Defective lysosome maturation and Legionella pneumophila replication in Dictyostelium cells mutant for the Arf GAP ACAP-A

The amoeba Dictyostelium discoideum is a genetically tractable model organism extensively used to study membrane trafficking in the endocytic pathway. In this professional phagocyte, engulfed material travels through a series of internal compartments (Maniak, 2003). Rapidly after uptake, newly formed macropinosomes and phagosomes recruit the vacuolar proton ATPase (H⁺-ATPase) and hydrolytic enzymes to form acidic lysosomes. About 30 min later, lysosomes mature into neutral post-lysosomes, and finally undigested material is ejected from the cells by exocytosis. One key step in lysosome maturation is the depletion of H⁺-ATPase from lysosomes leading to pH neutralization. H⁺-ATPase retrieval is mediated by small actin-coated vesicles budding from lysosomes (Carnell et al., 2011; Clarke et al., 2010). The role of actin is essential in this process. Hence, depolymerization of F-actin by drugs or the absence of Wiskott–Aldrich syndrome protein (WASH) or Scar family proteins, actin nucleation-promoting factors recruited on these recycling vesicles, completely blocks H⁺-ATPase retrieval and lysosome maturation (Carnell et al., 2011; King et al., 2013).

In mammalian cells, the ADP-ribosylation factors (Arfs) proteins regulate endosomal membrane traffic by cycling between active GTP-bound and inactive GDP-bound forms (D'Souza-Schorey and Chavrier, 2006; Donaldson and Jackson, 2011; Gillingham and Munro, 2007). Arf activity is regulated by Arf guanine-nucleotide-exchange factors (Arf GEFs) and by Arf GTPase-activating proteins (Arf GAPs) that promote hydrolysis of GTP bound to Arf. Among the Arf GAP family of proteins, the ACAP subfamily includes three members (ACAP1–ACAP3) in mammalian cells, and comprises four functional domains: a Bin, amphiphysin and Rvs (BAR) domain, a pleckstrin homology (PH) domain, a GAP domain and an ankyrin repeat (ANK) domain. ACAPs mainly regulate the dynamic reorganization of the actin cytoskeleton (Inoue and Randazzo, 2007; Kobayashi and Fukuda, 2012; Kobayashi and Fukuda, 2013; Randazzo et al., 2007), although functions in cargo recycling from endosomes have been reported (Li et al., 2005a; Li et al., 2007). D. discoideum has only one Arf protein (ArfA), which is a substrate for the two members of the ACAP subfamily (ACAP-A and ACAP-B) in this amoeba (Chen et al., 2010; Gillingham and Munro, 2007; Weeks, 2005). We have recently shown that ACAP-A is specifically involved in cytokinesis, cell migration and actin cytoskeleton dynamics in an ArfA-dependent manner (Dias et al., 2013). However, the role of ACAPs in the endocytic pathway of D. discoideum is still unknown.

Legionella pneumophila is the causative agent of Legionnaires' disease and leads to acute pneumonia in humans after inhalation of contaminated water aerosols (Lamoth and Greub, 2010). L. pneumophila pathogenic strains emerge after intracellular multiplication in protozoans and then infect human lung alveolar macrophages. L. pneumophila has developed a strategy to escape host-cell defenses mainly by altering membrane trafficking mechanisms in the endocytic and secretory pathways. Upon injection of effector proteins by a bacterial type IV secretion system, L. pneumophila drives the formation of an intracellular compartment, the ‘Legionella-containing vacuole’ (LCV), and avoids the normal host cell endocytic pathway (Allombert et al., 2013; Xu and Luo, 2013). Insights into L. pneumophila virulence have been advantageously provided by studies using D. discoideum as a host cell model because L. pneumophila can infect this amoeba with features highly comparable to that observed in human macrophages (Hägele et al., 2000; Solomon et al., 2000). Hence, newly formed LCVs are transported inside the cell on microtubules and rapidly acquire endoplasmic reticulum (ER) markers. Concomitantly, a reduced fusion rate with lysosomes results in a low amount of vacuolar H⁺-ATPase in these endomembrane structures (Lu and Clarke, 2005). Interestingly, the GTPase Arf1 has been shown to participate in the biogenesis of the LCV in macrophages (Derré and Isberg, 2004; Kagan and Roy, 2002; Kagan et al., 2004) and the L. pneumophila effector protein RalF has been described to recruit and activate Arf1 to the LCV (Nagai et al., 2002). Although ArfA is known to be present in newly formed LCVs (Urwyler et al., 2009), it is unknown whether ArfA plays a similar role to Arf1 in D. discoideum infected cells.

In this study, we explored the function of ACAP-A in the endocytic pathway. We report for the first time that ACAP-A, but not ACAP-B, is involved in maturation of lysosomes, through an Arf-dependent mechanism, and in fusion of post-lysosomes with the plasma membrane. Moreover, we show that ACAP-A is specifically required for the efficient intracellular growth of Legionella pneumophila, revealing thus an unexpected role of this Arf GAP in host-defense mechanisms against pathogens.

