HIV-1 Vpu Antagonizes BST-2 by Interfering Mainly with the Trafficking of Newly Synthesized BST-2 to the Cell Surface

Abstract

Bone marrow stromal cell antigen-2 (BST-2) inhibits human immunodeficiency virus type 1 (HIV-1) release by cross-linking nascent virions on infected cell surface. HIV-1 Vpu is thought to antagonize BST-2 by downregulating its surface levels via a mechanism that involves intracellular sequestration and lysosomal degradation. Here, we investigated the functional importance of cell-surface BST-2 downregulation and the BST-2 pools targeted by Vpu using an inducible proviral expression system. Vpu established a surface BST-2 equilibrium at ~60% of its initial levels within 6 h, a condition that coincided with detection of viral release. Analysis of BST-2 post-endocytic trafficking revealed that the protein is engaged in a late endosomal pathway independent of Vpu. While Vpu moderately enhanced cell-surface BST-2 clearance, it strongly affected the protein resupply to the plasma membrane via newly synthesized proteins. Noticeably, Vpu affected clearance of surface BST-2 more substantially in Jurkat T cells than in HeLa cells, suggesting a cell-dependent impact of Vpu on the pool of surface BST-2. Collectively, our data reveal that Vpu imposes a new BST-2 equilibrium, incompatible with efficient restriction of HIV-1 release, by combining an acceleration of surface BST-2 natural clearance, whose degree might be cell-type dependent, to a severe impairment of the protein resupply to the plasma membrane.

Although the accessory proteins (Vif, Vpr, Vpu and Nef) encoded by human immunodeficiency virus type 1 (HIV-1) were originally considered dispensable for virus replication in vitro, they are now thought to play a critical role in viral pathogenesis, given their functions in subduing a hostile target cell environment (1). Among these accessory proteins, Vpu is a small oligomeric type 1 transmembrane protein that is unique to HIV-1 and primate lentiviruses related to simian immunodeficiency virus found in chimpanzee. This protein, which resides primarily in the trans-Golgi network (TGN) and the endosomal system, exerts several biological activities during HIV-1 infection (reviewed in 2). First, Vpu was found to target newly synthesized CD4 molecules for proteasomal degradation through an endoplasmic reticulum (ER)-associated protein degradation (ERAD)-like process (3,4). The cytoplasmic domain of Vpu contains a DS₅₂GxxS₅₆ phosphoserine motif that recruits β-TrCP and the SCF^β–TrCP E3 ligase complex to ubiquitin-mark CD4 for degradation (5). The resulting depletion of CD4 receptor molecules is believed to promote the transport of Env glycoproteins to the plasma membrane and to minimize superinfection of infected cells (6,7). However, the Vpu biological activity that has attracted considerable attention in recent years is the antagonism of bone marrow stromal cell antigen-2 (BST-2, also designated Tetherin/CD317/HM1.24), an interferon (IFN)-inducible host restriction factor that strongly inhibits the release of HIV-1 from infected cells (8,9).

BST-2 is a type II membrane protein of 181 amino acids with a molecular weight of 29 and 33 kDa depending on its glycosylation state. The protein has an atypical topology with both extremities embedded in the cellular membranes through two different types of membrane anchors: a transmembrane domain proximal to an intracellular N-terminus cytosolic region and a C-terminal glycosyl-phosphatidylinositol (GPI) anchor (10). The two membrane anchors are connected by an extracellular domain that contains an extended coiled-coil structure that mediates BST-2 homodimerization (11,12). The protein is localized not only at the plasma membrane but also within several endosomal membrane compartments, including the TGN as well as early and recycling endosomes (10,13–16). Trafficking of BST-2 between the plasma membrane and the TGN was shown to require a tyrosine-based sorting signal in the BST-2 cytoplasmic domain. This YxYxxΦ motif was shown to interact with both clathrin adaptors AP1 and AP2, suggesting the involvement of the sequential action of AP-2 and AP-1 complexes in BST-2 internalization and delivery back to the TGN (15,17).

As a GPI-anchored protein, BST-2 is found within cholesterol-enriched lipid domains from which HIV-1 and other enveloped viruses preferentially assemble and bud (10,15,18). Up to now, BST-2 has been shown to www.traffic.dk restrict the release of a broad variety of enveloped viruses, including all classes of retroviruses (reviewed in 19). This broad spectrum antiviral activity appears to reside in the ability of the protein to target a virion component common to this diverse group of viruses, the host cell-derived lipid bilayer. Indeed, biochemical, structural and electron microscopic evidence, mostly from studies of HIV-1, currently favors a direct tethering mechanism of virus restriction, whereby parallel BST-2 dimers physically cross-link not only virion and cellular membranes but also virion membranes together (11,16,20–25).

The mechanism underlying BST-2 antagonism by HIV-1 Vpu is still a matter of debate. It is generally agreed that Vpu expression causes a downregulation of surface BST-2 levels (8) and a reduction of the total levels of the protein in most cellular systems (26–29). This Vpu-mediated depletion of BST-2, which indeed requires a physical interaction between the transmembrane domains of the two proteins (30–34), is thought to remove the restriction factor from the plasma membrane, the site of its tethering action, although this model has previously been challenged (35). Several studies reported the involvement of proteasomal (26–28) or endolysosomal pathways (29,36) in Vpu-mediated downregulation of BST-2 and suggested a dependence on β-TrCP2 engagement by Vpu. Consistent with the involvement of a lysosomal pathway, a recent study reported that Vpu accelerates BST-2 degradation by connecting the restriction factor to HRS (also called hepatocyte growth factor-regulated tyrosine kinase substrate), a component of the endosomal sorting complexes required for transport-0 (ESCRT-0) machinery that sorts ubiquitinated proteins to multivesicular bodies (MVB) and lysosomes for degradation (37). In addition, other studies reported that Vpu could also sequester BST-2 away from the plasma membrane, most probably in the TGN (32,38). As this sequestration was sufficient to inhibit the restricting activity of BST-2, several recent reports suggested that β-TrCP2-dependent targeting of the restriction factor to lysosomal compartments for degradation could represent a subsequent step required to achieve optimal depletion of BST-2 at the cell surface (32,37,39). In that regard, recent studies reported efficient enhancement of HIV-1 release by Vpu in the absence of BST-2 degradation (32,40,41).

As Vpu is expressed from a Rev-dependent bicistronic mRNA and, consequently, is made late during HIV-1 infection (42), it is unclear how fast and to what extent surface BST-2 levels need to be reduced before efficient HIV-1 particle release is initiated. Furthermore, it is unknown whether Vpu acts at pre- and/or post-endocytic steps to induce BST-2 sequestration and degradation. Using an inducible HIV-1 proviral expression system, we evaluated the kinetics of cell-surface BST-2 downregulation and HIV-1 release. We show that optimal cell-surface BST-2 downregulation is achieved within 6 h of Vpu expression. Importantly, the rapid emergence of this new BST-2 equilibrium coincided with detection of HIV-1 particle release. Analysis of BST-2 post-endocytic trafficking revealed that the protein is not sensitive to monensin, a drug interfering with fast recycling, but is rather engaged in a late endosomal pathway independent of the presence of Vpu. Lastly, our results show that, while Vpu accelerated surface BST-2 clearance and decay to varying degrees in HeLa and Jurkat T cells, it severely affected the restriction factor resupply to the plasma membrane via newly synthesized proteins independently of the cell type. Therefore, by combining a severe impairment of the restriction factor resupply to the plasma membrane to an acceleration of the surface BST-2 natural clearance process, whose degree might be cell-type dependent, Vpu imposes a new BST-2 equilibrium, which is incompatible with efficient restriction of HIV-1 particle release.

