Moesin is required for HIV-1-induced CD4-CXCR4 interaction, F-actin redistribution, membrane fusion and viral infection in lymphocytes

The human immunodeficiency virus 1 (HIV-1) envelope regulates the initial attachment of viral particles to target cells through its association with CD4 and either CXCR4 or CCR5. Although F-actin is required for CD4 and CXCR4 redistribution, little is known about the molecular mechanisms underlying this fundamental process in HIV infection. Using CD4+ CXCR4+ permissive human leukemic CEM T cells and primary lymphocytes, we have investigated whether HIV-1 Env might promote viral entry and infection by activating ERM (ezrin-radixin-moesin) proteins to regulate F-actin reorganization and CD4/CXCR4 co-clustering. The interaction of the X4-tropic protein HIV-1 gp120 with CD4 augments ezrin and moesin phosphorylation in human permissive T cells, thereby regulating ezrin-moesin activation. Moreover, the association and clustering of CD4-CXCR4 induced by HIV-1 gp120 requires moesin-mediated anchoring of actin in the plasma membrane. Suppression of moesin expression with dominant-negative N-moesin or specific moesin silencing impedes reorganization of F-actin and HIV-1 entry and infection mediated by the HIV-1 envelope protein complex. Therefore, we propose that activated moesin promotes F-actin redistribution and CD4-CXCR4 clustering and is also required for efficient X4-tropic HIV-1 infection in permissive lymphocytes.


Introduction
The entry of HIV into a target cell requires the interaction of multiple receptor and co-receptor molecules with each viral envelope (Env) trimer to promote the formation of fusion pores (Doms, 2000;Kuhmann et al., 2000). F-actin reassembly appears to be involved in Env-induced CD4, CXCR4 and LFA-1 redistribution to cell-cell junctions during viral spreading (Jolly et al., 2004), and drugs that affect actin dynamics appear to inhibit Env-mediated fusion (Eitzen, 2003;Pontow et al., 2004); but it is unclear whether these drugs affect HIV-1 Env activity (Campbell et al., 2004;Yonezawa et al., 2005). Although there is a clear correlation between HIV-1mediated recruitment, CD4 and co-receptor polarization, and efficient viral fusion and infection (Iyengar et al., 1998;Jolly et al., 2004;Manes et al., 2000), little is known about the molecular mechanisms regulating the CD4-co-receptor interaction. Recently, it has been reported that filamin A links HIV-1 receptors to the actin cytoskeleton to allow their clustering (Jimenez-Baranda et al., 2007).
HIV-1 viral entry occurs at specific cell-surface areas enriched in viral receptors, such as ruffles and microvilli (Singer et al., 2001;Steffens and Hope, 2003). The behavior of these structures is governed by cortical actin dynamics, which in turn depend on the activity of several actin cytoskeleton associated proteins such as those responsible for actin filament growth and capping (Mangeat et al., 1999). Ezrin-radixin-moesin (ERM) proteins provide an inducible and reversible link between membraneassociated proteins and the actin cytoskeleton (Bretscher, 1999;Mangeat et al., 1999), thereby regulating microvilli formation (Takeuchi et al., 1994). Soluble ERM proteins in the cytoplasm do not display plasma membrane and actin crosslinking capacity. This inactive state of ERM proteins is controlled by intramolecular associations between the N-terminal FERM domain (band 4.1ezrin-radixin-moesin) and the C-terminal domain (Chishti et al., 1998;Pearson et al., 2000). ERM activation requires dissociation of these domains to facilitate other intermolecular interactions (Gary and Bretscher, 1995;Hirao et al., 1996;Matsui et al., 1998). Phosphorylation of C-terminal threonine residues in ERM (moesin Thr558 and ezrin Thr567) is indicative of the activation of the ERM molecules (Fievet et al., 2004), which in their active conformation connect cortical F-actin to the plasma membrane. The FERM domain also interacts with the cytoplasmic tails of several integral membrane proteins such as CD43, CD44, VCAM1, and ICAM1, ICAM2 and ICAM3, and mediates their cell-surface clustering (Barreiro et al., 2002;Heiska et al., 1998;Helander et al., 1996;Hirao et al., 1996;Matsui et al., 1998;Serrador et al., 1997;Takeuchi et al., 1994;Yonemura et al., 1993). The C-terminal domain of ERM molecules binds to actin filaments (Algrain et al., 1993).

