Effector Proteins Encoded by Salmonella Pathogenicity Island 2 Interfere with the Microtubule Cytoskeleton after Translocation into Host Cells

Authors

  • Volker Kuhle,

    1. Institut für Klinische Mikrobiologie, Immunologie und Hygiene, FAU Erlangen-Nürnberg, Wasserturmstr. 3–5, D-91054 Erlangen, Germany
    Search for more papers by this author
  • Daniela Jäckel,

    1. Institut für Klinische Mikrobiologie, Immunologie und Hygiene, FAU Erlangen-Nürnberg, Wasserturmstr. 3–5, D-91054 Erlangen, Germany
    Search for more papers by this author
  • Michael Hensel

    Corresponding author
    1. Institut für Klinische Mikrobiologie, Immunologie und Hygiene, FAU Erlangen-Nürnberg, Wasserturmstr. 3–5, D-91054 Erlangen, Germany
    Search for more papers by this author

Michael Hensel, hensel@mikrobio.med.uni-erlangen.de

Abstract

The facultative intracellular pathogen Salmonella enterica has evolved strategies to modify its fate inside host cells. One key virulence factor for the intracellular pathogenesis is the type III secretion system encoded by Salmonella Pathogenicity Island 2 (SPI2). We have previously described SPI2-encoded SseF and SseG as effector proteins that are translocated by intracellular Salmonella. Detailed analysis of the subcellular localization of SseF and SseG within the host cell indicated that these effector proteins are associated with endosomal membranes as well as with microtubules. Specific association with microtubules was observed after translocation by intracellular Salmonella as well as after expression by transfection vectors. In epithelial cells infected with Salmonella, both SseF and SseG are required for the aggregation of endosomal compartments along microtubules and to induce the formation of massive bundles of microtubules. These observations demonstrate that SPI2 effectors interfere with the microtubule cytoskeleton and suggest that microtubule-dependent host cell functions such as vesicle transport or organelle positioning are altered by intracellular Salmonella.

The gastrointestinal pathogen Salmonella enterica has a remarkable ability to adapt to different niches within the infected host. This also includes the adaptation to the intracellular environment, as S. enterica is able to survive phagocytosis and to proliferate within eukaryotic cells. Throughout the intracellular phase, S. enterica remain in vesicles termed Salmonella-containing vacuoles (SCV). There are several observations indicating that intracellular S. enterica are able to interfere with the maturation of the SCV in order to prevent killing. Although the composition of SCV is not fully understood, this compartment has several unique characteristics, e.g. the presence of late lysosomal/endosomal markers such as the membrane protein LAMP-1, and an acidic pH.

The type III secretion system (T3SS) encoded by Salmonella Pathogenicity Island 2 (SPI2) has a central function in the intracellular fate of Salmonella (1,2). Mutant strains deficient in SPI2 are severely attenuated in systemic virulence in the murine model of salmonellosis (3), and the function of SPI2 is important for the intracellular proliferation (4). There are several cellular phenotypes reported for SPI2. By means of the SPI2-encoded T3SS, intracellular Salmonella can interfere with two aspects of the innate immune response of phagocytes. The delivery to the SCV of the enzymes NADPH oxidase and iNOS, involved in the production of reactive oxygen species and reactive nitrogen species, respectively, was inhibited (5,6). It has been observed that intracellular Salmonella can induce the formation of tubular aggregates of endosomal compartments in epithelial cells, referred to as Salmonella-induced filaments or SIF (7). The responsible bacterial virulence factor is SifA (8), and recently it turned out that SifA is translocated by the SPI2-encoded T3SS (9,10). SifA is required to maintain the integrity of the SCV, and mutant strains deficient in SifA lose their membrane-bound compartment during intracellular replication (9).

SifA is one member of a group of effector proteins of the SPI2 system that are encoded by genes outside of SPI2 (10). These effectors are referred to as Salmonella-translocated effectors (STE) and are characterized by conserved N-terminal translocation domains. There are also effector proteins that lack a conserved translocation domain. The SPI2-encoded protein SpiC has been proposed as an effector that inhibits cellular vesicle trafficking (11,12); however, the role of this protein is controversial (13,14). The T3SS-dependent secretion (15) and translocation of SseF and SseG (16) encoded by genes within SPI2 has been demonstrated. The target molecules of SPI2 effector proteins SseF and SseG in the host cell have not been defined so far.

There are many examples of the effects of bacterial virulence determinants on structure and function of the host cell cytoskeleton. The cytoskeleton of eukaryotic cells is composed of F-actin, microtubules and intermediate filaments. While the actin cytoskeleton is mainly important for cell migration and uptake of particulate material by phagocytosis, the microtubule cytoskeleton has important functions in cell division and intracellular transport of organelles. Many virulence factors involved in host cell invasion, and also toxins, interfere with functions of the actin cytoskeleton. The interference of microbial pathogens with microtubule-dependent processes is an emerging field of study (examples in 17–19).