Results

Characterization of an inducible HIV-1 proviral expression system

To study the early effect of Vpu on the surface levels of BST-2, we used an inducible system in which expression of HIV-1 genes is driven by tetO responsive elements in the presence of doxycycline (Dox). The wild-type (WT) and Vpu-defective proviral constructs were derived from the previously described HIV-1 rtTA provirus in which the Tat-TAR regulatory axis of transcription was inactivated. Viral transcription and replication were made Dox-dependent by introducing eight tetO elements in the long-terminal repeat (LTR) promoter region between NF-κB and Sp1 sites and by replacing the nef gene by a genetically selected Dox-regulated rtTA transactivator gene (43). In addition, we inserted enhanced green fluorescent protein (EGFP) in frame with the matrix protein (MA) as a marker of viral protein expression, thus generating inducible EGFP-encoding proviruses (HIV-1.rtTA.MA-EGFP) that are isogenic except for vpu expression (Figure 1A). Indeed, while transfection of HIV-1.rtTA.MA-EGFP in HeLa cells, a cell line that constitutively expresses BST-2 (8,9), revealed only background levels of p55-EGFP and no virus particle production 48 h post-transfection (Figure 1B; lanes 1, 5 and 9), cells treated with a low dose of Dox (250 ng/mL) displayed a very strong upregulation of all viral proteins, including Vpu in the case of Vpu-proficient constructs (Figure 1B; lanes 2, 6 and 10). Importantly, induction of HIV-1.rtTA.MA-EGFP.vpu+ expression led to downregulation of surface BST-2 on EGFP-positive cells and efficient virus particle release, whereas induction of HIV-1.rtTA.MA-EGFP.vpu− expression did not (Figure 1B, compare lanes 6–8 to lanes 2–4 and Figure 1C). Importantly, the reduction of cell-surface BST-2 levels observed with this inducible system was comparable to that detected with replication-competent HIV-1 [(32) and data not shown], indicating that the levels of Vpu expression reached with this system are physiological. Consistent with our previous observation (32), induction of a similar proviral construct encoding Vpu S52D,S56D, a mutant of Vpu that does not bind β-TrCP (32), resulted in an intermediate phenotype for both enhancement of virus particle release (Figure 1B; compare lanes 10–12 to lanes 6–8) and downregulation of cell-surface BST-2 (Figure 1C). Residual downregulation of surface BST-2 by Vpu S52D,S56D is likely to result from BST-2 intracellular sequestration because this mutant has the ability to bind and redistribute BST-2 to the TGN but is unable to induce its degradation (32). Therefore, this inducible proviral expression system recapitulates BST-2 antagonism by Vpu.

Figure 1. Characterization of inducible HIV-1 proviral constructs.

A) Schematic representation of the HIV-1.rtTA.MA-EGFP genomic organization. rtTA consists of the DNA binding domain (BD) of the tetracycline repressor fused to herpes simplex virus VP16 activation domains (AD). B and C) Characterization of HIV-1.rtTA.MA-EGFP proviral constructs. HeLa cells were transfected with the indicated constructs and treated or not with 0.25–1 μg/mL of Dox for 48 h. B) Virus and cell lysates were analyzed by western blot using anti-p24 or anti-Vpu Abs. C) Transfected cells from (B) were stained for surface BST-2. The graph depicts levels of surface BST-2 on EGFP-positive cells except for the EGFP-negative control (EGFP−).

A new BST-2 equilibrium is rapidly reached following Vpu induction

We next evaluated the kinetics of BST-2 downregulation at the cell surface upon induction of Vpu expression. HeLa cells were transfected with the inducible isogenic HIV-1.rtTA.MA-EGFP constructs and further treated with 1 μg/mL of Dox for different intervals of time (0–24 h) to induce viral protein expression. Cell-surface BST-2 levels were then analyzed 48 h post-transfection by flow cytometry. Vpu-mediated BST-2 downregulation was detectable as rapidly as 3 h after Dox induction, indicating expression of the viral protein, and reached its maximal effect of ~40% after 6 h of treatment (Figure 2A). Longer Dox treatments did not result in any further downregulation, indicating that surface BST-2 reached a new equilibrium. Interestingly, optimal cell-surface BST-2 downregulation at 6 h coincided with a detectable release of virus particle from Vpu-expressing cells (Figure 2B; compare lane 8 to lane 3). Noticeably, virus particle release was significantly delayed and inefficient in the absence of Vpu (compare lanes 2–5 to lanes 7–10). As expected, the new BST-2 equilibrium established by Vpu was reversible because extensive washing of the media to remove Dox and repress proviral expression restored initial surface BST-2 levels within 24 h (Figure 2C). Interestingly, the kinetics of BST-2 downregulation measured in Vpu S52D,S56D-expressing cells was delayed relative to that of WT Vpu-expressing cells. It peaked at ~32% of downregulation after 12 h of Dox treatment but failed to maintain this new equilibrium afterwards (Figure 2A). Similarly, the kinetics of WT Vpu-mediated cell-surface BST-2 down-regulation was found to be delayed in β-TrCP2-depleted HeLa cells and indeed recapitulated the kinetics observed with the Vpu S52D,S56D mutant (Figure S1), confirming that recruitment of β-TrCP2 is required for optimal surface BST-2 downregulation. Overall, these results indicate that Vpu imposes a reduction in the levels of BST-2 at the cell surface as rapidly as 6 h following its expression and this coincides with the detection of virus particle release.

Figure 2. Vpu achieves optimal cell-surface BST-2 downregulation within 6 h.

A) HeLa cells were transfected with the indicated HIV-1.rtTA.MA-EGFP constructs and analyzed 48 h post-transfection for surface BST-2 levels by flow cytometry. Prior to analysis, cells were treated for the indicated period of time with 1 μg/mL of Dox. Error bars represent the standard deviation calculated from three independent experiments. B) Virus-containing supernatants and transfected cells from (A) were processed as described in Materials and Methods to detect Gag proteins by western blot using anti-p24 Abs. The blot for analysis of virus particles was overexposed to detect p24 at early time-points. C) HeLa cells were transfected with the indicated HIV-1.rtTA.MA-EGFP constructs, treated with Dox for 24 h and analyzed 48 h post-transfection for surface BST-2 expression by flow cytometry. At the indicated time intervals, cells were washed to remove Dox.

Vpu reduces cell-surface BST-2 levels without altering the restriction factor distribution at the cell surface

To rapidly overcome BST-2 restriction on HIV-1 release, Vpu could exclude BST-2 from viral assembly sites. In this regard, Vpu was previously shown to decrease the colocalization between Gag and BST-2 at the surface of infected cells (8,36,44). To address this possibility, we characterized the distribution of the restriction factor at the surface of Jurkat cells infected with virus derived from GFP-marked HxBH10.GFP.IRES.nef-. In the absence of Vpu, surface BST-2 accumulated in patches that co-stained with p17, a marker of mature viruses and viral assembly sites (Figure 3A). As expected, in similar conditions of image capture, surface levels of BST-2 were drastically decreased in the presence of Vpu, while levels of p17 were barely detectable presumably as a result of efficient virus particle release (Figure 3A and flow cytometry analysis in panel B). Interestingly, when the staining signal was increased to an intensity comparable to that detected in Vpu-deficient HIV-1 infected cells (Figure 3A, upper panels), residual surface BST-2 still strongly colocalized with p17 (Figure 3A, bottom panel, long exposure, and quantification in Figure 3D), suggesting that Vpu is not displacing surface BST-2 outside membrane microdomains where viral assembly is taking place. Consistent with these observations, surface BST-2 colocalized with CD59, a lipid raft marker, as well as with CD81 and CD63, two markers of the tetraspanin-enriched microdomains, both in the absence or presence of Vpu (Figure 3C,D; the signal intensity in the Vpu+ images was increased). Overall, these results reveal that BST-2 colocalizes with markers associated with membrane microdomains that HIV-1 uses as assembly and budding platforms (45,46). They further show that Vpu downregulates cell-surface BST-2 levels without qualitatively affecting the distribution of the restriction factor at the plasma membrane.