The human immunodeficiency virus 1 (HIV-1) envelope regulates the initial attachment of viral particles to target cells through its association with CD4 and either CXCR4 or CCR5. Although F-actin is required for CD4 and CXCR4 redistribution, little is known about the molecular mechanisms underlying this fundamental process in HIV infection. Using CD4 + CXCR4 + permissive human leukemic CEM T cells and primary lymphocytes, we have investigated whether HIV-1 Env might promote viral entry and infection by activating ERM
(ezrin-radixin-moesin) proteins to regulate F-actin reorganization and CD4/CXCR4 co-clustering. The interaction of the X4-tropic protein HIV-1 gp120 with CD4 augments ezrin and moesin phosphorylation in human permissive T cells, thereby regulating ezrin-moesin activation. Moreover, the association and clustering of CD4-CXCR4 induced by HIV-1 gp120 requires moesin-mediated anchoring of actin in the plasma membrane. Suppression of moesin expression with dominant-negative N-moesin or specific moesin silencing impedes reorganization of F-actin and HIV-1 entry and infection mediated by the HIV-1 envelope protein complex. Therefore, we propose that activated moesin promotes F-actin redistribution and CD4-CXCR4 clustering and is also required for efficient X4-tropic HIV-1 infection in permissive lymphocytes.
Analyses of HIV-1 viral preparations indicate that cleaved ezrin and moesin fragments, as well as other cytoskeletal components, are found inside the virions (Lapham et al., 1996). Recently, ezrin has been shown to have a tubulin-dependent role in HIV-1 infection (Haedicke et al., 2008). ERM proteins seem to exert pleiotropic effects on HIV-1 infection, which are dependent of the tropism of the HIV-1 viral strain (Kubo et al., 2008). Although all these data suggest a role for ERM proteins during virus budding or infection, there has been no published evidence indicating an actin-dependent role of ERM in the first steps of the viral cycle.
The data presented here show that moesin-dependent reorganization of the actin cytoskeleton is a critical feature of HIV-1 Env-mediated CXCR4-CD4 colocalization and association. Moreover, moesin activity linking the plasma membrane to actin is required for efficient HIV-1 Env-mediated membrane fusion and infection.

HIV-1 Env induces ERM phosphorylation and F-actin redistribution in CD4 + CXCR4 + lymphocytes
It has recently been proposed that ERM proteins require phosphorylation to maintain their active state (Bretscher, 1999;Yonemura et al., 2002). To study the involvement of ERM proteins and F-actin redistribution in HIV-1 infection, we first investigated whether the HIV-1 envelope (Env) gp120 viral protein and/or HIV-1 viral particles activate ERM proteins in permissive primary lymphocytes or human leukemic CEM T cells. When CEM cells were incubated with the X4-tropic recombinant soluble (rs)-gp120 IIIB protein, the phosphorylation of moesin, and to a lesser extent ezrin, increased (Fig. 1A, top panel), whereas total moesin and ezrin expression was unaffected (Fig. 1A, bottom panel). Remarkably, ERM proteins were also phosphorylated during early viral infection of PHA-activated primary T lymphocytes, with a Journal of Cell Science 122 (1) Fig. 1. HIV-1 phosphorylates endogenous ERM and redistributes F-actin. (A) Western blot analysis of rs-gp120-induced phosphorylation of C-terminal Thr residues of ERM proteins (P-ezrin and P-moesin). Results from untreated control CEM cells are shown in the blot below. Total ERM (ezrin and moesin) is shown for all experimental conditions. A representative experiment of three is shown. The graph shows quantification of Thr phosphorylation of ERMs from three independent experiments. (B) Time course of ERM Thr phosphorylation in PHA-activated PBL, during early infection with the X4-tropic HIV-1 NL4.3 viral strain (MOI, 5) compared with non-infected cells. Basal phosphorylation of endogenous ERM-Thr, total ERM and total α-tubulin expression are shown for all experimental conditions. (C) Confocal microscopy shows polarized distribution (xy mid-sections) of activated endogenous ERM in CEM cells, 1 hour after HIV-1 infection (MOI, 1; HIV-1NL4.3). Untreated, non-infected cells. Quantification of HIV-1-induced ERM or ERM-P capping is indicated ± s.d. (D) Confocal analysis of ERM-P and endogenous ezrin and moesin localization and redistribution in non-infected or HIV-1-infected CEM cells (1 hour; MOI, 1). (E) Confocal analysis of F-actin and endogenous ERM localization and redistribution in non-infected or HIV-1infected CEM cells (1 hour; MOI, 1). Quantification of HIV-1-induced ERM actin capping is indicated ± s.d. (F) Western blot analysis of ezrin and moesin phosphorylation induced by rs-gp120 IIIB (1 hour