The environmental bacterium Legionella pneumophila causes a severe pneumonia termed Legionnaires' disease. L. pneumophila employs a conserved mechanism to replicate within a specific vacuole in macrophages or protozoa such as the social soil amoeba Dictyostelium discoideum. Pathogen–host interactions depend on the Icm/Dot type IV secretion system (T4SS), which translocates approximately 300 different effector proteins into host cells. Here we analyse the effects of L. pneumophila on migration and chemotaxis of amoebae, macrophages or polymorphonuclear neutrophils (PMN). Using under-agarose assays, L. pneumophila inhibited in a dose- and T4SS-dependent manner the migration of D. discoideum towards folate as well as starvation-induced aggregation of the social amoebae. Similarly, L. pneumophila impaired migration of murine RAW 264.7 macrophages towards the cytokines CCL5 and TNFα, or of primary human PMN towards the peptide fMLP respectively. L. pneumophila lacking the T4SS-translocated activator of the small eukaryotic GTPase Ran, Lpg1976/LegG1, hyper-inhibited the migration of D. discoideum, macrophages or PMN. The phenotype was reverted by plasmid-encoded LegG1 to an extent observed for mutant bacteria lacking a functional Icm/Dot T4SS.Similarly, LegG1 promoted random migration of L. pneumophila-infected macrophages and A549 epithelial cells in a Ran-dependent manner, or upon ‘microbial microinjection’ into HeLa cells by a Yersinia strain lacking endogenous effectors. Single-cell tracking and real-time analysis of L. pneumophila-infected phagocytes revealed that the velocity and directionality of the cells were decreased, and cell motility as well as microtubule dynamics was impaired. Taken together, these findings indicate that the L. pneumophila Ran activator LegG1 and consequent microtubule polymerization are implicated in Icm/Dot-dependent inhibition of phagocyte migration.
Legionella pneumophila, the causative agent of a life-threatening pneumonia termed Legionnaires' disease, is a natural parasite of free-living protozoa (Newton et al., 2010; Hilbi et al., 2011a). Upon inhalation, the bacteria employ a seemingly conserved mechanism to replicate in alveolar macrophages, by forming a specific pathogen compartment, the Legionella-containing vacuole (LCV) (Isberg et al., 2009; Hubber and Roy, 2010; Hilbi and Haas, 2012). LCVs communicate with the endosomal pathway but do not fuse with lysosomes, intercept early secretory vesicles emerging from endoplasmic reticulum (ER) exit sites and finally fuse with the ER. Moreover, L. pneumophila inhibits retrograde endosome to trans-Golgi vesicle trafficking to promote intracellular replication (Finsel et al., 2013). The complex process of LCV formation requires the acquisition of a number of small secretory and endosomal GTPases as well as GTPase-modulating factors (Nagai et al., 2002; Kagan et al., 2004; Urwyler et al., 2009; Hoffmann et al., 2014).
L. pneumophila governs the interaction with phagocytes by means of a type IV secretion system (T4SS) termed Icm/Dot, through which ∼300 different ‘effector’ proteins are translocated into host cells (Hubber and Roy, 2010; Gomez-Valero et al., 2011; Zhu et al., 2011). Some Icm/Dot substrates localize to the LCV membrane, where they subvert host phosphoinositide lipids (Hilbi et al., 2011b; Hsu et al., 2012; Toulabi et al., 2013; Weber et al., 2013), or modulate in an intricate manner the activity of small eukaryotic GTPases of the Rab and Ran family, which are implicated in vesicle trafficking (Itzen and Goody, 2011; Hilbi and Haas, 2012) or microtubule dynamics (Urwyler et al., 2009; Rothmeier et al., 2013) respectively. The Icm/Dot-translocated L. pneumophila protein Lpg1976 (alias PieG or LegG1, Legionellaeukaryotic gene) shows homology to the eukaryotic Ran GEF RCC1 (Regulator of chromosome condensation) and is encoded in the Pie (Plasticity island of effectors) gene cluster (de Felipe et al., 2005; 2008; Ninio et al., 2009). LegG1 contains a C-terminal CAAX tetrapeptide motif, which is lipidated (geranylgeranylated) by the host prenylation machinery to target the bacterial protein to host membranes (Ivanov et al., 2010). Recently, LegG1 has been shown to function as a bacterial Ran activator, which together with Ran and its effector RanBP1 (Ran binding protein 1) localizes to the LCV membrane, stabilizes microtubules, enhances LCV motility and promotes intracellular bacterial replication in phagocytes (Rothmeier et al., 2013).
In the course of these studies we observed that not only LCV motility within phagocytes, but also phagocyte motility per se appeared to be controlled by LegG1. This observation, together with the fact that modulation of cell migration is a common theme among pathogenic bacteria utilizing a T3SS (Aepfelbacher et al., 2003; McLaughlin et al., 2009; Konradt et al., 2011; Heymann et al., 2013), prompted us to investigate in detail the modulation of cell migration by L. pneumophila and Icm/Dot-translocated effectors. Here we show that L. pneumophila inhibits in a dose- and T4SS-dependent manner the migration and chemotaxis of D. discoideum amoebae, macrophages and polymorphonuclear neutrophils (PMN). Moreover, the L. pneumophila Ran activator LegG1 promoted the migration of phagocytes and epithelial cells, likely due to stabilization or increased polymerization of microtubules.
