PtdIns(3,4)P2, Lamellipodin, and VASP coordinate cytoskeletal remodeling during phagocytic cup formation in macrophages

Phosphoinositides are pivotal regulators of vesicular traffic and signaling during phagocytosis. Phagosome formation, the initial step of the process, is characterized by local membrane remodelling and reorganization of the actin cytoskeleton that leads to formation of the pseudopods that drive particle engulfment. Using genetically-encoded fluorescent probes we found that upon particle engagement a localized pool of PtdIns(3,4)P2 is generated by the sequential activities of class I phosphoinositide 3-kinases and phosphoinositide 5-phosphatases. Depletion of the enzymes responsible for this locally generated pool of PtdIns(3,4)P2 blocks pseudopod progression and ultimately phagocytosis. We show that the PtdIns(3,4)P2 effector Lamellipodin (Lpd) is recruited to nascent phagosomes by PtdIns(3,4)P2. Furthermore, we show that silencing of Lpd inhibits phagocytosis and produces aberrant pseudopodia with disorganized actin filaments. Lastly, vasodilator-stimulated phosphoprotein (VASP) was identified as a key actin-regulatory protein mediating phagosome formation downstream of Lpd. Mechanistically, our findings imply that a pathway involving PtdIns(3,4)P2, Lpd and VASP mediates phagocytosis at the stage of particle engulfment.


Introduction
Phagocytosis, the process whereby cells engulf and dispose of effete cells, microorganisms, and foreign particles, is pivotal for immunity and tissue homeostasis 1,2 .
Phagosome formation, the initial step during phagocytosis, entails marked reorganization of the actin cytoskeleton and membrane remodelling events that lead to pseudopod extension and particle engulfment 3,4 . Earlier studies have suggested that phosphoinositides are pivotal molecules in the regulation of phagocytosis. Phosphoinositides, which play crucial roles in signaling and membrane traffic 5 , reside primarily in the cytosolic leaflet of organelles including the plasma membrane and phagosomes. Phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P2) is enriched in the plasma membrane where, amongst its many roles, it acts as a positive regulator of actin polymerization 6 . The conversion of PtdIns(4,5)P2 to phosphatidylinositol 3,4,5-trisphosphate (PtdIns (3,4,5)P3) is a potent stimulus resulting in alterations in actin dynamics through the activation/inactivation of Rho-GTPases, as well as promoting the hydrolysis of PtdIns(4,5)P2 by activating phospholipase C Indeed, it has been appreciated for nearly two decades that IgG-opsonized particle engagement by the phagocytic Fc receptor leads to the activation of the phosphoinositide-3-kinase (PI3K) 7 that is required for the internalization of large phagocytic prey 8,9 . Remarkably, the dynamic changes in phosphoinositides and the actin cytoskeleton during phagocytosis are restricted to the site of particle engagement and do not propagate to the rest of the plasma membrane. Accordingly, phagosome formation is characterized by the focal generation of PtdIns (3,4,5)P3 10 at the site of contact and extending pseudopods, while a concomitant disappearance of PtdIns (4,5)P2 is observed at the base of the phagocytic cup prior to sealing of the nascent phagosome 11 .
The downstream effects of PI3K activation during phagocytosis have been mostly attributed to the generation of PtdIns (3,4,5)P3. However, some of these findings should be interpreted with caution, as most were obtained using probes with dual specificity for PtdIns (3,4,5)P3 and PtdIns (3,4)P2, such as the pleckstrin-homology (PH) domain of Akt 12,13 .

