Ezrin Is a Component of the HIV-1 Virological Presynapse and Contributes to the Inhibition of Cell-Cell Fusion

ABSTRACT

During cell-to-cell transmission of HIV-1, viral and cellular proteins transiently accumulate at the contact zone between infected (producer) and uninfected (target) cells, forming the virological synapse. Rearrangements of the cytoskeleton in producer and target cells are required for proper targeting of viral and cellular components during synapse formation, yet little is known about how these processes are regulated, particularly within the producer cell. Since ezrin-radixin-moesin (ERM) proteins connect F-actin with integral and peripheral membrane proteins, are incorporated into virions, and interact with cellular components of the virological presynapse, we hypothesized that they play roles during the late stage of HIV-1 replication. Here we document that phosphorylated (i.e., active) ezrin specifically accumulates at the HIV-1 presynapse in T cell lines and primary CD4⁺ lymphocytes. To investigate whether ezrin supports virus transmission, we sought to ablate ezrin expression in producer cells. While cells did not tolerate a complete knockdown of ezrin, even a modest reduction of ezrin expression (∼50%) in HIV-1-producing cells led to the release of particles with impaired infectivity. Further, when cocultured with uninfected target cells, ezrin-knockdown producer cells displayed reduced accumulation of the tetraspanin CD81 at the synapse and fused more readily with target cells, thus forming syncytia. Such an outcome likely is not optimal for virus dissemination, as evidenced by the fact that, in vivo, only relatively few infected cells form syncytia. Thus, ezrin likely helps secure efficient virus spread not only by enhancing virion infectivity but also by preventing excessive membrane fusion at the virological synapse.

IMPORTANCE While viruses, in principal, can propagate through successions of syncytia, HIV-1-infected cells in the majority of cases do not fuse with potential target cells during viral transmission. This mode of spread is coresponsible for key features of HIV-1 pathogenesis, including killing of bystander cells and establishment of latently infected T lymphocytes. Here we identify the ERM protein family member ezrin as a cellular factor that contributes to the inhibition of cell-cell fusion and thus to suppressing excessive syncytium formation. Our analyses further suggest that ezrin, which connects integral membrane proteins with actin, functions in concert with CD81, a member of the tetraspanin family of proteins. Additional evidence, documented here and elsewhere, suggests that ezrin and CD81 cooperate to prevent cytoskeleton rearrangements that need to take place during the fusion of cellular membranes.

INTRODUCTION

HIV-1 is transferred to uninfected cells most efficiently if infected and uninfected cells directly contact each other (for a recent review, see reference 1). Such cell-to-cell transmission requires not only successful assembly and release of viral particles from infected (producer) cells followed by their entry into adjoining uninfected (target) cells (2), but also necessitates precise spatial and temporal coordination of viral functions in producer and target cells (1). Cytoskeletal rearrangements in producer and target cells are required for the polarization of viral components and cellular factors toward the cell-cell interface and are thus critical for the formation of the so-called virological synapse (VS) (3–6). On the target cell side, cytoskeleton-interacting proteins regulate cortical actin and aid in the recruitment of CD4 and coreceptors to the site of viral attachment and entry (7–12). Also, virus-induced signaling further alters actin-based events in the target cell, likely to create a more accommodating environment for fusion and postfusion steps (13–17). In the producer cell, the subject of the current study, polarization of the microtubule-organizing center (MTOC) (18, 19), along with the presence of actin, adhesion molecules, and the clustering of tetraspanins and lipid raft components at the VS, suggests that this transient adhesion structure, i.e., the presynapse, is built comparably to the presynaptic face of the immunological synapse (IS) (3, 4, 20–22; also reviewed in reference 23). At the immunological presynapse, specific cytoskeleton-regulating proteins have been shown to play important roles (recently reviewed in reference 24), but it is unknown whether they also function at the virological presynapse.

While several studies documented that actin both colocalizes with HIV-1 assembly sites and is incorporated into newly formed viral particles (25–27), disruption of actin dynamics using drugs has been reported to have only small effects on HIV-1 assembly and release (25, 28, 29), suggesting that these steps are not dependent on the cytoskeleton. However, in those studies (as well as others showing that the cytoskeleton-interacting proteins KIF4 and filamin A are involved in the trafficking of the main structural protein of HIV-1, Gag, to the virus assembly site at the plasma membrane [30, 31]), assembly and release were analyzed in contexts that did not include target cells. Therefore, these analyses do not allow one to make predictions about whether or not these cytoskeletal structures are important for virus dissemination via cell-cell contacts. Likely more relevant for HIV-1 spread in vivo, in polarized, crawling T cells, HIV-1 assembly has recently been shown to be directed toward the uropod, a distinct (initially posterior) surface structure that gives rise to the presynaptic terminal of the VS (5, 32). Consequently, in these studies as well as in earlier ones, disruption of the cytoskeleton using actin- or microtubule-modulating drugs has been documented to inhibit the polarization of viral components toward the VS and to impede transmission of newly assembled particles (3–5, 32, 33). However, while the viral determinant for Gag's uropod localization has been identified (32, 34), it remains unclear whether specific, cytoskeleton-interacting proteins in producer cells help to secure the transmission of newly formed viral particles to target cells.

Ezrin-radixin-moesin (ERM) proteins are F-actin binding proteins that act as linkers between membrane proteins and the cytoskeleton. The three family members (ezrin, radixin, and moesin) have tissue-specific expression patterns, with ezrin and moesin being expressed in T cells. Through their ability to reversibly (via phosphorylation of C-terminal residues) link F-actin to specific membrane proteins, they have been recognized to be important regulators of diverse membrane-based processes and signaling events (reviewed in references 35 and 36). Of potential significance, phosphatidylinositol 4,5-bisphosphate (PIP₂), which is essential for Gag targeting to the plasma membrane, is required for optimal ezrin activation (37). Further, ERM proteins (which are incorporated into HIV-1 virions [26]) are known to both reside at the T cell uropod and to regulate its structure and protein composition (38–41), as well as to play multiple roles at the IS. For instance, ezrin is required for proper microcluster mobility in both T and B cells, as well as MTOC localization toward the synapse (42, 43). Additionally, the kinase ZAP-70 is positioned at the IS by ezrin while CD43 is drawn away from the contact site due to its interaction with moesin (44), although a recent study challenges the need for ezrin in ZAP-70 localization (45). ERM proteins also coordinate signaling at the IS, with defined roles for ERM proteins in interleukin-2 (IL-2) secretion and extracellular signal-regulated kinase (ERK) and nuclear factor of activated T cells (NFAT) activity, as well as in regulation of the GTPase Cdc42 (42, 45–47). As mentioned above, though obviously not being completely identical cellular structures, the IS and the HIV-1 VS share numerous specific features (23, 48), including regulation of cortical actin at the contact site, as well as the repositioning of the MTOC adjacent to the cell-cell interface. Additionally, ZAP-70 is recruited not only to the IS but also to the VS, where it regulates HIV-1 cell-to-cell transmission (18), further documenting that some of the same cellular machinery is required for proper functioning of these transient adhesion structures. For all these reasons, we decided to evaluate whether ERM proteins play a role(s) during the late phase of the HIV-1 replication cycle, i.e., on the producer cell side of the VS.

