TMEM55a localizes to macrophage phagosomes to downregulate phagocytosis

Phagocytosis by professional phagocytes is central for the elimination of microbial pathogens, innate and adaptive immunity, and clearance of apoptotic cells and debris. Although the precise mechanisms involved in phagocytosis vary depending on receptors involved or target particle sizes, they share the common feature of regulation through phosphoinositide derivatives. For example, F-actin assembly accelerated by PtdIns(4,5)P₂ facilitates the extension of pseudopods, which are indispensable for the initial capture of phagocytic targets regardless of the target species (Botelho et al., 2000; Freeman and Grinstein, 2014; Szymańska et al., 2008). However, the following phase, closure of the phagocytic cup, proceeds via different mechanisms between large (>5 µm) and small targets. Phagosomal accumulation of PtdIns(3,4,5)P₃, as catalyzed by the class I phosphoinositide 3-kinase (PI3K)-induced phosphorylation of PtdIns(4,5)P₂, is essential for the ingestion of large particles but is dispensable for the uptake of smaller targets (Araki et al., 1996; Cox et al., 1999; Schlam et al., 2015). The PI3K product functions to recruit Rho GTPase-activating proteins (RhoGAPs) to the phagocytic cup, which are then responsible for the inactivation of Rho family GTPases and the termination of actin assembly (Schlam et al., 2015). This coordinated disassembly of F-actin is required for the closure of the large phagocytic cup (Ikeda et al., 2017).

TMEM55a and TMEM55b, also known as type II (PIP4P2) and type I (PIP4P1) PtdIns(4,5)P₂ 4-phosphatase, respectively, were first identified as enzymes that catalyze the hydrolysis of PtdIns(4,5)P₂, but no other phosphoinositides, resulting in the generation of PtdIns(5)P (Ungewickell et al., 2005). Their enzymatic activities and substrate specificities have been determined in vitro, but the in vivo conversion of the polyphosphoinositide into PtdIns(5)P by the phosphatase has not yet been demonstrated and their roles have mostly been inferred from overexpression studies. Since the product of the phosphatase, PtdIns(5)P, predominantly exists in the nucleus and interacts with the plant homodomain (PHD) finger of inhibitor of growth protein-2 (ING2), it is suggested that TMEM55b (type I) generates nuclear PtdIns(5)P, which is responsible for activation of p53 (also known as TP53) and the resultant increase of apoptosis (Zou et al., 2007). However, when transfected in HeLa cells, both TMEM55a and TMEM55b are localized to late endosomal or lysosomal membrane, so it is likely that the phosphatases also play roles in the endosome (Ungewickell et al., 2005). Upon overexpression of these proteins in HEK293 cells, the degradation of EGF receptor following ligand binding is accelerated (Ungewickell et al., 2005). Accordingly, it was speculated that TMEM55a and TMEM55b are involved in the lysosomal degradation of plasma membrane receptors. However, ectopic expression of the virulence factor IpgD, a phosphatase with similar substrate specificity to that of TMEMs (Niebuhr et al., 2002), causes a delay in EGF receptor degradation (Boal et al., 2015; Ramel et al., 2011). Thus, the physiological role of the PtdIns(4,5)P₂ 4-phosphatases remains obscure.

In this paper, to investigate the physiological role of TMEM55a, we prepared Raw264.7 macrophages deficient in TMEM55a by using specific shRNA methodology. These cells possessed enhanced phagocytic activity against two distinct large particles, sheep red blood cells (RBCs) with membrane-bound IgG (hereafter IgG-RBCs) and zymosan, without affecting Escherichia coli phagocytosis. The accumulation of PtdIns(4,5)P₂ was increased, whereas the product of TMEM55a, PtdIns(5)P, was decreased on phagocytic cups in these cells, suggesting that TMEM55a dephosphorylates PtdIns(4,5)P₂ to PtdIns(5)P on phagosomes. This hypothesis was supported by the observation that fluorescent protein-tagged TMEM55a transfected in macrophages localizes to phagosomes during phagocytosis. The increased PtdIns(4,5)P₂ is likely immediately phosphorylated to PtdIns(3,4,5)P₃ by class I PI3K, because PtdIns(3,4,5)P₃ was also increased on phagocytic cups of TMEM55a-deficient cells. The augmentation of PtdIns(3,4,5)P₃ in addition to PtdIns(4,5)P₂ may thus be responsible for the target size-dependent activation of phagocytosis in the TMEM55a-knockdown cells.

To investigate the role of endogenous TMEM55a and TMEM55b, we generated RAW264.7 cells expressing shRNA directed against these phosphatases. Two cell lines (229 and 322 in Fig. 1A and Table S1) that produce shRNAs against distinct TMEM55a sequences (shTMEM55a) were prepared. The mRNA was reduced to 20% and 40% of that in control cells in the shTMEM55a-expressing lines 229 and 322, respectively (Fig. 1A). TMEM55b mRNA in the cells expressing shRNA against TMEM55b (shTMEM55b) (Table S1) decreased to 20% of that in control cells (Fig. 1B). We tried to estimate the endogenous expression of the TMEM55 proteins in the knockdown cells, but the antibodies from a commercial source did not work. To determine the knockdown efficiency, the knockdown cells were transfected with FLAG-tagged TMEM55a or TMEM55b and the expression was determined with anti-FLAG antibody. A large quantity of FLAG–TMEM55b protein was expressed in the control cells as well as shTMEM55a cells, but did not express in TMEM55b-knockdown cells (Fig. 1C, middle panel). The expression of FLAG–TMEM55a was far smaller than that of FLAG–TMEM55b but was detectable in a longer exposure (Fig. 1C, lower panel). The transfected TMEM55a was observed in the control cells but almost completely disappeared in TMEM55a-knockdown cells. Interestingly, the expression of FLAG–TMEM55a was increased in shTMEM55b cells, suggesting that FLAG–TMEM55a compensated for the deficit in TMEM55b. Hence, it is possible that the endogenous expression of TMEM55a may be also increased in shTMEM55b cells.

