LFA-1 Engagement Triggers T Cell Polarization at the HIV-1 Virological Synapse

ABSTRACT HIV-1 efficiently disseminates by cell-cell spread at intercellular contacts called virological synapses (VS), where the virus preferentially assembles and buds. Cell-cell contact triggers active polarization of organelles and viral proteins within infected cells to the contact site to support efficient VS formation and HIV-1 spread; critically, however, which cell surface protein triggers contact-induced polarization at the VS remains unclear. Additionally, the mechanism by which the HIV-1 envelope glycoprotein (Env) is recruited to the VS remains ill defined. Here, we use a reductionist bead-coupled antibody assay as a model of the VS and show that cross-linking the integrin LFA-1 alone is sufficient to induce active T cell polarization and recruitment of the microtubule organizing center (MTOC) in HIV-1-infected cells. Mutant cell lines coupled with inhibitors demonstrated that LFA-1-induced polarization was dependent on the T cell kinase ZAP70. Notably, immunofluorescent staining of viral proteins revealed an accumulation of surface Env at sites of LFA-1 engagement, with intracellular Env localized to a Golgi compartment proximal to the polarized MTOC. Furthermore, blocking LFA-1-induced MTOC polarization through ZAP70 inhibition prevented intracellular Env polarization. Taken together, these data reveal that LFA-1 is a key determinant in inducing dynamic T cell remodeling to the VS and suggest a model in which LFA-1 engagement triggers active polarization of the MTOC and the associated Env-containing secretory apparatus to sites of cell-cell contact to support polarized viral assembly and egress for efficient cell-cell spread. IMPORTANCE HIV-1 causes AIDS by spreading within immune cells and depletion of CD4 T lymphocytes. Rapid spread between these cells occurs by highly efficient cell-cell transmission that takes place at virological synapses (VS). VS are characterized by striking T cell remodeling that is spatially associated with polarized virus assembly and budding at sites of cell contact. Here, we show that the integrin LFA-1 triggers organelle polarization and viral protein recruitment, facilitating formation of the VS, and that this requires the T cell kinase ZAP70. Taken together, these data suggest a mechanism by which HIV-1-infected T cells sense and respond to cell contact to polarize viral egress and promote cell-cell spread. Understanding how cell-cell spread is regulated may help reveal therapeutic targets to specifically block this mode of HIV-1 dissemination.

H uman immunodeficiency virus type 1 (HIV-1) disseminates between T cells either by cell-free infection or by highly efficient cell-cell spread. Cell-cell spread is the predominant mode of HIV-1 dissemination and occurs at virus-induced intercellular contacts known as virological synapses (VS) (1). The HIV-1 VS can be broadly defined as a receptor-containing adhesive junction, characterized by the enrichment of the viral proteins envelope glycoprotein (Env) and Gag in the HIV-infected cell and CD4 and coreceptor (CCR5 or CXCR4) on the target cell, which are collectively polarized at the contact site (1)(2)(3)(4). In addition, adhesion molecules, such as lymphocyte function-associated antigen 1 (LFA-1), intercellular adhesion molecule 1 (ICAM-1), and intercellular adhesion molecule 3 (ICAM-3), are also enriched at the VS. Inhibiting either Env-CD4 or LFA-1-ICAM interactions reduces VS formation and cell-cell spread (2,3,5), suggesting that both sets of receptor-ligand interactions contribute to driving efficient HIV-1 dissemination by contact-mediated spread. However, an outstanding question remains as to whether integrins, as adhesion molecules, serve simply to stabilize the cell-cell contact, allowing subsequent receptor interactions to drive VS formation, or whether they can induce intracellular signaling that facilitates active VS formation, as is the case for the related human T cell lymphotropic virus type 1 (HTLV-1) VS (6).
Viral budding and assembly occur preferentially at the site of cell contact, resulting in highly efficient and rapid infection of the target T cell (1,2,7). Indeed, cell-cell spread of HIV-1 has been shown to be an order of magnitude more efficient than cell-free infection (2,4,5,(8)(9)(10)(11). Additionally, rapid and focused transfer of virions from one cell to another has been shown to reduce the window of exposure of HIV-1 to neutralizing antibodies and may allow evasion of cellular restriction factors or certain antiretroviral therapies (12)(13)(14)(15)(16)(17)(18)(19)(20). Recent intravital microscopy studies have also reported that HIV-1-infected cells show robust migration and form stable cell contacts within a humanized mouse model, providing evidence that cell-cell dissemination could occur in vivo (21)(22)(23). Thus, cell-cell spread confers many advantages on HIV-1 and potentially plays an important role in viral replication within the host.
