|Home | About | Journals | Submit | Contact Us | Français|
CD317/Bst-2/tetherin is a host factor that restricts the release of human immunodeficiency virus type 1 (HIV-1) by trapping virions at the plasma membrane of certain producer cells. It is antagonized by the HIV-1 accessory protein Vpu. Previous light microscopy studies localized CD317 to the plasma membrane and the endosomal compartment and showed Vpu induced downregulation. In the present study, we performed quantitative immunoelectron microscopy of CD317 in cells producing wild-type or Vpu-defective HIV-1 and in control cells. Double-labeling experiments revealed that CD317 localizes to the plasma membrane, to early and recycling endosomes, and to the trans-Golgi network. CD317 largely relocated to endosomes upon HIV-1 infection, and this effect was partly counteracted by Vpu. Unexpectedly, CD317 was enriched in the membrane of viral buds and cell-associated and cell-free viruses compared to the respective plasma membrane, and this enrichment was independent of Vpu. These results suggest that the tethering activity of CD317 critically depends on its density at the cell surface and appears to be less affected by its density in the virion membrane.
To counteract infection with pathogens, cells use a variety of strategies that can be regulated by signaling events. CD317 (Bst-2, HM1.24, tetherin) was recently discovered as an interferon-inducible cellular factor that inhibits the release of human immunodeficiency virus type 1 (HIV-1). In cells that express CD317, HIV-1 strains lacking the accessory protein Vpu accumulate at the cell surface and are not efficiently shed. This phenotype can be induced by the expression of CD317 in cells lacking significant levels of the protein and is overcome by Vpu-containing wild-type (wt) HIV-1 (24, 30; reviewed in references 2 and 25).
CD317 appears to block HIV-1 release by tethering newly budded virions to the cell surface, hence the name tetherin. It is therefore expected to localize to the plasma membrane and possibly to the virion envelope, at least in virions derived from Vpu-minus strains. Indeed, by light microscopy CD317 was detected at the plasma membrane with enrichment in cholesterol-rich microdomains and also at an intracellular site. Intracellular CD317 exhibited partial colocalization with markers of the trans-Golgi network (TGN) and early and recycling endosomes but not significantly of the late endosome (15, 19). Light microscopy and biochemical studies suggested that CD317 constantly cycles between these cellular membranes and is internalized from the cell surface by clathrin-mediated endocytosis (19, 27).
Vpu acts as a viral antagonist of CD317. Several studies showed that Vpu, either expressed alone or upon wt-HIV-1 infection, specifically reduced cell surface levels of CD317 (6, 21, 30; reviewed in reference 2), whereas Vpu-mediated relief of CD317 restriction independent of CD317 downregulation was also observed in one study (22). Vpu appears to reduce total levels of CD317, at least in most cell types (10, 12, 22, 30). Several recent studies reported that the downregulation of CD317 from the cell surface depends on the human β-transducin repeat-containing protein (β-TrCP), a protein that is also responsible for the Vpu-dependent proteasomal degradation of CD4 via the ERAD pathway (6, 21). For CD317, however, downregulation was reported to be blocked either by inhibitors of the proteasome (10, 12, 17) or by inhibitors of lysosomal degradation (6, 21), and the precise pathway of Vpu action is currently not understood. It appears likely, however, that Vpu mediates internalization and/or degradation of CD317 such that it is no longer available for its tethering, antiviral, activity at the cell surface (reviewed in reference 2).
The large clusters of virions at the cell surface, observed with Vpu-minus virus (14), suggest that CD317 could also be present in the virus envelope, tethering extracellular particles to the cell surface and to each other. Whether virions contain significant amounts of CD317 is not clear at present, however. One study suggested that CD317 is excluded from both wt and delVpu virions, as assessed by Western blot analyses after shearing of surface-bound virus (22). Another study reported incorporation of wt and defective CD317/tetherin into viruslike particles, when overexpressed in 293T cells (26).
