|Home | About | Journals | Submit | Contact Us | Français|
After cell entry, HIV undergoes rapid transport toward the nucleus using microtubules and microfilaments. Neither the cellular cytoplasmic components nor the viral proteins that interact to mediate transport have yet been identified. Using a yeast two-hybrid screen, we identified four cytoskeletal components as putative interaction partners for HIV-1 p24 capsid protein: MAP1A, MAP1S, CKAP1, and WIRE. Depletion of MAP1A/MAP1S in indicator cell lines and primary human macrophages led to a profound reduction in HIV-1 infectivity as a result of impaired retrograde trafficking, demonstrated by a characteristic accumulation of capsids away from the nuclear membrane, and an overall defect in nuclear import. MAP1A/MAP1S did not impact microtubule network integrity or cell morphology but contributed to microtubule stabilization, which was shown previously to facilitate infection. In addition, we found that MAP1 proteins interact with HIV-1 cores both in vitro and in infected cells and that interaction involves MAP1 light chain LC2. Depletion of MAP1 proteins reduced the association of HIV-1 capsids with both dynamic and stable microtubules, suggesting that MAP1 proteins help tether incoming viral capsids to the microtubular network, thus promoting cytoplasmic trafficking. This work shows for the first time that following entry into target cells, HIV-1 interacts with the cytoskeleton via its p24 capsid protein. Moreover, our results support a role for MAP1 proteins in promoting efficient retrograde trafficking of HIV-1 by stimulating the formation of stable microtubules and mediating the association of HIV-1 cores with microtubules.
During their replication cycle, most enveloped viruses must move from the cell surface membrane to the compartments containing the machinery for their replication and back to the plasma membrane for assembly and release. Passive diffusion through the cytoplasm is very ineffective and rarely leads to the required destination. Therefore, viruses have evolved mechanisms enabling them to “hijack” the cellular cytoskeleton, normally destined for the transport of cellular proteins and vesicles, for their active transport within infected cells (1). During the early phase of HIV-1 infection, incoming virions fuse with the host cell membrane, allowing the viral complex referred to as the reverse transcription complex (RTC)3 to be transported through the cytoplasm toward the nucleus. After crossing the cortical actin meshwork, HIV-1 uses both the microtubule and microfilament networks to traffic toward the nucleus (2, 3). Genome-wide RNA interference-based screens have identified numerous tubulin- and actin-associated proteins that may contribute to efficient infection (4,–6). However, as HIV interacts with the cytoskeleton at multiple points in its replication cycle and remodels the cytoskeleton to promote viral replication, it is unclear which cellular factors are specifically involved in trafficking.
The nature of the HIV RTC that traffics along the cytoskeleton to reach the nucleus is currently poorly characterized. After viral cell entry, the capsid structure of HIV disassembles to allow infection to proceed; however, the timing and location of this uncoating are still uncertain. Work based on capsid mutants or chimeras shows a functional link between HIV-1 capsid and nuclear import via the nuclear pore (7,–10), and HIV-1 cores have been shown to dock at the nuclear pore and interact directly with Nup358/RanBP2 (11,–13). Evidence further suggests that uncoating is closely coordinated with the reverse transcription of the viral genome. Acceleration of TRIM5α-mediated uncoating correlates with inhibition of reverse transcription (14,–17), mutations in capsid proteins that affect the stability of the HIV-1 core interfere with reverse transcription (18, 19), and inhibition of reverse transcription delays or prevents uncoating (11, 20). Recent work shows that the stability of the HIV capsid is key in preventing the sensing of viral DNA by cyclic GMP-AMP synthase (cGAS) (21, 22) and in dictating susceptibility to Mx2 after successful reverse transcription (23,–25), suggesting that the capsid remains intact until the viral DNA is fully synthesized, which is estimated to occur at 4–12 h post-infection and up to 16–24 h in primary macrophages (11, 26,–28). However, there is also evidence that uncoating or partial uncoating can happen earlier, within 30–45 min of infection (20, 29), which may be key in establishing productive infection, for instance by eliminating suboptimal complexes or, on the contrary, priming complexes for full uncoating at a later time.
Although some uncoating or partial uncoating can happen at the early post-infection stage, the importance of the viral capsid for efficient reverse transcription and its presence at the nuclear pore implies that many complexes that traffic along the cytoskeleton toward the nucleus still contain capsid. Because the capsid sequesters all other viral components remaining after cell entry until full uncoating, we hypothesized that the viral component that interacts with the cytoskeleton for active transport toward the nucleus might be the viral capsid. Here, we report the interaction of HIV-1 capsid cores with human MAP1A and MAP1S, two microtubule-associated proteins. This suggested a potential role for MAP1A and MAP1S in HIV-cytoskeleton interactions and viral transport, which we decided to investigate further. Depletion of MAP1A or MAP1S in indicator cell lines and primary human macrophages disrupted HIV-1 infection by hampering the efficient trafficking of incoming HIV-1 complexes to the nucleus. MAP1A/MAP1S did not impact microtubule network integrity or cell morphology but did stabilize microtubules in primary human macrophages. HIV-1 capsid association with both dynamic and stable microtubules was lost upon MAP1 depletion, suggesting that MAP1 proteins facilitate the tethering of HIV-1 complexes to microtubules. Our work shows for the first time that following entry into target cells, HIV-1 interacts with the cytoskeleton via its p24 capsid protein, supporting a role for MAP1A proteins in promoting the efficient retrograde trafficking of HIV-1 by stimulating the formation of stable microtubules and mediating their association with HIV-1 cores.
