Establishment of a Functional Human Immunodeficiency Virus Type 1 (HIV-1) Reverse Transcription Complex Involves the Cytoskeleton

293T cells (33) were maintained in DME supplemented with penicillin (100 U/ml), streptomycin (100 μg/ml), and 10% fetal bovine serum (FBS). For generation of viral stocks, 293T cells were transfected with HIV-1_LAI (34) proviral DNA (25 μg) using a modified calcium phosphate/DNA precipitation method (GIBCO BRL, Gaithersburg, MD) and incubated at 37°C. Virions in culture supernatants were collected at 24 and 48 h after transfection and either used directly for infection or amplified in MT-4 T cells in RPMI 1640 supplemented with 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 10% FBS. Pseudotype viruses were obtained after cotransfection of 293T cells with a vesicular stomatitis virus (VSV)-G envelope expression vector and an HIV-1_LAI clone containing a 600-bp deletion in gp120 coding sequences as described previously (35, 36). Calcium phosphate–transfected culture supernatants were collected at 48 h after transfection. Virus particles in culture supernatants were quantitated by measurement of reverse transcriptase activity. For preparation of ³²P-gag MA– labeled virions, productively infected CD4⁺ MT-4 cells were incubated in the presence of 0.5 mCi ortho [³²P]phosphate per ml in phosphate-free medium containing 0.1% calf serum. After a 6-h incubation, culture supernatants were cleared of cellular debris by low-speed centrifugation (1,500 g, 10 min), and virions in clarified supernatants were harvested after high-speed centrifugation (10,000 g, 2 h).

Subcellular fractions of HIV-1– infected MT-4 cells and HeLa-CD4-LTR/β-gal “Magi” cells comprising cell membrane, cytosolic, cytoskeletal, and nuclear extracts were prepared essentially as described elsewhere (20). Cells were pelleted and washed in a hypotonic solution (10 mM Hepes, pH 6.9, 10 mM KCl, 3 μl/ml aprotinin, 0.1 mM PMSF) and incubated on ice in hypotonic solution for 15–20 min. Cells were disrupted by dounce homogenization (20 strokes). Nuclei were pelleted at 1,000 g (3 min), supernatant was removed, and nuclei were further purified by three successive washes (retained as nuclear wash extract) in NTENT buffer (150 mM NaCl, 10 mM Tris-HCl, pH 7.2, 1 mM EDTA, 3 μl/ml aprotinin, 0.1 mM PMSF, 1% Triton X-100). Supernatant from pelleted nuclei was further centrifuged at 14,000 g for 30 min. The supernatant (cytosolic fraction) was separated, and the pellet was further resuspended in NTENT and recentrifuged at 14,000 g for 30 min. The resulting pellet comprises the cytoskeletal fraction, while the supernatant comprises the membrane fraction.

Subcellular fractions from an equivalent of 10⁶ cells were adjusted to 62.5 mM Tris-HCl, pH 6.8, 0.2% SDS, 5% 2-ME, 10% glycerol. Samples were denatured by boiling and separated on 12% SDS-polyacrylamide gels. After electrophoresis, proteins were transferred to nitrocellulose (Hybond C Extra; Amersham Pharmacia Biotech, Inc., Piscataway, NY), blocked in 5% milk, and incubated with the respective antibody. Bound antibodies were visualized with peroxidase anti-Ig followed by enhanced chemiluminescence (ECL; Amersham Pharmacia Biotech, Inc.). All Western blot images were within the linear range of the ECL assay. Rhodamine-conjugated phalloidin and antibodies to tubulin, vimentin, actin, and myosin were obtained from Sigma Chemical Co. (St. Louis, MO). For those experiments comparing protein distribution between different cellular compartments, cell lysate from an equivalent of 10⁶ cells was resolved in each lane. This allowed for qualitative comparison of protein distribution between cellular compartments. For those experiments comparing the effect of cytoskeletal inhibitors on protein distribution within the same cellular compartment, the amount of total protein in each lane was adjusted to equivalent levels and, as a consequence, more quantitative comparisons could be made.

