|Home | About | Journals | Submit | Contact Us | Français|
Human immunodeficiency virus type 1 (HIV-1) Gag is the primary structural protein of the virus and is sufficient for particle formation. We utilized the recently developed biarsenical-labeling method to dynamically observe HIV-1 Gag within live cells by adding a tetracysteine tag (C-C-P-G-C-C) to the C terminus of Gag in both Pr55Gag expression and full-length proviral constructs. Membrane-permeable biarsenical compounds FlAsH and ReAsH covalently bond to this tetracysteine sequence and specifically fluoresce, effectively labeling Gag in the cell. Biarsenical labeling readily and specifically detected a tetracysteine-tagged HIV-1 Gag protein (Gag-TC) in HeLa, Mel JuSo, and Jurkat T cells by deconvolution fluorescence microscopy. Gag-TC was localized primarily at or near the plasma membrane in all cell types examined. Fluorescent two-color analysis of Gag-TC in HeLa cells revealed that nascent Gag was present mostly at the plasma membrane in distinct regions. Intracellular imaging of a Gag-TC myristylation mutant observed a diffuse signal throughout the cell, consistent with the role of myristylation in Gag localization to the plasma membrane. In contrast, mutation of the L-domain core sequence did not appreciably alter the localization of Gag, suggesting that the PTAP L domain functions at the site of budding rather than as a targeting signal. Taken together, our results show that Gag concentrates in specific plasma membrane areas rapidly after translation and demonstrate the utility of biarsenical labeling for visualizing the dynamic localization of Gag.
While extensive progress has been made in the identification of cellular pathways and proteins involved in human immunodeficiency virus type 1 (HIV-1) assembly and budding, the location of Gag translation and the early trafficking steps in HIV-1 assembly are not known. Since several cellular components of the endosomal sorting complex required for transport (ESCRT) system that direct the budding of proteins into late endosomes and/or multivesicular bodies (MVBs) are important for virus release, it has been proposed that HIV-1 uses parts of or the whole late endosome/MVB sorting system as cellular partners in this process (13, 15, 19, 37, 41, 46, 65, 72, 73). In macrophages, it is clear that HIV-1 buds into these late endosomal compartments as a requisite step for trafficking virion particles to the cell surface in MVB-like structures (44, 47, 52, 59). However, the location of HIV-1 Gag trafficking and assembly in other cell types remains less clear. The first assembly stage visible by electron microscopy in nonmacrophage cells is the formation of bar-like complexes at the inner leaflet of the plasma membrane (14, 43). Fluorescence microscopy has also found that HIV-1 Gag exhibits punctate staining near the plasma membrane in epithelial and lymphoid cells (9, 25, 46, 47, 60). However, the late endosome/MVB pathway has been proposed to be involved in the trafficking of HIV-1 Gag from the interior of the cell to the plasma membrane (19, 53, 65). Despite the clear involvement of the MVB sorting and vesiculation machinery in HIV budding, most aspects of the localization of Gag synthesis, its trafficking, and its targeting to distinct membranes remain unexplored.
To examine the localization of HIV-1 Gag inside cells and examine the trafficking patterns of newly synthesized Gag, we chose to apply a recently developed biarsenical-binding technique to fluorescently label HIV-1 Gag and study virus-like particle (VLP) and virus assembly. This approach uses a relatively small tetracysteine tag that is genetically engineered into the protein of interest (1, 21). This tag, minimally two pairs of cysteines held in a hairpin configuration, i.e., C-C-P-G-C-C, specifically reacts with membrane-permeable biarsenical compounds that selectively fluoresce when covalently bound to the cysteine pairs. Since this genetic tag is relatively small and simple, it can be placed into target proteins with minimal disruption to the protein (1, 12, 21). Furthermore, this cysteine-based structure can form even under sodium dodecyl sulfate (SDS)-denaturing conditions, indicating that the dye-binding sequence does not require extensive structure for activity (30). Therefore, the tetracysteine tag should be competent to bind biarsenical dyes almost immediately after translation, thereby fluorescing much more rapidly than larger proteins used as fluorescent tags. An additional advantage is that two colors of biarsenical reagents are available, FlAsH and ReAsH, which fluoresce either green or red, respectively. This allows for examining where nascent Gag accumulates by labeling the existing target protein in the cell with one color and then labeling newly synthesized protein with the other (12, 29).
To study Gag trafficking inside the cell, we introduced tetracysteine tags at the C terminus of Pr55Gag in Gag expression constructs and an HIV-1 proviral molecular clone and examined Gag localization in different cell types. This approach revealed that Gag primarily associates with the plasma membrane both at steady state and just after synthesis. These results establish biarsenical labeling as an important method to dynamically observe Gag assembly in live cells.
