Regulation of Phospholipase D by Arf6 during FcγR-Mediated Phagocytosis

Specialized immune cells, such as macrophages, polymorphonuclear granulocytes, and dendritic cells, internalize and degrade large particles (>0.5 μm) such as pathogens and cellular debris by phagocytosis. This process is initiated by the binding of particle-associated ligands to specific receptors and lectins on the phagocyte surface. Subsequent clustering of these receptors stimulates tyrosine kinases, which in turn trigger an activation cascade that has been best described for the Fcγ portion of the Ig receptor (FcγR) (1). This cascade initiates the extension of pseudopods forming a “phagocytic cup” around the particle to generate a phagosome, the vacuole in which the particle is engulfed. Following internalization, the phagosome matures into a phagolysosome, which leads to the acidification of the vacuole, degradation of the ingested material, and eventual recycling of ligands for Ag presentation (2). Membrane homeostasis during phagocytosis is maintained by focal exocytosis (3) of endomembranes inserted into the plasma membrane, a process that requires extensive actin cytoskeleton remodeling (4) and membrane fusion events (5).

Membrane trafficking is essential for phagocytosis during both the formation and maturation of the phagosome. Different intracellular compartments, including early and late endosomes, lysosomes (6), and the endoplasmic reticulum (7), have been proposed as membrane sources for focal exocytosis at phagocytic sites. Formation of SNARE complexes between the plasma membrane and vesicular structures arising from these compartments are necessary for this focal exocytosis (8, 9). In addition, different lipids, including phosphoinositides and phosphatidic acid (PA) (10, 11), contribute to phagosome formation, internalization, and maturation. Using specific sensors, we have previously demonstrated that PA is produced and accumulates at phagocytic sites (12–14). Although diacylglycerol kinases may be an important source of PA during phagocytosis (15), an increase in phospholipase D (PLD) activity, which generates PA from phosphatidylcholine, has also been reported during the activation of several phagocytic receptors, including the FcγR (16, 17). Both PLD1 and PLD2 isoforms seem to be necessary for efficient phagocytosis (12). In RAW 264.7 macrophages, PLD2 is present at the plasma membrane, whereas PLD1 associated with the late endosome/lysosome compartment is recruited to phagocytic sites (12), suggesting a sophisticated regulation of PLD activity and PA synthesis during phagocytosis. Interestingly, PLD1 and PLD2 have also been shown to regulate different steps in FcεRI-mediated degranulation and anaphylactic reactions in mast cells (18–20), suggesting a widespread critical role for PLDs in immune cell functions. Note that immune cell activities are tightly controlled functions, which makes the issue of PLD regulation during immune responses particularly interesting.

Monomeric GTPases are master regulators of membrane trafficking processes, including receptor-mediated endocytosis, endosomal recycling, and exocytosis of secretory granules (21–23). The first GTPase that has been directly implicated in phagocytosis, in particular in the delivery of endomembranes to forming phagosomes, was the ADP ribosylation factor (Arf) 6 (24–26). Yet, the downstream pathway by which Arf6 participates in FcγR-dependent phagocytosis is still unknown. GTPases, together with protein kinase C, are also the best-characterized regulators of PLD. Initially, only PLD1 activity was thought as being regulated by these GTPases because of the high basal activity of PLD2. However, it was later shown that activities of both PLD isoforms are subject to regulation (27). Altogether, these findings prompt us to investigate whether Arf6 might be implicated in phagocytosis by regulating PLD activity and PA synthesis. We show, in this study, that Arf6 and PLDs partially colocalize and interact in macrophages undergoing phagocytosis. Arf6 silencing effectively reduces PLD activation and PA synthesis at the phagocytic sites, consistent with the idea that Arf6 contributes to optimal phagocytosis through the regulation of PLD.