With respect to intracellular transport processes, microtubules build transport routes for the intracellular movement of organelles, either towards the periphery of the cell or towards the microtubule-organizing center. As the maturation of phagosomes is also dependent on the function of microtubules (20), these structures might also form targets for intracellular pathogens.

We have previously analyzed the subcellular location of SPI2 effector proteins SseJ, SseF and SseG after translocation by intracellular S. enterica serotype Typhimurium (S. typhimurium). The three proteins were located with endosomomal membrane compartments in macrophages, and infection of an epithelial cell line indicated that these effectors are targeted to SIF structures (16). Interestingly, after translocation by mutant strains deficient in sseF or sseG, translocated effector proteins were associated with filamentous structures, but continuous aggregates of LAMP-1-positive vesicles were not induced. In order to characterize the nature of these structures, we investigated the association of SPI2 effector proteins with components of the host cell cytoskeleton. Previous reports indicated that disruption of the microtubule network inhibits SIF formation and result in the disintegration of SIF (7) and that SIF are formed along a scaffold of microtubules (21). We hypothesized that SPI2 effectors are interacting with components of the host cell cytoskeleton and investigated the association of the host cell cytoskeleton and translocated proteins.

During the characterization of the subcellular localization of translocated effector proteins of the SPI2-encoded T3SS, we observed that the effector proteins SseF and SseG show a remarkable association with microtubules. In this study, we report the role of SPI2 effector proteins SseF and SseG in novel microtubule-associated host cell phenotypes.

Results

Subcellular localization of translocated SPI2 effector proteins depends on microtubule integrity

The human epithelial cell line HeLa was infected with wild-type Salmonella harboring a plasmid for the expression of epitope-tagged SseF, and the subcellular localization of effector proteins was analyzed in the presence or absence of the microtubule-depolymerizing drug nocodazole. At 2 h, 4 h or 6 h after infection, the nocodazole was added and addition of the drug resulted in the complete disintegration of the microtubule cytoskeleton (data not shown). SIF-structures were absent in Salmonella-infected HeLa cells exposed to nocodazole (Figure 1). Addition of nocodazole had no effect on the translocation of SseF. However, if nocodazole was present, translocated SseF was not distributed along filamentous structures but was associated with LAMP-1-containing compartments also harboring Salmonella (Figure 1). These observations are in line with a recent report on the formation of SIF structures on a scaffold of microtubules (21). Translocation of SseF did not require a functional microtubule cytoskeleton, but the distribution of the effector was dependent on microtubule integrity.

Figure 1.

Figure 1.

Effect of microtubule integrity on translocation of SPI2 effectors and SIF formation. HeLa cells were infected with S. typhimurium wild type harboring a plasmid for the expression of SseF-M45. The microtubule disintegrating drug nocodazole was added to a final concentration of 10 μg/mL at the indicated time points after infection and main- tained throughout the experiment. As control, equal amounts of the solvent DMSO were added to the infected culture and maintained throughout the experiment. After an incubation time of 12 h, cells were fixed with 3% para- formaldehyde and immuno-stained for Salmonella lipopolysaccharide, late endosomal/lysosomal glycoprotein LAMP-1 and the translocated effector protein SseF. Representative infected cells are shown to indicate the effect of nocodazole on SIF formation and the association of translocated SseF with LAMP-1-positive compartments of SIF. Scale bar, 16 μm.

Translocated SPI2-effectors colocalize with host cell microtubules

The effect of nocodazole on the distribution of SseF prompted us to investigate the association of translocated effectors and the microtubule cytoskeleton in more detail. HeLa cells were infected with S. typhimurium strains harboring plasmids for the expression of epitope-tagged SPI2 effector proteins and the subcellular localization was analyzed after translocation by intracellular S. typhimurium. The translocated effector proteins appeared in the SCV and in SIF distant from the SCV (Figure 2A). As reported before (22), a certain proportion of the SCV showed accumulation of actin. Labeling of F-actin with Texas-Red-phalloidin did not reveal colocalization of translocated SseF-HA (Figure 2B) or SseG-HA (data not shown) with the actin cytoskeleton. In contrast, a proportion of the microtubule network showed colocalization with the translocated effector proteins SseF or SseG. Immuno-staining for β-tubulin indicated a frequent colocalization of SseF-HA (Figure 2B) and SseG-HA with the microtubule cytoskeleton. These observations demonstrate that SseF and SseG are present in endosomal compartments that aggregate to SIF along microtubules. The association of SIF, effector protein and microtubules was the most frequent observation and only few SIF without association to microtubules could be identified (Figure 2A).

Figure 2.

Figure 2.

