The Salmonella pathogenicity island 2 (SPI-2) type III secretion system (TTSS) of Salmonella typhimurium is required for bacterial replication within host cells. It acts by translocating effector proteins across the membrane of the Salmonella-containing vacuole (SCV). The SifA effector is required to maintain the integrity of the SCV membrane, and for the formation in epithelial cells of Salmonella-induced filaments (Sifs), which are tubular extensions of SCVs. We have investigated the role in S. typhimurium virulence of the putative SPI-2 effector genes sifB, srfJ, sseJ and sseI. An S. typhimurium strain carrying a mutation in sseJ was mildly attenuated for systemic virulence in mice, but strains carrying mutations in either srfJ, sseI or sifB had very little or no detectable virulence defect after intraperitoneal inoculation. Expression of SseJ in HeLa cells resulted in the formation of globular membranous compartments (GMCs), the composition of which appears to be similar to that of SCV membranes and Sifs. The formation of GMCs was dependent on the serine residue of the predicted acyltransferase/lipase active site of SseJ. Transiently expressed SseJ also inhibited Sif formation by wild-type bacteria, and was found to associate with Sifs, SCV membranes and simultaneously expressed SifA. Intracellular vacuoles containing sseJ mutant bacteria appeared normal but, in contrast to a sifA mutant, a sifA sseJ double mutant strain did not lose its vacuolar membrane, indicating that loss of vacuolar membrane around sifA mutant bacteria requires the action of SseJ. Collectively, these results suggest that the combined action of SseJ and SifA regulate dynamics of the SCV membrane in infected cells.
A distinguishing feature of TTSSs is that they couple bacterial secretion with the injection of effector proteins across host cell membranes. This is accomplished by the formation of a cylindrical, needle-like organelle (secreton) that exports proteins across the two membranes of the bacterial cell, and a pore in the host cell membrane (translocon), which allows transfer of effectors into the target cell (Cornelis and Van Gijsegem, 2000). In some systems, the secreton and translocon appear to be connected by other proteins (Knutton et al., 1998; Sekiya et al., 2001). Structural elements of the secreton and translocon are conserved to varying degrees among different pathogenic bacteria, but effector functions are distinct and confer functional specificity on each individual system.
Through the action of unknown effectors, the SPI-2 TTSS also enables Salmonella to avoid the respiratory burst by inhibiting trafficking to the SCV of NADPH oxidase subunits (Vazquez-Torres et al., 2000; Gallois et al., 2001). Additional functions associated with the SPI-2 TTSS include the induction of a delayed apoptotic-like cell death in macrophages (van der Velden et al., 2000), and the assembly of an F-actin meshwork in the vicinity of the SCV, which also helps to maintain vacuolar membrane integrity (Méresse et al., 2001).
Recently, several genes related to the SPI-2 TTSS have been identified in S. typhimurium. Worley and colleagues (Worley et al., (2000) identified a global regulon of genes under the control of SsrA-SsrB, many of them located outside SPI-2. A conserved N-terminal amino acid sequence of approximately 150 residues has been identified in several proteins, also encoded outside SPI-2, including SifA and a related protein, SifB (Beuzón et al., 2000; Guy et al., 2000; Miao and Miller, 2000; Brumell et al., 2001a). This sequence was shown to be required for translocation of five of these proteins to the host cell cytosol via either the SPI-2 TTSS (SspH2, SseI, SseJ), or both SPI-1 and SPI-2 TTSSs (SspH1, SlrP) (Miao and Miller, 2000). Apart from SifA and SpiC, the functions of the other putative effectors described above are currently unknown. A strain carrying mutations in both sspH1 and sspH2 was attenuated for calf enteropathogenesis but did not have a virulence defect in mice (Miao et al., 1999). slrP is not regulated by SsrA-SsrB (Miao and Miller, 2000), and is required for colonization of Peyer’s patches but not spleens of mice (Tsolis et al., 1999).
In this study, we have investigated the role of sifB, srfJ, sseJ and sseI during systemic infection of mice. SifB is similar to SifA over the length of the protein (Miao and Miller, 2000). SrfJ is similar to human lysosomal glucosyl ceramidase, suggesting that it may alter the host cell ceramide signal transduction pathway (Worley et al., 2000). The deduced amino acid sequence of sseI (originally referred to as srfH by Worley and colleagues; Worley et al., 2000) does not suggest a function for the protein but, like sifB, srfJ and sseJ, this gene is expressed in an SsrA-SsrB-dependent manner (Miao and Miller, 2000; Worley et al., 2000).
