The Salmonella typhimurium tyrosine phosphatase SptP is translocated into host cells and disrupts the actin cytoskeleton

Authors

  • Yixin Fu,

    1. Department of Microbiology, School of Medicine, and Program in Molecular and Cell Biology, State University of New York at Stony Brook, Stony Brook, NY 11794-5222, USA.
    Search for more papers by this author
  • Jorge E. Galán

    1. Department of Microbiology, School of Medicine, and Program in Molecular and Cell Biology, State University of New York at Stony Brook, Stony Brook, NY 11794-5222, USA.
    Search for more papers by this author

Jorge E. Galán, E-mail galan@asterix.bio.sunysb.edu; Tel. (516) 632 8780; Fax (516) 632 9797.

Abstract

The Salmonella typhimurium protein tyrosine phosphatase SptP is a target of the centisome 63 type III protein secretion system. This system is essential for the interaction of these bacteria with host cells. We have shown here by a combination of biochemical and microscopy techniques that S. typhimurium directs the translocation of SptP into cultured epithelial cells. Translocation requires the function of the secreted proteins, SipB, SipC and SipD, as strains carrying mutations in any of the genes encoding these proteins fail to translocate SptP. Microinjection of purified GST–SptP into cultured cells results in the disruption of the actin cytoskeleton and the disappearance of stress fibres. These changes are reversible, as microinjected cells regain the normal appearance of their actin cytoskeleton upon prolonged incubation. Microinjection of the catalytically inactive GST–SptP(C481S) protein results in changes similar to those induced by the wild-type toxin. Furthermore, microinjection of a fusion protein between GST and the first 285 amino acids of SptP also leads to identical disruption of the host cell actin cytoskeleton, indicating that the amino-terminal half of SptP is sufficient to mediate this effect. However, microinjection of a fusion protein between GST and the last 259 amino acids of SptP also disrupted the normal appearance of the cytoskeleton. These results support the hypothesis that SptP is an effector protein arranged in modular domains that may co-operate with each other to exert related functions.

Introduction

Salmonella spp. interact with host cells in an intimate manner, resulting in the stimulation of host cell signal transduction pathways that lead to a variety of responses, including cytoskeletal rearrangements, the production of cytokines and, in some cell types, programmed cell death or apoptosis. This interaction is largely dependent on the function of a type III or contact-dependent protein secretion system encoded in a pathogenicity island located at centisome 63 of the Salmonella chromosome (reviewed in Galán, 1996). This type of protein secretion system has been identified in a number of animal and plant pathogenic Gram-negative bacteria that share in common the ability to engage their host cells in complex interactions. It is generally agreed that the main function of type III protein secretion systems is the translocation into the host cell of bacterial proteins that can then stimulate or interfere with host cell signal transduction pathways (reviewed in Galán and Bliska, 1996).

Previous studies have identified a number of S. typhimurium proteins that are secreted via the centisome 63 type III secretion apparatus (Collazo et al., 1995; Hueck et al., 1995; Kaniga et al., 1995a,b; 1996; Wood et al., 1996; Hardt and Galán, 1997). These secreted proteins fall into four functional categories (Collazo and Galán, 1996): (i) proteins required for the secretion process itself (e.g. InvJ and SpaO); (ii) proteins involved in the regulation of the secretion process (e.g. SipD); (iii) proteins involved in the translocation of secreted proteins into host cells (e.g. SipB, SipC and SipD); and (iv) proteins with putative effector function (e.g. SptP, SipA and AvrA). Some secreted proteins may perform more than one function. For example, SipD is involved in both the regulation of the secretion process (Kaniga et al., 1995a) and the translocation of protein targets into the host cell (Collazo and Galán, 1997). Likewise, SipB and SipC are involved in the delivery of secreted proteins into the host cell, while they themselves are translocated into the host cell cytosol, establishing the possibility that they may also function as effector molecules (Collazo and Galán, 1997).