The role of ACAP-A in the formation of actin filopodia (Dias et al., 2013) prompted us to assess whether this Arf GAP is involved in fluid-phase and particle endocytosis, two other actin-driven processes. To evaluate the role of ACAP-A in fluid-phase endocytosis, cells in which the gene encoding ACAP-A has been disrupted (acapA⁻ cells) (Dias et al., 2013) were incubated with the fluid-phase marker Alexa-Fluor-647-labeled dextran, and uptake was monitored by flow cytometry (Fig. 1A). acapA⁻ and wild-type (WT) cells showed comparable rates of uptake for the first 20 min. However, whereas uptake began to plateau after 60 min in WT cells, uptake continued to rise in acapA⁻ cells (180.6% of WT uptake after 160 min). When acapA⁻ cells were stably transfected with GFP-tagged ACAP-A, fluid-phase uptake markers did not accumulate anymore and uptake was even slightly inhibited (Fig. 1A). Dextran accumulation in acapA⁻ cells suggested that the extracellular release of the fluid-phase marker might be defective in these cells. Alternatively, a prolonged intracellular transit time of dextran might also delay fluid-phase discharge in acapA⁻ cells. To discriminate between these two possibilities, we first measured fluid-phase exocytosis in acapA⁻ cells. Cells were loaded overnight with Alexa-Fluor-647–dextran, washed in dextran-free medium, and the fluorescence remaining in cells over time was measured. Similar rates of exocytosis were observed in acapA⁻ and WT cells for the first 40 min (Fig. 1B), indicating normal egress of post-lysosomes in the absence of ACAP-A. However, analysis of later time points revealed that there was a slight reduction in exocytosis kinetics in acapA⁻ cells. Next, the intracellular transit time of fluid-phase was determined. Cells were pulsed for 10 min with Alexa-Fluor-647–dextran, and intracellular fluorescence was recorded after different chase periods. The half-life for internalized dextran was 90 min in WT cells, and rose to 115 min in acapA⁻ cells (Fig. 1C). We thus conclude that the apparent increase of fluid-phase uptake observed in acapA⁻ cells is mainly due to a prolonged transit time of internalized fluid compared to WT cells, although a minor exocytosis defect might also contribute to this increase. Interestingly, disruption of the gene encoding ACAP-B (acapB⁻ cells) did not result in any significant defects in endocytosis (data not shown), exocytosis and intracellular transit time of fluid-phase markers (supplementary material Fig. S1A,B). These results indicate that even if ACAP-A and ACAP-B are both Arf GAPs for ArfA (Chen et al., 2010), ACAP-A has distinct functions in the endocytic pathway.

One essential step in the D. discoideum endocytic pathway is the neutralization of acidic lysosomes as they mature into post-lysosomes (Maniak, 2003). To determine whether the kinetics of this event was affected in the absence of ACAP-A, we first monitored lysosome neutralization by live-cell confocal microscopy. Cells were pulsed with a mixture of Oregon-Green-labeled dextran (a pH-sensitive probe) and TRITC-labeled dextran (a pH-insensitive probe) for 15 min, and chased with dextran-free culture medium for different periods of time. In this experimental setup, endosomes with neutral pH are yellow, but acidic endosomes appear red due to weak fluorescence of Oregon Green at low pH. After the pulse period, all endosomes were acidic in acapA⁻ and WT cells. However, neutralization was delayed in acapA⁻ cells; only 6.8% of endosomes were neutral after 35 min compared to 24.5% in WT cells (Fig. 2A, arrowheads). After 55 min, neutralization was similar in both cell types.

To analyze endosomal pH variations more precisely, we next used a previously described assay (Marchetti et al., 2009). Cells were pulsed with dextran coupled to pH-sensitive and pH-insensitive probes (Oregon Green and Alexa Fluor 647 respectively) for 10 min and chased with dextran-free culture medium. Fluorescence was analyzed here by flow cytometry, and the ratio of Oregon Green and Alexa Fluor 647 fluorescence intensities, reflecting pH variations, were calculated and compared to a pH calibration curve (Fig. 2B). In both acapA⁻ and WT cells, early endosome acidification, during the first 20 min, occurred at a comparable rate. This first step was followed by a neutralization phase that started with a slight delay in acapA⁻ cells compared to WT cells (starting at 25 min and 20 min, respectively). The kinetics of neutralization was also delayed in acapA⁻ cells. Indeed, neutralization was achieved at 45 min and 35 min in acapA⁻ and WT cells, respectively. It is noteworthy that acapB⁻ cells did not show endosome acidification and neutralization defects (supplementary material Fig. S1C). Taken together, these results suggest that lysosomal pH neutralization is slightly and specifically affected in the absence of ACAP-A.