SseJ shares a high level of similarity with several proteins belonging to the ‘GDSL’ family of lipolytic enzymes, exemplified by an Aeromonas hydrophila acyltransferase/lipase (Upton and Buckley, 1995). This similarity begins at amino acid 142 of SseJ, precisely where the conserved N-terminal translocation signal described for putative effectors of SPI-1 and SPI-2 TTSSs finishes (Miao and Miller, 2000). SseJ has the three residues (Ser-Asp-His) that constitute the catalytic triad characterized in the A. hydrophila enzyme, within three out of five conserved blocks of sequence present in SseJ and all proteins belonging to this family (Brumlik and Buckley, 1996). These include additional key residues such as a second serine flanking the GDSL signature pattern (Robertson et al., 1994).
We found that S. typhimurium strains carrying single mutations in either sseJ or srfJ were mildly attenuated for systemic virulence, whereas mutations in either sseI or sifB conferred no detectable virulence defect. Expression of SseJ in HeLa cells resulted in the formation of large globular structures with characteristics of SCV membranes and Sifs. Ectopically expressed SseJ also reduced Sif formation by wild-type bacteria, and was found to associate with the remaining Sifs, SCV membranes, and simultaneously expressed SifA. In contrast to a sifA mutant, a sifA sseJ double mutant strain did not lose its vacuolar membrane, indicating that loss of vacuolar membrane around sifA mutant bacteria (Beuzón et al., 2000) requires the action of SseJ. Collectively, our results indicate that the combined action of SseJ and SifA regulate dynamics of the SCV membrane in infected cells.
Effects of mutations in sifB, srfJ, sseJ and sseI on virulence and intracellular replication
To investigate the roles in virulence of sifB, srfJ, sseJ and sseI, single mutant strains were constructed which carry either disruptions in srfJ, sseJ or sseI, or a deletion of sifB. These were subjected to virulence tests by comparing them with the wild-type parental strain in mixed infections in mice. A competitive index (CI), which provides a value for the relative degree of virulence attenuation, was determined for each mutant after recovering bacteria from infected spleens, 48 h after intraperitoneal (i.p.) inoculation. The sifB and sseI mutant strains had no detectable virulence defect in the systemic phase of infection, whereas the srfJ and sseJ mutants had very slight and moderate virulence defects respectively (Table 1). In the case of sseJ, a CI of 0.47 ± 0.1 was obtained using a second mutant strain in which the entire coding region was deleted. Virulence was fully restored by the introduction of a plasmid expressing sseJ (Table 1). Therefore, the attenuated phenotype can be attributed to a null mutation in sseJ rather than a polar effect or secondary mutation in this strain. In view of these results, further studies were focused on SseJ.
Table 1. Competitive Index (CI) analysis of Salmonella typhimurium mutant strains.
a. Strains used in the mixed infection experiments are referred to by the relevant genotype and are described in Table 2.
b. The CI was calculated as the output ratio of mutant to wild-type bacteria, divided by the input ratio. The CIs shown are the means ± standard errors of at least three independent infections of mice. Mice were inoculated i.p. with a mixture of two strains comprising ~5 × 104 cfu of each strain. Mouse spleens were harvested after 48 h for enumeration of bacterial cfu. The different strains used were differentiated on the basis of antibiotic sensitivity. The same method was used to confirm the stability of each mutation after the infection.
A hallmark of S. typhimurium pathogenesis is its ability to replicate inside host cells. Therefore, intracellular growth of the sseJ mutant was compared with that of the wild-type strain and an ssaV mutant. ssaV encodes an essential component of the SPI-2 TTSS, the mutant strain being completely defective for SPI-2 mediated secretion (Beuzón et al., 1999). Replication assays were performed over 16 h in epithelial (HeLa) cells, a macrophage-like cell line (RAW 264.7), and in murine-elicited peritoneal macrophages. The ssaV mutant displayed a replication or survival defect in all three cell types, but the growth of the sseJ mutant was indistinguishable from that of the wild-type strain in HeLa and RAW cells. Only in the more restrictive growth environment within elicited peritoneal macrophages was a small but reproducible net survival defect detected (Fig. 1).
SseJ expression within HeLa cells induces membranous structures with characteristics of SCVs and Sifs
To gain further insight into the function of SseJ, an epitope-tagged version of the protein (myc::SseJ) was constructed and expressed in HeLa cells after transfection. The distribution of this protein was examined first in relation to that of LAMP-1, a lysosomal membrane glycoprotein (lgp) and a well established marker of the SCV membrane and Sifs (García-del Portillo and Finlay, 1995; Beuzón et al., 2000). Confocal immunofluorescence microscopy of untransfected cells using an anti-LAMP-1 antibody revealed characteristic punctate labelling throughout the cell (Fig. 2A). In contrast, in cells expressing myc::SseJ, unusual structures associated with both LAMP-1 and myc::SseJ were observed. The number and size of these structures correlated with the level of myc::SseJ expression. In cells containing the lowest detectable level of myc::SseJ expression, LAMP-1 labelling was punctate but, as the level of myc::SseJ increased, the punctate labelling was replaced by large aggregates of LAMP-1, in most cases enclosed by what appeared (in the XY plane of the confocal microscope) to be myc::SseJ-positive rings of up to 1 μm in diameter (Fig. 2B). When analysed by a stacked series of confocal images, these compartments were shown to have globular structure and seemed to result from aggregation of LAMP-1 vesicles (results not shown). These compartments were not visible under the phase-contrast microscope. For convenience, we refer to them as globular membranous compartments (GMCs). GMCs were not a consequence of expression of the myc epitope, because similar structures were observed in HeLa cells transfected with a green fluorescent protein (GFP)::SseJ chimera (results not shown), and were specific to SseJ, as no abnormal structures and typical LAMP-1 labelling were observed in cells transfected with a myc::SifB protein (results not shown).