In the present study, we applied quantitative immunoelectron microscopy (EM) to determine CD317 distribution in uninfected and infected cells and in virus particles. We used HeLa cells that were previously shown to constitutively express CD317 and an antibody that enabled us to localize the endogenous protein by EM. The majority of CD317 localized to the plasma membrane in uninfected cells and was significantly redistributed to the endosomal compartment upon wt-HIV infection. Unexpectedly, the protein was enriched in both wt and Vpu-minus virions compared to the plasma membrane, suggesting that the CD317 levels at the cell surface are critical for the inhibition of virus release. These results are discussed with respect to the possible way in which CD317 counteracts HIV-1 infection.
HeLa, 293T, and TZM-bl cells (31) were cultured in Dulbecco modified Eagle medium supplemented with 10% heat-inactivated fetal calf serum, penicillin, and streptomycin. Proviral plasmids pNL4-3wt (BH10 Env) and pNL4-3-delVpu (BH10 Env) were from Valerie Bosch (DKFZ, Heidelberg, Germany) (3). A derivative lacking both Vpu and Nef was constructed by inserting the respective fragment from a plasmid carrying a nonsense mutation in codon 13 of nef into pNL4-3-delVpu. The plasmids encoding hemagglutinin (HA)-tagged CD317 or CD317 in which the cysteine residues at position 53, 63, and 91 were replaced by alanine residues (C3A) have been described previously (1, 10).
For production of HIV-1 for EM analysis, HeLa or 293T cells were transfected with proviral plasmids. At 42 to 48 h posttransfection, supernatants were harvested, cleared by a brief centrifugation, filtered through 0.45-μm-pore-size filters, and purified through a 20% (wt/vol) sucrose cushion in phosphate-buffered saline (PBS) in the ultracentrifuge (24,000 rpm, 2 h, 4°C). Pellets were resuspended in PBS and repelleted in the ultracentrifuge (44,000 rpm, 1 h, 4°C). Finally, pellets were resuspended in 0.1 M PHEM buffer (60 mM PIPES, 25 mM HEPES, 2 mM MgCl2, 10 mM EGTA [pH 6.9]), fixed, and processed for cryo-sectioning as detailed below.
The antibody to CD317 has been described previously (22). Rabbit and sheep polyclonal antisera to HIV-1 capsid (CA) have been raised against recombinant protein. Antibodies to cellular marker proteins were from the following sources: anti-transferrin receptor (anti-TfR) monoclonal from Zymed (Invitrogen, Karlsruhe, Germany), rabbit anti-EEA1 was kindly provided by Arwyn Jones (University of Cardiff, Cardiff, United Kingdom), monoclonal to CD63 was from Sanquin (Amsterdam, The Netherlands), anti-LAMP-1 clone H4A3 monoclonal was from DSHB (Iowa City, IA), anti-Golgin97 monoclonal CDF4 from Molecular Probes (Invitrogen, Karlsruhe, Germany), rabbit anti-mitogen-activated protein kinase ERK2 from Santa Cruz Biotechnology (Heidelberg, Germany), and rabbit anti-GM130 was kindly provided by Antonella De Matteis (Mario Negri Sud, Santa Maria Imbaro, Italy). For labeling with mouse monoclonal antibodies, a bridging rabbit anti-mouse antibody from Cappel (MP Biomedicals, Heidelberg, Germany) was used.
For Western blot analyses, HeLa or 293T cells were transfected with proviral plasmids (or in combination with CD317 expression plasmids) and medium, and the cells were then harvested at 30 h posttransfection. Virus particles were concentrated by ultracentrifugation (44,000 rpm, 1 h, 4°C) through a 20% (wt/vol) sucrose cushion in PBS and either analyzed directly after lysis in sample buffer or further purified on Opti-Prep gradients as detailed elsewhere (4). Samples were analyzed by Western blotting with antisera against HIV-1 CA, CD317, and the cellular protein ERK2. Secondary antibodies were conjugated to Alexa 700/800 fluorescent dyes for detection by Odyssey infrared imaging system (Li-Cor Biosciences, Bad Homburg, Germany). The amount of HIV-1 CAp24 antigen in the supernatant was determined by an in-house antigen enzyme-linked immunosorbent assay (ELISA). The infectivity in the culture medium of virus-producing cells was determined 48 h after infection in a standardized 96-well titration assay on TZM-bl cells by luminometric analysis of firefly luciferase activity (7).