MT-4 cells were harvested 1.5 h after infection. Lysates of infected and uninfected cells were obtained by disruption of cells in immunoprecipitation buffer (50 mM Hepes, pH 6.9, 10 mM EDTA, 1% Triton X-100, 3 μl/ml aprotinin, 1 mM PMSF). The extracts were centrifuged at 1,000 g (3 min), and supernatants were precleared with normal rabbit antiserum (6 h, 4°C). Precleared lysates were adjusted to 100, 300, and 800 mM NaCl. Viral proteins were immunoprecipitated with rabbit anti-CA and anti-MA antibodies overnight at 4°C in the presence of protein A and G. Bound immunoprecipitates were washed five times in immunoprecipitation buffer and resolved on 15% SDS-PAGE gels. Proteins were transferred to Highbond C Extra (Amersham Pharmacia Biotech, Inc.) and analyzed by Western blotting with antibodies to tubulin, vimentin, actin, and myosin (Sigma Chemical Co.). Bound antibodies were visualized with peroxidase anti-Ig, followed by ECL (Amersham Pharmacia Biotech, Inc.).

Cells were seeded onto 12-well plates containing 0.13-mm-thick coverslips. Cells were grown for 24 h at 37°C in DME containing 10% FBS, Hygromycin (100 μg/ml), and G418 (200 μg/ml). For cytoskeletal staining, coverslips containing adherent cells were removed and rinsed two to three times with warm (37°C) PBS. Coverslips were placed in a petri dish and permeabilized with a solution of 0.1% Triton X-100 in PBS for 1 min, then washed twice with warm PBS. Cells were then immediately placed in warm (37°C), freshly prepared formaldehyde-Triton solution (4% formaldehyde, 0.2% Triton X-100, 60 mM Pipes, 25 mM Hepes, 10 mM EGTA, 3 mM MgCl₂, pH 6.1) for 10 min. The coverslips were then rinsed two to three times with room temperature PBS. To reduce nonspecific background staining, fixed cells were incubated with PBS containing 1% BSA for 15 min and washed twice with PBS before addition of staining solutions. For staining of actin microfilaments, cells were overlaid with 200 μl of a PBS/1% BSA solution containing 5 U/ml Texas Red phalloidin (Molecular Probes, Inc., Eugene, OR). Cells were incubated at room temperature for 20–30 min and then rinsed twice with PBS. For tubulin and vimentin visualization, permeabilized cells were overlaid with PBS/1% BSA containing tubulin (DM1A) and vimentin (V9) antibodies (Sigma Chemical Co.) for 20 min. Cells were washed twice in PBS and then incubated with FITC-conjugated donkey anti–mouse IgG (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA). Stained cells were mounted in Vectashield (Vector Laboratories, Inc., Burlingame, CA) to minimize photo-bleaching. Images were generated on a fluorescence microscope (Axioplan II; Carl Zeiss, Jena, Germany) using the appropriate FITC and Texas Red filter sets. Photographs were obtained with a Plan Apochromat objective (63× 1.40) under oil immersion.

Total cellular DNA was isolated from acutely infected cells using an IsoQuick DNA extraction protocol (Microprobe Corp., Bothell, WA). Minus-strand HIV-1 strong-stop cDNA products were amplified using LTR R (5′-g⁶³⁵ tgctagagattttccacactgac) and U5 (5′-g⁴⁹⁶ gctaactaggaacccactg) primers as described previously (37). Late reverse transcription products were amplified by LTR R (5′-g⁴⁸⁵ ggagctctctggctaact) and gag (5′-g⁹³¹ gattaactgcgaatcgttc) primers. Viral cDNA was amplified in 10 mM Tris-HCl, pH 8.3, 50 mM KCl, 1.5 mM MgCl₂, and 0.001% (wt/vol) gelatin. After the initial denaturation (95°C, 2 min), amplification proceeded for 25 or 35 cycles of PCR (95°C, 30 s; 58°C, 30 s; 72°C, 1 min) followed by a final 5-min extension at 72°C. HIV-1 copy number standards were generated by PCR on serial twofold dilutions of 8E5 cells (each cell containing a defective provirus [38]). PCR products were Southern blotted onto nylon membranes and hybridized with a mixture of two ³²P-labeled oligonucleotide probes (5′-g⁵⁶⁶ tctgttgtgactctggt, 5′-g⁵⁸⁴ taactagagatccctcagac) for detection of both early and late products of reverse transcription. Primers are numbered according to the HIV-1_HXB2 sequence (39). Hybridized PCR products were visualized on a PhosphorImager (Molecular Dynamics, Sunnyvale, CA). Hybridized products were quantitated by volume integration (local mean background subtraction was applied) using ImageQuant software (Molecular Dynamics).