Figure 3. Distribution of BST-2 at the plasma membrane of HIV-1-infected Jurkat cells.

Jurkat cells were infected with VSV-G-pseudotyped HxBH10.GFP.IRES.nef-vpu− or -vpu+ viruses. A) Forty-eight hours post-infection, cells were stained with anti-BST-2 Abs (cyan), fixed, permeabilized and then sequentially stained with anti-p17 Abs (red) and appropriate secondary Abs. In the lower panels (longer exposure), the gain was scaled up to obtain BST-2 and p17 signals comparable to those detected in the HIV-1(Vpu−)-infected cells from the upper panel. Magnifications (3×) are shown beside the panels. B) Jurkat cells infected with VSV-G-pseudotyped HxBH10.GFP.IRES.nef-vpu− or -vpu+ viruses were stained 48 h post-infection with anti-BST-2 Abs. Surface BST-2 levels were analyzed on EGFP-positive cells by flow cytometry. C) Cells were co-stained with anti-BST-2 Abs (cyan) and anti-CD59, anti-CD81 or anti-CD63 Abs (red) 48 h post-infection, and processed as described above prior to staining with appropriate secondary Abs. The gain from Vpu-expressing cells was scaled up as described above. Nuclei were counterstained with DAPI (blue). Green: GFP. White bar = 10 μm. D) Quantification of BST-2 colocalization with p17 or cellular markers. The values (%) represent percentages of BST-2 (cyan pixels) overlapping with each marker (red pixels). Error bars indicate the standard deviation of the mean from the quantitative analysis of at least 25 distinct cells.

The pool of BST-2 at the cell surface is modestly downregulated by Vpu

Previous studies, including ours, showed that Vpu does not increase the rate of BST-2 internalization (32,36,39). However, as these experiments were performed at steady state, once a new equilibrium was established at the plasma membrane, potential early effects of Vpu on the clearance of the pool of BST-2 at the surface may have been overlooked. To overcome this limitation, we took advantage of our inducible system. HeLa cells were transfected with HIV-1.rtTA.MA-EGFP constructs and analyzed 48 h post-transfection for surface BST-2 expression by flow cytometry. Prior to analysis, cells were incubated at 37°C for 1.5 h in the presence of a polyclonal anti-BST-2 rabbit serum to specifically label surface BST-2, washed and incubated for 0–24 h in the presence of 1 μg/mL of Dox to induce viral protein expression (Figure 4A). The fluorescent signal detected with the secondary antibody (Ab) at each time interval was indicative of the BST-2 levels that remained at the cell surface. The endogenous half-life of cell-surface BST-2 was ~8 h in the absence of Vpu (Figure 4B). The rate of surface BST-2 clearance measured with this Ab-based assay was found similar to that reported using a different assay system in which BST-2 resupply was blocked by the post-ER trafficking inhibitor Brefeldin A (47), indicating that the sum of processes governing the clearance of cell-surface BST-2 was not altered by the anti-BST-2 Ab labeling (data not shown). Upon induction of WT Vpu or Vpu S52D,S56D, the half-life of cell-surface BST-2 slightly decreased to ~6.5 and ~7 h, respectively. In fact at 6 h, the levels of BST-2 at the surface of Vpu-expressing cells were decreased by ~11% relative to control cells (Figure 4B). Although reproducible, this modest difference in the rate of BST-2 clearance from the surface cannot explain the ~40% decrease observed in Figure 2A at 6 h post-Dox induction. Therefore, Vpu must affect another pool of BST-2 to achieve optimal downregulation from the cell surface.

Figure 4. Kinetics of BST-2 clearance and decay.

A) Schematic representation of the experiment. B and C) HeLa cells were transfected with the indicated HIV-1.rtTA.MA-EGFP proviral constructs and analyzed 48 h post-transfection for surface BST-2 expression by flow cytometry. Prior to analysis, cells were incubated at 37°C for 1.5 h, washed and reincubated in the presence of 1 μg/mL of Dox for the indicated period of time. Cells were then harvested and (A) surface-stained or (B) fixed, permeabilized and stained with appropriate secondary Abs as described. For the samples treated with Dox, levels of BST-2 were evaluated on EGFP-positive cells. Levels of BST-2 at the cell surface at time 0 (no Dox) were monitored on the total population. The graphs depict the surface (A) or total levels (B) of BST-2 relative to time 0. Error bars represent the standard deviation calculated from three independent experiments.

We next analyzed the rate of decay of the BST-2 pool originating from the cell surface. To do so, half of the transfected cells processed in Figure 4B were permeabilized prior to staining with an appropriate secondary Ab. In this experiment, cell permeabilization allowed the detection of surface-derived BST-2 molecules engaged in endocytic compartments as well as those remaining cell surface-associated (Figure 4A). In the absence of Vpu, the turnover of cell-surface BST-2 was slow, with ~70% of the initial levels remaining after 24 h. Thus, the rate of cell-surface BST-2 decay appears much slower than its rate of clearance (compare Figure 4C to 4B). Induction of Vpu slightly accelerated this process because ~80% of the initial surface BST-2 levels were detected at 6 h post-induction, while ~50% remained after 24 h. Interestingly, induction of Vpu S52D,S56D led to a degradation kinetics that appeared similar to that of WT Vpu within the first 6 h, but which then progressively slowed down between 6 and 24 h post-induction to reach levels comparable to the Vpu-negative control (~70% at 24 h). These results suggest that a fraction of internalized BST-2 is slowly sorted toward degradative compartments and that Vpu modestly accelerates this process over time in a β-TrCP-dependent manner.

Internalized BST-2 accumulates in late endosomes

Having shown that BST-2 was endocytosed, we next evaluated the localization of internalized BST-2 by confocal microscopy. Given that degradation of internalized BST-2 appeared very slow, we reasoned that long incubation with Abs would be necessary to fully detect the association of BST-2–Ab complexes with degradative compartments. Therefore, untransfected HeLa cells were incubated with a polyclonal anti-BST-2 rabbit serum for 12 h at 37°C. Cells were then fixed, permeabilized and sequentially stained with Abs directed against specific cellular markers and secondary Abs that could recognize BST-2- and cellular marker-specific Abs. Localization of internalized BST-2 was compared to Rab5 (early endosomes) (48), Rab11 (recycling endosomes) (49), TGN46 (TGN) (50), Rab9 (51) and CD63 (52) (late endosomes) (Figure 5A). Surprisingly, only ~17–23% of internalized BST-2 signal overlapped with Rab5, Rab11 and TGN46 markers in these conditions, suggesting that internalized BST-2 is transiently trafficking through early and recycling endosomes and the TGN. In contrast, ~66–74% of the internalized BST-2 signal readily overlapped with CD63 and Rab9, two markers of late endosomes (Figure 5B). Importantly, internalized BST-2 similarly overlapped with late endosomal markers upon induction of Vpu expression in HIV-1.rtTA.MA-EGFP-transfected HeLa cells (Figure 5C,D). Colocalization of BST-2 with these late endosomal markers was also observed in fixed cells, thus excluding an aberrant intracellular distribution triggered by anti-BST-2 Abs (Figure S2). Trafficking of internalized BST-2 was also characterized at shorter time intervals to obtain additional information about the pathway followed by BST-2 to ultimately reach lysosomal compartments. As expected, internalized BST-2 was found at the plasma membrane and in vesicles beneath the plasma membrane after 10 min of incubation. Strong accumulation in a TGN46-positive compartment could be observed at 30–60 min, whereas association to CD63-positive vesicles became predominant after 6 h of incubation (Figure S3), suggesting a transient accumulation of internalized BST-2 within the TGN prior to its lysosomal sorting.