RPMI 1640 and FBS were purchased from Invitrogen. Latex beads (3 μm; Sigma-Aldrich) were coated with human IgG (Zymed). Escherichia coli coupled to Alexa Fluor 594 and zymosan particles were purchased from Molecular Probes. Rat anti-hemagglutinin (HA) affinity matrix was purchased from Roche. The following mAbs were used: anti-Arf6 (mouse IgG2b; Santa Cruz Biotechnology), anti-Arf1 (mouse IgG2a; Abcam), anti-CD64 (rat IgG2b; R&D Systems), anti-HA (mouse IgG1, HA.11; Covance), anti-GFP (mouse IgG1; Roche), anti–β-actin (mouse IgG1, clone AC-15; Sigma-Aldrich), and anti–β-tubulin (mouse IgG1, clone TUB 2.1; Sigma-Aldrich). Polyclonal anti-Arf6 (rabbit; Bethyl Laboratories) was also used for immunoblots. Goat secondary Abs coupled to Alexa Fluor 555 or 568 (Molecular Probes) or to peroxidase (Thermo Fisher Scientific) were used for immunofluorescence and immunoblots, respectively.

Culture conditions for the murine macrophage RAW 264.7 cell line in RPMI 1640 glutamax medium supplemented with 10% FBS were as previously described (12–14).

Arf6-pEGFP and small interfering RNA (siRNA)–resistant Arf6-HA-pXS and Arf6(N48I)-HA-pXS were described previously (23, 28). The plasmids pCGN-human PLD1 (PLD1-HA) and pCGN-mouse PLD2 (PLD2-HA) were described previously (29). MT2-GFP was used to visualize the localization of the active form of Arf6 (28). Similarly, Spo20p-GFP served as a probe to detect sites of PA production (30). RAW 264.7 cells were transfected with the plasmids by electroporation (12) or using Lipofectamine 2000 according to the manufacturer’s instructions (Invitrogen, Thermo Fisher Scientific). For confocal experiments, RNA interference (RNAi)–treated macrophages were collected, replated on glass inserts in 35-mm Petri boxes (Maktek), and transfected with MT2-GFP or Spo20-GFP.

IgG-opsonized 3-μm latex particles (IgG beads) or E. coli particles, washed with PBS and resuspended in medium, were added to cells grown on glass coverslips (20 beads per cell or 50 E. coli per cell). For these assays, phagocytosis was synchronized after adding particles by centrifuging cells for 2 min at 100 × g at 18°C and then initiating phagocytosis by placing them at 37°C as previously described (12). Briefly, phagocytosis was stopped 30 min later by washing twice in cold PBS. Fluorescence of noninternalized E. coli was bleached by trypan blue 0.4%. After fixation, external beads were labeled with goat anti-human IgG coupled to Alexa Fluor 555 (Molecular Probes). Unlabeled internalized beads were visualized with phase contrast optics (Axio Imager 2; ZEISS). The mean number of internalized beads or the mean fluorescence intensity (for E. coli assay) per cell was determined for randomly chosen fields (minimum of 80 cells for each field) using superimposed fluorescent and phase contrast images (Adobe Photoshop 9). The phagocytic index was normalized to 100% for cells transfected with the control RNAi.

Phagocytosis was initiated by adding IgG-opsonized particles to 1 × 10⁷ cells in suspension at 37°C and stopped at different times (t = 0, 5, 10, and 20 min) as previously described (31). For each time point, Arf-GTP was precipitated from lysates (2 mg total protein) with the Arf-GTP binding domain of Golgi-localized γ-adaptin ear homology Arf binding protein 3 (GGA3) linked to GST, according to the instructions in the Arf Activation Assay Kit (Thermo Fisher Scientific). Control samples with or without the addition of opsonized particles were maintained at 4°C. Arf-GTP precipitates and aliquots of total cell lysates (10–20 μg protein) were analyzed on immunoblots using Abs against Arf6, Arf1, and actin.

Macrophages were transfected with Arf6-GFP and either PLD1-HA, PLD2-HA, or βPIX-HA, and cells were collected 24 h later. Phagocytosis was then initiated by adding IgG beads to the cell suspensions (5 × 10⁶ cells) at 37°C for different times, and cell lysates were then prepared in 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM MgCl_2, 0.1 mM DTT, 0.5% Triton X-100, 0.5% DOC, and protease inhibitor mixture (Sigma-Aldrich). PLD proteins were immunoprecipitated from lysates (0.5 mg protein) using rat Anti-HA Affinity Matrix (Roche) and analyzed by immunoblots.

