Vacuolins and myosin VII are required for phagocytic uptake and phagosomal membrane recycling in Dictyostelium discoideum

Phagocytosis is the process used by animal phagocytes of the innate immune system to ingest and kill bacterial pathogens, which allows them to derive antigens necessary to trigger adaptive immune responses. In addition, phagocytosis is used to control tissue homeostasis through clearance of apoptotic, senescent or cancerous cells (Flannagan et al., 2012). The social amoeba, Dictyostelium discoideum, which uses this process primarily for nutrition purposes, has been widely used to study the evolutionary conservation of phagocytosis, owing to its experimental versatility. Furthermore, the generic endocytic and phagosomal maturation pathways are highly conserved between D. discoideum and animals, making it an excellent model to study these mechanisms (Dunn et al., 2018).

Particles are first bound and recognized by surface adhesion molecules and receptors, triggering signaling cascades that lead to polymerization of actin, deformation of the plasma membrane in a phagocytic cup and engulfment (Bozzaro et al., 2008; Dunn et al., 2018). After closure of the cup, the particle is enclosed in a phagosome, which matures by a series of fission and fusion events with other compartments of the endocytic pathway. To allow killing and digestion of bacteria, the compartment becomes acidic, proteolytic, oxidative and accumulates toxic levels of certain metals (Cosson and Lima, 2014; Dunn et al., 2018). In D. discoideum, lysosomes eventually mature into neutral and non-degradative compartments, called postlysosomes (PLs), by retrieval of the v-ATPase and, later, of the lysosomal hydrolases. This process is regulated by the highly conserved Wiskott–Aldrich syndrome protein and SCAR homolog (WASH) complex, an actin nucleation promoting factor that interacts with the retromer complex in both mammals and D. discoideum (Derivery et al., 2009; Gomez and Billadeau, 2009; Carnell et al., 2011; King et al., 2013). Interestingly, these recycling steps are reminiscent of the mammalian retromer-mediated retrograde trafficking from the endocytic compartments to the trans-Golgi network (Burd and Cullen, 2014). Eventually, PLs undergo exocytosis in a Ca²⁺-dependent manner (Lima et al., 2012). In addition to phagocytosis, D. discoideum uses a non-specific bulk internalization pathway termed macropinocytosis, which allows uptake of fluid inside a macropinosome that follows a maturation pathway similar to phagosomes (Buckley and King, 2017).

In D. discoideum, fast recycling of plasma membrane proteins from macropinosomes and phagosomes has recently been shown to require the WASH and the retromer complexes, as is the case in mammals (Seaman et al., 2013; Buckley et al., 2016; McNally and Cullen, 2018). These complexes, probably through the actin polymerization activity of WASH, segregate a specific set of cargoes and regulate their recycling to the plasma membrane. In addition, in mammals, other plasma membrane proteins can cycle through the slow recycling pathway, via a specific tubular juxtanuclear compartment termed the endocytic recycling compartment (ERC; Maxfield and McGraw, 2004), which harbors the GTPase RAB11. Myosins are actin-binding motors that play crucial roles in different steps of phagocytosis, including plasma membrane recycling, but also formation of lamellipodia, closure of the cup and trafficking of vesicles to the phagosome, in both mammals and D. discoideum (Araki, 2006; Dieckmann et al., 2010; Gopaldass et al., 2012). Interestingly, the class I myosin MyoB was also shown to participate in plasma membrane recycling, although at a later stage than the WASH complex (Neuhaus and Soldati, 2000). This suggests that two distinct steps of recycling, a fast and a slower one, are also found in D. discoideum.

Flotillin-1 and flotillin-2 are lipid raft proteins conserved throughout Metazoan evolution (Morrow and Parton, 2005; Otto and Nichols, 2011). They are composed of a prohibitin homology domain (PHB domain) at the N-terminus, responsible for their insertion into the cytosolic leaflet of membranes (Morrow et al., 2002). This domain is characterized by hydrophobic stretches and acylation sites, which are required for membrane binding (Morrow et al., 2002; Neumann-Giesen et al., 2004). At the C-terminus lies the so-called flotillin domain, which contains coiled-coil regions, necessary for protein-protein interactions and hetero- and homo-tetramerization (Solis et al., 2007). Flotillins are found in specific detergent-resistant lipid microdomains at the plasma membrane and at endosomal and phagocytic compartments (Dermine et al., 2001; Neumann-Giesen et al., 2004; Liu et al., 2005). At the plasma membrane, flotillins are thought to function as signaling platforms involved, for example, in axon regeneration and glucose uptake (Baumann et al., 2000; Stuermer, 2011). In endocytic compartments, flotillins have been shown to participate in trafficking of several cargoes across multiple cell types (Meister and Tikkanen, 2014). Recently, flotillin-1 was shown to bind to the endosomal sorting complex required for transport (ESCRT)-0 and -I and proposed to be involved in transferring cargoes destined for degradation from ESCRT-0 to ESCRT-I (Meister et al., 2017). On the other hand, the surface transferrin receptor (TfR) and E-cadherin (CDH1) require flotillins to be recycled to the plasma membrane through the ERC in a RAB11A- and SNX4-dependent manner (Solis et al., 2013). Moreover, flotillins interact directly with RAB11A at the ERC in epithelial cells as well as neurons, and participate in T-cell receptor sorting to the ERC in T cells, suggesting that their function in plasma membrane recycling might be conserved in multiple cell types (Bodrikov et al., 2017; Redpath et al., 2019).

D. discoideum expresses three flotillin-like proteins, called vacuolin A, B and C. These proteins are highly similar to each other and share a structure reminiscent of the flotillins, with a PHB-like domain and coiled-coil regions (Wienke et al., 2006). The coiled-coil regions at the C-terminus mediate oligomerization, and it was shown that vacuolin A (VacA) and vacuolin B (VacB) can oligomerize with themselves as well as with each other (Wienke et al., 2006). The PHB domain has also been implicated in membrane targeting, although how vacuolins are associated with membranes has not been investigated yet. Vacuolins bind membranes in a patchy distribution, thereby defining specific microdomains, and it was hypothesized that they might cycle between a cytosolic pool to a membrane-bound one (Rauchenberger et al., 1997). VacA and VacB have been previously described as PL markers in D. discoideum AX2-214 background (Rauchenberger et al., 1997; Jenne et al., 1998). However vacuolin C (VacC), which was only recently identified after the sequencing and annotation of the genome (Eichinger et al., 2005), has not been studied so far. vacA and vacB knockout (KO) mutants were generated in the AX2-214 background, revealing that VacB is involved in PL biogenesis. In fact, in the absence of VacB, cells exhibit a delayed lysosomal reneutralization phase, a longer transit time of fluid phase and delayed exocytosis (Jenne et al., 1998). In that study, no phenotype was observed for vacA^- cells, suggesting a non-redundant role for both vacuolins (Jenne et al., 1998). In addition, VacB was suggested to be a negative regulator of PL fusion, as PLs in vacB^- cells were enlarged (Jenne et al., 1998). Together, these results indicated a role for VacB in PL biogenesis.