HeLa, 293T, and Mel JuSo cells were cultured in Dulbecco's modified Eagle medium, and Jurkat T cells were cultured in RPMI 1640. Both media were supplemented with 10% (vol/vol) fetal bovine serum, 100 U of penicillin/ml, and 100 μg of streptomycin/ml. All media and components were obtained from Invitrogen (Carlsbad, Calif.). Transfections of HeLa and Mel JuSo cells for imaging were performed using Lipfectamine 2000 (Invitrogen) according to the manufacturer's protocol. Jurkat cells were electroporated by washing the cells twice with 20 mM HEPES in RPMI 1640 and resuspending them in 300 μl of HEPES-RPMI. Electroporation was carried out by using Bio-Rad (Hercules, Calif.) Gene Pulser II for a single 250-V, 975-μF pulse.
The human-codon-optimized Gag gene was PCR amplified from GagGFP (gift of Paul Spearman, Vanderbilt University, Nashville, Tenn.) (8) and cloned into the cytomegalovirus promoter/enhancer expression vector pCDNA3 (Invitrogen) as a Hind III-BamH I fragment to produce the Gagopt construct. Biarsenical-binding Gag constructs, Gag-TC, and Gag-LTC were produced by inserting tetracysteine tags with a minimal sequence, GSMPCCPGCCGC, or an extended linker sequence, GSDSSGSMHVVDSSGSMPCCPGCCGS, respectively, at the C terminus of Gag by oligonucleotide insertion into the BamH I site in Gagopt and the XhoI site of pCDNA3. A tetracysteine-tagged proviral clone HIV-TC, was constructed by adding a third tag sequence, GSESSGSMPCCPGCCGS, to the C terminus of Gag within the NL4-3 full-length molecular clone by PCR overlap extension (26). Myr− (Gag codon 2 change from Gly [GGA] to Ala [GCA]) and PTAP− (PTAPP to AIVSA) mutations were introduced into Gag-LTC and HIV-TC by PCR overlap extension.
Adherent cells were cultured in chambered-coverglass 35-mm-diameter dishes (Nalgene, Ashland, Mass.). Suspension cells were immobilized onto chambered coverglass (Nunc International, Rochester, N.Y.) treated with Cell Tak (BD Bioscience) before biarsenical labeling and immunostaining. For each flask or coverglass, biarsenical-labeling solution was freshly prepared by preincubating 1 μl of either 2 mM FlAsH-EDT2 or 2.5 mM ReAsH-EDT2 (Invitrogen) with 1 μl of 25 mM 1,2-ethanedithiol (EDT) (Fluka, St. Louis, Mo.) and 1 μl of dimethyl sulfoxide (Sigma, St. Louis, Mo.) for 15 min at room temperature. Then, 200 μl of Hanks' balanced salt solution (HBSS; Invitrogen) supplemented with 1 g of d (+)-glucose (Sigma)/liter was added to the biarsenical EDT-dimethyl sulfoxide mix and was incubated for 10 min. Cells were washed extensively with either HBSS-glucose for cells expressing Gag-LTC and its corresponding mutants or Dulbecco's modified Eagle medium with 2% (vol/vol) fetal bovine serum (serum-reduced medium) for those expressing Gag-TC. After washing, 1.8 ml of the respective medium was added to each dish or coverglass and 200 μl of the biarsenical mix was applied to each dish (or coverglass) and incubated for 1 h at 37°C. After labeling, cells were rinsed extensively with medium, followed by three separate 10-min incubations with 300 μM EDT in medium. The EDT solution was then replaced with HBSS-glucose, and cells were imaged live or fixed for staining. For costaining with antibodies or antiserum, cells were fixed with 2% (wt/vol) paraformaldehyde (Electron Microscopy Sciences, Hatfield, Pa.) in phosphate-buffered saline (PBS; Invitrogen) for 10 min. After washing with PBS, cells were permeabilized with 0.2% (vol/vol) Triton X-100 for 10 min, washed again with PBS, blocked with 1% bovine serum albumin (BSA) in PBS for 10 min, and incubated with primary antibody for 1 h at 37°C. After extensive washing with PBS, cells were blocked again with 1% (wt/vol) BSA in PBS for 10 min and then incubated with the appropriate secondary antibody for 30 min. After three additional PBS washes, cells were overlaid with 1% (wt/vol) BSA in PBS and examined by fluorescence microscopy. The following primary antibodies and antiserum were used: rabbit p6Gag antiserum (AIDS Vaccine Program, National Cancer Institute—Frederick, Frederick, Md.), mouse anti-CD63 clone H5C6 and anti-Lamp2 clone H4B4 (Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, Iowa), anti-EEA1 and anti-calnexin (BD Transduction Labs), and anti-human golgin-97 (Molecular Probes, Eugene, Oreg.). The following secondary antibodies were used: anti-mouse Texas Red-conjugated antibodies (Jackson ImmunoResearch, West Grove, Pa.) and anti-mouse or anti-rabbit antibodies conjugated to either Alexa Fluor 594 or Alexa Fluor 647 (Molecular Probes).