Our laboratory has recently identified a secreted protein with putative effector functions termed SptP (Kaniga et al., 1996). This protein has a modular structural organization, which may reflect the presence of different effector domains, although this has not been demonstrated formally. The amino-terminal half of SptP shares sequence similarity with YopE of Yersinia spp. (Forsberg and Wolf-Watz, 1990; Michiels et al., 1990) and ExoS of Pseudomonas aeruginosa (Kulich et al., 1994). Both YopE and ExoS have been shown to be targets of type III secretion systems and have been reported to be involved in bacteria-mediated damage to host cells (Michiels et al., 1990; Rosqvist et al., 1991; Apodaca et al., 1995; Yahr et al., 1996) The carboxy-terminal half of SptP shares sequence similarity with the prokaryotic tyrosine phosphatase YopH (Bolin and Wolf-Watz, 1988; Michiels and Cornelis, 1988), as well as with the catalytic domain of eukaryotic tyrosine phosphatases. Consistent with this sequence homology, purified SptP protein exhibits potent phosphatase activity against various tyrosine-phosphorylated peptides (Kaniga et al., 1996). Such activity is effectively blocked by sodium vanadate, a specific inhibitor of tyrosine phosphatases, and by a single amino acid substitution in the predicted catalytic site of SptP (Kaniga et al., 1996). These results demonstrate that the catalytic mechanism of SptP is related to that of other eukaryotic tyrosine phosphatases. A strain of S. typhimurium carrying a null mutation in sptP is defective in its ability to colonize deeper tissues of orally infected Balb/c mice, indicating that SptP is required for the full display of virulence in this pathogen (Kaniga et al., 1996). As tyrosine phosphate is largely absent from prokaryotic microorganisms, it is likely that SptP activity is directed to host cellular proteins, which would require the translocation of SptP into host cells. In this paper, we show that S. typhimurium indeed directs the translocation of SptP into cultured epithelial cells, and that such translocation is dependent on the function of the type III secretion system encoded at centisome 63 of the S. typhimurium chromosome. In addition, we show that microinjection of SptP results in the disruption of the host cell actin cytoskeleton.

Results

SptP is translocated into cultured epithelial cells via the centisome 63 type III secretion system

The nature of the enzymatic activity of SptP suggests that this protein exerts its function on targets in the host cell. This would require SptP to be translocated into host cells. The invasion-associated type III secretion system encoded at centisome 63 on the S. typhimurium chromosome has previously been shown to be capable of translocating proteins into the cytosol of infected cells (Wood et al., 1996; Collazo and Galán, 1997). Therefore, we investigated whether SptP was translocated into cultured cells by this secretion system. Henle-407 cells were infected with wild-type S. typhimurium strain SL1344 or the isogenic mutant strains, SB237, which carries a null mutation in sptP, and SB136, which carries a mutation in invA, a gene that encodes an essential component of the type III secretion system (Galán et al., 1992). Two hours after infection, cells were fixed and processed for immunofluorescence staining with a monoclonal antibody directed to SptP as described in Experimental procedures. Cells infected with wild-type S. typhimurium showed intensive SptP staining throughout the host cell cytoplasm (Fig. 1). The specificity of the SptP monoclonal antibody was demonstrated by the lack of staining in host cells infected with the sptP mutant strain, SB237 (Fig. 1). In contrast to the results observed in host cells infected with wild-type bacteria, cells infected with the invA mutant strain, SB136, did not exhibit SptP staining (Fig. 1). Confocal microscopy of cells infected with wild-type bacteria established that SptP was distributed throughout the cell cytoplasm but was absent from the nucleus and plasma membranes (Fig. 2). These results demonstrate that SptP is translocated into host cells during infection, and that such translocation requires the function of the centisome 63 type III protein secretion system.

Figure 1.

. Translocation of SptP into host cells. Henle-407 cells were infected with wild-type S. typhimurium or its isogenic invA or sptP mutant strains. Infected cells were subsequently stained with a monoclonal antibody directed to SptP. Left, micrographs of SptP-stained cells; right, DIC images of the same cells.

Figure 2.