To determine if the formation of GMCs is dependent on the predicted acyltransferase/lipase activity of SseJ, a mutant version of the protein was constructed by substituting the predicted serine of the catalytic triad with valine (SseJS151V). Replacement of the active site serine in the GDSL motif of the lipases of both A. hydrophila and Vibrio mimicus leads to loss of enzyme activity (Hilton and Buckley, 1991; Shaw et al., 1994). Expression of myc::SseJS151V in HeLa cells had no noticeable effect on the distribution of LAMP-1, and no GMCs were formed. However, this mutant protein colocalized extensively with LAMP-1, indicating that the myc::SseJ protein may be targeted to LAMP-1-containing vesicles independently of its acyltransferase/lipase activity (Fig. 2A).
To investigate the composition of the GMCs in more detail, we examined transfected cells for several other cellular markers. The lgp LIMP-1 (also known as LAMP-3 and CD63 antigen) shares the same lysosomal targetting sequence as LAMP-1 (Hunziker and Geuze, 1996). The vacuolar proton ATPase (vATPase), which acidifies endocytic vacuoles, has a different origin and targeting process, but both LIMP-1 and vATPase accumulate on SCV membranes (Steele-Mortimer et al., 1999; Beuzón et al., 2000). In cells expressing myc::SseJ, the distribution of both proteins was altered dramatically, and both accumulated within GMCs (Fig. 3A, upper and middle panels, and data not shown). Cholesterol has also been found recently to associate with Sifs and SCVs (Brumell et al., 2001b). In untransfected cells, cholesterol (as revealed by filipin staining), was mainly distributed in the plasma membrane and small perinuclear vesicles (data not shown). However, in cells expressing myc::SseJ this was replaced by a nearly exclusive localization in GMCs (Fig. 3A, lower panel). As for LAMP-1, redistribution of LIMP-1, vATPase and cholesterol was dependent on the active site serine residue of wild-type SseJ (data not shown). Recent work has shown that intracellular S. typhimurium induces actin polymerization in a SPI-2 dependent manner (Méresse et al., 2001). Surprisingly, Texas red-phalloidin staining of myc::SseJ-transfected cells showed that the GMCs also frequently contained F-actin (Fig 4A and B). Therefore, the interior of these compartments appears to be a complex aggregate of membranes and component(s) of the cytoskeleton.
Other subcellular compartments appeared to be un-affected by overexpression of SseJ. The distribution in transfected cells of the lysosomal marker cathepsin D, and several Golgi markers (p115, p14Kβ, GM130 and Giantin), was indistinguishable from that in untransfected cells, although occasional colocalization between myc::SseJ and cathepsin D was observed (Fig. 3B and results not shown). Therefore, overexpression of myc::SseJ in HeLa cells results in the formation of unusual membranous compartments, with a composition that is characteristic of SCV membranes and Sifs.
Effects of SseJ expression in infected cells and in cells expressing SifA
In view of these results, it was of interest to know if ectopic expression of myc::SseJ in infected cells would alter the development of SCVs and Sifs. HeLa cells were first transfected with myc::SseJ, infected for 10 h with wild-type S. typhimurium, then fixed and labelled with α-LAMP-1 antibody. Sif formation was inhibited by the presence of ectopically expressed SseJ, and this inhibition required the active site serine (Fig. 5A). Sifs that were detected in myc::SseJ transfected, infected cells had significantly less LAMP-1 and aberrant morphology, whereas Sifs in SseJS151V transfected, infected cells appeared normal (Fig. 5B). Ectopically expressed myc::SseJ colocalized with LAMP-1 on SCV membranes and Sifs, and this association was not dependent on the active site serine (Fig 5B and C, and results not shown). GMCs were also observed in infected cells (Fig. 5C). Ectopic expression of myc::SseJ significantly inhibited intracellular replication of wild-type S. typhimurium. Bacterial replication was also inhibited by expression of myc::SseJS151V or myc::SifB, but to a lesser extent than by myc::SseJ (Fig. 5D).
