HIV-1 Nef Triggers Macrophage Fusion in a p61Hck- and Protease-Dependent Manner Free

HIV-1 infection leads to the development of AIDS due to the degeneration of the immune system (1). Macrophages, an early cellular target of HIV-1, contribute importantly to the establishment of viral reservoirs (for reviews, see Refs. 2–4), because they contain and produce viruses for long periods of time. They even accumulate replication-competent viruses in patients receiving highly active anti-retroviral treatment (5). Because macrophages have the ability to migrate through anatomic barriers, they are also involved in HIV-1 dissemination. In addition, multinucleated giant cells (MGCs) formed by HIV-1–infected macrophages have been identified in several tissues of AIDS patients, such as in lymphoid organs, colon, and brain (6–9). In brain, the presence of giant macrophages has been associated with HIV-induced dementia (10) and is considered as a hallmark of AIDS progression (11, 12).

Nef, an HIV-1 accessory protein, is expressed abundantly in the early stages of the viral life cycle and plays a key role in the progression of infection toward AIDS in primates (13–15). Among Nef functions, its ability to affect cell surface protein trafficking, including CD4 or MHC I molecules, has been well characterized (16). Nef also influences intracellular signaling pathways by physically interacting with host cell proteins, thus modifying their localizations and functions (for reviews, see Refs. 17, 18). Src family protein tyrosine kinases (PTKs) are major targets of Nef. HIV-1 Nef interacts with the Src homology 3 (SH3) domains of Src PTKs through a proline-rich (P72xxP75) motif in its core domain, and substitution of a single amino acid in the Nef PxxP motif was sufficient to protect HIV-1 transgenic mice from the development of AIDS-like disease (19, 20). Binding of Nef to SH3 domains can inhibit the T cell-specific Lck kinase activity or activates the phagocyte-specific kinase hematopoietic cell kinase (Hck) (21). The binding affinity between Nef and Hck is the highest among the SH3-mediated interactions (22, 23); however, very little is known about the functional consequences of Hck activation by Nef in macrophages. It has been shown to interfere with M-CSF signaling (24, 25) and to activate Stat3 (26). Moreover, in a long-term HIV-positive nonprogressor, Trible et al. (27) identified an unusual Nef variant that failed to activate Hck, indicating that Hck/Nef interaction is important for AIDS progression.

Hck is expressed as two isoforms generated in equal amounts by alternative translation of a single mRNA (28). p59Hck and p61Hck differ by an additional 21-aa sequence at the N-terminal end of p61Hck that influences their subcellular localizations and functions (for review, see Ref. 29). Although p59Hck is found at the plasma membrane where its activation induces the formation of actin-dependent cell protrusions, p61Hck localizes at the cytoplasmic face of lysosomes and triggers the formation of F-actin–rich podosome rosettes (30–32). Hck was shown to play a role in the HIV-1 life cycle as: 1) expression of a dominant-negative Hck mutant inhibits HIV-1 entry and infectivity (33, 34), 2) pharmacological downregulation of Hck inhibits macrophage-tropic HIV-1 replication (35), and 3) AIDS-like symptoms in HIV-1 transgenic mice expressing Nef are delayed when they are crossed with hck^−/− mice (20). These results suggest that Hck is involved in some essential steps of the virus life cycle and may influence the development of immunological defects during HIV-1 infection.

Thus, we decided to further examine the functional consequences of Hck activation by Nef in macrophages. Human primary macrophages were infected with wild-type (wt) and nef-deleted HIV-1 (Δnef-HIV-1) for functional comparison. Unexpectedly, we observed that Nef was involved in HIV-1–induced giant macrophage formation and thus we focused on this phenomenon. We found that Nef-triggered macrophage fusion was dependent on p61Hck and protease activities.