Detection of PtdIns(3,4)P2 in the plasma membrane of resting macrophages
The distribution of PtdIns (3,4)P2 in unstimulated RAW 264.7 macrophages was investigated first. To this end we employed a recently described genetically-encoded biosensor based on tandem carboxy-terminal PH (cPH) domains of TAPP1 21,22 . Using these biosensors, NES-mCh-cPHx3 and NES-EGFP-cPHx2 (cPHx3 and cPHx2 respectively), we detected a discrete plasmalemmal pool of PtdIns (3,4)P2 in resting cells ( Figure 1A, left panel). To confirm the specificity and responsiveness of the probes we co-expressed inositol polyphosphate 4phosphatase type II (INPP4B), which selectively hydrolyzes the D-4 position of PtdIns(3,4)P2 23,24 . To ensure plasmalemmal targeting of INPP4B, the phosphatase was attached to a carboxyterminal CAAX motif (INPP4B-CAAX) that is prenylated and polycationic. As shown in Figure 1A (middle and right panels) the biosensors detached from the membrane upon co-expression of INPP4B-CAAX, but not when the catalytically inactive INPP4B(C842A)-CAAX mutant was coexpressed. These observations validate the selectivity of the probes and confirm that modest yet detectable amounts of PtdIns (3,4)P2 are indeed present in the membrane of resting macrophages. Furthermore, we documented that production of PtdIns(3,4)P2 is PI3Kdependent, since treatment with nanomolar concentrations of the PI3K inhibitor wortmannin completely released the cPHx3 probe from the plasma membrane ( Figure 1B,C).

PtdIns(3,4)P2 accumulates at the site of particle engagement during phagocytosis
Next, we examined the distribution of PtdIns(3,4)P2 during phagocytosis. Upon exposure to IgG-opsonized sheep erythrocytes (SRBCs), RAW 264.7 cells exhibited a marked accumulation of the PtdIns(3,4)P2 probe at the site of phagocytosis to levels greatly surpassing those of the neighbouring unengaged plasma membrane, an observation consistent with stimulated local production of the lipid (Figure 2A, left panel). PtdIns (3,4)P2 accumulated in the phagosomal cup and remained in the membrane even after scission of the nascent phagosome (Video S1.). Of note, there were marked differences in the distribution and dynamics of recruitment of cPHx2 and PH-BTKx2 (a sensor for PtdIns (3,4,5)P3) during phagocytosis. Specifically, PtdIns(3,4,5)P3 accumulated primarily -albeit transiently -at the base of the phagocytic cup, whereas PtdIns(3,4,)P2 was acquired slightly later, initially attaining higher levels at the tips of the extending pseudopods ( Figure S1., Video S2.) and later persisting for a few minutes in the sealed phagosome. The accumulation of the PtdIns (3,4)P2 probe at the phagosomal cup was largely obliterated by co-expression of the INPP4B-CAAX construct, but no depletion was observed when the catalytically-inactive INPP4B(C842A)-CAAX construct was used as negative control (Figure 2A and B). These findings imply that PtdIns (3,4)P2 is produced locally at sites of phagocytosis.

