Role of Abl Kinase and the Wave2 Signaling Complex in HIV-1 Entry at a Post-Hemifusion Step

Entry of human immunodeficiency virus type 1 (HIV-1) commences with binding of the envelope glycoprotein (Env) to the receptor CD4, and one of two coreceptors, CXCR4 or CCR5. Env-mediated signaling through coreceptor results in Gαq-mediated Rac activation and actin cytoskeleton rearrangements necessary for fusion. Guanine nucleotide exchange factors (GEFs) activate Rac and regulate its downstream protein effectors. In this study we show that Env-induced Rac activation is mediated by the Rac GEF Tiam-1, which associates with the adaptor protein IRSp53 to link Rac to the Wave2 complex. Rac and the tyrosine kinase Abl then activate the Wave2 complex and promote Arp2/3-dependent actin polymerization. Env-mediated cell-cell fusion, virus-cell fusion and HIV-1 infection are dependent on Tiam-1, Abl, IRSp53, Wave2, and Arp3 as shown by attenuation of fusion and infection in cells expressing siRNA targeted to these signaling components. HIV-1 Env-dependent cell-cell fusion, virus-cell fusion and infection were also inhibited by Abl kinase inhibitors, imatinib, nilotinib, and dasatinib. Treatment of cells with Abl kinase inhibitors did not affect cell viability or surface expression of CD4 and CCR5. Similar results with inhibitors and siRNAs were obtained when Env-dependent cell-cell fusion, virus-cell fusion or infection was measured, and when cell lines or primary cells were the target. Using membrane curving agents and fluorescence microscopy, we showed that inhibition of Abl kinase activity arrests fusion at the hemifusion (lipid mixing) step, suggesting a role for Abl-mediated actin remodeling in pore formation and expansion. These results suggest a potential utility of Abl kinase inhibitors to treat HIV-1 infected patients.


Introduction
HIV-1 enters cells in a pH-independent manner by fusion at the plasma membrane or from within endosomes [1][2][3]. HIV-1 entry requires multiple conformational changes in the HIV-1 glycoprotein, and rearrangement of the actin cytoskeleton. These events are triggered by binding of the viral envelope (Env) surface subunit gp120 to the primary receptor CD4 and one of two chemokine coreceptors, CCR5 or CXCR4 [1,4]. This interaction activates signaling events in the cell, similar to those initiated by natural ligands, such as Ca 2+ mobilization, activation of RhoGTPases, and phosphorylation of tyrosine kinases, pyk2, Zap70 and p56lck [4][5][6]. Rho family GTPases, which include the Cdc42, Rac, and Rho subfamilies, are responsible for regulating signaling from membrane receptors to the actin cytoskeleton. The Rho sub-family stimulates myosin based contractility, and drives the formation of stress fibers and focal adhesions. The Rac sub-family stimulates lamellipodia and membrane ruffles, and the Cdc42 subfamily stimulates the formation of filopodia [7][8][9]. We showed that HIV-1 Env binding to target cells induces activation of Rac, stimulates membrane ruffles and lamellipodia, and fusion is inhibited by dominant negative Rac [4,10]. Furthermore, HIV-1 Env-induced Rac activation depends on activation of Gaq, phospholipase C (PLC), Ca 2+ mobilization, protein kinase C (PKC), pyk2 and the GTPase Ras [5]. In the current study we identified the fusion-specific effectors of Rac required for actin cytoskeleton rearrangements that mediate membrane fusion and entry.
Guanine nucleotide exchange factors (GEFs) activate GTPases, facilitating the GDP to GTP switch, and regulate their downstream effects by participating in scaffolding protein complexes, thereby linking GTPase activity to specific effectors [7][8][9]. HIV-1 Env-induced Rac activation is mediated by a specific Rac GEF, either Tiam-1 or Trio [10,11]. There are multiple effectors of Rac, including serine/threonine kinases, lipid kinases, actin-binding proteins, and adaptor/scaffold molecules [7,12]. PAK is a downstream effector of Rac and Cdc42 that promotes stabilization of actin networks. Another downstream effector of Rac that nucleates actin polymerization is the Arp2/3 complex. The Arp2/3 complex is activated by the Wave2 complex through IRSp53, an adaptor protein that binds Rac and Wave2 [7]. The Wave2 complex includes Rac-associated protein 1, Nck-associated protein, Abl-interacting protein 2, and heat shock protein C300. Wave2 also associates with Abl, and Abl-mediated phosphorylation of Wave2 promotes its activation [13,14]. In addition to determining which Rac effectors are critical for membrane fusion, we studied the steps in the membrane fusion process affected by these signaling molecules. These data demonstrate that the Wave2 signaling complex and Abl are required for Env-mediated membrane fusion, entry, and infection and that Abl kinase inhibitors arrest the fusion process at hemifusion.