Figure 5. Localization of internalized BST-2 molecules.

A) HeLa cells were incubated at 3C for 12 h in the presence of anti-BST-2 Abs, fixed, permeabilized and then stained for the indicated intracellular markers. Stained cells were washed and incubated with appropriate secondary Abs to detect cellular markers (green) as well as internalized BST-2 (red). Nuclei were counterstained with DAPI (blue). Cells were analyzed by confocal microscopy. White arrow heads highlight examples of colocalization between BST-2 and specific cellular markers. White bars = 10 μm. B) Quantification of BST-2 colocalization with cellular markers. The values (%) represent the percentage of BST-2 (red pixels) overlapping with each cellular marker (green pixels). Error bars indicate the standard deviation of the mean from the quantitative analysis of at least 25 distinct cells. C) Coverslip-seeded HeLa cells were transfected with HIV-1.rtTA.MA-EGFP vpu− or vpu+. At 36 h post-transfection, cells were incubated at 37°C for 12 h in the presence of Dox and anti-BST-2 Abs, washed, fixed and permeabilized. Cells were then stained for the indicated intracellular makers. Stained cells were extensively washed and incubated with appropriate secondary Abs to detect cellular markers (red) as well as internalized BST-2 (cyan). Nuclei were counterstained with DAPI (blue). Cells were analyzed by confocal microscopy. White arrow heads highlight examples of colocalization between internalized BST-2 and cellular markers. White bars = 10 μm. D) Quantification of (C) as described in (B).

Overall, these results indicate that a large fraction of endogenous surface BST-2 is ultimately targeted to late endosomal compartments following its endocytosis. Although Vpu modestly accelerates the clearance of cell-surface BST-2, the intracellular distribution of internalized restriction factor molecules does not appear strongly affected by the viral protein.

Endogenous BST-2 is not sensitive to monensin, an inhibitor of fast recycling

BST-2 was previously reported to colocalize on endosomal membranes with the Transferrin receptor (TfR) (15,31), a protein that undergoes fast recycling back to the plasma membrane. To examine whether endogenous BST-2 is indeed recycling through a similar pathway, we assessed the effect of monensin, an inhibitor of the fast recycling pathway that was previously shown to block the recycling of the Transferrin and the low-density lipoprotein receptors (LDLR) (53,54). As expected, a 2-h treatment of HeLa or Jurkat cells with 100 μM monensin resulted in downregulation of LDLR at the surface as measured by flow cytometry (Figure 6A,C). Similar results were also obtained with the TfR (data not shown). In contrast, a 2-h (Figure 6B,D) or 5-h (data not shown) treatment of cells with similar concentrations of monensin did not have any effect on endogenous surface BST-2 levels, suggesting that BST-2 is not engaged in a recycling pathway similar to that used by LDLR or TfR.

Figure 6. Effect of Monensin on cell-surface BST-2 expression.

HeLa (A and B) and Jurkat cells (C and D) were treated with 100 μM monensin for 2 h at 37°C. Cell-surface LDLR (A and C) and BST-2 (B and D) levels were then monitored by flow cytometry as described.

Cell-surface downregulation of BST-2 by Vpu is primarily dependent upon BST-2 de novo synthesis

As the effect of Vpu on the pool of BST-2 derived from the cell surface appeared modest at best, we next examined the effect of the viral protein on the pool of newly synthesized BST-2. To do so, we took advantage of BST-2 inducibility by type 1 IFN. Upregulation of IFN-inducible genes by exogenous IFN involves endogenous expression of IFN and the establishment of a durable positive feedback loop (55), whereby newly produced IFN continues to act on its targets. This IFN-mediated positive feedback loop can be neutralized by B18R, a vaccinia-encoded secreted protein with a higher affinity than the cellular IFN receptor for type I IFN (56). The capacity of B18R to inhibit IFN-driven BST-2 expression was evaluated in HEK293T cells, which are devoid of BST-2 expression. Consistent with the establishment of a durable positive feedback loop (55), a strong upregulation of surface BST-2 was still detected 24 h after a 1-h pulse of exogenous IFN-α (Figure S4A). In contrast, addition of 10–100 ng/mL of B18R immediately after the IFN-α pulse completely prevented this upregulation. Indeed, using a type 1 IFN reporter system (HEK-blue IFN-α/β^™) (57), we confirmed that at concentration of 50 or 100 ng/mL, B18R completely prevented the initiation of an IFN positive feedback loop because no active IFN could be detected in the supernatant of IFN pulse-treated cells after 24 h (Figure S4B).

Having shown that we could interfere with induction of BST-2 neo-synthesis by type 1 IFN using B18R, we next examined whether blocking BST-2 synthesis could affect Vpu-mediated downregulation of surface BST-2. HEK293T cells were transfected with Dox-inducible HIV-1.rtTA.MA-EGFP proviral constructs and immediately treated with IFN-α for 42 h to ensure that steady-state levels of surface BST-2 were reached. Following extensive washes, Dox and recombinant B18R were added to cells to respectively induce expression of viral proteins and block IFN-α-induced BST-2 neo-synthesis. Six hours later, at the peak of Vpu-mediated downregulation of BST-2 (Figure 2), surface BST-2 levels were monitored by flow cytometry on EGFP-positive cells (Figure 7A). Consistent with an interruption of the IFN-mediated positive feedback loop and a block of IFN-induced BST-2 synthesis, B18R treatment resulted in an ~38% downregulation of cell-surface BST-2 [Figure 7B,C; compare mean fluorescence intensity (MFI) of the Vpu− condition in the absence (MFI:26) and presence (MFI:16) of B18R]. This downregulation was indeed very similar to that achieved upon Vpu induction in the absence of B18R treatment (Figure 7B,D). We hypothesized that if Vpu acted primarily on the pool of newly synthesized BST-2, Vpu-mediated downregulation of BST-2 would be affected in the presence of B18R. Indeed, as shown in Figure 7C,D, the effect of Vpu on surface BST-2 levels was strongly attenuated in the presence of B18R (downregulation of ~38% in the absence of B18R and ~11% in the presence of B18R). Taken together, these results suggest that downregulation of BST-2 by Vpu depends on a large part on BST-2 de novo synthesis. Furthermore, as inhibition of BST-2 neo-synthesis recapitulates to a large extent the effect of Vpu on BST-2 surface levels, it is therefore likely that Vpu targets primarily the pool of newly synthesized BST-2 en route to the plasma membrane.

Figure 7. Effect of B18R on Vpu-mediated BST-2 downregulation.