Icm/Dot-dependent inhibition of D. discoideum migration and aggregation by L. pneumophila
L. pneumophila translocates via the Icm/Dot T4SS a plethora of effector proteins into phagocytic host cells, where they target different cellular signalling pathways. To address the question, whether L. pneumophila affects migration and chemotaxis of phagocytes, D. discoideum producing the green fluorescent protein GFP was infected for 1 h at different multiplicities of infection (moi) with wild-type L. pneumophila or bacteria lacking a functional Icm/Dot T4SS (ΔicmT), and the infected amoebae were let migrate in an under-agarose assay towards folate for another 4 h (Fig. 1A). Within this time span, the bulk of uninfected D. discoideum amoebae migrated more than 1500 μm towards the chemoattractant, and cells infected with the ΔicmT mutant strain (moi 1–50) migrated to a similar extent. In contrast, the chemotaxis of D. discoideum infected at a low moi of 1 with wild-type L. pneumophila was significantly impaired. The migration of wild-type-infected D. discoideum was dose-dependently inhibited, and at an moi of 50 the migration was completely abolished (Fig. 1B). Under the conditions used, the uptake efficiency of wild-type and ΔicmT mutant L. pneumophila was comparable, and the cytotoxicity of the wild-type strain (6%) was in the low range of the ΔicmT strain (2%) (Fig. S1). Thus, the inhibition of amoebae migration by L. pneumophila is not caused by bacterial cytotoxicity, but seems to be governed by a more specific Icm/Dot-dependent process.
Upon starvation, vegetative D. discoideum cells aggregate in a cAMP-dependent manner and form ‘streams’ and ‘slugs’, which finally develop into stalks and fruiting bodies. To analyse whether L. pneumophila affects early steps in this developmental process, D. discoideum amoebae grown in rich HL-5 medium were placed in nutrient-poor SorC buffer, infected with wild-type or ΔicmT mutant L. pneumophila, washed with SorC and let develop for 12–48 h at 23°C (Fig. 1C). Compared with uninfected starved cells, wild-type but not ΔicmT mutant L. pneumophila retarded the aggregation of D. discoideum amoebae in the course of 12 h, 24 h and 48 h incubation. Also under these conditions, the toxicity of wild-type L. pneumophila was low and comparable with the ΔicmT strain (data not shown). Together, these observations indicate that L. pneumophila inhibits migration and aggregation of D. discoideum in an Icm/Dot-dependent manner.
Icm/Dot-dependent inhibition of macrophage and PMN migration by L. pneumophila
L. pneumophila infects not only free-living protozoa but also mammalian phagocytes. To determine whether the bacteria modulate the migration of leucocytes, murine RAW 264.7 macrophages and primary human PMN were used. RAW 264.7 macrophages stained with CellTracker Green BODIPY were infected at an moi of 10 for 1 h with wild-type or ΔicmT mutant L. pneumophila and let migrate towards the chemokine CCL5 for another 4 h. The bulk of uninfected macrophages, or cells infected at an moi of up to 50 with ΔicmT mutant L. pneumophila, migrated as far as approximately 500 μm towards CCL5 (Fig. 2A and S2) or towards TNFα (Fig. 2B) respectively. In contrast, already at low multiplicities of infection the migration of macrophages infected with wild-type L. pneumophila was abolished. The macrophages phagocytosed wild-type or ΔicmT mutant L. pneumophila with the same efficiency, and over the course of a 5 h infection, the cytotoxicity of either L. pneumophila strain was below 25% (Fig. S2). Finally, the migration of macrophages infected with Legionella longbeachae was inhibited in an Icm/Dot-dependent manner, similar to L. pneumophila-infected cells (Fig. S3).
Freshly isolated human PMN were stained with CellTracker Green BODIPY, infected at an moi of 10 for 1 h with wild-type or ΔicmT mutant L. pneumophila and also let migrate towards a gradient of the peptide formyl-methionyl-leucyl-phenylalanine (fMLP) for another 4 h. Similar to macrophages, PMN infected with ΔicmT mutant L. pneumophila migrated towards fMLP up to a range of 500–750 μm, which is comparable to uninfected cells, and wild-type L. pneumophila abolished cell migration (Fig. 2C). Analogous results were obtained using transwell (Boyden) chambers to analyse the migration of uninfected PMN or cells infected with wild-type or ΔicmT mutant L. pneumophila (Fig. 2D). The uptake efficiency of wild-type and ΔicmT mutant L. pneumophila was the same, and the toxicity was low (5–10%) over the course of a 5 h infection (Fig. S2). In summary, these results indicate that L. pneumophila inhibits leucocyte migration dependent on the Icm/Dot T4SS.