Dishes and slides were examined on a Delta Vision Workstation model DV base 3 with a Nikon Eclipse TE200 epifluorescence microscope fitted with an automated stage (Applied Precision Inc. Issaquah, Wash.), and images were captured in z-series with a charge-coupled digital camera (CH350E). Out of focus, light was digitally removed using SoftWoRx deconvolution software (Applied Precision Inc.).
HeLa cells at 60% confluency were transfected with 1.2 μg of Gag-TC DNA. After 24 h, cells were washed with reduced-serum medium and labeled with ReAsH in serum-reduced medium for 1 h as described above. Cells were then rinsed twice with serum-reduced medium and washed three times for 10 min with 150 μM EDT in serum-reduced medium with no rinses in between. After the EDT washes, cells were either immediately labeled with FlAsH in serum-reduced medium (for the 1-h analysis) or replaced in complete medium for 1.5 or 3 h at 37°C (for the 2.5- and 4-h analyses, respectively). For the latter analyses, at the end of the incubation, cells were again washed twice with serum-reduced medium and then labeled by FlAsH in serum-reduced medium for 1 h as described above. After FlAsH labeling, cells were rinsed twice with serum-reduced medium and washed three times with 250 μM EDT for 10 min in serum-reduced medium with no rinses in between treatments. Cells were then washed twice with PBS and fixed with 3.7% paraformaldehyde (in PBS) for 10 min and either imaged directly or stained with p6Gag antiserum.
VLP production from 293T cells was carried out by transfection in T75 flasks (Corning, Corning, N.Y.) by either using Transit-293T (Mirus Corporation, Madison, Wis.) according to the manufacturer's instructions or CaPO4 transfection (20). Particles from transfected-cell supernatants were purified by pelleting through a 20% sucrose pad at greater than 120,000 × g for 1 h. Immunoblotting was carried out using the Bio-Rad Immun-Star horseradish peroxidase substrate kit. Blots were developed by exposure to Lumifilm (Roche Applied Science, Indianapolis, Ind.) or by image capture with a VeraDoc camera imaging system (Bio-Rad). Antiserum against p24CA was produced by the AIDS Vaccine Program, National Cancer Institute—Frederick. Metabolic labeling of Transit-293T-transfected cells was carried out as follows: 48 h posttransfection, cells were released from the flask by agitation and transferred into 15-ml conical centrifuge tubes, pelleted at 900 × g, washed once in PBS, and preincubated in 1 ml of labeling medium (RPMI 1640 without Met or Cys [Sigma] supplemented with 5% [vol/vol] dialyzed fetal bovine serum [Invitrogen]) for 1 h. This Met-Cys starvation medium was then removed from the cells and replaced with RPMI 1640 with serum and 150 μCi of [35S]Met and 60 μCi of [35S]Cys (ProMix labeling mix, Amersham Biosciences, Piscataway, N.J.) per tube. After a 6-h labeling period, cells were pelleted and lysed in RIPA buffer: 50 mM Tris HCl (pH 8.0), 150 mM NaCl, 1% (vol/vol) Triton X-100, 0.1% (wt/vol) sodium deoxycholate, and 0.1% (wt/vol) SDS. Nuclei were removed from the cell lysates by centrifugation at 2,000 × g to produce cytoplasmic extracts. VLPs from culture supernatants were pelleted by sucrose density centrifugation as described above and lysed in RIPA buffer. HIV-1 Gag proteins were immunoprecipitated with a cocktail of goat p24CA with rabbit p7NC and p6Gag antisera and EZview Sepharose-protein G (Sigma). Beads were washed twice with RIPA buffer, and then precipitates were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE). Dried gels were analyzed by a Personal Molecular Imager FX phosphorimager (Bio-Rad) and quantitated by using Quantity One software (Bio-Rad).
Thin-section micrographs of positive-stained 293T-transfected cells were obtained as previously described (17). Cells were prepared by transfection with Transit-293T and then harvested and centrifuged at 900 × g for 5 min at room temperature before being placed in the 1.25% (vol/vol) glutaraldehyde fixative solution in PBS.