The p24 constructs were cloned into pDONR207, fused to the GAL4 DNA binding domain in pGBKT7 (pGBKT7-p24), and used to screen the human brain Matchmaker cDNA library (Clontech). The complexity of this cDNA library was about 5 × 106 independent inserts with an average size of 1.5 kb. The cDNA library was cloned into the pACT2 vector (Clontech), and then transformed into the Y187 yeast strain (MATα ura3–52 his3–200 ade2–101 trp1–901 leu2–3112 gal4Δ mel gal80Δ URA3::GAL1UAS-GAL1TATA-lacZ) using a standard large scale transformation procedure to obtain about 107 individual yeast colonies corresponding to a 2-fold coverage of the cDNA library. In parallel, pGBKT7-p24 constructs were established in the AH109 yeast strain (MATα trp1–901 leu2–3112 ura3–52 his3–200 gal4Δ gal80Δ LYS2::GAL1UAS-GAL1TATA-HIS3 GAL2UASGAL2- TATA-ADE2 URA3::MEL1UAS-MEL1TATA-lacZ). p24 baits were used to screen the cDNA library by mating. A mix of AH109 and Y187 yeasts were plated for 4 h at 30 °C on solid medium containing yeast extract, peptone, and dextrose at pH 3.5 (YCM). Mated yeast cells were grown for 5 days on synthetic dextrose medium lacking tryptophan, leucine, and histidine and containing 20 mm 3-aminotriazole to select for diploids that showed elevated expression of the GAL1::HIS3 Y2H reporter. The prey cDNAs corresponding to HIS3 positive colonies were PCR amplified and sequenced, and the resulting sequences were analyzed with BLAST.
P4-CCR5 reporter cells used were HeLa CD4+ CXCR4+CCR5+ carrying the LacZ gene under the control of the HIV-1 LTR promoter (30). The 293T cells are human embryonic kidney cells (ATCC). Citrate human blood was obtained from healthy donors (Etablissement Français du Sang). Monocytes were isolated using adherence on plastic (45 min) after Ficoll gradient and were differentiated to macrophages by incubation for 7–10 days with 50 ng/ml GM-CSF (Gentaur).
LAI and NL43 human immunodeficiency viruses were produced by transient transfection of 293T cells using calcium phosphate precipitation with a wild-type env proviral plasmid or Δenv construct co-transfected with the vesicular stomatitis virus glycoprotein (VSV-G) envelope expression plasmid pHCMV-G (31). Viruses were harvested at 48 h after transfection. The virus yield was measured by p24 ELISA according to the manufacturer's instructions (PerkinElmer Life Sciences).
Complementary oligonucleotides coding for shRNAs were annealed and cloned into BglII/HindIII of pSUPER (OligoEngine) downstream of the H1 promoter. The H1-shRNA cassette was then inserted into a polylinker within the U3 region of the TRIP-CMV-eGFP vector (32). Lentiviral vector (LV)-shRNAs were produced by transient co-transfection of 293T cells with the vector, encapsidation (pCMVΔR 8.74), and VSV-G plasmids. Vectors were harvested 48 h post-transfection and concentrated by ultracentrifugation for 1 h at 64,000 × g (Beckman Coulter) at 4 °C. LV-shRNAs were titered in P4-CCR5 cells using flow cytometry to assess GFP expression at 4 days post-transduction (dpt). P4-CCR5 cells and primary human macrophages (4 × 106) were transduced with LV-shRNA at m.o.i. 50. All knockdown (KD) cells were used at 3 dpt.
P4-CCR5 cells were reverse transfected with final concentrations of 10 and 50 nm siRNA (Dharmacon) in a final volume of 600 μl. Scramble siRNA was the Dharmacon ON-TARGET plus non-targeting pool. Briefly, 6 and 30 pmol of siRNAs were diluted in 90 μl of Opti-MEM. Then, Lipofectamine RNAiMAX (Invitrogen) was diluted 1:10 in Opti-MEM, and 10 μl was added to the 90 μl of diluted siRNAs. The transfection mix was incubated at room temperature for 20 min in a 24-well plate. P4-CCR5 (30,000) cells were diluted in 500 μl of DMEM plus 10% FCS without antibiotics, and 500 μl of cell suspension was added to the siRNA mix. Incubation was for 48 h at 37 °C.