. Confocal micrographs showing SptP localization in S. typhimurium-infected Henle-407 cells. Henle-407 cells were infected with wild-type S. typhimurium, stained with a monoclonal antibody directed to SptP and observed under the confocal microscope. A series of 0.25 μm optical Z-sections of infected cells are shown (from bottom to top). The entire data set consisted of 50 sections, of which only six (separated by 1.5 μm) are shown. The sample was scanned with a 15 μm slit, 60 ×, 1.4 NA Nikon PlanApo objective, 30% laser power and 60% setting on the photo multiplier.

SptP translocation into host cells is dependent on the function of SipB, SipC and SipD

We have shown previously that SipB and SipC, two targets of the centisome 63 type III secretion system, are translocated into cultured epithelial cells (Collazo and Galán, 1997). We have also demonstrated that mutations in sipB, sipC or sipD prevented the translocation of any of these proteins into host cells, although the mutations did not prevent their secretion from the bacteria (Kaniga et al., 1995a,b; Collazo and Galán, 1997). These results indicated that SipB, SipC and SipD may be involved in the translocation of secreted target proteins into host cells. We, therefore, investigated the effect of mutations in these genes on the translocation of SptP. Henle-407 cells were infected with wild-type S. typhimurium or the isogenic strains SB169, SB220 and SB241, which carry mutations in sipB, sipC and sipD respectively. Infected cells were then processed for immunofluorescence with an anti-SptP monoclonal antibody. As shown in Fig. 3, no SptP staining was observed in cells infected with the sipB, sipC or sipD mutant strains, indicating that the SipB, SipC and SipD proteins may act as translocases for SptP. In addition, no bacterial-associated staining was observed in host cells infected with these secretion-proficient strains, suggesting that SptP does not remain associated with the bacterial surface in the absence of translocation (Fig. 3). This is in contrast to what has been previously reported for YopE, a target of a related type III secretion system in Yersinia spp. (Rosqvist et al., 1994). In this case, the presence of a mutation in a gene that encodes a putative translocase resulted in the accumulation of YopE on the bacterial surface.

Figure 3.

. Effect of mutations in sipB, sipC and sipD on SptP translocation into host cells. Henle-407 cells were infected with wild-type S. typhimurium or its isogenic derivatives carrying null mutations in sipB, sipC or sipD. Infected cells were subsequently stained with a monoclonal antibody directed to SptP. Micrographs of SptP-stained cells and the DIC images of the same cells are shown.

To confirm the putative role of the Sip proteins in the translocation of SptP with a more sensitive technique, we carried out a biochemical fractionation of Henle-407 cells infected with various strains of S. typhimurium as described in Experimental procedures. As shown in Fig. 4, SptP was readily detected in all the fractions of cells infected with wild-type S. typhimurium and in the infection medium of cells infected with the translocation-deficient sipB, sipC or sipD mutant strains. SptP was not detected in the Triton X-100 soluble or insoluble fractions (which contain translocated proteins) of cells infected with the translocation-deficient sipB, sipC or sipD strains. These results further demonstrate that SptP is translocated into host cells via the type III secretion system encoded at centisome 63 of S. typhimurium and further establish a role for SipB, SipC and SipD in the translocation process.

Figure 4.

. Detection of SptP in fractions of Henle-407 cells infected with wild-type and sip mutants of S. typhimurium strains. A. Whole-cell lysate of non-cell-associated bacteria. B. Bacteria-free infection medium. C. Triton X-100 insoluble fraction containing internalized bacteria. D. Triton X-100 soluble Henle-407 cell lysate containing translocated proteins. SptP was detected by Western immunoblotting with an anti-SptP monoclonal antibody. Infecting strains are indicated at the top of the gel. Arrows to the left indicate the position of SptP. Only relevant parts of the gel are shown.