SPI2 effector proteins translocated by intracellular Salmonella interact with the microtubule cytoskeleton. HeLa cells were infected with S. typhimurium wild-type strains harboring plasmids for the expression of HA epitope-tagged SseF or SseG. Infection was performed at an MOI of 5, and cells were processed for immuno-staining 16 h after infection.A) Immuno-staining was performed for the epitope-tagged effector proteins with rat monoclonal antibodies against the HA tag (Cy5-labeled secondary antibody, blue), for LAMP-1 using mouse monoclonal antibodies (Cy2-labeled secondary antibodies, green) and for microtubules using monoclonal antibodies against β-tubulin (Cy3-labeled, red). Merged images are shown in higher magnification to demonstrate colocalization of β-tubulin, LAMP-1-containing compartments and SseF-HA or SseG-HA (indicated by arrows).B) Using an experimental setting as in (A), immuno-staining was performed for Salmonella using polyclonal antiserum against O4 lipopolysaccharide (Cy2-labeled Sec. antibody, green) for SseF-HA and β-tubulin or for F-actin using Texas-Red-phalloidin (red). Representative infected cells are shown and detailed sections indicate the colocalization of SseF-HA with tubulin (violet staining) and the absence of colocalization of SseF-HA and F-actin, respectively. Similar observations were made with strains harboring a plasmid for the expression of SseG-HA (data not shown). Scale bar, 16 μm.

The similar subcellular localization of translocated SseF and SseG suggested the targeting of both proteins to the same compartment. To analyze a possible colocalization of SseF and SseG after translocation in more detail, a strain for the simultaneous expression of sseF::HA and sseG::M45 was constructed. We observed that both SseF-HA and SseG-M45 appeared in continuous tubular aggregates after translocation (Figure 3). However, the overlay of immuno-staining for SseF-HA and SseG-M45 indicated that a certain proportion of both effectors colocalized, whereas another fraction of SseF-HA and SseG-M45 appeared to be present in separate compartments (detail in Figure 3). Co-immunoprecipitation of SseF-HA and SseG-M45 from Salmonella-infected host cells was not possible (data not shown). However, a coimmunoprecipitation of HA- and cMyc-tagged SseF and SseG from cotransfected HeLa cells has been observed (J. Deiwick, personal communication).

Figure 3.

Figure 3.

Subcellular localization of translocated SseF and SseG. HeLa cells were infected at an MOI of 5 with S. typhimurium wild type harboring plasmid p2888 for the expression of sseF::HA and sseG::M45. Cells were processed for immuno-staining of SseF-HA (green) and SseG-M45 (red) 16 h after infection. For controls, immuno-straining was performed without addition of primary antibody against the HA tag (αHA) or the M45 tag (αM45). Arrows and arrowheads indicate distinct localization of SseF and SseG or colocalization, respectively. Scale bar, 16 μm.

Different localization of SseG after translocation or expression from transfection vectors

To analyze whether the colocalization of SseF and SseG is an intrinsic property of these effectors, transfection vectors were used for the expression of individual effectors in HeLa cells. Constitutive expression of SseF appeared to be toxic for host cells, since only few transfected cells were observed. Expression of SseG was tolerated by the host cells, and allowed the analysis of colocalization to host cell structures. The distribution of SseG in transfected cells was heterogeneous, showing perinuclear aggregates and scattered globular distribution in about 80% of the transfected cells (Figure 4A, upper and middle panel). In about 20% of the cells SseG expressed from the transfection vector showed a filamentous distribution that resembled the distribution of translocated SseG as shown in Figure 2 and colocalized with microtubules (Figure 4A, lower panel and section detail). The heterogeneous distribution of SseG expressed from a transfection vector is in contrast to the distribution of SseG after translocation by the T3SS, suggesting that expression by transfection vectors results in non-physiological distribution of this effector protein within the host cell.

Figure 4.

Figure 4.

Subcellular localization of SseG after expression from a eukaryotic transfection vector or after translocation by intracellular Salmonella. HeLa cells were transfected with the eukaryotic expression vector pIRES2 harboring sseG::HA (A, B) or infected with S. typhimurium harboring plasmid p2644 for the expression of sseG::HA(C). Immuno-staining was performed for the detection of SseG-HA (green in panels A and B, blue in panel C) and β-tubulin (red, panelA) or the trans-Golgi network protein TGN46 (red, panels B and C). In addition, in the infection experiment, intracellular Salmonella were immuno-stained (green, panel C). For the different experimental setups, representative cells are shown 24 h after transfection or 16 h after Salmonella infection. Scale bar, 16 μm.

The speckled perinuclear localization of SseG expressed from a transfection vector suggested a possible colocalization with the Golgi apparatus. Immuno-staining for the trans-Golgi network marker TGN46 indicated colocalization of Golgi organelles with SseG expressed from a transfection vector (Figure 4B) as previously shown (23). In contrast, SseG translocated by intracellular Salmonella showed a completely different association. The translocated protein was located in the SCV and in peripheral regions of infected cells, and the major portion of SseG was excluded from Golgi organelles (Figure 4C). These observations demonstrate that SseG expressed from a eukaryotic expression vector is retained in a deviant subcellular localization. Therefore, the analysis of the subcellular localization of SseG and probably other SPI2 effector proteins depends on the analysis of protein translocated by intracellular Salmonella.