A) Schematic representation of the experiment. B and C) HEK293T cells were transfected with the indicated HIV-1.rtTA.MA-EGFP constructs and immediately treated with 1000 U/mL of IFN-α. Forty-two hours post-transfection, 1 μg/mL of Dox was added to the media without (B) or with 100 ng/mL of soluble B18R (C). Surface BST-2 levels were evaluated 6 h later by flow cytometry. D) Quantification of (B) and (C). The error bars represent the standard deviation calculated from two independent experiments. The IFN-treated Vpu-negative sample was set at 100% to illustrate the effect of B18R in the presence or absence of Vpu.

Vpu reduces the expression of newly synthesized BST-2 at the cell surface

To examine directly whether Vpu targets preferentially the pool of newly synthesized BST-2 en route to the cell surface, we developed a second system in which the effect of Vpu on the pools of cell-surface and newly synthesized BST-2 could be distinguished and simultaneously compared. HeLa cells were transfected with HIV-1.rtTA.MA-EGFP-vpu−/vpu+/vpu S52D,S56D constructs. Forty-one hours post-transfection, cells were incubated for 1 h at 37°C with a saturating concentration of the polyclonal anti-BST-2 rabbit serum to label all surface BST-2 molecules. Cells were then extensively washed and reincubated for 6 h at 37°C in the presence of 1 μg/mL Dox. Subsequently, half of the cells were harvested and sequentially stained for 45 min at 4°C with the polyclonal anti-BST-2 mouse serum and an anti-mouse secondary Ab to specifically label newly synthesized proteins reaching the surface. In parallel, the other half of the cells was similarly incubated for 45 min at 4°C and then directly stained with the anti-rabbit secondary Ab to specifically label surface BST-2 (Figure 8A). Conceptually, the mouse serum will only recognize unlabeled BST-2 molecules that were not present at the surface during the initial staining with the rabbit anti-BST-2 Abs. Consistent with the previous results, Vpu had only a marginal effect (~19%) on the levels of surface BST-2, as monitored by the signal issued from the anti-rabbit staining (Figure 8B,D). In contrast, analysis of the signal derived from the mouse Abs revealed a severe downregulation of the restriction factor (~76%), suggesting that Vpu strongly affected the resupply of BST-2 to the surface through newly synthesized proteins (Figure 8C,D). To assess how efficiently the rabbit anti-BST-2 Abs blocked subsequent binding by mouse anti-BST-2 Abs, transfected cells were cultured for 7.5 h in the presence of rabbit anti-BST-2 Abs. Cells were then immediately washed at 4°C to prevent intracellular transport of unlabeled molecules toward the plasma membrane, sequentially stained at 4°C with mouse anti-BST-2 Abs and anti-mouse secondary Abs and then analyzed by flow cytometry. As shown in Figure 8C, no BST-2 signal derived from the mouse Abs could be detected, indicating that rabbit anti-BST-2 Abs efficiently blocked the binding of mouse anti-BST-2 Abs (Figure 8C; dashed line) and confirming that the signal detected with mouse anti-BST-2 (Figure 8C; full lines) indeed represented newly synthesized BST-2 molecules transported to the plasma membrane during the 6 h of Dox treatment. Interestingly, expression of Vpu S52D,S56D, which lacks BST-2 degradation activity, still affected BST-2 resupply to the surface although not as efficiently as WT Vpu (decrease of ~55% for Vpu S52D,S56D versus ~76% for WT Vpu, Figure 8C,D). This indicates that both BST-2 sequestration and degradation are likely contributing to the inhibition of BST-2 resupply at the plasma membrane.

Figure 8. Effect of Vpu on the pool of newly synthesized BST-2.

A) Schematic representation of the experiment. HeLa (B–D) and Jurkat (E–G) cells were transfected with the indicated HIV-1.rtTA.MA-EGFP constructs. Forty-one hours post-transfection, cells were incubated at 37°C for 1 h with saturating amounts of anti-BST-2 rabbit Abs. Unbound Abs were then washed away and fresh media supplemented with 1 μg/mL Dox was added to cells for 6 h at 37°C. Cells were then harvested and either (B and E) immediately stained with secondary anti-rabbit Abs (surface BST-2) or (C and F) incubated with anti-BST-2 mouse Abs for 45 min at 4°C, washed and stained with secondary anti-mouse Abs (neo-synthesized BST-2). In (C), a blocking control is shown (dashed line). D and G) Shown are the quantification of (B and C) and (E and F), respectively. Vpu-negative samples were set at 100% to illustrate Vpu-mediated BST-2 downregulation for both types of staining. The error bars represent the standard deviation calculated from two distinct experiments.

To assess whether these findings could be extended to HIV-1 target cells, we performed similar experiments in transfected CD4+ Jurkat T cells. As shown in Figure 8F,G, newly synthesized BST-2 was similarly downregulated by Vpu in Jurkat cells (decrease of ~87% for WT Vpu and ~58% for Vpu S52D,S56D). Interestingly, in contrast to HeLa cells, the pool of cell-surface BST-2 appeared more significantly downregulated by Vpu and Vpu S52D,S56D (Figure 8E,G; decrease of ~42%, for WT Vpu versus ~17% for Vpu S52D,S56D). Overall, these results suggest that Vpu affects the global expression of cell-surface BST-2 primarily by interfering with the restriction factor resupply to the plasma membrane via newly synthesized proteins but also by increasing BST-2 clearance from the surface, the extent of which might be cell-type dependent.

Discussion

In this study, we investigated the mechanisms through which Vpu antagonizes the restriction imposed by the IFN-inducible host factor, BST-2, on HIV-1 release. Several previous studies reported that Vpu downregulates BST-2 from the surface, the site of its tethering action, and proposed that this process could indeed underlie BST-2 antagonism by Vpu (8,11,29,32,36,39). However, as the surface downregulation of BST-2 is never complete and efficient virus release was not found to correlate with a cell-surface reduction of BST-2 levels in the course of productive infection of CEMx174 and H9 T-cell lines (35), it remains to be determined whether depletion of surface BST-2 levels is a prerequisite for enhanced particle release or a downstream consequence of Vpu function. Using a proviral expression system that allows inducible HIV-1 expression and production, we chronologically monitored the surface downregulation of BST-2 and the enhancement of viral release induced by Vpu. Our data show that surface BST-2 levels decrease to ~60% of their initial levels by 6 h after initiation of Vpu expression and remain stable afterwards, indicating that a new BST-2 equilibrium was established at the cell surface (Figure 2). This rapid downregulation of BST-2 by HIV-1-encoded Vpu is consistent with the results of a recent study, which reported that transfection of plasmids encoding codon-optimized Vpu led to a detectable downregulation of BST-2 6 h after initiation of the transfection (47). Importantly, the emergence of this new surface BST-2 equilibrium coincided with detection of viral particle release in Vpu-expressing cells, suggesting that all processes involved in both Vpu-mediated BST-2 downregulation and antagonism were resolved within the same time-frame. Although we cannot completely rule out that this synchronism is coincidental, it is more likely to represent a causal relationship. Indeed, we were unable to find any evidence that Vpu would exclude BST-2 molecules from HIV-1 budding sites, as an alternative mechanism of antagonism. In fact, even though as previously reported (8,36,44), we observed a reduced colocalization between Gag and BST-2 in Vpu-expressing cells (8,36,44), residual surface BST-2 still colocalized with CD63− and CD81-positive viral budding sites (Figure 3). Thus, by establishing a new BST-2 equilibrium at the cell surface, Vpu appears to decrease the density of BST-2 molecules associated with HIV-1 budding platforms. The rather fast emergence of this new equilibrium is most probably necessary, so that Vpu, which is expressed relatively late during the HIV-1 life cycle, can reduce BST-2 levels before the onset of viral budding. Despite the fact that this reduction of surface BST-2 is incomplete, it seems that the residual levels are incompatible with efficient restriction of HIV-1 release. Whether or not these residual molecules underwent additional qualitative changes that prevent them to restrict the release of nascent virus particles remains to be determined.