As both SseJ and SifA localized to Sifs and SCV membranes (Fig 5B and C) (Beuzón et al., 2000), this raised the possibility that the two proteins might associate when expressed in the same cell. It has been shown previously that transfection of HeLa cells with SifA::GFP leads to the formation of large LAMP-1-containing compartments, and structures resembling Sifs (Brumell et al., 2001a). When HeLa cells were simultaneously transfected with plasmids expressing both myc::SseJ and SifA::GFP, GMC-like structures were formed. Interestingly, SifA::GFP was found within these structures, and no Sifs were observed (Fig. 6). Therefore, SifA and SseJ associate with the same structures when expressed in the same cell, but their lack of pixel-to-pixel colocalization suggests that the two effectors do not interact directly with each other.
Analysis of SCV membrane dynamics
The results described above suggest that the function of SseJ, like that of SifA, is related to the control of SCV membrane dynamics. However, no significant differences were observed between wild-type and sseJ mutant bacteria with respect to LAMP–1 association with SCVs in HeLa cells and macrophages, or Sif formation in HeLa cells, at different time-points (data not shown).
Therefore, to gain some insight into SCV membrane dynamics, we first investigated the role of SifA in more detail. sifA mutant bacteria gradually lose their vacuolar membranes several hours after bacterial entry into host cells, indicating that SifA is required to maintain the integrity of the vacuolar membrane (Beuzón et al., 2000). It is possible that SifA is required for recruitment and/or fusion of membrane vesicles to the SCV, which would lead to a depletion of LAMP-1 vesicles from elsewhere in the cell. Therefore, the distribution of LAMP-1 was re-examined in HeLa cells containing wild-type, ssaV or sifA mutant bacteria at different time-points. A redistribution of LAMP-1 was only observed in cells infected with wild-type bacteria, beginning approximately 6 h after bacterial entry. By 10 h, the typical uniform punctate distribution of LAMP-1 vesicles throughout the cell mostly disappeared and was replaced by a strong accumulation of LAMP-1 in the vicinity of the bacterial microcolony (Fig. 7). At this time-point, the distribution of LAMP-1-containing vesicles appeared normal in cells infected with either sifA or ssaV mutant bacteria, although as shown previously (Beuzón et al., 2000) only ssaV mutants were associated with LAMP-1-containing SCV membranes (Fig. 7). These results are consistent with the hypothesis that SifA is involved in recruiting membrane vesicles from other parts of the cell towards SCVs.
There are at least two possible explanations for the loss of the vacuolar membrane around sifA mutant bacteria. First, bacterial replication may itself be sufficient to cause their release into the host cell cytosol if insufficient membrane is recruited. Alternatively, the bacteria could possess an activity that destabilizes the SCV membrane and is opposed by SifA in wild-type SCVs. If the sifA mutant strain loses its vacuolar membrane because it cannot recruit sufficient membrane to enclose replicating bacterial cells, then combining a sifA mutation with additional, unrelated mutations which inhibit intracellular bacterial replication, would be predicted to result in a strain which undergoes less vacuolar membrane loss than a strain carrying only a single sifA mutation. It has been shown previously that sifA mutant strains are not deficient for replication in HeLa cells (Stein et al., 1996), but replicate poorly in macrophages (Beuzón et al., 2000; Brumell et al., 2001a). We therefore introduced auxotrophic aroC and purD mutations (which completely prevented intracellular replication of wild-type S. typhimurium (results not shown)) into a sifA mutant, and compared the numbers of SCVs associated with LAMP-1 (as a marker for the SCV membrane) 10 h after bacterial uptake into host cells. Although no replication of the sifA aroC purD triple mutant occurred in HeLa cells and macrophages (results not shown), the loss of LAMP-1 labelling was not significantly different to that of the sifA single mutant in either host cell type (Fig. 8A). Therefore, bacterial replication does not seem to play a significant role in the loss of vacuolar membrane around sifA mutant bacteria.
We then determined if the loss of vacuolar membrane associated with mutation of sifA requires bacterial protein synthesis. HeLa cells were infected with the sifA mutant and then treated with tetracycline at 6 h post uptake, when onset of loss of the vacuolar membrane occurs (Beuzón et al., 2000). Cells were fixed at 10 h and labelled for LAMP-1. Whereas only 27 ± 11.3% of sifA mutant bacteria were associated with LAMP-1 at 10 h, this increased to 82 ± 13.7% after exposure to tetracycline, a level similar to that of wild-type bacteria that had not been exposed to the antibiotic (Fig. 8B). Similar results were obtained in RAW macrophages (Fig. 8B). Therefore, loss of the vacuolar membrane around sifA mutant bacteria does require bacterial protein synthesis.