To visualize Gag within cells by biarsenical labeling, we produced Gag expression constructs that contained a tetracysteine (CCPGCC) biarsenical-dye-binding tag at the C terminus of Gagopt, a human-codon-optimized Gag which is Rev independent (28). The tetracysteine core sequence forms a putative hairpin and has been shown to strongly and specifically react with biarsenical reagents, either FlAsH (green) or ReAsH (red), to covalently label the protein of interest (1). Gag was fused to either a 12-amino-acid tetracysteine tag, Gag-TC, or a 26-amino-acid tag with an extended linker sequence, Gag-LTC. For this latter construct, unstructured spacer sequences were added to provide a flexible linker between Gag and the dye-binding tag to promote hairpin formation. The localization of Gag with the tetracysteine core within either short (Gag-TC) or extended (Gag-LTC) flanking sequences was tested in HeLa cells. While both tagged Gag proteins demonstrated specific FlAsH labeling (Fig. (Fig.1A),1A), Gag-TC was localized primarily to discreet areas at or near the plasma membrane with little, if any, punctate staining in the interior of the cell, likely from Gag associated with intracellular membranes. The Gag-LTC-expressing cells showed considerably more signal at these putative intracellular membranes than that observed for Gag-TC, although some Gag-LTC still was found at the plasma membrane (Fig. (Fig.1A).1A). Costaining FlAsH-labeled cells with p6Gag antiserum and Alexafluor 594-labeled secondary antibody (red) found that both reagents detected the same pattern (Fig. (Fig.1A).1A). The merged images showed that the two signals overlapped in both Gag-TC- and Gag-LTC-expressing cells, demonstrating the specificity of the FlAsH labeling.
In addition to Gag signal, we also observed a dim background of FlAsH signal mostly in the nucleus and at the perinuclear sites of cells in many, but not all, experiments, which was readily distinguished from the authentic Gag signal by p6Gag antiserum staining. Examples of this intermittent background problem are apparent in nontransfected cells in Fig. Fig.1A1A (upper left for Gag-TC and lower right for Gag-LTC) and D (upper left) and in the cell expressing Gag-LTC in Fig. Fig.1A.1A. Similar biarsenical background fluorescence has also been encountered by others (12) and, though manageable, is a significant difficulty in using the biarsenical-labeling technique.
To determine if the addition of these tags alters the localization of Gag, we stained permeabilized HeLa cells that expressed Gag, Gag-TC, or Gag-LTC with p6Gag antiserum and Alexafluor 594 anti-rabbit secondary antibody. The results showed that both Gag and Gag-TC localize mostly to the plasma membrane region while Gag-LTC is present at internal structures as well as the plasma membrane (Fig. (Fig.1B).1B). Therefore, Gag-TC recapitulated Gag localization better than Gag-LTC. In separate experiments, incubation of cells with the FlAsH reagent did not have any discernible effect on the localization of these Gag proteins or morphological effects on the treated cells (data not shown). These data indicate that biarsenical labeling of tetracysteine tags can faithfully detect Gag inside cells. However, our results with the Gag-LTC construct do show that some tag sequences (likely the flanking linker sequences) can alter the localization of Gag when placed at its C terminus.
Since Gag seems to localize differently in various cell types (46, 47, 52, 65), the distribution of Gag-TC was examined in the CD4+-Jurkat-T-cell line, a biologically relevant host for HIV-1, and the Mel JuSo melanoma cell line, which has an efficient endosomal trafficking system (38, 71). Gag-TC expressed in electroporated Jurkat cells produced a highly punctate signal near or at the plasma membrane that colocalized with the p6Gag staining (Fig. (Fig.1C).1C). Interestingly, Gag-TC was positioned mostly on only one side of the cell, suggesting that budding was restricted to limited regions of the plasma membrane, consistent with the polarized budding previously observed in HIV-1-infected lymphocytes by conventional staining methods and electron microscopy (9, 51, 61).
Gag-TC in Mel JuSo cells displayed relatively more internal signal associated with vesicular-like structures within the cytoplasm than either HeLa or Jurkat cells (Fig. (Fig.1C).1C). Still, the majority of the FlAsH signal was localized to the plasma membrane and overlapped that from p6Gag antiserum staining. While there was slightly more Gag-TC signal at these interior structures in Mel JuSo cells, this was still less than that seen in Gag-LTC in either HeLa (Fig. (Fig.1B)1B) or Mel JuSo (data not shown). It is possible that the increase in interior signal for Gag-TC is due to the more pronounced endosomal trafficking of Mel JuSo cells.