We sought to characterize the newly identified vacuolin, VacC, and, in the light of their potential homo- and hetero-merization, revisit the localization, biochemical behavior and role of all three vacuolins. For this purpose, we have generated overexpressing and chromosomally tagged GFP-fusions of each vacuolin, as well as single and multiple KOs. In this study, we bring further evidence that vacuolins are the functional homologs of flotillins in D. discoideum, and show that they are crucial for phagocytosis, plasma membrane recycling and, to a lesser extent, for phagosomal maturation.

To characterize the localization of VacC compared to the other two known vacuolins, we inserted by knock-in (KI) the coding sequence of GFP in-frame at the 3′ end of each vacuolin open reading frame in the genome of wild-type (WT) cells (Vac–GFP KI). Cells were immunostained for GFP, the A subunit of the v-ATPase, VatA, which resides in lysosomes and the contractile vacuole (CV), or p80, a predicted copper transporter that accumulates in PLs (Fig. 1A). As previously described (Rauchenberger et al., 1997; Jenne et al., 1998), VacA and VacB were present at PL compartments in a patchy distribution, suggesting that they are present in specific microdomains. Similarly, VacC colocalized with p80 at PLs and was found only sporadically at lysosomes with VatA (Fig. 1A,B). Localization of endogenous VacA and VacB with newly-produced specific recombinant nanobodies (Fig. S1) produced similar results (Fig. 1C).

Surprisingly, although overexpressed (OE) VacA– and VacB–GFP showed a similar localization to that observed in the KI cells, overexpressed VacC–GFP exhibited a more heterogenous localization, mostly colocalizing with VatA-positive lysosomal vesicles (85%) and only partially (15%) with p80 compartments (Fig. S2A,B). Further characterization of Vac–GFP OE strains revealed that cells overexpressing vacuolins, in contrast to the KIs, had a higher number of p80-positive compartments compared to WT cells (Fig. S2C). In addition, the number of GFP-positive, i.e. vacuolin-positive, vesicles was also increased in overexpressing cell lines with an average of over ten vesicles per cell, compared to three to four vesicles in KIs (Fig. S2D). Similar to GFP-staining in the KIs, the anti-VacA or anti-VacB nanobodies stained three to four vesicles per cell (Fig. S2D). These results are in agreement with the number of vesicles revealed by the 221-1-1 anti-vacuolin antibody (renamed here pan-vacuolin, see Materials and Methods; Rauchenberger et al., 1997). We conclude that overexpression of vacuolins might be detrimental to the cell and therefore not suitable to study the physiological localization and functions of vacuolins.

To study the recruitment dynamics of all three vacuolins to phagosomes, Vac–GFP KI cell lines were monitored using live-cell microscopy during phagocytosis (Fig. 1D; Fig. S3A). Interestingly, all three vacuolins were present on phagosomes as early as 2–4 min after closure. In fact, phagosomes presented clear, although very faint, vacuolin-positive patches as early as 2 min after closure. In addition, phagosomes were surrounded by highly dynamic and vacuolin-rich vesicles, which are better appreciated in live-cell microscopy movies (Movies 1,2). This result was in agreement with previously published phagosome proteome data, which showed that vacuolins were present on phagosomes throughout the maturation process (Gotthardt et al., 2002, 2006). Moreover, we observed vacuolin patches at the plasma membrane (Fig. S3B) that resulted from exocytic fusion of bead-containing phagosomes (BCP), as previously hypothesized (Jenne et al., 1998). Taken together, these results show that all three vacuolins, which are highly similar (Fig. S3C), are recruited very early to phagosomes and steadily accumulate on PLs until exocytosis.

Flotillins are well established markers of lipid rafts at the plasma membrane and phagosomes, and are tightly associated with membranes and partially resistant to alkaline extraction (Bickel et al., 1997; Dermine et al., 2001; Morrow et al., 2002). Given the similarity in the protein domains of flotillins with vacuolins (Fig. S3C), we investigated whether vacuolins biochemically behave like their putative mammalian homologs. To test the membrane association of vacuolins, Vac–GFP KIs were homogenized and ultracentrifuged to separate the cytosolic (Cyto) from the membrane-associated (MB) fraction, followed by alkaline extraction and western blotting (Fig. 1E). The integral membrane protein mitochondrial porin (PorA) was exclusively found in the MB fraction. Both the endogenous vacuolins and Vac–GFP KIs were also found exclusively associated with MBs. In addition, this association was quite strong, as alkaline pH only partially extracted vacuolins from MBs. To better characterize this interaction, additional extraction methods were applied on cells overexpressing VacB–GFP (Fig. 1F). Treatment with urea was not sufficient to extract vacuolins from the MB fractions, but both high salt concentration and alkaline pH partially extracted vacuolins, but not mitochondrial porin. This indicates that, like mammalian flotillins, vacuolins are exclusively and tightly associated with membranes.

In order to better characterize the role of vacuolins, we first examined the previously published gene disruption mutants vacA^- and vacB^- in the AX2-214 background (Jenne et al., 1998). However, we realized that the vacB^- mutant was not a single KO. To our surprise, although in vacA^- only the central portion of vacA was not amplified by PCR performed on genomic DNA, in the vacB^- mutant the central portions of both vacB and vacC were missing (Fig. S4A). RNA sequencing of both vacA^- and vacB^- mutants confirmed these results and revealed that, apart from vacB and vacC, no mRNA was detected over the neighboring slob2 gene in vacB^- mutants (Fig. S4B). For these reasons, we decided to generate new isogenic vacuolin KO mutants, in the AX2(Ka) background (see Materials and Methods). Single KO mutants of each vacuolin, as well as a double vacB/vacC KO and a triple vacuolin KO mutant were generated, which will be named hereafter ΔA, ΔB, ΔC, ΔBC and ΔABC, respectively. These were tested by PCR on genomic DNA, reverse transcription qPCR (RT-qPCR), western blotting and RNAseq (Fig. S4C,D) to ensure that only the intended gene(s) had been disrupted. We therefore used these mutants to completely revisit the role of vacuolins.

Given that vacuolins are recruited early and are present throughout phagosomal maturation, we investigated their role in uptake and phagolysosomal biogenesis. We first tested whether absence of one, two or three vacuolins impaired phagocytic uptake. Cells were incubated in the presence of various-sized fluorescent beads, GFP-expressing Gram-negative Klebsiella pneumoniae (KpGe) and pathogenic Mycobacterium marinum (Fig. 2). Uptake was measured by following the accumulation of fluorescence over time by flow cytometry. All vacuolin KO mutants were severely impaired in uptake of every particle tested. Both the rate and extent of phagocytosis were reduced, with as much as a 75–90% reduction in internalized particles. This defect was not dependent on the size (Fig. 2A) nor on the nature of the particle (Fig. 2B,C). Interestingly, vacuolin KO mutants were slightly less impaired in uptake of K. pneumoniae, with about 75% reduction (Fig. 2B), compared to 95% reduction of internalized beads or M. marinum (Fig. 2A,C).

