Tetherin inhibits the release of some enveloped viruses including HIV-1 in the absence of Vpu (Neil et al., 2008
) by bridging cellular and viral membranes (Perez-Caballero et al., 2009
). Our structural analyses demonstrate that the complete extracellular domain of tetherin adopts an extended conformation that spans a maximal distance of 170
. More than halve of this is provided by a 90
-long parallel coiled-coil. The low resolution model based on X-ray scattering data indicates a slightly bent orientation of the N-terminal domain with respect to the coiled-coil. This might be due to flexibility within the region (residues 79 to 89) connecting the N-terminal and coiled-coil domains as documented by the sensitivity to proteolysis and the absence of an ordered structure for residues 80 to 88. In addition, this region permits the insertion of a HA-tag epitope without substantial loss of tetherin function.
The extracellular rod-like structure must be connected to the TMR via three N-terminal residues and to the GPI anchor via one C-terminal residue. Consequently it is unlikely that tetherin is positioned parallel between cellular and viral membranes which would tether virions quite close to the plasma membrane. The distance between both membranes would be less than 3 to 5 nm. Thin section electron microscopy images support a larger distance between virions and the plasma membrane (Neil et al., 2008
; Perez-Caballero et al., 2009
). Thus, upon virion tethering, the dimeric tetherin rod has most likely one end anchored in the plasma membrane and the other one in the virus membrane, as hypothesized (Perez-Caballero et al., 2009
The length of the rod and its rather rigid structure in solution suggest that it functions as a molecular ruler that connects two entities via a 170
distance. The importance of the spacer function is documented by our mutagenesis studies of the coiled-coil region that abrogate HIV-1 retention and by the loss of function upon deletion of the coiled-coil (Perez-Caballero et al., 2009
). Such a molecular ruler function might be also required to connect adjacent lipid rafts within the plasma membrane (Kupzig et al., 2003
Single cysteine mutations do not affect tetherin function dramatically, but mutagenesis of all three cysteines lead to a complete loss of function during HIV-1 release (Andrew et al., 2009
; Perez-Caballero et al., 2009
) although the mutant is still active during Lassa and Marburg virus VLP release (Sakuma et al., 2009b
). We show that the presence of disulfide bonds is crucial for the stability of the extracellular domain, since the melting temperature drops to 35°C (tetherin(47–159)) under reducing conditions. The low stability of tetherin under reducing conditions is most likely due to instability of the coiled-coil, which shows an even lower Tm
under reducing conditions. The coiled-coil contains a number of coiled-coil destabilizing residues occupying central heptad positions. These positions do not follow classical knobs into holes packing but instead loosen the coiled-coil pitch and induce an expansion of its radius. Although the coiled-coil region contains two inter-helical salt bridges and one inter-helical hydrogen bond, which are employed to stabilize coiled-coils (Burkhard et al., 2002
), the solvent exposure of the apolar heptad positions (Li et al., 2003
) might contribute to the dramatic instability of the coiled-coil in the absence of the disulfide bond. Together these structural features explain the low Tm
in the absence of stabilizing disulfide bonds.
This mode of labile coiled-coil interactions might serve two functions. First, tetherin’s cellular function might involve the formation of heterodimers with a yet unknown ligand employing its coiled-coil to form more stable dimers. Secondly, the weak coiled-coil interactions together with the stabilizing disulfide bonds generate a dynamic structure, which permits disassembly and reassembly of the coiled-coil during dynamic processes. The latter function is in agreement with the presence of similar dynamic or destabilizing coiled-coil features in myosin (Blankenfeldt et al., 2006
; Li et al., 2003
), tropomyosin (Brown et al., 2001
) and the streptococcus M1 protein (McNamara et al., 2008
) that have been suggested to be important for their mode of action.
Despite its instability in vitro
, we demonstrate the importance of the coiled-coil in vivo
. Mutagenesis of N- and C-terminal sets of highly conserved heptad positions eliminates the tethering function, although the mutant proteins are still expressed on the plasma membrane. This indicates that the spacer function provided by proper coiled-coil formation is essential for tethering. We also identified a third set of residues within the highly conserved N-terminal extracellular region that are functionally required. Mutations within the stretch of residues 48 to 59 have no effect on tethering, whereas changes within residues 62 to 73 lead to a loss of the tethering function. Again both mutant proteins are expressed on the plasma membrane and the extracellular domains form dimers in vitro
. Since the set 4 mutant eliminates the glycosylation site at Asn65 and shows a less complex expression pattern as wild-type tetherin, we tested whether changes in posttranslational modification are responsible for loss of tetherin function. Although the expression pattern of Asn65Gln resembles that of the set 4 mutant, it only shows slightly reduced HIV-1 retention activity, consistent with previous findings reporting no effect on HIV-1 retention of single and double glycoslylation mutants of tetherin (Andrew et al., 2009
). This indicates that mutagenesis of this N-terminal region (set 4) either affects its spacer function or eliminates an important docking site, possibly for self-assembly. Although Perez-Caballero reported that the N-terminus can be replaced by a similar region derived from the transferrin receptor and the coiled-coil can be replaced by the dystrophia myotonica protein kinase coiled-coil, it is important to note that the activity of art-tetherin is ~10-fold lower (Perez-Caballero et al., 2009
). In contrast our data clearly demonstrate that the N-terminal domain and the dynamic features of the coiled-coil of tetherin are essential for HIV-1 retention.
Based on our structural analysis we propose the following interplay between the elongated shape and the conformational flexibility of tetherin. Although we do not know at which stage of assembly tetherin enters the virion membrane, it is likely that it is present from the beginning of assembly starting from lipid rafts. Since virus assembly and budding presents a dynamic process, tetherin cannot remain too rigid. The coiled-coil instabilities thus permit a certain degree of flexibility for the tetherin dimers to diffuse laterally into the budding site with four membrane anchors, while maintaining the strict distance between the membrane anchors. The conformational flexibility, which entails most likely opening and reassembly of the coiled-coil, is facilitated by the presence of the disulfide bonds. Consequently dimer dissociation and re-stabilization do not interfere with the dynamic process of virus assembly and budding while remaining anchored in the newly formed viral membrane and maintaining its spacer function. Furthermore the elongated rod-like structure might be involved in self-assembly as supported by the punctuate appearance of tetherin in the plasma membrane. Such clustering might require an intact N-terminal region, which could cluster tetherin around the membrane neck of a budding virion, consistent with the accumulation of tetherin at HIV-1 budding sites (Habermann et al., 2010
). This would ensure that at least one or several tetherin dimers can efficiently insert into the viral membrane to render the system efficient.