SptP disrupts the host cell actin cytoskeleton

The observation that SptP is translocated into infected host cells indicates that this protein exerts its function on host cellular substrates. This is consistent with the finding that SptP shares sequence similarity with at least three bacterial toxins that have been reported either to have antiphagocytic effects or to disrupt the host cell cytoskeleton (Kaniga et al., 1996). Presence or absence of SptP does not result in significant differences in the changes in the morphology of the actin cytoskeleton caused by these bacteria after 1 h of infection (Fig. 5). This is most probably because of the fact that the bacteria-induced disruption of the cytoskeleton is the result of the concerted action of several effector proteins, some of them perhaps with redundant functions. Therefore, to investigate the function of SptP without the interference of other effector proteins delivered by S. typhimurium via type III secretion systems, we introduced purified SptP into host cells by microinjection. A GST–SptP fusion protein was purified, and different amounts were microinjected into Ref-52 cells as described in Experimental procedures. Ref-52 cells have a very well-developed actin cytoskeleton, which facilitated the observation of potential effects of SptP on these cellular structures. Microinjected cells were stained by either an anti-GST or an anti-SptP monoclonal antibody, and by rhodamine-conjugated phalloidin, which has high affinity for F-actin and therefore allows the visualization of the actin cytoskeleton. One hour after microinjection of purified GST–SptP (1 mg ml−1), the ordered distribution of the actin network was disrupted (Fig. 6). Microinjected cells exhibited either a granular phalloidin staining pattern or the total disappearance of stress fibres (Fig. 6). No obvious effect on the host cell cytoskeleton was observed after microinjection of protein solutions of up to 2 mg ml−1 purified GST alone or GST fused to InvE, a S. typhimurium protein that is not translocated into host cells (Fig. 6). These results indicate that the changes in the host cell observed after microinjection of GST–SptP were specific for SptP function and were not caused by an activity of the GST moiety or the microinjection protocol itself. Disruption of the host cell actin cytoskeleton was not a consequence of cell death, as cells microinjected with GST–SptP were able to exclude the membrane-impermeant DNA stain, ethidium homodimer (data not shown). Although microinjection of toxin solutions with a concentration of 1 mg ml−1 or more resulted in occasional cell lifting, microinjection of toxin solutions with a concentration of 100–200 μg ml−1 resulted in similar cytoskeletal changes to those caused by microinjection of higher amounts of toxin without significant cell lifting (Fig. 7). In fact, 3 h after microinjection with this reduced amount of toxin, the cytoskeleton had regained its normal appearance, indicating that the SptP-induced changes are reversible (Fig. 7). Disruption of the cell cytoskeleton by SptP microinjection was not caused by the delivery of toxin in excess of that resulting from S. typhimurium infection, as immunofluorescence staining indicated that cytoskeletal changes occurred with SptP amounts that were equal or less than those observed in host cells after bacterial infection. Microinjection of solutions of at least 500 μg ml−1 was required in order for SptP to be detectable by immunofluorescence. However, microinjection of SptP solutions with a concentration as low as 100 μg ml−1 was able to induce obvious cytoskeletal changes, although in this case SptP was not detectable by immunofluorescence staining following the same protocol that readily detected translocated SptP after S. typhimurium infection.

Figure 5.

. Effects of wild-type S. typhimurium or its sptP isogenic mutant on the integrity of the host cell cytoskeleton. Ref-52 cells were infected with wild-type S. typhimurium or its isogenic sptP mutant strain for 1 h, and the actin cytoskeleton of the infected cells was stained with rhodamine-labelled phalloidin.

Figure 6.

. Microinjection of SptP disrupts the host cell actin cytoskeleton. Cultured Ref-52 cells were microinjected with purified GST–SptP (1 mg ml−1), GST–InvE or GST alone (2 mg ml−1). Microinjected cells were stained with a monoclonal antibody directed to GST and with rhodamine-labelled phalloidin to visualize the actin cytoskeleton.

Figure 7.

. The SptP-mediated disruption of the host cell actin cytoskeleton is reversible. Cultured Ref-52 cells were microinjected with purified GST–SptP (protein concentration 100 μg ml−1). Microinjected cells were stained with rhodamine-labelled phalloidin to visualize the actin cytoskeleton. A. 30 min after microinjection. B.150 min after infection.

In combination, these results indicate that SptP specifically disrupts the host cell actin cytoskeleton.