Contact of a T cell with an antigen-presenting cell (APC) at the immunological synapse (IS) results in T cell polarization characterized by distinct front and rear morphologies (24)(25)(26) and shares some similarities with VS (27). During IS formation, polarization of the microtubule organizing center (MTOC) serves to align the cytoskeleton and to recruit secretory granules and organelles to sites of cell-cell contact (25,26,(28)(29)(30)(31). The VS is associated with striking T cell polarization, with organelles such as mitochondria and the MTOC being found located proximal to viral budding at sites of cell contact (32)(33)(34). Furthermore, it has recently been shown that mitochondria are rapidly recruited to VS in response to cell-cell contact and that this supports efficient HIV-1 spread, as well as dynamic calcium flux that is suggestive of activation of T cell signaling at the VS (33). How HIV-1-infected T cells sense and respond to contact and what receptors trigger polarization at the VS remain unclear but have implications for understanding how HIV-1 disseminates between T cells and for the development of novel therapeutics to specifically target this mode of viral spread.
In this study, we developed an antibody-coupled bead assay as a reductionist model of the VS and used the MTOC as a marker to investigate triggers of polarization at VS. Our results identify the integrin LFA-1 as a mediator of T cell remodeling and show that, rather than simply stabilizing cell-cell contacts, LFA-1 signaling via a ZAP70-dependent pathway induces active T cell polarization at the VS. Furthermore, we found that the MTOC is spatially associated with intracellular Env-containing compartments and that LFA-1 engagement induces copolarization of the MTOC and intracellular Env located within the secretory apparatus. Thus, our results address the hitherto outstanding question of how Env is recruited to the VS for viral assembly and provide new insight into cellular processes regulating HIV-1 spread between T cells.
Antibody-coupled-bead assays. The antibodies used for bead labeling were as follows: anti-CD3 clone OKT3 (eBioscience) (44), anti-LFA-1 ␤ 2 subunit-specific clone L130 (BD Biosciences) (45), anti-LFA-1 ␣ L subunit mouse ascites 25.3.1 (45), and anti-ICAM-1 clone LB-2 (BD Biosciences) (46). Superparamagnetic polystyrene beads (2.8-m diameter) coated with sheep anti-mouse antibody (DynaBeads M-280 sheep anti-mouse IgG) were coupled with primary mouse antibody (0.1 g/10 6 beads; 1/400 for ascites antibodies). The correct coupling of a saturating amount of antibody to the beads was confirmed by flow cytometry. Equal numbers of T cells and beads were incubated in 1% FCS-RPMI on poly-L-lysinecoated coverslips for up to 60 min. For some experiments, T cells were pretreated for 30 min with inhibitors before mixing with the beads (1 M cytochalasin D, 1 M nocodazole, or 10 M piceatannol, all from Sigma-Aldrich). The cell-bead conjugates were fixed using 4% formaldehyde (Sigma-Aldrich) before permeabilization with ice-cold 100% methanol for 5 min or 0.1% Triton X in 5% bovine serum albumin (BSA)-phosphate-buffered saline (PBS) for 20 min at room temperature (only if staining for surface Env). The MTOC was stained with rabbit anti-␥-tubulin (Sigma-Aldrich). For some experiments, intracellular Env was also stained with human anti-gp120 2G12 (Polymun), surface Env was stained with human anti-gp41 (50-69; CFAR), and intracellular Gag was stained with rabbit anti-Gag p24 and p17 (donated by G. Reid and obtained from the CFAR). Primary antibodies were detected with anti-rabbit-conjugated secondary antibodies conjugated to Cy3, Cy5, fluorescein isothiocyanate (FITC), and tetramethyl rhodamine isocyanate (TRITC) (Jackson Immu-noResearch). Coverslips were mounted in Prolong Gold antifade with DAPI (4=,6-diamidino-2-phenylindole) (Life Technologies). To quantify MTOC polarization, single cell-bead conjugates were identified from random fields, and single xy slices through the middle of the cell were extracted from z-stacks taken through the entire volume of the cell. The cells were manually divided into equal thirds in relation to the position of the bead: proximal, middle, and distal. The MTOC was scored as polarized if it was located in the interface-proximal third of the cell. To quantify viralprotein polarization in response to the antibody-coupled beads, the same process was followed. Cells were scored as polarized if there was a significant accumulation of viral protein in the interface-proximal third of the cell. As an additional alternative measure of polarization, the distance of the MTOC from the site of the cell-bead interface in single cell-bead conjugates was measured manually for each conjugate in micrometers using Huygens Professional version 4.0 software. VS preparation and immunostaining. HIV-1-infected primary CD4 T cells were incubated with an equal number of autologous uninfected target T cells that had been loaded with CellTrace CFSE or calcein violet dye (Life Technologies) according to the manufacturer's instructions. Cell contacts were allowed to form at 37°C in 1% FCS-RPMI on poly-L-lysinecoated coverslips for 60 min. The cells were fixed in 4% formaldehyde for 30 min at 4°C. The cell contacts were permeabilized with ice-cold 100% methanol for 5 min according to published methods (1). The primary antibodies were rabbit anti-␥-tubulin (Sigma-Aldrich), mouse anti-␣tubulin (Sigma-Aldrich), mouse anti-Gag p24 clone 313 (donated by G. Reid and obtained from the CFAR, NIBSC, United Kingdom), human anti-gp120 2G12 (Polymun), mouse anti-giantin 9B6 (Abcam), mouse anti-lamp1 H4A3 (developed by J. T. August and J. E. K. Hildreth and obtained from the Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA, USA), mouse anti-EEA-1 clone 14 (BD Biosciences), and rabbit anti-Rab11a (a gift from Scott Lawrence). Primary antibodies were detected with anti-rabbit-or anti-mouse-conjugated secondary antibodies conjugated to Cy3, Cy5, FITC, and TRITC (Jackson ImmunoResearch). Coverslips were mounted in Prolong Gold antifade with DAPI (Life Technologies).