Phagosomal PtdIns(3,4)P2 is produced downstream of Class I PI3K and 5-phosphatases
We next sought to elucidate the pathway(s) responsible for the biosynthesis of PtdIns(3,4)P2 during phagocytosis. We considered two possibilities, which are not mutually exclusive: that the PtdIns(3,4)P2 was being produced by dephosphorylation of PtdIns (3,4,5)P3 and/or that it was synthesized from PI(4)P by class II PI3Ks. Class I PI3K is known to be activated during phagocytosis and its inhibition arrests engulfment by preventing full extension of the pseudopods around the particle 8,9 . To assess its involvement, we treated cells with nanomolar concentrations of the pan-PI3K inhibitor PI-103 or the class I PI3K-selective inhibitor GDC-0941 and monitored PtdIns(3,4)P2 formation during phagocytosis ( Figure 2C and D). In both cases the recruitment of cPHx2 was virtually eliminated, consistent with a major role for class I PI3Ks in the generation of PtdIns (3,4)P2. Nevertheless, we examined the possible contribution of Class II PI3Ks. For this, we acutely depleted the plasmalemmal pool of PtdIns(4)P using GSK-A1, a potent and specific inhibitor of the type III phosphatidylinositol 4-kinase PI4KA (PI4KIIIα) that is largely responsible for the generation and maintenance of plasmalemmal PtdIns(4)P; it is noteworthy that selective inhibition of PI4KIIIα by GSK-A1 does not acutely alter the plasmalemmal levels PtdIns(4,5)P 2 25 , the substrate of class I PI3Ks. Upon treatment with GSK-A1, the recruitment of mCherry-cPHx3 to the phagocytic cup remained largely unaffected despite a profound depletion of PtdIns(4)P monitored by the high-avidity biosensor 2xP4M ( Figure 2E). These findings strongly suggest that dephosphorylation of PtdIns (3,4,5)P3 is the main source of PtdIns(3,4)P 2 production at the phagocytic cup.
The involvement of PI3Ks that generate PtdIns (3,4,5)P3 suggests that phosphoinositide 5phosphatases are also required for the synthesis of PtdIns(3,4)P2 during phagocytosis. Multiple 5-phosphatases could in theory play a role in the production of PtdIns (3,4)P2. For instance, Src homology 2 (SH2) domain-containing inositol-5-phosphatase 2 (SHIP2) is involved in the production of PtdIns (3,4)P2 at sites of endocytosis 26 and at invadopodia 27 . Importantly, SHIP2 has also been reported to translocate to sites of phagocytosis 28 , making it an attractive candidate to mediate the production of PtdIns(3,4)P2. However, overexpression of the catalytically-inactive mutant of SHIP2 (D607A), which is expected to exert a dominant-negative effect, had no effect over phagosomal levels of PtdIns(3,4)P2 ( Figure S2 A and B). Similar negative results were observed when the cells were treated with either a SHIP2-selective inhibitor (AS1949490) or with the pan-SHIP1/2 inhibitor (K118) (Figure S2 C). We therefore turned our attention to other 5-phosphatases: there is compelling evidence that SHIP1 29 , INPP5E 30 , and OCRL 31 are all recruited to nascent phagosomes. In addition, we identified synaptojanin-2 (SYNJ2) and INPP5B as being present at the site of phagocytosis ( Figure 2F). It is therefore conceivable that multiple 5-phosphatases collaborate to dephosphorylate PtdIns(3,4,5)P3 to PtdIns(3,4,)P2 32 . Because of the likelihood of functional redundance, the identity of the specific phosphatases involved was not pursued further.

Depletion of PtdIns(3,4)P2 impairs phagosome formation
We next sought to determine whether PtdIns(3,4)P2 accumulation is necessary for efficient FcR-mediated particle uptake. To this end, phagocytic efficiency was compared in Thus, we hypothesised that a defect in pseudopod extension or sealing must be responsible for the observed decrease in phagocytic efficiency.
Actin remodelling, which drives pseudopod extension, is a hallmark of the early stages of the phagocytic process. To assess the role of PtdIns (3,4) necessary for phagocytosis and that it plays a direct role in regulating actin organization and pseudopod progression during the early stages of particle engulfment.

Lamellipodin, a PtdIns(3,4)P2 effector, accumulates at the phagocytic cup
To date, only a handful of proteins have been identified as specific effectors of PtdIns(3,4)P2 18 . These include the modular adaptor protein Ras-associated and pleckstrin homology domains-containing protein 1, more commonly referred to as Lamellipodin (Lpd).
Through its ability to cluster and tether Ena/VASP proteins to actin filaments, Lpd is a key regulator of the actin cytoskeleton 33 . As such Lpd has roles in lamellipodial formation 34 , stabilization of actin-dependent cellular protrusions 35 , cell migration and endocytosis 26,[36][37][38][39] .
Because the PH domain of Lpd has been shown to bind PtdIns(3,4)P2 34 , it appeared a likely candidate for the regulation of pseudopod extension during particle engulfment. To examine this possibility, Lpd was expressed in RAW 264.7 macrophages and its distribution assessed during phagocytosis of IgG-opsonized SRBCs. We observed robust accumulation of GFP-Lpd at the phagocytic cup, where PtdIns(3,4)P 2 was also enriched ( Figure 4A). Furthermore, upon treatment with wortmannin the levels of both PtdIns (3,4)P2 and Lpd at the phagocytic cup decreased in parallel fashion ( Figure 4B,C). Consistent with the notion that Lpd binds to PtdIns(3,4)P2 via its PH domain, we found that a GFP-tagged tandem PH domain of Lpd (Lpd-2xPH) was recruited to the phagocytic cup in cells expressing the inactive INPP4B(C842A)-CAAX ( Figure 4D) yet failed to accumulate in cells expressing INPP4B-CAAX. Furthermore, in resting RAW 264.7 cells, Lpd-2xPH showed modest accumulation at the plasma membrane, as we had seen earlier for the cPHx3 and cPHx2 probes, and this limited recruitment was similarly abolished by INPP4B-CAAX, but not by the catalytically-inactive INPP4B(C842A)-CAAX ( Figure   S3). From these experiments we conclude that Lpd is recruited early during phagocytosis at least in part by its PH domain-mediated interaction with PtdIns(3,4)P2.