The L. pneumophila Ran activator LegG1 modulates phagocyte migration
Based on the findings that L. pneumophila inhibits the migration and chemotaxis of D. discoideum amoebae, macrophages and PMN in an Icm/Dot-dependent manner, we sought to identify Icm/Dot substrates or regulatory factors that modulate phagocyte migration. To this end, D. discoideum amoebae producing GFP were infected at an moi of 10 for 1 h with L. pneumophila mutants lacking individual Icm/Dot-translocated effectors, components of the Icm/Dot T4SS or constituents of the Legionella quorum sensing (Lqs) system, and the infected amoebae were let migrate in an under-agarose assay towards folate for another 4 h (Fig. S4). L. pneumophila mutant strains lacking the ER interactor SidC or the Rab1 GEF SidM inhibited the migration of amoebae somewhat more strongly than wild-type bacteria, and the effect was reverted upon expression of the effector genes from a plasmid. L. pneumophila lacking the Icm/Dot component IcmG (alias DotF) partially inhibited the migration of amoeba, corresponding to a weak intracellular growth defect due to a partially compromised T4SS produced in this strain (Segal and Shuman, 1999; Albers et al., 2005). Moreover, L. pneumophila lacking constituents of the Lqs system (Tiaden et al., 2007; Kessler et al., 2013) also showed only a mild if any migration phenotype compared with wild-type bacteria.
Interestingly, an L. pneumophila strain lacking the Ran activator LegG1 (ΔlegG1) robustly caused a more efficient inhibition of D. discoideum migration than wild-type bacteria, and this ‘hyper-inhibition’ phenotype was not only complemented but reverted to an extent observed for ΔicmT mutant L. pneumophila (Fig. 3A and B and Fig. S4). Furthermore, the aggregation of D. discoideum infected with the ΔlegG1 strain was also arrested more profoundly than the aggregation of amoebae infected with wild-type L. pneumophila, and the effect was reverted upon expression of legG1 from a plasmid (Fig. S5). The uptake efficiency, cytotoxicity and LCV formation of L. pneumophila wild-type, ΔlegG1 or the complemented mutant strain were determined as controls and found to be the same. Similarly, the cell morphology (amoeboid versus rounded) and transient polarization dynamics [localization of the PI lipid phosphatidylinositol 3,4,5-triphosphate; PtdIns(3,4,5)P3] of amoebae infected with L. pneumophila wild-type, ΔlegG1 or the complemented mutant strain was indistinguishable (Fig. S1). Taken together, in absence of the Ran activator LegG1, Icm/Dot-proficient L. pneumophila obstruct the migration of amoebae completely. Under these conditions microtubule polymerization of L. pneumophila-infected cells is significantly diminished (Rothmeier et al., 2013). Cell migration was restored and even promoted by overexpression of legG1, which triggers Ran activation and microtubule polymerization (see below).
RAW 264.7 macrophages or PMN stained with CellTracker Green BODIPY were also infected for 1 h with L. pneumophila ΔlegG1 or with the mutant strain complemented with legG1 and let migrate in under-agarose assays for another 4 h towards CCL5 (Fig. 3C and D) or fMLP (Fig. 3E and F) respectively. Similar to the amoebae, macrophages or PMN infected with the ΔlegG1 strain caused a ‘hyper-inhibition’ of cell migration compared with wild-type L. pneumophila, and the phenotype was reverted by plasmid-encoded LegG1. The uptake efficiency and cytotoxicity towards macrophages or PMN of L. pneumophila wild-type, ΔlegG1 and the complemented mutant strain was similar (Fig. S2). In summary, in absence of the Ran activator LegG1, Icm/Dot-proficient L. pneumophila inhibited the migration of amoebae, macrophages or PMN even more efficiently than wild-type bacteria, and the phenotype was reverted by providing the legG1 gene on a plasmid, whereas the uptake and cytotoxicity of the mutant strains was not altered. These results suggest that activation of the Ran GTPase by LegG1 and consequent microtubule polymerization are required for phagocyte migration.
Single cell tracking of L. pneumophila-infected phagocytes
To analyse in more detail the effects on phagocyte migration of the L. pneumophila Icm/Dot T4SS and its substrate LegG1, the migration of L. pneumophila-infected phagocytes was tracked on a single cell level using the under-agarose assay and appropriate software. D. discoideum producing GFP was infected at an moi of 10 for 1 h with red fluorescent L. pneumophila wild-type, ΔicmT, ΔlegG1 or complemented ΔlegG1, and single infected amoebae were tracked for 15 min within a 2 h time window (Fig. 4A). The single cell analysis revealed that the forward migration index (FMI) and the velocity of D. discoideum infected with wild-type L. pneumophila was reduced 2–3 times, compared with amoebae infected with ΔicmT. Therefore, some Icm/Dot-translocated effector proteins appear to modulate the directionality and the speed of infected D. discoideum. Small effects due to the absence or overexpression of legG1 were statistically not significant.
The FMI and the velocity of uninfected D. discoideum was analysed in all experiments in parallel and found to be the same as for amoebae infected with L. pneumophila ΔicmT (Fig. S6). Furthermore, to address the point whether L. pneumophila-infected cells modulate the migration of uninfected bystander cells in a paracrine manner, we analysed the uninfected amoeba population in the under-agarose assay upon infection with either wild-type or ΔicmT bacteria. This single cell analysis revealed that the FMI and velocity of uninfected D. discoideum in infection assays was the same as for uninfected amoebae alone (Fig. S6).
