The motor protein myosin 1G functions in FcγR-mediated phagocytosis

Phagocytosis allows macrophages and other cells of the immune system to engulf and destroy particles, and hence plays a key part in the innate and adaptive immune response (Aderem and Underhill, 1999). This cellular process is initiated by the engagement of phagocyte surface receptors with ligands on a particulate target, which triggers the recruitment and activation of a variety of signalling pathways. Receptor signalling culminates in the local reorganisation of the actin cytoskeleton to form a phagocytic cup for internalisation of the bound particle (Aderem and Underhill, 1999; Swanson, 2008). The complexity of the signalling pathways is staggering, not only regulating engulfment, but also particle degradation in the phagosome, chemokine production, and gene expression (Underhill and Ozinsky, 2002). In addition to the molecular biology, phagocytosis also involves the generation of physical forces for cell-shape changes and engulfment. We previously found that the actin-rich cup works as a unidirectional ratchet for robust engulfment of large particles (Tollis et al., 2010). In addition to actin-driven protrusions, contractile forces are needed to close the cups (Swanson et al., 1999), but surprisingly little is known about the mechanism. Importantly, contractions need to be localised to the leading edge of the phagocytic cup and be highly coordinated with the actin cytoskeleton; if contraction occurs too early during engulfment, particles would be pushed out instead of being engulfed. Hence, what are the key molecular players and how are they regulated for such contraction?

The most extensively characterised phagocytic pathway is downstream of the Fcγ opsonin receptor (FcγR) in macrophages (Ravetch, 1994). Ligation of the FcγR by an immunoglobulin-G (IgG)-opsonised particle stimulates pseudopod extensions around the particle for cup formation. Specifically, binding to the particle triggers receptor clustering and tyrosine phosphorylation in the cytoplasmic ITAM (Immunoreceptor Tyrosine-based Activation Motif) domain of the receptors by Src kinases (Mitchell et al., 1994). In addition, ITAM phosphorylation leads to the recruitment and activation of phosphoinositide 3-kinases (PI3Ks) (Araki et al., 1996; Cox et al., 1999), which convert the lipid substrate phosphatidylinositol (4,5)-bisphosphate [PI(4,5)P₂] to the product phosphatidylinositol (3,4,5)-triphosphate [PI(3,4,5)P₃] (Cox et al., 1999; Ninomiya et al., 1994). Such changes in the phospholipid metabolism during FcγR-mediated uptake create a reactive surface, and are crucial in determining the recruitment and activation of downstream signalling molecules (Botelho et al., 2000; Grinstein, 2010). Although aspects of FcγR signalling are well described, relatively little is known about how motor proteins are regulated.

Several different actin-based motor proteins have been implicated in mammalian phagocytosis including members of the class-I myosins, as well as Myo2a, Myo5a and Myo10 (reviewed by Araki, 2006). However, it is unclear how forces for cup closure are created and if there are yet other, previously uncharacterised myosins essential during FcγR-mediated phagocytosis. Recently, one particular class-I myosin, myosin 1G (Myo1G), has been identified as being haematopoietic cell-specific and, interestingly, has been shown to regulate cell elasticity (Olety et al., 2010; Patino-Lopez et al., 2010). Furthermore, Myo1G exhibits a striking plasma membrane localisation produced by binding to membrane lipids, namely phospholipids through a PH-like domain (Olety et al., 2010; Patino-Lopez et al., 2010). Herein we determined that class-I myosin Myo1G is a key regulator of phagocytic cups during FcγR-mediated phagocytosis.

Several different classes of myosins have been shown to localise at phagocytic cups downstream of FcγR ligation (Allen and Aderem, 1995; Stendahl et al., 1980; Swanson et al., 1999). Consistent with the previous reports (Araki et al., 2003; Swanson et al., 1999), we found that Myo2 inhibition with the use of the MLCK (myosin light chain kinase) inhibitor ML-7 dramatically perturbs the progression of FcγR-mediated phagocytosis (Fig. 1). Interestingly, with the use of SEM we noted that the phagocytic cups formed in control (DMSO)-treated J774A.1 macrophages appear to comprise of large overlapping membrane ruffles that enclose the particles; an observation which differs from a couple of previous descriptions of cylindrical uniform cups and their inhibition by ML-7 in scanning electron micrographs (Araki et al., 1996; Araki et al., 2003). The possibility that this variability arises from the different cell types used in our study compared to those of others cannot be excluded, especially since the J774A.1 macrophages utilised in this work exhibited a high level of filopodial formation (Fig. 1A). Furthermore, we noted an overall reduction in the number of dorsal filopodia in ML-7-treated macrophages compared to those treated with DMSO, highlighting the importance of myosin function in filopodial formation.

