Vpu and Tsg101 Regulate Intracellular Targeting of
http://www.100md.com
病菌学杂志 2006年第8期
Department of Virology, Haartman Institute, FIN-00014 University of Helsinki, Finland
The Physiological Laboratory, University of Liverpool, Liverpool L69 3BX, United Kingdom
Department of Biosciences at Novum, Karolinska Institutet, S-141 57 Huddinge, Sweden
Department of Virology, Swedish Institute for Infectious Disease Control and MTC, Karolinska Institutet, 171 82 Stockholm and Department of Molecular Virology, Linkoping University, 581 83 Linkoping, Sweden
ABSTRACT
Assembly of human immunodeficiency virus type 1 (HIV-1) is directed by the viral core protein Pr55gag. Depending on the cell type, Pr55gag accumulates either at the plasma membrane or on late endosomes/multivesicular bodies. Intracellular localization of Pr55gag determines the site of virus assembly, but molecular mechanisms that define cell surface or endosomal targeting of Pr55gag are poorly characterized. We have analyzed targeting of newly synthesized Pr55gag in HeLa H1 cells by pulse-chase studies and subcellular fractionations. Our results indicated that Pr55gag was inserted into the plasma membrane and, when coexpressed with the viral accessory protein Vpu, Pr55gag remained at the plasma membrane and virions assembled at this site. In contrast, Pr55gag expressed in the absence of Vpu was initially inserted into the plasma membrane, but subsequently endocytosed, and virus assembly was partially shifted to internal membranes. This endocytosis of Pr55gag required the host protein Tsg101. These results identified a previously unknown role for Vpu and Tsg101 as regulators for the endocytic uptake of Pr55gag and suggested that the site of HIV-1 assembly is determined by factors that regulate the endocytosis of Pr55gag.
INTRODUCTION
Assembly and budding of human immunodeficiency virus type 1 (HIV-1) is driven by the viral core protein precursor Gag (Pr55gag), which assembles into enveloped virus-like particles (VLPs) when expressed in the absence of other viral proteins (9, 21). Pr55gag is a myristoylated, peripheral membrane protein (3, 9, 14, 40, 47). Concomitantly with assembly and budding, Pr55gag is processed into matrix, capsid (CA), nucleocapsid, and p6 proteins and two smaller peptides SP1 and SP2 by a viral protease present in the minor core component, the Gag-Pol precursor. An intriguing feature of the Pr55gag-mediated assembly of HIV-1 is that different cellular membranes serve as a platform for virus assembly in different cell types. Electron microscopy (EM) and fluorescence microscopy studies have indicated that Pr55gag predominantly accumulates at the plasma membrane, and progeny virions are formed at this site in T cells and model cell lines such as HeLa, 293T, and Cos-1 (see, for example, references 8, 16, and 27), although some Pr55gag has also been identified on late endosomes in these cells (6, 10, 26, 39, 46). In contrast, in macrophages Pr55gag almost exclusively concentrates on late endosomes/multivesicular bodies (MVBs) and progeny virions bud into the lumen of MVBs (25, 28, 31, 33). Egress of virus particles from the cell surface, and probably also from the endosomal assembly sites, is critically dependent on interaction of the Pr55gag p6 domain with the host protein Tsg101 (tumor susceptibility gene 101) (7, 23, 45, 46), which normally functions in formation of MVB vesicles (24). Since ablation of Pr55gag-Tsg101 interactions arrests virus budding at a stage after membrane distortion (7, 46), the primary function of Tsg101 in HIV-1 assembly and budding is thought to be the recruitment of downstream factors essential for the completion of budding and release of virions from the donor membrane (24, 46).
Molecular mechanisms that control cell surface versus endosomal accumulation of Pr55gag, and consequently the choice of virus budding site, are poorly understood. One aspect that remains to be clarified is the route by which Pr55gag reaches its cell surface or endosomal steady-state localization. Like other myristoylated proteins, Pr55gag is expected to be synthesized on cytosolic ribosomes and inserted into membranes posttranslationally (20). It is assumed that the classical endoplasmic reticulum-Golgi secretory route to the plasma membrane and endosomes is not involved in intracellular targeting of Pr55gag, since inhibitors of this transport pathway, such as brefeldin A or monensin, have no effect on the assembly of Pr55gag into extracellular particles (29, 30). Since Pr55gag has been observed on endosomal membranes in a number of different cell types (6, 10, 26, 39, 46), it is possible that the protein is first inserted into endosomal membranes and either retained in these compartments (macrophages) or transported further to the plasma membrane (e.g., HeLa, 293T, or T cells). Alternatively, newly synthesized Pr55gag could be first targeted to the plasma membrane, and endosomal localization could be achieved via endocytosis of the cell surface-associated protein. A third possibility is that a newly synthesized Pr55gag is selectively inserted either into the plasma membrane or endosomal membranes depending on the host cell type.
In the present study, we analyzed targeting of newly synthesized Pr55gag in HeLa H1 cells by pulse-chase studies and a subcellular fractionation assay that efficiently separated plasma membrane from internal membranes. Our results demonstrated that a newly synthesized Pr55gag was initially targeted to the plasma membrane. Pr55gag accumulated at the plasma membrane when coexpressed with the viral accessory protein Vpu but, in the absence of Vpu, a significant fraction of the cell surface-associated Pr55gag proteins was retargeted to internal membranes by endocytosis. This retargeting partially shifted virus assembly from the cell surface to internal membranes. RNA interference (RNAi)-mediated knockdown of Tsg101, or deletion of the p6 domain from Pr55gag suppressed the endocytic uptake of Pr55gag when the protein was expressed in the absence of Vpu. Taken together, these results suggest that the site of HIV-1 assembly is determined by factors that regulate endocytosis of Pr55gag, and the results identified a previously unknown role for Vpu and Tsg101 as regulators of the endocytic uptake of Pr55gag.
MATERIALS AND METHODS
Cell culture and transfection. HeLa H1 cells (American Type Culture Collection) were grown in Dulbecco’s modified Eagle medium supplemented with 7% fetal bovine serum (EuroClone) and nonessential amino acids (i.e., HeLa growth medium). 293T cells were cultured as previously described (18). Vesicular stomatitis virus (VSV) G-protein-pseudotyped infectious lentivirus vectors carrying modified HIV-1 genomes were produced by cotransfecting pCMVR8.91 and pMD2.VSV.G (generous gifts from Didier Trono [48]), together with a plasmid directing the synthesis of a recombinant HIV-1 genome into 293T cells by a calcium phosphate precipitation technique.
Plasmid constructs. pHXB2D and pNL4-3 were originally obtained from the NIBSC Centralized Facility for AIDS Reagents (the plasmids were donated by Robert Gallo and Mikulas Popovic and by Malcolm Martin, respectively). pHXB2D is a HIV-1 molecular clone with a defective vpu gene and truncated vpr and nef genes (38). Plasmid pHXB2D-Gag is a derivative of pHXB2D that expresses an unprocessed Pr55gag due to an inactivation of the Pol coding region. The plasmid was constructed by replacing the 423-bp ApaI-BclI fragment with a PCR-derived 297-bp ApaI-BclI fragment in which the stop codon of Pr55gag was immediately followed by SmaI and BclI sites. pHXB2D-GagHA and pHXB2D-Gagflag are derivatives of pHXB2D-Gag which express unprocessed Pr55gag proteins that contain hemagglutinin (HA) or FLAG epitopes at their carboxy termini. pNL4-3(Gag), a derivative of pNL4-3 (1), has an inactive Pol-region and expresses an unprocessed Pr55gag. The plasmid was constructed by replacing the 3,779-bp ApaI-SalI fragment with the equivalent ApaI-SalI fragment of pHXB2D-Gag. pNL4-3(Gag/Vpu), a derivative of pNL4-3(Gag), has a defective vpu gene due to a point mutation in the AUG start codon of vpu. The construct was created by replacing the 558-bp SalI-KpnI fragment with the equivalent fragment of pHXB2D. The plasmid pNL4-3(Gagp6/Vpu) encoding a carboxy terminally truncated unprocessed Pr55gag lacking the p6 domain was made from pNL4-3(Gag/Vpu) by replacing the 791-bp SpeI-SmaI fragment with a PCR-created 635-bp SpeI-SmaI fragment in which a stop codon and a SmaI site were placed immediately after the last codon of SP2 peptide. pNL4-3(Env/Nef) is a derivative of pNL4-3 with defective env and nef genes. The first step in construction of pNL4-3(Env/Nef) involved inactivation of the env gene on pNL4-3(Gag) by deleting the 580-bp BglII-BglII fragment from the Env coding region. This created the plasmid pNL4-3(Gag/Env). pNL4-3(Env/Nef) was made by replacing the 2,680-bp SalI-BamHI fragment of pNL4-3-R71.Nef- (35): the plasmid contains an inactivated nef gene) with the equivalent fragment from pNL4-3(Gag/Env). pHXB2D-(Vpu/Env/Nef) contains inactive vpu, env, and nef genes. The first step in the construction of pHXB2D-(Vpu/Env/Nef) was inactivation of the env gene on pHXB2D by deleting a 1,278-bp region between KpnI and BglII sites on env which in addition to an internal deletion also introduced a frameshift mutation on the gene. This created the plasmid pHXB2D-(Vpu/Env). pHXB2D-(Vpu/Env/Nef) was made from pHXB2D-(Vpu/Env) by inactivating the truncated nef gene via converting cohesive ends of XhoI-linearized pHXB2D-(Vpu/Env) into blunt ends by Klenow and subsequently religating the plasmid.
Subcellular fractionations. HeLa H1 cells were seeded on 10-cm plates at a density which gave 70% confluence after o/n growth. Cells were infected the following day with VSV-G-pseudotyped lentivirus vectors carrying modified HIV-1 genomes. At 24 h postinfection the now-confluent cell monolayers were metabolically labeled with 50 μCi of [35S]methionine for 30 min (18) and chased for various times in HeLa growth medium containing a 10-fold excess of cold methionine. After the chase, cell monolayers were washed twice with phosphate-buffered saline containing 0.02% EDTA (PBS-EDTA), and the cells were brought into suspension by a 2-min incubation in PBS-EDTA at 37°C. Subsequently, the suspension cells were coated at 4°C with cationic silica beads essentially as previously described (4, 41), except that the coating buffer used was 20 mM 2-[N-morpholine]ethanesulfonic acid (pH 6.66 to 6.68) and 150 mM NaCl. After coating, cells were homogenized with a tight-fitting Dounce homogenizer in buffer containing 25 mM HEPES (pH 7.4), 150 mM NaCl, 2 mM MgCl2, 20 μg of phenylmethylsulfonyl fluoride, 1 μg of CLAP (chymostatin, leupeptin, aprotinin, and pepstatin A), and 2.5 mg of heparin per ml (the heparin was included to suppress binding of Pr55gag to chromatin released from broken nuclei). The crude cell homogenates were mixed with 100% Nycodenz solution (AXIS-SHIELD PoC AS) to give a final 60% Nycodenz concentration. The sample (2 ml) was placed on a 1-ml cushion of 70% Nycodenz and overlaid with 1.5 ml of 50% Nycodenz and 0.5 ml of 25 mM HEPES (pH 7.4), 150 mM NaCl, and 1 mM EDTA. The gradient was centrifuged in an SW55 Ti rotor at 63,000 x g for 45 min at 4°C. The pellet (i.e., the plasma membrane fraction) was resuspended in hot 0.5% sodium dodecyl sulfate (SDS)-0.5 M NaCl buffer and then boiled for 5 min, and silica beads were removed from the resuspended samples by centrifugation in an Eppendorf centrifuge at full speed for 1 min. The internal membrane fraction at the 50% Nycodenz-buffer interphase was collected and diluted fivefold with 25 mM HEPES (pH 7.4), 150 mM NaCl, and 1 mM EDTA, and membranes were pelleted by centrifugation at 100,000 x g for 1 h at 4°C. The internal membrane pellet was solubilized as described above for the plasma membrane, and the samples were diluted with a 10-fold excess of lysis buffer (1% Triton X-100, 25 mM HEPES [pH 7.4], 150 mM NaCl, 1 mM EDTA, 20 μg of phenylmethylsulfonyl fluoride, and 1 μg of CLAP per ml). Pr55gag proteins were immunoprecipitated from the lysates with mouse anti-HIV-1 CA monoclonal antibody 38:9 (17), the Pr55gag processing intermediates and CA with a mixture of monoclonal antibody 38:9 and mouse monoclonal anti-p24 antibody 32/5.17.76 (Abcam, Ltd.), and the HA-tagged Pr55gag with a rabbit polyclonal anti-HA antibody (Sigma).