Because overexpression of vacuolins in WT cells led to an increased number of PL compartments and a different localization for VacC (Fig. S2), we tested whether overexpression also affected uptake. Vac–GFP OE cells had a 30% reduction in uptake of 1 μm beads (Fig. 2D). In contrast, expression of chromosomally GFP-tagged vacuolins did not significantly affect uptake (Fig. 2E). We conclude that both absence and overexpression of vacuolins, but not merely the fusion with GFP, severely impairs uptake of various particles, suggesting that both the level and stoichiometry of vacuolins are important for their function.

Vacuolins, unlike flotillins, are present only transiently and at low levels at the plasma membrane upon PL exocytosis, suggesting that their role in phagocytic uptake is indirect. Inability to take up particles could be explained by impaired motility towards the particle or defects in recognition or adhesion. In addition, actin-dependent rearrangements responsible for the formation and closure of the phagocytic cup, or signaling to initiate uptake might be affected.

To test whether formation of the phagocytic cup and its closure were impaired, WT and ΔABC cells were transfected with the phosphatidylinositol (3,4,5)-trisphosphate [PI(3,4,5)P₃] and PI(3,4)P₂ reporter PH_CRAC–GFP that is enriched at phagocytic cups and macropinosomes (Parent et al., 1998; Buckley et al., 2016). To observe PH_CRAC–GFP dynamics upon phagocytosis, cells were imaged by time-lapse microscopy in the presence of TRITC-labeled yeasts (Fig. 3A). We first observed that the morphology of the forming phagosome around the particle was often different in ΔABC cells compared with WT. Notably, in WT cells, the phagosomal membrane was tightly apposed to the particle, whereas in ΔABC cells, particles were more frequently taken up in structures that resembled widely open macropinocytic cups (Fig. 3A). Accumulation of PI(3,4,5)P₃ at the site of the forming phagosome is required for efficient closure of the phagocytic cup in mammalian cells (Dewitt et al., 2006). To determine whether phagosomal closure was affected in ΔABC cells, we measured the time from first appearance of the PH_CRAC–GFP patch until its disappearance, as an indication of efficient closure (Fig. 3B). No significant difference was observed between WT and ΔABC cells, suggesting that PIP-dependent signaling and phagosomal closure are not affected by vacuolin absence.

When imaged by time-lapse microscopy, vacuolin KO mutants were able to take up particles, although less frequently than WT cells and often after prolonged contact. We therefore wondered whether absence of vacuolins perturbed particle recognition. To test this, the plasma membrane of WT and ΔABC cells was labeled with the lipophilic dye, FM4-64, to follow formation of the phagosome after uptake of fluorescent beads. The time from the first contact with the particle until complete internalization was measured, as an estimation of the time it took to recognize and begin uptake of the particle (Fig. 3C; Movies 3–5). WT cells took on average 2–5 min to engulf the particle completely after first contact (Fig. 3D). On the other hand, ΔABC displayed a more heterogenous phenotype, with some cells taking as short a time as WT to engulf particles, and others as long as 30–40 min. This suggested that ΔABC cells were less able to recognize the particles and to trigger engulfment.

To determine whether vacuolin KO cells are less motile, and thus less capable of moving towards particles, we assayed their random motility and velocity in medium (Fig. 3E,F). Surprisingly, while ΔABC cells moved slightly, though significantly, faster than WT cells, the persistence of their movement was reduced. In addition, we tested whether vacuolin KO cells were still able to adhere to surfaces. We measured the area of adhesion of vacuolin KO cells on glass by reflection interference contrast microscopy (RICM; Fig. 3G), and observed that ΔABC cells were, on average, less attached to their substrate than WT cells. These results suggested that in absence of vacuolins, cells may be less adherent to surfaces. In conclusion, absence of vacuolins might indirectly impair initial recognition and/or adhesion to the particle and initiation of phagosome formation.

Flotillins have been recently shown to participate in the recycling of the plasma membrane proteins TfR, E-cadherin and TCR (Solis et al., 2013; Bodrikov et al., 2017; Redpath et al., 2019). Given the similarity between mammalian flotillins and vacuolins, we wondered whether they might also be involved in plasma membrane recycling. A recycling defect could affect the presence of adhesion and receptor molecules at the cell surface, and thus uptake. To verify this hypothesis, we biotinylated the cell surface of WT and ΔABC cells and purified phagosomes at different maturation stages, to assay the dynamics of plasma membrane protein retrieval from phagosomes (Fig. 4A,B). Latex bead-containing phagosomes were isolated by flotation on sucrose gradients and analyzed by western blot. Streptavidin–HRP was used to reveal all surface proteins present in the phagosomes. The presence of the β-integrin-like protein SibA and glycoprotein LmpB, two known plasma membrane proteins (Janssen et al., 2001; Cornillon et al., 2006; Sattler et al., 2018), was also assayed. Importantly, latex beads were first adsorbed on cells in the cold before induction of phagocytosis by warming up, ensuring synchronous phagosome maturation between WT and ΔABC cells, despite their phagocytic defect. Between 5 and 15 min after uptake, surface proteins were present in phagosomes of both cell lines as a consequence of membrane internalization during phagocytosis (Fig. 4A). In WT cells, biotinylated proteins, as well as SibA and LmpB, were rapidly retrieved from phagosomes, with a decrease reaching ∼50% after 30 min (Fig. 4A,B). On the other hand, in ΔABC cells, plasma membrane proteins lingered in phagosomes as long as 180 min, suggesting a defect in retrieval of these proteins.

To determine whether specific plasma membrane proteins were mistrafficked in the absence of vacuolins, we used biotinylation, streptavidin pulldowns and mass spectrometry to identify the surface proteome of WT and ΔABC cells (Fig. 4C,D; Tables S3, S4). Manual curation was used to exclude nuclear, mitochondrial or ribosomal contaminants, revealing 336 plasma membrane and cortical proteins (Fig. 4C; Table S4). Of these, 130 proteins contained transmembrane domains, or were known to be glycosylated or lipidated. In addition, 65 proteins mapped to the gene ontology (GO) cellular component term ‘plasma membrane’ (Fig. 4D). Moreover, proteins mapping to GO terms including membrane proteins, cytoskeleton or endocytic maturation were highly enriched in this surface proteome, indicating that our otherwise stringent procedure collected not only proteins inserted into the plasma membrane, but also co-purified peripheral cortical proteins strongly associated with membrane proteins. Twelve proteins were significantly under-represented in the surface proteome of ΔABC cells (Fig. 4C; Table S3). Of these, notably, three myosins were identified: MyoA, Myosin VII (also known as MyoI and referred to hereon as MyoVII) and MyoK. In addition, several uncharacterized proteins with transmembrane or EGF-like domains were less abundant at the surface of ΔABC cells. Of note, no known phagocytic or adhesion receptor (such as SibA, SibC, SibE or LmpB) was significantly less abundant at the surface of ΔABC cells at steady state level (Table S4). These results led us to conclude that D. discoideum vacuolins, like flotillins, are involved directly or indirectly in recycling of plasma membrane components from endosomes. However, in contrast to flotillins, vacuolins do not seem to participate in the recycling of a specific set of receptors, but rather in maintaining the composition of the plasma membrane and its underlying cortex, potentially by affecting clustering of surface proteins and their retrieval.