HeLa-CD4-LTR/β-gal Magi indicator cells (35) were plated in 96-well microtiter plates (2 × 10⁴ cells/well). HIV-1 virions (normalized to reverse transcription activity) were added in doubling dilutions to triplicate wells. At 48 h after infection, cells were harvested for quantitation of β-galactosidase production after hydrolysis of X-gal in vitro (Promega Corp., Madison, WI).

Gag MA is tightly associated with the viral transcription complex (2, 3, 7) likely through a high-affinity interaction with genomic viral RNA (40; our unpublished results), and thus represents a marker with which to follow the distribution of the reverse transcription complex in the infected cell. At 1 h after infection, MT-4 cells were fractionated into cytoplasmic, nuclear, cytoskeletal, and membrane fractions. The purity of the fractions was assessed by Western blotting with antibodies to cytoskeletal proteins (actin, tubulin), nuclear proteins (histone H1), and membrane proteins (CD4). Cytoskeletal proteins were detected primarily in cytosolic and cytoskeletal fractions and were absent from the nucleus (Fig. 1,A). Actin is present in the cell in two forms, as polymerized filaments and as unpolymerized monomers. Polymerized actin is enriched in cytoskeletal fractions, whereas much of the actin that is recovered from cytosolic extracts is in the unpolymerized form. The degree of actin polymerization varies between experiments and, as a consequence, the amount of actin recovered in cytosolic fractions varies between experiments. Histone H1 was present in nuclear fractions but not cytosolic or cytoskeletal fractions. CD4, the principal receptor for HIV-1, was restricted to membrane and cytoskeletal fractions, in agreement with previous reports (41, 42). Gag MA was found in all fractions with the exception of the cytoplasm (Fig. 1,B). We and others have previously demonstrated that gag MA that is associated with reverse transcription complexes is phosphorylated whereas that associated with membranes is not (6, 10, 14). Upon infection of cells with HIV-1 virions containing ³²P-gag MA, phosphorylated gag MA protein was localized exclusively to the nucleus and to the cytoskeleton (Fig. 1,C). In agreement with our previously published results (10), phosphorylated gag MA was excluded from the membrane despite the fact that the majority of gag MA in the acutely infected cell remains membrane associated (Fig. 1, B and C). Thus, gag MA in the cytoskeletal fraction was associated with viral reverse transcription complexes and did not represent contamination of that fraction with membrane-associated virion proteins. To examine the kinetics with which viral reverse transcription complexes associated with the cytoskeleton and the nucleus, cells were infected and subcellular fractions were prepared at early intervals after infection. Gag MA appeared in the cytoskeleton within 20 min of infection (Fig. 1,D), whereas its appearance in the nucleus was not evident until 60 min after infection (Fig. 1 D).

