Requirement for an Intact T-Cell Actin and Tubulin Cytoskeleton for Efficient Assembly and Spread of Human Immunodeficiency Virus Type 1

Abstract

Human immunodeficiency virus type 1 (HIV-1) infection of CD4⁺ T cells leads to the production of new virions that assemble at the plasma membrane. Gag and Env accumulate in the context of lipid rafts at the inner and outer leaflets of the plasma membrane, respectively, forming polarized domains from which HIV-1 buds. HIV-1 budding can result in either release of cell-free virions or direct cell-cell spread via a virological synapse (VS). The recruitment of Gag and Env to these plasma membrane caps in T cells is poorly understood but may require elements of the T-cell secretory apparatus coordinated by the cytoskeleton. Using fixed-cell immunofluorescence labeling and confocal microscopy, we observed a high percentage of HIV-1-infected T cells with polarized Env and Gag in capped, lipid raft-like assembly domains. Treatment of infected T cells with inhibitors of actin or tubulin remodeling disrupted Gag and Env compartmentalization within the polarized raft-like domains. Depolymerization of the actin cytoskeleton reduced Gag release and viral infectivity, and actin and tubulin inhibitors reduced Env incorporation into virions. Live- and fixed-cell confocal imaging and assay of de novo DNA synthesis by real-time PCR allowed quantification of HIV-1 cell-cell transfer. Inhibition of actin and tubulin remodeling in infected cells interfered with cell-cell spread across a VS and reduced new viral DNA synthesis. Based on these data, we propose that HIV-1 requires both actin and tubulin components of the T-cell cytoskeleton to direct its assembly and budding and to elaborate a functional VS.

Many intracellular pathogens, including parasites, bacteria, and viruses, invade, traffic within, and exit their target cells in a cytoskeleton-dependent manner (13, 16, 17, 20). Human immunodeficiency virus type 1 (HIV-1) is no exception. During HIV-1 entry into permissive cells, successful interaction between HIV-1 and its cellular receptor, CD4, or one of the chemokine receptors, CXCR4 or CCR5, is at least partially actin dependent (26). Efficient reverse transcription of incoming viral RNA appears to rely on an interaction between the HIV-1 core and actin (8). Uncoated preintegration complexes are shuttled to the nucleus along the microtubule network (35), and nuclear import leads to integration of the provirus (18). Newly synthesized viral RNA and structural proteins are transported to the site of HIV-1 assembly, where they assemble into budding virions at the appropriate target membrane. The mechanism of transport of HIV-1 Gag and Env to the site of virion assembly is a directed process that is currently under intense scrutiny. Finally, filamentous actin (f-actin) associates directly with HIV-1 Gag during HIV-1 budding and is found within virions (24, 33, 53, 65).

The location of viral budding is cell type dependent: HIV-1 buds predominantly into structures resembling multivesicular bodies in macrophages and is presumed to be released from these cells via an exocytic mechanism (31, 34, 41, 42, 48). In contrast, in T cells, HIV-1 is thought to assemble at and bud from glycosphingolipid-rich plasma membrane domains that share features with lipid rafts (40, 45, 46). The site of virus assembly and release appears to depend on targeting signals within Gag and Env. Env trafficking to the plasma membrane is regulated by interactions between dileucine and tyrosine motifs in Env and the clathrin adaptor proteins AP1 and AP2 (5, 10, 66), and Env localization into lipid rafts requires palmitoylation and other sorting signals in gp41 (4, 9, 56). Gag trafficking to and assembly at the plasma membrane are dependent on several factors. Gag p55 and p17 contain myristoyl groups and a cluster of basic amino acids that target them to membranes (52). Gag p6 interacts with components of the endosomal sorting complex required for transport (ESCRT) pathway to mediate transport to the site of virion assembly and subsequent virion budding (38). Additionally, the local membrane concentration of phosphatidylinositol 4,5-bisphosphate is important for Gag targeting (44), as is the association between Gag and the adaptor protein AP3 (11). The apparent overlap between HIV-1 morphogenesis and transport of luminal cargo within the cytoplasm suggests that HIV-1 enlists components of the cellular sorting pathway (6, 19, 42, 43, 49) and implies the utilization of the cytoskeleton to drive virion trafficking to the assembly site. To date, work on HIV-1 assembly and budding has been carried out mostly with transformed cell lines, usually of fibroblastic or epithelial origin, and less is known regarding the molecular mechanisms driving HIV-1 assembly and budding in T cells.

HIV-1 can spread by release of cell-free infectious virions and by direct cell-cell spread (27, 50). Recently, two retroviruses, namely, human T-cell leukemia virus type 1 (HTLV-1) (25) and HIV-1 (28), have been shown to mediate direct cell-cell transmission between T cells via the formation of a virological synapse (VS). The VS is comprised of a stable adhesive junction that forms at the interface between virally infected (effector) cells and uninfected (target) cells, across which virus is transmitted by directed transfer (2, 30, 51). The evolving VS is characterized by rapid, actin-dependent recruitment of CD4, CXCR4, and lymphocyte function-associated antigen 1 (LFA-1) from the target cell (28) and of Env and Gag in polarized lipid raft-like patches on the effector cell (28, 29) to the cell-cell interface. These events are followed by budding of virus into the synaptic cleft and transfer of HIV-1 Gag into the postsynaptic target cell (28). HIV-1-triggered VS formation in T cells is dependent on Env-receptor interactions, since specific inhibitors of these processes prevent VS assembly (28).

We and others have demonstrated that Env and Gag colocalize in HIV-1-infected T cells within plasma membrane caps with lipid raft-like properties, forming assembly platforms (23, 29, 32, 46). Moreover, lipid raft membrane domains are stabilized by actin scaffolds linked to the inner leaflet of the membrane via adaptor proteins, and these interactions are important for recruitment of receptors to the immunological synapse (IS) (36, 54). Additionally, microtubules interact with cortical actin via adaptors and may influence the spatial organization of actin-lipid raft complexes during cell adhesion, migration (14, 57), and IS formation (55, 60). Since there are numerous similarities between the VS and the IS (2, 30, 51), we wished to determine whether actin and tubulin functional integrity within HIV-1-infected T cells is required for conservation of viral assembly and budding platforms and for cell-cell spread across a VS.

To address the role of the actin-tubulin cytoskeleton in HIV exit and spread, we analyzed Gag and Env localization in infected T cells that were left untreated or treated with inhibitors of actin and tubulin remodeling. We show that polarized viral assembly platforms enriched in Gag, Env, and GM1 were disrupted by interference with either the actin or tubulin cytoskeleton. Coordinate with this, inhibitors of remodeling of either the actin or tubulin cytoskeleton reduced cell-cell transfer of HIV-1 across the VS, and the release of infectious cell-free HIV-1 was inhibited if the actin network was dissolved with latrunculin. These results demonstrate that the cytoskeleton participates actively in HIV-1 assembly in and exit from T cells and that interference with these processes inhibits viral propagation.