The uptake of both IgG-RBCs (Fig. 2B) and non-opsonized zymosan particles (Fig. 2C) was elevated in these TMEM55a-deficient cells, whereas the binding of IgG-RBCs to these cells was comparable to that in wild-type (WT) cells (Fig. 2A), indicating that the amount of the surface Fcγ receptor was not increased in the knockdown cells. Cells defective in TMEM55b, another PtdIns(4,5)P₂ 4-phosphatase, engulfed IgG-RBCs comparably to control cells (Fig. 2D). Thus, it is suggested that TMEM55a, but not TMEM55b negatively regulates phagocytosis. To further confirm the involvement of TMEM55a in phagocytosis, shTMEM55a cells were transfected with shRNA-resistant FLAG–TMEM55a. As shown in Fig. 2E,F, transfection of WT TMEM55a, but not its phosphatase dead mutant downregulated phagocytosis of IgG-RBCs both in control cells and shTMEM55a cells.

In contrast to the engulfment of the large targets, phagocytosis of a smaller target, E. coli, was not affected by the TMEM55a depletion (Fig. 2G). It has been reported that class I PI3K-mediated inactivation of Rho family GTPases is essential to complete the phagocytosis of large particles whereas it is dispensable for the phagocytosis of smaller targets (Araki et al., 1996; Cox et al., 1999; Schlam et al., 2015). In fact, as seen in Fig. 2H, the class I PI3K inhibitor wortmannin severely inhibited IgG-RBC phagocytosis but did not affect the uptake of IgG-opsonized E. coli (IgG-E. coli) (Fig. 2I). Thus, it is likely that the upregulation of the phagocytosis in the knockdown cells may be explained by an increase of phagosomal PtdIns(3,4,5)P₃.

We next investigated the phagosomal accumulation of PtdIns derivatives in TMEM55a-knockdown cells through live-cell imaging. For this purpose, EGFP-fused forms of MyoX(PH) and PLCδ(2×PH) were used as probes for PtdIns(3,4,5)P₃ and PtdIns(4,5)P₂, respectively. As presented in Fig. 3A, the control and the knockdown cells were transfected with EGFP–MyoX(PH). After the addition of IgG-RBCs, the fluorescence around the engulfed RBCs was monitored. In control cells, PtdIns(3,4,5)P₃ accumulated rapidly on the forming phagocytic cup but disappeared ∼5 min after the onset of phagocytosis (Fig. 3A,B), as previously reported (Segawa et al., 2014). In TMEM55a-deficient cells, however, the phagosomal level of PtdIns(3,4,5)P₃ was higher than that in control cells and was maintained for far longer (Fig. 3A,B). Since it has not yet been demonstrated that endogenous TMEM55a dephosphorylates PtdIns(4,5)P₂ to PtdIns(5)P, we next examined whether the phagosomal level of PtdIns(4,5)P₂ is altered in TMEM55a-knockdown cells (Fig. 3C,D). At the beginning of phagocytic cup formation, PtdIns(4)P on the plasma membrane is phosphorylated to PtdIns(4,5)P₂ on the cup (Coppolino et al., 2002). In control cells, PLCδ(PH) fluorescence on the phagosome peaked at 2 min and disappeared by ∼5 min after target incorporation (Fig. 3C). In cells lacking TMEM55a, the mean fluorescence on the phagocytic cup was higher than that on control cells. Additionally, the fluorescence surrounding phagosomes was prolonged after the completion of the incorporation in these cells (Fig. 3C). Expression of other lipid kinases and phosphatases that are involved in PtdIns(3,4,5)P₃ and PtdIns(4,5)P₂ production (PIP5K-I, class I PI3Ks and PTEN) was unchanged in shTMEM55a cells (Fig. S1). The total amount of PtdInsP₂ and PtdInsP at the steady state as determined by liquid chromatography mass spectrometry (LC/MS) system was almost comparable between the control cells and shTMEM55a cells (Fig. S2).