For biotinylation of the cell surface proteins, HeLa H1 cells were brought into suspension by treatment with PBS-EDTA and incubated in PBS (with MgCl2 and CaCl2) containing 0.5 mg per ml of EZ-Link Sulfo-NHS-SS-biotin (Pierce) for 30 min at 0°C. The sample was quenched with 50 mM NH4Cl before silica coating. Biotin-tagged proteins in the plasma membrane and the internal membrane fractions were visualized by dot blots using streptavidin-peroxidase polymer (Sigma). Biotin-tagged transferrin proteins (15 μg per ml; Molecular Probes) that had been bound to HeLa H1 cells at 0°C or internalized for 20 min at 37°C in serum-free medium were used as markers for the plasma membrane and recycling endosomes, respectively. In the case of the internalized transferrin, cells were incubated in a pH 4.5 buffer containing 25 mM citric acid, 24 mM trisodiumcitrate, 280 mM sucrose, and 10 μM deferoxamine mesylate for 2 min to remove cell surface-bound transferrin proteins prior to silica coating. Transferrin proteins in the plasma membrane and internal membrane fractions were visualized by Western blotting with streptavidin-peroxidase polymer. The amounts of the endoplasmic reticulum (ER) marker protein calnexin and the late endosome/lysosome marker protein Lamp-1 in the plasma membrane and internal membrane fractions were determined by Western blotting with rabbit polyclonal anti-calnexin (StressGen Biotechnologies Corp.) and goat polyclonal anti-Lamp-1 (C-20; Santa Cruz Biotechnology, Inc.), respectively.
Electron miscroscopy. Cells were fixed at 24 h postinfection with 2% glutaraldehyde in 100 mM HEPES (pH 7.4). The fixed samples were first stained with 1% osmium tetroxide, followed by staining with 0.5% alcoholic uranyl acetate, dehydration in ethanol, and embedding in Agar 100 resin. Sections (60 nm) were cut on an LKB ultramicrotome and imaged in an FEI Tecnai Spirit transmission electron microscope.
RNAi. HeLa H1 cells were seeded on 10-cm plates at a density which gave 50% confluence after overnight growth. siRNAs directed against the Tsg101 (7) or a control siRNA (GGUCUCUCUUUCGACGUGCdTdT; BLAST searches against the human genome sequence did not identify any target for this siRNA) were transfected into HeLa H1 cells by using Oligofectamine (Invitrogen) in accordance with the manufacturer's instructions. Transfections were carried out in HeLa growth medium containing 7% fetal bovine serum, and 40 μl oligofectamine and 20 pmol siRNA per 10-cm plate were used. Mock-transfected cells were treated with Oligofectamine alone. At about 22 h posttransfection the cells were infected with recombinant lentivirus vectors and then analyzed about 24 h postinfection. The intracellular levels of Tsg101 in mock- and siRNA-treated cells were determined from postnuclear supernatants of cell extracts by Western blotting with a monoclonal anti-Tsg101 antibody (C-2; Santa Cruz Biotechnology, Inc.). The monoclonal anti-actin antibody (Chemicon) was used as a control.
RESULTS
Subcellular fractionation method to separate plasma membrane from internal membranes. We have expressed Pr55gag in HeLa H1 cells and analyzed the targeting of the newly synthesized Pr55gag by pulse-chase studies and subcellular fractionations of cell extracts. HeLa H1 cells were chosen since our pilot experiments indicated that Pr55gag can accumulate in these cells either at the plasma membrane or at the internal membranes. Thus, these cells provide a good model system for studying how plasma membrane versus internal localization of the protein is achieved. To determine whether Pr55gag is initially targeted to the plasma membrane or internal membranes requires efficient separation of these membranes. During homogenization, plasma membrane breaks into vesicles and membrane sheets of various density (see, for example, reference 42) that are difficult to separate from, for example, endosomal membranes on conventional sucrose or iodixanol density gradients. Therefore, we chose a fractionation method in which intact cells were coated with cationic silica beads at 4°C prior to homogenization (4, 41), and crude cell extracts were fractionated on a Nycodenz step gradient. The step gradient consisted of a 70% cushion and a 60% sample loading zone overlaid with 50% Nycodenz and buffer solutions. Since the bound silica beads greatly increase density of the plasma membrane, the plasma membrane-derived vesicles and membrane sheets pellet through the 70% Nycodenz cushion during ultracentrifugation (the PM fraction), whereas internal membranes float to the 50% Nycodenz-buffer interphase (I, the internal membrane fraction). Figure 1 shows controls for this fractionation method. In Fig. 1A, externally exposed cell surface-associated proteins were tagged with biotin prior to coating, and biotinylated proteins in the PM and I fractions were visualized by dot blots with horseradish peroxidase-conjugated streptavidin. The majority of biotin-tagged proteins were found in the PM fraction. Similarly, biotin-conjugated transferrin that was bound to cells at 0°C prior to coating was predominantly in the PM fraction (Fig. 1B, Tf surface). In contrast, biotin-conjugated transferrin that had been concentrated into recycling endosomes by internalization at 37°C for 20 min was in the I fraction (Fig. 1B, Tf internalized). The ER protein calnexin, as well as the endosomal/lysosomal marker lysosome-associated membrane protein 1 (Lamp-1), were predominantly in the I fraction (Fig. 1B). Taken together, these results demonstrate that the fractionation method can separate the plasma membrane from internal membranes in HeLa H1 cell extracts.
Newly synthesized Pr55gag is inserted into the plasma membrane. To analyze targeting of newly synthesized Pr55gag, HeLa H1 cells were infected with VSV-G-pseudotyped lentivirus vectors carrying modified HIV-1 genomes. At 24 h postinfection, cells were metabolically labeled with [35S]methionine for 30 min and chased for various times prior to coating and fractionation. Confluent cell monolayers were used in the studies. Amounts of labeled Pr55gag proteins in the PM and I fractions were determined by immunoprecipitation and by SDS-polyacrylamide gel electrophoresis analysis of the immunoprecipitates. Figure 2A shows analysis of newly synthesized, unprocessed Pr55gag expressed from a NL4-3 proviral genome from which the Pol region was deleted [the NL4-3(Gag) genome]. After a 5-min chase, a significant amount of labeled Pr55gag was detected in the PM fraction, but essentially no labeled Pr55gag was present in the I fraction. Similar results were obtained with a 15-min chase (data not shown). Beginning from the 60-min chase point, small amounts of labeled Pr55gag appeared in the I fraction, but the majority of the protein (>80%) was found in the PM fraction at all chase points tested. The NL4-3(Gag) proviral construct efficiently produced VLPs, since large amounts of labeled Pr55gag was found in the culture medium already after 2 h of chase (data not shown). If silica coating was omitted, all labeled Pr55gag was in the I fraction (Fig. 2B), thus confirming that Pr55gag found in the PM fraction in coated cells represented cell surface-associated proteins, and no unspecific protein aggregates. A nonmyristoylated mutant of Pr55gag in which the amino-terminal glycine residue was mutated to alanine failed to associate with either the PM or I fractions (data not shown). Taken together, these results indicated that newly synthesized Pr55gag was inserted into the plasma membrane, and the majority of the proteins remained at the cell surface.
Deletion of the vpu gene induces endocytosis of the plasma membrane-associated Pr55gag. We next tested whether localization of Pr55gag to the plasma membrane was an intrinsic property of the protein, or whether other viral proteins had an effect on intracellular targeting of Pr55gag. Elimination of env or nef genes did not significantly alter the targeting of newly synthesized Pr55gag, i.e., the protein was inserted into the plasma membrane and the majority of labeled Pr55gag remained at the plasma membrane in the absence of Env or Nef expressions (data not shown). In contrast, deletion of the vpu gene produced a dramatic change in the intracellular localization of Pr55gag. As shown in Fig. 3A, after a 15-min chase, or 30-min chase (data not shown), the vast majority of labeled Pr55gag was present in the PM fraction in NL4-3(Gag/Vpu)-infected cells. Beginning from the 60-min chase point, significant amounts of labeled Pr55gag appeared in the I fraction. By 2 h of chase, 70% of labeled Pr55gag was shifted to internal membranes. As previously reported for HeLa cells (22), Vpu was also required for efficient virus production from HeLa H1 cells, since NL4-3(Gag)-infected cells produced 18-fold more VLPs than NL4-3(Gag/Vpu)-infected cells (data not shown). Since newly synthesized cytosolic Pr55gag is rapidly targeted to membranes, within 30 min of synthesis (43), the results presented in Fig. 3A suggest that in the absence of Vpu expression Pr55gag is initially targeted to the plasma membrane but is subsequently redistributed to the internal membranes by endocytosis. As shown in Fig. 3B, a carboxy-terminal HA tag, but not a FLAG tag, suppressed endocytosis of cell surface-associated Pr55gag. After 15-min chase, Pr55gag/HA and Pr55gag/FLAG expressed in the absence of Vpu were primarily at the plasma membrane (data not shown) but after 3 h of chase, 60% of labeled Pr55gag/FLAG had redistributed to internal membranes, whereas >80% of Pr55gag/HA remained at the cell surface even after 4 h of chase.
The effect of Vpu was also tested in the context of Gag and Gag-Pol coexpression. Figure 4A shows an analysis of processed Pr55gag that was expressed from a modified NL4-3 proviral genome lacking env and nef genes but expressing Vpu [the NL4-3(Env/Nef)]. At all of the time points tested, most of the labeled Pr55gag, its processing intermediates, or the p25/p24 CA forms were found in the PM fraction. Only trace amounts of the proteins could be detected in the I fraction. In contrast, if in addition to the env and nef genes the vpu gene was also deleted [the HXB2D-(Vpu/Env/Nef) construct], the intracellular localization of CA forms was changed (Fig. 4B). After 15 min of chase, Pr55gag and two of its early processing intermediates were detected in the PM fraction but not in the I fraction, thus indicating that Pr55gag was first targeted to the plasma membrane. After 2 h of chase, Pr55gag and its early processing intermediates were still found in the PM fraction but, significantly, 60% of the processed CA forms were in the internal membrane fraction. At the 4-h chase point, only the p25 and p24 CA forms were evident, and the majority of these proteins was found in internal membranes. Although CA was the only Gag form detected in the I fraction in the radioactively labeled samples, Western blot analysis of the PM and I fractions indicated that small amounts of Pr55gag and its processing intermediates were present in internal membranes as well (Fig. 4C). Taken together, the results in Fig. 4 demonstrate that deletion of vpu gene induced uptake of plasma membrane-associated core proteins also when Gag and Gag-Pol were coexpressed. These results were confirmed by EM. EM analysis of NL4-3(Env/Nef)-infected cells identified numerous budding profiles at the cell surface (Fig. 5A). Budding profiles were detected at the plasma membrane in HXB2D-(Vpu/Env/Nef)-infected cells as well (Fig. 5C), but these cells also contained endosome-like intracellular compartments full of VLP-sized vesicles (Fig. 5B). Occasionally, budding profiles resembling nascent VLPs could be observed on the limiting membrane of these compartments (Fig. 5D). When 10 randomly selected NL4-3(Env/Nef)- and HXB2D-(Vpu/Env/Nef)-infected cells, i.e., cells that had budding structures at the cell surface, were scored for VLP-containing endosomal structures, these structures were found in all of the inspected HXB2D-(Vpu/Env/Nef)-infected cells. In contrast, only one of the NL4-3(Env/Nef)-infected cells had clearly identifiable endosomal VLPs.