exposure) in CEM cells. Blots of total expression of ezrin, moesin and α-tubulin are also shown. Rho-K (ROCK) and PKC inhibitors failed to prevent rs-gp120 IIIB -induced Thr phosphorylation of ezrin and moesin. A representative of three independent experiments is shown. (G) Blockade of rs-gp120 IIIB (5 μg/ml)induced ERM Thr phosphorylation in CEM cells with neutralizing or non-neutralizing anti-CD4 mAbs (OKT4A or OKT4, respectively), or with the CXCR4 antagonist AMD3100. Blots of total ezrin and moesin proteins are shown. A representative example of three independent experiments is shown. peak 90 minutes after infection, which then later declined to basal levels ( Fig. 1B). Confocal analysis during early HIV-1 viral exposure showed that the X4-tropic HIV-1 NL4.3 strain induces Thr phosphorylation of ERM proteins (Fig. 1C). Endogenously phosphorylated ERM displayed a polarized distribution within 1 hour of viral infection (Fig. 1C). This polarized distribution and Thr phosphorylation is mainly due to moesin, not ezrin, as indicated by the specific colocalization of moesin protein and phosphorylated threonine (Thr-P) (Fig. 1D). Analysis of F-actin organization showed that F-actin and endogenous moesin are cortically distributed in non-infected cells, with no capping observed (Fig.  1E), and that during early HIV-1 infection or after exposure to rs-gp120, F-actin redistributes to a pole in about 30% of cells ( Fig.  1E and data not shown).
ERM C-terminal Thr residues are targets for phosphorylation in vitro by Rho kinase (ROCK) and PKC (Nakamura et al., 1999). We therefore assessed whether these kinases were responsible for the HIV-1-induced ERM phosphorylation. Phosphorylation of moesin and ezrin in response to treatment with rs-gp120 IIIB or rs-gp120 SF2 was not affected by the specific ROCK inhibitor Y-27632 or by the PKC inhibitors Ro-31-8220 and Gö-6976 (Fig. 1F), demonstrating that HIV-1 Env-induced moesin and ezrin phosphorylation is not mediated by ROCK or PKC in lymphocytes. Moreover, neither the CXCR4 antagonist AMD3100 nor a nonneutralizing anti-CD4 OKT4 mAb affected rs-gp120 IIIB -induced moesin and ezrin phosphorylation (Fig. 1G). By contrast, pretreatment of cells with a neutralizing anti-CD4 mAb (OKT4A) abolished rs-gp120 IIIB -mediated ERM activation (Fig. 1G). Active ERM-P (Thr-P) was monitored with a specific Thr-P antibody and Alexa Fluor 568-labeled secondary antibody. The yz and xz planes are shown for each xy mid-section presented (arrows in the right panel). Data are from three independent experiments, presented as means ± s.e.m. Quantification of basal (untreated) and HIV-1 Env-mediated co-distribution of ERM-P and ERM is shown in parentheses; the percentages represent the number of cells showing co-distribution per 200 cells counted. (C) Distribution of F-actin and nucleofected moesin-GFP proteins in non-infected or HIV-1-infected (1 hour; MOI, 1) CEM cells. Quantification of HIV-1 Env-mediated co-distribution of endogenous moesin, nucleofected moesin-GFP constructs and F-actin are shown in parentheses. Scale bars: 10 μm. (D) Quantified flow cytometry of F-actin in non-infected CEM T cells that were either untransfected (Control) or nucleofected with GFP or FL-, N-or C-moesin-GFP. Data are from three independent experiments, presented as mean ± s.e.m.
Therefore, it appears that HIV-1 Env induces moesin and ezrin phosphorylation and activation through specific CD4 engagement, ruling out the involvement of CXCR4.

Active moesin drives F-actin redistribution during initial HIV-1 to cell contacts
To further investigate the role of moesin in HIV-1 Envmediated F-actin reorganization, we transiently nucleofected CEM cells with one of three C-terminally GFP-tagged moesin constructs: FL-moesin (full-length moesin), N-moesin (the N-terminal FERM domain), or C-moesin (the C-terminal actin-binding region) (Amieva et al., 1999). These cells were then incubated with HIV-1 viral particles. Remarkably, overexpression of functional FL-moesin increased HIV-1 Envmediated phosphorylation of endogenous ERM ( Fig.  2A), thus suggesting a cooperative effect during moesin activation (Simons et al., 1998), which could reflect enhanced anchoring of cortical F-actin filaments to plasma membrane. In non-infected cells, endogenous moesin was restricted to the cell cortex ( , which has been previously described as a dominant-negative moesin construct that lacks the capacity to bind F-actin, thereby disconnecting plasma membrane from cortical actin (Amieva et al., 1999). Moreover, the dominant-negative N-moesin construct inhibited HIV-1 Env-induced phosphorylation of endogenous moesin ( Fig. 2A). These results indicate that moesin protein is activated and phosphorylated early during HIV-1 infection.