HIV-1 Env-Mediated Fusion Depends on the Wave2 Signaling Complex
To determine whether Abl, Trio, or Tiam-1 were required for HIV-1 Env-mediated cell-cell fusion, expression of these proteins was down regulated by RNA interference (RNAi) in U87.CD4.CCR5 cells. Cells expressing siRNA were then mixed with BSC40 cells expressing different Env subtypes and Envdependent cell-cell fusion was measured. Transfection of target cells with siRNA to Tiam-1 and Abl decreased levels of Envmediated cell-cell fusion by an average of 7965% and 7465% respectively for both HIV-1 R5 and dual-tropic Env-subtypes ( Figure 1A, left). There was no significant fusion observed with CCR5 expressing target cells and X4 Env expressing cells with or without siRNA, as expected. The decrease in the levels of fusion correlated well with the decreased steady-state level of Tiam-1, and Abl as detected by immunoblot ( Figure 1C). A siRNA directed against Trio had no effect on Env-induced cell-cell fusion despite a 70% reduction in expression of the Trio protein ( Figure 1A and C). To determine whether Tiam-1 and Abl are acting exclusively upstream of Rac, a constitutively active Rac mutant, RacV12 was expressed in siRNA transfected cells. Expression of RacV12 in cells expressing siRNA to Tiam-1 reversed the effects of this siRNA on fusion, suggesting that Tiam-1 is functioning upstream of Rac. In contrast, levels of fusion in cells expressing RacV12 and siRNA to Abl were only 5361% that of cells expressing RacV12 and control siRNA, suggesting a role for Abl upstream and downstream of Rac ( Figure 1A, right).
Tiam-1 binds to the Rac and Cdc42 effector IRSp53, enhancing IRSp53 binding to Rac and activation of the Wave2 scaffolding complex [15]. To determine the role of these Rac effectors in Env-mediated membrane fusion, their expression was down regulated by RNAi in U87.CD4.CCR5 cells. The siRNA expressing cells were mixed with Env-expressing cells and cell-cell fusion was measured. Expression of siRNA to IRSp53, Wave2, and Arp3 decreased fusion by 7465% 7764% and 7864%, respectively. The decrease in fusion with these siRNAs was not overcome by expression of RacV12, suggesting that these proteins are required downstream of Rac ( Figure 1B). The decrease in levels of fusion correlated with the decrease in protein expression in cells expressing these siRNAs, as seen by immunoblot ( Figure 1C), and each siRNA was specific for its target protein ( Figure S1A). Treatment of cells stably expressing siRNA resistant Arp3, with Arp3 targeted siRNA had no effect on Env-mediated cell-cell fusion ( Figure 1D, E). In contrast, with untransfected cells, and cells stably expressing siRNA resistant Arp3, treatment with siRNA to Rac decreased fusion by 7565% and 7663% respectively ( Figure 1D). These results show that the effects of RNAi on fusion were specific to inhibition of their target molecules.
To demonstrate the role of Tiam-1, Abl, Rac, IRSp53, Wave2 and Arp3 in virus-cell fusion, their expression was down regulated by RNAi in TZM-BL cells, a derivative of HeLa cells that express CD4, CCR5, and CXCR4, and these cells were then used in a Vpr-Blam assay [16,17]. In this assay siRNA expressing cells were mixed for 90 min with HIV-1 strains with cores carrying a blactamase (BlaM)-Vpr chimera, and pseudotyped with Env from ADA (R5), YU2 (R5) or HXB2 (X4), and fusion was quantified by measuring the cytosolic activity of viral core-associated BlaM [18]. Expression of siRNA to Tiam-1, Abl, Rac, IRSp53, Wave2, and Arp3 decreased virus-cell fusion by an average of 8064%, 8361%, 7664%, 8266%, 7763% and 8266%, respectively, for HIV-1 R5 and X4 Env subtypes ( Figure 1F). These results show that activation of the Wave2 signaling complex is required for Env-dependent cell-cell fusion and virus-cell fusion.

Small Molecule Inhibitors of Abl Kinase Activity Inhibit HIV-1 Entry
Since treatment of cells with Abl targeted siRNA led to a decrease in Env-dependent cell-cell fusion and virus-cell fusion we wanted to determine whether treatment of target cells with commercially available Abl kinase inhibitors, imatinib (IMB), nilotinib (NIL), and dasatinib (DAS), block fusion. IMB is a relatively specific inhibitor of Bcr-Abl, Abl, Arg, and class III receptor tyrosine kinases. NIL is an Abl kinase inhibitor 20-50 fold more potent than IMB at inhibiting Abl. DAS, originally designed as a Src family kinase inhibitor, antagonizes Abl, ephrin and platelet-derived growth factor receptor kinases, and kit. DAS is 300 fold more potent than IMB at inhibiting Abl [19,20]. To determine the concentrations of these Abl kinase inhibitors that inhibit Abl kinase activity and Env-mediated cell-cell fusion, without non-specific effects, Abl kinase activity, trypan blue analysis, vaccinia virus infection, and T7 polymerase activity were measured in addition to Env-dependent cell-cell fusion ( Figure  S1B, S2, and data not shown). Treatment of U87.CD4.CCR5 cells with 10 uM IMB, 500 nM NIL, and 300 nM DAS for 1 h prior to and during 3 h incubation with Env-expressing cells decreased Env-mediated cell-cell fusion by an average of 9562%, 9265%, and 9266%, respectively, and Abl kinase activity by 85-87% (Figure 2A and S1B). The CCR5 inhibitor TAK-779, which completely blocks Env-mediated cell-cell fusion and infection of CCR5 expressing cells, was included as a control, and it decreased Env-dependent cell-cell fusion by 9961% and Env-mediated Abl kinase activation by 98% (Figure 2A and S1B). Similar results were observed with U87.CD4.CXCR4 cells treated with CXCR4 inhibitor AMD3100 and Abl kinase inhibitors and incubated with cells expressing HIV-1 X4 or dual-tropic Env subtypes ( Figure  S3A). There was no decrease in T7 polymerase activity, or localization of CD4 and CCR5 on the cell surface ( Figure S4 and data not shown). Expression of RacV12 in U87.CD4.CCR5 cells treated with IMB, NIL and DAS increased the level of fusion by an average of 3.5-fold (*, P,0.05) compared to treated cells without RacV12, suggesting a role of Abl kinase activity upstream of Rac ( Figure 2B).