Knockdown of Tsg101 inhibits targeting of Pr55gag to internal membranes. The PTAP motif on the p6 domain of Pr55gag binds Tsg101, and this interaction is necessary for the efficient release of progeny particles (7, 23, 45). We tested the effect of a previously described Tsg101-specific small interfering RNAs (siRNAs) (7) in NL4-3(Gag/Vpu)-infected cells to probe whether Tsg101 is involved in endocytosis of Pr55gag. Western blot analysis of cell lysates demonstrated that Tsg101 siRNAs effectively reduced intracellular levels of Tsg101 (Fig. 6A). After a 30-min pulse and 3 h of chase, 58% of labeled Pr55gag remained at the cell surface in Tsg101 siRNA-transfected cells, whereas 73% of labeled Pr55gag was in internal membranes in mock-transfected cells (Fig. 6B). Cells transfected with control siRNAs of scrambled sequence had 63% of labeled Pr55gag in internal membranes (Fig. 6B). The less-efficient Pr55gag uptake in Tsg101 siRNA-transfected cells could be due to ablation of Pr55gag interaction with Tsg101, or, alternatively, due to unspecific and/or pleiotropic effects of the Tsg101 siRNAs. To distinguish between these possibilities, we tested uptake of a carboxy-terminally truncated Pr55gag lacking the p6 domain. The truncated protein was expressed from the recombinant proviral genome NL4-3(Gagp6/Vpu). After a 30-min pulse and a 15-min chase, the labeled Gagp6 protein was in the PM fraction (data not shown). Deletion of the p6 domain inhibited endocytosis of the cell surface-associated protein, since 75% of the truncated proteins were still at the plasma membrane after 3 h of chase (Fig. 6C). Taken together, these results suggest that endocytosis of Pr55gag is critically dependent on interaction of Tsg101 with the p6 domain of Pr55gag.
DISCUSSION
In the present study, we have analyzed targeting of newly synthesized Pr55gag in HeLa H1 cells by pulse-chase studies and by using a subcellular fractionation method that efficiently separates plasma membrane from internal membranes. Our results indicate that Pr55gag is initially targeted to the plasma membrane, and the protein is delivered to internal membranes by endocytosis. Our study also identified Vpu as a key regulator of intracellular targeting of Pr55gag. In the presence of Vpu, the bulk of Pr55gag remained at the plasma membrane after the initial membrane insertion, whereas in the absence of coexpressed Vpu a significant fraction of the cell surface-associated Pr55gag proteins was retargeted to internal membranes. Similar retargeting was observed regardless of whether Pr55gag was expressed as a nonprocessed precursor protein or whether Pr55gag was expressed together with Gag-Pol. These results are in agreement with previous reports indicating that deletion of the vpu gene increases virus budding into intracellular compartments in HeLa and T cells (13, 22). We observed a similar increase of VLPs in internal membrane compartments in the absence of Vpu when NL4-3(Env/Nef)- and HXB2D-(Vpu/Env/Nef)-infected cells were compared by EM. Furthermore, it is noteworthy that in all of the studies that have identified Pr55gag in internal membranes in transfected or infected, but otherwise nonperturbed HeLa, Cos-1, or 293T cells, the protein was expressed in the absence of Vpu (6, 10, 26, 39). Consistent with our results, Rudner et al. (34) recently identified plasma membrane as the primary target membrane for Pr55gag in HeLa cells. The method used was the fluorescent imaging of cells expressing a tetracysteine-tagged Pr55gag and sequential staining of the cells with two biarsenical compounds that produced different fluorescent signals upon binding to the tetracysteine tag. The dual-labeling procedure enabled the detection of newly synthesized Pr55gag. However, in contrast to our results, the presence or absence of Vpu was reported to have no effect on the association of Pr55gag with the plasma membrane. This discrepancy could perhaps be explained by the carboxy-terminal tetracysteine tag; as exemplified by our HA-tagged Pr55gag, a carboxy-terminal tag can inhibit the endocytosis of cell surface-associated Pr55gag. Since our subcellular fractionation method did not separate internal membrane compartments from each other, the identity of the internal membranes harboring Pr55gag was not established. Fluorescence microscopy analysis of HXB2D-(Vpu/Env/Nef)-infected cells indicated a vesicular staining pattern for intracellular Gag (data not shown), but the intracellular Gag-signal in this particular viral genetic background did not extensively colocalize with any of the tested early or late endosomal markers (CD63, Lamp-1, EEA1, or internalized fluorescein isothiocyanate-conjugated transferrin; data not shown).
It is formally possible that the Gag signal in the internal membrane fraction in our silica coating experiments originated from internalized extracellular VLPs rather than from endocytosed plasma membrane-associated Pr55gag. However, we think this is unlikely for the following reasons. First, very little Gag was detected in the internal membrane fraction in NL4-3(Gag)- and NL4-3(Env/Nef)-infected cells despite the fact that these cells released 18-fold more VLPs than cells expressing the respective Vpu-genomes (data not shown). Second, although rare, virus-like budding structures could be identified on internal membrane compartments, and early processing intermediates of Pr55gag, although not as abundant as CA, were observed in internal membrane fraction by Western blotting. Thus, unless VLPs produced in the absence of Vpu have some feature that promotes their rapid uptake after budding from the cell surface, these observations strongly suggest that Gag localization to internal membranes is the consequence of endocytosis of plasma membrane-associated proteins rather than internalization of extracellular VLPs. The mechanism by which Vpu suppresses endocytosis of Pr55gag remains to be elucidated. Since there are no indications that Vpu directly interacts with Pr55gag, Vpu most likely exerts its effect indirectly, e.g., by inducing modifications on the plasma membrane or by exerting broader modifying effects on the membrane trafficking pathways of the host cell. In order to elucidate how Vpu suppresses endocytosis of Pr55gag, it is essential to understand the molecular basis of endocytosis of Pr55gag when the protein is expressed in the absence of Vpu. Our initial analyses suggest that this uptake requires interactions between the cell surface-associated Pr55gag and Tsg101. siRNAs directed against Tsg101, or deletion of the p6 domain that mediates Pr55gag and Tsg101 interactions (7, 23, 45), inhibited the endocytosis of Pr55gag. siRNAs can cause various off-target effects that are difficult to control for in conventional experimental settings (32, 36), and therefore RNAi-mediated knockdown results have to be interpreted with caution. However, since the endocytosis of Pr55gag was suppressed by deletion of p6 domain, this suggests that the effect of Tsg101 siRNAs was due to ablation of Tsg101 and Pr55gag interactions. Tsg101 was initially identified as being essential for efficient egress of progeny particles from the cell surface assembly sites (7, 23, 45). Tsg101 is a component of the ESCRT-I (endocytic sorting complex required for trafficking) and recruits the so-called class E proteins, which normally function in MVB biogenesis, to the site of HIV-1 assembly and budding (24). The class E proteins also participate in the budding of other retroviruses, filoviruses, and arenaviruses (24). Since HIV-1 assembly is arrested at a late stage (after membrane distortion) when p6-Tsg101 interactions are disrupted (24), it has been assumed that Tsg101 and other class E proteins facilitate the final steps in budding. However, since our results indicate that Tsg101 also has an impact on intracellular localization of Pr55gag, this suggests that the function of Tsg101 in HIV-1 assembly and budding might be more complex than was previously thought.
Several biological activities have been attributed to Vpu (2, 19, 44). One of these is the enhancement of virus release from HIV-1-infected cells. Interestingly, Vpu can also increase virus release from cells infected by other retroviruses (13) and even alphaviruses (11). Since Vpu enhances virus production from most human cells, as well as from human-simian heterokaryons, but not from simian cells, Varthakavi et al. (44) postulated that Vpu counteracts a human cell-specific restriction factor that inhibits HIV-1 particle production. Although our present study by no means excludes the existence of such an assembly/release inhibitor, the data suggest a possible alternative explanation for the Vpu-mediated enhancement of HIV-1 particle release in human cells. Namely, Vpu could affect particle production indirectly via its effects on intracellular targeting of Gag. Since deletion of the vpu gene subjected plasma membrane-associated Gag to endocytosis and partially shifted virus assembly to internal membranes, and if these endosomal viruses were only inefficiently released into the extracellular medium, then Vpu-deficient viruses would be expected to produce less extracellular particles than their wild-type counterparts. Vpu has also been reported to enhance overall membrane association of Gag when Gag was expressed from a T7-driven vaccinia virus expression system (5). Although we did not rigorously test the distribution of Gag between cytosolic and membrane fractions in the present study, we, or others (15), have not observed any dramatic reduction in membrane-associated Gag with Vpu-deficient proviral expression systems.
The biological significance of the plasma membrane versus the endosomal assembly of HIV-1 is not fully understood at present, but emerging information suggests that the site of virus assembly could impact HIV-1 pathogenesis. For example, intracellular assembly may promote viral persistence since intracellular virions in macrophages retain infectivity for an extended period of time and can efficiently transfer the infection to lymphocytes upon contact (37). Furthermore, it has been speculated that the cellular site of assembly might have a direct impact on immune recognition and infectivity of the formed virions (12). Thus, understanding the molecular mechanisms that regulate the choice of viral assembly site is important. Our results suggest that cell surface versus internal assembly of HIV-1 in HeLa H1 cells is determined by factors that regulate endocytic uptake of plasma membrane-associated Pr55gag. Our results provide compelling evidence that Vpu promotes cell surface assembly of HIV-1 by inhibiting endocytosis of plasma membrane-associated Pr55gag and that newly synthesized Pr55gag proteins reach internal membrane compartments in HeLa H1 cells via endocytosis from the plasma membrane. The presence of VLPs in internal compartments in HXB2D-(Vpu/Env/Nef)-infected cells implies that internalized Pr55gag proteins can engage in productive particle assembly and that the internalization process does not merely represent garbage disposal, i.e., the removal of defective Gag forms from the cell surface. Since HeLa cells recapitulate the main features of HIV-1 assembly in T cells, our results most likely apply to T cells as well. Unfortunately, we could not directly test the targeting of Pr55gag in T cells, since, e.g., Jurkat T cells tended to disrupt during silica coating, and this lead to binding of silica beads to internal membranes as well (data not shown). Whether endocytosis from the plasma membrane is also the mechanism by which Pr55gag is targeted to the MVBs in macrophages is unclear at present, since intracellular targeting of Pr55gag in HeLa H1 cells and macrophages differs at least in two aspects. Namely, Gag is apparently targeted to MVBs in macrophages even when the protein is coexpressed with Vpu (25, 31, 33) and targeting to MVBs in macrophages appears to be independent of Pr55gag-Tsg101 interactions (27). Consequently, either the uptake mechanism is different in macrophages or, alternatively, the initial membrane insertion does not occur at the plasma membrane in these cells. We are currently testing whether the silica coating method can be used to determine the targeting phenotype of newly synthesized Pr55gag in macrophages.
ACKNOWLEDGMENTS
We thank Didier Trono and Kalle Saksela for providing plasmids and Birgitta Lindqvist and Cornelia Muncke for expert technical assistance. Plasmids pHXB2D and pNL4-3 were obtained from the NIBSC Centralized Facility for AIDS Reagents.
This study was supported by grants from Academy of Finland and Swedish Research Council to M.S.