HeLa and 293T cells were transfected using FuGENE HD (Roche Diagnostics) with the indicated proviral DNA. At the indicated times posttransfection the cells were fixed and processed either for resin embedding or for cryosectioning as described previously (32). Briefly, cells were fixed with 4% paraformaldehyde-0.1% glutaraldehyde in 0.1 M PHEM for at least 1 h at room temperature and then washed with 50 mM glycine in 0.1 M PHEM, followed by 100 mM sodium cacodylate (pH 7.4). Cells were scraped in 1% bovine serum albumin in 100 mM cacodylate, pelleted washed with 100 mM cacodylate, and postfixed with 2.5% glutaraldehyde (GA) for 1 h. Pellets were washed in 100 mM cacodylate, postfixed with reduced osmium (1% OsO4, 1.5% potassium hexacyanoferrate in 100 mM cacodylate), and stained for 16 h with 1% aqueous uranyl acetate. Cells were gradually embedded in epoxy resin after dehydration with a series of increasing concentrations of acetone. Ultrathin sections were poststained with lead citrate. For immuno-EM, cells were fixed and processed as detailed earlier (8), and thawed 60-nm cryosections were labeled as described previously (11). Double labeling was essentially performed as described by Slot et al. (28). All sections were examined with a Zeiss EM10 transmission electron microscope, and images were obtained using a Gatan MultiScan camera and digital micrograph software and further processed by using Adobe Photoshop CS3. For quantification of the CD317 labeling density per membrane length or per area, random pictures were taken at final magnifications of ×46,000 or ×58,000. The images were overlaid with a grid lattice with a distance between the grid lines of 20 mm at a ×46,000 or a ×58,000 magnification; the intersections of the plasma membrane or the defined area with the grid were counted, and the labeling density per μm of membrane length or per μm2 of area was calculated as described previously (11). Because of variations in transfection efficiencies, the average labeling density was calculated for 250 intersections, which corresponds to 70 μm of plasma membrane per grid. Averages and standard deviations were obtained from three different grids. Labeling densities on wt-HIV or delVpu virions and budding profiles were quantified as follows. The average membrane length of extracellular virions and budding profiles was calculated first by overlaying random images of virions at a ×115,000 final magnification and budding profiles with a grid lattice with a distance between the grid lines of 10 mm and then counting the intersections. The average membrane lengths of the virions and budding profiles were estimated to be 296 and 192 nm, respectively. The total amounts of gold particles associated with 46 virions and 72 budding profiles per grid (corresponding to 200 intersections with the grid lattice and 13.64 μm of membrane length) were counted and labeling densities were calculated by using the average membrane length of the viral structures. Averages and standard deviations were obtained by considering five grids.
Quantification of the relative CD317 distribution was carried out as described by Mayhew et al. (20). In brief, samples were chosen randomly for sectioning; grids with labeled sections were first examined at low magnification to find a grid square containing sections. Sections were then scanned systematically in an unbiased fashion at higher magnification, and every gold particle was counted and classified. Only CA- and CD317-double-positive cells were taken into account.
To localize endogenous CD317, we performed quantitative immunolabeling with a rabbit polyclonal antiserum raised against the recombinant protein (22). Thawed cryosections of HeLa cells showed low, but specific, labeling with anti-CD317. The specificity was assessed by parallel labeling of 293T cells (lacking endogenous CD317) showing virtually no background for untransfected 293T cells and a strong signal for 293T cells transfected with epitope-tagged CD317 (see Fig. S1 in the supplemental material). In HeLa cells, the endogenous protein localized to the plasma membrane (Fig. (Fig.1A),1A), membranes underneath the cell surface (Fig. (Fig.1B),1B), around the Golgi complex (Fig. (Fig.1C),1C), and structures reminiscent of late endosomes (Fig. (Fig.1D).1D). Only background labeling was found on mitochondria, the nucleus, and the cytoplasm (not shown). This localization of endogenous CD317 is similar to light microscopy studies with overexpressed CD317 showing colocalization with marker proteins of the endocytic pathway and the TGN (15, 19).