We next determined whether the association of the reverse transcription complex with the cytoskeleton was sensitive to cytoskeleton-disrupting agents. MT-4 cells were treated for 2 h with cytochalasin D (CCD, 5 μM), a specific inhibitor of actin polymerization (43, 44). Cells were then infected with HIV-1 and infection was allowed to proceed in the absence of CCD. At 1 h after infection, there was a reduction in the amount of gag MA present in the cytoskeletal and nuclear fractions, whereas there was a slight increase in the amount of gag MA associated with the membrane (Fig. 2,A). Since infection proceeded in the absence of CCD and the effects of CCD are reversible, there was a slight restoration of gag MA association with the cytoskeleton and nucleus when the distribution was examined at 5 h after infection (Fig. 2,A). In comparison, treatment of cells with demecolcine, which causes perinuclear aggregation of intermediate filaments (45), or with nocodazole, a specific inhibitor of tubulin polymerization into microtubules (46), did not significantly affect the association of gag MA with the cytoskeleton (Fig. 2,B). The inhibitory effects of CCD were dependent on the time of drug addition. When cells were treated with CCD for either 1 or 2 h before virus infection, the amount of gag MA in the cytoskeleton and in the nucleus was markedly reduced (Fig. 2,C). The presence of gag MA in the cytoskeleton or nucleus was not significantly affected if CCD was added as early as 1 h after infection (Fig. 2,C). Treatment of cells with CCD inhibits actin polymerization; however, monomeric actin could still be recovered from the cytoskeletal fraction of CCD-treated cells (Fig. 2 D). In addition, inhibition of actin polymerization resulted in an increase in the amount of actin recovered from the cytosolic fraction, likely as a consequence of increased levels of monomeric actin. The presence of gag MA in the cytoskeletal and nuclear fractions of cells treated with CCD after virus infection suggests that the interaction between the reverse transcription complex and the cytoskeleton is retained whether actin is monomeric or polymerized. Since gag MA rapidly associates with the cytoskeleton, and CCD did not affect this association when added 1 h after infection, the results suggest that, once established, the interaction between the reverse transcription complex and the cytoskeleton is not reversed upon actin depolymerization.

We next investigated the mechanism by which viral reverse transcription complexes associated with the cytoskeleton. Recent studies have suggested an interaction between HIV-1 gag polyprotein and F-actin present in microfilaments (47). Therefore, we determined whether the two major processed products of gag polyprotein in the virion, namely gag MA and gag CA, associated with actin in vivo. Cell extracts were prepared from uninfected MT-4 cells and from MT-4 cells 1 h after HIV-1 infection. Lysates of infected and uninfected cells were adjusted to 100, 300, and 800 mM NaCl and immunoprecipitated with antibodies to gag MA or gag CA (Fig. 3,A). The presence of cytoskeletal proteins in these immunoprecipitates was then determined by Western blotting with antibodies to cytoskeletal proteins including actin (microfilaments), vimentin (intermediate filaments), or tubulin (microtubules) and to gag MA and gag CA as outlined in Fig. 3,A. Antibodies to actin reacted with gag MA immunoprecipitates from HIV-1– infected cells at 100, 300, and 800 mM NaCl, suggesting a tight interaction between actin and gag MA (Fig. 3,B). Gag MA immunoprecipitates from uninfected cells also reacted with actin antibody but only at 100 mM NaCl, likely a result of a nonspecific interaction between actin and a cellular protein within gag MA immunoprecipitates (Fig. 3,B). In comparison, antibodies to actin bound to gag CA immunoprecipitates prepared in 100 mM NaCl but not in 300 or 800 mM NaCl. This weak interaction was again nonspecific since it was evident in both infected and uninfected cells (Fig. 3,B). No reactivity of gag MA or gag CA immunoprecipitates was noted with antibodies to vimentin or to tubulin (Fig. 3,B), suggesting that there was no association between these proteins in acutely infected cells. Each antibody recognized its cognate protein at all NaCl concentrations, indicating that the high salt immunoprecipitation conditions did not compromise antibody reactivity (Fig. 3 C). Collectively, these results suggest that gag MA associates with actin during early stages of HIV-1 infection.