Macrophages were transfected with 40 pmoles of a stealth RNAi duplex specific for mouse Arf6 (duplex A: 5′-GGAACAAGGAAATGCGGATCCTCA-3′; duplex B: 5′-CAGCCGGCAAGACAACGAUCCUGUA-3′; and duplex C: 5′-CCCAGGGUCUGAUCUUCGUGGUAGA-3′) or a control nontargeted oligonucleotide (CG medium) using Lipofectamine RNAiMAX according to the manufacturer’s instructions (Invitrogen). Using a control oligonucleotide Alexa Fluor 488 (Invitrogen), the transfection efficiency of stealth RNAi was estimated by flow cytometry to be >85%. After 48 h, the reduction in the Arf6 protein expression and activation was determined by immunoblots, and as a control, the expression of Arf1 was checked.

Lysates were prepared before and following FcγR activation. Total cell extracts were prepared in 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA, 1.0% Triton X-100, 0.5% deoxycholate, and a protease inhibitor mixture (Sigma-Aldrich) and cleared by centrifugation for 10 min at 10,000 × g at 4°C. For subcellular fractions, cells were suspended in 50 mM Tris-HCl (pH 7.5), 2 mM EDTA, containing protease inhibitors at 4°C and lysed in a Potter homogenizer. After an initial centrifugation at 800 × g for 10 min at 4°C, the lysate was centrifuged at 20,000 × g for 30 min at 4°C to separate soluble cytoplasmic supernatant and a membrane fraction composed of organelles and large-membrane fragments. Immunoblotting was carried out as previously described (12). Briefly, proteins were separated on 4–12% NuPAGE gradient gels in MES buffer (Invitrogen) and transferred to nitrocellulose. After immunolabeling, blots were revealed using the SuperSignal West Dura or Femto Chemiluminescent Substrate (Pierce Biotechnology, Thermo Fisher Scientific). For cell lysate samples, actin served as a control for equal protein loadings. Images were acquired using a Chemi-Smart 5000 and the Chemi-Capt program (Vilber Lourmat), and protein bands were quantified using the program Bio1D (Vilber Lourmat). Images were then processed with Photoshop 9.

Live cells were observed at 37°C 18–24 h after transfection in the absence or presence of IgG beads. Videos were obtained by acquiring images for GFP and phase contrast every minute for 20 min. Cells destined for immunofluorescent labeling were fixed for 10 min at 4°C with 4% paraformaldehyde in 0.125 M phosphate buffer, and for intracellular labeling, this was followed by a 10-min permeabilization step in fixative containing 0.2% Triton X-100. Samples were then blocked with 10% goat serum, and PLD was visualized using anti-HA Abs followed by goat anti-mouse Alexa Fluor 555 or 568. Images were obtained using a Zeiss LSM 510 or a Leica SP5 II inverted microscope equipped with a Plan APO oil (63×) immersion lens (numerical aperture = 1.4). Images were recorded with the same parameters and optimal pinhole and processed using Adobe Photoshop 9.

Using the Zeiss CLSM software 2.8, masks of double-labeled pixels were generated. The proportion of Arf6-GFP colocalized with PLD1-HA or PLD2-HA was determined in nonstimulated cells and in cells during particle ingestion using the weighted colocalization percentages generated for double-labeled pixels. Quantification of Arf6-GFP signal, or beads positive for GFP signal (MT2-GFP or Spo20p-GFP) were performed using Icy software. For each cell, a fixed region of interest of 6 pixels × 6 pixels were randomly selected at the plasma membrane or at the periphery of beads, and the fluorescence signal was measured and compared with the fluorescence signal obtained from five distinct region of interests of the same size randomly selected in the cytosol. Beads were considered as positive for staining if the bead fluorescence level was above 1.2-fold of averaged cytosolic signal.