Myo1G is a class-I myosin (Fig. 2A) reported to be predominantly expressed in immune cells (Olety et al., 2010; Patino-Lopez et al., 2010) and, therefore, represented a likely candidate for a myosin controlling the contractile activities associated with phagocytosis. As a first step to determine a possible role for Myo1G in FcγR-directed uptake, we examined the localisation of Myo1G in macrophages undergoing phagocytosis by immunofluorescence. Using an antibody specific to Myo1G, we found that Myo1G is highly enriched in early-forming FcγR phagocytic cups relative to the background intensity, similar to F-actin (Fig. 2B, magnified particle). Furthermore, Myo1G immunoreactivity was detected to co-localise with phagosomal membranes (supplementary material Fig. S2B). To reinforce the visual interpretation that Myo1G preferentially accumulates at phagocytic cups, we simultaneously quantified the pixel intensity using a line scan across the phagosomal membranes (Fig. 2C; supplementary material Fig. S2). The fluorescence intensity was corrected for the apparent enrichment by the membrane cup shape (Fig. 2C, inset; see Materials and Methods for details). As expected, the intensity of Myo1G fluorescence was significantly elevated in occurrence with a concomitant increase of F-actin, where the pseudopods of the phagocytic cup were making contact with the particle. These data indicate that Myo1G is localised in phagocytic cups of macrophages following binding to IgG-particles.

To better evaluate the distribution of Myo1G at phagocytic cups, we utilised a tractable system of FcγR-transfected COS-7 cells to isolate the signalling pathway being investigated. It was important to initially confirm that the accumulation of GFP-tagged Myo1G at phagocytic cups was a direct consequence of signalling through the FcγR and not simply an artefact of overexpression. To control for this, we transfected COS-7 cells with a signalling-dead mutant receptor FcγR(Y282F/Y298F) in which the ITAM tyrosines have been mutated to phenylalanines (Tollis et al., 2010) and GFP-Myo1G. We find that Myo1G is not enriched at sites of particle binding in this case, demonstrating that this myosin is selectively recruited and requires phosphorylation of the ITAM tyrosines by Src kinases immediately downstream of IgG-FcγR binding (supplementary material Fig. S3). We then used these COS-7 cells, expressing wild-type FcγR and a GFP-tagged version of the full-length Myo1G, to observe the localisation of this myosin over time during phagocytosis. Myo1G appears to accumulate at early phagocytic cups forming after 5 minutes of phagocytic challenge with 3 µm diameter particles. With increasing time after incubation with IgG-coated particles (10 and 15 minutes), the numbers of Myo1G-positive phagocytic cups was significantly greater (Fig. 3A,B) and followed the kinetics of phagosomal F-actin immunoreactivity (supplementary material Fig. S4A). Furthermore, Myo1G is also progressively recruited to phagocytic cups during engulfment of 6 µm particles (Fig. 3B; supplementary material Fig. S4B). More specifically it was noted that at the earliest time points, such as at 5 minutes, there was a greater differential localisation of GFP-Myo1G and F-actin at phagocytic cups indicating that Myo1G is enriched at phagosomal membranes before the appearance of F-actin (supplementary material Fig. S4A, inset; Fig. 3C). As expected, at later time points many GFP-Myo1G-positive cups were also labelled with F-actin (Fig. 3C). Orthogonal views from confocal microscopy revealed the distinct localisation of GFP-Myo1G and F-actin within the phagocytic cup formed from the uptake of a 6 µm particle after 10 minutes of phagocytic challenge (Fig. 3D). F-actin (red) is predominantly accumulating at the base of the cup with strong co-localisation with GFP-Myo1G observed at the cup edges or pseudopods.