Analogously to D. discoideum, RAW 264.7 macrophages (Fig. 4B) or human PMN (Fig. 4C) stained with CellTracker Green BODIPY were infected at an moi of 10 for 1 h with red fluorescent L. pneumophila wild-type, ΔicmT, ΔlegG1 or complemented ΔlegG1. The chemotaxis of single infected leucocytes towards CCL5 or fMLP, respectively, was tracked 2 h post infection for another hour. Again, the FMI and the velocity of the leucocytes infected with wild-type L. pneumophila were reduced by half, compared with phagocytes infected with ΔicmT, suggesting that Icm/Dot substrates modulate the directionality as well as the speed of the infected cells. However, compared with D. discoideum, the migration directionality and velocity of immune phagocytes infected with wild-type L. pneumophila appeared less severely affected. The FMI and the velocity of uninfected macrophages or PMN were the same as for phagocytes infected with L. pneumophila ΔicmT (data not shown).
Compared with leucocytes infected with wild-type L. pneumophila, the velocity (but not the directionality) of infected macrophages or PMN was significantly further reduced in absence of legG1, and both parameters were augmented upon overproduction of LegG1 (Fig. 4B and C). The FMI for macrophages or PMN infected with the complemented ΔlegG1 strain was even larger than the one for leucocytes infected with a ΔicmT mutant strain. Together, these results indicate that substrates of the L. pneumophila Icm/Dot T4SS, in particular the Ran activator LegG1, affect the FMI and the velocity of infected amoebae and leucocytes, and thereby modulate chemotaxis and migration of the phagocytes.
L. pneumophila LegG1 promotes random cell migration dependent on Ran GTPase
Next, we sought to determine whether L. pneumophila LegG1 affects either cell motility or chemotaxis. To this end, we performed scratch assays, where random cell migration occurs to close an area mechanically depleted of cells. Confluent layers of cells were scratched with a sterile pipette and infected, and ‘wound healing’ was followed over 24 h. RAW 264.7 macrophages infected with L. pneumophila wild-type or ΔlegG1 did not migrate and the size of the scratch area remained the same after 24 h as at the onset of the test (Fig. 5A). In contrast, macrophages infected with L. pneumophila ΔicmT or ΔlegG1 complemented with legG1 randomly migrated and significantly reduced the scratched area to an extent comparable to uninfected macrophages. Very similar results were obtained upon infection of A549 epithelial cells with the L. pneumophila strains (Fig. 5B). The depletion of the small GTPase Ran by RNA interference abolished the stimulation of migration by LegG1, and therefore, the process requires this small GTPase.
In an analogous approach, confluent layers of HeLa cells were subjected to ‘microbial microinjection’ with LegG1. To this end, derivatives of the T3SS-competent, effector-less Yersinia enterocolitica strain WA (pT3SS) were used, which produce YopE1–53, YopE1–53-LegG1, YopE1–138 or YopE1–138-LegG1 respectively (Rothmeier et al., 2013). Upon cell contact, these Yersinia strains inject YopE-LegG1 fusion proteins into HeLa cells, thus promoting microtubule polymerization. Y. enterocolitica strains producing the N-terminal YopE fragments, YopE1–53 or YopE1–138, abolished random cell migration in the scratch assay (Fig. 5C), and this effect was also observed with heat-inactivated bacteria (data not shown). In contrast, HeLa cells treated with Y. enterocolitica producing YopE-LegG1 fusion proteins readily migrated and significantly reduced the scratched area almost to the extent of uninfected cells. This result indicates that LegG1 is not only required but also sufficient to promote cell migration. In summary, the scratch assays revealed that LegG1 stimulates random migration of macrophages as well as of epithelial cells.
Real-time analysis of LegG1-dependent cell motility and microtubule dynamics
In order to analyse the role of LegG1 on cell motility and microtubule dynamics in real time, we used a D. discoideum strain producing GFP-α-tubulin. Compared with D. discoideum cells infected for 2 h with wild-type L. pneumophila, amoebae infected with the ΔlegG1 mutant strain exhibited impaired cell motility and much slower microtubule dynamics (Fig. 6; Movie S1). Moreover, the intracellular movement of L. pneumophila lacking legG1 was much less vivid compared with the wild-type strain. The latter observation is in agreement with the finding that the motility of LCVs harbouring ΔlegG1 mutant bacteria in D. discoideum producing calnexin–GFP is abolished (Rothmeier et al., 2013). The phenotypes could be complemented by providing the legG1 gene on a plasmid under control of its putative endogenous promoter. Finally, as outlined above the cell morphology and transient polarization dynamics of D. discoideum infected with L. pneumophila wild-type or ΔlegG1 was similar (Fig. S1), suggesting that the effects of LegG1 on cell motility and microtubule dynamics are rather specific. Together, these results indicate that LegG1 promotes cell migration by stimulating microtubule-dependent cell motility.
Using under-agarose migration, the Boyden chamber, scratch assays and (real-time) single cell tracking, we analysed the effects of L. pneumophila on motility and chemotaxis of D. discoideum, macrophages, PMN and epithelial cells. L. pneumophila was found to inhibit in a dose- and Icm/Dot-dependent manner the motility and chemotaxis of protozoan and mammalian cells. Furthermore, the Icm/Dot-translocated Ran activator LegG1 antagonized the inhibition of migration and stimulated cell motility.