MATERIALS AND METHODS

Cells and tissue culture.

The CD4⁺ CXCR4⁺ Jurkat CE6.1 T-cell line (American Type Culture Collection) and the A3.01 T-cell line (Center for AIDS Reagents [CFAR], United Kingdom) were maintained in suspension cell growth medium (RPMI 1640 [Gibco BRL] supplemented with streptomycin [100 μg/ml], penicillin [100 U/ml], and 10% fetal calf serum [FCS]). To prepare cells infected with the T-cell-line-adapted HIV-1 strain LAI (termed effector cells or Jurkat_LAI cells), 5 × 10⁶ Jurkat cells were infected at a multiplicity of infection of 0.01 and cultured for 7 days. Cells were phenotyped for surface Env expression by using pooled human anti-HIV immunoglobulin (HIV Ig; NIH AIDS Research and Reference Reagent Program, Rockville, MD) at 50 μg/ml, detected by anti-human immunoglobulin G (IgG)-phycoerythrin (Jackson Immunoresearch), and for CD4 expression by using monoclonal antibody (MAb) L120 (22) from CFAR. Cells were used on days 7 to 14 postinfection, when Env expression was readily detectable and CD4 expression was weak or undetectable. For conjugate experiments, peripheral blood mononuclear cells were separated from fresh blood of a healthy HIV-1-seronegative donor by using a Ficoll-Hypaque gradient and were negatively enriched for CD4⁺ T cells (termed target cells hereafter) by magnetic cell sorting according to the manufacturer's instructions (Miltenyi Biotec), and this routinely gave >90% pure CD4⁺ T cells. Cells were diluted in RPMI 1640-1% FCS (wash buffer [WB]) and used immediately.

Immunofluorescence and LSCM.

Jurkat_LAI cells (5 × 10⁵) were washed in WB and incubated on poly-l-lysine (Sigma)-treated coverslips at 37°C for up to 60 min. Cells were then fixed in 4% formaldehyde in PBS-1% bovine serum albumin (PBS-1% BSA) for 15 min at 4°C, followed by quenching in 50 mM ammonium chloride for 10 min at room temperature. Alternatively, Jurkat_LAI cells were mixed with an equal number of primary CD4⁺ target T cells, and conjugates were allowed to evolve by incubation on poly-l-lysine-treated coverslips for 60 min at 37°C. Conjugate evolution was arrested by fixation with cold 4% formaldehyde. For intracellular staining of Gag, the cells were permeabilized in 0.1% Triton X-100-5% FCS for 20 min at room temperature and washed extensively in PBS-1% BSA. For staining of tubulin, cells were fixed in 4% paraformaldehyde for 20 min at 37°C, quenched, and washed in PBS-1% BSA. The following antibodies were used for immunostaining: Env was stained with the gp41-specific MAb 50-69 (CFAR), and CD4 was stained with the MAb L120 (CFAR). The Env MAb 50-69 and CD4 MAb L120 were included during conjugate formation since they had previously been shown to be noninhibitory in this system (28). Rabbit antisera against HIV Gag p17 and p24 were obtained from CFAR. Tubulin was stained with rat anti-α-tubulin MAb 1864 (Chemicon), and actin was stained with phalloidin-tetramethyl rhodamine isocyanate (Sigma). GM1-rich lipid rafts were labeled after fixation with 10 μg/ml biotinylated cholera toxin B subunit (Sigma) at 4°C for 30 min in PBS-1% BSA with 0.01% NaN₃. Primary antibodies were detected with fluorescein isothiocyanate (FITC)-conjugated goat anti-rat (Molecular Probes), streptavidin-FITC (Jackson Immunoresearch), or FITC-, tetramethyl rhodamine isocyanate-, or Cy5-conjugated donkey anti-mouse, -human, or -rabbit secondary antibodies that were tested for the absence of interspecies reactivity (Jackson Immunoresearch). Coverslips were mounted with ProLong antifade mounting solution (Molecular Probes), and confocal analysis was performed using a Bio-Rad Radiance 2000 MP laser scanning confocal microscope (LSCM). Image processing and three-dimensional (3D) reconstruction of sequential z series were performed using MetaMorph, version 6.1, and Adobe Photoshop, version 7.

Quantification of conjugates and conjugates forming VS.

Jurkat_LAI cells, alone or in conjugates with target cells, were prepared and analyzed as described above, and multiple random sections of low-power fields were acquired. The total number of effector cells was counted, and the percentage of effectors within conjugates was quantified. Conjugates were defined as closely apposed pairs of cells containing at least one CD4⁺ and one Env⁺ or Gag⁺ cell. Each effector cell was analyzed for cocapping of Env and Gag or polarization to the cell-cell interface of CD4, Env, and Gag. Statistical analysis was performed using an unpaired, two-tailed Student t test with the Bonferroni correction for multiple comparisons, and statistical significance was assumed when P was <0.05.

Inhibition of microtubule and actin remodeling.

Actin remodeling was blocked by treating effector cells with 1 μM cytochalasin D (Sigma), jasplakinolide, or latrunculin A (Molecular Probes) in WB at 37°C for 60 min prior to fixation and single-cell analysis or prior to conjugate formation for VS analysis. To inhibit microtubule remodeling, Jurkat_LAI effector cells were treated with either 1 μM nocodazole or 1 μM colchicine (Sigma) in WB for 30 min at 37°C prior to single-cell analysis or conjugate formation. All inhibitors, with the exception of nocodazole, were removed by washing the effector cells twice in WB before mixing them with target cells for VS analysis. Aliquots of treated and untreated cells were taken, and the numbers of viable cells were calculated by trypan blue exclusion. The metabolic activities of treated and untreated cells were assayed using a colorimetric CellTiter96 Aqueous one-solution cell proliferation assay (Promega) according to the manufacturer's instructions. The assay measures the bioreduction of 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt (MTS) tetrazolium by NADPH or NADP by determining the absorbance at 490 nm.

Live-cell imaging.

