Antibody-Mediated Enhancement of HIV-1 and HIV-2 Production from BST-2/Tetherin-Positive Cells

Abstract

BST-2/CD317/HM1.24/tetherin is a B-cell antigen overexpressed on the surface of myeloma cell lines and on neoplastic plasma cells of patients with multiple myeloma. Antibodies to BST-2 are in clinical trial for the treatment of multiple myeloma and are considered for the treatment of solid tumors with high BST-2 antigen levels. Functionally, BST-2 restricts the secretion of retroviruses, including human immunodeficiency virus type 1, as well as members of the herpesvirus, filovirus, and arenavirus families, presumably by tethering nascent virions to the cell surface. Here we report that BST-2 antibody treatment facilitates virus release from BST-2⁺ cells by interfering with the tethering activity of BST-2. BST-2 antibodies were unable to release already tethered virions and were most effective when added early during virus production. BST-2 antibody treatment did not affect BST-2 dimerization and did not reduce the cell surface expression of BST-2. Interestingly, BST-2 antibody treatment reduced the nonspecific shedding of BST-2 and limited the encapsidation of BST-2 into virions. Finally, flotation analyses indicate that BST-2 antibodies affect the distribution of BST-2 within membrane rafts. Our data suggest that BST-2 antibody treatment may enhance virus release by inducing a redistribution of BST-2 at the cell surface, thus preventing it from accumulating at the sites of virus budding.

INTRODUCTION

BST-2 is an interferon (IFN)-inducible host factor responsible for the inhibition of human immunodeficiency virus type 1 (HIV-1) release (37, 58). A current model suggests that BST-2 tethers mature virions to the cell surface (37). This function of BST-2 is antagonized by HIV-1 Vpu. Recent data suggest that the human BST-2 transmembrane (TM) domain is crucial for sensitivity to HIV-1 Vpu (10, 15, 33, 34, 42, 45). This is consistent with the earlier reported critical importance of the Vpu TM domain for the regulation of virus release (51). More recently, simian immunodeficiency virus (SIV) Nef and the Env glycoprotein of some HIV-2 and SIV isolates were found to have Vpu-like activity capable of antagonizing BST-2 (16, 19, 22, 29, 47, 64, 65). Unlike Vpu, however, Nef and Env do not interact with the BST-2 TM domain but target its cytoplasmic domain and ectodomain, respectively (16, 22, 29, 30, 58, 64, 65), indicating that BST-2 offers multiple avenues for functional neutralization by viral factors.

BST-2 was originally identified as a membrane protein in terminally differentiated human B cells of patients with multiple myeloma (14, 38). BST-2 is a 30- to 36-kDa type II TM protein consisting of 180 amino acids (21). The protein is predicted to have an N-terminal TM domain and a C-terminal glycosyl-phosphatidylinositol (GPI) anchor (28). These two domains are separated by approximately 120 residues that constitute the protein's ectodomain and are predicted to form a rod-like coiled-coil structure (20, 50, 63). The BST-2 ectodomain encodes three cysteine residues (4, 14, 38, 42). Each of these cysteines can independently contribute to the formation of cysteine-linked dimers, which is critical for BST-2 function (4, 42). BST-2 is also modified by N-linked glycosylation (4, 28, 38); however, the functional significance of BST-2 glycosylation for inhibition of virus release is still debated (4, 42). BST-2 protein associates with lipid rafts at the cell surface and on internal membranes, presumably the trans-Golgi network (12, 19, 28, 31). Vpu has a tendency to associate with lipid rafts as well, and the protein accumulates in the Golgi/trans-Golgi network and early endosomes (12, 51, 59), and it is likely that Vpu's antagonism of BST-2 occurs in these intracellular compartments (5, 11, 12).

We have developed a polyclonal antibody recognizing endogenously, as well as exogenously, expressed BST-2 (35). The antibody was raised against the ectodomain of BST-2 and reacts with BST-2 in a variety of human cell types. Since BST-2's ectodomain contains functionally critical structural elements, including a coiled-coil domain and three cysteines involved in protein dimerization, we hypothesized that antibody binding to BST-2 could affect the formation of cysteine-linked dimers and/or affect coiled-coil-mediated protein-protein interactions and thus interfere with BST-2 function. Here, we analyzed the potential virus release-promoting effect of BST-2 antibody treatment. Indeed, we found that treatment of HeLa cells with BST-2 antibody neutralized BST-2 activity and significantly augmented virus release. Interestingly, antibody treatment not only increased the release of Vpu-deficient virus but also enhanced the release of wild-type (WT) HIV-1 virions. This suggests that Vpu expressed from WT NL4-3 is not sufficient to fully negate the inhibitory effect of endogenous BST-2 in HeLa cells. BST-2 antibody-induced enhancement of virus release was most efficient when the antibody was added early during virus assembly. Furthermore, kinetic studies demonstrate that virus particles already tethered to the cell surface prior to antibody addition were not released. We therefore conclude that antibody neutralization of BST-2's tethering activity requires antibody binding prior to BST-2's engagement with budding viruses. In contrast to Vpu, the expression of which is associated with reduced levels of cell surface BST-2 (10, 11, 17, 29, 34, 35, 46, 48, 58), antibody treatment did not cause cell surface down-regulation of BST-2 but, to the contrary, stabilized BST-2 at the cell surface. However, BST-2 antibody treatment reduced nonspecific shedding of BST-2 and limited its encapsidation into virions. Finally, flotation analyses indicate that BST-2 antibodies affect the association of BST-2 with membrane rafts. Our data suggest that BST-2 antibody treatment may enhance virus release by inducing a redistribution of BST-2 at the cell surface that prevents it from accumulating at the sites of virus budding.

MATERIALS AND METHODS

Cell culture, plasmids, and transfection.

HeLa cells and 293T cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (FBS). Activated peripheral blood mononuclear cells (PBMCs) were kindly provided by Olivia Donau. PBMCs were isolated from the leukapheresis-treated blood of HIV-seronegative donors, activated with concanavalin A (for 20 h), and cultured in RPMI containing 10% FBS and interleukin-2. The full-length molecular clone pNL4-3 (2) was used for the production of WT HIV-1. A Vpu-defective variant, pNL4-3/Udel (26), was used for the production of Vpu-deficient HIV-1 (Udel). For HIV-2, the pROD1014 and pROD1014.598TA constructs were used (6). pROD1014 exhibits a Vpu⁻ phenotype, while pROD1014.598TA has a Vpu⁺ phenotype. The latter clone differs from the pROD1014 clone by a single amino acid change at position 598 of the Env protein (6). All transfections were performed using TransIT-LT1 transfection reagent according to the manufacturer's recommendations (Mirus Corp., Madison, WI).

Antisera and purification of IgG.

A polyclonal antibody to BST-2 was raised in a rabbit immunized with recombinant BST-2 protein as previously described (35). The antibody can be obtained through the NIH AIDS Research and Reference Reagent Program (https://www.aidsreagent.org) (catalog no. 11721). A polyclonal Vpu antiserum was raised in rabbits against the hydrophilic C-terminal cytoplasmic domain of Vpu and was made available through the NIH AIDS Research and Reference Reagent Program (catalog no. 969). HIV-1- or HIV-2-positive patient sera were obtained from the NIH AIDS Research and Reference Reagent Program (catalog no. 3957) and used to detect HIV-specific Gag proteins. Tubulin was identified using a monoclonal antibody to α-tubulin (Sigma-Aldrich, Inc., St. Louis, MO; catalog no. T-9026). Anti-caveolin 1 (Cav-1) and anti-transferrin receptor (TfR) antibodies used in the flotation analyses were obtained from BD Biosciences (San Diego, CA; catalog no. 610060 and 612125, respectively). A Melon Gel IgG purification kit (Thermo Scientific, Rockford, IL) was used for the purification of IgG from preimmune or immune serum. Purified IgG fractions were subsequently washed twice with phosphate-buffered saline (PBS) and concentrated to their original volume using Amicon centrifugal filter units (10,000 molecular weight cutoff; Millipore Corp., Billerica, MA). IgG concentrations were determined by Coomassie staining of a gel using a bovine serum albumin (BSA) standard.

Fluorescence-activated cell sorter (FACS) analysis.