At 48 h after reverse transfection with siRNAs, or 72 h after transduction with LV-shRNAs, total RNA was isolated from P4-CCR5 cells using an RNeasy mini kit according to the manufacturer's instructions (Qiagen). RT-PCR was carried out by using a One-step RT-PCR kit (Qiagen) according to the manufacturer's instructions.
A β-galactosidase assay was performed 48 h post-infection (hpi) in indicator P4-CCR5 cells according to the manufacturer's instructions (Roche Applied Science). β-Galactosidase activity was normalized for protein concentration by the Bradford assay.
Total cellular DNA was isolated 24 hpi using the QIAamp DNA micro kit (Qiagen). Two long-terminal repeat (2-LTR)-containing circles were detected with primers MH535/536 and probe MH603 (33), using as the standard curve the pUC2LTR plasmid, which contains the HIV-1 2-LTR junction. Reactions were normalized by amplification of the late reverse transcript with primers MH531/532 and probe LRT-P (33). Alu-PCR was performed according to a protocol adapted from Brussel and Sonigo (34) as published previously (12).
The primary antibodies used were mouse monoclonal anti-β-actin, mouse monoclonal anti-α-tubulin, and acetylated tubulin (Sigma), rabbit polyclonal anti-detyrosinated tubulin (EMD Millipore), mouse monoclonal anti-p24 clone 183-H12-5C (NIH AIDS Reagent Program), rabbit anti-cyclophilin A (Cell Signaling, Denver, CO), mouse monoclonal anti-FLAG, rabbit polyclonal anti-MAP1A, and -MAP1S, and rat monoclonal anti-tubulin (YL1/2) (Abcam). Secondary antibodies were goat anti-mouse and anti-rabbit HRP conjugates (Thermo Fisher Scientific, Rockford, IL), goat anti-rabbit IgG (H+L) Alexa Fluor 488 (Molecular Probes), and goat anti-mouse Cy3 (GE Healthcare). Actin cytoskeleton was labeled with rhodamine phalloidin (Molecular Probes) staining, and nuclei were stained with Hoechst (Invitrogen).
Cultures were rinsed with PBS and fixed with 4% paraformaldehyde in PBS for 10 min at room temperature. They were then permeabilized with 0.5% Triton X-100 in PBS three times for 10 min. The same solution was used for all subsequent washing steps. Cells were incubated with primary antibodies for 1 h at 37 °C. After incubation, cells were washed for 30 min and incubated with Alexa Fluor- or Cy3-conjugated secondary antibodies for 1 h at 37 °C, and the nuclei were labeled with Hoechst dye.
The image acquisition of P4-CCR5 was carried out using a confocal microscope (Zen 2012 LSM700, Zeiss) with a Plan-Apochromat ×63/NA 1.4 objective, and 16-bit z-stack random field images were acquired as 1024 × 1024 pixels with a line averaging of 2. To perform a statistical analysis of capsid position in primary macrophages, we used the Cell Voyager spinning disk confocal system (CV1000, Yokogawa) for high-throughput mosaic analysis with an Olympus microscope (UPLSAPO ×40/NA 1.3). The sample area was acquired as a square of 10 × 10 fields of view (920 × 920 pixels for each) for each condition. In total ~600–700 macrophages were acquired per condition, and only those that were GFP positive (i.e. shRNA transduced) were analyzed, corresponding to ~10% of the cells. In these two acquisition approaches, three signals, Hoechst, eGFP, and Cy3, were detected over multiple z-slices (z = 8–12). For each z-position, the CV1000 reconstructed mosaic or random field from LSM700 image analysis was carried out using Acapella software (PerkinElmer Life Sciences). Cells were analyzed over a constant volume of four z-slices for P4-CCR5 and five z-slices for macrophages, selected around the focal plane of the center of each nucleus. The script was subdivided into three object subroutines segmenting successively the nucleus, the cytoplasm, and the spots within the perinuclear regions of each cell by z-position. Statistical analyses (one-way ANOVA with Dunnett's multiple comparisons test) were performed using Prism 6 (GraphPad).