The amino- and carboxy-terminal domains of SptP can mediate the disruption of the host cell cytoskeleton independently

The carboxyl-terminal half of SptP contains a signature sequence highly conserved in all tyrosine phosphatases (Kaniga et al., 1996). In addition, purified SptP showed potent tyrosine phosphatase activity against a variety of peptide substrates (Kaniga et al., 1996). Therefore, we investigated the possibility that this activity was required for SptP to disrupt the actin cytoskeleton. A GST–SptP fusion protein in which a critical cysteine residue located at position 481 of the amino acid sequence had been replaced by serine was purified and microinjected into Ref-52 cells. This protein fusion completely lacked tyrosine phosphatase activity (Kaniga et al., 1996; data not shown). Microinjected cells were then stained with anti-GST antibody and rhodamine-labelled phalloidin. Cells microinjected with GST–SptP(C481S) (0.2–1 mg ml−1) exhibited similar changes in the architecture of the actin cytoskeleton to those caused by the wild-type SptP protein (Fig. 8). These results indicate that the disruption of the actin cytoskeleton mediated by SptP can occur in the absence of its tyrosine phosphatase activity.

Figure 8.

. Toxicity of the catalytically defective SptP(C481S) mutant protein. Cultured Ref-52 cells were microinjected with purified GST–SptP(C481S) (1 mg ml−1), and microinjected cells were stained with a monoclonal antibody directed to GST and with rhodamine-labelled phalloidin to visualize the actin cytoskeleton.

The disruption of the host cell cytoskeleton mediated by SptP is reminiscent of the toxic effects mediated by YopE, a Yersina spp. toxin that is translocated into host cells by a related type III secretion system (Rosqvist et al., 1991; 1994). Indeed, strong amino acid sequence similarity to YopE exists throughout the amino-terminal half of SptP (Kaniga et al., 1996). We, therefore, investigated the toxic effects of the amino-terminal half of SptP. A fusion protein between GST and the first 285 amino acids of SptP was constructed and purified as described in Experimental procedures. Purified GST–SptP(1–285) was then microinjected into Ref-52 fibroblasts, and cells were subsequently stained with rhodamine-labelled phalloidin as described above. Microinjection of SptP(1–285) resulted in similar changes in the host cell cytoskeleton to those caused by either wild-type SptP or the catalytic mutant SptP(C481S) (Fig. 9), indicating that the amino-terminal domain of SptP is capable of mediating the disruption of the host cell cytoskeleton.

Figure 9.

. Toxicity of microinjected GST–SptP (1–285). Cultured Ref-52 cells were microinjected with purified GST–SptP(1–285) (200 μg ml−1), and microinjected cells were stained with rhodamine-labelled phalloidin to visualize the actin cytoskeleton.

Besides the tyrosine phosphatase signature, the carboxy-terminal half of SptP shares extensive sequence homology with YopH, a target of a Yersinia spp. type III secretion system that is also capable of disrupting the host cell actin cytoskeleton. We, therefore, investigated if the carboxy-terminal half of SptP was also capable of disrupting the host cell actin cytoskeleton. We microinjected a purified fusion protein between GST and the last 258 amino acids of SptP into Ref-52 cells and examined its effect on the actin cytoskeleton by rhodamine-labelled phalloidin staining. GST–SptP(286–544) retained tyrosine phosphatase activity in in vitro assays (data not shown). Microinjection of this protein also caused disruption of the host cell cytoskeleton. The changes were similar to those induced by the wild-type toxin, although the disruption in the cytoskeleton did not usually include the subcortical actin network (Fig. 10).

Figure 10.

. Toxicity of microinjected GST–SptP (286–544). Cultured Ref-52 cells were microinjected with purified GST–SptP(286–544) (500 μg ml−1), and microinjected cells were stained with a monoclonal antibody directed to GST and rhodamine-labelled phalloidin to visualize the actin cytoskeleton.

These results indicate that SptP is organized in two independent but related functional domains that presumably act on different cellular targets to effect similar cellular functions.