It is well established that the magnitude of pseudopod extension and phagosomal membranes increases with particle diameter (Cox et al., 1999); it therefore seems plausible that a greater need for pseudopod extension to enclose a large particle would simultaneously require increased contractile forces. Indeed, it has already been shown that Myo10 is necessary for maximal pseudopod extension associated with the ingestion of large particles (Cox et al., 2002). Hence, we investigated whether there was a greater dependence on Myo1G for the engulfment of large particles over small particles. FcγR-expressing COS-7 cells were transfected with either GFP or GFP-Myo1G and incubated with either 3 or 6 µm particles. Expression of GFP-Myo1G compared to GFP alone enhanced the phagocytosis of both 3 and 6 µm particles. Moreover, we observed a more significant increase in the uptake of 6 µm compared to 3 µm particles in the presence of GFP-Myo1G (Fig. 3E); this result fits with a positive role for Myo1G in uptake of larger diameter particles known to rely more on contractile activities for cup closure. It also corroborates a recent study combining in silico biophysical modelling with in vivo fluorescence imaging, which revealed that engulfment of 3 µm particles is not critically dependent on the acto-myosin contractile system, and might largely be driven by membrane fluctuations. In contrast, when challenged with 6 µm particles, cells require a fully functional acto-myosin cytoskeleton to support membrane fluctuations for particle wrapping, cup progression and closure, in order to overcome energy costs associated with larger membrane stretching and bending (Tollis et al., 2010).

Following ligation of the FcγR, there is a rise in PI3K activity and, accordingly, a temporal sequence of phosphoinositide metabolism, which is thought to be crucial for the ensuing cytoskeletal remodelling at the phagocytic cup (Botelho et al., 2000; Ninomiya et al., 1994). To explore a potential role for PI3K function in the localisation of Myo1G to sites of particle binding, we incubated FcγR- and Myo1G-expressing COS-7 cells with LY294002, an inhibitor of PI3Ks and then challenged these cells with either IgG-opsonised 3 µm (Fig. 4A, left panel; supplementary material Fig. S5) or 6 µm (Fig. 4A, right panel) particles. Irrespective of particle size, the accumulation of Myo1G to phagocytic cups was significantly reduced.

To pinpoint the spatio-temporal effect of the inhibitor LY294002, we reconstructed phagocytic cups by 3D image analysis in Fig. 4B. For this purpose, we determined the actin and myosin fluorescence intensities in cups, rotationally averaged around the particle-cell axis for multiple particles (see Materials and Methods for details). In untreated cells (Fig. 4Biiii) cups show a strong correlation (Pearson's correlation coefficient in blue) between Myo1G (green) and F-actin (red), especially at later stages of engulfment. This confirms the enrichment we observed in J774A.1 macrophages (Fig. 2C) and receptor/Myo1G-transfected COS-7 cells (Fig. 3). In contrast, the correlation in inhibitor-treated cells (Fig. 4Biv,v) is weaker, and cups are much broader and less defined. (Note that for treated cells we do not show specific time points as engulfment is much slower, which leads to a large spread of cup sizes with time.) This observation is further quantified in Fig. 4C which shows histograms of the Pearson's correlation coefficient calculated in nearly completed cups (at least 70% engulfment) for both treated and untreated cells. For treated cells the distribution is peaked around zero correlations, while for untreated cells correlations are mostly positive with a statistically significant shift of the distribution. These findings demonstrate that Myo1G is recruited in a PI3K-dependent manner, suggesting that the localisation of Myo1G at phagosomal membranes is dependent on PI(3,4,5)P₃ levels.

Since Myo1G (this work) and a number of other myosins (Cox et al., 2002; Swanson et al., 1999) are present in FcγR-mediated phagocytic cups, we sought to identify the specific role Myo1G is playing in uptake. This was achieved by analysing the contribution of the domains of Myo1G to phagocytosis by transiently transfecting truncation and point mutants of this myosin in FcγR-expressing COS-7 cells (Fig. 5A). In line with previous studies (Olety et al., 2010), deletion of the motor domain (truncated Myo1G constructs IQ+tail and tail alone) resulted in a preferential re-distribution of Myo1G from the plasma membrane to the nucleus (supplementary material Fig. S6). We therefore found that the constructs lacking the motor domain of Myo1G did not localise at FcγR phagocytic cups (supplementary material Fig. S6) and consequently inhibited engulfment downstream of the FcγR (70±1.5% reduction for the Myo1G tail construct compared to wild-type (full-length) Myo1G (Myo1G WT), and 63±0.5% for the Myo1G IQ+tail construct; Fig. 5B).