The pCiNeo3×FLAG plasmid was used for cloning FLAG fusions of candidate proteins under the control of the CMV promoter. The cDNA of candidate p24 CA interactors were transferred from the pENTR plasmid to the pCiNeo3×FLAG plasmid as a fusion dowstream of the 3×FLAG motif by in vitro recombination (LR cloning reaction, Gateway Technology, Invitrogen). Cells were either transfected at 60–80% confluency with FLAG pCiNeo3× constructs or left untransfected. After 24 h the cells were lysed in ice-cold cell lysis buffer (5 mm Tris, pH 8, 150 mm NaCl, 5 mm EDTA, and 0.01% Nonidet P-40) for 40 min on ice. The lysate was then briefly vortexed, and cell debris was cleared by centrifugation at 13,000 rpm at 4 °C for 30 min. HIV-1 capsids were prepared by centrifugation through a sucrose cushion by overlaying on ice in 1 ml of 50% sucrose (PBS) with 1 ml of 10% sucrose plus 0.1% Triton X-100. As a negative control, a sucrose cushion without Triton was produced. Viral supernatants (500 μl) were layered onto the 10% phase and centrifuged at 32,500 rpm for 2 h at 4 °C (SW55Ti rotor). After centrifugation, the presence of intact capsids (in the pellet) and soluble capsid (in 250 μl from the top of the supernatant) was measured by p24 ELISA. The presence of intact cores in the pellet was confirmed by negative staining followed by transmission electron microscopy (data not shown). Clarified cell lysate (120 μl) was incubated with capsids (100 μl) for 1 h on ice. Capsids were then ultracentrifuged at 15,000 rpm (TLS 55 rotor) at 4 °C for 30 min. After centrifugation the pellets were resuspended in 150 μl of ice-cold PBS and analyzed by Western blotting.
Paraformaldehyde-fixed P4-CCR5 cells were preincubated with blocking agent for 1 h. After washing in PBS for 10 min, cells were incubated with primary anti-MAP1A and anti-p24 antibodies for 1 h in a preheated humidity chamber at 37 °C. Slides were washed three times in PBS for 10 min and then they were incubated with Duolink proximity ligation assay probes (Olink Bioscience) diluted in the blocking agent (1:5) for 1 h in a preheated humidity chamber at 37 °C. Ligation of the connector oligonucleotides, rolling circle amplification, and detection of the amplified DNA products were done with Duolink Detection Reagents Red according to the manufacturer's instructions. Nuclei were labeled with Hoechst dye.
As a result of yeast two-hybrid screens that aimed at mapping HIV-host interactions at the protein level, four cytoskeleton-associated factors, namely MAP1A, MAP1S (BPY2), CKAP1 (CoB), and WIRE (WICH/WIPF2), were identified as putative partners of HIV-1 p24 capsid protein (Table 1). MAP1A and MAP1S belong to the same MAP1 family of proteins, which bind along the length of the microtubules with the primary role of stabilizing the microtubules (35). MAP1 proteins are translated as larger proteins that are proteolytically cleaved into heavy chains, MAP1A-HC (350 kDa) and MAP1S-HC (100 kDa), and light chains, MAP1A-LC2 (28 kDa) and MAP1S-LC (26 kDa). CKAP1 (cytoskeleton-associated protein) is a 38-kDa protein that binds to folded α-tubulin and promotes tubulin heterodimer quaternary structure formation, necessary for microtubule polymerization (36). Finally, WIRE belongs to the verprolin family of F-actin-binding proteins that regulate actin filament polymerization following signaling via WASPs (Wiskott-Aldrich syndrome protein) and Arp2/3 (37). This suggested that MAP1A, MAP1S, CKAP1, and WIRE might play a role in HIV-cytoskeleton interactions and viral transport, which we decided to investigate further.
We first asked whether the four identified cytoskeletal proteins participate in HIV-1 infection. We designed specific small interfering RNAs (siRNAs) to assess the impact of protein depletion on infection in single cycle infectivity assays. Multiple siRNAs were tested for each candidate, and the efficiency of knockdown was monitored by reverse transcription PCR. A pool of scramble siRNAs was used as the negative control. Expression of MAP1A, although predominant in the brain, was also detected in P4-CCR5 cells, human primary macrophages, and peripheral blood mononuclear cells and could be inhibited by specific siRNAs (Figs. 1 and and2).2). Depletion of MAP1A from infected cells led to a 10–20-fold decrease in the infectivity of both wild-type env and VSV-G-pseudotyped HIV-1 (Fig. 1). Similarly, knockdown of MAP1S led to a weaker but consistent (~5-fold) reduction of infectivity. The same infectivity profiles were observed with VSV-G-pseudotyped HIV-1 (Fig. 1B), indicating that viral entry via pH-dependent endocytosis, which results in a delivery of viral capsids deeper in the cytoplasm, does not alleviate the dependence of HIV-1 on MAPs for infection. In contrast, WIRE and CKAP1 depletion had no effect on HIV-1 infectivity, whether wild-type env or VSV-G-pseudotyped, despite strong knockdown levels (Fig. 1B). Using these single-cycle infectivity assays, we concluded that MAP1A and MAP1S play a role during the early steps of HIV-1 infection.