Microscopy. Laser scanning confocal microscopy imaging was performed using a Leica SP confocal microscope, and the images were analyzed using the Leica Application Suite Advanced Fluorescence Lite Version 2.2 and Metamorph V7. Alternatively, immunofluorescence (IF) microscopy was performed using a DeltaVision Elite image restoration microscope (Applied Precision) coupled with an inverted Olympus IX71 microscope and a CoolSnap HQ2 camera. Images were acquired and deconvolved with softWoRx 5.0. Processing and analysis were performed using Huygens Professional version 4.0, ImageJ, and Adobe Photoshop 7.
Colocalization analysis. Single xy slices through the middle of z-stacks of HIV-1-infected and uninfected T cell doublets were taken, and the images were deconvolved using softWoRx. To quantify colocalization, the Pearson correlation coefficient values were calculated using the Im-ageJ coloc 2 plug-in. The plug-in generates a scatter plot showing the pixel-by-pixel intensities of two fluorescent channels, and the Pearson coefficient is generated by dividing the covariances of each channel by their standard deviations. For each staining condition, 20 cell doublets from two independent experiments were analyzed. The Pearson coefficients reported are means and standard errors of the mean (SEM).
Statistical analysis. When comparing normally distributed data under two different conditions, statistical significance was calculated using the Student t test. For multiple comparisons, statistical significance was calculated using the parametric analysis of variance (ANOVA) test with Bonferroni correction. Significance was assumed when the P value was Ͻ0.05. All tests were carried out with GraphPad Prism 6 software.

The integrin LFA-1 triggers MTOC polarization in HIV-1-infected T cells.
To determine the plasma membrane triggers of T cell polarization at the VS, HIV-1-infected Jurkat cells were mixed with dye-loaded primary CD4 target T cells at a 1:1 ratio and incubated at 37°C for 60 min before formaldehyde fixation and antibody staining to visualize the MTOC and viral Gag protein; the latter is a well-defined marker of VS formation (1-3, 7, 33). A VS was defined as a doublet composed of an HIV-1-infected Gagpositive cell and an uninfected target T cell, with Gag enriched at the contact site in the infected cell (Fig. 1A). The MTOC is known to be polarized at VS and was used as a marker for organelle polarization within the HIV-1-infected cell (32)(33)(34) (Fig. 1A). Contacts were scored as polarized if the MTOC was located in the interface-proximal third in the HIV-1-infected cell. Figure 1A confirms that infected cells forming a VS with an uninfected target cell were significantly more likely to have the MTOC polarized toward the contact site than were two uninfected cells in contact (target-target [T-T] control) (71% polarized VS, 12% polarized T-T control contacts; Student's t test; P ϭ 0.0001) (Fig. 1B). Additionally, the distance of the MTOC from the site of cell-cell contact was measured in micrometers and was found to be significantly closer in VS than in T-T controls (2.1-m VS; 4.1-m T-T control contacts; Student's t test; P Ͻ 0.0001) (Fig. 1C), confirming that the MTOC is localized at the VS.
To identify specific triggers of polarization at VS, antibodycoupled beads were used as a reductionist model of the synapse. Magnetic beads were coupled to antibodies specific for receptors of interest and then incubated with HIV-1-infected Jurkat cells, stained, and analyzed by IF microscopy. Single cell-bead conjugates were scored as polarized if the MTOC was within the interface-proximal third of the cell (Fig. 1D), as described in Materials and Methods. Image analysis software was also used to measure the distance of the MTOC from the cell-bead interface. We tested receptors known to be involved in synapse formation. First, the adhesion molecules LFA-1 and ICAM-1 were examined. Notably, while LFA-1 and ICAM-1 are enriched at the VS (2, 3, 5), whether this interaction simply serves to stabilize cell-cell contacts or whether it plays a more specific role by inducing "outside-in" signaling into cells to facilitate T cell remodeling and VS formation remains unresolved. To investigate the effects of engagement of LFA-1, an antibody specific to the LFA-1 ␤ 2 subunit (L130) (45) was used. As a positive control for MTOC polarization, we used an antibody against CD3 (OKT3) (44), a component of the TCR complex, since it is well established that the MTOC is recruited to the T cell immunological synapse by engagement of T cell receptor (TCR) (28,47). Uncoupled "null" beads were used as a negative control. As expected, cross-linking CD3 induced HIV-1-infected T cells to polarize the MTOC to the cell-bead interface (59% of cells polarized with anti-CD3 beads compared to 24% polarized with null beads; one-way ANOVA; P Ͻ 0.0001). Notably, incubation with anti-LFA-1 ␤ 2 beads also resulted in significantly more MTOC polarization than incubation with null beads (47% polarized with anti-LFA-1 ␤ 2 beads and 24% polarized with null beads; one-way ANOVA; P Ͻ 0.01). Although fewer cells polarized in response to LFA-1 engagement than in response to CD3, the frequencies were not significantly different (P Ͼ 0.05). In contrast, cross-linking ICAM-1 did not induce significant polarization. Furthermore, uninfected Jurkat cells incubated with anti-LFA-1 ␤ 2 beads did not show this polarization phenotype, indicating that LFA-1-induced polarization is potentiated by HIV-1 infection (29% of uninfected T cells polarized compared to 47% of HIV-1infected Jurkat cells; one-way ANOVA; P Ͻ 0.05) (Fig. 1E).