Effector cells (5 × 10⁵) were washed in WB and stained for 40 min at 37°C with the gp41-specific MAb 50-69 premixed with an equimolar concentration of Qdot 655 goat F(ab′)₂ anti-human IgG (Quantum Dot Corporation). Cells were washed twice with PBS-1% FCS and added to poly-d-lysine-treated glass-bottomed culture dishes (MatTek Corporation). Low-magnification scanning of the coverslip was used to select the field for analysis. Cells were left untreated over the time period, or an actin or microtubule remodeling inhibitor was added directly to the cell mixture on the culture dish, and images were acquired immediately. The final concentration of all inhibitors was the same as that described for fixed-cell analysis (above). Live-cell movies were collected on a Zeiss Pascal Axiovert 200 M inverted LSCM, and single x-y images were acquired sequentially at a rate of 1 per min over 1 to 2 h.

Flow cytometry of Env surface expression.

Jurkat_LAI cells were either left untreated or treated with cytoskeletal inhibitors for 3 h as described above. The cells were then washed in cold fluorescence-activated cell sorter wash buffer (PBS with 1% FCS and 0.01% azide) and surface stained for HIV-1 Env with 10 μg/ml of the human anti-gp120 antibody 2G12 (Polymun Scientific) for 1 h on ice. The cells were then washed and stained with phycoerythrin-conjugated anti-human secondary antibody for 30 min on ice and fixed, and Env surface expression was quantified by flow cytometry. The average mean fluorescence intensity for two independent experiments was calculated.

SDS-PAGE and Western blotting.

Jurkat_LAI cells (2 × 10⁶) at 7 days postinfection were washed, resuspended in WB, and either left untreated or treated with 1 μM of nocodazole, colchicine, cytochalasin D, or latrunculin for 3 h at 37°C. Treated and untreated cells were pelleted by centrifugation, the supernatants were harvested, and cells were washed twice in cold PBS. Cell pellets and virus-containing supernatants were lysed (lysis buffer contained 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, and Complete protease inhibitor cocktail [Roche]) on ice for 10 min, and soluble protein was collected following centrifugation at 15,000 × g for 10 min at 4°C. Lysates were stored at −70°C, and the protein concentrations were determined using a BCA protein assay reagent kit (Pierce). Twenty micrograms of cell lysate or 10 μl of viral lysate was loaded into 10% Tris-glycine gels, and proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Proteins were transferred onto nitrocellulose, blocked in PBS, 0.1% Tween 20, and 5% skim milk (Marvel), and probed with rabbit antiserum raised against HIV-1 Gag (CFAR) followed by donkey anti-rabbit-horseradish peroxidase (HRP) conjugate (Jackson ImmunoResearch). Chemiluminescence detection was done with SuperSignal West Pico substrate (Pierce). Blots were then stripped, washed, and reprobed with rabbit anti-actin (Sigma) and goat anti-rabbit-HRP (Dako) to confirm equal loading. Densitometer analysis was performed on nonsaturated blots, using ImageJ software. The pixel intensities of each band corresponding to Gag p55 and p24 were quantified individually, and these values were combined to determine the total amount of Gag present in cell lysates and viral lysates. The ratio of total cell-free to total cell-associated Gag was then calculated for untreated and nocodazole-, colchicine-, cytochalasin-, and latrunculin-treated Jurkat_LAI lysates. The mean ratios for three independent virus release assays are shown, with standard errors.

For Western blotting of HIV-1 Env, Jurkat_LAI cells were either left untreated or treated with cytoskeletal inhibitors for 3 h at 37°C as described above. Viral supernatants were harvested and clarified, and virions were concentrated using a VivaSpin column (Sartorius) with a 300,000-Da molecular mass cutoff.

Viral lysates were prepared, and proteins were separated by SDS-PAGE, transferred to nitrocellulose, and probed with rabbit antiserum against HIV-1 gp120 (CFAR) followed by donkey anti-rabbit-HRP (Jackson ImmunoResearch), with chemiluminescence detection. Densitometer analysis of bands corresponding to gp120 was performed using ImageJ software, and the pixel intensity values from two independent virus release assays are shown.

Infectivity assay.

Jurkat_LAI cells were either left untreated or treated with nocodazole, colchicine, cytochalasin, or latrunculin as described above. After 3 h, the cells were washed extensively to remove all traces of inhibitors, and the cells were resuspended in RPMI with 10% FCS and incubated at 37°C. Virus-containing supernatants were harvested at either 1, 3 or 6 h postwashout and used to infect CD4⁺ CXCR4⁺ CCR5⁺ HeLa reporter cells containing the lacZ gene under the control of the HIV-1 long terminal repeat. HeLa cells were lysed at 3 days postinfection, and the supernatants were assayed for β-galactosidase activity (Promega) to measure relative levels of HIV-1 infection.

Quantitative real-time PCR.

PCR primers and the TaqMan probe for HIV-1 DNA quantification were selected using Primer Express 1.0 software and checked by a BLAST search of GenBank. The forward primer pol1 (5′-GTGCTGGAATCAGGAAAGTACTA-3′), the reverse primer pol2 (5′-ATCACTAGCCATTGCTCTCCAATT-3′), and the internal HIV-1 TaqMan probe (6-carboxyfluorescein-5′-TGTGATATTTCTCATGTTCATCTTGGGCCTTATCT-3′-6-carboxytetramethyl-rhodamine) were used to amplify and quantify a region of low variability in the HIV-1 B clade (IIIB) pol gene. Alb1 (5′-GCTGTCATCTCTTGTGGGCTGT-3′), Alb2 (5′-AAACTCATGGGAGCTGCTGGTT-3′), and the human albumin TaqMan probe (VIC-5′-CCTGTCATGCCCACACAAATCTCTCC-3′-6-carboxytetramethylrhodamine) were used to amplify and quantify albumin DNA in parallel in all samples. Albumin was measured as an endogenous reference in each sample, and pol DNA values were normalized against albumin DNA values. Effector cells (1 × 10⁶) and 1 × 10⁶ target cells were mixed, with or without microtubule-depolymerizing agents, in 200 μl growth medium. Conjugate formation was arrested after 0, 1, 3, and 6 h by lysis, and total genomic DNA was extracted immediately using a DNeasy tissue kit (QIAGEN). For real-time DNA quantification, each sample was diluted in a final volume of 25 μl Universal master mix (PE Applied Biosystems) containing 2.5 μl of template DNA, 300 nM of each primer, and 100 nM probe. Taq polymerase was activated for 10 min at 95°C, followed by 40 cycles of a two-step PCR procedure of 15 s at 95°C and 60 s at 60°C. Amplification and data acquisition were carried out using an Applied Biosystems 7500 fast real-time PCR system. Standard curves were prepared with serial dilutions of genomic DNA isolated from 2 × 10⁶ ACH-2 cells (containing one copy of integrated provirus). These were used to calibrate threshold cycles for each 96-well plate run. All test samples were run in triplicate, and all standard dilutions were run in duplicate. Statistical analysis was performed using an unpaired Student t test, with the Bonferroni correction for multiple comparisons.