SseF and SseG direct the aggregation of vesicles along microtubules

Infection with wild-type Salmonella leads to the formation of continuous tubular aggregates of late endosomes/lysosomes along a scaffold of microtubules ((21); this work). Triple staining for LAMP-1, β-tubulin and the translocated effector SseF-HA and SseG-HA demonstrated that SIF contain translocated effector protein and colocalize with microtubules (Figure 2A). In contrast, after infection with mutant strains deficient in sseF or sseG, continuous SIF were absent (24).

We analyzed the distribution of LAMP-1-positive vesicles, microtubules and SPI2 effector proteins after infection of HeLa cells with wild-type Salmonella or mutant strains deficient in sseF or sseG (Figure 5). In noninfected HeLa cells, a proportion of the LAMP-1-positive vesicles were associated with microtubules, but the vesicles appeared randomly distributed (Figure 5A). In host cells infected with wild-type Salmonella, SIF structures were induced and continuous tubular aggregates of vesicles were present along microtubules. In contrast, after infection with sseF- or sseG-deficient strains, LAMP-1-positive vesicles were associated with microtubules, but aggregation was not observed (Figure 5A). The LAMP-1-positive vesicles appeared lined-up along microtubules and we have previously described this phenotype as formation of pseudo-SIF (16).

Figure 5.

Figure 5.

SseF and SseG are required for the formation of endosomal aggregates along a scaffold of microtubules.A) HeLa cells were infected with S. typhimurium wild type or mutant strains deficient in sseF or sseG. Cells were fixed and immuno-stained for LAMP-1 (green) and β-tubulin (red) 16 h after infection. Infection with wild-type Salmonella induced the formation of continuous SIF along microtubules (indicated by arrowheads). Infection with mutant strains deficient in sseF or sseG resulted in the formation of discontinuous structures along microtubules (indicated by arrows), previously referred to as pseudo-SIF (16). The control (cont.) shows the distribution of LAMP-1-positive compartments and microtubules in noninfected cells.B) HeLa cells were infected with a mutant strain deficient in sseF harboring plasmid p2684 for the SPI2-dependent expression and translocation of SseJ-HA (left panel), or a mutant strain deficient in sseG harboring plasmid p2643 for the expression and translocation of SseF-HA (right panel). Immuno-staining was performed as before for the detection of LAMP-1 and β-tubulin, and SseJ-HA or SseF-HA were detected using rat monoclonal antibodies against the HA tag (Cy5-labelled secondary antibody, blue). Merged images and magnified sections show the distribution of SseJ and SseF along microtubules and the discontinuous distribution of LAMP-1-positive vesicles along microtubules and along translocated protein. Scale bar, 16 μm.

Figure 5.

Figure 5.

SseF and SseG are required for the formation of endosomal aggregates along a scaffold of microtubules.A) HeLa cells were infected with S. typhimurium wild type or mutant strains deficient in sseF or sseG. Cells were fixed and immuno-stained for LAMP-1 (green) and β-tubulin (red) 16 h after infection. Infection with wild-type Salmonella induced the formation of continuous SIF along microtubules (indicated by arrowheads). Infection with mutant strains deficient in sseF or sseG resulted in the formation of discontinuous structures along microtubules (indicated by arrows), previously referred to as pseudo-SIF (16). The control (cont.) shows the distribution of LAMP-1-positive compartments and microtubules in noninfected cells.B) HeLa cells were infected with a mutant strain deficient in sseF harboring plasmid p2684 for the SPI2-dependent expression and translocation of SseJ-HA (left panel), or a mutant strain deficient in sseG harboring plasmid p2643 for the expression and translocation of SseF-HA (right panel). Immuno-staining was performed as before for the detection of LAMP-1 and β-tubulin, and SseJ-HA or SseF-HA were detected using rat monoclonal antibodies against the HA tag (Cy5-labelled secondary antibody, blue). Merged images and magnified sections show the distribution of SseJ and SseF along microtubules and the discontinuous distribution of LAMP-1-positive vesicles along microtubules and along translocated protein. Scale bar, 16 μm.

Analysis of the localization of effector proteins was performed after translocation by mutant strains deficient in sseF or sseG. SseJ-HA (Figure 5B) and SseG-HA (data not shown) were translocated by the sseF mutant strain and colocalized with the microtubule network. A similar distribution was observed for SseF-HA (Figure 5B) and SseJ-HA (data not shown) after translocation by the sseG-deficient strain. SseF, SseG and SseJ showed a filamentous distribution that did not require the formation of continuous SIF. After translocation, these effector proteins associated with microtubules, leading to a filamentous distribution within the host cell. Both SseF and SseG were required to induce the aggregation of vesicles along microtubules.

Intracellular Salmonella induce the aggregation of microtubules

Analyses of the microtubule cytoskeleton after infection with wild-type Salmonella and various mutant strains also indicated the appearance of SPI2-dependent alterations in the architecture of the microtubule network. While uninfected cells predominantly showed a microtubule network of fine filaments distributed throughout the cell, remarkable changes in microtubule morphology were detected in host cell infected with wild-type S. typhimurium. In these cells, a proportion of microtubules appeared as massive bundles. As shown in Figure 2A (SseF-HA), the microtubule bundles were also associated with translocated effector protein and LAMP-1-positive vesicles.