To validate the Vpu phenotype in our experimental system, we transfected HeLa cells with proviral plasmids producing wt HIV-1 or a Vpu deletion variant (delVpu). As reported previously, wt-HIV-1 transfection yielded efficient virus release with low numbers of particles retained at the plasma membrane (Fig. (Fig.2A),2A), whereas delVpu HIV-1 was often found in large clusters at the cell surface (Fig. (Fig.2B)2B) (14, 24, 29, 30). At high magnification, virions attached to delVpu-transfected cells occasionally revealed a short stalk that seemingly connected the extracellular particles to each other (Fig. (Fig.2B,2B, inset). Quantification of 100 cells transfected with either wt or delVpu HIV-1 yielded ~2.5-fold more particles at the cell surface in the case of the deletion variant. These were predominantly found in clusters of 2 to 20 particles, whereas larger clusters were also detected in wt-HIV-1-producing cells (Fig. (Fig.2C).2C). Few virions were observed in endosomal structures, indicating that the clustered virions at the plasma membrane were not significantly internalized or were degraded rapidly after internalization (data not shown). The release of virus particles was strongly reduced in delVpu-transfected cells, as shown by Western blotting (Fig. (Fig.2D)2D) and ELISA analysis (Fig. (Fig.2E),2E), and infectivity was equally reduced by a factor of 10 to 20 (Fig. (Fig.2E).2E). Thus, consistent with the presence of CD317 in HeLa cells, transfection of these cells with a virus lacking Vpu led to clustering of virions at the cell surface and to a strong inhibition of virus release.
Immunolabeling was used to study changes in the localization of CD317 in HIV-1-producing cells. Virus-producing cells were identified and analyzed by double labeling with anti-CA (HIV-1 capsid protein) and anti-CD317 antibodies. In contrast to the expectation, increased immunolabeling for CD317 was observed in HIV-1-positive compared to untransfected HeLa cells (Fig. (Fig.3A,3A, compare upper and lower cells). Quantitative analysis showed an increase in gold per area for wt (about 3-fold) and delVpu (~2-fold) HIV-1-producing cells compared to control cells (Fig. (Fig.3D).3D). Compared to uninfected cells (Fig. (Fig.1),1), the plasma membrane seemed less labeled in infected cells and the labeling was preferentially associated with membranes located underneath the plasma membrane (Fig. 3B and C). From their ultrastructure and location (close to the plasma membrane), these membranes were reminiscent of early endosomes (see below). Consistently, CD317 localized to clathrin-coated vesicles (Fig. (Fig.3B)3B) and coated pits (Fig. (Fig.3C),3C), whereas the plasma membrane and extracellular virions generally displayed little labeling (Fig. 3B and C).
Double-labeling experiments were carried out to determine the subcellular localization of CD317 in wt-HIV-1-transfected cells. Virus producing cells were identified by the presence of typical HIV-1 budding profiles. CD317 showed significant colocalization with TfR over membrane structures underneath the cell surface (Fig. 4A and B), as well as on membranes located close to one side of the Golgi complex (Fig. (Fig.4C).4C). Both proteins showed low but detectable labeling on late-endosome-like structures (Fig. (Fig.4D).4D). TfR is known to cycle between the plasma membrane, early endosomes, and the so-called recycling compartment, indicating that CD317 localized predominantly to these intracellular membrane structures as well. Accordingly, we found colocalization with EEA1, an early endosome marker, on structures located underneath the cell surface (Fig. 5A and B). Little colocalization was found with CD63, a protein that localizes predominantly to late endocytic structures, since the two proteins were mostly found on different sets of membranes (Fig. (Fig.5C).5C). Consistently, we also found little labeling for CD317 on LAMP1-positive late endosomes (Fig. (Fig.5D).5D). At the Golgi complex, CD317 colocalized with Golgin97 (Fig. (Fig.5E),5E), a TGN protein (9, 16), but not significantly with GM130 (Fig. (Fig.5F),5F), a cis-Golgi protein (18, 23), confirming its TGN localization. In summary, CD317 localized predominantly to TfR-positive early and recycling endosomes and the TGN in HIV-1-producing cells, with some labeling on late endocytic structures.