To investigate whether ACAP-A plays a role in phagocytosis, acapA⁻ cells were incubated with fluorescent beads and the intracellular fluorescence was measured by flow cytometry. After 20 min of incubation, acapA⁻ and WT showed similar phagocytic rates (Fig. 3A). Furthermore, bead uptake constantly increased over time in acapA⁻ cells (to 162.4% of the amount of phagocytosis in WT cells after 80 min) whereas a plateau was rapidly reached in WT cells after 1 h (Fig. 3A). Despite enhanced phagocytic capacities, when acapA⁻ and WT cells were grown on Klebsiella pneumoniae, a commonly used strain permissive for D. discoideum growth, phagocytic plaques were observed in both cell types; however, the diameter of phagocytic plaques was smaller in acapA⁻ cells (Fig. 3B). This defect was not due to reduced phagocytic capacities because WT and acapA⁻ cells showed comparable uptake of live K. pneumonia (supplementary material Fig. S2). Furthermore, small phagocytic plaques were also detected in acapA⁻ cells fed upon several other bacteria strains (supplementary material Fig. S3), in agreement with a general growth defect on bacteria rather that a strain-dependent defect. In contrast, growth of acapB⁻ cells on K. pneumoniae was comparable to that of WT cells (supplementary material Fig. S1D). Next we tested whether acapA⁻ cells could properly sense bacteria. Given that folate is secreted by bacteria and acts as a chemoattractant for D. discoideum (Burkholder and McVeigh, 1942; Pan et al., 1972), we analyzed cell migration on non-nutrient agar in response to folate (Fig. 3C). In the absence of ACAP-A, cells were still able to sense folate; however, migration after 4 h was reduced by 12.6% compared to WT cells. Therefore, this slight chemotaxis defect might partially account for the growth defect. Alternatively, killing of internalized bacteria might be less effective in acapA⁻ cells and/or phagosomal proteolysis of ingested bacteria might be partially defective in these cells. As shown in Fig. 3D, WT and acapA⁻ cells were able to kill K. pneumoniae with similar efficiency. We also tested the enzymatic activities of some lysosomal enzymes involved in particle digestion. After 3 days of culture, we noticed increased intracellular enzymatic activities in acapA⁻ cells compared to WT cells, whereas very low glycosidase activities were detected in the extracellular medium in all tested cells (Fig. 3E). Note that these enhanced enzymatic functions might reflect intracellular accumulation of lysosomal enzymes because the amount of cathepsin D (a lysosomal protease) substantially rose in acapA⁻ cells, whereas extracellular secretion of cathepsin D was slightly reduced in these cells (Fig. 3F). Interestingly, only the mature form of cathepsin D was detected in WT and acapA⁻ cells (data not shown), suggesting that there was normal maturation of lysosomal enzymes in the absence of ACAP-A. Next, we measured the proteolytic activity of phagosomes in living cells using beads coupled to Alexa Fluor 594 and to BSA labeled with DQ-Green at a self-quenching concentration, which become dequenched upon proteolysis of BSA (Gopaldass et al., 2012; Le Coadic et al., 2013). We observed that protease activity was substantially increased in acapA⁻ cells (Fig. 3G). This observation strongly suggests that active lysosomal enzymes are correctly targeted to phagosomes in acapA⁻ cells, and their accumulation in this compartment might enhance phagosomal proteolysis capacities. Taken together, these results indicate that the slow growth rate of acapA⁻ cells fed upon K. pneumoniae is not due to defects in ingestion, killing or digestion of bacteria but instead might be caused by genuine bacteria recognition problems.

The above results suggested that the organization of the endocytic pathway might be defective in the absence of ACAP-A. To evaluate this possibility, acapA⁻ cells grown in culture medium were first observed by transmission electron microscopy and compared to WT cells. The overall morphology of the acapA⁻ cells was similar to that of WT cells (Fig. 4A). However, quantitative analysis revealed that mutant cells harbored more vacuoles per cell than WT cells (Fig. 4B). D. discoideum cells are characterized by the presence of acidic lysosomes that mature into neutral post-lysosomes, which can then fuse with the plasma membrane. Transmission electron microscopy observations do not allow us to distinguish between these various types of endocytic compartments. Therefore, to further characterize the endocytic pathway of acapA⁻ cells, WT and mutant cells were loaded for 5 h with a mixture of Oregon-Green- and TRITC-labeled dextran to distinguish neutral vesicles (Fig. 5A, yellow, arrowheads) and acidic vesicles (red). Both cell types showed numerous acidic vesicles and a few neutral post-lysosomes. However, acapA⁻ cells displayed more post-lysosomes as compared with WT cells (2.8 and 1.4 post-lysosomes per cell for acapA⁻ and WT cells respectively, 30 cells analyzed for each cell type). To better characterize this defect, we next analyzed the endosomal distribution of F-actin, the endocytic marker p80 and the membrane H⁺-ATPase (Neuhaus et al., 1998; Ravanel et al., 2001). In both acapA⁻ and WT cells, lysosomes (which are H⁺-ATPase-positive, p80-positive and F-actin-positive) (Fig. 5B, arrows) and post-lysosomes (which are H⁺-ATPase-negative, p80-positive and F-actin-negative) (Fig. 5B, arrowheads) were both detected. The size of lysosomes was similar in both cell types, whereas post-lysosomes were significantly smaller in acapA⁻ cells than those in WT cells (Fig. 5C). These results indicate that endocytic compartments in acapA⁻ cells display genuine characteristics of lysosomes and post-lysosomes as tested with three relevant markers (H⁺-ATPase, p80 and F-actin). In addition, the endocytic marker p25 (a marker of recycling endosomes in D. discoideum) (Ravanel et al., 2001), vacuolin (a protein binding to the surface of late endosomes; Rauchenberger et al., 1997) and two markers of the contractile vacuole (H⁺-ATPase and Rhesus 50) (Benghezal et al., 2001; Neuhaus et al., 2002) were correctly localized in acapA⁻ cells (data not shown), further indicating that the composition of intracellular compartments is preserved in acapA⁻ cells.