RESULTS

Actin- and tubulin-depolymerizing agents reduce Gag and Env polarization.

We initially investigated the role of actin and tubulin cytoskeleton function in the polarization of HIV-1 Env and Gag to capped assembly patches on the plasma membranes of infected Jurkat cells. Figure 1 shows the triple labeling of Jurkat_LAI cells for Gag, Env, and GM1 after they were left untreated or pretreated for 1 h with an inhibitor of actin (latrunculin) or tubulin (nocodazole or colchicine) remodeling. In untreated cells, Gag and Env colocalized to a polarized GM1-rich membrane cap in 36% (Gag) to 39% (Env) of infected cells (Table 1). In Jurkat_LAI cells where completely polarized Env/Gag/GM1 caps were not observed, HIV-1 antigens were generally distributed in a small number of distinct but closely clustered membrane patches (data not shown). Treatment of cells with the actin-depolymerizing agent latrunculin significantly reduced the percentage of cells with capped Gag and substantially reduced (44%) Env capping, although this did not reach statistical significance (Table 1). Figure 1A shows that in cells in which latrunculin had a dramatic effect, Env labeling was abolished and Gag staining was dispersed around the inner leaflet of the plasma membrane and within the cytoplasm (Fig. 1A). GM1 staining was redistributed into multiple patches scattered around the plasma membrane (Fig. 1A) but retained some weak colocalization with Gag. In a similar manner, the microtubule polymerization inhibitors nocodazole and colchicine eliminated Env, GM1, and Gag polarization within and under the membrane, respectively, substantially reduced Env detection (nocodazole), and reduced Gag colocalization with GM1 to a few small patches (Fig. 1A). These effects reached statistical significance for colchicine and were a substantial (44% and 38% reductions for Gag and Env, respectively), although nonsignificant, trend with nocodazole (Table 1). To exclude the possibility that the observed effects were artifacts of using an immortalized T-cell line (Jurkat), we repeated the process using HIV-1_IIIB-infected primary CD4⁺ T cells. Figure 1B (left panels) shows an HIV-1-infected, activated CD4⁺ T cell that has been fixed, permeabilized, and stained and in which Gag and Env are capped at one pole of the cell. It should be noted that Gag and Env staining is generally weaker in primary T cells than that observed in Jurkat cells, as exemplified in Fig. 1B. Similar to the case for Jurkat_LAI cells, treatment of the primary T cells with latrunculin, colchicine, or nocodazole resulted in a loss of detectable polarized Env and Gag staining. Latrunculin-mediated depolymerization of the cortical actin cytoskeleton resulted in the characteristic rounding up of the cells from their elongated shape, which is evidence of the collapse of the cortical actin cytoskeleton and of the appropriate function of this inhibitor (Fig. 1B). To further confirm that the inhibitors affected cytoskeletal function, Jurkat_LAI cells, left untreated or treated for 1 h with latrunculin or nocodazole, were fixed, permeabilized, stained for α-tubulin or f-actin, respectively, and counterstained for Gag, and reconstructed 3D images were generated by LSCM. Figure 1C shows Gag predominantly capped at one pole of the cell in untreated Jurkat_LAI cells, although some small clusters of Gag potentially corresponding to intracellular Gag in the process of trafficking, budding or mature virions at the cell surface, or a combination of these can be observed superimposed on f-actin (left panels) or microtubules (right panels). Subsequent to latrunculin treatment, little actin labeling was evident, and only a few remaining microclusters of Gag were detectable. Nocodazole treatment collapsed the microtubule network, resulting in a cytoplasmic ring of fragmented filamentous staining and, again, dispersed Gag staining. Similar results were obtained with colchicine (data not shown), confirming the loss of microtubule integrity induced by different inhibitors.

FIG. 1.

Polarization of Env and Gag in HIV-1-infected T cells requires an intact actin and microtubule cytoskeleton. (A) Jurkat_LAI cells (5 × 10⁵) were washed, resuspended in RPMI-1% FCS, and allowed to adhere to poly-l-lysine-coated coverslips at 37°C. The cells were either left untreated (far left panels) or treated with latrunculin (second panels from the left), nocodazole (second panels from the right), or colchicine (far right panels) and then were stained for HIV-1 Env with MAb 50-69 (blue). Cells were fixed and incubated with cholera toxin to label GM1 (green) and then permeabilized and stained for HIV-1 Gag (red). Images are single sections through the middle of the cell with corresponding Nomarski images (DIC), and areas of red/green colocalization and red/green/blue colocalization appear yellow and white, respectively. (B) HIV-1-infected primary CD4⁺ T cells were either left untreated (far left panels) or treated for 1 h with 1 μM latrunculin (second panels from the left), nocodazole (second panels from the right), or colchicine (far right panels) and then were stained for Env and Gag as described above. Images are single sections through the middle of the cell with corresponding Nomarski images (DIC). Red/blue colocalization appears mauve. (C) Jurkat_LAI cells (5 × 10⁵) were applied to coverslips and either left untreated (far left panels and second panels from the right) or treated with latrunculin (second panels from the left) or nocodazole (far right panels). The cells were then fixed, permeabilized, and stained for HIV-1 Gag (red) and either f-actin or α-tubulin (green). Images are 3D reconstructed z series with corresponding Nomarski images (DIC). Red/green colocalization appears yellow.

TABLE 1.

Quantification of HIV-1 antigen capping in T cells

Treatment	% Capped Env	% Capped Gag	No. of effector cellsa
None	39	36	247
Latrunculin	22b	10***	282
Nocodazole	24	20	268
Colchicine	17**	16*	267

^aNumber of Jurkat_LAI effector cells analyzed.

^bDifferences were analyzed by an unpaired Student t test with Bonferroni correction. Asterisks indicate significant differences from the untreated control. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Visualization of cytoskeletal disruption in living cells.