We also detected a subset of wt-HIV-1-producing cells that appeared to contain much higher levels of CD317 than neighboring cells. In such cells the protein was predominantly associated with accumulations of intracellular vesicles (see Fig. S2A to E in the supplemental material). This phenotype was most readily observed in cells that expressed high amounts of HIV-1 CA and thus were producing large amounts of virus (results not shown). Double-labeling experiments of such cells suggested that the CD317-enriched vesicles were derived from the Golgi complex and resulted from vesiculation of this compartment (see Fig. S2A to E in the supplemental material).
The apparent redistribution of CD317 upon transfection of a HIV-1 proviral plasmid was subsequently quantified. We determined the relative distribution of gold particles over different membranes, a method used successfully before to quantify the localization of ESCRT proteins (20, 32). To identify wt-HIV-1- or delVpu-transfected cells, we performed double labeling for HIV-1 CA and CD317. Triple labeling is not possible for EM, preventing us from using compartment-specific markers to unequivocally identify the various membrane structures, and these were therefore identified morphologically. This made it difficult to discriminate between early endosomes (EEA1 and TfR positive) and recycling endosomes (TfR positive and EEA1 negative) since they look morphologically very similar. These membranes were therefore counted as one pool and classified as early endosomes.
As shown in Fig. Fig.6,6, wt-HIV-1-producing cells exhibited a dramatic downregulation of CD317 from the plasma membrane (5% versus 35% in control cells) and a concomitant increase in labeling on early endosomes, where ca. 40% of the protein was found (only 10% in control cells). The labeling on other membranes remained mostly unchanged. HIV-1 delVpu-producing cells showed a distribution that was intermediate between control and wt-HIV-1-producing cells. The latter result was unexpected and suggested that besides Vpu additional factors play a role in the downregulation of CD317 from the plasma membrane. We expected the HIV-1 accessory protein Nef to be such a candidate since the latter can downregulate CD317 from the plasma membrane in retrovirus strains that do not contain Vpu (13, 33). Transfection with a double mutant lacking Vpu and Nef, however, led to the same relative distribution of CD317 as for HIV-1 lacking only Vpu, suggesting no additional role for HIV-1 Nef in CD317 cell surface downregulation in HeLa cells (see Fig. S3 in the supplemental material).
Relatively little labeling for CD317 was found to be associated with wt HIV-1 particles (1.5%) or virus buds (<1%), whereas these viral structures were relatively more frequently labeled in the case of delVpu (12 and 4.5% for virions and buds, respectively; Fig. Fig.66).
An important question was whether CD317 is enriched or depleted in the virion membrane compared to the cell surface and whether this is Vpu dependent. The quantification described above revealed the relative distribution of CD317 but did not permit a comparison of CD317 densities on cellular and viral membranes. We therefore directly determined the labeling density of CD317 over the plasma membrane, budding profiles, and extracellular virions for both wt HIV-1 and delVpu.
The labeling density on the plasma membrane was highest in control cells, ~3-fold lower in wt-HIV-1-producing cells and intermediate for delVpu (Fig. (Fig.7A).7A). Strikingly, analysis of budding profiles and apparently cell-free virions observed in thawed cryosections of wt and delVpu HIV-1-producing cells revealed a higher labeling density than in the plasma membrane of the respective cells (Fig. (Fig.7A).7A). This result suggested enrichment of CD317 in viral budding structures and virions. The CD317 labeling density was even higher in the viral structures than in the plasma membrane of control cells. In addition, there were only minor differences between wt and delVpu structures, suggesting that CD317 incorporation into HIV-1 may not be sensitive to Vpu-mediated downregulation from the plasma membrane in HeLa cells (Fig. (Fig.7A7A).
This apparent enrichment of CD317 in the wt and delVpu HIV-1 membrane was unexpected. One possible explanation would be that during sample preparation a subset of particles remained attached to the cell surface because it contained higher amounts of CD317, whereas viruses with little or no CD317 were lost from the cell sample. Accordingly, we determined the labeling density of CD317 in the membrane of cell-free HIV-1 collected from virus-producing cells and purified through a sucrose cushion. A very similar density was observed for these free virions as for the viruses attached to the cell surface (Fig. 7A and B). Thus, CD317 density appears to be increased in viral budding profiles and cell-free particles compared to the respective plasma membrane and appears to be unaffected by Vpu-mediated downmodulation. The presence of CD317 in the membrane of highly purified wild-type HIV-1 particles was confirmed by Western blot analyses of Opti-Prep-purified HIV-1 (Fig. (Fig.7C).7C). Silver staining of the same sample demonstrated that this preparation contained mostly HIV-1 structural proteins and was not significantly contaminated with cellular proteins, which were easily detectable prior to the Opti-Prep gradient (Fig. (Fig.7C7C).