Previous studies have shown that BST-2 downregulation does not result from an acceleration of the rate of BST-2 internalization from the surface (32,36,39). One caveat about these analyses is that they were all conducted at steady state, once a new equilibrium was achieved, a condition that could potentially mask effects of Vpu on surface BST-2 before the onset of this new equilibrium. Indeed, our inducible proviral expression system revealed that Vpu slightly accelerated the clearance of BST-2 from the surface (Figure 4). This rather small effect might have been overlooked in previous reports studying the rate of BST-2 internalization, especially because ~40% of surface BST-2 clearance appears to occur within 20–30 min (32,36,39). Interestingly, the rate of decay of cell-surface BST-2, which reflects its post-endocytic transport toward degradative compartments, also appeared slightly increased within the first 6 h of Vpu induction. This could represent an indirect consequence of accelerated BST-2 internalization or, alternatively, could indicate that Vpu enhances the targeting of internalized BST-2 molecules for lysosomal degradation. The accumulation of internalized BST-2 in Rab9- and CD63-positive late endosomes even in the absence of Vpu (Figure 5) suggests that Vpu modulates the natural endocytic and/or post-endocytic BST-2 trafficking, a notion indeed consistent with the role of the ESCRT-0 complex in the lysosomal sorting of the restriction factor both in the presence and absence of Vpu (37). Given that Vpu S52D,S56D did not display a similar acceleration of cell-surface BST-2 clearance and decay at least after 6 h suggests that the recruitment of β-TrCP by Vpu and the recently demonstrated Vpu-mediated ubiquitination of BST-2 (58,59) are likely involved in this process. Ubiquitination of internalized BST-2 could indeed interfere with its recycling back to the plasma membrane, thus forcing its trafficking toward lysosomes for degradation, as reported for the epidermal growth factor receptor (EGFR) (60). However, the poor association of internalized BST-2 to Rab11-positive compartments (Figure 5) and its insensitivity to monensin (Figure 6) do not support such hypothesis, although it is still conceivable that internalized BST-2 could rather engage in a different and/or slower recycling pathway. Alternatively, β-TrCP-mediated ubiquitination of BST-2 could represent a signal triggering its endocytosis and sorting toward lysosomes, a process reminiscent of the SCF-^β–TrCP-mediated degradation of type I IFN receptor (61). These models would perhaps explain the partial inhibition of Vpu-mediated BST-2 downregulation observed upon abrogation of endocytosis either through AP-2 depletion or overexpression of a dominant negative mutant of Dynamin (31,36,62). However, no matter how Vpu accelerates the clearance of BST-2 from the surface, this process cannot account for the overall 40–50% decrease of surface BST-2 observed 6 h post-Vpu induction. This conclusion contrasts with the study of Skasko et al. (47), which conferred to Vpu a more important role in the removal of surface BST-2. Indeed, their data revealed a greater downregulation of the restriction factor (fivefold in 6 h) than could be accounted for by spontaneous surface clearance (twofold in 8 h). These opposite interpretations rest on the magnitude of the Vpu-mediated downregulation of surface BST-2 reported by the two studies, which indeed might be attributed to the use of different expression systems, mainly virus-encoded Vpu constructs in our study versus codon-optimized Vpu-expressing constructs in the case of Skasko et al. (47). It is interesting to note that the effect of Vpu on the pool of cell-surface BST-2 appeared more pronounced in Jurkat CD4+ T cells (Figure 8). Similarly, Vpu appeared to remove quite efficiently BST-2 from the surface of COS cells exogenously expressing the restriction factor (63). Therefore, the efficiency with which Vpu affects the pool of cell-surface BST-2 may be cell-type specific and as such may depend on the presence of cellular determinant(s) that remain to be identified.

As the steady-state level of a cell-surface protein reflects a balance between its clearance from the plasma membrane (i.e. a balance between endocytosis, recycling and degradation) and its resupply through the biosynthetic route, we hypothesized that Vpu could affect the pool of newly synthesized BST-2. In support of this possibility, blocking BST-2 de novo synthesis in IFN-treated HEK293T cells by adding recombinant B18R decreased surface BST-2 by ~38% within 6 h (Figure 7), indicating that surface BST-2 could be efficiently downregulated by reducing its resupply. These results are indeed consistent with the recent data showing that treatment with Brefeldin A or cycloheximide, which prevents resupply of surface proteins, decreased by ~50% the levels of BST-2 at the plasma membrane within 8 h (47). In contrast, Vpu had only a marginal effect on residual BST-2 expressed at the surface of IFN-treated HEK-293T in the presence of B18R, suggesting that a large part of the viral protein antagonist activity depends on the pool of newly synthesized molecules. This notion is indeed further supported by the fact that surface expression of newly synthesized BST-2 is strongly reduced when Vpu expression is induced, while the pool of BST-2 molecules that reached the surface prior to induction of the viral protein is only slightly-to-moderately affected (Figure 8). These data provide direct evidence that Vpu blocks the resupply of BST-2 from the pool of newly synthesized proteins. In contrast to its differential effect on the pool of cell-surface BST-2 in HeLa and Jurkat cells, Vpu blocked the resupply of the restriction factor to the plasma membrane in these cells with a comparable efficiency, suggesting that this strategy is employed by the viral protein independently of the cell type. Interestingly, Vpu S52D,S56D, which is unable to mediate BST-2 degradation but is still able to redistribute the restriction factor particularly in the TGN (32), also blocked BST-2 resupply. Therefore, the inefficient resupply of surface BST-2 in the presence of Vpu appears mainly caused by a slower transport of newly synthesized BST-2 en route to the plasma membrane. Nonetheless, the role of β-TrCP in this process is clearly important because the WT Vpu protein was more efficient than Vpu S52D,S56D at downregulating de novo-synthesized BST-2 molecules (Figure 8). These results suggest that Vpu would optimally block the resupply of BST2 by using a β-TrCP-independent sequestration mechanism, as reported previously (32,38), together with a β-TrCP-dependent process that targets the restriction factor for degradation in the lysosomes (29,37,39).

While this manuscript was in preparation, Schmidt et al. (64) reported that Vpu blocked the recycling and biosynthetic transport of BST-2 to overcome the restriction to virion release. While, overall, our results agree with the conclusion of this study (i.e. effect of Vpu on both the surface and newly synthesized pools of BST-2), they are inconsistent with the conclusion that interference of Vpu with BST-2 biosynthetic and post-endocytic transport pathways does not require recruitment of β-TrCP. Schmidt et al. reported that whereas a phosphoserine-defective mutant of Vpu failed to interfere with both BST-2 transport pathways, siRNA-directed depletion of β-TrCP did not affect significantly the Vpu-mediated block of BST-2 recycling. While the phenotype of Vpu S52A,56A suggests a defect unrelated to β-TrCP, the concurring phenotypes of the Vpu S52D,56D mutant and WT Vpu in β-TrCP2-depleted cells confirm the importance of this cellular factor for optimal Vpu-mediated downregulation of BST-2 at the cell surface (Figures 2 and S1). This discrepancy could be attributed to the different Vpu mutants (S52A,56A versus S52D,56D) used in these studies.