REFERENCES
Adachi, A., H. E. Gendelman, S. Koenig, S. Folks, R. Willey, A. Rabson, and M. A. Martin. 1986. Production of acquired immunodeficiency syndrome-associated retrovirus in human and nonhuman cells transfected with an infectious molecular retrovirus clone. J. Virol. 59:284-291.
Bour, S., and K. Strebel. 2003. The HIV-1 Vpu protein: a multifunctional enhancer of viral particle release. Microbes Infect. 5:1029-1039.
Bryant, M., and L. Ratner. 1990. Myristoylation-dependent replication and assembly of human immunodeficiency virus 1. Proc. Natl. Acad. Sci. USA 87:523-527.
Chaney, L. K., and B. S. Jacobson. 1983. Coating cells with colloidal silica for high yield isolation of plasma membrane sheets and identification of transmembrane proteins. J. Biol. Chem. 258:10062-10072.
Deora, A., P. Spearman, and L. Ratner. 2000. The N-terminal matrix domain of HIV-1 Gag is sufficient but not necessary for viral protein U-mediated enhancement of particle release through a membrane-targeting mechanism. Virology 269:305-312.
Dong, X., H. Li, A. Derdowski, L. Ding, A. Burnett, X. Chen, T. R. Peters, T. S. Dermody, E. Woodruff, J.-J. Wang, and P. Spearman. 2005. AP-3 directs the intracellular trafficking of HIV-1 Gag and plays a key role in particle assembly. Cell 120:663-674.
Garrus, J. E., U. K. von Schwedler, O. W. Pornillos, S. G. Morham, K. H. Zavitz, H. E. Wang, D. A. Wettstein, K. M. Stray, M. Cte, R. L. Rich, D. G. Myszka, and W. I. Sundquist. 2001. Tsg101 and the vacuolar protein sorting pathway are essential for HIV-1 budding. Cell 107:55-65.
Gelderblom, H. R., E. H. S. Hausman, M. Ozel, G. Pauli, and M. A. Koch. 1987. Fine structure of human immunodeficiency virus (HIV) and immunolocalization of structural proteins. Virol. 156:171-176.
Gheysen, D., E. Jacobs, F. d. Foresta, C. Thiriart, M. Francotte, D. Thines, and M. D. Wilde. 1989. Assembly and release of HIV-1 precursor Pr55gag virus-like particles from recombinant baculovirus-infected cells. Cell 59:103-112.
Goff, A., L. S. Ehrlich, S. N. Cohen, and C. A. Carter. 2003. Tsg101 control of human immunodeficiency virus type 1 Gag trafficking and release. J. Virol. 77:9173-9182.
González, M. E., and L. Carrasco. 2001. Human immunodeficiency virus type 1 Vpu protein affects Sindbis virus glycoprotein processing and enhances membrane permeabilization. Virol. 279:201-209.
Gould, S. J., A. M. Booth, and J. E. K. Hildreth. 2003. The Trojan exosome hypothesis. Proc. Natl. Acad. Sci. USA 100:10592-10597.
Gttlinger, H. G., T. Dorfman, E. A. Cohen, and W. A. Haseltine. 1993. Vpu protein of human immunodeficiency virus type 1 enhances the release of capsids produced by gag gene constructs of widely divergent retroviruses. Proc. Natl. Acad. Sci. USA 90:7381-7385.
Gttlinger, H. G., J. G. Sodroski, and W. A. Haseltine. 1989. Role of capsid precursor processing and myristoylation in morphogenesis and infectivity of human immunodeficiency virus type 1. Proc. Natl. Acad. Sci. USA 86:3195-3199.
Handley, M. A., S. Paddock, A. Dall, and A. T. Panganiban. 2001. Association of Vpu-binding protein with microtubules and Vpu-dependent redistribution of HIV-1 Gag protein. Virol. 291:198-207.
Hermida-Matsumoto, L., and M. D. Resh. 2000. Localization of human immunodeficiency virus type 1 Gag and env at the plasma membrane by confocal imaging. J. Virol. 74:8670-8679.
Hinkula, J., L. Rosen, V.-A. Sundqvist, T. Stigbrand, and B. Wahren. 1990. Epitope mapping of the HIV-1 Gag region with monoclonal antibodies. Mol. Immunol. 27:395-403.
Holm, K., K. Weclewicz, R. Hewson, and M. Suomalainen. 2003. HIV-1 assembly and lipid rafts: Pr55gag complexes associate with membrane-domains that are largely resistant to Brij 98, but sensitive to Triton X-100. J. Virol. 77:4805-4817.
Hsu, K., J. Seharaseyon, P. Dong, S. Bour, and E. Marbán. 2004. Mutual functional destruction of HIV-1 Vpu and host TASK-1 channel. Mol. Cell 14:259-267.
Johnson, D. R., R. S. Bhatnagar, L. J. Knoll, and J. I. Gordon. 1994. Genetic and biochemical studies of protein N-myristylation. Annu. Rev. Biochem. 63:869-914.
Karacostas, V., K. Nagashima, M. A. Gonda, and B. Moss. 1989. Human immunodeficiency virus-like particles produced by a vaccinia virus expression vector. Proc. Natl. Acad. Sci. USA 86:8964-8967.
Klimkait, T., K. Strebel, M. D. Hoggan, M. A. Martin, and J. M. Orenstein. 1990. The human immunodeficiency virus type 1-specific protein Vpu is required for efficient virus maturation and release. J. Virol. 64:621-629.
Martin-Serrano, J., T. Zang, and P. D. Bieniasz. 2001. HIV-1 and Ebola virus encode small peptide motifs that recruit Tsg101 to sites of particle assembly to facilitate egress. Nat. Med. 7:1313-1319.
Morita, E., and W. I. Sundquist. 2004. Retrovirus budding. Annu. Rev. Cell Dev. Biol. 20:395-425.
Nguyen, D. H., and J. E. K. Hildreth. 2000. Evidence for budding of human immunodeficiency virus type 1 selectively from glycolipid-enriched membrane lipid rafts. J. Virol. 74:3264-3272.
Nydegger, S., M. Foti, A. Derdowski, P. Spearman, and M. Thali. 2003. HIV-1 egress is gated through late endosomal membranes. Traffic 4:902-910.
Ono, A., and E. O. Freed. 2004. Cell-type-dependent targeting of human immunodeficiency virus type 1 assembly to the plasma membrane and the multivesicular body. J. Virol. 78:1552-1563.
Orenstein, J. M., M. S. Meltzer, T. Phipps, and H. E. Gendelman. 1988. Cytoplasmic assembly and accumulation of human immunodeficiency virus types 1 and 2 in recombinant human colony-stimulating-factor-1-treated human monocytes: an ultrastructural study. J. Virol. 62:2578-2586.
Pal, R., R. C. Gallo, and M. G. Sarngadharan. 1988. Processing of the structural proteins of human immunodeficiency virus type 1 in the presence of monensin and cerulenin. Proc. Natl. Acad. Sci. USA 85:9283-9286.
Pal, R., S. Mumbauer, G. M. Hoke, A. Takatsuki, and M. G. Sarngadharan. 1991. Brefeldin A inhibits the processing and secretion of envelope glycoproteins of human immunodeficiency virus type 1. AIDS Res. Hum. Retrovir. 7:707-712.
Pelchen-Matthews, A., B. Kramer, and M. Marsh. 2003. Infectious HIV-1 assembles in late endosomes in primary macrophages. J. Cell Biol. 162:443-455.
Persengiev, S. P., X. Zhu, and M. R. Green. 2004. Nonspecific, concentration-dependent stimulation and repression of mammalian gene expression by small interfering RNAs (siRNAs). RNA 10:12-18.
Raposo, G., M. S. Moore, D. Innes, R. Leijendekker, A. Leigh-Brown, P. Benaroch, and H. J. Geuze. 2002. Human macrophages accumulate HIV-1 particles in MHC II compartments. Traffic 3:718-729.
Rudner, L., S. Nydegger, L. V. Coren, K. Nagashima, M. Thali, and D. E. Ott. 2005. Dynamic fluorescent imaging of human immunodeficiency virus type 1 Gag in live cells by biarsenical labeling. J. Virol. 79:4055-4065.
Saksela, K., G. Cheng, and D. Baltimore. 1995. Proline-rich (PxxP) motifs in HIV-1 Nef bind SH3 domains of a subset of Src kinases and are required for the enhanced growth of Nef+ viruses but not for down-regulation of CD4. EMBO J. 14:484-491.
Scacheri, P. C., O. Rozenblatt-Rosen, N. J. Caplen, T. G. Wolfsberg, L. Umayam, J. C. Lee, C. M. Hughes, A. Shanmugam, M. Meyerson, and F. S. Collins. 2004. Short interfering RNAs can induce unexpected and divergent changes in the levels of untargeted proteins in mammalian cells. Proc. Natl. Acad. Sci. USA 101:1892-1897.
Sharova, N., S. Swingler, M. Sharkey, and M. Stevenson. 2005. Macrophages archive HIV-1 virions for dissemination in trans. EMBO J. 24:2481-2489.
Shaw, G. M., B. H. Hahn, S. K. Arya, J. E. Groopman, R. C. Gallo, and F. Wong-Staal. 1984. Molecular characterization of human T-cell leukemia (lymphotropic) virus type III in the acquired immune deficiency syndrome. Science 226:1165-1171.
Sherer, N. M., M. J. Lehmann, L. F. Jimenez-Soto, A. Ingmundson, S. M. Horner, G. Cicchetti, P. G. Allen, M. Pypaert, J. M. Cunningman, and W. Mothes. 2003. Visualization of retroviral replication in living cells reveals budding into multivesicular bodies. Traffic 4:785-801.
Spearman, P., J.-J. Wang, N. Vander Heyden, and L. Ratner. 1994. Identification of human immunodeficiency virus type 1 Gag protein domains essential to membrane binding and particle assembly. J. Virol. 68:3232-3242.
Spector, D. L., R. D. Goldman, and L. A. Leinwand. 1998. Plasma membrane isolation using the cationic colloidal silica isolation technique, p. 35.1-35.14. In Cells: a laboratory manual, vol. 1. Cold Spring Harbor Laboratory Press, New York, N.Y.
Suomalainen, M., K. Hultenby, and H. Garoff. 1996. Targeting of Moloney murine leukemia virus Gag precursor to the site of virus budding. J. Cell Biol. 135:1841-1852.
Tritel, M., and M. D. Resh. 2000. Kinetic analysis of human immunodeficiency virus type 1 assembly reveals the presence of sequential intermediates. J. Virol. 74:5845-5855.
Varthakavi, V., R. M. Smith, S. P. Bour, K. Strebel, and P. Spearman. 2003. Viral protein U counteracts a human host cell restriction that inhibits HIV-1 particle production. Proc. Natl. Acad. Sci. USA 100:15154-15159.
VerPlank, L., F. Bouamr, T. J. LaGrassa, B. Agresta, A. Kikonyogo, J. Leis, and C. A. Carter. 2001. Tsg101, a homologue of ubiquitin-conjugating (E2) enzymes, binds the L domain in HIV type 1 Pr55Gag. Proc. Natl. Acad. Sci. USA 98:7724-7729.
von Schwedler, U. K., M. Stuchell, B. Muller, D. W. Ward, H.-Y. Chung, E. Morita, H. E. Wang, T. Davis, G.-P. He, D. M. Cimbora, A. Scott, H.-G. Krusslich, J. Kaplan, S. G. Morham, and W. I. Sundquist. 2003. The protein network of HIV budding. Cell 114:701-713.
Zhou, W., L. J. Parent, J. W. Wills, and M. Resh. 1994. Identification of a membrane-binding domain within the amino-terminal region of human immunodeficiency virus type 1 Gag protein which interacts with acidic phospholipids. J. Virol. 68:2556-2569.