Silencing of Lpd results in aberrant phagocytic cups and arrests phagocytosis
We next sought to determine whether Lpd is necessary for FcR-mediated phagocytosis. The expression of endogenous Lpd in RAW 264.7 macrophages and its susceptibility to shRNAmediated silencing were first validated by immunostaining ( Figure 5A and Figure S5). The enrichment of the endogenous Lpd during phagocytosis was similarly confirmed by immunostaining; its distribution at the phagocytic cup closely resembled that of F-actin stained with phalloidin ( Figure 5B). Upon silencing of Lpd, we observed an aberrant "flaring" of F-actin around the opsonized particle ( Figure 5C, top right panel). Lpd-silenced macrophages exhibited "loose" phagocytic cups characterized by multiple protrusions, in contrast to the tightly apposed pseudopods of control cells. In addition, Lpd-shRNA cells exhibited a ~60% decrease in phagocytic efficiency compared to control cells ( Figure 5D, E). Taken together, these findings suggest that Lpd is necessary for proper F-actin organization and pseudopod extension during phagocytosis.

The Lamellipodin ligand VASP localizes to the phagocytic cup and is necessary for phagocytosis
Ena/VASP proteins promote actin polymerization by accelerating filament elongation and opposing the action of capping proteins [40][41][42] . Ena/VASP proteins regulate the cytoskeleton during T cell receptor (TCR)-signalling 43 and play a role in Fc-mediated phagocytosis 44 and macroendocytosis in Dictyostelium discoideum 45 . It is relevant that Ena/VASP proteins harbour an EVH1 domain that interacts with the multiple proline-rich regions (FPPPP) present in Lpd 33,34 ( Figure 6A) and that they jointly regulate the dynamics of filopodia 46 . We sought to determine if Lpd and VASP associate within the phagocytic cup. Co-expression of the Lpd and VASP constructs revealed colocalization of these proteins during phagocytosis. Notably, Lpd, VASP and F-actin all share a similar distribution within the phagocytic cup ( Figure 6B). We also confirmed the enrichment of endogenous VASP at the phagocytic cup through immunostaining ( Figure 6E).
Next, we tested whether VASP was necessary for phagocytosis. To this end, we took advantage of the Listeria monocytogenes effector protein ActA which binds tightly to VASP. We expressed an N-terminally truncated ActA protein which includes the four repeats of the VASPbinding sequences attached to a C-terminal motif that targets the protein to the cytosolic surface of mitochondria (mRFP-Mito-FP4) 47,48 . Expression of mRFP-Mito-FP4 effectively sequestered the vast majority of VASP to the mitochondria and prevented its accumulation at the phagocytic cup, while a mutant version (mRFP-Mito-AP4) that also targets to mitochondria but is unable to bind VASP was without effect ( Figures 6C and S6). Tethering of VASP to the mitochondria by means of mRFP-Mito-FP4 reduced the phagocytic efficiency from 79.8% to 31.4% ( Figure 6D). These results demonstrate that VASP is a positive regulator of phagocytosis and support the notion that Lpd is exerting its effects at least in part by binding to VASP.