Because vacuolins accumulate at PLs, we wondered whether their depletion might also affect phagosomal maturation and PL biogenesis. Acidification and proteolysis are two hallmarks of phagosome maturation; we tested whether these functions were impaired in vacuolin KO mutants. WT and ΔABC cells were incubated with silica beads coupled to a pH-sensitive fluorophore (FITC) to measure phagosomal acidification over time (Sattler et al., 2013). To ensure that only efficiently phagocytosed beads were considered, cells were tracked by time-lapse microscopy and the normalized fluorescence of individual intracellular beads was measured (Fig. 5A). Upon uptake, the phagosomes of both WT and ΔABC cells rapidly acidified. In WT cells, reneutralization started at around 45 min, and phagosomes reached a neutral pH about 90 min post-uptake (Fig. 5A). Conversely, in ΔABC cells the reneutralization phase started earlier and a neutral pH was reached as rapidly as 40 min after uptake. To monitor phagosomal proteolysis a similar assay was used, using silica beads coupled to the self-quenched DQGreen–BSA. Proteolytic cleavage of BSA releases DQGreen into the lumen, thus increasing its fluorescence over time (Sattler et al., 2013). In WT and ΔABC cells, proteolysis started immediately after bead uptake, reaching a plateau after about 10 min (Fig. 5B). Interestingly, we observed a slight, although not striking, earlier onset of proteolysis in ΔABC cells.

To better investigate a possible role of vacuolins in lysosome biogenesis, we isolated latex bead-containing phagosomes from WT and ΔABC cells at different maturation stages, and analyzed the temporal profiles of maturation markers by western blotting (Fig. 5C,D). Actin, Abp1 (also known as AbpE) and the transmembrane v-ATPase subunit VatM had similar recruitment dynamics in WT and mutant cells (Fig. 5C; Fig. S5A). In D. discoideum, two different classes of lysosomal enzymes are distinguished by their sugar modifications (Freeze et al., 1990; Souza et al., 1997). Enzymes bearing the N-acetylglucosamine-1-phosphate (GlcNac1P) modification are first delivered to the phagosome, followed later on by those carrying the mannose-6-sulfate (Man-6-S, common antigen-1) modification (Souza et al., 1997). Cathepsin D, among others, belongs to the latter group (Journet et al., 1999). We observed that in ΔABC cells, Man-6-S-modified enzymes, including cathepsin D, appeared earlier than in WT cells, with a peak reached at 30–60 min instead of 120 min (Fig. 5C,D). Moreover, certain GlcNac1P enzymes were slightly under-represented in ΔABC phagosomes. LmpA, a PL glycoprotein involved in lysosomal biogenesis (Sattler et al., 2018), was also present earlier in phagosomes of cells lacking vacuolins (Fig. 5C,D). These results suggest that absence of vacuolins might impair delivery and/or retrieval of lysosomal proteins and thus impact lysosomal functions.

To test whether the differences in recruitment dynamics of lysosomal proteins in vacuolin KO mutants could affect digestion and killing of bacteria, we monitored growth of WT and vacuolin mutant cells on different Gram-negative and Gram-positive bacteria. For this, cells were plated in increasing dilutions on bacterial lawns, and the formation of phagocytic plaques was monitored over time (Fig. 5E; Fig. S5B). The large excess of bacteria in the lawn bypasses the uptake defects of vacuolin KO mutants. To our surprise, vacuolin KO mutants were able to grow (implying killing and digestion) on every bacterial species tested as efficiently as WT cells. To measure intracellular bacterial killing more precisely, we monitored, by time-lapse microscopy, killing of GFP-expressing K. pneumoniae, as previously described (Leiba et al., 2017). Disappearance of the GFP signal was used as a proxy for killing (Fig. 5F). On average, K. pneumoniae was killed in 3.75 min in both WT cells and ΔABC KO cells. To conclude, these results indicate that vacuolin KO mutants might have a shorter phagosomal acidic phase, as well as faster delivery or retrieval of certain lysosomal enzymes, although with no apparent impact on bacterial killing and digestion.

As a way to determine the mechanism by which vacuolins may affect plasma membrane recycling and lysosome biogenesis, we used a transcriptomic approach. The RNA of single, double and triple vacuolin KO mutants was collected, sequenced and compared to RNA from WT cells (Fig. 6A). We found that absence of one or several vacuolins affected transcription of myoVII. Strikingly, MyoVII was also completely absent from the surface proteome of ΔABC cells (Fig. 4C). Interestingly, KO of myoVII induces a severe phagocytosis defect similar to the one described here (Titus, 1999). In addition, we found transcription of talA was also affected, probably as a consequence of myoVII downregulation (Fig. 6A). Indeed, Talin A is an actin-binding protein known to interact with MyoVII and to be downregulated upon MyoVII depletion (Niewöhner et al., 1997; Gebbie et al., 2004; Tuxworth et al., 2005). Interestingly, none of these genes were downregulated in the previously described vacB^- cells, which did not exhibit the phagocytosis phenotypes observed in the present study (Fig. 6A). Notably, protein levels of MyoVII and TalA were as reduced in vacuolin KO cells as in a ΔMyoVII cell line (Fig. 6B,C). In addition, levels of endogenous TalA were clearly reduced in vacuolin mutants, as visualized by immunofluorescence, and its localization at filopodia was lost (Fig. S6A, left panel). Interestingly, when vacuolins were overexpressed, a similar reduction of MyoVII and TalA levels was observed (Fig. 6B,C). In contrast, vacuolin expression was not affected by the absence of MyoVII (Fig. 6B). We hypothesized that vacuolins might affect localization and/or stability of MyoVII and its partner TalA, and even consequently affect not only their protein but also mRNA levels. To test this hypothesis, GFP–MyoVII was ectopically and constitutively expressed in ΔABC cells. The overexpressed MyoVII localized, as in WT cells, at regions of cell-cell contacts, filopodia, and at macropinosomes and phagosomes (Fig. S6B), indicating that vacuolins are not absolutely required for the correct localization of MyoVII. In addition, endogenous TalA was stabilized by the ectopic expression of MyoVII and was enriched at filopodia with GFP–MyoVII in rescued cells, similar to its localization in WT cells, whether or not vacuolins were present (Fig. S6A, right panel). Moreover, the level of endogenous TalA in ΔABC cells correlated with the levels of GFP–MyoVII expression (Fig. S6A). We conclude that the stability of the MyoVII–TalA complex is regulated in an intricate manner both by the overall level of vacuolins and the correct balance of isoforms. In addition, forced expression of MyoVII in ΔABC cells is fully sufficient to restore endogenous levels of its partner TalA and the correct localization of the complex.