4-Amino-5-(-4-methylphenyl)-7-(tert-butyl)pyrazolo[3,4-d]-pyrimidine (PP1) and 4-amino-7-phenylpyrazol[3,4-d]pyrimidine (PP3) were purchased from TEBU-Bio (Le Perray en Yvelines, France). Recombinant human and mouse IFN-γ (used at concentrations of 50 U/ml with RAW264.7 cells and 100 U/ml with human macrophages) and recombinant M-CSF were purchased from PeproTech (London, U.K.). Aprotinin (2 μM from a 1 mM stock in water), leupeptin (2 μM), pepstatin A (2 μM), Con A (used at concentrations of 2.5 μg/ml with RAW264.7 cells and 5 μg/ml with human macrophages), E64C (100 μM from a 50 mM stock in DMSO), Bafilomycin A1 (1 nM and 100 pM from a 5 μM stock solution in DMSO), and DAPI were purchased from Sigma-Aldrich (Saint Quentin Fallavier, France). GM6001 (5 μM from a 10 mM stock in DMSO) was purchased from VWR International (Strasbourg, France). Rabbit polyclonal anti-Hck Abs (sc-72) were from Santa Cruz Biotechnology (TEBU-Bio, France), monoclonal anti-actin and anti–α-tubulin were from Sigma-Aldrich, secondary HRP-conjugated Abs were from Bio-Rad (Hercules, CA), monoclonal anti-hemagglutinin (HA) Abs were from Eurogentec (Seraing, Belgium), and monoclonal anti-p24 Abs were from DakoCytomation (Carpinteria, CA). Secondary Abs and Texas Red/Alexa Fluor 488/Alexa Fluor 633-coupled phalloidins were obtained from Molecular Probes (Invitrogen, Cergy Pontoise, France).

The cathepsin D-red fluorescent protein (RFP) construct was kindly provided by Dr. F. Darchen (Institut de Biologie Physico-Chimique, Paris, France). The constructs encoding HA-tagged Nef and GFP-tagged Nef and vectors for expression of wt Nef or mutated Nef_NL4-3 (NefAxxA, NefG2A, and NefE4A) fused to GFP were constructed as described (36–38). The construct encoding GFP-tagged NefSF2 was kindly provided by Dr. O. Fackler (University of Heidelberg, Heidelberg, Germany) and constructed as described in Ref. 39, and HA-tagged SIV Nef was a gift from Dr. H. Gottlinger (University of Massachusetts Medical School, Dartmouth, MA) and constructed as described in Ref. 40. The constructs encoding human p59Hck and p61Hck mutants have been subcloned into pEGFP and described elsewhere (30, 31, 41), except the p61Hck-dsRed and the kinase-dead (kd) variant of p61Hck. p61Hck-dsRed was obtained by replacing the GFP sequence of the p61Hck-GFP construct described above with the dsRed sequence of a pdsRedExpress1 vector kindly provided by P. Jurdic (Institut de Génomique Fonctionnelle, Lyon, France). The p61Hck^kd-GFP construct was obtained by inverse PCR on the p61Hck-GFP (31) vector by mutating the Lys²⁹⁰ (AAG) responsible of ATP binding into glutamic acid (GAG) and introducing an XcmI site. Two Hck small hairpin RNA (shRNA) constructs were used. Hck shRNA-1 (used in Fig. 3) was purchased from OpenBiosystems (Huntsville, AL) (www.openbiosystems.com/Query/?i=0&q=HCK reference V2LHS_133011). Hairpin sequence for V2LHS_133011 is 5′-tgctgttgacagtgagc gcgggctacatccc aagcaactatagtg aagccacagatgtata gttgcttgggatgtagcccttgcctactgcctcgga-3′. For the Hck shRNA-2 construct, a pair of 68-nt oligonucleotides encoding a 21-nt Hck shRNA (5′-ctagttccaaaaa ccgtatgcctcgaccagataa tctcttgaa ttatctggtcgaggcatacgg cggg-3′ and 5′-ctagttccaaaaa ccgtatgcctcgaccagataa tctcttgaa ttatctggtcgaggcatacgg cggg-3′) were designed that contained extra SalI and XbaI overhangs after annealing to facilitate cloning. The position of the core 21-nt sequence (underlined) targeted nucleotides 1536–1556 of murine Hck [accession no. Y00487, www.ncbi.nlm.nih.gov.gate1.inist.fr/gene/15162 (42)]. The oligonucleotides were annealed and cloned into the SalI and XbaI sites of the pSUPER-U6 vector. The resulting vector was subsequently cut with ClaI and NheI, and the generated fragment was ligated into the pRRLsinCMV, vector which was cut with ClaI and NheI to remove the CMV promoter, thus creating the pRRLsinU6+27-shHck vector. All of the constructs were sequenced to confirm identity (Millegen, Labège, France).

Control small interfering RNA (siRNA) (siGLO RISC-Free) and Hck siRNAs were purchased from Dharmacon (Denver, CO). A smart pool of four Hck siRNA sequences was tested for optimal efficacy. Each Hck siRNA was also tested individually, and all of them silenced Hck with a comparable efficiency to the smart pool used in the experiments herein.