Remarkably, acapA⁻ cells showed an increased number of p80-positive vacuoles compared to WT cells (Fig. 5B,D). acapA⁻ cells displayed 1.56 times more lysosomes and 2.4 times more post-lysosomes than WT cells (Fig. 5D). Moreover, the ratio between post-lysosomes and lysosomes for acapA⁻ and WT cells (0.57 and 0.37, respectively) indicated acapA⁻ cells have more post-lysosomes than would be expected if a normal post-lysosome:lysosome ratio was conserved. This accumulation of post-lysosomes suggested that exocytosis of these vesicles might be defective in acapA⁻ cells. To test this hypothesis, we counted the number of p80 patches on the surface of acapA⁻ and WT cells. These patches transiently appear upon the fusion of post-lysosomes with the plasma membrane, and are conveniently used to determine the efficiency of vesicle fusion (Charette and Cosson, 2006; Charette and Cosson, 2007). As shown in Table 1, acapA⁻ and WT cells displayed similar number of p80 patches. However calculation of the ratio between p80 patches and post-lysosomes numbers revealed that fusion efficacy of post-lysosomes with the cell surface was actually reduced in acapA⁻ cells (Table 1). This result supports a possible function of ACAP-A in the exocytic process. Strikingly, this post-lysosome fusion defect in acapA⁻ cells results only in a minor fluid-phase exocytosis deficiency. This apparent discrepancy might be explained by inherent sensibility differences in assays used here to analyze these endocytic functions.

To further test the role of ACAP-A in the maturation of lysosomes, we next followed the maturation of endocytic compartments containing internalized indigestible particles in acapA⁻ cells. Cells were pulsed with fluorescent latex beads for 15 min, and were then washed. The localization of particles in lysosomes and post-lysosomes was determined by analysis of p80 and H⁺-ATPase markers (see representative images in Fig. 6A), and quantified after different incubation times (Fig. 6B) (Charette and Cosson, 2007; Charette and Cosson, 2008). After 30 min, most ingested particles localized to lysosomes in all tested cell lines. In WT cells, particles were efficiently transferred to post-lysosomes over the next 30 min. By contrast, in acapA⁻ cells, transfer of particles to post-lysosomes still occurred but with slower kinetics than in WT cells, further indicative of a defect in lysosome maturation in these mutant cells. Whereas expression of ACAP-A–GFP restored the transfer of particles to post-lysosomes to normal kinetics, an ACAP-A mutant defective in the GAP activity (mutation R633Q) failed to complement the defect of acapA⁻ cells (Fig. 6B). Therefore, the role of ACAP-A in the maturation of lysosomes might rely on the regulation of ArfA activity by this GAP. Finally, efficient transfer of particles to post-lysosomes was also observed in acapB⁻ cells (supplementary material Fig. S1E), further highlighting the distinct functions of ACAP-A and ACAP-B.

Legionella pneumophila is an intracellular pathogenic bacterium that efficiently replicates in the LCVs formed in D. discoideum owing to bacteria-induced alteration of membrane trafficking events in this host cell (Hägele et al., 2000; Solomon et al., 2000). Because Arf is involved in the biogenesis of the LCV in macrophages (Derré and Isberg, 2004; Kagan and Roy, 2002; Kagan et al., 2004), we next tested whether the Arf GAP ACAP-A is required for intracellular growth of L. pneumophila in D. discoideum. Amoebae were infected with L. pneumophila (Lens strain) expressing fluorescent mCherry. After a lag period of 72 h post-infection, intracellular growth of bacteria was detected in both acapA⁻ and WT cells as measured by mCherry fluorescence analysis (Fig. 7A). At 144 h post-infection, a 5.4-fold increase in the initial fluorescence was observed in WT cells. In contrast, bacteria growth in acapA⁻ cells showed a marked plateau starting at 96 h, and was reduced by 34.7% compared to growth in WT cells at 144 h. Stable transfection of ACAP-A–GFP in acapA⁻ cells restored the normal L. pneumophila intracellular growth rates. Note that cell infection was dependent on the bacterial type IV secretion system (T4SS) Dot/Icm because the T4SS dotA L. pneumophila mutant failed to replicate in all tested amoeba cells (Fig. 7A). To exclude the possibility that inhibition of L. pneumophila growth in acapA⁻ cells resulted from a defect in bacteria uptake, we tested the uptake of L. pneumophila expressing GFP by flow cytometry. After 30 min of incubation with L. pneumophila, acapA⁻ and WT cells showed similar uptake of bacteria (supplementary material Fig. S4A).