In three independent experiments, approximately 100 infected cells per strain were examined for the formation of microtubule bundles (Figure 6B). While approximately 30% of cells infected with Salmonella wild type were positive for microtubule bundling, the appearance of the microtubule network in cells infected with mutant strains in ssaV, which encodes an essential part of the T3SS, was indistinguishable from those of uninfected cells. Cells infected with sseF and sseG mutants showed a significantly lower percentage of microtubule bundling (about 5%), comparable with cells infected with the ssaV strain. A similar extent of microtubule bundling was observed in uninfected cells and might represent the background level of microtubule aggregation in the host cells. Infection of HeLa cells with an sseF mutant complemented with a plasmid expressing sseF::M45 showed wild-type levels of microtubule bundling. We next analyzed the effect of mutations in further effector genes on microtubule bundling. SseI was recently shown to interact with the actin cross-linking protein filamin (25) while SseJ is associated with SIF (16,26). The function of both proteins is not required for induction of SIF ((27); our unpublished observations). Formation of microtubule bundles was observed after infection of HeLa cells with mutant strains deficient in sseI or sseJ to a similar extent as in cells infected with wild-type Salmonella.

Figure 6.

Figure 6.

Bundling of host cell microtubules is induced by SPI2-dependent translocation. HeLa cells were infected with S. typhimurium wild type or various mutant strains. A) 16 h after infection, the cells were fixed and immuno-staining was performed for Salmonella lipopolysacccharide and β-tubulin. In host cells infected with wild-type Salmonella, the appearance of bundles of microtubules was observed. Micrographs show the β-tubulin staining of representative infected cells with different extent of microtubule bundling (indicated by -/+, + , + +).B) At least 100 infected cells were analyzed and scored for the appearance of bundles of microtubules corresponding to –/+ to + + in panel A. The data shown are representative of the results of three independent experiments with similar outcome. Scale bar, 16 μm.

Analysis of the effect of the sifA mutation on microtubule bundling was difficult, since SifA-deficient bacteria were released from the SCV into the host cell cytoplasm, resulting in rapid replication and killing of HeLa cells. It has been reported that a double mutant in sifA and sseJ remains in a SCV (27). We analyzed the phenotype of the sseJ sifA mutant strain 10 h and 16 h after infection of HeLa cells. Rather than a localization of Salmonella in the SCV, we observed growth of the sifA sseJ strain in the cytoplasm similar to the sifA mutant strain. In most of these heavily infected cells the microtubule network was destroyed or showed an abnormal morphology, so that only weakly infected cells could be used for quantification of microtubule-bundling. About 10% of these cells infected with sseJ sifA or sifA mutant strains showed microtubule bundling in comparison to 30% in cells infected with Salmonella wild type.

Microscopic analyses of the microtubule morphology at various time points after infection showed that microtubule bundling appears rather late after infection (16 h), indicating that Salmonella-induced microtubule bundling is a time-dependent process. It is likely that microtubule bundling initiates earlier after infection but is not detectable before a certain extent of bundling has been reached. From these results, we concluded that SseF and SseG are both required for induction of bundling of microtubules.

Discussion

There are numerous examples of the interaction of bacterial pathogens with the actin cytoskeleton, either for invasion, the paralysis of phagocytes by extracellular bacteria, or for intracellular motility (for recent review see (28)). In contrast, there are far fewer observations of interactions of pathogens with microtubule-dependent functions. So far, involvement of microtubules has been demonstrated for invasion of Citrobacter and Campylobacter (17,18), and recently the interaction of the VirA of Shigella flexneri with microtubules was reported as an important event during invasion (19). The interaction of an intracellular bacterial pathogen with the microtubule cytoskeleton described here is a novel observation and might indicate the presence of an additional pathogenic strategy.

We have previously reported that SPI2-encoded proteins SseF and SseG are secreted (15) and translocated effector proteins (16). After translocation by intracellular S. typhimurium, both proteins were colocalized with endosomal membrane compartments rich in lysosomal glycoproteins (lpg) such as LAMP-1. In the absence of SseF or SseG, the formation of aberrant endosomal aggregates was observed, termed pseudo-SIF. These structures showed only punctate distribution of LAMP-1-positive compartments, but a continuous distribution of translocated effector proteins. The unique morphology of SIF and pseudo-SIF structures prompted us to investigate these structures in further detail and found that SseF, SseG and SseJ are colocalized with microtubules. We observed the formation of SIF along microtubules, confirming a recent study that showed microtubules acting as scaffolding for the formation of SIF (21). Targeting of SseG to microtubules was also observed after transfection with eukaryotic expression vectors; however, the subcellular localization was more heterogeneous than the distribution of translocated protein.