Fixed-cell immunofluorescence staining and LSCM analysis reveal snapshots of processes such as cytoskeletal depolymerization and viral antigen depolarization but cannot reveal these changes in individual cells over time. To investigate the effect of inhibition of cytoskeletal remodeling on HIV-1 antigen capping, we established a live-cell microscopy system in which cells labeled with quantum dot-tagged MAb 50-69, specific for viral gp41, were visualized over time, without inhibitors or during treatment with cytoskeletal inhibitors. Figure 2A shows individual frames of Jurkat_LAI cells over a 90-min period, and each series of frames is a representative of three or four independent movies captured for each condition. In the left panels, an untreated Jurkat_LAI cell expresses Env capping a single pole of the cell over the entire time course. In contrast, cells treated with latrunculin or colchicine demonstrated fragmentation of Env staining over time, resulting in dispersion of the Env signal into small patches by 90 min (colchicine) or in gradual reductions in the area of polarized label and in signal strength (latrunculin). Loss of Env staining is not a result of photobleaching, since quantum dots are resistant to photobleaching, as is evident from the bright labeling of Env in the control cells over the entire 90 min. Rather, we believe that the differences observed between the inhibitors (fragmentation of the Env cap after colchicine treatment compared to a slow loss of Env signal after latrunculin treatment) reflect differences in how Env is redistributed in the cell membrane after treatment with the two inhibitors. We speculate that depolymerization of microtubules with colchicine results in raft-actin membrane caps breaking up into smaller actin-raft aggregates or patches. In contrast, depolymerization of the actin scaffold results in a more profound loss of structural integrity of the capped rafts, which results in dispersion of the raft-like cap and its Env contents within the plasma membrane, yielding the effect of dilution, as observed in Fig. 2A (right panels). It is likely that the observed loss of Env polarization in Fig. 2A is an underestimate of the true rate of HIV-1 antigen depolarization because in these images, the Env cap is likely to be partially stabilized via cross-linking with primary and secondary antibodies; however, we were unable to image this process in the absence of antibodies. We did not see any obvious internalization of Env over the 90-min period, and to exclude the possibility that the loss of signal was due to inhibitor-induced shedding of Env from the cell surface, we treated cells with inhibitors for 3 h, surface stained them for Env, and performed flow cytometric analysis. No difference in Env expression was found between untreated and inhibitor-treated cells (Fig. 2B). Thus, it is most probable that Env is dispersed within the plane of the membrane, as observed when HIV-1-infected T cells are treated with β-cyclodextrin to disperse lipid raft-like domains (29).

FIG. 2.

Live-cell imaging of Jurkat_LAI cells treated with actin- or tubulin-depolymerizing agents. (A) Jurkat_LAI cells were stained for Env with MAb 50-69 and Qdot 655 (red), and images were acquired by live-cell LSCM. The left panels show individual frames selected from an approximately 90-min time-lapse analysis of an untreated Jurkat_LAI cell (note that Env stayed as a cap at one pole of the infected cell throughout the entire acquisition period). The selected time frames (from the top) were 0, 10, 23, 57, and 90 min. Treatment with colchicine (middle panels; time frames [from top], 0, 11, 28, 57, and 85 min) or latrunculin (right panels; time frames [from top], 0, 11, 31, 50, and 65 min) resulted in disruption of Env polarization and dispersal of the Env cap over the time-lapse series. Images are single x-y sections through the plane of the cell, selected for the strongest Env staining, and the bottom frames represent differential interference contrast (DIC) images of the last time points, with the fluorescence superimposed. (B) Total surface expression level of Env is unaffected by depolymerization of the cytoskeleton. Jurkat_LAI cells were either left untreated (white bar) or treated with inhibitors (gray bars) for 3 h and then surface stained for HIV-1 Env with the gp120 MAb 2G12. Surface expression was quantified by flow cytometric analysis, and values are the mean fluorescence intensities (MFI) for two independent experiments, with the background staining subtracted. Error bars show the standard errors of the means (SEM).

Disruption of actin remodeling reduces release of cell-free Gag, Env, and infectious virus.

Disruption of the HIV-1 polarized assembly platform might be expected to have functional consequences on virion budding and release. To investigate this, we performed quantitative Western blotting on cell and viral lysates prepared from HIV-1-infected Jurkat cells that were either left untreated or treated with cytoskeletal inhibitors for 3 h. We chose to use a 3-h treatment because this was the shortest time that allowed us to obtain supernatants with reproducibly detectable Gag levels. Cell and viral lysates from untreated and inhibitor-treated cells were separated by SDS-PAGE, and Western blots were probed for HIV-1 Gag to determine the relative amounts of cell-associated and cell-free Gag (Fig. 3A). Both Gag precursor polyprotein p55 and capsid protein p24 were readily detected in both cell and viral lysates. Similar amounts of Gag p55 and p24 were observed in cell lysates from untreated and inhibitor-treated Jurkat_LAI cells. Although there appeared to be a slight reduction in p24 levels in viral lysates prepared from inhibitor-treated cells, when the ratios of total cell-free to total cell-associated Gag were calculated for three independent virus release assays (Fig. 3B), only latrunculin treatment showed a reproducible decrease in Gag release (ratio = 0.03) compared to that of the untreated control (ratio = 0.06). Because we had observed an uncoupling of Env and Gag cocapping by immunofluorescence microscopy, we sought to determine whether this had any effect on Env incorporation into budding virions. To do this, we treated Jurkat_LAI cells with inhibitors for 3 h, harvested the supernatants, concentrated the virus by centrifugation, and probed viral lysates for Env by Western blotting. Figure 3C shows a representative Western blot of gp120 levels in concentrated virions released by untreated and inhibitor-treated cells. The intensities of the gp120 bands were determined by densitometer analysis, and the averages for two independent virus release assays are shown (Fig. 3D). Treatment of HIV-1-infected cells with the actin depolymerizers cytochalasin D and latrunculin caused marked decreases in released gp120 levels, a more modest effect was seen with nocodazole, and colchicine had no effect. The reduction of Env incorporation into virions was not the result of decreased total levels of membrane-incorporated Env, since Env cell surface expression remained unchanged after inhibitor treatment (Fig. 2B). Additionally, only latrunculin gave a substantial (∼50%) reduction in virion release, as measured by quantitative Gag Western blotting. Thus, the most likely explanation is that disruption of the cytoskeleton reduces Env incorporation into virions, possibly due to the uncoupling of Env and Gag colocalization (Fig. 1).

FIG. 3.