Comparison of CD317 incorporation into wt and delVpu HIV-1 was not possible for particles purified from HeLa cells because of the low virus yield for the delVpu variant (Fig. (Fig.2E).2E). We therefore performed immuno-EM and Western blot experiments on transfected 293T cells that lack endogenous CD317. Cotransfection of pNL4-3 or the delVpu variant with an expression vector for HA-tagged wt CD317 yielded cell-associated virions that were readily labeled with both anti-CD317 (Fig. 8A and B) and anti-HA (data not shown), whereas only background labeling was observed without cotransfection of the CD317 expression plasmid (Fig. (Fig.8C).8C). The labeling density was 5- to 10-fold higher on cell-associated virions from cotransfected 293T cells compared to isogenic cell-associated virions from HeLa cells, probably owing to overexpression of CD317 in the former case. Furthermore, labeling density was 2-fold higher for cell-associated delVpu virions compared to the wt in the case of 293T cells.
To directly compare CD317 incorporation into purified cell-free wt and delVpu HIV-1, we used the C3A variant of CD317 that does not restrict the release of Vpu-defective HIV-1, while still localizing to the plasma membrane and being sensitive to Vpu downmodulation (1). This phenotype is due to the replacement of three essential Cys residues in the extracellular domain of CD317 by alanine residues. Cotransfection of 293T cells with wt or delVpu proviral plasmids, together with an expression vector for CD317 C3A, led to efficient particle release in both cases, as expected. Cell-free virions were purified on Opti-Prep gradients, and the purity of the preparation was analyzed on silver-stained gels as described above (data not shown). Western blot analysis of purified virions revealed a strong CD317 signal in both cases with an ~2-fold higher incorporation of CD317 into HIV-1 delVpu compared to the wt (Fig. (Fig.8D8D).
Taken together, these results showed that CD317 is incorporated into both wt and delVpu HIV-1 and indicated that CD317 function critically depends on the amount of CD317 at the plasma membrane and not in the virus membrane. Assuming that CD317 acts by directly linking the viral and cellular membranes, this suggested that purified HIV-1 particles (containing CD317) may show differential binding to cells expressing or lacking Vpu (and thus having different cell surface levels of CD317). Quantitative fluorescence microscopy of virus attachment revealed no significant differences between Vpu-expressing and control cells, however (see Fig. S4 in the supplemental material). Vpu-dependent downregulation of CD317 was clearly detectable in these samples by fluorescence microscopy or fluorescence-activated cell sorting analysis, but comparable numbers of fluorescently labeled viruslike particles bound to cells with high or low CD317 surface expression.
The present study localized CD317 at the ultrastructural level and quantified its relative distribution in HeLa cells. We confirm that CD317 localizes to the plasma membrane and throughout the endocytic pathway. Upon HIV-1 infection, CD317 is downregulated from the plasma membrane and predominantly redistributes to TfR- and EEA1-positive early and recycling endosomes. Finally, EM enabled us to directly localize and quantify CD317 in virus particles and buds and revealed a Vpu-independent enrichment of CD317 in viral structures compared to the plasma membrane. CD317 incorporation into HIV-1 was confirmed biochemically on highly purified virus preparations.
Immunolocalization of CD317 on the ultrastructural level confirmed previous light microcopy data (15, 19) but provided a more detailed view on cellular substructures and permitted precise quantification of the protein's subcellular distribution. The majority (>30%) of CD317 was found at the plasma membrane of HeLa cells with lower amounts on early and recycling endosomes and little protein at the late endosome. Using this quantification we also confirmed previous studies showing that CD317 is downregulated from the plasma membrane in HIV-1-producing cells (30). This downregulation is the result of a major redistribution of the protein to early and recycling endosomes. Plasma membrane localization of CD317 was partly, but not completely, restored in HeLa cells producing HIV-1 with the Vpu gene deleted. The remaining effect was not due to the HIV-1 Nef protein which had been shown to downmodulate CD317 in the case of lentiviruses naturally lacking Vpu (13, 33). The contribution of factors other than Vpu appeared minor, however, when analyzing the labeling density at the plasma membrane (Fig. (Fig.7A),7A), and the more pronounced difference between delVpu-producing and control cells upon analysis of relative CD317 distribution (Fig. (Fig.6)6) may be due to the observed overall increase of CD317 labeling in HIV-1-producing HeLa cells.