HeLa cells were washed twice with ice-cold PBS containing 20 mM EDTA and then incubated in the same buffer for 15 min at 4°C to detach the cells. After that, cells were washed twice in ice-cold PBS containing 1% BSA. Cells were blocked for 10 min with mouse IgG (1.25 μg/ml; Millipore Corp., Billerica, MA). Cells were incubated with primary antibody (anti-BST-2) for 30 min at 4°C and washed twice with ice-cold 1% BSA-PBS; this was followed by the addition of allophycocyanin (APC)-conjugated anti-rabbit IgG secondary antibody (Jackson ImmunoResearch Laboratories Inc., West Grove, PA) in 1% BSA-PBS. Incubation was for 30 min at 4°C in the dark. Cells were then washed twice with ice-cold 1% BSA-PBS and fixed with 1% paraformaldehyde in PBS. Finally, cells were analyzed on a FACScalibur (BD Biosciences). Data analysis was performed using CellQuest Pro (BD Biosciences) and Flow Jo (Tree Star Inc., Ashland, OR).

Virus production and infections.

HIV-1 stocks were generated by transfection of 293T cells with 5 μg of the pNL4-3 WT or Vpu-defective pNL4-3/Udel (Udel) molecular clone. Virus-containing supernatants were harvested 48 h after transfection, filter sterilized, and assessed for infectivity by measuring reverse transcriptase (RT) activity as described previously (62). PBMCs from two donors were infected (1 × 10⁶ PBMCs/ml in 24-well plates) with NL4-3 WT or NL4-3/Udel virus. Input virus was normalized for equal units of RT activity.

Immunoblot analysis.

For immunoblot analysis of cell-associated proteins, whole-cell lysates were prepared as follows. Cells were washed once with PBS, suspended in PBS (250 μl/5 × 10⁶ cells), and mixed with an equal volume of sample buffer (4% sodium dodecyl sulfate, 125 mM Tris-HCl, pH 6.8, 10% 2-mercaptoethanol (ME), 10% glycerol, 0.002% bromophenol blue). To analyze virus-associated proteins, 3.5 ml of cell-free filtered supernatant from transfected 293T cells was pelleted (60 min, 35,000 rpm) through a 20% sucrose cushion (1.5 ml) in an SW55 rotor. The concentrated virus pellet was suspended in PBS (175 μl) and mixed with an equal volume of sample buffer. For analysis of BST-2 under nonreducing conditions, samples were prepared as described above, except that ME was omitted from the sample buffer. Proteins were solubilized by heating for 10 to 15 min at 95°C with occasional vortexing. Cell and virus lysates were subjected to SDS-PAGE; proteins were transferred to polyvinylidene difluoride membranes and reacted with appropriate antibodies as described in Results. Membranes were then incubated with horseradish peroxidase-conjugated secondary antibodies (Amersham Biosciences, Piscataway, NJ), and proteins were visualized by enhanced chemiluminescence (Amersham Biosciences).

Metabolic labeling and pulse-chase analysis of virus release.

Cells were transfected in T-25 flasks (∼3 × 10⁶ cells) with 5 μg of NL4-3 WT or NL4-3/Udel DNA. After 24 h, cells were scraped and washed with PBS. Cells were suspended in 1 ml of RPMI lacking methionine and cysteine (MP Biomedical, Solon, OH) containing 5% fetal calf serum (FCS) and incubated for 30 min at 37°C to deplete the intracellular methionine pool. Cells were then labeled for 30 min at 37°C in 500 μl methionine-free RPMI containing 5% FCS supplemented with 150 μCi of [³⁵S]Expre³⁵S³⁵S protein labeling mix (Perkin-Elmer, Waltham, MA; catalog no. NEG072). After the labeling period, the unincorporated isotope was removed and equal aliquots of cells were mixed with complete RPMI containing 10% FCS and chased for various times as indicated in Results. Cells and virus-containing supernatants were harvested at each time point and stored at −80°C until all samples had been collected. For steady-state labeling of viral proteins, cells were treated as described above but labeled for 90 min at 37°C without a subsequent chase. For immunoprecipitation of intracellular and virus-associated Gag proteins, cells were lysed in 500 μl of NP-40-DOC buffer (20 mM Tris-HCl, pH 8, 120 mM NaCl, 2 mM EDTA, 0.5% deoxycholate [DOC], 1% NP-40), virus-containing supernatants were filtered through 0.45-μm cellulose acetate Spin-X centrifuge tube filters (Corning Costar Corporation, Cambridge, MA) to remove residual cells and cell debris, and virus was lysed by adding 0.25 volume of 5× NP-40-DOC buffer (100 mM Tris-HCl, pH 8, 600 mM NaCl, 10 mM EDTA, 2.5% DOC, 5% NP-40).

Immunoprecipitation and protein quantitation.

Cell and virus lysates were immunoprecipitated (1 h, 4°C on a rotator) with HIV-1 patient serum or a 2:1 mixture of an HIV-1/HIV-2 patient serum pool immobilized to protein A-Sepharose beads (Sigma-Aldrich, Inc., St. Louis, MO; catalog no. P-3391). Immunoprecipitates were solubilized by boiling in sample buffer (2% SDS, 1% ME, 1% glycerol, 65 mM Tris-hydrochloride, pH 6.8) and separated by SDS-PAGE. Gels were fixed and dried. Gels were exposed to Kodak XMR film, and proteins were visualized by fluorography. For protein quantitation, gels were exposed to imaging plates and analysis of the relevant bands was performed using a FujiFilm FLA-7000 Phosphoimager.

Direct and indirect immunofluorescence.

HeLa cells were trypsinized, plated on coverslips (18 mm, no. 1) in a 12-well plate, and grown overnight at 37°C (5% CO₂). Cells were treated with antibody as indicated in Results. Cells were either fixed in methanol (−20°C, 10 min) and then stored in PBS at 4°C until use or fixed with 1% paraformaldehyde (in PBS) for 20 min at room temperature. Paraformaldehyde-fixed samples were not permeabilized prior to staining. Samples were stained in a humid chamber by placing coverslips on Parafilm and overlaying the coverslip with 100 μl of antibody solution. Antibody incubations were for 30 min at room temperature. For indirect immunofluorescence, cells were first exposed to primary antibody (i.e., rabbit anti-BST-2; 1:100) in PBS containing 1% BSA. Cells were then washed in a staining rack in PBS (150 ml) for 5 min prior to incubation with Cy2- or Texas Red-conjugated donkey anti-rabbit secondary antibodies (Jackson ImmunoResearch Laboratories Inc., West Grove, PA) diluted 1:100 in PBS containing 1% BSA. Finally, cells were washed twice in PBS and mounted on microscope slides using Glycerol Gelatin (Sigma-Aldrich, Inc., St. Louis, MO). For direct immunofluorescence, cells were directly stained with secondary antibodies. Images were acquired using a Zeiss LSM410 inverted confocal microscope equipped with a Plan-Apochromat 63×/1.44 oil immersion objective. Images were deconvoluted using Huygens Professional (v. 4.0) software.

Membrane flotation analysis (raft association).

Flotation analysis was done essentially as described previously (39). Cells (5 × 10⁶) were pelleted (2,000 × g for 2 min) and resuspended in 300 μl of 10 mM Tris-HCl, pH 7.5, supplemented with 4 mM EDTA and Complete protease inhibitor cocktail (Roche Diagnostics Corp., Indianapolis, IN). After 10 min of incubation on ice, cells were sonicated for 10 s and centrifuged for 3 min at 2,000 × g at 4°C in a microcentrifuge to remove insoluble material and nuclei. The postnuclear supernatants (120 μl) were mixed with 120 μl of TNE lysis buffer (100 mM Tris-HCl, 600 mM NaCl, 16 mM EDTA) containing 0.5% Triton X-100 and incubated on ice for 20 min. A total of 200 μl of each lysate was mixed with 1 ml of 85.5% (wt/vol) sucrose in TNE lysis buffer, placed at the bottoms of ultracentrifuge tubes, and overlaid with 1.5 ml of 65% (wt/vol) sucrose, followed by 0.75 ml of 20% (wt/vol) sucrose and 0.75 ml of 10% (wt/vol) sucrose (in TNE lysis buffer). The samples were centrifuged at 4°C in an SW55 rotor for 3.5 h at 35,000 rpm. Eight equal fractions (550 μl each) were collected from the top, mixed with 4× sample buffer (183 μl), and heated to 95°C for 10 min. Samples were analyzed by immunoblotting with appropriate antibody as specified in Results.

Knockdown of BST-2 expression in HeLa cells by siRNA.