To test whether absence of MyoVII and TalA was in fact responsible for the phenotypes we observed in vacuolin KO cells, different functions were assayed in ΔABC cells expressing GFP–MyoVII (Fig. 6D–H). Reintroduction of MyoVII completely rescued phagocytic uptake by vacuolin KO cells as well as their adhesion to glass surfaces (Fig. 6D,E). Interestingly, although overexpression of MyoVII did not affect the slightly increased speed of ΔABC cells (Fig. 6F), it was able to rescue the persistence of these cells (Fig. 6G). Finally, phagosome maturation in ΔABC+GFP–MyoVII cells was assayed (Fig. 6H,I). Whereas the early recycling of surface proteins was rescued by overexpression of MyoVII (streptavidin, SibA and LmpB), the recruitment dynamics of proteins involved in phagolysosomal biogenesis was only partially impacted, in that the phenotypes observed were intermediate between those of WT and ΔABC cells (Fig. 6H,I). MyoVII has been proposed, together with TalA, to form a complex at the plasma membrane and, with actin, to participate in localization of adhesion proteins at the cell surface and their clustering during uptake (Tuxworth et al., 2001). To test whether actin localization at the phagocytic cup was impaired in absence of vacuolins, and consequently of MyoVII, we performed immunofluorescence of cells during uptake of beads (Fig. S6C). While actin was present all around the nascent phagocytic cup in WT cells, in ΔABC cells, actin was mostly observed at the edges of the cup, but not at the base. Overexpression of MyoVII in these cells partially restored the presence of actin at the base of the cup (Fig. S6C), suggesting that the function of MyoVII in clustering receptors at the cup might be impaired in absence of vacuolins.

Taken together, these results show that reduced MyoVII expression in ΔABC cells is responsible for the adhesion, uptake and surface protein recycling phenotypes observed, but is probably not responsible for the later impairments of phagosome maturation (Fig. 7).

Metazoan flotillins are lipid raft proteins that function as signaling platforms at the plasma membrane and in membrane trafficking of an increasing number of cargoes in the endocytic pathway (Otto and Nichols, 2011; Meister and Tikkanen, 2014). Vacuolins have been proposed to act as flotillin homologs in the amoeba D. discoideum, based on protein structure similarities and sequence conservation of the PHB domain (Fig. S3C; Rauchenberger et al., 1997; Jenne et al., 1998). In this study, we further confirm that all three vacuolins share similar biochemical characteristics with mammalian flotillins, in that they partially associated with alkali-resistant membranes (Fig. 1E,F). Flotillins insert into the cytosolic leaflet of membranes through post-translational palmitoylations occurring in the PHB domain, which are a prerequisite for membrane association (Morrow et al., 2002; Neumann-Giesen et al., 2004). The PHB domain of vacuolins was also shown to be required for membrane association, as evidenced by deletion mutants lacking this domain (Wienke et al., 2006). It is important to note that the PHB domain of each vacuolin possesses two cysteines (Fig. S3C), and we speculate that these might also be palmitoylated to allow membrane association. In addition, like flotillins, vacuolins have been shown to oligomerize via protein-protein interactions of the C-terminal coiled-coil regions, and are proposed to form trimers (Wienke et al., 2006; Solis et al., 2007). Interestingly, the membrane insertion topology of flotillins and vacuolins may differ, as the PHB domain of flotillins is at the N-terminus of the protein, whereas in vacuolins the PHB domain is more central (Fig. S3C). Moreover, although flotillins are found across several compartments, from the plasma membrane to endosomal vesicles and the Golgi apparatus (Babuke and Tikkanen, 2007), we mainly detected vacuolins on endocytic compartments and only weakly at the plasma membrane after fusion of PLs during exocytosis (Fig. 1). How vacuolins are targeted to their destination compartment currently remains unknown, but our data show that vacuolins only reside in membranes and cycle through different compartments exclusively by vesicular trafficking.

We found that deletion of one or more vacuolin severely impaired phagocytic uptake, independently of the nature of the particle (Fig. 2). In addition, we provide evidence that absence of vacuolins affects particle recognition and adhesion (Fig. 3C–G), but does not significantly affect formation and closure of the phagocytic cup (Fig. 3A,B). Because vacuolin KO cells were still able to take up particles after prolonged contact or pre-incubation with particles in the cold (Figs 3, 4), we hypothesize that the limiting step is not the adhesion and binding to the particle per se, but rather the clustering of receptors, which leads to the signaling required to trigger actin rearrangements and engulfment in a zipper-like manner. The severe uptake defect observed is reminiscent of the phenotype of the KO strain of the unconventional myosin MyoVII (Titus, 1999). MyoVII depletion does not alter formation or closure of the phagocytic cup, but affects adhesion to particles and substrate, thereby impairing motility, migration and uptake (Tuxworth et al., 2001, 2005; Gebbie et al., 2004). The interacting partner of MyoVII, the talin TalA, is also necessary for efficient substrate and particle adhesion, and thus a TalA KO exhibits a similar uptake defect (Niewöhner et al., 1997; Gebbie et al., 2004; Tuxworth et al., 2005). Interestingly, we found that in vacuolin KO cells, both genes are downregulated at the mRNA and protein levels (Fig. 6A–C) and consequently, MyoVII was severely depleted from the surface proteome in these cells (Fig. 4C). Surprisingly, when MyoVII was ectopically expressed in vacuolin KO cells, the adhesion and uptake defects were completely rescued (Fig. 6D,E), as well as the surface recycling defects (Fig. 6H,I). This suggests that absence of MyoVII and TalA are in fact responsible for the uptake defects observed in vacuolin KO mutants. We thus propose that vacuolins are important for the correct and efficient function of MyoVII, by stabilizing the MyoVII–TalA complex formed at specific areas of the phagocytic cup and early endocytic compartments (Fig. 7). When vacuolins are depleted, the MyoVII–TalA complex is destabilized and degraded, perhaps via the proteasome, and MyoVII functions are impaired. This results in a disorganized phagocytic cup, as indicated by decreased actin persistence at its base (Fig. S6C), defective adhesion to particles and impaired uptake (Fig. 7). Conversely, when MyoVII is re-expressed, TalA and MyoVII resume their correct localization even in absence of vacuolins (Fig. S6A,B), rescuing the uptake and recycling defects (Figs 6, 7). How the absence of vacuolins affects the mRNA levels of myoVII and talA is not yet fully understood. In fact, knock-out of myoVII was already reported to affect TalA protein levels (Gebbie et al., 2004). In addition, feedback loops where the loss of one component of a complex affects the protein levels of the other components have already been reported in other systems. For example, flotillin-1 and flotillin-2 are known to form hetero-oligomers that stabilize both proteins, and when one is knocked-down, the level of the other is strongly reduced (Ge et al., 2011). Moreover, knockdown of some of the components of the WASH complex strongly affects the stability of the other components (Jia et al., 2010). Therefore, it is possible that by impairing the stability of the MyoVII–TalA complex, the transcription and/or stability of their mRNAs are in turn affected. It is important to note, however, that vacuolin KO cells do not display motility or cytokinesis defects as severe as myoVII and talA KO cells (Niewöhner et al., 1997; Gebbie et al., 2004). This indicates that vacuolins might be only involved in regulating the function of MyoVII in uptake, receptor clustering and recycling, but not in other actin-dependent processes. Analysis of the surface proteome of WT and ΔABC cells highlighted two other myosins that were under-represented or absent from the surface of vacuolin KOs (Fig. 4C). MyoK, one of the hits we identified, is found at the phagocytic cup and is involved in cup closure as well as in trafficking of ER and lysosomal proteins to the phagosome (Dieckmann et al., 2010, 2012). Our observation that actin was depleted from the base of the phagocytic cup when vacuolins were absent (Fig. S6C), together with the fact that this phenomenon was only partially rescued by forced expression of MyoVII, points to the hypothesis that downregulation of MyoK and/or MyoA in the surface proteome of ΔABC cells (Fig. 4) is responsible. For example, it was already shown that MyoK KO cells exhibit a reduced uptake rate and a disorganized actin cortex around particles (Schwarz et al., 2000; Dieckmann et al., 2010). Surprisingly, the phenotypes described here do not reproduce what has been found with the previously reported vacA^- and vacB^- mutants (Jenne et al., 1998). It is possible that these phenotypes are specifically found in the AX2(Ka) D. discoideum background; however, we have shown that the vacB^- mutant has deletions of several genes (Fig. S4), which may have altered the original description of the effects of vacB^- KO. In addition, depletion of MyoVII was not observed in the original vacB^- KO (Fig. 6A), corroborating that MyoVII is the common denominator and is responsible for the uptake and membrane recycling phenotypes documented in this study.