Class-I myosins have a putative PH-like domain in their tail region consisting of a β1-loop-β2 motif with two conserved basic residues crucial for binding phosphoinositides (Hokanson et al., 2006). Within murine Myo1G, these two residues are K883 and R893. To investigate the importance of these two residues and hence the involvement of the PH-like domain in FcγR-mediated phagocytosis, we utilised a Myo1G construct, Myo1G (K883A/R893A), in which these amino acids were mutated. As with the truncation mutants, GFP-Myo1G (K883A/R893A) was not localised at the plasma membrane and instead was often concentrated in the cytosol and the nucleus (Fig. 5C; supplementary material Fig. S6). Moreover, this point mutant did not accumulate at phagocytic cups and inhibited uptake by 60% compared to GFP-Myo1G WT-expressing COS-7 cells (Fig. 5B; supplementary material Fig. S6). Thus, both the motor domain and the PH-like domain are responsible for specific targeting of Myo1G to FcγR phagocytic cups which subsequently allows ingestion of particles to proceed.

In support of these data, we used SEM to examine how Myo1G directly influences phagocytic cup formation. Control COS-7 cells expressing the FcγR alone generated phagocytic cups consistent with those observed in J774A.1 macrophages (Fig. 5Di in comparison with Fig. 1A). These cups were formed from large membrane ruffles which were often tipped with thin membranous projections reminiscent of filopodia. Upon expression of Myo1G, the dorsal surface of the COS-7 cells appeared to have less pronounced membrane ruffles but instead had more prominent filopodia. Strikingly, the phagocytic cups seemed to be sealed once several filopodial tips had converged at the uppermost point of the particle (Fig. 5Dii; Fig. 5Diii for a schematic). Hence, filopodia may first grab the particle (Flannagan et al., 2010; Kress et al., 2007) and then guide pseudopods for cup formation.

To test which domains of Myo1G were important in generating the phagocytic cups following FcγR ligation, we transfected the truncation and point mutant constructs of Myo1G, along with the FcγR into COS-7 cells. Crucially, expression of the constructs, which were lacking the motor and IQ domains (Myo1G tail) or the motor domain alone (Myo1G IQ+tail), dramatically inhibited the formation of phagocytic cups, visible via fluorescence microscopy (supplementary material Fig. S6B,C, respectively) and scanning electron microscopy (Fig. 5D, iv and v, respectively). Interestingly, the dorsal surface of COS-7 cells expressing Myo1G IQ+tail exhibited small membrane ruffles contrasting with that of the Myo1G tail which showed an increased number of short filopodial-like structures. Moreover, COS-7 cells expressing Myo1G (K883A/R893A) were also unable to form phagocytic cups, despite small filopodia accumulating at the base of the particles (Fig. 5Dvi; supplementary material Fig. S6D). In addition, the dorsal surface of these cells was sparsely populated with filopodia. We believe that the inability of these Myo1G mutants to induce phagocytic cup formation is a direct result of their failure to localise at bound particles downstream of the FcγR ligation (Fig. 5B; supplementary material Figs S6, S7).