To further confirm TMEM55a as a phosphatase responsible for the dephosphorylation of PtdIns(4,5)P₂ to PtdIns(5)P on the phagosomes, we next tried to detect PtdIns(5)P. To date, visualization of PtdIns(5)P remains difficult but plant homodomain (PHD) finger of inhibitor of growth protein-2 (ING2) has been used as a bioprobe (Gozani et al., 2003). However, this probe sequesters PtIns(5)P, which limits the availability of the PtdIns and causes unfavorable effects on cytoskeleton organization (Viaud et al., 2014). To solve this problem, we tried a new overlay strategy: recombinant protein composed of 3×His-3×PHD was used to visualize PtdIns(5)P. A low-temperature expression system using chaperone plasmid and pCold™ vector enable creation of the recombinant protein. The construct was overlaid on the fixed cells and visualized with anti-His antibody and fluorescent secondary antibody. 3×PHD puncta surrounding zymosan particles were observed in the control cells (Fig. 4A, upper panels). It has been reported that the levels of PtdIns(5)P are profoundly decreased upon a hypo-osmotic stimulus (Sbrissa et al., 2002). To confirm 3×PHD puncta as PtdIns(5)P-containing endosomes, Raw264.7 cells were exposed to a hypo-osmotic medium after the engulfment. Surprisingly, 3×PHD was almost completely disappeared in these cells (Fig. 4A, lower panels). The validity of this probe was further supported by imaging of hydrogen peroxide-exposed cells, which have an increased PtdIns(5)P level (Jones et al., 2013), showing enormous accumulation of PtdIns(5)P in nuclei, as revealed by 3×PHD staining (Fig. 4B). Since the fluorescent intensity around zymosan was not bright enough to identify PtdIns(5)P-positive or -negative phagosomes in some cases, the profiling image of the phagosome was employed for the estimation (Fig. 4C). The phagosomes whose edges are brighter than the immediate outer areas were taken to be PtdIns(5)P-positive phagosomes (Fig. 4C). The quantitative analysis revealed that TMEM55a-deficient cells contained the less PtdIns(5)P on their phagosomes (Fig. 4D,E). When PtdIns(5)P was exogenously added to these cells, phagosomal PtdIns(5)P appeared to be increased (Fig. 4D). Hence, these results suggest that TMEM55a generates PtdIns(5)P on phagosomes.

To further estimate PtdIns(3,4,5)P₃ production, WT and TMEM55a-deficient cells were stimulated with aggregated IgG, and phosphorylation of Akt family proteins was determined by western blotting. As shown in Fig. 5, the phosphorylation of Akt was activated in the knockdown cells. The phosphorylation of the p38 MAPK family proteins appeared to be slightly but not significantly activated. The enhanced phosphorylation of Akt in TMEM55a-deficient cells could be explained by the accumulation of PtdIns(4,5)P₂, which is immediately phosphorylated to PtdIns(3,4,5)P₃ by class I PI3K in the presence of Fcγ receptor ligation, because the expression of the PtdIns(3,4,5)P₃ phosphatase PTEN and that of class I PI3Ks was comparable in control and shTMEM55a cells (Fig. S1).

Upon IgG-RBC uptake, PtdIns(4,5)P₂ on the phagocytic cup promotes F-actin polymerization (Botelho et al., 2000; Szymańska et al., 2008), which is followed by its immediate disassembly induced by PtdIns(4,5)P₂ disappearance for cup closure (Araki et al., 1996; Schlam et al., 2015). To monitor the F-actin organization, cells were transfected with EGFP–Lifeact (Riedl et al., 2008) and challenged with IgG-RBC. The F-actin accumulation on the phagocytic cup was increased in TMEM55a-knockdown cells (Fig. 6). Unexpectedly, the disassembly occurred earlier in the knockdown cells than in control cells (Fig. 6B). Although the reason why the F-actin disappeared more rapidly in TMEM55a-deficient cells remains to be clarified, it is possible that the increased production of PtdIns(3,4,5)P₃, as observed in Figs 3 and 5, activated RhoGAP proteins, and hence led to the inactivation of Rho family GTPases and termination of actin assembly (Schlam et al., 2015). The increased F-actin organization and its prompt disassembly in TMEM55a-deficient cells is advantageous for the phagocytosis of large particles.

As mentioned above, a defect in TMEM55a, but not TMEM55b enhanced phagocytosis (Fig. 2D). The discrepancy between these cells may be explained by the following hypotheses: (1) that the intracellular localization of TMEM55a and TMEM55b is different, and only TMEM55a is involved in phagocytosis; (2) that the negative regulation of phagocytosis by TMEM55a is not dependent on its phosphatase activity but can be attributed to structures that are specific to TMEM55a; or (3) that the intrinsic activity of TMEM55a is superior to that of TMEM55b and compensates for the function of TMEM55b. To clarify which of these hypotheses was responsible for the observed results, we visualized the localization of fluorescently tagged TMEM55a and TMEM55b (Fig. 7). WT Raw264.7 cells were transfected with mCherry–TMEM55a or mCherry–TMEM55b along with EGFP–EEA1(fyve) which binds to PtdIns(3)P, and were fixed with paraformaldehyde without stimulation (Fig. 7A). The cells were then stained with anti-Lamp1 antibody and Alexa-Fluor-647-conjugated secondary antibody. Both TMEM55a and TMEM55b colocalized with Lamp1 but not with PtdIns(3)P (Fig. 7A, panels 1–5 and 11–15) suggesting that both phosphatases localized to late endosomes but not to early endosomes in the resting state. The cells were then transfected with mCherry–TMEM55s along with EGFP–PLCδ(PH) and immunostained with anti-Lamp1 antibody. In the steady state, TMEM55a partially co-existed with PtdIns(4,5)P₂ (Fig. 7A, panels 6–10, left-hand cell). The cells whose plasma membrane contained little TMEM55a showed more bright PLCδ(PH) fluorescence (panels 6–9, right-hand cell), suggesting the accumulation of PtdIns(4,5)P₂. This may be explained by dephosphorylation of PtdIns(4,5)P₂ by TMEM55a. In this context, it is likely that TMEM55b could not dephosphorylate PtdIns(4,5)P₂ because TMEM55b obviously colocalized with PLCδ(PH) on the plasma membrane (Fig. 7A, 16–20). Further quantitative analysis is needed for a conclusive answer to this issue, because the transfection efficiency was different from cell to cell. However, TMEM55a more or less localized to the plasma membrane and possibly dephosphorylated PtdIns(4,5)P₂ in the steady state.