Lamellipodin-VASP interactions coordinate actin polymerization at the phagocytic cup
Next, we examined whether the interaction between Lpd and VASP was necessary for phagocytosis. We overexpressed a mutant version of Lpd in which all Ena/VASP-binding sites had been inactivated through mutations (Lpd EVmut ). Overexpression of GFP-Lpd EVmut produced a ~57% decrease in phagocytosis when compared to RAW 264.7 cells transfected with the wildtype GFP-Lpd construct ( Figure 7A-B). Furthermore, even though the initial recruitment of GFP-Lpd EVmut to the phagocytic cup was unaffected, the distribution of the construct became altered during phagocytic cup formation ( Figure S7). We noticed that the phagocytic cups formed by this mutant contained multiple filopodia-and ruffle-like projections that were rich in F-actin, as visualized through phalloidin staining ( Figure 7C). These atypical phagocytic cups and pseudopods closely resembled the ones we had previously observed upon silencing of Lpd ( Figure 5C). Additionally, these aberrant phagocytic cups were also enriched in PtdIns(3,4)P 2 , detected by the cPHx3 probe ( Figure 7D) demonstrating that the lipid metabolism itself was unperturbed by expression of the Lpd mutant. These findings suggest that interactions between Lpd and VASP are required for proper cytoskeleton organization and coordinated extension of pseudopods at the site of particle engulfment.