Loss of vacuolar membrane around sifA mutant bacteria requires SseJ
As myc::SseJ inhibited Sif formation in transfected, infected cells, this suggested that SifA and SseJ might have opposing effects on Salmonella-directed membrane dynamics. To investigate if SseJ accounts for the loss of vacuolar membrane around sifA mutant bacteria, we analysed the effect of the sseJ mutation in the absence of SifA. We constructed a sifA sseJ double mutant strain, and compared its level of association with LAMP-1 with that of a sifA single mutant in HeLa cells at 10 h after bacterial entry. In contrast to the sifA mutant, the majority of sifA sseJ double mutant bacteria remained within LAMP-1-positive compartments, although the overall percentage of bacteria associated with the marker was slightly lower than for wild-type or sseJ mutant bacteria (Fig. 9). An increased association of the double mutant with LAMP-1 over the sifA mutant was also observed in RAW macrophages (results not shown). These results show that a functional SseJ protein contributes to the loss of the Salmonella vacuolar membrane in the absence of SifA.
The presence of a vacuolar membrane around the sifA sseJ double mutant raised the possibility that SseJ compensates for the absence of SifA, both in terms of loss of vacuolar membrane, and for the strong virulence defect of the sifA mutant (Table 1). To analyse this possibility, the virulence of the sifA sseJ double mutant strain was compared with that of the sifA single mutant by mixed infection (Beuzón et al., 2001). The resulting COI (0.75 ± 0.004) is significantly different from 1.0 (P = 0.002); therefore, the addition of the sseJ mutation to a strain carrying a sifA mutation not only fails to compensate for its virulence defect, but actually increases its level of attenuation slightly.
In this study, we have investigated the function of the SseJ effector protein of the TTSS encoded by SPI-2. Many candidate effector proteins for this secretion system have been proposed (Hensel et al., 1998; Miao and Miller, 2000; Worley et al., 2000) but only SpiC and SifA have been demonstrated to act as effectors and have been functionally characterized to some degree. SpiC is an inhibitor of various aspects of endocytic trafficking (Uchiya et al., 1999), and SifA is required to maintain the integrity of the SCV membrane (Beuzón et al., 2000). Although strains carrying mutations in sifB and sseI had no detectable virulence defect by CI tests following i.p. inoculation of mice, it is possible that they contribute to SPI-2 TTSS function in ways that are not detectable by this assay. For example, they could function during stages of infection before colonization of the spleen, such as in intestinal epithelial cells or in Peyer’s patches (Cirillo et al., 1998). Alternatively they could be required for pathogenesis of other S. typhimurium hosts, may have genetically redundant roles or make no significant contribution to S. typhimurium virulence.
This study was focused on SseJ because the corresponding mutant had a detectable virulence and macrophage-survival defect. The sequence similarity shows that SseJ is a member of the ‘GDSL’ family of acyltransferases/lipases (Upton and Buckley, 1995; Miao and Miller, 2000), and its function was explored further by expression in epithelial cells, and examining its effects on markers of SCV membranes. Its effects were also compared with those of a mutant version containing a single amino acid substitution of the predicted nucleophilic serine of the catalytic triad, to abolish its enzymatic activity. Expression of the wild-type, but not the mutant form of this protein, resulted in the formation of novel globular structures (GMCs), whose composition appears to be similar to that of Sifs and SCV membranes. The distribution of Golgi and lysosomal markers appeared to be normal, suggesting that overexpression of SseJ did not perturb the biosynthetic or terminal endocytic machinery of these cells.
GMCs contain F-actin and membranes enriched in cholesterol, vATPase, LAMP-1 and LIMP-1. These structures are globular, up to 1 μm in diameter, and are surrounded by SseJ, which itself seems to be membrane-associated. Further detailed analysis of the organization of these structures will require electron microscopy and possibly biochemical fractionation. In view of their composition, GMCs may be partly of late-endosomal origin. The presence of F-actin in GMCs raises the possibility that SPI-2-dependent actin assembly on the SCV (Méresse et al., 2001) could be an inherent property of specific membranes recruited through the action of the SPI-2 TTSS, rather than a result of F-actin nucleation on the SCV membrane induced by a SPI-2 effector protein. The absolute dependence of GMCs on the serine residue of the putative active site strongly suggests that they arise by acyltransferase/lipase activity of this protein. Determination of the intracellular substrate of SseJ will help to elucidate the biogenesis of GMCs and, more importantly, to understand its role in regulating membrane dynamics of the SCV.
Although GMCs are unlikely to have any physiological significance (as no structures resembling these compartments have been observed in cells infected with wild-type S. typhimurium), their similar composition to SCV membranes and Sifs; (García-del Portillo et al., 1993; García-del Portillo and Finlay, 1995; Beuzón et al., 2000) is noteworthy. The fact that transfection of HeLa cells with a vector encoding SifA leads to extensive vacuolation of LAMP-1+, vATPase+ and cathepsin D− compartments (Brumell et al., 2001a) suggests that SseJ and SifA may target the same endocytic compartments but act on them differently. This observation, together with the effects of ectopic expression of SseJ on SCVs and Sifs in infected cells, led us to speculate that SseJ and SifA might have related functions.