Interestingly, upon vacuolin overexpression, cells were also impaired in uptake (Fig. 2D). Flotillins assemble on specific microdomains; however, when they are overexpressed, newly formed flotillin microdomains can be observed at the plasma membrane (Frick et al., 2007). In our case, we observed an increased number of vacuolin- and p80-positive compartments, suggesting that overexpression of vacuolins may affect formation and/or fusion of PLs (Fig. S2C). As vacuolins assemble in oligomers, we suggest that overexpression of single vacuolins causes an imbalance in the stoichiometry of the oligomers, thus affecting their function. This hypothesis is strongly corroborated by the fact that overexpression of vacuolins also affected MyoVII expression, and probably function, as shown by the uptake defects of these cells (Figs 6B,C and 2D). Interestingly, our data indicate that vacuolin acts upstream of MyoVII, and not vice versa, because in the absence of MyoVII, vacuolin expression was not affected (Fig. 6B). Thus, we propose that both absence and overexpression of vacuolins create an imbalance in the level and stoichiometry of vacuolin oligomers, which in turn affects MyoVII expression via the regulation of the stability of the MyoVII–TalA complex at the cortex.

Finally, we show here that vacuolins are involved in PL biogenesis, in that, in their absence, the reneutralization phase occurs slightly earlier and lysosomal enzymes and PL proteins accumulate earlier in phagosomes (Fig. 5A–C). It is plausible that vacuolins are involved in the delivery and/or retrieval of lysosomal and PL proteins from phagosomes, by creating specific sorting areas on the compartment. This function does not depend on MyoVII, because reintroduction of MyoVII only partially rescued these defects (Fig. 6H,I). Despite this slightly faster maturation, phagosomes of vacuolin KO mutants are still as bactericidal and proteolytic as those of WT cells. This confirms, once again, that acidification is not the only factor required for killing bacteria, which also requires proteolysis, oxidation and/or toxic metals (Cosson and Lima, 2014; Barisch et al., 2018).

To conclude, we report that D. discoideum vacuolins behave generally like flotillin homologs but also regulate the stability of the complex formed by MyoVII and TalA, which participates in phagocytosis by mediating receptor clustering and recycling of plasma membrane components.

D. discoideum strains and plasmids are listed in Table S1. Cells were axenically grown at 22°C in HL5c medium (Formedium, Hunstanton, UK) supplemented with 100 U/ml of penicillin and 100 μg/ml of streptomycin (Invitrogen, CA, USA).

To generate GFP knock-ins, 5′ and 3′ homologous regions of each vacuolin were cloned using the primers described in Table S2. The homology arms were cloned, using BglII/SpeI for the left arm (LA) and SalI/SacI for the right arm (RA), into the plasmid pPI183, which allows targeted in-frame integration of GFP at the C-terminus of the gene of interest (Paschke et al., 2018). Integration was confirmed by PCR, using flanking primers described in Table S2, and by western blot. In addition, the level of expression of KI genes was compared with that in WT cells by western blot and RT-qPCR to ensure comparable levels of expression. To generate single vacuolin KO strains, 5′ and 3′ homologous regions of each vacuolin were cloned using the primers described in Table S2 into the pKOSG-IBA-dicty1 plasmid following the manufacturer's instructions (IBA Lifesciences, Göttingen, Germany). Knockout was confirmed, using flanking primers described in Table S2, by PCR and RT-qPCR. The blasticidin resistance (BsR) cassette was then floxed from single mutants using the Cre recombinase. To generate double simultaneous ΔBC strains, the left arm of vacC and the right arm of vacB were cloned into pKOSG-IBA-dicty1 and clones were verified as before. To generate the triple ΔABC strain, the plasmid pJET-VacA-Hyg was transfected into the ΔBC strain and verified as before. Plasmids and primers used to generate GFP-tagged vacuolins are listed in Tables S1 and S2. Plasmids were transfected into D. discoideum by electroporation and selected with the relevant antibiotic. Hygromycin was used at a concentration of 15 μg/ml, blasticidin and G418 were used at a concentration of 5 μg/ml. The pDM631 PH_CRAC–GFP plasmid was provided by Dr J. King (University of Sheffield, Sheffield, UK), and the pDTI74 GFP–MyoVII plasmid was provided by Dr M. Titus (University of Minnesota, Minneapolis, Minnesota, USA).

The bacterial strains used in this study are listed in Table S1. Bacteria were cultured in LB medium at 37°C in shaking culture, as previously described (Froquet et al., 2009). Mycobacteria were grown in Middlebrook 7H9 (Difco, NJ, USA) supplemented with 10% OADC (Becton Dickinson, NJ, USA), 0.2% glycerol and 0.05% Tween 80 (Sigma Aldrich, MO, USA) at 32°C in shaking culture at 150 r.p.m. in the presence of 5 mm glass beads to prevent aggregation.