In cells overexpressing FL-moesin-GFP or Cmoesin-GFP, F-actin redistribution was readily observed during HIV-1 infection or treatment with rs-gp120 viral protein (Fig. 2C, and data not shown). By contrast, N-moesin abolished HIV-1-induced F-actin redistribution (Fig. 2C). However, F-actin levels, as determined by flow cytometry (Vicente-Manzanares et al., 2004), were not altered significantly by overexpression of any of the moesin constructs ( Fig.  2D). Therefore, it seems that active moesin drives Factin redistribution during the first HIV-1 to cell contacts.

Moesin regulates X4-tropic HIV-1 infection in permissive T cells
We next assessed the involvement of moesin in HIV-1 infection. We first confirmed the effect of moesin on F-actin distribution by suppressing endogenous moesin expression. Expression of ezrin, moesin or both proteins was suppressed with specific shortinterfering RNAs (siRNAs) (Fig. 3A) (oligonucleotides 1E, 2M or 2M+1E; an alternative oligo for moesin was also used) (see supplementary material Fig. S1A). Knockdown of moesin expression blocked HIV-1 Env-induced F-actin redistribution (Fig.  3B). Neither overexpression of the GFP-tagged moesin constructs nor knockdown of ezrin, moesin, or both proteins affected the cellsurface expression of CD4 and CXCR4 receptors by primary lymphocytes (Fig. 3C,D).
To determine whether moesin is involved in X4-tropic HIV-1 infection, we nucleofected CEM cells or PHA-activated peripheral blood lymphocytes (PBLs) with one of the three C-terminal GFPtagged moesin constructs and infected the cells with the X4-tropic Journal of Cell Science 122 (1) HIV-1 NL4.3 strain. Whereas overexpression of FL-moesin strongly enhanced HIV-1 infection, the level of infection with C-moesin was similar to non-transfected or GFP-transfected permissive cells (Fig.  4A,B). CEM cells or primary T cells overexpressing the N-moesin product were less infected than controls (Fig. 4A,B). This finding was confirmed by infection of permissive moesin-silenced CEM cells; suppression of endogenous moesin, but not ezrin, inhibited HIV-1 infection ( Fig. 4C; supplementary material Fig. S1B). The fact that the ezrin protein appears not to be involved in HIV-1mediated viral infection might be explained by the low expression of this actin adaptor in T cells (Serrador et al., 1997) (and data not shown).
We next studied the involvement of moesin in the early steps of HIV-1 infection. To exclude any influence of possible effects at late steps in the infection cycle, we infected cells with a single-cycle virus bearing either the reporter gene Luciferase (Luc-HIV-1)  or the reporter enzyme β-lactamase (Blam-HIV-1) (Cavrois et al., 2002). Non-replicative Luc-HIV-1 particles were incubated with CEM permissive cells or primary T lymphoblasts. Overexpression of FL-moesin potentiated the Luc-HIV-1 signal, whereas it was inhibited by overexpression of dominant-negative N-moesin (Fig. 4D). Therefore, the early steps of HIV-1 infection are regulated by moesin. Non-replicative Blam-HIV-1 virions were incubated with CEM permissive cells silenced for moesin to analyze the specific role of moesin in HIV-1 entry. Moesin knockdown rendered cells less susceptible to Blam-HIV-1 entry (Fig. 4E), indicating that moesin is necessary for the efficient entry of HIV-1. This process is mediated by the HIV-1 envelope because VSV-G virions penetrated equally in untreated and moesinsilenced target cells (Fig. 4E).
Moesin regulates X4-tropic HIV-1-mediated cell fusion and promotes fusion pore formation The effect of the different moesin constructs on HIV-1-Envdependent cell fusion was assessed in a quantitative T fusion assay, in which CEM T cells are the target for infection by Jurkat-Hxbc2 cells expressing functional X4-tropic Env. Overexpression of the N-moesin in CEM cells blocked cell fusion similarly to the neutralizing anti-CD4 mAb. By contrast, FL-moesin increased syncytia formation two-to threefold; overexpression of C-moesin did not significantly affect Env-mediated membrane fusion (Fig.  5A,B). Similar results were obtained in a HeLa-cell-based X4-tropic Env-induced fusion model (Fig. 5C).