Previous studies examining ERM proteins in the target cell have documented both positive (8, 49) and negative (50–52; also our unpublished results) roles for these proteins during the early phase of HIV-1 replication, as well as during the early phase of other viruses (53). Here, we focused our investigations on the potential importance of ERM proteins in producer cells by evaluating how these proteins contribute to (or counteract) successful cell-to-cell transmission, likely the predominant mode of virus transmission in vivo. As mentioned above, analysis of purified HIV-1 preparations identified ERM proteins in viral particles, demonstrating that they are present at the assembly/release sites (26). Further suggesting a potential involvement in the regulation of virus transmission, ERM-interacting proteins such as the tetraspanin CD81 are specifically recruited to viral assembly sites (54) and have been implicated in regulating viral processes such as release (55), virus-cell and cell-cell fusion (56–58), and cell-to-cell transmission (58). Based on these lines of evidence, along with the previously mentioned similarities between the VS and IS, we hypothesized that ERM proteins within the HIV-1 producer cell support efficient VS-mediated HIV-1 transmission.

MATERIALS AND METHODS

Cells.

HEK 293T, HeLa, and TZM-bl cells were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Invitrogen). Jurkat (clone E6-1) and CEMss cells were maintained in RPMI medium containing 10% FBS. The reagent CEM.NKR-CCR5-Luc (referred to throughout as CEM-luc) was obtained from John Moore and Catherine Spenlehauer through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH, and maintained in RPMI medium containing 10% FBS and 0.8 mg/ml G418 (Cellgro). Primary CD4⁺ cells were purified using a Ficoll separation of whole blood isolated from healthy donors, followed by negative selection (Miltenyi Biotec) according to the manufacturer's instructions. Primary CD4⁺ cells were cultured in RPMI medium supplemented with 10% FBS and 25 units/ml IL-2, activated with 5 μg/ml phytohemagglutinin (PHA) for 24 h, washed, and infected as described below.

Antibodies.

The following reagents were obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: mouse monoclonal antibody to HIV-1 p24 (AG3.0) from Jonathan Allan, human HIV-IG from North American Biologicals, Inc. (NABI), and the National Heart, Lung, and Blood Institute from Luiz Barbosa, mouse HIV-1 p24 hybridoma (183-H12-5C) from Bruce Chesebro and Hardy Chen, and antiserum to HIV-1 gp120 (sheep) from Michael Phelan. Rabbit anti-p6 was a kind gift from David Ott. The mouse anti-CD4 antibody Leu3A was purchased from BD Biosciences. Anti-CD81 was purchased from BD (catalog no. 555675). Goat anti-ezrin (sc-6407; Santa Cruz) was used for detection of total ERM proteins, while goat anti-p-moesin (sc-12895; Santa Cruz) and rabbit anti-pERM (catalog no. 3149; Cell Signaling) were used for detection of C-terminal phosphorylation of ERM proteins. To detect ezrin and moesin individually, we obtained rabbit anti-ezrin (catalog no. 3145) and rabbit anti-moesin (catalog no. 3150) from Cell Signaling Technology. Mouse anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Abcam) was used as a loading control when needed. Horseradish peroxidase (HRP)-conjugated secondary antibodies were purchased from Jackson, while Alexa Fluor-conjugated secondary antibodies were purchased from Molecular Probes.

Puromycin shRNA lentiviral knockdown system.

The FG12 short hairpin RNA (shRNA) lentiviral knockdown system has been described previously (58, 59). To introduce puromycin resistance, we amplified the puro resistance cassette out of the pEF Flag puro C vector (a kind gift from Oliver Dienz) using the PacI restriction site containing primers 5′-GTCGACTTAATTAATACCGGGTAGGGGAGGC-3′ (forward) and 5′-GCACCGGTTTAATTAATCAGGCACCGGGCTTGC-3′ (reverse) and ligated it into FG12. To introduce ERM-knockdown sequences into FG12-Puro, we used oligos ordered from Integrated DNA Technologies with restriction site overhangs BbsI and XhoI (ezrin sense, 5′-ACCGGCCGTGGAGAGAGAGAAAGATTCAAGAGATCTTTCTCTCTCTCCACGGCTTTTTTACCGGTC-3′, and anti-sense, 5′-TCGAGACCGGTAAAAAAGCCGTGGAGAGAGAGAAAGATCTCTTGAATCTTTCTCTCTCTCCACGGC-3′; moesin sense, 5′-ACCGGAGATCGAGGAACAGACTAATTCAAGAGATTAGTCTGTTCCTCGATCTCTTTTTTACCGGTC-3′, and anti-sense, 5′-TCGAGACCGGTAAAAAAGAGACTGAGGAACAGACTAATCTCTTGAATTAGTCTGTTCCTCGATCTC-3′). Note that these are complete shRNA sequences containing overhang restriction sites, the loop sequence, and an AgeI site for ligation verification. These oligos were annealed and ligated into PBS-hU6-1 using BbsI and XhoI restriction sites in order to obtain a Pol III promoter. This vector was then digested with XbaI and XhoI and ligated into the FG12-Puro vector described above.

Viral stocks and infections.

Vesicular stomatitis virus glycoprotein (VSV-G) pseudotyped viral strains NL4-3 and NL4-3 iGFP (kindly provided by Benjamin Chen, Mount Sinai Hospital) were produced by transfecting HEK 293T cells with the respective proviral construct and plasmid encoding VSV-G (10:1) using the calcium phosphate method according to the manufacturer's (Invitrogen) instructions. Supernatants were collected 2 days posttransfection, centrifuged at 1,000 × g for 10 min, aliquoted, and stored at −80°C. Enzyme-linked immunosorbent assays (ELISAs) were carried out to quantify the p24 content of the viral stocks. For spinoculations, 3 million CEMss or primary CD4⁺ cells were centrifuged at 1,200 × g for 2 h at 37°C with the indicated amount of p24 (the amount is indicated in each experimental section of Materials and Methods). Cells were then incubated at 37°C for 45 min, washed, and resuspended in RPMI plus 10% FBS medium. Cells were generally used 2 days postinfection or as indicated within each experimental section in Materials and Methods. In cases when spinoculation was not used, p24 (in the amount indicated in the corresponding experimental section of Materials and Methods) was incubated with the specified cells for 4 h, and the cells were washed, cultured, and used 2 days postinfection.