While Gag is sufficient for particle formation, other elements in HIV-1 might influence Gag localization and assembly. HIV-1 Env can direct HIV-1 budding to the basolateral surface in certain polarized cell systems (36) and Vpu, which increases virus release (3), can influence Gag localization (22). Also, RNA signals in the genomic RNA might target assembly by RNA trafficking (68). Thus, Gag in the context of assembling HIV-1 might be localized to different regions than Gag alone. Also, our Gagopt construct with a codon-optimized gene produced Gag at much higher levels than a proviral construct (data not shown), a feature that might also alter the localization of Gag. To examine these possibilities, we fused a tetracysteine tag to the C terminus of Gag in the pNL4-3 full-length proviral clone to produce HIV-TC. This tag interrupts the overlapping pol reading frame by introducing a nonsense mutation, prematurely terminating pol translation. FlAsH labeling of HeLa cells expressing HIV-TC showed a localization pattern similar to that of Gag-TC (Fig. (Fig.1D).1D). Gag signal from this construct was almost exclusively found at the plasma membrane surface with only a slight amount in the intracellular compartments. Elimination of Env, Vpu, or Vpr expression by site-directed mutagenesis had little effect on Gag localization versus HIV-TC (data not shown). Thus, expression of other viral proteins or expression of Gag from a viral RNA and the lower gag expression levels did not appreciably alter the localization of Gag.
While the C terminus of HIV-1 appears to be able to tolerate the addition of several types of sequences without grossly altering Gag function (46, 57, 60), we examined the ability of Gag-TC and Gag-LTC to produce VLPs by immunoblot analysis of particle preparations and VLP release by metabolic radiolabeling and immunoprecipitation. 293T human kidney fibroblast cells were transfected with Gag, Gag-TC, or Gag-LTC expression constructs, and particles from equal volumes of transfection supernatant were isolated by sucrose density centrifugation and examined by immunoblot analysis with p24CA antiserum (Fig. (Fig.2).2). All three constructs produced VLPs. As expected, the results of the immunoblot analyses showed that the Gag bands produced from Gag-TC and Gag-LTC migrated slightly higher in the gel than the Gag band, consistent with the addition of the two different-length tags (Fig. (Fig.2).2). To measure normalized particle release, the ratio of VLPs produced (i.e., particles isolated from supernatants by density centrifugation) to the total amount of Gag (VLPs and intracellular Gag) after 6 h was determined by 35S radiolabeling of 293T cells transfected with the various Gag expression constructs and immunoprecipitation of lysates with a cocktail of p24CA, p7NC, and p6Gag antisera. Radiolabeled Gag proteins were applied to SDS-PAGE gels and quantitated by phosphorimaging (Table (Table1).1). The results of duplicate experiments revealed that the release efficiencies of both tagged versions of Gag (Gag-TC and Gag-LTC) were somewhat better than that for Gag and similar to the release efficiency of HIV virions produced by the full-length NL4-3-based protease mutant PRR57G (50). It is interesting to note that even though considerable amounts of Gag-LTC were found in intracellular structures, these cells still efficiently released VLPs. Taken together, these data show that VLP production is not impaired by the addition of the tetracysteine tag.
To examine the morphology and particle budding of Gag, Gag-TC, and Gag-LTC, 293T cells transfected with the respective expression constructs were analyzed by thin-section transmission electron microscopy. Expression of any of the three constructs produced similar images (Fig. (Fig.3):3): numerous budding structures were present at the cell membrane, and released particles exhibited the doughnut-shaped morphology expected for VLPs. However, there were a few assembling and budding forms visible inside the cell apparently at intracellular membranes, suggesting that not all of Gag assembles at the plasma membrane. While these could be due to artifacts of surface invaginations or ruffles in the thin sections, these structures contain some staining material as expected for internal structures. Despite the increased localization of Gag to intracellular compartments in the Gag-LTC-expressing cells (Fig. (Fig.1A),1A), we did not find a corresponding increase in the amount of intracellular assembling particles versus Gag or Gag-TC. Perhaps the intracellular Gag-LTC does not begin assembly into highly ordered VLP forms until it reaches the plasma membrane, as previously observed for HIV-1 Gag (43). Alternatively, this internal Gag-LTC could be nonfunctional and destined for degradation as discussed below. Together, our electron microscopy, fluorescence microscopy, and VLP formation studies demonstrate that the tetracysteine tag in Gag-TC has little effect on Gag function.
To detect where Gag accumulates immediately after synthesis, we employed a two-color labeling procedure that takes advantage of the availability of the FlAsH and ReAsH compounds that covalently bind to the tetracysteine motif (12, 29). By labeling the existing (steady-state) pool of Gag with ReAsH (red), the Gag produced after this initial labeling can then be specifically labeled with FlAsH (green). Presently, this imaging technique is feasible using only the biarsenical system.