Through a putative PH domain, forming a β1-loop-β2 secondary structure, Myo1C has been identified to interact specifically with PI(4,5)P₂ (Hokanson and Ostap, 2006) although Myo1B can bind with high affinity to both PI(4,5)P₂ and PI(3,4,5)P₃ (Komaba and Coluccio, 2010). To ascertain whether Myo1G can interact with phosphoinositides, we used GFP-Myo1G WT-expressing COS-7 cell lysates in a lipid particle pull-down assay (Komaba and Coluccio, 2010). We found with this assay that Myo1G is primarily bound to PI(3,4)P₂ and PI(3,4,5)P₃ (Fig. 5E), confirming the dependence of Myo1G localisation on PI3K products. To independently validate the lipid particle pull-down and show that Myo1G specifically bound to acidic phospholipids through the putative PH domain, cell lysates expressing GFP-Myo1G WT or the Myo1G (K883A/R893A) mutant were incubated with a lipid array (PIP Strip). Bound proteins were detected with an anti-GFP antibody. Myo1G failed to bind most of the membrane lipids except for PI(4,5)P₂ and PI(3,4,5)P₃ (Fig. 5F). Furthermore, single alanine mutations at positions K883 and R893 in Myo1G were sufficient to abolish this binding to PIP₂ and PIP₃ (Fig. 5F) implying that these residues in the putative PH domain have a key role in the interaction between Myo1G and its phosphoinositide binding partners.

The contractile activity associated with FcγR-mediated phagocytosis is generated from the dynamic interplay between actin and myosin. Here we have focused on the immune cell-specific motor protein Myo1G, known to be a key regulator of cell elasticity (Olety et al., 2010; Patino-Lopez et al., 2010). We have shown by confocal fluorescence microscopy that this motor protein is recruited to phagocytic cups and is necessary for phagocytosis downstream of the FcγR (Figs 2, 5). Specifically, it is targeted to the phagocytic cup by the molecular signal of PI3K, activated by phosphorylation of the receptor ITAM domains (Fig. 4). In particular, our image analysis has revealed that when PI3K activity is inhibited, Myo1G and F-actin are no longer strongly co-localised, indicating that this kinase is important for the specific positioning of the myosin relative to F-actin (Fig. 4). We have found that the PH-like domain and the motor domain both contribute to targeting Myo1G to the phagocytic cup.

Myo1G predominantly binds PI(4,5)P₂, PI(3,4)P₂ and PI(3,4,5)P₃ (Fig. 5E,F). PI(3,4)P₂ and PI(3,4,5)P₃ are PI3K products recognised as regulating downstream target molecules via binding to PH domains (Isakoff et al., 1998). The association of Myo1G with the plasma membrane lipids is similar to that of the other class-I myosins. Both Myo1B (Komaba and Coluccio, 2010) and Myo1C (Hokanson and Ostap, 2006) can bind PI(4,5)P₂ and Myo1B can also bind PI(3,4,5)P₃. A specific lysine and arginine residue of Myo1B and Myo1C are important for their interaction with phosphoinositides (Hokanson et al., 2006; Komaba and Coluccio, 2010) and we have now shown that the corresponding amino acids 883 and 893 in the putative PH domain of the mouse sequence is essential for the binding of Myo1G to PI(4,5)P₂ and PI(3,4,5)P₃. However, we cannot discount that a third basic residue found in the β3 strand, lysine 904 in the mouse sequence (corresponding to K898 in the human sequence) could also confer Myo1G the specificity to bind phosphoinositides. This residue is only present in the Myo1G/Myo1D subfamily (and not in other class-I myosins), and facilitates Myo1G localisation to the membrane (Patino-Lopez et al., 2010). PIP₃ was identified as binding Myo1G in both the lipid bead-protein pull down assay and the lipid binding array, and given the dependence of Myo1G localisation on PI3K activity, of which PI(3,4,5)P₃ is the major product, we propose that Myo1G is more likely to associate with PI(3,4,5)P₃ than PIP₂ during phagocytosis. Although we have demonstrated that Myo1G binds both PIP₂ and PI(3,4,5)P₃, it may be that the affinity for PI(3,4,5)P₃ is greater than that of PIP₂. Upon FcγR ligation, PI(3,4,5)P₃ levels increase dramatically at cup membranes before phagosomal closure (Zhang et al., 2010). Moreover, high PI(3,4,5)P₃ concentrations allow the progression of phagosome formation to the later stages. It could be that Myo1G recruitment to and anchorage at early phagosomal membranes is mediated by PIP₂ although its continued presence at the phagocytic cup is regulated by binding to PI(3,4,5)P₃, from where it can function in the contractile activities associated with the later phases of phagocytic cup formation.