Microtubule bundling was detected at late time points after infection, while SIF formation initiated as early as 6 h after infection. As an alternative explanation for the bundling of microtubules, SseF and SseG might be initially involved in the aggregation of vesicles that are trafficking on microtubules. As a secondary effect of this aggregation, microtubule bundling could occur without a requirement of direct interactions of SPI2 effectors with microtubules. However, we observed that, in the absence of continuous SIF, SseF and SseJ colocalized with pseudo-SIF (16) and microtubules (Figure 5) and we were not able to detect any membrane compartment involved in this interaction ((16), this study). An association with LAMP-1-containing vesicles and SIF has also been reported for SifB (26), PipB (29) and PipB2 (30), other effector proteins translocated by the SPI2-encoded T3SS. It is conceivable that these proteins also show colocalization with microtubules and together with SseF, SseG and SseJ, belong to a large subset of effectors that are targeted to a similar subcellular localization. We have isolated microtubules from host cells infected with Salmonella strains translocating epitope-tagged effector proteins but were not able to copurify translocated SseF-HA or SseJ-HA with microtubules (data not shown). This may either indicate that there is no direct binding of the effector proteins to microtubules, or that a direct interaction is too weak to withstand the microtubule purification procedure. Further work has to confirm whether SseF, SseG and further effectors are directly interacting with microtubules, or indirectly affect the organization of the microtubule cytoskeleton.

In a recent study, SseG was described as a Golgi-targeting protein (23). Our experiments shown in Figure 4 confirm that SseG was colocalized with trans-Golgi membrane compartments if sseG was expressed from transfected plasmid. However, the localization of SseG was completely different if the protein was delivered by the natural way, i.e. by translocation through the SPI2-encoded T3SS. Translocated SseG was virtually excluded from trans-Golgi membranes and we cannot confirm the observation by Salcedo & Holden (23) that targeting to the Golgi apparatus is an intrinsic feature of SseG. Similar differences were observed for other SPI2 effector proteins after transfection or translocation (unpublished observation), suggesting that expression from eukaryotic expression vectors results in the deviant subcellular localization of SPI2 effector proteins and is of limited use for the study of the biological role of these proteins.

Our study indicates that intracellular Salmonella interact with the host cell cytoskeleton in two different ways. SspH2 and SseI, further effector proteins of the SPI2 system encoded by gene loci outside of SPI2, have been reported to interact with filamin and profilin, proteins involved in the organization of the actin cytoskeleton (25). In addition, SpvB, a protein encoded by the Salmonella virulence plasmid, was found to interfere with the actin cytoskeleton by ADP-ribosylation of G-actin (31,32). We here propose that a second subset of effector proteins SseF and SseG interferes with the microtubule cytoskeleton and modifies the organization of the microtubule network. The molecular nature of this interaction remains to be clarified by further investigations.

The presence of the different strategies for interaction with the cytoskeleton might be a consequence of the different requirements of intracellular Salmonella. On the one hand, intracellular Salmonella have to modify the maturation of the pathogen-containing vesicle in order to avoid killing and degradation in the endocytic pathway. A bacterial interference is also required to protect intracellular Salmonella from the bactericidal effects of enzyme systems that generate reactive oxygen or nitrogen intermediates. On the other hand, the replicating intracellular population depends on a continuous supply of membrane compartments in order to maintain the integrity of the SCV. Mutant strains deficient in SifA are attenuated due to the inability to maintain the SCV (9), resulting in the release into the cytoplasm and either killing of the bacteria or rapid killing of the host cell due to bacterial replication (33,34). Although both forms of intracellular activities are linked to the function of SPI2, is has not been possible so far to correlate the different activities to distinct subsets of effector proteins.

Interference with microtubule-dependent trafficking might be an important virulence trait of intracellular bacteria that remain within a membrane-bound compartment that contributes to avoidance of antimicrobial activities as well as acquisition of membrane compartments to the SCV. A large proportion of cellular transport depends on the vesicle trafficking along microtubules. Transport vesicles move bidirectionally along microtubules. In addition, the microtubule cytoskeleton is required for the maintenance of the structure and subcellular localization of organelles (reviewed in (20,35)). Microtubules maintain the Golgi apparatus in a perinuclear location, and are required to guide membranes of the endoplasmic reticulum (ER) to the periphery of a cell. It has been observed that the disruption of microtubule functions also affects the localization and function of ER and Golgi (20,35). We speculate that an interference of translocated SPI2 effector proteins with microtubule organization affects the normal subcellular localization of ER, Golgi and possibly further compartments. This pathogenic interference might allow access to membrane compartments for the growing SCV.

In phagocytic cells, the maturation of phagosomes requires the function of microtubules (36). At present it is not known how the subcellular localization of certain enzymes with antimicrobial activity such as the inducible NO synthase is controlled and if microtubules are involved in this process. In contrast, it is established that microtubule function is required for the presentation of epitopes by antigen-presenting cells such as dendritic cells (37). Interference of intracellular Salmonella with microtubules could reduce or even block microtubule-dependent vesicle transport. We propose that tubular aggregates of vesicles are a consequence of Salmonella-induced bundling of microtubules, resulting in close proximity and eventually fusion of vesicles that traffic along microtubules. This hypothesis is depicted in Figure 7.