Recombinant nanobodies with the Fc portion of rabbit IgG that specifically recognizes VacA (αVacA256) or VacB (αVacB259) were obtained via the Geneva Antibody Facility (University of Geneva, Geneva, Switzerland) and were used at 1:2. Cross-reaction and specificity were tested by immunofluorescence on knockout strains. Note that these antibodies only work in immunofluorescence and that unfortunately, we were not able to produce VacC-specific antibodies. Fig. 1E illustrates that the 221-1-1 pan-vacuolin antibody (1:10; Dr M. Maniak, University of Kassel, Kassel, Germany; Jenne et al., 1998) is not able to discriminate between the three different vacuolins. The following antibodies were also used to detect: VatA (1:10; Dr M. Maniak; Jenne et al., 1998), VatM (N2; 1:200; Dr R. Allen, University of Hawaii, Honolulu, USA; Fok et al., 1993), actin (1:10; Dr G. Gerisch, Max Planck Institut für Biochemie, Martinsried, Germany; Westphal et al., 1997), Abp1 (1:10,000; Dieckmann et al., 2010), LmpA and LmpB (1:5000; Dr M. Schleicher, Institut für Zellbiologie, Ludwig-Maximilians-Universität, Munich, Germany; Janssen et al., 2001), mitochondrial porin (1:3000; Dr G. Gerisch), p80 (1:10; Geneva Antibody Facility), SibA (1:2000; Dr P. Cosson, University of Geneva, Geneva, Switzerland; Cornillon et al., 2006), DymA (1:2000; Dr D. J. Manstein, Max-Planck-Institut für Medizinische Forschung, Heidelberg, Germany; Wienke et al., 1999), common antigen-1 (1:1000; Dr M. Maniak), GlcNac1P (AD7.5; 1:10; Dr Hudson Freeze, La Jolla Cancer Research Foundation, La Jolla, California, USA; Souza et al., 1997), cathepsin D (1:2000; Dr J. Garin, Laboratoire de Chimie des Protéines, CEA-Grenoble, France; Journet et al., 1999), MyoVII (1:2000; Dr M. Titus, University of Minnesota, Minneapolis, Minnesota, USA; Tuxworth et al., 2005), TalA (1:10; Dr P. Cosson; Gebbie et al., 2004), and GFP (1:5000, pAb from MBL Intl.; 1:5000, mAb from Abmart, Shangai, China). Goat anti-mouse or anti-rabbit IgG coupled to AlexaFluor488 or AlexaFluor594 (Invitrogen) or to HRP (Brunschwig, Basel, Switzerland) were used as secondary antibodies.

The lipophilic membrane dye FM4-64 [N-(3-Triethylammoniumpropyl)-4-(6-(4-(Diethylamino) Phenyl) Hexatrienyl) Pyridinium Dibromide; Invitrogen] was used at a concentration of 1 μg/ml.

After SDS–PAGE separation and transfer onto nitrocellulose membranes (Protran; Schleicher and Schuell, Maidstone, UK), immunodetection was performed as previously described (Schwarz et al., 2000) but with ECL Prime blocking reagent (Amersham Biosciences, Little Chalfont, UK) instead of non-fat dry milk. Detection was performed with ECL Plus (Amersham Biosciences) using a Fusion Fx device (Vilber Lourmat, Collégien, France). Quantification of band intensity was performed with ImageJ (NIH, MD, USA).

For immunofluorescence, D. discoideum cells were fixed with ultra-cold methanol and immunostained as previously described (Hagedorn et al., 2006). Images were recorded with a Leica SP8 confocal microscope using a 63×1.4 NA or a 100×1.4 NA oil immersion objective.

Cells were plated on a μ-dish (Ibidi, Gräfelfing, Germany) in filtered HL5c, to decrease autofluorescence. After adhesion, beads or yeasts were first added to cells at 4°C by brief spinoculation, then time-lapse movies were recorded at room temperature with a spinning disk confocal system (Intelligent Imaging Innovations, Göttingen, Germany) mounted on an inverted microscope (Leica DMIRE2; Leica, Wetzlar, Germany) using a 63× or a 100×1.4 NA oil objective. Images were processed with ImageJ. Quantifications were performed manually. A 1 mm-thin agarose sheet was overlayed onto cells before time-lapse microscopy to allow better imaging of the phagosomes.

D. discoideum cells were plated overnight at a density of 2×10⁵ cells/ml in filtered HL5c. The next day, about 2×10⁴ cells diluted in Sorensen-sorbitol (14.7 mM KH₂PO₄, 2.5 mM NaHPO₄, pH 6.2, 120 mM sorbitol) were deposited in 96-well plates (Perkin Elmer, MA, USA) and left to attach for 30 min at room temperature. Images were recorded every 15 s for 30 min using a 10× objective with the ImageXpress Micro XL high-content microscope (Molecular Devices). Cells were tracked with the MetaXpress software (Molecular Devices, CA, USA) and average speed and persistence were calculated with Microsoft Office Excel.

D. discoideum cells were plated at a density of 1×10⁵ cells in filtered HL5c in µ-Slide 8-well glass bottom slides (Ibidi). Images were recorded using a 60× objective with a Nikon Eclipse Ti microscope using the SRIC function (Nikon, Tokyo, Japan). The area of adhesion of each cell was measured using ImageJ.

Assessment of fluorescent particle uptake was performed as previously described (Sattler et al., 2013). Data was acquired using a FACSCalibur flow cytometer (BD Biosciences, Allschwill, Switzerland) or FACS Gallios (Beckman Coulter, Nyon, Switzerland) and analyzed with FlowJo software (TreeStar) or Kaluza (Beckman Coulter).

The kinetics of acidification and proteolytic activity inside phagosomes were monitored as previously described (Sattler et al., 2013). Phagosomal pH was monitored using the pH-sensitive fluorophore FITC (Molecular Probes, OR, USA) coupled to 3 µm silica particles (Kisker Biotech, Steinfurt, Germany). Beads were also coated with the pH-insensitive CF594 (Biotium, Fremont, CA, USA) as a reference dye. Cells were plated in filtered HL5c on 35-mm Ibidi dishes. At 30 min before imaging, HL5c was replaced with LoFlo (Formedium). Beads were added at a cell:beads ratio of 1:2, and a 1 mm-thin agar sheet was overlaid on top of the cells. Movies were recorded with a Leica AF6000LX widefield microscope using a 40× objective. Images were taken every 1 min for 4 h. Tracking of bead-containing phagosomes and calculations of their emitted fluorescence ratios were performed with the Imaris software (Bitplane, Belfast, UK). Proteolysis was measured using beads coupled to the reference stain CF594 (Biotium) and the reporter DQ Green–BSA (Molecular Probes) at a self-quenching concentration. Upon proteolysis of BSA, DQ Green is released and dequenched, which causes an increase in fluorescence. Movies were recorded as described above.