We have shown previously that an S. typhimurium ssaV mutant, which is unable to secrete any proteins (including SifA) through the SPI-2 TTSS, does not lose its vacuolar membrane (Beuzón et al., 2000). As the SPI-2 effector SpiC is thought to inhibit interactions between SCVs and endosomal compartments (Uchiya et al., 1999), we proposed that the vacuolar membrane around an ssaV mutant could, in contrast with that of the wild-type strain, be derived from interactions with the late endocytic pathway (Beuzón et al., 2000). Two findings have led us to revise this view. First, analysis of the trafficking of a SPI-2 TTSS null mutant in macrophages showed that this strain had a low level of association with late endosomes and lysosomes, similar to that of wild-type bacteria (Garvis et al., 2001). Hence, it seems unlikely that a SPI-2 null mutant would have greater access to these endocytic compartments than wild-type bacteria. Second, we show here that loss of vacuolar membrane from sifA mutant bacteria requires the action of SseJ. Therefore, the presence of SCV membrane around ssaV mutant bacteria could be explained simply by an inability to translocate SseJ.
Although the sifA sseJ double mutant retained a vacuolar membrane, we can rule out the possibility that the sseJ mutation has simply compensated for the loss of sifA, because the sifA sseJ double mutant was severely attenuated in virulence compared to wild-type bacteria, and was slightly more attenuated than a sifA mutant. Therefore, although the double mutant strain resides within a vacuole, the properties of this vacuole must be abnormal, and restrictive for bacterial growth.
What is the role of SseJ in relation to the SPI-2 TTSS and intracellular replication of S. typhimurium? To replicate intracellularly yet remain within a vacuole, S. typhimurium must ensure a net recruitment of membrane to accommodate the growing bacterial population. Intracellular S. typhimurium cells are closely associated with vacuolar membrane and often appear to be in distinct vacuoles (for example, see Fig. 9A). Therefore, the controlled expansion and separation of the vacuolar membrane are likely to be important activities regulated by S. typhimurium. Although we have not identified its (presumably lipid) substrate, the amino acid sequence and differential effects of the wild-type and mutant proteins in host cells indicate that SseJ is an acyltransferase/lipase. SseJ localizes to the SCV membrane and is required for the loss of this membrane around sifA mutant bacteria. SifA is required for the redistribution of LAMP-1 containing vesicles to the vicinity of the bacterial microcolony (Fig. 7), in which a proportion is presumably incorporated into the SCV membrane. The SCV membrane is likely to be a highly dynamic structure, undergoing fusion with selected vesicular compartments, and also modification, budding and scission. SseJ could be required for modification of SCV membrane lipid(s), to facilitate one or more of these processes. The lack of a dramatic intracellular phenotype for the sseJ single mutant indicates that its function is likely to be highly localized, specific and to involve other proteins, which may compensate in part for its absence in the sseJ mutant strain.
Bacterial strains and growth conditions
The Salmonella typhimurium strains used in this study are listed in Table 2. Bacteria were grown in Luria–Bertani (LB) medium supplemented with ampicillin (50 μg ml−1), kanamycin (50 μg ml−1), chloramphenicol (35 μg ml−1 for plasmid-containing strains, 10 μg ml−1 for chromosomal integrants) and tetracycline (15 μg ml−1), for strains resistant to these antibiotics (Ampr, Kmr, Cmr, and Tetr respectively). Insertion mutations in the sseI, sseJ and srfJ genes were obtained by site-directed chromosomal integration of the suicide plasmid pGP704 containing a short portion of the 5′-region of the corresponding coding sequence, as described previously (Miller and Mekalanos, 1988). The fragments of sseI, sseJ and srfJ ligated into pGP704 did not extend beyond the regions encoding the signal for translocation, and were obtained by polymerase chain reaction (PCR) from S. typhimurium strain 12023 genomic DNA, using the primers SSEI-1/–2, SSEJ-1/–2 and SRFJ-1/–2 respectively (Table 3). Mutations were transduced by phage P22 as described previously (Davis et al., 1980) into nalidixic acid-resistant strain 12023 (Shea et al., 1999) before use in this study. The stability of the mutations was verified after each infection experi-ment by plating on LB, and LB with the appropriate antibiotic. A deletion mutant for the sseJ gene was obtained following the method described by Datsenko and Wanner (2000), using SSEJ-CAT1 and SSEJ-CAT2 as primers. A non-polar deletion mutation in sifB was obtained by allelic exchange. The entire sifB gene and its flanking regions were amplified by PCR from S. typhimurium strain 12023 genomic DNA, using the primers SIFB-2 and SIFB-3. The 1.7 kb PCR product was ligated into pCRII-TOPO (Invitrogen), generating plasmid pID829. This plasmid was used as a template for an inverse PCR using the primers SIFB-6 and SIFB-7. The 5.1 kb PCR product, containing terminal SmaI sites, was digested and re-ligated to generate plasmid pID830. The SmaI fragment of pSL1 (Lukomski et al., 1996) containing a chloramphenicol resistance cassette without promoter or terminator se-quences was ligated into the SmaI site of plasmid pID830, to generate plasmid pID831. This plasmid was digested with BamHI and NotI, and ligated into the BglII–NotI sites of the suicide vector pKAS32 (Skorupski and Taylor, 1996) to generate plasmid pID832. The resulting plasmid pID832 was transferred by conjugation from Esche-richia coli S17–1λpir to S. typhimurium, and exconjugants were selected as described previously (Shea et al., 1999).