To determine the effect of these Abl kinase inhibitors on Envinduced Rac activation, U87.CD4.CCR5 cells were treated with inhibitors for 1 h prior to mixing with BSC40 cells expressing no HIV-1 Env, HIV-1 X4 Env, or HIV-1 R5 Env for 30 minutes in

Author Summary
Patients infected with HIV-1 are currently treated with highly active antiretroviral therapy (HAART) that efficiently suppresses the virus but does not cure the infection. HIV-1 envelope activates Rac-mediated actin cytoskeleton rearrangements in the target cell that promote membrane fusion and entry. We discovered that these rearrangements require activation of the actin polymerization machinery including the tyrosine kinase Abl. We also showed that Abl kinase inhibitors imatinib, nilotinib, and dasatinib, current drug therapies for chronic myeloid leukemia, block HIV-1 entry and infection. These results suggest that these inhibitors might be appropriate drugs for treatment of HIV-1. This strategy of using inhibitors that disable host signaling proteins rather than viral proteins, essential for pathogen survival, may have a general efficacy in developing drugs to combat HIV-1 and other pathogens that acquire drug resistance.
the presence of inhibitor. The mismatched X4 Env, that does not induce Rac activation in CCR5 expressing cells, and the CCR5 inhibitor TAK-779, which completely blocks Env-mediated Rac activation in CCR5 expressing cells, were included as controls [5]. Env-induced Rac activation was abolished in cells treated with TAK-779, and all three of the Abl kinase inhibitors ( Figure 2C). To validate these effects in a relevant HIV-1 target cell, peripheral blood lymphocytes (PBLs), which express CD4, CCR5 and CXCR4, were used as the target cell population in an Envdependent cell-cell fusion assay. Treatment of PBLs with IMB, NIL, and DAS decreased fusion by an average of 9261%, 9263%, and 99.561%, respectively, for HIV-1 R5, dual-tropic and X4 Env subtypes ( Figure 2D). The CCR5 inhibitor TAK-779, as expected, completely blocked fusion mediated by R5 Envexpressing cells, inhibited fusion mediated by dual-tropic Env by 5662%, and had no effect on fusion mediated by X4 Env ( Figure 2D).
A long term infection assay was also performed where PBLs were infected with 150 ng of the X4 HIV HXB2 virus after 1 h preincubation with no inhibitor, DMSO, 10 mM IMB, 250 nM NIL, or 75 nM DAS. After 3 h, virus and inhibitors were washed off, inhibitors were added back and the plate was incubated at 37u for 21 days with addition of the inhibitors every 24 h. After 21 days the samples were assayed for cell viability and p24 antigen content. Treatment with IMB, NIL, and DAS decreased cell viability of HIV HXB2 infected cells by 1764%, 865%, and 863% respectively and decreased infection by 52%, 51% and 94% compared to DMSO treated cells ( Figure S5).
To validate the specificity of these effects, we performed an Envdependent cell-cell fusion assay with cells stably expressing two different drug resistant Bcr-Abl mutants (Y253F and T315I), or expressing wild type (WT) Bcr-Abl [21]. Expression of the drug resistant Bcr-Abl mutants but not WT Bcr-Abl resulted in recovery of fusion ( Figure 2E), demonstrating that the effects of these inhibitors on Env-dependent cell-cell fusion are specific to inhibition of Abl.
To confirm these results using virus particles with relevant levels of virus-associated glycoprotein, we used a virus-dependent cellcell fusion assay based on the ability of virus particles to bridge two cells and allow transfer of cytoplasmic contents, and we also used the Vpr-BlaM assay described above [4,10]. For the virusdependent cell-cell fusion assay we used two populations of U87.CD4.CCR5 cells, one expressing the T7 polymerase and the other expressing the b-galactosidase (b-gal) gene under the T7 promoter. Both populations were incubated with inhibitors for 1 h prior to 3 h incubation with R5 virus HIV YU2 . In this assay, controls included untreated and inhibitor treated cells that were not incubated with virus, the CCR5 inhibitor TAK-779, and T-20 which blocks entry by inhibiting the conformational change in HIV-1 gp41 required for fusion [17]. R5 Virus-dependent cell-cell fusion was reduced by an average of 9463% in cells treated with IMB, DAS, and NIL compared to cells treated with DMSO alone, and treatment with TAK-779 and T-20 completely inhibited fusion ( Figure 2F). Treatment of U87.CD4.CXCR4 cells incubated with the X4 virus HIV HXB2 with AMD3100 IMB, NIL, and DAS decreased virus-dependent cell-cell fusion by 8867%, 98.661%, 8765%, and 96617%, respectively ( Figure S3B).