Having screened multiple siRNAs for each candidate for efficient knockdown (data not shown), we cloned the most efficient sequence as a small hairpin RNA in an HIV-1-derived lentiviral vector engineered to carry eGFP as the reporter gene (LV-shRNA) (Table 2). This provided controlled knockdown conditions and a straightforward readout for transduction efficiency for all subsequent RNAi experiments. Vectors were titered in target cells, and transductions were carried out at m.o.i. 50; knockdown was assessed by RT-PCR (Fig. 2B) and Western blotting (Fig. 2C). The effect of cytoskeletal protein depletion on cell viability was monitored by an MTT colorimetric assay, which measures mitochondrial activity in living cells. No statistically significant difference in viability was observed between any sample and the non-transduced (NT) control at 4 dpt when the measurements were made (Fig. 2D). Following infection of transduced cells, the same infectivity defects were observed with MAP1A (~10-fold) and MAP1S (~5-fold) shRNAs as previously with siRNAs, and similarly, CKAP1 and WIRE targeting shRNAs had no effect on HIV-1 infection, with wild-type env (Fig. 2E) and VSV-G-pseudotyped (Fig. 2F) HIV-1.
Having shown that MAP1A and MAP1S contribute to the efficiency of HIV-1 infection in single-cycle infectivity assays, we sought to identify the specific step in which they are implicated. To assess whether MAP1 proteins affect reverse transcription, we quantified late reverse transcripts (LRT) in KD cells at 24 hpi. As controls, the cells were either transduced with LV-shLuc or non-transduced. The reverse transcription inhibitor nevirapine (5 μm) was also included. In independent experiments, no statistically significant difference in LRT levels was observed between any of the samples when compared with shLuc, indicating comparable efficiencies in reverse transcription (Fig. 3A). A previous study shows that two microtubule-associated proteins, DNAL1 and MAP4, impact HIV-1 reverse transcription when depleted (38). However, in two independent experiments, siRNA-mediated depletion of DNAL1 and MAP4 had no statistically significant effect on HIV-1 reverse transcription when compared with scramble control (Fig. 3A). MAP1A and MAP1S did not contribute to the reverse transcription efficiency of HIV-1, arguing that HIV-1 trafficking to the nuclear pore and reverse transcription are independent processes (11, 12).
To assess whether MAP1 proteins affect nuclear import or integration, we quantified 2-LTR circles, which are an exclusively nuclear form of viral DNA and can therefore be used as a marker of nuclear import. We found that both MAP1A and MAP1S depletion led to a reduction in 2-LTR circles in infected cells at 24 hpi (2–4-fold), whereas WIRE and CKAP1 knockdown had no effect (Fig. 3B). This decrease was also seen when quantifying integrated HIV-1 proviruses by Alu-PCR (Fig. 3C), indicating a reproducible and statistically significant decrease in nuclear import. In the case of MAP1A, the measured defect in nuclear import was as low as expected when compared with the defect in infectivity. Indeed, we had noted previously that residual nuclear import (up to 10% of wild-type levels) is detected even for HIV-1 viruses that are unable to replicate in primary lymphocytes, regardless of the technique used to quantify nuclear viral DNA (Southern blotting or quantitative PCR) (39), which may account for an underestimation of the impact of MAP1A depletion on nuclear import.
Finally, we asked whether the depletion of MAP1 proteins leads to an overall defect in HIV-1 genome nuclear import or merely delays it, as might be expected to happen if viral trafficking could still occur without functional microtubular transport. We monitored 2-LTR circles over a 72-h time course post-infection. The proportional defect in the 2-LTR copy number remained the same in MAP1A- and MAP1S-depleted cells over the whole time course (Fig. 3D), indicating an overall defect in nuclear import rather than a delay.
Because MAP1A and MAP1S contribute to efficient HIV-1 nuclear access and are associated with microtubules, we asked whether MAP1 proteins might contribute to efficient trafficking of incoming HIV-1 to the nuclear membrane. We labeled infected KD cells with an anti-p24 antibody at 8 hpi (m.o.i. ~50), a time point at which viral complexes have uncoated or are in the process of uncoating and entering the nucleus (26). Lower m.o.i. values (m.o.i. 1–5) have been used with similar results; however, they do not produce a sufficient capsid signal to allow statistical analysis. LV-shRNA transduction efficiency was monitored by GFP expression from the internal cassette. Image acquisition was performed using confocal microscopy of multiple z-stack images of random fields and analyzed using an Acapella script (Fig. 4A). Overall, less p24 labeling was observed in control samples (NT and shLuc), suggesting that uncoating has already been initiated. In contrast, MAP1A and MAP1S depletion led to an accumulation of p24 labeling throughout the cytoplasm, suggesting an inability both to reach the nuclear membrane and to uncoat (Fig. 4B). Particularly, in MAP1A KD cells, capsid labeling formed characteristic clumps scattered throughout the cytoplasm, away from the nucleus. A similar accumulation was not apparent in the case of CKAP1 and WIRE depletion. Acapella quantification revealed a statistically significant decrease in the number of perinuclear spots per cell volume and perinuclear spot density in the case of MAP1A depletion, indicating that MAP1A knockdown leads to a reduced presence of HIV-1 capsid in the perinuclear area, regardless of the cell volume or output measured (Fig. 4C).