The localization of TMEM55s during phagocytosis was next examined. WT Raw264.7 cells were transfected with the various Rab GTPases tagged with EGFP along with mCherry–TMEM55a or mCherry–TMEM55b, supplemented with IgG-RBC and fixed with paraformaldehyde. TMEM55a colocalized with Rab7a, which exist on the late phagosome and phagolysosome and play an indispensable role in the fusion of phagosomes and lysosomes (Fairn and Grinstein, 2012; Vieira et al., 2003) (Fig. 7B). TMEM55a also colocalized with early endosome marker Rab5b (Horiuchi et al., 1997) on phagosome membrane, although this occurred to a lesser extent (Fig. 7B). Rab20, which is another Rab family member that is recruited to the early endosome and has been reported to be responsible for regulating phagosome acidification (Egami and Araki, 2012; Segawa et al., 2014), partly colocalized with TMEM55a (Fig. 7B). Rab11a is recruited to recycling endosomes during phagocytosis (Cox et al., 2000) and colocalized with TMEM55a (Fig. 7B). TMEM55b showed similar localization to that of TMEM55a, in that it localized substantially with the late endosome marker Rab7 and partially with the other Rab family members (Fig. 7C).

Since TMEM55s localized to plasma membrane in the quiescent state (Fig. 7A), we next tested whether the phosphatases translocated to the phagocytic cup, where its substrate PtdIns(4,5)P₂ accumulated (Fig. 7D). Thus, we transfected TMEM55 proteins along with PLCδ(PH) to confirm the colocalization of the phosphatases and PtdIns(4,5)P₂ on the cup. As shown in Fig. 7D, TMEM55b obviously colocalized with PLCδ(PH) on plasma membrane and phagocytic cup, while TMEM55a was partly observed at the inner leaflet of plasma membrane. Additionally, TMEM55a only slightly colocalized with PLCδ(PH) on the phagocytic cup. We speculate that the enzymatic activity of TMEM55a is superior to TMEM55b and TMEM55a dephosphorylates PtdIns(4,5)P₂ on the plasma membrane and phagocytic cup. This might be the reason why we detect less colocalization of TMEM55a with PLCδ(PH) than we do for TMEM55b.

Another reason why we assumed TMEM55a to be the superior enzyme was that the overexpression of TMEM55a but not TMEM55b was lethal. Similar lethality was observed when IpgD was transfected into the cells. The expression of these PtdIns(4,5)P₂ phosphatases peaked at 4–5 h after the transfection and then declined by the time of cell death. In contrast, cells transfected with TMEM55b proliferate normally and the protein expression was excellent (Fig. 1C).

To determine whether the negative regulation of phagocytosis by TMEM55a is dependent on its phosphatase activity, we next transfected IpgD, a PtdIns(4,5)P₂ phosphatase from Sigella flexineri, into Raw264.7 cells. The EGFP-tagged phosphatase was transfected to Raw264.7 cells, and the number of IgG-RBCs incorporated into the fluorescent cells was counted under a microscope (Fig. 8A). As shown in Fig. 8B, phagocytosis decreased in cells expressing IpgD. The expression of the phosphatase-dead mutants of IpgD did not affect the engulfment (Fig. 8B). Interestingly, the overexpression of PIP4K2a, which phosphorylates PtdIns(5)P to PtdIns(4,5)P₂, increased phagocytosis (Fig. 8D). It is noteworthy that the transfected EGFP–PIP4K2a localized to phagosomes (Fig. 8C). These data strongly suggest that PtdIns(4,5)P₂ 4-phosphatase inhibits but PtdIns(5)P 4-kinase enhances phagocytic activity. We next tested whether the product of PtdIns(4,5)P₂ 4-phosphatase, PtdIns(5)P, inhibited phagocytosis. The cells were exogenously supplemented with C16-PtdIns(5)P along with carrier protein before undergoing a phagocytosis assay. Unexpectedly, the PtdIns(5)P-treatment did not inhibit phagocytosis either in WT or TMEM55a-deficient cells (Fig. 8E). It is speculated that PtdIns(4,5)P₂ accumulation rather than PtdIns(5)P deficiency may be responsible for the up-regulation of phagocytosis due to TMEM55a knockdown.