Discussion

The function of the type III protein secretion system encoded at centisome 63 of the Salmonella spp. chromosome is central for the ability of these bacteria to modulate host cell responses (Galán, 1996). We have shown here by immunofluorescence, confocal microscopy and biochemical fractionation that this protein secretion system directs the translocation into cultured cells of SptP, a secreted tyrosine phosphatase required for the full display of virulence by S. typhimurium (Kaniga et al., 1996). SptP translocation into host cells is dependent on the function of SipB, SipC and SipD, three additional targets of the type III secretion apparatus that are essential for S. typhimurium induction of host cell responses (Kaniga et al., 1995a,b). Mutations in sipB, sipC or sipD did not prevent SptP secretion but completely abolished its translocation into host cells. These results indicate that SipB, SipC and SipD may act as translocases of effector proteins into the host cell. We have shown previously that membrane-bound intracellular S. typhimurium is capable of translocating proteins to the cytosol of infected cells (Collazo and Galán, 1997). As mutations in sipB, sipC or sipD also affect the ability of S. typhimurium to gain access to host cells, it remains formally possible that the effect of these proteins on SptP translocation is indirect.

The translocation of SptP did not appear to be coupled to its secretion from the bacteria, as a significant proportion of SptP was found in the infection medium. This is unlike the Yersinia Yop proteins, which are reported to be translocated into the host cell in a polarized manner (Rosqvist et al., 1994). The differences between the translocation processes in these bacteria may be related to the different manner in which these bacteria engage the host cell. Yersinia spp. bind to host cells in a very tight manner, and such tight binding is absolutely required for protein translocation (Rosqvist et al., 1994; Sory and Cornelis, 1994; Persson et al., 1995). In contrast, video and fluorescence microscopy studies indicate that intimate and continuous bacterial attachment to host cells is not a prerequisite for Salmonella spp. delivery of effector molecules to the host cell (Collazo and Galán, 1997; C. Collazo and J. E. Galán, unpublished results).

Infection of cultured cells with wild-type S. typhimurium results in very dramatic rearrangements of the host cell cytoskeleton (Finlay and Ruschkowski, 1991). These responses are strictly dependent on the function of the centisome 63 type III protein secretion system and are, therefore, expected to be the result of the translocation of effector molecules into the host cell (Galán et al., 1992; Ginocchio et al., 1992). Indeed, Salmonella spp. have been shown to translocate several bacterial proteins into the host cell cytoplasm (Wood et al., 1996; Collazo and Galán, 1997; Hardt and Galán, 1997). In order to examine the function of SptP independently of the action of other effector proteins, we microinjected purified SptP into cultured host cells. SptP microinjection resulted in marked disruption of the host cell actin cytoskeleton. Introduction of the toxin caused either a granular phalloidin-staining pattern or the total disappearance of the stress fibres. These effects, which were observed even after microinjection of SptP amounts significantly lower than those resulting from bacterial infection, were not observed after microinjection of purified GST or GST fused to InvE, a S. typhimurium protein that is not translocated into host cells.

SptP is arranged in modular domains (Kaniga et al., 1996). The amino-terminal half shares sequence similarity with two other bacterial toxins that are secreted via functionally homologous type III secretion systems, Yersinia YopE (Forsberg and Wolf-Watz, 1990; Michiels et al., 1990) and Pseudomonas ExoS (Kulich et al., 1994). The carboxy-terminal half shares homology with YopH (Bolin and Wolf-Watz, 1988; Michiels and Cornelis, 1988) and eukaryotic protein tyrosine phosphatases. Our results are consistent with the hypothesis that this modular structure indeed represents independent effector domains. Microinjection of purified SptP(C481S), which is completely devoid of tyrosine phosphatase activity, induced cellular changes similar to those induced by the wild-type toxin. Furthermore, similar results were obtained when cells were microinjected with the purified amino- or carboxy-terminal domains of SptP. These results indicate that SptP is organized in two independent domains that affect similar cellular functions. At present, it is not known how the amino- or carboxy-terminal domains of SptP exert their function, although it is likely that they may act on different cellular targets to effect similar cellular process. Candidate target molecules for SptP are components of signalling cascades involving the small GTP-binding protein Rho, which modulates the assembly of both stress fibres and focal adhesion complexes (Ridley and Hall, 1992). Indeed, a number of bacterial toxins known to disrupt the normal organization of the actin cytoskeleton modulate Rho function (reviewed in Aktories, 1997). Work is in progress to investigate this possibility.