Discussion
We found that comparatively small amounts of PtdIns (3,4)P2 are present in the plasma membrane of resting RAW 264.7 macrophages and that this phosphoinositide is greatly enriched at the sites of particle engagement during FcR-mediated phagosome formation.
Selective depletion using a highly specific phosphatase revealed that PtdIns(3,4)P2 has a critical role in supporting phagocytosis. This was previously unrecognized in part due to the use of probes like AKT-PH that have dual specificity and of PI3K inhibitors, rather than targeted enzymatic depletion by INPP4B. We also report that Lpd and its binding partner VASP are jointly required for robust phagocytosis: loss of either from the site of phagocytosis results in an atypical and seemingly uncoordinated actin assembly within the extending pseudopods that hence fail to encircle the target.
The importance of PI3Ks during the process of phagocytosis has been widely documented 49,50,51 . Genetic manipulation of PI3Ks as well as the use of pharmacological inhibitors impairs phagocytosis, particularly that of larger targets. Additionally, localized production of PtdIns(3,4)P2 or PtdIns (3,4,5)P3 at the phagocytic cup has been implicated in phagosomal closure and Ca 2+ signalling in HL60 neutrophils 10 . Although some of these defects can be directly linked to the role of PtdIns(3,4,5)P3 cognate effectors, based on this study it is likely that a second, underappreciated effect of PI3K inhibition is the impairment of PtdIns(3,4)P2 formation, which is itself required for phagocytosis. Curiously, in our studies we find that PtdIns(3,4)P 2 is found at both the base of the phagocytic cup as well as at the tips of the pseudopods, whereas PtdIns(3,4,5)P3 is largely restricted to the base of the phagocytic cup ( Figure S1). This raises the possibility that the relevant phosphoinositide 5-phosphatases are particularly active at the tips of the advancing pseudopods. Phenotypically, inhibition of PI3Kand thus attenuation of both PtdIns(3,4)P2 and PtdIns(3,4,5)P3 production-results in arrest of pseudopod progression roughly halfway around the opsonized particle 9,52 , while selective depletion of PtdIns(3,4)P2 prevents pseudopod extension ( Figure 3C right panel, 3D, and Video S3). PtdIns (3,4,5)P3 is thought to regulate actin dynamics during phagocytosis by locally activating Rho family GEFs and GAPs 53 , and by activating phospholipase C, thereby depleting PtdIns(4,5)P2 from the base of the phagocytic cup. In view of our observations, it seems worthwhile to reconsider whether the purported PtdIns(3,4,5)P3 effectors are in fact selective for this lipid or are instead responsive to PtdIns(3,4)P 2 .
An earlier report suggested that the inositol polyphosphate 4-phosphatase type 1 (INPP4A), which is expressed in RAW macrophages, is a negative regulator of phagocytosis 54 . That study documented enhanced ability to internalize particles by RAW cells expressing shINPP4A and by primary macrophages from INPP4A-knockout mice. While the authors did not provide a mechanistic explanation for the observed effects, their findings are consistent with our conclusion that increased PtdIns(3,4)P2 is essential for optimal phagocytosis, at least partly via Lpd and VASP.
Lpd belongs to the Mig-10, RIAM, Lpd (MRL) family of modular adaptor proteins 55 . RIAM, despite sharing homology with Lpd 56 , is thought to act downstream of Rap1, and evidence suggests that its PH domain binds preferentially to PtdIns(3)P, PtdIns(5)P 57 and PtdIns(4,5)P2 58,59 , not PtdIns(3,4)P2. Previously, RIAM was described to support complementdependent phagocytosis by relaying integrin signaling 60 . However, RIAM is not required for FcR-dependent phagocytosis 61 , the mode we studied. Furthermore, our results indicate that silencing of Lpd produces severe defects in the phagocytic cups of macrophages where RIAM was left intact. These findings suggest that Lpd, contrary to RIAM, is necessary for Fc-mediated phagocytosis and that these MRL proteins play non-redundant roles during this process.
We found that the PH domain of Lpd is sufficient for its localization at the phagocytic cup. It is possible however, that other interactions contribute to its recruitment and retention within the forming phagosome. Indeed, we noted that Lpd dissociates from the maturing phagosome prior to the disappearance of the PtdIns(3,4)P2, suggesting that it may function as a coincidence detector; other potential binding partners may well be present at the site of phagocytosis. In this regard, it is noteworthy that Lpd directly binds actin filaments through an unstructured and highly basic region in its carboxyl-terminus 33 . Furthermore, Lpd can also bind to active Rac1 62,63 , which is similarly recruited to forming phagocytic cups 64 . Additionally, the formin-binding protein 17 (FBP17) binds to Lpd and recruits it to the membrane during fast endophilin-mediated endocytosis 65 . Of note, FPB17 is also present within podosomes and phagocytic cups in macrophages 66 .
Lpd is central to the regulation of cytoskeletal dynamics 34 as it interacts with multiple actin regulators such as the Ena/VASP family proteins 33,37,67 and the SCAR/WAVE regulatory complex 37,62 . Ena/VASP proteins localize to the phagocytic cup and are indispensable for phagocytosis: Coppolino et al. 44 reported that upon binding of Ena/VASP to a cytosolic GFP-ActA construct, phagocytosis was inhibited. In the current study, we were able to validate these findings using a more thorough method that entailed mistargeting Ena/VASP proteins to the mitochondria. Additionally, we found that all three Ena/VASP family members (EVL, Mena, VASP) localize to the phagocytic cup when ectopically expressed ( Figure S8). Phosphorylation of Lpd by c-Abl kinase reportedly increases its interaction with Ena/VASP 68 . While we did not explore this possibility in the current study, Abl family kinases have been implicated in the positive regulation of phagocytosis 69 . Therefore, it is possible that a kinase-dependent increase in Lpd-Ena/VASP interaction is one of the mechanisms by which this positive regulation takes place. Furthermore, PI3K activation through Abl kinase was also recently reported to take place during podosome formation in macrophages 70 .
In addition to VASP, Lpd can also regulate the actin cytoskeleton through its interactions with the SCAR/WAVE complex. A recent study demonstrated that loss of the WAVE regulatory complex (WRC) impairs phagocytosis 71 . In our study, we compared the effects of Lpd mutants in which all Ena/VASP or all SCAR/WAVE binding sites had been mutated. The Lpd EVmut caused the most robust decrease in phagocytic efficiency and aberrant phagocytic cups (Figure 7). Nevertheless, the SCAR/WAVE binding-deficient mutant also exhibited a smaller, yet significant, decrease in phagocytic efficiency ( Figure S9B and C). Furthermore, Abi1, the component of the WRC which interacts directly with Lpd 62 , was also localized to the phagocytic cup ( Figure S9A).
Jointly, these findings suggest that Lpd is central to the regulation of actin dynamics during phagocytic cup formation not only through its interactions with VASP, but probably also through the regulation of the WRC, although additional experiments will be needed to validate the involvement of the latter pathway.
The diversity of phagocytic targets and receptors entails different molecular mechanisms during phagocytic cup formation. Despite these differences, all types of phagocytosis share an inherent dependence on the rearrangement of the actin cytoskeleton and on the dynamic remodelling of the plasma membrane. FcR-mediated phagocytosis is the best characterized model of phagocytosis and the one used throughout this study. However, other equally important modes of opsonin-dependent and independent phagocytosis exist. These include phagocytosis mediated by CR3, receptors that recognize activated complement components, such as iC3b, deposited on the phagocytic target 72 . As in the case of FcR-mediated phagocytosis, we were able to document that localized synthesis of PtdIns(3,4)P2 and recruitment of Lpd and VASP occur also at phagocytic cups generated by engagement of CR3 ( Figure S10A, B and C). Interestingly, the overall efficiency of complement-induced phagocytosis was largely unaffected by expression of INPP4B ( Figure S10D). One possibility for this difference is compensation by the Lpd homolog RIAM, which is associated with the activation of integrins like CR3 but not FcR and has the ability to bind multiple phosphoinositides including PtdIns(3,4,5)P3 and PtdIns(4,5)P2 59 , and recently shown to be required for CR3-mediated phagocytosis in HL60 cells 60 . In yet another mode of phagocytosis, the scaffolding protein SWAP70 was identified as a PtdIns(3,4)P2 effector critical for the phagocytosis of yeast particles in dendritic cells 73,74 . Collectively, these results demonstrate that PtdIns(3,4)P2 regulates actin dynamics through multiple effectors in at least a subset of phagocytic receptors and cell types.