In this study, we discovered a new role for TMEM55a in phagocytosis. The following is our present perspective. TMEM55a localizes mainly to late endosomes and partially to the plasma membrane in a steady state (Fig. 7A). Upon the approach of target particles, the plasma membrane extends to form phagocytic cups containing TMEM55a (Fig. 7C). PtdIns(4)P on the phagocytic cup is phosphorylated to PtdIns(4,5)P₂, which accelerates F-actin assembly. TMEM55a then dephosphorylates PtdIns(4,5)P₂ to limit actin polymerization (Fig. 6), which ultimately downregulates phagocytosis (Fig. 2). In addition, since the assembly–disassembly cycle of F-actin is required for large particle ingestion, it is also likely that a decrease in PtdIns(4,5)P₂ results in decreased PtdIns(3,4,5)P₃ production (Fig. 3), which in turn perturbs the RhoGAP-coordinated disassembly of F-actin and dampens phagocytosis in a target size-dependent manner (Fig. 2). This speculation is based on the experiments using TMEM55a-deficient cells. These cells exhibited enhanced phagocytic activity, accompanied by increased accumulation of PtdIns(4,5)P₂, PtdIns(3,4,5)P₃ and F-actin on the phagocytic cup. The finding that the overexpression of PIP4K2a, which catalyzes the reverse reaction from that of TMEM55a, also increased phagocytosis supports the hypothesis (Fig. 8C,D). Additional support is provided by the results seen upon ectopic expression of IpgD, a PtdIns(4,5)P₂ 4-phosphatase from bacteria, which inhibited the phagocytosis in a manner dependent on its enzymatic activity (Fig. 8A,B).

It is generally accepted that PtdIns(4,5)P₂ is responsible for the assembly and remodeling of F-actin during the formation of nascent phagosomes (Tuosto et al., 2015). The amount of PtdIns(4,5)P₂ on a phagocytic cup is regulated by various lipid kinases and phosphatases. Among these, PtdIns(4)P 5-kinases have been previously implicated in phagocytosis. In Raw264.7 macrophages, we detected small quantities of PtdIns(4)P 5-kinase β, as reported previously (Mao et al., 2009). Instead, the α and γ subtypes are more likely involved in phagocytosis. Bone marrow-derived macrophages from PtdIns(4)P 5-kinase γ-knockout mice have a highly polymerized actin cytoskeleton and are defective in IgG-RBC binding to the Fcγ receptor (Mao et al., 2009). In contrast, PtdIns(4)P 5-kinase α knockout has no effect on binding, but instead inhibits phagocytosis by decreasing Wiskott–Aldrich syndrome protein (WASP) and Arp2/3 activation, which, in turn, impairs de novo actin polymerization at the phagocytic cup (Mao et al., 2009). Hence, we speculate that PtdIns(4,5)P₂ produced by PtdIns(4)P 5-kinase is dephosphorylated by TMEM55a, which results in diminished F-actin assembly for large phagocytic cup formation.

The quiescent level of PIP₂, which is mostly PtdIns(4,5)P₂, shown in Fig. S2 was slightly but significantly lower in shTMEM55a cells than that in control cells. Since TMEM55a dephosphorylates PtdIns(4,5)P₂ to PtdIns(5)P, the reason for this is so far not known. However, it has been reported that a decrease in PtdIns(5)P dampens PtdIns(4,5)P₂ production (Sbrissa et al., 2002). Since most of PtdIns(4,5)P₂ is generated by PIP5K-mediated phosphorylation of PtdIns(4)P, it is intriguing to speculate that the PIP5K activity is regulated by PtdIns(5)P indirectly. A comprehensive analysis of protein–protein interaction recently revealed that PIP4K2a, PIP4K2b and PIP4K2c, which generate PtdIns(4,5)P₂ from PtdIns(5)P, associate with PIP5K (Huttlin et al., 2017). The decreased PIP₂ in the knockdown cells may be related to such an interaction.

TMEM55b is another PtdIns(4,5)P₂ 4-phosphatase with similar localization to TMEM55a (Fig. 7B). However, a reduction in TMEM55b expression did not increase phagocytosis (Fig. 2D). It is likely that the intrinsic enzymatic activity of TMEM55a is more robust than that of TMEM55b and can compensate for a loss in TMEM55b function. This concept is supported by the finding that the expression of transfected TMEM55a was increased in shTMEM55b cells compared to that in WT cells (Fig. 1C).

TMEM55a and TMEM55b were originally reported to be involved in endosome maturation (Ungewickell et al., 2005). Since both the phosphatases localize to phagosomes during phagocytosis (Fig. 7), it is reasonable that they also function in the maturation process. Further studies should investigate the function of TMEM55 proteins in phagosome acidification, the exchange of Rab GTPases, and Lamp1 or Lamp2 recruitment to phagosomes.

The sources of materials were as follows: zymosan, TexRed–zymosan, fluorescein isothiocyanate (FITC)–zymosan, E. coli-opsonizing reagent (E-2870), and RPMI 1640 medium (Life Technologies Co., Carlsbad, CA); protein assay kit (Bio-Rad, Hercules, CA); sheep red blood cells (RBCs) (Cosmo Bio Co., Tokyo, Japan); Alexa Fluor 488-conjugated anti-rabbit IgG and anti-pAkt (1:1000, #9271; Cell Signaling, Danvers, MA); anti-SRBC (1:50, A-6840, InterCell Technologies, Jupiter, FL); anti-pp38 (1:2500, #612280; BD Transduction Laboratories, Franklin Lakes, NJ); anti-Flag M2 monoclonal antibody (1:1000, #F3165; Sigma Aldrich, Japan); anti-actin antibody (1:5000, #612656; BD Transduction Laboratories); anti-Lamp 1 antibody (1:500, #121601, BioLegend, San Diego, CA); Na₂⁵¹CrO₄ (MP Biomedicals, Santa Ana, CA); Shuttle PIP™ and C16-PtdIns(5)P (Echelon Scientific, UT).