To produce shRNA lentiviral stocks, we transfected HEK 293T cells using the calcium phosphate method (Invitrogen) with FG12-Puro (containing either shScramble or ShEzrin sequences) along with the packaging vector ΔR8.2 and the envelope vector VSV-G. Supernatants were collected 2 days posttransfection, centrifuged at 1,000 × g for 10 min, aliquoted, and stored at −80°C.

Creation of shRNA stable cell lines.

Three million CEMss cells were spinoculated with 450 μl of shScramble or shEzrin lentivirus-containing supernatants (described above). Cells were then incubated at 37°C for 45 min, washed, and resuspended in complete medium. Two days later, the medium was replaced with RPMI medium containing 2 μg/ml puromycin (Sigma), and cells were passaged in this medium for at least 8 days before they were used for experiments. To confirm knockdown, we analyzed cells by immunofluorescence and flow cytometry, while we analyzed cell lysates by Western blotting, as described below.

Virological synapse visualization.

Three million CEMss or primary CD4⁺ cells were spinoculated with 20 to 50 ng p24 as described above. Two days later, infected producer cells were mixed with 7-amino-4-chloromethylcoumarin (CMAC) (Invitrogen)-labeled target cells (either CEMss or primary CD4⁺ cells) at a 1:3 ratio (infected:target) on poly-l-lysine (Sigma)-coated MatTek plates (MatTek Corporation). After 2 to 3 h, cells were fixed with 4% paraformaldehyde (PFA) for 10 min, blocked and permeabilized with 1% bovine serum albumin (BSA)–0.2% Triton X-100 in phosphate-buffered saline (PBS) for 10 min, and subsequently stained with the antibodies indicated on the figures. Image acquisition was performed on a Nikon Eclipse Ti-E microscope using a Nikon Plan Apo 1.4 NA 60× objective equipped with a Qimaging EXi Blue camera. Image analysis was done using NIS Elements 3.10 and AutoQuant 2.1.0 (for deconvolution). Accumulation of phosphorylated ERM protein (pERM), ezrin, or moesin was scored at the microscope and subsequently displayed using GraphPad Prism 5.

Quantifying total ERM and pERM signals in infected cells.

Three million CEMss or primary CD4⁺ cells were spinoculated with 20 ng p24 as described above. Two days later, infected producer cells were mixed with CMAC (Invitrogen)-labeled uninfected cells (either CEMss or primary CD4⁺ cells) at a 1:1 ratio (infected:uninfected) on poly-l-lysine (Sigma)-coated MatTek plates (MatTek Corporation). After 2 to 3 h, cells were fixed with 4% PFA for 10 min, blocked and permeabilized with 1% BSA–0.2% Triton X-100 in PBS for 10 min, and subsequently stained with the indicated antibodies. Random fields were imaged at a magnification of ×20 using the Nikon Eclipse Ti-E setup described above. Images were then imported into FiJi, infected and uninfected cells were circled using differential interference contrast (DIC), and the mean fluorescence was then recorded in the desired channel (either total ERM or pERM). Background fluorescence was obtained for each individual image and subtracted from each cell value. Data are from an experiment representative of two independent experiments and are displayed using GraphPad Prism 5.

Quantification of pERM and CD81 at the VS.

To quantify the relative levels of pERM and CD81 at the VS, 3 million shScramble or shEzrin cells were spinoculated with 5 ng of p24 as described above. Two days later, these infected producer cells were mixed with CMAC (Invitrogen)-labeled target CEMss cells at a 1:3 ratio (infected:target) on poly-l-lysine (Sigma)-coated MatTek plates (MatTek Corporation). After 2 to 3 h, cells were fixed with 4% PFA for 10 mins, blocked and permeabilized with 1% BSA–0.2% Triton X-100 in PBS for 10 min, and stained with the indicated antibodies. Images were acquired on a DeltaVision workstation (Applied Precision) with an Olympus IX70 base equipped with a CoolSnap HQ CCD camera (Photometrics) and an Olympus Plan Apo 1.42 NA 60× objective. Softworx software was used to deconvolve and export the images. Images were then loaded into FiJi. VS areas were selected in the Gag channel, and then the mean fluorescence intensity of that area was recorded in the pERM or CD81 channel. Two similar-sized, non-VS zones of the same cell were also measured for the mean intensity of pERM or CD81, and these numbers were used to represent the non-VS intensity of the selected channel. An area devoid of cells was measured to calculate the background intensity value for each channel, which was subtracted from the VS and non-VS values. The VS “enrichment factor” of pERM or CD81 was calculated by dividing the VS signal intensity by the non-VS signal intensity. For pERM, a total of 57 synapses were analyzed using shScramble cells as producers and 53 synapses using shEzrin cells as producers pooled from three independent experiments. For CD81, a total of 55 synapses were analyzed using shScramble cells as producers and 54 synapses using shEzrin cells as producers pooled from two independent experiments. Data were displayed using GraphPad Prism 5.

Release and infectivity assays.

Three million shScramble or shEzrin cells were incubated with 5 ng VSV-G pseudotyped NL4-3 for 4 h, washed, and cultured. Two days later, equal numbers of cells from each condition were plated, and 48 h later supernatants and lysates were harvested. Cellular debris was cleared from the supernatants by centrifugation at 1,000 × g for 10 min. Cells were lysed in TNE (50 mM Tris, 200 mM NaCl, 2 mM EDTA) containing 1% Triton X-100 and a protease inhibitor cocktail (catalog no. P834D; Sigma). p24 content was determined by ELISA, and relative viral release was calculated by dividing the supernatant p24 by the total p24 (supernatant + lysate). Three independent experiments were conducted, with three technical replicates per experiment.

To calculate the infectivity of released virions, we allowed infected shScramble or shEzrin cells to replicate for 6 days. Supernatants were then harvested as described above, and p24 content was measured by ELISA. TZM-bl cells were plated in a 96-well format, and 0.1 ng of virus produced from either shScramble or shEzrin cells was added per well. Forty-eight hours later, β-galactosidase (Pierce) activity was measured. Five independent experiments were conducted, with three technical replicates per experiment.

Transfer assay.