Although it is clear that the two positively charged residues in the PH-like domain of Myo1G are necessary for phagocytosis directed through the FcγR, they are not sufficient (Fig. 5B). The motor domain is also important since the Myo1G tail construct inhibited uptake (Fig. 5B). This may indicate that the truncated Myo1G, unable to bind F-actin, cannot be transported and enriched in the phagocytic cup, or that the head domain functions in the localisation and/or structural association with the phagosomal membranes. In line with these proposed mechanisms is the mislocalisation of the Myo1G tail construct upon phagocytic challenge (supplementary material Figs S6, S7).

The activity of PI3K is essential for mammalian phagocytosis as the PI3K product PI(3,4,5)P₃ represents a critical signature for commitment to engulfment (Zhang et al., 2010). The essential role of PI3K in phagocytosis has been explored in the well-characterised model organism Dictyostelium where the PI3K inhibitor LY294002 considerably decreases uptake of yeast, latex beads and bacteria (Clarke et al., 2010; Dormann et al., 2004). During engulfment of yeast particles by Dictyostelium, there is a rapid and transient accumulation of PI(3,4,5)P₃, which peaks around the time of phagosome closure (Dormann et al., 2004). Furthermore, there is a peak in the generation of PI(3,4)P₂, although the significance of this particular phosphoinositide species production in Dictyostelium has yet to be elucidated since no specific binding proteins have been identified (Dormann et al., 2004). Similarly to mammalian phagocytosis, Dictyostelium uptake of budding yeast is thought to involve several unconventional myosins with, specifically, class-I myosins MyoK, MyoC and MyoB all shown to localise at the constriction point of a phagocytic cup (Dieckmann et al., 2010). MyoB and the Arp2/3 complex are enriched close to the phagosome membrane and concentrated around the neck region of the budded yeast being engulfed (Clarke et al., 2010). Our image analysis revealed that the mammalian Myo1G is strongly co-localised with F-actin at phagocytic cups beyond half engulfment and dependent on PI3K (Fig. 4B).

Myo1G is targeted to the phagocytic cup by its preferential interaction with PI(3,4,5)P₃ and can additionally bind actin filaments via its head domain (Fig. 5G). It appears that the interaction between Myo1G and phosphoinositides is likely to be direct, given that the Myo1G (K883A/R893A) mutant could no longer bind PI(4,5)P₂ and PI(3,4,5)P₃ in the lipid binding assay. Indeed, it may be binding in a similar mechanism to kinesins, such as KIF16B which controls early endosome motility through a C-terminal PX domain that binds PI(3)P-containing early endosomes (Hoepfner et al., 2005). However, an indirect interaction to an additional binding partner cannot be ruled out as the lipid array in this study used cell lysates. A closely related myosin, Myo1E controls the movement of MHC-II vesicles along the actin cytoskeleton in dendritic cells through indirect interactions (Paul et al., 2011). This particular myosin regulates actin-based MHC-II transport by binding directly to actin and indirectly to PIP₂-containing vesicles through the effector protein ARF7EP of the GTPase ARL14/ARF7; it is the GEF PSD4 of ARL14/ARF7 which contains the PH domain which ultimately mediates the binding to phosphoinositides (Paul et al., 2011). Further studies using in vitro binding assays will reveal the specific nature of Myo1G interactions.

Due to its dual binding role, Myo1G is likely coupling the outward movement of the F-actin cytoskeleton with specific lipids in the plasma membrane, effectively leading to transport and enrichment of PI(3,4,5)P₃ at the leading edge of the phagocytic cup (Fig. 6A). To subsequently create a contractile force for cup closure, we envision that Myo1G pushes actin filaments away from the particle though its movement towards the barbed end of the filaments. This behaviour can increase the rate of actin polymerisation at the barbed end (Renkawitz et al., 2009) and thus create a particle-oriented force for contraction and cup closure (Fig. 6B). Hence, we have revealed how the temporal sequence of phosphoinositide metabolism generated by PI3K activity can control the precise spatial targeting of Myo1G to phagocytic cup membranes, wherein this motor protein can modulate pseudopod formation and contraction.

