The tetherin protein (HM1.24, BST-2, CD317) was identified as a type I interferon-inducible host cellular factor, which restricts release of progeny HIV-1 particles from infected cells (Neil et al., 2008; Van Damme et al., 2008). Tetherin is counteracted by the HIV-1 accessory protein Vpu (viral protein u), which allows efficient HIV-1 release form tetherin-expressing cells (Neil et al., 2008; Van Damme et al., 2008). The antiviral action of tetherin is not limited to HIV-1; several recent studies found that tetherin restricts release of VLPs and progeny particles of several enveloped viruses, including members of the Retroviridae, Arenaviridae, Paramyxoviridae, Filoviridae, Rhabdoviridae and Herpesviridae families (Jouvenet et al., 2009; Mansouri et al., 2009; Sakuma et al., 2009; Pardieu et al., 2010; Radoshitzky et al., 2010; Weidner et al., 2010; Watanabe et al., 2011; Yondola et al., 2011) (Fig. 5). The broad spectrum antiviral action of tetherin is intimately linked to its unusual structural organization. Tetherin encodes a short cytoplasmic domain at its N-terminus, followed by a single transmembrane (TM) domain, an extracellular coiled-coil region and a C-terminal glycosylphosphatidylinositol (GPI) anchor (Kupzig et al., 2003). Thus, the protein spans the membrane twice and is able to insert one end into the cellular membrane and the other end into the viral envelope. By this mechanism, tetherin retains budding virions at the surface of the infected cell and inhibits their transmission to new target cells (Perez-Caballero et al., 2009; Hammonds et al., 2010). An artificially constructed tetherin, which displays the same topology as the original molecule, but is assembled from completely unrelated sequences, is able to inhibit virus spread as well, confirming that it is tetherin’s unusual architecture that is responsible for tethering virions to cells (Perez-Caballero et al., 2009).
Antagonism of tetherin is not limited to Vpu. HIV-2 uses its envelope (Env) protein to counteract tetherin (Le Tortorec and Neil, 2009), while the simian immunodeficiency virus (SIV) uses either the accessory protein Nef or Vpu or Env, depending on the origin of the virus (Gupta et al., 2009; Sauter et al., 2009). The Kaposi’s sarcoma-associated herpes virus encodes the E3 ubiquitin ligase K5 as a tetherin antagonist (Mansouri et al., 2009). For the EBOV, the GP1,2 serves as tetherin antagonist (Kaletsky et al., 2009). This ability of GP1,2 is conserved between the GPs of the different EBOV species (Zaire, Sudan, Côte d’Ivoire and Reston), the proposed species BEBOV and the second filoviral genus MARV (Kaletsky et al., 2009; Lopez et al., 2010; Radoshitzky et al., 2010; Kühl et al., 2011a,b).
The EBOV glycoprotein and HIV-1 Vpu employ different strategies to counteract tetherin
The EBOV glycoprotein (EBOV-GP1,2) is the only viral surface protein and mediates viral entry into the host cell that requires binding of GP1,2 to the endosomal membrane protein Niemann-Pick C1 (NPC1) (Carette et al., 2011; Côté et al., 2011). EBOV-GP1,2 is synthesized as a precursor protein, which is post-translationally cleaved by subtilisin-like proteases into its two subunits GP1 and GP2 (Volchkov et al., 1998). The large and heavily glycosylated, extracellular domain GP1 mediates attachment to the host cell; the smaller TM unit GP2 facilitates fusion of the viral envelope with the membrane of host cell endosomes (Takada et al., 1997; Wool-Lewis and Bates, 1998). A recent study demonstrated that GP1,2 also inactivates one of the cell`s innate defences against infection, the tetherin protein (Kaletsky et al., 2009). Vpu allows HIV-1 to evade tetherin by mediating cell surface downregulation and relocalization of tetherin into intracellular compartments (Van Damme et al., 2008). In addition, Vpu facilitates degradation of tetherin in lysosomal or proteasomal compartments (Douglas et al., 2009; Goffinet et al., 2009, 2010; Mangeat et al., 2009; Mitchell et al., 2009; Dubé et al., 2010). In contrast, no evidence for downregulation, relocalization away from the cell surface or degradation of tetherin was observed in the presence of EBOV-GP1,2 or MARV-GP1,2 (Kaletsky et al., 2009; Lopez et al., 2010; Radoshitzky et al., 2010; Kühl et al., 2011a). Furthermore, EBOV-GP1,2 is active against tetherin homologues from different monkeys, in accordance with the ability of EBOV to infect several non-human primates (Kühl et al., 2011a). In contrast, tetherin antagonism by Vpu is limited to tetherin of human, chimpanzee and gorilla origin, and these species are infected by HIV-1 and Vpu encoding SIV, respectively (Goffinet et al., 2009; Jia et al., 2009; McNatt et al., 2009). Thus, Vpu and EBOV-GP1,2 have a different specificity for tetherin and employ different strategies to counteract tetherin.