NIH-3T3 fibroblasts were seeded in 24-well plates (5 × 10³ cells per well) and transfected using DNA-phosphate precipitates as previously described (31, 41).

RAW264.7 and RAW264.7-Hck_Low macrophages were cultured in complete RPMI (Invitrogen) (RPMI supplemented by 10% FCS, 1% l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin). A total of 2 × 10⁶ cells per condition of transfection were transfected with 2 μg DNA and/or shRNA (4 μg in total) using the Cell Line Nucleofector Kit V (Lonza, Levallois-Perret, France) according to the manufacturer’s instructions. Cells were seeded in 24-well plates (1.2 × 10⁵ cells per well), and the medium was renewed. In some experiments, 50 U/ml recombinant mouse IFN-γ and 2.5 μg/ml Con A were added to the culture medium to promote cell fusion.

Human monocytes were isolated from blood of healthy donors as previously described (43), except that monocytes were separated from lymphocytes by CD14 MACS (Miltenyi Biotec, Auburn, CA) according to the manufacturer’s protocol. CD14⁺ monocytes were transfected using the Amaxa nucleofector according to the manufacturer’s instructions. Monocytes were transfected with a fluorescent control siRNA (100 nM) or with a mixture of 50 nM Hck siRNA and 50 nM fluorescent control siRNA to detect transfected cells that reached 90–100% of the cells. Transfected monocytes were seeded in 24-well plates (3 × 10⁵ cells per well) for 24 h. Then, fresh complete medium was added with or without 100 U/ml recombinant human IFN-γ and 5 μg/ml Con A.

Proviral infectious clones of the macrophage-tropic virus isolate ADA and the same clone disrupted for the Nef open reading frame (ADAΔnef) were kindly provided by Luciana da Costa (Federal University, Rio de Janeiro, Brazil) and are described elsewhere (44). Virions were produced by transient transfection of 293T cells with proviral plasmids as described (45). Virus concentration in cell culture supernatants was measured by reverse transcriptase assay as described previously (45). Human monocytes were purified (43) and differentiated with 10 ng/ml M-CSF. After 11 d, monocyte-derived macrophages were seeded on slides in 24-well plates at a density of 1 × 10⁶ cells per well. After an overnight incubation, wt and Δnef ADA viruses (4000 reverse transcriptase cpm/ml) were added. After 4, 6, 8, 10, and 12 d postinfection, 1) half of the cell supernatants were collected, and virus was lysed with 0.5% NP-40 before determination of p24 levels by ELISA (Innotest, Innogenetics, Gent, Belgium), and 2) the slides were collected for immunostainings.

Fluorescent cells were visualized using a LSM710 confocal microscope (Zeiss, Oberkochen, Germany), ×40 1.3 numerical aperture objective (Fig. 1), a DM-RB fluorescence microscope (Figs. 2–4, 6; Supplemental Figs. 2, 4) (Leica Microsystems, Deerfield, IL), or a TCS-SP2 confocal scanning microscope (Leica) (Figs. 5, 7, Supplemental Fig. 3) as previously described (30, 46). In Supplemental Fig. 5, microscopy was performed with Deltavision RT equipment on an IX71 microscope, using a ×100 1.4 numerical aperture objective (Olympus, Melville, NY). Images were obtained with a CoolSnap HQ2 camera (Deltavision RT equipment) and deconvolved, and optical sections projected. All of the images were prepared with Adobe Photoshop.

Cells were lysed (47), and total proteins were separated through 7.5% SDS-PAGE, transferred, and immunoblotted as described (31). In mouse, Hck isoforms are p56Hck and p59Hck. Quantification of the intensity of the immunoblots (reported relative to α-tubulin or actin used as a loading control) was performed using Quantity One 4.5.0 software (Bio-Rad).

Cells were examined by fluorescence microscopy in areas with similar cell densities. All of the cells were counted, and cell fusion index (total cells) was calculated as follows: total number of nuclei in MGCs (cells with ≥2 nuclei)/total number of nuclei × 100. For each condition, at least 200 cells were counted in a double-blind manner. Cell fusion was controlled in some experiments 1) by measuring the number of yellow cells obtained by coincubating Nef-GFP–transfected RAW264.7 cells and mock-transfected RAW264.7 cells stained with PKH26 red fluorescence linker kit (Sigma-Aldrich) according to the manufacturer’s instructions (48), 2) by sequentially diminishing the cell density on plates with a consequent decrease of the fusion index as a result of decreased cell–cell interactions, and 3) by only counting GFP-expressing cells instead of all of the cells and showing that the fusion index for Nef-GFP is 58 ± 7% (n = 4) 48 h after cell transfection.