Three million shScramble or shEzrin cells were spinoculated with 10 ng VSV-G pseudotyped NL4-3 iGFP as described above. Two days later, the infected producer cells were counted and mixed with 7-hydroxy-9H-(1,3-dichloro-9,9-dimethylacridin-2-one) (DDAO; Invitrogen)-labeled CEMss target cells at a 2:1 ratio. The cells were incubated for 3 h in the presence of the antibodies indicated on the figures (0.25 μg/ml), washed in fresh media, trypsinized for 1 min, washed, and fixed in 4% PFA. In parallel, shScramble and shEzrin producers alone were fixed in 4% PFA, blocked and permeabilized with 1% BSA–0.2% Triton X-100 in PBS for 10 min, and stained for Gag using AG3.0 followed by Alexa Fluor 647-conjugated secondary antibody. Samples were analyzed on a BD LSR II flow cytometer. FlowJo, version 10 (Tree Star), was used to analyze and quantify the data.

Coculture luciferase assays.

Three million shScramble or shEzrin cells were spinoculated with 5 ng VSV-G pseudotyped NL4-3 as described above. Two days later, equal numbers of cells from each condition were mixed with CEM-luc cells with either dimethyl sulfoxide (DMSO), zidovudine (AZT; 10 μM), or plerixafor (AMD3100; 500 nM), and luciferase activity was measured 24 h and 48 h later. For cell-free conditions, 20 ng VSV-G pesudotyped NL4-3 was incubated with the same number of CEM-luc cells for 6 h (on the same day as the cocultures described above were begun), washed, and plated. Luciferase activity was measured 24 h and 48 h later.

Cell-cell fusion assay.

Three million shScramble or shEzrin cells were spinoculated with 5 ng VSV-G pseudotyped NL4-3 as described above. Two days later, equal numbers of cells from each condition were mixed with CMAC (Invitrogen)-labeled target cells at a 1:3 ratio. Cells were incubated for 3 h, fixed with 4% PFA for 10 min, blocked and permeabilized with 1% BSA–0.2% Triton X-100 in PBS for 10 min, and stained for p24. Cells were then placed on poly-l-lysine (Sigma)-coated MatTek plates (MatTek Corporation), allowed to settle for 1 h at room temperature, and then imaged on the Nikon Eclipse Ti-E setup described above. Infected cells were located based on positive Gag staining and then evaluated in the blue (CMAC) channel. Cells that stained positive for both Gag and CMAC were counted as syncytia, and the percentages of infected cells forming syncytia over three separate experiments were quantified (total, n = 271 infected shScramble cells, n = 256 infected shEzrin cells) and displayed using GraphPad Prism 5.

Western blotting.

Equal numbers of shScramble or shEzrin cells were lysed in TNE (50 mM Tris, 200 mM NaCl, 2 mM EDTA) containing 1% Triton X-100 and a protease inhibitor cocktail (catalog no. P834D; Sigma). Loading buffer was added, and samples were electrophoresed in a 10% polyacrylamide gel and transferred to a nitrocellulose membrane (Pall Corporation). Membranes were blocked with 5% milk in Tris-buffered saline (TBS)–0.15% Tween 20 for 1 h at room temperature. The primary antibodies indicated on the figures were incubated overnight at 4°C in blocking buffer, washed, and probed with HRP-conjugated secondary antibodies (Jackson). Signals were revealed using Western blotting enhanced chemiluminescence (ECL) kits from Thermo Scientific.

For Western blotting of pelleted virus particles, supernatants from infected shScramble and shEzrin cells were centrifuged at 400 × g for 5 min followed by centrifugation at 1,200 × g for 10 min. Supernatants were then filtered through a 0.22-μm filter (VWR) and centrifuged at 80,000 × g for 2 h at 4°C. Viral pellets were lysed in TNE containing 1% Triton X-100 and a protease inhibitor cocktail (catalog no. P834D; Sigma), loading buffer was added, and the samples were run as described above.

For Western blot quantification using the Li-Cor Odyssey system, protein was transferred to a polyvinylidene difluoride (PVDF) membrane (Immobilon-FL; Millipore), blocked with Li-Cor blocking buffer (Li-Cor), and probed with primary antibodies as described above. Membranes were probed with IR800 secondary antibodies (Li-Cor), imaged, and quantified on a Li-Cor Odyssey system (Image Studio software).

Flow cytometry.

shScramble or shEzrin cells were fixed with 4% PFA for 10 min, blocked and permeabilized with 1% BSA–0.2% Triton X-100 in PBS for 10 min, and stained with either rabbit anti-ezrin followed by anti-rabbit Alexa Fluor 647 (Molecular Probes), or Alexa Fluor 647-conjugated phalloidin (Molecular Probes). Samples were analyzed on a BD LSR II flow cytometer. FlowJo, version 10 (Tree Star), was used to analyze and quantify the data.

Statistical analysis.

All statistical analysis was carried out in GraphPad Prism 5. Student's t test or one-way analysis of variance (ANOVA) with a Bonferroni's post hoc test was used as indicated in the figure legends. Values were considered significantly different if P was ≤0.05.

RESULTS

pERM proteins localize at HIV-1 assembly sites in T cells.

To evaluate the potential role(s) of ERM proteins in the late stages of HIV-1 replication, we first sought to determine, in T lymphocytes, the localization of these proteins relative to virus assembly sites. As previously mentioned (and documented by others; e.g., see references 38–41), ERM proteins localize to the uropod in polarized T cells (Fig. 1). Since these distinct membrane areas serve as sites of polarized HIV-1 assembly (5, 32) and also because ERM proteins are incorporated into virions (26), we anticipated seeing coaccumulation of ERMs and Gag at these locations. To visualize viral assembly sites, we infected T cells with NL4-3 and immunostained for Gag and with an antibody that recognizes the C-terminally phosphorylated (and thus active) version of all three ERM proteins (pERM). We found that in Jurkat, CEMss, and primary CD4⁺ T cells, pERMs localized to polarized HIV-1 assembly sites (Fig. 1A).

FIG 1.

pERM and HIV-1 Gag colocalize in T cells. (A) Jurkat (top), CEMss (middle), or primary CD4⁺ T cells (bottom) were infected with NL4-3, fixed, and stained for Gag and pERM. (B) Uninfected CEMss cells were stained for pERM and imaged. Three representative cells are shown. Scale bar = 10 μm.

Phosphorylated ERM proteins accumulate at the virological synapse, but overall pERM levels are diminished in HIV-1-infected T lymphocytes.