We also assessed the involvement of ERM proteins in HIV-1 Env-mediated cell fusion by specific knockdown of lymphocyte ezrin and/or moesin protein expression. Suppression of endogenous moesin expression, but not ezrin, inhibited HIV-1 Env-mediated cell fusion (Fig. 5D).
To define the mechanism of moesin in fusion pore formation, we quantitatively assayed F-actin and moesin redistribution, as well Journal of Cell Science 122 (1) as dye diffusion from charged-Hxbc2-Env + cells to fused target cells. After 30 minutes, F-actin was recruited to the Env + cell-target cell contact area (Fig. 5E). ERM-P also accumulated in this region. Overexpression of functional FL-moesin accelerated dye diffusion from Env + cells to fused target cells, indicating enhanced membrane fusion. Moreover, F-actin recruitment to cell-cell contact areas was enhanced in cells expressing FL-moesin. By contrast, overexpression of N-moesin blocked F-actin redistribution and moesin phosphorylation at cell-cell contact areas. Moreover, Nmoesin abrogated dye diffusion from Env + cells to associated T cells, indicating failure of membrane fusion. C-moesin did not affect F-actin redistribution or fusion pore formation. These results indicate that virus-activated moesin regulates an early step in HIV-1 infection by controlling fusion-pore formation, which is related to the capacity to reorganize F-actin and to promote HIV-1-Envmediated CD4-CXCR4 interaction.
Clustering of CD4 and CXCR4 receptors during the first HIV-1 to cell contacts was assessed by confocal microscopy. HIV-1 induced an early co-distribution of CD4-CXCR4 (Fig. 6C). To further analyze how HIV-1 induces this polarization and clustering, the capping of the redistributed molecules was quantified over time by microscopy. ERMs and F-actin were the first to be recruited to the capping area, followed by CD4 and CXCR4 (Fig. 6D). Enhanced co-distribution of CD4-CXCR4 was observed in cells overexpressing FL-moesin construct (Fig. 6F), whereas the clustering in GFP-transfected cells was the same as in untransfected controls (Fig. 6E). The effect of C-moesin on receptor clustering varied in different experiments, inducing a slight enhancement of clustering (mean average 27% in C-moesin-transfected cells compared with 20% in control cells). However, no other functional effects were exerted by C-moesin in any other assays. Consistently, overexpression of N-moesin (Fig. 6E,F) or moesin mRNA silencing (Fig. 6G) impaired CD4 and CXCR4 redistribution at capping domains. Therefore, moesin regulates HIV-1-induced CD4 and CXCR4 clustering in permissive lymphocytes during the first viruscell contacts, and this event correlates with the requirement of moesin activation for HIV-1 membrane fusion and infection.

Discussion
We have shown that HIV-1 activates endogenous ERM proteins during early viral infection. The rs-gp120 viral envelope proteins activate moesin and ezrin through interaction with CD4, but not with the CXCR4 co-receptor. Moesin phosphorylation in response to HIV-1 to T cell contact is required for cell fusion, and this is related to the capacity of phosphorylated moesin to promote CD4-CXCR4 interaction and the redistribution of F-actin and CD4-CXCR4 to sites of T-cell-virus contact. Jimenez-Baranda et al. have recently described a role for Filamin A, another actin binding protein, in the clustering of CD4-CXCR4 facilitated by the actin cytoskeleton. It is conceivable that several actin-binding proteins contribute to the regulation of transmembrane protein clustering through the actin cytoskeleton to provide the cell with the plasticity needed to respond to the environment.
The identity of the kinase responsible for moesin phosphorylation is not known. ROCK and PKC can phosphorylate ERM Thr residues in vitro (Bretscher, 1999;Nakamura et al., 1999), but the in vivo relevance of this has been controversial (Bretscher, 1999;Matsui et al., 1999;Yonemura et al., 2002). Results of the present study suggest that these kinases are not involved in HIV-1-mediated ERM phosphorylation. Increased moesin phosphorylation has been correlated with enhanced association of moesin with the cortical actin cytoskeleton (Simons et al., 1998). Moesin may thus potentiate HIV-1 Env-mediated F-actin to plasma membrane anchoring events at sites of virus to cell contact. Further studies are needed to identify the kinase or kinases that phosphorylate the C-terminal ERM-Thr in vivo in this Env-triggered CD4-signaling pathway.