Distinct chemo-attractants were used to assay the effects of L. pneumophila on the migration and chemotaxis of protozoan or immune phagocytes: folate (D. discoideum), CCL5 (macrophages) or fMLP (PMN). Compared with phagocytes infected with a ΔicmT mutant strain, the FMI as well as the velocity of the cells decreased significantly upon infection with wild-type L. pneumophila, regardless of the chemotactic pathway involved (Fig. 4). Together with the finding that L. pneumophila blocks random migration of macrophages and epithelial cells in scratch assays (Fig. 5), our results suggest that the pathogen targets conserved eukaryotic components to inhibit cell migration rather than (or in addition to) distinct signal transduction pathways involved in chemotaxis. Moreover, the accumulation of PtdIns(3,4,5)P3 at the plasma membrane of D. discoideum infected with wild-type L. pneumophila was indistinguishable from amoebae infected with a ΔicmT mutant strain, suggesting that the underlying signalling pathways are not affected by the bacteria (Fig. S1).
The Icm/Dot-translocated L. pneumophila effector(s) inhibiting phagocyte migration are currently unknown. L. pneumophila mutant strains lacking the ER interactor SidC or the Rab1 GEF/AMPylase SidM did not inhibit cell migration to a smaller extent than wild-type bacteria (Fig. S4). Interestingly, however, L. pneumophila lacking the Icm/Dot substrate legG1 inhibited the migration of amoebae and mammalian cells even more pronouncedly than wild-type bacteria, suggesting that LegG1 stimulates cell migration. Accordingly, the overexpression of legG1 promoted the migration of D. discoideum, macrophages, PMN and epithelial cells (Figs 3 and 5), increased the FMI as well as the velocity of the phagocytes (Fig. 4) and was sufficient to stimulate random migration of epithelial cells upon ‘microbial microinjection’ by Y. enterocolitica (Fig. 5C). These results are in agreement with the notion that LegG1 promotes cell migration by antagonizing Icm/Dot-translocated effectors inhibiting cell migration.
It is intriguing to speculate on how L. pneumophila might benefit from the LegG1-dependent promotion of cell migration. Perhaps, LegG1 counteracts Icm/Dot-translocated L. pneumophila effector proteins, which destabilize microtubules and thereby impair a number of crucial cellular pathways, including phagocytosis, vesicle trafficking, cytokinesis and migration. Thus, by activating the eukaryotic small GTPase Ran and consequent microtubule stabilization (Rothmeier et al., 2013), LegG1 might dampen or revert an otherwise deleterious impact of other effectors on the host cytoskeleton.
LegG1 has recently been characterized as a Ran activator, which promotes intracellular replication of L. pneumophila (Rothmeier et al., 2013). Ran is a pleiotropic small GTPase involved in a number of cellular processes, including nucleo-cytoplasmic transport (Stewart, 2007), post-mitotic nuclear envelope formation, as well as microtubule nucleation and spindle assembly during mitosis (Goodman and Zheng, 2006; Clarke and Zhang, 2008). In the context of L. pneumophila infection, microtubule polymerization represents a major downstream effect of Ran activation by LegG1 (Rothmeier et al., 2013).
Microtubule dynamics and polarization are pivotal determinants of eukaryotic cell migration (Etienne-Manneville, 2013). In D. discoideum microtubules can be reversibly depolymerized with pharmacological compounds such as nocodazole (Cappuccinelli et al., 1979) or thiabendazole (Kitanishi et al., 1984). Pharmacological studies using these reagents revealed that microtubule polymerization promotes locomotion of D. discoideum (Kitanishi et al., 1984). Furthermore, live cell microscopy indicated that repositioning of the microtubule-organizing centre (centrosome) stabilizes a pre-selected direction of movement (but does not control cell motility), most probably through the microtubule system (Ueda et al., 1997).
Whereas microtubule polymerization clearly represents an important effect of Ran activation and plays an obvious role for cell migration, we cannot rule out at this point that Ran activation by LegG1 has other migration-relevant implications for the host cell. Moreover, while LegG1 is sufficient to stimulate cell migration, other substrates of the Icm/Dot T4SS likely inhibit host cell components involved in cell migration and/or might target distinct pathways implicated in motility and chemotaxis (Fig. S4).
Ran activation by LegG1 might also be reverted by L. pneumophila, e.g. by a hypothetical bacterial Ran GAP. The reversible activation of a small GTPase by L. pneumophila has been analysed in great detail for Rab1, which is targeted by distinct bacterial effectors showing antagonistic activities as GEF/GAP (SidM/LepB) (Ingmundson et al., 2007), AMPylase/deAMPylase (SidM/SidD) (Müller et al., 2010; Neunuebel et al., 2011; Tan and Luo, 2011b) or phosphocholinase/dephosphocholinase (AnkX/Lem3) (Mukherjee et al., 2011; Tan et al., 2011a; Goody et al., 2012). Furthermore, LegG1 might be inhibited or degraded to modulate its activity. This has been described for the Icm/Dot substrate SidH, which is polyubiquitinylated by the ubiquitin ligase LubX and thus earmarked for proteolysis by the proteasome (Kubori et al., 2010).