Zufferey, R., D. Nagy, R. J. Mandel, L. Naldini, and D. Trono. 1997. Multiply attenuated lentiviral vector achieves efficient gene delivery in vivo. Nat. Biotechnol. 15:871-875.(Kirsi Harila, Ian Prior, )
The Physiological Laboratory, University of Liverpool, Liverpool L69 3BX, United Kingdom
Department of Biosciences at Novum, Karolinska Institutet, S-141 57 Huddinge, Sweden
Department of Virology, Swedish Institute for Infectious Disease Control and MTC, Karolinska Institutet, 171 82 Stockholm and Department of Molecular Virology, Linkoping University, 581 83 Linkoping, Sweden
ABSTRACT
Assembly of human immunodeficiency virus type 1 (HIV-1) is directed by the viral core protein Pr55gag. Depending on the cell type, Pr55gag accumulates either at the plasma membrane or on late endosomes/multivesicular bodies. Intracellular localization of Pr55gag determines the site of virus assembly, but molecular mechanisms that define cell surface or endosomal targeting of Pr55gag are poorly characterized. We have analyzed targeting of newly synthesized Pr55gag in HeLa H1 cells by pulse-chase studies and subcellular fractionations. Our results indicated that Pr55gag was inserted into the plasma membrane and, when coexpressed with the viral accessory protein Vpu, Pr55gag remained at the plasma membrane and virions assembled at this site. In contrast, Pr55gag expressed in the absence of Vpu was initially inserted into the plasma membrane, but subsequently endocytosed, and virus assembly was partially shifted to internal membranes. This endocytosis of Pr55gag required the host protein Tsg101. These results identified a previously unknown role for Vpu and Tsg101 as regulators for the endocytic uptake of Pr55gag and suggested that the site of HIV-1 assembly is determined by factors that regulate the endocytosis of Pr55gag.
INTRODUCTION
Assembly and budding of human immunodeficiency virus type 1 (HIV-1) is driven by the viral core protein precursor Gag (Pr55gag), which assembles into enveloped virus-like particles (VLPs) when expressed in the absence of other viral proteins (9, 21). Pr55gag is a myristoylated, peripheral membrane protein (3, 9, 14, 40, 47). Concomitantly with assembly and budding, Pr55gag is processed into matrix, capsid (CA), nucleocapsid, and p6 proteins and two smaller peptides SP1 and SP2 by a viral protease present in the minor core component, the Gag-Pol precursor. An intriguing feature of the Pr55gag-mediated assembly of HIV-1 is that different cellular membranes serve as a platform for virus assembly in different cell types. Electron microscopy (EM) and fluorescence microscopy studies have indicated that Pr55gag predominantly accumulates at the plasma membrane, and progeny virions are formed at this site in T cells and model cell lines such as HeLa, 293T, and Cos-1 (see, for example, references 8, 16, and 27), although some Pr55gag has also been identified on late endosomes in these cells (6, 10, 26, 39, 46). In contrast, in macrophages Pr55gag almost exclusively concentrates on late endosomes/multivesicular bodies (MVBs) and progeny virions bud into the lumen of MVBs (25, 28, 31, 33). Egress of virus particles from the cell surface, and probably also from the endosomal assembly sites, is critically dependent on interaction of the Pr55gag p6 domain with the host protein Tsg101 (tumor susceptibility gene 101) (7, 23, 45, 46), which normally functions in formation of MVB vesicles (24). Since ablation of Pr55gag-Tsg101 interactions arrests virus budding at a stage after membrane distortion (7, 46), the primary function of Tsg101 in HIV-1 assembly and budding is thought to be the recruitment of downstream factors essential for the completion of budding and release of virions from the donor membrane (24, 46).
Molecular mechanisms that control cell surface versus endosomal accumulation of Pr55gag, and consequently the choice of virus budding site, are poorly understood. One aspect that remains to be clarified is the route by which Pr55gag reaches its cell surface or endosomal steady-state localization. Like other myristoylated proteins, Pr55gag is expected to be synthesized on cytosolic ribosomes and inserted into membranes posttranslationally (20). It is assumed that the classical endoplasmic reticulum-Golgi secretory route to the plasma membrane and endosomes is not involved in intracellular targeting of Pr55gag, since inhibitors of this transport pathway, such as brefeldin A or monensin, have no effect on the assembly of Pr55gag into extracellular particles (29, 30). Since Pr55gag has been observed on endosomal membranes in a number of different cell types (6, 10, 26, 39, 46), it is possible that the protein is first inserted into endosomal membranes and either retained in these compartments (macrophages) or transported further to the plasma membrane (e.g., HeLa, 293T, or T cells). Alternatively, newly synthesized Pr55gag could be first targeted to the plasma membrane, and endosomal localization could be achieved via endocytosis of the cell surface-associated protein. A third possibility is that a newly synthesized Pr55gag is selectively inserted either into the plasma membrane or endosomal membranes depending on the host cell type.
In the present study, we analyzed targeting of newly synthesized Pr55gag in HeLa H1 cells by pulse-chase studies and a subcellular fractionation assay that efficiently separated plasma membrane from internal membranes. Our results demonstrated that a newly synthesized Pr55gag was initially targeted to the plasma membrane. Pr55gag accumulated at the plasma membrane when coexpressed with the viral accessory protein Vpu but, in the absence of Vpu, a significant fraction of the cell surface-associated Pr55gag proteins was retargeted to internal membranes by endocytosis. This retargeting partially shifted virus assembly from the cell surface to internal membranes. RNA interference (RNAi)-mediated knockdown of Tsg101, or deletion of the p6 domain from Pr55gag suppressed the endocytic uptake of Pr55gag when the protein was expressed in the absence of Vpu. Taken together, these results suggest that the site of HIV-1 assembly is determined by factors that regulate endocytosis of Pr55gag, and the results identified a previously unknown role for Vpu and Tsg101 as regulators of the endocytic uptake of Pr55gag.
MATERIALS AND METHODS
Cell culture and transfection. HeLa H1 cells (American Type Culture Collection) were grown in Dulbecco’s modified Eagle medium supplemented with 7% fetal bovine serum (EuroClone) and nonessential amino acids (i.e., HeLa growth medium). 293T cells were cultured as previously described (18). Vesicular stomatitis virus (VSV) G-protein-pseudotyped infectious lentivirus vectors carrying modified HIV-1 genomes were produced by cotransfecting pCMVR8.91 and pMD2.VSV.G (generous gifts from Didier Trono [48]), together with a plasmid directing the synthesis of a recombinant HIV-1 genome into 293T cells by a calcium phosphate precipitation technique.
Plasmid constructs. pHXB2D and pNL4-3 were originally obtained from the NIBSC Centralized Facility for AIDS Reagents (the plasmids were donated by Robert Gallo and Mikulas Popovic and by Malcolm Martin, respectively). pHXB2D is a HIV-1 molecular clone with a defective vpu gene and truncated vpr and nef genes (38). Plasmid pHXB2D-Gag is a derivative of pHXB2D that expresses an unprocessed Pr55gag due to an inactivation of the Pol coding region. The plasmid was constructed by replacing the 423-bp ApaI-BclI fragment with a PCR-derived 297-bp ApaI-BclI fragment in which the stop codon of Pr55gag was immediately followed by SmaI and BclI sites. pHXB2D-GagHA and pHXB2D-Gagflag are derivatives of pHXB2D-Gag which express unprocessed Pr55gag proteins that contain hemagglutinin (HA) or FLAG epitopes at their carboxy termini. pNL4-3(Gag), a derivative of pNL4-3 (1), has an inactive Pol-region and expresses an unprocessed Pr55gag. The plasmid was constructed by replacing the 3,779-bp ApaI-SalI fragment with the equivalent ApaI-SalI fragment of pHXB2D-Gag. pNL4-3(Gag/Vpu), a derivative of pNL4-3(Gag), has a defective vpu gene due to a point mutation in the AUG start codon of vpu. The construct was created by replacing the 558-bp SalI-KpnI fragment with the equivalent fragment of pHXB2D. The plasmid pNL4-3(Gagp6/Vpu) encoding a carboxy terminally truncated unprocessed Pr55gag lacking the p6 domain was made from pNL4-3(Gag/Vpu) by replacing the 791-bp SpeI-SmaI fragment with a PCR-created 635-bp SpeI-SmaI fragment in which a stop codon and a SmaI site were placed immediately after the last codon of SP2 peptide. pNL4-3(Env/Nef) is a derivative of pNL4-3 with defective env and nef genes. The first step in construction of pNL4-3(Env/Nef) involved inactivation of the env gene on pNL4-3(Gag) by deleting the 580-bp BglII-BglII fragment from the Env coding region. This created the plasmid pNL4-3(Gag/Env). pNL4-3(Env/Nef) was made by replacing the 2,680-bp SalI-BamHI fragment of pNL4-3-R71.Nef- (35): the plasmid contains an inactivated nef gene) with the equivalent fragment from pNL4-3(Gag/Env). pHXB2D-(Vpu/Env/Nef) contains inactive vpu, env, and nef genes. The first step in the construction of pHXB2D-(Vpu/Env/Nef) was inactivation of the env gene on pHXB2D by deleting a 1,278-bp region between KpnI and BglII sites on env which in addition to an internal deletion also introduced a frameshift mutation on the gene. This created the plasmid pHXB2D-(Vpu/Env). pHXB2D-(Vpu/Env/Nef) was made from pHXB2D-(Vpu/Env) by inactivating the truncated nef gene via converting cohesive ends of XhoI-linearized pHXB2D-(Vpu/Env) into blunt ends by Klenow and subsequently religating the plasmid.
Subcellular fractionations. HeLa H1 cells were seeded on 10-cm plates at a density which gave 70% confluence after o/n growth. Cells were infected the following day with VSV-G-pseudotyped lentivirus vectors carrying modified HIV-1 genomes. At 24 h postinfection the now-confluent cell monolayers were metabolically labeled with 50 μCi of [35S]methionine for 30 min (18) and chased for various times in HeLa growth medium containing a 10-fold excess of cold methionine. After the chase, cell monolayers were washed twice with phosphate-buffered saline containing 0.02% EDTA (PBS-EDTA), and the cells were brought into suspension by a 2-min incubation in PBS-EDTA at 37°C. Subsequently, the suspension cells were coated at 4°C with cationic silica beads essentially as previously described (4, 41), except that the coating buffer used was 20 mM 2-[N-morpholine]ethanesulfonic acid (pH 6.66 to 6.68) and 150 mM NaCl. After coating, cells were homogenized with a tight-fitting Dounce homogenizer in buffer containing 25 mM HEPES (pH 7.4), 150 mM NaCl, 2 mM MgCl2, 20 μg of phenylmethylsulfonyl fluoride, 1 μg of CLAP (chymostatin, leupeptin, aprotinin, and pepstatin A), and 2.5 mg of heparin per ml (the heparin was included to suppress binding of Pr55gag to chromatin released from broken nuclei). The crude cell homogenates were mixed with 100% Nycodenz solution (AXIS-SHIELD PoC AS) to give a final 60% Nycodenz concentration. The sample (2 ml) was placed on a 1-ml cushion of 70% Nycodenz and overlaid with 1.5 ml of 50% Nycodenz and 0.5 ml of 25 mM HEPES (pH 7.4), 150 mM NaCl, and 1 mM EDTA. The gradient was centrifuged in an SW55 Ti rotor at 63,000 x g for 45 min at 4°C. The pellet (i.e., the plasma membrane fraction) was resuspended in hot 0.5% sodium dodecyl sulfate (SDS)-0.5 M NaCl buffer and then boiled for 5 min, and silica beads were removed from the resuspended samples by centrifugation in an Eppendorf centrifuge at full speed for 1 min. The internal membrane fraction at the 50% Nycodenz-buffer interphase was collected and diluted fivefold with 25 mM HEPES (pH 7.4), 150 mM NaCl, and 1 mM EDTA, and membranes were pelleted by centrifugation at 100,000 x g for 1 h at 4°C. The internal membrane pellet was solubilized as described above for the plasma membrane, and the samples were diluted with a 10-fold excess of lysis buffer (1% Triton X-100, 25 mM HEPES [pH 7.4], 150 mM NaCl, 1 mM EDTA, 20 μg of phenylmethylsulfonyl fluoride, and 1 μg of CLAP per ml). Pr55gag proteins were immunoprecipitated from the lysates with mouse anti-HIV-1 CA monoclonal antibody 38:9 (17), the Pr55gag processing intermediates and CA with a mixture of monoclonal antibody 38:9 and mouse monoclonal anti-p24 antibody 32/5.17.76 (Abcam, Ltd.), and the HA-tagged Pr55gag with a rabbit polyclonal anti-HA antibody (Sigma).