Forty-eight hours after siRNA transfection, RAW 264.7 macrophages (6 × 10⁶ cells) were washed and then incubated at 37°C in the absence (resting) or presence (stimulated) of IgG-opsonized zymosan particles. Cell lysates were prepared in 600 μl of ice-cold 50 mM Tris-HCl (pH 8) by three freeze and thaw cycles. Aliquots of the lysates (corresponding to 1 × 10⁶ cells) were mixed with an equal amount of the Amplex Red reaction buffer (Amplex Red Phospholipase D Assay Kit; Molecular Probes), and the PLD activity was estimated after 1-h incubation at 37°C with a Mithras fluorometer (Berthold Technologies) as described previously (32). A standard curve was established with purified PLD from Streptomyces chromofuscus (Sigma-Aldrich), and an average activity (milliunits per milliliter) was calculated from four determinations made for each condition.

Number of experiments and repeats are indicated in figure legends. Normality of the data distribution was verified with ANOVA test, and statistical analysis was performed with t tests relative to the indicated control, except for Fig. 3A, for which ANOVA test was used for the analysis.

The datasets generated during and/or analyzed during the current study are available from the corresponding author on request.

Based on the expression of constitutively active and dominant-negative GTP binding–deficient mutants, Arf6 was the first monomeric GTPase described to play a role in phagocytosis (24–26). To investigate the downstream effector of Arf6 in phagocytosis, we decided to use an RNAi strategy to decrease the expression level of the endogenous Arf6 protein in RAW 264.7 macrophages. Three Arf6-RNAi were compared. As seen by Western blot and densitometric scan analyses (Fig. 1A), the three Arf6-RNAi specifically reduced endogenous Arf6 levels by 60–80% without affecting the expression levels of actin or the related isoform Arf1. Decrease of endogenous Arf6 in RAW 264.7 macrophages reduced phagocytosis of IgG beads by ∼25% (Fig. 1B). Interestingly, Arf6 silencing inhibited phagocytosis of E. coli (Fig. 1B) and of nonopsonized beads (data not shown) as well. Thus, Arf6 seems to be involved in both FcγR-mediated and FcγR-independent phagocytosis. It is of note that Arf6 silencing did not alter CD64 receptor expression level (Fig. 1C), indicating that the inhibition of FcγR-dependent phagocytosis by reducing endogenous Arf6 is not the consequence of an eventual FcγR downregulation.

We then investigated the subcellular localization of Arf6 in RAW 264.7 macrophages undergoing FcγR-mediated phagocytosis. Distribution of endogenous Arf6 was first compared with the distribution of expressed Arf6-GFP in macrophage subcellular fractions by Western blot analysis (Fig. 2A). In both resting cells and cells stimulated with IgG beads, Arf6 was found to be concentrated in the crude membrane fraction containing the plasma membrane and organelles. Of note, expressed Arf6-GFP behaved like endogenous Arf6 regarding its subcellular localization (Fig. 2A). Arf6-GFP was then used to study the distribution of Arf6 during phagocytosis in living cells by time-lapse confocal imaging (Fig. 2B). Incubation with IgG beads triggered the rapid recruitment of Arf6-GFP to the cell periphery, especially at sites of phagocytosis (arrow, Fig. 2B). Arf6-GFP was found to accumulate on pseudopods and membrane ruffles formed around the beads (Fig. 2B, times 0–5 min). Note, however, that Arf6-GFP was not detected on fully internalized beads (Fig. 2B, times 7–20 min). Quantification of the GFP fluorescence confirmed the preferential accumulation of Arf6 at the phagocytic cup (nascent phagosomes) and, to some extent, on the plasma membrane but not on internalized beads (Fig. 2C), suggesting that Arf6 is mainly involved in the early stages of the internalization process.