Effects of cytoskeletal inhibitors on HIV-1 release and viral infectivity. Jurkat_LAI cells were either left untreated or treated with 1 μM of nocodazole, colchicine, cytochalasin D, or latrunculin for 3 h at 37°C. (A) Cell and viral lysates from untreated or inhibitor-treated cells were separated by SDS-PAGE, and HIV-1 Gag was detected by Western blotting. C, lysates prepared from cell pellets; V, lysates prepared from cell-free virus-containing supernatants. For each condition, 20 μg of cell lysate was used and an equal volume of viral lysate was loaded to allow a direct comparison of virus outputs. The Gag polyprotein p55 and the capsid protein p24 are indicated. (B) Densitometer analysis was performed on Gag p55 and p24 bands from nonsaturated Western blots, and for each treatment condition, the ratio of total cell-free to total cell-associated Gag was calculated. Values are the averages for three independent virus release assays, and the error bars show the SEM. (C) Env incorporation into HIV-1 virions from cells treated with cytoskeletal inhibitors. Supernatants were harvested from Jurkat_LAI cells that were either left untreated or treated with cytoskeletal inhibitors for 3 h, and virions were concentrated by centrifugation. Viral lysates were separated by SDS-PAGE, and HIV-1 Env was detected by Western blotting with rabbit anti-gp120 serum. A representative blot is shown. (D) Densitometer analysis was performed on gp120 bands from nonsaturated blots. Values are the average pixel intensities of individual bands corresponding to gp120 and are the averages for two independent virus release experiments. Error bars show the SEM. (E) Infectivity assay on viral supernatants harvested from inhibitor-treated cells. Jurkat_LAI cells were either left untreated (white bar) or treated for 3 h with cytoskeletal inhibitors (gray bars). The cells were then washed to remove inhibitors and incubated for 1 h at 37°C, and virus-containing supernatants were collected and used to infect HeLa reporter cells. Viral infectivity was measured by performing a β-galactosidase assay, and values show the percentages of infectivity relative to that of the untreated control, which was normalized to 100%. Data are the averages for two independent virus release assays, and error bars show the SEM. (F) After 3 h of incubation in the presence or absence of inhibitors, the number of viable Jurkat_LAI cells was calculated by trypan blue exclusion. Error bars show the standard deviations (SD) for a single experiment performed in triplicate. (G) The metabolic activity of treated or untreated cells was determined using a nonradioactive colorimetric cellular proliferation assay (Promega) that measures bioreduction of the MTS tetrazolium compound. The absorbance at 490 nm was measured, and values are shown for each inhibitor and the untreated control. Error bars show the SD.

To demonstrate that the changes in virion assembly and release that we observed have functional consequences for the virus, we performed an infectivity assay. Viral supernatants were harvested from inhibitor-treated cells, and these were used to infect HeLa P5 reporter cells. HIV-1 infection of these cells drives β-galactosidase expression from the HIV-1 long terminal repeat, and thus relative levels of viral infectivity can be quantified by performing a β-galactosidase assay. To ensure that the supernatants were free of any inhibitors that would depolymerize the cytoskeleton in the reporter cells, it was necessary to wash out the inhibitors after treatment and subsequently incubate the cells in inhibitor-free medium to allow virus to accumulate. Figure 3E shows that cells treated with latrunculin contained 50% less infectious virus than supernatants from untreated cells (P = 0.004), supporting our virus release results. We did not observe any statistically significant reduction in viral infectivity for cells treated with nocodazole, colchicine, and cytochalasin D; however, unlike latrunculin, these inhibitors are rapidly reversed upon washout, and thus any effect on viral infectivity would be rapidly lost.

To ensure that the effects we observed were not due to reduced cell viability, we determined the total number of viable cells present after 3 h of inhibitor treatment by trypan blue exclusion. No difference between untreated and inhibitor-treated cells was evident (Fig. 3F). As a further control, we compared the metabolic activities of treated and untreated cells by using a colorimetric cell proliferation assay (Fig. 3G). Treatment with colchicine, cytochalasin D, or latrunculin had no significant effect on the metabolic activity of cells compared to that of untreated control cells (P = 0.08, 0.36, and 0.4, respectively). We did detect a statistically significant (50%) reduction in metabolic activity after nocodazole treatment (P = 0.0004); however, this was not reflected in reduced cell viability, as measured by trypan blue exclusion (Fig. 3F), or in reduced levels of cell-associated Gag (Fig. 3A) and, moreover, was not observed after treatment with colchicine. Taken together, the data show that depolymerization of the actin cytoskeleton with latrunculin reduces cell-free virion release and that depolymerization of the actin and tubulin network reduces Env levels in virions in the absence of a global cellular perturbation in either Gag or Env expression. This implies a generalized phenomenon of reduced efficiency of viral assembly, budding, and/or release in response to disruption of components of the cytoskeletal network.

Cytoskeleton inhibitors interfere with VS assembly.

We have previously demonstrated that HIV-1 cell-cell spread between T cells can take place via formation of a VS. In these studies, treatment of the target cell with inhibitors of actin, but not tubulin, remodeling prevented formation of the VS and blocked transfer of Gag into the target cell (28). To investigate whether interference with cytoskeletal function in the effector cell influenced VS formation, we pretreated Jurkat_LAI cells with cytochalasin D or nocodazole, added them to primary CD4⁺ T-cell targets for 1 h at 37°C on coverslips, and fixed, permeabilized, and stained the cells for CD4, Env, and Gag. Untreated cell conjugates showed characteristic VS formation, with CD4 capping on the target cell and Gag and Env cocapping the membrane of the opposing effector cell (Fig. 4). Pretreatment of the effector cells with cytochalasin D or nocodazole resulted in similar phenotypes for the two inhibitors: although target-effector conjugates still formed at a similar frequency to that for untreated cells, there was no evidence for capping by CD4, Gag, or Env at the conjugate interface. Indeed, as observed previously on infected cells, Env staining became undetectable and Gag was evenly distributed within the cytoplasm, whereas CD4 on the target cell was evenly distributed around the plasma membrane. These results demonstrate that CD4 capping on the target cell towards the VS is dependent on polarized viral antigens in the effector cell, with the most probable candidate being Env. Quantification of Env and Gag capping to the VS was carried out, and the results are presented in Table 2. All inhibitors yielded highly significant inhibition of Gag and Env capping of the VS. Latrunculin was the most potent inhibitor, although this treatment also reduced the number of conjugates formed, as we observed previously (28).

FIG. 4.

Cytoskeletal inhibitors reduce VS formation. Jurkat_LAI cells (5 × 10⁵) were either left untreated (left panels) or treated with cytochalasin D (middle panels) or nocodazole (right panels) and then were incubated with an equal number of freshly isolated primary CD4⁺ target T cells from a healthy donor. Conjugates were allowed to evolve on coverslips for an hour at 37°C in the presence of noninhibitory MAbs against CD4 (green) and Env (blue). Synapse evolution was arrested by fixation, and the conjugates were permeabilized and stained for HIV-1 Gag (red). Images are single sections through the middle of a conjugate, with the target cell at the top and the effector at the bottom. Areas of three-color colocalization appear white.

TABLE 2.

Inhibition of HIV antigen recruitment to VS

Treatment	% Conjugated cellsa	% Polarized cellsb,c		No. of effector cells
		Env	Gag
None	19	69	69	136
Latrunculin	5	0***	0***	127
Nocodazole	31	35***	35***	227
Colchicine	20	0***	0***	300

^aPercentage of total effector cells in conjugates with target cells. Conjugates were defined as closely apposed pairs of cells containing at least one effector cell and one CD4⁺ target cell.

