Figure 7. VPS4 gene knockdown protects mice against Ebola virus cells. (EBOV) challenge. A, Western blot of lysates from human myeloid dendritic We then sought to examine the effect of this PMO in a mouse cells treated with a phosphorodiamidite morpholino oligonucleotide(PMO) against VPS4A. The position of the endogenous full-length VPS4A and lethal challenge model. We tested PMOs for 2 homologs of the product of the VPS4A alternate reading frame (Alt. ORF) are indicated. VPS4 found in both mice and humans: VPS4A and VPS4B. VPS4A was detected using rabbit anti-VPS4. B, Protection of mice from The mice that received PBS began to die of infection on the lethal EBOV challenge with PMO treatment. Mice were treated with the seventh day after challenge. More than half were dead by the indicated PMOs 24 h before infection with a lethal dose of mouse-adapted eighth day, and only 20% survived. In contrast, mice that re- Zaire EBOV and again at 24 h and 48 h after challenge. Control mice ceived the mouse/human VPS4A PMO had a delayed time to received PBS. death at day 8, and 70% of the mice survived the challenge. Even those mice that were given the human VPS4B PMO (de- cytoplasm (figure 5). Coexpression of wild-type VP40 caused signed for a region of human VPS4B with just 66% homology clear accumulation of VPS4A K173Q at the plasma membrane to mouse sequence) had a 50% survival rate. Scrambled-se- (figure 5). These results show that the enzymatic activity of quence PMOs did not protect mice against lethal EBOV in- VPS4A is not required for redirecting VPS4A to the plasma fection ([20] and data not shown). These data show that PMO membrane by VP40. inhibitors against VPS4 can be protective against lethal EBOV Requirement of enzymatic activity of VPS4A for efficient infection, indicating the importance of this molecule in regu- VLP release. Although our results show that the membrane lation of EBOV replication in vivo. recruitment of VPS4 by VP40 is independent of its enzymatic activity, a previous report suggested that the VPS4 enzymatic DISCUSSION activity plays a role in the release of virions [12]. We examined this question in our VLP-release assay by transfecting cells with In the present study, we demonstrated the ability of VP40 to constructs encoding VP40 and GP plus wild-type VPS4A or utilize the vps machinery to support its budding at the plasma enzymatically inactive VPS4A mutants K173Q and E228Q, [13]. membrane in a manner independent of TSG101. VP40 mutants The data are expressed as a ratio of the amount of VP40 de- incapable of recruiting TSG101 to the plasma membrane were tected in the media supernatant to the amount of VP40 con- able to redirect other components of the vps pathway, such as tained in the cell lysates. The level of VLP release obtained by VPS4, VPS28, and VPS37B, to the site of budding and to sup- transfecting VP40 and GP was set at 100%. Figure 6 shows that port VLP release. Our studies also demonstrate that the ATPase expression of exogenous wild-type VPS4A did not have an VPS4 can be used as a potential therapeutic target against EBOV impact on VLP release. However, mutants K173Q and E228Q and, possibly, against other vps-using viruses. reduced the amount of released VLPs by 60% to 80%, respec- Previous studies have suggested a key role for TSG101 in

VPS Pathway in Ebola Virus Release • JID 2007:196 (Suppl 2) • S269 linkage between the vps pathway and EBOV VP40 [5, 10] as References a result of TSG101 binding to the PTAP motif of VP40. How- 1. Aman MJ, Bosio CM, Panchal RG, Burnett JC, Schmaljohn A, Bavari ever, our findings demonstrate that this linkage is not entirely S. Molecular mechanisms of filovirus cellular trafficking. Microbes In- dependent on TSG101. In fact, the significant residual VLP fect 2003; 5:639–49. release by L domain–deficient mutants was shown to still be 2. Geisbert TW, Jahrling PB. Differentiation of filoviruses by electron microscopy. Virus Res 1995; 39:129–50. dependent on the VPS pathway. In the absence of TSG101 3. Bavari S, Bosio CM, Wiegand E, et al. Lipid raft microdomains: a recruitment to the plasma membrane, VP40 was able to redirect gateway for compartmentalized trafficking of Ebola and Marburg vi- other vps proteins to the plasma membrane. When coexpressed, ruses. J Exp Med 2002; 195:593–602. 4. Timmins J, Scianimanico S, Schoehn G, Weissenhorn W. Vesicular VP40 translocated VPS28, VPS37B, and VPS4 to the site of release of ebola virus matrix protein VP40. Virology 2001; 283:1–6. budding at the membrane. This effect was most pronounced