Analysis of average labeling per area indicated an increased level of CD317 in HIV-1-producing HeLa cells compared to control cells. These results are consistent with Mitchell et al. (21) showing CD317 redistribution that is not followed by major degradation, whereas other studies reported CD317 relocation accompanied by its degradation (6, 12). The immuno-EM data should be interpreted with caution, however, since this method is not optimal for determining quantitative differences in protein turnover. Independent of its subsequent fate, however, our quantitative immuno-EM results indicate that downregulation of cell surface CD317 and accumulation within the early/recycling endosome is sufficient to induce a full Vpu phenotype in HIV-1-producing HeLa cells. Redistribution rather than degradation presents the advantage of being quickly reversible with endocytic recycling occurring within minutes, depending on the protein (reviewed in reference 5).
The most surprising finding of the present study is the apparent enrichment of CD317 in the membrane of both wt and delVpu HIV-1. There is currently only limited information about CD317 incorporation into virus particles, with one study suggesting the protein to be absent from purified virions as assessed by Western blotting (22) and a very recent study reporting the incorporation of wt and defective CD317 into viruslike particles when overexpressed in 293T cells (26). Quantitative EM can visualize and count labeling densities over various membranes in the most direct way, thus allowing detection of proteins incorporated in low copy numbers. In the present study, we found the density of CD317 to be similar in wt HIV-1 and delVpu particles from HeLa cells and significantly enriched compared to the plasma membrane the viruses were derived from. Lack of CD317 detection in the virion fraction in a previous study (22) may have been due to insufficient sensitivity. Based on the labeling densities of HeLa cells and derived HIV-1 particles, their respective membrane surface area, and the amount of cell lysates needed for CD317 detection by immunoblotting, we calculated that a virus amount corresponding to ca. 0.5 to 1 μg of CA would be required for CD317 detection in the virion fraction in our system. Indeed, by Western blotting we were able to detect CD317 in highly purified wt HIV-1 from HeLa cells when 750 ng of CA or more was loaded.
Enrichment of CD317 in the viral membrane was especially surprising for wt virus, considering the much lower levels of CD317 at the plasma membrane in this case. CD317 was equally enriched in wt and delVpu virions tethered to the cell surface, and its C3A variant was almost equally incorporated into extracellular virions. CD317 is a glycosylphosphatidylinositol (GPI)-anchored protein that has been reported to be enriched in raftlike microdomains of the plasma membrane (15, 19, 27). Given that the HIV-1 membrane is strongly enriched in all putative raft lipids (4), it appears likely that virion incorporation of endogenous CD317 is affected by its sorting into membrane microdomains, thus potentially allowing local escape from Vpu-mediated downregulation. Overexpression of CD317 variants lacking either the N-terminal trans-membrane domain or the C-terminal GPI anchor revealed that both were incorporated into HIV-1 delVpu, while the variant lacking the trans-membrane domain was insensitive to Vpu-mediated downregulation from the cell surface (26). Local escape from Vpu downmodulation in membrane microdomains could explain why there is no significant difference in CD317 labeling density for wt and delVpu HIV-1 budding sites and cell-associated particles in HeLa cells and why the restriction-defective C3A variant of CD317 was almost equally incorporated into wt and delVpu HIV-1 from cotransfected 293T cells. We could exclude that the enhanced labeling density for CD317 on viral buds and cell-associated virus was due to a selected virus population trapped at the cell surface, because labeling densities on purified, cell-free virions were comparable and highly purified wt HIV-1 contained the predicted amount of CD317. Furthermore, the lower curvature of the plasma membrane compared to the virus membrane would favor surface protein labeling by immuno-gold, making it unlikely that the observed difference is caused by technical aspects.