Initial labeling of HeLa cells expressing Gag-TC with ReAsH produced typical results: steady-state Gag was observed mostly near or on the plasma membrane with little, if any, cytoplasmic signal (Fig. (Fig.4A,4A, ReAsH). Immediately labeling these cells for 1 h with FlAsH detected signal both at the plasma membrane and inside the cell (Fig. (Fig.4A,4A, FlAsH). The amount of FlAsH fluorescence at the plasma membrane was considerably less than the ReAsH labeling, likely due to the small amount of Gag expressed over the 1-h time period (Fig. (Fig.4A,4A, FlAsH). In addition to the plasma-membrane-associated signal, there were localized, brightly labeled FlAsH-positive loci within the cell. To confirm that the signals detected by ReAsH and FlAsH were Gag, we stained these cells with p6Gag antiserum and Alexa-674 labeled secondary antibody (colored blue) (Fig. (Fig.4A,4A, anti-p6Gag). The results showed a strong overlap between the plasma-membrane-associated Gag detected by the biarsenical dyes and p6Gag antiserum staining (Fig. (Fig.4A,4A, ReAsH/anti-p6Gag, FlAsH/anti-p6Gag, and all merge). However, little, if any, of the bright intracellular FlAsH signal overlapped with the antiserum staining (Fig. (Fig.4A,4A, FlAsH/p6Gag merge). Thus, these internal signals are likely background staining and not produced from authentic Gag. Alternatively, these could be from Gag molecules that somehow have their epitopes for this polyclonal antiserum masked, although this seems unlikely, as we saw no evidence for this in our steady-state experiments. Given the hour interval between the ReAsH and FlAsH labeling, the average age of the Gag observed by FlAsH labeling in these experiments is 30 min. Therefore, these data show that Gag is present at the plasma membrane within approximately 30 min postsynthesis.
Another set of experiments examined the localization of nascent Gag at 2.5 h after the ReAsH labeling (1.5 h of medium-only incubation followed by 1 h of FlAsH medium). These results showed a stronger FlAsH labeling at the plasma membrane as expected, since more Gag is produced over the 2.5-h post-ReAsH labeling (Fig. (Fig.4B,4B, FlAsH). Both the ReAsH and FlAsH signals overlapped that of a p6Gag antiserum costain (Fig. (Fig.4B,4B, anti-p6Gag and corresponding merged images). The localization of Gag after 4 h showed essentially the same pattern (Fig. (Fig.4C),4C), although more Gag was present at the surface due to the increased incubation time between the two labelings.
At all three time points, the green pattern displayed by the nascent Gag was quite similar to the red-colored steady-state Gag labeling (Fig. (Fig.4,4, ReAsH/FlAsH merge); thus, both species accumulated at the same areas on the plasma membrane. This is consistent with Gag budding from specific sites on the cell, as previously observed in lymphocytes (9, 36, 51), rather than randomly from the plasma membrane. While Gag might be present in the interior of the cell, no convincing localized labeling or costaining was observed other than that at the membrane. Thus, newly formed Gag predominantly accumulates rapidly at the plasma membrane.
We applied our biarsenical-imaging technique to analyze mutants that disrupt two important aspects of Gag assembly, Gag myristylation and L-domain function. Since either of the mutations might significantly alter the intracellular localization of Gag, we chose to introduce this mutation into Gag-LTC since, unlike Gag-TC, it localizes both to intracellular structures and to the plasma membrane (Fig. (Fig.1A).1A). In this way, we might see any effect on both internal and plasma membrane targeting. Costaining of FlAsH-labeled Gag-LTC with various markers showed that much of the intracellular Gag-LTC signal overlapped with CD63- (Fig. (Fig.5A)5A) and Lamp-2- (data not shown) positive structures, consistent with Gag-LTC being concentrated in late endosomes and early lysosomes. Gag-LTC did not appreciably overlap with other diagnostic cellular markers for endoplasmic reticulum, Golgi apparatus, or early endosomes (data not shown). Thus, Gag-LTC should allow us to see even subtle changes in the association of Gag with intracellular compartments (e.g., MVB) as well as changes at the plasma membrane.
The cotranslational addition of myristate to the N terminus of Gag is critical for its binding to the plasma membrane (4, 24, 25, 42, 49, 67). To block the myristylation of Gag (4), we altered glycine 2 to alanine to disrupt the myristylation signal in Gag-LTC, producing Gag-LTCMyr−. HeLa cells transfected with the Gag-LTCMyr− construct failed to release significant amounts of particles (Fig. (Fig.2;2; Table Table1)1) as expected (4). Consistent with this result, FlAsH-labeled Gag-LTCMyr− was detected throughout the cell with no noticeable localization at any specific organelles or structures and displayed a conspicuous absence of staining at the plasma membrane (Fig. (Fig.5B).5B). Similar results were obtained with an HIV-TC myristylation mutation (data not shown). We examined the colocalization of this Gag-LTC mutant with our panel of markers and failed to find any specific association of Gag-LTCMyr− with either the secretory or endosomal/lysosomal pathways (Fig. (Fig.5B5B and data not shown).