To determine the earliest time point at which the absence of MAP1 proteins leads to a defect in HIV-1 trafficking and ascertain whether the defect persists with time, we performed a time course experiment by localizing the HIV-1 capsid signal at different time points post-infection in KD and control cells. We acquired images at 30 min and at 2, 4, 8, and 24 hpi. Results show that the defect in HIV-1 trafficking to the nuclear membrane is apparent in the absence of MAP1A and MAP1S as early as 4 hpi and persists at 24 hpi, indicating that the defect is not a delay but an impediment in trafficking (Fig. 5). The p24 signal observed throughout the cytoplasm of MAP-depleted cells at late time points post-infection suggests that complexes not only failed to traffic toward the nucleus but also had impaired uncoating. This may imply that complexes that fail to traffic to the nuclear pore, where the capsid can interact physically and functionally with nucleoporins and karyopherins, do not achieve proper uncoating. Alternatively, the p24 signal seen in MAP-depleted cells may be aggregates rather than individual cores, suggesting that a defect in trafficking to the nucleus may lead to core aggregation that might prevent efficient uncoating.
To confirm the implication of MAP1 proteins in HIV-1 trafficking in a physiological target cell of HIV-1, we knocked down their expression in primary macrophages. Knockdown was confirmed by Western blotting (Fig. 2C). Similarly to P4-CCR5 cells, depletion of MAP1A and MAP1S led to an accumulation of p24 signal in the cytoplasm (Fig. 6A), confirming that MAP1A and MAP1S contribute to efficient trafficking of HIV-1 to the nuclear membrane. To perform a statistical analysis of the capsid position in primary macrophages, which have lower culture density and permissiveness to infection than P4-CCR5 cells, we used a spinning disk confocal microscope combined with powerful Acapella scripts for high-throughput mosaic analysis (Fig. 4A). Because of the important number of fields acquired for each condition and the high performance computing required, we focused our analysis on MAP1S knockdown and control primary macrophages. We infected with HIV-1 primary macrophages transduced either with control or shMAP1S vector and labeled the infected cells for capsid localization. For each condition, 100 fields were acquired as non-overlapping z-stack mosaics in channels enabling the visualization of nuclei (Hoechst), transduced cells (GFP), and capsid. Acapella quantification revealed a statistically significant decrease in the number of perinuclear spots per cell volume, as well as perinuclear spot density in the case of MAP1S depletion, confirming in primary macrophages that MAP depletion leads to a reduced presence of HIV-1 capsid in the perinuclear area (Fig. 6B). Taken together, these experiments suggest that MAP1A and MAP1S contribute toward efficient trafficking of HIV-1 complexes toward the nucleus.
To address how MAP1 proteins facilitate transport of HIV-1 RTCs to the nucleus, we first asked whether they affect the overall architecture of microtubule and microfilament networks, because loss of cytoskeletal network integrity would severely hamper infection. We labeled microtubules and microfilaments to assess their polymerization in KD cells and quantified cell morphology to test for possible structural collapse. No difference in cytoskeletal integrity was observed compared with non-transduced cells or shLuc-transduced cells (Fig. 7A), suggesting an underlying redundancy in the function of these proteins in the regulation of cytoskeletal polymerization. We noted, however, that WIRE-depleted cells were larger than control and other KD cells with an increased cytoplasmic volume of 65% but a constant nuclear volume (Fig. 7B). With the exception of this change, we concluded that the inhibition of HIV-1 infection following MAP1A/MAP1S depletion is not caused by the disruption of cytoskeletal trafficking that would result from an abnormal or depolymerized cytoskeleton.
Recent work has shown that HIV-1 infection induces, within 1–2 hpi, tubulin post-translational modifications that are associated with the formation of a more stable subset of microtubules, which facilitates the translocation of HIV particles across the cytoplasm (40). We therefore asked whether MAP1 proteins might promote microtubule stabilization in infected cells and thus facilitate overall retrograde trafficking of HIV-1. First, we tested the ability of HIV-1 to induce microtubule stability in primary macrophages, using as markers the prevalent post-translational modifications on stable microtubules, acetylation and detyrosination (41). Post-translational modifications characteristic of stable microtubules increased in primary human macrophages from three independent donors upon infection by both HIV-1 and VSV-G-pseudotyped HIV-1 (data not shown), ranging from 20 to 250% increase depending on experiments, a result concordant with previous findings (40) (Fig. 7C). Similarly, the protein levels of the microtubule plus end tracking protein (+TIP) EB1, which is required for microtubule stabilization and HIV-1 infection (40), were induced by HIV-1 infection (Fig. 7C).