Coaccumulation of pERMs with Gag at areas of viral assembly, as documented in Fig. 1, strongly suggested that pERMs would also be found at the virological presynapse. To visualize the VS, we cocultured infected CEMss or primary CD4⁺ T cells with CMAC-labeled autologous target cells, fixed and stained them with the indicated antibodies, and imaged them as described in Materials and Methods. Synapses were identified as cell-cell contact sites where Gag polarized toward an adjacent (target) cell, and images were then scored for enrichment of pERM, ezrin, or moesin. As shown in Fig. 2, pERMs accumulated at the HIV-1 VS in the majority of conjugates examined in both CEMss (Fig. 2A to C) and primary CD4⁺ T cells (Fig. 2D to F). Comparable localization of pERMs at the VS was observed when Jurkat cells were used as producers (data not shown). Interestingly, in primary CD4⁺ T cells, ezrin seemed to be the predominant ERM protein at the synapse (Fig. 2F), while in CEMss cell-cell conjugates, ezrin was observed to accumulate at the VS to only a slightly higher degree than moesin (Fig. 2C). Because T cells do not express radixin, we did not analyze the localization of this ERM family member. Altogether, these data demonstrate that pERMs are enriched at the HIV-1 VS; they also reveal that, similar to what has been documented for the IS (44–46), ezrin is the predominant ERM protein at the VS.

FIG 2.

pERM accumulates at the HIV-1 virological synapse. CEMss T cells (A to C) or primary CD4⁺ lymphocytes (D to F) were infected with NL4-3, allowed to form synapses with CMAC-labeled autologous target cells, fixed, and stained with the indicated antibodies. Polarized staining at the VS of either pERM (B and E) or ezrin and moesin (C and F) was compared to that of uninfected cell contacts. Bars represent the averages of three independent experiments (± standard deviations [SD]). Scale bar = 10 μm. Infected CEMss T cells (G) or primary CD4⁺ lymphocytes (H) were mixed with autologous uninfected T cells, fixed, and stained for Gag and either total ERM or pERM. Cells were circled using DIC, and the mean fluorescence intensity of total ERM or pERM was calculated per cell. Results are representative of two independent experiments. For statistical analysis, Student's t test or a one-way ANOVA was performed with a Bonferroni's post hoc test. Means were considered significantly different if the P value was ≤0.05.

While examining pERM localization using deconvolution microscopy, we noticed that although pERMs accumulated at the VS, overall, unconjugated HIV-1-infected cells seemed to express far less pERM than their uninfected counterparts. To quantify this apparent difference, we mixed infected CEMss or primary CD4⁺ T cells with uninfected CMAC-labeled autologous T cells and stained them for Gag and either total ERM or pERM. Random fields were imaged, cells were circled, and the mean fluorescence intensity of total ERM or pERM per cell was measured. As shown in Fig. 2G (CEMss cells) and Fig. 2H (primary CD4⁺ cells), HIV-1-infected cells indeed contained significantly less pERM than uninfected cells, with no apparent change in the total ERM signal, confirming that HIV-1 infection specifically alters the levels of the active, i.e., phosphorylated forms of these proteins.

Ezrin knockdown results in the production of virus with decreased infectivity.

Together, the data presented in Fig. 1 and 2 suggested that ERM proteins play roles during the late stages of HIV-1 replication. Based on their localization at the cell periphery and specifically at the VS, we reasoned that ERM proteins could affect (i) HIV-1 assembly and release, (ii) the infectivity of newly released particles, or (iii) virus transmission to target cells. To address these hypotheses, we sought to create stable ezrin and moesin knockdown CEMss cell lines using a lentiviral shRNA system (see Materials and Methods for details). While shMoesin cells always failed to recover during the selection phase, suggesting that moesin expression is essential for continued growth of CEMss cells, shEzrin cells did recover. However, when ezrin levels in these shEzrin cells were assessed by Western blotting (Fig. 3A), immunofluorescence microscopy (Fig. 3B), and flow cytometry (Fig. 3C), the average ezrin reduction in multiple (independently generated) cell lines was never more than 50%, suggesting that these cells do not tolerate complete loss of this ERM member either (Fig. 3C).

FIG 3.

Ezrin knockdown in producer cells results in decreased viral infectivity. Levels of ezrin were determined in shScramble and shEzrin CEMss cells via Western blotting (A), immunofluorescence (B), and flow cytometry (C). Data are representative of multiple separate cell lines, and panel C shows an average of the results for three independent cell lines. MFI, mean fluorescence intensity. Scale bar = 10 μm. Viral release (D) and infectivity (E) of particles produced in shScramble or shEzrin cells were measured. To calculate release, infected cells were plated, and 2 days later, supernatant and lysate p24 were measured using ELISA. Release was calculated using the formula release = supernatant p24/total p24 and subsequently normalized to shScramble. Bars represent the averages of three independent experiments (±SD). To measure infectivity, we collected supernatants from infected cultures of either shScramble or shEzrin cells and added equal amounts of p24 to TZM-bl cells. Bars represent the averages of five independent experiments (±SD). (F) VLPs from shScramble and shEzrin cells were pelleted, lysed, and immunoblotted for gp120 and p24. For statistical analysis, Student's t test was performed. Means were considered significantly different if the P value was ≤0.05.

We first examined whether (partial) ezrin depletion in producer cells alters HIV-1 assembly and release. shScramble or shEzrin CEMss cells were incubated with NL4-3 for 4 h, washed, and cultured. Two days later, equal numbers of cells were plated, and 48 h after plating, the supernatants and lysates were harvested and assayed for p24 content by ELISA. Relative viral release was calculated by dividing the supernatant p24 by the total p24 (supernatant + lysate). As summarized in Fig. 3D, ezrin does not appear to be required for viral assembly or release, as ezrin knockdown did not significantly affect the levels of released viral p24.

To evaluate the infectivity of viral particles produced in shEzrin cells, we harvested supernatants from infected cultures and measured the p24 content by ELISA. Equal amounts of p24 from supernatants of either shScramble or shEzrin cells were then added to TZM-bl reporter cells, which produce β-galactosidase under the control of the HIV-1 long terminal repeat (LTR). After 48 h, β-galactosidase production was measured, and a clear drop in infectivity was seen for virus harvested from shEzrin cells compared to that for virus released from shScramble cells (Fig. 3E). To test if the defect in infectivity was the result of decreased Env incorporation into virions produced from shEzrin cells, we harvested, pelleted, and immunoblotted virus particles for gp120 and p24. As documented in Fig. 3F, however, comparable levels of virion-associated gp120 were found in particles released from shScramble and shEzrin cells.

Ezrin depleation decreases the relative amount of pERM proteins at the VS.

Because pERM and ezrin accumulate at the VS (Fig. 2), we next quantified how ezrin knockdown in the producer cell affected pERM levels at the VS. Infected shScramble or shEzrin cells were cocultured with CMAC-labeled CEMss cells, fixed, and stained for Gag and pERM (Fig. 4A). Contact zones with polarized Gag staining were scored as VSs, and the enrichment of pERMs at the synapse was calculated by dividing the pERM intensity at the synapse by the pERM intensity at noncontact membrane areas of the same cell (see Materials and Methods for a more detailed description of this quantification). The relative enrichment of pERM proteins at the VS, as shown in Fig. 4B, was ∼10-fold in shScramble producer cells but was reduced to ∼5-fold in shEzrin cells.