RAW264.7 cells (ATCC #2142765) lacking TMEM55a or TMEM55b were produced as follows (Kubo et al., 2005). Oligonucleotides targeting each molecule (Table S1) were cloned into a pH1 vector downstream of the H1 RNA promoter as previously described to express small interfering RNA (siRNA) hairpins. For each of the targeted sequences, a pair of oligonucleotides were synthesized (eurofins, Tokyo, Japan) with the following sequences: 5′-CCC(X)₁₉TTCAAGAGA(Y)₁₉TTTTTGGAAA-3′ and 5′-CTAGTTTCCAAAAA(Y)₁₉TCTCTTGAA(X)₁₉GGGTGCA-3′, where (X)₁₉ is the coding sequence and (Y)₁₉ is the complementary sequence. This oligonucleotide pair was annealed and ligated downstream of the H1 RNA promoter at the PstI and XbaI sites of the pH1 vector. Plasmids were transfected into RAW264.7 cells (5×10⁶–10×10⁶ cells) at 250 V/950 μF (Gene Pulser II; Bio-Rad). At 24 h after transfection, puromycin (3 μg/ml) was added to the cells, and the incubation was continued to select for resistant cells. To determine the gene silencing efficiency, total RNA was isolated with Sepasol (Nakarai Tesque, Kyoto, Japan), and the mRNA was quantified by reverse transcription PCR with the primer pairs shown in Table S2. Control cells were prepared as above with a pH1 vector containing a 400-bp stuffer sequence instead of the target sequence. For microscopic analysis, cells were seeded in tissue culture-coated glass bottom dishes (Greiner Bio-One, Frickenhausen, Germany) in RPMI 1640 medium containing 4.5 g/l glucose and 10% fetal calf serum (FCS) in a humidified 5% CO₂ atmosphere at 37°C. Immediately before starting the assay, media were aspirated and replenished with incubation buffer (complete RPMI 1640 medium without NaHCO₃, fortified with 20 mM HEPES-NaOH, pH 7.4). Activities were then determined by incubating the cells at 37°C under ambient conditions. Alternatively, cells were seeded in 24-well plates in normal RPMI 1640 medium to determine the level of Akt phosphorylation in response to aggregated IgG. The incubation was then undertaken as above in the incubation buffer.

Sheep red blood cells (RBCs) were labeled with ⁵¹Cr as previously described (Ninomiya et al., 1994). IgG-RBCs were prepared by incubating labeled cells with rabbit anti-SRBC antibody at 37°C for 10 min in 5 mM veronal buffer (pH 7.5) supplemented with 0.1% gelatin, 75 mM NaCl, 0.15 mM CaCl₂, 0.5 mM MgCl₂ and 10 mM EDTA, followed by incubation on ice for 15 min. The IgG-RBCs were washed three times with 5 mM veronal buffer (pH 7.5) supplemented with 0.1% gelatin, 75 mM NaCl, 0.15 mM CaCl₂, and 0.5 mM MgCl₂ (GVB) and finally suspended in incubation buffer (detailed above). The binding and phagocytosis of IgG-RBCs were measured as previously reported (Niomiya et al., 1994). RAW264.7 cell monolayers (2×10⁵ cells/well in a 24-well plate) were incubated with ⁵¹Cr-labeled IgG-RBCs (2×10⁷ cells) at 37°C for the indicated times. Monolayers were then washed three times with PBS to remove unbound IgG-RBCs and then briefly exposed to 0.1 ml of hypotonic PBS (diluted fivefold). The radioactivity released into the supernatant during the hypotonic shock was measured as a proxy for the amount of IgG-RBCs bound to the surface of the phagocytes. Monolayers were washed an additional three times with PBS and finally solubilized in 0.5% Triton X-100. Solution radioactivity was measured to determine the amount of engulfed IgG-RBCs.

As shown in Figs 2E and 8, cells were transfected with FLAG-tagged or EGFP-tagged constructs and seeded into tissue culture-coated glass bottom dishes 4–6 h before the phagocytosis assay. Cells were incubated with non-labeled IgG-RBCs prepared as above and kept on ice for 15 min, followed by an incubation at 37°C for 15 min. Cells were then washed three times with PBS, fixed with PBS containing 4% formaldehyde for 15 min at room temperature, permeabilized with TBS containing 0.1% saponin and 3% BSA for 60 min and finally subjected to immunostaining with Alexa Fluor 647-conjugated anti-rabbit IgG to assess the IgG-RBC engulfment. In Fig. 2E, FLAG tag was immunostained with anti-FLAG antibody (1:1000, Sigma #F3165) and Alexa Fluor 488-conjugated anti-mouse IgG (Cell Signaling) secondary antibody. The numbers of engulfed IgG-RBCs were determined by using a BZ-H2C analysis system (Keyence, Osaka, Japan).