For the Vpr-BlaM assay, TZM-BL cells were treated with 1 mM AMD3100, 1 mM TAK-779, 10 mM IMB, 500 nM NIL and 150 nM DAS for 1 hr prior to and during the 90 min incubation CD4.CCR5 cells were infected with vCB21R alone, or with (B) vRacV12 overnight, then treated with DMSO alone, TAK-779, IMB, NIL, or DAS for 1 h and the inhibitors were also present during 3 h incubation with HIV-1 Env-expressing cells and b-gal activity as measured. (C) U87.CD4.CCR5 cells were treated for 1 h with TAK-779, IMB, NIL, or DAS and during 30 min incubation with BSC40 cells expressing no Env (subtracted as background), HIV ADA Env or HIV HXB2 Env. Whole cell lysates were analyzed by Rac specific G-LISA activation assay. Average A490 of triplicate wells 6 standard deviation are shown. (D) PBMCs were infected with vCB21R in complete media overnight, treated with DMSO, TAK-779, IMB, NIL, or DAS for 1 h prior to addition of HIV-1 Env-expressing cells and b-gal activity was measured. (E) U87.CD4.CCR5 cells engineered to express indicated clones of Bcr-Abl were treated with Abl inhibitors and HIV UNC or HIV ADA Env as described above or analyzed by Western blot with anti-Abl or anti-actin antibody (inset). (F) U87.CD4.CCR5 cells were infected overnight with vCB21R or vPT7-3, then mixed (1:1) in triplicate wells, treated for 1 h with DMSO, TAK-779, T20, IMB, NIL, DAS, and with 100 ng of HIV YU2 for 3 h at 37uC. b-gal activity was measured and cell fusion was normalized using DMSO treated cells mixed with HIV YU2 as 100%. (G) TZM-BL cells were treated with DMSO, 1 mM TAK-779 or AMD3100, 10 mM IMB, 500 nM NIL, or 150 nM DAS for 1 h prior to 90 min incubation with indicated HIV viruses and BlaM activity was measured. Activity was normalized using DMSO treated cells mixed with HIV HXB2 virus as 100%. All data are representative of results from three similar experiments performed in triplicate. doi:10.1371/journal.ppat.1000956.g002 with HIV-1 Vpr-BlaM viruses expressing R5 and X4-tropic Env. AMD3100 treatment decreased X4-Vpr-BlaM activity by 8461%, but had no effect on R5-Vpr-BlaM activity. TAK-779 treatment decreased R5-Vpr-BlaM activity by an average of 8962%, but had no effect on X4-Vpr-BlaM activity, as expected. However, treatment of TZM-BL cells with IMB, NIL, and DAS decreased virus-cell fusion by an average of 8164%, 8965%, and 9061%, respectively, for both HIV-1 R5 and X4 Env subtypes ( Figure 2G). These results together with the results of the Envdependent and virus-cell fusion assay demonstrate that Abl kinase is required for HIV-1 entry mediated by CXCR4 and CCR5.

Infection of TZM-BL Cells with HIV-1 Particles, but not Particles Pseudotyped with Amphotropic Murine Leukemia Virus (A-MLV) Env or Vesicular Stomatitis Virus Glycoprotein (VSV-G), Depends on Abl and the Wave2 Signaling Complex
To determine whether the Wave2 signaling complex and Abl are required exclusively for HIV-1 entry, or virus-induced fusion and infection in general, we examined infection with HIV-1 versus A-MLV Env (A-MLV-ENV-HIV-1) or VSV-G pseudotyped HIV-1 (VSV-G-HIV-1) using the TZM-BL assay. HIV-1 Env induces pH independent virus-cell fusion to facilitate entry, whereas viruses pseudotyped with VSV-G or A-MLV Env induce pH-dependent clathrin mediated endocytosis or caveolin-mediated endocytosis, respectively [22][23][24][25]. TZM-BL cells, a derivative of HeLa cells that express CD4, CCR5, CXCR4, and luciferase (luc) under the control of the HIV-1 LTR, were pretreated with the 10 mM IMB, 500 nM NIL and 150 nM DAS for 1 h prior to incubation with virus for 3 h, and a subsequent 24 h incubation with inhibitor only [16,17]. The CCR5 inhibitor TAK-779, the CXCR4 inhibitor AMD3100, and ammonium chloride (NH 4 Cl) which inhibits endosomal acidification required for VSV-G mediated entry, were included as controls [22,23,26]. The top two panels of Figure 3A shows that treatment with IMB, NIL, and DAS decreased infection with R5 HIV YU2 virus and X4 HIV HXB2 virus by an average of 9167%, 8864%, and 9165%, respectively, comparable to the reductions observed with Env-dependent cell-cell fusion, virus-dependent cell-cell fusion and virus-cell fusion ( Figure 2). The Abl kinase inhibitors had no effect on infection of TZM-BL cells with A-MLV-ENV-HIV-1 or VSV-G-HIV-1, but treatment of cells with NH 4 Cl blocked infection with VSV-G-HIV-1 as expected ( Figure 3A, bottom two panels). These data show that Abl-kinase inhibitors were able to block HIV-1 Envmediated fusion specifically and had no effect on infection via pHdependent clathrin-mediated or caveolin-mediated endocytosis, and post-entry steps were not affected by these inhibitors.