Defects in intracellular L. pneumophila growth in acapA⁻ cells suggested that the formation and the composition of LCVs might be altered in the absence of ACAP-A, thus preventing optimal replication of L. pneumophila. To test this hypothesis, we analyzed the presence of ER and lysosomal markers on the LCV in acapA⁻ cells because this compartment is characterized by the presence of ER markers and the exclusion of lysosomal proteins. At 2 h post-infection, L. pneumophila localized in the LCVs containing the ER membrane proteins calnexin fused to GFP (Fig. 7B) and protein disulfide isomerase (PDI; data not shown) to a similar extent in both acapA⁻ and WT cells (Fig. 7C). As previously reported (Ragaz et al., 2008; Weber et al., 2006), the endocytic marker p80 was observed on the LCV (Fig. 7D), whereas the lysosomal marker H⁺-ATPase was only found on a minority of these vacuoles in WT cells (Fig. 7E). Such a composition was also observed in acapA⁻ cells (Fig. 7D,E). These results strongly suggest that ACAP-A might be dispensable for the formation of the LCV.

At a few hours post-infection, LCVs displaying ER markers have been described to change from a tight to a spacious morphology (Li et al., 2005b; Lu and Clarke, 2005). This morphological transition appears to be essential for efficient bacteria intracellular replication because several D. discoideum mutants defective in this step show reduced bacteria growth rates (Li et al., 2005b; Ragaz et al., 2008; Weber et al., 2006). Careful analysis of the LCV morphology in acapA⁻ cells revealed no major LCV maturation defects that could account for defective L. pneumophila intracellular replication (data not shown).

Finally, proteomic analysis of purified LCVs has revealed that ArfA is detected 1 h post-infection and then rapidly lost (Urwyler et al., 2009), although our microscopy studies did not provide convincing evidence for the localization of ArfA–GFP on the LCV (data not shown). Given that ACAP-A is a GAP for ArfA, we next tested the localization of ACAP-A on the LCV. Analysis of WT cells expressing ACAP-A–GFP revealed that ACAP-A did not accumulate on membranes of LCVs positive for the ER marker PDI (at least 200 vacuoles were analyzed; see a representative image in supplementary material Fig. S4B) but was typically observed at the plasma membrane and the cytosol, as previously reported in non-infected WT cells (Dias et al., 2013). Although a transient recruitment of ACAP-A–GFP to the LCV cannot be formally excluded, this observation suggests that the localization of ACAP-A to the LCV is not required for the LCV biogenesis. This result is in agreement with the absence of ACAP-A in the proteome of purified LCVs previously reported (Urwyler et al., 2009).

The growth defect of L. pneumophila in acapA⁻ cells prompted us to analyze the role of ArfA in intracellular growth of this bacterium because ArfA is a substrate for ACAP-A (Chen et al., 2010). ArfA-Q71L–GFP and ArfA-T31N–GFP mutants were conditionally expressed in cells 48 h before infection. These ArfA mutants mimic constitutively activated (Arf bound to GTP) and dominant-negative (Arf bound to GDP) forms of ArfA, respectively (Dias et al., 2013). In ArfA-T31N–GFP-expressing cells, L. pneumophila growth was reduced by 24.7% compared to growth in WT cells at 144 h post-infection (Fig. 7F). In contrast, no bacteria growth defect was observed in ArfA-Q71L–GFP-expressing cells. Moreover, the formation of the LCV was altered in ArfA-T31N–GFP-expressing cells. These cells showed a 28.8% reduction in LCVs containing the ER marker calnexin (Fig. 7G) whereas the lysosomal marker H⁺-ATPase was still massively excluded from the LCV (data not shown).