Plasmids
The plasmids used in this study are summarized in Table 1. Where indicated, plasmids were constructed using In-Fusion HD EcoDry Cloning Kits (Takara) and were verified by Sanger sequencing prior to transfection into mammalian cells.

Cell Culture
The murine macrophage RAW 264.7 cell line was obtained from the American Type Culture Collection (Manassas, VA) and cultured in DMEM (Wisent Bioproducts, Saint-Jean-Baptiste, Canada) supplemented with 10% heat-inactivated fetal bovine serum and incubated at 37°C under 5% CO2.

Phagocytosis assays
Quantitative phagocytosis assays have been described previously 75 . Briefly, ∼5 × 10 4 cells were seeded onto 1.8-cm glass coverslips and grown for 18-24 h. Opsonization of SRBC was performed by incubating 100 μL of a 10% SRBC suspension with 3 μL of rabbit IgG fraction against SRBCs at 37°C for 1 h. SRBCs were then labelled with a fluorescent secondary antibody against rabbit. 10 μL of labelled and opsonized SRBCs was added to individual coverslips containing RAW cells within 12-well plates or directly into imaging chambers for live-cell imaging. Synchronization of phagocytosis was achieved by centrifugation (300 ×g for 10 s) of the phagocytic targets onto the cell-containing coverslips. Phagocytosis was terminated by replacing the medium with ice-cold PBS. To obtain the differential inside-outside staining used throughout this study, cells were incubated with cold PBS containing a secondary antibody against rabbit to label non-internalized SRBCs, previously internalized labelled-SRBCs are not accessible to this secondary antibody staining and therefore are not additionally labelled.

Antibodies and reagents
Rabbit polyclonal Lpd antibody 34

Confocal microscopy and image processing
Confocal fluorescence microscopy was performed using spinning-disk confocal microscopes

Statistics and reproducibility
Statistical analyses were performed using GraphPad PRISM. Statistical details for each experiment are also stated in the figure legends. Statistical significance (****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05 and NS (not significant)) is denoted in figures when applicable. All microscopy-based experiments were performed independently at least three times.