As MAPs can alter the dynamic behavior of microtubules (35), we next asked whether MAP1A/MAP1S might contribute to microtubule stabilization. In both uninfected and infected cells, the knockdown of MAP1A and MAP1S led to a reproducible decrease in markers of stable microtubules by 2–3-fold on average, indicating that MAP1 proteins contribute, albeit modestly, to the stabilization of microtubules in primary human macrophages (Fig. 7C). To determine whether the microtubule-stabilizing properties of MAP1 proteins could alone account for their impact on HIV-1 replication, we asked whether the infectivity defect observed in MAP1-depleted cells could be rescued in the presence of taxol, a drug that stabilizes microtubules and increases α-tubulin acetylation (42). LV-shRNA-transduced cells were treated with increasing doses of taxol and infected with HIV-1. The results show that microtubule stabilization in the presence of taxol does not rescue infection in MAP1-depleted cells (Fig. 7D). Taken together, these results suggest that the ability of MAP1 proteins to stimulate the formation of stable microtubules is unlikely to contribute to retrograde trafficking of HIV-1.
As MAP1 proteins were identified in yeast two-hybrid screens using p24 protein as bait, we next asked whether they might be able to bind to HIV-1 RTCs and thus mediate their association with microtubules. HIV-1 uncoating has been shown to occur within 30–45 min of infection (20, 29) or up to 4–12 h post-infection (11, 28) depending on studies and cell type, indicating that RTCs that latch onto the cytoskeleton in the minutes following cell entry still contain the capsid shell. The four candidate cytoskeletal protein coding sequences were cloned into pCi-Neo-3×FLAG plasmids for overexpression in 293T cells. In the case of MAP1A, the clones identified by our yeast two-hybrid screen all covered regions of the light chain, LC2, indicating that LC2 is the binding target of p24 (Table 1). This domain, which contains both microtubule and F-actin binding sequences, constitutes a good candidate based on our previous work that suggests that HIV-1 uses both microtubules and actin filaments for retrograde transport (3). We isolated HIV-1 cores using ultracentrifugation through a sucrose cushion in a protocol adapted from Shah and Aiken (43) and incubated them with cell lysates prepared from transfected 293T cells. To monitor for the nonspecific pulldown of proteins, lysates alone were ultracentrifuged without cores. The capsid-binding protein CypA was included as a positive control and the mRNA-binding protein hnRNP E1 as a negative control. The results showed a strong interaction for MAP1A-LC2 and MAP1S with isolated HIV-1 cores (Fig. 8A). A weaker interaction with CKAP1 and WIRE was also detected. These results indicate that both MAP1 proteins interact with the HIV-1 cores, suggesting that they can serve as anchor points on microtubules for HIV RTCs.
To confirm the interaction of HIV-1 capsid with MAP1A in infected cells, we performed a Duolink in situ proximity ligation assay. NT and shMAP1A-transduced cells, infected or not with HIV-1, were probed with anti-p24 and anti-MAP1A primary antibodies followed by species-specific proximity ligation assay probes. As expected, no signals were observed in uninfected control cells, and few signals were detected in the MAP1A-depleted cells. Our results show an interaction of MAP1A with viral capsid in HIV-1-infected cells, confirming the interaction at a cellular level in the context of infection (Fig. 8B). Similar results were obtained following infection with VSV-G-pseudotyped HIV-1 (data not shown).
Because MAP1 proteins associate with both microtubules and HIV-1 capsid, we asked whether MAP1 proteins might mediate the interaction between HIV-1 capsids and microtubules. We labeled infected cells with p24 antibody and probed for co-localization with both dynamic and stable microtubules. Close proximity events, highlighted by a yellow signal where green and red channels overlap, were readily observed in control infected cells, indicating that HIV-1 cores can associate with both dynamic (Fig. 8C) and stable microtubules (Fig. 8D). In contrast, the proximity of capsid and microtubules was rarely observed in MAP1A-depleted cells, suggesting that capsids overlap less well with microtubules in the absence of MAP1 proteins (Fig. 8, C and D). Similar results were obtained following infection with VSV-G-pseudotyped HIV-1 (Fig. 8, C and D). Taken together, our results support a model in which MAP1 proteins promote HIV-1 retrograde trafficking by tethering HIV-1 capsids to microtubules.
After cell entry, HIV-1 is propelled toward the nucleus by cytoskeleton-directed trafficking (44). Although trafficking along microtubules has been characterized in terms of velocity and movement dynamics (2, 3), the nature of the interactions between the HIV RTC and the host cytoskeleton that mediate retrograde trafficking are relatively unknown. This study reports the importance of microtubule-associated proteins MAP1A and MAP1S for efficient HIV-1 infection and retrograde trafficking. We found that MAP1 proteins interact with HIV-1 cores in vitro and in infected cells and that depletion of MAP1 proteins reduces the association of HIV-1 capsids with both dynamic and stable microtubules. Taken together, our data suggest that MAP1 proteins help tether incoming viral capsids to the microtubular network, thus promoting cytoplasmic trafficking.
MAP1 proteins bind along the length of microtubules, and an in situ proximity ligation assay showed the association of HIV-1 capsid and MAP1 proteins throughout the cell cytoplasm, suggesting that interaction with MAP1 proteins is likely to occur repeatedly during transport along microtubules and not just as an initial latching onto the cytoskeleton at the cell periphery. We showed previously that HIV-1-directed transport along microtubules toward the nucleus is saltatory (3). Intermittent transport along microtubules suggests binding to and detachment from filaments. MAP1 proteins may provide important anchor points for HIV-1 complexes traveling toward the nucleus, enabling reattachment following dissociation.