In this study, we report the role of ACAP-A, a GAP for ArfA, in the endocytic pathway of the amoeba D. discoideum. In the absence of ACAP-A, maturation of lysosomes into post-lysosomes is significantly perturbed, and expression of a GAP-defective ACAP-A mutant does not restore normal functions in acapA⁻ cells. These results provide thus the first evidence that ACAP-A might play a role in lysosome maturation through an Arf-dependent activity. It is noteworthy that lysosome maturation is only partially blocked in acapA⁻ cells. This suggests that additional factors might also regulate this process. Such a moderate lysosome maturation defect has been previously described in lvsB-null cells (Charette and Cosson, 2007; Charette and Cosson, 2008). LvsB is D. discoideum ortholog of mammalian lysosomal trafficking regulator (LYST) and regulates the fusion between lysosomes and post-lysosomes (Kypri et al., 2013). In contrast, more severe lysosome maturation defects have been reported in two other well-characterized mutants, μ3- and WASH-null cells, both of which are impaired in vesicle fission events. Hence, cells lacking the μ3 subunit of the AP-3 clathrin adaptor complex show defects in protein recycling from early endosomes because of a vesicle-fission-dependent mechanism that results in major defects along the endocytic pathway (Charette and Cosson, 2006</emph>; Charette and Cosson, 2008). Moreover, in the absence of the nucleation-promoting factor WASH, the removal of H⁺-ATPase from late lysosomes is impaired and neutralization of lysosomes to form post-lysosomes is fully inhibited (Carnell et al., 2011). In contrast to WASH-null cells, neutralization of lysosomes containing a pH-sensitive fluid-phase marker still occurs in acapA⁻ cells but with reduced kinetics. This result strongly suggests that ACAP-A might only partially control the retrieval of H⁺-ATPase from lysosomes leading to pH neutralization. The retrieval of H⁺-ATPase from maturing lysosomes relies on vesicles (Clarke et al., 2010) whose biogenesis depends on actin polymerization driven by WASH (Carnell et al., 2011). Because ACAP-A has been shown to regulate the actin cytoskeleton dynamics in an ArfA-dependent manner (Dias et al., 2013), we propose that ArfA and ACAP-A could also participate in the control of actin polymerization on these recycling vesicles by regulating the activity of actin nucleation-promoting factors such as WASH. In agreement with this hypothesis, mammalian Arf1 has been recently shown to cooperate in vitro with the Rho GTPase Rac1 to fully activate the WAVE regulatory complex that controls actin polymerization to membranes (Koronakis et al., 2011). More detailed studies, especially at the single-cell level using live-cell microscopy to record trafficking of fluorescently-tagged key proteins (e.g. H⁺-ATPase and WASH), will be required to further determine the precise function of ACAP-A in lysosome maturation.

In addition to lysosome maturation, our results suggest that ACAP-A might play a role in fusion of post-lysosomes with the plasma membrane. Hence, acapA⁻ cells show a reduced rate of fusion of post-lysosomes with the cell surface leading to accumulation of post-lysosomes. Although ACAP-A might control some important steps in the membrane fusion process per se, we favor the hypothesis that ACAP-A might control the fusion competency of post-lysosomes or the targeting of post-lysosomes to the plasma membrane. For instance, ACAP-A might regulate tethering factors involved in vesicle docking to the plasma membrane, such as the exocyst complex (Hsu et al., 1996; TerBush et al., 1996). Consistent with this hypothesis, Arf6 has been shown to associate and regulate the exocyst during endocytic membrane recycling in mammalian cells (Prigent et al., 2003). Although the molecular mechanism of exocytosis is unknown in D. discoideum, a yeast two-hybrid screen with Sec15 (an exocyst subunit) has revealed highly that it interacts with ArfA (Essid et al., 2012), in accordance with the proposed regulatory function of ACAP-A in vesicle docking during exocytosis.

Here, we show that intracellular replication of L. pneumophila is partially inhibited in acapA⁻ cells, but that the morphology and the composition of the LCV still appear to be intact. This normal biogenesis of the LCV in acapA⁻ cells indicates that ACAP-A is dispensable for the formation of the replicative niche. In macrophages, Arf1 plays essential functions of in the formation of the LCV (Derré and Isberg, 2004; Kagan and Roy, 2002; Kagan et al., 2004). Here, we found that expression of a dominant-negative variant of ArfA in D. discoideum affects both the growth of L. pneumophila and the formation of the LCV. This LCV biogenesis defect is similar to that described in macrophages overexpressing the Arf1 dominant-negative variant Arf1-T31N–GFP (Kagan and Roy, 2002). We thus provide the first evidence that ArfA might play a role similar to Arf1 in cells infected with D. discoideum. Although ArfA mutants might not fully simulate the actual fine-tuning of ArfA activity, our results suggest that ArfA might be in an activated state (GTP-bound) to support the LCV biogenesis. Accordingly, we observe that the lack of ACAP-A in acapA⁻ cells, inferred to induce excess activated ArfA, does not impede the formation of the LCV. So, why is the replication of L. pneumophila inhibited in acapA⁻ cells in the absence of defects in LCV biogenesis? Although proteomic analysis of the LCV composition in acapA⁻ cells would probably give some hints for answering this question, the ACAP-A functions in the endocytic pathway revealed here might also readily provide interesting clues on the role of ArfA and ACAP-A in L. pneumophila intracellular replication. Indeed, our results point to a role of ACAP-A in the control of actin dynamics, and this function might be essential for optimal intracellular growth of L. pneumophila. In agreement with this hypothesis, mounting evidence reveals that the actin cytoskeleton dynamics plays a fundamental role in L. pneumophila virulence (Franco and Shuman, 2012). Besides classical functions in the uptake mechanism by host cells and its presence on the LCV at later times post-infection, actin has recently been shown to be the target of the L. pneumophila effector VipA. This bacterial effector binds and nucleates actin polymerization in vitro and it associates with actin filaments and early endosomes during macrophage infection (Franco et al., 2012). The control of actin dynamics by VipA is thought to interfere with host cell organelle trafficking pathways, preventing deleterious connections of the LCV with the endocytic machinery. This report on VipA identifies actin as a key factor in L. pneumophila virulence, hence ACAP-A might be an important regulatory element in actin-dependent steps. Alternatively (or in addition) to the modulation of actin dynamics, the role of ACAP-A in L. pneumophila virulence could rely on other well-known Arf activities. Among those, Arf proteins have been described to regulate vesicular traffic and phospholipid metabolism (D'Souza-Schorey and Chavrier, 2006; Donaldson and Jackson, 2011; Gillingham and Munro, 2007). These functions are essential for L. pneumophila infection. For instance, several translocated effectors have been described to bind phosphoinositides on the LCV and promote a modification in lipid membrane composition required for the LCV biogenesis (Haneburger and Hilbi, 2013).