What is the role of SptP in Salmonella infection? The modulation of the host cell cytoskeleton suggests that SptP may play a role in the mechanisms of induction of membrane ruffling and subsequent bacterial uptake into non-phagocytic cells. It is possible that the disruption of stress fibres may facilitate the formation of membrane ruffles by increasing the pool of available G-actin necessary for the assembly of novel actin structures, such as filopodia and lamellopodia. Indeed, it has been proposed that filopodia and lamellopodia are functionally linked to the formation of stress fibres by the co-ordinated action of CDC42, Rac and Rho, three members of the Rho subfamily of low-molecular-weight GTP-binding proteins (Nobes and Hall, 1995). Furthermore, we have previously shown that CDC42 is required for S. typhimurium-induced membrane ruffling and bacterial entry into host cells (Chen et al., 1996a). It should be pointed out that SptP may exert other effects on the host cell cytoskeleton that our assays may not have detected, because they can only be revealed in the presence of other S. typhimurium effector molecules. An alternative potential function of SptP might be the disruption of the actin cytoskeleton to avoid phagocytosis by professional phagocytes or to prevent further bacterial infection of an already infected cell. However, an S. typhimurium strain carrying an sptP null mutation did not exhibit any significant defect in its ability to interact with either macrophages or epithelial cell lines in vitro (Chen et al., 1996b; Kaniga et al., 1996). It is possible that this is due to the presence of other Salmonella determinants with redundant functions. Indeed, mutations in all putative type III effector target proteins identified so far have resulted in rather weak virulence phenotypes, further suggesting that type III secreted targets often perform redundant functions (Kaniga et al., 1995a; Wood et al., 1996; Hardt and Galán, 1997).

Experimental procedures

Plasmids, bacterial strains and growth conditions

Plasmid pSB651, which encodes a fusion between glutathione-S-transferase (GST) and SptP, has been described previously (Kaniga et al., 1996). The wild-type S. typhimurium strain SL1344 (Hoiseth and Stocker, 1981) and its invA (SB136) (Galán et al., 1992), sipB (SB169), sipC (SB220) (Kaniga et al., 1995b), sipD (SB241) (Kaniga et al., 1995a) and sptP (SB237) (Kaniga et al., 1996) derivatives have been described previously. Bacteria were grown in LB containing 0.3 M sodium chloride to allow expression of components and targets of the invasion-associated type III system (Galán and Curtiss, 1990).

Fluorescence microscopy

Henle-407 cells grown on poly L-lysine-treated 12 mm round glass coverslips to 70% confluency were infected with the different strains of S. typhimurium at a multiplicity of infection (MOI) of 50 in Hanks' balanced salt solution (HBSS) for 2 h. Cells were briefly washed with phosphate-buffered saline (PBS), fixed in 3.7% formaldehyde and permeabilized in the presence of 0.2% Triton X-100 for 3 min. Permeabilized cells were incubated with anti-SptP monoclonal antibody in PBS containing 1.5% BSA and 2% goat serum overnight at 4°C. Cells were then washed with PBS and incubated with fluorescein isothiocyanate (FITC)-conjugated anti-mouse secondary antibody for 1 h. The coverslips were subsequently mounted onto glass slides with vectashield mounting solution (Vector Laboratories) and visualized under a fluorescence (Nikon Diaphot 300) or a confocal (Odyssey; Noran Instruments) microscope.