^bPercentage of effector cells that showed polarized Env and Gag at the conjugate interface and formation of a VS.

^cDifferences were analyzed with an unpaired Student t test with Bonferroni correction. ***, significantly different from the untreated control (P < 0.001).

Cytoskeleton inhibitors reduce HIV-1 cell-cell spread.

To quantify the spread of HIV-1 from effector to target T cells, we developed a real-time, quantitative PCR (qPCR) assay to detect de novo synthesis of late (pol) viral DNA sequences subsequent to mixing effector and target cells. Detection of new viral DNA synthesis does not indicate viral integration into target cell chromatin but demonstrates successful reverse transcription of incoming viral RNA genomes, a proportion of which would be expected to integrate. Viral DNA detection was expressed as a function of invariant human serum albumin (HSA) DNA. Initial studies were designed to evaluate the specificity of the intercellular interactions leading to synthesis of viral DNA in target cells. Target A2.01 cells that express CXCR4 but not CD4 (CD4⁻ CXCR4⁺), and are therefore nonpermissive for VS assembly and cell-cell spread of HIV-1, were mixed with Jurkat_LAI cells for the stated times and DNA extracted and analyzed. The values obtained were used as the baseline and were subtracted from the test values (Fig. 5A). Test values were obtained by mixing permissive A3.01 cells (CD4⁺ CXCR4⁺) with effector cells for the stated times and DNA extracting and analyzing as described above. At 1 h, an increase of 0.5 over baseline was observed. At 3 h, this had risen to 1.5, and by 6 h, an increase of 4.5 was observed. Pretreatment of target cells with the blocking CD4 MAb Q4120 or a small-molecule antagonist of the gp120-CXCR4 interaction (AMD3100) resulted in values that were significantly lower (P < 0.0002) at both the 3-h and 6-h time points than those for the untreated sample, confirming the requirement for Env-receptor interactions in cell-cell HIV-1 spread (28). We pretreated effector cells with actin-depolymerizing agents cytochalasin D and latrunculin and the actin-stabilizing agent jasplakinolide and assayed their effects on de novo HIV-1 DNA synthesis in A3.01 target cells (Fig. 5B). At 3 h, an index of 0.8 above baseline was observed in the presence of latrunculin and cytochalasin D, and a value of 0.6 above baseline was observed in the presence of jasplakinolide. These were significantly less than the untreated control index of 1.8 (for latrunculin, P = 0.0009; for cytochalasin D, P = 0.035; and for jasplakinolide, P = 0.0004). At 6 h, the index had increased to 1.6 above baseline in the presence of latrunculin or cytochalasin and to 1.7 above baseline in the presence of jasplakinolide, but all values were still significantly less than the signal (7.5) in the absence of inhibitor (for latrunculin, P = 0.0054; for cytochalasin D, P = 0.002; and for jasplakinolide, P = 0.0054). Pilot experiments demonstrated that the effects of the microtubule-depolymerizing agents colchicine and nocodazole are rapidly reversible after washing. Since we previously demonstrated that no interference with VS assembly and function appears to occur when target cells are pretreated with nocodazole (28) or colchicine (unpublished results), we assayed de novo DNA synthesis in cultures in which the drugs were maintained. We cannot exclude an effect of the inhibitors on the target cells but assume that the dominant inhibitory activity is within the effector cells. Although a small increase in DNA synthesis was observed over the time course (Fig. 5C), this was significantly less than that observed in the absence of inhibitor, and at 3 h postmixing the index was 0.8 for colchicine and 0.7 for nocodazole (P = 0.0006 and 0.0003, respectively), in contrast to 1.8 in the absence of inhibitor. By 6 h postmixing, the index had increased to 1.7 and 2.1 for colchicine (P = 0.0016) and nocodazole (P = 0.0022), respectively, compared to 7.5 for the untreated control cells, and the differences were still statistically significant. Thus, inhibition of remodeling of both the actin and tubulin cytoskeleton in the effector cell interferes dramatically with HIV-1 cell-cell spread leading to de novo viral DNA synthesis.

FIG. 5.

Cytoskeletal inhibitors reduce cell-cell spread, as measured by qPCR. Equal numbers of Jurkat_LAI cells and uninfected target cells were mixed and incubated for 1, 3, and 6 h prior to cell lysis and purification of total cellular DNA. qPCR using pol primers was performed to measure de novo viral DNA synthesis. Data were normalized to HSA and are shown as ratios of HIV-1 DNA to HSA DNA. The data obtained with A2.01 target cells were used as baseline HIV-1 DNA synthesis at the different time points and were subtracted from the signal with A3.01 target cells. (A) Jurkat_LAI cells were either left untreated or incubated with Q4120 or AMD3100 prior to being mixed with A3.01 target cells to inhibit Env-CD4 or Env-CXCR4 interactions, respectively. (B) Jurkat_LAI cells were either left untreated or incubated with an inhibitor of actin remodeling (latrunculin, cytochalasin D, or jasplakinolide) prior to being washed and mixed with A3.01 cells. (C) Jurkat_LAI cells were either left untreated or incubated with an inhibitor of microtubule remodeling (colchicine or nocodazole). Bars represent the means for triplicate samples plus 1 SD.

DISCUSSION

There is a growing consensus that the cytoskeleton is intimately involved in providing a scaffold upon which plasma membrane cellular receptors interact with the exterior environment in the context of lipid rafts (15, 21, 55). An example of this is leukocyte IS assembly, which relies upon coordination of lipid raft-containing receptors by both actin and tubulin functional elements of the cytoskeleton (36, 37, 64). Here we demonstrate that similar to IS assembly, HIV-1 assembly and budding take place within a lipid raft-rich platform that appears to be constrained and/or maintained by the actin and tubulin cytoskeleton. Disruption of either actin or tubulin remodeling dispersed this platform, resulting in reduced cell-cell HIV-1 spread via a VS, and disruption of the actin cytoskeleton reduced cell-free virion release and viral infectivity.

Our study relied upon the actions of a series of pharmacological inhibitors of actin and tubulin polymerization to disrupt function. Although there are potential pitfalls to this approach, including cellular toxicity and nonspecific effects on other cellular functions, we believe that the data from our cell viability and metabolic assays demonstrate that neither of these is likely to be of major significance in this study. This is because we chose the lowest doses and the shortest treatment times that yield functional effects on the target structures to avoid toxicity and demonstrated (unpublished results) that the effects of these inhibitors are reversible. Moreover, we used more than one inhibitor of each cytoskeletal element, each with a different mode of action, adding weight to our conclusions regarding the proposed inhibitory effects. Other, more specific approaches, such as the use of small interfering RNA, dominant-negative mutants, and antibody-mediated inhibition, are difficult to apply to the T-cell cytoskeleton, as its disruption tends to be toxic over extended periods, and moreover, T cells are notoriously difficult to manipulate in these ways.