The modulation of immune cell migration is a common theme among intra- and extracellular bacterial pathogens. The extracellular agent of plague, Yersinia pestis, inhibits chemotaxis and migration of neutrophils (Welkos et al., 1998) and dendritic cells (DC) (Velan et al., 2006), and enteropathogenic Y. enterocolitica disrupts macrophage immune function including chemotaxis through the action of the YopT cysteine protease targeting the small GTPase RhoA (Aepfelbacher et al., 2003). Thus, the pathogens apparently dampen the host immune response and thereby promote the infection process. Analogously, L. pneumophila might modulate the immune response during lung infection by subverting chemotaxis and migration of alveolar macrophages and inflammatory neutrophils.
A number of intracellular pathogens were also shown to subvert cell migration. The enteropathogen Shigella flexneri invades activated human CD4+ T lymphocytes and inhibits chemokine-induced cell migration by means of the type III-secreted PI 4-phosphatase IpgD in vitro (Konradt et al., 2011) as well as in vivo (Salgado-Pabon et al., 2013). Furthermore, the vacuolar enteropathogen Salmonella enterica serovar Typhimurium inhibits the motility of immune phagocytes such as primary macrophages and DC by means of the type III-secreted effector SseI (alias SrfH) (Worley et al., 2006; McLaughlin et al., 2009). Finally, obligate intracellular bacteria of the genus Chlamydia also modulate cell migration. Whereas C. trachomatis was found to inhibit the non-directed migration of HeLa cells (Heymann et al., 2013), C. pneumoniae promotes cell adhesion and migration of vascular smooth muscle cells (Zhang et al., 2012), and endosymbiotic environmental Chlamydia control the motility of host Acanthamoeba (Okude et al., 2012).
Whereas interference with leucocyte migration promotes bacterial infection by dampening the immune response of metazoan hosts, it is less clear how amoeba-resistant bacteria such as environmental Legionella or Chlamydia species might benefit from subverting the cell-autonomous migration of protozoa. A complex process like motility might negatively affect the survival and replication of intracellular bacteria in several manners, and thus is specifically targeted by the pathogens to tip the balance in favour of the invaders. Cell migration is energetically very costly. Thus, by inhibiting protozoa motility, intracellular pathogens might reduce the energy expenditure of their hosts and exploit intracellular energy sources to promote their own growth. The findings presented here provide the basis for a further mechanistic analysis of how Legionella species modulate phagocyte motility and chemotaxis and benefit from this strategy.
Cells, bacteria and plasmids
D. discoideum strains (Table S1) were grown as described (Ragaz et al., 2008). Murine RAW 264.7 macrophages and human A549 or HeLa cells were cultivated in RPMI 1640 medium amended with 10% heat-inactivated fetal calf serum (FCS) and 1% glutamine (all from Life Technologies). Human PMN were isolated freshly as described (Arnould et al., 1994). Briefly, 20 ml whole blood was obtained using a Vacutainer device (BD Biosciences) and a syringe (Braun) precoated with 50 μl heparin (5000 U ml−1). The blood was centrifuged (1000 rpm, 5 min), and the plasma fraction was transferred to fresh tubes and incubated for 1 h at room temperature (RT). Using isotonic Percoll (Sigma Aldrich) solutions in PBS, a discontinuous gradient was established in 15 ml tubes through layering 4 ml of 74% Percoll followed by 3 ml of 55% Percoll. Two millilitres of plasma was added on top of the Percoll gradient, and centrifuged for 20 min at 3000 rpm (600 g) without brake. The fraction containing PMN was washed with 40 ml PBS (1200 rpm, 5 min) and suspended in 1 ml PBS.
L. pneumophila and L. longbeachae (Cazalet et al., 2010) strains (Table S1) were grown for 3 d on CYE agar plates containing charcoal and yeast extract, buffered with N-(2-acetamido)-2-amino-ethanesulfonic acid (ACES). Liquid cultures were inoculated in AYE medium at an OD600 of 0.1 and grown at 37°C to an OD600 of 3.0 (21–22 h). Chloramphenicol (5 μg ml−1) or IPTG (1 mM) were added when needed. The infection of phagocytes by L. pneumophila was performed as described (Weber et al., 2006; 2009; Tiaden et al., 2007; Ragaz et al., 2008; Brombacher et al., 2009). In brief, the cells were infected (moi 1–50) with L. pneumophila grown for 21–22 h in AYE broth, the infection was synchronized by centrifugation (450 g, 10 min, RT), and the infected phagocytes were incubated at 23°C (D. discoideum) or at 37°C/5% CO2 (mammalian cells) for the indicated time.
The plasmids used and constructed in this study are listed in Table S1. DNA manipulations were performed according to standard protocols using commercially available kits. To construct plasmid pER22, the primers oSYS3 (TTTTTGGATCCTCATAACAAATTGCATGG) and oSYS6 (AAAAAGGATCCAATAACCCCAAATTATGC) were used to amplify a genomic region of L. pneumophila JR32 encompassing the genes lpg1975 and lpg1976, as well as the putative upstream promoter region (300 nucleotides). The PCR fragment was cut with BamHI, ligated into the vector pSW001 cut with the same restriction enzymes and sequenced.