Viral receptors on the surface of non-infected T cells rapidly polarize after contact with HIV-1-infected cells (Piguet and Sattentau, 2004), but the molecular signaling pathway that triggers this recruitment remains to be elucidated. The constitutive association between CD4 and CXCR4 (Lapham et al., 1999;Wang et al., 2004) is augmented after gp120-CD4 engagement. Remarkably, functional FL-moesin strongly enhanced the HIV-1 gp120-induced CD4-CXCR4 interaction and co-clustering at one pole of the lymphocyte, whereas dominant-negative N-moesin or moesin knockdown impaired this association. An important finding was that N-moesin overexpression inhibits HIV-1 Env-induced phosphorylation of endogenous moesin, thereby blocking rs-gp120or HIV-1-induced F-actin reorganization and CD4-CXCR4 capping and interaction. Silencing of moesin impairs HIV-1 Env-mediated CD4-CXCR4 interaction, and inhibits HIV-1 viral entry and infection. By contrast, in T cells overexpressing FL-moesin, HIV-1 particles and rs-gp120 viral protein induce the formation of prominent pseudopodia, to which F-actin and CD4 and CXCR4 viral receptors redistributed. These FL-moesin-overexpressing cells were readily infected by HIV-1 replicative and non-replicative viral particles. It is thus conceivable that moesin controls the interaction of Env with CD4-CXCR4 by increasing the density of CD4-CXCR4 clusters at one pole of the cell. In turn, this would favor viral synapse and fusion-pore formation and therefore HIV-1 entry and infection.
F-actin disorganization negatively affects CD4-CXCR4 clustering and infection by X4-tropic viruses (Iyengar et al., 1998), and efficient HIV-1 viral spreading requires an intact F-actin skeleton (Jolly et al., 2007). It has recently been described that moesin and ezrin limit HIV-1 viral replication by affecting tubulin cytoskeleton at a post-entry step (Haedicke et al., 2008;Naghavi et al., 2007). Strikingly, in that study, dominant-negative N-moesin was equally as effective as the functional full-length protein at blocking viral replication, which apparently contradicts previous knowledge about how these proteins exert their function. Although the mechanism of inhibition by N-moesin, acting as a dominant negative, is not completely known, it is thought that the inhibitory activity of constructs containing only the FERM domain relies on the capacity of this domain to interact with the C-terminal regions of active endogenous ERM, thereby preventing their interaction with F-actin (reviewed by Bretscher, 1999). Therefore, when HIV-1 Env triggers moesin activation, a large fraction of the C-terminal exposed regions of endogenous moesin would be sequestered by the overexpressed FERM domain of the N-moesin construct. In agreement with our data, ERM proteins have very recently been identified as positive regulators of X4-tropic HIV-1 infection in non-lymphoid cell lines (Kubo et al., 2008). However, the molecular mechanism of the effects on HIV infection described in the study was not explored. Here, our study provides compelling evidence that ERM proteins regulate HIV-1-mediated membrane fusion and infection through the promotion of F-actin redistribution and the molecular association Journal of Cell Science 122 (1) Fig. 6. Moesin drives HIV-1 Env-mediated CD4-CXCR4 interaction and polarized redistribution, during early viral infection. (A) Effect of FL-and N-moesin constructs on rs-gp120 IIIB (10 μg/mL)-induced CD4/CXCR4 direct association. Immunoprecipitation assays were performed with anti-CD4 OKT4 mAb. Inducible co-immunoprecipitation of both receptors was measured as the ratio between the intensities of CXCR4 and CD4 immunoblotted bands. A representative experiment of three is shown. (B) Effect of the silencing of endogenous moesin on HIV-1 Env-induced direct CD4-CXCR4 interaction, in CEM cells. One representative experiment of three is shown. (C) Confocal microscopy of CD4 (anti-CD4 mAb HP2/6) and CXCR4 (biotin-conjugated mAb anti-CXCR4 12G5) in non-infected (Untreated) or 1 hour HIV-1-infected CEM cells (MOI, 1). The yz and xz planes are shown for each xy mid-section presented (arrows). Scale bars: 10 μm. (D) Percentage of colocalization of ERM and CD4, Actin or CXCR4, and of CD4 and CXCR4 in the presence of HIV-1 (MOI, 3) during a tim-course assay. (E) CD4 and CXCR4 redistribution in 1 hour HIV-1-infected cells overexpressing the different moesin-GFP proteins. xy maximal projections (left) and xy midsections (right) are shown. Scale bars: 10 μm. (F) Quantification of HIV-1-induced CD4-CXCR4 co-distribution in CEM T cells overexpressing FL-moesin-GFP and N-moesin-GFP constructs. (G) Quantification of HIV-1-induced CD4-CXCR4 capping in untransfected cells (Control), cells overexpressing the scrambled siRNA or lacking endogenous moesin (siRNA-moesin oligo 2). In F and G, quantification of HIV-1-induced CD4-CXCR4 capping is indicated as a percentage of each 200 and 100 cells counted, respectively, under each experimental condition. and clustering of CD4 and CXCR4. Our results obtained from infection with single-cycle, non-replicative HIV-1 particles exclude effects that require late gene expression, and confirm the involvement of moesin in HIV-1 viral entry.