The cell viability was systematically controlled by Trypan blue exclusion 24 h after the end of experiments with drug treatments.

Statistically significant differences were determined using the Student t test. Differences were considered significant if *p < 0.05, **p < 0.01, or ***p < 0.001.

To study the functional consequences of HIV-1 Nef expression in macrophages, we first examined the phenotypes of human monocyte-derived macrophages infected with wt-HIV-1 or Δnef-HIV-1. F-actin–labeled macrophages were immunostained with p24 Abs to detect infected cells. As previously described (11), wt-HIV-1–infected macrophages formed MGCs. In Δnef-HIV-1–infected macrophages, fewer MGCs were observed (Fig. 1A). The kinetics of viral replication measured by p24 Ag production in cell culture supernatants for wt-HIV-1–infected macrophages and Δnef-HIV-1–infected cells showed a peak at day 8 postinfection (Supplemental Fig. 1B). The formation of wt-HIV-1–infected multinucleated giant macrophages (p24-positive cells) was observed 4 d postinfection and was further increased 8 d postinfection (Fig. 1B, Supplemental Fig. 1A). Therefore, we decided to study the role of Nef in macrophage fusion at day 8 postinfection. The percentage of MGCs (cells with ≥2 nuclei) was decreased by 40% in Δnef-HIV-1–infected macrophages compared with that in wt-HIV-1–infected cells (Fig. 1B, red bars), whereas infection of macrophages with Δnef-HIV-1 was slightly higher than infection with wt-HIV-1 (Fig. 1B, gray bars), as previously reported (49). Further analysis of the number of nuclei per cell showed that Δnef-HIV-1–infected macrophages formed fewer cells with a high number of nuclei than wt-HIV-1–infected macrophages (Fig. 1C).

These results indicate that Nef is involved in the fusion of HIV-1–infected human macrophages.

To further study the role of Nef in MGC formation, we transiently expressed Nef_NL4-3 coupled to GFP in the easily transfectable RAW264.7 mouse macrophage cell line. As shown in Fig. 2A, Nef triggered the formation of multinucleated macrophages. In Nef transfection experiments, the fusion index (total cells) at 48 h was 36 ± 4% compared with GFP-transfected cells with a fusion index of 15 ± 5% (Fig. 2B). In some experiments, the fusion index was calculated only for cells expressing Nef (GFP-positive, 27 ± 6% of total cells) and found to be 58 ± 7%.

Given the multifunctionality of Nef, we wanted to determine whether this novel function of Nef is conserved among Nef proteins from other HIV-1 and SIV isolates. As shown in Fig. 2C, induction of macrophage fusion was similarly obtained when Nef alleles from the HIV-1 isolate SF2 (HIV-1 Nef _SF2) or from SIV were expressed in RAW264.7 cells.

Thus, Nef is sufficient to trigger macrophage fusion independently of other viral proteins.

To examine the possibility that Src PTKs could be involved in Nef-elicited signaling leading to macrophage fusion, we first used pharmacological inhibition of Src kinases using PP1 and 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP2). As shown in Fig. 3A and 3B, PP1 or PP2 (data not shown) decreased Nef-induced macrophage fusion. Quantification of the fusion index indicated that there was a dose-dependent reduction of Nef-induced MGC formation, whereas the nonfunctional analogue PP3 had no effect (Fig. 3B). Because the interaction of Nef with Src kinases involves a PxxP motif (21, 50), we examined the role of a Nef mutant with substitutions in the polyproline sequence (NefAxxA). The cell fusion index was significantly decreased in RAW264.7 macrophages expressing the NefAxxA mutant (21 ± 4%) compared with that in cells expressing wt Nef (39 ± 3%) (Fig. 3C). In macrophages, three kinases of the Src family are expressed: Hck, Lck/Yes-related novel protein, and Gardner-Rasheed feline sarcoma (51). Interaction of Nef with Hck has been described as the highest affinity among the Src family kinases (21). Nef triggers Hck activation, whereas it has no effect on Lck/Yes-related novel protein activity and Gardner-Rasheed feline sarcoma is not a target of Nef (52). Interestingly, Hck activation is the only cellular function of Nef known to be independent of Nef myristoylation (18, 26). Thus, we also investigated a Nef mutant in which the N-terminal glycine residue essential for myristoylation and membrane attachment was replaced by alanine (NefG2A mutant). We showed that NefG2A significantly increased the formation of giant RAW264.7 macrophages (Fig. 3C) to the same extent as wt Nef, indicating that myristoylation is not required for Nef-mediated MGC formation.