Biochemical fractionation of infected cells

Fractionation of infected cultured Henle-407 cells was carried out as described previously with minor modifications (Collazo and Galán, 1997). Briefly, semiconfluent intestinal Henle-407 cells grown in 100 mm tissue culture dishes were infected with wild-type S. typhimurium or its isogenic derivative strains at an MOI of 50 for 90 min in 2.5 ml of HBSS. The infection medium containing non-cell-associated bacteria was removed, and the Henle-407 cells were washed three times with HBSS. The infection medium and the cell washes were pooled, and the bacteria were recovered by centrifugation (non-adherent bacteria fraction). Proteins from the bacteria-free supernatant were recovered by 10% trichloroacetic acid (TCA) precipitation (infection medium fraction). Infected cells were incubated in 2.5 ml of Dulbecco's modified Eagle medium (DMEM) containing 100 μg ml−1 gentamicin for 1 h to kill extracellular bacteria, followed by three washes with PBS. The Henle-407 cells were subsequently treated with proteinase K (50 μg ml−1) for 15 min at 37°C to remove extracellularly associated SptP. The proteinase K treatment was terminated by the addition of 1 mM phenymethylsulphonyl fluoride (PMSF). The cells were then collected by low-speed centrifugation and lysed in the presence of 0.1% Triton X-100. Triton X-100 soluble and insoluble fractions were separated by centrifugation at 15 000 g.

Western blot analysis

Proteins were separated by SDS–PAGE, transferred to nitrocellulose filters and subsequently reacted with an anti-SptP monoclonal antibody. Reactive bands were visualized using enhanced chemiluminescence with a kit from Amersham according to the instructions provided by the manufacturer.

DNA manipulation

All recombinant DNA procedures were carried out according to standard procedures. Plasmid pSB675, which encodes a fusion of GST to the first 285 amino acids of SptP, was constructed by complete digestion of pSB651 with HindIII, partial digestion with PvuII and subsequent religation after treatment with the Klenow fragment of DNA polymerase to ‘fill in’ the ends. Plasmid pSB676 was constructed by complete digestion of pSB651 with XbaI, partial digestion with PvuII and subsequent religation after treatment with the Klenow fragment of DNA polymerase to ‘fill in’ the ends. This procedure created an in frame fusion between GST and the carboxy-terminal 259 amino acids of SptP. Expression and purification of all GST fusion proteins was carried out as described previously (Guan and Dixon, 1991).

Tyrosine phosphatase assay

The phosphatase activity of purified fusion proteins was determined using an enzyme-linked immunosorbent assay (ELISA; non-radioactive tyrosine phosphatase assay kit; Boehringer Mannheim) as described previously (Kaniga et al., 1996).

Microinjection of GST fusion proteins

Microinjection studies of the different GST–SptP fusion proteins were carried out in rat fibroblast Ref-52 cells. These cells were chosen because they possess a very well-developed actin cytoskeleton, which facilitates the observation of alterations in its organization. Cells were grown to subconfluency on gridded coverslips and serum starved for at least 3 h before microinjection. Purified GST–SptP fusion proteins (0.1–1 mg ml−1) were microinjected into the cytoplasm of Ref-52 cells using a semiautomatic microinjector and a micromanipulator (Eppendorff). At different times after microinjection, cells were fixed in 3.7% formaldehyde for 1 h and permeabilized in the presence of 0.2% Triton X-100 for 3 min. Cells were then sequentially incubated with a 1:500 dilution (in PBS containing 3% BSA) of anti-GST (Sigma) or anti-SptP monoclonal antibodies for 1 h, a 1:100 dilution of FITC-conjugated anti-mouse antibody for 1 h and a 1:500 dilution of rhodamine-conjugated phalloidin (Molecular Probes). Coverslips were finally mounted onto slides with vectashield mounting solution (Vector Laboratories) and visualized under a Nikon Diaphot 300 fluorescence microscope. The viability of microinjected cells was tested using the membrane-impermeant DNA stain, ethidium homodimer, as described elsewhere (Chen et al., 1996b).

Acknowledgements

We would like to thank Sumati Murli for critical review of this manuscript and useful discussions. This work was supported by Public Health Service grants AI30492 and GM52543 from the National Institutes of Health to J. E. G. who is an investigator from the American Heart Association.

Ancillary