BST-2-specific small interfering RNA (siRNA; Accell SMARTpool, catalog no. E-011817-00) and nonspecific control siRNA (Accell control siRNA, catalog no. D-001910-01-20) were purchased from Dharmacon (Thermo Scientific Dharmacon, Lafayette, CO). HeLa cells (2 × 10⁶ in 25-cm² flask) were transfected with siRNAs (25 μM) using Mirus TransIT-TKO transfection reagent (catalog no. MIR2150; Fisher Scientific, Pittsburgh, PA) according to the manufacturer's instructions. Twenty-four hours later, cells were transfected with 5 μg of NL4-3 or NL4-3/Udel DNA and incubated for another 24 h. Cells were then harvested by scraping and suspended in 4.5 ml of RPMI lacking methionine and cysteine. One aliquot of the cells (1.5 ml) was pelleted and used to prepare whole-cell lysates, which were then processed for immunoblot analysis as described above. The remaining cells (2 × 1.5 ml) were processed for metabolic labeling with or without antibody pretreatment as described above (metabolic labeling and pulse-chase analysis of virus release). Purified IgG (5 μg) was added 10 min prior to the beginning of the labeling period.

RESULTS

BST-2-specific antibodies enhance HIV-1 release from BST-2-expressing cells.

BST-2 is expressed at the cell surface and is believed to inhibit virus release by tethering nascent virions to the plasma membrane (for a review, see reference 3). We recently generated a polyclonal antibody to human BST-2 (35). Our antibody is directed against the ectodomain of BST-2 and reacts with exogenously and endogenously expressed protein. Since BST-2 surface expression is thought to be critical for its antiviral effect and since BST-2's ectodomain encompasses functionally critical sequences, in particular, cysteine residues required for BST-2 dimerization (4, 42), we tested the ability of ectopically applied antibody to antagonize BST-2 function. In an initial experiment, HeLa cells were transfected with WT HIV-1 NL4-3 or a vpu-defective derivative, NL4-3/Udel. Cells were metabolically labeled for 90 min in a total volume of 500 μl with [³⁵S]methionine as described in Materials and Methods in the presence of either preimmune serum (Fig. 1A, lanes 1 and 3) or various volumes of BST-2-specific antiserum, as indicated (Fig. 1A, lanes 2 and 4 to 6). The total volume of antiserum added to each culture was 50 μl and was adjusted as appropriate by adding appropriate volumes of preimmune serum. Cell lysates and virus-containing supernatants were immunoprecipitated using HIV-positive patient serum and processed for fluorography as described in Materials and Methods. As anticipated, the addition of preimmune serum did not overcome the Vpu dependence of HIV-1 particle release (Fig. 1A, compare lanes 1 and 3). Interestingly, addition of BST-2-specific antibody reversed the inhibitory effect of BST-2 and resulted in a dose-dependent increase in virus release in the absence of Vpu (Fig. 1A, lanes 4 to 6). Of note, even the smallest amount of BST-2-specific antibody was sufficient to increase the release of Udel virus to the level observed for NL4-3 WT virus (Fig. 1A, compare lanes 1 and 4). Therefore, this amount of antibody was used in subsequent experiments unless stated otherwise. Adding more antibody led to further enhancement of virus release. These results demonstrate that the antagonistic effect of BST-2 on HIV-1 release from HeLa cells can be neutralized by the ectopic application of BST-2-specific antibody. Surprisingly, the release of WT virus was also enhanced by BST-2 antiserum (Fig. 1A, lane 2) and the overall efficiency of virus release in the presence of 50 μl of BST-2 antiserum was indistinguishable for the NL4-3 WT and Udel viruses (Fig. 1A, compare lanes 2 and 6). It is unclear why BST-2 antibody treatment leads to enhanced release of NL4-3 WT virions. However, it is conceivable that levels of Vpu expressed by NL4-3 WT virus are insufficient to fully counteract the activity of endogenous BST-2 in HeLa cells. This conclusion is supported by the finding that siRNA silencing of BST-2 in HeLa cells can result in the increased release of not only Vpu-deficient virus but WT virus as well (58).

Fig. 1.

BST-2-specific antibody enhances HIV-1 release. (A) HeLa cells were transfected with 5 μg of pNL4-3 WT (lanes 1 and 2) or pNL4-3/Udel (lanes 3 to 6). Twenty-four hours later, cells were incubated for 30 min in medium lacking methionine to deplete the endogenous methionine pool and then metabolically labeled for 90 min with [³⁵S]methionine in a total volume of 500 μl. Rabbit sera were added as follows: lanes 1 and 3, 50 μl of preimmune serum; lanes 2 and 6, 50 μl of BST-2 antiserum; lanes 4 and 5, 5 and 10 μl, respectively, of BST-2 antiserum adjusted to 50 μl with preimmune serum. Cell lysates and cell-free supernatants were prepared as described in Materials and Methods and subjected to immunoprecipitation with HIV-positive patient serum. Immunoprecipitated proteins were separated by SDS-PAGE and visualized by fluorography. (B) HeLa cells were transfected with 5 μg of pNL4-3 WT (lanes 1 to 4) or pNL4-3/Udel (lanes 5 to 8). Cells were metabolically labeled 24 h later for 90 min with [³⁵S]methionine in a total volume of 500 μl supplemented with 20 μl of PBS (no antiserum, lanes 1 and 5) or 5 μg each of purified IgG from preimmune serum (pre-I, lanes 2 and 6), nonspecific immune IgG (Ctrl IgG, lanes 3 and 7), or BST-2-specific IgG (lanes 4 and 8). Immunoprecipitation and visualization of the proteins were done as for panel A.

Our initial experiment involved whole rabbit antiserum. To rule out nonspecific effects of the serum and to make sure the virus release-promoting effect was caused by the BST-2 antibodies, we purified IgG from antiserum as described in Materials and Methods. We purified IgG from preimmune serum (Fig. 1B, pre-I), an unrelated rabbit immune serum (Fig. 1B, Ctrl IgG), as well as BST-2 antiserum (Fig. 1B, BST-2 IgG). Purified IgG preparations were reconstituted with PBS. To study the effects of purified IgG on virus release, HeLa cells were transfected with the NL4-3 WT and NL4-3/Udel viruses and metabolically labeled as described for panel A. Purified IgGs (5 μg of each) were added during the labeling period, and cell lysates and virus-containing supernatants were processed as described for Fig. 1A. A sample treated with PBS in lieu of antibody was included as a negative control (Fig. 1B, no IgG). As predicted, only BST-2 IgG enhanced the release of Vpu-deficient virus (Fig. 1B, lane 8) while addition of PBS (Fig. 1B, lane 5), preimmune serum (Fig. 1B, lane 6), or unrelated immune Ig (Fig. 1B, lane 7) had no effect. Consistent with the results in Fig. 1A, BST-2 IgG also enhanced the release of WT NL4-3 (Fig. 1B, lane 4) relative to the PBS-treated sample, while preimmune or control immune IgG had no enhancing effect (Fig. 1B, lanes 1 to 3). The slightly reduced virus output in the sample treated with control IgG (Fig. 1B, lane 3) is due to overall lower viral protein expression in that sample (see cellular fraction) and was not consistently seen in replicate experiments. These results confirm that the increase in virus release observed after the addition of BST-2 antibody is indeed attributable to BST-2-specific immunoglobulin.

Knockdown of BST-2 negates the antibody-induced enhancement of virus release.

To verify that the enhanced virus release observed after BST-2 antibody treatment was specific for BST-2 and not caused by a general nonspecific effect of the antibody on the virus-producing cells, we silenced BST-2 expression in HeLa cells prior to transfection with NL4-3 or NL4-3/Udel virus as detailed in Materials and Methods. A fraction of the cells was processed for immunoblotting to monitor BST-2 down-regulation (Fig. 2A, BST-2 siRNA). BST-2 expression was compared to that in untreated cells (Fig. 2A, no siRNA) and cells treated with nonspecific siRNA (Fig. 2A, Ctrl siRNA). Indeed, siRNA treatment effectively suppressed BST-2 expression, while treatment with control siRNA had no effect on BST-2 levels in HeLa cells. As expected, siRNA treatment did not affect the production of viral Gag or Vpu protein (Fig. 2A, CA and Vpu).

Fig. 2.