Then, the specific role of Hck in the formation of MGCs triggered by Nef was examined by transfecting RAW264.7 cells with shRNA against Hck. Immunoblot analyses showed that Hck expression was reduced by 82% 48 h after Hck shRNA transfection (Fig. 3D). Because Hck isoforms are derived from a single mRNA (28), expression of both isoforms was markedly decreased. When Nef was expressed in Hck knockdown cells, it failed to trigger the formation of MGCs when compared with cells expressing Nef and Hck (Fig. 3E), demonstrating that the presence of Hck is necessary for Nef-induced giant macrophage formation. In addition, with a subclone of RAW264.7 cells (RAW264.7-Hck_Low) in which Hck expression was decreased by 93% when compared with that in parental cells (Fig. 3F), Nef expression did not trigger macrophage fusion (Fig. 3G, compare to Fig. 2B), confirming the results obtained with Hck shRNA.

Taken together, these data show that the role of HIV-1 Nef in macrophage fusion is Hck-dependent.

Next, we questioned whether Hck could be more generally involved in the formation of giant macrophages. To this end, we knocked down Hck expression in human macrophages derived from blood monocytes using Hck siRNA. Expression of Hck isoforms was markedly decreased (80% at day 4 and 65% at day 7 posttransfection) (Fig. 4A). Macrophage fusion was induced by treating cells with IFN-γ, which triggers the formation of MGCs with morphological and functional similarities to HIV-induced giant macrophages (12), and Con A to optimize cell–cell interactions (53). MGC formation was strongly reduced in Hck knockdown macrophages compared with that in control (Fig. 4B). The macrophage fusion index induced by IFN-γ/Con A treatment was decreased by ∼50% in Hck-depleted macrophages compared with that in control cells at day 6 (Fig. 4C). These results show that the role of Hck in the formation of giant macrophages is not restricted to the effect of Nef.

Similarly, in RAW264.7 macrophages transfected with Hck shRNA-2 and treated with IFN-γ/Con A to trigger MGC formation, a significant decrease in the cell fusion index was observed that was partially restored when Hck expression was partially recovered (Supplemental Fig. 2). Nef expressed in RAW264.7 cells treated with IFN-γ/Con A further stimulated the formation of MGCs, which was abolished in the absence of Hck (Supplemental Fig. 2C).

Taken together, these data show that Hck plays a key role in the formation of giant macrophages triggered either by Nef or by IFN-γ/Con A treatment, suggesting that HIV-1 could hijack an Hck-mediated cell function.

Although the activation of Hck by Nef is clearly established (23, 52, 54, 55), it has not been investigated whether Nef activates one or both Hck isoforms. Our previous works showed that ectopic expression in fibroblasts of constitutively active (ca) p59Hck^ca or p61Hck^ca mutants triggers the formation of plasma membrane protrusions (Fig. 5A, arrow) and F-actin–rich podosome rosettes (Fig. 5B, arrow), respectively (30, 41), whereas expression of wild-type p59Hck (p59Hck^wt) or p61Hck (p61Hck^wt) in fibroblasts did not have any effect on the cell shape (Supplemental Fig. 3A) (31). Thus, to study the activation by Nef of each Hck isoform, human p59Hck^wt or p61Hck^wt were expressed ectopically in combination with Nef in 3T3 fibroblasts. As reported previously (18, 37, 56–58), Nef expressed alone was mainly localized in the Golgi area, in perinuclear vesicles, and at the plasma membrane (Supplemental Fig. 3B). Coexpression of p59Hck^wt with Nef led to induction of plasma membrane protrusions (Fig. 5C, arrow, Fig. 5D), and coexpression of Nef and p61Hck^wt led to F-actin–rich podosome rosette formation (Fig. 5E, arrowheads, Fig. 5F), indicating that both Hck isoforms were activated by Nef. The percentage of cells with plasma membrane protrusions or podosome rosettes was similar in cells expressing NefAxxA with wt Hck isoforms and in cells expressing p59Hck^wt or p61Hck^wt alone (Fig. 5D, 5F), indicating that the PxxP motif of Nef is important to induce the morphologic changes that are characteristic of Hck activation.