To test the effect of Wave2 complex targeted siRNAs on infection, TZM-BL cells were transfected with 200 nM control siRNA or siRNA directed towards Tiam-1, Trio, Abl, IRSp53, Wave2 and Arp3. These cells were incubated with virus for 3 h, and media alone for 24 h. The decreased levels of HIV-1 YU2 and HIV-1 HXB2 infection of TZM-BL cells expressing siRNA targeted to Tiam-1, Abl, IRSp53, Wave2, and Arp3 were comparable to levels of Env-mediated cell fusion with U87.CD4.CCR5 cells expressing these siRNAs, whereas siRNA to Trio had no effect ( Figure 3B, top two panels). Steady state levels of target proteins in cells expressing targeted siRNAs were decreased to similar levels as in U87 cells ( Figure 1C and data not shown). Infection of TZM-BL cells with A-MLV-ENV-HIV-1 or VSV-G-HIV-1 was not affected by expression of the targeted siRNAs, suggesting that Tiam-1, Abl, IRSp53, Wave2, and Arp3 are required for HIV-1 Env-mediated entry and are not necessary for post-fusion steps in the virus life cycle ( Figure 3B, bottom 2 panels).

The Abl Kinase Inhibitors Arrest Fusion at the Hemifusion Step
HIV-1 Env-induced fusion, and release of the viral capsid into the cytosol is a multistep process. First, gp120 binds to CD4 inducing conformational changes in gp120, and actin cytoskeletal rearrangements in the target membrane that bring the coreceptor CCR5 or CXCR4 into close proximity with CD4. Next, coreceptor binding to gp120 triggers conformational changes in gp41 to produce a prebundle conformation that inserts into the target cell membrane, allowing lipid mixing or hemifusion, and then pore formation. Additional conformational changes induce formation of the gp41 6-helix-bundle which prevents pore closure and facilitates pore enlargement and full fusion [2,27,28]. To determine which step(s) in the membrane fusion process are blocked by the Abl kinase inhibitors, we examined the effect on infection of membrane curving agents. Oleic acid (OLA), chlorpromazine (CPZ), and trifluoperazine (TFP) are lipid analogs that insert into the inner leaflet of the cell membrane. OLA induces negative curvature in the membrane that promotes formation of a hemifusion intermediate (i.e. lipid mixing), but cannot induce pore formation if there is a block at hemifusion. CPZ and TFP are membrane-permeable weak bases that partition into inner leaflets of cell membranes, induce positive curvature, and relieve a block at hemifusion [29][30][31][32].
To determine the effect of inhibitors and lipid analogs on HIV-1 infection, TZM-BL cells were treated with 1 mM AMD3100, 1 mM TAK-779, 10 mM IMB, 500 nM NIL, and 150 nM DAS for 1 h, prior to and during 1 h incubation with no virus, HIVD ENV , R5 HIV YU2 , X4 HIV HXB2 , A-MLV-ENV-HIV-1, or VSV-G-HIV-1. After 1 h, cells were treated with CPZ or TFP for 1 min or OLA for 5 min, followed by 2 h incubation with inhibitor and virus, and subsequent 24 h incubation with inhibitor only. Addition of CPZ and TFP to cells treated with Abl kinase inhibitors and infected with HIV YU2 or HIV HXB2 resulted in an 8 fold increase in infection compared to inhibitor treated cells infected in the absence of lipid analogs ( Figure 4A), The exogenous cone shaped lipid OLA, which induces negative curvature of the membrane resulting in lipid mixing, had no affect on infection ( Figure 4A). TAK-779 mediated inhibition of HIV YU2 infection and AMD3100 mediated inhibition of HIV HXB2 infection was not affected by these lipid analogs. No increase in luc activity was observed with lipid analog treatment of cells infected with HIVD ENV versus no virus, indicating that Env is required to observe an increase in infection ( Figure S6A). Treatment of A-MLV-ENV-HIV-1 and VSV-G-HIV-1 infected cells with CPZ and TFP decreased overall infection by 2 fold and had no effect on cells treated with Abl kinase inhibitors, indicating that the increase in HIV-1 infection observed with Abl kinase inhibitor treated cells was specific ( Figure 4A, lower panels). CPZ also partially reversed the inhibitory effects of nilotinib as measured by the Vpr-BlaM assay ( Figure S6B). Similar increases in virus-dependent cell-cell fusion were observed when U87.CD4.CCR5 cells were treated with inhibitors and lipid analogs and HIV YU2 mediated fusion was measured after 3 h ( Figure S6C). Cells were also incubated with the lipid analogs in the absence of HIV YU2 to account for the effects of these agents on the cells and on T7 polymerase activity. Addition of OLA did not increase fusion in cells treated with any of the inhibitors ( Figure S6B). To confirm the results obtained with the Abl kinase inhibitors we incubated TZM-BL cells transfected with Tiam-1, Abl, Rac, IRSp53, Wave2, and Arp3 targeted siRNA, for 1 h with no virus, HIVD ENV , R5 HIV YU2, or X4 HIV HXB2 . After 1 h cells were treated with CPZ for 1 min or OLA for 5 min, followed by 2 h incubation with virus, and subsequent 24 h incubation with media alone. As with the Abl kinase inhibitors, treatment of siRNA transfected cells with CPZ increased infection by an average 8.4 fold compared to untreated cells, and OLA had no effect ( Figure 4B). These results suggest that inhibition of Tiam-1, Abl, Rac, IRSp53, Wave2 or Arp3 arrests fusion at hemifusion, preventing pore formation, pore enlargement and content mixing.