Knockdown of BST-2 negates antibody-induced enhancement of virus release. HeLa cells were transfected with BST-2-specific siRNA (BST-2 siRNA) or control siRNA (Ctrl siRNA) or left untreated (no siRNA), as described in Materials and Methods. One set of cells was transfected 24 h later with 5 μg of NL4-3 WT DNA; the second set was transfected with 5 μg of pNL4-3/Udel DNA. Cells were harvested 24 h later. (A) One portion of the cells was processed for immunoblotting to identify BST-2 (top panel), HIV-1 capsid (middle panel), or Vpu (lower panel). (B) The remaining cells were divided in half and metabolically labeled for 90 min with [³⁵S]methionine/cysteine in the presence or absence of BST-2-specific IgG (5 μg/ml). Gag proteins present in cell lysates and viral supernatants were immunoprecipitated with HIV-positive patient serum, separated by SDS-PAGE, and visualized by fluorography. Gels shown for viral supernatants were exposed approximately three times longer than the cell lysates. (C) Virus release was quantified by Phosphoimager analysis of the radioactive bands. All data were collected simultaneously using identical exposure times. Signals for intra- and extracellular Gag were determined and added to yield total intra- and extracellular Gag protein for each sample. Virus release was calculated as the fraction of total Gag protein released during the 90-min observation period. Virus release observed for NL4-3 WT in untreated cells in the absence of BST-2 IgG was defined as 100%. All other samples were calculated relative to the reference sample.

The remaining cells were divided into two equal aliquots. One sample was pretreated for 10 min with 5 μg of purified BST-2-specific IgG (Fig. 2B, +BST-2 IgG); the other sample was left untreated (Fig. 2B, −BST-2 IgG). Cells were then metabolically labeled for 90 min as described for Fig. 1B. Cell extracts and virus-containing supernatants were immunoprecipitated with HIV-1-positive patient serum and processed for fluorography as described in Materials and Methods. Results were quantified by Phosphoimager analysis (Fig. 2C). As expected, release of Vpu-defective viruses was inefficient in untreated and control-siRNA-treated samples in the absence of BST-2 IgG (Fig. 2C, gray bars) but was comparable to that of NL4-3 WT virus in BST-2 IgG-treated samples (Fig. 2C, solid bars). Importantly, down-modulation of BST-2 led to increased secretion of Vpu-defective viruses, irrespective of whether the cells had been treated with BST-2 IgG or not. In fact, addition of BST-2 IgG did not further increase the release of WT or Vpu-defective virions in a significant manner. These results demonstrate that the antibody-induced enhancement of virus release from HeLa cells is indeed dependent on the expression of BST-2.

Pretreatment with BST-2 IgG is required to antagonize BST-2-dependent inhibition of virus release.

In the previous experiment, BST-2 antibody was added during the starvation period that precedes metabolic labeling and was present for the duration of the 90-min labeling step. We next tested the kinetic requirements of antibody-mediated neutralization of BST-2 function. For that purpose, HeLa cells transfected with NL4-3/Udel were metabolically labeled for 90 min as described for Fig. 1. BST-2 antiserum was added at different times during the experiment, beginning with the starvation period (−30 min) and up to 5 min prior to the end of the labeling period (85 min), as indicated in Fig. 3A. An untreated sample (no antiserum) and a sample treated with preimmune serum (pre-I) at −30 min were included as controls. Cell lysates and virus-containing supernatants were collected and immunoprecipitated with anti-HIV serum as described for Fig. 1. Virus release was quantified by Phosphoimager analysis as described in Materials and Methods, and the results of three independent experiments are summarized in Fig. 3B. Consistent with the results in Fig. 1, the addition of BST-2 antibody at the beginning of the starvation period (Fig. 3, lane 3) efficiently enhanced the release of Vpu-deficient NL4-3 while treatment with preimmune serum had no effect compared to an untreated sample (Fig. 3, compare lanes 1 and 2). Addition of BST-2 antibody at the beginning of the labeling period (time zero) was also very effective in counteracting BST-2 (Fig. 3, lane 4). In contrast, addition of BST-2 antiserum at later time points resulted in a gradual loss of effect of the antibody-induced enhancement of virus release (Fig. 3, lanes 5 to 8). Since release of particle-associated de novo-synthesized viral proteins can be observed as early as 30 to 60 min following synthesis (53), our results suggest that BST-2 antibodies must be present either prior to or early during virus assembly in order to avoid subsequent tethering of virus particles by BST-2.

Fig. 3.

Pretreatment is required for antagonizing the BST-2 effect by antibody for enhancement of HIV-1 release. (A) HeLa cells were transfected with 5 μg of pNL4-3/Udel. Cells were metabolically labeled for 90 min with [³⁵S]methionine in a total volume of 500 μl. A 5-μl volume of BST-2 antiserum (AS) was added during starvation (−30 min; lane 3), at the start of labeling (0 min; lane 4), or 30 min (lane 5), 60 min (lane 6), 75 min (lane 7), or 85 min (lane 8) after the beginning of labeling. Samples receiving no antiserum (no AS; lane 1) or receiving preimmune serum (pre-I; lane 2) were included as controls. Cell lysates and cell-free supernatants were subjected to immunoprecipitation and analyzed as described for Fig. 1. Fluorographs of a representative result are shown. (B) Virus release was quantified by Phosphoimager analysis using a Fujifilm FLA-7000 system. Virus release was calculated for each sample at each time point by determining the percentage of cell-free CA protein relative to the total intracellular and extracellular Gag protein. The amount of virus released from samples treated with antiserum 30 min prior to labeling (lane 3) was independently determined for the three replicate samples, and its mean was defined as 100%. Results are expressed as the mean ± the standard error of the mean of three independent experiments.

BST-2 antibody treatment does not affect cell surface distribution of BST-2.

Antibody treatment can cause cap formation of cell surface proteins due to cross-linking (55). It is conceivable that antibody-induced cap formation is the cause of the functional inactivation of BST-2 in the previous experiments. On the other hand, cap formation was shown to be caused primarily through the cross-linking of antigen-bound primary antibodies by secondary antibodies in indirect immunofluorescence studies (32), which can occur in live cells but also in fixed cells that are not properly cross-linked by the fixative (18). Since our experiments did not involve secondary antibodies, antibody-induced cross-linking of BST-2 at the cell surface is relatively unlikely. In fact, we had previously demonstrated that antibody-bound BST-2 is rapidly internalized from the cell surface and subsequently recycled back to the cell surface to establish a relatively stable equilibrium (5), suggesting that BST-2-antibody complexes are stable and remain mobile.

To experimentally address the possible effects of antibody treatment on the distribution of BST-2 at the HeLa cell surface, we performed a series of immunofluorescence studies. In the first experiment, we looked at the long-term effects of antibody treatment. HeLa cells were treated with purified BST-2 IgG (5 μg/ml) for 18 h (Fig. 4A, panels b and d). As a control, untreated cells were analyzed in parallel (Fig. 4A, panels a and c). Cells were washed in PBS to remove unbound antibody and then fixed with either 1% paraformaldehyde (Fig. 4A, PFA) or methanol (Fig. 4A, MeOH), followed by staining with dye-conjugated anti-rabbit secondary antibodies. As expected, untreated cells were negative because of the absence of a primary BST-2 antibody (Fig. 4A, panels a and c). In PFA-fixed nonpermeabilized cells, punctate surface fluorescence was observed, indicating the presence of antibody-bound BST-2 at the HeLa cell surface (Fig. 4A, panel b). In contrast, methanol fixation, which causes permeabilization of cells, revealed significant cytoplasmic fluorescence (Fig. 4A, panel d). This result confirms that BST-2 antibody is efficiently internalized and that antibody-BST-2 complexes are stable for at least 18 h.

Fig. 4.

Antibody pretreatment does not affect cell surface distribution of BST-2. (A) HeLa cells were seeded on coverslips and grown in 2 ml of DMEM in 12-well plates. Four hours later, 5 μl of BST-2 IgG (2 mg/ml) was added to half of the wells. Cells were incubated for 18 h at 37°C in the presence of antibody. Cells were then washed once with ice-cold PBS to remove unbound antibody and fixed with either paraformaldehyde (PFA; 1% in PBS, 20 min at room temperature) or methanol (MeOH; 10 min, −20°C). Cells were then stained for 30 min at room temperature with secondary anti-rabbit antibody (diluted 1:100 in PBS containing 1% BSA). Images were acquired on a Zeiss LSM410 confocal microscope using a Plan-Apochromat 63×/1.44 oil immersion objective and processed using the Huygens Professional deconvolution software package (v4.0). Images a and b are paraformaldehyde-fixed untreated and BST-2 IgG-treated samples, respectively. Images c and d are methanol-fixed samples. (B) HeLa cells were treated as for panel A, except that all samples were fixed with paraformaldehyde and the time of pretreatment with BST-2 IgG was modified as indicated above the images. Samples in images a to d were stained directly with secondary anti-rabbit antibody (diluted 1:100 in PBS containing 1% BSA) for 30 min at room temperature. Samples in images e to h were first incubated for 30 min at room temperature with primary BST-2-specific antibody (1:100 diluted in PBS containing 1% BSA), washed once with PBS, and then incubated with secondary anti-rabbit antibody. Image acquisition and processing were done as described for panel A.