5. Panchal RG, Ruthel G, Kenny TA, et al. In vivo oligomerization and Downloaded from https://academic.oup.com/jid/article/196/Supplement_2/S264/859891 by guest on 24 September 2021 with VPS4, which was entirely cytoplasmic when expressed raft localization of Ebola virus protein VP40 during vesicular budding. alone but colocalized with VP40 at the plasma membrane when Proc Natl Acad Sci USA 2003; 100:15936–41. 6. Ruigrok RW, Schoehn G, Dessen A, et al. Structural characterization coexpressed with VP40. This effect was completely independent and membrane binding properties of the matrix protein VP40 of Ebola of both L domains of VP40. Although the VP40-dependent virus. J Mol Biol 2000; 300:103–12. translocation of VPS4 to the plasma membrane did not require 7. Scianimanico S, Schoehn G, Timmins J, Ruigrok RH, Klenk HD, Weis- the enzymatic activity of VPS4, this enzymatic activity was re- senhorn W. Membrane association induces a conformational change in the Ebola virus matrix protein. EMBO J 2000; 19:6732–41. quired for VLP release. 8. Pornillos O, Garrus JE, Sundquist WI. Mechanisms of enveloped RNA The profound effect on VLP release caused by expression of virus budding. Trends Cell Biol 2002; 12:569–79. dominant-negative VPS4, as well as previous reports showing a 9. Freed EO. Viral late domains. J Virol 2002; 76:4679–87. similar effect on replication of EBOV [12], prompted us to ex- 10. Martin-Serrano J, Zang T, Bieniasz PD. HIV-1 and Ebola virus encode small peptide motifs that recruit Tsg101 to sites of particle assembly amine the potential of this enzyme as a therapeutic target in a to facilitate egress. Nat Med 2001; 7:1313–9. lethal mouse model of EBOV infection. Specific knockdown of 11. VerPlank L, Bouamr F, LaGrassa TJ, et al. Tsg101, a homologue of VPS4A and partial knockdown of VPS4B genes in mice resulted ubiquitin-conjugating (E2) enzymes, binds the L domain in HIV type 1 Pr55(Gag). Proc Natl Acad Sci USA 2001; 98:7724–9. in partial protection against EBOV challenge. These data validate 12. Licata JM, Simpson-Holley M, Wright NT, Han Z, Paragas J, Harty VPS4 as a promising target for filovirus therapeutics. RN. Overlapping motifs (PTAP and PPEY) within the Ebola virus VP40 These data show, for the first time, that VP40 can use mech- protein function independently as late budding domains: involvement anisms other than those requiring the known L domains to of host proteins TSG101 and VPS-4. J Virol 2003; 77:1812–9. 13. Garrus JE, von Schwedler UK, Pornillos OW, et al. Tsg101 and the hijack the vps pathway and use it for efficient budding. How- vacuolar protein sorting pathway are essential for HIV-1 budding. Cell ever, the molecular mechanism by which L domain mutants 2001; 107:55–65. of VP40 interact with VPS4 remains to be determined. In sum- 14. Kallstrom G, Warfield KL, Swenson DL, et al. Analysis of Ebola virus mary, our findings suggest the existence of a novel pathway for and VLP release using an immunocapture assay. J Virol Methods 2005; 127:1–9. recruitment of the cellular protein sorting machinery by EBOV 15. Committee on the Care and Use of Laboratory Animals of the Institute for efficient budding. It remains to be seen whether such a of Laboratory Animals Resources Commission of Life Sciences, Na- pathway is also operational in other viruses. If so, targeting the tional Research Council. Guide for the care and use of laboratory VPS4 enzymatic activity by specific inhibitors or gene knock- animals. Washington, DC: National Academy Press, 1996. 16. Gomis-Ruth FX, Dessen A, Timmins J, et al. The matrix protein VP40 down by antisense or RNA interference may lead to the de- from Ebola virus octamerizes into pore-like structures with specific velopment of broad-spectrum antivirals. RNA binding properties. Structure 2003; 11:423–33. 17. Dessen A, Forest E, Volchkov V, Dolnik O, Klenk HD, Weissenhorn W. Crystallization and preliminary X-ray analysis of the matrix protein from Ebola virus. Acta Crystallogr D Biol Crystallogr 2000; 56:758–60. Acknowledgments 18. Dessen A, Volchkov V, Dolnik O, Klenk HD, Weissenhorn W. Crystal We thank Wesley Sundquist (University of Utah) for providing anti- structure of the matrix protein VP40 from Ebola virus. Embo J 2000;19: VPS4 and anti-VPS28 polyclonal antibodies. 4228–36. Supplement sponsorship. This article was published as part of a sup- 19. Stuchell MD, Garrus JE, Muller B, et al. The human endosomal sorting plement entitled “Filoviruses: Recent Advances and Future Challenges,” complex required for transport (ESCRT-I) and its role in HIV-1 bud- sponsored by the Public Health Agency of Canada, the National Institutes ding. J Biol Chem 2004; 279:36059–71. of Health, the Canadian Institutes of Health Research, Cangene, CUH2A, 20. Warfield KL, Swenson DL, Olinger GG, et al. Gene-specific counter- Smith Carter, Hemisphere Engineering, Crucell, and the International Cen- measures against Ebola virus based on antisense phosphorodiamidate tre for Infectious Diseases. morpholino oligomers. PLoS Pathog 2006; 2:e1.

S270 • JID 2007:196 (Suppl 2) • Silvestri et al.