FIG 4.

Ezrin knockdown reduces pERM levels at the VS and affects cell-to-cell transmission. Relative levels of pERM at the VS were measured by allowing infected shScramble or shEzrin cells to form synapses with CMAC-labeled CEMss target cells. Cells were then fixed and stained for Gag and pERM. Synapses were identified by the polarization of Gag toward the target cell, and pERM intensity was measured at the synapse and compared to a non-synapse area of the same cell to calculate the relative enrichment (see Materials and Methods). Representative images (A) and quantification of all synapses (B) pooled from three independent experiments are shown. (C and D) Spinocculated shScramble and shEzrin cells were lysed, and the ratio of p55 to GAPDH was measured by Western blotting and quantified using a Li-Cor Odyssey system. A representative blot (C) and quantification of three independent infections (D) are shown (±SD). To measure transmission, we mixed infected shScramble or shEzrin cells with CEM-luc cells and measured luciferase 48 h later (E). For statistical analysis, Student's t test was performed. Means were considered significantly different if the P value was ≤0.05.

Next, we evaluated whether the decreased pERM levels at the VS affects cell-to-cell transmission. We applied a commonly used coculture assay in which CEM-luc cells (which express luciferase under the control of the HIV-1 LTR) are used as target cells. To ensure comparable levels of initial viral production from shScramble and shEzrin producers, and to overcome the known effects of ERM knockdown on entry and postentry steps, we spinoculated the cells with VSV-G pseudotyped NL4-3. Quantitative Western blotting using the Li-Cor Odyssey system was used to measure the amounts of Gag (p55) and GAPDH present in the infected producer cells. Figure 4C displays a representative blot, while Fig. 4D is a quantification of three separate experiments. To then measure cell-to-cell transmission, we cocultured equal numbers of infected shScramble or shEzrin cells with CEM-luc cells for 48 h. The luciferase signal was found to be 2-fold higher when shEzrin cells were used as producers than when shScramble cells were used (Fig. 4E). These data suggested that ezrin, somewhat paradoxically, might act as a barrier to cell-to-cell transmission, even though ezrin is required for the production of fully infectious particles (Fig. 3E).

Ezrin knockdown does not affect viral transfer to target cells.

To test whether the increased luciferase signal was indeed the result of augmented particle transfer from shEzrin producer cells to target cells, we adapted an assay developed by the Chen group (33), which allowed us to specifically measure viral transfer, irrespective of whether the viral transfer results in productive infection, i.e., viral entry followed by reverse transcription and integration. shScramble and shEzrin cells were spinoculated with NL4-3 iGFP and cultured for 48 h. The cells were then either fixed and stained for Gag (Fig. 5A) or mixed at a 2:1 ratio with uninfected DDAO-labeled target cells (Fig. 5B) and analyzed by flow cytometry. As shown in Fig. 5A, at the time of coculture, shScramble and shEzrin cells were producing equal amounts of cell-associated Gag. Following a 3-h coculture with labeled targets, the cells were trypsinized, fixed, and analyzed by flow cytometry. DDAO cells were gated and assessed for their green fluorescent protein (GFP) fluorescence (a measure of the amount of virus being transferred). There was no significant difference between the amount of viral transfer when either shScramble or shEzrin cells were used as producers (Fig. 5B and C). Note that the majority of viral transfer was sensitive to the antibody Leu3A, confirming that the measured viral transfer is CD4 dependent (Fig. 5B and C).

FIG 5.

Viral transfer to target cells is unaffected by ezrin knockdown. Viral transfer to target cells was measured using a flow-based coculture assay. iGFP-infected shScramble or shEzrin cells were either fixed and stained for Gag (A) or cocultured with DDAO-labeled targets with the indicated antibodies for 3 h, trypsinized, and fixed (B). DDAO target cells that are GFP positive are indicative of viral transfer. (C) Percentages of viral transfer determined from three independent experiments (normalized to infected shScramble without antibody treatment).

Evaluating virus transmission and syncytium formation with a commonly used coculture assay.

Because we did not see any difference in total viral transfer to target cells when either shScramble or shEzrin cells were used as producers, we decided to more closely examine the source of the luciferase signal in our cell-to-cell transmission coculture assay (Fig. 4E). While the luciferase coculture assay can be used to measure the efficiency of HIV-1 cell-to-cell transmission, it is evident that the luciferase signal can be the result of either “true” viral infection derived from cell-to-cell transmission (when Tat, which activates luciferase expression, is synthesized after integration of the incoming, newly reverse transcribed viral genome) or that luciferase gene expression can be triggered by preexisting Tat, i.e., Tat that has been synthesized in the producer cell and that quickly reaches the HIV-1 LTR upon fusion of producer and target cells (see Fig. 6A for a comparison of the two modes of activation).

FIG 6.

shEzrin cells show elevated levels of HIV-1-induced cell-cell fusion. (A) The outline of a cell-to-cell transmission assay shows how luciferase production in this assay can be stimulated by newly synthesized Tat (upper half) and/or by Tat that can transfer into target cells upon cell-cell fusion. (B and C) Either free virus (B) or infected CEMss cells (C) were incubated with CEM-luc cells for 24 h or 48 h in the presence of the indicated drugs. (D) Under the conditions described for panel C, infected CEMss cells were imaged at a magnification of ×10. Syncytia are indicated by red arrows. (E and F) Equal numbers of infected shScamble or shEzrin cells were mixed with CEM-luc cells in the presence of the indicated drugs for 24 h (E) or 48 h (F) (n = 3, ± standard errors of the means [SEM]). (G and H) Infected shScramble or shEzrin cells were mixed with CMAC-labeled CEMss target cells, fixed, stained for Gag, and imaged at a magnification of ×40. Syncytia were determined by the presence of Gag and CMAC in the same cell. A representative image (G) and quantification (H) of three independent experiments (±SEM) are shown. Scale bar = 10 μm. For statistical analysis, Student's t test was performed. Means were considered significantly different if the P value was ≤0.05.

In order to first evaluate the timing of luciferase production in this assay, we mixed either cell-free virus (Fig. 6B) or infected shCEMss cells (Fig. 6C) with equal numbers of CEM-luc cells in the presence of the indicated drugs, and luciferase activity was measured 24 h and 48 h later. As shown in Fig. 6B, addition of free virus did not yield signals above that produced upon mock infection at 24 h, but an appreciable luciferase signal appeared at 48 h. This signal was largely sensitive to treatment with the reverse transcriptase inhibitor AZT. Together, these data show that although viral entry, reverse transcription, integration, and viral Tat transcription/translation can of course occur within the first 24 h after the initiation of infection, detectable levels of luciferase are not yet produced, and hence, productive infection, involving the above-mentioned steps, cannot be evaluated at the 24-h time point using this assay.