To confirm that Abl kinase inhibitors cause arrest at hemifusion, we used a modification of a fusion assay described previously [33]. To demonstrate the effects of the lipid analog CPZ on HIV-1 Env mediated cell-cell fusion and to observe the effect of CPZ and the Abl kinase inhibitors on A-MLV Env or VSV-G induced cellcell fusion we treated U87.CD4.CCR5 cells with DMSO, TAK-779, or IMB for 1 hr prior to incubation with CHO-K1 cells expressing no Env, HIV ADA , A-MLV Env or VSV-G for 1 hr. After 1 h cells were treated with CPZ for 1 min and OLA for 5 min, then washed and incubated with inhibitor for an additional 2 h prior to fixation and GM1 staining. Incubation of IMB treated cells with HIV ADA and CPZ promoted the transition from hemifusion to full fusion as expected ( Figure S8). Fusion of A-MLV Env and VSV-G Env expressing cells with U87.CD4.CCR5 cells was unaffected by treatment with IMB or CPZ ( Figure S9) and all Env-mediated fusion was unaffected by OLA treatment (data not shown). These results confirm that Abl kinase activity is required at a post-hemifusion step for HIV-1 Env mediated fusion and entry.

Discussion
Dynamic regulation of the actin cytoskeleton is required for fusion of biological membranes. Multiple reports have demonstrated that actin remodeling is required for HIV-1 mediated fusion and entry [4,5,10,11,[34][35][36]. Some studies showed that treatment of target cells expressing physiologically relevant levels of receptor and coreceptor with the actin filament capping drug cytochalasin D prevented the formation of the gp120-CD4coreceptor complex [35,37,38]. Another more recent study, demonstrated a role for CD4 and coreceptor-mediated filamin-A interactions in receptor clustering that is dependent on RhoA and ROCK mediated phosphorylation of ADF/cofilin [34]. Previous work from our lab with the actin filament stabilizing drug jasplakinolide and the actin monomer sequestering drug latrunculin A (LA) suggested a role for actin remodeling at a post binding step in fusion [4]. To further substantiate the role of actin polymerization in HIV-1 entry, we treated cells with 1 mM LA and 5 mM latrunculin B (LB). Both drugs blocked HIV-1 fusion for multiple cell types, as measured by the Env-dependent cell-cell fusion assay, the virus-dependent cell-cell fusion assay, the viruscell fusion assay, and infection ( Figure S10).
Our previous data demonstrated that the GTPase Rac was activated by HIV-1 Env ligation of CCR5, resulting in membrane ruffles and lamellipodia in the target cell membrane. Inhibition of this activation by dominant negative Rac or by a Rac GEF inhibitor completely abolished Env-dependent cell-cell fusion, virus dependent cell-cell fusion and infection [4,5,10,11]. Our lab went on to show that Env-induced Rac activation is mediated by Gaq and its downstream effectors, including Ras. Other studies showed that Ras promotes Rac activation via direct interaction with Tiam-1, or by phosphatidylinositol 3-kinase (PI3K)-mediated activation of Tiam-1 [39]. Env-dependent Rac activation likely occurs through the first mechanism, since treatment of target cells with PI3K inhibitors had no effect on Env-dependent cell-cell fusion [40].
The nonreceptor tyrosine kinase, Abl, modulates actin upstream and downstream of Rac [41,42]. In the current study, we used siRNAs and specific inhibitors to show that the activity of Abl kinase is required both upstream and downstream of Rac for Envinduced membrane fusion. Upstream of Rac, Abl phosphorylation of the Ras GEF complex promotes the activity of the Rac GEF Tiam-1, which was shown in the current study to be required for HIV-1 fusion. Downstream of Rac, Abl promotes phosphorylation and activation of Wave2 and its interaction with the Arp2/3 complex, events also demonstrated here to be critical for HIV-1 infection, but not VSV-G or A-MLV Env-mediated infection. These results suggest that these signaling mediators are important There is some conflict in the literature as to the location and mechanism of virus cell fusion. A recent report used microscopic imaging to track HIV-1 Env-pseudotyped MLV virus particles and observed virus-membrane fusion in endosomes [3]. This study also showed that virus-cell fusion and infection were inhibited in the presence of the dynamin inhibitor dynasore (DYN) which is known to block both clathrin and caveolin-mediated endocytosis [3]. The results in our current study suggest that fusion is occurring via a mechanism that is distinct from that of VSV (clathrin-mediated endocytosis) or A-MLV (caveolin-mediated endocytosis). In order to address this conundrum, we treated cells with the dynamin inhibitor DYN, and then used these cells for the Env-dependent cell-cell fusion assay, the virus-dependent cell-cell fusion assay, the virus-cell fusion assay and the TZM-BL infection assay. DYN treatment decreased HIV-1 Env-mediated infection and virus-cell fusion by an average of 5867% and 5063% respectively ( Figure  S10 and Figure S11). However treatment with DYN decreased A-MLV-Env-HIV-1 infection and VSV-G-HIV-1 infection by 7565% and 8961% respectively, showing that the affect on HIV-1 Env-mediated infection was not as significant ( Figure S11). DYN treatment also decreased Env-dependent cell-cell fusion and virus-dependent cell-cell fusion by 5368% and 50610%, respectively which was unexpected since these assays both measure cell-cell plasma membrane fusion. Dynamins are a group of large GTPases that are involved in multiple processes in addition to endocytosis, such as vesicle transport, cytokinesis, organelle division, cell movement and cell signaling [43][44][45]. Therefore, the inhibition observed with the dynamin inhibitor DYN could be due to nonspecific effects on cellular processes. In support of this conclusion, a recent study used the Rev-dependent indicator cell line Rev-CEM to study the effects of DYN on HIV-1 replication and VSV-G-HIV-1 infection [44]. Using this assay they observed a dosage dependent decrease in VSV-G-HIV-1 infection with DYN treatment but did not see any decrease in HIV-1 infection [44]. These results as well as the results in Figure 3 show a clear distinction between HIV-1 Env-mediated entry and VSV-G-and A-MLV-mediated entry.