To determine whether the cytoskeletal association of viral reverse transcription complexes influenced the function of this complex, we initially determined whether cytoskeletal depolymerizing agents influenced cDNA synthesis within the complex. According to current models of reverse transcription (48), the earliest viral cDNA detectable in the cell after fusion and uncoating is minus-strand strong-stop cDNA which spans the R-U5 region of the viral LTR (Fig. 4,A), whereas a late step in reverse transcription involves synthesis of plus-strand DNA spanning the LTR U5/gag boundary (Fig. 4,A). Thus, LTR R-U5 products represent a marker with which to gauge the extent of virus entry and early reverse transcription, whereas LTR U5-gag products provide an indicator of the extent of late reverse transcription in the infected cell (37). CCD (5 μM) was added to cells 1 h before or 2 h after virus infection. Cells were harvested 4 h after virus infection for isolation of viral nucleic acids from membrane/ cytosol, cytoskeletal, and nuclear extracts. Early (LTR R-U5) and late (LTR U5-gag) cDNA products of reverse transcription in these extracts were analyzed by PCR under limiting dilution conditions. In untreated (control) cells, early reverse transcripts were detectable in all subcellular fractions (Fig. 4,B). Surprisingly, late products of reverse transcription were found only in cytoskeletal and nuclear fractions but were absent from membrane/cytosol extracts (Fig. 4,B). When added 1 h before infection, CCD reduced the amount of early reverse transcripts in all fractions by approximately four- to fivefold (Fig. 4,B). Late products were not detectable in the cytoskeleton of CCD-treated cultures, whereas in the nucleus their abundance was reduced by ∼25-fold (Fig. 4,B). In contrast, addition of CCD to cells 2 h after virus infection had no effect on abundance of either early or late products of reverse transcription in any cell fraction (Fig. 4,B). Both early and late cDNA products were the result of de novo reverse transcription rather than virion-associated cDNA products, since the amount of PCR products was markedly reduced in cultures that had been pretreated with the reverse transcriptase inhibitor, AZT (Fig. 4,B). The results suggest that steps in virus entry leading up to initiation of reverse transcription (fusion/uncoating, activation of the reverse transcription complex) involve the cytoskeleton. The results further indicate that the reverse transcription process is localized primarily on the cytoskeleton and that efficient synthesis of late cDNA transcripts requires the presence of actin microfilaments. This interaction with the cytoskeleton appears to be initiated early upon infection and is also irreversible, since addition of CCD to cells after virus infection did not compromise the reverse transcription process (Fig. 4 B).

A confounding factor in experiments that examine early events in the virus life cycle is that much of the viral material present in the inoculum is noninfectious. Thus, it is possible that the viral nucleic acids and proteins found in association with the cytoskeleton may not be the component of the inoculum that was actually mediating bona fide virus infection. To address this issue, we examined the effects of cytoskeletal inhibitors on HIV-1 infection. After completion of reverse transcription, viral nucleic acids integrate into host cell DNA, leading to the formation of a “provirus” from which genomic and subgenomic viral transcripts are derived. Therefore, we examined the effects of cytoskeletal inhibitors on provirus establishment in a single-cycle Magi infectivity assay (35). We initially determined drug concentrations required to impair organization of their cognate cytoskeletal targets in Magi cells (Fig. 5,A). In adherent cells, actin is organized in “stress fibers” of polymerized actin microfilaments (Fig. 5,A). Treatment of cells with cytochalasins A, D, and E (5 μM) led to a collapse of actin microfilaments into condensed bundles of monomeric actin that was randomly distributed through the cell cytoplasm (Fig. 5,A). This concentration of cytochalasin did not compromise cell viability during the course of the experiment (not shown). Complete disruption of actin microfilaments typically occurred within 1 h of drug addition. All three cytochalasins (CCA, CCD, and CCE) markedly impaired viral infectivity when added 2 h before (2h pre) virus infection (Fig. 5,B). In contrast, the drugs did not significantly affect HIV-1 infectivity when added 2 h after (2h post) virus infection, and in some cases enhanced infection (Fig. 5,B). Thus, the inhibitory effects of the cytochalasins on HIV-1 infectivity were not due, for example, to indirect effects of the drugs on host cell viability. This pattern of inhibition is in strong agreement with the effect of these inhibitors on the association of gag MA with the cytoskeleton (Figs. 2 and 3) and on reverse transcription (Fig. 4) in MT-4 cells, in that CCD was inhibitory when added at least 1 h before but not after virus infection. We next determined whether there was a concomitant effect on the reverse transcription process in Magi cells. Early (LTR R-U5) and late (LTR U5-gag) products of reverse transcription were quantitated at different intervals after infection (Fig. 6). Treatment of CCD only slightly impaired the levels of minus-strand strong-stop cDNA in all fractions (Fig. 6). In agreement with the pattern observed in MT-4 cells, late products of reverse transcription were evident only in the cytoskeleton and the nucleus. However, pretreatment of the cells with CCD delayed the appearance of late products of reverse transcription (Fig. 6). In untreated cells, late products of reverse transcription were detectable in the cytoskeleton and the nucleus at 1 and 3 h after infection, respectively. After CCD treatment, late reverse transcription products could not be detected until 5 h after infection (Fig. 6). In contrast to the inhibitory effects of cytochalasins on HIV-1 entry, the microtubule disrupting agent, nocodazole, had a very modest inhibitory effect on HIV-1 infectivity in several independent experiments (Fig. 7). Although a slight inhibitory effect on HIV-1 infectivity was observed at lower levels of virus inoculum, these effects were equally apparent whether nocodazole was added before or after HIV-1 infection. Since nocodazole had no effect on the interaction between gag MA and the cytoskeleton or on the abundance of gag MA in the acutely infected cell (Fig. 2 B), it is unlikely that nocodazole is directly influencing establishment of the reverse transcription complex. Since the slight inhibitory effect was evident when the inhibitor was added before and after virus infection, we cannot exclude the possibility that nocodazole influenced the infectivity assay indirectly through its effect on host cell function. Collectively, these results indicate that the reverse transcription process in a newly infected cell is localized to the host cell cytoskeleton and that intact actin microfilaments are necessary for efficient reverse transcription.