Our work suggests a model of moesin involvement in X4-tropic HIV-1 membrane fusion and infection of lymphocytes. In this model, HIV-1 virus activates lymphocyte moesin, as monitored by phosphorylation of Thr558. Activated moesin stabilizes and reorganizes F-actin, which may be necessary for CD4-CXCR4 colocalization at one pole of the cell. Activated moesin thus drives the HIV-1 Env-induced CD4-CXCR4 interaction. This process increases the probability of HIV-1 Env-CD4-CXCR4 interactions, potentiating fusion pore formation and thus HIV-1 viral fusion and infection. It is also conceivable that moesin-directed F-actin reorganization might drive the final steps of lipid mixing, perhaps by producing mechanical strain on the lipid bilayer (Eitzen, 2003), thus favoring viral fusion. In this regard, open-activated moesin allows F-actin anchoring in the plasma membrane, which is enhanced by FL-moesin and prevented by N-moesin. Interfering with this process inhibits HIV-1 viral fusion and infection. Therefore, the function of lymphocyte moesin is required for efficient HIV-1 viral fusion and infection.

Western blotting
Treated cells were resuspended in 60 μl MES buffer (10 mM MES at pH 7.4, 150 mM NaCl, 5 mM EGTA, 5 mM MgCl 2 , 1 mM Na 3 VO 4 , 1 mM NaF, and a protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany), boiled for 5 minutes and immunoblotted with specific antibodies. Protein bands were analyzed using the LAS-1000 CCD system and Image Gauge 3.4 software (Fuji Photo Film Co., Tokyo, Japan).

Actin polymerization assay
Actin polymerization was analyzed as described previously (Vicente-Manzanares et al., 2004). Briefly, 24 hours after transfection, CEM T cells were fixed to stop intracellular actin polymerization with a solution containing 4% formaldehyde in PBS, 1% Triton X-100, and 5 μg/ml Alexa Fluor 647-phalloidin from Molecular Probes (Eugene, OR). Cells were incubated for 15 minutes at 37°C and washed once in PBS, and the intracellular polymerized actin was then determined with a FACScan flow cytometer (BD Biosciences, Mountain View, CA) using CellQuest software.

HIV-1 viral preparation and infection
Preparation of HIV-1 NL4.3 and measurement of viral replication were performed as described (Valenzuela-Fernandez et al., 2005). Highly infectious preparations of HIV-1 NL4.3 viral strain were generated by several consecutive passages of the original HIV-1 isolates in peripheral blood mononuclear cells (PBMCs). Briefly, PBMCs were infected with one synchronous dose of HIV-1 NL4.3 , and culture supernatants were recovered 3 days later and stored at -70°C. Freshly thawed aliquots were filtered through 0.22 μm filters before use. HIV-1 NL4.3 entry at multiplicity of infection (MOI=1) was assayed in phytohemagglutinin (PHA; 1 μg/ml)-activated PBLs or CEM T cells for 90 minutes. Cells were then trypsinized and extensively washed with fresh medium to remove viral input. Infected cells were kept in culture, and viral entry and infection was monitored every 48 hours by measuring the concentration of p24 in the culture supernatant by enzyme-linked immunosorbent assay (Innotest HIV-1 antigen mAb; Innogenetic, Ghent, Belgium). When indicated, permissive cells were pretreated with anti-CD4 mAbs (5 μg/ml) or 3Ј-azido-2Ј,3Ј-dideoxythymidine (AZT) (5 μg/ml) before addition of virus.

Luciferase virus assay
Luciferase-HIV-1 viral particles deficient for replication were kindly provided by Suryaram Gummuluru (Boston University, Boston, MA). Replication-deficient viral particles were produced by transfecting a luciferase-expressing reporter virus, HIV/Δnef/Δenv/luc + , which contains the luciferase gene inserted into the nef ORF and does not express env glycoprotein (Yamashita and Emerman, 2004), with a CXCR4-tropic (Lai) env glycoprotein. Virus stocks were generated by PolyFect transient transfection of HEK293T cells . Two days after transfection, cell-free virus-containing supernatants were clarified of cell debris and concentrated by centrifugation (16,000 g 1 hour at 4°C) and stored at -80°C until required. HIV-1 virus preparations were titrated by ELISA and determination of the p24 Gag content. Untreated or nucleofected CEM or PBLs activated over 2 days with PHA (1 μg/ml), were infected with a synchronous dose of luciferase-based virus for 2 hours. Virus was removed by washing infected cells. After 32 hours of infection, luciferase activity was determined with a luciferase assay kit (Promega Corporation, Madison, WI) and a 1450 Microbeta Luminescence Counter (Walax, Trilux). Protein contents were measured by the bicinchoninic acid method (BCA protein assay kit from Pierce, Rockford, IL) according to the manufacturer's instructions.