We next analyzed the short-term effects of antibody treatment with regard to the surface distribution of BST-2. HeLa cells were treated with purified BST-2 IgG (5 μg/ml) for 10, 30, or 60 min. An untreated sample was included as a control. Unbound antibody was removed by washing in PBS, and cells were fixed with 1% paraformaldehyde. One set of samples was stained directly with dye-conjugated anti-rabbit secondary antibody (Fig. 4B, panels a to d). The second set was first incubated with BST-2-specific primary antibody and then stained with anti-rabbit secondary antibody (Fig. 4B, panels e to h). As expected, direct staining of untreated cells with secondary antibody was negative due to the absence of primary antibodies (Fig. 4B, panel a). In contrast, pretreatment with BST-2 IgG followed by staining with secondary antibody resulted in a punctate surface fluorescence similar to that in Fig. 4A (panel b). Of note, pretreatment of HeLa cells with BST-2 IgG for different times had no obvious effect on the staining pattern (compare Fig. 4B, panels b to d), suggesting that BST-2 IgG treatment does not induce cap formation in HeLa cells. In addition, indirect fluorescence analysis of untreated HeLa cells (Fig. 4B, panel e) produced a staining pattern very similar to that observed after antibody pretreatment (Fig. 4B, panels f to h). These results indicate that antibody-induced enhancement of virus release does not involve a gross redistribution of BST-2 at the cell surface.

BST-2 antibody does not release already tethered virions.

To further investigate the requirements for antibody addition, we tested the effect of BST-2 antibody on already tethered virions. For that purpose, we first determined the kinetics of virus release from transfected HeLa cells by pulse-chase analysis as performed in earlier studies (53). HeLa cells were transfected with NL4-3 WT or NL4-3/Udel virus, metabolically labeled with [³⁵S]methionine for 30 min, and chased for up to 3 h, followed by immunoprecipitation of cell lysates and virus-containing supernatants using HIV-1-positive patient serum (Fig. 5A). Under the conditions of this experiment, very little, if any, Vpu-deficient HIV-1 was released during the 3-h observation period (Fig. 5A, right panel). In contrast, a significant amount of WT virus was released within 3 h (Fig. 5A, left panel). Thus, the conditions of this experiment, i.e., labeling for 30 min followed by a 3-h chase, provide sufficient time for Vpu-deficient particles to accumulate at the cell surface.

Fig. 5.

BST-2 antibody treatment does not enhance release of already tethered virions. (A) HeLa cells were transfected with 5 μg of pNL4-3 WT (left side) or pNL4-3/Udel (right side). After 24 h, cells were pulse-labeled for 30 min with [³⁵S]methionine and chased for up to 3 h. Cell lysates and cell-free supernatants were subjected to immunoprecipitation using HIV-positive patient serum and analyzed as described for Fig. 1A. (B) HeLa cells were transfected with 5 μg of pNL4-3 WT or pNL4-3/Udel DNA. After 24 h, cells were pulse-labeled for 30 min with [³⁵S]methionine and then incubated for an additional 3 h in complete RPMI in the absence of isotope. Virus-containing supernatants were then harvested and immunoprecipitated with HIV-positive patient serum to assess spontaneous virus release (middle). The cell pellet was either vortexed for 30 s (bottom, lanes 1 and 2) or treated with subtilisin (100 μg/ml) for 2 min at room temperature and then suspended in the same volume of culture medium (bottom, lanes 3 and 4) and subjected to immunoprecipitation with HIV-positive patient serum. Immunoprecipitates of the cell lysates are shown at the top. The arrow shows the order in which samples were collected. (C) HeLa cells were transfected with 5 μg of pNL4-3 WT (W) or pNL4-3/Udel (U) DNA. Cells were pulse-labeled after 24 h for 30 min with [³⁵S]methionine and incubated for an additional 3 h in the absence of isotope. Cells were then treated for 90 min with 5 μl of either preimmune serum (pre-I) or BST-2 antiserum (α-BST-2). A control sample that did not receive antiserum was included as well (no AS). Cell lysates and cell-free supernatants were subjected to immunoprecipitation and analyzed as described for panel A. A representative result is shown. Virus release was quantified as described for Fig. 3B. Release of WT virus from untreated cells (Ctrl) was defined as 100%. Shaded bars represent WT NL4-3; white bars represent NL4-3/Udel. Results are expressed as the mean ± the standard error of the mean of three independent experiments.

To ascertain that Vpu-deficient virions were tethered at the cell surface following 3 h of chase and were not already internalized for degradation, we stripped the virus from the cell surface by either physical force (vortexing) or protease treatment (subtilisin). Both methods were previously employed to release Vpu-deficient virions from virus-producing cells (26, 35, 36). The spontaneous release of WT and Vpu-deficient virions in the absence of vortexing or protease treatment is shown in Fig. 5B, middle panel. We found that Udel virus was indeed released by both vortexing (Fig. 5B, bottom panel, lane 2) and subtilisin treatment (Fig. 5B, bottom panel, lane 4), suggesting that a significant fraction of tethered virions were still at the cell surface. As expected, some WT virus could also be released by vortexing (Fig. 5B, bottom panel, lane 1) and subtilisin treatment (Fig. 5B, bottom panel, lane 3).

Next we tested the effect of BST-2 antibodies on the release of already tethered virions. HeLa cells were transfected with NL4-3 WT or NL4-3/Udel virus, metabolically labeled for 30 min, and chased for 3 h to allow virus accumulation at the cell surface. At that time, BST-2 antiserum was added to the cultures and samples were incubated for an additional 90 min at 37°C to allow the antibody to take effect. Cell lysates and virus-containing supernatants were then immunoprecipitated with HIV-positive patient serum (Fig. 5C, α-BST). Parallel samples that were not treated with antiserum (Fig. 5C, no antiserum) or that had been treated with preimmune serum (Fig. 5C, pre-I) were included as controls. We found that under these conditions, BST-2 antibody treatment did not increase the release of HIV-1 virions above the levels observed without antibody treatment or following treatment with preimmune serum. These results demonstrate that BST-2 antibody is unable to induce the release of already tethered virions from BST-2-expressing HeLa cells.

Enhancement of virus release by BST-2 antibody is cell type independent.

All of the experiments described so far were performed in HeLa cells, which express comparatively high levels of endogenous BST-2 (35). To rule out the possibility that the antibody-induced enhancement of virus release is a cell type-specific phenomenon, we analyzed the effects of BST-2 antibody on the release of virions from 293T cells. 293T cells generally express undetectable levels of endogenous BST-2 (35). This is exemplified in Fig. 6A, where different volumes of HeLa cell lysate (lanes 1 to 5) were compared to different volumes of 293T cell extracts using a BST-2-specific antibody (lanes 6 and 8). Expression of tubulin served as a reference (Fig. 6A, lower panel). As reported previously (35, 37), BST-2 expression in 293T cells could be induced by treatment with IFN-α (Fig. 6A, lanes 7 and 9). To analyze the effects of BST-2 antibody on virus release from IFN-α-treated 293T cells, cells were transfected with NL4-3 WT or NL4-3/Udel provirus. IFN-α (10 ng/ml) was added to the cultures 4 h after transfection. Untreated 293T cells were included as a control. Virus release was assessed 24 h later by pulse-chase analysis in the presence or absence of BST-2 antibody. The results of two independent experiments are shown in Fig. 6B. As expected, virus release from unstimulated 293T cells was Vpu independent (Fig. 6B, lanes 1 and 2) and addition of BST-2 antibody did not further increase virus release (Fig. 6B, lanes 5 and 6). In contrast, IFN-α treatment rendered virus release Vpu dependent and reduced the release of Vpu-deficient virus 2- to 3-fold relative to that of WT virus (Fig. 6B, lanes 3 and 4; P = 0.032). Release of treated WT virus was also reduced compared to that of untreated WT virus, presumably due to the induction of BST-2 (Fig. 6B, lanes 1 and 3). Of note, the IFN effect was reversed by BST-2 antibody treatment (Fig. 6A, lanes 7 and 8), demonstrating that the antibody-enhanced release of virus is not restricted to HeLa cells.