During preparation of the present study, Perez-Caballero et al. (26) reported the incorporation of CD317 variants into HIV-1 after transfection of 293T cells. That study mainly used HIV-1 delVpu and defective CD317 variants, however, and no direct comparison between wt and delVpu HIV-1 was made. Importantly, these authors also observed the incorporation of wt CD317 into viruslike particles upon overexpression of Gag. This result was suggested to be due to titration of CD317 at the cell surface but clearly confirms that particles carrying wt CD317 can be released and is thus in accordance with our results. Incorporation of wt CD317 also indicates that particles carrying a single CD317 dimer are not necessarily tethered to the cell surface, suggesting that efficient tethering requires multiple dimer interactions.
The observation that CD317 is enriched in wt and delVpu HIV-1 (and clearly not excluded) has implications for its mechanism of action. Based on the unusual topology of CD317 with a GPI anchor and a membrane-spanning domain at either end connected by a helical domain prone to form a disulfide-linked coiled coil (1, 26), two models for CD317 tethering have been suggested: (i) the virus and cell membrane are bridged by a dimer of CD317 with its N and C termini in the two different membranes or (ii) CD317 molecules have their N and C termini in the same membrane (either cellular or viral), and tethering occurs through dimerization by the helical region. The observation of enriched CD317 levels in the virus and depletion in the host cell plasma membrane appears to favor the second model, because loss of protein density on either membrane would be sufficient for the full phenotype in this case. Tethering of HIV-1 would then be largely determined by the surface density of CD317 on the producer cell membrane and largely independent of the density on the virion surface. The recent structural analysis of the CD317 coiled-coil domain suggests, however, that such a topology would place the tethered virion in very close (3 to 5 nm) proximity with the plasma membrane (W. Weissenhorn, unpublished data), which appears to be inconsistent with the available EM data. A dimer of CD317 spanning both membranes, on the other hand, would place the virion in a distance of ~17 nm. This topology is difficult to reconcile with the largely unaltered CD317 incorporation into wt HIV-1, however. Perez-Caballero et al. (26) attempted to determine the membrane topology of CD317 by protease digestion and observed that either the N or the C terminus can be incorporated into the membrane of viruslike particles if only one membrane anchor is present and at least some CD317 molecules have both N termini of the wt dimer in the viral membrane. This does not distinguish between dimers spanning the cell and virion membrane and dimers anchored in only one membrane, however. Alternatively, CD317 may occur in several topologies with more than one CD317 dimer being required for efficient tethering. In this case, modest downmodulation of CD317 from the cell surface by Vpu (2- to 3-fold difference in plasma membrane labeling per area for wt and delVpu-producing HeLa cells) may already be sufficient to overcome the tethering effect of CD317, especially with constitutively low CD317 surface expression. Furthermore, Vpu may act differentially on differently anchored CD317 molecules, and some of the virion-associated CD317 may thus be functionally inactive. The presence of inactive CD317 molecules in the virion would also be consistent with the equal binding efficiency of CD317-carrying particles to cells with low and high CD317 surface densities. Further studies are clearly required, however, to reveal the precise mechanism of CD317 action on virus release. Our results indicate that CD317 tethering of HIV-1 in HeLa cells is largely controlled by cell surface levels and not by virion membrane levels of the protein. Vpu-dependent downregulation into the early/recycling endosome correlates with its release function but appears to have little effect on the virion membrane density of CD317.
After acceptance of this manuscript, another immuno-EM study was reported, also describing a direct association of CD317/tetherin with HIV budding sites and particles (J. Hammonds, J.-J. Wand, H. Yi, and P. Spearman, PLoS Pathog. 6:e1000749, 2010).
This study was supported in part by grants from the Deutsche Forschungsgemeinschaft (SFB638, A9) and within the EU FP7 program (HIV-ACE). H.-G.K. is an investigator in the CellNetworks Cluster of Excellence (ECX81).
We are grateful to V. Bosch for proviral plasmids and to A. De Matteis and A. Jones for antibodies. We thank B. Müller for support and G. Griffiths, O. Fackler, and O. Keppler for critical reading of the manuscript.
Published ahead of print on 10 February 2010.
‡Supplemental material for this article may be found at http://jvi.asm.org/.