As described above, luciferase production in CEM-luc cells upon cocultivation with infected cells can result from transmission of viral particles followed by integration and subsequent transcription of the viral genome or directly upon fusion of infected producer and CEM-luc target cells. In the latter case, Tat that is already present in the infected cell can stimulate luciferase expression immediately after cell-cell fusion, and consequently, the luciferase signal should appear at 24 h and be insensitive to AZT, yet sensitive to AMD-3100, a small molecule that binds CXCR4, thus preventing conformational changes required for fusion (Fig. 6A). Indeed, when infected shScramble cells were cocultured with CEM-luc cells, an appreciable luciferase signal appeared at 24 h, the majority of which was insensitive to AZT but completely sensitive to the addition of AMD-3100 (Fig. 6C). This result, together with the one shown in Fig. 6B, suggests that in the coculture system, the luciferase signal at 24 h results from cell-cell fusion and not productive transmission, as further evidenced by the visual observation of large syncytia if cells were cocultured without AMD-3100 (Fig. 6D). Also note that AZT treatment inhibited productive transmission in the coculture assay only ∼3.5-fold at the 48-h time point (Fig. 6C), while AZT treatment of free virus inhibited transmission about 11-fold (Fig. 6B). These results are consistent with recent data showing that cell-to-cell transmission of HIV-1 is more resistant to treatment with reverse transcriptase inhibitors than cell-free infection (60).

Ezrin knockdown enhances HIV-1-induced cell-cell fusion.

To then evaluate whether ezrin knockdown in producer cells affects productive transmission to target cells or cell-cell fusion, we cocultured equal numbers of infected shScramble or shEzrin cells with CEM-luc cells. Evaluation of the luciferase signal upon mixing of producer and CEM-luc target cells for 24 h showed that luciferase production was 2.5-fold greater when shEzrin cells were used as producers (Fig. 6E). As before, this signal was largely resistant to AZT, suggesting that shEzrin cells fused with target cells more readily than their shScramble counterparts. To specifically confirm that shEzrin producer cells formed more syncytia than shScramble cells, we infected the different cells as described above and incubated them at a 1:3 ratio with CMAC-labeled CEMss cells. After 3 h, the cells were resuspended, fixed, labeled for Gag, and subjected to immunofluorescence imaging. Individual infected cells and the products of cell-cell fusion events, i.e., syncytia, were identified by the presence of either just Gag (individual cells) or Gag and CMAC (syncytia). Figure 6G displays a representative image, while Fig. 6H shows the average percentage of infected cells forming syncytia in three separate experiments. Altogether, these results demonstrate that a 50% ezrin depletion in producer cells (Fig. 3) leads to a doubling of HIV-1-induced syncytium formation.

Levels of CD81 at the virological synapse and overall levels of F-actin are reduced in shEzrin cells.

We have previously documented that a known interaction partner of ezrin, CD81, is also present at the virological synapse and negatively affects membrane fusion (57). Therefore, we tested whether reduced amounts of ezrin at the presynapse directly correlated with decreased CD81 levels at this site. Infected producer cells (either shScramble or shEzrin) were allowed to form contacts with CMAC-labeled targets, and CD81 enrichment at the VS was measured as for Fig. 4 (also see Materials and Methods). We found that when shEzrin cells were used as producers, the relative enrichment of CD81 at the VS was significantly lower than when shScramble cells were used (Fig. 7A and B).

FIG 7.

Ezrin knockdown affects CD81 levels at the VS. Relative levels of CD81 at the VS were measured by allowing infected shScramble or shEzrin cells to form synapses with CMAC-labeled CEMss target cells. Cells were then fixed and stained for Gag and CD81. Synapses were identified by the polarization of Gag toward the target cell, and CD81 intensity was measured at the synapse and compared to a non-synapse area of the same cell to calculate the relative enrichment (see Materials and Methods). Representative images (A) and quantification of all synapses (B) pooled from two independent experiments are shown. F-actin levels were visualized and measured based on phalloidin staining using immunofluorescence (C) and flow cytometry (D). Data are representative of multiple independent cell lines per technique. Scale bars = 10 μm.

Because cortical F-actin is also known to inhibit the fusion process (61), we next assessed the overall levels of F-actin in both shScramble and shEzrin cells by immunofluorescence (Fig. 7C) and flow cytometry (Fig. 7D). We found that, overall, F-actin levels are decreased in shEzrin cells, consistent with the increased cell-cell fusion observed in these cells.

DISCUSSION

HIV-1 cell-to-cell transmission is a multifaceted event which relies heavily on the cytoskeleton, yet very few cytoskeleton-interacting proteins within the producer cell have been found to contribute to VS formation and cell-to-cell transmission. Here, we report that phosphorylated (active) ERM proteins accumulate at the HIV-1 VS. Because our initial investigations revealed that ezrin is the predominant ERM protein at the presynaptic terminal of the VS, we attempted to ablate this protein and test how this affected virus transmission to target cells. It was found that, while T lymphocytes did not tolerate complete loss of ezrin, even a partial knockdown in producer cells decreased the amount of pERMs at the synapse, decreased the infectivity of released virions, and resulted in increased syncytium formation. At the onset of our studies aimed at identifying a potential role for ERM proteins in HIV-1-producing cells, we documented their localization relative to the viral protein Gag. In Fig. 1 and 2, we document specific accumulations of pERM proteins, particularly ezrin, at both polarized assembly sites (Fig. 1) and at the VS (Fig. 2). Of note, while pERMs were present at assembly sites, these proteins only partially colocalized with Gag. Localization of ERM proteins at the VS is consistent with data which documented that HIV-1 assembly takes place at the uropod of T lymphocytes, a structure that is known to be regulated by pERM proteins (38–41). How ezrin is actively phosphorylated at the VS remains to be investigated. As mentioned in the Introduction, PIP₂ has been show to play a role in ezrin phosphorylation, and since this phospholipid is also required for Gag recruitment to the plasma membrane (62), it could potentially link ezrin activation with its localization at the VS. Also of possible relevance, we note that ERM proteins are intimately tied to Rho family GTPase activity: they reside both upstream and downstream of specific Rho signaling reactions (reviewed in reference 63). Of potential significance, Rho GTPases have been implicated in HIV-1 replication (28, 64, 65). Specifically, during viral assembly, the cytoplasmic tail of HIV-1 Env interacts with a Rho family guanine nucleotide exchange factor (GEF), possibly providing a spatial and temporal link between Rho activity, pERMs, and the VS (66). Further, numerous effector kinases have been documented to phosphorylate ERM proteins, and several of them, including protein kinase C (PKC) and protein kinase A (PKA), have been found in purified HIV-1 virions, revealing that they localize at viral assembly sites (67, 68). Finally, Lck and citron kinase have been shown to modulate HIV-1 release from producer cells, while ZAP-70 (also in producer cells) has been documented to regulate HIV-1 cell-to-cell transmission (18, 69, 70).