Podosome rosettes were small and immature in cells coexpressing Nef and p61Hck^wt when compared with rosettes observed in cells expressing p61Hck^ca (Fig. 5B, 5E), and they formed only in 19 ± 8% of the cells compared with 38 ± 6% in p61Hck^ca-expressing cells (Fig. 5F). When Nef was coexpressed with p61Hck^ca, the percentage of cells forming podosome rosettes was decreased by 2-fold compared with that in cells expressing p61Hck^ca alone (Fig. 5F), whereas it did not affect the percentage of cells forming membrane protrusions when Nef was coexpressed with p59Hck^ca (Fig. 5D). In the light of our previous observations showing that the SH3 domain of p61Hck^ca is essential for its functional effects whereas the SH3 domain of p59Hck^ca is not (30, 41), the results obtained here indicate that although Nef activates p61Hck, it may also partially affect its signaling pathway by occupying the SH3 domain.

Next, we examined whether Nef also triggers the phenotypes characteristic of endogenous Hck activation in RAW264.7 macrophages. We observed that Nef-expressing macrophages formed plasma membrane protrusions (Supplemental Fig. 4Ad–Af, 4B), as previously shown for p59Hck^ca (41). Macrophages spontaneously form single podosomes (Supplemental Fig. 4Ab, 4Ac, arrowhead) that can arranged into rosettes by activation of p61Hck (30). In Nef-expressing macrophages, 27% of the cells had podosome rosettes versus 3% in control cells (Supplemental Fig. 4Ag–Ai, arrow, 4C).

Altogether, these results indicate that Nef is able to trigger activation of both Hck isoforms.

We then studied the role played by each Hck isoform in the Nef-mediated formation of giant macrophages. To this end, we used the macrophage clone RAW264.7-Hck_Low in which Nef expression was inefficient to form giant cells (Figs. 3G, 6A, white bars). In this cell line, a 2-fold increase in MGC formation was observed when Nef was coexpressed with p61Hck^wt (Fig. 6A, black bars) but not when Nef was coexpressed with p61Hck^kd (Fig. 6A, gray bars) or with p59Hck^wt (Fig. 6A, striped bars). Overexpression of each Hck isoform in parental RAW264.7 cells similarly showed that p61Hck, but not p59Hck, is able to increase macrophage fusion induced by Nef expression (data not shown). Moreover, expression of p61Hck^ca triggered the formation of MGCs (fusion index at 24 h is 24 ± 1% compared with 9 ± 3% in control cells) but not p59Hck^ca (Fig. 6B), indicating that p61Hck^ca was sufficient to promote macrophage fusion in the absence of Nef.

Therefore, macrophage fusion triggered by Nef is p61Hck- but not p59Hck-dependent and required p61Hck kinase activity. Interestingly, activation of p61Hck is sufficient to induce the biogenesis of giant macrophages.

The specific involvement of p61Hck in MGC formation suggests that lysosomes might play a role in this process. Indeed, coexpression of Nef with the p61Hck_G2A mutant, which is redirected from the lysosomal membranes to the cytosol due to a point mutation of the myristoylation site (30, 31), had no effect on Nef-mediated cell fusion in RAW264.7-Hck_Low (Fig. 7A, gray bars). In contrast, when p59Hck^wt was redirected from the plasma membrane to the membranes of lysosomes using the p59_C3SHck^wt mutant defective for palmitoylation (30, 31), we observed a significant increase in giant macrophage formation (Fig. 7A, black bars). The increase in Nef-induced fusion obtained with p59_C3SHck^wt was similar to that obtained with p61Hck^wt (compare Fig. 6A to Fig. 7A). These results demonstrated that association of p61Hck with lysosomes is important to mediate Nef-dependent macrophage fusion, suggesting that Nef could play a role at lysosomes. In fact, when cathepsin D-RFP, a lysosomal marker, was coexpressed with Nef-GFP or p61Hck^wt-GFP in RAW264.7 macrophages, they partially colocalized, as shown in Fig. 7B.

Because the NefG2A mutant unable to attach to the membrane was able to trigger the formation of giant cells (Fig. 3C), we wanted to determine whether it could colocalize with p61Hck at the membranes of lysosomes. We observed that part of NefG2A was associated with p61Hck-positive lysosomes (Supplemental Fig. 4), whereas it was fully in the cytosol in the absence of Hck (data not shown). This indicates that p61Hck could drive the subcellular localization of Nef at the lysosomal membranes.