Under-agarose assay and single cell tracking
Under-agarose assays and folic acid chemotaxis using D. discoideum were performed as described (Laevsky and Knecht, 2001). Briefly, 0.7% UltraPure agarose (Invitrogen) was melted in SM medium [10 g bacteriological peptone (Oxoid), 1 g Bacto yeast extract (BD Biosciences), 1.9 g KH2PO4, 0.6 g K2HPO4, 0.43 g MgSO4, 10 g glucose per litre, pH 6.5] and filled into microscopy dishes (μ-Dish, 35 mm; Ibidi). After solidification, 3 parallel slots of 2 × 4 mm (for cell suspensions and chemoattractant solution) were cut into the agarose in a distance of 5 mm (Fig. 1A). As a chemoattractant 1 mM folic acid (Sigma-Aldrich) in SM medium was applied to the centre slot 20–30 min before the cells were filled into the neighbouring slots.
For infections 1 × 106D. discoideum cells producing GFP (pSW102) were seeded onto 6-well plates in HL-5 medium and incubated overnight. The amoebae were washed once with MB medium [14 g bacteriological peptone (Oxoid), 7 g Bacto yeast extract (BD Biosciences), 4.26 g MES (Sigma-Aldrich) per litre, pH 6.9] and kept in 3 ml MB medium for infection. The infection with L. pneumophila was done at the indicated moi as described above (1 h, 23°C), followed by washing twice with MB to remove extracellular bacteria. Cells were detached by scratching into 300 μl MB, and 30 μl of the cell suspension was filled into the slots. The agarose dishes were incubated for 4 h in a humid chamber at 23°C.
Under-agarose assays using RAW 264.7 macrophages or human PMN were performed as described (Heit and Kubes, 2003). The microscopy dishes (μ-Dish, 35 mm; Ibidi) were incubated with 10% FCS (30 min, RT). After washing twice with PBS, the dish was filled with 1% UltraPure agarose in a 1:1 mixture of RPMI/HBSS (Life Technologies). Slots (5 mm apart) were formed using a template. The chemoattractant solutions (macrophages: 100 ng ml−1 CCL5 or TNFα; PMN: 100 ng ml−1 fMLP) were filled into the central slot 45 min before the cells were filled into the neighbouring slots.
For infections 1 × 106 macrophages or PMN were seeded onto 6-well plates in RPMI medium and incubated overnight. Cells were washed with RPMI, incubated for 45 min with 1 μM of BODIPY (green fluorescent cell tracker; Invitrogen), washed and kept in 3 ml RPMI for the infection. After infection with L. pneumophila (1 h, 37°C, 5% CO2), the cells were washed twice with RPMI to remove extracellular bacteria. The infected phagocytes were detached by scratching into 500 μl, and 150 μl of the cell suspension was filled into the agarose slots. The agarose dishes were incubated for 4 h in a humid chamber at 37°C/5% CO2.
Bulk cell migration was monitored by fluorescence microscopy (GFP, D. discoideum; BODIPY, macrophages or PMN) using a Leica TCS SP5 confocal microscope (HCX PL APO CS 10x/0.40 dry UV objective, Leica Microsystems). Merged overview pictures of the under-agarose assay were obtained using the tile scan function of the Leica software. Cell migration was quantified using ImageJ 1.45 software (function ‘plot profile’). The fluorescence intensities of infected cells relative to uninfected cells were plotted against the migration distance.
Individual phagocytes (D. discoideum, RAW 264.7 macrophages, human PMN) were tracked in the under-agarose assay 1 h after the cells were filled into the slots, using a SP5 confocal microscope (HCX PL APO CS 40x/1.25 oil UV objective). D. discoideum cells were monitored at 23°C for 15 min within a 2 h time window by taking 1 frame per 25 s. Macrophages and PMN were tracked at 37°C for 1 h with 1 frame per 35 s at 2 h post infection. Cells were tracked using the ImageJ manual tracking plugin and analysed with the ‘chemotaxis and migration tool 2.0’ (Ibidi).
In vitro scratch assay
In vitro scratch assays were performed essentially as described (Liang et al., 2007; Heymann et al., 2013). RAW 264.7 macrophages, A549 or HeLa cells were seeded into 35 mm μ-Dishes (Ibidi) at a density of 1.5 × 105 cells ml−1 (3 × 105 cells/dish) and incubated for 24 h to reach confluency. Confluent cell layers were washed with fresh medium and infected either with the L. pneumophila strains indicated (moi 10) or with the Y. enterocolitica ‘toolbox’ strain WA (pT3SS), which produces the Ysc type III secretion system (T3SS), yet lacks all endogenous effectors (Trülzsch et al., 2003; Wölke et al., 2011; Wölke and Heesemann, 2012). Y. enterocolitica strains expressing YopE1–53, YopE1–53-LegG1, YopE1–138 or YopE1–138-LegG1 were used for ‘microbial microinjection’ of LegG1 into HeLa cells, or as controls respectively (Rothmeier et al., 2013). Y. enterocolitica cultures used for infection (moi 10) were grown overnight in LB medium at 27°C, diluted 1:20 in LB medium and further grown to an OD600 of 0.3–0.4. After 1.5 h infection, extracellular L. pneumophila or Y. enterocolitica were washed off with fresh RPMI medium, the cell layer was scratched with a sterile pipette tip, and detached cells were washed off with fresh medium. Images of the scratched positions were taken at time point zero and after 24 h using a Leica SP5 confocal microscope (HCX PL APO CS 10x/0.40 dry UV objective). The percentage of ‘wound healing’ was quantified using ImageJ software (function ‘analyse particle’) by comparing the remaining area with the initial cell-free area.