Like all GTPases, Arf6 cycles between inactive GDP-bound and active GTP-bound forms. The level of endogenous activated Arf6 in resting and IgG-stimulated RAW 264.7 macrophages was assessed by pull-down experiments using GST–GGA3 as bait for Arf6-GTP. Whereas the total level of Arf6 did not significantly change upon stimulation, the level of active Arf6 increased by more than 2.5-fold after 20 min of incubation with IgG beads (Fig. 3A). Based on the total amount of Arf6 and Arf6-GTP levels, we estimated the ratio of activated Arf6 and found an increase from 0.3% in control conditions to 0.75% following 20 min of IgG stimulation. This is probably an underestimation of the actual levels of Arf6-GTP. As the pull-down efficiency in our assay ranges from 50 to 95%, the actual Arf6-GTP levels might range from 0.3–0.6% in resting cells to 0.75–1.5% in cells stimulated for 20 min with IgG beads.

Because the amount of activated Arf6-GTP remains relatively low in many cell types (33, 34) including macrophages, the residual endogenous Arf6 expressed in Arf6-RNAi–transfected cells might be sufficient to provide levels of active Arf6 similar to that found in control cells. To probe this possibility, we used the GST–GGA3 probe to pull down the GTP-bound forms of Arf6 and Arf1 in Arf6-RNAi–expressing cells stimulated for phagocytosis. As observed for the endogenous Arf6 protein, the amount of Arf6-GTP was significantly reduced in cells expressing an Arf6-RNAi (Fig. 3B). Thus, Arf6-RNAi not only reduced the total level of endogenous Arf6 but also the amount of activated Arf6 detected in macrophages stimulated for phagocytosis. Surprisingly, Arf6-RNAi also significantly increased the level of activated Arf1 (Fig. 3B). Because Arf1 and Arf6 share several common downstream effector pathways, it is therefore quite possible that in cells expressing an Arf6-RNAi, active Arf1 may, to some extent, compensate for the reduced level of active Arf6, reducing the inhibitory impact of the Arf6-RNAi on phagocytosis.

To visualize the distribution of activated Arf6 in RAW 264.7 macrophages, MT2 fused to GFP was used as a specific sensor for Arf6-GTP because it does not recognize other members of the Arf family (28). In resting RAW 264.7 cells, the MT2-GFP sensor was found in the cytosol and the nucleus (Fig. 3C). IgG stimulation for phagocytosis led to an apparent accumulation of the MT2-GFP probe at phagocytic cups and around phagosomes (Fig. 3C). This apparent recruitment of the MT2-GFP sensor to nascent phagosomes was clearly reduced in cells expressing Arf6-RNAi (Fig. 3C). Quantification revealed that the percentage of phagosomes labeled by MT2-GFP dropped from 86 to 23% after Arf6-RNAi transfection (Fig. 3C). Altogether, these observations confirm that endogenous Arf6 is activated in stimulated macrophages and that activation occurs at the plasma membrane near phagocytic sites.

PLD1 and PLD2 have been shown to be involved in FcγR-mediated phagocytosis (12). Because Arf6 is a well-known regulator of PLD activity in membrane trafficking (35), we first investigated whether Arf6 might be able to interact with PLD during phagocytosis. The lack of effective Abs for immunocytochemistry prompted us to use overexpressed proteins to study the distribution of Arf6, PLD1, and PLD2. In unstimulated RAW 264.7 macrophages coexpressing both Arf6-GFP and PLD1-HA, Arf6-GFP was mostly found on vesicular structures in the cytoplasm, where it partially colocalized with PLD1-HA (Fig. 4A). In contrast, when macrophages coexpressed both Arf6-GFP and PLD2-HA, Arf6-GFP was mostly found at the plasma membrane together with PLD2-HA (Fig. 4A). These observations indicate that the subcellular distribution of Arf6-GFP is affected by the overexpression of PLD1 or PLD2, suggesting that Arf6 might interact with these PLDs in RAW 264.7 cells. Stimulation with IgG beads triggered an accumulation of Arf6-GFP/PLD1-HA on nascent phagosomes, whereas Arf6-GFP/PLD2-HA was mostly detected on the plasma membrane (Fig. 4A). Quantification revealed a substantial colocalization of Arf6 with either PLD1 or PLD2 in resting conditions but also a significant increase of Arf6/PLD1 and Arf6/PLD2 colocalization in cells undergoing phagocytosis (Fig. 4B). Note the disappearance of Arf6/PLD1-positive vesicular structures in the cytoplasm of macrophages stimulated for phagocytosis (Fig. 4A), suggesting that these structures might fuse with the plasma membrane or with the phagosomes to provide an additional membrane.