For biotinylation of the cell surface proteins, HeLa H1 cells were brought into suspension by treatment with PBS-EDTA and incubated in PBS (with MgCl2 and CaCl2) containing 0.5 mg per ml of EZ-Link Sulfo-NHS-SS-biotin (Pierce) for 30 min at 0°C. The sample was quenched with 50 mM NH4Cl before silica coating. Biotin-tagged proteins in the plasma membrane and the internal membrane fractions were visualized by dot blots using streptavidin-peroxidase polymer (Sigma). Biotin-tagged transferrin proteins (15 μg per ml; Molecular Probes) that had been bound to HeLa H1 cells at 0°C or internalized for 20 min at 37°C in serum-free medium were used as markers for the plasma membrane and recycling endosomes, respectively. In the case of the internalized transferrin, cells were incubated in a pH 4.5 buffer containing 25 mM citric acid, 24 mM trisodiumcitrate, 280 mM sucrose, and 10 μM deferoxamine mesylate for 2 min to remove cell surface-bound transferrin proteins prior to silica coating. Transferrin proteins in the plasma membrane and internal membrane fractions were visualized by Western blotting with streptavidin-peroxidase polymer. The amounts of the endoplasmic reticulum (ER) marker protein calnexin and the late endosome/lysosome marker protein Lamp-1 in the plasma membrane and internal membrane fractions were determined by Western blotting with rabbit polyclonal anti-calnexin (StressGen Biotechnologies Corp.) and goat polyclonal anti-Lamp-1 (C-20; Santa Cruz Biotechnology, Inc.), respectively.
Electron miscroscopy. Cells were fixed at 24 h postinfection with 2% glutaraldehyde in 100 mM HEPES (pH 7.4). The fixed samples were first stained with 1% osmium tetroxide, followed by staining with 0.5% alcoholic uranyl acetate, dehydration in ethanol, and embedding in Agar 100 resin. Sections (60 nm) were cut on an LKB ultramicrotome and imaged in an FEI Tecnai Spirit transmission electron microscope.
RNAi. HeLa H1 cells were seeded on 10-cm plates at a density which gave 50% confluence after overnight growth. siRNAs directed against the Tsg101 (7) or a control siRNA (GGUCUCUCUUUCGACGUGCdTdT; BLAST searches against the human genome sequence did not identify any target for this siRNA) were transfected into HeLa H1 cells by using Oligofectamine (Invitrogen) in accordance with the manufacturer's instructions. Transfections were carried out in HeLa growth medium containing 7% fetal bovine serum, and 40 μl oligofectamine and 20 pmol siRNA per 10-cm plate were used. Mock-transfected cells were treated with Oligofectamine alone. At about 22 h posttransfection the cells were infected with recombinant lentivirus vectors and then analyzed about 24 h postinfection. The intracellular levels of Tsg101 in mock- and siRNA-treated cells were determined from postnuclear supernatants of cell extracts by Western blotting with a monoclonal anti-Tsg101 antibody (C-2; Santa Cruz Biotechnology, Inc.). The monoclonal anti-actin antibody (Chemicon) was used as a control.
RESULTS
Subcellular fractionation method to separate plasma membrane from internal membranes. We have expressed Pr55gag in HeLa H1 cells and analyzed the targeting of the newly synthesized Pr55gag by pulse-chase studies and subcellular fractionations of cell extracts. HeLa H1 cells were chosen since our pilot experiments indicated that Pr55gag can accumulate in these cells either at the plasma membrane or at the internal membranes. Thus, these cells provide a good model system for studying how plasma membrane versus internal localization of the protein is achieved. To determine whether Pr55gag is initially targeted to the plasma membrane or internal membranes requires efficient separation of these membranes. During homogenization, plasma membrane breaks into vesicles and membrane sheets of various density (see, for example, reference 42) that are difficult to separate from, for example, endosomal membranes on conventional sucrose or iodixanol density gradients. Therefore, we chose a fractionation method in which intact cells were coated with cationic silica beads at 4°C prior to homogenization (4, 41), and crude cell extracts were fractionated on a Nycodenz step gradient. The step gradient consisted of a 70% cushion and a 60% sample loading zone overlaid with 50% Nycodenz and buffer solutions. Since the bound silica beads greatly increase density of the plasma membrane, the plasma membrane-derived vesicles and membrane sheets pellet through the 70% Nycodenz cushion during ultracentrifugation (the PM fraction), whereas internal membranes float to the 50% Nycodenz-buffer interphase (I, the internal membrane fraction). Figure 1 shows controls for this fractionation method. In Fig. 1A, externally exposed cell surface-associated proteins were tagged with biotin prior to coating, and biotinylated proteins in the PM and I fractions were visualized by dot blots with horseradish peroxidase-conjugated streptavidin. The majority of biotin-tagged proteins were found in the PM fraction. Similarly, biotin-conjugated transferrin that was bound to cells at 0°C prior to coating was predominantly in the PM fraction (Fig. 1B, Tf surface). In contrast, biotin-conjugated transferrin that had been concentrated into recycling endosomes by internalization at 37°C for 20 min was in the I fraction (Fig. 1B, Tf internalized). The ER protein calnexin, as well as the endosomal/lysosomal marker lysosome-associated membrane protein 1 (Lamp-1), were predominantly in the I fraction (Fig. 1B). Taken together, these results demonstrate that the fractionation method can separate the plasma membrane from internal membranes in HeLa H1 cell extracts.
Newly synthesized Pr55gag is inserted into the plasma membrane. To analyze targeting of newly synthesized Pr55gag, HeLa H1 cells were infected with VSV-G-pseudotyped lentivirus vectors carrying modified HIV-1 genomes. At 24 h postinfection, cells were metabolically labeled with [35S]methionine for 30 min and chased for various times prior to coating and fractionation. Confluent cell monolayers were used in the studies. Amounts of labeled Pr55gag proteins in the PM and I fractions were determined by immunoprecipitation and by SDS-polyacrylamide gel electrophoresis analysis of the immunoprecipitates. Figure 2A shows analysis of newly synthesized, unprocessed Pr55gag expressed from a NL4-3 proviral genome from which the Pol region was deleted [the NL4-3(Gag) genome]. After a 5-min chase, a significant amount of labeled Pr55gag was detected in the PM fraction, but essentially no labeled Pr55gag was present in the I fraction. Similar results were obtained with a 15-min chase (data not shown). Beginning from the 60-min chase point, small amounts of labeled Pr55gag appeared in the I fraction, but the majority of the protein (>80%) was found in the PM fraction at all chase points tested. The NL4-3(Gag) proviral construct efficiently produced VLPs, since large amounts of labeled Pr55gag was found in the culture medium already after 2 h of chase (data not shown). If silica coating was omitted, all labeled Pr55gag was in the I fraction (Fig. 2B), thus confirming that Pr55gag found in the PM fraction in coated cells represented cell surface-associated proteins, and no unspecific protein aggregates. A nonmyristoylated mutant of Pr55gag in which the amino-terminal glycine residue was mutated to alanine failed to associate with either the PM or I fractions (data not shown). Taken together, these results indicated that newly synthesized Pr55gag was inserted into the plasma membrane, and the majority of the proteins remained at the cell surface.
Deletion of the vpu gene induces endocytosis of the plasma membrane-associated Pr55gag. We next tested whether localization of Pr55gag to the plasma membrane was an intrinsic property of the protein, or whether other viral proteins had an effect on intracellular targeting of Pr55gag. Elimination of env or nef genes did not significantly alter the targeting of newly synthesized Pr55gag, i.e., the protein was inserted into the plasma membrane and the majority of labeled Pr55gag remained at the plasma membrane in the absence of Env or Nef expressions (data not shown). In contrast, deletion of the vpu gene produced a dramatic change in the intracellular localization of Pr55gag. As shown in Fig. 3A, after a 15-min chase, or 30-min chase (data not shown), the vast majority of labeled Pr55gag was present in the PM fraction in NL4-3(Gag/Vpu)-infected cells. Beginning from the 60-min chase point, significant amounts of labeled Pr55gag appeared in the I fraction. By 2 h of chase, 70% of labeled Pr55gag was shifted to internal membranes. As previously reported for HeLa cells (22), Vpu was also required for efficient virus production from HeLa H1 cells, since NL4-3(Gag)-infected cells produced 18-fold more VLPs than NL4-3(Gag/Vpu)-infected cells (data not shown). Since newly synthesized cytosolic Pr55gag is rapidly targeted to membranes, within 30 min of synthesis (43), the results presented in Fig. 3A suggest that in the absence of Vpu expression Pr55gag is initially targeted to the plasma membrane but is subsequently redistributed to the internal membranes by endocytosis. As shown in Fig. 3B, a carboxy-terminal HA tag, but not a FLAG tag, suppressed endocytosis of cell surface-associated Pr55gag. After 15-min chase, Pr55gag/HA and Pr55gag/FLAG expressed in the absence of Vpu were primarily at the plasma membrane (data not shown) but after 3 h of chase, 60% of labeled Pr55gag/FLAG had redistributed to internal membranes, whereas >80% of Pr55gag/HA remained at the cell surface even after 4 h of chase.
The effect of Vpu was also tested in the context of Gag and Gag-Pol coexpression. Figure 4A shows an analysis of processed Pr55gag that was expressed from a modified NL4-3 proviral genome lacking env and nef genes but expressing Vpu [the NL4-3(Env/Nef)]. At all of the time points tested, most of the labeled Pr55gag, its processing intermediates, or the p25/p24 CA forms were found in the PM fraction. Only trace amounts of the proteins could be detected in the I fraction. In contrast, if in addition to the env and nef genes the vpu gene was also deleted [the HXB2D-(Vpu/Env/Nef) construct], the intracellular localization of CA forms was changed (Fig. 4B). After 15 min of chase, Pr55gag and two of its early processing intermediates were detected in the PM fraction but not in the I fraction, thus indicating that Pr55gag was first targeted to the plasma membrane. After 2 h of chase, Pr55gag and its early processing intermediates were still found in the PM fraction but, significantly, 60% of the processed CA forms were in the internal membrane fraction. At the 4-h chase point, only the p25 and p24 CA forms were evident, and the majority of these proteins was found in internal membranes. Although CA was the only Gag form detected in the I fraction in the radioactively labeled samples, Western blot analysis of the PM and I fractions indicated that small amounts of Pr55gag and its processing intermediates were present in internal membranes as well (Fig. 4C). Taken together, the results in Fig. 4 demonstrate that deletion of vpu gene induced uptake of plasma membrane-associated core proteins also when Gag and Gag-Pol were coexpressed. These results were confirmed by EM. EM analysis of NL4-3(Env/Nef)-infected cells identified numerous budding profiles at the cell surface (Fig. 5A). Budding profiles were detected at the plasma membrane in HXB2D-(Vpu/Env/Nef)-infected cells as well (Fig. 5C), but these cells also contained endosome-like intracellular compartments full of VLP-sized vesicles (Fig. 5B). Occasionally, budding profiles resembling nascent VLPs could be observed on the limiting membrane of these compartments (Fig. 5D). When 10 randomly selected NL4-3(Env/Nef)- and HXB2D-(Vpu/Env/Nef)-infected cells, i.e., cells that had budding structures at the cell surface, were scored for VLP-containing endosomal structures, these structures were found in all of the inspected HXB2D-(Vpu/Env/Nef)-infected cells. In contrast, only one of the NL4-3(Env/Nef)-infected cells had clearly identifiable endosomal VLPs.