Measuring the position of the MTOC relative to the cell-bead interface supported these data. Consistent with CD3 engagement providing the strongest stimulus, the MTOC was closest to the interface under these conditions (2 m from the interface) and furthest away under the null condition (3.5 m from the interface). In HIV-1-infected Jurkat cells incubated with anti-LFA-1 ␤ 2 beads, the MTOC was located significantly closer to the cell-bead interface than when incubated with null beads (2.7 m with anti-LFA-1 compared to 3.5 m with null beads; one-way ANOVA; P Ͻ 0.01). Additionally, uninfected Jurkat cells incubated with anti-LFA-1 ␤ 2 beads had their MTOCs significantly further away from the cell-bead interface, reflecting a nonpolarized phenotype (2.7 m in HIV-1-infected Jurkat cells compared to 3.6 m in uninfected Jurkat cells; one-way ANOVA; P Ͻ 0.01) (Fig. 1F). To provide some context for these distances, the average diameter of a Jurkat cell is approximately 8 m. We also developed a measure- ment, termed the "polarization index," that takes into account the size of a cell and is the ratio between the distance of the MTOC from the interface and the diameter of the cell. When the polarization index was calculated for anti-LFA-1 and anti-ICAM-1 cellbead conjugates, it mirrored the results seen in Fig. 1F, supporting the observation that LFA-1, but not ICAM-1, was associated with MTOC polarization (data not shown). Together, these data show that LFA-1 engagement alone is sufficient to trigger T cell polarization in HIV-1-infected Jurkat cells.
The integrin LFA-1 is a transmembrane heterodimer composed of two subunits, ␤ 2 and ␣ L (48). Upon ligand binding, the cytoplasmic tail of ␤ 2 contributes to outside-in signaling via its interactions with proteins, such as talin and Vav 1 (48). To investigate if LFA-1-associated polarization in HIV-1-infected cells was associated with the active ␤ 2 component, HIV-1-infected Jurkat cells were incubated with antibody-coupled beads specific to the ␤ 2 subunit or the ␣ L subunit. Figure 1G shows that cells incubated with the anti-␤ 2 beads were significantly more likely to be polarized (47% polarized with anti-␤ 2 compared to 29% polarized with anti-␣ L ; Student's t test; P ϭ 0.001), confirming that induction of polarization in HIV-1-infected T cells is mediated by the active ␤ 2 subunit of LFA-1. As anti-␣ L beads did not trigger a polarization effect, they were used in further assays as a negative control.
To investigate if LFA-1 was required for T cell polarization in response to cell-cell contact, a cell-based assay using mutant Jurkat derivative cell lines was employed. LFA-1-negative Jurkat cells (J␤2.7), LFA-1-negative Jurkat cells reconstituted with WT LFA-1 (J␤2.7/LFA-1 WT), or the parental LFA-1-expressing cells (Jn9) (35,36) were infected with HIV-1, mixed with dye-loaded primary CD4 target T cells for 60 min at 37°C, and analyzed by IF. VS (as defined above) were scored as polarized if the MTOC was found in the interface-proximal third of the HIV-1-infected T cell. Because LFA-1-negative cells show reduced, but not completely abolished, Gag recruitment and VS formation (5), it was possible to use these cells to score MTOC polarization at sites of cell-cell contact. LFA-1-negative cells showed a significant defect in MTOC polarization at VS (71% of WT compared to 27% of LFA-1 negative cells; one-way ANOVA; P Ͻ 0.0001) that was rescued by restoring LFA-1 expression (27% in LFA-1-negative compared to 60% in LFA-1-reconstituted cells; P Ͻ 0.01) (Fig. 1H), confirming the importance of LFA-1 binding for contact-induced polarization in HIV-1-infected T cells at the VS.