Polarized HIV-1 assembly and virion release in infected leukocytes are well documented in the literature, although the underlying mechanisms directing these phenomena are not yet understood. Most reports of polarized HIV-1 release describe it as occurring at sites of cell-cell contact, suggesting that intercellular interactions may stimulate polarization of HIV-1 assembly and budding (12, 27, 28, 50). However, we and others (10, 29, 40) have reported polarization of HIV-1 antigens in the T-cell plasma membrane in the absence of cell-cell contacts, which may be mediated in part by the self-assembling nature of Gag (1, 46, 52) and a tyrosine-based polarization signal in Env (10). The concept of intrinsic polarization is reinforced in the current study, in which we demonstrate that between 36 and 39% of HIV-1-infected Jurkat T cells have viral Env and Gag capping to one pole of the cell. Although we cannot exclude the possibility that transient cell-cell contacts may have induced these antigen caps, this seems unlikely, as they existed even in dilute cultures, in which cell-cell contact is presumably rare (C. Jolly and Q. Sattentau, unpublished results). We therefore hypothesize that both intrinsic and cell contact-induced polarization (in a VS) of viral budding is likely to occur in HIV-1-infected T cells. Polarized budding in single cells may represent a later stage of viral infection in which GM1-rich patches of Env and Gag self-assemble into membrane caps, whereas acutely infected cells may require signals from intercellular contacts that drive polarized viral assembly at cell-cell contact areas. Intrinsic polarization is likely to favor budding and release of cell-free virions, since assembly is focused to one region of the plasma membrane (10, 12, 27, 50), whereas both intrinsic and induced polarization would be predicted to increase the efficiency of cell-cell HIV-1 spread, as they would favor multiple Env-receptor ligation events. Our observations that capping of Gag and Env and cell-cell HIV-1 transmission are inhibited by agents that interfere with either actin or tubulin remodeling are consistent with these models of HIV-1 assembly and release. The reduction in cell-cell spread of HIV-1 measured by quantitative reverse transcription-PCR following depolymerization of the actin-tubulin network is dramatic, and taking this together with the substantial reduction in Env detection by live-cell imaging of inhibitor-treated cells, we conclude that a functional cellular cytoskeleton within the HIV-1-infected cell is necessary for efficient cell-cell spread across the VS. We cannot exclude the possibility that the reduction in cell-cell spread is a consequence, at least in part, of a decrease in cell-free virus infection, which would, in turn, impact the efficiency of cell-cell spread. However, we believe that since cell-cell transfer of HIV-1 is mediated by cell surface Env engaging viral receptors on the target cell and since Env is essentially absent from the VS in inhibitor-treated cells, this represents the dominant mechanism underlying the block to cell-cell spread and subsequent de novo DNA synthesis.

Our data obtained with actin remodeling inhibitors are consistent with two previous studies from Sasaki and colleagues, who demonstrated that treatment of the HIV-1-infected, immortalized T-cell line H9 or the monocytic line U937 with the actin-depolymerizing agents cytochalasin D and mycaloid B resulted in altered virion morphogenesis and reduced virion release (58, 59). Although the underlying mechanism was not elucidated and the potential toxic effects of the inhibitors were not investigated, the authors speculated that actin-myosin interactions participate in the transport of newly synthesized HIV-1 components to the assembly site and that their disruption reduces viral assembly and subsequent release from the cell surface. Our own data suggest that actin is required to bring the viral components together at the membrane to maintain the integrity of the budding platform. An additional function of actin may be to influence the localization and/or function of the ESCRT complex, which is required for viral budding; in support of this possibility, an interaction between actin and ESCRT-1 has been proposed for Drosophila developmental pathways (61).

We showed here that the microtubule-depolymerizing agents colchicine and nocodazole had an inhibitory effect on the polarization of Gag and Env and on cell-cell spread of virus that was phenotypically distinct from but functionally similar to the effect of actin inhibitors. We propose that this is because microtubules play related or complementary roles in facilitating assembly and maintenance of the polarized viral budding platform. In contrast, Pearce-Pratt and colleagues (47) reported that treatment of the chronically HIV-1-infected, immortalized MOLT4 T-cell line with the microtubule-depolymerizing agent colchicine induced the polarization of HIV-1 budding. Since the same virus isolate (HIV-1_IIIB) was used in both studies and since MOLT4 cells are unlikely to differ dramatically from Jurkat T cells in microtubule function, we assume that the difference may result from the concentrations of inhibitor used: we used 1 μM colchicine, whereas Pearce-Pratt et al. used a 20-fold higher concentration. Colchicine is a substoichiometric inhibitor with a complex mode of action, and different concentrations can yield dramatically different functional effects on microtubule function (62). Moreover, we found that both colchicine and nocodazole had similar effects on Jurkat_LAI cells, adding weight to our observation.

The existing evidence, mainly indirect, for links between the microtubule cytoskeleton and HIV-1 assembly is as follows. (i) Vesicles, including multivesicular bodies and secretory lysosomes, are trafficked to the plasma membrane on microtubules (39). (ii) The adaptor protein AP3, implicated in microtubule-mediated transport of secretory lysosomes, is important in HIV-1 trafficking and assembly (11). (iii) In both HTLV-1 (3, 25)- and HIV-1 (C. Jolly, unpublished data)-infected cells, the microtubule organizing center is polarized proximal to the site of cell-cell contact in the VS, in a manner analogous to that observed in the IS (7). Microtubules associate with the cortical actin cytoskeleton and/or the plasma membrane via a network of adaptors and linkers (15, 21, 55). External stimuli promote reorganization of the cytoskeleton in leukocytes to mediate a diversity of functions, including adhesion, migration, and synapse formation (36, 37, 57, 63). Once junctions with target cells or the extracellular matrix have been made, cargo can then be transported towards the plasma membrane along microtubules and may then transfer to cortical actin prior to being incorporated into or released at the plasma membrane. Indeed, viruses such as vaccinia virus exploit this network to gain access to cortical actin, from which they induce formation of actin tails that propel the virus into neighboring cells (13, 16, 17).

Based on the data presented here, we propose that HIV-1 may follow a similar microtubule-directed trajectory to gain access to the T-cell plasma membrane, from which it can then disseminate via cell-free or cell-cell spread. Elucidating the molecular interactions driving this pathway will allow a greater understanding of viral subversion of lymphocyte secretory pathways and may lead to novel insights into therapeutic intervention.