The current study also showed that the block in fusion caused by inhibition of Tiam-1, Abl, Rac, IRSp53, Wave2 and Arp3 occurs after hemifusion and before cytoplasmic mixing. This conclusion was based on the 1) confocal microscopy demonstration that addition of IMB to the fusion reaction allowed membrane but not cytoplasmic mixing, and 2) observation that lipid analogs that overcome a block at hemifusion overcame inhibition of HIV-1 virus dependent cell fusion, virus-cell fusion and infection caused by Abl kinase inhibitors and siRNA expression. These results support a model whereby HIV-1 Env binding to CCR5 stimulates activation of Gaq resulting in activation of Rac and activated Rac interacts with IRSp53. IRSp53 promotes Rac activation of the Wave2 complex, which is also activated by Abl, and activated Wave2 induces subsequent activation of Arp2/3-mediated actin rearrangements which facilitate pore formation, pore enlargement, and entry of HIV-1.
Many microbial pathogens depend on Abl family kinases to mediate efficient infection of their targeted host, including Shigella flexneri, enteropathogenic Escherichia coli, Helicobacter pylori, Anaplasma phagocytophilum, coxsackievirus, poxvirus, and murine AIDS virus. Abl kinases are involved in pathogen entry, intracellular movement, and exit from target cells; proliferation of target cells; and phosphorylation of microbial effectors. Many of these processes involve reorganization of the target cell actin cytoskeleton and depend on the same signaling pathways as HIV-1 [4,5,46]. Discovery of these signaling mediators as fundamental components of microbial pathogenesis provides new targets for therapeutic intervention. The clinical application of IMB, NIL, and DAS, which block deregulated Abl kinases in leukemia patients, demonstrate that inhibition in vivo is possible with manageable side effects [19,20]. In addition IMB has been shown to be an effective inhibitor of antiapoptotic pathways induced by HIV-1 in macrophages [47]. Most current antiviral therapies target viral proteins and mutation of the virus leads to therapy resistance. Therefore, using inhibitors that target host signaling proteins essential for HIV-1 entry may be an efficient new strategy for treatment of infected patients.

Reagents and Cell Lines
U87.CD4.CCR5 cells are astroglioma cells expressing CD4, CCR5-GFP or HA-CCR5. U87.CD4.CXCR4 cells are astroglioma cells expressing CD4 and CXCR4-GFP. CHO-K1 cells (ATCC) were grown in F-12K media with 10% serum and other cells maintained as described [48]. pMSCVneo-WT, Y253F, and T315I Bcr-Abl were gifts from Dr. R. Van Etten [21]. The siRNA resistant mutations were generated in Arp3 based on sequences obtained from Santa Cruz Biotechnology, Inc (SCBT, Santa Cruz, CA,), by PCR-mediated mutagenesis of a sub fragment that was sequenced to confirm the presence of mutations before sub cloning into the corresponding cDNA. WT and mutant cDNAs were cloned in pcDNA3.1 + zeo for expression by transduction. IMB, NIL, and DAS were from LC Laboratories and were used at 10 uM, 500 nM, and 300 nM respectively unless indicated; CPZ (0.5 mM), TFP (0.3 mM), OA (100 nM), OLA (50 uM) and NH 4 Cl (50 mM) were from Sigma; TAK-779 (1 uM), and T-20 (10 ug/ml) were from the AIDS Research and Reference Reagent Program. The control siRNA constructs (non-targeting 20-25 nt siRNA designed as a negative control), the siRNA constructs and antibodies used for Western blots were from SCBT [5]. The siRNA constructs were transfected using GeneEraser siRNA Transfection Reagent or Lipofectamine RNAiMAX Transfection Reagent according to the manufacturer's instructions (Stratagene, La Jolla, CA, Invitrogen, Carlsbad, CA).

BlaM Assay for Virus-Cell Fusion
TZM-bl cells were serum starved for 24-36 h then plated (4610 4 cells/well) in 96-well plates in complete media overnight. Cells were treated with indicated concentrations of inhibitors for 1 hr prior to and during 90 min incubation with DEAE-dextran (20 mg/ml) alone or DEAE-dextran (20 mg/ml) and 150 ng p24 HIV YU2 Vpr-BlaM, HIV ADA Vpr-BlaM, or HIV HXB2 Vpr-BlaM. After 90 min virus and media were aspirated off cells and 100 ul 1X Lysis and Detection Solution was added to wells (LyticBlazer-BODIPY FL, Invitrogen). The plate was incubated at room temperature in the dark overnight. The BlaM activity was quantified using TECAN fluorescence plate reader (Tecan, Switzerland). The extent of virus-cell fusion was measured with excitation centered at 485 nm and emission centered at 535 nm. The green signal for samples incubated with no inhibitors or inhibitors and no virus was subtracted as background from their respective virus treated samples.