In conclusion we found that Myo1G is a key regulator of contractility in phagocytic cup closure during FcγR-mediated phagocytosis of large particles, linking actin cytoskeleton, myosin contraction, and lipid metabolism. Exploring further the differential regulation of multiple myosins in phagocytosis represents an exciting avenue of future research. For instance, Myo10, also modulated by PI3K, is important for pseudopod extension in FcγR-mediated phagocytosis of large particles (Cox et al., 2002). Therefore, the coordinated, concerted functions of a series of molecular motors, perhaps acting at different stages of uptake may play a crucial role in the force generation associated with phagocytic pathways. Since Myo1G is a significant player in phagocytosis, we speculate that it may also have roles in other varied cellular functions requiring contractile forces, such as cytokinesis and cell motility.

ML-7 and LY294002 were purchased from Calbiochem (Merck, Darmstadt, Germany). Rabbit anti-Myo1G antibody was from Rockland Inc. (Pennsylvania, USA). Wheat germ agglutinin (WGA), Texas Red-X conjugate was from Invitrogen.

Eukaryotic pRK5 expression vectors encoding the FcγRIIA and the FcγRIIA(Y282F/Y298F) have been previously described (Caron and Hall, 1998; Tollis et al., 2010). The mammalian expression vectors pEGFP-Myo1G, pEGFP-Myo1G tail, pEGFP-Myo1G IQ+tail and pEGFP-Myo1G (K883A/R893A) (Olety et al., 2010) were kind gifts from Martin Bähler.

All cell lines were maintained in Dulbecco's modified Eagle's medium (Invitrogen, Paisley, UK) supplemented with 10% heat-inactivated fetal bovine serum (PAA Laboratories), 100 units/ml penicillin (Sigma, Dorset, UK) and 0.1 mg/ml streptomycin (Sigma) at 37°C, 5% CO₂. COS-7 cells were transiently transfected using the Amaxa nucleoporation system (Amaxa Inc., Gaithersburg, USA; program A-024) and incubated at 37°C for 24 h to allow for expression of the corresponding proteins. HEK-293T cells were transiently transfected using the Calcium Phosphate transfection kit (Sigma) according to manufacturers' instructions.

Transfected COS-7 cells and J774A.1 macrophages were serum-starved for 1 h at 37°C in serum-free Dulbecco's modified Eagle's medium (SFM) with 10 mM HEPES (Sigma). Latex-polystyrene particles of 3 µm (Sigma) or 6 µm diameter (Polysciences Inc.) were opsonised by first incubating overnight at 4°C with 5% BSA fraction V (Sigma) in PBS followed by incubation with 1∶100 dilution of rabbit anti-BSA (Sigma) in PBS for 1 h at room temperature. Particles were re-suspended in ice-cold SFM at a concentration of 1.5×10⁶ particles per ml and 500 µl of particle suspension was added to each 13 mm glass coverslip. The onset of phagocytosis was synchronised by allowing the particles to bind at 4°C for 10 minutes. Uptake was then initiated by addition of pre-warmed medium and further incubation at 37°C for specified time periods. Washes with PBS followed before fixation in ice-cold 4% paraformaldehyde for 10 minutes at 4°C. Where indicated, macrophages were pre-incubated with 5 µM ML-7 for 10 minutes in 10 mM HEPES-buffered SFM before challenging with IgG-particles. ML-7 was also maintained in the medium during challenge. COS-7 cells were treated with 50 µM LY294002 in 10 mM HEPES-buffered SFM for 30 minutes before FcγR challenge. LY294002 was also added to the medium during challenge. Control conditions for these two inhibitors were cells treated with equivalent volumes of the vehicle DMSO.

Total bound and internalised IgG-coated particles were readily distinguished by utilising differential labelling of particles before and after a 3-minute permeabilisation with 0.2% Triton X-100. Generally, Alexa-Fluor-488-conjugated donkey anti-rabbit IgG was used to label external particles pre-permeabilisation and Alexa-Fluor-555-conjugated donkey anti-rabbit IgG (Invitrogen) to detect engulfed particles post-permeabilisation. In some cases, particles were labelled with Alexa-Fluor-633-conjugated donkey anti-rabbit IgG (Invitrogen) post-permeabilisation to detect total particles when looking at recruitment of GFP-tagged proteins to phagocytic cups. F-actin was stained using Alexa-Fluor-conjugated phalloidin and membrane was stained with Texas-Red-X-conjugated WGA (Invitrogen). Coverslips were mounted onto glass slides on 5 µl Mowiol 4–88 (Calbiochem) with ρ-phenylene diamine and viewed under an Olympus BX50 epifluorescence microscope. Confocal images were taken using the 100× oil objective of a LSM510 Zeiss Axiovert 100 M microscope. Images captured from the confocal microscope were typically 512×512 pixels and z-stacks were acquired at a minimum of 20 slices at 0.4 µm apart.