Following infection, HIV-1 complexes undergo uncoating (loss of capsid) prior to nuclear entry. Some uncoating or partial uncoating occurs in early post-infection in the cell periphery. However the importance of capsid for efficient reverse transcription, as well as its presence deep in the cytoplasm and at the nuclear pore, implies that many complexes that traffic along the cytoskeleton toward the nucleus still contain capsid. Our work suggests that the viral structure that interacts with the cytoskeleton for trafficking to the nucleus is the viral capsid protein.
MAP1 proteins are thought to stabilize microtubules by altering their intrinsic dynamic instability (35). Here, MAP1 did not impact microtubule network integrity or cell morphology but contributed to microtubule stabilization, which was shown previously to facilitate infection (40). Therefore, although the contribution of MAP1 to microtubule stabilization is modest in primary human macrophages, it is likely that their ability to promote microtubule stabilization also contributes to their overall effect on HIV-1 trafficking.
The cytoskeletal proteins CKAP1 and WIRE, which are two other potential candidates identified as HIV-1 p24 binding partners in the yeast two-hybrid screen, were not found to be involved in the early phases of infection. This study does not exclude their involvement in the later phases of the cycle or redundancy with other family proteins for their function.
Evidence suggests that HIV-1 trafficking may involve both microtubules and microfilaments (2, 3, 45, 46). The importance of both actin filaments and microtubules for HIV-1 infection has also been highlighted by a series of siRNA genome-wide screens for cellular factors necessary for infection (4,–6), which identified a number of proteins that regulate the polymerization and stability of microfilaments (e.g. gelsolin, spectrin, and cofilin) and microtubules (e.g. calpain, microtubule-associated protein, and kinesin superfamily proteins). Interestingly, MAPs have been reported to facilitate microtubule-actin cross-talk, and the interaction of HIV-1 cores with MAP1A is mediated by its light chain, LC2, which contains both microtubule and microfilament binding domains. Therefore, MAP1A and MAP1S, in which the light chains are 39% homologous, may be involved in the transition between the actin cortex and microtubules immediately following cell entry or between microtubules and microfilaments around the nuclear membrane before docking at the nuclear pore (3). Indeed, an inability to engage onto microfilaments might result in an anterograde movement back toward the periphery, which might account for the scattering of HIV-1 complexes that we observed throughout the cytoplasm in KD cells.
Alternatively, MAPs have been reported to interact with molecular motors and interfere directly with microtubule or cargo binding or affect their overall function (47). Although MAP1A and MAP1S can localize along the length of microtubules, the different localization densities of MAP1A and MAP1S along microtubular tracks (47) suggest that they might participate in cargo engagement onto microtubules at the cell periphery or in the drop-off of cargo by molecular motors near the nuclear membrane. Although the molecular motor(s) involved in HIV-1 retrograde trafficking has not yet been identified, evidence suggests that dynein might be involved; treatment with the dominant negative inhibitor p150 CC1 domain of dynactin, which uncouples dynein-based transport, or microinjection of anti-dynein antibodies leads to a clustering of HIV-1 in the cell periphery (2, 3, 48). Moreover, the detected speeds and directionality of incoming HIV-1 (>1 μm/s (3)) suggest dynein-mediated microtubular movement. In the case of spumaviruses, a direct interaction between human foamy virus Gag and LC8, the light chain of dynein, has been demonstrated (48), although another study suggests that interaction with LC8 might be involved in other cellular processes and thus not be specific for retrograde transport (49). No molecular motors were identified in our yeast two-hybrid screen. This could be because of the sheer size and complexity of molecular motors; dynein, for instance, comprises two heavy chains, three intermediate chains, and four light chains and typically functions in conjunction with the dynactin complex, which consists of at least 10 additional proteins. Moreover, because movement is bidirectional (3), it is possible that incoming complexes simultaneously bind to molecular motors of opposite polarity, as has been shown for the herpes simplex virus HSV-1 (50).
Cargo loading onto microtubule tracks, navigation along its route, and subsequent cargo delivery are likely to be regulated by multiple, interdependent mechanisms, such as microtubule polarity, post-translational modifications, and microtubule-associated proteins (47). Our results and previous studies together suggest that HIV-1 may also use multiple mechanisms to regulate its retrograde transport, such as interaction with minus end-directed motors, induction of stable microtubules, and interaction with microtubule-associated proteins (Refs. 40 and 51 and this study).
Taken together, our work reveals that HIV-1 cores interact with MAP1 proteins for trafficking to the nucleus, leading to efficient nuclear import and infection. Further work will shed light on how these proteins participate in the trafficking dynamics and transitions among microtubules, microfilaments, and nuclear pores.
3The abbreviations used are: