WASH-driven actin polymerization is required for efficient mycobacterial phagosome maturation arrest

Authors


Summary

Pathogenic mycobacteria survive in phagocytic host cells primarily as a result of their ability to prevent fusion of their vacuole with lysosomes, thereby avoiding a bactericidal environment. The molecular mechanisms to establish and maintain this replication compartment are not well understood. By combining molecular and microscopical approaches we show here that after phagocytosis the actin nucleation-promoting factor WASH associates and generates F-actin on the mycobacterial vacuole. Disruption of WASH or depolymerization of F-actin leads to the accumulation of the proton-pumping V-ATPase around the mycobacterial vacuole, its acidification and reduces the viability of intracellular mycobacteria. This effect is observed for M. marinum in the model phagocyte Dictyostelium but also for M. marinum and M. tuberculosis in mammalian phagocytes. This demonstrates an evolutionarily conserved mechanism by which pathogenic mycobacteria subvert the actin-polymerization activity of WASH to prevent phagosome acidification and maturation, as a prerequisite to generate and maintain a replicative niche.

Introduction

The genus Mycobacterium comprises important pathogens, such as Mycobacterium tuberculosis, with a global impact on human health. M. tuberculosis is restricted to the human host; however, it shares essential virulence properties with other mycobacterial species. Among these is the capability to survive within phagocytes and to modify the intracellular trafficking of host cells. To understand the intracellular course of mycobacterial infection and its underlying molecular mechanisms, cellular models have to be applied in a comparative manner.

To enter a host cell, pathogenic mycobacteria exploit phagocytosis, an evolutionarily highly conserved process by which cells engulf particles or microbes. In the amoeba Dictyostelium, phagocytosis allows grazing on bacteria. In higher multicellular organisms, professional phagocytic cells kill and eliminate invading bacteria as part of the host defence system (Flannagan et al., 2009). After formation, the phagosome follows a defined sequence of fusion and fission events referred to as phagosomal maturation (Fairn and Grinstein, 2012). A hallmark of this progression is the delivery of the proton pumping V-ATPase to the phagosomal membrane, which lowers its luminal pH. Subsequently, the phagosome, which has already acquired endosomal hydrolases and toxic oxidative compounds, matures into an acidic and bactericidal compartment equipped to kill and digest engulfed bacteria (Haas, 2007).

Pathogenic mycobacteria are able to survive phagocytosis through interference with intracellular trafficking and fusion events (Jordao and Vieira, 2011). A range of pathogen effectors and host target molecules are involved in this process. For example, pathogenic mycobacteria can modulate Rab-GTPases and their effectors (Seto et al., 2011), conversion of phosphoinositides, especially PI(3)P (Vergne et al., 2003), as well as calcium fluxes (Vergne et al., 2004; 2005). It has also been proposed that mycobacterial phagosome maturation is brought to a halt by recruitment of the actin-binding protein coronin-1 and its activation of calcineurin (Jayachandran et al., 2007). In Dictyostelium lacking coronin (corA–), pathogenic M. marinum were shown to replicate better for several days (Solomon et al., 2003). One of the first and most critical steps for mycobacterial phagosome maturation arrest is the exclusion of the V-ATPase from the newly formed mycobacterial vacuole to diminish acidification (Sturgill-Koszycki et al., 1994). One player in this exclusion process is the phosphatase ptpA, which is secreted by M. tuberculosis and directly interacts with the H-subunit of the V-ATPase in mammalian cells (Wong et al., 2011).

The cytoskeletal component actin is a major regulator in the endocytic pathway involved in both, the fusion and fission of endosomes and phagosomes (Jahraus et al., 2001; Kjeken et al., 2004; Liebl and Griffiths, 2009; Marion et al., 2011; Gopaldass et al., 2012). Mycobacteria, like many other intracellular pathogens have been shown to interfere with the interactions of their enclosing membrane and the host cell actin cytoskeleton (Defacque et al., 2000; Anes et al., 2003). In addition, Guerin and de Chastellier have shown (Guerin and de Chastellier, 2000b), that the overall organization of the cellular actin is altered in M. avium infected macrophages. These authors suggested that actin might contribute to the limited fusion and fission events that occur at the M. avium vacuole (Guerin and de Chastellier, 2000a).

The regulation of actin nucleation is complex and dynamic. The multi-subunit complex, Wiskott-Aldrich syndrome protein and SCAR Homolog (WASH) drives actin polymerization on endosomal membranes by activating the Arp2/3 complex (Derivery et al., 2009; Gomez and Billadeau, 2009; Duleh and Welch, 2010). Through many interactions, this complex links the actin cytoskeleton to local membrane dynamics by forming functional membrane domains (Duleh and Welch, 2010; Derivery et al., 2012). Most relevant to the present study, the WASH-complex has been shown to be required for recycling of the V-ATPase from maturing phagosomes (Carnell et al., 2011; Park et al., 2013). In WASH deficient Dictyostelium cells, the V-ATPase accumulates on phagosomes, which cannot neutralize and do not mature further to post-lysosomes (Carnell et al., 2011).

To address whether WASH and actin-dependent recycling of the V-ATPase play a role in phagosomal maturation arrest induced by mycobacteria, we initially made use of the established model system Dictyostelium combined with M. marinum (Solomon et al., 2003; Hagedorn and Soldati, 2007; Hagedorn et al., 2009), a close relative of M. tuberculosis (Tobin and Ramakrishnan, 2008). Dictyostelium can be easily genetically manipulated and has long been used as a single-cell organism to shed light on basic cellular processes, such as phagocytosis and membrane transport (Muller-Taubenberger et al., 2012) including a more recent focus on mycobacterial trafficking (Solomon et al., 2003; Hagedorn and Soldati, 2007; Hagedorn et al., 2009). The M. marinum infection in Dictyostelium follows a course similar to the infection of mammalian cells (Hagedorn and Soldati, 2007; Hagedorn et al., 2009). M. marinum diminishes the accumulation of the V-ATPase at its vacuolar membrane and remodels it into a compartment which allows mycobacterial replication. M. marinum subsequently escapes from the vacuole and enters the host cell cytosol at later stages in infection (Hagedorn and Soldati, 2007; Hagedorn et al., 2009). In contrast to M. marinum, our knowledge on the intracellular behaviour of M. tuberculosis is limited. M. tuberculosis seems to interfere with phagosome maturation but can also survive in acidified compartments (Vandal et al., 2008; 2009; Abramovitch et al., 2011). At later time points of infection in human host cells, M. tuberculosis also seems to be able to escape from the phagosome into the cytosol (van der Wel et al., 2007; Houben et al., 2012; Simeone et al., 2012).

In this study we show that F-actin is deposited on the mycobacterial vacuole at an initial stage of infection in a WASH-dependent manner. Disruption of WASH-activity or depolymerization of actin associated with the mycobacterial compartment results in partial reversion of phagosome maturation arrest in Dictyostelium. These observations were reproduced in two different murine phagocytic cells infected with M. marinum and M. tuberculosis, respectively, indicating evolutionary conserved mechanisms. We propose that the mycobacterial compartment undergoes a phase, in which the WASH-driven polymerization of actin is essential to maintain a vacuole competent for bacterial replication.

Results

WASH drives actin polymerization on the M. marinum compartment in Dictyostelium at 6 hpi

To investigate a role for actin in the initial phase of infection (Phase 1) (Hagedorn and Soldati, 2007) we imaged the localization of actin in Dictyostelium cells infected with M. marinum by confocal fluorescence microscopy at 6 hours post infection (hpi). At this stage no net-proliferation of M. marinum is yet observed but the mycobacterial vacuole is diverted from the usual phagosomal maturation pathway as indicated by loss of the V-ATPase and exclusion of cathepsin D (Hagedorn and Soldati, 2007). In order to follow the composition and morphology of the M. marinum vacuole we monitored p80 in addition to F-actin. P80 is a putative endosomal copper transporter (Ravanel et al., 2001), which is delivered to phagosomes and enriched towards later phagosomal stages. From early time-points onwards, p80 associates with the mycobacterial vacuole and has been previously used to follow the mycobacterial vacuole over the complete course of infection (Hagedorn and Soldati, 2007).

At 6 hpi we found that F-actin was present on 23% of vacuoles containing fluorescent M. marinum (Fig.  1A and B). In comparison to the prominent signal at the cell cortex, the presence of F-actin on the vacuoles was less pronounced and showed a patchy distribution (Fig. 1A). This was very different from the thick coat of actin that is present on phagosomes during the internalization of budded yeast (Clarke et al., 2010a) and Salmonella (Meresse et al., 2001; Odendall et al., 2012) or late endosomes (Drengk et al., 2003). This actin coat also differed in appearance from the accumulation of F-actin on the poles of M. avium vacuoles and the phagosomes containing killed M. tuberculosis (Anes et al., 2003). During the later stages of infection actin association was lost from most vacuoles [24 hpi, Phase 2 (Hagedorn and Soldati, 2007)], suggesting that it was a transient and stage specific event (Fig. 1B).

Figure 1.

WASH dependant actin on the mycobacteria compartment (6 hpi).

A. Confocal fluorescence analysis of fixed Dictyostelium cells infected with M. marinum (red) shows F-actin (green) around the mycobacteria vacuole (arrowheads) at 6 hpi.

B. Actin association is significantly reduced (**P ≤ 0.01, n = 3, mean ± SEM) at 24 hpi compared with 6 hpi.

C. Confocal immunofluorescence micrograph of GFP-WASH (right panel) and F-actin (centre panel) accumulation on the M. marinum (blue) vacuole at 6 hpi (left panel, Merge).

D. The accumulation of p80 (red) and GFP-WASH (green) on the M. marinum (blue) vacuole.

E. Quantifications of GFP-WASH accumulation on the vacuole over time shows an increase up to 6 hpi (n = 4, mean ± SEM).

F. In wshA cells no F-actin (green) is observed around the M. marinum (red) vacuole. Bars show 3 μm (A-F).

G. The colocalization of coronin-GFP (green) and F-actin (red) at the 6 hpi M. marinum (blue) vacuole. Bar shows 2 μm (G).

The actin coat on the vacuole appeared remarkably similar to that described by Carnell et al. (2011) for WASH-driven F-actin polymerization. Immunofluorescence analysis of cells expressing GFP-WASH showed that WASH and F-actin (Fig. 1C), as well as the Arp2/3–complex subunit ArpC4 and F-actin (Fig. S1), colocalized on the mycobacterial vacuole. Quantifications of cells showed that GFP-WASH associated with the vacuole over the first 6 hpi, then remained stable for up to 24 hpi (Fig. 1E). Similar to the actin coat WASH accumulated in a punctate pattern on the vacuole (Fig. 1D). It should be noted that the expression of GFP-WASH did not significantly change the appearance or number of actin associated with the M. marinum vacuole (Fig. S2A). To prove a functional link between WASH and actin on the mycobacterial compartment, Dictyostelium mutant cells lacking functional WASH (wshA–) (Carnell et al., 2011) were infected. Quantitative fluorescence microscopy revealed that neither at 6 nor 24 hpi vacuoles were associated with F-actin (Fig. 1F, Fig. S2B). WASH is therefore recruited to the early mycobacterial compartment and required for subsequent actin association.

The actin binding protein coronin has been previously reported to be present on the M. marinum vacuole in Dictyostelium (Solomon et al., 2003). We therefore monitored, whether coronin-GFP was present on M. marinum vacuoles at 6 hpi. Immunofluorescence microscopy of infected cells showed that coronin-GFP localized to M. marinum vacuoles that were associated with F-actin (Fig. 1G).

WASH-driven actin polymerization is required for efficient phagosome maturation arrest

To test whether WASH-mediated actin nucleation is instrumental for the biogenesis or maintenance of the mycobacterial replication niche, we analysed the mycobacterial vacuole in wshA– cells. P80 and VatA were visualized at 6 hpi in wshA– and wild-type cells by immunofluorescence and quantified for their association with the mycobacterial vacuole. In contrast to non-pathogenic mycobacteria the V-ATPase has been shown to be significantly less associated with M. marinum vacuoles and proposed to be only transiently present during Phase 1 (Hagedorn and Soldati, 2007). Mycobacterial vacuoles were either fully associated with VatA (Fig.  2A and A') or completely free of any VatA-specific signal (Fig. 2A and 2A''). Most importantly, at 6 hpi the vacuoles were either exclusively associated with actin or VatA, but never with both (Fig. S3). In comparison to wild-type cells, the p80 decoration of wshA– cells showed a different distribution. Examples of mycobacteria scored as either p80-associated or non-associated are given in Fig. 2B and C respectively. Partial decoration not fully enclosing the bacteria was scored as associated. However, when WASH was disrupted, at 6 hpi the V-ATPase was present on significantly more mycobacterial vacuoles, which concomitantly lost p80 (Fig. 2D). This indicates that WASH activity at the mycobacterial vacuole participates in the exclusion of the V-ATPase.

Figure 2.

Loss of actin association changes the M. marinum vacuole.

A. Representative fluorescence micrographs of M. marinum (red, asterisk in A' and A'') with VatA (A', green) or without (A'') at 6 hpi.

B and C. M. marinum (red) were scored as associated (B) or non-associated (C) with p80 (green) by immunofluorescence microscopy; indicated by arrowheads.

D. wshA cells were infected with M. marinum and association with p80 (n = 3) and VatA (n = 4) was quantified at 6 hpi.

E. Immunofluorescence micrograph of Dictyostelium expressing VMC (red, anti-myc staining; right panel) infected with M. marinum (blue). No actin (green, centre panel) was found at the vacuole, which was decorated with VMC (arrowhead, right panel).

F. VMC cells were infected with M. marinum and association with p80 and VatA was quantified by immunofluorescence microscopy at 6 hpi (n = 4) (***P ≤ 0.001; **P ≤ 0.01, mean ± SEM). Bars show 2 μm.

Removal of F-actin from the established M. marinum vacuole overcomes phagosome maturation arrest

The phenotype of wshA– cells suggests that WASH-driven actin polymerization at the mycobacterial vacuole controls its biogenesis. We therefore investigated whether the removal of F-actin from the replication compartment established at 6 hpi affects maintenance of the vacuole. To achieve this, we made use of the observation that flotillin-like vacuolin accumulates at the mycobacterial compartment from 6 hpi onwards (Hagedorn and Soldati, 2007). By expressing a fusion construct between vacuolin and the actin-severing cofilin (VMC) (Drengk et al., 2003), we were able to locally disrupt the actin coat on vacuolin-positive vesicles. Subsequently we infected VMC expressing Dictyostelium cells with M. marinum and monitored VMC localization. At 6 hpi we observed accumulation of VMC on the M. marinum vacuole (Fig. 2E, right panel) and exclusion of F-actin (Fig. 2E, centre panel; Fig. S4). Infected VMC expressing cells were analysed at 6 hpi by quantitative immunofluorescence microscopy. Consistent with our observations in wshA– cells, VMC expressing cells showed significantly more V-ATPase and less p80 on the mycobacterial vacuole at 6 hpi than wild-type ones (Fig. 2F).

In a secondary approach we made use of the actin-depolymerizing drug Latrunculin A (LatA) to reversibly remove F-actin (Morton et al., 2000) at any given stage during infection. To investigate whether the phenotype observed in the wshA– and VMC cells was specific for the Phase 1 of infection, we compared infected cells, which were untreated, with those treated with LatA (5 μM, up to 2 h) at 6 (Phase 1) and 24 hpi (Phase 2). Quantification of vacuolar marker molecules at 6 hpi revealed that exposure to LatA continuously decreased association of p80 with an accumulation of the V-ATPase at the M. marinum vacuole over time (Fig.  3A, left graph). At 24 hpi however, LatA exposure did neither affect association of p80 nor VatA (Fig. 3A, right graph). Intriguingly, overall these quantifications indicate gradual changes over time in appearance and composition of the M. marinum vacuole rather than a sudden change.

Figure 3.

LatA treatment at 6 hpi leads to gradual accumulation of VatA at the M. marinum vacuole.

A. Dictyostelium cells were infected with M. marinum and treated with LatA (5 μM) at 6 and 24 hpi. The association of the mycobacterial vacuole with p80 and VatA was quantified and plotted over the time indicated. The percentage of vacuoles with the respective marker is shown (n = 4, mean ± SEM) with exposure to LatA and solvent only (Control).

B. Live fluorescence imaging of Dictyostelium expressing VatM-GFP (green) infected with M. marinum (red) before and after the addition of LatA at 6 hpi. Time before the addition of LatA is shown in the upper right corner with a minus. The minutes the cell is exposed to LatA is indicated with a plus (3 frames min−1). VatM-GFP positive vesicles (arrowheads) deliver the V-ATPase to the mycobacterial vacuole. Bar represents 2 μm.

C. Dictyostelium cells were infected with wild-type M. marinum, avirulent M. marinum L1D and non-pathogenic M. smegmatis and treated with LatA (5 μM, 2 h) at 6 hpi. Quantification of p80 association by immunofluorescence microscopy shows that only the pathogen–vacuole exhibits a significant reduction (n = 3, **P ≤ 0.001, mean ± SEM). Control was exposure to solvent alone.

The delivery of VatM-GFP to the mycobacterial compartment is a fast, synchronous event

To determine whether the gradual changes in VatA association with the mycobacterial vacuole were due to the averaging over time, the effect of LatA treatment on VatA-distribution was followed at the single-cell level by live cell fluorescence microscopy. In Dictyostelium, the transmembrane subunit of the proton pumping V-ATPase, VatM, fused to GFP (Liu and Clarke, 1996) localizes correctly to endolysosomal membranes and assembles into the V-ATPase enzyme complex (Clarke et al., 2002). Therefore we monitored M. marinum vacuoles before and after LatA treatment in infected cells expressing VatM-GFP. Within 30 min of LatA addition, accumulation of VatM at mycobacterial vacuoles was frequently observed. As shown in Fig. 3B, upon addition of LatA the mycobacterial vacuole was surrounded by a cloud of small VatM-positive vesicles (+9:40 and +10:00 min), which fused within 20 s (+10:20 min). The V-ATPase decorated the complete surface of the vacuole (+10:40 and +12:00 min). This sequence of events resembled V-ATPase delivery to newly formed phagosomes (Clarke et al., 2002; Clarke et al., 2010b).

These observations suggest that upon actin depolymerizing, the V-ATPase is delivered to the mycobacterial vacuole in an asynchronous manner. It should also be noted, that in Dictyostelium the V-ATPase may not only be derived from the fusion with lysosomes but also from the contractile vacuole (CV) (Heuser et al., 1993; Clarke et al., 2002), the osmoregulatory organelle of amoeba (Gabriel et al., 1999). However, using live microscopy, we neither observed association between both compartments nor did we find accumulation of the CV-marker calmodulin on mycobacterial vacuoles in LatA-treated cells (Fig. S5).

Because p80 is present on endosomes carrying the V-ATPase it should concomitantly be delivered during fusion with lysosomal endosomes (Ravanel et al., 2001). Intriguingly we found, that, in contrast to phagosomes containing latex beads or yeast, association of p80 with mycobacterial vacuoles decreased when actin was depolymerized. Furthermore, despite that live-imaging of single cells suggested occurrence of fusion events with endosomes, the mycobacterial vacuole did not increase in size. Taken together, these data suggest fission of p80-positive vesicles from the mycobacterial vacuole. It also demonstrates that, even under actin depolymerizing conditions, dynamic membrane trafficking events occur at the mycobacterial vacuole.

Actin-depolymerization does not alter p80 abundance on vacuoles containing non-pathogenic and avirulent mycobacteria

We wanted to know whether vacuoles carrying other mycobacterial species were also sensitive to LatA as found for M. marinum. During maturation of phagosomes carrying latex beads or non-pathogenic mycobacteria the V-ATPase is delivered to the phagosome, which coincides with an increase in p80. Therefore, instead of following the V-ATPase we used p80 association as an indirect read-out for alterations caused by LatA treatment in Dictyostelium infected with non-pathogenic M. smegmatis (Hagedorn and Soldati, 2007) and the avirulent mutant M. marinum L1D (Ramakrishnan et al., 2000; Hagedorn and Soldati, 2007). Both strains are transported along the classic phagosomal maturation pathway in Dictyostelium (Hagedorn and Soldati, 2007). At 6 h after uptake, the cells were treated with LatA, fixed and the association with p80 was compared between all mycobacterial strains. In comparison to wild-type M. marinum vacuoles, M. smegmatis and M. marinum L1D vacuoles maintained p80 association (Fig. 3C) demonstrating that the F-actin dependent stage is specific to the M. marinum vacuole.

Actin is required to prevent acidification and bactericidal activity of the M. marinum vacuole

The accumulation of both, VatA and VatM, on the mycobacterial phagosome suggests that the V-ATPase enzyme complex is fully assembled and functional (Clarke et al., 2002). We therefore tested whether depolymerization of actin promotes acidification of M. marinum vacuoles. Infected Dictyostelium cells (6 hpi) were treated with LatA for 60 min and subsequently loaded with the acidophilic dye neutral red (NR) (Clarke and Maddera, 2006). The cells were monitored by fluorescence microscopy and accumulation of NR in mycobacteria-compartments was quantified (Fig. 4A, micrographs). The results show that in contrast to control cells, significantly more M. marinum were found in acidified compartments (Fig. 4B) after treatment with LatA. The increase in NR colocalization (1.6-fold) was comparable to the increase observed in VatA-association with the vacuole after LatA treatment (Fig. 3A, 1.6-fold). Assessing the colony-forming units (cfu) over time after a 2 h pulse of LatA at 6 hpi revealed a significant decrease in live bacteria between 32 and 48 hpi in LatA treated cells compared to untreated cells (Fig. 4C and D). It should be noted that LatA itself had no impact on mycobacterial growth in the absence of host cells (Fig. S6). In addition, growth of M. marinum is limited in wshA– cells in comparison to wild-type cells as determined by FACS-analysis (Fig. S7A). In addition, the infection cycle appears disrupted because ejectosome structures, which allow cell-to-cell transmission of mycobacteria (Hagedorn et al., 2009), were not observed in the absence of WASH (Fig. S7B). Therefore, removal of F-actin associated during phase 1 leads to acidification and renders the mycobacterial vacuole bactericidal.

Figure 4.

LatA treatment generates a bactericidal environment. (A) Infected Dictyostelium (phase contrast: upper image, fluorescence micrograph: lower image: M. marinum, red) were treated with LatA or solvent (control) at 6 hpi and loaded with NR (green). Bacteria were scored as being associated (indicated by white arrowheads) or non-associated with NR (asterisk) and (B) the number plotted as a bar graph. In treated cells significantly more M. marinum colocalize with NR (n = 3, mean ± SEM, ***P ≤ 0.001).

C. Dictyostelium cells infected with M. marinum were treated with LatA or solvent (Control) at 6 hpi. LatA was removed after 2 h and cells maintained in culture. Up to 48 hpi samples were taken and live bacteria determined by counting cfu. Normalized numbers are shown (n = 5, mean ± SEM). At 48 hpi treated cells show a significant reduction in the number of live bacteria (*P ≤ 0.05).

D. The absolute number of cfu at 48 hpi. Bar represents 2 μm.

WASH and F-actin colocalize with M. tuberculosis and M. marinum vacuoles in mammalian phagocytes

It was important to investigate whether the phenotype observed for M. marinum in Dictyostelium was also true for M. marinum as well as the human pathogen M. tuberculosis in mammalian phagocytes. We therefore infected the murine microglial cell line BV2 with M. marinum and RAW 264.7 macrophages with M. tuberculosis. At 2 hpi, infected cells were fixed and association of WASH and F-actin to mycobacterial vacuoles was assessed (Fig. S8). Consistent with our Dictyostelium data, we frequently observed the accumulation of WASH and F-actin on mycobacterial vacuoles often in a very tight apposition to the mycobacteria (Fig. 5A and B). However, at 4 hpi, association of F-actin with these vacuoles was reduced (Fig. S8). In mammalian cells, F-actin was present on mycobacterial vacuoles in similar amounts as observed in Dictyostelium and showed a punctate pattern interspersed and partially overlapping with WASH (Fig. 5A and B).

Figure 5.

WASH and F-actin colocalize to M. marinum and M. tuberculosis vacuoles at 2 hpi. Confocal fluorescence micrograph of BV2 cells infected with M. marinum (A) and RAW 264.7 cells infected with M. tuberculosis (B). At 2 hpi infected cells were stained against F-actin (red) and WASH (green; immunofluorescence labelling), which localize in puncta on the vacuole (indicated by arrowheads) on both, the M. marinum (A) and M. tuberculosis vacuole (B). Bars represent 3 μm.

Actin also prevents acidification of M. tuberculosis and M. marinum vacuoles in mammalian phagocytes

To determine whether LatA treatment also drives M. marinum and M. tuberculosis into acidified lysosome-like compartments, we studied association of the mycobacteria with lysotracker, an acidophilic dye similar to NR. Accumulation of lysotracker in mycobacterial vacuoles was quantified by fluorescence microscopy before and after treatment with LatA (10 μM, 30 min). In addition, we monitored the accumulation of lamp-1 on M. tuberculosis vacuoles by immunofluorescence.

Upon LatA treatment of BV2 and RAW 264.7 the numbers of mycobacteria in acidified vacuoles were significantly increased at 2 hpi (Fig. 6A–C) as observed for both mycobacterial species. In contrast, at 4 hpi LatA had no effect on lysotracker association with mycobacteria (Fig. 6B and C). Lamp-1 association with M. tuberculosis was also significantly enhanced at 2 but not 4 hpi (Fig. 6D and E). Taken together this confirms that also in mammalian, cells actin association prevents acidification of mycobacterial vacuoles and, for M. tuberculosis, stimulates the delivery of lysosomal protein lamp-1.

Figure 6.

Treatment with LatA leads to accumulation of lysotracker in M. marinum and M. tuberculosis vacuoles at 2 hpi. BV2 (A, B) and RAW 264.7 cells (C) cells were infected with M. marinum (A, B) and M. tuberculosis (C) respectively. At 2 and 4 hpi the cells were treated with LatA or solvent (control). Twenty minutes before quantification the cells were loaded with lysotracker and colocalization with the bacteria was analysed in fixed cells. (A) shows a representative fluorescence micrograph of lysotracker (green) association. A BV2 cell (grey channel) infected with M. marinum (red) shows a bacterium which associates (green, inset indicated by white arrowhead) and another one, which is free of colocalization (inset, white arrow). Quantification showed that colocalization with lysotracker for both, M. marinum (n = 6) (B) and M. tuberculosis (n = 3, mean ± SEM) (C) increased significantly at 2 hpi (***P ≤ 0.001). No difference is found at 4 hpi (n = 3, mean ± SEM, n.s., not significant).

D. Fluorescence micrograph of an infected cell (RAW 264.7, grey; M. tuberculosis, red) at 2 hpi. Lamp-1 is shown in green. Bacteria not associated with lamp-1 are indicated with an arrow, lamp-1 on the M. tuberculosis vacuole is shown in the inset (arrowhead).

E. Association of lamp-1 on vacuoles was quantified. LatA-treated cells at 2 hpi showed a significant increase in colocalization (n = 3, mean ± SEM, **P ≤ 0.01) in contrast to 4 hpi, where treatment had no significant impact on the association with lamp-1 (n = 3, mean ± SEM, n.s., not significant). Bars represent 3 μm.

Discussion

Phagosome maturation is a highly dynamic and well-defined program of fusion and fission events (Fairn and Grinstein, 2012). While pathogenic mycobacteria can arrest this process early on, their vacuole still remains connected to the endocytic pathway (Sturgill-Koszycki et al., 1996). In the present study we show that WASH-driven F-actin polymerization on the mycobacterial vacuole protects mycobacteria from delivery to a bactericidal environment. Importantly, the protective function of this F-actin was determined by the stage of the mycobacterial vacuole. In both, Dictyostelium and murine phagocyte cell lines, depolymerization of actin by LatA enhanced the delivery of lysosomal markers to the mycobacterial vacuole only at certain time points of infection, i.e. 6 and 2 hpi respectively. At later stages (24 and 4 hpi respectively) LatA treatment had no effect. This suggests that the mycobacterial compartment undergoes a phase in which accumulation of F-actin is essential for the maintenance of a vacuole at an arrested stage.

In contrast to our observations at the 2 hpi time point, M. avium-containing vacuoles isolated from murine macrophages at 24 hpi reveal reduced capacity to nucleate actin polymerization (Anes et al., 2003), which seems to be different from the early F-actin dependent phase. These authors also observed strong actin association with phagosomes containing killed M. tuberculosis but not live bacteria. Actin has been shown to accumulate strongly on mature phagosomes (Defacque et al., 2000; Drengk et al., 2003; Kjeken et al., 2004) containing latex beads. The early (6 and 2 hpi respectively) F-actin associations observed in this study in Dictyostelium and mouse macrophage-like cells infected with M. marinum or M. tuberculosis respectively show a much weaker accumulation and a rather patchy localization without restriction to the poles.

In addition to WASH and F-actin, we also monitored accumulation of the actin-binding protein coronin on the M. marinum vacuole in Dictyostelium at 6 and 24 hpi (Fig. 1G and Fig. S9 respectively). In mammalian cells coronin-1 (also known as TACO, Tryptophan Aspartate containing Coat protein) has been shown to be present on M. bovis BCG vacuoles and to play a role in the survival of these attenuated mycobacteria (Ferrari et al., 1999). More recently, coronin-1 has been shown to inhibit lysosomal fusion with M. bovis BCG vacuoles in mammalian cells by activating the phosphatase calcineurin (Jayachandran et al., 2007). In liver cryosections from coronin-1 knock-out but not wild-type mice (1 day post infection) these authors observed a significant increase in the numbers of virulent M. tuberculosis present in lamp-1-positive compartments (Jayachandran et al., 2007). Our quantitative immunofluorescence microscopy in corA– Dictyostelium cells (Fig. S10A) shows a transient, but significantly enhanced accumulation of VatA at 6 hpi on the M. marinum vacuole. Intriguingly, when determining the number of live M. marinum over the period of several infection cycles (7 days), Solomon et al. (2003) observed an increased replication in Dictyostelium corA– cells.

While the lack of either WASH and coronin in Dictyostelium enhanced VatA levels on M. marinum vacuoles at early time points of infection (6 hpi), association with VatA in the corA– mutant returned to wild-type levels at 24 hpi without a change in p80 association (Fig. S10). The presence of WASH and coronin on M. marinum vacuoles in Dictyostelium at 6 hpi suggests a function of both proteins in regulating F-actin dynamics on mycobacterial vacuoles by initiating polymerization of and interactions with F-actin. Thereby interference with V-ATPase-accumulation in the early phase of infection and promotes mycobacterial survival.

We therefore conclude from our studies that F-actin nucleation on mycobacterial vacuoles is distinctly regulated over the course of infection: actin polymerization is required to maintain phagosomal arrest early in infection but is suppressed at later stages when the vacuole has progressed to an established vacuole competent for mycobacterial replication in a stable physiological environment.