The ability of D. discoideum to form plaques on a lawn of bacteria was monitored as previously described (Froquet et al., 2009). Briefly, 50 µl of an overnight bacterial culture was plated on SM-agar [10 g peptone (Oxoid), 1 g yeast extract (Difco), 2.2 g KH₂PO₄, 1 g K₂HPO₄ and 1 g MgSO₄×7H₂O with 20% glucose in 1 liter of ddH₂O] in wells from a 24-well plate. Serial dilutions of D. discoideum (10, 10², 10³ or 10⁴ cells) were plated onto the bacterial lawn, and the plates were incubated at 21°C for 4–7 d until plaque formation was visible. To quantify cell growth on bacteria, a logarithmic growth score was assigned as follows: plaque formation up to a dilution of 10 cells received a score of 1000; when cells were not able to grow at lower dilutions, they obtained the corresponding lower scores of 100, 10 and 1.

Intracellular killing of individual bacteria was measured as previously described (Leiba et al., 2017). Briefly, K. pneumoniae-GFP bacteria were mixed with D. discoideum cells at a ratio of 3:1 in Sorensen-sorbitol, deposited on an Ibidi dish for 10 min, then imaged every 30 s for 2–3 h with a Leica AF6000LX widefield microscope using a 40× objective with the autofocus function. Images were processed and quantified using ImageJ. Survival analysis of phagocytosed fluorescent bacteria was computed using the Kaplan–Meier estimator.

A total of 10⁹ D. discoideum cells were washed in Sorensen-sorbitol and resuspended in HESES buffer (20 mM HEPES, 250 mM sucrose, 5 mM MgCl₂ and 5 mM ATP) supplemented with proteases and phosphatase inhibitors (cOmplete EDTA-free and PhosSTOP, Roche, Basel, Switzerland). Cells were homogenized in a ball homogenizer with 10 μm clearance. The post-nuclear supernatant was diluted in HESES buffer and centrifuged at 35,000 rpm in a Sw60 Ti rotor (Beckman Coulter) for 1 h at 4°C. The cytosol (supernatant) and membrane (MB, pellet) fractions were recovered. The membrane fraction was further treated for 1 h on ice by addition of: HESES buffer (negative control), 1 M Urea (in HESES), 500 mM NaCl (in HESES) or 200 mM sodium carbonate at pH 11 (pure) in the same final volume as the recovered cytosol fraction (about 3 ml). Membranes were centrifuged at 35,000 rpm in a Sw60 Ti rotor for 1 h at 4°C. The supernatant (SN) and pellet (P) fractions were recovered. After acetone precipitation of the SN, both SN and P fractions were resuspended in equal volumes (1.5 ml each). The protein concentration of the cytosol fraction was further quantified by Bradford assay, and an equivalent (1:1) amount of MB, SN and P fractions were loaded for western blotting.

A total of 8×10⁹ D. discoideum cells were washed once in Sorensen-sorbitol buffer at pH 8 at 4°C. The cell surface was then biotinylated by incubating cells for 3 min with 30 mg of EZ-Link-Sulfo-NHS-LC-Biotin (Thermo Scientific, MA, USA) in Sorensen-sorbitol buffer at pH 8 on ice. After this step, latex beads were added and phagosomes were purified at different stages of maturation, as described in Gotthardt et al. (2006). Equivalent amounts of proteins from each time point were loaded for western blotting.

A total of 2×10⁷ D. discoideum cells were washed twice in Sorensen-sorbitol buffer at pH 6 at 4°C. The cell surface was biotinylated by incubating cells for 10 min with 1 mg of EZ-Link-Sulfo-NHS-LC-biotin (Thermo Scientific) in Sorensen-sorbitol buffer at pH 8 on ice. Cells were pelleted and incubated for 5 min on ice in PBS containing 10 mM glycine to quench unbound biotin. Cells were carefully washed four times in Sorensen-sorbitol buffer pH 6, then lyzed in RIPA buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% SDS, 2 mM EDTA, 1% Triton X-100, 0.5% sodium deoxycholate, 1× EDTA-free Roche protease Inhibitors) and incubated overnight with SpeedBeads Neutravidin magnetic beads (GE Healthcare, IL, USA). Pulled-down proteins and beads were then washed twice in RIPA, followed by a 15 min wash in 6 M urea and 2 washes in RIPA. Beads were then washed and collected in PBS and sent for mass spectrometry analysis. Protein were digested on-beads and peptides were analyzed by nanoscale liquid chromatography-tandem mass spectrometry (nanoLC-MSMS) using an easynLC1000 (Thermo Scientific) coupled with a Qexactive Plus mass spectrometer (Thermo Scientific). A database search was performed with Mascot (Matrix Science) using the D. discoideum reference proteome database from Uniprot. Data were analyzed and validated using Scaffold (Proteome Software), with quantitation based on spectral counting and normalized spectral abundance factors (NSF) with statistical t-test. GO-term enrichment analysis was performed using PANTHER (http://www.pantherdb.org/).

One 80% confluent dish of D. discoideum cells was collected and centrifuged. The cell pellet was resuspended in 400 μl of Trizol (Invitrogen, #AM9738) and RNA extracted using the Direct-zol RNA MiniPrep kit, according to manufacturer's instructions (Zymo Research, CA, USA). The quality and the quantity of RNA was confirmed with a Bioanalyzer (RNA 6000 Nano Kit, Agilent, CA, USA) and Qubit 2.0 fluorometer (Thermo Scientific). Libraries were constructed from 100 ng of RNA using the Ovation Universal RNA-Seq System kit (Nugen, CA, USA). The quality of the libraries was verified by TapeStation (Agilent, High Sensitivity D1000 ScreenTape). Samples were pooled and run in a single read 50 flow cell (Illumina, CA, USA) and on a Hiseq 4000 (Illumina). Bioinformatic analysis and quality controls were performed as previously described (Hanna et al., 2019 preprint). Moderated t-statistics were used for P-value calculation (limma-voom) and adjusted P-value correction was performed using the Benjamini–Hochberg method. The RNA-seq data analyzed here are presented in Hanna et al. (2019 preprint).

We gratefully acknowledge Dr P. Cosson (University of Geneva) for discussions and suggestions, Romain Bodinier (University of Geneva) for help with RICM experiments, and the staff of the Bioimaging Center for Microscopy, the FACS core facility, the Proteomics core facility, the Geneva Antibody facility and the IGE3 Genomics Platform at the Faculty of Sciences and Faculty of Medicine of the University of Geneva for their precious help. We thank Dr J. King for sharing plasmids (pDM631 PH_CRAC–GFP) and critical reading of the manuscript, Dr M. Titus for sharing plasmids (pDTI74 GFP–MyoVII) and antibodies (anti-MyoVII) and Dr D. Moreau at the ACCESS Geneva Imaging Facility of the University of Geneva for help with the high-content microscope experiments.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.242974.reviewer-comments.pdf