The HIV-1 L domain is centered around the PTAP sequence of p6Gag and is required for efficient budding from the plasma membrane (18, 27). Recent findings have shown that Tsg101 binds to this PTAP sequence and mediates interactions between Gag and components of the ESCRT I/III pathway which, in turn, promote virion budding (6, 11, 13, 15, 16, 40, 41, 47, 55, 72, 73). Interestingly, Tsg101 is also a primary protein involved in sorting proteins into late endosomes and MVBs through the ESCRT I/III pathway (2, 58), presenting the possibility that the L domain might direct HIV-1 Gag into this internal system as a means of translocating Gag through the cytoplasm to the plasma membrane. To examine the importance of the L domain on Gag localization, we altered the wild-type PTAPP sequence in Gag-LTC to AIVSA, producing Gag-LTCPTAP−. VLP production of this mutant Gag was lower than that of wild type (Fig. (Fig.2),2), and virus release in 293T cells was decreased fourfold (Table (Table1),1), consistent with results from other groups (7, 64). Despite this defect in budding, the FlAsH labeling of Gag-LTCPTAP− in HeLa cells produced the same localization pattern as the Gag-LTC protein (Fig. (Fig.5).5). Also, costaining these labeled cells with either anti-CD63 or anti-Lamp2 (Fig. (Fig.5C5C and data not shown) produced a pattern similar to Gag-LTC. Similar results were obtained for Gag-LTC Mel JuSo cells (data not shown). Thus, the absence of an L domain and its associated Tsg101 binding had no apparent effect on the localization of Gag-LTC either at the plasma membrane or in late-endosomal-like compartments.
We have applied tetracysteine tagging and biarsenical labeling to examine the location of Gag in live cells. Using this approach, we found that steady-state Gag, expressed either alone or within the context of a pol− virus, was present mostly at the plasma membrane with little protein found associated with intracellular compartments. The FlAsH-binding tag had no observed effect on localization or VLP production of Gag: both Gag and Gag-TC were distributed in a similar pattern in our model HeLa cells as observed by FlAsH labeling and antiserum staining. Moreover, electron microscopy in 293T cells revealed virus assembly taking place at the plasma membrane with assembly structures only rarely found in the cytoplasm, consistent with the findings from our fluorescence imaging. Thus, Gag-TC faithfully recapitulates the properties of Gag and reveals that Gag localizes mostly to the plasma membrane at steady state.
Using a fluorescent two-color technique, we found that newly synthesized Gag also accumulates primarily at the plasma membrane within 30 min of synthesis. Thus, it is likely that the majority of Gag is primarily synthesized at or very near the inner leaflet of the plasma membrane. This corresponds well with a previous biochemical report that assembly-competent Gag associates with membranes within 10 min of synthesis (69). Both of these results, the rapid membrane binding and our plasma membrane localization of Gag, are consistent with our previous proposal that Gag translation, RNA packaging, and assembly are spatial and temporally linked at or near the plasma membrane (54).
Two-color analysis also showed that both existing and nascent Gag proteins are mostly found in the same limited regions of the plasma membrane (Fig. (Fig.4,4, all merge), supporting the concept that HIV-1 egress occurs through defined regions of the plasma membrane, possibly through late endosomal membranes (19, 33, 44, 65) and/or cholesterol-rich regions such as lipid rafts (34, 35, 45, 48). This may also be due to the presence of the appropriate endosomal sorting machinery, e.g., ESCRT complexes, at these sites (15, 39, 40, 55, 72).
One point to consider in interpreting any Gag localization study is that not all Gag proteins produce particles: a sizable fraction of synthesized Gag (25 to 80%) fails to form into particles and is degraded (43, 63, 69). One fate of defective Gag is degradation by the proteasome pathway (63). Thus, it is unlikely that all of the Gag molecules observed by any fluorescence imaging technique are functional assembling molecules. Despite this caveat, the predominant localization of Gag to the plasma membrane observed by our fluorescence and electron microscopy makes it likely that some of the Gag proteins we observe by fluorescence were in the process of assembly and budding.
Rous sarcoma virus Gag can pass through the nucleus and might select its genomic RNA for packaging during this journey (62). Since the matrix region of HIV-1 Gag has a nuclear localization and export signal (10), it is possible that HIV-1 Gag could act similarly. However, we did not observe any specific nuclear staining in any of our experiments. Therefore, any nuclear import or export would involve only a small percentage of HIV-1 Gag that is beyond our limit of detection.
Our FlAsH-labeled imaging of Gag expressed from a full-length HIV-1 pol− molecular clone found that Gag assembling into virions was distributed similar to Gag-TC and Gag. Thus, the expression of other viral proteins that are involved in assembly and have been previously implicated in Gag localization, primarily Env (36) and Vpu (22), the presence of cis-acting RNA sequences (68), or levels of Gag expression did not appreciably alter the Gag localization pattern in HeLa cells.