Tetherin interacts with the GP2 subunit of the EBOV-GP1,2
Four different proteins are expressed from the EBOV-GP gene: (i) The primary transcript from the GP gene is sGP, a soluble, secreted form of the GP (Volchkov et al., 1995; Sanchez et al., 1996), which has an anti-inflammatory property (Wahl-Jensen et al., 2005). (ii) A soluble Δ-peptide is released upon proteolytic processing of sGP by furin (Volchkova et al., 1999). (iii) The full-length, membrane-bound form GP1,2, which is incorporated into the virions, is produced by RNA editing of the primary transcript (Volchkov et al., 1995; Sanchez et al., 1996). (iv) Furthermore, a small soluble GP (ssGP) has recently been identified, but the function of this protein is at present unknown (Mehedi et al., 2011). In addition, the TNF-α converting enzyme induces shedding of GP1,2Δ from the surface of infected cells by cleaving GP1,2 near the TM domain (Dolnik et al., 2004). All different forms of the GP might play a role in combating the host’s immune system as, for example the shedded form of GP might be able to act as a decoy for neutralizing antibodies (Dolnik et al., 2008). However, neither sGP nor GP1,2Δ are able to counteract tetherin (Kaletsky et al., 2009). In addition, a GP1,2 mutant that is retained in the ER failed to counteract tetherin, suggesting that the correct localization of the EBOV-GP1,2 at sites of viral budding is crucial for tetherin antagonism (Kaletsky et al., 2009). Indeed, because of the expression of the GPI-anchor, tetherin is localized to lipid rafts, which are used by HIV-1 and EBOV as platforms for budding (Ono and Freed, 2001; Bavari et al., 2002).
Vpu and tetherin interact via their TM domains, and the interaction is critical for tetherin antagonism (McNatt et al., 2009). In contrast, the sequences of tetherin’s cytoplasmic tail (CT) and TM domain do not determine counteraction by EBOV-GP1,2. Thus, tetherin chimeras in which the TM region and the N-terminus of tetherin were exchanged against similar domains of the transferrin receptor type 1 (TfR) displayed antiviral activity and were counteracted by EBOV-GP1,2 (Lopez et al., 2010). Nevertheless, ZEBOV-GP1,2 is able to efficiently coimmunoprecipitate human tetherin (Kaletsky et al., 2009), and the GP2 subunit of EBOV-GP1,2 was shown to be critical for the interaction (Kühl et al., 2011a). Which domain of tetherin interacts with ZEBOV-GP2 remains to be determined, and, because of topological constraints, the TM unit and the CT are interesting candidates. However, the TM and CT have been shown to be irrelevant for tetherin counteraction by ZEBOV-GP1,2 (Lopez et al., 2010), and the importance of tetherin binding for tetherin inhibition by EBOV-GP1,2 remains to be determined.