Finally, the L. pneumophila infection mechanisms in D. discoideum and human macrophages are highly comparable (Hägele et al., 2000; Solomon et al., 2000). Therefore, mammalian Arf GAPs homologous to ACAP-A might play similar roles in L. pneumophila infection of macrophages. There are three different mammalian ACAPs with high similarity to ACAP-A (Gillingham and Munro, 2007; Chen et al., 2010). Given that ACAP1 and ACAP2 have preferential GAP activities towards Arf6 in vivo (Jackson et al., 2000), an involvement in L. pneumophila infection seems unlikely. ACAP3 has not been characterized yet. In addition to ACAPs, ACAP-A also shares a domain organization comparable to mammalian ASAPs, which also have with BAR, PH, Arf GAP and ankyrin repeat domains (Gillingham and Munro, 2007; Chen et al., 2010). Interestingly, ASAP1 has been shown to function as a GAP for Arf1 in vivo (Furman et al., 2002), a requisite for a hypothetic role in L. pneumophila infection. Further studies will be needed to evaluate the functions of ACAPs and ASAPs during L. pneumophila infection of macrophages.

D. discoideum strain DH1-10 (Cornillon et al., 2000), and mutants acapA⁻ and acapB⁻ (Dias et al., 2013) were grown at 22°C in HL5 medium. Plasmids encoding calnexin–GFP (Müller-Taubenberger et al., 2001), ACAP-A–GFP and the ACAP-A–GFP mutant R633Q (Dias et al., 2013) were transfected in cells by electroporation as reported previously (Alibaud et al., 2003). ArfA(Q71L) and ArfA(T31N) mutants cloned in the inducible expression vector pDM370 were as previously described (Dias et al., 2013). Expression was induced by adding 10 mg/ml doxycycline 2 or 3 days before analysis. L. pneumophila strains (virulent L. pneumophila serogroup 1, strain Lens, and avirulent dotA mutant Lens lpl2613::Km) were grown at 30°C on buffered charcoal yeast extract (BCYE) agar or in BYE liquid medium. The laboratory strain of K. pneumonia (Benghezal et al., 2006) was grown overnight at 37°C in LB medium.

Internalization of fluid phase (Alexa-Fluor-647–dextran; Molecular Probes, Eugene, OR) or phagocytic particles (FITC-conjugated fluorescent 1-µm diameter latex beads; Polysciences, Warrington, PA) was carried out as previously described (Lima et al., 2012) and analyzed by flow cytometry (FACS Calibur, Becton Dickinson, San Jose, CA).

Uptake of L. pneumophila strains by D. discoideum was analyzed by flow cytometry using GFP-labeled bacteria. L. pneumophila cells harboring a plasmid expressing GFP were grown on BYE liquid medium containing 2 mM IPTG and chloramphenicol for 4 days at 30°C. Amoeba were seeded in 24-well microplates (5×10⁵ cells per well) and then infected with 5×10⁷ fluorescent L. pneumophila [multiplicity of infection (MOI) = 100]. Infection was synchronized by centrifuging bacteria at 880 g for 10 min and plates were incubated 20 min at 25°C. Extracellular bacteria were removed by washing three times with SorC (2 mM Na₂HPO₄, 15 mM KH₂PO₄, 50 µM CaCl₂, pH 6.0). Infected cells were detached in HL5 containing 0.2% azide by vigorously pipetting. Cells were then centrifuged at 3000 g for 1 min and resuspended in 0.5 ml of HL5 medium containing 0.5% paraformaldehyde.