What is the precise function of actin and WASH on the replication compartment? One possibility is that the WASH-dependent actin coat may simply block fusion with lysosomes (depicted in Fig. 7), as has been proposed for actin cage-like structures of Salmonella vacuoles (Meresse et al., 2001). Alternatively, subunits of the V-ATPase complex have been shown to directly and indirectly interact with filamentous actin (Vitavska et al., 2005). This interaction has been proposed to control V-ATPase localization. Accordingly WASH, and subsequently F-actin have been shown to promote recycling of the V-ATPase from maturing phagosomes (Carnell et al., 2011). WASH may therefore be subverted to exert this function at the initial mycobacterial compartment to prevent acidification. In accordance with this model, the V-ATPase has been proposed to only transiently associate with the mycobacterial vacuole (Sturgill-Koszycki et al., 1996; Hagedorn and Soldati, 2007). In Hagedorn and Soldati (2007) it was shown that the V-ATPase is rapidly delivered (15 min) to the vacuole of pathogenic M. marinum but subsequently removed, which could be promoted by WASH-initiated actin patches as described in this study.

Figure 7.

Proposed model for events upon actin depolymerization. In the initial phase (Dictyostelium phase 1) a pathogenic mycobacterium (grey) has successfully arrested phagosome maturation and blocks accumulation of the V-ATPase (orange) either by inhibiting its delivery and/or efficiently recycling the V-ATPase. The arrested phagosome shows the presence of WASH (W), accumulation of actin (green) and the marker p80 (blue). Upon loss of actin (LatA treatment), the vacuole accumulates the V-ATPase either by fusion with lysosomes (efficient delivery) or inhibition of V-ATPase recycling. The compartment subsequently acquires lamp-1, acidifies, accumulates hydrolases and restricts mycobacterial growth. In a later phase (Dictyostelium phase 2) mycobacteria (grey) have proliferated. WASH (W) is present on the vacuole but no association with actin is observed. When cells in this late phase are treated with LatA no change of the vacuole is observed.

Many pathogens are known to manipulate the host cell actin cytoskeleton to establish, maintain and spread to other cells. For example Shigella flexneri induces membrane ruffles on the surface of its host cell in order to be internalized by injecting effectors into the host cell. These effectors modulate signalling cascades, which lead to the remodelling of the actin cytoskeleton (Ehsani et al., 2012). Once escaped into the host cell cytosol, a number of pathogens including Listeria monocytogenes (Tilney and Portnoy, 1989) utilize the Arp2/3 complex to stimulate the assembly of actin filaments on their surface. Tails of actin allow these bacteria to move within the cell and in some cases to pass over into neighbouring cells (e.g. Rajabian et al., 2009). In principal, pathogens apply two strategies to activate the Arp2/3 complex, either by mimicking NPFs on their surface such as L. monocytogences' actA (Pistor et al., 1994; Welch et al., 1997; 1998) or by recruiting host family WASP proteins as shown for S. flexneri (Suzuki et al., 1998) and M. marinum (Stamm et al., 2003; 2005). Recently the M. tuberculosis Rv1626 gene product was identified as an interactor for the Arp2/3 complex (Ghosh et al., 2013).

It is not clear how WASH is maintained on the M. marinum compartment. As further outlined below WASH is localized to the retromer complex via its FAM-21 subunit in mammalian cells (Jia et al., 2012). Further experiments will be necessary to clarify, whether the FAM-21 subunit, retromer or mycobacterial factors are necessary for WASH localization. In addition, the observation that the exclusion of the V-ATPase from the mycobacterial vacuole is partially attributed to the secreted phosphatase ptpA (Wong et al., 2011) suggests that this or additional secreted mycobacterial factors can directly or indirectly interact with the actin-machinery.

It should be noted that, in contrast to Dictyostelium cells, it has been established in mammalian cells that the retromer recruits the WASH complex to endosomes via the WASH subunit FAM-21 (Jia et al., 2012). WASH stimulates actin polymerization on endosomes and generates actin patches, which function as sorting platforms (Puthenveedu et al., 2010) into either the endosome-to-Golgi (Gomez and Billadeau, 2009) and the endosome-to-plasma membrane pathway (Derivery et al., 2009). Our observations in both, Dictyostelium and mammalian cells, suggest a conserved function of the WASH-complex which is executed on the mycobacterial compartment. In both systems, WASH stimulates the Arp2/3 complex to generate actin patches on the vacuole, which leads to the exclusion of components that would turn the vacuolar lumen acidic thereby activating anti-microbial effectors. However, it is intriguing that WASH remains associated with the M. marinum and M. tuberculosis compartment at later stages in Dictyostelium (Fig. 1) and RAW 246.7 macrophages (Fig. S8), respectively, without the presence of F-actin.

Many studies describe that virulence traits of intracellular bacteria have evolved from the early interactions between bacteria with environmental amoeba (Greub and Raoult, 2004; Cosson and Soldati, 2008). Here, we show WASH-driven polymerization of actin on both, the M. marinum and M. tuberculosis phagosomes, which are arrested at an early stage, in both Dictyostelium and murine phagocytes. This F-actin is necessary to shield the bacteria from exposure to acidic and bactericidal conditions. This process is conserved between the amoebal and murine phagocytes and two mycobacterial pathogens, of either cold-blooded animals or humans. Our data indicate that the recruitment of active WASH to mycobacterial vacuoles is an evolutionary conserved and universal virulence strategy of pathogenic mycobacteria to escape microbicidal effectors in mature phagosomes.