Fig. 6.

Enhancement of virus release by BST-2 antibodies is not cell type specific and is not restricted to HIV-1. (A) Comparative analysis of BST-2 expression in HeLa cells and untreated or IFN-stimulated 293T cells. HeLa and 293T cells were grown to approximately 80% confluence. For IFN induction of BST-2, 293T cells were treated overnight with IFN-α (10 ng/ml). Whole-cell lysates were prepared as described in Materials and Methods. Different amounts of lysate, as indicated at the top, were separated by SDS-PAGE and subjected to immunoblot analysis using a BST-2-specific antibody (top panel). The same blot was subsequently reprobed with an antibody to tubulin as an internal reference. Based on the tubulin blot, we estimate that 5 μl of 293T lysate corresponds to approximately 20 μl of HeLa extract. (B) 293T cells were transfected with 5 μg of pNL4-3 WT or pNL4-3/Udel (Udel) DNA. After 4 h, IFN-α (10 ng/ml) was added to the samples as indicated (columns 3, 4, 7, and 8). Cells were metabolically labeled 24 h later for 90 min with [³⁵S]methionine in the presence of 5 μl preimmune serum (columns 1 to 4) or BST-2 antiserum (columns 5 to 8). Cell lysates and cell-free supernatants were subjected to immunoprecipitation as shown for Fig. 3A, and virus release was quantified as described for Fig. 5C. Shaded bars represent NL4-3 WT; open bars represent NL4-3/Udel. Results of two independent experiments are expressed as the mean ± the standard error of the mean. (C) PBMCs from two healthy donors were infected with equal RT units of NL4-3 WT and NL4-3/Udel virus stocks produced in 293T cells. Preimmune serum or BST-2 antiserum (1 μl/ml each) was added 4 h after infection. At each time point, all of the culture supernatants were collected and replaced with fresh medium supplemented with preimmune serum or BST-2 antiserum, respectively (1 μl/ml). Virus replication was monitored by measuring the virus-associated RT activity in the culture supernatants. Results are plotted as a function of time. Error bars represent the mean ± the standard error of the mean of two independent infections. (D) HeLa cells were transfected with 5 μg of pROD1014TA (1014TA) or pROD1014 (1014). Cells were metabolically labeled 24 h later for 90 min with [³⁵S]methionine in the presence of 5 μl preimmune serum (lanes 1 and 2) or BST-2 antiserum (lanes 3 and 4). Cell lysates and cell-free supernatants were subjected to immunoprecipitation using HIV-2-positive patient serum. (E) Quantitation of the results in panel D was done as described for Fig. 5C. Shaded bars represent ROD1014TA; white bars represent ROD1014. Error bars represent the mean ± the standard error of the mean of two independent experiments.

The effect of BST-2 antibody treatment on the replication of NL4-3 WT or NL4-3/Udel virus in primary cells was tested in human PBMCs from two independent donors (Fig. 6C). Virus stocks used for the infection of PBMCs were prepared in 293T cells. BST-2 antibody (α-BST) or preimmune serum (pre-I) was added at 4 h postinfection. Culture supernatants were collected and replaced with fresh medium on the days indicated in Fig. 6C. Fresh preimmune serum or BST-2 antibody (1 μl/ml) was added with each medium change. Virus replication was determined by measuring the virus-associated RT activity in the culture supernatants (62). In the presence of preimmune serum, the amounts of cell-free virus produced from the Udel-infected culture were lower than those in the WT-infected culture (Fig. 6C, compare open and closed circles). Overall, however, the Vpu dependence of virus replication in PBMCs was very modest, consistent with our previous observations on the Vpu dependence of virus release from PBMCs (51) and attributable to the fact that levels of BST-2 in PBMCs are generally low (35). Nevertheless, addition of BST-2 antibody increased the release of Vpu-deficient virus (Fig. 6C, open triangles), as well as WT virus (Fig. 6C, closed triangles). We conclude that the positive effect of BST-2 antibody on virus release is cell type independent and can be observed in primary cells, as well as in BST-2-expressing immortalized cell lines.

The effect of BST-2 antibody on virus release is not HIV-1 specific.

To test the virus specificity of antibody-enhanced virus release, we chose HIV-2 as a model system. HIV-2 does not contain a vpu gene, but the Env glycoprotein of certain HIV-2 isolates has Vpu-like properties and can functionally substitute for Vpu in virus release assays (1, 7, 44). Of note, the virus release activity of HIV-2 Env can be regulated by a single amino acid change (A598T) that was identified in a natural HIV-2 isolate (6). To analyze the effect of BST-2 antibody on the release of HIV-2 from HeLa cells, we used two previously described HIV-2 constructs, pROD1014 and pROD1014.598TA (6). The pROD1O14 chimera expresses the ROD14 Env glycoprotein in the backbone of the ROD10 isolate and exhibits a Vpu⁻ phenotype. The pROD1014.598TA chimera is identical to pROD1014, except for a single amino acid change (threonine to alanine) at position 598 of the Env protein, and exhibits a Vpu⁺ phenotype (6). HeLa cells were transfected with pROD1014 and pROD1014.598TA, and virus release in the presence of either preimmune serum (Fig. 6D, pre-I) or BST-2-specific antibody (Fig. 6D, α-BST) was determined by immunoprecipitation of metabolically labeled proteins using HIV-2-specific antibody. Virus particle release was quantified as described for Fig. 3. The quantitative analysis of two independent experiments is summarized in Fig. 6E. In the presence of preimmune serum, release of ROD1014 virus was low compared to that of ROD1014.598TA virus (Fig. 6D/E, lanes 1 and 2), consistent with the Vpu⁻ phenotype of the ROD1014 chimera. However, addition of BST-2 antibody increased the release of ROD1014 virions to levels comparable to those of the ROD1014.598TA virus (Fig. 6D and E, compare lane 4 with lane 1). BST-2 antibody also improved the release of the ROD1014.598TA virus, despite its inherent Vpu⁺ phenotype (Fig. 6E, compare column 3 with column 1). These results demonstrate that enhancement of virus release by BST-2 antibody is not virus specific.

BST-2 antibody treatment neither inhibits dimerization nor causes internalization or degradation of BST-2.

Our next goal was to investigate how BST-2 antibody treatment antagonizes BST-2 function. Several possibilities were considered, including more rapid degradation of BST-2 in the presence of antibody or interference of the antibody with cysteine-linked dimerization, which is critical for BST-2 function (4, 42). Both possibilities were analyzed simultaneously in a single experiment. HeLa cells were incubated for 24 h at 37°C in the presence of preimmune serum or BST-2 antibody. Cells were then lysed by heating in sample buffer lacking ME to preserve disulfide bonds (Fig. 7A, −ME) or in the presence of ME to reduce disulfide bonds (Fig. 7A, +ME). Samples were then analyzed by immunoblotting using BST-2-specific antibody (Fig. 7A, top panel). The same membrane was subsequently reprobed with antibodies to tubulin to control for equal sample loading (Fig. 7A, lower panel). The results of this experiment suggest that BST-2 antibody treatment neither accelerated BST-2 degradation nor inhibited BST-2 dimerization. In fact, BST-2 protein appeared to be stabilized by the antibody, as an increased BST-2 signal is evident, especially in the unreduced sample (Fig. 7A, compare BST-2 dimer in lanes 1 and 2) but also in the reduced sample (Fig. 7A, compare BST-2 monomer in lanes 3 and 4). Also, most, if not all, of the BST-2 appears to form covalently linked dimers in the absence of ME (Fig. 7A, lanes 1 and 2), irrespective of the antibody treatment. Of note, IgG bands were visible in BST-2 antibody-treated samples but not in samples treated with preimmune serum, indicating that binding of the antibody to the cell and/or its uptake into the cell requires the specific interaction with the BST-2 ligand.

Fig. 7.

BST-2 antibody treatment does not inhibit dimerization or cause internalization of BST-2 protein. HeLa cells were treated for 24 h with preimmune serum (lanes 1 and 3) or BST-2 antiserum (lanes 2 and 4) (1 μl/ml each). Cells were then harvested and split into three fractions. (A) One fraction was used to prepare whole-cell extracts under nonreducing conditions (−ME); the second fraction of cells was used to prepare whole-cell extracts under reducing conditions (+ME) as described in Materials and Methods. Cell lysates were subjected to immunoblotting using antibodies to BST-2 (top). The blot was reprobed with an antibody to tubulin as an internal control for sample loading (bottom). Proteins are identified on the right. Ig marks the position of immunoglobulin heavy chain. (B) The remaining cells were used for cell surface staining of BST-2 at 4°C using the same BST-2-specific primary antibody as used in the pretreatment. Prebound and newly bound BST-2 antibodies were then decorated at 4°C with APC-conjugated anti-rabbit secondary antibodies, followed by FACS analysis. Untreated cells stained with preimmune serum served as a negative control (Ctrl). Untreated cells stained with BST-2 antibody served as a positive control (untreated; dotted line). Cells pretreated with preimmune serum (pre-I) or with BST-2 antibody (α-BST) were stained with BST-2 antibody prior to FACS analysis.