Phagocytosis was quantified as the proportion of bound particles that were internalised from 100 cells. Protein recruitment to phagocytic cups was scored as the number of cups formed at sites of particle attachment positive for a particular marker out of the total number of phagocytic cups (typically ≧100) from 20 cells.

For SEM, cells were grown on glass coverslips and challenged with IgG-coated particles as described above and in our previous work (Hemrajani et al., 2010). Phagocytosis was arrested by fixation in 3% glutaraldehyde. The samples were then postfixed in 1% OsO₄ (Sigma). Dehydration was carried out with an ascending series of ethanol solutions. Finally, samples were critical point-dried using an Emitech K850 Critical Point Drier, mounted onto stubs and sputter coated with a thin layer of gold with the use of a Polaron SC7620 Mini Sputter Coater. Images were acquired with a JEOL JSM-6390 scanning electron microscope.

This assay was performed using the PIP Particle Sample Pack (Echelon Biosciences Inc., Salt Lake City, USA) consisting of 50% slurries of six different species of phosphoinositide-bound beads and uncoated control particles. 50 µl of each 50% slurry were washed in 1 ml of binding buffer (10 mM HEPES pH 7.4, 150 mM NaCl and 0.25% Igepal CA630) and spun at 100 g for 5 minutes at 4°C. COS-7 cells expressing the GFP-Myo1G construct were lysed in 250 µl modified lysis buffer [50 mM Tris pH 7.2, 10 mM MgCl₂, 150 mM NaCl, 0.2% Igepal CA630 and Complete Protease inhibitor (Roche Applied Science)] and then incubated with the beads for 90 minutes at 4°C with gentle rotation. The beads were pelleted at 100 g for 5 minutes at 4°C followed by three washes each with 500 µl binding buffer at 100 g for 3 minutes. The supernatant was removed and the bound protein was eluted with 50 µl 2× Laemmli sample buffer. Samples were separated by SDS-PAGE and analysed by western blotting using the rabbit anti-Myo1G antibody.

This assay was performed using PIP Strips membranes (Molecular Probes, Invitrogen) containing 100 pmol samples of 15 different phospholipids and a blank sample. HEK-293T cells expressing GFP-Myo1G WT and mutant GFP-Myo1G (K883A/R893A) were lysed in lysis buffer (50 mM Tris pH 8.0, 10 mM EDTA, 100 mM NaCl, 0.5% Triton X-100 and Complete Protease inhibitor). Membranes were blocked for 1 hour in 3% fatty acid-free BSA (Calbiochem) in TBS-T before being incubated with cell lysate in 3% BSA in TBS-T overnight at 4°C with gentle rotation. After three 10-minute washes with TBS-T, the membranes were incubated with mouse anti-GFP antibody (Roche) in 1% BSA in TBS-T for 1 hour. Following a further three washes, membranes were then incubated with goat anti-mouse IgG-HRP (Dako, UK) for 1 hour. Lipid binding was detected using standard ECL (Pierce, Thermo Fisher Scientific, USA).

Statistical analyses were carried out using Prism 4.0 software (GraphPad). Unpaired, two-tailed Student's t-test was performed to determine statistical significance of the results. Data sets were considered different for P values <0.05.

We would like to thank Martin Bähler for the GFP-fusion proteins of Myo1G, and Thierry Soldati for a critical reading of the manuscript. The author(s) have made the following declarations about their contributions: A.E.D. and R.G.E. conceived and designed the experiments. A.E.D. and M.D.B. performed the experiments. A.E.D. and S.T. analysed the data. G.F. contributed reagents, materials and analysis tools. A.E.D. and R.G.E. wrote the paper. The authors declare that they have no conflict of interest.