Translational movement along actin microfilaments is mediated by myosins (for a review, see reference 49). Myosin function is regulated by phosphorylation of the regulatory light chains by myosin light-chain kinase (50), which promotes myosin ATPase activity and polymerization of actin cables (51, 52). ML7 (53) and KT5926 (54; both obtained from Calbiochem Corp., La Jolla, CA) are well-described inhibitors of myosin light-chain kinase (K_i; 21 μM and 18 nM, respectively). Inhibition of myosin light-chain phosphorylation inhibits myosin-mediated translational movement. Thus, we examined whether HIV-1 infectivity was sensitive to agents that affect myosin-mediated translational movement along actin microfilaments. The myosin light-chain kinase inhibitors ML7 (Fig. 8,A) and KT5926 (Fig. 8,B) both impaired viral infectivity. This inhibition was similar to that observed with cytochalasins, in that ML7 inhibited HIV-1 infection when added 2 h before virus infection, but not 2 h after infection. Subcellular distribution of viral reverse transcription complexes in cells treated with KT5926 was determined in order to identify at which point the inhibition of infectivity was manifest. Pretreatment of cells with increasing concentrations of KT5926 between 100 and 500 nM resulted in a gradual decrease in the amount of gag MA that localized to the host cell nucleus, while there was a corresponding increase in the level of gag MA in the cytoskeletal fraction (Fig. 8,C). One explanation for this result is that inhibition of myosin phosphorylation delays the transit of viral reverse transcription complexes from the cytoskeleton to the nucleus. KT5926 also reduced the amount of gag MA that was present in the cytosol and membrane fractions by ∼30%, indicating that viral uptake was also being influenced (Fig. 8 C). However, this modest effect on gag MA uptake was not dose dependent and was unlikely to account for the rather dramatic effect of this inhibitor on virus infection. Thus, although additional effects of the inhibitor on virus uptake cannot be excluded, the inhibitory profiles are more consistent with the hypothesis that the effects of the inhibitor on gag MA redistribution between cytoskeletal and nuclear fractions account for the majority of the inhibitory effects on virus infection.

HIV-1 entry is mediated by pH-independent fusion between viral and host cell membranes (55). When HIV-1 virion cores are pseudotyped with envelope proteins of VSV, infection is mediated by endocytosis (35, 56). Thus, we compared the relative cytoskeletal requirements for virus entry via the fusion and endocytosis routes. In contrast to the marked effects of CCD on wild-type HIV-1 infection (Fig. 5,B), CCD had no significant effect on entry of VSV-pseudotyped HIV-1 when added before virus inoculation (Fig. 9,A). Similarly, CCD had no effect on the association of gag MA with the cytoskeleton (Fig. 9,B). Although a slight inhibitory effect on virus infection was evident in cultures in which the inhibitor was added 2 h after virus inoculation (Fig. 9 A), this very slight effect was not consistent over several independent infectivity experiments and likely fell within the variability of the single-cycle infectivity assay. These results suggest that cytoskeletal association can be uncoupled from infectivity when HIV-1 entry is mediated by endocytosis. It is possible that engagement of the cytoskeleton is coupled to fusion, which places the reverse transcription complex in an appropriate context within the acutely infected cell (i.e., in association with the cytoskeleton). The experiments also indicate that, upon establishment of a functional reverse transcription complex within the cell (whether after fusion or endocytosis), events in the viral life cycle leading up to de novo gene expression from the integrated provirus do not require actin microfilaments.