We then used Bafilomycin A1 to inhibit vacuolar adenosine triphosphatase (v-ATPase), a proton pump associated with acidic organelles, including lysosomes, previously involved in the process of macrophage fusion (59). Formation of MGCs triggered by Nef was inhibited in a dose-dependent manner by Bafilomycin A1 (Fig. 7C). We wondered whether lysosome enzymes such as proteases might contribute to macrophage fusion, because they have been implicated in the formation of multinucleated myotubes (60, 61). Therefore, we used a protease inhibitor mix, mostly non-cell-permeant, containing pepstatin A (inhibitor of aspartic proteases including cathepsin D), leupeptin (a broad inhibitor of cysteine proteases and cathepsin B), aprotinin (inhibitor of serine proteases), E64C (inhibitor of cysteine proteases such as cathepsins B, L, H, and K), and GM6001 (a broad metalloproteases inhibitor) (62–64). As shown in Fig. 7D, macrophage fusion triggered by Nef expression was >2-fold diminished by the protease inhibitor mix.

In conclusion, the presence of Nef at lysosomes and the inhibition of macrophage fusion with v-ATPase and protease inhibitors suggest that lysosomes could play a role in the process of MGC formation.

Finally, we asked whether human macrophage fusion induced by HIV-1–infection also depends on protease activities. HIV-1–infected macrophages were treated with the protease inhibitor mix [except pepstatin A, which could strongly interfere with HIV-1 replication (65)], which did not affect macrophage viability (data not shown) or HIV replication (p24 in HIV-1–infected macrophage supernatants in the absence [63 ± 12 ng/ml] or presence of protease inhibitors [52 ± 13 ng/ml], n = 3 experiments). Formation of MGCs by HIV-1–infected macrophages was significantly decreased by 40% in the presence of protease inhibitor mix (Fig. 7E), and the number of cells with a high number of nuclei was reduced (data not shown). These results indicate that proteases are also implicated in the macrophage fusion process triggered by HIV-1 infection.

The molecular bases for Hck–Nef interaction, the highest affinity known for a SH3-mediated interaction, which leads to activation of the kinase, have been well-established (21–23, 50, 54, 55). Because Hck is a Src family kinase specifically expressed in phagocytes, our objective was to study the functional consequences of Hck activation by Nef in HIV-1–infected macrophages. We show that 1) HIV-1 Nef triggers the formation of giant macrophages in an Hck-dependent manner, 2) p61Hck and its lysosome attachment are required in this process, and 3) v-ATPase and proteases, most probably of lysosomal origin, are implicated in the formation of MGCs triggered by Nef during HIV-1 infection.

When human macrophages infected with Δnef-HIV-1 or wt viruses were compared, we observed that formation of MGCs (or syncytia) was diminished in Δnef-HIV-1–infected cells, whereas Δnef-HIV-1 and wt-HIV-1 infectivities of human macrophages were similar. In AIDS patients, MGCs have been shown to be of macrophage origin (7–9), but the mechanisms involved in their formation in vivo or in vitro have not been studied. Also HIV-1–infected CD4⁺ lymphocytes form giant cells in vitro, but until now, they have not been observed in AIDS patients. Other mechanisms could account for the remaining MGCs formed with Δnef-HIV-1–infected macrophages (53, 66–70) [i.e., fusion of HIV- 1 infected T cells is in part mediated by interaction of the virally encoded env gene expressed at the plasma membranes of infected cells with CD4 receptors (66, 67, 71–75)]. Because giant macrophages rapidly form upon expression of Nef alone, it indicates that Nef is sufficient to trigger macrophage fusion independently of Env or other viral proteins. Interestingly, among HIV-1 proteins, Nef was the only protein detected in giant macrophages of the brain in AIDS patients (76).

We clearly demonstrate that the process of MGC formation induced by Nef involves Hck. Nef did not trigger formation of giant macrophages when Hck was knocked down or when Nef was expressed in RAW264.7-Hck_Low macrophages. Formation of MGCs triggered by Nef was polyproline motif-dependent as expected for a molecular process involving a SH3-domain protein. This previously unknown function of Nef was conserved among various HIV-1 and SIV isolates, all described to activate Hck (77, 78). Therefore, this is the first report showing that Nef and Hck contribute to the formation of giant macrophages.