Knockdown of Tsg101 inhibits targeting of Pr55gag to internal membranes. The PTAP motif on the p6 domain of Pr55gag binds Tsg101, and this interaction is necessary for the efficient release of progeny particles (7, 23, 45). We tested the effect of a previously described Tsg101-specific small interfering RNAs (siRNAs) (7) in NL4-3(Gag/Vpu)-infected cells to probe whether Tsg101 is involved in endocytosis of Pr55gag. Western blot analysis of cell lysates demonstrated that Tsg101 siRNAs effectively reduced intracellular levels of Tsg101 (Fig. 6A). After a 30-min pulse and 3 h of chase, 58% of labeled Pr55gag remained at the cell surface in Tsg101 siRNA-transfected cells, whereas 73% of labeled Pr55gag was in internal membranes in mock-transfected cells (Fig. 6B). Cells transfected with control siRNAs of scrambled sequence had 63% of labeled Pr55gag in internal membranes (Fig. 6B). The less-efficient Pr55gag uptake in Tsg101 siRNA-transfected cells could be due to ablation of Pr55gag interaction with Tsg101, or, alternatively, due to unspecific and/or pleiotropic effects of the Tsg101 siRNAs. To distinguish between these possibilities, we tested uptake of a carboxy-terminally truncated Pr55gag lacking the p6 domain. The truncated protein was expressed from the recombinant proviral genome NL4-3(Gagp6/Vpu). After a 30-min pulse and a 15-min chase, the labeled Gagp6 protein was in the PM fraction (data not shown). Deletion of the p6 domain inhibited endocytosis of the cell surface-associated protein, since 75% of the truncated proteins were still at the plasma membrane after 3 h of chase (Fig. 6C). Taken together, these results suggest that endocytosis of Pr55gag is critically dependent on interaction of Tsg101 with the p6 domain of Pr55gag.
DISCUSSION
In the present study, we have analyzed targeting of newly synthesized Pr55gag in HeLa H1 cells by pulse-chase studies and by using a subcellular fractionation method that efficiently separates plasma membrane from internal membranes. Our results indicate that Pr55gag is initially targeted to the plasma membrane, and the protein is delivered to internal membranes by endocytosis. Our study also identified Vpu as a key regulator of intracellular targeting of Pr55gag. In the presence of Vpu, the bulk of Pr55gag remained at the plasma membrane after the initial membrane insertion, whereas in the absence of coexpressed Vpu a significant fraction of the cell surface-associated Pr55gag proteins was retargeted to internal membranes. Similar retargeting was observed regardless of whether Pr55gag was expressed as a nonprocessed precursor protein or whether Pr55gag was expressed together with Gag-Pol. These results are in agreement with previous reports indicating that deletion of the vpu gene increases virus budding into intracellular compartments in HeLa and T cells (13, 22). We observed a similar increase of VLPs in internal membrane compartments in the absence of Vpu when NL4-3(Env/Nef)- and HXB2D-(Vpu/Env/Nef)-infected cells were compared by EM. Furthermore, it is noteworthy that in all of the studies that have identified Pr55gag in internal membranes in transfected or infected, but otherwise nonperturbed HeLa, Cos-1, or 293T cells, the protein was expressed in the absence of Vpu (6, 10, 26, 39). Consistent with our results, Rudner et al. (34) recently identified plasma membrane as the primary target membrane for Pr55gag in HeLa cells. The method used was the fluorescent imaging of cells expressing a tetracysteine-tagged Pr55gag and sequential staining of the cells with two biarsenical compounds that produced different fluorescent signals upon binding to the tetracysteine tag. The dual-labeling procedure enabled the detection of newly synthesized Pr55gag. However, in contrast to our results, the presence or absence of Vpu was reported to have no effect on the association of Pr55gag with the plasma membrane. This discrepancy could perhaps be explained by the carboxy-terminal tetracysteine tag; as exemplified by our HA-tagged Pr55gag, a carboxy-terminal tag can inhibit the endocytosis of cell surface-associated Pr55gag. Since our subcellular fractionation method did not separate internal membrane compartments from each other, the identity of the internal membranes harboring Pr55gag was not established. Fluorescence microscopy analysis of HXB2D-(Vpu/Env/Nef)-infected cells indicated a vesicular staining pattern for intracellular Gag (data not shown), but the intracellular Gag-signal in this particular viral genetic background did not extensively colocalize with any of the tested early or late endosomal markers (CD63, Lamp-1, EEA1, or internalized fluorescein isothiocyanate-conjugated transferrin; data not shown).
It is formally possible that the Gag signal in the internal membrane fraction in our silica coating experiments originated from internalized extracellular VLPs rather than from endocytosed plasma membrane-associated Pr55gag. However, we think this is unlikely for the following reasons. First, very little Gag was detected in the internal membrane fraction in NL4-3(Gag)- and NL4-3(Env/Nef)-infected cells despite the fact that these cells released 18-fold more VLPs than cells expressing the respective Vpu-genomes (data not shown). Second, although rare, virus-like budding structures could be identified on internal membrane compartments, and early processing intermediates of Pr55gag, although not as abundant as CA, were observed in internal membrane fraction by Western blotting. Thus, unless VLPs produced in the absence of Vpu have some feature that promotes their rapid uptake after budding from the cell surface, these observations strongly suggest that Gag localization to internal membranes is the consequence of endocytosis of plasma membrane-associated proteins rather than internalization of extracellular VLPs. The mechanism by which Vpu suppresses endocytosis of Pr55gag remains to be elucidated. Since there are no indications that Vpu directly interacts with Pr55gag, Vpu most likely exerts its effect indirectly, e.g., by inducing modifications on the plasma membrane or by exerting broader modifying effects on the membrane trafficking pathways of the host cell. In order to elucidate how Vpu suppresses endocytosis of Pr55gag, it is essential to understand the molecular basis of endocytosis of Pr55gag when the protein is expressed in the absence of Vpu. Our initial analyses suggest that this uptake requires interactions between the cell surface-associated Pr55gag and Tsg101. siRNAs directed against Tsg101, or deletion of the p6 domain that mediates Pr55gag and Tsg101 interactions (7, 23, 45), inhibited the endocytosis of Pr55gag. siRNAs can cause various off-target effects that are difficult to control for in conventional experimental settings (32, 36), and therefore RNAi-mediated knockdown results have to be interpreted with caution. However, since the endocytosis of Pr55gag was suppressed by deletion of p6 domain, this suggests that the effect of Tsg101 siRNAs was due to ablation of Tsg101 and Pr55gag interactions. Tsg101 was initially identified as being essential for efficient egress of progeny particles from the cell surface assembly sites (7, 23, 45). Tsg101 is a component of the ESCRT-I (endocytic sorting complex required for trafficking) and recruits the so-called class E proteins, which normally function in MVB biogenesis, to the site of HIV-1 assembly and budding (24). The class E proteins also participate in the budding of other retroviruses, filoviruses, and arenaviruses (24). Since HIV-1 assembly is arrested at a late stage (after membrane distortion) when p6-Tsg101 interactions are disrupted (24), it has been assumed that Tsg101 and other class E proteins facilitate the final steps in budding. However, since our results indicate that Tsg101 also has an impact on intracellular localization of Pr55gag, this suggests that the function of Tsg101 in HIV-1 assembly and budding might be more complex than was previously thought.
Several biological activities have been attributed to Vpu (2, 19, 44). One of these is the enhancement of virus release from HIV-1-infected cells. Interestingly, Vpu can also increase virus release from cells infected by other retroviruses (13) and even alphaviruses (11). Since Vpu enhances virus production from most human cells, as well as from human-simian heterokaryons, but not from simian cells, Varthakavi et al. (44) postulated that Vpu counteracts a human cell-specific restriction factor that inhibits HIV-1 particle production. Although our present study by no means excludes the existence of such an assembly/release inhibitor, the data suggest a possible alternative explanation for the Vpu-mediated enhancement of HIV-1 particle release in human cells. Namely, Vpu could affect particle production indirectly via its effects on intracellular targeting of Gag. Since deletion of the vpu gene subjected plasma membrane-associated Gag to endocytosis and partially shifted virus assembly to internal membranes, and if these endosomal viruses were only inefficiently released into the extracellular medium, then Vpu-deficient viruses would be expected to produce less extracellular particles than their wild-type counterparts. Vpu has also been reported to enhance overall membrane association of Gag when Gag was expressed from a T7-driven vaccinia virus expression system (5). Although we did not rigorously test the distribution of Gag between cytosolic and membrane fractions in the present study, we, or others (15), have not observed any dramatic reduction in membrane-associated Gag with Vpu-deficient proviral expression systems.
The biological significance of the plasma membrane versus the endosomal assembly of HIV-1 is not fully understood at present, but emerging information suggests that the site of virus assembly could impact HIV-1 pathogenesis. For example, intracellular assembly may promote viral persistence since intracellular virions in macrophages retain infectivity for an extended period of time and can efficiently transfer the infection to lymphocytes upon contact (37). Furthermore, it has been speculated that the cellular site of assembly might have a direct impact on immune recognition and infectivity of the formed virions (12). Thus, understanding the molecular mechanisms that regulate the choice of viral assembly site is important. Our results suggest that cell surface versus internal assembly of HIV-1 in HeLa H1 cells is determined by factors that regulate endocytic uptake of plasma membrane-associated Pr55gag. Our results provide compelling evidence that Vpu promotes cell surface assembly of HIV-1 by inhibiting endocytosis of plasma membrane-associated Pr55gag and that newly synthesized Pr55gag proteins reach internal membrane compartments in HeLa H1 cells via endocytosis from the plasma membrane. The presence of VLPs in internal compartments in HXB2D-(Vpu/Env/Nef)-infected cells implies that internalized Pr55gag proteins can engage in productive particle assembly and that the internalization process does not merely represent garbage disposal, i.e., the removal of defective Gag forms from the cell surface. Since HeLa cells recapitulate the main features of HIV-1 assembly in T cells, our results most likely apply to T cells as well. Unfortunately, we could not directly test the targeting of Pr55gag in T cells, since, e.g., Jurkat T cells tended to disrupt during silica coating, and this lead to binding of silica beads to internal membranes as well (data not shown). Whether endocytosis from the plasma membrane is also the mechanism by which Pr55gag is targeted to the MVBs in macrophages is unclear at present, since intracellular targeting of Pr55gag in HeLa H1 cells and macrophages differs at least in two aspects. Namely, Gag is apparently targeted to MVBs in macrophages even when the protein is coexpressed with Vpu (25, 31, 33) and targeting to MVBs in macrophages appears to be independent of Pr55gag-Tsg101 interactions (27). Consequently, either the uptake mechanism is different in macrophages or, alternatively, the initial membrane insertion does not occur at the plasma membrane in these cells. We are currently testing whether the silica coating method can be used to determine the targeting phenotype of newly synthesized Pr55gag in macrophages.
ACKNOWLEDGMENTS
We thank Didier Trono and Kalle Saksela for providing plasmids and Birgitta Lindqvist and Cornelia Muncke for expert technical assistance. Plasmids pHXB2D and pNL4-3 were obtained from the NIBSC Centralized Facility for AIDS Reagents.
This study was supported by grants from Academy of Finland and Swedish Research Council to M.S.
REFERENCES
Adachi, A., H. E. Gendelman, S. Koenig, S. Folks, R. Willey, A. Rabson, and M. A. Martin. 1986. Production of acquired immunodeficiency syndrome-associated retrovirus in human and nonhuman cells transfected with an infectious molecular retrovirus clone. J. Virol. 59:284-291.