Another possible effect of BST-2 antibody could be down-modulation of BST-2 from the cell surface. To test this, we incubated HeLa cells for 24 h at 37°C in the presence of either preimmune serum or BST-2 antibodies. Surface expression of BST-2 was then determined by FACS analysis using the same BST-2 antibody as used for pretreatment. This allowed us to quantitatively identify both preexisting BST-2 antibody complexes and newly synthesized BST-2 molecules with a single anti-rabbit secondary antibody and thus identify the entire steady-state population of cell surface BST-2. We found that treatment with BST-2 antibody did not reduce the cell surface expression of BST-2 compared to that on cells treated with preimmune serum; in fact, pretreatment of cells with BST-2-specific antibody increased the cell surface expression of BST-2 (Fig. 7B, compare pre-I and α-BST profiles). Similar results were observed in cells producing NL4-3 WT or NL4-3/Udel virus, except that the set point was lower in the case of NL4-3 WT virus due to the activity of Vpu (data not shown). Thus, BST-2 antibody treatment did not adversely affect two functionally critical parameters of BST-2: surface expression and formation of cysteine-linked dimers.

BST-2 antibody treatment reduces packaging of BST-2 into virions.

Previous studies suggested that BST-2 is packaged into cell-free virions (17, 42). The observation that antibody treatment increased, rather than decreased, the cell surface expression of BST-2 raised the question of whether antibody treatment affected the encapsidation of BST-2. To investigate this possibility, we analyzed the packaging of BST-2 protein into WT or Vpu-deficient HIV-1 virions produced in the presence of preimmune serum or BST-2 antiserum. HeLa cells were transfected with pNL4-3 (Fig. 8A, WT) or pNL4-3/Udel (Fig. 8A, Udel). Mock-transfected samples were included as a negative control (Fig. 8A, mock). Preimmune serum (pre-I) or BST-2 antibody (α-BST-2) was added 4 h after transfection. Cells and virus-containing supernatants were collected 24 h later. Supernatants were concentrated by pelleting through a 20% sucrose cushion as described in Materials and Methods. Whole-cell extracts (cell) and concentrated supernatants (sup) were subjected to immunoblotting to identify BST-2. Two different exposures are shown (Fig. 8A, top two panels). The same blot was sequentially reprobed in the following order: HIV-positive patient serum (Fig. 8A, CA), antibodies to Vpu (Fig. 8A, Vpu), and antibodies to tubulin (Fig. 8A, tub). Consistent with our previous observation (35) BST-2 was undetectable in supernatant fractions using normal exposure times (Fig. 8A, top panel). However, upon overexposure, small amounts of BST-2 were detectable in concentrated culture supernatants, even in mock-transfected samples (Fig. 8A, second panel, lanes 7 to 12). Quantitation of BST-2 from three independent experiments is shown in Fig. 8B. Our results indicate that the relative amounts of BST-2 identified in supernatants containing WT virus were nearly identical to those secreted nonspecifically from mock-transfected cultures, irrespective of whether the samples had been treated with preimmune serum or with BST-2 antiserum (Fig. 8B, compare open and shaded bars). In contrast, supernatants from cultures treated with preimmune serum and producing Vpu-deficient virus contained about 7 times higher levels of BST-2 (Fig. 8B, pre-I; open bar). Of note, pretreatment with BST-2 antibody did not further increase the packaging of BST-2; quite to the contrary, levels of extracellular BST-2 were lower in all samples pretreated with BST-2 antibody, even in the mock-transfected culture and the sample producing vpu-defective virus (Fig. 8B, α-BST-2). These results suggest that treatment of cells with BST-2 antibody, despite increased surface expression, as shown in the previous figure, reduces nonspecific shedding and minimizes packaging of BST-2 into virus particles. Thus, treatment with BST-2 antibodies may cause a change in the surface distribution of BST-2 that diverts the protein away from the site of virus assembly. Alternatively, antibody treatment may induce cross-linking of surface BST-2 and render the protein packaging incompetent.

Fig. 8.

BST-2 antibody treatment reduces encapsidation of BST-2 into virions. (A) HeLa cells were transfected with 5 μg of pNL4-3 WT (lanes 2, 5, 8, and 11) or pNL4-3/Udel (lanes 3, 6, 9, and 12) DNA. Mock-transfected samples were analyzed in parallel (lanes 1, 4, 7, and 10). Preimmune serum (pre-I) or BST-2 antiserum (α-BST-2) was added 4 h after transfection (each at 10 μl/ml). Cells were harvested 24 h later, and virus-containing supernatants (sup) were concentrated by pelleting through 20% sucrose. Cell lysates and concentrated supernatants were subjected to immunoblotting using BST-2-specific antibody. Short and long exposures of the same blot are shown (top two images). The same membrane was then sequentially reprobed with HIV-positive patient serum to detect viral capsid protein (CA), as well as antibodies to Vpu or tubulin (tub). Proteins are identified on the right. A representative experiment is shown. (B) BST-2 in the culture supernatant was quantified by densitometric scanning of the blots. BST-2 present in the supernatant of samples treated with preimmune serum and expressing WT virus (panel A, lane 8) was defined as 1. Levels of BST-2 in the other supernatants were calculated relative to that sample. BST-2 in virus-containing supernatants was normalized for CA levels (an exposure shorter than that shown was used for quantification) to take into account the less efficient release of Udel virions. Error bars represent the mean ± the standard error of the mean of three independent experiments.

BST-2 antibody treatment alters localization of BST2 in raft-associated membranes.

Virus assembly and release involve lipid raft structures at the cell surface (40). Interestingly, BST-2 has been reported to associate with raft structures, presumably mediated by its C-terminal GPI anchor (28). Our own analyses confirmed the near quantitative partitioning of endogenous BST-2 in untransfected HeLa cells with raft fractions in a flotation assay as described in Materials and Methods (Fig. 9A). We then tested whether BST-2 antibody treatment could interfere with the raft association of BST-2 and/or could lead to separation of BST-2 and viral Gag protein in virus-producing cells. For that purpose, HeLa cells were transfected with pNL4-3/Udel. Preimmune serum or BST-2 antiserum was added 4 h after transfection. Cells were also treated with ritonavir (1 μM) to prevent Gag maturation and thus enrich for the precursor protein, which—unlike mature capsid protein—is membrane associated due to its N-terminal myristic acid moiety. Ritonavir treatment did not interfere with the enhancing effect of BST-2 antibody on virus release (data not shown). Cells were processed 24 h after transfection for flotation analysis, followed by immunoblotting as described in Materials and Methods. Membranes were sequentially probed with antibodies to BST-2, Cav-1, TfR, and HIV-1 Gag (Fig. 9B). Quantitation of the BST-2, Cav-1, and Pr55Gag signals from three independent analyses is shown in Fig. 9C. As in untransfected HeLa cells, BST-2 was enriched in the raft fractions in virus-producing cells (Fig. 9B, fractions 2 and 3). Interestingly, however, while the bulk of BST-2 was found in fraction 2 in untransfected cells and in virus-producing samples treated with preimmune serum (Fig. 9A to C, lanes 2), BST-2 was equally distributed between fractions 2 and 3 in samples treated with BST-2-specific antibody (Fig. 9B and C, lanes 10 to 11). The relative distribution of neither Cav-1 nor Pr55Gag in the flotation gradient was affected by BST-2 antibody treatment (Fig. 9B and C). Thus, the shift in distribution was specific for BST-2. The same shift was also observed in the absence of ritonavir treatment, indicating that it was not caused by the protease inhibitor (data not shown). Pr55Gag was identified in both raft and nonraft fractions of the gradient. Surprisingly, the majority of the raft-associated Gag precursor was identified in fraction 3. Only small amounts of Gag protein were identified in fraction 2. Thus, paradoxically, treatment of cells with BST-2 antibody appeared to increase, rather than decrease, the cofractionation of BST-2 and Gag in the flotation gradients.