Besides its broad antiviral activity, BST-2 was identified as the ligand of the immunoglobulin-like transcript 7 (ILT7) receptor expressed at the surface of plasmacytoid dendritic cells (pDCs) and, as such, was proposed to play a critical role in regulating the human pDCs’ IFN responses through ILT7 in a negative feedback manner (65). Such a role would imply that a tight and highly dynamic control of cell-surface BST-2 expression must exist. Based on our results, it is tempting to speculate that a reduced BST-2 resupply in response to a decreased IFN concentration in the extracellular milieu coupled to a two-step clearance process from the plasma membrane that includes a rapid internalization step that allows ~40% of molecules to be internalized in 20–30 min (32,36,39) and a slower step that clears remaining molecules within ~24 h (Figure 4) would allow cells to rapidly re-establish surface BST-2 levels present prior to IFN exposure. The results of our current study reveal that Vpu would use a similar strategy to rapidly downregulate BST-2 prior to the onset of viral budding. This strategy, which leads to a net but incomplete depletion of BST-2 from the surface, would involve two processes: (i) a β-TrCP-dependent acceleration of the natural clearance of BST-2 from the plasma membrane, with a variable efficiency depending on the cell type and importantly, (ii) a block of BST-2 resupply via the biosynthetic route through sequestration of molecules in a post-ER compartment, likely the TGN, and their β-TrCP-mediated sorting toward lysosomes for degradation.

Materials and Methods

Plasmid constructs

HxBH10-vpu+, HxBH10-vpu− and HxBH10-vpu S52D,S56D are isogenic infectious molecular clones of HIV-1 that differ only for vpu expression (32). HxBH10.GFP.IRES.nef- variants were constructed by introducing the BamHI-GFP.IRES.nef-XhoI fragment from HxBru.ADA.GFP.IRES.nef- (66) into the HxBH10 molecular clones. To construct HIV-1.rtTA.MA-EGFP-vpu−/vpu+/vpu S52D,S56D, the BssHII-SphI fragment from HIV.rtTA (43) was first replaced by the corresponding fragment from pNL4.3.MA-EGFP (67) to generate HIV-1.rtTA.MA-EGFP. Then, the SalI-BamHI fragments from HxBH10-vpu−, HxBH10-vpu+ or HxBH10-vpu S52D,S56D were introduced into HIV-1.rtTA.MA-EGFP to generate the isogenic HIV-1.rtTA.MA-EGFP.vpu−, HIV-1.rtTA.MA-EGFP.vpu+ and HIV-1.rtTA.MA-EGFP.vpu S52D,S56D proviral constructs. All constructs were validated by automatic DNA sequencing. The vesicular stomatitis virus (VSV) glycoprotein G-expressing plasmid, pSVCMVin-VSV-G, was previously described (32).

Abs and chemical compounds

The anti-Vpu and anti-BST-2 rabbit sera were described previously (32). Monoclonal anti-p17 Abs (catalog no. HB-8975), which recognize p17 but not the p55^Gag precursor, and monoclonal anti-p24 Abs (catalog no. HB9725) were isolated from the supernatant of cultured hybridoma cells obtained from the American Type Culture Collection (ATCC). Polyclonal sheep anti-TGN46 (Serotec), mouse anti-BST-2 (Abnova), monoclonal mouse anti-CD81 (BD Pharmingen), mouse anti-CD63 [Hybridoma Bank (NICHD, University of Iowa)], mouse anti-LDLR (Santa Cruz Biotechnology), rabbit anti-Rab11a (Abcam) and mouse anti-Rab5 (BD transduction laboratories), mouse anti-Rab9 (Calbiochem), mouse anti-CD59 (BD Pharmingen) Abs, PE-coupled anti-p24 (Beckman Coulter) and transferrin conjugates-594 (Molecular Probe) were all obtained, as indicated, from commercial sources. All secondary Alexa-conjugated immunoglobulin G (IgG) Abs were obtained from Invitrogen. Human IFN-α and recombinant B18R were purchased from PBL Interferon Source and eBioscience, respectively, concanamycin A was obtained from MP Biochemicals, whereas Dox and monensin were purchased from Sigma. All reagents were stored according to the manufacturer’s instructions.

Cells and transfection

HEK 293T, HeLa and Jurkat cells were obtained from ATCC. HEK-blue IFN-α/β^™ cells were obtained from InvivoGen. All cells were maintained as described previously (14).

The β-TrCP2-depleted cell line was generated by transducing HeLa cells with a pGIPZ lentiviral vector expressing a shRNAmir targeting specifically β-TrCP2 (shRNAmir #187 provided by Open Biosystems). The sequence of β-TrCP2 targeted by the shRNAmir is TGCCAATTATCTGTTTGAAATA, located in the 3′ UTR of the complete β-TrCP2 cDNA. Lentiviral vectors were produced in 293T cells according to the manufacturer’s instructions. Transduced cells were selected using 1 μg/mL puromycin (Sigma-Aldrich). The control HeLa cell line was generated by the same procedure using a non-silencing pGIPZ lentiviral shRNAmir control obtained from Open Biosystem.

HEK 293T, HeLa and Jurkat cells were transfected using the calcium-phosphate method, lipofectamine 2000^™ and lipofectamine LTX^™ (Invitrogen), respectively. Transfected cells were analyzed 48 h post-transfection.

Virus particle release assay

The viral particle release assay was described previously (14). Briefly, supernatants of transfected cells were clarified by centrifugation and filtered through a 45-μm filter. Virus particles were pelleted by ultracentrifugation onto a 20% sucrose cushion in PBS for 2 h at 130 000 × g at 4°C. Viruses and cells were lysed in radio-immunoprecipitation assay (RIPA-DOC) buffer (10 mM Tris pH 7.2, 140 mM NaCl, 8 mM Na₂HPO₄, 2 mM NaH₂PO₄, 1% Nonidet-P40, 0.5% sodium dodecyl sulfate, 1.2 mM deoxycholate). Proteins from lysates were resolved on 12.5% SDS–PAGE, electro-blotted and analyzed by western blot as described previously (14).

Flow cytometry

Cells were washed in PBS, resuspended at a concentration of 5 × 10⁶ cells/mL and stained with the specific rabbit (or mouse, where indicated) polyclonal anti-BST-2 rabbit serum for 45 min at 4°C. Cells were then washed and stained using appropriate Alexa Fluor-647-coupled secondary Abs for 30 min at 4°C, washed and fixed in 2% paraformaldehyde (PFA). Jurkat cells were also stained with a LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Invitrogen) to exclude dead cells. EGFP-expressing cells or p24-positive cells (as determined by an intracellular staining following Beckman and Coulter’s procedures) were then analyzed for surface BST-2 expression by flow cytometry. Normal rabbit serum served as a staining control (gray shades in histograms). Fluorescence intensities were acquired using a FACScalibur flow cytometer (BD Biosciences) and data were analyzed using the FlowJo software v. 7.25 (Treestar). MFI values presented in histograms correspond to the specific signal obtained after subtraction of the MFI value from the staining control.