Table 2. S. typhimurium strains used in this study.
Double and triple mutant strains were constructed by P22-mediated transductions (Davis et al., 1980).
Plasmid pFVP25.1, carrying gfpmut3A under the control of the rpsM constitutive promoter (Valdivia and Falkow, 1997), was introduced into bacterial strains and used for fluorescence visualization when indicated. The complementing plasmid psseJ is a derivative of pACYC184 (Chang and Cohen, 1978) carrying the sseJ gene under the control of a constitutive promoter. A DNA fragment including the complete open reading frame (ORF) of sseJ and its ribosomal binding site (rbs) was amplified using PCR from 12023s genomic DNA using SSEJ-C1 and SSEJ-C2 primers. The 1.2 kb PCR product, containing terminal BamHI and SalI sites, was digested and ligated into pACYC184 within the TetR gene, generating psseJ. The expression plasmid pmyc::sseJ was constructed as follows: a DNA fragment including the complete ORF of sseJ was amplified using PCR from genomic DNA using SSEJ3-AB and SSEJ5-AB primers. The 1.2 kb PCR product, containing terminal BamHI and EcoRI sites, was digested and ligated into the plasmid pRK5-myc (Lamarche et al., 1996), within the same reading frame as that of the c-myc epitope tag of the plasmid, generating pmyc::sseJ. This plasmid expresses a full-length version of SseJ bearing an N-terminal fusion to the c-myc epitope tag. Following the same strategy, pmyc::sifB was constructed using SIFB3-AB and SIFB5-AB primers.
The expression plasmid pmyc::sseJS151V was constructed as follows: two PCRs with genomic DNA as template were performed, using SSEJ5-AB with SSEJ3-S151V and SSEJ3-AB with SSEJ5-S151V as primers. The products of the two PCRs were used as templates for another PCR, using SSEJ3-AB and SSEJ5-AB as primers. The resulting PCR product was digested and ligated into the plasmid pRK5-myc, into the same reading frame as the c-myc epitope tag coding sequence included in the plasmid, generating pmyc::sseJS151V. This plasmid expresses a full-length version of SseJ bearing an N-terminal fusion to the c-myc epitope tag, and where the AGC codon encoding serine residue 151 of SseJ has been replaced with a GTC codon, encoding valine. Expression plasmid pgfp::sseJ was constructed as follows: a DNA fragment including the complete ORF of sseJ was amplified using PCR from 12023s genomic DNA using SSEJ-GFPC1 and SSEJ-GFPC2 primers. The 1.2 kb PCR product, containing terminal EcoRI and BamHI sites, was digested and ligated into the plasmid pEGFP-C1 (Clontech) into the same reading frame as the gfpmut1 coding sequence included in the plasmid, generating pgfp::sseJ. This plasmid expresses a full-length version of SseJ bearing an N-terminal fusion to the green fluorescent protein (GFP). All the constructions were verified by DNA sequencing.
Mouse mixed infections
Female BALB/c mice (20–25 g) were used for Competitive Index studies (CIs and COIs). At least three mice were inoculated i.p. with a mixture of two strains comprising ≈5 × 104 colony-forming units (cfus) of each strain in physiological saline, and the CIs or COIs were determined from spleen homogenates 48 h post inoculation as described previously (Beuzón et al., 2001).
Each CI or COI value is the mean of at least three independent mice infections. Student’s t-test was used to analyse the CIs or COIs of single mutant strains versus the wild-type strain (wt v a−) with the null hypothesis: mean is equal to 1.0. Probabilities (P) of 0.05 or less were considered significant.