Fig. 9.

BST-2 antibody treatment alters the relative distribution of BST-2 in raft fractions. (A) Untransfected HeLa cells were processed for flotation centrifugation as described in Materials and Methods. Eight fractions were harvested from the top to the bottom of the gradient. Fractions were separated by SDS-PAGE and subjected to immunoblotting using BST-2-specific antiserum (top). The same membrane was reprobed with antibodies to cellular Cav-1, which served as a raft marker, and TfR, representing a non-raft-associated control. Proteins are identified on the right. (B) HeLa cells were transfected with 5 μg of pNL4-3/Udel DNA. Preimmune serum (pre-I) or BST-2 antiserum (α-BST-2) was added 4 h after transfection (1 μl/ml). In addition, ritonavir (1 μM) was added to enrich for Gag precursor proteins. Cells were harvested 20 h later and processed for flotation analysis and subsequent immunoblotting as described for panel A. Viral Gag protein (Pr55Gag) was identified with HIV-positive patient serum. Results of a representative experiment are shown. (C) BST-2, Cav-1, and Pr55Gag signals were quantified by image analysis, and the relative distribution of each protein across the 8 fractions of the gradient was plotted. For BST-2 and Cav-1, fractions 4 to 8 were excluded from the quantitation because there was no protein detectable in these fractions. Results are shown as the mean ± the standard error of the mean of three independent experiments.

DISCUSSION

The BST-2-mediated tethering of nascent viral particles to the infected cell surface is thought to be an intrinsic host response aimed at preventing or delaying the spread of a viral infection through the body. However, in the case of HIV, the success of this strategy is unclear since HIV can establish spreading infection not only through secretion of cell-free virions but also through the infection of adjacent target cells via direct membrane fusion in a process known as cell-to-cell transmission. There is no doubt that cell-free transmission of HIV is sensitive to tetherin. However, the effect of BST-2 on cell-to-cell transmission of HIV is still under debate. Two recent studies suggested that BST-2 expression inhibits the cell-to-cell transmission of HIV-1 (9, 27), while a third study concluded that BST-2 expression in fact promotes the cell-to-cell transmission of HIV (23). The latter view is supported by traditional virological studies that consistently demonstrated that Vpu-deficient HIV-1 does spread in tissue culture—including primary T cells and macrophages—with kinetics similar to those of the WT virus; only the yield of detectable cell-free virus is reduced (13, 26, 52, 54, 56). We therefore prefer to view BST-2 less as a host factor inhibiting HIV replication than as a host factor modulating the mode of virus transmission.

In fact, HIV-1 may employ BST-2 to switch between the cell-free and cell-cell transmission modes. This is suggested by the fact that some HIV-1 isolates, including ADA clone AD8 (57), LAV Mal (GenBank accession no. A07116), and Yu2 isolates (GenBank accession no. HIVYU2X), carry a point mutation in the vpu initiation codon that disables the expression of Vpu. Vpu expression in these isolates can easily be restored by a single nucleotide change. Similarly, the HIV-2 Env protein has the ability to turn its Vpu-like activity on or off through a single point mutation in the TM subunit without losing its function as a viral envelope protein (1, 6). This allows the virus to adapt to changes in the host milieu by switching between cell-free and cell-to-cell modes of transmission. Cell-free virus transmission is likely to favor rapid dissemination of the virus to susceptible organs throughout the body while cell-to-cell transmission of HIV-1 favors localized virus spread within an organ.

Our data demonstrate that antibody treatment effectively neutralizes BST-2 and enhances virus secretion only if the antibody is added within 30 min of viral protein synthesis. This effect was seen not only with our own BST-2 rabbit polyclonal antibody but with a commercial mouse monoclonal antibody (BioLegend, San Diego, CA; clone RS 38E) as well (data not shown). Importantly, antibody treatment did not release already tethered virions. We prepared Fab and F(ab′)₂ fragments of our BST-2 antibody to test if antibody-induced cross-linking of BST-2 at the cell surface may lead to functional inactivation. Fab fragments are monovalent, while F(ab′)₂ fragments retain their bivalent character. Thus, F(ab′)2, but not Fab, fragments retain the potential to cross-link antigen at the cell surface. While our Fab and F(ab′)2 preparations were able to identify surface BST-2 in indirect immunofluorescence analyses, neither one worked in immunoblot analyses and neither one enhanced virus release (data not shown). Nevertheless, antibody treatment had no overt effects on the surface distribution of BST-2. Thus, antibody-induced cross-linking of surface BST-2 as the mechanism of BST-2 inactivation is unlikely, although it cannot be formally ruled out. Taking into account the facts that BST-2:antibody complexes rapidly cycle between cell surface and intracellular compartments (5, 11, 34) and that antagonism of virus release seems to be exerted primarily by de novo-synthesized BST-2 (5), it seems plausible that antibody-induced interference with BST-2 function occurs intracellularly. In fact, BST-2-antibody complexes appear to be stable and were readily detected intracellularly at least 18 h after antibody addition (Fig. 4A). Several scenarios can be envisioned. For instance, antibody-loaded BST-2 molecules could be redirected upon recycling to the surface away from viral budding sites. Alternatively, antibody binding could induce structural changes in BST-2 that interfere with structural properties critical to BST-2 function. However, it is also possible that BST-2 antibodies act at the cell surface to inhibit the association of BST-2 with budding virion membranes. In such a scenario, the observed inability of antibody treatment to release already-tethered virions could be attributable to steric limitations (e.g., an inability of the antibody to bind BST-2 that is engaged with tethered virions).

It remains to be investigated whether HIV-1 Vpu, HIV-2 Env, SIV Nef, and BST-2 antibody treatment interfere with BST-2 activity through a common mechanism. All three viral proteins and antibody treatment appear to interfere with BST-2 function through physical interaction with BST-2. There are, of course, differences in the way these proteins interact; in particular, the domains in BST-2 involved in the interaction differ for each of these proteins. Also, there are differences in the reported effects of these factors on the cell surface expression of BST-2. While HIV-1 Vpu and HIV-2 Env can efficiently down-modulate BST-2 from the cell surface (10, 11, 17, 19, 29, 34, 35, 46, 48, 58), SIV Nef appears to have only a very modest or no effect on BST-2 cell surface expression (22, 65). Interestingly, BST-2 antibody treatment also did not reduce the surface expression of BST-2 but instead increased surface levels, presumably by stabilizing the total cellular pool of BST-2. In that regard, it should be pointed out that while Vpu has the inherent ability to down-modulate cell surface BST-2, it can antagonize BST-2 in acutely infected T cells without BST-2 down-modulation (35). Thus, while viral particles are tethered to the virus-producing cell through cell surface BST-2, interference with BST-2 function does not necessitate cell surface down-modulation, implying that interference may not happen at the cell surface but prior to tetherin reaching the plasma membrane.

It remains to be determined exactly how BST-2 antibody treatment affects BST-2 function. We demonstrated that antibody treatment did not accelerate BST-2 degradation, reduce surface expression, or affect the formation of cysteine-linked BST-2 dimers. Instead, we found that antibody treatment affected the relative distribution of BST-2 between raft fractions 2 and 3 in the gradient. This effect was highly reproducible and specific to BST-2. Paradoxically, BST-2 antibody treatment appeared to increase the copartitioning of BST-2 and Gag protein in the flotation assay, although this effect could be coincidental. However, this does not indicate increased interaction between these proteins. It simply reflects a change in the density of BST-2-containing membrane complexes. Clearly, more detailed biochemical analyses are needed to fully understand the implications of this phenomenon.

Our study focused on HIV and demonstrates that BST-2 antibody treatment enhances the production of HIV-1, as well as HIV-2. However, BST-2 is known to also control the release of other enveloped viruses, including members of the herpesvirus, filovirus, and arenavirus families. It is likely that antibody treatment will increase the secretion of these viruses as well and potentially aggravate virus-induced disease. This is of concern since BST-2 antibody therapy has been proposed for the treatment of patients with multiple myeloma and related plasma cell-derived immunoproliferative disorders (25, 41, 49), as well as for the treatment of lung cancer and other solid tumors with high BST-2 antigen expression levels (8, 24, 60, 61). In fact, humanized BST-2 antibodies have been developed and are in clinical trial for the treatment of patients with multiple myeloma. In those studies, the antibody therapy seems to be well tolerated, with low cytotoxicity (43). However, considering the virus-inducing properties described in the present study, potential side effects, such as the aggravation of existing viral infections, should be taken into consideration, especially in immunosuppressed patients.