In cells expressing Nef, both Hck isoforms, located in distinct subcellular compartments (31), were activated, but only p61Hck was involved in the process of MGC formation. In fact, coexpression of p61Hck^wt with Nef triggered macrophage fusion, whereas coexpression of p59Hck and Nef or p59Hck^ca alone did not. We show that the catalytic activity of p61Hck and its association with lysosomes are critical for macrophage fusion and that expression of p61Hck^ca alone was sufficient to trigger the formation of MGCs. Nef is associated with lysosomes in macrophages where it colocalizes with cathepsin D. Because p61Hck activation has been associated with mobilization and exocytosis of lysosomes (29, 30, 47, 79–82), it is conceivable that lysosomes, which contain several enzyme subsets, including proteases, hydrolases, and v-ATPase, could facilitate the cleavage of surface proteins and glycocalyx from opposite cells to interact more closely and hence facilitate cell fusion. Additionally, lysosomal proteases could modulate surface receptor activity and signaling involved in cell–cell fusion (53, 68, 69, 83–85). Lysosomal proteases have been implicated in the formation of myotubes (60, 61) and found in the secretome of HIV-1–infected giant human macrophages (86). In the presence of protease inhibitors, the formation of MGCs is altered in Nef-expressing RAW264.7 cells or HIV-1–infected human macrophages. Moreover, v-ATPase, previously involved in the formation of giant osteoclasts (59), is also implicated in the biogenesis of giant macrophages triggered by Nef, probably because an acidic pH is required for lysosomal enzyme processing, activity, and secretion (87). Thus, we propose that activation of p61Hck by Nef at lysosomes contributes to the mobilization of lysosome-associated enzymes that play a pivotal role in the formation of giant macrophages.

In contrast to the interaction between Nef and Lck in T cells (88, 89), Nef did not induce any visible delocalization of Hck isoforms. Actually, the phenotypes triggered by Nef expression in fibroblasts and RAW264.7 macrophages are consistent with activation of Hck isoforms at their original localizations because the formation of plasma membrane protrusions or podosome rosettes is strictly dependent on the correct intracellular locations of Hck isoforms (30, 41). In addition, RAW264.7 macrophages expressing p61Hck^ca formed MGCs in the presence (data not shown) or in the absence of Nef, further indicating that the kinase subcellular localization is not altered by Nef. In fact, association of p61Hck with lysosomes is critical for MGC formation, as demonstrated with the cytosolic mutant of p61Hck (p61Hck_G2A).

We also report that Nef does not divert Hck from its normal functions but rather uses them. Actually, Hck appeared to be also involved in the formation of giant macrophages triggered by a combination of IFN-γ and Con A.

It is interesting to question the role of giant macrophages in AIDS pathogenesis. Several pathologies involve the formation of giant macrophages, but it is not clear whether it is beneficial or detrimental for patients (12). Giant macrophages have been described in the brains of AIDS patients as a marker for disease progression associated with HIV-induced dementia (7) and in lymphoid tissues of Waldeyer’s ring in 66.7% of HIV-infected patients (8). HIV-infected T cells also form giant cells, but they rapidly undergo apoptosis (10), probably explaining why giant HIV-1–infected lymphocytes have not been observed in vivo. In vitro experiments performed with giant T cells have shown that cell–cell fusion brings together diverse HIV-1 subtypes, thus facilitating intersubtype recombination and, consequently, viral diversity (90). In addition, HIV-1–infected giant macrophages harbor higher numbers of proviral copies than a mononuclear cell and could be long-lasting reservoirs (7, 11, 91). Taken together, these findings suggest that the formation of giant cells could be detrimental for AIDS patients. However, specific functions of HIV-infected giant macrophages compared with those of mononuclear macrophages remain to be elucidated to better understand their role and determine whether pharmacological modulation of macrophage fusion could be beneficial for HIV-1–infected patients.

We thank C. Cougoule for shRNA transfection, F. Darchen for the cathepsin D-RFP construct, O. Fackler for NefSF2 constructs, H. Gottlinger for SIV Nef, O. Neyrolles for critical reading of the manuscript, M. Cazabat for help with HIV-1 experiments, A. Bouissou for deconvolution images, and, finally, R. Poincloux of the imaging platform of the Institut de Pharmacologie et de Biologie Structurale.

Disclosures The authors have no financial conflicts of interest.