Interestingly though, and as documented in Fig. 2G and H, despite being specifically enriched at the VS, pERM levels, overall, were clearly decreased in infected cells, a finding that was also confirmed by flow cytometry (data not shown). Since HIV-1 typically alters the expression of specific cellular proteins to ensure that viral replication can proceed unhindered (reviewed in references 71 and 72), it seems likely that the virus benefits from pERM downregulation, though we cannot rule out the possibility that the decreased pERM levels in infected cells are an unrelated consequence of HIV-1 infection.

To evaluate whether ezrin plays a specific role during the late stages of HIV-1 replication, we established an ezrin knockdown cell line (shEzrin) and used these cells as producers in experiments aimed at analyzing whether this ERM protein is required for virus production and/or subsequent transmission. Though it was not possible to generate cell lines that had ezrin expression levels lower than 50% of those seen in control cells (shScramble), we found that HIV-1 virions produced in shEzrin cells were significantly less infectious than those produced in shScramble cells, while overall particle release remained unaffected (Fig. 3). This decrease in infectivity was not the result of altered Env incorporation into virions (Fig. 3F). While this suggests that these virions enter target cells less efficiently than particles released from shScramble cells, it is also conceivable that postentry steps are affected. Contrasting with these data, ezrin knockdown in HIV-1-producing cells was documented in a recent paper to increase free virus infectivity (73). In the latter study, however, Bregnard et al. used 293T cells as producers. 293T cells not only express overall levels of ezrin that differ from those seen in T lymphocytes (which were used in our study), but these adherent cells also express all three ERM proteins, possibly explaining the different outcomes.

Because cell-to-cell transmission is the predominant mode of HIV-1 dissemination and because this transmission mode can overcome multiple producer and target cell barriers (74), we next sought to test how efficiently virus produced in shEzrin cells would be transmitted to target cells (via the VS). We first examined synapses using immunofluorescence and found that ezrin depletion led to a deceased amount of pERM at the contact site, demonstrating that phosphorylated ezrin (p-ezrin) is the predominant pERM protein at the VS. As documented in Fig. 5 and 6, such decreased p-ezrin at the viral presynapse did not affect viral transfer to target cells but clearly affected cell-to-cell transmission when a reporter-based assay was used. Upon further investigation, we found that when infected shEzrin cells were allowed to form VSs with target cells, twice as many syncytia were observed than when shScramble cells were used as virus producers. These results suggest that the presence of ezrin at the VS inhibits syncytium formation. We also measured a significant increase in luciferase signal derived from productive transmission (48-h time point) when shEzrin cells were used as producers. However, although this might suggest that depletion of ezrin in the producer cell also results in increased productive transmission, the 48-h time point is difficult to interpret, since the initial difference in syncytium formation may skew later luciferase readings. Furthermore, our data document that shEzrin cells transfer similar amounts of virus to target cells (Fig. 5). Nevertheless, overall these data suggest that ezrin does indeed promote viral spread, not by enhancing particle release (Fig. 3) or particle transfer (Fig. 5), however, but by preventing membrane fusion and thus facilitating the separation of producer and target cell upon particle transmission at the VS.

The finding that ezrin regulates cell-cell fusion at the VS is reminiscent of a recent report demonstrating that this ERM protein can also regulate non-virus-induced membrane fusions in vivo: ezrin depletion leads to a destabilization of the intestinal brush border in adult mice, with clearly observable fusions between adjacent villi (75). How ezrin regulates membrane fusion remains to be determined in both cases. It seems reasonable to assume, based on its subcellular localization, that this ERM protein does not directly interfere with conformational changes in viral or cellular fusogens required for membrane fusion. Rather, it would seem conceivable that ezrin, which connects membrane proteins with the underlying cortical actin network, regulates the physical properties of the membrane, such as stiffness and tension (76), which are important factors in determining how readily membranes will fuse (77–80; also hypothesized in reference 53). Indeed, ERM proteins have been found to directly regulate membrane tension, with possible consequences for membrane fusion processes (81, 82). Furthermore, cortical actin itself can inhibit cell-cell fusion (61), which is consistent with our finding that shEzrin cells have an overall decrease in cortical F-actin.

Alternatively (or additionally), ezrin may affect fusion through an interaction with other known fusion regulators. We and others have previously documented that tetraspanins negatively regulate HIV-1-induced fusion (56, 57, 83), while others have shown they are also involved in nonviral fusion processes (e.g., see reference 84). Because ezrin directly interacts with the tetraspanin CD81 (85), we sought to evaluate the effect of ezrin knockdown on the accumulation of CD81 at the VS. In Fig. 7 we document that ezrin depletion correlated with decreased CD81 levels at the cell-cell contact site, indicating that the fusion inhibition mediated by these two proteins may be mechanistically linked. In support of this, we recently found (using single-molecule tracking) that CD81 trapping at HIV-1 assembly sites requires ezrin (P. Rassam, L. Fernandez, M. Thali, P. Dosset, E. Rubinstein, and P. E. Milhiet, unpublished results). Given our recent report showing that tetraspanin overexpression blocks syncytium formation at a posthemifusion stage (86), it is tempting to speculate that both ezrin and CD81 may be involved in the pore opening and/or expansion phase of fusion, the phase of fusion which involves cytoskeletal rearrangements. Future experiments will be needed to evaluate how ezrin and CD81 functions are coordinated, likely together with those of other cellular and viral factors, to repress the formation of syncytia.

While syncytia likely help spread the virus, as evidenced by the finding that these entities can be found in infected individuals (e.g., see reference 87) as well as in lymph nodes of HIV-1-infected humanized mice (88), typically producer and target cells disengage upon particle transfer at the VS. Previously, we have shown that tetraspanins repress HIV-1 Env-induced cell-cell fusion, and more recently we documented that viral Gag at the VS, before being cleaved upon particle release, prevents Env-triggered fusion of producer and target cells (89), similar to how it represses the activity of particle-associated Env (90–93). The current study adds yet another component, p-ezrin, to the list of factors that secure the integrity of the VS by regulating the fusion of producer and target cells.