The EBOV-GP1,2 might relocalize tetherin within the plasma membrane or interfere with the structural integrity of tetherin
Tetherin and VP40, which is essential for EBOV budding, colocalize at the plasma membrane and release of VP40-based VLPs is inhibited by tetherin (Jouvenet et al., 2009). It is thus possible that the EBOV-GP1,2 rescues the block to particle release by interfering with the integrity of tetherin at filoviral budding sites (Kühl et al., 2011a). Alternatively, the extracellular, heavily glycosylated EBOV-GP1 subunit might interfere with the formation of the ‘tetherin-clamp’ between the cellular and the viral membrane, because of steric hindrance. Finally, EBOV-GP1,2 might relocalize tetherin within the plasma membrane, thereby excluding it from membrane domains used by EBOV for budding. Indeed, it was observed that tetherin is excluded from plasma membrane sites positive for GP1,2 in EBOV-infected cells (Radoshitzky et al., 2010), lending support to the idea that EBOV-GP1,2 blocks tetherin’s antiviral action by inducing its mislocalization within the plasma membrane.
Regardless of the mechanism underlying tetherin counteraction by EBOV-GP1,2, it remains to be determined whether endogenous tetherin can reduce EBOV release from infected cells, as modest effects were observed in one study (Kühl et al., 2011a) but not in another applying a substantially higher multiplicity of infection (MOI) (Radoshitzky et al., 2010), and might thus restrict viral spread in the infected host. In addition, it will be interesting to determine whether African fruit bats, the potential natural reservoir of EBOV, encode a tetherin-like protein and whether this tetherin homologue restricts EBOV spread.
The IFITM 1, 2 and 3 were recently discovered as inhibitors of host cell entry of several enveloped viruses, including EBOV (Fig. 5). IFITMs 1, 2 and 3 are ubiquitously expressed in human cells and tissues upon exposure to type I (α) and type II (γ) IFN, and homologous proteins are present in many vertebrates (Siegrist et al., 2011). IFITM proteins were shown to play a role in early development, cell adhesion and control of cell growth (Siegrist et al., 2011). Their antiviral activity was discovered in a siRNA screen designed to identify host cell factors modulating FLUAV infection (Brass et al., 2009). IFITM3 was identified as a potent inhibitor of host cell entry of FLUAV and members of the family Flaviviridae, West Nile virus and Dengue virus serotype 2, but not hepatitis C Virus (Brass et al., 2009; Jiang et al., 2010), and the induction of IFITM3 expression was shown to be largely responsible for the blockade to FLUAV entry imposed by treatment of target cells with IFNs (Brass et al., 2009). IFITM1 and 2 were also shown to restrict viral infection although to a lower extent, and the antiviral activity of IFITMs was conserved between human proteins and murine orthologues (Brass et al., 2009). Thus, IFITMs are novel IFN-induced antiviral effector proteins, which could modulate viral spread in humans and animals.
IFITMs restrict viral entry into host cells
While antiviral activity of IFITMs was initially reported for influenza and flaviviruses (Brass et al., 2009; Jiang et al., 2010), subsequent studies showed that IFITMs inhibit entry of additional enveloped viruses such as vesicular stomatitis Indiana virus (VSIV), severe acute respiratory syndrome coronavirus (SARS-CoV) and filoviruses (Weidner et al., 2010; Huang et al., 2011), which, very much like FLUAV and flaviviruses, depend on endo-/lysosomal acidic pH for host cell entry. Inhibition of filoviruses and SARS-CoV was demonstrated employing lentiviral vectors pseudotyped with the respective viral GPs and with replication competent virus (Huang et al., 2011), but the inhibitory efficiency was modest. In contrast to FLUAV, IFITM1 showed the most prominent antiviral effects against replication competent EBOV and MARV, while inhibition by IFITM3 was less efficient (Huang et al., 2011). Entry of MARV and EBOV was reduced by the expression of several IFITM orthologues of mouse and chicken, although with different efficacies compared to the human proteins (Huang et al., 2011). Depletion of IFITM3 by shRNA was sufficient to rescue entry of FLUAV pseudotypes, whereas a knockdown of both, IFITM1 and IFITM3, was required to increase entry of EBOV and MARV pseudotypes (Huang et al., 2011). Finally, a different study also detected inhibition of HIV-1 infection (Lu et al., 2011), which does not depend on low pH, and it is at present unclear which structure or process shared by the viruses listed above for host cell entry is targeted by IFITMs.