TexasRed- or FITC-labeled zymosan particles were mixed with an equal amount of unlabeled zymosan to expedite counting, and the mixture was sonicated for ∼1 min. RAW264.7 cells (10⁵ cells/well in a glass bottom dish) were incubated at 37°C for the indicated durations with 2×10⁶ zymosan particles. Phagocytosis was stopped by the addition of ice-cold PBS. Cells were then washed three times with PBS, fixed with 4% paraformaldehyde for 15 min at room temperature and finally rinsed with PBS. Fluorescence and phase-contrast images of at least 100 macrophages from three randomly selected fields were taken using confocal microscopy. The mean numbers of ingested zymosan particles were expressed as particles per RAW264.7 cell.

mCherry was amplified by PCR using oligonucleotide primers, introducing BamHI and EcoRI sites at the 5′ and 3′ ends, respectively, with pmCherry-N1 (Clontech, Mountain View, CA) as a template. Amplification products were then cloned into the pGEX-4T-3 vector. The resulting recombinant plasmid was then used to transform BL-21 E. coli. Induction of mCherry was carried out as follows. A single colony was inoculated into 10 ml of Luria–Bertani (LB) medium and grown overnight at 37°C. Bacteria were expanded in 100 ml of LB medium and growth was allowed to continue at 37°C until the optical density at 600 nm (OD₆₀₀) reached 0.8. Protein expression was induced by the addition of isopropyl-β, d-thiogalactopyranoside at a final concentration of 0.5 mM and incubation was continued for an additional 4 h at 30°C. mCherry-expressing E. coli were then washed and fixed with 4% paraformaldehyde for 15 min at room temperature and stored at 4°C in the presence of NaN₃ until use. Labeled E. coli were opsonized with the opsonizing reagent in GVB, washed with the same buffer and sonicated for 3 min immediately before use. Raw264.7 cells in glass bottom dishes were incubated with labeled E. coli (1:50 ratio) at 0°C for 20 min, transferred to 37°C and further incubated for 20 min. The phagocytosis was stopped by washing three times with acidic medium (pH 5.5) to strip the binding E. coli. Cells were fixed in 4% paraformaldehyde for 15 min at room temperature and finally rinsed with PBS. Fluorescence images and phase-contrast images of at least 100 macrophages from three randomly selected fields were taken using confocal microscopy.

C16-PtdIns(5)P was dissolved in CHCl₃:MeOH:H₂O (1:2:0.8), and aliquots were stored at −80°C. The solvent was dried with liquid nitrogen gas. The dried PtdIns(5)P was mixed with the carrier at a 1:1 molar ratio, sonicated for 2 min and incubated 10 min at room temperature. The resulting phosphoinositide–carrier complex was then diluted in medium and added to cells (Vicinanza et al., 2015).

The tandem domain of myosin X, PH1(N)-PH2-PH1(C) (Cox et al., 2002; Lu et al., 2011), hereafter referred to as MyoX(PH), was isolated from mouse macrophages by RT-PCR with the following primer pair: 5′-AGATCTCCCTATTTCCACAGTTTTC-3′ and 5′-GAATTCCTACTTGGATCTCTGCAGCA-3′. This domain was then subcloned into pEGFP-C1 (Clontech) at the BglII and EcoRI sites. pEGFP-C1-IpgD (WT and phosphatase dead mutants), pEGFP-N1-PLCδ(PH), and TMEM55a, TMEM55b and PIP4K2a from mouse origin cloned into pcDNA3.1 were kindly provided by Dr Takehiko Sasaki (Akita University, Akita City, Japan). TMEM55a and TMEM55b were subcloned into pmcherry-C1 at the EcoRI and BamHI sites and the HindIII and BamHI sites, respectively. For the rescue experiments, silent mutations were introduced into the target sequence of TMEM55a constructs using primer pair shown in Table S3. PIP4K2a was subcloned into pEGFP-C1 at the KpnI and XbaI sites. The EGFP-tagged Rab5, Rab7, Rab11 and Rab20 constructs were kind gifts from Dr Mitsunori Fukuda (Tohoku University, Sendai, Japan) (Fukuda et al., 2008). EGFP-Lifeact was prepared as previously described (Riedl et al., 2008).

Plasmids were transfected using the Neon™ transfection system (Invitrogen). Briefly, Raw264.7 cells were suspended in 25 µl of T buffer at 5×10⁷ cells/ml and mixed with 2 µg of plasmid DNA. Cells were then subjected to electroporation using a single 20-ms pulse of 1800 V with a 10 µl tip two times. Electroporated cells were immediately transferred to culture-coated glass bottom dishes (∼2 cm²) in RPMI 1640 medium containing 20% FCS without antibiotics. At 5–24 h (the incubation length varied depending on the plasmid) after transfection, cells were subjected to microscopic analysis.

Cells were incubated with IgG-RBCs and placed on a BIOREVO BZ9000 microscope (Keyence, Osaka, Japan) equipped with a CFI Plan Apo VC60×H oil immersion lens, and phagocytosis was allowed to proceed at 37°C (Segawa et al., 2017). Unless otherwise stated, the fluorescent images were collected every 1 min, and the intensity of the phagosome-associated fluorescence was analyzed by using a BZ-II analysis system (Keyence, Osaka, Japan).