Figure 7.

Figure 7.

Working model for SPI2-induced phenotypes in infected host cells. After entry into a host cell, the function of the SPI2-encoded T3SS is essential to direct the intracellular fate of Salmonella. Wild-type Salmonella (WT) deliver a set of effector proteins that modify the SCV into a compartment that permits bacterial proliferation. This process is accompanied by modification of the cytoskeleton such as bundling of microtubules (MT). Aggregation of endosomes or SIF formation might be consequence of the blockage of traffic of vesicles (V) along microtubules. In contrast, SPI2 null mutant strains (e.g. ssaV) are unable to modify the maturation of the SCV into degradative compartments and fail to proliferate. Mutant strains deficient in sseF or sseG are unable to interfere with the microtubule cytoskeleton, but are only slightly attenuated in intracellular proliferation. Target proteins for the effectors are not known and the mode of interference remains to be explained by further studies.

Taken together, we propose that SseF and SseG are proteins that interfere with the microtubule cytoskeleton as well as with endosomal membrane compartments. We observed that SseF and SseG are both required for induction of massive bundles of microtubules and tubular endosomal aggregates. As formation of continuous SIF structures is absent in mutant strains deficient in either sseF or sseG, the gene products are not redundant in function but rather appear to interact in the modification of endosomal transport processes. This notion is supported by the observation that bundling of host cell microtubules requires the function of both SseF and SseG and not of other SPI2 effector proteins. Mutant strains deficient in sseF, sseG or sseFG double mutants only show a mild reduction in intracellular proliferation and in systemic virulence in the mouse model (16,38). This suggests either that further effector proteins are involved in modification of intracellular trafficking, or that this modification is not essential for the survival and replication of Salmonella in host tissues.

Further work is needed to determine the specific host cell molecules interacting with SseF and SseG and how such an interaction can result in alterations of the microtubule cytoskeleton.

Materials and Methods

Bacterial strains and culture conditions

Bacterial strains (Table 1) were routinely grown in Luria Bertani (LB) broth or on LB agar plates. If required to maintain plasmids, kanamycin (50 μg/mL) carbenicillin (50 μg/mL) or chloramphenicol (30 μg/mL) was added.

Table 1. Strains and plasmids used in this study
Strains or plasmidRelevant characteristicsReference
S. typhimurium strains
 NCTC 12023wild typeNCTC, Colindale
 P2D6ssaV::mTn5(3)
 HH107ΔsseF::aph(38)
 HH108ΔsseG::aph(38)
 MvP377ΔsseJ::aph(5)
 MvP378ΔsseI::aph(5)
 MvP373ΔsscBsseFG(16)
 MvP392ΔsseJthis study
 MvP450ΔsseJ sifA::mTn5this study
Plasmids
 pWSK29cloning vector, low copy number(41)
 pIRES2-EGFPeukaryotic expression vectorClontech, Heidelberg
 p2095ProsseAsscBsseF::M45 in pWSK29(15)
 p2643ProsseAsscBsseF::HA in pWSK29this study
 p2644ProsseAsscBsseFsseG::HA in pWSK29this study
 p2684ProsseJsseJ::HA in pWSK29(42)
 p2691sseG::HA in pIRESthis study
 p2888ProsseAssecBsseF::HA sseG::M45 in pWSK29this study

DNA biochemistry and strain construction

Plasmids (Table 1) for the expression of HA epitope-tagged effector proteins were constructed basically as described before (16). Briefly, sseF, sseG or sseJ were amplified with primers introducing restriction sites and the HA sequence at the 3′-terminus of the respective gene. Plasmids were introduced into S. typhimurium wild type or various mutant strains by electroporation.

A plasmid for the simultaneous expression of sseF::HA and sseG::M45 was generated by amplification of the sseG::M45 fusion of p2096 with primers SseG-XbaI-Fw (5′-ACTTCTAGATCTCGGGGAGAACCATGAAAC-3′) and Seq-Rev (5′-AGCGGATAACAATTTCACACAGGA-3′) introducing XbaI restriction sites. The product was digested with XbaI and cloned into XbaI-digested plasmid p2643. In the resulting plasmid p2888 a hybrid operon consisting of sscB sseF::HA sseG::M45 was under control of the promoter of sseA.

For construction of a eukaryotic expression vector, sseG::HA was amplified by primers SseG-For-EcoRI (5′-CGCGAATTCTGATCATACATCTCGGG-3′) and HA-Rev-BamHI (5′-TGCGGATCCTTAAGCGTAGTCTGGGA-3′) from template p2644. The resulting DNA fragment was digested with EcoRI and BamHI and subcloned in pIRES2-EGFP to obtain plasmid p2691.