Determinants of the antiviral activity of IFITM proteins
Domains and modifications important for antiviral activity of the IFITMs have been identified, although most studies were not conducted in the context of filovirus infection. Abrogation of S-palmitoylation of IFITM3 by mutation of crucial cysteine residues inhibited IFITM3 clustering in membrane compartments as well as the antiviral activity of IFITM3 against FLUAV (Yount et al., 2010). The sites of S-palmitoylation are highly conserved in members of the IFITM protein family, suggesting a general role in the antiviral activity. In addition, the sequence of the N-terminus is distinct for the different IFITMs as IFITM1 is lacking an N-terminal amino acid stretch present in IFITM2 and 3. Part of the N-terminus might be crucial for their different antiviral activity (Siegrist et al., 2011). Indeed, the characterization of IFITM1/IFITM3 chimeras revealed that sequences within the N- and C-terminus of IFITM3 are important for antiviral activity against VSIV (Weidner et al., 2010). Furthermore, Lu et al. (2011) could show that only IFITM-2 and IFITM-3 inhibit the HIV-1 life cycle at the step of viral entry, and that deletion of the N-terminal region abrogated this ability. IFITM1, in contrast, needs an intact intracellular domain to inhibit HIV-1 replication, whereas the N- and C-terminus are dispensable (Lu et al., 2011).
IFITM block cellular entry after transport of viruses into endosomal compartments
What is known about the mechanism underlying IFITM dependent inhibition of host cell entry of EBOV and other viruses? One possibility is that IFITMs interfere with receptor expression. However, IFITM proteins do not interfere with the level of surface expression of sialic acids and ACE2, the receptors for FLUAV and SARS-CoV, respectively (Brass et al., 2009; Huang et al., 2011). IFITM expression is compatible with FLUAV access to low pH compartments, and inhibition of SARS-CoV by IFITMs could be rescued by forcing the virus to fuse with the plasma membrane instead of an internal membrane (Huang et al., 2011). Thus, viruses might reach internal compartments in IFITM expressing cells in which fusion of viral and compartment membrane could normally occur, but membrane fusion might be blocked by IFITMs. One possibility could be that IFITMs interfere with the activity of cathepsins, pH-dependent endo-/lysosomal proteases essential for proteolytic activation of SARS-CoV and filoviruses in vitro (Chandran et al., 2005; Simmons et al., 2005). No appreciable decrease of cathepsin activity was observed in IFITM expressing cells (Huang et al., 2011). The endosomal membrane protein NPC1 has recently been reported as an essential host cell factor for EBOV entry (Carette et al., 2011; Côté et al., 2011). According to the model of Côté et al., NPC1 expression or activity is required for endosomal fusion after cathepsin-mediated processing of GP1,2. It is conceivable that IFITMs interfere with NPC1 expression or activity. Furthermore, it is possible that a step in the filovirus life cycle different from membrane fusion could be affected by IFITMs, as it was reported that IFITM2 and 3 inhibit HIV-1 entry, whereas IFITM1 acts later in the viral life cycle by suppressing Gag translation (Lu et al., 2011). Thus, the mechanisms underlying inhibition of filoviruses and other viruses remain to be defined, including the possibility that IFITMs require cellular cofactors to exert their antiviral effects.
In sum, IFITMs inhibit filovirus entry into host cells, and the level of constitutive IFITM expression might shape the choice of early target cells in filovirus infection.
Further research is needed to define the basal expression of IFITMs in filovirus target cells and to elucidate the mechanism by which the different IFITM proteins restrict filovirus infection.