Experimental procedures

Cell strains and culture conditions

Wild-type Dictyostelium discoideum (Ax2) was cultured axenically as described (Hagedorn and Soldati, 2007). Dr R. Insall (Beatson Institute, Glasgow, UK) provided the wshA– cells (Carnell et al., 2011) and cells with GFP-WASH (pLP108). Ax2–VMC cells (Drengk et al., 2003) and VatM-GFP were a gift from Dr M. Maniak (Kassel University, Germany) and Dr T. Soldati (University of Geneva, Switzerland) respectively. Dr A. Müller-Taubenberger provided the coronin-GFP expression plasmid (pDdA15coronin-GFP). BV2 mouse microglial cells (gift from Dr T. Soldati) and RAW 264.7 cells were maintained as described (Axelrod et al., 2008).

Mycobacteria strains and culture conditions

Mycobacteria strains were cultured as described (Hagedorn and Soldati, 2007). M. marinum M-strain (wild-type), M. marinum L1D as well as the msp12-GFP plasmid were a kind gift of Dr L. Ramakrishnan (Washington University, Seattle, USA). M. smegmatis was a gift from Dr G. Griffiths (University of Oslo, Norway). The pCHERRY3 plasmid (No. 24659; Carroll et al., 2010) was provided by Addgene (Tanya Parish). M. tuberculosis (H37Rv) GFP was a kind gift from Dr T. Parish (QMUL) and described before (Corleis et al., 2012).

Antibodies and reagents

Monoclonal antibodies against the following antigens were used at a dilution of 1:10 and obtained from: p80 (Ravanel et al., 2001) from Dr P. Cosson (University of Geneva, Switzerland); VatA (Neuhaus et al., 1998) from Dr M. Maniak (Kassel University, Germany), 9E11 anti-myc from Santa Cruz Biotechnology. The following polyclonal antibodies were used: GFP (1:500, MBL); WASH1 (1:100, Sigma, St Louis, USA) and lamp-1 (1:100, Abcam, Cambridge, UK). Anti-calmodulin (1:1000) was a gift from Dr T. Soldati. Secondary antibodies (1:1000) coupled to AlexaFluor 488, 568, 594 or AlexaFluor 647 from Life Technologies (Carlsbad, USA) were used. Actin was visualized using AlexaFluor 568 Phalloidin (1:500) or AlexaFluor 647 Phalloidin (1:200) from Life Technologies. LatA was purchased from Enzo Life Sciences (Farmingdale, USA), lysotracker from Life Technologies and Neutral Red (NR) from Merck (Darmstadt, Germany).

Infection assays

The infection of Dictyostelium and BV2 cells were performed as described in (Hagedorn et al., 2009) at a moi of 10 and 3 respectively. Infection of RAW 264.7 cells with M. tuberculosis was done as described before (Axelrod et al., 2008). For both infection procedures the time of addition of bacteria to the cells was defined as 0 hpi.

Treatment with LatA

At the appropriate time points infected Dictyostelium cells were incubated with LatA (5 μM in HL5c; stock solution 500 μg ml−1 ethanol) for 15, 30, 60 and 120 min. Subsequently, the cells were fixed and processed for immunofluorescence. For BV2 and RAW 264.7 cells, LatA stock solution (1 mg ml−1 DMSO) was added to RPMI and DMEM respectively (final conc. 10 μM). Infected BV2 and RAW 264.7 cells were grown on coverslips up to 2 and 4 hpi. At the appropriate time points the cells were incubated with medium containing LatA for 30 min. When indicated the cells were loaded with 75 nM lysotracker in the appropriate medium for 20 min prior to LatA treatment.

Immunofluorescence staining and microscopy of fixed cells

Cells were fixed in 4% PFA (paraformaldehyde) for 1 h at 4°C. Immunofluorescence and phalloidin labelling were performed as described (Hagedorn et al., 2006). Analysis was done using an Olympus FV1000 confocal microscope and Fluoview software v1.7b. Image data was analysed using ImageJ (http://rsb.info.nih.gov/ij/). All statistical analyses were performed using the Student's t-test (unpaired).

Live cell microscopy

VatM-GFP-expressing Dictyostelium cells infected with mCherry-expressing M. marinum were transferred into a μ-slide with 8 wells (ibidi) at 4 hpi. At 6 hpi recordings were performed using an Olympus IX81 microscope. The imaging was performed for 10 min with time intervals of 20 s at 25°C. Subsequently, LatA (5 μM) or only solvent (control) was added to the medium and cells were imaged for another 35 min with intervals of 20 s. The images were deconvoluted (no neighbour, haze removal of 40%) with the Xcellence rt software, analysed and brightness adjusted using ImageJ (http://rsb.info.nih.gov/ij/).

Colony-forming unit measurements

The cfu measurements were performed as described (Hagedorn and Soldati, 2007). Dictyostelium was infected as described above. At 6 hpi 5 μM LatA was added for 2 h. The drug was taken away, cells were washed once with HL5c medium and 300 μl of cells were collected at 10, 24, 32 and 48 hpi in the medium. The pelleted cells were dissolved and lysed in PBS with 0.1% Triton X-100 for 10 min. Serial dilutions in PBS were plated onto Middlebrook 7H11 plates (Difco) and incubated at 32°C for 5–7 days.

Acknowledgements

We would like to acknowledge Markus Maniak, Thierry Soldati, Robert Insall and dictybase (http://www.dictybase.org) for providing strains, reagents and support. Special thanks go to Jason King for constructs and reading the manuscript.

Financial support came from the Leibniz Association, the DFG (HA3473/3-1) to MH and MK, and from DFG-SPP-1580 and BMBF ‘Medical Infection Genomic’ to US and AG.

Ancillary