TZM-BL Assay
TZM-BL cells were serum starved for 12-24 h then plated overnight in complete media in 96 well plate at 2610 4 cells per well. Cells were treated for 1 h with indicated concentrations of inhibitors prior to addition of media alone or 150 ng p24 of HIV YU2 HIV HXB2 or VSVG or A-MLV-pseudotyped HIV in the presence of 20 ug/ml DEAE-dextran for 3 h at 37uC. After 3 h cells were washed 3 times with PBS and inhibitors were added in fresh media. Following a 24 h incubation cells were lysed and luciferase (luc) units determined. Infected wells and uninfected wells with inhibitor were compared to wells with no inhibitor. For the TZM-Bl assay with lipid analogs serum starved TZM-BL cells were treated with inhibitors for 1 h, then 150 ng of indicated virus was added for 1 h prior to treatment with CPZ or TFP for 1 min or OLA for 5 min. Cells were washed three times with PBS and virus and inhibitors were added back. After 2 h cells were washed two times with PBS and incubated in inhibitor overnight and luc activities were measured.

Long-Term PBMC Infection and Viability Assay
PBMCs that were isolated and stimulated as previously described [5]. They were plated at 5610 5 cells per well in 96 well plate and were treated with 10 mM IMB, 250 nM NIL, or 75 nM DAS for 1 h prior to addition of 150 ng p24 of HIV HXB2 in the presence of 20 ug/ml DEAE-dextran for 3 h at 37uC. After 3 h cells were washed three times with PBS and incubated in inhibitor for 24 h. Inhibitors were added back at the same concentration every 24 h for three weeks. 100 ul of supernatant was collected every fourth day and all samples were assayed for p24 antigen content by enzyme-linked immunosorbent assay. Two separate plates were set up under the exact same conditions and one plate was used for p24 measurement and the other was incubated with 20 ul cell viability substrate per 100 ul of sample (Promega, Madison, WI).

Fusion and Hemifusion Assays
Envelope-mediated and virus-dependent fusion assays were described. Average fusion compared to untreated control reactions were detected by b-galactosidase activity 6 standard deviation [5]. To account for any effect of inhibitors on vaccinia virus infection and/or on T7 polymerase function, vCB21R and vPT7-3 coinfected cells were similarly treated with inhibitors. Concentration curves were performed with all of the inhibitors to determine the concentration that resulted in the maximum decrease in fusion without altering vaccinia virus infection or T7 polymerase activity. Hemifusion assays were performed with 2610 6 CHO-K1 cells nucleofected with a GFP expression plasmid, and after 24 h infected with vaccinia virus expressing HIV ADA Env or no Env. After 16 h, 4610 5 U87.CD4.CCR5.HA cells were added for 3 h, fixed with paraformaldehyde, stained with TRITC-conjugated CTX-555 (EMD), and analyzed on a 510 Meta LSM confocal microscope.

Statistical Analysis
Fusion and infectivity results were compared using a two-tailed t-test. All p values, unless indicated, were ,0.03.   Figure S3 Abl kinase is required for X4 Env-dependent and X4 virus-dependent cell-cell fusion. For Env-dependent fusion assays, U87.CD4.CXCR4 cells were infected with vCB21R overnight then treated with DMSO alone, AMD3100, IMB, NIL, or DAS for 1 h and the inhibitors were also present during 3 h incubation with HIV-1 Env-expressing cells and b-gal activity as measured. For X4 virus-dependent fusion assays, U87.CD4.CXCR4 cells were infected overnight with vCB21R or vPT7-3, then mixed (1:1) in triplicate wells, treated for 1 h with DMSO, AMD3100, IMB, NIL, DAS, and with 250 ng of HIVHXB2 for 5 h at 37uC. b-gal activity was measured and cell fusion was normalized using DMSO treated cells mixed with HIV HXB2 as 100%. After another 16 h, CHO-K1 cells were overlayed for 3 h with U87.CD4.CCR5 cells pre-treated for 1 h with DMSO, 1 mM TAK-779, or 10 mM IMB. Cells were fixed and stained with TRITC-conjugated CTX (CTX-555, red), and counterstained with TO-PRO3 (blue). Images were collected using an oil objective (magnification X63). Images were cropped but relative cell size was maintained. Found at: doi:10.1371/journal.ppat.1000956.s007 (7.02 MB TIF) Figure S8 IMB induced arrest at hemifusion step is overcome with membrane curving lipid analog CPZ. U87.CD4.CCR5 cells were pre-treated for 1 h with DMSO, 1 mM TAK-779, or 10 mM IMB, incubated with CHO-K1 cells expressing no Env or HIV ADA Env for 1 h, treated with CPZ for 1 min, then washed and inhibitors were added back for additional 2 h incubation at 37u. Cells were fixed and stained with TRITC-conjugated CTX (CTX-555, red), and counterstained with TO-PRO3 (blue). Images were collected using an oil objective (magnification X63). Images were cropped but relative cell size was maintained. Found at: doi:10.1371/journal.ppat.1000956.s008 (7.06 MB DOC) Figure S9 A-MLV Env-and VSV-G-induced cell-cell fusion is unaffected by IMB or CPZ. U87.CD4.CCR5 cells were pretreated for 1 h with DMSO or 10 mM IMB, incubated with CHO-K1 cells expressing A-MLV Env or VSV-G for 1 h, treated with CPZ for 1 min, then washed and inhibitors were added back for additional 2 h incubation at 37u. Cells were fixed and stained with TRITC-conjugated CTX (CTX-555, red), and counterstained with TO-PRO3 (blue). Images were collected using an oil objective (magnification X63). Images were cropped but relative cell size was maintained.