In addition to integrins, the HIV-1 envelope glycoprotein (Env) is expressed on the surface of HIV-1-infected T cells, and binding to CD4 on opposing target T cells is a key requirement for VS formation (1-4). To investigate whether Env contributes directly to contact-induced polarization in HIV-1-infected T cells, the cell-based VS assay was again employed. CD4 T cells were infected with WT or VSV-G-pseudotyped ⌬Env HIV-1, mixed with dye-loaded autologous CD4 target T cells for 60 min at 37°C, and analyzed by IF (Fig. 1l). As expected, infected cells not expressing HIV-1 Env (⌬Env) showed a profound defect in MTOC polarization at infected cell-target cell contacts compared to WT HIV-1 VS, which is consistent with a lack of stable cell-cell contacts and VS formation (1)(2)(3)(4). Importantly, T cells infected with a variant of the NL4-3 molecular clone with a 144-amino-acid deletion in the cytoplasmic tail (HIV-1 ⌬Env-CT) (49) showed no defect in MTOC polarization. This deletion leaves only a small portion of Env-CT, anchoring it in the cell membrane and permitting CD4coreceptor binding via gp120, but removes any potential cytoplasmic signaling domains. Thus, we conclude that direct signaling mediated by the Env-CT is not required to recruit the MTOC, whereas surface Env-gp120 expression is necessary in order to engage CD4 ϩ T cells and establish stable contacts for subsequent T cell remodeling mediated by cell-cell receptor interactions.
LFA-1-induced MTOC polarization is an active process. Next, we sought to confirm that LFA-1-associated polarization in infected cells was an active process indicative of signaling and not simply due to the MTOC already being positioned close to a plasma membrane domain enriched in LFA-1 at which antibodycoupled beads could preferentially bind. To do this, HIV-1-infected Jurkat cells were conjugated with antibody-coupled beads for 15 or 60 min before formaldehyde fixation and staining to visualize the MTOC. Figure 2A shows a time-dependent increase in MTOC polarization in response to LFA-1 ␤ 2 engagement suggestive of active recruitment, similar to the positive-control CD3. In contrast, HIV-1 Jurkat cells conjugated to anti-LFA-1 ␣ L , anti-ICAM-1 beads, or null beads did not show a time-dependent increase in polarization ( Fig. 2A). The MTOC is recruited to immune cell contacts in a microtubule-dependent manner (24,50,51). Therefore, pharmacological inhibitors were used to disrupt microtubule and actin networks to provide further evidence to support active recruitment. Nocodazole is an inhibitor that depolymerizes microtubules and can prevent MTOC recruitment (52). Cytochalasin D is an inhibitor of actin polymerization and so should not affect MTOC movement (53). HIV-1-infected Jurkat cells were pretreated with either drug for 30 min or left untreated and then incubated with antibody-coupled beads for 60 min before staining to visualize the MTOC. Nocodazole treatment significantly inhibited polarization induced by anti-CD3 and anti-LFA-1 ␤ 2 beads. In contrast, cytochalasin D had little effect (Fig.  2B). Taken together, these data show that the MTOC is actively trafficked in response to LFA-1 engagement in a manner that requires a functional microtubule network in HIV-1-infected cells. (1-3, 7, 33), we wished to confirm these data using primary T cells. To do this, antibodycoupled bead assays were also performed using HIV-1-infected primary CD4 T cells. Figure 2C shows that, similar to Jurkat cells, anti-LFA-1 ␤ 2 engagement induced polarization so that the MTOC was located significantly closer to the bead-cell interface than with null beads (1.6 m with anti-LFA-1 ␤ 2 beads versus 2.8 m with null beads; one-way ANOVA; P Ͻ 0.0001). Additionally, uninfected primary T cells incubated with anti-LFA-1 ␤ 2 beads were found to have the MTOC significantly further away from the cell-bead interface, which is indicative of HIV-1 infection conditioning cells to polarize, similar to Jurkat T cells (1.6-m interface distance with anti-LFA-1 ␤ 2 beads in HIV-1 primary T cells compared to 2.6 m in uninfected T cells; one-way ANOVA; P Ͻ 0.0001). Again, anti-LFA-1 ␣ L beads did not induce significant polarization. Interestingly, anti-ICAM-1 beads were associated with a polarized phenotype in primary cells (1.3-m average distance from the interface) (Fig. 2C). To explore this in more detail, a time course analysis in which cells were conjugated with antibody-coupled beads for 15 or 60 min was performed. As expected, either anti-CD3 or anti-LFA-1 ␤ 2 induced a significant increase in polarization over time (anti-CD3, 40% increase in polarization over time [Student's t test; P Ͻ 0.0005]; anti-LFA-1 ␤ 2 , 39% increase in polarization over time [Student's t test; P Ͻ 0.0001])  (Fig. 2D). Importantly, engaging ICAM-1 did not increase polarization over time, suggesting that there is no active recruitment of the MTOC. To further analyze MTOC recruitment in HIV-1infected primary CD4 T cells, the inhibitors nocodazole and cytochalasin D were again used to interfere with microtubule or actin polymerization. As expected, nocodazole abrogated CD3and LFA-1 ␤ 2 -induced polarization (25% reduction; Student's t test; P Ͻ 0.05) but had no effect on anti-LFA-1 ␣ L or anti-ICAM-1 (Fig. 2E). Together, these data show that engaging LFA-1 actively recruits the MTOC along microtubules to sites of cell contact in HIV-1-infected primary CD4 T cells, in agreement with our observations in Jurkat T cells. Because ICAM-1 engagement is associated with a polarized phenotype but without the requirement for active MTOC trafficking, we conclude that anti-ICAM-1 beads likely bind to a plasma membrane compartment that is enriched in ICAM-1 and already spatially associated with the MTOC; thus, signaling through ICAM-1 does not induce active polarization. This membrane domain may be the uropod that in polarized primary T cells contains the MTOC and is enriched in ICAM-1 and Gag (54).