The potential interaction of Arf6 with PLD isoforms in resting and stimulated macrophages was further probed by immunoprecipitation using cells coexpressing Arf6-GFP with PLD1-HA or PLD2-HA. In control experiments performed on RAW 264.7 cell extracts, anti-HA Abs were found to precipitate around 90% of PLD1-HA or PLD2-HA, validating the efficiency of the precipitation (data not shown). We found significant levels of Arf6-GFP coimmunoprecipitating with either PLD1-HA or PLD2-HA (Fig. 5A). Coprecipitation of Arf6-GFP with PLDs appeared to be specific as no Arf6-GFP was found to precipitate with β-PIX, a protein that is not supposed to interact with either Arf6 or PLDs (Fig. 5A). PLD2 seemed to be expressed at higher levels than PLD1 as seen by the stronger HA signal, but equal levels of Arf6-GFP were recovered in the immunoprecipitates (Fig. 5A, 5B), suggesting that Arf6 binds more tightly to PLD1 or to a complex containing PLD1 than to PLD2. Of note, triggering phagocytosis with IgG-coated beads increased the amount of Arf6 coprecipitating with PLD1 and PLD2 but with a rather distinct time course (Fig. 5C), suggesting different kinetics of interaction for Arf6/PLD1 and Arf6/PLD2 during phagocytosis. This observation is in line with our previous report describing the sequential activation of PLD2 followed by PLD1 in FcγR-mediated phagocytosis (12).

To explore the possibility that Arf6 regulates PLD activation during phagocytosis, we assayed PLD activity in resting or IgG-stimulated macrophages expressing Arf6-RNAi to reduce endogenous Arf6 levels. As illustrated in Fig. 6A, expression of Arf6-RNAi did not significantly modify the localization of PLD1-HA and PLD2-HA in resting or stimulated cells. However, Arf6 silencing significantly decreased the PLD activity induced by IgG stimulation of phagocytosis (Fig. 6B). Rescue experiments were performed by expressing in resting and IgG-stimulated macrophages Arf6 constructs resistant to the Arf6-RNAi. Expression of a resistant wild-type form of Arf6 largely prevented the Arf6-RNAi–mediated inhibition of PLD activity, whereas the expression of the Arf6 (N48I) mutant, unable to stimulate PLD (23), failed to rescue the PLD activity during phagocytosis (Fig. 6C). Altogether, these results strongly support a positive link between Arf6 and PLD activation during FcγR-mediated phagocytosis.

Activated PLD produces PA, which can be visualized using the PA-binding domain of the yeast homolog of SNAP25, Spo20p, fused to GFP (Spo20p-GFP) (13). In nonstimulated macrophages, Spo20p-GFP was mostly found in the nucleus and discretely at the plasma membrane (Fig. 7A). In contrast, in macrophages stimulated for phagocytosis, the PA sensor was actively recruited to the plasma membrane, including the phagocytic sites (Fig. 7A), in line with the idea that PLD mediates the formation of PA at the sites of particle ingestion (12, 13). To further assess the importance of Arf6 in the upstream signaling pathway of PLD, we examined the distribution of the Spo20p-GFP PA sensor in cells expressing Arf6-RNAi. In these experiments, RAW 264.7 cells were first transfected with RNAi and then, 24 h later, transfected with the Spo20p-GFP plasmid. As shown in Fig. 7A, expression of Arf6-RNAi largely inhibited the recruitment of Spo20p-GFP to the cell periphery. The percentage of phagosomes displaying a Spo20p-GFP staining in control and Arf6-RNAi–expressing cells decreased from 83% in control cells to 30% in cells with reduced Arf6 levels (Fig. 7B). Thus, Arf6 is required for PLD activation and PA production at phagocytic sites. Finally, in line with the functional importance of PA synthesis during phagocytosis, expression of Spo20p-GFP and binding to PA potently inhibited FcγR-dependent phagocytosis (Fig. 7C), whereas the mutant Spo20p(L67P) that bound PA with a reduced affinity (14) failed to block phagocytosis (Fig. 7C).