For RNA interference experiments, A549 cells grown in 96-well plates were treated for 2 days with 10 nM (end concentration) of a mixture of four siRNA oligonucleotides against Ran, as described (Rothmeier et al., 2013). The cells were then infected (moi 10) for 1 h with GFP-producing L. pneumophila strains diluted in RPMI, washed 3 times with pre-warmed medium containing 10% FCS and incubated for 24 h.
Live cell fluorescence microscopy
For live cell fluorescence microscopy 5 × 105 exponentially growing D. discoideum cells harbouring plasmid pPH_CRAC-GFP or pSW102 (GFP) were seeded in 35 mm μ-Dishes (Ibidi) in HL-5 medium overnight. Infection with L. pneumophila strains expressing DsRed was performed as described above. One hour post infection extracellular bacteria were removed by washing 3× with HL-5 and the infected cells were further kept at 23°C. Cell morphology and localization of PtdIns(3,4,5)P3 was monitored with a Leica SP5 confocal microscope (HCX PL APO CS 63×/1.4–0.60 oil objective). Pictures were taken every 20 s for 30 min starting 1.5 h post infection. A total of 17–50 infected cells per strain and experiment were analysed.
Alternatively, 5 × 105 exponentially growing D. discoideum amoebae producing GFP-α-tubulin were seeded in 35 mm μ-Dishes (Ibidi) in LoFlo medium (Formedium) containing G418 (10 μg ml−1) one day prior to the experiment. Before infection, the cells were washed with LoFlo containing ascorbic acid (2 mg ml−1). Two hours post infection with L. pneumophila strains producing DsRed (moi 10) at 25°C (no centrifugation), an agarose overlay was prepared, and the infected cells were monitored with a SP5 confocal microscope (HCX PL APO CS 63×/1.4–0.60 oil objective) for 2 min. Images were taken every 15 s.
Boyden chamber assay
Chemotaxis of PMN towards fMLP was investigated in a Transwell (Boyden) chamber assay using 8 μm BD Falcon cell culture inserts for 24-well plates (BD Biosciences) as described in the manufacturer's protocol. The infection of PMN was done as described above for the under-agarose assay. As a chemoattractant fMLP (100 ng ml−1 in RPMI) or RPMI alone (control) was filled into the bottom chambers, and infected or uninfected PMN were loaded onto the cell culture inserts. After incubation for 3–4 h (37°C, 5% CO2) cells transmigrated into the bottom chamber were counted with a Neubauer hemocytometer.
For starvation/aggregation assays, 5 × 106D. discoideum Ax3 amoebae were seeded into 6-well plates in HL-5 medium the day before infection. The cells were washed twice with SorC (2 mM Na2HPO4, 15 mM KH2PO4, 50 μM CaCl2, pH 6.0) and infected with L. pneumophila strains (moi 10) as described above. After 1 h extracellular bacteria were removed by washing twice with SorC. To allow aggregation, the amoebae were further incubated in SorC at 23°C for 12–48 h and stained with propidium iodide (PI; 2.5 μg ml−1, 10 min, 23°C) prior to fluorescence microscopy.
Uptake, immunofluorescence microscopy and cytotoxicity assays
For uptake experiments 5 × 105D. discoideum cells, or 2.5 × 105 RAW 264.7 macrophages or PMN, respectively, were infected (moi 10, 1 h) with GFP-producing L. pneumophila strains in 24-well plates, washed, and detached by scraping. Fluorescence of GFP-positive amoebae harbouring L. pneumophila was determined by flow cytometry (Tiaden et al., 2013). To quantify uptake, the percentage of cells with GFP fluorescence above a defined threshold was determined. LCV formation was analysed in D. discoideum infected with DsRed-producing L. pneumophila strains by immunofluorescence staining with a rabbit anti-SidC antiserum and a secondary Cy5-conjugated goat anti-rabbit IgG (Invitrogen) (Weber et al., 2006).
To determine cytotoxicity of different L. pneumophila strains, 5 × 105D. discoideum or 2.5 × 105 RAW 264.7 macrophages were seeded in a 6-well plate. The cells were infected with L. pneumophila (moi 10, 1 h) as described above, detached by scraping, and collected in 15 ml tubes. After centrifugation (240 g, 10 min), the cells were suspended in 1 ml SorC (D. discoideum) or PBS (macrophages) containing 2.5 μg μl−1 PI, incubated for 10 min at 25°C in the dark, and PI-positive cells were quantified by flow cytometry (Tiaden et al., 2013).
We thank Gudrun Pfaffinger for excellent technical assistance and Carmen Buchrieser (Institute Pasteur) for providing L. longbeachae wild-type and ΔdotA mutant strains. The work in the group of H. H. was funded by the Max von Pettenkofer Institute, Ludwig-Maximilians University Munich, and the German Research Foundation (DFG; HI 1511/1-1, SFB914, SPP1580). A. M.-T. was supported by the DFG (SFB914). The authors declare no competing interests.