LFA-1-induced polarization in HIV-1-infected T cells proceeds through a pathway involving the T cell kinase ZAP70.
Active LFA-1-induced T cell polarization is indicative of localized synaptic signaling. To investigate downstream mediators, we took advantage of a panel of Jurkat cell lines defective in key cellular proteins involved in T cell remodeling and organelle polarization at immune cell contacts (55,56). Specifically, we interrogated ZAP70 (37), SLP76 (41), LAT (40), and Lck (39), which are key regulators of T cell signaling pathways in response to immune cell interactions (57). Western blot analysis confirmed the phenotype of the cells (Fig. 3A), and nearly 100% infection of the cell lines was achieved by using VSV-G-pseudotyped NL4-3 (Fig. 3B). Next, flow cytometry analysis was used to quantify surface LFA-1 expression in the cell lines. With the exception of LAT-negative cells, all the cell lines expressed LFA-1 (Fig. 3C). Next, HIV-1-infected WT or signaling-defective cells were incubated with anti-LFA-1 ␤ 2 beads for 60 min, and MTOC polarization was scored as previously described. Figure 3D shows that ZAP70-negative cells were significantly impaired in LFA-1 ␤ 2 -induced MTOC polarization to sites of cell-cell contact (one-way ANOVA; P ϭ 0.01), but no defect was seen in the absence of SLP76 or Lck. Since LAT-negative cells did not express LFA-1, the failure of these cells to polarize was expected. Restoring expression of ZAP70 in negative Jurkat cells completely rescued LFA-1-induced MTOC polarization in HIV-1-infected cells to WT levels (Fig. 3E). To confirm the importance of ZAP70, HIV-1 primary CD4 T cells were pretreated with piceatannol, a Syk kinase inhibitor that has been shown to specifically inhibit the activation of ZAP70 (58). Treatment with 10 M piceatannol resulted in reduced polarization of the MTOC at sites of LFA-1 engagement (52% polarized contacts with dimethyl sulfoxide [DMSO] control; 25% polarized contacts with piceatannol treatment; P ϭ 0.05) (Fig. 3F). Taken together, these data indicate that LFA-1-induced polarization at the VS proceeds down a pathway requiring ZAP70 in HIV-1-infected T cells.

LFA-1-induced T cell remodeling is associated with polarization of viral proteins.
To examine the consequences of LFA-1 crosslinking for polarized virus assembly at the VS, antibody-coupledbead assays were performed using HIV-1-infected primary CD4 T cells coupled with immunofluorescent staining for surface Env (monoclonal antibody [MAb] 50-69) and intracellular Gag. Infected cells were incubated with beads for 5, 15, or 60 min, and contacts were scored as polarized if there was a significant accumulation of viral protein at the cell-bead interface. Unfortunately, the staining conditions required to permeabilize and visualize the MTOC by IF result in poor surface Env staining; therefore, we quantified viral protein polarization at the cell-bead interface in the absence of concomitant MTOC polarization. Despite this, it was striking that anti-LFA-1 ␤ 2 beads induced a 6-fold increase in the enrichment of surface Env and intracellular Gag to the cell-bead interface over time (4% of contacts at 5 min; 25% of contacts at 60 min; P ϭ 0.0003) (Fig. 4A and B). In contrast and as expected, uncoupled null beads induced no recruitment of surface Env or intracellular Gag to the cell-bead interface (Fig. 4A). Likewise, anti-LFA-1 ␣ L beads did not induce viral-protein recruitment (8% of contacts at 5 min; 7% of contacts at 60 min; P Ͼ 0.05) (Fig. 4A and B).
Little is known about how Env is recruited to the VS. In contrast, Gag has been better studied, and it has been reported that the nucleocapsid domain is required to direct Gag to plasma membrane compartments enriched in uropod markers, termed uropod-directed microdomains (UDMs) (54). As Gag multimerizes, the highly basic region in the matrix domain promotes its targeting to specific UDMs containing PSGL-1, CD43, and CD44 (59). These UDMs have been suggested to carry Gag to the uropods of infected T cells, which forms a prepolarized Gag assembly platform that can participate in VS formation (54). As the intracellular route that Env takes to the VS remains unclear, we decided to further investigate the relationship between Env and T cell polarization induced by LFA-1. Notably, while analyzing intracellular Env staining, we observed that intracellular Env-positive compartments were very frequently clustered around the polarized MTOC. Quantification of this revealed that 91% (Ϯ6%) of primary CD4 T cells in contact with anti-LFA-1 ␤ 2 beads in which the MTOC was polarized to the contact site showed this staining pattern (Fig. 4C). Figure 4C shows xz and xy images of a three-dimensional (3D)-volume reconstruction of a polarized HIV-1-infected T cell in contact with an anti-LFA-1 ␤ 2 bead built using z-stacks taken 0.2 m apart in which Env (green) is clearly seen clustered in a compartment around the MTOC (red). We frequently observed this compartment taking a ring-like shape with the MTOC located in the central hole. Importantly, blocking MTOC polarization by inhibiting ZAP70 with piceatannol prevented the recruitment of intracellular Env in response to LFA-1 cross-linking, with Env remaining clustered around the nonpolarized MTOC (Fig. 4D), indicating a functional coupling of MTOC and intracellular Env recruitment to the VS.