Over the last two decades, accumulating evidence has implicated the GTPase Arf6 in various membrane trafficking events, including endocrine and neuroendocrine exocytosis (23, 36, 37), endocytosis (21), and myoblast fusion (38). In these membrane trafficking processes, downstream effector(s) generally link Arf6 to lipid synthesis, mainly through the regulation of phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P₂] (37) and PA synthesis (35). Arf6 has also been involved in phagocytosis, first in macrophages (24–26, 39) and more recently in human monocytes (40). Further highlighting the contribution of Arf6 in phagocytosis, the Arf6 guanine nucleotide-exchange factor and GTPase-activating proteins, Arf-GEP100/BRAG2 and PAG3, have been identified as regulators of phagocytosis (41, 42). Yet, despite this set of findings, our understanding of the role of Arf6 in phagocytosis is still limited and awaits the identification of its downstream effectors. The present report provides evidence that IgG-stimulated phagocytosis leads to the activation of Arf6, its recruitment to the phagocytic cup, and the subsequent stimulation of PLD to produce PA at the phagosome. Arf6 silencing was found to inhibit phagocytosis of opsonized and nonopsonized beads as well as bacteria, suggesting that the Arf6-regulated PLD pathway described in this study in detail for FcγR-dependent phagocytosis might be generalized for different types of phagocytic pathways.

We have previously described a positive effect of Ral A on phagocytosis that could be explained in part by its interaction with PLD, which itself positively regulates phagocytosis (31). A similar relationship between the Ral GTPases and PLD has also been described in exocytosis (43), receptor endocytosis (44), and neurite branching (45). In this study, we observed that Arf6 localizes with PLD1 and PLD2 on nascent phagosomes and interacts with both PLD isoforms in resting and stimulated macrophages as seen by colocalization and immunoprecipitation experiments. Noteworthy, Ral A was also found to coimmunoprecipitate with PLD1 and PLD2 but only in cells undergoing phagocytosis (31). At this stage, it is not possible to state whether the interactions of Arf6 and Ral A with PLDs are direct or indirect, but the persistent coimmunoprecipitation of Arf6 with PLDs compared with the activity-dependent coimmunoprecipitation of Ral A with PLDs, suggests that in macrophages, the regulation of PLD activity and PA production by these two GTPases involves different modalities. Interestingly, a Ral-Arf6 cross-talk has been recently described in exocyst trafficking and other Ral functions in cells in which active Ral uses a Ral-RalBP1-ARNO-Arf6 pathway to mediate Arf6 activation (46). Thus, a possible explanation for our observations would be that Arf6 and PLD are part of a common complex in resting macrophages and that IgG binding to FcγR mediates activation and recruitment of Ral A to Arf6/PLD complex to trigger Arf6 activation and subsequent PLD stimulation. Regarding the production of PA, it was recently reported that macrophages and immature dendritic cells display elevated basal plasma membrane levels of PA required for immunosurveillance that appear to be generated through the phosphorylation of DAG (15). Thus, macrophages seem to control intracellular PA levels through several alternative pathways, highlighting the central role of this lipid in the various immune functions of macrophages.