Semi-quantitative RT PCR

The depletion of β-TrCP2 was controlled by reverse transcriptase polymerase chain reaction (RT PCR). Total cellular RNA was extracted using the QIAGEN® RNA extraction kit. After DNase (Invitrogen) treatment, total RNA was converted to cDNA using the Superscript II reverse transcriptase kit (Invitrogen). cDNAs were engaged in the PCR reaction using cloned Pfu (Agilent). The sequences of the primers used for β-TrCP2 cDNA amplification were: 5′-ACGAATGGTACGCACTGATCC-3′(sense) and 5′-ACTTCACCCGTGTTCACATCC-3′ (antisense). GADPH was also amplified as control using primer sequences 5′-GCCATCAATGACCCCTTCATT-3′ (sense) and 5′-TTGACGGTGCCATGGAATTT-3′ (antisense). The products of the PCR reactions were analyzed using a conventional semi-quantitative approach on 1.5% agarose gel.

Cell-surface BST-2 clearance assay

Transfected cells were pre-incubated in DMEM + 5% FBS supplemented with the rabbit polyclonal anti-BST-2 serum for 1.5 h at 37°C. Cells were then washed and incubated at 37°C in DMEM + 5% FBS with 1 μg/mL Dox for 0, 3, 6, 12 or 24 h to initiate internalization of Abs–BST-2 complexes. To measure the BST-2 levels remaining at the surface, cells from each sample were washed and harvested 48 h post-transfection in cold PBS–ethylenediaminetetraacetic acid (EDTA), stained with an anti-rabbit Alexa Fluor-647 secondary Ab for 30 min at 4°C, washed, fixed in 2% PFA and analyzed by flow cytometry.

To detect intracellular BST-2, harvested cells were fixed, permeabilized using Cytofix/Cytoperm^™ and intracellularly stained with the appropriate secondary Abs following the manufacturer protocol (BD Biosciences) and analyzed by flow cytometry.

Production of VSV-G-pseudotyped HIV-1 virus

HEK 293T cells were transfected with HxBH10 proviral constructs and pSVCMVin-VSV-G as described previously (32). Supernatants of transfected cells were clarified, filtered and pelleted by ultracentrifugation as described above and resuspended in DMEM supplemented with 10% bovine serum (FBS). Viruses were titrated using a standard MAGI assay as previously described (32).

Confocal microscopy

Jurkat cells were infected with VSV-G-pseudotyped HxBH10-derived viruses at a multiplicity of infection (MOI) of 0.125 with polybrene. Forty-eight hours post-infection, cells were immunostained with the indicated Abs (BST-2, CD81, CD59 and/or CD63) for 45 min at 4°C prior to extensive washes. Cells were then centrifuged at 4°C for 20 min on poly-D-lysine-treated coverslips and fixed for 30 min in 4% PFA. To detect p17, fixed cells were permeabilized in Triton 0.2% for 5 min, incubated for 2 h at 37°C in 5% milk-PBS containing anti-p17 Abs, washed and incubated with the appropriate secondary Ab for 30 min at room temperature. Steady-state intracellular localization of BST-2 was performed as described previously (14). To study intracellular localization of internalized BST-2, coverslip-seeded HeLa cells were transfected with HIV-1.rtTA.MA-GFP proviral constructs. Thirty-six hours later, cells were incubated at 37°C for 0–12 h in media containing rabbit anti-BST-2 serum as well as 1 μg/mL Dox to trigger viral protein expression. Twelve hours later, transfected cells were fixed, permeabilized with 0.2% Triton and stained for the indicated cellular markers for 2 h at 37°C. Following extensive washes, cells were stained at room temperature for 30 min with Alexa Fluor-594 or -647-coupled secondary Abs to detect internalized BST-2 and the cellular markers. All analyses were acquired using a 63× Plan Apochromat oil immersion objective with an aperture of 1.4 on an LSM710 Observer Z1 laser scanning confocal microscope coupled with a Kr/Ar laser (488-, 594- and 633-nm lines) (Zeiss).

Type I IFN detection

Twenty microliters of the supernatant of HEK293T was added to a suspension of 50 000 HEK-blue IFN-α/β^™ cells in 180 μL DMEM containing 10% heat-inactivated FBS and incubated for 24 h at 37°C. To quantify secreted placental alkaline phosphatase (SEAP) in culture supernatants, 40 μL of HEK-blue cell supernatant was added to 160 μl of QUANTI-Blue reagent (InvivoGen) and incubated for 20 min at 37°C. SEAP levels were evaluated using a spectrophotometer at 620 nm.

Supplementary Material

Figure S1. Figure S1: Recruitment of β-TrCP2 by Vpu optimizes cell-surface BST-2 downregulation.

A) Semi-quantitative analysis of β-TrCP2 depletion in HeLa cells by RT PCR. Total RNA isolated from cell lines stably expressing the indicated shRNA was used to monitor the depletion of β-TrCP2. The cellular gene GADPH was used as a specificity and loading control. B) Control and β-TrCP2-depleted HeLa cells were transfected with the indicated HIV-1.rtTA.MA-EGFP constructs and analyzed 48 h post-transfection for surface BST-2 levels on p24-positive cells by flow cytometry. Prior to analysis, cells were treated for the indicated period of time with 1 μg/mL of Dox. Error bars represent the standard deviation calculated from three independent experiments.

Figure S2. Figure S2: Analysis of BST-2 subcellular localization in fixed cells.

HeLa cells were fixed, permeabilized, stained for intracellular BST-2 (green) and the indicated intracellular markers (red), and then incubated with appropriate secondary Abs. Nuclei were counterstained with DAPI (blue). Cells were analyzed by confocal microscopy. ConA indicates a condition in which cells were pretreated before fixation with 50 nM of the lysosome acidification inhibitor concanamycin A. White arrow heads highlight examples of colocalization between BST-2 and specific cellular markers. White bars = 10 μm.

Figure S3. Figure S3: Pathway followed by internalized BST-2 molecules to reach lysosomal compartments.

A and B) HeLa cells were incubated at 37°C for 0–6 h in the presence of anti-BST-2 Abs, fixed, permeabilized and then stained for (A) TGN46 or (B) CD63. Stained cells were washed and incubated with appropriate secondary Abs to detect cellular markers (red) as well as internalized BST-2 (green). Nuclei were counterstained with DAPI (blue). Cells were analyzed by confocal microscopy. White arrow heads highlight examples of colocalization between BST-2 and CD63. White bars =10 μm. C) Quantification of BST-2 colocalization with cellular markers. The values (%) represent the percentage of BST-2 (green pixels) overlapping with each cellular marker (red pixels). Error bars indicate the standard deviation of the mean from the quantitative analysis of at least 25 distinct cells.

FIgure S4. Figure S4: Recombinant B18R prevents BST-2 expression at the surface of type 1 IFN-treated HEK293T cells.

A) HEK293T cells were treated with 1000 U/mL of IFN-α for 1 h at 37°C. Cells were then washed and reincubated for 24 h at 37°C in the presence of 0–100 ng/mL of recombinant B18R. Cells were then harvested, stained with an anti-BST-2Abs, washed, stained with an appropriate secondary Ab and analyzed by flow cytometry. MFI values for each condition are depicted within the histogram. B) HEK-blue IFN-α/β^™ cells were cultured for 24 h in the presence of different dilutions of the media collected from (A). Supernatants from HEK-blue IFN-α/β^™ cells were subsequently incubated in the presence of QUANTI-blue to detect the levels of SEAP secreted upon exposure of HEK-blue IFN-α/β^™ cells to type I IFN. The graph depicts the relative light unit (RLU) detected by luminometry and is indicative of the levels of active type I IFN secreted by HEK293T cells.