Intracellular Env localizes to a Golgi compartment proximal to the MTOC at the VS. To further investigate the MTOC-associated Env-positive compartment, we performed IF microscopy and costained for intracellular organelles. HIV-1-infected primary CD4 T cells were incubated with autologous target cells, fixed, permeabilized, and stained for Env (MAb 2G12); the MTOC (␥-or ␣-tubulin); and either the Golgi (giantin), lysosome (lamp 1), early endosome (EEA-1), or recycling endosome (Rab11a) compartments, which reflect the intracellular itinerary of newly synthesized and endocytosed Env (60). Analysis of VS with the MTOC polarized to the contact site revealed that Env predominantly colocalized with the Golgi marker giantin in a compartment proximal to the polarized MTOC (Pearson coefficient for Env/giantin ϭ 0.61 Ϯ 0.08) (Fig. 5A). Partial colocalization of Env was also seen with the early endosome marker EEA-1 (Pearson coefficient for Env/EEA-1 ϭ 0.24 Ϯ 0.08) (Fig. 5C) and the recycling endosome marker Rab11a (Pearson coefficient for Env/Rab11a ϭ 0.25 Ϯ 0.05) (Fig. 5D). In contrast, there was little colocalization between Env and the late endosome/lysosomal marker lamp 1 (Pearson coefficient for Env/lamp 1 ϭ 0.07 Ϯ 0.02) (Fig. 5B), consistent with efficient Golgi network retrieval of Env following endocytosis from the plasma membrane (61). Taken together, these data show that intracellular Env is associated close to the MTOC, with most steady-state Env localized to the MTOC-proximal Golgi network that copolarizes to the VS in response to cell-cell contact.

DISCUSSION
Cell-cell contact at the VS is associated with extensive T cell remodeling, with organelles such as mitochondria and the MTOC reorienting to be proximal to the site of viral egress (32,33). We have previously shown that calcium-dependent T cell remodeling during cell-cell contact is required for efficient HIV-1 dissemination between T cells. However, the triggers and pathways involved in this process are currently unknown and may have implications for future antiviral strategies to specifically target cell-cell spread.
Furthermore, while the requirement for LFA-1/ICAM in VS formation has been demonstrated by many studies (2,3,5), the precise role of LFA-1/ICAM at the VS remain unclear. Here, we show that engagement of the T cell integrin LFA-1 is sufficient to ac- tively recruit the MTOC in HIV-1-infected T cells to the VS; that this is mediated by the ␤ 2 subunit of LFA-1, which is the known signaling component of LFA-1 (48,62,63); and that LFA-1-induced T cell polarization requires the kinase ZAP70. In addition to inducing MTOC polarization, LFA-1 cross-linking was also found to recruit viral proteins. Most notably, we found that intracellular Env is localized in a Golgi compartment proximal to the MTOC at the VS and that inhibition of MTOC polarization by pharmacological blockade of ZAP70 signaling also results in loss of intracellular Env clustering at sites of cell contact.
Our data suggest a model (Fig. 6) in which physical contact between an unpolarized HIV-1-infected T cell and a suitable target T cell triggers remodeling that is mediated, at least in part, by LFA-1 signaling into the infected cell through a ZAP70-dependent pathway. This remodeling provides a trigger for T cell polarization and a focal point for the recruitment of MTOC-associated organelles that support cell-cell spread, including the Golgi compartment that resides proximal to the MTOC, thus potentiating delivery of intracellular Env to the plasma membrane (PM) to facilitate full VS formation and polarized virus assembly and budding for rapid infection of target cells by cell-cell spread. While contactinduced T cell remodeling at the VS has been shown to support efficient viral dissemination (1-3, 7, 33), a significant outstanding question remained as to how Env is recruited to sites of cell-cell contact at the VS to support polarized viral assembly and cell-cell spread. During the course of examining MTOC polarization at the VS, we noticed that intracellular Env was frequently found localized around the MTOC in a ring-like structure. Further analysis of intracellular compartments revealed that this was the Golgi compartment and that most intracellular Env clustered around the MTOC was colocalized to the Golgi and to a lesser extent to early and recycling endosomes. Furthermore, LFA-1 binding was able to trigger the polarization of this intracellular Env clustered around the MTOC to the contact zone, while inhibition of MTOC polarization by pharmacological inhibition of ZAP70 blocked this and resulted in Env and the MTOC remaining distal to the VS. In this way, MTOC polarization at the VS may also help facilitate cell-cell spread by recruiting the secretory apparatus and Golgi compartment, which contains the majority of Env (60), allowing specific targeting of Env to the plasma membrane at sites of cellcell contact. It should be noted that this model of contact-induced T cell polarization is entirely consistent with previous work reporting Gag localization to uropods and uropod-mediated cellcell contact (54), as it provides an explanation for relocalization of the uropod toward a target cell to form a VS through LFA-1 signaling. Even under conditions where the uropod (which contains the MTOC) had formed a prepolarized Gag-containing platform for cell-cell contact (54), the ability of infected T cells to dynamically remodel, depending on the status of the infected cell and where it forms contact with the target T cell, would afford HIV-1 with the most opportunities to form VS and to spread by highly efficient cell-cell means.