Antibodies and reagents
The mouse monoclonal anti-c-myc antibody 9E10 was provided by E. Caron (London University College) and was used at a dilution of 1:200. Rabbit polyclonal anti-c-myc antibody PRB-150C was purchased from Babco Laboratories (Richmond, CA) and was used at a dilution of 1:100. The mouse monoclonal antibodies anti-LAMP-1 H4A3 and anti-CD63 H5C6, developed by J. T. August and J. E. K. Hildreth, were obtained from the Developmental Studies Hybridoma Bank, developed under the auspices of the NICHD and maintained by the University of Iowa (Department of Biological Sciences, IA), and were used at a dilution of 1:1000. Anti-vATPase mouse monoclonal antibody OSW2 was obtained from Dr S. B. Sato (Kyoto University, Kyoto, Japan) and was used at a dilution of 1:1000. Anti-LAMP-1 rabbit polyclonal antibody 156 (Steele-Mortimer et al., 1999) was used at a dilution of 1:1000. Anti-Salmonella goat polyclonal antibody CSA-1 was purchased from Kirkegaard and Perry Laboratories (Gaithensburg, MD) and was used at a dilution of 1:400. Anti-cathepsin D rabbit polyclonal antibody RC242 was purchased from Scripps Laboratories (San Diego, CA) and was used at a dilution of 1:1000. The rabbit polyclonal antibodies anti-GM130, anti-P115, anti-P14Kβ and antigiantin were obtained from Dr A. De Matteis (Consorzio Mario Negri SUD, Chieti, Italy) and used at dilutions of 1:1000, 1:500, 1:2000 and 1:2000 respectively. Texas red sulphonyl chloride (TRSC)- and fluorescein isothiocyanate (FITC)-conjugated donkey anti-mouse and anti-rabbit antibodies, and cyanine 5 (Cy5)-conjugated donkey anti-goat antibody were purchased from Jackson Immunoresearch Laboratories (West Grove, PA), and used at a dilution of 1:400. TRSC-conjugated phalloidin was purchased from Molecular Probes and used at a dilution of 1:50. The fluorescent cholesterol binding drug filipin was obtained from Dr D. Young (Imperial College, London, UK) and used at a concentration of 50 μg ml−1.
RAW 264.7 cells were obtained from ECACC (ECACC 91062702). HeLa (clone HtTA1) cells were kindly provided by Dr H. Bujard (Heidelberg, Germany). Cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS) and 2 mM glutamine at 37°C in 5% CO2. Peritoneal cells were harvested 4 days after BALB/c mice were inoculated i.p. with 5 mM periodate (De Groote et al., 1997). Cells were plated at a density of 5.5 × 105 cells per well in 24-well microtitre dishes and allowed to adhere for 2 h. Non-adherent cells were flushed out with pre-warmed RPMI containing 10% FCS. The adherent macrophages were incubated for a further 48 h before infection.
Bacterial infection of cultured cells and replication assays
HeLa cells were infected with exponential phase S. typhimurium as described previously (Beuzón et al., 2000). Macrophages were infected with opsonized, stationary phase S. typhimurium as described previously (Beuzón et al., 2000). To follow a synchronized population of bacteria, host cells were washed after 15 min (HeLa cells) or 30 min (macrophages) of exposure to S. typhimurium and sub-sequently incubated in medium containing gentamicin to kill extracellular bacteria. For enumeration of intracellular bacteria, macrophages were washed three times with phosphate-buffered saline (PBS), lysed with 0.1% Triton X-100 for 10 min and a dilution series was plated onto LB agar. Tetracycline was used at a final concentration of 15 μg ml−1 when indicated, and was shown to prevent bacterial replication, but did not affect bacterial viability over the course of the experiment.
For immunofluorescence, cell monolayers were fixed in 3.7% paraformaldehyde in PBS, pH 7.4, for 15 min at room temperature and washed three times in PBS, except when using the anti-cathepsin D polyclonal antibody when cells were fixed as described by Brumell and colleagues (Brumell et al., 2001b). Antibodies were diluted in 10% horse serum; 0.1% saponin in PBS. Coverslips were washed twice in PBS containing 0.1% saponin, incubated for 30 min with primary antibodies, washed twice with 0.1% saponin in PBS and incubated for 30 min with secondary antibodies. Coverslips were washed twice in 0.1% saponin in PBS, once in PBS and once in H2O, and mounted in Mowiol. Samples were analysed using a fluorescence microscope (BX50; Olympus Optical) or a confocal laser scanning microscope (LSM 510, Zeiss).
Transfection of HeLa cells
HeLa cells were transiently transfected by the calcium phosphate DNA precipitation method (Matthias et al., 1982), using 2 μg of plasmid DNA per 60 mm-diameter Petri dish, with each dish containing 6 × 105 cells. Then, 24 h after DNA addition, coverslips were collected and washed three times with PBS, and fixed for immunofluorescence microscopy. For the double transient transfection, equivalent amounts of plasmid DNA from pmyc::sseJ and psifA::gfp (Beuzón et al., 2000) were mixed and used for transfection as described. Bacterial infection of transfected HeLa cells was performed as described above, 14 h after DNA addition, and cells were fixed 24 h after DNA addition (10 h after infection).
We are very grateful to Lalita Ramakrishnan for suggesting the use of auxotrophic mutants to investigate the relationship between vacuolar membrane loss and bacterial replication. We would like to thank Steve Garvis for help with the tetracycline experiments, and Mei Liu for practical assistance. We also thank Stéphane Méresse for plasmid psifA::gfp, Josep Casadesús for strain TT311, Microscience Ltd for strain ST030 and Kate Unsworth for helpful comments on the manuscript. This work was supported by grants from the MRC to David Holden, and from the BBSRC to Edouard Galyov. Abigail Blakey was supported by a BBSRC postgraduate studentship.