DH1-10 and acapA⁻ cells grown in HL5 were fixed for 3 h in 0.1 M sodium cacodylate buffer (pH 7.3) containing 2% glutaraldehyde and 0.3% osmium tetroxide and then processed for transmission electron microscopy as previously described (Paquet et al., 2013). Samples were examined using a transmission electron microscope (JEOL 1230) at 80 kV.

Antibodies against protein disulfide isomerase (221-64-1, mouse monoclonal), p80 (H161 mouse monoclonal), cathepsin D and vacuolar H⁺-ATPase (221-35-2, mouse monoclonal) were as described previously (Journet et al., 1999; Monnat et al., 1997; Neuhaus et al., 1998; Ravanel et al., 2001). Actin was labeled using tetramethylrhodamine B isothiocyanate (TRITC)-labeled phalloidin (Sigma-Aldrich, St Quentin Fallavier, France). For immunofluorescence analysis, cells were applied on glass coverslips for 3 h, then fixed with 4% paraformaldehyde for 30 min, washed and permeabilized with 0.1% Triton X-100 for 10 min. Cells were incubated with the indicated antibodies for 1 h, and then stained with appropriate fluorescent (Alexa Fluor 488, 568 or 633) secondary antibodies (Molecular Probes, Invitrogen, Eugene, OR) for 30 min. When required, anti-p80 monoclonal antibody directly coupled to Alexa Fluor 633 was used. Cells were observed by laser scanning confocal microscopy (Zeiss LSM 510). The quantification of the number and diameter of lysosomes and post-lysosomes, and kinetics of lysosome maturation (pH neutralization) were carried out as already described (Charette and Cosson, 2007). The visualization of neutral post-lysosomes was performed using dextran coupled to Oregon Green and TRITC-conjugated dextran (pH-sensitive and pH-insensitive dextran, respectively) as previously described (Jenne et al., 1998; Rivero and Maniak, 2006). Briefly, cells were seeded onto glass-bottomed dishes and incubated in LoFlo medium for 1 h. Cells were then pulsed with LoFlo medium containing 4 mg/ml Oregon-Green–dextran and 40 mg/ml TRITC–dextran for 15 min. Cells were washed twice, incubated in fresh LoFlo medium for the indicated times, and observed by confocal microscopy. For immunofluorescence analysis of cells infected with L. pneumophila, D. discoideum cells were seeded on coverslips, incubated overnight, and then infected with fluorescent L. pneumophila (MOI = 100). After centrifugation (880 g for 10 min) to initiate cell–bacterium contact, cells were incubated at 25°C for the indicated times. Coverslips were then processed for immunofluorescence as described above.

The ability of D. discoideum strains to kill internalized Klebsiella bacteria was tested as described previously (Benghezal et al., 2006). D. discoideum growth on bacteria lawn was tested as reported (Froquet et al., 2009). Chemotaxis was assayed using a semi-quantitative assay (Wallace and Frazier, 1979). Briefly, cells were first starved in phosphate buffer for 6 h, and then 1 µl of 2.5×10⁸ cells/ml was deposited on a phosphate agar (1.6%) plate, 4 mm away from a trough filled with 250 µM folate. After 4 h at 22°C, migration distances were calculated by measuring displacement of cell front. Glycosidase activities (N-acetyl β-glucosaminidase and α-mannosidase) and cathepsin D content in cell pellet and culture medium after 3 days of culture were assessed as previously described (Froquet et al., 2008; Le Coadic et al., 2013). For enzymatic activities and western immunoblot analysis, equal amounts of protein were analyzed. Phagosomal proteolysis of DQ-BSA-labeled beads was determined as reported previously (Gopaldass et al., 2012; Le Coadic et al., 2013). After bead uptake, the cell suspension was analyzed by flow cytometry to determine the intensity of the DQ-green fluorescent marker in cells containing phagocytosed beads (Le Coadic et al., 2013).

Monitoring of L. pneumophila intracellular growth in D. discoideum was adapted from an assay described previously (Hervet et al., 2011). L. pneumophila bacteria harboring a plasmid expressing mCherry were grown on BYE liquid medium containing 2 mM IPTG and chloramphenicol for 4 days at 30°C. D. discoideum cells were seeded in 96-well microplates (10⁵ cells/well) and then infected with 10⁶ fluorescent L. pneumophila (MOI = 10). Infection was synchronized by centrifuging plates at 880 g for 10 min. Plates were further incubated for 1 h at 25°C and then washed three times. Intracellular growth was automatically monitored by measuring the fluorescence of mCherry at an excitation of 580 nm and an emission of 620 nm in an Infinite 200 Pro plate reader every hour for 144 h (TECAN, Männedorf, Switzerland).

We thank Virginie Molle (UMR5235, Montpellier, France) for hosting this project in her laboratory. We also wish to thank Richard Janvier (Electron microscopy platform of Institut de Biologie Intégrative et des Systèmes) for his technical assistance.