Determining the cellular processes implicating Arf6-mediated PLD activation in the course of phagocytosis remains an important unresolved issue. Our observation that the expression of the PA sensor Spo20p-GFP decreases IgG beads internalization supports the idea that PA synthesis at the phagocytic cup contributes to efficient phagocytosis. As lipids and lipid-anchored proteins coordinate spatial and temporal patterns within the nascent phagosomal membrane (47), the modulation of PA production by PLD at phagocytic sites (12) may actually contribute to the organization of the phagocytic cup. For instance, PA can be converted into other bioactive lipids, like diacylglycerol or PtdIns(4,5)P₂, both of which are implicated in the phagocytic machinery (10, 48–50). Alternatively, PA may act by activating or recruiting regulators of actin cytoskeleton implicated in membrane trafficking (51). In line, it was recently reported in PLD knockout mice that macrophages present disorganized phagocytic cups with abnormal F-actin structures (17). PA could also participate in the fusion process either directly as its small polar head promotes negative membrane curvatures favoring membrane hemifusion (36) or indirectly by serving as a lipid anchor for proteins involved in the fusion process as, for example, Rab11FIP on recycling endosomes (52) or Q-SNARE Syntaxin1A (53) at the plasma membrane. The fact that Arf6 is recruited during phagosome formation and thereafter dissociates from internalized phagosomes supports the idea that Arf6 might play its role in the early phases of the phagocytic process.

We show in this study that Arf6-RNAi only moderately inhibited phagocytic activity and had much less effect than the expression of Arf6 (T27N), a dominant-negative form of Arf6 (25). In fact, in contrast to Arf6-RNAi (36), expression of Arf6 (T27N) also strongly affects the dynamics of the recycling endosomal compartment and PtdIns(4,5)P₂ biosynthesis (54). It is thus likely that the strong effect on phagocytosis observed upon Arf6 (T27N) expression is the consequence of a major alteration of PtdIns(4,5)P₂ biosynthesis, in addition to reduced PLD activation. Altogether, these observations suggest that either Arf6 activation of PLD plays only a modulatory role in phagocytosis or the amount of Arf6 remaining in cells expressing Arf6-RNAi is enough to sustain a large part of Arf6 function in phagocytosis. Alternatively, upregulation of related GTPases, which can activate PLD, may compensate for the reduction of Arf6 activity. Indeed, we observed that Arf1 activation significantly increased after Arf6-RNAi treatment (Fig. 3B). PLDs are modular enzymes that can be activated by a variety of GTPases, including Arfs, Rals, and Rac (27). Interestingly, Arf6 and Rac have been shown to transiently form a complex that is dependent on their activation status and their expression levels (55), highlighting the complexity of interactions between the various GTPases regarding their effector pathways. Both Arf6 and Rac have been shown to be involved in membrane ruffling in macrophages (56), and Rac is implicated in phagosome formation (57, 58). Another possibility, although not exclusively, is that the implication of the Arf6–PLD pathway may depend on the size and number of particles to be internalized because phagosome formation has been proposed to require more or less focal exocytosis of internal membranes (7). It is indeed likely that the type and level of PA required at phagocytic sites may differ for different phagocytic processes. For instance, we have recently shown that frustrated phagocytosis induces a nearly 2-fold increase in PA levels at the plasma membrane occurring only partly at the expense of the endoplasmic reticulum pool of PA (14). Lipidomic analysis revealed that more than 40 different PA species were detected in different intracellular membrane compartments in RAW 264.7 macrophages, suggesting that these different forms of PA contribute differently to phagocytosis (14). In this direction, it is likely that different regulatory pathways involving various GTPases actually contribute to the synthesis of different PA species required for phagocytosis. The recent development of novel PA sensors displaying some preference for specific PA species, depending on the lipid environment and the subcellular localization (14), will undoubtedly prove to be useful to investigate this possibility. Finally, Arf6 promotes not only PA but also PtdIns(4,5)P₂ production for autophagosome formation (59), further underlining the complexity and interconnections of Arf6 and lipid-modifying enzymes in the dynamics of membrane remodeling.

To conclude, together with several other lipids (60), local PA production appears to be essential for at least an important step of the life cycle of phagosomes (61). The exquisite dynamics of all these key lipids remain, however, to be precisely studied. The present results indicate that Arf6 is implicated in the regulation of phagocytosis in part by controlling PLD activation at the nascent phagosome, and we propose that Arf6 may serve to link the PA production to phagosome formation.

We thank the microscopy facilities of Plateforme Imagerie In Vitro of the Institut des Neurosciences Cellulaires et Intégratives.

The authors have no financial conflicts of interest.