For the construction of strain MvP450 deficient in sseJ and sifA the aph gene replacing sseJ of strain MvP377 was deleted by FLP-mediated recombination as described (39) to generate sseJ deletion strain MvP392. Subsequently, the sifA::mTn5 allele of mutant strain P3H6 was moved into MvP392 by P22 transduction according to standard procedures (40) to generate MvP450.

Bacterial infection of HeLa cells and nocodazole treatment

The human epithelial cell line HeLa was cultured in Dulbecco's modified Eagle medium (DMEM, Invitrogen, Heidelberg, Germany) containing 10% fetal calf serum (FCS, Invitrogen) and 2 mm glutamine at 37 °C in 5% CO2. HeLa cells with a passage number lower than 25 were used.

For infection of HeLa cells, S. typhimurium strains were grown in LB broth, containing antibiotics if required, at 37 °C with agitation to stationary phase. The cultures were then diluted 1 : 30 with fresh LB broth and incubated for another 3.5 h at 37 °C with agitation to reach late logarithmic phase. The OD600 of the cultures was adjusted with phosphate-buffered saline (PBS; pH 7.3) to 0.2. The inocula were diluted in DMEM containing FCS and glutamine and added at a multiplicity of infection (MOI) of about 1 : 5 to cells seeded in 24-well tissue culture plates. The bacteria were centrifuged onto the cells at 500 × g for 5 min and incubated for 25 min at 37 °C in 5% CO2. After infection, the epithelial cells were washed three times with PBS and incubated for 1 h in medium containing FCS, glutamine and 100 μg/mL gentamicin. The medium was replaced with medium containing FCS, glutamine and 10 μg/mL gentamicin for the remainder of the experiment.

For inhibition of microtubule-dependent processes, nocodazole (Calbiochem, Heidelberg, Germany) was added to the cell culture media at various time points after infection to a final concentration of 10 μg/mL. As controls, equal amounts of the solvent DMSO were added to the cells. Samples were taken 12 h after infection.

Transfection of HeLa cells

HeLa cells were seeded 24 h before transfection in 24-well plates on glass cover slips at a density of about 3.5 × 104 cells/well. Directly before transfection, cells were washed with PBS and 600 μL of fresh medium was added. Per well, 0.2 μg DNA prepared using the EndoFree kit (Qiagen, Hilden, Germany) was diluted to a final volume of 20 μL with serum-free DMEM medium before addition of 3 μL Polyfect transfection reagent (Qiagen). The solution was briefly mixed and incubated for 5–10 min at room temperature before addition of 120 μL DMEM containing serum. After mixing by pipetting up and down, the solution was added immediately to the cells. Cells were incubated at 37 °C in 5% CO2 for different periods (24–48 h). Generally, medium was changed after approximately 16 h.

Immuno-staining

For immuno-staining, cells were grown in 24-well tissue culture plates on glass cover slips. After infection or transfection and incubation for different periods, the cells were fixed in 3% para-formaldehyde in PBS for 15 min at RT and then washed three times with PBS. The antibodies were diluted in a blocking solution consisting of 10% goat serum, 1% bovine serum albumin (BSA) and 0.1% saponin (Sigma, Deisenhofen, Germany) in PBS. The cover slips were incubated for 1 h at room temperature with the various antibodies and for 1–3 h with the antibody against the M45 epitope. After each incubation step, the coverslips were washed three times with PBS. The coverslips were mounted on Fluoroprep (bioMèrieux, Nürtingen, Germany) and sealed with Entellan (Merck). Samples were analyzed using a confocal laser-scanning microscope (Leica TCS-NT).

Antibodies

Following primary antibodies were used at the indicated dilutions: rabbit anti Salmonella-O4 Bacto testsera, 1 : 1000 (Difco); rat MAb anti HA, 1 : 500 (Roche, Mannheim, Germany); mouse anti β-tubulin Cy3, 1 : 200 (Sigma); sheep anti human TGN46 1 : 100 (Biozol, Eching, Germany); mouse anti human LAMP-1, 1 : 100 (A4H3; generated by Drs August and Hildret and provided by DSHB, Iowa City, IA); mouse anti M45 epitope, 1 : 10–1 : 200.

Secondary antibody fluorochrome conjugates were obtained from Dianova (Hamburg, Germany) and used as follows: donkey anti rat Cy2 at 1 : 500; goat anti rabbit Cy2 at 1 : 1000; goat anti mouse Cy3 1 : 500; goat anti rat Cy5 at 1 : 200; goat anti mouse Cy2 at 1 : 300; rabbit anti sheep Cy3 at 1 : 200 and goat anti mouse Cy5 at 1 : 200.

For labeling of host cell F-actin, Texas-Red-phalloidin (Molecular Probes, Leiden, The Netherlands) was used at a dilution of 1 : 200.

Acknowledgments

This work was supported by DFG grant 1964/8-1 and in part by the Fonds der Chemischen Industrie. We wish to thank Dr Jörg Deiwick, Lübeck, for sharing data prior to publication and the members of the Hensel lab for critical comments and useful suggestions.

Ancillary