On T cells, the integrin LFA-1 plays an important role in cell arrest during migration and formation of the immunological synapse (64,65). The ␤ 2 subunit is involved in outside-in signaling when LFA-1 binds its ligand, ICAM-1 (48). At the immune synapse, it is well known that TCR ligation triggers MTOC recruitment that is required to recruit cellular organelles for polarized secretion (25,28,30,31,47). However, it has also been demonstrated that under certain chemokine conditions, LFA-1 activation alone can induce fast recruitment of the MTOC and mitochondria to a potential APC (64). We now show that HIV-1 infection of T cells can also serve as a necessary stimulus to prime or condition T cells to undergo contact-induced polarization. The fact that HIV-1-infected T cells are more prone to polarize in response to LFA-1 engagement is consistent with previous reports that describe polarization in HIV-1-infected T cells, but not target T cells, at the VS (32,33). Here, we have shown that while surface Env expression is necessary for polarization at the VS, consistent with the requirement for Env-CD4 binding in VS formation, the cytoplasmic tail of Env appears to be dispensable, indicating that cytoplasmic signaling directly through Env is not required. Because uninfected CD4 T cells typically form only transient contacts with one another (4), it is likely that Env acts to allow T cells to form sustained homotypic contacts and that this facilitates prolonged engagement of LFA-1 by its ligand, promoting signaling into the cell to induce T cell remodeling. Further work to define the viral protein(s) responsible for conditioning infected cells to polarize would clearly be informative.
LFA-1-induced polarization is indicative of localized signaling, and this study found that the T cell adaptor ZAP70, but not SLP76 or Lck, is required for this process in both HIV-1-infected Jurkat cells and primary CD4 T cells. ZAP70 and Lck have been reported to associate with LFA-1 and are activated by ligand binding (58,(66)(67)(68). In support of our results, one other study has also reported that ZAP70-negative cells show a defect in MTOC polarization to the VS (34); however, the molecular mechanisms were less clear, and the triggers for polarization were not examined. Here, we provide some mechanistic insight into the requirement for ZAP70 by linking ZAP70 to a pathway involving cell-cell contact and engagement of LFA-1 that induces T cell remodeling and  1) The Env glycoprotein binds to its receptors, CD4 and coreceptor. This allows two CD4 T cells to form a stable contact. (2) On the infected cell, LFA-1 is engaged by its ligand, ICAM-1, which is present on the target cell plasma membrane. This may precede, follow, or be contemporaneous with Env-receptor binding. (3) LFA-1-ICAM binding triggers a signaling event into the infected cell, mediated by the ␤ 2 subunit of LFA-1 and proceeding through a pathway involving the kinase ZAP70. (4) This signal acts to recruit the MTOC and associated organelles containing intracellular Env. (5) The MTOC moves to be proximal to the cell-cell interface, bringing associated Env-containing components of the secretory apparatus, such as the Golgi complex. This could allow more focused and rapid trafficking of intracellular Env toward a contact zone. (6) The infectious-virus assembly is polarized toward the cell-cell interface, resulting in further elaboration of the VS, directed viral budding, and highly efficient infection of the target cell. polarization of organelles and viral proteins, potentially facilitating VS formation and efficient viral spread. In contrast, the role of Lck in T cell polarization is less well defined at both the IS and the VS. At the IS, Lck has been shown by one group (47) to be important for polarization, whereas another group found the related Src family kinase (SFK) Fyn was required for proximal MTOC movement and that Lck was necessary only for centrosome docking at the cell interface (56). Additionally, Fyn has been shown to compensate for an absence of Lck signaling at the IS (69). Further work will be required to determine whether Fyn plays any role in compensating, in part, for the loss of Lck signaling at the VS. Moreover, while we have shown a requirement for LFA-1 in triggering MTOC polarization at the VS, the complexity of multiple receptor interactions that occur during cell-cell contact at the VS means that other cell surface interactions also undoubtedly contribute functionally to VS formation and efficient HIV-1 cell-cell spread through activating synaptic signaling. It would clearly be of interest to further define the molecular details of signaling in recruitment of specific viral and cellular proteins to the VS, as this will provide greater insight into the mechanisms of HIV-1 dissemination and pathogenesis and potentially reveal future strategies to specifically target this mode of HIV-1 dissemination.