The results outlined in this study suggest that an essential early event in the virus life cycle, namely establishment of the viral reverse transcription complex, is promoted by actin microfilaments of the host cell cytoskeleton. During preparation of this manuscript, Iyengar et al. (57) demonstrated that HIV-1 infection of primary T cells could be completely inhibited by treatment of cells with CCD. The investigators went on to demonstrate that colocalization of the HIV-1 receptors, CD4 and CXCR4, on primary T cells required actin, and suggested that actin is necessary for a step leading up to fusion between viral and host cell membranes. However, the results presented here clearly indicate that the inhibitory effects of the cytochalasins in T cells and Magi cells can only be partly accounted for by inhibition of a step leading up to entry of the reverse transcription complex. In the experiments presented here, minus-strand strong-stop cDNA was detected in CCD-treated T cells and Magi cells which completely lack actin microfilaments. Our results indicate that the reverse transcription complex rapidly associates with actin microfilaments and that efficient reverse transcription is dependent on the presence of actin microfilaments. Our results also suggest that actin microfilaments of the host cell cytoskeleton are a major site for reverse transcription in the infected cell. Although actin depolymerization did not significantly affect the presence of early reverse transcription products in the cytoskeleton and in the nucleus, these early reverse transcription products did not go on to complete reverse transcription if actin polymerization was inhibited before virus infection. This suggests that interaction with actin microfilaments is an essential intermediate step that is required for efficient reverse transcription, and early reverse transcripts that bypass this intermediate step still translocate to the nucleus but are incapable of undergoing complete reverse transcription. The basis for the dependence of an intact cytoskeleton for efficient reverse transcription is unclear at present, and we know of no precedent in the animal oncoretrovirus or primate lentivirus systems to explain this dependence. Cytoskeletal proteins have also been implicated in the activation of transcription of paramyxoviruses (measles virus, Sendai virus, respiratory syncytial virus) and rhabdoviruses (VSV) (58–60). For example, in paramyxoviruses such as human parainfluenza virus type 3, polymerized actin has been shown to be necessary for the activation of transcription from the ribonucleoprotein complex in vitro and in vivo (61–63). It has been suggested that upon interaction with actin, the paramyxovirus ribonucleoprotein complex undergoes a conformational alteration from a loosely coiled to a moderately condensed form, which facilitates transcription (62). It is possible that activation of HIV-1 reverse transcription also requires a conformational change in the reverse transcription complex, and that this is mediated through interaction with the actin microfilaments. Although there is no direct evidence to suggest that activation of the reverse transcription complex requires interaction with the cytoskeleton, such a model is suggested by the demonstration that the large subunit of HIV-1 reverse transcriptase interacts with β-actin (64). Although this interaction has been implicated in promoting the release of virus particles from the infected cell, in light of our current findings it is tempting to speculate that such an interaction may provide the basis by which association with actin microfilaments was required for efficient reverse transcription. Alternatively, the uncoating process itself may be facilitated by association of the incoming capsid core with the cytoskeleton, due to the interaction between capsid protein and actin. However, the results presented here would favor the former hypothesis in that gag MA within the reverse transcription complex associated with actin, whereas capsid, the major component of the viral core, did not exhibit any significant binding. The infectious virus particle does not contain all the components necessary for efficient completion of cDNA synthesis, since only early reverse transcription products are reconstituted in in vitro reverse transcription reactions using extracts from disrupted virions (for a review, see reference 65). Thus, it is also possible that the cytoskeleton is a favored site for reverse transcription because cellular factors important in the reverse transcription process may also localize to the cytoskeleton.