Production of non-replicative viral particles containing BlaM-Vpr X4-tropic HIV-1 viral particles deficient for replication and containing the BlaM-Vpr chimera were produced by co-transfecting HEK293T cells with the following vectors: pNL4-3.Luc.R-E (20 μg; NIH-AIDS Reagent Program); CXCR4-tropic (HXB2-env; NIH-AIDS Reagent Program) env glycoprotein vector (10 μg) and the pCMV-BlaM-Vpr vector (10 μg). The BlaM-Vpr chimera was kindly provided by Warner C. Greene (University of California, San Francisco, CA). Co-transduction of the pNL4-3.Luc.R-E-(20 μg) vector with the pHEF-VSV-G (10 μg; NIH-AIDS Reagent Program) and pCMV-BlaM-Vpr (10 μg) vectors was used to generate nonreplicative viral particles that fuse with cells in a VSV-G-dependent manner. Viral plasmids were transduced in HEK293T cells by using linear polyethylenimine with an average molecular mass of 25 kDa (PEI25k) (Polyscience, Warrington, PA). The PEI25k was prepared as a 1 mg/ml solution in water and adjusted to neutral pH. After addition of PEI25k to the viral plasmids (at a plasmid: PEI25k ratio of 1:5 w/w), the solution was mixed immediately, incubated for 20-30 minutes at room temperature and then added to HEK293T cells in culture. Viruses were harvested 40 hours after transfection. The supernatant was clarified by centrifugation at 3000 rpm for 30 minutes. Viral stocks were normalized by p24-Gag content measured by ELISA (Innogenetics, Gent, Belgium).
Virion-based fusion assay 1ϫ10 6 CEM T permissive cells were incubated for 3 hours with equivalent viral inputs of BlaM-Vpr-containing virions (500 ng of p24) in 500 μl RPMI-1640 medium. Cells were then extensively washed to remove free virions and incubated (1 hour, room temperature) with CCF2-AM loading mix, as recommended by the manufacturer (GeneBLAzer TM detection kit; Invitrogen, Carlsbad, CA). Next, excess dye was washed off and cells were incubated for 16 hours at room temperature prior to fixation with 4% paraformaldehyde. The percentages of cells infected were determined by measuring the fluorescence intensities of intact and cleaved CCF2 probe in virusinfected CCF2-loaded target cells in a fluorescence spectrophotometer (Cary Eclipse, Varian; Melbourne, Australia). An increase in the ratio of blue (447 nm; cleaved CCF2) to green (520 nm; intact CCF2) fluorescent signals indicates more virions fused to target cells. The background blue and green fluorescence was determined in non-infected CCF2-loaded cells (without β-lactamase activity).

CXCR4-CD4 capping assay
For capping assays, cells were co-cultured with the HIV-1 NL4.3 strain (MOI, 1) or with rs-gp120 IIIB (5 μg/ml) for 1 hour at 37°C as described previously (Gordon-Alonso et al., 2006). Viral particles were then washed out and the cells were fixed with 2% paraformaldehyde. CD4 and CXCR4 were detected by immunofluorescence, and molecule distribution was analyzed with a Leica TCS-SP confocal microscope (Leica. Heidelberg, Germany).

HIV-1 Env-mediated cell-cell fusion assays
A dual-fluorescence cell-fusion assay was also performed as described previously (Gordon-Alonso et al., 2006). Briefly, CMTMR-loaded Env+Jurkat-Hxbc2 cells were mixed with Calcein-AM-loaded parental or transfected CEM cells. The double-labeled cells were detected 6 or 12 hours later by flow cytometry. The target cells were then washed extensively to remove the drugs before co-culturing with Env-expressing cells. Anti-CD4 mAb (5 μg/ml for 30 minutes at 37°C) was used as a control for the blockage of cell fusion. The extent of fusion was calculated as the percentage fusion=(number of bound cells positive for both dyes/number of bound cells positive for calcein-labeled target cells) ϫ 100. The β-Galactosidase and X-Gal staining cell fusion assays in the HeLa system were performed as described (Gordon-Alonso et al., 2006).