pEGFP-C1-3×PHD was kindly donated by Dr Bernard Payrastre (Université Paul Sabatier, Toulouse, France) (Gozani et al., 2003). 3×PHD was amplified by PCR using oligonucleotide primers with the sequences 5′-GGTACCCGGACTCAGATCTCGGCA-3′ and 5′-GGATCCCCGTCGACTCATCACTAC-3′. The amplification product was then cloned into pCold II vector (Takara) at the KpnI and BamHI sites. The resulting recombinant plasmid was used to transform BL-21 previously transformed with chaperone plasmid (pG-KJE8, Takara) and cultured overnight on LB plates fortified with chloramphenicol (50 µg/ml) and ampicillin (50 µg/ml). Induction of 3×His-3×PHD was carried out as follows. A single colony was inoculated into 10 ml of LB medium with chloramphenicol and ampicillin and grown overnight at 37°C. Bacteria were expanded in 100 ml of LB medium containing chloramphenicol, ampicillin, arabinose (0.5 mg/ml) and tetracycline (5 ng/ml). Growth was allowed to continue at 37°C until the OD₆₀₀ reached 0.5–0.7, at which point the culture was transferred to 15°C for an additional 30 min. Protein expression was induced by the addition of isopropyl-β, D-thiogalactopyranoside at a final concentration of 0.5 mM and incubation was continued for a further 24 h at 15°C. The purification of 3×His-3×PHD was performed using TALON metal affinity resin (Clontech) according to the manufacturer's protocol. Briefly, the cells were collected by centrifugation (5000 g for 5 min), resuspended in 4 ml of lysis buffer (50 mM NaH₂PO₄·H₂O, 300 mM NaCl, 10 mM imidazole, 0.5% Triton X-100 and 1 mM PMSF, pH 7.4) and sonicated for 3 min. After centrifugation (15,000 g for 30 min), the supernatant was mixed with a 2 ml bed of TALON resin pre-washed with lysis buffer, and then incubated at 4°C for 30 min with rotation. After three washes with lysis buffer, the bound proteins were eluted with lysis buffer supplemented with 160 mM imidazole. The eluate was then passed through a PD-10 column (GE Healthcare) to remove imidazole. The prepared protein was identified as single band by Coomassie Blue staining after SDS-PAGE. The purified protein was identified as a single band by Coomassie Blue staining after SDS-PAGE and by western blotting with anti-His antibody (Amersham Biosciences).

After phagocytosis of FITC-zymosan, cells were washed thrice with acidic medium (pH 5.5) to strip binding zymosan. Cells were fixed with 4% paraformaldehyde for 15 min at room temperature, rinsed with PBS and permeabilized with PBS containing 0.1% saponin and 3% BSA for 30 min. After incubation with 3×His-3×PHD described above at 4°C overnight, the cells were washed three times with 0.1% saponin in PBS, incubated with anti-His antibody (1:1000, #27-4710-01, GE Healthcare) for 4 h and finally stained with Alexa Fluor 647-labeled secondary antibody (1:1000 dilution) for 2 h at room temperature. Microscopic analysis was performed by using a Keyence BZ-9000 fluorescence microscope with CFI Plan Apo VC60×H lens (Keyence, Osaka, Japan). Fluorescence images of at least 100 macrophages from three randomly selected fields were counted to analyze the PtdIns(5)P-positive phagosomes.

Cells in the 24-well plates were incubated with 30 µg/ml aggregated mouse IgG, which was prepared by heating at 62°C for 30 min. Cells were then washed with PBS and lysed in 50 µl of lysis buffer containing 25 mM Tris-HCl pH 7.4, 0.5% Nonidet P-40, 150 mM NaCl, 1 mM sodium orthovanadate (Na₃VO₄), 1 mM EDTA, 20 mM sodium fluoride, 1 mM phenylmethylsulfonyl fluoride, protease inhibitor cocktail (Sigma-Aldrich) and 20 µM p-amidinophenylmethylsulfonyl fluoride. Cell lysates were centrifuged at 20,000 g for 10 min. Supernatants were collected, and the protein concentration was determined by using a Bio-Rad assay kit. Total cell lysates (100 µg protein) were mixed with 10 µl of 5× sample buffer (62.5 mM Tris-HCl pH 6.8, 1% SDS, 10% glycerol, 5% 2-mercaptoethanol, and 0.02% Bromophenol Blue) and heated at 100°C for 5 min. Proteins were separated by SDS-PAGE and transferred electrophoretically onto a polyvinylidene difluoride (PVDF) membrane (Millipore). This membrane was blocked with 5% skimmed milk and incubated with the appropriate antibodies. Antibody binding was detected by using a chemiluminescent substrate (Perkin-Elmer).

Total RNA prepared from the indicated cells was subjected to quantitative RT-PCR analysis. In brief, cDNA were amplified with FastStart Universal SYBER Green Master and primer sequences provided in Table S2 by ABI HT-7900. Data analysis was performed using the ΔΔCT methods as described in the Applied Biosystem protocol. A ΔCT value was determined for each sample using the CT value from input DNA. The CT value for β-actin was used to normalize RT-PCR loading. ΔΔCT values were then obtained by subtracting control ΔCT values (vector control) from sample ΔCT values from the knockdown cells. The ΔΔCT values were converted to fold difference compared with the control by raising 2 to the ΔΔCT power.

Two-tailed Student's t-tests were performed. All P-values were considered significant at P<0.05.

We thank Dr Takehiko Sasaki (Akita University) for pEGFP-C1-IpgD (wild type and phosphatase dead mutants), pEGFP-N1-PLCδ(PH), TMEM55a, TMEM55b and PIP4K2a, Dr Mitsunori Fukuda (Tohoku University) for EGFP-Rab5, Rab7, Rab11 and Rab20, and Dr Bernard Payrastre (Université Paul Sabatier) for EGFP-3×PHD.