VSV-G envelope–pseudotyped HIV-1 entry was independent of actin microfilaments, suggesting that virus entry by endocytosis circumvents the aspect of the virus life cycle that requires cytoskeletal interaction. Entry of pseudotyped HIV-1 by endocytosis requires an acidification step which promotes fusion between the viral membrane and the membrane of the endocytic vesicle (56). It is possible that the low pH environment of the endosome may also alter conformation of the viral reverse transcription complex in a way that obviates the requirement for actin in this process. By analogy, the requirement for actin microfilaments in the entry of MLV also depends on the route of virus entry. MLV infection of NIH 3T3 cells occurs by endocytosis, whereas that of XC Sarcoma cells proceeds by surface fusion. Both entry mechanisms are dependent on polymerized actin, implicating a critical role for this network in an early step common to both entry pathways, perhaps at the level of receptor clustering (66, 67). However, events in the entry process subsequent to virus uptake via endocytosis, but not by fusion, were dependent on intact microtubules, implicating the microtubule component of the cytoskeleton in a postpenetration step (67).

Previous studies have implicated the host cell cytoskeleton in late events in the virus life cycle. The release of HIV-1 from polarized cells was shown to be dependent on an intact cytoskeleton (68), whereas inhibitors of myosin light-chain kinase and actin polymerization inhibited the release of HIV-1 from the infected cells (69). Polymerized actin has been implicated in directional budding of HIV-1 from infected monocytes as evidenced by colocalization of actin and gag CA protein at the sites of virus release (23). A role for the cytoskeleton in virus release was also suggested by the observation that HIV-1 gag polyprotein associated with polymerized actin (47) and that HIV-1 virions contain actin in a 10% molar ratio relative to gag polyproteins (70). The accessory gene product, Vif, which has an important role in the regulation of virus infectivity, has been proposed to regulate virus maturation by mediating interaction with the cellular cytoskeleton (71). The role of gag–cytoskeleton interaction in virus maturation may be paradoxical with the potential role of gag MA–cytoskeletal interaction during virus entry. Presumably, the interaction between gag precursor proteins and the cytoskeleton at late stages in the virus life cycle would favor accumulation of virion precursor proteins at the cell surface, which is necessary for efficient budding from the infected cell. However, as shown in this study, at early stages in the virus life cycle this interaction appears to be promoting the establishment of infection after virus uptake. Our data suggest that phosphorylated gag MA proteins of the virion associate with the cytoskeleton. This differential phosphorylation likely dictates the fate of the replication intermediates that have associated with the cytoskeleton. We have previously suggested a model in which gag MA phosphorylation regulates the opposing targeting properties of this protein during virus assembly and during virus entry (10, 14).

Recent studies have suggested that nuclear targeting of incoming human foamy virus proteins was dependent on the presence of intact microtubules (72). Similar to the localization of viral reverse transcription complexes to actin microfilaments outlined here, the study of Saib et al. (72) demonstrated aggregation of viral reverse transcription complexes on microtubules. That study also suggested that virus entry was not dependent on microtubules but that a subsequent step, either at the level of reverse transcription or nuclear import, was microtubule dependent and highly sensitive to colchicine, a microtubule-disrupting agent. These results are reminiscent of observations in other viral systems such as herpes simplex, where transport of incoming viral capsids to the nucleus is mediated by microtubules (73). Intriguingly, recent studies in our laboratory (our unpublished results) suggest that intermediate filaments, another component of the cytoskeleton, do not appear to be involved in events leading up to completion of viral cDNA synthesis on the cytoskeleton but may be important for subsequent translocation of nascent viral reverse transcription complexes to the host cell nucleus.

We thank M. Emerman for the pSfi clone and the VSV-G expression vector, S. Doxsey for guidance in the cytoskeletal analysis, and L. Henderson for the rabbit anti–HIV-1 gag MA antisera. Magi cells, 293T cells, and MT-4 T cells were provided by the National Institutes of Health AIDS Research and Reference Reagent Program (National Institute of Allergy and Infectious Diseases, Bethesda, MD).

This work was supported by grants from the National Institutes of Health and the Fogarty Program to M. Stevenson.

CA

capsid

CCA

CCD, and CCE, cytochalasin A, D, and E, respectively

FBS

fetal bovine serum

MA

matrix

MLV

murine leukemia virus

VSV

vesicular stomatitis virus