Bour, S., and K. Strebel. 2003. The HIV-1 Vpu protein: a multifunctional enhancer of viral particle release. Microbes Infect. 5:1029-1039.
Bryant, M., and L. Ratner. 1990. Myristoylation-dependent replication and assembly of human immunodeficiency virus 1. Proc. Natl. Acad. Sci. USA 87:523-527.
Chaney, L. K., and B. S. Jacobson. 1983. Coating cells with colloidal silica for high yield isolation of plasma membrane sheets and identification of transmembrane proteins. J. Biol. Chem. 258:10062-10072.
Deora, A., P. Spearman, and L. Ratner. 2000. The N-terminal matrix domain of HIV-1 Gag is sufficient but not necessary for viral protein U-mediated enhancement of particle release through a membrane-targeting mechanism. Virology 269:305-312.
Dong, X., H. Li, A. Derdowski, L. Ding, A. Burnett, X. Chen, T. R. Peters, T. S. Dermody, E. Woodruff, J.-J. Wang, and P. Spearman. 2005. AP-3 directs the intracellular trafficking of HIV-1 Gag and plays a key role in particle assembly. Cell 120:663-674.
Garrus, J. E., U. K. von Schwedler, O. W. Pornillos, S. G. Morham, K. H. Zavitz, H. E. Wang, D. A. Wettstein, K. M. Stray, M. Cte, R. L. Rich, D. G. Myszka, and W. I. Sundquist. 2001. Tsg101 and the vacuolar protein sorting pathway are essential for HIV-1 budding. Cell 107:55-65.
Gelderblom, H. R., E. H. S. Hausman, M. Ozel, G. Pauli, and M. A. Koch. 1987. Fine structure of human immunodeficiency virus (HIV) and immunolocalization of structural proteins. Virol. 156:171-176.
Gheysen, D., E. Jacobs, F. d. Foresta, C. Thiriart, M. Francotte, D. Thines, and M. D. Wilde. 1989. Assembly and release of HIV-1 precursor Pr55gag virus-like particles from recombinant baculovirus-infected cells. Cell 59:103-112.
Goff, A., L. S. Ehrlich, S. N. Cohen, and C. A. Carter. 2003. Tsg101 control of human immunodeficiency virus type 1 Gag trafficking and release. J. Virol. 77:9173-9182.
González, M. E., and L. Carrasco. 2001. Human immunodeficiency virus type 1 Vpu protein affects Sindbis virus glycoprotein processing and enhances membrane permeabilization. Virol. 279:201-209.
Gould, S. J., A. M. Booth, and J. E. K. Hildreth. 2003. The Trojan exosome hypothesis. Proc. Natl. Acad. Sci. USA 100:10592-10597.
Gttlinger, H. G., T. Dorfman, E. A. Cohen, and W. A. Haseltine. 1993. Vpu protein of human immunodeficiency virus type 1 enhances the release of capsids produced by gag gene constructs of widely divergent retroviruses. Proc. Natl. Acad. Sci. USA 90:7381-7385.
Gttlinger, H. G., J. G. Sodroski, and W. A. Haseltine. 1989. Role of capsid precursor processing and myristoylation in morphogenesis and infectivity of human immunodeficiency virus type 1. Proc. Natl. Acad. Sci. USA 86:3195-3199.
Handley, M. A., S. Paddock, A. Dall, and A. T. Panganiban. 2001. Association of Vpu-binding protein with microtubules and Vpu-dependent redistribution of HIV-1 Gag protein. Virol. 291:198-207.
Hermida-Matsumoto, L., and M. D. Resh. 2000. Localization of human immunodeficiency virus type 1 Gag and env at the plasma membrane by confocal imaging. J. Virol. 74:8670-8679.
Hinkula, J., L. Rosen, V.-A. Sundqvist, T. Stigbrand, and B. Wahren. 1990. Epitope mapping of the HIV-1 Gag region with monoclonal antibodies. Mol. Immunol. 27:395-403.
Holm, K., K. Weclewicz, R. Hewson, and M. Suomalainen. 2003. HIV-1 assembly and lipid rafts: Pr55gag complexes associate with membrane-domains that are largely resistant to Brij 98, but sensitive to Triton X-100. J. Virol. 77:4805-4817.
Hsu, K., J. Seharaseyon, P. Dong, S. Bour, and E. Marbán. 2004. Mutual functional destruction of HIV-1 Vpu and host TASK-1 channel. Mol. Cell 14:259-267.
Johnson, D. R., R. S. Bhatnagar, L. J. Knoll, and J. I. Gordon. 1994. Genetic and biochemical studies of protein N-myristylation. Annu. Rev. Biochem. 63:869-914.
Karacostas, V., K. Nagashima, M. A. Gonda, and B. Moss. 1989. Human immunodeficiency virus-like particles produced by a vaccinia virus expression vector. Proc. Natl. Acad. Sci. USA 86:8964-8967.
Klimkait, T., K. Strebel, M. D. Hoggan, M. A. Martin, and J. M. Orenstein. 1990. The human immunodeficiency virus type 1-specific protein Vpu is required for efficient virus maturation and release. J. Virol. 64:621-629.
Martin-Serrano, J., T. Zang, and P. D. Bieniasz. 2001. HIV-1 and Ebola virus encode small peptide motifs that recruit Tsg101 to sites of particle assembly to facilitate egress. Nat. Med. 7:1313-1319.
Morita, E., and W. I. Sundquist. 2004. Retrovirus budding. Annu. Rev. Cell Dev. Biol. 20:395-425.
Nguyen, D. H., and J. E. K. Hildreth. 2000. Evidence for budding of human immunodeficiency virus type 1 selectively from glycolipid-enriched membrane lipid rafts. J. Virol. 74:3264-3272.
Nydegger, S., M. Foti, A. Derdowski, P. Spearman, and M. Thali. 2003. HIV-1 egress is gated through late endosomal membranes. Traffic 4:902-910.
Ono, A., and E. O. Freed. 2004. Cell-type-dependent targeting of human immunodeficiency virus type 1 assembly to the plasma membrane and the multivesicular body. J. Virol. 78:1552-1563.
Orenstein, J. M., M. S. Meltzer, T. Phipps, and H. E. Gendelman. 1988. Cytoplasmic assembly and accumulation of human immunodeficiency virus types 1 and 2 in recombinant human colony-stimulating-factor-1-treated human monocytes: an ultrastructural study. J. Virol. 62:2578-2586.
Pal, R., R. C. Gallo, and M. G. Sarngadharan. 1988. Processing of the structural proteins of human immunodeficiency virus type 1 in the presence of monensin and cerulenin. Proc. Natl. Acad. Sci. USA 85:9283-9286.
Pal, R., S. Mumbauer, G. M. Hoke, A. Takatsuki, and M. G. Sarngadharan. 1991. Brefeldin A inhibits the processing and secretion of envelope glycoproteins of human immunodeficiency virus type 1. AIDS Res. Hum. Retrovir. 7:707-712.
Pelchen-Matthews, A., B. Kramer, and M. Marsh. 2003. Infectious HIV-1 assembles in late endosomes in primary macrophages. J. Cell Biol. 162:443-455.
Persengiev, S. P., X. Zhu, and M. R. Green. 2004. Nonspecific, concentration-dependent stimulation and repression of mammalian gene expression by small interfering RNAs (siRNAs). RNA 10:12-18.
Raposo, G., M. S. Moore, D. Innes, R. Leijendekker, A. Leigh-Brown, P. Benaroch, and H. J. Geuze. 2002. Human macrophages accumulate HIV-1 particles in MHC II compartments. Traffic 3:718-729.
Rudner, L., S. Nydegger, L. V. Coren, K. Nagashima, M. Thali, and D. E. Ott. 2005. Dynamic fluorescent imaging of human immunodeficiency virus type 1 Gag in live cells by biarsenical labeling. J. Virol. 79:4055-4065.
Saksela, K., G. Cheng, and D. Baltimore. 1995. Proline-rich (PxxP) motifs in HIV-1 Nef bind SH3 domains of a subset of Src kinases and are required for the enhanced growth of Nef+ viruses but not for down-regulation of CD4. EMBO J. 14:484-491.
Scacheri, P. C., O. Rozenblatt-Rosen, N. J. Caplen, T. G. Wolfsberg, L. Umayam, J. C. Lee, C. M. Hughes, A. Shanmugam, M. Meyerson, and F. S. Collins. 2004. Short interfering RNAs can induce unexpected and divergent changes in the levels of untargeted proteins in mammalian cells. Proc. Natl. Acad. Sci. USA 101:1892-1897.
Sharova, N., S. Swingler, M. Sharkey, and M. Stevenson. 2005. Macrophages archive HIV-1 virions for dissemination in trans. EMBO J. 24:2481-2489.
Shaw, G. M., B. H. Hahn, S. K. Arya, J. E. Groopman, R. C. Gallo, and F. Wong-Staal. 1984. Molecular characterization of human T-cell leukemia (lymphotropic) virus type III in the acquired immune deficiency syndrome. Science 226:1165-1171.
Sherer, N. M., M. J. Lehmann, L. F. Jimenez-Soto, A. Ingmundson, S. M. Horner, G. Cicchetti, P. G. Allen, M. Pypaert, J. M. Cunningman, and W. Mothes. 2003. Visualization of retroviral replication in living cells reveals budding into multivesicular bodies. Traffic 4:785-801.
Spearman, P., J.-J. Wang, N. Vander Heyden, and L. Ratner. 1994. Identification of human immunodeficiency virus type 1 Gag protein domains essential to membrane binding and particle assembly. J. Virol. 68:3232-3242.
Spector, D. L., R. D. Goldman, and L. A. Leinwand. 1998. Plasma membrane isolation using the cationic colloidal silica isolation technique, p. 35.1-35.14. In Cells: a laboratory manual, vol. 1. Cold Spring Harbor Laboratory Press, New York, N.Y.
Suomalainen, M., K. Hultenby, and H. Garoff. 1996. Targeting of Moloney murine leukemia virus Gag precursor to the site of virus budding. J. Cell Biol. 135:1841-1852.
Tritel, M., and M. D. Resh. 2000. Kinetic analysis of human immunodeficiency virus type 1 assembly reveals the presence of sequential intermediates. J. Virol. 74:5845-5855.
Varthakavi, V., R. M. Smith, S. P. Bour, K. Strebel, and P. Spearman. 2003. Viral protein U counteracts a human host cell restriction that inhibits HIV-1 particle production. Proc. Natl. Acad. Sci. USA 100:15154-15159.
VerPlank, L., F. Bouamr, T. J. LaGrassa, B. Agresta, A. Kikonyogo, J. Leis, and C. A. Carter. 2001. Tsg101, a homologue of ubiquitin-conjugating (E2) enzymes, binds the L domain in HIV type 1 Pr55Gag. Proc. Natl. Acad. Sci. USA 98:7724-7729.
von Schwedler, U. K., M. Stuchell, B. Muller, D. W. Ward, H.-Y. Chung, E. Morita, H. E. Wang, T. Davis, G.-P. He, D. M. Cimbora, A. Scott, H.-G. Krusslich, J. Kaplan, S. G. Morham, and W. I. Sundquist. 2003. The protein network of HIV budding. Cell 114:701-713.
Zhou, W., L. J. Parent, J. W. Wills, and M. Resh. 1994. Identification of a membrane-binding domain within the amino-terminal region of human immunodeficiency virus type 1 Gag protein which interacts with acidic phospholipids. J. Virol. 68:2556-2569.
Zufferey, R., D. Nagy, R. J. Mandel, L. Naldini, and D. Trono. 1997. Multiply attenuated lentiviral vector achieves efficient gene delivery in vivo. Nat. Biotechnol. 15:871-875.(Kirsi Harila, Ian Prior, )