Unlike the other tagged constructs, the Gag-LTC construct localized more to internal structures than at the plasma membrane. Staining for CD63 and Lamp2 markers demonstrated that these Gag-containing structures were late endosomes or MVBs. This alternate localization pattern did not alter the release of Gag-LTC versus Gag-TC or cause any observable increase in internal particle assembly by electron microscopic analysis. Perhaps these late-endosomal-associated Gag molecules are degraded as discussed above. While the residues flanking the tetracysteine motif (underlined) (GSDSSGSMHVVDSSGSMPCCPGCCGS) in Gag-LTC do not have any obvious intracellular targeting sequences, they appear to be responsible for its increased intracellular accumulation when placed at the C terminus of Gag. This finding suggests that caution should be used when producing Gag fusion proteins for imaging; it is important to validate the localization of Gag or any protein that contains even minimal additional sequences.
We applied biarsenical labeling to study the localization of two different assembly mutants. Blocking Gag myristylation caused Gag-LTC to lose plasma membrane and late endosomal targeting and exhibit a generalized cytoplasmic localization. These data essentially agree with those previously obtained by immunofluorescence of HIV-1 Gag myristylation mutants in African green monkey (66), Cos-7 monkey (25), and HeLa cells (49), as well as the imaging of Gag-GFP in the presence of a myristylation inhibitor (25). Therefore, whatever the role for myristylation in Gag localization, it is essential for plasma membrane and late-endosomal targeting, likely due to an early event in Gag localization.
The mutation of the PTAP sequence in Gag-LTC did not apparently alter its localization compared to the wild type: both plasma membrane and interior vesicular sites contained Gag with little difference in CD63 overlap. This is consistent with recent findings by Ono and Freed, who found little difference in localization between wild-type Gag and a p6Gag deletion mutant (47). Similar results have also been obtained for an Ebola VP40 L-domain mutant (32). Together, these results suggest that the defect in L-domain-mediated budding in HeLa cells occurs mostly at the plasma membrane rather than due to a deficiency in Gag trafficking. Thus, Tsg101 and other L-domain-interacting proteins more likely act in the budding process at the plasma membrane immediately during release, as previously proposed (2, 5, 6, 11, 13, 15, 16, 40, 41, 47, 72, 73), rather than acting in a sorting process into the late endosome/MVB pathway (19, 53, 65, 73).
While fusing proteins of interest with fluorescent proteins (FPs) from various Cnidarians is the workhorse for imaging proteins (70), the addition of a short tetracysteine tag and biarsenical labeling is emerging as an alternate approach to observing proteins in living cells. An important advantage to this method is that the tetracysteine hairpin can bind biarsenical dyes rapidly after synthesis (30), so newly synthesized Gag can be visualized. In contrast, FPs need to fold and autooxidize to form a functional chromophore after synthesis, processes with combined experimental half-life values ranging from 27 min to 4 h (23, 31, 56, 70). Given the rapid kinetics of nascent Gag for membrane binding (starting at 10 min) and budding (starting at 1 h) (64, 69), the tetracysteine tag allowed us to visualize the earlier stages of Gag assembly.
Despite these advantages, we have observed intermittent background problems, mostly internal staining that remains a significant problem. However, with the proper controls, it is manageable. Thus, while the biarsenical labeling is clearly a valuable tool, it remains more technically demanding than FP techniques. Current research aimed at optimizing the sequences flanking the tetracysteine motif promises to bring even greater sensitivity and lower backgrounds. Additionally, while the primary motif itself did not alter the properties of Gag, in Gag-LTC, some of the sequences that flank the tetracysteine tag did affect the localization of Gag. Therefore, tagged and untagged proteins need to be compared in some fashion to rule out this potential problem.
One of the important challenges for retrovirology is to understand the spatial organization of the viral proteins within the cell and the cellular components that interact with the various viral proteins during assembly and budding. We are currently introducing biarsenical tags into various proteins in replication-competent HIV-1 to study their localization in live primary cells. The approach that we have developed here should assist this effort and promises to allow us to study HIV-1 Gag, Pol, and Env translation and trafficking in both model and primary cell systems. In turn, this should assist our understanding of HIV-1 assembly and biology.
We thank Paul Spearman for Gagopt-GFP; Stephen Adams for reagents; Guido Gaietta for essential technical assistance; and Eric Freed, Karine Goussett, Akira Ono, and Rob Gorelick for critical reading of the manuscript. The anti-CD63 and anti-Lamp2 developed by J. T. August and J. E. Hildreth were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa Department of Biological Sciences, Iowa City, Iowa.
This project was funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under contract no. NO1-CO-12400 and by RO1 grant AI 47727 to M.T.